Suppression of Hypoxia-inducible Factor 1α (HIF-1α) Transcriptional Activity by the HIF Prolyl Hydroxylase EGLN1*

The cellular response to hypoxia is, at least in part, mediated by the transcriptional regulation of hypoxia-responsive genes involved in balancing the intracellular ATP production and consumption. Recent evidence suggests that the transcription factor, HIF-1α, functions as a master regulator of oxygen homeostasis by controlling a broad range of cellular events in hypoxia. In normoxia, HIF-1α is targeted for destruction via prolyl hydroxylation, an oxygen-dependent modification that signals for recognition by the ubiquitin ligase complex containing the von Hippel-Lindau tumor suppressor. Three HIF prolyl hydroxylases (EGLN1, EGLN2, and EGLN3) have been identified in mammals, among which EGLN1 and EGLN3 are hypoxia-inducible at their mRNA levels in an HIF-1α-dependent manner. In this study, we demonstrated that apart from promoting HIF-1α proteolysis in normoxia, EGLN1 specifically represses HIF-1α transcriptional activity in hypoxia. Ectopic expression of EGLN1 inhibited HIF-1α transcriptional activity without altering its protein levels in a von Hippel-Lindau-deficient cell line, indicating a discrete activity of EGLN1 in transcriptional repression. Conversely, silencing of EGLN1 expression augmented HIF-1α transcriptional activity and its target gene expression in hypoxia. Thus, we proposed that the accumulated EGLN1 in hypoxia acts as a negative-feedback mechanism to modulate HIF-1α target gene expression. Our finding also provided new insight into the pharmacological manipulation of the HIF prolyl hydroxylase for ischemic diseases.

Interestingly, both EGLN1 and EGLN3 genes are up-regulated by hypoxia at their mRNA levels (31,38) in an HIF-1␣-dependent manner (39,40). These observations have led to the hypothesis that the buildup of the prolyl hydroxylases during hypoxia primes for rapid degradation of HIF-1␣ upon reoxygenation. In this study, however, we sought the role of EGLN1 in hypoxia. Our results indicated that EGLN1 binds HIF-1␣ in hypoxia as well as in normoxia and functionally inhibits HIF-1␣ N-terminal transcriptional activity.

EXPERIMENTAL PROCEDURES
Cell Culture-HEK293 and U-2 OS cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen), supplemented with 10% fetal bovine serum (Hyclone, Logan, UT). Hep3B cells were grown in minimum essential medium with 10% fetal bovine serum. A VHL-deficient renal clear cell carcinoma cell line, UMRC2 or C2, was obtained from L. Neckers (NCI, National Institutes of Health, Bethesda, MD). C2 cells were cultured in Dulbecco's modified Eagle's medium, supplemented with 10% fetal bovine serum. For hypoxic treatment, cells were incubated in a hypoxic chamber (NAPCO, Winchester, VA), maintained with 1% O 2 and 5% CO 2 , and balanced with N 2 .
Prolyl Hydroxylase Binding-Gal4-HIF-1␣(C-ODD) (amino acids 498 -603) or the corresponding L574S mutant (1.5 g) was transfected with or without the V5-tagged EGLN1 expression plasmid (0.2 g). Cells were exposed to hypoxic conditions or treated with 12.5 M Cbz-LLL (Sigma) for 4 h and then were lysed in a buffer containing 25 mM Tris (pH 7.5), 300 mM NaCl, and 1% Triton-X. The lysate was immunoprecipitated with anti-Gal4 monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) followed by immunoblotting with anti-EGLN1 * This research was supported by the Intramural Research Program of the NCI, National Institutes of Health, Center for Cancer Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  antibody (Novus Biologicals, Littleton, CO) and polyclonal anti-Gal4 antibody (Santa Cruz Biotechnology). RNA Interference-The design of the 21-nucleotide small interfering RNA (siRNA) duplexes against EGLN1, EGLN2, and EGLN3 is as described previously (40). All of them were chemically synthesized by Qiagen (Valencia, CA). SMARTpool siRNA against VHL was purchased from Dharmacon (Lafayette, CO). An siRNA targeting the firefly luciferase coding sequence of the pGL2 vector (41) was employed as a negative control. Cells were seeded at 30% confluence in antibiotic-free medium 24 h prior to transfection. Oligofectamine reagent (Invitrogen) was used for transfection with 20 nM siRNA duplex twice at a 24-h interval according to the manufacturer's instructions. The efficacy of the siRNA transfection in each experiment was determined by Western blot. Polyclonal antibodies against EGLN isoforms were purchased from Novus Biologicals (Littleton, CO), and anti-VHL monoclonal antibody was from Pharmingen.
HIF-1␣ Transcriptional Activity-Transient transfection of pGal4luc and pEpoE-luc reporter plasmids and luciferase assays were essentially as described previously (18,42). In brief, 0.25 g of reporter plasmid and 0.1 g of each effecter plasmid were used for each well in a 12-well plate with FuGENE 6 (Roche Applied Science). 0.1 g of pEYFP-Nuc expression vector (Clontech) was co-transfected for normalization of transfection efficiency. For reporter experiments involving siRNAs, cells were first transfected with the plasmid DNA at a confluency of ϳ30 -40%. The next day, cells were fed with Opti-MEM (Invitrogen) and transfected with siRNAs as described above. Luciferase and EYFP were measured as described previously (43). The data are presented as means plus standard deviations from three independent experiments in duplicate.
Western Blot Analysis-Western blot was performed essentially as described previously (18,43). The effect of EGLN1 on HIF-1␣ C-terminal oxygen-dependent degradation domain (C-ODD) protein level was examined after transfection of Hep3B and C2 cells, respectively, with 1 g of Gal4-HIF-1␣-(498 -603) and 0.2 g of EGLN1. pEYFP-Nuc (0.05 g) was also included as an internal control. Cells were treated in hypoxia for 4 h before harvest.
Quantitative Real-time RT-PCR-RNA was extracted with TRIzol (Invitrogen). Total RNA (500 ng) was reverse-transcribed with the Taq-Man reverse transcription reagents (Applied Biosystems, Branchburg, NJ), and 1 g of cDNA was then amplified using the TaqMan universal PCR master mix (Applied Biosystems). The PGK1 probe and primers were designed using ABI Primer Express version 2.0 software (Applied Biosystems): probe, 6FAM-5Ј-ATTTATCTAATTGTCCCATCTCT-CCACTGCTGCT-MGBNFQ-3Ј; forward primer, 5Ј-TCTTGAGGA-ACGGATCAGATGTC-3Ј; and reverse primer, 5Ј-AGTAGGCCCTT-GATAAAGAATGGA-3Ј. Human ACTB (Taqman primers and probe, ABI 7700) was used as an endogenous control (VIC). Gene-specific PCR products were measured continuously by an ABI PRISM 7700 sequence detection system (Applied Biosystems) during 40 cycles. The difference in threshold number of cycles between PGK1 and ACTB for each sample was then expressed relative to the normoxic mock-transfected samples and converted into -fold difference. All experiments were repeated three times in triplicates, and representative results were presented as means Ϯ S.E.

EGLN1 Binds HIF-1␣ Irrespective of Oxygen Tension-HIF-1␣
possesses an ODD (amino acids 401-603) (21) that can be divided into N-terminal and C-terminal fragments (43). Previously, we demonstrated that in HEK293 cells, EGLN1 co-precipitated with the C-ODD (amino acids 498 -603) after treatment with the proteasome inhibitor, Cbz-LLL (44) (Fig. 1A). In contrast, no co-precipitation could be detected when Leu-574, a molecular determinant of Pro-564 hydroxylation (44), was mutated (Fig. 1B). Moreover, no such interaction was shown when EGLN2 or EGLN3 was tested (44). Interestingly, in this study, we observed that EGLN1 also bound C-ODD during a 4-h  hypoxic treatment (Fig. 1C). A similar result was obtained when cells were treated with desferrioxamine, a hypoxia mimetic (Fig. 1D). To rule out the possibility that the detected EGLN1 binding was a remnant of hydroxylated HIF-1␣ from normoxia, we increased the hypoxic treatment to 8 h, given the turnover rate of HIF-1␣ in hypoxia being less than 2 h (21). Again, the binding of EGLN1 to C-ODD was readily detectable from cells with the prolonged hypoxic treatment (data not shown). Together, these findings indicated that EGLN1 interacts with HIF-1␣ irrespective of oxygen tension. EGLN1 is presumably inactive in hypoxia because of the strict requirement of oxygen for its enzymatic activity. This observation prompted us to search for additional functions of EGLN1.
EGLN1 Inhibits HIF-1␣ N-terminal Transcriptional Activity-In addition to the CAD, C-ODD also possesses transcriptional activity (19,43,45), referred to here as N-terminal transcriptional activity. To examine whether EGLN1 modulates HIF-1␣ transcriptional activities, we tested effects of transient expression of EGLN1 in Hep3B cells in a Gal4 reporter system (42,43). Results in Fig. 2A show that EGLN1 markedly inhibited the transcriptional activity of C-ODD but not of CAD. Even under hypoxic conditions, there was Ͼ50% decrease of the N-terminal transcriptional activity. To further test the specificity of such inhibition, we took advantage of mutations (P564A and L574S) that disrupt EGLN1 binding. As expected, each mutation gave rise to an increased reporter activity and a loss of hypoxic induction. Of note, the elevated transcriptional activity of each mutant was apparently correlated well with its increased protein levels (data not shown). However, EGLN1 had no significant effects on these mutants in hypoxia ( Fig. 2A). Similar results were obtained in U-2 OS cells (Fig. 2B). In addition, overexpression of EGLN2 and EGLN3 also inhibited the N-terminal transcriptional activity (data not shown).
EGLN1 Inhibition of HIF-1␣ Transcriptional Activity Is Independent of the VHL Degradation Pathway-As mentioned above, HIF-1␣ hydroxylation leads to VHL-mediated polyubiquitination and proteasomal degradation. It is possible therefore that the EGLN1 repression of HIF-1␣ transcriptional activity in hypoxia is a result of increased HIF-1␣ proteolysis despite the scarce hydroxylase activity. To test this possibility, we used the renal cell carcinoma cell line C2 (46) that lacks a functional VHL protein for HIF-1␣ proteolysis. However, the N-termi-nal transcriptional activity was still inhibited significantly by EGLN1 irrespective of oxygen tension (Fig. 3A). Of note, EGLN1 bound C-ODD in both normoxic and hypoxic C2 cells (Fig. 3B) as in HEK293 cells. Again, the transcriptional activities of the P564A and the L574S mutant was unaffected by EGLN1 overexpression. These results suggested a novel function of EGLN1 for inhibiting HIF-1␣ transcriptional activity.
To provide further evidence, we determined C-ODD protein levels in cells co-expressed with EGLN1 by Western analysis. In Hep3B, ectopic expression of EGLN1 led to a Ͼ20% reduction of the protein levels in hypoxic conditions ( Fig. 3C; also see Fig. 1). However, EGLN1 overexpression failed to do so in the VHL-deficient C2 cells. Therefore, these results supported that EGLN1, in addition to promoting HIF-1␣ proteolysis, also independently represses the N-terminal transcriptional activity.
We also examined the effect of EGLN1 on endogenous HIF-1␣-mediated transcription with a reporter plasmid driven by the human EPO hypoxia-responsive element (18). EGLN1 overexpression significantly inhibited hypoxia-induced EPO reporter activity in Hep3B and C2 cells (Fig. 3D). Thus, these results indicated the biological relevance of EGLN1 as a repressor for HIF-1␣-mediated transcription.

EGLN1 Silencing Enhances HIF-1␣ Transcriptional Activity and Its Target Gene Expression in Hypoxia-To corroborate the role of EGLN1
in inhibiting HIF-1␣ transcriptional activity, we took a complementary approach in which EGLN1 expression was silenced by siRNA. Knockdown of EGLN1 in Hep3B cells resulted in a significant increase in HIF-1␣ N-terminal transcriptional activity under normoxic and hypoxic conditions (Fig. 4A). However, siRNA targeting EGLN2 or EGLN3 showed no effect, which differs from the EGLN2 and EGLN3 overexpression results above (see "Discussion").
To confirm the effectiveness of the siRNA, we determined protein levels of targeted genes by Western blot analysis. The extent and specificity of siRNA knockdown are illustrated in Fig. 4B. The protein level of each EGLN was strikingly and specifically reduced by its corresponding siRNA; neither control siRNA nor those against its isoforms had any significant effect. Importantly, knockdown of EGLN1 expression rendered HIF-1␣ stable in normoxia, whereas knockdown of the other two showed no obvious effect, consistent with the hypothesis that EGLN1, generally most abundant (47), is the key oxygen sensor setting low

. EGLN1 inhibition of HIF-1␣ transcriptional activity is not a result of increased HIF-1␣ proteolysis.
A, EGLN1 inhibition of HIF-1␣ transcriptional activity was tested as above with C-ODD and its mutants, as indicated, in the VHLdeficient C2 (C2) cell line. B, EGLN1 binding to C-ODD was examined in normoxic and hypoxic C2 cells expressed ectopically with EGLN1 and Gal4-C-ODD, as described in the legend for Fig. 1. The arrows denote detected protein complexes. C, the effect of ectopic EGLN1 on HIF-1␣ C-ODD expression levels was analyzed in VHL-proficient (Hep3B) and -deficient (C2) cell lines, respectively. After a 4-h hypoxic treatment, Western blot was performed with anti-Gal4 and anti-EGLN1 antibodies. Co-transfected YFP served as an internal control. The arrowhead denotes nonspecific detection, and the bracket specifies remaining signals from the blot above. D, Hep3B and C2 cell lines were transfected with the pEpoE-luc reporter in the absence or presence of ectopic EGLN1. EGLN1 overexpression inhibited endogenous HIF-1␣ transcriptional activity under hypoxic conditions. Relative luciferase units (RLU) of EpoE-luc in the y axis were plotted as means Ϯ S.D. from three independent experiments in duplicate. *, p Ͻ 0.01; **, p Ͻ 0.001.
To demonstrate that EGLN1 inhibits HIF-1␣ target gene expression in hypoxia, we asked whether EGLN1 siRNA alters hypoxic induction of PGK1, a hypoxia-responsive gene. Results in Fig. 4C from real-time RT-PCR show that only EGLN1 siRNA further elevated PGK1 mRNA levels in hypoxia. This result supported the hypothesis that EGLN1 modulates the HIF-1␣ target gene expression under hypoxic conditions, although its enzymatic activity is inactivated by limited oxygen. It is noted that although the relative abundance of the EGLN isoforms and the efficiency of corresponding siRNAs may vary in the cells, neither EGLN2 nor EGLN3 siRNA altered PGK1 expression under these conditions. Taken together, these results argued that only EGLN1 inhibits HIF-1␣ transcriptional activity in hypoxia by virtue of binding to HIF-1␣.
Stimulation of HIF-1␣ Transcriptional Activity and Enhancement of Its Target Gene Expression by EGLN1 siRNA Is VHL-independent-To demonstrate further that the increased expression of PGK1 in hypoxia by EGLN1 siRNA is not simply a result of increased HIF-1␣ stability, we asked whether knockdown of VHL expression in Hep3B cells would produce a similar result. Fig. 5A shows that the VHL siRNA effectively down-regulated VHL protein levels, in concomitance with an increased level of HIF-1␣. However, in contrast to the augmentation by EGLN1 siRNA, VHL silencing failed to further enhance PGK1 mRNA levels in hypoxia despite a marked increase in normoxia (Fig. 5B). Of note, EGLN1 expression was essentially unaffected by the VHL siRNA (Fig. 5A).
Likewise, in VHL-deficient C2 cells, only EGLN1 siRNA was able to stimulate HIF-1␣ N-terminal transcriptional activity (Fig. 6A) and to further elevated PGK1 expression under both normoxic and hypoxic conditions (Fig. 6B) without an additional increase in HIF-1␣ protein levels (Fig. 6C). Again, siRNA against EGLN2 or EGLN3 had no effects on the transcriptional activity and PGK1 expression. These results confirmed that EGLN1 specifically represses HIF-1␣ transcriptional activity in hypoxia.

DISCUSSION
The known function of the HIF prolyl-4-hydroxylases is to sense and transduce oxygen signals via prolyl hydroxylation of HIF-1␣, thereby B, EGLN isoforms in Hep3B cells were knocked down with their corresponding siRNA as indicated. EGLN1-3 denotes a mix of siRNAs targeting all three EGLN isoforms. Various antibodies as specified were used sequentially for Western blot analysis. Endogenous ␤-actin was used as a loading control. Note that HIF-1␣ protein expression was increased in normoxic conditions (N) by EGLN1 silencing. However, none of the siRNAs further elevated HIF-1␣ protein levels in hypoxic conditions (H). C, the effects of EGLN1 siRNA were examined on the HIF-1␣ target gene PGK1 expression. Hep3B cells were transfected with the corresponding siRNAs as above and subjected to analysis by real-time RT-PCR after a hypoxic treatment. PGK1 mRNA levels in reference to those of endogenous ACTB are presented in means Ϯ S.E.  NOVEMBER 11, 2005 • VOLUME 280 • NUMBER 45 triggering VHL binding, polyubiquitination, and proteasomal degradation (31,32). The observation that EGLN1 and EGLN3 themselves are hypoxia-inducible (31) has prompted investigations of the biological relevance of the increased hydroxylase levels. It has been shown that the rate of HIF-1␣ proteolysis, upon reoxygenation, depends on the duration of hypoxic stress; after prolonged hypoxia, accelerated HIF-1␣ degradation supervenes (48). Therefore, the hypoxic up-regulation of EGLN1 and EGLN3 apparently explains the buildup of the prolyl hydroxylases during hypoxia for rapid degradation of HIF-1␣ upon reoxygenation (40). However, the biological role for the accumulated hydroxylases during prolonged hypoxia remains unclear. It is conceivable that a negative-feedback mechanism would take place by modulating HIF-1␣ transcriptional activities. In fact, it has been demonstrated that HIF-1␣ CAD activity is inhibited during hypoxia by CITED2/ p35srj, a hypoxia-inducible gene product that competes for p300 binding (15). In contrast, the regulation of the HIF-1␣ N-terminal transcriptional activity remains obscure. Here we demonstrated that EGLN1 binds HIF-1␣ in hypoxia as well. Apart from its role in regulating HIF-1␣ protein levels, EGLN1 also modulates HIF-1␣ N-terminal transcriptional activity as a feedback mechanism to avoid excessive hypoxic response.

EGLN1 Suppresses HIF-1␣ Transcriptional Activity
During this study, the possibility of altered HIF-1␣ stability in hypoxia resulting from the manipulation of EGLN1 expression was always a concern. Although the increased protein degradation overshadowed the direct transcriptional repression by EGLN1, the effect of transcriptional repression was uncovered, nevertheless, with the use of VHL-deficient cells defective in HIF-1␣ degradation. Moreover, EGLN1 inhibited endogenous HIF-1␣ transcriptional activity. EGLN1 silencing specifically enhanced HIF-1␣ transcriptional activity and PGK1 expression, especially under hypoxic conditions without further increase in HIF-1␣ levels. Furthermore, EGLN1 silencing in VHL-deficient cells confirmed the inhibitory effect of EGLN1 on HIF-1␣ transcriptional activity. In addition, the lack of enhanced PGK1 expression, when VHL expression was knocked down, further supported that EGLN1-mediated HIF-1␣ degradation in normoxia and its inhibition of HIF-1␣ transcriptional activity in hypoxia are two distinct functions. Accordingly, we proposed that both EGLN1 and CITED2 fine-tune HIF-1␣ transcriptional activities under hypoxic conditions. It is noteworthy that forced expression of the HIF prolyl hydroxylases in cells often down-regulates HIF-1␣, as shown in this study as well as in a previous report (32). Obviously, results from such an approach often "violate" in vivo rules that govern the specificity of intracellular biochemical reactions. Therefore, the data should be interpreted with extreme caution. Although overexpression of the three hydroxylases all inhibited HIF-1␣ N-terminal transcriptional activity in our initial reporter assays, only EGLN1 silencing augmented HIF-1␣ transcriptional activity and PKG1 expression. These findings are consistent with our co-immunoprecipitation result that only EGLN1, but not EGLN2 and EGLN3, interacts with HIF-1␣ C-ODD (44). Thus, alternative approaches to gene overexpression are necessary to avoid potential pitfalls. While this report was in preparation, Baek et al. (49) reported that the protein OS-9 interacts with both HIF-1␣ and EGLN1 in normoxic and hypoxic conditions. OS-9 promotes HIF-1␣ hydroxylation, VHL binding, and proteasomal degradation. In addition, it inhibits HIF-1-mediated transcription. Apparently, OS-9 is not involved in EGLN1 inhibition of HIF-1␣ transcriptional activity because HIF-1␣ stability was not altered in this study. Nevertheless, further studies are warranted to identify the molecular mechanism(s) by which EGLN1 inhibits HIF-1mediated transcription. Our current data seemingly suggested that the EGLN1 hydroxylase activity per se is not necessarily required for inhibiting HIF-1␣ transcriptional activity, and therefore, it would be interesting to search for additional factors that interact with EGLN1 for mechanistic dissections. Very recently, Ozer et al. (50) reported that EGLN1, in addition to regulating HIF stability, modulates HIF function through the recruitment of the candidate tumor suppressor ING4.
Targeting the HIF prolyl hydroxylases is regarded as a potential therapeutic principle for manipulating HIF activity (5,51). The differential effects on the HIF-1␣ target gene expression by the respective silencing of EGLN1 and VHL also suggest that although inhibiting EGLN1 activity alone may give rise to a marked increase in HIF-1␣ levels, the resulting HIF-1␣ transcriptional activity may depend on the intracellular oxygen tension, i.e. hypoxia specifically augments HIF-1␣ target gene expression. By contrast, inhibiting VHL activity would render HIF-1␣ stable and constitutively active regardless of oxygen tension. Therefore, understanding the regulation of HIF-1␣ transcriptional activity by the prolyl hydroxylase may provide new insight into the pharmacological FIGURE 6. Stimulation of HIF-1␣ transcriptional activity and augmentation of PGK1 expression by EGLN1 siRNA in VHL-deficient cells. A, C2 cells were transfected with C-ODD to analyze the effect of EGLN1 siRNA on HIF-1␣ N-terminal transcriptional activity. Mock siRNA transfection (Mock), together with siRNAs targeting the respective firefly luciferase gene (luc), EGLN2, and EGLN3, served as controls. Relative luciferase units (RLU) of Gal4luc in the y axis were plotted as means Ϯ S.D. from three independent experiments in duplicate. *, p Ͻ 0.01; **, p Ͻ 0.001. B, PGK1 mRNA levels in C2 cells were determined by real-time RT-PCR after transfection with the indicated siRNAs. PGK1 mRNA levels in reference to those of endogenous ACTB are presented in means Ϯ S.E. C, Western blot analyses were performed to examine the specific effect of each siRNA as indicated. Endogenous ␤-actin was used as a loading control. None of the siRNAs increased HIF-1␣ protein levels under either normoxic (N) or hypoxic (H) conditions. The arrowhead denotes nonspecific detection. manipulation of the HIF prolyl hydroxylase activities for the treatment of ischemic/hypoxic disease.