Ubiquitination of Inositol-requiring Enzyme 1 (IRE1) by the E3 Ligase CHIP Mediates the IRE1/TRAF2/JNK Pathway*

Background: Post-translational modification is involved in regulating IRE1 signaling. Results: CHIP ubiquitinated IRE1 and regulated the IRE1/TRAF2/JNK pathway. Conclusion: Ubiquitination is a post-translational modification that is required for IRE1 signal transduction. Significance: This study may provide the first evidence that the IRE1/TRAF2/JNK pathway is regulated by ubiquitination. Deciphering the inositol-requiring enzyme 1 (IRE1) signaling pathway is fundamentally important for understanding the unfolded protein response (UPR). The ubiquitination of proteins residing on the endoplasmic reticulum (ER) membrane has been reported to be involved in the UPR, although the mechanism has yet to be fully elucidated. Using immunoprecipitation and mass spectrometry, IRE1 was identified as a substrate of the E3 ligase CHIP (carboxyl terminus of HSC70-interacting protein) in HEK293 cells under geldanamycin-induced ER stress. Two residues of IRE1, Lys545 and Lys828, were targeted for Lys63-linked ubiquitination. Moreover, in CHIP knockdown cells, IRE1 phosphorylation and the IRE1-TRAF2 interaction were nearly abolished under ER stress, which may be due to lacking ubiquitination of IRE1 on Lys545 and Lys828, respectively. The cellular responses were evaluated, and the data indicated that CHIP-regulated IRE1/TRAF2/JNK signaling antagonized the senescence process. Therefore, our findings suggest that CHIP-mediated ubiquitination of IRE1 contributes to the dynamic regulation of the UPR.

Deciphering the inositol-requiring enzyme 1 (IRE1) signaling pathway is fundamentally important for understanding the unfolded protein response (UPR). The ubiquitination of proteins residing on the endoplasmic reticulum (ER) membrane has been reported to be involved in the UPR, although the mechanism has yet to be fully elucidated. Using immunoprecipitation and mass spectrometry, IRE1 was identified as a substrate of the E3 ligase CHIP (carboxyl terminus of HSC70-interacting protein) in HEK293 cells under geldanamycin-induced ER stress. Two residues of IRE1, Lys 545 and Lys 828 , were targeted for Lys 63linked ubiquitination. Moreover, in CHIP knockdown cells, IRE1 phosphorylation and the IRE1-TRAF2 interaction were nearly abolished under ER stress, which may be due to lacking ubiquitination of IRE1 on Lys 545 and Lys 828 , respectively. The cellular responses were evaluated, and the data indicated that CHIP-regulated IRE1/TRAF2/JNK signaling antagonized the senescence process. Therefore, our findings suggest that CHIPmediated ubiquitination of IRE1 contributes to the dynamic regulation of the UPR.
In the endoplasmic reticulum (ER), 3 synthesized proteins mature and are assembled for delivery to target organelles. The homeostasis of the folded proteins in the ER is monitored and maintained via a coordinated adaptive pathway termed the unfolded protein response (UPR) (1). Pathophysiological insults lead to the accumulation of unfolded proteins in the ER and cause ER stress. In response to the accumulation of unfolded proteins, cells adapt to this stressful condition via the UPR, initially to restore normal cellular function by halting protein translation and activating signaling pathways that increase the expression of molecular chaperones involved in protein folding (2,3). In the UPR, inositol-requiring enzyme 1 (IRE1) is one of three key sensors that reside on the ER membrane. It is generally accepted that IRE1 has dual functions to cope with ER stress. On the one hand, in response to ER stress, IRE1 dimerizes to activate its cytosolic RNase domain and to cleave a 26-nucleotide intronic sequence from XBP1 (X-box DNA binding protein 1) mRNA (4). The resulting XBP1, an activator of UPR genes, activates the expression of numerous chaperone genes to improve the protein folding capacity in the ER (5,6). On the other hand, activated IRE1 recruits TNF receptor-associated factor 2 (TRAF2) to the ER membrane and activates the pro-apoptotic ASK1 (apoptosis signal-regulating kinase)-JNK pathway (7).
The mechanisms by which IRE1 is regulated in the UPR have yet to be explored. Recently, the functions of IRE1 have been reported to be regulated via post-translational modifications such as ADP-ribosylation by poly(ADP-ribose) polymerase 16 (8), dephosphorylation by protein phosphatase 2C on endoplasmic reticulum (9), and ubiquitination by the E3 ligase HRD1 (10). These studies really have initiated greater understanding about the working mechanism of IRE1, but they have also introduced further questions. Except the proteins anchored on ER membrane, can the proteins not located on ER perform modifications to IRE1 under ER stress? Are there additional types of post-translational modifications added to IRE1 under ER stress? As IRE1 responds to ER stress through two pathways, XBP-1 splicing and IRE1/TRAF2/JNK, how does an IRE1 modification regulate them, one or both pathways? These questions prompted us to investigate IRE1 modifications more thoroughly.
CHIP (carboxyl terminus of HSC70-interacting protein) has been identified as a regulator that maintains protein homeostasis, particularly by targeting various pathological unfolded proteins in numerous tissues and cells (11). It is generally accepted that CHIP is not localized to a specific organelle, whereas it is found on the ER outer membrane (12). Many studies have shown that CHIP is an E3 ubiquitin ligase for endoplasmic reticulum-associated degradation (ERAD) (13,14), a process involved in cell adaptation to ER stress. Because ERAD is often detected around the ER membrane, a logical deduction is that CHIP may exert some functions on the ER membrane under ER stress. However, whether CHIP participates in the UPR, the specific branch of the UPR that CHIP participates in and how the information of CHIP ubiquitination is delivered in response to ER stress are still largely unknown. In this study, we performed IP assays followed by a proteomic screen and identified one E3 ligase, CHIP, in the IRE1 immunoprecipitates. CHIP interacted with and ubiquitinated IRE1 in a Lys 63 -linked manner. The ubiquitination event was likely critical for regulating the IRE1/TRAF2/JNK pathway and affecting cellular senescence. Our findings suggest that the post-translational modification of IRE1 by the E3 ligase CHIP contributes to the dynamic control of IRE1 signaling in response to ER stress.

EXPERIMENTAL PROCEDURES
Vector Construction for Recombinant Expression, Site-directed Mutagenesis, and RNA Interference-For recombinant expression in HEK293 cells, full-length human IRE1, CHIP, TRAF2, OTUB1 (OTU domain-containing ubiquitin aldehydebinding protein 1), and ubiquitin (Ub) cDNAs were amplified from HEK293 cDNA by RT-PCR and cloned into pCMV-HA and pCMV-MYC. For pulldown experiments with GST-fused recombinant protein, the CHIP gene was inserted into pGEX-4T-1 to generate the pGEX-4T-1-CHIP fusion. For in vitro ubiquitination assay with His tag-fused recombinant protein, the CHIP gene was inserted into pET-30a(ϩ) to generate the pET-30a-CHIP fusion.
For the site-directed mutagenesis, plasmids containing the CHIP, IRE1, and Ub genes were mutated via single site-directed mutagenesis (Agilent, Santa Clara, CA) using the corresponding mutated primers that contained the indicated mutated amino acid. The mutation sites were predicted using the BDM-PUB (Computational Prediction of Protein Ubiquitination Sites with a Bayesian Discriminant Method) algorithm.
Immunofluorescent Confocal Microscopy-HEK293 cells were placed on glass cover slips, and subconfluent cells were treated. The treated and untreated cells were fixed in 3.7% paraformaldehyde and permeabilized with 0.3% Triton X-100 in PBS at room temperature. The indicated primary antibodies diluted in PBS containing 1% bovine serum albumin and 0.3% Triton X-100 were incubated with the cells at 4°C overnight. The secondary antibodies, goat anti-rabbit IgG-TRITC, or goat anti-mouse IgG-FITC (Santa Cruz Biotechnology), were incubated with the cells for 1 h at room temperature. All of the labeled cells were examined using a Leica TCSSP5 confocal fluorescence imaging microscope (Leica Microsystems, Wetzler, Germany) equipped with a 63ϫ oil immersion objective with a numerical aperture of 1.4 at room temperature. The confocal images were acquired using LAS AF software, version 2.0 (Leica Microsystems). DAPI was used to stain the nuclei.
Western Blot-The lysates or affinity beads were boiled in 20 l of Laemmli buffer containing 50 mM Tris-HCl, pH 6.8, 30% glycerol, 4% SDS, and 1% ␤-mercaptoethanol. The samples were subjected to 8% or 10% SDS-PAGE, and the electrophoresed gels were electrotransferred onto polyvinylidene difluoride membranes (Millipore, Billerica, MA). The blotted membranes were probed with the indicated primary antibodies at various dilutions and then incubated with HRP-conjugated secondary antibodies. The immunorecognition signals were detected using ECL and were acquired using ImageQuant ECL (GE Healthcare). Immunoprecipitation and Pulldown to Capture Associated Proteins-Transiently or stably transfected HEK293 cells were collected from culture dishes and lysed with radioimmune precipitation assay buffer, pH 7.5, containing 50 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, and a protease inhibitor mixture. For co-immunoprecipitation, the cell lysates were precleared with 25 l of protein A/G PLUS-agarose beads (Santa Cruz Biotechnology) for 4 h and then incubated with antibody (2 g) overnight at 4°C, after which 25 l of protein A/G PLUS-agarose beads were added. After washing with ice-cold radioimmune precipitation assay buffer, the proteins bound to the beads were analyzed by Western blot.
The GST-CHIP fusion protein was expressed in Escherichia coli BL21 cells, and the overexpressed recombinant fusion protein was purified by affinity chromatography with glutathione-Sepharose beads (GE Healthcare). The GST or GST-CHIP recombinant protein was incubated with the precleared HEK293 cell lysates harvested from transiently IRE1-expressing cells. Glutathione-Sepharose beads were added to the mix-ture, which was incubated overnight at 4°C. After washing with ice-cold radioimmune precipitation assay buffer, the proteins bound to the beads were analyzed by Western blot.
Transient and Stable Transfection-The constructed vectors were transfected into HEK293 cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Six hours after transfection, the medium was refreshed, and the cells were maintained in an incubator at 37°C in 5% CO 2 . Genes expression or knockdown was examined by Western blot 24 -48 h after transient gene transfection in HEK293 cells. For stable gene knockdowns, the transfected cells were cultured and selected in medium containing 600 g/ml G418. G418resistant cell colonies were isolated for further culture, and the expression of the target gene was assessed by Western blot.
In Vitro Ubiquitination Assay-The His-CHIP fusion protein as the E3 ligase was expressed in E. coli BL21 cells, and the recombinant fusion protein was purified by affinity chromatography with nickel-nitrilotriacetic acid beads. The recombinant protein GST-IRE1 (amino acids 465-977) as the ubiquitination substrate was purchased from Sino Biological Inc. (Beijing, China). The in vitro ubiquitination assay kit was purchased from Enzo Life Sciences (Plymouth Meeting, PA). Ubiquitination was analyzed following protocols recommended by the manufacturer.
SA-␤-gal Assay to Evaluate Cellular Senescence-HEK293 cells were seeded into six-well plates, and after 24 h, the cells were treated as indicated. SA-␤-gal staining was performed according to the manufacturer's protocol (Cell Signaling Technology). To estimate the amount of senescence, the percentage of cells positive for SA-␤-gal staining was calculated in several randomly selected representative fields (n ϭ 10).
Identification of IRE1-interacting Proteins-The immunocomplexes precipitated with protein A/G beads bound to an IRE1 antibody were treated with Laemmli buffer and analyzed by 12% SDS-PAGE. The electrophoresed gels were visualized using silver nitrate staining and then cut into several pieces. The gel slices were dried using acetonitrile followed by alkylation with iodoacetamide and reduction with DTT. The treated gel pieces were incubated with trypsin (Promega, Madison, WI) overnight at 4°C. The peptides generated from tryptic digestion were injected into a Dionex LC-Packings Ultimate3000 HPLC (Amsterdam, the Netherlands) mounted with a C18 column (75 m ϫ 150 mm) and run with a linear acetonitrile gradient (0 -60%) at a flow rate of 400 nl/min. The eluted peptides were injected directly into a Q Exactive hybrid quadrupole-Orbitrap MS (Thermo Fisher, Bremen, Germany) using nanoESI spray at an ion spray of 1800 V. Intact masses were measured at a resolution of 70,000 (at m/z 200) in the Orbitrap using an AGC target value of 3 ϫ 10 6 charges. The top 20 peptide signals (charge states 2 ϩ and higher) were submitted to MS/MS in the HCD cell (m/z 2 isolation width, 27% normalized collision energy). The output files from the LC-MS/MS analyses were converted to mgf files using Protein Discovery 1.3 (Thermo Fisher, Bremen, Germany) and delivered to the Mascot 2.3 search engine (Matrix Science, Boston, MA), in which the data were queried against the Swiss-Prot database 57.15 for human proteins with the following search parameters: peptide mass tolerance, 15 ppm; fragment mass tolerance, 20 mmu; methionine oxidation and the amino terminus of peptides changed from Gln to pyroGlu as variable modifications; and carbamidomethylation of cysteine as a static modification.
Statistical Analysis-All experimental data were generated from experiments that were performed at least in triplicate, and the corresponding data are presented as the means Ϯ S.D. The significant differences between two groups were determined using Student's t test, and the null hypothesis was rejected in cases of p Ͻ 0.05. The quantitative evaluation of the Western blot images was performed using ImageQuant TL software (Piscataway, NJ).

RESULTS
CHIP Interacted with IRE1-In HEK293 cells with IRE1 overexpression, the IRE1 antibody was used to immunoprecipitate protein complexes associated with IRE1. The immunoprecipitated proteins were separated by SDS-PAGE and stained with silver nitrate. All the stained bands were excised, digested with trypsin, and delivered to LC-MS/MS for protein identification. As shown in Table 1, five ubiquitination-related proteins were solely identified in the immunoprecipitates with the IRE1 antibody: three E3 ligase proteins (CHIP and two SCF (Skp1-Cullin-F-box) subunits (FBXL21 and FBXO11)) and two deubiquitinases (OTUB1 and UBHL1). This finding indicates that ubiquitination may be a critical modification on IRE1. The detection of CHIP is intriguing based on its function in ERAD. We conducted co-immunoprecipitation assays to confirm the interaction of CHIP and IRE1 found by mass spectrometry. CHIP was found in the IRE1 immunoprecipitates from the cells co-transfected with HA-IRE1 but was limitedly detected in the cells transfected with CHIP alone (Fig. 1A). When IgG was used as an immunoprecipitation reagent in the same experiment, no CHIP was detected by the MYC antibody (Fig. 1B). Furthermore, endogenous IRE1 and CHIP were co-immunoprecipitated using an IRE1 antibody and normal IgG as the control (Fig. 1C). Next, GST and recombinant GST-CHIP were utilized to pull down overexpressed IRE1 in HEK293 cells. As depicted in Fig. 1D, the immunosignal against IRE1 was present only in the pulldown products of GST-CHIP. Finally, the co-localization of CHIP and IRE1 in cells was examined by confocal microscopy, which implicates that CHIP partially co-localizes with IRE1 (Fig. 1E). Because the nucleus occupies two-thirds of the volume of the entire HEK293 cell, we chose another cell line, A549, to repeat the co-localization experiment. The confocal images depicted in Fig. 1F demonstrated that IRE1 was present in the region around the nucleus where the ER was located, whereas CHIP was found throughout the cytoplasm with certain distribution patterns. Moreover, CHIP was enriched in some areas, and these areas co-localized with IRE1 as well.
These results indicate that CHIP interacts with IRE1.
CHIP Ubiquitinated IRE1-IRE1 is a substrate of an E3 ligase, and the ubiquitination of IRE1 could increase its degradation rate (10). Because CHIP is a typical E3 ligase that ubiquitinates its substrates and regulates their turnover rates, we inquired whether CHIP ubiquitinated IRE1. In transfected cells under normal conditions, IRE1 was not obviously ubiquitinated, regardless of whether the cells were overexpressing CHIP ( Fig.  2A). Because IRE1 is a transducer of and is activated in response to ER stress, we utilized geldanamycin (GA) (15) to induce ER stress; the effects of GA were confirmed using the ER stress indicators BIP (binding immunoglobulin protein) and protein disulfide isomerase (Fig. 2, B and C). In response to ER stress, IRE1 was significantly ubiquitinated, and this ubiquitination was augmented by CHIP overexpression (Fig. 2A). Considering the possibility of CHIP autoubiquitination as reported by Woo et al. (16), we performed experiments to test whether ER stress could induce CHIP autoubiquitination to rule out the possibility that the ubiquitination bands found in Fig. 2A were the ubiquitinated forms of CHIP. The results indicate that GA could not induce CHIP autoubiquitination (data not shown). Next, we transfected the CHIP shRNA vector into HEK293 cells and obtained CHIP knockdown stable cell lines (Fig. 2D). The ubiquitination of IRE1 was markedly attenuated by endogenous CHIP knockdown, further indicating that endogenous CHIP ubiquitinates IRE1 (Fig. 2E). The two functional sites in CHIP, Lys 30 (interacts with HSC70) and His 260 (catalytic activity) (17), were mutated and used to evaluate which site was necessary for CHIP to ubiquitinate IRE1. As presented in Fig. 2F, K30A CHIP slightly weakened IRE1 ubiquitination compared with wildtype CHIP; however, the ubiquitination was nearly abolished by H260Q CHIP. The evidences above indicate that CHIP ubiquitinates IRE1 in vivo. We next investigated whether CHIP directly ubiquitinated IRE1. We obtained His-CHIP and GSTfused IRE1 cytoplasmic tail (amino acids 465-977) recombinants and measured the IRE1 ubiquitination using an in vitro ubiquitination kit. As shown in Fig. 2G, CHIP could directly ubiquitinate GST-IRE1 (amino acids 465-977) in vitro, whereas the ubiquitination only occurred with UbcH13/Uev2. These results demonstrate that CHIP acts as an E3 ligase to ubiquitinate IRE1. with an IRE1 antibody and analyzed by Western blot with antibodies against IRE1 and MYC. B, the lysates of HEK293 cells co-transfected with HA-IRE1/MYC-CHIP were immunoprecipitated with IgG, IRE1 antibody, or without antibody, respectively. The immunoprecipitated proteins were analyzed by Western blot with antibodies against IRE1 and MYC. C, HEK293 cell lysates were immunoprecipitated with an IRE1 antibody or normal IgG (control), followed by Western blot with antibodies against IRE1 and CHIP. D, HA-IRE1-transfected HEK293 cell lysates were pulled down using GST-CHIP or GST (25 g), followed by Western blot with antibodies against IRE1 and GST. E, confocal microscopy images were obtained from HEK293 cells co-transfected with HA-IRE1 (red) and MYC-CHIP (green); the nuclei were imaged by DAPI staining (blue). Bar, 10 m. F, confocal microscopy images were obtained from A549 cells co-transfected with HA-IRE1 (red) and MYC-CHIP (green); the nuclei were imaged by DAPI staining (blue). Bar, 10 m. IP, immunoprecipitation; WB, Western blot.
CHIP Ubiquitinated IRE1 on Lys 545 and Lys 828 in a Lys 63linked Manner-Generally, CHIP brings the ubiquitin chain to its substrates and stimulates their degradation (18,19). Therefore, we investigated whether CHIP-mediated IRE1 ubiquitination increased the turnover of IRE1. IRE1 abundance was consistent in both CHIP knockdown cells (Fig. 2E) and CHIP overexpressing cells (Fig. 2A). Cycloheximide, an inhibitor of protein synthesis, was used to further evaluate the impact of CHIP on IRE1. As illustrated in Fig. 3A, exogenous IRE1 expression gradually decreased after cycloheximide treatment in a time-dependent manner, with a nearly 70% decrease in IRE1 by 8 h. When CHIP was overexpressed in these cells, the degradation of IRE1 was not enhanced but blocked in some level. A similar phenomenon occurred in cells with increased ER stress (Fig. 3B).
For further evaluation of the effect of CHIP abundance on the degradation of endogenous IRE1, we performed the degradation assay in HEK293 stable cell lines with/without CHIP RNAi. The Western blot images showed similar patterns in these two stable cell lines, and the same phenomenon was found under ER stress condition (data not shown). These results indicate that CHIP does not induce endogenous IRE1 degradation. Combining the evidence from these figures, we conclude that CHIP does not induce IRE1 degradation.
Substrate ubiquitination by CHIP does not always result in enhanced degradation. Ubiquitination at Lys 48 favors protein degradation, whereas ubiquitination at Lys 63 likely participates in regulating signal transduction (20). Therefore, we identified the ubiquitin ligation site(s) on IRE1 by co-transfecting wildtype or mutant ubiquitin with CHIP and IRE1. Compared with wild-type ubiquitin in cells under ER stress, K48R ubiquitin did not significantly change IRE1 ubiquitination; however, K63R and double-mutated ubiquitin nearly depleted IRE1 ubiquitination (Fig. 3C). These results suggest that CHIP ubiquitinates IRE1 in a Lys 63 -linked manner. Furthermore, it is believed that the manner of ubiquitination depends primarily on the ubiquitin-conjugating enzyme (E2) that is involved in the ubiquitination process (20). As shown in Fig. 2G, we found that CHIP could only ubiquitinate the IRE1 cytoplasmic tail (amino acids 465-977) with UbcH13. This observation is consistent with the report by Zhang et al. (21) that the E2 UbcH13 collaborated with CHIP to specifically forms a Lys 63 -linked ubiquitin chain, which assisted CHIP to ubiquitinate its substrates. Taking all the evidence above into account, we come to a conclusion that CHIP ubiquitinates IRE1 in a Lys 63 -linked manner.
Next, we examined the ubiquitin-conjugating sites on IRE1. Utilizing the Bayesian discriminant method for ubiquitination sites, the 16 top ranked sites were individually mutated from lysine to arginine (Table 2) to determine their ubiquitination potential. Except for K267R, the other 15 IRE1 mutants could be overexpressed in cells and recognized by the IRE1 antibody. Compared with wild-type IRE1, the K545R and K828R IRE1 mutants exhibited a significant change in ubiquitination in response to ER stress (Fig. 3D), demonstrating that these two lysine residues are likely to be CHIP modification sites.
CHIP Participated in the IRE1/TRAF2/JNK Pathway-Because Lys 63 ubiquitination by CHIP has been reported to be involved in signal transduction (22), we determined whether CHIP played a functional role in IRE1 signaling pathways in response to ER stress. We first verified that GA activated IRE1related pathways. We found that GA induced IRE1 phosphorylation in a dose-dependent manner within the GA concentration from 0 to 1 M and in a time-dependent manner within 12 h. However, the phosphorylation of IRE1 was attenuated at 24 h or with 5 M GA (Fig. 4, A and B). The influence of CHIP on IRE1/XBP1 function was then examined. There was no difference in XBP1 activation between CHIP knockdown and control cell lines, implying that the CHIP-IRE1 interaction does not regulate the IRE1-XBP1 pathway (data not shown). Next, the effect of CHIP on the IRE1/TRAF2/JNK function was evaluated. In response to ER stress, the IRE1-TRAF2 interaction activates ASK1, a downstream protein that relays the UPR signal to the stress-activated protein kinase JNK (7). Therefore, the phosphorylation status of the related proteins was examined. The phosphorylation of IRE1, JNK and p38 decreased in  Table 2. The cells in C and D were treated with or without 1 M GA for 12 h, followed by IP with an IRE1 antibody and Western blot with IRE1 and MYC antibodies. All data are from at least three independent experiments. The p values were calculated using Student's t test. CHX, cycloheximide; CTRL, control; IP, immunoprecipitation; WB, Western blot. response to CHIP knockdown under ER stress (Fig. 4C). The macroautophagy phenotype, which is regulated by JNK, downstream of IRE1 (23), was then assessed. The result corresponded with the IRE1/TRAF2/JNK activation status (Fig. 4D). Then we wondered whether the phosphorylation of IRE1 correlated with its ubiquitination. We found that the phosphorylation status of IRE1 was increased when it was ubiquitinated by CHIP. However, the K545R but not K828R mutant form of IRE1 abolished its phosphorylation under ER stress (Fig. 4E). This finding strengthens the deduction that the phosphorylation of IRE1 depends on its ubiquitination. In addition, we found that the interaction between exogenous IRE1 and TRAF2 was enhanced under ER stress in control cells, which agrees with the previous observation that the IRE1-TRAF2 complex only forms in response to ER stress (7); however, in CHIP knockdown cells, this interaction was completely inhibited (Fig. 4F). To examine which residue, Lys 545 or Lys 828 , ubiquitination of IRE1 is critical for the IRE1/TRAF2 interaction, the two vectors with IRE1 mutated as K545R or K828R were transfected into HEK293 cells, followed by the same experiment as above to assess the IRE1/TRAF2 interaction. As illustrated in Fig. 4G, the K828R mutant nearly abolished this interaction even under the same ER stress; however, the K545R mutant did not significantly affect the interaction. This result suggests that the ubiquitination of IRE1 at Lys 828 is critically important for IRE1/TRAF2 interaction, which is possibly involved in the regulation of the IRE1/TRAF2/JNK branch. Therefore, the evidence presented above demonstrates that CHIP likely regulates the IRE1/ TRAF2/JNK pathway. Moreover, as shown in Table 1, the deubiquitinase OTUB1 was also present in the IRE1 immunoprecipitates. Therefore, we hypothesized that OTUB1 might deubiquitinate IRE1 and thereby regulate the IRE1/TRAF2/JNK pathway. In HEK293 cells, overexpression of OTUB1 abolished CHIP-mediated IRE1 ubiquitination under ER stress (Fig. 4H). Moreover, the activation of IRE1, JNK, and p38 was also remarkably inhibited in cells overexpressing OTUB1 (Fig. 4I). Based on these observations, it is reasonable to deduce that OTUB1 removes the ligated ubiquitin on IRE1 under ER stress, thereby regulating the IRE1/TRAF2/JNK pathway. Thus, ubiquitination is indeed a modification involved in the activation of the IRE1/TRAF2/JNK pathway.
The IRE1/TRAF2/JNK Pathway Antagonized Cellular Senescence under ER Stress-The involvement of IRE1 in multiple cellular processes, such as proliferation, apoptosis, and autophagy, has been well established (23)(24)(25). A recent report showed that IRE1 was involved in senescence. Denoyelle et al. (26) determined that IRE1 and XBP1 played a pro-senescence role in mutant Ras-driven senescence. In this study, we examined whether the IRE1/TRAF2/JNK pathway played a role in senescence. A stable cell line was selected after transfection of an shRNA vector targeting IRE1 (Fig. 5A). There was no difference in the ratio of senescent cells between IRE1 knockdown cells and control cells; however, senescence markedly increased in IRE1 knockdown cells under ER stress (Fig. 5B). This result indicates that IRE1 potentially regulates cellular senescence in response to ER stress. Additionally, by counting cell numbers, we generated the growth curves of IRE1 knockdown stable cell lines and normal stable cell lines under the condition with/ without GA treatment. The result shown in Fig. 5C was consistent with the observation obtained from SA-␤-gal staining (Fig.  5B). Furthermore, to clarify whether GA performs a specific effect to induce senescence process, we tested GA and two other ER stress inducers, DTT and tunicamycin, in a parallel experiment. All three reagents were able to cause senescence augmentation in IRE1 knockdown stable cell lines, implying that GA does have a function on cell behavior comparable with those of other ER stress inducers (data not shown). We then studied whether this senescence event correlated with the IRE1/TRAF2/JNK pathway. A JNK inhibitor (SP600125) and a p38 inhibitor (SB203580) were used to inhibit the activation of this pathway. As depicted in Fig. 5D, the senescence ratio after GA treatment was significantly increased regardless of which inhibitor, SP600125 or SB203580, was added, suggesting that activation of the IRE1/TRAF2/JNK pathway antagonizes cellular senescence.
CHIP Antagonized the Senescence of HEK293 Cells under ER Stress-The involvement of CHIP in cellular senescence is well documented (27). As described above, CHIP facilitated IRE1 ubiquitination, and IRE1 participated in the senescence pathway in response to ER stress. Therefore, we ascertained the role of CHIP in IRE1-related senescence. First, senescence and cell growth were examined in cells with or without CHIP knockdown. The results showed that ϳ20% of the cells were positive for ␤-gal (Fig. 6A), and the cell growth rate was slightly decreased in CHIP knockdown cells (Fig. 6B). Next, we evaluated cellular behavior under ER stress. In CHIP normal cells, the ␤-gal staining was almost unchanged, but the cell proliferation rate decreased by nearly 50% in response to ER stress. In CHIP knockdown cells, ϳ60% of the cells were strongly positive for ␤-gal, and cell proliferation nearly ceased in response to ER stress (Fig. 6, A and B). Finally, after GA treatment, the recovery ability of these two cell lines was determined. The stressed and reseeded CHIP knockdown cells continued without proliferating until cell death occurred; however, the cells with normal CHIP expression almost completely recovered their proliferation ability within 48 h (Fig. 6C). As irreversibility is a common feature of senescent cells, the data suggests that CHIP is indeed involved in the cellular senescence process under ER stress.

DISCUSSION
We have systematically investigated one of the post-translational modifications of IRE1 in HEK293 cells under ER stress. Our data revealed that CHIP interacted with IRE1, leading to its ubiquitination, and that this ubiquitination event regulated the IRE1/TRAF2/JNK pathway, which possibly affected cellular senescence. These discoveries expand our understanding of the UPR, especially regarding the functions of CHIP under ER stress. Although some results described above are consistent with previous observations or conclusions, some remain controversial with others.
CHIP has been demonstrated to participate in the ERAD pathway. Cytoplasmic CHIP is believed to enhance the ubiquitination of retrotranslocated proteins and then to expedite the degradation of unfolded proteins (28). In contrast to the traditional view, the discoveries described in this study suggest that the ubiquitination of IRE1 by CHIP is likely to mediate signaling rather than protein degradation under ER stress. Similar to our results, Yang et al. (22) claimed that the CHIP ubiquitination could affect signal transduction but not protein degradation in RAW264.7 cells, whereas Gao et al. (10) perceived that an ER resident E3-ligase did ubiquitinate IRE1. We therefore postulate that CHIP performs as a regulator with dual   OCTOBER 31, 2014 • VOLUME 289 • NUMBER 44 functions. The ERAD machinery is abundantly activated once ER stress is induced. We believe that because the proteasomedependent degradation of unfolded proteins is generally a timeconsuming process, whereas the UPR is a rapid response to ER stress, the degradation path is not always efficient enough to remove unfolded proteins while coping with greater ER stress. According to our model of dual functions, we assume that CHIP-mediated ubiquitination is an economical response to ER stress, which functions both by facilitating the degradation of unfolded proteins and by modifying IRE1 to activate the UPR pathway.

CHIP Regulates IRE1 Signaling via Ubiquitination
We identified one ubiquitin-conjugating site, Lys 545 , on IRE1 that was likely to be associated with IRE1 phosphorylation. It is well accepted that phosphorylation of IRE1 is required for the formation of IRE1 dimers, whereas IRE1 dimerization leads to its conformational changes and further stimulates IRE1 kinase and RNase activity (29 -31). Based on topological analysis, Lys 545 is located at the connecting domain that links the transmembrane and cytosolic domains. Therefore, we assume that the K545R mutation abolishes the juxtaposition of the cytosolic domains of two IRE1 monomers so as to impair the IRE1 phosphorylation. In addition, we observed that the IRE1-TRAF2 interaction was weakened in CHIP knockdown cells and was dependent on IRE1 ubiquitination at Lys 828 under ER stress. OTUB1, a deubiquitinase, could almost completely abolish CHIP-mediated IRE1 ubiquitination and thus inactivated IRE1/JNK and p38 signaling. An analysis of IRE1 topology suggests that this ubiquitination site is not located in the RNase domain (mapped to 837-977) but likely resides in the kinase domain, where it is essential for TRAF2 binding (7). A logical deduction is that the ubiquitination at Lys 828 acts as a platform for TRAF2 anchoring. Our observation is in partial agreement with the findings from Arshad et al. (32). Under ER stress induced by staurosporine, the authors observed that the knockdown of RNF13, an E3 ligase, impaired the IRE1/TRAF2 signaling axis and caused JNK inactivation. Therefore, we propose IRE1 ubiquitination as a key modification for regulation of IRE1/TRAF2/JNK signaling under ER stress. Summarizing all the analysis above, a mechanism model for the hypothesis is illustrated in Fig. 7.
Senescence has been reported to be regulated by the UPR. As mentioned above, the ectopic expression of H-RAS G12V in primary human melanocytes regulated the UPR and induced senescence (26); on the other hand, aging was accelerated in CHIP knock-out mice (27). CHIP was gradually activated during senescence and regulated the abundance of p53 and oxidized proteins (33). The question remains as to what senescence mechanisms are associated with the UPR and CHIP. Our results summarized above seem to offer a proper answer to this question, which is basically in agreement with the observation that the activity of JNK is tightly correlated with senescence development (34). Furthermore, we determined that CHIP knockdown impaired the IRE1/TRAF2/JNK pathway, and this impairment was accompanied by decreased autophagy in HEK293 cells. Abundant evidence has demonstrated that this pathway provides important stimulation for initiating autophagy (23,35), which may antagonize senescence (36,37). Our findings are consistent with these reports. It is well known that CHIP could ubiquitinate lots of substrates that may regulate senescence. The directly experimental evidence that supports senescence regulated by IRE1 ubiquitination related with CHIP is not gained yet. However, the correlation evidence described in this study has set a solid foundation for the hypothesis that CHIP participates in the UPR, at least through IRE1-related signaling transduction, and partially regulates senescence.