Reviewer #1 (Public Review):

This is an interesting manuscript providing important new information on the mechanism of action of EROS in the generation of superoxide by the NADPH oxidase of neutrophils. The authors have shown in previous publications that EROS deficiency results in defective NOX2 activity and thus represents a hitherto unrecognised, rare form of chronic granulomatous disease. They now show how EROS is involved in oligosaccharide transfer during the maturation of gp91phox and also extend what is known about the role for EROS in regulating expression of the P2x7 ion channel.

The results presented in the manuscript are supported by findings from a variety of techniques and for the most part, are convincing and well presented. However, I do have queries about certain aspects of the manuscript.

1. Figure 1<br /> The much lower EROS expression when gp91phox is expressed warrants a comment.<br /> Fig 1 G. Please explain what fold change represents. From F, zero time expression appears much more than the 1.5 fold higher shown in G for the EROS-expressing cells. This needs explaining. With the very high error bars (presumably for the EROS sample although this is not clear) overlapping zero I find it hard to conclude anything from this figure.

2. P 9 line 9 states that Fig 1H shows that cycloheximide increases expression. Yet it appears from the legend that cycloheximide is present in all samples and it is EROS that increases expression. Please clarify.

3. Fig 3A&B and p12 1st para. The identification of OST as a binding partner is interesting and a significant novel finding. However, the presentation of this information appears to me to be unduly complex and more information is required. Not all the readers will be familiar with the details of SAINTexpress methodology and more explanation of what is being shown would be helpful. At the least, a supplementary Table of the 59 identified proteins would be helpful, plus information on controls to establish selective pull down by EROS and on how the blue spots in A relate to the proteins. Also please make it clearer which of the proteins in B were identified and the relevance of showing all the steps in the pathway.

4. Figure 6. This contains a large amount of information. Although interesting, I am concerned that the authors may be trying to include too much at the expense of the necessary detail for some of the experiments. For example, the EROS -/- +ATP scattergram on the left of Fig 6E does not seem to agree with the right hand graph. I would also like to see the mean values for the 5 experiments in Fig 6G shown. Most importantly, insufficient information is given for Fig 6H. I don't think I missed it but I could find no details about the experiment in the Methods section. We need to know more about exactly how many animals in were in each group (death of 1 animal appears to equate to 5% of total - how does this relate to >10 in total), how signs of illness were monitored and related to death, and generally more about the conditions of the experiment. Alternatively, this may be better left to a more detailed study.

Money suggests recourses. Things kids can obtain, chances they can have and people they can meet. I had a former roommate who said that we come from different hierarchies, because her family income is more than 10 times of mine. I do see some difference between us, but I think the difference is not as big as the poverties and the riches. Middle class families can basically make sure that their children get enough resources. The richer families may hold better resources, but this is a gap that somehow not that big. The problem for now is how to give the kids from poverty families get the resources, no matter how good the resources are, I hope at least they can have the basic needs being met.

Author Response:

Reviewer #1 (Public Review):

The manuscript by Liu et al investigates how MRI can be used to detect the earliest stages of CNS infections and how MRI can also be used as a surrogate readout for treatment efficacy. Authors demonstrate convincingly that microbleeds, as evidenced by unusual dark spots in the brain of mice infected with a virus that infects the brain, occurred at the earliest stages of viral infection. Authors also convincingly demonstrate that the infusion of virus-specific immune cells, when delivered at the right time and at the right dose, could reduce these microbleeds. Importantly, authors showed that the wrong dose could be detrimental.

The authors cast this study as a method for improving research and discovery in immunotherapy context and the study is convincing in its conclusions regarding imaging microbleeds and the immunotherapy tested herein. While authors do not directly suggest so, these findings extend the significance of this work beyond research and development of immunotherapies by providing a potential early detection mechanism for viral infection in the brain. This may be feasible as the MRI methodologies for detecting these phenomena are generally translatable to clinical imaging scenarios, though the imaging resolution may not.

Weaknesses in the report revolves around the value of and the ability to image magnetically labeled T cells in the presence of microbleeds.

1) Authors developed a magnetic particle coated with fluorescent molecules and antibodies specific for CD8+ T cells. They labeled these T cells with particles for detection by MRI. They then wanted to follow the accumulation of these cells in the brain following infusion and viral infection by performing MRI using parameters that amplify the signal of the attached label. The rationale for these experiments was to determine if immune cell infiltration preceded vascular compromise. This suggests the expectation for active chemotactic migration or other signaled accumulation rather than leakage. When authors tested their magnetically labeled T cells for functional impairment due to the presence of attached magnetic particles, they did not test for deficits to migratory capabilities, such as in standard transwell migration assays. Others have shown (see https://doi.org/10.1038/nm.2198 for example) that T cell migration is very sensitive to the type of attached nanoparticle as well as the surface coverage. Perhaps authors should temper their claims that magnetically labeling of T cells does not alter T cell function without at least an assay of this critical function. Further, the fluorescence microscopy shown in Figure 7D is of insufficient resolution to claim that MPIOs are inside cells. Electron microscopy should be used to determine this.

We thank this Reviewer for the comments. In this Revision, we added EM data to confirm the cellular location of MPIOs (Fig 7D and S7D). The EM experiment also added another layer of information for improving our cell isolation method. We improved our FACS experiment by narrowing down the MPIO positive gating to exclude the T cell population that labeled with high numbers of MPIO particles, which may affect T cell functions, and some crosslinked MPIO particles that formed during conjugation (Fig 7B and S7A). The yield of FACS of MPIO-labeled T cells is ~8.3%. As quantified from EM images, 91% MPIOs were localized intracellularly (Fig 7E). We agree that labeling T cells with nanoparticles might alter key T cell functions. We have improved the manuscript by putting this caution and reference. We also added T cell migration assay results (Fig 7G). Labeling CD8 T cells with MPIO did not affect T cell migration. This adds to our other in-vitro assays that T cell function is not significantly affected. There is in-vivo evidence as well that labeled T cells are functional. In Fig 8E-I, MPIO-labeled T cells were found in the brain, which showed that labeled T cells can migrate into the brain. In addition, a key phenotype of virus specific CD8 T cells in this model is the therapeutic function described in the manuscript. Labeling virus specific CD8 T cells with MPIO did not affect their therapeutic function. Quantification of bleeding in the OB and brain on day 6 and 11 verified the therapeutic effects of MPIOlabeled OT-I T cells (Fig 1E and 2C vs Fig S9C and D). We added discussion of these points in this Revision.

2) Regarding the use of imaging the accumulation of magnetically labeled T cells, authors show evidence that magnetically labeled T cells accumulate in areas of the brain that as yet do not present with microbleeds but do have the histological hallmarks of vascular inflammation. This corroboration is intriguing but only provable with a serial imaging study in the same animal, which was not performed. Authors are also encouraged to report on the frequency in which a magnetically labeled T cell was present in a pre-vascular compromised inflammatory environment. The bulk of the results on imaging magnetically labeled T cells essentially show that the accumulation of magnetically labeled T cells enhances the ability to detect microbleeeds that otherwise were perhaps too small to detect (Sup Fig 8). Given the lack of data supporting the retained migratory capacity of magnetically labeled T cells, one wonders then, whether magnetically labeled T cells are indeed trafficking to the brain or are passively arriving in the brain, and might some vascular magnetic particle accumulate in an early inflammation or leak into the microbleed on its own and similarly enhance the ability to detect the otherwise undetectable microbleed. A series of controls would be useful to answer these questions, perhaps testing the administration of magnetic particles alone, and/or magnetically labeled non-CD8+ T cells. Authors are also encouraged to report on the frequency in which a magnetically labeled T cell was present in a pre-vascular compromised inflammatory environment versus in the microbleed, as measured by MRI and histology.

Distinguishing bleeding from T cells is a key challenge for doing a serial MRI study in the same animal. In the new Fig 8I and Fig S8, we did a study using time-lapse MRI on the same mouse from 20 to 24 hr-post infection. We observed the appearance of hypointensities at the center of the bulb at 22 hr which is prior to bleeding in this area. Bleeds were observed at the GL, but not at the center of the bulb by IHC. Thus, we were able to time the entrance of T cells in this area of the brain. We were not able to find migration tracks of T cells from the outer GL layer into the center of the bulb. This is consistent with the idea that T cells infiltrate directly into areas with virus prior to vessel breakdown and microbleeds. We didn’t observe a very significant change in the location of T cells from 22 to 24 hr on the distance scale of MRI. There are two possibilities to explain our inability to detect T cell movement over a 2 hr time interval: 1.) the T cells under investigation may have been attached to blood vessels and required more time to extravasate. surface due to inflammation, and it might take some time for extravasation, or 2.) although T cell velocities in the CNS have been clocked at ~10 µm/min (Herz et al., 2015), their paths are often tortuous and influenced by antigen presenting cells displaying cognate peptide MHC as well as local chemokine gradients. Thus, upon entering a site of viral infection, the labeled T cells may not have traveled far enough in 2 hrs for us to detect their movement by MRI. We did not image mice beyond 24 hrs post-infection due to the possibility of bleeding. We added this discussion. Quantification of the frequency in which a MPIO labeled T cell was present in a region where no bleeding was detected versus in a region with a microbleed was added in Fig 8H. In the ONL/GL, 85% of MPIO-labeled T cells were in the region with microbleeds and 15% were in a region where no tissue bleeding was detected. In the MCL/GCL areas, no evidence for bleeding was detected. Magnetic labeling of CD8 T cells doesn’t reduce their migratory capacity in an in-vitro migration assay (Fig 7G). This adds to other in-vitro assays that the labeled T cells are functioning. Labeled T cells had therapeutic efficacy like unlabeled T cells and labeled T cells were found at the center of the bulb (Fig 8F-I) with no bleeds as well as in other brain regions. Based on these observations, we think that MPIO-labeled T cells are functioning and trafficking in the brain. A previous study showed that non-CD8 T cells, such as monocytes/macrophages, CD4 T cells, and neutrophiles also migrate into the OB and are involved in the immune responses in this model [(Moseman et al., 2020), Fig 2E]

Reviewer #2 (Public Review):

[...]

Weaknesses:

• Individuals with systemic infections or other underlying condition may have microbleeds due to inflammation or hypertension. The etiology of microbleeds is thus not necessarily tied to CNS infections. Investigation of potential cerebrovascular microbleeds following systemic or respiratory infections not affecting the CNS may shed light on this possibility which may also provide alternative interpretation of neurological symptoms associated with on CNS invasive infections.

This is an important issue. Prior work has shown that virus in this model is cleared quickly (2 to 3 days) from the periphery (Ramsburg et al., 2005; Roberts et al., 1999). This is likely due to the fact the virus is inoculated through the nose. It is clear in this model that virus infects the brain, that bleeding corresponds to sites of high viral load, and bleeding can be modulated by blocking immune infiltration into the brain. However, the quantitative role of peripheral influences such as high blood pressure could be important and will be checked as this work proceeds.

• Representative colocalization of virus infected endothelial cells with red blood cells (RBCs) is shown in Fig 4. However, a more quantitative assessment indicating how many areas or hypointensities were evaluated for virus-localization with RBCs, and how many of these revealed colocalization versus virus or RBC only would strengthen interpretation.

Fig 4 shows that VSV can infect vascular endothelial cells and cause bleeding. Hypointensities were not measured in this Figure. We quantified the numbers of VSV infected vessels, colocalizing and not colocalizing with bleeds. Fig 4D was added with this new data.

• A limitation clearly acknowledged by the authors is that hypointensity spots detected by MRI cannot distinguish microbeads from MPIO-labeled T cells.

As in our response to Reviewer 1, this is a critical next step since bleeding so often occurs with immune cell infiltration in the brain. We have discussed potential approaches and have added the idea that development of more sensitive MRI contrast agents and quantitative T2* analysis especially at different magnetic field strengths may be approaches to accomplish this. It will be crucial for MRI cell tracking under the condition of bleeding, which is one common pathology associated with many diseases.

Author Response:

Reviewer #1 (Public Review):

In this manuscript, the authors exploit retinal cell proliferation and neurogenesis in zebrafish to study banp, a protein that is essential in humans and embryonic lethal in mice. The authors performed large-scale mutagenesis and identified a mutant known as "rw337" that compared to WT cells the mutant zebrafish have smaller eyes and optic tectum. They found that the retinas of these mutants have mitotic-like round cells that accumulate indicating mitotic arrest. Sequencing of these mutants identified that the rw337 mutant gene encodes a truncated banp protein. Expression of WT Banp occurs primarily in retinal and neuronal cells in Zebrafish. Interestingly, rw337 showed significant decrease in retinal photoreceptors number and neuronal formation within the OPL and IPL were morphologically disrupted and had fewer cells. The authors found that rw337 cells have increased numbers of DSBs in the retina over time (via TUNEL) assays. They found that mitotic defects and apoptosis are spatially and temporally occurring in distinct regions of the retina as prolonged phosphorylation of histone H3, which indicates an issue in exit of mitosis, occurred in apical surface of the neural retina whereas apoptosis occurred in retinal progenitor cells (via Caspase 3 staining). The authors then went on to examine the role of replication stress regulators like p53, atm, and atr and showed that protein and RNA levels of banprw337 were increased and upregulated. As p53 binds banp in zebrafish, it was not surprising that regulators of p53 were enhanced in banprw337 mutants. Intriguingly, the authors found that two genes which are essential for chromatin segregation were downregulated in banprw337 mutants and banp morphants as a result of chromatin accessability decreases near the TSS of resulting in decreased transcriptional activity of cenpt and ncapg genes. Finally, the authors temporally monitored mitosis in mitosis of banprw337 mutants and found that chromosomal segregation is abnormal and takes longer. The authors have performed a thorough analysis of the impact of the banp gene on retinal biology and its importance regulating replication stress response and cenpt and ncapg expression. This paper is important to retinal biology, genome stability, and replication stress response fields and requires minor revision.

Strengths:<br /> • These studies exploit zebrafish retinal development and its cell-cycle regulation as knockout of Banp/ SMAR1 is an essential gene in human cells and embryonic lethal in mice.<br /> • The authors show that this gene is involved in replication stress responses involving p53, atm, and atr signaling.<br /> • The authors show that banp is required for chromatin segregation factors and chromatin accessability by binding to banp sequences (TCTCGCGAGA) upstream of specifically cenpt and ncapg. Interestigly the mutant rw337 had decreased chromatin accessability near the transcript start sites of these genes. This is an elegant study of how a gene is regulating the transcription of two genes essential for chromatin segregation.<br /> •<br /> Weaknesses:<br /> • The authors could highlight the protein names of both zebrafish and humans throughout the text using standard nomenclature description with humans proteins all capitalized etc... This will enable the reader to understand their findings in the context of fascinating biology and human disease/cancer.

We have revised nomenclature of genes and proteins throughout the text, consistent with nomenclature conventions as follows.

species /gene/ protein zebrafish / banp / Banp mouse / Banp / BANP human / BANP / BANP

In the revised manuscript, we have used human/mouse/zebrafish nomenclature in sentences relating findings that were achieved using human/mouse/zebrafish samples, respectively.

• As banprw337 mutants show such severe morphological disruption a discussion on the impact of this work for the vision community could strengthen the importance of understanding how this gene functions.

We appreciate this suggestion. In response to comments from the editor and reviewer #2, we have revised the Introduction to mention that vertebrate retina is an excellent model system to dissect mechanisms of cell-cycle regulation and DNA damage response-mediated neuronal cell death. We believe that our banp paper will have an impact on the retinal community. Furthermore, in addition to the role of Banp in cell-cycle regulation, most photoreceptors fail to differentiate in banp mutants, whose phenotypes are more severe than other retinal cell-types. Nuclear architecture, especially heterochromatin and euchromatin patterns, are quite differently organized in photoreceptor neurons and dynamically changed during rod photoreceptor differentiation, so we suspect that Banp may be important for photoreceptor differentiation through regulation of its nuclear organization. In the future, we will investigate this underlying mechanism. There are very interesting perspectives on retinal phenotypes in banp mutants, which may attract retinal and vision community researchers. However, these are diverse topics. So, in the current manuscript, we have limited the discussion to within cell-cycle regulation.

• Gamma H2AX phosphorylation is a global marker of DSBs and stalled forks. The authors did not note that H2AX phorylation is present and a marker of stalled replications forks.<br /> o PMID: 11673449, PMID: 20053681, doi:10.1101/gad.2053211, https://doi.org/10.1016/j.cell.2013.10.043 etc.

We appreciate this suggestion. We have added a statement on gamma-H2AX and cited appropriate references.

• As gamma H2AX phosphorylation recruits DNA repair factors like BRCA2, speculation of importance of these genes may be of interest to the DNA repair community.

We agree that to clarify which step or steps of DNA replication stress and the DNA repair mechanism are direct targets of Banp, it is important to consider how DNA repair factors are affected in banp mutants. Among Banp transcriptional target genes, we found that wrnip1 mRNA expression is significantly reduced in banp mutants. We have added these data to a new Figure 6-figure supplement 2. wrnip1 protects stalled replication forks from degradation and promotes fork restart during replication stress by cooperating with BRCA2. It was recently reported that WRNIP1 functions in translesion synthesis (TLS) and template switching (TS) at stalled forks, and also interstrand crosslink repair (ICR). It is possible that the loss of Wrnip1 causes defects in fork stabilization for restart, and ICR, leading to genomic instability. We have added this material to the Discussion and have revised a summary figure (Figure 7).

Reviewer #2 (Public Review):

Babu et al report the role of the zebrafish banp gene in the developing retina. They find that banp is required for faithful S-phase as well as mitosis.

Manuscript strengths: 1- The authors performed a large-scale mutagenesis screen and successfully identified a causative banp gene mutation from these efforts, which represent a significant amount of work. 2- The authors provide a substantial amount of cellular-level analysis of a host of cell cycle-related phenotypes in the banp mutant retina. The data are of high technical quality and the experiments are well-executed. For the most part, the data support the conclusions.

We are grateful for the reviewer’s high estimation of our work.

Manuscript weaknesses: 1- Banp mutants have numerous defects, and perhaps this is not unexpected for a nuclear matrix protein. I'm left wondering what insights are gained from the study beyond that the nuclear matrix is required for numerous cell cycle events?

As we mentioned in the Introduction, BANP was originally identified as a nuclear protein that binds matrix-associated regions (MARs). MARs are regulatory DNA sequences mostly present upstream of various promoters. MAR-binding proteins interact with numerous chromatin-modifying factors and regulate gene transcription. In addition, it was reported that BANP suppresses tumor growth, and that loss of BANP heterozygosity is associated with several cancers in humans. So, before we started this banp mutant analysis, we expected that loss of Banp might cause defects in the cell cycle. However, because the majority of prior studies on BANP have been done using in vitro systems, its physiological function was still ambiguous. Very recently, it was reported that BANP functions as a transcription factor that binds to Banp motifs and regulates essential metabolic genes. In this study, rather than focusing on the MAR domain, we used this Banp motif to search for direct transcriptional targets of Banp that may function in cell proliferation and differentiation in zebrafish retina. Our study provides the first in vivo evidence that Banp serves as an essential transcription activator of cell cycle genes, including cenpt, ncapg, and wrnip1 via Banp motifs. We believe that such a list of Banp direct target genes provides a new research avenue to discover more precisely how Banp functions in tumor suppression and that it will contribute to medical research on cancer therapy.

Our study did not investigate how the nuclear matrix itself is involved in Banp mutant phenotypes. However, since it is likely that the interaction between MAR domains and nuclear matrix may influence chromatin organization in the nucleus, BANP functions must depend on nuclear matrix configuration. So, while this question is interesting, we think it is beyond the scope of our current study. In addition, we are afraid that the term “matrix-associated nuclear protein” might mislead people to think that Banp is a regulator of nuclear matrix. To better clarify the relationship between Banp and nuclear matrix, we have revised “nuclear matrix-associated protein” -> “nuclear matrix associated region-binding protein” in the text.

2- Why did the authors focus on the eye? It is unclear whether this study revealed a sensitivity to eye development regarding nuclear matrix function specifically, or it was just a convenient place in the animal to look.

Historically, molecular and cellular mechanisms that regulate cell proliferation and differentiation in the nervous system has been intensively studied using the vertebrate retina, because retinal neuronal cell types are fewer than those of other brain regions and its neural circuits are also simpler than those of other brain regions. Furthermore, many research groups, including us, have identified zebrafish retinal mutants, including mutants that show defects in cell-cycle regulation and DNA damage response. Indeed, our group has investigated this topic using retinal apoptotic mutants for the last 20 years. Thus, we focus on the zebrafish retina, because the retina is an excellent in vivo model system to dissect mechanisms of cell-cycle regulation and DNA damage response. To emphasize the importance of this excellent in vivo model system to researchers beyond the retinal community, we have revised in the Introduction as follows. "The developing retina is a highly proliferating tissue, in which a spatiotemporal pattern of neurogenesis is tightly coordinated by cell-cycle regulation. So, vertebrate retina provides a great model for studying how cell-cycle regulation, including DNA damage response ensures neurogenesis and subsequent cell differentiation."

3- I found the conclusions regarding mitosis to be contradictory. The authors at first emphasize mitotic arrest, but then characterize chromosome segregation defects. How can chromosomes segregate if cells are arrested in mitosis?

We apologize for the confusion due to our incorrect usage of the term “mitotic arrest.” Mitotic arrest was one of possibilities that we considered when first examining banp mutant phenotypes, in which we just observed accumulation of mitotic (pH3+) cells. However, when we examined mitosis in Banp morphants using live imaging, we found that mitosis duration is significantly prolonged because of chromosome segregation defects in Banp morphants, but that all 28 mitoses we examined eventually completed cytokinesis. Thus, we finally concluded that mitotic cells are not permanently arrested in M phase, but that mitosis is prolonged. To prevent confusion, we have changed “mitotic arrest” to “mitotic cell accumulation” or simply “mitotic defects” in the Results section on banp mutant phenotype analysis (shown in Figures 2 and 4).

4- It would be important to know whether the authors can rule out that S-phase defects cause the M phase defects, or vice versa. Could there be a primary defect, rather than multiple independent defects as the authors conclude?

We thank reviewer #2 for this suggestion. Interdependence between S phase defects and M phase defects is important to correctly interpret the data on cell-cycle regulation, especially cell-cycle checkpoint and DNA damage response. Indeed, there are interesting reports using in vitro cell culture systems indicating that replication stress induces mitotic death, through specific pathways (for example, Masamsetti et al., 2019, Nat. Comm. 10.4224. However, this topic is still challenging to dissect in vivo. In terms of our findings on Banp functions in zebrafish, we found that two chromosome segregation regulators, ncapg and cenpt, are direct transcription targets of Banp, and that it is likely that loss of Banp causes mitotic defects through downregulation of cenpt and ncapg. From this point, we conclude that mitotic defects are primary effects of the loss of Banp. The next question is how the loss of Banp stalls DNA replication forks and causes subsequent cell death. To address this question, we examined whether Banp direct targets include cell-cycle regulators, especially in S phase. We found that wrnip1 is an interesting candidate, because Wrnip1 reportedly protects stalled replication forks and promotes fork restart after DNA replication stress. In addition, Wrnip1 functions in interstrand crosslink repair (ICR). We found that the mRNA expression level of wrnip1 is markedly decreased in banp mutants, suggesting the possibility that DNA replication stress may be caused by reduction of wrnip1 expression in banp mutants. We present these data in new Figure 6-figure supplement 2. We have revised the possible role of Banp in cell-cycle regulation in new Figure7. Under this scenario, we consider it likely that loss of Banp may cause DNA replicationstress through downregulation of S phase regulators, independent of mitotic defects. However, we cannot exclude the possibility that DNA replication stress causes mitotic defects in banp mutants. Masamsetti et al., 2019, Nat. Comm. 10.4224. revealed that replication stress induces spindle assembly checkpoint (SAC)-dependent mitotic arrest and subsequent mitotic death when tp53 activity is inhibited. We showed that cell death in zebrafish banp mutant retinas was fully suppressed by tp53-MO at 48 hpf, but still occurred at 72 hpf, although there was no significant difference between wildtype and banp mutants (Figure 3GH). In the manuscript, we mentioned the possibility that some tp53-independent mechanism induces retinal apoptosis in banp mutants after 48 hpf. An alternative possibility is that most cell death in banp mutants depends on tp53; however, replication stress persisting in banp mutants injected with MO-tp53 may cause SAC-mediated mitotic death, as reported by Masamsetti et al., 2019. Future studies will be necessary to clarify this possibility.

Reviewer #3 (Public Review):

Babu and colleagues demonstrate that banp is expressed in the retina progenitor cells among other locations, and mutational loss of it results in increased mitosis, increased apoptosis, increased DNA damage, and the failure to differentiate photoreceptors. Importantly, these phenotypes are seen at a time period when retina progenitors undergo rapid cell cycles and differentiate into multiple cell types that make up the fully developed retina. Rescue with the wild type and phenocopy with another mutant allele provide strong support that the phenotypes results from loss of banp. Mutant animals show elevated p53 protein and reduction of p53 delays the onset of apoptosis by 24 hours. Mutant animals show altered transcriptional profile, with increased p53 expression and decreased expression of two genes that encode proteins needed for chromosome segregation. The authors propose that loss of banp results in defective DNA replication and DNA damage as well as mitotic chromosome segregation failures, all of which contribute to p53-dependent apoptosis to reduce cell number and cause developmental defects.

Banp is a very interesting protein. Also known as Scaffold/matrix attachment region binding protein 1, it is known to regulate the transcription of a number of genes including those important in oncogenesis. In vivo function of Banp, especially in the context of normal development, remains to be better understood. The current study fills this knowledge gap but I have some concerns about the interpretation of the data, the presentation and the potential impact. Specifically:

We are very pleased that reviewer #3 understood and appreciated the significance of our study.

Increased expression of atm and atr is observed and the authors suggest that replication stress and DNA damage activate the checkpoints to cause cell cycle arrest. There are several problems with this conclusion, which is depicted in Fig. 4G. Checkpoint activation occurs via phosphorylation changes in ATM/ATR and not through their transcriptional upregulation, which would take too long for a response that occurs within minutes.

We agree with the referee that upregulation of ATR/ATM mRNA expression may represent chronical activation of DNA replication stress and DNA damage response. In addition to ATR/ATM mRNA upregulation, RNA-seq analysis revealed that exo5 is one of the TOP15 upregulated genes in banp mutants (Fig. 3B). exo5 plays a critical role in ATR-dependent replication restart (Hambarde et al., 2021), suggesting that chronic replication stress occurs in banp mutants. We have mentioned exo5 upregulation in the Results section. As Referee 1 suggested, phosphorylation of H2AX is induced by ATR prior to DSBs, indicating that gammaH2AX is a marker of DNA replication stalling as well as of DSBs. We showed that gamma-H2AX+ cells are more numerous in banp mutants (Figure 4CF) and morphants (Figure 4-figure supplement 1AB) and in S phase banp mutant cells (Figure 4-figure supplement 1CDEFF’), suggesting that DNA replication stress and subsequent DNA damage linked to fork breakage are induced in banp mutants. We have revised the text by adding this statement in the Results section. In addition, we have revised Fig. 4G and its legend, in order to more clearly show the role of ATR and ATM in DNA replication fork repair and HR-mediated DNA repair in response to DSBs, and tp53-mediated regulation of cell survival and death.

ATM/ATR-dependent checkpoints arrest cells in G1 or G2 so you would expect reduced S and M phases. Yet, the authors saw increased M and no change in S.

It is puzzling that BrdU+ cell number does not change because if cells are indeed arrested in mitosis, they should be prevented from going into S phase and BrdU+ cell numbers should decrease.

There is no significant difference in the BrdU+ fraction of total retinal cells between wild-type and banp mutants at 48 hpf (Fig. 2-figure supplement 1AC), suggesting that cell-cycle arrest in S phase does not occur at significant levels in banp mutants at 48 hpf. At present, we have no good tool to detect G1 phase in zebrafish developing retina, because the Cdt1 fluorescent protein of the FUCCI zebrafish line cannot be stably driven in highly proliferating tissues such as zebrafish retina due to its very short G1 duration. Thus, we cannot determine whether G1 arrest occurs in banp mutant retina. However, we found that mRNA expression of p21 cdk inhibitor is upregulated in banp mutants, using bulk RNA-seq (Figure 3AB) and RT-PCR (Figure C), so it is still possible that banp mutant retinal cells are (probably partially) arrested in G1 phase. We have added this possibility to the Discussion. Further study is necessary to evaluate this point.

It is not addressed whether cenpt and ncapg expressed in the retina and whether are their expressions decreased in banp mutants. The RNAseq data is from whole animals.

RNA-seq data (Fig 3AB) were obtained from embryonic heads, but not whole bodies (see Materials and Methods). In accordance with this suggestion, to examine whether cenpt and ncapg mRNAs are expressed in retina, we performed in situ hybridization. We confirmed that these mRNAs are expressed in proliferative cells in zebrafish retina and have added these data to new Figure 5-figure supplement 1. In addition, we also confirmed that cenpt and ncapg mRNA expression is absent in banp mutants (see panels at 48 hpf in Fig. 5-figure supplement 1).

The rescue by banp-EGFP in Fig.1G is very nice. But it looks like there is partial rescue also with EGFP-banp(rw337) in the same panel. The defects the last panel do not seem as severe as in non inj controls. There are fewer pyknotic nuclei and the cell layers lack gaps. Quantification of the extent or reproducibility of the rescue is lacking.

We conducted acridine orange (AO) staining of retinas of wild-type, banp mutants, and banp mutants injected with banp(wt)EGFP and with EGFP-banp(rw337). We confirmed that banp(wt)EGFP significantly suppressed apoptosis in banp mutant retinas, whereas EGFP-banp(rw337) did not. We have added these data to new Figure 1-figure supplement 5. So, there is no partial rescue by EGFP-banp(rw337).

Some of the conclusions lack supporting data. For example, line 99: "Thus, Banp is required for integrity of DNA replication and DNA damage repair." There are no data for the integrity (meaning 'fidelity'?) of DNA replication and there are no DNA repair assays.

Thank are grateful for this suggestion. We understand that the term “integrity” could be too strong and changed it to “regulation.”

In another example, non-overlap of pH3 (M phase) and caspase+ cells is interpreted to mean that cells are dying in S phase (Figure 2 supplement 1). But the data are equally consistent with cells dying in G1 and G2.

In addition to non-overlap of the pH3+ and caspase+ areas along the apico-basal axis of the retina (Fig.2-figure supplement 1DG), we did not observe mitotic death in our live imaging of mitosis in banp morphant retinas. Considering the very short G2 phase of retinal cells in zebrafish, we conclude that apoptosis occurs mostly in retinal progenitor cells undergoing G1 or S phase, or differentiating neurons. However, we cannot exclude the possibility that apoptosis occurs in G2 phase. So, we have revised the text. Furthermore, caspase 3+ cells were mostly located in the intermediate zone of the neural retina along the apico-basal axis, whereas pH3+ cells were localized at the apical surface of the neural retina (Fig. 2-figure supplement 1G), suggesting that apoptosis occurs mostly in retinal progenitor cells during G1, S or G2 phase, or in differentiating neurons. Accordingly, we have revised Fig. 2-figure supplement 1L, to suggest that apoptosis may be induced in G1, S, or G2 phase.

The model in Figure 7 includes components without accompanying supportive data. For example, the arrow from Banp to DNA repair that indicates a direct role and the arrow from tp53 to delta113 tp53 that indicates direct activation.

Thank appreciate this suggestion. We have revised Figure 7 and its legend. In new Figure 7, we used solid arrows for regulatory pathways confirmed by us and previous other groups, and dotted arrows for proposed regulatory pathways. We already cited a reference (Chen et al., 2009), indicating direct activation of ∆113 tp53 by FL tp53.

The data that together support a single point are often split up among figures. For example, increased pH3+ cells shown in Fig. 2 and is interpreted as mitotic arrest. But it is equally possible that cells are undergoing extra divisions (and then dying). Support for mitotic arrest is provided by live imaging of mitosis, which is not presented until the last figure (Fig. 6). There are many such instances in the manuscript.

A similar concern was raised by reviewer #2. Please see our response.

Banp is already known for roles in p53-dependent transcription and in apoptosis (e.g. Sinha et al papers cited in the manuscript). Banp is also known to bind to the promoter regions of cenpt and ncapg (Grand et al and Mathai et al papers cited in the manuscript). These genes are known to be involved in mitosis in zebrafish (Hung et al and Seipold et al papers cited in the manuscript). In terms of what is new about banp function in this report, the requirement for banp in a critical phase of retina development and spontaneous induction of DNA damage come to mind. Unfortunately, how loss of banp leads to this defect remains to be addressed.

A related concern was raised by the editors and also by reviewer #2. Please see our responses. We found that wrnip1 mRNA expression is drastically reduced in banp mutants, which may cause DNA replication stalling and abnormal phenotypes.

Author Response:

Reviewer #3 (Public Review):

Two cell types in the parasubthalamic nucleus (a region of the posterior hypothalamus) are activated following food intake. The authors determine that the Tac1 expressing population is sufficient to suppress food intake and the Crh population does not influence food intake. Further, the authors demonstrate that only the Tac1 population projects to the PBN. The Tac1 neurons are transiently activated following food presentation or satiation hormones (for about 1 minute). This transient change in activity is interesting and fits into a lot of other recently published work showing transient neural activity changes that are involved in longer term behavior. Longer term activation of these neurons reduces food intake and the authors begin to explore the circuits/networks that these neurons influence. Overall, the work is well done and the experiments support the conclusions. Some minor clarifications could enhance the manuscript and could be addressed through further analysis or adding in text.

1. What % of the overall PSTN neurons are tac1/crh (ie, how many other cell types are there?). Or what % of the vglut2 neurons do they make. This just requires further analysis of the current dataset. And, are there any GABAergic cells (like are the PV GABAergic)?

We thank the Reviewer for suggesting this analysis because it is interesting and other readers are likely to ask the same questions. In our original submission we were hesitant to report these values because they ultimately represent an approximation. Because the neurons that surround the PSTN are also glutamatergic (including the subthalamic nucleus and the lateral hypothalamic area), it is impossible to precisely delineate the border of the PSTN using Slc17a6 as a marker. However, this is an important question and we feel that reporting these values while qualifying them as an estimation will be impactful. Therefore, in the revised manuscript, we now include the following statement:

“Although it is impossible to delineate a precise border for the PSTN using Slc17a6 because adjacent regions are also glutamatergic, we estimate that ~22% of Slc17a6- expressing neurons within the PSTN region do not express either Tac1 or Crh, indicating the presence of glutamatergic PSTN cell types that may express other unique genetic markers.”

We did not examine GABAergic expression in the PSTN because the Allen Brain Atlas and recent RNA-Seq studies (e.g., Wallén-Mackenzie et al., 2020) found an almost complete absence of Gad1- and Gad2-expressing cells in the PSTN region. We report this previous finding within the Results:

“Expression of the GABAergic markers Gad1 and Gad2 are notably absent from the PSTN region (Shah et al., 2022).”

2. The 60 second increase in tac1 neuron activity is interesting. In the discussion, the authors present some plausible arguments for how that may affect feeding for hours. Additionally, it would be nice to point out that this is a recurring theme. This occurs in other neuron populations that influence food intake. Although this is seemingly counterintuitive, I think it is good to mention as these short-term neural activity changes are clearly having large effects on behavior and it is important for everyone to realize this.

This point is an excellent observation and we agree that we could highlight other studies showing transient activation of neural activity controlling food intake. Therefore, we added to our Discussion:

“Indeed, many other neural populations that regulate food intake behavior also show a transient increase in neural activity on the timescale of seconds (Berrios et al, 2021; Luskin et al., 2021; Mohammad et al., 2021; Wu et al., 2022).”

3. Something a little strange with the meal frequency. I thought CCK reduced meal size not frequency. Why does the rescue then increase frequency? Could it be that the rescue to the CCK is by a different means than just blocking the effect of CCK? Adding some language to the discussion about how to interpret the satiation peptide data would be useful.

We thank the Reviewer for bringing up this interesting point. Previous studies do indicate that CCK (and also amylin, to a large extent) reduces meal size and does not have much of an effect on meal frequency. We therefore added a paragraph to the Discussion to note and discuss this point:

“It is also noteworthy that chemogenetic inhibition of PSTN^Tac1 neurons attenuates the effects of amylin, CCK, and PYY by decreasing the frequency of meals as opposed to meal size or meal duration (Figure 5). Previous studies of these anorexigenic hormones, especially amylin and CCK, indicate that they affect food intake primarily by decreasing meal size as opposed to meal frequency (Drazen and Woods, 2003; Lutz et al., 1995; West et al., 1987). Therefore, inhibition of PSTN^Tac1 neurons might attenuate the effects of these hormones indirectly, perhaps by reducing activity in downstream populations such as the NTS or PBN. In this model, infusion of anorexigenic hormones activate PSTN^Tac1 neurons that, in turn, cause sustained activation of downstream populations. Without this sustained activity, downstream populations may not have sufficient activity to cause a reduction in the intermeal interval, leading to increased bouts of feeding. The mechanism by which anorexigenic hormones activate PSTN^Tac1 neurons, as well as how decreases in PSTN^Tac1 neuronal activity affect downstream populations, are important topics for future investigation.”

4. The axonal stimulation data needs qualification - as axons could project to multiple target regions (like the projections to the PVT could also have a collateral to the CEA). For this type of experiment, I prefer to use the phrase "neurons with a projection to region X do behavior Y". Otherwise, the implication in reading the results is that the particular projection is mediating the behavior. Also, the collateral issue, which is qualified in the discussion, should be mentioned here.

We see the Reviewer’s point and have revised the language to highlight this important qualification of our results. Specifically, we added text in the Results section in regard to Figure 8:

“Because it is unknown whether PSTNneurons send collateral projections to multiple brain regions, it is possible that stimulation in a single projection target causes antidromic activation to one or more other target areas. Therefore, these results indicate that PSTNTac1 neurons with projections to the CeA, PVT, PBN, and NTS can suppress food intake, although the exact functional role of each downstream target region on food intake behavior remains undetermined.”

Author Response:

Reviewer #2 (Public Review):

In their supplementary section A.3-1.5 the authors perform QTL simulations to assess the performance of their analysis methods. Of particular interest is the performance of their cross-validated stepwise forward search methodology, which was used to identify all the QTL. However, a major limitation of their simulations was their choice of genetic architectures. In their simulations, all variants have a mean effect of 1% and a random sign. They also simulated 15, 50, or 150 QTL, which spans a range of sparse architectures, but not highly polygenic ones. It was unclear how the results would change as a function of different trait heritability. The simulations should explore a wider range of genetic architectures, with effect sizes sampled from normal or exponential distributions, as is more commonly done in the field.

As suggested, we have expanded the range of simulations we explore in the revised manuscript. We note that the original simulations discussed in the manuscript involve exponentially distributed effect sizes (with a mean of 1% and random sign) at multiple different heritability values. These are described in Figures A3-4 and A3-5. We also simulated epistatic terms (Figure A3-3.3). In the revision, we have broadened the simulations to add more ‘highly polygenic’ architectures (1000 QTL). We find that the algorithm still performs well, though worse than when 150 QTL are simulated. The forward search behaves in a fairly intuitive way: QTLs get added when the contribution of a true QTL to the explained phenotypic variance overcomes the model bias and variance. QTLs are only missed if their effect size is too low to contribute significantly to phenotypic variance, or if they are in strong linkage and thus their independent discovery barely increases the variance explained (which is all finally controlled by the trait heritability). At much higher polygenicity, composite QTL can be detected as a single QTL when their sum contribute to phenotypic variance, and get broken up if and only if independent sums also contribute significantly to phenotypic variance. Of course, there are many ways to break up composite QTL, but the algorithm proceeds in a greedy fashion focusing on unexplained variance. We have also explored cases with multiple QTL of the same effect, and with different mean effects or different number of epistatic terms, but we found these results were largely redundant. To summarize these conclusions, we have added the following discussion at the end of the results section: “The behavior of this approach is simple and intuitive: the algorithm greedily adds QTL if their expected contribution to the total phenotypic variance exceeds the bias and increasing variance of the forward search procedure, which is greatly reduced at large sample size. Thus, it may fail to identify very small effect size variants and may fail to break up composite QTL in extremely strong linkage.”

We have also added additional clarification in the Appendix: “These results allow us to gain some intuition for how our cross-validated forward search operates. […] However, while our panel of spores is very large, it remains underpowered in several cases: 1) when QTL have very low effect size, therefore not contributing significantly to the phenotypic variance, and 2) when composite QTL are in strong linkage and few spores have recombination between the QTL, then the individual identification of QTL only contributes marginally to the explained variance and the forward search may also miss them.”

In this simulation section, the authors show that the lasso model overestimates the number of causal variants by a factor of 2-10, and that the model underestimates the number of QTL except in the case of a very sparse genetic architecture of 15 QTL and heritability > 0.8. This indicates that the experimental study is underpowered if there are >50 causal variants, and that the detected QTL do not necessarily correspond to real underlying genetic effects, as revealed by the model similarity scores shown in A3-4. This limitation should be factored into the discussion of the ability of the study to break up "composite" QTL, and more generally, detect QTL of small effect.

We agree with some aspects of this comment, but the details are a bit subtle. First, we note that the definition of underpowered depends on the specifics of the QTL assumed in the simulation. In addition, many of the simulations were performed at 10,000 segregants, not at 100,000, with no effort to enforce a minimum effect size, or minimum distance between QTL. For example, if 100 QTL are all evenly spaced (in recombination space) and all have the same effect such that they all contribute the same to the phenotypic variance, then the algorithm is in principle maximally powered to detect these. This is why our algorithm is capable of finding >100 QTL per environment. On the other hand, just 2 QTL in complete linkage cannot be distinguished and no panel size will be able to detect these.

However, we do agree with the general need to discuss the limitations in more detail and have clarified these concerns in the ‘Polygenicity’ result section. We have also reiterated the limitations of the LASSO approach within the simulation section. The motivation for an L0 normalization in this data was first discussed in the section A3-1.3: “Unfortunately, a harsh condition for model consistency is the lack of strong collinearity between true and spurious predictors (Zhao & Yu, 2006). This is always violated in QTL mapping studies if recombination frequencies between nearby SNPs are low. In these cases, the LASSO will almost always choose multiple correlated predictors and distribute the true QTL effect amongst them.”

In section A3-2.3, the authors develop a model similarity score presented in A3-4 for the simulations. The measure is similar to R^2 in that it ranges from 0 to 1, but beyond that it is not clear how to interpret what constitutes a "good" score. The authors should provide some guidance on interpreting this novel metric. It might also be helpful to see the causal and lead QTLs SNPs compared directly on chromosome plots.

We agree that this was unclear, and have added additional discussion in the main text describing how to interpret the model similarity score. Essentially, the score is a Pearson’s correlation coefficient on the model coefficient (as defined in section A3-2.3, after equation A3-28). However, given a single QTL that spans two SNPs in close linkage, a pure Pearson’s correlation coefficient would have high variance, as subtle noise in the data could lead to one SNP being called the lead SNP vs the other, and two models that call the same QTL might have either 100% correlation, or 0% correlation. Instead, our model similarity score ‘aligns’ these predicted QTL before obtaining the correlation coefficient. The degree at which QTL are aligned are based on penalties with respect to collinearity (or linkage) between the SNPs, and the maximum possible score is obtained by dynamic programming. Similar to sequence alignments between two completely unrelated sequences, a score of 0 is unlikely to occur on sufficiently large models as at least a few QTL can usually be paired (erroneously). We have also added a mention in the main text referring to Figures A3-3, A3-7, A3-8, A3-9, which show the causal and lead QTL SNP directly on the chromosome plots.

The authors performed validation experiments for 6 individual SNPs and 9 pairs of RM SNPs engineered onto the BY background. It was promising that the experiments showed a positive correlation between the predicted and measured fitness effects; however, the authors did not perform power calculations, which makes it hard to evaluate the success of each individual experiment. The main text also does not make clear why these SNPS were chosen over others-was this done according to their effect sizes, or was other prior information incorporated in the choice to validate these particular variants? The authors chose to focus mostly on epistatic interactions in the validation experiments, but given their limited power to detect such interactions, it would probably be more informative to perform validation for a larger number of individual SNPs in order to test the ability of the study to detect causal variants across a range of effect sizes. The authors should perform some power calculations for their validation experiments, and describe in detail the process they employed to select these particular SNPs for validation.

We agree with the thrust of the comment, but some of the suggestions are impossible to implement because of practical constraints on the experimental methods (and to a lesser extent on the model inference). First, we chose the SNPs to reconstruct based on three main factors: (a) to ensure that we are validating the right locus, the model must have a confident prediction that that specific SNP is causal, (b) the predicted effect must be large enough in at least one environment that we would expect to reliably measure it given the detection limits of our experimental fitness measurements, and (c) the SNP must be in a location that is amenable to CRISPR-Cas9 or Delitto Perfetto reconstruction. In practice, this means that it is impossible to validate SNPs across a wide range of effect sizes, as smaller-effect SNPs have wider confidence intervals around the lead SNP (violating condition a) and have effects that are harder to measure experimentally (violating condition b). In addition, because the cloning constraints mentioned in (c) require experimental testing for each SNP we analyze, it is much easier to construct combinations of a smaller set of SNPs than a larger set of individual SNPs. Together, these considerations motivated our choice of specific SNPs and of the overall structure of the validation experiments (6 individual and 9 pairs, rather than a broader set of individual SNPs).

In the revised manuscript, we have added a more detailed discussion of these motivations for selecting particular SNPs for validation, and mention the inherent limitations imposed by the practical constraints involved. We have also added a description of the power and resolution of the experimental fitness measurements of the reconstructed genotypes (we can detect approximately ~0.5% fitness differences in most conditions). We are unsure if there are any other types of power calculations the reviewer is referring to, but we are only attempting to note an overall positive correlation between predicted and measured effects, not making any claims about the success of any individual validation (these can fail for a variety of reasons including experimental artifacts with reconstructions, model errors in identifying the correct causal SNP, unresolved higher-order epistasis, and noise in our fitness measurements, among others).

In section A3-1.4, the authors describe their fine-mapping methodology, but as presented is difficult to understand. Was the fine-mapping performed using a model that includes all the other QTL effects, or was the range of the credible set only constrained to fall between the lead SNPs of the nearest QTL or the ends of the chromosome, whichever is closest to the QTL under investigation? The methodology presented on its face looks similar to the approximate Bayes credible interval described in Manichaikul et al. (PMID: 16783000). The authors should cite the relevant literature, and expand this section so that it is easier to understand exactly what was done.

We have attempted to clarify section A3-1.4. As the reviewer correctly points out, the fine mapping for a QTL is performed by scanning an interval between neighboring detected QTL (on either side) and using a model that includes all other QTL. For example, if a detected QTL is a SNP found in a closed interval of 12 SNPs produced by its two neighboring QTL, 10 independent likelihoods are obtained (re-optimizing all effect sizes for each), and a posterior probability is obtained for each of the ten possible positions. We have cited the recommended paper, as our approach is indeed based on an approximate Bayes credible interval similar to the one described in that study (using all SNPs instead of markers). We have added the following sentence to the A3-1.4 section at the end of the second paragraph (similar to the analogous paragraph in Manichaikul et al): “[…] as above by obtaining the maximum likelihood of the data given that a single QTL is found at each possible SNP position between its neighboring QTL and given all detected other QTL (thus obtaining a likelihood profile for the considered positions of the QTL). We then used a uniform prior on the location of the QTL to derive a posterior distribution, from which one can derive an interval that exceeds 0.95.” Some typos referring to a ‘confidence’ interval were also changed to ‘credible’ interval.

The text explicitly describes an issue with the HMM employed for genotyping: "we find that the genotyping is accurate, with detectable error only very near recombination breakpoints". The genotypes near recombination breakpoints are precisely what is used to localize and fine-map QTL, and it is therefore important to discuss in the text whether the authors think this source of error impacts their results.

This is a good point, we have added a reference in the main text to the Appendix section (A1-1.4) that has an extensive discussion and analysis of the effect of recombination breakpoint uncertainties on finemapping.

The use of a count-based HMM to infer genotypes has been previously described in the literature (PMID: 29487138), and this should be included in the references.

We now also add this citation to our text on the count-based HMM.

Author Response:

Reviewer #1 (Public Review):

Major Comments

I am concerned that a lot of these studies had relatively low n numbers (n=5 in some cases) and that some of the studies may have been underpowered. Given the variability with in vivo studies, some endpoints may have been significant with more numbers. Along these lines, what is the justification for using the (parametric) ANOVA test. I'm not a statistician but I thought that the rule of thumb was that non-parametric tests should be used if n<12 since you cannot verify that the data is normally distributed. In this case, I would recommend having a statistician look at it and/or increasing some of the N's, or using the non-parametric Kruskal-Wallis test. Indeed, in some cases, the variation the variation is quite large (ie Fig 6, 7). Whilst I do not think that the low N's change the ultimate conclusions, but more rigor (ie more N's) would help solidify the paper given that it will likely be of great interest and scrutinized by the scientific community.

We conducted power analyses prior to the start of the studies to identify the number of animals per group to use, based on our past studies of inflammatory changes induced by inhalants, infections and asthma. We set the target number of mice (n) at that time, such that these studies would be powered to detect a 25% change in cytokine expression. We did go through and reviewed all of the data with our biostatisticians, we came to the conclusion that it would not be statistically appropriate to run more mice to increase the n when our primary outcome remains the same. We double-checked that the ANOVAs with corrections for multiple comparisons were correct for each particular experiment. Discussion with our statistician confirmed that ANOVA is correct as long as the data passed normality testing, which was done. An additional point, and most relevant to this specific recommendation, JUUL Mint and JUUL Mango flavors are no longer on the market, such that extensive further studies are not feasible. While these two flavors are not available anymore, they were composed of an array of chemicals commonly found in other flavors (but in different combinations), such that we believe that these data are most likely relevant to other vapes. In particular, JUUL Mint shares chemical features with JUUL Menthol, which took its place as one of the most popular JUUL flavors. The discontinuation of these flavors has been added as a limitation within the Discussion

Fig S3. For the lung histology, please quantify the mean linear intercept per ATS guidelines and show representative BAL images.

We have conducted the mean linear intercept (MLI) measurements on e-cigarette aerosol exposed lungs and controls per ATS guidelines and have added these data to the manuscript (new Appendix 1- Figure 4M). We paired these data with the original histology images (Appendix 1 – Figure 4A-4L). We have added appropriate methods (pages 21-22) and results (page 9) as well. Of note, the MLI data matches our original physiologic assessments of lung function (Appendix 1 – Figure 2A-2J), including elastance and compliance, which are known to change in the setting of emphysema. MLI, lung elastance and compliance were no different across inhalant groups and controls. Further, we have taken representative images of Giemsa Wright stained BAL samples, and have added these to the manuscript (new Appendix 1 Figure - 3E-3J and 3O-3T) paired with BAL cell count data.

One of the most novel conclusions from this paper is increased inflammation in the brain which the authors speculate could lead to altered moods and or change the addiction threshold. I would tend to agree with this conclusion, but could the authors perform additional mouse psychological tests to confirm this? Also, were there observable physiological responses in the vaped mice that could be reported which may correlate this conclusion, ie changes in grooming, fur ruffling or other behavioral changes?

We are thrilled that the Reviewer is as interested in these implications as we are, because we believe the neuroinflammation detected is quite frightening, particularly because it is likely to impact both behavior and mood. We have added further discussion regarding the potential consequences of inflammation in each of the organs (pages 13-19), with an emphasis on the effects of neuroinflammation on behavior and psychology. We have subdivided the Discussion section to highlight potential effects on each distinct organ.

While we are not a behavioral lab, and thus running behavioral studies in mice is beyond the scope of both our lab and this manuscript, we agree that the neuroinflammation is of great interest and further studies are needed to best assess potential psychological and behavioral changes. Of note, we did not observe any overt behavioral changes - we closely observe the mice both during and after exposures and make notes regarding grouping, fur, and activity level - none of which were changed by the different vaping exposures. We have added the lack of dedicated behavioral and psychological evaluations as a limitation of this work and as an opportunity for discovery in future studies (page 19- 20).

Minor comments Change title to state "in mouse". That this study was performed in rodents should be apparent from the outset.

Actually, our original title does contain “in mice” at the end. Apologies if these words were cut off on your end. We do agree that the title should be apparent that the study was conducted in mice. We wanted to make the title even clearer, so replaced the brand name JUUL with the type of e-device. The title is as follow: “Effects of Mango and Mint pod-based e-cigarette aerosol inhalation on inflammatory states of the brain, lung, heart and colon in mice”

No changes in collagen deposition were detected using basic histology. Have the reviewers considered performing immunohistochemistry and staining for alpha-smooth muscle actin which may be a more sensitive assay?

We agree with the reviewer that there are more sensitive tools that can be used. We believe that, in our system, and at 3 months of exposure, JUUL Mint and Mango are not very likely to induce fibrosis, since our data of inflammatory markers and fibrosis associated genes (in homeostatic conditions, Figure 3) show that there are not significant differences, and in some markers, JUUL Mint and Mango exposed mouse lungs are even showing less inflammation than Air controls. In addition, we also showed no differences were obtain in physiological assessment (heart rate, heart rate variability or blood pressure, Appendix 1 – Figure 1). Thus, we do not expect to find significant differences even with additional assays. We are planning on challenging mice with bleomycin in the future, as it may be possible to detect differences in fibrosis in the setting of this pro-fibrotic challenge.

"Thus long term exposure to Juul does not lead to significant changes...". I would argue that 1-3 months is not long term. Indeed, other researchers have performed 6-12 month ecigarette exposures and it takes a lifetime in humans to develop lung disease after smoking. Since you can detect pro-inflammatory changes but no altered physiology, it may be that alterations in airway physiology are only just beginning.... The authors should modify this sentence and maybe not call their studies "long term".

We agree with the reviewer and have modified the sentence as follows for a more accurate interpretation of our results (page 9): “Thus, 1 and 3 month exposure to JUUL Mint and Mango aerosols may not cause significant changes in airway physiology, but this does not preclude the possibility that changes may occur with longer exposures, such as 6-12 months.” We have also gone through the entire the manuscript to focus on describing our exposure in terms of months instead of the descriptive terms acute / sub-acute / chronic, and we have removed the word chronic from the title.

"Differences in LPS induced cytokine levels were no longer observed after 3 month JUUL exposure versus Air control groups". As per the major comments, this might be a power issue - there is certainly a trend for some cytokines.

It has been seen in prior studies that chronic inhalant use (including and most notably cigarette smoke) can lead to proinflammatory changes in the first days to weeks, but opposite effects thereafter. For example, cigarette smoke inhalation leads to inflammatory changes at 4 weeks that resolve by 12 weeks. Thus, we feel that some of the cytokine findings are not unusual or surprising versus other patterns of inhalant use. However, we agree with the reviewer that IL-1b in cardiac tissue trends in the same direction at 3 months in both JUUL Mint and JUUL Mango exposed mice (Figure 8C and 8D). As per one reviewers’ comments, we combined 1 and 3 month data for merged graphs (Appendix 1 – Figure 4) and when analyzed together (data passed normality testing) further differences at 3 months were identified (see IL-1b in Appendix 1 – Figure 4 panel 4B). We have included these additional figures for each dataset in the Appendix 1 files.

Of note, because some JUUL flavors are no longer on the market, including JUUL Mint and JUUL Mango, we are unable to run additional studies with these flavors. We are running new studies of the impact of JUUL Tobacco and JUUL Menthol, the two remaining JUUL flavors on the market. However, these studies will take an additional 1- 2 years and thus are beyond the scope of this manuscript. We have expanded the limitation section within the discussion with regards to power, in order to clarify to the readers that some findings are limited by the number of subjects.

Reviewer #2 (Public Review):

Under homeostasis conditions, the authors observed sign of inflammatory responses in the brain, the heart and the colon, while no inflammation was detected in the broncho-alveolar lavage fluid of the mice following exposures to JUUL aerosols. Also, JUUL aerosol exposures mediated airway inflammatory responses in the acute lung injury model (LPS). Further, this infection affected the inflammatory responses in the cardiac tissue. Most of the biological adverse effects induced by JUUL aerosols were flavor-specific.

Strengths include evaluating inflammation in multiple organs, as well as assessing the physiological responses in the lungs (lung function) and cardiovascular system (heart rate, blood pressure), following exposures to JUUL aerosols. Weaknesses include the fact that only female mice were used in this study. Further, the daily exposures to either air or to the JUUL aerosols lasted only 20 min per day. It is unclear how a 20-min exposure is representative of human vaping product use. Also, although daily exposures were conducted for a duration of both 1 and 3 months, time-course effects associated with JUUL aerosols are barely addressed.

We would like to thank the reviewer for their positive comments on our manuscript. We apologize for our error; in reality we exposed mice for 20 minutes three times daily, so one hour in total per day. We have corrected this error within our Methods. We designed the exposures this way to better mimic human e-cigarette use throughout the day (instead in one intense vaping session per day, which is not the norm). We agree that there is a limitation in using only female mice in the study in case that there are sex-dependent effects, which is definitely an interesting question. We typically start with one sex of mice and then run repeat experiments with the other sex. Unfortunately, this study faced problems beyond our control that prevented us from performing further experiments. In late 2019 the FDA was moving to ban specific flavors for pod devices, which include those for Mint and Mango. In anticipation of the new regulations, JUUL ultimately decided to discontinue JUUL Mint and Mango, and soon they were out of the market. The same process occurred with the other popular JUUL flavors such as Crème Brûlée and Cucumber. We have expanded the limitation section within the Discussion, and have pointed out that because these studies were conducted in female mice alone, the results may not represent effects in males.

Although there are a few limitations related to this study, which should be included in the manuscript, overall, the authors' claims and conclusions are based on the data that is presented through multiple figures.

We appreciate the Reviewers comments and have added limitations about the study size, power, lack of male subjects, etc. to the discussion section.

Reviewer #3 (Public Review):

Weaknesses

1. The authors observed neuroinflammation in brain regions responsible for behavior modification, drug reward and formation of anxious or depressive behaviors after exposure to JUUL. The importance of the neuroinflammation is still unclear. It would help demonstrate the pathogenic role of the neuroinflammation by testing animal behaviors. Similar issue for other organ inflammation.

We are an immunology, inflammation, and lung physiology lab, thus, behavioral studies are beyond the scope of both our lab and this manuscript. However, we agree that the neuroinflammation is of great interest and is highly likely to impact behavior and mood. Further studies are needed to best assess potential psychological and behavioral changes. We believe this work is important to share such that dedicated behavioral science labs can undertake these important studies. We have added these important limitations to the discussion.

1. Majority of the data are inflammatory cytokine mRNA expression. Other methods would be needed to confirm their expression.

Of note, in the original submission, we included protein quantification data for both the brain and the lung. We have taken the reviewers comments to heart and have conducted protein-level assays on the cardiac tissues as well, yielding additional data (new Figure 4) that has been added to the methods, results, figures and discussion. Unfortunately, we do not have any additional colonic tissue for protein-level assessments, as all of the tissue was used for the gene transcription and histologic studies. But to take a step back, these studies were originally intended to examine the broad reaching impact of e-cigarette aerosols across the body. This work, and thus this manuscript, was designed to highlight changes at the gene expression level, to demonstrate that e-cigarette use is not benign and does have broad-reaching effects on gene expression. We agree that more work is needed to fully define the impact of e-cigarette use at the protein, cellular, and organ level, but the majority of that work is beyond the scope of this manuscript. To bring the focus back to gene expression, we have conducted RNAseq on the lungs of JUUL exposed mice, and have included those data herein to highlight the effects of ecigarette aerosols on gene expression in the lung, with a particular focus on differences between Mint and Mango flavors (the most popular JUUL flavors at the time of this study). These new data (new Figure 6) support the hypothesis that e-cigarette aerosol inhalation fundamentally alters the lung, which raises the specter of downstream health effects.

1. The author seemed to assume the difference between JUUL Mango and JUUL Mint is flavor and then came up with the conclusion regarding flavor-dependent changes in several inflammatory responses. Evidence is needed to approve the assumption.

Although the formulation of JUUL e-liquids is proprietary, their website claims simplicity (https://www.juul.com/learn/pods) in that they use pharmaceutical grade propylene glycol and glycerol (which makes up the majority of their e-liquids), in order to form an aerosol which carries pharmaceutical grade nicotine and benzoic acid (when combined, create a nicotine salt), and flavors (which can be a mixture of natural and artificial ingredients). Thus, according to their website the only difference among the different JUUL pods would be the flavoring components. Hence, we concluded that differences observed in our study between Mint vs Mango should be most likely due to flavor-dependent effects, since base components should be the same. To support this flavor-dependent effect, a study from Omaiye et al in 2019 (PMID: 30896936) showed the variety of different flavoring chemical in all JUUL flavors and how the different JUUL vapors induce different level of cytotoxicity in BEAS-2B cells in vitro based their flavor. We have added relevant discussion to the manuscript.

1. In most cases, the change of inflammatory cytokines is mild ~2 fold. The author should demonstrate how these marginal changes could affect pathophysiology.

We agree with the reviewer that the majority of changes in cytokines were relatively small. However, the fact that multiple cytokines are changing in concert indicates a significant shift in immunophenotyping across organs. We are most concerned about how these shifts in the inflammatory state will alter an e-cigarette vapers response to common clinical challenges. In Dr. Kheradmand’s recent work, mice exposed to e-cigarette aerosols with and without nicotine were much more susceptible to acute lung injury in the setting of viral pneumonia. In our work, we utilized the LPS model of acute lung injury to take a first look at the potential impact of JUUL inhalation in particular on susceptibility to lung inflammation. Further work is needed to truly define how the subtle, broad shifts in the cytokine milieu across organs will impact the health of e-cigarette vapers. We have added relevant discussion to the manuscript.

1. To fully evaluate the health impact of evolving cigarette, it would be informative to included other tobacco or vaping device as control.

We agree that such comparisons are likely to provide insight into the differences between devices and formulations and versus cigarette smoke, and thus will be incredibly important for the field. However, these comparisons were beyond the scope of this study, whose main goal was to assess the inflammatory and physiological aspects of JUUL in particular. We believe this to be important because JUUL e-cigarettes are the most popular of all e-cigarette devices, and many young users do not use other e-devices or conventional tobacco. Thus, our primary objective of this work was to specifically assess the safety or risk of this device in particular (versus not using any inhalant at all). However, because we have run parallel studies in the past with vape pens, box mods, and conventional tobacco, we are hopeful to start combining data to look for trends and differences across inhalant exposures. For example, we recently published our work on differences in metabolites in the circulation of mice exposed to a wide variety of ecigarette based inhalants (Moshensky et al. Vaping induced metabolomic signatures in the circulation of mice are driven by device type, eliquid, exposure duration and sex. ERJ Open. July 2021 PMID: 34262972). This study is one of the few studies that have employed animal models to test JUUL devices and the only one assessing their effects in different organs, and although we agree that comparisons with other devices is important, it was not the goal of this study.

1. The longest exposure in the study is 3 months. It is not convicting to come up with conclusions regarding chronic exposure. Some organ showing no difference may be due to the timing.

We have altered the wording throughout the manuscript to clarify that the 3-month duration is equivalent to 10 to 20 years of inhalant use versus 40 to 50 years for a 6 to 12 month model. We have also removed many instances of the descriptive terms acute, sub-acute and chronic across the manuscript, as focused on using the absolute duration of exposure instead, to avoid accidental extrapolation to longer exposures. Because we utilized cellular and molecular based assays, we were not relying on identifying organ level pathology such as fibrosis, emphysema, and organ dysfunction, all of which would require longer exposures.

I think this website is a great way for educators to communicate and share their work and ideas. However, it is important to note that all information may not be as reliable as we expect!

research continues to show that echo chambers are not as prevalent or important as we may think

• reminds me of how facebook is known for this (Zucked)

I think Cognitive Flexibility is a very interesting concept which requires personal effort as it encourages us to change our mentality regarding something which is important because we need to have the ability to think differently.It helps us to think of new ideas.This skill is very useful in academic and work environments as it allows us to think with keeping in mind another person's point of view.

Author Response:

Reviewer #1

1: “A major weakness was that the simulation algorithm was both highly complex, but insufficiently explained. As a consequence, it was not clear what the underlying assumptions of the simulations were and how these assumptions were based on and/or constrained by the experiments.”

We have revised the section related to the simulation algorithm. This reviewer also raised a similar issue and suggested adding pseudocode or explaining it in plain language. We have therefore included two sections, “Cell-fate simulation algorithm” and “Cell-fate simulation options with Operation data”, as well as Figure 7, Figure 8 and Supplementary Figure 9.

In our previous version of the manuscript, we named the data used for the simulation as “Source data”. However, we realize that this journal uses this term for other purposes. We have therefore changed “Source data” to “Operation data” to avoid confusion.

1. “The single-cell analysis, including measuring lineages, by itself is not cutting-edge and has been done before, and so the novelty should be in the analysis.”

We agree that single-cell tracking per se is not a new technology, and was carried out as early as 1989 using 16 mm film. However, it has not been used frequently in the field of cell biology because of its extremely laborious nature. Our focus was thus on the development of a single-cell tracking technique that could be used routinely in cell biological research. We therefore computerized the analysis (preprint, BioRxiv 508705; doi: https://doi.org/10.1101/508705 (2018)) to allow the generation of large amounts of single-cell tracking data for bioinformatics analysis. We have mentioned this in the Results (“System to investigate the functional implications of maintaining low levels of p53 in unstressed cells”).

1. “However, in many cases, the resulting data is presented in a manner that does not rely on the single-cell tracking (e.g. total cell number vs time in Fig. 2, average frequency of events in Fig. 4).”

We realize that we did not adequately explain the data relating to Figure 2. Counting experiments were performed to validate the results of single-cell tracking data, because such verification has not previously been performed. We therefore intended to produce a figure including both the actual counting data and single-cell tracking data together, to allow the readers to compare the results obtained by the different approaches. Although this reviewer commented that some data did “not rely on the single-cell tracking”, we would like to stress that the counting data were only used for the purpose of comparison. We have thus rewritten the “Effect of silencing the low levels of p53 on cell population expansion” in the Results, to clarify this.

1. “The impact of p53 was only assessed on level of differences between experimental conditions (p53 siRNA or not), but p53 levels themselves were not measured and therefore not incorporated in the single-cell analysis.”

To the best of our knowledge, there are currently no techniques that allow the expression levels of proteins or genes of interest to be determined in individual live cells that are being tracked, and which could thus be used to generate data for bioinformatics analysis. It may be possible to use cells expressing a fluorescence-tagged protein, but as noted by this reviewer, frequent excitement of fluorophores in cells could affect cell growth (phototoxicity). We have thus been searching for a suitable technique that could be combined with single-cell tracking since 2012. If it becomes possible to perform an experiment similar to that suggested by this reviewer, it could potentially reveal many unknown cellular characteristics. We have revised the Discussion to consider this matter.

1. “In general, differences between wild-type and p53 siRNA data were small, while cell-to-cell variability in p53 knock-down appears high (as judged by Supplementary Fig. 4). This leaves open whether the relatively minor difference between wild-type and p53 siRNA cells reflects variability in p53 knockdown between cells, which is currently not directly assessed.”

With regard to the “differences between wild-type and p53 siRNA data were small”, we would like to make a comment related to the small difference. In a typical study of p53, a lethal dose of an agent that could kill a majority of growing cells within e.g. 24-48 hrs has been used to detect a difference with control cells. A reason to use the lethal dose of agents is to make the status of cells homogeneous to detect any alteration of interest using average-based techniques, which represent the alteration that occurred in a majority of cells. On the other hand, when lower doses of agents are used, cell-to-cell heterogeneity has to be talking into account, as only a certain group of cells in a cell population may respond to the agents. In this case, only a small or no difference may be able to detect by the average-based analyses, if only a small number of cells in a cell population respond. Distance from the average-based analysis, single-cell tracking is a technique that allows quantitative analysis of alteration that occurred in individual cells in a cell population. By Western blotting, which is an average-based assay, (Supplementary Fig. 4), the level of p53 in unstressed cells was reduced to 30%. As the levels of p53 in unstressed cells are already low, a 70% reduction of the amount of p53 may be considered to be small. However, at the individual cell levels, it was sufficient to increase cell death, multipolar cell division, and cell fusion (Fig. 4). Thus, analysis of cells at the single-cell level could allow obtaining information that is difficult to find by the average-based analysis.

The comment related to “reflects variability”, however, made an important point. It is currently technically difficult to determine the expression levels of p53 or other proteins in individual live cells that are being tracked by long-term live-cell imaging. We therefore assumed that silencing reduced the levels of p53 in all the tracked cells. However, it is reasonable to expect variations in the silencing levels of p53 among individual cells, and it may be possible that cells in which p53 levels were reduced, e.g. to 0%, underwent cell death, while cells in which expression was only reduced to 50% underwent cell fusion, etc. Information on the levels of silencing in each cell would allow us to evaluate the relationship between p53 levels and the type of induced events. However, this analysis is currently technically difficult, as explained above. Nevertheless, the fact that silencing induced changes in cell fate suggested that the low background levels of p53 may have some functions. We have revised “Silencing of p53 and single-cell tracking” in the Results.

Reviewer #2

“The study's main weakness is the lack of empirical evidence from the simulation predictions of biology, and that the cellular consequences of p53 function were predictable and mostly confirmatory.”

We appreciate these interesting comments regarding the similarities and differences of the empirical and simulation approaches. In empirical studies, a model or hypothesis is often based on the results of an analysis that aims to reveal characteristics of interest e.g. of cells. However, such a model or hypothesis generally needs to be confirmed or tested independently. We therefore considered simulation as a tool to build a model or hypothesis, which also needed to be confirmed or tested.

Simulation could thus be considered as an additional tool, e.g. in addition to western blotting and DNA sequencing, which could generate different types of data than other existing techniques. We therefore think that such simulations could provide new options for cell biological studies. Regarding its “confirmatory” use, we think that simulation can be used to confirm existing models, but may also be used as a discovery tool. For example, p53-knockout cells are known to produce tetraploid cells, but how such cells are formed remains unclear. Single-cell tracking analysis can be used to fill the gap between the loss of p53 and tetraploid cell formation, and simulation can then be used to simulate the fate of cells generated by this loss.

Although we focused on describing our approach using single-cell tracking and cell-fate simulation in our manuscript, we believe these methods could be used in combination with empirical studies, to widen the cell biological research options.

We have discussed these issues in “Cell fate simulation and its applications” in the Discussion.

Reviewer #3

"Yet it is unclear how these results can be generalized because the authors only studied one cell line."

The current work focused on addressing a biological question using single-cell tracking and cellfate simulation; however, it will also be interesting to see if the proposed models can be generalized. Given that HeLa cells, in which p53 function is neutralized by papillomavirus E6 protein, also frequently undergo cell fusion followed by multipolar cell division and cell death (Sato, Rancourt, Sato and Satoh Sci Rep (2916) 6:23328), we believe that the low levels of p53 may also play a similar role in suppressing those events in many other types of cells.

"The results are not compared to other cell lines or primary cells, in terms of baseline expression of p53. "

We agree that it will be interesting to apply the methods in various types of cells and primary cell lines. However, there are significant variations in growth profiles among cell types. We have created live-cell imaging videos for > 30 cell lines, and found that each cell type showed unique characteristics in terms of growth patterns, frequencies of cell death, cell fusion, and multipolar cell division, and in the degree of cell-to-cell heterogeneity, implying that each cell type must be characterized using single-cell tracking analysis before moving on to studies using those cells, given that no such data are currently available. We believe that establishing a public data archive of single-cell tracking data will be useful for cell biological research, as well as for testing the current model.

"In addition, it is unclear how this model is superior to testing homeostatic p53 compares to models that use mutated p53.”

Most cancer cells carrying p53 gene mutations still express mutant p53 in the cytoplasm, and mutant p53 is suggested to confer gain-of-function in cancer cells. The characteristics of the cells used in the current study were related to the p53 null phenotype, but it will be interesting to determine if cancer cells carrying mutant p53 have a null+gain-of-function phenotype, or if gainof-function alters the null phenotype, in order to further understand the role of p53 in tumorigenesis. Such a study will require a large amount of work, but is probably feasible.

In addition to our responses, we would like to take this opportunity to discuss the cell biological meaning of “generality”. For example, if a response is detected in cell types A, B, and C by e.g. enzymatic assay, quantitation of protein expression levels, and staining of cells, it is often concluded that the response is commonly induced in those cells (generalized). However, as noted by this reviewer, the levels of responses may vary among cells, and commonly induced responses may thus only occur in a specific group of cells in the A, B, and C cell populations. In this case, such responses may not be generally induced in cell types A, B, and C, but only in certain subpopulations of these cell populations. In the current study, cell death etc. were induced in the A549 cell population following p53 silencing, but not in the majority of A549 cells, indicating that this might not be “general” for A549 cells, according to the definition of “generality” used for classical experimental approaches. We have thus been considering the meaning of the term “general”. Each cell in a cell population may have a different status, and without knowing the context affecting the status of each cell, it is not possible to establish “generality”. Information regarding the context of each cell in various types of cell populations is currently lacking, and we do not know how many contexts exist. In the current study, we described one context related to A549 cells, but there will be many other contexts, which may be similar to or distinct from A549 cells. We therefore consider that we are still at the stage of revealing such contexts, e.g. contexts for cancer cells carrying p53 mutation and for metastatic cells, and some commonality may begin to emerge after more contexts have been revealed. However, revealing these contexts will require extensive work, and we hope that other groups will also show an interest in this type of study.

We have addressed some these points in the revised Discussion.

“The tools described, including the DIC tracking software and the simulation algorithms would be useful additions to the biologist's toolkit. The direct visualization of siRNA transfection agents through DIC, and its integration with western blotting is novel, and the authors may consider preparing a protocol or methods paper that describes this in more detail, as it may be useful for trouble-shooting when encountering difficulties with siRNA transfections. ”

We appreciate the encouraging comments and would be happy to publish a protocol.

“The use of white-light imaging is refreshing, as many of us in the field default to fluorescence imaging, which has the potential to interfere with cell proliferation. Overall, the approach is innovative by extracting the most information possible from optical imaging data sets, in the less invasive way possible.”

We have been working on live-cell imaging since 2000 and had difficulty maintaining cell viability using fluorescent imaging. We therefore tried various light sources and found that nearinfrared light (not white light) was less toxic to the cells, allowing us to maintain cell cultures for at least a month on a microscope stage. We mentioned that near-infrared was used in the current study (“System to investigate the functional implications of maintaining low levels of p53 in unstressed cells” in the Results.

Indeed, the power of photographs lies in the fragments of history that we cannot take in. On December 16, 2014, Taliban stormed a children school in Peshawar, where more than a 100 children were killed. The photographs of blood bath and massacre in school still invites the most horrible memory our city Peshawar ever witnessed. It is that fragment of history we cannot take in. In contrast to this, when we see photographs of those young children dressed in uniforms, as a memory of who passed away in the attack still invokes a different kind of a meaning. A photograph freezes the moment between life and death. In that very moment, when a child was posing in school uniform, he was very well alive, unaware of what will happen to him. When today their parents hold photographs by protesting on roads to find justice hurts even more. After reading this, I think there is a need to do similar work which emphasizes that those killed in wars were human too. For instance placing the pictures of people in some seminar project where people could come and see who died in Drone strikes or military operations. It may evoke some anti-war sentiments that those killed were not just numbers or stats but human beings.

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity):

The authors present further investigation of the Sox transcription factors in the model Cnidarian Hydractinia. They showcase the Hydractinia as now a relatively technically advanced model system to study animal stem cells, regeneration and the control of differentiation in animal cells. In this study they characterise the neural cells in hydractinia using FACS and sing cell transcriptome sequencing, investigate the sequential expression of SoxB genes in the i-cells and presumptive lineage giving rise to i-cells and investigate the neuronal regeneration making good use of transgenic rules. Finally, they investigate the role of SoxB genes in embryonic neurogenesis.

There are no major or minor issues effecting the conclusions

Reviewer #1 (Significance):

This study helps to confirm the role of an important group of transcription factors is conserved across the metazoan as well as showcasing an exciting model organism for regeneration and stem cell biology. This will of interest to a broad audience of developmental and biologists.

My own research is in the same field, using a different model system

Referees cross-commenting

I agree with the comments from the other reviewers, and am sure the authors can address these adequately with further explanation.

Reviewer #2 (Evidence, reproducibility and clarity):

Summary

Chrysostomou et al. investigate the role of three putative SoxB genes in embryonic neurogenesis in the colonial hydrozoan Hydractinia. They show that SoxB1 is co-expressed with Piwi in the multipotent i-cells and, using transgenics, they show that these Piwi/SoxB1 cells become neurons and gametes, consistent with the cell types that differentiate from i-cells. They further suggest that SoxB2 and SoxB3 are expressed downstream of SoxB1 in the progeny of the i-cells and, using shRNAs, investigate the role of SoxB genes on embryonic neurogenesis. The primary conclusions center on the similarity between neural differentiation in humans and Hydractinia as both systems pattern neurons using sequential expression of SoxB genes during the differentiation of neurons. The manuscript presents a large and diverse set of data derived from analysis of transgenic animals, single-cell sequencing, and investigation of gene function; despite this, the conclusions are either not particularly novel or not well-supported. The co-expression of SoxB1 in Piwi-expressing i-cells appears to be both novel and significant but the implications are not clearly indicated. Additional specific concerns are detailed below.

Major comments

1. SoxB genes act sequentially<br /> Knockdown of SoxB2 has already been shown to result in the loss of SoxB3, so the sequential action of SoxB genes in this animal does not seem to be a terribly novel conclusion.

Sequential expression of Soxb1-Soxb2 has not been demonstrated previously. Flici et al. did show some data on Soxb1 expression but these were not detailed. Furthermore, they have not shown in vivo transition to Soxb2. Our new single-molecule fluorescence in situ hybridization, and the transgenic reporter animals have been developed to address these issues.

While this manuscript does appear to report the most comprehensive analysis of SoxB1 expression, the evidence for sequential activation of SoxB1 and then SoxB2 in the same lineage (Figure 4) is a bit troubling. Panel A of this figure appears to show complete overlap between SoxB1 and SoxB2, suggesting all the cells in this field are synchronously passing through the transition point from SoxB1 to SoxB2 expression. While this may reflect reality, it would be more convincing to see adjacent cells expressing SoxB1 only or SoxB2 only, reflecting the dynamic progression of cell type specification along the main body axis.

As shown in Figures 1, Soxb1 is expressed by i-cells (together with Piwi1) in the lower body column of feeding polyps and in germ cells in sexual polyps. These cells do not express Soxb2. Figure 2 shows that Soxb2 is expressed more orally in a population of putative i-cell progeny as they migrate towards the head. These cells still express Soxb1. In the upper part of the body column, just under the tentacle line, there are Soxb2+ cells that do not express Soxb1. Therefore, cells expressing Soxb1 but not Soxb2 are present in the basal part of the polyp, Soxb1+/Soxb2+ double positive cells in the mid body region (i.e., the interface between the two domains where Soxb1+ cells start to express Soxb2 and downregulate Soxb1.), and cells expressing Soxb2 but not Soxb1 in the upper part of the polyp, just under the tentacle line. In Figure 4, we show the interface between these two domains using in vivo imaging of double transgenic reporter animals to visualize the Soxb1 to Soxb2 transition. Indeed, in the mid body area, most Soxb1+ cells also express Soxb2 (Figure 2). Hence, Figure 4 should be seen keeping Figure 2’s data in mind. At the mRNA level, the overlap between the Soxb1 and Soxb2 domains is smaller (Figure 2) than the one shown in Figure 4 because the latter constitutes a lineage tracing, showing fluorescent proteins with a long half-life. Therefore, when i-cells downregulate Soxb1 while starting to express Soxb2, the long half-life of tdTomato results in red fluorescence persisting longer than the mRNA encoding it. We have added cartoons to Figure 4 to indicate the position along the main body axis that are depicted.

Panel B is more concerning; while the authors have highlighted a cell that does appear to transition from SoxB1+ to SoxB1+/SoxB2+, there are several cells in the background that appear to gain SoxB2 expression without first expressing SoxB1. Do these cells constitute a fundamentally different, SoxB1-indpenendent, lineage of SoxB2+ cells? This would be noteworthy but is not mentioned or characterized.

The panels included in Figure 4 constitute selected confocal slices of stacks acquired in vivo. During imaging, cells move in three dimensions, making them appear and disappear in given optical planes over time. In other words, the individual time frames shown (T0-T5) were not always found in the same plane due to cell migration in the Z dimension. The cells that appear to gain Soxb2+ w/o having expressed Soxb1 first are an example of such cells. They are probably Soxb2+ cells that had already downregulated Soxb1 and migrated into the respective plane of image. We have added the explanation to Figure 4's legend.

Figure 7 shows the effect of SoxB1 knockdown (by shRNA) on the number of Piwi-expressing cells, nematocytes, etc but why not show that SoxB2 and SoxB3 are also knocked down in these experiments? Figure S11 shows no effect of SoxB2 and SoxB3 knockdown on SoxB1 expression but why wasn't the reciprocal experiment performed? If SoxB2 and SoxB3 are really downstream of SoxB1, the authors should demonstrate that with the shRNA experiments.

Our data show that Soxb1 is expressed in i-cells and its KD reduces the number of these stem cells (assessed by expression of Piwi1, an i-cell marker). Because i-cells give rise to all Hydractinia somatic lineages (and to germ cells), focusing specifically on Soxb2+ cells would provide no further insight because all cell types are expected to be affected. Indeed, injection of shRNA targeting Soxb1 resulted in smaller animals with multiple defects, including but not limited to the neural lineage.

1. Knockdown of SoxB genes resulted in complex defects in embryonic neurogenesis<br /> The manuscript aims to detail the roles of SoxB1, SoxB2, and SoxB3 in embryogenesis but only one of the main figures even shows pre-polyp life stages (Figure 7) and the results presented in in this figure are confusing. The authors suggest that knockdown of SoxB3 had no effect on embryonic neurogenesis but another interpretation of these data is that the SoxB3 shRNA simply did not work. The authors should provide additional support to show that this reagent is working as expected.

This information is included in Figure S11. Using mRNA in situ hybridization, we show that injection of shRNA targeting Soxb3 causes transcriptional downregulation of Soxb3 but not of Soxb2. The figure also shows the specificities of the shRNAs targeting Soxb1 and Soxb2.

Further, the results for SoxB1 and SoxB2 knockdown do not support the previous investigation of the role of SoxB2 in neurogenesis (Flici et al 2017). If SoxB1 is upstream of SoxB2, how does knockdown of SoxB1 have such a dramatic effect on RFamide neurons and nematocytes but knockdown of SoxB2 has an effect only on RFamide neurons? Is it possible the SoxB2 shRNA also wasn't working as expected? Can the results of the Flici et al 2017 paper showing SoxB2 knockdown in polyps be recapitulated using these shRNAs? If the point is to argue that embryos and adults (polyps) use fundamentally different mechanisms to drive neurogenesis, then the results presented in Figures 1-6 (which investigate SoxB genes in polyps) can't really be used to make inferences about embryonic neurogenesis. I think the authors have more work to do to demonstrate that embryonic and adult neurogenesis fundamentally differ.

The Soxb2 shRNA specificity is shown in Figure S11 (i.e., it KD Soxb2 but not Soxb1). We were equally surprised to discover that Soxb2 KD resulted in somewhat different phenotypes than the ones obtained by Flici et al. (2017) in polyps. At this stage, we cannot explain the difference. However, one could speculate that it resulted from slightly different regulation logic between embryonic and adult neurogenesis. More specifically, we propose different priorities for generating neural subtypes as explanation. Unfortunately, shRNAs work only with embryos, and long dsRNA mediated KD works only with polyps. CRISPR/Cas9-mediated KO is feasible in Hydractinia, but knocking out developmental genes, such as these Sox genes, would likely cause embryonic lethality. Other conditional KO/KD approaches are not available for Hydractinia. We believe we have made all possible efforts to clarify the roles of these genes using currently available techniques. Neurogenesis is a complex process that is only partially conserved among different animals and poorly studied in non-bilaterians. Furthermore, it is not possible to answer all questions in one study. As many studies before, our work contributes to the understanding of neurogenesis but also raises new questions. Addressing them is matter for future research. We have toned down the statement in the last sentence of the results and in the discussion and do not claim that embryonic and adult neurogenesis are fundamentally different.

Minor comments

Methods: A large bit of data from this manuscript relies on quantitative analysis of cell number but there's not enough information in the methods to understand how quantification was performed. How many slices from the z-stack were analyzed? Were counts made relative to the total tissue area in the X/Y dimension or relative to the number of total nuclei in the same section? How many individuals were examined for each analysis?

All cell counting analysis was performed using ImageJ/Fiji software. Counts were made relative to the total tissue area in the X/Y dimension (for the shRNA experiments). A Z-stack covering the whole depth of each larva was obtained. Counting was performed on cells positive for the respective cell type marker based on antibody staining and numbers were compared between shControl and shSoxb1/2/3 animals. At least 4 animals were counted per condition.

Page 11 - "Piwi2low cells, which are presumably i-cell progeny" - how were "high" and "low quantified?

“High” and “low” were not quantified. This is because i-cells progressively downregulate Piwi genes (i.e., Piwi1 and Piwi2) as they differentiate but this is a continuous process. Hence, it is difficult to put a threshold of Piwi1/Piwi2 protein level below which a cell ceases to be an i-cell while becoming a committed progeny. This is a similar process that is well documented in other animals where stemness markers are gradually downregulated during differentiation.

Page 13 - "a role in maintaining stemness" - this comment is not totally clear to me. Why would the number of EdU+ cells increase if the role of SoxB1 is to maintain stemness? Wouldn't SoxB1 knockdown then force stem cells to exit their program, resulting in early differentiation of i-cell progeny? This should be clarified.

KD of Soxb1 resulted in a decrease in the number of i-cells (i.e., Piwi1+ ones), suggesting that the gene is required for stemness maintenance. The increase in the numbers of cells in S-phase in this context was not related to i-cells because most of them were Piwi1-negative (Figure 7B). The identity of the cells in S-phase remains unknown, but a plausible explanation is that i-cell progeny (e.g., nematoblasts; see also next comment) increase their proliferative activity when i-cells numbers are low as a compensatory mechanism. This is merely a speculation. We have rephrased the paragraph to increase clarity.

Page 13 - "if progenitors are limiting" - if progenitors are limited why would there be an increase in nematocytes?

We do not have a definitive answer to this question but speculate that nematoblasts (i.e., stinging cell progenitors) account, at least in part, for the excessive proliferation seen under Soxb1 KD. This may constitute a mechanism allowing a depleted i-cell population to recover by self-renewal (instead of differentiation), moving temporarily the proliferation task to committed progeny (e.g., nematoblasts) until i-cell numbers return to normal. However, in the absence of evidence we refrain from expanding on this in the text.

Figures 1 and 2 claim to show "partial overlap" but they look perfectly overlapping to me. This makes the situation in Figure 4B difficult to interpret.

Figure 1 shows full overlap between Piwi1 and Sox1 expression and this is reflected in the text. Figure 2 shows no overlap between Soxb1 and Soxb2 in the lower body column (where only Soxb1 is expressed), overlap in the mid body region, and Soxb2 only expressing cells in the upper part of the body, just under the tentacle line. Similarly, the figure shows overlap between Soxb2/Soxb3 under the tentacle line, and predominantly Soxb3 above it in the head region. The small cartoons at the left side of each panel indicate its position along the oralaboral axis. See also our reply to the second part of comment #1.

Figure 4 - No indication of which part of the animal or which stage is shown in these images.

We have added cartoons to indicate the area in the polyp from where the images were taken.

Figure 5 - No indication of where these dissociated cells came from - polyps? Larvae?

All tissue samples were taken from feeding polyps; this is now mentioned in the Materials and Methods section.

Panel D is a bit perplexing - what are the "progeny" of Piwi+ cells if not SoxB2+ cells and their derivatives?

In Panel D, we show three cell fractions. One constitutes i-cells, based on high Piwi1 expression (green fluorescence of the Piwi1::GFP reporter transgene) and morphology; one fraction includes nematocytes, based on the characteristic nematocyst capsule, and one constitutes a mixture of other i-cell progeny. The latter includes different cell types, given that i-cells are thought to contribute to all lineages. They have only dim GFP fluorescence because the Piwi1 promoter-driven GFP shuts down upon i-cell differentiation. Soxb2+ cells are also among them but are not the only i-cell progeny.

Why are nematocytes but not neurons indicated?

Neurons are shown on Panels E & F. See also next comment.

Piwi seems to be maintained in Ncol-expressing cells but not in SoxB2- or RFamide-expressing cells? Does this suggest that Piwi is turned on in i-cells, off in SoxB2-expressing cells, and on again in terminally differentiating nematocytes? This would be quite surprising and should be verified with antibody labeling/imaging in Piwi transgenics to confirm the result. The resolution for Panel M is too low to evaluate this part of the figure.

The Piwi1i gene is downregulated upon i-cell differentiation. In the Piwi1:GFP reporter animal, residual GFP fluorescence persists post differentiation due to GFP's long half-life. The brightness of which depends on the time elapsed since differentiation. Because nematocytes are short living cells with high turnover, most nematocytes have recently differentiated and are therefore relatively bright green in the Piwi1::GFP animal. Neuron turnover is lower, making most neurons in the same transgenic animal appear dim. The resolution of the imaging flow cytometer is limited because the machine images 1000s of cells per second through all optical channels. However, it is high enough to allow the identification of features such as cell shape, some organelles (e.g., nematocytes), nuclear size and shape, and fluorescence intensity.

Figure 7 - the low magnification images provide nice overall context but the authors should also provide high magnification panels for the same images. Without them it is not possible to assess "defects in ciliation" or to determine if there are defects in GLWamide neurons from these knockdowns (e.g., neurite vs cell body defects). There's no mention of the fact that SoxB1 knockdown resulted in complete loss of RFamide cells, which is strange. Are there SoxB2-independent populations of RFamide? Panel B could be interpreted multiple ways - downregulation of Piwi in SoxB1 shRNA or upregulation in SoxB2/B3. The authors should provide an image of control shRNA-injected larvae with the same co-labeling of Piwi/EdU for context. From the images, it's not clear that there were differential effects of SoxB2 and SoxB3 on nematocytes.

The resolution of the images is, in fact, high, allowing it to be blown up on the screen. Even higher magnification of ciliation can be seen in Figure S12. KD of Soxb1 resulted in complete or nearly complete loss of Rfamide+ neurons. We have added this statement to the text as requested. Panel B shows the relative difference in Piwi1+ and S-phase cells between shSoxb1, shSoxb2, and shSoxb3-treated animals. The quantification relative to the control is presented in Figure 7C.

Figures 6 and S9 - why piwi2 and not piwi1?

In Figure 6, we co-stained the regenerates with two antibodies: one was a rabbit anti-GFP (to visualize the RFamide+ neurons), and the other was a guinea pig anti-Piwi2 (to visualize icells). The anti-Piwi1 antibody that was used in other images to visualize i-cells was raised in rabbit and could not be used in conjunction with the anti-GFP one.

Figure S1 - Kayal et al 2018 is the most recent phylogeny of cnidarians and should probably be cited in place of Zapata throughout the manuscript. Independent of this, the polytomy in Figure S1 panel A is not supported by either Zapata or Kayal and should be fixed.

We have cited Kayal et al. 2018 and revised the tree in Figure S1 as pointed.

Figure S3 - is this mRNA? Protein? Panels E-G are too small to interpret. Please provide stage/time for cartoons in panel H.

As per the legend, Panels A, B, D, E, F refer to protein; C is lectin staining (DSA), and G is EdU. The resolution of Panels E-G is actually high, allowing blowing up of the images on the screen to view the details. The stages of the cartoon in Panel H are now provided in the figure legend.

Figure S11 - please provide images of whole larvae as shown for Piwi knockdown in Fig S9 and some additional support (e.g., qPCR) to demonstrate the shRNAs are actually working.

Figure S9 represents immunostaining using the anti-Piwi1 antibody. In Figure S11, we show the specificity of the shRNA treatments; we used highly sensitive single-molecule mRNA in situ hybridization. Whole animal imaging is not informative due to the punctuated nature of the single-molecule staining.

Figure S12 - it's not clear what ciliary "defects" are being shown.

In the control, cilia are uniformly distributed along the oral-aboral axis whereas in the shSoxb1-injected animals, the pattern is patchy. Additionally, shSoxb1-injected larvae could not swim (planulae swim by coordinated cilia beat).

Reviewer #2 (Significance):

Generally, the results are either equivocal or the conclusions are not well supported by the results (as detailed above). The significance of this work to vertebrate neurobiology is somewhat weak. (Especially considering the orthology of these genes to bilaterian SoxB genes is not well supported.) Why not compare these results to other cnidarians - the expression patterns of SoxB1 and SoxB2 in corals and sea anemones seem to differ quite a lot (Shinzato et al 2008; Magie et al 2005), suggesting these genes are almost certainly not behaving in the same way across cnidarians. This is exciting! What's happening in Hydra? Seems like it should be possible to mine the single-cell data set from Siebert et al to test these hypothesized relationships between the Sox genes in another hydrozoan which constantly makes new neurons.

We have modified the concluding section in the discussion, in line with this comment. See also comment to Reviewer #3.

Reviewer #3 (Evidence, reproducibility and clarity):

This paper characterizes the role of Soxb genes in neurogenesis in Hydractinia. The authors use cutting edge approaches including FISH, transgenics, image flow cytometry, FACS and shRNA knock downs to characterize SoxB in Hydractinia. The images are beautiful, the data is sound and the interpretation of the data is appropriate.

I have only minor suggested listed by section below:

Abstract<br /> - The abstract and introduction should make clear that this is a colonial animal and the cell migration occurs from the aboral to the oral end of the polyp (not the animal, as there are many oral ends). This is relevant to the interpretation of the data as the polyps do not act in isolation as they interconnected and may communicate via the stolonal network that connects the polyps in the colony.

We have added a section to the Introduction to address the reviewer's comment. The Abstract, however, is too short to include this explanation.

• The human disease justification is a relatively weak one and does not need to be included. Using Hydractinia to understand the role of SoxB in the evolution of neurogenesis in animals is enough justification for the study.

We have adopted the reviewer's comment and modified the statement in the discussion (see also comment to Reviewer #2).

Introduction<br /> - Instead of Sox phylogenies (the term phylogeny is more appropriate for species trees), consider substituting, for Sox gene trees. And instead of "phylogenetic relation" use the term "orthology"

This has been done.

• The number of times the sentences that have the sentiment "....remain unknown." "....little is known.." "...unclear..." , "....difficult to establish...." etc. is distracting and detracts from what IS known about these genes. It is not necessary to continually justify the study throughout the introduction. Instead a clearer description of the background and setting up the question/hypothesis of SoxB paralog subfuctionalization in space and time - would be more informative to the reader.

We have reduced the number of occasions as recommended.

• The authors state that there are three SoxB genes in the Hydractinia genome? What genome? For several years there has been multiple papers published by subsets of these authors have used unpublished genome data, but the complete genome has yet to be released to the public. This is especially egregious because they cite their NSF funded EDGE proposal to CEF and UF which is supposed to develop tools to the community, and yet the community at large doesn't have access to the genome. If these data came from the genome, then the genome should be released. If these data came from a previously published transcriptome as in the previous SoxB paper then this should be stated explicitly.

The Hydractinia genome assembly, annotation, RNA-seq data, and genome browser are now available in the Hydractinia genome project portal at the National Human Genome Research Institute (NIH) website (https://research.nhgri.nih.gov/hydractinia/). The raw data have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject PRJNA807936. This information has been added to the 'Resource availability' section.

Results<br /> - I assume there was no expression of Soxb2 and Soxb3 in the reproductive polyps? This should be stated explicitly.

Soxb2 expression in sexual polyps was consistent with the nervous system and with maternal deposition in oocytes. It was not detected in male germ cells. We have added a new in situ hybridization image of Soxb2 to Figure 12.

• The word "progeny" is used throughout to describe terminally differentiated cells. However, progeny implies offspring, but these are actually later stages of differentiation of the in a cell's ontogeny, thus the term should be changed to "differentiated cells"

We used "progeny" to indicate that the corresponding cells derived from a specific progenitor cell type. We did try replacing it with "differentiated cells" but this completely changes the meaning of the sentence: first, it does not include the cell of origin info and second, not all progeny are already fully differentiated.

• Typo on page 11 "This predictable generation of many new neurons provides an opportunity to study neurogenesis in [a ]regeneration." - Remove the "a"

Corrected.

• While the regeneration study is interesting, there is nothing revealed about the role of Soxb and there is not a lot of new information revealed about regenerations. Authors should better justify this section or consider omitting.

These sections demonstrate de novo neurogenesis in head regeneration. This was not known in this animal before.

Discussion<br /> - The authors assume that in the transgenic lineage, the fluorescent marker in differentiated cells is due to retention of fluorescence, but it is unclear if they can rule out that Soxb2 is still being expressed in those cells" Please clarify.

We conclude this by comparing the mRNA expression (Figures 1 & 2) with the fluorescent proteins (Figure 3).

• How did the authors determine that the shSoxb3 knockdown worked? Please discuss relevant controls and validation (either in discussion or methods). This is particularly important given that it didn't have an apparent phenotypic effect.

The efficacy of all shRNAs determined by in situ hybridization, showing that each shRNA downregulates its own target mRNA but not the others (Figure S11).

• Again, the connection to human health is a bit of a stretch. Instead, what is most interesting is the similarity of Soxb paralogs acting sequentially as has been found in vertebrates. This suggests a highly conserved mechanism of subfunctionization following gene duplication at the base of animals.

We agree. This is now also better highlighted in the discussion.

Figures<br /> - Its very hard to distinguish the overall abundance of Soxb2 and Soxb3 expression along the polyp body axis from the panels figure 2. A lower magnification or larger area in each region would be helpful

In Figure 2, we performed single-molecule in situ hybridization. While highly sensitive, this method generates spotty images because they highlight single molecules and are not coupled to an enzymatic reaction as in other methods. They mostly looks poor when showing low magnification images. Because a previous study (Flici et al. 2017) has already shown the general expression pattern, we aimed at providing the details of the transition.

• Figure 4 - either the figure is upside down or the text is upside down. It is also difficult to see the double staining (if any).

The figure is oriented to position the oral end up. The resolution of the panels is high, enabling blowing-up on the screen. The quality of in vivo time lapse images cannot match that of fixed and antibody stained ones, or of single in vivo images. This is because the animals are imaged for many hours during which they tend to bleach.

• Figure 5M is difficult to read due to the small print. Consider enlarging and moving it to Supplementary Material

The size of the text is small but the resolution is very high, enabling blowing up the image on the screen. We thought that the information was important enough to be presented in the main text and given that most readers would use the electronic version we preferred this option on another supplemental figure on top of the 12 we already have.

Reviewer #3 (Significance):

This is an interesting and important study because although it is well known that SoxB genes function in neurogenesis in animals, it is unclear how and if subfunctionalization occurs outside of vertebrates. Hydractinia is an excellent model to study SoxB genes because of its colonial organization and continuous development of nerve cells throughout the life of the animal. In addition, it is part of the early diverging cnidarian lineage and thus can provide insight into the relative conservation of SoxB genes across animals.

Hugo places a certain scrutiny onto riots, and here argues that, even those who think that the June 1832 insurrection was a riot, "cannot speak of it without respect." I understand Hugo's distinction between riots and insurrections, but is a riot against injustice not a noble act to him? Why must an uprising be an insurrection in order to gain our respect? Yes, riots are violent and seemingly random, but they also serve a purpose in society and are one viable option for the oppressed.

I have't read on yet, but I wonder if there are any biases that we are unaware of that contribute to this mistreatment. We did a study similar to this, on race and age using the tool "Implicit Association Test", which is said to unveil hidden biases that we may have. I have added the link if you want to try this out for yourself. Personally, I believe that there are many things that are unconscious to us, and we try to avoid negative biases, but sometimes they can be apart of our nature based on the way we were raised, and the environment we are exposed to.

https://implicit.harvard.edu/implicit/selectatest.html

It is really sad to hear the data on this as we think of teachers to be loving caretakers.

Author Response:

Reviewer #1 (Public Review):

This article focuses on a quantitative description of airineme morphology and its consequences for contact and communication between cells via these long narrow projections. The primary conclusions are

1) Airineme shapes are consistent with a persistent random walk model (analogous to a wormlike polymer chain), unhindered by the presence of other cells.

The authors convincingly demonstrate, using analysis of the mean-squared-displacement along the airineme contour, that the structures cannot be described by a diffusive growth process (ie: a Gaussian chain) as would be expected if there were no directional correlations between consecutive steps. Furthermore, by observing the airineme growth and looking at the distribution of step-sizes, they show that these steps do not exhibit the expected long-tail distributions that would imply a Levy-walk behavior. The persistent random walk (PRW) is presented as an alternative that is not inconsistent with the data. However, given the high level of noise due to low sampling, the claimed scaling behavior of the MSD at long lengths is not fully convincing. Nevertheless, the PRW provides a plausible potential description of the airineme shapes.

To reiterate the comment: the MSD analysis allows us to reject the simple random walk model, and it is consistent but alone is not strongly supportive of the PRW model, especially at high time of around 15 minutes (long lengths of around 65 microns). As the Reviewer points out, this is due to low numbers of long airinemes.

This prompted us to investigate the long-length data using multiple analysis approaches. In the new manuscript, new Fig 2B, we took all airinemes whose growth time was greater than 15 min, and plotted their final angle, i.e., the angle between the tangent vector at their point of emergence from the source cell and the tangent vector at their tip. At long times (>1/D_theta), the PRW model predicts that the angular distribution should become isotropic.

In new 2B, we find that the angular distribution is uniform, i.e., isotropic, using a Kolmogorov-Smirnov test (p-value 0.37, N=26).

Since there are relatively few data points, we repeated this analysis under various airineme selection criteria, and in all cases found the final angular distribution to be consistent with uniformity (new Supplemental Data Figure 1). For example, if we set the threshold at 10min, which includes N=49 airinemes, the Kolmogorov-Smirnov test against a uniform angular distribution gives a p-value of 0.32.

We here add a few additional notes

● Note that there is significantly less data used in this test than in the MSD analysis or the autocorrelation function maximum likelihood analysis. In order to perform a hypothesis test, we wanted to be sure that the data points are independent, so we take only one from each airineme (unlike MSD and autocorrelation analyses, for which we take every interval of a particular length, whether in the same airineme or not.)

● Finally, although the >10min KS test has more data than the >15min KS test (N=49 compared to N=26), we have chosen to present the >15min KS test in the Main Text. As we mentioned above, the conclusion is unchanged for >10min (see Supporting Data). The reason is that >15min is the first test we ran to check angular distribution against a uniform (-pi,pi) distribution, and we did not want to bias our testing.

Taken together, the data are even more strongly supportive of the PRW model. We are grateful for the Reviewer in encouraging us to further explore the high-time data.

2) The flexibility (ie: persistence length) of the airineme shapes is one that maximizes the probability of a given airineme (of fixed length) contacting the target cell.

This optimum arises due to the balance between straight-line paths that reach far from the source but cover a narrow region of space and diffusive paths that compactly explore space but do not reach far from the starting point. Such optimization has previously been noted in unrelated contexts both for search processes of moving particles and for semiflexible chains that need to contact a target. The authors present a compelling case (Fig 4B) that the measured angular diffusion of the airinemes falls close to the predicted optimum. Furthermore, the measured probability of hitting the target cell also lies close to the model prediction, providing a strong test of the applicability of their model.

3) Airineme flexibility engenders a tradeoff between contact probability and directional information (ie: the extent to which the target cell can determine the position of the source).

This calculation proposes an alternative utility metric for communication via airinemes. The observed flexiblity is shown to be at a Pareto optimum, where changes in either direction would decrease either the probability of contact or the directional information. Again the absolute value of the metric (Fisher information for angular distribution) is within the predicted order of magnitude from the model. Thus, while the importance of maximizing this metric remains speculative, its quantitative value provides an additional test for the applicability of the PRW model.

Overall, this paper provides an interesting exploration of optimization problems for communication by long thin projections. A particular strength is the quantitative match to experimental data -- indicating not just that the experimental parameters fall along a putative optimum but also that the metrics being optimized are well-predicted by the model. Defining an optimization problem and showing that some parameter sits at the optimum is a common approach to generating insight in biophysical modeling, albeit invariably suffering from the fact that it is difficult to know which optimization criteria actually matter in a particular cellular system. The authors do an excellent job of exploring multiple optimization criteria, quantifying the balance between them, and pointing out inherent limitations in knowing which is most relevant.

A minor weakness of the manuscript is its focus on a very narrowly defined cellular system, with the general applicability of the results not being highlighted for clarity. For example, the fact that the same flexiblity optimizes contact probability and the balance between contact and directional information is an interesting conclusion of the paper. Is this true in general? Is it applicable to other systems involving a semiflexible structure reaching for a target or a moving agent executing a PRW?

The Reviewer’s question is an excellent question: Is the trade-off between contact and directional information a general property of searchers that obey persistent random walks? To address this question, we now include the analysis previously contained in Figure 5D, but for a full parameter space exploration. This is done in new Figure 5 Supplemental Figure 1. In doing so, we found fascinating behavior that sheds some light on the loop in Fig 5D.

At low d_targ, the trade-off is amplified, and the parametric curve resembles bull's horns with two tips representing the smallest and largest D_theta in our explored range, pointing outward so the shape is concave-up. Intuitively, we understand this as follows: since the target is fairly close (relative to l_max), contact is easy. The only way to get directional specification is by increasing D_theta to be very large, effectively shrinking the search range so it only reaches (with significant probability) the target at the near side (“3-o-clock'' in Fig. 5A). At low d_targ, the parametric curve is concave-up, and there is no Pareto optimum.

At high d_targ, the searcher either barely reaches (when D_theta is high), and does so at 3-o-clock, therefore providing high directional information, or D_theta is low, and the searcher fails to reach, and therefore also fails to provide directional information. So, at high d_targ, there is no trade-off.

At intermediate d_targ, the curve transitions from concave-up bull's horn to the no-tradeoff line. To our surprise, it does so by bending forward, forming a loop, and closing the loop as the low-D_theta tip moves towards the origin. At these intermediate d_targ values, the loop offers a concave-down region with a Pareto optimum.

So, to answer the specific question of the Reviewers: No, the Pareto optimum is not a general feature of persistent random walk searchers. It only exists in a particular parameter regime, sandwiched between a regime where there is a strict trade-off with no Pareto optimum and a regime in which there is no trade-off.

All of these results are now discussed in the main text.

(Note that although we do not explicitly explore lmax, since these plots have not been nondimensionalized, the parametric curve for a different lmax can be obtained by rescaling the results).

Reviewer #2 (Public Review):

Signalling filopodia are essential in disseminating chemical signals in development and tissue homeostasis. These signalling filopodia can be defined as nanotubes, cytonemes, or the recently discovered airinemes. Airinemes are protrusions established between pigment cells due to the help of macrophages. Macrophages take up a small vesicle from one pigment cell and carry it over to the neighbouring pigment cell to induce signalling. However, the vesicle maintains contact with the source cell due to a thin protrusion - the airineme. In support of these data, the authors find that the extension progress of the airinemes fits an "unobstructed persistent random walk model" as described for other macrophages or neutrophils.

The authors describe the characteristics of an airineme as it would be a signalling filopodia, e.g. a nanotube or a cytoneme, which sends out to target a cell. An airineme, however, is fundamentally different. Here, a macrophage approaches a pigment cell binds to the airineme vesicle. Then, the macrophage approaches a target pigment cell and hands over the airineme vesicle. During this process, the airineme vesicle maintains a connection to the source pigment cell by a thin protrusion. Then, the macrophage leaves the target cell, but the airineme vesicle, including the protrusion, is stabilized at the surface and activates signalling. Indeed nearly all airinemes observed have been associated with macrophages (Eom et al., 2017).

Therefore, it is essential to focus on the "search-and-find" walk of the macrophage and not the passively dragged airineme. In the light of this discussion, I am not sure if statements like "allow the airineme to hit the target cell" are helpful as it would point towards an actively expanding protrusion like a filopodium.

We have added a new paragraph in the Introduction emphasizing the role of the macrophage, and we have changed the language. In particular, we want to remove agency from the airineme, since it is indeed moving with the macrophage. In the mathematical sections, we opt for the phrase “search process”.

We have also clarified that, in the biological system, the details of contact are unclear (e.g., what mechanism in the macrophage-airineme-vesicle is responsible for distinguishing the target cell). Therefore, in the model, we have clarified that contact is declared when the airineme tip arrives at a distance r_targ from the center of the target cell, and this critical distance might be larger than the size of the target cell, since it might include part or all of the macrophage.

Reviewer #3 (Public Review):

This paper studies statistical aspects of the role of long-range cellular protrusions called airinemes as means of intracellular communication. The mean square distance of an airineme tip is found to follow a persistent random walk with a given velocity and angular diffusion. It is argues that this distribution with these parameters is the one that optimise the probability of contact with the target cell. The authors then evaluate the directional information (where in space did the airineme come from) and found that, again, the measure diffusion coefficient optimise the trade-off between high directional information (small diffusion) and large encounter probability.

I found this paper well written and clear, and addressing an interesting problem (long-range intracellular communication) using rigorous quantitative tools. This is a very useful approach, which appears to have been appropriately done, that in itself makes this paper worthy of interest.

1) The main conclusion of this paper is that the airineme properties optimises something that has to do with their function. Although rather appealing, I find this kind of conclusion often questionable considering the large uncertainty surrounding many parameters.

We agree that conclusions about optimality need to be expressed carefully, to avoid teleological statements and overstating our knowledge about the constraints and variability faced by the living system. In the revised manuscript, we strive to use language to point out that the parameter extracted from data (an average) and the parameter predicted to be optimal (on average) are approximately equal, and avoid speculation about the evolutionary process that may have led to these parameters.

Here, optimality is shown from a practical perspective, using measure parameters. For instance, the optimal diffusion coefficient for hitting the target varies by 2 orders of magnitude when the distance between cells is varied (Fig.3A). The measured coefficient is optimal for cells about 25 µm distant. Does this reflect anything about the physiological situation in which these airinemes operate?

To find the physiological regime in which the airinemes operate, we extracted distance-to-target measurements from imaging data, and found an average distance of 51 microns (note possible typo in referee comment), with a range of 33𝜇m − 84𝜇m, 𝑁 = 70. We report this in updated Table 1). The optima we find is in the average number of attempts before success (so, a single instance of an airineme may either succeed or fail, stochastically), when the distance to the target is 50 microns. These are both averages, across an entire fish epithelium (which contains ~10^5 source cells). So, for a particular cell generating airinemes, there may be different optimal parameters given a priori knowledge of its environment, but, across the whole fish epithelium, we assume the overall success corresponds to the average single-cell success we simulate.

Another rather puzzling claim is that the diffusion coefficient is optimised both for finding the target, AND for finding the best compromised between finding the target and providing directional information, while the latter must necessarily require weaker diffusion. Hence the last paragraph of p.6 ("the data is consistent with either conclusion that the curvature is optimized for search, or it is optimized to balance search and directional information"), although quite honest, gives the feeling that the conclusions are not very robust. I would welcome a discussion of these points.

We have clarified the result about directional information in the new manuscript.

First, it is not optimized for maximal directional information, in the sense that there are other parameters that would give more directional information – we apologize for the lack of clarity. Rather, the parameters observed are such that changing them would either reduce search success or directional information. In the study of multiple optimization, this property is called “Pareto optimality”.

Second, the Reviewer’s intuition is that weaker diffusion (straighter airinemes) would provide more directional information. This was indeed our intuition as well, prior to this study. To our surprise, we found that very weak diffusion or very strong diffusion both give local maxima of directional information. The intuitive explanation is that the searchers are finite-length, and high diffusion leads to a smaller search extent which only reaches the target cell at its very nearest region. We provide this intuitive explanation (which was indeed a surprise to us) in the Results section.

Third, the Reviewer asks about the generality of the result about directional information. This is an excellent question. The comment, and similar comments from other Reviewers, prompted us to perform a parameter exploration study. This is contained in a new Supplemental Figure and new paragraphs in the Results section.

The Reviewer’s question is an excellent question: Is the trade-off between contact and directional information a general property of searchers that obey persistent random walks? To address this question, we now include the analysis previously contained in Figure 5D, but for a full parameter space exploration. This is done in new Figure 5 Supplemental Figure 1. In doing so, we found fascinating behavior that sheds some light on the loop in Fig 5D.

At low d_targ, the trade-off is amplified, and the parametric curve resembles bull's horns with two tips representing the smallest and largest D_theta in our explored range, pointing outward so the shape is concave-up. Intuitively, we understand this as follows: since the target is fairly close (relative to l_max), contact is easy. The only way to get directional specification is by increasing D_theta to be very large, effectively shrinking the search range so it only reaches (with significant probability) the target at the near side (“3-o-clock'' in Fig. 5A). At low d_targ, the parametric curve is concave-up, and there is no Pareto optimum.

At high d_targ, the searcher either barely reaches (when D_theta is high), and does so at 3-o-clock, therefore providing high directional information, or D_theta is low, and the searcher fails to reach, and therefore also fails to provide directional information. So, at high d_targ, there is no trade-off.

At intermediate d_targ, the curve transitions from concave-up bull's horn to the no-tradeoff line. To our surprise, it does so by bending forward, forming a loop, and closing the loop as the low-D_theta tip moves towards the origin. At these intermediate d_targ values, the loop offers a concave-down region with a Pareto optimum.

So, to answer the specific question of the Reviewers: No, the Pareto optimum is not a general feature of persistent random walk searchers. It only exists in a particular parameter regime, sandwiched between a regime where there is a strict trade-off with no Pareto optimum and a regime in which there is no trade-off.

All of these results are now discussed in the main text.

(Note that although we do not explicitly explore lmax, since these plots have not been nondimensionalized, the parametric curve for a different lmax can be obtained by rescaling the results).

2) on p.4: "the airineme tips (which are transported by macrophages [30]) appear unrestricted in their motion". I don't understand what it means that the airineme tips are transported by macrophage, and I missed the explanation in the cited article. Is airineme dynamics internally generated (i.e. by actin/microtubule polymerisation) or does it reflect to motility of cells dragging the airineme along? This is discussed in passing in the Discussion, but I think that this should be explainde in more detail right from the start. Aslo, if a cell is indeed directing the tip, what does contact mean? Does it mean that the driving macrophage must contact the target cell and somehow attached the airineme to it? IF yes, that means that the airineme tip has a large spatial extent, which will certainly affect the contact probability.

These are very good questions. Airinemes have been characterized in a few studies since their discovery in 2015. We are saddened (and excited) to say that: the answers to all of these questions are currently unknown. To paraphrase the Reviewer, the questions are: First, what is the force generation mechanism that leads to airineme extension (additionally, if there are multiple coordinated force generators, e.g., the airineme’s internal cytoskeleton and the macrophage, how are these forces coordinated)? And second, what are the molecular details of airineme tip contact establishment upon arrival at a target cell?

We present an extended biological background discussion addressing these questions, including what is known and what remains unknown. We have incorporated a shortened version of this as a new paragraph in the introduction.

Airinemes are produced by xanthophore cells (also called yellow pigment cells) and play a role in the spatial organization of pigment cells that produce the patterns on zebrafish skin. Xanthophores have bleb-like structures at their membrane, and those blebs are the origin of the airineme vesicles at the tip. Those blebs express phosphatidylserine (PtdSer), an evolutionarily conserved ‘eat-me’ signal for macrophages. Macrophages recognize the blebs, ‘nibble,’ and ‘drag’ as they migrate around the tissue and the filaments trailing and extending behind. Airineme lengths have a maximum, regardless of whether they reach their target. If the airineme reaches a target before this length, the airineme tip complex recognizes target cells (melanophores) and the macrophage and airineme tip disconnect.

The airineme tip contains the receptor Delta-C, which activates Notch signaling in the target cell. The mechanism by which a macrophage hands off the airineme tip is still mysterious, due to temporal and spatial resolution limits. It is also known what other signals, if any, are carried by the airineme. If no target cell is found by the maximum length, the macrophage and airineme disconnect, and the airineme the extension switches to retraction. Thus, macrophages do not keep dragging the airineme vesicles until they find the target melanophores. However, how macrophages determine when to engulf the untargeted airineme vesicles is not understood.

Regarding the Reviewer’s specific question about the implications for the macrophage on how we model contact establishment: This would indeed change the interpretation of the model parameter r_targ. Specifically, contact is declared when the airineme tip arrives at a distance r_targ from the center of the target cell, and this critical distance might be larger than the size of the target cell, since it might include part or all of the macrophage. We have added this to the first part of Results, when the parameter is introduced.

3) Fig. 2A shows the airinemes MSD and the fit using the PRW model. I don't find the agreement so good. The power law t^2 seems good almost up to 10 minutes, and the scaling above that, if there is one, is clearly larger than linear. So I would say that the apparent agreement with the PRW model reflects the fact that there is a crossover from a ballistic motion to something else, but that this something else is not a randow walk. The MSD does look quite strange at long time, where it apparently decays. This made me wonder whether there might be a statistical biais in the data, for instance, the longest living airinemes are those who didn't find their target and hence those who travel less far, on average. I tried to get more information on the data from the ref.[29,30], but could not find anything. The authors should discuss these data and possible biais in more detail. For instance, do the data mix successful and unsuccessful airinemes? This is somewhat touched upon in Fig.s$, but I did not gain any useful information from it, except that the authors find the agreement "good" while it does not look so good to me.

To reiterate the comment, which is closely related to comments from other Reviewers: the MSD analysis allows us to reject the simple random walk model, and it is consistent but alone is not strongly supportive of the PRW model, especially at high tau of around 15 minutes (long lengths of around 65 microns). As the Reviewer points out, this is due to low numbers of long airinemes.

We agree, and have performed new analysis. The following is repeated here for convenience:

This prompted us to investigate the long-length data using multiple analysis approaches. In the new manuscript, new Fig 2B, we took all airinemes whose growth time was greater than 15 min, and plotted their final angle, i.e., the angle between the tangent vector at their point of emergence from the source cell and the tangent vector at their tip. At long times, the PRW model predicts that, for long times >1/D_theta, the angular distribution should become isotropic. In new 2B, we find that the angular distribution is uniform, i.e., isotropic, using a Kolmogorov-Smirnov test (p-value 0.37, N=26).

Since there are relatively few data points, we repeated this analysis under various airineme selection criteria, and in all cases found the final angular distribution to be consistent with uniformity (new Supplemental Data Figure 1). For example, if we set the threshold at 10min, which includes up to N=49 airinemes, the Kolmogorov-Smirnov test against a uniform angular distribution gives a p-value of 0.32.

We here add a few additional notes

● Note that there is significantly less data used in this test than in the MSD analysis or the autocorrelation function maximum likelihood analysis. In order to perform a hypothesis test, we wanted to be sure that the data points are independent, so we take only one from each airineme (unlike MSD and autocorrelation analyses, for which we take every interval of a particular length, whether in the same airineme or not.)

● Finally, although the >10min KS test has more data than the >15min KS test (N=49 compared to N=26), we have chosen to present the >15min KS test in the Main Text. As we mentioned above, the conclusion is unchanged for >10min (see Supporting Data). The reason is that >15min is the first test we ran to check angular distribution against a uniform (-pi,pi) distribution, and we did not want to bias our testing.

Taken together, the data are even more strongly supportive of the PRW model. We are grateful for the Reviewer in encouraging us to further explore the high-time data.

4) Regarding the directionality discussion, some aspect are a bit vague so that we are left to guess the assumptions made. For instance, the source cell is place at \theta=0 "without loss of generality" (p.6). Apparently (sketch Fig.5A) this also means that the airineme starting point from the source is at \theta=0, which clearly involves loss of generality, since the airineme could start from anywhere, its path could be hindered by the body of the source cell, and its contact angle would then be much less likely to be close to 0. It might be that in practice, only those airineme starting close to theta=0 do in fact make contact, but this should be discussed more thoroughly. Also, why is there to maxima in the Fisher information (Fig.5C) for very high and very low diffusion coefficient at short distance?

The sketch was indeed not clear about generality, so we have edited it so that the angles are no longer perpendicular. We also now also clarify in the Main Text that, in all simulations (both measuring contact probability and directional sensing), the airineme begins at a specified point in an orientation uniformly random in (-pi,pi). We apologize that this was not clear in the previous sketch.

Regarding hindrance by the source cell: While the tissue surface is crowded, the airineme tips appear unrestricted in their motion on the 2d surface, passing over or under other cells unimpeded (Eom et al., 2015, Eom and Parichy, 2017). We therefore do not consider obstacles in our model. This includes the source cell, i.e., we allow the search process to overlie the source cell. We now state this explicitly in the Main Text.

Regarding two maxima in Figure 5C (which was a surprise to us): We understand it with the following intuitive picture. For low D_theta, i.e., for very straight airinemes, the allowed contact locations are in a narrow range (by analogy, imagine the day-side of the planet Earth, as accessible by straight rays of sunlight), resulting in high directional information. For high D_theta, i.e., for very random airinemes, we initially expected low and decreasing directional information, since there is more randomness. However, these are finite-length searches, and the range of the search process shrinks as D_\theta increases. This results in a situation where the tip barely reaches only the closest point on the target cell, resulting again in high directional information. We have added this intuitive reasoning in the Main Text.

In addition to this inadvertent tendency to create stereotypes, I think that in only making a couple of personas to represent the learning audience you may fall into stereotyping just by lack of a sufficient sampling. How do you determine how many personas would be a representative enough sample? If you are looking at a diverse group of learners, you need more personas, and you need to have instructional materials that cover diverse needs. In larger groups would you break the group into sections to better address individual needs? Or have additional instructors?

This is such an important point! I think that often designers, SMEs, coders...everyone involved in the design team can become so enthralled with and focused on their design, that they lose sight of the learner experience. It seems to me that true LXD requires that the design team set their egos aside and be flexible and open to change in order to provide the best and most effective learning experience for the learner. If the design itself induces frustration, the learner may give up and never get to the actual learning process. Designers need to strive for ease of use and provide design with limited barriers for the learner

I have seen the role of designer-ego play out in the real world. In my role as a virtual math teacher, my colleagues and I regularly reached out to the curriculum team to request a change to the virtual book, activities, or assessments in order to enhance our students' experiences. Too often, we were told no, with no regard for the learner.

My frustration with this led me to want to move into curriculum so that the learner's point of view would be better understood and represented. That is part of what led me to my current role in Gifted and to the ID program at UF.

This week really put into perspective the idea of empathy vs. sympathy. It's much easier to sympathize with someone because you're still creating understanding from your personal perspective, while empathy requires what Baaki & Maddrell suggest: putting ourselves in someone else's shoes. However, I think that's a lot more difficult than these activities suggest. In Dr. Schmidt's example of creating a course for parents dealing with a child's diagnosis, while we may sympathize with them as instructional designers and human beings, it's far more difficult to truly understand the depths and nuances and of their experiences. While empathy interviews certainly help, it can never replace the experience. I saw this as a mother whose son recently received a diagnosis and dealing with the feelings and thoughts associated with it. I don't know if there's truly an empathy interview that an instructional designer can use to gauge and truly understand and feel what I do.

I think this idea is so important! Creating accommodations or new accessibility features is not just helpful for people with disabilities. They can serve as useful tools for anyone regardless of their abilities. If the accessible features can make everyone's time using a tool easier, why is there a lack of emphasis on creating these features? Tool creators may not prioritize them to begin with because they do not value people with disabilities as much as they should, but they need to realize these features can help everyone.

This is something that I think is very important to recognize. In regards to inclusion, it is essential that we are constantly aware of how or who we may be excluding others, even if we do not realize it. Nobody can be perfect 100% of the time, but as long as we are making an effort to respect others, that's all that we can ask for.

Author Response:

Reviewer #1 (Public Review):

In this detailed study the authors show that in isolated islets the polarity of the secretory apparatus is largely lost while it is preserved in slices where the capillary network remains intact. The authors then go on to show that the integrin/FAK pathway appears to be responsible for inducing and maintaining polarity, which involves concentration of active zone proteins and calcium channels at the contact sites and a higher sensitivity and potency of insulin secretion to glucose stimulation.

Generally, the data appear to be of high quality, being carried out with state-of-the-art technology, and the manuscript is lavishly illustrated. Since as a neuroscientist I am not sufficiently familiar with the field of the cell biology of insulin release it is difficult for me to judge whether there is sufficient advance in knowledge. A higher degree of organization of release sites including a role of active zone proteins was previously demonstrated from other endocrine organs involving the release of large dense-core vesicles such as chromaffin cells. Thus, the differences between the highly organized and rapidly responding exocytotic sites in neurons and the slower reacting release sites of peptide/protein containing granules are not fundamental but rather gradual, despite the principal cell biological differences between the biogenesis and recycling pathways of the secretory organelles.

In summary, the work adds new aspects to the understanding of the regulation of exocytosis in pancreatic beta cells. Aside from corrections of figure descriptions and experimental details, my only major comment relates to the data shown in Fig. 4. It appears that the difference in the time-to-peak between the two preparation is mainly caused by a (rather variable?) delay between glucose addition and the onset of the rise since the rate of increase is apparently not different between the preparations. Is this due a delay in depolarization, i.e. a delay in the closure of the ATP-K channels? This should be clarified. Also, the authors should show a comparative histogram of the delay times (between glucose addition and the inflection point at the onset of the rise).

The delay observed is due to a slower response in islets vs slices, which given the potentiating effects we show of the KATP channel drugs (diazoxide and now glibenclamide) is likely explained by a delay in KATP closure. However, since we are measuring the Ca2+ response we cannot directly prove this. We feel this is adequately discussed with reference to glucose-dependent triggering (where the KATP channel is a key component). In direct response to the referee’s comment about variability, we have re-expressed the data to show frequency histogram comparisons of the delay to peak (new Fig 4J).

Reviewer #2 (Public Review):

1) The authors present an investigation of subcellular distribution and dynamics of known presynaptic proteins in a relatively new approach, pancreatic slices, mastered by a limited number of laboratories, and which is currently the best method to largely preserve capillary networks. They demonstrate the advantage of this method by detailed cellular and subcellular optical analysis comparing isolated islets, islets in pancreatic slices, isolated islet cells and isolated islet cells on ECM (laminin) covered surfaces. This work provides good proof that preservation of capillary networks and corresponding distribution of proteins (laminin, liprin, integrin beta1 etc) is required for insulin secretion at the apical surface of islet cells. Moreover, in these pancreatic slices they observe a restriction of exocytotic sites at the vascular surfaces. The role of the extracellular matrix is also well investigated here by experiments on dispersed or single beta cells attached either to a glass-BSA interface or to a glass-laminin interface. However, the authors have already previously published in 2014 a restricted polarized insulin secretion in cultured islets as well as the preservation of localized liprin and laminin distribution (as well as RIM2 and piccolo; DOI 10.1007/s00125-014-3252-6). It is not clear why these data cannot be reproduced now again in isolated islets (see Fig. 1 and 2) .

We thank the referee for their comments. To clarify the specific issue around our past work. All our live sub-cellular resolution experiments have previously been performed with isolated islets – we have not, until recently been able to reliably get the slice to work. In contrast, our work with immunofluorescence of active zone proteins has been performed with fixed slices (including DOI 10.1007/s00125-014-3252-6, Low et al 2014).

2) The authors try to gain insight which mechanisms control this specific spatial restriction and they provide evidence that Focal Adhesion kinase activity is implicated in glucose-induced calcium fluxes and insulin secretion by the use of a small molecule antagonist and the use of a purified monoclonal antibody. They conclude that FAK is a master regulator of glucose induced insulin secretion that controls positioning of presynaptic scaffold proteins and the functioning of calcium channels. Although FAK may be a regulator, the claim that FAK controls functioning of calcium channels can certainly not be made. Ratio measurements of cellular calcium levels do not suffice for that (patch or sharp would be required). Moreover, the fact that KCl-induced insulin secretion (which bypasses nutrient metabolism and leads directly to opening of voltage-dependent calcium channels) is not altered by the FAK antagonist strongly argues against a role of FAK in calcium channel regulation. Indeed, the presented data suggest that FAK may intervene far more upstream from exocytosis such as in nutrient metabolism or granule mobility/maturation.

Our data clearly shows that integrin/FAK activation is part of the glucose dependent control of Ca2+ and insulin secretion. It is not relevant to this conclusion how we measure Ca2+ responses – they are obviously affected by all manipulations of integrin/FAK. We note that the referee is specifically correct in saying that we do not have evidence that Ca2+ channel function is a direct target of integrins/FAK and we have reworded the text to make this clear.

Further, our work does not define where in the glucose pathway integrin/FAK are acting. The referee is correct in saying the KCl data suggests it is upstream of the final stages of Ca2+ channel and exocytosis. Consistent with this we see effects of integrin/FAK manipulation on ELKS and liprin positioning (Figs 7 and 8) and, given the published data showing that ELKS enhances Ca2+ channel current (Ohara-Imaizumi et al 2019) we think it is plausible integrin/FAK intersect with this pathway to regulate Ca2+ channel activity. With reference to the high K responses, KCl rapidly depolarises the cells to recruit Ca2+ channels, in contrast glucose slowly depolarises cells. This difference will affect Ca2+ channel behaviour and altered CaV1.2 function, such as lowered voltage threshold might specifically only be apparent in the glucose responses.

3) The authors present data that islets in pancreatic slices are considerably more sensitive to glucose, inducing a response already at basal glucose levels (2.8 mM). In the same vein the authors observe a considerably shortened delay between stimulus and response (this delay is general due to nutrient metabolism and initial filling of intracellular calcium stores). The authors take these phenomena as evidence for a superior and more physiological quality of their islet slices as compared to conventional purified islets.

However, contrary to their interpretation, these observations considerably questions whether the slice preparation used here in this work has physiological qualities. Indeed, the authors observe considerable activity of islet beta-cells already far below the set-point of around 6 or 7 mM in rodents, very well characterized through a number of studies in-vivo, in-vitro and even in-situ (10.1113/jphysiol.1995.sp020804), and their preparations reach almost full activity around the set-point. This is also surprising as such a hypersensitivity has not been reported by several other groups using the same preparation, i.e. pancreatic slices (10.1152/ajpendo.00043.2021; 10.1371/journal.pone.0054638; 10.3389/fphys.2019.00869; 10.1371/journal.pcbi.1009002; 10.1038/nprot.2014.195) even using patch clamp (10.3390/s151127393). >Moreover, even human islets, known for a lower set-point, are inactive in slices at 3 mM (10.1038/s41467-020-17040-8) in line with the physiological requirement to avoid insulin secretion in low glucose states as to avoid life-threatening hypoglycaemia. The same applies for the shortened delay between application of a stimulus (glucose) and start of the response, which has also not been observed by other groups in pancreatic slices (refs see above).

We are cognisant that our data challenges the dogma and talked around this point in the discussion. Evidence that our findings might be correct include the responses seen by Henquin to glucose concentrations below 6 mM (Gembal et al 1992) and the long-standing evidence of heterogeneous responses in isolated cells that show responses to very low glucose concentrations (Van Schravendijk et al 1992). As such, our data is not as unusual as it might initially appear. Furthermore, as discussed in detail below the findings from others using the slice preparation is not directly or easily compared to our work.

In general, such an increased glucose sensitivity is observed in prediabetic states or experiments mimicking such a condition. To the best of my recollection such an apparently increased sensitivity can also be observed in brain slices due to leakage. Unfortunately, no independent measures of islet quality in slices are provided.

We have previously characterised increased insulin secretion in “prediabetes” in mice and demonstrated a clear effect on the mechanisms of granule fusion such as an increase in compound exocytosis (Do et al 2016). We do not think this is relevant to this slice preparation where normal mice were used for both the slice and the islet experiments and our data in slices and islets both show normal granule fusion and not compound exocytosis.

Within the same vein the comparison between slices and islets (Fig 5) is not in favour of a more physiological aspect of slices and the different cell morphology and small number of observations shed more doubt, especially in view of the well known normal beta-cell heterogeneity (which may explain differences and may have been missed here due to a small sample size).

We acknowledge that beta cell heterogeneity is a potential confounding factor. However, our sample sizes are not small, in each islet or slice we record Ca2+ responses from ~10 cells (see Fig 3) and have repeated preparations from each mouse with the total dataset from >3 mice. It is true that the sample size for Ca2+ waves is small for the isolated islets, but this is because these are such rare events which is explained by the fragmented capillaries and compromised cell structure (eg Fig 1) in isolated islets.

In a larger context this glucose supersensitivity may also shed doubts on the proposed important role of FAK as its role may be far less preponderant in preparations corresponding to physiological criteria.

We agree that the relative importance of FAK might be different in different in vitro models. But it is clear that FAK plays an important role in vivo and the data from FAK KO mice show both defective glucose homeostasis and lower insulin secretion (Cai et al 2012) directly demonstrating physiological relevance.

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

Summary: This manuscript documents a very thorough biophysical, structural and functional dissection of interactions between the RNA-binding protein Rrm4 and the endosomal adaptor Upa1 in the filamentous fungus Ustilago maydis. It has been shown previously that the Rrm4-Upa1 interaction is critical for mRNA transport in this system as mRNAs hitchhike on motor-associated endosomes. Here, the authors reveal using modelling that Rrm4 has three MLLE domains, including a cryptic one that had not been identified previously. They then report the crystal structure of MLLE2 and analyze the distribution anf arrangement of the MLLE domains in the protein using SAXS. They then show using pulldowns and isothermal titration calorimetry that MLLE3 is critical for the Upa1 interaction (via the PAM2L domains of Upa1) and that MLLE2 contributes to Rrm4 localization in vivo when the MLLE3-Upa1 interaction is partially impaired. The study suggests that Rrm4 has a platform of MLLE domains for orchestrating Rrm4 function. Overall, this is technically a high quality study. However, a number of points (mostly minor) should be addressed.

Major comments:

__A key part of the study if the in vivo work illustrating a role for MLLE2 in regulating Rrm4 localization when the system is sensitized. Some aspects of this part of the work need clarifying.

a) The authors should show that the abberant staining is indeed microtubule-related with the benomyl experiment that they used in Jankowski et al. 2019. __

We included this important control in Figure EV5F demonstrating that the aberrant staining is no longer visible after the microtubule inhibitor benomyl treatment

b) The authors claim from these experiments that MLLE2 contributes to endosomal targeting (as there is ectopic protein on other structures (presumptive microtubules)). However, to make this claim, the authors would need to measure the intensity of the mutant Rrm4 protein on endosomes and/or the colocalization of these Rrm4 variants with endosomes, as they do in other experiments in this paper. Otherwise, it is possible that the MLLE2 deletion has another effect, e.g. increasing protein stability, and thus increasing the likelihood of binding to structures other than endosomes. If available, data on the relative abundance in the cell of the protein expressed from the wild-type control (rrm4-kat) and MLLE2 deletion constructs (e.g. rrm4-m1,2delta-kat) should be provided.

As indicated by the reviewer, a critical point is identifying a function of MLLE2. Surprisingly, the domain is conserved in evolution, but , we do not see a mutant phenotype under optimal culture conditions. Therefore, we challenged the system and observed the mislocalisation of Rrm4, if the MLLE2 domain is deleted. However, the overall amount of shuttling Rrm4-positive endosomes was not strongly affected according to our kymograph experiments. We observe aberrant staining, which is not seen with the Rrm4 wild-type protein. Thus, under challenging conditions, we do see a function of MLLE2.

To address the valid point of the reviewer, we quantified the signal intensities in kymographs of the most important Rrm4 variants. As indicated in Figure 5E, we observed that the maximum fluorescence intensity in kymograph signals was reduced when Rrm4 variants are mislocalised to microtubules while the minimum intensities were comparable in all strains. This underlines that a subset of Rrm4 molecules are no longer shuttling through the cell and most likely are attached to microtubules (to prove the involvement of microtubules, we did benomyl treatment which is now shown in Figure EV5F). We also included a Western Blot experiment (Figure EV5G) demonstrating that neither MLLE1 nor MLLE2 deletion impacts the total protein amount of Rrm4. These data support the notion that MLLE2 contributes to endosomal targeting.

c) Was the data in Figure 5D scored blind of the identity of the samples? Given that the classification has to be done manually, it is important to confirm the phenotypes are robust to blinding (at least for the key comparisons).

We agree entirely that manual evaluation of microscopic images has to be carried out with utmost care. The phenotype of aberrant microtubule staining is not easily detectable, and it needs an experienced person to quantify this. The data were analyzed by a second experimentalist with experience in evaluating microscopy images to validate the system’s robustness. Notably, the key findings were confirmed in both cases aberrant microtubule staining was only observed when the MLLE domain was mutated. However, the second person reported difficulties in differentiating a bundle of Rrm4 signals or stained microtubules. Therefore, this person quantified higher values with less experience in Rrm4 movement. In essence, we can rely on the key findings. We included the information in the section “Materials and methods” and gave the comparison in Figure EV5H.

If points b and c are addressed, it should be possible to draw an arrow between the gray question mark protein in Figure 6 and the endosome surface, which is what I assume the authors believe to be case based on their discussion.

Having addressed both points, we have also improved the model. To this end, we added a second unknown protein component (grey oval with a question mark) that interacts with MLLE2 and the endosomal surface. Thereby the hierarchical order with the accessory role of MLLE2 during endosomal attachment is stressed.

Minor comments:

1. The first line of the abstract is quite bold. It is hard to quantify the role of transport vs RNA stability for example, so I suggest this sentence is toned down. Correct, the first line now reads, “Spatiotemporal expression can be achieved by transport and translation of mRNAs at defined subcellular sites”.

Line 269: change "amount of motile Rrm4-M12delta-Kat positive signals" to "number of motile Rrm4-M12delta-Kat positive signals".

Changed as mentioned above.

Figure 3 legend: Insert "Variant" before "amino acids of the FxP and FxxP..." to indicate what is labeled in gray. Change "fond" to "font" in the same sentence.

Corrected as mentioned above.

The cartoons of the different protein variants are very helpful but I had problems spotting the Upa1-Pam2L deletions due to the similar gray to the background of the protein. This would perhaps be clearer if the gray used for the background was lighter than it currently is.

We improved the contrast by reducing the background of Upa1 to a lighter grey tone in all the corresponding figures.

The residual motility of wild-type Rrm4 when PAM2L1 and PAM2L2 are both mutated (Figure 5C) is reminiscent of what is seen in a complete Upa1 deletion in the group's previous work. It would be helpful to point this out to the reader, as well as the implication that other proteins are contributing to Rrm4's linkage to endosomes. After all, some of these other adaptors might contact MLLE2 of Rrm4.

We addressed this point by referring to our previous publication with the following sentence: “Comparable to previous reports, we observed residual motility of Rrm4-Kat on shuttling the endosomes if both PAM2L motifs are mutated or if upa1 is deleted. This indicates that additional proteins besides Upa1 are involved in the endosomal attachment of Rrm4 (Pohlmann et al., 2015).”

Some of the y-axes of the charts should be more descriptive so that the reader can understand the plots even before they consult the legends. For example, in Figure EV4A and EV5D and E, which protein is being to referred to in each 'number of signals' plot should be included. In Figure 5D, 'Hyphae [%]' would be clearer as 'Hyphae with MT staining of Rrm4 [%]'

We improved this in Figures EV4, 5D and EV5.

Figure EV5 legend title: this could be misleading as the authors are seeing ectopic MT localization rather than a deficit in microtubule association.

Corrected to “Deletion of MLLE1Rrm4 and -2 cause aberrant staining of microtubules”.

Reviewer #1 (Significance (Required)):

__The Feldbrugge group has previously mapped interactions between Upa1 and Rrm4 (Pohlmann et al., 2015) and some conclusions are corroborated in the paper by Boehm et al. The paper under review is, however, a significant advance due to the identification of the third MLLE domain, detailed biophysical characterization of the interactions, the structural insights, and evidence of a subsidiary role of MLLE2. The work would of course be stronger if the target of MLLE2 had been identified but I think this is beyond the scope of this initial work. To my knowledge, this is one of the most extensive analyses of the interactions mediated by MLLE and PAM domains and will be of interest to others working on these protein features. The work will also appeal to those interested in the links of localizing mRNAs with motor-associated membranes, which is an emerging field.

Reviewer expertise: I have a long-standing interest in molecular analysis of mRNA trafficking mechanisms. I do not have experience in fungal genetics. __

**Referee Cross-commenting**

It seems that we are in agreement that this is solid work and that biochemical and biophysical analysis of the MLLE-PAM interactions will be of significant interest to those working on those domains (or proteins containing those domains). I agree with the comments of the other reviewers and there are clearly some essential minor revisions needed to strengthen the evidence for their conclusions and some clarifications. I think it is a long shot that RNA binding to the RRMs will affect the MLLE-PAM interactions and would require quite a lot of work to show this conclusively. The study would, however, be more impactful if this was shown to be the case, or the target of MLLE2 was found. Nonetheless, I would not say these new avenues of research are necessary to find a home in one of the Review Commons journals.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

Devan, Schott-Verdugo et al.

Summary

In this study the putative MLLE RNA-binding motifs of the endosomal RNA-binding protein, Rrm4, from Ustilago maydis were examined using structural and genetic analyses. MLLE motifs are conserved in polyA-binding proteins (Pab1/PABPC1) and found also in Rrm4, which was shown to reside on motile endosomes and deliver septin mRNAs for endosome-localized translation during polarized growth. Upa1 on the endosome interacts with Rrm4 via its PAM2L domain that itself interacts with the MLLE domains of proteins like Pab1. Mutations in the known MLLE domain of Rrm4 were earlier shown to affect localization to endosomes. Here, the C-terminal domain of Rrm4 was revealed to have three divergent MLLE motifs using comparative modeling; only two of which were previously predicted. Crystallization and X-ray diffraction analysis of a truncated version of bacterially produced Rrm4, showed MLLE2 is most similar to that of PABPC1 and UBR5, although MLLE1 and 2 are somewhat divergent in the key region of PAM2 binding. Small angle X-ray scattering of recombinant full-length or truncated Rrm4 revealed that the MLLE domains might form a platform that could allow for multiple contacts with different binding partners. In vitro binding studies with different N-terminal GST-tagged versions of the Rrm4 were used to examine for interactions with PAM2 sequences of Upa1 using N-terminal hexa-histidine-SUMO fusions. It was found that Pab1-MLLE interacts with the PAM2, but not PAM2L, domain of Upa1. In contrast, the complete Rrm4 MLLE region (G-Rrm4-NT4) interacted with the PAM2L domain, but not the PAM2 of Upa1. Notably, the interaction with PAM2L required the third MLLE and neither MLLE1 nor MLLE2, nor both. No significant differences in affinity were observed and were similar to that of the Pab1 MLLE. The results also show that the MLLE3 has a higher affinity for the PAM2L2 than PAM2L1 of Upa1.

To examine the biological role of the Rrm4 MLLEs, U. maydis strains bearing deletions in the domains of Rrm4 were examined for hyphal growth and endosomal transport (latter using Upa1-GFP and Rrm4-mKate2). Only the loss of the MLLE3 domain inhibited polarized growth (as seen with the full deletion of RRM4) and not the deletion of either MLLE1 or 2. Similar results were obtained regarding endosome shuttling. Thus, in line with the biochemical experiments performed the MLLE3 domain alone (of the three identified) is necessary for the biological actions of Rrm4. This suggested the MLLE1 and 2 are not necessary for function under these conditions.

To examine this further, Upa1 carrying mutations in the PAM2L 1or PAM2L2 domains were examined. It was found that the deletion of both PAM2L domains affected unipolar growth resulting in bipolar growth similar to the deletion of UPA1 alone. This phenotype was observed even upon the deletion of Rrm4 MLLE1 and 2 in the same background as the PAM2L mutants. The mutation of both PAM2L domains led to a reduction in Rrm4-labeled shuttling endosomes, which suggests that these domains help anchor Rrm4 to endosomes. When only the PAM2L1 domain is present in Upa1 there was a larger increase in hyphae with aberrant microtubule staining than upon the loss of PAM2L1. The authors suggest that this indicates PAM2L2 is more important and prescribes an accessory role for MLLE2 in endosome association.

Comments: Overall, the study seems well conducted. We cannot comment on the structural aspect of the work since this is not our field of expertise. That said, the biochemical and genetic/functional studies appear solid, well thought-out, and clearly presented. No new experiments are necessary to support the general claims of the paper, however, experiments suggested below might make it more revealing with regards to the connection between RNA binding and MLLE-PAM2L interactions (i.e. endosome localization and RNA binding functions).

1. Line 286 - It reads the they "Next, we investigated the association of Rrm4 -M12D-Kat in strains expressing PAM2L1. Thus, the endosomal attachment was solely dependent on the interaction of MLLE3 with the PAM2L2 sequence of Upa1." Unclear - wouldn't lacking PAM2L1 (and not expressing) fit the logic of the sentence? We corrected this with the sentence, “Next, we investigated the association of Rrm4-M1,2D-Kat in strains expressing Upa1 with mutated PAM2L1”.

Several questions regarding the specificity of PAM2 vs. PAM2L domains. What happens when you switch/replace the PAM2L1 or 2 of Upa1 with Upa1 PAM2 domains? Are they exclusive? What happens when the MLLE3 of Rrm4 is switched with that of Pab1? And if one does both - does that restore functionality to Rrm4?

These are very interesting suggestions. Previously, we have shown that a single PAM2L1 or PAM2L2 sequence of Upa1 is sufficient for unipolar growth and recruitment of Rrm4 to endosomes. Please note that Upa1 with mutated PAM2L1 and L2 still contains a PAM2 motif. Furthermore, mutating the PAM2 motif of Upa1 did not affect Rrm4 shuttling or unipolar growth. Thus, switching the domains would mostly address whether the precise location within Upa1 would be important. This is interesting but, unfortunately very labour-intensive and beyond the manuscript’s current scope.

Switching MLLE3 with MLLE of PAB1 is an interesting approach. One might expect that Rrm4 can be recruited to endosomes again. However, Rrm4 would also interact with numerous other proteins containing PAM2 motifs like deadenylase Not4. Here it would compete with the MLLE of Pab1. Thus, it would be expected that Rrm4 is on the surface, but the protein will be mistargeted to other proteins causing pleiotropic alterations. It will be difficult to judge whether Rrm4 functionality is restored or whether other processes are disturbed. In essence, these are stimulating ideas, but we believe that these experiments are beyond the scope of the current study. In the future, we might address this point by using a heterologous peptide-binding pocket or tethering approach.

Likewise, what happens if Upa1 only has PAM2L2 instead of only PAM2L1 domains? Does that alter function - perhaps now one can observe a contribution of MLLE1? If it it's there it's likely to have function. Anything known about the post-translational modification of these MLLE or PAM domains? Does it change during unipolar vs. bipolar growth? Perhaps the different MLLE domains are regulated in such a fashion?

Again also very valid points. Upa1 with two PAM2L2 motifs might interact stronger. The problem is that one PAM2L motif is sufficient for interaction, and we do not see a strong phenotype.

Currently, we do not know if post-translational modifications regulate the MLLE domains. This could alter the binding affinity or specificity, and by expressing fungal proteins in E. coli, we might have missed this type of regulation. However, we addressed the function of MLLE1 and MLLE2 in U. maydis using a genetic approach. We deleted the corresponding domains and interfered with potential regulation by posttranslational modification. Thus, we cannot exclude post-translational modification, but it appears to be not essential for function. We will address the posttranslational regulation of Rrm4 in more detail in the future.

Can the authors show whether the binding of mRNA cargo (e.g. Cdc3 mRNA) to the RRM motifs of Rrm4 affects the interaction between any of the MLLE-PAM2L pairs, or vice versa (i.e. does the MLLE-PAM2L interaction affect mRNA binding)?

In previous studies, we have investigated a version of Rrm4 carrying a mutation in the first RRM motif of Rrm4. According to RNA live imaging, the respective strains exhibit a loss of function phenotype and mRNA transport is strongly affected. However, the endosomal association of Rrm4-mR1-Gfp is not affected, indicating no direct cross-talk between RNA-binding via RRM1 and endosomal attachment via MLLE3. Also, a version of Rrm4 carrying a deletion of all three RRM domains is still shuttling on endosomes. The two functions, i.e. RNA binding and endosomal binding, appears to be carried out by two independent platforms, i.e. three RRMs and three MLLEs, respectively. The overall structure of the protein also reflects this. The RRM domains are structurally clearly separated from the flexible MLLE domains.

Discussion line 311 It is written that the three MLLE domains "collaborate for optimal functionality..." Perhaps there's a misunderstanding here, but the authors show that MLLE3 domain alone is necessary & sufficient for function, so where is the collaboration? MLLE2 may have an accessory role according to the authors, but we do not know if it is in collaboration with MLLE3 or independent thereof. Since the KD of MLLE3 is not affected by the presence or absence of MLLE1,2 in vitro at least, it may be that they have independent, and not collaborative, roles.

Correct, we rephrased this more carefully. We omitted the collaboration aspect. It now reads, ”but a sophisticated binding platform consisting of three MLLE domains with MLLE2 and MLLE3 functioning in linking the key RNA transporter to endosomes.”

Reviewer #2 (Significance (Required)):

This paper concerns functional domains found in an endosome-localized RNA binding protein, U. maydis Rrm4, which is necessary for localized translation on endosomes and subsequent unipolar growth. Here the authors show using structural, biochemical, and genetic studies that instead of one or two MLLE protein-protein interacting domain in Rrm4 there are three, although one (MLLE3) is necessary and sufficient for full function. This work is for an audience interested in those studying RNA trafficking and its role in cell physiology, which is our expertise. The work is interesting, but it could be made more so especially if a connection was established between the RNA-binding function of the RRM domains and the MLLE-PAM2L interaction(s). At this point it is solid technical work and could be published after minor revisions.

**Referee Cross-commenting**

I concur with the comments of the other reviewers in that the work is solid and necessitates minor revisions in order to be published. Clearly, establishing a connection between the RNA-binding function and the MLLE-PAM interactions of Rrm4 would be an interesting and worthy pursuit that might enhance the novelty of the work, but I agree that it could belong to future studies.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

__ Summary: Long-distance subcellular transport of mRNAs is achieved through selective and dynamic interaction with the transport machinery. Using the highly polarized hyphae of Ustilago maydis, the authors previously showed i- that mRNAs can hitchhike on actively transported endosomes for proper distribution, and ii- that the connection between mRNAs and endosomes is mediated by the interaction between a C-terminal MademoiseLLE (MLE) domain of the RNA binding protein Rrm4 and the Upa1 adapter protein. In this study, the authors aimed at more precisely characterizing the structural and molecular bases underlying the Rrm4-Upa1 interaction. Combining structural modeling and X-ray analyses, they discovered a non-canonical, and previously missed, MLE domain (MLE1) in Rrm4, and characterized the structure of the second MLE domains (MLE2) of Rrm4. Through binding assays, they showed that the three MLE domains exhibit different binding properties, and that MLE3 is the only domain capable of binding to the PAM2 domain of Upa1. Consistent with this finding, functional assays performed in U. maydis revealed that MLE3 is the main domain involved in interaction with endosomes and trafficking, MLE1 and 2 having either no or minor functions in this process.

The manuscript is very-well written, the data are of high quality and clearly presented. A wide range of complementary approaches has been used to molecularly and functionally characterize the different MLE domains of Rrm4. From an "RNA transport" perspective, this manuscript falls short of a main novel findings as the domains characterized in this study (MLE1 and 2) don't have a clear function in connecting mRNAs to the transport machinery. From an "MLE domain" perspective, this work however provides interesting information about non-canonical domains and structures, and about binding and function specificity. As described below, my major concern relates to the role played by the ML2 domain of Rrm4, a role referred to as "accessory" by the authors. __

__

Major comments: __

The authors conclude from their results that ML2 has an accessory role in promoting association with endosomes.

1- This conclusion is made based on in vivo experiments showing that a form of Rrm4 lacking the M2 domain, in contrast to wild-type Rrm4, aberrantly attached to MTs in a context where the Rrm4-Upa1 interaction mediated by MLE3Rrm4 has been weakened (Upa1-pl2m). Although the results are convincing, their interpretation is less. The authors, indeed, claim that the observed phenotype results from "the static accumulation of Rrm4" due to reduced interaction with endosomes. Why then don't they see a decrease in the motility/transport properties of Rrm4-M2Δ in this context then? Also, do the authors see a decrease in the co-localization of Rrm4-M2Δ with endosomes (which would be expected if the interaction is decreased)? Can the authors perform IP or co-sedimentation experiments to strengthen their hypothesis?

This is a fair criticism that was also raised by reviewer 1. In the improved version of the manuscript, we now include important control experiments demonstrating that (i) the aberrant localisation is microtubule-dependent (Fig. EV5F) (ii) the mutations do not cause differences in protein amounts of Rrm4 (Fig. EV5G) (iii) the key findings of the aberrant microtubule staining, which were scored manually in microscopic images were verified independently by two persons (Fig. EV5H) and (iv) most importantly, Rrm4 signal intensity is decreased in processive signals of our kymograph analysis (Fig. 5E). We firmly believe that this set of experiments strengthens our conclusion that MLLE2 plays an accessory role in the endosomal attachment (Fig. 6).

2- Whether MLE2Rrm4 mediates interaction with endosomes through association with Upa1 is unclear, as the binding assays performed in Figure 3 test for association of Rrm4 variants with single isolated domains of Upa1, not with the full-length protein. Assessing the binding of Rrm4-M2Δ variants with Upa1-PL2m would help interpreting the phenotypes described in Figure 5.

Unfortunately, it is difficult to express full-length Upa1 protein in E. coli due to the presence of extended unstructured regions. To overcome this limitation, we performed yeast two-hybrid experiments with full-length proteins of Rrm4 and Upa1. We were able to recapitulate qualitatively the results observed in vitro using the individual domains.

Notably, the Rrm4 version carrying a deletion in MLLE1 and MLLE2 interacted with Upa1 versions carrying mutations in PAM2L1 or PAM2L2 (Fig. EV3C), suggesting that both MLLE domains of Rrm4 are dispensable for interaction with Upa1. MLLE3 is sufficient to interact with a single PAM2L sequence of Upa1. This suggests the presence of additional interaction partners for MLLE1 and MLLE2 and is entirely consistent with our genetic and cell biological analysis described in Fig. 5.

__

Minor comments: __

1- The authors have previously characterized the effect of a C-terminal deletion of Rrm4 on Rrm4 motility and binding to Upa1 (Becht et al., 2006; Pohlmann et al., 2015). How their previously-described construct compares to the Rrm4-M3Δ used in this study is unclear (is it the same?).

It is the identical mutation to allele rrm4GPD from Becht et al. 2006. We indicate the information in the text “(Fig. 4B-C; mutation identical to allele rrm4GPD in Becht et al., 2006).”

2- page 6, line 141: refer to Fig. 1B rather than Fig. EV1A ?

We included the reference to Fig. 1B.

3- page 10, line 274: "Rrm4-Kat was found"

We corrected this.

4- page 11, line 286: "in strains expressing Upa1-PAM2L1", replace by "in strains expressing Upa1 with mutated PAM2L1"?

We corrected this.

5- The Figures and accompanying legends are overall very clear and detailed. In Figures EV4A and EV5D-E, it would however help if the authors would indicate on the Figure itself, left to each panel which markers/signals is being analyzed (e.g Rrm4-Kat (top) and Upa1-GFP (down) for Figure EV4).

We clarified this.

Reviewer #3 (Significance (Required)):

Active transport of mRNAs along microtubule tracks has been shown to play a key role in the spatio-temporal control of gene expression in various cell types and species. How specific mRNAs mechanistically connect to molecular motors for their transport to their subcellular destination has however for long remained largely unclear. Recent work, including work from the authors, has uncovered that RNAs can hitchhike on membranous organelles through adapter proteins linking mRNAs and RNA binding proteins with trafficking membrane-bound organelles.

This study aimed at investigating the structural and molecular bases underlying the interaction between RNA binding proteins and endosomes. While their identification and characterization of the MLE1 and MLE2 domains of Rrm4 did not provide significant new insight into the mechanisms involved in the endosome-mediated transport of mRNAs, it uncovered interesting new properties of MLE domains, including structural variations, selective binding and functional specificity. This work should thus be of interest for structural biologists and researchers interested in protein-protein interaction platforms.

**Referee Cross-commenting**

Our comments all converge to the idea that this study is solid as it is and requires only minor revision work to support the authors conclusions. Although characterizing further MLE/PAM2 binding specificity and MLE2 interactors would be of great interest and indeed provide a more complete understanding of interaction networks at play, I feel that this is beyond expected revision work.

Author Response:

Reviewer #1:

Hu and colleagues employ computed-tomography methods and provide a detailed description of and inferences about the dental system in three early-diverging ceratopsian dinosaur genera represented by rare specimens from China. Their study identifies nuanced tooth replacement rates and patterns. Furthermore, combined with the analysis of dental wear patterns, their study not only elucidates ontogenetic aspects of these early ceratopsians but also explores the implication of such patterns for dietary adaptations among these taxa. The manuscript, therefore, provides unique insights into the anatomical and ecological contexts of ceratopsians in such deep time.

The manuscript is rich in data that are summarized in multiple tables and figures. It is also well-written and easy to follow. The inference and conclusions made are also overall well supported by the data presented.

Thank you for your positive comments!

The only main comment I have concerns the inference made about the dietary adaptation of Yinlong, which is inferred to be characterized by "feeding strategies other than only grinding food with their teeth." I think that this could be expanded a bit more to incorporate dietary breadth as an additional possible explanation, particularly given the lack of conclusive evidence for the predominance of a single plant species. As it stands, the inference (made across lines 475 through 485) may only imply processing the same food resource using non-chewing methods (e.g., gastroliths to triturate fern). Could the incorporation of other, less abrasive plat foods--in addition to the fibrous ferns--in the diet of Yinlong be a possible, additional explanation for the relatively slow tooth replacement and lack of a heavy tooth wear from chewing-related stress?

We have provided more explanations and discussion for feeding strategies based on analysing the environmental condition and internal features. Firstly, we analyzed the flora of the Shishugou Formation and the environment that Yinlong lived. Then its feeding strategy can be inferred from its body size and tooth characters. The relatively small body length implies that Yinlong likely feeds on some low plants. The morphology of dentitions, the primitive jaw morphology, and the low tooth replacement rate suggest that Yinlong is unlikely to grind tough foods like derived ceratopsians. Yinlong possibly has other feeding strategies such as processing the foodstuffs by gastroliths, which have been found in some other dinosaurs. We have added more comparison with other dinosaurs (i.e., an armoured dinosaur preserved stomach contents and gastroliths). We suggest that ferns such as Angiopteris, Osmunda, and Coniopteris are suitable to be food choices of Yinlong. Some low and tender leaf and other less abrasive plant foods could also be possible.

Reviewer #2:

The authors of the present work aimed to describe tooth replacement in early ceratopsian species from the Lower Jurassic of China, and with this novel information, discuss new hypotheses of successive changes in jaw evolution that led to the highly specialized replacement and jaw function of derived ceratopsids. Major strengths of this study include not only the use of microCT-scans and 3D reconstructions to address tooth replacement in three different species of early ceratopsians (Yinlong, Hualianceratops, and Chaoyangsaurus), but also the observation of wear development, pulp cavity development, zahnreihen, and z-spacing and replacement rate to compare between taxa and address the succession of mandibular and replacement changes in the phylogeny of ceratopsian dinosaurs. The aims were achieved and the conclusions are strongly supported by the evidence discussed and the cited bibliography. Figures are clear and captions are concise. The presented information gives evidence for the comparison and discussion of the order of acquisition of different craniomandibular adaptations that lead to a specialized herbivorous diet, useful not only for ceratopsians and ornithischians, but also for other lineages of dinosaurs in the Mesozoic, and further for comparing with extant and extinct lineages of mammals. Dinosaurs not only were fantastic creatures from the past but also achieved different morphologic, physiologic, and behavioral traits unknown to any other creature, even mammals. For ceratopsians, the appearance of dental batteries corresponds to a unique trait only functionally similar to that in hadrosaurs and some sauropods, and understanding the steps that led to that specialized structure allows us to also understand the drivers that later guided their diversification during the Late Cretaceous.

Thank you for your positive comments!

Reviewer #3:

The major strengths of the paper are its thorough level of detail, rich dataset, and easy readability. The figures are excellent and clear.

One shortcoming of the paper is the lack of measurements -- a table of measurement for each functional and replacement tooth's length, mesiodistal width, and linguolabial width should be provided.

We thank the reviewer for pointing out this. We have provided each functional and replacement tooth’s total height, maximum mesiodistal width, maximum labiolingual width of all specimens presented in TABLE S1. These data help to support our conclusions.

Unfortunately the manuscript is not publishable in its current form because the conclusions are not testable based on the limited data provided. The authors stated "All data generated or analysed during this study are included in the manuscript and supporting file." This is not true. Only the 3D models derived from segmentations are provided, not the raw scans. Segmentation-derived models are interpretations, akin to publishing a drawing of a fossil instead of a photograph, which is not generally acceptable under today's publishing standards (drawings can be published alongside photographs). Please upload the raw scans to an appropriate repository such as Morphosource, Dryad, or Morphobank. Scans can be cropped to the dentigerous regions only, so long as scaling information is preserved.

We have added raw micro-CT scans of all scanned specimens (all cropped to the dentigerous regions) in Dryad as .TIF or .BMP file format. The file object details are also provided in a TXT file ‘README_file.txt’ saved in Dryad, at https://doi.org/10.5061/dryad.9ghx3ffk0.

Indeed. The machine learning system will tend to learn the most from people's initial wealth. I think we may combine other facts as input (like education background, occupation etc. ) to the machine learning systems to weaken the effect of initial wealth.

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1: General comments:

Fujimoto and collaborators use Nanopore-based cDNA sequencing for genome-wide transcriptome analysis of a collection of hepatocellular carcinomas (HCCs) and matched normal liver tissues. To improve detection of alternatively spliced isoforms and hybrid transcripts potentially deriving from genomic rearrangements, they develop a dedicated pipeline SPLICE, which they benchmark against available software used for the same analysis. Besides having dual functionality (calls both alternative transcripts and fused transcripts), SPLICE seems to outperform previous software in calling alternative/fused transcripts and accuracy. They use the SPLICE pipeline to call isoforms and gene fusions in normal liver cells and HCCs and perform basic functional validations on novel fusions identified. The manuscript is well written, and the analyses are well performed. Perhaps the benchmarking of the SPLICE pipeline could have been more extensive (i.e., performed on additional independent datasets).

Major points: 1. Line 149-150: "We compared the results of mapping to the reference genome and the reference transcriptome sequences, and removed candidates if both were inconsistent (removal of mapping errors). " Please specify what "both were inconsistent" means.

Our reply; Thank you for this comment. The accuracy of fusion gene detection is influenced by mapping errors. To remove possible mapping errors, SPLICE aligned reads to the reference genome and the reference transcriptome sequences and compared the results. If the results are inconsistent (for example, GeneA-GeneB in the reference genome and GeneA-GeneB in the transcriptome genome, or GeneA-GeneB in the reference genome and GeneA in the transcriptome genome), SPLICE considers the candidates as false positive and removes them from the analysis.

          We changed the sentence “We compared the results of mapping to the reference genome and the reference transcriptome sequences, and removed candidates if both were inconsistent (removal of mapping errors).” to “we compared the results of mapping to the reference genome and the reference transcriptome sequences, and removed candidates if both results did not detect same fusion genes (removal of mapping errors).”  (line 150-152).

• Concerning TE-derived novel exons, in principle, this may lead to altered expression of the TE-transcript (as the Authors report for L1-MET) or to altered splicing of the transcript (i.e., other exon/introns could be retained or excluded). Can the Authors assess whether the inclusion of the TE in a transcript enhances its expression or affects the splicing of the "parental" transcript? If so, can they verify if the position of the insertion of the TE has any effect on expression and splicing?*

Our reply; Thank you very much for this important comment. As the reviewer mentioned, exonization of TE may affect the splicing patterns and gene expression levels of transcripts. To determine the effect of TE on expression levels, we compared the expression levels of transcripts with TE-derived novel exons with those of known transcripts of the gene. We found that the expression levels of transcripts with TE-derived novel exon were lower than those of known transcripts (Figure 1 in the reply). Since the same results were observed in all novel transcripts (Fig. 1E,F), most TE exonization would not affect the expression level of transcripts.

          We then analyzed the effects of TE in the splicing change, we compared the numbers of novel splicing junctions between transcripts with TE-derived novel exons and other transcripts in each gene. The proportions of genes with novel splicing junctions were not significantly different between the transcripts with TE-derived novel exons and others (transcripts with TE-derived novel exons; 9.1% and others; 11.9%)  (Figure 2 in the reply). As observed in L1-*MET* and L2-*RHR1*, transposons can affect expression levels and structures of transcripts, however, their effect would be limited to a part of genes.

Figure 1

Comparison of expression levels of transcripts with TE-derived novel exon and known transcripts. Only transcripts derived from genes with TE-derived novel exons were compared. The total number of transcripts is shown below the plot. Transcript abundance was measured in reads per million reads (RPM), and log10 converted values for RPM were shown in the violinplot. P-values were calculated by Wilcoxon rank-sum test.

Figure 2

Comparison of the percentage of novel splicing junction in transcripts with novel TE-derived exon and other transcripts. The total number of genes are shown below the plot. Transcripts with TE-derived novel exons and other transcripts were compared. P-value was calculated by Fisher’s exact test.

• Can the Authors explain why the NBEAL1-RPL12 was not detected by SPLICE?*

Our reply; Thank you for this comment. Although NBEAL1-RPL12 fusion was detected by SPLICE, mapping results to the reference genome and the reference transcriptome were inconsistent and removed from the final result. AsNBEAL1-RPL12 was not validated by PCR (Supplemental Fig. S4B) (line 183-184), we consider that this fusion-gene is a false positive, and filtering of SPLICE successfully removed false-positive fusions.

• Line 332: Can the Authors explain how the total amount of HVB mRNA was determined in each sample? Is it a relative amount calculated from the sequencing data? If so, it should be made clear in the text that this is a fractional measure.*

Our reply; Thank you very much for this comment. Expression levels were calculated by log10 converted reads per million reads (log10(RPM)) for each sample. We added the following sentences to the "Expression from HBV" subsection in the Results (line 337-338); “Expression levels were estimated by log10 converted support reads per million reads (log10(RPM)) for each sample.”.

• Fig4a: please specify if the y-axis "number of support reads" reports library normalized values.*

Our reply; Thank you for this comment. The values of the y-axis are row read counts. We added the following sentences to the Figure legend (line 348); “Y-axis shows the total number of support reads (raw counts).”.

• HCCs have more HBV-human genome fusion transcripts than normal liver. Could the authors clarify if these HCC transcripts are selectively found in tumors? or whether they are also expressed in normal liver samples? The paragraph starting from line 356 is confusing, and it is difficult to retrieve the above information for both HBs and HBx fusions.*

Our reply; We apologize for the confusing description. All HBV-human genome fusion transcripts were selectively expressed in tumor or normal liver. We added the following sentence to the "Expression from HBV" subsection in the Results (line 365-366); “All of these HBV-human genome fusion transcripts were selectively expressed in the HCCs and the livers.”.

• Figure 4C: what was the control used to calculate the relative viability in these analyses?*

Our reply; Thank you for this comment. Fig. 4C shows the number of HBV-human fusion transcripts in the six categories. If this comment refers to Fig. 4H, cell lines transfected with the empty vector (pIRES2-AcGFP1-Nuc) was used as controls. This has been described in the "Gene overexpression" subsection of Methods (line 716-717).

• MYT1L: the Authors report the identification of a novel MYT1L transcript downregulated in HCC, and argue it may have a potential tumor-suppressive function. For the sake of clarity, it will be advisable to show also the differential expression (HCC vs. Liver) of the other transcripts expressed from the same locus.*

Our reply; Thank you for this important comment. In HCCs and normal livers, only the novel MYT1L transcript was expressed from this locus, and no known transcript of MYT1L was expressed. We changed the sentence “In the MYT1Lgene, a highly-conserved novel exon was detected (Fig. 2E), and this transcript was significantly down-regulated in the HCCs” to “In the MYT1L gene, a highly-conserved novel exon was detected (Fig. 2E), and only a transcript with the novel exon was expressed.” (line 471-472).

• *

Minor points: 1. Table S4: there is a typo, correct “secific” in “specific”

Our reply; Thank you very much for this comment. We corrected the typo of Table S4.

• *

• *

*Reviewer #2: General comments:

Summary: This is both a presentation of a pipeline for analysis of Nanopore RNA-seq data, as well as an analysis of a cohort of 44 hepatocellular carcinomas against matched-normal liver tissue. It presents a number of quite intriguing results from the long-read RNA analysis, and suggests potential new targets for study in HCC. It is also worth noting that the current version of guppy (6) has functionality to detect primer sequences in the middle of reads and split those reads, which may obviate one of the steps in SPLICE.*

*Major comments:

1) The work done in this study used data that was basecalled using guppy 3.0.3. Since that version, I am aware of at least two major upgrades to the base caller accuracy, which would likely also improve the accuracy of isoform resolution. Given that the data is relatively low-coverage and that you have an automated workflow for the analysis, I would recommend re-basecalling using an updated basecaller and re-running your analysis using that. This is especially important given your comments in the paper about splice site misalignment.*

Our reply; Thank you very much for this important comment. We performed basecalling of a sequence data of MCF7 using the latest guppy v6.0.6 and compared the result with that by guppy v3.0.3. We randomly extracted 1M reads from MCF-7 reads that passed qscore filtering in guppy basecaller. The same reads were extracted and basecalled by guppy v3.0.3. These two data were analyzed by SPLICE.

The average error rate was 4.6 % for v6.0.6 and 6.8 % for v3.0.3. The number of transcripts was 9,674 for v6.0.6 and 9,329 for v3.0.3. Of these, the number of novel transcripts was 446 and 410, respectively. The number of fusion genes was 2 (BCAS3-BCAS4, and BCAS3-ATXN7) by v6.0.6 and one (BCAS3-BCAS4) by v3.0.3. As the reviewer mentioned, we found that using the latest version of guppy improved the accuracy and detected a larger number of transcripts.

We added the results to Supplemental Table S12. We also changed the sentences from “Second, our analysis removed the change of splicing sites within 5 bp to remove alignment errors (Fig. 1B). We consider that this cutoff value is necessary due to currently available high-error reads (S____upplemental Data S____2). However, sequencing technologies and basecallers are improving, and in the near future, we should be able to use a smaller cutoff value and identify larger numbers of splicing changes.” to “Second, the accuracy of the analysis depends on the sequencing error rate. Although several filters are used for currently available high-error reads (Fig. 1B and ____Supplemental____ Fig. S1), sequencing errors would affect the accuracy of the result. Sequencing technologies and basecallers are improving, and in the near future, we should be able to identify larger numbers of splicing changes with high accuracy (Supplemental Table S10).” (line 538-542).

2) You have compared your software to another tool for isoform analysis on Nanopore sequencing data, TALON. But a number of other tools exist for this purpose, including stringtie2, flair and bambu. My own testing has shown that stringtie2 outperforms TALON in terms of concordance with Illumina RNA-seq. It is quite important that you perform a complete comparison of your software to the state of the art for this purpose.

Our reply; Thank you very much for this important comment. We compared our tool with four tools (TALON, FLAIR, StringTie, and bambu). For this comparison, we used sequence data of MCF-7 and HCC (RK107C). We randomly extracted 1 M reads from MCF-7 and HCC (RK107C) sequence data using Seqtk (v1.3) (params: sample -s1 1000000). Reads were mapped to the reference genome sequence (hg38) with minimap2 (v2.17) (params: -ax splice --MD), and the output SAM files were converted to BAM files and sorted with samtools (v1.7) (Li et al. 2009).

For benchmarking of TALON (v5.0), we corrected aligned reads with TranscriptClean (v2.0.3) (Wyman and Mortazavi 2018). Next, we ran the talon_label_reads module to flagging reads for internal priming (params: --ar 20). TALON database was initialized by running the talon_initialize_database module (params: --l o --5p 500 --3p 300). Then, we ran the talon module to annotate the reads (params: --cov 0.8 --identity 0.8). To output transcript abundance, we first obtained a whitelist using the talon_filter_transcripts module (params: --maxFracA 0.5 --minCount 5), and then quantified transcripts using the talon_abundance module based on the whitelist. For FLAIR (v1.5), the sorted BAM file was converted to BED12 using bin/bam2Bed12.py. We then corrected misaligned splice sites with the flair-correct module. High-confidence isoforms were defined from the corrected reads using the flair-collapse module (params: -s 3 --generate_map). For benchmarking of StringTie (v2.2.1), Stringtie was performed with input files consisting of long-read alignment and reference annotation (params: -L -c 3). For benchmarking of bambu (v2.0.0), Bambu was performed with input files consisting of long-read alignment, reference annotation and reference genome (hg38) (params: min.readCount = 3). Candidates with low expression levels (support reads As a result, SPLICE identified the third-highest number of transcripts followed by FLAIR and StringTie (Supplemental Fig. S3A). In MCF-7 the concordance rate with IsoSeq MCF-7 transcriptome data was the highest in SPLICE for known transcripts and the second highest in SPLICE for novel transcripts (Supplemental Fig. S3B). These results indicate that SPLICE has sufficient accuracy for analyzing transcript aberrations.

We added the text to the "Comparison of SPLICE method with other tools" subsection of the Results (line 165-177) and the "Benchmarking" subsection of the Methods (line 640-679). We added the results to Supplemental Fig. S3.

3) Likewise, for fusion detection, you compare to LongGF. You should also compare to (and cite) JAFFAL.

Our reply; Thank you very much for this important comment. We compared our tool with the two tools (LongGF and JAFFAL). We used 1 M reads randomly extracted from MCF-7 and HCC (RK107C) sequence data as described above.

          For benchmarking of LongGF (v0.1.2), reads were mapped to the reference genome sequence (hg38) with minimap2 (v2.17) (params: -ax splice --MD), and the output SAM files were converted to BAM files and sorted with samtools (v1.7). We then ran the *longgf* module and obtained the list of fusion genes (params: min-overlap-len 100 bin_size 50 min-map-len 200 pseudogene 0 secondary_alignment 0 min_sup_read 3). For benchmarking of JAFFAL (v2.2), we ran the *JAFFAL.groovy* module with zipped fastq files.

          In this comparison, close gene pairs (We added the text to the "Comparison of SPLICE method with other tools" subsection in the Results (line 178-186) and the "Benchmarking" subsection in the Methods (line 667-679). We showed the results in Supplemental Fig. 4.

4) In terms of the source code, I have questions. Why did you use BASH to run the Python code, instead of making this into a Python package? Why did you not use the functionality already available in BioPython for a number of basic sequence data handling tasks? Why is there not even a single function defined anywhere, let alone classes?

At some level, if it works, it works. But I have serious concerns about the long-term maintainability of the code in its current state.

Our reply; Thank you very much for this critical comment. As the reviewer mentioned, we think it is better to make a python package and use BioPython for maintenance and long-term maintainability of the code. We have been building our analysis pipeline by trial and error, and at this stage, the current scripts are convenient for us (our group may need to learn software development). We provided a Docker package (see the reply to comment 5)), and this would promote usability.

5) Also related to the code, it is generally the standard now to create a BioConda package or Docker container for a bioinformatics package. BioConda has the advantage that the BioContainers project automatically generate Docker and Singularity containers from it. Please provide one of these.

Our reply; Thank you very much for this critical comment. We made a Docker file and provided it from our github page. It is available from the "Installation and usage via Docker" section.

6) There is some quite nice functional validation work done on some of the DE transcripts that would have been hidden in a gene-level analysis. There is also some nice work on detecting HBV fusion genes. These both contain important results which are not mentioned at all in the abstract. I feel like the abstract as it stands is selling the paper short.

Our reply; Thank you very much for this important comment. We added the following sentences to the abstract; “Comparison of expression levels identified 9,933 differentially expressed transcripts (DETs) in 4,744 genes. Interestingly, 746 genes with DETs, including LINE1-MET transcript, were not found by the gene-level analysis. We also found that fusion transcripts of transposable elements and hepatitis B virus (HBV) were overexpressed in HCCs. In vitro experiments on DETs showed that LINE1-MET and HBV-human transposable elements promoted cell growth.”.

7) Fig 5C shows a Venn diagram of fusions detected by short-read vs long-read sequencing, in which there is quite low overlap between these. You make the statement in the paper that "a combination of short- and long-reads can detect more fusion genes". I find it more likely that the short-read ICGC data had much greater depth of coverage than the MinION data you produced, which allowed for the detection of fusions that were expressed at much lower levels. This could be easily tested by downsampling the ICGC data to the same amount of sequence data as was generated on the MinION, and re-creating the Venn diagram with the fusions detected that way.

Our reply; Thank you very much for this very important comment. We compared the amount of data between our long-reads and the previous short-reads. However, the amounts of data were not quite different (Supplemental Fig. S14A). Therefore, differences in depth are not likely to be the cause of the low overlap. We considered that two possibilities could explain the low overlap. First, most of the fusion genes missed by short-read were very low expression levels, less than 1 reads per million reads (RPM) (Supplemental Fig. S14B), therefore, there are many fusion-genes with low expression levels, and they are difficult to be detected. Second, 28.9 % of transcripts in long-reads lacked 5' region (Supplemental Fig. S5 and Supplemental Fig. S14C,D). Therefore fusion-genes whose breakpoints are located in the 5' region were difficult to detect by long-read.

We added the following sentences to the "Fusion genes" subsection in the Results (line 400-405); “We considered that two possibilities could explain the low overlap. Since the most of the fusion genes missed by short-reads had very low expression levels (Supplemental Fig. S14B), many fusion-genes with low expression levels would be missed by a single approach. In addition, 28.9 % of transcripts in long-reads lacked 5' region (Supplemental Fig. S5 and Supplemental Fig. S14C, D). Therefore fusion-genes whose breakpoints are located in the 5' region would be difficult to detect by long-read.”. We also added a figure on the amount of data to Supplemental Information (Supplemental Fig. S14A).

8) Figure 5D is very interesting. What do you conclude from that result? Please comment in the manuscript.

Our reply; Thank you very much for this important comment. We used samples that used for whole-genome sequencing in our previous study. Therefore, a list of SVs is available. We classified fusion-gene to these supported by SVs (SV detected fusion-genes) and others (no SV detected fusion-genes), and compared the expression levels of them (Figure 5D).

Whole-genome sequencing can accurately identify clonal (high frequency) SVs, however, would miss sub-clonal (low frequency) SVs. Therefore, we considered that no SV detected fusion-genes were generated by sub-clonal SVs. This result suggests that there are a lot of sub-clonal fusion genes, and their expression levels are lower than clonal fusion genes. Although the functional importance of sub-clonal fusion genes is currently unknown, deeper RNA sequencing would detect a larger number of fusion genes.

          We added the following sentences to the “Fusion genes” subsection in the Results (line 410-412); “This result suggests that there are a lot of sub-clonal fusion genes, and their expression levels are lower than clonal fusion genes. Although the functional importance of sub-clonal fusion genes is currently unknown, deeper RNA sequencing would detect a larger number of fusion genes.”.

*Minor comments:

1) The manuscript has many small errors in English grammar, spelling and style. I would strongly recommend sending it for copy editing before submitting it to a journal.*

Our reply; Thank you very much for this comment. Due to the limitation of time, the current version has not been proofread by a native-English speaker. We are planning to review English grammar by a native-English speaker.

2) Neither the results section nor the methods section describing the sequencing that was performed specify whether it was done on a MinION or PromethION (or flongle). While this is implied elsewhere in the paper, it should definitely be specified in the methods at a minimum.

Our reply; Thank you for this comment. We used a MinION for sequencing. We added the following sentences to the Method section (line 579-580); “Libraries were sequenced on a SpotON FlowCell MKⅠ(R9.4) (Oxford Nanopore), using the MinION sequencer (Oxford Nanopore)”.

3) You also write in the introduction that your method, SPLICE, was developed for the MinION specifically. Please comment on its applicability to data generated on the PromethION and flongle Nanopore sequencers.

Our reply; Thank you very much for this comment. We consider that our method is applicable to data from MinION, PromethION, and flongle. We added the following sentence to the Methods section (line 592-593); “In the present study, we analyzed sequence data from MinION. We consider that our method is applicable to data from MinION, PromethION, and flongle.”.

4) The volcano plot in Fig 3A is missing its dots.

Our reply; Thank you very much for this comment. We modified the Fig. 3A.

*Reviewer #3: General comments:

Summary: In this manuscript, Kiyose et al have developed and tested a novel methodology for identifying splicing alterations, and fusions, from full-length transcript or long read sequencing data. They apply this approach to liver cancer and paired, non-cancerous liver tissue from a prior publication, and use wet-lab/experimental methods to validate their in silico findings. They conclude that their new methodology, SPLICE, outperforms one existing method, and is uniquely suitable to identifying fusion genes.*

Major Comments: 1) Figure 1B shows a schematic of common error patterns from MinION cDNA sequencing, and the text of the manuscript describes how the authors' new approach (SPLICE), overcomes several of these, e.g. sequencing errors, artificial chimeras, and mapping errors of highly homologous genes. However, there is a fundamental disconnect between the text and the graphic in Figure 1B. This should either be revised for clarity, or an additional graphic or flowchart placed in the supplementary materials to clearly show *how* SPLICE overcomes each of these limitations.

Our reply; We apologize for the insufficient explanation in Figure 1. We showed a detailed explanation of the data analysis procedure in Supplemental Fig. S1.

2) Why was TALON the only alternative approach chosen for validation of SPLICE performance? There are a number of other, more advanced pipelines such as SUPPA2, and IsoformSwitchAnalyzeR. It would strengthen the manuscript, and its conclusions, to incorporate at least one of these methods as a second comparator. This is particularly true for IsoformSwitchAnalyzeR, since Kiyose et al identify a number of differentially expressed transcripts (DETs) for genes that are not differentially expressed.

Our reply; Thank you very much for this important comment. Another reviewer also requested additional benchmarking, therefore we performed an additional performance comparison for the revised manuscript. As SUPPA2 and IsoformSwichAnalyzeR are used to analyze the annotated output GTF files, and direct comparison with SPLICE is difficult. Since IsoformSwichAnalyzeR recommends StringTie as an annotation soft, we compared using StringTie instead.

We compared the performance of SPLICE with that of four other methods (TALON, FLAIR, StringTie and Bambu) for splicing variant detection. SPLICE identified the third-highest number of transcripts followed by FLAIR and StringTie (Supplemental Fig. S3A). In MCF-7 the concordance rate with IsoSeq MCF-7 transcriptome data was the highest in SPLICE for known transcripts and the second highest in SPLICE for novel transcripts (Supplemental Fig. S3B).

We added the text to the "Comparison of SPLICE method with other tools" subsection of the Results (line 165-177) and the "Benchmarking" subsection of the Methods (line 640-665). We added the results to Supplemental Fig. 3.

3) The Venn diagram in Figure 5C appears to show that conventional short read sequencing identifies 46 fusion genes that are not also detected by long read sequencing. However, this result, and its implications are never addressed in the text.

Our reply; Thank you very much for this important comment. We apologize for the insufficient explanation. We considered that two possibilities could explain the low overlap. First, most of the fusion genes missed by short-read were very low expression levels, less than 1 reads per million reads (RPM) (Supplemental Fig. S14B), therefore these are many fusion-gene with low expression level and they are difficult to be detected. Second, 28.9 % of transcripts in long-reads lacked 5' region (Supplemental Fig. S5 and Supplemental Fig. S14C,D). Therefore fusion-genes whose breakpoints are located in the 5' region were difficult to detect by long-read.

          We added the following sentences to the "Fusion genes" subsection in the Results (line 400-405); “We considered that two possibilities could explain the low overlap. The most of the fusion genes missed by short-reads had very low expression levels (Supplemental Fig. S14B). This result suggests that there are many missed fusion-genes with low expression levels. In addition, 28.9 % of transcripts in long-reads lacked 5' region (Supplemental Fig. S5 and Supplemental Fig. S14C, D). Therefore fusion-genes whose breakpoints are located in the 5' region would be difficult to detect by long-read.”. We also added a figure on the amount of data to Supplemental Information (Supplemental Fig. S14A).

Minor Comments: 1) On pages 20-21, the language used to describe the HBV and/or HCV postive vs negative materials is very confusing. Please clarify that by "HBV- and HCV-related tissues" you in fact mean "HBV-and HCV-infected samples."

Our reply; We apologize for the confusing wording. We converted "HBV and HCV-related tissues" to " HBV and HCV-infected samples" in the manuscript.

I believe multilingualism and multiculturalism are what define today's societies, being able to speak more than one language is a need since there is so much language contact around us. I believe that as teachers we have to recognize, respect, and protect the different cultures present in our classroom. Some examples that I can think about are: reading about a legend or myth from different cultures, learning about a holiday from different countries, having students share with each other their country's traditional food, games, music, etc. Finally, I believe CSP practices are about creating a welcoming and safe space for all students.

Considere un dispositivo futuro... en el que un individuo almacene todos sus libros, registros y comunicaciones, y que esté mecanizado para que pueda consultarse con una velocidad y flexibilidad extraordinarias. Es un suplemento íntimo ampliado a su memoria.

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

1. General Statements [optional]

Overall we were elated to have received such positive comments on the manuscript, with requests for only minor changes. We have made all suggested changes to clarify or tone down the language as suggested.

We would like to thank each of the three reviewers for their assessment of our work. We note that all three reviewers agreed the phylogenetic analysis was interesting and convincing. Two of the three reviewers felt the study sufficiently demonstrated roles for Baramicin in the nervous system. We have responded to comments from Reviewer 2 to draw attention to some aspects of the data that they may have been overlooked, which we hope reassures them that our proposal of BaraB and BaraC involvement in the nervous system is robust, coming from different approaches that show consistent results.

Reviewer 1 and Reviewer 3 compliment the study as being very worthwhile, and for suggesting concrete routes for how an AMP evolved non-immune functions. Both compliment its comprehensiveness, and describe the study as having striking findings that should have broad appeal to audiences interested in the crosstalk between the nervous system and the innate immune system.

2. Point-by-point description of the revisions

In the revised manuscript file, we have highlighted all text where changes were made.

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

The authors provide convincing evidence for an evolutionary scenario in which duplications of an AMP gene with ancestral immune function led to paralogs specialist for neural functions. They focus on the Baramicin genes, coding for major Toll signalling targets in the context of antifungal defence. Their study uses infection experiments in several Drosophila species, a careful annotation of the Baramicin genes of D. melanogaster, the demonstration of neural expression of BaraB and BaraC, the KD analysis of Bara B revealing lethality and neurological phenotypes, a reconstruction of the evolutionary history of Baramicn genes in Drosophilids and an analysis of the sequence evolution of the IM24 domain providing the neural functions. In general the paper is well written. There are a few places in the manuscript where the language can be improved and one point, which needs clarification: - ine 297: ...,which did not present with... - line 314/315: ...to just 14% that of...to 63% that of - line 459: ..., we this motif... - line 518: What does "... genomic relatedness (by speciation and locus)..." mean? - line 527/528: ...drive behaviour or disease through interactions... - line 532: ... ancestrally encodes distinct peptides involved with either the nervous system or the immune response... line 535: ...with either the nervous system (IM24) or.... Do the data provide enough evidence suggesting that IM24 had a neural function in the ancestor? Ideally the authors should look at neural expression of the Baramicin gene in the ourgroup, S. lebanonensis. The authors later (line571) admit, that they cannot rule out that IM24 is also antimicrobial.

We thank reviewer #1 for drawing attention to these points. We have made changes to each line to be more concise, clarify our meaning, or fix typos.

Reviewer #1 (Significance (Required)):

This is a very comprehensive study, which, to my knowledge for the first time, suggests concrete routes of how an AMP evolved non-immune functions. One of the striking findings of this paper is that duplications and subsequent truncations of the ancestral Baramicin locus linked to specialisation for neural functions occurred independently in different Drosophila lineages.

We thank reviewer #1 for their very positive comments. We also agree with all suggested changes, including more careful phrasing to emphasize that we have not described a mechanism, just an involvement in the nervous system. For instance, see lines 556-568 are reworked to soften language and explicitly state the ancestral function of IM24 is unknown, and our suggestion that IM24 could underlie Dmel\BaraA interactions with the nervous system is speculation that should be tested.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

Hanson and Lemaitre present a genomic and phylogenetic characterization of the Baramicin family of antimicrobial peptide genes in different species. They discover new Baramicin paralogs, united by the presence of an IM24 domain at the N-terminus. They show that among Baramicins, those that are not inducible by infection (which they improperly call non-immune since a protein can be non-inducible by infection and have very important immune functions), are truncated. They propose that an ancestor peptide with immune functions evolved into a neuronal regulator/effector via truncation.

Although the hypothesis is interesting, the data do not really support it. This manuscript is rather descriptive at this point. The demonstration that IM24 is necessary for neural function is very tenuous. For example, in the paragraphs titled Dmel\BaraB is required in the nervous system during development and Baramicin B plays an important role in the nervous system, I did not find convincing data demonstrating that BaraB is required in the nervous system. The only data that links BaraB to the nervous system is a weak locomotion defect observed in the BaraB mutant. But how many genes, when inactivated, give a locomotion defect? This remains totally unexplained at the molecular level. The authors also mentioned that BaraB is expressed in a subset of mechanosensory neuron cells in the wing. What is the link between this expression and the nubbin phenotype? The authors also mention that data in the literature indicate that BaraC is expressed in glial cells but also in other tissues. Finally, we have no idea what role, if any, these peptides have in the nervous system.

While the characterization of the Baramicin gene family and its evolution across species is convincing, the link between these AMPs and the nervous system is really too preliminary to be convincing. The manuscript would greatly benefit from being more concise.

Reviewer #2 (Significance (Required)):

see above

We thank reviewer #2 for their fair assessment. We have made edits to soften our phrasing, and to emphasize that we have not described a mechanism, just an involvement, in the nervous system.

Examples:

line 270: “integral development role” -> “important for development”

line 277: “Baramicin B plays an important role in the nervous system“ -> “Baramicin B suppression in the nervous system mimics mutant phenotypes”

line 532: “Here we demonstrate that the Baramicin antimicrobial peptide gene of Drosophila ancestrally encodes distinct peptides involved with either the nervous system or the immune response.“ -> “Here we demonstrate that the Baramicin antimicrobial peptide gene of Drosophila ancestrally encodes distinct peptides that may interact with either the nervous system (IM24) or invading pathogens (IM10-like, IM22).”

line 562 new text: “Thus while our results suggest that IM24 of different Baramicin genes might underlie Baramicin interactions with the nervous system, we cannot exclude the possibility that IM24 is also antimicrobial, or even that antimicrobial activity is IM24’s ancestral purpose. Future studies could use tagged IM24 transgenes or synthetic peptides to determine the host binding partner(s) of secreted IM24 from the immune-induced Dmel\BaraA, and/or to see if IM24 binds to microbial membranes.”

We have also changed all instances of “non-immune Baramicins” to “Baramicins lacking immune induction” or something to that effect (e.g. new Lines 25,464, 469,478-82).

We also made some small changes to be more concise (e.g. line 387, 447, cut lines 492-495 from previous version, cut lines 506-507 from previous version).

We have responded below in the reviewer-to-reviewer comments for a few of the specific points raised there, which we hope further assuage some of Reviewer 2’s concerns.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Antimicrobial peptides are main effectors in (insect) immune defenses. It is becoming more and more clear, that AMPs can have pleiotropic effects or even acquire new functions. In the present paper, the authors investigate Baramicin, an antifungal AMP that they described first in publication last year. Here they show that in Drosophila melanogaster Baramicin A, which they described before, has paralogs, that are not immune-inducible. They then show that these paralogs, named BarB and BarC, which are truncated versions of BarA, are expressed in the head and neural tissues. That they have neural functions is supported by targeted gene-silencing experiments. They go on to show, using a comparative approach across Drosophila, that Baramicin A with its antimicrobial function constitutes the ancestral state. Moreover, Baramicin is also enriched in head samples of some of the other Drosophila species they study. This manuscript, which according to the acknowledgements has already been seen by reviewers, is in a very good shape.

I have only a number of minor points, that might help to clarify the presentation.

Lines 34-36: I would delete this sentence and replace it with a statement based on the main findings of the manuscript

We now conclude the abstract with “As many AMP genes encode polypeptides, a full understanding of how immune effectors interact with the nervous system will require consideration of all their peptide products.”

Lines 56-60. May be tone down a bit. Anti-inflammatory activities of AMPs have been known for a long time. I think the next paragraph makes a very good case what is already known and is hence a nice motivation for the current study.

Toned down. This part now reads: “However AMPs and AMP-like genes in many species have recently been implicated in non-immune roles in flies, nematodes, and humans, suggesting non-immune functions might help explain AMP evolutionary patterns.”

Line 125: classical instead of classically

done

Line 200: what is a 'novel' time course? I would just describe what has been done.

Now reads: “We next measured Baramicin expression over development from egg to adult.”

Line 268: hypomorph, I guess in the literature usually hypomorphic is used.

done

Line 279: I would suggest to tone this headline down. This is not a criticism of the paper, but the actual mechanisms of the roles in the nervous system are not studied here.

Done. Now reads: “Baramicin B suppression in the nervous system mimics mutant phenotypes”

Line 505: what does not really become clear is whether IM24 plays an important role in the nervous system of fly species that only have BarA.

Edits from lines 556-568 now help highlight this question.

Line 540-549. This comparison I find a bit far-fetched, or maybe it needs clarification how doublesex expression is related to Baramicins.

Being completely honest: the doublesex discussion was requested during previous review at another journal. We agree that it is a bit of a tangent, and so we have removed these sentences.

Line 584-585. I think that this has been known for much longer from studies in frogs and beetles.

Our use of “in vivo” might have been a bit squishy here. We have edited this to reflect endogenous loss-of-function study, rather than simply “in vivo,” to clarify our intended sentiment.

Reviewer #3 (Significance (Required)):

Overall, I think that this is a very worthwhile and convincing story about the evolution AMPs and how they can acquire new functions. All the main statements are supported by careful experiments and data analysis. The paper does not go into any detail, of how the neurological role of BarB and BarC is achieved, but I think this is beyond the scope of the current manuscript. In short, this is a very worthwhile contribution to the growing literature of the role of AMPs in the nervous system. The authors provide the context of the main published papers in the area in the introduction. As opposed to most papers on this so far, the current manuscript also provides very interesting data on the evolutionary history of the Baramicin genes, both within the main study species, and within other Drosophila species. This paper should appeal to a rather broad audience of researchers interested in innate defenses, AMPs and the crosstalk between the nervous system and the innate immune system.

My background is insect immunology with a focus on AMPs and evolutionary approach.

We thank reviewer #3 for their very positive comments. We agree with all suggested changes.

**Referees cross-commenting**

This session contains the comments of all reviewers

Reviewer 3

Reviewer 2 and I share the view, that the evidence for the effects of BarB and C on the nervous system is rather limited. But I still think, that the paper provides enough new and interesting data that make it a very useful contribution. Though not a neurobiologist, I would assume that providing functional evidence for the role of BarA and B in the nervous system would justify a paper on its own. I agree though, that the relevant sections should be toned down.

Reviewer 2

As I mentioned in my review, I found the genomic and phylogenetic analysis interesting and convincing. I therefore totally agréé with reviewers 2 and 3 on that. Whether BarA and B are playing a role in the nervous system and how it does remain speculative. BaraB mutants show locomotion defects. But mutants in mitochondrial genes have locomotion defects. Can we conclude that mitochondria play a role in the nervous system? If I understand correctly, downregulating Bara in neurons only (With Elav-Gal4 driver) does not show the locomotion phenotype. it induces early lethality. How many genes when inactivated in neurons will give rise to such a phenotype? A lot. I really think that the implication of Bara in the nervous system should be seriously toned done and more presented as an hypothesis than a validated fact.

We would like to note for Reviewer 2 here that it is specifically elav> BaraB-IR that results in lethality, and in weaker gene silencing experiments, adult elav>BaraB-IR flies emerge, and they do suffer locomotor defects. Often, they got stuck in the food shortly after emerging, or would move haphazardly (which was common in flies with nubbin-like wings). We have added explicit mention that elav>BaraB-IR also results in locomotor defects (Line 288-289).

Our private speculation is that the reason flies fail to emerge from their pupae is because they are so uncoordinated that they sometimes cannot wriggle out of the pupal case before their cuticle hardens. In some instances, both using mutants and RNAi, we observed fully developed adults with mature abdominal pigmentation that died trapped inside their pupal cases.

We’d also like to emphasize here that despite testing many other Gal4 drivers, including mef2-Gal4 (muscle/myocytes), nubbin-like wings and lethality were only found using elav-Gal4. A role interacting with mitochondria would likely have been revealed using mef2-Gal4, given the importance of mitochondrial function in muscle.

For BaraC: expression in other tissues (like the rectal pad) could nevertheless be from e.g. nerves innervating the muscles controlling the sphincter. Or it could indeed be entirely unrelated to the nervous system. However we feel the nearly perfect overlap with Repo-expressing cells is a strong argument for a neural role. We also made an effort using RNAi to validate this pattern suggested by scRNAseq, which confirmed a strong knockdown of BaraC-IR with Repo-Gal4 (Fig. 3, Fig. S4).

We hope these comments clarify for Reviewer 2 why we feel confident in proposing a role for Baramicins in the nervous system, even if we do not investigate a mechanism in this study.

Reviewer 1

I agree with reviewer 3 that the main message of the paper providing a concrete scenario of how non-immune functions of AMPs may evolve is an important contribution. A deep investigation of the neural function is definitely going beyond the scope of the paper. Indeed this might be quite tricky. But it would help if the authors could clarify their idea about the ancestral condition. Is there the possibility that IM24 had ancestrally already non-immune function? They are not really clear about this point.

Reviewer 2

I agree with the other reviewers that determining the exact role of Bara peptides could be complicated. I just ask that the authors limit themselves to proposing that the peptides have lost their immune function. I stress that this argument is not very strong. It relies solely on the lack of inducibility of these peptides following infection. I still think that the demonstration of the role of Bara in the nervous system is not provided.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Antimicrobial peptides are main effectors in (insect) immune defenses. It is becoming more and more clear, that AMPs can have pleiotropic effects or even acquire new functions. In the present paper, the authors investigate Baramicin, an antifungal AMP that they described first in publication last year. Here they show that in Drosophila melanogaster Baramicin A, which they described before, has paralogs, that are not immune-inducible. They then show that these paralogs, named BarB and BarC, which are truncated versions of BarA, are expressed in the head and neural tissues. That they have neural functions is supported by targeted gene-silencing experiments. They go on to show, using a comparative approach across Drosophila, that Baramicin A with its antimicrobial function constitutes the ancestral state. Moreover, Baramicin is also enriched in head samples of some of the other Drosophila species they study. This manuscript, which according to the acknowledgements has already been seen by reviewers, is in a very good shape.

I have only a number of minor points, that might help to clarify the presentation.

Lines 34-36: I would delete this sentence and replace it with a statement based on the main findings of the manuscript

Lines 56-60. May be tone down a bit. Anti-inflammatory activities of AMPs have been known for a long time. I think the next paragraph makes a very good case what is already known and is hence a nice motivation for the current study.

Line 125: classical instead of classically

Line 200: what is a 'novel' time course? I would just describe what has been done.

Line 268: hypomorph, I guess in the literature usually hypomorphic is used.

Line 279: I would suggest to tone this headline down. This is not a criticism of the paper, but the actual mechanisms of the roles in the nervous system are not studied here.

Line 505: what does not really become clear is whether IM24 plays an important role in the nervous system of fly species that only have BarA.

Line 540-549. This comparison I find a bit far-fetched, or maybe it needs clarification how doublesex expression is related to Baramicins.

Line 584-585. I think that this has been known for much longer from studies in frogs and beetles.

Significance

Overall, I think that this is a very worthwhile and convincing story about the evolution AMPs and how they can acquire new functions. All the main statements are supported by careful experiments and data analysis. The paper does not go into any detail, of how the neurological role of BarB and BarC is achieved, but I think this is beyond the scope of the current manuscript.

In short, this is a very worthwhile contribution to the growing literature of the role of AMPs in the nervous system. The authors provide the context of the main published papers in the area in the introduction. As opposed to most papers on this so far, the current manuscript also provides very interesting data on the evolutionary history of the Baramicin genes, both within the main study species, and within other Drosophila species.

This paper should appeal to a rather broad audience of researchers interested in innate defenses, AMPs and the crosstalk between the nervous system and the innate immune system.

My background is insect immunology with a focus on AMPs and evolutionary approach.

Referees cross-commenting

This session contains the comments of all reviewers

Reviewer 3

Reviewer 2 and I share the view, that the evidence for the effects of BarB and C on the nervous system is rather limited. But I still think, that the paper provides enough new and interesting data that make it a very useful contribution. Though not a neurobiologist, I would assume that providing functional evidence for the role of BarA and B in the nervous system would justify a paper on its own. I agree though, that the relevant sections should be toned down.

Reviewer 2

As I mentioned in my review, I found the genomic and phylogenetic analysis interesting and convincing. I therefore totally agréé with reviewers 2 and 3 on that. Whether BarA and B are playing a role in the nervous system and how it does remain speculative. BaraB mutants show locomotion defects. But mutants in mitochondrial genes have locomotion defects. Can we conclude that mitochondria play a role in the nervous system? If I understand correctly, downregulating Bara in neurons only (With Elav-Gal4 driver) does not show the locomotion phenotype. it induces early lethality. How many genes when inactivated in neurons will give rise to such a phenotype? A lot. I really think that the implication of Bara in the nervous system should be seriously toned done and more presented as an hypothesis than a validated fact.

Reviewer 1

I agree with reviewer 3 that the main message of the paper providing a concrete scenario of how non-immune functions of AMPs may evolve is an important contribution. A deep investigation of the neural function is definitely going beyond the scope of the paper. Indeed this might be quite tricky. But it would help if the authors could clarify their idea about the ancestral condition. Is there the possibility that IM24 had ancestrally already non-immune function? They are not really clear about this point.

Reviewer 2

I agree with the other reviewers that determining the exact role of Bara peptides could be complicated. I just ask that the authors limit themselves to proposing that the peptides have lost their immune function. I stress that this argument is not very strong. It relies solely on the lack of inducibility of these peptides following infection. I still think that the demonstration of the role of Bara in the nervous system is not provided.

Arguments not only involve evidence for a claim but for limits, counterarguments, and rebuttals.

The Green Knight is informing Gawain that none of the strikes were due to the covenant. Instead, he explains that he pretended to strike Gawain the first two times because "Gawain gave him the gifts he received from the lady" (Sparknotes summary part 4 page 1). He then goes on to say that he hurt Gawain on the third strike because Gawain was dishonest about the girdle from Bertilak's wife. However, the Green Knight adds on that he did not kill Gawain because Gawain valued his life, which the Green Knight understood.

"Sir Gawain and the Green Knight," Sparknotes.com. www.sparknotes.com/lit/gawain/section4/ <accessed 18 May 2022>

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

We thank the reviewers for carefully reading our manuscript. We found their comments to be incredibly thoughtful and constructive and greatly appreciate their feedback. We are confident that addressing the reviewers’ concerns has strengthened our manuscript.

Reviewer #1 (Evidence, reproducibility and clarity (Required)): Camuglia, Chanet and Martin investigate the mechanisms that control cell division orientation in vivo, using the mitotic domains (MDs) in the head of the Drosophila embryo as their main model system. They find that cells in the head mitotic domains rotate and align their spindles within 30 degress of the anterior-posterior axis of the embryo. The Pins protein, implicated in spindle orientation in other systems, is planar polarized in mitotic cells. Pins polarization precedes spindle rotation and is correlated with the division angle (but cell shape is not, violating Hertwig's rule). Overexpression of myristoylated Pins results in uniform Pins distribution on the membrane and affects spindle orientation. alpha-catenin RNAi (but not canoe RNAi) disrupts Pins polarity and spindle orientation in MDs 1, 3 and 5. Low dose CytoD injections (which should disrupt force transmission) also result in defective Pins polarity and spindle orientations. Finally, mechanical isolation by laser ablation also disrupts spindle orienttion. The authors find that preventing mesoderm invagination by snail dsRNA disrupts Pins polarity and spindle orientation in the head. MAJOR 1. Is there a certain chirality in the rotation of the spindles? From Movie 1, it seems like in MDs 1 and 3 at least, a majority of spindles on the right side of the embryo rotate clockwise, while spindles on the left side rotate counter-clockwise? Is that so, and in that case, are there geometric/molecular considerations that could explain that chirality?

We thank the reviewer for pointing this out. They are correct in that there is a tilt to the spindle orientation relative to the AP axis. To illustrate this tilt, we performed our spindle analysis separately on the right and left sides of MD1 and found that spindles on the left side align with an average division angle of about 30 from the AP axis whereas spindles on the right side align with an average division angle of -30 from the AP axis. To determine whether spindles on either side rotated with a certain chirality, we found there was no preference in rotating clockwise or counterclockwise on the left and right sides (on the left side of MD1 53% of measured spindles rotated counterclockwise and 47% rotated clockwise, on the right side 46% rotated counterclockwise and 54% clockwise). We have added this data as Fig. 1I-J and discussed in the Results lines 134-145.

1. The authors are experts in mesoderm invagination, and understandably concentrate on the role that forces from that process may have in the orientation of head MD divisions. However, the cephalic furrow forms much closer to the head MDs, and in an orientation that might also explain the alignment of spindles in the head. Is cephalic furrow formation important for Pins polarity and spindle orientation in the head MDs?

This was certainly a possibility, but our experimental results strongly argues that mesoderm invagination is most relevant.

1) Perturbing the ventral furrow (e.g. by Snail depletion) does not block the cephalic furrow (Vincent et al., 1997; Leptin and Grunewald, 1990), but does block mesoderm invagination. Snail depletion strikingly disrupted spindle orientation and Pins localization, which suggests mesoderm is most important.

2) In addition, depletion of -catenin blocks ventral furrow invagination but not cephalic furrow formation. We see a disruption in spindle orientation and Pins localization in -catenin RNAi, which suggests cephalic furrow itself cannot orient spindles.

3) Furthermore, light sheet imaging of the Drosophila embryo has shown that the head region of the embryo undergoes tissue movement in the direction of the cell division and that this is associated with mesoderm invagination (Streichan et al., 2018; Stern et al., 2022).

See movies here: https://www.youtube.com/watch?v=kC11Upr30JY

To further test the importance of mesoderm invagination, we will perform additional ablation experiments trying to disrupt forces transmitted to the mitotic domains from distinct directions. Once we get this experimental result we will include language in the Discussion that will summarize the experimental results and the weight of the evidence for the roles of either ventral or cephalic furrow.

1. Does expression of myristoylated Pins affect mesoderm invagination (or cephalic furrow formation)? From Table S1 it seems that a maternal Gal4 driver was used to express myristoylated Pins, which could affect other tissues in the embryo. So it is in principle possible that effects of myristoylated Pins on mesoderm internalization/cephalic furrow formation could affect cell division orientation much like sna loss of function does, but in a mechanism that does not depend on Pins polarity. There is definitely an effect on mesoderm invagination in alpha-catenin RNAi (but not in canoe RNAi) embryos, so I wonder if the effect could be consistently through defects in mesoderm invagination (or cephalic furrow formation), and Pins polarity is really dispensable for spindle orientation. Are there head-specific Gal4 drivers that could be used to drive myristoylated Pins exclusively in the head?

We apologize that we did not clarify this in the text. Maternal overexpression of myr-Pins does not obviously disrupt mesoderm internalization/cephalic furrow formation. But, we do see that targeted disruption of mesoderm internalization via a Snail depletion affects the orientation of division. Note that our paper demonstrates the effect of force transmission on Pins polarity and division orientation, which is new and the main conclusion. The role of these divisions in morphogenesis is more complicated and is beyond the scope of this study.

In response to this comment we: 1) added language in the Results that states that gastrulation proceeds in myr-Pins expressing embryos (lines 206-208), 2) Added to the Discussion of the role of these oriented divisions to morphogenesis (lines 443-449), and 3) will add a figure showing ventral furrow and cephalic furrow formation in embryos ectopically expressing the myr-Pins.

1. Related to the previous point, does mechanical isolation by laser ablation (Figure 6I-N) affect Pins polarity? This experiment could alleviate some of my concerns above, as it certainly does not (should not?) disrupt neither mesoderm invagination nor cephalic furrow formation.

We agree that it would be useful to look at Pins polarity in laser ablated embryos. Currently, we have been unable to analyze Pins polarity after laser ablation, because the ablation to fully isolate the mitotic domain has bleached our Pins::GFP signal. Also, we have shown that Pins polarity is disrupted by 1) alpha-catenin-RNAi, 2) low dose CytoD injection, and 3) Snail depletion, all of which are expected to disrupt force generation and transmission through tissues.

In response to the reviewer comment, we will determine if Pins::GFP can be analyzed in less aggressive (directional) laser ablations. Again, remember that myr-Pins does not affect mesoderm internalization and that Snail depletion affects Pins polarity.

MINOR 1. Figure S5: I am a bit confused about the role of Toll 2, 6, 8 in orienting spindle orientation. In Figure S5D it seems that dsRNA treatment against these genes does not disrupt spindle orientation, but Figure S5F shows quite a significant (p=0.0057) effect in triple mutants. The authors favor the idea that Toll receptors do not affect spindle orientation, but the difference with the mutant should be addressed. Furthermore, what happens in MDs 3, 5 and 14 (if the germband extension defect does not affect those divisions)? Is there a difference between dsRNA and triple mutant embryos in these other MDs?

We think this is a great point. We stated in the text that TLRs are not solely responsible (line 247) for spindle orientation as they do not recapitulate the random pattern of division seen in the myr-Pins expression condition. We acknowledge the differences between the dsRNA injection and TLR triple mutant in the manuscript (lines 242-247), but our data show a greater importance for the role of force transmission. We favor the idea that other mechanisms contribute to spindle orientation because of the small effect of mutating all three Tolls and the dramatic effects of depleting AJs, inhibiting actin (with CytoD), laser ablation, and blocking mesoderm invagination. The planned laser ablation experiments (described above) will also contribute to addressing this point.

1. No statistical analysis is provided for any of the differences in polarity between Pins and Gap43, and this should be done to demonstrate the significance of the polarization of Pins. Also, particularly for MD14, they should compare anterior vs. posterior polarity, as based on the images in Figure 2H it is not clear that there is a difference between the anterior and posterior side of cells.

We thank the reviewer for this point. We have added the statistical comparison.

1. Figure 2A-D: the authors propose that Pins localizes preferentially to the posterior end of cells (instead of both anterior and posterior ends) in MDs 1, 3 and 14 (and anterior in MD 5). How is the asymmetry in the distribution of Pins along the AP axis accomplished, and is there any significance to it? This should be discussed in a bit more detail (currently no potential mechanisms provided in the discussion, just an acknowledgment of the question).

__We agree the localization of Pins to the posterior end of cells in MDs 1, 3, and 14 and anterior end in MD 5 is of great interest. The details and further mechanism of this preferential localization are beyond the scope of this paper, but we have added an acknowledgment of the question and discuss possible models that could explain the result (lines 458-460). __TYPOS 1. Line 49: "one daughter cells" should be "one daughter cell". 2. Line 193: "rotation. (Figure 3E-F)." should be "rotation (Figure 3E-F)." 3. Lines 232-237: please review. 4. Line 238: "epithelia cells" should be "epithelial cells".

We thank the reviewers for carefully reading our manuscript. We have fixed the typos mentioned.

Reviewer #1 (Significance (Required)): This is the first study to my knowledge that demonstrates the role of mechanical forces in polarizing Pins, and provides a nice model to further investigate how mechanical forces generated in one tissue may affect cell division orientation in distant ones. The paper is clear, well written, and quantitative analysis is present for most results. I have some issues with the statistics (or lack thereof) for a couple of results, and potential alternative interpretations for some experiments that in my opinion should be addressed prior to publication. Specifically, it is not clear to me if Pins polarity is at all necessary for spindle orientation in any of the examined MDs.

Reviewer #2 (Evidence, reproducibility and clarity (Required)): Overview: In this manuscript, Camuglia et al. show Pins/LGN, which is understood to drive spindle orientation, can localize asymmetrically (with respect to the tissue plane) in the Drosophila embryo. Experimental work (including drug treatments, laser ablation, and knockdowns) lead the authors to propose that this asymmetry is driven by tissue-level tension. The findings are quite interesting and the manuscript is well-written overall. Major Comments: • The authors propose that localization is driven by tissue-level tension, but the direction of the tension isn't clear from the experimental work. For example, the laser ablation experiments cut around the entire perimeter of the mitotic domain, rather than along just one tension axis. Similarly, the finding that disruption of the ventral furrow (by Snail RNAi) interferes with spindle orientation in the head is very puzzling; the furrow is A) outside the embryonic head and B) runs in the parallel direction to the divisions considered. The authors need to address the directionality of tension experimentally.

We thank the reviewer for this comment and agree that better defining the direction of tension would strengthen our manuscript. We showed that blocking mesoderm invagination with Snail depletion disrupts spindle orientation, despite Snail not being required for cephalic furrow formation (refs). Recent light sheet data has shown that mesoderm invagination is associated with global movements throughout the embryo. Furthermore, the ventral furrow extends into the head region just past the anterior of MD5. To address the reviewer’s comments, we plan to: 1) Perform directional laser ablations to determine the directionality of the tension that orients the spindle, 2) Analyze strain rates in the mitotic domains prior to and during division, and 3) Add to our Discussion more about what is said in the literature about the movements that occur in the head during mesoderm invagination.

• As acknowledged in the text, the asymmetric enrichment of Pins in MD14 is fairly weak. Since the cells being examined here border a divot in the tissue, and might therefore be curving relative to the focal plane, it would be good to rule out the possibility that some of the asymmetry in Pins intensity is just a consequence of cell/tissue geometry. One way this could be achieved is by showing multiple focal planes.

Good point. We do not think that the asymmetric Pins enrichment in MD14 is due to tissue geometry or junction tilt. 1) MD14 divides ~10-15 minutes after mesoderm invagination is completed, so the cells do not border a divot (as seen with Gap43::mCh, Fig. 2I). The cells do round up, which can be seen as gaps between cells (Fig. 3E). 2) We compare Pins to GapCh and only see an enrichment with Pins (Fig. 2H-K). If the enrichment was due to tissue curvature or junction orientation relative to imaging axis, we would see the same enrichment in GapCh. 3) Expression of myr-Pins randomizes spindle orientation in MD14 (Fig. 3M, N).

• In Figure 3I (and 3M?), it appears that there are fewer cell divisions in the presence of myr-Pins. Is this the case? Since cell shapes change during division, and cell shapes influence tissue tension, an increase in cell divisions could lead to a change in tissue tension. This would be important to address, since tissue tension plays an important role in the proposed model.

These images are not taken at the same point of MD1 division ‘wave’, there are the same number of divisions in each condition. These mitotic domains exhibit a ‘wave’ of cell division (Di Talia and Wieschaus, 2012), and so the number of divisions in each image reflect the timing at which we captured the image. Quantifications involved divisions throughout this wave, but we have chosen images for figures which are most representative of what we see. We will add this to the text in the final version of the manuscript.

• The alpha-catenin and Canoe results are a bit confusing: - The rose plot in Figure 4D doesn't show a random distribution of spindle angles, but rather a modest change; most spindles still orient in the normal range. The p value in the figure legend (0.0012) is very different from the one in the figure (5.8284e-04). - Alpha-catenin is the strongest way to disrupt AJs, but A) the epithelium appears to be intact in the knockdown condition and B) spindle orientation is impacted but not randomized. Does this mean that the knockdown is incomplete? Or is Cadherin-mediated adhesion (in which alpha-catenin participates) only partially responsible for force transduction?

We acknowledge that perturbation using ____alpha-cat RNAi does not recapitulate the complete disruption of division orientation seen in embryos expressing myr-Pins. This is likely due to the variability in the strength of RNAi knockdown, which is observed for most RNAi lines that we use. To address the reviewer’s comment, we have added rose plots for individual embryos showing extremes in the severity of division orientation disruption (Fig. 4E and F). For the main plot (Fig. 4D), we have included all the data that we took because we obviously did not want to pick and choose which embryos were used for analysis. So Fig. 4D includes all the variability.

• Given that previous studies implicate Canoe in Pins localization, it seems important to lock down the question of whether Canoe is participating in the mechanism described in this paper. How do the authors know the extent of Canoe knockdown? As suggested by the alpha-catenin results (described above), is it possible that Canoe knockdown is simply not strong enough to impact spindle orientation? Aren't there genetic nulls available? We thank the reviewer for bringing these points to our attention. There are certainly genetic nulls available (Sawyer et al., 2009), but the experiment suggested by the reviewer would not establish the necessity of Canoe in mitotic domain cells. This is because Canoe nulls severely disrupt mesoderm invagination (Sawyer et al., 2009; Jodoin et al., 2015), as well as affecting junctions in the ectoderm during germband extension (Sawyer et al., 2011). Therefore, we would not be able to distinguish what effect of Canoe would be responsible for the spindle orientation using a null mutation. We did better experiments, we used 1) a mutant which specifically compromised mesoderm invagination (snail), 2) laser isolation to show the importance of external force transmission in orienting mitotic domain divisions, and 3) RNAi to deplete Canoe so that mesoderm invagination initiates and pulls on the ectoderm, but where there is clearly compromised Canoe function. This treatment did not cause any effect on spindle orientation arguing against a role of Canoe in this case. In response to the reviewers comment, we added language to the Results to indicate that it is possible that the Canoe knockdown is not strong enough and our rationale for why we did not perform the experiment in a Canoe null (lines 279-282).

Minor Comments:

• It can be difficult to interpret some of the spindle orientation data since the AP axis is vertical in the diagrams but horizontal in the rose plots. Can one of these be flipped so they go together?

We thank the reviewer for this suggestion and have flipped the rose plots so they match the images. Note that because of the large size of the figures, we have had to consistently orient anterior towards the top, which we establish at the beginning of the Results.

• Figure S3 is important information for the reader and should be ideally moved into the main paper. - Protein localizations referred to in text should be annotated on images, as they can be hard to see.

We disagree that S3 should be included in the main paper. The myr-Pins reagent has been used previously so the information in S3 is not new (Chanet et al., 2017).

• There are some discrepancies between figures, legends and text. - p-values differ between figures, legends, and/or text. - Fluorescent markers are labelled differently in figures and legend (CLIP170 in Figure 1) - Graphs appear to show that MD3 polarizes on posterior side, but figure legend says anterior in Figure S1. Vice versa for MD5.

We thank the reviewer for catching these typos. We have fixed these issues.

• Ideally, multichannel image overlays should be shown along with individual channels (b/w). However, it is appreciated that the fluorescent signals are exceptionally weak in this study, presenting a challenge to presentation and to quantification.

We agree the overlays would be nice. However, the Pins::GFP signal is weak compared to the tubulin and Gap43 signals, the merge does not provide more clarity, and the figures are already quite large. Therefore, we have only included the separated the images.

• Graph axes depicting spindle orientation would be more clear if shown in degrees, instead of normalized or in radians.

We thank the reviewer for this suggestion. We have changed the graph axes to be in degrees.

Reviewer #2 (Significance (Required)): Several recent studies have demonstrated that division orientation (in the tissue plane) is governed by tissue level tension. Remarkably, it appears that diverse mechanisms link tension with spindle orientation. Here the authors provide the first in vivo evidence connecting tension to the asymmetric localization of Pins, an important and evolutionarily conserved spindle orientation factor.

Reviewer #3 (Evidence, reproducibility and clarity (Required)): This beautiful manuscript uncovers a role for planar polarized PINS/LGN in orienting the mitotic spindle in Drosophila epithelia. In response to morphogenetic forces acting on adherens junctions, PINS/LGN localises to junctions in a planar polarized fashion to orient the spindle, and de-polarization of PINS/LGN prevents planar spindle orientation. The experiments are very well performed and the findings are robust. The conclusions are well supported by the data. Reviewer #3 (Significance (Required)): These important findings mirror previous work in human cell culture, but crucially reveal that the same phenomenon occurs in vivo in the Drosophila embryo. Thus, the findings underscore the highly conserved nature and in vivo relevance of this phenomenon.

We thank this reviewer for reading the manuscript and their encouraging words.

Author Response:

Reviewer #1 (Public Review):

This paper is of potential interest to researchers performing animal behavioral quantification with computer vision tools. The manuscript introduces 'BehaviorDEPOT', a MATLAB application and GUI intended to facilitate quantification and analysis of freezing behavior from behavior movies, along with several other classifiers based on movement statistics calculated from animal pose data. The paper describes how the tool can be applied to several specific types of experiments, and emphasizes the ease of use - particularly for groups without experience in coding or behavioral quantification. While these aims are laudable, and the software is relatively easy to use, further improvements to make the tool more automated would substantially broaden the likely user base.

In this manuscript, the authors introduce a new piece of software, BehaviorDEPOT, that aims to serve as an open source classifier in service of standard lab-based behavioral assays. The key arguments the authors make are that 1) the open source code allows for freely available access, 2) the code doesn't require any coding knowledge to build new classifiers, 3) it is generalizable to other behaviors than freezing and other species (although this latter point is not shown) 4) that it uses posture-based tracking that allows for higher resolution than centroid-based methods, and 5) that it is possible to isolate features used in the classifiers. While these aims are laudable, and the software is indeed relatively easy to use, I am not convinced that the method represents a large conceptual advance or would be highly used outside the rodent freezing community.

Major points:

1) I'm not convinced over one of the key arguments the authors make - that the limb tracking produces qualitatively/quantitatively better results than centroid/orientation tracking alone for the tasks they measure. For example, angular velocities could be used to identify head movements. It would be good to test this with their data (could you build a classifier using only the position/velocity/angular velocities of the main axis of the body?

2) This brings me to the point that the previous state-of-the-art open-source methodology, JAABA, is barely mentioned, and I think that a more direct comparison is warranted, especially since this method has been widely used/cited and is also aimed at a not-coding audience.

Here we address points 1 and 2 together. JAABA has been widely adopted by the drosophila community with great success. However, we noticed that fewer studies use JAABA to study rodents. The ones that did typically examined social behaviors or gross locomotion, usually in an empty arena such as an open field or a standard homecage. In a study of mice performing reaching/grasping tasks against complex backgrounds, investigators modified the inner workings of JAABA to classify behavior (Sauerbrei et al., 2020), an approach that is largely inaccessible to inexperienced coders. This suggested to us that it may be challenging to implement JAABA for many rodent behavioral assays.

We directly compared BehaviorDEPOT to JAABA and determined that BehaviorDEPOT outperforms JAABA in several ways. First, we used MoTr and Ctrax (the open-source centroid tracking software packages that are typically used with JAABA) to track animals in videos we had recorded previously. Both MoTr and Ctrax could fit ellipses to mice in an open field, in which the mouse is small relative to the environment and runs against a clean white background. However, consistent with previous reports (Geuther et al., Comm. Bio, 2019), MoTr and Ctrax performed poorly when rodents were fear conditioning chambers which have high contrast bars on the floor (Fig. 10A–C). These tracking-related hurdles may explain, at least in part, why relatively few rodent studies have employed JAABA.

We next tried to import our DeepLabCut (DLC) tracking data into JAABA. The JAABA website instructs users to employ Animal Part Tracker (https://kristinbranson.github.io/APT/) to convert DLC outputs into a format that is compatible with JAABA. We discovered that APT was not compatible with the current version of DLC, an insurmountable hurdle for labs with limited coding expertise. We wrote our own code to estimate a centroid from DLC keypoints and fed the data into JAABA to train a freezing classifier. Even when we gave JAABA more training data than we used to develop BehaviorDEPOT classifiers (6 videos vs. 3 videos), BehaviorDEPOT achieved higher Recall and F1 scores (Fig. 10D).

In response to point 1, we also trained a VTE classifier with JAABA. When we tested its performance on a separate set of test videos, JAABA could not distinguish VTE vs. non-VTE trials. It labeled every trial as containing VTE (Fig. 10E), indicating that a fitted ellipse is not sufficient to detect fine angular head movements. JAABA has additional limitations as well. For instance, JAABA reports the occurrence of behavior in a video timeseries but does not allow researchers to analyze the results of experiments. BehaviorDEPOT shares features of programs like Ethovision or ANYmaze in that it can classify behaviors and also report their occurrence with reference to spatial and temporal cues. These direct comparisons address some of the key concerns centered around the advances BehaviorDEPOT offers beyond JAABA. They also highlight the need for new behavioral analysis software targeted towards a noncoding audience, particularly in the rodent domain.

3) Remaining on JAABA: while the authors' classification approach appeared to depend mostly on a relatively small number of features, JAABA uses boosting to build a very good classifier out of many not-so-good classifiers. This approach is well-worn in machine learning and has been used to good effect in highthroughput behavioral data. I would like the authors to comment on why they decided on the classification strategy they have.

We built algorithmic classifiers around keypoint tracking because of the accuracy flexibility and speed it affords. Like many behavior classification programs, JAABA relies on tracking algorithms that use background subtraction (MoTr) or pattern classifiers (Ctrax) to segment animals from the environment and then abstract their position to an ellipse. These methods are highly sensitive to changes the experimental arena and cannot resolve fine movement of individual body parts (Geuther et al., Comm. Bio, 2019; Pennington et al., Sci. Rep. 2019; Fig. 10A). Keypoint tracking is more accurate and less sensitive to environmental changes. Models can be trained to detect animals in any environment, so researchers can analyze videos they have already collected. Any set of body parts can be tracked and fine movements such as head turns can be easily resolved (Fig. 10E).

Keypoint tracking can be used to simultaneously track the location of animals and classify a wide range of behaviors. Integrated spatial-behavioral analysis is relevant to many assays including fear conditioning, avoidance, T-mazes (decision making), Y-mazes (working memory), open field (anxiety, locomotion), elevated plus maze (anxiety), novel object exploration, and social memory. Quantifying behaviors in these assays requires analysis of fine movements (we now show Novel Object Exploration, Fig. 5 and VTE, Fig. 6 as examples). These behaviors have been carefully defined by expert researchers. Algorithmic classifiers can be created quickly and intuitively based on small amounts of video data (Table 4) and easily tweaked for out of sample data (Fig. 9). Additional rounds of machine learning are time consuming, computationally intensive, and unnecessary, and we show in Figure 10 that JAABA classifiers have higher error rates than BehaviorDEPOT classifiers, even when provided with a larger set of training data. Moreover, while JAABA reports behaviors in video timeseries, BehaviorDEPOT has integrated features that report behavior occurring at the intersection of spatial and temporal cues (e.g. ROIs, optogenetics, conditioned cues), so it can also analyze the results of experiments. The automated, intuitive, and flexible way in which BehaviorDEPOT classifies and quantifies behavior will propel new discoveries by allowing even inexperienced coders to capitalize on the richness of their data.

Thank you for raising these questions. We did an extensive rewrite of the intro and discussion to ensure these important points are clear.

4) I would also like more details on the classifiers the authors used. There is some detail in the main text, but a specific section in the Methods section is warranted, I believe, for transparency. The same goes for all of the DLC post-processing steps.

Apologies for the lack of detail. We included much more detail in both the results and methods sections that describe how each classifier works, how they were developed and validated, and how the DLC post-processing steps work.

5) It would be good for the authors to compare the Inter-Rater Module to the methods described in the MARS paper (reference 12 here).

We included some discussion of how BehaviorDEPOT Inter-Rater Module compares to the MARS.

6) More quantitative discussion about the effect of tracking errors on the classifier would be ideal. No tracking is perfect, so an end-user will need to know "how good" they need to get the tracking to get the results presented here.

We included a table detailing the specs of our DLC models and the videos that we used for validating our classifiers (Table 4). We also added a paragraph about designing video ‘training’ and test sets to the methods.

Reviewer #2 (Public Review):

BehaviorDEPOT is a Matlab-based user interface aimed at helping users interact with animal pose data without significant coding experience. It is composed of several tools for analysis of animal tracking data, as well as a data collection module that can interface via Arduino to control experimental hardware. The data analysis tools are designed for post-processing of DeepLabCut pose estimates and manual pose annotations, and includes four modules: 1) a Data Exploration module for visualizing spatiotemporal features computed from animal pose (such as velocity and acceleration), 2) a Classifier Optimization module for creating hand-fit classifiers to detect behaviors by applying windowing to spatiotemporal features, 3) a Validation module for evaluating performance of classifiers, and 4) an Inter-Rater Agreement module for comparing annotations by different individuals.

A strength of BehaviorDEPOT is its combination of many broadly useful data visualization and evaluation modules within a single interface. The four experimental use cases in the paper nicely showcase various features of the tool, working the user from the simplest example (detecting optogenetically induced freezing) to a more sophisticated decision-making example in which BehaviorDEPOT is used to segment behavioral recordings into trials, and within trials to count head turns per trial to detect deliberative behavior (vicarious trial and error, or VTE.) The authors also demonstrate the application of their software using several different animal pose formats (including from 4 to 9 tracked body parts) from multiple camera types and framerates.

1) One point that confused me when reading the paper was whether BehaviorDEPOT was using a single, fixed freezing classifier, or whether the freezing classifier was being tuned to each new setting (the latter is the case.) The abstract, introduction, and "Development of the BehaviorDEPOT Freezing Classifier" sections all make the freezing classifier sound like a fixed object that can be run "out-of-the-box" on any dataset. However, the subsequent "Analysis Module" section says it implements "hard-coded classifiers with adjustable parameters", which makes it clear that the freezing classifier is not a fixed object, but rather it has a set of parameters that can (must?) be tuned by the user to achieve desired performance. It is important to note that the freezing classifier performances reported in the paper should therefore be read with the understanding that these values are specific to the particular parameter configuration found (rather than reflecting performance a user could get out of the box.)

Our classifier does work quite well “out of the box”. We developed our freezing classifier based on a small number of videos recorded with a FLIR Chameleon3 camera at 50 fps (Fig. 2F). We then demonstrated its high accuracy in three separately acquired data sets (webcam, FLIR+optogenetics, and Minicam+Miniscope, Fig. 2–4, Table 4). The same classifier also had excellent performance in mice and rats from external labs. With minor tweaks to the threshold values, we were able to classify freezing with F1>0.9 (Fig. 9). This means that the predictive value of the metrics we chose (head angular velocity and back velocity) generalizes across experimental setups.

Popular freezing detection software including FreezeFrame, VideoFreeze as well as the newly created ezTrack also allow users to adjust freezing classifier thresholds. Allowing users to adjust thresholds ensures that the BehaviorDEPOT freezing classifier can be applied to videos that have already been recorded with different resolutions, lighting conditions, rodent species, etc. Indeed, the ability to easily adjust classifier thresholds for out-of-sample data represents one of the main advantages of hand-fitting classifiers. Yet BehaviorDEPOT offers additional advantages above FreezeFrame, VideoFreeze, and ezTrack. For one, it adds a level of rigor to the optimization step by quantifying classifier performance over a range of threshold values, helping users select the best ones. Also, it is free, it can quantify behavior with reference to user-defined spatiotemporal filters, and it can classify and analyze behaviors beyond freezing. We updated the results and discussions sections to make these points clear.

2) This points to a central component of BehaviorDEPOT's design that makes its classifiers different from those produced by previously published behavior detection software such as JAABA or SimBA. So far as I can tell, BehaviorDEPOT includes no automated classifier fitting, instead relying on the users to come up with which features to use and which thresholds to assign to those features. Given that the classifier optimization module still requires manual annotations (to calculate classifier performance, Fig 7A), I'm unsure whether hand selection of features offers any kind of advantage over a standard supervised classifier training approach. That doesn't mean an advantage doesn't exist- maybe the hand-fit classifiers require less annotation data than a supervised classifier, or maybe humans are better at picking "appropriate" features based on their understanding of the behavior they want to study.

See response to reviewer 1, point 3 above for an extensive discussion of the rationale for our classification method. See response to reviewer 2 point 3 below for an extensive discussion of the capabilities of the data exploration module, including new features we have added in response to Reviewer 2’s comments.

3) There is something to be said for helping users hand-create behavior classifiers: it's easier to interpret the output of those classifiers, and they could prove easier to fine-tune to fix performance when given out-ofsample data. Still, I think it's a major shortcoming that BehaviorDEPOT only allows users to use up to two parameters to create behavior classifiers, and cannot create thresholds that depend on linear or nonlinear combinations of parameters (eg, Figure 6D indicates that the best classifier would take a weighted sum of head velocity and change in head angle.) Because of these limitations on classifier complexity, I worry that it will be difficult to use BehaviorDEPOT to detect many more complex behaviors.

To clarify, users can combine as many parameters as they like to create behavior classifiers. However, the reviewer raises a good point and we have now expanded the functions of the Data Exploration Module. Now, users can choose ‘focused mode’ or ‘broad mode’ to explore their data. In focused mode, researchers use their intuition about behaviors to select the metrics to examine. The user chooses two metrics at a time and the Data Exploration Module compares values between frames where behavior is present or absent and provides summary data and visual representations in the form of boxplots and histograms. A generalized linear model (GLM) also estimates the likelihood that the behavior is present in a frame across a range of threshold values for both selected metrics (Fig. 8A), allowing users to optimize parameters in combination. This process can be repeated for as many metrics as desired.

In broad mode, the module uses all available keypoint metrics to generate a GLM that can predict behavior. It also rank-orders metrics based on their predictive weights. Poorly predictive metrics are removed from the model if their weight is sufficiently small. Users also have the option to manually remove individual metrics from the model. Once suitable metrics and thresholds have been identified using either mode, users can plug any number and combination of metrics into a classifier template script that we provide and incorporate their new classifier into the Analysis Module. Detailed instructions for integrating new classifiers are available in our GitHub repository (https://github.com/DeNardoLab/BehaviorDEPOT/wiki/Customizing-BehaviorDEPOT).

MoSeq, JAABA, MARS, SimBA, B-SOiD, DANNCE, and DeepEthogram are among a group of excellent opensource software packages that already do a great job detecting complex behaviors. They use supervised or unsupervised machine learning to detect behaviors that are difficult to see by eye including social interactions and fine-scale grooming behaviors. Instead of trying to improve upon these packages, BehaviorDEPOT is targeting unmet needs of a large group of researchers that study human-defined behaviors and need a fast and easy way to automate their analysis. As examples, we created a classifier to detect vicarious trial and error (VTE), defined by sweeps on the head (Fig. 9). Our revised manuscript also describes our new novel object exploration classifier (Fig. 5). Both behaviors are defined based on animal location and the presence of fine movements that may not be accurately detected by algorithms like MoTr and Ctrax (Fig. 10). As discussed in response to reviewer 1, point 3, additional rounds of machine learning are laborious (humans must label frames as input), computationally intensive, harder to adjust for out-of-sample videos, and are not necessary to quantify these kinds of behaviors.

4) Finally, I have some concerns about how performance of classifiers is reported. For example, the authors describe "validation" set of videos used to assess freezing classifier performance, but they are very vague about the detector was trained in the first place, stating "we empirically determined that thresholding the velocity of a weighted average of 3-6 body parts ... and the angle of head movements produced the bestperforming freezing classifier." What videos were used to come to this conclusion? It is imperative that when performance values are reported in the paper, they are calculated on a separate set of validation videos, ideally from different animals, that were never referenced while setting the parameters of the classifier. Otherwise, there is a substantial risk of overfitting, leading to overestimation of classifier performance. Similarly, Figure 7 shows the manual fitting of classifiers to rat and mouse data; the fitting process in 7A is shown to include updating parameters and recalculating performance iteratively. This approach is fine, however I want to confirm that the classifier performances in panels 7F-G were computed on videos not used during fitting.

Thank you for pointing this out. We have included detailed descriptions of the classifier development and validation in the results (149–204) and methods (789–820) sections and added a table that describes videos used to validate each classifier (Table 4).

To develop the classifier freezing, we explored linear and angular velocity metrics for various keypoints, finding that angular velocity of the head and linear velocity of a back point tracked best with freezing. Common errors in our classifiers were identified as short sequences of frames at the beginning or end of a behavior bout. This may reflect failures in human detection. Other common errors were sequences of false positive or false negative frames that were shorter than a typical behavior bout. We included the convolution algorithm to correct these short error sequences.

When developing classifiers (including adjust the parameters for the external videos), videos were randomly assigned to classifier development (e.g. ‘training’) and test sets. Dividing up the dataset by video rather than by frame ensures that highly correlated temporally adjacent frames are not sorted into training and test sets, which can cause overestimation of classifier accuracy. Since the videos in the test set were separate from those used to develop the algorithms, our validation data reflects the accuracy levels users can expect from BehaviorDEPOT.

5) Overall, I like the user-friendly interface of this software, its interaction with experimental hardware, and its support for hand-crafted behavior classification. However, I feel that more work could be done to support incorporation of additional features and feature combinations as classifier input- it would be great if BehaviorDEPOT could at least partially automate the classifier fitting process, eg by automatically fitting thresholds to user-selected features, or by suggesting features that are most correlated with a user's provided annotations. Finally, the validation of classifier performance should be addressed.

Thank you for the positive feedback on the interface. We addressed these comments in response to points 3 and 4. To recap, we updated the Data Exploration Module to include Generalized Linear Models that can suggest features with the highest predictive value. We also generated template scripts that simplify the process of creating new classifiers and incorporating them into the Analysis Module. We also included all the details of the videos we used to validate classifier performance, which were separate from the videos that we used to determine the parameters (Table 4).

Reviewer #3 (Public Review): There is a need for standardized pipelines that allow for repeatable robust analysis of behavioral data, and this toolkit provides several helpful modules that researchers will find useful. There are, however, several weaknesses in the current presentation of this work.

1) It is unclear what the major advance is that sets BehaviorDEPOT apart from other tools mentioned (ezTrack, JAABA, SimBA, MARS, DeepEthogram, etc). A comparison against other commonly used classifiers would speak to the motivation for BehaviorDEPOT - especially if this software is simpler to use and equally efficient at classification.

We also address this in response to reviewer 1, points 1–3. To summarize, we added direct comparisons with JAABA to a revised manuscript. In Fig. 10, we show that BehaviorDEPOT outperforms JAABA in several ways. First, DLC is better at tracking rodents in complex environments than MoTr and Ctrax, which are the most used JAABA companion software packages for centroid tracking. Second, we show that even when we use DLC to approximate centroids and use this data to train classifiers with JAABA, the BehaviorDEPOT classifiers perform better than JAABA’s.

In a revised manuscript, we included more discussion of what sets BehaviorDEPOT apart from other software, focusing on these main points:

BehaviorDEPOT vs. commercially available packages (Ethovision, ANYmaze, FreezeFrame, VideoFreeze)

1) Ethovision, ANYmaze, FreezeFrame, VideoFreeze cost thousands of dollars per license while BehaviorDEPOT is free.

2) The BehaviorDEPOT freezing classifier performs robustly even when animals are wearing a tethered patch cord, while VideoFreeze and FreezeFrame often fail under these conditions.

3) Keypoint tracking is more accurate, flexible, and can resolve more detail compared to those that use background subtraction or pixel change detection algorithms combined with center of mass or fitted ellipses.

BehaviorDEPOT vs. packages targeted at non-coding audiences (JAABA, ezTrack)

1) DLC keypoint tracking performs better than MoTr and Ctrax in complex environments. As a result, JAABA has not been widely used in the rodent community. Built around keypoint tracking, BehaviorDEPOT will enable researchers to analyze videos in any type of arena, including videos they have already collected. Keypoint track also allows for detection of finer movements, which is essential for behaviors like VTE and object exploration.

2) Hand-fit classifiers can be creative quickly and intuitively for well-defined laboratory behaviors. Compared to machine learning-derived classifiers, they are easier to interpret and easier to fine-tune to optimize performance when given out-of-sample data.

3) Even when using DLC as the input to JAABA, BehaviorDEPOT classifiers perform better (Figure 10)

4) BehaviorDEPOT integrates behavioral classification, spatial tracking, and quantitative analysis of behavior and position with reference to spatial ROIs and temporal cues of interest. It is flexible and can accommodate varied experimental designs. In ezTrack, spatial tracking is decoupled from behavioral classification. In JAABA, spatial ROIs can be incorporated into machine learning algorithms, but users cannot quantify behavior with reference to spatial ROIs after classification has occurred. Neither JAABA nor ezTrack provide a way to quantify behavior with reference to temporal events (e.g. optogenetic stimuli, conditioned cues).

5) BehaviorDEPOT includes analysis and visualization tools, providing many features of the costly commercial software packages for free.

BehaviorDEPOT vs. packages based on keypoint tracking (SimBA, MARS, B-SOiD)

Other software packages based on keypoint tracking use supervised or unsupervised methods to classify behavior from animal poses. These software packages target researchers studying complex behaviors that are difficult to see by eye including social interactions and fine-scale grooming behaviors whereas BehaviorDEPOT targets a large group of researchers that study human defined behaviors and need a fast and easy way to automate their analysis. Many behaviors of interest will require spatial tracking in combination with detection of specific movements (e.g. VTE, NOE). Additional rounds of machine learning are laborious (humans must label frames as input), computationally intensive, and are not necessary to quantify these kinds of behaviors.

2) While the idea might be that joint-level tracking should simplify the classification process, the number of markers used in some of the examples is limited to small regions on the body and might not justify using these markers as input data. The functionality of the tool seems to rely on a single type of input data (a small number of keypoints labeled using DeepLabCut) and throws away a large amount of information in the keypoint labeling step. If the main goal is to build a robust freezing detector then why not incorporate image data (particularly when the best set of key points does not include any limb markers)?

While one main goal was to build a robust freezing detector, BehaviorDEPOT is a general-purpose software. BehaviorDEPOT can classify behaviors from video timeseries and can analyze the results of experiments similar to Ethovision or FreezeFrame. BehaviorDEPOT is particularly useful for assays in which behavioral classification is integrated with spatial location, including avoidance, decision making (T maze), and novel object memory/recognition. While image data is useful for classifying behavior, it cannot combine spatial tracking with behavioral classification. However, DLC keypoint tracking is well-suited for this purpose. We find that tracking 4–8 points is sufficient to hand-fit high performing classifiers for freezing, avoidance, reward choice in a T-maze, VTE, and novel object recognition. Of course, users always have the option to track more points because BehaviorDEPOT simply imports the X-Y coordinates and likelihood scores of any keypoints of interest.

3) Need a better justification of this classification method

See response to reviewer 1, points 1–3 above.

4) Are the thresholds chosen for smoothing and convolution adjusted based on agreement to a user-defined behavior?

Yes. We added more details in the text. Briefly, users can change the thresholds used in both smoothing and convolution in the GUI and can optimize the values using the Classifier Optimization Module. Smoothing is performed once at the beginning of a session and has an adjustable span for the smoothing window. The convolution is a feature of each classifier, and thus can be adjusted when adjusting the classifier. When developing the freezing classifier, we started with a smoothing window that had the largest value that did not exceed the rate of motion of the animal and then fine-tuned the value to optimize smoothing. In the classifiers we have developed, window widths that are the length of the smallest bout of ‘real’ behavior and count thresholds approximately 1/3 the window width yielded the best results.

5) Jitter is mentioned as a limiting factor in freezing classifier performance - does this affect human scoring as well?

We were referring to jitter in terms of point location estimates by DeepLabCut. In other words, networks that are tailored to the specific recording conditions have lower error rates in the estimates of keypoint positions. Human scoring is an independent process that is not affected by this jitter. We changed the wording in the text to avoid any confusion.

6) The use of a weighted average of body part velocities again throws away information - if one had a very high-quality video setup with more markers would optimal classification be done differently? What if the input instead consisted of 3D data, whether from multi-camera triangulation or other 3D pose estimation? Multianimal data?

From reviewer 2, point 3: MARS, SimBA, and B-SOiD are excellent open-source software packages that are also based on keypoint tracking. They use supervised or unsupervised methods to classify complex behaviors that are difficult to see by eye including social interactions and fine-scale grooming behaviors. Instead of trying to improve upon these packages, which are already great, BehaviorDEPOT is targeting unmet needs of a large group of researchers that study human defined behaviors and need a fast and easy way to automate their analysis. Additional rounds of machine learning are laborious (humans must label frames as input), computationally intensive, and are not necessary to quantify these kinds of behaviors. However, keypoint tracking offers accuracy, precision and flexibility that is superior to behavioral classification programs that estimate movement based on background subtraction, center of mass, ellipse fitting, etc.

7) It is unclear where the manual annotation of behavior is used in the tool as currently stands. Is the validation module used to simply say that the freezing detector is as good as a human annotator? One might expect that algorithms which use optic flow or pixel-based metrics might be superior to a human annotator, is it possible to benchmark against one of these? For behaviors other than freezing, a tool to compare human labels seems useful. The procedure described for converging on a behavioral definition is interesting and an example of this in a behavior other than freezing, especially where users may disagree, would be informative. It appears that manual annotation doesn't actually happen in the GUI and a user must create this themselves - this seems unnecessarily complicated.

Manual annotation of behavior is used in the four classifier development modules: inter-rater, data exploration, optimization, and validation. The inter-rater module can be used as a tool to refine ground-truth behavioral definitions. It imports annotations from any number of raters and generates graphical and text-based statistical reports about overlap, disagreement, etc. Users can use this tool to iteratively refine annotations until they converged maximally. The inter-rater module can be used to compare human labels (or any reference set of annotations) for any behavior. To ensure this is clear to the readers, we added more details to the text and second demonstration of the inter-rater module for novel object exploration annotations (Fig. 7). The validation module imports reference annotations which can be produced by a human or another program, which can benchmark classifier performance against the reference. We added more details to this section as well.

Freezing is a straightforward behavior that is easy to detect by eye. Rather than benchmark against an optic flow algorithm, we benchmarked against JAABA, another user-friendly behavioral classification software that uses machine learning algorithms. We find that BehaviorDEPOT is easier to use and labels freezing more accurately than JAABA. We also made a second freezing classifier that uses a changepoint algorithm to identify transitions from movement to freezing that may accommodate a wider range of video framerates and resolutions.

We plan to incorporate an annotation feature into the GUI, but in the interest of disseminating our work soon, we argue that this is not necessary for inclusion now. There are many free or cheap programs that allow framewise annotation of behavior including FIJI, Quicktime, VLC, and MATLAB. In fact, users may already have manual annotations or annotations produced by a different software and BehaviorDEPOT can import these directly. While machine learning classifiers like JAABA require human annotations to be entered into their GUI, allowing people to import annotations they collected previously saves time and effort.

8) A major benefit of BehaviorDEPOT seems to be the ability to run experiments, but the ease of programming specific experiments is not readily apparent. The examples provided use different recording methods and networks for each experimental context as well as different presentations of data - it is not clear which analyses are done automatically in BehaviorDEPOT and which require customizing code or depend on the MiniCAM platform and hardware. For example - how does synchronization with neural or stimulus data occur? Overall it is difficult to judge how these examples would be implemented without some visual documentation.

We added visual documentation of the experimental module graphical interface to figure 1 and added more detail to the results, methods and to our GitHub repository (https://github.com/DeNardoLab/Fear-Conditioning-Experiment-Designer). Synchronization with stimulus data can occur within the Experiment Module (designed for fear conditioning experiments) or stimuli timestamps can be easily imported into the Analysis Module. Synchronization with neural data occurs post hoc using the data structures produced by the BehaviorDEPOT Analysis Module. We include our code for aligning behavior to Miniscope on our GitHub repository https://github.com/DeNardoLab/caAnalyze).

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

Summary

In their manuscript, Hattori et al., put forward evidence that the knock-out of CD38 expression in astrocytes at approximately post-natal day 10 (referred to as CD38 AS-cKO P10) leads to a specific deficit in social memory in adult mice, while other types of memory remain unaltered. Using immunohistochemistry (IHC), the authors found a reduced number of excitatory synapses in the medial prefrontal cortex (mPFC) of CD38 AS-cKO P10 mice. Switching to in vitro primary cell culture models, the authors identify the astrocyte secreted protein SPARCL1 as a relevant synaptogenic factor. Using pharmacological dissection of relevant signaling pathways, Hattori et al., propose that cADPR formation and calcium released from intracellular stores, is essential for SPARCL1 secretion from astrocytes. Finally, the authors analyzed the transcriptome of primary CD38 KO astrocytes using bulk mRNA sequencing, and found that genes related to calcium signaling were downregulated in these cells.

Major commments:

• Are the key conclusions convincing?
  1. From a global perspective, the multiple lines of evidence provided by the authors strongly suggest that expression of CD38 in astrocytes is important for synaptogenesis in the mPFC of P10 mice, with ablation of CD38 and reduced synapse formation leading to social memory deficits at P70. However, the data concerning the role of astrocyte-secreted SPARCL1 is not particularly strong: further experiments are needed to support this claim (see below).
• Are the claims preliminary or speculative?
  1. As it stands, there is no proof that the claimed astrocyte-specific deletion of CD38 is actually astrocyte specific. This evidence is crucial: without it the reported effects could be due to non-specific CD38 knock-out in other CNS cells. In this respect, the Western Blot in Supplementary Figure 1A does not provide information on astrocyte-specific deletion, merely that CD38 was globally reduced in the mPFC. Interestingly, the authors have previously published data (Hattori et al., 2017, 10.1002/glia.23139) showing that CD38 expression is mostly astrocyte-specific, peaking at p14, which coincides with the peak period of synaptogenesis. The degree of CD38 heterogeneity is also an issue that I think the authors need to consider. Do they information on this? Is CD38 expressed in every astrocyte of the CNS, or are there some astrocytes that are CD38 negative at P14? Is the mPFC a region specifically enriched in CD38 positive astrocytes and does this explain the observed behavioral deficit? I think if this is known, the authors should mention it in the "Introduction" or "Discussion". If this is not known, maybe the authors could provide data addressing the issue.
  2. I think the authors should take more caution in claiming that SPARCL1 is the main factor secreted through the CD38 signaling pathway and responsible for increased synaptogenesis. This is for several reasons, all centered on data displayed in Figure 4 and Supplementary Figure 6:
     • a) Western Blot (WB) data: The "Materials and Methods" section for WB does not indicate how protein loading and transfer efficiency were controlled for. Normalizing to β-Actin levels is an acceptable way to control for loading and transfer efficiency when using cell lysates. However, in the absence of such an abundant structural protein in conditioned media it is unclear how loading and transfer was controlled for under these conditions. Do the authors normalized the CD 38 KO AS ACM data by expressing protein levels relative to those from WT AS ACM? Is BDNF being used as a control, based on proteomics data? If so, why is proteomics data not given in the manuscript and why is this control not shown for all ACM blots? I realize that (quantitative) blotting using ACM is difficult, but I am also not convinced that the methodology used is sufficiently rigorous. Simple steps to give confidence would be Coomassie staining of gels both before and after membrane transfer, to show that i) the total protein amount loaded was the same in each lane of the gel and ii) the transfer to the nitrocellulose membrane was complete. In addition, Ponceau S staining of the nitrocellulose membrane should also have been performed and displayed, to show (roughly) equal amounts of protein were transferred for each lane. In summary, the WB data quantification needs to be better controlled. The values of the Y axis in these graphs (and throughout the manuscript) are simply too small to be read properly. Finally, I want to highlight the general lack of precision regarding the nature of the replication unit (the "n"). For example, the legend of Figure4C-D states "n = 6", but we have no idea if these are 6 independent primary cultures originating from 6 mice, 6 independent cultures from the same mouse, 6 repeats of the Western Blot using the same sample etc. This issue is valid for the whole manuscript: in my opinion, the authors should be more much careful when it comes to these crucial elements of scientific reporting.
     • b) While the data hint at an important role of SPARCL1 in synapse formation, when the authors tested if ACM from CD38 KO astrocytes supplemented with exogenous SPARCL1 could rescue synapse formation, the effect was incomplete, with only a trend to an increase in synapse number (Figure 4J-K). Perhaps the authors simply forgot to indicate the statistical significance of differences between the experimental groups (Figure 4K)? However, if there really were no statistically significant differences observed, the authors should reduce the strength of their conclusions regarding SPARCL1. This protein may well be pro-synaptogenic but, as it stands, other factors could well be in play. Perhaps the authors should have tried higher concentrations of SPARCL1 to further boost synaptogenesis? In this respect, the SPARCL1 knockdown (KD) experiment in Supplementary Figure 6B-D is an important addition, but should be supplemented by rescue with an siRNA-resistant recombinant SPARCL1? If SPARCL1 is a major player in synaptogenesis, the prediction is that synapse numbers would be close to wild type levels with this approach.
     • c) In my opinion, there are also issues with the data displayed in Figure 4H-I. The authors want to convince the reader that SPARCL1 is mostly an astrocytic protein using immunohistochemistry on mouse mPFC sections, co-labelled with antibodies against neuronal and astrocytic markers. In these panels, we are presented with images showing a few cells, in which it seems SPARCL1 is absent from NeuN positive cells, present in WT astrocytes and reduced in CD38 AS-cKO P10 astrocytes. However, the numbers of cell counted and lack of quantification severely impact on the strength of this conclusion. In my opinion, the authors should have quantified their IHC data by counting cells and establishing the ratios of SPARCL1 positive over NeuN or S100β positive cells, in both control and CD38 AS-cKO P10 animals. This experiment would provide critical information that the conditional gene targeting strategy is robust. The authors should also consider quantifying the intensity of the SPARCL1 signal in astrocytes. This is recommended as the image displayed in Figure 4I for the CD38 AS-cKO is problematic: are the authors really claiming that the reduction in SPARCL1 expression following cKO of CD38 in astrocytes is at best only partial? Is 11 days between the first tamoxifen injection and tissue fixation actually sufficient to allow for CD38 turnover? With low levels of protein turnover, the possibility exists that residual levels of CD38 are still sufficient to impact SPARCL1 levels. What would happen if there is a greater interval between tamoxifen administration and tissue recovery? Would levels of synaptogenesis be further reduced? Is this an issue of production versus secretion or a combination of factors?
  3. The heatmap (Figure 5E-F) is simply too small to interpret. The color choice is also not accessible for colorblind readers. The authors might consider displaying this heatmap in a separate figure. The authors should also provide a supplementary table where all the genes detected are listed along with their respective counts. Furthermore, it is surprising that the authors only found genes being downregulated in CD 38 KO astrocytes. Were they really no genes up-regulated? The authors might also want to indicate the genes belong to each of the ontological categories listed in Figure 5F. On p. 11, Figure 5E: The authors should indicate in the main text they performed bulk RNA-sequencing and not another type of RNA sequencing (like single cell RNA sequencing for instance). The authors indicate n = 2 but we have no indications of the nature of the replicate (also see earlier comments). Please amend.
• Are additional experiments necessary? I think supplementary experiments are essential to support the claims of the paper. Most are described in the section above, but to summarize:
  1. Show data to prove that the CD38 AS-cKOP10 model is astrocyte-specific and leads to a total loss of CD38 in these cells.
  2. WB data: The issue of protein loading and transfer efficiency should be dealt with. Quantifications should be revisited.
  3. The authors should quantitatively analyze the different IHC performed in Figure 4H-I.
  4. The authors should provide more information on their RNA sequencing data: list of genes detected with their FPKM values etc. The authors should display the RNA sequencing data in a separate figure, allowing the heatmap to be enlarged.
  5. LC-MS/MS data: the authors should provide the list of all the proteins they identified in their LC-MS/MS experiment. As a supplementary table for instance? The majority of these experiments should be able to be performed with pre-existing samples/tissue slices. If not, the experimental pipeline necessary exits and these supporting experiments should not be too burdensome.
• Data and methods presentation Methods: The authors need to work on this aspect of the manuscript. Most of the important details are already described, but some crucial ones are missing, while the phrasing used to describe methods is sometimes misleading. I will give some examples here, but this is not an exhaustive list. The fact that the manuscript is riddled with small mistakes, inconsistencies and/or oversights makes it difficult to read and creates a negative impression. The whole manuscript would benefit from a thorough proof-reading, preferably by a native speaker.
  1. in the "Immunohistochemistry and Synaptic Puncta Analysis" section on p. 21-22, we have no indication of which antibodies against "GFAP, NDRG2, VGlut1, PSD95, S100β, NenN(?) and SPARCL1" were used. It is standard practice to indicate the company, product number and lot number. The authors must also indicate the dilution at which they use these antibodies. On p.22, the authors write the cells were incubated with "Alexa- or Cy3-conjugated secondary antibodies". The excitation wavelengths of the Alexa dyes used need to be given.
  2. The authors need to provide more details on the microscope they used. Merely writing "using a 63× lens on a fluorescence microscope" (p.23) is insufficient.
  3. In the "LC-MS/MS" method the authors wrote: "Briefly, these proteins were reduced, alkylated, and digested by trypsin". I think that in the reduction and alkylation steps, chemicals other than trypsin were actually used. This sentence should be modified to reflect this.
  4. p.19: "uM" is written when the authors very likely mean "µM". Please check the whole manuscript for repeat examples. I know this is often lab "short-hand", but it should be avoided in scientific publications.
  5. The authors should be careful when describing their data to always indicate whether they referring to experiments performed using cultured astrocytes or not. As it stands, the text is confusing: for instance, when describing RNA-sequencing data in Figure 5, the main text appears to indicate that these astrocytes were acutely isolated from adult mice, when in fact they were obtained from primary cultures. Given concerns in the literature about potential differences between acutely isolated and cultured astrocytes (Foo et al., Neuron, 2011), this is essential. Data presentation: The figures appear to have been produced in a rush - and almost have a "screenshot" feel to them. This is not a scientific issue per se, but does impact on the overall impression given by the manuscript. The following is a non-exhaustive list of issues with the figures. I list the major ones that the authors should correct.
  6. Almost all Y axis labels are too small. The authors should comply to the basic journal requirements in terms of font sizes. Some axes do not end on a tick (e.g. Figure 3R). This is not dramatic, but should be corrected. Globally, the authors need to display bigger bar plots - most of them are extremely hard to read. Labeling should also be checked: Figure 4K, the Y axis label indicates values displayed are in %, when I think the axis graduation displays ratio values. Some of the IHC pictures are also too small to be easily interpreted.
  7. The heatmap in Figure 5E is impossible to read and, as such, has little or no value for the manuscript.
  8. Scale bars: where is the scale bar in Figure 2A? Figure 3A-H: Is the scale bar really representing 10 millimeters? Supplementary Figure 3A: scale bar is missing. Please check for similar issues throughout the manuscript.
  9. Figure Legends are problematic, and often contain incorrect or incomplete information. Examples include: Supplementary Figure 1: The description of panels J, L and N appears to be missing. Please also use the Greek letter beta and not 'b' for S100β. Supplementary Figure 5: I think the term "KO" is missing after CD 38 in the legend title. Figure 3: why state that nuclei were counterstained with DAPI in Figure 3P,Q, when this precision is not given for panels Figure 3A-H? Figure 3A-H: If the authors choose to explicitly state PSD95 is a post-synaptic marker, why not indicate that VGlut1 is a pre-synaptic marker? Same issue in Supplementary Figure 4.
  10. There are multiple instances of panels being wrongly referred to in the main text. On p.10, Figure 4H is referenced, when I think the authors mean Figure 4I; on p.10, Figure 4I-J are referred to when the authors clearly describe data found in Figure 4J-K. These types of mistakes are problematic and recur throughout the manuscript.
• Statistical analysis As mentioned above, the exact nature of the replicates is often not stated, when the "n" number is indicated. The authors must correct this issue and give the information either at the appropriate point in the main text or in the figure legend.

The authors should also be more consistent in the way they indicate which statistical tests were performed. This should also be indicated either at the appropriate point in the main text or in the figure legend. Furthermore, care should be taken to ensure statistics are presented in an appropriate manner: at the end of legend for Figure 4, it is indicated #p < 0.05 vs. CD38 KO ACM. This hashtag symbol is completely absent from the figure. In Figure 4F-G, the lack of statistical symbols seems to indicate no statistical tests were performed on these data, when the legend covering these panels states "*p < 0.05 versus P70", indicating some tests were done. We cannot interpret this panel without knowing which comparisons were done exactly and which were significant.

In the "Materials and Methods", the authors give no indication that the assumptions of the statistical test they used were met (normality of data distribution for t-tests, homogeneity of variances for ANOVA...). This needs to be checked, and if not met, appropriate non-parametric tests should be used instead.

Minor commments:

• Specific experimental issues that are easily addressable. Most of the experimental issues that need to be addressed are given in previous sections and should be easily addressable.
• Citation of previous studies? Adequate
• Clarity and accuracy of text and figures There are issues with the clarity and accuracy of text and figures - which are described above. The text is also often problematic in its phrasing and other, more fundamental aspects. For instance, the authors spent a considerable amount of time speaking about the role of oxytocin, when they only performed one measurement of oxytocin levels in mice.
• Suggestions to improve the presentation of data and conclusions? All my suggestions to improve the presentation of data can found in previous sections. As for improving the authors presentation of their conclusions, the authors should make a considerable re-drafting effort, particularly for the "Discussion", which lacks clarity in how supporting arguments are built and presented. For example, on p.13, I am confused with the argument made by the authors. Their data are focused on synapses onto pyramidal neurons of the mPFC, but here the discussion states that the behavioral phenotype they observed in CD38 AS-cKOP10 might be explained by a lack of mPFC neurons synapsing onto neurons in the Nucleus Accumbens (assuming that "NAc" really refers to this brain region, as the definition is missing from the text). I think the authors should make it clear if this is their interpretation of their own result, which essentially renders their focus on mPFC pointless, or a speculation on possible other mechanisms that could also explain their behavioral results. Personally, given the data shown, I believe the authors should focus on explaining how their data in mPFC might explain the behavioral output observed. The authors could also provide perspectives on how the hypothesis laid down in this paragraph would be tested. When the authors write on p.14 "We identified SPARCL1 as a potential molecule for synapse formation in cortical neurons" why use the word "potential"? Does this mean the authors consider their data on SPARCL1 (one of the key messages of the paper) invalid? If the authors themselves think the role of SPARCKL1 is ambiguous based on their own data, they should perform further experiments. P. 13, the authors write: "Moreover, many studies have shown that astrocyte-specific molecules, including extracellular molecules such as IL-6, are involved in memory function"; Interleukin 6 (Il-6, abbreviation not defined in the manuscript) is definitely not an astrocyte-specific molecule (see, for example, Erta et al., 2021 10.7150/ijbs.4679).

Significance

NATURE AND SIGNIFICANCE OF THE ADVANCE: I think that despite the issues described above, this manuscript, once revised, could have a strong impact in the field. It would fuel the current paradigm shift which puts astrocytes at the forefront of neuronal circuit wiring during development with links to adult behavior. By identifying clear molecular targets involved in astrocyte-driven synaptogenesis, this article could help the clinical field to find new druggable targets, which may help reverse aging-related cognitive decline.

COMPARISON TO EXISTING PUBLISHED KNOWLEGDE: This work adds new data in the specific and growing line of research that study how astrocytes control synaptogenesis. Recent reviews have summarized advances in this field (Shan et al., 2021, 10.3389/fcell.2021.680301; Baldwin et al., 2021, 10.1016/j.conb.2017.05.006).

AUDIENCE: Neuroscientists in general, clinicians interested in cellular and molecular causes of neurodevelopmental disorders leading to social dysfunctions.

REVIEWER EXPERTISE: Astrocyte biology; Astrocyte-neuron interactions and synapse assembly; Neuronal circuit formation and plasticity

Referees cross-commenting

After careful reading of the other comments, I feel that there is considerable agreement/overlap between the reviewers on the main issues with this manuscript. Perhaps the major difference relates to the amount of further work necessary for the manuscript to be publication ready.

As Reviewer 3 rightly points out, this is always a moot point: how much is it reasonable for reviewers to ask authors to do? While I agree with all of Reviewer 1's comments regarding the rigour of the mass-spec/western blot analysis, it seems to me that from a molecular/cell biological point of view, the key issue is whether Sparcl1 is a synaptogenic factor released from astrocytes following CD38/cADPR/calcium signaling (irrespective of whether other factors may be in play); and whether raising Sparcl1 levels is sufficient to recover spine morphology and synapse numbers. Of course, if these experiments were performed in vivo using AAV-mediated overexpression of Sparcl1, it is also reasonable to think that the deficit in social memory may be reversed on testing.

The issues of whether there is a difference in observable behavioral phenotypes between the astrocyte-specific and constitutive CD38 knock-outs is an interesting one, as is why there is only a deficit in social memory seen following astrocyte-specific CD38 ablation. These issues should at least be discussed.

Author Response:

Evaluation Summary:

This study adds to the considerable, but often conflicting, work on how neurotransmitter systems contribute to auditory processing dysfunction. The paper details a thorough and careful analysis of an important hypothesis from the point of view of schizophrenia research: do muscarinic and dopaminergic receptors contribute to mismatch negativity effects? The answers could be useful for future treatment allocation in psychosis. The analysis was pre-registered and departures from the planned analysis were well-motivated and clearly described.

Thank you for this positive statement. We would like to make sure that the nature of our pre-registration is fully understood: we did not formally pre-register our study (i.e., there was no independent peer review). Instead, we defined an analysis plan ex ante (i.e., before beginning the data analysis for examining drug effects), and time-stamped and uploaded this plan on our institutional Git repository, prior to the unblinding of the analysing researcher. This a priori analysis plan is publicly available as well as our analysis code, and we report any departures from the analysis plan in our manuscript.

Reviewer #1 (Public Review):

The reduced amplitude of the mismatched negativity (MMN) in Schizophrenic patients has been associated with NMDA receptor malfunction. Weber and colleagues adjusted the systemic levels of two neurotransmitters (acetylcholine and dopamine), that are known to modulate NMDA receptor function, and examined the effects on mismatch related ERPs. They examined mismatch related ERPs elicited during a novel passive auditory oddball paradigm where the probability of hearing a particular tone was either constant for at least 100 trials (stable phases) or changed every 25-60 trials (volatile phases). Using impressive statistical testing the authors find that mismatch responses are selectively affected by reduced cholingeric function particularly during stable phases of the paradigm, but not by reduced dopamine function. Interestingly neither enhanced cholingeric or dopamine function affected MM responses at all. While the presented data support the main conclusions mentioned above, there are some claims in the abstract and text that are not supported by the results.

1) The authors state in the abstract that "biperiden reduced and/or delayed mismatch responses......", while the results (Figure 2) support the statement that biperiden delayed mismatch responses, the claim that biperiden reduced mismatch responses is misleading as on P13 the authors actually report that "mismatch signals were stronger in the biperiden group compared to the placebo group at right central and centro-parietal sensors" around 200ms. This is close both in time and spatially to the traditional temporal and spatial locations of the MMN component. If one were to only read the abstract they would take away the result that the muscarinic acetylcholine receptor antagonist biperiden has an attenuative effect on MMN which is not what the results show.

Thank you for this comment. We agree that the description in the abstract might be misleading and have changed our wording there. We now say (in the overall shortened abstract):

“We found a significant drug x mismatch interaction: while the muscarinic acetylcholine receptor antagonist biperiden delayed and topographically shifted mismatch responses, particularly during high stability, this effect could not be detected for amisulpride, a dopamine D2/D3 receptor antagonist.”

2) The conclusion that biperiden reduced mismatch responses may be due to the finding that at pre-frontal sensors mismatch responses were significantly smaller in the biperiden group than in the amisulpride (a dopaminergic receptor antagonist) group (P9) around 164ms. However, it is difficult to interpret if this is a meaningful result as amisulpride was found not to significantly alter mismatch responses in any way compared to placebo. It would be more convincing if the significant difference here were between biperiden and placebo groups. Or are we to think of amisulpride as being comparable to a placebo?

We agree with your previous point and have adjusted our wording in the abstract accordingly (see response to previous comment).

Furthermore, we have included an additional section in the Discussion in which we address the points you raise:

"One might wonder whether the early difference between the biperiden and the amisulpride group at pre-frontal sensors is difficult to interpret, given the lack of differences of either drug group compared to placebo. However, given our research question – i.e., whether auditory mismatch signals are differentially susceptible to muscarinic versus dopaminergic receptor status – showing a significant difference between biperiden and amisulpride is critical.

Clearly, such a differential effect would be even more compelling if biperiden differed significantly from amisulpride and placebo at the same time (and in the same sensor locations). While we do not find this in our main analysis, we do see it for the analysis using the alternative pre-processing pipeline and the trial definition (Figure 2—figure supplement 3) that was also specified a priori in our analysis plan. In this alternative analysis, mismatch responses under biperiden did differ significantly from both placebo and amisulpride."

We suspect this difference in results between the analysis pipelines might partly be due to the different re-referencing. Compared to the average reference used in the main analysis, the linked mastoid reference in the alternative pre-processing pipeline subtracts the effects at sensors which show positive mismatch signals from those at fronto-central channels (with opposite sign), effectively enhancing the signal at the fronto-central channels (for evidence of this effect see also current Figure 3—figure supplement 1) but weakening it at temporal and pre-frontal sensors.

We now discuss the question of sensitivity of both our paradigm and processing strategy in the discussion.

3) The authors use the words mismatch negativity (MMN) and mismatch responses interchangeably however in some cases it is clearly mismatch responses being described and not the classical MMN ERP component. This occurs especially in the Introduction where the authors describe the study and that they plan to focus on the MMN but in the results section, since the initial analysis focuses on all sensors, other mismatch responses are consistently discussed. These differences in wording need to be precisely defined and used consistently in the text.

We agree that it is important to use precise definitions of the terms and be consistent in their use. The dipole source signal of mismatch detection shows up with different signs across different sensor locations, and “MMN” traditionally refers to the effect in fronto-central channels, where it is a deviant-induced negativity. However, even when we constrain the use of “MMN” to the (difference in) negative deflection at fronto-central channels between 100 and 250ms (or similar) there remains some ambiguity due to the choice of reference. A common choice in MMN research is a linked mastoid reference. Because the mismatch signal shows up at the mastoids with opposite sign to fronto-central channels, this reference maximizes the observed difference at fronto-central channels (see also our Figure 3—figure supplement 1 and our reply to the previous comment) and minimizes it elsewhere, effectively forcing all (drug or other) effects to show up at frontocentral channels. This demonstrates that we typically think of the effects at different sensor locations as (caused by) one and the same (dipole source) signal. In our average referenced data (our main analysis), we observe some effects at fronto-polar sensors, where they are expressed as a modulation of a positive deflection, however, we think of these as being part of what is typically referred to as “MMN” for the above reasons.

However, to avoid any confusion that this may cause, we have adapted the wording in our manuscript everywhere and mention this distinction in the methods section:

“To avoid confusion, we will only use the term “MMN” when we talk about effects in the classical time window (100-200ms) and sensor locations (frontocentral sensors) for the MMN, and use “mismatch responses” for all other effects.”

4) A weakness of the paper would be that the authors offer no prediction in the Introduction about what the expected effects of these specific neurotransmitter modulations would be on mismatch responses.

Thank you for this suggestion and apologies for this oversight. We have now added a sentence to the Introduction, describing the effects we expected based on previous literature.

“Based on previous literature, one would expect mismatch responses in our paradigm to be sensitive to (1) volatility, with larger mismatch amplitudes during more stable phases (Dzafic et al., 2020; Todd et al., 2014; Weber et al., 2020), and (2) cholinergic manipulations, with galantamine increasing and biperiden reducing mismatch amplitudes (Moran et al., 2013; Schöbi et al., 2021). Furthermore, we expected a differential effect of cholinergic (muscarinic) and dopaminergic receptor status on mismatch responses, as postulated by initial work on MMN-based computational assays (Stephan et al., 2006). Our results suggest that muscarinic receptors play a critical role for the generation of mismatch responses and their dependence on environmental volatility, whereas no such evidence was found for dopamine receptors.”

5) A nice aspect of this paper is that the authors re-analyzed their data using pre-processing settings identical to those used in comparable research papers examining the effect of cholinergic modulation on MMN. The main findings did not differ following this re-analysis.

Reviewer #2 (Public Review):

The authors found that Biperiden (M1 antagonist) delayed and altered the topography of MMN responses, particularly in the stable condition. Amisulpride did not do so, and neither did Galantamine or L-DOPA. The analysis using an ideal Bayesian observer (the HGF) detailed in the Appendix showed that Biperiden reduced the representation of lower-level prediction errors and increased that of higher-level prediction errors (about volatility).

The methods were rigorous (including obtaining drug plasma levels and detailing alternative preprocessing techniques) and I have no suggestions for improvement from that point of view.

I only have one main comment that I think could be discussed. I'm not an expert on this but as I understand it, Olanzapine is most selective for M2 receptors rather than M1 (https://www.nature.com/articles/1395486), although Clozapine metabolites do have some M1 selectivity (https://www.pnas.org/content/100/23/13674) - I'm not sure about Clozapine itself. So Biperiden (very M1 selective) might not be the ideal drug to use to explore a treatment allocation paradigm, at least for Olanzapine? I suspect the options are quite limited but it would probably be worth commenting on this.

Thank you for pointing this out, this is indeed an important point for the discussion.

First, clarifying the pharmacodynamics of psychopharmacological drugs and their relative affinity to different receptor subtypes is notoriously difficult as this depends on many methodological factors. The seminal paper on the binding profile of olanzapine (which, at the same time, also examined clozapine) is (Bymaster et al., 1996). Using in vitro assays, this study found that both olanzapine and clozapine showed by far the greatest affinity for the M1 receptor (see the Table 5). By contrast, using SPECT data from seven patients with schizophrenia treated with olanzapine, the paper you mentioned (Raedler et al., 2000) estimated the affinity of olanzapine to the M2 receptor as being roughly twice as high as to the M1 receptor. Both studies have methodological pros and cons (as discussed by (Raedler et al., 2000)). From our view, an important limitation by the study of (Raedler et al., 2000) is that they used the ligand [I-123]IQNB which is not selective and "does not allow discrimination between the different subtypes of the muscarinic receptors" (Raedler, Knable, Jones, Urbina, Gorey, et al., 2003). Instead, the M1/M2 comparison by (Raedler et al., 2000) rested on conclusions from a mathematical approximation – under various assumptions and with only 7 data points available. We note that subsequent studies by the same group on muscarinic receptors in schizophrenia (Raedler, Knable, Jones, Urbina, Egan, et al., 2003; Raedler, Knable, Jones, Urbina, Gorey, et al., 2003) no longer used this approach and refrained from making statements about relative selectivity of olanzapine and clozapine with regard to M1/M2 receptors. Furthermore, the results by (Raedler et al., 2000) are potentially confounded by the fact that they were not obtained from healthy controls, but from patients with schizophrenia. This is potentially problematic: if schizophrenia is characterised by an aberration related to M1 receptors (see below), this would affect the interpretability of the results by (Raedler et al., 2000). Overall, the relative affinity of olanzapine and clozapine to M1/M2 receptors remains a matter of debate, but it seems safe to say that both drugs affect both receptors.

Second, we would like to explain that we think of biperiden as a model of a (potential) impairment, rather than a treatment. A series of studies have provided compelling evidence for a role of muscarinic (M1) receptor dysfunction in the pathophysiology of schizophrenia. In particular, there is compelling evidence for a subgroup of patients with markedly decreased M1 availability in the prefrontal cortex ((E. Scarr et al., 2009); see also (Gibbons et al., 2013) and (Elizabeth Scarr et al., 2018)). Moreover, multiple studies have found antipsychotic effects of xanomeline, an M1/M4 agonist (Bodick et al., 1997; Shekhar et al., 2008).

Against this background, clozapine and olanzapine may seem counterintuitive as treatment options since they antagonize muscarinic receptors. However, the muscarinic system is complex, and the mechanisms by which muscarinic receptors are involved in the therapeutic effects of clozapine and olanzapine are far from being understood. One interesting observation is that both clozapine and olanzapine have been found to elevate extracellular acetylcholine concentrations in cortical regions (Ichikawa et al., 2002; Shirazi-Southall et al., 2002), potentially by blocking muscarinic autoreceptors (Johnson et al., 2005), although this is debated (Tzavara et al., 2006). There is clinical evidence that clozapine or its metabolites may exert their pro-cognitive effects by increasing the release of actetylcholine (Weiner et al., 2004), and preclinical evidence that clozapine is able to normalize M1 receptor availability in cortex (Malkoff et al., 2008).

Irrespective of the exact mechanism by which clozapine and olanzapine exert their antipsychotic effects, their much higher affinity to muscarinic cholinergic receptors compared to dopaminergic receptors sets them apart from other antipsychotics. If a functional readout of the relative contribution of cholinergic versus dopaminergic deficits could be obtained in individual patients, this might be predictive of whether this patient would profit from clozapine, olanzapine, or, in the future, potential new treatments targeting the muscarinic system specifically.

Given the above considerations, we have amended the relevant paragraph in the discussion to state this rationale more clearly.

“Notably, there is compelling evidence for a subgroup of patients with markedly decreased M1 availability in the prefrontal cortex ((E. Scarr et al., 2009); see also (Gibbons et al., 2013) and (Elizabeth Scarr et al., 2018)). This is consistent with the possibility that a key pathophysiological dimension of the heterogeneity of schizophrenia derives from a differential impairment of cholinergic versus dopaminergic modulation of NMDAR function (Stephan et al., 2006, 2009). Distinguishing these potential subtypes of schizophrenia could be highly relevant for treatment selection, as some of the most effective neuroleptic drugs (e.g., clozapine, olanzapine) differ from other atypical antipsychotics (e.g., amisulpride) in their binding affinity to muscarinic cholinergic receptors. The exact mechanisms by which muscarinic receptors are involved in the therapeutic effects of clozapine and olanzapine are still under debate and include, for example, elevation of extracellular levels of acetylcholine in cortex (Ichikawa et al., 2002; Shirazi-Southall et al., 2002; Weiner et al., 2004), possibly via blocking presynaptic muscarinic autoreceptors (see (Johnson et al., 2005; Tzavara et al., 2006) for conflicting data), and normalization of M1 receptor availability in cortex (Malkoff et al., 2008). Irrespective of the exact mechanism by which clozapine and olanzapine exert their antipsychotic effects, their much higher affinity to muscarinic cholinergic receptors compared to dopaminergic receptors sets them apart from other antipsychotics. If a functional readout of the relative contribution of cholinergic versus dopaminergic deficits could be obtained in individual patients, this might be predictive of whether this patient would profit from clozapine, olanzapine, or, in the future, potential new treatments targeting the muscarinic system specifically. Indeed, muscarinic receptors have become an important target of drug development for schizophrenia (Yohn & Conn, 2018).”

Author Response:

Reviewer #2 (Public Review):

The manuscript addresses an important question regarding sensory processing related to self-motion. The main experiment is clearly described and demonstrates that neurons display a diversity of responses from purely reflecting vestibular input (head-in-space motion) to predominantly body motion, and any combination between. Of particular interest, is that the response of the Purkinje cells are profoundly different than its downstream target, the fastigial neurons which signal only head-in-space or body motion. This substantive difference in neural representations between these two connected brain regions is surprising.

The manuscript also provides a simple population model to show that fastigial responses could be generated from Purkinje cell activity, but only from combining at least 40 neurons. While the model provides some insight on the potential interaction between Purkinje cells and fastigial neurons, I think the model assumes no other input to the fastigial neurons. However, I would assume that there is likely a strong input from mossy fibers onto the fastigial neurons that also target the Purkinje cells. This mossy fiber input will certainly provide vestibular and neck proprioceptive input to the fastigial nucleus. Thus, the Purkinje cell input may be essential for countering the mossy fiber input leading to separate representations for head and body motion in the fastigial nucleus.

We agree this is an important point. To address the reviewer’s concern, we performed additional modeling in order to consider the influence of mossy fiber inputs. Specifically, following the reviewer’s suggestion below, mossy fiber input was modeled using random patterns of vestibular and neck proprioceptive input. Prior studies have shown that the dynamics of vestibular nuclei neuron responses strongly resemble those of unimodal fastigial neurons in rhesus monkeys (i.e., they encode vestibular input and are insensitive to neck proprioceptive inputs, Roy & Cullen, 2001). In contrast, reticular formation neurons responses to such yaw head and/or neck rotations have not yet been described. We therefore simulated mossy fiber input first as a summation of vestibular and neck proprioceptive inputs, for which the gains and phases were randomly drawn from a distribution, comparable to that previously reported (Mitchell er al. 2017) in the vestibular nuclei (Fig. 7-figure supplement 3). We then further explored the effect of systematically altering this simulated mossy fiber input - relative to the reference distribution of mossy fiber inputs - by i) doubling the gain, ii) reducing the gain by half, iii) doubling the phase, and iv) reducing the phase by half (Fig. 7-figure supplement 4). Overall, we found that the addition of such simulated mossy fiber did not dramatically alter our estimate of the population Purkinje cell population size required to generate rFN neurons responses (~50 versus 40; Fig. 7-figure supplement 3&4).

Another issue is the limited number of neurons recorded in the secondary experiment with only 12 bimodal neurons and 5 unimodal (although there appears to be only 4 neurons in Figure 5C). Such a small sample impacts the estimated tuning properties of Purkinje neurons in Figure 5D and the results from the population model. This needs to be clearly recognized.

We have revised the RESULTS to clarify the numbers of Purkinje cells that were tested (13 bimodal and 4 unimodal Purkinje cells). For comparison, in our Brooks and Cullen study, tuning curves were computed for 10 bimodal and 12 unimodal rFN. We note that i) unimodal Purkinje cells make up a relatively small percentage of anterior vermis Purkinje cells and ii) similar to unimodal rFN, our small sample of unimodal 9 Purkinje cells did not demonstrate significant tuning. In contrast, all bimodal Purkinje cells in our sample demonstrated significant tuning. To simulate responses for the bimodal Purkinje cells that were not held long enough to test during gain-field paradigm (i.e., Fig 5), we generated tuning curves drawn from a normal distribution estimated from 13 bimodal Purkinje cells. We appreciate this was not clear in the original submission and have revised the METHODS section to clarify our approach. Overall, while we recognize that our sample size is small, we nevertheless found it interesting that including this our results from this protocol did not increase the estimated population size relative to that estimated using our other dynamic protocols.

Reviewer #3 (Public Review):

In this study, the authors characterize the simple spike discharges of Purkinje cells in the anterior vermis of the macaque during passive vestibular and neck proprioceptive stimulation. The activity of most Purkinje cells encoded both vestibular (whole-body rotation) and proprioceptive (body-under-head rotation) stimuli. Although the vestibular and proprioceptive responses were, on average, antagonistic in the preferred direction, consistent with a partial transformation from head to body coordinates, response properties for both modalities were highly variable across neurons. Most cells responded under combined vestibular and proprioceptive stimulation (head-on-body rotation), and these responses were well-approximated by the average of the responses to each modality individually. Vestibular responses exhibited gain-field-like tuning with changes in head-on-body position, though these changes were significantly smaller than the shifts observed for neurons downstream in the rostral fastigial nucleus. Finally, a weighted average of the responses of approximately 40 Purkinje cells provided a good fit to the responses of postsynaptic fastigial neurons.

Overall, these results provide important and novel insights into the implementation of coordinate transformations by cerebellar circuitry. The experiments are well-designed, the data high quality, the analyses reasonable, and the conclusions justified by the data. The manuscript is clear and well-written, and will be of interest to a broad neuroscientific audience. I have no major concerns. I have a few minor suggestions for improving this manuscript, described below.

1 - The authors may wish to discuss earlier work in the decerebrate cat by Denoth et al. (1979, Pflügers Archiv), which provided evidence that the responses of Purkinje cells in the anterior vermis to head-on-body tilt is relatively well-approximated by averaging the responses to neck and macular stimulation alone.

We thank the reviewer for bringing this reference to our attention and have revised the INTRODUCTION and DISCUSSION to include the early work of Denoth et al.,1979.

2 - To better convey the heterogeneity of responses across the sample of Purkinje cells, two additional supplemental figure panels might be useful: (1) the vestibular, proprioceptive, summed, and combined sensitivities in each direction (as in the Fig. 3C insets) for each individual neuron (perhaps as a series of subpanels), and (2) scatterplots of response phase for proprioceptive vs vestibular stimulation for bimodal neurons (with separate panels for preferred and non-preferred directions).

We agree that this is a useful way to emphasize the heterogeneity of bimodal Purkinje cells responses and have added the requested response phase scatterplots for proprioceptive vs vestibular stimulation (Fig 2 - figure supplement 2C&D). We have also made a figure showing the summation model for each individual neuron. However, because our Purkinje cell population included 73 neurons, this figure includes a corresponding 73X2 =146 polar plots (i.e., two plot each cell, one for ipsi and contralateral motion). Given the immense size of this figure, we elected not to include this figure in the supplementary material in the revised manuscript.

3 - Can the authors provide additional information on the approximate location of the recorded neurons (lobule and zone or mediolateral position)? Is it possible that some project to the vestibular nuclei, rather than the rFN? This consideration seems especially relevant for the interpretation of the pooling analysis in Fig. 6, which seems to assume that Purkinje cells are sampled from a sagittal zone with overlapping projections in the rFN (or, at least, that the response properties of the sampled neurons are representative of the properties in a corticonuclear zone). Some additional discussion on this point would be helpful.

The recorded neurons were located in the lobules II-V of the anterior vermis, ~0 to 2 mm from the midline. We now include this information in the revised METHODS. As noted by the reviewer, Purkinje cells in this region of the anterior vermis project to the vestibular nuclei as well as to the rFN (Voogd et al. 1991). Nevertheless, using comparable stimulation protocols, we have previously shown that the responses of vestibular nuclei neurons are comparable to those of unimodal rFN neurons (Brooks et al., 2015). Specifically, both vestibular nuclei and unimodal rFN neurons are insensitive to proprioceptive stimulation and demonstrated comparable responses to vestibular stimulation. Thus, our present modeling results regarding the population convergence required to account for unimodal rFN neurons can be directly applied to vestibular nuclei neurons. We have revised the DISCUSSION to consider this point.

4 - When weighted averages of Purkinje cell responses are used to model rFN responses, my intuition would be that w_i is near zero for v-shaped and rectifying Purkinje cells. That is, the model would mostly ignore them, as data from both directions appear to be included. Is this the case? A more detailed description of the fitting procedure would also be helpful.

To address the reviewers’ concerns regarding the Purkinje cell weights, we have added a new inset to Fig 7C. As can be seen, model weights are well distributed across different Purkinje cells. Further, to confirm that the distribution of the weights of Purkinje cells inputs are distributed over different classes of PCs we now illustrate the weight distributions for (a) linear vs. v-shaped vs. rectifying Purkinje cells, (b) bimodal vs. unimodal Purkinje cells, (c) Type I vs. Type II Purkinje cells and (d) Purkinje cells with agonistic vs. antagonistic vestibular and proprioceptive sensitivities. These results are shown in Figure 7-supplemental figures 1&2. Overall, we found that distribution of the weights was not biased towards linear cells, but rather were similarly distributed across all three groups. This was true for our modeling of both bimodal and unimodal rFN cells (compare Fig 7- figure supplement 1 vs. Fig 7- figure supplement 2). As can be seen in this Figure, we likewise found comparable results for the weights of Type I vs. Type II Purkinje cells, unimodal vs. bimodal Purkinje cells, and/or vestibular / proprioceptive agonist vs. antagonist bimodal neurons. Finally, as detailed above in our response to the reviewers’ consensus comments, we have also revised the METHODS section to provide a more detailed description of linear regression method.

5 - Another potential interpretive issue in the averaging analysis concerns the presence of noise on single trials. The authors could briefly comment on whether more Purkinje cells might be needed to predict rFN responses on a single trial in real time.

This is an interesting question; we have revised the DISCUSSION to consider this point.

Author Response:

Reviewer #1:

The aim of this paper to reveal the mechanisms that establish the Wnt gradient combining a mathematical model and experiments is of general importance. The results of computer simulations and biological experiments are interesting because they consider multiple extracellular components. They successfully demonstrated that the ligand/receptor feedback and the other extracellular components shape the morphogen gradient of Wnt ligand so that the fine patterning found in heart development can be explained. However, I feel that quantification of the experimental data, explanation of the mathematical model and discussion of the results are not sufficient in the current manuscript.

Major points:

1. Experimental validation of the results of computer simulations is very important in this study. However, many of experimental data were not properly quantified or statistically tested. The authors would need to quantify the experimental results when appropriate and perform statistical tests (e.g. Figs. 1E, 2A, 4A-B, Supplemental Figs. 6, 7).

We are sorry for the lack of quantitative and statistical analyses in many experiments. We revised all the points (graphs and statistical analyses in Figs 1, 2, 4; Figure 1-figure supplement 1; Figure 3-figure supplement 7; Figure 4-figure supplement 1, 2).

1. Design of the mathematical model is not sufficiently explained in the main text. Besides details in the method section, the basic design of the model and simulation should be briefly explained. For example, initial distribution of Fzd7, regions that produce Wnt6 and sFRP1, and interpretation of the simulation results should be added for Fig. 3 (page 10, line 11-16).

We are sorry for the inconvenience. In this revision, we wrote the basic design of the model and simulation in the main text.

As an interpretation of the simulation results, we added an explanation as follows:

The Wnt signaling gradient became steeper with increased feedback strength. Considering a threshold of signal activation (Fig. 3A, dashed line), feedback results in restriction of the Wnt-activated region.

1. The authors demonstrated the roles of Wnt6/Fzd7 feedback and sFRP/Heparan sulfate binding. A typical simulation data showing the roles of sFRP and Heparan sulfate would need to be shown in the main figure.

Thank you for your suggestions. We moved a typical result of sFRP/HS simulation from the original supplemental figure to a main figure (Fig. 4G).

Unfortunately, they did not sufficiently discuss their actions using the mathematical model. They would need to at least qualitatively discuss these points. How do they control Wnt gradient? What are the roles of these two mechanisms? What are the difference? How do they influence with each other? Simplified models may be necessary to reveal the relationship between these two mechanisms and to gain mechanistic insights.

Thank you for pointing out these critical points.

For Wnt gradients, receptor feedback, sFRP, and HS are synergistically acting for the restriction of signal activated region (steep gradient).

However, there are some differences. The receptor-feedback can overcome the variation of Wnt production but sFRP1 and HS cannot because sFPR1 expression is inhibited by Wnt, which forms a positive feedback loop for Wnt signaling (Gibb et al., 2013). Thus, sFRP1/HS cannot buffer the variation of Wnt production.

In this revision, we added these explanations.

[They will influence each other] Because sFRP1 inhibits Wnt signaling, sFRP1 reduces fzd7 expression. This occurs mainly in the right side (because sFRP1 is expressed in the right side), resulting in a short-range activation of Wnt signaling.

Deeply considering your comments, we recognized that we did not describe sFRP1/HS function in the title of the previous version. We revised it as follows:

Previous) Positive Feedback Regulation of fzd7 Expression Robustly Shapes Wnt Signaling Range in Early Heart Development

Current) Positive feedback regulation of fzd7 expression robustly shapes a steep Wnt gradient in early heart development, together with sFRP1 and heparan sulfate

Additionally, the situation studied in this paper would need to be compared with the other examples of ligand/receptor feedback, and the similarity and difference should also be discussed (e.g. Hedgehog/Patched and Wingless/Frizzled2 in the fly wing).

Thank you for your helpful comments.

As you mentioned, the gene regulatory circuit of our Wnt6/Fzd7 is similar to that of Hedgehog (Hh)/Patched (Ptc): both of the morphogens commit self-enhanced degradation via induction of receptor expression (Eldar et al., 2003; Hh induces Ptc expression, and this increases Hh degradation). In the case of Wingless/Frizzled2, the gene regulatory circuit is different from that of Wnt6/Fzd7: Wingless commits self-enhanced degradation via repression of receptor expression. Wingless inhibits Fzd2 expression, and Fzd2 inhibits Wingless degradation. Both gene regulatory circuits function as a robust system for morphogen variations (Alon, 2006).

There is also a little difference between Wnt6/Fzd7 and Hh/Ptc. In the Hh, the receptor Ptc inhibits downstream signaling. Thus, the network of Hh restricts the ligand distribution as is the case with Wnt, but the signal activity is not as steep as Wnt (highly Ptc expression inhibits the signaling).

We added these explanations.

Reviewer #2:

In this work, the authors tried to understand the effect of receptor and diffusible inhibitors on the Wnt morphogen gradient during heart development by combining experiment and computational modeling. The experimental part seems to be a solid contribution to this academic field, and I appreciate the interdisciplinary attempt to combine the results with the computational model. However, their results may be interpreted more clearly using classical mathematical models.

First of all, we greatly thank you for evaluating our manuscript. And thank you very much for explaining classical models in detail.

1. Classical models may be enough.

Previous mathematical models provided stronger predictions than numerical simulations, and I am not sure numerical results provided by the authors give us new insights. For example, Eldar et al. (2003) have provided analytical results on why the concentration becomes robust. In normal SDD model

u'(x,t) = -d_1 u(x,t) + d_u \Delta u(x,t),

the steady-state solution is exponential function,

u_s(x) = u_0 exp(- \sqrt (d_1/d_u)x)

, and the amount of morphogen production at the boundary critically affects the result (If the production becomes 1/2, the concentration becomes 1/2 everywhere). On the other hand, if the degradation is promoted by the morphogen itself (in this case, by the upregulation of the receptor expression), the governing equation becomes

u'(x,t) = -d_2 u(x,t)^2 + d_u \Delta u(x,t),

the solution is

u_s(x) =A/(x+x_b)^2

($A$ and $x_b$ are constants determined by $d_u$ and $d_2$). It converges to

u_s(x) =A/x^2

and the morphogen gradient profile does not change much when the morphogen production is relatively high (that means there is a condition to be robust).

Similarly, a linear approximation is enough to understand the diffusion length change - diffusion length of the morphogen gradient (the length necessary to become morphogen concentration 1/e) is in general $\sqrt{D_u /d_1}$, and feedback mechanism should increase d_1 in first-order estimation, hence decreasing the diffusion length. Binding to HSPG may have a similar effect (in the case of FGF, HSPG is necessary to the binding of FGFR, and the situation is very different).

Thank you again for your explanations. Our explanations in the previous manuscript were not enough.

–Difference of our computational simulation and the classical analysis:

We think we need numerical simulation to consider points not addressed with previous analytical methods. The following two points are the new points that are too complicated to handle with analytical methods.

1. Transient state is considered, which is hard to analyze without computer simulation.

Considering the in vivo situation, we cannot determine whether the fate determination takes place at a transient or steady state (as described in page 7, line 14). So, we analyzed it not limited to a steady state but including transient state in our simulation.

1. Receptor has multiple functions in interaction with multiple molecule species: (i) binds to the ligand and restricts the ligand spreading, (ii) activates the intracellular signaling, and (iii) degrade the ligand (new Supplementary Fig. 1A). We would like to include these different functions separately in the simulation. In addition, we considered sFRP1 and N-acetyl-rich HS. Thus, we need a multivariate nonlinear reaction-diffusion equation, which is hard to handle without computer simulation.

To clarify these points, we added an explanation of the multiple receptor functions with a schematic figure (Supplementary Fig. 1A).

–Importance/significance of our simulation:

We first confirmed that our simulation reached a similar conclusion as the classical simulation at a certain time point (~ 1 day after the onset of simulation): the network was robust against variation of Wnt production. In addition, examining the time change of activation level, we have found that this network is robust against changes in speed of the differentiation. We added these explanations.

1. Biological example of Wnt fluctuation

The authors examine the effect of Wnt production fluctuation, but their motivation is not clear. Eldar et al. (2003) is motivated by the fact that the Shh heterozygote knockout has no phenotype, although the amount of mRNA is halved. Theoretically, it should have a major effect on the organs utilizing the Shh morphogen gradient (actually, haploinsufficiency is observed, but the phenotype is mild). The authors would need to provide some argument why they are interested in the robustness to the Wnt expression fluctuation.

We all agree with your opinion. Compared with Eldar et al. (2003), our motivation is not clear to set 50% for the variance of ligand production.

It is generally accepted that gene expression is different between individuals. In contrast, the proportions of the patterned tissues are almost the same among individuals.

We examine this general question in our specific example of Wnt production. Here we focused on an extreme example (50% increase) among various sizes of gene expression.

We added a phrase “as an extreme case” to clarify that it is an example in the revised manuscript.

1. Wnt signal distribution

It is difficult for general readers to understand why the Wnt signal distribution in the simulations (0 around 0-10 µm, Sudden disappearance at 40 µm) is appropriate. The authors can provide the profile plot of the actual measurement, which corresponds to the modeling result.

Sorry for this inconvenience. As indicated in Figure 1—figure supplement 1B, Fzd7 shows a limited expression in pericardium. Fzd7 expression was not detected in epidermis (Figure 1—figure supplement 1B), which is the Wnt source (Lavery et al., 2008), indicating that the sudden increase of Fzd7 expression near Wnt source (at x = 10 μm) is reasonable (because the amount of Wnt at x = 10 μm is considered to be above the threshold for Fzd7 expression). In the prospective myocardium region, Fzd7 expression was also disappeared suddenly (Figure 1—figure supplement 1B), suggesting that the activity of Wnt signaling is also disappeared suddenly in the region. We added the explanations.

In addition to the indirect estimation of Wnt signaling from Fzd7 expression, to directly confirm the “sudden disappearance” of Wnt signaling, we tried following three ways, but they failed. We examined (i) a transgenic reporter line of Wnt signaling (TCF-promoter-driven GFP) and (ii) immunohistochemistry (IHC) of beta-catenin (nucleus localization of beta-catenin is an indicator of the activation of Wnt signaling) and (iii) IHC of active beta-catenin (which only detect the active form of beta-catenin), expecting more gradual signal distribution, compared to the readout of Fzd7 expression which may have a threshold to express. But (i) the background signal was high in the transgenic. (ii) The background signal was also high with IHC maybe because beta-catenin is abundant also in the cytoplasm in heart region. (iii) The signal of active beta-catenin was not changed by Wnt addition in Xenopus.

In addition, about the width of wnt6 and fzd7 expression, we measured the actual size of the fzd7-expressed region (Figure 1—figure supplement 1B), which was around 32 μm. It was almost the same as that in the model (30 μm). The width of Wnt6-expressed region was set to be 10 μm following a previous report (Lavery et al., 2008). We added explanations for the width of the expressions.

1. Variable "Wnt signal"

It is not clear what the variable "Wnt signal" means. As far as I understand, the signal inside the cell changes quickly (in the case of FGF, the ERK phosphorylation state changes within a minute). The author should provide a concrete example of this "Wnt signal" (maybe mRNA expression of some marker gene?).

We agree with your opinion. As an indicator of Wnt signal activation, we think of the translocation of β-catenin (a transcriptional regulator) into the nucleus. Indeed, the translocation is observed at least in a 15 min and concurrently the transcription of the target gene is observed (Kafri et al., 2016), suggesting this translocation (the activation of the signal in the cells) is recognized enough by the cells within a minute. We added this explanation.

1. Use of BMP measurement values.

In addition, I am not sure whether using BMP values for the estimate of Wnt dynamics is appropriate. I have an impression that BMP is a fast-diffusing molecule that has a less binding affinity to ECM compared to FGFs. Although I have not dealt with Wnts, they are reported to bind strongly to ECM.

Thank you for the comments. In this revision, we used all of the reported Wnt values. According to this parameter change, we performed computer simulation again. All the conclusions were not changed.

Reviewer #3:

A summary of the study and the strengths of this manuscript: The authors found several new molecular interactions that may be essential for understanding the mechanism of steep gradient formation of Wnt ligands in the prospective cardiac field.

One of the new findings is that expression of a Wnt receptor, Frizzled7, in the prospective heart field is activated by Wnt/b-catenin signaling, as well as by Wnt6 ligands, which is involved in the patterning of this field. They also found that the diffusing Wnt6 ligand is trapped at the surface of cells in which Frizzled7 is ectopically expressed. It seems reasonable that the combination of signal-dependent receptor expression and receptor-dependent ligand capture would result in a steep gradient of morphogen molecules. In fact, this idea is supported by mathematical modeling. In addition, this modeling suggests that the receptor feedback mechanism provides robustness to morphogen-mediated patterning against fluctuations in morphogen production.

Another highlight of their study is that the soluble Wnt antagonist, sFRP1, specifically binds to N-acetyl HS, and this modification of HS is specifically detected in the outer of the cardiogenic field. The localized N-acetyl HS may also be involved in Wnt gradient formation by inhibiting Wnt signaling around myocardium region.

The weaknesses of this manuscript: Although the issue they address in this manuscript is very important for understanding the mechanism of morphogen-based tissue patterning, most of the experimental data presented in this manuscript are preliminary.

We added and revised many experiments (including computational analysis) in this revision. In particular, in Figs 1, 2, 4; Figure 1-figure supplement 1; Figure 3-figure supplement 7; Figure 4-figure supplement 1, 2.

Therefore, interpretations other than the ones they have argued for in this manuscript are quite possible. any other interpretations except those they claimed in this manuscript are still possible.

For example, the authors argue that receptor feedback is essential for the formation of steep Wnt gradients (lines 8-9 in the abstract), but their model does not rule out an alternative possibility that high levels of receptor expression in the cardiogenic field form steep gradients.

We agree.

As you mentioned, high levels of receptor expression can form steep gradients. In a case distributions are similar with and without feedback, the changes in the boundary position in response to Wnt production change seemed smaller with feedback than without (Fig. 3B), providing a possibility that feedback has higher robustness to the variation.

These explanations were poor in the previous version. We added explanation.

In addition, it would be a waste of energy because too much receptor expression is needed. If the initial expression of receptor is critical for the patterning (not the receptor feedback), the amount and the area should be tightly controlled by an additional mechanism.

We added these explanations to the result and discussion sections.

Furthermore, they have not succeeded in directly examining the effect of receptor feedback on Wnt6 gradient formation. Although the data shown in Supplementary Figure 6E appear to support the contribution of feedback mechanisms to patterning, the results do not exclude another interpretation that an increase in Wnt trapper molecules simply inhibits the receptor-mediated clearance of Wnt ligands from the extracellular space in the pericardial region, resulting in an increase of extracellular Wnt ligands and their long-range transport.

Thank you for your comment. As you mentioned, the Wnt trapper inhibits clearance. However, at the same time as it inhibits clearance, it also inhibits diffusion of Wnt. These two inhibitions happen simultaneously for the same duration. Thus, the trapper will not promote long-range transport via competitive inhibition of the Wnt clearance.

Thus, from the results using the trapper, we can conclude that the receptor expressed after the activation of Wnt signal (not the initial amount of receptor) is critical for determining the range of Wnt signaling (e.g. the width of the resulting pericardium).

We added these explanations in the new text.

With regard to the restriction of sFRP1 diffusion, no evidence has been presented to show that N-acetyl modification of HS is actually involved in the restriction of sFRP1 diffusion, the formation of Wnt gradient, and the patterning of prospective cardiac fields. This lack of data significantly undermines the credibility of the conclusions presented in this paper.

We performed a new experiment.

We overexpressed Ndst1 enzyme that modifies N-acetyl to N-sulfo HS to eliminate N-acetyl HS, and analyzed if heart patterning is changed. We revealed that Ndst1 expression results in a reduced pericardium but an increased myocardium region, suggesting that N-acetyl HS promotes pericardium differentiation and inhibits myocardium differentiation.

We added these explanations and figures (Fig. 4F; Figure 4-figure supplement 2A-C).

A list of all the questions that Vannevar Bush poses in the piece:

• What are the scientists to do next?
• Of what lasting benefit has been man's use of science and of the new instruments which his research brought into existence?
• Is this all fantastic?
• Will there be dry photography?
• What would it cost to print a million copies?
• The preparation of the original copy?
• To consider the first stage of the procedure, will the author of the future cease writing by hand or typewriter and talk directly to the record?
• Is it not possible that some day the path may be established more directly?
• Might not these currents be intercepted, either in the original form in which information is conveyed to the brain, or in the marvelously metamorphosed form in which they then proceed to the hand?
• Is it not possible that we may learn to introduce them without the present cumbersomeness of first transforming electrical vibrations to mechanical ones, which the human mechanism promptly transforms back to the electrical form?
• True, the record is unintelligible, except as it points out certain gross misfunctioning of the cerebral mechanism; but who would now place bounds on where such a thing may lead?
• Must we always transform to mechanical movements in order to proceed from one electrical phenomenon to another?

Author Response:

Evaluation Summary:

This work provides new insights into how surface-exposed lipoproteins of Gram-negative bacteria reach their destination in the outer membrane. Authors find that the outer membrane protein complex Slam serves as a translocon for the lipoproteins and the periplasmic chaperone Skp mediates their targeting to Slam. This work may contribute to the elucidation of host invasion mechanisms by pathogenic bacteria, in which surface lipoproteins play an important role.

Reviewer #1 (Public Review):

Previously, using rigorous genetic, bioinformatic and cell-based biochemical analyses, the same group discovered SLAM1, an outer membrane protein in Neisseria spp., which mediates the membrane translocation of surface lipoproteins (SLPs) (Hooda et al. 2016 Nature Microbiology 1, 16009). Here, authors reconstituted this system in proteoliposomes using minimal purified components including the translocon Slam1 and the client lipoprotein TbpB. Authors further coupled the system to TbpB-expressing E. coli spheroblasts and LolA, the Slam1-specific periplasmic shuttle system. Using the digestion pattern of TbpB by Proteinase K as a readout, authors confirmed that Slam1 indeed served as a translocon for SLPs. As a step forward, authors found that Skp, a periplasmic chaperone (holdase), was critical to the membrane-assembly and translocation of TbpB. Strengths: Overall, this is a solid biochemical study that demonstrates the role of Slam1 as a translocon for SLPs. The experimental design is neat and straightforward. The specific role of Skp in SLP translocation is interesting. This reconstituted system will serve as a novel platform for further elucidation of the Slam1-mediated SLP translocation mechanisms. The manuscript is overall well written. Weakness: There are several major concerns, however. 1) It is not fully convincing whether these findings are novel and significantly advance the field. Identification of minimal components in a biological process and their reconstitution are always challenging and thus, this study is an achievement. Nonetheless, I am not sure whether we have learned novel molecular insights besides the confirmation of the group's previous discovery. The specific role of Skp in translocation is interesting but not surprising, considering that periplasmic holdases are already known to be extensively involved in the biogenesis of periplasmic and outer membrane proteins.

We thank the reviewer for their time and thorough review of the manuscript. In the previous paper (Hooda et al. 2016 Nature Microbiology 1, 16009), we discovered that the outer membrane protein Slam is “important/responsible” for the surface display for SLPs (TbpB, LbpB, fHbp). In this mechanism focused manuscript, we were able to demonstrate Slam’s role as an outer membrane translocon. One of the achievements in this paper is to demonstrate that Slam as an autonomous translocon – importantly this is unlike the two-partner secretion systems, as it does not require the Bam complex for the translocation of TbpB.

2) Although authors developed nice assays (Figs. 1 and 2), it was not verified whether TbpB protected from Proteinase K digestion had "correct" conformation and membrane-topology. Authors performed a functional assay on TbpB (Fig. 5a), but this result was obtained from a cell-based assay, not from the reconstituted system.

We have performed pulldown assay for the TbpB that has been translocated into Slam-proteoliposomes using human transferrin conjugated beads to show that this TbpB protein is correctly folded and functional. Blots and explanations are attached in the revised manuscript (see new Figure 2 – figure supplement 2 and line 197-207). (As addressed in major scientific concerns point 2-i).

Although the data in Figs. 1 and 2 clearly show that the membrane association of TbpB depends on Slam1, it does not mean that the "translocation" has actually occurred in the proteoliposomes. Probably, more rigorous analysis on the Proteinase K-protected portion of TbpB (for example, mass spec) seems necessary (that is, whether the proteolytic product is expected based on the predicted topology).

The TbpB is flag-tag at its C-terminus and the protected band on our blots (detected by α-flag antibody) corresponds to the expected Mw (~75kDa) for Mcat TbpB flag tagged protein. Therefore, we believed the band at 75kDa is our full length processed TbpB. Moreover, we have confirmed that TbpB can be detected at the top of the sucrose gradient with our Slam-proteoliposomes in this assay. This would only occur if TbpB was actually translocated inside the intact liposomes, otherwise we should not see any TbpB in the top layer of the sucrose gradient (Figure 4d). Furthermore, we have performed a pulldown assay for TbpB in proteoliposomes to check for their functional binding to human transferrin beads after translocation. These results are explained in the updated new Figure 2 – figure supplement 2 and line 197-207.

3) The manuscript has a couple of missing supporting data. 3a) Lines 87-89: "From our analysis, we found that the Slam1 from Moraxella catarrhalis (or Mcat Slam1) expressed well and the purified protein was more stable than other Slam homologs." I cannot find the expression and stability data of various homologs supporting this sentence.

In general, what we meant was that we chose Mcat Slam as the target of this study because it is more stable during the purification and resulted in a higher yield of protein. We needed higher yields of Slam to be able to reconstitute the protein into the liposomes for the translocation assay. We have purification data for Mcat Slam1, Nme Slam1 and Ngo Slam2 but we think including them in the supplementary is not necessary. We have changed and rewritten this section dedicated to Mcat Slam1 purification (Figure 1 – figure supplement 1 and 2).

3b) "Lines 216-219: Furthermore, the processing of TbpB by signal peptidase II and subsequence release from the inner membrane was unaffected suggesting the defect in surface display by Skp occurs after the release of TbpB from the inner membrane (Fig. 4a)." The result supporting this sentence seems missing or this sentence points to a wrong figure.

Yes, this sentence is misleading. What we meant was that the processed TbpB (TbpB has 2 bands, unprocessed TbpB – upper band and signal peptidase processed, lipidated TbpB - lower band) is similar for all samples indicating that the knockout of Skp did not affect the expression or processing of the signal peptide of TbpB up until it is ready (processed and lipidated in the periplasm) for translocation by Slam to the surface. We have added an explanation in the figure legend of Figure 4a –line 267-269.

4) Some statistical analysis results are not clear, making some conclusions not convincing. 4a) Figure 4a top "Exposure of TbpB on the surface of K12 E. coli" Apparently, all three data points for (Delta_DegP+Slam1+TbpB) are very closely distributed. Accordingly, (WT+Slam1+TbpB) vs (Delta_DegP+Slam1+TbpB) data look significantly different (difference is ~0.2). But the two data were assigned as "Not Significant". Similarly, in the comparable in vitro data (Figure 4b), the intensity for Slam1 (WT+Proteinase K - Triton) looks larger than that for Slam1 (Delta_DegP + Proteinase K - Triton). So, the DegP contribution should not be ignored.

For figure 4a, the ONE WAY ANOVA test was performed using Prism with 4 biological replicates (we can include the analysis report in the revised submission if this is requested we have updated the figure to include data points. In general, both our in vitro liposomes translocation assay and in vivo surface exposure assay for TbpB showed that delta-DegP only slightly reduces the translocation of TbpB to the surface but could not detect statistically significant differences.

4b) Figure 5a top "Exposure of TbpB on the surface of N. meningitidis" What is the p-value for WT vs Delta_Skp data? Are the two data significantly different? The p-value range for (*) is not shown.

We have included the p-value range for (*) in the revised manuscript, figure 5a.

Reviewer #2 (Public Review):

The article addresses the function of SLAM, a protein which the authors have shown previously to be involved in the traffic of lipoproteins to the bacterial surface. The authors have performed a series of experiments to assess the impact of SLAM on the delivery into proteoliposomes of the model lipoprotein TbpB either added exogenously or presented by E coli spheroplasts. They identify a periplasmic chaperone, Skp, which enhances transport of TbpB and other lipoproteins to proteoliposomes, and show the contribution of endogenous Skp to lipoprotein transport in Neisseria meningitidis. The authors set up an in vitro translocation assays using purified components from different bacteria. This is reasonable as the assays can be challenging to establish and require proteins that can be expressed and are stable. It would be helpful however if the sources of the proteins and how they are tagged (for their detection) is clearly documented in the article and the figures. In keeping with this, the figures describing the assays could be improved (ie 1A, 2A, 3A and C). Despite this, the results presented in Fig 1 and 2 clearly demonstrate the role of SLAM as a translocase, and the authors have included appropriate controls for their assays; the translocation of a OmpA to demonstrate that the Bam complex is functional in their hands in an important control and should be included in the main figures. Experiments outlined in Figure 3 and Table 1 demonstrate the interaction specific of TbpB and another lipoprotein HpuA with Skp, a previously characterised periplasmic chaperone. This is performed by pull-downs and MS as well as immunobloting. A critical result is shown in Figure 4 in which SLAM and TbpB are introduced into E coli, and the role of endogenous Skp is assessed. Importantly, the absence of Skp reduces but does not eliminate TbpB surface expression. The authors could speculate on the nature of Skp-indendent surface expression of TbpB, as this result mirrors what they find in a meningococcal strain lacking Skp (Figure 5A). It appears that Skp might be required for the correct insertion/folding of lipoproteins given their result in Figure 5B (currently, this could be changed into 5C) which tests the binding of transferrin to the bacterial surface. Clearly this could be influenced by an effect of Skp on TbpA, which acts as a co-receptor with TbpB. In summary, the authors have used appropriate assays to reach their conclusions about the role of SLAM as a translocase and the contribution of Skp to the localisation of lipoproteins to the surface of bacteria. The findings presented are robust and shed new insights into the sorting of proteins in bacteria, an incompletely understood process which is central to microbial physiology, viurlence and vaccines.

Reviewer #3 (Public Review):

Slam was identified as an outer membrane protein involved in the translocation of certain lipoproteins to the cell surface in Neisseria meningitidis. Slam homologs were also identified in other proteobacteria. However, direct evidence that Slam is an outer membrane translocation device is still missing. In this paper, the authors set up an in vitro translocation assay to probe the role of Slam proteins in the translocation of the lipoprotein TbpB. Although they provide strong data supporting the role of Slam in lipoprotein translocation, further molecular dissection is required to unambiguously establish Slam as a lipoprotein translocator. The work is interesting and the paper clearly written. The authors also discovered a functional link between the periplasmic chaperone Skp and Slam-dependent lipoproteins, which is a novel and interesting finding.

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Manuscript number: RC-2021-01015

Corresponding author(s): Jordan, Raff

1. General Statements [optional]

We thank the reviewers for their thoughtful and constructive comments and have now revised our manuscript accordingly. We apologise that it has taken so long to send in these revisions, but this is in part because both first authors have now left the lab.

2. Point-by-point description of the revisions

Reviewer #1

This reviewer was generally supportive. They note that it is unfortunate that our data suggests the CP110/Cep97 complex does not play a major part in controlling daughter centriole growth—although we believe that this is an important negative result—but feel that other aspects of our data are interesting. They requested no further experiments, but did comment that it would be interesting to determine when g-tubulin is incorporated into growing centrioles. Unfortunately, we cannot test this as the centrioles in these embryos recruit large amounts of g-tubulin to their PCM, so we cannot specifically assay the small amount of protein in the centriolar fraction.

Reviewer #2

Major Points:

__Figure 1: The reviewer notes that Sas-4 and CP110 have antagonistic roles in promoting/repressing centriole growth and asks if Sas-4 is involved in promoting centriole elongation and whether it also oscillates. __It is unclear if Sas-4 directly promotes centriole elongation in flies. We have previously shown that centriolar Sas-4 levels do oscillate during S-phase, but with a timing that is distinct from CP110/Cep97 (Novak et al., Curr. Biol., 2014). These observations do not shed much light on the potential antagonistic relationship between CP110/Cep97 and Sas-4, so we do not comment on this here.

Figure S1B: The reviewer requests that we image the centrioles with greater laser intensity to test whether some residual CP110 or Cep97 protein can be recruited in the absence of the other protein. The quantification of this data suggests that some residual CP110 or Cep97 can still be recruited to centrioles in the absence of the other (Graphs, Figure S1B,C), so we do not think it necessary to repeat this experiment at higher laser intensity to further test this point. We now state that the centriolar recruitment of one protein may not be completely dependent of the other (p6, para.2).

Figure 3: The reviewer questions whether the reduction in CP110/Cep97 levels at the mother centriole that we observe during S-phase could be due to photobleaching. This is an interesting point that we now analyse in more detail (p8, para.2). We do not think the decrease in mother centriolar CP110/Cep97 levels is due to photobleaching as our new analysis (which includes more data points during mitosis) strongly suggests that centriolar levels on the mother rise again at the start of the next cycle (New Figure 3C,D).

The reviewer asks whether the CP110/Cep97 oscillations occur at the tip of the growing centriole, and whether we can use super-resolution imaging to address this. A large body of evidence indicates that CP110/Cep97 are highly concentrated at centriole distal tips, and all our experiments suggest that it is this fraction that is oscillating. In Figure 3, for example, we use Airy-scan super-resolution imaging to follow the oscillation on Mother and Daughter centrioles in living embryos. Although the resolution in these experiments is not as high as we can achieve using 3D-SIM in fixed cells, it seems reasonable to assume that the dots of fluorescence we observe oscillating on these centrioles (Fig. 3) are the same fluorescent dots we observe localised at the distal tips of the mother and daughter using 3D-SIM in fixed cells (Fig. 1A).

The reviewer requests additional quantification of the western blots shown in Figure S1 that we use to judge relative expression levels. As we now describe in more detail in the M&M, these ECL blots are very sensitive, but highly non-linear, so we usually estimate relative expression levels by comparing serial dilutions of the different fractions (see, for example, Figure 1B, Franz et al., JCB, 2013). As we now clarify, the key point is not precisely by how much these proteins are over- or under-expressed, but that we observe a similar oscillatory behaviour when they are either over- or under-expressed.

__The reviewer points out that our statement that the CP110/Cep97 oscillation is entrained by the Cdk/Cyclin oscillator (CCO) is too strong as it is based only on a correlation. __We agree and apologise for this overstatement. To address this, we have now perturbed the CCO by halving the dose of Cyclin B (New Figure 5E—H). This extends S-phase length and we now show that the period of the CP110/Cep97 oscillation is also extended. This suggests that the CCO directly influences the period of the CP110/Cep97 oscillation.

The reviewer notes that our conclusion that the centriole cartwheels are longer or shorter when CP110 or Cep97 are absent or overexpressed, respectively, is based only on Sas-6-GFP fluorescence intensity. They ask if this fluorescence intensity perfectly reflects cartwheel length, and if we can confirm these conclusions using EM. Sas-6 is the main structural component of the cartwheel, so the amount of Sas-6 at the centriole should be proportional to cartwheel length, and we have published two papers that support this conclusion and that use the incorporation of Sas-6 as a proxy to measure cartwheel length (Aydogan et al., JCB, 2018; Aydogan et al., Cell, 2020). Importantly, our previous EM studies support our conclusions about the relationship between cartwheel length and CP110/Cep97 levels: the centrioles in wing-disc cells are slightly longer in the absence of CP110 and slightly shorter when CP110 is overexpressed (Franz et al., JCB, 2013). The new findings reported here provide a potential explanation for this EM data, which was puzzling at the time.

Minor Points:

Figure 1C: The reviewer noted that our schematic illustrations in this Figure could be misleading____. We agree and have now redrawn them.

Reviewer #3

Major points:

The reviewer requested that we clarify our use of the term oscillation, pointing out that oscillations are repetitive variations in levels/activity over time, whereas the “oscillations” we describe here occur during each round of centriole assembly. This is a fair point, and one that is often debated in the oscillation field, with many believing that too many biological processes are termed “oscillations”, when they are not truly driven by the passage of time. To avoid any ambiguity, we now no longer describe the behaviour of CP110/Cep97 as an oscillation (although, for ease of discussion, we still use the term in this letter).

The reviewer thought that the data we show in Figure 1 was not relevant as we largely analyse centrioles in living embryos whereas the data in Figure 1 is derived from fixed wing-disc cells—and similar fixed-cell data has been shown in previous studies. The reviewer suggests we use super-resolution methods to analyse Cp110/Cep97 dynamics in the syncytial embryo, and show this relative to Sas-6 and Plk4. They ask if Plk4 and CP110/Cep97 colocalise at any time. While CP110/Cep97 localisation has been analysed by super-resolution microscopy previously (e.g. Yang et al., Nat. Comm., 2018; LeGuennec et al., Sci. Adv., 2020), CP110/Cep97 was a minor part of these studies and our data is the first to show that this complex sits as a ring on top of the centriole MTs in fly centrioles (that lack the complex distal and sub-distal appendages present in the previously analysed systems). As this localisation is important in thinking about how CP110/Cep97 might influence centriole MT growth, we would like to include it. We cannot show this detail in living embryos as the movement of the centrioles reduces resolution and we cannot observe the ring structure.

Although we do use Airy-scan super-resolution microscopy to study CP110/Cep97 dynamics in living embryos (Figure 3), we cannot do this in two colours (to compare these dynamics to Sas-6 or Plk4 dynamics) as red-fluorescent proteins bleach too quickly. We now show the relative dynamics of CP110/Cep97 and Plk4 recruitment using standard resolution microscopy (New Figure S2). While it is well established that Plk4 and CP110/Cep97 are concentrated at opposite ends of centrioles, they are all recruited to the nascent site of daughter centriole assembly, effectively “colocalising” at this timepoint. This could provide an opportunity for the crosstalk we observe here, and we now mention this possibility (p17, para.1).

The Reviewer questioned whether the loading of Sas-6-GFP onto centrioles can be used as a proxy for cartwheel length, pointing out that Sas-6 could load into centrioles in a way that does not change the cartwheel structure, and that EM is required to test this. As described in our response to Reviewer #2, Sas-6 is the main structural component of the cartwheel, and we have published two papers that use the incorporation of Sas-6 into the cartwheel as a proxy to measure cartwheel length (Aydogan et al., JCB, 2018; Aydogan et al., Cell, 2020). While we cannot exclude that Sas-6 might also associate with the cartwheel in a way that does not involve its incorporation into the cartwheel, it is not clear how EM might address this question. Moreover, even if such a fraction existed, it should not affect our conclusions—as long as Sas-6 is binding to the cartwheel in some way, then the amount bound should remain proportional to the length of the cartwheel. Perhaps the reviewer is suggesting that we perform an EM time course of cartwheel growth to back up our conclusions from the Sas-6 incorporation assay? If so, we think this impractical. The changes in cartwheel length shown in Figure 6 are revealed from analysing several thousand images of centrioles compared at precise relative time points. Such an analysis cannot be done in fixed embryos by EM.

Similar to the point above, the reviewer notes that we use the length of the cartwheel to infer centriole MT length, but we never directly measure MT length. They suggest we perform either an EM analysis or use MT markers to directly measure the kinetics of centriole MT growth. In flies (and many other organisms), the centriole MTs grow to the same length as the centriole cartwheel (Gonzalez, JCS, 1998), so we can be confident that the final length of the cartwheel reflects the final length of the centriole MTs. Moreover, we previously measured the distance between the mother centriole and the GFP-Cep97 cap that sits at the distal tip of the centriole MTs as a proxy for centriole MT length, and found that the inferred kinetics of MT growth were similar to the kinetics of cartwheel growth (inferred from Sas-6 incorporation) (Aydogan et al., 2018). This manual analysis was very time consuming, and we have tried to implement computational analysis methods, but so far without success. For similar reasons to those described in the point above, it is not feasible to accurately measure centriole MT growth kinetics by EM (nobody has been able to do this). Moreover, the centrosomes in these embryos are associated with too much tubulin and the centriole MTs are not yet modified (e.g. by acetylation) as the cycles are so fast—so we cannot directly stain the centriole MTs in fixed embryos. We have now toned down our conclusions about MT length throughout the paper, and we make it clear that we cannot directly measure this.

All of the experiments shown here are performed in the presence of endogenous untagged proteins, and the reviewer wonders if recruitment dynamics might be influenced by competition for binding from the endogenous protein. We have compared the behaviour of many centriole and centrosome proteins in the presence and absence of the untagged WT protein. In all cases, less tagged-protein binds to centrioles/centrosomes in the presence of untagged protein, presumably due to competition. Apart from this, however, we usually observe no real difference in overall dynamics and in Reviewer Figure 1 (see below) we show that CP110-GFP and GFP-Cep97 both oscillate even in the absence of any endogenous protein. As we feel this result is not very surprising, we do not show it in the manuscript.

The reviewer correctly noted that our data was not strong enough to conclude that the CP110/Cep97 oscillation is influenced by the CCO. This was also raised by Reviewer #2 and, as described above (p2, para.3 above), we have now performed additional experiments to more directly demonstrate this point (new Figure 5G—H).

The reviewer requests more discussion of why our conclusion that CP110/Cep97 levels oscillate on the growing daughter centrioles during S-phase is different to that reached by Dobbelaere et al, (Curr. Biol., 2020), who conclude that Cep97-GFP only starts to incorporate into the new daughter centrioles late in S-phase when the daughters are fully grown. We have discussed this discrepancy with these authors and they kindly shared their reagents with us (so our endogenous Cep97-GFP oscillation data comes from the same line they used in their experiments), but we have not come to a clear conclusion on this point. We have shown robust oscillations for CP110 and Cep97 by quantifying many hundreds of centrioles using multiple transgenes (both over- and under-expressed) in multiple backgrounds. Cep97 dynamics were a very minor part of the Dobbelaere et al., study, and they analysed a much smaller number of centrioles. We now briefly mention this discrepancy (p9, para.1), but do not discuss it in detail as we have no definitive explanation for it.

The reviewer requests more experiments or more discussion to address the mechanism(s) of crosstalk between CP110/Cep97 and Plk4, and they suggest several avenues for further investigations. These are excellent ideas, and we are working hard on these approaches. These are all long-term experiments, however, and we feel it is important that the field be made aware of these surprising findings as soon as possible, as others may be better-placed to provide mechanistic insight into how this system ultimately works. We now briefly mention some of the future directions the reviewer highlights in the Discussion.

The reviewer thought we should highlight the previous publications showing that Plk4-induced centriole amplification requires CP110 and that Plk4 can phosphorylate CP110. These studies (Kleylein-Sohn et al, Dev. Cell, 2007; Lee et al., Cell Cycle, 2017) were mentioned, but we now discuss them more prominently (p17, para.2).

Minor Points:

The reviewer raised a number of minor concerns that we have now addressed: (1) We discuss the model the reviewer suggests; (2) we no longer state that the crosstalk between CP110/Cep97 and Plk4 is unexpected; (3) We have clarified our description of the shift in timing of the peak levels of CP110/Cep97, which we no longer refer to as an oscillation; (4) We define mNG as monomeric Neon Green; (5) We have changed our schematics in Figure 1 as suggested by the reviewer; (6) We have corrected the mistake in the legend to Figure 8.

Reviewer #4

Major points:

1. The reviewer noted that the amplitude of the CP110/Cep97 oscillations depended on protein expression levels, so the oscillations might not reflect the behaviour of the endogenous proteins. They requested that we either repeat our experiments with CRISPR knock-in alleles, or conduct experiments with the lines driven by the endogenous promotors but in their respective mutant backgrounds. We have not generated CRISPR knock-ins for CP110/Cep97, but have done so for many other centriole/centrosome proteins (>8) and found that most such lines are expressed at higher or lower levels than the endogenous allele (and sometimes very significantly so). This is also true for our standard transgenic lines, where genes are expressed from their endogenous promoters, but are randomly integrated into the genome. The blots in Figure 4 show that CP110-GFP and GFP-Cep97 expressed from a ubiquitin (u) promoter or from their endogenous promoters (e) are expressed at ~2-5X higher or ~2-5X lower levels than the endogenous proteins, respectively. As we observe CP110/Cep97 oscillations in all cases, it seems unnecessary to generate new CRISPR knock-ins (that are also likely to be somewhat over- or under-expressed) to show this again. As the reviewer asks, we show that Cep97-GFP and CP110-GFP still oscillate in in the absence of the endogenous proteins (Reviewer Figure 1). As this does not seem a surprising result, we do not show this in the main manuscript. In the same point the reviewer requests that we use antibody staining in fixed embryos to show that the untagged proteins also oscillate. Analysing protein dynamics is much harder in fixed embryos, as the levels of fluorescent staining are more variable and we can only approximately infer relative timing, rather than precisely measuring it (as we can in living embryos). Moreover, as both proteins in the CP110/Cep97 complex exhibit a very similar oscillatory behaviour when tagged with either GFP or RFP (e.g. Figure 2C), and this behaviour is distinct to that observed with several other GFP- or RFP-tagged centriole proteins (e.g. Novak et al., Curr. Biol., 2014; Conduit et al., eLife, 2015; Aydogan et al., JCB, 2018; Aydogan et al., Cell, 2020) it seems very unlikely that this behaviour is induced by the GFP (or RFP) tag.

The reviewer also suggests that we show the data with the endogenous promoter before we show the data with the ubiquitin promoter. As we now explain better (and show in Figure 4), this seems unnecessary as the proteins expressed from the ubiquitin promotor are probably actually expressed at levels that are more similar to the endogenous protein.

The reviewer questions whether the oscillations we observe might be due to the centrioles simply moving up and down in the embryo during the cell cycle, and they suggest we monitor Asl behaviour to rule this out. We have previously shown that Asl-GFP levels do not oscillate; they remain constant throughout the cell cycle on old-mother centrioles, and grow approximately linearly throughout S-phase on new-mother centrioles (see Figure 1D in Novak et al., Curr. Biol., 2014).

We were not sure we understood this point properly, so we copy the reviewers comment in full here: ____The authors mention (for instance on p. 3) that the inner cartwheel and the surrounding microtubules assemble at opposite ends of the daughter centriole. However, my understanding is that the short centrioles present in the fly embryo have an inner cartwheel that extends throughout the organelle, such that it might be moot to make a distinction between the two ends in this case. Moreover, it is also my understanding that this inner cartwheel is itself surrounded by microtubules, so that microtubule assembly might not be expected to occur strictly at the distal end no matter what. The reviewer is correct that Drosophila centrioles are short (~150nm) and that the cartwheel extends throughout the centriole. We think the reviewer is suggesting that it may not be relevant therefore whether the cartwheel and centriole MTs grow from opposite ends—as the activities that govern their growth may not be spatially separated? However, because cartwheels grow preferentially from the proximal-end (Aydogan et al., JCB 2018) while centriole MTs are assumed to grow preferentially from the distal (plus) end, there is an intrinsic problem in ensuring they grow to the same size—no matter how short or long the centrioles are. The reviewer is correct that one possible solution to this problem is that the centriole MTs actually grow from their minus ends, but this is not widely accepted (or even proposed). We have tried to explain this issue more clearly throughout the revised manuscript.

The reviewer points out that the schematic illustrations in Figure 1A and 1C are inaccurate and unhelpful. We agree and have now redrawn these.

The reviewer asks that we provide information about the eccentricities of the centrioles in the different datasets used to calculate the protein distributions shown in Figure 1, particularly as the data for Sas-4-GFP and Sas-6-GFP were obtained previously using a different microscope modality, making comparisons complicated. The point that comparing distance measurements across different datasets is difficult is an important one, and we now state that such comparisons should be treated with caution. However, we have not provided information on the distribution of centriole eccentricities in the different experiments as it wasn’t clear to us how this information could be used to make such comparisons more accurate (presumably the reviewer is suggesting we could apply a correction factor to each dataset?). The very tight overlap in the positioning of CP110/Cep97 fusions (Figure 1C) strongly suggests that any difference in the average centriole eccentricities of the different populations of centrioles analysed, which are already tightly selected for their en-face orientation (i.e. eccentricity

The reviewer requested that we show the “noisy data” we obtained during mitosis that we excluded from our analysis in Figure 3. As we now explain in more detail (p8, para.2), there are two reasons why the data for mitosis in this experiment is “noisy”: (1) The protein levels on the centrioles are low in mitosis and the centrioles are more mobile, so they are hard to track; (2) The Asl-mCherry marker used to identify the mother centriole starts to incorporate into the daughter (now new mother) centriole during mitosis, making it difficult to unambiguously distinguish mothers and daughters. As a result, we cannot track and assign mother/daughter identity to very many centrioles during mitosis—although we now include some extra data-points during mitosis for the centrioles where we could do this (revised Figure 3C,D). Importantly, it is clear that this “noisy” data hides no surprises: one can see (Figure 3C,D) that the signal on the centrioles is simply low during mitosis and then starts to rise again as the embryos enter the next cycle. This is confirmed in the normal resolution data (Figure 2B,C; Movies S1 and S2) where we can track many more centrioles due to the wider field of view and because we do not have to discard centrioles in mitosis that we cannot unambiguously assign as mothers or daughters.

The reviewer requests that we conduct a super-resolution Airy-scan analysis of CP110/Cep97 driven from their endogenous promoters (eCP110 or eCep97) to ensure that the oscillations we see with these lines (shown in Figure 4C,D) are also occurring at the daughter centriole—as we already show for the oscillations observed with the uCP110 and uCep97 lines (shown in Figure 4C,D, and analysed at super-resolution on the Airy-scan in Figure 3). This is technically very challenging as super-resolution techniques require a lot of light and the centriole signal in the eCP110/Cep97 embryos is very dim compared to uCP110/Cep97 embryos (Figure 4C,D). We have managed to do this for eCep97-GFP and confirmed that—even in these embryos that express Cep97-GFP at much lower levels than the endogenous protein (Figure 4A)—the “oscillation” is primarily on the daughter (Reviewer Figure 2). As this data is very noisy, and as the ubiquitin uCP110/Cep97 lines express these fusions at levels that are closer to endogenous levels (Figure 4A,B), we do not show this data in the main text.

The reviewer also asks for clarification as to why we use the Airy-scan for some experiments and 3D-SIM for others. As we now explain (p8, para.1), 3D-SIM has better resolution than the Airy-scan, but it takes more time and requires more light—so we cannot use it to follow these proteins in living embryos. Thus, for tracking CP110/Cep97 throughout S-phase in living embryos we had to use the Airy-scan.

The reviewer questions why in some experiments we analyse the behaviour of 100s of centrioles, whereas in others the numbers are much smaller (1-14 in Figure 3—note, the reviewer quoted this number as coming from Figure 4, but it actually comes from Figure 3, so we have assumed they mean Figure 3). We apologise for not explaining this properly. The super-resolution experiments in Figure 3 are performed on a Zeiss Airy-scan system, which has a much smaller field of view than the conventional systems we use in other experiments. Thus, we inherently analyse a much smaller number of centrioles in these experiments. In addition, as explained in point 6 above, in these experiments we need to analyse mother and daughter centrioles independently, and in many cases we cannot unambiguously make this assignment, so these centrioles have to be excluded from our analysis.

The reviewer questions why we selected the 10 brightest centrioles for the analysis shown in Figure S1B,C (note, the reviewer states Figure S2 here, but it is the data shown in Figure S1B,C that is selected from the 10 brightest centrioles, so we assume this is the relevant Figure). We apologise for not explaining this properly. In these mutant embryos very little CP110-GFP localises to centrioles in the absence of Cep97, and vice versa, so we cannot track centrioles using our usual pipeline and instead have to select centrioles using the Asl-mCherry signal. As the difference between the WT and mutant embryos is so striking, we simply selected the brightest 10 centrioles (based on Asl-mCherry levels) in both the WT and mutant embryos for quantification. We could select more centrioles, or select centrioles based on different criteria, but our main conclusion—that the centriolar localisation of one protein is largely dependent on the other—would not change.

The reviewer also questioned why we performed the analysis shown in Figure S2 (new Figure S3) during S-phase of nuclear cycle 14, when the rest of the manuscript focuses on nuclear cycles 11-13. We apologise for not explaining this properly. In cycles 11-13 centriolar CP110/Cep97 levels rise and fall during S-phase, whereas both proteins reach a sustained plateau during the extended S-phase (~1hr) of nuclear cycle 14—making it easier to analyse CP110/Cep97 levels in embryos when their centriole levels are maximal. We now explain this.

The reviewer requests that we quantify the western blots shown in Figure 4 in the same way we do in figure 8. To do this we would need to perform multiple repeats of these blots and we did not perform these because the blots shown in Figure 4 largely recapitulate already published data (Franz et al., JCB, 2013; Dobbelaere et al., Curr. Biol., 2020). Moreover, as described in our response to Reviewer #2, these ECL blots are very sensitive, but highly non-linear, so we always compare multiple serial dilutions of the different extracts to try to estimate relative levels of protein expression. We now explain this in the M&M.

The reviewer suggests the data shown in Figure 8 is a “straw man”: we really want to test whether modulating CP110/Cep97 levels modulates centriolar Plk4 levels, but instead we test how they modulate cytoplasmic Plk4 levels. The language here is harsh, as it suggests that our intention was to mislead readers into thinking that we have addressed a relevant question by addressing a different, irrelevant, one. We apologise if we have missed something, but we believe we do perform exactly the experiment that the reviewer thinks we should be doing—quantifying how centriolar Plk4 levels change when we modulate the levels of CP110 or Cep97 (Figure 7). It is clear from this data that modulating the levels of CP110/Cep97 does indeed modulate the centriolar levels of Plk4. In Figure 8 we seek to address whether this change in centriolar Plk4 levels occurs because global Plk4 levels in the embryo are affected—a very reasonable hypothesis, which this experiment addresses quite convincingly (although negatively).

Minor Points:

The reviewer highlights a small number of mistakes and omissions, all of which have been corrected.

Finally, we would like to thank the reviewers again for their detailed comments and suggestions. We hope that you and they will agree that the changes we have made in response to these comments have substantially improved that manuscript and that it is suitable for publication in The Journal of Cell Science.

Sincerely,

Jordan Raff

__Reviewer Figure 1. CP110/Cep97 dynamics remain cyclical even when Cep97-GFP and CP110-GFP are expressed from their endogenous promotors in the absence of any endogenous protein. __Graphs show how the levels (Mean±SEM) of centriolar CP110/Cep97-GFP change during nuclear cycle 12 in (A) Cep97-/- embryos expressing eCep97-GFP or (B) CP110-/- embryos expressing eCP110-GFP. CS=Centrosome Separation, NEB=Nuclear Envelope Breakdown. N≥11 embryos per group, average of n≥15 centrioles per embryo.

__Reviewer Figure 2. ____The cyclical recruitment of Cep97-GFP expressed from its endogenous promoter occurs largely at the growing daughter centriole. __The graph quantifies the fluorescence intensity (Mean±SD) acquired using Airy-scan microscopy of eCep97-GFP on mother (dark green) and daughter (light green) centrioles in individual embryos over Cycle 12. CS = Centrosome Separation, NEB = Nuclear Envelope Breakdown. Data was averaged from 3 embryos as the number of centriole pairs that could be measured was relatively low (total of 2-8 daughter and mother centrioles per time point; in part due to the much dimmer signal of eCep97-GFP in comparison to uGFP-Cep97).

This work has been peer reviewed in GigaScience (see paper https://doi.org/10.1093/gigascience/giac011), which carries out open, named peer-review.

These reviews are published under a CC-BY 4.0 license and were as follows:

Reviewer 2: Gregg Thomas

This paper presents 17 new insect genomes from the order of caddisflies (Trichoptera). The authors combine these genomes with 9 previously sequenced genomes to analyze genome size evolution across the order. They find that genome size tends to correlate with evolution of repeat elements, specifically expansion of transposable elements (TEs). Interestingly, the authors also notice that TE expansions also correlate with gene copy-number (or gene fragment copy-number), even of highly conserved genes used to assess genome completeness. Overall, I find this paper very well written and easy to follow. The genomic resources and analyses presented provide novel new resources and findings for insects in the order Trichoptera, with potential implications beyond. I have only minor suggestions before publication, outlined below.

1. Regarding the TE and BUSCO gene fragment associations, while I think this is a really interesting analysis, I found the underlying models a bit difficult to understand. Line 236 reads, "To test whether repetitive fragments were due to TE insertions near or in the BUSCO genes or, conversely, due to the proliferation of 'true' BUSCO protein-coding gene fragments…" Is the idea that a BUSCO gene has been duplicated itself and then one copy is either fragmented by a TE insertion or hitch-hikes with a TE (as mentioned on line 501)? Or are these fragments only of BUSCO genes that didn't match a full BUSCO gene at all, but the fragments that did match had unexpectedly high coverage? I guess I'm just confused as to whether a gene duplication needs to precede the TE insertions/hitch-hiking, which is subsequently pseudogenized either prior to or because of the TE activity, or if these are gene losses. I understand how the TE could inflate the coverage of these fragments, but I guess I'm still not clear on how these fragments arise in the first place. Any clarification would be helpful! Also, if the case is that these are fragments of BUSCO genes that have no full matches in the genome, how might assembly contiguity or quality be affecting these matches?

2. One thing that I noticed throughout the figures is that branch B1, leading to A. sexmaculata, the branch leading to clade A, and the branch leading to clade B (as labeled in Figures 1 and 2) appear to form a polytomy. I don't find this mentioned in the text and am wondering why this relationship remains unresolved with these data. I don't think this has any bearing on the results, since all analyses are done on the tips of the tree, but I think readers looking at these trees will want to know what is going on at that node.

3. The authors use custom scripts for their BUSCO-TE correlation analysis and provide a link to a Box folder on line 514. I would request that these scripts be put somewhere more stable and accessible (e.g., github). Not only was I asked to login when clicking the link, but after I had done so that link didn't seem to exist.

Minor/editorial points

1. Would the authors be able to report concordance factors for the species tree? I think this should be easy enough with IQ-tree and is something I ask everyone to do. This may also help answer my question about the polytomy.

2. The authors do a good job of mentioning and citing programs used throughout the manuscript but seem to skip this in the Assembly section (starting on Line 398). "First, we applied a long-read assembly method…" Which one? Same for "de novo hybrid assembly approaches." I see that assembly is covered in detail in the Supplement, but I think naming the main programs used (wbtdbg2 and Masurca) should be in the main text.

3. Line 281-282: I think some of the brackets and parentheses here are mismatched or un-closed.

This work has been peer reviewed in GigaScience (see paper https://doi.org/10.1093/gigascience/giac005), which carries out open, named peer-review.

These reviews are published under a CC-BY 4.0 license and were as follows:

Reviewer 1: Paul Stewart

Fahrner et al have produced a very nice manuscript and corresponding pipeline. They describe a collection of DIA tools in the Galaxy framework for reproducible and version-controlled data processing. These DIA tools are an excellent addition to the growing number of proteomics-centric tools already available in Galaxy. The reviewer could find no major revisions needed and therefore only requests a few minor revisions before this is ready for publication:

Please include page numbers in the revised manuscript to make referencing the text easier.

Page 6

OpenSwath and PyProphet are cited and are also used in the manuscript. Please cite one or two alternatives.

Please consider citing a tool the each time it is used in a new paragraph (e.g. MSstats).

There is heavy reliance on conjunctive adverbs (However, ...; Thus, ...) on this page and throughout the manuscript. These can make passages a bit hard to read. Please consider rephrasing.

Page 7

Why "so-called histories"? Aren't they simply "Histories"?

Page 14

'To decrease the analysis time of the semi-supervised learning, the merged OSW results can be first subsampled using the PyProphet subsample tool and subsequently scored using the PyProphet score tool. '

The reviewer is not familiar with this approach. Can you please give additional justification (maybe under methods?) or provide a citation that this is a reasonable approach?

Page 15

Please check your reference software and/or work with the journal to ensure that the web addresses are linked properly. For example, the reviewer tried copying the link "https://training.galaxyproject.org/training- %20material/topics/proteomics/tutorials/DIA_lib_OSW/tutorial.html" but a "%20" (or a space) is inserted into the URL after "training-" so the link as it appears did not work until this was removed. A less technically savy reader may think the links are broken and will not be able to access the materials.

Page 16

'We identified and quantified between 25.000 to 27.000 peptides ...'

Please be consistent with number formatting (25000 vs 25.000). Other values in the tables did not use this formatting. Please check with journal editor for convention.

Figures

Please be consistent with axes labels. Some are upper case and some are lower case.

Figure 2B

Please round R2 to 2 or 3 decimals.

Figure 3

Please change the red-green color scheme to a more color-blind friendly color scheme (e.g. red blue)

This work has been peer reviewed in GigaScience (see paper https://doi.org/10.1093/gigascience/giac001), which carries out open, named peer-review.

These reviews are published under a CC-BY 4.0 license and were as follows:

Reviewer 1: Bo Li

Single-cell RNA-seq has revolutionized our abilities of investigating cell heterogeneity in complex tissue. Generating a high-quality gene count matrix is a critical first step for single-cell RNA-seq data analysis. Thus, a detailed comparison and benchmarking of available gene-count matrix generation tools, such as the work described in this manuscript, is a pressing need and has the potential to benefit the general community.

Although this work has a great potential, the benchmarking efforts described in the manuscript are not comprehensive enough to justify its publication at GigaScience unless the authors address my following major and minor concerns.

Major concerns:

1) The authors should discuss related benchmarking efforts and the differences between previous work and this manuscript in the Background section instead of the Discussion section. For example, Du et al. 2020 G3: Genes, Genomics, Genetics. and Booeshaghi & Pacther bioRxiv 2021 should be mentioned and discussed in the Background section. In addition, STARsolo manuscript (https://www.biorxiv.org/content/10.1101/2021.05.05.442755v1), which contains a comprehensive comparison of CellRanger, STARsolo, Alevin and Kallisto-Bustools should be cited and discussed. Zakeri et al. 2021 bioRxiv (https://www.biorxiv.org/content/10.1101/2021.02.10.430656v1) should also be included and discussed in the Background section.

2) Benchmark with latest versions of the software. The choice of Cell Ranger, STARsolo, Alevin and Kallisto-BUStools is good because they are four major gene count matrix generation tools. However, I urge the authors also include CellRanger v6 and Alevin-fry (Alevin_sketch/Alevin_partialdecoy/Alevin_full-decoy, see STARsolo manuscript), which are currently lacking, into their benchmarking efforts. The authors may also consider add STARsolo_sparseSA into the benchmark. Since single-cell RNA-seq tool development is a fast-evolving field, benchmarking of the up-to-date versions of tools is super critical for a benchmarking paper.

3) Conclusions. The authors summarized the observed differences between tools based on the benchmarking results. This is good but very helpful for end-users. I recommend the authors to emphasize their recommendations for end-users more clearly in the discussion/results section. For example, do the authors recommend one tool over the others under certain circumstances? If so, which tool and which circumstance and why? I like Figure 5 a lot and hope the authors can summarize this figure better in the manuscript.

4) This manuscript concluded that differential expression (DEG) results showed no major differences among the alignment tools (Figure 4). However, the STARsolo manuscript suggested DEG results are strongly influenced by quantification tools (Sec. 2.6, Figure 5). Please explain this discrepancy.

5) This manuscript suggested simulated data is not as helpful as real data. However, the STARsolo manuscript reported drastic differences between tools using simulated data. Please comment on this discrepancy.

6) I have big concerns regarding the filtered vs. unfiltered annotation comparison. In particular for pseudogenes, we know that many of them are merely transcribed or lowly transcribed. As a result, many of these pseudogenes would not be captured by the single-cell RNA-seq protocol. At the same time, because these pseudogenes share sequence similarities with functional genes, they would bring trouble for read mapping. This is one of the main reasons for using a carefully filtered annotation. Actually, whether and how to filter annotation is in active debate in big cell atlas consortia such as Human Cell Atlas. Thus, I would be super careful about describing results comparing filtered vs. unfiltered annotation. For example, in Suppl. Figure 8D, there are 6 mitochondrial genes that have 100% sequence similarity to their corresponding pseudogenes. It is impossible to distinguish if a read comes from a gene or a pseudogene for these 6 genes and it is also not necessary --- the transcribed RNA should also be exactly the same. Thus, I encourage the authors remove their pseudogenes from the annotation and I suspect the mouse data results should look similar to the human data in the Suppl. Figure 8A.

7) The endothelial dataset was only run on CellRanger 3 because the UMI sequence is one base shorter. Could the authors augment the UMI sequence with one constant base and run this dataset through CellRanger 4/5/6?

8) I think it is more appropriate to call the tools benchmarked as "gene count matrix generation tools" instead of "alignment tools".

Minor concerns:

1) The Suppl Table 2 mentioned in the main text corresponds to Suppl. Table 3 in the attachment. In addition, there is no reference to Suppl Table 2.

2) Suppl Table 3 PBMC, why do I see endothelial cell markers in PBMC dataset?

3) Suppl Figure 7 is never referenced in the main text.

4) Suppl Figure 8D is never referenced in the main text.

This work has been peer reviewed in GigaScience (see paper https://doi.org/10.1093/gigascience/giab099), which carries out open, named peer-review.

These reviews are published under a CC-BY 4.0 license and were as follows:

Reviewer 3: idoia ochoa

The authors present a novel tool for the compression of collections of bacterial genomes. The authors present sound results that demonstrate the performance gain of their tool, MBGC, with respect to the state-of-the-art. As such, I do not have concerns about the method itself. My main concerns are with respect the description of the tool, and how the results are presented. Next I list some of my suggestions (in no particular order):

Main Paper: - Analysis section: Before naming MBGC specify that it is the proposed tool. - Analysis section: Reference for HRCM. Mention here also that other tools such as iDoComp, GDC2, etc. are discussed in the Supplementary (this way the reader knows more tools were analyzed or at least tried on the data).

• Analysis section: The paragraph "Our experiments with MBGC show that... " is a little misleading, since it seems that the tool has the capacity to compress a collection and just extract a single genome from it. This becomes clear later in the text when it is discussed how the tool could be used to speed up the download of a collection of genomes from a repository. So maybe explain that in more detail here, or mention that it could be used to compress a bunch of genomes prior to download. And then point to the part of the text where this is discussed in more detail.

• Analysis section: The results talk about the "stronger MGBC mode", the "MGBC max", but in the tables it reads "MBGC default" or "MBGC -c 3". I assume "MBGC -c 3" refers to "MBGC max", but it is not stated anywhere. maybe better to call it "MBGC default" and "MBGC max".

• Analysis section: Although the method is explained later in the text, it would be a good idea to give a sense of the difference between the default and max modes of the tool. Or some hints on the trade-off between the two. Also, the parameter "-c 3" is never explained.

• Analysis section: Figures, it is difficult to see the trade-off between relative size and relative time, can you use colored lines? such that the same color refers to the same set of genomes. Also, in the caption, explain if we want small or high relative size and time. it may be clear, but better to clearly state it.

• Analysis section: there is a sentence that says "all figures w.r.t. the default mode of MBCG". It would be good also to state that in the caption, so that the reader knows which mode of the tool is being used to generate the presented results. and if the input files are gzipped or not. For example, for the following paragraph that starts with Fig. 1, it is not clear if the files are gzipped or not.

• Analysis section: First time GDC2 is mentioned, the first thing that comes to mind is why it was not used for the bacterial experiments. See my previous point on having a couple of sentences about the other tools that were considered, and why they are not included in the main tables/figures.

• Methods:

-- Here I am really missing a diagram explaining the main steps of the tool. It seems the paper has been rewritten slightly to fit the format of the journal and some things are not in the correct order. For example, it says the key ideas are already sketched, but i do not think that is true.

-- (offset, length) i assume refers to the position of the REF where the match begins, and the length of the match, but again, not really explained. A diagram would help. Also, when it is time to compress the pairs, are the offset delta encoded? or encoded as they are with a general compressor?

-- How are the produced tokens (offset, length, literals, etc.) finally encoded?

-- First time parameter "k" is mention, default value? Also, how can you do a left extension and "swallow" the previous match? is it because the previous match could have been at another position? otherwise if it was in that position it would have been already extended to the right, correct? i mean, it would have generated a longer match.

-- The "skip margin" idea is not well explained. not sure why the next position after a match is decreased by m. please explain better or use a diagram with an example.

-- when you mention 1/192, maybe already state that this is controlled by the parameter u. otherwise when you mention the different parameters is difficult to relate them to the explanation of the algorithm.

Availability of supp...

-- from from (typo) Tables

-- Specify the number of genomes in each collection.

-- change MBGC -c 3 to MBGC max or something similar. (see my previous comment -c flag is not explained!)

Supplementary Material

-- move table 1 after the text for ease of reading

-- not clcear if the tool has random access or not. it is discussed the percentage of time (w.r.t. decompreessing the whole collection i believe) that it would take to decompress one of the first gneomes vs one of the last ones. this should be better explained. for example, if we decompress the last genome of the collection we will employ 100% of the time, right? given that previous genomes are part of REF (potentially). please explain better and discuss this point in the analysis part, not only in the supplementary. seems like an important aspect of the algorithm.

-- I assume this is not possible, but should be discussed as well. can you add a genome to an already compressed collection? this together with the random access capabilities will highlight better the main possible uses of the tool.

-- section 4.3: here HT is used, and then HT is introduced in the next paragraph. please revise the whole text and make sure everything is in the right order.

-- parameter m, please explain better.

-- add colors to figures, it will be easier to read them. Overall, as I mentioned before, I believe the tool offers significant improvements with respect to the competitors for bacterial genomes, and performs well on non bacterial genomes as well. What should be improved for publication is the description of the method, since at the end of the day is the main contribution, and how the text is presented.

Many schools and daycares are sadly closed at the moment because of COVID19 pandemic.

Rashida, as an SLA student, talked about the issue she met before, and explained why this process can help. Her letter is a great evidence to show that our action is effective.

Author Response:

Reviewer #1:

The authors of this study carried out two carefully designed field and a glasshouse experiment simulating effects of rapid warming on soil carbon loss. They did this by transplanting alpine turfs from their cold environment to lowland warm environment. They found that when lowland plants were inserted into alpine turfs under these lowland climatic conditions (referred to as warming treatment combined with warm-adapted plant introduction) they rapidly increased soil microbial decomposition of carbon stocks due to root exudates feeding the microbes.

The question is how well this experimental setup mimics what would happen if lowland plants would be inserted into alpine turfs in situ (which have already experienced considerable warming over the past decades), perhaps with an additional warming treatment there.

The Reviewer alludes to two pertinent points here. The Reviewer’s first point considers whether lowland plants would function similarly (and, by extension, have the same effect on the soil system) if moved from the warmer lowland site to the cooler alpine site. This is a fascinating question in its own right, in that it raises questions about how migrations of non-adapted genotypes far beyond range edges (e.g. via human activity) impact recipient ecosystems. However, although we agree that alpine ecosystems have warmed considerably in recent decades, we cannot be confident that the high elevation sites in our study are already within the climate niche of the lowland focal species. As such, to address our research questions in situ at the high sites would have required additional warming treatments, which come with their own set of disadvantages (see our second point to this comment, below). We also refer the Reviewer to specific questions about adaptation below (see R6), although we see that we were not careful enough about the rationale for our design in the previous version of the manuscript. We have therefore added a clarifying sentence to the Main Text as follows:

L101: “In short, the experiments used here examined how the arrival of warm-adapted lowland plants influences alpine ecosystems in a warmed climate matching lowland site conditions (i.e. turf transplantation to low elevation plus lowland plant addition) relative to warming-only (i.e. turf transplantation to low elevation) or control (i.e. turf transplantation within high elevation) scenarios.”

Second, the Reviewer implicitly raises a point about whether our chosen approach of simulating warming plus lowland plant arrival (i.e. transplantation plus addition of lowland plants) is the most appropriate, specifically by suggesting an alternative option of adding lowland plants to (possibly experimentally-warmed) alpine turfs at the high elevation origin site. Here, it was essential to create a climate scenario in which lowland plants would survive and operate within their climatic niche (i.e. relative to their home conditions) once planted into alpine turfs, rather than perform sub-optimally (e.g. be in a potentially inferior competitive position) or be unable to persist at all. The most parsimonious and reliable way to ensure this was to transplant alpine turfs to a site with a lowland temperature regime, with transplantations also being shown to outperform other methods when novel species interactions are involved (Yang et al. 2018). Most importantly, it was crucial to select a method that warmed the entire plant-soil system rather than only the air (e.g. open-top chambers, IR lamps; Marion et al. 1997; Aronson et al. 2009) or soil (e.g. heating cables; Hanson et al. 2017), and did so realistically throughout the year regardless of the weather (e.g. open-top chambers only work on sunny days in the summer; Marion et al. 1997) or a power supply (e.g. IR lamps, heating cables). Transplantation remains the only way to achieve this (Hannah 2022; Shaver et al. 2000). We now clarify our logic in the manuscript as follows:

L91: “Elevation-based transplant experiments are powerful tools for assessing climate warming effects on ecosystems because they expose plots to a real-world future temperature regime with natural diurnal and seasonal cycles while also warming both aboveground and belowground subsystems. This is especially true if they include rigorous disturbance controls (here, see Methods) and are performed in multiple locations where the common change from high to low elevation is temperature (here, warming of 2.8 ºC in the central Alps and 5.3 ºC in the western Alps). While factors other than temperature can co-vary with elevation, such factors either do not vary consistently with elevation among experiments (e.g. precipitation, wind), are not expected to strongly influence plant performance (e.g. UV radiation) or in any case form part of a realistic climate warming scenario (e.g. growing-season length, snow cover).”

A further question is if alpine plants inserted in turfs at alpine climatic conditions would have a similar effect as lowland plants inserted in turfs at lowland climatic conditions.

We interpret “turfs” to mean “lowland turfs” here, since we did insert lowland plants into alpine turfs under lowland climatic conditions (i.e. the WL treatment). We found that adding alpine plants to alpine turfs in alpine climatic conditions (i.e. planting disturbance control, see Methods) had no effect on alpine soil carbon content. By extension, we would expect that adding lowland plants to lowland turfs in lowland climatic conditions would have no effect on lowland soil carbon content. While not explicitly tested, including this treatment would not change our finding that adding lowland plants to alpine turfs causes a reduction in soil carbon content relative to adding alpine plants to alpine turfs. Given this, we have left the text as is, but are happy to revisit this issue based on further discussion with the Reviewer/Editor.

I suggest that the authors consider these questions when they draw conclusions about the results from their experiments. It would also be interesting to discuss the relevance of sudden strong warming effects relative to slower warming, potentially allowing ecosystems to adjust via changes in genetic composition of species (i.e. evolution) or species composition of communities (i.e. community assembly).

Thank you for this excellent suggestion. We absolutely agree that anything short of a decadal experiment is unable to detect the role of longer-term evolutionary or community processes on soil carbon dynamics. While this doesn’t eliminate the need for experiments that consider shorter timescales, it is important to explicitly state this limitation. As suggested, we have added a sentence discussing this possibility in the concluding paragraph:

L387: “While our findings demonstrate that lowland plants affect the rate of soil carbon release in the short term, short-term experiments, such as ours, cannot resolve whether lowland plants will also affect the total amount of soil carbon lost in the long term. This includes whether processes such as genetic adaptation (in both alpine and lowland plants) or community change will moderate soil carbon responses to gradual or sustained warming.”

We also agree that it is extremely challenging to undertake warming experiments that do not initially “shock” the system through a sudden change in temperature. Having said this, alpine ecosystems are adapted to rapid within- and between-season temperature changes, making such shocks less relevant here.

Reviewer #2:

The authors were trying to test whether the migration of lowland plants into alpine ecosystems affects the warming impact on soil carbon. To achieve this goal, the authors first did two field experiments (moving intact turf from high-elevation to low-elevation to simulate warming) in the Alps, and then did a greenhouse pot study to explore the potential mechanisms for the results observed in the field experiments.

The main strenghs of this work are the combination of a field experiment (conducted at two sites) and a greenhouse pot experiment (to explore the detailed mechanisms). Moreover, a number of techniques were used to measure plant traits, soil DOM and microbial properties (e.g. CUE, growth) which help to find the potential mechanisms.

We thank the Reviewer for this positive comment.

The main weaknesses of this work are below:

1) The two field experiments are very short-term (<1 year), but the results were that warming and/or warming+lowland plants led to very high amount of soil C loss (up to ~40%, Fig. 1). I was shocked to see these results as many field warming studies have shown undetectable change in SOC even after years or decades. The authors did not provide a good explanation for this rapid and large change in SOC.

We apologise for the confusion. We’re unsure where “up to ~40%” comes from here, so we have taken the Reviewer’s later suggestion of changing the annotation on Fig. 1 to contrast C versus WL treatments (Western Alps = 25.6 ± 7.2 mg g-1; Central Alps = 25.3 ± 8.6 mg g-1) rather than W versus WL treatments.

With regards to the magnitude of soil carbon loss observed, we express soil carbon content in mg g-1 (i.e. mass-based per-mil), not cg g-1 (i.e. mass-based percent). This is so that we could use percent changes in the text to highlight the numeric magnitude of differences between treatments without confusing them with mass-based percent soil carbon – although we appreciate that this also caused confusion. To clarify, converting the above C versus WL treatment contrasts from mg g-1 to mass-based percent yields 2.56% ± 0.72% for the Western Alps experiment and 2.53% ± 0.86% for the Central Alps experiment. While it is striking that the WL treatments lost ~2.5% (~25 mg g-1) soil carbon in one year, such a loss is not extraordinary. To avoid future confusion, we have clarified the units in the Fig. 1 caption as follows:

L77: “Mean ± SE soil carbon content (mg C g-1 dry mass; i.e. mass-based per-mil) in alpine turfs transplanted to low elevation (warming, W; light grey), transplanted plus planted with lowland plants (warming plus lowland plant arrival, WL; dark grey) or replanted at high elevation (control, C; white). Data are displayed for two experiments in the western (left) and central (right) Alps, with letters indicating treatment differences (LMEs; N = 58).”

2) The greenhouse experiment was used to explore the potential reasons for the amplified loss of soil C in the field experiment. However, a key result was based on incubation of disturbed soils (8 g) and a two-pool modeling of the respiration data from the short-term incubation. This may not provide a good estimate of the true turnover rate of SOC under different plant species (even in the greenhouse condition). If rhizosphere priming was the proposed mechanism (as hinted by the authors), a better approach (such as 13C labeling) is needed to measure microbial respiration from intact soils (with plant/root presence).

We agree with the Reviewer that using an approach such as 13C-labelling would have provided more direct evidence that lowland plants cause a rhizosphere priming effect. However, although some of our evidence comes from disturbed soils (i.e. microbial respiration), some (i.e. soil pore water) also comes from intact pots prior to harvest and we now also include another line of evidence from plant root biomass. In short, we draw on multiple lines of evidence suggesting that root exudates were involved, and note that Reviewer #3 thought our approach and interpretation on this aspect of the study was robust.

Having said this, we acknowledge that we were too confident in our interpretation here, so we have added caveats to the text as follows:

L207: “While not directly measured here, a nine-day decay period corresponds to the time expected for newly photosynthesised CO2 to be released through root exudation and respired by soil microbes, suggesting that this carbon pool was mostly root exudates.”

L215: “While further directed studies are required to resolve whether root exudates are truly involved, our findings collectively suggest that lowland plants have the capacity to increase total root exudation into alpine soil relative to resident alpine plants.”

3) Some details of the sampling or measurement are very crucial and affect the results/interpretations. For example, in the field experiment, the soil core was only 1-cm diameter. Considering the spatial heterogeneity of soil carbon in field plots, this small volume may not well represent the true soil condition. Moreover, in the field plots, did soil bulk density change after planting of lowland plants or warming? This will affect the measured SOC concentration (mg/g) even the SOC stock (g/m2) did not change.

We agree with the Reviewer that taking a single soil core of 1 cm diameter in each plot would not have been robust. We did not do this. While we used 1 cm diameter cores to minimise disturbance, we took three cores per plot to account for within-plot heterogeneity and combined them into a composite sample. This is stated in the Methods as follows:

L523: “In each plot, we created a composite sample from three cores (ø = 1 cm, approx. d = 7 cm) no closer than 7 cm from a planted individual and from the same quarter of the plot used for ecosystem respiration measurements (see below; Supplementary Fig. S1).”

We also agree that bulk density measurements were an important omission in the initial submission. We note that this point was fleshed out by Reviewer #3, below, so we refer the Reviewer to our response to that comment for further details.

Reviewer #3:

The authors investigated the effect of warming and herbaceous plant migration on soil carbon (C) content using an ecosystem monolith transplant experiment along an elevation gradient in the Swiss Alp mountains. They observed, approximately 1 year after the transplant, that warming alone had little effect on soil carbon content (monoliths transplanted to a lower elevation with higher temperature remained unchanged in C content) but that the presence of lowland (warm-adapted) herbaceous plants in combination with warming had a negative effect on soil C content. The authors then conducted a glasshouse experiment and used a series of field and laboratory measurements to explore potential mechanisms explaining the observed changes in soil C content in the field. They concluded that soil C losses under lowland plant migration were likely mediated via increased microbial activity and CO2 release from soil C decomposition.

The research questions are extremely relevant to our understanding of the feedback between soil C dynamics and climate warming and remain an unexplored part of this debate. Moreover, both field and laboratory experimental designs are robust, with all the relevant and necessary validation checks needed for transplant experiments; the laboratory techniques employed to measure the range of microbial and plant variables potentially explaining soil C dynamics are adequate and modern; and the statistical analyses are appropriate. These elements make the present data set very relevant and valuable. The manuscript is also very well and clearly written.

We thank the Reviewer, and are delighted that they think the study is extremely relevant, novel, experimentally robust, cutting-edge and valuable.

However, I have two major concerns, casting doubt respectively on the main field results and on the proposed explanatory mechanisms.

First, at no point is bulk density mentioned and it does not appear to have been measured. This is critical because changes in soil C concentration (which was measured and reported here, in mg C g-1 soil) does not necessarily indicate an actual change in the quantity of C present in the soil (C stock, in unit mass C per unit soil volume, or per unit surface area to a constant depth) if this is accompanied by a change in bulk density: if less C per unit mass of soil (lower C concentration) is concurrent with more mass of soil in a constant volume (higher bulk density), this could mean that no change in C stocks actually occurs (or that even an increase occurs). In the present study, it is possible that the presence of lowland plants increased bulk density as compared to only alpine plants, compensating the lower C concentration and resulting in no change in C stocks. This is perhaps not likely, but it is too critical an issue not to be quantified (or at the very least discussed).

This is an excellent point, and one also raised by Reviewer #2. To clarify, we initially decided against measuring bulk density because it is destructive and the experiments were still being used for other studies. Having said this, we agree with the Reviewer that more consideration of soil bulk density was needed, so we have rectified this in three ways. First, although the western Alps experiment has now been taken-down, to address this comment we took new soil cores to measure bulk density in the central Alps experiment in 2021 to indirectly confirm that no changes occurred in the presence versus absence of lowland plants. They did not, and we now include these data in the Methods as follows:

L539: “It was not possible to take widespread measurements of soil bulk density due to the destructive sampling required while other studies were underway (e.g. ref 28). Instead, we took additional soil cores (ø = 5 cm, d = 5 cm) from the central Alps experiment in 2021 once other studies were complete to indirectly explore whether lowland plant effects on soil carbon content in warmed alpine plots could have occurred due to changes in soil bulk density. We found that although transplantation to the warmer site increased alpine soil bulk density (LR = 7.18, P = 0.028, Tukey: P < 0.05), lowland plants had no effect (Tukey: P = 0.999). It is not possible to make direct inferences about the soil carbon stock using measurements made on different soil cores four years apart. Nevertheless, these results make it unlikely that lowland plant effects on soil carbon content in warmed alpine plots occurred simply due to a change in soil bulk density.”

Second, in the Main Text we now caution readers against translating soil carbon content changes to soil carbon stock in absence of coupled measurements of soil bulk density as follows:

L113: “We caution against equating changes to soil carbon content with changes to soil carbon stock in the absence of coupled measurements of soil bulk density (Methods). Nevertheless, these findings show that once warm-adapted lowland plants establish in warming alpine communities, they facilitate warming effects on soil carbon loss on a per gram basis.”

Finally, we have altered the language throughout the manuscript (including the title) to make it clearer that we focussed on soil carbon content/concentration – not stock.

Second, even assuming that no changes in bulk density occurred and that indeed soil C stocks decreased under warming combined with lowland plant migration, the interpretation of the results are, in my view, at least incomplete. Certainly, the results do not support the claim that soil C losses were mediated via increased microbial decomposition of soil C with the certainty suggested by the authors. Generally speaking, I see three issues with the interpretation:

• Very schematically, increased microbial respiration and soil C losses from decomposition is only one of two equally likely pathways potentially explaining soil C losses (the other being decreased C inputs to the soil from the plant community). The possibility that decreased soil C content was simply mediated by decreased inputs of C to the soil is hardly explored at all in the study (there is a quick mention of it (L155), but differences in plant biomass are interpreted only for their correlations with microbial activity (L160-166), not as a component of the C balance. Plant traits are measured and analysed but not in a way that can be used to test the hypothesis of changing C inputs. The presence of "more productive traits" (L141) for the lowland plants does not directly relate to differences in the quantity of C inputs to the soil, nor is it interpreted in relation to inputs. Even the interpretation of changes in ecosystem respiration seem to omit the possibility of changes in plant respiration (L208): "depressed microbial respiration per unit of soil was also evident at the ecosystem scale in that warming accelerated total ecosystem respiration but its effect was dampened in plots containing lowland plants". This statement was made despite no significant differences in microbial respiration per unit soil in the field data, and disregards the possibility that the dampened effect in plots with lowland plants could be due to lower plant respiration.

This is an excellent point. We have performed new analyses of the plant trait/biomass data from the field experiment, included additional measurements/analyses of NEE and GPP from the field experiments (originally omitted due to space, which was a mistake!) and have rewritten all relevant sections in the manuscript to change the focus to a shifting balance between soil carbon inputs and outputs. Importantly, our original interpretation remains robust – i.e. that lowland plants most likely operate by accelerating soil carbon outputs, not decelerating soil carbon inputs – but we are careful to present our conclusions with an appropriate level of caution.

• For the glasshouse experiment, I agree that the results indicate that (L115); "lowland plants accelerated microbial activity by increasing the quantity of root exudates", but not that (L112): "these findings together imply that lowland plants accelerate alpine soil C loss" because stimulating microbial activity is not per se an indicator of soil C loss. It is now well-known that the activity of microbes is not only a motor for soil C losses, but also a key mechanism leading to transformation of C inputs from plants that leads to the subsequent stabilisation of C in the soil. This is actually clearly stated further down in the manuscript when interpreting the field microbial data (L190. Furthermore, there is no direct evidence that the pots with lowland plants were losing more C than those without. Therefore, results from the glasshouse experiment could be interpreted differently: a larger fast cycling pool of soil C constituted of recently photosynthetically fixed exudates associated with higher microbial activity could well be interpreted as an early indicator of more C stabilisation, particularly since the absorbance index seems to indicate more microbially derived product in the DOC. It would have been great to measure microbial biomass C over time (as well as CUE, and mass specific growth and respiration), to see if higher respiratory activity was associated with higher biomass. The lack of differences in microbial biomass between the plant community treatments at the end of the 6 weeks does not show that the quantity of microbial biomass produced over the whole incubation period remained constant. In a word, more respiration of a larger fast cycling pool is not an indicator of future soil C loss (in the presence of plants).

We thank the Reviewer for raising this important point. On reflection, we agree that the previous version of the manuscript did not give sufficient consideration to the possibility for increased microbial activity (and, indeed, respiration) in the glasshouse experiment to signal soil carbon accumulation via increased microbial growth. Having said this, all pots began with the same soil and microbial biomass remained unchanged between alpine and lowland plant treatments at the end of the six-week experiment. By extension, no net microbial growth occurred during this timeframe, making it unlikely that the accelerated respiration observed under lowland plants was indicative of soil carbon accumulation. Sadly, while we can deduce that intrinsic rates of respiration were higher, we can only speculate that growth remained unchanged (no new measurements can be done since growth measurements require fresh soil). We have rewritten the respective section in the manuscript in light of this and the Reviewer’s other comments, which includes the following caveat:

L181: “These findings support the hypothesis that lowland plants have the capacity to increase soil carbon outputs relative to alpine plants by stimulating soil microbial respiration and associated CO2 release. While accelerated microbial respiration can alternatively be a signal of soil carbon accumulation via greater microbial growth, such a mechanism is unlikely to have been responsible here because it would have led to an increase in microbial biomass carbon under lowland plants, which we did not observe.”

• The interpretation of the microbial variables measured in the field line up better with current conceptualisations of the role of microbes in C cycling (but overall interpretation still lacks consideration for plant C inputs). However, interpreting those data measured once 1 year after the transplant to explain the changes that happened gradually over this whole year is a risky and difficult exercise. How do we know that CUE, Rmass, Gmass etc… measured then represent what they were a day, a week, a month before? There is an attempt to deal with this timing issue by comparison with the glasshouse experiment, but only Cmic and Rmass can really be compared and it only very partially fills in the gap in time. Besides, the interpretation of this comparison can be questioned: in the glasshouse, Rmass was higher for the lowland plant pots (as compared to alpine plant at constant temperature) but actually remained constant between the comparable treatments W and WL in the field (Fig 2m). The results from the field, therefore, do not "support observations from the glasshouse experiment" in this context (L197) and neither do they "confirm (…) that this persists for at least one season" (L199). Finally, the thinking around the pulsed nature of C losses seems misplaced because there are no evidence that soil C losses had stopped after a year in the field (no measurements of soil C content are presented after that year).

With regards to plant carbon inputs, we refer the Reviewer to their previous comment for corresponding revisions. With regards to specific comparisons between the glasshouse and field experiments, we have now deleted the sentences in question and have interpreted our results as follows:

L329: “Thus, despite lower rates of ecosystem respiration overall, alpine soil microbes still respired intrinsically faster in warmed plots containing lowland plants. Moreover, accelerated microbial respiration, but not growth, implies that alpine soils had a higher capacity to lose carbon under warming, but not to gain carbon via accumulation into microbial biomass, when lowland plants were present. These findings align with observations from the glasshouse experiment that lowland plants generally accelerated intrinsic rates of microbial respiration (Fig. 3), although in field conditions this effect occurred in tandem with warming.”

With regards to soil carbon loss being pulsed, while there is support for such a mechanism, we agree that this is one of several hypotheses and with only two timepoints we were too confident about it in the original submission. We have now reshaped this section of the manuscript entirely to be more cautious about the temporal dynamics involved. For instance, the section title now reads “Lowland plant-induced soil carbon loss is temporally dynamic”. Some other notable changes are:

L286: “Importantly, lowland plants had no significant bearing over net ecosystem exchange (Fig. 5a), implying that although lowland plants were associated with soil carbon loss from warmed alpine plots (Fig. 1), this must have occurred prior to carbon dioxide measurements being taken and was no longer actively occurring.”

L293: “By contrast, ecosystem respiration in warmed alpine plots was depressed in the presence versus absence of lowland plants (Fig. 5c). These findings generally support the hypothesis that lowland plants affect the alpine soil system by changing carbon outputs. However, they contrast with expectations that lowland plants perpetually increase carbon outputs from the ecosystem and thus raise questions about how soil carbon was lost from warmed plots containing lowland plants (Fig. 1).”

L320: “Carbon cycle processes are constrained by multiple feedbacks within the soil system, such as substrate availability and microbial acclimation, that over time can slow, or even arrest, soil carbon loss. We thus interrogated the state of the soil system in the field experiments in the western Alps experiment to explore whether such a feedback may be operating here, in particular to limit ecosystem respiration once soil carbon content had decreased in warmed alpine plots containing lowland plants.”

L354: “Taken together, one interpretation of our findings is that the establishment of lowland plants in warming alpine ecosystems accelerates intrinsic rates of microbial respiration (Fig. 3, Fig. 6a), leading to soil carbon release at baseline levels of microbial biomass (Fig. 1, Fig. 3c), a coupled decline in microbial biomass (Fig. 6c) and a cessation of further carbon loss from the ecosystem (Fig. 5a, Fig. 6d).”

L358: “Although such a mechanism has been reported in other ecosystems, applying it here is speculative without additional timepoints because field soil measurements came from a single sampling event after soil carbon had already been lost from the ecosystem. For instance, an alternative mechanism could be that soil microbes acclimate to the presence of lowland plants and this decelerated microbial processes over time.”

L368: “Beyond the mechanism for lowland plant effects on alpine soil carbon loss, it is conceivable that soil carbon loss is not isolated to a single season, but will reoccur in the future even without further warming or lowland plant arrival. This is especially true in the western Alps experiment where warming yielded a net output of carbon dioxide from the ecosystem (Fig. 5a). Moreover, in our field experiments we simulated a single event of lowland plant establishment and at relatively low abundance in the community (mean ± SE relative cover: 4.7% ± 0.7%), raising the possibility that increases in lowland plant cover or repeated establishment events in the future could facilitate further decreases in alpine soil carbon content under warming.”

Reviewer #4:

This manuscript took alpine grasslands as a model system and investigated whether lowland herbaceous plants contributed to the short-term dynamics of soil carbon under the context of climate warming. The authors find that warming individually does not render significant changes in alpine soil carbon, but corporately causes ~52% of carbon loss with lowland herbaceous plants in two short periods of field experiments. They further show that alpine soil carbon loss is likely mediated by lowland herbaceous plants through root exudation, soil microbial respiration, and CO2 release. This work adds in an interesting way to the ongoing debate on whether a positive climate feedback will be mediated by plant uphill range expansion in alpine grasslands, where climate warming may lead to a rapid loss of soil carbon.

The claims of this manuscript are well supported, but some aspects of background information in the studied alpine systems and field experiment design need to be clarified.

1) There is an extremely high level of carbon stored in the alpine soils (Figure 1). Climate warming will certainly lead to a great loss of soil carbon in the study systems that could contribute to the positive climate feedback. However, it is unclear for me how the effects of climate warming on soil carbon are relevant to the ongoing climate change in the studied alpine grasslands. It is therefore reasonable to provide more background information about ongoing climate change, and whether the simulated climate warming (i.e., 2.8 oC in central alps and 5.3 oC in western alps, Line 328-329) is realized as real-world climate change in the local systems. In addition, it seems that the manuscript aims to address a question that is of global concern, but my concern is about how the findings could be generalized to other regions.

We thank the Reviewer for pointing this out. With regards to the amount of soil carbon stored in the alpine soils, we refer the Reviewer to comments from Reviewer #2. With regards to the magnitude of warming expected in mountain regions, we agree with the Reviewer that the original submission lacked context. We have therefore added specific values as suggested:

L59: “They are experiencing both rapid temperature change (0.4 to 0.6 ºC per decade) and rapid species immigration…”

With regards to how findings could be generalised to other regions or ecosystems, this is an important point that requires further research – and which we raise in the concluding paragraph. However, we see that we could have been more explicit about validating our findings in other mountain regions, so we have amended the sentence in question as follows:

L400: “Future work should focus on testing the conditions under which this feedback could occur in different mountain regions, as well as other ecosystems, experiencing influxes of range expanding plant species, on quantifying how deeply it occurs in shallow alpine soils, and on estimating the magnitude of the climate feedback given both ongoing warming and variation in rates of species range shifts.”

2) I understand that the manuscript considers elevation as a natural gradient of climate change, which makes it possible to compare soil carbon dynamics in lowlands with alpine grasslands under climate warming. I also understand that the authors have done everything they can to control for the disturbances caused by transplanting that has been well justified by the supplementary data (e.g., Figure S6). However, it is unclear how the authors controlled for the influences of other factors given there are huge differences between lowlands and alpine grasslands, such as differences in wind, solar radiation, humidity, and the length of growing season.

This is an excellent point. We note that Reviewer #1 also raised this point, so we refer the Reviewer to our response to that comment for further details.

3) It is generally known that different species respond to climate warming differently. Some species may be sensitive to climate warming and have traits aiding to dispersion that could expand their living ranges to some degree, while others may adjust themselves to adapt to climate warming and may not migrate to alpine systems. It is therefore cautious to assume that all the lowland species have the same dispersal ability. In other words, it is unclear how lowland plant species are selected for the field transplanting experiment (Line 284-290). Do all the lowland plant species selected have the potential to migrate to alpine systems?

This is an excellent question. In short, the specific dispersal abilities of lowland species used are currently unknown and will certainly vary. However, all are widespread and we assume have the capacity to migrate to higher elevations, given that horizontal distances between high and low elevation sites were in both cases less than 2 km. We now clarify this in the manuscript as follows:

L433: “While exact dispersal distances for selected lowland species are unknown, all species are widespread and are expected to migrate uphill under warming and the horizontal distance between high and low sites in the field experiments was always less than 2 km.”

4) The authors acknowledge that "we did not perform a reverse transplantation (that is, from low to high elevation), so we cannot entirely rule out the possibility that transplantation of any community to any new environment could yield a loss of soil carbon" (Line 318-320). When I read the title "lowland plant migrations into alpine grasslands …", I thought lowland plant species that were transplanted from low to high elevation. In fact, it is just the opposite to my thoughts. Without performing a reverse transplantation experiment, I am not sure the conclusion will stand that "lowland plant migrations into alpine grasslands amplify soil carbon loss under climate warming". In addition, it is unclear whether lowland plant effects stand alone or depend on climate warming based on the results in Figure 1 that lowland plant treatment is missing, and it is impossible to test the interactions between lowland plant and climate warming.

We apologise for the confusion. This comment echoes other comments from Reviewer #1 asking us to be more explicit about the treatments used when interpreting findings, to caveat the step in logic from transplantation to warming and to acknowledge throughout the manuscript that lowland plant effects were dependent on transplantation in the field experiment. We therefore refer the Reviewer to our responses to those comments for details on how we resolved this. We have also modified the title and abstract to more accurately represent the experimental design, as follows:

Title: “Lowland plant arrival in alpine ecosystems facilitates soil carbon loss under experimental climate warming”

L30: “Here we used two whole-community transplant experiments and a follow-up glasshouse experiment to determine whether the establishment of herbaceous lowland plants in alpine ecosystems influences soil carbon content under warming. We found that warming (transplantation to low elevation) led to a negligible decrease in alpine soil carbon content, but its effects became significant and 52% ± 31% (mean ± 95% CIs) larger after lowland plants were introduced at low density into the ecosystem.”

With regards to testing the interaction between warming and lowland plants, while we acknowledge that not performing a fully-factorial design limited our ability to explicitly separate lowland plant versus warming effects on alpine soil, both are occurring simultaneously due to climate warming and we thus focussed effort on simulating such a scenario with greater experimental replication and at multiple locations. We note that Reviewers #1, #2 and #3 thought that this approach was robust. Importantly, the statistical analyses performed are valid for such an experimental design, and we have clarified and nuanced our interpretation throughout to avoid reaching beyond it.

Author Response:

Reviewer #1 (Public Review):

This paper uses a combination of confocal and electron microscopy to localize gap junctions in the outer retina. Electrical coupling between photoreceptors is an important aspect of retinal function, and past work provides (often indirect) evidence for rod-rod, rod-cone and cone-cone coupling. The work described here indicates that rod-cone coupling dominates. The combination of techniques is quite convincing and very elegant. My concerns are primarily about the appeal of the work to non-retina readers. Some of these concerns could be mitigated by a more accessible presentation of some of the results. Suggestions along these lines, and a few other minor issues, follow.

Introduction:

The introduction is a bit retina-centric. I think more needs to be done to explain how each type of coupling (rod-rod, rod-cone, cone-cone) could impact retinal processing, and why it is important to resolve which are present or dominant. One issue that could get emphasized is the difference between gap junctions between like cell types (presumably involved in lateral spread of signals, averaging, etc) and between unlike cells (potentially providing an alternate path for signal flow - as in the secondary rod pathway).

We have included new text in the introduction to address this issue. We have tried to provide background material of a general nature and we have included some introductory text about different types of gap junctions, as requested. We thank reviewer 1 for this helpful suggestion.

Cone-cone coupling:

It would be helpful to put the conclusions about rod-cone and cone-cone coupling together. The paragraph starting on line 585 is a bit confusing that way. It starts by summarizing evidence that blue cones are not coupled with red/green cones. But then (in mouse) all the cones are coupled to rods, so that specific exclusion of blue cones seems unlikely to hold. You come back to this a bit later in the discussion, and there indicate that there appears to be weak cone-cone coupling. Merging the text in those two locations might help. It might also help to make the (seemingly clear) prediction that blue and green cone signals in mouse will get mixed.

Thank you for pointing out that this section is not clear. It seems two different points are muddled: 1) Blue cones do not make gap junctions with other cones, perhaps to minimize spectral mixing: the evidence from primate and ground squirrel suggests that blue cones are not coupled to red/green cones or green cones. 2) In contrast, we find no evidence of color selectivity in rod/cone coupling: green cones and blue cones are both coupled to all nearby rods. Thus, rod signals can be injected into the downstream pathways of both blue and green cones.

We have rewritten the text and separated these points into separate paragraphs for clarity, as below.

Revised Text:

Blue cone pedicles are also coupled to rods.

In the cone networks of primate and ground squirrel retina, there is good evidence that blue cones are not coupled to neighboring red/green (primate) or green cones (ground squirrel) (Hornstein et al., 2004; Li and DeVries, 2004; O’Brien et al., 2012). In the primate retina, the telodendria of blue cones are few in number and too short to reach the neighboring red/green cones (O’Brien et al., 2012). Thus, blue cones appear to be electrically separated from other cones in these two species, perhaps to maintain spectral discrimination (Hsu et al., 2000). In the mouse retina, although the blue cones were identified by Behrens et al., (2016), we were unable to find any cone to cone gap junctions, regardless of color (see below).

In contrast to the selective connections between cones in some species, rods were coupled to both blue and green cones indiscriminately in the mouse retina (present work) and in primate retina (O’Brien et al., 2012). Blue cones, identified in confocal work by the presence of S-cone opsin, and in SBF-SEM by their connections with blue cone bipolar cells (Behrens et al., 2016; Nadal-Nicolás et al., 2020), and green cones both made telodendrial contacts at Cx36 clusters with all nearby rod spherules (Fig. 4). Thus, we find no evidence for color specificity in rod/cone coupling. In fact, a single rod spherule may be coupled to both blue and green cones (Fig. 5, supplement 5). Therefore, rod signals can pass via the secondary rod pathway into both blue and green cones and their downstream pathways. Considering blue cone circuits specifically, rod input to blue cone bipolar cells and downstream circuits is predicted via the secondary rod pathway, in addition to the previously reported primary rod pathway inputs from AII amacrine cells to blue cone bipolar cells (Field et al., 2009; Whitaker et al., 2021).

Relation to other circuits:

Are there implications of the present results for gap junctional coupling in other circuits that could be emphasized? Things like the open probability how strongly it can be modulated seem like points of general interest - but I don't have enough expertise to know if those are established facts on other systems. Some of that is touched on in the Discussion, but quite briefly.

In an effort to keep the discussion short, we have perhaps been too abrupt. We have added text to the discussion to include some general issues concerning gap junctions.

Location of Cx36:

Can you speculate on why Cx36 is generally located at the mouth of the synaptic opening in the rod spherule? This was a very clear result, but it was unclear (at least to me) if it was important.

This is an interesting topic and we have expanded the discussion to consider potential functions and mechanisms.

Added to discussion:

The position of rod/cone gap junctions, at the base of the rod spherule, close to the opening of the post-synaptic cavity, appears to be systematic in that the vast majority of rod/cone gap junctions occur at this site. We may speculate that gap junctions are localized with some of the same scaffolding proteins that occur at the rod synaptic terminal, but the functional significance of this repeated motif is unknown. In mutant mouse lines, where Cx36 has been deleted from either rods or cones, cone telodendria are still present and they still reach out to contact nearby rod spherules in the absence of rod/cone gap junctions. Therefore, the specificity of synaptic connections is not determined or maintained by the presence of Cx36 gap junctions.

Reviewer #2 (Public Review):

Previous studies demonstrate that modulation of gap junctional coupling in the outer plexiform layer of the mouse retina regulates the balance between sensitivity and resolution. The authors use optical and electron microscopy to structurally characterize this coupling. They find that gap junctional coupling in mouse OPL is produced by a dense meshwork of cone photoreceptor telodendrions that selectively innervate the rim surrounding the synaptic openings of rod photoreceptor spherules. The density of this coupling network is such that each cone is coupled to dozens of rods and each rod is coupled to multiple cones. Rod/rod and cone/cone gap junctions were not detected.

The combination of antibody labeling, reconstruction of the photoreceptor terminal network, and ultrastructural analysis provides a remarkably clear view of the gap junctional connectivity that constitutes the first stage of visual processing. A few results are only weakly supported due to sample size or technical limitations. However, the overall conclusions are well supported and the data is presented with unusual transparency. The map of the network organization of photoreceptor coupling generated here is an important contribution to visual science.

Optical imaging:

The quality of the confocal imaging is high and the images of the Cx36 distribution relative to rod spherules is convincing. There does seem to be a significant amount of processing in the images and a lack of background signal in antibody images. Whether this processing is due to the airy scan software or additional filtering and thresholding, it can be difficult to judge the distribution of signal in several images.

In general, there was no filtering or processing of any confocal images, except for adjusting brightness and contrast. However, we may have been over-zealous in reducing the background. Therefore, we have adjusted Figures 1 and 2 to include more background as requested, to enable the reader to better judge the specificity of the immunolabeling. In addition, we have prepared supplementary figures to show the individual channels with background, as well as the combined images, to be absolutely clear and transparent. Finally, for each confocal image, the confocal series from which it was derived has been archived and is publicly accessible.

Former Figure 1D, now Fig. 2D is an exception because it shows a 3D projection of the colocalization between a single EGFP labeled cone pedicle and Cx36. We have revised this figure, providing new 2D optical sections to show how the image was prepared, in addition to revising the final 3D projection, labeling it as a 3D projection with colocalized Cx36.

Electron microscopy:

The authors perform annotations on two previously acquired volume EM datasets. The first serial blockface EM dataset is relatively low resolution and lacks ultrastructural labeling but is used effectively to reconstruct the terminal morphology and points of contacts between photoreceptors. The second EM data set uses FIB SEM to obtain smaller voxel sizes from tissue stained in such a way that the darkened membranes of putative gap junctions are distinct from surrounding membrane. Most measures of gap junction number come from the ultrastructure free dataset. In isolation, counting of gap junctions in this type of image volume could be unreliable. However, comparing the putative gap junctions in this dataset to the morphology and distribution of Cx36 antibody clusters in the confocal imaging and the darkened plaques in the FIB SEM images greatly increases confidence that the network description of rod/cone gap junctional coupling is accurate.

Quantification:

Most quantification is presented with an unusually high degree of transparency, with scatterplots showing all data points, data source files showing the animals that data came from, and standard deviations being supplied in descriptive statistics. There are a few places where Ns are difficult to determine or the analysis is not quite clear. For several results, claims are made when the sample size is too small to be sufficiently confident. The reconstruction of 5 blue cones suggests that, overall, blue cones are not radically different from other cones in their terminal morphology or gap junctional coupling to rod spherules. Claims that the blue cones are identical to other cones in most measures or that their telodendrions are smaller, but not statistically smaller are not well supported by the sampling. Similarly, the fact that the 6 nearby cones closely analyzed for cone/cone gap junctions yield no junctions, strongly suggests that vast majority of gap junctions are cone/rod gap junctions. However, the sample is too small to argue that there could not be infrequent, atypical, or region-specific cone/cone gap junctions.

We have addressed the issues of blue cones and cone/cone coupling to soften our conclusions and explicitly point out the small numbers.

Estimate of open channels:

The authors estimate that 89% of gap junction channels are open during times of maximum rod/cone coupling and point out that this number is surprisingly high relative to previous estimates. However, this estimate appears to be subject to many significant potential errors. The estimate combines previous freeze fracture studies of the density of gap junctions from various species and various parts of the retina the measurements of the length and width of the gap junctions in the current study. Differences in tissue processing, density variation within and between systems, reconstruction error, and variation and error in the inputs to the model could all contribute to an underestimate of the total number of channels linking mouse rods and cones. Moreover, without an accounting of these issues, the real error bars on the range of possible open channels would seem to include both surprising and less surprising estimates of open gap junction fractions.

This is a major issue. In short, for the calculations of open probability, we have estimated the cumulative errors, added these numbers to the text and attached an appendix showing the statistical analysis. We have also added a section to the discussion to address the possible sources of error enumerated by reviewer 2.

Reviewer #3 (Public Review):

In the presented work, Ishibashi and colleagues combine immunohistochemistry, analysis of a publicly available large scale 3D EM dataset and smaller but more detailed newly acquired EM datasets to qualitatively and quantitatively study gap junctions of mouse rod and cone axon terminals. The existence of rod-to-cone gap junctions has been known before, but the use of larger 3D EM data allows to determine an average number of contacts as well as an estimate of the strength of gap junctions. This as well as the (very likely) exclusion of direct cone-to-cone coupling in the mouse as opposed to some other mammals are the main contributions of this paper and one more puzzle piece of the big picture of mouse retinal connectivity. However, while the findings are a valuable addition towards a complete picture of the connectivity in the mouse retina, the novelty of the findings is limited to the number of contacts per photoreceptor and gap junction sizes.

In my opinion, while the authors present a thorough analysis of their data, the manuscript in its current state has stylistic flaws on the motivational side. To me, abstract and introduction lack a motivation or stronger statement of relevance for this analysis. Similarly, while each individual analysis is discussed one by one, I'm missing a broader discussion of the implications of the findings for the field and possible directions for future research to highlight relevance for a broader readership.

Thank you for the positive comments. We have rewritten and added material to the Abstract, Introduction and Discussion in an attempt to explain the reasoning for this study and to explain the findings to a broader audience.

I believe these questions are so important to consider when creating any lesson! As someone who needs time adjusting to new tools or apps, I have experience many times that I have had to focus more on how to use the tool than the content we were using it for. I think when you consider these questions, it helps make sure the students will all benefit from the tool. It may take some extra time during planning, but it will be more beneficial in the long run.

Author Response:

First we would like to thank the reviewers for their very kind words regarding our manuscript and for their helpful suggestions for how to improve our paper. We believe their suggestions have helped to strength the paper as a whole. We will address below the specific weaknesses that the reviewers have brought up and describe how we have modified our manuscript in response to these suggestions.

Reviewer #1:

This is an interesting study of the relation between vividness of visual imagery and the pupillary light response that can result from it. The authors collected data in two experimental paradigms, which they ran in two independent samples. One of these samples was a larger group of psychology students; the other a self-reported group of people with aphantasia. In a first paradigm, the authors show that a lack of vivid imagery is associated with a smaller (or even absent) pupillary light response. Using a second paradigm, binocular rivalry, they show that the degree to which imagery primes binocular rivalry is correlated (to a degree that is quite striking) with the magnitude of the pupillary light response to imagined stimuli. These results were obtained both for low-scoring individuals in the large sample as well as for the aphantasics. The study provides objective evidence for the absence of imagery in individuals that self-report as aphantasic.

The paper is well written and all the necessary controls for potentially confounding variables are in place. For instance, age or visual persistence are discussed and excluded as alternative explanations based on convincing analyses. A particular strength of the manuscript is that the authors report positive results for pupillary responses in the group with aphantasia. That is, these individuals show regular pupillary responses to changes in physical stimulus brightness as well as to cognitive load. Another strength is that the group of aphantasics was invited separately and not determined post-hoc in the initial sample.

In summary, there is a lot to like about this paper. I have three comments / questions that I think should be addressed, however.

1. A point that I would like to see analyzed and discussed is the role of eye movements. The authors do not report any analyses of fixation behavior or the frequency of saccades in the two groups. These should be analyzed and reported. The only mention of fixation control is in lines 423-424, but the authors remain at a very superficial level, stating that footage from this scene camera of the pupil labs eye tracker was "assessed to ensure fixation on the computer monitor". Does this mean that participants could look anywhere provided they looked at the monitor?



We have now analysed the eye-movements of participants to assess whether or not they might be driving some of our findings, which we agree is a very important additional analysis to add to this paper to confirm our findings are not being driven by eye-movements. When analysing both eccentricity and the number of saccades participants made there was no differences between the two groups when imagining the triangles (see supplementary figures s7 and s11). There was also no correlation between eccentricity data and either the imagery pupillary light reflex or binocular rivalry priming. Taken together it seems unlikely that the observed pupillary light response during imagery is being driven by eye-movements.

1. In Figure 1D (also lines 120-124), the authors show a correlation between vividness ratings and the pupillary light response. I assume that participants differ substantially in their distributions of responses. So these correlations could be a consequence of individual differences or they could provide evidence for trial-by-trial variation. There might be ways to find out. For instance, is there evidence for these correlations at the level of individuals? Does the correlation persist if individual vividness-response distributions are normalized to span the same range for each observer?

We would like to clarify the analysis we ran. Figure 1D is the results of 2 x 4 linear mixed-effects analysis, not correlations. This model included subject identity as a random effect (see Methods section of our paper) and therefore the effects reported were computed at the subject level. We report in the text, effects that are significant at the level of the sample. This does not exclude the possibility of inter-individual differences, but we are not sure how interpretable a single-subject analysis is in the current study.

1. In lines 314-315, the authors state that the pupillary light response to imagined stimuli may serve as an objective indicator of aphantasia. I think this is taking the interpretation of the data too far, mainly for two reasons. First, the authors haven't shown that low pupillary light response predicts aphantasia in a group of people that does not self-report as aphantasics before the test. Second, the absence of a pupillary light response (in a new sample with no additional controls) could also indicate a lack of motivation to engage in imagery. The authors should thus clarify that such tests would always have to be combined with positive tests that show the commitment of participants to the task instructions.

We agree that it is very important to include positive controls in not only pupillary light response imagery tasks, but any task that measures imagery or any other internal experience. We have now expanded on this point in our discussion as well as reporting on the mock binocular rivalry trials that were included in the priming imagery task as a control for potential response biases.

Reviewer #2:

Kay et al. investigated visual mental imagery in the general population and the lack thereof in individuals with aphantasia by measuring the pupillary light response to imagined light and dark shapes. Their findings are twofold. First, they show a link between pupil size change and perceived vividness of imagery and corroborate this finding using another established objective measure of vividness. Secondly, they found a lack of such a pupillary light response in a group of individuals who maintain no visual experience of imagery. This demonstrates the usefulness of using the pupillary light response as a measure of subjective vividness of imagery and potentially demonstrates the first physiological finding in aphantasia.

Strengths

The experiment incorporates several different dimensions into a single clean design that is useful for isolating and tracking multiple relevant measures. First, by having the brightness of the perceived and imagined shapes vary across trials, the authors could show that changes in the pupillary light response correspond to changes in imagined brightness. The authors also added in an independent number-of-objects dimension since pupil size also varies with cognitive effort. This provided evidence that aphantasic subjects were attempting to imagine, since the pupil size did change with set size, even when it didn't change with brightness. Finally, by having subjects report the perceived vividness of each imagined image, the authors could link subjective experience of imagery to the pupillary light response.

The authors also strengthen their findings by comparing changes in pupil size to an objective measure of imagery vividness. By leveraging the fact that imagery mimics vision's ability to bias a perception during binocular rivalry, the authors avoid the severe limitations present in measures that rely on introspection only.

Weaknesses

Due to the inherently private nature of mental imagery, ruling out fabrication or demand characteristics is extremely difficult. This is especially true in aphantasia research, as we are often looking for the absence of an effect rather than an enhancement. Readers should keep in mind that, while the authors made some effort to confirm that the aphantasic subjects were attempting to imagine, the potential for this and other biases were not ruled out. Without the use of probes to test subjects on the remembered/imagined objects and reporting the outcomes of catch trials, it is difficult to tell whether subjects were fully engaging with the stimuli.

Readers should also take the pupillary light response as a tool to add to the battery of assessments for aphantasia, not as one that a diagnosis can be based on alone. While the authors do show a group level difference in pupil size in response to imagined shapes and claim it as a "new low-cost objective measure for aphantasia", it should be remembered that this manuscript does not demonstrate the tool's efficacy in identifying individual subjects with aphantasia. The absence/presence of an imagery pupillary light response does not confirm/rule out aphantasia.

Overall, the manuscript helps characterize an intriguing condition that until relatively recently received little empirical attention. These findings support the internal experiences described by aphantasic individuals, experiences that are often met with skepticism. Importantly, the authors have also offered the field a new objective physiological approximation of imagery vividness which can be incorporated into a number of study designs examining changes in imagery. The majority of previous measures relied on self-report alone and often suffered from the limitations of language (e.g., what it means for something to be "like vision" can be very different for different people). This manuscript also adds to the growing body of evidence of the power of internally generated signals, which can apparently reach all the way down the visual hierarchy to the eyes themselves.

We are in full agreement that when we investigate the internal contents of the mind we need to be mindful of the many caveats that exist when relying on people’s ability to introspect. We agree that future studies should expand on our research by adding in further controls, such as having participants report what item they were asked to remember at the end of the trial. However researchers should also keep in mind that changing the demands of a task can alter how participants undertake a given task. For example by emphasising remembering the items, rather than creating detailed vivid images in mind, participants may revert to a non-visual imagery strategy to remember the items, such as labelling the items. This may be particularly easy to do in the current study as the items being imagined are simple geometric shapes. Indeed it was important to avoid this potential pitfall here with our aphantasic population as we have previously shown that aphantasic individuals can perform a wide array of visual working memory tasks despite their lack of visual imagery. We believe that the addition of a set-size/cognitive load condition, plus our added reporting on mock trails helps to answer some of these potential response bias issues, but future research can and should further investigate these potential biases in greater detail.

The second point Reviewer #2 brings up is a very good one, that no one singular measure in isolation, at this point in time, can be used to ‘diagnose’ aphantasia. The field is very young and we are still in the process of understanding exactly what aphantasia is. For example there may be many subtypes of aphantasia, with previous work from our group and others showing that aphantasic individuals are heterogenous in their reporting of how other imagery modalities are affected. We agree with Reviewer #2’s point that a battery of tests, potentially comprising questionnaires (e.g. VVIQ), psychophysical tasks (e.g. binocular rivalry paradigm) and physiological (e.g. skin conductance, pupillometry) should be aimed for where possible in testing aphantasic populations. The pupillary light response is a new tool that can be added to this arsenal.

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity (Required)): *

In this manuscript by Wang and colleagues, the authors analyse single-cell RNA-seq (scRNAseq) data by applying transition path theory to infer gene regulatory network (GRN) changes along the transition (reaction coordinate, trajectory) between free energy stable states (i.e. cell types). The work aims to understand how stable cell types, and their regulatory programs (combination of active and repressed genes) switches during differentiation/reprogramming/response (i.e. cell phenotypic transition/CPT). The premise of the work is to assess whether genes within GRNs undergo step-wise repression, state-change and activation (& vice-versa; analogous to SN1) or concurrently regulate gene expression (analogous to SN2). The GRNs are inferred based on highly variable genes and their expression dynamics from RNA velocity over CPT, across 3 scRNA-seq datasets.

The authors first analyse public scRNA-seq dataset of 3003 human A549 adenocarcinomic basal epithelial cells treated with TGF-b for 0hrs, 8hrs, 1 day and 3 days (4 timepoints). The authors select two stable states (Day0-untreated; Epithelial and Day 3-treatment; Mesenchymal) using local kernel densities and set transition paths using Dijkstra shortest path, dividing state space into Voronoi cells (i.e. reaction coordinate value), and constructed single-cell GRNs based on RNA velocity differences (n=500 genes) and a linear model (from Qiu et al). This GRN is based on expression and velocity estimates, and does not distinguish direct from indirect regulation. Calculating interaction frequency (edges) across two stable states over 4 louvain clusters, the authors find global increase in effective edges that correlates with increased active genes; but with variable trend within inter-cluster edges. To quantify the concerted GRN changes between clusters, the authors utilise a "frustration" score (from Tripathi et al 2020). The average frustration score increases and peaks at day 1 treatment, followed by a decline over terminal stable state (day 3-treatment); similar to interaction frequency trends. The author also separately measure network heterogeneity and repeat analysis using alternative transition matrix. The authors conclude that EMT proceeds through concerted regulation of multiple genes first with an increase in inter-cluster edges, frustration and heterogeneity followed by a decrease into final stable state. The authors apply the analysis to scRNA-seq data from (i) pancreatic endocrine differentiation where Ngn3-low progenitors give rise to Ngn3-high, then Fev-high and into glucagon producing a-endocrine cells; (ii) dendate gyrus; radial glial cell differentiation into nIPCs, neuroblast, immature granule and mature granule cells. In both cases, the authors observe concerted regulation with initial increase in inter-community edges, heterogeneity during differentiation followed by decrease towards final stable state. **

The study and ideas in the manuscript are interesting and the methods would be potentially be useful. However, there are a few specific and general comments stated below, which the authors should try to address.

1 • P4: "RC increases first and reaches a peak when cells were treated with TGF-β for about one day, then decreases (Fig. 1G)". It would be better to label the figure with the treatment information. *

Reply: Thanks for your advice. In the revised manuscript, we analyzed two additional datasets, and moved the EMT result in the supplemental Fig. EV8. In the new Fig. 1d, we marked the cell types along the reaction coordinate.

*2 • Fig. 1G and EV1D: Why are the trends different? *

Reply: In the original figures, ____Fig____.1g is the frustration score and EV1D shows the variation of pseudo-Hamiltonian along the reaction coordinate. The frustration score is the focus of this work. We also calculated the pseudo-Hamiltonian since it has been used in the literature. However, we realized that showing both of the results might lead to confusion, so we deleted all pseudo-Hamiltonian results in the revised manuscript.

* 3 • How is the appropriate community/cluster/Louvain resolution selected? This can have a major impact on number of cell states, types and transition path from initial to final state. *

Reply: The number of cell states, types and transition path from initial to final state____ are not determined from the community/cluster/Louvain analyses. For the EMT data, we assume most cells in the initial treatment time are epithelial cells, and those in the final time point are mesenchymal cells. For other datasets, we followed the original publications to assign cell types based on known marker expression.

The Louvain method was applied to coarse grain the gene regulation network, and it does not affect the number of cell states, types and transition path, which were determined separately. To address the reviewer’s question, we also use the Leiden method to adjust the resolution ____(1)____. The resolution does not affect the result. The results are added to Fig. EV12. We tried three different resolution values 0.8,1.0 and 1.2. The number of inter-community edges consistently shows the trend that it increases first then decreases.

Figure EV12 Cell-specific variation of the number of effective inter-community edges between communities calculated with different resolution parameter values for dentate gyrus neurogenesis (a), pancreatic endocrinogenesis (b), and bone marrow marrow hematopoiesis (c). Each dot represents a cell and the color represents the number of inter-community edges____.

• * What effect does the Louvain resolution have on e.g. frustration scores? * Reply: The resolution of community division algorithm doesn’t affect the frustration scores, since the frustration score is based on the gene-gene interactions instead of community assignment.

• * The authors match resolution to samples/timepoints/known prior cell types i.e. 3-4 communities. However it is unclear whether this is enough to describe entire differentiation/transition process. * Reply: This is a good question. In one above reply we have explained how the cell types were determined____. We also agree with the reviewer that these coarse-grained communities cannot reflect the overall heterogeneity and dynamics of the whole process. Notice in most of our analyses (e.g., reaction coordinate and transition paths), we treated the transition as continuous and the distribution of single cell data points in all datasets cover the whole space that involved in cell phenotype transition. The coarse-grained analyses are for further mechanistic insights on how gene regulatory networks are reorganized during the transition process.

• * Gene selection: The selection based on minimum 20 counts as highly expressed genes is arbitrary and dependent on sequencing depth. Perhaps the authors could show distribution of gene counts for the datasets and have a data-driven filtering criteria * Reply: Thanks for the advice. The number 20 is a default value suggested in the package (scVelo) we use, and in another package dynamo the default number is 30. Following the reviewer’s suggestion (together with the next question on the influence of all highly variable genes), we looked for a data-drive filtering criterion. The method has been described in different tools ____(2-4)____. We first grouped the genes into 20 bins by their mean expression values, and____ scaled their dispersions by subtracting the mean of dispersions and dividing standard deviation of dispersions____. Figure EV9 shows the distribution of the minimum shared counts. ____As one can see, most genes counts are larger than 10, and using a smaller value causes error in the following velocity analysis. Therefore we set the minimum shared counts as 10 in the new results.

Figure EV9 Shared counts distribution of the datasets. (a) Dentate gyrus neurogenesis; (b) Pancreatic endocrinogenesis; (c) Bone marrow hematopoiesis.

• * The choice of 500 variable genes (for human A549 cells) is also quite arbitrary. Perhaps, the authors could compare how additional genes (all highly variable genes) affects their analysis and interpretation. * Reply: ____Thanks. Following previous question on shared counts and ____data-driven filtering criteria____,____ we take all the highly variable genes into consideration. The details of gene selection and binarization are given in the Materialss and Methods (Materials and Methods 2) section.

• * How are other factors (sequencing depth, genes detected, #of cell types, multiple branches) affects the connectivity between communities at different phases of transition/development? * Reply: This is a good question. The A549 EMT dataset has a sequence depth of 40000-50000. The ____dentate gyrus neurogenesis dataset____ has a sequence depth of 56,700 reads. A saturation depth would be close to 1,000,000, but there is a compromise between cell number and depth. There are genes that are not detected even under the saturation reads setting. That is why the preprocessing is needed. On the other hand, the network we inferred include both direct and indirect interaction, so the influence of sequence depth and gene number detected can be reduced to a certain extent. We used a random subset of the selected gene and performed the same analyses. The results are consistent with what we obtained using all the genes (Fig. EV11b). With the new gene selection criteria (Materials and Method 2), our analyses are not related with the number of cell types.

We did analysis on another beta branch of pancreatic endocrinogenesis data. The other branches show the same results (Fig. EV4). There are two additional branches in the pancreatic endocrinogenesis dataset. It has been reported that the RNA velocity estimation for the epsilon branch is incorrect ____(3)____. There are too few cells in the delta branch for reliable analyses. Therefore we didn’t present results for these two branches.

Figure EV4 Analyses on the branch of glucagon producing β-cells in pancreatic endocrinogenesis.

(a) Transition graph based on RNA velocity.

(b) The RCs and corresponding Voronoi cells. The large colored dots represent the RC points (start from blue and ends in red). The small dots represent cells with color as cell type.

(c) Frustration score along the RCs.

(d) Cell-specific variation of effective intercommunity regulation. Each dot represents a cell. Color represents the number of effective intercommunity edges within each cell in the GRN.

• • Are the velocity graph, transition matrix and further shortest path estimation derived in a reduced latent space, and if so, how much (nPCs) and what impact does it have. Presumably, the density estimation is not performed in expression space. Reply: Yes. ____The calculation of transition matrix is based on neighbor information. The calculation of neighbors was in the reduced latent space in scVelo and Dynamo. We performed the same analysis by varying number of principal components. The results are similar because the first several components account for large proportion of variance. Figure R1 shows the results of dentate gyrus neurogenesis with the number of principal components being 10, 20 and 30, respectively. In the revised manuscript, we delete the step of using density estimation constrain to simplify the procedure. __Figure R1 Frustration scorer along RCs (left) and cell specific variation of number of effective intercommunity edges (Each dot represents a cell and color represents the number of effective intercommunity edges) in the GRN within each cell (right) when using different number of PCs in analyses (dentate gyrus neurogenesis): (a) number of PCs is 10.*__

(b) number of PCs is 20. (c) number of PCs is 30

* - The figure legends and labels were hard to read. These should be improved for better readability. *

Reply: Thanks. We modified the figure legends and labels.

* - A suggestion would be move the initial results section to methods and highlight the biological interpretation. *

Reply: Thanks for your advice. We moved large part of this section to the Materials and Methods.

*The authors could highly which GRN and representative genes/edge pairs are highest ranked within inter-community and to overall final stable states. *

Reply: Thanks. We list some representative gene pairs in the Table. EV 2&EV 3 &EV 4 for different datasets. And we performed gene enrichment analysis for each community.

* - How does the GRN inference compare to current state-of-the-art GRN inference scRNA-seq methods? *

Reply: we used the method GRISLI to perform the same analysis ____(5)____. The results are similar to what obtained with our current method (Figure EV6). We want to emphasize that the focus of this work is not on another GRN inference method, but discussing some general principles of GRN reorganization during a cell phenotypic transition process.

Figure EV6 Analyses of datasets of dentate gyrus neurogenesis (a), pancreatic endocrinogenesis (b), and hematopoiesis (c) based on GRN inferred with GRISLI.

(a) Frustration score along the RCs of dentate gyrus neurogenesis (left) and cell-specific variation of the number of inter-community edges (right). Each dot represents a cell and color represents the number of inter-community edges in GRN within each cell.

(b) Same as in panel (a), except for pancreatic endocrinogenesis.

(c) Same as in panel (a), except for hematopoiesis.

* - How do extremely noisy/stochastic genes vary in metrics between final stable states? How are the metrics affected by number of cells and stochasticity of expression within a given cluster/community. *

Reply: To address this question, we selected two genes, Id2 and Cdkn1c, with high variance and compare their distributions in the initial and final states. ____The gene distributions show significant shift between the Ngn3 low EP cells and Alpha cells (Fig. R2 a &b left).____ Then we randomly selected a subset (half) of cells and compared the distributions of these high-variance genes in the sub-population (Fig. R2 a&b right). The results are similar to the full-set results.

Fig. R2 Comparison of gene distribution in the initial and final states in pancreatic endocrinogenesis. (a) Comparison of the distribution of gene Id2 at the initial and final states (left), and in the randomly selected sub-population at the initial and final states (right). (b) Comparison of the distribution of Cdkn1c at the initial and final states (left), and in the randomly selected sub-population at the initial and final states (right).

* - Given that the author's approach includes both direct and indirect genes effects, the authors could further prune genes based on existing TF databases or protein-protein validated networks. *Reply: This is a good suggestion. We will work on this idea in future work. As we mentioned, due to constrains of data quality, only tens of transcription factors can be analyzed in these dataset. We list some regulations of transcription factors inferred with current method in Table EV1.

• *It is unclear which GRNs are already known and which ones are novel and biologically relevant * Reply: We compare some regulations inferred with the method and compare these interactions w____ith some references in Table. EV1____.

* - It would be good for authors to comment when there are multiple bifurcations instead of A-B transitions. Particularly in datasets with multiple discrete stable states. *Reply: This is a good question.____ In our analysis, we focus on the transition from one stable state to another stable state. For transition process with multiple bifurcations like____ the pancreatic endocrinogenesis, the results are similar across different branches. For the transition that goes through multiple discrete stable states, for example, a transition from state A____à____B____à____C, we expect to observe two peaks in the frustration score and the number of inter-community edges. We added some discussions in the Discussion section.

• *Another suggestion would be to highlight gene expression of selected markers based on f-regression and mi over the trajectory * Reply: As we modified the criteria of gene selection, we plotted trajectories of some high-variance genes versus the reaction coordinate obtained with different datasets in Fig. EV10 based on current criteria.

Figure EV10 ____Typical trajectories of high variance genes versus RCs of dentate gyrus neurogenesis (a), pancreatic endocrinogenesis (b) and bone marrow ____hematopoiesis ____(c).

* - If possible, a proof of principle could be re-analysis of a perturbation scRNA-seq dataset (e.g. where one path/transition path is stalled) *

Reply: Thanks. This is a really a good suggestion. We will perform more systematic studies in future work.

* Reviewer #1 (Significance (Required)): Nature and significance of advance: The study and ideas in the manuscript are interesting and the methods would be potentially be useful to community. Compare to existing published knowledge: *

*Audience: Predominantly computational audience *

*Your Expertise: PI with background in experimental, computational biology and expertise in single-cell genomic tools and developmental biology *

*

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

Understanding the cellular and molecular basis of cell type or cell state transitions occurring during development or reprogramming is a fundamental challenge. scRNA-seq has provided a window into gene expression programs across thousands of cells undergoing such transitions. Wang and colleagues leverage scRNA-seq and develop an approach to reverse engineer gene regulatory network underlying cells along a path from one cell type/state to another, and characterize community-level properties of this network associated with various stages of the cell phenotype transition. The study is innovative and rigorous, and their results point to how intercommunity interactions increase and then decrease, indicating a concerted regulatory rewiring that orchestrates transitions. Application of their approach to three different datasets also shows that this trend is consistent across three different transitions and maybe a general trend. However, there are some major and minor concerns that need to be addressed.

**Major comments and questions**

1. The analogy to SN1 and SN2 mechanisms of chemical bond formation is very nice.
2. What is the basis for the two statements made in paragraph 3 of Introduction (beginning with "A question arises ...") about transitions being sequential or concurrent? Please *Reply: Thanks. We added references in this paragraph.

* 2.1. Provide references to previous experimental and computational studies that have investigated developmental and reprogramming gene expression programs. *

Reply: Thanks. We added a paragraph in the Introduction.

*

2.2. Describe specific examples of findings that support the two possible transitions highlighted here. Why couldn't transitions happen through an entirely gradual process involving changes to overlapping subsets of genes. *

Reply: Thanks. In the review paper of Naomi Moris et. al., they proposed the hypothesis that cell phenotype transition is similar to a chemical reaction ____(6)____. Thus we extrapolate this hypothesis and test it in our study. For the example of SN1 mechanism, ____Kalkan et al. showed that mouse embryonic stem cells can exit from ____naïve pluripotency____ but remain uncommitted ____(7)____.

Just like the SN1 and SN2 mechanisms are two extremes in chemical reactions and there are cases lie in between, for cell phenotypic transitions we agree with the reviewer that such gradual process may exist. Actually the result in Fig. EV4d shows that the frustration score remains flat for the Fev+ ____à____ Beta transition, suggesting a possible gradual process. With the analyses provided in this work, such as the reaction coordinate, frustration score, heterogeneity, and inter-/intra- community edges, one may perform more systematic studies on a larger number of datasets and enumerate/classify possible patterns of transitions.

• Please make plots of the number of effective intra-community edges vs. number of active genes to support the statement that these two numbers are correlated. *

Reply: We plotted the corresponding intra-community active genes and calculated its correlation coefficient with the number of effective intra-community edges in dentate gyrus neurogenesis (Fig. EV1d). ____The correlation coefficients are 0.91,0.96, 0.99 and 0.96 for community 0, 1, 2 and 3 separately.

* A bunch of notations are not clear:

4.1. What is the "r" in "strongest intercommunity interactions at r = 10 (Fig. 1F)"? Is it the same as the "r" mentioned in the Methods section? *

Reply: r____ is the index number of the discretized reaction coordinate. We added it when we define the reaction coordinate. We modified the conflict usage of r in Materials and Method 4.

4.2. What is "s_i" in "cell-specific effective matrix, Fbar_ij = (2*s_i - 1)*F_ij"? Also, that description of F_ij, f_ij, and H should be moved to the Methods section, and a more high-level, intuitive description should instead be included in this Results paragraph. Reply: represent the binarized gene expression state. is 0 for when gene is in low expression level (silence) and is 1 when gene is in high express level (active). We modified this part following your advice.

* How were the h_f and h_m thresholds chosen? *

Reply: and are based on the distribution of each dataset. Following suggestions from another reviewer, we modified this part. All the highly variable genes were selected and the genes were binarized with the Silverman’s bandwidth method and ____K____means (Materials and Methods 2).

* What is the "density of each single cell" ("⍴_t")? The formulation of the penalty of the distance between cells i and j (the expression with -logP_ij...) is unclear. What is the intuition behind it? What is r? How were the values of r (0.5 and 0.8) chosen? *

Reply: The probability density of cells in the expression space is based on the kernel density estimation. Intuitively, a region in the expression space with more cells is more likely passed by more cell trajectories. The values are based on the distribution of kernel density estimation in different datasets.

In the modified manuscript, we used trajectory simulation and deleted this assumption for simplification.

* One of the reasons the authors state to justify the choice of PLSR is "In the scRNA dataset, the number of genes is often comparable to or larger than the number of cells." This is not true most of the time. In nearly all recent studies, the number of cells is way larger than the number of genes measured. *

Reply: The PLSR method definitely can be used for the data whose number of cells is larger than the number of genes. Also the PLSR method was applied on cells that are the k nearest neighbors of each reaction coordinate, which are a subset of the whole dataset (Materials and Methods 5). While we mainly presented results with the PLSR method, in this revised manuscript we also added results with another method of GRISLI (Materials and Methods 9). The results are similar with what we obtained with PLSR.

* There is a fleeting reference to a nice previous finding that supports their observations: "several lines of evidence support that EMT proceeds through a concerted mechanism. Indeed, both in vivo and in vitro studies have identified intermediate states of EMT that have co-expressed epithelial and mesenchymal genes (Pastushenko et al, 2018; Zhang et al, 2014)". The authors should thoroughly survey the literature related to EMT transition, development of pancreatic endocrine cells, and development of the granule cell lineage in dentate gyrus, to find more previously identified molecular/cellular features relevant to cell state/type transitions, compared and contrasted with findings from this study. *

Reply: Thanks. We added references on these cell phenotype transitions and modified the corresponding part. We do want to point out that the main focus of this work is that all these processes share a common feature of transient increase of intercommunity interactions.

* What is the "dynamo" package, which is supposed to contain a Python notebook? As of now, the code and data have not been made available. Both need to be released along with thorough documentation on how to run the code to reproduce the analyses described here. *Reply: Thanks. Dynamo is a python package accompanying our recent publication ____(8)____. We uploaded the code on Github and added the link of Dynamo.

* **Minor comments and questions**

1. Replace "confliction" throughout the manuscript with "conflict" or "conflicting" as appropriate. *

Reply: Thanks. We modified them.

* Paragraph two of the Introduction (beginning with "Another example of transitions ...") is missing multiple references, esp. for the last four sentences. *

Reply: Thanks. We added references.

* There are direct quotes from previous papers like "predicts the future state of individual cells on a timescale of hours". The authors are highly encouraged to check for usage of exact phrasing using available text software such as iThenticate. *

Reply____: ____Thanks a lot for pointing out this severe mistake. We re-edited the manuscript and checked with iThenticate. *

*

• "Each community contains both E and M genes": what does this mean? *

Reply: The E (M) genes are defined as those genes that are active or have high expression levels in epithelial (mesenchymal) state or sample. As we reorganized the manuscript, we add this explanation for all datasets in the caption of Fig.1i.*

*

• Reference to Qui 2021 is missing in the "Path analysis" subsection under Methods. *

Reply: We added it in the Methods.

* Fix: "transition between the cells that their sample time points are successive" in Methods. *

Reply: Thanks. ____We modified it.

* In Methods, under "Network inference", it is "partial least square regression" (not *least* s square). *

Reply: Thanks. We modified it.

* Figure 1: The cyan, magenta, and lime in 1C are very hard to see and, perhaps, the grey of the points can be made lighter. Also, change the red and green colors for the arrows in 1I to something else. These colors are not colorblind-friendly. *

Reply: Thanks. We re-plotted the figures and changed the colormap.*

*

• Periods and commas are missing at several places. Reply: Thanks. We modify these and re-edit the manuscript.

Reviewer #2 (Significance (Required)):

The study uses RNA-velocity calculated from scRNA-seq data in an inventive way to characterize paths that reflect cell phenotype transitions. Then, a sparse gene regulatory network is reverse engineered from the data and the community structure within this network is examined at various stages along the transition to make observations about inter- and intra-community regulation and network "frustration". However, the study lacks the context of existing literature in terms of previous work studying cell transitions both experimentally and computationally. Adding this context (as suggested in the comments) will considerably improve the utility and significance of the findings. Overall, this study will be of broad interest to researchers interested in development and reprogramming as well as computational scientists developing and applying methods for scRNA-seq data analysis, trajectory inference, and network reconstruction. All the comments and questions raised here are based on my background and expertise in omics data (including scRNA-seq) analysis and network biology.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

The authors analyze three datasets of Single cell RNA velocity measured during phenotypic transition. They infer the gene regulatory network in each case and characterize the transition between the initial and final expression states (in which different sets of genes are expressed). Their motivating question was to find whether during such transitions first genes characterizing the initial state are no longer expressed and only then the genes associated with the final state start expressing or alternatively there is gradual transition through an intermediate state in which subsets of both initial and final state genes are transiently expressed.

They define a measure of regulatory frustration representing the mismatch between regulatory signals a gene receives and its current expression state. They conclude that phenotypic transitions involve transient interactions between otherwise non-interacting gene modules and a temporary increase of gene frustration, which is relaxed once the final expression state is reached.

The study uses of advanced inference and machine learning methods.

I find the question studied in this manuscript interesting, opening avenue to further questions and studies and relevant to different scientific communities. Personally I think that the focus of the paper should be the exposition of the methods used this manuscript would benefit from a longer format, but that depends of course on the journal they are aiming at. *

*

Statistical analysis is missing. Especially since the authors mention the potential of over-fitting due to large number of genes (on the order of the number of cells) - I think the authors should provide a sensitivity analysis testing how sensitive are the conclusions to the choice of cells or genes by applying the methods to subsets of the cells / genes. *

Reply: Thanks. For the subset of cells, we randomly selected cells from the dataset and performed the analyses (Fig. EV11a). For the subset of genes, we selected a subset of genes randomly and performed the analyses (Fig. EV 11b). We found the results are not affected. We also perform another statistical analysis by varying the value of resolution in community detection algorithm. And we found that the conclusion on variation of inter-community edges is not affected (Fig. EV12).

Figure EV11 Statistical analyses of dentate gyrus neurogenesis. Each dot represents a cell and color represents the number of inter-community edges.

(a) Frustration score along the RCs (left) and cell-specific variation of the number of inter-community edges (right) of a randomly selected sub-population of 2000 cells (from a total of 3184 cells);

(b) Frustration score along the RCs (left) and cell-specific variation of the number of inter-community edges) (right) of cells on the space of 400 randomly selected genes (from a total of 678 genes).

*What is the meaning of the distribution in the frustration plots? *

Reply: For each cell we calculated a frustration score. Therefore for cells in each Voronoi cell (which is a geometric cell, don’t be confused with the biological “cells”) along the reaction coordinate (Fig.1d, Fig. 2b &2g), we obtained a distribution of the frustration scores.*

In general, the conclusions are well-justified, but I think some statements in the discussion are inaccurate: "intercommunity interactions of a GRN are indeed minimized' - are they minimal or are they only lower at the stable states? There are two stable states - for which of them is intercommunity interaction lower? *

Reply: Thank. We agree with the reviewer and modified the writing. Comparing with the transition state, the number of intercommunity interactions is less for the stable states. ____The datasets' quality are not high enough for us to investigate whether ____"intercommunity interactions of a GRN are indeed minimized”.*

It is written in the discussion that 'for all three datasets frustration decreases with differentiation', but then Fig. 1g shows the opposite (final state is more frustrated than initial state). It is interesting to discuss the differences between the datasets analyzed in that respect and what could cause transition to a more frustrated state. I suggest that the authors also refer in the discussion to related questions and possible follow-up studies, such as: what determines the duration of the phenotypic transition? A relevant number is the switching time of a single gene. *

Reply: Good suggestion. Compared to other datasets, we found that the result of EMT shows larger variances. The relative difference of the frustration score is also affected by the GRN inference algorithm. For example, the difference between initial and final frustration scores of the pancreatic endocrinogenesis is more significant when using the GRISLI method (Figure EV6b). Given these, the trend that the frustration scores in the transition states transiently increase keep consistent.

Our conclusion is limited by the quality of the data. So we delete this part of discussion in the manuscript.

Qiu et al. have shown that splicing-based ____RNA velocities are relative, while metabolic-labeling-based RNA velocities are more quantitative and accurate____(8)____. We will re-analyze this problem if data with metabolic labeling becomes available.

* The authors mention at the end that the networks can often reach multiple final states from a common initial states. Do such transitions share some of their path (and in particular the intermediate frustrated state)? Given the intermediate connected state, it would be interesting to characterize the network stability to perturbations. *

Reply: This is a very important question. To reliably address these questions, we need higher quality data. We plan to characterize the network stability to perturbations in future studies, while in our recent paper using a full nonlinear modeling framework____(8)____, we performed in silico perturbations.

* While interesting, the manuscript itself is unfortunately hard to read and would benefit from major editing, including better exposition of the science and language editing. *

Reply: Thanks. We revised the manuscript extensively.*

Methods: Description of PCA and 'revised finite temperature string method' are missing in the Methods section. *

Reply:____ Thanks. PCA is used in RNA velocity analysis for dimension reduction. We added this in Materials and Methods 3. The revised string method is in Materials and Methods ____4.

*

Some examples:

Figure captions are very short and often non-informative. Some variables are not defined (or only defined later on) and the reader then needs to guess their meaning: it took me a while to understand what is 'r' in Fig. 1f and what 'r=10' (p. 4) means. *

Reply: Thanks. ____r____ represents the index number of reaction coordinates. We added this in the manuscript where we define reaction coordinates.*

p. 4: what are 'f' (as opposed to F) and 's_ij' and 's_j' (expression states?) Or is fs_ij one variable? What does a Hamiltonian of a cell mean (p. 4, bottom)? *

Reply: is the regulation of gene ____j on gene i, and is the expression state of gene i (0 for silence, and 1 for active expression). is the frustration value of regulation from gene j to gene i.

The pseudo Hamiltonian value is proposed in the literature as an analogy of ____the magnetic systems following the work of Boolean model in EMT ____(9)____. A high Hamiltonian value indicates that the cell is in an unstable state. In the original manuscript we included this quantity since it has been discussed in the literature. However we found it causes confusion and is not necessary for our discussions, so we removed the pseudo-Hamiltonian results in the revised manuscript. * P. 4: how are 'E and M genes' defined? *

Reply: The E (M) genes are defined as those genes that are active or have high expression levels at the epithelial (mesenchymal) state or sample. We explained our general strategy in the caption of Fig.1i . * What does 'network heterogeneity' (p. 5) mean? *

Reply: Network heterogeneity measures how homogenously the connections are distributed among the genes____(10)____. A high heterogeneity ____means that some genes have high degree of connectivity (the so-called hubs), while some have low degree of connectivity.

*

Fig. 1 is too tiny and hard to read and details are missing. *

Reply: Thanks. We modified this figure and caption.*

A glossary for all the acronyms used would be very helpful. *

Reply: Thanks. We added glossary in the manuscript.*

Language (some examples):

p. 5 bottom: Another system is on development... invitro -> in vitro

p. 6: 'measure on developmental potential' -> measure of... *

Reply: Thanks. We modified these and re-edited the whole manuscript.*

Reviewer #3 (Significance (Required)):

This study presents a methodological advance in demonstrating the application of data analysis methods to study developmental phenotypic transitions. High throughput measurements and computation power available today enable putting to test theoretical conjectures, as made by Waddington. I think this is a promising line of research, which could be used to further develop the computational methods as well as to further our understanding of developmental transitions and potentially develop associated mathematical modeling frameworks.

This study should be of interest to a diverse readership composed of developmental biologists as well as to quantitative biologists and CS researchers applying optimization techniques and data analysis methods to high-throughput biological data.

I am not an expert on the computational methods applied in this manuscript and hence cannot assess their correct use and statistical analysis.

*

1. Traag VA, Waltman L, & van Eck NJ (2019) From Louvain to Leiden: guaranteeing well-connected communities. Scientific Reports 9(1):5233.
2. Stuart T, et al. (2019) Comprehensive Integration of Single-Cell Data. Cell 177(7):1888-1902.e1821.
3. Bergen V, Lange M, Peidli S, Wolf FA, & Theis FJ (2020) Generalizing RNA velocity to transient cell states through dynamical modeling. Nature Biotechnology 38(12):1408-1414.
4. Wolf FA, Angerer P, & Theis FJ (2018) SCANPY: large-scale single-cell gene expression data analysis. Genome Biology 19(1):15.
5. Aubin-Frankowski P-C & Vert J-P (2020) Gene regulation inference from single-cell RNA-seq data with linear differential equations and velocity inference. Bioinformatics (Oxford, England) 36(18):4774-4780.
6. Moris N, Pina C, & Arias AM (2016) Transition states and cell fate decisions in epigenetic landscapes. Nature reviews. Genetics 17(11):693-703.
7. Kalkan T, et al. (2017) Tracking the embryonic stem cell transition from ground state pluripotency. Development 144(7):1221-1234.
8. Qiu X, et al. (2022) Mapping Transcriptomic Vector Fields of Single Cells. Cell 185(4):690-711.
9. Font-Clos F, Zapperi S, & La Porta CAM (2018) Topography of epithelial–mesenchymal plasticity. Proceedings of the National Academy of Sciences 115(23):5902-5907.
10. Gao J, Barzel B, & Barabási A-L (2016) Universal resilience patterns in complex networks. Nature 530(7590):307-312.

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

We are very grateful to the three referees for their constructive comments and suggestions which have helped improve the quality of our manuscript.

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

In the publication HAT-field: a very cheap, robust and quantitative point-of-care serological test for Covid-19 by Joly and Ribes the authors describe an adaption and an improved protocol to their previously published haemagglutination based test to detect antibodies to SARS-CoV-2 in patient blood (Towsend et al., 2021). In detail, they analyzed the effect of several adaptions including buffer optimization, plate coating, usage of patient whole blood instead of washed RBCs and plasma. Additionally they tested different temperatures and stability of the reagents, namely the nanobody-RBD construct IH4-RBD. For validation they compared their optimized HAT-field assay with Jurkat-S&R as a FACS-based assay.

Major comments:

Introduction: This section is rather short and could benefit from a broader overview of currently established methods and assays to detect appropriate immune responses against SARS-CoV-2. The author are advised to summarize the current literature in the field more comprehensively and not only focus on their own work.

Response: Hundreds of different tests to monitor immune responses against SARS-CoV-2 have been described to date, and the literature on these various tests is vast, with new articles coming out almost on a daily basis. We would not feel either that the introduction of our rather technical paper would benefit from being lengthened by such a review of the current literature, or even competent to carry out such a summary. Following the referee’s suggestion, we have, however, introduced a new sentence and given three references providing relatively recent overviews on the subject of immune-monitoring.

Cross-reactivity with IH4-RBD. In Figure 6, the authors highlight the samples in red and orange that showed cross-reactivity with IH4-RBD. In their discussion, however, the authors state that only 2 of 60 (3%) were cross-reactive. In making this statement, they ignore the proportion of cross-reactive samples that were also positive in the Jurkat S&R assay. Therefore, the authors should acknowledge in the discussion that the actual number of cross-reactive samples was higher.

Response*: The statement in the discussion about 2 cross reactive samples out of 60 concerns the results obtained after an incubation of one hour under normal gravity, and not the two red dots in each of the three graphs of figure 6, which correspond to the two negative samples which gave false-positive results in HAT plasma titrations after spinning (Figure 6C), for which we correctly state in the discussion that 12 samples showed cross-reactivity on IH4 alone. The data presented in Figure 6B corresponds to HAT-field after spinning, for which we correctly state in the discussion that 5 out of 60 showed cross-reactivity (4 orange dots + 1 red dot, the second red dot having a score of 0, in accordance with the fact that this sample showed no cross reaction on IH4 alone in HAT-field after spinning). *

*To try to prevent this possible confusion, we have now clarified what data we are referring to at the start of that paragraph in the discussion. *

Quantitative Assay. Since the HAT assay does not allow determination of the absolute number of antibodies reactive to SARS-CoV-2 in the blood samples, the authors should refrain from claiming that the HAT-field is a quantitative assay.

Response*: Since immune sera are inherently polyclonal, they contain a multitude of different types of antibodies of different affinities and avidities, and we are not aware of any technique that allows to determine the “absolute number” of antibodies directed against a given antigen in such samples. *

*For many serological tests, including ELISA and the initial protocol of HAT, serum or plasma titrations are used as a means to obtain what is widely considered as a quantitative evaluation of the amounts of antibodies in blood samples. Even FACS-based assays such as the Jurkat-S&R-flow test we have used, are commonly considered as quantitative but those only provide relative results and not absolute numbers. *

We perceive that the close correlations we find between the results of the HAT-field protocol and those of the Jurkat-S&R-flow test as well as with serum titrations using the standard HAT protocol warrants considering the results of HAT-field as being as quantitative as those obtained with all those other tests.

Morphological read out For field application, the morphological description of the observed deposits ("teardrop" vs. "button") could be problematic and might lead to bias depending on the user. Thus, the authors should provide a clearer description for phenotype classification.

Response: We have now introduced a specific paragraph detailing how to score HAT assays in the Methods section, as well as a new figure providing a graphic description of positive, partial and negative RBCs deposits.

Minor comments: Title: the authors should remove "very"

Response*: We have now removed the word ‘very’ from the title, and thank the referee for this helpful suggestion. *

By the way: What are the costs of IH4-RBD for a 96 well plate? Who will produce this reagent? Is the sequence of the IH4 fully disclosed?

Response*: As specified in our original paper (see Townsend et al. 2021), the plasmid coding for the IH4-RBD is available upon request from Alain Townsend (Oxford, UK). Furthermore, his laboratory funded the production of 1 gram of the IH4-RBD reagent by a commercial company, and professor Townsend has been graciously sending aliquots of 1 mg of this reagent, which suffice for several thousand tests, to all the laboratories that have requested it from him. *

*In its initial format, HAT only required 100 ng of IH4-RBD per well, corresponding to a cost of 0.0027 £ per well. For the HAT-field protocol, 5 times more reagent is needed, thus bringing the cost of the reagent to 1.5 cts per test, to which one would have to add a similar cost for the IH4-reagent alone. This would thus bring the cost of the two reagents to approximately 3 cts, which is still lower than the price of any of the cheap disposable plasticware necessary for the test (lancet, pipet, plastic tube and portion of a plate). *

The sequence of the IH4 nanobody is indeed fully disclosed (see figure 1 of Townsend et al. 2021), and has actually been protected by a patent ( US9879090B2 ). Whilst IH4 can be used freely for research purposes, licensing rights would have to be taken into consideration by any health authority wishing to use the technique broadly, or for any commercial distribution.

The usage of the CR3022 as positive control for neutralizing antibodies should be reconsidered since this antibody does not confer viral neutralization. Other well describe antibodies blocking the ACE2:RBD interface might be better suited.

Response*: CR3022 was the one that we had at our disposal, but other mAbs can certainly be used instead of as positive controls, and this is actually indicated in the detailed HAT-field protocol provided. Since the use of a positive control is only to ensure that the IH4-RBD has not been degraded and works as well as expected, and that any negative samples are not due to a very rare glycophorin mutation that could prevent IH4 from binding to it at the surface of RBCs, we are not sure why using a mAb with neutralizing activity would necessarily be better than the CR3022 mAb. *

Figure 2: Please state the concentration of IH4-RBD used. As stated in the figure legends for Figure 2 B, the authors should show the result all 4 replicates (incl. SD)

Response: The concentration of IH4-RBD was 1 m*g/ml, i.e. the normal concentration for standard HAT tests. This was already indicated in the Methods section, but has now been added to the legend of Figure 2. *

Whilst 4 experiments were indeed carried out, which all gave similar results, i.e. showed that using PBS-N3 or PBN did not hinder HAT performance, but could instead result in a slight increase in HAT sensitivity, those various experiments were not all exact replicates of the experiment shown on figure 2. Furthermore, performing of those various experiments was spread over a period of over a year, using different reagents, thus precluding numerical comparisons between the various results. We have clarified this issue by rewording the final statement to “Comparable results were obtained in four similar experiments.”

Figure 3: Although the authors showed stability of IH4-RBD at 2 µg/ml they do not provide data for the stabilities at higher dilutions. As the authors suggest to predistribute the IH4-RBD in plates they should at least discuss this issue.

We thank the referee for raising this valid point, which has now been discussed in the paragraph entitled “Practical considerations for performing HAT assays” in the Methods section: “One aspect that will have to be considered for the design and use of such individual strips of wells will be to ensure that, upon storage, the various dilutions of IH4-RBD are as stable in such strips as the working stocks of IH4-RBD (2 mg/ml) tested in Figure 3.”

Figure 6/Supplementary Figure 1 and 3 The presentation of the data is not accurate, as many of the points (samples) are obviously identically positioned in the graph. The authors should choose a different representation of their data. E.g. they could adjust the size of the points to the number of overlapping samples.

Response: We thank the referee for raising this issue, which was also pointed to by referee #2. This apparent inaccuracy is due to the fact that, on these plots, the scales for both x and Y axes used discrete values, which indeed results in multiple points overlapping on top of one another. This was resolved by adding numbers next to the positions where several dots overlapped

Wording / text length In the current manuscript the text is very long. Thus, the authors should shorten it to report the essential findings more appropriately. Additionally they should check for correct English wording.

Response*: We thank the referee for this remark, which helped us realize that the excessive length of the manuscript was mostly due to an extensive discussion of highly technical and practical points. The corresponding paragraphs were indeed out of place in the general discussion, and have not been deleted but have been moved to the Methods section since we feel that they contain very important information for people who would actually start to performing HAT assays. *

Reviewer #1 (Significance (Required)):

In summary, the authors describe the HAT-field test as a simple PoC test for the detection of SARS-CoV-2 antibodies in patients. Because of its ease of use and robustness, the test appears to be particularly well suited for use in countries with underdeveloped health care or limited testing facilities, as also reported previously. The value of this manuscript lies mainly in the detailed description of the protocol and its validation. In this context, the adaptations described are certainly useful and helpful from a practical point of view, but do not provide significant new scientific insights. In light of these considerations, we recommend that this work be submitted to an appropriate journal specializing in the publication of such methods

Expertise The reviewers have established and published different serological assays to monitor immune responses against SARS-CoV-2

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

In this paper, the authors developed a feasible protocol for an affordable point-of-care serological test for SARS-CoV-2. This method was adapted from the HAT plasma titration test that the authors previously published. Specifically, the test utilizes a 96-well plate pre-coated with the RBD of SARS-CoV-2 spike glycoprotein fusing to a red blood cell targeting nanobody (IH4). By adding microliters amount of the blood or plasma samples to the plate, it allows the detection of antibodies against RBD by measuring the level of hemagglutination. In the current upgraded protocol (so called HAT-field), the authors made major modifications including optimizations of buffer and experimental protocol and the use of pre-titrated IH4-RBD on the plate, which collectively helped to lower the sample consumptions, improved the stability and the sensitivity of detection, and made the test more user-friendly under non-clinical settings.

Major comments: My major concerns are related to the robustness and quantitative capability of this approach. Specifically: It seems that multiple variables may impact the results. These include volume of droplets, the presence/absence of serum IH4 or BSA cross-reactive antibodies, and the amount (%) of red blood cells which may vary substantially among samples. Could you find a way to normalize the results (e.g., the discrepancy shown in Figure 6) instead of only leaving them as false-positives or false-negative?

Response*: Regarding the volume of the droplets, in other words, the amount of blood collected and used in an assay, two sentences in the manuscript underline the fact that this is not a critical variable: *

In the Results section “the precise volume of blood collected is not critical; it may vary by as much as 30% with no detectable influence on the results.”

In the discussion: “On this subject, we have found that increasing the amount of whole blood per well (in other words using blood that is less dilute) has very little influence over the HAT-field results, and, if anything, adding more blood can sometimes reduce the sensitivity, albeit never by more than 1 dilution.”

Consequently the % of RBCs in samples seem unlikely to influence the HAT-field scores significantly. This is supported by the fact that, although men tend to have higher hematocrits than women, we have not noticed any detectable difference between men and women in the correlation of the HAT-field scores with those of the Jurkat-S&R-flow test.

We are not sure that we fully understand what discrepancy shown in Figure 6 the referee is pointing to, but if it is about the increase in the number of samples found to be cross reacting on IH4 alone when the sensitivity increases, in the discussion, we propose to perform tests using titrations of the IH4 nanobody alone simultaneously to using the IH4-RBD reagent, so as to minimize the number of samples that would be identified as false positives if only one concentration of IH4 alone was used as negative control. Comparing the titers obtained with IH4-RBD and IH4 alone will then provide some level of normalization for the samples cross reacting on IH4. As for the hypothetical presence of antibodies cross reacting on BSA alluded to by the referee, since such antibodies would not bind to RBCs, we do not think they would affect the HAT results.

Second, the score of the HAT-field ranges from 0 - 8. However, based on the current manuscript, it is not clear how the scoring and scaling works. How is the noise (non-specific antibody signal) defined here?

Response: We have now introduced a specific paragraph and a new figure detailing how to score HAT assays in the Methods section.

In addition, it is unclear how to translate the HAT-field score into a meaningful measure of protection by serum antibodies.

Response*: Documenting the correlation between HAT-field scores and levels of protection against SARS-CoV-2 infections and/or Covid-19 severity would indeed be extremely interesting. This would, however, require setting up a large scale clinical trial carried out over several months. This type of work could only be carried out by a large consortium including clinicians or even preferably a national health agency. This was, however, far beyond the reach of this initial project, which was based on the work of a single person on a shoestring budget. *

Can you provide more evidence to demonstrate that the test is quantitative? For example, performing additional orthogonal experiments to better validate the scoring and generate a correlation function?

Response*: Inasmuch as it would have been very interesting to perform additional serological tests from commercial sources on the samples of our cohort, such tests are all very expensive (e.g. ca. 500 € for one ELISA plate). This was in fact the main reason for developing the Jurkat-S&R-flow test in the first place, since it is much cheaper, more modular, and at least as sensitive as ELISA (see Maurel Ribes et al. 2021). The funds for this whole project came from a single 15 k€ grant obtained from the ANR, and we simply did not have access to the funds, or to the human resources to carry out such experiments based on commercial serological tests. *

Minor comments: Figure 6: are all results included? To me, it does not seem that all 60 samples data were included in the plot.

Response: We thank the referee for raising this issue, which was also pointed to by referee #1. This apparent inaccuracy is due to the fact that the scales for both x and Y axes used discrete values, which results in multiple points overlapping on top of one another. This was resolved by adding numbers next to the positions where several dots overlapped.

There are several redundant statements in the discussion and results section. Please make the text more concise.

Response: The discussion has now been shortened considerably, mostly by moving the paragraphs pertaining to technical considerations to the Methods section.

Reviewer #2 (Significance (Required)):

The current paper is built upon the improvement of previous published work. In addition, there are similar approaches that have been published. It was unclear if the current method is superior to other works.

Response: Whilst we have made no statement regarding whether the method we describe is superior to other methods, we are pretty confident that very few alternatives will be as frugal and simple as the HAT-field protocol described here. As alluded to in the final paragraph of the discussion, two recent reports have described that HAT could be performed on cards rather than in V-shaped wells, with semi-quantitative results being obtained in minutes. If such card-based approaches turn out to provide sensitivity and reliability comparable to those of the HAT-field protocol, they will certainly represent very interesting alternatives. As stated in our manuscript, we would be very interested if the comparative evaluation of the two approaches could be carried out by one or several independent third party.

My research involves the development of antiviral antibody therapeutics. This method may be used as a point-of-care tool for the measurement of serologic response to RBD in less developed countries. However, due to the high vaccination rate and large infected populations, the overall needs for such detection drastically decrease. The significance of the work and utilities of the test may expand with more experiments related to the variants.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

This paper describes a low-cost robust and quantitative serological test based on haemgglutination, which could be used in resource limited settings for evaluating population-based and vaccine induced immunity. Neutralising antibodies to the receptor binding domain (RBD) on the SARS CoV-2 spike protein are an immunological correlate of protection. The HAT has a single reagent the RBD domain of SARS CoV-2 linked to a monomeric anti-erythrocyte single domain nanobody. When human polyclonal serum antibodies bind to the RBD they cross-link and agglutinate human red blood cells, resulting in haemagglutination which can be read visually.

This paper thoroughly evaluate the stability of the HAT reagents used to measure human and monoclonal antibodies examining the robustness of the HAT reagent. It provides a comprehensive protocol for conducting field based HAT with limited reagents. The test can evaluate is subjects have been infected using a simple finger prick to detect RBD specific antibodies. The field HAT can also be used to define people that can be susceptible to reinfection or in need of vaccination, With the use of RBDs from the variants of concern the test can be rapidly adapted to evaluate antibodies as new variants arise to evaluate surrogate correlates of protection to allow timely evaluation of vaccine effectiveness and predict the need for vaccine booster doses. The data are very comprehensively presented with good figures demonstrating the most appropriate buffer to store the IH4-RBD reagent and the robustness of the HAT over time at different temperatures. No additional experiments are needed and suitable numbers of replicates are included. All data, methods and reagents are comprehensively described.

Minor comments: The paper is well written but rather long in places and may have benefited from being more succinct.

Response: The excessive length of the manuscript was mostly due to an extensive discussion of highly technical and practical points. The discussion has now been shortened considerably, mostly by moving the paragraphs pertaining to technical considerations to the Methods section.

Panels in figures could be labelled as A, B, C etc to help in identifying the correct panel..

Response: We thank the referee for this helpful suggestion, which we have followed.

I would avoid the use of experiment and project and refer to next we confirmed... or in this paper or our results show Please make sure all abbreviation are defined upon first use. Perhaps include early in the paper that most of the work was conducted with the Wuhan RBD

Response: We thank the referee for these helpful suggestions, which we have followed to the best of our abilities. The abstract now contains a mention of the fact the work on optimizing the protocol was carried out with the IH4-RBD carrying the Wuhan version.

Figure 2: I would suggest placing either a solid line between the two halves of the plates to make it easier for the reader to differentiate between the two antibodies. It also would have been easier to read if the bottom PBS, PBS-N3 and PBN were at 45 degree angle. In B include the serum name (e.g. serum 197).

Response: We thank the referee for these helpful suggestions, which we have followed.

Legend to figure 4: please include the serum numbers after covid-19 patients. Perhaps include arrows to demonstrate the dilutions of serum and IH4-RBD in the figure.

Page 6 it might be easiest to use the same times as in figure 6 and use for example more than one year in the discussion

Response: We thank the referee for these helpful suggestions, which we have all followed.

Legend figure 6 perhaps replace dots with circles page 10 include the R values from figure 6 in the description of results.

Response: We are grateful to the referee for these helpful suggestions, but have not followed them since we do not feel that these changes would be real improvements.

Page 12 of note perhaps this can be moved to the methods ?

Response: This, and several other paragraphs of the Discussion, have now been moved to the Methods section.

Supplementary figure 2 A can be seen, is something missing here?

Response: An s was indeed missing : “A can be seen” corrected to “As can be seen “

*

Reviewer #3 (Significance (Required)):

This paper describes a simple rapid field test for evaluating antibodies to the receptor binding domain of the spike of SARS CoV-2 using the Wuhan and delta variant. Whilst high income countries can provide booster doses and extensive testing (either lateral flow or RT-PCR based) and contact racing to control the waves of the pandemic, low income countries have had limited access to Covid vaccine and the extent of previous waves of the pandemic in the populations are unknown.

This paper describes a robust and simple test for investigating human antibodies to SARS-CoV-2 which could be performed in resource limited settings providing a very useful tool for monitoring infection in the community and potentially for prioritising this scarce COVID-19 vaccines available.

This study builds upon the work conducted on the HAT and has extensively studied and optimised the test so that it could be used globally. This paper provides a comprehensive protocol and has simplified the test to ensure it could be used in LMICs.

This paper would be of great interest to a wide scientific audience who are interested in a rapid low-cost test to evaluate population based and vaccine induced immunity.

Reviewer: serological assays for use in virology and vaccinology. Suitable competence to review the whole paper *

Author Response:

Evaluation Summary

Challa and Ryu et al. systematically evaluated various combinations of ADP-ribose-binding modules to make sensors detecting poly(ADP-ribose). They developed and tested two indicator designs optimized for analyses in cell culture (dimerization-dependent GFP-based) or intact tissues (split Nano luciferase-based). Overall, with further experimental controls and quantification, this timely set of cell biology probes will be useful to study the biological functions of ADP-ribosylation in cultured cells and whole organisms.

We appreciate the positive and encouraging words from the reviewer. We also appreciate the helpful comments, criticisms, and suggestions, which we have endeavored to address fully.

Reviewer 1 (Public Review):

While these tools are more sensitive than existing tools, it is unclear whether a dynamic range of 6-fold (GFP) and 3-fold (luciferase) provide sufficient sensitivity for properly understanding the PAR dynamics (which was thought to increase as much as 100-fold in DNA damage settings). In addition, it is unclear whether the fold increases in both fluorescence and luminescence linearly correlate with the traditional measures by western blot.

We are pleased that the reviewer found our sensors to potentially useful. The reviewer provided a number of excellent comments and suggestions that have served as a useful guide for improving our paper. We have carefully considered all of the comments, insights, and suggestions from the reviewer and revised the manuscript accordingly. We think this has strengthened our conclusions and improved the paper considerably. We thank the reviewer for the careful and thorough review of our paper.

Figure 1F indicates on the western blot that there was a precipitous drop of PARylation after 5 min, but the GFP signal indicated a linear drop. It will be important to quantify the signals on western blots and test how correlate their data with the GFP/luciferase data in scatter plots for their various sets of data. Would this system under-estimate the changes and be not sensitive enough to subtle changes that may be 1-2 fold measured by traditional means

We agree with the reviewer that a comparison with existing PAR detection technologies will improve the manuscript. We now performed a comparative analysis of ELISA, Western blot, and immunofluorescence assays with live cell imaging using PAR-T GFP (Figures 6A, 6C, 6D). The results indicate that the detection range of PAR-T ddGFP is comparable to the established PAR detection assays. In addition, we also compared the live cell luciferase assays using PAR-T NanoLuc to Western blotting (Figure 6B) and found that these two assays are able to detect PAR changes at comparable levels. We would also like to emphasize that these sensors were developed to improve our ability to detect PAR changes in living cells and animals, which the existing techniques are not capable of doing.

Similarly, how is their quantitation in Figure 2 compared with traditional immunofluorescence?

We performed this comparison and observed that the changes in PAR levels as detected by live cell imaging using PAR-T ddGFP are comparable to the changes detected in immunofluorescence assays using the WWE-Fc reagent (Figure 6D and 6E).

Lastly, for the luciferase signal in Figure 3B and C, the corresponding signal in western blots are missing. Therefore, it is difficult to estimate the background signal. If Niraparib, as in other figures, eliminates PAR signals on western blot, these data would indicate half of the basal signal are background, which is rather high. Having said that, tool development is an evolution process. These tools will provide a good foundation for future development. Therefore, understanding these limitations (dynamic range, quantitative sensitivity correlation, and background) will provide a better assessment of the utility of these new tools for investigating PAR biology.

We appreciate the reviewer’s concern about the high background signal in Niraparibtreated samples. To answer this concern, we compared the dynamic range of PAR-T NanoLuc to Western blotting (Figure 6B) and found that the results from live cell luciferase assays using PAR-T NanoLuc are comparable to Western blotting using WWE-Fc. Of note, we were able to detect decreases in PAR levels with Niraparib using live cell luciferase assays using PAR-T NanoLuc, but not Western blotting. Based on these analyses, we can conclude that the changes in PAR levels at the basal level are very minimal, leading to only 50% decrease in PAR-T NanoLuc signal with Niraparib treatment (Figure 6B, Figure 5A-5C). Note that the decrease in PAR-T NanoLuc signal is greater when UV-treated cells were pre-treated with Niraparib, which is consistent with the results from Western blot analysis (Figure 5A).

Reviewer 2 (Public Review):

In this study, the authors attempted to extend their own work and that of others in the field in developing probes to detect the signaling molecule, poly-ADPribose (PAR) that can be used in the test tube, in cells and in tumor models. Major strengths include the development of a set of probes with data demonstrating utility and efficacy. Further, the authors show the assay to be useful in cell models and tumor models. Some weaknesses include what appears to be a high level of background in the assay. Further, regarding methods, the exact probes (sequences) being evaluated are not defined. This is one of several new PAR probes being developed over the last few years but may have widespread utility due to the quantitative nature of the bioluminescent assay.

We thank the reviewer for these thoughtful and encouraging comments, as well as the interesting, thought-provoking, and constructive criticisms that have prompted us to dig deeper and provide more evidence to support our claims

Reviewer 3 (Public review):

The major drawback is that, while the authors demonstrated some applications of these PAR trackers (PAR-T) in both culture cells and in animals, the data of PAR-T ddGFP on cancer cells and the data of PAR-T Nano luciferase may not be sufficient to support the authors' claim that the new tool can detect spatial and temporal dynamics of PAR in cells and in animals. That said, the new tools can potentially expand the capability of cell biologists to visualize and study the PAR production process in both normal and disease states with improved sensitivity and tissue compatibility.

We thank the reviewer for appreciating the potential utility of the PAR-T sensors, as well as the detailed and constructive criticisms that have prompted us to provide more evidence to support our claims. Addressing these comments has helped us to improve the paper.

One of the major issues of this manuscript is the lack of time-course data for PAR-T luminescent sensors to demonstrate temporal monitoring of PAR levels in animals. If the binding of two split Nano Luciferase parts is irreversible, the application might be limited. However, according to the literature (Scientific Reports volume 11, Article number: 12535 (2021)), the split Nanoluc technology should be able to detect dynamic changes. Either way, a set of time-course data would be necessary. The authors need to provide evidence to support their statement "The high sensitivity and low signal to noise ratios of the PAR-Trackers described here enable spatial and temporal monitoring of PAR levels in cells and in animals.

We agree with the reviewer’s comment that the original manuscript did not demonstrate that the PAR-T sensors can be used to detect spatio-temporal changes in PAR. To demonstrate that PAR-T NanoLuc can be used to detect time-dependent changes in PAR levels, we performed a time course of UV-mediated PARP-1 activation (Figure 5D). The results from this assay demonstrated that the dynamic changes in PAR in live cells, in response to DNA damage, can be recaptured using the PAR-T NanoLuc sensors. In addition, we also measured PARGi-mediated PAR accumulation in vivo in xenograft tumors (Figure 8 - figure supplement 1B-1D). We found that PAR can be detected readily in breast cancer cells when injected into mice. Upon treatment with PARGi, the luminescence from PAR-T NanoLuc increased significantly by 6 hours and then diminished by 24 hours. These data demonstrate that PAR-T NanoLuc can be used to track dynamic changes in PAR levels both in cells and in animals. While not in vivo, our work with spheroids also addresses this concern. See our response to the next comment below.

Figure 2- figure supplement 2. For the detection of spatial dynamics of PAR signals in cancer spheroids, the authors did not provide sufficient evidence as only static images of different spheroids in different conditions were provided. And 2 out of 3 fields of view only include one spheroid. In addition, there is no time-course image data showing the spatial patterns of PAR in cancer cells are dynamic.

We have now performed a quantitative analysis of multiple spheroids. As indicated in Figure 3B, we observed a significantly higher GFP fluorescence signal in spheroids derived Challa et al. (Kraus) – Rebuttal February 2, 2022 10 from PAR-T ddGFP expressing cells compared to those expressing ddGFP or those treated with Niraparib. To address the reviewer’s concern about using PAR-T ddGFP for spatio-temporal changes in cells, we included a video for live cell imaging of H2O2-mediated increase in PAR-T ddGFP (Figure 2 - figure supplement 2, video). We also developed an analysis approach that allows us to quantify the signals from the core of the spheroids separately from the periphery of the spheroids. We also performed a time course in 3D cancer spheroids to visualize the spatio-temporal changes in PAR levels (Figure 3C and 3D). The results from this experiment demonstrate that the PAR levels in cells at the core of the spheroids are relatively resistant to Niraparib treatment, as the PAR levels in cells at the core of the spheroid decrease at a lower rate when compared to PAR in the cells at the outer layer of the spheroid.

In the caption of Figure 2 -figure supplement 1 (B and C), it states "Immunofluorescence assay to track PAR formation in response to H2O2.", but there is no evidence showing any antibodies were used there.

We thank the reviewer for pointing out this error. It should have been written as live cell imaging, not immunofluorescence assay. We made this correction.

It seems that Figure 3 B and C does not support the statement "we observed specific detection of firefly luciferase with D-Luciferin and NanoLuc with furimazine with no cross-reactivity" And it is unclear why the authors refer Fig. 3B and C after that statement as those data seems not supporting this claim. Similarly, the statement "Moreover, the luminescence of PAR-T Luc is only 30-fold lower than intact firefly luciferase." Was not supported by Fig. 3B. In fact, the differences between PAR-T Luc and intact firefly luciferase were ~1000 fold in vivo, judging from Fig 5B. It is also unclear which data of the construct was used to plot Fig. 3C.

We thank the reviewer for this comment. We changed the scale bar to represent the true scale for the luminescence from Nano luciferase and Firefly luciferase. This indicates that the brightness of PAR-T NanoLuc is 30-fold lower than intact firefly luciferase. In Figure 3C, we plotted the ratio of PAR-T NanoLuc to firefly luciferase.

Fig. 4C, it seems that Firefly luciferase was consistently brighter with PARGi, and I wonder if such difference is statistically significant. The authors did not perform a twoway ANOVA test for the firefly luciferase dataset.

We included the statistics to indicate that these changes were not significant.

The statement "Moreover, none of these sensors can detect PAR accumulation in vivo." seems to lack support. Have the authors proved that with evidence? I would recommend using the following statement instead: "Moreover, none of these sensors has yet demonstrated detection of PAR accumulation in vivo

We made this change.

For the in vivo experiment, it is unclear about the benefits of normalizing the PAR-T radiance to the Firefly luciferase since the signals from Firefly luciferase did not overlap well with that from the PAR-T nano luciferase, which may cause bigger variations.

We thank the reviewer for raising this point. We normalize the luminescence from PAR-T NanoLuc to that from firefly luciferase to account for the variability in tumor size between the mice. We think this is an important control in the analysis. The luminescence from firefly luciferase represents the differences in tumor size between the mice. Hence, that signal is greater than the signal from PAR-T NanoLuc and is spread over a larger area.

Judging from the data of Fig 3 supplement 1E, the signal intensity from the split firefly luciferase-based PAR-T sensors was ~10000 fold less than intact firefly luciferase, not ~1000 fold. It makes more sense to give up the split firefly luciferase for ~10000 fold differences since the signal intensity from the split nano luciferase was ~1000 fold less than intact firefly luciferase (Fig 5B).

We noted the reviewers concern about the split firefly luciferase PAR-T. We agree with the reviewer that the split nano luciferase is brighter than the split firefly luciferase (Figure 4C and Figure 4 - figure supplement 1E). Although split nano luciferase is 1000-fold dimmer than the intact firefly luciferase in vivo (Figure 8B and Figure 8 - figure supplement 1A), this difference is only 30-fold in in vitro assays (Figure 4C). Hence, the comparison of sensors based on split firefly luciferase to split nano luciferase highlights our efforts to make a brighter sensor. Moreover, we included the split firefly luciferase data to compare the performance of WWE vs macrodomain in the development of the PAR-T NanoLuc sensor. Since firefly luciferase is frequently used for sensor development, we believe that it is important to include the results obtained from this sensor.

Therefore, developing tools to measure ADPR dynamics in cells and in vivo is critical for better understating the various biological processes mediated by ADPR". "understating" should be "understanding".

We corrected this error.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary:

Estrach and colleagues seek to identify the ECM components that are key to regulating hair follicle stem cell (HFSC) activation using the highly-characterized mouse hair follicle as a model. They first use a targeted approach to examine key ECM components expressed by HFSC and find that Fibronectin (FN) is highly expressed. Further, wholemount analysis of the hair follicle reveals a meshwork of FN enveloping the hair follicle. They hypothesize that FN is a fundamental regulator of hair follicle (HF) cycling and then proceed to carry out longterm studies required to examine hair follicle cycling and knockout FN with two different HFSC Cre lines (Lrig1 and Krt19), as well as integrin coreceptor SLC3A2. They clearly show that absence of Fibronectin (FN) and SLC3A2 is detrimental to hair follicle stem cell activation and cycling (FN) and hair follicle identity (SLC3A2).

Overall comments:

The authors use the tail hair follicles as a model similarly to the highly-characterized, synchronous back skin hair follicles. However, the tail hair follicles are asynchronous (Braun et al. 2003, PMID: 12954714), thus reporting the age of the mouse from which the tail whole mounts came from is not sufficient to claim a HF cycle disorder - HF should be imaged in an unbiased manner and subsequently quantified for phase. The manuscript would greatly benefit from including more information in the figure legends, such as age of mice, number of mice and HF quantified, as well as what the error bars represent. Further, in samples where many HF were counted per mouse, these should be averaged and then the average per mouse displayed; super plots would be great to use here.

Major comments:

1. In Figure 1, the use of tail whole mount images indeed provides striking display of the fibronectin meshwork that envelops the hair follicle. However, addition of a marker of the regenerative phase (e.g. proliferation) and resting phase would provide more convincing evidence that this is the particular phase of the hair cycle that you have captured, especially given my overall comment regarding the asynchronous nature of the tail HF cycle.
2. The authors show that FN is expressed in early-mid anagen and conclude that FN is a regenerative signal. This claim should be substantiated with FN staining on more time points across the HF cycle to substantiate the argument that it is a regeneration-specific signal, found only in the telogen-anagen transition.
3. Lrig1-cre and K19-cre-mediated FN knockout result in HF that are thinner at D158 - this is not immediately apparent from histological sections. Can you use your thick sections to give better perspective?
4. The authors measure the width of the infundibulum from lightsheet microscope images. It is a bit difficult to position whole tissues using this technique, and the images that are shown are not from the same perspective, and thus measurement of the width is not accurate from these images. I suggest either removing this analysis or using more comparable images. Further, if this is a true phenotype, can you speculate on what the thickened infundibulum might mean?
5. The authors then show mislocalization of Lrig1+ cells to the infundibulum in absence of FN. Are other stem markers localized to the infundibulum or outside of the bulge? Further, what might the mislocalization of Lrig1+ cells might mean?
6. Please explain your conclusion after Figure 3i and at the end of the manuscript that states that FN is required for stem cell anchorage. I think that a very plausible explanation is that FN is required for stem cell function and identity, but anchorage of the SC lacks sufficient evidence. Further, your only evidence to support the anchoring theory only comes from expression of Lrig1 in FN knockout and no other markers. Are they also mislocalized? Please either tone down this conclusion on SC anchorage or provide stainings for more SC markers to show mislocalization in absence of FN.
7. In Figure 3l-o, you examine proliferation on the control vs the conditional deletion of FN in D30 and D158. However, in D30, these tissues are not at all directly comparable since one is obviously in anagen and the knockout in telogen. You must compare the anagen knockout sample, although this occurs a bit later than the control. Further, how was the infundibulum distinguished from the bulb in these control images?
8. In Figure 3P, you carry out RT-qPCR on whole skin to detect HFSC markers. This should have been carried out on sorted epithelial cells as isolation of whole back skin introduces bias to the system in that the number of stem cells may artificially look different in skin that is in anagen vs skin that is in telogen as the anagen skin has a different proportion of SC to progenitor cells to dermal cells. This concern is also similar to point 9 - the control and FN knockout at D30 are not comparable given that they are in different phases of the hair cycle.
9. Figure 4a these images need to be of the whole mouse - it is not possible to determine what we are looking at or where - there is not even a scale bar.
10. After Figure 4, you argue that because fibronectin expression resolves from healing dermis is the reason that hair follicles do not form, and site Dekonick and Blanpain (PMID: 30602767) - however this review makes no mention of the dynamics of fibronectin in wound healing. Further, evidence from Driskell et al (2013, PMID: 30602767) would suggest that it is the fibroblast population that responds to the wound that determines whether HF regenerate. And further, very large wounds do regenerate HF (Ito et al PMID: 30602767). In addition, this would all be fibroblast-derived FN, as opposed to the current study which examines keratinocyte-derived FN. Please reconsider this argument.
11. The authors knockout SLC3A2, an integrin coreceptor that is localized to the plasma membrane. They show a very similar, yet more severe phenotype to the Lrig1- and K19- mediated knockout of FN. Given the bidirectional communication that SLC3A2 is responsible for, can you reconcile whether the defects in the HF cycle and the HFSC are a result of outside-in or inside-out signaling? Further, is it possible that integrin function regulated by SLC3A2 is necessary for more than FN assembly? This could be especially relevant given that your targeted screen also identified Col17A1, which is well known to be required for HFSC function (Matsumura et al., PMID: 26912707)
12. It is intriguing that in the absence of HFSC-derived SLC3A2 that no FN network forms. Is FN expressed or is the assembly perturbed in the absence of properly functioning integrins? The authors conclude that the signaling cascade flows from fibronectin to integrin to SLC3A2, but do not test where the FN phenotype arises in the SLC3A2 knockout - is it due to aberrant assembly of the FN meshwork or a change in transcriptional or translational levels?
13. In the grafting assay in Supplemental Figure 3, keratinocytes undergo a de novo hair follicle morphogenesis - is Lrig1 expression maintained in order to carry out cre-mediated deletion? Further, the fibroblasts in this assay may adopt a wound-like phenotype, expressing FN, which you earlier claim to be required for hair follicle production in wounds. Yet in the absence of epithelial FN, no HF form. Can the authors reconcile this?

Minor comments:

1. In Figure 1a, the two populations are Lgr5+ and basal; please define what the basal population is in this experiment.
2. Significative is not a word.
3. In Figure 4 figure legend, there is reference to a grafting experiment but no experiment shown.
4. The authors delete FN in Lrig1+ or K19+ cells starting D19 and harvest at D30, and conclude that the hair follicles do not enter anagen after the second telogen, can you please include the data supporting the statement that mutant HF did not reenter the hair cycle after D65.

Significance

The authors show for the first time that fibronectin is expressed during cutaneous homeostasis and that it is required for normal function of the hair follicle stem cells. This is significant conceptual advance for the field of skin biology because fibronectin is thought to only be present in wounds: derived first from infiltrating serum and second from fibroblasts to act as provisional dermal ECM to support epithelialization during wound-response, which is ultimately resolved upon the conclusion of wound healing (reviewed in: Singer and Clark, PMID: 10471461). Further, FN has also been characterized as an EMT marker during cancerous progression (Lamouille et al, PMID: 24556840). Estrach and colleagues show that fibronectin is actually expressed by hair follicle stem cell keratinocytes and then is assembled into a meshwork that envelops the hair follicle and is in fact necessary for hair follicle stem cell homeostasis. This work would be broadly interesting to the field of stem cell biology as well as those working on extra cellular matrix signaling. My field is epithelial stem cells and more specifically hair follicle development and cycling.

Referee Cross-commenting

I have no disagreement with any of the points raised by the other reviewers. In fact, we seem to agree on the majority of the concerns. This includes the use of the tail wholemount model, the use of Lrig1-cre, selection of timepoint vs phase of the hair cycle, the appropriateness of the link between Fibronectin and SLC3A2, and further significant issues related to display of data and their reproducibility. Further, all of the major comments raised need to be addressed in order to properly evaluate the conclusions that the authors make. In my opinion, none of the comments raised here are unreasonable.

Authors Response:

Reviewer #2 (Public Review):

The authors use representational similarity analysis on a combination of behavioral similarity ratings and EEG responses to investigate the representation of actions. They specifically explore the role of visual, action-related, and social-affective features in explaining the similarity ratings and brain responses. They find that social-affective features best explain the similarity ratings, and that visual, action-related, and social-affective features each explain some of the variance in the EEG responses in a temporal progression (from visual to action-related to social-affective).

The stimulus set is nicely constructed, broadly sampled from a large set of naturalistic stimuli to minimize correlations between features of interest. I'd like to acknowledge and appreciate the work that went into this in particular.

The analyses of the behavioral similarity judgments are well executed and interesting. The subject exclusion criteria and catch trials for online workers are smart choices, and the authors have tested a good range of models drawn from different categories. I find the case that the authors make for social features as determinants of behavioral similarity ratings to be compelling.

I have a few questions and requests for additional detail about the EEG analyses. I appreciate that the authors have provided the code they used for all the analyses, and I'm sure that the answers to many if not all of my questions are there, but I don't have access to a Matlab license to run the code. Also, since the code requires familiarity with not just Matlab but with specific libraries to understand, I think that more description of the analysis in the paper would be appropriate.

Some more detail is needed in the description of the multivariate classifier analysis. The authors write (line 597-599): "The two pseudotrials were used to train and test the classifier separately at each timepoint, and multivariate noise normalization was performed using the covariance matrix of the training data (Guggenmos et al., 2018). "

I suspect I'm missing something here, because as written this sounds as if there was only one trial on which to train the classifier, which does not seem compatible with SVM classification. If only one trial was used to train the classifier, that sounds more like nearest-neighbor classification (or something else). Alternatively, if all different pseudo-trial averages - each incorporating a different subset of trials - were used for training, then that would seem to mean that some of the training pseudo-trials contained information from trials that were also averaged into the pseudo-trials used for testing. I don't know if this was done (probably not) but if it was it would constitute contamination of the test set. I think this part of the methods needs more detail so we can evaluate it. How many trials were used to train and to test for each iteration?

Thank you for raising this issue; we agree that our Methods section was unclear on this point. We used split-half cross-validation. There was one pseudotrial for training per condition (which was obtained by averaging trials). There was no contamination between the training and test sets, because the data was first divided into separate training and test sets, and only afterwards averaged into pseudotrials for classification. This procedure was repeated 10 times with different data splits to obtain more reliable estimates of the classification performance. We rewrote the corresponding section to make this clearer:

“Split-half cross-validation was used to classify each pair of videos in each participant’s data. To do this, the single-trial data was divided into two halves for training and testing, whilst ensuring that each condition was represented equally. To improve SNR, we combined multiple trials corresponding to the same video into pseudotrials via averaging. The creation of pseudotrials was performed separately within the training and test sets. As each video was shown 10 times, this resulted in a maximum of 5 trials being averaged to create a pseudotrial. Multivariate noise normalization was performed using the covariance matrix of the training data (Guggenmos et al., 2018). Classification between all pairs of videos was performed separately for each time-point. […] The entire procedure, from dataset splitting to classification, was repeated 10 times with different data splits.”

We also performed the decoding procedure with a higher number of cross-validation folds and found very similar results.

I think a bit more detail is also necessary to clarify the features used for the classification. My understanding is that each timepoint was classified as one action vs each other action on the basis of all the electrodes in the EEG for a given temporal window. Is this correct? (I'm guessing / inferring more than a little here.)

This is correct, and we agree that further clarification was needed in text. We have added this:

“Classification between all pairs of videos was performed separately for each time-point. Data were sampled at 500 Hz and so each time point corresponded to non-overlapping 2 ms of data. Voltage values from all EEG channels were entered as features to the classification model.

The entire procedure, from dataset splitting to classification, was repeated 10 times with different data splits. The average decoding accuracies between all pairs of videos were then used to generate a neural RDM at each time point for each participant. To generate the RDM, the dissimilarity between each pair of videos was determined by their decoding accuracy (increased accuracy representing increased dissimilarity at that time point).”

It would be useful to know how many features constituted each feature space. For example, was motion energy reduced to one summary feature (total optic flow for whole sequence?) For "pixel value", is that luminance? (I suspect so, since hue is quantified separately, but I don't think this was specified).

For motion energy, we used the magnitude of the optic flow, and calculated Euclidean distances between the vectorized magnitude maps rather than reducing it to summary features. We have included the dimensionality of each feature in Supplementary File 1b and we now refer to it in text:

“These features were vectorized prior to computing Euclidean distances between them (see Supplementary File 1b for the dimensionality of each feature).”

Pixel value was indeed the luminance, and we have clarified this in text.

More broadly, I would appreciate a bit more discussion of the role of time in these analyses. Each clip unfolds over half a second, so what should we make of the temporal progression of RDM correlations? Are the social and affective features correlated with later responses because they take more time to compute (neurally speaking), or because they depend on longer temporal integration of information? These two are not even exactly mutually exclusive, and I realize that it may be difficult to say with certainty based on this data, but I think some discussion of this issue would be appropriate.

This is a great point, although it is difficult to speculate based on this data. One way to get at this would be to examine how much social-affective processing relies on previously extracted features. Future work could look at the causality between early and later-stage EEG features (unfortunately our post-hoc attempts to address this via Granger-causal analysis were unsuccessful, likely due to insufficient SNR with our specific experimental design). Alternatively, this could be investigated in a follow-up experiment that varies how social information unfolds over time (e.g., images vs. videos or varying video duration). We now discuss this possibility in the manuscript:

“Given the short duration of our videos and the relatively long timescale of neural feature processing, it is possible that social-affective features are the result of ongoing processing relying on temporal integration of the previously extracted features. However, more research is needed to understand how these temporal dynamics change with continuous visual input (e.g. a natural movie), and whether social-affective features rely on previously extracted information.”

Reviewer #2 (Public Review):

The authors use representational similarity analysis on a combination of behavioral similarity ratings and EEG responses to investigate the representation of actions. They specifically explore the role of visual, action-related, and social-affective features in explaining the similarity ratings and brain responses. They find that social-affective features best explain the similarity ratings, and that visual, action-related, and social-affective features each explain some of the variance in the EEG responses in a temporal progression (from visual to action-related to social-affective).

The stimulus set is nicely constructed, broadly sampled from a large set of naturalistic stimuli to minimize correlations between features of interest. I'd like to acknowledge and appreciate the work that went into this in particular.

The analyses of the behavioral similarity judgments are well executed and interesting. The subject exclusion criteria and catch trials for online workers are smart choices, and the authors have tested a good range of models drawn from different categories. I find the case that the authors make for social features as determinants of behavioral similarity ratings to be compelling.

I have a few questions and requests for additional detail about the EEG analyses. I appreciate that the authors have provided the code they used for all the analyses, and I'm sure that the answers to many if not all of my questions are there, but I don't have access to a Matlab license to run the code. Also, since the code requires familiarity with not just Matlab but with specific libraries to understand, I think that more description of the analysis in the paper would be appropriate.

Some more detail is needed in the description of the multivariate classifier analysis. The authors write (line 597-599): "The two pseudotrials were used to train and test the classifier separately at each timepoint, and multivariate noise normalization was performed using the covariance matrix of the training data (Guggenmos et al., 2018). "

I suspect I'm missing something here, because as written this sounds as if there was only one trial on which to train the classifier, which does not seem compatible with SVM classification. If only one trial was used to train the classifier, that sounds more like nearest-neighbor classification (or something else). Alternatively, if all different pseudo-trial averages - each incorporating a different subset of trials - were used for training, then that would seem to mean that some of the training pseudo-trials contained information from trials that were also averaged into the pseudo-trials used for testing. I don't know if this was done (probably not) but if it was it would constitute contamination of the test set. I think this part of the methods needs more detail so we can evaluate it. How many trials were used to train and to test for each iteration?

I think a bit more detail is also necessary to clarify the features used for the classification. My understanding is that each timepoint was classified as one action vs each other action on the basis of all the electrodes in the EEG for a given temporal window. Is this correct? (I'm guessing / inferring more than a little here.)

It would be useful to know how many features constituted each feature space. For example, was motion energy reduced to one summary feature (total optic flow for whole sequence?) For "pixel value", is that luminance? (I suspect so, since hue is quantified separately, but I don't think this was specified).

More broadly, I would appreciate a bit more discussion of the role of time in these analyses. Each clip unfolds over half a second, so what should we make of the temporal progression of RDM correlations? Are the social and affective features correlated with later responses because they take more time to compute (neurally speaking), or because they depend on longer temporal integration of information? These two are not even exactly mutually exclusive, and I realize that it may be difficult to say with certainty based on this data, but I think some discussion of this issue would be appropriate.

What "creative solutions" did her doctors come up with?

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

1. General Statements [optional]

This section is optional. Insert here any general statements you wish to make about the goal of the study or about the reviews.

We are grateful to the reviewers for their honest opinion regarding this work and plan to address the majority of the comments in a revised version either through new analysis or revision to the text, as we believe these will improve the manuscript by making some of the details clearer. There were few suggestions that will lead to substantiative changes to the findings. Here, we address the most salient critiques, the primary one being related to novelty.

We respectfully disagree, as our detailed analysis of the DNA methylome in Octopus bimaculoides represents a significant advance to understanding how the epigenome is patterned in non-model invertebrates in general, and cephalopods in particular. We acknowledge that the previous report that the octopus methylome resembles the few other invertebrates where low DNA methylation has been found, the finding was part of a multi-organism study last year (de Mendoza et al., 2021), which lacked any detailed investigation. Our study provides the first in depth analysis on methylation patterning, the relationship with transposons and gene expression, and reports the finding of other key epigenetic marks in O. bimaculoides, and in other cephalopods.

In short, we believe our study to be highly novel and that it represent the first analysis of this kind in cephalopods and one of the few existing in non-model invertebrate organisms. In addition, we identify the conservation of the histone code in cephalopods. While this may be expected, this is the first experimental evidence in this class and represents an important step forward to understand the epigenetic regulation of genes and transposons in invertebrates. Finally, we plan to provide an updated transcriptome annotation for O. bimaculoides that will be available for the scientific community as a new valuable resource. We believe these features will make this study highly cited.

We believe that findings like ours will complement several recent studies that extend the epigenetics field out of the current narrow focus on model organisms to understand how epigenetic mechanisms function in diverse animals. This provides new insights regarding the epigenetic mechanism of gene regulation in an emerging invertebrate model.

2. Description of the planned revisions

Insert here a point-by-point reply that explains what revisions, additional experimentations and analyses are planned to address the points raised by the referees.

Reviewer 1 raised the following points that we are planning to address:

*- It is unclear why the authors did not use the original gene models of O. bimaculoides or tried to improve them. By only relying on adult tissue (but the relatively late hatchling stage), they would have omitted most developmentally expressed genes, that are incidentally also the ones that are subjected to extensive spatiotemporal gene regulation (which is also a problem to assess the role of methylation). I think more comparisons with existing gene models and how the newly generated stringtie models should be provided. *

We agree that using as many tissues and developmental stages as possible will expand the octopus transcriptome.

We plan to:

• Add RNA-seq data from stage 15 embryos to improve this.
• Compare the gene model used in the original version of the manuscript (Stringtie model to use in Trinotate for improving the annotation of the genes) to the existing annotation model and report on which has superior performance for annotating the * bimaculoides* transcriptome.
• Extend the annotation of the transcriptome which we undertook in a focused fashion in the first iteration of this manuscript. Reviewer 2 raised the following points that we are planning to address:

*- It is not exactly clear to me why the authors look for expression clusters in the first part of the manuscript? This information, while interesting, does not seem to be used in the methylation analysis. It is also somewhat contradictory because the authors first claim that, based on their GO-term enrichment analysis, that different expression clusters are associated with "complex regulatory mechanisms, potentially based in the epigenome". Yet at the end they conclude that, due to the global and tissue-overarching nature of methylation, this "argues against this epigenetic modification as a player in the dynamic regulation of gene expression". *

We thank the reviewer for pointing out this issue and we plan to clarify the point through changing the text and additional analysis. Since we found that the methylation pattern was stable across tissues, and that it corresponded to gene expression levels regardless of tissues, we concluded that the methylation pattern is not likely relevant for the tissue-specific gene expression pattern reported in Figure 1.

We plan to:

• Ask whether there is a correlation between the gene clusters generated in Figure 1 and the DNA methylation patterns identified in Figure 4. *- At least for the trees that are shown in the main figures it would be great to show support values. *

We thank the reviewer for this request.

We plan to:

• Add full Supplementary information regarding the support values in Supplemental Files for all the trees present in the main Figures. Reviewer 3 raised the following points that we are planning to address:

*- It would be great to see more data on cephalopod TET and MBD structure. For example, it would be interesting to know whether octopus TETs have a CxxC domain or whether MBD proteins harbor functional 5mC - binding domains. *

We agree that it would be of interest to examine the conservation of TET genes to expand upon the initial analysis by Planques et al 2021 showing that O. bimaculoides have one TET homolog, one MBD4 homolog and one MBD1/2/3 homolog. Detailed analysis of MBD4 protein has been already performed in de Mendoza et al. 2021 by using the protein sequence of O. vulgaris, as the MBD4 gene in the O. bimaculoides genome appears truncated.

We plan to:

• add the PFAM domain analysis for TET proteins This will be added as a new figure panel.
• Update the text to include the reference to the identification of MBD4/MECP2 as the invertebrate homologs of vertebrate MBD4. *- Even though RRBS provides limited insight into DNA methylation patterns, the authors could have done more to explore read-level 5mC information. For example, by studying single reads, the authors could deduce the numbers of fully methylated, unmethylated or partially methylated reads. Such analyses might provide valuable insight into potentially different modes of epigenetic inheritance in different tissues i.e are there tissues that favor fully methylated or unmethylated stretches of DNA vs tissues that favor partial methylation? *

We think this is a really interesting point. This has been partially addressed in a previous work (de Mendoza et al., 2021) which found limited to no partially methylated reads in whole-genome bisulfite sequencing from O. bimaculoides brain.

We plan to:

• Use Proportional Discordant Reads on the RRBS data we generated from 30 dpf hatchlings to assess the presence of discordant methylated reads across different tissues to assess possible different modes of epigenetic inheritance. We will use “WSHPackage” in R: https://academic.oup.com/nar/article/48/8/e46/5760751?login=true https://github.com/MPIIComputationalEpigenetics/WSHPackage/blob/master/vignettes/WSH.md#x1-50003.1 This will be added as a new figure panel.

3. Description of the revisions that have already been incorporated in the transferred manuscript

Please insert a point-by-point reply describing the revisions that were already carried out and included in the transferred manuscript. If no revisions have been carried out yet, please leave this section empty.

Reviewer 1 raised the following points that we have already addressed:

We addressed all the comments raised by this Reviewer by revising the text, fixing references, typos and improving clarity.

Reviewer 2 raised the following points that we have already addressed:

We addressed all the minor Comments raised by this Reviewer regarding spelling errors and Supplementary Figures.

- The finding that less than 10% of all possible sites are methylated is surprising. I could not (easily) find statistics of RRBS experiment read mapping to the genome.

We have now provided this data and new Supplemental Table 1 (refereed in the text as Table S1).

*- It is very exciting to see methylation of gene bodies and some correlation to their expression levels, but the authors may need to include a disclaimer that the methylation of TEs may go undetected due to the gapness of the genome. In fact, the authors may try to map their data onto a somewhat closely related Octopus sinensis genome sequenced with long reads available at NCBI to confirm overall pattern. It is likely though that due the evolutionary distance only gene bodies will have mapping. *

The thank the reviewer for this suggestion and we included a sentence in the Result session indicating that methylation of TEs may go undetected due to the poor annotation of the octopus genome.

*- The statistical reasoning (and methodology) behind how clusters in Figures 1 and 4 were defined is unclear. In particular, in Figure 4, it seems that the authors had asked the program to give four clusters in total - why was this number chosen? It seems that using the same generic clustering approach as in Figure 1 may benefit or confirm the results in Figure 4. *

We clarified the rationale in the Material and Methods session to describe the bioinformatic analysis. We will put the full code used in the manuscript in our GitHub page (https://github.com/SadlerEdepli-NYUAD/) to have a more comprehensive understanding of the Method used.

Reviewer 3 raised the following points that we have already addressed:

We addressed all the minor comments in the text and figures raised by this reviewer regarding typos and clarity.

*- There is little info on the generated 5mC data. To bolster its value as a resource, the manuscript should have a link to the table describing RRBS metrics. This should include: non-conversion rates, numbers of sequenced and mapped reads, read length and other info that the authors deem useful. *

We have now provided this data in a new Supplemental Table 1 (refereed in the text as Table S1).

4. Description of analyses that authors prefer not to carry out

Please include a point-by-point response explaining why some of the requested data or additional analyses might not be necessary or cannot be provided within the scope of a revision. This can be due to time or resource limitations or in case of disagreement about the necessity of such additional data given the scope of the study. Please leave empty if not applicable.

Reviewer 1 raised the following points that we are not planning to address:

*- The newly sequence RNA-seq samples are using a ribodepletion protocol (RiboZero) while the other ones are using a polyA selection. This might be a slight problem to compare them quantitatively. Actually in the Figure 1, all 4 newly generated samples group together in the hierarchical clustering. *

We acknowledge the reviewer’s point here and agree that heterogeneity in library prep and batch is a common issue when comparing public available with newly generated datasets. This could account for the clustering of the Ribosomal RNA depleted (i.e. RiboZero) from polyA selected RNA libraries. While this could potentially introduce bias, we do not believe that it substantially alters any of the main findings or the interpretations of this data. Our purpose for carrying out the cluster analysis of transcriptomic data from multiple tissues was to identify distinct gene patterns that defined different tissue types. This was accomplished regardless of the potential confounding variable introduced by different library preparations. In addition, we used TMP which seems to help in the comparison across different samples when used for qualitative analysis such as PCA and cluster analysis (Zhao et al. 2020; DOI: http://www.rnajournal.org/cgi/doi/10.1261/rna.074922.120). Therefore, even if not ideal we think that this approach is still valuable.

*- I am not so sure about the way the authors used z-score normalized logTPMs and applied hierarchical clusters, this most likely would not fully alleviate the impact of expression level on the outcome compared to more advanced form of normalization and clustering. *

We agree with the reviewer that applying z-score or a logTPMs normalization would not fully resolve the technical variance in the direct comparison of libraries generataed with different RNA selection methods. We did not apply z-score on logTPMs but these 2 methods were applied separately: z-score on TPMs in Figure 1B to define the gene clusters and log2(TPM+1) in Figure 4E. We have clarified the text to reflect this.

*- I am not convinced that differences in western blot for histone modification could really provide a clear insight into their regulatory role. *

We agree with the reviewer that Western blotting for histone modifications does not provide deep insight into their regulatory role. However, this is the first description of these marks in any cephalopod, and we believe that reporting a finding from experimental evidence is important, even if the result is aligned with the existing paradigm. Moreover, the marked difference in levels of distinct histone marks across tissues supports the hypothesis that they play a regulatory role. We observed this in mice where difference abundance in western blot correspond to different abundance and enrichment also by ChIP-seq (Zhang et al., 2021 DOI: https://doi.org/10.1038/s41467-021-24466-1). Considering the limited tools available in this species, we still consider this an important finding.

Reviewer 2 raised the following points that we are not planning to address:

*- The finding that less than 10% of all possible sites are methylated is surprising. I could not (easily) find statistics of RRBS experiment read mapping to the genome. I also wonder how much the gap-richness of the genome may affect the overall methylation estimate. If assembly permits, would it make sense to limit the sampled sites to areas where no flanking gaps are present (and sufficient scaffold length is available, maybe excluding very short scaffolds)? *

We added all the statistical values regarding the RRBS in a NEW Supplemental Table 1. We used a single base pair analysis approach (not tiling windows), so the data we extracted is not biased by the length of the scaffolds. This is confirmed by the fact that the DNA methylation value obtained in our RRBS data matches the findings observed in Whole Genome Bisulfite Sequencing (WGBS). Moreover, global DNA methylation values assessed by Slot blot analysis as a technique independent from genome assembly confirmed what observed with RRBS.

I think that the is possible and we could see it at more progressive schools. It is just like the banking industry, one bank got rid of overdraft fees and slowly we are seeing major banks do away with the fees as well. All it takes is for one brave institution and another to change the landscape of things.

• Fast Car is a song written and performed by American folk rock singer Tracy Chapman in 1986.
• As quoted by Tracy Chapman in an interview in 1986 with Canadian radio station CIDR, “I think that it was a song about my parents…..a new life together and my mother was anxious to leave home. [They]tried to make a life for themselves and it was very difficult going….my mother didn’t have a high school diploma and my father was a few years older….hard for him to create the ideal life that he wanted…in a sense I think they came together thinking that they would have a better chance of making it”
• The song became popular around the 1980s after Tracy Chapman performed it at the 70th birthday Nelson Mandela tribute.
• The song is narrative, and tells the story from the point of view of a woman who wants to get out of her hometown to leave her problems behind, such as taking care of her alcoholic father, after her mother left the family. She finds the addressee, a lover with the titular fast car, and sees it as an opportunity to get out of her hometown at last and start life anew. However, the reality is far from that, as the song continues, and reveals how even after moving to the city, she is still working a dead-end job and living in a homeless shelter, while her lover is unemployed. It parallels her own parents’ relationship, as her lover becomes a father and an alcoholic, and their relationship grows rocky. Eventually she decides that she does not need him anymore, and tells him to either clean up his act or leave.
• Hope and resilience are major themes in this song, as the persona continues to persevere despite the many challenges that are thrown at her in life, from her upbringing to her newfound problems.However, she remains optimistic and presses on, and we find hope in how she eventually achieves independence and supposedly leaves her problems behind.
• One can interpret the title ‘Fast Car’ as a symbol of escapism. However, as the song goes on, it is revealed that getting out of a situation is far from easy and even if you start anew, there will always be problems and challenges in life to overcome. Additionally, sometimes, you will find that people who are close to you make irresponsible and selfish decisions which affect you .In order to carry on a meaningful and purposeful life, you may have to give up on them. To prevent this from happening, it is essential to become independent and not place all your faith in someone else, just as the persona has done in Fast Car.

I think that this is so important. Being aware of the different privacy or security issues that may arise when using different technologies is something that everyone should be aware of in order to be safe. We have learned that not all tools are safe and there are many times where passwords and personal information can be stolen. Also, students may not actually fully research the privacy of a website so having a parent also research the tool/site can help ensure that all information will be safe.

It is so easy to click "accept" online which may be sharing your information and we often don't even think twice and should be paying closer attention to this.

I think this is a huge "however" when looking at how helpful an accessibility statement is. If there are huge gaps then the statement is not as helpful but also if it is quite extensive you have to remember that it is still the company writing it up, and they may be lacking in knowledge. I think one day regulations will be put in place for the VPAT but that is not where we are yet.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Summary:

Formation of tubes in a developing organism may arise from the closure of a pre-existing polarized epithelium or from de novo polarization and cavity formation in group of dividing cells. The concept of apical membrane initiation site (AMIS) refers to the fact that polarity proteins as PAR3 accumulate at a point where the apical membrane will be created. This accumulation occurs as early as the two cell stage. Previous reports have demonstrated the importance of the division process in defining this AMIS, however, in the present work the authors in vitro 3D cultures of mESC to report a mitosis independent mechanism that creates an AMIS, induces the polarization of groups of two or more cells, and permits the formation of a central cavity. The report shows that the mechanism is fully dependent on the polarized accumulation of E-cadherin at the cell membrane in contact with the other cells. Moreover, the mechanism does not require mitosis or interaction with the extracellular matrix.

Major comments:

The main objective of the work is to demonstrate that AMIS creation and cavity formation can be mitosis independent and that it is dependent on the accumulation of E-cadherin at the midline between two cells in contact. To demonstrate these objectives, the authors perform 3D cultures of mESC. To rule out the requirement of mitosis the authors perform cultures that are treated with mitomycin C and the purify single cells that are cultured again. The authors show time-laps experiments demonstrating that individual cells that do not dived create an AMIS when they contact one to each other. With this cultures they demonstrate that the process does not require an interaction with ECM (provided by the matrigel) but requires E-cadherin, to demonstrate, that they use E-cadherin KO cells (the same line where E-cadherin has been deleted). The work is well written and the objectives very clear. The technology used and the experiments done are adequate and sufficient to accomplish the proposed objectives and the results obtained clearly support the conclusions reached. The methods are well explained and transparent to be reproduced elsewhere and the number of replicas and the statistical methods applied seem corrects to me, although I am just a biologist, not a mathematician. Although the objectives of the work, that are: to demonstrate that AMIS formation can be independent of mitosis and that AMIS requires E-cadherin, there are parts of the results that could be farther studied or at least discussed more thoroughly. Firstly, the authors show that in non-dividing cells an AMIS is formed at the first contact site between the two cells, they also show that in the absence of E-cadherin the cell maintains the polarization of centrioles and Golgi apparatus, in spite that no AMIS is formed, this indicates that the deposition of E-cadherin at the midline membrane is part of a more global polarization event that most likely is initiated by the a directional activity of the Golgi apparatus that may direct the delivery of mature E-cadherin in that particular direction, initiating or maintaining the basis for an AMIS, since recent work (already cited in the manuscript) has demonstrated the importance of cadherin maturation for polarity establishment and maintenance (Herrera et al, 2021), the actual results should be farther discussed in this context. Secondly, it was previously shown that in different epithelia, upon cell-cell contact, the aPKC complex (that includes Par3 and Par6) is recruited early to the contact site where with the participation of Cdc42, aPKC is activated generating an initial spot-like adherent junction (AJs) (Suzuki et al., 2002). In that case it is thought to be mediated by a direct interaction between the first PDZ domain of PAR-3 and the C-terminal PDZdomain-binding sequences of immunoglobulin-like cell adhesion molecules: JAM-1 and nectin-1/3 (Fig. 3) (Ebnet et al., 2001; Itoh et al., 2001; Takekuni et al., 2003). Thus it wold be interesting to know if AMIS formation in absence of cell division depends on JAM-1 or nectin and whether JAM-/Nectin signalling is sufficient to initiate the Golgi and centriole polarization and which is the mechanism governing it.

Minor comments:

As I mentioned before, the paper is well presented and very clear, yes it is simple, but simple is always better, no complicated graphics or letterings, thank you. Although in my opinion the work is very well written, I have to admit that I am not qualified to evaluate the literary style of the work since English is not my mother tongue, also I have not reviewed typographical errors since I think that is the work of the editorial, not of scientific reviewers. Please include the full reference of all the antibodies used, including the company and not just the catalog number

Quoted references:

Ebnet, K., Suzuki, A., Horikoshi, Y., Hirose, T., Meyer Zu Brickwedde, M. K., Ohno, S. and Vestweber, D. (2001). The cell polarity protein ASIP/PAR-3 directly associates with junctional adhesion molecule (JAM). EMBO J. 20, 3738-3748.

Itoh, M., Sasaki, H., Furuse, M., Ozaki, H., Kita, T. and Tsukita, S. (2001). Junctional adhesion molecule (JAM) binds to PAR-3: a possible mechanism for the recruitment of PAR-3 to tight junctions. J. Cell Biol. 154, 491-497.

Takekuni, K., Ikeda, W., Fujito, T., Morimoto, K., Takeuchi, M., Monden, M. and Takai, Y. (2003). Direct binding of cell polarity protein PAR-3 to cell-cell adhesion molecule nectin at neuroepithelial cells of developing mouse. J. Biol. Chem. 278, 5497-5500

Suzuki, A., Ishiyama, C., Hashiba, K., Shimizu, M., Ebnet, K. and Ohno, S. (2002). aPKC kinase activity is required for the asymmetric differentiation of the premature junctional complex during epithelial cell polarization. J. Cell Sci. 115, 3565-3573.

Significance

The paper describes for the first time that contrary to what was previously believed an AMIS can be generated without a cell division. This is very important because it opens the possibility that the mechanisms that originate the biologic cavities are in fact not really how we believed. The work is of interest of all cell biology scientists, specially working in developmental biology, cancer research.

My particular field of expertise is cell biology and signaling, always applied to particular events as nervous system development or cancer, in particular I am interested in Wnt/b-catenin and Sonic Hedgehog pathways.

The speaker foreshadows about how their relationship would undoubtedly end. The word ‘story’ symbolises their relationship, which both may have a start and an end. The different parts of a story can resemble the life in a relationship with someone you love. The ‘introduction’ resembles how they meet, the ‘rising action’ resemble how they have taken things further and started to date, the ‘climax’ can resemble the fights or arguments that start to take place inferred from the previous line. And it carries on until the ‘conclusion’, where they separate. From this, I can reflect on the theme of relationship as how it may only last for a brief moment, and shatter, which can hurt so much so it is hard to let go. I feel proud for the speaker even though he fails repetitively to let go but tries his best to do so

Reviewer #1 (Public Review):

In this manuscript, Yang et al. trained monkeys to play the classic video game Pac-Man and fit their behavior with a hierarchical decision making model. Adapting a complex behavior paradigm, like Pac-Man, in the testing of NHP is novel. The task was well-designed to help the monkeys understand the task elements step-by-step, which was confirmed by the monkeys' behavior. The authors reported that the monkeys adopted different strategies in different situations, and their decisions can be described by the model. The model predicted their behavior with over 90% accuracy for both monkeys. Hence, the conclusions are mostly supported by the data. As the authors claimed, the model can help quantify the complex behavior paradigm, providing a new approach to understanding advanced cognition in non-human primates. However, several aspects deserve clarification or modification.

1. The results showed that the monkeys adopted different strategies in different situations, which is also well described by the model. However, the authors haven't tested whether the strategy was optimal in a given situation.

Our approach to analyze monkeys’ behavior is not based on optimality. Instead, we centered around the strategies and showed that they described the monkeys’ behavior well. The model and its fitting process does not assume the monkeys were optimizing for something. Nevertheless, the fitting results suggested that the strategies that the monkeys chose were rational, which suggests validity of our model. As we have pointed out above, optimality is hard to define in such a complex game. In particular, most of the game is about collecting pellets, strategies that are only used in a small portion of the game can be ignored when searching for optimal solutions. We feel that further analyses on the issue of optimality would dilute the center message of the paper and choose not to include them here.

According to the results, the monkeys didn't always perform the task in an optimal way, as well. Most of the time, the monkeys didn't actively adopt strategies in a long-term view. They were "passively" foraging in the task: chasing benefit and avoiding harm when they were approached. This "benefit-tending, harm-avoiding" instinct belongs to most of the creatures in the world, even in single-cell organisms. When a Paramecium is placed in a complex environment with multiple attractants and repellents, it may also behave dynamically by adopting a linear combination of basic tending/avoiding strategies, although in a simpler way. In other words, the monkeys were responding to the change of environment but not actively optimizing their strategy to achieve larger benefits with fewer efforts. The only exception is the suicides. Monkeys were proactively taking short-term harms to achieve large benefits in the future.

One possible reason is that the monkeys didn't have enough pressure to optimize their choices since they will eventually get all the rewards no matter how many attempts they make. The only variable is the ghosts. Most of the time, the monkeys didn't really choose between different targets/ strategies. They were making choices between the chasing order of the options, but not the options themselves. It is similar to asking a monkey to choose either to eat a piece of grape or cucumber first, but not to choose one and give up the other one. A possible way to avoid this is to stop the game once the ghost catches the Pac-Man or limit each game's time.

The game is designed to force the players to make decisions quickly to clear the pellets, otherwise the ghosts would catch Pac-Man. Even in the monkey version of the game where the monkeys always get another chance, Pac-Man deaths lead to long delays with no rewards. They will not be able to complete the game if they do not actively plan their route, especially in the late stage when they must reach the scarcely placed dots while escaping from the ghosts. In addition, we provided additional rewards when a maze is cleared in fewer rounds (20 drops if in 1 to 3 rounds; 10 drops if in 4 to 5 rounds; and 5 drops if in more than 5 rounds), which added motivation for the monkeys to complete a game quickly.

The monkeys’ behavior also suggested that they did not just adopt a passive strategy. Our analyses of the planned attack and suicide behavior clearly demonstrated that the monkeys actively made plans to change the game into more desirable states. Such behavior cannot be explained with a passive foraging strategy.

2. It is well known that the value of an element is discounted by time and distance. However, in the model, the authors didn't consider it. A relevant problem will be the utility of the bonus elements, including the fruits and scared ghosts. Their utilities were affected not only by their value defined by the authors but also by effects, including their novelty and sense of achievement when they were captured, as the ghosts attracted relatively much more attention than the other elements (considering the number is 2 for them, see in figure 3E).

These are good points, and our strategies could be built with more complexity to account for other potential factors. However, we focused our investigation on how to account for monkeys’ behavior with a set of strategies. A set of simple strategies with a small number of parameters would make a strong argument.

Using a complex game such as Pac-Man allows us to investigate all of these interesting cognitive processes. We can certainly look at them in the future.

3. The strategies are not independent. They are somehow correlated to each other. It may result in, in some conditions, false alarming of more strategies than the real, as shown in figure 2A.

We have computed the Pearson correlations between the action sequences chosen with each basis strategy within each coarse-grained segment determined by the two-pass fitting procedure. As a control, we computed the correlation between each basis strategy and a random strategy, which generates action randomly, as a baseline. Most strategy pairs' correlations were lower than the random baseline. The results were now included in Supplementary (Appendix Figure 3).

Sometimes two strategies may give exactly the same action sequence in a game segment. To deal with this problem, now we include an extra step when we fit the model to the behavior, which was described in Methods:

“To ensure that the fitted weights are unique (Buja et al., 1989) in each time window, we combine utilities of any strategies that give exactly the same action sequence and reduce multiple strategy terms (e.g., local and energizer) to one hybrid strategy (e.g., local+energizer). After MLE fitting, we divide the fitted weight for this hybrid strategy equally among the strategies that give the same actions in the time segments.“

Moreover, as the reviewer correctly reasoned, correlations between the strategies would yield possibly more strategies. However, our finding is that the monkeys were using a single strategy most of the time. This possible false alarm would go against our claim. Our conclusions stand despite the strategy correlations.

It is hard to believe that a monkey can maintain several strategies simultaneously since it is out of our working memory/attention capacity.

Exactly, and we are among the first to quantitatively demonstrate that the monkeys’ mostly relied on single strategies to play the game.

Reviewer #2 (Public Review):

In this intriguing paper, Yang et al. examine the behaviors of two rhesus monkeys playing a modified version of the well-known Pac-Man video game. The game poses an interesting challenge, since it requires flexible, context-dependent decisions in an environment with adversaries that change in real time. Using a modeling framework in which simple "basic" strategies are ensembled in a time-dependent fashion, the authors show that the animals' choices follow some sensible rules, including some counterintuitive strategies (running into ghosts for a teleport when most remaining pellets are far away).

I like the motivation and findings of this study, which are likely to be interesting to many researchers in decision neuroscience and animal behavior. Many of the conclusions seem reasonable, and the results are detailed clearly. The key weakness of the paper is that it is primarily descriptive: it's hard to tell what new generalizable knowledge we take away from this model or these particular findings. In some ways, the paper reads as a promissory note for future studies (neural or behavioral or both) that might make use of this paradigm.

I have two broad concerns, one mostly technical, one conceptual:

First, the modeling framework, while adequate, is a bit ad hoc and seems to rely on many decisions that are specific to exactly this task. While I like the idea of modeling monkeys' choices using ensembling, the particular approach taken to segment time and the two-pass strategy for smoothing ensemble weights is only one of many possible approaches, and these decisions aren't particularly well-motivated. They appear to be reasonable and successful, but there is not much in the paper to connect them with better-known approaches in reinforcement learning (or, perhaps surprisingly, hierarchical reinforcement learning) that could link this work to other modeling approaches. In some ways, however, this is a question of taste, and nothing here is unreasonable.

Thanks for the suggestion. In the new revision, we include a linear approximate reinforcement learning model (LARL) (Sutton, 1988; Tsitsiklis & Van Roy, 1997). The LARL model shared the same structure with a standard Q-learning algorithm but used the monkeys’ actual joystick movements as the fitting target. The model, although computationally more complex than the hierarchical mode, achieves a worse fitting performance.

Second, there is an elision here of the distinction between how one models monkeys' behavior and what monkeys can be said to be "doing." That is, a model may be successful at making predictions while not being in any way a good description of the underlying cognitive or neuroscientific operations. More concretely: when we claim that a particular model of behavior is what agents "actually do," what we are usually saying is that (a) novel predictions from this model are born out by the data in ways that predictions from competing models are not (b) this model gives a better quantitative account of existing data than competitors. Since the present study is not designed as a test of the ensembling model (a), then it needs to demonstrate better quantitative predictions (b).

We concede to the point that our model, while fitting to the behavior well, does not directly prove that the monkeys actually solved the task in this way. The eye movement and pupil dilation analyses partly addressed this issue, as their results were consistent with what one would expect from the model. We also hope future recording experiments will provide neural evidence to support the model.

But the baselines used in this study are both limited and weak. A model crafted by the authors to use only a single, fixed ensemble strategy correctly predicts 80% of choices, while the model with time-varying ensembling predicts roughly 90%. This is a clear improvement and some evidence that *if* the animals are ensembling strategies, they are changing the ensemble weights in time. But there is little here in the way of non-ensemble competitors. What about a standard Q-learning model with an inferred reward function (that is, trained to replicate monkeys' data, not optimal performance). The perceptron baseline as detailed seems very poor as a control given how shallow it is. That is, I'm not convinced that the authors have successfully ruled out "flat" models as explanations of this behavior, only found that an ensembled model offers a reasonable explanation.

We hope the new LARL model provides a better baseline control as a flat model. It performs better than the perceptron, yet much worse than our hierarchical model. Yet, we must point out that any hierarchical models can be matched in performance with a flat model in theory (Ribas-Fernandes et al., 2011). The advantage of hierarchical models mainly lies in their smaller computational cost for efficient planning. Even in a much simpler task such as a four-room navigation task, a hierarchical model can plan much faster than a flat model, especially under conditions with limited working memory (M. Botvinick & Weinstein, 2014). Our Pac-Man task contains an extensive feature space while requiring real-time decision-making. The result is that a reasonably performing flat model would go beyond the limits of the cognitive resources available in the brain. Even for a complex flat model such as Deep Q-Network (it can be considered to be similar a flat model since it does not explicitly plan with temporal extended strategies (Mnih et al., 2015)), the game performance is much worse than a hierarchical model (Van Seijen et al., 2017). The performance of the monkeys was unlikely to be achieved with a flat model. In addition, we trained the monkeys by introducing the game concepts gradually, with each training stage focusing on certain game aspects. The training procedure may have encouraged the monkeys to generalize the skills acquired in the early stages and use them as the basis strategies in the later training stages when the monkeys faced the complete version of the Pac-Man task.

Reviewer #3 (Public Review):

Yang and colleagues present a tour de force paper demonstrating non-human primates playing a full on pac-man video game. The authors reason that using a highly complex, yet semi controlled video game allows for the analysis of heuristic strategies in an animal model. The authors perform a set of well motivated computational modeling approaches to demonstrate the utility of the experimental model.

First, I would like to congratulate the authors on training non-human primates to perform such a complex and demanding task and demonstrating that NHP perform this task well. From previous papers we know that even complex AI systems have difficulty with this task and extrapolating from my own failings in playing pac-man it is a difficult game to play.

Overall the analysis approach used in the paper is extremely well reasoned and executed but what I am missing (and I must add is not needed for the paper to be impactful on its own) is a more exhaustive model search. The deduction the authors follow is logically sound but builds very much on assumptions of the basic strategy stratification performed first. This means that part of the hierarchical aspect of the behavioral strategies used can be attributed to the heuristic stratification nature of the approach. I am not trying to imply that I do not think that the behavior is hierarchically organized but I am implying that there is a missed opportunity to characterize that hierchical'ness (maybe in a graph theoretical way, think Dasgupta scores) further.

All in all this paper is wonderful. Congratulations to the authors.

We thank the reviewer for the encouraging comments. We have included a new flat model in the new revision for comparison against our hierarchical model and discussed other experimental evidence to support our claim.

While multiple readings of a text in antiquity may have been rarer, due to the cheap proliferation of books, one can more easily "blaze a trail" through their reading to make it easier or quicker to rebuild context on subsequent readings.

Look at history of reading to see which books would have been more likely re-read, particularly outside of one's primary "area" of expertise.

Link to the trails mentioned by Vannevar Bush in As We May Think.

Author Response:

Reviewer #1:

In this work Warneford-Thomson et al. developed an approach for surveillance screening for SARS-CoV-2, which involves the isothermic amplification of a region of the SARS-CoV-2 nucleocapsid gene using RT-LAMP, followed by detection with deep sequencing. High-throughput and cost effectiveness is achieved by two sets of barcodes that allow up to about 37,000 samples to be combined into one deep sequencing run. Moreover, the authors demonstrate they can do the detection from saliva collected on paper, which should make sample collection easier.

The main strength of the work lies in the technical aspects, including setting up multiple controls such as a detection of a human gene, and multiplexing with detection of the influenza virus.

The main weakness is that there are multiple other papers either published or archived that use RT-LAMP for SARS-CoV-2 detection, deep sequencing for SARS-CoV-2 detection, or both. These are cited in the current work, which is very well written and presented. Whether this method is better than the others which have the same aim of developing cost-effective and high-throughput detection is not conclusively demonstrated as only 8 clinical saliva samples are examined.

We do not wish to claim that our method is better than the others. We think it has advantages and disadvantages and certainly it should be further optimized before scaling it up to population level. We have added these considerations to the text (lines 376–80).

Furthermore, the requirement for deep sequencing and batching many samples for cost-effectiveness will, in most situations, greatly increase turn-around time. This will make surveillance much less effective, since by the time results are fed back, the asymptomatically infected individual would have had more opportunity to transmit the infection to others.

We argue that time from sample to result is a mostly a function of logistics and not of the method. With proper set ups the time from sample collection to results could be < 16 hours, which would be compatible with population-level surveillance. We added these considerations to the text.

However, the deep sequencing step may be very useful for surveillance of circulating SARS-CoV-2 spike sequences to detect emerging variants within a population, provided this method can be modified to do it.

We agree and we mention this possibility in the discussion.

Reviewer #2:

In 'COV-ID: A LAMP sequencing approach for high-throughput co-detection of SARS-CoV-2 and influenza virus in human saliva', Warneford-Thomson et al. present a novel methodology to perform large numbers of COVID-19 tests in parallel. Their approach takes unprocessed saliva and requires only a small number of experimental steps before the results are sequenced overnight to generate many thousands of results. This straightforward experimental design should allow the protocol to be expanded to a number of settings where population-level monitoring is required in order to contain outbreaks and reduce transmission. In this paper, the authors demonstrate the efficacy of their approach and perform a large number of benchmarking experiments to quantify its sensitivity, specificity and limitations of detection. They are able to detect artificially created infections (spike-ins) with as low as 5 virions per µL and all clinically available samples agreed with the standard RT-qPCR test. This method can detect both SARS-CoV-2 and Influenza infection and can also be applied to saliva samples which have been collected on filter paper, a strategy which will further simplify the testing regime.

The authors have spent much time testing this approach but these have largely been limited to analysing artificially created infections. The only results which were obtained were from eight clinically derived samples which are presented in Figure 2E. Although all results from this approach agreed with the standard clinical test this is a small number of tests compared to the total number of tests which are reported in this paper. It is also only a small proof-of-principle experiment to justify a quick rollout of this technology.

We have now performed COV-ID on 120 additional patient samples (new Figure 2-figure supplement 2). These new results are described in the text.

The potential for this technology to perform rapid, high-throughput SARS-CoV-2 testing alongside the potential for very low sequencing costs (Figure 4G) is impressive. It is noted in the manuscript that this will require 96 unique barcodes but only 32 are tested here. All but three of these 32 work for the SARS-CoV-2 N2 primers and required STATH control but how will the remaining 67 primers be derived (i.e. is it realistic that this can be made to work to deliver the promise of this approach)?

The current COV-ID patient barcodes are 5 base pairs long. This allows for 4^5 = 1,024 combinations. Out of an abundance of caution, we excluded barcodes with homology to the reverse complement of the RT-LAMP primers used in any of the experiments (i.e. primers for SARS-CoV-2 N2, STATHERIN, ACTIN, and influenza virus) and then selected a set of 32 with Hamming distances of at least 2 from each other. This is now described more in detail in the methods.

Regarding the numbers, out of 1,024 5-bp barcodes, 404 were removed due to homology, leaving 620. Of these, we could find at least 163 with Hamming distance ≥ 2 from each other. Even with a substantial failure rate, this should allow for 96 working barcodes. If we had only considered clashes with N2 and STATHERIN primers, the number of available barcodes would be substantially higher.

Overall, this is an interesting paper which has very clear real-world application to helping to defeat the ongoing COVID-19 pandemic, but some extra validations are needed to fully demonstrate its performance in clinical and/or public health settings.

Author Response:

Reviewer #1:

The experiments are well designed, generally well controlled, and carefully conducted, and are thoughtfully and appropriately discussed. The authors make conclusions that are well supported by their results.

When describing the aptamer knockdown of the PPS, the authors explain that the western blot was too noisy for monitoring the knockdown, which is frustrating for the reader and must have been frustrating for the authors. The authors instead counter-intuitively use qRT-PCR to monitor the transcript abundance of the PPS transcript in the aptamer system - this aptamer system is thought to be a modifier of protein, not transcription or transcript abundance. The authors describe that this has been seen once before (using aptamer knockdown of PfFis1), and the authors of that study speculate that the TetR-DOZI aptamer might be degrading the target mRNA. This is a plausible explanation, but it isn't quite clear from the description how this experiment was performed. The authors explain that the knockdown parasites grew normally for three days, but the parasites may be becoming sicker over this period. It's therefore possible that the decrease in PPS mRNA abundance is a product, rather than a cause of the growth defect. Sick or dying parasites could plausibly impact the PPS differently to the two chosen controls, particularly since both control genes chosen have substantially longer half-lives than the PPS mRNA (according to the Shock and DeRisi datasets). I therefore I suggest that this experiment be performed in an IPP rescue scenario (where the parasites aren't dying) with biological replicates. There is no explanation of the replicates here, but the error bars in 6C are implausibly small for real biological replicates.

To address these concerns, we have added western blot data showing down-regulation of PPS expression in -aTc +IPP conditions, relative to a loading control. We have also repeated the growth assay and RT-qPCR experiment (in biological triplicate) under IPP-rescue conditions. Parasites samples harvested on day 3 of the IPP-rescue assay were analyzed by RT-qPCR and show reduced PPS mRNA abundance that is similar to (and slightly lower than) that observed without IPP supplementation. This similarity is not surprising to us, since the day 3 harvest in the original growth assay (without IPP) was 3 days before observing a parasite growth defect in -aTc conditions. With respect to the mechanism of transcript loss in the aptamer/TetR-DOZI system, the fate of transcripts in this system has not been investigated in depth. However, DOZI is believed to target bound mRNA to P-bodies, which are a known site of mRNA degradation in cells. We have unpublished data with multiple parasite proteins tagged with the aptamer/TetR-DOZI system. In all cases, we see strong reductions in mRNA abundance in -aTc conditions, suggesting that such decreases are a general property of this knockdown system.

Line 342 "These results directly suggest that apicoplast biogenesis specifically requires synthesis of linear polyprenols containing three or more prenyl groups." - I think that this might be overinterpreting those results - there could be a number of different reasons why polyprenols of different sizes do or don't rescue, including different solubility, diffusion, availability of transporters, predisposition to break down to useable subunits. Perhaps this needs a caveat.

We have modified the text here to remove “directly” and to acknowledge uncertainty in beta-carotene uptake: “Although it is possible that β-carotene is not taken up efficiently into the apicoplast, rescue by decaprenol, which is similar in size and hydrophobicity to β-carotene, suggests that apicoplast biogenesis specifically requires synthesis of linear polyprenols containing three or more prenyl groups.” We have also added the statement that “this hypothesis is further supported by additional results described in the next two sections”, referring to our identification of an apicoplast-targeted polyprenyl synthase.

Line 361 " the cytosolic enzyme, PF3D7_1128400" - I don't think we know the localisation of this protein based on the published data. The Gabriel et al study makes it clear the protein isn't apicoplast or mitochondrial, but it is punctate at stages in a pattern that doesn't look to me to be a straightforward cytosolic localisation (and the original authors don't describe it as cytosolic).

We agree that the localization of PF3D7_1128400 requires further investigation. The Gabriel study, which (surprisingly) is the only study we found that has examined localization of this protein by microscopy, observed diffuse signal in trophozoites consistent with cytoplasmic localization, in additional to focal, punctate signals in schizonts that were distinct from the apicoplast or mitochondrion. The authors described their results as, “Analysis by fluorescence microscopy of live parasites confirms expression along the intra-erythrocytic cycle and shows FPPS/GGPPS localization throughout the cytoplasm and also forming spots, which increase in number as parasites mature from trophozoite to schizont stages.” For simplicity we referred to FPPS/GGPPS localization as cytoplasmic but agree that available data suggest more a complex localization that requires further studies to understand. We have modified the text to indicate that available data suggests a complex cellular distribution that includes both the cytoplasm and additional sub-cellular foci outside the apicoplast and mitochondrion.

Line 423 "with strong prediction of an apicoplast-targeting transit peptide but uncertainty in the presence of a signal peptide". I don't think this describes well the bioinformatic analysis of the N-terminus. Although the experimental data are convincing that this is an apicoplast-targeted protein, bioinformatically this would not be predicted as an apicoplast protein. There is no obvious signal peptide, and "uncertainty" is too vague a descriptor. None of the versions of signalP, nor psort, predict this as possessing a signal peptide (which by definition means that PlasmoAP absolutely rejects it), and there is no obvious hydrophobic segment at the N-terminus that we would normally expect of a signal peptide. The toxoplasma hyperlopit doesn't suggest that the Toxoplasma orthologue is apicoplast, and the protein isn't found in the Boucher et al apicoplast proteome. This is somewhat of a mystery. It doesn't diminish the solid localisation data, with the excellent complementary data from IFA as well as the doxycycline+IPP experiment, but it should be pointed out clearly that this localisation isn't to be expected from the sequence analysis.

We thank the reviewer for this perspective and agree that SignalP is unable to identify a signal peptide at the N-terminus of PPS. We have modified the text to remove our description of “uncertainty” and explicitly state that SignalP is unable to identify a canonical signal peptide at the N-terminus of PPS.

We note that multiple proteins detected in the Boucher et al. apicoplast proteome also lack an identifiable signal peptide by SignalP yet are clearly imported into the apicoplast. These proteins include the key MEP pathway enzymes DXR (PF3D7_1467300) and IspD (PF3D7_0106900), holo ACP synthase (PF3D7_0420200), FabB/F (PF3D7_0626300), and the E1 subunit of pyruvate dehydrogenase (PF3D7_1446400). Thus, apicoplast import despite lack of identifiable signal peptide by SignalP is not unique to PPS but general to multiple (if not numerous) apicoplast-targeted proteins. These observations suggest to us that protein N-termini in Plasmodium can have sequence properties compatible with ER targeting that are broader and more heterogenous than other eukaryotic organisms that comprise the training sets upon which SignalP is currently based. It remains a future challenge to fully understand these properties.

With respect to the lack of PPS detection in the Boucher et al. apicoplast proteome, PPS appears to have a very low expression level and unusual solubility properties that require overnight extraction of parasite pellets in 2% SDS (or LDS) for detection. In our experience, the RIPA extraction conditions (which contained 0.1% SDS) used in the Boucher et al. study are insufficient to solubilize PPS, which may explain lack of PPS detection in their study.

To explicitly address these questions regarding PPS targeting to the apicoplast, we have added a new section to the Discussion to explore PPS targeting in the absence of a recognizable signal peptide, its unusual solubility properties and lack of detection in the Boucher et al. proteome, and planned future studies to further test, refine, and understand targeting determinants.

With respect to Toxoplasma, T. gondii appears to also express two polyprenyl synthase homologs, TGME49_224490 and TGME49_269430, that are ~30% identical (in homologous regions) to PF3D7_1128400 (FPPS/GGPPS) and PF3D7_0202700 (PPS), respectively. TGME49_224490 appears to be targeted to the mitochondrion in T. gondii (based on MitoProt and HyperLOPIT analysis), in contrast to its P. falciparum homolog, PF3D7_1128400, which localizes to the cytoplasm and other cellular foci outside the mitochondrion. TGME49_269430 does not appear to target the apicoplast in T. gondii (based on HyperLOPIT data), which contrasts with our determination of apicoplast targeting for the P. falciparum homolog, PF3D7_0202700. These differing localizations may suggest distinct cellular roles for these homologs in T. gondii compared to P. falciparum. We are also aware of a recent study (Pubmed 34896149) showing that loss of MEP pathway activity in T. gondii (due to loss of apicoplast ferredoxin) does not impact apicoplast biogenesis, in contrast to our observations in P. falciparum based on FOS treatment, DXS deletion, and PPS knockdown. These distinct phenotypes further suggest differences in isoprenoid utilization and metabolism between T. gondii and P. falciparum that remain to be understood. We have added a new section to the Discussion to address these considerations.

The section after line 344 "Iterative condensation of DMAPP with IP…", up until line 377 doesn't sit well within the section that has the heading "Apicoplast biogenesis requires polyprenyl isoprenoid synthesis". I suggest either creating a separate subheading for this material, or moving it into the start of the subsequent section "Localization of an annotated polyprenyl synthase to the apicoplast.".

We thank the reviewer for this suggestion, which we have followed. We have moved the referenced text to the beginning of the subsequent section to better align the text with that section heading.

Reviewer #2:

Minor comments:

The authors emphasize that this study reveals a previously unnoted interconnection between apicoplast maintenance and pathways that produce an output from the apicoplast to serve the cell. But is the prevailing view really that these two are separate? Isn't the interconnection already clear from many other studies and observations? E.g., the fatty acids produced inside the apicoplast provide membrane- and lipid- precursors for the rest of the cell as well as for the apicoplast itself (Botte et al., PNAS, 2013) (although not essential in Plasmodium blood stages). Other pathways that function inside the apicoplast such as the Fe-S cluster synthesis are critical to support enzymes that provide exported metabolites (e.g., IPP synthesis, IspG/H) and function in maintenance (e.g., MiaB) (Gisselberg et al., PLoSPath, 2013). Perhaps the authors could tone this conclusion down and acknowledge that maintenance and output are interconnected in other cases, which have been acknowledged in the literature.

We thank the reviewer for this perspective and agree that in Toxoplasma as well as in mosquito- and liver-stage Plasmodium there are multiple apicoplast outputs (i.e., metabolic products exported from the apicoplast) that contribute to parasite fitness, including IPP, fatty acids, and coproporphyrinogen III. To clarify, we are specifically referring to blood-stage Plasmodium in our manuscript, when heme and fatty acid synthesis are dispensable and where the prior literature has intensely focused on IPP as the key essential output of the blood-stage apicoplast and consistently stated that IPP is not required for organelle maintenance.

We agree that prior work has firmly established that apicoplast housekeeping functions (e.g., synthesis of proteins and Fe-S clusters) are required for organelle maintenance and to support IPP synthesis. However, our work is the first to demonstrate in blood-stage Plasmodium that the reverse is also true- that IPP as an essential apicoplast output is also required for organelle maintenance and that apicoplast maintenance and IPP synthesis are thus reciprocally dependent. We have modified the Discussion section to clarify these points and to explicitly acknowledge that apicoplast maintenance and other metabolic outputs may also be interdependent in Toxoplasma and other Plasmodium stages.

Could the authors elaborate more on the leader sequence predicting apicoplast localization for the PPS characterized here and discuss why it might have been missed in previous detailed study of apicoplast localised proteins (Boucher et al., PlosBiol, 2018)?

Please see our response above to Reviewer #1.

Could the authors discuss conservation of the PPS gene(s) in other Apicomplexa with (e.g., T. gondii) and without (e.g., Cryptosporidium spp.) an apicoplast? This could be relevant for other people in the field and could give further insights into the enzyme's role in apicoplast maintenance.

Please see our response above to Reviewer #1. Polyprenyl synthases are diverse enzymes that perform a variety of cellular functions, whose specific roles can differ between organisms. Although the two Plasmodium prenyl synthases show preferential homology with each of two different prenyl synthase homologs in Toxoplasma and Cryptosporidium (CPATCC_003578 and CPATCC_001801), the differing localizations of these homologs in each parasite suggest differing cellular roles. The differing dependence of apicoplast biogenesis on MEP pathway activity in T. gondii and P. falciparum and the absence of an apicoplast in Cryptosporidium further support differences in isoprenoid utilization and metabolism in these organisms. We have added a new section to the Discussion to address these considerations.

Reviewer #3:

The paper is very nicely written and was a true pleasure to read. The introduction is concise yet dense with all relevant background of our current understanding of functioning of the apicoplast in relation to IPP production and utilization. The rational of the experiments and the interpretation of the results are presented clearly and everything is discussed well in the context of the current understanding of the field. The main conclusion of the paper that isoprenoid is not solely essential for critical functions elsewhere in the cell, such as prenylation-dependent vesicular trafficking but also for apicoplast biogenesis via its processing by an essential polyprenyl synthase conserved with plants and bacteria is well substantiated and very exciting. The authors demonstrate an equally beautiful and clever use of available and newly generated genetic mutants in combination with complementary pharmacological interventions and metabolic supplementation. There are no true major weaknesses that could jeopardize the conclusions or change the interpretation of the results. However, the authors do consistently perform statistical analyses on data obtained from individual cells obtained in no more than two independent experiments, which in my humble opinion does not qualify for statistical analysis. That said, the results are so clear-cut that no statistics are required to convince me, or to quote Ernest Rutherford: '"If your experiment needs statistics, you ought to have done a better experiment."

We thank the reviewer for these positive comments and suggestions. For growth assays, we have performed a third biological replicate and updated those figures and the indicated statistical analyses. For microscopy experiments, we have removed p values.

Reviewer #1 (Public Review): 

The experiments are well designed, generally well controlled, and carefully conducted, and are thoughtfully and appropriately discussed. The authors make conclusions that are well supported by their results. 

When describing the aptamer knockdown of the PPS, the authors explain that the western blot was too noisy for monitoring the knockdown, which is frustrating for the reader and must have been frustrating for the authors. The authors instead counter-intuitively use qRT-PCR to monitor the transcript abundance of the PPS transcript in the aptamer system - this aptamer system is thought to be a modifier of protein, not transcription or transcript abundance. The authors describe that this has been seen once before (using aptamer knockdown of PfFis1), and the authors of that study speculate that the TetR-DOZI aptamer might be degrading the target mRNA. This is a plausible explanation, but it isn't quite clear from the description how this experiment was performed. The authors explain that the knockdown parasites grew normally for three days, but the parasites may be becoming sicker over this period. It's therefore possible that the decrease in PPS mRNA abundance is a product, rather than a cause of the growth defect. Sick or dying parasites could plausibly impact the PPS differently to the two chosen controls, particularly since both control genes chosen have substantially longer half-lives than the PPS mRNA (according to the Shock and DeRisi datasets). I therefore I suggest that this experiment be performed in an IPP rescue scenario (where the parasites aren't dying) with biological replicates. There is no explanation of the replicates here, but the error bars in 6C are implausibly small for real biological replicates. 

Line 342 "These results directly suggest that apicoplast biogenesis specifically requires synthesis of linear polyprenols containing three or more prenyl groups." - I think that this might be overinterpreting those results - there could be a number of different reasons why polyprenols of different sizes do or don't rescue, including different solubility, diffusion, availability of transporters, predisposition to break down to useable subunits. Perhaps this needs a caveat. 

Line 361 " the cytosolic enzyme, PF3D7_1128400" - I don't think we know the localisation of this protein based on the published data. The Gabriel et al study makes it clear the protein isn't apicoplast or mitochondrial, but it is punctate at stages in a pattern that doesn't look to me to be a straightforward cytosolic localisation (and the original authors don't describe it as cytosolic). 

Line 423 "with strong prediction of an apicoplast-targeting transit peptide but uncertainty in the presence of a signal peptide". I don't think this describes well the bioinformatic analysis of the N-terminus. Although the experimental data are convincing that this is an apicoplast-targeted protein, bioinformatically this would not be predicted as an apicoplast protein. There is no obvious signal peptide, and "uncertainty" is too vague a descriptor. None of the versions of signalP, nor psort, predict this as possessing a signal peptide (which by definition means that PlasmoAP absolutely rejects it), and there is no obvious hydrophobic segment at the N-terminus that we would normally expect of a signal peptide. The toxoplasma hyperlopit doesn't suggest that the Toxoplasma orthologue is apicoplast, and the protein isn't found in the Boucher et al apicoplast proteome. This is somewhat of a mystery. It doesn't diminish the solid localisation data, with the excellent complementary data from IFA as well as the doxycycline+IPP experiment, but it should be pointed out clearly that this localisation isn't to be expected from the sequence analysis. 

The section after line 344 "Iterative condensation of DMAPP with IP...", up until line 377 doesn't sit well within the section that has the heading "Apicoplast biogenesis requires polyprenyl isoprenoid synthesis". I suggest either creating a separate subheading for this material, or moving it into the start of the subsequent section "Localization of an annotated polyprenyl synthase to the apicoplast.".

Author Response:

Reviewer #1:

Significance: A central puzzle in evolutionary biology (and philosophy of biology) is the evolution of new (collective) entities that can evolve on their own right (e.g. the evolution of multicellular organisms from single cells). These evolutionary transitions are often conceptualized in terms of fitness decoupling (a fitness increase of the collective even as the fitness of the component particles decreases). Using a life-history model, the authors show that fitness decoupling is not possible when the conditions for fitness are the same. Thus, this paper has the potential to change how we think about the evolution of new collective entities.

Strengths: This paper is conceptually rich and the overall argument is clear. Re-analyzing previous data/models using their new framework highlights new patterns of fitness change in these transitions of individuality, and as such, it provides novel and exciting avenues of research.

Weaknesses: While the overall argument is clear, some of the details can be hard to follow (even as someone familiar with the literature). The initial description of their model is fairly clear, but given its conceptual novelty, the paper does not spend enough time developing the different concepts of fitness at the particle level.

Moreover, it is not entirely clear what is at stake: what is the role of fitness decoupling in our understanding of fitness transitions? And how does the proposed mechanistic ("trade-off breaking") model serve as a replacement? It seems to me like trade-off breaking is a characteristic of many evolutionary innovations, not only of major transitions. It seems even possible to envision groups that allow for an escape in a trade-off without leading to the evolution of a new "Darwinian" individual.

For example, one could conceive of a trade-off in zebras between time spent foraging and protection against predators. Coming together temporarily as a group is likely to allow for values outside this trade-off space (similar to those in Fig. 6). One could even imagine a new mutation that makes zebras switch activities (foraging/watching) depending on their position within the group. This mutation is only available to zebras that form groups (the phenotype does not exist in the absence of a group). But I would still want to argue that there is more to the evolution of new levels of individuality. Trade-off breaking seems (potentially) a necessary, but not sufficient step in these transitions.

And while the language of the authors is careful to not suggest sufficiency, it is not entirely clear how this approach helps us understand the particularity of these transitions.

Reviewer #1 asks first to clarify the stakes: what is the role of fitness decoupling in the explanation and how does tradeoff-breaking replace or supplement it? Second, they requested us to make a statement about the necessary or sufficient nature of tradeoff-breaking.

With respect to the second point, we argue that tradeoff breaking is not sufficient, but is probably necessary for an ETI to occur.

Let us now clarify the role of fitness decoupling and tradeoff breaking in the explanation of ETIs. It must be stressed that tradeoff breaking does not “replace” fitness decoupling; rather, tradeoff breaking is an event that cannot be understood readily in the framework of fitness decoupling. Thus, we claim that ETIs are better understood when seen through the lenses of traits and the evolutionary constraints that link them (i.e., tradeoffs) than via the export-of-fitness model (i.e., fitness decoupling). To illustrate this, we use the zebra herd example proposed by the reviewer. Coming together temporarily as a collective does not, in itself, constitute a tradeoff-breaking event, but rather simply a collective-formation event (similar to the first ace2 mutation in snowflake yeast or the first WS mutation in the Pseudomonas system). From this starting point, a number of mutations (i.e., change in traits values) can be fixed in the population that improve the performance of zebras within this environment. This is the “fast” part of the evolutionary trajectory that occurs on the ancestral tradeoff, which we called “low hanging fruit mutations” in the manuscript. As a consequence, “optimal herds within the ancestral tradeoff” evolve. As stated in the manuscript, if we assume that the tradeoff on traits is identical for lone zebra and zebra herd and also assume that the ancestral lone zebra exhibit trait values that are optimal (within these constraints) for lone zebras, it follows that the low-hanging fruit mutations that improve the zebra herd will probably reduce counterfactual fitness. This lowering of counterfactual fitness is not due to a “transfer” between real and counterfactual fitness (because there is nothing to transfer between real and counterfactual worlds), but is a consequence of the differential contribution of the traits involved in the tradeoff to the two fitness quantities. However, this specificity of the tradeoff might be significant because it could lead to stabilisation of the new collectives through ratchetting.

There is, indeed, “more to the evolution of new levels of individuality,” as pointed out by Reviewer #1. We claim that it involves rare mutations that would overcome the ancestral constraint and call them “tradeoff breaking mutations”. Tradeoff-breaking mutations are not bound by ancestral tradeoff; therefore, there is no a priori theoretical or biological reason to think they would have any positive or negative effect on counterfactual fitness. Here, we must stop using the zebra herd example because no tradeoff-breaking mutation occurred. However, the tradeoff-breaking lineages in the Pseudomonas example exhibit an improvement of both counterfactual and within-collective fitness. This observation does not fit within an export-of-fitness framework, but makes perfect sense in a traits-based view of ETIs—as a tradeoff-breaking mutation.

Reviewer #2:

This work reviews the influential "fitness decoupling" heuristic for understanding evolutionary transitions in individuality (ETIs), describes some of its limitations, and clarifies its interpretation. The review of the fitness decoupling account capably describes an interpretation of this framework that has frequently occurred in the literature, for example in Okasha 2006, Godfrey-Smith 2011, Hammerschmidt et al. 2014, Black et al. 2019, and Rose et al. 2020. However, it does not address the interpretation advanced by its authors, Richard Michod and colleagues, which they have clarified in several papers cited in the present work. Michod and colleagues have argued that the fitness decoupling account describes a changing relationship between the fitness of groups and the "counterfactual" fitness of their component cells, that is, the fitness the cells would have if they were removed from the group. This point is made explicitly in Shelton & Michod 2104 and Shelton & Michod 2020 and was present (though perhaps not as obvious) in Michod 2005 and later works, in contrast to the claim in the Glossary that this is a "relatively recent development of the fitness decoupling literature." The interpretation that Michod embraces is similar to what is here described as f2, the fitness of a "theoretical mono-particle collective", but that interpretation is not mentioned in the present work until Section 2.3. It is possible that an argument could be made that Michod and colleagues have not consistently interpreted fitness decoupling this way, or have made statements inconsistent with this interpretation, but no such argument is present in this work. Thus the impression conveyed is that Michod and colleagues consider decoupling of "commensurably computed fitnesses" possible, which is counter to their explicit statements on the topic.

The description of the limitations of the fitness decoupling heuristic (Section 2) is useful and goes a considerable distance toward clarifying the ways in which fitness decoupling can rigorously be interpreted. However, the final assessment (Section 2.3) does not make a compelling case for its central argument, the lack of utility of the fitness decoupling concept. Elsewhere in the work, the ratcheting model of Libby and colleagues is referenced in comparison to the tradeoff-breaking approach, but Section 2.3 does not acknowledge the relationship between Libby and colleagues' model and the counterfactual interpretation of the fitness decoupling heuristic. For example, the argument in Libby and Ratcliff 2014 that "If any of the yeast that evolved high rates of apoptosis within clusters were to leave the group and revert to a unicellular lifestyle, they would find themselves at a competitive disadvantage relative to other, low-apoptosis unicellular strains." and in Libby et al. 2016 that "…if G cells were to revert to unicellular I cells, they would be quickly outcompeted" are counterfactual fitness arguments essentially similar to that of Shelton and Michod 2020 that "the fitness a cell would have on its own declines as the transition progresses." Section 2 makes a convincing case that commensurable fitnesses cannot be decoupled, but by fixating on commensurability, which is not relevant to the counterfactual interpretation of fitness decoupling, Section 2.4 fails to make a convincing case that "fitness-decoupling observations do little to clarify the process of an ETI." That is, "because they are not commensurable" does little to explain why the counterfactual interpretation of fitness decoupling "does little on its own to clarify the process of an ETI," since commensurability is not a claim that the the counterfactual interpretation of fitness decoupling makes.

We agree with the reviewer on two essential points: (1) the decoupling of commensurably computed fitness is impossible when collectives have a finite size and (2) counterfactual fitness is not commensurable to particle or collective fitness.

While we recognise that Michod and collaborators did clarify that fitness decoupling referred to counterfactual fitness (although, to us, this becomes clear from 2015 onward), we argue that the fitness transfer (or export-of-fitness) metaphor implies (by its wording) a commensurability of fitnesses that undermine this welcome clarification.

Indeed, for a quantity to be transferred from one place—or component—to another, the source and destination must be commensurable. It is incorrect to talk about a transfer between counterfactual and actual quantities. A better choice of words to discuss the relative change of counterfactual and actual quantities would avoid the physical transfer metaphor and focus instead on the correlation of the two quantities. It must be noted that, despite the clarification of counterfactual fitness, the word “transfer” continues to be used in recent work (Davison & Michod, 2021).

This may seem like nitpicking; however, there is a real advantage in being careful about this. We do agree that, under some assumptions, counterfactual fitness would decrease while whole–life cycle particle fitness (or collective fitness) increases. From there, one might ask: what needs explaining? If one assumes an export-of-fitness framework, the transfer of fitness explains why it cannot be otherwise. If fitness decreases on one side, it must increase on the other. In other words, the existence of a tradeoff is taken for granted based on the improper physical metaphor. While there are strong reasons to think that such tradeoffs exist, they should be assessed in their own right and on a case-by-case basis rather than being assumed to hold. Otherwise, there is no way to make sense of the tradeoff-breaking scenario described in Section 4.

By the same token, the metaphor of “decoupling” often associated with the export-of-fitness model is misleading because it is used to describe a part of the evolutionary dynamics where counterfactual particle fitness and whole–life cycle particle fitness are strongly dependant on one another (even if their changes are anticorrelated), through the existence of the tradeoff.

Nevertheless, we welcome the reviewer’s urge to clarify our position and how this relates to Michod and colleagues’ counterfactual fitness proposal.

The model based on trade-offs and trade-off breaking is useful and likely to be of interest to theorists interested in ETIs. The observation that this model can reproduce the (counterfactual) fitness-decoupling observation is a useful in showing the how the two models relate. The result that counterfactual fitness decoupling is a consequence rather than a cause of the evolutionary dynamics is an important point (though perhaps obvious in retrospect, since counterfactuals, things to do not happen, can't be the causes of anything).

The caution in Section 3.3 that "the same [counterfactual fitness decoupling] observation will be made in any situation in which short-term costs are compensated by long-term benefits, not solely during ETIs" is a good point, and it sets up the argument that trade-off breaking is a "genuine marker for an ETI". However, no convincing case is made that the same criticism, that the observed phenomenon is not unique to ETIs, is not equally true of trade-off breaking. Some nice examples of trade-off breaking in the context of ETIs are given, but these do not amount to an argument that trade-off breaking is only observed during ETIs. The life history literature includes examples of trade-off breaking that are not related to ETIs, so it is not clear that trade-off breaking is either a reliable indicator of ETIs or superior in this respect to counterfactual fitness decoupling.

This point is in line with one of the points made by Reviewer #1. We have now clarified our position with respect to the generality of the tradeoff-breaking approach.

In the Discussion, the "inconveniences" associated with the fitness decoupling are cogent limitations of this heuristic. The "impossibility of decoupling between commensurable measures of fitness" is an important result, but it is not new and should thus probably not be presented as "[o]ur first main finding". Shelton and Michod 2014 includes a mathematical proof in the appendix that, given the model assumptions, "consideration of the births and deaths of colonies gives us exactly the same bottom line (fitness) as consideration of the births and deaths of lone cells." The second main finding, that "fitness decoupling observations cannot be reliably used as a marker for ETIs," is valid, but as described above, a convincing case is not made that trade-off breaking can be reliably used in this manner, either. Trade-off breaking may, however, be a useful way to think about ETIs in the other ways that are suggested, for example as key events and as stepping stones to new hypotheses.

We have now clarified our position.

I think here we are calling sensors/acuators/transducers as "applications". But this may be saying that the endpoint is where the data comes from, and not the hardware device. The question is, can application endpoints be at coordinator and router devices?

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

The paper tackles an important problem regarding the effect of demographic dependent vaccination protocols on the reduction in the number of deaths with respect to the situation of no vaccination (say J). A compartmental SIRD model with reinfection Y is proposed, stratified in two (age dependent) groups, based on a binary reduction of a given contact map, and given infection fatality risk (IFR). Several countries are then analyzed.

As far as I understand we have a control variable v, parameters of the stratified model (i=1,2) tuned to match IFRi, and a control objective, i.e. minimization of J over one year.

The paper is well written. The final message and some theoretical passages are not completely clear, at least to me. I have the following observations that the authors may want to consider.

We thank the referee for the revision and are very glad that the overall evaluation is positive. Comments and suggestions have been thoroughly addressed, as we discuss in the following.

1) The study of stability of infection free and endemic equilibria should be better developed. The 5 equations can be reduced to 4 (neglecting D) and the characteristic of the reduced Jacobian used to characterize the local asymptotic stability of equilibria, instability, bifurcation points etc... Alternatively, one can use a co-positive Lyapunov function (LF). For instance, if we take the LF V=S+I+Y+R, we get $\dot V=-\mu_I I-\mu_Y Y \le 0$. If $\mu_I$ and $\mu_y$ are strictly positive all equilibria are characterized by (S*,0 0,R*) and D=1-S*-R*. So, I don't understand the phrase after (7,8), notice that Y cannot be zero in finite time. For $\mu_y=0$ then Y* can be nonzero. I guess that closed-form computation of S* and R* is possible as function of the parameters at least in the case v=0. The stability result should be cast in function of the current reproduction number (not explicitated) wrt to S and R.

The authors are invited to have a look at

1.1) Pagliara et al, "Bistability and Resurgent Epidemics in Reinfection Models", IEEE CSLetters, 2018,

for a theoretical analysis of stability on a similar (just a little bit simpler) model.

We appreciate the suggestions of the referee for improvement of this material. We have carried out an in-depth revision of the stability analysis and significantly extended it. The major addition has been, as suggested, a section relating the current reproductive number at equilibrium (we call it the asymptotic reproductive number in the text) to the fixed points of the dynamics for three different scenarios: general model, no vaccination, and zero mortality of reinfected individuals. As Pagliara et al. show in their paper, the connection between the fixed points and the reproductive number is not trivial, but it is possible to derive it through the next-generation matrix technique, as we now do. Additional references regarding this technique have been added. We have included a Table summarizing the stability analysis (page 2 in SI 3) at the end of this new section.

Other modifications include the reduction of 5 equations to 4 for the stability analysis and a clarification of possible equilibria (page 1 of SI 3), rephrasing and correcting our sentence after eqs. (7) and (8). We also attempted to obtain a closed-form computation of S* and R* but, to the best of our knowledge, concluded that it is not possible. We would be happy to pursue any insight in this respect the referee may have.

What said before should be also extended to the stratified model, where a "network" Rt could be defined, see for instance

1.2) L. Stella et al, "The Role of Asymptomatic Infections in the COVID-19 Epidemic via Complex Networks and Stability Analysis", SIAM J Cont. Opt., 2021, (arxiv.org/pdf/2009.03649.pdf)

We thank the referee for pointing out this reference. Following the analysis in Stella et al., we have carried out a stability analysis for the stratified model as well. The results are included in a new section (pages 7-10 in the SI 3).

2) It is not clear whether the free contagion parameters of the model have been fitted on real data (identification from infection and reinfection data). Notice that the interplay between vaccination strategies and NPI is important, see e.g.

*2.1) Giordano et al, Modeling vaccination rollouts, SARS-CoV-2 variants and the requirement for non-pharmaceutical interventions in Italy", Nature Medicine 2021, *

where progressive vaccination in reverse age order is considered together with different enforced NPI countermeasures.

In the first part of our study, parameters are intendedly left free because we aim at describing the generic behavior of the model. Still, we derive several inequalities and relationships between parameter ratios that seem to be sensible attending to what the different classes in the model stand for. This is as described in sections regarding model parameters when the two generic models (SIYRD and S2IYRD) are introduced. The aim is to represent both the generic dependence with some variables and a broad class of contagious diseases, so parameters are mostly free. In agreement with this approach, parameters can be also freely varied in the companion webpage.

In the second part of our study, the model is applied to COVID-19. In that case, we have used parameter values in agreement with observations, as (admittedly poorly) explained in pages 9-10 of the main text. Indeed, not enough information on parameter estimation was provided in the main text, and the SI 2 also needed some additional information. This has been amended. Let us explicitly mention that we have not fitted the dynamics of the model to any actual data set to fix specific values, as Giordano et al. do. In our case, we have first used different demographic data sets to evaluate contact rates and IFRs of the two population groups (these are parameters Mij and Ni in eqs. (7-10)). Secondly, recovery and death rates are estimated through the IFRi values for each age group i and the infectious period of COVID-19, that we fix at dI=13 days. Third, infection rate βSI=R0/dI has been estimated fixing R0=1, since the reproductive number of COVID-19 all over the world fluctuates around this value (Arroyo-Marioli et al. (2020) Tracking R of COVID-19: A new real-time estimation using the Kalman filter, PLoS ONE 16(1):e0244474). The reinfection rate is defined through its relationship with the infection rate, βRI= α1 βSI, where α1 was in the range 0-0.011 at early COVID-19 stages (Murchu et al. (2022), Quantifying the risk of SARS‐CoV‐2 reinfection over time, Rev Med Virol 32:e2260) and seems to be about 3-4 fold larger for the omicron variant (Pulliam et al., Increased risk of SARS-CoV-2 reinfection associated with emergence of the Omicron variant in South Africa, www.medrxiv.org/content/10.1101/2021.11.11.21266068v2). Given the relationships derived among parameters, our only free parameter was α2RY= α2 βRI, and we fixed it to α2=0.5 (i.e., reinfected individuals recover twice as fast as individuals infected for the first time).

Once more, it was not our goal to precisely recover specific trajectories of COVID-19 or to point at possible future scenarios, but to illustrate the dependence of major trends with model parameters. Also, the appearance of new variants requires the reevaluation of parameters. For example, omicron has different IFR (therefore different mortality and recovery rates), a different infectious period, and higher infection and reinfection rates. In this context, the interactive webpage (where we will update demographic profiles and IFR data as they become available) is a useful resource to simulate any situation different from current or past ones.

3) In the model the immunity waning is not explicitly considered (flux from R to S or better from a vaccinated compartment to S). It is clear that this complicates the model. Please discuss why the indirect way the waning is considered here is justified.

3.1) Batistela et al, "SIRSi compartmental model for COVID-19 pandemic with immunity loss", Chaos Soliton and fractals, 2021.

3.2) McMahon et al, "Reinfection with SARS-CoV-2: Discrete SIR (Susceptible, Infected,Recovered) Modeling Using Empirical Infection Data", JMIR Public health and surveillance, 2020.

Though the model does not consider an incoming flux of individuals to compartment S, the existence of a "backward" flux from R to Y yields a transient phenomenology analogous to models with increases in the S class. Indeed, it is these fluxes that cause persistent endemic states; otherwise, the S class is monotonously depleted until infection extinction.

In Batistela's et al. work, the possibility that individuals become reinfected is effectively implemented through a flux between the R and S classes, since only one class of infected individuals is considered and recovered individuals cannot be infected again. In our case, feeding back to S would mean that previous immunity is completely lost or that vaccines are not effective at all for some individuals. This is neither what McMahon et al. conclude when evaluating real data nor what more recent surveys indicate (see for instance the Science Brief published in October 2021 by the CDC, SARS-CoV-2 Infection-induced and Vaccine-induced Immunity, https://www.cdc.gov/coronavirus/2019-ncov/science/science-briefs/vaccine-induced-immunity.html).

This nonetheless, complete immunity waning (feedback to the S class) and reinfections (feedback to a partly immune class experiencing overall lower severity of the disease) are equivalent to a large extent: the trend of COVID-19 seems to indicate that our Y class will be the "new S", and that fully naive individuals would arrive mostly due to demographic dynamics (birth and death processes, as also implemented by Batistela et al.). Summarizing, complete immunity waning is rare in the time scales considered in our simulations, while partial immunity that decreases the severity of the disease (after infection or vaccination) is the rule, in agreement with our choices.

4) Reduction of deaths wrt no vaccination is of course important, but also reduction of stress in hospitals. This is particularly important now with the advent in Europe of the omicron variant. Please discuss on the real message you want to convey to policy makers in the actual scenario of the pandemic.

The model in this work is deliberately simple. Our main goal was to explore the qualitative effects of demographic structure and disease parameters in protocols for vaccine administration. This was the reason to consider a mean-field model in a population structured into two groups. The main conclusion is that optimal vaccination protocols are demography- and disease-dependent. If this is so in our streamlined model, the more it will be in more realistic models, where one should include a finer stratification and, in all likelihood, heterogeneity in contagions. Our main message, therefore, is that there is no unique protocol for vaccine roll-out, valid for all populations and diseases. The abstract has been modified to highlight this conclusion.

Some qualitative considerations also allow us to draw preliminary conclusions on the reduction of stress in hospitals. Since the number of hospital admissions is proportional to the incidence of the disease, the number H of hospitalized individuals can be represented as H=a I + b Y, with a>>b due to the partial immunity of vaccinated or recovered individuals (which belong to class Y upon (secondary) contagion). Therefore, minimizing the burden on the healthcare system amounts to minimizing the number of individuals in the I class. Beyond non-pharmaceutical measures, I is minimized when individuals are transferred as fast as possible to the Y class, that is, maximizing vaccine supply and acceptance. In terms of our model parameters, this entails maximizing v and also θ (the maximum fraction of individuals eventually vaccinated), for instace through devoted awareness campaigns. These ideas have been included in the Discussion section.

Reviewer #1 (Significance (Required)):

The final message and some theoretical passages are not completely clear, at least to me.

Please discuss on the real message you want to convey to policy makers in the actual scenario of the pandemic.

As discussed above, we have modified the manuscript following the advice given by the Reviewer. We think that both the presentation and the theory are clearer now.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

In this paper, a compartmental model of the propagation of an infection with vaccination and reinfection is studied. The impact that changes in the rates of these two processes have on disease progression and on the number of deaths is analyzed. In order to highlight the overall effect of the demographic structure of populations and the propagation of a given disease among different groups, the population is divided into two subpopulations and the model is extended to the two-dimensional case. In addition to the study of equilibria and their relative stability, the model is then applied in the case of COVID-19. Different vaccination strategies are studied using real demographic data and with a population split between under 80 and over 80 individuals. It is observed that for low vaccination rates, the advisable strategy is to vaccinate the most vulnerable group first, in contrast to the case of sufficiently high rates, where it is appropriate to vaccinate the most connected group first. The simulations show also that with a low fatality ratio, the strategy that yields the greatest reduction in deaths is vaccination of the group with the most contacts, while the situation is reversed for higher fatality ratio.

The model and simulations presented are interesting and valuable. The comparison of the behavior of the model in the 4 different countries is very interesting, as well as the webpage created by the authors.

We thank the referee for the very positive evaluation and are very glad that the study is found interesting and valuable.

As minor comment, I think that the introduction of the model needs a more extensive literature review. For example, there is no mention of the classic SIR model of Kermack and McKendrick (1927) and other works on the introduction to epidemic models, which form the basis of the model presented by the authors.

The referee is right. There is a long history of extensions and applications since Kermack & McKendrick introduced the SIR model that we obviated. This has been amended by adding an introductory paragraph with several new references at the beginning of the Models section, page 3 in the main text.

Reviewer #2 (Significance (Required)):

The model presented by the authors is quite original and simple enough to be suitable to different contexts and scenarios.

Compared to previous work, this paper makes a twofold contribution, as explained by the authors. First, the introduction of reinfections shows the existence of long transients (or quasi-endemic states) that may precede the transition to a truly endemic state predicted for COVID-19. Second, the simplicity of model allows the characterization of systematic effects due to, at least, group size, demographic composition, and IFRs.

I am involved in the study and analysis of epidemic models accompanied by network effects. I think this paper is a good contribution, although preliminary, in the analysis of the vaccination process and in the search for the optimal strategy.

We thank the Reviewer and are glad that our goal, offering a model as simple as possible to obtain meaningful conclusions, is appreciated.

Author Response:

Reviewer #1 (Public Review):

In this paper, Qin et al. investigated the molecular mechanism of phospholamban (PLN) linked dilated cardiomyopathy (DCM), using structural approaches combined with biophysical measurements. Structures of the catalytic domain of protein kinase A (PKAc) in complex with PLN peptides (both wild-type and the R9C and A11E DCM mutants) provide insights into the mechanism of substrate recruitment and how it is perturbed in the disease state. Qin et al. show convincingly that the mutant peptides all have lower affinity for PKA than the wild-type peptide, suggesting models in which heterozygous DCM mutations act via sequestering PKA and thereby preventing phosphorylation of the wild-type peptide may be incorrect.

The authors highlight significant differences between their structure of the WT-PLN:PKAc complex, which has a 1:1 stoichiometry, and a previous structure of the complex (PDB 3O7L), which has 1 PLN bound between two PKAc monomers (a 1:2 complex). The authors posit that the stoichiometry observed in 3O7L is an artifact of the crystal lattice, and does not occur in solution, supporting this with analysis of the elution volumes of the peptide complexes on size exclusion chromatography compared to PKAc alone. They further suggest that the AMP-PNP ligand included in the 3O7L structure is not bound, based on analysis of Fo-Fc maps calculated from the deposited coordinates. Inspecting 3O7L I am not convinced of this last point - it seems more likely that a technical error was made in assigning or refining the B-factor of the ligand in 3O7L, because there is clearly density present in SA-omit maps for the nucleotide.

Taking these results together, the authors suggest a mechanism for DCM, whereby mutations in PLN result in lower affinity for PKA, and consequently reduced phosphorylation. This seems plausible and well supported by the data, although in the ADP-Glo assay used here, the reductions in phosphorylation observed for some of the mutant peptides are rather modest. However, as the authors state, it is plausible that even relatively subtle changes in PLN phosphorylation could have substantial effects on Ca2+ homeostasis via increasing SERCA inhibition.

We thank the reviewer for the appreciation of our work.

Reviewer #2 (Public Review):

Strengths:

The authors presented new high-resolution 3D crystal structures of the PKA catalytic domain (PKAc) in complex with PLN WT or mutant peptides (residues 8-22) containing the DCM-associated PLN mutations (R9C or A11E). These are novel and important data given that the present structures are dramatically different from those reported previously. The authors made convincing argument that the 3D model reported previously may result from a crystallization artifact.

By characterizing the interactions between the PKAc domain and PLN WT or DCM-associated mutant peptides using surface plasmon resonance (SPR) analysis, the authors convincingly showed that the DCM-associated PLN mutations at positions 9, 14, and 18 alter the conformation of the PLN peptide and reduce the binding affinity of the PLN peptide with PKAc. These data provide an explanation how some DCM-associated PLN mutations at these positions reduce the level of PKA-dependent phosphorylation of PLN.

The authors also performed nuclear magnetic resonance (NMR) to determine the structural dynamics of PLN WT, R9C, P-Ser16, and P-Thr-17 peptides. These NMR structures combined with the SPR analysis also support their conclusion that PLN phosphorylation and DCM-associated PLN mutations have an impact on its conformation.

We thank the reviewer for the comments.

Weakness:

The present study used PLN-derived peptides (aa 8-22). Although technically challenging, it is important to consider if the full-length WT or mutant PLN will behave the same as those observed with the peptides. This is especially crucial in light of the prior work showing substantially different structures using a different segment of PLN.

We are fully aware of the potential risk to draw conclusion from an isolated peptide instead of the full-length PLN as a transmembrane protein. In the previous study, people showed that the PLN peptide could be used as a good model substrate that gets phosphorylated as efficiently as the full-length PLN protein (L. R. Masterson et al., Dynamics connect substrate recognition to catalysis in protein kinase A. Nat Chem Biol 6, 821-828 (2010); D. K. Ceholski, C. A. Trieber, C. F. Holmes, H. S. Young, Lethal, hereditary mutants of phospholamban elude phosphorylation by protein kinase A. The Journal of biological chemistry 287, 26596-26605 (2012)). These results together with our biochemistry results suggest the tail peptides are indeed active substrates of PKA. Due to the technical difficulty, we were not able to crystallize PKAc in complex with the full-length PLN. To explain the potential difference between the peptides and the full-length PLNs, we added more text in the discussion section “Additionally, the trend of the reduced phosphorylation by DCM mutations can be significantly affected by the oligomerization state of PLN. Ceholski et al. showed that R9C severely inhibits PKA phosphorylation in the context of full-length pentameric PLN, but has a much milder effect in the context of full-length monomeric PLN or an isolated tail peptide [41].”

Although it is convincing that DCM-associated PLN mutations likely reduce the interaction between PKAc and PLN (assuming that the peptides behave the same as the full-length PLN with respect to interaction with PKA) and, as a result, the PKA dependent phosphorylation of the mutant PLN, it is unclear how this impaired interaction between PKA and PLN mutant could explain the effects of the DCM-associated PLN mutations on SERCA function (either reduced or enhanced PLN-dependent inhibition of SERCA, as proposed previously). In this regard, can the authors predict if the DCM-associated PLN R9C mutation reduces or increases SERCA inhibition based on the results of their present study?

It is indeed controversial how PLN mutations cause DCM. Previous studies have shown that the DCM mutations in PLN might change this regulation in either a phosphorylation-dependent or phosphorylation-independent manner. Our results show that the mutations may act through both manners: 1) the mutations reduce the phosphorylation level of PLN, which has been shown to enhance the inhibition of SERCA and inhibit the uptake of Ca2+; 2) the mutations change the conformation of PLN before binding to PKA or SERCA, which could have additional consequences, such as altered assembly state of PLN, phosphorylation of PLN by CaMKII, or changes in interactions of PLN with the lipid membrane. This could impact in either directions, reducing or increasing SERCA inhibition, which is difficult to predict based on our data. We added the explanation in the discussion “While decreased PLN phosphorylation is likely an important contributor to the physiological dysfunction associated with familial DCM, disease-causing mutations in PLN may have additional consequences, such as altered assembly state of PLN, phosphorylation of PLN by CaMKII, or changes in interactions of PLN with the lipid membrane. The influence of such factors on SERCA inhibition are unclear. In principle, they might further increase inhibition of SERCA and act in conjunction with lower PKA-mediated phosphorylation to manifest the disease symptoms. Conversely, it is possible that these factors could decrease the inhibition of SERCA, partially compensating for the decreased phosphorylation level, and mitigating the symptoms.”

It is also unclear how reduced PKA phosphorylation of mutant PLN could lead to DCM. PLN is unlikely to be significantly phosphorylated by PKA at rest (in other words, PLN is likely to be phosphorylated by PKA during stress, i.e. during the adrenergic fight-or-flight response). Therefore, it is puzzling how such reduced PKA-dependent phosphorylation of PLN would significantly affect the PLN function during the absence of flight-or-flight response.

As explained above, we think that this regulation could be through both phosphorylation-dependent and phosphorylation-independent manner. Even only considering the phosphorylation-dependent manner, the DCM phenotype could be due to an accumulation of the Ca2+ imbalance in the cell over repeated cycles of cardiac muscle contraction upon chronic accumulation of the sporadic phosphorylation events. It is also possible that the mutations affect the CaMKII-dependent regulation of PLN, which leads to DCM.

Given that the DCM-associated PLN mutations have significant effects on the conformation of PLN itself, at least in the form of short-peptides, it is possible that these mutations could affect the folding, oligomerization, trafficking, degradation, etc., in addition to PKA-dependent phosphorylation. The relevance and contribution of reduced PKA-dependent PLN phosphorylation to DCM remain unresolved.

We agree with the reviewers that both phosphorylation-dependent and phosphorylation-independent manners could contribute to the DCM disease phenotype. It remains unresolved which factor is the major contributor. We have added a statement in the discussion (see point above).

Reviewer #3 (Public Review):

This manuscript describes an elegant study utilizing the crystal structures for the elucidation of the disease mechanism of familial dilated cardiomyopathy. It has been known for decades that the mutations in PLN are associated with DCM, but the underlying mechanism remains controversial. In my opinion, Prof Yuchi and co-authors did excellent job on revealing the high-resolution crystal structures of PKA-phospholamban complexes, representing both the native and diseased states. Combined with various of biophysical and biochemical methods, including SPR, ADP-glo, thermal melts, NMR, etc, the authors systematically investigated the correlations between the PLN conformation, the binding affinity, and the phosphorylation level. The mechanism of PKA phosphorylation on another related substrate, ALN, was also convincingly revealed. The results are very helpful for understanding the pathological mechanism of PLN-related DCM. More importantly, the atomic structures of PKA-phospholamban complexes lay a solid foundation for the structure-based rational design of therapeutic molecules that can reverse the effects of the DCM-causing mutations in the future, e.g. by stabilizing the interactions between PLN and PKA.

We thank the reviewer for the appreciation of our work.

Author response:

Reviewer #2 (Public Review):

This work by Castledine et al. addresses the important question of whether results from in vitro (laboratory-based) evolution studies may be useful for predicting evolution during phage therapy in a clinical setting. In order to explore this question, the authors cultured a set of bacterial isolates from a patient pre- and during phage therapy, as well as phages from several time points during therapy. They then experimentally evolved (in vitro) a mixture of the bacterial isolates from the patient in the absence of phage, or in the presence of phage using two different treatments (phage added once or added repeatedly). Overall, they observed similarities between the evolutionary outcomes (genomic and phenotypic) in vitro and in the patient. Resistance evolved rapidly in the patient and in vitro under phage selection, and similar genomic changes were observed in both environments. The approach of using bacterial isolates directly from the patient (as well as the phages used for therapy) in vitro is clever, and the observed similarities are compelling.

We thank the reviewer for appreciating the novelty in our results and methodology.

However, I think there are some limitations with the study that should be addressed in the text.

In particular, (1) While the similarities in vitro and in the patient are quite interesting, there are some differences that were dismissed as being minor without justification. Calling the results "highly parallel" is a bit subjective - in vitro in the repeated phage treatment (which is suggested to be most similar to the clinical context), there did appear to be phage coevolution that was not observed in vivo. The tradeoffs/relationships between traits (as shown in Fig. 3) also differed to some extent.

We agree this could have been more objectively phrased at the start of the discussion – this has been edited to reflect this. We have highlighted the differences between in vivo and in vitro treatments with respect to phage evolution. Moreover, we have also highlighted that the observed trade-offs had different underlying mechanisms which may not always result in parallel evolutionary changes between in vivo and in vitro environments.

Additionally, for the genomic results only a subset of variants were plotted (those in genes of known function), but there were far more significant variants in genes of unknown function that were not included. It is difficult to assess whether the genomic findings are truly similar across environments if only a fraction of those results were presented in the manuscript.

We chose to concentrate on genes of only known function so that we could better understand their potential significance, and also because the figures and analyses (Figures 4 and 5) would become extremely complex and large and uninterpretable with genes of unknown function included. This is especially true for Figure 5, which would have required us to show 284 rows if all genes would have been included. Ultimately, whichever way we do this exploratory analysis, it is going to be difficult to see if findings are truly similar across environments because we only have a single patient who had phage therapy.

However, we have redone the analysis with all of the significant genetic changes (SNPs and indels from both known and unknown genes) included.

Figure 5 has been recreated and is now included as "Figure 5 - Figure supplement 1". All of the statistical analysis on (a) the number of SNP/indels seen (b) genetic distance from ancestor and (c) alpha diversity give quantitatively similar results. That is, although all the estimates are generally much higher after including many more genetic variants, all of the significant results from both the overall model fit and post-hoc multiple comparisons remain the same. One interesting result that came out of looking at all the genetic changes was that for genetic variants occurring in a gene of known function, 56% (28 out of 50) were de novo mutations, whereas this value was only 42% (98 out of 234) for variants in genes of unknown function.

We then looked at the proportion of genetic variants (both in known and unknown genes) found in vitro that were also found in vivo. For genes of known function, 62% of genetic variants were found in vivo (31 of 50) and this was comparable to the 65% of genetic variants in genes of unknown function (153 of 234). Of the 26 genes of known function with differences identified in the in vitro analysis, 16 (61%) were also found to have genetic changes in vivo. The equivalent metric for genes of unknown function was 86% (85 of 99). Similar to in vitro, variants occurring in a gene of known function were more likely to be de novo mutations (77%) compared to variants occurring in a gene of unknown function (46%).

While these patterns and exploratory analyses are interesting, they have extremely limited statistical power and therefore do not alter the conclusions or results of the work presented. For these reasons, we have chosen not to include these results in the already long manuscript. We have added a line to say we have done it both way:

“We performed all downstream statistical analyses on (a) only genetic variants in genes of known function and (b) all genetic variants.”

And we also added a line at the beginning of the genomic analysis results section:

“Results were not affected whether we included only genetic variants occurring in genes of known function or all genetic variants (Figure 5-Figure supplement 1). As we were interested in attributing potential functions to the variants identified, we only present the results for genetic variants occurring in genes of known function.”

(2) Much of the text is framed around whether in vitro outcomes are predictive of those in vivo, but this study only included results from a single patient. Thus, it is impossible to know whether these findings are by chance or representative of a more general relationship between in vitro and in vivo evolution.

We agree that having a single patient for our in vivo comparison limits the generalisability of our results. We have highlighted this in the revised manuscript. However, that our replicated in vitro experiments agreed broadly with our in vivo results and that of other studies (finding resistance-virulence trade-offs) suggests that at least in some circumstance in vitro dynamics are predictive of in vivo dynamics. Further studies are clearly needed (and hopefully will arise as a consequence of this work) to determine the generalisability of this finding and the circumstances where this parallelism might break down.

(3) Although the evolutionary outcomes appear to be similar, the pathogen was successfully cleared from the patient but persisted throughout experimental evolution. Whether the pathogen is successfully eliminated or not is presumably the most important clinical outcome, and while this difference is not surprising, it is an important one to point out to the reader. Essentially, evolution was similar to some extent but the consequences of evolution for bacterial persistence in each environment were quite different.

We have now highlighted this difference to the reader in the revised manuscript.

Author Response:

Reviewer #1 (Public Review):

• Line 141: It would be beneficial to better understand how the sequenced sample of the population corresponds to the PCR confirmed sample of the population, in order to understand possible selection biases in the sequence data. Could you elaborate on how the composition of sequence PCR confirmed cases matches the composition of PCR confirmed cases, by the demographic characteristics listed in Table 1.

Early in the pandemic (March-April), we tried to sequence every SARS-CoV-2 positive case diagnosed in our KWTRP laboratory from Coastal Kenya. However, with the sharp increase in the number of identified cases from the month of May 2020 onwards, and a limited in-house sequencing capacity, we changed strategy to sequence only a sub-sample of the identified positives. The criteria for sub-sampling included having a cycle threshold of < 30.0, spatial representation (at county level) and temporal representation (at month level). The consequent number and proportion of samples sequenced across the study period months and across the counties is summarized in Fig. 2C-E with the sample flow provided in Figure 2-figure supplement 1.

In the revised manuscript we have provided a comparison of the demographic characteristics of the sequenced cases versus non-sequenced cases (shown as Table 2). The participants providing the sequenced and non-sequenced positive samples had a similar gender distribution and similar probabilities of being from either from Wave one or Wave two. However, the distribution of sequenced vs non-sequenced cases differed significantly in age distribution, nationality and travel history. Specifically in the sequenced sample, there were more participants in 30–39 years age bracket compared to the non-sequenced samples, a disproportionately representation of non-Kenyan nationals and persons with a recent international travel history in the sequenced sample.

• Line 283: I am particularly interested in the observed inter county flows, but it is hard to interpret the numbers. Considering population sizes in each county, what are the phylogenetically observed import rates per 100,000? What are the rate ratios? Based on the observed data, is there any evidence that imports into coastal Kenya occurred statistically significantly through Mombasa?

We thank the reviewer for these comments.

In the revised manuscript we have added two new tables (1 & 4) which detail the population size in each of the six Coastal Kenya counties, population density and estimated import/export rates (per 100,000) for the counties.

The alluvial plots are descriptive regarding genome flows. The underlying data on the pattern of virus movement is inferred using the ancestral state reconstruction which an established phylogenetic approach that has been applied elsewhere to infer SARS-CoV-2 local and global movement (Wilkinson et al, Science 2021, Tegally et al, Nature, 2021).

The results we obtained from ancestral state reconstruction of Mombasa being a major gateway for variants entering the coastal region of Kenya is consistent with (a) the county showing the highest number circulating of lineages (n=28) compared to the other five remaining counties of Coastal Kenya, (b) approximately half (n=21, 49%) of the detected lineages in coastal Kenya had their first case identified in Mombasa and (c) Mombasa had an early wave of infections compared to the other Coastal counties.

We are not aware of an approach to consider statistical significance on these plots. The graphical display is based on the observed number events, and we would argue this is more appropriate than presenting absolute rates which would be susceptible to sampling bias.

Is it possible to account for potential bias in sequence sampling in these calculations, perhaps as done in Bezemer et al AIDS 2021? It should be possible to adjust for the proportion of sequenced individuals in PCR confirmed individuals, and it might also be possible to back calculate infected cases from cumulative reported deaths and to adjust for the proportion of sequenced individuals in infected individuals?

The reviewer suggests helpful methods to examine sampling bias, but we found this beyond scope here. Our method was based on ancestral location state reconstruction of the dated phylogeny. The approach has been used elsewhere to answer similar questions (Wilkinson et al, Science 2021, Tegally et al, Nature, 2021). The Bezemer paper uses maximum parsimony ancestral state reconstruction algorithm implemented in phyloscanner, and the Bayesian method applied to impute incomplete sampling is applicable to chains of transmission which we have not tried to reconstruct in our analysis.

Considering my earlier recommendation to document sequence sampling representativeness in Table 1, if Mombasa is found to be oversampled relative to infections, then it might also be helpful to perform sensitivity analyses in which sequences from over-represented locations are down-sampled. Another option might be to consider the approaches considered in de Maio PLOS Comp Bio 2015, or Lemey Nat Comms 2020. Thank you for investigating potential caveats and substantiating your findings in more detail.

In the revised manuscript we have clarified that our sequenced sample was proportional the number of positive cases reported in the respective Coastal Kenya counties (see-Fig.2E and Table 1).

The De Maio method uses BASTA (BAyesian STructured coalescent Approximation) into BEAST for purposes of phylogeographic analysis to compare ability to discriminate a zoonotic reservoir vs the implausible alternative cryptic human transmission. Analyses developed from these methods would be valid and interesting to apply to our dataset but would be a major new analysis and beyond the scope of the present paper. We have therefore taken the approach of: a) more clearly acknowledging sampling bias (see below) and b) undertaking sensitivity analyses (Supplementary File 5, see below). Using the larger global background sequence sets selected in a different way (more geographically balanced relative to the first round that was random), we still find that most of the virus introductions into coastal Kenya occurred via Mombasa consistent with our previous analysis.

The results are consistent with the case numbers in that (i) Mombasa experienced an earlier peak during wave one relative to other counties and (ii) had in total more cases than all the other five counties, and (iii) was commonly the first county of detection for many of the identified lineages in the region. However relative to its population, the border county of Taita Taveta had a higher import rate (13.5. per 100,000 people) compared to that of Mombasa (11.6 per 100,000 people), Table 4

Observations from our sensitivity analyses (Supplementary File 5) are included in the revised manuscript. We found that the absolute number of estimated viral imports/exports and intercounty transmission events fluctuated depending on the number of Coastal Kenya sequences and size of global comparison dataset but with a clear pattern of (a) counted events increasing with sample size (b) with Mombasa County consistently leading in the number of events; imports or exports.

• Line 292: The results are of course subject to differences in sequencing rates in each of the countries listed, and differences in reporting of these data.

This is a valid concern; to mitigate the bias that arises with these differences, unlike in the previous comparison dataset where we randomly selected a specified number of samples per month for each continent, in the revised analysis we have done the selection at country level. We limited the comparison data to maximum of 30 genomes per country per month per year. In this way, countries with high sequencing rates do not become overrepresented in our comparison dataset.

Some of these biases could be elicited through comparison to international travel data. For example, are the US and England also the top two countries from which most travellers arrive into Kenya? If such additional analyses are out of scope, it seems warranted to either strongly point to the substantial limitations of this analysis, or remove it altogether.

We concur with the reviewer on the potential bias that could exist in conclusions that arise from inferring sources of importations based on genomic data alone, available from only a few countries. However, vital quality and curated international travel data into Kenya during the study period was not available to us at the time of this analysis. We have therefore agreed to remove the previous analysis on potential origins and destinations of observed Kenya lineages from the revised manuscript.

What is perhaps striking is that Tanzania is entirely missing from this list, given extensive spread there. Another analysis that could be useful is a comparison of country specific lineage compositions, which might bypass some of the difficulties associated with substantial differences in sequence sampling/reporting rates.

SARS-CoV-2 genomic data from Tanzania has not been publicly shared to date, and hence is not included. And as indicated above, we have removed the analysis that was trying to infer sources of SARS-CoV-2 importations into Kenya.

To hypothesize on the potential lineages circulating in Tanzania, we have added a sentence detailing that 5 Pango lineages were identified among the 34 Tanzanian nationals who provided samples that were sequenced: B.1 (n=10), B.1.1 (n=10), B.1.351 (n=8), A (n=5) and A.23.1 (n=1)

• Line 536: it seems problematic that the data used in the import/export analysis did not contain all available African sequences. Can these be included in the corresponding analysis please.

In the revised manuscript we have included all accessible, good quality and contemporaneous Africa genomes in the revised manuscript (n=21,150). However due to the huge computational processing power need to process the phylogenetics for such large sequence data sets, we split the analysis into two parts, each with approximately 10,000 genomes (see Figure 3-figure supplement 1).

Notably with the increased sample size (including the analysis of 390 more genomes from coastal Kenya), we detected far more imports of SARS-CoV-2 into Coastal Kenya compared to our previous analysis (n=280 vs n=69) but only a modest change in exports (n=95 vs n=105) and inter-county virus movement events (239 vs 190).

Reviewer #2 (Public Review):

Agoti et al. analyzed SARS-CoV-2 samples collected from infected patients in coastal Kenya, collected between March 2020 and February 2021. This period spans the first two waves of COVID-19 in Kenya, and the authors aimed to understand the lineages circulating throughout the region, in comparison to the virus circulating elsewhere in Kenya and in the world. The manuscript is clearly written, and the figures and results are thorough and well described throughout. These data add to our understanding of COVID-19 in Kenya and in East Africa, and the discussion of how different lineages spread in Kenya (single clusters versus dispersed over several regions) is both interesting and potentially useful for informing public health measures.

The analyses are well done and excellently presented, but this paper is significantly lacking in a discussion of how sampling bias may affect the stated conclusions. Additionally, the paper focuses almost exclusively on genomic data and fails to closely examine epidemiological factors that may better contextualize the results presented.

We thank the reviewer for bringing this to our attention, we have added the paragraph below to the revised manuscript.

“Sampling bias is a potential limitation of this study arising from the fact that (a) demographic characteristics (age distribution, travel history and nationality) of the sequenced versus non-sequenced sub-sample differed significantly, (b) <10% of confirmed SARS-CoV-2 infections in Coastal Kenya were sequenced, prioritizing samples with a Ct value of <30.0 (Table 1); (c) the Ministry of Health case identification protocols were repeatedly altered as the pandemic progressed (Githinji et al., 2021) and (d) sampling intensity across the six Coastal counties differed, probably in part due to varied accessibility of our testing center that is located in Kilifi County (Figure 1A and Table 1). This may have skewed the observed lineage and phylogenetic patterns. To better contextualize the genomic analysis results, close examination of the case metadata is important, but unfortunately there was a lot of the metadata was missing (e.g., travel history, nationality, Table 2) which made it hard to integrate genomic and epidemiological data in an analysis. Although all analyzed genomes had > 80% coverage, very few were complete or near complete (>97.5%, n=344) due to amplicon drop-off or low sample quality and this may have reduced the overall phylogenetic signal.”

Specifically:

1) The authors do not discuss the potential effects of sampling on their import/export analyses. For example, they find that the USA and England are in the top six country sources of SARS-CoV-2 importation into coastal Kenya, as well as in the top six country destinations of viral export from the region. These two countries have generated huge numbers of sequences compared to the rest of the world, which may clearly bias these findings. While the authors do evaluate the sensitivity of their analyses by repeating them with different global subsamples, it is unclear if these subsamples corrected for large discrepancies in available data from different parts of the world.

We concur and appreciate that sampling bias is indeed a common limitation in the type of analysis we have undertaken given the variation in data collection across geographies. Some of the approaches we took to correct for this have been highlighted in our responses to reviewer #1.

In the revised manuscript, we have undertaken a reanalysis with a larger and more representative dataset at all scales of observation (Figure 3-figure supplement 1). Specifically, for the global dataset, we have revised our sub-sampling script to pick up the comparison dataset uniformly across months and countries for non-African countries. All the available African genomes have been included in our analysis including 605 collected in Kenya outside the coastal regional.

Similarly, the authors find that new variant introductions were mainly through Mombasa city, but most of the Kenyan sequences were from this region, so it is perhaps unsurprising that more lineages were found there. The authors should repeat their analyses with a more representative global subsample, or at the very least discuss these caveats in the discussion and discuss what other evidence there may be to support their findings.

Our sequencing rate by county is approximately proportional to the total number of cases seen in the county (Table 1 and Figure 2E). For Coastal Kenya, the revised manuscript included 389 additional genomes from coastal Kenya that became available while the manuscript was under review.

Thus, in the revised manuscript, we have addressed the valid sampling bias concerns of the reviewers and editor by: (i) increasing the number of analyzed genomes in our dataset for previously under-represented periods and regions, (ii) including contemporaneous Kenyan genomes from outside the coastal counties in our import/export analysis, (iii) including all available Africa genomes into the analysis and selecting a balanced global sub-sample for inclusion into the analysis. In addition, were have also provided a paragraph in the discussion section highlighting sampling bias as a caveat to interpretation of the findings of the current study:

“The accuracy of the inferred patterns of virus importations to and exportations from coastal Kenya are in part dependent on both the representativeness of our sequenced samples for Coastal Kenya and the comprehensiveness of the comparison data from outside Coastal Kenya. Our sequenced sample was proportional the number of positive cases reported in the respective Coastal Kenya counties (Figure 2E and Table 1). Also, we carefully selected comparison data to optimize chances of observing introductions occurring into the coastal region (e.g. by using all Africa data). But still there remained some important gaps e.g. non-coastal Kenya genomic data was limited (n=605). Despite this, we think the results from ancestral state reconstruction indicating that Mombasa is a major gateway for variants entering coastal Kenya is consistent with (a) the county showing the highest number lineages circulating (n=28) during the study period compared to the other five remaining Coastal counties Kenya, (b) approximately half (n=21, 49%) of the detected lineages in coastal Kenya had their first case identified in Mombasa and (c) Mombasa had an early wave of infections compared to the other Coastal counties and (d) is the most well connected county in the region to the rest of the world (large international seaport and airport and major railway terminus and several bus terminus).”

2) Restriction measures enforced by the Kenyan government are briefly introduced at the very beginning of the manuscript and then mentioned at the very end as a possible explanation for observed transmission patterns. However, there is very limited discussion of the potential effect of restriction measures throughout, and no formal analyses are presented using this kind of epidemiological information. Adding formal analyses to back up the hypothesis that relaxation of interventions may have driven the second wave of infections would make this paper much stronger and potentially more interesting.

In the revised manuscript, we have detailed the restriction measures the government of Kenya put in place in the introduction, methods, and results sections and discussed where appropriate on how we think they impacted the observed transmission patterns. We have added Supplementary Table 1 that provides the dates the various measures took effect or were relaxed.

In a separate piece of work (Brand et al, 2021 published in Science journal, 10.1126/science.abk0414), we investigated the potential drivers of the first three waves of infection observed in Kenya and we have appropriately referenced this in the revised manuscript.

We feel that additional analyses on the impact of the restriction measures on SARS-CoV-2 epidemiology and the lineage patterns observed are beyond the scope of this work whose focus was primarily genomic epidemiology.

3) Generally, the text of the manuscript focused on waves of SARS-CoV-2 transmission, while the analyses presented data aggregated by month. A clearer connection between month and wave (particularly visually, on the figures themselves) would aid in interpretation of the data presented.

This is a valid concern and a good suggestion. In the revised manuscript, for all temporal plots, we have added a line to demarcate when we switched from wave one to wave two period. Similarly, for several analyses, we have provided aggregations by wave period rather than by month.

4) One of the strengths of this manuscript is the depth to which the authors discuss the detection of specific lineages in coastal Kenya. However, there is limited discussion of these results in the context of when various lineages appeared or disappeared globally, though these details are presented in a table. Discussing the appearance of the various lineages (was it surprising to see a particular lineage at a certain time or in a certain place?) would also improve this manuscript.

In the revised manuscript, we have compared the patterns of lineage detection locally compared to all Kenya and to all continents in the newly added Figure 3. We have also discussed this aspect for the most frequent 4 lineages in both Wave one and Wave two.

Author Response:

Reviewer #1:

Hauser et al, analyze two large datasets of GPCR-G protein interactions/couplings ("Inoue" and "Bouvier"), comparing and combining them with the widely-used literature-based Guide to Pharmacology (GtP) database. As the Inoue and Bouvier datasets were based on different experimental setups, this enables the identification of which couplings are supported by more than one method. The authors also establish a normalization protocol that enables to move from qualitative to quantitative comparisons and identify couplings that might be either below are above a rigid threshold. Overall, the paper describes a new resource and the methodologies used to build this resource. The resulting coupling map is available through the GPCRdb website, a widely used resource in the field.

The authors have thus improved the ability of researchers to assess prior results and compare them to their own new data. This resource clearly and significantly upgrades options currently available and will likely be of interest and prove quite useful to scientists both in academia and in industry.

We thank the reviewer for so nicely describing the study and its prospective application.

Weaknesses include:

• The data is described mostly by broad numbers, such as the number of receptors or coupling in a subset, or percentages. While this is helpful to understand the data, this reviewer found it hard to follow the mountain of numbers. A suggestion would be to add a section where the authors pick selected examples of particular experimental data and show how their combine database can resolve previously unanswered (or wrongly answered) questions of GPCR/G protein coupling.

We have removed numbers in several places throughout Results where we had included multiple measures e.g., absolute numbers and percentages. Furthermore, where an overall number has been broken down into distributions, e.g., across different G proteins of families thereof, we moved other numbers to parentheses.

The different sections of Results that answer questions of GPCR-G protein coupling have now been presented more clearly by updating their headings and grouping them all in a subsection of part of Results called “Research Advances – Insights on GPCR-G protein selectivity”. These sections are all based on our “combined database”/coupling map. In each such section, we start at the overall level – covering all GPCRs and/or G proteins – but then give selected examples thereof that are weaved into and exemplifies the text. This approach has also been used in the new Results section “Differential tissue expression gives G proteins in the same family large spatial selectivity”, which gives selected examples of G proteins with specific tissue expression profiles.

Given that the paper has already exceeded the maximum of 5,000 words by quite a bit, we think that this approach of weaving selected examples into each selectivity insight section is the most appropriate, and that it brings most clarity. Furthermore, we hope that readers will be inspired to use our coupling map to generate additional questions for future experiments.

• The paper does not reveal new biological findings. For example, while some emphasis is placed on new data on G15, it would be helpful to take the extra step and use this to suggest new biological insights.

eLife’s author guidelines (https://reviewer.elifesciences.org/author-guide/types) state that “Tools and Resources articles do not have to report major new biological insights or mechanisms, but it must be clear that they will enable such advances to take place, for example, through exploratory or proof-of-concept experiments.” In case this manuscript is published as a Tools and Resources paper, it may therefore be sufficient to provide the foundation for future studies to reveal new biological findings.

Nevertheless, the coupling map led to biological findings relating to patterns and mechanisms of GPCR-G protein selectivity that were not described in the original studies. I.e., while this study did not generate new data, it arrived at new insights based on published data. This seems to be in line with eLife’s publication format “Research Advances” (https://reviewer.elifesciences.org/author-guide/types), and the Analysis format of several other journals. Some insights described herein have not been presented before while others have been updated in scope and precision. Furthermore, we have added a new section of Results with insights on G protein expression profiles and co-expression.

We have clarified this by updating the headings of the sections that present these insights, and grouped them under a common subheading of Results termed “Research Advances – Insights on GPCR-G protein selectivity”. However, in case we have overlooked very recent studies describing some of the same biological insights, we would please like to ask for their references and would be more than willing to revise the manuscript again to incorporate them. Furthermore, if the Reviewer is missing a particular analysis that is critical to understand GPCR-G protein coupling, please let us know.

• The authors cautiously label couplings supported by only one dataset as "unsupported". It would seem more helpful to grade couplings by a reliability scale, providing users with a wider set of data. Perhaps only couplings that are directly conflicted by negative data should be labeled as unsupported?

We understand that the term “unsupported” has been used in a confusing way. We have now replaced this term with “unique” and explained all terms in Table 1 of the revised manuscript.

To address the need for a means to grade or filter couplings by reliability, we have added the following paragraph to the manuscript:

“To enable any researcher to use the coupling map, we have availed a “G protein couplings” browser (https://gproteindb.org/signprot/couplings) in GproteinDb (2). By default, this browser only shows “supported” couplings with evidence from two datasets, but there is an option (first blue button) to changes the level of support to only one (for most complete coverage of GPCRs) or to three (for the highest confidence) sources. We propose a standardized terminology to describe couplings based on their level of experimental support from independent groups (Table 1). The criterion of supporting independent data, and the terms “proposed” and “supported”, are already used by the Nomenclature Committee of the International Union of Basic and Clinical Pharmacology (NC-IUPHAR) for GPCR deorphanization. Furthermore, the online coupling browser allows any researcher to use only a subset of datasets, or to apply filters to the Log(Emax/EC50), Emax, and EC50 values. Finally, users can filter datapoints based on a statistical reliability score in the form of the number of SDs from basal response."

Furthermore, we have added references to the online G protein coupling browser in the:

(1) Introduction ending: “On this basis, we develop a unified map of GPCR-G protein couplings that can be filtered or intersected in GproteinDb …”, (2) Fig. 2 legend ending: “Note: Researchers wishing to use this coupling map, optionally after applying own reliability criteria or cut-offs, can do so for any set of couplings in GproteinDb (1).” (3) Fig. S2 ending: “Unique couplings are hidden by default in the online G protein couplings browser in GproteinDb, as they await the independent support by a second group.”

To many scientists the most reliable option is to involve NC-IUPHAR. Gloriam is a corresponding member of NC-IUPHAR, which has mentioned the possibility of involving its many worldwide pharmacological experts to update GtP on a case-by-case basis for receptors. For example, many of the “novel” couplings jointly supported by Bouvier and Inoue may be added. This option is advantageous as it involves experts in each receptor system (often with knowledge of other relevant studies) and is backed by the authoritative organization.

• Given that this manuscript includes authors from both the Inoue and Bouvier studies, I can understand why they are not directly assessing which of the two datasets (in relation to the GtP) might be more accurate. Nevertheless, I believe this assessment should be done and that the advantages and disadvantages of the two experimental systems discussed clearly.

We believe that the three-way intersection of couplings is the most informative and therefore preferred over individual comparison of each of the Inoue and Bouvier datasets to GtP. GtP is unfortunately not suitable as a stand-alone resource – neither to contradict nor support couplings (on the G protein subtype level). This is because GtP is incomplete (especially for G12/13) and does not provide any information on the level of G protein subtypes, only families. The three-way interactions will always use GtP but adds a second dataset on top of this when validating a third dataset. Our manuscript already included a three-way intersection of datasets, allowing readers to conclude which dataset might be more accurate (then Fig. 3 and Spreadsheet 3) on a per-G protein basis.

In the revised manuscript, we have rewritten this section, which now has the heading “Bouvier’s and Inoue’s biosensors appear more sensitive for G15 and, Gs and G12, respectively. We have also made a completely new figure, Fig. 7, which more clearly illustrates for which G proteins that Bouvier and Inoue may have overrepresented or underrepresented couplings. This section specifically investigates the question of whether differential sensitivity can explain “unique” couplings. However, such unique couplings can either be due to overrepresentation or instead be true positives that are missing in GtP because of incompleteness and in the other biosensor due to lower sensitivity. Unfortunately, we will not be able to distinguish these possibilities until the research community has gained additional datasets from independent biosensors with as high sensitivity.

Whereas our study compares datasets rather than experimental systems, we have added a paragraph in the Discussion describing which aspects should be considered when choosing a biosensor. There, we reference a review from last year dedicated to biosensors and describing their pros and cons (3), and the accompanying paper by Bouvier et al. (4), comparing several aspects of the experimental system used by Inoue et al. It is also important to note that the most advantageous biosensor may be one of the two for which data is analyzed in our paper. For many studies, researchers may instead be better off with another biosensor, for example those from Lambert/Mamyrbekov (5), Roth (2) (Gαβγ sensors first described in (6-11)) or Inoue (unpublished dissociation assays using wt G proteins fused with LgBit and HiBit). These are all referenced in the Discussion.

References:

1. Pandy-Szekeres G, Esguerra M, Hauser AS, Caroli J, Munk C, Pilger S, et al. The G protein database, GproteinDb. Nucleic Acids Res. 2022;50(D1):D518-D25. 10.1093/nar/gkab852
2. Olsen RHJ, DiBerto JF, English JG, Glaudin AM, Krumm BE, Slocum ST, et al. TRUPATH, an open-source biosensor platform for interrogating the GPCR transducerome. Nat Chem Biol. 2020;16(8):841-9. 10.1038/s41589-020-0535-8
3. Wright SC, Bouvier M. Illuminating the complexity of GPCR pathway selectivity – advances in biosensor development. Curr Opin Struct Biol. 2021;69:142-9. https://doi.org/10.1016/j.sbi.2021.04.006
4. Avet C, Mancini A, Breton B, Gouill CL, Hauser AS, Normand C, et al. Effector membrane translocation biosensors reveal G protein and B-arrestin profiles of 100 therapeutically relevant GPCRs. bioRxiv. 2021:2020.04.20.052027. 10.1101/2020.04.20.052027
5. Masuho I, Martemyanov KA, Lambert NA. Monitoring G Protein Activation in Cells with BRET. Methods Mol Biol. 2015;1335:107-13. 10.1007/978-1-4939-2914-6_8
6. Gales C, Rebois RV, Hogue M, Trieu P, Breit A, Hebert TE, et al. Real-time monitoring of receptor and G-protein interactions in living cells. Nat Methods. 2005;2(3):177-84. 10.1038/nmeth743
7. Gales C, Van Durm JJ, Schaak S, Pontier S, Percherancier Y, Audet M, et al. Probing the activation-promoted structural rearrangements in preassembled receptor-G protein complexes. Nat Struct Mol Biol. 2006;13(9):778-86. 10.1038/nsmb1134
8. Schrage R, Schmitz AL, Gaffal E, Annala S, Kehraus S, Wenzel D, et al. The experimental power of FR900359 to study Gq-regulated biological processes. Nat Commun. 2015;6:10156. 10.1038/ncomms10156
9. Breton B, Sauvageau E, Zhou J, Bonin H, Le Gouill C, Bouvier M. Multiplexing of multicolor bioluminescence resonance energy transfer. Biophys J. 2010;99(12):4037-46. 10.1016/j.bpj.2010.10.025
10. Bunemann M, Frank M, Lohse MJ. Gi protein activation in intact cells involves subunit rearrangement rather than dissociation. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(26):16077-82. 10.1073/pnas.2536719100
11. Janetopoulos C, Jin T, Devreotes P. Receptor-mediated activation of heterotrimeric G-proteins in living cells. Science. 2001;291(5512):2408-11. 10.1126/science.1055835

Reviewer #2:

This study is a meta-analysis of previously reported studies on G protein-coupled receptor (GPCR) coupling to G proteins. The data sets are from three distinct sources: a compendium compiled by the International Union of Basic & Clinical Pharmacology (IUPHAR), and two data sets compiled by two separate laboratories. Each of these data sets describes the coupling of members of the superfamily of non-sensory GPCRs (~200 genes) to the large family of G protein alpha subunits (~20 genes). The authors try to arrive at a consensus for receptor-G protein coupling from the three data sets, as well as identify and highlight differences or incongruencies. Compiling these vast data sets into a unified format will be extremely useful for investigators to understand receptor and effector relationships. The meta-analysis will help to deconvolute the complex physiology and pharmacology underlying hormone or drug actions acting on receptor superfamilies. A better understanding of receptor-G protein selectivity and/or promiscuity will ultimately help in identifying safer therapeutics.

We appreciate the summary and the explanation of the usefulness of our meta-analysis and its potential impact.