|Home | About | Journals | Submit | Contact Us | Français|
Hypoxia-inducible factors (HIFs) are critical transcription factors that mediate cell survival during reduced oxygen conditions (hypoxia). At regular oxygen conditions (normoxia), HIF-1α and HIF-2α are continuously synthesized in cells and degraded via the ubiquitin–proteasome pathway. During hypoxia, these proteins are stabilized and translocate to the nucleus to activate transcription of target genes that enable cell survival at reduced oxygen levels. HIF proteins are tightly regulated via post-translational modifications including phosphorylation, acetylation, prolyl-hydroxylation and ubiquitination. Here we show for the first time that exogenous and endogenous HIF-2α are also regulated via the ubiquitin-like modifier small ubiquitin-like modifiers (SUMO). Using mutational analysis, we found that K394, which is situated in the sumoylation consensus site LKEE, is the major SUMO acceptor site in HIF-2α. Functionally, sumoylation reduced the transcriptional activity of HIF-2α. Similar to HIF-1α, HIF-2α is regulated by the SUMO protease SENP1. The proteasome inhibitor MG132 strongly stabilized SUMO-2-conjugated HIF-2α during hypoxia but did not affect the total level of HIF-2α. The ubiquitin E3 ligases von Hippel–Lindau and RNF4 control the levels of sumoylated HIF-2α, indicating that sumoylated HIF-2α is degraded via SUMO-targeted ubiquitin ligases.
Many different post-translational modifications including phosphorylation, acetylation, methylation and glycosylation are involved in regulating the activity of proteins. Sumoylation, the process of attaching a small ubiquitin-like modifier (SUMO) protein to a target protein, is a more recently discovered reversible post-translational modification (1,2). SUMOs are covalently conjugated to acceptor lysines in sumoylation consensus motifs (ψKx(E/D), where ψ stands for V, L, I, M or F and x can be any amino acid) in target proteins. Sumoylation regulates many different cellular processes such as gene expression, signal transduction, chromatin structure and the maintenance of the genome by attachment to and modulation of target proteins.
Three SUMO family members have been described in vertebrates, SUMO-1, -2 and -3, encoded by three different genes (3). The mature forms of SUMO-2 and SUMO-3 are very similar (~95% identical) but differ from SUMO-1 (~50% identical) (4). SUMO-1 and SUMO-2 each have a unique subset of substrates as well as a shared subset to which both isoforms can be conjugated (5).
Low oxygen concentration leads to adaptive changes in the transcription of a range of genes. The Hypoxia-Inducible Factors (HIFs) mediate the transcriptional activation of genes that allow cells and tissues to cope with low oxygen conditions (6,7). HIFs are heterodimers, composed of one α and one β subunit. While the HIF-1β/ARNT (aryl hydrocarbon receptor nuclear translocator) subunit is stable, the HIF-α proteins (HIF-1α, HIF-2α and HIF-3α) are continuously synthesized and degraded under normoxic conditions. During normoxia, two conserved proline residues in the oxygen-dependent degradation domain (ODD) of HIF-α are hydroxylated by HIF-specific prolyl hydroxylase-domain proteins (PHD 1, 2 and 3). Hydroxylation of the prolyl residues mediates binding of the von Hippel–Lindau (VHL) ubiquitin ligase complex responsible for ubiquitination and degradation of HIF-1α in a proteasome-dependent manner (8,9). The PHD proteins require molecular oxygen for their enzymatic activity and no longer function during hypoxia. Consequently, degradation no longer occurs during hypoxia and the HIF-α subunits rapidly accumulate and translocate to the nucleus (10–12).
HIF-1α and HIF-2α are 48% identical and their stability and transcriptional activity are regulated via shared mechanisms (13,14). Both proteins play important and similar, but non-redundant roles in fetal development and tumor angiogenesis. Mouse embryos in which HIF-1α expression was disrupted exhibited multiple defects in cardiovascular development and died early during development (E11.5) (15). HIF-2α–/– mice generated by different groups differed somewhat in phenotype, possibly due to differences in genetic backgrounds. Phenotypes observed include defective vascular remodeling with local hemorrhage (16), defective fetal catecholamine production (17) or altered lung maturation secondary to impaired surfactant secretion by alveolar type 2 cells (18).
The activity of HIF-2α is tightly controlled by post-translational modifications including prolyl-hydroxylation, ubiquitination and phosphorylation. We have investigated the sumoylation of HIF-2α, which contains two consensus sumoylation sites, LK394EE and LK497IE. We have generated HIF-2α mutants in which these consensus sumoylation sites were disrupted and demonstrate that K394 is used for SUMO conjugation and plays a role in the regulation of HIF-2α activity. Interestingly, SUMO-2-conjugated HIF-2α is rapidly degraded during hypoxia via SUMO-targeted ubiquitin ligases.
Plasmids containing the human wild-type HIF-1α and -2α cDNA tagged at the N-terminus with three consecutive FLAG tags were a kind gift from Dr A. Groot (Utrecht University, Utrecht, The Netherlands) (19,20). The 5xHREpGL3-Luciferase reporter was a kind gift from Dr M. Duyndam (VUMC, Amsterdam, The Netherlands). The FLAG-SUMO-2 lentiviral construct was generated by inserting his6-SUMO-2 into the EcoRV site of plasmid pLV-CMV-IRES-eGFP (21). The FLAG epitope was subsequently introduced in this plasmid via oligonucleotide insertion into the PstI site. Plasmids encoding shRNAs were obtained from the MISSION shRNA Library (Sigma Aldrich): control shRNA, SHC-002; RNF4 shRNA, TRCN0000017053; RSUME shRNA, TRCN0000004095; SENP1 shRNA, TRCN0000004399 and VHL shRNA, TRCN0000010460.
The K394R, K497R, E396A and E499A mutations in HIF-2α were generated by site-directed mutagenesis using the Quickchange II kit according to the instructions of the manufacturer (Stratagene). The following oligonucleotides were used to generate these mutations: K394R, forward 5′-ctattcaccaagctaagggaggagcccgag-3′ and reverse 5′-ctcgggctcctcccttagcttggtgaatag-3′; K497R, forward 5′-gataacgacctgaggattgaagtgattg-3′ and reverse 5′-caatcacttcaatcctcaggtcgttatc-3′; E396A, forward 5′-caagctaaaggaggcgcccgaggagctg-3′ and reverse 5′-cagctcctcgggcgcctcctttagcttg-3′; E499A, forward 5′-gacctgaagattgcagtgattgagaag-3′ and reverse 5′-cttctcaatcactgcaatcttcaggtc-3′. All mutants were sequence verified.
HeLa cells stably expressing his6-SUMO-1 or his6-SUMO-2 were previously described (5). The HeLa cell line stably expressing FLAG-SUMO-2 was generated by infection with the lentivirus encoding FLAG-SUMO-2. HeLa cells were grown in Dulbecco’s modified Eagle medium (DMEM), supplemented with 10% FCS and 100 U/ml penicillin and streptomycin (Invitrogen). Hypoxic stabilization of HIF-α was performed at 1% O2 for 24 h. Transient transfections were performed using 2.5 µl Polyethylenimine (PEI, 1 mg/ml, Alpha Aesar) per µg DNA. Trichostatin A (TSA, Sigma Aldrich) was added to the culture medium at 300 nM for 24 h where indicated. MG132 (Sigma Aldrich) was added to the culture medium at 10 µM for 4 or 7 h where indicated. For both drugs, control cells were treated with dimethyl sulphoxide (DMSO). Cycloheximide was added to the culture medium at 10 µg/ml where indicated. Lentiviral infections were performed at an MOI of 1 for 24 h. Cells were subsequently incubated for another 48 h prior to lysis.
Cells were cultured in 24-well plates and cotransfected with 0.2 µg luciferase reporter plasmid and 0.5 µg expression plasmid as indicated. Experiments were carried out in triplicate and additional wells were prepared for control immunoblotting experiments. Cells were lysed in Reporter Lysis Buffer (Promega) for luciferase activity measurements or in NuPage LDS protein sample buffer (Invitrogen) for immunoblotting.
HeLa cells were cultured on glass cover slips in 24-well plates and transfected with 0.5 µg expression plasmid. The cells were fixed for 20 min in 3.7% paraformaldehyde at room temperature and permeabilized in phosphate-buffered saline (PBS) containing 0.5% Triton X-100 for 10 min. Cells were blocked with 10% goat serum for 30 min. The cells were incubated with primary antibody (mouse anti FLAG, 1:1000 in PBS/T, 10% NGS) for 1 h, washed and incubated with secondary antibody (goat anti mouse-Alexa 488, 1:1000 in PBS/T, 10% NGS) for 1 h. The cells were washed 5× with PBS/T, mounted in Vectashield, sealed with nail varnish and stored at −20°C.
Microscopy experiments were carried out using a confocal microscope system (model TCS/SP2, Leica). Images were acquired with a 100× NA 1.4 plan Apo objective and were analysed with Leica confocal software.
Peptide antibody AV-SM23-0100 (Eurogentec) against SUMO-2/3 was described previously (22). Monoclonal antibody T5326 against γ-tubulin, polyclonal antibody HPA011765 against SENP1 and monoclonal antibody M2 against FLAG were obtained from Sigma Aldrich. Polyclonal antibody NB100-122 against HIF-2α was obtained from Novus Biologicals. Polyclonal antibody against RNF4 was a kind gift of Dr J. Palvimo. Secondary antibodies used were anti-rabbit HRP and anti-mouse HRP (Pierce Chemical Co.). The secondary antibody used for immunofluorescence was goat anti mouse Alexa 488 (Invitrogen).
Cells were cultured in 14 cm diameter culture dishes and transfected with 10 µg FLAG-HIF-1α or -2α expression plasmids as indicated. His6-SUMO conjugates were purified essentially as previously described (23). Cells were scraped in ice-cold PBS. Two small aliquots of each sample were lysed in LDS protein sample buffer (Invitrogen) as input control, or in 8 M Urea, 100 mM Na2HPO4/NaH2PO4, 10 mM Tris–HCl, pH 7.0 to determine the protein concentration. The remaining cells were solubilized in lysis buffer (6 M Guanidinium-HCl, 100 mM Na2HPO4/NaH2PO4, 10 mM Tris–HCl, pH 8.0, 20 mM Imidazole, 10 mM β-mercaptoethanol) and sonicated to reduce the viscosity. His6-SUMO conjugates were enriched on Ni-NTA Agarose beads (Qiagen) and washed using wash buffers A–D. (Buffer A: 6 M Guanidinium–HCl, 100 mM NaH2PO4/Na2HPO4, 10 mM Tris–HCl pH 8.0 and 0.2% Triton-X-100. Buffer B: 8 M Urea, 100 mM NaH2PO4/Na2HPO4, 10 mM Tris–HCl pH 8.0 and 0.2% Triton-X-100. Buffer C: 8 M Urea, 100 mM NaH2PO4/Na2HPO4, 10 mM Tris–HCl pH 6.3 and 0.2% Triton-X-100. Buffer D: 8 M Urea, 100 mM NaH2PO4/Na2HPO4, 10 mM Tris–HCl pH 6.3 and 0.1% Triton-X-100). These wash buffers also contained 10 mM β-mercaptoethanol. Samples were eluted in 6.4 M Urea, 80 mM NaH2PO4/Na2HPO4, 8 mM Tris–HCl pH 7.0, 200 mM imidazole.
Cells were grown in 14 cm diameter dishes and cultured for 7 h at 1% O2 before lysis. Cell extracts were prepared in SDS buffer [2% SDS, 50 mM Tris–HCl, pH 7.5 and 10 mM iodoacetamide (23)] and incubated at room temperature for 15 min. The viscosity of the samples was reduced by sonication. SDS concentration of the samples was reduced by dilution in buffer B (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5% NP40 and 0.5 mM β-mercaptoethanol). Samples were centrifuged at 9000g for 30 min at 4°C and incubated for 3 h at 4°C with Anti-FLAG M2 beads (M8823, Sigma Aldrich). The beads were collected, washed 5 times with wash buffer (50 mM Tris–HCl, pH 7.5, 150 mM NaCl and 0.5% NP40) and eluted with 15 µl wash buffer containing 100 µg/ml FLAG peptide for 15 min at 4°C. Supernatant was collected for immunoblotting.
Protein samples were size fractionated on Novex 4–12% Bis-TRIS gradient gels using 4-morpholineprop-anesulfonic acid buffer (Invitrogen) or on regular SDS-PAGE gels with a tris-glycine buffer. Note that these different methods cause slight differences in the running behavior of proteins (Invitrogen). Size fractionated proteins were subsequently transferred onto Hybond-C extra membranes (Amersham Biosciences) using a submarine system (Invitrogen). The membranes were incubated with specific antibodies as indicated. Bound antibodies were detected via chemiluminescence with ECL Plus (Amersham Biosciences).
The transcription factor HIF-1α is regulated by SUMO conjugation. As a homologue of HIF-1α, HIF-2α shows close similarity in sequence, function and regulation. HIF-1α contains three consensus sumoylation sites, two of which are involved in the regulation of its transcriptional activity (6,24–26). Two potential SUMO acceptor sites matching the consensus sequence are present in HIF-2α (Figure 1). These sites are conserved in a wide range of species including Mus musculus, Rattus norvegicus, Bos taurus, Gallus gallus and Danio rerio.
To compare SUMO conjugation between HIF-1α and -2α in cells, we made use of published stable cell lines expressing his6-SUMO-1 and his6-SUMO-2 (5). HeLa, HeLahis6-SUMO-1 and HeLahis6-SUMO-2 cells were transfected with plasmids encoding FLAG-HIF-1α, -2α or an empty vector. The cells were placed at 1% O2 24 h post-transfection and cultured for another 24 h before lysis and purification of his6-SUMO conjugates. Expression levels of HIF-1α and -2α were checked in total cell lysate samples. SUMO-modified forms of FLAG-HIF-1α and -2α could be detected in both the his6-SUMO-1 and his6-SUMO-2 purified samples (Figure 1C). Note that unmodified, but not SUMO-modified, FLAG-HIF could be detected in the HeLa negative control. These results indicate that analogous to HIF-1α, HIF-2α is a substrate for sumoylation. HIF-2α was found to be conjugated to both SUMO proteins at least as efficiently as HIF-1α.
To investigate whether the consensus sumoylation sites present in HIF-2α are used for SUMO conjugation in cells, we created plasmids encoding mutant forms of FLAG-HIF-2α lacking one or both SUMO acceptor lysines. HeLahis6-SUMO-2 cells were transfected with plasmids encoding FLAG-HIF-2α wild-type and mutants and cultured at 1% O2 for 24 hours. His6-SUMO-2 conjugates were purified from these cells and analysed by immunoblotting. At least 2 SUMO-modified forms of FLAG-HIF-2α wild-type could be detected (Figure 2, lane 3) that were absent in the control samples (lanes 1 and 2). Disrupting the first consensus site by mutating K394 (lane 4) strongly reduced sumoylation of FLAG-HIF-2α compared to wild-type. In contrast, disrupting the second consensus site had no effect on sumoylation levels (lane 5). A mutant in which both sites were disrupted did not exhibit a further reduction in sumoylation compared to the K394R single mutant (lane 6 compared to 4). These results indicate that lysine 394 is the major SUMO acceptor site of HIF-2α.
To rule out the possibility that our results could be attributed to potential modifications of these lysine residues other than sumoylation, a second set of mutants was prepared. In these mutants, the consensus sumoylation sequence was disrupted by mutating the glutamic acid residues, while leaving the lysines intact. This second set of mutants was used to repeat this experiment and confirmed our findings with the lysine to arginine mutants (Supplementary Figure S1). Mutating E396 resulted in a similar reduction of sumoylation compared to the K394R mutant. Disrupting the second consensus site by mutating E499 had no effect on sumoylation.
Since sumoylation of transcription factors often leads to a reduction in their activity (27), we compared the transcriptional activity of wild-type and mutant FLAG-HIF-2α using an HRE-luciferase reporter plasmid (Figure 3). Mutating either K394 or E396 increased the transcriptional activity of HIF-2α while the K497R and E499A mutants showed a level of activity similar to that of the wild-type (Figure 3A). Neither double mutant was transcriptionally more active than HIF-2α proteins carrying only the single K394R or E396A mutation. Increased transcriptional activity of the sumoylation-impaired HIF-2α mutants was not due to an increase in expression of these mutant proteins (Figure 3B).
Sumoylation has previously been shown to alter the subcellular localization of a subset of target proteins including RanGAP1 (28,29). To investigate whether sumoylation has an effect on the subcellular localization of HIF-2α, cells were transfected with plasmids encoding wild-type or mutant FLAG-HIF-2α and cultured for 24 h at 1% O2 before fixation. The localization of the FLAG-HIF-2α proteins was determined using immunofluorescence (Figure 4). FLAG-HIF-2α wild-type was found to be predominantly localized to the nucleoplasm. Disrupting the major sumoylation consensus site by mutating K394 or E396 did not influence the localization of FLAG-HIF-2α.
We have previously reported that proteasome inhibition strongly increases the total pool of SUMO-2/3 conjugates (30). Moreover, we have identified a subset of SUMO-2 conjugates that are stabilized by proteasome-inhibition and verified that the ubiquitin–proteasome system specifically degraded the SUMO-2-conjugated forms of hnRNP M, MCM-7 and PIAS-1 without affecting the total pools of these proteins. To investigate whether sumoylation regulates HIF-2α stability, we purified his6-SUMO-2 conjugates from HeLa cells that were transfected with plasmids encoding FLAG-HIF-2α wild-type or mutant proteins. The cells were cultured at 1% O2 in the presence or absence of the proteasome inhibitor MG132 for 4 h before being lysed. His6-SUMO-2 conjugates were analysed by immunoblotting. As expected, proteasome inhibition strongly stabilized SUMO-2 conjugates (Figure 5A). SUMO-modified FLAG-HIF-2α exhibited a very strong accumulation upon proteasome inhibition by MG132, whereas the stability of unmodified HIF-2α was not significantly affected. Accumulation of SUMO-modified HIF-2α could be strongly reduced by disrupting the LK394EE sumoylation site but not by disruption of the LK497IE site. The E to A mutants were used for this experiment to interfere as little as possible with potential direct ubiquitination of lysines.
In order to further investigate whether sumoylation has an effect on the stability of the total pool of HIF-2α, we compared the stability of FLAG-HIF-2α wild-type versus the E396A mutant employing cycloheximide to block protein synthesis. HeLa cells were transfected with plasmids encoding wild-type FLAG-HIF-2α or the E396A mutant. The cells were kept at 21% oxygen and 24 h post-transfection cells were cultured for 3 or 5 h in the presence or absence of cycloheximide. Total cell lysates were prepared and analysed by immunoblotting to compare total FLAG-HIF-2α levels. Cells treated with cycloheximide showed greatly reduced levels of FLAG-HIF-2α compared to control samples (Figure 5B). No differences in stability were observed for the total pools of wild-type and E396A mutant FLAG-HIF-2α.
Subsequently, we investigated the conjugation of endogenous HIF-2α to SUMOs by employing a HeLa cell line stably expressing FLAG-SUMO-2. HeLaFLAG-SUMO-2 and control HeLa cells were cultured in the presence or absence of MG132 at 1 or 21% O2 for 7 h before lysis and purification of FLAG-SUMO-2 conjugates. Several SUMO-modified forms of endogenous HIF-2α could specifically be detected in FLAG-SUMO-2 purified samples from hypoxic or MG132-treated HeLaFLAG-SUMO-2 cells (Figure 6). A substantial increase in the amount of SUMO-modified HIF-2α could be observed after treating the cells with MG132. We conclude that endogenous HIF-2α is a target for SUMO-2 during hypoxia and normoxia.
Proteasome inhibition strongly stabilized SUMO-2-conjugated HIF-2α without affecting the total levels of HIF-2α (Figures 5 and and6),6), indicating that HIF-2α might be controlled by SUMO-targeted ubiquitin ligases. RNF4 has previously been shown to act as a SUMO-targeted ubiquitin ligase (31), therefore we have performed RNF4 knockdown experiments (Figure 7). RNF4 depletion resulted in a significant accumulation of SUMO-modified HIF-2α, consistent with a role for RNF4 as a SUMO-targeted ubiquitin ligase for HIF-2α.
Previously, the sumoylation of HIF-1α was shown to be regulated via the SUMO protease SENP1 (25) and via the SUMO E3 ligase RSUME (32). Cheng et al. (25) also proposed a role for VHL as a SUMO-targeted ubiquitin ligase in SUMO-dependent HIF-1α degradation during hypoxia. We used shRNA interference to reduce the expression levels of endogenous SENP1, RSUME, and VHL (Figure 7). Downregulating the expression of VHL resulted in a significant accumulation of SUMO-modified HIF-2α, consistent with a role for VHL as a second SUMO-targeted ubiquitin ligase for HIF-2α. Furthermore, we observed an increase in the levels of sumoylated HIF-2α upon SENP1 knockdown, consistent with a role for this SUMO protease in HIF-2α desumoylation. SENP1 knockdown did not affect the non-sumoylated pool of HIF-2α but did decrease the total pool of HIF-1α, consistent with the results published by Cheng et al. (25) (Figure 7 and Supplementary Figure S2). The shRNA against RSUME did not lead to a change in the level of SUMO-modified HIF-2α. However, we cannot rule out a role for RSUME in HIF-2α regulation as we were unable to verify the efficiency of the downregulation of this protein in our experiment.
We have shown here that HIF-2α is a target protein for SUMOs; both endogenous and exogenous HIF-2α were found to be sumoylated in cells. HIF-2α contains two consensus sumoylation sites, LK394EE and LK497IE, and mutational analysis showed that K394 is the major SUMO acceptor site in HIF-2α. The FLAG-HIF-2α K394R and E396A mutants showed a clear reduction in sumoylation; however, sumoylation was not totally absent, indicating the presence of at least one non-consensus site for SUMO modification. Functionally, sumoylation inhibited the transcriptional activity of HIF-2α.
Recently, we have shown that SUMO-2 conjugation affects the stability of a subset of conjugates (30). HIFs are primarily controlled via stability; they are degraded during normoxia and stabilized during hypoxia (33). Interestingly, SUMO-2 conjugation enabled the degradation of HIF-2α during hypoxia. Since the total pool of HIF-2α was not affected, SUMO-2 conjugation appeared to be a requirement for HIF-2α degradation during hypoxia. This indicated that HIF-2α can be controlled by SUMO-targeted ubiquitin ligases (25,31,34). RNF4 is an important mammalian SUMO-targeted ubiquitin ligase that controls the stability of sumoylated PML, PML-RARα and PEA3 (35–37). Consistently, SUMO-2-conjugated HIF-2α accumulated upon RNF4 knockdown. Similarly, VHL is a key regulator of HIF stability during normoxia and part of an ubiquitin E3 ligase complex (8,9). In addition, Cheng et al. (25) have previously reported the involvement of VHL in the degradation of sumoylated HIF-1α under hypoxic conditions. Our results suggest that VHL also controls the stability of sumoylated HIF-2α during hypoxia since VHL knockdown strongly increased the amount of SUMO-2-modified HIF-2α.
Different studies have reported conflicting roles for HIF-1α sumoylation (6,24–26,32). Carbia-Nagashima et al. (32) and Bae et al. (24) reported that sumoylation stabilizes and activates HIF-1α whereas Cheng et al. (25) and Berta et al. (6) reported that sumoylation inhibits HIF-1α. Cheng et al. (25) showed that in mice deficient for SENP1, the stabilization of HIF-1α during hypoxia was reduced, resulting in embryonic lethality due to severe fetal anemia. We have addressed whether SENP1 and RSUME also regulate HIF-2α activity. Knockdown of SENP1 increased the amount of HIF-2α-SUMO-2 conjugates, whereas knockdown of RSUME did not influence HIF-2α sumoylation.
Conjugated SUMOs can also act as a binding site for repressor proteins such as histone deacetylases (HDACs) (38,39). It appears that HDACs control HIF-2α since the HDAC-inhibitor TSA increased the activity of HIF-2α (Supplementary Figure S3). However, this was not dependent on HIF-2α sumoylation since the E396A mutant was similarly affected by TSA.
In contrast to the K394R and E396A mutations, the K497R and E499A mutations did not appear to reduce sumoylation of HIF-2α. Nevertheless, both sumoylation consensus sites are well conserved from zebrafish to human. Three consensus sumoylation sites are present in HIF-1α and two of these sites are situated in sequences which show a high degree of homology to sequences surrounding the consensus sites found in HIF-2α (Figure 1). Interestingly, the K391 sumoylation site in HIF-1α and the K394 sumoylation site in HIF-2α show a high degree of sequence similarity and both are used for SUMO conjugation. HIF-1α K532 and HIF-2α K497 also lie in highly similar sequence stretches and are both not used for SUMO conjugation. The presence of a second sumoylation site that is well conserved but appears not to be used for SUMO conjugation is puzzling. The site could potentially be inaccessible for the SUMO conjugation machinery. Alternatively, specific E3 factors might be required that mediate the conjugation of SUMOs to K497 but are missing in our assay. Furthermore, it is possible that sumoylation of K497 is restricted to specific stages of embryonic development or to specific cell types. The third HIF-1α consensus site, K477, has been shown to be used for SUMO conjugation, but no equivalent site is present in HIF-2α (6,25).
Rapidly growing cancer cells induce a hypoxic environment. Survival of these cells in such an environment is enabled, at least in part, by the HIF system. The HIF pathway is activated in many solid tumors and is therefore a major drug target for anti-cancer therapy (40). Overexpression of HIF-2α has been shown in several different tumor types and signifies an advanced tumor stage and/or poor patient survival outcome. By inducing the transcription of target genes, it contributes to vascularization of these tumors and thus to tumor growth and malignant behavior (14). Our results indicate that sumoylation inhibits HIF-2α. Inhibiting HIF-2α activity is expected to limit the growth rate of solid tumors. Potentially, stimulating the sumoylation or inhibiting the desumoylation of HIF-2α in tumors could be a novel strategy to limit tumor progression.
Supplementary Data are available at NAR Online.
Funding for open access charge: The Netherlands Organisation for Scientific Research (NWO).
Conflict of interest statement. None declared.
The authors thank Drs Groot, Duyndam and Palvimo for generously providing reagents. They also thank M. Rabelink for assistance with lentivirus production and Dr Th. van Laar for insightful discussions.