Hypoxia-inducible factor 1 (HIF-1) activates transcription of genes encoding glucose transporters, glycolytic enzymes, vascular endothelial growth factor, and other proteins involved in O2 homeostasis and tumor progression. The expression and transcriptional activity of the HIF-1α subunit is regulated by the cellular O2 concentration. We demonstrate that insulin, insulin-like growth factor (IGF)-1, and IGF-2 induce expression of HIF-1α, which is required for expression of genes encoding IGF-2, IGF-binding protein (IGFBP)-2 and IGFBP-3. These data provide a novel mechanism by which HIF-1α overexpression may occur in tumor cells and contribute to an autocrine growth factor loop.
Complex cellular and systemic regulatory mechanisms control O2, glucose, and energy homeostasis. A key regulator of O2 homeostasis is HIF-1, 3 a basic-helix-loop-helix-PAS transcription factor (1) . HIF-1 activates transcription of genes whose protein products either: (a) increase O2 availability; or (b) promote metabolic adaptation to hypoxia. Examples include (a) erythropoietin and VEGF, which control erythropoiesis and angiogenesis, respectively; and (b) glucose transporters and glycolytic enzymes, which are up-regulated to produce adequate ATP in the absence of oxidative phosphorylation (reviewed in Ref. 2 ). HIF-1 is a heterodimer of HIF-1α and HIF-1β subunits (1) . The expression and activity of HIF-1α are regulated by the cellular O2 concentration and determine the transcriptional activity of HIF-1 (2) . Hypoxia-inducible genes regulated by HIF-1 contain a cis-acting HRE that includes a binding site for HIF-1, which has the consensus sequence 5′-RCGTG-3′ (2) . In mouse ES cells in which HIF-1α expression was eliminated by targeted mutation, expression of genes encoding glucose transporters, glycolytic enzymes, and VEGF was markedly reduced (3, 4, 5) .
Other than hypoxia, known inducers of HIF-1 activity are divalent cations such as cobalt chloride (CoCl2) and iron chelators such as desferrioxamine, which act by undetermined mechanisms (1 , 2) . Exposure of cultured cells to the organomercurial compound mersalyl was recently shown to induce expression of HIF-1α protein, HIF-1 DNA-binding activity, and downstream genes encoding VEGF and enolase 1 (6) . HIF-1α protein and HIF-1 DNA-binding activity were induced by mersalyl in wild-type MEFs but not in MEFs in which expression of the gene encoding the IGF-1R was eliminated by targeted mutation. In contrast, HIF-1 activity was induced in IGF-1R-null cells exposed to hypoxia, CoCl2, or desferrioxamine. Furthermore, induction of HIF-1 by mersalyl was inhibited by treatment with PD098059, an inhibitor of mitogen-activated protein kinase kinase activity, whereas induction by hypoxia was unaffected (6) . These results indicated that expression of HIF-1 and downstream genes could be induced via IGF-1R and mitogen-activated protein kinase activity. These conclusions were supported by studies demonstrating that exposure of cultured cells to insulin induced expression of HIF-1 DNA-binding activity, genes encoding glucose transporters and glycolytic enzymes, and reporter genes containing a HRE (7) . These responses were observed in hepatoma cell lines that lacked IGF receptors, which suggests that the responses to insulin were mediated by another pathway. In this paper, we report that exposure of cells to insulin, IGF-1, or IGF-2 results in the induction of HIF-1α protein expression. In addition, we demonstrate that HIF-1α is required for expression of mRNAs encoding IGF-2 and the IGF binding proteins IGFBP-2 and IGFBP-3. These results indicate complex cross-talk between the HIF-1 and IGF systems that has important implications for the understanding of cellular energy metabolism, growth control, and tumor progression.
Materials and Methods
The Igf1r+/+ and Igf1r−/− MEF cell lines W and R− (8) were provided by Dr. Renato Baserga (Thomas Jefferson University, Philadelphia, PA) and were maintained in DMEM (Mediatech) supplemented with 10% FBS (Life Technologies, Inc.). Hif1a+/+ and Hif1a−/− ES cells were generated as described previously (4) and maintained in high-glucose DMEM supplemented with 10% FBS, 0.1 mm nonessential amino acids, and 10 μg/ml bovine insulin. To generate MEFs, E9.25 embryos were isolated from timed matings of Hif1a+/− males and females and genotyped by PCR (4) . Individual embryos were rinsed with Ringer’s solution to remove any adherent maternal cells, incubated in trypsin, disrupted by pipetting to generate a single cell suspension, cultured in high-glucose DMEM containing 15% FBS and 0.1 mm nonessential amino acids for 48 h, and then transfected with the plasmid pOT (provided by Dr. Randall Johnson, University of California, San Diego, CA), which contains the SV40 early region encoding T antigen (9) , using Lipofectamine-Plus reagent (Life Technologies, Inc.). From a single Hif1a+/+ embryo, a pool of eight colonies was expanded to form the N3 line. From a Hif1a−/− embryo, a pool of 12 colonies was expanded to form the M5 line.
Cells were subjected to hypoxia in a modular incubator chamber (Billups-Rothenberg) flushed with 1% O2, 5% CO2, and balance N2. For growth factor stimulation, cells were cultured in 0.1% FBS for 48 h and then exposed to bovine insulin (Life Technologies, Inc.), recombinant human EGF (Sigma Chemical Co.), basic FGF-2 (Sigma Chemical Co.), IGF-1, or IGF-2 (Calbiochem) by the addition of a 1000× aqueous stock solution containing 0.1% BSA. Control (untreated) cells received 0.1% BSA only.
Nuclear extracts were prepared and aliquots were fractionated by 7% SDS-PAGE and transferred to a nitrocellulose membrane (1 , 4) . Blots were incubated with a 1:1000 dilution of anti-HIF-1α monoclonal antibody H1α67, which was protein G-purified from ascites fluid (Novus Biologicals, Inc.), 4 followed by sheep antimouse immunoglobulin (1:2000) and ECL reagent (Amersham).
RNA Blot Hybridization.
Total RNA was isolated by guanidine isothiocyanate extraction, and 10-μg aliquots were fractionated by 1.4% agarose-2.2 m formaldehyde gel electrophoresis, transferred to a nylon membrane, and hybridized with a 32P-cDNA probe in Quik-Hyb (Stratagene; Ref. 4 ). cDNA probes were provided by Dr. Cunming Duan, University of Michigan (IGFBP-2, -4, -5; Ref. 10 ) and Dr. Amato Giaccia, Stanford University; (IGF-1; Ref. 11 ) or were purchased from Research Genetics (GenBank accession number listed): (a) IGF-1 (AA277619); (b) IGF-2 (AA259833); (c) IGFBP-2 (AA712031); (d) IGFBP-3 (W89951); and (e) IGFBP-4 (AA289121).
We previously demonstrated that exposure of cells to the organomercurial compound mersalyl induced expression of HIF-1α in cells that contained an intact IGF-1R but failed to do so in cells lacking IGF-1R (6) . We tested whether the endogenous ligands of IGF-1R could also induce HIF-1α expression in cultured cells. Human embryonic kidney 293 cells were serum-starved for 48 h and then exposed to 6.8 nm insulin (40 ng/ml), IGF-1, or IGF-2 for 6 h; nuclear extracts were prepared, and subjected to immunoblot assay using a monoclonal antibody specific for HIF-1α (Fig. 1A) ⇓ . Compared with untreated cells (Lane 4), HIF-1α protein levels were increased in growth factor-treated cells (Lanes 1–3) and cells exposed to 1% O2 for 6 h (Lane 5). HIF-1α expression was induced when wild-type (W) or IGF-1R-deficient (R−) MEFs were exposed to hypoxia, insulin, or IGF-1 (Fig. 1B) ⇓ . Thus, in contrast to mersalyl, IGFs and insulin can use receptors other than, or in addition to, IGF-1R to induce HIF-1α expression.
To determine whether other growth factors induce HIF-1α expression, serum-starved cells were exposed to EGF or FGF-2. A dose-dependent increase in HIF-1α expression was observed over a three-log concentration range for both EGF and FGF-2 (Fig. 2) ⇓ , indicating that members of multiple growth factor families can induce HIF-1α expression in cultured human cells.
To determine whether binding of growth factors to their receptors was sufficient to induce HIF-1α expression, we analyzed the effect of cell density. Equal numbers of wild-type MEFs were plated and incubated in the presence of serum for 24, 48, or 72 h before serum starvation and exposure to insulin. HIF-1α expression was induced in cells cultured for 24 h but not in cells cultured for 48 or 72 h (Fig. 3) ⇓ . Thus, insulin induced HIF-1α expression only under conditions of low cell density when cells were most actively proliferating (data not shown), which suggests that HIF-1α may be expressed as part of a mitogenic response pathway.
The transcription of several genes encoding IGFs and IGFBPs is induced by hypoxia in certain cell types (11, 12, 13, 14) . We analyzed the expression of IGF-1, IGF-2, IGFBP-1, IGFBP-2, IGFBP-3, IGFBP-4, and IGFBP-5 mRNA in ES cells and MEFs that were either wild-type or homozygous for a targeted mutation that resulted in complete absence of HIF-1α expression (4) . Expression of IGF-1, IGFBP-1, and IGFBP-5 was not detected in the ES or MEF cells (data not shown). Expression of IGFBP-3 and IGFBP-4 mRNA was detected only in ES cells. IGFBP-3 mRNA expression was induced by hypoxia in wild-type ES cells whereas in mutant ES cells, no IGFBP-3 mRNA was detected (Fig. 4) ⇓ . Neither IGF-2 nor IGFBP-2 mRNA expression was induced by hypoxia in wild-type ES cells or MEFs, but expression was dramatically reduced in HIF-1α-deficient cells. IGFBP-4 mRNA expression was not affected by O2 concentration or HIF-1α deficiency. Thus, IGF-1 and IGF-2 induce expression of HIF-1α, which in turn regulates the expression of genes encoding IGF-2, IGFBP-2, and IGFBP-3.
We have demonstrated that insulin, IGF-1, IGF-2, EGF, and FGF-2 each induces HIF-1α expression in cultured cells. The observation that HIF-1α expression is stimulated by exposure of cells to insulin, IGF-1, or IGF-2 suggests that HIF-1 may play a previously unrecognized role in cellular glucose and energy metabolism in addition to its well-established role in O2 homeostasis. Considering the interrelatedness of these physiological processes and previous data demonstrating the regulation of genes encoding glucose transporters and glycolytic enzymes by HIF-1 (4 , 5) , it is not surprising that they would be integrated at the transcriptional level. Although the signaling pathway remains undefined, the results presented here and elsewhere (7) indicate that in addition to IGF-1R, activation of the insulin receptor or IGF-2R also induces HIF-1α expression.
The observation that all of the growth factors tested stimulated HIF-1α expression, particularly under conditions of low cell density, suggests that HIF-1α also plays an important role in cellular proliferation. Several previous studies support this hypothesis:
(a) HIF-1α-deficient ES cells manifested reduced rates of cellular proliferation relative to wild-type cells under nonhypoxic and, to an even greater extent, under hypoxic culture conditions (4) ;
(b) PC-3 and LNCaP human prostate cancer cells showed increased expression of HIF-1α under both nonhypoxic and hypoxic conditions when cultured at low density as compared with high density (15) ;
(c) in a recent immunohistochemical analysis of human cancers, a strong statistical correlation was observed between HIF-1α overexpression and cellular proliferation as measured by the Ki67 labeling index.
HIF-1α expression in proliferating cells may be important for the transcriptional activation of genes encoding glucose transporters and glycolytic enzymes to allow increased nonoxidative generation of ATP and reduced mitochondrial generation of reactive oxygen species that might otherwise result in damage to replicating DNA (16) . Interestingly, compared with W cells, R− cells proliferated at a higher rate under standard culture conditions (data not shown) and showed higher levels of HIF-1α expression under both noninduced and induced conditions (Fig. 1) ⇓ .
The observation that HIF-1α was required for maximal expression of genes encoding IGF-2, IGFBP-2, and IGFBP-3 establishes a positive reciprocal relationship between the HIF-1 and IGF systems. The expression of IGF-2, IGFBP-2, and IGFBP-3 was previously shown to be induced by hypoxia in bovine pulmonary artery endothelial cells (14) , human hepatoma cells (13) , and neonatal rat retina (12) , respectively. The data presented in this paper implicate HIF-1 as a transcriptional activator of these genes, although in the absence of a defined HRE containing a HIF-1 binding site it remains possible that HIF-1α deficiency has an indirect effect on the expression of one or more of these genes. However, hypoxia-induced IGFBP-1 mRNA expression in hepatoma cells was mediated by a HRE containing a putative HIF-1 site (11) , providing a precedent for direct regulation. IGFBP-1 mRNA expression was not detected in the ES and MEF cell lines studied (data not shown). Expression of IGFBP-4 mRNA was not induced by hypoxia, in agreement with previous studies (14) , and was not affected by HIF-1α deficiency, demonstrating that HIF-1 regulates a subset of IGF family members in a cell-type-specific manner.
The positive reciprocal relationship between HIF-1 and IGF-2 may be of considerable importance in tumor progression. IGF-2 was the most highly up-regulated gene in colon cancer (17) , and colon cancers showed the highest frequency and degree of HIF-1α overexpression among the 19 different types of human cancer assayed by immunohistochemistry. 4 Although IGFBP-3 has antiproliferative effects, these can be counteracted in cancer cells by overexpression of matrix metalloproteinase-9 (18) . Mutant hepatoma cells that lacked HIF-1 expression showed dramatically reduced rates of xenograft growth and vascularization compared with wild-type parental cells (19 , 20) . HIF-1α is overexpressed in human breast, colon, prostate, and lung cancer, 4 the most common causes of cancer mortality in the United States. The data presented in this paper provide a novel mechanism by which HIF-1α and IGF-2 overexpression may occur in tumor cells and contribute to an autocrine growth factor loop.
We thank Renato Baserga, Cunming Duan, Amato Giaccia, and Randall Johnson for providing cell lines and plasmid clones. We thank Jonathan Simons for helpful discussions.
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.
↵1 G. L. S. is an Established Investigator of the American Heart Association, and this work was supported by grants from the American Heart Association as well as by NIH Grants R01-DK39869 and R01-HL55338.
↵2 To whom requests for reprints should be addressed, at Johns Hopkins Hospital, CMSC-1004, 600 North Wolfe Street, Baltimore, MD 21287-3914. Phone: (410) 955-1619; Fax: (410) 955-0484; E-mail:
↵3 The abbreviations used are: HIF-1, hypoxia-inducible factor 1; VEGF, vascular endothelial growth factor; HRE, hypoxia response element; ES, embryonic stem; MEF, mouse embryo fibroblast; IGF, insulin-like growth factor; IGF-1R, IGF-1 receptor; IGFBP, IGF-binding protein; FBS, fetal bovine serum; EGF, epidermal growth factor; FGF, fibroblast growth factor.
↵4 H. Zhong, A. M. De Marzo, E. Laughner, M. Lim, D. A. Hilton, D. Zagzag, P. Buechler, W. B. Isaacs, G. L. Semenza, and J. W. Simons. HIF-1α is overexpressed in common human cancers and their metastases, submitted for publication.
- Received May 12, 1999.
- Accepted June 30, 1999.
- ©1999 American Association for Cancer Research.