The Lifetime of UDP-galactose:Ceramide Galactosyltransferase Is Controlled by a Distinct Endoplasmic Reticulum-associated Degradation (ERAD) Regulated by Sigma-1 Receptor Chaperones*

Background: UDP-galactose:ceramide galactosyltransferase (CGalT) is a glycoprotein that synthesizes galactosylceramides at the endoplasmic reticulum (ER). Results: Molecular chaperone sigma-1 receptors promote degradation of CGalT by forming a complex with Insig. Conclusion: CGalT and its activity are post-translationally regulated by ER-associated degradation (ERAD) involving sigma-1 receptor chaperones. Significance: The sterol-sensing ERAD system controls the enzyme involved in glycosphingolipid biosynthesis. The glycosphingolipid biosynthesis is initiated by monoglycosylation of ceramides, the action of which is catalyzed either by UDP-glucose:ceramide glucosyltransferase or by UDP-galactose:ceramide galactosyltransferase (CGalT). CGalT is expressed predominantly at the endoplasmic reticulum (ER) of oligodendrocytes and is responsible for synthesizing galactosylceramides (GalCer) that play an important role in regulation of axon conductance. However, despite the importance of ceramide monoglycosylation enzymes in a spectrum of cellular functions, the mechanism that fine tunes activities of those enzymes is largely unknown. In the present study, we demonstrated that the sigma-1 receptor (Sig-1R) chaperone, the mammalian homologue of a yeast C8-C7 sterol isomerase, controls the protein level and activity of the CGalT enzyme via a distinct ER-associated degradation system involving Insig. The Sig-1R forms a complex with Insig via its transmembrane domain partly in a sterol-dependent manner and associates with CGalT at the ER. The knockdown of Sig-1Rs dramatically prolonged the lifetime of CGalT without affecting the trimming of N-linked oligosaccharides at CGalT. The increased lifetime leads to the up-regulation of CGalT protein as well as elevated enzymatic activity in CHO cells stably expressing CGalT. Knockdown of Sig-1Rs also decreased CGalT degradation endogenously expressed in D6P2T-schwannoma cells. Our data suggest that Sig-1Rs negatively regulate the activity of GalCer synthesis under physiological conditions by enhancing the degradation of CGalT through regulation of the dynamics of Insig in the lipid-activated ER-associated degradation system. The GalCer synthesis may thus be influenced by sterols at the ER.

The glycosphingolipid biosynthesis is initiated by monoglycosylation of ceramides, the action of which is catalyzed either by UDP-glucose:ceramide glucosyltransferase or by UDP-galactose:ceramide galactosyltransferase (CGalT). CGalT is expressed predominantly at the endoplasmic reticulum (ER) of oligodendrocytes and is responsible for synthesizing galactosylceramides (GalCer) that play an important role in regulation of axon conductance. However, despite the importance of ceramide monoglycosylation enzymes in a spectrum of cellular functions, the mechanism that fine tunes activities of those enzymes is largely unknown. In the present study, we demonstrated that the sigma-1 receptor (Sig-1R) chaperone, the mammalian homologue of a yeast C8-C7 sterol isomerase, controls the protein level and activity of the CGalT enzyme via a distinct ER-associated degradation system involving Insig. The Sig-1R forms a complex with Insig via its transmembrane domain partly in a sterol-dependent manner and associates with CGalT at the ER. The knockdown of Sig-1Rs dramatically prolonged the lifetime of CGalT without affecting the trimming of N-linked oligosaccharides at CGalT. The increased lifetime leads to the up-regulation of CGalT protein as well as elevated enzymatic activity in CHO cells stably expressing CGalT. Knockdown of Sig-1Rs also decreased CGalT degradation endogenously expressed in D6P2Tschwannoma cells. Our data suggest that Sig-1Rs negatively regulate the activity of GalCer synthesis under physiological conditions by enhancing the degradation of CGalT through regulation of the dynamics of Insig in the lipid-activated ER-associated degradation system. The GalCer synthesis may thus be influenced by sterols at the ER.
Glycosphingolipids play a variety of roles in cell biology, not only as structural components of membranes, but also as regulators of signal transductions, cell death, cell adhesion, protein/ lipid trafficking, cellular differentiation, receptor clustering, and conduction of electrical impulses at axons (1)(2)(3)(4)(5)(6). Ceramide, the backbone of glycosphingolipids, is synthesized at the endoplasmic reticulum (ER) 3 and acquires glucosylation or galactosylation catalyzed by specific transferases at the ER or the Golgi (7)(8)(9). The monoglycosylation step of ceramides influences the synthesis activity of gangliosides and sulfatides, intramembrane flip-flop movements of sphingolipids, membrane curvature, and cytotoxic activity of ceramides (9 -11). Glucosylation is catalyzed at the outer surface of the Golgi apparatus by the UDG-glucose:ceramide glucosyltransferase for the synthesis of glucosylceramides (GlcCer) (12). Conversely, galactosylation of ceramides is catalyzed by the UDGgalactose:ceramide galactosyltransferase (CGalT) at the lumen of the ER (7,13).
The CGalT amino sequence contains a KKVK motif, the potential ER retrieval signal, at the C terminus (14,15). A study demonstrated that CGalT remains sensitive to endoglycosidase H, supporting its ER localization (7). Immunocytochemistry using specific antibodies also showed that CGalT immunoreactivity is found exclusively at the ER and nuclear envelope and not at the Golgi and plasma membrane (7,13). These series of studies confirmed that CGalT is a class I integral ER protein possessing a long ER lumenal catalytic domain and a single transmembrane domain at the C terminus (7). A study using knock-out mice of CGalT confirms that there is only one galactosylceramide (GalCer)-synthesizing enzyme in the brain. CGalT is highly and specifically expressed in oligodendrocytes and in human epithelia (14). GalCer synthesized by CGalT comprise a major lipid in the myelin sheath of oligodendrocytes or Schwann cells that insulates axons to regulate electric impulses conducted by neuronal depolarization (4). Although highly enriched in myelins, GalCer is shown to serve as a negative regulator of oligodendrocyte differentiation and myelin formation (16).
The sigma-1 receptor (Sig-1R) is an integral membrane protein ubiquitously expressed in multiple organs, including the brain (17,18). The Sig-1R binds a variety of structurally different drugs ((ϩ)-isoforms of benzomorphans, haloperidol, and fluvoxamine) as well as endogenous molecules, such as sterols (e.g. progesterone) and simple sphingolipids (e.g. D-erythrosphingosine, ceramide, and GalCer) (18 -22). The structure of the Sig-1R shares no similarity with those of any mammalian proteins but shares 70% similarity with that of a yeast C8-C7 sterol isomerase (23). In animals, Sig-1Rs have been shown to promote neuronal survival under ischemia or ␤-amyloid depositions, potentiate morphine-induced analgesia, improve learning and memory, and contribute to the development of addictive behaviors induced by psychostimulants (24 -29). A recent study identified the Sig-1R as being a novel ligand-operated molecular chaperone that can regulate ER stress and signal transductions (30). When ER Ca 2ϩ is depleted, Sig-1Rs chaperone inositol 1,4,5-trisphosphate receptors localized at the interface between ER and mitochondria to ensure proper Ca 2ϩ transmission between the two organelles (30,31). Whether Sig-1Rs stabilize any other ER proteins or regulate ER-associated degradation (ERAD) has not been investigated in detail.
Sig-1Rs are highly expressed in oligodendrocytes of rat brains (32,33). We previously demonstrated that Sig-1Rs colocalize with GalCer at the ER and up-regulate in oligodendrocyte-type 2 astrocyte progenitors during their oligodendrocyte differentiation (33). Knockdown of Sig-1Rs almost completely blocks the differentiation and formation of myelin sheets in oligodendrocytes (33). Because both Sig-1Rs and CGalT are localized at the ER, and GalCer synthesized by CGalT is a major regulator of oligodendrocyte differentiation, we hypothesized that Sig-1Rs may regulate CGalT activity at the ER and may thus regulate cellular differentiation. Here we employed a simple cellular model (i.e. CHO cell line stably expressing CGalT) to examine whether Sig-1Rs regulate the activity of CGalT. We found that Sig-1R chaperones, in lieu of stabilizing CGalT proteins, promote instead the acceleration of CGalT degradation by utilizing a distinct class of ERAD systems. Sig-1Rs do so by forming a protein complex with the sterol-sensing protein Insig to regulate degradation of CGalT.
Thin Layer Chromatography (TLC)-Total lipids were extracted by Bligh and Dyer partitioning as described previously (22,36). Lipid extracts dried under an N 2 flow were dissolved in a chloroform/methanol mixture (2:1) and spotted on a high performance thin layer chromatography plate (Merck). After resolution on TLC plates with a chloroform/methanol/ water mixture (13:5:0.5), lipids were visualized by a diphenylamine-aniline or 0.2% 8-anilinonaphthalene-1-sulfonate spray. Respective lipids were identified by resolving purified lipids on the same TLC plate. Lipids were quantified by a Kodak Image Station 440 CF.
[ 14 C]Serine Labeling for Analysis of Lipid Synthesis-After washing with serum-free medium, cells in 6-cm dishes were incubated at 37°C in 1 ml of minimum essential medium containing 0.5 Ci/ml [ 14 C]serine/ml. Cells harvested on ice were suspended in 200 l of ice-cold H 2 O. Following vortex (15 times for 1 s each), 5 l of cell suspensions were transferred into a 96-well plate (duplicate) for the Bio-Rad protein assay. Cells were lysed by adding 20 l of 1% SDS (on ice for 5 min). Lipids in the lysates were extracted by successively adding the following reagents: ϳ563 l of CHCl 3 /methanol (1:2) (vortex for 10 min), ϳ188 l of CHCl 3 (vortex for 1 min), and 187.5 l of H 2 O (vortex for 1 min). Following a centrifugation for 3 min at 5000 ϫ g, the lower phase was transferred to a new tube. After adding ϳ280 l of CHCl 3 to the upper water phase in the original tubes, samples were vortexed for 1 min and centrifuged for 3 min at 5000 ϫ g. The yielded lower phase was combined with the previously obtained lower phase in a new tube. Lipids in the lower phase were dried under a N 2 flow and resolved by TLC as described above. The TLC plate was air-dried and exposed to a ␤-Max film (Amersham Biosciences) for 5 days at room temperature.
Sucrose Fractionation-CHO cells grown on a 15-cm dish were homogenized with a Dounce glass homogenizer (30 strokes) in 400 l of homogenization buffer (10 mM HEPES (pH 7.4), 15 mM KCl, 1.5 mM MgCl 2 , protease inhibitor mixture (Sigma)). Cell homogenate was centrifuged at 380 ϫ g for 10 min at 4°C. The supernatant was pooled. The pellet was homogenized (20 strokes) again in 200 l of homogenization buffer. After centrifugation (380 ϫ g, 10 min), the supernatant was combined with one obtained after the first centrifugation. 300 l of the supernatant was placed on top of a discontinuous sucrose gradient (0.5, 0.7, 0.9, 1.1, 1.3, and 1.5 M sucrose; 600 l for each fraction). The sample was centrifuged in a SW 55i rotor at 180,000 ϫ g for 3 h at 4°C. 12 fractions were obtained from the top (325 l/fraction) and stored at Ϫ80°C until the assays.
Cholesterol Deprivation and Drug Treatment-The intracellular cholesterol level was reduced according to the previously established procedure (37,38). Briefly, 6 h after transfection of plasmids, CHO cells were seeded onto 6-cm dishes with normal culture medium. On the next day, medium was replaced with 3 ml of minimum essential medium containing 2% lipoproteindeficient serum (LPDS), 10 M compactin, and 50 M mevalonate with or without sterols. After 16 h, cells were incubated in the MEM-␣ containing 2% LPDS, 25 M compactin, and 5 mM mevalonate with/without sterols or (ϩ)-pentazocine for 4 h. Sterols and compactin were dissolved in ethanol. Mevalonate and (ϩ)-pentazocine were dissolved in H 2 O. Controls received the same vehicles without drugs.

CHO Cell Line Stably Expressing
CGalT-To examine the possibility that Sig-1Rs regulate the CGalT activity, the CHO cell line stably expressing rat CGalT (CHO-CGalT) was employed (7). This cell line constitutively expresses a considerably high level of endogenous Sig-1Rs (22,30), and stably transfected rat CGalT is shown to behave normally regarding its subcellular distribution and enzymatic activity (7,13). The CGalT antibody (7) revealed strong CGalT-like immunoreactivity in this line but not in wild-type CHO cells (CHO-WT) (Fig. 1A). CGalT distributed in the cytoplasmic region, showing a typical ER pattern (Fig. 1A).
In immunoblotting, anti-CGalT antibodies detected a single band with a 55-kDa molecular mass (7). Expression of Sig-1Rs was similar between CHO-WT and CHO-CGalT cells (Fig. 1B). Because CGalT accepts both ceramides and diacylglycerols as substrates (7), both GalCer (doublets in Fig. 1B) and monogalactosylated diacylglycerols were present in lipid extracts from CHO-CGalT cells but not in those from CHO-WT (Fig. 1B). The upper and lower bands of GalCer represent those containing non-hydroxylated and hydroxylated acyl-chains, respectively (40).
Monoclonal antibodies against GalCer expressed strong GalCer-like immunoreactivities only in CHO-CGalT (Fig. 1C). The level of GalCer immunoreactivities is fairly correlated with the CGalT level in most CHO-CGalT cells (Fig. 1C). However, In the left panels, 20 g of total cell lysates prepared from CHO-WT or CHO-CGalT were resolved by 13% SDS-PAGE and immunoblotted with respective specific primary antibodies. In the right panels, total lipid extracts were resolved by TLC followed by a 0.2% 8-anilinonaphthalene-1-sulfonate spray. Cer, ceramide. The asterisk in the CHO-WT represents endogenous GlcCer. Lipids marked with dots were not identified. The images represent the result of 4 -5 repeated experiments. C, immunostaining of CGalT and GalCer expressed in CHO-CGalT. CHO-CGalT cells were fixed with paraformaldehyde. Following a mild permeabilization with Triton X-100 (0.01%, 1 min), samples were immunostained with anti-CGalT antiserum and anti-GalCer antibodies. Images were captured by a confocal microscope. The arrows indicate CHO-CGalT expressing very low levels of CGalT. Note that the level of GalCer in individual cells is fairly proportional to that of CGalT. Bars, 10 m.
there are some cells that express a high level of CGalT but express less GalCer immunoreactivity. The lipid environment that may vary between individual cells and/or cell permeabilization procedures during immunofluorescence may affect the immunoreactivity of anti-GalCer antibodies.
Sig-1Rs Negatively Regulate Production of GalCer-As shown by TLC ( Fig. 2A), we found that CHO-CGalT cells overexpressing Sig-1R-EYFP, when compared with those expressing EYFP, contained lower GalCer. In contrast, knockdown of Sig-1Rs caused an increase in GalCer ( Fig. 2A). The same tendency was also observed in the level of monogalactosylated diacylglycerols ( Fig. 2A), suggesting that Sig-1Rs negatively regulate the enzymatic activity of CGalT.
Next, sphingolipid synthesis between control and Sig-1R knockdown cells was monitored in [ 14 C]serine-labeled CHO-CGalT cells. Because serine is also utilized for phospholipid synthesis, autoradiography of TLC plates revealed both 14  There was no apparent difference in the synthesis of ceramides and phospholipids between control and Sig-1R knockdown cells (Fig. 2B). In this study, we also serendipitously found that knockdown of Sig-1Rs decreased the GlcCer synthesis (see the upper band of [ 14 C]GlcCer in Fig. 2B). Because the GlcCer synthase, which shares no homology with CGalT, localizes mainly at the Golgi complex, this result may suggest that Sig-1Rs regulate stability and/or the transport of the enzyme to the Golgi complex. Although this is a striking finding, we focused solely on exploring the regulation of CGalT by Sig-1Rs in this paper.
Sig-1Rs Down-regulate CGalT Proteins-ER chaperones are known to promote protein folding but also serve as subcomponents in ERAD systems (41,42). We therefore examined whether overexpression or knockdown of Sig-1Rs affects CGalT protein levels. Immunoblotting demonstrated that the level of CGalT was significantly increased by Sig-1R knockdown, whereas it was decreased by overexpression of Sig-1Rs (Fig. 3A). Knockdown of Sig-1Rs also caused an increase of CGalT proteins that are endogenously expressed in D6P2Tschwannoma cells (Fig. 3B), indicating the physiological relevance of our finding. Neither knockdown nor overexpression of Sig-1Rs affected the mRNA level of CGalT in CHO-CGalT (Fig.  3C). Therefore, Sig-1Rs may post-translationally regulate the protein level of CGalT.
Sig-1Rs Associate with CGalT at the ER of CHO Cells-Because Sig-1Rs physically associate with structurally different proteins to regulate degradation of associated proteins (20, 30), we next examined whether Sig-1Rs also associate with CGalT. Confocal microscopy demonstrated that both Sig-1R-EYFP (expressed at a low level with 0.5 g of plasmid/6-cm dish) and CGalT distribute over the punctate/reticular ER structures and nuclear envelopes (Fig. 4A). Similarity of subcellular distribution of these proteins was also examined by using subcellular fractionation. The result showed that distribution of membranes containing CGalT possesses two peaks: the small peak at low density fractions (F2 and F3) and the large peak at high density fractions (F8 -F11). Membranes containing Sig-1Rs also showed a similar pattern (Fig.  4B). Sig-1Rs were also contained in fractions F10 and F11 (Fig.  4B). Fractionation patterns of GM130 (cis-Golgi marker) and Mcl-1 (mitochondria marker) are completely different from that of CGalT.
Knockdown of Sig-1Rs Prolonged the Lifetime of CGalT-To delineate the molecular mechanism by which Sig-1Rs regulate the CGalT protein level, the pulse-chase experiment was performed. As shown in Fig. 5A, CGalT was shown to have a t1 ⁄ 2 of approximately 200 min. The knockdown of Sig-1R did not affect the synthesis of [ 35 S]CGalT during the 10-min pulselabeling (chase 0 min in Fig. 5A) but significantly delayed the decline of the [ 35 S]CGalT during chasing (Fig. 5A). The knockdown of Sig-1Rs thus prolongs the lifetime of CGalT. The similar effect of Sig-1R knockdown was seen with endogenously expressed CGalT in D6P2T cells (Fig. 5B).
We previously demonstrated that N-linked oligosaccharides on CGalT remain sensitive to endoglycosidase H digestion even 16 h after chasing, suggesting that CGalT is an ER-resident glycoprotein and is not transported through the Golgi complex (7). Here, we found that in the pulse-chase experiments, CGalT bands are downward shifted slightly at the 30-min chasing point and then shifted further at 60-and 360-min chasing points (arrowheads in Fig. 5A), indicating the active trimming of the N-linked oligosaccharides being processed on CGalT. In fact, the downward shifts were partially blocked by castanospermine (ER glucosidase I and II inhibitor) and kifnensine (ER mannosidase I inhibitor) but not by swainsomine, an ER mannosidase II inhibitor (Fig. 5C). Notably, the effect of Sig-1R knockdown on delaying degradation of CGalT was still observed in the presence of those inhibitors. Thus, the action of Sig-1Rs regulating the degradation of CGalT is probably independent of the trimming status of CGalT (i.e. in apparent contrast to that seen in well defined ERAD of glycoproteins that is operated by lectin chaperones, such as calnexin) (41,42).
A specific ERAD pathway is selected based on topological and structural configurations of individual proteins to be degraded (41). For example, ER proteins with a defect on the cytosolic domain utilize ERAD-C, whereas ER proteins having defects at the ER lumenal domain utilize ERAD-L for degradation. The latter is further classified into ERAD-L M and ERAD-L S , based on whether the substrate proteins possess a trans- membrane domain (i.e. membrane proteins) or not (i.e. soluble protein) (41,43). Because CGalT is a glycoprotein having nearly its entire sequence inside the ER (Fig. 6A), Sig-1Rs might promote degradation of CGalT via ERAD-L. To test this possibility, we employed well characterized substrates of ERAD-L M (CD3-␦) and ERAD-L S (NHK; CD3-␦⌬ lacking its transmembrane domain) (43) and examined whether knockdown of Sig-1Rs compromises degradation of those proteins. As shown in Fig. 6B, we found that knockdown of Sig-1Rs does not affect degradation of any ERAD-L substrates tested. Knockdown of Sig-1Rs also did not affect the lifetime of the cytosolic protein YFP (Fig. 6B).
Sig-1Rs Form a Protein Complex with Insig1-There is a distinct subclass of ERAD complexes that processes protein degradation by sensing lipid levels at the ER membrane. When the cholesterol level is high at the ER, Insig1 or -2 delivers a specific E3 ligase to HMG-CoA reductase to degrade the enzyme (37). Insig also forms a different protein complex with SCAP to regulate the translocation of the transcription factor SREBPs from the ER to Golgi (44). In both systems, Insig interacts with counterpart proteins when they bind sterols (44). Because the Sig-1R possesses a putative sterol-binding site, we speculated that the Sig-1R might form a complex with Insig, thus regulating ERAD of CGalT. The small amount of Sig-1R-EYFP vectors (0.5 g/6-cm dish) were transfected into CHO-CGalT 2 days before the observation. CGalT was immunocytochemically labeled with anti-rat CGalT antiserum, followed by labeling with Alexa 590 goat anti-rabbit IgG secondary antibodies. Bar, 10 m. The image represents the results from five independent immunolabelings. B, subcellular fractionation of CHO-CGalT cells. Fractions 1-12 were obtained by a sucrose gradient centrifugation of cell homogenates. The levels of respective proteins were measured by immunoblotting in each fraction. The percentage of the total in fraction x was calculated by the equation, 100 ϫ (level of the protein in fraction x)/(the sum of the protein levels in 12 fractions). CGalT activity was measured by incubating each fraction with 2 Ci/ml [ 14 C]UDP-galactose and 100 nM BSA-conjugated C 6 -ceramides at 37°C for 120 min, as described in the legend to Fig. 2C. The data represent three independent experiments. C, immunoprecipitation assay for the detection of the CGalT-Sig-1R association. Two days before immunoprecipitation assays (IP), CHO cells were transfected with empty vectors (Ϫ), CGalT-Myc (2 g/6-cm dish), and/or rat Sig-1R-V5 (0.5 g/6-cm dish). After cross-linking with DSP, cell lysates were prepared as described under "Experimental Procedures." Each lysate was then divided in half to use one half for Myc immunoprecipitation (top two panels) and the other for V5 immunoprecipitation (bottom two panels). The images represent results from three independent experiments.
We designed an immunoprecipitation study (i) to examine the potential interaction of Sig-1Rs with Insig1 and (ii) to clarify which Insig complex (i.e. Insig associating with SCAP or Insig coupling to ERAD) associates with Sig-1Rs. As shown in Fig.  7A, anti-FLAG antibodies immunoprecipitated Sig-1R-FLAG together with Insig1-Myc but not with SCAP. Insig1-Myc was pulled down with anti-FLAG antibodies only when Sig-1R-FLAG was co-expressed (Fig. 7A), verifying the specific co-immunoprecipitation of Insig1-Myc by Sig-1R-FLAG proteins. The result also indicates that Sig-1R-FLAG does not associate with SCAP. Conversely, anti-Myc antibodies immunoprecipitated Insig1-Myc together with SCAP and Sig-1R-FLAG, suggesting the existence of two Insig1 complexes: one associating with SCAP and the other with Sig-1Rs. Notably, when Sig-1R-FLAG was expressed, thus leading to the increase of Sig-1R⅐Insig1 complexes, SCAP associating with Insig1 was significantly reduced (Fig. 7A,  lane 5 versus lane 6 in the third panel from the top). There-  DECEMBER 14, 2012 • VOLUME 287 • NUMBER 51

JOURNAL OF BIOLOGICAL CHEMISTRY 43163
fore, Sig-1Rs associating with Insig1 may prevent the interaction of Insig1 with SCAP.
Sig-1Rs Are Involved in Insig-regulated ERAD-To test whether Sig-1Rs are involved in Insig-mediated ERAD, we employed a well characterized model of Insig-induced protein degradation. HMG-CoA reductase overexpressed in CHO cells is stable under sterol-reducing conditions, whereas 25-hydroxycholesterol promotes a rapid down-regulation of the enzyme via Insig (37). In this assay, Insig was routinely co-overexpressed to enhance Insig-induced protein degradation (37). We confirmed that the application of 25-hydroxycholesterol to sterol-reduced CHO cells indeed causes a reduction of V5-tagged HMG-CoA reductase (67.6% reduction, n ϭ 4). Interestingly, knockdown of Sig-1Rs per se increased the level of HMG-CoA reductase-V5 (Fig. 7B). Further, in CHO cells knocking down Sig-1Rs, the 25-hydroxycholesterol-induced decrease of HMG-CoA reductase-V5 was significantly smaller ( Fig. 7B; 30.6% reduction; *, p Ͻ 0.05 compared with the control siRNA samples by paired t test, n ϭ 6), indicating that Sig-1Rs play a role in Insig-mediated ERAD.
Sig-1Rs Associate with Insig1 via the Second Transmembrane Domain in a 25-Hydroxycholesterol/Ligand-sensitive Manner-To gain insight into the functional relevance of the interaction between Sig-1Rs and Insig, the effect of sterols on their association was examined. We found that Sig-1Rs constitutively associate with Insig1 to a certain degree even under sterol-reducing conditions (Fig. 7C, lane 1), where the association of SCAP with Insig was completely abolished (45). 25-Hydroxycholesterol (1 g/ml for 16 h), the potent inducer of Insig⅐SCAP association, potentiated the association between Sig-1R-FLAG and Insig1-Myc (Fig. 7C). The association of Sig-1R-FLAG with Insig1-Myc was unaffected by cholesterol (10 g/ml for 16 h) or lanosterol (10 g/ml for 16 h), the relatively less potent inducer of the Insig⅐SCAP association (Fig. 7C). Sig-1Rs bind sterols but also diverse classes of hydrophobic or amphipathic drugs (18). Notably, the prototypic Sig-1R ligand (ϩ)-pentazocine potentiates the association of Sig-1R-FLAG with Insig1-Myc (Fig.  7C).
The domain of the Sig-1R responsible for the association with Insig was also explored by expressing various truncated Sig-1R-V5 (⌬Sig-1R-V5). ⌬Sig-1R-V5 lacking a part of or the entire ER-lumenal domain (amino acids 116 -223) associated with Insig1-Myc to a similar degree as shown with the fulllength Sig-1R-V5 (Fig. 7D). The Sig-1R-V5 mutants that lack the second transmembrane domain (amino acids 90 -116) no longer co-immunoprecipitated Insig1-Myc. Thus, the second transmembrane domain, which possesses a putative sterol-binding pocket (23), seems to be essential for the formation of the Sig-1R⅐Insig1 complex. In contrast, the ER lumenal domain (Sig-1R-V5(117-223) with/without the KDEL ER retrieval sequence) that exhibits innate chaperone activity of the Sig-1R (30) failed to associate with Insig1. We also found that knockdown of Sig-1Rs did not affect the protein level of Insig1-Myc (data not shown), suggesting that Sig-1Rs do not associate with Insig1 merely to stabilize/degrade Insig1.
Adapting the experimental paradigm used in Fig. 7B, we examined whether 25-hydroxycholesterol induces the downregulation of CGalT under sterol-reducing conditions. The application of 25-hydroxycholesterol causes a significant reduction of CGalT in control cells (Fig. 8B, 54.5% reduction, n ϭ 7). In Sig-1R knockdown cells, the 25-hydeoxycholesterolinduced reduction of CGalT was significantly smaller when compared with CHO-CGalT transfected with control siRNA (Fig. 8B; 29.3% reduction in siSig-1R cells; *, p Ͻ 0.05 compared with the control by paired t test, n ϭ 7). Taken together, these results suggest that similar to HMG-CoA reductase, the CGalT protein level is regulated by sterol-sensing ERAD that involves Sig-1Rs and Insig.

DISCUSSION
This study demonstrated for the first time that Insig-mediated ERAD involves the novel ER chaperone Sig-1R and that this ERAD system regulates the degradation of the sphingolipid enzyme CGalT. Our findings therefore suggest that expression of the CGalT enzyme and the production of GalCer can be regulated by sterols inside of the cell. Exactly how the Sig-1R, particularly its chaperone activity, is involved in the Insig-mediated ERAD machinery needs to be examined in the future at a detailed molecular level. Also, how Insig-mediated ERAD recognizes CGalT is unclear at present. Nevertheless, according to our amino acid sequence analysis, the transmembrane domain of rat CGalT contains a cholesterol-binding domain motif of the sequence (L/V)X 1-5 YX 1-5 (K/R) (amino acids V 495 TKFIY 500 RK 502 ), as described in the benzodiazepine receptor (25). Further, we previously found that CGalT is photoaffinity-labeled with [ 3 H]azicholesterol. 4 Thus, it is plausible to speculate that CGalT interacting with sterols may be recognized by Insig (Fig. 9). It is also noteworthy that progesterone receptor membrane component 1 (PGRMC1) was recently found to constitute the binding activity of the sigma-2 receptor (46), another subtype of sigma receptors (47). Interestingly, PGRMC1 was previously found to associate with Insig (48). Thus, the two pharmacologically defined subtypes of sigma receptors, namely Sig-1R and sigma-2 receptor, may form heteropolymeric complexes with Insig to regulate the lipid-sensing ERAD.
Because the level of endogenous CGalT in primary cells and schwannoma cell lines is considerably low, experiments for exploration of detailed molecular mechanisms of CGalT regulation are substantially limited with those cell types. Therefore, we used the CHO cell line stably expressing CGalT as a model to examine the post-translational modification of the enzyme. Therefore, it remains possible that the processing of ectopically expressed CGalT in CHO cells might be different from that of endogenously expressed CGalT. Nonetheless, our data from D6P2T schwannoma (Figs. 3B and 5B) support the notion that the mechanism we found in this study is utilized in cells endogenously expressing CGalT. Furthermore, because amino acid sequences of both Sig-1R and CGalT are highly conserved between mammals (Ͼ90.1% for the Sig-1R and Ͼ93.7% for CGalT between human, mouse, and rat; NCBI COBALT alignment analysis) and the cholesterol and sphingolipid metabolism is omnipresent in vertebrates, the mechanism found in our CHO cell model could be extrapolated to other mammalian cells, such as the human oligodendrocyte. 4 H. Sprong, unpublished data.  associates with Insig to form an ERAD complex at the ER membrane. The association is strengthened by 25-hydroxycholesterol or Sig-1R ligands, such as (ϩ)-pentazocine. In the presence of high sterols (e.g. 25-hydroxysterol), CGalT is recruited to the Sig-1R-Insig machinery for degradation. Other components involved in the Sig-1R-regulated ERAD machinery (e.g. ubiquitin ligases) are not defined. The transmembrane domain of CGalT (a gray box) contains a putative sterol-binding motif that, upon binding to sterols, might be recognized by the Insig-mediated ERAD complex.
It is known that CGalT expression begins at the late progenitor stage of the oligodendrocyte-type 2 astrocyte lineage. The enzymatic product GalCer, which together with sulfatide comprises 27% of total myelin lipids (49), is transported from the ER to the outer leaflet of the oligodendrocyte plasma membrane to serve as a negative regulator of myelination (16). It is interesting that myelin is also highly enriched in cholesterol, the synthesis of which is increased at the same stage of oligodendrocyte differentiation (49). In contrast to GalCer, however, sterols are known to stimulate oligodendrocyte differentiation (50). Sterols and GalCer may therefore serve as an accelerator and a brake, respectively, on the oligodendrocyte differentiation, and our data suggest that the two systems might cross-talk via the Sig-1R-Insig complex. A future study using primary oligodendrocyte progenitors is certainly necessary to support this possibility and to explore a direct relationship between sterol levels and GalCer levels.
Previous studies have demonstrated that the second transmembrane domain of the Sig-1R is involved in the formation of the sterol/ligand-binding domain (23,51). Notably, amino acids that are identical between Sig-1R and yeast C8-C7 sterol isomerase are in fact clustered within the second transmembrane domain of the Sig-1R (23). The segment of those amino acids corresponds to the sterol-binding pocket of a yeast C8-C7 sterol isomerase (23). In the presence of sterols, Insig associates specifically with the so-called sterol-sensing domain of SCAP or HMG-CoA reductase that binds 25-hydroxycholesterol (44). The association between Sig-1Rs and Insig1 is similarly potentiated by 25-hydroxicholesterol (Fig. 7C). Thus, it is conceivable that the second transmembrane domain of the Sig-1R may serve as a sterol-sensing domain that promotes the Insig association (Fig. 7A) and that 25-hydroxicholesterol may serve as a high affinity endogenous Sig-1R ligand. Importantly, however, there are some biochemical differences between Sig-1R-Insig and SCAP-Insig associations. First, under sterol-deprived conditions that can promote nearly complete dissociation of Insig from SCAP (45), the Sig-1R is still able to hold a certain degree of association with Insig. Second, cholesterol and lanosterol have no effect on the Sig-1R-Insig association, at least under our assay condition. Third, the association between Sig-1R and Insig can be altered by Sig-1R ligands as shown by the prototypic Sig-1R ligand (ϩ)-pentazocine. The Sig-1R binds a variety of hydrophobic/amphipathic drugs (e.g. antidepressants) as well as certain endogenous compounds, such as progesterone, sphingolipids (e.g. D-erythro-sphingosine, and monoglycosylceramides), and N,N-dimethyltryptamine (20 -22, 51). Although more studies are needed to confirm whether these synthetic and endogenous compounds similar to (ϩ)-pentazocine also contribute to tightening the association between Sig-1R and Insig, the unique drug/lipid-binding profile of the Sig-1R may explain why Sig-1Rs still maintain a certain degree of the association with Insig under sterol-depriving conditions.
Our results indicate that the Insig1⅐Sig-1R complex is devoid of SCAP. The increased association of Sig-1R with Insig1 leads to the decrease of SCAP associating with Insig1 (Fig. 7A). This may suggest that the association of Sig-1Rs with Insig keeps Insig away from SCAP, thus influencing the equilibrium in the Insig⅐SCAP association. Because the dissociation of Insig from SCAP is a critical step for the departure of the SCAP⅐SREBP complex from ER to Golgi for SREBP activation (44), Sig-1Rmediated sequestration of Insig from SCAP may potentially result in activation of SREBP. Particularly under ER stress, where Sig-1Rs are highly up-regulated (30), the inhibitory action of Sig-1Rs on the Insig⅐SCAP association might become relevant. Indeed, we have found that knockdown of Sig-1Rs decreases the active form of SREBP1 (i.e. the form cleaved at Golgi), whereas the overexpression increases activated SREBP1 in CHO cells. 3 This mechanism may also partly explain why ER stress activates SREBPs (52).
In summary, we found that a novel ERAD complex involving Sig-1R chaperones and Insig regulates the lifetime of CGalT under physiological conditions. The identified ERAD components may provide a mechanistic insight into understanding of the cross-talk between sterol and sphingolipid metabolisms. Last, we would like to mention that CGalT, based on its structure, belongs to the glucuronyltransferase family of ER enzymes (7,39). In collaboration with cytochrome p450 that comprises the phase I drug metabolism pathway, glucuronyltransferases comprise the phase II drug metabolism pathway that plays a crucial role in biotransformation of drugs, sterols, and xenobiotics (39). At present, it is unknown why Sig-1R possesses the unusual binding profile that allows the protein to bind numerous types of drugs and sterols. It will be interesting to see whether Sig-1Rs, which bind xeno-/endobiotics, regulate the lifetime of drug metabolism enzymes that include the glucuronyltransferase family at the ER.