To identify proteins that associate with gp78, we began by generating a line of CHO-7 cells that stably overexpress the full-length enzyme with a C-terminal TAP tag. The TAP tag is composed of three copies of a T7 epitope and Protein A separated by a cleavage site for the TEV protease. In previous studies (15
), we found that appending epitope tags to the C terminus of gp78 impairs ubiquitin ligase activity. When overexpressed in cells, epitope-tagged gp78 exhibits dominant-negative activity toward reductase ERAD by competing with the endogenous enzyme for binding to Insig. In A
, we compared the sterol-accelerated degradation of reductase in non-transfected cells with that in cells stably overexpressing gp78-TAP (designated CHO/gp78-TAP). The cells were first depleted of sterols through incubation for 16 h in medium containing lipoprotein-deficient serum and the reductase inhibitor compactin. The cells also received a low concentration of mevalonate (50 μm
), which allows for the synthesis of essential nonsterol isoprenoids but not of cholesterol (24
). The cells were then treated for 5 h with various concentrations of the regulatory oxysterol 25-HC plus 10 mm
mevalonate, after which they were harvested and subjected to subcellular fractionation. Immunoblot analysis of the resulting membrane fractions with monoclonal anti-reductase antibody revealed that 25-HC stimulated degradation of reductase in a dose-dependent manner (A
, top panel
, lanes 1–4
). In contrast, reductase resisted sterol-induced ERAD in CHO/gp78-TAP cells (lanes 5–8
) even though the amount of transfected gp78 drastically exceeded that of the endogenous enzyme (bottom panel
, lanes 5–8
). This result implies that gp78-TAP is inactive but retains the ability to associate with Insigs or some other component required for sterol-accelerated ERAD of reductase.
We next conducted a large scale TAP experiment in which detergent lysates of CHO/gp78-TAP cells treated with the proteasome inhibitor MG-132, 25-HC, and mevalonate were subjected to affinity chromatography using IgG- and anti-T7-coupled agarose beads. The eluted proteins, which include gp78 and any associated proteins, were fractionated by SDS-PAGE and visualized by silver staining. Segments of the gel containing visible bands that were not present in mock-purified samples (supplemental Fig. 1A
) were excised and digested with trypsin, and protein identities were determined by mass spectrometry. One of these proteins, SPFH2, was chosen for further study because of its recently appreciated role in the regulated ERAD of membrane-bound inositol trisphosphate receptors (25
SPFH2 (also known as Erlin-2) belongs to a family of proteins that contain a stretch of ~340 amino acids of unknown function called the SPFH domain (named for the founding members S
lotillin, and H
). SPFH2 is a 45-kDa glycoprotein that localizes to lipid-rich regions of the ER and is anchored to membranes through a single N-terminal membrane-spanning segment (25
). The remainder of the protein, including the SPFH domain, is located within the ER lumen. SPFH2 forms a high molecular weight complex with its close homolog SPFH1 (or Erlin-1); these proteins associate with activated inositol trisphosphate receptors before their ubiquitination and degradation. Both of these processes are significantly blunted when expression of SPFH1 and SPFH2 are reduced through RNAi (25
). In addition to inositol trisphosphate receptors, SPFH2 associates with other ERAD substrates including the cystic fibrosis transmembrane conductance receptor ΔF508 mutant (26
To confirm the association of gp78 with SPFH2, sterol-depleted cells were treated in the absence or presence of 25-HC plus mevalonate and MG-132 before lysis in detergent-containing buffer. The resulting lysates were then immunoprecipitated with polyclonal anti-gp78 IgG or control preimmune IgG. Immunoblot analysis of the resulting immunoprecipitates revealed specific pulldown of endogenous gp78 (B
, top panel
, compare lanes 1
, and 6
with lanes 3
, and 8
). Endogenous SPFH2 co-immunoprecipitated with gp78 when cells were treated in both the absence and presence of 25-HC plus mevalonate (second panel
3 and 4
); MG-132 had no effect on this interaction (lanes 7
). We also probed the anti-gp78 precipitates for reductase and VCP/p97. Treatment of the cells with 25-HC plus mevalonate caused the disappearance of reductase from the supernatant fraction of the immunoprecipitate (third panel
, lanes 9–12
), indicating accelerated degradation of the enzyme. Regulated degradation of reductase was blocked by MG-132 (lanes 13–16
), which also caused the appearance of enzyme in the pellet fraction of 25-HC and mevalonate-treated cells (compare lanes
7 and 8
). In MG-132-treated cells, VCP/p97 also co-precipitated with gp78 in a sterol-independent manner (bottom panel
7 and 8
). The constitutive, non-regulated interaction can be explained by our previous results that gp78 bridges the reductase-Insig complex to VCP/p97 (14
In the next set of experiments we compared the association of SPFH2 and SPFH1 with gp78 and two other membrane-bound ubiquitin ligases, Trc8 and Hrd1. Cells were transfected with expression plasmids encoding T7-tagged SPFH1 or SPFH2 together with increasing amounts of plasmids encoding Myc-tagged gp78 (A), Hrd1 (B), or Trc8 (C). After transfection, the cells were harvested for preparation of detergent lysates, which were immunoprecipitated with anti-T7-coupled agarose beads to pull down transfected SPFH1 or SPFH2. The results show that SPFH1 co-immunoprecipitated with gp78 (A, top panel, lanes 5–8) and Hrd1 (B, top panel, lanes 5–8) but not with Trc8 (C, top panel, lanes 5–8). In contrast, SPFH2 formed a complex with all three ubiquitin ligases (, A–C, top panels, lanes 10–12). It should be noted that the affinity of SPFH2 for gp78 and Hrd1 exceeds that of SPFH1 as indicated by the appearance of the ubiquitin ligases in the SPFH2 immunoprecipitates when low amounts of the enzymes were transfected.
To appraise a role for SPFH2 in the sterol-induced ubiquitination of endogenous reductase, we transfected HEK-293 cells with various combinations of siRNAs targeting vesicular stomatitis virus glycoprotein (VSV-G), a control gene not expressed in the cells, SPFH1, and SPFH2 (A). After treatment with MG-132 in the absence or presence of 25-HC plus mevalonate, the cells were harvested, and detergent lysates were immunoprecipitated with polyclonal anti-reductase antibodies. Immunoblot analysis of the resulting precipitated material with anti-ubiquitin revealed that in control-transfected cells, reductase became ubiquitinated upon 25-HC plus mevalonate treatment (A, top panel, compare lanes 1 and 2). RNAi-mediated knockdown of SPFH1 had no effect on this reaction (lanes 3 and 4). In contrast, knockdown of SPFH2 significantly blunted the regulated ubiquitination of reductase (lanes 5 and 6); similar results were obtained when expression of both SPFH1 and SPFH2 were reduced by RNAi (lanes 7 and 8). To confirm this result, we conducted another RNAi experiment with an additional siRNA duplex targeting SPFH2 (B). The results show that the alternative siRNA blunted sterol-regulated ubiquitination of reductase, albeit the effect was reduced compared with the original duplex (B, top panel, compare lane 4 with lane 6). Quantitative real-time PCR revealed that SPFH1 and SPFH2 expression was reduced by 90 and 80%, respectively, by RNAi (data not shown). A similar experiment was conducted in the absence of MG-132 to examine whether SPFH2 knockdown blocks sterol-accelerated degradation of reductase. The results show that 25-HC stimulated reductase degradation in cells transfected with control vesicular stomatitis virus glycoprotein siRNA duplexes (B, top panel, lanes 1–3), but this degradation was significantly blunted in cells that received siRNA duplexes targeting SPFH2 (lanes 4–6).
Having established a role for SPFH2 degradation of reductase, experiments were next designed to define the interactions between gp78, SPFH2, and reductase. In the experiment of A, endogenous reductase was immunoprecipitated from lysates of cells treated in the absence or presence of 25-HC plus mevalonate and MG-132. In the absence of MG-132, 25-HC plus mevalonate caused reductase to become degraded, as indicated by reduced levels of the protein in the immunoprecipitate (A, first and fourth panels, compare lanes 1 and 2). This degradation was blocked by MG-132 (lanes 3 and 4), and immunoprecipitation of reductase pulled down endogenous SPFH2, gp78, and VCP/p97 in a manner that was enhanced by the presence of 25-HC plus mevalonate (second, third, and fifth panels, compare lanes 3 and 4). The low level of SPFH2-reductase binding in the absence of 25-HC is likely caused by residual sterols in the lipid-deprived cells.
The Insig requirement for SPFH2-reductase binding was examined by comparing co-immunoprecipitation of the two proteins with or without co-expression of pCMV-Insig-1-Myc encoding human Insig-1 followed by six copies of the Myc epitope. As shown in B, immunoprecipitation of overexpressed reductase brought down a small amount of SPFH2 from lysates of cells treated in both the absence and presence of 25-HC plus mevalonate (second panel, lanes 1 and 2). When Insig-1 was co-expressed, 25-HC plus mevalonate treatment caused a significant increase in the amount of SPFH2 that was brought down by reductase immunoprecipitation (compare lanes 5 and 6).
The experiment of C shows that the Insig-SPFH2 interaction is mediated in part by gp78. Immunoprecipitation of Insig-1-Myc brought down a small amount of SPFH2 (C, second panel, lane 3); this co-immunoprecipitation was markedly enhanced by the overexpression of gp78 (lane 4). Considered together, the results of , A–C, indicate the reductase-SPFH2 complex that forms in sterol-treated cells is mediated by interactions between gp78 and Insig-1.
To identify the region of gp78 that mediates its association with SPFH2, we conducted the co-immunoprecipitation experiment of D. In this experiment cells were transfected with SPFH2 together with full-length gp78, the truncated membrane domain of gp78, or the cytosolic domain of gp78 anchored to membranes through the membrane attachment region of cytochrome P450 2C1. As expected, precipitation of SPFH2 brought down full-length gp78 (D, top panel, lanes 1 and 2); a similar result was obtained for the cytosolic domain of gp78 (lanes 5 and 6). In contrast to this, the membrane domain of gp78 failed to co-precipitate with SPFH2 (lanes 3 and 4), indicating that its binding to gp78 is mediated by the cytosolic domain of the enzyme.
The observation that the cytosolic domain of gp78 mediates its association with SPFH2 is surprising considering the majority of SPFH2 resides in the ER lumen (25
). This finding suggests the existence of a protein that bridges SPFH2 to gp78. To identify this protein, we employed HEK-293 cells stably overexpressing SPFH2 with a C-terminal TAP tag (designated SPFH2-TAP) and conducted a two-step affinity purification similar to that used to identify gp78-associated proteins (see ). This procedure led to identification of several proteins that appeared in purifications from lysates of SPFH2-TAP-expressing cells (supplemental Fig. 1B
). One of these proteins, TMUB1, was chosen for further study. The cDNA for TMUB1 predicts a 245-amino acid protein that contains one N-terminal and two C-terminal hydrophobic domains (A
). Protease protection studies indicate that the first hydrophobic domain (HD) inserts into membranes in a hairpin fashion, whereas the two C-terminal hydrophobic domains traverse the membrane. Previous studies demonstrate that the TMUB1 UBL domain projects into the cytosol (29
To confirm the interaction between SPFH2 and TMUB1, we transfected cells with an expression plasmid encoding T7-tagged SPFH2 together with plasmids encoding Myc-tagged TMUB1, UbxD2, and UbxD8. UbxD8 and UbxD2 are Ubx-domain containing proteins that are known to play key roles in ERAD (30
). Overexpression of TMUB1 gave rise to two bands in immunoblots (B
, top panel
, lanes 1
); the molecular weight of the slower migrating band corresponds to full-length TMUB1 as predicted by its cDNA. Mutagenesis studies show that the faster migrating band results from the use of an alternative site of translational initiation at methionine 56 (data not shown). Immunoprecipitation of transfected TMUB1 brought down SPFH2 (second panel
, lanes 1
) but not UbxD2 (lanes 3
) or UbxD8 (lanes 5
We next designed experiments to determine which region of TMUB1 mediates its binding to SPFH2. Wild type, full-length TMUB1 co-immunoprecipitated with SPFH2 as expected (C, second panel, lane 2). This binding was abolished by deletion of the first hydrophobic domain of TMUB1 (second panel, compare lanes 2 and 4). In contrast, mutant forms of TMUB1 lacking both TM1 and TM2 or only TM2 continued to bind SPFH2 (lanes 6 and 8, respectively). TMUB1 lacking the UBL domain also bound to SPFH2 (data not shown).
We next used co-immunoprecipitation to compare the binding of wild type and mutant forms of TMUB1 to gp78 (D). The results show that as expected, gp78 co-immunoprecipitated with wild type TMUB1 (second panel, lane 1). Deletion of the N-terminal HD completely abolished this interaction (lane 2), whereas deletion of TM1 and TM2 had an intermediate effect on gp78-TMUB1 binding (lane 3). In contrast, mutant TMUB1 lacking only TM2 continued to associate with gp78 (lane 4); deletion of all three hydrophobic domains in TMUB1 renders the protein cytosolic (data not shown) and abolished gp78 binding. Together, these results indicate that the N-terminal HD and TM1 of TMUB1 contribute significantly to the association of the protein with gp78.
The experiment of E was designed to establish that TMUB1 bridges SPFH2 to gp78. Immunoprecipitation of SPFH2 brought down a small amount of gp78 when the proteins were overexpressed together in cells (third and fourth panels, lanes 1 and 2). The co-immunoprecipitation between gp78 and SPFH2 was markedly enhanced when TMUB1 was co-expressed, indicating the protein bridges SPFH2 to gp78 (third and fourth panels, compare lanes 1 and 2 with 5 and 6). It should be noted that the larger form of TMUB1 corresponding to the full-length protein exhibited a significantly higher affinity for SPFH2 than the shorter form. Considering that the shorter form of TMUB1 is translationally initiated at methionine 56, this result is consistent with a role for the N-terminal hydrophobic domain of TMUB1 in mediating association with gp78 and SPFH2.
A role for TMUB1 in sterol-regulated ubiquitination and degradation of reductase was addressed in . In cells transfected with control siRNA duplexes, reductase became ubiquitinated (, A and B, top panels, lane 2) and degraded (B, top panel, lanes 1–3) in the presence of 25-HC and mevalonate. The RNAi-mediated knockdown of TMUB1 significantly blunted both the sterol-induced ubiquitination (, A, top panel, lanes 4, 6, and 8, and B, top panel, lane 4) and sterol-accelerated degradation (B, top panel, lanes 4–6) of endogenous reductase.