Hypoxia Influences the Vascular Expansion and Differentiation of Embryonic Stem Cell Cultures Through the Temporal Expression of Vascular Endothelial Growth Factor Receptors in an ARNT-Dependent Manner§


  • Yu Han,

    1. Case Cardiovascular Research Institute and University Hospitals Harrington-McLaughlin Heart and Vascular Institute, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    Search for more papers by this author
  • Shu-Zhen Kuang,

    1. Department of Pathology, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Alla Gomer,

    1. Case Cardiovascular Research Institute and University Hospitals Harrington-McLaughlin Heart and Vascular Institute, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    Search for more papers by this author
  • Diana L. Ramirez-Bergeron

    Corresponding author
    1. Case Cardiovascular Research Institute and University Hospitals Harrington-McLaughlin Heart and Vascular Institute, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    • 2103 Cornell Rd Rm. 4-532, Cleveland, OH 44106, USA
    Search for more papers by this author
    • Telephone: 216-368-2036; Fax: 215-368-0556

  • Author contributions: Y.H., S.K.: Collection and assembly of data, data analysis and interpretation; A.G.: Collection of data; D.L.R.: Conception and design, financial support, collection and assembly of data, data analysis and interpretation, manuscript writing. Y.H. and S.-Z.K. contributed equally to this article.

  • First published online in STEM CELLS EXPRESS February 4, 2010.

  • §

    Disclosure of potential conflicts of interest is found at the end of this article.


Adaptive responses to low oxygen (O2) tension (hypoxia) are mediated by the heterodimeric transcription factor hypoxia inducible factor (HIF). When stabilized by hypoxia, bHLH-PAS α- and β- (HIF-1β or ARNT) HIF complex regulate the expression of multiple genes, including vascular endothelial growth factor (VEGF). To investigate the mechanism(s) through which hypoxia contributes to blood vessel development, we used embryonic stem cell (ESC) differentiation cultures that develop into embryoid bodies (EBs) mimicking early embryonic development. Significantly, low O2 levels promote vascular development and maturation in wild-type (WT) ESC cultures measured by an increase in the numbers of CD31+ endothelial cells (ECs) and sprouting angiogenic EBs, but refractory in Arnt−/− and Vegf−/− ESC cultures. Thus, we propose that hypoxia promotes the production of ECs and contributes to the development and maturation of vessels. Our findings further demonstrate that hypoxia alters the temporal expression of VEGF receptors Flk-1 (VEGFR-2) and the membrane and soluble forms of the antagonistic receptor Flt-1 (VEGFR-1). Moreover, these receptors are distinctly expressed in differentiating Arnt−/− and Vegf−/− EBs. These results support existing models in which VEGF signaling is tightly regulated during specific biologic events, but also provide important novel evidence that, in response to physiologic hypoxia, HIF mediates a distinct stoichiometric pattern of VEGF receptors throughout EB differentiation analogous to the formation of vascular networks during embryogenesis. STEM CELLS 2010;28:799–809


Mammalian embryonic development unfolds under conditions of limited oxygen (O2) levels and poor nutrient delivery, and to survive, embryos must adapt to these environmental constrains. Ultimately, the early formation of the vascular system is essential to the embryo as it provides factors to promote further growth. Vascular development is believed to originate with the emergence of progenitor cells (hemangioblasts) that further differentiate into angioblasts and form a primitive vascular network through the process of vasculogenesis [1]. In subsequent angiogenic mechanisms, the cells undergo remodeling that includes pruning and sprouting, recruitment of supporting cells, and production of extracellular matrix proteins.

The hypoxia inducible factor (HIF-1) is a transcription factor that is regulated by changes in O2 tension and is responsible for upregulating multiple genes, including those involved in glycolysis, cell proliferation and survival, erythropoiesis, and angiogenesis (reviewed in Semenza [2]). HIF-1 is a ubiquitously expressed heterodimeric transcription factor comprised of members of the basic helix-loop-helix (bHLH)-Per-Arnt-Sim (PAS) family of proteins. In the presence of O2, prolyl hydroxylases target the HIF-α (HIF-1α, HIF-2α) subunits for rapid proteosomal degradation. In contrast, as oxygen becomes limiting, HIF-α subunits stabilize, heterodimerize with constitutively expressed ARNT (Aryl hydrocarbon Receptor Nuclear Translocator, ARNT, HIF-1β), bind to and transactivate multiple gene promoters containing hypoxia-response elements (HREs). Not surprisingly, HIF-deficient mice die in utero with hematopoietic, vascular, placental, and cardiac defects reviewed in Simon et al. [3].

As ARNT is the obligate HIF- β binding partner responsible for the regulation of hypoxia-specific gene expression in most cell types, the use of genetic models that target deletion of ARNT serve to test the consequences of HIF transcriptional inactivity in response to hypoxia. In previous analyses, we determined that the expression of ARNT in early differentiating ES cell cultures enhances the expression of vascular endothelial growth factor (VEGF), numbers of Flk-1+ (VEGF-R2, KDR) cells, and the formation of hemangioblast precursor cells in response to hypoxia [4]. Furthermore, using mouse embryonic para-aortic splanchnopleural explant cultures, we demonstrated that ARNT is necessary for proper hematopoietic, vasculogenic, and angiogenic intraembryonic processes [5]. However, the precise mechanistic role of HIF's downstream effects in response to hypoxic embryonic environment during the emergence and maturation of functional vessels remains unclear.

The aim of this study was to discern the effects of ARNT and hypoxia in the development and maturation of the vasculature. To overcome inherent difficulties in analyzing the biologic and molecular consequences to in vivo changes in O2 tension during embryogenesis, we employed mouse embryonic stem cells (ESCs). ESCs can be differentiated into embryoid bodies (EBs) resembling early embryogenesis and facilitate the analysis of reduced O2 levels in the programming of particular developmental pathways. EBs can mimic embryonic vascular development whereby, around day 3 of EB differentiation, early vasculogenesis is evident by the presence of hemangioblast precursors expressing Brachyury transcription factor and Flk-1 receptor [6–8]. The EBs further mature and express endothelial cell (EC) markers [9]. As during embryogenesis, the vessels in EBs undergo vascular remodeling beginning at day 6. When replated in 3-Dimensional (3D) collagen type I gel matrix, angiogenic potential can be characterized and quantified by the outgrowth and sprouting of ECs [10, 11].

In this study, we examine the influence of hypoxia on the angiogenic potential of differentiating ESC cultures. Arnt−/− ESCs cells provide an important genetic tool to examine the requirements for HIF transcriptional activity in vessel outgrowth and sprouting. In addition, as VEGF is believed to play a critical role in EC proliferation, differentiation, and sprouting, we examined the direct role of hypoxia independent of VEGF using Vegf−/− ES cells [12]. Our data indicate that hypoxia promotes angiogenesis in a VEGF independent manner and that HIF transcriptional activity is important for this process. We show distinct requirements for VEGF in that, whereas Vegf−/− ES cell cultures are refractory to hematopoietic defects, they display angiogenic deficiencies similar to Arnt−/− cultures [13]. Also, our results demonstrate that hypoxia temporally regulates the expression of Flk-1 and the antagonistic receptor Flt-1 (VEGFR-1) in the developing EB. We conclude that hypoxia promotes vascular development in part by regulating VEGF signaling through multiple mechanisms linked to the temporal alteration in the expression of VEGF receptors.


ESC Culture and Differentiation

The generation and maintenance of Arnt−/−, Vegf+/−, and Vegf−/− ES cells have been previously described [14–16]. ESCs were then differentiated into suspension EBs as described [7, 17]. Collagen differentiation of EBs has also been described [18].

To examine vascular differentiation, EBs were differentiated in methylcellulose (M3131, SCT) as previously described with a few modifications [6]. Briefly, ES cells were trypsinized, washed, and resuspended in 10% IMDM. Cells ranging from 1,000 to 5,000 in number were mixed in 1% methylcellulose containing 15% fetal bovine serum (FBS), stem cell technologies (SCT), (Vancouver, Canada, http://www.stemcelltechnologies.com), 450 μM monothioglycerol (MTG) (Sigma Life Science), 1% L-glutamine, 1% penicillin-streptomycin, and various concentrations of basic fibroblast growth factor (bFGF) or VEGF (Pharmingen, BD Biosciences, San Diego CA http://www.bdbiosciences.com) in a final volume of 1.5 ml and incubated under either 21% or 3% O2 for 7 to 14 days. On day 6, cultures were supplemented with an additional 1 ml of the methylcellulose mixture.

To quantify the angiogenesis, the differentiated EBs were replated in collagen. Briefly, after dissolving the methylcellulose and washing in phosphate buffered solution (PBS) at 37°C, EBs were then enumerated and triplicate cultures containing about 50 to 100 EBs were mixed in 1.25 ml of 1.2 mg/ml of type I collagen (BD Biosciences, Bedford, MA) containing 15% FBS (Gemini Bioproducts West Sacramento, CA http://www.gembio.com), 450 μm MTG, 5 ng/ml VEGF, 25 ng/ml bFGF, and 10 ug/ml insulin in Iscove's Modified Dulbecco's Medium (IMDM) (Invitrogen, Carlsbad, CA, http://www.invitrogen.com). Finally, 1N Sodium Hydroxyde (1N NaOH), was added to polymerize the cultures in 35-mm bacterial dishes ((SCT), Vancouver, Canada http://www.stemcelltechnologies.com). After 4 days, EBs were individually scored for the extent of angiogenic outgrowth and sprouting and data were assessed using Student's t test.

For hematopoietic progenitor cell analysis, ESCs were differentiated as previously described [15]. D9 EBs were disassociated with collagenase (SCT) and 15,000 were replated in M3434 methylcellulose media (SCT). Colony forming Cell (CFC) were scored after 6 days of culture.

Gene Expression Analysis

RNA was isolated by the trizol method (Invitrogen). After treatment with DnaseI (Invitrogen), reverse transcription was performed with Superscript-II reverse transcriptase (Invitrogen) using supplied oligo dT primers. Real-time detection polymerase chain reaction (RTD-PCR) was performed as previously described [19]. Reactions were carried out in a 15 μl volume with 2x Taqman Master Mix (Roche) using specific ROCHE universal probes (ROCHE Diagnostics, Indianapolis IN). Primer sequences are listed in supporting information Table 1.


Collagen gels were transferred onto glass slides and dehydrated with adsorbent cards (SCTs), fixed with 4% paraformaldehyde for 30 minutes, washed with PBS, and blocked in 1% bovine serum albumin (BSA), 0.1% Triton X-100 in PBS for 2 hours. Slides were incubated overnight at 4°C with 1:250 anti-CD31 (MEC13.3, Pharmingen) or for 1 hour at room temperature with 1:500 anti-SMA (Sigma Life Science) or 1ug/ml of anti-NG-2 in blocking buffer. After three washes, samples were incubated with secondary immunofluorescent conjugated antibodies (Pharmingen, or Invitrogen) and mounted with 4',6'-diamidino-2-phenylidole (DAPI) containing mounting media (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). Alternatively, biotinlylated antirat IgG antibody (Pharmingen) was used with a diaminobenzidine kit (Vector Laboratories, DRB1: DRB 1: Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com).

Flow Cytometry

EBs were dissociated in 0.25% collagenase to obtain single-cell suspension. Cells were then blocked with CD16/32 for 10 minutes before staining with 1:200 FITC-CD31, 1:50 PE-Flk-1 or 1:100 FITC-mFlt1 (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com) in PBS containing 1% BSA on ice. After washing, surface expression was detected by flow cytometry (BD LSR I) and analyzed using FloJo software (Tree Star, Ashland, OR, http://www.treestar.com).

Enzyme-Linked Immunosorbent Assay

Cell lysates or suspension media were used to quantify sFlt-1 as described by the manufacturer (Biosciences, R&D Systems, Minneapolis, MN, http://www.rndsystems.com).


Hypoxia Promotes EC Expansion and Vascular Gene Transcription

To evaluate the requirements of ARNT for angiogenesis in a system that emulates embryonic vascular development, we used mouse ESCs differentiation cultures. In the initial characterization of angiogenic events, aggregated ESCs from drop cultures were differentiated with reduced growth factor and serum conditions in suspension into cystic EBs (CEBs) or collagen type I matrix (Fig. 1A, 1B). Vessel maturation was observed in most of the developing WT EBs with the ability to differentiate into vessel-containing CEBs in suspension and into CD31+ ECs that emerge, sprout, lumenize, and have been shown to recruit supporting cells in 3D collagen cultures (Fig. 1B, upper panels). In contrast, under these limiting growth conditions, Arnt−/− ES cells failed to form CEBs and sprouting vessels consistent with a vascular deficiency phenotype (Fig. 1B, lower panels).

Figure 1.

Loss of Arnt disrupts proper vascular differentiation in ES differentiation cultures. (A,B): Arnt−/− embryonic stem cells (ESCs) stained for CD31 demonstrate a failure to generate vascular cystic embryoid bodies in 20-day EBs grown in suspension (A) or sprouting EBs differentiated in type I collagen (B) compared with WT controls. (C): Flow cytometry was used to assess CD31 levels from cells obtained by dissociating EBs from WT, Arnt−/−, or Vegf−/− ESCs differentiated in methylcellulose or suspension under normoxia or hypoxia (3% O2). (D): Transcript levels of angiogenic genes from cDNA isolated from WT or Arnt−/− day-7 and day-10 EBs were determined by real-time PCR, normalized against 18S ribosomal RNA, and expressed as mean ± SEM relative to the mean of D7 WT normoxia sample. Abbreviations: EBs, embryoid bodies; ES, embryonic stem; WT, wild-type.

We have previously shown that hypoxia promotes the early emergence of Flk-1+ cells associated with the early temporal presence of hemangioblast progenitor cells in suspension EB cultures [4]. We first examined whether chronic exposure of EB cultures to physiological levels of O2 affects the stability of HIF-1α. By Western blot, we observed that 3% O2 promotes the accumulation of HIF-1α protein. To assess the effects of hypoxia on the differentiation and proliferation of ECs in EB cultures, the surface expression of CD31 (platelet endothelial cell adhesion molecule, [PECAM]) EC cell marker was quantified by flow cytometry. CD31lo and CD31hi cells from methylcellulose and suspension EB cultures in the absence of exogenous VEGF were analyzed over 5 to 9 days of differentiation. Hypoxia promoted the expression of CD31 in both culture conditions, but interestingly, suspension cultures promoted the earlier production and an increase in numbers of CD31+ cells (Fig. 1C). Specifically, hypoxia induced the early expression of CD31lo, followed by the induction of CD31hi cell numbers in WT EB (Fig. 1C). In contrast, whereas the percentage of CD31lo cell numbers increased over time in either Arnt−/− EB culture conditions, the numbers of CD31hi cells were greatly reduced (7 to 12% for methylcellulose and 2 to 12% for suspension) compared with WT control (2 to 25% for methylcellulose and 11 to 40% for suspension) cultures (Fig. 1C). Moreover, hypoxia either doubled (18% for normoxia and 40% for hypoxia) or tripled (8% for normoxia and 24% for hypoxia) the peak production of CD31hi cells in WT suspension or methylcellulose cultures, respectively. Under these conditions, no significant numbers of Annexin V+ apoptotic cells were detected by flow cytometry, suggesting that the phenotypes of Arnt−/− EBs are not caused by decreased cell survival.

To begin to characterize the molecular events that result from hypoxia stimulated vascular development, the expression of various vascular markers was examined by real-time polymerase chain reaction (PCR). WT and Arnt−/− EBs were differentiated under normoxic and hypoxic (3% O2) conditions for 7 and 10 days, and complementary DNA (cDNA) was used to amplify transcripts, including a known HIF target adrenomedullin (ADM) whose levels are enhanced in hypoxic WT cultures (Fig. 1D). VEGF messenger RNA (mRNA) levels were comparable between normoxic and sustained hypoxic treatment. By day 7 of differentiation, WT cells displayed elevated Angiopoietin-1 [Ang-1 (3.0X)], Angiopoietin-2 [Ang-2 (10X)], Flk-1 [VEGFR-2, KDR (2.5X)], Tie-2 (4X) and Tie-1 (5X) transcript levels. In day-10 EBs, the expression of EPO (3.5X), Tie-2 (1.5X) and Tie-1 (1.8X) message levels in WT cultures were also induced by hypoxia. The overall transcript levels of angiogenic genes were greatly reduced in Arnt−/− EBs. Of interest, independent of ARNT, the Tie receptors were induced by hypoxia in D7 EBs, although significant differences were indistinguishable in D10 Arnt−/− EBs. Taken together, these results show that in an EB-differentiation system, hypoxia treatment promotes vascular development in an ARNT-dependent manner as measured by the expression of CD31 and select blood vessel gene profile.

Hypoxia Promotes the Emergence of Highly Angiogenic EBs

To quantitatively assess hypoxia's role in sprouting angiogenesis, an established angiogenesis assay was used to differentiate ESCs into 14-day EBs in methylcellulose containing 25 and 100 ng/ml of exogenous VEGF and bFGF, respectively [20]. Angiogenic competence was measured by replating the EBs in 3D collagen type I gel matrix for 4 days and evaluating the extent of vessel length and sprouting of individual EBs (Fig. 2A, upper panel). From these EBs, the outgrowth of sprouting endothelial tubes was further characterized by staining for SMA+ or NG2+ pericytes (Fig. 2A, middle panels). Following this protocol, the requirement of HIF transcriptional activity for angiogenic growth was assessed by contrasting WT and Arnt−/− ESCs differentiated in either normoxic (21% O2) or physiologic hypoxic (3% O2) conditions. The replated EBs were individually categorized into least (type A) to most (type E) angiogenic with types B and C (low angiogenic) having few emergent ECs, whereas types D and E (high angiogenic) displayed significant numbers of vessels infiltrating the collagen with profuse sprouting (Fig. 2A, lower panels). Since we observed no measureable differences between normoxia and hypoxia in 14-day EBs, the differentiation period was reduced to 10 days, revealing a significant increase of angiogenic WTs EB differentiated under hypoxic (59%) compared with normoxic (37%) conditions (data not shown, Fig. 2B). In contrast, Arnt−/− ES cultures contained reduced numbers of angiogenic EBs under either condition (Fig. 2B). These results show that hypoxia promotes more rapid vascular differentiation.

Figure 2.

The presence of ARNT in response to hypoxia influences angiogenic growth in differentiating ESC cultures independent of exogenous VEGF. (A): Experimental model used to differentiate EBs and quantify their angiogenic potential. Representative fields of vessels sprouts immunostained to reflect vessels containing CD31+ EC and α-SMA+ vascular smooth muscle cells. At least 50 individual EBs replated in 3D type I collagen are classified as low or high angiogenic EBs. Experiments were performed in triplicate and data are represented as percentage mean values ± SEM, *p < .05, **p < .005, *p < .001. (B–E): Quantitative analysis of angiogenesis from wild-type (WT) or Arnt−/− ES cells differentiated under normoxia or hypoxia (3% oxygen) and replated in type I collagen. (B): Percentage of high angiogenic EBs of cultures differentiated for 10 days in the presence of 25 ng/ml VEGF and 100 ng/ml bFGF. (C): Results from similar EB cultures differentiated for 7 days with growth factors reduced to 5 ng/ml VEGF and 25 ng/ml bFGF. (D, DRB2): Results from EB differentiated for 7 or 10 days in the absence of exogenous VEGF. DRB3 (E): DRB3 (E): See DRB2 and 3 representative micrographs of 10-day WT cultures grown under normoxia (a,b,c,d) or hypoxia (e,f,g,h) with (a,b,e,f) or without (c,d,g,h) exogenous VEGF and replated in type I collagen to assess angiogenic potential. (F): WT or Arnt−/− ES cells differentiated in methylcellulose with 25, 5, or 0 ng/ml VEGF for 7 or 10 days and then replated in type I collagen and individual EBs were scored for displaying low or high angiogenic outgrowth. Abbreviations: EBs, embryoid bodies; ES, embryonic stem; VEGF, vascular endothelial growth factor; WT, wild-type.

To further optimize our experimental system, the concentration of exogenous growth factors was lowered to 25 ng/ml bFGF with either 5 (Fig. 2C) or 0 (Fig. 2D, 2E) ng/ml of exogenous VEGF. As a result, replated day-7 WT EBs differentiated with 5 ng/ml displayed elevated numbers of highly angiogenic EBs (types D and E) induced by hypoxia compared with normoxia conditions (Fig. 2C). In the absence of exogenous VEGF, hypoxia significantly induced numbers of highly angiogenic EBs in WT EB cultures that were further expanded by day 10 of differentiation (Fig. 2D). Phenotypically, no discernable differences were observed between control and VEGF-treated WT EBs differentiated under 3% O2 (Fig. 2E). In contrast, in cultures treated with 5 ng/ml VEGF, Arnt−/− EBs failed to generate highly angiogenic EBs or to demonstrate significant hypoxic effects (Fig. 2C). Whereas in the absence of VEGF, Arnt−/− day 7 EBs in response to hypoxia, but the numbers of low angiogenic EBs in response to hypoxia, but the numbers of highly angiogenic EBs were significantly reduced in both day-7 and day-10 cultures (Fig. 2D). Thus, these findings demonstrated that chronic hypoxia promotes angiogenic sprouting in an ARNT- dependent manner.

Exogenous VEGF Is Not Sufficient to Rescue the Angiogenic Defects in Arnt−/− EBs

It has been debated whether differing degrees of VEGF signaling affect specific aspects of vessel development [21–23]. Since VEGF is significantly regulated by HIF, experiments were undertaken to examine the potential for VEGF to rescue and promote angiogenesis in Arnt−/− EB cultures and to test whether hypoxia alone promotes vascular differentiation. Day-7 and day-10 EBs cultures were supplemented with 25, 5, or 0 ng/ml VEGF. In day-7 EBs, hypoxia promoted the differentiation of angiogenic EBs (EB types C, D, and E) in WT cultures in the absence of exogenous VEGF (Fig. 2F). Furthermore, by day 10 of differentiation, whereas overall numbers of angiogenic EBs were reduced in WT cultures with no VEGF, hypoxia significantly increased (16%) highly angiogenic EBs compared with control normoxic conditions (6%) (Fig. 2F). Titration experiments further revealed that an addition of 25 ng/ml of exogenous VEGF was not sufficient to promote further angiogenesis in Arnt−/− EB cultures. Surprisingly, the percentage of highly angiogenic EBs in Arnt−/− cultures never exceeded 10%. Therefore, since VEGF did not rescue Arnt−/− EBs, we conclude that dysregulation of other important vascular growth factors, including angiopoietin-1 and angiopoietin-2, might contribute to the vascular phenotype as a result of HIF deficiency (Fig. 1C).

Hypoxia Promotes Angiogenic Outgrowth and Sprouting in Collagen

To test whether hypoxia could affect the extent of outgrowth and sprouting of vessels, replated EBs were further exposed to two differing O2 tensions (Fig. 3A). Exposure of Arnt−/− EBs to hypoxia during their replating into collagen did not promote further sprouting (Fig. 3B). In contrast, hypoxia promoted angiogenic outgrowth from differentiated WT EBs in the collagen matrix, the sprouting phase of the experiment (Fig. 3B). Furthermore, hypoxic stimulation during differentiation and vascular growth stages resulted in an enriched percentage of highly angiogenic WT EBs. These results suggest that 1) hypoxia stimulates vascular differentiation of EBs by enhancing their angiogenic potential, and 2) hypoxia promotes angiogenesis by increasing the numbers of EBs eliciting vessel outgrowth and sprouting.

Figure 3.

Hypoxia promotes vascular differentiation and angiogenic sprouting in an ARNT-dependent manner. (A): Experimental model outlining the normoxic or hypoxic (3% oxygen) conditions used for the differentiation of embryoid bodies (EBs) in methylcellulose (in the absence of exogenous vascular endothelial growth factor) and during vascular outgrowth in type I collagen. (B): Individual EBs were scored for angiogenic outgrowth for each condition. Data represent mean ± SEM from triplicate experiments, **p < .01; ***p < .001. Abbreviations: Lo, low; Hi, high.

Hypoxia Promotes EB Vascular Differentiation Independent of VEGF

To assess a requirement for VEGF in hypoxia-induced EB vascular differentiation, soluble Fc-Flt1 was used in EB methylcellulose cultures to compete with endogenous VEGF receptors, thereby inhibiting VEGF activity [24]. The addition of 1 μg/ml Fc-Flt1 (solid blue line) during normoxic differentiation of WT EBs substantially reduced the outgrowth of angiogenic EBs relative to cultures treated with 0.25 μg/ml (stippled blue line) of the inhibitor (Fig. 4A). The numbers of angiogenic EBs were also reduced in hypoxic cultures treated with 1 μg/ml (solid red line) compared with 0. 25 μg/ml (stippled red line) of Fc-Flt1. Yet, hypoxia protected the EBs from the antagonizing effects of Fc-Flt1 as the percentage of angiogenic EBs (types C, and/or D, in particular) remained significantly higher compared with normoxic conditions (Fig. 4A).

Figure 4.

Hypoxia promotes vascular differentiation independent of vascular endothelial growth factor (VEGF). (A–C): Inhibiting VEGF is not sufficient to block hypoxia's angiogenic effects. Inhibitors were added during the differentiation of embryoid bodies (EBs) for 7 or 10 days in methylcellulose EBs exposed to normoxic or hypoxic (3% oxygen) conditions and at least 50 EBs per treatment were then replated in type I collagen and EBs were individually scored for angiogenic outgrowth. (A): Fc-Flt1 was used at 1 and 0.25 μg/ml, and results are represented as percentage of each EB angiogenic type from total numbers of EBs per treatment. (B,C): WT or Arnt−/− EBs were differentiated under normoxic or hypoxic conditions with and without with 50 μM SU5416 and replated in type I collagen. (B): Micrographs of representative cultures demonstrate that WT EBs preserve some angiogenic EBs. (C): Quantification of low and high angiogenic EBs in response to SU5416 treatment. Data represent means from triplicate samples ± SEM. (D,E): Vegf−/− ESCs contain all hematopoietic colony-forming potential but are angiogenic deficient. (D): Hematopoietic progenitor activity of day 9 DRB5 Vegf−/−, DRB 6 Vegf−/−, and WT EB-derived cells. One 1500 cells were replated in triplicate and individual colonies were scored 6 to 7 days after replating. Error bars represent SEM. (E): Quantitative analysis of angiogenesis from WT, Arnt−/−, or DRB 7 Vegf−/− ES cells differentiated for 7 days under normoxia or hypoxia (3% O2) in the absence of exogenous VEGF. More than 50 EBs replated in type I collagen in triplicate and individual colonies were scored for angiogenic sprouting. Abbreviations: CFU, colony forming units; WT, wild-type.

SU5416, a potent and selective inhibitor of Flk-1 receptor tyrosine kinase activity, was also employed to downregulate VEGF signaling [24]. While there was no consequence to the numbers of low angiogenic EBs by hypoxia as a result of SU5416 treatment, a statistically significant reduction of highly angiogenic EBs from day-10 WT hypoxic cultures was observed (Fig. 4B, 4C). In contrast, no vessel growth or sprouting was observed with the exclusive addition of SU5416 to the collagen matrix (data not shown). Taken together, extrinsic and intrinsic inhibition of VEGF reduced but did not inhibit vascular development in EB cultures in response to hypoxia. These observations further suggest that whereas intracellular VEGF signaling is required for EC outgrowth and sprouting of replated EBs, enhanced angiogenic differentiation by hypoxia is not exclusively mediated through VEGF.

Vegf−/− ES Cells Generate Hematopoietic Progenitor Colonies but Are Angiogenic Deficient

We previously determined that whereas Flk-1+ ES cells numbers present in day-3 Vegf−/− EB cultures can be induced by hypoxia, the numbers of hemangioblast progenitor colonies were limited compared with WT cultures [4]. Since VEGF-A plays a critical role during embryonic development, we tested Vegf−/− mESCs in differentiation conditions that promote either hematopoietic or vascular differentiation [14]. Using clonogenic hematopoietic assays, ES cells were differentiated into 9-day EBs, dissociated into single-cell suspension by collagenase treatment, and replated in methylcellulose with a cocktail of growth factors for 6 days. The hematopoietic potential from these EBs were evaluated by categorizing and quantifying hematopoietic progenitor colony-forming units (CFUs). Under these conditions, Vegf± and Vegf−/− EBs phenocopied WT control cultures, effectively generating all types of hematopoietic progenitor colonies (Fig. 4D). This contrasts with Arnt−/− ES cells which have been reported to ineffectively generate multiple hematopoietic colony types [13].

Vascular development was first assessed by quantifying CD31lo and CD31hi cell numbers derived from methylcellulose and suspension EB cultures (Fig. 1C). Hypoxia promoted the temporal upregulation of CD31hi expression in Vegf−/− EBs, albeit the numbers were distinctly reduced compared with WT cultures. Next, the angiogenic capacity of Vegf−/− EBs was analyzed by replating in 3D collagen gels. In the absence of exogenous VEGF, Vegf−/− and Arnt−/− ES cultures inadequately generated sprouting EBs, even in response to hypoxic stimulus (Fig. 4E). Collectively, these results suggested that EBs have specific requirements for VEGF during the differentiation of progenitor, blood, and vascular cells.

Hypoxia Influences the Temporal Expression of VEGF Receptors

Distinct VEGF receptors are functionally associated with VEGF-A in ECs whereby Flk-1 and Flt-1 are considered positive and negative regulators of VEGF signaling, respectively [25]. Using suspension cultures, we previously showed that Flk-1 levels are transiently induced in early differentiating day-2.5 to day-3 WT EBs (during the emergence of early progenitor cells, which we also measured as hemangioblast colonies, BL-CFCs) in response to hypoxia and that Arnt−/− EB cultures are Flk-1+ cell deficient [4]. In the present study, we examined the levels of VEGF receptors from angiogenic EB cultures during later time points when they are considered to be actively angiogenic [26] (Fig. 5A). Suspension EB cultures were also employed since they permit analyses of soluble molecules from conditioned media. Single cell suspensions obtained from collagenase-treated EBs were used to evaluate the surface expression of Flk-1 and mFlt-1 (Fig. 5B). In more mature suspension WT EBs, Flk-1 expression ranged from 2% at D5-7% by D9 of differentiation and with a maximum 6% percent of double positive (DP) Flk-1+mFlt-1+ cells. In contrast, single positive (SP) mFlt-1+ cell numbers were transiently induced by hypoxia in WT EBs with a peak level (57%) achieved by D7 of differentiation and tapering in D9 EBs. The numbers of SP Flk-1+ cells in Arnt−/− EB cultures peaked with 18% and 45% at day 5 of differentiation for normoxia and hypoxia, respectively. A high percentage of Flk1+ cells (SP and DP combined) were detected in Arnt−/− cultures (ranging from 46-17%) relative to WT cultures (<13%). In contrast, a delay in the emergence of SP or DP mFlt-1+ cells was observed, and levels remained below those seen for hypoxic WT EBs. (Fig. 5B, upper panels).

Figure 5.

Hypoxia regulates the temporal expression of vascular endothelial growth factor (VEGF) receptors. (A): Embryoid bodies (EBs) were differentiated in suspension or methylcellulose cultures in the absence of exogenous VEGF. Cultures were exposed to normoxia or hypoxia (3% oxygen) and analyzed for the surface expression of Flk-1 or mFlt-1 by flow cytometry (B,C) and secreted Flt-1 by enzyme-linked immunosorbent assay (D). (C): In WT cultures, hypoxia (solid red squares) alters the expression of each receptor compared to normoxic conditions (solid blue squares) and are further dysregulated in DRB8 Arnt−/− EB cultures. (D): Although distinct from each other, the concentration of sFlt-1 from protein cell lysates isolated from methylcellulose EB cultures and conditioned media from suspension EB cultures is increased by hypoxic treatment in both WT and Arnt−/− EB cultures. Abbreviations: EBs, embryoid bodies; WT, wild-type.

Hypoxia's effect on the expression of VEGF receptors from EBs differentiated in methylcellulose in the absence of exogenous VEGF was also evaluated. In a time-dependent manner, the expression of either DP Flk-1+mFlt-1+ cells or SP mFlt-1+ cells in WT cultures was measured. Consistently, hypoxia treatment increased the expression of VEGF receptors under these differentiation conditions. The percent number of cells for normoxia and hypoxia in WT cultures were respectively: day 5, SP Flk-1+ 3% and 17%; day 7, total mFlt-1+ 19.6% and 46%; and day 9 total number of cells expressing either Flk-1 and/or Flt-1 were 42% and 50% (Fig. 5B, lower panels). In contrast, a distinct expression pattern for VEGF receptors in Arnt−/− EB cultures was observed. First, the relative levels of SP Flk-1+cells were significantly higher in Arnt−/− than WT EB cultures in both normoxic and hypoxic day 5 cultures (42% and 26%, respectively) and maintained in day 7 cultures (22% and 23%, respectively). Second, mFlt-1+ cell numbers were reduced overall (3-20%) and no significant differences between normoxia and hypoxia were observed. In contrast, SP or DP Flk-1+ cell numbers were distinctly different between both culture conditions.

To examine the presence of sFlt-1, conditioned media from EB suspension cultures were tested by ELISA. While sFlt-1 levels were induced by hypoxia in a time-dependent manner in WT cultures, there were no significant differences in s-Flt1 protein levels between normoxic and hypoxic Arnt−/− EBs at day 5 or day 7, though concentrations were higher than supernatants from control WT normoxia cultures (Fig. 5D). We also analyzed sFlt-1 levels from protein cell lysates obtained from D5-9 EBs differentiated in methylcellulose. A similar trend of hypoxic induction of sFlt-1 levels in WT EBs was observed. While the concentrations of sFlt-1 in Arnt−/− EB cultures were increased over control WT normoxia cultures, the levels of cellular and supernatant sFlt-1 were inversely proportional in these mutant cultures. Surprisingly, more sFlt-1 was retained within the cells of Arnt−/− EBs when compared with WT cultures. Thus, it appears that while hypoxia enhances the expression of sFlt-1 in an Arnt -independent manner, Arnt may be responsible for the formation of a sFlt-1 gradient by affecting the expression of mFlt-1 and retention/release of sFlt1.

To examine the contribution of VEGF to hypoxia-regulated expression of VEGF receptors, Vegf−/− ES cells were differentiated in the absence of exogenous VEGF. While Flk-1+ cell numbers were relatively high at day 5, the overall numbers were progressively lower in hypoxia-treated Vegf−/− cultures (Fig. 6A). Moreover, mFlt-1 cell numbers increased in a time-dependent manner. Overall, the expression profile of VEGF receptors in Vegf−/− EBs appeared intermediate of Arnt−/− and WT cultures, and sFlt-1 levels, either from cell lysates or conditioned media, were increased in hypoxia-treated Vegf−/− EB cultures (Fig. 6B). Together, these results argue that HIF activity in response to physiologic hypoxia contributes to the stringent control of the VEGF signaling pathway by influencing the temporal expression of VEGF receptors independent of VEGF.

Figure 6.

Hypoxia c regulates the expression of VEGF-receptors independent of VEGF. (A): DRB 9 Vegf−/− embryonic stem cells (ESCs) were differentiated in suspension or methylcellulose in the absence of exogenous VEGF. Cultures were exposed to normoxia or hypoxia (3% O2) and dissociated cells were analyzed for the surface expression of Flk-1 or mFlt-1 by flow cytometry. (B): Concentrations of secreted Flt-1 from supernatant or protein lysates of Vegf−/− EB cultures were analyzed by enzyme-linked immunosorbent assay. Abbreviations: EBs, embryoid bodies.


We had previously described that hypoxia alters the kinetics and promotes the emergence of hemangioblast progenitor cells in early differentiating EBs [4]. Our present analyses demonstrate how HIF, in response to hypoxic cues, participates in the differentiation of the vasculature and angiogenic expansion using an ESC differentiation model that permits the quantification of vascular competence.

While VEGF is a critical transcriptional target of HIF, VEGF itself plays an important autocrine role in ECs during vascular development and survival [27]. Specific effects for VEGF have been extensively documented, including its ability to induce EC and reduce hematopoietic differentiation [28–30]. While VEGF is required for the formation of blood vessels, its necessity for vessel maintenance differs between embryogenesis and adult homeostasis [14, 21, 27]. We previously showed that early Vegf−/− EBs expressed significant levels of Flk-1 inducible by hypoxia, although they ineffectively generate hemangioblast colonies [4]. We now confirm reports that Vegf−/− derived EBs fail to organize vascular networks, and that while Vegf−/− EBs have altered levels of VEGF receptors, our analyses demonstrate that hypoxia additionally affects their expression [31, 32]. The results from our present study corroborate the concept that there are unique requirements for intrinsic VEGF, as Vegf−/− EBs successfully generate hematopoietic progenitor cell colonies but are poorly angiogenic.

While competition (Fc-Flt1) or inhibition (SU5416) of Flk-1 during the differentiation of WT ES cultures affected angiogenesis, hypoxic treatment was nonetheless able to promote vascular differentiation. One plausible explanation is that the experimental parameters do not permit adequate inhibitory effects. On the other hand, we observed that EBs treated with 1 μg/ml of Fc-Flt1 significantly inhibited vascular growth from normoxic cultures and that the addition of SU5416 inhibitor exclusively during the replating of EBs completely blocked outgrowth of vessels in WT cultures in both culture conditions. These results corroborate our previous interpretation that reduced PECAM+ vessels observed in Arnt−/− embryos may be caused by limited numbers of endothelial and hematopoietic precursors and that hypoxic induction of VEGF is required for further development and maturation [5].

VEGF receptors are known to regulate EC proliferation, migration, and branching. HIF has been demonstrated to induce the transcription of Flk-1 and Flt-1 [33–35]. Moreover, exposure of in vitro human EC cultures to chronic hypoxia resulted in the downregulation of Flk-1, thus attenuating VEGF signaling activity [36]. By analyzing the expression of these receptors, our present study demonstrates that hypoxia can influence the levels of Flk-1, membrane- and soluble-Flt1. Indeed it appears that in a physiologic hypoxic environment, HIF initially upregulates the expression of Flk-1 in day-2 and day-3 WT EBs, and then in a reciprocal manner, downregulates its expression while “secreted” and “cellular” sFlt-1 levels increase. A recent ESC study describing the effect of hypoxia on the expression of sFlt-1 emphasized that the balance between Flk-1 and sFlt-1 directs the differentiation of hemangioblasts into ECs and hematopoietic progenitor cells, respectively [37]. In Vegf−/− or Arnt−/− genetic ESCs systems, we determined that hypoxia alters the expression of these receptors independent of VEGF or HIF, respectively, although in a pattern clearly distinct from WT cells. Moreover, the overall stoichiometry of VEGF receptors is significantly delayed in Arnt−/− cultures relative to WT cultures, indicative of inadequate vascular differentiation.

The abundance of mFlt-1 ad sFlt-1 can vary physiologically, suggesting their importance during various vascular processes [23, 25, 26, 38–40]. In recent observations in an angiogenic model of tumorigenesis, the inhibition of HIF degradation in PHD2+/Tie2Cre mice promoted the “normalization” of ECs by upregulating sFlt1 and Ve-Cadherin, which respecify ECs to become quiescent [41]. Also, sFlt-1 isoform modulates a VEGF-concentration gradient by affecting ligand availability correlating with patterns of Flk-1 activation in EBs [42]. In a previous analysis, the amount of Flt-1 mRNA from EB derived Flk-1+ cells was upregulated when cells were further differentiated and may explain the greater requirement for VEGF during the late phase of EC differentiation [22]. Indeed, we observed that in WT EB vascular cultures, Flk-1+ cell numbers decreased relative to earlier differentiation phase of hemangioblast expansion, whereas the inhibiting Flt-1 receptor increased in a time-dependent manner. Specifically, hypoxia expanded the numbers of mFlt-1+ or DP cells, and as vascular remodeling proceeded, sFlt-1 levels were also enhanced. Thus, as a result of physiologic hypoxia, HIF influenced the critical stoichiometric pattern of VEGF dose responsiveness during specific vascular events (Fig. 7). A local gradient of sFlt-1 may function to direct the guidance of emerging and sprouting vessels consistent with the Arnt−/− mouse embryonic phenotype featuring congested vessel growth and reduced sprouting [5, 16, 43].

Figure 7.

Model illustrating that hypoxia influences the expression of VEGF receptors either promoting (Flk-1) or inhibiting (sFlt-1) vessel sprouting, thereby controlling the formation of vascular networks. Angiogenic defects in absence of Arnt or Vegf may result in part from the altered expression of these receptors. Abbreviations: VEGF, vascular endothelial growth factor.

ECs isolated from human hemangiomas (hemECs) are associated with increased GLUT-1 and VEGF, both important HIF transcriptional targets. Strikingly, the abundance of Flt-1 is significantly reduced in hemECs [44]. Since we observed opposing expression patterns for VEGF receptors during vascular differentiation of EBs, we suspect that the hemECs that arise exclusively in neonates may develop as a consequence of ineffective responses to physiologic hypoxia [45]. While we propose that a physiologic hypoxic environment during embryogenesis directs HIF to regulate the temporal expression of VEGF receptors, a major challenge is to clarify how these events are orchestrated during the various morphogenic events in vessel generation that include proliferation, migration, sprouting, remodeling, and survival of vascular cells contributing to our understanding of vascular pathologies.


The use of ESC differentiation cultures permitted us to demonstrate that hypoxia is able to promote the emergence of CD31+ ECs, hasten vascular development, and promote angiogenic outgrowth and sprouting. However, in the absence of ARNT, and thereby ineffective HIF transcriptional activity, vascular development is delayed and significantly reduced in Arnt−/− EBs. In our present model we consider that the transcriptional activity of HIF, in response to hypoxia, is important in effectively directing vascular development. In particular, in this study we report a novel mechanism of the ability of HIF to moderate the temporal expression of VEGF receptors known to impart either positive or negative effects on VEGF signaling, thus affecting the stoichiometric pattern of VEGF dose responsiveness critical to the formation of vascular networks. Further studies on the mechanisms of hypoxia-induced angiogenesis may define novel elements that control vascular development and illuminate new therapeutic approaches against pathologic angiogenesis. Moreover, these experiments impart an important experimental tactic to promote the expansion of ECs for further experimental analyses and ultimately regenerative cell therapies.


We would like to thank Dr. Andras Nagy for the Vegf+/− and Vegf−/− ES cells. We would also like to thank Aaron Proweller for critically reading this manuscript. This work was supported by the NIH (HL- 073,153 D.R.B).


The authors indicate no potential conflicts of interest.