Oxysterol Binding Protein–related Protein 9 (ORP9) Is a Cholesterol Transfer Protein That Regulates Golgi Structure and Function

Plasmids

Tetracycline-inducible expression vectors for ORP9S and ORP9L (pTRE-ORP9L-V5, pTRE-ORP9S-V5), as well as FFAT domain mutants (pTRE-ORP9L-FF/AA-V5 and pTRE-ORP9S-FF/AA-V5) were previously described (Wyles and Ridgway, 2004). The ORP9 PH (pTRE-ORP9L-R22E-V5) and sterol-binding domains (pTRE-ORP9L-Δ375-378-V5) were mutagenized using the QuikChange II XL kit (Stratagene, La Jolla CA). pcDNA3.1-VSVG-tsO45-(myc)3 was kindly given by Dr. Johnny Ngsee (Ottawa Health Research Institute, University of Ottawa, ON, Canada). Bacterial expression and purification of GST-VAP was previously reported (Wyles et al., 2002).

Cell Culture and Transfections

Chinese hamster ovary (CHO)-K1 and CHO Tet-on cells were cultured in DMEM containing 5% FBS and 34 μg proline/ml (medium A). For filipin fluorescence experiments, cells were also cultured in DMEM containing 5% lipoprotein-deficient serum (LPDS) and proline. Expression of ORP9 in Tet-on cells were induced in medium A containing doxycycline (1 μg/ml) for times indicated in figure legends.

ORP9L expression was knocked down by transfection with a pool of three short interfering RNA (siRNA) duplexes (25 nM each, Dharmacon, Lafayette, CO) or a nontargeting siRNA (100 nM) with Trans-IT TKO transfection reagent (Mirus, Madison, WI) for 48 h (Lessmann et al., 2007).

Expression and Purification of Recombinant OSBP and ORP9

A EcoRI-BamHI fragment encoding the ORP9 PH domain (aa 1-104) was generated by PCR amplification, cloned into pGEX-3X, and transformed into BL21 Escherichia coli. Expression of GST-ORP9-PH and GST-OSBP-PH was induced with 100 μM IPTG for 3 h at 25°C and purified by glutathione-Sepharose affinity chromatography (Wyles et al., 2002).

Full-length, C-terminal His-tagged OSBP, ORP9L, ORP9L Δ375-378 (ΔSB), and OSBP PH-RR_109,110EE (RR/EE) were expressed and purified from baculovirus-infected Sf21 cells. cDNAs were amplified by PCR and cloned into pENTR/D-Topo (Invitrogen). Inserts were verified by sequencing, inserted into linearized BaculoDirect (C-terminal V5-His tagged) by recombination, and transfected into Sf21 cells. For large-scale expression and purification, Sf21 cells were infected at an MOI of 0.1 for 1 h at 20°C, resuspended in 200–500 ml of Sf-900 II media with 0.5% FCS and incubated with shaking at 27°C. After 72 h, cells were collected by centrifugation at 300 × g for 10 min, resuspended in one-tenth volume of phosphate buffer (50 mM potassium phosphate and 300 mM NaCl, pH 7.4) containing 50 mM imidazole and EDTA-free protease inhibitors, and lysed. OSBP or ORP9L were purified from the supernatant fraction with Talon metal-affinity resin (Clontech, Palo Alto, CA). OSBP required an additional purification step on DEAE-Sepharose. Purified recombinant proteins were stored at −80°C.

Fluorescence Microscopy

After transfections, cells were cultured to 60% confluence on glass coverslips, washed with phosphate-buffered saline (PBS), fixed in 4% (wt/vol) paraformaldehyde for 10 min, and permeabilized in 0.05% (vol/vol) Triton X-100 at −20°C for 10 min before probing with primary and secondary antibodies. Images were captured using a Zeiss LSM 510 Meta laser scanning confocal microscope (0.8-μm sections; Thornwood, NY) using a 100× oil emersion objective (NA 1.4).

CHO cells were fixed as described above, incubated with ORP9L primary and secondary antibodies, and then with filipin (50 μg/ml) for 1 h. Filipin fluorescence images were captured at identical exposure times using a Zeiss Axiovert 200M inverted microscope (Thornwood, NY) equipped with a 63× oil immersion objective (NA 1.4) and axioCam HRm CCD camera. Using NIH Image J software (http://rsb.info.nih.gov/ij/), the background and threshold on image fields of 20–30 cells were normalized and then subjected to fluorescence intensity (integrated density/area) analysis. Intensity values were normalized to cell number and expressed relative to siNT-transfected CHO cells or uninduced CHO Tet-on cells cultured in media A (FCS).

Phospholipid Overlay Assay

Lipid (100 pmol) dissolved in chloroform:methanol:water (1:2:0.8, vol/vol) were spotted on Hybond-C nitrocellulose membrane and dried at room temperature for 1 h. Membranes were incubated with blocking buffer (Tris-buffered saline, 3% fatty acid free BSA, and 0.1% Tween-20 (vol/vol) for 2 h. Membranes were incubated with 100 nM glutathione S-transferase (GST), GST-ORP9-PH, or GST-ORP9-PH-R22E for 1 h in blocking buffer, incubated with an anti-GST monoclonal followed by a goat anti-mouse HRP secondary antibody, and detected by ECL. Full-length OSBP or ORP9L (100 nM) were incubated with HyBond-C–immobilized lipids under the same conditions and detected with monospecific antibodies.

045-Vesicular Stomatitus Virus Glycoprotein EndoH Processing Assay

CHO-K1 cells were transfected with 75 nM siNT or siORP9 for 48 h before transfection with pcDNA3.1-VSV-G-myc3 using Lipofectamine 2000. CHO Tet-on cells were treated with or without doxycycline (1 μg/ml) for 24 h and then transfected with pcDNA3.1-VSV-G-myc3 using Lipofectamine 2000. CHO-K1 or CHO Tet-on cells were then shifted to 40°C for 16 h and incubated with 20 μg/ml cyclohexamide for an addition 20 min before shifting to 32°C for the times indicated in figure legends. Cells were harvested in PBS, collected by centrifugation, sonicated in 10 mM Tris-HCl (pH 7.4), and centrifuged at 100,00 × g for 30 min. Pellets were resuspended in 0.1% SDS, 50 mM sodium citrate (pH 5.5), and incubated for 16 h with 25 mU EndoH at 37°C overnight. Samples were precipitated in 10% trichloroacetic acid and glycosylated forms of vesicular stomatitus virus glycoprotein (VSVG) was analyzed by SDS-PAGE and immunoblotting using an anti-myc monoclonal.

Sterol-binding Assays

25-[³H]Hydroxycholesterol binding by OSBP and ORP9L was assayed by the charcoal-dextran method in the presence or absence of a 40-fold molar excess of unlabeled sterol (Kandutsch and Shown, 1981; Ridgway et al., 1992). [³H]Cholesterol binding to OSBP and ORP9L could not be quantified using the charcoal-dextran method because of high background. Instead, binding assays contained 20 mM HEPES (pH 7.4), 150 mM KCl, and 0.05% Triton X-100 (to disperse [³H]cholesterol). After incubation for 2 h at 20°C, OSBP or ORP9L were bound to Talon metal-affinity resin for 15 min and washed three times with HEPES/KCl buffer, and bound sterols released with 150 mM imidazole. Specific binding was determined in the presence of a 40-fold molar excess of unlabeled cholesterol.

The capacity of OSBP and ORP9L to bind and extract sterols from liposomes was assayed as follows. PC liposomes (0.5 mM) with 1 mol% [³H]cholesterol (specific activity 720 dpm/pmol) were prepared by drying down lipids in chloroform under nitrogen and rehydration in 25 mM HEPES (pH 7.4) and 150 mM NaCl for 1 h at room temperature. Liposomes (400-nm diameter) were prepared by filter extrusion using the Lipofast system (Avestin, Ottawa, ON, Canada) and incubated with recombinant OSBP or ORP9L (1:1; mol protein/mol sterol) for 30 min at 25°C. Liposomes were sedimented by centrifugation at 100,000 × g for 25 min at 4°C and radioactivity in the supernatant was measured by liquid scintillation counting. Specific extraction of [³H]cholesterol from liposomes was determined by subtraction of background radioactivity in the absence of OSBP or ORP9L. The distribution of OSBP and ORP9L in supernatant and pellet fractions were analyzed by SDS-PAGE and Coomassie staining.

[³H]Cholesterol transfer assays were based on previous methods (Kasper and Helmkamp, 1981; Raychaudhuri et al., 2006). Donor PC liposomes (400-nm diameter) containing 10 mol% PE, 1 mol% cholesterol with or without 10 mol% PI-4P, PI-3P, or PI-5P and acceptor PC liposomes (400-nm diameter) containing 10 mol% PE, 10 mol% lactosyl-PE with or without 10 mol% PI4P, PI-3P, or PI-5P were prepared by extrusion. Before the assay, liposomes were cleared by centrifugation at 13,000 rpm for 5 min. The transfer assay consisted of 20 μl of acceptor and donor liposomes (100 pmol cholesterol), recombinant OSBP or ORP9L (100 pmol) in 120 μl of 25 mM HEPES (pH 7.4), 150 mM NaCl, 1 mM EDTA, and 3 μg of fatty acid–free BSA. After incubation at 25°C for 25 min, 30 μg of R. communis agglutinin was added on ice for 15 min and acceptor liposomes were sedimented at 15,000 rpm for 5 min. Radioactivity in the supernatant was measured by scintillation counting and background (absence of OSBP or ORP9L) was subtracted.

ORP9L Is a PI-4P–dependent Cholesterol Transfer Protein

Endogenous ORP9L is primarily on the Golgi apparatus (Wyles and Ridgway, 2004), but whether the PH and sterol-binding domains regulate this interaction is unknown. To establish the PI-phosphate–binding specificity of the OPR9L PH domain, a lipid-overlay assay was used to compare the GST-PH domains of ORP9L and OSBP, as well as purified, full-length His-tagged ORP9L and OSBP expressed and purified from insect cells. Purified GST-fusion proteins (Figure 1A) and ORP9L and OSBP (Figure 1B) had the predicted molecular mass on SDS-PAGE and were incubated with phosphorylated PIs immobilized on nitrocellulose membranes (Figure 1C). GST-ORP9-PH bound to phosphatidylinositol 3-phosphate (PI-3P), PI-4P, and phosphatidylinositol 5-phsophate (PI-5P), but not other 4-phosphorylated derivatives. GST-OSBP-PH displayed a similar pattern but with reduced binding to PI-5P and increased binding of PI-4,5 bisphosphate. The ORP9L PH domain contains several basic residues (R11, R22, and R48) that were implicated in PI-4P and PI-4,5 bisphosphate binding by the related OSBP, CERT, and FAPP PH domain (Perry and Ridgway, 2005). Mutation of one of these arginine residues (ORP9 PH-R22E) dramatically reduced binding to all phosphorylated PIs (Figure 1C). ORP9L and OSBP had similar binding specificities as the corresponding GST-PH domains, suggesting that this activity resides solely in the PH domain. Binding of OSBP and ORP9L to phosphatidylcholine (PC) liposomes containing increasing PI-4P was determined by sedimentation (Figure 1D). By this method, ∼50% of ORP9L was associated with liposomes containing 10 mol% PI-4P, compared with 80–90% for OSBP.

The sterol-binding domain of ORP9 is homologous to OSBP and other ORPs yet was reported not to bind photoderivatives of 25-hydroxycholesterol and cholesterol (Suchanek et al., 2007). As well, it is unclear if ORP9L or OSBP can transfer sterols from donor to acceptor membranes. This was investigated by comparing the in vitro sterol-binding, extraction, and transfer activity of recombinant OSBP and ORP9L (Figure 2). As expected, OSBP bound [³H]25-hydroxycholesterol (Figure 2A) and [³H]cholesterol (Figure 2B) under conditions where the ligand was dispersed in solution or 0.05% Triton X-100, respectively. ORP9L did not bind either sterol under these conditions (Figures 2, A and B) nor over a range of ligand concentrations (results not shown).

Next, an assay was used that measured extraction of [³H]cholesterol from PC/cholesterol liposomes by measuring soluble radioactivity after centrifugation. Extraction of cholesterol was linear with increasing ORP9L protein (Figure 2C). Approximately 35% of liposome [³H]cholesterol was extracted into the supernatant by 200 pmol of ORP9L (Figure 2C) compared with extraction of only <5% liposome phospholipid (results not shown). To demonstrate that cholesterol extraction required the sterol-binding domain, an ORP9L mutant with a deletion between the α-helical lid and β1-sheet of the sterol-binding fold (Δ375-378, referred to hereafter as ΔSB) was assayed for [³H]cholesterol extraction activity. A similar mutation in OSBP and ORP4 reduced cholesterol and 25-hydroxycholesterol binding by >90% (Perry and Ridgway, 2006; Wyles et al., 2007). The activity of ORP9L ΔSB was reduced by 80% compared with ORP9L (Figure 2D), but was still present in the soluble fraction similar to the wild-type protein (Figure 2E). OSBP also extracted cholesterol from liposomes to a similar extent as ORP9L (Figure 2D). Attempts to assay extraction of [³H]25-hydroxycholesterol from liposomes by OSBP and ORP9L was not successful because of increased solubility of the oxysterol and resulting high background. Finally, ORP9L- and OSBP-mediated cholesterol transfer between donor and acceptor liposomes was assayed using the lactosyl-phosphatidylethanolamine (PE)/R. communis agglutinin method (Kasper and Helmkamp, 1981; Raychaudhuri et al., 2006; Figure 2F). Both ORP9L and OSBP transferred <5% cholesterol when PI-4P was excluded from the assay or present only in acceptor liposomes. In contrast, ORP9L transferred 20% of donor liposome cholesterol when PI-4P was present in donor liposomes or both donor and acceptor liposomes. OSBP displayed significantly greater cholesterol transfer under the same conditions. Although OSBP and ORP9L bound PI-3P and PI-5P on nitrocellulose (Figure 1C), sterol transfer was not stimulated when either of these lipids was included in donor and acceptor liposomes (Figure 2G). PI-4P–dependent transfer activity required the PH domain as shown by reduced transfer activity of OSBP PH-RR/EE, a mutation that prevents PI-4P binding (Levine and Munro, 2002; Figure 2G). Thus the physical state of sterol ligands presented to OSBP does not affect binding, but ORP9L clearly shows a preference for sterol ligands presented in phospholipid bilayers. More importantly, both proteins display cholesterol transfer activity that was stimulated by PI-4P.

PH and Sterol-binding Domains Control ORP9L Partitioning between the ER and trans-Golgi/TGN

Dual targeting of ORP9L to the ER and Golgi has striking similarity to OSBP and CERT, suggesting a role in sterol trafficking between these compartments. To establish which Golgi compartment contained ORP9L, coimmunofluorescence experiments were conducted with cis (giantin), cis/medial (PI-4 kinase IIIβ), and trans-Golgi/TGN (γ-adaptin) markers (de Graaf et al., 2004; Weixel et al., 2005; Figure 3). Immunofluorescence was performed in control cells and cells treated with nocodazole to fragment the Golgi into mini-stacks to enhance visualization. Giantin was closely associated but did not overlap with endogenous ORP9L in control or nocodazole-treated CHO cells. PI-4 kinase IIIβ generates a pool of PI-4P that targets the isolated PH domains of CERT and OSBP to the trans-Golgi (Balla et al., 2005; Toth et al., 2006). However, PI-4 kinase IIIβ did not colocalize with ORP9L. ORP9L colocalized with γ-adaptin in the TGN, but only partially because there were distinct regions of nonoverlap (see magnified area in Figure 3, bottom panel). This suggests that ORP9L is restricted to distal trans-Golgi/TGN compartments.

We next identified how the lipid and VAP-binding domains of ORP9L affected partitioning between the ER and Golgi apparatus (Figure 4). To this end, vectors encoding V5-tagged wild-type ORP9L and mutants in the sterol-binding (ORP9L-ΔSB), VAP (ORP9L-FY/AA), or PH domains (ORP9L-R22E), were expressed in CHO cells under the control of the TET-repressor and induced with the doxycycline. To confirm that the PH and sterol-binding domain mutations did not affect the interaction with VAP, cells were induced with doxycycline, and interaction of ORP9L mutants with VAP was determined by GST-VAP pulldown and coimmunoprecipitation. ORP9L, ORP9L-ΔSB, and ORP9L-R22E bound VAP in GST-pulldown and immunoprecipitation assays (Supplementary Figure S1, A and B). As expected, ORP9L-FY/AA did not bind to GST-VAP, but some background binding was detected by coimmunoprecipitation. The effect of inactivating the PH, sterol-binding and FFAT domains of ORP9L on intracellular localization was determined by immunofluorescence (Figure 4). As we previously reported, inducibly overexpressed ORP9L colocalized in CHO cells with VAP in vacuolated ER structures (Figure 4A). There was also limited perinuclear staining that colocalized with giantin but significantly less than observed for endogenous ORP9L, indicating that other factors are limiting for Golgi interaction. Vacuolated ER and Golgi localization was lost when the FFAT motif was mutated (Figure 4B). ORP9L FY/AA was diffusely localized and occasionally in large punctuate structures that contained the ERGIC marker p58 (Wyles and Ridgway, 2004; Figure 3B). The PH domain mutant ORP9L R22E was strongly localized to condensed ER structures containing VAP (Figure 4C) that were reminiscent of OSBP PH domain mutants (Wyles et al., 2002) that caused ER membrane stacking due to weak protein–protein interactions between opposing bilayers (Snapp et al., 2003; Amarilio et al., 2005). Lastly, sterol-binding–defective ORP9L ΔSB was primarily localized to the Golgi apparatus, but was also dispersed to a peripheral ER compartment that did not localize with the perinuclear ER marker calnexin or the PM marker caveolin (results not shown; Figure 3D). ORP9L ΔSB interaction with VAP also resulted in mislocalization of this partner protein to the Golgi apparatus. Collectively this shows that the ORP9L FFAT and sterol-binding domains promote ORP9L retention in the ER, whereas the PH domain promotes partitioning to the trans-Golgi/TGN.

ORP9L Is Required for Golgi Organization and Protein Transport

Similarities between ORP9L and OSBP domain organization, sterol binding/transfer, and intracellular localization suggested that ORP9L could also be directly or indirectly involved in regulation of CERT and SM synthesis (Perry and Ridgway, 2006). This was tested by depleting CHO cells of ORP9L with siRNA as previously described (Lessmann et al., 2007) and monitoring CERT-dependent SM synthesis by [³H]serine incorporation in the presence and absence of 25-hydroxycholesterol (Figure 5A). As previously reported in CHO and HEK293 cells (Perry and Ridgway, 2006), silencing of OSBP by RNA interference (RNAi) minimally affected basal SM synthesis but profoundly inhibited 25-hydroxycholesterol–activated synthesis. In contrast, depletion of ORP9L caused a slight increase in both basal and 25-hydroxycholesterol–stimulated SM synthesis. SM synthesis in cells depleted of both OSBP and ORP9L was increased slightly compared with OSBP knockdown. ORP9L depletion did not affect synthesis of GlcCer or ceramide, indicating no obvious role in sphingolipid synthesis in the ER or Golgi apparatus. Immunoblots showing the siRNA-mediated knockdown that resulted in 70–90% depletion of OSBP and ORP9L (Figure 5B). Sphingomyelin synthesis was also measured in CHO cells inducibly expressing ORP9L (Figure 5C). In this case, induction of ORP9L expression with doxycycline did not affect basal SM and GlcCer synthesis but blocked stimulation by 25-hydroxycholesterol. Because OSBP and ORP9L share protein and lipid ligands, the inhibitory effect of ORP9L overexpression could be the results of competition for these factors.

While analyzing ORP9L depleted cells, we noted that immunofluorescence staining of Golgi resident proteins was frequently dispersed. For example, Figure 6 shows coimmunofluorescence localization of ORP9L with giantin or γ-adaptin in fields of CHO cells transfected with an ORP9L siRNA. Cells with relatively low ORP9L expression (indicated by an arrow) displayed conspicuous dispersion of giantin and γ-adaptin staining compared with adjacent cells that expressed more ORP9L.

Fragmentation of the giantin-positive cis-Golgi in ORP9L-depleted cells, a compartment in which ORP9L does not reside (Figure 3), suggested that structural changes could be due to defective retrograde or anterograde transport. To test whether anterograde transport was affected, export of myc-tagged, temperature-sensitive 045-VSVG from the ER to the Golgi was monitored by endoH sensitivity. Control and siORP9 transfected cells were maintained at the nonpermissive temperature of 40°C to retain 045-VSVG in the ER, followed by a shift to 33°C to initiate ER export and modification of trimmed N-linked carbohydrates in the Golgi apparatus (Figure 7A). At 40°C, 045-VSVG carbohydrates were completely endoH sensitive in siNT- and siORP9-transfected cells. After a 2-h shift to 32°C, 70% of 045-VSVG in siNT-transfected cells was transported to the Golgi where it acquired endoH resistance, compared with 50% in ORP9L-depleted cells. A comparison of the ratio of resistant to sensitive 045-VSVG (Figure 7B) indicated consistently less 045-VSVG transport to the ER in ORP9L-depleted cells; however, this result did not reach statistical significance.

Cholesterol is essential for transport through the secretory pathway (Wang et al., 2000; Ying et al., 2003), where it is progressively enriched from the ER to the plasma membrane (Ridgway, 2000). Based on in vitro cholesterol transfer activity, a primary function of ORP9L could be maintenance of cholesterol distribution within the secretory pathway. RNAi depletion of ORP9L and resulting effects on cholesterol transport could be reflected in altered regulatory responses in the ER. However, ORP9L depletion had no effect on cholesterol or cholesterol ester synthesis, indicating its activity is segregated from sterol-regulatory events in the ER (Supplemental Figure 4A and results not shown). To determine whether ORP9L regulated cholesterol content of other compartments, the cholesterol-binding polyene macrolide filipin was used to visualize cholesterol in control and ORP9L-depleted CHO cells (Figure 8). On visual inspection, it was apparent that, compared with siNT-treated cells, cells transfected with siORP9L and cultured in FCS had increased filipin fluorescence primarily in punctate structures corresponding to endosomes and lysosomes (Figure 8A). There was no apparent change in filipin fluorescence in the PM or other diffusely stained structures. Quantitation of these images showed that knockdown cells cultured in FCS or LPDS had a 50–75% increase in filipin fluorescence relative to siNT controls (Figure 8B). Measurement of cellular cholesterol mass under the same conditions revealed that although ORP9L-depleted cells cultured in FCS or LPDS had a consistent 10–15% increase in total and unesterified cholesterol, this result was not significant (Figure 8C). This indicates that ORP9L maintains cholesterol distribution in the post-Golgi, endosomal compartment rather than influencing regulatory events in the ER.

ORP9S Is a Dominant Inhibitor of Cell Growth and Protein Export from the ER

Several ORP genes encode truncated or “short” versions encoding the C-terminal sterol binding and FFAT domains (Lehto and Olkkonen, 2003). Although ORP9S protein and mRNA is expressed in a variety of tissues and cells (Wyles and Ridgway, 2004; Lessmann et al., 2007), its function relative to the full-length proteins is obscure. The absence of a PH domain in ORP9S would restrict it to the ER where it could interfere with the normal function of ORP9L or other VAP-interacting proteins. To test this, CHO cell lines were generated that expressed ORP9S, as well as VAP-binding (ORP9S-FF/AA) and sterol-binding (ORP9S-ΔSB) mutants under the control of a doxycycline-inducible promoter. GST-pulldown and coimmunoprecipitation assays confirmed that wild-type and ORP9S-ΔSB interacted with VAP, but the FF/AA mutant was devoid of VAP-binding activity (Supplementary Figure S1B). Overexpressed ORP9S was dispersed though out the cell but also colocalization with VAP at several foci (Figure 9A). Similar to ORP9L-depleted cells (Figure 6), the Golgi apparatus (visualized with giantin) was fragmented in ORP9S-overexpressing cells (+Dox) compared with uninduced cells (−Dox, Figure 9B). Dispersion of the Golgi apparatus and colocalization with VAP was not evident in cells expressing VAP (ORP9S-FF/AA) and sterol-binding domain (ORP9S-ΔSB) mutants (Figure 9, C and D).

During initial characterization of overexpressing cells, it became apparent that ORP9S caused growth inhibition. Quantification of cell number over a 72-h induction period revealed that ORP9S expression caused profound inhibition of cell proliferation relative to uninduced controls (Supplemental Figure S2A). This was not due to activation of an irreversible cell death pathway because removal of doxycycline after induction of ORP9S for 36 h led to recovery of cell growth (Supplemental Figure S2B). As well, cells expressing ORP9S did not display any indications of necrosis or apoptosis (result not shown). VAP and sterol binding were required for this phenotype because expression of ORP9S-FY/AA or ORP9S-ΔSB did not affect cell growth (Supplemental Figure S2, C and D). Overexpression experiments with wild-type and mutant ORP9L failed to effect cell growth or viability (Supplementary Figure S3). Interestingly, overexpression of ORP9L-R22E, which had defective PI-4P binding and Golgi localization (Figures 1C and 4D), did not affect cell proliferation, indicating that deletion of the entire interval between amino acids 1-165 is required for the growth arrest phenotype.

Fragmentation of the Golgi apparatus and cessation of cell growth by ORP9S expression suggested that ORP9S was a dominant-negative inhibitor of ORP9L functions related to cholesterol distribution and function in the ER–Golgi pathway. We initially determined whether enforced expression of ORP9S influenced cholesterol homeostasis and cellular distribution. Overexpression of ORP9S, but not ORP9L, caused a two-fold decrease in cholesterol synthesis measured by [³H]acetate incorporation (Supplemental Figure S4A). This was accompanied by a significant reduction in filipin fluorescence in the endosomes/lysosomes of ORP9S cells cultured in FCS or LPDS, but no significant change in cholesterol mass (Supplemental Figure S4, B and C). Enforced expression of ORP9L did not affect either of these parameters (Supplemental Figure S4, D and E).

The effect of ORP9S expression on ER–Golgi protein transport was determined by assessing 045-VSVG transport and sensitivity to endoH. For these experiments, CHO cells were cultured in the absence or presence of doxycycline to induce ORP9S, transfected with 045-VSVG at the nonpermissive temperature, and 045-VSVG transport to the Golgi was assessed by endoH sensitivity after shifting to the permissive temperature (Figure 10). At the permissive temperature, 045-VSVG processing to an endoH-resistant form in cells expressing ORP9L was evident after 120 min (Figure 10A), although there appeared to be a slight delay compared with uninduced cells. In contrast, cells expressing ORP9S had a complete block in processing at all time points compared with uninduced cells. In a related experiment, CHO cells expressing ORP9L, ORP9S, and ORP9S mutants were transiently transfected with 045-VSVG and treated without and with doxycycline at 33°C for 16 h to assess steady-state export and carbohydrate processing in the Golgi apparatus (Figure 10B). Under these conditions, 045-VSVG was processed to an endoH-resistant form in ORP9L expressing cells, although there appeared to be slightly less relative to uninduced cells. In contrast, expression of ORP9S prevented transport of 045-VSVG to the Golgi as indicated by a complete lack of the endoH-resistant form. Consistent with the lack of effect of ORP9S FY/AA and ΔSB on cell growth (Supplemental Figure S3), processing of 045-VSVG carbohydrates was similar to matched uninduced cells (Figure 10B). Results from Figure 10B were quantified by densitometry and expressed relative to matched uninduced controls (Figure 10C). It is evident there was a significant block in steady-state processing in cells expressing ORP9S relative to those expressing ORP9L or the ORP9S mutants. Thus ORP9S-mediated inhibition of ER–Golgi transport is correlated with cell growth arrest and requires interaction with VAP and sterol ligand(s).