HIF-2α-Haploinsufficient Mice Have Blunted Retinal Neovascularization Due to Impaired Expression of a Proangiogenic Gene Battery

Abstract

purpose. To characterize the effect of HIF-2α haploinsufficiency on retinal neovascularization and angiogenic signaling in neonatal mice.

methods. Retinal samples were obtained from HIF-2α-haploinsufficient (Epas1 ^+/−) and wild-type (Epas1 ^+/+) neonatal mice subjected to an oxygen-induced retinopathy (OIR) protocol. Histologic and molecular studies were performed immediately, 12 hours, or 5 days after initiation of the hypoxia phase of the OIR protocol. Molecular profiling was performed in mouse brain endothelial cells maintained in normoxia or hypoxia. Transfection studies assessed the response of isolated promoter regions from proangiogenic genes to HIF-1α or -2α overexpression.

results. Epas1 ^+/− mice exhibited no significant differences in retinal vasculature during normal development but had reduced retinal neovascularization in an OIR protocol. Multiple proangiogenic factors were induced during the hypoxia phase in Epas1 ^+/+ OIR retinal samples, whereas Epas1 ^+/− OIR retinal samples had absent or blunted induction of these same factors. Several, but not all, proangiogenic factors were induced in mouse brain endothelial cells after hypoxia. In transfection assays, most proangiogenic promoter regions were preferentially activated by HIF-2α relative to HIF-1α.

conclusions. HIF-2α deficiency results in reduced neovascularization and blunted inducibility of multiple proangiogenic factors in the retinas of mice with OIR. The authors propose that HIF-2α is a master regulator of proangiogenic factors in retinal vascular endothelial cells, the predominant cell type of the retina in which HIF-2α is expressed. Future studies will address whether the molecular and functional roles for HIF-2α identified from these studies can be generalized to other pathophysiological states involving neovascularization.

Retinal vascular development in the mouse involves prenatal and postnatal developmental processes, ^{1 2 3 4} with spatial, temporal, and cell-specific requirements as defined by in vitro and in vivo models. ^{5 6} As the retina grows, astrocyte differentiation, endothelial cell recruitment, and pericyte association results in mature retinal vessel formation. ^{4 7 8} Proangiogenic factors essential for proper retinal vascular formation are produced by astrocytes, endothelial cells, and pericytes, ^{9 10 11 12} with environmental stresses such as physiological hypoxia serving as stage-specific developmental cues. ¹³ Abnormal environmental stresses during development perturb the otherwise carefully orchestrated retinal vascular developmental program. 

Pathophysiological hypoxia results in marked neovascularization in retinopathy of prematurity (ROP) and other eye diseases. ¹⁴ In the mouse ROP model, a paradigm for oxygen-induced retinopathy (OIR), ¹⁵ hyperoxia leads to prematurely and irreversibly growth arrested retinal vascular development, as well as the onset of several environmental stresses, including hypoxia on return to room air. The exact mechanisms underlying the impaired retinal vascular development during the hyperoxia phase ¹⁶ and the subsequent neovascular response after return to room air remain undetermined. ¹⁷ However, it is believed that the determinants of proliferative retinopathies are similar, but not identical, to those of normal vascular development. ¹⁸  

Retinal neovascularization in the mouse OIR model is probably driven largely by hypoxia, although other coexisting environmental stresses may also contribute. Hypoxia inducible factor (HIF) members are stress-responsive transcriptional regulators implicated in regulation of several proangiogenic growth factors and their receptors during normal vascular development. Therefore, HIF factors are logical candidates as regulators of neovascularization in proliferative retinopathies. HIF-1α expression correlates with expression of VEGF in a mouse ROP model. ¹⁹ However, a direct role for HIF-1α in normal retinal vascular development has not been demonstrated. Moreover, endothelial cell conditional HIF-1α knockout mice have no reported overt vascular abnormalities, ²⁰ suggesting global HIF-1α elimination has a developmental consequence involving multiple cell populations. 

HIF-2α, the second HIF-α member encoded by Epas1, ²¹ is expressed at high levels in vascular endothelial cells. However, the lack of ²¹ or minimally abnormal ²² vascular phenotype of Epas1 ^−/− embryos suggests that HIF-2α does not play a significant role in vasculogenesis. In contrast, a role for HIF-2α in retinal vascular development is much more evident. Epas1 ^−/− mice have markedly abnormal retinal vascular development. ²³ Molecular characterization reveals reduced expression of major antioxidant enzymes and erythropoietin (Epo), HIF-2α target genes. Exogenous Epo reverses several aspects of the Epas1 ^−/− state, but not the retinal abnormalities, ²⁴ suggesting that multiple HIF-2α target genes are dysregulated in Epas1 ^−/− retinas. 

A second HIF-2α-deficient model has also provided insight into the role of HIF-2α in retinal vascular function. ²⁵ HIF-2α knockdown mice, generated by disruption of the 5′ untranslated region of the Epas1 gene, have an 80% reduction in HIF-2α levels in the eye and have normal retinal vascular development, but impaired neovascularization in a mouse ROP protocol. The molecular abnormalities include impaired Epo expression. However, as with Epas1 ^−/− mice, repletion of Epo in HIF-2α knockdown mice had only a minor effect on the neovascular response. ²⁵  

Missense or nonsense mutations in HIF-2α may result in a haploinsufficient state with no overt effect under ambient conditions, but with significant abnormalities in the presence of increased environmental stress. Indeed, although Epas1 ^+/− mice have no overt abnormalities at baseline, these mice exhibit molecular as well as physiological abnormalities in response to systemic hypoxia. ²⁴ Thus, we reasoned HIF-2α haploinsufficiency would be likely to have demonstrable effects on neovascularization with systemic hypoxia. In this study, we address the effect of HIF-2α haploinsufficiency on in vivo retinal vascular form and function in the mouse ROP model, an OIR model that results in hypoxia-dependent neovascularization in the retina. 

Materials and Methods

Mice

All data were obtained from F1 progeny originating from crosses of hemizygous C57/BL6J (male) or 129S6/SvEvTac (female) HIF-2α mice maintained in a standard light–dark cycle and on a standard diet. Procedures were approved by the University of Texas Southwestern Medical Center (UTSWMC) Institutional Animal Care and Use Committee and performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

OIR Protocol

OIR was induced by placing postnatal day (P)7 pups with their dams in a sealed chamber containing 75% O₂ and 25% N₂, monitored with an oxygen sensor (Model AD300; Teledyne, City of Industry, CA). Mice remained in the chamber for 5 days (hyperoxic period, P7–P12) and were then housed in room air for another 5 days (hypoxic period, P12–P17). During the hyperoxic period, the mothers were provided with water and standard chow (4% fat) and exposed to room air for 6 hours each day. The pups received nutrition from their mothers in room air for 2 hours each day. Age-matched animals were maintained identically, except they were exposed to room air (21% O₂, 79% N₂) for the duration of the experiment. 

Histology

Enucleated eyes were fixed overnight at 4°C in 10% formalin, and the retinal vasculature was examined by adenosine diphosphatase (ADPase) staining. ²⁶ For the assessment of retinal neovascularization, the eyes were fixed overnight in 10% formalin/PBS, paraffin-embedded, and sliced into 6-μm-thick sagittal sections, 30 μm apart from each other, spanning the entire retina. After period acid-Schiff (PAS)/hematoxylin staining, neovascular cell nuclei anterior to the internal limiting membrane (on the vitreous side) were counted. Eyes from 9 to 10 mice for each genotype/treatment group were analyzed. In each eye, 30 retinal sections were evaluated, and the mean number of neovascular nuclei per section was determined in a blinded manner. Staining was performed in groups containing Epas1 ^+/− and Epas1 ^+/+ mice, to minimize variability in staining techniques. 

Cell Culture

Human embryonic kidney 293 (HEK293) and mouse brain endothelial cells (bEnd3 for transfections; bEnd5 for hypoxia) were maintained in a humidified incubator (37°C and 5% CO₂) in complete Dulbecco’s modified Eagle’s medium (DMEM, 4.5 g/L glucose; Mediatech, Inc., Herndon, VA) supplemented with 10% heat-inactivated fetal bovine serum (Atlanta Biologicals, Inc., Lawrenceville, GA) and penicillin (100 U/mL)/streptomycin (100 μg/mL). For hypoxia, the bEnd5 cells were plated in complete medium in a 60-mm dish at a density of 1.6 × 10⁵ cells/well and allowed to attach overnight. The following morning, the cells were exposed to hypoxia (5% CO₂, 1% O₂, and 94% N₂) in a humidified hypoxic chamber (Coy Laboratory Products, Inc., Grass Lake, MI) or were maintained in normal tissue culture conditions (5% CO₂, 21% O₂, and 74% N₂) at 37°C. Twenty-four hours after initiation of hypoxia or normoxia, the cells were used to prepare total RNA (Qiagen, Inc., Valencia, CA). 

Real-Time RT-PCR

Quantitative real-time RT-PCR was performed by using a sequence-detection system and software (model 7000 Prism; Applied Biosystems, Inc. [ABI], Foster City, CA) with gene-specific primers for proangiogenic factors (Vegf-a, Flk1, Flt1, Tie2, Angpt1, and Angpt2), erythropoietin (Epo), HIF-1 target genes (Car9, Pfkfb3), and a housekeeping gene for the internal standard β-actin (Actb). Expression was compared by using the threshold cycle method. Data are the overall mean of three sets of means, each generated from triplicate real-time RT-PCR data points. Each independent set consisted of four different RNA pools: Epas1 ^+/+ mice in room air, Epas1 ^+/+ mice with hyperoxia treatment, Epas1 ^+/− mice in room air, and Epas1 ^+/− mice with hyperoxia treatment. Each pool was generated from the tissues of three pups matched for genotype, age, and treatment. The bEnd5 real-time RT-PCR studies were performed in triplicate from three separate plates maintained in normoxia or hypoxia. 

Reporter and Expression Plasmids

The upstream regulatory regions from the human CAIX (−440/+31) or mouse Vegf-A (−1106/+123), Flt1 (−970/+188), Flk1 (−639/+298), Tie2 (−1792/+288), Angpt1 (−1121/+95), Angpt2 (−1626/+272), Pecam1 (−1491/+19), and Epo (−410/+1) genes, as well as the mouse Epo enhancer region, were PCR amplified from human or mouse genomic DNA. Promoters were identified by using the start of transcription as annotated in the NCBI (National Center for Biotechnology Information, Bethesda, MD) murine genomic DNA database and mouse RefSeq (NCBI) data (Angpt1, Angpt2, Tie2, Epo, and Pecam1), or as previously described in other promoter studies (Vegf-A, ²⁷ Flt1, ²⁸ and Flk1 ²⁸ ). The amplified regulatory regions were subcloned into the firefly luciferase reporter plasmid pGL3-basic (Promega Corp., Madison, WI) and sequenced before use. For the Epo reporter, the enhancer region was fused upstream of the promoter. Expression plasmids encoding constitutively active human HIF-1α and -2α, designated as P1P2N HIF-1α and -2α, were rendered by alanine substitution of the conserved proline or asparagine residues normally modified in an oxygen-dependent manner and cloned into pIRES-hrGFP (Stratagene, La Jolla, CA). ²⁹  

Transfections

DNA-liposome complexes were formed with reporter plasmid (30 ng), expression plasmid (200 ng), and 1 μL transfection agent (Lipofectamine 2000; Invitrogen Corp., Carlsbad, CA). After 30 minutes, DNA-liposome complexes were added to 50% to 60% confluent cells in 48-well plates, incubated overnight, and harvested for luciferase activity 24 hours after transfection, as described. ²⁹ Each transfection was performed in triplicate. Data are presented as the elevation (x-fold) relative to basal activity of each promoter and represent the mean obtained from the means of three independent sets of transfections. 

Statistics

We report the data as the mean with SD for means generated from individual samples or with SEM for means generated from several group means. Statistical analyses were performed with commercially available software (Excel; Microsoft Corp., Redmond, WA). Comparisons were by Student’s t-tests for paired data. 

Results

We first evaluated whether HIF-2α haploinsufficiency has any effect on retinal vascular development under ambient oxygen conditions. Retinal vasculature from Epas1 ^+/− mice raised in normoxic conditions was indistinguishable from that of Epas1 ^+/+ mice at P12 and P17 (Fig. 1)or earlier time points (P0 and P7, data not shown). To evaluate whether Epas1 haploinsufficiency affects neovascularization, we exposed the mice to an OIR protocol. Retinal flatmounts from Epas1 ^+/+ OIR mice possessed prominent neovascularization (Fig. 2) . In comparison, retinal flatmounts from Epas1 ^+/− OIR mice were suggestive of less neovascularization than those of Epas1 ^+/+ OIR mice (Fig. 2) . To quantify the extent of neovascularization, we examined retinal cross sections, to visualize neovascular tufts (Fig. 3) . Morphometric analyses reveal less neovascular structures in Epas1 ^+/− compared with Epas1 ^+/+ OIR retinas (Fig. 4) . There were no significant differences in neovascularization between control Epas1 ^+/+ or Epas1 ^+/− mice maintained in normoxic conditions. 

To establish a molecular profile for proangiogenic factors that induce neovascularization associated with OIR, we first defined gene expression patterns in Epas1 ^+/+ mice maintained in an OIR protocol compared with normoxia-housed control animals (Fig. 5) . The real-time RT-PCR data revealed three distinct patterns of induction for retinas from the OIR-protocol Epas1 ^+/+ mice (Fig. 5 , white dashed lines) versus normoxia-housed Epas1 ^+/+ mice (Fig. 5 , white solid lines). The first pattern involved peak induction in the early hypoxia phase (P12.5) and unchanged or reduced expression in the late hypoxia phase (P17) compared with the start of the hypoxia phase (P12). The genes that reflect this pattern include Epo, Angpt2, VegfA, Car9, and Pfkfb3. In the second pattern of induction, gene expression increased at P12.5, but reached its maximum at P17. The genes with expression patterns that mirrored this behavior include Pecam1, Tie2, and Flt1. The third pattern is that of unchanged or minimally affected levels of expression from P12 through P17. The genes that reflected this pattern are Angpt1 and Flk1. 

We next defined the effect of Epas1 haploinsufficiency on proangiogenic gene expression in OIR mice. For Pecam1, Tie2, Flt1, Angpt1, and VegfA, expression levels and patterns in Epas1 ^+/− OIR retinal samples (Fig. 5 , black dashed lines) were similar to those in normoxia-housed Epas1 ^+/− retinal samples (Fig. 5 , black solid lines). For Angpt2, the pattern of gene expression fell or remained constant at P12.5 and P17 in OIR versus normoxia-housed Epas1 ^+/− retinal samples, in contrast to the increase in expression observed with Epas1 ^+/+ retinal samples. For Flk1, the pattern of gene expression also fell or remained constant at P12.5 and P17 in OIR versus normoxia-housed Epas1 ^+/− retinal samples, although there was more variability of Flk1 expression in the Epas1 ^+/− retinal samples. Epo gene expression in Epas1 ^+/− OIR retinal samples was substantially blunted at P12.5, although the P17 levels were slightly higher. To assess HIF-1 signaling, we also determined the expression of Car9 and Pfkfb3, the target genes activated in an HIF-1-selective manner. Levels of Car9 were increased in both Epas1 ^+/+ and Epas1 ^+/− retinal samples obtained from OIR mice. However, the maximum induction for Car9 was at P12.5 in Epas1 ^+/+ retinal samples and at P17 in Epas1 ^+/− retinal samples. Similar results were obtained for Pfkfb3. 

HIF-2α is prominently expressed in retinal vascular endothelial cells. To determine whether the changes in gene expression of proangiogenic factors in whole retina samples during OIR are primarily attributable to changes in vascular endothelial cell gene expression patterns, we subjected mouse brain vascular endothelial cells to continuous hypoxia and measured expression of the same panel of proangiogenic factors as examined in the whole retina samples (Table 1) . The changes in gene expression induced in the bEnd5 cells after hypoxia revealed marked induction of Car9 and VegfA, modest induction of Flt1 and Pecam1, no change in Flk1 and Tie2, and reduced expression of Angpt2. The levels of Angpt1 and Epo were below the levels of detection in the bEnd5 cells. 

To assess whether HIF-2α signaling has selective action on proangiogenic gene expression, we determined the effect of HIF-1α or -2α overexpression on the transcriptional activity of the upstream regulatory regions for the proangiogenic factors in an endothelial and a nonendothelial cell line (Fig. 6) . The transfection data revealed that all the regulatory regions were efficiently and preferentially induced by constitutively active HIF-2α, with the notable exceptions of the CAIX promoter (activated preferentially by constitutively active HIF-1α) and the Angpt1 and Pecam1 promoters (not activated by HIF-1α or -2α). 

Discussion

In this study, we characterized the effect of Epas1 haploinsufficiency on ROP-associated neovascularization by using histologic and molecular analyses. The three time-points examined—P12, P12.5, and P17—correspond to 0 hours, 12 hours, or 5 days of relative hypoxia exposure (for OIR mice), respectively. To determine whether the induction of proangiogenic factors in ROP-treated retinas is due to increased expression in vascular endothelial cells, a prominent site of HIF-2α expression in mouse retinas, we exposed mouse brain endothelial cells to hypoxia and measured gene expression patterns. Finally, to determine whether the increased expression of endogenous proangiogenic factors in OIR retinas and hypoxia-treated brain endothelial cells is attributable to increased HIF signaling, we performed transfection assays in vascular endothelial and epithelial cells by using isolated promoter regions derived from the proangiogenic genes. The results in aggregate support the following conclusions. 

At P12.5 in the OIR model, we expected that the changes in gene expression would reflect predominantly hypoxia-induced transcriptional effects. Epo, Angpt2, and VegfA were maximally induced at P12.5 in the Epas1 ^+/+ retinal samples, but were not induced in the Epas1 ^+/− retinal samples, suggesting that these genes may be regulated in a predominantly HIF-2-dependent manner. Consistent with this interpretation, the Epo, VegfA, and Angpt2 regulatory regions were preferentially stimulated by HIF-2α in transfection assays. Only VegfA was induced in the hypoxia-treated bEnd5 cells. Epo was not detectable in the hypoxia-treated bEnd5 cells, probably because the principal cellular compartment in the retina expressing Epo is a nonendothelial cell type. Whether the increased Angpt2 expression in whole retina samples was due to activation by HIF-2α in non–vascular endothelial cells remains to be determined. 

At P17 in the OIR model, we expected that changes in gene expression would reflect an increase in the number of cells associated with neovascularization or would indicate the delayed or sustained effects of hypoxia on gene expression patterns. Pecam1, Tie2, and Flt1 expression were maximal at P17 in Epas1 ^+/+ retinal samples and were blunted in Epas1 ^+/− retinal samples. In hypoxia-treated bEnd5 cells, Flt1, but not Tie2 and Pecam1, was induced. In transfection assays, the Tie2 and Flt1, but not Pecam1, promoter regions were activated with HIF-2α overexpression. As for Epo and Angpt2, the increased Tie2 expression in whole-retina samples may be due to HIF-2α-dependent activation in non–vascular endothelial cells. ³⁰ If in fact increased by HIF signaling, hypoxia-mediated Pecam1 expression may be regulated by another region of the Pecam1 gene, besides the proximal promoter, which contains functional HIF-responsive elements. Alternatively, Pecam1 expression may involve activation by other hypoxia-inducible transcription factors or may involve posttranscriptional regulatory processes. 

Several factors were not clearly dependent on HIF or hypoxia signaling in the mouse retina. Flk1 expression was not increased substantially in the Epas1 ^+/+ OIR retinal samples or in the bEnd5 cells on exposure to hypoxia. However, Flk1 promoter activity was preferentially increased with HIF-2α overexpression. Furthermore, other studies have implicated HIF-2α in regulation of Flk1 expression in mouse brain capillaries. ³¹ Flk1 expression in the hypoxic OIR retina may be restricted to endothelial progenitor cells destined to differentiate into vascular endothelial cells, ^{32 33} and therefore changes in Flk1 expression may be difficult to assess in whole-retina samples. Similarly, bEnd5 cells may not serve as a model for endothelial progenitor cells and therefore may not exhibit increased Flk1 expression in the presence of HIF-2α. 

Angpt1 expression was not increased in Epas1 ^+/+ OIR retinas, nor was Angpt1 detected in the hypoxia-treated bEnd5 cells. The latter was probably due to prominent expression of Angpt1 mRNA in pericytes, ^{34 35} rather than in vascular endothelial cells. ³⁶ The isolated Angpt1 promoter region was not induced by HIF-1α or -2α overexpression in either the vascular endothelial or epithelial cell lines, consistent with Angpt1’s being insensitive to hypoxia. Thus, normalization to Angpt1 gene expression may provide a useful comparison when using whole retina samples from OIR or other experimental paradigms involving hypoxia. We note that expression of the HIF-1-selective target genes Car9 and Fkbfb3 was increased at the P12.5 and P17 time-points in both Epas1 ^+/+ and Epas1 ^+/− retinal samples, indicating the presence of hypoxia and possibly other environmental stimuli that activate HIF-1-dependent signaling pathways in Epas1 ^+/+ and Epas1 ^+/− retinal samples. 

Epo has been suggested to be the primary basis for disease in an HIF-2α knockdown model. ²⁵ In our model, HIF-2α haploinsufficiency indeed resulted in lower Epo levels that were associated with a blunted neovascular response in OIR mice. However, we also noted reduced expression of additional proangiogenic factors that are probably direct HIF-2α target genes. Furthermore, reconstitution of Epo signaling in the HIF-2α knockdown model did not fully compensate for the abnormal retinal vascular phenotype, suggesting the participation of additional HIF-2α-regulated factors in the development of retinal neovascularization in OIR mice. ²⁵ Further study is needed to dissect the molecular mechanisms and cellular compartments that define expression in the mouse retina of previously identified (Epo, VegfA, Tie2, Flt1, and Flk1) as well as novel (Angpt2) HIF-2 target genes. 

HIF-2α haploinsufficiency in humans may not result in an overt retinal phenotype, but may have beneficial effects in reducing the severity of proliferative retinopathies. Similarly, targeted reductions in HIF-2α signaling may offer novel therapeutic opportunities for the treatment of proliferative retinopathies in humans. However, a full complement of HIF-2α gene dosage may be necessary to maintain adequate retinal vascular homeostasis. Long-term studies are needed to assess whether partial reductions in HIF-2α gene dosage, as in HIF-2α-haploinsufficient mice, adversely affects retinal form and function. 

²  Contributed equally to the work and therefore should be considered equivalent authors.

 Supported by March of Dimes Grant 6-FY06-319.

 Submitted for publication November 14, 2007; revised January 2 and 29, 2008; accepted April 11, 2008.

 Disclosure: E.M. Dioum, None; S.L. Clarke, None; K. Ding, None; J.J. Repa, None; J.A. Garcia, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Retinal flatmounts from P12 and P17 EPAS1 ^+/− mice maintained in room air (RA) were indistinguishable from those in EPAS1 ^+/+ mice. Low-magnification views of ADPase-stained retinal flatmounts from (A, B) wild-type (EPAS1 ^+/+) and (C, D) HIF-2α-haploinsufficient (EPAS1 ^+/−) mice maintained in RA conditions and examined at P12 or P17. The radial distributions of the superficial retinal vessels were similar between the two genotypes.

Figure 2.

 

Retinal flatmounts from P12 and P17 Epas1 ^+/− mice exposed to an OIR protocol had reduced neovascularization compared with that in Epas1 ^+/+ mice. Low-magnification views of ADPase-stained retinal flatmounts from (A, B) wild-type (Epas1 ^+/+) and (C, D) HIF-2α-haploinsufficient (Epas1 ^+/−) mice in an OIR model. In this model, mice were maintained in hyperoxia from P5 to P12 and then were transferred to room air. The retinas examined at P12 (A, C), therefore, correspond with the start of the relative hypoxic phase on re-exposure to room air. At P17 (B, D), there was a robust neovascular response in Epas1 ^+/+ mice, whereas this response appeared to be blunted in Epas1 ^+/− mice.

Figure 3.

 

Retinal cross-sections from P17 Epas1 ^+/− OIR mice revealed decreased evidence of neovascularization compared with P17 Epas1 ^+/+ OIR mice. Low-magnification views of ADPase-stained, retinal flatmounts from (A, B) wild-type (Epas1 ^+/+) and (C, D) HIF-2α-haploinsufficient (Epas1 ^+/−) mice maintained in (A, C) room air (RA) or in an OIR protocol (B, D). At P17, there was a greater number of neovascular tufts (B, D, white arrows) observed on the retinal surface for Epas1 ^+/+ compared to Epas1 ^+/− OIR samples. There is no evidence of neovascularization in RA samples for mice from either genotype.

Figure 4.

 

Morphometric analysis confirms a reduced number of neovascular tufts in retinas from Epas1 ^+/− compared with Epas1 ^+/+ OIR mice. Morphometric comparisons at P17 of ADPase-stained, retinal flatmount samples from wild-type (Epas1 ^+/+ □) or HIF-2α-haploinsufficient (Epas1 ^+/− ▒) mice maintained in room air (RA) or in an OIR protocol (n = 6 for each category). There was a significant reduction in the number of neovascular tufts on the retinal surface in Epas1 ^+/+ compared with Epas1 ^+/− samples in the OIR group (P = 0.03).

Figure 5.

 

Kinetics of induction for proangiogenic factors were altered in Epas1 ^+/− compared with Epas1 ^+/+ mice during the hypoxic phase in an OIR protocol. Semiquantitative real-time RT-PCR studies of retinal samples from wild-type (Epas1 ^+/+, white lines) or HIF-2α-haploinsufficient (Epas1 ^+/−, black lines) mice maintained in room air (RA, solid lines) or in an OIR protocol (dashed lines) conditions performed at P12, P12.5, and P17 (P12, P12.5, and P17). The samples from P12 were harvested in RA from RA mice or harvested immediately after removal from the hyperoxic environment from OIR mice. The samples at P12.5 and P17 were harvested from mice maintained at in RA from RA mice or harvested during the indicated time during the relative hypoxia phase in RA from OIR mice.

Table 1.

 

Molecular Profiling of Hypoxia-Treated bEnd5 Cells

Gene	Change (x-fold)
VegfA	51.1 ± 4.1*
Flt1/Vegfr1	5.1 ± 0.4*
Flk1/Vegfr2	1.2 ± 0.4
Angpt1	ND
Angpt2	0.6 ± 0.2
Tie2	1.1 ± 0.1
Pecam1	2.1 ± 0.2, †
Epo	ND
Car9	678.2 ± 61.0*

 Levels of gene expression were measured by real-time RT-PCR for selected proangiogenic factors in hypoxia- versus normoxia-treated mouse brain endothelial cells (bEnd5). There were significant increases in expression of VegfA, Flt1, Pecam1, and Car9 after 16 hours of hypoxia. ND, nondetectable.

*   P ≤ 0.01.

†   P ≤ 0.05.

Figure 6.

 

The regulatory regions of multiple proangiogenic factors were preferentially activated by HIF-2α compared with HIF-1α. Reporter constructs containing regulatory regions for CAIX, Epo, Pecam1, Tie2, Flt1, Flk1, Ang1, Ang2, and VegfA were introduced into (A) HEK293 cells or (B) bEnd3 cells along with control (□), HIF-1α (▪), or HIF-2α (▒) expression plasmids and assayed for luciferase activity 20 hours after transfection. Data represent the mean of three independent transfections with each data point for each transfection generated from triplicate samples. Error bars, SEM. Each reporter was normalized to the respective basal activity of the reporter when transfected with the control expression plasmid (normalized basal activity set at 1).

The authors thank Alok Das and Julian Restrepo for technical assistance and the UTSWMC Molecular Pathology Core for use of their facilities.