Glycine 420 Near the C-terminal Transmembrane Domain of SR-BI Is Critical for Proper Delivery and Metabolism of High Density Lipoprotein Cholesteryl Ester*

Scavenger receptor BI, SR-BI, is a physiologically relevant receptor for high density lipoprotein (HDL) that mediates the uptake of cholesteryl esters and delivers them to a metabolically active membrane pool where they are subsequently hydrolyzed. A previously charac-terized SR-BI mutant, A-VI, with an epitope tag inserted into the extracellular domain near the C-terminal transmembrane segment, revealed a separation-of-function between SR-BI-mediated HDL cholesteryl ester uptake and cholesterol efflux to HDL, on one hand, and cholesterol release to small unilamellar phospholipid vesicle acceptors and an increased cholesterol oxidase-sensi-tive pool of membrane free cholesterol on the other. To further elucidate amino acid residues responsible for this separation-of-function phenotype, we engineered alanine substitutions and point mutations in and around the site of epitope tag insertion, and tested these for various cholesterol transport functions. We found that changing amino acid 420 from glycine to histidine had a profound effect on SR-BI function. Despite the ability to mediate selective HDL cholesteryl ester uptake,

in reverse cholesterol transport (RCT). RCT is the process whereby cholesterol is transported from peripheral tissues via plasma HDL to the liver for bile acid synthesis and secretion or to endocrine tissues for steroid production (5). Contrary to the classic low density lipoprotein receptor endocytic pathway, in which the entire lipoprotein is internalized in clathrin-coated pits and degraded (6), SR-BI mediates the selective uptake of HDL cholesteryl ester (CE), the process whereby HDL core CE is taken into the cell without degradation of the whole particle and its apolipoproteins. The HDL CE-selective uptake pathway and SR-BI are the major route for delivery of HDL CE to the liver and steroidogenic tissues in rodents (7)(8)(9)(10)(11)(12) and appear to be a major route for cholesterol delivery in human steroidogenic cells as well (13,14). SR-BI-mediated HDL CE uptake is a two-step process: the first step is lipoprotein binding to the extracellular domain of SR-BI, and the second step is the actual lipid transfer from the HDL particle to the plasma membrane (3,4). In addition to selective HDL CE uptake, SR-BI stimulates the bi-directional flux of free cholesterol (FC) between cultured cells and lipoproteins (15)(16)(17), an activity that may be responsible for net cholesterol efflux from peripheral cells as well as the rapid hepatic clearance of free cholesterol from plasma HDL and its resultant secretion into bile (18,19). Therefore, SR-BI participates in both ends of the RCT pathway: efflux of FC to HDL from peripheral cells and the delivery of HDL CE and FC to the liver for secretion into bile.
Evidence is starting to emerge as to what happens to the HDL CE after transfer by SR-BI to the plasma membrane. It has been known for some time that low density lipoprotein CE delivered by the low density lipoprotein receptor are hydrolyzed in the lysosomal pathway by an acidic CE hydrolase (6,20), whereas HDL CE delivered by SR-BI are hydrolyzed extralysosomally (21) by a neutral CE hydrolase (22,23). In fact, SR-BI has been shown to deliver HDL CE into a metabolically active membrane pool where they are efficiently hydrolyzed by cell type-specific neutral CE hydrolases (24). In conjunction with these activities, SR-BI increases cellular cholesterol mass and alters cholesterol distribution in plasma membrane domains as judged by the enhanced sensitivity of membrane cholesterol to extracellular cholesterol oxidase (25,26). Together these data support the idea that SR-BI delivers HDL CE and FC into a metabolically active membrane pool where they are efficiently metabolized.
Recently, we described an SR-BI mutant with a separation-of-function between SR-BI-mediated HDL binding, HDL CE uptake, and HDL FC efflux on the one hand and FC efflux to small unilamellar vesicle acceptors and an increased cholesterol oxidase-sensitive pool of membrane FC on the other (27). This mutant receptor, A-VI, contains an epitope tag from the adenovirus E4/5 protein inserted in the extracellular domain of SR-BI near the C-terminal transmembrane domain. To further elucidate amino acid residues that might be responsible for this separation-of-function phenotype, we engineered alanine substitutions and point mutations in and around the epitope tag insertion site and tested the mutants for various cholesterol transport functions. From this study we found that changing a single glycine residue, in the extracellular domain near the C-terminal transmembrane domain, to a histidine has a profound effect on the ability of SR-BI to deliver HDL CE in a way that allows its efficient metabolism.
All plasmids were prepared using endotoxin-free Qiagen Maxi-prep kits and sequenced throughout the SR-BI coding region to confirm the correct mutation or epitope tag addition and to ensure that no undesired point mutations had been generated during the amplification process. DNA sequencing was performed by the automated sequencing facility at the State University of New York at Stony Brook. Reactions were prepared using a dye termination cycle sequencing kit and analyzed on an Applied Biosystems Model 3100 DNA Sequencer with an Excel upgrade as recommended by the manufacturer (PE Applied Biosystems).
Transient Transfection of COS-7 Cells-COS-7 cells were maintained and transfected as previously described (28). The cells were assayed 48 h post-transfection unless otherwise indicated. Cell lysates were made as previously described (30,31), and protein concentrations were determined by the Lowry method (32). Protein lysates were electrophoresed, transferred onto nitrocellulose membranes, and detected using a polyclonal anti-SR-BI C-terminal antibody (Novus Biologicals, Inc.) (1:5,000), a horseradish peroxidase-conjugated anti-rabbit secondary antibody (Jackson ImmunoResearch Laboratories) (1:10,000), and SuperSignal West Pico reagent (Pierce H]COE-HDL particles were added at a concentration of 10 g of protein/ml (unless otherwise indicated) in serum-free DMEM, 0.5% BSA. After incubation for 1.5 h at 37°C, the medium was removed and the cells were washed three times with PBS, 0.1% BSA (pH 7.4) and one time with PBS (pH 7.4). The cells were lysed with 1.1 ml of 0.1 N NaOH, and the lysate was processed to determine trichloroacetic acid-soluble and insoluble 125 I radioactivity and organic solvent-extractable 3 H radioactivity. The values for cell-associated HDL apolipoprotein, total cell-associated HDL COE, and the selective uptake of HDL COE, were obtained as previously described (28).
Cholesterol Oxidase-sensitive FC, FC Efflux, and Cholesterol Mass Assays-Assays were performed as previously described (27). Statistical comparisons were made by unpaired Student's two-tailed t test (GraphPad Prism version 4.0, GraphPad Software).
Cholesteryl Ester Hydrolysis Assay-Transiently transfected COS-7 cells (35-mm wells) were preincubated for 2 h at 37°C in the presence of an acyl CoA:cholesterol acyltransferase (ACAT) inhibitor, CP113,818 (gift from Pfizer), at a final concentration of 2 g/ml in DMEM containing 0.5% BSA. [ 3 H]CE-HDL particles were added at a concentration of 10 g of protein/ml in DMEM containing 0.5% BSA and 2 g/ml ACAT inhibitor. After incubation at 37°C for 2 h (unless otherwise indicated), the medium was removed and the monolayers were washed three times with PBS and allowed to dry. The lipids were extracted from the cell monolayers with isopropanol. One-tenth of the extracted lipids were counted for total cell-associated radioactivity. The remainder of the lipids was dried under N 2 and resuspended in 40 l of chloroform/ methanol (1:1, v/v) with FC and CE as unlabeled standards. The samples were run on activated silica gel ITLC plates (VWR or Nuncleon) in a TLC tank with 95:5:4 (petroleum ether:diethyl ether:acetic acid, v/v) as the solvent system. The plates were air-dried, and spots corresponding to CE and FC were visualized with iodine, cut from the strip, and analyzed by liquid scintillation counting to determine the fraction of Co-immunoprecipitation Assay-Cell lysate (ϳ100 g of protein), 2.5 mg/ml BSA, 100 l of Nonidet P-40 cell lysis buffer (30, 31) containing 1 g/ml pepstatin, 0.2 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 10 g/ml aprotinin, and 2 g of myc-or FLAG-antibody were rotated at room temperature for 90 min. Protein-G-Sepharose (50 l of a 1:1 slurry in cell lysis buffer) was added to each microcentrifuge tube and rotated at room temperature for an additional hour. Samples were centrifuged and washed twice in cell lysis buffer (with protease inhibitors) and once in cell lysis buffer (without protease inhibitors). Sample treatment buffer (60 l) was added to each lysate pellet. Samples immunoprecipitated with the myc antibody were boiled at 100°C for 5 min, and samples immunoprecipitated with the FLAG antibody were heated at 42°C for 15 min. Samples (50 l) were separated by 8% SDS-PAGE and analyzed by immunoblotting using antibodies against the myc or FLAG epitopes (28). The following antibodies were used: polyclonal anti-SR-BI C-terminal and anti-SR-BI extracellular domain antibodies (Novus Biologicals, Inc., 1:5,000 for immunoblot); monoclonal anti-FLAG IgG (M2, Stratagene, 1:2,000 for immunoblot); monoclonal anti-myc IgG (Invitrogen, for immunoprecipitations); polyclonal c-Myc (A-14:sc-789, Santa Cruz Biotechnology, Santa Cruz, CA, 1:1,000 for immunoblot); and peroxidase-conjugated goat anti-mouse or antirabbit secondary IgG (Jackson ImmunoResearch Laboratories, 1:10,000 for immunoblot). Protein-G-Sepharose was from Sigma Aldrich.
Fluorescence Staining-Transiently transfected COS-7 cells were plated onto coverslips 24 h after transfection and processed 24 h after plating. For fluorescent HDL experiments, HDL was labeled with an Alexa-568 protein labeling kit using the protocol provided by Molecular Probes, Inc. The transiently transfected cells were incubated with 20 g/ml Alexa-labeled HDL for 30 min on ice, then washed with PBS and fixed for 1 h with 4% (w/v) paraformaldehyde in 77 mM PIPES, pH 7.5. Cells were washed in PBS and treated with the Slowfade Light Antifade kit from Molecular Probes, Inc. For immunofluorescence experiments, cells were fixed with 4% paraformaldehyde (w/v) for 1 h, permeabilized with 0.3% Triton X-100 for 15 min, and blocked with 3% BSA, 10 mM glycine in PBS for 1 h. Cells were then incubated with primary antibody directed against the C-terminal cytoplasmic domain of SR-BI (Novus Biologicals, Inc., 1:500) for 1 h at room temperature. Cells were washed with PBS and treated with secondary antibody, Alexa 488-conjugated goat anti-rabbit IgG (Molecular Probes, Inc., 1:1000), for 30 min at room temperature. Cells were washed and mounted as above and examined using a Nikon Eclipse E800 microscope, and confocal images were collected with a Bio-Rad Radiance 2000 system.

Alanine Substitutions Reveal a Span of Five Amino Acids
That, When Mutated, Allows SR-BI to Mediate Selective HDL COE Uptake, but Inhibits Its Ability to Increase the Size of the Oxidase-sensitive Pool of Membrane Cholesterol-We recently reported an SR-BI mutant, A-VI, that contains an epitope tag, from the adenovirus E4/5 protein, inserted in the extracellular region of SR-BI near its C-terminal transmembrane domain (27). To further define the residues responsible for the A-VI separation-of-function phenotype, we substituted blocks of five amino acids, near the epitope tag insertion site (indicated in Fig. 1A by an asterisk), with alanines ( Fig. 1A). When expressed in COS-7 cells, all of the alanine substitution mutants exhibited at or near wild type SR-BI levels of expression as observed by immunoblot (Fig. 1D). Therefore, we proceeded to assess the ability of these alanine substitution mutants to bind HDL and mediate lipid transport.
Comparison of cells expressing mutant receptors to cells expressing wild type SR-BI revealed that the alanine substitution mutants displayed four different phenotypes (Fig. 1). K425A-T429A expressing COS-7 cells exhibited wild type lev-els of HDL binding (Fig. 1B), selective HDL COE uptake (Fig.  1C) and cholesterol oxidase-sensitive FC (Fig. 1E). W415A-S419A-expressing cells, on the other hand, did not bind HDL (Fig. 1B), did not mediate selective HDL COE uptake (Fig. 1C), and did not show an increase in the size of the oxidase-sensitive pool of membrane FC (Fig. 1E). This was expected, because this mutation overlaps the glutamine residue (Gln-418) that was previously reported to be involved in HDL binding (36). F430A-L434A-and V435A-Q439A-expressing cells exhibited reduced HDL binding (Fig. 1B), reduced selective HDL COE uptake (Fig. 1C), and a reduced ability to produce an increase in the oxidase-sensitive pool of membrane FC (Fig. 1E) when compared with wild type SR-BI-expressing cells. However, we did not determine whether the decrease in HDL binding for these two mutants was due to decreased cell surface receptor expression or due to a decreased ability of the mutant to bind HDL. Despite a reduced level of cell surface HDL binding, G420A-G424A-expressing COS-7 cells exhibited wild type levels of selective HDL COE uptake (Fig. 1C). Interestingly, these cells also exhibited a reduced oxidase-sensitive pool of membrane FC compared with wild type SR-BI-expressing cells (Fig. 1E).

Point Mutations in SR-BI Reveal a Glycine to Histidine Mutation That Reduces the Ability of the Receptor to Mediate
Selective HDL COE Uptake and Increase the Size of the Oxidase-sensitive Pool of Membrane FC-The ability of the G420A-G424A mutant receptor to mediate efficient selective uptake of HDL COE, but to produce only a minimal increase in the size of the oxidase-sensitive pool of membrane FC, is similar to what we reported previously for the A-VI epitope tag insertion mutant (27). Therefore, we mutated the G420A-G424A residues individually, to determine which were most responsible for the separation-of-function phenotype. Analysis of the protein sequences from seven different mammals (mouse, rat, human, hamster, rabbit, pig, and cow) showed the residues in the region of SR-BI near the C-terminal transmembrane segment (amino acids 412-462) that are conserved in evolution (Fig. 2). Due to their high degree of cross-species homology, residues Gly-420, Met-422, and Gly-424 were chosen for further mutation (see residues highlighted in boldface, Fig.  2). We modified the SR-BI cDNA to encode alanines (G420A, M422A, and G424A) or histidines (G420H and G424H) at these sites (Fig. 3A). In addition, double alanine and double histidine mutants were generated for Gly-420 and Gly-424 (G420A/ G424A and G420H/G424H). We then expressed the mutant receptors in COS-7 cells and assessed the ability of these point mutants to mediate lipid transport. COS-7 cells transiently expressing each of the point mutant receptors exhibited at or near wild type levels of HDL binding (Fig. 3B). In addition, all of the point mutants were expressed at or near wild type levels as observed by immunoblot (Fig. 3D). Although cells expressing G420A, M422A, G424A, G420A/ G424A, and G424H exhibited wild type levels of selective HDL COE uptake, cells expressing G420H and the double-histidine mutant G420H/G424H exhibited levels of HDL COE-selective uptake that were 50 -60% wild type levels (Fig. 3C). Furthermore, G420H-and G420H/G424H-expressing cells showed no increase in the size of the oxidase-sensitive pool of membrane FC when compared with wild type SR-BI cells (Fig. 3E).
Histidine Replacement of Glycine 420, G420H, Greatly Reduces the Ability of SR-BI to Efflux Cholesterol to HDL-In addition to mediating HDL CE-selective uptake, SR-BI accelerates the release of FC from cells (15,16,25). To determine whether the G420H mutation had an effect on the ability of the receptor to mediate cholesterol efflux, COS-7 cells expressing the point mutant receptors were assayed for efflux of cholesterol to HDL (200 g/ml) (Fig. 4A). The cells expressing G420A, M422A, G424A, G420A/G424A, and G424H were able to efflux cholesterol to HDL in 4 h with the same efficiency as wild type SR-BI-expressing cells (Fig. 4A). G420H-and G420H/G424Hexpressing cells, however, showed a reduced ability to release cholesterol to HDL (Fig. 4A). In addition, we tested the ability of G420H to efflux cholesterol to increasing concentrations of HDL. The G420H mutant-expressing cells showed a greatly reduced ability to release cholesterol, even to lower HDL ac- ceptor concentrations, when compared with wild type SR-BIexpressing cells (Fig. 4B).

G420H Reduces the Ability of SR-BI to Deliver HDL CE, and the CE That It Does Deliver Is Not Efficiently Hydrolyzed-In
earlier studies it was shown that SR-BI delivers HDL CE to a metabolically active membrane compartment, where the CE is efficiently hydrolyzed by a neutral CE hydrolase (24). To determine whether the point mutations affected the ability of SR-BI to direct HDL CE to a neutral CE hydrolytic pathway, we expressed the mutant receptors in COS-7 cells and assayed the cells for the amount of HDL CE incorporated and the fraction of this CE that was hydrolyzed. Although cells expressing G420A, M422A, G424A, G420A/G424A, and G424H exhibited wild type levels of HDL CE uptake, G420H and the doublehistidine mutant G420H/G424H-expressing cells exhibited levels of HDL CE uptake that were about 25-30% of wild type levels (Fig. 5A). In addition, the HDL CE taken into the cells by cells expressing G420A, M422A, G424A, G420A/G424A, and G424H was hydrolyzed as efficiently as that delivered by wild type SR-BI (Fig. 5B). On the contrary, much of the HDL CE delivered to the cells by G420H and G420H/G424H remained esterified (Fig. 5B). To determine if the decreased hydrolysis in the G420H-and G420H/G424H-expressing cells was due to less uptake of HDL CE or to a change in the way the HDL CE is delivered to the cell, we compared the HDL CE taken into the cell to the amount of FC generated for cells expressing each of the mutants discussed so far. We found a very good correlation (r 2 of 0.9991) between the amount of HDL CE uptake and the amount of FC generated by CE hydrolysis (Fig. 5C). In fact, all of the mutants, including the G420A-G424A mutant, fall on the line (indicating ϳ77% hydrolysis), except for the two mutants with the G420H mutation. These two mutants, G420H and G420H/G424H, clearly fall below the line (indicating ϳ38% hydrolysis). This indicates that these two mutants generated less FC than would be expected for the amount of CE that they delivered to the cells.
Relative to SR-BI-expressing cells, G420H-expressing cells took up considerably less HDL CE (30% compared with SR-BI) than they did HDL COE (60% compared with SR-BI). To test whether the reduced uptake of HDL CE in the G420H-expressing cells was due to a direct inability of this mutant receptor to mediate uptake or due to a saturation of the membrane with unhydrolyzed HDL CE, we performed a time course of HDL CE uptake and hydrolysis. At the earliest time points, 15 and 30 min, the G420H mutant-expressing cells incorporated 85 and 58%, respectively, the amount of HDL CE incorporated by wild type SR-BI-expressing cells (Fig. 6A). After 1 h, however, G420H-expressing COS-7 cells showed about 32% of the HDL CE uptake seen with wild type SR-BI-expressing cells (Fig. 6A). In addition, when the amount of HDL CE accumulation was plotted over time, it became clear that HDL CE accumulated to the same extent in both the SR-BI-and G420H-expressing cells (Fig. 6B). The amount of FC, on the other hand, increased with time in the SR-BI-expressing cells due to hydrolysis of the CE delivered to the cell from the HDL particles by SR-BI. In contrast, the amount of FC in the G420H-expressing cells remained the same as those observed with cells transfected with vector alone (Fig. 6C). Clearly, the amount of HDL CE uptake in the G420H-expressing cells slowed down greatly with time, presumably due to the failure to hydrolyze the HDL CE.
G420H Mutation Reduces the Ability of SR-BI to Accumulate Cellular CE and FC-A striking characteristic of SR-BI is that it alters cholesterol content and distribution in cells (26). To test whether or not this mutation had an effect on the mass distribution of cholesterol, we expressed G420H, G420H/ G424H, and G420A-G424A mutants in COS-7 cells and measured total and free cholesterol by gas chromatography. As previously observed, when compared with vector-transfected cells, SR-BI-expressing cells had an increased amount of total cholesterol, which included a statistically significant increase in free and esterified cholesterol (Table I). Like SR-BI, G420A-G424A-expressing cells also exhibited an increase in the levels of total, free, and esterified cholesterol, which was consistent with its ability to mediate wild type levels of selective HDL CE uptake (Fig. 1C) and to deliver the HDL CE for hydrolysis (Fig.  5C). G420H-and G420H/G424H-expressing cells, on the other hand, had a greatly reduced pool of esterified cholesterol compared with wild type SR-BI-expressing cells. This observation is consistent with the ability of these mutant receptors to deliver some HDL CE to the cell, but not to a membrane pool of CE that would allow its efficient hydrolysis. Notably, the reduced pool of CE seen in the G420H-and G420H/G424Hexpressing cells was still statistically higher than that observed with cells transfected with vector alone, which is also consistent with the observation that these receptors do allow some uptake of HDL CE to occur. Although there is a reduction in the cellular accumulation of CE in G420H-and G420H/ G424H-expressing cells, the values for cellular FC mass are intermediate to cells expressing wild type SR-BI. Although we have not tested this directly, G420H-and G420H/G424H-ex- pressing cells probably retained the ability to mediate influx of HDL FC, which could account for the increase in cellular FC and not CE. We repeated this experiment in the presence of acyl CoA:cholesterol acyltransferase (ACAT) inhibitor and found that in cells that deliver HDL CE for hydrolysis, such as SR-BI-and G420A-G424A-expressing cells, there is a decrease in the amount of CE that accumulated in the cells. This was presumably because the cells that were not treated with ACAT inhibitor were able to hydrolyze and re-esterify a fraction of the HDL CE that was incorporated, and the cells treated with the ACAT inhibitor could not re-esterify the FC generated by hydrolysis. The G420H-and G420H/G424H-expressing cells, on the other hand, exhibit the same CE accumulation in the presence or absence of ACAT inhibitor, which indicates that the HDL CE came directly from the HDL particle and was not due to hydrolysis and re-esterification. This further supports the observation that, unlike wild type SR-BI, the G420H mutation does not permit the receptor to deliver the HDL CE for efficient CE hydrolysis.
G420H Mutation Has No Effect on the Ability of the Receptor to Homo-oligomerize-Interestingly, the region of SR-BI near and including part of the C-terminal transmembrane domain contains a leucine zipper motif (37). In addition, the critical glycine residue, Gly-420, is in register with the six highly conserved leucines that form this motif (see arrowheads, Fig.  2). Because leucine zipper regions are protein dimerization motifs and recent evidence shows that SR-BI is able to form homo-multimeric complexes (38), 2 we decided to test whether the G420A-G424A or G420H mutations had an effect on homo-

FIG. 5. Uptake and hydrolysis of HDL CE by cells expressing wild type SR-BI and mutant SR-BI receptors with point mutations.
COS-7 cells transiently expressing SR-BI or point mutant receptors were incubated at 37°C for 2 h with 10 g/ml [ 3 H]CE-labeled HDL in the presence of an ACAT inhibitor at a final concentration of 2 g/ml. After incubation, the cells were processed to determine the total cell associated dpm/mg cell protein (A) and the percentage of the total number of counts that remained as [ 3 H]CE (B). COS-7 cells transiently expressing wild type SR-BI, alanine substitution mutants or point mutants were incubated at 37°C for 2 h with 10 g/ml [ 3 H]CE-labeled HDL in the presence of an ACAT inhibitor at a final concentration of 2 g/ml. After incubation, the cells were processed to determine the total cell-associated dpm/mg cell protein, and the amount of free cholesterol generated (FC dpm/mg cell protein) was plotted against the amount of HDL CE uptake (dpm) (C). Values represent the mean Ϯ S.D. of three replicates and are representative of two separate experiments. oligomerization. To demonstrate oligomer formation of wild type SR-BI, COS-7 cells were transfected with either of two SR-BI cDNA constructs containing C-terminal additions of either the c-myc (myc-SR-BI) or FLAG (FLAG-SR-BI) epitope tags (29). Cell lysates from COS-7 cells expressing myc-SR-BI or FLAG-SR-BI were immunoprecipitated with anti-myc or anti-FLAG antibodies, respectively, and immunoblots, probed with the appropriate antibodies, showed high levels of expression of both epitope-tagged receptors (Fig. 7, top panels, lanes 1  and 4). Cells transfected with myc-SR-BI and immunoprecipitated with anti-FLAG antibody or cells transfected with FLAG-SR-BI and immunoprecipitated with anti-myc antibody showed no detectable protein on the immunoblot, indicating that there was no nonspecific immunoprecipitation (lanes 2 and 3, Fig. 7). When both myc-SR-BI and FLAG-SR-BI were co-transfected, both epitope-tagged receptors could be co-immunoprecipitated by their opposing antibody, indicating that they were complexed with each other (lanes 5 and 6, Fig. 7). As a control for nonspecific complex formation, or aggregation, lysates of either myc-SR-BI-or FLAG-SR-BI-transfected cells were mixed and then immunoprecipitated. In this case, neither of the epitopetagged receptors could be immunoprecipitated by their opposing antibody (lanes 7 and 8, Fig. 7), indicating that there was no co-immunoprecipitation of the epitope-tagged receptors if they were not expressed at the same time in the COS-7 cells. These results support the previous conclusion that wild type SR-BI is able to homo-oligomerize. In a similar fashion, the same series of co-transfections with myc-tagged and FLAG-tagged G420A-G424A (Fig. 7, middle panels) or myc-tagged and FLAG-tagged G420H (Fig. 7, bottom panels) demonstrated that these receptors also homo-oligomerize.
At the Light Microscope Level, G420H Shows No Change in Cellular Localization When Compared with Wild Type SR-BI-Fluorescence microscopy was used to confirm cell surface expression of the G420H mutant receptor as well as to determine if the changes in G420H function were due to differences in its cellular localization pattern when compared with wild type SR-BI. COS-7 cells expressing G420H or SR-BI showed a remarkably similar pattern of cell surface fluorescent HDL binding in patches and in membrane extensions as observed previously (39) and showed a similar cellular distribution as judged by antibody staining (data not shown).

DISCUSSION
The key finding in this study is the identification of glycine 420 as a critical residue for the function of SR-BI. Substitution of histidine for glycine in G420H did not disrupt HDL binding and only modestly reduced the uptake of cholesteryl ether from HDL. In contrast, the uptake of HDL cholesteryl ester was much more compromised. The uptake of HDL cholesteryl ester via G420H was only modestly reduced at early times but progressively decreased with incubation time. This decrease was not due to a change in HDL CE accumulation in the cell as compared with SR-BI but to the failure to hydrolyze the CE (see Fig. 6B). Thus, in the absence of SR-BI-directed CE hydrolysis, membrane CE content appeared to saturate. This is the expected result, because the capacity of the plasma membrane to accommodate CE is limited to 2-3 mol% with respect to membrane phospholipid. Thus, in the absence of ongoing CE hydrolysis, the rate of net CE transfer from HDL declines. The G420H receptor provides support for the importance of the coupling between selective CE uptake and an efficient CE hydrolysis process that is necessary for selective uptake to continue. Additionally, results with the G420H receptor indicate that SR-BI is directly responsible for the efficient delivery of HDL CE to the CE hydrolysis process. The mechanism of this delivery is unclear at present. It could reflect the ability of SR-BI to recruit a neutral CE hydrolase to the plasma membrane, although the ability of SR-BI to facilitate CE hydrolysis by apparently different CE hydrolases in diverse cell types (24) argues against a direct protein-protein interaction in such a recruitment process. Alternatively, the failure of the G420H receptor to facilitate HDL CE hydrolysis might reflect the ability of SR-BI to organize membrane lipids in a manner that permits access of a cytoplasmic CE hydrolase to the CE within the membrane bilayer.
In addition to the loss of SR-BI-mediated HDL CE hydrolysis, the G420H receptor shows a complete loss of the cholesterol oxidase-sensitive pool of membrane FC, suggesting an inability of G420H to alter membrane lipid organization. One possibility is that the inability of G420H to facilitate HDL CE hydrolysis is responsible for the loss of the cholesterol oxidase-sensitive FC pool, because insufficient FC is generated within the membrane bilayer. The previously described A-VI mutant, that has a 14-residue epitope tag inserted after amino acid 424, also exhibits no cholesterol oxidase-sensitive pool of FC, although the A-VI receptor does show facilitated HDL CE hydrolysis (data not shown). In the case of A-VI, however, less of the cell surface receptor binds HDL and the amount of HDL CE taken into the cell and hydrolyzed to FC is very low compared with wild type SR-BI (27). The inefficiency of HDL CE delivery by A-VI may account for the failure to generate a cholesterol oxidase-sensitive pool of membrane FC. Interestingly, the A-VI receptor has near wild type levels of FC efflux to low levels of HDL (27), whereas the G420H receptor is more severely disabled in FC efflux to HDL, even at low HDL levels (Fig. 4). These findings as well as the alanine scanning mutations described in Fig. 1 illustrate that the SR-BI sequence from Gly-420 to the C-terminal transmembrane domain is crucial for SR-BI functions other than HDL binding. The importance for cellular cholesterol homeostasis of disruptions in SR-BI in this region is evident in the altered FC and CE mass in cells expressing the G420H receptor.
As judged by fluorescence microscopy (data not shown) and the oligomerization assay (Fig. 7), disruption of SR-BI function in the G420H receptor is not due to an altered cell surface distribution of SR-BI or to the inability of the receptor to form homo-oligomers. This result was surprising considering the mutated glycine is in register with seven leucines that form a heptad repeat reminiscent of a leucine zipper motif (37,40). The context of the Gly-420 residue (GAMGG) also closely resembles a GXXXG motif, which is another motif responsible for dimerization of membrane channel/transport proteins (41)(42)(43)(44)(45). Therefore, we initially hypothesized that the alpha helical region, which includes the C-terminal transmembrane segment and the extracellular domain to residue 413 (see Fig. 2), might assist the receptor in homo-oligomerization as has been found for many other leucine-zipper containing proteins. Although this region, and Gly-420, may participate in such interactions, it does not appear necessary for homo-oligomerization. Intriguingly, the extracellular part of this predicted alpha helix, which lies just outside of the C-terminal transmembrane segment of SR-BI, is amphipathic. As would be expected, the side of the alpha helix that contains the leucine residues, as well as glycine 420, is hydrophobic, whereas the opposite side of the alpha helix has polar and charged residues. One possibility is that one side of the amphipathic helix interacts with apoA-I on HDL, and the opposite side interacts with the plasma membrane to draw the HDL close enough to properly deliver the HDL CE. Previous studies showed that the class A amphipathic alpha helix of apoA-I binds directly to SR-BI (46,47). The apoA-I helix could interact with the polar face of the predicted SR-BI amphipathic helix in the Leu-413 through Gln-439 region. Histidine at residue 420 may disrupt the orientation of this alpha helix due to its bulk or a potential charge on the hydrophobic face. In contrast, glycine or alanine at this position is well tolerated. Further mutational analysis should be informative as to how residue 420 of SR-BI influences the interaction with HDL and the formation of the hydrophobic channel that transports the HDL CE to its site of metabolism.