Etv2 rescues Flk1 mutant embryoid bodies

Abstract

Independent mouse knockouts of Etv2 and Flk1 are embryonic lethal and lack hematopoietic and endothelial lineages. We previously reported that Flk1 activates Etv2 in the initiation of hematopoiesis and vasculogenesis. However, Flk1 and its ligand VEGF are expressed throughout development, from E7.0 to adulthood, whereas Etv2 is expressed only transiently during embryogenesis. These observations suggest a complex regulatory interaction between Flk1 and Etv2. To further examine the Flk1 and Etv2 regulatory interaction, we transduced Etv2 and Flk1 mutant ES cells with viral integrants that inducibly overexpress Flk1 or Etv2. We demonstrated that forced expression of Etv2 rescued the hematopoietic and endothelial potential of differentiating Flk1 and Etv2 mutant cells. We further discovered that forced expression of Flk1 can rescue that of the Flk1, but not Etv2 mutant cells. Therefore, we conclude that the requirement for Flk1 can be bypassed by expressing Etv2, supporting the notion that disruption of Etv2 expression is responsible for the early phenotypes of the Etv2 and Flk1 mutant embryos.

Introduction

Flk1 is the primary receptor for Vascular Endothelial Growth Factor (VEGF). Flk1 mutant embryos maintain Flk1 reporter expression as demonstrated in LacZ knockin/knockout embryos, but fail to develop hematopoietic or endothelial lineages (Shalaby et al., 1995). Flk1 mutant embryonic stem cells (ESCs) that are differentiated on a stromal cell layer using OP9 co-culture techniques do not yield hematopoietic cells, although Brachyury positive mesoderm clusters form under these conditions (Hidaka et al., 1999). Surprisingly, however, Flk1 mutant ESCs can develop hematopoietic and endothelial lineages, albeit at a reduced level, when induced to differentiate using embryoid body (EB) formation (Hidaka et al., 1999). Previous studies have revealed that hematopoietic progenitors are present in the E7.5 Flk1 mutant embryos and that Flk1 deficient cells cannot migrate from the posterior primitive streak to the yolk sac (Schuh et al., 1999). These results support the notion that Flk1⁺ hematopoietic and endothelial progenitors encounter inductive signals that allow further differentiation as the progenitors migrate into the yolk sac (Hidaka et al., 1999). While there is a three dimensional component in EB differentiation, structures such as the primitive streak and yolk sac are not present and therefore migration may not be necessary to receive these inductive cues in the context of EB differentiation.

Mouse knockouts of Etv2 and Flk1 result in similar embryonic lethal phenotypes, which lack hematopoietic and endothelial lineages (Ferdous et al., 2009; Koyano-Nakagawa et al., 2012; Lee et al., 2008; Shalaby et al., 1995). Previous discrepant observations regarding the genetic hierarchy of Etv2 and Flk1 can be explained with a model in which Etv2 and Flk1 reciprocally regulate each other dependent on the developmental stages. At early stages of mesodermal induction and endothelial specification, Flk1 signaling is required for Etv2 expression, as Flk1 mutant embryos lack Etv2 expression (Rasmussen et al., 2012). Flk1 is expressed in the Etv2 mutant embryos and in Etv2 mutant ESCs after 4 days of differentiation with OP9 coculture techniques, indicating that Flk1 is necessary for Etv2 expression but not vice versa, at these developmental stages (Kataoka et al., 2011; Rasmussen et al., 2012). Further evidence that Flk1/VEGF signaling can activate the Etv2 promoter demonstrates that this signaling cascade is operational (Rasmussen et al., 2012). In contrast, after completion of gastrulation and specification of the endothelial lineage, Etv2 may reinforce Flk1 expression. Transcriptional assays and promoter analysis provide evidence that Etv2 binds and activates the Flk1 promoter (De Val et al., 2008; Lee et al., 2008). This notion is further supported as there is decreased expression of Flk1 in Etv2 mutant embryos (Ferdous et al., 2009; Kataoka et al., 2011; Lee et al., 2008; Rasmussen et al., 2011; Rasmussen et al., 2012). However, these reports vary widely as to the degree of this effect, suggesting that while Etv2 may reinforce Flk1 expression, this could be a minor effect or could occur in a tissue or temporal specific pattern.

Although ample evidence demonstrates that Flk1 is upstream of Etv2 at the onset of vasculogenesis and hematopoiesis (see above), the expression patterns of these genes are significantly different; Etv2 is expressed transiently during development whereas Flk1 and its ligand VEGF are expressed throughout development, from E7.0 into adulthood (Dumont et al., 1995). Thus, it is possible that the Flk1 signal activates targets other than Etv2 to initiate vascular and hematopoietic development. To definitively address whether Etv2 is the Flk1 target necessary for hematopoietic and endothelial differentiation, we undertook a rescue strategy of the Flk1 and Etv2 mutant EB phenotypes with forced expression of Flk1 or Etv2. We demonstrated using the ES/EB system that Etv2 could rescue the Flk1 mutant phenotype, but Flk1 could not rescue the Etv2 mutant phenotype, further supporting the notion that the Flk1-Etv2 cascade is operational during hemato-endothelial development.

Results

In order to perform the rescue experiments, we chose to use Etv2 mutant ES cells derived in our laboratory and Flk1 mutant ES cells, generously provided by the Stanford laboratory (Hidaka et al., 1999). We initially characterized these cells by analyzing transcript levels throughout differentiation. A wildtype 2.1 line that was derived from a littermate blastocyst of the Etv2 mutant ESCs and a wildtype R1 line, the line genetically engineered to generate the Flk1 mutant were used as controls. We differentiated the wildtype 2.1 ESCs, the Etv2 mutant ESCs, the R1 ESCs, and the Flk1 mutant ESCs side by side using a hanging-drop method and harvested cells every 24 hours from embryoid body day (EB D) 2 through EB D6. Etv2 expression peaks at EB D3 in the 2.1 EBs and is absent in the Etv2 mutant EBs (Fig. S1 A). Flk1 expression is minimally affected or unaffected in the Etv2 mutant cell line (Fig. S1B), which supports our previous observation that Etv2 plays a minor, if any, role upstream of Flk1 signaling. Brachyury (Bry) was used as a marker of early mesoderm and Fgf5 was used as a marker of embryonic ectoderm and the epiblast state (Leahy et al., 1999). Both of these undifferentiated germ layer markers are increased in the Etv2 mutant compared to wildtype cell lines (Fig S1 C-D). This may be due to impaired differentiation and therefore, more cells may be committed to the germ layers, but not fully differentiated. Finally, we observed that in the Etv2 mutant cells, markers of the endothelial and hematopoietic lineages are absent, such as Lmo2 (Fig. S1E), Scl (Fig. S1F), Gata1 (Fig. S1G), and Embryonic Globin (Fig. S1H).

The wildtype R1 line showed slightly different expression patterns from the wildtype 2.1 line. For example, the expression of Etv2 was detectable as early as EB D2 and was more broadly expressed, peaking at EB D4 (Fig. S1I). In the Flk1 mutant, Etv2 expression was significantly delayed and was undetectable until EB D4 (Fig. S1I). This supports the notion that Flk1 signaling is one of the early regulators of Etv2 expression, but other factors can compensate, at least at later time points during EB development. The presence of Etv2 may also explain why Flk1 mutant EBs are capable of generating some endothelial and hematopoietic cells. As expected, the Flk1 transcript was undetectable in the Flk1 mutant compared to the R1 wildtype (Fig. S1J). We also observed that Bry and Fgf5 are expressed at similar levels in the mutant compared to the wildtype, although the timing is delayed by a day (Fig. S1K-L). Furthermore, markers of endothelial and hematopoietic lineages are reduced or delayed in the Flk1 mutant compared to R1 control EBs (Fig. S1 M-P).

These mutant mouse ES cells were further genetically modified using a stably-integrating, doxycycline inducible lenti-viral vector. cDNAs encoding Flk1 and Etv2 with an HA tag (HA-Etv2) were subcloned into the pSAM2 vector (Darabi et al., 2012), where GFP was replaced with the mCherry reporter gene, and will be referred to as Flk1-pSam2 and Etv2-pSam2, respectively. The reporter gene was expressed in response to doxycycline along with the subcloned gene, and allowed monitoring of the response of the transduced cells to doxycycline. After enrichment, all four resulting transduced cell stocks, including Etv2 mutant:Flk1-pSam2, Flk1 mutant:Flk1-pSam2, Etv2 mutant:Etv2-pSam2, and Flk1 mutant:Etv2-pSam2, responded to doxycycline by expressing mCherry in 60-80% of the cell population (Fig. 1 A,B,D,E). We confirmed that this inducible expression strategy resulted in an increase in the appropriate transcript levels using qPCR (data not shown). Furthermore, we confirmed that the respective protein was induced by performing western blots of the cell extracts for Flk1 and the HA tag. Neither the Etv2 nor the Flk1 mutant cell lines transduced with Flk1-pSam2 had Flk1 expression in the absence of doxycycline. However, in the presence of doxycycline, both cell lines showed robust Flk1 expression (Fig. 1C). Likewise, the Etv2 and Flk1 mutant ES/EB cell lines transduced with Etv2-pSam2 had no detectable levels of HA expression in the absence of doxycycline. In the presence of doxycycline, the Flk1 mutant ES/EBs transduced with Etv2-pSam2 displayed a robust HA-Etv2 band, and the Etv2 mutant ES/EBs transduced with the same virus displayed a clear induction, albeit at a lower level (Fig. 1F). Therefore, we established four ESC stocks that expressed the transduced proteins only in the presence of doxycycline.

Figure 1. Transduced cell lines overexpress Etv2 or Flk1.

Etv2 mutant (A) or Flk1 mutant (B) ESCs stably transduced with Flk1-pSam2 express mCherry in response to 1μg/ml doxycycline as detected by flow cytometric analysis. Both cell lines transduced with Flk1-pSam2 express Flk1 protein only in the presence of doxycycline as shown by western blot analysis (C). Etv2 mutant (D) or Flk1 mutant (E) embryonic stem cells stably transduced with Etv2-pSam2 express mCherry in response to 1μg/ml doxycycline by flow cytometric analysis. Both cell lines transduced with Etv2-pSam2 express HA-tagged Etv2 protein only in the presence of doxycline as shown by western blot analysis (F). The control lanes contain cell extract from 293T cells transiently transfected with either a Flk1 expression construct (C) or an Etv2-HA expression construct (F).

First, we examined whether overexpression of Etv2 or Flk1 could rescue the Etv2 mutant phenotype. We differentiated the wildtype 2.1 ES cells, the Etv2 mutant: Etv2-pSam2 ES cells, and the Etv2 mutant: Flk1-pSam2 ES cells side by side using a hanging-drop method, and cultured without doxycycline or with 1 μg/ml of doxycycline from days (D) 2-4, or D4-6 and analyzed the EBs at day 6 (EB D6). As expected, the wildtype cells were unaffected using any doxycycline treatments (data not shown). At EB D6, these wildtype cells contained approximately 3% endothelial cells (Flk1⁺/CD31⁺) (Fig. 2A, upper panel; Fig. 2D, red bar), 20% hematopoietic progenitors (CD41⁺/CD45^-) (Fig. 2A, lower panel: Fig. 2E, red bar), and a small number of hematopoietic cells (CD41⁺/CD45⁺) (Fig. 2A, lower panel; Fig. 2F, red bar, Table S1). As expected the Etv2 mutant: Etv2-pSam2 cells with no doxycycline treatment had essentially no (0.3%) endothelial cells (Flk1⁺/CD31⁺) (Fig. 2B, upper panel; Fig. 2D, Table S1) and essentially no hematopoietic progenitors (CD41⁺/CD45^-) or hematopoietic cells (CD41⁺/CD45⁺) (Fig. 2B, lower panel; Fig. 2E and 2F, Table S1). Doxycycline treatment from EB D2-4 increased the Flk1⁺/CD31⁺ endothelial population to normal levels (Fig. 2B, upper panel; Fig. 2D, Table S1); treatment from EB D4-6 further increased the same population. The lack of Flk1 protein on the cell surface of the Etv2 mutant EBs and the rescue of Flk1 protein expression after doxycycline induction of Etv2 supports the notion that Etv2 may activate Flk1 through positive feedback. Due to the similar expression levels of Flk1 in Etv2 wildtype and mutant EBs (Fig S1B), Etv2 may induce pathways that affect Flk1 translation or localization instead of directly regulating Flk1 transcription. Doxycycline treatment also had a strong positive effect on the hematopoietic compartment. Treatment from EB D2-4 rescued both the CD41⁺/CD45^- and CD41⁺/CD45⁺ cell populations (Fig. 2B, lower panel; Fig. 2E and 2F, Table S1). Treatment from EB D4-6 yielded an increase of the CD41⁺/CD45^- population, but with minimal contribution to the CD41⁺/CD45⁺ cell population (Fig. 2B, lower panel; Fig. 2E and 2F, Table S1). These data demonstrate that forced expression of Etv2 could, at least partially, rescue the defect in endothelial and hematopoietic differentiation of Etv2 mutant EBs.

Figure 2. Etv2, but not Flk1, rescues the hematopoietic and endothelial compartments of Etv2 mutant embryoid bodies.

(A-C) Flow cytometric analysis of a representative differentiation experiment comparing Wildtype 2.1 EBs (A), Etv2 mutant:Etv2-pSam2 (B) and Etv2 mutant:Flk1-pSam2 (C) at EB D6. Wildtype and mutant cells were analyzed for endothelial (CD31, Flk1) (upper panel) and hematopoietic markers (CD41, CD45) (lower panel) in multiple doxycycline induction schemes: no doxycycline (A, B left panels, C left panels), 1μg/ml doxycycline from EB D2-4 (B middle panels, C middle panels), and 1μg/ml doxycycline from EB D4-6 (B right panels, C right panels). (D-F) Quantification of the FACS profiles is graphically displayed for CD31+/Flk1+ endothelial cells (D) CD41+/CD45- hematopoietic progenitor cells (E), and CD41+/CD45+ hematopoietic cells (F). Results from the wildtype control are shown in red and results from Etv2 mutant cells are shown in black. Error bars indicate SEM, (n equals at least 3, see Table S1).

We next tested whether overexpression of Flk1 could rescue the Etv2 mutant phenotype. Similar to the Etv2 mutant: Etv2-pSam2 cells, the Etv2 mutant: Flk1-pSam2 cells without doxycycline treatment also showed a minimal contribution to the CD31⁺/Flk1⁺, CD41⁺/CD45^-, or CD41⁺/CD45⁺ populations (Fig. 2C-F, Table S1). While the doxycycline treatment from EB D2-4 did not have an increase in Flk1 expression at EB D6, doxycycline treatment from EB D4-6 showed an increase in Flk1⁺ cells (Fig. 2C), as expected with forced overexpression of Flk1. However, forced expression of Flk1 did not allow these cells to re-express CD31⁺/Flk1⁺, CD41⁺/CD45^-, or CD41⁺/CD45⁺ (Fig 2C-F, Table S1). Therefore, Etv2 mutant EBs could not be rescued by forced expression of Flk1.

In parallel experiments, we examined whether overexpression of Etv2 or Flk1 could rescue the phenotype of Flk1 mutant ESCs. We differentiated the wildtype R1 ES cells, the Flk1 mutant: Flk1-pSam2 ES cells, and the Flk1 mutant: Etv2-pSam2 ES cells side by side and cultured with or without doxycycline from EB D2-4, or EB D4-6. The wildtype cells were minimally affected by the doxycycline treatment, with a slight reduction in CD31⁺/Flk1⁺ cells (data not shown). The Flk1 mutant: Flk1-pSam2 cells with no doxycycline treatment had CD31⁺/Flk1⁺ endothelial cells present, albeit at a much reduced level than the wildtype control (Fig. 3A, 3B, 3D, Table S2). The presence of a Flk1⁺ population is likely due to leaky expression of the Flk1-pSam2 vector as this was almost entirely absent in the Flk1 mutant: Etv2-pSam2 cells (Fig. 3C, 3D). While the Flk1 mutant: Flk1-pSam2 cells without doxycycline induction had nearly as many CD31⁺/Flk1^- and CD41⁺/CD45^- cells as the wildtype control (Fig. 3A, 3B, 3D, 3E, Table S2), there was a striking reduction of CD41⁺/CD45⁺ hematopoietic cells (Fig. 3B, 3F, Table S2) consistent with the previously reported phenotype (Hidaka et al., 1999). Forced expression of Flk1 from EB D2-4 or EB D4-6 nearly doubled the CD31⁺/Flk1^- population, and induction from EB D4-6 led to a robust increase of Flk1⁺/CD31⁺ population (Fig. 3B, 3D, Table S2). We also observed a 2-3 fold increase in CD41⁺/CD45^- cells when induced from either EB D2-4 or EB D4-6 and a rescue of the CD41⁺/CD45⁺ population when induced from EB D2-4 (Fig. 3B, 3E, 3F, Table S2). Similar to the observation that D4-6 induction of Etv2 mutant: Etv2-pSam2 cells was unable to rescue the CD41⁺/CD45⁺ population (Fig. 2C, 2F, Table S2), D4-6 induction of Flk1 mutant: Flk1-pSam2 cells did not rescue this population (Fig. 3B, 3F, Table S2). This may be due to the ability of Flk1 to rescue during this time period and a longer chase period may be needed to observe the CD45⁺ cells. Nevertheless, our data clearly demonstrate that Flk1 can rescue the Flk1 mutant EB phenotype, indicating that the cells, timing of doxycycline treatment, and the viral construct are functional and appropriate to evaluate the rescue of the respective cell populations.

Figure 3. Etv2 or Flk1 rescues the phenotype of Flk1 mutant embryoid bodies.

(A-C) Flow cytometric analysis of a representative differentiation experiment comparing Flk1 wildtype (R1) EBs (A), Flk1 mutant:Flk1-pSam2 (B) and Flk1 mutant:Etv2-pSam2 (C) at EB D6. Wildtype and mutant cells were analyzed for endothelial (CD31, Flk1) (upper panels) and hematopoietic markers (CD41, CD45) (lower panels) in multiple doxycycline induction schemes: no doxycycline (A, B left panels, C left panels), 1μg/ml doxycycline from D2-4 (B middle panels, C middle panels), and 1μg/ml doxycycline from D4-6 (B right panels, C right panels). (D-F) Data from all experiments are graphically displayed for CD31+/Flk1- (D, left panel) cells and CD31+/Flk1+ endothelial cells (D, right panel), CD41+/CD45- hematopoietic progenitor cells (E), and CD41+/CD45+ hematopoietic cells (F). Results from the wildtype control are shown in red and results from mutant cells are shown in black. Error bars indicate SEM (n equals at least 3, see Table S2).

Next, we examined whether overexpression of Etv2 could rescue the Flk1 mutant phenotype. The Flk1 mutant Etv2-pSam2 cells without doxycycline treatment had a significant defect in CD31⁺/Flk1^- and CD31⁺/Flk1⁺ cells as well as a two fold reduction in the CD41⁺/CD45^- population and an almost complete loss of CD41⁺/CD45⁺ cells (Fig. 3C-3F, Table S2). To quantify the endothelial cells, we analyzed the CD31⁺/Flk1^- population, because the Flk1 mutant cells are not able to express Flk1. Doxycycline induction at EB D2-4 and EB D4-6 resulted in an increase of CD31⁺ cells to approximately the level observed in the wildtype (Fig. 3C, 3D, Table S2). Doxycycline induction at EB D2-4 and EB D4-6 also resulted in an increase in CD41⁺/CD45^- cells and induction of EB D2-4 but not EB D4-6 rescued the CD41⁺/CD45⁺ population (Fig. 3C, 3E, 3F, Table S2), similar to the observation with Flk1 overexpression. Therefore, Etv2 can rescue the Flk1 mutant endothelial and hematopoietic phenotypes.

To further ensure that these phenomena were not just limited to expression of the selected cell surface markers, we performed qPCR to analyze the transcript levels of known lineage markers. Representative hematopoietic (CD41) and endothelial (vWF and Cdh5) genes were examined. Etv2 mutant cells without doxycycline treatment had an almost complete loss of Cdh5 (Fig. 4B), and CD41 (Fig. 4C), and a significant decrease in vWF (Fig. 4A) compared to the wildtype cells. Forced expression of Etv2 from EB D2-4 or EB D4-6 rescued or partially rescued expression of all 3 of these genes. However forced expression of Flk1 from EB D2-4 or EB D4-6 had no effect on these genes. These results further support the previous observations that Flk1 cannot rescue the defects of Etv2 mutant EBs.

Figure 4. Etv2 expression rescues the expression of hematopoietic and endothelial transcripts in Etv2 and Flk1 mutant embryoid bodies.

(A-C) Expression levels vWF (A), Cdh5 (B), and CD41 (C) in Wildtype 2.1 (grey) and transduced Etv2 mutant D6 EBs (black). (D-F) Expression levels of vWF (D), Cdh5 (E), and CD41 (F) in Wildtype R1 (grey) and transduced Flk1 mutant D6 EBs (black). All results were normalized to the respective wildtype cells. The doxycycline induction scheme, transduced virus, and cell lines are indicated under each graph. (n= 3), Error bars represent SEM.

Flk1 mutant cells without doxycycline treatment had a reduction in expression of vWF, Cdh5, and CD41. Forced expression of Flk1 rescued (and increased above wildtype level) expression of vWF, Cdh5, and CD41 (Fig. 4D, 4E, 4F). Furthermore, forced expression of Etv2 in Flk1 mutant cells rescued the expression levels of vWF, Cdh5, and CD41 (Fig. 4D-4F). These data further support the finding that Etv2 expression can bypass the need for Flk1 in developing EBs.

Finally, to examine whether the rescue of cell surface markers and gene expression leads to functional recovery of the hematopoietic lineages and to address whether treatments from EB D2-4 and/or EB D4-6 can rescue the hematopoietic lineage, we performed colony forming assays by plating dissociated cells from D6 EBs in methylcellulose. Etv2 mutant cells without doxycycline treatment (Fig. 5A-5C) displayed no hematopoietic colonies (Fig. 5A-5C). Forced expression of Etv2 in Etv2 mutants from EB D2-4 partially rescued the number of Granulocyte-macrophage (GM) colonies (Fig. 5A), but had little effect on erythroid burst forming units (BFUs) (Fig. 5B) and GM, erythrocyte, and megakaryocyte mixed colonies (GEMMs) (Fig. 5C). Forced expression of Etv2 from EB D4-6 completely rescued the number of observed GM, BFU, and GEMM colonies (Fig. 5A-5C). In contrast, forced expression of Flk1 in the Etv2 mutant EBs from EB D2-4 or EB D4-6 had no effect on any colony types (Fig. 5A-5C). Therefore, Etv2 can rescue the hematopoietic colony formation of the Etv2 mutant ESCs, but Flk1 is unable to rescue the hematopoietic colony formation of the Etv2 mutant EBs.

Figure 5. Etv2 rescues hematopoietic progenitors in the Etv2 mutant and Flk1 mutant EBs.

Cells were dissociated from Wildtype 2.1 (red bar) and Etv2 mutant D6 EBs (A-C) transduced with Etv2-pSam2 (white bars) or Flk1-pSam2 (grey bars) or Wildtype R1 and Flk1 mutant EBs (D-F) transduced with Etv2-pSam2 (white bars) or Flk1-pSam2 (grey bars). 20,000 cells were plated in methocult M3434 in triplicate and cultured for 7 days. Colonies were counted and identified as either GM (granulocyte or macrophage containing) (A,D), BFU (erythrocyte burst forming units) (B,E), or GEMM (granulocyte, erythrocyte and macrophage and megakaryocyte mixed) colonies (C,F). n=3, error bars represent SEM.

Flk1 mutant cells with no doxycycline (Fig. 5D-5F) produced roughly 2 fold fewer GM, BFU, and GEMM colonies than the wildtype R1 control as described previously (Fig. 5D-5F) (Hidaka et al., 1999). Forced expression of Etv2 or Flk1 from EB D2-4 did not increase the number of any colony types (Fig. 5D-5F). However, forced expression of Etv2 or Flk1 from EB D4-6 completely restored or increased the number of GM (Fig. 5D), BFU (Fig. 5E), and GEMM (Fig. 5F) colonies observed. Therefore, forced expression of Etv2 in the Flk1 mutant cells from EB D4-6 rescues the phenotype of the Flk1 mutant equally as well as Flk1 itself.

Discussion

It is interesting to note that Etv2 is capable of rescuing endothelial and hematopoietic progenitor population (CD41⁺/CD45^-) of both the Flk1 and Etv2 mutant ESCs when induced from either EB D2-4 or EB D4-6 using flow cytometric analysis. However, at EB D6, forced expression of Etv2 in Etv2 mutant EBs rescues the hematopoietic compartment (CD41⁺/CD45⁺) cells only when expressed from EB D2-4. In contrast, in methylcellulose assays, we observed a much more pronounced increase of GM colonies when Etv2 is overexpressed from EB D4-6 than from EB D2-4 and only see rescue of BFU and GEMM colonies when Etv2 is overexpressed from EB D4-6. These apparently discordant observations can be explained in light of the notion that hematopoietic progenitors, but not hematopoietic cells, form colonies in methylcellulose culture. Thus, the cells that were rescued from EB D2-4 have further differentiated to hematopoietic cells by D6, the time of analysis, and have retained limited progenitor capabilities compared to wildtype controls, whereas cells that were rescued from EB D4-6 are still progenitors at D6 and are capable of giving rise to GM, BFU, and GEMM colonies as efficiently, if not more efficiently than wildtype controls. This supports the notion that Etv2 expression is important for the initiation of differentiation.

Furthermore, the observation that Etv2 induction from EB D4-6 generates hematopoietic progenitors raises a possibility that either Etv2 is capable of inducing trans-differentiation of progenitors of one or more non-hematopoietic lineage to the hematopoietic lineage, or that in the Etv2 and Flk1 mutant EB cultures, there are still a significant number of mesodermal progenitors with hematopoietic potential at EB D4. Either of these possibilities would be unexpected because it has been shown that 1) at least a subset of Etv2-mutant progenitors in embryos and EBs trans-differentiate to the cardiac lineage in the absence of Etv2, suggesting that in the absence of Etv2, these progenitors will not just stall and retain their progenitor characteristics (Liu et al., 2012; Palencia-Desai et al., 2011; Rasmussen et al., 2011) and 2) cells with non-hematopoietic or endothelial potential are unaffected by forced (and persistent) expression of Etv2 (Hayashi et al., 2012). While it has been shown that some of the progenitors can differentiate to the cardiac lineage (Liu et al., 2012; Palencia-Desai et al., 2011; Rasmussen et al., 2011), it is unknown to what degree this occurs or if there are other differentiation pathways that these cells could follow. The limitations of the previous lineage tracing analyses in the Etv2 mutant and the lack of an effect when Etv2 is expressed in non-hematopoietic or vascular cells in vivo leads us to believe the second hypothesis is more likely that, in the Etv2 and Flk1 mutant EBs, the differentiation of a subset of hematopoietic and endothelial progenitors are simply arrested, and that a progenitor cell population is maintained without trans-differentiating down other pathways or undergoing cell death, such that, upon introduction of Etv2, some cells are still capable of resuming differentiation towards the hematopoietic and endothelial lineages. If this were not the case, we would expect that rescue could only occur within the narrow window that Etv2 functions during wildtype development.

In summary, we have shown that forced expression of Etv2 rescues both the Etv2 and the Flk1 mutant EB phenotype. In contrast, forced overexpression of Flk1 rescues the Flk1 mutant embryoid body phenotype but is unable to rescue the Etv2 mutant EB phenotype. These data further support the notion that Etv2 is genetically downstream of Flk1 and that, at least in early hematopoietic development (EB differentiation parallels primitive hematopoiesis), Etv2 induction by the Flk1 signaling cascade is the critical pathway regulating the initiation of hematopoiesis and vasculogenesis. Future studies are warranted to analyze the extent of which forced expression of Etv2 can rescue the Flk1 mutant phenotype in vivo.

Methods

Cell lines

Etv2 mutant and wildtype mouse embryonic stem (ES) cells were derived using standard techniques (Bryja et al., 2006; Nagy and Nichols, 2011) by harvesting blastocysts from Etv2 heterozygous female mice bred with Etv2 heterozygous males (Ferdous et al., 2009). Flk1 mutant cells were provided by William L. Stanford (Hidaka et al., 1999) and were compared to unmodified wildtype R1 mouse ES cells. HA-Etv2 cDNA or Flk1 cDNA were cloned into pSAM2-IRES-mCherry (Darabi et al., 2012), a lentiviral construct with a doxycycline inducible promoter and the internal ribosome entry site (IRES) sequence followed by mCherry at the 3′ end. Viral particles were prepared and cells were infected as previously described (Darabi et al., 2012). Infected cells were enriched by doxycycline induction followed by FACS for mCherry positive cells. The sorted cells were amplified by growing for three passages and the enrichment process was repeated twice. During the last enrichment, cells were sorted by FACS for mCherry and SSEA-APC (1:1000, eBiosciences 51-8813-71) expression to enrich for virally integrated, undifferentiated ES cells. Expression was confirmed by flow cytometric analysis of m-Cherry, qRT-PCR using FAM-labeled Gapdh (4352339E) or VIC-labeled Flk1 (mm00440085_m1) and Etv2 (mm01176581_g1) Taqman probes (Applied Biosystems) and western blot using the following antibodies: rat-anti-HA serum (1:2000, Roche 3F10 Roche), rabbit anti-Flk1 serum (1:1000, Cell Signaling 55B11), mouse-anti-α-tubulin serum (1:500, Sigma B-5-1-2), anti-rabbit-HRP serum (1:2000, Pierce), anti-mouse-HRP serum (1:2000, eBiosciences), and anti-rat-HRP serum (1:2000, Pierce). The resulting ES cells will be available to the research community upon acceptance of this manuscript.

EB differentiation

ES cells were preplated for 20 minutes on gelatin coated flasks to remove the mouse embryonic fibroblasts (MEFs). Cells were resuspended at a concentration of 10,000 cells/ml in mouse EB differentiation medium (IMDM, 15% Fetal Bovine Serum, 2mM Glutamax, 1× Penicillin and Streptomycin, 450μM 1-thioglycerol (MTG), 200μg/ml Holo-Transferrin, and 50μg/ml Ascorbic acid). 10μl hanging drops were plated and cultured at 37°C with 5% CO₂ to induce differentiation. After 48 hours (at EB D2), hanging drops were washed from plates and further incubated in petri dishes on a rotating platform (80 rpm) at 37°C with 5% CO₂. Cells were fed with fresh medium with or without doxycycline every 48 hours, as previously described (Koyano-Nakagawa et al., 2012).

Flow Cytometry

ES cells or EBs were dissociated using 0.25% trypsin-EDTA and filtered through 50 μm cell strainers. Cells were incubated for 30 minutes on ice in the dark with combinations of the following antisera, SSEA-APC (eBiosciences 51-8813), CD31-PECy7 (eBiosciences 25-0311) and Flk1 APC (eBiosciences 17-5821), or CD41-PECy7 (eBiosciences 25-0311) and CD45-PE (BD Pharmingen 553081) and analyzed on a BD FACS Aria.

qPCR

D6 EBs were washed with PBS and resuspended in Trizol (Life Technologies). RNA was isolated using standard techniques and reverse transcribed using SuperScript Vilo cDNA Synthesis Kit (Life Technologies). Expression was analyzed in triplicate by qRT-PCR using FAM-labeled Gapdh (4352339E) or VIC-labeled vWF (Mm00550376_m1), Cdh5 (Mm00486938_m1), CD41 (Mm11439741_m1), Etv2 (Mm00468389_m1), Flk1 (Mm00440085_m1), Bry (Mm01318252_m1), Fgf5 (Mm 00438918_m1), Lmo2 (Mm00493153_m1), Scl (Mm01187033_m1), and Gata1 (Mm00484678_m1) Taqman probes (Life Technologies) and a VIC labeled Embryonic Globin (IDT 51863328) probe.

Methylcellulose assays

D6 EBs were dissociated and resuspended at 220,000cells/ml. Cells were further diluted 1:10 in 3ml methylcellulose (Methocult M3434, Stem Cell Technologies). 0.9ml of the resulting cells were plated in 35mm petri dishes in triplicate and allowed to differentiate for 7 days. Resulting colonies were counted and identified as GM (granulocyte and/or macrophage containing colonies), BFU (burst forming units or erythrocytes), and GEMM (GM, megakaryocyte and BFU mixed colonies) as previously described (Koyano-Nakagawa et al., 2012).

Supplementary Material

Supp Fig S1

Figure S1. Etv2 and Flk1 mutant EBs have altered transcript levels for germ layer markers and endothelial and hematopoietic markers. Etv2 wildtype (2.1) and mutant (A-H) and Flk1 wildtype (R1) and mutant (I-P) cell lines were differentiated to EB D6 and collected in 24 hour intervals for expression analysis of Etv2 (A,I), Flk1 (B, J), Bry (C, K), Fgf5 (D, L), Lmo2 (E,M), Scl (F, N), Gata1 (G, O), and Embryonic Globin (H, P). EB Day vs. expression relative to Gapdh levels are shown and error bars represent SEM, (n=3).