The binding activity of the macrophage lipoprotein(a)/apolipoprotein(a) receptor is induced by cholesterol via a post-translational mechanism and recognizes distinct kringle domains on apolipoprotein(a).

Elevated plasma levels of lipoprotein(a) (Lp(a)) can be a risk factor for atherosclerosis, and the interaction of Lp(a) with cholesterol-loaded macrophages (foam cells) in atheromata may be important in Lp(a)-induced atherogenesis. We have previously shown that when cultured macrophages are loaded with cholesterol, they acquire the ability to internalize and lysosomally degrade Lp(a) via interaction between a novel cell-surface receptor activity and the apolipoprotein(a) (apo(a)) moiety of Lp(a). Herein we explore the cell-surface binding of recombinant apo(a) (r-apo(a)) by foam cells. Whereas the induction of degradation of r-apo(a) by cholesterol loading of macrophages depended on new protein synthesis, the induction of binding of r-apo(a) did not. Furthermore, J774 macrophages bound r-apo(a) in a cholesterol-regulatable and specific manner but degraded r-apo(a) poorly. Thus, the binding and internalization/degradation functions of the receptor activity are distinct. To explore which domains on r-apo(a) interact with the foam cell receptor, we conducted a series of competitive and direct binding and degradation experiments using 12 r-apo(a) constructs that differed in their content of specific kringle subtypes. These data, as well as complementary data with anti-apo(a) monoclonal antibodies, indicated that the region centered around kringle type IV, subtypes 6-7 (KIV6-7) is important in receptor binding. Remarkably, a cholesterol-induced receptor activity with similar structural specificity was also found on Chinese hamster ovary cells. In conclusion, the foam cell Lp(a)/apo(a) receptor consists of a cholesterol-regulatable binding activity and a short-lived component necessary for internalization or lysosomal degradation; the binding activity interacts with a distinct region of apo(a) that is different from that involved in competition for plasminogen binding.

Lp(a) 1 is an LDL-like lipoprotein in which the apoB-100 moiety of LDL is covalently attached to a glycoprotein called apo(a) (1,2). Apo(a) consists of multiple domains called kringles, which are regions of protein folds each stabilized by three disulfide bonds (3). Apo(a) shares 80% homology with another kringle-containing protein, plasminogen (3). Although the physiological role of Lp(a) is not known, elevated levels of this lipoprotein in certain human populations are often associated with increased risk for atherosclerotic coronary artery disease and stroke (4). Furthermore, Lp(a) transgenic mice (5,6) and, in one report ( (7); cf. Ref. 8), apo(a) transgenic mice have been found to have accelerated atherosclerosis.
The mechanism of Lp(a)-induced atherosclerosis is not known. Several groups of investigators have postulated that the ability of Lp(a) to compete for plasminogen binding sites on cells is important in certain potentially atherogenic processes, such as decreased fibronolysis (9 -11) and increased smooth muscle cell proliferation (12). Another possible clue to the potential atherogenicity of Lp(a) comes from the observation that Lp(a) and apo(a) are often physically in contact with cholesterol-loaded macrophages (foam cells) (13), which are prominent components of atherosclerotic lesions (14 -16). In previous work, we have demonstrated that mouse peritoneal and human monocyte-derived macrophages have a receptor activity that can bind, internalize, and lysosomally degrade Lp(a) (17,18). Importantly, the receptor activity is induced by cholesterol loading (17) and down-regulated by interferon-␥ (19), suggesting possible roles in atherosclerosis and inflammation. The receptor recognizes the apo(a) moiety of Lp(a) and interacts similarly with Lp(a) and a 17-kringle recombinant apo(a) construct (17,18,20). We have shown that this foam cell receptor is distinct from the LDL receptor, the scavenger receptor, the LDL receptor-related protein, and plasminogen receptors (17,18); furthermore, antibodies against the macrophage MAC-1 receptor do not block interaction of foam cells with apo(a). 2 To further elucidate how the foam cell Lp(a)/apo(a) receptor interacts with apo(a), the present study was designed to determine which sites on apo(a) bind to the receptor. By conducting a series of competitive and direct binding and degradation experiments using 12 recombinant apo(a) constructs and two anti-apo(a) monoclonal antibodies, we show that the region centered around KIV 6 -7 is important in receptor binding. Furthermore, additional studies revealed that the receptor consists of distinct binding and internalization/degradation activities. Finally, we took advantage of a CHO cell line that could be cholesterol loaded to show that this cell type also has a cholesterol-inducible receptor that recognizes the KIV 6 -7 region of apo(a).
Antibodies and Immunoaffinity Chromatography-Monoclonal antibody 12C11 raised against the apo(a) moiety of human Lp(a) was purchased from Perimmune, Inc. (Rockville, MD). Monoclonal antibody 8B4 raised against human Lp(a) was generously provided by Dr. Gunther Fless, University of Chicago. These monoclonal antibodies were previously shown to recognize Lp(a) and apo(a), but not plasminogen, by immunoblot analysis, but neither had been further characterized before this report.
Immunoaffinity chromatography (Table I) was performed by covalently linking anti-apo(a) monoclonal antibodies 8B4 and 12C11 to Affi-Gel-10 matrix according to the manufacturer's instructions (Bio-Rad). Briefly, 200 g of each antibody was reacted with 1 ml of Affi-Gel-10 in 100 mM MOPS buffer, pH 7.5, for 3 h at 4°C on a rotating wheel. After 3 h, the matrix was centrifuged in a microcentrifuge and the supernatant was removed. All remaining active sites on the matrix were blocked by the addition of 100 mM ethanolamine, pH 7.4, by incubating for 1 h at 4°C. Matrix not linked to antibody (control matrix) was made by incubating the Affi-Gel-10 with ethanolamine only. The control and antibody-conjugated matrices were then washed extensively with NET buffer (150 mM NaCl, 20 mM Tris-HCl, 1 mM EDTA, 1.0% Triton X-100, 0.1% SDS, pH 7.4). Each 125 I-kringle construct (6.0 g, 100 -800 cpm/ng) was first incubated with control matrix for 1 h at 4°C; after this initial pre-clearing step, the matrix was pelleted, and aliquots of the supernatant, which contained greater than 99% each of the 125 I-recombinant kringle peptides, were incubated overnight at 4°C with each of the two antibody-conjugated matrices. The matrices were then washed extensively in NET buffer with an SDS concentration of 1.0%, followed by several washes in NET buffer containing 1.5 M NaCl. The matrices were re-equilibrated in the original NET buffer (0.1% SDS and 150 mM NaCl), and antibody-125 I-kringle complex was disassociated by the addition of NET buffer containing 50 mM glycine HCl, pH 2.5; the eluate was neutralized using NaOH. Recovery of radioactivity during this procedure was greater than 90%.
Lipoproteins and Non-lipoprotein Cholesterol-LDL (density, 1.020 -1.063 g/ml) from fresh human plasma and ␤-VLDL (density, Ͻ1.006 g/ml) from the serum of cholesterol-fed rabbits (22) were isolated by preparative ultracentrifugation (23). Acetyl-LDL was prepared by treating LDL with acetic anhydride as described by Goldstein and co-workers (24). Human Lp(a), isolated and purified by density ultracentrifugation and lysine-Sepharose chromatography as described previously (17), was generously provided by Dr. Lars Berglund, Columbia University. All concentrations of lipoproteins are given in terms of their protein content with bovine serum albumin as a standard. All lipoprotein preparations were stored under argon at 4°C and used within 4 weeks. Lp(a) was iodinated using the iodine monochloride method, as described previously (17). Non-lipoprotein cholesterol was added to media from a 20 mg/ml stock in ethanol; the final concentration of ethanol (also added to control wells) was 0.25%.
Cells-All cells were maintained in a 37°C tissue culture incubator with a 5% CO 2 atmosphere. J774.A1 macrophages (from the American Type Culture Collection, Rockville, MD (25)) were cultured in suspension in high glucose Dulbecco's modified Eagle's medium (DMEM), containing 7% (v/v) fetal calf serum, penicillin (100 units/ml), streptomycin (100 mg/ml), and glutamine (292 g/ml). Two days prior to an assay, the cells were harvested from the spinner, pelleted by centrifugation, resuspended in DMEM containing 10% fetal calf serum, and plated on 16-mm tissue culture wells in 24-well dishes. Resident mouse peritoneal macrophages from female ICR mice (20 -25 g) were plated in 16-mm dishes in DMEM containing 10% fetal calf serum as described (26). CHO-mSRAII cells, which were transfected with a murine macrophage scavenger receptor (class AII) cDNA (27), were generously provided by Dr. Monty Krieger (Massachusetts Institute of Technology). These cells were grown in monolayer culture in Ham's F12 medium containing 10% fetal bovine serum.
Unless indicated otherwise, the cells were washed twice with phosphate-buffered saline (PBS) 1 h after plating and then incubated for 2 days in DMEM, 10% LPDS alone or containing the indicated sources of cholesterol. Four hours before the beginning of the binding or degradation assay, the cells were washed twice with PBS and then incubated with DMEM containing 0.2% BSA in order to remove residual surfacebound lipid.
Cloning of Recombinant Apo(a) Constructs-The recombinant apo(a) (r-apo(a)) derivatives used to identify the domains in apo(a) that mediate binding to the macrophage receptor are shown schematically in Fig.  1. All of the apo(a) derivatives illustrated were assembled in the expression vector pRK5, which contains the cytomegalovirus (CMV) promoter and SV40 termination sequences (28). The construction and expression of pRK5ha17, which encodes a 17-kringle form of r-apo(a) (17K in Fig. 1), has been described previously (20). The construction of plasmids encoding the r-apo(a) derivatives 12K, 6K, KIV 6-P , KIV 7-P , KIV 8-P , and KIV 9-P has also been described elsewhere (29).
For the construction of (KIV 2 ) 5 , the cDNA clone a6 (3) was partially digested with BamHI. A fragment encoding four copies of kringle IV subtype 2 was isolated and ligated into the single kringle IV type 2 expression construct (30) that had been digested with BamHI.
An expression plasmid encoding apo(a) kringle IV subtype 10 followed by the kringle V and protease domain sequences was generated as follows: a construct encoding single kringle IV subtype 10 (30) was digested with AvrII; the resultant 580-bp fragment was replaced with a 1,678-bp AvrII fragment isolated from pRK5ha17, which contains the latter 22 bp of kringle IV subtype 10, followed by sequences encoding the kringle V and protease domains. The resultant expression construct was designated pRK5haIV 10-P .
An expression plasmid encoding apo(a) kringle IV subtypes 6 -7 was constructed using a three-part ligation. A 735-bp EcoRI/AluI fragment (encoding the signal sequence followed by the complete sequence of KIV 6 and the first 325 bp of KIV 7 ) was obtained by digestion of KIV 6-P (see above) with EcoRI and AluI. In order to obtain the latter 17 bp of kringle IV subtype 7 followed by a stop codon, we digested an expression plasmid encoding single kringle IV subtype 7 with AluI and SalI; the latter construct was generated using overlapping synthetic oligonucleotides as described previously (30). The EcoRI/AluI fragment and the AluI/SalI fragment were ligated into pRK5 that had been digested with EcoRI and SalI; the resulting expression plasmid was designated as pRK5haIV 6 -7 .
For the construction of a plasmid encoding kringle subtypes 5-8, we utilized the pRK5ha6 construct, which has been previously described (29). We digested the latter plasmid with HindIII and AluI; this released a 1,509-bp fragment that encodes the latter 242 bp of kringle IV subtype 5, followed by the complete sequence of kringle IV subtype 6, and the first 325 bp of kringle IV subtype 7. In order to obtain the latter 17 bp of kringle IV subtype 7, followed by the complete sequence of kringle IV subtype 8 and a stop codon, we utilized PCR to generate a 365-bp fragment spanning this region; the 5Ј PCR primer was designed to anneal to a sequence 5Ј to the AluI site in kringle IV subtype 7, while the 3Ј PCR primer was designed to anneal to the 3Ј end of the kringle IV subtype 8 sequence and contained a stop codon and SalI restriction site. The HindIII/AluI fragment and the AluI/SalI-digested PCR product were ligated into pRKha6 that had been digested with HindIII and SalI; the resultant expression construct was designated pRK5ha [5][6][7][8] .
In order to construct an expression plasmid encoding apo(a) kringle IV subtypes 6 -8, we utilized a three-part ligation as follows. An expression plasmid encoding KIV 6-P (see above) was digested with EcoRI and AluI, giving rise to a 735-bp fragment that encoded the signal sequence, followed by the complete sequence of KIV 6 and the first 325 bp of KIV 7 . In order to obtain the final 17 bp of KIV 7 , followed by the complete sequence encoding KIV 8 with a stop codon at the 3Ј end, we digested the plasmid encoding KIV 5-8 (above) with AluI and SalI. This fragment (380 bp) and the 735-bp EcoRI/AluI fragment (above) were ligated into pRK5 that had been digested with EcoRI and SalI. The final expression plasmid was designated as pRK5haIV 6 -8 .
Expression, Purification, and Iodination of r-Apo(a) Derivatives-The r-apo(a) derivatives shown in Fig. 1 were used to stably transfect human embryonic kidney cells (293 cells) (31) by the method of calcium phosphate co-precipitation (32). Briefly, 10 g of each expression plasmid was co-transfected with 1 g of plasmid encoding the neomycin resistance gene (33). Stable transfectants were selected by culturing cells in the presence of the antibiotic G418 (800 g/ml) (Life Technologies, Inc.) as described previously (20). Stable transfectants expressing the various apo(a) derivatives were identified by enzyme-linked immunosorbent assay as described previously (29).
All of the r-apo(a) derivatives, with the exception of (KIV 2 ) 5 , were purified by affinity chromatography on lysine-Sepharose (Pharmacia Biotech Inc.) columns (29). The r-apo(a) derivative (KIV 2 ) 5 was purified by immunoaffinity chromatography using an anti-apo(a) polyclonal antibody, raised in rabbits (34), immobilized on Affi-Gel (see above). Briefly, conditioned medium (OptiMEM; Life Technologies, Inc.) from 293 cells expressing (KIV 2 ) 5 was applied to a 5-ml immunoaffinity column, and the column was washed with 5 volumes of Tris-buffered saline (TBS; 25 mM Tris-HCl, pH 7.4, 135 mM NaCl), followed by 10 volumes of TBS containing 1 M NaCl and final 5 volumes of TBS. Specifically bound protein was eluted with 100 mM glycine, pH 2.3, and 1-ml fractions were collected into 100 l of 1 M Tris base. Fractions containing protein were pooled and dialyzed against HEPES-buffered saline (20 mM HEPES, pH 7.4, 150 mM NaCl). All proteins were judged to be pure by SDS-PAGE followed by silver staining. Following purification, proteins were stored at Ϫ70°C and shipped from Kingston to New York on dry ice via overnight courier. Proteins were stored at Ϫ70°C prior to use, at which time the integrity of the preparations was verified by SDS-PAGE.
The r-apo(a) constructs were iodinated by the IODO-BEAD method (Pierce) as described (18). Briefly, 500 g of each construct was reacted with 200 Ci of carrier-free 125 I and one IODO-BEAD for 4 min. Unreacted 125 I was removed either by dialysis or by PD-10 gel filtration chromatography (Pharmacia). The 125 I-labeled constructs were at least 95% trichloroacetic acid-precipitable, had a specific activity of 100 -800 cpm/ng, and were used within 2 weeks of the labeling procedure.
Binding and Degradation Assays-After the preincubations described in the text and figure legends, the cells were washed with PBS, incubated for the indicated times with 10 or 25 nM 125 I-labeled ligand in 0.5 ml of DMEM, 0.2% BSA, and then assayed for binding (cf. Ref. (35) and below) or degradation (17,18). For the binding assay, the cells were cooled for 1 h at 4°C before the addition of ligand and then incubated with 125 I-labeled r-apo(a) at 4°C for 3 h, at which point 125 I-r-apo(a) binding had reached equilibrium (not shown). The assay was terminated by removing the 125 I-ligand and washing the cells twice in succession with each of the following solutions: PBS containing 1 mM CaCl 2 and 0.5 mM MgCl 2 (solution I); solution I plus 2% BSA; and PBS. The cells were then solubilized in 0.1 N NaOH, and the amount of cellassociated 125 I radioactivity was determined by gamma counting. The data are expressed per mg of cellular DNA, which was determined by a minor modification (36) of the method of Labarca and Paigen (37).
Statistics-Unless otherwise indicated, results are given as mean Ϯ S.D. (n ϭ 3 separate dishes of cells in the same experiment). Absent error bars signify S.D. values smaller than the graphics symbols.

Cholesterol-induced Induction of r-Apo(a) Binding to the Cell Surface Does Not Require New Protein Synthesis and Can be Functionally Distinguished from r-Apo(a) Internalization and
Degradation-In our previous studies, we assayed the macrophage foam cell Lp(a)/apo(a) receptor assay by measuring internalization and lysosomal degradation of 125 I-labeled ligand (17,18). With this assay, we showed that cholesterol-mediated induction of receptor activity was dependent upon new protein synthesis (18). To further explore this point, we assayed cellsurface binding of 125 I-r-apo(a) to macrophages and foam cells. The r-apo(a) used for these studies, called 17K, is shown as the top construct in Fig. 1; it consists of all 10 of the kringle IV (KIV) subtypes, including eight copies of KIV 2 (17 KIV domains in all), plus kringle V and the protease domain (20). As shown previously, the foam cell receptor interacts with the apo(a) moiety of Lp(a), and experiments utilizing 125 I-17K have yielded similar results as those using 125 I-Lp(a) (17,18).
We first tested whether the binding activity of the receptor was induced by cholesterol loading and, if so, whether the induction required new protein synthesis. The surprising results with mouse peritoneal macrophages are shown in Fig. 2A.
The data depicted by the diagonal-hatched bars verify our previous results mentioned above, namely that 125 I-17K degradation is induced in mouse peritoneal macrophage foam cells and that the induction is blocked by treatment of the foam cells with the protein synthesis inhibitor, cycloheximide (18). Binding of 125 I-17K (cross-hatched bars) is also induced in these foam cells, but, remarkably, the induction of binding is not inhibited by cycloheximide treatment. These data suggest that the receptor consists of at least two distinct functional components: a binding component that is post-translationally induced by cellular cholesterol loading, and a relatively shortlived component that mediates internalization and degradation of the ligand. Note that the internalization and lysosomal degradation of 125 I-acetylated LDL, which enters cells via the scavenger receptor (38), is not blocked by cycloheximide (18), indicating that cycloheximide is not simply a nonspecific inhibitor of receptor internalization or lysosomal degradation of endocytosed ligands.
In previous unpublished work, 2 we made the intriguing finding that the J774 murine macrophage cell line, unlike mouse peritoneal and human monocyte-derived macrophages, degraded very little 125 I-17K or 125 I-Lp(a), and the degradation was not induced by cholesterol loading. This point was confirmed by the data depicted by the diagonal-hatched bars in Fig. 2B. At the initial time of this finding, we concluded that J774 macrophages lacked the Lp(a)/apo(a) receptor. Our new data, however, prompted us to examine 125 I-17K binding to these cells. As shown by the cross-hatched bars in Fig. 2B, The data in Fig. 3 show that 10-fold molar excess of 17K, 12K, 6K, and KIV 6-P were good competitors of 125 I-17K binding. In contrast, the more truncated forms lacking KIV 6 through KIV 9 , as well as the repeating KIV 2 construct ((KIV 2 ) 5 ), were relatively poor competitors or did not compete at all. In particular, the greatest loss of competitive inhibition was noted in constructs lacking KIV 6 and KIV 7 . These data can be interpreted in two ways, either specific kringle domains in the area of KIV 6 -7 region are important for receptor binding or the total number of kringles is critical.
The goals of the next series of experiments were to confirm the competitive binding data in Fig. 3 with direct and competitive 125 I-ligand degradation studies as well as to distinguish between the two interpretations of the data in Fig. 3 mentioned above. In Fig. 4A, unloaded mouse peritoneal macrophages, shown by the cross-hatched bars, and cholesterol-loaded mouse peritoneal macrophages (foam cells), shown by the solid bars, were incubated directly with 125 I-labeled 6K in the absence or presence of excess unlabeled 6K or 17K and assayed for 125 Iligand degradation. Consistent with the competitive binding data in Fig. 3, 125 I-6K was degraded by the macrophages, and degradation was induced 3-fold by cholesterol loading. Both unlabeled 6K or 17K competed well for receptor activity in the cholesterol-loaded macrophages. To determine the relative importance of specific kringles versus number of kringles, we tested the ability of unloaded macrophages and foam cells to degrade 125 I-KIV 5-8 (Fig. 4B). The data clearly show that this 4-kringle construct was recognized by the foam cell receptor and was competed by both unlabeled KIV 5-8 as well as by unlabeled 17K. Qualitatively similar results were obtained using these unlabeled ligands to compete for 125 I-lipoprotein(a) degradation by foam cells: inhibition (relative to unloaded macrophages) by unlabeled Lp(a), 17K, and KIV 5-8 was 79.1, 74.6, and 63.8%, respectively. To further define the receptorbinding domain, we examined the interaction of macrophages with 125 I-KIV 6 -8 (Fig. 4C). Degradation of this ligand was induced greater than 7-fold by cholesterol loading. Although in this particular experiment the absolute level of degradation was less than that seen with 125 I-6K or 125 I-KIV 5-8 , in another experiment the levels of degradation between 125 I-KIV 6 -8 and 125 I-17K were similar (data not shown). Furthermore, adding 20-fold excess unlabeled KIV 6 -8 and 17K instead of 10-fold excess inhibited 125 I-KIV 6 -8 degradation to values similar to or less than the unloaded macrophages value (data not shown). Consistent with the competitive binding data in Fig. 3, the degradation of 125 I-KIV 6 -7 (Fig. 6D) was considerably less than those of the three other ligands in Fig. 6, A-C; in addition, there was no statistically significant induction of 125 I-KIV 6 -7 degradation in foam cells, and unlabeled 17K failed to compete. Finally, 125 I-(KIV 2 ) 5 was poorly degraded by macrophages or foam cells (Fig. 4E), also consistent with the competitive binding data in Fig. 3. Thus, the failure of some of the ligands to compete for receptor binding in Fig. 3 cannot be explained simply by their having a small number of kringles. From these data, we conclude that the foam cell receptor recognizes a domain in the KIV 5-8 region. Recall that deletion of KIV 5 or KIV 8 had only minor effects on the ability of r-apo(a) constructs to compete for receptor binding (Fig. 3). Thus, the data in Fig.  4 most likely indicate that the KIV 6 -7 domain is most critical but needs to be presented to the receptor in the context of a larger sequence.
Given the importance of the data in Fig. 4, we conducted a set of experiments with a similar set of labeled and unlabeled ligands, except cell-surface binding was assayed. Fig. 5, A-C, shows that both 125 I-6K and 125 I-KIV 5-8 , but not 125 I-(KIV 2 ) 5 , bound specifically and in a cholesterol-regulatable manner to J774 macrophages. Although there were trends toward slight specific binding of 125 I-KIV 6 -7 (Fig. 6D), none of the differences shown are statistically different. Thus, these binding data are entirely consistent with the mouse peritoneal degradation data in Fig. 4, and, as expected, similar cell-surface binding data were obtained with mouse peritoneal macrophages (not shown). Furthermore, consistent with the data in Fig. 2A Finally, we utilized two anti-apo(a) monoclonal antibodies raised against apo(a) or Lp(a) (see "Experimental Procedures"). Based upon immunoaffinity data, we found that both antibodies recognized 17K and (KIV 2 ) 5 to a similar extent (Table I). In contrast, whereas 8B4 recognized 6K and KIV 5-8 very well, 12C11 interacted poorly with these structures. Thus, 8B4, but  Table I). All assay incubations contained blockers of the macrophage Fc-␥ receptors as described previously (18) (500 nM rat anti-murine Fc-␥ 2,3 receptor antibody (2.4 G 2 ; isotype 2B) from Pharmingen (San Diego) plus 1000 nM monomeric IgG 2A from Chemicon Int., Inc. (Temecula, CA)). not 12C11, recognizes the region of apo(a) that we postulate interacts with the foam cell receptor (see above). The effect of these two antibodies on 125 I-17K degradation and binding by mouse peritoneal foam cells is shown in Fig. 6. Antibody 8B4 inhibited both degradation (A) and binding (B), whereas 12C11 inhibited these processes hardly at all. These data further support the conclusion that the foam cell receptor recognizes a domain in the KIV 5-8 region.
Ligand Dose Relationship and Scatchard Analysis of 125 I-KIV [5][6][7][8] Cell-surface Binding to Mouse Peritoneal Macrophages-Based upon the data presented above, we conducted a binding study with 125 I-KIV [5][6][7][8] to determine the affinity and number of sites of the cholesterol-inducible receptor on macrophages. The data in Fig. 7 display total (open squares), specific (close circles), and nonspecific (open diamonds) binding data; there was very little nonspecific binding, and the specific and total binding curves showed a tendency toward saturation at the concentrations of 125 I-KIV 5-8 used. Scatchard analysis (inset) of these data was consistent with a single class of binding sites with an affinity of 2.5 ϫ 10 Ϫ11 M and a number of sites equal to approximately 2 ϫ 10 5 per cell. Whether a lower affinity, higher capacity site would be evident at higher concentrations of ligand cannot be determined from these data.
CHO Cells Have an Apo(a) Receptor Activity with Similar Properties as the Macrophage Receptor-To determine if cells other than macrophages have an apo(a) receptor similar to that described above, it was necessary to work with a cell type that could be loaded with substantial amounts of cholesterol. For this purpose, we utilized CHO-mSRAII cells, which are CHO cells that have been transfected with a murine macrophage scavenger receptor (class AII) cDNA (27). When incubated with acetyl-LDL, these cells accumulate large amounts of free and esterified cholesterol (39). As shown in Fig. 8A, unloaded cells bound relatively little 125 I-17K, but cells loaded with cholesterol bound an amount of ligand similar to or greater than that bound by cholesterol-loaded macrophages (see previous figures). The 125 I-17K was not simply sticking to surface-bound acetyl-LDL: the cells were routinely incubated overnight in the absence of acetyl-LDL prior to the binding assay (Fig. 8), and even when fucoidin was included in this overnight incubation to displace any residual cell-surface acetyl-LDL (40), there was no decrease in 125 I-17K binding (data not shown). Furthermore, as shown in the inset to Fig. 8A, 125 I-17K binding was further enhanced by increasing the free cholesterol content of the cells (by incubation with acetyl-LDL plus the ACAT inhibitor 58035; see Ref. 39). These data are similar to our previous findings in macrophages (17). Cholesterol-loaded CHO cells also degrade 125 I-17K (Fig. 8B, solid bars) as well as the more  monoclonal antibodies The four listed 125 I-r-apo(a) constructs were assayed for interaction with 8B4 or 12C11 monoclonal antibodies (mAb) attached to Affi-Gel-10 as described under "Experimental Procedures." The data are expressed as ((total cpm Ϫ non-pelleted cpm) Ϭ (total cpm)) ϫ 100, where total cpm refers to the cpm originally incubated with each antibody-matrix complex.

DISCUSSION
The original goal of this study was to determine the domains on apo(a) that are recognized by the foam cell Lp(a)/apo(a) receptor. In this regard, the combined binding, degradation, and monoclonal antibody data indicate that a region centered around KIV 6 and KIV 7 is critical, with flanking kringles probably necessary for proper conformation of these kringles or for optimal "presentation" of this region to the receptor (Figs. 3-6). Analysis of our initial 125 I-r-apo(a) cell-surface binding data (Fig. 2), however, revealed other important properties of the receptor that were not evident from our previous work (17,18), which used solely the ligand degradation assay. In particular, we obtained evidence that the binding and internalization/ degradation functions of the receptor are distinct. It is the binding function that is induced by cholesterol, and this induction does not require new protein synthesis.
One hypothesis to explain these data is that the receptor exists in a relatively inactive conformation in the plasma membrane of unloaded macrophages. Cholesterol loading of macrophages enriches the plasma membrane with cholesterol (41), and this is known to affect the conformation and function of a wide variety of plasma membrane proteins (42)(43)(44)(45)(46)(47). Thus, cholesterol loading may change the conformation of the Lp(a)/ apo(a) receptor to a more active form. Alternatively, cholesterol loading may affect the trafficking of the receptor to the plasma membrane, as was recently found for a glycosylphosphatidylinositol-anchored protein that is localized to cholesterol-rich domains of the plasma membrane (48). Since inhibition of protein synthesis blocks lysosomal degradation of r-apo(a) but not binding or lysosomal degradation of other receptor ligands, we conclude that there is a separate short-lived component of the Lp(a)/apo(a) receptor that is somehow necessary for internalization or delivery of ligand to lysosomes. Precedents for this idea include the mediation of G protein-coupled receptor internalization by ␤-arrestins (49) and the requirement for receptor ubiquitination in the internalization of certain yeast plasma membrane receptors (50,51). Interestingly, J774 macrophages can bind r-apo(a) in a cholesterol-regulatable (Fig. 2B) and specific (Fig. 5) manner, but these cells degrade r-apo(a) poorly (Fig. 2B). Thus, these cells may be lacking the putative internalization component. Proof of these ideas awaits molecular identification of the binding and internalization/degradation activities of the receptor.
The most important issue surrounding the foam cell Lp(a)/ apo(a) receptor is its physiological function and possible pathophysiological roles. Its regulation by cholesterol may indicate importance in atherosclerosis, and its regulation by interferon-␥ (36) may suggest a role in inflammation. Along these lines, we have previously postulated that the receptor may function to focally clear Lp(a) and apo(a) in atherosclerotic and inflammatory lesions (18,36), but this idea awaits experimental testing. Moreover, the fact that the receptor is present in species without Lp(a) (i.e. mice) suggests the presence of at least one other ligand, and the finding that the receptor is present on CHO cells (Fig. 8) may also indicate a broader function. Perhaps one clue for receptor function is revealed by its affinity (K D ϭ 2.5 ϫ 10 Ϫ11 M), which is quite high compared with most other mouse macrophage receptors whose affinities have been published. For example, mouse peritoneal macrophage receptors for the Fc moiety of IgG (52), acetyl-LDL (53), and mannose albumin (54)  tors with affinities similar to the one described here include the receptor for platelet activating factor (55) and the receptor for granulocyte macrophage-colony stimulating factor (56), both of which mediate cell signaling responses. Whether the high affinity nature of the cholesterol-inducible apo(a) receptor indicates a role in signal transduction remains to be explored.
To address these and other issues, our future goal is to identify the binding component of the receptor by expression cloning. In this regard, we may be able to take advantage of the presence of the receptor on CHO cells (above), which are wellsuited for somatic cell genetic strategies. Once cloned, studies utilizing targeted disruption of the receptor gene in mice should help shed light on receptor function. Cloning will also facilitate molecular interaction studies, including the identification of other ligands for the receptor and of the putative internalization component suggested by the data in this report.