Purified aa-Macroglobulin ReceptorLDL Receptor-related Protein Binds Urokinase. Plasminogen Activator Inhibitor Type- 1 Complex EVIDENCE THAT THE a2-MACROGLOBULIN RECEPTOR MEDIATES CELLULAR DEGRADATION OF UROKINASE RECEPTOR-BOUND COMPLEXES*

Complexes between 125I-labeled urokinase-type plasminogen activator (uPA) and plasminogen activator inhibitor type-1 (PAI-1) bound to purified alpha 2-macroglobulin (alpha 2M) receptor (alpha 2MR)/low density lipoprotein receptor-related protein (LRP). No binding was observed when using uPA. The magnitude of uPA.PAI-1 binding was comparable with that of the alpha 2MR-associated protein (alpha 2MRAP). Binding of uPA.PAI-1 was blocked by natural and recombinant alpha 2MRAP, and about 80% inhibited by complexes between tissue-type plasminogen activator (tPA) and PAI-1, and by a monoclonal anti-PAI-1 antibody. In human monocytes, uPA.PAI-1, like uPA and its amino-terminal fragment, bound to the urokinase receptor (uPAR). Degradation of uPAR-bound 125I-uPA.PAI-1 was 3-4-fold enhanced as compared with uncomplexed uPAR-bound uPA. The inhibitor-enhanced uPA degradation was blocked by r alpha 2MRAP and inhibited by polyclonal anti-alpha 2MR/LRP antibodies. This is taken as evidence for mediation of internalization and degradation of uPAR-bound uPA.PAI-1 by alpha 2MR/LRP.

As expected for a GPI-anchored receptor (23), uPAR-bound uPA and ATF are poorly internalized and degraded. Surprisingly, uPA a PAI-1 complex, when added in solution or formed on the cell surface by the addition of PAI-1 to uPAR-bound uPA, is readily internalized and degraded in lysosomes. These results were obtained with the following aeMR/LRP-bearing cells: JAR cells (13), monocytoid U-937 cells (24), and bloodderived monocytes (21). Analogous results were obtained in monocytes with uPA .plasminogen activator inhibitor type-2 (25). On the other hand, LAK cells, not expressing a2MR/ LRP, degrade uPA.PA1-1 complex poorly (21). These observations led us to investigate whether a2MR/LRP might be involved in binding and endocytosis of uPA. PAI-1 complex.
Low M, uPA was from Green Cross Corporation, Osaka, Japan. ATF was a gift from Dr. J. Krause, Thomae GmbH, Biberach, Germany. Human PAI-1 was purified and cleaved in the reactive center as described (26). Complex between PAI-1 and "T-uPA ('251-~PA.PAI-1) was prepared as described previously (13). Complex between tissue type plasminogen activator (tPA) and PAI-1 was purified from conditioned medium of cocultured Bowes and HT-1080 cells by sequential immunoaffinity chromatography on columns directed against tPA and PAI-1.
were iodinated following previously published procedures. The specific activities were about 6 X 10l6 Bq/mol. tr2MRAP Cloning, Expression, and Purification-The coding reading frame corresponding to mature n'MRAP (nucleotide position 116-1087) was produced by first-strand cDNA synthesis on total RNA from JAR cells (27) and ensuing PCR amplification by standard methods. The amplified 1-kb n2MRAP DNA fragment was cloned into the expression plasmid pT;-PL (28,29). The resultingexpression plasmid pT7HeFXn,MRAP was used to transform Escherichia coli HL21 cells which upon induction (30) produced (U~MRAP protein linked to an NH2-terminal fusion tail with the amino acid sequence MGSHHHHHHSIEGR. The hybrid n2MRAP protein was solubilized in 6 M guanidinium-chloride, 50 mM Tris, 0.1 M dithioerythritol, pH 8.0, and purified by Sephadex G25 gel filtration in 8 M urea, 0.5 M NaCI, 50 mM Tris, 1 mM methionine, pH 8.0, and affinity chromatography on an NP'NTA column (31). The NH2-terminal fusion tail was cleaved off the purified and folded n2 MRAP with FX. (32). FX. was removed with Sepharose-immobilized bean pepsin-trypsin inhibitor. The cleaved fusion tail and uncleaved fusion protein was removed with NPNTA.
Antibodies-The anti-human a2MR antibodies have been described (9, 12). Monoclonal anti-human PAI-1 IgG from hybridoma clone 3 was that described by Nielsen et al. (33). The anti-PAI-1 from clone 7 was the result of a separate immunization.' The anti-human uPA antibodies, clones 2 and 6, have been described (34).
Blotting Procedures-SDS-PAGE, blotting of proteins, incubation o f strips with labeled ligands, and autoradiography followed previously published procedures (9, 21).
Incubations with Purified Receptor-Microtiter wells from NUNC (Denmark) were coated for 2 h a t 20 "C to provide approximately 25 fmol of two-chain tr2MR/LRP per well. After wash and blocking with 10% BSA for 2 h, incubations (100 pl final volume) were performed for 16 h a t 4 "c in 140 mM NaCI. 10 mM Hepes, 2 mM CaCI', 1 mM MgC12, 2% BSA, pH 7.8. Following wash, bound radioactivity was removed by the addition of 2 X 100 pl of 10% SDS. In the absence of immobilized tr,MR/LRP, apparent binding was 0.3-0.4%, and the measurements were corrected accordingly.
Whole Cell Incubations-Human monocytes were prepared and incubated as described (14,20). Cells were acid-treated in 50 mM glycine-HCL, 100 mM NaCI, pH 3.0, a t 0 "C for 3 min (24). Degradation of labeled ligands was assessed by measuring radioactivity soluble in 10% trichloroacetic acid. Non-cell-mediated degradation in monocyte-conditioned medium was about 3 1 after a 5-h incubation, and the measurements were corrected accordingly (21).
The displayed experiments were performed a t least three times. The results are presented as mean values & S.D. Fig. 1 (lane 1 ) shows the SDS-PAGE-resolved components of the human placental a,MR/LRP preparation. Lane 2 shows binding of the established ligand a,MRAP to the a-chain. Binding of labeled natural and ra2MRAP were not distinguishable (not shown). Labeled uPA.PAI-1 (lane 3), but not ATF (lune 4 ) , bound to the a-chain. Lanes 5 and 6 show that a2MRAP and uPA. PAI-1 bound at the position of the a2MR/ LRP a-chain in SDS-PAGE resolved placental membranes, but not at other positions. Fig. 2 illustrates that immobilized two-chain (inset, compare with Fig. 1, lane 1 ) a,MR/LRP, at the given receptor density, bound 22% of the added l2'1-uPA. PAI-1 as compared with 32 and 25% for the established ligands a,M-methylamine and a2MRAP. At lower receptor density, uPA. PAI-1 binding was higher than a,M-methylarnine binding (not shown). The reason is that the uPA-PAI-1 binding increases linearly," but a?M-methylamine increases near-exponentially (9), as a function of receptor density. The polyclonal antibody raised against the a2MR preparation inhibited 'Z'I-a2MRAP and "'I-uPA.PAI-1 binding by 70-75%. Binding of "'I-uPA and "'I-ATF was minimal or absent.

RESULTS
These results raised the question whether the established ligands compete for binding. Fig. 3 shows that 100 nM natural aaMRAP or ra2MRAP blocked the binding of "'I-uPA. PAI-1. Control experiments (not shown) with immobilized a2MRAP demonstrated absence of "'I-uPA. PAI-1 binding to this protein. The binding of uPA.PAI-1 to a,MR/LRP was sensitive to EDTA and heparin. a2M-methylamine (100 nM) did not inhibit. ATF and low M, uPA had no effect, whereas 100 nM uPA inhibited slightly. Fig. 3 also shows that tPA. PAI-1 complex, but not tPA, inhibited 76%, and reactive center-cleaved PAI-1 inhibited 55%, suggesting that the PAI-1 moiety might contribute to the binding. This hypothesis was supported by the finding that one anti-PAI-1 monoclonal antibody inhibited 85%. The monoclonal anti-uPA antibody against the COOH-terminal serine proteinase domain inhibited 77%, whereas no inhibition (not shown) was observed with the anti-uPA antibody, clone 6, inhibiting binding of uPA to uPAR (21).
The functional consequence of uPA. PAI-1 binding to a,MR/LRP was explored in monocytes. Table I shows that binding of labeled uPA. PAI-1 a t 4 "C was blocked by unlabeled ATF, but not by ra2MRAP, confirming binding to uPAR (21,24). Under these conditions '251-ATF was 14% bound (not shown). Conversely, binding of labeled ra2MRAP was blocked by unlabeled raaMRAP, but not by ATF. Acid treat-

Rinding of uPA . PAI-I and rN2MRAP to monocytes at 4 "c
Monocytes (1 X lO'/ml) were incubated with the 12"I-labeled ligands (1.5-20 PM) with or without 100 nM unlabeled ligand for 16 h a t 4 "C, pelleted, and counted for radioactivity. Some cells, incubated with the tracers alor.e, were subjected to acid treatment (3 min, 0 " C )  Fig. 4 shows the results of incubation experiments with '"I-uPA.PA1-1 and I2"I-ATF for 5 h a t 37 "C. The upper panel shows a higher cell-associated radioactivity when using ATF, presumably reflecting the low turnover of this tracer. Unlabeled ATF inhibited the cell association of both tracers. Unlabeled ra2MRAP caused no major changes. The inset shows that binding of ra2MRAP occurs specifically to anMR/ LRP in solubilized, SDS-PAGE-resolved monocyte membranes. The radioactivity remaining associated to monocytes incubated a t 37 "C was about 90% acid-releasable (not shown) and represented a t least 90% authentic ligand as judged by SDS-PAGE and the ability to rebind to cells (21). Fig. 4  (lower panel) shows degradation of 25% of the '2sI-uPA. PAI-1 and "'1-ra2MRAP present in monocyte suspensions by 5 h, as compared with 3% for '""I-ATF. Under this condition, I2,'I-uPA was degraded 6-7% (not shown). Similar results, including the time courses of degradation, have been reported for JAR cells (13) and U-937 cells (24). Unlabeled ATF blocked l'JI-ATF degradation and strongly inhibited '251-uPA. PAI-1 degradation. Unlabeled ra,MRAP had no effect on "'I-ATF degradation. However, ra2MRAP inhibited "'I-uPA. PAI-1 Additions: -ATF ra2MRAP AT+F Ab The inset shows the presence of n2MR/LRP and uPAR in Triton X-114 detergent phase extract of 7 days cultured monocytes; lane I , ligand blot showing binding of '"I-re2MRAP; lane 2, binding of ""I-ATF. Louer panel, the bars show the percentage of degraded tracers in the monocyte suspension after 5 h. The inhibition by the polyclonal anti-n2MR/LRP antibody (Ab) was corrected for the effect of nonimmune IgG (8% inhibition). The trichloroacetic acid-precipitable radioactivity in the medium (not shown) accounted for the part of the added radioactivity which was not cell-associated or degraded. and '*'I-ra2MRAP degradation to about 7%. When added together, ATF and rarMRAP essentially abolished the cellular degradation of "'I-uPA. PAI-1 complex. Polyclonal anti-a2MR/LRP antibodies inhibited the degradation of complex in the monocyte suspension to 12%.
These data suggested that uPAR-bound uPA. PAI-1 can be degraded via binding to a2MR/LRP. The following experiment supports this hypothesis. '"I-uPA and '"I-uPA-PAI-I were prebound to monocytes a t 4 "C and washed. ra2MRAP was present during the incubation a t 4 "C to prevent any minor binding to a2MR/LRP. Wash was in the presence of EDTA since uPA-PAI-1 binding to a2MR/LRP (Fig. 3), but not to uPAR;' is Ca"-dependent. The cells were then transferred to 37 "C. Fig. 5 shows that complexing with PAI-1 caused a 3.2-fold increase in degradation of uPAR-bound "' 1-uPA by 4 h in agreement with the previously described time courses (21). This inhibitor-enhanced degradation was essentially blocked by razMRAP (Fig. 5).

DISCUSSION
The data show that uPA.PAI-1 complex can bind to the purified a2MR/LRP, while uPA alone is unable to do so. This Effect of r(r2MRAP on the degradation of "'I-uPA.PA1-1 prebound to monocyte uPAR. Monocytes (2 X IO6/ ml) were incubated for 16 h at 4 "C with 20 pM tracer in the presence of 100 nM ra2MRAP followed by wash (4 "C, 10 min) in the presence of 10 mM EDTA. Cells with prebound '251-~PA (left) or '251-~PA. PAI-1 (right) were transferred to 37 'C and incubated in buffer for 4 h followed by measurement of cell-associated radioactivity (column 1 ) and radioactivity released into the medium in trichloroacetic acidprecipitable (column 2 ) or acid-soluble form (column 3 ) . Column 4 shows the degradation of prebound tracer to acid-soluble radioactivity when 100 nM ra2MRAP was present in the 4-h incubation at 37 "C. The cell-associated radioactivity and that released in acid-precipitable form were at least 90% authentic ligand as judged by SDS-PAGE.
binding seems to involve the PAI-1 moiety, as judged by the inhibitions with monoclonal anti-PAI-1 antibody, reactive center-cleaved PAI-1, and tPA.PA1-1 complex, but does not involve the NHz-terminal receptor-binding domain of uPA.
It is remarkable that azM-methylamine did not inhibit the binding of uPA. PAI-1 to the purified receptor. Previous experiments have shown that receptor-active azM is a poor inhibitor of azMRAP binding to purified azMR/LRP (9). In fibroblasts, azM inhibits binding of apoE-activated P-VLDL, but only at high concentration, suggesting that competition might result from steric hindrance (35). On the other hand, binding of uPA.PAI-1, azM complex (9), and apoE-activated P-VLDL (36) is effectively inhibited by azMRAP. The picture emerges that aZMRAP binds to multiple sites and inhibits binding of several ligands interacting with different domains of the multifunctional azMR/LRP.
In experiments with monocytes we found that primary uPA.PA1-1 binding is to uPAR (blocked by ATF), and that azMRAP, the competing ligand on azMR/LRP, blocked the inhibitor-enhanced degradation. In addition, polyclonal anti-azMR/LRP antibodies inhibited this degradation. We therefore propose that in monocytes uPA.PAI-1, initially bound to the high affinity uPAR, is internalized and degraded following interaction with a2MR/LRP. The GPI-anchored and mobile (23) uPAR concentrates uPA.PA1-1 complexes on the cell surface and may thereby facilitate its binding to azMR/ LRP via a separate domain. Future studies should show whether azMR/LRP might serve as a receptor for proteinasecomplexed serpins other than PAI-1.
A presentation of uPA. PAI-1 by uPAR for binding to azMR/LRP would, to our knowledge, be the first example of a receptor for an adjacent receptor-bound ligand. It is unknown whether uPA. PAI-1 will dissociate from uPAR following binding to the secondary receptor. In any case, azMR/ LRP would provide a means for internalizing and targeting uPAR-bound uPA. PAI-1 to lysosomes (13). Possibly, other GPI-anchored receptors may be linked to secondary receptors through their ligands.
The cDNA of a2MRAP is preceded by a hydrophobic sequence, which appears to be a somewhat atypical cleaved signal peptide, and a2MRAP is observed on the surface of human fibroblasts (10). We have been unable to detect this peptide in the blood and in conditioned medium from cells containing abundant intracellular (Y~MRAP.~ This peptide may cross the plasma membrane bound to azMR/LRP or, after secretion, become rapidly and efficiently bound to a multitude of sites on a2MR/LRP. In either case, a2MRAP will modulate the efficiency of internalization of other ligands via the constitutively endocytosing a2MR/LRP. azM. proteinase and related complexes constitute only one out of several ligand families transported over the plasma membrane by a2MR/LRP-mediated endocytosis. The present names of this receptor do not reflect its function as a clearance receptor for multiple ligands. We propose the name Charon Receptor (ChaR) after the ferryman who, according to Greek mythology, sailed the souls across the river Styx to Hades.