The Endothelial Transcription Factor ERG Promotes Vascular Stability and Growth through Wnt/β-Catenin Signaling

1Angiogenesis is essential during embryogenesis and is a critical component of many diseases. Coordination of growth and stability signals is required for effective angiogenesis. Diseases such as cancer, diabetic retinopathy, and vascular malformations are associated with vascular instability, which causes increased permeability and edema, excessive and/or dysfunctional angiogenesis, and hemorrhage. New strategies that target the maturation of blood vessels and restore vascular integrity could therefore have important therapeutic implications.

2Multiple interactions at endothelial cell-cell junctions control vascular integrity. Crucial among these are the adhesion molecule vascular endothelial (VE)-cadherin and its intracellular partner β-catenin, an essential component of the canonical Wnt pathway (reviewed in ). β-catenin is a multifunctional protein that can act as a scaffold between VE- and N-cadherins and the actin cytoskeleton, and as a coregulator for the T cell factor (TCF)/lymphoid enhancer-binding factor transcription factor complex. β-catenin levels are controlled by phosphorylation through a cytoplasmic degradation complex (reviewed in ). In the presence of Wnt ligands, which bind to a receptor complex containing members of the Frizzled (Fzd) family, the degradation complex is inactivated; β-catenin is stabilized and translocates to the nucleus to promote transcription. In the vasculature, the Wnt/β-catenin pathway controls vascular stability through remodeling, junction assembly, and pericyte recruitment (reviewed in ).

3The ETS transcription factor family is implicated in vascular development and angiogenesis (reviewed in ). The ETS related gene (ERG), expressed throughout the life of the endothelium, regulates multiple pathways involved in vascular homeostasis and angiogenesis, such as monolayer integrity, endothelial permeability, and survival. Previous studies have indicated a role for ERG in vascular development and angiogenesis. Vijayaraj et al. reported that global deletion of a subset of endothelial ERG isoforms in mice results in defects in vascular and cardiac morphogenesis, causing embryonic lethality. The mechanisms through which ERG controls blood vessel formation are still unclear, and its therapeutic potential is unexplored.

4In this study, we use genetic lineage-specific mouse models and multiple in vitro models to show that ERG promotes vascular growth and stability, through control of the canonical Wnt/β-catenin pathway. Crucially, we demonstrate that overexpression of ERG in vivo enhances vascular endothelial growth factor (VEGF)-dependent angiogenesis and promotes stability of VEGF-induced new blood vessels.

5To investigate the role of endothelial ERG in vivo, we used a Cre/loxP strategy. The floxed allele of Erg was obtained by inserting two loxP sites flanking exon 6 (Figure S1A available online). Constitutive endothelial-specific deletion of Erg was achieved by breeding floxed Erg mice with mice expressing the Cre transgene under the control of the Tie2 promoter and enhancer. Homozygous deletion of endothelial Erg (Tie2Cre-Ergfl/fl; henceforth referred to as ErgcEC-KO) resulted in embryonic lethality between embryonic day (E)10.5 and E11.5, with no live offspring (Figure S1B). Analysis of the yolk sacs from ErgcEC-KO embryos at E10.5 showed a significant reduction in perfused large vessels, consistent with defects in yolk sac vascular remodeling (Figures 1A, 5E, and 5F). Between E10.5 and E11.5, some mutant embryos appeared pale with no evidence of vessel blood flow, whereas others displayed hemorrhages and an enlarged pericardial cavity, suggesting defective heart function (Figures S1C and S1D). ErgcEC-KO embryos were reduced in size compared to littermate controls (Figure 1B), with growth retardation clearly visible at E9.5 (Figure S1E). These results are in agreement with the phenotypes caused by global deletion of endothelial ERG isoforms. Endomucin staining of blood vessels in ErgcEC-KO embryos at E10.5 revealed an immature disorganized vascular plexus, with significant disruption of the large vessels in the cranial vasculature (Figure 1B, panels a and b), altered development of the hyaloid vessels of the eye (Figure S1F), and irregular blind ending vessels in the head microvasculature (Figure S1G). Mutant embryos also exhibited disorganization of the intersomitic vessels in the trunk (Figure 1B, panel c). These data confirm that ERG is required for vascular development in mice.

6To investigate the role of ERG in physiological and pathological postnatal angiogenesis, floxed Erg mice were bred with mice carrying tamoxifen-inducible Cre recombinase under the control of the Cdh5 promoter (Cdh5(PAC)-iCreERT2). Following tamoxifen administration, efficient Cre-recombinase deletion of Erg was confirmed by PCR in Ergfl/fl/Cdh5(PAC)-iCreERT2 mice (henceforth referred to as ErgiEC-KO) (Figure S1H). The reduction in ERG protein levels was demonstrated by western blotting and immunofluorescence microscopy (Figures S1I and S1J). In the retinal vasculature of littermate controls (Ergfl/fl), ERG was strongly expressed in endothelial cells (EC) from all regions of the vascular plexus, including tip and stalk cells, arteries, veins, and capillaries (Figure S1K), in line with previous studies. Deletion of endothelial Erg resulted in significant reduction of vascular coverage (Figure 1C) and density (Figure 1D) in the retinal plexus, and reduction in the number of vascular sprouts at the front (Figure 1E). Next, we investigated whether ERG is involved in pathological angiogenesis, using the B16F0 melanoma tumor model, which depends on angiogenesis for growth. At day 14, tumor size and microvessel density were significantly reduced in adult ErgiEC-KO mice compared to controls (Figures 1F–1H). These studies confirm that ERG is required for postnatal angiogenesis and show that endothelial ERG is involved in tumor angiogenesis and tumor growth.

7Co-staining for isolectin B4 and the basement matrix component collagen IV showed a greater number of empty collagen IV sleeves in the capillary plexus (Figure 2A) and at the angiogenic front (Figure S2A) in retinas from ErgiEC-KO mice compared to controls, indicating increased vessel regression. Pericyte recruitment, measured by staining with neuron-glial antigen 2 (NG2) and desmin, was significantly decreased along all vessels in the vascular plexus, including veins and arteries in ErgiEC-KO mice (Figures 2B, S2B, and S2C). Similar signs of decreased vessel stability were observed in tumors grown in ErgiEC-KO mice, with a marked increase in the number of empty collagen IV sleeves (Figure 2C) and a reduction in pericyte coverage of blood vessels (Figure 2D). These data suggest that ERG controls both physiological and pathological angiogenesis through pathways that promote vascular stability.

8In vitro studies have shown that ERG is essential to maintain the integrity of endothelial junctions, by driving expression of VE-cadherin. A marked reduction in VE-cadherin expression and junctional localization was also observed in the retinal vasculature of ErgiEC-KO mice (Figure 3A), demonstrating that loss of endothelial ERG leads to a disruption of cell-cell junctions in vivo. Isolated primary mouse lung EC from heterozygous Tie2Cre-Ergfl/+ mice (ErgcEC-het) (Figure S3A) showed an approximate 50% decrease in VE-cadherin expression (Figure 3B), as expected.

9At the junctions, VE-cadherin binds β-catenin, protecting it from degradation; this interaction is required for the control of junction stabilization. β-catenin plays a pivotal role in Wnt signaling; interestingly, transcription profiling in ERG-deficient human umbilical vein EC (HUVEC) identified several Wnt-related genes as candidate ERG targets (Figure S3B). We therefore speculated that ERG might regulate β-catenin and Wnt signaling in EC.

10Inhibition of ERG expression in HUVEC (Figures S3C and S3D) significantly reduced β-catenin junctional staining and protein expression (Figures 3C and 3D). ERG regulates β-catenin protein levels in confluent (Figure 3C) and subconfluent (Figure S3E) EC, suggesting that this pathway functions both in quiescent and angiogenic endothelium. However, β-catenin mRNA levels were unaffected by ERG inhibition in HUVEC (Figures 3E and S3F) or in primary mouse EC from ErgcEC-het mice (Figure 3H). The decrease in β-catenin protein expression correlated with a decrease in canonical Wnt signaling: Wnt3a stimulation of β-catenin transcriptional activity was lost in ERG-deficient EC (Figure 3F). Moreover, expression levels of the Wnt target genes Cyclin D1, Axin-2, and TCF-1 were decreased in ERG-deficient HUVEC (Figure 3G) and primary mouse EC from ErgcEC-het mice (Figure 3H). Endothelial β-catenin signaling regulates blood brain barrier maintenance through concomitant activation of the tight junction molecule Claudin-3 and repression of plasmalemma vesicle-associated protein (PLVAP). In line with these findings, Claudin-3 expression was significantly downregulated, while PLVAP was strongly upregulated in ErgiEC-KO mouse brains (Figure 3I). Together, these results indicate that ERG regulates canonical Wnt/β-catenin signaling in both human and mouse EC.

11To confirm the relationship between ERG and β-catenin pathways, we used gene set enrichment analysis (GSEA) to compare the data set from transcriptome profiling of ERG-deficient HUVEC with the data set from transcriptome analysis of human pulmonary artery EC (HPAEC) following β-catenin inhibition. GSEA showed a significant positive correlation between the genes regulated by ERG and β-catenin (Figure 3J, left). Interestingly, gene ontology analysis of shared genes identified a significant number of regulators of angiogenesis, cell adhesion, migration, and apoptosis (Figure 3J, right). These results suggest a strong relationship between these two pathways.

12Since ERG inhibition decreases β-catenin protein, but not mRNA levels, we tested whether ERG regulates β-catenin degradation. β-catenin protein expression was restored in ERG-deficient EC in the presence of the proteosomal degradation inhibitor MG132 (Figure 4A). We investigated whether ERG controls β-catenin stability through VE-cadherin. In ERG-deficient HUVEC, GFP-tagged VE-cadherin overexpression (Figure S4A) partially restored junctional β-catenin protein levels (Figures 4B and 4C, lane 6). However, cellular fractionation studies showed that ERG also controls the nuclear pool of β-catenin (Figure 4C, lane 3), which was not corrected by VE-cadherin overexpression (Figure 4C, lane 5). This suggests that ERG controls β-catenin also through a Wnt signaling-dependent, VE-cadherin-independent pathway. Activation of Wnt signaling by lithium chloride (LiCl) inhibits GSK3β, and thus degradation of cytoplasmic β-catenin, allowing its nuclear translocation. LiCl was able to partially normalize β-catenin nuclear levels in ERG-deficient EC (Figure 4C, lane 7). Finally, combined Wnt signaling activation (through LiCl) and VE-cadherin overexpression were able to rescue both nuclear and junctional β-catenin pools in ERG-deficient EC (Figure 4C, lanes 9 and 10). These results demonstrate that the balance between VE-cadherin-dependent and Wnt signaling-dependent pathways, which modulates canonical Wnt/β-catenin signals in EC, is controlled by the transcription factor ERG.

13Wnt ligands bind to receptors of the Fzd family to inhibit the β-catenin degradation complex and activate Wnt signaling. LiCl treatment was able to partially stabilize β-catenin expression in ERG-deficient EC (Figure 4C, lanes 7 and 8). However, the upstream ligand Wnt3a was unable to do so (Figure S4B), suggesting a receptor-mediated defect upstream of the degradation complex. Wnt3a interacts with the Frizzled-4 (Fzd4) receptor, which is highly expressed in cultured EC. Fzd4 was identified as a putative ERG target by transcriptome analysis in HUVEC. We confirmed that Fzd4 mRNA (Figure 4D) and protein levels (Figure 4E) were significantly decreased in ERG-deficient HUVEC. Consistently, Fzd4 expression was decreased in mouse EC isolated from ErgcEC-het mice (Figure S4C).

14Comparative genomic analysis of the Fzd4 promoter revealed the presence of three highly conserved ERG DNA binding motifs in the 800 base pair (bp) region upstream of the Fzd4 transcription start site (Figures 4F and S4D). Analysis of the Encyclopedia of DNA Elements (ENCODE) chromatin immunoprecipitation sequencing (ChIP-seq) data for histone marks H3K4me1 and H3K27Ac and RNA polymerase II occupancy, markers of active promoters, show that the location of these marks correlates with the position of the ERG binding motifs (Figure 4F). ChIP-quantitative (q)PCR demonstrated that ERG interacts directly with the human Fzd4 promoter (Figure 4G); specificity of the interaction was confirmed in ERG-deficient EC (Figure 4G). ERG overexpression resulted in a 6-fold transactivation of a Fzd4 promoter luciferase construct in HUVEC (Figure 4H). Finally, Fzd4 overexpression in ERG-deficient EC was able to partially rescue Wnt3a activation of β-catenin transcriptional activity (Figure 4I). These data demonstrate that ERG controls transcription of the Fzd4 receptor in EC and point to a molecular mechanism for the VE-cadherin-independent control of Wnt signaling by ERG.

15Wnt/β-catenin signaling can promote EC proliferation and induce cell cycle progression through transcriptional activation of Cyclin D1. Therefore, we tested whether ERG may also control EC proliferation through Wnt signaling. As shown in Figure 5A, inhibition of ERG expression by siRNA decreased HUVEC proliferation; LiCl, which prevents β-catenin degradation, rescued proliferation of ERG-deficient HUVEC. These results indicate that ERG controls endothelial proliferation through the Wnt/β-catenin pathway. We have previously shown that ERG deficiency causes increased EC apoptosis. Combination of VE-cadherin overexpression and LiCl treatment could completely prevent cell death in ERG-deficient cells, indicating that ERG controls EC survival through Wnt/β-catenin signaling (Figure S5A).

16To test the functional relevance of Wnt signaling in ERG-dependent angiogenesis, we used an in vitro sprouting assay. ERG-deficient HUVEC formed markedly decreased numbers of significantly shorter sprouts (Figures 5B, panel b, 5C, and 5D). However, pretreatment of ERG-deficient cells with LiCl to inhibit β-catenin degradation was able to partially restore normal sprouting behavior of HUVEC (Figure 5B, panel d), by rescuing the number (Figure 5C) and length (Figure 5D) of the sprouts. These results suggest that Wnt/β-catenin signaling is required for ERG to control sprout formation during angiogenesis.

17To confirm that ERG controls angiogenesis and vascular development in a Wnt/β-catenin-dependent manner in vivo, we carried out a rescue experiment by pharmacological stabilization of Wnt/β-catenin signaling. Light microscopy examination of the yolk sacs from NaCl (control) and LiCl treated mice revealed a dramatic increase in perfused vessels in the yolk sacs of ErgcEC-KO mutants following LiCl treatment (Figure 5E). Endomucin staining revealed disrupted vessel morphology in the yolk sacs from NaCl-treated ErgcEC-KO embryos, with reduced microvasculature branching and decreased diameter of the larger vitelline vessels (Figures 5E and 5F). LiCl treatment of ErgcEC-KO mutants resulted in significant increase in vitelline vessel diameter and in the remodeling of the microvascular plexus (Figures 5E and 5F), in line with the increase in perfusion.

18Wnt targets CyclinD1 and Axin2, previously shown to be decreased in ERG-deficient endothelium (see Figures 3G and 3H), were significantly decreased in NaCl-treated ErgcEC-KO yolk sacs compared to controls (Figure S5B). LiCl-treatment of ErgcEC-KO yolk sacs normalized Cyclin D1 and Axin2 levels to those observed in LiCl-treated control embryos (Figure 5G). Interestingly, ERG targets VE-cadherin and Fzd4 were not normalized, in line with the direct transcriptional role of ERG in their regulation.

19These results demonstrate that endothelial ERG controls embryonic vascular development and angiogenesis through the Wnt/β-catenin signaling pathway.

20The data presented so far show that the transcription factor ERG controls angiogenesis through pathways mediating vascular stability and growth. The importance of the coordinated regulation of these pathways is highlighted by the variable and disappointing results of clinical trials for therapeutic angiogenesis in ischemic diseases, using the proangiogenic growth factor VEGF. VEGF has been shown to induce the formation of unstable and highly permeable vessels in vivo, giving rise to local edema and inefficient tissue perfusion. Therefore, we investigated the ability of ERG to stabilize new vessels induced by VEGF in vivo. C57BL/6 mice received a subcutaneous injection of Matrigel supplemented with VEGF-A165 and ERG (Ad.ERG) or Lacz (Ad.Lacz) adenovirus; basic fibroblast growth factor (bFGF), which can induce stable new vessels in this model, was used as control. Immunofluorescence staining for the adenovirus hexon coat protein showed localization of the adenovirus to endomucin-positive neovessels in the Matrigel plugs (Figure S6A). In addition, qPCR analysis confirmed significant expression of V5-tagged ERG in Matrigel samples treated with Ad.ERG compared to Ad.Lacz control (Figure S6B).

21To evaluate the stability of the new vessels, vascular permeability was measured using two different sized dextran tracers. In the presence of bFGF (Figure 6A, top panel), the lower molecular weight tetramethylrhodamine (TRITC)-dextran was fully contained within the vascular structures and colocalized with the larger molecular weight Fluorescein isothiocyanate (FITC)-tracer, confirming that bFGF induces the formation of stable, nonleaky vessels. Plugs containing VEGF-A165 and Ad.Lacz revealed a less organized vascular network with the smaller molecular weight dextran dispersed both inside and outside of the vessels (Figure 6A, middle panel, and Movie S1A). Interestingly, overexpression of ERG in the presence of VEGF-A165 resulted in reduced diffusion of the smaller molecular weight dextran (Figure 6A, bottom panel, and Movie S1B), indicating a more stable, less permeable vasculature. Quantification of the net amount of extravasated TRITC-dextran tracer shows that Ad.ERG caused an approximate 4-fold reduction in tracer extravasation in VEGF-A165-induced new vessels compared to Ad.Lacz control (Figure 6B). Consistently, quantification of FITC-dextran area showed that ERG overexpression in the presence of VEGF-A165 resulted in an increase in perfused vessels within the Matrigel plug after 7 and 10 days, compared to control (Figures 6C and S6C). ERG overexpression also resulted in a significant increase in the number of new vessels within the Matrigel plug; however this difference was observed only at the later time point (Figure 6D), suggesting that the increase in blood vessel number is secondary mainly to stabilization of VEGF-induced angiogenesis. These results confirm that ERG promotes stabilization of VEGF-induced angiogenesis in vivo.

22Pericyte recruitment is a critical step in vascular stability and maturation, and lack of pericytes has been shown to cause increased permeability. Since pericyte recruitment was decreased in two models of angiogenesis in the ErgiEC-KO mice (see Figure 2), we investigated whether ERG overexpression could increase the recruitment of vascular pericytes in the in vivo Matrigel plug model. Indeed, pericyte recruitment as measured by desmin staining was increased in the Ad.ERG-treated plugs compared to controls (Figures 6E and 6F). These results suggest that ERG may promote stabilization of angiogenesis also through control of pericyte recruitment.

23Over the last decade, major progress has been made in understanding the molecular mechanisms that regulate angiogenesis. However, the pathways that control vessel stability are less well characterized. In this study, we identify a transcriptional program regulated by ERG that controls vascular stability and growth through the Wnt/β-catenin pathway, in both a physiological and pathological context.

24We show that constitutive deletion of endothelial ERG in the mouse embryo causes embryonic lethality with severe vascular disruption. These observations are in line with a previous report where global deletion of a subset of endothelial ERG isoforms resulted in vascular defects and lethality between E10.5 and E11.5. Instead of a strategy based on a posteriori knowledge of ERG isoform expression, the Cre/LoxP system allowed us to delete all endothelial isoforms of Erg, by targeting exon 6, which encodes a region of the protein present in all isoforms. A previous transgenic model, where ERG’s function was disrupted by a mutation in the DNA binding ETS domain (ErgMld2/Mld2), caused embryonic lethality at a later stage (E13.5) and did not appear to display early vascular defects, suggesting that ERG’s functions in the vasculature are not exclusively mediated by its DNA binding activity.

25Using the inducible endothelial specific Cdh5(PAC)-iCreERT2 line, we show that ERG is required for angiogenesis in the developing retina of newborn mice and for tumor blood vessel growth in adult mice. ERG deficiency results in vessel regression and reduced pericyte recruitment, demonstrating that ERG controls vascular stability. Interestingly, ERG overexpression in the in vivo Matrigel plug model resulted in increased pericyte recruitment to vessels. This suggests that ERG may promote stabilization of angiogenesis in part through control of pericyte recruitment. A recent paper has described a role for ETS factors (including ERG) in arterial specification and reported increased ERG expression in arterial-derived EC in vitro. However, in the mouse retinal vasculature, ERG was strongly expressed in all EC with no detectable difference between arteries and veins.

26The in vivo developmental vascular defects in the ErgcEC-KO embryos are reminiscent of those associated with deletion of endothelial β-catenin. Endothelial deletion of ERG causes embryonic lethality earlier than the E12.5 reported for endothelial deletion of β-catenin. This study proposed that EC might not require β-catenin for early vascular development, but rather for maintenance of vascular integrity and vascular patterning at later stages. ERG’s regulation of other genes involved in earlier stages of vascular development, such as VE-cadherin, may be partly accountable for this difference in phenotypes. Constitutive endothelial deletion of ERG also causes diffuse hemorrhages and defects in vascular remodeling, similar to those observed in the β-catenin deficient embryos. In both lines, vitelline vessels of the yolk sac are significantly smaller in diameter; however, unlike ErgcEC-KO embryos, endothelial-specific loss of β-catenin does not affect vessel formation in the head, but causes hyperbranching of intersomitic vessels. Thus, embryonic mouse phenotypes of ERG versus β-catenin endothelial deletion show similarities, but not complete overlap, as expected, given the complex role of ERG as a transcriptional regulator of multiple vascular pathways. Crucially, the yolk sac vascular defects in the ErgcEC-KO and expression of Wnt targets were rescued by in vivo treatment with LiCl. Although we cannot completely rule out non-EC effects of LiCl, these experiments clearly demonstrate that ERG controls vascular development through Wnt signaling.

27Interestingly, similar angiogenic defects are observed in the retinas from ErgiEC-KO and from the reported endothelial-specific β-catenin and Fzd4 knockout mice. Whether ERG is implicated in human ocular diseases, including Norrie disease and familial exudative vitreoretinopathy, which are associated with Fzd4 and its ligand Norrin, remains to be established. In line with our data, a link between Fzd4 and ERG has been previously observed in prostate cancer. Interestingly, our results show that ERG deficiency results in about 50% reduction in Fzd4 protein, but completely abrogates Wnt luciferase reporter activity in response to Wnt3a. This suggests that ERG’s control of other nodes in this pathway, including repression of the Wnt inhibitor DACT1 and activation of the transcription factor TCF4, may be important.

28Canonical Wnt signaling promotes EC survival, junction stabilization, proliferation, and pericyte recruitment and is essential for vessel stability (; reviewed in ). In this study, we establish ERG as a regulator of canonical Wnt/β-catenin signaling, and therefore identify a connection between two key transcriptional regulators essential for EC function. We show that ERG controls cell survival, proliferation, angiogenesis, and vessel stability through β-catenin. Whether ERG controls pericyte recruitment via Wnt signaling remains to be elucidated; preliminary evidence suggests that ERG regulates expression of the junction molecule and β-catenin transcriptional target N-cadherin (data not shown), which plays a crucial role in pericyte attachment during vessel formation. Our data show that ERG controls Wnt/β-catenin levels and signaling through VE-cadherin-dependent and -independent pathways both in confluent, quiescent monolayers and in subconfluent, proliferating cells. The balance between VE-cadherin and Wnt-dependent signals controls β-catenin cellular localization and activity. It has been suggested that β-catenin could function to increase cell plasticity and sensitivity to extracellular signals. Transcriptional activity of ERG itself has been shown to be modulated by extracellular signals. Thus, we propose that in EC, ERG is required to maintain homeostatic levels of β-catenin protein, the output of which can be modulated according to the growth and survival signals it encounters, providing the balance between proliferation and stability required in a nascent blood vessel.

29Dysregulation of the Wnt/β-catenin signaling pathway is frequently observed in many types of cancer. Constitutive Wnt signaling activation caused by mutations in β-catenin or genes that control β-catenin stability has been associated with aberrant cell proliferation and subsequent cancer progression (reviewed in ). Wnt signaling has been shown to be a critical mediator of ERG-induced oncogenesis in several types of cancer, where aberrant ERG overexpression is a marker of aggressive malignancy and associated with increased proliferation. This is in contrast with its role in healthy endothelium, where ERG promotes homeostasis and stability. The reasons for this discrepancy are unknown and may be linked to the lack of balance between growth and survival signals, due to disrupted cell-cell signaling in malignant cells, thus driving the cells to a proliferative fate. Thus, strategies to control ERG’s activity in malignant cells through cell-cell adhesion signals might be worthy of investigation.

30Finally, in this study, we explore the potential for ERG in promoting vascular stability during VEGF-induced angiogenesis. Numerous studies have shown that the new vasculature induced by VEGF in vivo, to promote revascularization in ischemic diseases, can be dysfunctional due to vascular instability and increased permeability. Here, we show that overexpression of ERG can reduce permeability and promote VEGF-induced angiogenesis in vivo. Combined with the homeostatic and anti-inflammatory role of ERG, these results establish the ERG pathway as a potential target to promote vascular quiescence and stability.

31Detailed methods are available in the Supplemental Experimental Procedures.

32Generation of Erg floxed mice was carried out by genOway. LoxP sequences were inserted around exon 6. Deletion of this exon leads to a frameshift mutation resulting in a premature stop codon in exon 7. Ergfl/fl mice were crossed with the following Cre transgenic deleter lines: Cdh5(PAC)-CreERT2 and Tie2-Cre. All experiments were conducted in accordance with the Animals (Scientific Procedures) Act of 1986.

33Mice were administered Tamoxifen (50 μg per mouse; Sigma) by intraperitoneal injection (IP) at postnatal (P) day 1, P2 and P3. Retinas were collected at P6 and processed as described.

34Mouse melanoma B16F0 tumors were grown in tamoxifen-treated adult Ergfl/fl and ErgiEC-KO mice as described.

35LiCl or NaCl (400 mg/kg, dissolved in water) was injected IP into pregnant female mice at E8.5 and E9.5. Embryos were harvested at E10.5 and yolk sac vasculature was analyzed by light microscopy and immunostaining.

36C57BL/6 mice received a subcutaneous injection of Matrigel (BD Biosciences), as described. Matrigel was supplemented with 80 nanogram (ng)/ml bFGF (R&D Systems), or with 100 ng/ml murine VEGF-A165 (Peprotech) containing 109 plaque-forming unit adenovirus expressing either Lacz or ERG. After 3, 7, or 10 days, 100 μl of a 1:1 mixture of 10 mg/ml Dextran:FITC (2×106 MW) and Dextran:TRITC (4.4×104 MW) was injected intravenously 15 min prior to harvesting plugs. Plugs were imaged whole-mount using confocal microscopy. Volocity software (Perkin Elmer) was used to reconstruct 3D images of the vessels from serial Z-sections. The extent of TRITC-dextran tracer extravasation was quantified by subtracting the signal corresponding to the FITC-dextran tracer (intravascular) from the signal corresponding to the TRITC-dextran tracer (intravascular + extravascular).

37Primary mouse lung EC were isolated from control Ergfl/fl and ErgcEC-het mice as described. Rat APC-CD31, anti-ICAM-2, and anti-rat PE antibodies (all BD Biosciences) were used to assess the EC purity by flow cytometric analysis using a Cyan flow cytometer (Beckman Coulter).

38Primary HUVEC were harvested from umbilical cords. Human ERG expression was inhibited using either GeneBloc antisense oligonucleotides (Silence Therapeutics) or siRNA against ERG (Hs_ERG_7); both denoted as siERG in the text. Control GeneBloc antisense or AllStars Negative Control siRNA (QIAGEN) are denoted as siCtrl.

39ChIP was performed using ChIP-IT express (Active Motif) as previously described.

40For Fzd4 transactivation assays, a 1,010-bp region of the Fzd4 promoter (SwitchGear, Active Motif) was cloned into the pGL4 Luciferase Reporter Vector (Promega). TCF reporter constructs TOPFLASH and FOPFLASH were used to measure the transcriptional activity of β-catenin/TCF. In some experiments, cells were cotransfected with a pCMV6-Fzd4 expression construct (Origene). Reporter assays were performed using the Dual-Luciferase Reporter Assay System (Promega).

41Cell proliferation was determined in vitro using a BrdU proliferation ELISA kit (Roche) according to the manufacturer’s instructions.

42The 3D in vitro model of angiogenesis was performed as described previously.

43Values are presented as means ± SEM. Statistical significance was determined by using unpaired two-tailed Student’s t test. Differences were considered significant with a p value < 0.05.

44G.M.B. designed, carried out, and supervised in vivo and in vitro experiments, analyzed and interpreted results, and wrote the manuscript. A.V.S. designed and carried out in vitro experiments, analyzed, interpreted, and conceptualized results, and wrote the manuscript. N.D. designed and performed in vivo experiments, analyzed, and interpreted results. B.G. provided advice on bioinformatic analysis and interpretation and contributed to scientific discussion; Y.Y. performed bioinformatic analysis and interpretation; and L.R. and K.H.D. performed and supervised the tumor angiogenesis experiments, analyzed results, and contributed to scientific discussion. L.O.A. and S.T.K. performed experiments and analyzed results. I.M.A. provided advice on retina isolation and optimized ERG retinal staining; E.D. provided reagents and contributed to scientific discussion; J.C.M. contributed to scientific discussion; H.G. provided advice and contributed to scientific discussion; R.H.A. assisted in the design of the transgenic mice, provided reagents, advice, and contributed to scientific discussion; and A.M.R. provided funding, conceived, designed, and supervised the study, interpreted results, and wrote the manuscript.