Squalene monooxygenase (SM) is an essential rate-limiting enzyme in cholesterol synthesis. SM degradation is accelerated by excess cholesterol, and this requires the first 100 amino acids of SM (SM N100). This process is part of a protein quality control pathway called endoplasmic reticulum-associated degradation (ERAD). In ERAD, SM is ubiquitinated by MARCH6, an E3 ubiquitin ligase located in the endoplasmic reticulum (ER). However, several details of the ERAD process for SM remain elusive, such as the extraction mechanism from the ER membrane. Here, we used SM N100 fused to GFP (SM N100-GFP) as a model degron to investigate the extraction process of SM in ERAD. We showed that valosin-containing protein (VCP) is important for the cholesterol-accelerated degradation of SM N100-GFP and SM. In addition, we revealed that VCP acts following ubiquitination of SM N100-GFP by MARCH6. We demonstrated that the amphipathic helix (Gln62–Leu73) of SM N100-GFP is critical for regulation by VCP and MARCH6. Replacing this amphipathic helix with hydrophobic re-entrant loops promoted degradation in a VCP-dependent manner. Finally, we showed that inhibiting VCP increases cellular squalene and cholesterol levels, indicating a functional consequence for VCP in regulating the cholesterol synthesis pathway. Collectively, we established VCP plays a key role in ERAD that contributes to the cholesterol-mediated regulation of SM.
Squalene monooxygenase (SM) catalyses the second rate-limiting step of cholesterol synthesis , downstream of the first rate-limiting step catalysed by 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR). SM has recently been implicated in cancers [2–4] and was previously investigated as a target for lowering cholesterol levels [5–7]. Post-translationally, both yeast SM (commonly referred to as Erg1) and human SM are degraded by the ubiquitin-proteasome system under excess sterols [1,8,9]. SM is ubiquitinated and degraded by MARCH6 [8,10], an E3 ubiquitin ligase located in the endoplasmic reticulum (ER) containing 14 transmembrane domains . The degradation of SM by MARCH6 is conserved in yeast, where Doa10p, the mammalian ortholog of MARCH6 , ubiquitinates Erg1 . Doa10p is an important E3 ubiquitin ligase that ubiquitinates ER proteins in a process called endoplasmic reticulum-associated degradation (ERAD) [13,14]. Using select substrates, the yeast Doa10p pathway has been examined in great detail [13–18]. The yeast Doa10p and mammalian MARCH6 pathways have been presumed to be similar, but the MARCH6 degradation pathway is still largely unexplored, raising the question of whether the mechanisms are indeed entirely conserved.
Valosin-containing protein (VCP) mediates the degradation of ubiquitinated ERAD substrates . It is an AAA+-type ATPase with substrate threading being a proposed mechanism for how VCP can pull proteins out of the membrane and unfold them [20,21]. In yeast, the VCP homologue Cdc48 carries out this role . The role of VCP in ERAD has been demonstrated to require several protein cofactors, including deubiquitinases (DUBs) . However, the role of these proteins and processes have predominantly been examined for other mammalian ERAD E3 ubiquitin ligases such as Hrd1 and gp78 [24–27]. Thus, there is a need to address how the MARCH6 ubiquitination complex operates and whether these general proteins could also be involved in regulating MARCH6 substrates, including SM, a canonical substrate, used to measure MARCH6 ubiquitinating activity .
The cholesterol-accelerated degradation of SM requires the first 100 amino acids (SM N100) and ubiquitination by MARCH6 [1,8]. We proposed that cholesterol triggers conformational changes within the re-entrant loop and amphipathic helix of SM N100 [28,29]. These observations suggest that SM N100 belongs to the class of ERAD-M and ERAD-C substrates. ERAD-M substrates have misfolded regions in the membrane while ERAD-C substrates have misfolded regions in the cytosol. Consistent with this, Doa10p has been reported to target both ERAD-M and ERAD-C substrates [14,15,30,31].
Here, we aimed to investigate the dislocation mechanism of SM, which is an important step in ERAD . We used SM N100-GFP as a model degron which undergoes cholesterol-accelerated degradation and is ubiquitinated by MARCH6 for proteasomal degradation [1,8]. Inhibiting or knocking down VCP reduced the cholesterol-accelerated degradation of SM N100-GFP and SM. Ataxin-3 , USP13  and YOD1  are reported VCP-associated DUBs that regulate ERAD substrates but knocking down these DUBs did not affect SM N100-GFP cholesterol-accelerated degradation. Depleting MARCH6 reduced the stabilising effects of VCP inhibition, indicating that ubiquitination is a requirement for VCP-mediated regulation of SM N100-GFP. Furthermore, deleting or replacing the amphipathic helix region (residues Gln62–Leu73) of SM N100 affected regulation of SM N100-GFP by MARCH6 and VCP. VCP inhibition increased SM levels and abrogated the cholesterol-mediated decrease of SM, and also increased flux through the cholesterol synthesis pathway. Overall, we revealed that VCP is involved in regulating the cholesterol-accelerated degradation of SM N100-GFP and SM, which is a first for a MARCH6 substrate, other than the autodegradation of MARCH6 itself .
Materials and methods
Inhibitors used in this study were MG132 (Sigma–Aldrich, C2211), CB-5083 (Cayman Chemical, 19311), PR-619 (Cayman Chemical, 16276) and WP1130 (Cayman Chemical, 15227). Cholesterol-methyl-β-cyclodextrin (Chol/CD) (Sigma–Aldrich, C4951) was dissolved in water to 2 mg/ml and frozen in dry ice. The frozen Chol/CD was dried down with a SpeedVac vacuum concentrator (Thermo Fisher Scientific) and kept at −20°C for storage. Chol/CD was reconstituted to 2 mg/ml in water prior to use. All siRNAs and primers were obtained from Sigma–Aldrich.
The wild type pTK-SM N100-GFP-V5 and pTK-SM-V5 constructs were generated previously  as were the amphipathic helix deletion and replacement constructs of pTK-SM N100-GFP-V5 . The amphipathic helix deletion construct of pTK-SM-V5 was generated in this study using the polymerase incomplete primer extension method .
Cells were maintained as monolayers at 37°C in 5% CO2. HEK-293 Flp-InTM T-RExTM cells stably expressing SM N100-GFP-V5 (HEK-SM N100-GFP-V5)  were cultured and maintained in DMEM (high glucose) supplemented with 10% (v/v) foetal calf serum, 200 µg/ml of hygromycin B, penicillin (100 units/ml) and streptomycin (100 µg/ml). HEK-293T, HEK-293 Flp-In™ T-Rex™ and HEK-MARCH6-CRISPR  cells were cultured and maintained in DMEM (high glucose) supplemented with 10% (v/v) foetal calf serum, penicillin (100 units/ml) and streptomycin (100 µg/ml). CHO-7 cell lines (gifts of Drs. Goldstein and Brown, University of Texas Southwestern) were maintained in DMEM/F-12 media supplemented with 5% (v/v) newborn calf lipoprotein-deficient serum, penicillin (100 units/ml), and streptomycin (100 µg/ml).
Prior to transfection, cells were refreshed with 10% (v/v) foetal calf serum / DMEM (high glucose) medium without antibiotics. For plasmid transfections, HEK-293T, HEK-293 Flp-In™ T-Rex™ and HEK-MARCH6-CRISPR cells were plated in 12-well plates and transfected with 1 µg of expression plasmid DNA using 2 µl of P3000 supplemental reagent and 1.5 µl Lipofectamine 3000 (Thermo Fisher Scientific). For siRNA transfections, HEK-SM N100-GFP-V5 cells were plated in 12-well plates and transfected with 25 nM siRNA using 2 µl Lipofectamine RNAiMAX (Thermo Fisher Scientific). For plasmid and siRNA co-transfections, HEK-239T cells were plated in 12-well plates and transfected with 1 µg of expression plasmid DNA and 25 nM siRNA using 4 µl of Lipofectamine 2000 (Thermo Fisher Scientific). All transfections were carried out for 24 h before cells were washed with PBS and medium was changed for further experiments. CHO-7 transfections for SM-V5 were carried out as described previously .
Cholesterol regulation and inhibitor treatments
HEK-SM N100-GFP-V5 cells were grown for 24 h in 12-well plates. If required, cells were then transfected with siRNA for 24 h. After transfection or 24 h after plating, cells were pre-treated overnight in 10% (v/v) foetal calf lipoprotein-deficient serum/DMEM (high glucose) medium (referred to as sterol-depleted media in Figure legends). All sterol-depleted media condition contained compactin (5 µM) and mevalonate (50 µM) except Figure 7. Cells were then treated for 8 h with Chol/CD (20 µg/ml) before cells were collected for Western blot analyses. Growth medium, concentrations and incubation time for the inhibitors CB-5083, MG132, PR-619 and WP1130 are included in the Figure legends. Testing of SM-V5 cholesterol regulation in CHO-7 cells was carried out as previously described .
Quantitative real-time PCR
Cells were grown in triplicates in 12-well plates for each condition. RNA was extracted using TRI Reagent (Sigma–Aldrich). cDNA was synthesised using 1 µg of RNA using the SuperScript III First-Strand Synthesis System (Thermo Fisher Scientific). Quantitative real-time PCR of synthesised cDNA was carried out using SensiMix™ SYBR® No-ROX Kit (Bioline) on the Rotor-Gene Q (Qiagen). The mRNA expression levels were determined after normalising to GAPDH and made relative to control siRNA condition using the ΔΔCT method. GAPDH , MARCH6  and VCP  primers for quantitative real-time PCR have been described previously. Real-time primers for Ataxin-3 (forward, GCTCACTTTGTGCTCAACATTG, and reverse, CCTGAGCCTCTGATACTCTGG), USP13 (forward, CGGGTGTGAAAACATCTCGC, and reverse CCTCTCCTGGCTGTAACCC), YOD1 (forward, CATCCAATCTGGTGACATGCTG and reverse, GCACAAGCTGGATTCAAGACTC) and Ube2J2 (forward, CGCAGCGACAGGGAGAGATGA, and reverse, CTCTGGGCCTCGGACGACATAG) were made in this study.
Cells were grown in 12-well plates on cover slips for 24 h and transfected the next day for another 24 h. CHO-7 cells were transfected with 0.05 µg DsRed-ER (Clontech) and 1 µg pTK-SM N100-GFP-V5 or GFP-V5. HEK-SM N100-GFP-V5 cells were transfected with 0.10 µg DsRed-ER for 24 h. After transfection, cells were grown in sterol-depleted media containing compactin (5 µM) and mevalonate (50 µM) overnight. Cells were fixed with 4% (v/v) paraformaldehyde and mounted on glass slides with ProLong Gold Antifade Mountant (Thermo Fisher Scientific). Mounted slides were visualised using a Leica TCS SP8 DLS with a 63× Plan Apo oil-immersion objective (NA = 1.4). Fluorescence signals of GFP (excitation 488 nm/Ar laser) and DsRed-ER (561 nm/DPPS laser) were detected with sequential scanning.
Gas chromatography–mass spectrometry
GC–MS was carried out as described previously . Cells were pre-treated in sterol-depleted media overnight and treated the next day for 8 h with Chol/CD (20 µg/ml) and 0.1% (v/v) DMSO or 5 µM CB-5083. Cells were lysed in NaOH (0.1 M), saponified, and lipids were extracted with hexane, then dried down with a SpeedVac vacuum concentrator (Thermo Fisher Scientific). Lipids were suspended in BSTFA and analysed with a Thermo Trace gas chromatograph coupled with a Thermo DSQIII mass spectrometer and Thermo Triplus autosampler (Thermo Fisher Scientific) with the protocols described previously . Samples were analysed in selective ion monitoring mode for squalene and the internal standard, 5α-cholestane. Data acquisition and analyses were performed using Thermo Xcalibur software.
Cells were treated as described in Figure legends. Cells were harvested and lysed in modified RIPA buffer (50 mM Tris–HCl pH 8.0, 150 mM NaCl, 0.1% (w/v) SDS, 1.5% (w/v) IGEPAL CA-630, 0.5% (w/v) sodium deoxycholate, and 2 mM MgCl2). The 17 000×g supernatant was used for bicinchoninic acid and cholesterol assay. Cholesterol content was measured using the Amplex Red Cholesterol Assay Kit (Thermo Fisher Scientific) and normalised to protein content, determined using the Pierce Bicinchoninic Acid Protein Assay Kit (Thermo Fisher Scientific). Fluorescence was measured with a CLARIOstar plate reader (BMG LabTech) with an excitation wavelength of 544 nm and an emission wavelength of 590 nm.
Western blotting was carried out as described previously [28,29]. Briefly, cells were lysed in 2% SDS lysis buffer (2% (w/v) SDS, 10 mM Tris–HCl, pH 7.6, 100 mM NaCl) containing 2% (v/v) protease inhibitor cocktail (Sigma–Aldrich), passed 20 times through a 21-gauge needle, vortexed for 20 min at room temperature, and normalised in lysis buffer containing 1 × Laemlli buffer. Samples were boiled for 5 min and equal protein amounts were loaded for SDS–PAGE followed by transfer to nitrocellulose membranes. Membranes were blocked with 5% (w/v) skim milk dissolved in PBST at room temperature for 1 h. Membranes were incubated with the indicated antibodies; anti-V5 (1 h incubation at room temperature with 1 : 5000 dilution in 5% (w/v) BSA/PBST; Thermo Fisher Scientific, R960–25), anti-β-tubulin (overnight incubation at 4°C with 1:5 000 dilution in 5% (w/v) BSA/PBST; Abcam, ab6046), anti-α-tubulin (1 h incubation at room temperature with 1 : 20 000 dilution in 5% (w/v) BSA/PBST; Sigma–Aldrich, T5168), anti-ubiquitin P4D1 (overnight incubation at 4°C with 1 : 5000 dilution in 5% (w/v) BSA/PBST; Santa Cruz Biotechnology, sc-8017), anti-SQLE (overnight incubation at 4°C with 1 : 2500 dilution in 5% (w/v) BSA/PBST; Proteintech, 12544-1-AP). After incubations with primary antibodies, blots were washed three times with PBST. Blots were then incubated with IRDye® 680RD donkey anti-rabbit IgG (1 h incubation at room temperature with 1 : 10 000 dilution in 5% (w/v) BSA/PBST; LI-COR, 925–68073) and IRDye® 800CW donkey anti-mouse IgG (1 h incubation at room temperature with 1 : 10 000 dilution in 5% (w/v) BSA/PBST; LI-COR, 926–32212). Membranes were then visualised using the Odyssey CLx (LI-COR). Western blots were quantified by densitometry using Image Studio Lite (version 5.2.5). Locations of molecular mass standards or size estimates acquired from ImageJ (v1.52p) are indicated on the blots.
The number (n) of independent experiments are presented in the Figure legend. Densitometry data from at least three independent experiments are presented as bar graphs. Densitometry data in bar graphs are presented as mean + S.E.M. Statistical differences were determined by the Student's paired t-test (two-tailed), where P values of P < 0.05 (*) and P < 0.01 (**) were considered statistically significant.
Relative protein levels for SM N100-GFP-V5, SM-V5 and endogenous SM were determined by normalising to wild type, DMSO, control siRNA or HEK-293 Flp-In™ T-Rex™ cells, which was set to 1. For instance, a value of 2 would indicate double the protein level, compared with wild type construct, DMSO, control siRNA and HEK-293 Flp-In™ T-Rex™ cells.
Cholesterol regulation was defined as the proportion of SM N100-GFP-V5, SM-V5 or endogenous SM degraded for each condition, normalised to the proportion of SM N100-GFP-V5, SM-V5 or endogenous SM degraded for wild type, DMSO or control siRNA, which were all set to 1. Values approaching 0 on the cholesterol regulation scale indicate little or no cholesterol regulation, which means no degradation in the presence of excess cholesterol. A value closer to 1 would indicate similar levels of degradation in the presence of excess cholesterol when compared with wild type, DMSO or control siRNA.
The relative MARCH6 response was determined as a relative fold-increase compared with wild type. After densitometric analyses, the fold-increase for each mutant construct after MARCH6 siRNA knockdown was normalised to the fold-increase for the wild type SM N100-GFP-V5 or SM-V5 construct, which was set to 1. Values approaching 0 on the MARCH6 response scale indicate less of a rescue compared with wild type when MARCH6 was knocked down. A value closer to 1 would indicate the MARCH6 knockdown rescue effect on SM-V5 or SM N100-GFP-V5 mutants is like that of wild type. The relative CB-5083 response was also presented in the same manner as the relative MARCH6 response. Values approaching 0 on the CB-5083 response scale indicate less of a rescue of SM N100-GFP-V5, SM-V5 or endogenous SM, when compared with control siRNA, wild type or HEK-293 Flp-In™ T-Rex™ cells.
VCP regulates the cholesterol-accelerated degradation of SM and SM N100-GFP
VCP is a core component of ERAD that mediates the extraction of a number of ERAD substrates from the ER, enabling proteasomal degradation , but this widely generalised ERAD component has yet to be tested for SM. To test if VCP is necessary for ERAD of SM, we utilised SM N100-GFP, an ER-localised fusion protein (Figure 1) containing the first 100 amino acids of SM (SM N100) fused to GFP, which recapitulates the cholesterol-accelerated degradation of SM . This construct has been utilised in our previous studies to understand how the SM N100 degron functions in enabling cholesterol-accelerated degradation of SM [8,28,29].
SM N100-GFP-V5 is an ER-localised protein.
We used HEK-SM N100-GFP-V5 cells  which stably express the SM N100-GFP fusion protein, and co-treated these cells with cholesterol and the VCP inhibitor, CB-5083. This inhibitor has been used to identify endogenous ERAD substrates in a quantitative ubiquitin proteomics strategy . CB-5083 blocked the cholesterol-accelerated degradation of SM N100-GFP (Figure 2a,b). MG132, a proteasome inhibitor, also reduced the cholesterol-accelerated degradation although it was not as effective as VCP inhibition (Figure 2a,b). MG132 has been shown to block the cholesterol-mediated decrease of endogenous SM [1,8]. Here, we observed a similar effect with CB-5083 blunting the cholesterol-mediated decrease of endogenous SM (Figure 2c,d).
VCP inhibition stabilises SM N100-GFP and SM and reduces cholesterol-accelerated degradation.
We repeated the experiments by inhibiting VCP and the proteasome prior to cholesterol treatment. These pre-treatments were first carried out at lower concentrations of CB-5083 (1 µM) and MG132 (5 µM), before the cells were co-treated with the inhibitors and cholesterol the next day. We observed a greater stabilisation of SM N100-GFP with both inhibitors (Figure 2e,f) than the previous experiments (Figure 2a,b). Likewise, CB-5083 increased endogenous SM protein levels and blocked the cholesterol-mediated decrease (Figure 2g,h) to a greater extent than the previous experiments (Figure 2c,d).
To complement the VCP inhibitor findings, we depleted VCP with siRNA (Figure 3a), which resulted in a 3- to 4-fold increase in SM N100-GFP protein levels and blunting of cholesterol regulation (Figure 3b,c). In contrast, MARCH6 knockdown increased SM N100-GFP-V5 protein levels by ∼8-fold. VCP knockdown also stabilised endogenous SM and prevented its cholesterol-mediated decrease (Figure 3d,e). Collectively, these data indicate a role for VCP in ERAD of SM and SM N100-GFP, consistent with its role in extracting ERAD substrates from the ER [19,39].
VCP depletion stabilises SM N100-GFP and SM and reduces cholesterol-accelerated degradation.
DUB inhibition alters protein levels of SM and SM N100-GFP
VCP carries out its role in ERAD with various other associated cofactors [39–41]. UBXD8 is a ubiquitin-binding cofactor of the VCP complex , and our previous study demonstrated that siRNA-mediated knockdown of UBXD8 blunted SM N100-GFP cholesterol-accelerated degradation . We hypothesised that disrupting some of these VCP cofactors would impair SM N100-GFP degradation.
We turned our attention to DUBs, as VCP associates with DUBs to regulate the degradation of other ERAD substrates [27,39,44]. The broad-spectrum DUB inhibitor, PR-619, increased SM N100-GFP levels without any changes in cholesterol regulation (Figure 4a,b). Intriguingly, PR-619 treatment decreased endogenous SM protein levels, without affecting cholesterol regulation (Figure 4c,d). Using a more specific DUB inhibitor, WP1130, which inhibits DUBs beyond those reported to associate with VCP [23,39,45], we observed no changes in SM N100-GFP protein levels but significant loss of cholesterol regulation (Figure 4e,f). This loss of cholesterol regulation may be due to WP1130 inducing formation of aggresomes containing ubiquitinated misfolded proteins , as suggested by the increased ubiquitin levels (Figure 4e). WP1130 decreased endogenous SM protein levels and blunted cholesterol regulation (Figure 4e and Supplementary Figure S1).
SM N100-GFP and SM are regulated by deubiquitinases.
Next, we focused on more specific targets, the VCP-associated DUBs. We knocked down VCP-associated DUBs identified as direct or indirect regulators of other ERAD substrates, including Ataxin-3 [33,44], USP13  and YOD1  (Figure 4g), but observed no changes in SM N100-GFP and endogenous SM protein levels or cholesterol regulation (Figure 4h, i and Supplementary Figure S2). VCP associates with many DUBs beyond those reported to be linked to ERAD [23,39] and we cannot rule out the possibility that other DUBs regulate SM during cholesterol-accelerated degradation. This may require alternative approaches beyond targeted knockdowns or knockouts.
VCP regulates SM N100-GFP in a MARCH6-dependent manner
The dedicated cofactors of VCP allow the VCP complex to bind to ubiquitinated substrates, consequently dislodging them from the ER . We hypothesised that VCP inhibition would not further stabilise SM N100-GFP if ubiquitination activity is absent. SM is a well-established substrate that is ubiquitinated and degraded by the E3 ubiquitin ligase MARCH6, as shown by independent groups [8–10]. One reported output for verifying knockout of MARCH6 in cell lines is by detecting an increase in SM levels, as reliable MARCH6 antibodies have yet to be reported [10,35].
Following MARCH6 knockdown (Figure 5a), VCP inhibition stabilised SM N100-GFP to a lesser extent than when compared with the control knockdown (Figure 5b,c). Furthermore, knocking down Ube2J2 (Figure 5a), the E2 ubiquitin conjugating enzyme of SM and SM N100-GFP [10,47,48], also showed a blunting of CB-5083 mediated stabilisation (Figure 5b,c). This supports our hypothesis that VCP regulation of SM N100-GFP is dependent on the existence of the ubiquitination machinery, notably MARCH6 and Ube2J2.
VCP regulates SM N100-GFP and SM in a MARCH6-dependent manner.
We also tested the effects of CB-5083 on transiently transfected SM N100-GFP in a MARCH6-depleted cell line which was generated previously using CRISPR-Cas9 by our laboratory . Here, we also observed the lack of CB-5083-mediated stabilisation in the MARCH6-depleted cells (Figure 5d,e), which provides additional evidence that VCP regulates SM N100-GFP downstream of MARCH6, most likely requiring MARCH6 ubiquitination activity. Moreover, our findings with SM N100-GFP are supported by observations that endogenous SM showed a lack of rescue with CB-5083 when MARCH6 is knocked down or knocked out (Figure 5b,d and Supplementary Figure S3).
The amphipathic helix of SM N100 enables regulation of SM N100-GFP by MARCH6 and VCP
Having established that VCP is likely to act on SM N100 downstream of ubiquitination by MARCH6 and Ube2J2 (Figure 5), we focused on the SM N100 degron itself and how it is regulated by both MARCH6 and VCP. One study showed that MARCH6 recognises the 16-amino acid CL1 degron. These residues form an amphipathic helix associated with the ER membrane . Doa10p also recognises structural features in degrons including hydrophobic amino acid sequences, amphipathic helices and intramembrane regions [13,15,49].
We hypothesised that the hydrophobic amphipathic helix of SM N100 (Gln62–Leu73)  may be recognised by MARCH6, controlling the degradation rate of SM N100-GFP in a VCP-mediated process. Here, we utilised a series of previously cloned SM N100-GFP constructs to investigate the 12-amino acid stretch of SM N100 (Gln62–Leu73). These contain variations of the amphipathic helix including helix deletion (Δ62–73), replacement with a charged helix from myoglobin (Myo), and replacement with hydrophobic re-entrant loops from ghrelin O-acyltransferase (GOAT) and hedgehog acyltransferase (HHAT) [50–52].
Deleting the amphipathic helix (Δ62–73) resulted in SM N100-GFP becoming less responsive to MARCH6 knockdown (Figure 6a), consistent with our previous reports in the CHO-7 cell line of hamster origin . Replacing the amphipathic helix with a charged helix from myoglobin (Myo) also blunted the stabilising effects from MARCH6 knockdown (Figure 6a). The hydrophobic re-entrant loops from GOAT and HHAT resulted in lower SM N100-GFP protein expression than wild type (Figure 6a), which was also observed in CHO-7 cells . Interestingly, the GOAT construct was more responsive to MARCH6 knockdown, whereas the HHAT construct was not, suggesting that hydrophobicity alone was not enough for enabling regulation by MARCH6 (Figure 6a). Next, we tested if these mutants could be rescued with VCP inhibition. The SM N100-GFP amphipathic helix deletion (Δ62–73) and myoglobin (Myo) constructs were not as responsive to VCP inhibition as wild type (Figure 6b). On the other hand, the SM N100-GFP hydrophobic re-entrant loop constructs (GOAT and HHAT) were stabilised by VCP inhibition to a greater extent than wild type (Figure 6b).
Regulation of SM N100-GFP by VCP and MARCH6 is dependent on its amphipathic helix.
We next questioned if the amphipathic helix could also influence full-length SM regulation by VCP and MARCH6. Unlike SM N100-GFP, the ectopic SM amphipathic helix deletion mutant (Δ62–73) could still be regulated by VCP and MARCH6 to a similar degree as wild type (Figure 6c,d). We tested the cholesterol-responsiveness of this mutant (Δ62–73) in CHO-7 cells, which allow flexibility in manipulating cell cholesterol status as they can be grown in lipoprotein-deficient serum . We showed that the amphipathic helix deletion mutant exhibits blunted cholesterol regulation (Supplementary Figure S4), supporting the model that this amphipathic helix is required for cholesterol-sensing . Thus, while SM N100 is a transferrable cholesterol-responsive degron, its inherent properties in regulating protein stability is impacted by what it is fused to, with differences between GFP or the native catalytic domain of SM.
Our findings suggested that the hydrophobic re-entrant loops strongly promoted instability, indicating an increased rate of ERAD for the hydrophobic variants of SM N100-GFP. Taken together, our data revealed that the regulation of the SM N100 degron by MARCH6 and VCP is strongly affected by the nature of the region encompassing the amphipathic helix.
VCP inhibition has functional consequences on the cholesterol synthesis pathway
We hypothesised that the cholesterol and VCP-mediated ERAD of SM allows accumulation of its substrate squalene. To test this, we co-treated cells with cholesterol to decrease SM levels, accompanied with CB-5083 treatment to block this decrease. SM N100-GFP and endogenous SM levels were greatly increased with CB-5083 treatment (Figure 7a), consistent with our earlier data (Figure 2).
VCP inhibition increases squalene and cholesterol levels.
We observed that VCP inhibition increased squalene levels by 2-fold (Figure 7b,c). We asked if this could impact overall cholesterol levels. To measure cholesterol levels, we excluded exogenous cholesterol addition in the subsequent experiments. Treating cells with CB-5083 stabilised SM N100-GFP and endogenous SM (Figure 7d), accompanied by an increase of ∼15% in total cholesterol levels (Figure 7e).
The increase of squalene may be due to increased HMGCR activity upstream in the cholesterol synthesis pathway, as VCP is also required for ERAD of HMGCR [54,55]. This observation mirrors that of our previous work showing that MARCH6 knockdown increases SM levels, and yet squalene levels increase, due to elevated HMGCR protein levels upon MARCH6 knockdown .
A more systematic approach analysing the abundance of enzymes and intermediates in the mevalonate pathway, before and after squalene, is required to better reconcile the contribution of VCP inhibition on flux through the pathway. However, our current result supports the notion that modulating VCP has functional consequences on lipid metabolism, in this case, the cholesterol synthesis pathway [32,56].
Our study has focused on a critical process in ERAD, the dislocation mechanism for the model MARCH6 substrate, SM N100-GFP, which constitutes the regulatory domain of SM, a key flux controlling enzyme in cholesterol synthesis. Our results showed that VCP regulates SM through the SM N100 degron in a MARCH6-dependent manner. We examined the SM N100 degron in more detail and found that the amphipathic helix region is an intrinsic feature of SM N100 that enables ERAD of SM N100-GFP by MARCH6 in a VCP-mediated process.
The SM N100 degron contains a membrane embedded re-entrant loop and an amphipathic helix [28,29]. These membrane embedded regions are probably subjected to protein quality control from the moment SM N100 is synthesised. We showed that the amphipathic helix is an important structural feature allowing MARCH6 regulation, with hydrophobicity being a key determinant in ERAD of SM N100-GFP (Figure 6). As this helix becomes disordered in response to cholesterol excess , MARCH6 may recognise structural changes in the loosely membrane-associated amphipathic helix. Consistent with this idea, deleting the amphipathic helix or replacing it with a charged helix of myoglobin reduced the rescue response of SM N100-GFP (Figure 6a,b).
Recently, the term ‘mallostery’ has been coined to describe specific and reversible ligand binding to yeast Hmg2, causing structural transitions and regulation by the ERAD machinery [57,58]. To date, it is unknown where cholesterol binds to SM N100, and the strength of this interaction is unknown. Unsaturated fatty acids increase SM N100-GFP stability by blunting MARCH6 interaction , while plasmalogens enhance SM degradation by augmenting MARCH6 interaction . This model raises new questions, but it could be that unsaturated fatty acids shield the SM N100 hydrophobic amphipathic helix while plasmalogens expose the hydrophobic patch for MARCH6 detection. This is distinct from the highly specific potency of ligands as described by the ‘mallostery’ phenomenon, but it certainly strengthens the idea that misfolding can occur in synthesised folded proteins when exposed to certain conditions, such as changing lipid levels to affect feedback regulation.
VCP interacts with a number of DUBs, and this affects the degradation of ERAD substrates [19,27,34,44,60]. It is possible that our siRNA-mediated knockdown may be insufficient to yield a significant change (Figure 4). We note other studies reporting subtle effects on model ERAD substrates with DUB knockdowns: USP13 knockdown stabilised TCRα-GFP, a model ERAD substrate, by 1.5- to 2-fold while Ataxin-3 knockdown decreased TCRα-GFP by only 20% . Another group showed Ataxin-3 knockdown had no apparent effect on TCRα, but overexpressing the catalytically inactive C14A Ataxin-3 mutant stabilised TCRα . One study showed YOD1 knockdown had no effect on TCRα-GFP , but another showed overexpressing the catalytically inactive C160S YOD1 mutant stabilised TCRα . The precise roles of DUBs in ERAD are still unclear, but it is likely that SM N100 is deubiquitinated to some extent as the trimming of polyubiquitin chains may facilitate processing by VCP [20,23].
Significantly, SM was detected in a large-scale proteomic screen of endogenous ERAD substrates  in the membrane fraction of HepG2 liver cells, treated with the same VCP inhibitor, CB-5083, used in the current study. However, the tryptic SM peptide identified, which centred on ubiquitinated Lys-90, was not enriched after VCP inhibition. Thus, SM failed to meet the criterion for a candidate ERAD substrate in that study . But perhaps Lys-90 ubiquitination is not needed for ERAD of SM. Indeed, we have shown that Lys-90 and the other four lysines present in SM N100 are not required for cholesterol-dependent turnover, whereas four serine residues are (Ser-59, Ser-61, Ser-83, Ser-87); likely by serving as non-canonical ubiquitination sites . In fact, ubiquitination at Ser-83 was confirmed by mass spectrometric analysis . Ser-83 occurs on the same tryptic peptide as Lys-90 (S83PPES87ENK90EQLEAR), and it is likely that VCP inhibition may accumulate other SM peptides containing multiple ubiquitinated amino acids including serine, which would require further validation by mass spectrometry.
Insights into the mechanistic regulation of MARCH6 substrates are still limited in comparison with the other ERAD E3 ubiquitin ligases such as yeast Hrd1p [61,62] and Doa10p [16,62,63], or mammalian gp78 [64,65] and Hrd1 [66–68]. Here, we showed that the cholesterol-accelerated degradation of SM N100-GFP and SM requires VCP, and that regulation of the SM N100 degron by MARCH6 and VCP in ERAD is dependent upon a hydrophobic amphipathic helix (Figure 8). The conservation of MARCH6 and Doa10p ubiquitination complexes have been presumed to be similar, and we showed that VCP is one such conserved component, although the role of DUBs requires further investigation. Degrons have also been thought to possess similar features, raising the question of how MARCH6 recognises other substrates. Whether these features are simply generic hydrophobic amino acids or highly sequence-specific amino acids remain to be determined. We postulate that VCP may be required for other MARCH6 substrates, but this needs to be investigated further. Overall, our investigation of the SM N100 degron has established a role for VCP in the cholesterol-regulated demolition of a canonical ERAD substrate by MARCH6.
Model for the cholesterol-accelerated degradation of the SM N100 cholesterol-responsive degron.
N.K.C. and A.J.B. conceptualised the project and writing of the manuscript. N.K.C. and N.A.S. carried out the experiments and contributed to the data collection and data analyses. All authors participated in the design of the experiments.
This work was supported by the Australian Research Council Grant DP170101178 (to A. J. B.). N.K.C. is funded by a University International Postgraduate Award (UIPA) from UNSW Sydney.
We thank Miss Florence Tomasetig (UNSW Biomedical Imaging Facility) for training and technical assistance with confocal microscopy, Dr Martin Bucknall (UNSW Bioanalytical Mass Spectrometry Facility) for training and technical assistance with mass spectrometry, and members of the Brown laboratory for critically reviewing this manuscript.
The Authors declare that there are no competing interests associated with the manuscript.