The Role of a Novel p97/Valosin-containing Protein-interacting Motif of gp78 in Endoplasmic Reticulum-associated Degradation*

Improperly folded proteins in the endoplasmic reticulum (ER) are eliminated via ER-associated degradation, a process that dislocates misfolded proteins from the ER membrane into the cytosol, where they undergo proteasomal degradation. Dislocation requires a subclass of ubiquitin ligases that includes gp78 in addition to the AAA ATPase p97/VCP and its cofactor, the Ufd1-Npl4 dimer. We have previously reported that gp78 interacts directly with p97/VCP. Here, we identify a novel p97/VCP-interacting motif (VIM) within gp78 that mediates this interaction. We demonstrate that the VIM of gp78 recruits p97/VCP to the ER, but has no effect on Ufd1 localization. We also show that gp78 VIM interacts with the ND1 domain of p97/VCP that was shown previously to be the binding site for Ufd1. To evaluate the role of Ufd1 in gp78-p97/VCP-mediated degradation of CD3δ, a known substrate of gp78, RNA interference was used to silence the expression of Ufd1 and p97/VCP. Inhibition of p97/VCP, but not Ufd1, stabilized CD3δ in cells that overexpress gp78. However, both p97/VCP and Ufd1 appear to be required for CD3δ degradation in cells expressing physiological levels of gp78. These results raise the possibility that Ufd1 and gp78 may bind p97/VCP in a mutually exclusive manner and suggest that gp78 might act in a Ufd1-independent degradation pathway for misfolded ER proteins, which operates in parallel with the previously established p97/VCP-Ufd1-Npl4-mediated mechanism.

The endoplasmic reticulum (ER) 3 is the first compartment in the secretory pathway (1,2). Proteins destined for the secretory pathway are translocated into the lumen or inserted into the membrane of the ER where they are properly folded and modified before being delivered to their functional destinations (1,2). However, protein folding in the secretory pathway is an imperfect process that generates misfolded products (3,4). These misfolded proteins are retained in the ER by a qualitycontrol mechanism and eliminated through a process known as ER-associated degradation (ERAD) (3)(4)(5)(6). ERAD requires retrotranslocation, in which misfolded proteins are dislocated from the ER membrane into the cytosol where they are degraded by the proteasomes (7). Genetic evidence from yeast indicates that the ATPase CDC48 in combination with its cofactors, the Ufd1-Npl4 dimer, are essential for retrotranslocation (8). The mammalian homolog of CDC48, p97/VCP, in concert with Ufd1-Npl4 dimer, appears to play a similar role in mammalian cells (9,10). p97/VCP, Ufd1, and Npl4 are cytosolic proteins. They must be recruited to the cytosolic surface of the ER to carry out the retrotranslocation process. In yeast, an ER membrane-anchored protein, Ubx2, plays a central role in coupling ubiquitination with retrotranslocation. Ubx2 recruits CDC48 along with the Ufd1-Npl4 dimer using its Ubx domain (11,12), thereby linking the CDC48-Ufd1-Npl4 complex to ER-anchored ERAD E3s, that include Hrd1 and Doa10, as well as ERAD substrates (see a diagrammatic illustration of this complex in Fig. 1a) (11,12). Recent studies suggest that more complicated mechanisms exist for coupling ubiquitination with retrotranslocation in higher eukaryotes (Fig. 1b) (13)(14)(15). In mammalian systems, p97/VCP interacts with a protein complex that contains at least four ER membrane-anchored proteins, the p97/VCP-interacting membrane protein (VIMP), Derlin-1, Derlin-2, and ERAD E3 Hrd1. The Ufd1-Npl4 dimer is also recruited along with p97/VCP by the four-protein complex (Fig. 1b) (16). This is consistent with previous reports that the p97/VCP-Ufd1-Npl4 complex is functionally required for ERAD in mammalian cells (9,10). gp78, a multimembrane-spanning ERAD E3, also interacts with p97/VCP (13,(17)(18)(19). Functionally, the gp78-p97/VCP interaction enhances the binding of p97/VCP to polyubiquitinated proteins, as well as degradation of ERAD substrates, including CD3␦, apolipoprotein B100, and 3-hydroxy-3-methylglutaryl-CoA reductase (18 -20). However, the role of p97/VCP cofactors, Ufd1 and Npl4, in gp78-mediated ERAD is yet to be determined.
In this study, we have identified a novel p97/VCP-interacting motif (VIM) in the cytosolic tail of gp78 and in the N-terminal region of the small p97/VCP-interacting protein (SVIP) (21). The VIM of gp78 interacts with the ND1 domain of p97/VCP, the reported binding site for SVIP, Ufd1, and p47, another p97/ VCP cofactor required for Golgi and ER membrane fusion (22,23). SVIP, Ufd1, and p47 are known to bind p97/VCP in a mutually exclusive manner. In the current study, our data suggest that gp78 and Ufd1 also bind p97/VCP in a mutually exclusive fashion. Functionally, gp78-mediated ERAD apparently requires p97/VCP but not Ufd1. We have also confirmed the presence of an Ufd1-dependent ERAD pathway. Collectively, these data suggest that gp78 may mediate an Ufd1-independent degradation pathway for misfolded ER proteins. This pathway appears to operate in parallel with the previously established p97/VCP-Ufd1-Npl4-mediated mechanism.
Fractionation, Immunoprecipitation, and Immunoblotting-For subcellular fractionation, 293 cells were transfected as indicated in the figure legends. Cell fractionation was carried out by homogenization in fractionation buffer containing 50 mM Tris-HCl (pH 8), 1 mM 2-mercaptoethanol, 1 mM EDTA, 10 mM triethanolamine, and 0.32 M sucrose via passage through a 27-gauge syringe 20 times. Homogenates were centrifuged at 1000 ϫ g to sediment unbroken cells, cell debris, and nuclei (pellet fraction (p)). The remaining supernatant was separated by centrifugation at 105,000 ϫ g for 60 min. The resulting pellet contained the microsomal fraction (m) and the supernatant contained the cytosolic fraction (c). All three fractions (m, c, and p) were further processed for immunoblotting. Immunoprecipitation and immunoblotting were carried out following previously published protocols (17).
Fluorescence Microscopy-U2OS cells were cultured on coverslips in 6-well dishes and transfected with the indicated plasmids using Lipofectamine 2000. For double immunofluorescent staining, gp78 and endogenous p97/VCP were labeled using rabbit anti-gp78 and mouse anti-p97/VCP antibodies in conjunction with Alexa Fluor-594 and Alexa Fluor-488. A similar procedure was used to label gp78 and Ufd1, as well as Hrd1/ Hrd1 plus VIM and p97/VCP. Staining was performed as described previously (17). Fluorescent microscopy was performed using a Zeiss Axiovert 200M fluorescent microscope.
Establishment of 293 Cell Lines That Stably Express HA-CD3␦ or HA-CD3␦ and gp78-293 cells were transfected with plasmids encoding HA-CD3␦ or HA-CD3␦ and gp78. pBABE vector conferring puromycin-resistance was co-trans-FIGURE 1. Current models of protein complexes that links ubiquitination to retrotranslocation during ERAD. a, in yeast the ER-anchored Ubx2 (labeled with x2) recruits a CDC48-Ufd1-Npl4 complex from the cytosol to the ER, thereby linking the CDC48 complex to E3 (Hrd1 and Doa10), ubiquitinated ERAD substrate (black rectangle), and Der1. This Ubx2-assembled complex couples ubiquitination with retrotranslocation and leads to degradation of ERAD substrates. U represents ubiquitin. b, in mammals, four ER-anchored proteins, including the E3 Hrd1, Derlin-1 and -2 (behind major histocompatibility complex class I heavy chain: HC in the black rectangle), and VIMP (cylinder behind E3), all interact with p97/VCP. The interaction leads to recruitment of the p97/VCP-Ufd1-Npl4 complex from the cytosol to the ER, thereby linking p97/VCP complex to Hrd1, substrate (HC), Derlin-1 and -2, and VIMP to couple ubiquitination with retrotranslocation.
RNA Interference-pSUPER-gp78 siRNA1 was used to knockdown endogenous gp78 expression as previously reported (18). Silencer pre-designed siRNAs target p97/VCP and Ufd1, as well as serves as a negative control, were purchased from Ambion (Austin, TX). CD3␦ clone-5 and CD3␦/gp78 clone-20 cells were transfected with appropriate siRNA(s) using Lipofectamine 2000. 72 h after transfection cells were harvested and processed for further experiments.
Pulse-chase Analysis-Pulse-chase analysis was performed essentially as we have previously described (17). CD3␦ clone-5 and CD3␦/gp78 clone-20 cells with a knockdown of either p97/VCP or Ufd1 were pulse-labeled with a mixture of L-[ 35

RESULTS
Identification of a Novel p97/VIM in gp78 and SVIP-Many p97/VCPinteracting proteins have been reported (26 -28). In search of a common p97/VCP-interacting domain that might be present in different p97/VCP-interacting proteins, we identified a conserved amino acid sequence within the C-terminal portion of gp78 and the N-terminal region of SVIP, both of which have previously been found to bind p97/VCP (18,21). We named this sequence the p97/VCP-interacting motif (VIM). VIM comprises the C-terminal 30 amino acids of gp78 and is well conserved in vertebrates (examples are given in Fig. 2a). However, it is not present in the yeast Cue1p protein or Caenorhabditis elegans gp78. Consistently, Cue1p and C. elegans gp78 do not interact with p97/VCP (supplemental Fig. S1). Bioinformatic analysis predicts a helix formed by 17 amino acids (aa 625-641 in gp78 and aa 20 -35 in SVIP, underscored in Fig. 2a) within the VIM of human gp78 and SVIP (29). To ascertain whether the VIM of gp78 is sufficient to bind p97/VCP, we assessed its ability to recruit p97/VCP to microsomal membranes. 293 cells were transfected with wt gp78 or its VIM deletion mutant (gp78⌬VIM) (Fig. 2b). Transfected cells were fractionated into cytosol (c) and ER-containing microsomal membrane (m) fraction where gp78 is generally localized, which we have previously reported (17). Wild-type (wt) gp78 efficiently moves p97/VCP from the cytosol to the microsomes as determined by immunoblotting (Fig. 2, b and c, lanes 4 and 5 versus lanes 1 and 2). Deletion of VIM from gp78 (gp78⌬VIM) largely diminishes this effect (Fig. 2b, lanes 10 and 11 versus lanes 4 and 5). To investigate the role of the E3 activity of gp78 in relocalization of p97/VCP, we examined the effects of FIGURE 2. Identification of a p97/VCP-interacting motif (VIM) in gp78 and SVIP. a, alignments of the VIM of gp78 and SVIP. Hs, human; Rn, rat; Mm, mouse; aa, amino acid. Underscored is the predicted sequence that forms a helix. b, 293 cells were transfected with plasmids encoding wt gp78, or gp78⌬VIM, or gp78R2m. Empty vector was transfected as a negative control. Eighteen hours after transfection, cells were fractionated into membrane (m) and cytosol (c). The initial pellet (p) obtained from clearing the homogenates was retained to monitor total levels of proteins. Endogenous p97/VCP was revealed by immunoblotting using monoclonal antibody. p97/VCP in the microsome (m) and cytosolic (c) fractions are framed for easy comparison. Transfected gp78 was detected by immunoblotting using polyclonal anti-gp78 antibodies that do not recognize endogenous gp78 as previously reported (17). gp78 tends to form high molecular weight smears (arrow) due to oligomerization and autoubiquitination (17). *, residual signal of p97/VCP from the first blotting for p97/VCP. Calnexin was blotted as an ER marker. c, 293 cells were transfected with plasmids encoding wt gp78 or empty vector as control. Transfected cells were analyzed as in b. Endogenous p97/VCP and Ufd1 were blotted using monoclonal antibodies. NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 gp78R2m on p97/VCP localization in cytosol and microsomal fractions. gp78R2m is an E3-inactive mutant form of gp78 with two defective zinc ligand residues in the RING finger (17). The results showed that gp78R2m recruits p97/VCP from the cytosol to the microsomes (Fig. 2b, lanes 7 and 8 versus lanes 1 and 2) as efficiently as wt gp78 (Fig. 2b, lanes 4 and 5 versus lanes  7 and 8). These results strongly suggest that the VIM of gp78 interacts with and recruits p97/VCP from the cytosol to the ER and gp78 E3 activity is not required for p97/VCP relocalization.

gp78-mediated ERAD
A previous study has shown that VIMP, another ER-anchored p97/VCP-interacting membrane protein, co-localizes with p97/VCP in the ER (16). Importantly, VIMP interacts with p97/VCP along with its cofactor, the Ufd1-Npl4 dimer (16). To determine whether gp78 could recruit p97/VCP along with Ufd1, a similar fractionation experiment to that shown in Fig.  2b was performed and the effects of gp78 on localization of endogenous Ufd1 were examined by immunoblotting. According to this data, gp78 has no apparent effect on Ufd1 localization despite the fact that p97/VCP is relocalized to the microsomes (Fig. 2c, lanes 1 and 2 versus lanes 3 and 4). These results suggest that gp78 interacts only with p97/VCP and that Ufd1 is not a part of the gp78-p97/VCP complex (Fig. 2c, lanes 1 and 2  versus lanes 4 and 5).
GST-VIM Interacts with p97/VCP but Not the p97/VCP-Ufd-Npl4 Complex-The interaction of gp78 with the ND1 domain of p97/VCP suggests that gp78 and Ufd1 might form mutually exclusive complexes with p97/VCP. To test this possibility, GST-VIM pull-down assays of lysates from 293 cells transfected with FLAG-tagged Ufd1 and FLAG-His-tagged Npl4 were conducted. In these experiments, p97/VCP alone was associated with GST-VIM but not Ufd1-Npl4 or complexes between p97/VCP and Ufd1-Npl4 (Fig. 4a, lanes 4 and 12).
Overexposing the blot did not reveal any additional bands (data not shown). These results further support the notion that the VIM of gp78 mediates the formation of the gp78 and p97/VCP complex from which Ufd1 is excluded. However, a second possibility also exists: p97/VCP might not be able to form a complex with Ufd1-Npl4 dimer under the conditions we used for the pull-down assay shown in Fig. 4a. To distinguish between the two possibilities, we carried out a pull-down of Npl4 with nickel-nitrilotriacetic acid-agarose beads. 293 cells were transfected with plasmids encoding FLAG-Ufd1, FLAG-His-Npl4, or both (Fig. 4b, lanes 1-4). Lysates prepared from the transfected cells were used for a pull-down assay with nickel-nitrilo-triacetic acid-agarose beads. Analysis of the precipitates by immunoblotting indicates that, indeed, p97/VCP, Ufd1, and Npl4, appear to be capable of forming a complex (Fig. 4b, lane 8) under conditions used for the GST-pull-down assay (Fig. 4a). Therefore, it is plausible that GST-VIM forms a complex with p97/VCP and excludes Ufd1-Npl4 from the complex.
Next, we sought to determine whether GST-VIM could directly interact with p97/VCP. Purified recombinant p97/ VCP, Ufd1, and Npl4 from bacteria were co-incubated with GST-VIM immobilized glutathione-agarose beads. The results show that GST-VIM selectively binds p97/VCP despite the presence of Ufd1 and Npl4 (Fig. 4c, lane 6). Importantly, the purified p97/VCP, Ufd1, and Npl4 appear to be able to form a complex under the same conditions in the absence of gp78 (data not shown). Together, these data suggest that gp78 and Ufd1-Npl4 dimer form mutually exclusive complexes with p97/ VCP in cells (Fig. 4d).

gp78-mediated ERAD
gp78 Recruits p97/VCP but Not Ufd1 Complex to the ER through Its VIM-We have previously reported that gp78 is largely localized in the ER based on the fact that it co-localizes with the ER-anchored ERAD E2 MmUbc6 (17,30). In this report, we have demonstrated that gp78 can recruit p97/VCP from the cytosol to microsomes (Fig. 2, b and c). Here, we asked whether gp78 could recruit p97/VCP to the ER in intact cells. Using double immunofluorescent staining, we examined the effects of gp78 and gp78⌬VIM on the localization of endogenous p97/VCP. U2OS cells were chosen for these experiments because their flattened morphology allows excellent imaging of the ER network (17). A plasmid construct encoding wt gp78 or gp78⌬VIM was transfected into U2OS cells and cells were double-stained with a rabbit polyclonal anti-gp78 (recognizing both gp78 and gp78⌬VIM) and mouse monoclonal anti-p97/ VCP. Cells transfected with empty vector exhibited diffuse distribution of endogenous p97/VCP in the cytosol and nuclei (Fig.  5a). Wt gp78 expression relocalizes p97/VCP from the cytosol to the ER where gp78 resides (Fig. 5b). Deletion of VIM from gp78 (gp78⌬VIM) markedly reduces this effect (Fig. 5c), suggesting that gp78 recruits p97/VCP to the ER via its VIM.
To test this possibility further, we generated a construct that expresses a fusion protein of VIM to the C terminus of Hrd1, another ER-anchored E3. Double immunofluorescent staining revealed that Hrd1 alone has minimal effects on the localization of p97/VCP (Fig. 5d). However, fusion of VIM to Hrd1 (Hrd1 plus VIM) dramatically increases co-localization of p97/ VCP with Hrd1 plus VIM (Fig. 5e). Interestingly, the co-localized p97/ VCP and Hrd1 plus VIM exhibit a punctate pattern in the cytoplasm of the cells. Because Hrd1 is an ER protein, the punctate structures are likely to be derived from the ER. These results further indicate that the VIM of gp78 mediates the recruitment of p97/VCP from the cytosol to the ER. gp78 recruits p97/VCP but does not appear to recruit Ufd1 to the microsomes by fractionation assay (Fig. 2c). To confirm this unexpected result, immunofluorescent staining was utilized to further test the effect of gp78 on Ufd1 localization in cells. We co-transfected U2OS cells with plasmids that encoded FLAG-tagged Ufd1 and wt gp78. Transfected cells were doubly labeled with monoclonal anti-FLAG antibody for Ufd1 and polyclonal anti-gp78 antibodies. Consistent with the fractionation results (Fig.  2c), we found that gp78 does not affect Ufd1 localization in cells (Fig. 5f). Together, these results suggest that gp78 recruits p97/ VCP but not the p97/VCP-Ufd1-Npl4 complex to the ER.
We next asked whether the difference in the steady-state levels of CD3␦ between CD3␦/gp78 clone-20 and CD3␦ clone-5

gp78-mediated ERAD
cells affects the outcomes of RNAi for Ufd1 as we see in Figs. 7a and 8a. We therefore expressed lower levels of CD3␦ in 293 cells with physiological levels of gp78 and higher levels of CD3␦ in cells overexpressing gp78. Effects of knockdown of p97/VCP and Ufd1 on the levels of CD3␦ in these cells were evaluated by immunoblotting. The results show that, although the steadystate levels of CD3␦ are lower, knockdown of Ufd1 produces significant stabilization of CD3␦ in cells with physiological levels of gp78 (Fig. 8c, lanes 1 and 3). Little effect on CD3␦ levels is observed in cells that overexpress gp78 (Fig. 8c, lanes 4 and 6). However, knockdown of p97/VCP leads to marked stabilization of CD3␦ in both cell types (Fig. 8c, lanes 2 and 5). These results are identical to those for CD3␦/gp78 clone-20 (Fig. 7a) and CD3␦ clone-5 (Fig. 8a). Therefore, we conclude that any differences we observed in responses to Ufd1 knockdown in CD3␦/ gp78 clone-20 and CD3␦ clone-5 are not due to differences in their steady-state levels of CD3␦. Rather, the p97/VCP-Ufd1-Npl4 complex is involved in CD3␦ degradation in CD3␦ clone-5 cells, whereas Ufd1 is dispensable for CD3␦ degradation in clone-20 cells that overexpress gp78. To determine if this Ufd1independent degradation is true for other gp78 substrate, we examined the effects of p97/VCP and Ufd1 knockdown on the levels of the Z variant of ␣-1-antitrypsin, a newly identified gp78 substrate. 4 The knockdown was carried out in 293 cells that stably express the Z variant of ␣-1-antitrypsin. The results showed that silencing p97/VCP but not Ufd1 stabilizes the Z variant of ␣-1-antitrypsin (supplemental Fig. S2), which further supports the possibility that gp78 mediates a Ufd1-independent ERAD pathway.
We then examined the Ufd1 dependence of the gp78-mediated ERAD in CD3␦ clone-5 cells. The Ufd1 dependence was determined by comparing the stabilization effects on CD3␦ in gp78 or Ufd1 single knockdown and gp78-Ufd1 double knock- RNAi, CHX chase, and immunoblotting were performed as in Fig. 6. c, the effects of p97/VCP or Ufd1 knockdown on CD3␦ stabilization were determined in 293 cells that express lower levels of CD3␦ (lanes 1-3) and 293 cells that express higher levels of both CD3␦ and gp78 (lanes 4 -6). Equal amounts of total protein prepared from cells with indicated treatments were analyzed by immunoblotting for the specified proteins. Tubulin was blotted as a loading control. d, RNAi of endogenous gp78 inhibits CD3␦ degradation. CD3␦ clone-5 cells were transfected with siRNA expression empty vector pSUPER (C) or pSUPER-gp78siRNA1 (G) as previously reported (18). CD3␦ degradation was monitored by CHX chase. Endogenous gp78 was detected by monoclonal anti-gp78 antibody 2G5. e, simultaneous knockdown of gp78 and Ufd1 produces synergistic inhibitory effects on CD3␦ degradation. CD3␦ clone-5 cells transfected with siRNA for Ufd1 (U) or gp78 (G) alone, or both (U ϩ G) simultaneously were analyzed by CHX chase followed by immunoblotting for the specified proteins. FIGURE 7. Effects of inhibiting p97/VCP or Ufd1 expression by RNAi on CD3␦ degradation in CD3␦/gp78 clone-20 cells. a, inhibition of p97/VCP by siRNA stabilizes CD3␦, but Ufd1 siRNA has minimal effects in CD3␦/gp78 clone-20. Control siRNA (C) or siRNA for p97/VCP (V1 and V2 represent two different siRNAs targeting different regions of p97/VCP mRNA; shown in duplicate, lanes [2][3][4][5] or Ufd1 (U in duplicate, lanes 6 and 7) were transfected into CD3␦/gp78 clone-20 cells. Protein levels of CD3␦, p97/VCP, and Ufd1 were examined by immunoblotting. Actin was blotted as a loading control. b, inhibition of p97/VCP but not Ufd1 expression by RNAi markedly inhibits CD3␦ degradation in CD3␦/gp78 clone-20 cells. CD3␦/gp78 clone-20 cells transfected as indicated were analyzed by CHX chase. RNAi and immunoblotting for HA, p97/VCP, and Ufd1 were performed as in a. Tubulin was blotted as loading control. c, gp78 is associated with p97/VCP but not Ufd1 in CD3␦/gp78 clone-20 cells. gp78 or p97/VCP was immunoprecipitated from CD3␦/gp78 clone-20 cells (lane 4 and 5). Anti-HA and anti-GFP antibodies were used as negative controls (lanes 2 and 3). Ufd1 in immunoprecipitates was detected by immunoblotting. NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 down cells. If gp78 ubiquitinates CD3␦, and then the p97/VCP-Ufd1-Npl4 complex retrotranslocates the ubiquitinated CD3␦ from the ER for degradation, we would expect little or no additive inhibitory effects on CD3␦ degradation by a double knockdown of gp78 and Ufd1. However, the results from our CHX chase experiments show that the double knockdown of Ufd1 and gp78 produces a significant additive effect in the inhibition of CD3␦ degradation (Fig. 8e, lanes 9 -12). In addition, individual knockdown of gp78 or Ufd1 resulted in similar inhibitory effects (Fig. 8e, lanes 1-4 versus [5][6][7][8] suggesting that both gp78-and Ufd1-mediated pathways may contribute significantly to CD3␦ degradation in CD3␦ clone-5 cells. Interestingly, knockdown of Ufd1 appears to stabilize gp78 (Fig. 8e,  lanes 1-4). This stabilization effect may represent a compensatory mechanism to ensure ERAD in the shortage of Ufd1.

gp78-mediated ERAD
CD3␦ degradation in an Ufd1-independent manner. In addition, the results help explaining why the steady-state levels of CD3␦ are lower in clone-16 and -20 cells (Fig. 6). Collectively, our results suggest the presence of at least two pathways for CD3␦ degradation in cells with physiological levels of ERAD machinery: the p97/VCP-Ufd1-Npl4 complex-mediated pathway and the gp78-mediated and Ufd1-independent pathway.

DISCUSSION
Studies of both yeast and higher eukaryotes have indicated that the Ufd1-Npl4 dimer is indispensable for CDC48/p97/VCP-mediated retrotranslocation of ubiquitinated ER proteins (7,8). Here, we provide evidence to suggest a novel mechanism by which gp78mediated ubiquitination is coupled with p97/VCP-mediated retrotranslocation without requiring Ufd1 (see a diagrammatic illustration of this model in Fig. 10). The molecular mechanism underlying this Ufd1-independent pathway is mediated through a novel p97/VCP-interacting motif, i.e. VIM of gp78. The VIM of gp78 interacts with the ND1 domain of p97/VCP, which is also the binding site for Ufd1 (36,37). Sharing of the interaction site leads gp78 and Ufd1 to form mutually exclusive complexes with p97/ VCP (Fig. 4d). Therefore, gp78 physically excludes Ufd1 from the gp78-p97/VCP complex, thereby functionally excluding Ufd1 from gp78-mediated ERAD. However, our data does not exclude the possibility that other p97/VCP cofactor(s) might be involved in gp78-p97/VCP-mediated ERAD (Fig. 10).
The Ufd1-independent degradation of CD3␦ becomes dominant in cells that overexpress gp78 as demonstrated by our RNAi experiments in CD3␦/gp78 clone-20 cells. Our data also support the presence of this pathway in cells that express endogenous levels of gp78. First, knockdown of gp78 stabilizes CD3␦ in CD3␦ clone-5 cells, indicating that endogenous gp78 is involved in ERAD (Fig. 8c). Second, although knockdown of Ufd1 accumulates CD3␦, CD3␦ degradation still occurs although at a slower rate (Figs. 8b, 9a, and 9b). This suggests that an Ufd1-independent pathway exists in CD3␦ clone-5 cells. Third, simultaneous knockdown of gp78 and Ufd1 results in a significant additive effect on CD3␦ stabilization in CD3␦ clone-5 cells (Fig. 8d). When expression of both gp78 and Ufd1 are almost abolished (Ͼ90% reduction), this additive inhibitory effect would only be possible if gp78 and Ufd1 act in parallel independent pathways. If gp78 and Ufd1 work sequentially, that is, gp78 mediates ubiquitination of CD3␦ and then Ufd1 in conjunction with p97/VCP and Npl4 retrotranslocates the ubiquitinated CD3␦ for degradation, knockdown of Ufd1 would have little effect when the function of gp78 is diminished. If gp78 and Ufd1 work in parallel in different degradation pathways, that is, they act in independent pathways on CD3␦ degradation, simultaneous knockdown would block both pathways and produce significant additive effects of inhibition, which is what we have demonstrated. In contrast, p97/VCP is essential to both gp78 and Ufd1-mediated pathways. p97/VCP knockdown significantly stabilizes CD3␦ regardless the levels of gp78 expression (Figs. 7, 8a, 8b, 9a, 9b, and 9c). This suggests that both Ufd1-dependent and -independent pathways require the function of p97/VCP.
Studies have suggested that the p97/VCP-Ufd1-Npl4 complex is recruited to the ER by an ER-localized multiprotein complex through multiple interactions (13, 14, 16, 38) (Fig. 1b). In addition, other cytosolic ERAD components are also recruited to the ER by the multiprotein complex. For examples, both Derlin-1 and p97/ VCP interact with the peptide:N-glycanase, a key enzyme that removes N-glycan from misfolded ER proteins before proteasomal degradation (39,40). Peptide:N-glycanase also interacts with hHR23B, a polyubiquitin-binding protein that chaperones polyubiquitinated ERAD substrates to the proteasomes for degradation (39,41). HERP, an ER stress-inducible protein, has recently been identified as a new member of the ERAD complex (14). However, how HERP regulates ERAD remains unknown. Nevertheless, there appears to be a large protein complex in the ER, which integrates recruitment, ubiquitination, retrotranslocation, deglycosylation, and proteasomal targeting of misfolded ER substrates during ERAD. The functional details and the dynamics of the complex are yet to be determined.
This report and recent studies by others suggest that gp78 may be the central player in another ERAD pathway. gp78 acts not only as an ERAD E3, but also as a scaffold protein to assemble a complex that couples ubiquitination, retrotranslocation, and deglycosylation. gp78 has a multifunctional cytosolic tail that conveys E3 activity, binds polyubiquitin, and recruits Ubc7 and p97/VCP (17,18,35). Recent evidence indicates that p97/ VCP recruited by gp78 is associated with peptide:N-glycanase (39,42). We have now demonstrated that gp78 has a p97/VIM responsible for recruiting p97/VCP to the ER. Because gp78 and Ufd1 compete for binding to p97/VCP, gp78 recruits only p97/ VCP but not the p97/VCP-Ufd1-Npl4 complex to the ER. Consistent with the absence of Ufd1 in the gp78-p97/VCP complex, our evidence also suggests that Ufd1 is dispensable in gp78mediated CD3␦ degradation. Other p97/VCP cofactor(s), however, may be involved in gp78-mediated ERAD (Fig. 10).
Ufd1-independent ERAD may not be limited to gp78. VIM is also present in SVIP, another ER-membrane localized p97/VCPbinding protein (21). 5 SVIP, p47, and Ufd1 form mutually exclusive complexes with p97/VCP (21). It is likely that the VIM of SVIP interferes with the binding of p47 or Ufd1 to p97/VCP, as does the VIM of gp78. Therefore, SVIP may cooperate with other ERAD E3 to mediate Ufd1-independent ERAD. Thus, Ufd1-independent ERAD could represent one of the major pathways for disposal of misfolded ER proteins. In addition, many additional CDC48/p97/ VCP cofactors with ubiquitin-binding capability have been identified (28,43). Once the functions of the cofactors are fully uncovered, other mechanisms of coupling ubiquitination with retrotranslocation and degradation may be discovered. The presence of multiple ERAD pathways may be required to degrade different substrates and to facilitate efficient removal of misfolded proteins under different circumstances.