Dermomyotome-derived endothelial cells migrate to the dorsal aorta to support hematopoietic stem cell emergence

  1. Pankaj Sahai-Hernandez
  2. Claire Pouget
  3. Shai Eyal
  4. Ondrej Svoboda
  5. Jose Chacon
  6. Lin Grimm
  7. Tor Gjøen
  8. David Traver  Is a corresponding author
  1. Department of Cell and Developmental Biology, University of California, San Diego, United States
  2. Department of Pharmacy, University of Oslo, Norway

Abstract

Development of the dorsal aorta is a key step in the establishment of the adult blood-forming system, since hematopoietic stem and progenitor cells (HSPCs) arise from ventral aortic endothelium in all vertebrate animals studied. Work in zebrafish has demonstrated that arterial and venous endothelial precursors arise from distinct subsets of lateral plate mesoderm. Here, we profile the transcriptome of the earliest detectable endothelial cells (ECs) during zebrafish embryogenesis to demonstrate that tissue-specific EC programs initiate much earlier than previously appreciated, by the end of gastrulation. Classic studies in the chick embryo showed that paraxial mesoderm generates a subset of somite-derived endothelial cells (SDECs) that incorporate into the dorsal aorta to replace HSPCs as they exit the aorta and enter circulation. We describe a conserved program in the zebrafish, where a rare population of endothelial precursors delaminates from the dermomyotome to incorporate exclusively into the developing dorsal aorta. Although SDECs lack hematopoietic potential, they act as a local niche to support the emergence of HSPCs from neighboring hemogenic endothelium. Thus, at least three subsets of ECs contribute to the developing dorsal aorta: vascular ECs, hemogenic ECs, and SDECs. Taken together, our findings indicate that the distinct spatial origins of endothelial precursors dictate different cellular potentials within the developing dorsal aorta.

Editor's evaluation

In this important study, the authors identify an additional source of dorsal aorta endothelium derived from the somites that are conventionally thought of as defined blocks of skeletal muscle. The authors present convincing data that these "somite-derived endothelial cells" (SDECs) arise from bipotential precursors in the dermamyotome that give rise to endothelium or muscle. The authors conclude that these cells support but do not produce emergent hematopoietic stem cells in the dorsal aorta, findings that will be of interest to stem cell and developmental biologists.

https://doi.org/10.7554/eLife.58300.sa0

Introduction

The primitive vascular network, which integrates into all organ systems in the developing organism, arises from endothelial precursors termed angioblasts (Risau and Flamme, 1995). To form a functional vascular network, angioblasts must first differentiate into a variety of distinct arterial and venous endothelial cell (EC) types, including hemogenic, endocardial, and blood brain barrier ECs (Aird, 2007; Herbert and Stainier, 2011). EC differentiation is thought to initiate midway through somitogenesis, during the migration of angioblasts to the embryonic midline, where they coalesce to form the vascular tube (Herbert et al., 2009; Isogai et al., 2003; Jin et al., 2005). The current view is that angioblasts are initially equipotent and undergo successive steps of differentiation that, along with cues from local microenvironments, give rise to specialized subsets of ECs (Atkins et al., 2011; Marcelo et al., 2013). This view, however, does not consider possible differences in EC function due to different developmental origins. Instead, exposure to embryonic signaling cascades mediated via Wnt (Hübner et al., 2017), Hedgehog (Hh) (Gering and Patient, 2005; Vokes and McMahon, 2004; Wilkinson et al., 2012; Williams et al., 2010), Vascular Endothelial Growth Factor (VEGF) (Casie Chetty et al., 2017; Hong et al., 2006; Lawson et al., 2003; Lawson et al., 2002; Wythe et al., 2013), and Notch molecules are thought to differentially instruct equipotent angioblasts to each distinct endothelial cell fate (Fang et al., 2017; Lawson et al., 2001; Siekmann and Lawson, 2007; Zhong et al., 2001).

Lateral plate mesoderm (LPM) is known to be the primary source of ECs across vertebrate phyla (Potente and Mäkinen, 2017). However, recent findings suggest that ECs can arise from distinct mesodermal derivatives, including extraembryonic-derived erythromyeloid progenitors (EMPs) that contribute extensively to the murine kidney vasculature (Plein et al., 2018). Furthermore, lineage tracing studies in zebrafish have demonstrated that ECs supportive of hematopoiesis can derive from endoderm (Nakajima et al., 2023). To better understand the development of EC subsets, we performed single-cell RNA sequencing (scRNA-seq) of ECs purified by flow cytometry over a range of time points during zebrafish embryogenesis. Following the end of gastrulation, the earliest developmental timepoint that nascent ECs can be identified, a variety of distinct molecular signatures were present, including those similar to kidney-specific ECs, brain-specific ECs, and paraxial mesoderm-derived ECs. These data suggest that specific EC fates may be specified much earlier than previously appreciated and that embryonic origins may dictate the development of tissue-specific EC subsets.

Within these datasets, we detected a signature indicative of paraxial mesoderm (PM) origins; we thus further explored this possibility since previous findings suggested that somite-derived ECs (SDECs) are intimately involved with both the development of the dorsal aorta and the emergence of hematopoietic stem and progenitor cells (HSPCs; Nguyen et al., 2014; Pouget et al., 2006). There is evidence in rodents and other amniotes that PM generates a contingent of ECs that contribute to embryonic vasculature (Ambler et al., 2001; Esner et al., 2006; Noden, 1989; Pardanaud et al., 1996; Pouget et al., 2008; Pouget et al., 2006; Wilting et al., 1995; Yvernogeau et al., 2012). Specifically, these ECs arise from a transient somitic compartment known as the dermomyotome (Eichmann et al., 1993; Ema et al., 2006; Pouget et al., 2008; Yvernogeau et al., 2012), where skeletal hypaxial muscle progenitors (skMPs) reside (Tozer et al., 2007). In the chick embryo, SDECs migrate to the dorsal aorta to replace hemogenic endothelium that exits the ventral aorta as emerging HSPCs (Pouget et al., 2006). A recent report in zebrafish has also suggested the existence of an endothelial-producing compartment (termed ‘endotome’) within the central somite that generates SDECs that migrate to the dorsal aorta (Nguyen et al., 2014). Whereas experiments in birds have suggested that SDECs only replace hemogenic endothelium, the experiments in zebrafish by Nguyen et al., 2014 suggested that SDECs may help induce the HSPC program. However, the genetic profile of SDECs and their possible role in HSPC induction remain poorly understood.

Here, using a combination of molecular, genetic, and computational approaches, we characterize a rare population of SDECs in zebrafish that emerges from the trunk dermomyotome. Trunk SDECs migrate and contribute exclusively to rostral regions in the dorsal aorta. Within the somite, EC-fate acquisition occurs sequentially, concomitant with the epithelialization of each somite and the migration of angioblasts toward the midline of the embryo. We show that Wnt signaling is a key regulator of the distribution of ECs within the somite, whereas Notch signaling is necessary for skeletal muscle progenitor cell maintenance. Finally, epistasis experiments indicate that SDECs arise from bipotent precursors within the somite, with skMPs showing competency to become ECs in a meox1- and npas4l (cloche Stainier et al., 1995) – dependent manner. While SDECs integrate into the dorsal aorta, indelible lineage tracing demonstrates that they do not harbor hematopoietic potential. Instead, they appear to act at least in part as a developmental niche to facilitate HSPC emergence. Collectively, our findings indicate that distinct EC subsets are molecularly and functionally distinct as early as the end of gastrulation and describe a cellular mechanism by which the somite regulates the production of hematopoietic precursors.

Results

Molecular differences in endothelial cells underlie cellular diversity within the vasculature

In zebrafish, most ECs originate from the LPM (Jin et al., 2005). However, recent studies have suggested that somites also produce ECs that integrate into the vascular cord (Nguyen et al., 2014), but the role and nature of these ECs remain incompletely defined. To better understand how different ECs are specified in the zebrafish embryo, we utilized single-cell RNA sequencing (scRNA-seq) on a collection of purified ECs from distinct vascular transgenic animals to represent diverse endothelial cohorts. Specifically, we purified cells via fluorescence-activated cell sorting (FACS) from TgBAC(etv2:Kaede)ci6, Tg(fli1:DsRed)um13; Tg(tp1:GFP)um14, and Tg(drl:H2B-dendra) (hereafter termed mixed-vasculature) (Kohli et al., 2013; Mosimann et al., 2015; Parsons et al., 2009; Villefranc et al., 2007) between 22 and 24 hpf, and performed scRNA-seq on each sample. Following quality control (QC), we merged the individual datasets. Altogether, we obtained 1994 single cells that belonged to the endothelial lineage and an additional 5446 erythroid and myeloid cells, which we omitted from our analyses as our focus was on the EC subsets. Following unsupervised clustering of single-cell transcriptomes, we identified erythroid cells, myeloid cells, and eight distinct endothelial cell clusters (Figure 1A and B), which we named based on known marker genes within established tissue lineages (Figure 1—source data 1 and Figure 1—figure supplement 1). First, we identified a sizeable endothelial cluster, general endothelium (GE), that co-expressed canonical endothelial genes and a putative hemogenic endothelium (HE) cluster that co-expressed endothelial and hematopoietic genes, including etv2, dab2, and stab2. We identified a smaller EC cluster corresponding to pre-HSCs based on expression of genes such as cmyb, cebpa, and gfi1aa. We found two clusters, brain vascular endothelial cells 1 and 2 (BVECs-I and BVECs-II), that co-expressed brain and neuronal-associated genes, including tncb, elavl3, and fmoda. Another cluster likely represents endocardial EC (EEC) based upon co-expression of canonical heart genes, including gata5, hand2, and spock3. A kidney vascular EC (KVECs) cluster co-expressed kidney-associated genes, such as pax2a, ap1m2, and myh9a. Finally, we identified an EC cluster that contained the expression of paraxial mesoderm (PM) signature genes, including igf2a, fbn2b, and fn1a. This PM signature suggested that this population may represent somite-derived endothelial cells (SDECs). We used differential expression analysis among clusters to identify distinct gene programs which are enriched within each specific subset (Figure 1B and Figure 1—source data 1).

Figure 1 with 3 supplements see all
Cell-type-specific endothelial cell markers highlight cellular diversity within the vasculature.

(A) Uniform manifold approximation projection (UMAP) plots of scRNA-seq data of total endothelial lineage cells collected from TgBAC(etv2:Kaede)ci6, Tg(fli1:DsRed)um13; Tg(tp1:GFP)um14, and Tg(drl:H2B-dendra) embryos at 22–24 hpf. Clusters were named according to their gene expression: Erythroid, Lymphoid, General Endothelium (GE), Hemogenic Endothelium (HE), Pre-HSCs, Brain Vascular Endothelial Cells (BVECs-I and BVECs-II), Kidney Vascular Endothelial cells (KVECs), Endocardial Endothelial Cells (EECs), and somite-derived endothelial cells (SDECs). Color-coded marker gene expression levels are shown on corresponding clusters. A pink circle highlights the SDEC cluster. (B) Expression heatmap of 22–24 hpf single-cell transcriptome shows the top predicted differentially expressed marker genes across the different clusters. A red box highlights the SDEC cluster. (C’,C”) A list of somite-annotated genes was curated from the AmiGo annotation database and compared with the SDEC transcriptome. 32 genes were commonly expressed. Interestingly, several of these 32 genes were enriched within the SDEC cluster (C”; white boxed genes, D; enlarged circles).

Figure 1—source data 1

Transcriptomes of all endothelial cell clusters, myeloid, and erythroid cells.

The transcriptomes were extracted and read from cells purified collected from TgBAC(etv2:Kaede)ci6, Tg(fli1:DsRed)um13; Tg(tp1:GFP)um14, and Tg(drl:H2B-dendra) embryos at 22–24 hpf.

https://cdn.elifesciences.org/articles/58300/elife-58300-fig1-data1-v1.xlsx
Figure 1—source data 2

Comparison between Genes expressed in EC clusters (e.g. SDECs) and gene annotation of the same anatomical structure (e.g., somite) based on annotation from AmiGO (Consortium, 2019).

Expression of overlapping genes was compared in the same cluster (e.g. SDECs) between 15 ss and 22 hpf and divided into DE genes that are upregulated (Up) on downregulated (Down). Each DE group was then annotated using AmiGO (Consortium, 2019).

https://cdn.elifesciences.org/articles/58300/elife-58300-fig1-data2-v1.xlsx

To test our clustering predictions and verify the somitic origin of the SDEC cluster, we compared its genetic signature to annotated genes expressed within the somite compartment based on the AmiGO annotation database (Consortium, 2019; Figure 1—source data 2). Our comparative analysis resulted in 32 genes that were commonly expressed within the SDEC fraction and the AmiGO annotation database (Figure 1C’ and C”). Interestingly, several of these 32 genes were enriched within the SDEC cluster (Figure 1C” white boxed genes and 1D). We repeated our analysis on the BVEC-I and KVEC clusters. To validate their cluster identity, we compared their transcriptomes to the annotated brain and kidney genes on AmiGO. Our comparative analysis resulted in 32 and 63 common genes, respectively (Figure 1—figure supplement 2A–C” and Figure 1—figure supplement 3A–C”). Furthermore, as in the SDEC cluster, our analyses confirmed that many of these genes were enriched in both clusters (Figure 1—figure supplement 2D and Figure 1—figure supplement 3D). Thus, our scRNA-seq analysis approach identified various tissue-specific EC subsets in the 22–24 hpf embryo, including HE, pre-HSCs, and a rare population of SDECs.

Identified EC populations can be traced back as early as the tailbud stage

After identifying the discrete genetic signatures of each endothelial cell cluster at 22–24 hpf, we queried if these signatures could also be identified at earlier stages. We followed the same protocol to purify and scRNA-seq samples collected at different developmental stages, namely drl:H2B-dendra+ embryos at tailbud (10 hpf) and 12 somite stages (ss) (15 hpf), as well as etv2:Kaede+ embryos at 15 ss (16.5 hpf) (Figure 2A–D). Next, we crossed-referenced the transcriptomes of all EC subsets from the mixed vasculature 22–24 hpf samples (Figures 1A and 2A) with those of cells purified from earlier tailbud, 12-, and 15-ss samples and clustered them. We identified EC clusters with distinct transcriptomes as early as the tailbud stage, suggesting that EC specification started by the end of gastrulation and the initiation of somitogenesis. While the EC clusters were still heterogeneous prior to the 12 ss stage, we could identify most EC subsets, including HE, pre-HSC, KVECs, BVECs, and SDECs, after the 15 ss when their transcriptomes became more defined (Figure 2B). Together, these data suggest a gradual EC specification process that begins as early as the tailbud stage (Figure 2B–D).

Figure 2 with 1 supplement see all
Cellular diversity within the vasculature can be traced back to the tailbud stage.

(A) Uniform manifold approximation projection (UMAP) plots of scRNA-seq data of total endothelial lineage cells collected from TgBAC(etv2:Kaede)ci6, Tg(fli1:DsRed)um13; Tg(tp1:GFP)um14, and Tg(drl:H2B-dendra) embryos at 22–24 hpf. Clusters were named according to their gene expression: General Endothelium (GE), Hemogenic Endothelium (HE), Pre-HSCs, Brain Vascular Endothelial Cells (BVECs-I and BVECs-II), Kidney Vascular Endothelial cells (KVECs), Endocardial Endothelial Cells (EEC), and somite-derived endothelial cells (SDECs). Color-coded marker gene expression levels are shown on corresponding clusters. A pink circle highlights the SDEC cluster. (B–D) Referenced uniform manifold approximation projection (RefUMAP) plots of scRNA-seq data of total endothelial lineage cells collected from etv2:Kaede+ embryos at 15 ss (B) and drl:H2B-dendra+ embryos at 12 ss (C) and tailbud stage (D). By cross-referencing the transcriptomes of EC subsets at each developmental stage to the 22–24 hpf ECs, we identified EC clusters with distinct transcriptomes as early as the tailbud stage. (E–H) Comparison of expression patterns of EC populations from early TgBAC(etv2:Kaede)ci6 15 ss, and later 22 hpf etv2:Kaede+ in the 32 overlapping genes between the SDEC transcriptome data and the AmiGo somite annotated genes. (E,F) Representative genes that were upregulated in the etv2:Kaede+ 22 hpf samples compared to the 15 ss sample (F) and their suggested role in EC differentiation, according to GO biological processes (E). (G,H) Representative genes that were downregulated in the etv2:Kaede+ 22 hpf samples compared to the 15 ss sample (H) and their suggested role in somitogenesis, according to GO biological processes (G). The expression and downregulation of somitic genes within etv2+ ECs between 15 ss and 22 hpf highlight their somitic origin and loss of myogenic cell fate.

To explore how early tissue-specific EC segregation occurs, we compared the changes in expression patterns of EC populations from early TgBAC(etv2:Kaede)ci6 15 ss and later 22 hpf scRNA-seq data. Specifically, we focused on the changes in expression in the 32 overlapping genes between the scRNA-seq transcriptome data and the AmiGo annotated genes (Figure 1C’ and Figure 1—source data 2). From the 32 SDEC genes compared, 16 genes were upregulated in the etv2:Kaede+ 22 hpf sample compared to the 15 ss sample (Figure 2E and F), whereas 16 genes were downregulated (Figure 2G and H). GO function analysis indicated a role for upregulated genes in arterial EC differentiation, whereas downregulated genes were involved in somitogenesis (Figure 2E and G, respectively). The expression and downregulation of somitic genes within etv2+ ECs between 15 ss and 22 hpf highlight their somitic origin and the loss of myogenic cell fate of these cells as they emerge from the somite compartment.

We repeated our analysis on the BVEC-I and KVEC clusters to validate this observation by comparing samples from etv2:Kaede+ 15 ss and 22 hpf. In the BVEC-I cluster, we found 17 upregulated genes that indicate epithelium development, whereas seven genes indicating brain development were downregulated (Figure 2—figure supplement 1A–D). In the KVEC cluster, we found 18 upregulated genes indicative of epithelial cell differentiation in the kidney and 14 downregulated genes indicating kidney development (Figure 2—figure supplement 1E–H). As with SDECs, both BVEC-I and KVEC cells showed evidence of their commitment to a brain and kidney EC fate as early as 15 ss.

Rare endothelial cells emerge from trunk somites located above the dorsal aorta

Our transcriptome analysis identified that SDECs arise as a distinct EC population as early as the tailbud stage. Next, we aimed to characterize the development of SDECs and their potential to contribute to DA formation at later stages. To label somitic cells and follow their trajectories, we crossed a photoconvertible Tg(actb2:nls-Eos) animal with a Tg(fli1:eGFP)y1 vasculature reporter line (Cruz et al., 2015; Isogai et al., 2003). Resulting embryos ubiquitously express a photoconvertible nuclear GFP that can be instantly converted to RFP by UV exposure. To determine which somites generate SDECs and to quantify how many SDECs are made by each somite pair, we collected embryos at different somitogenesis stages ranging from 4 ss to 18 ss and converted single somite pairs using directed UV exposure via confocal microscopy. We avoided converting any LPM-derived EC on the lateral or ventral sides of the somite (Figure 3A and A’). Embryos were allowed to develop to 32–36 hpf, then imaged and examined for the presence of RFP+ SDECs within the dorsal aorta (Figure 3B). We observed that the trunk somites (numbers 5–18), located above the yolk tube extension generated the most SDECs. In contrast, photoconverted somites anterior to the 4th somite rarely gave rise to SDECs (Figure 3C and D). In sum, most SDECs were generated by somites 5–18. Each somite pair contributed between 0–6 SDECs to the DA (Figure 3D and Figure 3—source data 1). Interestingly, we occasionally noticed SDECs within the roof of the DA, suggesting a migratory path that passes through the roof of the DA before moving to the floor of the DA, consistent with previous reports (Zhao et al., 2022). On rare occasions, we also detected SDECs in the posterior caudal vein or intersomitic vessels, consistent with previous findings (Nguyen et al., 2014).

Rare SDECs emerge from trunk somites and migrate to the dorsal aorta.

(A–D) Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 embryos were collected at developmental stages ranging from 4 to 18 ss. (A) Newly developed posterior somite pairs were selected by setting a region of interest and photoconverted by UV light. (A’) A sagittal section through a pair of converted somite showing somite-specific conversion and lack of converted LPM-derived ECs. (B) At 32 hpf, embryos were laterally staged, and images of the dorsal aorta taken. SDECS were quantified in the dorsal aorta by examining individual z-stacks and visualizing colocalization of fli1+; actb2:nlsEosRFP converted cells in Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 embryos (Ci-Cv). In fli1:eGFP - embryos, we identified SDECs by observing actb2:nlsEosRFP cells on the floor of the DA based on the brightfield channel. (D) We observed that trunk somites (numbers 5–18), located above the yolk tube extension, generated the most SDECs. Each somite pair contributed between 0–6 SDECs to the DA. s, somites; DA, dorsal aorta. In each converted somite pair, n6, with each point representing the SDEC count from one embryo. The median for each somite pair is indicated as a column, and the standard error of the mean (SEM) is indicated as an error bar.

Figure 3—source data 1

A table summarizing all converted somite pairs and SDECs found in Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 embryos that were included in the final SDECs quantification assay.

Embryo’ somites were converted at developmental stages ranging from 4 to 18 somite stage (Column C) and imaged at 32–36 dpf. The imaging date (Column A), sample number within a cohort (Column B), and the number of observed SDECs (Column D) used for the quantification were documented, and the quantification and presented graph were done in Prism 9 (GraphPad).

https://cdn.elifesciences.org/articles/58300/elife-58300-fig3-data1-v1.xlsx

Somites give rise to etv2+ endothelial cells concomitant with somite epithelialization

Next, we aimed to further characterize the development of SDECs by following etv2:eGFP+ cells and their trajectories from the somite. As the SDEC subset became more distinct between 12–15 ss, we focused our analysis on that developmental time frame. To do so, we used confocal time-lapse imaging of Tg(etv2.1:eGFP)zf372 embryos (Veldman and Lin, 2012) injected with mOrange2:CAAX mRNA. These embryos have early ECs marked by GFP with cell boundaries demarcated by mOrange2. As previously described (Veldman and Lin, 2012), beginning at the 10 ss, we observed a line of etv2:eGFP+ along the most medial part of the LPM (Figure 4A). We did not detect any etv2:eGFP+ in the somites at this stage. Initiating at the 12 ss, we noted etv2:eGFP+ in the lateral lip of the somitic compartment (Figure 4B). After the onset of etv2:eGFP expression, within the next 3 hr, somite-derived etv2:eGFP+ rounded up and delaminated from the somite, then integrated into the cohort of LPM-derived etv2:eGFP+ that migrated toward the midline (Figure 4C–E). Time-lapse imaging of Tg(etv2.1:eGFP)zf372; Tg(phlbd1:Gal4-mCherry) embryos resulted in similar observations that support the emergence of ECs from the somites as early as 12 ss (Figure 4—video 1 and Figure 4—video 2).

Figure 4 with 2 supplements see all
Endothelial cells emerge from the dermomyotome at 12 ss.

(A–E) Time-lapse imaging from a dorsal view of Tg(etv2.1:EGFP)zf372 embryos injected with mOrange:CAAX mRNA and imaged between 10 ss and 15 ss. (A) The expression of Etv2:GFP+ cells is visible along the LPM region (arrow) at 10 ss. At this stage, no Etv2:GFP+ cells are visible in the somites. (B) Starting at 12 ss, the first Etv2:GFP+ SDECs are detected in the lateral lip of the dermomyotome (arrowheads). Simultaneously, the LPM Etv2:GFP+ cells start migrating to the midline. (C) Soon after emergence, SDECs change shape and become rounder (arrowheads). (D–E) Etv2:GFP+ SDECs bud off from the somite as individual cells (arrowhead). (F) Dorsal view of a 12 ss embryo that was submitted to double fluorescent in situ hybridization for muscle progenitor maker pax3a (green) and endothelial marker etv2 (red). pax3a expression reveals the dermomyotome compartment that contains muscle progenitor cells. An etv2+ SDEC (red and arrowhead) is found in the dermomyotome, co-expressing pax3a (green), showing colocalization of an endothelial and muscle progenitor cell marker. We observed 1–2 etv2-positive cells per somite in each of the embryos examined (n=6). (G) Somitic etv2+ SDECs (green) do not co-express the muscle differentiation marker myoD (red), suggesting that etv2 expression is restricted to the muscle progenitor region of the somite. Dashed white lines delimitate somite from the LPM (arrow). We observed 1–2 etv2-positive cells per somite in each of the embryos examined (n=6). s, somites; LPM, lateral plate mesoderm; SDECs, somite-derived endothelial cells.

In mice and chick, SDECs emerge from the same region as skeletal muscle progenitor cells in the hypaxial dermomyotome compartment (Pouget et al., 2006; Tozer et al., 2007). To examine the spatial origins of SDECs in the zebrafish, we performed double fluorescent in situ hybridization (FISH) for the endothelial marker etv2 and the skeletal muscle progenitor marker pax3a (Relaix et al., 2005). At 12 ss, we observed colocalization of etv2 and pax3a within rare cells in the somite (Figure 4F). However, FISH for etv2 and myod, a marker of differentiated muscle cell types (reviewed in Hernández-Hernández et al., 2017), did not show colocalization (Figure 4G). These results suggest that SDECs emerge within the somite from precursor cells shared with the muscle lineage. We observed that somitic etv2+ were localized specifically within the dermomyotome region, the muscle progenitor cell compartment of the somite. As this was previously shown in other organisms, we suggest that a conserved mechanism of SDEC generation is shared amongst vertebrates (Mayeuf-Louchart et al., 2016; Mayeuf-Louchart et al., 2014; Pouget et al., 2006).

The dermomyotome contains progenitors with muscle and endothelial potential

Since our results above suggested the presence of bipotent progenitors with competence for muscle and endothelial cell differentiation, we sought to determine if blocking skeletal muscle differentiation may lead to enhanced SDEC generation by knocking down mesenchyme homeobox 1 (meox1), an essential regulator of muscle cell formation (Mankoo et al., 2003). To knock down meox1, we used a meox1 translation blocking morpholino (MO) that can block translation of a functioning Meox1 protein but does not affect the transcription of a viable meox1 mRNA; thus, it can still be detected by an RNA probe. Using FISH to label meox1 and etv2 expression, we observed ectopic expression of etv2 in morphant animals and an increased number of SDECs (Figure 5A–F). Like normal SDECs, ectopic etv2:eGFP+ emerged at 12 ss. However, in contrast to wild type, meox1 morphants showed SDEC generation as late as 23 hpf in Tg(phldb1:mCherry); Tg(etv2.1:eGFP)zf372 embryos (Distel et al., 2009; Kobayashi et al., 2014; Figure 5G and H). Thus, meox1 loss of function leads to enhanced and prolonged production of SDECs.

Figure 5 with 1 supplement see all
notch is required for the maintenance of a bipotent skeletal muscle progenitor population in the somite.

(A–F) Dorsal view of 12 ss control (A–C) and meox1 morphant embryos (D–F). Embryos were submitted to double fluorescent in situ hybridization for meox1 (green) and etv2 (red). In control and morphant embryos, meox1; etv2 double-positive cells are detected within the somite compartment (arrowheads). (C,F) Overlay of meox1 (green), etv2 (red), and DAPI (blue). (D–F) Knockdown of meox1 results in ectopic formation of double-positive cells within the somite (arrowheads). We observed 3–4 etv2 positive cells per somite in the meox1 morphants compared to 1–2 etv2-positive cells per somite in the siblings (n=3). (G–H) Time-lapse imaging of a 22 hpf Tg(etv2.1:eGFP)zf372; Tg(phldb1:mCherry) embryo, injected with meox1 morpholino and mOrange2:CAAX mRNA to delineate cell boundaries. Knockdown of meox1 results in an extension of the period that the dermomyotome can generate Etv2:GFP+ cells (arrowheads). (I–K) Cross section of 12 ss Tg(etv2.1:eGFP)zf372 embryo. In absence of meox1 (J), ectopic Etv2:GFP+ cells are visible in epithelialized layer of the somites, compared to controls (I). In embryos coinjected with mib and meox1 morpholinos, the number of Etv2:GFP+ cells within the somite compartment (dotted line) is substantially increased (arrowheads) (K), suggesting that Notch signaling is dispensable for SDEC specification. (L–N) Lateral view of 12 ss embryos analyzed by FISH for meox1 (green), etv2 (red), and DAPI (blue). In notch3+/- heterozygote controls (L) and notch3-/- mutant embryos (M), etv2+ SDECs are detected in the somites. (N) notch3-/- mutant embryos co-injected with meox1 morpholino results in ectopic formation of etv2; meox1 double positive cells (arrowheads). We observed 2–4 etv2-positive cells per somite in the notch3 mutants and >6 etv2-positive cells in the notch3 mutants; Mib morphants (n=3). (O) qRT-PCR in 24 hpf notch3-/- mutant embryos and sibling controls. Genetic ablation of notch3 results in decreased expression of muscle progenitor markers pax3a and pax7b; increased expression of muscle differentiation genes, myod and myog, and endothelial markers, etv2, and fli1. Asterisks denote a statistically significant difference (p<0.05, unpaired, two-tailed Student’s t-test; n=3.) (P,Q) notch3-/- mutant embryos show premature expression of MyoHII in 48 hpf embryos (Q) compared to sibling controls (P). (R) Summary cartoon for the role of Notch signaling in the maintenance of bipotent-muscle progenitors (bipotent muscle progenitors in purple and green; muscle cells in green; SDECs in purple). s, somites; LPM, lateral plate mesoderm; SDECs, somite-derived endothelial cells.

Previous work in mice identified the Notch signaling pathway as a positive regulator of muscle fate in the somite (Mayeuf-Louchart et al., 2014). We tested the requirement for Notch signaling in zebrafish SDEC generation by knocking down the essential notch regulator mindbomb (mib) (Itoh et al., 2003) in Tg(etv2.1:eGFP)zf372 animals and examining SDEC formation between 12–14 ss. While loss of meox1 alone led to the ectopic formation of a few SDECs (Figure 5I and J), knockdown of both mib and meox1 led to a profound expansion of etv2:eGFP+ within the somite (Figure 5K).

To examine the role of Notch signaling more precisely in these cells, we examined notch3fh332/fh332 animals (referred to as notch3-/-) since notch3 is the primary notch receptor in the somites at this stage (Kim et al., 2014). We injected meox1 MO into notch3-/- embryos and performed FISH for meox1 and etv2 (Figure 5L–N). Using this strategy, we observed that in notch3-/- homozygous mutant embryos, the formation of SDECs was not impaired (Figure 5L and M). However, following the combined loss of notch3 and meox1, we observed the ectopic formation of etv2+; meox1+ double positive cells by FISH (Figure 5N and Figure 5—figure supplement 1).

To determine which cell populations were explicitly affected by the loss of notch3, we compared the expression levels of markers for muscle progenitors (pax3a, pax7b), differentiated muscle cells (myod, myog), and differentiated endothelial cells (etv2, fli1a) by qRT-PCR in notch3 mutant and heterozygous embryos at 24 hpf (Figure 5O). In notch3-/- animals, we observed a decrease in the expression of muscle progenitor markers (pax3a, pax7b) concomitant with an increase in the expression of muscle differentiation markers (myod, myog) and endothelial differentiation markers (etv2, fli1a) (Figure 5O). Furthermore, we observed premature expression of the muscle differentiation marker myoHII (Beier et al., 2011; Salucci et al., 2015; Sjöblom et al., 2008) by antibody staining at 48 hpf in notch3-/- animals (Figure 5P and Q). Together, these results suggest that Notch signaling is dispensable for the specification of SDECs but is required for pax3+ and pax7+ skeletal muscle progenitor maintenance, as previously described (Relaix et al., 2005; Zhang et al., 2021). Moreover, the absence of notch3 leads to premature differentiation of muscle and SDEC fates (Figure 5R).

npas4l is required for the formation of SDECs npas4l (cloche) is regarded as the most upstream gene required for blood and endothelial cell specification (Reischauer et al., 2016; Stainier et al., 1995). Therefore, we sought to determine if npas4l is also required for SDEC development. WISH in 12 ss embryos showed a complete absence of etv2 expression along the embryonic A-P axis in Clos9/s9 mutants (referred to as cloche) compared to controls (Figure 6A–B), showing that npas4l function is required for both PLM-derived ECs and PM-derived SDECs. Knockdown of meox1 to increase SDEC differentiation from bipotent somitic progenitors was insufficient to rescue SDEC generation, indicating npas4l function is necessary to generate the SDEC lineage. In addition, suppressing muscle specification in the absence of npas4l was insufficient to induce SDEC fate (Figure 6C–D). Similarly, in Tg(etv2.1:eGFP)zf372; cloche embryos injected with mib MO and meox1 MO, we observed a complete absence of etv2:eGFP+, compared to cloche control siblings (Figure 6—figure supplement 1).

Figure 6 with 1 supplement see all
npas4l is required for the specification of SDECs.

(A–D) WISH for etv2 in 12 ss npas4l-/- (cloche) mutant and control embryos. (B) cloche mutant embryos show an absence of etv2 expression along the A-P axis of the embryo, compared to sibling control (A). (D) Similarly, cloche mutant embryos injected with meox1 morpholino show loss of etv2 expression, compared to sibling control (C). (E) qRT-PCR of cloche mutant embryos shows expected loss of endothelial genes (fli1 and etv2) and concomitant increase of muscle differentiation genes (myod and myog), compared to sibling control. All genes analyzed between cloche mutant and cloche het embryos showed a statistically significant difference (p<0.001, unpaired, two-tailed Student’s t-test; n=3.) (F, G) Summary cartoon for the effect of npas4l on endothelial cell competence in PM progenitors (early mesoderm progenitor in grey; bipotent muscle progenitor in purple and green; muscle cells in green; endothelial cells in purple). LPM, lateral plate mesoderm; SDECs, somite-derived endothelial cells.

Lastly, we performed qRT-PCR for endothelial and muscle cell genes from 48 hpf cloche mutant and control embryos (Figure 6E). As expected, cloche mutant embryos showed decreased endothelial gene expression (etv2 and fli1a). Interestingly, we also observed a concomitant increase in the expression of skeletal muscle differentiation genes (myod and myogenin) (Figure 6E). Together, these results confirm that npas4l is required for specification of all EC subsets, including SDECs, and suggest a negative regulatory role in muscle differentiation in the shared bipotent progenitors of SDECs and skeletal muscle cells (Figure 6F and G).

Wnt signaling regionalizes the formation of SDECs

Previous work has shown that Wnt signaling is required for the differentiation of muscle progenitors through activation of the required skeletal muscle factor myf5 (reviewed in von Maltzahn et al., 2012). In addition, inhibition of Wnt signaling in early presomitic mesoderm leads to an increase in endothelial cells that can integrate into the zebrafish vasculature (Veldman et al., 2013), suggesting that Wnt signaling may also act later in the somite to balance muscle and endothelial cell production from shared muscle progenitor cells. To investigate whether Wnt signaling is active in meox1+ muscle progenitor cells during SDEC development, we performed FISH for meox1 in the background of Tg(7xTCF-Xla.Siam:GFP)ia4, a destabilized Wnt/TCF reporter line +. At the 12 ss, we observed cells positive for GFP and meox1 (Figure 7A), indicating that Wnt signaling is active while SDEC fate decisions occur. Next, to determine if Wnt inhibition affects SDEC development, we treated Tg(etv2.1:eGFP)zf372 embryos with IWP-L6, a potent inhibitor of Wnt protein secretion (Wang et al., 2013), from 2 to 12 ss. Inhibition of the Wnt signal was confirmed by qRT-PCR for the canonical target gene axin2 (Figure 7B), which led to a decrease in axin2 and an increase in etv2 transcripts. In addition, we observed a slight reduction in meox1 expression; however, it was not statistically significant. By examining serial sections, Wnt inhibition led to the formation of ectopic etv2:eGFP+ in the somites (Figure 7C–D’). Finally, we also determined that Wnt signaling is required for meox1 expression (Figure 7E–F’). Together, these results demonstrate that Wnt signaling compartmentalizes the formation of SDECs within the somite to the most lateral region. In addition, they show that Wnt inhibition results in downregulation of meox1 and the expansion of etv2:eGFP+.

Wnt signaling is required for the regionalization of SDECs.

(A) FISH for meox1 (red) and antibody staining for a destabilized Wnt/TCF reporter line (green) show co-expression of GFP and meox1 within the somite. (B) Inhibition of Wnt signaling using the chemical inhibitor IWP2 from 2 ss to 15 ss results in decreased expression of axin2 and meox1 with a concomitant increase of the expression of etv2 by qRT-PCR. We observed a reduction in meox1 expression, although not statistically significant. All genes analyzed between Wnt inhibitor and control embryos, except meox1, showed a statistically significant difference (p<0.001, unpaired, two-tailed Student’s t-test; n=3.) (C–D’) Cross section of Tg(etv2.1:EGFP)zf372 embryos treated with IWP2 from 2 ss to15 ss. wnt inhibition results in ectopic formation of Etv2:GFP+ cells within the somite (arrowheads). (C’,D’) enlargement of somite compartment (dashed lines). Notice LPM cells migrating under the sclerotome (arrows). (E–F’) IWP2 control and treated embryos. wnt inhibition results in decreased expression of meox1 by WISH, compared to control embryos. (E,F) Lateral view. (E’,F’) Dorsal view. s, somites; LPM, lateral plate mesoderm; SDECs, somite-derived endothelial cells.

SDECs integrate into the dorsal aorta but do not generate HSPCs

Our results confirmed previous studies in the chick (Pouget et al., 2006) and zebrafish (Nguyen et al., 2014) that ECs emerge from the somites and incorporate into the DA (Figure 3). Next, we were interested in understanding the role of SDECs in the developing zebrafish vasculature and hematopoiesis. To this end, we obtained a PM-specific Gal4 transgenic line, Tg(tbx6:Gal4FF:GFP-nls) (Yabe et al., 2016), which has been shown to recapitulate tbx6 mRNA expression (referred to as tbx6:Gal4). We crossed this line to Tg(UAS-Cre) (Butko et al., 2015) to drive Cre recombinase only within the PM. We then crossed it to a ubiquitously expressed reporter line Tg(actb2:loxP-BFP-loxP-DsRed)sd27 (Kobayashi et al., 2014), which upon genetic recombination, switches from a BFP to DsRed cassette (abbreviated as A2BD for clarity). Upon confocal imaging of the triple transgenic tbx6:Gal4; Tg(UAS-Cre); A2BD animals at 48 hpf, we observed DsRed+ cells within the region of the axial vasculature (Figure 8A–C). To confirm that these cells were SDECs arising from a tbx6+ somitic population, we crossed the tbx6:Gal4; Tg(UAS-Cre) line with a previously published endothelial-specific CFP-to-YFP switch line, TgBAC(kdrl:LOXP-AmCyan-LOXP-ZsYellow) (referred to as kdrl:CSY) (Zhou et al., 2011). Imaging the DA in tbx6:Gal4; Tg(UAS-Cre); kdrl:CSY triple transgenic embryos at 4 dpf showed YFP+ cells corresponding to SDECs to localize preferentially to the trunk region of the dorsal aorta (Figure 8D–G and K). We found this labeling pattern consistent among different embryos over a range of developmental time points.

Figure 8 with 1 supplement see all
SDECs contribute to the dorsal aorta but do not generate HSPCs.

(A–C) Lineage tracing of SDECs using tbx6:Gal4; Tg(UAS-Cre); A2BD shows dsRed+ cells in the vasculature region at 48 hpf (arrowheads). (E–J) Using a vasculature-specific switch line TgBACkdrl:LOXP-AmCyan-LOXP-ZsYellow (referred to as kdrl:CSY), we observe the contribution of SDECs or LPM-derived endothelial cells to the vasculature. (E–G) For SDEC labeling, a PM-specific driver tbx6:Gal4; Tg(UAS-Cre) was used. PM-derived YFP+ SDECs are observed in the vasculature of imaged embryos. (H–J) For LPM-specific EC labeling, a Tg(drl:CreERT2) was used and treated with 10 µm tamoxifen starting at 8 hpf. YFP+ ECs are observed in all regions of the vasculature. (K) Quantification of YFP+ SDECs and ECs from tbx6 or drl switched embryos, respectively. Quantifications were based on independent experiments per transgenic background with n=23 for tbx6 switched embryos and n=9 for drl switch embryos. (L) Analysis of the adult kidney marrow of tbx6:Gal4; Tg(UAS-Cre); A2BD animals shows no contribution to hematopoietic cells from switched DsRed+ SDECs through flow cytometry analysis, whereas the FSC/SSC distribution of the unswitched BFP+ ECs corresponds to all blood lineages (quantifications based from independent experiments with a total of n=21 samples). SDECs, somite-derived endothelial cells.

To complement these results, we performed lineage tracing using an LPM-specific switch line, Tg(drl:CreERT2) (Henninger et al., 2017). Likewise, we crossed it to kdrl:CSY for EC-specific tracing. We incubated Tg(drl:CreERT2); kdrl:CSY embryos with tamoxifen (4-OHT) to induce Cre-based recombination between 8 and 24 hpf. At 4 dpf, confocal imaging showed complementary results to our PM-specific tbx6+ lineage tracing experiments. We observed that drl-derived YFP+ ECs contributed to all regions of the vasculature, as would be expected from an LPM source (Figure 8H–J and K). However, we observed that the contribution of these cells was more robust within the posterior region of the DA and inverse to the contribution of the PM-specific, tbx6-labeled SDECs, which contributed ECs mainly to the trunk region of the vasculature (Figure 8D–K). Together, these results further demonstrate that SDECs integrate into the DA in zebrafish and appear to contribute preferentially to the trunk portion of the dorsal aorta.

Since HSPCs derive from hemogenic endothelium, specifically within the DA, we examined whether or not SDECs can generate HSPCs. We lineage traced PM-derived cells into the kidney, the adult hematopoietic organ in teleosts. We dissected kidneys of adult tbx6:Gal4; Tg(UAS-Cre); A2BD transgenic animals and observed no contribution of switched DsRed+ cells to any hematopoietic lineage (Figure 8L). A ubiquitous vascular-specific transgenic Cre driver was utilized as a positive control for hemogenic endothelium, Tg(kdrl:Cre)s898 (Bertrand et al., 2010; Figure 8—figure supplement 1). Taken together, these results demonstrate that SDECs integrate into the dorsal aorta but do not generate HSPCs.

SDECs support the emergence of HSPCs

Since SDECs did not directly contribute to HSPCs, we wanted to explore their role by analyzing the differential expression of SDECs versus lateral plate mesoderm-derived ECs (LPMDEC) that populated the DA. We compared the SDEC to HE and Pre-HSC clusters at 22–24 hpf. From SDECs, we identified many genes previously attributed to ‘niche’ functions required to induce aortic hemogenic endothelium (Figure 9A; Charbord et al., 2014). These results suggested that SDECs might be acting in a paracrine manner to support hematopoietic induction. If so, we reasoned that changes in the number of SDECs might result in the altered formation of hematopoietic cells. Increasing the numbers of SDECs via enforced expression of etv2 with a muscle-specific mylz2 promoter (Ju et al., 2003) or knockdown of meox1 led to an increase in runx1+ by WISH (Figure 9B–E). In contrast, decreasing the number of SDECs via over-expression of meox1 mRNA led to the loss of runx1+ and gata2b+hematopoietic cells in the DA (Figure 9F–I). These results suggest that SDECs act in a paracrine manner to support the induction of hematopoietic cells from neighboring hemogenic endothelial cells.

SDECs act as a vascular niche for hemogenic endothelium.

(A) Heatmap of genes differentially regulated between LPM-derived endothelial cells (LPMDEC sample is composed of pre-HSCs and HE clusters) and SDECs. (B,C) Zebrafish embryos were injected with an expression vector containing a somite-specific promoter, mylz2, driving an etv2 transgene to ectopically induced SDECs. By WISH, an increase in runx1 expression was observed at 24 hpf compared to embryos injected with an empty mylz2 vector. (D,E) Similarly, meox1 morphants exhibit an increased expression of runx1 by WISH compared to uninjected control embryos. (F–I) Conversely, overexpression of meox1 by mRNA injection strongly reduces the hemogenic markers runx1 and gata2b. SDECs, somite-derived endothelial cells.

Discussion

Here, we present an extensive analysis of how a bipotent skeletal muscle progenitor population gives rise to a rare contingent of ECs that are functionally distinct from endothelial cells specified in the LPM (Garcia-Martinez and Schoenwolf, 1992; Psychoyos and Stern, 1996; Schoenwolf et al., 1992; Selleck and Stern, 1991). Through single-cell analysis, we identify unique molecular signatures sufficient to identify subsets of ECs, including somite-derived endothelial cells (SDECs), brain endothelial cells (BVECs), kidney endothelial cells (KVECs), and endocardium as early as the tailbud stage. Together, these results indicate that cell fate restriction of EC subsets occurs earlier than previously thought. Our results show that SDECs emerge from the trunk dermomyotome during somitogenesis and contribute to the dorsal aorta of the zebrafish embryo. We show that modulators of muscle differentiation, such as meox1, notch, and wnt, regulate SDEC number and the timeframe of SDEC differentiation. Furthermore, our findings elucidate how SDECs are needed to support the hemogenic program required for the formation of HSPCs.

Avian and murine studies have previously identified a population of endothelial cells that arise from the somites (Ambler et al., 2001; Esner et al., 2006; Kardon et al., 2002; Mayeuf-Louchart et al., 2016; Mayeuf-Louchart et al., 2014; Pardanaud et al., 1996; Pouget et al., 2006; Wilting et al., 1995; Yvernogeau et al., 2012). Previously, the common belief was that in zebrafish, all endothelial and blood cells arise exclusively from the LPM (Childs et al., 2002; Jin et al., 2005; Kohli et al., 2013; Lawson and Weinstein, 2002; Zhang and Rodaway, 2007). Furthermore, angioblasts are thought to be specified as an equipotent population that undergoes progressive cell-fate restriction only upon migration to the midline (Kobayashi et al., 2014; Lawson et al., 2001). To our knowledge, only two studies have shown that paraxial mesoderm can generate endothelial cells in zebrafish (Martin and Kimelman, 2012; Nguyen et al., 2014). The first noted that presomitic mesoderm could generate ECs that incorporate into caudal blood vessels following early inhibition of the Wnt signaling pathway after gastrulation (Martin and Kimelman, 2012). The second study highlighted a rare population of ECs generated in the somite that migrated to the developing dorsal aorta (Nguyen et al., 2014). Each of these studies was based on retrospective analyses, making it difficult to ascertain precisely when and where SDECs arise and the extent of their contribution. In addition, studies by Murayama et al., 2023; Murayama et al., 2015 have shown that a population of stromal cells important in the maintenance of zebrafish HSPCs derive from the sclerotome, the ventrolateral domain of the somite. These do not appear to give rise to ECs but rather to stromal cells that reside adjacent to blood vessels. As we have occasionally observed rare cells outside of the vasculature in our lineage-tracing studies, some of our marked cells may include this stromal cell subset. Further studies are required to ascertain how this population of somite-derived cells compares to SDECs in the trophic support of hematopoietic precursors.

Labeling individual somite pairs using Tg(actb2:nls-Eos), a pan-nuclear photoconvertible transgenic line, allowed us to quantify the contribution of SDECs from trunk somites. Using Tg(etv2.1:eGFP)zf372, with etv2 as one of the earliest markers of endothelial cell fate acquisition, allowed us to visualize precisely when and where SDECs emerge. Lastly, we could permanently and differentially mark ECs from the LPM or PM using an endothelial-specific switch transgene and follow their respective contributions to the vasculature. Together, both temporal and indelible lineage tracing approaches showed that somites 5–18 have potential to generate SDECs. Emergence of SDECs initiated at the 12-somite stage, concomitant with the epithelization of the somite and the migration of LPM-derived ECs to the midline. Notably, contribution of SDECs to adult hematopoietic tissues was not observed, consistent with many previous studies concluding that the adult hematopoietic program is LPM-derived (Henninger et al., 2017; Jin et al., 2007; Murayama et al., 2006). These results are consistent with a previous study by Nguyen et al., 2014 that identified a rare population of EC precursors born in a central location within somite, which they termed the ‘endotome’, that incorporated into the axial vessels but did not appear to generate blood (Nguyen et al., 2014). By contrast, in our study, we observed that somite-derived etv2+ ECs arise from the hypaxial dermomyotome, which is consistent with earlier observations in avian and mouse embryos (Eichmann et al., 1993; Ema et al., 2006; Pouget et al., 2006; Tozer et al., 2007). Notably, the hypaxial dermomyotome comprises bipotent-skeletal muscle progenitors that generate ECs or myoblasts upon local cues. Since these bipotent progenitors are present in the territory where the hypaxial muscle precursors originate (Christ and Ordahl, 1995), we believe a more appropriate term for this somitic compartment would be the ‘myo-endotome’.

To better understand the signaling pathways regulating SDEC emergence, we tested the roles of notch, wnt, and npas4l. We find that Notch signaling is essential for the maintenance of muscle progenitors, as previously shown in mice (Schuster-Gossler et al., 2007). Loss of the Notch signaling components mib or notch3 resulted in a premature depletion of skeletal muscle progenitor markers and a concomitant increase of differentiated muscle and endothelial cell gene expression. Our results thus suggest that Notch signaling is required to maintain muscle progenitors rather than determining cell fate choice between muscle and endothelium. However, further work is needed to investigate whether the reduced expression of skeletal muscle progenitor markers results from progenitor cell depletion or simply reflects lower expression levels within these cells. In addition, Wnt signaling is necessary for the regionalization of muscle and endothelial cells within the somite, in agreement with previous work (Borello et al., 2006; Martin and Kimelman, 2012). We find that during homeostasis, a few etv2:eGFP+ delaminate from the hypaxial dermomyotome. However, combined loss of Notch signaling and meox1 shows that EC competence within the somite encompasses a much larger territory, extending as far as the median dermomyotome. Interestingly, we find that endothelial potential within bipotent skeletal muscle progenitors is npas4l-dependent. These results are surprising since npas4l has been shown to be the earliest gene required for EC specification and was thought to be restricted to the LPM starting at the tailbud stage (Reischauer et al., 2016). Equally surprising, a recent paper demonstrated that ECs populating the zebrafish caudal hematopoietic tissue are endoderm-derived (Nakajima et al., 2023). Similar to SDECs, this novel population is likewise dependent upon npas4l function (Nakajima et al., 2023). Together, these results suggest that npas4l is required to generate endothelial cell fate regardless of tissue origins.

Recent studies have elucidated a role for non-hemogenic ECs in supporting the hemogenic program and in amplifying HSPC number in various systems (Butler et al., 2012; Butler et al., 2010; Butler and Rafii, 2012; Gori et al., 2017; Gori et al., 2015; Guo et al., 2017; Hadland et al., 2015; Kim et al., 2014; Kobayashi et al., 2010; Lis et al., 2017; Raynaud et al., 2013; Sandler et al., 2014). Interestingly, several groups have successfully differentiated endothelial precursors into transplantable HSPCs, a process that requires an endothelial ‘niche’ population (Lis et al., 2017; Sandler et al., 2014). Our work indicates that changes to the number of SDECs via downregulation or overexpression of meox1 resulted in increased or decreased hematopoietic factors, such as runx1 and gata2b, respectively. This finding and evidence that SDECs lack hemogenic potential suggest that SDECs are important as niche support cells to induce HSPC formation. Future work will elucidate the molecular nature of this interaction and will forward efforts to instruct HSPC fate from human pluripotent precursors.

Materials and methods

Zebrafish husbandry

Request a detailed protocol

Wild-type AB* and transgenic TgBAC(etv2:Kaede)ci6 (Kohli et al., 2013) referred to as etv2:Kaede, Tg(fli1:DsRed)um13 (Villefranc et al., 2007) referred to as fli1:DsRed, Tg(tp1:GFP)um14 (Parsons et al., 2009) referred to as tp1:GFP, Tg(drl:H2B-dendra) (Mosimann et al., 2015) referred to as drl:H2B-dendra, Tg(actb2:nls-Eos) (Cruz et al., 2015) referred to as actb2:nls-Eos, Tg(fli1:eGFP)y1 (Isogai et al., 2003) referred to as fli1:eGFP, Tg(etv2.1:eGFP)zf372 (Veldman and Lin, 2012) referred to as etv2:eGFP, Tg(phlbd1:Gal4-mCherry) (Distel et al., 2009) referred to as phlbd1:mCherry, notch3fh332/fh332 (Alunni et al., 2013) referred to as notch3-/-, Clos9/s9 (Reischauer et al., 2016) referred to as npas4l-/- (cloche), Tg(7xTCF-Xla.Siam:GFP)ia4 (Moro et al., 2012) referred to as TCF:d2GFP, Tg(tbx6:Gal4FF:GFP-nls) (Yabe et al., 2016) referred to as tbx6:Gal4; Tg(UAS-Cre)* (Butko et al., 2015) referred to as UAS-Cre, Tg(actb2:loxP-BFP-loxP-DsRed)sd27 (Kobayashi et al., 2014) referred to as A2BD, TgBAC(kdrl:LOXP-AmCyan-LOXP-ZsYellow) (Zhou et al., 2011) referred to as kdrl:CSY, Tg(drl:CreERT2) (Henninger et al., 2017) referred to as drl:CreERT2, and Tg(kdrl:Cre)s898 (Bertrand et al., 2010) referred to as kdrl:Cre. Zebrafish embryos and adult fish were raised in a circulating aquarium system (Aquaneering) at 28  °C and maintained in accordance with UCSD Institutional Animal Care and Use Committee (IACUC) guidelines (Protocol number: S04168), the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) (Accreditation number: 000503), and the Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals (Assurance number: A3033-1 or D16-00020). All experiments were performed under approved methods of anesthesia, and every effort was made to minimize suffering.

FACS

Request a detailed protocol

TgBAC(etv2:Kaede)ci6 embryos were collected at 15 ss and 22 hpf. Tg(drl:H2B-dendra) embryos were collected at tailbud stage, 12 ss, and 22 hpf. Tg(fli1:DsRed)um13; Tg(tp1:GFP)um14 embryos were collected at 24 hpf. Samples were separately dissociated in phosphate-buffered saline (PBS) supplemented with 10% fetal bovine serum (FBS) by homogenizing them with a sterile plastic pestle or pipette. Dissociated cells were filtered through a 35 μm nylon cell strainer (Falcon 2340) and then rinsed with PBS with 10% FBS. Propidium iodide (Sigma) was added (1  μg ml−1) to exclude dead cells and debris. FACS was performed using GFP, Dendra, and DsRed fluorescence gating with a FACS Aria II flow cytometer (Beckton Dickinson).

Single-cell RNA sample preparation

Request a detailed protocol

After FACS, total cell concentration and viability were ascertained using a TC20 Automated Cell Counter (Bio-Rad). Samples were resuspended in 1XPBS with 10% bovine serum albumin (BSA) at a concentration between 8E2-3E3 cells per ml. Samples were loaded on the 10 X Chromium system and processed as per manufacturer’s instructions (10 X Genomics). Single-cell libraries were prepared per the manufacturer’s instructions using the Single Cell 3’ Reagent Kit v2 (10 X Genomics). Single-cell RNA-seq libraries and barcode amplicons were sequenced on an Illumina HiSeq2500 platform.

Single-cell RNA sequencing analysis

Request a detailed protocol

The Chromium 3’ sequencing libraries were generated using Chromium Single Cell 3’ Chip kit v3 and sequenced with an Illumina HiSeq2500 platform. The Ilumina FASTQ files were used to generate filtered matrices using CellRanger 3.0.0 (10X Genomics) with default parameters and imported into R for exploration and statistical analysis using a Seurat package V4 (La Manno et al., 2018). Counts were normalized according to total expression, multiplied by a scale factor (10,000), and log-transformed. For cell cluster identification and visualization, gene expression values were also scaled according to highly variable genes after controlling for unwanted variation generated by sample identity. Cell clusters were identified based on UMAP of the first 20 principal components of PCA using Seurat’s method, Find Clusters, with an original Louvain algorithm and resolution parameter value 0.5. To find cluster marker genes, Seurat’s method, FindAllMarkers was used. Only genes exhibiting a significant p-value of (<0.022), and a minimal average absolute log2-fold change of 1 between each cluster and the rest of the dataset, were considered differentially expressed. To merge individual 22–24 hpf datasets and remove batch effects, Seurat v3 Integration and Label Transfer standard workflow (Stuart et al., 2019) was used. To project samples of embryonic stage <22 hpf onto the reference 22–24 hpf dataset, Seurat’s MapQuery function was used.

UV-driven photoconversions

Request a detailed protocol

Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 embryos were collected and staged at developmental stages ranging from 4 to 18 ss in a glass bottom Petri dish that was precast using a 2% low melting agarose (Fisher Bioreagents, BP160-500) and a custom-made stamp (Idylle-lab) to created semi-spherical wells. Embryos were then staged and photoconverted with newly emerging somites directly at the bottom of the dish using a Leica SP8 confocal microscope. Recently developed somite pairs were selected by setting a region of interest, then photoconverted by applying maximum UV exposure (405 nm; 100% laser power) using a 20 x objective lens and high digital zoom (>X4) for 90 s. To validate the quality and specificity of the somite photoconversion, images were taken of each channel, eGFP (488 nm; 10% laser power) and mCherry (564 nm; 25% laser power). Photoconverted zebrafish embryos were then transferred to a petri dish with 1-phenyl 2-thiourea (PTU) media and incubated at 28.5  °C in the dark. At 32 hpf, embryos were embedded in a glass bottom petri dish covered by 1% agarose and imaged on a Leica SP8 confocal microscope using resonant scanning. Images of the dorsal aorta were taken in a sequential stack manner for eGFP (488 nm; 10% laser power) and mCherry (564 nm; 35% laser power) using a 20 x objective lens.

Quantification of SDECs emerging from somites

Request a detailed protocol

Maximum intensity projection (MIP) images were taken for each converted embryo to identify the converted somite pair. Then, SDECS were identified and quantified in the dorsal aorta by examining individual z-stacks and visualizing colocalization of fli1+; actb2:nlsEosRFP converted cells in Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 embryos. Particularly in fli1:eGFP- embryos, we identified SDECs by observing actb2:nlsEosRed cells on the floor of the DA based on the brightfield channel. Statistical analysis of the number of SDECs per somite was performed using Prism 9 (GraphPad).

Statistical analysis and graphic display

Request a detailed protocol

Graphic data representation and statistical analysis were performed using Prism 9 (Graphpad Version 9.5.1). In Figure 3, All graphs with single dot scatter plots indicate the mean as column height, and standard error of the mean (SEM) is represented as error bars. In Figures 57, the statistics of the qPCR were performed using unpaired, two-tailed t-tests (n=3), and standard error of the mean (SEM) is represented as error bars.

Microscopy and timelapse videos

Request a detailed protocol

Embryos were embedded and sectioned as described by Kobayashi et al., 2014. Live transgenic embryos and flat-mount or whole-mount two-color double FISH samples were imaged using confocal microscopy (SP5 or SP8, Leica). Since the emission spectra of CFP and YFP overlap, images were captured in two separate sequences to filter overlapping signals by limiting the PMT detection. For CFP a 476 nm laser was used, and the PMT detector was set to collect signals ranging between 480 nm and 505 nm. For YFP, a 514 nm laser was used, and the PMT detector was set to collect signals ranging between 520 nm and 570 nm.

Microinjections of morpholinos, mRNA, and plasmids

Request a detailed protocol

Antisense morpholinos (MOs; Gene Tools, LLC) were diluted as 1- or 3 mM stock in H2O. meox1-MO (CTGGCTGACTGTTCCATACTGAAGA) and mib-MO (GCAGCCTCACCTGTAGGCGCACTGT) were injected at the 1–2 cell stage of development. mOrange-CAAX mRNA was synthesized from linearized mOrange-CAAX with the mMessege mMaching kit (Ambion). 100  pg mOrange-CAAX mRNA was injected into one- to two-cell stage embryos. A total of 25 pg of mylz2:etv2 construct was coinjected with 50 pg of transposase mRNA. All microinjections were performed with the indicated RNA or MO concentration in a 1  nl using a PM 1000 cell microinjector (MDI).

WISH

Request a detailed protocol

Whole-mount single or double enzymatic in situ hybridization was performed on embryos fixed overnight with 4% paraformaldehyde (PFA) in PBS. Fixed embryos were washed briefly in PBS and transferred into methanol for storage at −20  °C. Embryos were rehydrated stepwise through methanol in PBS +0.1% Tween 20 (PBT). Rehydrated embryo samples were then incubated with 10  μg ml−1 proteinase K in PBT for 5  min for 5–10 somite stage (12–15 hpf) embryos and 15  min for 24–36 hpf embryos. After proteinase K treatment, samples were washed in PBT and refixed in 4% PFA for 20  min at room temperature. After washes in two changes of PBT, embryos were prehybridized at 65  °C for 1 hr in hybridization buffer (50% formamide, 5 x saline-sodium citrate (SSC), 500  μg ml−1 torula tRNA, 50  μg ml−1 heparin, 0.1% Tween 20, 9  mM citric acid (pH 6.5)). Samples were then hybridized overnight in hybridization buffer including digoxigenin (DIG)- or fluorescein-labeled RNA probe. After hybridization, experimental samples were washed stepwise at 65  °C for 15  min each in hybridization buffer in 2×SSC mix (75%, 50%, 25%), followed by two washes with 0.2×SSC for 30  min each at 65  °C. Further washes were performed at room temperature for 5  min each with 0.2×SSC in PBT (75%, 50%, 25%). Samples were incubated in PBT with 2% heat-inactivated goat serum and 2  mg ml−1 BSA (block solution) for 1 hr and then incubated overnight at 4  °C in block solution with diluted DIG-antibodies (1:5,000) conjugated with alkaline phosphatase (AP) (Roche). To visualize WISH signal, samples were washed three times in AP reaction buffer (100  mM Tris, pH 9.5, 50  mM MgCl2, 100  mM NaCl, and 0.1% Tween 20) for 5  min each and then incubated in the AP reaction buffer with NBT/BCIP substrate (Roche).

For two-color double FISH, embryos were blocked in maleic acid buffer (MAB; 150  mM maleic acid, 100  mM NaCl, pH 7.5) with 2% Roche blocking reagent (MABB) for 1  hr at room temperature, after hybridizing at 65  °C with probes as described above. Embryos were incubated overnight at 4  °C in MABB with anti-fluorescein POD. (Roche) at a 1:500 dilution. After four washes in MABB for 20  min, each followed by washes in PBS at room temperature, embryo samples were incubated in TSA Plus Fluorescein Solution (Perkin Elmer) for 1  hr. Embryos were washed for 10  min each in methanol in PBS (25%, 50%, 75%, 100%). Embryos were incubated in 1% H2O2 in methanol for 30  min at room temperature and washed for 10  min each in methanol in PBS (75%, 50%, 25%) and 10  min in PBS. After blocking for 1  hr in MABB, embryos were incubated overnight at 4  °C in MABB with anti-DIG POD. (Roche) at a 1:1000 dilution. As described above, samples were washed and incubated in TSA Plus CY3 solution (Perkin Elmer). Embryos were washed thrice for 10  min each in PBT and refixed in 4% PFA after the complete staining. Antisense RNA probes for the following genes were prepared using probes containing DIG or fluorescein-labeled UTP: etv2 (Clements et al., 2011), pax3a (Minchin et al., 2013), myod (Minchin et al., 2013), meox1 (Nguyen et al., 2014), runx1 (Burns et al., 2005), and gata2b (Butko et al., 2015).

Quantitative real-time RT-PCR

Request a detailed protocol

Total RNA was collected from whole embryos (~20 embryos) using TRIzol reagent (Ambion, Life Technologies) and isolated from notch3fh332/fh332 mutant embryos or control siblings at 48 hpf; cloche mutant embryos or control siblings at 48 hpf; or IWP2 treated and control embryos at 15 hpf with the RNeasy kit (Qiagen). cDNA was generated from total RNA with iScript cDNA synthesis kit (Bio-Rad). The following primers were used for cDNA quantification: ef1α (forward, 5′- GAGAAGTTCGAGAAGGAAGC –3′; reverse, 5′- CGTAGTATTTGCTGGTCTCG –3′), etv2 (forward, 5′- CGAGGTTCTGGTAGGTTTGAG –3′; reverse, 5′- GCACAAAGGTCATGTTCTCAC –3’), fli1a (forward, 5′- CGTCAAGCGAGAGTATGACC –3′; reverse, 5′- AGTTCATCTGAGACGCTTCG –3’), myod1 (forward, 5′- GAAGACGGAACAGCTATGAC –3′; reverse, 5′- GGAGTCTCTGTGGAAATTCG –3’), myf5 (forward, 5′- CCAGACAGTCCAAACAACAGACC –3′; reverse, 5′- TGAGCAAGCAGTGTGAGTAAGCG –3’), pax3a (forward, 5′- ATTCCTTGGAGGTCTCTACG –3′; reverse, 5′- CTACTATCTTGTGGCGGATG –3’), pax7b (forward, 5′- CAGTATTGACGGCATTCTGGGAG –3′; reverse, 5′- TCTCTGCTTTCTCTTGAGCGGC –3’), myf5 (forward, 5′- GAATAGCTACAACTTTGACG –3′; reverse, 5′- GTAAACTGGTCTGTTGTTTG –3’), myog (forward, 5′- GTGGACAGCATAACGGGAACAG –3′; reverse, 5′-TCTGAAGGTAACGGTGAGTCGG-3’).

Immunofluorescence

Request a detailed protocol

Whole-mount immunofluorescence staining for myoHII was performed on notch3fh332/fh332 embryos at 48 hpf using a MF-20 (DSHB) antibody. Immunofluorescence was performed as previously described in Alexander et al., 1998.

Pharmacological treatment

Request a detailed protocol

IWR (Tocris) was dissolved in DMSO at a concentration of 10  mM. Zebrafish AB* embryos were incubated in 10 μM IWR solution from 11 to 15 hpf, then fixed with 4% PFA or snap frozen for RNA collection.

Analysis of whole kidney marrow cells

Request a detailed protocol

Whole kidney marrow cells (WKM) were prepared as previously described in Kobayashi et al., 2008. Briefly, adult tbx6:Gal4; Tg(UAS-Cre); A2BD or kdrl:Cre; A2BD transgenic animals were anesthetized and sacrificed following IACUC regulations. The kidney of individual transgenic animals was dissected then minced with scissors. Hematopoietic cells were mechanically dissociated by trituration in PBS supplemented with 2% of FBS and filtered through 40 µm nylon mesh. Prior to flow cytometry analysis, samples were stained with a viability dye (5 nM of Sytox Red, ThermoFisher) to exclude dead cells from the analysis. Flow cytometry was performed on a FACS Aria II (BD Bioscience). Collected data were analyzed using FlowJo software.

Plasmid construction

Request a detailed protocol

Plasmid expression constructs were generated by standard means using PCR from cDNA libraries generated from zebrafish larvae at 24 hpf. These were cloned into pCS2 +downstream of an SP6 promoter. Previously described mylz2 promoter (Ju et al., 2003) was cloned upstream of etv2 CDS flanked by Tol2 sites.

Replicates

Request a detailed protocol

All experiments assessing phenotype and expression patterns were replicated in at least two independent experiments. Embryos were collected from independent crosses, and experimental processing (staining or injections) was carried out on independent occasions. Exceptions to this include the analysis of the kidney hematopoietic cells by flow from tbx6:Gal4; Tg(UAS-Cre); A2BD and the imaging of tbx6:Gal4; Tg(UAS-Cre); kdrl-CSY where many embryos were processed, and the corresponding n are reported in the associated figure legends. For the analysis of qRT-PCR, three independent experiments were performed per condition. One sequencing run for the single-cell sequencing experiments was performed per timepoint from a pool of at least 100 embryos within the corresponding transgenic background.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files. scRNA-seq data are available in ArrayExpress under accession number E-MTAB-13196.

The following data sets were generated
    1. Sahai-Hernandez P
    (2023) Dryad Digital Repository
    Single Cell Sequencing Data for Endothelial Cell Types.
    https://doi.org/10.6075/J0GB22J0
    1. Svoboda O
    (2023) ArrayExpress
    ID E-MTAB-13196. Single-cell RNA-seq (10x Chromium) of FACS-sorted drl:H2B-Dendra2, etv2:Kaede, fli1:DsRed, and tp1:eGFP transgenic zebrafish embryos.

References

    1. Risau W
    2. Flamme I
    (1995) Vasculogenesis
    Annual Review of Cell and Developmental Biology 11:73–91.
    https://doi.org/10.1146/annurev.cb.11.110195.000445

Decision letter

  1. Owen J Tamplin
    Reviewing Editor; University of Wisconsin-Madison, United States
  2. Didier YR Stainier
    Senior Editor; Max Planck Institute for Heart and Lung Research, Germany

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

Thank you for sending your article entitled "Dermomyotome-derived endothelial cells migrate to the dorsal aorta to support hematopoietic stem cell emergence" for peer review at eLife. Your article is being evaluated by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Didier Stainier as the Senior Editor.

Given the list of essential revisions, including new experiments, the editors and reviewers invite you to respond as soon as you can with an action plan for the completion of the additional work. We expect a revision plan that under normal circumstances can be accomplished within two months, although we understand that in reality revisions will take longer at the moment. We plan to share your responses with the reviewers and then advise further with a formal decision.

In particular, the reviewers had concerns regarding the novelty of some of the major points of the manuscript, such as the derivation and contribution of somite-derived endothelial cells to hematopoietic stem cell development, in light of recent high profile publications. However, they also mentioned several interesting aspects of the work that might be developed further to enhance the suitability for publication. These include but are not limited to further understanding of the role of Notch signaling in the development, migration and function of the cell type of interest, additional transcriptomics profiling at earlier timepoints, as well as the use of alternative lines for expanded lineage tracing studies to further document the functional potential and role of the somite-derived endothelial population. If after reading the comments and suggestions below, you feel that these points can be sufficiently addressed in a reasonable timeframe, we look forward to receipt of a revision plan detailing your plans for the manuscript.

Reviewer #1:

The manuscript entitled: Dermomyotome-derived endothelial cells migrate to the dorsal aorta to support hematopoietic stem cell emergence, by Sahai-Hernandez et al. aims to explore the contribution of somite-derived endothelial cells (SDECs) to hematopoietic competence in the embryonic dorsal aorta. This study builds on and significantly extends recent work identifying a paraxial mesoderm derived endothelial population under the regulation of Meox that contributes to hemogenic fate in the dorsal aorta. Here, the authors have examined the interplay of several pathways, including Wnt and Notch, in regard to their contribution to the generation of these somite-derived endothelial cells (SDECs) from specific skeletal muscle progenitors and their subsequent role in HSC formation. The authors find that this population of SDECs, marked by ETV2-reporter expression, is derived specifically from the dermomyotome, in balance with the production of muscle cell types. Consistent with early studies in the chick, they describe integration of the SDEC population into the dorsal aorta itself, and then provide novel insight into their preferential localization as well as an unexpected role in paracrine BMP4 mediated HSC regulation. Interestingly, the authors also show, using the reporter line, that the SDEC population is missing in cloche embryos. Through elegant scRNA-seq analysis and lineage trajectory studies, they identified several transcriptionally distinct endothelial populations in the early embryo, prior to HSC specification and vascular maturation, and show how modulation of Notch signaling, known to be essential for hemogenic fate, does or does not impact their development. Finally, the authors investigate the role of ID1 in Notch associated regulation of HSC development. There are several important novel insights provided by the study, however the flow of the report, and strong initial emphasis on the SDEC population, masks some of the other key observations uncovered regarding endothelial biology and hematopoietic fate in the early vertebrate embryo.

1. The flow of the manuscript is somewhat problematic. The authors perform a very elegant unbiased profiling study which identifies several distinct endothelial populations in the embryo, including the SDECs, at a much earlier timepoint than previously anticipated. Given recent publications, rather than starting with characterization of the SDEC population, and evaluating their production and function with additional precision, the authors might consider inverting the study to describe and then validate the relationships and regulation of the formation of each distinct subset, at least in regard to those impacting blood formation. In essence they end up doing this for SDECs, arterial and hemogenic endothelium already in the later figures, but altering the flow may help to highlight distinctions from prior work, as well as alleviate some of confusion/ redundancy regarding the multiple roles/timing of Notch signaling, and differing mechanistic influences hemogenic fate.

2. With the current layout of the manuscript, the very intriguing studies at the end of the paper regarding ID1 seem disconnected from the main focus on the SDECs. If the authors tell a more generalized story about establishing several distinct populations of endothelial cells that influence HSC formation (it would be fine to almost ignore the few that don't impact HSCs if that is the intended focus), then the ID1 data being associated with LPM-derived endothelium, but not PM-derived cells (in contrast to BMP signaling) is more logical.

3. The authors use an ETV2-reporter to follow the onset of SDEC production as well as their later integration in the aorta. However, they also indicate that ETV2 is expressed before the somite-derived population is present, and the RNAseq analysis seems to indicate its expression is found in many endothelial populations, such that it might not be specific to SDECs. Given those caveats, it would be helpful to find another marker (from the RNAseq?) that could be used to verify some of the data presented. As an earlier study also showed localization of a similar population of somite-associated cells in the dorsal aorta, this conclusion not in doubt, but the reporter could be inclusive of more than one endothelial cell type, impacting specificity of the BMP analysis and/or loss of function/epistasis studies; use of a second marker may help solidify the novel insights.

4. As the dermomyotome is transient, it is interesting to understand how the biology overlaps with HSPC development. It appears the authors presume that the cells of different origin (LPM vs PM) have experienced differential signaling that impacts their later function. The elegant RNAseq data and trajectory analyses imply they are indeed transcriptionally distinct at a later time point. Is it possible to find and compare the hematopoiesis-associated endothelial subsets earlier in development to learn more about the onset of their development and differences, or validate additional distinguishing marks by qPCR or LOF studies prior to integration in the aorta? Given prior reports suggesting a role for CXCL12 in hematopoietic regulation from somite-derived endothelial cells, it would be interesting to know when this signature appears and in which cohort(s).

Reviewer #2:

The article by Sahai-Hernandez et al. describes endothelial cells in the somite of developing zebrafish, their delamination to colonize the developing dorsal aorta and a trophic role in aortic hematopoiesis. The paper shows that these cells are derived from a muscle-endothelial precursor present in the lateral lip of the somite starting from the 12-somite stage. Differentiation of these cells are dependent on the Notch or the Wnt pathways Single cell transcriptome approaches have revealed the molecular identity of somitic endothelial cells, in particular the expression of genes related to the hematopoietic support. These single cell studies have been extended to a context of Notch mutant and show that the Notch pathway plays a moderate role on somitic endothelial cells.

This is an interesting article that, unfortunately comes late regarding the paper by Nguyen et al., Nature 2014. A number of data presented here have already been published in the Nature paper in a more convincing way. In particular the dual expression of muscle- and endothelial-specific genes, the muscle/endothelium balance and the trophic role of somitic endothelial cells on aortic hematopoiesis among the most important ones.

In its current form, the article is not acceptable for eLife because many of the data and conclusions presented here have already been published in the Nature paper. However, this article brings some novelties that are the timing of appearance and migration of the somite endothelial cells, the single cell characterization and the extensive study of the Notch pathway.

If acceptable by the Editor and feasible by the authors, this reviewer would recommend to significantly reinforce

1. the descriptive aspect of the emergence, migration and integration of cells in the dorsal aorta by insisting on the beginning and the end of the phenomenon, the number of cells per somite, the dynamic aspect of development, Draculin and Tbx6 approaches. In other words, make a kind of atlas for these cells with 3D visualizations if possible.

2. to focus on the Notch pathway. The role of Wnt and Meox have already been published in the Nature paper.

3. to further exploit the single cell transcriptome to show the existence of bipotent precursors and how these precursors enter the endothelial or muscular pathway depending on Notch conditions.

Reviewer #3:

This manuscript explores the origins of endothelial cells that make up the dorsal aorta (DA) in the developing zebrafish. The DA is the site of the hemogenic endothelium that produces definitive hematopoietic stem cells (HSCs). The prevailing model is that a subset of lateral plate mesoderm is fated for the dorsal aorta. These cells are further segregated into cells that make up the aortic endothelium, and those with hemogenic potential. In this study the authors identify an additional source of DA endothelium derived from the somites, which are conventionally thought of as defined blocks of skeletal muscle. These cells in the dermamyotome are positive for etv2:gfp+ cells and are proposed as bipotential precursors that can give rise to endothelium or muscle. The authors conclude that these cells are supportive of HSC emergence, but do not themselves produce HSCs. This is a novel concept in the ontogeny of hemogenic endothelium.

1. In Figure 1, it would be clearer if the etv2:gfp line was crossed to a red endothelial-specific line (e.g. flk:mCherry). Without additional landmarks in S1 it is unclear if these cells are continuously migrating towards the midline and becoming incorporated in the DA, or if they are moving out of the focal plane but are still part of the somite. Additionally, the movies S1 and S2 need to have features of interest labeled including boundaries between different tissue types.

2. Following the point above, the migration between etv2:gfp+ cells in the dermamyotome and the DA needs to be better defined. There should be more discrete lineage tracing, such as a photoactivatable marker, or time-lapse live imaging to show incorporation of these cells in the DA. The existing time-lapse shows delamination of etv2:gfp+ cells from the dermamyotome, but not migration and incorporation into the DA.

3. A key conclusion of the study is that SDEC cells are integrated into the DA are supportive of HSCs. If that is the case, the authors could perform time-lapse of tbx6:Cre;kdrl:CSY or some other reporter line combination to observe where budding HSCs are emerging relative to SDECs. This would capture the spatial relationships between the proposed supportive SDECs and HSCs.

4. Does meox1 knockdown result in enhanced proliferation of etv2+ cells in the dermamyotome, or do more cells emerge de novo? Single cell lineage tracing or tracking of single cells in time-lapse movies could help distinguish between these possibilities.

5. In Figures 2O and 3O the change in expression levels from bulk mRNA of pooled whole embryos does not necessarily mean that there is a change in the numbers of specific cell types. This result could also occur if there are similar cell types present that have altered expression profiles. These results could be clarified by FACS analysis of different populations to detect a change in ratios.

6. The authors do not go back further in ontogeny to explore possible heterogeneity within the dermamyotome. For example, how early do the etv2:gfp cells arise from the PSM and can be distinguished as a bipotential dermamyotome precursor?

7. What is the significance or conclusions drawn from the regional anterior-trunk-posterior observations presented in Figure 5?

8. The rationale for choosing lines for scRNA-seq (Etv2:kaede, Fli1:dsRed; Tp1:GFP, and Draculin:Dendra:H2B) is not clear because these lines are not specific to ECs and are known collectively to mark blood, endocardium, and other cell types. How do the authors distinguish between subtypes of ECs versus other cell types altogether?

9. The authors could show in Figure 6 where etv2:kaede and the other sorted reporter lines map onto these clusters. The authors could also provide scRNA-seq data for the same etv2:gfp+ cells that were presented in previous figures to confirm the presence of SDECs.

10. A recent paper comparing a new etv2 knock-in line with existing transgenic lines shows differences in expression patterns (Chestnut & Sumanas Dev Dyn 2020). Although data is presented in this study that shows endogenous expression of etv2 using FISH, it is possible that expression and conclusions drawn from the etv2:gfp line does not precisely follow endogenous cell types. This caveat should be considered and discussed within the larger context of the study.

11. It is unclear from the FISH images if the etv2+ cells are in fact double positive with meox1, or if etv2+ cells are intercalated between meox1 cells. The green meox1 signal appears low within the boundaries of etv2+ cells. Could this be resolved by crossing the etv2:gfp line with a red muscle progenitor reporter line, if available?

12. The authors do not discuss the paper by Murayama et al. (Nat. Comm. 2015) that shows stromal cells in the caudal hematopoietic tissue (CHT) directly delaminate from the somites. There are important similarities between this manuscript and the previous study that are not explored or referenced. For example, in Figure 5G, what are the round cells in the CHT that have switched? Could these be the stromal cells observed by Murayama et al? The switched cells in Figure 5J have endothelial morphology, but not the cells in Figure 5G.

13. The authors say that lineage tracing of tbx6:cre cells in the DA are consistent between multiple embryos over different stages, however this should be better quantified with more of the data represented in the analysis. It would be informative to show data before the 90 hpf timepoint in Figure 5E-J.

14. Are the Cre reporter lines single insertions? It appears from the FACs plots in S3 there are a range of signal intensities, and that some cells may express both colors, suggesting incomplete recombination of multiple insertions. This is possible too in Figure 5G that shows cell that appear double positive. Are there additional controls that could be presented?

15. The draculin-cre result is not surprising as the line markers endothelial and blood precursors that, as expected, are labeled in this experiment. These results are independent of SDEC ontogeny. This should be clarified.

16. In Figure 7A, bmp4 only appears upregulated in a small number of SDECs, and yet the conclusion is made that this is an important factor produced by SDECs.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Dermomyotome-derived endothelial cells migrate to the dorsal aorta to support hematopoietic stem cell emergence" for further consideration by eLife. Your revised article has been evaluated by Didier Stainier (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The reviewers are very enthusiastic about the revised manuscript. The only outstanding issue is the presentation of new data that does not have quantification or statistical analysis.

Reviewer #3 (Recommendations for the authors):

In this study, the authors identify an additional source of dorsal aorta endothelium derived from the somites, which are conventionally thought of as defined blocks of skeletal muscle. These cells in the dermamyotome are positive for endothelial cell markers and are proposed as bipotential precursors that can give rise to endothelium or muscle. The authors conclude that these cells are supportive of HSC emergence, but do not themselves produce HSCs.

Most of the previous reviewers' comments have been addressed. However, there are new imaging data presented that has not been analyzed quantitatively or rigorously tested for significant changes using statistical tests.

1. Quantification of results in Figure 4F,G, 5A-F, 5I-N, 7C-D.

2. Is decrease in meox1 significant in Figure 7B? If not, this should be clarified.

3. Need more detail about microscopy that allowed resolution of CFP and YFP from kdrl:CSY line in Figure 8E-J.

4. The drl:Cre-ERT2 data in Figure 8H-K is not clearly convincing and could be removed. The tbx6:Cre data makes the point well.

https://doi.org/10.7554/eLife.58300.sa1

Author response

Essential revisions:

Reviewer #1:

The manuscript entitled: Dermomyotome-derived endothelial cells migrate to the dorsal aorta to support hematopoietic stem cell emergence, by Sahai-Hernandez et al. aims to explore the contribution of somite-derived endothelial cells (SDECs) to hematopoietic competence in the embryonic dorsal aorta. This study builds on and significantly extends recent work identifying a paraxial mesoderm derived endothelial population under the regulation of Meox that contributes to hemogenic fate in the dorsal aorta. Here, the authors have examined the interplay of several pathways, including Wnt and Notch, in regard to their contribution to the generation of these somite-derived endothelial cells (SDECs) from specific skeletal muscle progenitors and their subsequent role in HSC formation. The authors find that this population of SDECs, marked by ETV2-reporter expression, is derived specifically from the dermomyotome, in balance with the production of muscle cell types. Consistent with early studies in the chick, they describe integration of the SDEC population into the dorsal aorta itself, and then provide novel insight into their preferential localization as well as an unexpected role in paracrine BMP4 mediated HSC regulation. Interestingly, the authors also show, using the reporter line, that the SDEC population is missing in cloche embryos. Through elegant scRNA-seq analysis and lineage trajectory studies, they identified several transcriptionally distinct endothelial populations in the early embryo, prior to HSC specification and vascular maturation, and show how modulation of Notch signaling, known to be essential for hemogenic fate, does or does not impact their development. Finally, the authors investigate the role of ID1 in Notch associated regulation of HSC development. There are several important novel insights provided by the study, however the flow of the report, and strong initial emphasis on the SDEC population, masks some of the other key observations uncovered regarding endothelial biology and hematopoietic fate in the early vertebrate embryo.

1. The flow of the manuscript is somewhat problematic. The authors perform a very elegant unbiased profiling study which identifies several distinct endothelial populations in the embryo, including the SDECs, at a much earlier timepoint than previously anticipated. Given recent publications, rather than starting with characterization of the SDEC population, and evaluating their production and function with additional precision, the authors might consider inverting the study to describe and then validate the relationships and regulation of the formation of each distinct subset, at least in regard to those impacting blood formation. In essence they end up doing this for SDECs, arterial and hemogenic endothelium already in the later figures, but altering the flow may help to highlight distinctions from prior work, as well as alleviate some of confusion/ redundancy regarding the multiple roles/timing of Notch signaling, and differing mechanistic influences hemogenic fate.

We thank our reviewer for this excellent suggestion and agree that restructuring the manuscript has provided an emphasis on novelty, a more logical progression, and improved clarity. As suggested, we have now structured the paper to lead with the divergence of the endothelial lineages and the early onset of their specification, as highlighted by the single-cell work. We then focus upon the uniqueness of the somite-derived endothelial cell (SDEC) population and characterize their cellular and molecular nature. We now quantify SDEC formation per somite pair, better characterize their cellular emergence from the somites, characterize the molecular regulation of their emergence, and focus upon their role as niche support cells within the dorsal aorta to regulate the emergence of HSPCs.

2. With the current layout of the manuscript, the very intriguing studies at the end of the paper regarding ID1 seem disconnected from the main focus on the SDECs. If the authors tell a more generalized story about establishing several distinct populations of endothelial cells that influence HSC formation (it would be fine to almost ignore the few that don't impact HSCs if that is the intended focus), then the ID1 data being associated with LPM-derived endothelium, but not PM-derived cells (in contrast to BMP signaling) is more logical.

We agree that in the original layout of the manuscript, the roles of bmp4 and id1 seemed out of focus. In the revised manuscript, we have better addressed the specification and contribution of the different EC subsets, followed by a more in-depth characterization of the SDEC population. We have now omitted the section regarding the roles of id1 and bmp4 and will address them in another manuscript.

3. The authors use an ETV2-reporter to follow the onset of SDEC production as well as their later integration in the aorta. However, they also indicate that ETV2 is expressed before the somite-derived population is present, and the RNAseq analysis seems to indicate its expression is found in many endothelial populations, such that it might not be specific to SDECs. Given those caveats, it would be helpful to find another marker (from the RNAseq?) that could be used to verify some of the data presented. As an earlier study also showed localization of a similar population of somite-associated cells in the dorsal aorta, this conclusion not in doubt, but the reporter could be inclusive of more than one endothelial cell type, impacting specificity of the BMP analysis and/or loss of function/epistasis studies; use of a second marker may help solidify the novel insights.

etv2 expression is indeed a marker of commitment to the endothelial lineages and is believed to be expressed by all subsets of ECs. It thus is certainly not a specific marker of SDECs. To specifically mark SDECs, we used a PM-specific lineage tracer, tbx6:Cre, and two “switch” lines. In the first combination, tbx6:Gal4; Tg(UAS-Cre); A2BD animals show switching of PM-derived lineages from BFP+ to dsRed+ cells, allowing SDECs to be followed upon leaving the somite and integrating into the DA. In the second combination, tbx6:Gal4; Tg(UAS-Cre); kdrl:CSY animals will switch out a CFP cassette to YFP in PM-derived cells. Because this switched transgene will only be expressed in endothelial cells due to the kdrl driver, only SDECs will be marked with YFP, whereas all other ECs will remain CFP+ due to derivation from non-PM-derived precursors. In both lineage tracing lines, we show PM-derived ECs lodged within the floor of the DA (Figure 8). In addition, we have added a meticulously performed analysis of the migration and localization of SDECs to the DA by using a photoconvertible Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 transgenic line. By applying precise photoconversion of single somite pairs and following their trajectories into the DA, we further support SDEC contribution to the DA, provide a detailed map of the somites involved (trunk somites), and quantify the number of SDECs contributed by each somite pair (Figure 3). We believe that these additional observations provide further evidence of SDEC emergence, migration, and contribution to the DA. We have not yet identified a single gene product that can be used to specifically identify SDECs.

4. As the dermomyotome is transient, it is interesting to understand how the biology overlaps with HSPC development. It appears the authors presume that the cells of different origin (LPM vs PM) have experienced differential signaling that impacts their later function. The elegant RNAseq data and trajectory analyses imply they are indeed transcriptionally distinct at a later time point. Is it possible to find and compare the hematopoiesis-associated endothelial subsets earlier in development to learn more about the onset of their development and differences, or validate additional distinguishing marks by qPCR or LOF studies prior to integration in the aorta? Given prior reports suggesting a role for CXCL12 in hematopoietic regulation from somite-derived endothelial cells, it would be interesting to know when this signature appears and in which cohort(s).

This is an excellent point. We have now obtained scRNA sequencing data sets from ECs purified at earlier development timepoints. We have added differential expression analyses comparing selected EC subsets between tailbud, 12 ss, 15 ss, and 22 hpf stages. Surprisingly, the tissue-specific signatures we detected at 22 hpf can be traced backward in developmental time. We identified EC clusters with distinct transcriptomes as early as the tailbud stage, suggesting that EC specification starts by the end of gastrulation and the initiation of somitogenesis. While the EC clusters were still heterogeneous prior to the 12 ss stage, we could identify most EC subsets, including HE, pre-HSC, KVECs, BVECs, and SDECs, after the 15 ss when their transcriptomes became more defined. Together, these data suggest a gradual EC specification process that begins as early as the tailbud stage. These findings are highlighted in our revised Figures 1, Figure 1 —figure supplement 1-3, Figure 2, and Figure 2 —figure supplement 1.

Specifically, for the cxcl12 signaling axis, we have found expression of its components only within early brain ECs, which was downregulated by 22 hpf (Figure 2 —figure supplement 1). The sequencing depth within our scRNA-Seq libraries is limited, which may have precluded us from finding cxcl12 components within the SDEC subset.

Reviewer #2:

The article by Sahai-Hernandez et al. describes endothelial cells in the somite of developing zebrafish, their delamination to colonize the developing dorsal aorta and a trophic role in aortic hematopoiesis. The paper shows that these cells are derived from a muscle-endothelial precursor present in the lateral lip of the somite starting from the 12-somite stage. Differentiation of these cells are dependent on the Notch or the Wnt pathways Single cell transcriptome approaches have revealed the molecular identity of somitic endothelial cells, in particular the expression of genes related to the hematopoietic support. These single cell studies have been extended to a context of Notch mutant and show that the Notch pathway plays a moderate role on somitic endothelial cells.

This is an interesting article that, unfortunately comes late regarding the paper by Nguyen et al., Nature 2014. A number of data presented here have already been published in the Nature paper in a more convincing way. In particular the dual expression of muscle- and endothelial-specific genes, the muscle/endothelium balance and the trophic role of somitic endothelial cells on aortic hematopoiesis among the most important ones.

In its current form, the article is not acceptable for eLife because many of the data and conclusions presented here have already been published in the Nature paper. However, this article brings some novelties that are the timing of appearance and migration of the somite endothelial cells, the single cell characterization and the extensive study of the Notch pathway.

If acceptable by the Editor and feasible by the authors, this reviewer would recommend to significantly reinforce

1. the descriptive aspect of the emergence, migration and integration of cells in the dorsal aorta by insisting on the beginning and the end of the phenomenon, the number of cells per somite, the dynamic aspect of development, Draculin and Tbx6 approaches. In other words, make a kind of atlas for these cells with 3D visualizations if possible.

2. To focus on the Notch pathway. The role of Wnt and Meox have already been published in the Nature paper.

3. To further exploit the single cell transcriptome to show the existence of bipotent precursors and how these precursors enter the endothelial or muscular pathway depending on Notch conditions.

We acknowledge that our study revisits some previous findings from Pete Currie’s group and many seminal studies of SDECs in the chick embryo. We want to point out that our work brings a variety of new insights into the formation and function of SDECs, as outlined above in our response to Reviewer 1. In our revised manuscript, we further these novel findings by improved imaging of SDEC formation and migration from the somite to the aorta by utilizing a photoconvertible Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 transgenic line (Figure 3). These animals were used to produce a detailed atlas of which somites contribute SDECs to the trunk DA and quantify the range of SDEC contribution per somite pair. Crossing the photoconvertible Tg(actb2:nls-Eos) line with a Tg(fli1:eGFP)y1 transgenic line demonstrates how SDECs migrate into the DA. We also utilized several fluorescent transgene combinations, including tbx6:Cre; kdrl:CSY and tbx6:Cre; A2BD, which demarcates SDECs from LPM-derived ECs (Figure 8).

We extended our studies and discussion on the role of notch in SDEC formation and function, as suggested. We show that Notch signaling is dispensable for SDEC fate and is rather required to enforce the muscle cell program from their bipotent precursors. To our knowledge, our results show for the first time the role of Wnt signaling in the muscle vs. endothelial cell fate decision in the dermomyotome by reducing meox1 expression to yield increased numbers of SDECs. We have thus kept these results in our manuscript. The Wnt signaling referred to in the (Nguyen et al., 2014) paper was specific to wnt16, a non-canonical wnt ligand that operates exclusively in the sclerotome and thus irrelevant to our current studies.

As for adding more on the bipotency of the early ECs, we have obtained scRNA-Seq data sets from ECs purified at earlier time points in development (Figure 2, and Figure 2 —figure supplement 1). We have now added a differential expression analysis comparing selected EC subsets between 15 ss and 22 hpf. We have found that genetic programs driving endothelial differentiation were upregulated in all subsets tested, in contrast to tissue-specific differentiation programs (i.e., somitogenesis), which were downregulated. These results suggest an early commitment to the EC lineages as early as 15 ss.

Reviewer #3:

This manuscript explores the origins of endothelial cells that make up the dorsal aorta (DA) in the developing zebrafish. The DA is the site of the hemogenic endothelium that produces definitive hematopoietic stem cells (HSCs). The prevailing model is that a subset of lateral plate mesoderm is fated for the dorsal aorta. These cells are further segregated into cells that make up the aortic endothelium, and those with hemogenic potential. In this study the authors identify an additional source of DA endothelium derived from the somites, which are conventionally thought of as defined blocks of skeletal muscle. These cells in the dermamyotome are positive for etv2:gfp+ cells and are proposed as bipotential precursors that can give rise to endothelium or muscle. The authors conclude that these cells are supportive of HSC emergence, but do not themselves produce HSCs. This is a novel concept in the ontogeny of hemogenic endothelium.

1. In Figure 1, it would be clearer if the etv2:gfp line was crossed to a red endothelial-specific line (e.g. flk:mCherry). Without additional landmarks in S1 it is unclear if these cells are continuously migrating towards the midline and becoming incorporated in the DA, or if they are moving out of the focal plane but are still part of the somite. Additionally, the movies S1 and S2 need to have features of interest labeled including boundaries between different tissue types.

We agree that our previous figures could have been more clear. We have now overhauled our imaging figures and present improved timelapse imaging movies. To better target the somites and image their contribution, we lineage traced PM-specific ECs, and photoconverted somite-derived cells (see explanation below). In addition, we annotated features and boundaries in Figure 4 – video 1 and 2.

2. Following the point above, the migration between etv2:gfp+ cells in the dermamyotome and the DA needs to be better defined. There should be more discrete lineage tracing, such as a photoactivatable marker, or time-lapse live imaging to show incorporation of these cells in the DA. The existing time-lapse shows delamination of etv2:gfp+ cells from the dermamyotome, but not migration and incorporation into the DA.

Agreed. We now present much improved lineage tracing experiments. To specifically mark SDECs, we used a PM-specific lineage tracer, tbx6:Cre, and two “switch” lines. In the first combination, tbx6:Gal4; Tg(UAS-Cre); A2BD animals show switching of PM-derived lineages from BFP+ to DsRed+ cells, allowing SDECs to be followed upon leaving the somite and integrating into the DA. In the second combination, tbx6:Gal4; Tg(UAS-Cre); kdrl:CSY animals will switch out a CFP cassette to YFP in PM-derived cells. Because this switched transgene will only be expressed in endothelial cells due to the kdrl driver, only SDECs will be marked with YFP, whereas all other ECs will remain CFP+ due to derivation from non-PM-derived precursors. In both lineage tracing lines, we show that the vast majority of PM-derived ECs migrate to the floor of the DA. In addition, we have added a meticulously performed analysis of the migration and localization of SDECs to the DA by using a photoconvertible Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 transgenic line. By applying precise photoconversion of single somite pairs and following their trajectories into the DA, we further support the contribution of SDECs to the DA, provide a detailed map of the somites involved (trunk somites), and quantify the number of SDECs contributed by each somite pair.

3. A key conclusion of the study is that SDEC cells are integrated into the DA are supportive of HSCs. If that is the case, the authors could perform time-lapse of tbx6:Cre;kdrl:CSY or some other reporter line combination to observe where budding HSCs are emerging relative to SDECs. This would capture the spatial relationships between the proposed supportive SDECs and HSCs.

See response to comment 2. Capturing the precise interplay over time between SDECs and HE within the DA is difficult and requires generation of new transgenic lines. We are currently building these as we feel this is an exciting avenue for our future studies.

4. Does meox1 knockdown result in enhanced proliferation of etv2+ cells in the dermamyotome, or do more cells emerge de novo? Single cell lineage tracing or tracking of single cells in time-lapse movies could help distinguish between these possibilities.

This issue was addressed in the studies of (Nguyen et al., 2014), where they performed phosphohistone H3 staining in meox1 mutants. They showed no increase in proliferation within the somites in meox1 mutants, suggesting that the increase in SDECs is likely due to a difference in cell fate outcomes. Our data support this postulate in that further genetic perturbations (in notch3-/-; meox1 KD animals, e.g.) reveal broad endothelial potential within the somite. Ectopic EC formation appears to occur via lineage skewing within bipotent muscle / EC precursors within the dermomyotome.

5. In Figures 2O and 3O the change in expression levels from bulk mRNA of pooled whole embryos does not necessarily mean that there is a change in the numbers of specific cell types. This result could also occur if there are similar cell types present that have altered expression profiles. These results could be clarified by FACS analysis of different populations to detect a change in ratios.

We agree that the results from bulk mRNA of pooled whole embryos in notch3 and cloche mutants are suggestive and do not necessarily indicate a change in cell number. FACS analysis for these different populations in their respective mutant backgrounds is not feasible due to the crosses required. Instead, we toned down our conclusions and suggested further inquiry into the idea that loss of notch3 results in premature depletion of muscle progenitors or reduced expression/activity within these cells.

6. The authors do not go back further in ontogeny to explore possible heterogeneity within the dermamyotome. For example, how early do the etv2:gfp cells arise from the PSM and can be distinguished as a bipotential dermamyotome precursor?

We have obtained scRNA-Seq data sets from ECs purified at earlier development times (Figure 2, and Figure 2 —figure supplement 1). We show that specification of EC subsets, including SDECs, can already be observed as early as tailbud stage. As noted above in response to Reviewer 1, by comparing selected EC subsets between 15 ss and 22 hpf, we have found evidence suggesting an early commitment to the EC lineages following the end of gastrulation. Similar to observations in the avian model (Pouget et al., 2006), we have never detected endothelial transcripts or Etv2:GFP+ cells in the PSM. In addition, the earliest Etv2:GFP+ cells were observed at 12 ss. We also confirmed this observation by FISH, where we observed etv2+ / meox1+ and etv2+ / pax3+ cells at 12 ss. Furthermore, when we knocked down meox1, we observed ectopic formation of meox1+ / etv2+ cells. Together, these data indicate that SDECs arise from a bipotent progenitor population as early as the 12ss.

7. What is the significance or conclusions drawn from the regional anterior-trunk-posterior observations presented in Figure 5?

That, relative to cells marked by a drl:CreERT2 lineage tracer that labels all ECs, a tbx6:Cre driver specifically targeting SDECs generates ECs that traffic only to the anterior / trunk portions of the aorta. We have clarified this in the text.

8. The rationale for choosing lines for scRNA-seq (Etv2:kaede, Fli1:dsRed; Tp1:GFP, and Draculin:Dendra:H2B) is not clear because these lines are not specific to ECs and are known collectively to mark blood, endocardium, and other cell types. How do the authors distinguish between subtypes of ECs versus other cell types altogether?

We chose a variety of lines to ensure we would mark all EC subsets and have reference populations built into our data sets. For example, etv2:Kaede is a general marker of endothelium, fli1:DsRed/tp1:GFP marks arterial endothelium, and draculin (drl) labels LPM, ECs, and HECs depending upon the developmental timepoint. Having distinct overlapping and nonoverlapping populations allowed us to validate our results via the presence or absence of the respective populations of interest. For example, SDECs do not appear in drl-marked cells as this line marks LPM descendants. Following unbiased clustering of all ECs, we used the coexpression of landmark genes to assign ECs with different tissue origins (Figure 1 —figure supplement 1 and source data 1). We have clarified this in the text.

9. The authors could show in Figure 6 where etv2:kaede and the other sorted reporter lines map onto these clusters. The authors could also provide scRNA-seq data for the same etv2:gfp+ cells that were presented in previous figures to confirm the presence of SDECs.

Good ideas. We have revised Figure 6 (now Figure 1) to include a UMAP and heatmap with EC cluster identity and differentially expressed genes used to help identify each cluster.

10. A recent paper comparing a new etv2 knock-in line with existing transgenic lines shows differences in expression patterns (Chestnut & Sumanas Dev Dyn 2020). Although data is presented in this study that shows endogenous expression of etv2 using FISH, it is possible that expression and conclusions drawn from the etv2:gfp line does not precisely follow endogenous cell types. This caveat should be considered and discussed within the larger context of the study.

See response to comment 8. Because each EC marker captures only a subset of the total EC populations, we used a broader strategy to capture and analyze ECs, by sorting ECs not only from TgBAC(etv2:Kaede)ci6 embryos but also from Tg(fli1:eGFP)y1; Tg(tp1:GFP)um14, and Tg(drl:H2B-dendra). For imaging and lineage-based experiments, we have used a validated version of etv2 reporter line, Tg(etv2.1:eGFP)zf372 (Veldman and Lin, 2012) that, was shown to be more precise than the previous BAC-based etv2 reporter line. In addition, most of our major conclusions based upon the Tg(etv2.1:eGFP)zf372 reporter lines are complemented by our FISH studies for endogenous etv2 expression; we observed close correlations with the transgenic pattern in all cases analyzed.

11. It is unclear from the FISH images if the etv2+ cells are in fact double positive with meox1, or if etv2+ cells are intercalated between meox1 cells. The green meox1 signal appears low within the boundaries of etv2+ cells. Could this be resolved by crossing the etv2:gfp line with a red muscle progenitor reporter line, if available?

We do not have a red reporter line for muscle cells available. However, our results show coexpression of etv2 with pax3a (Figure 4F); thus, both experiments support that SDECs are derived from shared progenitors.

12. The authors do not discuss the paper by Murayama et al. (Nat. Comm. 2015) that shows stromal cells in the caudal hematopoietic tissue (CHT) directly delaminate from the somites. There are important similarities between this manuscript and the previous study that are not explored or referenced. For example, in Figure 5G, what are the round cells in the CHT that have switched? Could these be the stromal cells observed by Murayama et al? The switched cells in Figure 5J have endothelial morphology, but not the cells in Figure 5G.

We believe the cells under study in the Murayama paper are vascular smooth muscle cells or stromal cells based upon their morphology and lineal origin from the sclerotome. We thus did not think the papers were directly comparable. That said, on some rare occasions, we occasionally detected lineage-traced cells in the posterior caudal vein, intersomitic vessels or extravascular space. We now discuss these findings and the possibility of having labeled some of the stromal cells identified by Murayama and colleagues.

13. The authors say that lineage tracing of tbx6:cre cells in the DA are consistent between multiple embryos over different stages, however this should be better quantified with more of the data represented in the analysis. It would be informative to show data before the 90 hpf timepoint in Figure 5E-J.

Agreed. We have now included photoconvertible Tg(actb2:nls-Eos); Tg(fli1:eGFP)y1 transgenic animals that describe the migration of the SDECs with greater precision and show their contribution to the DA at 36 hpf (Figure 3). In addition, our imaging experiments using tbx6:Gal4; Tg(UAS-Cre); A2BD show PM-derived SDECs incorporated into the floor of the DA at 48 hpf (Figure 8A-C).

14. Are the Cre reporter lines single insertions? It appears from the FACs plots in S3 there are a range of signal intensities, and that some cells may express both colors, suggesting incomplete recombination of multiple insertions. This is possible too in Figure 5G that shows cell that appear double positive. Are there additional controls that could be presented?

We found that all cre lines utilized segregated in mendelian ratios, consistent with only single insertions. In addition, we found that the range of signal intensities in the switched RFP+ (Figure 8 —figure supplement 1A) is mirrored by the range of intensities observed in the unswitched BFP+ cells (Figure 8 —figure supplement 1B). Therefore, we believe that fluorophore expression levels are based on the hematopoietic cell type in which they are expressed and not on transgene copy number.

15. The draculin-cre result is not surprising as the line markers endothelial and blood precursors that, as expected, are labeled in this experiment. These results are independent of SDEC ontogeny. This should be clarified.

The drl:CreERT2 results were added as a control / contrast to those obtained with tbx6:Cre. We have clarified this information in the text.

16. In Figure 7A, bmp4 only appears upregulated in a small number of SDECs, and yet the conclusion is made that this is an important factor produced by SDECs.

It is true that bmp4 is only expressed in a few SDECs. However, we observe no bmp4 expression in endothelial cells from the LPM and believe that this difference could be due to heterogeneity within SDECs and/or resolution of the scRNA-Seq datasets. That said, we also felt the bmp4 results should be further investigated and reported in future work. We have removed the results on bmp4 and id1 from this manuscript.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The reviewers are very enthusiastic about the revised manuscript. The only outstanding issue is the presentation of new data that does not have quantification or statistical analysis.

Reviewer #3 (Recommendations for the authors):

In this study, the authors identify an additional source of dorsal aorta endothelium derived from the somites, which are conventionally thought of as defined blocks of skeletal muscle. These cells in the dermamyotome are positive for endothelial cell markers and are proposed as bipotential precursors that can give rise to endothelium or muscle. The authors conclude that these cells are supportive of HSC emergence, but do not themselves produce HSCs.

Most of the previous reviewers' comments have been addressed. However, there are new imaging data presented that has not been analyzed quantitatively or rigorously tested for significant changes using statistical tests.

1. Quantification of results in Figure 4F,G, 5A-F, 5I-N, 7C-D.

Figure number 4F and 4G: For each experiment, sections from 6 independent embryos were taken. We observed 1-2 etv2 positive cells per somite in each of the embryos examined.

Figure 5A-F: For each experiment, sections from 3 independent embryos were taken. We observed 3-4 etv2 positive cells per somite in the meox1 morphants compared to 1-2 etv2 positive cells per somite in the siblings.

Figure 5I-K: For each experiment, sections from 5 independent embryos were taken. We observed 3-4 etv2 positive cells per somite in the meox1 morphants and >8 etv2 positive cells in the meox1; Mib double morphants.

Figure 5L-N: For each experiment, sections from 3 independent embryos were taken. We observed 2-4 etv2 positive cells per somite in the notch3 mutants and >6 etv2 positive cells in the notch3 mutants; Mib morphants.

Figure 7C-D: The number of repeats for the control (n=7) and the WNT inhibitor (n=5) is already depicted in the image. For the WNT inhibited embryos, 4 out of 5 showed abnormal migration of the SDECs.

2. Is decrease in meox1 significant in Figure 7B? If not, this should be clarified.

We thank the reviewer for his comment. We repeated the statistical analysis of the WNT inhibitor versus sibling data set. We observed decreased expression of axin2 with a concomitant increase in etv2 by qRT-PCR. In addition, we observed a reduction in meox1 expression, although not statistically significant. All genes analyzed between Wnt inhibitor and sibling embryos, except meox1, showed a statistically significant difference (p<0.001, unpaired, two-tailed Student's t-test; n=3.). We updated the, text, figure legend and methods section.

3. Need more detail about microscopy that allowed resolution of CFP and YFP from kdrl:CSY line in Figure 8E-J.

We thank the reviewer for his comment. We added the following description of the imaging settings and filters used to distinguish between CFP and YFP in the material and methods sections. “Since the emission spectra of CFP and YFP overlap, images were captured in two separate sequences to filter overlapping signals by limiting the PMT detection. For CFP a 476nm laser was used, and the PMT detector was set to collect signals ranging between 480nm and 505nm. For YFP, a 514nm laser was used, and the PMT detector was set to collect signals ranging between 520nm and 570nm.”

4. The drl:Cre-ERT2 data in Figure 8H-K is not clearly convincing and could be removed. The tbx6:Cre data makes the point well.

We thank the reviewer for his comment. Draculin is an early broad marker gene for the lateral plate mesoderm. Therefore using the drl:CreERT2 as a driver for converting endothelial cells is a good control to show the potential of these cells to convert, hence, the efficiency of the reporter line. Thus, the data in Figure 8 serves not only to show that most cells in the DA are descendants of LPM but also as control for our kdrl:CFP to YFP reporter fish line.

https://doi.org/10.7554/eLife.58300.sa2

Article and author information

Author details

  1. Pankaj Sahai-Hernandez

    Department of Cell and Developmental Biology, University of California, San Diego, La Jolla, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing
    Contributed equally with
    Claire Pouget, Shai Eyal and Ondrej Svoboda
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7867-8346
  2. Claire Pouget

    Department of Cell and Developmental Biology, University of California, San Diego, La Jolla, United States
    Present address
    Angiocrine Bioscience, San Diego, United States
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing
    Contributed equally with
    Pankaj Sahai-Hernandez, Shai Eyal and Ondrej Svoboda
    Competing interests
    No competing interests declared
  3. Shai Eyal

    Department of Cell and Developmental Biology, University of California, San Diego, La Jolla, United States
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing
    Contributed equally with
    Pankaj Sahai-Hernandez, Claire Pouget and Ondrej Svoboda
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8131-8077
  4. Ondrej Svoboda

    Department of Cell and Developmental Biology, University of California, San Diego, La Jolla, United States
    Present address
    Department of Cell Differentiation, Institute of Molecular Genetics, Academy of Sciences of the Czech Republic v.v.i, Prague, Czechia
    Contribution
    Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – review and editing
    Contributed equally with
    Pankaj Sahai-Hernandez, Claire Pouget and Shai Eyal
    Competing interests
    No competing interests declared
  5. Jose Chacon

    Department of Cell and Developmental Biology, University of California, San Diego, La Jolla, United States
    Contribution
    Formal analysis, Investigation, Writing – review and editing
    Competing interests
    No competing interests declared
  6. Lin Grimm

    Department of Cell and Developmental Biology, University of California, San Diego, La Jolla, United States
    Contribution
    Formal analysis, Validation, Investigation, Writing – review and editing
    Competing interests
    No competing interests declared
  7. Tor Gjøen

    Department of Pharmacy, University of Oslo, Oslo, Norway
    Contribution
    Software, Formal analysis, Investigation, Writing – review and editing
    Competing interests
    No competing interests declared
  8. David Traver

    Department of Cell and Developmental Biology, University of California, San Diego, La Jolla, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Writing – original draft, Writing – review and editing
    For correspondence
    dtraver@ucsd.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9652-7653

Funding

National Institute of Diabetes and Digestive and Kidney Diseases (R01-DK074482)

  • David Traver
  • Pankaj Sahai-Hernandez

NIH Office of the Director (TR01-OD026219)

  • David Traver

National Institutes of Health (T32-HL086344)

  • Pankaj Sahai-Hernandez

California Institute for Regenerative Medicine (EDUC4-12804)

  • Shai Eyal

American Heart Association (19POST34380328)

  • Ondrej Svoboda

Ministry of Education of the Czech Republic (RVO: 68378050-KAV-NPUI)

  • Ondrej Svoboda

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank members of the Traver laboratory for helpful discussions and Stephanie Grainger for discussions and careful editing of the manuscript. We thank the UCSD Institute for Genomic Medicine sequencing core for supporting the scRNA-seq sample preparation and sequencing. We would like to thank Dr. Peng Guo from the Nikon Imaging Center and Jennifer Santini from the Microscopy core at UCSD for their support. We want to thank Anthony Santella from the Bao laboratory at Memorial Sloan Kettering Cancer Center and Kurt Weiss from the Huisken laboratory at the Morgridge Institute for Research for their support and expert advice. Funding This work was supported by the National Institute of Health [R01-DK074482 to DT and PS-H, TR01-OD026219 to DT]; National Institute of Health [T32-HL086344 to PS-H]; California Institute of Regenerative Medicine [EDUC4-12804 to SE]; American Heart Association [19POST34380328 to OS]; and Ministry of Education of the Czech Republic [RVO: 68378050-KAV-NPUI to OS].

Ethics

Zebrafish embryos and adult fish were raised in a circulating aquarium system (Aquaneering) at 28 °C and maintained by UCSD Institutional Animal Care and Use Committee (IACUC) guidelines (Protocol number: S04168), the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) (Accreditation number: 000503), and the Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals (Assurance number: A3033-1 or D16-00020). All experiments were performed under approved methods of anesthesia, and every effort was made to minimize suffering.

Senior Editor

  1. Didier YR Stainier, Max Planck Institute for Heart and Lung Research, Germany

Reviewing Editor

  1. Owen J Tamplin, University of Wisconsin-Madison, United States

Version history

  1. Received: April 27, 2020
  2. Preprint posted: May 15, 2020 (view preprint)
  3. Accepted: September 3, 2023
  4. Version of Record published: September 11, 2023 (version 1)

Copyright

© 2023, Sahai-Hernandez, Pouget, Eyal et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 176
    Page views
  • 28
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Pankaj Sahai-Hernandez
  2. Claire Pouget
  3. Shai Eyal
  4. Ondrej Svoboda
  5. Jose Chacon
  6. Lin Grimm
  7. Tor Gjøen
  8. David Traver
(2023)
Dermomyotome-derived endothelial cells migrate to the dorsal aorta to support hematopoietic stem cell emergence
eLife 12:e58300.
https://doi.org/10.7554/eLife.58300

Further reading

    1. Developmental Biology
    2. Neuroscience
    Igor Y Iskusnykh, Nikolai Fattakhov ... Victor V Chizhikov
    Research Article

    Development of the nervous system depends on signaling centers – specialized cellular populations that produce secreted molecules to regulate neurogenesis in the neighboring neuroepithelium. In some cases, signaling center cells also differentiate to produce key types of neurons. The formation of a signaling center involves its induction, the maintenance of expression of its secreted molecules, and cell differentiation and migration events. How these distinct processes are coordinated during signaling center development remains unknown. By performing studies in mice, we show that Lmx1a acts as a master regulator to orchestrate the formation and function of the cortical hem (CH), a critical signaling center that controls hippocampus development. Lmx1a co-regulates CH induction, its Wnt signaling, and the differentiation and migration of CH-derived Cajal–Retzius neurons. Combining RNAseq, genetic, and rescue experiments, we identified major downstream genes that mediate distinct Lmx1a-dependent processes. Our work revealed that signaling centers in the mammalian brain employ master regulatory genes and established a framework for analyzing signaling center development.

    1. Developmental Biology
    2. Evolutionary Biology
    Salvatore D'Aniello, Stephanie Bertrand, Hector Escriva
    Feature Article

    Cephalochordates and tunicates represent the only two groups of invertebrate chordates, and extant cephalochordates – commonly known as amphioxus or lancelets – are considered the best proxy for the chordate ancestor, from which they split around 520 million years ago. Amphioxus has been an important organism in the fields of zoology and embryology since the 18th century, and the morphological and genomic simplicity of cephalochordates (compared to vertebrates) makes amphioxus an attractive model for studying chordate biology at the cellular and molecular levels. Here we describe the life cycle of amphioxus, and discuss the natural histories and habitats of the different species of amphioxus. We also describe their use as laboratory animal models, and discuss the techniques that have been developed to study different aspects of amphioxus.