Dissociation of the high density lipoprotein and low density lipoprotein binding activities of murine scavenger receptor class B type I (mSR-BI) using retrovirus library-based activity dissection.

The murine class B, type I scavenger receptor (mSR-BI) is a receptor for both high density lipoprotein (HDL) and low density lipoprotein (LDL) and mediates selective, rather than endocytic, uptake of lipoprotein lipid. We have developed a "retrovirus library-based activity dissection" method to generate mSR-BI mutants in which some, but not all, of the activities of this multifunctional protein have been disrupted. This method employs three techniques: 1) efficient in vitro cDNA mutagenesis (here error-prone PCR was used), 2) efficient retroviral delivery and high expression of single mutant cDNAs into individual cells, and 3) isolation of infected cells expressing the desired mutant phenotype using high sensitivity positive/negative screening by two-color fluorescence-activated cell sorting. A set of mutants, all having arginine substitutions at two common sites (positions 402 or 401 and position 418), were isolated and characterized. Mutation at either site alone did not generate as strong a mutant phenotype (loss of DiI uptake from DiI-HDL) as did the double mutations. "Activity-dissected" double mutants were as effective as wild-type mSR-BI in functioning as LDL receptors, mediating high affinity LDL binding and uptake of metabolically active cholesterol from LDL, but they lost most of their corresponding HDL receptor activity. Thus, these mutants provide support for the proposal that the interaction of SR-BI with HDL differs from that with LDL. Examination of the in vivo function of such mutants may provide insights into the differential roles of the LDL and HDL receptor activities of SR-BI in normal lipoprotein metabolism and in SR-BI's ability to protect against atherosclerosis.

Scavenger receptors (1,2) are multifunctional proteins defined by their ability to bind chemically modified lipoproteins (3). Often scavenger receptors can mediate the binding of a wide array of other types macromolecules or macromolecular complexes (3)(4)(5). For example, scavenger receptor class B, type I, SR-BI 1 (1,2), which was the first molecularly well defined native HDL receptor identified (6), can also bind unmodified LDL, unmodified VLDL, anionic phospholipids, as well as a number of modified proteins including acetylated LDL, oxidized LDL, and maleylated-BSA (1,7,8). In transfected cultured cells expressing high levels of murine (m)SR-BI, HDL is a very effective competitor for mSR-BI-mediated 125 I-HDL cell association (6). In contrast, LDL, which binds more tightly to mSR-BI than HDL (1,6), only poorly competes for HDL association (Ref. 6, and see below). These findings suggested distinct modes of binding, and perhaps distinct binding sites, for these lipoproteins on mSR-BI.
Unlike the classic LDL receptor, which mediates endocytosis of the intact LDL particle via coated pits and vesicles and its subsequent hydrolysis in lysosomes (9), SR-BI mediates the selective uptake of the cholesteryl esters from HDL (6). Selective lipid uptake is fundamentally different from LDL receptormediated endocytosis. It involves efficient transfer of the lipid, but not the protein (apolipoprotein), components from HDL to cells, and it does not involve the internalization and subsequent degradation of the intact lipoprotein particle (10,11,reviewed in Ref. 2). Recent studies suggest that SR-BI-mediated selective lipid uptake is a two-step process, in which high affinity lipoprotein binding is followed by receptor-mediated transfer of lipid from the lipoprotein particle to the cell membrane (12,13). After lipid transfer, the lipid-depleted lipoprotein particle is released from the cells and re-enters the extracellular space. SR-BI can mediate selective uptake of lipid from LDL as well as HDL (14 -16). The detailed molecular mechanism(s) underlying selective uptake have not yet been elucidated.
SR-BI is highly expressed in liver and steroidogenic tissues and plays a central role in controlling plasma HDL levels and cholesterol stores in steroidogenic tissues (2,6,17). Liverspecific overexpression of SR-BI in mice can result in decreases in cholesterol in VLDL, LDL/intermediate density lipoprotein, and HDL and reductions in apoB and apoA-I (18 -20). Furthermore, SR-BI overexpression has been reported to reduce diet (high fat, high cholesterol) induced increases in VLDL and LDL apoB in transgenic mice (19,20). These data suggest that SR-BI might play a role in LDL as well as HDL metabolism in vivo. An SR-BI null mutation in the chow-diet fed apoE KO mouse model of atherosclerosis dramatically accelerates atherosclerosis (21), and overexpression of SR-BI in either fat-fed LDL receptor-deficient or apoE-deficient mice decreases atherosclerosis (22)(23)(24). Although these studies show that SR-BI is atheroprotective in mice, and thus a potential therapeutic target in humans, the relative importance of the activities of SR-BI as a receptor for HDL versus LDL for this protection from atherosclerosis remains uncertain (22,24).
Structure/function analysis of multifunctional proteins such as SR-BI can be facilitated by the identification of mutant forms of those proteins in which a subset of their activities or properties are disrupted. Here we describe a method to isolate such mutants, called retrovirus library-based activity dissection. This approach is based on three techniques: 1) efficient mutagenesis of all or part of the cDNA encoding the multifunctional protein to generate a large, complex library of mutant cDNAs (here error-prone PCR was used); 2) efficient delivery of single mutant cDNAs from the library into individual cells and high level expression using retroviruses and retrovirus-receptor expressing cells; and 3) isolation of infected cells expressing the desired mutant phenotype (retention of some functions, loss of others) using high sensitivity positive/negative screening by two-color fluorescence-activated cell sorting (FACS). Others have previously reported the use of random mutagenesis, retroviral cDNA transduction, and selection to isolate gain of function mutants with a single mutant phenotype (e.g. altered cell morphology or hormone-independent cell growth, see Refs. 25 and 26).
Using the retrovirus library-based activity dissection method, we identified two sites in mSR-BI that, when simultaneously mutated to arginines, dramatically reduced native HDL binding to the receptor without substantially altering the levels of high affinity LDL binding or uptake of metabolically active cholesterol from LDL. Such "activity dissected" mutants should help clarify the molecular basis for the complex binding properties of SR-BI and, if expressed in animal models, may help provide insights into the mechanism(s) underlying the atheroprotective effects of SR-BI.

EXPERIMENTAL PROCEDURES
Materials-Human HDL, LDL, 125 I-labeled HDL ( 125 I-HDL), 125 Ilabeled LDL ( 125 I-LDL), and DiI (1,1Ј-dioctadecyl-3,3,3Ј,3Ј-tetramethylindocarbocyanine perchlorate)-labeled HDL (DiI-HDL) were prepared as described previously (6,12). All 125 I-HDL preparations were monitored by SDS-polyacrylamide gel electrophoresis to ensure that the preparations were free of radiolytic or oxidative damage. Alexa 488 was purchased from Molecular Probes (Eugene, OR). Alexa 488-labeled HDL (Alexa-HDL) was prepared following the manufacturer's suggestions. In brief, 50 l of 1 M bicarbonate were added to 0.5 ml of 2 mg of protein/ml HDL in PBS and the mixture was then transferred to a vial of reactive dye (Alexa 488 carboxylic acid, succinimidyl ester, dilithium salt). The reaction mixture was stirred for 1 h at room temperature before the reaction was stopped by adding 15 l of hydroxylamine. The labeled HDL was separated from the free dye by exclusion chromatography using the column supplied by the manufacturer. Based on SDSpolyacrylamide gel electrophoresis, Alexa 488, which was expected to react with primary amines of proteins (Molecular Probes, Alexa 488 Protein Labeling Kit Manual, catalog number A-10235), primarily covalently labeled the two major apolipoproteins on the surface of HDL, apoA-I and apoA-II (27, 28) (data not shown). In contrast, in DiI-HDL the DiI associated with the HDL particles but had not covalently labeled the apolipoproteins (not shown). The human CD36 (hCD36) expression vector (29) was the generous gift of B. Seed (Massachusetts General Hospital, Boston). All of the expression vectors for mSR-BI were based on pCDNA1 (Invitrogen). Wild-type mSR-BI expression vectors used for these studies included pmSR-BI number 77 (6) and minor variants (e.g. ex47 and ex68) 2 involving small differences in the linker region to facilitate cloning and restriction digestion. CHO cells were grown or incubated in medium A (Ham's F-12 containing 50 units/ml of penicillin, 50 g/ml streptomycin, and 2 mM glutamine) supplemented with serum or other additions as indicated.
Error-prone PCR-PCR-based random mutagenesis was performed under two different conditions according to the protocols by Leung et al. (30) and Cadwell et al. (31) with some modifications. Under the first condition (30), the reaction mixture contained 1 mM dGTP, 1 mM dCTP, 1 mM dTTP, 0.2 mM dATP, 0.5 mM MnCl 2 , and 5.5 mM MgCl 2 . pmSR-BI number 77 (6) containing the wild-type mSR-BI was used as the template. Primers XG65 (5Ј-GGAAGATCTCCGTCTCCTTCAGGTCCT-GAGC-3Ј) and XG66 (5Ј-TTATCGTCGACGCGTGGGCATCCATGT-GCCGT-3Ј) were used at a concentration of 500 pmol/ml. Under the second condition, 0.2 mM dGTP was used in order to reduce the bias for A/T Ͼ G/C change (31). Amplification was performed on a Perkin-Elmer DNA cycler: initial denaturation (94°C for 4 min) followed by 40 cycles of annealing (50°C for 1 min), elongation (72°C for 2 min), and denaturation (94°C for 1 min). The PCR products from 25 replicate tubes of each PCR under both conditions were pooled together to generate the collection of mutated 1.7-kb products used to construct the mutant library.
Generation of pMX-mSR-BI and the Sub4 Mutant mSR-BI Library-Plasmid pMX-mSR-BI was generated by blunt-end ligating the HindIII/ XbaI fragment of pmSR-BIЈ (12) which contains the coding sequence of the wild-type mSR-BI into the NotI/EcoRI site of the pMX vector (32). For generating the Sub4 mutant library, the 1.7-kb fragments generated by error-prone PCR were digested with BlpI/SalI and cloned into the corresponding position in pMX-mSR-BI. Plasmids were isolated from pooled Escherichia coli colonies without further amplification.
Production of Retroviral Supernatants-Retroviruses were generated in the Phoenix packaging cell line (ATCC SD 3444, grown in DMEM containing 10% fetal bovine serum, 50 units/ml penicillin, 50 g/ml streptomycin, 2 mM glutamine) provided by G. Nolan (Stanford Medical Center) by transfecting the cells with the pMX-mSR-BI plasmid or the Sub4 plasmid library of the mSR-BI mutants using the calcium phosphate method (33). Supernatants were collected 24 and 48 h after transfection, and either were used immediately to infect CHO[Eco] cells (see below) or were frozen in liquid nitrogen and stored at Ϫ80°C before use.
Generation . Cells were selected in medium A containing 5% fetal bovine serum (medium B) and 250 g/ml hygromycin B. Individual colonies were then tested for the expression of human placental alkaline phosphatase (PLAP) after infection with a PLAP-expressing retrovirus construct (MSCV-PLAP, gift from D. Baltimore, Caltech, Pasadena, CA) and the clone which gave the highest infection rate was isolated. This clone, designated CHO[Eco], was used for subsequent studies to express retrovirally encoded mSR-BI proteins. For transduction, CHO[Eco] cells were set at 500,000 cells/100-mm dish in medium B on day 0. On day 1, the cells were transduced with 6 ml of medium B containing 6 g/ml of the adjuvent Polybrene and retrovirus supernatants diluted (1:3 or 1:6) in this medium. After incubation for 24 h at 34°C, the retrovirus containing medium was replaced with fresh medium B. On day 3, cells were harvested with trypsin, reset into two 100-mm dishes, and maintained in medium B without selection. On day 5, cells were processed for labeling with fluorescent lipoproteins and FACS (see below) or were harvested and stored frozen at Ϫ80°C for future FACS screening.
Cell Labeling and Fluorescence-activated Cell Sorting-Cells transduced with retrovirus (see above) were washed twice with Ca 2ϩ -and Mg 2ϩ -free PBS and were then incubated at 37°C for 2 h with DiI-HDL (5 g of protein/ml) in medium A containing 0.5% (w/v) fatty acid free BSA (medium C). Cells were harvested with trypsin, the trypsin was quenched with medium B, and the cells were then washed once with medium C. Cells were then incubated at room temperature for 1 h with 10 g of protein/ml Alexa-HDL in Ca 2ϩ -and Mg 2ϩ -free PBS containing 0.5% BSA. Immediately before FACS analysis and sorting, the cells were pelleted at 500 ϫ g for 2 min, and resuspended in Ca 2ϩ -and Mg 2ϩ -free PBS containing 0.5% BSA. FACS analysis was performed with a Becton-Dickson FACStar instrument. ldlA[mSR-BI] cells (6) labeled separately with either DiI-HDL or Alexa-HDL were used to set the compensation of the instrument to correct the fluorescence signal spill-over between optical channels. Doubly labeled CHO[Eco] cell pools transduced with MX-mSR-BI virus at low multiplicity of infection (m.o.i.) were used as a reference to set the region for collecting the DiI Ϫ /Alexa ϩ cells. Cells showing low DiI/Alexa signal ratio (DiI Ϫ /Alexa ϩ ) were isolated, expanded, and subjected to additional rounds of fluorescence labeling, FACS analysis, and sorting. 2 X. Gu, R. Lawrence, and M. Krieger, unpublished data.
Recovery of the Integrated Mutant mSR-BI cDNAs-Genomic DNAs from the DiI Ϫ /Alexa ϩ CHO[Eco] cells were isolated (35) and the mutant mSR-BI transgenes were cloned using high fidelity Platinum Taq DNA polymerase (Life Technology) and primers MX1(5Ј-CCACCGCCCT-CAAAGTAGACG-3Ј, upstream primer in the 5Ј-viral long terminal repeat region) and XG66 (downstream primer). The resulting 1.7-kb PCR fragments were digested with BglII/BstXI and cloned into the BamHI/BstXI sites in plasmid ex47.
To generate the double mutant with Q402R/Q418R, we cloned the XbaI/FspI fragment from plasmid pmSR-BI number 77 into the corresponding site of plasmid M4-D (see "Results") to generate plasmid VM12Ј. The XbaI/BlpI fragment from VM12Ј was then cloned into the corresponding site in plasmid ex68 to generate plasmid VM54, which encodes the mSR-BI mutant with the Q402R/Q418R double mutation. All the site-specific mutations mentioned above were confirmed by DNA sequencing.
Fluorescence Flow Cytometric Analysis of Protein Surface Expression-Cells grown in 6-well dishes were labeled with 1 ml of Ca 2ϩ -and Mg 2ϩ -free PBS containing 0.5% BSA and rabbit polyclonal antibody against the extracellular domain of mSR-BI (1:1000 dilution, generous gift from K. Kozarsky) for 1 h at room temperature. Cells were then washed 3 times with Ca 2ϩ -and Mg 2ϩ -free PBS and labeled with FITCconjugated goat anti-rabbit IgG (1:1000 dilution, Cappel, West Chester, PA) in 1 ml of Ca 2ϩ -and Mg 2ϩ -free PBS containing 0.5% BSA, 2 mM EDTA. After 20 min, cells were detached from the plastic by gentle pipetting and incubated at room temperature for another 20 min. Cells were then pelleted and resuspended in 1 ml of Ca 2ϩ -and Mg 2ϩ -free PBS containing 0.5% BSA and subjected to fluorescence flow cytometric analysis. The mean value of the fluorescence intensity was used as a measure of mSR-BI surface expression.
ACAT Assay-Cells were maintained in medium A containing 5% newborn calf lipoprotein-deficient serum (NC-LPDS, medium D) for at least 2 weeks before the ACAT assay to prevent differential buildup of intracellular cholesterol in cells expressing different lipoprotein receptors. Cells were set in 6-well dishes at 80,000 cell/well on day 0 in medium D and refed with the same medium on day 2. On day 3, cells were washed twice with Ca 2ϩ -and Mg 2ϩ -free PBS and refed with 1 ml of medium C containing the indicated amounts of either HDL or LDL. After a 5-h incubation at 37°C, 10 l (100 nmol) of [ 3 H]oleate/BSA mixture (21,804 dpm/nmol, a gift from Dr. L. Liscum, Tufts University, Ref. 36) was added to each well. After incubation at 37°C for 1.5 h, the cells were washed twice with ice-cold Tris wash buffer (50 mM Tris-HCl, 0.15 M NaCl, pH 7.4) containing 2 mg/ml BSA, followed by two quick washes with Tris wash buffer without BSA. Two ml of hexane/isopropyl alcohol (3:2) were then added to each well. After 15 min of shaking at room temperature, 10 l of recovery standard (2 mg/ml cholesterol oleate, 1 mg/ml triolein, 50 Ci/ml cholesterol [ 14 C]oleate in chloroform/ methanol (2:1)) were added to each well. The plates were shaken for another 15 min before the organic extracts were transferred to 12 ϫ 75-mm glass tubes. Cells were extracted with addition of another 1 ml of hexane/isopropyl alcohol and the extracts were pooled and dried under nitrogen gas. Samples were then dissolved in 50 l of chloroform and separated by silica gel thin layer chromatography (developed with heptane/diethyl ether/glacial acetic acid (90:30:1)). The cholesteryl esters were visualized by staining with iodine vapor and cut away from the silica gel plates (Silica Gel 60, Fisher Scientific). The amounts of cholesteryl [ 3 H]oleate formed were measured using a Tri-carb liquid scintillation analyzer (Packard). For protein determinations, cells were dissolved in 1 ml of 0.1 N NaOH after solvent extraction and the protein levels determined using the method of Lowry et al. (37).
ACAT assays were also performed using a slightly modified protocol with cells maintained as stock cultures in medium B. For these experiments, the cells were set on day 0 in medium B as described above. On day 2, cells were washed once with Ca 2ϩ -and Mg 2ϩ -free PBS and refed with medium D containing 10 M compactin and 100 M mevalonate to suppress cholesterol synthesis. ACAT activity was then determined on day 3 as described above.
Miscellaneous-Transient transfection, fluorescence flow cytometric analysis, 125 I-HDL binding, and 125 I-LDL binding were performed as described previously (6,12,38). For measuring 125 I-LDL binding at 4°C, transiently transfected COS cells were first treated with DMEM containing 10% newborn calf lipoprotein-deficient serum, 10 g/ml cholesterol, 1 g/ml 25-hydroxycholesterol overnight to reduce the background level of 125 I-LDL binding activity due to endogenous LDL receptor activity. DNA sequencing was performed by the biopolymers laboratory at MIT. Stable ldlA-7 (39) cell lines expressing the mutant SR-BI Q402R/Q418R were generated using the GenePORTER transfection reagent according to the manufacturer's (GTS INC.) instructions. The transfected cells were screened by flow cytometry using an anti-mSR-BI polyclonal antibody (see above) to select a clone (ldlA[Q402R/ Q418R], clone number 6) which exhibited surface expression of receptor protein comparable to that of the wild-type receptor expressed by previously described ldlA[mSR-BI] cells (6).

RESULTS
The approach we have used to isolate novel, activity dissected, mSR-BI mutants relies on the complex characteristics of the interaction of mSR-BI with HDLs labeled with distinct fluorescent dyes, and on three powerful techniques: retrovirusbased expression of single cDNAs in cells infected at a low multiplicity of infection, dual-color FACS, and error-prone PCR in vitro mutagenesis.
Complex Characteristics of Native and Modified Lipoprotein Binding to SR-BI-mSR-BI can bind both native HDL (apparent K d ϳ16 g of protein/ml (ϳ160 nM)) and native LDL (apparent K d ϳ 5 g of protein/ml (ϳ10 nM) at 37°C) (1). 3 However, LDL is a poor competitor (compared with HDL itself) for HDL binding to mSR-BI (6). 2 These data suggest that the mode of binding of LDL is distinct from that of HDL (2,6).
Previous studies indicated that HDL noncovalently labeled with the fluorescent lipid DiI (DiI-HDL) binds to cells expressing mSR-BI in a fashion similar to that of native HDL, and that the mechanism by which DiI is transferred by selective uptake to the cells is similar to that for the cholesteryl esters in HDL (6,40). Thus, the accumulation of DiI fluorescence by cells expressing SR-BI is dependent on both DiI-HDL binding and the subsequent selective uptake of the dye. We have also seen that HDL covalently labeled on its protein constituents by the fluorescent dye Alexa (Alexa-HDL) binds to SR-BI. Most of the Alexa-HDL binding appears to resemble the binding of native or 125 I-HDL (12). Indeed, results from studies examining the distinct properties of SR-BI and another class B scavenger receptor, CD36, using Alexa-HDL, 125 I-HDL, and DiI-HDL (12), or HDL labeled with [ 3 H]cholesteryl ethers and 125 I-labeled protein (13) were essentially identical. (Both SR-BI and CD36 bind HDL, but only SR-BI efficiently mediates the cellular uptake of lipid from HDL.) However, alexylation involves the covalent modification of positively charged lysine residues on the apolipoproteins of HDL and their conversion to negatively charged side chains. This raised the possibility that some features of Alexa-HDL binding to wild-type or mutant forms of SR-BI might differ from those of native HDL. The results reported below show that this is the case.
Rationale for Retrovirus Library-based Activity Dissection of the Binding Properties of SR-BI-We have developed a technique to isolate SR-BI mutants which have lost some, but not all of SR-BIs distinctive functional properties. Our goal was to identify mutants which could bind Alexa-HDL but exhibited reduced uptake of DiI from DiI-HDL. A DiI uptake negative (DiI Ϫ ) but Alexa-HDL binding positive (Alexa ϩ ) phenotype could arise because of specific loss of the capacity to selectively take-up lipids from HDL without loss of HDL binding activity. The properties of such mutants would resemble those of native CD36, an HDL receptor which can bind HDL but does not efficiently mediate the cellular uptake of HDL-associated lipids (12,13). Alternatively, a DiI Ϫ /Alexa ϩ phenotype could arise if the mutant SR-BI lost the ability to bind DiI-HDL, and perhaps native HDL, but retained the ability to bind Alexa-HDL because of its altered surface charge. Indeed, we have isolated and characterized such HDL binding defective (yet Alexa-HDL and LDL binding positive) SR-BI mutants (see below).
Generation of SR-BI Mutant Library Sub4 by Error-prone PCR-To generate a plasmid library of mutant mSR-BI cDNAs (see "Experimental Procedures"), we used error-prone PCR to amplify the full-length mSR-BI cDNA. The 1.7-kb PCR products were digested with BlpI/SalI and cloned into the corresponding position of pMX-mSR-BI (a plasmid containing the full-length, wild-type mSR-BI cDNA cloned into the expression cassette (viral long terminal repeat driven) of the pMX plasmid) to generate the Sub4 library. The region of the mSR-BI protein corresponding to the mutagenized segment of cDNA in the Sub4 library included the C-terminal cytoplasmic domain, the transmembrane domain adjacent to the C-terminal domain, and a portion of the extracellular domain (see crosshatched area in Fig. 1). Libraries in which other segments of the wild-type mSR-BI cDNA were replaced with their corresponding error-prone PCR fragments were also generated 2 but will not be considered here. Sequencing of portions of randomly selected clones isolated from the original full-length mutated PCR products indicated an average mutation rate of about 9 nucleotide changes per 1527 base pairs (the length of the complete SR-BI coding sequence). 2 No mutational hot spots were found; however, as has been reported previously (30), there was a much higher incidence of A Ͼ G and T Ͼ C changes than other mutations (data not shown). We estimate that the Sub4 library had a complexity of ϳ1.2 ϫ 10 5 .

Retrovirus-based Expression of SR-BI in Cells Infected at a Low Multiplicity of Infection and Two-color Fluorescence Flow
Cytometric Analysis-The identification and isolation of rare mutant clones with the desired phenotype (DiI Ϫ /Alexa ϩ ) from our randomly mutagenized libraries required efficient, high level expression of single cDNA clones in individual cells which express a low background of endogenous SR-BI activity. Sensitive detection and isolation of rare mutants was also essential. We chose retroviruses as the vectors for the mutant library expression, because of their high efficiency in stably transducing cells expressing retrovirus receptors and their ability to very efficiently mediate high level expression of genes of interest, expression can be readily detected in cells infected by a single virion (41). Therefore, we converted the PCR-mutated plasmid library Sub4 into a library of retroviruses by transfecting the plasmids into the Phoenix retrovirus packaging cell line. Viral supernatants were harvested by collecting the media from the transfected cells (see "Experimental Procedures"). As a target for the retrovirus infection, we generated an essentially mSR-BI-negative CHO cell line which stably expresses the murine ecotropic retrovirus receptor (34) . Cell-to-cell variations in the levels of wild-type mSR-BI expression were presumably due to differences in the sites of provirus integration into the genomes of the host cells. Some of the variation at the higher levels of expression might have been due to multiple viruses infecting individual cells despite the low m.o.i. These data indicate that the efficiency of transduction and levels of mSR-BI expression were sufficient to permit the isolation of cells expressing the desired mutants should they be generated by the error-prone PCR. The relatively cell-free region in the lower right of the cytogram for the transduced cells is demarcated by a box (Fig. 2, right Fig. 3 shows the cytograms from each of the four rounds of sorting. The boxes in the lower right quadrants represent the cells selected (approximately 5,000 out of a total of 4 million screened during the first round of FACS) and grown to mass culture. After 3 rounds of sorting, a clear enrichment of cells with a low DiI signal but a high Alexa signal was observed.
Cloning of cDNAs Encoding Mutant DiI Ϫ /Alexa ϩ mSR-BIs-Genomic DNAs from pooled cells from the fourth round of sorting were isolated and the integrated mSR-BI transgenes were amplified by high fidelity PCR and cloned into the mammalian cell expression vector pCDNA1. Plasmids were isolated from individual E. coli colonies and transfected into COS cells. These transiently transfected cells were incubated with DiI-HDL and Alexa-HDL and analyzed by dual-color fluorescence flow cytometry. Fig. 4 shows the cytograms for the COS cells transfected with three mutant plasmids derived from the Sub4 library (bottom cytograms) and control plasmids encoding: pCDNA1 (empty vector negative control (top left), CD36 (a DiI Ϫ /Alexa ϩ class B scavenger receptor, top middle), and wildtype mSR-BI (top right). As described previously (12), the cells transfected with the negative control pCDNA1 ("control") were DiI Ϫ /Alexa Ϫ (most cells in the lower left quadrant of the cytogram), the CD36 expressing cells exhibited a DiI Ϫ /Alexa ϩ phenotype (many cells in the lower right quadrant), and the SR-BI expressing cells were DiI ϩ /Alexa ϩ (essentially linear distribution of positive slope in the upper right quadrant). The cytograms of mutants M4-A, M4-B, and M4-C (bottom) were similar and resembled that of CD36, i.e. a DiI Ϫ /Alexa ϩ phenotype (also see Refs. 13 and 42), although the mutants differed substantially from CD36 (see below). These data show that the DiI Ϫ /Alexa ϩ phenotypes of the stably transfected CHO[Eco] cells isolated by flow cytometry were directly attributable to the characteristics of individual mSR-BI cDNAs isolated from the cells.
Seven cDNAs derived from DiI Ϫ /Alexa ϩ cells infected with the Sub4 library were sequenced. Fig. 5 shows the encoded amino acid sequences (corresponding to the crosshatched region in Fig. 1). All of the cDNAs contained multiple mutations, including frameshift mutations due to deletion of either an A or T near the C termini of mutants M4-C and M4-D. Six of the seven cDNAs represented independent mutants (i.e. only two of the plasmids had identical sequences). Strikingly, two sites in the extracellular portion of all 6 independent clones had either an identical amino acid substitution, Q418R, or a substitution of arginine at one of two adjacent residues, G401R or Q402R. The Q418R and Q402R substitutions were due to A/T 3 G/C substitutions. The G401R in two of the mutants resulted from a C/G 3 T/A change while the third such substitution was due to a C/G 3 G/C change.
The presence of the same type of mutation at two conserved sites (G401R (or Q402R) and Q418R) in all 6 independent clones suggested that both of these mutations, but none of the non-conserved mutations were important for conferring the DiI Ϫ /Alexa ϩ phenotype. To test this, we constructed a series of mutants with single or double point mutations at these positions in an otherwise wild-type backbone, and examined by dual-color flow cytometry the phenotypes of COS cells transfected with these plasmids. Fig. 6 shows the results of two independent experiments. The first (panel A) shows that mSR-BI carrying both the Q402R and Q418R mutations, but no other mutations exhibited the strong DiI Ϫ /Alexa ϩ phenotype seen in cells expressing the cDNAs from the Sub4 library (compare far right cytogram with those in Fig. 4). All of the

FIG. 2. Dual-color fluorescence flow cytometric analysis of mSR-BI expression in CHO[Eco] cells transduced with the MX-mSR-BI-derived retrovirus. CHO[Eco]
cells which express the ecotropic retrovirus receptor were plated on day 0 and either were untransduced (left panel) or were transduced on day 1 at a low multiplicity of infection (ϳ26% of the cells were transduced) with the MX-mSR-BI retrovirus (a construct containing the wild-type mSR-BI cDNA cloned into the pMX vector) as described under "Experimental Procedures." On day 5, cells were incubated for 2 h at 37°C in medium C containing DiI-HDL (5 g of protein/ml), harvested, incubated for 1 h at room temperature in Ca 2ϩ -and Mg 2ϩ -free PBS containing 0.5% BSA and Alexa-HDL (10 g of protein/ml), and then were analyzed by dual-color fluorescence flow cytometry. For each cell analyzed the relative intensities of Alexa fluorescence (horizontal axes) and DiI fluorescence (vertical axes) are indicated by a dot and presented on log scales. Each panel represents analysis of 10,000 cells. The relatively cell-free region in the lower right of the cytogram for the transduced cells (right cytogram) demarcated by a box represents the region which would include signals from cells exhibiting a DiI Ϫ /Alexa ϩ phenotype.

FIG. 3. Isolation of rare DiI ؊ /Alexa ؉ CHO[Eco] cells by multiple rounds of dual-color FACS. CHO[Eco] cells infected with the
Sub4 library were double-labeled with DiI-HDL and Alexa-HDL and subjected to dual-color FACS. Cells which exhibited low DiI fluorescence relative to Alexa fluorescence (boxed region on the cytogram) were collected. About 4 ϫ 10 6 cells were analyzed and 5,000 cells collected during the first round of sorting (Sort 1). After expansion, this population of cells was subjected to three additional rounds of labeling and sorting. The dual-color flow cytograms (see Fig. 2  single-site mutants (G401R, Q402R, or Q418R) were, as expected, Alexa ϩ . Although the DiI uptake activity in the singlesite mutants was reduced relative to wild-type mSR-BI, it was quantitatively less severe than in the double mutant. The effects of the single mutations at positions 401 and 402 were similar and more severe than those of the single mutation at 418. Thus, the retrovirus library-based method successfully permitted the isolation of DiI Ϫ /Alexa ϩ activity dissected SR-BI mutants. These mutants have led to the identification of residues at two sites in the sequence which play an important and mutually reinforcing roles in establishing the ability of SR-BI to take up DiI from DiI-HDL.
To determine if generation of the DiI Ϫ /Alexa ϩ phenotype required substitution of positively charged arginine for glutamine at positions 402 and 418, we constructed single site mutants in which the glutamine mutations at these positions FIG. 4. Flow cytometric analysis of COS M6 cells transiently transfected with expression vectors for wild-type and mutant class B scavenger receptors. COS cells were plated in DMEM containing 10% fetal bovine serum, 50 units/ml penicillin, 50 g/ml streptomycin, 2 mM glutamine on day 0, transiently transfected with the indicated plasmids on day 1, and reset at 150,000 cells/well in 24-well dishes on day 2 as described previously (6,12,38). On day 3, the cells were incubated with DiI-HDL and Alexa-HDL and analyzed by fluorescence flow cytometry as described in the legend to Fig. 2. Cells were transfected with the pCDNA1 vector without an insert (Control, top left), or with pCDNA1 plasmids encoding wild-type mSR-BI (top right), three putative mutant mSR-BI cDNAs (M4-A, M4-B, and M4-C, bottom panels) and human CD36 (hCD36, top center). hCD36 transfected cells were included because they exhibit a DiI Ϫ /Alexa ϩ phenotype due to the inability of this receptor to efficiently take up lipid from bound HDL (12). The distribution of the DiI ϩ /Alexa ϩ wild-type mSR-BI-expressing cells (e.g. top right) is indicated in each cytogram by an arrow to facilitate comparisons with the other cell distributions. Each panel represents analysis of 5,000 cells. were replaced by other residues with either positive (lysine), negative (glutamic acid), small neutral (alanine) or moderately large hydrophobic (leucine) side chains. The results of dualcolor flow cytometric analysis of COS cells transfected with these constructs are shown in Fig. 6 (panel B). The effects of substitutions at position 402 with lysine, alanine, and leucine were similar to those for the arginine mutation: reduced DiI uptake relative to Alexa-HDL binding (reduced slope of the distributions in the cytograms). Thus, normal levels of SR-BIdependent DiI uptake from DiI-HDL depended on the presence of the glutamine at position 402 (Q402R is a loss-of-function mutation). Indeed, Gln 402 is conserved in SR-BI from all species reported to date. Substitution of Gln 402 with Glu dramatically decreased both DiI uptake and Alexa-HDL binding. Additional experiments will be required to determine if this loss of both activities was due to reduced or abnormal synthesis, folding, post-translational processing, subcellular localization (e.g. surface expression), stability, and/or intrinsic binding capacity.
In contrast to position 402, the presence of glutamine at position 418 was not essential for normal DiI uptake and Alexa-HDL binding. The cytograms for Glu 418 , Ala 418 , and Leu 418 were essentially identical to that of the wild-type Gln 418 . However, the substitution of a positively charged lysine side chain at 418 resulted in a cytogram similar to that for the positively charged Arg 418 mutant. Thus, substitution of a positively charged residue at position 418, rather than the loss of the side chain of the glutamine at that site, appears to have been required to generate the mutant phenotype. The finding that Gln 418 is not conserved in SR-BI from all species (e.g. Glu 418 in human SR-BI, also called CLA-1 (43)) is consistent with this interpretation. 4 Why did all of the mutants isolated from the Sub4 library have substitutions at positions 401/402 and 418? A definitive answer is not available, but there are several possible explanations. The wild-type codons at positions 402 (CAA) and 418 (CAG) in conjunction with the bias for A/T 3 G/C substitutions in the error-prone PCR mutagenesis increased the likelihood of glutamine to arginine substitutions at these sites. Also, the double substitution of Arg at positions 401/402 and 418 may have resulted in a quantitatively stronger mutant phenotype (reduced DiI uptake relative to Alexa-HDL binding) than with other DiI Ϫ /Alexa ϩ mutations present in the original library. This, in combination with stringent selection criteria applied in multiple rounds of section, may have lead to a substantial enrichment of one class of mutations (double R) over others. It is important to note that in the current studies we have analyzed the results of only one infection/selection experiment using the Sub4 library. It is not certain that bias for mutations to arginines at positions 401/402 and 418 would be observed in independent experiments.
Effects of the 401/402-418 Double Arginine Mutations on SR-BI Binding to Native Lipoproteins- Fig. 7 shows the concentration dependence of the cell association of 125 I-HDL and 125 I-LDL with COS cells transiently expressing wild-type mSR-BI, the Q402R/Q418R double mutant, or control cells. Most of the high affinity association of 125 I-HDL with wild-type mSR-BI was lost in the double mutant, whose activity was only 4 The residues in CD36 from several species at positions equivalent to 401, 402, and 418 are Glu or Gln, Ala or Val, and Glu, respectively (29,44,45).

FIG. 6. Flow cytometric analysis of COS M6 cells transiently transfected with expression vectors for modified mSR-BIs bearing specific point mutations.
The indicated single (e.g. Gly 401 to Arg 401 (G401R)) or double point mutations at amino acid positions 401, 402, and/or 418 in mSR-BI were introduced into the otherwise wild-type cDNA of mSR-BI in the pCDNA1 expression vector as described under "Experimental Procedures." COS cells were transiently transfected in two independent experiments (panels A and B) with these plasmids and a control plasmid encoding the wild-type mSR-BI (not shown), incubated with DiI-HDL and Alexa-HDL and analyzed by fluorescence flow cytometry as described in the legends to Figs. 2 and 4. The arrows in each cytogram represent the distribution of cells transfected with wild-type mSR-BI, as shown in Fig. 4. Single letter amino acid abbreviations used are as follows: G, glycine; R, arginine; Q, glutamine; K, lysine; E, glutamate; A, alanine; L, leucine. Each panel represents analysis of 5,000 cells.
slightly above that of the control cells (panel A). Similar results were observed when the binding of 125 I-HDL to cells was measured at 4°C (data not shown). We have recently developed an immunoblotting-based quantitative HDL binding assay in which the binding of unlabeled HDL to SR-BI expressing cells can be measured. 5 Virtually identical results were obtained with this immunoreceptor assay as with the radioreceptor assay shown in Fig. 7, the double mutant containing receptors had lost much of their ability to bind to native HDL (not shown). Thus, the radioiodination of HDL did not significantly influence the ability of the lipoprotein to interact with wildtype or mutant mSR-BI.
Differences in cell surface expression of the wild-type and mutant receptors might influence their relative HDL binding activities in transfected cells. Thus, we used a polyclonal antibody which recognizes the extracellular domain of mSR-BI and flow cytometric analysis to determine the relative amounts of surface expression of the receptors (see "Experimental Procedures"). The surface expression data along with background association of 125 I-HDL with the control cells were used to correct the binding data in Fig. 7A for differences in surface expression. The results, Fig. 7B, show that, although there was a reduced level of cell surface expression of the double mutant (mutant Q402R/Q418R, ϳ56% in this experiment) relative to that of the wild-type receptor, when the data were corrected it is clear that the substantial decrease in binding to the mutant was not a consequence of lower surface expression of the pro-tein. There was an intrinsic, substantial reduction in the HDL binding activity of the mutant receptor. Fig. 7C shows the concentration dependence of the cell association of 125 I-LDL with the transfected COS cells, and Fig. 7D shows these data corrected for control cell binding and surface expression levels of the receptors. In contrast to the substantial effects on cell association of 125 I-HDL, the Q402R/Q418R double mutations had little influence on 125 I-LDL cell association. The apparent K d for 125 I-LDL cell association was similar for the cells expressing the wild-type and double mutant receptors (apparent K d values of approximately 3 and 2 g of protein/ml, respectively). Similar results (loss of HDL, but not LDL, binding activity) were obtained when we analyzed the activities of this mutant in stably transfected clones of the LDL receptordeficient cell line ldlA-7 (39) at 37°C (data not shown). We also tested native lipoprotein association with COS cells expressing two mutants isolated from the library screening (M4-A and M4-F, see Fig. 5). These mutants contain non-conserved mutations in addition to the 401/402 and 418 mutations. Essentially the same results were observed: both mutants lost HDL binding, but not LDL binding activities at 37°C (data not shown). Furthermore, we examined binding to COS cells expressing either wild-type mSR-BI or the M4-A mutant at 4°C and observed similar results (Fig. 8). Thus, the Q402R/Q418R double mutations effectively dissected away the bulk of the HDL binding activity, but did not dramatically alter LDL binding. There was a very low level of residual HDL binding to the mutant (e.g. compare the mutant expressing and control cells in Fig. 7, panel A); however, this residual activity was too low  A and B) or 125 I-LDL (panels C and D) were added in 0.5 ml of medium C in the absence (duplicate incubations) or presence (single incubations) of a 40-fold excess of unlabeled HDL or LDL, respectively. After incubation at 37°C for 1.5 h, the amounts of specific 125 I-HDL or 125 I-LDL cell association (nanograms of HDL or LDL protein per 1.5 h/mg of cell protein, panels A and C) were determined by calculating the differences between measurements made in the absence or presence of unlabeled ligand as described (6,12). The data shown are representative of results observed in multiple independent experiments. The error bars (representing the range of the values) for all data were smaller than the symbols presented, except for those data indicated in panels C and D. The data in panels A and C, corrected to normalize for surface receptor expression as indicated below, are illustrated in panels B and D. Data were corrected by first subtracting the specific values from the control cells (COS[Control]) to remove any background from transfection-independent cellular specific binding activity, and then those values for the COS[Q402R/Q418R] cells were multiplied by 1.8 to correct for the lower level of cell surface expression of the Q402R/Q418R mutant relative to the wild-type receptor. Relative levels of surface expression were determined using a polyclonal anti-mSR-BI antibody and flow cytometry as described under "Experimental Procedures." to permit reliable determination of binding parameters.
Effect of 402/418 Double Arginine Mutations on SR-BI-mediated Stimulation of Cholesteryl Esterification by HDL and LDL-Binding of HDL and LDL to the surfaces of the wild-type mSR-BI-expressing cells results in the transfer of cholesteryl esters from these lipoproteins to the cell and the stimulation of cholesterol esterification by ACAT (14). To determine if the double arginine mutations affect mSR-BI mediated lipoprotein stimulation of cellular cholesterol esterification, we measured the HDL and LDL dependence of cholesteryl ester formation in stably transfected ldlA-7 cell lines expressing wild-type mSR-BI (6) or the Q402R/Q418R double mutant (clone number 6). These two lines expressed comparable levels of mSR-BI protein on their surfaces as determined by quantitative flow cytometry using a polyclonal anti-mSR-BI antibody (not shown) or by immunoblotting (Fig. 9, panel AЈ, inset) and similar levels of 125 I-LDL cell association (data not shown). Fig. 9 shows that neither HDL nor LDL substantially stimulated cholesterol esterification in the control, untransfected receptordeficient ldlA-7 cells (panels A and AЈ). In contrast, both lipoproteins stimulated cholesterol esterification in cells expressing the wild-type mSR-BI (panels B and BЈ), as has been reported previously (14). As expected, the stimulation by HDL of ACAT activity in cells expressing the mutant receptor (panel C) was substantially lower than that for the wild-type mSR-BI expressing cells. In the experiment shown in Fig. 9, A-C, the extent of stimulation of ACAT activity by LDL was substantially lower for the ldlA[Q402R/Q418R] cells (panel C) than for the ldlA[mSR-BI] cells (panel B, note the differences in the scales of the ordinates). Basal levels of cellular cholesterol can influence the sensitivity of the ACAT response (46). 6 We suspected that much of the difference in absolute ACAT activities in panels B and C might be eliminated if cells were grown for several weeks prior to the experiment in medium supplemented with lipoprotein-deficient serum rather than complete serum. This was expected to limit the potential effects of the serum lipoproteins on basal ACTA levels. Indeed, Fig. 9, panels A'-C', show that most of the difference in the amount of LDL simulated ACAT activity was eliminated by prior growth in medium containing lipoprotein-deficient serum. Thus, despite their effects on HDL-stimulated ACAT activity, the mutations had little, if any, effect on the capacity of mSR-BI to mediate uptake of metabolically active cholesterol from LDL.
Taken together with the binding data (Fig. 7), these data support the conclusion that the mechanism of binding of LDL to mSR-BI differs significantly from that of HDL. The Q402R/ Q418R double mutations of mSR-BI generated a receptor with apparently normal LDL, but dramatically reduced HDL, binding, and cholesterol delivery activities. It may be useful to compare the consequences of expressing these mutants and wild-type receptors in animals to help differentiate the physiologic consequences of the HDL and LDL binding activities of SR-BI. DISCUSSION We have developed a "retrovirus library-based activity dissection" method to generate mSR-BI mutants in which some, but not all, of the activities of this multifunctional protein have been ablated. The method is based on three techniques: 1) efficient mutagenesis of all or part of the cDNA of the target gene to generate a large, complex library of mutant cDNAs (here error-prone PCR was used), 2) efficient delivery and high expression of single mutant cDNAs into individual cells using retroviruses and retrovirus-receptor expressing cells, and 3) high sensitivity positive/negative screening and isolation of infected cells expressing the desired mutant phenotype by dual-color FACS. Similar approaches, involving single phenotype selections (e.g. altered cell morphology or hormone-independent cell growth), have been described previously for the isolation of gain of function mutants (25,26). The use of dual-color FACS enabled us to simultaneously monitor two distinct activities of the target protein. Thus, by selecting cells which lost one, but not the other, activity (positive/negative screening), we were able to functionally "dissect" the target protein. In the current experiments we analyzed the properties of SR-BI using fluorescent ligands with distinct properties. In principle, the method could be applied to other types of cell surface proteins or to intracellular proteins, provided that at least two nontoxic, distinct fluorescent probes of the protein's activities or structure could be added to living cells and gain access to the target protein. Double selections or screens, positive for one phenotype and negative for another, need not depend on fluorescence flow cytometry; however, this very sensitive and versatile technique is well suited for this approach.
Where practical (e.g. availability of appropriate fluorescent probes), the retrovirus library-based method provides a powerful complement to, and even some advantages over, the alanine-scanning mutagenesis method (47) for analyzing structure/function relationships. For example, this new method permitted the identification of mutants bearing simultaneous mutations at two sites in a single cDNA. One of these mutations involved a loss-of-function when a glutamine side chain (Gln 402 ) was replaced with one of several diverse residues. 6 L. Liscum, personal communication.

FIG. 8. Concentration dependence of 125 I-HDL (A) and 125 I-LDL (B) binding at 4°C to COS M6 cells transiently transfected with expression vectors for wild-type mSR-BI and the M4-A mutant.
COS cells were plated in DMEM containing 10% fetal bovine serum, 50 units/ml penicillin, 50 g/ml streptomycin, 2 mM glutamine on day 0, and transiently transfected with plasmids encoding the indicated genes on day 1 as described previously (6,12,38). On day 2, cells were reset at 150,000 cells/well in 24-well dishes in DMEM containing 10% fetal bovine serum, 1 mM sodium n-butyrate (for 125 I-HDL binding assay), or DMEM containing 10% newborn calf lipoprotein-deficient serum, 1 mM sodium n-butyrate, 10 g/ml cholesterol, and 1 g/ml 25-hydroxycholesterol to suppress endogenous LDL receptor activity (for 125 I-LDL binding assay). On day 3, the indicated amounts of either 125 I-HDL (panel A) or 125 I-LDL (panel B) were added in 0.5 ml of ice-cold medium C in the absence (duplicate incubations) or presence (single incubations) of a 40-fold excess of unlabeled HDL or LDL, respectively. After incubation at 4°C for 2 h, the amounts of specific 125 I-HDL or 125 I-LDL cell association (nanograms of HDL or LDL protein per 2 h/mg of cell protein were determined by calculating the differences between measurements made in the absence or presence of unlabeled ligand as described previously (6,12).
Mutation at the other position (418) required the replacement of a glutamine side chain by a positively charged side chain (Lys or Arg, a substitution with Ala was not effective). The ability to generate mutants of altered phenotypes that require multiple mutations and/or specific types of amino acid substitutions is an especially attractive feature of this method. Indeed, using a retrovirus-based approach and single selection for a constitutively active form of the regulatory protein STAT5, Onisi et al. (26) isolated a STAT5 mutant whose phenotype depended on two amino acid substitutions. Future development of this approach should include refinement of the mutagenesis step (improve complexity of libraries, vary extent and diversity, as well as the method, of mutagenesis). It should also involve exploring the effects on the diversity of the isolated clones of varying: (i) the length of the cDNA segment subjected to mutagenesis, (ii) the stringency of the selection/screening and the characteristics of the fluorescent ligands employed, and (iii) the number of rounds of selection/screening. SR-BI was the subject of this analysis because: 1) it binds both HDL and LDL (1, 6); 2) earlier studies suggested that the mechanisms by which these lipoproteins bind to SR-BI are distinct (6); and 3) SR-BI mediates the selective uptake of lipid (cholesteryl esters, DiI) from HDL and LDL (6, 14 -16). One goal of the study was to identify residues in SR-BI which differentially contribute to these activities. Another was to generate activity dissected forms of SR-BI which could help in evaluating the relative roles of SR-BI's diverse activities on its in vivo functions (e.g. control of lipoprotein metabolism, atheroprotective effects in murine models of atherosclerosis) (2,(21)(22)(23)(24). In the experiments reported here, we screened for mutants which could bind HDL covalently modified on lysine residues with the negatively charged green fluorescent dye Alexa (Alexa-HDL), but not take up the lipophilic red fluorescent dye DiI when it was noncovalently incorporated into HDL (DiI-HDL).
The DiI Ϫ /Alexa ϩ phenotypes of the mutants isolated from a library of SR-BI mutagenized in its C-terminal region (Sub4) were initially expected to represent mutants which could bind native HDL, but could not efficiently mediate selective lipid uptake. This phenotype would resemble that of CD36, another class B scavenger receptor (Refs. 12 and 13, also see Ref. 42). Instead, the first experiment employing the technique lead to the isolation of a set of conserved double mutants (G401R or Q402R and Q418R) which lost most of their ability to bind native and DiI-HDL, but not Alexa-HDL. It seems likely that the increased positive charge on the mutant receptors might have had dual effects: interfering with native HDL binding while simultaneously promoting binding of the more negatively charged Alexa-HDL. These mutations, however, had little apparent direct effect on LDL binding. These results support the suggestion that the mechanism of binding of LDL to SR-BI differs significantly from that of HDL. The G401R (or Q402R) and Q418R double mutations effectively dissected the bulk of SR-BIs HDL binding activity away from the LDL binding activity of this receptor, thus generating novel mutant forms of ACAT activities were measured on cells grown in medium supplemented with either 5% fetal bovine serum (medium B, panels A-C)) or 5% newborn calf lipoprotein-deficient serum (medium D, panels A'-C'). The cells were maintained in medium D for at least 2 weeks prior to the ACAT assay (day 0). The indicated stably transfected cells were seeded in 6-well dishes at 80,000 cells/well in medium B (panels A-C, Experiment 1) or medium D (panels A'-C', Experiment 2). Cells were refed on day 2 with either medium D containing 10 M compactin, 100 M mevalonate (Experiment 1) or medium D only (Experiment 2). On day 3, the cells were washed twice with Ca 2ϩ -and Mg 2ϩ -free PBS and then incubated for 5 h at 37°C in 1 ml of medium C containing either no additions (None or 0) or the indicated amounts of HDL or LDL. Afterward, 10 l (100 pmol) of [ 3 H]oleate (ϳ21.840 dpm/nmol) in a BSA complex (36) were added to each well and the 37°C incubation was continued for an additional 1 h (Experiment 1) or 1. SR-BI. While it is possible that the size difference between large LDL particles and smaller HDL particles might contribute to the different modes of binding of these lipoproteins, it seems likely that other differences in the structures of these lipoproteins (e.g. apolipoprotein composition and conformation) play important roles in establishing the differential binding. Additional screening of the Sub4 and other mutant libraries will be required to isolate HDL binding-positive, selective uptake-negative (CD36-like) mutants.
There is currently some uncertainty regarding the relative importance of the HDL and LDL receptor activities of SR-BI for SR-BIs atheroprotective effects in mice (hepatic overexpression reduces atherosclerosis while genetic ablation promotes atherosclerosis; Refs. [21][22][23][24]. Some experiments suggest that the HDL receptor activity of SR-BI is predominant (24) while others indicate that the non-HDL receptor activity may be key (22). Examination of the in vivo functions of activity dissected mutants such as those described here may help resolve this question.