Carboxyl-Terminal Transactivation Activity of Hypoxia-Inducible Factor 1{alpha} Is Governed by a von Hippel-Lindau Protein-Independent, Hydroxylation-Regulated Association with p300/CBP

Hypoxia-inducible factor 1 complex (HIF-1) plays a pivotal rolein oxygen homeostasis and adaptation to hypoxia. Its functionis controlled by both the protein stability and the transactivationactivity of its alpha subunit, HIF-1 ${alpha}$ . Hydroxylation of at leasttwo prolyl residues in the oxygen-dependent degradation domainof HIF-1 ${alpha}$ regulates its interaction with the von Hippel-Lindauprotein (VHL) that targets HIF-1 ${alpha}$ for ubiquitination and proteasomaldegradation. Several prolyl hydroxylases have been found tospecifically hydroxylate HIF-1 ${alpha}$ . In this report, we investigatedpossible roles of VHL and hydroxylases in the regulation ofthe transactivation activity of the C-terminal activating domain(CAD) of HIF-1 ${alpha}$ . We demonstrate that regulation of the transactivationactivity of HIF-1 ${alpha}$ CAD also involves hydroxylase activity butdoes not require functional VHL. In addition, stimulation ofthe CAD activity by a hydoxylase inhibitor, hypoxia, and desferrioxaminewas severely blocked by the adenoviral oncoprotein E1A but notby an E1A mutant defective in targeting p300/CBP. We furtherdemonstrate that a hydroxylase inhibitor, hypoxia, and desferrioxaminepromote the functional and physical interaction between HIF-1 ${alpha}$ CAD and p300/CBP in vivo. Taken together, our data provide evidencethat hypoxia-regulated stabilization and transcriptional stimulationof HIF-1 ${alpha}$ function are regulated through partially overlappingbut distinguishable pathways.

INTRODUCTION

Top

Introduction

Lack of oxygen (hypoxia) is a common pathological cause in manyhuman diseases, such as stroke, coronary disease, and pulmonarydisorders. Maintenance of oxygen homeostasis and adaptationto hypoxia in higher eukaryotes require hypoxia-inducible factor1 (HIF-1), a heterodimeric DNA-binding complex that acts asa key regulator of hypoxia-responsive genes. Among its transcriptionaltargets, HIF-1 activates genes with critical roles in erythropoiesis,angiogenesis, vasomotor function, and intracellular energy metabolism(43). The HIF-1 complex is composed of two subunits (49): HIF-1ß(ARNT), which is constitutively expressed, and HIF-1 ${alpha}$ , whichis constantly translated but rapidly degraded under normoxicconditions. The degradation of HIF-1 ${alpha}$ is mediated by the ubiquitin-proteasomesystem (14, 22, 38) and requires a highly conserved domain,termed the oxygen-dependent degradation domain (ODD) (15, 33,45). Lack of oxygen stabilizes HIF-1 ${alpha}$ , allowing its translocationto the nucleus and its interaction with HIF-1ß toform the HIF-1 complex (8, 21). Genetic knockout studies revealan essential role of HIF-1 ${alpha}$ in hypoxia-mediated apoptosis, cellproliferation, and tumorigenesis (6, 36, 37). Nonhypoxic stabilizationof HIF-1 ${alpha}$ is observed in response to growth factors and cytokinesand also in solid tumors, suggesting a more general role ofHIF-1 in tumorigenesis and inflammation (35, 48, 53).

HIF-1 ${alpha}$ degradation depends on its interaction with the von Hippel-Lindauprotein (VHL) (47). VHL is a tumor suppressor whose defect definesa dominantly inherited cancer syndrome, von Hippel-Lindau syndrome(27). Patients with the VHL syndrome are predisposed to developmultiple hemangioblastomas of the central nerve system and retina,renal cell carcinomas, pheochromocytomas, pancreatic carcinoma,and cysts in the kidneys, liver, and pancreas (17, 24, 50).While the functions of VHL have not been completely characterizedyet, several studies indicate that VHL is a component of anE3 ubiquitin protein-ligase complex that targets HIF-1 ${alpha}$ for ubiquitinationand proteasomal degradation under normoxic condition (9, 18,23, 30, 31, 46). Loss of VHL function leads to stabilizationof HIF-1 ${alpha}$ , providing an explanation for the hypervascular natureof VHL-associated neoplasms.

The interaction between VHL and HIF-1 ${alpha}$ depends on the hydroxylationof proline residues (Pro-402 and Pro-564) in the oxygen-dependentdegradation domain of HIF-1 ${alpha}$ (17, 19, 29, 51). These hydroxylationsare mediated by recently identified HIF-specific prolyl hydroxylases(HIF-PHDs) that require oxygen, 2-oxoglutarate, and iron fortheir full activity (5, 12).

In addition to regulating protein stabilization, hypoxia enhancesthe transactivation activity of HIF-1 ${alpha}$ , thus providing a multistepprocess of regulation (20, 33, 47). Earlier reports have indicatedthat a region adjacent to the transactivation domains inhibitsthe transactivation activity of HIF-1 ${alpha}$ , and this inhibition canbe relieved by hypoxia (20). HIF-1 ${alpha}$ transactivation involvesthe physical interaction with p300/CBP (1, 10, 21), two coactivatorsrequired for the transcriptional activity of a variety of transcriptionfactors, such as MyoD and p53 (2, 34, 41, 52). It has been suggestedthat hypoxia enhances the recruitment of p300/CBP through aredox-mediated mechanism (7, 11, 13, 21, 26).

In this report, we examine the possible role of VHL and hydroxylasesin the regulation of the carboxyl-terminal transactivation activityof HIF-1 ${alpha}$ . Our data indicate that while both protein stabilityand the transactivation activity are regulated by the same initialsignals and involve hydroxylase activity, two partially overlappingbut distinguishable pathways are utilized: one requires VHL,and the other one is VHL independent.

MATERIALS AND METHODS

Top

Materials and Methods

Plasmids and DNA recombination. The pM vector encoding the DNA binding domain (DBD) of yeasttranscription factor GAL4 and pRL-CMV were purchased from Clontech.pGH1 ${alpha}$ 530-778 and pGH1 ${alpha}$ 740-826 were constructed by inserting aPCR fragment encoding this region into pM as described previously(44, 45). pGH1 ${alpha}$ 786-826 was generated by PCR and subcloned intothe pM vector at EcoRI-BamHI sites. pGH1 ${alpha}$ 740-813 was generatedby digesting pGH1 ${alpha}$ 740-826 with PstI followed by self-ligationthat deletes a small DNA fragment encoding the last 13 aminoacids of HIF-1 ${alpha}$ . Plasmids expressing PHD1, PHD2, and PHD3 werea generous gift from P. Ratcliffe (Oxford University, Oxford,United Kingdom) (12). pFR-luc, a luciferase reporter under thecontrol of a promoter with GAL4 binding sites, was purchasedfrom Stratagene. pEpo-luc was described previously (38). pG4CH1(amino acids [aa] 300 to 523 of p300 fused with the Gal4 DNAbinding domain) and pVP16H1 ${alpha}$ 723-826 (aa 723 to 826 fused withVP16) were kindly provided by D. M. Livingston (Harvard MedicalSchool) (4, 25). pCMVßp300 and pSV40-lacZ were describedpreviously (39).

Antibodies and special chemicals. Antibodies against HIF-1 ${alpha}$ (monoclonal; catalog no. 610959), VHL(mouse monoclonal; clone Ig32) and p300 (monoclonal; clone NM11)were purchased from Transduction Lab/Pharmagen. Monoclonal anti-Gal4-DBDantibodies were purchased from Clontech (catalog no. 5399-1)or Santa Cruz (catalog no. SC-510). Monoclonal anti-E1A antibodyM73 was a gift from A. Giordano and was described previously(42). Desferrioxamine and cobaltous chloride were purchasedfrom Sigma. Dimethyloxalylglycine (DOG) and anti-HIF-2 ${alpha}$ werea generous gift from Ratcliffe (Oxford University) (19, 30).Horseradish-peroxidase (HRP)-labeled goat anti-mouse immunoglobulinG (IgG) or goat anti-rabbit IgG antibodies were purchased fromGIBCO/Invitrogen. HRP-labeled goat anti-mouse IgG Fc fragmentswere purchased from Sigma.

Cell lines, tissue culture, and transfection. HeLa, Hep3B, 143B, and the VHL-deficient 786-O cell line werepurchased from American Type Culture Collection. 786-O#52, expressingwild-type VHL under the Tet-off* control system (Clontech),and 786-O#126, transfected with the backbone vector lackingVHL, were a generous gift from N. Kley (Bristol Myers Squibb).HeLa, 143B, 786-O, and derivative lines were cultured in Dulbecco'smodified Eagle medium (DMEM) (Mediatech) supplemented with 10%fetal bovine serum (HyClone), 100 U of penicillin/ml, and 100µg of streptomycin/ml. Hep3B cells were cultured in minimalessential medium supplemented with 10% fetal bovine serum, penicillin,and streptomycin. Transient transfection was performed withthe LipofectAmine Plus reagent (GIBCO/Invitrogen) followingthe manufacturer's instructions. Serum-free procedures wereused for transfection. pRL-CMV (Promega) or pSV40-lacZ werecotransfected for normalization of transfection efficiency.Twenty-four hours after transfection, cells were trypsinized,split into multiwell plates, and exposed to the various experimentalconditions.

Hypoxia treatment. Cell cultures were treated as previously described in a modularincubator (Billups-Rothenburg) and flushed with a gas mixtureof 0.5% O₂-5% CO₂, and balanced with nitrogen (3). In some experiments,cells were directly incubated in a hypoxic workstation (In VIVO2)and the percentage of oxygen in mixed air was described foreach experiment.

Luciferase assays. For single luciferase assays, cells were washed once with 1xphosphate-buffered saline and lysed on plates with 1x reporterlysis buffer (Promega). After incubation at room temperaturefor 5 min, cells were scraped and collected into Eppendorf tubes.After vigorous vortex and centrifugation, the supernatant wastaken for the luciferase assay. For dual luciferase assays,cells were washed and lysed on plates with 1x passive cell lysisbuffer (Promega). The cells were scraped, collected, and subjectedto three cycles of freezing-thawing. Luciferase assays wereperformed with the assay reagents purchased from Promega byfollowing the manufacturer's instruction. The luminescence wasmeasured in a TD20/20 luminometer (Promega). When applicable,ß-galactosidase activity was assayed in 1x Z-buffer(42) containing 4 mg of o-nitrophenyl ß-D-galactopyranoside(Sigma)/ml.

Cell lysate preparation, immunoprecipitation, and Western blot. For whole-cell lysates, cell pellets were collected and lysedon ice for 1 h with 1x hydroxylase buffer (50 mM Tris-Cl, 250mM NaCl, 1% Triton-100, 5 mM EDTA, 50 mM NaF, 0.1 mM Na₃VO₄,1 mM phenylmethylsulfonyl fluoride, 100 µM desferrioxamine,1x protease inhibitor mix). Immunoprecipitation and Westernblot procedures were as previously described except that HRP-labeledgoat anti-mouse IgG-Fc was used in immunoblotting (39, 40).The protein concentration was determined by using Bio-Rad reagents.

RESULTS

Top

Results

Common stimuli regulate both protein stability and transactivation activity of HIF-1 ${alpha}$ . Two transactivation domains of HIF-1 ${alpha}$ , the N-terminal transactivationdomain (NAD) and the carboxyl-terminal transactivation domain(CAD), have been previously described (20, 33) as shown schematicallyin Fig. 1A. We have reported that the protein levels of bothGH1 ${alpha}$ 530-826 and GH1 ${alpha}$ 530-778 were up-regulated by hypoxia and definedan ODD (aa 557 to 571) (45), and the protein stabilities ofthe different GH1 ${alpha}$ constructs tested are summarized in Fig. 1A.Since the ODD overlaps NAD, the regulation of NAD activity appearsto be determined, at least partially, by variations in proteinlevels (22, 33). Although the protein levels of GH1 ${alpha}$ 740-826 intransfected HeLa cells were not affected, their transactivationactivity was significantly up-regulated by hypoxia and the hypoxiamimics, cobalt and desferrioxamine (Fig. 1B). Similar resultswere observed in the 143B and Hep3B cell lines (data not shown).Figure 1C shows the increase in endogenous HIF-1 ${alpha}$ protein levelsthat accompanies the stimulatory response to hypoxia, desferrioxamine,and cobalt. The finding that the same stimuli that regulateHIF-1 ${alpha}$ protein stability also affect its transcriptional activitysuggests that a similar or identical regulatory pathway is involvedin both processes.

View larger version (20K):
[in this window]
[in a new window]
FIG. 1. (A) Schematic structures of HIF-1 ${alpha}$ and GAL4-H1 ${alpha}$ constructs used in this study. (B) Effects of hypoxia and hypoxia mimics on the transactivation activity of GH1 ${alpha}$ 740-826. HeLa cells were cotransfected with GH1 ${alpha}$ 740-826 (2.5 µg), pFR-luc reporter (2.5 µg), and pCMVLacZ (0.5 µg). Twenty-four hours later the transfected cells were trypsinized. Fifty percent of the cells were split equally into 12-well culture plates for luciferase assays (top), and the other 50% of the cells were split into 60-mm-diameter dishes to extract proteins for Western blotting with monoclonal antibody against GAL4-DBD (bottom). The cells were cultured under either normoxic (Nmx) or hypoxic (Hpx) (0.5% O₂) conditions or with desferrioxamine (Dfx) (130 µM)or cobalt chloride (Cbt) (75 µM), respectively, for 16 h before harvest. (C) Response of endogenous HIF-1 to hypoxia and hypoxic mimics in Hep3B cells. Hep3B cells were transfected with a pEpo-luc reporter (5 µg) and pCMVLacZ (0.5 µg), and the transfected cells were treated as described in panel B. The protein levels of HIF-1 ${alpha}$ in whole-cell lysates were detected with anti-HIF-1 ${alpha}$ monoclonal antibody. The luciferase activity was corrected by ß-galactosidase activity. RLU, relative luciferase activity.

Characterization of VHL-deficient and VHL-expressing cell lines. Since VHL is involved in the regulation of HIF-1 ${alpha}$ stability,we investigated the role of VHL in the regulation of the CADtransactivation activity. A renal carcinoma cell line, 786-O,which lacks one VHL allele and expresses a truncated nonfunctionalprotein from the second allele (16), was stably transfectedwith a vector expressing a tetracycline-regulatable (Tet-off*)wild-type VHL protein (786-O#52) or with the vector backbone(786-O#126). The expression of HIF- ${alpha}$ and VHL in 786-O cells wasexamined by immunoblot analysis. As shown in Fig. 2A, HIF-2 ${alpha}$ but not HIF-1 ${alpha}$ is detectable in 780-O cells, confirming the observationsmade by others (30). In addition, HIF-2 ${alpha}$ levels were constitutivelyexpressed and were not affected by hypoxia in the VHL-deficientcells. Figure 2B shows the lack of VHL expression in 786-O cellor 786-O#126, whereas it was detected in the 786-O#52 cells.Treatment of 786-O#52 cells with tetracycline for 36 h increasedHIF-2 ${alpha}$ levels, while VHL expression became undetectable (Fig.2C). Hypoxia, desferrioxamine, and cobalt did not repress theexpression of VHL but stabilized HIF-2 ${alpha}$ (Fig. 2C and data notshown). These data confirmed that tetracycline efficiently repressesthe expression of VHL in 786-O#52 cells and that the expressedVHL functions normally.

View larger version (43K):
[in this window]
[in a new window]
FIG. 2. (A) Hypoxia-regulated expression of HIF- ${alpha}$ proteins in VHL-deficient cells (786-O). 786-O cells and HeLa cells were cultured under either normoxic (Nmx) (21% O₂) or hypoxic (Hpx) (0.5% O₂) conditions for 6 h. Whole-cell lysates were prepared and examined for the expression levels of HIF-1 ${alpha}$ and HIF-2 ${alpha}$ . (B) Expression of VHL and HIF-2 ${alpha}$ in 786-O-derived cell lines. Whole-cell lysates prepared from 786-O, 786-O#52, and 786-O#126 cells were fractionated on an SDS-10% polyacrylamide gel and transferred onto a PVDF membrane, and the expression of VHL and HIF-2 ${alpha}$ was detected by immunoblotting. (C) Regulated expression of VHL in 786-O#52 cells by tetracycline. Cells were seeded and cultured in DMEM with or without 3 µg of tetracycline/ml for a total of 36 h, and the media were changed every 12 h to maintain the tetracycline concentration. Immediately before harvest, cells were exposed to normoxia (21% O₂) (lanes 1 and 2) or hypoxia (2% O₂) (lane 3) for 12 h. The expression status of VHL and HIF-2 ${alpha}$ was detected by immunoblot. The same membrane was stripped and detected by a monoclonal antibody against ß-tubulin (bottom). NS, nonspecific band.

VHL is not required for the up-regulation of HIF-1 ${alpha}$ transactivation activity in response to hypoxia or hypoxia mimics. Next, we studied whether VHL was required for the stimulationof HIF CAD transactivation activity in response to hypoxia andhypoxia mimics. A GAL4 fusion construct, GH1 ${alpha}$ 740-826 (Fig. 1A),was cotransfected into 786-O#52 cells with a luciferase reporterunder the control of five GAL4 binding sites, and the cellswere maintained with or without tetracycline. As shown in Fig.3A, in response to hypoxia or desferrioxamine, the transactivationactivity of GH1 ${alpha}$ 740-826 increased regardless of the treatmentof tetracycline. In addition, hypoxia and hypoxia mimics increasedthe transactivation activity of GH1 ${alpha}$ 740-826 in the 786-O#126cell line, which lacks functional VHL (Fig. 3B). These datademonstrate that expression of a functional VHL is not necessaryfor hypoxia-induced up-regulation of HIF CAD activity.

View larger version (19K):
[in this window]
[in a new window]
FIG. 3. (A) Regulation of CAD activity in 786-O#52 (Tet-off VHL). Cells were seeded and cotransfected with GH1 ${alpha}$ 740-826 and pFR-Luc reporter. After transfection, cells were cultured in DMEM with or without tetracycline (3 µg/ml) for 36 h. Twelve hours before harvest and luciferase assay, cells were treated with hypoxia (Hpx) (2% O₂) or desferrioxamine (Dfx) (130 µM). The activity was corrected by cotransfected pRL-CMV. (B) Regulation of CAD activity in 786#126 (VHL^-) cells. Cells were transfected, treated, and assayed as for panel A, except for the addition of tetracycline. Nmx, normoxic conditions.

Effect of hydroxylase inhibitor on CAD activity. PHD enzymes are involved in the regulation of oxygen-dependentdegradation of HIF-1 ${alpha}$ (5, 12). These enzymes require Fe²⁺, O₂,and 2-oxoglutarate to hydroxylate prolyl residues at the ODDof HIF-1 ${alpha}$ , thus promoting its interaction with VHL. Since thecarboxyl-terminal transactivation domain is responsive to hypoxia,iron chelators, and transition metals as well, we hypothesizedthat the regulation of CAD activity might involve a similarenzymatic process. To test this hypothesis, we treated cellswith a 2-oxoglutarate analogue, DOG, which serves as a cell-permeativecompetitive inhibitor of hydroxylases (19). As shown in Fig.4A, DOG stimulated the transactivation activity of CAD in HeLacells under normoxic conditions as efficiently as desferrioxamineor hypoxia. This regulation requires the negative regulatoryregion (NRR) (aa 740 to 785) of HIF-1 ${alpha}$ because another Gal4 fusionconstruct, GH1 ${alpha}$ 786-826, is constitutively active and is not stimulatedby DOG or desferrioxamine (Fig. 4B). To examine if VHL participatedin this hydroxylation-dependent modulation of CAD activity,we utilized the VHL^- 786-O#126 or VHL⁺ 786-O#52 cells. Figure4C and D shows that both cell lines responded normally to DOG,indicating that the responsiveness of HIF CAD to hydroxylaseinhibition is also independent of the expression of functionalVHL. To examine the effect of a hydroxylase inhibitor on proteinlevels of GH1 ${alpha}$ 740-826, we transfected a plasmid expressing GH1 ${alpha}$ 740-826into HeLa cells. Western blot analysis revealed that while treatmentof cells with DOG significantly increased the protein levelsof HIF-1 ${alpha}$ , the protein levels of GH1 ${alpha}$ 740-826 did not change (Fig.4E), confirming that inhibition of hydroxylation enhanced thetransactivation activity, not the stability, of GH1 ${alpha}$ 740-826.

View larger version (15K):
[in this window]
[in a new window]
FIG. 4. PHD inhibitor induces CAD activity under normoxic conditions. (A and B) HeLa cells were transfected with GH1 ${alpha}$ 740-826 and pFR-luc reporter (A) or GH1 ${alpha}$ 786-826 and pFR-luc (B). Luciferase assays were performed after the cells were treated with DOG (1 mM), hypoxia (Hpx) (2% O2), and desferrioxamine (Dfx) (130 µM) for 12 h. (C and D) Effects of hydroxylase inhibitor on GH1 ${alpha}$ 740-826 in 786-O#52 (VHL⁺) and 786-O#126 (VHL^-) cells. (E) Effect of hydroxylase inhibitor on protein stability of GH1 ${alpha}$ 740-826. HeLa cells were transfected with GH1 ${alpha}$ 740-826, pooled, and reseeded in two dishes that were treated with or without DOG. Whole-cell lysates were prepared, and the expression levels of HIF-1 ${alpha}$ and GH1 ${alpha}$ 740-826 were examined by Western blotting by using anti-HIF-1 ${alpha}$ and anti-GAL4-DBD monoclonal antibodies, respectively. Cbt, cobalt chloride.

Effect of overexpression of HIF-PHDs on HIF CAD activity. The recently identified HIF-PHD family includes three structurallyand functionally related enzymes (5, 12). Since overexpressionof these enzymes enhanced the degradation of HIF-1 ${alpha}$ , hydroxylaseactivity was considered limiting for the normoxic degradationof HIF-1 ${alpha}$ (5). If hydroxylase activity is a limiting factor forthe repression of the transactivation activity of HIF-1 ${alpha}$ , andany of these PHDs is involved in this repression, overexpressionof that hydroxylase would repress the basal transactivationactivity of GH1 ${alpha}$ 740-826. We cotransfected plasmids expressingPHD1, PHD2, and PHD3 with GH 1 ${alpha}$ 740-826 in HeLa cells and observedthat at least PHD3 repressed the transactivation activity ofGH1 ${alpha}$ 740-826 under normoxic conditions, while VP16 showed no effect(Fig. 5A). In contrast, none of the enzymes inhibited the constitutivetransactivation activity of GH1 ${alpha}$ 786-826 (Fig. 5B) or a controlGal4-VP16 protein (Fig. 5C). As expected, none of these enzymesinhibited hypoxia or desferrioxamine-stimulated activity ofCAD (data not shown). It should be pointed out that the proteinlevels of PHD1, PHD2, and PHD3 were not examined because ofunavailability of antibodies. Therefore, these data do not ruleout the involvement of PHD1 in normoxic repression. Nevertheless,these results suggest that a hydroxylase activity is involvedin the normoxic repression of HIF CAD and that this activityis limiting.

View larger version (18K):
[in this window]
[in a new window]
FIG. 5. Effects of overexpression of PHDs on CAD activity. HeLa cells were cotransfected with GH1 ${alpha}$ 740-826 (1 µg), pFR-Luc (1 µg), and pRL-CMV (0.1 µg). In addition, either empty vector (3 µg) or plasmids coding for PHD1, PHD2, PHD3, or VP16 (3 µg) was cotransfected (A). Similarly, the effects of PHDs on GH1 ${alpha}$ 786-826 and GVP16 were examined (B and C). RLU, relative luciferase activity.

Hydroxylase inhibitor enhances the functional interaction between HIF-1 ${alpha}$ and p300/CBP. Since the transactivation activity of HIF-1 ${alpha}$ involves its interactionwith p300/CBP, we next tested if inhibition of the PHDs enhancesHIF-1 ${alpha}$ activity by promoting the functional interaction betweenHIF-1 ${alpha}$ and p300/CBP. As shown in Fig. 6A, exogenous expressionof p300 mildly increased the activity of GH1 ${alpha}$ 740-826 under normoxiccondition. Treatment of cells with DOG, hypoxia, and desferrioxaminesubstantially enhanced the activity. The adenoviral oncoproteinE1A, which sequesters endogenous p300/CBP (41), blocked theresponse of GH1 ${alpha}$ 740-826 to DOG, hypoxia, and desferrioxamine.An E1A mutant that is defective in binding p300/CBP is defectivein blocking the response, demonstrating that recruitment ofp300/CBP is required for GH1 ${alpha}$ 740-826 responses. Since DOG, hypoxia,and desferrioxamine do not directly increase the protein levelsor transactivation activity of p300/CBP (N. Sang and J. Caro,data not shown), the results presented suggest that these stimuliare affecting the interaction between HIF CAD and p300/CBP.

View larger version (31K):
[in this window]
[in a new window]
FIG. 6. (A) Hydroxylase inhibitor enhances the ability of p300 to potentiate CAD activity. HeLa cells were cotransfected with GH1 ${alpha}$ 740-826 and pFR-Luc. In addition, empty vector, p300, E1A, or an E1A mutant was cotransfected to test its effects on CAD activity. The transfected cells were split and exposed to various conditions as indicated. (B) Hydroxylase inhibitor enhances functional interaction between HIF-1 ${alpha}$ and the CH1 domain of p300. Indicated combinations of mammalian two-hybrid partners (2 µg of each) were cotransfected with pFR-luc reporter (1.5 µg). Twenty-four hours after transfection, cells were split and subjected to various conditions as indicated. The relative luciferase activity in pDBD-, pHVP-transfected cells was too low to be seen in the chart. Nmx, normoxia; Hpx, hypoxia; Dfx, desferrioxamine; Cbt, cobalt chloride.

Using the mammalian two-hybrid system, we further tested thehypothesis that DOG, hypoxia, and desferrioxamine enhance HIFCAD-p300 interactions. In this system, the CH1 domain (aa 300to 528) of p300, which is responsible for the interaction withHIF-1 ${alpha}$ , was fused with the DBD of GAL4 (GCH1) (4). A DNA fragmentencoding aa 723 to 826 of HIF-1 ${alpha}$ was fused in frame with theVP16 activating domain (VP16H1 ${alpha}$ 723-826) (4, 25). G4CH1 givesrise to basal transactivation activity because CH1 is capableof interacting with the basal transcription complex (TBP) directly(52). Enhanced transactivation activity would be detected ifVP16H1 ${alpha}$ 723-826 interacts with G4CH1, because VP16 is a much strongertransactivator than CH1 (4). As shown in Fig. 6B, cotransfectionwith VP16H1 ${alpha}$ 723-826 increased the transactivation activity aroundtwofold under normoxic conditions. Treatment with DOG, hypoxia,and desferrioxamine further enhanced the transactivation activity,confirming that they promoted the functional interaction betweenp300-CH1 and HIF CAD.

Hydroxylase inhibitor and hypoxia enhance the physical interaction of HIF-1 ${alpha}$ -p300 in vivo. We next examined the effects of hydroxylase inhibitor, hypoxia,and desferrioxamine on the physical interaction between HIFCAD and p300/CBP in vivo. Plasmids expressing GH1 ${alpha}$ 740-826 andGH1 ${alpha}$ 786-826 were transfected into HeLa cells. Protein complexeswere immunoprecipitated by an anti-p300 monoclonal antibodyand were resolved by sodium dodecyl sulfate-polyacrylamide gelelectrophoresis (SDS-PAGE) and transferred onto polyvinylidenedifluoride (PVDF) membranes. An anti-GAL4 (DBD) antibody wasused to detect whether GAL4-HIF CAD copurified with p300 underdifferent experimental conditions. GH1 ${alpha}$ 786-826 was found to copurifywith p300 under all conditions tested (Fig. 7A). However, theinteraction between p300 and GH1 ${alpha}$ 740-826 was only detectablewhen cells were exposed to DOG, hypoxia, or desferrioxamine(Fig. 7B). These results suggest that the NRR (aa 740 to 785)inhibits the interaction between p300 and HIF CAD and that thisinhibition can be overridden by a hydroxylase inhibitor, hypoxia,and desferrioxamine.

View larger version (55K):
[in this window]
[in a new window]
FIG. 7. Hydroxylase inhibitor enhances the physical interaction between HIF-1 ${alpha}$ and p300 in vivo. HeLa cells were transfected with GH1 ${alpha}$ 786-826 (A) or GH1 ${alpha}$ 740-826 (B). Twenty-four hours after transfection, transfected dishes were trypsinized, pooled, and reseeded to fresh dishes to normalize the transfection efficiency. The dishes then were treated overnight with various conditions as indicated (samples 1 to 5). As controls, none-transfected HeLa cells were exposed to normoxia (Nmx) or hypoxia (Hpx) (samples 6 and 7). Whole-cell lysates (WCL) were prepared. WCL (5%) were separated directly through an SDS-10% polyacrylamide gel (top) to check the expression of transgenes. The remaining lysates (95%) were immunoprecipitated (IMP) with a monoclonal antibody against human p300 (bottom, (lanes 1 to 4, 6, and 7) or an anti-E1A monoclonal antibody as the control (bottom, lane 5). The precipitated protein complexes were resolved through an SDS-10% polyacrylamide gel and transferred onto PVDF membranes. The membranes were detected with a monoclonal antibody specific for GAL4-DBD and developed by an ECL kit. Dfx, desferrioxamine.

Overexpression of the negative regulatory region up-regulates HIF CAD activity in vivo. The NRR (aa 740 to 813) may directly inhibit the interactionbetween p300 and HIF-1 ${alpha}$ by itself or indirectly by recruitmentof an inhibitory factor. If recruitment of an inhibitory factoris required for the inhibition, overexpression of NRR in transshould be able to sequester the inhibitory factor and thus alleviatethe inhibition. For this purpose, we constructed a plasmid expressingthe NRR fused in frame with a fragment of VP16 that providesa nuclear localization signal. As shown in Fig. 8, transfectionof the plasmid expressing H1 ${alpha}$ 740-813 increased the transactivationactivity of GH1 ${alpha}$ 740-826 in normoxic Hep3B cells, while it hadno stimulatory effect on the activity of GH1 ${alpha}$ 786-826. Transfectionwith the VP16 backbone plasmid had no effect. These data supporta model in which the NRR interacts with a cellular inhibitoryfactor.

View larger version (19K):
[in this window]
[in a new window]
FIG. 8. Effect of overexpression of NRR on CAD activity. In Hep3B cells, pFR-Luc (1 µg) and pRL-CMV (0.1 µg) were cotransfected with either GH1 ${alpha}$ 740-826 (1 µg) (A) or GH1 ${alpha}$ 786-826 (B). In addition, the indicated amount (in micrograms) of empty vector, plasmids expressing HIF-1 ${alpha}$ 740-813, or VP16 was cotransfected. Luciferase assays were performed 48 h after transfection, and the luciferase activity (RLU) was corrected by cotransfected pRL-CMV.

DISCUSSION

Top

Discussion

Since HIF-1 ${alpha}$ stability and the transactivation activity of itsCAD domain are responsive to the same stimuli (Fig. 1), it seemsreasonable to suggest that a common signaling pathway controlsboth protein degradation and transactivation activity. HIF-1 ${alpha}$ degradation is determined by its interaction with VHL, whichtargets it for ubiquitination and proteasomal destruction. Inturn, HIF-VHL interaction depends on the activity of oxygen-dependentPHD enzymes that hydroxylate specific residues in the ODD ofHIF-1 ${alpha}$ . In this study, we examined the possible involvement ofVHL and hydroxylase in the regulation of the CAD activity. Weprovide evidence that up-regulation of HIF CAD activity in responseto hypoxia can be mimicked by hydroxylase inhibitors but doesnot depend on the presence of VHL. Thus, it appears that whileboth regulatory pathways overlap partially, the pathway regulatingHIF-1 ${alpha}$ stability is distinguishable from the one that regulatesthe CAD activity.

Three structurally and functionally related PHDs (PHD1 to PHD3)have been cloned and proven to be highly conserved in distantspecies recently (5, 12). Our data show that the regulationof CAD also involves hydroxylase activity. It is not clear whetherthe different hydroxylases have equivalent activities on HIF-1 ${alpha}$ degradation and transcriptional activation. During our studies,we observed that while cobalt is a very effective inducer ofHIF-1 ${alpha}$ accumulation, it is not an effective stimulator of HIFCAD transactivation activity (Fig. 1B and C). The involvementof different hydroxylases in the regulation of protein degradationand transactivation may provide an explanation for that discrepancy.It is assumed that cobalt replaces iron in the hydroxylase enzymaticcore, thus inhibiting its activity. Since cobalt has limitedeffects on CAD activity, it is possible that the hydroxylaseinvolved in regulation of CAD may have a higher affinity forFe²⁺ binding. In addition, while there are already three membersin the PHD family, the number may still grow, and functionallythey may not be redundant. The existence of multiple hydroxylasesin human cells reflects complexity in oxygen sensing and hypoxiaadaptation.

The role of p300/CBP in hypoxic response was first suggestedby its interaction with HIF-1 ${alpha}$ in a protein-protein interactionscreening (1). Since then, several studies have linked the recruitmentof p300/CBP to the hypoxia-induced up-regulation of HIF-1 ${alpha}$ transactivationactivity (7, 10, 11, 13, 21). Mutations in the CH1 domain ofp300 disrupting the interaction between p300/CBP and HIF CADcompletely abolish the transactivation activity of HIF-1 ${alpha}$ (13,25). Furthermore, blocking the interactions between p300 andHIF-1 ${alpha}$ in vivo by overexpressing polypeptides corresponding tothe p300-CH1 or HIF-1 ${alpha}$ 786-826 domain results in down-regulationof HIF-1 ${alpha}$ transactivation activity and a significant inhibitionof downstream gene expression (25). These observations indicatethat the interaction with p300/CBP is indispensable for thetransactivation activity of HIF-1 ${alpha}$ CAD and that HIF-1 ${alpha}$ is notable to interact directly with the basic transcriptional machinery.

Previous studies have shown that the HIF-1 ${alpha}$ contains an internaldomain that inhibits HIF-1 ${alpha}$ transcriptional activity, and mutationsof aa 781 to 783 (RLL to AAA) within this domain inactivateits repressive activity (20, 32, 33). Consistent with previousobservations, we find that GH1 ${alpha}$ 786-826 is constitutively activeand is hardly regulated by hypoxia or hypoxia mimics. Correspondingly,a HIF-1 ${alpha}$ fragment encompassing aa 786 to 826 interacts with p300in vivo under normoxia conditions. In contrast, GH1 ${alpha}$ 740-826 showsa striking transcriptional response to hypoxia, desferrioxamine,and hydroxylase inhibitors. Therefore, the region of aa 740to 785 serves as an NRR that prevents p300/CBP from bindingto the active 786-826 region in vivo under normoxic conditions.This hypothesis is confirmed by our finding that coimmunoprecipitationof GH1 ${alpha}$ 740-826 and p300 only becomes detectable under hypoxicconditions or following treatment with desferrioxamine or hydroxylaseinhibitors. How NRR blocks p300/CBP interaction with HIF-1 ${alpha}$ invivo is not clear. The repressive effect appears not to be causedby intramolecular interactions between the NRR and the 786-826region, since studies utilizing a mammalian two-hybrid systemfailed to show any interaction between these two fragments (Sangand Caro, data not shown). Since HIF-1 ${alpha}$ 723-826 is able to interactwith p300 efficiently in vitro (4), the repressive effect ofthe NRR appears to depend on posttranslational modification.One possibility is that this region serves as a destabilizingfactor that represses the interaction between HIF CAD and p300.In this context, normoxic hydroxylation would affect the propertiesof either p300 or HIF-1 ${alpha}$ , making it more difficult for p300 toaccess CAD. Although there are two proline residues within theregion of aa 740 to 785, neither of them conforms to the suggestedconsensus hydroxylation motif LXXLAP (12, 29). Similarly, whilemultiple prolyl residues exist, no prolyl hydroxylation consensusis found in the CH1 domain of p300. Another possibility is thatthe NRR interacts with protein factors that function as hydroxylation-dependentrepressors that prevent the association between p300/CBP andCAD under normoxic conditions. Inhibition of hydroxylation destabilizesthe interaction of these factors and the NRR, thus allowingp300/CBP to access the CAD domain. Our finding that overexpressionof the NRR relieves the normoxic repression of CAD supportsthe notion that NRR interacts with a negative regulatory factor.

Based on previous reports and the data presented here, we proposea model for the regulation of the protein stability and transactivationactivity of HIF-1 ${alpha}$ (Fig. 9) in which hydroxylases are the keyfactors. Since hydroxylase activity depends on iron, oxygen,and 2-oxoglutarate, a decrease in the concentration of any ofthese three factors leads to a hypoxia-like response. Whilethe signaling pathways upstream of the hydroxylases are identical,the mechanism underlying the regulation of protein stabilityand the modulation of the transactivation activity are differentiallycontrolled. The former involves direct hydroxylation of HIF-1 ${alpha}$ at its prolyl residues in the ODD, thus promoting the interactionwith VHL and its subsequent targeting for degradation. The latterinvolves the hydroxylation of protein factor(s) that could bep300, HIF-1 ${alpha}$ , a putative inhibitory factor, or any combinationthereof. This hydroxylation would most likely promote the interactionbetween HIF-1 ${alpha}$ and the inhibitory factor, thus disrupting theinteraction between HIF-1 ${alpha}$ and p300/CBP. Conversely, hypoxiaand hypoxia mimics directly or indirectly inhibit hydroxylaseactivity, thus resulting in activation of the CAD.

View larger version (22K):
[in this window]
[in a new window]
FIG. 9. Proposed roles of VHL and hydroxylases (HDS) in HIF-1 ${alpha}$ stabilization and transactivation. Filled arrows indicate stimulation or enhancement, and empty arrows stand for inhibition or repression.

During the preparation of this report, it was reported by othersthat an inhibitory factor, FIH-1 (factor inhibiting HIF), interactswith HIF-1 ${alpha}$ and inhibits the transcriptional activity of CAD(28). It was shown that the transactivation activity of HIF-1 ${alpha}$ is regulated by VHL but not through protein stabilization. Instead,FIH-1 binds to VHL wherein VHL functions as a transcriptionalcorepressor by recruiting histone deacetylases. In addition,these interactions between HIF-1 ${alpha}$ , FIH-1, and VHL were suggestedto be independent of hydroxylation (28). While their findingssupport the involvement of a cellular inhibitory factor, ourdata demonstrate that the regulation of the CAD activity ofHIF-1 ${alpha}$ involves a hydroxylase activity and is, at least partly,a VHL-independent event.

ADDENDUM IN PROOF

Top

Addendum in proof

While this paper was in press, Lando et al. published resultsthat are consistent with our experimental findings and conclusions(D. Lando, D. J. Peet, D. A. Whelan, J. J. Gorman, and M. L.Whitelow, Science 295:858-861, 2002).

ACKNOWLEDGMENTS

We thank P. Ratcliffe, D. Livingston, N. Kley, and A. Giordanofor kindly providing expression plasmids, cell lines, and otherreagents necessary for this study. We thank R. Silvano for professionalproofreading and editing of the manuscript. N.S. and J.F. thankY. Zhang (Huashan Hospital of Fudan University Medical Center,Shanghai) for his support.

N.S. is supported by a training grant from the National Institutesof Health (T32HL007821). This work was supported in part bya grant from American Heart Association (9950122N) and by agrant from National Institutes of Health (RO1CA89212-02) toJ.C. V.S. is supported by a grant from American Heart Association(0060194U). J.F. is a visiting researcher from Huashan Hospitalof Fudan University Medical Center, Shanghai, China.

FOOTNOTES

* Corresponding author. Mailing address: Cardeza Foundation for Hematologic Research, Jefferson Medical College of Thomas Jefferson University, 1015 Walnut St., Curtis Bldg., Rm. 810, Philadelphia, PA 19107-5099. Phone: (215) 955-7775. Fax: (215) 923-3836. E-mail: jaime.caro{at}mail.tju.edu.

REFERENCES

Top