Functional Expression of the Chicken Low Density Lipoprotein Receptor-related Protein in a Mutant Chinese Hamster Ovary Cell Line Restores Toxicity of Pseudomonas Exotoxin A and Degradation of α2-Macroglobulin*

The low density lipoprotein receptor-related protein (LRP) is responsible for the clearance of several physiological ligands including a complex of proteinase and α2-macroglobulin (α2M) and for the entrance of Pseudomonas exotoxin A (PEA) into cells. We have prepared expression plasmids for the full-length chicken LRP (designated LRP100) and two intermediates encoding 25 and 67% of the receptor (designated LRP25 and LRP67, respectively) using overlapping cDNA fragments. LRP25 and LRP67 encode the N-terminal 22 and 64%, respectively, of LRP100 plus the transmembrane and intracellular domains. Transient transfection of these plasmids into COS-7 cells yielded recombinant proteins of expected molecular mass and immunoreactivity. However, LRP100 was incompletely processed into α- (515-kDa) and β- (85-kDa) chains and was poorly transported from the endoplasmic reticulum to the Golgi compartment. Stable transformants of LRP100, LRP67, and LRP25 were generated in a mutant Chinese hamster ovary cell line that lacked expression of endogenous LRP and was resistant to PEA. All forms of recombinant LRP proteins were transported from the endoplasmic reticulum to the Golgi apparatus in Chinese hamster ovary cells as shown by their sensitivity to endoglycosidase H and resistance to neuraminidase. Cell surface iodination and subcellular fractionation studies indicated that all three LRP variants were expressed on the plasma membrane. Furthermore, expression of the three LRP variants restored, to various degrees, sensitivity to PEA and the ability to degrade methylamine-activated α2M (α2M*). These data suggest that deletion of large internal portions of LRP, including the processing site, does not prevent transport of LRP to the plasma membrane, nor does it abolish the interaction of LRP with α2M* or PEA. This LRP expression system may allow for the characterization of domains within LRP responsible for its multifunctionality.

The low density lipoprotein receptor-related protein (LRP) 1 of chicken somatic cells has been characterized (1). Comparison of amino acid sequences between the chicken LRP (4522 amino acids) and the human counterpart (4525 amino acids) (2) has revealed 83% sequence identity between the two receptors. LRP is translated as a single polypeptide chain (600 kDa) and subsequently processed into ␣-(515-kDa) and ␤-(85-kDa) chains during its transport from the endoplasmic reticulum (ER) to the cell surface (3). Processing of LRP is catalyzed by the trans-Golgi endopeptidase furin (4) that recognizes the RXRR consensus sequence (RHRR in human LRP and RNRR in chicken LRP). The resulting ␣and ␤-chains are noncovalently associated on the cell surface (3). Intracellular transport of the nascent LRP polypeptide from the ER to the Golgi apparatus requires a 39-kDa chaperone called the receptor-associated protein (RAP) (5,6). Recently, it has been suggested that RAP facilitates LRP folding in the ER and prevents premature association of physiological ligands concomitantly synthesized with the receptor (7,8).
LRP is a type I membrane protein. Its extracellular ␣-chain contains 31 class A ligand-binding motifs, arranged into four clusters (I through IV), and 22 class B epidermal growth factor type repeats. The ␤-chain contains a single membrane-spanning segment followed by two Asn-Pro-X-Tyr motifs at the carboxyl terminus (1,2). Probably because of the structural similarities between chicken and human LRP, both receptors exhibit identical binding to a number of diverse ligands, including lipid-associated apolipoprotein E (apoE), activated ␣ 2 -macroglobulin (␣ 2 M*), vitellogenin, and RAP (1,9). The structural basis for the multifunctionality of ligand binding to LRP is not completely understood. Analyses of recombinant LRP minireceptors (7,11) and proteolytic fragments (10) of human LRP have demonstrated that ligand binding activity is associated mainly with class A motifs in clusters II and IV. Conversely, clusters I and III displayed only weak ligand binding activity. Accumulating experimental evidence suggests that LRP plays an important role in hepatic clearance of chylomicron remnants, a process probably mediated by interplay between many factors including apoE (13,14), hepatic lipase (15), lipoprotein lipase (16), and heparan sulfate proteoglycans (17), in addition to LRP. With the recognition of the possible link between apoE metabolism and the development of Alzheimer's disease, the potential role of LRP in the pathophysiology of the central nervous system has also been suggested (18).
LRP has also been suggested to serve as a receptor for Pseudomonas exotoxin A (PEA) (19,21). PEA consists of three functional domains and exerts its toxicity by irreversibly inhibiting protein translation. Following internalization, PEA protein is cleaved by a cellular protease and translocated to the cytosol. The N-terminal domain of PEA mediates cell binding, while the central domain contains the translocating activity and acts as the substrate for proteolytic cleavage. The C-terminal domain of PEA possesses the enzymatic activity and catalyzes ADP-ribosylation of elongation factor 2, resulting in inhibition of protein synthesis and cell death (22,23). The primary target of PEA is the liver (25,26). A mutant Chinese hamster ovary (CHO) cell line has been isolated that lacks LRP and is resistant to the toxic effects of PEA (20).
In this study, we prepared expression plasmids for the fulllength chicken LRP and two deletion variants using LRP cDNA fragments and expressed them in an LRP-deficient CHO cell line (LRP-null). We used the cytotoxicity of PEA to monitor the function of the recombinant receptor and found that expression of the full-length receptor in the LRP-null cell line restored its sensitivity to the toxin. Furthermore, the expressed chicken LRP also mediated the uptake and degradation of ␣ 2 M*. This expression system will facilitate investigations of the pathophysiology and cell biology of this enormous cell surface receptor.

EXPERIMENTAL PROCEDURES
Materials-DNA restriction enzymes and endoglycosidase H (Endo H) were purchased from New England Biolabs. Neuraminidase was purchased from Boehringer Mannheim or New England Biolabs. All reagents for cell culture were purchased from Life Technologies, Inc. ProMix TM (a mix of [ 35 S]methionine and [ 35 S]cysteine; 1000 Ci/mmol), carrier-free Na 125 I, horseradish peroxidase-conjugated goat anti-rabbit IgG antibody, and the enhanced chemiluminescence (ECL) reagents for immunoblotting were obtained from Amersham Corp. Bicinchoninic acid protein assay reagent and D-Salt Excellulose™ columns were obtained from Pierce. An expression plasmid encoding human furin (pcDNA3hfurin) was a gift from N. Seidah (Clinical Research Institute of Montreal). The expression plasmid pcDNA3RAP that encodes human RAP (5) and a polyclonal antibody against human LRP (27) were gifts from G. Bu (Washington University, St. Louis, MO). Recombinant Pseudomonas exotoxin A was expressed in Escherichia coli and isolated from the periplasm according to procedures described previously (28).
Preparation of Expression Plasmids-The expression plasmids pcLRP25, pcLRP67, and pcLRP100 were constructed by combining 12 chicken LRP cDNA fragments that span the total length of 15598 base pairs (Fig. 1A). Plasmids pcLRP25 and pcLRP67 encode the N-terminal 22 and 64%, of the full-length LRP, respectively, plus the transmembrane and intracellular domains. Inserts were cloned into the polylinker region of the pCMV5 vector (29) positioned between the cytomegalovirus promoter and enhancer sequences and the human growth hormone transcription termination and polyadenylation signals (Fig. 1B). A BamHI-EagI fragment (nucleotides 1229 -4222), prepared from clones A5, E4, and N3 was ligated together with an EagI-BglII fragment (nucleotides 14524 -15248, obtained from clone L-1) into the pCMV5 vector that had been digested with XbaI and BamHI to create pcLRP25. Next, an EagI-EagI fragment (nucleotides 4222-9865) was prepared from clones N1, Z19, and 18 and inserted into pcLRP25 that had been digested with EagI to produce pcLRP67. Finally, an XhoI-XhoI fragment that encoded the C-terminal portion of LRP (nucleotides 9112-15248) plus the hGH region of the pCMV5 vector was prepared by ligation of clones 28, N2, 2B1-2, 112, 2B4, and L-1 and inserted into pcLRP67 to create pcLRP100 encoding the full-length chicken LRP.
Preparation of Cell Extract and Immunoblot Analysis-Transfected cells were washed with phosphate-buffered saline (PBS), resuspended in buffer S (200 mM Tris-maleate, pH 6.0, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 2.5 M leupeptin, and 1.4% Triton X-100) and incubated on ice for 30 min. Insoluble protein was removed by centrifugation (14,000 rpm, 40 min, 4°C) in an Eppendorf microcentrifuge. The Triton X-100-soluble cell proteins were collected from the supernatant, incubated with an equal volume of buffer R (8 M urea, 2% SDS, 10% glycerol, 10 mM Tris-HCl, pH 8.3, and 5% ␤-mercaptoethanol) at 70°C for 15 min, and resolved by electrophoresis on 3-8% gradient polyacrylamide gels containing 0.1% SDS (SDS-PAGE). Proteins were electrophoretically transferred (6 h) onto nitrocellulose membranes for immunoblot analysis. Rabbit antiserum against the carboxyl-terminal 17 amino acids of the ␤-chain of chicken LRP or the C-terminal 15 amino acids of the ␣-chain was used as primary antibody (1) for ECL detection.
Fractionation of Subcellular Membranes-Transfected CHO cells (ten 100-mm dishes) were washed and harvested into PBS. Cells were collected by low speed centrifugation and resuspended in 2 ml of buffer A (20 mM HEPES, pH 7.4, 250 mM sucrose, 1 mM EDTA, 0.1 mM PMSF, 0.1 mM leupeptin, 40 g/ml acetyl-leucyl-leucyl-norleucinal, 10 kallikrein-inactivating units/ml aprotinin) and homogenized by 20 passes through a ball bearing homogenizer (H and Y Enterprise, Redwood City, CA). After centrifugation of the homogenate (16,000 ϫ g, 20 min), the supernatant was subjected to another centrifugation (250,000 ϫ g, 1.5 h) in a TLA100.4 rotor to isolate microsomal membranes. The 16,000 ϫ g pellet was resuspended in 1 ml of buffer B (20 mM HEPES, pH 7.4, 1 mM EDTA, 0.1 mM PMSF, 0.1 mM leupeptin, 40 g/ml acetylleucyl-leucyl-norleucinal, 10 kallikrein-inactivating units/ml aprotinin) using a glass homogenizer (10 strokes of a Teflon pestle), layered on a 1.12 M sucrose cushion, and subjected to centrifugation (100,000 ϫ g, 1 h) in an SW41 rotor. The plasma membrane fraction was collected from the interface of the sucrose cushion and pelleted by centrifugation (30,000 ϫ g, 30 min) in a TLA100.4 rotor. The microsomal and plasma membranes were resuspended in buffer B.
Endoglycosidase H or Neuraminidase Digestion of LRP Proteins-For Endo H digestion, cell extracts (Triton X-100-soluble proteins) or subcellular membrane fractions (in buffer B) were mixed with SDS (final concentration of 0.5%) and ␤-mercaptoethanol (final concentration of 1%) and heated to 100°C for 10 min. The sample was then acidified by the addition of sodium citrate, pH 5.5, to a final concentration of 50 mM. Endo H (500 units) was then added, and the mixture was incubated at 37°C for 5 h. For neuraminidase experiments, cell extract or membrane fraction was acidified by the addition of sodium citrate buffer, pH 6.2, to a final concentration of 50 mM. Neuraminidase (0.75 units) was added, and the mixture was incubated at 37°C for 4 h. Both Endo H and neuraminidase reactions were stopped by the addition of 100 l of buffer R and resolved by SDS-PAGE under reducing conditions. LRP was detected by immunoblotting as described above.
Pseudomonas Exotoxin A Cytotoxicity Assay-Cells were seeded (1 ϫ 10 5 cells/well) in 24-well dishes and allowed to adhere for 24 h. PEA was added to the medium at indicated concentrations, and cells were incubated for up to 18 h. The cells were washed with PBS (37°C) and labeled with [ 35 S]methionine/cysteine (200 Ci/ml) for 1 h in a methionine-and cysteine-free medium. After labeling, cells were washed with cold PBS and lysed with 250 l of lysis buffer (1 mM EDTA, 1% Triton X-100, 1% deoxycholic acid, 1% SDS, 1 mM dithiothreitol, 0.015% PMSF, 50 mM Tris-HCl, pH 8.0). Complete solubilization of cell protein was achieved by heating the samples at 75°C for 30 min. Each lysate was then diluted 10-fold with water, and an aliquot was mixed with an equal volume of fetal bovine serum and spotted under gentle vacuum onto 23-mm nitrocellulose membranes (saturated with ice-cold 10% trichloroacetic acid) using a Millipore 1225 manifold system. Membranes were washed twice with ice-cold 10% trichloroacetic acid and once with water and then dried for liquid scintillation counting.
Cell Surface Iodination-Transfected CHO cell monolayers (80 -90% confluent in 60-mm dishes) were washed twice, and the cells were collected into 1 ml of PBS (pH 6.5). Na 125 I (300 Ci) and two IODO-BEADs (Pierce) were added, and the cell suspension was mixed by rocking for 15 min at room temperature. The cell suspension was separated from the IODO-BEAD, and the cells were pelleted by centrifugation (3000 ϫ g, 2 min). Unincorporated radioiodine was removed by washing the cells three times with PBS. The cell pellet was resuspended in 160 l of buffer I (200 mM Tris-maleate, pH 6.0, 2 mM CaCl 2 , 0.5 mM PMSF, 2.5 M leupeptin) and were lysed with 40 l of 7% Triton X-100 at 4°C overnight. The lysate was then diluted to 0.5 ml with buffer I containing 1.4% Triton X-100 and precleared with protein A-agarose beads (40 l of 50% suspension) for 2 h. After removing the protein A beads, an aliquot of the supernatant was incubated with anti-␣-chain or anti-␤-chain antibody (overnight at 4°C) to precipitate LRP. The immune complex was recovered with protein A-agarose beads, washed extensively with buffer I containing 1.4% Triton X-100, and eluted into buffer R. Following separation by SDS-PAGE, the radiolabeled LRP was visualized by autoradiography.
Preparation of RAP-GST Fusion Protein-The Salmonella japonicum glutathione S-transferase (GST)/39-kDa expression plasmid containing human 39-kDa protein (RAP) cDNA was obtained from D. Strickland. The purification of GST/39-kDa protein from E. coli (DH5␣) and of the 39-kDa protein after thrombin cleavage were carried out essentially as described by Herz et al. (31).
Purification, Iodination, and Uptake/Degradation of ␣ 2 M*-Human ␣ 2 M was purified from fresh plasma (obtained from the Blood Bank of the Ottawa Civic Hospital) by Zn 2ϩ -chelate affinity chromatography as described previously (32). Preparation of ␣ 2 M* was performed according to previously described procedures (12). The purified ␣ 2 M* (100 g in 250 l of PBS, pH 7.3) was iodinated with one IODO-BEAD and 1 mCi of Na 125 I for 15 min. Unincorporated 125 I was removed by passing the reaction mixture over a D-Salt Excellulose column equilibrated with PBS containing 0.1% bovine serum albumin. 125 I-␣ 2 M* (8.9 ϫ 10 3 cpm/ ng) was collected in the void volume of the column effluent. LRP-null and LRP-transfected cells (confluent in 12-well dishes) were incubated with 125 I-␣ 2 M* (2.5 nM, 0.4 ml/well) in the presence or absence of RAP in F-12 medium containing 5 mM CaCl 2 and 6 mg/ml bovine serum albumin for up to 4 h at 4°C for ␣ 2 M* binding or 37°C for ␣ 2 M* degradation. For degradation, the cells were then placed on ice, and the medium was collected into tubes containing ice-cold trichloroacetic acid (final concentration, 20%). After 30 min, trichloroacetic acid-insoluble material was pelleted by centrifugation, and the supernatant was removed for determination of trichloroacetic acid-soluble, non-iodide radioactivity as described previously (33). For ␣ 2 M* binding, the cells were washed twice with cold PBS containing 6 mg/ml bovine serum albumin and twice with cold PBS. The cells were then solubilized with 0.1 N NaOH, and radioactivity was quantified.
Protein Assays-Proteins were determined by the bicinchoninic acid method (Pierce) according to the manufacturer's instructions.

Transient Expression of LRP in COS-7 Cells-
The expression plasmid encoding the full-length chicken LRP (LRP100) was prepared by combining 12 LRP cDNA fragments as indicated in Fig. 1A. Two intermediate plasmids, LRP25 and LRP67, representing ϳ25 and ϳ67%, respectively, of the fulllength receptor, were also generated during the preparation of LRP100 (Fig. 1B). Fig. 2 shows immunoblot analysis of the LRPs treated with and without Endo H ( Fig. 2A) or neuraminidase (Fig. 2B) LRP. LRP25 encodes the N-terminal 984 amino acids (including the 21-residue signal peptide, cluster I, and the first three class A repeats of cluster II) plus the C-terminal 125 amino acids (including the transmembrane domain and the intracellular domain) of LRP100. LRP67 contains the N-terminal 2865 amino acids (including clusters I and II and nine class A repeats of cluster III) plus the transmembrane and intracellular domains.
All three LRP variants exhibited the expected molecular mass in transfected COS-7 cells and reacted with a specific antibody raised against the C-terminal 17 amino acids of the chicken LRP ␤-chain (Fig. 2, anti-␤ chain). Recognition of the full-length recombinant LRP by the anti-␤-chain antibody (right two lanes in Fig. 2, A and B) and the absence of the 85-kDa subunit indicated that the precursor protein was not processed into ␣and ␤-chains. Processing was not expected for LRP25 or LRP67, since they do not contain the furin recogni-tion site (Fig. 2C). However, the native LRP in the chicken liver membrane (cLM lanes in Fig. 2, A and B) and endogenous LRP in COS-7 cells (probed with an antibody against human LRP; data not shown) were fully processed into ␣and ␤-chains. Incomplete proteolytic cleavage of the recombinant LRP100 is most likely attributable to the impaired intracellular transport from the ER to the trans-Golgi network (where furin resides) based upon analysis of the carbohydrate moiety. Thus, while the native LRP in the chicken liver membrane was fully resistant to Endo H and sensitive to neuraminidase (left two lanes in Fig. 2, A and B), all three recombinant LRPs were sensitive to Endo H and resistant to neuraminidase. The high molecular mass species (ϳ200 kDa) found in the LRP25-transfected cells ( Fig. 2A) were probably self-associated dimer. Dimerization of LRP25 and LRP67 was also observed when PAGE was performed under nonreducing conditions (data not shown). We attempted to improve LRP processing by cotransfection of LRP100 with RAP (to enhance ER-to-Golgi transport) or furin. Although RAP and furin expression was increased, processing of LRP100 was not significantly improved (data not shown). In ligand blotting studies, the unprocessed full-length LRP100 and LRP67 expressed by COS cells could bind to Ca 2ϩ , vitellogenin, and RAP (data not shown). However, to demonstrate LRP function at the cellular level, we sought a cell culture system in which proteolytic processing and transport to the plasma membrane could be demonstrated.
Expression of LRP100 in CHO-K1 Cells-We tested several different cell lines and found that CHO-K1 cells were a suitable host for functional expression of recombinant LRP. Stably transfected cells expressing LRP100 were generated in a mutant CHO-K1 cell line that lacked expression of endogenous LRP (Fig. 3, A and B, LRP-null). This LRP-null cell line was resistant to PEA toxicity and was unable to internalize activated ␣ 2 M (20). In LRP100-transfected cells, the majority of LRP protein was processed into ␣and ␤-chains that could be detected in both the microsomes (Fig. 3A) and plasma membrane (Fig. 3B). The unprocessed LRP100 was found in the microsome fraction but was not detectable in the plasma membrane (middle two lanes in Fig. 3, A and B), indicating that the majority of the surface-presented LRP is proteolytically processed. Surface iodination experiments with intact LRP100transfected cells also demonstrated that the processed ␣-chain was the predominant form presented on the cell surface, although a small amount of unprocessed LRP was iodinated in these experiments (data not shown). As expected, the LRP ␣-chain was sensitive to neuraminidase (right two lanes in Fig.  3, A and B). The LRP ␤-chain, like the native ␤-chain (left two Top, cells (ϳ230 g of protein/well) were incubated with 125 I-␣ 2 M* (5.8 ϫ 10 3 cpm/ ng, 20 g/ml) at 37°C, and the trichloroacetic acid-soluble, non-iodide radioactivity in the medium was determined at the indicated times. Bottom, the inhibitory effect of RAP on ␣ 2 M* degradation was determined by co-incubation of 125 I-␣ 2 M* with increasing concentrations of RAP for 4 h at 37°C.

TABLE I Degradation of 125 I-␣ 2 M* and PEA toxicity index in CHO cells transfected with wild-type or truncated mutant chicken LRP cDNAs
The wild-type CHO-K1 cell, mutant LRP-null cell (13-5-1) and stably transfected cells expressing LRP100 (P2B3), LRP67 (P2B4, P2A2, P5A3), or LRP25 (P7B3, P4B2, P6B4) were incubated with 125 I-␣ 2 M* (8.9 ϫ 10 3 cpm/ng) for 2 h at 4°C to determine surface binding or for 4 h at 37°C to determine degradation. lanes in Fig. 2B), was also sensitive to neuraminidase (middle two lanes in Fig. 3, A and B), indicating that the LRP ␤-chain is sialylated. Functional Analysis of Recombinant LRP in Transfected Mutant CHO Cells-We tested if expression of LRP100 in LRPnull cells would restore the toxicity of PEA. Preliminary time course experiments indicated that at 200 ng/ml, PEA exerted maximal inhibitory effect on protein synthesis after a 12-h incubation (data not shown). The effect of PEA dose was assessed in PEA toxicity assays using cells that had been treated with PEA for 18 h (Fig. 4A). In wild-type CHO-K1 cells, incorporation of [ 35 S]methionine/cysteine into cell protein decreased with increasing PEA dose, while LRP-null cells were insensitive to the toxin. In LRP100-transfected cells, expression of the full-length LRP restored the toxicity of PEA to LRP-null cells. The PEA dose required to reduce protein synthesis to 50% of untreated cells (IC 50 ) decreased from Ͼ500 ng/ml in LRP-null cells (19) to 50 ng/ml in the LRP100-transfected cells (Fig. 4A). The IC 50 for wild-type CHO-K1 cells was approximately 25 ng/ml, similar to an earlier observation (19). When the apparent PEA toxicity observed in wild-type CHO-K1 cells and in LRP100-transfected cells was corrected for the number of receptor molecules on the surface (as determined by 125 I-␣ 2 M* binding at 4°C), the chicken LRP gave a value that was twothirds of the endogenous LRP (Table I, sixth column).
Expression of LRP100 also restored the ability to bind, internalize, and degrade ␣ 2 M*. While LRP-null cells were unable to degrade 125 I-␣ 2 M*, LRP100-transfected cells released trichloroacetic acid-soluble, non-iodide radioactivity at a rate similar to the wild-type CHO-K1 cells (Fig. 4B, top). Measurement of the cell-associated radioactivity (at 37°C) revealed that the failure of LRP-null cells to degrade 125 I-␣ 2 M* was attributable to their inability to bind or internalize the ligand (data not shown). Degradation of 125 I-␣ 2 M* by CHO-K1 or LRP100transfected cells could be effectively prevented, in a dose-dependent manner, by RAP (Fig. 4B, bottom) or unlabeled ␣ 2 M* (data not shown). Analysis of the ␣ 2 M* binding and degradation data indicated that the ability of the chicken LRP to degrade ␣ 2 M* was equivalent (87%) to that of the endogenous receptor (Table I, fourth column).
Functional Analysis of Two Deletion LRP Variants-We next tested whether LRP25 and LRP67 could function as receptors for PEA and ␣ 2 M*. In three LRP67-transfected cell lines, two bands representing the sialylated LRP67 (mature form) and the asialyl LRP67 proteins were observed (Fig. 5A). The mature LRP67 was resistant to Endo H digestion (Fig. 5B) and sensitive to neuraminidase digestion (Fig. 5C) and was the major species (in comparison with the asialyl form) in the plasma membrane fraction (Fig. 5, B and C, bottom). Cell surface presentation of the mature form of LRP67 was also demonstrated by surface iodination experiment (data not shown). In contrast, the asialyl form of LRP67 was found predominantly in the microsomal membrane fraction (Fig. 5, B  and C, top) and was sensitive to Endo H digestion (Fig. 5B).
Among the three LRP67-transfected cell lines, clone P2A2 (a high expressor) was as sensitive to PEA as CHO-K1 cells, whereas in clones P2B4 and P5A3 (low expressors) the toxicity of PEA was only partially restored when compared with LRPnull cells (Fig. 6A). The difference in toxicity of PEA between the cell lines is attributable to the differing levels of expression and cell surface presentation of LRP67 among the clones (Fig.  5, A-C). When the high expressor P2A2 and low expressor  Fig. 4A. B, degradation of ␣ 2 M* (left) and the effect of RAP on ␣ 2 M degradation (right), as described in Fig. 4B. TCA, trichloroacetic acid.
P2B4 were tested for their ability to degrade ␣ 2 M*, we found that both clones degraded 125 I-␣ 2 M* and that the ability to degrade 125 I-␣ 2 M* was correlated closely to the level of LRP67 expression (Fig. 6B, left). The LRP67-mediated 125 I-␣ 2 M* degradation could be prevented by RAP (Fig. 6B, right) and unlabeled ␣ 2 M* (data not shown). When the ␣ 2 M* degradation data were corrected for the differing level of expression, the LRP67transfected cells gave results (64 and 124% in two clones) that were comparable with that of LRP100 (87%) ( Table I, fourth  column). Similarly, the sensitivity to PEA observed in LRP67transfected cells (31 and 56% in two clones) was comparable with that in LRP100-transfected cells (66%) ( Table I,  Similar analyses were performed with the LRP25-transfected cells. Three stable transformants that expressed different levels of LRP25 were analyzed (Fig. 7A). In each cell line, the mature form of LRP25 (sialylated) exhibited Endo H resistance (Fig. 7B) and neuraminidase sensitivity (Fig. 7C). In addition, LRP25 was expressed on the plasma membrane as determined by subcellular fractionation (Fig. 7, B and C) and cell surface iodination experiments (data not shown). Two of the LRP25 clones (P4B2 and P7B3) displayed modest sensitivity to PEA, whereas one clone (P6B4) remained resistant to the toxin (Fig. 8A). Expression of LRP25 also restored the ability to degrade 125 I-␣ 2 M*, and the extent of ␣ 2 M* degradation correlated with the level of LRP25 expression (Fig. 8B). The LRP25-mediated binding of 125 I-␣ 2 M* could be abolished by RAP (Fig.  8C). The efficiency of ␣ 2 M* degradation (ϳ30% of normal) and the cytotoxicity of PEA (Ͻ10% of normal) were much lower in cells expressing LRP25 than in cells expressing LRP100 or LRP67 (Table I). DISCUSSION In this study, the full-length chicken LRP was stably expressed in a mutant CHO-K1 cell line that lacks endogenous LRP. Biochemical experiments, together with functional analyses, indicated that the recombinant LRP protein was glycosylated, proteolytically processed into ␣and ␤-chains, and presented on the cell surface as a functional receptor. Thus, we have provided conclusive experimental evidence using expressed recombinant protein that LRP indeed serves as a receptor for ␣ 2 M (34) as well as confirmed that it serves as a gate for receptor-mediated entrance of PEA into cells (18). In addition, analysis of two receptors with internal deletions has revealed that at least part of the sequence elements responsible for the binding of ␣ 2 M, RAP, and PEA may be located within the amino terminus of the receptor.
It has been proposed that LRP contains multiple binding sites for RAP. Existence of RAP-binding sites in clusters II and IV of LRP has been shown by studies using anchored minireceptors (11), soluble LRP fragments (35), or proteinase and CNBr digests of LRP (10). Using anchor-free, soluble LRP fragments that contained each of the four clusters of the class A ligand binding repeats, Bu and co-workers (7) have shown that RAP binds avidly to clusters II and IV and less avidly to cluster III but does not bind to cluster I. They have also shown that there are at least five independent RAP-binding sites within LRP (two in cluster II, one in cluster III, and two in cluster IV) and that RAP binding activity seems to be conferred primarily by the class A motifs (8). The current studies with the anchored minireceptor LRP25 (Fig. 8B) have provided new evidence that a RAP-binding site may reside within the first three class A repeats of cluster II.
The ␣ 2 M binding site has been assigned to cluster II of LRP, but the binding activity may not be conferred solely by the class A repeats. Studies with proteolytic fragments of LRP have suggested that some epidermal growth factor type repeats flanking cluster II of class A motifs may also contribute to binding of ␣ 2 M-proteinase complexes (10). A membrane-anchored minireceptor that contained all eight class A repeats of cluster II but not the neighboring fourth epidermal growth factor repeat did not show binding to ␣ 2 M (11). Our observations of binding and degradation of ␣ 2 M* by LRP25-transfected cells indicate that structural determinants essential for binding of ␣ 2 M* are encoded by the amino-terminal 22% of the LRP molecule, a region that also contains sequence determinants for binding of RAP.
This study is the first attempt to define sequence elements within LRP that are required for the entrance of PEA. Evidence that at low concentrations PEA might enter cells via LRPmediated endocytosis includes (i) binding of PEA to LRP ␣-chain and inhibition of binding by RAP (19) and (ii) resistance to PEA toxicity of cells lacking LRP expression (20,21). The LRP-null cell line (13-5-1) used in the present study was initially selected for its increased (100-fold) PEA resistance and inability to internalize the ␣ 2 M-proteinase complex (20). The current transfection experiments have confirmed that the receptor responsible for PEA entrance certainly is the receptor for ␣ 2 M, and they have also demonstrated that deletion of a significant portion of the LRP molecule including the processing site (in LRP67) does not impair the ability of PEA to exert its toxicity. It is noted that even in cells that express an extremely high level of LRP25 (Fig. 8A), the toxicity of PEA is only restored slightly (Ͻ10%). Therefore, the affinity of PEA for LRP may be a function of the receptor length. The LRP expression system established in the current study should be a suitable model to further define the retrograde trafficking of PEA in cells.
Among all members of the low density lipoprotein receptor gene family, LRP is unique in its post-translational processing into two subunits. Similar processing occurs in the insulin receptor. Defective processing of the insulin receptor is associated with impaired insulin binding (Ͻ10% of normal) in transfected mutant CHO cells (36) and in some insulin-resistant diabetes mellitus patients (37). However, the functional expression of the membrane-anchored LRP25 and LRP67 demonstrated in this study suggests that the proteolytic cleavage of LRP may not be essential for the receptor's function. Abolishing the LRP processing in the two deletion variants does not seem to adversely affect their post-translational modification (e.g. N-glycosylation and sialylation; although in the mutant cell lines screened there appear to be two glycosylated forms of LRP25 and LRP67), ER-to-Golgi transport, cell surface presentation, or endocytosis. Willnow et al. (11) have previously expressed a functional membrane-anchored minireceptor that contained cluster II plus the transmembrane and intracellular domains but lacked the processing site of LRP in ldlA7 cells. The minireceptor bound and degraded RAP and the tissue-type plasminogen activator/plasminogen activator inhibitor-1 complex (11). To date, the physiological significance of post-translational processing of LRP remains unknown.
An important finding of this study is that the chicken LRP cannot be expressed on the cell surface in COS-7 cells. Biochemical analysis of the carbohydrate moiety indicated that the recombinant LRP was retained within the ER and was not processed into ␣and ␤-chains, suggesting that the transport of the chicken receptor to the distal Golgi was impaired. Previous transfection studies of human low density lipoprotein receptor showed that the recombinant receptors could be functionally expressed in COS cells (38,39). The inability of COS cells to transport recombinant LRP onto the cell surface may be attributable to the failure of the cellular trafficking machinery to correctly recognize the chicken receptor, since endogenous LRP was shown to be fully processed into ␣and ␤-chains and was presented on the cell surface. Since the molecular chaperone RAP is essential for LRP folding and also to prevent premature binding of the receptor to its ligands (7,8), we have considered the possibility that RAP might be limiting in COS-7 cells. However, co-expression of human RAP with LRP did not improve trafficking of the receptor. It is unlikely that the lack of an effect of the human RAP expression on LRP trafficking is the result of its inability to react with the chicken LRP, since our in vitro studies demonstrated clearly that the human RAP can effectively abolish uptake and degradation of ␣ 2 M. Currently, the reason for the inability of COS-7 cells to transport recombinant LRP onto the cell surface is unexplained.
In summary, we have established an expression system for LRP, a multifunctional cell surface receptor involved in the catabolism of proteinases and lipid-associated proteins. The availability of an in vitro LRP expression system will assist in the identification of structural determinants that are responsible for the multiligand binding activity of LRP and will also facilitate investigations of the involvement of LRP in the development of premature atherosclerosis and Alzheimer's disease.