Expression Cloning and Characterization of a Novel Glycosylphosphatidylinositol-anchored High Density Lipoprotein-binding Protein, GPI-HBP1*

By expression cloning using fluorescent-labeled high density lipoprotein (HDL), we isolated two clones that conferred the cell surface binding of HDL. Nucleotide sequence of the two clones revealed that one corresponds to scavenger receptor class B, type 1 (SRBI) and the other encoded a novel protein with 228 amino acids. The primary structure of the newly identified HDL-binding protein resembles GPI-anchored proteins consisting of an N-terminal signal sequence, an acidic region with a cluster of aspartate and glutamate residues, an Ly-6 motif highly conserved among the lymphocyte antigen family, and a C-terminal hydrophobic region. This newly identified HDL-binding protein designated GPI-anchored HDL-binding protein 1 (GPI-HBP1), was susceptible to phosphatidylinositol-specific phospholipase C treatment and binds HDL with high affinity (calculated K d = 2–3 μg/ml). Similar to SRBI, GPI-HBP1 mediates selective lipid uptake but not the protein component of HDL. Among various ligands for SRBI, HDL was most preferentially bound to GPI-HBP1. In contrast to SRBI, GPI-HBP1 lacked HDL-dependent cholesterol efflux. The GPI-HBP1 transcripts were detected with the highest levels in heart and, to a much lesser extent, in lung and liver. In situhybridization revealed the accumulation of GPI-HBP1 transcripts in cardiac muscle cells, hepatic Kupffer cells and sinusoidal endothelium, and bronchial epithelium and alveolar macrophages in the lung.

High density lipoprotein (HDL) 1 plays a key role in the transportation of cholesterol to extrahepatic tissues including steroidogenic tissues and in the reverse transportation of cholesterol from extrahepatic tissues to the liver (1). Unlike the low density lipoprotein (LDL) receptor pathway, the delivery of cholesterol from HDL to cells is mediated by selective lipid uptake from HDL particles and is independent of internalization of HDL. Reverse cholesterol transportation requires the extraction of cholesterol from extrahepatic cells by HDL and the subsequent delivery of cholesterol to hepatocytes.
Several HDL-binding proteins have been identified including class B type I scavenger receptor (SRBI) (2, 3), two candidate hepatic HDL receptors designated HDL-binding proteins 1 and 2 (4,5), 80-and 130-kDa GPI-anchored HDL-binding proteins expressed in human macrophages (6), 110-kDa GPI-anchored HDL-binding protein expressed in HepG2 cells (7), and recently characterized 95-kDa HDL-binding protein (8). To date, only SRBI appears to be a physiological HDL receptor based on the selective uptake of cholesterol esters into cells and the efflux of cholesterol from cells to HDL mediated by SRBI (1). Consistent with the postulated physiological role, SRBI is highly expressed in tissues that selectively take up cholesterol esters from HDL including liver, adrenal gland, testis, and ovary (3). Although hepatic overexpression of SRBI mediated by an adenovirus encoding SRBI resulted in a dramatic reduction of plasma cholesterol (9), the targeted disruption of the murine SRBI gene led to a modest increase in plasma cholesterol (10). This finding may suggest the presence of other HDL receptors that cooperate with SRBI in the metabolism of HDL.
In this study, we identified a novel HDL-binding protein by expression cloning from a murine hepatic cDNA library. This newly identified HDL-binding protein designated GPI-anchored HDL-binding protein 1 (GPI-HBP1) belongs to the GPIanchored lymphocyte differentiation antigen Ly-6 family, binds HDL with high affinity on the cell surface, and mediates selective lipid uptake from HDL particles. We also describe the ligand specificity and tissue expression of GPI-HBP1. * This work was supported in part by Grant RFTF97L00803 from the Japan Society for the Promotion of Science. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB095543.
Standard Procedures-Standard molecular biology techniques were performed essentially as described by Sambrook and Russell (15). cDNA clones were subcloned into pBluescript vectors and sequenced by the dideoxy chain termination method with a BigDye Terminator Cycle Sequencing Ready Reaction kit (PE Biosystems) and a DNA sequencer (model 310, PE Biosystems). To analyze RNA in murine tissues, commercially available Northern blots (Clontech laboratories) were used for Northern blot analysis. 32 P-Labeled probes were prepared by priming with random hexanucleotides.
Expression Cloning-A cDNA library was constructed from poly(A) RNA isolated from the livers of LDL receptor-deficient male mice (16) in the pZeoSV2 vector (Invitrogen) using a NotI unidirectional primer. The cDNA library consisted with ϳ3 ϫ 10 5 clones, and these clones were divided into small pools (300 clones/pool). Plasmid DNA from each pool was prepared using the QIAprep 96 Turbo Miniprep kit (Qiagen). On day 0, LDL receptor-lacking Chinese ovary cells, ldlA7 (17), were plated into 96-well plates (5 ϫ 10 4 /well) in minimum essential medium ␣ supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin (medium A). On day 1, the cells in each plate were transfected with 0.1 g of a cDNA pool using Effectene transfection reagent (Qiagen) according to the manufacturer's protocol. After incubation for 16 h, the transfection mixture was replaced with medium A. On day 3, the monolayers were refed with minimum essential medium ␣ containing 2 g/ml protein of DiI-HDL and 5% lipoprotein-deficient serum. After a 2-h incubation at 37°C, the plates were washed twice with PBS containing 10 mg/ml BSA and then twice with PBS, and the cells were fixed with 3% formaldehyde in PBS for 15 min at room temperature. The presence of DiI in the fixed cell was detected by visual inspection under fluorescent microscopy. Positive pools were serially subdivided and retested to obtain positive cDNA clones.
Cell Culture and Transfection-The entire coding regions of murine GPI-HBP1 and SRBI were subcloned into the pRC/CMV vector (Invitrogen) for stable transfection. CHO ldlA7 cells were transfected with the expression plasmid using the LipofectAMINE reagent (Invitrogen) according to the manufacturer's instructions. Stably transfected cells were selected in Ham's F-12 medium containing 50 units/ml penicillin, 50 g/ml streptomycin, and 2 mM glutamine (medium B) supplemented with 5% fetal bovine serum and 250 g/ml G418 for 2 weeks. For 125 I-HDL and DiI-HDL binding studies and [ 3 H]cholesterol efflux assays (see below), cells were plated in 6-well (250,000 cells/well) dishes in medium B supplemented with 5% newborn calf lipoprotein-deficient serum and cultured for 48 h.
Phosphatidylinositol-specific Phospholipase C (PIPLC) Treatment-GPI-HBP1-expressing cells were incubated for 1 h in medium B supplemented with 5% newborn calf lipoprotein-deficient serum with or The amino acid sequence of murine GPI-HBP1 deduced from the cDNA is compared with that of murine sHBP1. Identical amino acids are indicated by asterisks. The N-and C-terminal hydrophobic amino acids were boxed. A cluster of negatively charged amino acids is underlined. The Ly-6 motif is indicated by a dotted underline.

FIG. 3. Effect of PIPLC treatment on DiI-HDL uptake.
A, murine GPI-HBP1 was transiently expressed in ldlA7 cells as described under "Experimental Procedures." After a 1-h incubation with or without PIPLC at 37°C, cells were washed and incubated for the measurement of DiI-HDL uptake. B, DiI-HDL uptake. The amount of cellassociated DiI was determined as described under ''Experimental Procedures.'' Values are the average of triplicate determinations, and error bars represent the range of the three measurements. without 1 unit/ml PIPLC. Cells were then washed and incubated for the measurement of DiI-HDL uptake. 125 I-HDL-binding and Association Assays-Cells were washed once with medium B and then refed with medium B containing 0.5% (w/v) fatty acid-free BSA and the indicated concentrations of 125 I-HDL. After a 2-h incubation at 37°C, cells were washed once with 50 mM Tris-HCl, pH 7.4, and 0.15 M NaCl (buffer A) containing 2 mg/ml BSA followed by two quick washes with buffer A without BSA. Cells were then solubilized with 0.1 N NaOH for 30 min at room temperature on a shaker, and we determined the amounts of cell-associated radioactivity using a ␥-counter. The protein content was determined using the method of Lowry et al. (18). For 4°C binding studies, the protocol was identical to that at 37°C with the exception that the cells were prechilled on ice for 15 min and incubated with 125 I-HDL at 4°C for 2 h. Specific cell association or binding was determined by subtracting the values obtained with parental vector (pRC/CMV) transfected cells from those obtained with a given expression plasmid.
Fluorimetric Assay of DiI-HDL Uptake-DiI-HDL was used to measure the cellular association and uptake of fluorescence by cells stably expressing GPI-HBP1 or SRBI according to the procedure as described by Acton et al. (3). Cells were washed once with medium B and then refed with medium B containing 0.5% (w/v) fatty acid-free BSA and the indicated concentrations of DiI-HDL. After incubation at 37°C for 2 h, cells were washed twice with PBS containing 5 mM CaCl 2 and 5 mM MgCl 2 (5 min/wash). Cell-associated DiI was then solubilized in 0.5 ml of Me 2 SO at room temperature for 2 h, and the fluorescence was measured using a spectrofluorimeter. The amount of DiI in each sample expressed as equivalent amounts of DiI-HDL (micrograms of protein) was calculated by comparing the fluorescence intensity of the sample to that from a standard curve generated by dissolving DiI-HDL in Me 2 SO.
Specific DiI uptake was determined by subtracting the values obtained with parental vector transfected cells from those obtained with a given expression plasmid.
[ 3 H]Cholesterol Efflux from Stably Transfected CHO ldlA7 Cells-HDL-dependent cholesterol efflux study was performed according to the procedure described by Gu et al. (19). Cells (ϳ70% confluent) were incubated for 48 h with 0.2 Ci/ml [1,2-3 H]cholesterol (40 -60 Ci/mmol, Amersham Biosciences). After washing five times with PBS containing 1% fatty acid-free BSA, radiolabeled cells were incubated overnight in Ham's F12 containing 1% fatty acid-free BSA to allow for the equilibration of cellular cholesterol pools. Cells were then washed and incubated for the indicated times in efflux medium (Ham's F12, 0.5% fatty acid-free BSA) with or without 40 g/ml HDL. The efflux medium was collected and clarified by centrifugation for 1 min with a Microcentrifuge, and the radioactivity of each supernatant was determined by liquid scintillation counting. Cells were solubilized with 1% Triton X-100 in PBS for 30 min at room temperature, and the amount of In Situ Hybridization-Digoxigenin-11-UTP-labeled single-stranded RNA probes were prepared with a digoxin/RNA-labeling mixture and the corresponding T3 or T7 RNA polymerase (Roche Molecular Biochemicals) according to the manufacturer's instructions. The entire coding region of murine GPI-HBP1 cDNA was subcloned into the pBluescript II vector (Stratagene) and used to prepare singlestranded RNA probes. Tissues from a C57BL/6J male mouse (10weeks old) were fixed in PBS containing 4% paraformaldehyde at 4°C for 12 h, dehydrated, and embedded in paraffin using a standard procedure. In situ hybridization was performed using 6-m thick sections mounted on silane-coated glass slides. After digestion with 10 mg/ml proteinase K at 37°C for 20 min, the tissue sections were hybridized with antisense or sense RNA probes at 50°C for 16 h. For the reaction with antidigoxigenin antibodies, slides were washed with 100 mM Tris-HCl, pH 7.5 and 150 mM NaCl (buffer B), treated with 0.5% blocking reagent (Roche Molecular Biochemicals) in buffer B, and then incubated with alkaline phosphatase-coupled antidigoxigenin antibodies (diluted 1:750 in buffer B, Roche Molecular Biochemicals) for 1 h.

RESULTS AND DISCUSSION
Expression Cloning of HDL-binding Proteins-We screened a cDNA library from murine liver for cDNAs that facilitate the binding of HDL when transiently expressed in LDL receptorlacking ldlA7 cells. After screening a total of 960 pools of ϳ300 cDNA each, we obtained two pools of cDNAs that stimulated HDL binding to levels that were significantly higher than background. We then plated a total of 960 colonies from two positive pools into individual wells of 96-well plates. We prepared cDNAs from the pooled rows and columns of the plates. Four of these pools gave positive results. We then assayed individual clones from the wells at the intersections of the positive rows and columns and obtained two positive clones (Fig. 1). Nucleotide sequencing of these clones and subsequent BLAST search revealed that one corresponds to murine SRBI cDNA (3) and the other (designated pHRC7) encoded a protein of 228 amino acids (Fig. 2).
Structure of a GPI-anchored HDL-binding Protein-A hydropathy plot of the deduced amino acid sequence of the cDNA shows the presence of two hydrophobic regions (Fig. 2), one at the N terminus and the other at the C terminus. The former corresponds to a classical signal sequence of probable 19 amino acids in length, whereas the latter strongly resembles the hydrophobic region of GPI-anchored cell surface proteins. Excluding the N-terminal 19 amino acids, the mature protein would then consist of 209 amino acids with a calculated M r of 22,603. This value is greatly smaller than those values of the previously characterized candidate receptors for HDL (4 -8).
The predicted amino acid sequence revealed the presence of a region highly enriched with acidic amino acids (aspartate and glutamate) and a region similar to a highly conserved domain termed Ly-6 motif, which occurs singly in the GPI-anchored lymphoid differentiation antigen Ly-6 family, and is repeated 3-fold in urokinase-type plasminogen activator receptor. These results predicted that the cloned protein consists of an Nterminal signal sequence, an acidic region with a cluster of aspartate and glutamate residues, an Ly-6 motif, and a Cterminal hydrophobic region.
Based on the structural similarity with the GPI-anchored Ly-6 family proteins, we analyzed the effects of phosphatidylinositol-specific PIPLC treatment on the HDL binding. As shown in Fig. 3, the PIPLC treatment almost completely abol- RNA from the indicated murine tissues was probed with 32 P-labeled murine GPI-HBP1 (upper panel). The filters were exposed to Kodak X-Omat AR film with an intensifying screen at Ϫ80°C for 48 h. The same samples were subsequently hybridized with a control probe for mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (lower panel) and exposed to Kodak X-Omat AR with an intensifying screen at Ϫ80°C for 6 h.

FIG. 8. In situ hybridization analysis of GPI-HBP1 transcripts in mouse.
Sections A, C, and E were hybridized with an antisense probe to murine GPI-HBP1. Sections B, D, and F are negative controls with a sense probe. Hybridization signals were visualized in blue. Tissue sections prepared from liver (A and B), heart (C and D), and lung (E and F) of a normal male mouse were analyzed by in situ hybridization as described under ''Experimental Procedures.'' GPI-HBP1 transcripts are localized in the Kupffer cells (indicated by arrows in A) and sinusoidal endothelium (arrowheads in A), but no significant accumulation is detected in the parenchymal cells. In the heart, the hybridization signal is detected in cardiac muscle cells (C). The intense hybridization signals are seen in bronchial epithelium (arrow in E) and alveolar macrophages (arrowhead in E) in the lung. Bars, 50 m ished the uptake of DiI-HDL, suggesting that the cloned protein is indeed GPI-anchored. Therefore, we designate the cloned HDL-binding protein as GPI-HBP1.
EST data base search identified an isoform lacking the C-terminal half of GPI-HBP1. This isoform (designated sHBP1) shares the N-terminal signal sequence, the acidic region, and part of the Ly-6 motif with GPI-HBP1 but lacks the C-terminal hydrophobic region, suggesting that the isoform is a secreted form generated by alternative splicing. Consistent with the deduced structural feature, ldlA7 cells transiently transfected with sHBP1 failed to bind DiI-HDL on the cell surface (Fig. 1C).
BLAST searches of the GenBank TM databases revealed that GPI-HBP1 is structurally related to Ly-6 molecules, a set of GPI-anchored cell surface proteins belonging within a superfamily that includes the urokinase-type plasminogen activator receptor, the complement inhibitor CD59, the sperm antigen SP10, and more distantly, the snake venom neurotoxin family. The Ly-6 motif consists of ϳ90 amino acids with ten highly conserved cysteine residues (20). A comparison of the sequence of the Ly-6 motif in GPI-HBP1 with those in the Ly-6 family members revealed that these cysteine residues are completely conserved in the Ly-6 motif of GPI-HBP1 (data not shown). The phylogenetic tree of the Ly-6 family proteins of various origins indicates that GPI-HBP1 is most closely related to Ly-6 molecules (data not shown).
HDL Binding-To characterize lipoprotein-binding properties, cDNAs encoding GPI-HBP1 and SRBI were stably expressed in ldlA7 cells. The binding of 125 I ( 125 I-HDL) or fluorescent (DiI-HDL) labeled HDL was measured following incubation of the cells. As shown in Fig. 4, A and B, the saturation binding of 125 I-HDL was observed in both GPI-HBP1-and SRBI-expressing cells at 4 and 37°C. Although the maximal binding of 125 I-HDL in GPI-HBP1-expressing cells was 3-fold lower than that of SRBI-expressing cells, the calculated K d values of the two proteins were within the same range (2-3 g/ml). The relative maximal binding activity between SRBI and GPI-HBP1 could not be determined because the expression levels of these two proteins in ldlA7 cells were unknown. Similar saturation binding was observed when cells were incubated with DiI-HDL at 4°C (Fig. 4C). In contrast, when GPI-HBP1-expressing cells were incubated with DiI-HDL at 37°C, the amounts of DiI-HDL binding by the cells were ϳ3-fold higher than those at 4°C and unsaturable (Fig. 4D). Similarly, the amounts of DiI-HDL binding at 37°C by SRBI-expressing cells were ϳ6-fold higher than those at 4°C. Furthermore, compared with 125 I-HDL binding, the degradation of 125 I-HDL (trichloroacetic acid-soluble 125 I) at 37°C was almost negligible in both cells (data not shown). These data indicate that GPI-HBP1, similar to SRBI (3), mediates selective lipid uptake but not the protein component of HDL.
Effects of SRBI Ligands-We next analyzed the effects of various ligands for SRBI (2,14) on the binding of 125 I-HDL to GPI-HBP1-expressing cells. As shown in Fig. 5, in the presence of excess unlabeled HDL, the binding of 125 I-HDL to GPI-HBP1-and SRBI-expressing cells was strongly reduced. Compared with the strong inhibition of 125 I-HDL binding to SRBI-expressing cells by human apoAI (free form), phosphatidylserine, and acetylated LDL (inhibited by ϳ75%), the inhibitory effects by these compounds were relatively lower in GPI-HBP1-expressing cells (reduced by ϳ50%). Oxidized LDL, native LDL, and human apoAII had relatively weak inhibitory effects on 125 I-HDL binding to GPI-HBP1-and SRBI-expressing cells. These data show that HDL is bound to GPI-HBP1 most preferentially among various SRBI ligands including HDL, phosphatidylserine, and acetylated LDL.
Cholesterol Efflux-In addition to selective lipid uptake from HDL particles, SRBI mediates HDL-dependent cholesterol efflux. To test whether GPI-HBP1 also mediates cholesterol efflux, the cells were labeled with [ 3 H]cholesterol, allowed for the equilibration of cellular cholesterol pools, and then incubated with or without HDL. As shown in Fig. 6, SRBI-expressing cells exhibited HDL-dependent cholesterol efflux with time-dependent manner, whereas almost no cholesterol efflux was seen in GPI-HBP1-expressing cells in the absence or the presence of HDL. The lack of cholesterol efflux in GPI-HBP1 cells indicates that GPI-HBP1 mediates selective lipid uptake only, whereas SRBI mediates both influx and efflux of cholesterol.
Expression of GPI-HBP1 Transcripts-Northern blot analysis of RNA from various murine tissues revealed hybridization of the HBP1 probe to a major transcript of 0.8 kilobases with the highest expression in heart and, to a much lesser extent, in lung and liver (Fig. 7). Apparently, no transcripts were detected in other tissues including brain, kidney, skeletal muscle, spleen, and testis.
To locate cells expressing GPI-HBP1 transcripts, in situ hybridization was performed using tissue sections from murine liver, heart, and lung. In the liver, the accumulation of hybridization signals for GPI-HBP1 transcripts appearing dark blue were detected most intensely in the Kupffer cells and sinusoidal endothelium, but no significant accumulation was detected in the parenchymal cells (Fig. 8, panel A). In the heart, the intense hybridization signal was detected in cardiac muscle cells (Fig. 8, panel C). The intense hybridization signals were also detected in bronchial epithelium and in alveolar macrophages in the lung (Fig. 8, panel E).
In contrast to the abundant expression of SRBI in steroidogenic tissues (3,21), GPI-HBP1 transcripts were highly accumulated in the Kupffer cells as well as in alveolar macrophages of the lung. Based on the abundant expression in these scavenger cells and the lack of cholesterol efflux, it is suggested that GPI-HBP1 plays a role in the initial entry of HDL cholesterol into these scavenger cells for further transportation of cholesterol. To elucidate the precise biological role of GPI-HBP1 and to determine any disorders caused by the absence of the protein, the generation of mice lacking GPI-HBP1 is currently underway.