ETV2 expression marks blood and endothelium precursors, including hemogenic endothelium, at the onset of blood development


  • Sarah Wareing,

    1. Cancer Research UK Stem Cell Hematopoiesis Group, Paterson Institute for Cancer Research, University of Manchester, Manchester, United Kingdom
    Search for more papers by this author
  • Alexia Eliades,

    1. Cancer Research UK Stem Cell Hematopoiesis Group, Paterson Institute for Cancer Research, University of Manchester, Manchester, United Kingdom
    Search for more papers by this author
  • Georges Lacaud,

    1. Cancer Research UK Stem Cell Biology Group, Paterson Institute for Cancer Research, University of Manchester, Manchester, United Kingdom
    Search for more papers by this author
  • Valerie Kouskoff

    Corresponding author
    1. Cancer Research UK Stem Cell Hematopoiesis Group, Paterson Institute for Cancer Research, University of Manchester, Manchester, United Kingdom
    • Paterson Institute for Cancer Research, University of Manchester, Wilmslow Road, M20 4BX Manchester, UK
    Search for more papers by this author


Background: ETV2 has been identified as an important player in embryonic hematopoiesis. However, the cell populations in which this transcription factor is expressed and operates during blood specification remain to be fully characterized. Here we address these issues using ES cells and a transgenic mouse line expressing green fluorescent protein (GFP) under the control of ETV2 regulatory elements, allowing us to observe the tight association between ETV2 expression and the initiation of hematopoiesis. Results: Both in differentiating ES cells and gastrulating embryos ETV2::GFP is mostly found co-expressed with endothelial markers and defines a subset of cells with greatly enriched primitive erythroid potential. Upon culture ETV2::GFP cells rapidly up-regulate CD41, down-regulate endothelium cell surface markers and generate definitive hematopoietic progenitors. Altogether these characteristics represent the hallmark of hemogenic endothelium cells, a specialized endothelium originating from the hemangioblast and giving rise to hematopoietic cells. Importantly, ETV2 deficiency results in a complete absence of hemogenic endothelium in differentiating ES cells and gastrulating embryos. Conclusions: Altogether our data reveal that ETV2 marks hemogenic endothelium in gastrulating embryos and is absolutely required for the formation of this precursor at the onset of hematopoiesis. These results enhance our understanding of embryonic hematopoiesis and the factors driving hemogenic endothelium specification. Developmental Dynamics 241:1454–1464, 2012. © 2012 Wiley Periodicals, Inc.


During embryogenesis, hematopoiesis occurs in sequential waves, producing specific blood cell types to meet the needs of the growing embryo at different developmental stages. The first transient wave of primitive erythrocyte production occurs in the yolk sac soon after gastrulation, and is rapidly followed by the production of definitive blood cell types (Palis et al., 1999). Shifting to an intra-embryonic site, a second wave of hematopoiesis occurs along the ventral aspect of the dorsal aorta and in all major arteries of the developing embryo. This wave of hematopoiesis has been shown to produce the hematopoietic stem cell population that will subsequently seed and expand in the fetal liver before migrating to the bone marrow, where it maintains the life-long supply of blood cells in the adult organism (Dzierzak and Speck, 2008). Due to the close proximity of the endothelial and hematopoietic lineages within the extra-embryonic yolk sac blood islands, hematopoietic and endothelial development in the early embryo have long been believed to be closely linked (Sabin, 1920). Studies using differentiating embryonic stem (ES) cells and gastrulating embryos have shown this to be the case by identifying the hemangioblast, a multi-potent precursor that gives rise to hematopoietic, endothelial and smooth muscle lineages (Kennedy et al., 1997; Choi et al., 1998; Nishikawa et al., 1998; Huber et al., 2004). The hemangioblast emerges in the primitive streak from Brachyury-expressing mesoderm and expresses the vascular endothelial growth factor receptor fetal liver kinase 1 (FLK1) (Fehling et al., 2003; Huber et al., 2004). Furthermore, there is a large body of evidence demonstrating that definitive blood cells, emerging from intra-embryonic sites, have an endothelial origin (Jaffredo et al., 1998; de Bruijn et al., 2000; Zovein et al., 2008; Bertrand et al., 2010; Boisset et al., 2010; Kissa and Herbomel, 2010), but it was not until recently that yolk sac hematopoiesis was also demonstrated to proceed from hemangioblast to hematopoiesis by means of an endothelial intermediate with hematopoietic potential, known as hemogenic endothelium (Eilken et al., 2009; Lancrin et al., 2009). This yolk sac hemogenic endothelium population expresses TIE2 and cKIT, but is negative for the αIIb integrin CD41 (Lancrin et al., 2009). Acquisition of CD41 expression followed by the down-regulation of endothelial markers is indicative of the progression to fully committed hematopoietic progenitors (Ferkowicz et al., 2003; Mikkola et al., 2003; Sroczynska et al., 2009a).

Studies of the ETS family transcription factor ETV2 within the context of early embryonic development have revealed a vital role for this protein during mesoderm specification. ETV2 has been shown to be critically required for both hematopoiesis and vasculogenesis as ETV2-deficient embryos die by E11.0 with a complete absence of blood progenitors and severe vascular defects (Lee et al., 2008). Interestingly, while the function of ETV2 in vasculogenesis seems to be well conserved across evolution in higher vertebrates, the role of this transcription factor in hematopoiesis is much less conserved with no apparent role in Xenopus blood formation and only a limited function in the formation of myeloid progenitors in zebrafish (Sumanas and Lin, 2006; Pham et al., 2007; Liu and Patient, 2008; Sumanas et al., 2008; Neuhaus et al., 2010; Ren et al., 2010; Salanga et al., 2011). Nevertheless, despite significant advances in understanding the roles of ETV2 in development and more recently its fundamental requirement for hematovascular mesoderm specification (Kataoka et al., 2011), the precise stage and requirement of ETV2 expression in early hematopoietic precursor populations remains to be fully characterized. To address this, we generated ES cells and corresponding mice carrying a transgene expressing green fluorescent protein (GFP) under the control of ETV2 regulatory sequences. Analysis of ETV2::GFP expression both in vitro and in vivo allowed us to accurately define ETV2 expression at the onset of hematopoiesis. Specifically we demonstrate that both in differentiating ES cells and gastrulating embryos ETV2 is detected in the FLK1+ population but that it also marks the next stage of specification, the hemogenic endothelium. Importantly, knocking out ETV2 function reveals that ETV2 is not just expressed in the hemogenic endothelium, it is also essential for its formation.


Generation and Validation of a Transgene to Track ETV2 Expression

To easily track ETV2 expression, a transgenic construct was created by placing the IRES-VenusGFP sequence downstream of the ETV2 coding sequence (Fig. 1A), incorporating both 5′ and 3′ regulatory elements sufficient to drive ETV2 expression during embryogenesis (Ferdous et al., 2009). ES cells carrying this ETV2-IRES-VenusGFP construct (hereafter referred to as Etv2GFP) differentiated normally as assessed by polymerase chain reaction (PCR) detection of genes up-regulated as cells become committed to the hematopoietic and endothelial lineages (Fig. 1B) and by flow cytometry analysis of cell surface markers of early hematopoietic differentiation (Fig. 1C). To assess the suitability and accuracy of this Etv2GFP construct to mark Etv2-expressing cells, we next analyzed by PCR the presence of Etv2 transcripts in GFP and GFP+ cells isolated from in vitro differentiated embryoid bodies (EBs). The expression of Etv2 was highly enriched in the GFP+ fraction isolated from day 2.2 Etv2GFP EBs compared with the GFP fraction and to the presorted cells (Fig. 1D). Next, we generated a mouse transgenic line from these ETV2GFP ES cells to also track and analyze ETV2 expression during embryogenesis. To similarly demonstrate the accuracy of the Etv2GFP construct in vivo, GFP+FLK1+, GFPFLK1+ and GFPFLK1 cell populations were sorted from pooled E7.5 Etv2GFP embryos and analyzed by real-time PCR for the presence of Etv2 transcripts (Fig. 2). The highest level of Etv2 expression was found in the GFP+FLK1+ fraction while very low levels were detected in the two other subpopulations. Altogether, these data establish that both in vitro and in vivo our transgenic tracking construct accurately marks cells expressing ETV2.

Figure 1.

Etv2 expression is associated with hemangioblast and hemogenic endothelium markers. A: Schematic representation of the transgene used to express Venus-green fluorescent protein (GFP) under Etv2 regulation. To validate the Etv2GFP ES cell line, single cell suspensions isolated from wild-type (WT) and Etv2GFP embryoid bodies (EBs) between day 1 and 6 of differentiation were analyzed by (B) semiquantitative polymerase chain reaction (PCR) and (C) fluorescence-activated cell sorting (FACS) for the expression of markers indicative of the early differentiation steps of endothelium and hematopoiesis. These markers appear with similar timing in the two cell lines. D: The GFP and GFP+ fractions were isolated from day 2.2 Etv2GFP EBs and, along with the nonsorted population (presort), were analyzed for Etv2 expression by semi-quantitative PCR. All data are representative of at least three experiments.

Figure 2.

Green fluorescent protein (GFP) accurately tracks ETV2 expression in Etv2GFP embryos. Based on their relative expression of the ETV2::GFP and FLK1 markers, the three subpopulations ETV2::GFP FLK1, ETV2::GFP+ FLK1+ and ETV2::GFP FLK1+ were sorted from pooled Etv2GFP embryonic day (E) 7.5 embryos and analyzed by real-time quantitative polymerase chain reaction (PCR) for ETV2 mRNA which is strongly detected only in the ETV2::GFP+ fraction. Data are representative of at least three experiments. Error bars represent 1 SD for triplicate values.

ETV2 Is Associated With the Emergence of the Hemogenic Endothelium In Vitro

To characterize the dynamics of ETV2 expression at the onset of hematopoietic specification, we next analyzed its expression relative to the hemangioblast marker FLK1 during EB differentiation. ETV2::GFP was detected from day 2 of differentiation and mostly associated with FLK1 expression (Fig. 3), a result anticipated in light of the importance of ETV2 in FLK1 regulation (De Val et al., 2008; Lee et al., 2008). During the time course of EB differentiation, a small fraction of cells expressed only ETV2::GFP and at day 3, a subpopulation of cells was positive for FLK1 expression but negative for ETV2::GFP. To gain insight into the dynamics of these two markers and the events occurring at the transition from hemangioblast to hemogenic endothelium, we analyzed the biological potential of the four subpopulations defined by their relative expression of ETV2::GFP and FLK1 (Fig. 4A). Day 2.2 EB cells were sorted into the four fractions (Fig. 4B), cultured for 24 hr and then analyzed for immunophenotypic changes. Before day 3, EB cells expressed little of the hemogenic endothelium marker TIE2 and virtually no CD41 (Fig. 1C). However, after 24 hr in culture high levels of TIE2 expression were found co-expressed with ETV2::GFP for all fractions (Fig. 4C, middle panel), suggesting that TIE2+ hemogenic endothelium precursors are derived from cells expressing ETV2. Furthermore, only the ETV2::GFP+FLK1+ sorted fraction generated a significant population of CD41+ hematopoietic precursors after 24 hours, suggesting that this population was the most advanced toward hematopoiesis (Fig. 4C, lower panel). Notably, upon culture most cells in both ETV2::GFP FLK1+ and ETV2::GFP+FLK1 fractions up-regulated the second marker for which they were initially negative hence becoming ETV2::GFP+ FLK1+ double positive (Fig. 4C, upper panel), a rather unexpected finding given the suggestion that ETV2 controls FLK1 expression in mesoderm (Lee et al., 2008). The data shown here instead indicate that either gene can be switched on first, priming mesodermal cells to initiate hematopoiesis. Alternatively, ETV2::GFPFLK1+ cells could represent mesoderm that has a high degree of plasticity and converts to ETV2::GFP+FLK1+ hematopoiesis when cultured under permissive conditions. Altogether, these data demonstrate that in vitro hemogenic endothelium and CD41+ progenitors are generated from FLK1+ ETV2-expressing cells.

Figure 3.

ETV2 and FLK1 expression are tightly correlated upon in vitro differentiation. Fluorescence-activated cell sorting (FACS) analysis of single cell suspensions obtained from WT and Etv2GFP EBs between day 1 and 5 of differentiation reveals a strong association between ETV2::GFP and hemangioblast marker FLK1. Data are representative of at least three experiments. GFP, green fluorescent protein.

Figure 4.

Hemogenic endothelium cells emerge from the green fluorescent protein (GFP) -expressing population in differentiating Etv2GFP embryonic stem (ES) cells. A: Schematic representation of the experimental design. B: Representative dot plots of the purity of the four populations sorted from day 2.2 Etv2GFP embryoid bodies (EBs) based on their relative expression of ETV2::GFP and FLK1 expression. C: Fluorescence-activated cell sorting (FACS) analysis of FLK1, TIE2, and CD41 markers following the re-aggregation of sorted fractions for 24 hr in hemangioblast media. Data are representative of three independent experiments.

ETV2 Expression Marks the Yolk Sac Wave of Hemogenic Endothelium

Next, we analyzed ETV2::GFP expression in gastrulating embryos and this revealed a clear localization of ETV2::GFP+ cells within the extra-embryonic yolk sac at all stages of gastrulation (Fig. 5A). Similar to the in vitro data, ETV2::GFP was found mostly co-expressed with FLK1 in gastrulating embryos (Fig. 5B,C). However, the fraction of FLK1+ cells not expressing ETV2::GFP, most likely representing other mesoderm, was larger in vivo than in vitro. Strikingly, ETV2::GFP was mostly found associated with the hemogenic endothelium markers TIE2 and cKIT. However, in the early stages of gastrulation (primitive streak and neural plate), we observed a small subset of ETV2::GFP+ cells that were positive for FLK1 but negative for TIE2, and it is likely that this fraction represents hemangioblasts which have previously been isolated based on FLK1 expression alone (Huber et al., 2004). In addition, the expression of Brachyury, a mesodermal and hemangioblast marker, was much higher in ETV2::GFP+ cells isolated from embryonic day (E) 7.5 embryos than from E8.5 embryos, further suggesting the presence of hemangioblast within the ETV2::GFP+ population of early gastrulating embryos (Fig. 5D). Etv2 and Brachyury were similarly shown to be co-expressed in FLK1+ cells derived from in vitro differentiated ES cells (Wareing et al., 2012). Of interest, while in primitive streak embryos ETV2::GFP+ cells were negative for CD41 expression, CD41+ cells with low ETV2::GFP levels were detected in older headfold embryos (gated population; Fig. 5C), suggesting a progressive down-regulation of ETV2 expression as commitment toward hematopoiesis occurs. To further characterize the cell population expressing ETV2 at the onset of hematopoiesis, cells from pooled E7.5 embryos were sorted based on FLK1 and ETV2::GFP expression (Fig. 6A). By necessity several litters were combined to obtain sufficient cells for a sort so the embryos were therefore a mixture of primitive streak, neural plate and headfold stage. Gene expression analysis revealed that the early endothelial-hematopoietic transcription factors Scl, Gata2, and Fli1 were highly expressed in the ETV2::GFP+FLK1+ subpopulation, with minimal or undetectable levels in ETV2::GFPFLK1 and ETV2::GFPFLK1+ cells (Fig. 6B). The specific and restricted expression of these transcriptional regulators strongly suggest that at this stage of embryonic development all endothelial and hematopoietic potential is restricted to the subset of FLK1+ cells expressing Etv2.

Figure 5.

Etv2 expression localizes to yolk sac hemogenic endothelium in Etv2GFP embryos. A: Localization of ETV2::GFP+ cells to the yolk sac region of Etv2GFP gastrulating embryos (LS, late streak; EB, early bud; LB, late bud; EHF, early headfold; LHF, late headfold; staging according to Downs and Davies, 1993). Images were taken using a Leica DMI3000 microscope with a 10X objective (numerical aperture 0.25) equipped with a DCF310F camera. Merged bright field and fluorescent images are shown. B: Transversal section of neural plate stage ETV2GFP embryos stained for green fluorescent protein (GFP; green) and FLK1 (red) expression. C: Dynamics of expression of ETV2 in gastrulating embryos. Fluorescence-activated cell sorting (FACS) analyses of GFP alongside TIE2, FLK1, and CD41 in E7.5 Etv2GFP embryos, with one wild-type (WT) embryo shown as a GFP control. Representative data for Etv2GFP embryos at primitive streak, neural plate, and headfold stages are shown. D: Real-time quantitative polymerase chain reaction (PCR) of Brachyury marker was performed on RNA extracted from FLK1+GFP+CD41 cells isolated from embryonic day (E) 7.5 and E8.5 ETV2GFP embryos. Error bars represent SD for three independent experiments. All data are representative of at least three independent experiments.

Figure 6.

ETV2-expressing cells co-express hematopoietic transcription factors. A: Representative purity of the fractions sorted from pooled embryonic day (E) 7.5 Etv2GFP embryos. Embryos were sorted into three fractions based on their relative expression of FLK1 and ETV2::GFP defining the three subpopulations ETV2::GFP FLK1, ETV2::GFP FLK1+, and ETV2::GFP+ FLK1+. The middle panels show the purity of the sorted fractions. The lower panels show the relative level of CD41 expression in the sorted populations. B: Real-time quantitative polymerase chain reaction (PCR) of hematopoietic markers was performed on RNA isolated from the sorted populations. Error bars represent 1 SD for triplicate values. Data are representative of three independent experiments. GFP, green fluorescent protein.

To formally establish whether ETV2 truly marks a hemogenic endothelium population, E7.5 FLK1 and ETV2::GFP-expressing fractions were again sorted from pooled embryos excluding any potential CD41-expressing cells (Fig. 7A,B). Following 3 days of culture in hemogenic endothelium-supporting conditions, ETV2::GFP+FLK1+ CD41 cells acquired high levels of CD41 and down-regulated endothelial markers as well as ETV2::GFP (Fig. 7C). To assess the presence of hematopoietic precursors, clonogenic replating assays were performed and revealed a dramatic increase in the frequency of hematopoietic progenitors after 3 days of culture in this subpopulation when compared with the initial sorted population (Fig. 7D). In contrast, both ETV2::GFPFLK1 and ETV2::GFPFLK1+ populations produced very few hematopoietic colonies either at the time of sorting or following culture. Interestingly, primitive erythroid potential was found exclusively in the ETV2::GFP+ subsets at the time of sort (Fig. 7D). These data indicate that the ETV2-expressing population is highly enriched for all hematopoietic potential and also defines the hemogenic endothelium population in gastrulating embryos. The replating assays revealed that the ETV2::GFPFLK1+ population also contained a very small number of definitive progenitors at the time of sorting. It is possible that this population contained a minority of cells which had progressed to ETV2::GFP+FLK1+ but subsequently down-regulated ETV2::GFP, and were primed to immediately up-regulate CD41 and to generate hematopoietic progenitors upon further culture. Hematopoietic populations which co-express the endothelial marker CD34 together with CD41 are highly enriched for definitive hematopoietic precursors (Pearson et al., 2008) so it is not unexpected that this fraction, expressing endothelial marker FLK1, would give rise exclusively to definitive colonies. The ETV2::GFP+FLK1 population, from which the CD41+ cells were not excluded, contains a very high number of primitive erythroid progenitors at the time of sorting. These cells are likely to represent the earliest wave of yolk sac hematopoiesis, known to be highly enriched for primitive progenitors. Upon culture, this population did not express endothelium markers or produce blood progenitors. Altogether these data demonstrate that the ETV2::GFP+FLK1+ cell population is highly enriched for bona fide hemogenic endothelium that progressively up-regulates CD41, down-regulates endothelial markers and generates hematopoietic progenitors upon further culture.

Figure 7.

ETV2::GFP+ CD41 cells from gastrulating embryos give rise to CD41+ hematopoietic progenitors. A: Schematic representation of the experimental design. B: Representative dot plots of the purity of fractions sorted from pooled embryonic day (E) 7.5 Etv2GFP embryos. Embryos were sorted into four populations based on FLK1 and green fluorescent protein (GFP) expression: ETV2::GFPFLK1, ETV2::GFPFLK1+, ETV2::GFP+FLK1+ and ETV2::GFP+FLK1. CD41+ hematopoietic cells were excluded from all subpopulations except the GFP+FLK1 population. The upper panels show the purity of the sorted fractions based on GFP and FLK1 expression, the lower panels show the successful exclusion of CD41+ cells from three populations. C: Purity of ETV2::GFP+ FLK1+ CD41 cells sorted from pooled E7.5 Etv2GFP embryos (left) and fluorescence-activated cell sorting (FACS) analysis of these cells after 3 days of culture on a monolayer of OP9 stroma (right). OP9 cells were excluded based on their size and high levels of GFP expression. D: Clonal hematopoietic progenitor assays of ETV2::GFP+ FLK1+ CD41, ETV2::GFP FLK1 CD41 and ETV2::GFP FLK1+ CD41 populations at the time of sorting or after 3 days in hemogenic endothelium culture on OP9. In the top graph, the ETV2::GFP+ CD41+/− sorted population was included as a positive control for the presence of committed hematopoietic progenitors. EryP, primitive erythroid colonies; EryD, definitive erythroid colonies; Mac, macrophage colonies; Mix, mixed lineage colonies; Myeloid, granulocyte-macrophages colonies. Data are representative of two independent experiments.

Etv2−/− ES Cells and Mice Lack Hemogenic Endothelium

We next investigated the impact of ETV2 deficiency on the formation of yolk sac hemogenic endothelium using Etv2−/− ES cells and the corresponding Etv2-deficient mouse line. Differentiating Etv2−/− ES cells failed to form hemogenic endothelium populations as assessed by a complete lack of cell surface expression of TIE2 and CD41, and a total absence of the immunophenotypic TIE2+cKIT+ population of hemogenic endothelium cells when compared with wt differentiating ES cells (Fig. 8A). The Etv2−/− ES cells additionally did not express significant levels of the endothelial markers VE-Cadherin and SOX7 at any of the time points tested. Interestingly, low levels of Tie2 transcripts were observed at day 5 and 6 of EB differentiation (Fig. 8B). Upon culture of FLK1+ cells isolated from day 3.5 EBs Etv2−/− cells also failed to produce the typical hemogenic endothelial cores and the hematopoietic cells that emerge from those cores (Fig. 8C), previously shown to stain positive for endothelial and hematopoietic markers, respectively (Lancrin et al., 2009; Costa et al., 2012). Similarly, E7.5 Etv2−/− embryos, while containing FLK1+ cells, lacked the immunophenotypic FLK1+ TIE2+ cKIT+ population representing hemogenic endothelium emerging in gastrulating embryos (Fig. 8D). Importantly, as described elsewhere, ETV2-deficient ES cells and embryos were unable to generate primitive and definitive hematopoietic progenitors (Wareing et al., 2012). Altogether these data reveal that, both in vivo and in vitro, formation of hemogenic endothelium at the onset of yolk sac hematopoiesis requires ETV2 function.

Figure 8.

ETV2 is absolutely required for the formation of hemogenic endothelium. A: Representative fluorescence-activated cell sorting (FACS) analysis of CD41, TIE2, and cKIT expression on single cell suspensions obtained from wild-type (WT) and Etv2−/− day 5 embryoid bodies (EBs). B: Semiquantitative polymerase chain reaction (PCR) of endothelial marker expression in WT and Etv2−/− cells isolated from EBs between day 1 and 6 of in vitro differentiation. C: FLK1+ cells were sorted from day 3.5 WT and Etv2−/− EBs and cultured for 2 days in hemangioblast culture conditions on gelatin-coated dishes. The white arrows point to endothelial cores and the black arrow indicates detaching round hematopoietic cells, as observed in the WT culture only. D: Representative FACS analysis of WT and Etv2−/− embryonic day (E) 7.5 embryos shows absence of the FLK1+TIE2+cKIT+ population in the absence of ETV2 expression. Data are representative of three independent experiments.


One of the main benefits of the ES/EB system is that it provides a means to generate sizeable populations of a cell type of interest, which may be virtually inaccessible or present in too small numbers in vivo. Studies which test the idea that differentiating ES cells are a very close model for early embryonic development are always important to refine our ideas of the accuracy of using solely in vitro data to make general conclusions (Sroczynska et al., 2009b; Pearson et al., 2010). Our in vitro and in vivo data are overall in agreement, providing further evidence for the suitability of the ES/EB system for modeling early embryonic development. However, the two systems did show some discrepancies. In vitro, early EBs contained a small ETV2::GFP+ FLK1 population which up-regulated FLK1 to become double-positive on further culture. In vivo, however, an equivalent population was not observed in the earliest embryos studied, and a similar population only appeared during the later stages of gastrulation when ETV2::GFP+ FLK1low cells were present, likely a result of the down-regulation of FLK1 in the ETV2::GFP+ FLK1+ population. In primitive streak embryos, ETV2::GFP was first seen in cells that were already FLK1+. This is in concordance with another recent study that claims that ETV2 is up-regulated within committed FLK1+PDGFRα+ mesoderm and specifies vascular mesoderm from cells which are therefore already FLK1+ (Kataoka et al., 2011). Additionally, the ETV2::GFP FLK1+ population is much smaller in vitro than in vivo. This likely reflects the tendency of our specific EB culture conditions, optimized for the production of hematopoietic cells, to drive the formation of hematovascular mesoderm in preference to other mesodermal types.

Sorting populations from pooled embryos is a necessity when working at such early time points in development, but our fluorescence-activated cell sorting (FACS) analysis of individual E7.5 embryos at discrete stages of gastrulation revealed the rapidly changing dynamic of ETV2 expression even over this very short developmental window. While it is clear from our study that the hemogenic endothelium marked by ETV2::GFP represents a relatively large and easily isolated population within pooled E7.5 embryos, the hemangioblast has proved more difficult to isolate and has only been derived at very low frequencies from a heterogeneous population of FLK1+ cells (Huber et al., 2004). Our results suggest that before the up-regulation of TIE2 and other endothelial markers associated with progression to the hemogenic endothelium, ETV2 and FLK1 may together mark a very transient population of cells which could contain the elusive hemangioblast. This would require sorts of early stage embryos, limiting cell number to the point where sorting becomes extremely challenging.

Because ETV2 is expressed strongly in a population of cells which is very highly enriched for hemogenic endothelium potential, this rather suggests that it has an important role to play here. We have performed microarray analyses in Etv2−/− ES cells revealing that hematopoietic genes are not expressed in these cells, and a considerable number of endothelial genes are either absent or down-regulated when ETV2 is not present (Wareing et al., 2012). Finding out which genes are directly regulated by ETV2 will help us to elucidate how this Ets family member fits into the transcriptional network. As ETV2 appears to be important for expression of both endothelial and hematopoietic genes (Kataoka et al., 2011; Wareing et al., 2012), teasing apart the specific requirements of ETV2 in hematopoietic and endothelial development will be necessary to fully comprehend its importance to each lineage. We do not currently know how much the hemogenic endothelium contributes to true endothelium, if it even contributes at all, so vascular defects present in ETV2 knock-out embryos might be due to defects in a different population of ETV2-expressing cells. Whether this is linked to the hemogenic endothelium which we have further characterized here, or is an entirely independent population of mesodermal cells, is another issue that needs to be resolved.

Another question outside the scope of this study but which would be of great significance is whether ETV2 is required to generate the hemogenic endothelium present in the dorsal aorta and major arteries from which hematopoietic stem cells (HSCs) are derived. Both our model and the work of others suggest that ETV2 is down-regulated rapidly in the developing embryo, undetectable by the time HSCs begin to emerge from the dorsal aorta (Lee et al., 2008; Kataoka et al., 2011), but it is conceivable that ETV2 is required initially to generate precursors of the cells of the dorsal aorta that will later go on to produce HSCs. To address this question, better in vitro culture conditions for supporting pre-HSC populations and the emergence of HSCs would have to be established. Only then would we be able to more fully understand the events in emergence of hematopoietic progenitors from different embryonic sites which are regulated by ETV2. A recent publication has revealed a role for ETV2 in adult hematopoiesis (Lee et al., 2008). Bone marrow and spleen cells from adult Etv2GFP mice used in our study were assessed for ETV2::GFP expression but no clear ETV2::GFP+ population was observed (data not shown). One possibility could be that the transgene is inactivated in adult mice, or potentially that it does not contain regulatory elements required for ETV2 expression within the adult hematopoietic system. This raises the intriguing possibility that ETV2 could be under distinct regulation in different cell types and at different stages of development. How ETV2 is regulated, and indeed what factors are involved in its activation, is another area that warrants investigation.


In summary, our study defines the expression pattern of ETV2 at the onset of blood development both in vitro and in vivo (Fig. 9). We show that ETV2 is expressed in hemogenic endothelium and is necessary for the formation of this specialized endothelium population. These results make an important contribution to our understanding of embryonic hematopoiesis, but are also of interest for the expanding field of reprogramming and potential derivation of blood cells for therapy. The hemogenic endothelium represents an extremely important intermediate step in the production of blood cells, and fully understanding the factors involved in its formation is crucial for the directed differentiation of stem cells to the various hematopoietic lineages.

Figure 9.

Schematic representation depicting the sequential stages of hematopoietic specification during early embryonic development. [Color figure can be viewed in the online issue, which is available at]


Transgenic ES Cells

Recombineering was used to create a plasmid with IRES-H2B-VenusGFP-Neomycin downstream of ETV2 coding sequence together with 5 kb upstream region (Copeland et al., 2001). Briefly, the ETV2 locus was retrieved from a BAC by phage λ-mediated recombination. An IRES-H2B-VenusGFP-Neomycin cassette flanked with ETV2 homology arms was inserted into the ETV2 locus by means of recombination. Following electroporation ES cell clones were selected with G418. Transient Flpe recombinase expression removed the Neomycin cassette to generate the finished Etv2GFP ES cell line. Similar recombineering techniques were used to generate a construct for targeting and deletion of the ETV2 locus in ES cells. Details of oligo sequences used for the various cloning steps and screens of ES cell clones are available on request.


Etv2GFP ES cells were injected into blastocysts to generate an Etv2GFP mouse line. ES cells with one copy of ETV2 targeted for conditional deletion were injected into blastocysts and chimeric mice were backcrossed onto the C57BL/6 line. Mice were crossed with a PGK-Cre-expressing line to generate the knocked out allele. Timed matings were set up between Etv2GFP males and ICR females, or between Etv2−/+ heterozygous mice. The morning that a plug was observed was taken to be E0.5. Morphological landmarks were used to accurately stage gastrulating embryos. All animal work was performed under regulations set out by the Home Office Legislation under the 1986 Animal Scientific Procedures Act.

ES and Embryo Cell Culture

ES cell lines were maintained, differentiated, and cultured in hemangioblast media as described (Lancrin et al., 2009; Sroczynska et al., 2009b). Populations sorted from E7.5 embryos were seeded onto irradiated OP9 stromal cells in hemogenic endothelium media (Mukouyama et al., 2000).

Flow Cytometry

Flk1-Bio, Tie2-PE, CD41-APC, and cKit-APC-eFluor780 monoclonal antibodies were purchased from eBioscience. For cell surface marker staining, cells were incubated at +4°C in the presence of various combinations of these mAbs. When Flk1-Bio was used, a secondary staining was performed with streptavidin-PE-Cy7 (eBioscience). Cells were sorted using an Aria or Influx and analyzed with an LSRII cytometer (BD Biosciences).

Gene Expression Analysis

Total RNA was extracted from ES or embryo-derived cells using the RNeasy Plus Mini Kit or RNeasy Micro Kit (Qiagen). cDNA was generated using the Omniscript RT Kit (Qiagen) with random hexamer oligos. PCR was performed using GoTaq and real-time PCR was performed on an ABI7900 (Applied Biosystems) with universal probe library (Roche).

Immunofluorescence Staining

Immunostaining was performed on E7.5 ETV2GFP embryos using rat anti-mouse Flk1 (BD Pharmingen), rabbit anti-GFP (MBL) followed by anti-rat Alexa Fluor-555 and anti-rabbit Alexa Fluor-647 (Invitrogen) staining as described previously (Pearson et al., 2010). Slides were then mounted with ProLong Gold antifade reagent with DAPI (Molecular Probes). Images were taken with a Low-light Zeis Axiovert 200M system and an Andor iXon DU888+ camera.


We thank the BRU team for help with all mouse work and lab members for critical reading of the manuscript. This work is funded by Cancer Research UK, grant number C147/A6058 and a grant from the Biotechnology and Biological Sciences Research Council (BBSRC, Swindon, UK). S.W. and A.E. designed and performed the research, analyzed the data and wrote the manuscript. V.K. and G.L. designed and supervised the research project, analyzed the data, and wrote the manuscript. The authors declare no competing financial interests.