The Low Density Lipoprotein Receptor-related Protein/a2-Macroglobulin Receptor Binds and Mediates Catabolism of Bovine Milk Lipoprotein Lipase*

Lipoprotein lipase (LPL), the major lipolytic enzyme involved in the conversion of triglyceride-rich lipoproteins to remnants, was found to compete with binding of activated a2-macroglobulin (a2M*) to the low density lipoprotein receptor-related protein (LRP)/a2-macro-globulin receptor. Bovine milk LPL displaced both 1251-labeled a2M* and 39-kDa azM receptor-associated protein (RAP) from the surface of cultured mutant fibroblasts lacking LDL receptors with apparent K , values at 4 "C of 6.8 and 30 nM, respectively. Further-more, LPL inhibited the cellular degradation of 1251-a2M* at 37 "C. Because both a2M* and RAP interact with LRP, these data suggest that LPL binds specifically to this receptor. This was further supported by observing that an immunoaffinity-isolated polyclonal antibody against LRP blocked cellular degradation of '*'I-LPL in a dose-dependent manner. In addition, 1251-LPL bound to highly purified LRP in a solid-phase assay with a KO of 18 nM, and this binding could be partially displaced with a2M' (K, = 7 nM) In some experiments, normal human foreskin fibroblasts or NRK cells purchased from the American Type Culture Collection, Rock- ville, MD were used. Cells were grown to confluence in 25- or 35-mm plastic wells, and binding of lZ6I-ligands to cells at 4 "C for 3 h or 37 "C for 5 h was determined using previously described methods (25). Some binding assays at 37 "C were performed in the presence of lipoprotein-deficient serum (LPDS) at 4 mg of protein/ml, whereas other assays at 37 "C and all assays at 4 "C were performed in buffer containing 4 mg/ml BSA. Surface-bound ligands were measured using a modification of the method of Van Leuven et al. (26) as radioactivity released by incubation of cells for 1 h at 4 "C in phosphate-buffered saline (pH 7.4) containing 50 pg/ml trypsin (GIBCO), 50 pg/ml proteinase K (Sigma), and 5 mM sodium EDTA. During this incubation, the cells detached from the well. The detached cells were pelleted in microfuge tubes by centrifugation for 90 s, washed once in protein-free buffer, and recentrifuged. Radioactivity in the supernatant fluid and the cell pellet was measured separately. This method, which gave results comparable to those using 10 mg/ml heparin for lZ5I-LPL release, was used because heparin did not adequately displace surface-bound lZ5I-a2M' (data not shown). Degraded ligands were measured as trichloroacetic acid-soluble radioactivity. The

** Recipient of Individual Research Service Award HLO8467 from the National Heart, Lung and Blood Institute.
The abbreviations used are: LPL, lipoprotein lipase; apo-, apolipoprotein; LDL, low density lipoproteins; @-VLDL, @-migrating very low density lipoproteins; LRP, LDL receptor-related protein; ~z M ' , activated a,-macroglobulin; RAP, receptor-associated protein; GST, glutathione-S-transferase; LPDS, lipoprotein-deficient serum; KLH, keyhole limpet hemocyanin; BSA, bovine serum albumin. thought that lipoprotein remnant clearance occurs via a specific receptor that is distinct from the LDL receptor (4). The identity of the remnant receptor remains unknown. However, recently the low density lipoprotein receptor-related protein (hereafter referred to as LRP), which is identical to the a2macroglobulin (a2M) receptor ( 5 , 6), became a candidate for a remnant receptor because of its high sequence homology to the LDL receptor ( 5 ) and its ability to mediate cellular catabolism of apoE-enriched @-migrating very low density lipoproteins (@-VLDL); these are abnormal, remnant-like particles produced in cholesterol-fed rabbits (7).
Strickland et al. (6) discovered that LRP is identical to the a2M receptor. This was a surprising observation because no previous relationship between aZM and lipoprotein metabolism was suspected. a2M can be activated to bind LRP by a variety of proteases and primary amines (8). When bound to its receptor, activated aZM (azM') is known to enter cells via clathrin-coated pits and undergo degradation in lysosomes (9,10). In addition to azM* and apoE-enriched P-VLDL, other ligands for LRP include the 39-kDa receptor-associated protein (RAP) and Pseudomonas exotoxin A (6,11). The 39-kDa RAP is particularly important because it is cosynthesized and copurifies with LRP and thus appears to act as an endogenous cellular ligand that can inhibit binding of other molecules to Recently, the possibility that LPL might be involved in lipoprotein remnant recognition by LRP was suggested by Beisiegel et al. (15). These investigators chemically crosslinked lz5I-LPL to a protein resembling LRP in size in intact cells. Also, addition of LPL to incubation media markedly increased surface binding of P-VLDL by cells at 4 "C (15).
The effect of LPL on @-VLDL binding was observed in mutant fibroblasts that could not express LDL receptors but which expressed LRP normally. In the current studies, we investigated the ability of LPL to bind and compete for ligands binding to LRP both on cultured fibroblasts lacking LDL receptors and in a solid-phase assay using highly purified LRP.

EXPERIMENTAL PROCEDURES
Proteins-Human n2M was purified by polyethylene glycol precipitation of plasma followed by zinc chelate and size-exclusion chromatography as described (14). azM was activated by incubation in buffer containing 200 mM methylamine as described (14). LRP was isolated from human placenta by affinity chromatography (16) followed by heparin-Sepharose chromatography to remove the 39-kDa RAP which copurifies with LRP. A rabbit polyclonal antibody to purified LRP was prepared and isolated by affinity chromatography over LRP conjugated to Sepharose as described (11). Another rabbit polyclonal antibody against the cytoplasmic carboxyl-terminal 11 amino acids of LRP conjugated to keyhole limpet hemocyanin (KLH) was used as a control. The peptide comprised of the carboxyl-terminal 11 amino acids of LRP was synthesized by a t-Boc-benzyl protection strategy using a T-bag method and conjugated to KLH as described (5,17). The 39-kDa RAP used in ligand binding assays was produced as a fusion protein with glutathione-S-transferase (GST) in an expression system utilizing human placental RAP cDNA (14). In initial studies, the intact fusion protein (RAP-GST) was used (Mr 70,000). Subsequently, thrombin digestion at a cleavage site in the fusion protein was performed to generate recombinant RAP (Mr 39,000) (14).
LPL was isolated by heparin-Sepharose chromatography of bovine milk as described (18) and stored at -20 "C in buffer containing 10 mM sodium phosphate (pH 7.4) and 50% glycerol. Because native LPL is dimeric (18,19), the calculated molecular weight of the protein dimer ( M , 101,000) was used in calculations of binding constants (20). Human apoCII, which activates LPL's enzymatic activity (l), was isolated from the d < 1.02 g/ml fraction of plasma by sizeexclusion and ion-exchange chromatography (21). Human LDL (d = 1.02-1.05 g/ml) and VLDL (d < 1.006 g/ml) were isolated from normal subjects as described (22). Proteins were iodinated to specific activities of 1,000 to 3,000 cpm/ng using IODOBEADS (Pierce Chemical Co.) or ENZYMO-BEADS (Bio-Rad) as described (14,23). Free lZ5I was removed from lZ5I-LPL by dialysis at 4 "C in buffer containing 10 mM sodium phosphate (pH 7.4) and 50% glycerol.
Binding Assays-Mutant skin fibroblasts that are incapable of expressing LDL receptors (24) were purchased from the NIGMS Human Genetic Mutant Cell Repository (GM00486A), Camden, NJ.
In some experiments, normal human foreskin fibroblasts or NRK cells purchased from the American Type Culture Collection, Rockville, MD were used. Cells were grown to confluence in 25-or 35-mm plastic wells, and binding of lZ6I-ligands to cells at 4 "C for 3 h or 37 "C for 5 h was determined using previously described methods (25). Some binding assays at 37 "C were performed in the presence of lipoprotein-deficient serum (LPDS) at 4 mg of protein/ml, whereas other assays at 37 "C and all assays at 4 "C were performed in buffer containing 4 mg/ml BSA. Surface-bound ligands were measured using a modification of the method of Van Leuven et al. (26) as radioactivity released by incubation of cells for 1 h at 4 "C in phosphate-buffered saline (pH 7.4) containing 50 pg/ml trypsin (GIBCO), 50 pg/ml proteinase K (Sigma), and 5 mM sodium EDTA. During this incubation, the cells detached from the well. The detached cells were pelleted in microfuge tubes by centrifugation for 90 s, washed once in proteinfree buffer, and recentrifuged. Radioactivity in the supernatant fluid and the cell pellet was measured separately. This method, which gave results comparable to those using 10 mg/ml heparin for lZ5I-LPL release, was used because heparin did not adequately displace surfacebound lZ5I-a2M' (data not shown). Degraded ligands were measured as trichloroacetic acid-soluble radioactivity.
The solid-phase assays were performed in 96-well microtiter plates as described (14). For each assay, binding to BSA-coated wells was determined as a control. For both fibroblast and solid-phase assays, equilibrium binding data obtained by homologous or heterologous ligand displacement at 4 "C were analyzed by a nonlinear, leastsquares curve fitting program (LIGAND) using a model that included one saturable site for specific binding and one nonsaturable site for nonspecific binding (27). Nonequilibrium dose-response data obtained at 37 "C were analyzed by a nonlinear, least-squares curve fitting program (ALLFIT) and reported as the effective concentration for 50% inhibition (EC50) of the response in the absence of competitor (28). Average binding constants are reported as geometric means.

RESULTS
Mutant fibroblasts that are incapable of expressing LDL receptors were used so that data interpretation would not be confounded by the possibility that binding to LDL receptors contributed to the results. Fig. lA shows that bovine milk LPL displaced lZ5I-azM' from the surface of fibroblasts at 4 " c ; t h e average KI for two separate experiments was 6.8 nM. In contrast, '251-a2M' bound with a KO of 0.37 nM (data not shown). Qualitatively similar results were obtained when the ability of LPL to inhibit 1251-a2M' binding to normal fibroblasts or NRK cells was studied (data not shown). LPL also inhibited cellular degradation of lZ5I-a2M* at 0.7 nM with an average EC50 of 15 k 1. The assay in Fig. 1B was performed in the presence of LPDS at 4 mg of protein/ml. Preparation of LPDS involved activation of coagulation with thrombin which is also known to activate a2M (29). LPDS was heat-inactivated prior to use but, nevertheless, could have contained thrombin-activated a2M. In addition, LPDS contains apoCII, the cofactor for LPL's enzymatic activity (1). However, results obtained when BSA was substituted for LPDS in the incubation buffer were similar (data not shown) which suggests that neither a2M* nor apoCII present in LPDS affected the outcome of the assay. Also, addition of purified apoCI1 in &fold molar excess to t h e LPL concentration did not affect LPL binding affinity at 4 "C (data not shown).
LPL displaced lZ5I-RAP-GST from the surface of fibroblasts lacking LDL receptors (Fig. 2 A ) ; the average Kr from two separate experiments at 4 "C was 30 nM. Fig. 2B shows the ability of various proteins to compete for lZ5I-LPL binding.
By homologous ligand displacement, LPL bound with an average KD of 11 nM ( n = 2). RAP-GST competed for 1251-LPL binding with an average K, of 2 nM ( n = 2) which is comparable to the affinity of RAP for purified LRP (14). In contrast, azM*, native a2M (not shown), and GST competed poorly (Fig. 2B). RAP-GST also inhibited degradation of lZ5I-LPL at 37 "C (data not shown). Thus, LPL and RAP appeared to compete for the same cell-surface-binding sites. As dis- cussed later, the inability of a2M* to inhibit the majority of 1251-LPL binding may have been due to the attachment of lZ5I-LPL to cell-surface proteoglycans that cannot bind azM'.
A rabbit polyclonal antibody to LRP was able to block degradation of 1251-a2M' and lZ5I-LPL by fibroblasts at 37 "C in a dose-dependent manner (Fig. 3C). A control antibody against KLH conjugated to the cytoplasmic carboxyl-terminal 11 amino acids of LRP had no effect on catabolism. However, in contrast to its effects on 1251-a2M', the antibody that blocked degradation did not cause parallel changes in surface binding and cell uptake of lZ5I-LPL and, in fact, did not significantly inhibit these parameters (Figs. 3, A and B ) . This suggests that, unlike 1251-~2M', a major fraction of lZ5I-LPL was bound to sites that were not blocked by the antibody. In Fig. 3, the amounts corresponding to 100% for surface binding, uptake, and degradation of lZ5I-LPL were 64, 65, and 55 ng/ well, respectively, and 3.0, 1.5, and 14 ng/well, respectively, for lZ5I-azM'. Thus, relative to the amounts of the ligands degraded, much more lZ5I-LPL was present on the surface or intracellularly than was the case for 1251-a2M'. The possibility that binding to slowly internalized cell-surface proteoglycans contributed to lZ5I-LPL catabolism is addressed under "Discussion." In other experiments, denaturation of lZ51-LPL by boiling for 30 min reduced degradation by about 10-fold as compared t o native LPL. However, the degradation remaining could still be blocked by the antibody to LRP. Also, denaturation reduced but did not eliminate the ability of LPL to compete for lz5I-azM' degradation (data not shown). Qualitatively similar results to those presented above were obtained in two other assays in which lower antibody concentrations were tested (data not shown).
To more definitively establish that LPL bound specifically to LRP, we studied its binding to highly purified receptors using a solid-phase assay. Fig. 4A shows that lZ5I-LPL exhibited high affinity, saturable binding to LRP-coated wells at 4 "C in contrast to binding to BSA-coated wells. The average K D from three separate assays was 18 f 5 nM which is close to the value of 11 nM obtained in cell-surface assays. aZM* partially displaced Iz5I-LPL binding with a KI at 4 "C of 7 nM (Fig. 4B), whereas RAP partially displaced lZ5I-LPL with a K , of 3 nM (Fig. 4C). Denaturation of LPL by boiling for 30 min reduced but did not destroy binding to LRP (data not shown).

DISCUSSION
Beisiegel ct al. (15) showed that LPL can be cross-linked in intact cells to a protein resembling LRP in size. The current studies establish that LPL binds directly to LRP and further indicate that LPL can enter cells and be degraded via an LRP-mediated process. LPL inhibits binding to cell surfaces by two previously described ligands for LRP, a2M*, and RAP and competes for intracellular degradation of a2M*. Degradation of LPL can be blocked by a polyclonal antibody against LRP. Furthermore, LPL displays saturable, high affinity binding to purified LRP in a solid-phase assay. Denaturation of LPL by boiling reduces but does not destroy its binding to LRP. Several important aspects of the binding interaction remain to be determined, including the binding stoichiometry and the sites of molecular contact.
It appears that LRP contributes only a fraction of the cellsurface-binding sites available to LPL. Although LPL can displace azM' binding to cells, the reverse is not true despite the fact that azM* can partially compete for LPL binding to purified LRP in a solid-phase assay (see Figs. lA, 2B, and 4B). Further evidence that binding of LPL to LRP is only a portion of the total LPL bound to the cell surface comes from studies using a polyclonal antibody to LRP. The majority of intracellular degradation of LPL can be blocked by the antibody without significantly diminishing surface binding or uptake (Fig. 3). Also, unlike azM', 39-kDa RAP can completely displace cell-surface binding of LPL and vice uersa (Fig. 2). Apparently, LPL and RAP bind cell-surface sites that cannot bind a' ".
One possible explanation for the data is that both LPL and RAP, which are heparin-binding proteins (1,12,30), bind to cell-surface proteoglycans under our assay conditions. If so, LPL and RAP might compete for proteoglycan attachment sites that cannot bind a2M*. Preliminary, unpublished studies suggest that a2M* does not bind heparin-Sepharose as strongly as does LPL. Also, digestion of fibroblasts with a mixture of heparinase and heparatinase can reduce LPL binding to the cell surface by up to 70% without affecting aZM* binding.' Detailed data on RAP and a2M' binding to cell-surface proteoglycans are unavailable, but studies by other investigators indicate that LPL binds proteoglycans on endothelial cells with a KO of 140-560 nM (31,32). A 220-kDa proteoglycan on the surface of endothelial cells has been identified that mediates slow internalization of surface-bound LPL (28% in the catabolic fate of LPL in cells. Conceivably, proteoglycans are involved in presenting LPL to LRP. This scheme is analogous to the recently described interaction of urokinase/ plasminogen activator inhibitor type-1 complexes with LRP in which the complexes appear to first bind urokinase-type plasminogen activator receptors, and then bind to LRP (35).
A major, unanswered question is whether LPL bound to triglyceride-rich lipoproteins i n vivo can provide a recognition site for lipoprotein clearance via hepatic receptors as originally proposed by Felts et al. (36) or, more specifically, via LRP as suggested by recent data. Normally, LPL circulates in the picomolar range in plasma and almost exclusively bound to lipoproteins (37,38). After intravenous heparin, LPL's concentration increases about 10-fold into the nanomolar range (37), although still somewhat lower than the KO of LPL binding to LRP at 4 "C (18 nM). I n viuo, LPL is rapidly cleared by the liver, which is also the major site for a2M* clearance (39,40). The enzyme is located in the space of Disse and inside both hepatocytes and endothelial cells in the livers of rats (41). The amount of LPL present in the liver increases after fat feeding or intravenous infusion of triglyceride-rich emulsions (42, 43). We find that the presence of normal human VLDL or LDL in 5-fold excess by protein mass do not inhibit the ability of LPL to compete for aZM' binding or degradation.' Thus, it seems possible that LPL could first attach to lipoproteins either in the extrahepatic or hepatic circulation and then bind to LRP, thereby facilitating lipoprotein uptake into cells. Although also potentially relevant to lipoprotein clearance, the present studies did not test LPL binding to LDL receptors or whether hepatic lipase, another lipolytic enzyme with major effects on lipoprotein catabolism (l), might also bind to LRP. Further studies are needed to address these possibilities.