Ca2+ and Receptor-associated Protein Are Independently Required for Proper Folding and Disulfide Bond Formation of the Low Density Lipoprotein Receptor-related Protein*

The low density lipoprotein receptor-related protein (LRP) is a cysteine-rich, multifunctional receptor that binds and endocytoses a diverse array of ligands. Recent studies have shown that a 39-kDa receptor-associated protein (RAP) facilitates the proper folding and subsequent trafficking of LRP within the early secretory pathway. In the current study, we have examined the potential role of Ca2+ and its relationship to RAP during LRP folding. We found that depletion of Ca2+ following either ionomycin or thapsigargin treatment significantly disrupts the folding process of LRP. The misfolded LRP molecules migrate as high molecular weight aggregates under nonreducing SDS-polyacrylamide gel electrophoresis, suggesting the formation of intermolecular disulfide bonds. This misfolding is reversible because misfolded LRP can be re-folded into functional receptor molecules upon Ca2+ restoration. Using an LRP minireceptor representing the fourth ligand binding domain of LRP, we also observed significant variation in the conformation of monomeric receptor upon Ca2+ depletion. The role of Ca2+ in LRP folding is independent from that of RAP because RAP remains bound to LRP and its minireceptor following Ca2+ depletion. Furthermore, Ca2+depletion-induced LRP misfolding occurs in RAP-deficient cells. Taken together, these results clearly demonstrate that Ca2+ and RAP independently participate in LRP folding.

The low density lipoprotein (LDL) 1 receptor-related protein (LRP) is an endocytic receptor that belongs to the LDL receptor gene family (1,2). Recent biological studies on LRP have been facilitated by the identification of an array of structurally and functionally distinct LRP ligands that include various circulating lipoproteins and protease/protease inhibitor complexes. LRP is a widely expressed, extremely large glycoprotein of 4525 amino acids, the extracellular domain of which structurally roughly resembles four combined LDL receptor molecules (1)(2)(3). Studies by Herz et al. (4) and Willnow et al. (5) have shown that LRP is synthesized as a single polypeptide chain of approximately 600 kDa and in the trans-Golgi network is cleaved by furin into two subunits of a 515-kDa ligand binding domain and an 85-kDa transmembrane domain, which remain associated with one another as they travel to the cell surface. The 515-kDa NH 2 -terminal subunit binds ligands and remains attached to the membrane through noncovalent association with the 85-kDa transmembrane subunit (4). The primary structure of LRP is remarkable for its high content of cysteine residues. Most of these cysteine residues are found within clusters of tandemly arranged complement-type or EGF-type repeats within the extracellular domain. Each of these repeats contains ϳ40 amino acid residues, including six cysteine residues, which form three disulfide bonds (6,7). Thus, a single LRP molecule may contain at least 159 disulfide bonds. This extensive degree of posttranslational modification presents a challenging task for the cell to correctly fold the receptor within the endoplasmic reticulum (ER). Thus, the folding of LRP, including the formation of disulfide bonds, is likely assisted by enzymes and molecular chaperones.
Indeed, recent studies on the 39-kDa receptor-associated protein (RAP) have defined the participation of this protein in receptor biogenesis, including its proper folding and trafficking along the early secretory pathway. Using anchor-free, soluble minireceptors that represent each of the four putative ligand binding domains of LRP (sLRPs), our previous studies (8) showed that co-expression of RAP is both necessary and sufficient for the correct folding and subsequent secretion of the sLRPs. In the absence of RAP co-expression, sLRPs misfolded as a result of formation of intermolecular disulfide bonds and are retained within the ER with little secretion. In addition to assisting receptor folding, continuous interaction between RAP and LRP along the early secretory pathway is also important for preventing premature ligand binding to LRP. This latter function of RAP correlates its ability to universally antagonize ligand interactions with the receptor (9). The role of RAP in the maturation and trafficking of LRP is also supported by gene knockout studies (10,11), which demonstrate that cells lacking RAP exhibit ER-retention of aggregated LRP and a 75% reduction in functional LRP.
RAP contains three internal repeats (12,13). Our recent studies (14) showed that the carboxyl-terminal repeat of RAP functions similarly to the full-length RAP in terms of assisting the receptor to fold. However, this repeat of RAP did not emulate full-length RAP in the inhibition of ␣ 2 -macroglobulin, a ligand for LRP. In contrast, the amino-terminal and central repeats of RAP, which were unable to assist receptor to fold, were found to inhibit ␣ 2 -macroglobulin binding to LRP (14). These differential roles of the RAP repeats suggest that the effects of RAP in receptor folding and inhibition of ligand interaction are independent functions.
It is well known that Ca 2ϩ binds to the LDL receptor (15) and LRP (1) and that binding of ligands to members of the LDL receptor gene family is Ca 2ϩ -dependent (2,3). However, the molecular basis of Ca 2ϩ interaction with these receptors was not clear until recently. Examination of the crystal structure of a ligand binding repeat from the LDL receptor revealed that each repeat contains a single Ca 2ϩ ion trapped in an octahedral cage formed primarily by four conserved acidic residues (7). Interactions between Ca 2ϩ and these acidic residues appear to be important for stabilizing and maintaining the receptor in its native conformation. In the current study, we examined the potential role for Ca 2ϩ in the folding process of LRP. We found that Ca 2ϩ and RAP are independently required for LRP folding, including the formation of correct disulfide bonds.

EXPERIMENTAL PROCEDURES
Cell Culture-Human hepatoma HepG2 cells were cultured in minimum essential medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin and maintained at 37°C in humidified air containing 5% CO 2 (16). Mouse embryonic fibroblast (MEF)-1 and MEF-7 cells were cultured under the same conditions, except that Dulbecco's minimum essential medium was used in place of minimum essential medium.
Metabolic Pulse-Chase Labeling-Metabolic labeling with [ 35 S]cysteine was performed essentially as described before (16). For pulsechase experiments, cells were generally pulse-labeled for 30 min with 200 Ci/ml [ 35 S]cysteine in cystine-free medium and chased with serum-containing medium for various times as indicated in each experiment. For calcium depletion, either 5 M ionomycin (A23187, Calbiochem), or 100 nM thapsigargin (Sigma) was included in the chase medium. Cells were lysed in ice-cold phosphate-buffered saline supplemented with 1 mM CaCl 2 and 0.5 mM MgCl 2 containing 1% Triton X-100 and 1 mM phenylmethylsulfonyl fluoride. Unless otherwise indicated, 10 mM N-ethylmaleimide (NEM) was also included in the lysis buffer to protect free -SH groups.
Antibodies, Immunoprecipitation, and SDS-PAGE-Rabbit polyclonal anti-LRP (generated against purified human LRP) and anti-RAP (generated against recombinant human RAP) antibodies have been described before (13). Monoclonal anti-HA antibody was obtained from Babco (12CA5). Immunoprecipitations were carried out essentially as described before (16), except that 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 were used to precipitate protein-IgG complexes. The immunoprecipitated material was released from the beads by boiling each sample for 5 min in Laemmli sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% (w/v) SDS, 10% (v/v) glycerol) (19). If the immunoprecipitated material was analyzed under reducing conditions, 5% (v/v) ␤-mercaptoethanol was included in the Laemmli sample buffer. The percentage of SDS-polyacrylamide gels is indicated in each figure legend. Rainbow molecular weight markers (Bio-Rad) were used as the molecular weight standards.
Construction of LRP Minireceptor-The construction of the membrane-containing minireceptor of LRP (mLRP4; see Fig. 4) via PCR was performed essentially as described previously (8). All oligonucleotides were synthesized in the Washington University School of Medicine Protein Chemistry Laboratory.
Transient Transfection-HepG2 cells at 20 -30% confluence were transiently transfected with various plasmids using a calcium phosphate precipitation method (17). Initial transfections were performed in 10-cm dishes using 40 g of DNA in a total volume of 15 ml of medium. Sixteen h after the start of transfection, cells were washed with medium and cultured continuously for an additional 4 h before being split into multiple six-well dishes (3.5 cm in diameter) for various chase points.
Chemical Cross-linking-Intracellular cross-linking with membrane-permeable cross-linker DSP (Pierce) was performed as described previously (13). Quenching of the cross-linking reaction was achieved by washing cell monolayers three times, 5 min each, with Tris-buffered saline.

RESULTS
Depletion of Ca 2ϩ Results in LRP Misfolding-To examine whether Ca 2ϩ is required for LRP folding, we analyzed the states of LRP folding upon Ca 2ϩ depletion. We used two ap-proaches to deplete ER Ca 2ϩ : ionomycin (A23187, a Ca 2ϩ ionophore) and thapsigargin (an ATPase inhibitor). Human hepatoma HepG2 cells, which express abundant LRP (16), were metabolically pulse-labeled with [ 35 S]cysteine for 30 min and chased for 0, 30, 60, or 120 min with complete medium or complete medium containing either 5 M ionomycin, 100 nM thapsigargin, or both at the given concentrations. After each chase period, cell lysates were immunoprecipitated with anti-LRP antibody and analyzed by SDS-PAGE under either nonreducing ( Fig. 1A) or reducing (Fig. 1B) conditions. As seen in Fig. 1A, at the beginning of the chase, small amounts of LRP existed in aggregated forms (lanes 1, 5, 9, and 13). These LRP aggregates are likely intermolecular disulfide bond-linked LRP molecules because they disappeared when analyzed under reducing conditions (Fig. 1B). When cells were chased in the presence of ionomycin (Fig. 1A, lanes 6 -8), thapsigargin (Fig.  1A, lanes 10 -12), or both (lanes 14 -16), most of the LRP migrated as intermolecular disulfide bond-linked aggregates, which disappeared upon reduction (Fig. 1B). In each case, the increase in LRP aggregates resulted in a corresponding decrease in monomeric LRP. These data suggest that Ca 2ϩ is required for LRP folding and, in particular, the formation of correct intramolecular disulfide bonds.
To further examine whether Ca 2ϩ depletion retards the intramolecular disulfide bond formation in addition to promoting the formation of intermolecular disulfide bonds, we compared LRP migration patterns in the presence or absence of NEM, which protects free -SH groups during cell lysis. HepG2 cells were pulse-labeled with [ 35 S]cysteine for 30 min and chased for 0, 30, or 60 min with complete medium or complete medium containing 5 M ionomycin. After each chase period, cells were lysed in either the presence or absence of NEM, immunoprecipitated with anti-LRP antibody, and analyzed under either nonreducing ( Fig. 2A) or reducing (Fig. 2B) conditions. In the absence of NEM during cell lysis, exposed free sulfhydryl groups within cysteine residues can link randomly to other free sulfhydryl groups, including those in other LRP molecules forming intermolecular disulfide bonds. Thus, by comparing the extent of aggregated LRP in the presence or absence of NEM, the free sulfhydryl groups at the time of cell lysis can be compared under various conditions. As seen in Fig. 2A, and similar to the results shown in Fig. 1A, LRP aggregation due to Ca 2ϩ depletion was seen when ionomycin was present during the chase (lanes 5 and 6). In the absence of NEM, more aggregated LRP was seen (lanes 10, 11, and 12). Most notably, after 30 or 60 min of ionomycin treatment, almost all of the monomeric LRP became aggregated into intermolecular disulfide bond-linked aggregates when cells were lysed in the absence of NEM, whereas without ionomycin treatment, LRP remained predominantly monomeric with little aggregation (lanes 8 and 9). These results suggest that depletion of Ca 2ϩ during LRP folding results in the exposure of free sulfhydryl groups, which probably then form nonnative disulfide bonds with thiols in neighboring molecules.
Misfolded LRP Can Be Refolded upon Ca 2ϩ Restoration-To examine whether misfolded LRP molecules can be rescued upon Ca 2ϩ restoration, we examined the states of LRP folding following ionomycin removal. HepG2 cells were metabolically labeled with [ 35 S]cysteine for 30 min, chased with complete medium containing 5 M ionomycin for 60 min, and continuously chased with complete medium without ionomycin for an additional 0, 60, and 120 min. After each chase, cell lysates were either immunoprecipitated with anti-LRP antibody or incubated with activated ␣ 2 -macroglobulin-Sepharose, and analyzed under either nonreducing (Fig. 3A) or reducing (Fig. 3B) conditions. As seen in Fig. 3A, LRP was refolded from aggre-gated forms into monomeric form upon Ca 2ϩ restoration. The refolded LRP appears to be functional because it is capable of binding its ligand ␣ 2 -macroglobulin. of misfolded monomeric LRP forms with nonnative intramolecular disulfide bonds into discrete species. Therefore, we generated an LRP minireceptor that could mimic the folding process of LRP. This LRP minireceptor encodes residues 2462-4525 (1) and represents from the beginning of the fourth cluster of ligand binding repeats (2) through the carboxyl terminus of the receptor (designated mLRP4, with "m" denoting that this minireceptor contains the membrane-spanning sequence; see Fig. 4). To facilitate immunoprecipitation following metabolic labeling, an HA epitope was included near the NH 2 -terminal end of mLRP4.
Our previous studies using soluble LRP minireceptor 4 (sLRP4, see Ref. 8) have shown that proper folding and secretion of sLRP4 requires the co-expression of RAP. Thus, we transiently transfected cDNA for mLRP4 into HepG2 cells with co-transfection of either empty vector (pcDNA3) or vector containing RAP cDNA (pcDNA-RAP; see Ref. 13). The transfected cells were metabolically pulse-labeled with [ 35 S]cysteine for 30 min and chased for 0, 30, 60 or 120 min. Cell lysates were then immunoprecipitated with anti-HA antibody and analyzed by SDS-PAGE under either nonreducing (Fig. 5A) or reducing (Fig. 5B) conditions. As seen in Fig. 5, mLRP4 is folded and transported from the ER to the Golgi faster following co-transfection with RAP (see plot in Fig. 5B). For example, after 60 min of chase, 66% of mLRP4 had been converted to post-ER forms (including the Golgi form and the processed forms) when co-transfected with RAP, compared with only 29% in the absence of RAP co-transfection (Fig. 5B). These results are consistent with our previous conclusion that RAP facilitates LRP folding and trafficking through the early secretory pathway. It is interesting to note that in the absence of RAP co-transfection, significant amounts of mLRP4 migrated at the top of the stacking gel under nonreducing conditions (Fig. 5A), suggesting the formation of excessive intermolecular disulfide bonds and a concomitant delay for their folding and trafficking.
We next examined the effects of ionomycin on the folding and processing of mLRP4. As in the experiment shown in Fig. 5, HepG2 cells were transiently transfected with cDNA for mLRP4, with co-transfection of either pcDNA3 (-RAP) or pcDNA-RAP (ϩRAP). The transfected cells were metabolically pulse-labeled with [ 35 S]cysteine for 30 min and chased in the presence of ionomycin for 0, 30, 60, or 120 min. Cell lysates were then immunoprecipitated with anti-HA antibody and analyzed by SDS-PAGE under either nonreducing (Fig. 6A) or reducing (Fig. 6B) conditions. As shown in Fig. 6, the depletion of Ca 2ϩ caused misfolding of mLRP4 in both the absence and the presence of RAP co-expression, consistent with the notion that RAP and Ca 2ϩ are independently required for LRP folding (see below). More noticeably, misfolding of mLRP4 can be seen not only as intermolecular-linked aggregates but also as a broad band indicative of heterogeneously folded minireceptor monomers (Fig. 6A, lanes 3, 4, 7, and 8), which likely arise because of mislinked intramolecular disulfide bonds.
Ca 2ϩ and RAP Are Independently Required for LRP Folding-The interaction between RAP and LRP is known to be Ca 2ϩ -dependent (9,13). Ca 2ϩ depletion with ionomycin decreases the Ca 2ϩ concentration within the ER but does not completely eliminate Ca 2ϩ (18). To analyze whether RAP still interacts with LRP and its minireceptor under the conditions of ionomycin treatment, we performed chemical cross-linking experiments using a membrane-permeable cross-linker, DSP (13). HepG2 cells were transiently co-transfected with cDNAs for mLRP4 and RAP. The transfected cells were then metabolically pulse-labeled with [ 35 S]cysteine for 30 min, either followed or not followed by a 30-min chase of complete medium containing ionomycin. Cells were then incubated in the absence or presence of DSP cross-linker, immunoprecipitated with anti-RAP antibody (13), and analyzed by SDS-PAGE under either nonreducing (Fig. 7A) or reducing (Fig. 7B) conditions. As seen in Fig. 7, following chemical cross-linking with DSP, LRP and mLRP4 were co-immunoprecipitated with anti-RAP antibody, both without (lane 2) and with (lane 4) ionomycin treatment. Thus, under the conditions of Ca 2ϩ depletion with ionomycin, RAP remains associated with LRP. These results indicate that even when RAP remains associated with LRP, depletion of Ca 2ϩ results in LRP misfolding.
To further analyze whether Ca 2ϩ and RAP are independ- FIG. 3. LRP misfolding is reversible upon Ca 2؉ restoration. HepG2 cells were metabolically labeled with [ 35 S]cysteine for 30 min, chased for 60 min with complete medium containing 5 M ionomycin, and continuously chased with complete medium without ionomycin for 0, 60, or 120 min. Cells were then lysed in the absence of NEM, divided into two equal parts, either immunoprecipitated with anti-LRP antibody or incubated with ␣ 2 -macroglobulin-Sepharose, and analyzed via 5% SDS gels under either nonreducing (A) or reducing (B) conditions. ently required for LRP folding, we compared the effects of Ca 2ϩ depletion on LRP folding in RAP-expressing MEF-1 cells and RAP-deficient MEF-7 cells (11). Previous studies by Willnow et al. (11) have shown that folding of LRP and its trafficking in RAP-deficient MEF-7 cells is not significantly impaired. In the present study, MEF-1 and MEF-7 cells were metabolically pulse-labeled with [ 35 S]cysteine for 30 min and chased for 0, 30, or 60 min in the absence or presence of ionomycin. Cell lysates were then immunoprecipitated with anti-LRP antibody and analyzed by SDS-PAGE under either nonreducing (Fig. 8A) or reducing (Fig. 8B) conditions. As seen in Fig. 8A, in the presence of ionomycin during chase, aggregation of LRP was seen in both MEF-1 (lanes 5 and 6) and MEF-7 (lanes 11 and 12) cells, suggesting that irrespective of whether RAP is present, Ca 2ϩ depletion results in significant LRP misfolding. DISCUSSION Ligand binding to members of the LDL receptor gene family requires Ca 2ϩ . The underlying mechanism has recently been defined; Ca 2ϩ binds to conserved acidic residues at each of the cysteine-rich repeats and stabilizes the receptors in their native conformation (7). In the present report, we analyzed the potential role of Ca 2ϩ in LRP folding and examined its relationship to the function of RAP as a folding chaperone. We found that Ca 2ϩ and RAP are independently required for LRP folding. When Ca 2ϩ and RAP are limited during receptor folding, LRP becomes misfolded with the formation of intermolecular disulfide bonds. Intermolecular (also termed interchain) disulfide bonds have been observed in other cases of glycoprotein misfolding, e.g. influenza hemagglutinin (19,20). In addition to intermolecular disulfide bond-linked molecules, hetero-geneously migrating monomeric LRP minireceptors were also observed upon Ca 2ϩ depletion. This suggests that binding of Ca 2ϩ to LRP is important not only in preventing the formation of intermolecular disulfide bonds, but also in facilitating the formation of correctly linked intramolecular disulfide bonds. Because ER Ca 2ϩ is also required for the folding process of at least one other endocytic receptor, the asialoglycoprotein receptor (21), Ca 2ϩ may serve as an essential folding chaperone for some of the endocytic receptors within the ER.
The ER is an oxidative environment in which disulfide bondcontaining proteins may fold correctly (22). The oxidative redox state of the ER maintained in part by the ratio of oxidized: reduced glutathione (GSH:GSSH) at 1:1-3:1 is favorable toward disulfide bond formation and rearrangement (23). Such an oxidative environment within the ER would also facilitate the spontaneous formation of mislinked disulfide bonds, especially during folding of cysteine-rich proteins. For example, in the pulse-chase experiments performed in the present study (see Figs. 1 and 2), we consistently observed intermolecular disulfide bond-linked LRP aggregates immediately following pulse labeling. Binding of Ca 2ϩ and RAP may reduce the amounts of mislinked disulfide bonds, perhaps by protecting certain cysteine residues against oxidation at early stages of LRP folding. In addition, because mislinked disulfide bonds are present during the normal folding process, it is possible that Ca 2ϩ and RAP facilitate the disulfide bond reshuffling process catalyzed by protein disulfide isomerase (PDI, 24). Thus, the process of folding for proteins with extensive disulfide bonds is likely dynamic and may involve constant trial and error, especially in the formation of correct disulfide bonds. Rearrange- ment of disulfide bonds during the normal folding process is also suggested by our observation that misfolded LRP containing extensive intermolecular disulfide bonds can be refolded into functional monomeric LRP (Fig. 3). These results suggest that the ER possesses all the factors and machinery necessary to convert LRP molecules from misfolded states to correctly folded states. Such functions not only support the existence of the trial and error theory of protein folding but also provide a means for cells to recover protein functions following various stress conditions. It is interesting to note that the structure of the fifth complement-type ligand binding repeat of the LDL receptor is organized around a calcium ion (7). Six residues, including four acidic residues, contribute to the octahedral coordination geometry around the Ca 2ϩ ion. It is possible that the formation of the Ca 2ϩ coordination within each repeat juxtaposes at least two pairs of cysteine residues to form native disulfide bonds. Thus, Ca 2ϩ ions may play critical roles in the initial structure organization during receptor folding, which precedes the formation of disulfide bonds. This hypothesis is supported by in vitro folding studies on the fifth repeat of the LDL receptor (25). In an experiment not shown here, we found that depletion of Ca 2ϩ following the completion of LRP folding (i.e. 30-min pulse labeling and 60-or 120-min chase) did not induce further misfolding of LRP, suggesting that Ca 2ϩ is not required for maintaining correctly disulfide-bonded LRP structure. However, because Ca 2ϩ is required for ligand binding to LRP (2), this metal ion may play a certain role in maintaining the receptor in a conformation that is competent for ligand binding (7).
The present study demonstrates that Ca 2ϩ and RAP are independently involved in LRP folding. However, whether these factors function in similar ways remains to be determined. It appears from the current study that ER Ca 2ϩ is essential for LRP folding. Depletion of Ca 2ϩ results in the eventual total misfolding of LRP. In contrast, participation of RAP in LRP folding appears to be important but not essential. For example, deletion of the RAP gene by homologous recombination resulted in a 75% reduction but not total elimination of functional LRP molecules (10,11). In addition, when LRP minireceptors are expressed in the absence of RAP-co-expression, some of them appear able to fold and be processed correctly (see Fig. 5 and Ref. 26). The presence of less efficacious redundant factors or chaperones remains to be defined. Thus, RAP likely functions as a facilitator during LRP folding. The fact that RAP functions in both LRP folding and subsequent trafficking along the early secretory pathway (11,13) emphasizes the specialized chaperone function of RAP during the biogenesis of LRP, and likely other members of the LDL receptor gene family (27).
In summary, our current results demonstrate that Ca 2ϩ and RAP are independently involved in LRP folding. Ultimately, understanding the folding dynamics of LRP, a molecule with potential complex secondary structure and a high disulfide bond content, should provide us with additional insight as to how such a protein can fold correctly and efficiently within the ER.