Ring finger protein 145 (RNF145) is a ubiquitin ligase for sterol-induced degradation of HMG-CoA reductase

Cholesterol biosynthesis is tightly regulated in the cell. For example, high sterol concentrations can stimulate degradation of the rate-limiting cholesterol biosynthetic enzyme 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMG-CoA reductase, HMGCR). HMGCR is broken down by the endoplasmic reticulum membrane–associated protein complexes consisting of insulin-induced genes (Insigs) and the E3 ubiquitin ligase gp78. Here we found that HMGCR degradation is partially blunted in Chinese hamster ovary (CHO) cells lacking gp78 (gp78-KO). To identify other ubiquitin ligase(s) that may function together with gp78 in triggering HMGCR degradation, we performed a small-scale short hairpin RNA–based screening targeting endoplasmic reticulum–localized E3s. We found that knockdown of both ring finger protein 145 (Rnf145) and gp78 genes abrogates sterol-induced degradation of HMGCR in CHO cells. We also observed that RNF145 interacts with Insig-1 and -2 proteins and ubiquitinates HMGCR. Moreover, the tetrapeptide sequence YLYF in the sterol-sensing domain and the Cys-537 residue in the RING finger domain were essential for RNF145 binding to Insigs and RNF145 E3 activity, respectively. Of note, amino acid substitutions in the YLYF or of Cys-537 completely abolished RNF145-mediated HMGCR degradation. In summary, our study reveals that RNF145, along with gp78, promotes HMGCR degradation in response to elevated sterol levels and identifies residues essential for RNF145 function.

Cholesterol is the most abundant sterol in mammalian cells. It regulates membrane function and serves as the precursor for bile acids and steroid hormones. Cholesterol can either be synthesized through the mevalonate pathway (1) or taken up from diets via Niemann-Pick C1-like 1 (NPC1L1)-mediated absorption (2)(3)(4)(5).
HMG-CoA 4 reductase (HMGCR) catalyzes the rate-limiting step in cholesterol biosynthesis in which HMG-CoA is converted to mevalonate. The half-life of HMGCR varies with cellular sterol levels: more than 12 h in sterol-depleted cells and less than 1 h in sterol-overloaded cells (6,7). HMGCR is tightly regulated by sterols at both transcriptional and posttranslational levels. High concentrations of cholesterol decrease transcription of the HMGCR gene by inhibiting the activation of sterol regulatory element-binding protein 2 (SREBP-2) (8,9). In addition, excess levels of 24,25-dihydrolanosterol, an intermediate in the mevalonate pathway, promote ubiquitination and degradation of the HMGCR protein (10 -12). Oxysterols can inhibit HMGCR transcription and stimulate HMGCR degradation (13,14). Besides sterols, geranylgeraniol, a nonsteroid product downstream of mevalonate, acts on the postubiquitination step to accelerate sterol-induced HMGCR degradation (7).
The sterol-induced degradation of HMGCR initiates when the endoplasmic reticulum (ER)-localized Insig-1 and -2 proteins bind to HMGCR (15) and recruit the ubiquitin ligase (E3) gp78 to catalyze ubiquitination (16). The HMGCR protein is eventually degraded in the proteasome. Ufd1 enhances the E3 activity of gp78 and accelerates the degradation of HMGCR (17). Ablation of gp78 in mouse liver increases the stability of HMGCR, Insig-1, and Insig-2 (18,19). Elevated levels of Insigs inhibit the SREBP pathway and decrease cholesterol synthesis (18). These data suggest that gp78 is a major E3 essential for HMGCR degradation in the hepatocytes.
Besides gp78, TRC8 and MARCH6 are two other ER-localized E3s involved in HMGCR degradation (20,21). TRC8 interacts with Insig-1 and -2 and ubiquitinates HMGCR for proteasomal degradation. In addition to sterol-regulated degradation, the basal turnover of HMGCR is mediated by Hrd1, an ER-anchored E3 homologous to gp78 (22,23). Interestingly, sterolinduced HMGCR degradation has been found to persist in gp78-deficient primary mouse embryonic fibroblasts (24), which leads us to speculate that there might be other E3(s) compensating for the function of gp78 in cultured cells.
In this study, we identified that an ER-anchored E3 named RNF145 catalyzed sterol-induced ubiquitination of HMGCR. Knockout of gp78 or Rnf145 alone had partial or little effect on HMGCR degradation in Chinese hamster ovary (CHO) cells. However, knockout of both genes dramatically blunted sterolinduced degradation of HMGCR. The E3 activity-deficient RNF145 (C537A) failed to promote sterol-induced ubiquitination and degradation of HMGCR. Moreover, we found that Insigs were required for RNF145-catalyzed HMGCR degradation and that RNF145 interacted with Insigs constitutively through its transmembrane domains. We therefore conclude that RNF145 is a new E3 promoting sterol-induced degradation of HMGCR.

Identification of Rnf145 involved in HMGCR degradation
To determine whether gp78 is exclusively responsible for HMGCR degradation, we treated WT CHO and gp78 knockout (gp78-KO) cells with increasing concentrations of 25-hydroxycholesterol . Although the degradation of HMGCR was partially impaired when cells were treated with low concentrations of 25-HC (0.03 and 0.1 g/ml), high concentrations of 25-HC (0.3 and 1 g/ml) almost or completely diminished HMGCR in gp78-KO cells (Fig. 1, A and B), suggesting that gp78 is not the only E3 mediating HMGCR degradation. Because previous studies identified a total of 24 ER membranespanning E3s (25), we next transfected plasmids expressing shRNAs targeting each E3 together with those expressing HMGCR-T7 and Insig-1-Myc into gp78-KO cells and examined sterol-induced degradation of HMGCR (Fig. S1). In contrast to control shRNA transfected cells where HMGCR degraded rapidly in response to sterols (Fig. 1C, compare lanes 1 and 2), no HMGCR degradation was detected upon Rnf145 deficiency in gp78-KO cells (Fig. 1C, lanes 3-6). These results suggest that RNF145 is involved in HMGCR degradation. We further generated Rnf145-KO and gp78 plus Rnf145 knockout (double KO) CHO cells using the CRISPR/Cas9 technique (26). Knockout of gp78 slightly affected HMGCR degradation relative to WT cells (Fig. 1D, lanes 4 -6), and knockout of Rnf145 alone had little influence on HMGCR degradation (Fig. 1D, lanes 7-9). However, HMGCR degradation was largely blunted in the double KO cells (Fig. 1D, lanes 10 -12). Next we measured sterol-induced ubiquitination of HMGCR in cells lacking either E3 or both. Knockout of gp78, Rnf145, or both genes decreased sterol-induced ubiquitination of HMGCR (Fig. 1E). These results suggest that RNF145 plays a critical role in sterolinduced degradation of HMGCR.

RNF145 is an ER-localized ubiquitin ligase mediating HMGCR degradation
RNF145 is a putative ER transmembrane protein ( Fig. 2A) (25). To confirm the subcellular location of RNF145, we performed immunofluorescence experiments by co-staining transfected RNF145 together with the endogenous ER marker Calnexin. RNF145-FLAG was largely colocalized with Calnexin, indicating that RNF145 is indeed an ER-localized protein ( Fig. 2B). Next, we purified the recombinant cytosolic domain (aa 511-663) of RNF145 and performed an in vitro ubiquitination assay. The recombinant cytosolic domain of gp78 (309 -643) was used as a positive control. We found that RNF145 (511-663) could efficiently catalyze the formation of polyubiquitin chains in the presence of E1, E2, FLAG-ubiquitin, and ATP (Fig. 2C). Replacement of the conserved Cys-537 residue with alanine (C537A) in the RING finger domain of RNF145 (27), however, completely abolished the E3 ligase activity of RNF145 (Fig. 2D). These results indicate that RNF145 is an ER-localized ubiquitin ligase and that the Cys-537 residue is required for its E3 activity.
We then sought to determine whether Cys-537 is required for sterol-induced HMGCR ubiquitination. As shown in Fig.  3A, lanes 1 and 2, sterols substantially increased the ubiquitination of HMGCR, as evidenced by the high-molecular-weight smears of the immunoprecipitates. Addition of RNF145 (C537A) competitively blocked the ubiquitination of HMGCR induced by sterols (Fig. 3A, lane 3). Consistent with the ubiquitination results, the WT form of RNF145 accelerated HMGCR degradation (Fig. 3B, lanes 1-4), whereas the C537A mutant completely abrogated HMGCR degradation (Fig. 3B, lanes 5-10). These results suggest that RNF145 acts as a ubiquitin ligase promoting HMGCR degradation.

The sterol-sensing domain of RNF145 is crucial for HMGCR degradation
It is known that SREBP cleavage-activating protein (SCAP) and HMGCR harbor the sterol-sensing domain (SSD) for Insig

Degradation of HMG-CoA reductase mediated by RNF145
binding and that the tetrapeptide YIYF is highly conserved within the SSD (7, 15). There is a putative SSD in RNF145 ( Fig.  2A). Sequence alignment revealed a YLYF tetrapeptide in RNF145 corresponding to the YIYF found in HMGCR and SCAP (Fig. 6A). To test whether this motif was essential for RNF145-mediated degradation of HMGCR, we mutated

Degradation of HMG-CoA reductase mediated by RNF145
YLYF to AAAA and evaluated its function in 4KO cells. Unlike WT RNF145, the RNF145 (YLYF-AAAA) mutant could not elicit the degradation of HMGCR following sterol treatment (Fig. 6B). We next combined the YLYF-AAAA mutation with E3 activity deficiency together. As shown in Fig. 6C, the RNF145 (C537A and YLYF-AAAA) mutant no longer blocked sterol-regulated degradation of HMGCR. Moreover, the RNF145 (YLYF-AAAA) mutant could not be co-immunoprecipitated with Insig-1 (Fig. 6D). Collectively, these results suggest that the YLYF motif is required for Insig binding as well as RNF145-mediated HMGCR degradation.

Discussion
In this study, we identified RNF145 as another E3 mediating the degradation of HMGCR through unbiased shRNA screening. RNF145 resides in the ER with 14 putative transmembrane segments, one to five of which constitute an SSD that is also present in SCAP, HMGCR, NPC1, NPC1L1, and TRC8 (29 -31). The RING finger domain localized in the cytosolic C terminus confers RNF145 E3 activity. RNF145 and TRC8 display similar structures and sequences.
In CHO cells, knockout of gp78 partially delayed the turnover of HMGCR in response to low concentrations of sterols, and ablation of Rnf145 alone also had little effect (Fig. 1). Notably, knockout of both genes largely abolished sterol-induced degradation of HMGCR (Fig. 1D). Similar to gp78 and TRC8, Insigs are required by RNF145 for HMGCR ubiquitination. The binding between Insigs and HMGCR is regulated by sterol levels, whereas the Insig and E3 (gp78, TRC8, and RNF145) interaction is constitutive. The inactivated mutation (C537A) of RNF145 had a dominant-negative effect on HMGCR degradation. A similar C356G mutation was found in gp78 (16). Interestingly, HMGCR, SCAP, and RNF145 all bind Insigs through the SSD and require the conserved YI(L)YF tetrapeptide for binding.

Degradation of HMG-CoA reductase mediated by RNF145
The question why multiple E3s are involved in sterol-induced degradation of HMGCR is intriguing. One possibility is that different cells express different levels of E3s in response to distinct signals. In fact, recent studies have identified Rnf145 as a Liver X receptor (LXR) target gene (32,33). We hypothesize that activation of LXR might elevate the RNF145 level and subsequently down-regulate cholesterol biosynthesis through degrading HMGCR. Another possibility is that the existence of multiple E3s for HMGCR degradation prevents saturation of specific E3(s) and ensures that ER-associated degradation functions properly when HMGCR is degraded.
The protein machineries involved in HMGCR degradation may also participate in other cholesterol-regulating processes. gp78 is the first characterized E3 catalyzing HMGCR ubiquitination (16). It is highly expressed in the liver. Knockout of gp78 in the hepatocytes largely blunted the degradation of HMGCR (18). However, gp78 deficiency also stabilizes Insigs (especially Insig-2), resulting in suppressed processing of SREBP and subsequently decreased expression of Hmgcr and other genes in the mevalonate pathway (18). As the protein levels of Insigs were dramatically increased in gp78-deficient cells (18) (Fig.  S2), the effects of other E3s such as TRC8 and RNF145 might be boosted, which may explain why HMGCR is still degraded in gp78-KO cells (24). gp78 can also catalyze ubiquitination of acyl-CoA:cholesterol acyltransferase (ACAT)-2 on a cysteine residue (34). In addition to degrading HMGCR, RNF145 triggers the ubiquitination of SCAP and interferes with SCAP binding to coat protein complex II (COPII), thus inhibiting SREBP-2 maturation (33).
In summary, we here identify that RNF145 is an E3 governing sterol-regulated degradation of HMGCR. Together with the previous findings that Rnf145 is an LXR-regulated gene and that RNF145 inhibits the SREBP pathway through ubiquitinating SCAP, RNF145 serves as an important negative regulator of cholesterol biosynthesis. Activation of RNF145 may be effective for treating hypercholesterolemia through inhibiting endogenous cholesterol synthesis.

Reagents
We obtained lovastatin, mevalonate, and 25-hydroxycholesterol from Sigma, MG132 from Calbiochem, FLAG-ubiquitin from Boston Biochem, and linear polyethyleneimine from Polysciences; lipoprotein-deficient serum was prepared from newborn calf serum as described before (35).

Antibodies
Primary antibodies used were as follows. Mouse monoclonal antibody against the T7 tag (Novagen), mouse monoclonal antibody P4D1 against ubiquitin and goat polyclonal antibody against Calnexin (Santa Cruz Biotechnology), mouse monoclonal antibody (clone 16B12) against the HA tag (Biolegend), mouse monoclonal antibody (clone AC-15) against ␤-actin and monoclonal antibody (clone M2) against the FLAG tag (Sigma), mouse monoclonal antibody (clone 10E2) against the His tag (Abmart), and mouse monoclonal antibody (clone 9E10) against the Myc tag and monoclonal antibody (clone A9) against HMGCR were prepared from hybridomas (ATCC).

Degradation of HMG-CoA reductase mediated by RNF145
Rabbit polyclonal antibodies against gp78 and HMGCR were prepared as described before (18). Rabbit polyclonal antibody against RNF145 (aa 511-663) was generated by immunizing rabbits followed by affinity purification with antigens.

Cell culture
Medium A contained a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 medium containing 100 units/ml penicillin and 100 g/ml streptomycin sulfate. Medium B contained medium A supplemented with 5% fetal bovine serum. Medium C contained medium A supplemented with 5% lipoprotein-deficient serum, 1 M lovastatin, and 50 M mevalonate. All cells were kept in medium B at 37°C and 5% CO 2 and treated as indicated in the figure legends.

Transient transfection of CHO cells
Cells were set up on day 0 at 8 ϫ 10 5 cells per 60-mm dish. On day 1, a total of 3 g of DNA and 6 g of linear polyethyleneimine were mixed and transfected to each dish. The medium was refreshed 6 h after transfection.

Immunoblot analysis
Cells were lysed with 200 l of radioimmune precipitation assay buffer and mixed with the loading buffer (23.4 mM Tris-HCl, 5.625% (w/v) SDS, 1 M urea, 3.75% (v/v) glycerol, and 37.5 mM DTT in final concentrations) and incubated at 37°C for 30 min. The protein concentration of each lysate was quantified by Pierce BCA protein assay, and equal amounts of total proteins were loaded for SDS-PAGE gels and transferred to polyvinylidene difluoride membranes. Membranes were blocked by TBS-Tween (1‰) supplemented with 5% skim milk for 1 h at room temperature and then incubated with the indicated primary antibodies overnight at 4°C. Membranes were washed three times with TBS-Tween and incubated with secondary antibodies (1:5000) diluted in TBS-Tween supplemented with 5% skim milk for 1 h at room temperature, followed by at least three washes with TBS-Tween. Quantification of the immunoblot was performed with ImageJ.

Immunoprecipitation
Cells were harvested and lysed in 1 ml of immunoprecipitation (IP) buffer (1 ϫ PBS, 0.5% (v/v) Nonidet P-40, 5 mM EDTA, 5 mM EGTA, and protease inhibitors), followed by centrifugation at 12,000 ϫ g for 10 min at 4°C. Supernatants were immunoprecipitated with anti-Myc beads for 4 -6 h at 4°C. Beads were washed three times with IP buffer at 4°C and boiled at 95°C for 10 min. Aliquots were subjected to immunoblot analysis.

Ubiquitination of HMGCR
CHO cells were transfected and treated as described in the figure legends. Cells were lysed in HMG-IP buffer (1 ϫ PBS containing 1% (v/v) Nonidet P-40, 1% (w/v) deoxycholate, 5 mM EDTA, 5 mM EGTA, 0.1 mM leupeptin, protease inhibitors, 10 M MG132, and 10 mM N-ethylmaleimide). Lysates were first immunoprecipitated at 4°C with 3 g of polyclonal antibody against GFP plus 40 l of protein A/G-agarose beads for 1 h. Beads were removed by centrifugation, and supernatants were then immunoprecipitated with 40 l of anti-FLAG-agarose beads at 4°C for 5 h. After incubation, beads were washed three times with HMG-IP buffer at 4°C and boiled at 95°C for 10 min. Aliquots were subjected to immunoblot analysis.
For detecting ubiquitination of endogenous HMGCR, CHO cells were set up on day 0 at 2.5 ϫ 10 6 per 100-mm dish.

Degradation of HMG-CoA reductase mediated by RNF145
Duplicate dishes were treated as described in the figure legends. 6 g of polyclonal antibody against HMGCR plus 100 l of protein A/G-agarose beads were used to precipitate HMGCR. Aliquots were subjected to immunoblot analysis.

Immunostaining
Cells were fixed with 4% paraformaldehyde for 15 min and then permeabilized with 1ϫ PBS containing 0.1% Triton X-100 for 5 min. Cells were blocked with 1% BSA for 30 min, followed by incubation with primary and secondary antibodies diluted in 1% BSA for 1 h at room temperature (36). Cells were analyzed with a Leica confocal scanning laser microscope (Leica SP8) with a ϫ63 oil immersive objective.