Hypoxia-inducible Factor-2α (HIF-2α) Is Involved in the Apoptotic Response to Hypoglycemia but Not to Hypoxia*

Deprivation of oxygen (hypoxia) and/or glucose (hypoglycemia) represents a serious stress that affects cellular survival. The hypoxia-inducible transcription factor-1α (HIF-1α), which has been implicated in the cellular response to hypoxia (Semenza, G. L. (1999) Annu. Rev. Cell Dev. Biol. 15, 551–578), mediates apoptosis during hypoxia (Halterman, M. W., Miller, C. C., and Federoff, H. J. (1999) J. Neurosci. 19, 6818–6824 and Carmeliet, P., Dor, Y., Herbert, J. M., Fukumura, D., Brusselmans, K., Dewerchin, M., Neeman, M., Bono, F., Abramovitch, R., Maxwell, P., Koch, C. J., Ratcliffe, P., Moons, L., Jain, R. K., Collen, D., and Keshet, E. (1998)Nature 394, 485–490), but the function of its homologue HIF-2α remains unknown. Therefore, the role of HIF-2α in cellular survival was studied by targeted inactivation of theHIF-2α gene (HIF-2α−/−) in murine embryonic stem (ES) cells. In contrast to HIF-1α deficiency, loss of HIF-2α did not protect ES cells against apoptosis during hypoxia. Both HIF-1α−/− and HIF-2α−/− ES cells were, however, resistant to apoptosis in response to hypoglycemia. When co-cultured with wild type ES cells, HIF-2α−/− ES cells became rapidly and progressively enriched in hypoglycemia but not in hypoxia. Thus, HIF-1α and HIF-2α may have distinct roles in responses to environmental stress, and despite its name, HIF-2α may be more important in the survival response to environmental variables other than the level of oxygen.

HIF-1␣ 1 is a basic helix-loop-helix transcription factor, which mediates the cellular adaptation to hypoxia (1,3). During hypoxia, HIF-1␣ up-regulates the expression of a number of genes involved in erythropoiesis, glycolysis, and angiogenesis by formation of a heterodimer with HIF-1␤, which binds to a hypoxia-response element in the promoter of these target genes (1,3,(5)(6)(7). Several studies have shown that loss of HIF-1␣ or HIF-1␤ impaired gene expression in response to hypoxia and/or hypoglycemia (5)(6)(7)(8), even though the precise molecular mechanisms for the latter condition remain largely undetermined. HIF-1␣ has also been implicated in the induction of apoptosis during stressful conditions of hypoxia and hypoglycemia (5).
Recently, a novel hypoxia-inducible factor, HIF-2␣ (also known as EPAS1* (9), HLF (10), HRF (11), or MOP2 (12)), was identified, which also binds as a heterodimer with HIF-1␤ to the hypoxia-response element. Like HIF-1␣, HIF-2␣ is subject to oxygen-dependent proteosomal destruction, mediated by the von Hippel-Lindau tumor suppressor protein (13), and the protein levels of HIF-2␣ are increased under hypoxic conditions (14). Gene inactivation studies revealed that HIF-2␣ is essential for cardiovascular development and angiogenesis (15,16), but it remains unknown whether these embryonic defects resulted from insufficient hypoxic up-regulation of target genes. In fact, expression of vascular endothelial growth factor (VEGF), VEGF receptor-2 (VEGFR-2/Flk-1), and the endothelial receptor Tie-2 was comparable in HIF-2␣ Ϫ/Ϫ and wild type embryos (16). Although ectopic overexpression of HIF-2␣ in vitro stimulates reporter gene expression in hypoxic conditions, the induction is variable and lower than that by HIF-1␣, and surprisingly, HIF-2␣ already stimulates reporter gene expression in normoxia (9,12,14,17,18). Thus, the precise role of endogenous HIF-2␣ in mediating the cellular responses to hypoxia remains uncertain. In addition, the role of HIF-2␣ as compared with HIF-1␣ in apoptotic processes remains to be elucidated. Because HIF-dependent gene regulation is emerging as a target for anti-or proangiogenic treatments (19,20), it is important to define the role of HIF-2␣. In the present study, the endogenous role of HIF-2␣ in cellular survival was examined by targeted gene inactivation in murine embryonic stem (ES) cells.

Generation of Targeted ES Cell Clones-ES cells deficient in HIF-1␣
(HIF-1␣ Ϫ/Ϫ ) were generated previously (5). The HIF-2␣ targeting vector pPNT.HIF-2␣ (Fig. 1A) contained a 5Ј-flanking 3.5-kb HindIII/StuI fragment upstream of exon 2 and a 3Ј-flanking 7.0-kb fragment (a 2-kb NheI/BamHI fragment and an immediately downstream 5-kb BamHI/ EcoRI fragment) downstream of exon 2. Culture and targeting of undifferentiated ES cells, including selection in high G418 and Southern blot analysis, were done as described (5). Inactivation of the HIF-2␣ gene was confirmed by RT-PCR and immunoblot analysis. Forward * This work was supported by BIOMED Grant PL963380 (to P. C. and D. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Instituut voor de aanmoediging van innovatie door wetenschap en technologie in Vlaanderen fellow.
Hypoxic and Hypoglycemic Treatment of ES Cells and Apoptosis Measurements-ES cells were seeded at low cell density (10,000 cells/ 35-mm culture dish) and cultured in medium supplemented with 20 ng/ml leukemia inhibitory factor and 5% fetal calf serum (FCS) for 24 h. Afterward, ES cells were stressed in the same medium without leukemia inhibitory factor under normoxia (20% O 2 ), hypoxia (0.5% O 2 ), hypoglycemia (0% glucose or 6 mM 2-deoxyglucose), or after the addition of a mixture of cytokines (TNF-␣ (30 ng/ml), IFN-␥ (5 ng/ml), and interleukin-1␤ (3 ng/ml)). Incubations were done at 37°C under normoxia by maintaining the cells in 95% air and 5% CO 2 or under hypoxia by incubating cells in a humidified sealed chamber gassed with 0.5% O 2 , 5% CO 2 , and 94.5% N 2 (AGA, Lille, France) at a flow rate of 30 liters/h. After incubation in normoxia or hypoxia for 24 h, ES cells were analyzed for apoptosis or used for immunoblotting. Apoptosis was measured by quantitative detection of cytoplasmic histone-associated oligonucleosomes or by counting TUNEL-positive cells as described (5). Bcl-2 and p53 levels were quantified by enzyme-linked immunosorbent assay as described (5).
ES Cell Stress Cycle Experiment-Stress cycle experiments were performed in a manner similar to that described by Graeber et al. (22). HIF-2␣ Ϫ/Ϫ ES cells were stably transfected by electroporation with a plasmid bearing the ␤-galactosidase gene (under control of the cytomegalovirus promoter) and the hygromycin phosphotransferase gene (selection marker). Wild type (WT) ES cells were transfected with a control plasmid bearing only the selection marker. Transfected HIF-2␣ Ϫ/Ϫ and WT ES cells were cultured as single populations or mixed at different ratios (20:1, 1:1, and 1:20) in 10-cm culture dishes until 30% confluency was reached. Subsequently, culture medium containing 15% FCS was replaced, and ES cells were cultured in normoxia, hypoxia, or hypogly-cemia (described above) for 20 h in medium containing 5% FCS. At the end of every cycle, stressed ES cells were allowed to recover for 4 h in normoxia/normoglycemia (medium with 15% FCS). Thereafter, ES cells were split, cultured until they reached 30% confluency, and challenged with another cycle of 20 h of stress. At every split, 300,000 ES cells were replated in a 24-well plate, subcultured for 24 h, and subsequently fixed with 2% formaldehyde, 0.5% glutaraldehyde. After staining for ␤-galactosidase as described (15), the percentage of HIF-2␣ Ϫ/Ϫ and WT ES cells in the mixed populations was quantified using the Scion Image 1.60c system (Meyer Instruments, Inc., Houston, TX).
We have previously demonstrated that apoptosis of ES cells in response to hypoxia and hypoglycemia is mediated by HIF-1␣ via a mechanism involving up-regulation of p53 and down-regulation of Bcl-2 (5). The role of HIF-2␣ in cellular apoptosis during hypoxia was compared with that of HIF-1␣ by measuring the number of oligonucleosomes (or TUNEL-positive cells; similar results; not shown), which are liberated during apoptosis. WT, HIF-1␣ Ϫ/Ϫ , and HIF-2␣ Ϫ/Ϫ ES cells did not undergo apoptosis under normal conditions (25 mM glucose and 20% oxygen; Table I). During hypoxia (0.5% O 2 ) or hypoglycemia (0 mM glucose), apoptosis significantly increased in WT ES cells but, as previously reported (5), not in HIF-1␣ Ϫ/Ϫ ES cells (Table I). In contrast, HIF-2␣ Ϫ/Ϫ ES cells were refractory to hypoglycemia but, surprisingly, not to hypoxic stress (Table I). The role of HIF-1␣ and HIF-2␣ in hypoxia and hypoglycemia was specific, because serum deprivation (not shown) and stimulation with Fas ligand (not shown) or a cytokine-mixture (IL-1␤/IFN-␥/TNF-␣) comparably stimulated apoptosis in WT, HIF-1␣ Ϫ/Ϫ , and HIF-2␣ Ϫ/Ϫ ES cells (Table I). When apoptosis occurred, protein levels of p53 (a mediator of genotoxic apoptosis, up-regulated during hypoxia) were up-regulated, whereas those of Bcl-2 (an apoptosis inhibitor) were reduced (Table I).
These data indicate that, in contrast with HIF-1␣, HIF-2␣ specifically induces up-regulation of p53 and down-regulation of Bcl-2 in ES cells under hypoglycemia but not under hypoxia.
To confirm the role of HIF-2␣ in the control of ES cell survival during hypoglycemia, we studied whether HIF-2␣ Ϫ/Ϫ ES cells would exhibit a survival advantage and become enriched over WT ES cells when both cell types were co-cultured in hypoglycemia. Therefore, HIF-2␣ Ϫ/Ϫ and WT ES cells were  mixed at a ratio of 1:20 (5% HIF-2␣ Ϫ/Ϫ and 95% WT ES cells) ( Fig. 2A) and cultured for repetitive cycles of 20-h hypoglycemia or hypoxia alternating with 28-h normoxia/normoglycemia. To distinguish HIF-2␣ Ϫ/Ϫ from WT ES cells, the former were stably transfected with a ␤-galactosidase gene (allowing easy detection after LacZ staining), whereas the latter were mock-transfected. Pools of HIF-2␣ Ϫ/Ϫ and WT ES cells were used to avoid clonal selection. When each of these cell types was grown separately in normoxia/normoglycemia, HIF-2␣ Ϫ/Ϫ ES cells grew slightly faster than WT cells (doubling time: ϳ15 h for HIF-2␣ Ϫ/Ϫ cells versus 16 h for WT cells). However, when the mixture of both cell types was intermittently and repetitively challenged by hypoglycemia (20-h hypoglycemia followed by 28-h normoxia/normoglycemia), HIF-2␣ Ϫ/Ϫ ES cells became progressively enriched over WT cells, and after 13 cycles of hypoglycemic stress, WT cells were overgrown and absent from the culture dish as revealed by LacZ staining (Fig. 2, B, G, and  J). In contrast, intermittent hypoxia or normoxia failed to enrich LacZ-positive HIF-2␣ Ϫ/Ϫ ES cells within 13 cycles (Fig.  2, C, D, G, H, and I). Because of their slightly faster growth potential (see doubling times above), HIF-2␣ Ϫ/Ϫ ES cells ultimately also became enriched in hypoxia or normoxia/normoglycemia but only after many more cycles of hypoxia (n ϭ 33) than of hypoglycemia (n ϭ 13) (Fig. 2, G-J). This is not surprising considering the slightly faster intrinsic growth rate of HIF-2␣ Ϫ/Ϫ cells in normoxia/normoglycemia, e.g. any cell type with a slightly shorter doubling time will overgrow another more slowly growing cell population, even if the difference in doubling time is minimal. Importantly, however, enrichment of HIF-2␣ Ϫ/Ϫ cells required a comparable number of passages in normoxia and hypoxia, indicating that these cells had no survival advantage over WT cells under hypoxia. In contrast, under hypoglycemia, the enrichment of HIF-2␣ Ϫ/Ϫ cells occurred much faster than in normoxic or hypoxic conditions (Fig.  2G), indicating that these cells indeed had a survival advantage over WT cells in hypoglycemic conditions. Similar results were obtained when different ratios of HIF-2␣ Ϫ/Ϫ and WT ES cells were mixed and challenged (not shown). Thus, the fast and progressive enrichment of HIF-2␣ Ϫ/Ϫ ES cells in hypoglycemia but not in hypoxia is consistent with their refractoriness to hypoglycemia-induced apoptosis and with the minimal involvement of HIF-2␣ in hypoxia-induced apoptosis.
To examine whether HIF-1␣ and HIF-2␣ always had a distinct role in mediating the response to hypoxia and hypoglycemia, we determined by quantitative Real Time RT-PCR the expression levels of known HIF-1␣ target genes in ES cells under normoxia, hypoxia, and hypoglycemia. The role of HIF-1␣ and HIF-2␣ in controlling gene expression was found to be dependent on the specific target gene (Fig. 3). Similar to its role in regulating cell survival, HIF-2␣ controlled expression of the glucose transporter-1 (Glut-1) in hypoglycemia but not in hypoxia, whereas HIF-1␣ up-regulated its expression in both conditions, consistent with previous findings (1,2,(5)(6)(7). In contrast, expression of phosphoglycerate kinase-1 (PGK-1) in hypoxia and hypoglycemia was only dependent on HIF-1␣ but not on HIF-2␣, whereas induction of VEGF and Flk-1 transcripts was dependent on both HIF-1␣ and HIF-2␣ (Fig. 3,  A-D). The role of HIF-2␣ in hypoxic gene expression of VEGF (10,14,24) and Flk-1 (17) confirms previous findings. HIF-2␣ induced expression of Flt-4/VEGFR-3 in hypoglycemia and hypoxia, whereas HIF-1␣ only up-regulated its expression in hypoglycemia. Expression of the endothelial receptor Tie-2 was not induced by hypoxia, but its up-regulation under hypoglycemia was more dependent on HIF-2␣ than HIF-1␣ (Fig. 3, E-F). Previous studies also observed that reporter gene expression under control of the Tie-2 promoter was only induced by HIF-2␣ but not by HIF-1␣ (9). Taken together, these findings indicate that HIF-1␣ and HIF-2␣ have differential and distinct roles in biological processes such as apoptosis, angiogenesis, and glycolysis dependent on the particular target gene and stress condition.
In conclusion, these genetic data indicate that HIF-1␣ and HIF-2␣ have distinct roles in the adaptation of the cellular response to deprivation of oxygen or nutrients. Whereas HIF-1␣ is involved in regulating the expression of both angiogenic factors (VEGF, Flk-1) and glycolytic enzymes (Glut-1, PGK-1), HIF-2␣ is more restricted to the regulation of angiogenic processes but is only minimally involved in glycolysis. In addition, this study provides evidence that HIF-2␣ may act more as a hypoglycemia than a hypoxia response factor in apoptotic processes. How HIF-2␣ is involved in the cellular response to hypoglycemia remains to be unraveled because HIF-2␣ protein does not accumulate during hypoglycemia, 2 suggesting that the latter condition activates HIF-2␣ via mechanisms distinct from those whereby hypoxia activates HIF-2␣ and HIF-1␣ (1,3,13,14). Our data indicate that gene induction by either HIF-1␣ and/or HIF-2␣ may not only depend on the particular target gene but also on the type of stress. Because the expression pattern of HIF-1␣ and HIF-2␣ in vivo is distinct (9 -11, 25), gene induction by a particular HIF is likely to depend on the cell type.
The role of HIF-2␣ in the cellular response to hypoglycemia might be significant for a number of biological processes and pathological disorders. For instance, hypoglycemia may cause congenital cardiac (26) and neural malformations (27). In preterm or small for gestational age infants, hypoglycemia causes cerebral damage and edema (28) and impairs psychomotor development (29), in particular in children of diabetic mothers (30). Glucose levels are often undetectable in tumors (31), whereas neurons in Alzheimer's or Parkinson's disease have impaired glucose uptake, causing intracellular glucose deprivation (32). Non-coma hypoglycemia or stroke may also induce neuronal apoptosis (33). Because all these processes are characterized by glucose deprivation in tissues, known to express HIF-2␣ (10,11,25), an intriguing yet outstanding question remains whether HIF-2␣ is involved in the pathogenesis of these disorders, and if so, whether modulation of its activity might bear any therapeutic potential.