Hypoxia Regulates Vascular Endothelial Growth Factor Gene Expression in Endothelial Cells
Identification of a 5′ Enhancer
Abstract Vascular endothelial growth factor (VEGF) is a potent mitogen specific for endothelial cells. Its expression is dramatically induced by low oxygen tension in a variety of cell types, and it has been suggested to be a key mediator of hypoxia-induced angiogenesis. Although VEGF action is targeted to endothelial cells, it is generally believed that these cells do not express VEGF. In addition, the mechanisms by which hypoxia regulates VEGF production remain unclear. We report in the present study that pulmonary artery endothelial cells do not express VEGF under basal conditions; however, significant VEGF mRNA levels accumulate when these cells are exposed to hypoxia. Using a DNA fragment containing human VEGF promoter sequence, we identified a 28-bp element that is necessary and sufficient to upregulate transcription in response to hypoxia. This element can act as a hypoxia-specific enhancer when placed upstream or downstream from a heterologous promoter. The enhancer includes, in addition to an octamer homologous to the hypoxia-inducible factor-1 (HIF-1) consensus, a sequence that resides 3′ to the consensus. Although this sequence may not be involved in the binding of HIF-1, it is absolutely required for the enhancer activity and may be the binding site for certain constitutive binding proteins. The expression of VEGF by endothelial cells in response to hypoxia may provide an important mechanism by which endothelial cell permeability and proliferation is regulated in an autocrine manner.
Hypoxia is an important regulator of blood vessel tone and structure. It has also been shown to be a potent stimulus of angiogenesis, a process whereby neovascularization arises from existing blood vessels. For the past several years, many growth factors have been characterized and shown to have angiogenic activities (for review, see Reference 11 ). Among these, VEGF, also known as vascular permeability factor,2 3 4 has attracted special attention and is believed to be the prime regulator of angiogenesis.5 This secreted protein has mitogenic activity that is specific for vascular endothelial cells.6 7 8 9 10 Recent studies have shown that the expression of VEGF can be induced by hypoxia,11 12 a condition that causes tumor necrosis and stimulates angiogenesis.13 14 Northern blot analysis indicated that VEGF mRNA in cultured glioma cells was dramatically increased when cells were grown under hypoxic conditions.11 In situ hybridization revealed that VEGF and its receptor mRNAs were elevated in tumor cells and that the expression was adjacent to tumor necrotic areas.11 12 These in vitro and in vivo data support a model whereby low oxygen tension associated with tumor necrosis induces the expression of VEGF, which in turn stimulates the proliferation of vascular endothelial cells in a paracrine manner, leading to the sprouting of new capillary vessels. It is generally believed that endothelial cells, the target of VEGF action, do not express VEGF under physiological or pathophysiological (such as hypoxic) conditions.
Hypoxia has been reported to regulate the expression of many genes,15 16 17 18 but the mechanisms involved are poorly understood. An activity termed HIF-1 has been shown to increase the expression of Epo in response to hypoxia,19 and more recently, HIF-1 binding sites have been identified in the genes encoding a number of glycolytic enzymes.18 However, the molecular mechanisms by which hypoxia regulates VEGF expression are not known.
In the present study, we present data showing that hypoxia increases the transcriptional rate of the VEGF gene in vascular endothelial cells. Furthermore, from a systematic analysis of the promoter, we present the identification and characterization of a 5′enhancer that is responsible for the hypoxia-induced increases in VEGF gene expression.
Materials and Methods
Cell Culture and Media
BPAE cells were cultured in DMEM supplemented with 10% fetal bovine serum, 2 mmol/L l-glutamine, and 0.1 mg/mL gentamicin sulfate. The cells were exposed to a normoxic environment consisting of 21% O2/5% CO2/remainder N2 or to hypoxic conditions with 95% N2/5% CO2. The measured Po2 in the media was 18 to 20 mm Hg under 0% O2 conditions and 130 mm Hg under 21% O2 conditions, as we previously reported.15 Cells were cultured at 37°C.
Plasmids and Oligonucleotides
A plasmid that contains the human VEGF 5′ sequence and part of the coding region was kindly provided by Judy A. Abraham at Scios Nova Inc, Mountain View, Calif. A DNA fragment covering the 5′ upstream region and part of the untranslated coding sequences of VEGF was excised from this plasmid and was used to replace the TK promoter (Xba I–Xho I) in the plasmid pBLCAT2.20 The resulting plasmid pVR47/CAT contains the VEGF sequence from −2362 to +61, relative to the transcription initiation site determined by Tischer et al21 (or from 1 to 2423 if using the numbering system of GenBank, accession No. M63971). The plasmids that contain the truncated 5′ sequences described in Fig 2⇓ were all generated from pVR47/CAT by restriction digestion and religation or by unidirectional deletion using Exonuclease III from New England Biolabs. Plasmid pRSV–β-gal has been previously described.22 Plasmid pCAT promoter is from Promega. The oligonucleotide primers used to amplify different segments in the VEGF promoter region all contain 5′-gcggatcccggg-3′ (sense strand) or 5′-gaagatct-3′ (antisense strand) linked to 19 to 23 nucleotides of the native sequence. The sense strands of the oligonucleotides used in the electrophoretic mobility shift assays are as follows: W18 (5′-agcttGCCCTACGTGCTGTCTCAgaatt-3′), A-C′ (5′-gcggatcccgggCCACAGTGCATACGTGGGCTCagatcttc-3′), A-G (5′-gcggatcccgggCCACAGTGCATACGTGGGCTCCAACAGGTCCTCTTagatcttc-3′), and an NF-κB binding site (5′-agcttCAGAGGGGACTTTCCGAGAGGtcga-3′), where the native sequences are in uppercase letters and linker sequences are in lowercase letters.
Total cellular RNA was prepared by guanidinium isothiocyanate extraction from BPAE cells exposed to hypoxic (0% O2) or normoxic (21% O2) environments for various periods. Total RNA (15 μg per lane) was electrophoresed in 1% agarose gels containing formaldehyde, and Northern analysis was performed. As probes, we used an 800-bp DNA fragment, which hybridizes to the first exon of the human VEGF gene,21 and an 800-bp Pst I fragment of the mouse β-actin gene. The DNA was labeled with [α-32P]dCTP in a standard random-primed reaction to a specific activity of 1 to 2×109 cpm/μg. Hybridization was performed for 2 hours at 68°C in QuikHyb solution (Stratagene), followed by low-stringency and high-stringency washes at room temperature and 60°C, respectively. Autoradiography was performed at −80°C for 2 days. The filters were stripped and reprobed with β-actin to normalize for RNA loaded.
Transfections and CAT Assays
Transfections of BPAE cells were carried out with LipofectAmine from GIBCO-BRL according to the manufacturer’s protocol. After transfection, cells were placed in either normoxia or hypoxia. After incubation for 24 to 48 hours, cells were lysed, and β-galactosidase and CAT activity were measured according to previously described protocols.23 Normalized CAT activity was the ratio of radioactivity (in counts per minute) of labeled acetylchloramphenicol to the optical density units from the cleavage product of o-nitrophenyl-β-d-galactopyranoside catalyzed by β-galactosidase.
Nuclear Extract Preparation and EMSAs
Cell nuclear extracts were prepared according to the method of Schreiber et al,24 and total protein was quantified by using the Bio-Rad protein assay. For EMSA, 5 μg nuclear proteins were incubated for 10 minutes at room temperature in a 20 μL binding mixture containing 1× HM buffer (1× HM buffer consists of 10 mmol/L HEPES, pH 7.9, 0.5 mmol/L MgCl2, 0.1 mmol/L EDTA, and 5% glycerol), 100 to 130 mmol/L KCl, 1 μg polydeoxyinosinic-deoxycytidylic acid, and in some cases competitor oligonucleotides. Radiolabeled oligonucleotides (22 000 cpm) were then added, and the incubation was continued for 20 minutes. The binding mixture was fractionated on a 4% polyacrylamide gel in 0.5× TBE (1× TBE consists of 89 mmol/L Tris base, 89 mmol/L boric acid, and 5 mmol/L EDTA) at 4°C. After it was dried, the gel was autoradiographed at −80°C.
Hypoxia Increases VEGF mRNA Levels in Endothelial Cells
BPAE cells were exposed to 0% O2 or to an ambient environment for 18 hours. Northern blot analysis of total RNA showed a single 3.7-kb VEGF message in RNA isolated from hypoxic endothelial cells, but no signal was detected in normoxic cell RNA (Fig 1⇓). β-Actin mRNA levels were not altered by hypoxia under these conditions. VEGF mRNA levels increased by 6 hours of hypoxia, when the Po2 is ≈60 mm Hg,16 and remained elevated at 48 hours (data not shown). The heavy metals CoCl2 and NiCl2, known to have the same effects on gene expression as hypoxia, also increased VEGF mRNA levels in endothelial cells. CdCl2, on the other hand, had no effect on VEGF mRNA (authors’ unpublished data, 1994).
Hypoxia Increases the Transcriptional Rate of the VEGF Gene
To examine whether the increases in VEGF mRNA in endothelial cells are regulated at the transcriptional level, we constructed a plasmid in which the promoter of the human VEGF gene was fused to the reporter CAT gene by replacing the TK promoter in plasmid pBLCAT2 with that of VEGF (Fig 2⇓). The resulting plasmid pVR47/CAT contains VEGF sequences from −2362 to +61 (relative to the transcription initiation site, see “Materials and Methods”). This plasmid was then transfected into BPAE cells, which were incubated under either hypoxic (0% O2) or normoxic (21% O2) conditions. Plasmid pRSV–β-gal, containing the β-galactosidase gene under the control of the Rous sarcoma virus long terminal repeat, was cotransfected into BPAE cells. After incubation, cells were lysed, and CAT activity was measured and normalized to that of β-galactosidase. As shown in Fig 2⇓, CAT activity was induced 5.5-fold by hypoxia. These results indicate the presence of a positive hypoxia response element in the 5′ region of the VEGF gene that regulates its transcription.
Delineation of the Hypoxia Response Element of VEGF
A series of deletions were made upstream from the transcription initiation site in the plasmid pVR47/CAT (Fig 2⇑). These new plasmids were transfected into BPAE cells, and CAT activity was determined. As shown in Fig 2⇑, hypoxia inducibility was retained until the deletions were extended to −795 [plasmid pV(−795)/CAT]. Further deletions resulted in loss of a hypoxic response; however, the basal activity was retained and was similar in all constructs. Since plasmid pV(−985)/CAT still retains the hypoxia inducibility and is 190-bp longer in 5′ sequence than pV(−795)/CAT, we postulated that the hypoxia response element may reside in this 190-bp fragment. Indeed, when this 190-bp fragment (−985 to −790) was excised from the surrounding sequence and inserted in front of the TK promoter in plasmid pBLCAT2, it provided a strong (8.4-fold) induction on CAT gene expression (Fig 3⇓).
Further localization of this element was carried out by amplifying different segments within the 190-bp region by polymerase chain reaction and testing their ability to increase CAT gene expression in response to hypoxia. The smallest region that still provides hypoxia inducibility (4.7-fold) is a 28-bp fragment (−978 to −951) close to the 5′ end of the 190-bp sequence (Fig 3⇑).
To investigate whether the hypoxia regulatory element has the characteristics of a classic enhancer, we inserted the 28-bp fragment mentioned above into the BamHI site upstream from the TK promoter in pBLCAT2 in both orientations. The resulting plasmids both showed a 4.7-fold induction by hypoxia. We have also inserted a second fragment (−985 to −863) in either orientation downstream from the CAT gene and obtained similar results (data not shown), demonstrating that the hypoxia response element is an enhancer.
cis-Acting Elements in the VEGF Enhancer in Addition to the HIF-1 Binding Consensus Are Needed for the Stimulation of the VEGF Gene by Hypoxia
A recent study by Semenza et al18 has shown that the HIF-1 recognition sequence contained in the hypoxia response element of the Epo gene is conserved in a number of glycolytic enzyme genes that are hypoxia inducible. The consensus in these elements is (G/C/T)ACGTGC(G/T) (see Fig 4⇓). In sequence comparison analysis, we have found a short region (underlined in Fig 4⇓) in the VEGF hypoxia response element that is highly homologous to this consensus. The sequence of this region is TACGTGGG and differs from the consensus in the seventh nucleotide, which is a G in the VEGF enhancer (double-underlined) and a conserved C in the Epo and glycolytic enzyme genes. We have made several deletions in the 35-bp fragment containing the minimal 28-bp hypoxia response element and tested their function in transfection experiments (Fig 4⇓). We observed that the disruption of the core HIF-1 binding sequence (construct V1714) eliminated hypoxia inducibility. Interestingly, we found that the downstream sequence (3′ to the core) is also important, because the replacement of this region with linker or vector sequences without changing the core either dramatically reduced the hypoxic induction (construct V1316b) or abolished it completely (construct V1318). Therefore, critical elements (TCCTCTT) in the region 3′ to the consensus are required for VEGF enhancer function.
A Hypoxia-Inducible Factor Binds to the Enhancer
An enhancer element exerts its function through the binding of a trans-acting factor. To demonstrate the presence of such a factor, we have radiolabeled fragment A-G (see Fig 4⇑) and performed in vitro binding followed by EMSA. As shown in Fig 5A⇓, a distinct band was induced in nuclear extracts isolated from hypoxia-treated cells (lane 3) but not by normoxia-treated cells (lane 2), indicating the presence of an inducible trans-acting factor. Competition EMSA was carried out by incubating nuclear proteins with labeled fragment A-G in the presence of several unlabeled competitors. The binding was competed by increasing amounts (20-, 100-, and 500-fold excess) of unlabeled fragment A-G (lanes 4 through 6) but not by the same amounts of an arbitrary oligonucleotide (in this case, an NF-κB binding sequence) (lanes 7 through 9), indicating that the binding activity is specific for the VEGF enhancer.
The only hypoxia-inducible factor that has been investigated so far is the one that binds to the enhancer region of the Epo gene, named HIF-1.19 25 26 To explore potential common mechanisms of transcriptional regulation by hypoxia between VEGF and Epo, competition EMSA was carried out. A double-stranded oligonucleotide competitor (W18), which is contained in the Epo enhancer and has been shown to bind to HIF-1,19 was added to the cell nuclear extract before the addition of labeled VEGF enhancer (fragment A-G). The competitor was in 20-, 100-, 500-, and 1000-fold excess to the labeled VEGF enhancer (Fig 5A⇑, lanes 10 through 13). The EMSA results showed that W18 can compete with labeled A-G for nuclear factors, although to a lesser extent than the unlabeled A-G fragment. Thus, factors binding to the VEGF enhancer may be related to, if not the same as, the component(s) that binds to the Epo enhancer.
Interestingly, a 21-bp fragment (contained in construct V1318), which retains an intact HIF-1 site but has a deletion of the sequence 3′ to the consensus, allowed for hypoxia-induced binding to occur in vitro (Fig 5B⇑). Since this fragment does not have a function in the absence of the downstream sequence (Fig 4⇑), the binding of HIF-1 (or a similar factor) to the VEGF enhancer is not sufficient to provide hypoxia inducibility.
Hypoxia is a recognized stimulus for blood vessel remodeling, altered cellular permeability, and the process of angiogenesis. VEGF is dramatically induced by hypoxia in a number of tumor cells as well as in fibroblasts and smooth muscle cells.11 12 27 Although endothelial cells are the target cells of VEGF action, they are not known to produce VEGF. We report in the present study that pulmonary artery endothelial cells are capable of expressing VEGF under conditions of hypoxia by increasing the transcriptional rate of the gene.
We have identified a hypoxia-responsive enhancer in the promoter region of the human VEGF gene, which includes a 28-bp element that is sufficient to mediate the upregulation of VEGF gene transcription. Gel retardation assays revealed a hypoxia-inducible factor that can bind to this element. In addition, we showed that oligonucleotides corresponding to the Epo enhancer sequence could compete for the binding of this element, suggesting that the binding protein for these two may be related.
Semenza et al18 recently reported the HIF-1 binding sequence originally described in the Epo enhancer to be conserved in several genes encoding enzymes in the glycolytic pathway. The consensus in these elements is (G/C/T)ACGTGC(G/T). Interestingly, a computer-aided search in the VEGF promoter region has revealed several sequences that closely resemble the HIF-1 binding consensus. One of them, CGCACGTA, at position −313 (or 2050 if using the numbering system of GenBank, accession number M63971), is 100% homologous to this consensus when read from the antisense strand. However, VEGF promoter constructs containing this sequence, but lacking the 28-bp element, did not show any hypoxia inducibility in endothelial cells [plasmids pV(−795)/CAT and pV(−416)/CAT; see Fig 2⇑]. Furthermore, in our deletion experiments, we demonstrated that the replacement of nucleotides 3′ to the consensus in the VEGF element (constructs V1316b and V1318; see Fig 4⇑) resulted in either a dramatically reduced or a completely abolished induction by hypoxia despite the presence of the intact HIF-1 consensus sequence (Fig 4⇑). Therefore, it appears that the VEGF enhancer function requires not only the conserved HIF-1 recognition sequence but also its flanking context, which may be bound by other factors. Further dissection of the VEGF enhancer is in progress.
Previous studies have indicated that the expression of AP-1 proteins c-jun and c-fos are upregulated by hypoxia.28 29 Since there are three potential AP-1 binding sites in the promoter region of VEGF,21 it has been speculated that hypoxia induction of VEGF expression might be mediated by AP-1. Our results suggest that the AP-1 site may not be a necessary element for hypoxic regulation, because the 28-bp enhancer is sufficient to provide hypoxia response when inserted in front of a reporter gene. However, to completely exclude the possibility of the involvement of AP-1 (or any other suspected elements in the promoter) in hypoxic regulation, it is necessary to test a construct in which only the enhancer is mutated and all the surrounding sequences are intact.
The identification of a 5′ enhancer that upregulates VEGF transcription presents a molecular mechanism that links hypoxia and angiogenesis. However, there may be other mechanisms (eg, other transcriptional elements or posttranscriptional events) that also contribute to the hypoxic regulation of VEGF expression. Minchenko et al30 found two regions in the human VEGF gene that were shown to be hypoxia responsive in transient transfection experiments using HeLa cells. The enhancer identified in the present study using endothelial cells is distinct from the above-reported regions in experiments using the HeLa cell system. It is possible that more than one element contributes to the hypoxic induction of the human VEGF gene. Additionally, the relative contributions of each element may vary according to cell type, such that the 28-bp enhancer identified in the present study may be important in the hypoxic response of endothelial cells, whereas the regions reported by Minchenko et al may play a role in regulating VEGF expression in HeLa cells. According to our findings, the expression of VEGF by hypoxic endothelial cells implicates these cells not only as passive responders to this potent mitogen but also as possible regulators of their growth and permeability in an autocrine manner.
Selected Abbreviations and Acronyms
|BPAE||=||bovine pulmonary artery endothelial|
|EMSA||=||electrophoretic mobility shift assay|
|VEGF||=||vascular endothelial growth factor|
Dr Liu was supported by Individual National Research Service Award HL-09008. Dr Kourembanas was supported in part by a grant from the American Heart Association and by National Institutes of Health grant P50 HL-46491. We thank Dr Judy A. Abraham for providing plasmids, Dr S. Alex Mitsialis for critical reading of the manuscript, and Kelly Ames for her secretarial assistance.
- Received May 26, 1995.
- Accepted July 10, 1995.
- © 1995 American Heart Association, Inc.
Folkman J, Shing Y. Angiogenesis. J Biol Chem. 1992;267:10931-10934.
Connolly DT, Heuvelman DM, Nelson R, Olander JV, Eppley BL, Delfino JJ, Siegel NR, Leimgruber RM, Feder J. Tumor vascular permeability factor stimulates endothelial cell growth and angiogenesis. J Clin Invest. 1989;84:1470-1478.
Connolly DT, Olander JV, Heuvelman D, Nelson R, Monsell R, Siegel N, Haymore BL, Leimgruber R, Feder J. Human vascular permeability factor: isolation from U937 cells. J Biol Chem. 1989;264:20017-20024.
Leung DW, Cachianes G, Kuang WJ, Goeddel DV, Ferrara N. Vascular endothelial growth factor is a secreted angiogenic mitogen. Science. 1989;246:1306-1309.
Jakeman LB, Winer J, Bennett GL, Altar A, Ferrara N. Binding sites for vascular endothelial growth factor are localized on endothelial cells in adult rat tissues. J Clin Invest. 1992;89:244-253.
Gospodarowicz D, Abraham JA, Schilling J. Isolation and characterization of a vascular endothelial cell mitogen produced by pituitary-derived folliculo stellate cells. Proc Natl Acad Sci U S A. 1989;86:7311-7315.
Conn G, Soderman DD, Schaeffer MT, Wile M, Hatcher VB, Thomas KA. Purification of a glycoprotein vascular endothelial cell mitogen from a rat glioma-derived cell line. Proc Natl Acad Sci U S A. 1990;87:1323-1327.
Adair TH, Gay WJ, Montani JP. Growth regulation of the vascular system: evidence for a metabolic hypothesis. Am J Physiol. 1990;259:R393-R404.
Knighton DR, Hunt TK, Scheuenstuhl H, Halliday BJ, Werb Z, Banda MJ. Oxygen tension regulates the expression of angiogenesis factor by macrophages. Science. 1983;221:1283-1285.
Kourembanas S, Hannan RL, Faller DV. Oxygen tension regulates the expression of the platelet-derived growth factor-B chain gene in human endothelial cells. J Clin Invest. 1990;86:670-674.
Kourembanas S, Marsden PA, McQuillan LP, Faller DV. Hypoxia induces endothelin gene expression and secretion in cultured human endothelium. J Clin Invest. 1991;88:1054-1057.
Rocha-Singh KJ, Honbo NY, Karliner JS. Hypoxia and glucose independently regulate the β-adrenergic receptor-adenylate cyclase system in cardiac myocytes. J Clin Invest. 1991;88:204-213.
Semenza GL, Roth PH, Fang H-M, Wang GL. Transcriptional regulation of genes encoding glycolytic enzymes by hypoxia-inducible factor 1. J Biol Chem. 1994;269:23757-23763.
Semenza GL, Wang GL. A nuclear factor induced by hypoxia via de novo protein synthesis binds to the human erythropoietin gene enhancer at a site required for transcriptional activation. Mol Cell Biol. 1992;12:5447-5454.
Luckow B, Schutz G. CAT constructions with multiple unique restriction sites for the functional analysis of eukaryotic promoters and regulatory elements. Nucleic Acids Res. 1987;15:5490.
Tischer E, Mitchell R, Hartman T, Silva M, Gospodarowicz D, Fiddes JC, Abraham JA. The human gene for vascular endothelial growth factor. J Biol Chem. 1991;266:11947-11954.
Edlund T, Walker MD, Barr PJ, Rutter WJ. Cell-specific expression of the rat insulin gene: evidence for role of two distinct 5′ flanking elements. Science. 1985;230:912-916.
Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1989.
Schreiber E, Matthias P, Muller MM, Schaffner W. Rapid detection of octamer binding proteins with ‘mini-extracts,’ prepared from a small number of cells. Nucleic Acids Res. 1989;17:6419.
Wang GL, Semenza GL. Characterization of hypoxia-inducible factor 1 and regulation of DNA binding activity by hypoxia. J Biol Chem. 1993;268:21513-21518.
Wang GL, Semenza GL. General involvement of hypoxia-inducible factor 1 in transcriptional response to hypoxia. Proc Natl Acad Sci U S A. 1993;90:4304-4308.
Brogi E, Wu T, Namiki A, Isner JM. Indirect angiogenic cytokines upregulate VEGF and bFGF gene expression in vascular smooth muscle cells, whereas hypoxia upregulates VEGF expression only. Circulation. 1994;90:649-652.
Goldberg MA, Schneider TJ. Similarities between the oxygen-sensing mechanisms regulating the expression of vascular endothelial growth factor and erythropoietin. J Biol Chem. 1994;269:4355-4359.
Ausserer WA, Bourrat-Floeck B, Green CJ, Laderoute KR, Sutherland RM. Regulation of c-jun expression during hypoxic and low-glucose stress. Mol Cell Biol. 1994;14:5032-5042.