Identification of SVIP as an Endogenous Inhibitor of Endoplasmic Reticulum-associated Degradation*

Misfolded proteins in the endoplasmic reticulum (ER) are eliminated by a process known as ER-associated degradation (ERAD), which starts with misfolded protein recognition, followed by ubiquitination, retrotranslocation to the cytosol, deglycosylation, and targeting to the proteasome for degradation. Actions of multisubunit protein machineries in the ER membrane integrate these steps. We hypothesized that regulation of the multisubunit machinery assembly is a mechanism by which ERAD activity is regulated. To test this hypothesis, we investigated the potential regulatory role of the small p97/VCP-interacting protein (SVIP) on the formation of the ERAD machinery that includes ubiquitin ligase gp78, AAA ATPase p97/VCP, and the putative channel Derlin1. We found that SVIP is anchored to microsomal membrane via myristoylation and co-fractionated with gp78, Derlin1, p97/VCP, and calnexin to the ER. Like gp78, SVIP also physically interacts with p97/VCP and Derlin1. Overexpression of SVIP blocks unassembled CD3δ from association with gp78 and p97/VCP, which is accompanied by decreases in CD3δ ubiquitination and degradation. Silencing SVIP expression markedly enhances the formation of gp78-p97/VCP-Derlin1 complex, which correlates with increased degradation of CD3δ and misfolded Z variant of α-1-antitrypsin, established substrates of gp78. These results suggest that SVIP is an endogenous inhibitor of ERAD that acts through regulating the assembly of the gp78-p97/VCP-Derlin1 complex.

We have recently reported that gp78 directly interacts with p97/VCP through a newly identified p97/VCP-interacting motif (VIM) (25). This interaction is essential for degradation of ERAD substrates CD3␦ and the Z variant of ␣-1-antitrypsin (ATZ). The putative retrotranslocation channel Derlin1 has been shown to be part of the gp78 and p97/VCP complex (12,26). Interestingly, a highly conserved VIM is also found in the small p97/VCP-interacting protein (SVIP) (25). SVIP was isolated by a yeast two-hybrid screen using p97/VCP as bait (27). Overexpression of SVIP causes vacuolization of cells, but its physiological role is unknown (27). The sharing of VIM between gp78 and SVIP prompted us to evaluate a possible regulatory role of SVIP on gp78-mediated ERAD. We found that SVIP forms a complex with Derlin1 and p97/VCP. We provided evidence that SVIP regulates the formation of gp78-p97/VCP-Derlin1 complex, which leads to a change in substrate association with gp78 and p97/VCP and, eventually, changes in the efficacy of ERAD. These findings suggest that SVIP is an endogenous inhibitor of ERAD, acting through the inhibition of the assembly of the gp78-p97/VCP-Derlin1 complex. Thus, SVIP may act to prevent excessive ERAD that may cause damage to cells.

Stable Cell Lines
293 cells that stably express HA-CD3␦ alone (clone-5) and HA-CD3␦ together with gp78 (clone-20) have been previously reported (6). To generate cell lines that stably express ATZ, plasmid encoding hATZ was co-transfected with pBABE vector that confers puromycin resistance. 24 h after the transfection, 2.5 g/ml puromycin was added to eliminate non-transfected cells. Positive clones were characterized by immunoblotting.

Immunoblotting (IB) and Immunoprecipitation (IP)
293 cells were seeded at 2.5 ϫ 10 5 /well in 6-well plates or 1.2 ϫ 10 6 /100-mm dish prior to the day of transfection with Lipofectamine 2000 (Invitrogen) (for plasmids and siRNAs) or by calcium phosphate precipitation (for plasmids only). Cells transfected with plasmids were collected 24 h after transfection, whereas those transfected with siRNA were harvested 72 h post-transfection. IB and IP were performed as we previously described (24).

Subcellular Fractionation
Alkaline Extraction-Microsomes were isolated as described (25) and then incubated with 0.1 M Na 2 CO 3 , pH 11, for 20 min at room temperature. After the incubation, microsomes were pelleted by centrifugation at 105,000 ϫ g for 10 min at 4°C. The resulting microsomes and supernatants were processed for IB.
N-Myristoylation Inhibition-1 mM 2-OHM (2-hydroxymyristic acid) (Sigma) was delivered to 293 cells in a complex with fatty acid-free bovine serum albumin (Sigma) as previously described (30), and cells were incubated for 24 h. Because unmyristoylated cytosolic SVIP is not very stable, cells were treated with proteasome inhibitor for 5 h before being processed for fractionation and IB.
Gradient Fractionation-293 cells were homogenized in buffer B (0.25 M sucrose, 1 mM EDTA, 10 mM HEPES-NaOH, pH 7.4) and then centrifuged at 3000 ϫ g for 10 min to remove nuclei and unbroken cells. The post-nuclear supernatant was layered on top of a preformed 0 -25% iodixanol gradient in buffer B and centrifuged at 200,000 ϫ g for 2.5 h, after which fractions were collected from the bottom of the tube. Equal volume of each fraction was processed for IB.

RESULTS
SVIP Is a Membrane-anchored Protein-SVIP has been reported to be a membrane-associated protein, but it possesses neither a transmembrane domain nor a signal peptide (27). How it associates with the membrane is not clear. To address this issue, we carried out alkaline extraction of the microsomes isolated from 293 cells. We found that Na 2 CO 3 has no effect on SVIP-membrane association, as seen on the membrane-spanning protein Derlin1 (Fig. 1A). As a positive control, a significant amount of p97/VCP that is known to be peripherally associated with the microsomes is released into the cytosol (Fig. 1A, lane 2 versus 4). This result is consistent with the previous report that SVIP may be attached to membrane through myristoylation (27). To test this possibility, we mutated the putative myristoylation site glycine-2 to alanine and generated a mutant SVIP (SVIP(G2A)) that is defective in myristoylation. We found that SVIP(G2A) is exclusively localized in the cytosol (Fig. 1B, lanes 3, 4 versus 5, 6). The difference is not due to overexpression, since SVIP(G2A) was still in the cytosol when its expression was markedly reduced (lanes 7, 8 versus 9, 10). Furthermore, the membrane anchorage of SVIP can be abolished by the N-myristoylation inhibitor 2-hydroxymyristic acid (2-OHM). Treatment with 2-OHM slightly increased the mobility of SVIP and resulted in its cytosolic localization (Fig. 1C), which is con-sistent with inhibition of SVIP myristoylation. Thus, it is likely that SVIP is membrane-anchored through myristoylation on glycine 2. In agreement, we demonstrated that SVIP is attached on the cytosolic surface of microsomes, since proteinase K completely digested SVIP, while the intraluminal domain of calnexin remained protected by the microsomal membrane (Fig. 1D).
SVIP Forms a Complex with p97/VCP and Derlin1-SVIP is localized to the ER and other membranes (27). To investigate whether the two known VIM-containing proteins, SVIP and gp78, are in the same membrane fraction, post-nuclear supernatants from 293 cells were fractionated on an iodixanol (Optiprep) gradient. Indeed, SVIP cofractionated with gp78 along with other ER membrane proteins such as Derlin1, Hrd1, and calnexin ( Fig. 2A). As expected, a fraction of p97/VCP cofractionated with SVIP and gp78 ( Fig. 2A).
SVIP Regulates gp78-mediated ERAD-Overexpression of SVIP induces cellular vacuolation (27), and electron microscopy revealed that the vacuoles represent dilated ER (27). We speculated that such vacuoles might be caused by accumulation of misfolded ER proteins and that SVIP might be an inhibitor of ERAD. To test this possibility, we determined the effects of SVIP on ERAD. Increasing amounts of SVIP were expressed in 293 cells that stably express CD3␦, a well known ERAD substrate (31). We found that SVIP causes a dose-dependent accumulation of CD3␦ (Fig. 3A). The accumulation was due to decreased degradation as shown by cycloheximide (CHX) chase analysis in 293 cells with approximately three times overexpression of SVIP (31) (Fig. 3B). When a similar exper-FIGURE 1. SVIP is anchored to microsomal membrane, probably through myristoylation. A, alkaline extraction does not affect SVIP-microsome association. 293 cells were homogenized and the microsomes (m) were isolated from the post-nuclear homogenate as we previously described (25). For alkaline extraction, the microsomes were incubated in 0.1 M Na2CO3, pH 11, or in 1ϫ phosphate-buffered saline as a control. After the incubation, the microsomes were pelleted, and the supernatants (s) were collected. These fractions were analyzed by IB using anti-SVIP, anti-Derlin1, and anti-p97/VCP antibodies. p97/VCP and Derlin1 were used as controls for peripheral protein and ER membrane-anchored protein, respectively. B, wild-type (Wt)-SVIP, but not myristoylation-deficient mutant SVIP(G2A), localizes to the microsomes. 293 cells transfected either with pCIneo, Wt-SVIP, or SVIP(G2A) were processed for fractionation into microsomes (m) and cytosol (c). The localization of Wt-SVIP and SVIP-G2A was determined by blotting with anti-His 6 antibody to detect their His 6 tags. C, 2-hydroxymyristic acid (2-OHM) inhibits endogenous SVIP anchorage to membrane. 293 cells were treated with 2-OHM or its vehicle (ethanol: EtOH) followed by fractionation into microsomes (m) and cytosol (c). D, proteinase K digestion assay used to determine the membrane topology of SVIP. Microsomes prepared from 293 cells were incubated with increasing amounts of proteinase K (0 -100 g/ml) for 30 min before processing for IB with anti-SVIP and anti-calnexin (Cnx) antibodies.
Next, we asked whether the endogenous SVIP plays an inhibitory role in gp78-mediated ERAD. We again utilized 293 cells that stably express CD3␦. RNA interference was employed to silence endogenous SVIP expression. Degradation of CD3␦ was assessed by CHX chase analysis. Inhibition of SVIP expression significantly increased CD3␦ degradation (Fig. 4A). A similar experiment was performed on ATZ, a newly identified luminal substrate for gp78 (32). As predicted, silencing SVIP also decreased the intracellular levels of ATZ, which correlates with reduced ATZ secretion (Fig.  4B). These data further support the hypothesis that endogenous SVIP negatively regulates the function of gp78s. To determine whether endogenous SVIP inhibits gp78 interaction with p97/VCP and Derlin1, we silenced the expression of SVIP and evaluated changes in the formation of gp78-p97/ VCP-Derlin1 complex by co-immunoprecipitation. Inhibition of SVIP expression markedly augmented the association of p97/VCP and Derlin1 with gp78, consistent with the fact that silencing SVIP enhances ERAD. To further substantiate gp78 as a target inhibited by SVIP, we studied the effects of SVIP silencing on CD3␦ degradation in 293 cells that overexpress both gp78 and CD3␦. We predicted that overexpression of gp78 will overcome the inhibitory effect exerted by endogenous SVIP and that silencing SVIP would have no effect on CD3␦ degradation in cells overexpressing gp78. This was indeed the case (Fig. 4D). Collectively, these results suggest that SVIP is an endogenous inhibitor for gp78-mediated ERAD that acts through its common motif shared with gp78 and uncouples gp78 from its substrates, p97/VCP and Derlin1. Changes in the levels of SVIP and gp78 proteins appear to control the efficacy of ERAD (Fig. 3, A and B, and  Fig. 4, A, B, and D). The next question is whether the expressions of SVIP and gp78 proteins are modulated under ER stress, a condition known to regulate the efficacy of ERAD. Interestingly, tunicamycin-induced ER stress inversely regulated the levels of SVIP and gp78 proteins in a time-dependent manner, which correlates well with previous reports that ER stress enhances, but prolonged ER stress (17 h) inhibits, ERAD (33,34).

DISCUSSION
The central components of the ERAD pathways include the ER resident E3s and the cytosolic CDC48/p97/VCP ATPase (13,14). Multiple ER membrane-anchored proteins interact with and link CDC48/p97/VCP to ERAD E3s, thereby promoting association of ubiquitinated ER proteins with the ATPase for retrotranslocation (12,20). In this study, we found that myristoylation targets SVIP to the ER membrane where it competes with gp78 to bind to p97/VCP and Derlin1. As a result, SVIP reduces the association of ERAD substrates with gp78 and p97/VCP, thereby inhibiting sub-  , n ϭ 3). B, silencing SVIP expression reduces the levels of intracellular ATZ accompanied by decreased ATZ secretion. 293 cells stably expressing ATZ were transfected with control (C) or SVIP (S) siRNA. Three days after transfection, CHX chase was performed. Intracellular level of ATZ (i) was detected by IB with horseradish peroxidaseconjugated anti-ATZ antibody. To determine the secreted ATZ (s), media were processed for IP with anti-ATZ antibody. Actin was blotted as loading control. * indicates IgG HC in lanes 4,5,9,10. C, silencing endogenous SVIP expression enhances the formation of gp78-p97/VCP-Derlin1 complex. Control (C) or SVIP (S) siRNA was transfected into 293 cells. Three days after transfection, cells were processed for IP for gp78. Co-immunoprecipitated Derlin1 and p97/VCP were detected by IB with anti-Derlin1 and anti-VCP, respectively. SVIP was blotted to determine knock down efficiency. D, gp78 overexpression diminishes the inhibitory effects of SVIP on CD3␦ degradation. 293 cells that stably express both gp78 and CD3␦ (gp78-CD3␦ clone-20 cells) (25) were transfected as the indicated siRNA. Three days after transfection, cells were used for CHX chase analysis. Densitometric analysis of HA-CD3␦ levels is shown in the lower panel (mean Ϯ S.D., n ϭ 3). E, tunicamycin treatment inversely regulates the levels of SVIP and gp78 proteins. 293 cells were treated with tunicamycin (2.5 g/ml) for the indicated time and then processed for IB. The relative expression levels of SVIP and gp78 proteins were quantified by densitometry and expressed in the graph. strate ubiquitination and subsequent ERAD steps. This study identifies SVIP as the first endogenous inhibitor of ERAD that uses a novel mechanism through inhibiting the assembly of the gp78-p97/VCP-Derlin1 complex.
Previous studies have focused on the mechanisms of ERAD and how ERAD is enhanced under ER stress (1,2,11,35). However, the control mechanism of ERAD activity is largely unknown. As a general mechanism, activated cellular events have to be turned off after having fulfilled their tasks, such as in the cases of the activated receptor tyrosine kinases and transcription factors (36,37). Failure to control the durations of receptor tyrosine kinase signaling and gene transcription can cause devastating diseases. Therefore, it is conceivable that activated ERAD in response to accumulation of misfolded proteins in the ER has to be controlled once the misfolded proteins have been removed. A similar mechanism by which ERAD is controlled may be through degradation of key components of the ERAD machineries. For example, gp78 itself is degraded by ERAD (15). Importantly, gp78 degradation is inhibited under ER stress, which correlates with an increase in ERAD (24). Thus, when misfolded proteins are accumulated in the ER, gp78 is stabilized and it aids to remove misfolded proteins. Once the accumulated proteins in the ER have been eliminated, gp78 is degraded, thereby preventing excessive ERAD that may cause damage to cells. The present finding on the regulation of the assembly of ERAD machinery by SVIP represents another mechanism of control for ERAD activity.
Although p97/VCP binds gp78 and SVIP in a mutually exclusive manner, both p97/VCP-gp78 and p97/VCP-SVIP can associate with Derlin1. This type of interaction is reminiscent of the Hrd1-p97/VCP-Derlin1 complex, in which p97/VCP simultaneously binds Derlin1 and Hrd1 (12). Our study also revealed an interaction between the transmembrane domains of gp78 and Derlin1, supporting the possibility that gp78 and Derlin1 may be part of the protein retrotranslocation channel as previously suggested (12,24). Derlin1 has two homologues, Derlin2 and Derlin3. Derlin2 also forms complex with p97/VCP and Hrd1 (20). It would be important to know whether SVIP also interacts with Derlin2 and Derlin3. If it does, SVIP may play a more general inhibitory role on ERAD. CDC48/p97/VCP ATPase is the converging point of probably all the characterized ERAD pathways. Inhibition of this ATPase is expected to disrupt the ERAD process. The question is why SVIP also uncouples Derlin1 from gp78. Derlin1 has been proposed as the long sought after channel for retrotranslocation (12). Thus, simultaneous sequestration of p97/VCP and Derlin1 by SVIP would inhibit the targeting of misfolded proteins to the retrotranslocation channel and subsequent retrotranslocation. Additionally, SVIP diminished the gp78-CD3␦ interaction, which results in stabilization of non-ubiquitinated CD3␦. By doing so, SVIP may enhance protein trafficking through the ER/Golgi. In support of this possibility, we found that silencing SVIP expression increases ATZ degradation and decreases ATZ secretion (Fig. 4B).
Our data presented strongly suggest that p97/VCP and Derlin1 can form a complex with either gp78 or SVIP. The gp78 complex facilitates ERAD, whereas the SVIP complex inhibits ERAD; the stoichiometry between these two complexes dictates the efficacy of gp78-mediated ERAD (Fig. 5). Importantly, the stoichiometry is apparently regulated under ER stress (Fig. 4E).