VCIP135 deubiquitinase and its binding protein, WAC, in p97ATPase-mediated membrane fusion

WAC directly binds to VCIP135 and forms a complex with VCIP135 and p97

Although VCIP135 is required for p97/p47 and p97/p37-mediated Golgi membrane fusion, its deubiquitinating activity is required only in the p97/p47 pathway, and not in the p97/p37 pathway. This indicates that VCIP135 has at least two functions, one ubiquitin-dependent and the other ubiquitin-independent. The question is how VCIP135 chooses either of these distinct functions. One possible answer is that VCIP135 may have its adaptor proteins, which guide it towards one of the functions. Based on this idea, we performed a yeast two-hybrid screen to identify VCIP135-binding proteins, and identified a clone encoding cDNA of WAC. As shown in Figure 1A, WAC was immunoprecipitated from various rat tissues and its amounts were estimated by western blotting. WAC was widely expressed in all the tissues we tested.

The intracellular distribution of WAC was determined by immunofluorescence microscopy with paraformaldehyde fixation, and the results are presented in Figure 1B. Double immunofluorescence staining with polyclonal antibodies to WAC and monoclonal antibodies to GM130, a Golgi marker, showed that WAC mainly localizes to the Golgi as well as the nucleus, although a small amount is present in the cytosol.

We next aimed to confirm the interaction between WAC and VCIP135. Since both WAC and VCIP135 exist in the Golgi (Figure 1B), Golgi membrane extracts were used for immunoprecipitation experiments. Golgi membranes were extracted and WAC and its binding proteins were co-immunoprecipitated with antibodies to WAC. As shown in Figure 2A, both VCIP135 and p97, its binding protein, were co-precipitated with WAC (top and middle panels, lane 2), indicating that WAC in Golgi membranes is involved in a complex containing VCIP135 and p97.

We further investigated the binding between WAC and VCIP135 in vitro, using purified recombinant proteins. GST-tagged WAC was incubated with either VCIP135 or p97, as presented in Figure 2B. WAC bound VCIP135 (top panel, lane 2), but not p97 (middle panel, lane 3). When VCIP135 and p97 were added together, WAC bound p97 as well as VCIP135 (lane 4). These results indicate that WAC directly binds to VCIP135 and that p97 is included in the complex via its binding to VCIP135. The amount of VCIP135 bound to WAC increased in the presence of p97 (Figure 2B, top panel, lanes 2 and 4; Supplementary Figure S2A), suggesting that the binding of p97 to VCIP135 increases the binding affinity between WAC and VCIP135. To test this, we performed another binding experiment as presented in Figure 2C. WAC and p97 were incubated with GST-tagged VCIP135 and the amounts bound to VCIP135 were assayed. The amount of WAC bound to VCIP135 was increased in the presence of p97 (Figure 2C, middle panel, lanes 2 and 4; Supplementary Figure S2B), which is consistent with the result in Figure 2B. Regarding the amount of p97 bound to VCIP135, there was no difference between the absence and presence of WAC (Figure 2C, top panel, lanes 3 and 4). Taken together, these results indicate that WAC, VCIP135 and p97 form a complex and the binding affinity between VCIP135 and WAC is enhanced by the binding of p97 to VCIP135.

Isolation of a VCIP135 mutant, VCIP135(Q381A), which lacks binding affinity to WAC

As WAC directly binds to VCIP135, we carried out a rough mapping experiment of VCIP135 to determine the WAC-binding region in VCIP135, as presented in Figure 3A. WAC bound to the N-terminal half of VCIP135, but not to its C-terminal half (bottom panel, lanes 2 and 3). In contrast, p97 bound to the C-terminal half of VCIP135 (top panel, lane 3). These results strongly support our finding that WAC does not compete with p97 for its binding to VCIP135 (Figure 2B and C).

We next aimed to produce VCIP135 mutants with a point mutation in the WAC-binding region. The cDNA of the N-terminal half of VCIP135 was amplified by a PCR reaction with a limiting amount of dATP, resulting in the introduction of random mutations. Full-length cDNAs of VCIP135 including these mutated N-terminal fragments were expressed in yeast cells and their WAC-binding affinities were tested using a yeast two-hybrid system. We screened ∼1 × 10³ colonies and successfully obtained one mutant (Q381L) showing low WAC-binding affinity. The Gln-381 is conserved in all VCIP135 orthologues we found (Supplementary Figure S5).

Since full-length VCIP135 protein with Q381L mutation expressed in Escherichia coli became insoluble, VCIP135 with the Q381A mutation (VCIP135(Q381A)) was instead expressed in E. coli and its biochemical characteristics were analysed. Figure 3B shows the binding of VCIP135wt and VCIP135(Q381A) to GST-tagged WAC. VCIP135wt bound to WAC (top panel, lane 1), while VCIP135(Q381A) showed almost no binding to WAC (top panel, lane 2). As to their binding to p97, there was no difference between VCIP135wt and the Q381A mutant (Figure 3C, top panel).

The deubiquitinating activity of VCIP135(Q381A) was also assayed (Figure 3D). Following the method of Wang et al (2004), a mixture of ubiquitin chains (Ub2–7) was used as a substrate, and VCIP135(C218A) and (C218S) were used as negative controls. In order to detect very low deubiquitinating activities of VCIP135wt and its mutants, a smaller amount of substrate was used in this assay compared with that in the other assays. VCIP135(Q381A) generated almost the same amount of monoubiquitin after the reaction as VCIP135wt (top panel, lane 3). Therefore, the Q381A mutation in VCIP135 appears to have no effect on its deubiquitinating activity.

WAC activates VCIP135 deubiquitinating activity and is required for p97/p47-mediated Golgi membrane fusion

VCIP135 is reported to have deubiquitinating activity, which is essential for p97/p47-mediated Golgi reassembly (Wang et al, 2004). We therefore investigated the effect of WAC on the deubiquitinating activity of VCIP135. We first used a mixture of ubiquitin chains as a substrate in the deubiuitinating assay. As presented in Figure 4A, VCIP135 alone showed very low deubiquitinating activity (top panel, lane 1). Addition of WAC significantly increased the deubiquitinating activity of VCIP135 (top panel, lane 2), while the addition of p97 had almost no effect or a very slight effect (top panel, lane 3). When both WAC and p97 were added together, deubiquitinating activity was markedly increased (top panel, lane 4), to greater than 10 times compared with that of VCIP135 alone (Figure 4B). This increase is likely to be caused by the p97-mediated increase in binding affinity between VCIP135 and WAC (Figure 2B and C; Supplementary Figure S2A and B). These results indicate that WAC enhances VCIP135 deubiquitinating activity through its binding to VCIP135. We next used the extract from salt-washed mitotic Golgi fragments as a substrate in the assay (Figure 4C). Although WAC and p97 still remained in salt-washed membranes (third and fourth panels from the top, lane 1), VCIP135 was not detected (second panel from the top, lane 1) and either recombinant VCIP135 or its mutant was added instead. The addition of VCIP135wt increased the generation of mono-Ub (top panel, lane 2), while the addition of VCIP135(Q381A), which lacked binding affinity to WAC (Figure 3), did not (top panel, lane 3). There still remain some possibilities that this might be an indirect effect caused by using crude substrates. However, considering that VCIP135(Q381A) per se has almost the same deubiquitinating activity as VCIP135wt (Figure 3D), this result rather suggests that the binding of VCIP135 to WAC is necessary for the deubiquitinating function of VCIP135. Taking together, it is thought that WAC enhances VCIP135 deubiquitinating activity through its binding to VCIP135.

Next, we aimed to clarify the in vivo function of WAC and carried out WAC-knockdown experiments using siRNA (Figure 5A–D). Two distinct WAC siRNA duplexes, specific for the nucleotide sequence of WAC, were introduced in HeLa cells. In WAC siRNA-treated cells, WAC was undetectable by immunofluorescence staining (Figure 5A, panels e and f). Western blotting also showed that both WAC siRNA duplexes decreased WAC levels to <5% of its original level (Supplementary Figure S1B). In WAC-depleted cells, although the Golgi still localized to a perinuclear region, it was slightly dispersed (panels b and c). The ultrastructures of the Golgi apparatus were further investigated by electron microscopy (Figure 5B and C). In WAC-depleted cells, the typical big stacks of Golgi were rarely observed and large-scale vesiculation was present (Figure 5B, top right and bottom panels). Membrane profiles in the Golgi area were calculated and the results are shown in Figure 5C. Both WAC siRNAs decreased the percentage of membranes in Golgi cisternae by ∼40%, while increased the percentage of tubules and vesicles within the Golgi area, compared with the mock control. These results strongly suggest that WAC is important for the biogenesis of the Golgi.

Since WAC forms a complex with VCIP135 in Golgi membranes (Figure 2A), another question is whether the subcellular localization of VCIP135 is altered when WAC expression is suppressed in living cells. The localization of VCIP135 in WAC siRNA-treated cells is shown in Figure 5D. The depletion of WAC (Supplementary Figure S1C) caused the partial dissociation of VCIP135 from the Golgi (Figure 5D, bottom left panel), although some VCIP135 still remained in the Golgi. Considering our biochemical result that VCIP135 directly binds to WAC (Figure 2B and C), it is likely that part of VCIP135 in the Golgi requires its binding to WAC for its Golgi localization.

To clarify the molecular function of WAC in Golgi biogenesis, we performed the in vitro Golgi reformation assay (Uchiyama et al, 2006). As shown in Figure 6A, cisternal growth was observed by incubating Golgi fragments with the p97/p47 complex (the control). Polyclonal antibodies to WAC inhibited the p97/p47-mediated cisternal growth by ∼75% compared with the control. This inhibition was rescued by quenching the antibodies with recombinant WAC protein. The same antibodies had no effect on p97/p37-mediated cisternal growth. Together, these results strongly suggest that WAC is specifically involved in the p97/p47 membrane fusion pathway.

We also tested VCIP135(Q381A), which lacks binding affinity to WAC but maintains its binding affinity to p97 as well as its deubiquitinating activity at the same level as the wild type (Figure 3B–D), in the in vitro Golgi reformation assay (Figure 6B). Golgi membranes were washed with 1 M KCl to remove membrane-bound VCIP135 before being used for the assays. WAC still remained in Golgi membranes after the salt-wash treatment. When the salt-washed membranes were incubated with p97/p47 and VCIP135, cisternal growth was observed. However, when VCIP135(Q381A) was added together with p97/p47, there was no cisternal growth. Other VCIP135 mutants, C218A and C218S, which lack deubiquitinating activities (Figure 3D, lanes 4 and 5) (Wang et al, 2004), also caused no cisternal growth even in the presence of p97/p47. In contrast to the p97/p47 pathway, all mutants, including VCIP135(Q381A), still functioned at the same level as the wild type in the p97/p37 pathway. All of these results indicate that WAC is required specifically for p97/p47-mediated Golgi membrane fusion.

Since the p97/p47 pathway is required only for organelle reassembly at mitosis (Uchiyama et al, 2003), the function of WAC is thought to be necessary for Golgi reassembly at mitosis. To confirm this, we investigated Golgi reassembly at the end of mitosis in WAC-depleted cells (Supplementary Figure S3). In WAC-depleted cells, Golgi reassembly was partially inhibited at cytokinesis (Supplementary Figure S3, panels m–o). This in vivo result supports the requirement of WAC in p97/p47-mediated Golgi reassembly at mitosis.

VCIP135 is required for ER membrane fusion mediated by p97/p47 and p97/p37, although its deubiquitinating activity is not necessary

Although immunofluorescence staining revealed that WAC mainly localizes to the Golgi and the nucleus, some WAC was present in the cytosol (Figure 1B). Biochemical experiments using subcellular fractionation also showed that WAC existed in the cytosol (Supplementary Figure S1A). As VCIP135, a binding partner of WAC, localizes to the ER as well as the Golgi (Uchiyama et al, 2002), there are some possibilities that WAC may also have a role in ER membrane fusion together with VCIP135. On the other hand, the role of VCIP135 in ER biogenesis still remains unclear. We previously reported that VCIP135 also functions in ER biogenesis by showing that the microinjection of anti-VCIP135 antibodies into living cells disrupted ER network structures (Uchiyama et al, 2002). There has been no detailed study since then on its role in ER biogenesis. Therefore, before investigating the requirement of WAC for ER membrane fusion, we first tried to clarify the function of VCIP135 using an in vitro ER reformation assay.

In our in vitro ER reformation assay, the membranes labelled with a fluorescent dye were incubated with cytosol in a narrow space between two coated coverslips (at a distance of 6.0 μm). The coated coverslip allowed the adhesion of the branched portion of the ER network on its surface and stabilized the reformed network structures, which enabled us to measure the degree of network formation. After incubation, the ER networks reformed on the surface of a coverslip were observed and quantified by confocal microscopy without fixation. When membranes were incubated together with cytosol, ER networks were reformed (Figure 7A, panel b). The addition of antibodies to p47, p37 and VCIP135 partially inhibited this ER reformation, and these inhibitions were rescued by quenching the antibodies with their antigens (Figure 7Ac–e and B). These in vitro results show that p47, p37 and VCIP135 function in ER membrane fusion. In contrast, anti-WAC antibodies showed no inhibition (Figure 7Af and B). This is the first report on the in vitro functions of VCIP135 and p37 in ER network formation, although we previously showed their in vivo functions (Uchiyama et al, 2002, 2006).

In the case of Golgi membrane fusion, the p97 pathways require VCIP135, and the p97/p47 pathway especially requires its deubiquitinating function as well as its activator, WAC (Figure 6). In ER membrane fusion, do the p97 pathways also require VCIP135 and its deubiquitinating activity? To answer this question, we performed in vitro ER reformation assays and the results are presented in Figure 8. Endogenous VCIP135, p47 and p37 were removed from the membranes by salt-wash and from the cytosol by the depletion with p97-beads (Hetzer et al, 2001), and recombinant VCIP135, p47 and p37 were added instead. When recombinant VCIP135 was not added, ER reformation was not observed even in the presence of either recombinant p47 or p37 (Figure 8Ab and Cb). When VCIP135wt was added together with either p47 or p37, ER network structures were reformed (Figure 8Ac and Cc). These data indicate that the p97/p47 and p97/p37 pathways require VCIP135 for their ER membrane fusion.

We next tested using VCIP135 mutants (C218S, C218A), which lack deubiquitinating activities (Figure 3D, lanes 4 and 5), whether the deubiquitinating activity of VCIP135 is necessary in these two p97 pathways. Interestingly, even when VCIP135(C218S) was added instead of VCIP135wt together with either p47 or p37, ER network reformation was observed (Figure 8Ad and Cd) and there was no significant difference in the degree of ER network formation between VCIP135wt and its mutants (C218S and C218A) (Figure 8B and D). These results demonstrate that ER membrane fusion mediated by p97/p47 and p97/p37 requires VCIP135 but not its deubiquitinating activity.

We finally investigated the requirement of WAC, an activator of VCIP135 deubiquitinating enzyme, in ER membrane fusion. In an in vitro ER reformation assay, we also tested VCIP135(Q381A), which lacks WAC-binding affinity. As presented in Figure 9A and B, VCIP135(Q381A) was as effective as VCIP135wt in both the p97/p47 and p97/p37 pathways. These results indicate that the binding of VCIP135 to WAC may be unnecessary in ER membrane fusion. Figure 9C shows the results of WAC siRNA experiments. Since ER structures are easily damaged and lost after fixation, we used a stable cell line expressing GFP-tagged HSP47, an ER protein, for in vivo experiments and observed ER structures in living cells without fixation, by confocal microscopy. The depletion of WAC by WAC siRNA caused no obvious morphological changes in the ER structures of living cells (middle and right panels). Additionally, anti-WAC antibodies did not inhibit ER reformation in the in vitro assay (Figure 7Af and B). Considering both in vivo and in vitro results, it can be concluded that WAC is not required for ER membrane fusion, although VCIP135 is involved in p97-mediated ER membrane fusion. This strongly supports the finding that VCIP135 deubiquitinating activity is unnecessary for p97-mediated ER membrane fusion.

Proteins and antibodies

p97, p37, p47, VCIP135 and p115 were prepared as reported previously (Rabouille et al, 1995; Kondo et al, 1997; Uchiyama et al, 2002, 2006). Polyclonal antibodies to p47, p37, p97 and VCIP135 were prepared as described previously (Uchiyama et al, 2002, 2003, 2006). Monoclonal antibodies to p97, His-tag, GM130 and ubiquitin were purchased from Progen, Qiagen, BD Transduction and Santa Cruz, respectively.

Identification and cDNA cloning of WAC

The ORF of VCIP135 was subcloned into the yeast expression vector pGBKT7. The plasmid was transformed into yeast strain AH109 and a yeast two-hybrid screen was performed essentially according to the manufacturer's instructions using a GAL4 DNA activation domain fusion library in pACT2 (mouse 17d embryo MATCHMAKER cDNA Library, Clontech). In all, 5 × 10⁶ colonies were screened and 98 colonies were positive. Two of them encoded WAC. Its full cDNA sequence (BC080851) was identified by an EST database search. The clone containing the full cDNA sequence of WAC was isolated from a rat liver cDNA library (Clontech) by PCR (primers: 5′-CTGCCGCCCCGTAGTTGGCAC-3′, 5′-GTTACACATGGCTTCCATGGCTTC-3′). This clone was sequenced, and a start codon was confirmed: it had a stop codon immediately upstream of the start codon. Its nucleotide sequence is available in the DNA data bank of Japan database under the accession code AB574169.

The ORF of WAC was subcloned into pET22b and pGEX6P-1 for the production of His-tagged and GST-tagged recombinant protein, respectively. His–WAC and GST–WAC were expressed in E. coli and purified with Ni-beads and GSH-beads, respectively, followed by further purification using 5–30% sucrose gradient centrifugation. His-tagged WAC was used to raise rabbit anti-WAC polyclonal antibodies.

Isolation of a VCIP135 mutant, which lacks binding affinity to WAC

VCIP135 mutants were generated as previously described (Reddy and Seaman, 2001; Kaneko et al, 2010). In order to generate PCR products with mutations randomly distributed throughout VCIP135(1–611), the cDNAs of VCIP135(1–312) and (312–611) were amplified by a PCR reaction in which dATP was limited to 10% of the other dNTPs. Full-length cDNAs of VCIP135 in which mutations were randomly introduced at its amino-terminal half were generated using each PCR product by a PCR reaction and then subcloned into pGBKT7 vector. The ORF of WAC was also subcloned into pACT2 vector. Both plasmids were co-transformed into yeast strain AH109 and a yeast two-hybrid screen was performed.

Biochemical experiments

Binding experiments were carried out in buffer A (20 mM Hepes, 1 mM MgCl₂, 1 mM ATP, 1 mM DTT, 10% glycerol, pH 7.4) containing 0.15 M KCl and 1% Triton X-100. For immunoprecipitation experiments using Golgi membranes, highly purified Golgi membranes (Hui et al, 1998) (0.45 mg) were solubilized in 100 μl of buffer A containing 0.1 M KCl, 0.5% Triton X-100 and protease inhibitor cocktail (EDTA-free, Roche), and mixed with polyclonal antibodies to WAC (3 μg). WAC and its binding proteins were precipitated by protein G-beads.

The tissue distribution of WAC was studied with denaturing immunoprecipitation as previously described (Uchiyama et al, 2003). Each rat tissue (10 mg wet) was homogenized and extracted, followed by immunoprecipitation using an excess amount of anti-WAC antibodies (1 μg). The amounts of WAC in the precipitates were estimated by western blotting.

The deubiquitinating activity of VCIP135 was assayed basically according to the method of Wang et al (2004). Briefly, VCIP135 and the indicated proteins were incubated with a mixture of oligo-ubiquitin chains (UC-230, BostonBiochem) in 20 μl of buffer A containing 0.1 M KCl and 1 mg/ml BSA for 4 h at 30°C. Aliquots of the reaction mixture were analysed by western blotting using a monoclonal antibody to ubiquitin.

The deubiquitinating functions of VCIP135 and its mutants were also tested using the extract from mitotic Golgi fragments, as follows. Mitotic Golgi fragments (15 μg) were prepared as previously reported (Uchiyama et al, 2003) and washed with 1 M KCl, followed by solubilization with 0.5% CHAPS. The extracts were incubated together with either VCIP135 or its mutants (1.8 μg) in 25 μl of buffer A containing 0.1 M KCl and 0.5% CHAPS for 2 h at 30°C. The amounts of generated monoubiquitin were estimated by western blotting using a monoclonal antibody to ubiquitin.

Knockdown of WAC by siRNA

WAC was targeted with independent siRNAs (B-Bridge International): 5′-CCAGUUACUCUCCACAAGA -3′ (WAC siRNA1), 5′-CCAGUGGAAUGGAAGACAA-3′ (WAC siRNA2). Each 15 nM siRNA duplex was transfected into either HeLa or NRK cells using lipofectamine RNAimax (Invitrogen).

Microscopic studies on cultured cells

For observation of Golgi structures, cells grown on coverslips were fixed with 3% PFA/PBS for 5 min, permeabilized with 0.1%. saponin for 2 min and stained with polyclonal antibodies to either WAC or VCIP135 and monoclonal antibodies to GM130. For observation of ER structures, we used a stable Hela cell line expressing GFP-tagged HSP47, an ER protein (Uchiyama et al, 2002). ER structures were observed in living cells, without fixation, using confocal microscopy.

HeLa cells grown on coverslips were fixed 32 h after WAC siRNA treatment, embedded into Epon, ultrathin sectioned and observed by an electron microscope (JEM-1230, JEOL). The Golgi area was defined by the boundary enclosing the Golgi stacks, tubulo-reticular networks and tubules and all the vesicles that were within 100 nm of these membranes. Membrane profiles in the Golgi area were divided into three categories: cisternae, vesicles and tubules. The relative proportion of each category of membranes was counted using an intersection method, as previously described (Shorter et al, 1999).

In vitro Golgi reformation assay

The in vitro Golgi reformation assay was performed as reported previously (Shorter et al, 1999). All proteins added in this assay were prepared as recombinant proteins from E. coli.

In vitro ER reformation assay

Xenopus egg cytosol and membrane were prepared according to the method of Hetzer et al (2001) with the following modifications. The light membrane fraction was stained with 0.05 mg/ml CM-DiI (Molecular Probes), diluted with 20 vol. of buffer B (50 mM KCl, 20 mM HEPES, 2.5 mM MgCl2, pH 7.4) containing 0.25 M sucrose, and centrifuged onto a cushion of buffer B containing 1 M sucrose at 20 000 g for 30 min at 4°C. The membrane was suspended in buffer B with 0.5 M sucrose, aliquoted and stored at −80°C. For the preparation of salt-washed membrane, membranes were washed with 0.5 M KCl immediately before the assay. HeLa cytosol was prepared as reported previously (Rabouille et al, 1995). The cytosol was dialysed against buffer B containing 0.25 M sucrose and 1 mM DTT, aliquoted and stored at −80°C. To remove p97-binding proteins, the desalted Xenopus egg cytosol (40 mg) was depleted with p97-beads (100 μg p97) three times. Neither p47, p37 nor VCIP135 could be detected in the depleted cytosol by western blotting.

Membrane (not salt-washed, 20 μg: salt-washed, 10 μg), cytosol (60 μg) and indicated proteins were mixed in 10 μl of reaction buffer B containing 0.25 M sucrose, 1 mM DTT, 0.2 mM GTP and ATP regenerating system (1 mM ATP, 20 mM creatine phosphate, 1 mg/ml creatine kinase). The reaction mixture was incubated for 2 h at 25°C in a narrow space between two coverslips. The coverslips coated with polyethyleneimine were used to stick the branched portion of the reformed ER network on their surfaces. Some silica-based spacer beads (6.0 μm in diameter, Ube-Nitto Kasei, Japan), which are usually used in liquid crystal displays and semiconductor chips, were also added into the reaction mixture to control the precise distance between the two coverslips. After incubation, the ER network reformed on the surface of a coverslip was observed by confocal microscopy (LSM510, Zeiss) without fixation. Obtained images were analysed using Image-Pro Plus software (Media Cybernetics).