HIF1α Protein Stability Is Increased by Acetylation at Lysine 709*

Background: HIF1α and p300 are key components of HIF-1 transcription complex. Results: Lysine 709 of HIF1α is acetylated by p300, which increases protein stability and HIF-1 activity. Conclusion: p300 has a novel function in stabilizing HIF1α by Lys-709 acetylation. Significance: New insights in how HIF1α is post-translationally regulated by its cofactor to ensure HIF-1 activity. Lysine acetylation regulates protein stability and function. p300 is a component of the HIF-1 transcriptional complex and positively regulates the transactivation of HIF-1. Here, we show a novel molecular mechanism by which p300 facilitates HIF-1 activity. p300 increases HIF-1α (HIF1α) protein acetylation and stability. The regulation can be opposed by HDAC1, but not by HDAC3, and is abrogated by disrupting HIF1α-p300 interaction. Mechanistically, p300 specifically acetylates HIF1α at Lys-709, which increases the protein stability and decreases polyubiquitination in both normoxia and hypoxia. Compared with the wild-type protein, a HIF1α K709A mutant protein is more stable, less polyubiquitinated, and less dependent on p300. Overexpression of the HIF1α wild-type or K709A mutant in cancer cells lacking the endogenous HIF1α shows that the K709A mutant is transcriptionally more active toward the HIF-1 reporter and some endogenous target genes. Cancer cells containing the K709A mutant are less sensitive to hypoxia-induced growth arrest than the cells containing the HIF1α wild-type. Taken together, these data demonstrate a novel biological consequence upon HIF1α-p300 interaction, in which HIF1α can be stabilized by p300 via Lys-709 acetylation.

Hypoxia-inducible factor 1 ␣ (HIF1␣) 2 is a critical component of the HIF-1 transcriptional complex that regulates cellular response to hypoxia (1). In non-hypoxic conditions, HIF1␣ protein is constantly synthesized, but rapidly degraded by oxygen. The primary mechanism consists of the post translational hydroxylation of HIF1␣ at proline residues 402 and 564 by oxygen-dependent prolyl hydroxylases, which is followed by von Hippel-Lindau (VHL)-mediated polyubiquitination and 26 S proteasome-dependent protein degradation (2). In hypoxic conditions, oxygen is not available for the hydroxylation of HIF1␣ protein, which is stabilized and dimerized with the oxy-gen-independent HIF1␤ to form the core of HIF-1 transcriptional complex (2,3). The HIF-1 target genes regulate a vast variety of biological activities that are critical for cell viability, growth, and differentiation (1). Therefore, the inappropriate accumulation of HIF1␣ proteins in cancer cells can lead to inappropriate HIF-1 activities, which underpins malignant phenotypes including angiogenesis, metabolic reprogramming, metastasis, and therapy resistance (4,5).
HIF1␣ protein stability can also be regulated by additional mechanisms that are independent and cross-talk with the oxygen/VHL-mediated HIF1␣ protein degradation pathway. HSP90 is a key cellular chaperone and is positively associated with HIF1␣ proteins in both normoxia and hypoxia (6). In normoxia, the HIF1␣ chaperone function of HSP90 is countered by RACK1 (7). Although significantly stabilized in hypoxia due to the lack of oxygen/VHL pathway, the HIF1␣ protein can still be reduced by HSP70-CHIP via polyubiquitination and degradation (8). In addition, the HIF1␣ protein stability and HIF-1 transcriptional activity are associated with the protein lysine acetylation-deacetylation system (9). Multiple members within the lysine acetylase and deacetylase families have been reported to interact with HIF1␣ (10 -14). The biological consequence is a change of HIF-1 activity due to a change in HIF1␣ protein stability in many cases (9,15,16). Lysine acetylation changes protein stability and function (17). Multiple acetylation sites have been discovered within the HIF1␣ protein (11,14,18). However, the biological consequence due to these different acetylation sites can be very different. Currently, both the lysine acetylase and deacetylase are potential therapeutic targets against cancer and other human diseases, in which HIF1␣ plays a pathological role. It is important to delineate the functional consequence of HIF1␣ acetylation at specific lysine locations.
p300 is a critical component in the HIF-1 transcriptional complex (19). In the non-hypoxic condition, p300 and HIF1␣ are not physically associated due to the rapid HIF1␣ protein degradation and the hydroxylation at asparagine 803 within the HIF1␣ (20). Recently, the acetylation at lysine 674 of HIF1␣ by PCAF has also been reported to play a role in determining HIF1␣-p300 interaction (14). Upon hypoxia, the C-terminal of HIF1␣ interacts with the CH1 (cysteine/histidine-rich 1) domain of p300, which leads to an increase of HIF-1 transcriptional activity (21). A small molecule inhibitor, chetomin, has been reported to disrupt the interaction between HIF1␣ and p300 and inhibit the HIF-1 target gene expression (22). p300 is a multiple domain protein with acetylase activity (23). Despite being closely associated with HIF1␣ in hypoxia, it is unclear whether p300 has a direct effect on HIF1␣. In the current study, we identified a novel function of p300 in stabilizing HIF1␣ proteins by lysine acetylation at Lys-709 within HIF1␣.
Plasmid Construction-Alanine substitutions in HIF1␣ protein were generated by two-step PCR amplification using FLAG-HIF1␣ wild-type plasmid as a template. The PCR products carrying the mutant HIF1␣ fragments were digested with restriction enzymes and inserted into the FLAG empty vector used for carrying the wild-type FLAG-HIF1␣. Each mutation was verified by sequencing.
Mass Spectrometry-Mass spectrometry was performed by Proteomics Shared Resource at Oregon Health & Science University. Briefly, HIF1␣ proteins were recovered from SDS-PAGE gel, reduced with DTT, and digested by trypsin. Each protein digest was analyzed by LC-MS/MS using an Agilent 1100 series capillary LC system (Agilent Technologies) and a VELOS linear ion trap mass spectrometer (Thermo Fisher). Electrospray ionization was performed with an ion max source fitted with a 34-gauge metal needle (Thermo Fisher). MS/MS spectra data were searched using SEQUEST against a database containing human protein sequences from UniProtKB/ Swiss-Prot. In-house analysis software was used to calculate discriminant scores and filter out incorrect peptide identifications using sequence-reversed matches to estimate false discovery rates. Scaffold (Proteome Software, Portland, OR) was used to validate protein identifications derived from MS/MS sequencing results.
In Vitro Acetylation Assay-HIF1␣ and p300, purified by immunoprecipitation, were mixed and incubated at 30°C in buffer (25 mM Tris-HCl, pH 8.0, 0.5 mM EDTA, 5 mM DTT, and 0.7 mM acetyl-CoA). Reactions were stopped at indicated time by adding 6ϫ SDS loading buffer and analyzed by Western blots.
Cycloheximide HIF1␣ Half-life-HIF1␣ and GFP proteins were expressed in Hek293T cells. Following exposure to normal or hypoxic condition for 5 h, cells were treated with cycloheximide and harvested at different time points. Western blots were used to detect HIF1␣ and GFP proteins, which were quantified using LI-COR odyssey Infrared software (version 2.0) as described previously (18).
Real-time PCR and Cell Proliferation-The ⌬⌬C t method was used to calculate mRNA fold change. The trypan blue exclusion method was used to calculate viable cells in the proliferation experiments as described (18,24).
Statistical Analysis-Differences between the means of unpaired samples were evaluated by the Student's t test.

RESULTS
p300 Up-regulates the HIF1␣ Protein Level-Although p300 is known to increase HIF-1 transcriptional activity, its effect on HIF1␣ protein is less clear. In the cancer cell line U2OS, we used siRNA to knockdown p300 and cultured the cells in either normal ambient oxygen (21% O 2 ) or hypoxic (1% O 2 ) conditions for 6 h to induce HIF1␣ protein accumulation. We found that p300 siRNA knockdown (si-p300) significantly decreased the HIF1␣ in hypoxia (Fig. 1A, U2OS, and supplemental Fig. S1). This decrease was on the protein level because the mRNA of HIF1␣ was not changed by p300 knockdown (supplemental Fig.  S1). We also tested the effect of p300 siRNA on HIF1␣ proteins in a renal cell carcinoma cell line RCC10 (25), in which a loss of function in VHL renders HIF1␣ protein stable in normal oxygen condition (Fig. 1A, RCC10). The siRNA knockdown of p300 significantly decreased HIF1␣ proteins in both normal and hypoxic conditions (Fig. 1A, RCC10). Next, we measured the effect of p300 overexpression on HIF1␣ protein levels. In VHL wild-type U2OS cells, p300 overexpression increased HIF1␣ proteins in the hypoxic condition (Fig. 1B, U2OS). In VHL-negative RCC10 cells, the HIF1␣ protein was significantly increased by p300 transfection in both normal and hypoxic conditions (Fig. 1B, RCC10). The effect of p300 on HIF1␣ appears to be specific because HIF2␣ and HIF1␤ protein levels were not changed by p300 (supplemental Fig. S2). Next, we co-overexpressed a FLAG-tagged HIF1␣ wild-type plasmid with either empty vector (Ev) or p300 in Hek293T cells. To ensure the regulation of FLAG-HIF1␣ by p300 is specific, we also included a plasmid encoding the green florescence protein (GFP) that is regulated by the same promoter of FLAG-HIF1␣ as a control. In Ev cells, hypoxia increased the FLAG-HIF1␣ protein level, and the overexpressed HIF1␣ protein was present in cells cultured in normal oxygen condition (Fig. 1C); presumably, this was because that the vector-based FLAG-HIF1␣ overexpression was very robust and overcame the endogenous oxygen/VHL-mediated HIF1␣ degradation. Importantly, p300 overexpression further increased the FLAG-HIF1␣ in both normal and hypoxic conditions (Fig. 1C). As a control, the co-overexpressed GFP was not changed by either hypoxia or p300 overexpression. To further confirm the HIF1␣ regulation by p300, we transfected the FLAG-HIF1␣ into Hek293T cells and treated cells with chetomin, a chemical inhibitor that disrupts the HIF1␣-p300 interaction (22). We found that chetomin treatment decreased the FLAG-HIF1␣ protein expression in both normal and hypoxic conditions (Fig. 1D). These data suggest that p300 can enhance the HIF1␣ stability in hypoxia or in conditions that the oxygen/VHL-mediated HIF1␣ degradation is either inactive (RCC10 cell line) or overwhelmed (vectorbased HIF1␣ overexpression).
p300 Acetylates HIF1␣-One of the primary functions of p300 is protein lysine acetylation (23). To understand the molecular mechanism underlying the HIF1␣ up-regulation by p300, we transfected the FLAG-HIF1␣ and p300 in Hek293T cells, immunoprecipitated the FLAG-HIF1␣ protein from whole cell lysates (WCL), and measured the HIF1␣ lysine acetylation and its interaction with p300. In agreement with the reported role of p300, in Ev cells, hypoxia increased the interaction between FLAG-HIF1␣ and the endogenous p300 ( Fig.  2A). The overexpressed p300 can also interact with FLAG-HIF1␣ proteins in normoxia, and the interaction was further increased by hypoxia ( Fig. 2A). Importantly, the level of FLAG-HIF1␣ lysine acetylation was increased by hypoxia in the Ev cells and was significantly increased by p300 overexpression in both normal and hypoxic conditions ( Fig. 2A). Because both hypoxia and p300 overexpression increased the FLAG-HIF1␣ protein level (Figs. 1C and 2A), we wanted to know whether the increase of HIF1␣ acetylation was due to the increase of total HIF1␣ proteins or the increase of de novo HIF1␣ acetylation. We performed an in vitro HIF1␣ acetylation assay by mixing equal amount of IP-purified FLAG-HIF1␣ proteins with IPpurified p300 proteins in the acetylation assay buffer. After 30 min of incubation at 30°C, the acetylation level of FLAG-HIF1␣ was measured by Western blot. As shown in Fig. 2B, the de novo HIF1␣ acetylation was significantly increased in vitro by incubating with the p300 immunocomplex compared with IgG complex, whereas the total FLAG-HIF1␣ input remained the same. Next, we repeated the HIF1␣ and p300 co-overexpression experiment as in Fig. 2A using a plasmid coding for a FLAG-HIF1␣ proline double mutant (HIF1␣-dm) that is resistant to the oxygen/VHL-mediated degradation. In Ev cells, the HIF1␣-dm protein level in WCL was not significantly changed between normal and hypoxic conditions, and hypoxia increased the interaction between HIF1␣-dm and the endogenous p300 ( Fig. 2C) and increased the HIF1␣-dm protein acetylation in the IP samples (Fig. 2C). Co-overexpression of p300 slightly increased the HIF1␣-dm total protein levels in the WCL, increased the interaction between HIF1␣ and p300, and significantly increased the HIF1␣-dm protein lysine acetylation (Fig.  2C). These results indicated de novo acetylation of HIF1␣ by p300. To test whether the HIF1␣-p300 interaction is required for the increase of HIF1␣ protein acetylation, we transfected FLAG-HIF1␣ into Hek293T cells and treated cells with chetomin. In hypoxia, chetomin treatment significantly reduced the HIF1␣-p300 interaction, which coincided with the reduction of HIF1␣ acetylation and total HIF1␣ protein level (Fig. 2D). Because HIF1␣ can also be associated with lysine deacetylases (26,27), we further tested whether the p300-induced acetylation can be countered by deacetylation. We co-overexpressed FLAG-HIF1␣ with p300 or in combination with either HDAC1 or HDAC3. We found that HDAC1, but not HDAC3, significantly reduced the p300-induced HIF1␣ acetylation, which also coincided with a reduction of total FLAG-HIF1␣ protein level (Fig. 2E). Recently, HIF1␣ was reported to be acetylated by a p300-interacting protein, PCAF (14). We compared the ability of p300 and PCAF in terms of HIF1␣ acetylation. When an HA-tagged HIF1␣ was co-overexpressed with either p300 or FLAG-PCAF, we observed that the HIF1␣-PCAF interaction was not changed by hypoxia, and the interaction was less robust than HIF1␣-p300 (Fig. 2F). Furthermore, p300 also induced a more robust HIF1␣ acetylation than FLAG-PCAF (Fig. 2F). Taken together, these data suggest that HIF1␣-p300 interaction has two novel biological consequences: the increases of HIF1␣ lysine acetylation and protein stability, which are FIGURE 1. p300 regulates HIF1␣ proteins. A and B, Western blot analysis of the p300 effect on endogenous HIF1␣ in U2OS and RCC10 cell lines. Cells were transfected with siRNA control (siC) or si-p300 in A, and empty vector or vector encoding p300 in B. C, Western blot analysis of Hek293T cells co-transfected with FLAG-HIF1␣ and p300 or Ev. D, Western blot analysis of Hek293T cells that were transfected with FLAG-HIF1␣ and treated by 50 nM p300 inhibitor, chetomin (CTM), or solvent control (V) for 6 h. In A-D, cells were cultured under normal or hypoxic conditions for 6 h before whole cell protein lysates were harvested. In C and D, a vector encoding GFP was used for transfection efficiency controls. The HIF1␣ protein level (R) was calculated based on the ratio of band intensity of HIF1␣/tubulin or GFP. All data were representatives of at least three independent experiments. dependent on HIF1␣-p300 interaction that can be achieved by hypoxia or p300 overexpression.
Lysine 709 of HIF1␣ Is the Acetylation Target of p300-To identify the p300 acetylated lysine residue(s) within HIF1␣, we co-transfected the full-length wild-type FLAG-HIF1␣ with either p300 or Ev in Hek293T cells. Next, we immunoprecipitated and gel-purified FLAG-HIF1␣ proteins, digested them with trypsin, and analyzed the samples by liquid chromatography-tandem mass spectrometry (LC-MS/MS). We observed that lysine 674 within the FLAG-HIF1␣ was acetylated in both Ev and p300-overexpressed samples (data not shown). Significantly, lysine 709 was only acetylated in p300-overexpressed cells ( Fig. 3A and supplemental Fig. S3). Lys-674 has recently been reported as an acetylation target for PCAF (14), and Lys-709 is novel. Based on the FLAG-HIF1␣ plasmid, we used sitedirected mutagenesis to convert lysine to alanine and generated FLAG-HIF1␣ mutants containing a single point mutation of either K674A or K709A. Then, we overexpressed FLAG-FIGURE 2. p300 acetylates HIF1␣. A, Hek293T cells were co-transfected with FLAG-HIF1␣ and p300 or Ev. 48 h after transfection, cells were exposed to either normoxia (n) or hypoxia (h) for 6 h. WCLs were harvested and used for Western blot (IB) or IP with anti-FLAG antibodies and followed by Western blot analysis. Protein acetylation was detected by anti-ace-lysine (Ace-K) antibodies. A GFP vector was also co-transfected as control. B, FLAG-HIF1a and p300 were expressed in Hek293T cells separately and IP purified by anti-FLAG or anti-p300. The immunocomplex containing FLAG-HIF1␣ was mixed with p300 or IgG immunocomplex for 0 or 30 min in the in vitro acetylation buffer. Afterward, the acetylated (Ace) HIF1␣, the total p300, and FLAG-HIF1␣ protein levels were measured by Western blot. C, similar experiments as described in A were done using a vector encoding for FLAG-HIF1␣ proline mutant (HIF1␣-dm) containing proline to alanine mutation at 402 and 564. D, FLAG-HIF1␣ was overexpressed in Hek293T cells. 24 h later, cells were treated with vehicle (V) or 50 nM of chetomin (CTM) and exposed to hypoxia for 6 h. Afterward, WCL was used for anti-FLAG IP and Western blots. E, FLAG-HIF1␣ was co-transfected with Ev and p300 with or without HDAC1 or HDAC3. 48 h after transfection, WCLs were harvested, and used for Western blot (IB) or IP with anti-FLAG antibodies and followed by Western blot analysis. F, a vector encoding for HA-HIF1␣ was co-overexpressed with Ev, p300, or PCAF in H293T cells. 48 h after transfection, cells were exposed to normal (n) or hypoxia (h) for 6 h, and the HA-HIF1␣ was IP-purified from WCL and analyzed by Western blot (IB). All data were representative of at least three independent experiments. OCTOBER 12, 2012 • VOLUME 287 • NUMBER 42

JOURNAL OF BIOLOGICAL CHEMISTRY 35499
HIF1␣-WT, FLAG-HIF1␣-K674A, or FLAG-HIF1␣-K709A in Hek293T cells with either Ev control or p300. Forty-eight hours after transfection, the FLAG-HIF1␣ wild-type and mutant proteins were immunopurified from the whole cell lysates. Western blots showed that all HIF1␣ proteins interact with p300 similarly (Fig. 3B). In both FLAG-HIF1␣-WT and FLAG-HIF1␣-K674A mutant, the acetylation levels were robustly increased by p300 (Fig. 3B). In contrast, the acetylation level of FLAG-HIF1␣-K709A mutant was not significantly increased by p300 (Fig. 3B). To further confirm the specific acetylation at Lys-709 by p300, we co-overexpressed HIF1␣-WT or Lys-709 mutant with either the wild-type p300 or a p300 mutant that is defective in the acetyltransferase activity (28). Similar to the results in Fig. 3B, p300 wild-type was able to increase the acetylation in HIF1␣-WT, but not in HIF1␣-K709A mutant proteins (Fig. 3C). However, the p300 acetylase mutant was unable to increase the acetylation in HIF1␣-WT and K709A mutant (Fig.  3C). These data suggest that p300 acetylates HIF1␣ primarily at Lys-709.
p300 Regulates HIF1␣ Protein Stability via Lys-709 Acetylation-To investigate the biological significance of Lys-709 acetylation, we overexpressed equal amounts of FLAG-HIF1␣-WT or FLAG-HIF1␣-K709A mutant in Hek293T cells with GFP plasmids as transfection controls. Then, we measured the normal and hypoxic FLAG-HIF1␣ protein levels. We observed that the FLAG-HIF1␣ K709A mutant protein expressed at a higher level than FLAG-HIF1␣-WT in both normal and hypoxic conditions by Western blots coupled with florescence densitometry (Fig. 4A). Next, we used cycloheximide to inhibit protein synthesis and measured the HIF1␣ WT and K709A mutant protein stability in the absence (Ev) or presence of p300 overexpression. In normal oxygen condition, the K709A mutant protein was significantly more stable than HIF1␣-WT protein in the Ev cells (#, p Ͻ 0.01, Fig. 4B). p300 overexpression significantly increased the HIF1␣-WT protein stability (a, p Ͻ 0.01, Fig. 4B) but had no effect on HIF1␣-K709A mutant. A similar trend was observed in the hypoxic condition, in which the HIF1␣-K709A mutant was more stable than the HIF1␣-WT in Ev cells (##, p Ͻ 0.01, Fig. 4C), and p300 overexpression increased the FLAG-HIF1␣-WT protein stability (aa, p Ͻ 0.01, Fig. 4C) and had no effect on the K709A mutant (Fig. 4C).
Lys-709 within the HIF1␣ has been reported to be a site for polyubiquitination (29). We hypothesize that p300 increases HIF1␣-WT protein stability by acetylation at Lys-709, which prevents polyubiquitination and protein degradation. To investigate the relationship between polyubiquitination and acetylation at Lys-709, we performed similar co-overexpression experiments as in Fig. 4, B and C, and used proteasomal inhibitor MG132 to prevent FLAG-HIF1␣ degradation. FLAG-HIF1␣ FIGURE 3. HIF1␣ is acetylated by p300 at lysine 709. A, mass spectrum of HIF1␣ acetylation (Ace) by p300 at lysine 709. FLAG-HIF1␣ and p300 were co-overexpressed in Hek293T cells. The FLAG-HIF1␣ was IP and gel-purified. The gel slice containing FLAG-HIF1␣ was digested by trypsin and subjected to LC-MS/MS analysis. Mass spectrum of b ions and y ions were shown in red and blue, respectively. Acetylation at the location was marked with an ampersand by additional mass of 42 Da. Image was prepared with Scaffold. B, FLAG-HIF1␣ WT, FLAG-HIF1␣ K674A, and FLAG-HIF1␣ K709A was co-transfected with Ev (Ϫ) or p300 (ϩ) in Hek293T cells. 48 h later, WCLs were harvested, IP purified with anti-FLAG antibodies, and analyzed by Western blots. C, FLAG-HIF1␣ WT and HIF1␣ K709A were co-transfected with Ev (Ϫ), p300 WT, or p300 mutant that is defective in acetyltransferase activity. 48 h later, WCL was harvested, IP-purified by anti-FLAG antibodies, and analyzed by Western blots. Data in B and C were representatives of at least three independent experiments. mut, mutant.
WT and mutant proteins were IP-purified, and the level of polyubiquitination and acetylation were measured by Western blot. In normal oxygen condition, the polyubiquitination level of HIF1␣-WT was much higher than HIF1␣-K709A mutant in Ev cells (Fig. 5A). p300 overexpression reduced polyubiquitination and increased acetylation in HIF1␣-WT proteins (Fig. 5A). In contrast, p300 did not change the polyubiquitination and acetylation levels of the HIF1␣-K709A mutant (Fig. 5A). In hypoxic condition, although both wild-type and mutant HIF1␣ proteins were less polyubiquitinated than in normoxia, the HIF1␣-WT still had a significantly higher level of polyubiquitination than the HIF1␣-K709A mutant (Fig. 5A). p300 only decreased polyubiquitination and increased acetylation levels in the HIF1␣-WT, but not in the HIF1␣-K709A mutant (Fig.  5A). Next, we overexpressed HIF1␣-WT or K709A mutant in the presence of p300 siRNA in U2OS cells. In both normoxia and hypoxia, the loss of p300 significantly reduced the HIF1␣-WT protein level, and did not change the HIF1␣-K709A mutant (Fig. 5B). These data suggest that p300 regulates HIF1␣ protein stability by Lys-709 acetylation. Lys-709 within the HIF1␣ protein can be a target of both polyubiquitination and acetylation. Polyubiquitination at normal and hypoxic conditions cause HIF1␣ protein degradation. Acetylation by p300 or the point mutation (Lys3 Ala) eliminates the polyubiquitination possibility and leads to HIF1␣ protein stabilization.
Lys-709 Regulates HIF-1 Reporter and Target Gene Expression-We hypothesized that the up-regulation of HIF1␣ protein level due to Lys-709 acetylation can lead to an increase of HIF1␣ transcriptional activity. To directly compare the function of HIF1␣ WT and K709A mutant in normal and hypoxic conditions, we transduced Hek293 cells with lentivirus containing shRNA against the 3Ј-UTR of HIF1␣. As a result, the transduced Hek293 cell line contained stable shRNA knockdown against the endogenous, but not the exogenous, HIF1␣ (supplemental Fig. S4). Then, we performed HIF-1 activity reporter assay using the Dual-Luciferase plasmid system, in which the firefly luciferase reporter gene (p2.1) is under the control of hypoxia and HIF-1 and the Renilla luciferase gene is constitutive (30,31). The HIF1␣ shRNA effectively inhibited the HIF-1 activity driven by the endogenous HIF1␣ (supple- Then, cells were exposed to normal (N) or hypoxia (H) for 6 h, followed by Western blot analysis. A vector of GFP was co-transfected as the control. The band intensities of FLAG-HIF1␣ and GFP were quantified by fluorescence densitometry. For each sample, HIF1␣ signal was divided by the GFP signal, and normalized to WT at normoxia, expressed as mean and S.D. of at least three experiments. *, p Ͻ 0.05; **, p Ͻ 0.01, t test. B and C, the WT and K709A plasmids were co-transfected with Ev or p300 in Hek293T cells for 48 h. A vector of GFP was co-transfected as controls. Then cells were exposed to normal (B) or hypoxic mimetic CoCl 2 (C) for 6 h. Afterward, cells were treated by cycloheximide (CHX) for the indicated time, and WCL was harvested and analyzed by Western blots using anti-FLAG and anti-GFP antibodies. The FLAG-HIF1␣ WT and mutant protein stability was measured and expressed as in A. Line graphs in B and C were mean and S.D. of three experiments. p Ͻ 0.01 for all statistical comparisons. OCTOBER 12, 2012 • VOLUME 287 • NUMBER 42 mental Fig. S5) and had no effect on HIF-1 activity due to FLAG-HIF1␣ overexpression in normal and hypoxic conditions (supplemental Fig. S5). In the Hek293 sh-HIF1␣ cell line, p300 overexpression did not significantly increase the HIF-1 reporter activity in normal and hypoxic conditions (Fig. 6A). Consistent with the Western blot results in Fig. 4, the K709A mutant had significantly higher activities than the HIF1␣-WT in both normoxia (*) and hypoxia (**), respectively (Fig. 6A). When p300 was co-overexpressed, it significantly increased the HIF1␣-WT activities in both normoxia and hypoxia as reported. In contrast, p300 did not significantly change the activity of K709A mutant (Fig. 6A). We also compared the HIF1␣-WT and K709A mutant activity under the condition of p300 siRNA. In both normoxia (Fig. 6B, *) and hypoxia (Fig. 6B, **), we observed that HIF1␣-WT activity was significantly reduced by p300 knockdown; in contrast, HIF1␣-K709A mutant activity was not significantly affected (Fig. 6B). This was consistent with the Western blot data in Fig. 5B, in which the protein expression of HIF1␣-WT, not K709A mutant, was reduced by p300 knockdown.

HIF1␣ Lys-709 Acetylation Increases Protein Stability
We also compared the HIF-1 target gene expression due to HIF1␣-WT or the K709A mutant. The Hek293-shHIF1␣ cell line was transfected with the wild-type or mutant plasmid, and cultured in normal or hypoxic conditions overnight. Similar to Fig. 4A, the Lys-709 mutant protein was more stable than the wild-type (supplemental Fig. S6). The HIF-1 target gene expressions were significantly up-regulated by HIF1␣-WT in hypoxia. For a subset of these genes such as VEGFa, Glut 1, and PDK1, the fold of hypoxia up-regulation was significantly higher by an equal amount of HIF1␣-K709A plasmid compared with the wild-type (Fig. 6C and supplemental Fig. S6). In contrast, HIF-1 target genes such as hexokinase 2 (HK2) and JMJD1A were not further up-regulated by the K709A mutant ( Fig. 6C and supplemental Fig. S6). Taken together, these results suggest that Lys-709 is critical for p300 to exert its positive effect on HIF1␣ protein stability and HIF-1 activity. The elimination of this lysine residue provides a gain of function effect to HIF-1 to phenotypically mimic p300. Finally, to investigate the effect of HIF1␣-WT and K709A mutant on cancer cell growth, we used shRNA to knockdown the endogenous HIF1␣ in the liver cancer cell line Hep3B and subsequently overexpressed either HIF1␣-WT or K709A mutant. Proliferation studies showed that cells expressing the WT or mutant protein had similar growth patterns in normal oxygen condition (Fig. 6D). Importantly, cells with the K709A mutant grew significantly better than the wild-type in hypoxia (Fig. 6D).

DISCUSSION
In hypoxic conditions, the increased HIF1␣-p300 interaction can enhance HIF-1 transcriptional activity. In this study, we identified a novel biological consequence of HIF1␣-p300 interaction. HIF1␣ is acetylated at lysine 709 by p300. The consequence of this acetylation is to allow the hypoxic cells to have a more sustainable level of HIF1␣ protein, which can lead to a more sustainable HIF-1 transactivation and HIF-1 target gene transcription. Our data indicate that p300 can acetylate HIF1␣ Lys-709 in both normal and hypoxic conditions. However, the acetylation process requires HIF1␣-p300 interaction, which is significantly enhanced by hypoxia (26). Therefore, the physiologically relevant Lys-709 acetylation and the biological effect should primarily occur in hypoxia. As suggested in Fig. 6E, in hypoxia, the dehydroxylation at asparagine 803 leads to the increase of HIF1␣-p300 interaction, which shifts Lys-709 to an acetylated state and contributes to the HIF1␣ protein stability and HIF-1 transactivation. HIF1␣ is known to interact directly and/or indirectly with lysine deacetylases (class I-III HDACs) (10,13,14,18,27), including HDAC1, which specifically counters the Lys-709 acetylation and down-regulates HIF1␣. Furthermore, it has been shown that HDAC1 interacts with HIF1␣ via VHL (27); therefore, it is likely that in normoxia, one or multiple HDACs keep the HIF1␣ (including Lys-709 and Lys-674) in a deacetylated state. This contributes to the inhibition of HIF-1 activity via increasing protein degradation and/or decreasing HIF1␣-p300 interaction. Afterward, cells were treated with proteasomal inhibitor MG132 at normoxia or hypoxia for 4 h, and the WCL was harvested for anti-FLAG IP and Western blots using anti-ubiquitin, antiacetylated lysine (Ace-K), anti-p300, or anti-FLAG antibodies. B, U2OS cells were treated with p300 siRNA or control siRNA. 24 h later, cells were cotransfected with FLAG-HIF1␣ WT or K709A mutant and GFP. 24 h after the transfection, cells were exposed to normoxia or hypoxia for 6 h, WCL was harvested for Western blot analysis of FLAG-HIF1␣ WT, mutant, and GFP. The FLAG-HIF1␣ protein level (R) was calculated based on the ratio of band intensity of FLAG-HIF1␣/GFP. All data were representative of three independent experiments.
We have shown that the Lys-709 acetylation is not the result of hypoxia-induced HIF1␣ protein stability, but rather, it is one of the causes for HIF1␣ to be stable and/or more sustainable in hypoxic condition. Although the oxygen/VHL pathway provides a major regulatory mechanism for HIF1␣ protein stability, additional mechanisms do exist to fine-tune the HIF1␣ protein stability in both normoxia and hypoxia. For example, despite being significantly stabilized by hypoxia, HIF1␣ still undergoes polyubiquitination (via HAF and CHIP/HSP70) and 26 S-dependent proteasomal degradation (8,32). In the current study, the post-translational modification at Lys-709 provides a novel mechanism in regulating HIF1␣ polyubiquitination and protein stability in hypoxia. Our data suggest that the acetylation at Lys-709 by p300 reduces HIF1␣ polyubiquitination, which serves as a possible mechanism of enhancing the protein stability in hypoxia. Recently, HIF1␣ Lys-709 has been identified as a possible site of polyubiquitination in a proteomic study (29). In this study, the polyubiquitination of HIF1␣-WT was significantly reduced by p300 in normal and hypoxic conditions. In contrast, the polyubiquitination of HIF1␣ K709A mutant was not dramatically affected by p300. We speculate that Lys-709 is an oxygen/VHL-independent polyubiquitination site. In normoxia, the VHL-dependent mechanism primarily determines the HIF1␣ protein stability. In hypoxia or VHL loss-of-function conditions, there is a possible competition between the polyubiquitination (degradation) and acetylation by p300 (stability) in regulating HIF1␣ protein. A more detailed study is required to elucidate the polyubiquitination machinery at Lys-709, and the cross-talk or competition between polyubiquitination and acetylation.
Although the interaction between HIF1␣-p300 is functionally critical for HIF-1 transactivation, it is unknown whether all HIF-1 target gene transactivation depends on the presence of p300 to the same level. Because p300 can also serve as a chro-FIGURE 6. Lys-709 regulates HIF-1 transcriptional activity and cancer cell proliferation in hypoxia. A, Hek293 cells with stable shRNA knockdown of endogenous HIF1␣ (Hek293-shHIF1␣) were transfected with Ev, p300, FLAG-HIF1␣ WT, FLAG-HIF1␣ K709A mutant (mut) as single plasmid or as HIF1␣ and p300 combined. 24 h later, cells were transfected with hypoxia response element-driven firefly (p2.1) and constitutive Renilla luciferase vectors and cultured in normoxia or hypoxia overnight before Dual-Luciferase measurement. *, p Ͻ 0.01; **, p Ͻ 0.05. B, Hek293-shHIF1␣ cells were treated with p300 siRNA or control siRNA. 24 h later, cells were transfected with Ev, FLAG-HIF1␣ WT, or K709A mutant. After culturing overnight, cells were further transfected with HRE firefly and constitutive Renilla luciferase vectors and cultured in normoxia or hypoxia overnight before Dual-Luciferase measurement. ⌬, the difference between si-C and si-p300. *, p Ͻ 0.05; **, p Ͻ 0.01. C, Hek293-shHIF1␣ cells were transfected with equal amount of Ev, FLAG-HIF1␣ WT, or K709A mutant plasmid. 48 h after transfection, cells were cultured in normoxia (N) or hypoxia (H) overnight, and the HIF-1 target gene expressions were measured by real-time quantitative RT-PCR. *, p Ͻ 0.01. D, Hep3B-shHIF1␣ cells were transfected with FLAG-HIF1␣ WT or K709A mutant. 48 h later, cells were equally divided into 24-well plates and cultured in normoxia (N) or hypoxia (H). The cell proliferation over a 48-h period was measured. *, p Ͻ 0.01. E, summary of how HIF1␣ protein stability can be regulated by p300 via Lys-709. ub, ubiquitination. The bar and line graphs in A-D represent mean and S.D. of four experiments; all comparisons were made by t test. Ace, acetylated. matin histone acetylase (23,33), it is possible that the HIF1␣-p300 interaction and acetylation occurs at the chromatin level and is only relevant to HIF-1 target genes that require p300. This is a likely interpretation that only a subset of HIF-1 target gene expressions were further increased by p300 or HIF1␣ Lys-709 mutant overexpression. Therefore, the functional significance of HIF1␣-p300 interaction is both gene-and cell type-dependent.
Multiple sites of the HIF1␣ protein can be modified by lysine acetylation with different biological consequences. Acetylation within the oxygen-dependent degradation domain is related to the VHL-dependent HIF1␣ degradation (11), whereas acetylation within the N-terminal region is related to VHL-independent degradation (18). The acetylation at Lys-674 is related to HIF1␣-p300 interaction and HIF-1 transactivation (14). In the current study, the acetylation at Lys-709 is related to VHLindependent HIF1␣ stability. Furthermore, the acetylation/ deacetylation enzymes regulating these specific modifications can also be different. Our site-directed mutagenesis study indicates that p300 primarily acetylates Lys-709 within the HIF1␣. In contrast, a p300 associated protein, PCAF, has no effect on Lys-709 acetylation and is primarily responsible for Lys-674 acetylation (14).
Therapeutically, HIF1␣ protein stability and HIF-1 transcriptional activity can be targeted by deacetylase inhibitors targeting class I, II, or III HDACs. Currently, most of the inhibitors are not specific to an isozyme; therefore, the current study suggest less than optimal inhibitory results using these deacetylase inhibitors. For example, the HIF1␣ protein can be stabilized by deacetylation of N-terminal lysine residues via HDAC4 (18). In the current study, HIF1␣ protein can be destabilized by deacetylation of Lys-709 via HDAC1. This suggests that a hydroxamic-based HDAC inhibitor such as TSA or SAHA can cause both stabilization and destabilization activities toward HIF1␣. The protein level after the treatment is just a net result of these two conflicting mechanisms. Interestingly, this may also explain the observation that inhibiting class I HDACs (HDAC1-3) does not inhibit HIF1␣ proteins (10). Therefore, the current study provides mechanistic rationale of discovering and developing the HDAC isozyme-specific inhibitors, some of which can potentially be used to either increase or decrease HIF1␣ and HIF-1 activities.