Dissection of receptor folding and ligand-binding property with functional minireceptors of LDL receptor-related protein

Members of the LDL receptor (LDLR) gene family are cysteine-rich endocytic receptors. In mammals, seven LDLR family members have been described and variably characterized (Hussain et al., 1999; Krieger and Herz, 1994). Among these, the LDLR-related protein (LRP) and megalin are extremely large in size (approx. 600 kDa) and each binds over ten structurally and functionally distinct ligands (Herz et al., 1988; Saito et al., 1994). These ligands are mostly circulating plasma proteins that range from lipoprotein particles to protease and protease/protease inhibitor complexes. While members of the LDLR family bind overlapping ligands, they appear to exhibit distinct tissue and cellular expression patterns (Krieger and Herz, 1994; Zheng et al., 1994). In addition to mediating ligand uptake and degradation, several of these receptors have been shown to mediate signal transduction (Goretzki and Mueller, 1998; Trommsdorff et al., 1998; Misra et al., 1999; Trommsdorff et al., 1999).

A unique feature of the LDLR family members is that ligand binding to these receptors is universally antagonized by a soluble receptor-associated protein (RAP) (Bu, 1998; Strickland et al., 1991). Using LRP as a model receptor, studies have shown that, under physiological conditions, RAP associates with LRP within the early secretory pathway and keeps the receptor in a ligand-binding inactive state (Bu et al., 1995; Bu and Schwartz, 1998; Willnow et al., 1996). This function is presumably to ‘safeguard’ the receptor during its trafficking through the early secretory pathway such that premature binding to ligands is prevented. Another function of RAP as a chaperone for LRP is to promote proper folding of the receptor following its translation within the endoplasmic reticulum (ER) (Bu and Rennke, 1996; Bu and Schwartz, 1998; Obermoeller et al., 1998). Although the mechanisms underlying RAP’s role as a folding chaperone are still unclear at present, it is likely that binding of RAP to LRP prevents the formation of mislinked intra- and intermolecular disulfide bonds (Bu and Rennke, 1996; Obermoeller et al., 1998).

The molecular nature of ligand interactions with LRP and megalin are still not clear. However, each of these two receptors contains four putative ligand-binding domains, each consisting of a cluster of ligand-binding repeats. These repeats share homology with cysteine-rich complement proteins and each contains three disulfide bonds. Previous studies have shown that RAP binds with high affinity to domain II and IV, and with low affinity to domain III, of soluble LRP minireceptors, and promotes their folding and secretion (Bu and Rennke, 1996). Similar interactions of RAP with LRP domains have been demonstrated via solid phase analyses (Neels et al., 1999). Using LRP proteolytic fragments and LRP minireceptors, it was shown that domain II of LRP may contain binding sites for tissue-type plasminogen activator (tPA) complexed with plasminogen activator inhibitor type 1 (PAI-1) as well as for α₂-macroglobulin (α₂M) (Moestrup et al., 1993; Willnow et al., 1994). Finally, a recent study using secreted soluble fragments of LRP and surface resonance binding assays has shown that many of the LRP ligands bind domain II and IV with similar affinity (Neels et al., 1999). Since several LRP ligands also interact with other cell-surface proteins such as heparan sulfate proteoglycans (HSPG) and the urokinase receptor, it is not clear whether solid-phase analysis with purified LRP fragments reflects the physiological state of these LRP domains in terms of ligand binding when they are present on the cell surface. The primary aims of the current study were to generate functional membrane-containing LRP minireceptors, which are expressed on the cell surface, to map ligand-binding domains for a variety of LRP ligands, and to systematically examine the role of RAP in LRP domain folding. We found that LRP minireceptors that exhibit high affinity for RAP also depend on this chaperone for proper folding. In addition, we show that different domains of LRP exhibit both overlapping and distinct ligand-binding properties.

The method for constructing membrane-containing LRP minireceptors (mLRPs) has been described previously (Bu and Rennke, 1996; Obermoeller et al., 1998) and is illustrated in Fig. 1. Briefly, human LRP cDNA (Herz et al., 1988) was used as the template for polymerase chain reaction (PCR). Each mLRP was constructed in three consecutive steps by subcloning PCR products into vectors. The first PCR product includes the signal peptide, the first five amino acids of the mature protein, and a nine-amino acid hemagglutanin (HA) epitope (Bu and Rennke, 1996), and was subcloned into pcDNA3 vector (Invitrogene). The second PCR product represents each of the four ligand-binding domains (residues 6-94 for mLRP1; 787-1164 for mLRP2; 2462-2923 for mLRP3; and 3274-3764 for mLRP4; see Herz et al., 1988, for amino acid numbering). The third PCR product includes the entire region from immediately after domain IV to the carboxyl terminus of LRP (residues 3765-4525). To facilitate cloning, a BamHI restriction site that encodes amino acids glycine and serine was inserted between the first and second PCR products, and an EcoRI site that encodes amino acids glutamate and serine was inserted between the second and the third PCR products. All oligonucleotides were synthesized at Washington University School of Medicine Protein Chemistry Laboratory. All constructs derived from PCR were verified by DNA sequencing.

Human glioblastoma U87 cells were cultured in MEM supplemented with 10% fetal calf serum and antibiotics (Bu et al., 1994). Wild-type CHO-K1 and CHO-LRP null cells (FitzGerald et al., 1995) were cultured in Ham’s F-12 medium supplemented with 10% fetal calf serum and antibiotics.

U87 cells at 20-30% confluence were transiently transfected with various mLRP plasmids with cotransfection of either pcDNA3 or pcDNA-RAP (Bu et al., 1995) using a calcium phosphate precipitation method (Bu et al., 1995; Obermoeller et al., 1998). Initial transfections were performed in 10 cm dishes using 40 μg of DNA in a total volume of 10 ml medium. 16 hours after the start of transfection, cells were washed with medium and cultured continuously for an additional 4 hours before being split into multiple 6-well dishes (3.5 cm in diameter) for various chase points. For stable transfection, CHO-LRP null cells were transfected at 20-30% confluence in 10 cm plates with various mLRP plasmids using the calcium phosphate precipitation method. 16 hours after the start of transfection, cells were washed and cultured for an additional 4 hours before being split into four 15 cm plates. Stably transfected cells were selected using 700 μM G418 (Sigma) and maintained with 350 μM G418. Stably transfected cell lines were utilized within the first ten passages following stable selection.

Metabolic labeling with [³⁵S]cysteine was performed essentially as described before (Bu et al., 1995). For pulse-chase experiments, cells were generally pulse-labeled for 30 minutes with 200 μCi/ml [³⁵S]cysteine in cystine-free medium, and either lysed directly without chase or chased with serum-containing medium for 120 minutes. Cells were lysed in ice-cold PBSc (phosphate-buffered saline supplemented with 1 mM CaCl₂, and 0.5 mM MgCl₂) containing 1% Triton X-100, 1 mM PMSF and 10 mM N-ethylmaleimide (NEM). ‘Cell lysates’, as utilized in the current manuscript, are ‘Triton X-100 extracts’.

Rabbit polyclonal anti-LRP (generated against purified human LRP) and anti-RAP (generated against recombinant human RAP) antibodies have been described before (Bu et al., 1995). Monoclonal anti-HA antibody was obtained from BabCo (12CA5). Immunoprecipitations were carried out essentially as described before (Bu et al., 1995), except the washing buffer for monoclonal anti-HA antibody contained 0.1% SDS instead of 1% SDS. Preliminary experiments were performed to ensure that the primary antibody used in each immunoprecipitation was in excess. Protein A-agarose beads (Repligen) were used to precipitate protein-IgG complexes. The immunoprecipitated material was released from the beads by boiling each sample for 5 minutes in Laemmli sample buffer containing 62.5 mM Tris-HCl, pH 6.8, 2% (w/v) SDS and 10% (v/v) glycerol (Laemmli, 1970). If the immunoprecipitated material was analyzed under reducing conditions, 5% (v/v) β-mercaptoethanol was included in the Laemmli sample buffer. For western blot and RAP-ligand blotting, cells were lysed as described above and 50 μg of cell lysates applied to each lane for SDS-PAGE. The percentage of SDS-polyacrylamide gels is indicated in each figure legend. Rainbow molecular mass markers (Amersham) were used as the molecular mass standards.

For cell-surface mLRP analysis, living cells were used (Li et al., 1997). Briefly, CHO cells expressing mLRPs were detached by incubation with non-enzymatic cell dissociation solution (Sigma). Successive incubations with affinity-purified anti-HA IgG (25 μg/ml) and goat antimouse IgFITC (Becton Dickinson) were carried out at 4°C for 45 minutes. Background fluorescence intensity was assessed in the absence of primary monoclonal antibody. The antibody binding capacities were evaluated from the standardized Quantum Simply Cellular bead calibration plot (Zagursky et al., 1995). The bead standards (Flow Cytometry Standards Corporation) consist of four populations of microbeads coated with goat antimouse antibody, which bind different numbers of mouse IgG monoclonal antibody molecules (5686, 18,329, 50,908 and 150,477 molecule binding capacities) in addition to a blank population. The beads were stained in the same way as the CHO cells.

RAP, tPA (Bu et al., 1992), tPA: PAI-1 complexes (Bu et al., 1992), methylamine-activated α₂M (Bu et al., 1992), scuPA (Li et al., 2000), and TFPI (Warshawsky et al., 1994) (50 μg each) were iodinated by using the IODO-GEN method as described previously (Bu et al., 1992). Stably transfected CHO cells were plated in 12-well plates at a density of 2×10⁵ cells/well and used after overnight culture. Cells were rinsed twice in prewarmed (37°C) ligand-binding buffer, and individual ¹²⁵I-labeled ligand was added to the concentration as indicated in each experiment, in the absence or presence of excess unlabeled RAP (500 nM). Following incubation for 4 hours at 37°C, the overlying buffer was removed and subjected to precipitation with

10 mg/ml BSA and 20% trichloroacetic acid. Degradation of radioligand was defined as the appearance of radioactive fragments in the overlying buffer that were soluble in 20% trichloroacetic acid (TCA). Nonreceptor-mediated degradation was determined in the presence of excess RAP and was subtracted from the total degradation. The protein concentration of each cell lysate, as well as cell numbers for each cell line, were measured in parallel dishes that did not contain LRP ligands.

CHO-LRP null cells expressing each mLRP or the empty pcDNA3 vector were maintained in RPMI1640, 5% FBS and G418 (400 μg/ml). Wild-type CHO-K1 cells were grown in the same medium without G418. To measure toxicity, cells were seeded (1×10⁵/well) in a 24-well plate and then incubated with increasing concentrations of Pseudomonas exotoxin A (PEA) for 18 hours. At the end of the incubation period, protein synthesis was determined by pulsing cells with [³H]leucine (3 μCi/ml) for 1 hour. Cell monolayers were then washed free of unincorporated leucine, precipitated with TCA (12% w/v), solubilized with NaOH (0.1 N) and counted in a Beta counter. Toxin treatments were in triplicate, where one standard deviation was usually less than 5% of the total counts. Results are expressed as ‘percentage of control’ where toxin-treated cells were compared with cells receiving no toxin.

The extracellular subunit of LRP contains four clusters of ligand-binding repeats. Although these domains have been referred to as ‘putative ligand-binding domains’, the contribution of each of these domains to ligand binding has not been systematically examined on the cell surface. Thus, we generated membrane-containing mLRPs representing each of the four ligand-binding domains. These mLRPs (referred as mLRP1, mLRP2, mLRP3 and mLRP4; see Fig. 1), which contain the identical carboxyl-terminal region to that of endogenous LRP, include the furin-cleavage site (Herz et al., 1990; Willnow et al., 1996), and extend from immediately after domain IV to the carboxyl terminus of the receptor. To facilitate immunodetection, an HA epitope was included near the amino terminus of each mLRP. Following furin cleavage, each mLRP is processed into two subunits. The carboxyl terminal subunit is identical to the 85 kDa subunit of the endogenous full-length LRP (Herz et al., 1990), whereas the amino-terminal subunits represent each of the four ligand-binding domains (LBDs).

To examine the folding and trafficking of these mLRPs, we transiently transfected their cDNAs into U87 cells, with cotransfection of either empty vector or pcDNA-RAP (Bu et al., 1995). Cells were then pulse-labeled for 30 minutes with [³⁵S]cysteine, and either lysed directly or chased for 120 minutes prior to lysis (Fig. 2). Following immunoprecipitation with anti-HA antibody, mLRPs were analyzed via SDS-PAGE under either nonreducing (Fig. 2A) or reducing (Fig. 2B) conditions. Misfolding of mLRPs was observed following 120 minutes chase under nonreducing conditions (Fig. 2A) for mLRP2 (lane 6), mLRP3 (lane 10), and mLRP4 (lane 14). These misfolded mLRPs migrated as large molecular size ‘aggregates’ on SDS-PAGE. Although formation of these ‘aggregates’ could be due to other protein features and/or other post-translational events, they probably were the result of mislinked intermolecular disulfide bonds since they were reduced to monomeric receptor bands under reducing conditions (comparing the corresponding lanes in Fig. 2A,B). Coexpression of RAP prevented the formation of misfolded receptors for each mLRP (Fig. 2A, lanes 8, 12 and 16). No misfolding was seen for mLRP1 either in the absence or presence of RAP coexpression. Under reducing conditions (Fig. 2B) following a 120 minute chase, four distinct bands are visible: the precursor ‘ER form’, the fully glycosylated ‘Golgi form’, and the furin-processed LBD and 85 kDa subunits. Since a portion of the ‘Golgi form’ may also be present on the cell surface, they can also be referred to as ‘plasma membrane form’. The LBD and 85 kDa subunits migrated in close proximity under nonreducing conditions (Fig. 2A). LBD1 has a small molecular size (approx. 40 kDa; see Fig. 3) and is not visible in these gels. These results suggest that proper folding of domain II, III and IV, but not domain I, of LRP is facilitated with coexpression of RAP.

In order to evaluate the contribution of each LRP domain to ligand binding, we generated stably transfected cell lines expressing each of the four mLRPs. To exclude any potential contribution from endogenous LRP to ligand binding, we utilized a CHO cell line that lacks endogenous LRP (CHO-LRP null). This cell line was derived from CHO-K1 cells via mutagenesis and selection for its resistance to PEA (FitzGerald et al., 1995). To compare the expression of LRP and potentially other members of the LDLR family in CHO-K1 and CHO-LRP null, we prepared cell lysates from these cells, which were used either for western blotting with anti-LRP antibody (Fig. 3A) or ligand-blotting with RAP (Fig. 3B). As seen in the figure, LRP (515 kDa subunit) is present in CHO-K1, but not in CHO-LRP null cells. Ligand-blotting with RAP gave only the LRP 515 kDa band in CHO-K1 cells (Fig. 3B), suggesting that other members of the LDLR family are not significantly expressed in CHO-K1 and CHO-LRP null cells. Following generation of all the stable cell lines for further analysis, we selected one stable line for each mLRP that exhibited a similar level of expression. To compare the expression of these mLRPs at steady state, we performed western blot analysis with cell lysates using anti-HA antibody (Fig. 3C). As seen in the figure, mLRPs are detected in each stably transfected cell line. The most significant bands detected are the ER forms and LBDs. The Golgi forms are also visible for mLRP2 and mLRP3. The 85-kDa subunit is not detected in this analysis since the HA epitope is present at only the amino termini of these mLRPs. To detect LBD1, we analyzed mLRP1 with a higher percentage polyacylamide gel (Fig. 3C, right). LBD for mLRP1 is detected as approx. 40 kDa band. Thus, each of the four mLRPs was abundantly expressed in the stable cell lines with different degrees of processing.

To examine the folding and processing of mLRPs in stably transfected CHO-LRP null cells, we performed pulse-chase analyses with these cell lines similar to the experiment described in Fig. 2. We suspected that RAP would be limiting for the folding of mLRPs since only the endogenous RAP in CHO-LRP null cells would be available. However, as endogenous LRP and other RAP-binding receptors are not expressed in these cells (see Fig. 3), endogenous RAP was expected to be available exclusively for the stably expressed mLRPs. Results from a representative experiment are shown in Fig. 4. As seen in the figure, each of the four mLRPs was variably processed following 120 minutes of chase. To compare the folding and processing of mLRPs directly, we quantitated the post-ER forms as the percentage of total radiolabeled receptors. We found that mLRP4 is processed up to 54.3% after 120 minutes chase (Fig. 4D), whereas mLRP2 (Fig. 4B) and mLRP3 (Fig. 4C) were processed at only 19.8% and 16.0%, respectively. The poor processing of mLRP2 and mLRP3 is probably due to a higher degree of dependency of these two mLRPs on RAP coexpression. The processing of mLRP1 (Fig. 4A) is only moderate (29.9%) compared to other mLRPs, although the folding of this mLRP is not affected by the level of RAP expression (see Fig. 2).

To quantitate the steady-state cell surface expression of mLRPs in stably transfected cells, we performed flow cytometric analyses in intact cells using anti-HA antibody. Shown in Fig. 5 are results from the representative cell lines examined in the experiments described in Figs 3 and 4. As seen in the figure, CHO cells transfected with only the pcDNA3 vector yield no specific staining with anti-HA antibody, whereas specific signals were detected for each of the four mLRP cell lines. Using Quantum Simply Cellular bead standards (Zagursky et al., 1995) (Fig. 5B), the numbers of cell-surface mLRPs per cell were assessed (Fig. 5C). Higher numbers of cell-surface mLRP4 (77,622 receptors/cell) were found compared to other mLRPs (21,682 for mLRP1, 32,488 for mLRP2 and 32,826 for mLRP3, respectively), consistent with better processing for mLRP4 (see Fig. 4). It is important to point out that there may also be differences in the trafficking patterns of various LRP minireceptors along the exocytic pathway since mLRP1, which is folded better than mLRP2 and mLRP3 (see Fig. 4), exhibited the least cell surface expression.

To analyze the functionality of cell-surface expressed mLRPs, we examined their ability to endocytose and degrade ¹²⁵I-RAP. The degradation experiments were performed for each of the four mLRP cell lines along with CHO-K1, CHO-LRP null and pcDNA3-transfected CHO-LRP null cells. As seen in Fig. 6A, when the degradation was analyzed following a 4 hour incubation and normalized via the amount of cellular protein, mLRP2, mLRP3 and mLRP4, but not mLRP1, stable cells were capable of degrading ¹²⁵I-RAP. These results are consistent with the fact that domain II, III and IV, but not domain I, of LRP possess RAP-binding sites (Bu and Rennke, 1996; Bu, 1998; Neels et al., 1999). Interestingly, when the results were normalized via the number of cell surface receptors (Fig. 6B), we found that mLRP3 cells exhibited a higher efficiency in cellular degradation of ¹²⁵I-RAP. This is interesting in that previous studies have indicated a lower affinity of LRP domain III in RAP-binding compared to domains II and IV (Bu and Rennke, 1996; Neels et al., 1999).

To further analyze the functionality of cell-surface expressed mLRPs, we examined the ability of mLRP-stably transfected cells to endocytose and degrade LRP’s physiological ligands. We examined ligands that bind LRP directly and exclusively (tPA, tPA:PAI-1 complexes, α₂M), bind to another cell surface receptor (i.e. uPA receptor) in addition to LRP (scuPA), or bind to cell-surface HSPG prior to LRP binding and uptake (TFPI) (see Krieger and Herz, 1994, for a review). As seen in Fig. 7, cellular degradation of ¹²⁵I-tPA:PAI-1 complexes (Fig. 7A), ¹²⁵I-scuPA (Fig. 7B), and ¹²⁵I-TFPI (Fig. 7C) was mediated by both mLRP2 and mLRP4, and was somewhat comparable to that of the endogenous LRP. ¹²⁵I-tPA was also degraded by these two mLRP stable cell lines, although the efficiency was significantly lower than that observed with the endogenous LRP in CHO-K1 cells (data not shown). No degradation was detected for ¹²⁵I-α₂M with any of the mLRP stable cell lines, although CHO-K1 cells, which express endogenous LRP, degrade this ligand with high efficiency (data not shown). These results suggest that tPA:PAI-1 complexes, scuPA and TFPI bind to individual domains II and IV of LRP for endocytosis and degradation, whereas uncomplexed tPA and α₂M probably require interaction with multiple domains of LRP for high-affinity binding and degradation.

Previous studies have shown that LRP is the only cell-surface receptor to which PEA binds and thus gains entry to cells (FitzGerald et al., 1995; Kounnas et al., 1992; Willnow and Herz, 1994). To analyze which LRP domain(s) are responsible for PEA binding, we examined the sensitivity of each mLRP stable cell line to this toxin. As shown in Fig. 8, only the mLRP4 stable cell line exhibited similar sensitivity to PEA when compared to wild-type CHO-K1 cells. Each cell line was incubated with toxin (1-1000 ng/ml) and susceptibility was determined by measuring inhibition of protein synthesis. An IC₅₀ value of approx. 20 ng/ml was recorded for both wild-type CHO-K1 and mLRP4 cells. The other lines required 20- to 50-fold more toxin to achieve the same level of toxicity. These results suggest that, in contrast to many other LRP ligands, PEA binds exclusively to domain IV of LRP.

In this study, we have generated membrane-containing mLRPs that represent each of the four putative ligand-binding domains of LRP. We found that proper folding of LRP domains II, III and IV was facilitated by the coexpression of RAP. Stable expression of these mLRPs in CHO cells that lack endogenous LRP showed that these mLRPs are functional in the endocytosis and degradation of both RAP and several of LRP’s physiological ligands.

LRP is an extremely large (approx. 600 kDa) endocytic receptor that contains four clusters of ligand-binding repeats. Thus, analysis of the biogenesis and ligand-binding properties of LRP is hampered by the difficulties of efficient expression of full-length receptor. Several previous studies have utilized minireceptor techniques to express either soluble forms or membrane-containing forms of LRP minireceptors that represent individual domains (Bu and Rennke, 1996; Neels et al., 1999; Obermoeller et al., 1998; Willnow et al., 1994). However, functional cell-surface expression for each of the four ligand-binding domains has not been carried out. Using surface plasmon resonance, a recent study by Neels et al. analyzed ligand interaction with immobilized soluble LRP minireceptors (Neels et al., 1999). Among several LRP ligands examined, they found that domains II and IV of LRP had similar abilities to bind ligands. Although our current results using functional mLRPs confirm the similarities of domain II and domain IV in binding LRP ligands, we found that at least one LRP ligand, PEA, binds only domain IV. Thus, our results suggest that the epitope(s) within domain II and domain IV of LRP that are utilized by LRP ligands are not always identical. Interestingly, another LRP ligand α₂M, although able to bind to soluble domain II and domain IV (Neels et al., 1999) as well as to a proteolytic fragment representing the amino terminus of domain II (Moestrup et al., 1993), did not bind to our membrane-containing mLRPs expressed on the cell surface. Similarly, Willnow et al. (Willnow et al., 1994) found that although tPA:PAI-1 complexes bind to domain II minireceptor on the cell surface, α₂M did not bind to this minireceptor. Thus, the epitope utilized by LRP for α₂M-binding must be present and differently expressed on the immobilized soluble receptor from that expressed on the membrane-containing minireceptor at the cell surface. It is possible that the presence of other domains is essential for the proper exposure of the α₂M-binding epitope in full-length LRP. Alternatively, the spatial relationship between different regions of the LRP minireceptor may present some type of steric hindrance that prevents α₂M from binding to the receptor. Interestingly, free tPA, which others found binds to domain II (Willnow et al., 1994), was less efficiently endocytosed and degraded by our domain II and domain IV mLRPs. Comparison of the domain II mLRPs in these studies shows that a large region between domain II and III of LRP that includes several EGF precursor homologous domains is present in that of Willnow et al., but not in our domain II minireceptor. Thus, it is possible that this region may be important for the binding of free tPA either directly, or indirectly by enhancing the presentation of a binding epitope for tPA within domain II. Alternatively, the amount of PAI-1 that is available for complex formation with tPA at the cell surface may be different in the host cell lines used for each study. Future investigations will be necessary to examine these possibilities.

PEA clearly binds to mLRP4 in preference to mLRP1-3. Whether this is due to differences in primary sequence or is defined by different structural folds remains to be determined. However, initial comparisons of the Class A repeats in each minireceptor, using Clustal W multiple sequence alignment, have not revealed any obvious sequence in mLRP4 that was absent in mLRP1-3 (data not shown).

To date, except for the LRP chaperone RAP, no ligand has been found to bind domain III. Thus, the function of this domain in ligand binding and endocytosis remains unclear. It is possible that a physiological LRP ligand that binds this domain exists, yet has not been identified. Alternatively, this domain may not bind ligand directly. Its presence may either enhance the overall affinity of ligands to domain II and IV, or alter the spatial presentation of these domains such that ligand-binding epitopes are sustained in native LRP.

Our previous studies using soluble forms of the LRP minireceptors have shown that proper folding and secretion of domains II, III and IV requires the coexpression of RAP (Bu and Rennke, 1996). However, using membrane-containing forms of the minireceptors, we note a lesser dependency on RAP coexpression. This suggests that membrane-anchoring of the receptor reduces the likelihood of formation of mislinked intermolecular disulfide bonds (Bu and Rennke, 1996). Our current study further indicates that, although coexpression of exogenous RAP is not absolutely required for the proper folding and processing of mLRPs, it does facilitate these processes. These results are consistent with those of Willnow et al. (Willnow et al., 1995) in vivo. Therein, mice with homozygous deletion of RAP displayed a significant reduction, but not a complete absence, of functional LRP.

In addition to analyzing receptor folding, the functional mLRPs described in the present study will be useful for dissection of the structural and functional elements involved in LRP trafficking and signaling. For example, using mLRP4 and mutagenesis strategies, we have recently found that the YXXL, but not the two NPXY motifs, serves as the dominant endocytosis signal for LRP (Li et al., 2000). Thus, each member of the LDLR family may utilize different potential signal(s) within their cytoplasmic tails for receptor-mediated endocytosis. The common NPXY motif present in all members of the LDLR family may play a role in other cellular functions. For example, a study by Herz and his colleagues has demonstrated that two cytoplasmic adaptor proteins, mammalian Disabled-1 (Dab1) and FE65, interact with the two NPXY motifs in the cytoplasmic tail of LRP, and probably serve as an intermediate in LRP-mediated signal transduction (Trommsdorff et al., 1998). Thus, functional expression of mLRPs both in vitro and in vivo should facilitate further dissection of structural and functional relationship between LRP-mediated signal transduction and endocytosis. Finally, the smaller sizes of these mLRPs should allow efficient expression of these molecules in vivo using transgenic techniques for analysis of the physiological functions of LRP.

This work was supported by grants from the National Institutes of Health (HL59150 and DK56783 to G.B. and HL53280 to A.L.S.). G.B. is an Established Investigator of the American Heart Association, and Y.L. is an American Heart Association postdoctoral fellow. We thank Joachim Herz for providing full-length LRP cDNA.