Hypoxia-inducible Factor α Subunit Stabilization by NEDD8 Conjugation Is Reactive Oxygen Species-dependent*

Hypoxia-inducible factor α proteins (HIF-αs) are regulated oxygen dependently and transactivate numerous genes essential for cellular adaptation to hypoxia. NEDD8, a member of the ubiquitin-like family, covalently binds to its substrate proteins, and thus, regulates their stabilities and functions. In the present study, we examined the possibility that the HIF signaling is regulated by the neddylation. HIF-1α expression and activity were inhibited by knocking down APPBP1 E1 enzyme for NEDD8 conjugation but enhanced by ectopically expressing NEDD8. HIF-1α and HIF-2α were identified to be covalently modified by NEDD8. NEDD8 stabilized HIF-1α even in normoxia and further increased its level in hypoxia, which also occurred in von Hippel-Lindau (VHL) protein- or p53-null cell lines. The HIF-1α-stabilizing effect of NEDD8 was diminished by antioxidants and mitochondrial respiratory chain blockers. This suggests that the NEDD8 effect is concerned with reactive oxygen species driven from mitochondria rather than with the prolyl hydroxylase (PHD)/VHL-dependent oxygen-sensing system. Based on these findings, we propose that NEDD8 is an ancillary player to regulate the stability of HIF-1α. Furthermore, given the positive role played by HIF-αs in cancer promotion, the NEDD8 conjugation process could be a potential target for cancer therapy.

prolyl hydroxylases (PHD1-3) (2); subsequently, HIF-1␣ is ubiquitinated by von Hippel-Lindau protein (pVHL) and finally degraded by 26 S proteasome (3). However, this hydroxylation is limited under hypoxic conditions, which stabilizes HIF-␣s. Given the essential roles played by HIFs in tumor promotion, the HIF inhibition has become a frontline topic in research on new cancer therapies (4).
NEDD8 (neural precursor cell-expressed developmentally down-regulated 8) is conserved in eukaryotes and is ubiquitously expressed in most mammalian tissues. As NEDD8 is structurally similar to ubiquitin, it is classified as a member of the ubiquitin-like family (5). Furthermore, like ubiquitin, NEDD8 conjugates to its substrate proteins, which is named "neddylation," via a sequential process involving activation, conjugation, and ligation. Neddylation requires a unique set of conjugating enzymes, namely NEDD8-activating E1 complex, which is composed of APPBP1 and UBA3, NEDD8-conjugating E2 enzyme (UBC12), and various NEDD8-ligase E3 enzymes (6). Functionally, neddylation is essential for cell viability, development, and responses to stress, and thus, disorganized neddylation is regarded to be associated with the pathogeneses of neurodegenerative and neoplasmic diseases (7,8). Accordingly, neddylation is viewed as a potential therapeutic target. Recently, an E1 inhibitor MLN4924 was found to induce cell cycle arrest and apoptosis in cultured cancer cells and to effectively inhibit tumor growth in mice (9), which aroused much interest in the neddylation-targeting cancer therapy.
The ubiquitin-proteasome system is viewed as a promising target for the development of anticancer agents, and several proteasome inhibitors have been subjected to clinical trials in patients with multiple myeloma and in patients with some solid tumors (10). Proteasome inhibitors induce tumor cell death and inhibit tumor adaptation to hypoxia by inactivating HIF-1 (11), and thus, the double targeting of tumor cells and their microenvironments with proteasome inhibitors might synergistically inhibit tumor growth in vivo. Likewise, neddylation is now being introduced as a novel target for anticancer treatment. However, the role of neddylation in tumor growth has only been investigated in the context of tumor cell death and not with respect to tumor adaptation to hypoxia. Therefore, in the present study, we tested the possibility that NEDD8 is an ancillary player to regulate the stability of HIF-1␣. It is hoped that this work provides a rationale for the development of neddylation-targeting cancer therapy.
Luciferase Assay-Luciferase reporter genes, containing hypoxia responsive element of the human vegfa gene and the erythropoietin (epo) enhancer region, were constructed as described previously. HEK293 cells were cotransfected with 1 g each of reporter gene, 50 nM siRNAs, and ␤-gal plasmid using the calcium phosphate method. After stabilization for 48 h, cells were incubated under either normoxic or hypoxic conditions for 16 h. Luciferase activities were measured using a Biocounter M1500 luminometer (Lumac) and normalized to ␤-gal activities.
Immunoblotting and Immunoprecipitation-Total cell lysates in a SDS sample buffer were separated on SDS-polyacrylamide gels and transferred to Immobilon-P membranes (Millipore Bedford, MA). Membranes were blocked from nonspecific proteins with 5% skim milk in TTBS (Tris-buffered saline containing 0.1% Tween 20) for 30 min and incubated overnight with a primary antibody diluted at 1:1000 to 5000 in the blocking solution. Then, the membranes were incubated with a horseradish peroxidase-conjugated secondary antibody (diluted 1:5000 in the blocking solution) for 1 h, and blots were visualized using an ECL Plus kit (Amersham Biosciences). For immunoprecipitation, cell lysates (1 mg of protein) were incubated with 5 l of antibody for 2 h and then incubated with 10 l of protein A/G-Sepharose beads (GE Healthcare) for 4 h at 4°C. After washing, the immunoprecipitated proteins were eluted in the SDS sample buffer and subjected to SDS-PAGE and Western blotting.
Identification of His 6 -tagged NEDD8 Conjugates-Identification of NEDD8 conjugation was performed and modified based on the description in Jaffray and Hay (14). After transfection of plasmid expressing His 6 -tagged NEDD8 or NEDD8⌬GG, cells were divided into two dishes. One was lysed with SDS sample buffer and analyzed by Western blotting to confirm the expression level of proteins (input samples). The other was lysed by adding denaturing buffer (6 M guanidine hydrochloride, 0.1 M Na 2 HPO 4 /NaH 2 PO 4 , 0.01 M Tris-Cl (pH 8.0), plus 10 mM imidazole and 10 mM ␤-mercaptoethanol) directly to the cells. The lysates were mixed with Ni 2ϩ -NTA-agarose beads (Qiagen, Valencia, CA), prewashed with lysis buffer, and rotated for 4 h at room temperature. The beads were successively washed for 5 min in each step with the following solutions: lysis buffer (pH 8.0); washing buffer (pH 8.0) (8 M urea, 0.1 M Na 2 HPO 4 / NaH 2 PO 4 , 0.01 M Tris/HCl, pH 8.0, plus 20 mM imidazole, 10 mM ␤-mercaptoethanol); washing buffer (pH 6.3) plus 0.2% Triton X-100; and washing buffer (pH 6.3) plus 0.1% Triton X-100. Then, the beads were eluted with SDS sample buffer and analyzed by Western blotting.
Preparation of Nuclear Extract-Cells were harvested, washed twice with ice-cold PBS, resuspended in buffer A (20 mM Tris, pH 7.8, 1.5 mM MgCl 2 , 10 mM KCl, 0.2 mM EDTA, 0.5 mM DTT, 0.5 mM PMSF, protease inhibitor mixture, and 1 mM Na 3 VO 4 ), and cooled on ice. After adding 0.6% Nonidet P-40, lysed cells were centrifuged, and the supernatants were collected as the cytosolic fraction. Nuclear pellets were resuspended in a buffer B (400 mM NaCl and 5% glycerol in buffer A) and incubated at 4°C for 30 min. After centrifugation, the supernatants were kept as the nuclear fraction.
Assay of ROS-293T cells in the experimental conditions were washed by prewarmed PBS, culture media were replaced with a Hank's balanced salt solution that had been preincubated at 37°C in 5% CO 2 atmosphere, and then cells were treated with DCFDA (50 M) for 30 min in the dark and detached with trypsin-EDTA solution. After a brief washing, the oxidized form of DCFDA, fluorescent dichlorofluorescein, was excited at 488 nm and detected at 530 nm wavelength, using a FACStar flow cytometer.
Statistical Analysis-All data were analyzed using Microsoft Excel 2002 software, and results are expressed as means Ϯ S.D. We used the unpaired Student's t test to compare reporter activities. Differences were considered statistically significant at the p Ͻ 0.05 level. All statistical tests were two-sided.

RESULTS
Neddylation-dependent Activation of Hypoxic Signaling due to HIF-1␣ Stabilization-To examine whether the HIF-dependent hypoxic signaling is regulated by neddylation, we inhibited the initial step of neddylation by knocking down APPBP1. Under hypoxic conditions, VEGF and PDK1 mRNA levels increased in Hep3B cells, which was reversed by appbp1 siRNAs (Fig. 1A, upper). We also found that HIF-1␣ is downregulated by appbp1 siRNAs (Fig. 1A, lower). Furthermore, vegfa promoter and epo enhancer reporter analyses showed that the APPBP1 inhibition attenuated the transcriptional activity of HIF-1 under hypoxic conditions (Fig. 1B). To understand the mechanism underlying HIF-1␣ suppression by APPBP1 knockdown, we checked the synthesis and degradation of HIF-1␣ protein. Newly synthesized HIF-1␣ accumulated after blocking its degradation with MG132, but the rate of HIF-1␣ synthesis was constant regardless of APPBP1 expression (Fig. 1C). The rate of HIF-1␣ degradation was analyzed by reoxygenating cells having been subjected to hypoxia. Fig. 1D shows that the HIF-1␣ degradation was accelerated about three times in APPBP1 knockdown cells. These results suggest that neddylation is required for HIF-1␣ stabilization in hypoxia.
NEDD8 Conjugation of HIF-␣s-To investigate whether NEDD8 directly regulates HIF-␣s, we checked HIF-1␣ expression and activity in HEK293 cells that had been transfected with FLAG-NEDD8 or FLAG-NEDD8⌬GG (conjugation-defective due to Gly-75/76 deletion) plasmid. Endogenous HIF-1␣ level and activity were increased by NEDD8 overexpression, but not by NEDD8⌬GG (Fig. 2A). In addition, ectopically expressed HIF-1/2␣s were also stabilized by NEDD8 (Fig. 2B). We examined the interaction between HIF-1␣ and NEDD8 and found that the two proteins are co-precipitated by each other in hypoxic HEK293 cells (Fig. 2C). Then is HIF-1␣ conjugated with NEDD8? HEK293 cells, which had been transfected with His-NEDD8 or His-NEDD8⌬GG, were subjected to hypoxia, and proteins were pulled downed by Ni 2ϩ under a denaturing condition. Endogenous HIF-1␣ was induced in conjugation with NEDD8, but not by NEDD8⌬GG (Fig. 2D). Also, ectopically expressed HIF-␣s were identified to be conjugated with NEDD8 (Fig. 2E).

Calculation of Neddylated Form in the Stabilized HIF-1␣ by
Hypoxia-To determine the portion of neddylated HIF-1␣ under hypoxia, we measured how much HIF-1␣ is reduced by silencing APPBP1. HIF-1␣ levels were determined by immunoblotting (upper panel) and quantified using the ImageJ program (National Institutes of Health). The hypoxic expression of HIF-1␣ was reduced by 34% after blocking neddylation (Fig.  2F). We confirmed this result in a different way. After cells were The proteins in total cell lysates were analyzed by immunoblotting using specific antibodies. B, neddylation is required for the hypoxic activation of HIF. The luciferase reporter plasmid containing vegfa promoter or epo enhancer was cotransfected with siRNAs and ␤-gal plasmid into HEK293. The transfected cells were incubated under normoxic or hypoxic conditions for 16 h, and then luciferase activities were measured and normalized to ␤-gal activity. Results (means Ϯ S.D., n ϭ 8) are presented as relative values versus the normoxic si-Con groups. †, p Ͻ 0.01 versus the normoxic si-Con; *, p Ͻ 0.01 versus the hypoxic si-Con. C, HIF-1␣ synthesis is not affected by neddylation. HEK293 cells were transfected with siRNAs and stabilized for 2 days and then treated with 100 M cycloheximide for 1 h to remove the remaining HIF-1␣ completely. After washing out cycloheximide, cells were further incubated for the indicated times with MG132. Protein levels were analyzed by Western blotting (upper panel) and quantified using the ImageJ program (lower panel). D, HIF-1␣ stability depends on neddylation. HEK293 cells were transfected with siRNAs and stabilized for 48 h. Cells were exposed to 4 h of hypoxia and then subjected to normoxia for indicated times. Protein levels were analyzed by Western blotting (upper panel) and quantified using ImageJ (lower panel). Half-lives (t1 ⁄2 ) of HIF-1␣ were calculated from the slopes of the first-order decay curves. Points represent the means Ϯ S.D. of three experiments.
Identification of Neddylated Domains of HIF-␣s-We co-expressed HIF-1␣ fragments with His-NEDD8 and found that the N-terminal region of HIF-1␣ (amino acids 1-400) is neddylated (Fig. 3A). To further specify the neddylated domain, we rechecked the NEDD8 conjugation in N terminus-deleted mutants of HIF-1␣. The ⌬N1 mutant (amino acids 201-826) was still neddylated, but ⌬N2 mutant (amino acids 401-826) was scarcely neddylated (Fig. 3B). These results indicate that  (lower panel). B, ectopic HIF-1/2␣ proteins are stabilized by NEDD8. HA-HIF-1␣ or HA-HIF-2␣ (1 g) were cotransfected with FLAG-NEDD8 or FLAG-NEDD8⌬GG (3 g) into HEK293 cells. After stabilization for 48 h, cell lysates were analyzed by Western blotting. C, endogenous HIF-1␣ associates with NEDD8 under hypoxia. HEK293 cells were incubated in normoxia (N) or hypoxia (H) for 8 h. Cell lysates were immunoprecipitated using anti-HIF-1␣ or anti-NEDD8, and coprecipitated proteins were analyzed by Western blotting (IB) using anti-NEDD8 or anti-HIF-1␣, respectively. D, neddylation of endogenous HIF-1␣. His-NEDD8/⌬GG plasmid was transfected into HEK293 cells, and cells were lysed. His-NEDD8-conjugated proteins were bound to Ni 2ϩ -NTA-agarose beads under denaturing conditions and eluted with a SDS sample buffer. Proteins were analyzed by Western blotting. E, neddylations of ectopically expressed HIF-1␣ and HIF-2␣. After cotransfection with HIF-1/2␣ and His-NEDD8/⌬GG plasmids, HEK293 cells were lysed under denaturing conditions. His-NEDD8-conjugated HIF-␣s were isolated with Ni 2ϩ affinity resin. Proteins were analyzed by Western blotting. F, calculating neddylated HIF-1␣ using appbp1 siRNA. HEK293 cells were transfected with siRNAs and stabilized for 48 h, and then cells were exposed to 4 h of hypoxia. Protein levels were analyzed by Western blotting (upper panel) and quantified using ImageJ (lower panel). The amount of HIF-1␣ stabilized by neddylation was calculated as the difference between si-Con and si-APPBP1 groups. Bars represent the means Ϯ S.D. of six experiments. G, calculating neddylated HIF-1␣ using Ni 2ϩ pulldown. HEK293 cells were transfected with His-NEDD8 and stabilized for 24 h, and then cells were exposed to 4 h of hypoxia. Cell lysate was separated into six tubes, and Ni 2ϩ -NTA-agarose was added into the tubes at the indicated amounts. The mixtures were incubated for 2 h, and then Ni 2ϩ -bound proteins were separated by centrifugation. Ni 2ϩ -bound proteins were eluted as the same volume as that of supernatant. HIF-1␣ levels of supernatant or Ni 2ϩ -bound HIF-1␣ were analyzed by Western blotting (upper panel) and quantified using ImageJ (lower panel). Neddylated HIF-1␣ was calculated based on the ratio of Ni 2ϩ -bound HIF-1␣ level to total level. the neddylation site of HIF-1␣ is present between amino acids 201 and 400, which corresponds to the PAS-B domain.

DISCUSSION
Neddylation is now being introduced as a novel target for anticancer treatment. However, the role of neddylation in tumor growth has only been investigated in the context of tumor cell death and not with respect to tumor adaptation to hypoxia. In the present study, we found that the hypoxic gene regulation is ensured by neddylation of HIF-␣s and that the neddylation is required for the stabilization of HIF-1␣ by mitochondrial ROS.
The oxygen-dependent degradation of HIF-1␣ is initiated by the hydroxylation of prolines 402 and 564. Then, pVHL targets the hydroxylated HIF-1␣ and recruits elongin B/C, Cul2, and Rbx1, which forms the VBC-Cul2 E3-ubiquitin ligase complex to ubiquitinate HIF-1␣ (19). Of these components, pVHL and Cul2 have been identified to be covalently conjugated by NEDD8. In functional aspects, neddylated pVHL cannot recruit Cul2 to HIF-1␣ and rather preferentially interacts with fibronectin (20). Accordingly, the pVHL neddylation could negatively regulate the ubiquitination of HIF-1␣, but this possibility has not been investigated. On the contrary, Cul2 is neddylated after recruited by pVHL, and the neddylated Cul2 facilitates the polyubiquitination of HIF-1␣ (21)(22)(23). Therefore, the net effect of neddylation on HIF-1␣ stability is obscure. Nonetheless, Ohh et al. (24) suggested that the neddylation process is required for the ubiquitination and subsequent degradation of HIF-1␣. They demonstrated that HIF-1␣ is up-regulated at a non-permissive temperature in CHO cells having a temperature-sensitive mutation of APPBP1, which contradicts our results. Therefore, we checked whether or not the effects of NEDD8 on HIF-1␣ are linked with pVHL-mediated HIF-1␣ ubiquitination. Consequently, HIF-1␣ expression and activity were both still regulated by NEDD8 in VHL-defective cells. Given our results and a report of Ohh et al. (24), it is concluded that the direct neddylation of HIF-1␣ may overcome the HIF-1␣-destructive effect of neddylated Cul2. The reciprocal regulations of HIF-1␣ by neddylation remain to be investigated.