Detergent-Resistant Microdomains Determine the Localization of σ-1 Receptors to the Endoplasmic Reticulum-Mitochondria Junction^S⃞

Materials and Methods.

Reagents for cell culture were purchased from Invitrogen (Carlsbad, CA). Sources of antibodies included the following: anti-N-CAM was from BD Biosciences Pharmingen (San Diego, CA); anti-IP3R3 and anti-BiP were from BD Biosciences (San Jose, CA); anti-Src, anti-cytochrome P450 reductase (CYP450R), and anti-calreticulin were from Santa Cruz Biotechnology (Santa Cruz, CA); anti-ceramide was from Sigma-Aldrich (St. Louis, MO), anti-mitofusin-2 and anti-ERp57 were from Abcam Inc. (Cambridge, MA); anti-ATP synthase inhibitor and anti-cytochrome c oxidase subunit I were from Invitrogen; and anti-green fluorescent protein (GFP) was from Clontech (Mountain View, CA). Anti-Sig-1R antibodies were developed as described previously (Hayashi and Su, 2007). 6-((N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)hexanoyl) sphingosine (NBD-Cer) and N-((4-(4,4-difluoro-5-(2-thienyl)-4-bora-3a,4a-diaza-s-indacene-3-yl)phenoxy)acetyl)sphingosine (BODIPY-Cer TR) were from Invitrogen. Chemicals and lipids were from Sigma-Aldrich or Avanti Polar Lipids (Alabaster, AL). Expression vectors for Mito-DsRed, Mito-GFP, and KDEL-DsRed were purchased from Clontech. The structures of lipids are provided in the supplemental figure.

Cell Culture.

Chinese hamster ovary (CHO; American Type Culture Collection, Manassas, VA) cells were maintained in minimum essential medium-α containing 2 mM GlutaMax (Invitrogen) and 10% heat-inactivated fetal calf serum at 37°C with 5% CO₂. The vector encoding the mouse Sig-1R with an enhanced yellow fluorescent protein (EYFP)-tag on the C terminus was constructed in the pEYFP-N1 vector (Clontech) as described previously (Hayashi and Su, 2003b). CHO cell lines stably expressing EYFP or Sig-1R-EYFP were established by transiently transfecting pEYFP-N1 or pEYFP-Sig-1R vectors followed by colony selections with G418 (Geneticin; Invitrogen). Lipofectamine 2000 (Invitrogen) was used for gene transfection. CHO cells were treated with fumonisin B1 (FB1), methyl-β-cyclodextrin (MβC), or cholesterol-conjugated MβC in culture medium without serum. FB1 was dissolved in ethanol and used at 5 μg/ml. MβC was dissolved in phosphate-buffered saline (PBS) and used at 5 mM with 5 μM lovastatin. Cholesterol-conjugated MβC was prepared by rotating 1 ml of MβC solution (100 mM) with N₂-dried cholesterol (4 mg) overnight at room temperature.

DRM Preparation.

For the sucrose gradient fractionation of DRMs, CHO cells were cultured in a 10-cm dish at 80 to 90% confluence. CHO cells (2 × 10⁶ cells) were lysed at 4°C in 0.5 ml of TNE-Tx buffer (10 mM Tris at pH 7.4, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 5 mM EDTA, 48 Kallikrein inactivation units/ml aprotinin, and 0.5% Tx). After being mixed with an equal volume of 80% sucrose, the cell lysate was subjected to sucrose gradient centrifugation at 100,000g, and a total of 13 fractions was collected from the top, as described previously (Hayashi and Su, 2003b). The same volume from each fraction was applied to SDS-PAGE.

To prepare the Tx-insoluble pellet enriched with DRMs and the Tx-soluble supernatant by differential centrifugations, cell membranes were incubated in TNE-Tx for 30 min at 4°C. After sonication (10 s, three times), samples were centrifuged at 12,000g for 20 min. The supernatant was further centrifuged at 100,000g for 1 h in a ⁵⁰Ti rotor. The pellet and the supernatant were collected as Tx-insoluble (pellet) and Tx-soluble (supernatant) fractions, respectively.

MAM Fractionation.

The MAM fraction was prepared as described previously (Rusiñol et al., 1994). In brief, CHO cells were grown on two 15-cm dishes at 90 to 100% confluence. Harvested cells were homogenized by a glass Dounce homogenizer in a homogenization buffer (0.25 M sucrose and 10 mM HEPES/KOH, pH 7.4). The homogenate was centrifuged twice at 500g for 5 min to yield the P1 nuclear fraction. The supernatant was spun down at 10,300g for 20 min to yield crude mitochondria membranes as a pellet (P2 fraction). The supernatant was centrifuged at 100,000g for 1 h in a 50Ti rotor to obtain the P3 microsomal and cytosolic fractions. Crude mitochondrial membranes were suspended in 0.5 ml of isolation medium (250 mM mannitol, 5 mM HEPES/KOH pH 7.4, and 0.5 mM EGTA/KOH), layered on Percoll solution [225 mM mannitol, 25 mM HEPES/KOH, pH 7.4, 1 mM EGTA/KOH, and 30% (v/v) Percoll (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK)], and centrifuged at 95,000g for 30 min in an SW 55Ti rotor. Purified mitochondrial and MAM fractions were collected and washed twice with an isolation medium by centrifugation.

Immunocytochemistry and Confocal Microscopy.

Immunocytochemistry and confocal microscopy were performed as described earlier (Hayashi and Su, 2003b). Primary antibodies were used at 1:100 for anti-Sig-1R or 1:50 for anti-IP3R3. The filipin staining was performed according to a previous report (Hayashi and Su, 2003b).

Lipid Overlay Assay.

CHO cells were grown in a 15-cm dish at 90% confluence. Cells were harvested in ice-cold PBS and centrifuged at 2000g for 10 min. The cell pellet was suspended in 1 ml of 50 mM Tris, pH 7.4, containing 0.2% CHAPS. The cell lysate was rotated at 4°C for 2 h, followed by a centrifugation at 100,000g in a 50Ti rotor for 1 h. The supernatant was filtered by a Millipore 0.22-mm filter unit (Millipore, Billerica, MA). The total protein concentration was measured with a BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA). The lysate was stored at −80°C until use.

Lipids were dissolved in a chloroform/methanol mixture (2:1) and spotted on a Hybond C nitrocellulose membrane (200–10,000 pmol/spot; GE Healthcare). The membrane was air-dried for 1 h at room temperature. The membrane was blocked with TNC buffer (50 mM Tris, pH 7.4, 150 mM NaCl, and 0.2% CHAPS) containing 30% nonfat dry milk and 5% chicken ovalbumin at 4°C for 6 h with gentle shaking. After washings twice with TNC buffer, the membrane was incubated with the CHAPS cell lysate (28–60 μg of protein/cm²) in TNC containing 150 mM NaCl and 4% BSA at 4°C for 16 h with shaking at 200 rpm. After three washings with TNC containing 3% BSA (each for 10 min), the membrane was incubated with either anti-full-length GFP antibodies (1:1500) or anti-Sig-1R antibodies (1:1000) in TNC with 3% BSA for 3 h at room temperature or 16 h at 4°C. After extensive washings with TNC containing 3% BSA, Sig-1R complexes associated with lipids were detected by horseradish peroxidase-conjugated secondary antibodies, followed by WestFemto chemiluminescence (Thermo Fisher Scientific). Images were captured with Kodak Image Station 440CF (Carestream Health, Rochester, NY).

Lipid Assays.

Total lipids were extracted by the Bligh and Dyer method. Free cholesterol in the sample was measured enzymatically by the Free-Cholesterol E kit (Wako Bioproducts, Richmond, VA) or the Cholesterol Assay kit (Sigma-Aldrich). For high-performance thin-layer chromatography (HPTLC; Merck, Darmstadt, Germany), lipid extracts dissolved in the chloroform/methanol mixture (2:1) were spotted on an HPTLC plate followed by resolution with the hexane/ether/acetic acid mixture (16:4:0.2) for cholesterol and neutral lipids and the chloroform/methanol/water mixture (13:5:0.5) for sphingolipids and phospholipids. Lipids were visualized on HPTLC plates using a ferric chloride spray (for cholesterol) or a diphenylamine-aniline spray (for neutral glycosphingolipids). Phospholipids and ceramides were visualized under UV light after spraying 0.2% 8-anilinonaphthalene-1-sulfonate (ANS) in ethanol. The ganglioside-enriched fraction was obtained by applying total lipid extracts to a silicic acid column fractionation followed by Dowex 50 (Na⁺ form) and Sep-Pak C18 column chromatography (Waters, Milford, MA). Gangliosides were resolved by HPTLC with chloroform/methanol/0.25% CaCl₂ (55:45:10) and visualized by resorcinol or diphenylamine-aniline spray. After separation from other lipids, the ceramides were extracted from an HPTLC plate (Merck) and spotted on a Hybond-C membrane (GE Healthcare). The membrane was blocked with 10% nonfat milk (Bio-Rad Laboratories, Hercules, CA) in PBS followed by incubation with ceramide antibodies (1:200) in the same solution at room temperature for 16 h. After five wash steps with PBS, ceramides on the membrane were visualized as described in the lipid overlay assay.

SDS-PAGE and Silver Staining.

Samples were dissolved in a 2× Laemmli sample buffer after protein quantifications (BCA assay kit; Pierce, Rockford, IL). Proteins were separated by SDS-PAGE as described previously (Hayashi and Su, 2003b). Proteins in the acrylamide gel were electrically transblotted to a polyvinylidene difluoride membrane (Bio-Rad Laboratories) in a Towbeen buffer and detected by immunoblotting. Silver stainings were performed according to the company's instruction (Silver Stain Plus; Bio-Rad Laboratories).

Living-Cell Labeling with Fluorescent Ceramides.

Fluorescent ceramides (NBD-Cer and BODIPY-Cer TR) were dissolved in the chloroform/methanol mixture (19:1) to a 1 mM concentration. After they were completely dried under N₂, the lipids were dissolved in 200 μl of ethanol. This ethanol solution was injected into 4 ml of Hanks' balanced salt solution supplemented with 10 mM HEPES, pH 7.4, and 12.5 μM BSA under vigorous vortexing. Yielded ceramide-BSA complexes were added at 4°C for 30 min to the culture medium at the final concentration of 1.25 μM followed by washings with ice-cold PBS three times. In the chase experiment, ceramide-labeled cells were returned to the culture incubator with a normal medium. Cells were fixed with 4% paraformaldehyde for 15 min followed by two wash steps with ammonium chloride (5 mM) for 5 min twice and with PBS once.

ER Proteins Associate with DRMs.

We examined whether Sig-1Rs and other ER proteins could be found in DRMs extracted from CHO cells. In agreement with previous studies using different types of cells (Hayashi and Su, 2003b, 2004), the sucrose gradient centrifugation of the Tx lysate found that a portion of Sig-1R was indeed present in DRM fractions (Fig. 1A, fractions 4–7). As seen in NG108 cells (Hayashi and Su, 2003b), DRMs associated with Sig-1R showed lower buoyancy in the sucrose gradient (fractions 4–7) than those of Src (fractions 2–5), the protein associated mainly with plasma membrane DRMs. Some ER proteins that are enriched at the MAM, including IP3R3 and BiP, were also present in DRM fractions, although IP3R3 was present at a much lower level (Fig. 1A). Mitofusin-2, the protein tethering MAM to mitochondria (de Brito and Scorrano, 2008), was also detected in DRMs (Fig. 1A). Some ER proteins such as ERp57 were completely excluded from DRMs (Fig. 1A).

Unique Detergent-Solubility Profile of DRMs Associating with Sig-1R.

Although Sig-1Rs were consistently detected in DRM fractions in different cell types, the proportion of Sig-1Rs present in DRM fractions is only 10 to 20% of total Sig-1R proteins (Fig. 1A). On the other hand, at least 50% of Src or caveolin was fractionated in DRMs (Fig. 1, A and B). However, it is noteworthy that recent studies demonstrate that DRMs at internal membranes have lower resistance to extraction by the prototypic detergent Tx but exhibit somewhat higher resistance to other types of detergents (Marwali et al., 2003), implying that there are different subclasses of DRMs. Thus, although the marginal level of Sig-1R was present in Tx-insoluble DRMs, the result may not merely imply few Sig-1Rs associated with DRMs. We therefore conducted sucrose gradient centrifugations with different types of detergents. More than 50% of Src and caveolin consistently associated with DRMs in the presence of any type of detergent tested (Fig. 1B). The highest detergent resistance for both proteins was seen with Tx. The detergent resistance was slightly decreased with Brij-35 or Lubrol WX (Fig. 1B). In contrast, detergent resistance of DRMs associated with Sig-1R and IP3R3 dramatically changed depending on the type of detergent used. Among them, DRMs associated with Sig-1R showed remarkable resistance against extraction with Triton X-114, in which more than 80% of Sig-1Rs were present in DRM fractions (Fig. 1B). Buoyancy of the DRMs was significantly higher in Triton X-114 (fractions 2–4) than that in Tx (Fig. 1B). DRMs associated with IP3R3 exerted significantly higher resistance to extraction by CHAPS, Triton X-114, Brij-35, or Lubrol WX compared with Tx (Fig. 1B). These findings suggest that the high proportion of Sig-1R may be associated with DRMs at the ER, but those are probably much less resistant to extraction by Tx.

The MAM Is the Specialized ER Subdomain Accommodating DRMs.

We examined whether the Triton X-114-insoluble DRMs associated with Sig-1Rs is derived from the MAM. As a result, MAM and microsomes were purified from CHO cells, and enrichment of Sig-1R-containing DRMs and DRM-forming lipids was examined in those membranes. In agreement with a previous study (Hayashi and Su, 2007), Sig-1Rs were enriched in the MAM fraction (Fig. 2A). The levels of IP3R3 and certain ER chaperones such as BiP and calreticulin were also the highest in the MAM fraction (Fig. 2A). ER chaperones involved in vesicular transport of proteins were also detected in the cytosolic fraction (Fig. 2A). The ER chaperone ERp57 was more enriched in the microsomal fraction (Fig. 2A). The microsomal protein CYP450R was present exclusively in microsomal and nucleus fractions (Fig. 2A). Mitofusin-2 was present both in mitochondrial and MAM fractions (Fig. 2A). Mitochondria-specific proteins were detected only in the mitochondrial fraction (Fig. 2A). The distribution patterns of these organelle markers thus verified the sufficiently high purity of MAM and microsomal membranes.

When each fraction was solubilized with Tx, a marginal but still detectable level of Sig-1R was present in DRMs from the MAM fraction but not in those from microsomes (Fig. 2B, top). As shown in Fig. 1B, a much higher level of Sig-1Rs was present in DRMs extracted from the MAM with Triton X-114 compared with the extract with Tx (Fig. 2B, bottom). Sig-1Rs were present in DRMs only from the MAM fraction (Fig. 2B), strongly indicating that the Sig-1R-associated DRMs are mostly derived from the MAM. Furthermore, the silver staining showed a much larger amount of proteins present in DRMs derived from the MAM compared with DRMs from microsomes (Fig. 2C), suggesting that the MAM accommodates a higher level of DRMs than the microsome. Indeed, HPTLC revealed that the MAM contains higher levels of DRM-forming lipids, such as cholesterol, ceramides, glucosylceramide (GlcCer), and sphingomyelin, compared with microsomes (Fig. 2D). The higher level of ceramides at the MAM was further confirmed by the lipid overlay assay using anti-ceramide antibodies (Fig. 2D, bottom). Among phospholipids, phosphatidylserine and phosphatidylcholine were present in both membranes at similar levels, but phosphatidylethanolamine was higher in the MAM fraction (Fig. 2D). Lactosylceramide and GM3, the major ganglioside in CHO cells, were not detected in either MAM or microsomes under our HPTLC conditions (data not shown).

The Sig-1R Complex Associates with Specific Sphingolipids.

It is well known that DRMs prepared from the plasma membrane mostly consist of cholesterol and complex sphingolipids such as gangliosides. In contrast, DRMs from the MAM seem to be more enriched in simple sphingolipids such as ceramides and GlcCer (Fig. 2D). To examine whether the simple sphingolipids might contribute to the formation of DRMs and the localization of Sig-1R at the MAM, we examined whether Sig-1Rs preferentially associate with those MAM-enriched lipids by using a modified lipid overlay assay. The assay is particularly advantageous in studies using highly hydrophobic molecules as ligands (e.g., lipids), because the assay avoids problematic insolubility/aggregation of ligands and partitioning of ligands into membranes. Because CHAPS (0.2%) solubilization is shown to maintain the ligand-binding property of Sig-1Rs well (McCann and Su, 1991), we used CHAPS cell lysates in the assay. It should be noted, however, that the assay using the CHAPS whole-cell lysate might detect the direct association between lipids and Sig-1R molecules as well as the association between lipids and protein complexes composed of Sig-1R (designated here as the “Sig-1R complex”). The commercially available SphingoStrips Array membrane (200 pmol/spot; Invitrogen) demonstrated that Sig-1R-EYFP complexes from the CHAPS lysate of Sig-1R-EYFP-overexpressing CHO cells specifically associate with ceramide and sulfatide of 15 different lipids immobilized on the array membrane (Fig. 3A). No association of Sig-1R-EYFP complexes with sphingosine, sphingomyelin, or sphingosine-1-phosphate was observed under these conditions (Fig. 3A); however, the association between Sig-1R complexes and sphingosine was consistently detected when the lipid on the array was increased to 1 nmol/spot. Solubilized EYFP complexes obtained from CHO cells stably overexpressing EYFP did not associate with any lipids on the array (Fig. 3A).

To further examine lipids interacting with Sig-1R complexes, we prepared a nitrocellulose array membrane with 30 different lipids immobilized (200 pmol/spot). The assay found that Sig-1R-EYFP complexes associated preferentially with sphingolipids with few or no sugar moieties but not with gangliosides [association potency: lactosylceramide > galactosylceramide > glucosylceramide > sulfatide > hydroxy or nonhydroxy ceramide (from bovine brains) ⋙ gangliosides] (Fig. 3B). The two oligodendrocyte-specific glycosphingolipids, galactosylceramide and sulfatide, were both capable of associating with Sig-1R-EYFP complexes (Fig. 3B). It is interesting that among sterols tested under this condition, only lathosterol showed a detectable yet low association with Sig-1R-EYFP complexes (Fig. 3B). This result is in agreement with the fact that the Sig-1R shares 66% similarity with a yeast C8–C7 sterol isomerase catalyzing the conversion of zymosterol to lathosterol (Hanner et al., 1996). Endogenous Sig-1R complexes also showed a similar profile in the lipid binding, but with lower signals, thus prohibiting the detection of some interactions (Fig. 3C). This seemed to be mainly due to the lower protein level of endogenous Sig-1R complexes in the lysate. Increasing the total protein level in the lysate (from 28 to 60 μg of protein lysate/cm² of membrane), accompanied by increased lipids on the membrane (400 ≥ pmol/spot), ensured the reproducible detection of endogenous Sig-1R complexes associated with ceramides or lathosterol (Fig. 3D).

It has been demonstrated that Sig-1Rs associate with certain steroids and cholesterol (Su et al., 1988; Pal et al., 2007; Palmer et al., 2007). Indeed, the assay detected the association of Sig-1Rs with some sterols (lathosterol ≫ progesterone ≥ testosterone ⋙ cholesterol), but much higher amounts of lipids (10 nmol/spot) were necessary, except for lathosterol, to obtain reliable signals (Fig. 3E).

d-erythro-Sphingosine associated with Sig-1R complexes (Fig. 3F). However, in contrast to the previous report (Ramachandran et al., 2009), the interaction between the Sig-1R complex and ceramide was consistently detected. Ceramides possessing a long fatty acid acyl chain (i.e., C16–18), which is naturally present, associated better with Sig-1R complexes, whereas synthetic ceramides with short acyl chains (e.g., C6, C2) failed to associate with Sig-1R-EYFP or Sig-1R complexes (Fig. 3, B and F). The estimated order of potencies of ceramides/sphingosine in associating with Sig-1R complexes was C18-ceramide = C16-ceramide > d-erythro-sphingosine > C22-ceramide ⋙ C6-, C2-ceramides. The structures of these lipids are provided in the supplemental figure.

Because the MAM is shown to possess high ceramide biosynthesis activity (Bionda et al., 2004), we focused particularly in the following studies on ceramides to determine whether the lipid may play a role in the localization of Sig-1R to the MAM.

MAM Is Highly Capable of Accumulating Ceramides.

An early study demonstrates that exogenously applied fluorescent NBD-Cer distributes heterogeneously in living cells. NBD-Cer accumulate initially at the plasma membrane and ER and then at mitochondria in living cells when they are incubated at 2°C, a temperature at which energy-dependent vesicle transport of lipids is shut down (Lipsky and Pagano, 1983). This finding clearly illustrates that the intermembrane diffusion of ceramides from the ER to the mitochondrial membranes is restricted at the ER-mitochondria junction. Thus, the MAM might be highly capable of accumulating ceramides to serve as the checkpoint regulating the diffusion of ceramides from the ER to the mitochondria. To examine whether the MAM accumulates ceramides, CHO cells were incubated with NBD-Cer at 4°C, and the cellular distribution was monitored by confocal microscopy. As expected, a striking accumulation of NBD-Cer was observed at MAM-like small structures (Fig. 4A, large arrows) juxtaposed to mitochondria that bear the typical spaghetti-like structure (Fig. 4A, small arrows). NBD-Cer also accumulated at the cisternae of the ER (Fig. 4A, arrowheads) and the mitochondria, but at much lower levels. The observed NBD signals in living CHO cells seem to represent mostly the distribution of NBD-Cer, but not that of those metabolites, because few NBD-Cer were metabolized to GlcCer in the living cells under 4°C (Fig. 4A, lane 1). The structures accumulating NBD-Cer clearly apposed mitochondria that were visualized by expressing Mito-DsRed (Fig. 4B), and the staining pattern was identical with that of Sig-1R immunostaining (Fig. 4C). The other fluorescent ceramide (BODIPY-Cer TR) accumulated similarly at the MAM-like structures that colocalized with Sig-1R-EYFP (Fig. 4D). Membrane fractionation confirmed that the level of NBD-Cer is the highest in the MAM fraction (Fig. 4E).

To examine whether the accumulation of ceramides at the MAM is adversely affected by the low-temperature incubation, under which vesicular transports of lipids from ER to Golgi are stalled, cells labeled with NBD-Cer were returned to the normal culture condition at 37°C. Under these conditions, NBD-Cer accumulated at the Golgi apparatus but decreased at the ER (Fig. 4F), indicating that the energy-dependent transport of ceramides from ER to Golgi was now resumed. Under these conditions, a portion of NBD-Cer was converted to GlcCer (Fig. 4A, lane 2 of TLC); the step is known to be mainly catalyzed at the Golgi. It is important to note that even under these conditions, the MAM still accommodated a significant level of NBD-Cer, suggesting that the MAM may possess the innate ability to accumulate ceramides at the locus.

Although the distribution of NBD-Cer may not necessarily represent precise cellular dynamics of endogenous ceramides, the finding that MAM can attract NBD-Cer and accommodate the high level of endogenous ceramides (Fig. 2D) suggests that Sig-1R complexes that possess affinities for ceramides might be preferentially recruited to the ER subdomain.

Sig-1Rs Require Ceramides for Their Partitioning into DRMs.

To examine the importance of ceramides in localization of Sig-1Rs at the MAM, we used pharmacological approaches that enable the manipulation of lipid contents and DRM integration at the ER membranes. MβC is a well known compound that depletes free cholesterol in biological membranes, thus disrupting DRMs (Simons and Toomre, 2000). Because water-soluble MβC does not readily penetrate plasma membranes, we assumed that MβC might extract cholesterol initially at the cell surface. On the other hand, the long exposure to MβC is known to cause a nearly complete depletion of cholesterol in the whole cell (Naslavsky et al., 1999). As a result, CHO cells were treated with MβC for different periods of time to preferentially destroy DRMs either at the cell surface or in the entire cell. Incubation with MβC for 15 min depleted 30% of total cholesterol in CHO cells, whereas incubation for 60 min depleted approximately 80% of total cholesterol (Fig. 5A). Filipin staining confirmed that a short exposure to MβC reduced cholesterol mainly on the cell surface, but a longer exposure decreased cholesterol in the entire cell (Fig. 5B). Next, we examined the effect of MβC on partitioning of proteins to DRMs. The treatment with MβC for 15 min significantly decreased Src associated with DRM but not Sig-1R (Fig. 5E), suggesting that Sig-1R-residing DRMs are intracellularly localized. On the other hand, a long exposure to MβC (for 60 min) significantly decreased Sig-1R in DRMs (Fig. 5E), indicating that Sig-1R-residing microdomains also require cholesterol for stabilization/formation.

FB1, which blocks the conversion of sphinganine to dihydroceramide, is a potent inhibitor of the ceramide biosynthesis at the ER (Naslavsky et al., 1999). Although FB1 rapidly inhibits the synthesis of ceramides at the ER, FB1 requires prolonged incubation (e.g., 24–72 h) to deplete gangliosides or sphingomyelin in the cell because of their slow turnover rates and/or the alternative synthesis pathways independent of de novo synthesis of ceramides (Naslavsky et al., 1999). Because the MAM serves as the locus for ceramide biosynthesis (Bionda et al., 2004) and it is assumed that the major pool of ER ceramides relies on activity of the de novo synthesis at the ER (Sprong et al., 2001), we expected that FB1 might efficiently and rapidly decrease the focal concentration of ceramides at the MAM. In fact, FB1, even with its treatment for 4 h, significantly reduced ceramides at the MAM without largely affecting total levels of ganglioside GM3 in CHO cells (Fig. 5C). As shown previously (Naslavsky et al., 1999), prolonged incubation with FB1 (24 h) significantly decreased GM3 in CHO cells (data not shown). Next, we examined whether the depletion of ceramides caused by FB1 might affect the association of Sig-1R with DRMs. As shown in Fig. 5E, treatment with FB1 for 4 h significantly decreased Sig-1R associated with DRMs. However, little effect on Src association with plasma membrane DRMs was observed, suggesting that ceramides are important molecules for partitioning of Sig-1R to DRMs.

FB1 may also reduce metabolites such as GlcCer, which might be components of DRM at the MAM, immediately downstream of the ceramide synthesis. GlcCer is synthesized by a GlcCer synthase at the cytoplasmic surface of the Golgi apparatus (Sprong et al., 2001). The GlcCer synthesis depends largely on ceramides supplied by the de novo synthesis at the ER (Sprong et al., 2001). Thus, inhibition of the GlcCer synthesis by (±)-threo-1-phenyl-2-decanoyl-amino-3-morpholino-1-propanol hydrochloride (PDMP) decreased GlcCer, whereas it concomitantly accumulated ceramides in the cell (Fig. 5D). We next examined whether PDMP affects the association of Sig-1R with DRMs. PDMP (4 h) unexpectedly increased Sig-1R association with DRMs (Fig. 5E), indicating that ceramides, but not GlcCer, seem to serve as a positive regulator of Sig-1R for their partitioning into DRMs.

DRMs Anchor Sig-1R at the ER-Mitochondria Junction.

To assess the importance of DRMs in MAM-specific distribution of Sig-1R, we tested whether disruption of DRMs may affect the cellular localization of Sig-1R. Treatment with MβC (60 min) or FB1 (4 h) caused a dramatic redistribution of Sig-1R-EYFP from MAM to ER cisternae (Fig. 6A), resulting in a significant reduction in Sig-1R-EYFP colocalization with Mito-DsRed (Fig. 6B). MβC presaturated with cholesterol (MβC-Chol), which provides free cholesterol to the cell, failed to promote the translocation of Sig-1R-EYFP (Fig. 6A). MβC-Chol, however, did cause swelling of the mitochondria (Fig. 6A; round mitochondria shown in red), as reported previously (Colell et al., 2003). This redistribution seemed to be due to the translocation of Sig-1R-EYFP at ER membranes but not due to structural disorganization of ER membranes caused by MβC or FB1, because the treatments did not significantly alter the overall structure of ER membranes or the spatial interaction between ER membranes and mitochondria at least under observations with confocal microscopy (Fig. 6, C and D, middle). The confocal analysis revealed that these treatments do not alter colocalization of fluorescence signals derived from Mito-GFP and KDEL-DsRed (Fig. 6C). The treatment with MβC or FB1 also caused a similar redistribution of IP3R3 originally present at the MAM (Fig. 6D). Thus, cholesterol and ceramides are important membrane components at the ER for anchoring Sig-1R at the MAM.