Caveolin-1 Interacts with Derlin-1 and Promotes Ubiquitination and Degradation of Cyclooxygenase-2 via Collaboration with p97 Complex*

Background: Caveolin-1 assists in COX-2 degradation through the proteasome pathway. Results: Caveolin-1 enhances interactions among COX-2, Derlin-1, and the p97-Ufd1 complex and assists in COX-2 retrotranslocation and ubiquitination. Conclusion: Caveolin-1 is a cofactor facilitating COX-2 degradation via a Derlin-1-p97 pathway. Significance: Results represent a novel model for Derlin-1-mediated N-glycosylated protein degradation facilitated by caveolin-1. Caveolin-1 (Cav-1) interacts with and mediates protein trafficking and various cellular functions. Derlin-1 is a candidate for the retrotranslocation channel of endoplasmic reticulum proteins. However, little is known about how Derlin-1 mediates glycosylated protein degradation. Here, we identified Cav-1 as a key player in Derlin-1- and p97-mediated cyclooxygenase 2 (COX-2) ubiquitination and degradation. Derlin-1 augmented the interaction of Cav-1 and COX-2 and mediated the degradation of COX-2 in a COX-2 C terminus-dependent manner. Suppression of Cav-1 decreased the ubiquitination of COX-2, and mutation of Asn-594 to Ala to disrupt N-glycosylation at the C terminus of COX-2 reduced the interaction of COX-2 with Cav-1 but not Derlin-1. Moreover, suppression of p97 increased the ubiquitination of COX-2 and up-regulated COX-2 but not COX-1. Cav-1 enhanced the interaction of p97 with Ufd1 and Derlin-1 and collaborated with p97 to interact with COX-2. Cav-1 may be a cofactor in the interaction of Derlin-1 and N-glycosylated COX-2 and may facilitate Derlin-1- and p97 complex-mediated COX-2 ubiquitination, retrotranslocation, and degradation.

facilitate cyclooxygenase 2 (COX-2) degradation in vivo and in vitro (25). These studies suggest that Cav-1 plays important roles in various cellular and physiological functions.
COX converts arachidonic acid to prostaglandin G 2 (PGG 2 ) and PGG 2 to PGH 2 . PGH 2 is converted to various prostanoids by a specific isomerase (26). Two COX isoforms (COX-1 and COX-2) have been identified. The COX-1 isoform is constitutively expressed in most mammalian cells, whereas COX-2 can be markedly induced by various stimuli, including proinflammatory and mitogenic factors. Prostanoids generated by COX mediate inflammation and cellular functions and are important mediators in regulating and maintaining physiological homeostasis (27). Dysregulation of COX-2 is implicated in a number of human diseases, such as cancer and cardiovascular diseases (28 -30). Moreover, expression of COX-2 is controlled by both transcription and post-translational regulation (31). Although proteasome-mediated COX-2 degradation through ERAD has been suggested for some time (25,32), the cellular machinery remains mostly unknown.
In this study we aimed to elucidate how Cav-1 interplays with the components of the ERAD complex to mediate COX-2 degradation. Our results suggest that Cav-1 interacts with Derlin-1 and COX-2 to facilitate COX-2 ubiquitination. Moreover, Cav-1 enhances interactions among Derlin-1, p97, and Ufd1 to assist in COX-2 extraction and degradation. Cav-1 may be a cofactor in the ERAD pathway promoting COX-2 degradation.
Transfection involved the Lipofectamine 2000 method (Invitrogen). The amount of pCOX-2 for transfection was adjusted to ensure similar COX-2 levels in input samples for immunoprecipitation assay. To ensure an equal amount of protein expression, for co-transfection of a plasmid with siRNA the plasmid was transfected for 6 h, then cells were trypsinized and subcultured in smaller dishes overnight, followed by siRNA transfection for another 24 h.
Sucrose Gradient Fractionation and Western Blot Analysis-H1299 cells were lysed in radioimmuneprecipitation assay buffer, cell lysates underwent centrifugation by linear sucrose gradients (5-40% sucrose), and fractions were collected as described (34). Cell lysates or targeted proteins were resolved in SDS-polyacrylamide gels and detected by Western blot analysis as described (35). GADPH or ␤-actin was an internal control. The protein levels were quantified by densitometry (Gel Pro v.3.1; Media Cybernetics).
Immunofluorescence and Confocal Microscopy-Cells were fixed in 4% paraformaldehyde in PBS for 15 min and permeabilized with methanol for 15 min at room temperature, then blocked in PBS with 10% FBS for 30 min. Cells were doublestained by incubation with anti-COX-2 antibody along with anti-p97 or anti-Derlin-1 antibody for 1 h, washed three times, then incubated with FITC-conjugated donkey-anti-rabbit IgG and rhodamine-labeled donkey anti-mouse IgG for 1 h. After a wash, cells were examined by confocal microscopy.
Immunoprecipitation Assay-Cells were washed in lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM NaF, 2 mM NaVO 4 , 1 mM EDTA, 1% Triton X-100, and protease inhibitor mixture (Roche Applied Science)) and sonicated. Cell lysates were centrifuged at 13,000 rpm for 15 min at 4°C. The supernatant was incubated with antibodies for 2 h, then 50% AϩG agarose beads were added for incubation for 1 h at 4°C. The precipitates were washed three times with lysis buffer for Western blot analysis.

Recombinant Proteins and in Vitro Protein Interaction-
Mouse p97 cDNA was subcloned into pRSET to generate pHis-p97. Cav-1 cDNA was cloned into pGEX-5X-1 to generate pGST-Cav-1. His-p97 and GST-Cav-1 were overexpressed in Escherichia coli BL21 and then purified by use of nickel-agarose (Invitrogen) and GST beads (Pharmacia), respectively. For in vitro protein interaction assay, GST and GST-Cav-1 were incubated with GST beads in immunoprecipitation buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 10% glycerol, 0.1% Nonidet P-40, 1 mM EDTA, 5 mM MgCl, and protease inhibitor mixture) for 1 h at 4°C, then His-p97 or COX-2 was added for 2 h. After centrifugation at 1200 rpm for 2 min, bound proteins were washed with immunoprecipitation buffer three times and eluted for Western blot analysis.
In Vivo Ubiquitination Assay-HEK293 or H1299 cells were transfected to overexpress COX-2 and His-tagged ubiquitin plus Derlin-1 or p97 siRNA for 24 h and then treated with MG-132 (20 M) for another 8 h. The ubiquitinated proteins were bound by nickel-agarose beads for protein ubiquitination assay as described (25).
Statistical Analysis-One-way analysis of variance followed by a post hoc (Bonferroni) test was used to account for multiple testing. Differences between two groups were tested by Student's t test. Statistical significance was set at p Ͻ 0.05.
To examine the role of Asn-594 in Derlin-1-and Cav-1-mediated degradation, we constructed a COX-2-N594A mutant. In agreement with previous results, the COX-2 level was higher with the N594A mutant than wild-type COX-2 (Fig. 2, D-F).
Overexpression of HA-Derlin-1 down-regulated the wild-type COX-2 but not the COX-2-N594A level (Fig. 2D). siRNA suppression of Derlin-1 or Cav-1 also up-regulated COX-2 but had no effect on the COX-2-N594A level compared with controls (Fig. 2, E and F). Moreover, the N594A mutation markedly prolonged the COX-2 half-life from ϳ1.3 h to longer than 6 h (Fig.  2G). Therefore, Asn-594 may play a key role in COX-2 degradation. In addition, the N594A mutation attenuated the interaction of COX-2 with HA-Cav-1, as evidenced by less COX-2-N594A than wild-type COX-2 co-precipitated with HA-Cav-1 in H1299 cells (Fig. 2H). However, the N594A mutation did not reduce the interaction of Derlin-1 with COX-2-N594A (Fig. 2I). Therefore, the interaction of Cav-1 but not Derlin-1 with the C terminus of COX-2 depends on Asn-594. N-Glycosylation at Asn-594 may play a key role in the interaction of Cav-1 but not Derlin-1 with COX-2.
Interplay of Derlin-1 and Cav-1 in COX-2 Retrotranslocation and Ubiquitination-Contrary to the ER lumen localization of COX-2, Cav-1 is located in the outer membrane of the ER (14). Because Derlin-1 is a candidate for protein transmembrane translocation from the ER lumen to cytosol, we investigated the interplay of Derlin-1 and Cav-1 in interacting with COX-2 by immunoprecipitation assay. siRNA suppression of Derlin-1 reduced the interaction of Cav-1 and COX-2 (Fig. 3A); however, siRNA suppression of Cav-1 did not alter the interaction of Derlin-1 with COX-2 (Fig. 3B). Therefore, Derlin-1 enhances the interaction of Cav-1 and COX-2, but Cav-1 does not affect the interaction of Derlin-1 and COX-2. We further investigated the roles of Cav-1 and Derlin-1 in COX-2 ubiquitination. Overexpression of Derlin-1 increased the ubiquitination of COX-2-FLAG (Fig. 3C); however, overexpression of Derlin-1 did not alter the ubiquitination of COX-2 ⌬32aa -FLAG (Fig. 3D), which suggests that the C terminus of COX-2 is involved in Derlin-1mediated retrotranslocation. Also, in line with our previous results (25), siRNA suppression of Cav-1 decreased the ubiquitination of COX-2 (Fig. 3E). These results suggest a sequential step of Derlin-1 and Cav-1 in COX-2 degradation, whereby Derlin-1 mediates the retrotranslocation of COX-2 from the ER to the cytosol to interact with Cav-1 for efficient ubiquitination.
We then examined the role of the C terminus of COX-2 in p97-mediated degradation. Transfection of p97 shRNA up-regulated both COX-2-myc and COX-2 ⌬32aa -myc levels but the latter to a lesser extent (Fig. 4H, lane 1 versus 2 and lane 3 versus  4, respectively). In addition, we constructed a COX-1 insertion mutant with the 32 aa inserted into the C terminus of myctagged COX-1 (COX-1 ϩ32aa -myc). In H1299 cells, COX-1-myc levels were similar in cells transfected with control or p97 shRNA (Fig. 4H, lane 5 versus 6). The COX-1 ϩ32aa -myc level was lower than the COX-1-myc level with control shRNA (Fig.  4H, lane 5 versus 7) but was higher with p97 than control shRNA (Fig. 4H, lane 7 versus 8). Therefore, the p97-Ufd1 complex mediates COX-2 but not COX-1 degradation, and the C terminus of COX-2 plays an important role in p97-mediated degradation.
We next examined whether Cav-1 promoted the interaction among p97, Derlin-1, and COX-2. Co-precipitation of p97 with HA-Derlin-1 and Derlin-1 with HA-p97 was lower in H1299 cells with Cav-1 siRNA knockdown than the control (Fig. 5, D  and E), which suggests that Cav-1 enhances the interaction of p97 and Derlin-1. In addition, co-precipitation of COX-2 with HA-p97 and with HA-Cav-1 was lower in H1299 cells with Cav-1 and p97 siRNA knockdown, respectively, than the control (Fig. 6, A and B). Moreover, the GST pulldown assay revealed a direct interaction of GST-Cav-1 with COX-2, and the addition of His-p97 augmented this interaction (Fig. 6C). Thus, Cav-1 and p97 collaborated with each other to interact with and mediate COX-2 degradation.

DISCUSSION
In this study, we demonstrated a novel role of Cav-1 in interacting with Derlin-1, VIMP, p97, and Ufd1 to facilitate degradation of COX-2, a glycosylated protein. Our results suggest that COX-2 is targeted to Derlin-1 for initial retrotranslocation and, with the help of Cav-1, for ubiquitination. Although a number of studies show important functions of Cav-1 in caveolae near the plasma membrane, our results demonstrate a critical role of Cav-1 in the ER in interacting and enhancing the interaction among COX-2, Derlin-1, p97, and Ufd1 to facilitate COX-2 degradation.
COX-2 is a membrane-associated protein with a hydrophobic domain that interacts with the membrane and is located in the luminal leaflet of the ER membrane (36). Degradation of COX-2 may represent a specific category of ER proteins besides soluble and integral membrane proteins. Our results reveal a critical role of Derlin-1 in mediating COX-2 retrotranslocation and a sequential step of COX-2 degradation in that Derlin-1 mediates the initial translocation of COX-2 from the ER to the cytosol to interact with Cav-1. Cav-1 may interact with both Derlin-1 and COX-2 to help in Derlin-1-mediated retrotranslocation, thereby enhancing COX-2 ubiquitination.
N-Glycosylation of the Asn-594 residue in the C-terminal region of COX-2 has been reported to mediate COX-2 degradation (32,37). Here, we further demonstrated that the C terminus of COX-2 plays a critical role in Derlin-1-, p97-, and Cav-1-mediated COX-2 degradation. Studies suggest that Derlin-1 interacts with and mediates nonglycosylated protein retrotranslocation (11), and knockdown of Derlin-1 stabilized the nonglycosylated mutant but not the wild type of human breast cancer-resistant protein (13). However, our results showed that knockdown of Derlin-1 increased the level of wild-type COX-2 but not that of the nonglycosylated N594A COX-2 mutant compared with the control (Fig. 2E). Notably, although deletion of the C terminus of COX-2 reduced its interaction with Derlin-1, the N594A mutation failed to suppress the interaction (Fig. 2I). Derlin-1 interacting with the C terminus of COX-2 did not depend on glycosylation. Conversely, Cav-1 interaction  NOVEMBER 15, 2013 • VOLUME 288 • NUMBER 46 with the C terminus of COX-2 did depend on glycosylation because the COX-2 N594A mutation suppressed the interaction (Fig. 2H). These results support a novel model of Derlin-1mediated glycosylated protein degradation in that after the initial translocation of COX-2 by Derlin-1, Cav-1 may interact with the N-glycosylated C terminus of COX-2 to assist with Derlin-1-mediated translocation and ubiquitination and then retrotranslocation and degradation.

Caveolin-1 Assists Derlin-1/p97-mediated COX-2 Degradation
Cav-1 has been suggested to interact with specific motifs containing aromatic residues such as ⌽X⌽XXXX⌽ or ⌽XXXX⌽XX⌽, where ⌽ is an aromatic residue and X is any amino acid (39). However, the C terminus of COX-2 contains no such motif. Cav-1 may interact with the specific secondary structure of the COX-2 C terminus containing the N-glycosylated residues (32). Of note, despite reduced interaction, the antibody of Cav-1 co-precipitated with COX-2 with the C-terminal 32 aa deleted (25) or with the N594A mutation (Fig. 2H), so Cav-1 may also interact with other regions of COX-2.
Although COX-1 and COX-2 are highly similar in protein sequence and structure (40), our results demonstrate that p97 mediates the degradation of COX-2 but not COX-1 (Fig. 4, B and  C). Moreover, besides Derlin-1 and Cav-1, the C terminus of COX-2 plays an important role in p97-mediated COX-2 degradation, as evidenced by insertion of this region in the C terminus of COX-1 accelerating p97-mediated COX-1 degradation (Fig. 4H). Nevertheless, siRNA suppression of p97 augmented the COX-2 ⌬32aa -myc level compared with the control (Fig. 4H), but up-regulation or suppression of Derlin-1 did not alter COX-2 ⌬32aa -FLAG level (Fig. 2, A and B). Therefore, although the C terminus of COX-2 plays a key role in Derlin-1-mediated retrotranslocation, p97 may also mediate the degradation of COX-2 ⌬32aa but with less efficiency than degradation of wild-type COX-2.
Interaction of p97 with Cav-1 has been reported in that p97 binds with monoubiquitinated Cav-1 in endosomes and mediates its transport to endolysosomes (41,42). Our results support the direct interaction of Cav-1 with p97 and reveal that this interaction facilitates p97-mediated COX-2 protein degradation. Our results suggest a critical role of Cav-1 in p97-mediated COX-2 degradation in that Cav-1 enhances the interaction of p97 with Ufd1 (Fig. 5C) and Derlin-1 (Fig. 5, D and E) and Cav-1 and p97 collaborate with each other to interact with COX-2 ( Fig. 6, A-C). These results reveal an important role of Cav-1: besides interacting with N-glycosylated COX-2, Cav-1 may be a cofactor to enhance and stabilize the interaction among COX-2, Derlin-1, and the p97-Ufd1 complex to facilitate ERAD-mediated COX-2 ubiquitination and degradation.
In conclusion, we have demonstrated a novel role of Cav-1 in mediating COX-2 degradation through ERAD (Fig. 6D). Derlin-1 interacts with the C terminus of COX-2 and mediates its initial translocation. Cav-1 interacts with Derlin-1 and N-glycosylated COX-2 to facilitate COX-2 ubiquitination and enhances the interaction of p97 with Derlin-1 and COX-2. Moreover, Cav-1 enhances the interaction of p97 with Ufd1 and collaborates with p97 to interact with COX-2, which may help with COX-2 retrotranslocation and degradation. Cav-1 may be a cofactor to enhance the interaction among ERAD components and COX-2 to facilitate COX-2 degradation.