To whom correspondence should be addressed: Berrie Research Pavilion, Rm. 412C, Inst. for Cancer Genetics, Columbia University, 1150 St. Nicholas Ave., New York, NY 10032. Tel.: 212-851-5282 (office) or 212-851-5285/5286 (laboratory); Fax: 212-851-5284
* This work was supported in part by grants from the Leukemia & Lymphoma Society, Avon Foundation, the Stewart Trust, the Irma T. Hirschl Trust, and National Institutes of Health/NCI (to W. G.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The p53 tumor suppressor is maintained at low levels in normal cells by Mdm2-mediated degradation and strongly stabilized in response to various types of stress including hypoxia. Although hypoxia-inducible factor 1α (HIF-1α) has been implicated to be involved in p53 stabilization, the precise mechanism by which HIF-1α regulates p53-mediated function remains unknown. Here, we found that HIF-1α directly binds Mdm2 both in vitro andin vivo; in contrast, p53 fails to directly interact with HIF-1α in vitro. Interestingly, Mdm2 expression can significantly enhance the in vivo association between p53 and HIF-1α, indicating that Mdm2 may act as a bridge and mediate the indirect interaction between HIF-1α and p53 in cells. Furthermore, HIF-1α protects p53 degradation mediated by Mdm2, and leads to activation of p53-mediated transcription in cells. To elucidate the mechanism of HIF-1α-mediated effect, we also found that HIF-1α can significantly suppress Mdm2-mediated p53 ubiquitination in vitro and blocks Mdm2-mediated nuclear export of p53. These results have significant implications regarding the molecular mechanism by which p53 is activated by HIF-1α in response to hypoxia.
Cellular hypoxia is an important phenomenon in developmental biology, normal physiology, and many pathological conditions, including cancer. Hypoxia triggers a multifaceted adaptive response that is primarily mediated by the heterodimeric transcription factor hypoxia-inducible factor (HIF).
). On the other hand, HIF-1α specifically mediates hypoxic responses. In normoxia, HIF-1α is maintained at low and often undetectable levels. HIF-1α is targeted for degradation by the ubiquitination-proteasome pathway through directly binding to the von Hippel-Lindau tumor suppressor gene (pVHL), which forms the recognition component of an E3 ubiquitin-protein ligase leading to ubiquitination of HIF-1α (
). Recent reports demonstrate that HIF-1α undergoes an iron- and oxygen-dependent modification before it can interact with pVHL. This modification is catalyzed by a specific family of enzymes termed HIF-1α-proline hydroxylases (
). During hypoxia, this specific hydroxylase is inactive since it requires dioxygen for its activity. As a result, HIF-1α accumulates due to the failure of pVHL to recognize its non-hydroxylated form. HIF-1α then translocates to the nucleus and dimerizes with the constitutively present HIF-1β (
The importance of the HIF-1α response pathway in human tumorigenesis is underscored by the finding that HIF-1α is overexpressed in multiple human cancers, because tumor cells, unlike normal cells from the same tissue, are often chronically hypoxic (
). Wild-type p53 is expressed at low levels in most cells because of its short half-life under normal conditions. In contrast, the p53 protein is stabilized, and its level increases in response to various stresses such as DNA damage, hypoxia, and inappropriate oncogene signaling (
). In its active form, p53 can bind DNA in a sequence-specific manner and activate transcription of target genes. p53 levels are regulated in large part by Mdm2, the product of a p53-inducible gene. Mdm2 can interact with the N terminus of p53, which also contains the major acidic transcriptional activation domain. The interaction between Mdm2 and p53 can inhibit p53 transcriptional activity by interfering with the ability of p53 to contact transcriptional coactivators such as p300/CBP (
). However, the molecular mechanism by which HIF-1α stabilizes p53 remains unknown. Moreover, it is not clear whether HIF-1α interacts with p53 directly despite previous indications that p53 associated with HIF-1α in cells. Nevertheless, an array of immobilized peptide assay showed that the core domain of p53 has an affinity for the oxygen-dependent degradation domain of HIF-1α (
). As described in this report, we have found a strong interaction between HIF-1α and Mdm2, while we failed to detect any direct interaction of p53 with HIF-1α. Our data demonstrate that HIF-1α regulates p53 activity including stability and nuclear export through interactions with Mdm2. These results provide a potential mechanism for p53 stabilization by HIF-1α in response to hypoxia.
Cell Culture and Transfection
p53-null H1299 lung carcinoma cells and mouse embryonic fibroblast cells (MEFs) were cultured in Dulbecco's modified Eagle's medium (Mediatech, Richmond, VA) supplemented with penicillin/streptomycin and 10% fetal bovine serum (Mediatech, VA).
1 × 106 or 2 × 105 cells were plated in 10-cm or six-well plates, respectively, and 24 h later transfections were done by calcium phosphate precipitation procedures. After a 24-h incubation with 20% O2, the cells were harvested.
Full-length HIF-1α DNA was generated by PCR using hemagglutinin-tagged HIF-1α from David Livingston (Dana Farber Cancer Institute) and was subcloned into pcDNA3.1/v5-His-Topo (Invitrogen) or p3×FLAG-CMV-14 (Sigma). Green fluorescence protein (GFP)-tagged wild-type p53 was obtained from Yanping Zhang (M. D. Anderson Cancer Center). Plasmid DNA for transfections was isolated using Qiagen plasmid maxi kit (Qiagen).
Recombinant Protein Preparation and Glutathione S-Transferase (GST) Pull-down Assay
GST fusions of p53 and Mdm2 were expressed in Escherichia coli BL-21 (DE3) (Promega) and induced with 0.1 mmisopropyl-1-thio-β-d-galactopyranoside. Bacterial pellets were lysed in BC500 (25 mm Tris, pH 7.8, 500 mmNaCl, 1 mm EDTA, 1 mm DTT, 10% glycerol, 0.2% Nonidet P-40, fresh 1 mm PMSF) with sonication. Levels of expressed GST fusion proteins were estimated by incubation with glutathione-Sepharose beads, washing, and quantification by SDS-PAGE followed by staining with Coomassie Brilliant Blue R-250. Known amounts of bovine serum albumin were used as standards.
35S-Labeled in vitro translated HIF-1α was prepared by using the TNT system (Promega). Equal amounts (1 μg) of GST, GST-p53, and GST-Mdm2 immobilized on glutathione-Sepharose beads were incubated with in vitro translated HIF-1α in BC200 for 3 h at 4 °C. After washing, the bound proteins were eluted with SDS sample buffer and were separated by SDS-PAGE, followed by autoradiography.
Immunoblotting and Co-immunoprecipitation
Cells were lysed in FLAG lysis buffer (50 mm Tris, 137 mm NaCl, 10 mm NaF, 1 mm EDTA, 1% Triton X-100, 0.2% Sarkosyl, 1 mm DTT, 10% glycerol, pH 7.8) with fresh protease inhibitors (1 mm PMSF, protease inhibitor mixture (Sigma)). Aliquots (30 μg) of cell extracts were resolved in SDS, 8% polyacrymide gels and then transferred to nitrocellulose membranes in 20 mm Tris-HCl, pH 8.0, 150 mm glycine, 20%(v/v) methanol. Membranes were blocked with 5% (v/v) nonfat dry milk, TBST (20 mm Tris-HCl, pH 7.6, 137 mmNaCl, 0.1% Tween 20), incubated with α-p53 (DO-1) antibody (Santa Cruz), or anti-GFP (Clontech), and detected with ECL reagents (Amersham Biosciences).
Coimmunoprecipitation assay was performed essentially as described previously (
). In brief, 50 μl of proteasome inhibitorN-acetyl-leucyl-leucyl-norleucinal was added to the cotransfected culture 6 h before harvest. 24 h after transfection, the cells were lysed in BC100 (25 mm Tris, pH 7.8, 100 mm NaCl, 1 mm EDTA, 1 mmDTT, 10% glycerol, 0.2% Nonidet P-40, fresh 1 mm PMSF) and incubated with anti-FLAG M2 beads (Sigma) overnight at 4 °C. The beads were washed five times with 1 ml of lysis buffer, after which the associated proteins were eluted with BC100, 0.2% Nonidet P-40 plus 0.2 mg/ml FLAG peptide (Sigma). The eluted proteins were resolved on 8% SDS-PAGE and Western blot with the anti-p53 or anti-Mdm2 (SMP14, Santa Cruz) monoclonal antibody.
Luciferase activity was determined using a dual luciferase assay system (Promega) following the manufacturer's protocol. Cells in six-well plates were removed by scraping into 100 μl of reporter lysis buffer. Cell lysate was collected by centrifugation for 15 min at 12,000 × g. Luciferase activity was measured using a Lumat LB 9507 luminometer (EG&G Wallac, Gaithersburg, MD).
In Vitro Ubiquitination Assays
The in vitroubiquitination assay was performed as previously described (
) with some modifications. For a standard reactions, the purified FLAG-p53 proteins from H1299 cells were mixed with other purified components, including E1, E2 (GST-UbcH5C), E3 (GST-Mdm2), and His-ubiquitin in reaction buffer (40 mm Tris, 5 mmMgCl2, 2 mm ATP, 2 mm DTT, pH 7.6). The reaction was stopped after 60 min at 37 °C by additions of SDS sample buffer and subsequently resolved SDS-PAGE gels for Western blot analysis with α-p53 (DO-1).
Nuclear Export Assay for p53
H1299 cells were plated on six-well-containing glass coverslips, and GFP-p53 was transfected as described. 50 μm proteasome inhibitorN-acetyl-leucyl-leucyl-norleucinal was added for 6 h before fixation. Twenty-four hours after transfection, cells on the coverslips were washed three times with phosphate-buffered saline (PBS) and then fixed in 4% paraformaldehyde for 10 min at room temperature. After fixation, cells were washed in PBS three times and then permeabilized in ice-cold PBS containing 0.2% Triton X-100 for 10 min. Cells were blocked in PBS containing 1% bovine serum albumin and 1 μg of DAPI (Sigma)/ml at room temperature for 30 min. Cells were washed three times with PBS, and the stained cells were mounted with mounting medium (Polysciences, Inc., Warrington, PA) and sealed with nail polish. Immunofluorescence was recorded using an immunofluorescence microscope.
Interaction between HIF-1α and Mdm2
Although earlier studies indicated that p53 can bind HIF-1α in cells, we repeatedly failed to detect any direct interaction between these two protein in a GST pull-down assay (data not shown), suggesting that HIF-1α may interact with p53 through other factors. To examine the notion that p53 may interact with HIF-1α through Mdm2, we tested whether HIF-1α directly binds Mdm2 in vitro by GST pull-down assays. Both GST-Mdm2 and GST were expressed in bacteria and purified to near homogeneity. As shown in Fig.1a, the35S-labeled in vitro translated HIF-1α strongly bound to immobolized GST-Mdm2 (lane 2) but not to immobilized GST (lane 3). However, using the same assay, no significant interaction was detected between GST-p53 and HIF-1α. Furthermore, we tested the interaction between HIF-1α and Mdm2in vivo by coimmunoprecipitation assays. FLAG-tagged full-length HIF-1α was transiently co-transfected with Mdm2 in H1299 cells. Immunoprecipitations were performed with anti-Flag M2 beads, and the precipitated proteins were analyzed by Western blot with an anti-Mdm2 antibody. As shown in Fig. 1b, the Mdm2-HIF-1α complexes were readily detected in cells cotransfected with FLAG-HIF-1α and Mdm2 (lane 4) but not by Mdm2 alone (lane 3), indicating that there is a specific in vivo interaction between Mdm2 and HIF-1α.
Mdm2 Enhances the in Vivo Binding between p53 and HIF-1α
To examine the possibility that the HIF-1α-p53 interaction is mediated by Mdm2, we tested whether Mdm2 expression is required for the HIF-1α-p53 interaction in cells. p53 and FLAG-HIF-1α were transiently transfected with or without Mdm2 in H1299 cells. Immunoprecipitations were carried out from cell extracts using anti-Flag M2 beads. As shown in Fig. 2, p53 was barely detectable in the HIF-1α-associated immunocomplexes in the absence of Mdm2 expression (lane 5), further confirming the idea that p53 cannot directly bind HIF-1α in vivo. In contrast, when Mdm2 was expressed in the cells, p53 was efficiently coprecipitated with HIF-1α by the same method (lane 6). These results indicate that Mdm2 acts as a bridge mediating the binding between HIF-1α and p53.
HIF-1α Abrogates Mdm2-mediated p53 Degradation and Transcription Repression
Mdm2 is a key regulator of p53 and can both inhibit p53 transcriptional activity and target it for degradation (
). We tested the possibility whether HIF-1α could protect p53 from degradation mediated by Mdm2. The p53-null H1299 cells were transfected with CMV-p53, CMV-Mdm2, CMV-GFP, and pTOPO-HIF-1α. 24 h after transfection the cells were lysed in a FLAG lysis buffer and analyzed by Western blot. As shown in Fig.3a, HIF-1α effectively rescues p53 from Mdm2-mediated degradation. Overexpression of Mdm2 significantly induced p53 degradation (lane 2 versus lane 1), whereas this degradation was inhibited in a dose-dependent manner upon overexpression of HIF-1α (lanes 3 and 4 versus lane 2). Interestingly, HPV E6 protein can induce p53 degradation through the E6/E6 AP ubiquitin ligase complex. However, overexpression of HIF-1α cannot protect p53 degradation mediated by E6 (Fig. 3b). Taken together, these results demonstrated an effect of HIF-1α directly on Mdm2-mediated degradation of p53. Furthermore, we used p53 transcriptional activity assay to support the notion. To this end, we cotransfected MEFs (p53−/−) with vectors expressing p53, Mdm2, and HIF-1α, along with a reporter construct containing the minimal p21 promoter (p21min-luc). As indicated in Fig. 3c, cotransfection of Mdm2 with p53 strongly repressed p21 luciferase activity due to p53 degradation. However, HIF-1α significantly abrogated the inhibitory effect of Mdm2 in a dose-dependent manner. The above data demonstrate that HIF-1α can strongly stabilize p53 by abrogating Mdm2-mediated effects, which leads to activation of p53-mediated transcription.
HIF-1α Suppresses p53 Ubiquitination Mediated by Mdm2
To investigate whether HIF-1α can suppress p53 ubiquitination mediated by Mdm2, we carried out an in vitro ubiquitination assay. As shown in Fig. 3d, ubiquitinated p53 was easily detected at the reaction with Mdm2 but no HIF-1α (lane 2 versus lane 1). However, in the presence of HIF-1α, the levels of ubiquitinated p53 decreased dramatically (lane 3 versus lane 2). This result demonstrates that HIF-1α can significantly suppress Mdm2-mediated p53 ubiquitination.
HIF-1α Blocks Nuclear Export Mediated by Mdm2
Current models indicate that Mdm2 can also induce nuclear export of p53 (
). To test the possibility that HIF-1α can block this export, we transfected H1299 cells with vectors expressing GFP-p53, Mdm2 or HIF-1α and examined them by immunofluorescence microscope (Fig.4a). The distribution of GFP-p53 was almost entirely nuclear when expressed alone, but it localized to the cytoplasm to varying extents when coexpressed with Mdm2. We scored the extent by which p53 localized to the cytoplasm when expressed alone or when expressed together with Mdm2 (Fig.4b). Cells cotransfected with p53 and Mdm2 had a 76% cytoplasmic expression versus 24% when p53 was transfected alone, indicating that nuclear Mdm2 can promote the cytoplasmic localization of p53 (
). Interestingly, only 26% of cytoplasmic expression occurred in cells cotransfected with HIF-1α in addition to p53 and Mdm2 (Fig. 4, a and b). These results indicate that HIF-1α can strongly block Mdm2-mediated p53 nuclear export.
How cells sense changes in ambient oxygen is a central question in biology. In mammalian cells, lack of oxygen, or hypoxia, leads to stabilization of a sequence-specific DNA binding transcriptional factor called HIF. Downstream genes of HIF are linked to processes such as angiogenesis and glucose metabolism (
); however, the nature of this interaction has not been elucidated. In this study, we demonstrate for the first time that HIF-1α directly binds to Mdm2 both in vitro and in vivo. In the absence of Mdm2, the binding between HIF-1α and p53 is almost undetectable. Furthermore, Mdm2 expression significantly induces the indirect interaction between p53 and HIF-1α in cells, indicating that Mdm2 may act as a bridge mediating the p53-HIF-1α interaction. In this regard, we failed to detect any significant interaction between p53 and HIF-1α in MEF Mdm2−/− cells, and stabilization of p53 induced by HIF-1α expression was also severely abrogated (data not shown). Our findings seem at odds with the recent report of tight binding between the ODD domain of HIF-1α and the core domain of p53 by an array of immobilized peptide assay (
). It is likely that the experiment was based on p53 pepetide fragments, which may not be equivalent to the native, folded protein. However, the existence of Mdm2 somehow might cause the conformational change of p53 native protein, thereby favoring the exposure of binding sites of p53 to HIF-1α.
In addition, our results demonstrate that HIF-1α protects p53 from degradation mediated by Mdm2 and can abrogate p53 transcriptional repression by Mdm2. Considering that p53 degradation is mainly induced by Mdm2 in normal cells, we also found that Mdm2-mediated ubiqutination of p53 is significantly inhibited by HIF-1α. Since Mdm2-mediated p53 ubiquitination promotes its nuclear export (
), we further demonstrate that HIF-1α expression can block Mdm-2-mediated nuclear export of p53. Thus, these results have significant implications regarding the molecular mechanism by which HIF-1α modulates p53 function in vivo.
Our results are also consistent with published results indicating HIF-1α interacts with the wild-type p53 protein but not the tumor-derived p53 mutant form in cells (
). Since wild-type p53 protein is capable of inducing Mdm2 expression in cells, the observed interaction between p53 and HIF-1α is most likely mediated by endogenous Mdm2. In contrast, because the tumor-derived p53 mutant is completely inactive in transcriptional activation of endogenous Mdm2, HIF-1α failed to interact with p53 since there is no (or very low levels) Mdm2 in cells expressing mutated p53. Recent studies also indicate that Mdm2 is involved in modulating HIF-1α stability under hypoxic conditions (
), further supporting the notion that HIF-1α directly interacts with Mdm2 but not p53. Mdm2 is a potent E3 ubiquitin ligase and induces both p53 degradation and nuclear export of p53 through ubiquitination. Therefore, it is possible that Mdm2 also directly mediates degradation and nuclear export of HIF-1α. Our study, together with several other studies (
), strongly implicates the important role of HIF-1α in the regulation of the p53-Mdm2 pathway in response to hypoxia.
We thank C. Brooks for carefully reading the manuscript and other members of W. Gu laboratory for sharing unpublished data and critical comments. We thank David Livingston and Yanping Zhang for providing plasmids.