Binding of high density lipoprotein (HDL) and discoidal reconstituted HDL to the HDL receptor scavenger receptor class B type I. Effect of lipid association and APOA-I mutations on receptor binding.

The binding of apoA-I-containing ligands to the HDL receptor scavenger receptor class B type I (SR-BI) was characterized using two different assays. The first employed conventional binding or competition assays with (125)I-labeled ligands. The second is a new nonradioactive ligand binding assay, in which the receptor-associated ligand is detected by quantitative immunoblotting ("immunoreceptor assay"). Using both methods, we observed that the K(d) value for spherical HDL (density = 1.1-1.13 g/ml) was approximately 16 microgram of protein/ml, while the values for discoidal reconstituted HDL (rHDL) containing proapoA-I or plasma apoA-I were substantially lower (approximately 0.4-5 microgram of protein/ml). We also observed reduced affinity and/or competition for spherical (125)I-HDL cell association by higher relative to lower density HDL and very poor competition by lipid-free apoA-I and pre-beta-1 HDL. Deletion of either 58 carboxyl-terminal or 59 amino-terminal residues from apoA-I, relative to full-length control apoA-I, resulted in little or no change in the affinity of corresponding rHDL particles. However, rHDL particles containing a double mutant lacking both terminal domains competed poorly with spherical (125)I-HDL for binding to SR-BI. These findings suggest an important role for apoA-I and its conformation/organization within particles in mediating HDL binding to SR-BI and indicate that the NH(2) and COOH termini of apoA-I directly or indirectly contribute independently to binding to SR-BI.

I-labeled ligands. The second is a new nonradioactive ligand binding assay, in which the receptor-associated ligand is detected by quantitative immunoblotting ("immunoreceptor assay"). Using both methods, we observed that the K d value for spherical HDL (density ‫؍‬ 1.1-1.13 g/ml) was ϳ16 g of protein/ml, while the values for discoidal reconstituted HDL (rHDL) containing proapoA-I or plasma apoA-I were substantially lower (ϳ0.4 -5 g of protein/ml). We also observed reduced affinity and/or competition for spherical 125 I-HDL cell association by higher relative to lower density HDL and very poor competition by lipid-free apoA-I and pre-␤-1 HDL. Deletion of either 58 carboxyl-terminal or 59 amino-terminal residues from apoA-I, relative to full-length control apoA-I, resulted in little or no change in the affinity of corresponding rHDL particles. However, rHDL particles containing a double mutant lacking both terminal domains competed poorly with spherical 125 I-HDL for binding to SR-BI. These findings suggest an important role for apoA-I and its conformation/organization within particles in mediating HDL binding to SR-BI and indicate that the NH 2 and COOH termini of apoA-I directly or indirectly contribute independently to binding to SR-BI.
SR-BI 1 is a high affinity HDL receptor that recognizes apoA-I and other apolipoproteins (1,2). Following HDL binding to SR-BI, cells selectively take up cholesteryl esters, and their intracellular concentration increases as a function of time (2). SR-BI is predominantly expressed in steroidogenic tissues and the liver (2,3). Overexpression of SR-BI in mice following infection with recombinant adenoviruses or in transgenic mice dramatically decreases plasma HDL cholesterol levels, increases biliary cholesterol levels, and suppresses atherosclerosis (4 -8, 82). Disruption of the SR-BI gene in mice increases by more than 2-fold plasma HDL cholesterol levels without altering plasma apoA-I levels; increases the size of HDL; decreases the cholesterol content of the adrenal gland, ovary, and bile (9,10); and induces dramatically accelerated atherosclerosis in apoE knockout mice (10). These data suggest that SR-BI plays an important role in the last steps of murine reverse cholesterol transport and in the delivery of HDL cholesterol to steroidogenic tissue for maintenance of cholesteryl ester stores and the synthesis of steroid hormones. In addition, female homozygous null SR-BI knockout mice are infertile (10).
In addition to mediating binding to SR-BI, apoA-I in HDL can directly or indirectly promote the efflux of cholesterol from peripheral cells (11), and regulate lecithin:cholesterol acyltransferase activity (12,13). Thus, apoA-I may play an important role in regulating the cholesterol content of peripheral tissues either through selective cholesterol delivery or cholesterol efflux pathways (11, 14 -16).
In the current study, we have focused on the SR-BI binding of different apoA-I-containing ligands, including spherical HDL isolated by ultracentrifugation or immunoaffinity chromatography, pre-␤-1 HDL, lipid-free apoA-I, and reconstituted HDL (rHDL) particles containing wild-type and truncated apoA-I forms. We have used standard "radioreceptor" binding assays (2), in which the ligands are labeled with 125 I on their protein moieties. To ensure that 125 I labeling of the ligands did not affect their binding to SR-BI, we developed a new "immunoreceptor assay," which permits analysis of binding without the need to radiolabel the ligands. The binding parameters determined using the immunoreceptor assay were similar to those determined using the standard radioreceptor assay. Using both assays for direct binding measurements as well as competition experiments, we have shown that apoA-I's affinity for SR-BI was higher when it was incorporated into discoidal rHDL par-ticles than when it was present in spherical HDL (HDL 2 or HDL 3 ), ␣ HDL, pre-␤-1 HDL, or when it was lipid-free. Higher density spherical HDLs bound less tightly to SR-BI than did lower density HDLs. Analysis of rHDL particles containing truncated apoA-I forms showed that deletion of either the carboxyl-terminal sequence 185-243 (apoA-I:⌬(185-243)) or the amino-terminal sequence 1-59 (apoA-I:⌬(1-59)) resulted in little or no change in the affinity of the corresponding rHDL particles for SR-BI. In contrast, rHDL particles containing a double mutant lacking both of these amino and carboxyl-terminal regions competed poorly for the binding of native HDL to SR-BI.

Plasmid Constructions and Generation of Cell Lines Expressing
ProapoA-I and ApoA-I:⌬(185-243)-The fourth exon of the human apoA-I gene was amplified and mutagenized by polymerase chain reaction, using a set of specific mutagenic primers (185S, 185A), containing the mutation of interest (a stop codon at nucleotide 185), and a set of flanking universal primers (AINotI, AIXhoI), containing the restriction sites NotI and XhoI. The sequences of the primers used are shown in Table I. The pUCA-I N * vector, which contains a NotI site in intron 3 and an XhoI site in the 3Ј-end of the apoA-I gene, was used as a template in the amplification reactions (18). The DNA fragment containing the mutation of interest was digested with NotI and SalI and cloned into the NotI and XhoI sites of the pBMT3X-AI vector. The variant apoA-I sequences were verified by DNA sequencing.
Permanent cell lines in mouse mammary tumor C127 cells expressing the wild-type proapoA-I and the carboxyl-terminal truncation apoA-I:⌬(185-243) were generated as described previously (18). C127 cell clones overproducing the wild-type and the variant apoA-I form were grown in roller bottles on collagen-coated lead microspheres (Verax Corp.), and the protein was purified from the serum-free medium as described previously (18).
Generation of Amino-and Carboxyl-terminal ApoA-I Deletion Constructs: Expression in the Baculovirus System and Protein Purification-To generate the recombinant baculoviruses expressing wild-type and mutant apoA-I forms, a BamHI-SalI fragment containing human apoA-I cDNA was cloned into the polylinker region of pFASTBAC donor plasmid, which contains the ampicillin and gentamycin resistance genes (Life Technologies). This recombinant plasmid also contains a histidine tag and the tobacco etched viral protease cleavage site. The apoA-I-containing plasmid was used to transform DH10 Bac Escherichia coli cells (Life Technologies). These cells had been transformed previously with the baculovirus genome containing the lacZ and kanamycin resistance genes, along with a helper plasmid containing the tetracycline resistance gene and the transposase genes. Transposition of the apoA-I gene into the lacZ gene disrupts the expression of lacZ and provides a recombinant plasmid named bacmid. For the generation of the donor plasmid expressing the apoA-I:⌬(1-59) form, the apoA-I region between amino acid 60 and the end of 3Ј-untranslated region was amplified using 5Ј (AIM2-5) and 3Ј (AIM1-3) primers (see Table I). The 5Ј and 3Ј primers contain BamHI and SalI restriction sites, respectively. The amplified fragment was digested with BamHI and SalI and cloned into the corresponding sites of the pFASTBAC donor plasmid. For the generation of the donor plasmid expressing the (⌬1-59 and ⌬185-243) A-I, the region between amino acids 60 and 184 was amplified using the 5Ј (AIM2-5) and 3Ј (AIM3-3) primers, respectively (see Table I). The AIM3-3 primer introduces a stop codon at position 185 and contains a SalI restriction site. The amplified fragment was digested with BamHI and SalI and cloned into the corresponding sites of the pFASTBAC donor plasmid. For the generation of the wild-type apoA-I donor plasmid (pFASTBAC-A-I), the 3Ј apoA-I Bsu36I-SalI region between codon 81 and the SalI polylinker site of the plasmid was obtained by restriction digestion. The region corresponding to amino acids ϩ1 to 82 was amplified using the 5Ј (AIC-5) and 3Ј (AIC-3) primers of Table I. The 5Ј and 3Ј primers contain restriction sites for BamHI and Bsu36I, respectively. The amplified fragment was digested with BamHI and Bsu36I and ligated along with the 3Ј Bsu36I-SalI fragment of apoA-I into the BamHI and SalI sites of the pFASTBAC TM HTb plasmid to generate the pFASTBAC-A-I donor plasmid. Cells containing recombinant bacmids were selected by kanamycin, tetracycline, and gentamycin resistance, as white colonies, due to the disruption of lacZ sequence in the recombinant bacmid. Recombinant bacmid DNA was isolated from minipreps and used to infect a monolayer of Sf-9 insect cells (19 -21). Recombinant viruses were isolated, amplified, titrated, and used to infect larger Sf-9 cell cultures grown in suspension at 27°C. Sf-9 cells were pelleted and resuspended in a lysis buffer. The supernatant was used for the purification of apoA-I fusion proteins using a   (17). c oligonucleotide position (ϩ) relative to the transcription initiation site (i.e. ϩ1 from ATG).
Ni 2ϩ -nitrilotriacetic acid resin affinity column (22,23). The pure apoA-I without the His tag, when needed, was obtained by cleavage with recombinant tobacco etched viral protease and purified by a second Ni 2ϩ -nitrilotriacetic acid resin affinity column.
Lipoprotein Isolation, Labeling, and Characterization-Blood was obtained from healthy fasting human donors. Spherical HDL (density range 1.09 -1.18 g/ml) was prepared from pooled plasma (two donors for each preparation) by zonal centrifugation as described previously (24) and was stored under nitrogen at 4°C. 125 I-Labeled HDL (296 -649 cpm/ng protein) and 125 I-labeled apoA-I (358 -470 cpm/ng protein) were prepared using the iodine monochloride method (25). Unless otherwise noted, the HDL fractions that were used for iodination and subsequently in the competition experiments were in the density range 1.1-1.14 g/ml. The protein concentrations of the HDL preparations, apolipoproteins, and cell lysates were determined by the method of Lowry et al. (26). The apolipoprotein composition of the preparations was assessed by Coomassie Brilliant Blue staining of gradient (6 -20%) SDS-polyacrylamide gels. Native and radiolabeled HDL preparations were periodically monitored, and preparations were discarded if evidence of abnormal electrophoretic mobility, presumably due to radiolysis/oxidation, was observed. The binding properties of such modified HDLs differ from those of native HDL (e.g. increased binding affinity for SR-BI; not shown).
Pre-␤-1 HDL Purification from Plasma-Pre-␤-1 HDL and ␣ HDL were purified from freshly drawn venous blood under nondenaturing conditions as described by Kunitake et al. (27) with the following modifications. Typically, blood pooled from two donors (360 ml) is drawn into a mixture containing 1 mM sodium EDTA, 0.02% NaN 3 , 10 g/ml ␣-2 macroglobulin, 0.13% ⑀-aminocaproic acid, 0.3 mg/ml benzamidine, 1 g/ml gentamycin sulfate and cooled immediately on ice, and plasma is separated by low speed centrifugation (1,800 ϫ g, 30 min). Plasma aliquots (20 ml) were subjected to anti-apoA-I (affinity-purified IgG) immunoaffinity column chromatography. The binding fraction consisting of apoA-I-containing lipoproteins (LpA-I) was eluted with 0.2 M acetic acid (pH 3.0) and immediately neutralized to pH 7.0 with 2 M Tris buffer. The fraction was further processed over Sepharose columns covalently coupled with affinity-purified anti-human serum albumin IgG and protein A. The pooled LpA-I material was reduced in volume by ultrafiltration (spiral ultrafiltration cartridge, type S3Y10 (Amicon), Ultrafree (5,000 molecular weight cut-off; Millipore Corp.) and resolved by preparative slab gel-agarose electrophoresis into five discrete fractions (␤, pre-␤-1, pre-␤-2, ␣-1, ␣). Pre-␤-1 HDL from the pre-␤-1 and pre-␤-2 zones was recovered by electroelution and subjected to molecular sieve and anion exchange chromatography. Molecular sieve chromatography consisted of two Superose 12 columns (Amersham Pharmacia Biotech) connected in series operating at 0.03 ml/min at 3°C in 10 mM Tris (pH 7.4), 0.15 M NaCl, 1 mM EDTA, 0.02% NaN 3 . Anion exchange chromatography consisted of two (1-ml) Hi Trap Q columns (Amersham Pharmacia Biotech), connected in series and operated at 0.5 ml/min at 3°C. Pre-␤ HDL eluted in 10 mM Tris (pH 8.5), 40 mM NaCl. Pre-␤-1 HDL was characterized for particle size distribution by electrophoresis through a 2-45% concave acrylamide gradient gel in a vertical gel apparatus (MiniProtean; Bio-Rad) modified for cooling at 10°C for 3,000 V-h. Calibrator proteins (high molecular weight; Pharmacia) were supplemented with human low density lipoprotein (d ϭ 1.030 -1.050 g/ml) and molecular sieve-purified ovalbumin to yield a dynamic Stokes radii range of 3.0 -12.5 nm. Analysis by SDS-polyacrylamide gel electrophoresis (2-25% linear acrylamide gradient) revealed the presence of a single apoA-I band free of proteolytic peptides. Preparations exhibiting evidence of apoA-I proteolysis were discarded. Typically, 2 mg of pre-␤-1 HDL were obtained with an overall recovery of 15%. Yields depend in part on the starting plasma levels, which vary among individual donors, as determined by quantitative agarose immunoblotting (28) and isotope dilution ultrafiltration assays (29).
Preparation of rHDLs of Apolipoprotein, POPC, and Cholesterol-Purified apolipoprotein A-I was prepared from plasma as described previously (30). Complexes comprising apolipoprotein, POPC, and cholesterol were prepared using the sodium cholate dialysis method (31) using an apoA-I/POPC/cholesterol molar ratio of 1:100:10, as previously reported (1). Apolipoprotein-lipid complex formation was verified by analysis with native polyacrylamide gradient (8 -25%) gel electrophoresis (Pharmacia Phast gel system; Amersham Pharmacia Biotech). The lipid-free apoA-I that was not incorporated in the lipid protein complexes was removed from the preparation either by dialysis using a dialysis membrane with a molecular weight cut-off of 50,000 or by gel filtration using an Amersham Pharmacia Biotech Superose 6HR column, 10/30 (total bed volume 24 ml).
Cell Cultures for Receptor Binding Assays-Control (ldlA-7) cells and a cell line expressing murine SR-BI receptor derived from ldlA-7 cells (ldlA[mSR-BI]) have been described previously (2,(32)(33)(34). The ldlA-7 cells were maintained in monolayer culture in Ham's F-12 medium containing 5% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM glutamine (medium A). The ldlA[mSR-BI] cells were maintained as stock cultures in medium A supplemented with 500 g/ml G418 (medium B). All incubations with cells were performed at 37°C in a humidified 5% CO 2 , 95% air incubator. Cell Association and Competition Assays Using 125 I-Labeled Ligand (Radioreceptor Assay)-SR-BI activity at 37°C was assessed by measuring cell association of radiolabeled ligands, as described previously (1,2,33,35,36). For direct association saturation curves, incubations were performed in the absence (total cell association, duplicate determinations) or presence (nonspecific cell association, single determinations) of a 40-fold protein mass excess of unlabeled HDL. For competition curves, duplicate incubations were performed in the absence or presence of the indicated unlabeled competitors. The cell association of 125 I-ligands was analyzed as a function of concentration by nonlinear regression (using the Prism program). The best fit was found for the model with high affinity binding to a specific site plus low affinity binding to other sites (background) as described below for the immunoreceptor assay. The specific, high affinity cell association activities presented in the saturation curves represent the differences between the average total cell association and nonspecific cell association values and are expressed as ng of protein of HDL or apolipoprotein complex associated per mg of total cell protein. The total cell association values presented in the saturation curves are expressed as ng of protein of HDL or apolipoprotein complex associated with the cells per mg of total cell protein. Cell association for the competition assays is presented as the average percentage of control values determined in the absence of inhibitors. Each of the binding/competition experiments shown is representative of the results obtained in at least two and often four or more independent experiments.
Immunoreceptor Assay-To measure the cell association of unlabeled ligands, we used the following assay. On day 0, ldlA-7 and ldlA[mSR-BI] cells were plated in six-well dishes at 3 ϫ 10 5 cells/well in medium A (ldlA-7) or medium B (ldlA[mSR-BI]). On day 2, the monolayers were washed twice with Ham's F-12 medium and then refed with 0.7-1 ml of medium C (Ham's F-12 medium containing 0.5% (w/v) fatty acid-free bovine serum albumin, 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM glutamine) with the indicated amounts of lipid-free apoA-I, apoA-I-POPC-cholesterol complexes, or spherical HDL. After a 1.5-h incubation at 37°C, the cells were washed twice at 4°C with 50 mM Tris-HCl, pH 7.4, 0.15 M NaCl containing 2 mg/ml fatty acid-free bovine serum albumin, followed by one rapid wash with buffer B alone. The cells were then solubilized in lysis buffer (PBS, 1% Triton X-100, 50 g/ml phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, 20 g/ml aprotinin, and 10 g/ml leupeptin) and incubated, with shaking, at 4°C for 30 min. The collected lysates (after scraping) were clarified by centrifugation at 14,000 rpm in a Beckman microcentrifuge for 20 min at 4°C. The protein concentrations of the cell lysates were determined by the method of Lowry et al. (26). An aliquot of 20 -30 g of the lysate was analyzed by 12% SDS-polyacrylamide gel electrophoresis, and in independent lanes the same ligand used in the binding assay, ranging from 1 to 25 ng, was mixed with the corresponding amount (20 -30 g) of SR-BI cell lysate from control cells not incubated with the ligand. These samples were used to generate the standard curves for quantitation (see below). After electrophoresis, the proteins were transferred to polyvinylidene difluoride membranes at 50 V in 500 mM CAPS-NaOH buffer (pH 11.5) for approximately 1 h. The membranes were blocked overnight at 4°C, or for 1.5 h at room temperature in PBS containing 0.1% Tween 20 and 5% nonfat dried milk. The membranes were washed three times for 5 min each with PBS 0.1% Tween 20 and 0.5% nonfat dried milk and were incubated for 1 h or overnight in the same solution using monoclonal human anti-apoA-I antibodies diluted 1:2000 (5F-6 antibody, the epitope is amino acids 118 -141 of the human apoA-I protein). The membranes were then washed three times for 15 min/ wash with PBS, 0.1% Tween 20 and incubated for 1 h in the same solution containing anti-mouse IgG conjugated to horseradish peroxidase diluted 1:5000. Finally, the membranes were washed three times as above in PBS, 0.1% Tween 20. 3Ј,3Ј-Diaminobenzidine or enhanced chemiluminescence (ECL) reagent was used as substrate of the horseradish peroxidase, and the protein bands on the filters were detected and captured by exposing the blots to Kodak film (Kodak Digital Science Image Station 440CF), according to the manufacturer's instructions. The quantitation of the intensities of the bands was performed using the ImageQuant software. A rectangular box was drawn containing the band of interest, and the net intensity of the band was calcu-lated by subtracting the intensity of an equivalent band-free adjacent area in each lane (background value) from the total intensity of the band-containing box. A standard curve was used to convert the intensity values to ng of ligand protein. The conversion was performed by plotting the standard amount of ligand versus the corresponding intensity. Because the standard curves were reproducibly nonlinear, a polynomial second order regression was used as the best fit for the standard curve. The total binding is expressed as ng of cell-associated HDL, lipid-free apoA-I, or apolipoprotein complex protein per mg of total cell protein.
The cell-associated ligands were analyzed as a function of concentration by nonlinear regression (using the Prism program). The best fit to equations for three models was tested. These are 1) high affinity and saturable binding to a single class of sites on SR-BI; 2) high affinity and saturable binding to two nonequivalent sites; or 3) high affinity and saturable binding to one site plus low affinity and nonsaturable binding to other sites. For the immunoreceptor assays, the best fit was found for the model with high affinity binding to a specific site plus low affinity binding to other sites (background). The equation is as follows, where "total" represents specific plus nonspecific, B total is the measured amount of ligand bound, B max is the amount of ligand bound at saturating concentrations, K d is the apparent high affinity constant, and NS is the slope of the low affinity nonsaturable process (background). The nonspecific binding values for ldlA[mSR-BI] cells were similar to the total binding values for control ldlA-7 cells.

Development of a New
Binding Assay for SR-BI-We have shown previously that apoA-I is one of the ligands of SR-BI (1). In the current study, we compared the binding of different apoA-I-containing ligands of SR-BI, including HDLs of different densities, discoidal rHDLs containing full-length and truncated apoA-I forms, pre-␤-1 HDL, and lipid-free apoA-I.
The standard binding assay for lipoprotein receptors, such as SR-BI, involves measurement of the association of lipoproteins radioiodinated on their apolipoproteins with cells (radioreceptor assay). Since iodination directly or indirectly (for instance as a consequence of oxidation) might alter the properties of the ligand, we have developed an assay in which unlabeled ligand binding to the receptor is detected by quantitative immunoblotting (immunoreceptor assay). A typical saturation curve for the binding of 125 I-HDL (d ϭ 1.1-1.13 g/ml) to ldlA[mSR-BI] cells using a radioreceptor assay is shown in Fig. 1A. The total binding values were experimentally determined, whereas the specific and the nonspecific values were obtained by nonlinear regression analysis, as described under "Experimental Procedures." HDL bound to a single class of sites on ldlA[mSR-BI] cells with high affinity, apparent K d of 16.3 g of protein/ml and B max of 450 ng/mg cell protein. B max varied from experiment to experiment, presumably due to variation in the expression of the SR-BI transgene in the cell line. The apparent K d value was highly reproducible from experiment to experiment when freshly iodinated HDL was used. However, the K d did vary depending on the age of the 125 I-HDL preparation, pre- in medium B (ldlA-7) at a density of 3 ϫ 10 5 cells/well. On day 2 after plating, cells were incubated with the indicated concentrations of unlabeled spherical HDL (density 1.13 g/ml) for 1.5 h, at 37°C. Cell lysates containing receptor-ligand complexes were prepared, separated by 12% SDS-polyacrylamide gel electrophoresis, and blotted into polyvinylidene difluoride filters as described under "Experimental Procedures." The receptor-associated ligand was detected by immunostaining using an enhanced chemiluminescence reagent and was quantitated by photoimaging. The total binding values (Ⅺ) were calculated from the observed intensities using a standard curve (C) as described under "Experimental Procedures." The specific (q) and the nonspecific (E) binding values and the apparent K d value of 16.9 g of protein/ml were calculated by nonlinear regression analysis. C, a standard curve was obtained by fitting a second order polynomial to the immunostaining intensity values from known amounts of the ligand.
sumably because of radiolytic/oxidative damage. 2 For example, the same preparation used in Fig. 1A 3 days after iodination was retested 15 and 55 days postiodination, and the corresponding apparent K d values were 10.4 Ϯ 2 and 4.9 Ϯ 1 g of protein/ml, respectively. We typically observed increased binding affinity as the preparations aged. Because of this iodinationdependent variability and the practical limitations of labeling ligands available in low quantities, we developed an alternative assay, the immunoreceptor assay (see "Experimental Procedures"). Fig. 1B shows the binding of unlabeled, spherical HDL (d ϭ 1.13 g/ml) to SR-BI as determined by the immunoreceptor assay. The observed total binding values were determined from the intensity of the bound apoA-I signals in quantitative immunoblots using the standard curve in Fig. 1C. The specific binding and the nonspecific binding values were calculated by nonlinear regression, as described under "Experimental Procedures." The calculated nonspecific binding values for ldlA[mSR-BI] cells were similar to the observed total binding values for untransfected ldlA-7 cells (not shown), which express very little endogenous HDL receptor activity (2). The apparent K d (16.9 g of protein/ml) and B max (370 ng of protein/mg of cell protein) values determined by the immunoreceptor assay were similar to those obtained using the radioreceptor assay and freshly iodinated HDL. Thus, these two assays give similar results, and the new immunoreceptor assay can be used as a simple and powerful alternative to evaluate receptor-binding parameters of a variety of ligands without the need for iodination.
High Affinity Binding of Discoidal rHDLs Containing ApoA-I-In order to characterize further the binding of apoA-I ligands to SR-BI, we tested the role of lipid association on ligand binding affinity. Plasma apoA-I (apoA-Ip) was purified and used for the preparation of rHDL with POPC and unesterified cholesterol, as described under "Experimental Procedures." Most of the discoidal HDL particles in the preparations (80 -90%) had diameters of 96 -104 Å (not shown). There were less abundant, larger forms with approximate diameters of 120 Å. Fig. 2 shows that 125 I-rHDL{apoA-Ip} exhibited high affinity, saturable binding to mSR-BI-expressing cells. The apparent K d value was 1.5 Ϯ 0.2 g of protein/ml. The data best fit a model for binding of apoA-I to a single class of sites. Similar results were obtained with the immunoreceptor assay (data not shown). The rHDL{apoA-Ip} (predominant population with diameter of 96 Å) is expected to contain approximately two apoA-I molecules per particle (37). The apparent binding affinity of rHDL{apoA-Ip} was at least 1 order of magnitude higher than that of spherical plasma HDL (d ϭ 1.13 g/ml) (16 versus 1.5 g of protein/ml). Control experiments showed that POPCcholesterol (2:1) liposomes without apolipoprotein do not bind to SR-BI (data not shown; also see Ref. 1), indicating that the high affinity binding of rHDL to SR-BI was apoA-I-dependent, presumably due to direct association of apoA-I with SR-BI.
Comparison of the Binding of High and Low Density Spherical HDL, ␣ HDL, Pre-␤-1 HDL, and Lipid-free ApoA-I to SR-BI-The observation that the affinities of spherical HDL and discoidal rHDL{apoA-Ip} for SR-BI were different prompted us to analyze the binding of different naturally occurring HDL species, including spherical HDL particles of different densities. This analysis involved both direct association and competition assays. Fig. 3A shows the direct association saturation curves for two 125 I-HDL fractions of different densities. The lower density HDL fraction (filled circles) was 1.11-1.13 g/ml, and the higher density HDL fraction (open circles) was 1.14 -

FIG. 3. Effects of the density of spherical HDL on SR-BI binding.
A, on day 2 after plating, ldlA[mSR-BI] cells were incubated with the indicated concentrations of 125 I-labeled HDL fractions of lower (1.11-1.13 g/ml) (q) or higher (1.14 -1.17 g/ml) (E) densities (isolated by zonal ultracentrifugation). Nonlinear regression analysis of the observed total cell association (see Fig. 1 and "Experimental Procedures") was used to calculate the specific cell association values shown. B, on day 2 after plating, ldlA[mSR-BI] cells were incubated with 10 g of protein/ml of 125 I-HDL (d ϭ 1.1-1.2 g/ml) and either 25 g (hatched bars) or 75 g (stippled bars) of protein/ml unlabeled spherical HDL preparations with the indicated densities (g/ml). The values presented are the averages of duplicate determinations. The 100% of control value of cell association, measured in the absence of unlabeled HDL, was 162 ng of 125 I-HDL protein/mg of cell protein.
1.17 g/ml. These preparations were assayed 15 days after iodination. The binding constants were determined by nonlinear regression, and the apparent K d values were 11 Ϯ 2 g of protein/ml for 125 I-HDL of density 1.11-1.13 g/ml and 37 Ϯ 5 g of protein/ml for 125 I-HDL of density 1.14 -1.17 g/ml. Fig. 3B shows the inhibition of binding of 125 I-HDL (d ϭ 1.1-1.12 g/ml) to ldlA[mSR-BI] cells by unlabeled HDL fractions of different densities (1.095-1.149 g/ml). The ability of the HDL fractions to compete for the binding of this low density, high affinity 125 I-HDL decreased as the density of the HDL particles increased. We have also observed that HDL of ␣ electrophoretic mobility, isolated by immunoaffinity chromatography (27), and the spherical HDL (d ϭ 1.12 g/ml), isolated by zonal ultracentrifugation, did not show significant differences in their abilities to compete for the binding of 125 I-HDL to SR-BI-expressing cells (data not shown).
Pre-␤-1 HDL particles are thought to have novel physiological functions that differ from those of spherical HDL (15,27). In order to compare the interaction with SR-BI of pre-␤-1 HDL, discoidal rHDL{apoA-Ip}, and spherical HDL particles (d ϭ 1.14 g/ml), we examined their abilities to compete for the binding of spherical 125 I-HDL (d ϭ 1.13 g/ml) to SR-BI. Fig. 4A shows that pre-␤-1 HDL was a poor competitor, compared with the spherical or discoidal rHDL{apoA-Ip} particles. Similar results were observed for the inhibition of the transfer of the lipid dye DiI from DiI-HDL to ldlA[mSR-BI] cells by unlabeled spherical HDL and pre-␤-1 HDL (data not shown). Preliminary direct binding assays, using the immunoreceptor assay did not show SR-BI-dependent binding of pre-␤-1 HDL (data not shown). We also examined the ability of lipid-free apoA-Ip, isolated from plasma, to compete for the binding of spherical 125 I-HDL (d ϭ 1.14 g/ml) to SR-BI. As previously observed (1), lipid-free apoA-Ip was a poor competitor compared with spherical HDL (d ϭ 1.14 g/ml) or discoidal rHDL{apoA-Ip} particles (Fig. 4B). Numerous direct binding experiments, using either the radioreceptor or the immunoreceptor assay, could not establish an SR-BI-dependent binding of lipid-free apoA-I (data not shown).
Effect of ApoA-I Mutations on the Binding of rHDL Particles to SR-BI-The common feature of all the experiments presented above and in previous studies (1) is that apoA-I is a ligand for SR-BI. This led us to ask if specific domains of the apoA-I molecule were involved in the receptor recognition. To test if the carboxyl-terminal or amino-terminal domains of apoA-I were essential for binding to SR-BI, we designed the following mutant apoA-I forms: a carboxyl-terminal truncated mutant, apoA-I:⌬(185-243); an amino-terminal truncated mutant, apoA-I:⌬(1-59); and a double truncated mutant, lacking both the amino-and the carboxyl-terminal domains, apoA-I: ⌬(1-59)(185-243). The apoA-I:⌬(185-243) form was generated by expression of the protein in mammalian C127 cells and contains a 6-residue prosegment, which is not present in plasma apoA-I (38). The amino-terminal truncation, apoA-I B : ⌬(1-59) and the double truncation, apoA-I B :⌬(1-59)(185-243) were generated by expression of apoA-I using the baculovirus system (the B subscript indicates protein obtained by baculovirus expression) (19,20). In the baculovirus expression system, the mature apoA-I B (1-243) synthesized contains a histidine tag at its amino terminus that can be cleaved with the etched virus protease. Discoidal rHDL particles were prepared containing apoA-I recombinant forms and analyzed by native polyacrylamide gel electrophoresis (Table II). The predominant size of rHDL particles containing the wild-type apoA-I, proapoA-I, the amino-terminal deleted, and the double truncated forms was 96 -104 Å. The predominant size of rHDL particles containing the carboxyl-terminal truncation was 77-79 Å. These particles are expected to contain two apoA-I molecules per rHDL particle (37). These rHDL particles were tested for binding to SR-BI by both competition for 125 I-HDL binding and direct association assays. Fig. 5A shows the inhibition of binding of spherical 125 I-HDL (d ϭ 1.13 g/ml) by rHDL particles containing proapoA-I:⌬(185-243) or apoA-I B :⌬(1-59), compared with rHDL{apoA-Ip} and native spherical HDL (d ϭ 1.14 g/ml, dashed line). Both truncated forms, when reconstituted in rHDLs, competed very effectively for the binding of native spherical 125 I-HDL to SR-BIexpressing cells. rHDL particles containing the full-length proapoA-I recombinant protein competed as well as particles containing the plasma apoA-Ip (data not shown), indicating that the presence of the prosegment did not substantially affect   the binding of apoA-I to SR-BI. Similar competition experiments also showed that the presence of the histidine tag in full-length apoA-I in rHDL particles, relative to rHDL particles containing apoA-I in which the tag was removed by proteolysis, did not affect the efficiency of competition for the binding of native spherical 125 I-HDL to SR-BI (data not shown). Furthermore, the direct binding of rHDL particles containing baculovirus expressed wild-type apoA-I gave similar apparent K d values in the presence or absence of the His tag on the apoA-I molecule (4 Ϯ 0.5 g of protein/ml or 135 Ϯ 14 nM equivalent of apoA-I for no His-containing wild-type apoA-I versus 5 Ϯ 1 g of protein/ml or 168 Ϯ 27 nM equivalent of apoA-I for His-tag containing wild-type apoA-I; data not shown).
The efficient competition of rHDL particles containing the amino-or carboxyl-terminal apoA-I truncated forms was confirmed in direct association studies, presented in Fig. 5, B and D. Discoidal rHDLs containing proapoA-I:⌬(185-243) or the apoA-I B :⌬(1-59) exhibited high affinity saturable binding to a single class of sites on ldlA[mSR-BI] cells, while there was almost no binding to the control untransfected ldlA-7 cells. The respective apparent K d values were 3 Ϯ 0.5 g of protein/ml and 7 Ϯ 2 g of protein/ml. These findings suggest that deletion of either the amino-terminal or the carboxyl-terminal domains of apoA-I alone did not affect substantially the ability of the protein incorporated into particles to bind to SR-BI. In contrast, rHDL particles containing the double truncated mutant apoA-I B :⌬(1-59)(185-243) competed very poorly for the binding of spherical 125 I-HDL (Fig. 5C), compared with rHDL{apoA-Ip} particles and spherical HDL. The double truncations of apoA-I may alter the size, shape, and/or composition of the rHDL particles and thus affect the affinity of binding to SR-BI. It seems likely that the conformation of apoA-I may be important for the recognition and binding of different ligands to SR-BI. It appears that optimal binding of discoidal rHDL to SR-BI can be achieved by the central domain of apoA-I (residues 60 -184) combined with either the amino or carboxyl-terminal domain. Simultaneous truncation of both these domains might cause conformational changes in rHDL that result in reduced binding affinity. The present study confirms the important role of apoA-I in HDL binding to SR-BI and indicates that the NH 2 and COOH termini of apoA-I independently influence binding to SR-BI, either directly or through changes in the conformation of the core of apoA-I (residues 60 -184).

Affinities of Lipid-associated Forms of ApoA-I for SR-BI-To
assess the binding to SR-BI of apo-A-I-containing ligands without iodinating the apoA-I, we developed an immunoreceptor binding assay. In this assay, the unlabeled receptor-associated ligand is detected by quantitative immunoblotting. This assay is not influenced by potential alterations of the properties of the ligand as a result of radioiodination. It also provides an easier, effective way to assess the binding of ligands that are only available in small quantities and for which iodination presents practical difficulties (e.g. recombinant truncated apoA-I forms). Using either the standard radioreceptor or the new immunoreceptor assay, we have been able to show that spherical HDL (d ϭ 1.1-1.13 g/ml) binds to a single class of sites on SR-BI with an apparent K d of approximately 16 g of protein/ml. Discoidal rHDL particles containing plasma apoA-I bind approximately 1 order of magnitude more tightly than spherical HDL (d ϭ 1.1-1.13 g/ml). The conformation of apoA-I varies when it is lipid-free or incorporated on spherical HDL or discoidal rHDL particles (15,37). Thus, the differences we observed in the apparent affinities of the spherical HDL or the rHDL particles may be due to differences in the conformation of apoA-I on the surfaces of these particles. These differences may alter the way the ligand is presented to the receptor. The discoidal rHDLs used were formed by the association of apoA-I with phospholipids and cholesterol at a molar ratio of 1 apoA-I:100 POPC:10 cholesterol. It has been previously reported that the rHDL discs prepared with specific phospholipids, such as POPC or DPPC, and apoA-I have discrete and reproducible sizes, apoA-I compositions, and apoA-I conformations. For example, the POPC-A-I particles have been described that consist of discs with diameters of 96, 86, and 77 Å and contain two apoA-I molecules per particle (37). It has been suggested that in the 96-, 86-, and 77-Å particles, eight, seven and six apoA-I helices, respectively, make contact with lipids. rHDL particles with three and four apoA-I molecules have also been reported (37, 39 -41). Therefore, it is possible that the different number of apoA-I molecules per particle and/or the different conformation of apoA-I within a specific particle could affect the affinity of apoA-I binding to SR-BI. Two models have been proposed for the arrangement of apoA-I helices on these particles. The classic picket fence model suggests that the helices are arranged parallel to the fatty acid chains of the phospholipid (42,43). The belt model for discoidal HDL suggests that A/B dimer containing two antiparallel apoA-I chains wraps around the disc (44 -46). In the belt model, the carboxyl-terminal apoA-I helix appears to be important for dimer formation (44,46). Our results suggest that carboxyl terminus-dependent structural features do not appear to play an important role in discoidal rHDL binding. In both models, polyvalent binding of multiple apoA-I molecules could facilitate ligand-receptor interactions by cross-linking adjacent SR-BI molecules or binding to multiple independent sites on a single receptor, thus increasing the apparent binding affinity.
Particle Density, Lipid, and Apoprotein Composition Affect Ligand Binding Affinity-It has been suggested that newly synthesized lipid-free apoA-I, or apoA-I that dissociates from HDL or other lipoproteins, recruits phospholipids from cell membranes to form discoidal pre-␤-1 HDL (15). These particles can accept cholesterol from cell membranes and are converted into larger discoidal pre-␤-2 HDL (47,48). Lecithin:cholesterol acyltransferase-mediated cholesterol esterification converts the discoidal pre-␤-2 particles into spherical ␣-migrating HDL particles (49,50). HDL 3 may then receive lipids from other lipoproteins or from cells to be converted into HDL 2 . Plasma HDL may also consist of particles containing apoA-I (LpA-I) or apoA-I and apoA-II (Lp(AI:AII)) (51-55). The majority of Lp(AI) floats at the density of HDL 2 , whereas the majority of Lp(AI: AII) floats at the density of HDL 3 . It has been shown that HDL 2 and HDL 3 contain four and three apoA-I molecules per particle, respectively (56).
The current study suggests that in addition to the shape of the HDL particles (discoidal versus spherical) the lipid and apolipoprotein composition of the HDL particles may affect their affinity for SR-BI. The first experiment showed a reduction in the affinity of the HDL isolated by zonal ultracentrifugation as its density increased. The gradual reduction in affinity may result from differences in the sizes or the compositions of the particles (e.g. number of apoA-I molecules/particle) or both, which may alter the manner in which apoA-I is presented to the receptor. The second experiment showed strikingly diminished ability of both pre-␤-1 HDL isolated from plasma and lipid-free apoA-I, relative to spherical HDL, to compete with native 125 I-HDL binding to SR-BI. We previously reported that lipid-free apoA-I could compete partially with 125 I-HDL for binding to SR-BI (ϳ36% inhibition of the binding of 10 g of protein/ml 125 I-HDL by 50 g of protein/ml lipid-free apoA-I (1)). While we confirmed this finding of partial inhibition in the current study, the more extensive analysis reported here, using both radioreceptor and immunoreceptor assays, clearly shows that lipid-free apoA-I is a much less effective competitor than native HDL or rHDL particles. The findings for lipid-free apoA-I and pre-␤-1 HDL, which comprises apoA-I combined with only a small amount of phospholipid (15), both indicate an important role for the interaction of apoA-I with lipids in controlling the interaction of apoA-I with SR-BI. Previous studies established that the lipid composition of pre-␤-1 HDL species as well as the conformation of apoA-I in the pre-␤-1 particles differ from those of spherical HDL (15,27). These differences presumably contribute to the differences in interactions with SR-BI. Based on these data and many previous studies (4,7,9,10,(57)(58)(59), we suggest the following. SR-BI binds most tightly to large, relatively low density, cholesteryl ester-rich HDL particles to maximize the efficiency of cholesterol transport via selective uptake. As a consequence of such selective uptake, the particles become smaller. Indeed, in the absence of SR-BI in vivo (9), HDL particles are abnormally large. Larger HDL particles have been shown to have longer in vivo plasma residence, as determined by the fractional catabolic rates, than smaller HDL particles (60). It seems likely that SR-BI mediates both the transfer of cholesteryl ester from lipid-rich HDL to target cells (liver, steroidogenic tissue) and participates with proteins such as lipases and cholesteryl ester transfer protein (59,(61)(62)(63)(64)(65)(66)(67) in the remodeling of HDL to generate smaller particles. The smaller particles, which bind less tightly to SR-BI, either serve as substrates for regenerating larger cholesteryl ester-rich HDLs, presumably by the action of ABC1 (68 -71) and lecithin:cholesterol acyltransferase, or are catabolized after filtration by the kidney, presumably via cubilinmediated endocytosis (72,73).
Effect of ApoA-I Mutations on the Binding of rHDL Particles to SR-BI-Recent studies have explored the roles of the amino and carboxyl termini of apoA-I on the protein's structure and function. For example, deletion of the amino-terminal residues 1-43 reduces the stability of apoA-I in the lipidfree state (44,74). In addition, the carboxyl terminus (residues 185-243) has been shown to be important for binding to phospholipids and lipoproteins (18,75) and may have other functions (76). Furthermore, the central core region, 68 -185, which contains six amphipathic ␣-helices, is important for other apoA-I functions. In particular, helices 6 and 7 are important for the activation of lecithin:cholesterol acyltransferase (18,(77)(78)(79). It has been shown that deletions of the amino (residues 1-43) or the carboxyl-terminal residues (residues 185-243) of apoA-I did not alter significantly the helical content of lipid-free apoA-I (80).
In the current studies, we examined the association with SR-BI of rHDL particles containing mutant apoA-I forms with either single deletions of the 59 amino-terminal residues or the 58 carboxyl-terminal residues or the double deletion mutant in which both sets of termini were removed. The affinities of rHDL particles containing the single terminal deletion mutant forms were similar to those of rHDL particles comprising the full-length apoA-I. The small changes in the apparent K d values of the rHDLs with the single truncations may reflect differences in the particle sizes of the different apoA-I forms. Simultaneous deletion of both terminal domains was associated with substantially reduced capacity of the corresponding rHDL particles to compete for the binding of spherical 125 I-HDL. These data raise the possibility that both the amino or the carboxyl-terminal domains can bind independently to SR-BI. In support of this interpretation, a recent independent report, which appeared during the preparation of this manuscript, showed that discoidal rHDL particles containing peptides representing the amino-terminal (residues 1-85) or the carboxyl-terminal region (residues 149 -243) of apoA-I bind with high affinity to SR-BI (81). An alternative interpretation could be that both terminal domains influence a feature(s) of the structure of the core of apoA-I that is critical for binding to SR-BI. Thus, it is possible that the central helices (residues 60 -184) in the proper conformation contribute independently to receptor binding. The diminished affinity of the double mutant may then have resulted from an altered conformation of the central helices because of these mutations. The current data do not allow us to identify precisely where on apoA-I the binding site(s) for SR-BI is located.
Additional in vitro and in vivo studies will be required to determine if alterations in the conformation, sequence, or structure of apoA-I significantly influence selective uptake of cholesterol mediated by SR-BI.