Hsp70 and CHIP Selectively Mediate Ubiquitination and Degradation of Hypoxia-inducible Factor (HIF)-1α but Not HIF-2α*

Hypoxia-inducible factors (HIFs) are transcription factors that mediate adaptive responses to reduced oxygen availability. HIF-α subunits are stabilized under conditions of acute hypoxia. However, prolonged hypoxia leads to decay of HIF-1α but not HIF-2α protein levels by unknown mechanisms. Here, we identify Hsp70 and CHIP (carboxyl terminus of Hsc70-interacting protein) as HIF-1α-interacting proteins. Hsp70, through recruiting the ubiquitin ligase CHIP, promotes the ubiquitination and proteasomal degradation of HIF-1α but not HIF-2α, thereby inhibiting HIF-1-dependent gene expression. Disruption of Hsp70-CHIP interaction blocks HIF-1α degradation mediated by Hsp70 and CHIP. Inhibition of Hsp70 or CHIP synthesis by RNA interference increases protein levels of HIF-1α but not HIF-2α and attenuates the decay of HIF-1α levels during prolonged hypoxia. Thus, Hsp70- and CHIP-dependent ubiquitination represents a molecular mechanism by which prolonged hypoxia selectively reduces the levels of HIF-1α but not HIF-2α protein.

Understanding the fundamental mechanisms regulating HIF-1 activity may lead to novel therapies for these diseases.
In contrast to rapid degradation under aerobic conditions, HIF-1␣ is resistant to prolyl hydroxylation and translocated to the nucleus under hypoxic conditions (1,17). The accumulated nuclear HIF-1␣ dimerizes with HIF-1␤, recruits the co-activators p300/CBP, and binds to cis-acting hypoxia-response elements (HREs) in target genes, leading to the transcriptional activation of genes encoding proteins that mediate adaptive responses to hypoxia (18,19). Although HIF-1␣ protein is stabilized in response to acute hypoxia, HIF-1␣ protein levels are attenuated during prolonged hypoxia (20).
HIF-2␣ has a similar structure to HIF-1␣ but is only expressed in certain tissues (21). HIF-2␣ protein stability is also O 2 -regulated and degraded through the PHD/VHL/Elongin pathway in well oxygenated cells. HIF-2␣ is stabilized during hypoxia, and protein levels increase with prolonged hypoxia (22). Like HIF-1␣, HIF-2␣ dimerizes with HIF-1␤ and regulates gene transcription in response to hypoxia. In addition to common target genes regulated by both HIF-1 and HIF-2, HIF-2 mediates transcription of some unique genes, such as POU5F1, CCND1, and TGFA (23,24). The amino-terminal transactivation domains of HIF-1␣ and HIF-2␣ are essential for target gene specificity (25). In this study, using a stable isotope labeling by amino acids in cell culture (SILAC)-based quantitative proteomic screen, we identified Hsp70 and CHIP (carboxyl terminus of Hsc70-interacting protein) as HIF-1␣-interacting proteins that selectively regulate ubiquitination and degradation of HIF-1␣ but not HIF-2␣.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-The complete coding sequence of human Hsp70 or CHIP was amplified from HEK293 cell cDNA by PCR and inserted into pcDNA3. 1-V5-His vector (Invitrogen). Hsp70-V5 or CHIP-V5 cDNA was excised with BamHI and PmeI, polished by Klenow, and ligated into lentiviral vector EF.v-CMV.GFP. Full-length HIF-1␣ cDNA was subcloned to pGex-6P-1 vector (GE Healthcare). Deletion mutants of Hsp70 or CHIP were generated by PCR and cloned into pcDNA3.1-V5-His or pGex-6P-1 vectors. The CHIP(H260Q) and CHIP(K30A) mutations were generated using QuikChange sitedirected mutagenesis kit (Stratagene). Other constructs have been described previously (14). The DNA sequences of plasmid constructs were confirmed by nucleotide sequencing.
Lentivirus Production-Recombinant lentivirus was generated by transfection of HEK293T cells with the transducing vector containing Hsp70-V5 or CHIP-V5 cDNA and packaging vectors pMD.G and pCMVR8.91 (26). After 48 h, lentivirus was harvested and transduced into RCC4 cells.
SILAC-based Proteomic Screen-HEK293 cells cultured in Dulbecco's modified Eagle's medium containing 12 C 6 -14 N 2lysine/ 12 C 6 -14 N 4 -arginine (light isotope media) or 13 C 6 -15 N 2lysine/ 13 C 6 -15 N 4 -arginine (heavy isotope media) were lysed in modified RIPA buffer containing 50 mM Tris-HCl (pH 7.5), 1 mM ␤-mercaptoethanol, 150 mM NaCl, 1 mM Na 3 VO 4 , 1 mM NaF, 1 mM EDTA, 0.25% sodium deoxycholate, 1% Igepal, and protease inhibitor mixture (Roche Applied Science). 72 mg FIGURE 1. Identification of Hsp70 and CHIP as HIF-1␣-interacting proteins. A, proteomic screen was performed to identify HIF-1␣-interacting proteins. B, mass spectrum of a tryptic peptide of Hsp70 identified by SILAC was determined. C, fragmentation spectrum of a representative Hsp70 peptide shown in B was determined. D, mass spectrum of a tryptic peptide of CHIP identified by SILAC was determined. E, fragmentation spectrum of a representative CHIP peptide shown in D was determined. For the peptides shown in B and D, the heavy and light forms differ by m/z 5. LC-MS/MS, liquid chromatography-tandem mass spectrometry; aa, amino acids.
of whole cell lysates were incubated overnight with 200 g of GST or GST-HIF-1␣-(531-826) immobilized on glutathione-Sepharose beads. The bound proteins were eluted, mixed, fractionated by SDS-PAGE, and stained with colloidal blue staining kit (Invitrogen). Stained protein bands were excised and digested with trypsin. After in-gel digestion, the tryptic peptides were extracted, dried, and reconstituted in 0.1% formic acid. The peptide mixture was analyzed by reverse phase liquid chromatography-tandem MS. MS spectra were acquired on a quadrupole time-of-flight MS (Q-TOF US-API, micromass) in a survey scan (m/z range from 350 to 1200) in a data-dependent mode selecting the four most abundant ions for tandem MS (m/z range from 100 to 1800). The acquired data were processed using MassLynx (version 4.02) and searched against NCBI Protein Database (released on March 5, 2007) using Mascot search engine (version 2.2.0, Matrixscience). Proteins with at least two reliable peptides (rank 1; unique; individual score higher than or equal to 30) were considered as positively identified proteins. Relative quantitation of stable isotope-labeled peptides was performed using MSQuant (version 1.4.3a39) and manually verified (27).
GST Pulldown Assays-GST and GST fusion proteins were expressed in Escherichia coli BL21-Gold (DE3) and purified (14,28). His-tagged proteins were expressed in HEK293 or HEK293T cells, captured by Ni-NTA beads (Qiagen) from whole cell lysates, and eluted with 250 mM imidazole. Equal amounts of GST and GST fusion proteins immobilized on glutathione-Sepharose beads were incubated overnight with whole cell lysates or purified Hsp70-V5-His. After washing three times, the bound proteins were fractionated by SDS-PAGE, followed by immunoblot assays.
Reverse Transcription-PCR-Total RNA was isolated from cells and reverse-transcribed (14). cDNA was amplified by PCR and analyzed by 1.5% agarose/ethidium bromide gel electrophoresis.
shRNA Assays-The 19-nucleotide sequences targeting Hsp70 or CHIP are shown in supplemental Table S1. Oligonucleotides encoding shRNAs were annealed and ligated into BglII/HindIII-digested pSUPER.retro.neo.GFP vector (OligoEngine). A scrambled shRNA with no significant homology to any mammalian gene sequence was also prepared (28). Cells were transfected with shRNA expression vectors for 72 h prior to analysis.
Statistical Analysis-Data were expressed as mean Ϯ S.E. Differences were examined by Student's t test between two groups or oneway analysis of variance within multiple groups. p Ͻ 0.05 was considered significant.

Identification of HIF-1␣-interacting
Proteins-To study novel molecular mechanisms that regulate HIF-1␣ protein stability during hypoxia, we screened for HIF-1␣-interacting proteins using SILAC combined with a GST pulldown approach. HEK293 cells were cultured in standard culture media (light media) or media supplemented with heavy isotope-labeled lysine/arginine (heavy media), so that the cellular proteomes incorporated light or heavy lysine and arginine. The heavy amino acids introduce a mass shift, which is distinguishable and quantifiable by MS (29). The cell lysates containing light isotope-or heavy isotope-labeled proteins were incubated with GST or a GST fusion protein containing residues 531-826 of A, immunoblot assays were performed to determine FLAG-HIF-1␣, HIF-1␤, Hsp70-V5, and actin protein levels in co-transfected HEK293T cells exposed to 20% or 1% O 2 for 4 h. EV, empty vector. B, immunoblot assays were performed to determine HIF-1␣, Hsp70-V5, and actin protein levels in transfected HEK293T cells exposed to 20% or 1% O 2 for 4 h. C, HEK293 cells were co-transfected with HIF-1-dependent firefly luciferase reporter p2.1, control reporter pSV-Renilla, and EV or Hsp70-V5 expression vector and exposed to 20% or 1% O 2 for 24 h. The ratio of firefly/Renilla luciferase activity was determined and normalized to the nonhypoxic EV condition (mean Ϯ S.E., n ϭ 5). *, p Ͻ 0.05; ***, p Ͻ 0.001 compared with EV. D, expression of GLUT1, vascular endothelial growth factor (VEGF), and Hsp70 mRNA and 18 S rRNA was analyzed by reverse transcription-PCR in transfected HEK293T cells exposed to 20% or 1% O 2 for 24 h. E, HEK293T cells were co-transfected with vectors encoding His-ubiquitin, FLAG-HIF-1␣, and Hsp70 shRNA (shHsp70) or scrambled control shRNA (shSC) and treated with 10 M MG132 for 8 h. Whole cell lysates were used for co-immunoprecipitation (IP). Band intensities of the ubiquitinated HIF-1␣ species (indicated by vertical bar at right) were quantified by densitometry and normalized to shSC. F, immunoblot assays were performed to determine FLAG-HIF-1␣, Hsp70-V5, and actin protein levels in co-transfected HEK293T cells exposed to 20% or 1% O 2 for 4 h in the presence or absence of MG132.
human HIF-1␣, respectively, immobilized on glutathione-Sepharose beads. After extensive washing, proteins that bound to GST or GST-HIF-1␣ were boiled in sample buffer, mixed, and fractionated by SDS-PAGE. Following in-gel trypsin digestion and peptide extraction, the tryptic peptide mixtures were analyzed by liquid chromatography-tandem mass spectrometry (Fig. 1A). The peptide spectra were searched against the NCBI Protein Database by Mascot search engine, verified, and quantified by MSQuant software. Among the identified proteins that showed a ratio of heavy to light peak Ն2, we found known HIF-1␣-binding proteins, including the VHL-Elongin-C-Elongin-B-Cullin-2 complex, factor inhibiting HIF-1, and p300 (data not shown), which validated our approach. Hsp70 and CHIP exhibited SILAC ratios of 10 and 4.5, respectively (Fig. 1, B and D). The tandem MS spectrum demonstrated the accurate assignment of each representative peptide of Hsp70 or CHIP (Fig. 1, C and E). Together, MS data indicate that Hsp70 and CHIP are HIF-1␣-interacting proteins. Hsp70 is a widely expressed, inducible heat shock protein, which is involved in protein folding and unfolding, signal transduction, cell survival, and inflammation (30). CHIP is an Hsp70-associated E3 ubiquitin ligase that promotes protein degradation by the 26 S proteasome (31).
To confirm the proteomic finding, we performed a co-immunoprecipitation assay. HEK293T cells were co-transfected with FLAG-tagged HIF-1␣ and V5-tagged CHIP expression vectors and treated with the proteasome inhibitor MG132. CHIP-V5 was specifically precipitated from cell lysates by anti-FLAG antibody but not by control IgG (Fig. 2A). Endogenous Hsp70 was also detected in immunoprecipitates of FLAG-HIF-1␣ ( Fig. 2A). Conversely, FLAG-HIF-1␣ was pulled down with CHIP-V5 from lysates of co-transfected HEK293T cells by anti-V5 antibody (Fig. 2B). Endogenous Hsp70 was also precipitated by anti-V5 antibody (Fig. 2B). These data indicate that Hsp70/CHIP and HIF-1␣ form a protein complex in human cells. In vitro GST pulldown assays further revealed that V5/His-tagged Hsp70 purified by Ni-NTA beads from transfected HEK293 cell lysates associated with purified GST-HIF-1␣ but not with GST indicating that Hsp70-HIF-1␣ interaction is direct (supplemental Fig. S1).
To map the HIF-1␣ domains that are involved in Hsp70/ CHIP binding, we performed in vitro GST pulldown assays. GST-HIF-1␣ fusion proteins were purified from bacteria and incubated with lysates from HEK293T cells expressing CHIP-V5 and endogenous Hsp70. As shown in Fig. 2C, Hsp70 bound strongly to GST fusion proteins containing HIF-1␣ amino acid residues 331-427, 432-528, and 531-826 but not to GST. CHIP-V5 bound strongly to HIF-1␣ residues 81-200 and 201-329. These results suggest that the initial identification of CHIP based on interaction with HIF-1␣-(531-826) in the proteomic screen was probably due to dual interaction of Hsp70 with both HIF-1␣-(531-826) and CHIP.
HIF-1 regulates the transcription of numerous hypoxiaadaptive genes (5). To test the effect of Hsp70 on HIF-1 transcriptional activity, HEK293 cells were co-transfected with the following: HIF-1-dependent reporter plasmid p2.1, which contains a 68-bp HRE from the human ENO1 gene upstream of SV40 promoter and firefly luciferase coding sequences (32); FIGURE 5. Hsp70 knockdown increases HIF-1␣ protein levels and HIF-1dependent gene transcription during prolonged hypoxia. A, HEK293T cells were transfected with expression vector encoding shRNA against Hsp70 (shHsp70) or a scrambled control shRNA (shSC) and exposed to 20% or 1% O 2 for the indicated time. HIF-1␣ and actin levels were determined by immunoblot assays and quantified by densitometry. The ratio of HIF-1␣/actin was normalized to the nonhypoxic shSC condition (mean Ϯ S.E., n ϭ 3). **, p Ͻ 0.01, relative to shSC. B, HEK293T cells were transfected with shHsp70 or shSC and exposed to 20% or 1% O 2 for 24 h, followed by exposure to 20% O 2 for the indicated time. HIF-1␣ and actin levels were determined by immunoblot assays and quantified by densitometry. The ratio of HIF-1␣/actin was normalized to hypoxic shSC group (mean Ϯ S.E., n ϭ 3). *, p Ͻ 0.05; ***, p Ͻ 0.001, relative to shSC. C, HEK293 cells were co-transfected with p2.1, pSV-Renilla, and expression vector encoding shSC or shHsp70 and exposed to 20% or 1% O 2 for 24 h. The ratio of firefly/Renilla luciferase activity was determined and normalized to nonhypoxic shSC condition (mean Ϯ S.E., n ϭ 4). ***, p Ͻ 0.001. control plasmid pSV-Renilla, where Renilla luciferase expression is driven by the SV40 promoter alone; and EV or Hsp70-V5 expression vector. The transcriptional activity of HIF-1 was significantly decreased, at both 20% and 1% O 2 when Hsp70-V5 was overexpressed in HEK293 cells (Fig. 3C). Analysis of HIF-1 target gene expression by reverse transcription-PCR demonstrated that Hsp70-V5 reduced expression of GLUT1 and vascular endothelial growth factor mRNA, but it had no effect on 18 S rRNA levels in HEK293T cells exposed to 20% or 1% O 2 (Fig. 3D). Interestingly, Hsp70 mRNA levels were also increased in hypoxic cells (Fig. 3D), suggesting that Hsp70 may be encoded by a HIF-1 target gene. Taken together, these data indicate that Hsp70 reduces HIF-1␣ protein levels, thereby decreasing HIF-1 transcriptional activity and downstream target gene expression.
To investigate the mechanism underlying Hsp70-mediated HIF-1␣ down-regulation, we first studied the effect of Hsp70 on HIF-1␣ ubiquitination using an shRNA targeting both the Hsp70-1A and Hsp70-1B isoforms (shHsp70) or a scrambled control shRNA (shSC) that does not target any known human gene (28). shHsp70, but not shSC, markedly reduced levels of Hsp70 mRNA and protein in HEK293T cells (supplemental Fig.  S2E). The levels of 18 S rRNA and actin protein were not affected by shHsp70. HEK293T cells were co-transfected with vectors encoding His-tagged ubiquitin, FLAG-HIF-1␣, and shHsp70 or shSC and treated with the proteasome inhibitor MG132 to block degradation of ubiquitinated HIF-1␣. Analysis of immunoprecipitated FLAG-HIF-1␣ by anti-His antibody demonstrated that ubiquitination of FLAG-HIF-1␣ was inhibited by 40% in Hsp70 knockdown cells, compared with cells transfected with shSC, whereas total FLAG-HIF-1␣ levels were similar (Fig. 3E). Further experiments revealed that MG132 treatment inhibited FLAG-HIF-1␣ protein degradation mediated by Hsp70-V5 in HEK293T cells at both 20% and 1% O 2 ( Fig. 3F and supplemental Fig. S2C). Taken together, our gain-of-function data indicate that Hsp70 pro- FIGURE 7. CHIP promotes ubiquitination and degradation of HIF-1␣. A, immunoblot assays were performed to determine FLAG-HIF-1␣, HIF-1␤, CHIP-V5, and actin protein levels in co-transfected HEK293T cells exposed to 20% or 1% O 2 for 8 h. B, immunoblot assays were performed to determine HIF-1␣, CHIP-V5, and actin protein levels in transfected HEK293T cells exposed to 20% or 1% O 2 for 8 h. C, immunoblot assays were performed to determine HIF-1␣, CHIP-V5, or CHIP (H260Q)-V5 and actin protein levels in transfected HEK293T cells exposed to 20% or 1% O 2 for 8 h. D, immunoblot assays were performed to determine HIF-1␣, CHIP-V5, and actin protein levels in transfected HEK293T cells exposed to 20% or 1% O 2 for 8 h in the presence or absence of 10 M MG132. E, HEK293T cells were co-transfected with vector encoding His-ubiquitin, FLAG-HIF-1␣, shSC, shHsp70, shCHIP1172, or shCHIP501 and treated with MG132 for 8 h. Total ubiquitinated proteins were precipitated from WCLs by Ni-NTA beads and analyzed by immunoblot assay with anti-FLAG antibody. FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 motes ubiquitination and degradation of HIF-1␣ through a 26 S proteasome-dependent pathway.
To determine whether endogenous Hsp70 physiologically regulates HIF-1␣ protein turnover, we next performed loss-of-function studies using shHsp70. HIF-1␣ protein levels were increased after 4 h of hypoxia and then progressively decreased in shSCtransfected HEK293T cells (Fig.  5A). However, Hsp70 knockdown led to a delay in, and reduced magnitude of, the down-regulation of HIF-1␣ protein levels during prolonged hypoxia (Fig. 5A). Reoxygenation time course experiments showed that Hsp70 knockdown did not affect the rate of post-hypoxic decay of HIF-1␣ protein in HEK293T cells, as compared with that in shSC-transfected cells (Fig.  5B). The increased HIF-1␣ levels at time points up to 12 min were due to the higher starting levels of HIF-1␣ in shHsp70 cells. The p2.1 HRE reporter assay revealed that Hsp70 knockdown by shHsp70 significantly increased the transcriptional activity of HIF-1 in HEK293 cells that were subjected to hypoxia for 24 h (Fig. 5C). Taken together, the loss-of-function data indicate that Hsp70 physiologically regulates HIF-1␣ protein stability and transcriptional activity in hypoxic human cells but does not affect posthypoxic decay of HIF-1␣.

JOURNAL OF BIOLOGICAL CHEMISTRY 3659
Hsp70-HIF-1␣ interaction abolishes Hsp70-induced degradation of HIF-1␣, suggesting that their interaction is required for the physiological regulation of HIF-1␣ by Hsp70.
CHIP is a U-box-dependent E3 ubiquitin ligase (31). To determine whether the ubiquitin ligase activity is required for CHIP to promote HIF-1␣ degradation, we used the dominant-negative CHIP(H260Q), a U-box mutant that does not bind to the E2 ubiquitin-conjugating enzyme and lacks E3 activity (33). Expression of CHIP(H260Q)-V5 led to increased HIF-1␣ protein levels in hypoxic HEK293T cells (Fig. 7C). Similar results were also observed when U-box domain-deleted CHIP-V5 was expressed in HEK293T cells (data not shown). Thus, the ubiquitin ligase activity of CHIP regulates HIF-1␣ degradation. MG132 treatment also blocked CHIP-induced HIF-1␣ degradation in hypoxic HEK293T cells (Fig. 7D and supplemental Fig. S3C). We further investigated HIF-1␣ ubiquitination by CHIP loss-of-function using two shRNAs targeting different CHIP mRNA sequences. In HEK293T cells, shCHIP1172 and shCHIP501 both efficiently knocked down levels of CHIP mRNA and protein, compared with the negative control shSC and the untransfected control (supplemental Fig. S3D). Next, HEK293T cells were co-transfected with vectors encoding His ubiquitin, FLAG-HIF-1␣, shSC, shCHIP1172, or shCHIP501 and then treated with

HIF-1␣ Degradation by Hsp70 and CHIP
MG132. Ubiquitinated proteins were precipitated by Ni-NTA beads from whole cell lysates. Analysis of Ni-NTA precipitates with anti-FLAG antibody revealed that FLAG-HIF-1␣ was highly ubiquitinated in nonhypoxic HEK293T cells transfected with shSC. However, CHIP knockdown markedly reduced FLAG-HIF-1␣ ubiquitination, similar to the effect of Hsp70 knockdown (Fig. 7E). Taken together, these results indicate that CHIP promotes degradation of HIF-1␣ by a ubiquitinand proteasome-dependent mechanism.
To complement the CHIP gain-of-function data, we performed loss-of-function analysis. Relative to shSC, transfection of shCHIP1172 increased levels of HIF-1␣ and impaired HIF-1␣ protein decay during prolonged hypoxia (Fig. 9A). Although HIF-1␣ protein levels were increased in CHIP knockdown cells during the first 10 min of reoxygenation following 24 h of hypoxia, loss of CHIP did not attenuate the rate of HIF-1␣ degradation (Fig. 9B). Similar results were also observed in HEK293T cells transfected with shCHIP501 (supplemental Fig. S3, E-H), which indicates that the observed effects were specifically due to CHIP loss of function. The p2.1 HRE reporter assay demonstrated that CHIP knockdown by shCHIP501 increased HIF-1 transcriptional activity in HEK293 cells at 20% and 1% O 2 (Fig. 9C). Therefore, as in the case for Hsp70 (Fig. 5), CHIP knockdown increases HIF-1␣ protein levels and HIF-1 transcriptional activity.
Hsp70 Recruits CHIP and Cooperates with CHIP to Regulate HIF-1␣ Degradation-The experiments described thus far demonstrate that both Hsp70 and CHIP mediate HIF-1␣ degradation. CHIP associates with Hsp70 and functions as a chaperone-dependent E3 ubiquitin ligase (30,31). Therefore, we hypothesized that CHIP mediates Hsp70-induced proteasomal degradation of HIF-1␣. To test this hypothesis, we investigated the effect of Hsp70 on HIF-1␣ stability in CHIP-deficient cells. Hsp70 significantly inhibited HIF-1␣ protein induction by hypoxia in HEK293T cells transfected with shSC (Fig. 10, A and  B), consistent with data presented above. However, CHIP knockdown by shCHIP501 prevented Hsp70-induced HIF-1␣ degradation in hypoxic HEK293T cells (Fig. 10, A and B). Similar results were obtained using shCHIP1172 (data not shown).
These results indicate that CHIP is required for Hsp70-induced degradation of HIF-1␣.
The finding that Hsp70 is required for CHIP-mediated HIF-1␣ degradation suggested that CHIP-HIF-1␣ interaction may depend on Hsp70. As shown in Fig. 10F, the interaction between CHIP-V5 and GST-HIF-1␣-(1-826) was dramatically reduced by the K30A mutation, which disrupts the binding of CHIP to Hsp70 (33). These data demonstrate that stable CHIP-HIF-1␣ interaction in human cells requires CHIP-Hsp70 interaction, indicating that Hsp70 recruits CHIP to HIF-1␣.
Hsp70 and CHIP Fail to Regulate HIF-2␣ Stability-To investigate whether Hsp70 and CHIP have a similar effect on HIF-2␣ stability, we first performed a co-immunoprecipitation assay. HEK293T cells were co-transfected with HIF-2␣ and CHIP-V5 expression vectors and treated with MG132. Hsp70 co-immunoprecipitated with HIF-2␣ (Fig.  11A), and its binding to HIF-2␣ was comparable with its binding to HIF-1␣ (Fig. 11B). However, co-immunoprecipitation of CHIP-V5 with HIF-2␣ was markedly reduced (Fig.  11, A and B) compared with its interaction with HIF-1␣ ( Fig.  2A). Western blot analysis demonstrated that expression of Hsp70-V5 or CHIP-V5 did not reduce HIF-2␣ protein levels in hypoxic HEK293T cells (Fig. 11, C and D). Loss-of-function studies revealed that knockdown of either Hsp70 (Fig.  11E) or CHIP (Fig. 11F) also failed to affect HIF-2␣ protein levels in nonhypoxic and hypoxic HEK293T cells. These results indicate that Hsp70 and CHIP selectively regulate the protein stability of HIF-1␣ but not HIF-2␣. FIGURE 11. Hsp70 and CHIP fail to regulate HIF-2␣ protein stability. A, co-immunoprecipitation (IP) of HIF-2␣ and CHIP-V5 or Hsp70 from transfected HEK293T cells treated with 10 M MG132 for 8 h was performed. B, binding of Hsp70 and CHIP to HIF-2␣ (as shown in A) or to HIF-1␣ (as shown in Fig. 2A) was quantified by densitometric analysis of the immunoblots. The band intensity for the precipitated protein was normalized to the band density for the total protein in WCL used for immunoprecipitation (mean Ϯ S.E., n ϭ 3). **, p Ͻ 0.01, compared with HIF-1␣. C, immunoblot assays were performed to determine HIF-2␣, Hsp70-V5, and actin protein levels in transfected HEK293T cells exposed to 20% or 1% O 2 for 4 h. D, immunoblot assays were performed to determine HIF-2␣, CHIP-V5, and actin protein levels in transfected HEK293T cells exposed to 20% or 1% O 2 for 8 h. E, immunoblot assays were performed to determine HIF-2␣, Hsp70, and actin protein levels in transfected HEK293T cells exposed to 20% or 1% O 2 for 16 h. F, immunoblot assays were performed to determine HIF-2␣, CHIP, and actin protein levels in transfected HEK293T cells exposed to 20% or 1% O 2 for 16 h. Representative blots from three experiments are shown. G, mechanism by which Hsp70 induces the ubiquitination and degradation of HIF-1␣ through recruitment of CHIP is shown.