Doxycycline-inducible HA-Etv2 overexpressing mouse embryonic stem cells (iHA-Etv2-mES) were generated in our laboratory and have been described previously (Koyano-Nakagawa et al., 2012). Similarly, to generate doxycycline-inducible Vezf1 mouse embryonic stem cells, we subcloned Vezf1 cDNA from pCMV-Vezf1-myc-flag (Origene) plasmid into the p2Lox plasmid. The authenticity of p2Lox-Vezf1-Myc-Flag plasmid was confirmed by sequencing using p2lox-FP:5′-CTTTCGACCTGCATCCATCT-3′ and p2lox-RP: 5′-TGG​TTT​GTC​CAA​ACT​CAT​CAA-3’.

Mouse ES cells were cultured and differentiated into embryoid bodies as previously described (Iacovino et al., 2014). EBs were prepared using the hanging drop technique and the appropriate mES cell lines, and then cultured in suspension beginning on day 2. For induction of HA-Etv2 or Vezf1-Flag separately or together, the respective mES cells were treated with 0.5 μg/mL doxycycline. Western blot analysis and qPCR were performed (as described below) to verify the over-expression of ETV2 and VEZF1.

To generate the Vezf1 knockout mES, the exon two of Vezf1 was deleted in V6.5 ES cells using the CRISPR/Cas9 system as previously described (Ran et al., 2013). The guide RNAs (gRNAs) targeting the 5′ and 3’ ends of Vezf1 exon 2 (Table 1) were cloned into the Cas9 expressing plasmid pSpCas9(BB)-2A-Puro (PX459) (Addgene, # 48139). V6.5 mES cells were transfected with both gRNA expression plasmids and 24 h post-transfection, cells were treated with 2 μg/mL puromycin for 36 h followed by seeding on mouse embryonic fibroblast (MEF) cells. Single colonies were picked after 5–7 days for expansion and genotyped using PCR. The wild-type (wt) cells yielded a 1754 bp product, whereas a homozygous deletion of the exon two yielded a 677 bp product. The PCR amplified products were cloned into the pCR2.1 TOPO plasmid using TOPO^® TA Cloning^® Kit (Invitrogen), and the biallelic deletion was verified using the Sanger sequencing technique.

Total RNA was extracted from mouse ES cells or embryoid bodies (harvested on different days of differentiation) using TRIzol Reagent (Invitrogen) and purified with the RNeasy Mini Kit (Qiagen). Following DNAse treatment, reverse transcription was performed with 500 ng of RNA using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quantitative Real-Time PCR analyses were performed using an ABI PRISM 7,900 sequence detection system and gene specific TaqMan probes. Probes used included: VIClabeled Gapdh: 4352339E, FAM labeled Etv2: mm01176581_g1, Vezf1: Mm00497288_m1, Flk1: mm00440099_m1, Pecam: mm01246167_m1, Tie2: mm01256892_m1, VEGFa: mm00437304_m1, Scl: mm01187033_m1, Cdh5: mm00486938_m1, Gata1: mm00484678_m1, Gata2: mm00492300_m1, c-kit: mm00445212_m1, Lmo2: mm00493153_m1, Fli1: mm00484410_m1, PU1: mm0048140_m1, CD41: mm00439741_m1, Runx1: mm00486762_m1, CD34: mm00519283_m1, and Endoglin: mm00468262_g1. Each transcript level was normalized to Gapdh expression.

In studies using inducible overexpression Vezf1-ES cells, the primer pair mVezf1-qRT-F1: 5′-ACA​TTA​GCT​GCC​CCT​CTC​AA-3′ and mVezf1-3′UTR-QRT-R1: 5′-ATT​TAC​TTT​AAC​CCA​ATT​TTG ACTTT-3′ were used to detect the endogenous Vezf1 gene specifically (not the overexpressed Vezf1). On the other hand, the transcript level of over-expressed Vezf1 (not the endogenous Vezf1 transcript) was assessed using mVezf1-qRT-F1 and myc-mVezf1-qRT-R1: 5′-TTGCTG CCAGATCCTCTTCT-3’. The Vezf1 expression was normalized to Gapdh expression using mGapdh-RT-FP: 5′-TGT​GTC​CGT​CGT​GGA​TCT​GA-3′ and mGapdh-RT-RP: 5′-CCT​GCT​TCA​CCA​CCT​TCT​TGA-3’.

To probe the interaction between HA-Etv2 and endogenous VEZF1, EBs (derived from dox-inducible HA-Etv2 mouse ES cells) were treated with doxycycline (0.5 μg/mL) from days 2–3 and then lysed using Co-IP lysis buffer (20 mM HEPES-KOH pH 7.9, 150 mM NaCl, 0.5% NP40, 1 mM EDTA, 10% Glycerol) supplemented with 1X protease inhibitor cocktail (Sigma), 1 mM PMSF (Sigma), 1 mM sodium orthovanadate, 10 mM sodium fluoride on day 4 for immunoprecipitation. Following centrifugation at 13.2 K rpm at 4°C, protein concentration was estimated in cell lysates using the BCA protein assay kit (Pierce). Immunoprecipitation and immunoblotting reactions were performed using Mouse IgG TrueBlot^® Protocol (Rockland, # 88–7,788–31). Immunoprecipitation reactions were performed using 1 mg of cell lysates and 2–5 μg of antibodies [Mouse anti-HA (12CA5) antibody from Roche, Mouse IgG TrueBlot antibody and ETV2 antibody (abcam #ab181847)] for the co-immunoprecipitation assays. Briefly, immunoprecipitated and input samples were separated by SDS-PAGE and transferred to activated PVDF membrane (Bio-Rad). After blocking with 5% non-fat milk for 1 h at room temperature, the membranes were incubated with specific primary antibody rat Anti-HA (Roche) at 1:1,000 dilution or VEZF1 antibody (Abcam, #ab50970) at 1:150 dilution, overnight at 4°C. After washing with 1X TBST, the membrane was incubated with rat or mouse HRP-conjugated secondary antibody (Santa Cruz Biotechnologies, United States of America, sc-2006 or #sc-2005) at 1:2,000 dilution) for 1 h. The detection was carried out using chemiluminescence kit (Pierce) and the signal was captured using the GelDoc Chemiluminescence Detection system (Bio-Rad).

For the pull down assays, GST-tagged constructs harboring full length ETV2 (1–335), N-terminal ETV2 (1–226) and C-terminal ETV2 (227–335) were generated using pGEX-4T1 vector as the backbone in our laboratory and have been previously described. The GST fusion proteins were induced using 0.5 mM isopropyl-D-thiogalactopyranoside (IPTG), at an optical density of 0.6, for 3 h at 25°C in BL21 cells. Cells were lysed using the B-PER protein extraction reagent (Pierce) containing 1X protease inhibitor cocktail, 1 mM PMSF, DNAse and lysozyme according to manufacturer’s instructions. GST fusion proteins were purified using Glutathione Sepharose 4B beads (GE Healthcare). The expression of GST fusion proteins in the input samples was confirmed by western blotting using anti-GST antibody (Bethyl laboratories, #A190-122A) and anti-rabbit secondary antibody (Santa Cruz Biotechnologies, United States of America, #sc-2004). Mouse VEZF1 was in vitro translated using TNT^® T7 Quick Coupled Transcription/Translation Systems (Promega) and pCMV-VEZF1-Flag-Myc (Origene) plasmid. Following the binding of GST fusion proteins to Glutathione Sepharose 4B beads, the beads were washed three times with cold 1X PBS and blocked overnight with BSA (2 mg/mL) to prevent the non-specific binding of VEZF1 with the beads. Beads bound with GST fusion proteins were incubated with 6 μL of in vitro translated VEZF1 protein for 2 h at 4°C with end-over-end rotation. After extensive washes with cold 1X PBS, proteins bound to the beads were eluted by the addition of elution buffer (10 mM glutathione, pH 8.0). Input and the eluted proteins were analyzed by western blotting using standard procedures. The myc and flag tagged VEZF1 was detected using mouse ant-Flag antibody (Sigma, #F1804 at 1:2,500) and anti-mouse secondary antibody (Santa Cruz Biotechnologies, United States of America, #sc-2005, at 1:2,000).

NIH3T3 and HEK-293T cells were maintained in DMEM containing high glucose, L-glutamine and sodium pyruvate (HyClone) supplemented with 10% (v/v) fetal bovine serum (HyClone), penicillin G and streptomycin sulfate (Invitrogen) in a humidified incubator at 37°C with 5% CO₂. Luciferase assays were performed using a pGL3T-Flt1 promoter reporter construct that was generated in our laboratory and has been previously described. For co-transfection and luciferase assays, NIH3T3 cells were grown in 24-well plates (70% confluence) and transfections were performed in triplicates with 250 ng/well of the pGL3T-Flt1 promoter reporter construct and different doses (50, 100, 200, 400 ng) of pcDNA-HA-Etv2 or pCMV-Vezf1-Flag-Myc (OriGene) expression plasmid using Lipofectamine 3,000 (Invitrogen) according to manufacturer’s instructions. Separately, the pGL3T-Flt1 construct was co-transfected into NIH3T3 cells with single dose of Etv2/Vezf1 and different doses of Vezf1/Etv2 expression plasmids. In all these co-transfection experiments, the total amount of DNA was made equal using pcDNA3 as balancing plasmids in different transfection mixtures. Cells were lysed and luciferase assays were performed using the Dual-Luciferase assay (Promega) following the directions outlined in the user manual.

Gel shift assays were performed as described previously (Sasi et al., 2014; Sonawane et al., 2014). Nuclear protein extracts from HEK-293T cells were prepared using the NE-PER^® Nuclear and Cytoplasmic Extraction Reagents kit (Pierce) and stored in aliquots at −80°C until use. Wild type and mutant oligonucleotides (Table 2) were synthesized at IDT. The single stranded oligos were labeled at the 3′-end by using Biotin-11-UTP (Thermo Fisher Scientific) and terminal transferase (NEB) following the manufacturer’s protocol. Following the biotinylation, the terminal transferase was heat-inactivated at 70°C and oligos were purified using chloroform:isoamyl alcohol (25:24) (Sigma). Equal amounts of each oligomer (labeled or unlabled) and its complementary strand (labeled or unlabled) were annealed. Electrophoretic Mobility Shift assays were performed by using the LightShift Chemiluminescent EMSA kit (Pierce) and the signal was captured using the GelDoc Chemiluminescence Detection system (Bio-Rad). In some EMSA experiments, the nuclear extracts were incubated with 2X-, 4X-, and 8X-fold molar excess of the unlabeled oligomers having either wild type or mutated sequences (Table 1) in the binding buffer prior to the addition of biotin-labeled probes. Supershift assays were carried out using anti-VEZF1 antibodies (#sc-365560x, Santa Cruz Biotechnologies, United States).

To perform the ChIP assay, embryoid bodies (EBs) were prepared from doxycylcine-inducible HA-Etv2 mouse ES cells, harvested at day 4.0 and dissociated with trypsin to obtain a single cell suspension. ChIP assays were performed as previously described (Sasi et al., 2014; Sonawane et al., 2014). Overnight immunoprecipiation reactions were performed using 5 µg of ChIP-grade antibodies against VEZF1 (#sc-365560x Santa Cruz Biotechnology) and mouse non-immune IgG as the negative control (Sigma). The immune complexes were captured using Rec protein G-Agarose beads (Invitrogen). The immune complexes were reverse cross-linked, treated with RNAse A/proteinase K, purified by PCR product purification kit (Promega) and stored at −20°C until used. End-point PCR as well as quantitative real-time PCR were carried out using immunoprecipitated chromatin as the template and the following mFlt1-specific primers: (forward: 5′-CTG​TCC​GGC​GAC​CCG-3′ and reverse 5′-TCC​CCA​GCC​CAC​TTC​CTA​C-3′). To demonstrate the specificity of the ChIP assay, Gapdh specific primer pair (forward: 5′-CCC​TTT​TCT​GCC​TTC​CTA​CC -3′ and reverse 5′-TGC​TGA​AGT​GCT​CCC​TAC​CT-3′) were designed and used to amplify a 176 bp promoter fragment without the VEZF1 binding domain. qPCR was performed using SsoAdvanced universal SYBR Green Supermix (Bio-Rad) and the fold enrichment of the Flt1 promoter domain in the immunoprecipitates was quantified.

All co-transfection experiments were performed at least three times and results were expressed as mean ± S.E. of triplicates from representative experiments. Statistical analyses were carried out using Student’s t-test or one-way ANOVA with Bonferroni’s multiple comparisons post-test, as appropriate, by using GraphPad Prism 5.

To evaluate the relationship between Etv2 and Vezf1 transcript expression patterns were examined during EB differentiation (derived from iHA-ETV2-mES cells), beginning on day 0 until day 8 (Figures 1A, B). As previously reported (Lee et al., 2008), Etv2 expression showed a transient expression pattern peaking at day 3.0 and subsequently downregulated to baseline levels (Figure 1B). On the other hand, Vezf1 expression remained similar from Day 0 to Day 2 and gradually increased from day 3 through day 6 EB differentiation (Figure 1A). Thus, Etv2 and Vezf1 had overlapping expression during EB mediated differentiation of ES cells on day 2–4.

Initially, VEZF1 was selected from a yeast two hybrid screen as a candidate to investigate its interaction with ETV2 because of its expression in the nucleus and its role in endothelial cell biology (Bruderer et al., 2013). Co-expression of Etv2 and Vezf1 in mES cells on days 2–4, further supported the hypothesis that ETV2 and VEZF1 could function via a protein–protein interaction mechanism. We performed co-immunoprecipitation (Co-IP) assay in differentiated EBs derived from doxycycline inducible iHA-Etv2-mES cell line which over expressed HA-tagged ETV2. Following Dox (0.5 μg/mL) treatment on day 2–3, the induction of Etv2 transcript and protein was confirmed by qPCR and western blot techniques (Supplementary Figures S1A–C). As positive controls, input samples showed HA-tagged ETV2 and VEZF1 in cell lysates (Figure 1C). Endogenous VEZF1 was successfully immunoprecipitated using an anti-HA antibody while the negative control, non-immune mouse IgG did not precipitate VEZF1 (WB: anti-VEZF1, Figure 1D), thus confirming the specificity of the interaction. Furthermore, we pulled down VEZF1 from D4 EB lysates using an antibody against the endogenous ETV2 (Figure 1E). Together these studies establish the interaction between endogenous ETV2 and VEZF1.

To confirm the endogenous interaction between ETV2 and VEZF1, we performed an in vitro pulldown assay using GST-tagged ETV2 constructs as bait proteins and in vitro translated myc-flag-tagged VEZF1 as prey. Expression and purification of GST tagged full length (1–335 amino acids) or deletional N-terminal ETV2 (1–226 amino acids) and C-terminal ETV2 (227–335 amino acids) were verified by western blot (Figures 1F, G). In vitro translated myc-flag-tagged VEZF1 was pulled down with the full length, N-termina, and C-terminal ETV2 constructs, but not with the GST alone as control bait (Figure 1H). Collectively, these results supported the conclusion that ETV2 can directly bind with VEZF1 through the N- and C-terminal domains.

The 0.7 kb Flt1 promoter is essential for endothelial cell formation (Morishita et al., 1995; Jin et al., 2009) and we have previously shown that ETV2 potently transactivated the Flt1 promoter (Koyano-Nakagawa et al., 2015). Therefore, in order to probe the functional consequences of ETV2 and VEZF1 interaction, we analyzed the Flt1 promoter for ETV2/VEZF1 binding motifs and identified four putative VEZF1 binding motifs (Figure 2A). Hence, we tested the effect of ETV2/VEZF1 over-expression on Flt1 promoter activity. Indeed, ETV2 and VEZF1 caused a dose-dependent increase in Flt1 promoter reported luciferase activity by ∼12.0 (p < 0.001) and ∼7.3-fold (p < 0.001), respectively (Figures 2B, C). Both ETV2 and VEZF1 were able to enhance the activation of the Flt1 promoter activity in a dose-dependent manner (Supplementary Figures S2A–B) when either of them were kept at a constant level. Next, we examined whether transcription of the Flt1 gene was regulated by these transcription factors in an additive or synergistic manner. Accordingly, NIH3T3 cells were co-transfected with Flt1 promoter plasmid and different concentrations of ETV2 and VEZF1. While ETV2 and VEZF1 enhanced the Flt1 promoter activity by ∼ 4-fold (p < 0.01) and ∼ 2.5-fold (p < 0.05), respectively, the combination of both caused a more pronounced increase (∼7.5-fold, p < 0.001) in promoter activity (Figure 2D), suggesting an additive effect of ETV2 and VEZF1 in Flt1 gene regulation.

Because the 0.7 kb Flt1 promoter domain governing the Flt1 gene transcription contained several potential binding sites for VEZF1 (Figure 2A), we tested the binding capacity of VEZF1 with synthetic double-stranded oligonucleotides (wild type/mutant) (Table 1) representing the Flt1 promoter segments using EMSAs. The biotin-labeled Flt1 oligonucleotide (Wt-L-Oligo3) yielded the specific protein/DNA complex (Figure 3A, lane 5) with HEK293 nuclear protein extracts, whereas the labeled mutant oligo did not yield any specific complex (Figure 3A, lane 3). Intensity of this shift was reduced in a dose dependent manner with addition of molar excesses of unlabeled wild-type oligo (Figure 3A, lanes 6–8), whereas the unlabeled mutated oligo did not show any competition effect (Figure 3A, lanes 9–11). Similarly, an EMSA with another Flt1 oligonucleotide (Wt-L-Oligo4) displayed formation of one specific complex (Figure 3B, lane 5) that was inhibited by unlabeled wild type oligonucleotide in a dose dependent manner (Figure 3B, lanes 6–8); the intensity of this specific band was not changed in the presence of unlabeled mutant oligonucleotide (Figure 3B, lanes 9–11). Incubation with VEZF1 antibody decreased the intensity of the specific bands using either Wt-L-Oligo3 or Wt-L-Oligo4 (Figures 3C, D), lane 4), respectively, suggesting the specificity of VEZF1 binding to the with Flt1 promoter oligoes.

We next validated VEZF1 binding to the endogenous Flt1 promoter in vivo by pursuing ChIP assays. Formaldehyde-cross-linked, sonicated chromatin from day 4 EBs (derived from iHA-Etv2 mESCs) were immunoprecipitated using a VEZF1 antibody. PCR amplification resulted in the detection of Flt1 promoter specific product from the VEZF1 pull-down (Figure 3E, lane 4), whereas the IgG control ChIP had no enrichment (Figure 3E, lane 3). PCR amplification of the Gapdh promoter region having no putative VEZF1 binding sites demonstrated the specificity of the VEZF1 antibody (Figure 3F) and the ChIP assay. Quantitative PCR analysis revealed a 4.0-fold enrichment of VEZF1 binding to the Flt1 promoter region over the IgG control (Figure 3G). These results established that VEZF1 displayed robust interaction with the chromatin segments harboring the Flt1 promoter domain.

The ES/EB system recapitulates the mechanisms that govern embryonic development (Keller, 1995). To establish the functional role of VEZF1 in hematoendothelial cell lineage development and differentiation, we generated Vezf1 ^−/− mouse ES cells (ko) using CRISPR/Cas9 technology (Figure 4A) and deleted exon 2. The ES cell clone with homozygous deletion of the exon two was verified by PCR (Figure 4B). We used the ES/EB differentiation system to further characterize the Vezf1 ^−/− ES cells. Western blot analysis of protein lysates from D5 and D6 EBs confirmed the absence of VEZF1 protein in the ko cells (Figure 4C). To assess whether the loss of VEZF1 leads to changes in expression levels of endothelial/hematopoetic markers, qPCR analysis was performed with total RNA isolated from D4, D6 EBs (wt and ko). The hematoendothelial markers (Flk1, Tie2 and CD31) were significantly downregulated in the Vezf1 ^−/− D4 EBs (Figure 4D), whereas early hematopoiesis markers Gata1, Lmo2, Runx1, Scl, and Flt1 were downregulated in Vezf1 ^−/− D6 EBs compared to the WT control EBs (Figure 4E).

Our data showed that lack of VEZF1 diminished the expression of hematopoietic and endothelial markers in the differentiating ES/EB system. To further examine these results, we engineered an ES cell line in which VEZF1 could be induced by doxycycline using a cassette exchange recombination strategy (Iacovino et al., 2014) (Figure 5A). Induction of VEZF1 expression was confirmed by western blot and qPCR techniques, respectively (Figures 5B, C). The expression of VEZF1 was induced in differentiating EBs from D2 to D4. Gene expression analysis of hematoendothelial markers using RNA from D4 EBs showed the upregulation of Flt1, Gata2, and Flk1 transcripts following VEZF1 overexpression compared to the control EBs, which further established the role of VEZF1 in hematoendothelial progenitor cells.