Binding analysis of amino-terminal and carboxyl-terminal regions of the 39-kDa protein to the low density lipoprotein receptor-related protein.

A 39-kDa protein binds to the low density lipoprotein receptor-related protein/alpha 2-macroglobulin receptor (LRP/alpha 2MR) and inhibits the binding of ligands to this receptor. We recently reported that inhibition of tissue-type plasminogen activator binding to LRP/alpha 2MR is mediated by both amino-terminal and carboxyl-terminal regions of the 39-kDa protein, whereas inhibition of alpha 2-macroglobulin-proteinase binding is mediated only by amino-terminal regions. In this report we show that amino-terminal and carboxyl-terminal regions of the 39-kDa protein bind specifically and with high affinity to LRP/alpha 2MR on rat hepatoma MH1C1 cells. Following binding, these amino-terminal and carboxyl-terminal regions of the 39-kDa protein are each rapidly endocytosed and degraded with kinetics identical to the full-length 39-kDa protein. Competition binding experiments with these constructs demonstrate that amino-terminal and carboxyl-terminal regions of the 39-kDa protein compete with one another for binding to LRP/alpha 2MR. A model is proposed in which amino-terminal and carboxyl-terminal regions of the 39-kDa protein bind to different sites on LRP/alpha 2MR in order to inhibit ligand binding.

tein (RAP), blocks the binding andor uptake of all known ligands to LRP/a2MR (1, 10-13, 16, 17). LRP/a2MR is a widely expressed plasma membrane glycoprotein of 4525 amino acids that structurally resembles four LDL-receptor molecules arranged in series. The domain structure of LRP/a2MR includes 22 epidermal growth factor repeats and 31 cysteine-rich ligandbinding complement-type repeats (18). The 31 complementtype repeats are organized into four clusters of 2,8, 10, and 11 repeats. Recently the binding of RAP, a,-macroglobulin light chain, and u-PA.PAI-1 complexes was identified within a 624amino acid region of LRP/a2MR containing a cluster of 8 complement-type repeats (19). I t seems likely that al-macroglobulin light chain and u-PA.PAI-1 complexes bind to different sites within this 624-amino acid region because LRP/a2MRspecific ligands have been demonstrated to not cross-compete for binding (8,10). The mechanism of RAP inhibition of almacroglobulin light chain and u-PA.PAI-1 binding within this 624-amino acid region has not been elucidated.
We recently utilized glutathione S-transferase (GST)-fusion proteins encoding different regions of the 39-kDa protein to demonstrate that multiple regions on the 39-kDa protein can interact with LRP/a2MR to inhibit ligand binding (20). GSTfusion proteins encoding amino-and carboxyl-terminal regions of the 39-kDa protein inhibited t-PA binding to LRP/a2MR on rat hepatoma MHICl cells, whereas only amino-terminal constructs inhibited a2M* binding. Inhibition of t-PA and azM* binding to LRP/a2MR required residues 18-24 and 100-107 within amino-terminal constructs. Inhibition of t-PA binding by carboxyl-terminal constructs required residues 200-225 and 311-319 of the 39-kDa protein (20). The purpose of the present study was to examine the mechanism by which amino-and carboxyl-terminal regions of the 39-kDa protein interact with LRP/a2MR on MHICl cells to regulate ligand binding. We found that residues 1-114, 115-319, and 200-319 of the 39-kDa protein, generated as fusion proteins with GST, behaved identically to the full-length protein, GST/l-319, in terms of binding, uptake, and degradation by MHlCl cells. We also found that amino-and carboxyl-terminal constructs competed with one another for binding to LRP/a2MR. We propose that amino-and carboxyl-terminal regions of the 39-kDa protein bind to separate sites on LRP/a2MR to inhibit ligand binding.
EXPERIMENTAL PROCEDURES Materials-Carrier-free sodium [12611iodide was purchased from Amersham Corp. IODO-GEN was from Pierce Chemical Co. Pronase was obtained from Calbiochem.
Ligand Binding Assays-Rat hepatoma MHICl cells were cultured as described previously (21). MHICl cells were seeded into multiwell(12 welldplate) culture plates 2 days before the assay. Ligand binding buffer was phosphate-buffered saline containing 1 m~ CaCl, and 0.5 n m MgC1, (PBSc). Cell monolayers were washed three times on ice with 4 "C PBSc. Binding was initiated by adding 0.5 ml of PBSc containing the indicated '=I-labeled ligand in the absence or presence of competitor protein. After incubation a t 4 "C for 2 h, unbound ligand was removed by washing the cells three times with PBSc. Cells were lysed in 62.5 m~ Tris-HC1, pH 6.8, containing 0.2% (w/v) SDS and 10% (v/v) glycerol. Radioactivity of cell lysates was quantified by y scintillation spectrometry (Packard model C5304). Nonspecific binding was determined in the presence of excess unlabeled ligand. Ligand Uptake, Internalization, and Degrudation-MHICl cells were seeded into 6-well dishes and '2sI-GST/1-114 (8 m), 1261-GST/115-319 (5 I & , 12SI-GST/20&319 (8 m), or 1261-GST/1-319 (5 m) was allowed to bind for 2 h a t 4 "C. Nonspecific binding (generally less than 10% of total binding) was determined in the presence of excess unlabeled ligand (500 m GST/l-114, 300 m GST/115-319,500 IM GST/200-319, 500 m 1261-GST/1-319). After binding, cells were washed three times with 4 "C PBSc to remove unbound ligand. To initiate uptake, cells were warmed rapidly to 37 "C by adding prewanned PBSc containing 200 m unlabeled ligand. Following incubation a t 37 "C for selected intervals, the overlying medium was removed and precipitated by the addition of bovine serum albumin to 10 mg/ml and trichloroacetic acid to 20%. The cell monolayers were washed three times with 4 "C PBSc and incubated with PBSc containing 0.25% Pronase for 30 min at 4 "C. Cells were detached from the dishes by gentle pipetting and separated from the buffer by centrifugation. Radioactivity of cell pellets (defining internalized radioligand) and supernatant fractions (defining cell surface radioligand) were determined separately. Degradation of ligand was defined as the appearance of radioactive ligand fragments in the overlying media that were soluble in trichloroacetic acid.   Table I). of pre-bound amino-and carboxyl-terminal constructs was identical to that observed for the full-length protein, 1251-GST/ 1-319 ( Fig. 2 0 1 , in MHICl cells.  structs was similar for different 1251-carboxyl-termina1 constructs, the ability of amino-and carboxyl-terminal regions to inhibit 1251-GST/200-319 binding to MHICl cells was examined. As seen in Fig. 5 l 0.3 as 0.7 0.9 1.1 1.3 1 Fig. 7B is a double-reciprocal plot of the data in Fig. 7A and shows that the lines in the absence and presence of inhibitor (GST/l-114) intersect along the abscissa. The intersection point corresponds to a Kd value of 8 m. Fig. 7B also shows the lines have distinct intercepts along the ordinate and correspond to values of 330, 220, and 120 fmol of 1251-GST/200-319 bound per well in the absence and presence of 5 and 20 n~ unlabeled GST/l-114, respectively. Table I -0100.1 0.3 0.5 0.7 0.9 1.1 1. 3 1 were performed independently (i.e. GST/1-114 alone, GSTI 115-319 alone, or GST/1-319 alone) or simultaneously (i.e. GST/l-114, GST/115-319, and GST/l-319 assayed in separate dishes at the same time) and therefore all results were averaged. The average Kd values for GST/l-114 (9.7 n~) and GST/ 115-319 (8.9 n~) were similar to the full-length protein, GST/ 1-319 (8.1 m). However, the average number of binding sited cell was different for amino-and carboxyl-terminal regions of the 39-kDa protein with the carboxyl-terminal region having approximately four times as many binding sitedcell as the amino-terminal region (2,410,000 versus 583,000). The average number of binding sitedcell for the full-length protein, GST/l-319, was 978,000. Although amino-and carboxyl-terminal regions each bind independently to MHICl cells, it is not clear whether these regions bind to the same site on LRP/a2MR (see below) or whether multiple regions within the full-length 39-kDa protein can simultaneously bind to LRP/a2MR. It is possible that only one region (i.e. amino-or carboxyl-terminal) within the full-length protein binds to LRP/a2MR at any one time and that this binding induces a conformational change in LRP/a2MR such that other regions cannot bind. Therefore the number of binding sitedcell for the full-length protein may reflect a mixed population of different regions of the full-length protein binding to LRP/a2MR.

DISCUSSION
We previously demonstrated that both amino-and carboxylterminal regions of the 39-kDa protein inhibited t-PA binding to LRP/a2MR on MHICl cells, whereas only amino-terminal regions inhibited azM* binding (20). These observations prompted us to examine whether amino-and carboxyl-terminal regions bound to the same site or to separate sites on LRP/ azMR to inhibit t-PA binding. Binding competition experiments demonstrated that amino-and carboxyl-terminal regions of the 39-kDa protein competed with one another for binding to LRP/ a2MR. However, the extent of competition varied whereby unlabeled carboxyl-terminal constructs completely inhibited specific 1251-amino-terminal construct binding (Fig.  3B) and unlabeled amino-terminal constructs inhibited only 40430% of '251-carboxyl-termina1 construct binding (Figs. 4A and 5 A ) . The differences in these extents of inhibition are most likely due to a decrease in the number of binding sites available for binding. Since saturation binding experiments yielded an average of four carboxyl-terminal binding sites to one amino-terminal binding site (Table I), one would predict in competition binding experiments, using a fixed non-saturating concentration of '251-amino-terminal constructs and saturating concentrations of unlabeled carboxyl-terminal constructs (i.e. 50-500 nM), that all binding sites would be occupied with the carboxyl-terminal constructs and thus no '251-amino-terminal construct binding would be detected. Indeed, as seen in Fig. 3B, carboxyl-terminal constructs were able to completely inhibit '251-GST/1-114 binding. Conversely in competition binding experiments with a fured non-saturating concentration of 1251-carboxyl-termina1 constructs and saturating concentrations of unlabeled aminoterminal constructs, approximately 20-25% of the binding sites would be occupied with amino-terminal constructs and therefore the remaining binding sites would be available for Iz5Jcarboxyl-terminal construct binding. Figs. 4A and 5A demonstrated that saturating concentrations of unlabeled aminoterminal constructs maximally inhibited 4040% of lZ5Icarboxyl-terminal construct binding.
Although the binding competition experiments demonstrated that amino-and carboxyl-terminal constructs competed with one another for binding, these data did not demonstrate whether the constructs bound to the same site or to distinct sites on LRP/aZMR. To characterize the nature of competition (i.e. competitive inhibition versus noncompetitive inhibition), saturation binding experiments were performed with lZ5I-amino-and 1251-carboxyl-termina1 constructs in the presence of various fured concentrations of unlabeled competitor protein.
The amino-and carboxyl-terminal regions of the 39-kDa protein contain no sequence identity and, therefore, would be predicted to bind to different sites on LRP/a2MR. Double-reciprocal plots of the data using 1251-GST/1-114 in the absence and presence of unlabeled GST/115-319 showed that all lines intersected at the abscissa and had distinct intercepts on the ordinate (Fig. 6B). Thus the Kd value in the presence and absence of GST/115-319 was 5.5 nM, whereas the maximum femtomoles bound per well was reduced in the presence of inhibitor. Similarly, double-reciprocal plots of the data using lz5I-GST/2O0-319 in the absence and presence of unlabeled GST/1-114 showed that all lines intersected at the abscissa (corresponding to a Kd value of 8 IN) and had distinct intercepts on the ordinate (Fig. 7B). These double-reciprocal patterns are consistent with a mechanism of noncompetitive inhibition (24), whereby the inhibitor and substrate bind to different sites on LRP/a2MR.
A model is depicted in Fig. 8 to summarize t-PA and a2M* binding to LRP/a2MR and the regulatory role of the 39-kDa protein on ligand binding. As seen, t-PA and azM* bind to separate sites on LRP/a2MR (arrow 1 )because these ligands do not cross-compete for binding (10). We previously demonstrated that several amino-terminal regions of the 39-kDa protein, generated as fusion proteins with GST (GST/l-114 and GST/12-107), and several carboxyl-terminal GST-fusion proteins (GST/ 187311) bound directly to LRP/azMR immobilized on nitrocellulose. Based on our observations that carboxyl-terminal constructs may function as noncompetitive inhibitors for '251-amino-terminal construct binding to MHICl cells (Fig. 6) and amino-terminal constructs may function as noncompetitive inhibitors of '2SI-carboxyl-termina1 construct binding (Fig. 71, we propose that the amino-and carboxyl-terminal regions of the 39-kDa protein bind to different sites on LRP/a2MR (arrow 2). We have previously shown that the GST-39-kDa protein inhibits both t-PA and azM* binding to MHICl cells whereas t-PA and azM* inhibit 1251-GST-39-kDa protein binding only slightly (10). These results may suggest that the GST-39-kDa protein inhibits ligand binding indirectly, for example by steric hindrance. Since the amino-terminal region of the 39-kDa protein inhibited both t-PA and azM* binding to MHICl cells, whereas the carboxyl-terminal region only inhibited t-PA binding, it is possible that the amino-terminal region overlaps both the t-PA and azM* binding sites on LRP/a2MR to sterically hinder t-PA and azM* binding (arrow 3 1, whereas the carboxylterminal region overlaps only the t-PA binding site and thus inhibits t-PA binding but not azM* binding (arrow 4). Arrows 5 and 6 demonstrate a possible model for the noncompetitive inhibition by amino-and carboxyl-terminal regions of the 39-kDa protein. Binding of the amino-terminal region to LRP/ a2MR may induce a conformational change in LRP/a2MR such that the carboxyl-terminal region of the 39-kDa protein is unable to bind (arrow 5). Similarly, binding of the carboxyl-terminal region to LRP/a2MR may induce a conformational change in LRP/azMR such that the amino-terminal region of the 39-kDa protein cannot bind (arrow 6).
In summary, we have demonstrated that amino-and carboxyl-terminal regions of the 39-kDa protein each bind specifically to LRP/a2MR on rat hepatoma MHIC1 cells with similar affinities as the full-length protein. Following binding, these regions are rapidly endocytosed and degraded with kinetics 115-319, GST/151-319, GST/187319, GST/200-319, and GST/ identical to the full-length protein. In addition, the amino-and carboxyl-terminal regions bind to separate sites on LRP/a2MR, and this implies that these regions may differentially regulate ligand binding to LRP/a2MR.