Single-cell transcriptomic analysis identifies the conversion of zebrafish Etv2-deficient vascular progenitors into skeletal muscle

Cell fate decisions involved in vascular and hematopoietic embryonic development are still poorly understood. An ETS transcription factor Etv2 functions as an evolutionarily conserved master regulator of vasculogenesis. Here we report a single-cell transcriptomic analysis of hematovascular development in wild-type and etv2 mutant zebrafish embryos. Distinct transcriptional signatures of different types of hematopoietic and vascular progenitors are identified using an etv2ci32Gt gene trap line, in which the Gal4 transcriptional activator is integrated into the etv2 gene locus. We observe a cell population with a skeletal muscle signature in etv2-deficient embryos. We demonstrate that multiple etv2ci32Gt; UAS:GFP cells differentiate as skeletal muscle cells instead of contributing to vasculature in etv2-deficient embryos. Wnt and FGF signaling promote the differentiation of these putative multipotent etv2 progenitor cells into skeletal muscle cells. We conclude that etv2 actively represses muscle differentiation in vascular progenitors, thus restricting these cells to a vascular endothelial fate.

overexpression in many cell types will push cells toward a vascular fate. Lack of myod expression only reflect a lack of skeletal muscle cells but does not directly implicate etv2 as a repressor of skeletal muscle fate.
In the dkk1 treated embryos (fig 4), what happens globally to myocytes? Are they reduced to the same extend as the GFP+ myocytes? The authors state that "Wnt signaling promotes muscle differentiation in multipotent etv2+ progenitors." Is this unique to the etv2+ derived myocytes?
Page 14, lines 43-44: "our work has demonstrated that cells in the vascular endothelial lineage arise from multipotent progenitors in the lateral plate mesoderm which can differentiate into skeletal muscle in the absence of Etv2 function". To fully demonstrate this, lineage tracing would be required.

Minor comments
Page 10, lines 10-11:" Double heterozygous etv2gal4/ci33 embryos displayed a dramatic increase in GFP+ myocytes compared to etv2:gal4+/-embryos". It looks like a 3 to 4 fold increase in GFP+ myocytes which is far from "dramatic". Same remark for the sentence: "SU5402-treated embryos showed a dramatic decrease in the number of ectopic GFP+ muscle cells", data in which GFP+ myocytes decrease about two-fold." Page 10, lines 23-24: "The number of myocytes was significantly increased in the etv2:gal4+/embryos injected with the previously validated scl morpholino (MO)". Does this sentence refer to the overall number of myocytes per embryos? Figure 5: it is very difficult to see co-expressing cells. Close-up images should be shown.
Page 14, lines 10-11: "previous studies have demonstrated the existence of a common vascular and endothelial progenitor, the hemangioblast". Is that not rather a common hematopoietic and endothelial progenitor?
Reviewer #2: Remarks to the Author: In this manuscript Chestnut et al., perform scRNA-seq in etv2-expressing cells from zebrafish embryos. They perform a thorough analysis and based on the expression of known markers for specific lineages they identify clusters of endothelial cells, myeloid and erythroid cells, LPM progenitors etc. Interestingly, they observe that some of the Etv2 expressing cells differentiate into skeletal muscle cells in the absence of Etv2. They then show that this fate is promoted by the Wnt and FGF pathways and assume that Etv2 is required to suppress this alternative fate. Apart from this main observation of the paper, they also identify the transcriptional signature of multipotent progenitors in LPM and show that arterial progenitors co-express arterial and venous markers. Single cell RNA seq technology gave a tremendous boost to developmental and lineage tracing studies and this paper provides important information on the molecular fingerprint of vasculogenesis. However, many aspects need to be addressed or clarified.
Major comments: 1. The authors need to find a way to isolate the etv2 positive cells and they employ a transgenic line that has gal4 inserted into the coding part of the gene. As a result, they use the heterozygous and homozygous cells of this line and treat the heterozygous as almost wild type. This can lead to wrong conclusions, especially if the authors don't show in detail that heterozygosity of etv2 does not disturb main functions. Another way to solve the problem would be to use the Tg(-2.3 etv2:GFP) line, perform scRNA-seq and compare it to their current results. It is understandable that these two lines do not mark exactly the same cell populations, but some commonalities should be present. In addition, the authors claim that the phenotype of their transgenic animals is more severe than the one of the actual mutants. This may be caused by deletion of regulatory elements that are located close to etv2 gene. Can the authors check the expression of these adjacent genes? Such a problem leads to an even more urgent need to sequence cells from the Tg(-2.3 etv2:GFP) or any other transgenic that could verify the current results. 2. The number of heterozygous and homozygous cells sequenced differs greatly between them. This maybe a problem in comparing these two experiments and identifying missing or enriched populations. Can the authors equilibrate these differences or prove that they do not affect the results? 3. The authors claim that Wnt and FGF signaling promote skeletal muscle fate. Can they show in their scRNA seq data that genes regulated by these pathways have altered expression? 4. Can scl overexpression rescue the phenotype?
Minor comments 1. Page 7 line 30: Supp. Fig. S2 should be mentioned here. 2. On Figure 1 the authors claim that they show also markers of erythroid cells and other populations (specifically in Fig 1c,d) but that is not so. Can the authors add all the markers that are mentioned in the text? 3. The authors claim that the homozygous transgenic animals have more erythroid cells. Can the authors verify this using an independent method like WISH or by FACS analysis of an ertyhroid reporter animal? 4. The authors show results on heterozygous and homozygous animals, but never on real wild type animals. Can the authors include wild type animals in any case that is pertinent, for example in WISH experiments? 5. Page 8 Line 16,17. Apoptosis and cell cycle genes do not exist in Fig 1c,d. 6. Page 10 line 2 myod expression is missing from the figure. 7. Can the authors clarify how they count the GFP+ myocytes?
Reviewer #3: Remarks to the Author: The authors have made a zebrafish transgenic line to characterise the development of cells derived from etv2-expressing progenitors by inserting a Gal4 protein into the etv2 locus. They then isolated GFP+ cells from wildtype and etv2 heterozygote animals at two stages of development when haematopoetic and vascular progenitor cells are specified and performed single cell sequencing. They used outputs from this sequencing to highlight a population of cells that express genes expressed in myogenic lineages and ask whether this reveals that etv2-expressing cells can form muscle. They test this by performing time-lapsed analysis of the same transgenic line to argue that a population of lateral-plate mesoderm (LPM) derived cells form myofibres during development and that this is prevented by expression of etv2 in these cells. They then argue that etv2 acts to promote a fate decision of LPM cells to form blood vessel cells and in an absence of etv2 function there is a fate switch to myoblasts. They support this proposed mechanism by showing that knockdown of the etv2 target gene scl increased the number of GFP+ myoblasts that form. They also show that over-expression of etv2 RNA appears to reduce expression of myod at early stages. They then investigate the importance of Fgf and Wnt signalling in directing this fate decision using transgenic lines to inhibit Wnt and Fgf and a small molecule inhibitor of Fgf receptor function. Based on changes to the number of GFP+ cells forming myofibres they argue that both Wnt and Fgf promotes etv2-expressing cells to assume a muscle fate.
The central premise of this work is that etv2-expressing cells in the LPM make a fate decision controlled by etv2/ scl function relative to Wnt and Fgf activity. An important consideration when making this assumption is whether there is an effect on progenitor cells in the LPM that may affect how they respond to signals from their environment. It was not possible to discern from the time-lapse movies how the GFP+ progenitor cells migrate in etv2 heterozygotes and whether this is altered. If they show a different spatial organisation prior to or during migration they are likely to experience different cues that affect their specification. In order to address this point the authors should carefully characterise etv2-expressing cells in heterozygotes and homozygous animals using dorsal and lateral views to highlight cell movement to answer whether cells migrate differently in homozygous animals. A similar point holds for animals in which Wnt and Fgf are altered. I was also not clear if there are changes to the relative number of GFP+ cells at different stages of development in both genotypes or in when Fgf or Wnt are inhibited as this may also explain why there are more or fewerGFP+ myofibres. Careful quantification of cell number relative to their position is needed to address this issue.
Given that the entire paper relies on the new etv2 line I was surprised that this is not described in full in this paper. It is important to do this as I had the impression that the expression differs from a previously published line made with a BAC and it is not clear in this work how accurately the new CRISPR-generated line expresses GFP relative to etv2. A similar caveat holds for descriptions for cell behaviour -it is simply not possible to see the different cell populations described in the text from the figures shown, specifically how the LPM progenitors migrate. Clearer figures demonstrating the spatial distribution of GFP+ cells is needed to clarify how this may affect their fate switch in an absence of etv2 function.
The hypothesis that etv2 acts to inhibit myogenesis is tested by injecting RNA for etv2 and assaying myod expression. A striking result is shown indicating a unilateral loss of myod expression at early somite stages. This should be corroborated by qPCR for myod and other myogenic genes including myf5. The statement that GFP+ cells are forming myofibres should be tested by antibody staining to show that these express markers of differentiated myofibres. One important piece of evidence for the role of etv2 in dictating fate of this early LPM progenitor population is to show where progenitor cells switch to a myogenic fate and when this occurs. This could be accomplished by examining myod expression relative to GFP+ cells in etv2 mutants at different stages of development to show if there is ectopic expression in GFP+ cells and when this occurs.
Reviewer #4: Remarks to the Author: Chestnut B. et al performed a comprehensive study for Etv2 deficient zebrafish embryos during hematovascular development. It is known that Etv2 expression alone is sufficient to transdifferentiate skeletal muscle cells into functional blood vessels (Ref: Transdifferentiation of fast skeletal muscle into functional endothelium in vivo by transcription factor Etv2. PLoS biology, 11(6), e1001590.). The major finding of this study is that in the absence of Etv2 function, vascular progenitors can acquire a skeletal muscle fate. Hence, this manuscript further confirmed the role of Etv2 in blood vessel development.

Major comments:
(1) Over-expression of Etv2 have been extensively studied. It will be interesting to compare this study (Etv2-deficient) with published Etv2-overexpression datasets, and see the overlap (e.g., whether key upregulation markers in this study are downregulated in Etv2-overexpression datasets).
(2) Page 5 (Line 17-19): To perform QC, it is reasonable to remove cells with low number of genes expressed (<200). But it is unclear why cells with more genes expressed (>3,500 genes) were also filtered out. Sometimes, cells with unexpectedly high counts and a large number of detected genes may represent doublets. But this is not always true depending on sequencing protocols, batches, and cell types. Different cell types have huge variations for the number of genes detected. Also, what does ">5% of genes mapping to the mitochondria" mean is unclear. Does it mean (a) >5% of expressed genes are mitochondria genes or (b) >5% of mapped reads came from mitochondria?
(3) Page 5 (Line 28-30): A high fraction of mitochondrial counts are indicative of cells whose cytoplasmic mRNA has leaked out through a broken membrane. For cells with high % of mitochondria reads should be removed in QC step. Since gene expression levels calculated by RNA-seq are relative numbers, they should be normalized and scaled by total non-mitochondrial genes. It is unclear how gene expressions can be normalized by mitochondria genes.
(4) PCAs were used for t-SNE analysis. However, the choose of top PCs are fairly random and are not consistent: Page 5 Line 33: 13 PCs were selected; Page 6 Line 9: 14 PCs were selected; Page 6 Line 21: 7 PCs are selected. The top PCs selection should either have specific reason for each study or be consistent for all studies.
(5) Page 7 Line 8: p-values was determined by t-test. The basic assumption of using t-test is that "in-group" is very homogeneous and the variations follow normal distribution. Since the data is "log-transformed fold change", other statistical methods, such as Wilcoxon signed-rank test, will be more appropriate.
(6) Page 6 Line 18-19: "Cells were filtered based on UMI<10^6, to exclude doubles/triplets". Low number of UMI are mainly due to PCR bias. This is because amplification of low amount of RNAs can result in substantial bias towards to certain fragments. Low UMI has nothing to do with doubles/triplets. The only way to detect doubles/triplets from scRNA-seq data is to use complicated computational approaches, such as Scrublet (Ref: Scrublet: computational identification of cell doublets in single-cell transcriptomic data." Cell systems 8.4 (2019): 281-291.). In fact, "doubles/triplets" will cause an increased number of UMI due to "more cells".

RESPONSE TO REVIEWERS
We thank all reviewers for their comments. Based on the comments, we have substantially revised the manuscript, including multiple new experiments and new figures (new panels in Fig. 3, new figs. 4, 5, revised fig. 7, and new Suppl. Figs. S9, S10, S12, S13, S14, new Table 1 and Movies 1-5). Please note that based on the advise of ZFIN nomenclature committee, we have renamed etv2:Gal4 line into etv2 Gt(2A-Gal4)ci32 , abbreviated as etv2 ci32Gt , to reflect the fact that it is a gene trap construct which has interrupted etv2 locus. We had previously submitted the description of the generation and characterization of the etv2 ci32Gt line as a separate manuscript which has been recently published (Chestnut and Sumanas, Dev Dyn 2019; doi: 10.1002/dvdy.130).
The specific comments by reviewers were addressed as follows.

Reviewer #1:
Specific comments: Why was the Etv2:gal4, UAS:GFP rather than the Etv2:GFP or Etv2:mCherry used for the scRNA-seq experiments? Why not do an ETV2-2A-GFP genomic insertion? Is there an advantage in using the gal4/UAS transgenic system? How accurate and stringent is Etv2 expression represented by GFP in this model? In addition to the immunofluorescent data presented, was any molecular analysis performed? What is the half-life of GFP in this transgenic model? As judged by etv2 expression in single cell RNA-seq data (fig 1 c), only a low fraction of the cells analyzed do express etv2.
Response: As we reported recently in a separate manuscript (Chestnut and Sumanas, Dev Dyn 2019; doi: 10.1002/dvdy.130) which describes generation and characterization of the etv2 ci32Gt gene-trap line, previously generated etv2 reporter lines display some non-specific expression. We expected that the gal4 knock-in line will recapitulate endogenous pattern of etv2 expression more accurately, and will also allow to visualize etv2-deficient cells present in homozygous embryos. In addition, Gal4/UAS system allows for signal amplification and typically results in a brighter GFP fluorescence, compared with other approaches. We have performed detailed characterization and comparison of GFP fluorescence with endogenous etv2 mRNA expression in a separate study (Chestnut and Sumanas, Dev Dyn 2019) and showed that GFP expression pattern recapitulates well the endogenous etv2 expression. Reported half-life of EGFP is 26 hours, therefore GFP fluorescence is expected to stay much longer even after etv2 mRNA expression has been downregulated. Our previous research has suggested (Sumanas et al 2008, Blood 111, 4500-4510;Glenn et al 2014, Dev Biol 393, 149-159) that etv2 is expressed in early hematopoietic progenitors but its expression is downregulated as they differentiate. This explains low levels of etv2 mRNA expression in many other cell populations identified by scRNA-seq. In addition, due to "dropouts" which happen when preparing libraries for RNA-seq, only a fraction of endogenous transcripts are present in the scRNA-seq libraries. As a result, cells that have low expression of etv2, may show 0 transcripts in scRNAseq data which explains apparent absence of etv2 in some cells.

Comment:
In the scRNA-seq experiment, 2049 GFP+ cells were sequenced from 100-150 het embryos and 588 GFP+ cells from 75-100 from homo embryos. It is not clear how representative the scRNA-seq data are of overall GFP+ embryonic populations. What are the frequencies of GPF+ cells in het and homo embryos? This becomes very important when reporting the proportions of identified subsets as shown in figure 1e for example. The authors state that: "the LPM, cardiomyocyte and skeletal muscle (myocyte) populations were greatly increased in the homozygous embryos" but is this increase only relative within the context of the single cell populations analysed? It is not clear that these conclusions can be extrapolated to the full embryo.
Response: During FACS sorting of GFP-positive cells from heterozygous and homozygous embryos, 30,984 GFP+ cells out of total 1.639x10 -6 cells and 9,067 GFP+ cells out of 458,341 total cells were isolated from etv2 ci32Gt heterozygous and homozygous embryos, respectively, resulting in the frequencies of 1.89% and 1.98% GFP+ cells in heterozygous and homozygous embryos, respectively. Thus, the relative number of GFP+ cells was similar in heterozygous and homozygous embryos. We have now included this data in the main manuscript text.

Comment:
Page 8 lines 40-41: "The presence of GFP-positive cardiomyocytes in etv2:gal4 heterozygous embryos could be explained by the loss of one functional etv2 allele, which resulted in some etv2+ cells differentiating as cardiomyocytes." What are the evidences that Etv2+ progenitors do not give rise to cardiomyocytes? Has lineage tracing with Etv2 been performed?
Response: GFP expression has not been observed in cardiomyocytes using any other etv2 trangenic reporter lines in wild-type background (Palencia-Desai et al 2011, Development 138, 4721-4732). In addition, etv2 is not expressed in the myocardial field and endogenous etv2 expression does not overlap with early myocardial markers in wild-type embryos. Although, to our knowledge, no lineage tracing of etv2 cells has been performed in zebrafish embryos as yet, this makes it very unlikely that etv2+ cells differentiate into cardiomyocytes in wild-type embryos. In the manuscript we merely suggest a likely explanation of our results; this study was not designed to test differentiation of etv2 progenitors into cardiomyocytes.

Comment:
Page 3, lines 38-39: "Etv2 function is required to actively repress alternative cell fates in multipotent mesodermal progenitors." What are the evidences that these progenitors are multipotent and that Etv2 is expressed in these progenitors? Based on the scRNA-seq data, few LPM cells express etv2. If those represent progenitors, one might expected to find etv2 still expressed.

Response:
Appearance of only few LPM cells that express etv2 can be explained by low level of etv2 expression in these cells. The drop out rate in scRNA-seq experiments is estimated at 80% or even higher due to many technical reasons, meaning that only 20% or less of mRNA transcripts are captured in each cell. For genes that are expressed at a low level, some or multiple cells will show no transcripts. This is also evident in etv2 expression in EC1 and EC2 populations (Suppl. Fig. S2); there are multiple EC1 / EC2 cells with no etv2 expression. Even expression of cdh5, a classical marker for EC populations, is absent in a significant number of EC1 and EC2 cells (Suppl. Fig. S2). It is expected that gal4; UAS system results in signal amplification, therefore even low level of etv2 expression can result in substantial GFP expression which helps to capture these cells during FACS sorting. We have confirmed overlap of GFP and LPM gene co-expression in Fig. 7 and Suppl. Fig. S14.
We suggest that these progenitors are multipotent because they can differentiate into alternative cell fates (skeletal muscle) in the absence of Etv2 function. We admit that we have not demonstrated that a progeny of a single cell can acquire different fates, a classical definition of multipotency. Therefore we softened the statement and removed multipotency claim from that sentence. Comment: Page 9 lines 5-6: "A population of RBC cells was expanded in the etv2:gal4 homozygous embryos". What is the basis for claiming that RBC cells expanded? Are there more RBCs in Etv2:gal4+/+ embryos? This should be supported by additional evidences.
Response: Indeed, expansion of RBCs has not been previously reported. Based on scRNA-seq data, there is a higher percentage of RBC cells in etv2 ci32Gt homozygous embryos than in heterozygous, while the overall fraction of GFP+ cells is similar. To confirm this result, we used three different approaches. 1) We have recently published global RNA-seq analysis of etv2 ci32Gt embryos (Chestnut and Sumanas 2019, Dev Dyn). We reanalyzed this data to investigate expression of RBC-specific genes. Indeed, multiple RBC-specific genes, including six different hemoglobins were upregulated in etv2 ci32Gt homozygous embryos at 24 hpf (Suppl. Table S3). Note that this was a completely different analysis using global RNA-seq of embryos at 24 hpf stage compared to scRNA-seq at the 20-somite stage presented in the current study. 2) We performed differential expression analysis of RBC marker genes identified in scRNA-seq analysis between homozygous and heterozygous embryos. Several RBC-specific genes including three different globins were upregulated in etv2 ci32Gt homozygous embryos (Suppl . Table  S4). 3) We could not directly count RBC cells in etv2 ci32Gt homozygous embryos because we did not have the line crossed into RBC-reporter line to do this analysis in a timely manner. Instead, we used a previously validated etv2 MO to knock down etv2 in gata1:dsRed reporter embryos, which express dsRed in RBCs. Based on the cell counts of FACS-sorted fluorescent cells, etv2 knockdown embryos show a significant and reproducible increase in the number of gata1:dsRed-positive cells (Suppl. Fig. S9). All of these data point to increased erythropoiesis in etv2-deficient embryos. The nature and the mechanism of this phenotype will require further investigation.
Comment: Differentiation of etv2+ progenitors into skeletal muscle: it is not clear how prevalent this is? What is the frequency of etv2+ derived skeletal muscle relative to the overall number of skeletal muscle cells at the embryonic stage analyzed ( figure 3, 4)? Is this a marginal event or a widespread phenomenon? The scl morpholino data are rather weak and would need additional molecular-based evidences to support the involvement of Scl in this process.

Response:
We have observed 3±2.5 GFP+ myocytes per embryo in etv2 ci32Gt heterozygous embryos and 18±5.4 GFP+ myocytes per embryo in homozygous embryos. We now included these numbers in the manuscript text and Fig. 3e. Based on scRNA-seq data (Fig. 1e), GFP+ myocytes represent 1.2% of total GFP cells in heterozygous and 3.9% of total GFP cells in homozygous embryos. It is challenging to estimate the total number of skeletal muscle cells. Nevertheless, it is clear that the number of GFP+ myocytes is relatively small compared to the total number of skeletal muscle cells. Muscle cell counts in etv2 ci32Gt+/embryos injected with scl MO are based on 20 randomly selected embryos which were analyzed by confocal microscopy in two independent experiments. Increase in GFP+ myocytes was reproducible and highly statistically significant (p<0.01). Control uninjected embryos from the same batch were analyzed in parallel. Scl MO-injected embryos were morphologically normal and showed previously reported defects in hematopoiesis and vacular development. To confirm that the analyzed cells are indeed skeletal muscle, we analyzed muscle actin reporter actc1b:GFP expression in etv2 ci32Gt ; UAS:mCherry embryos injected with scl MO. We show that these cells are positive for actc1b:GFP expression (Fig. 3o-q), thus confirming that scl knockdown results in increased muscle differentiation of etv2 reporter cells.

Comment:
Page 10, lines 31-32: "Muscle-specific myod expression was greatly inhibited upon etv2 overexpression". This does not strengthen the conclusion given that it is well established that etv2 overexpression in many cell types will push cells toward a vascular fate. Lack of myod expression only reflect a lack of skeletal muscle cells but does not directly implicate etv2 as a repressor of skeletal muscle fate.

Response:
The data in Fig. 3t show that etv2 overexpression results in a partiall loss of myod expression. We now confirmed this result by qPCR analysis (Fig. 3u). We agree that there could be different interpretations of this result possible. Indeed, etv2 overexpression will drive many cells towards vascular fate. However, myod-expressing cells are positioned in the somites and normally are exposed to the signals which induce muscle fate in wild-type embryos. It is not well understood why activation of endothelial program in these muscle progenitors prevents these cells from also initiating myogenesis in response to muscle inducing signals. Previous work by Org et al (EMBO J 34, 759-777 (2015) has demonstrated that Scl can occupy cardiac enhancers preventing their activation. We propose that Etv2 may function through or together with Scl to inhibit muscle development through a similar mechanism. However, inhibition of muscle expression of etv2-overexpressing embryos could also be indirect. Further studies would be needed to identify the mechanism of how muscle differentiation is inhibited in etv2expressing cells.

Comment:
In the dkk1 treated embryos (fig 4), what happens globally to myocytes? Are they reduced to the same extend as the GFP+ myocytes? The authors state that "Wnt signaling promotes muscle differentiation in multipotent etv2+ progenitors." Is this unique to the etv2+ derived myocytes? Response: Previous study by Martin and Kimelman, 2012 (Dev Cell 22, 223-232) showed that heatshock induction of dkk1 expression at the 8-somite stage results in reduced myocyte differentiation which is limited to the tail region because Wnt signaling is required for myocardial differentiation of tailbudderived multipotent progenitors during tail extension. We also observed similar reduction in myod expression in the tail region (Suppl. Fig. S13a,b) . Because the heat-shock is performed at the 8-somite stage and some additional time is needed for induction of Dkk1 expression, it is not expected that Dkk1 would inhibit already established somites in the anterior and middle portions of the trunk. Wnt role in myocyte differentiation in multipotent progenitors in the tailbud has been previously demonstrated (Martin and Kimelman, 2012, Dev Cell 22, 223-32). Clearly, not all tailbud progenitors in the tailbud are etv2+ positive, and etv2 is not expressed in muscle progenitors in wild-type embryos, therefore Wnt role would not be unique to the etv2+ derived myocytes.
Comment: Page 14, lines 43-44: "our work has demonstrated that cells in the vascular endothelial lineage arise from multipotent progenitors in the lateral plate mesoderm which can differentiate into skeletal muscle in the absence of Etv2 function". To fully demonstrate this, lineage tracing would be required.

Response:
We revised this phrase to "suggested"

Minor comments
Page 10, lines 10-11:" Double heterozygous etv2gal4/ci33 embryos displayed a dramatic increase in GFP+ myocytes compared to etv2:gal4+/-embryos". It looks like a 3 to 4 fold increase in GFP+ myocytes which is far from "dramatic". Same remark for the sentence: "SU5402-treated embryos showed a dramatic decrease in the number of ectopic GFP+ muscle cells", data in which GFP+ myocytes decrease about two-fold." Response: These phrases were revised. Comment: Page 10, lines 23-24: "The number of myocytes was significantly increased in the etv2:gal4+/embryos injected with the previously validated scl morpholino (MO)". Does this sentence refer to the overall number of myocytes per embryos?
Response: No, this refers to etv2 ci32Gt -expressing myocytes. We revised this sentence to avoid confusion. Figure 5: it is very difficult to see co-expressing cells. Close-up images should be shown.

Comment:
Response: Higher magnification images have been included (currently Fig. 7) Comment: Page 14, lines 10-11: "previous studies have demonstrated the existence of a common vascular and endothelial progenitor, the hemangioblast". Is that not rather a common hematopoietic and endothelial progenitor?

Response:
We apologize, this error has been corrected.
--Reviewer #2: Major comments: 1. The authors need to find a way to isolate the etv2 positive cells and they employ a transgenic line that has gal4 inserted into the coding part of the gene. As a result, they use the heterozygous and homozygous cells of this line and treat the heterozygous as almost wild type. This can lead to wrong conclusions, especially if the authors don't show in detail that heterozygosity of etv2 does not disturb main functions. Another way to solve the problem would be to use the Tg(-2.3 etv2:GFP) line, perform scRNA-seq and compare it to their current results. It is understandable that these two lines do not mark exactly the same cell populations, but some commonalities should be present. In addition, the authors claim that the phenotype of their transgenic animals is more severe than the one of the actual mutants. This may be caused by deletion of regulatory elements that are located close to etv2 gene. Can the authors check the expression of these adjacent genes? Such a problem leads to an even more urgent need to sequence cells from the Tg(-2.3 etv2:GFP) or any other transgenic that could verify the current results.

Response:
We have performed detailed characterization of vascular development in the etv2 ci32Gt line and previously generated TgBAC(etv2:GFP) and Tg(-2.3 etv2:GFP) lines and we did not find significant differences in vascular development between etv2 ci32Gt heterozygous embryos and the other reporter lines (Chestnut and Sumanas, Dev Dyn 2019; doi: 10.1002/dvdy.130). The two differences that we have observed include the presence of a few GFP+ skeletal muscle cells and also a cluster of apoptotic cells observed in etv2 ci32Gt+/embryos which we attribute to the partial loss of etv2 function, and this is discussed in the current manuscript. We did perform single-cell RNA-seq of Tg(-2.3 etv2:GFP) embryos at similar embryonic stages using Fluidigm approach (Fig. 8). Despite the difference in sequencing platforms, we have identified in Tg (-2.3 etv2:GFP) embryos many of the same cell populations with the same marker genes, including endothelial progenitor cells, vascular endothelial cells (three different subpopulations were observed in Tg(-2.3 etv2:GFP) embryos likely due to better separation of marker genes because of greater sequencing depth), red blood cells, macrophages and presumptive tailbud progenitors. Cell populations which were observed in etv2 ci32Gt line but not identified in Tg(-2.3 etv2:GFP) embryos include cardiomyocytes, skeletal muscle cells, apoptotic cells and LPM progenitors. We did not observe any GFP expression in cardiomyocytes, skeletal muscle cells or apoptoting-looking cells in Tg(-2.3 etv2:GFP) or TgBAC (etv2:GFP) embryos, therefore the presence of these populations is a likely consequence of partial loss of etv2 function in etv2 ci32Gt embryos. Etv2 inhibition is known to result in EC apoptosis (Pham et al., 2007, Dev Biol 303, 772-783), and we have previously demonstrated that etv2-positive cells can differentiate into cardiomyocytes in etv2-inhibited embryos (Palencia-Desai et al., 2011, Development 138, 4721-4732). Regarding LPM progenitors, this population was not identified in Tg(-2.3 etv2:GFP) embryos either because fewer cells were sequenced in this line and low number of LPM cells could not be separated into a distinct cluster, or because this population may have low GFP expression as these cells may have just initiated etv2 expression. etv2 ci32Gt line is significately brighter than Tg(-2.3 etv2:GFP) line, likely due to the amplication of GFP expression by Gal4:UAS system, therefore low GFP cells may be missed in etv2:GFP line. We have now included the comparison of the sequencing results between the two different platforms in the Discussion section.
Comment: 2. The number of heterozygous and homozygous cells sequenced differs greatly between them. This maybe a problem in comparing these two experiments and identifying missing or enriched populations. Can the authors equilibrate these differences or prove that they do not affect the results?
Response: During bioinformatical analysis GFP+ cells from heterozygous and homozygous etv2 ci32Gt embryos were combined for clustering. After clustering, cells were then plotted on separate graphs based on their genotype. Thus they were analyzed together in the same experiment.
Comment: 3. The authors claim that Wnt and FGF signaling promote skeletal muscle fate. Can they show in their scRNA seq data that genes regulated by these pathways have altered expression?
Response: Wnt and FGF pathways can induce different effector genes depending on the cell type and signaling context. Myogenic regulator factors MyoD and Myf5 are known to be transcriptionally activated by Wnt signaling (Tajbakhsh S, et al. Differential activation of Myf5 and MyoD by different Wnts in explants of mouse paraxial mesoderm and the later activation of myogenesis in the absence of Myf5. Development. 1998;125:4155-4162; Chen AE, et al. Protein kinase A signalling via CREB controls myogenesis induced by Wnt proteins. Nature. 2005;433:317-322). In our scRNA-seq results, zebrafish homologs myod1 and myf5 are among the top markers expressed in zebrafish myocytes, and the number of myocytes is greatly increased in homozygous etv2 ci32Gt embryos.