Allosteric Modulation of Ligand Binding to Low Density Lipoprotein Receptor-related Protein by the Receptor-associated Protein Requires Critical Lysine Residues within Its Carboxyl-terminal Domain*

The low density lipoprotein receptor-related protein (LRP) is a large endocytic receptor that recognizes more than 30 different ligands and plays important roles in protease and lipoprotein catabolism. Ligand binding to newly synthesized LRP is modulated by the receptor-associated protein (RAP), an endoplasmic reticulum-resident protein that functions as a molecular chaperone and prevents ligands from associating with LRP via an allosteric-type mechanism. RAP is a multidomain protein that contains two independent LRP binding sites, one located at the amino-terminal portion of the molecule and the other at the carboxyl-terminal portion of the molecule. The objective of the present investigation was to gain insight into how these two regions of RAP interact with LRP and function to modulate its ligand binding properties. These objectives were accomplished by random mutagenesis of RAP, which identified two critical lysine residues, Lys-256 and Lys-270, within the carboxyl-terminal domain that are necessary for binding of this region of RAP to LRP and to heparin. RAP molecules in which either of these two lysine residues was mutated still bound LRP but with reduced affinity. Furthermore, the mutant RAPs were significantly impaired in their ability to inhibit α2M* binding to LRP via allosteric mechanisms. In contrast, the mutant RAP molecules were still effective at inhibiting uPA·PAI-1 binding to LRP. These results confirm that both LRP binding sites within RAP cooperate to inhibit ligand binding via an allosteric mechanism.

The low density lipoprotein receptor-related protein (LRP) 1 is one of twelve or more receptors that make up the LDL receptor superfamily (for reviews, see Refs. 1 and 2) and is essential for embryonic development in mice (3). LRP recognizes more than 30 different ligands and plays important roles in protease and lipoprotein catabolism. The deduced amino acid sequence of LRP reveals that it is composed of a cytoplasmic domain containing two copies of an NPXY consensus sequence, a transmembrane domain, and a large extracellular region containing a total of 22 growth factor repeats and 31 complement-type repeats that are arranged into four clusters (clusters I-IV). Most ligands seem to bind to repeats located within clusters II and IV (4 -6). One ligand, the activated form of ␣ 2 -macroglobulin (␣ 2 M*), requires repeats from both cluster I and cluster II for binding (7).
While purifying LRP by ligand affinity chromatography, a 39-kDa protein, termed the receptor-associated protein (RAP) was identified (8 -10). Analysis of the primary sequence of RAP revealed a possible internal triple repeat structure (11,12) leading to the suggestion that this molecule may contain three domains, termed D1, D2, and D3, that roughly correspond to thirds of the molecule. The independent nature of the domains was confirmed by expressing individual fragments and performing structural analysis by 3 H NMR spectroscopy (12) and differential scanning calorimetry (DSC) (13). Interestingly, the DSC measurements suggested that D2 might be composed of two subdomains (13). The structure of the amino-terminal domain (residues 18 -112, termed D1) of RAP has been solved by NMR spectroscopy (14) and contains three helices composed of residues 23-34, 39 -65, and 73-88 that are arranged in an anti-parallel topology.
Within the cell, RAP is localized primarily to the endoplasmic reticulum due to the presence of an HNEL sequence at its carboxyl terminus (9,(15)(16)(17). Here RAP appears to function as a molecular chaperone for LRP and other LDL receptor family members by binding to the newly synthesized receptors and preventing them from associating with ligands also present within the ER (15)(16)(17)(18)(19). The mechanism by which RAP antagonizes the binding of all known ligands to LRP is not fully understood. LRP contains at least three independent binding sites for RAP, one each in cluster II, cluster III, and cluster IV. By preparing soluble fragments encompassing portions of cluster II, Vash et al. (20) demonstrated that RAP binds complement repeats 5-7 within this repeat. Repeats 5-7 are also responsible for binding two LRP ligands, uPA⅐PAI-1 complexes and lactoferrin. Thus, one mechanism by which RAP antagonizes ligand binding is by direct competition for their LRP binding site. However, other ligands, such as ␣ 2 M*, bind to a different set of complement-type repeats found in clusters I and II, and thus the ␣ 2 M* binding site does not strictly overlap with the RAP binding site (7). These results indicate that RAP also inhibits ligand binding to LRP by allosteric-type mechanisms. * This work was supported by National Institutes of Health Grants HL50784 and HL54710. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Current address: Biacore, Inc., 200 Centennial Ave., Suite 100, Piscataway, NJ 08854.
The binding of RAP to LRP is complicated by the fact that two independent binding regions within RAP bind LRP (11). One of these binding sites is localized at the amino-terminal region of RAP containing D1 and a portion of D2 (and includes amino acid residues 1-164) that binds to LRP with a K D of 9 nM (13). In addition, the carboxyl-terminal domain of RAP, which encompasses amino acid residues 216 -323, termed D3, binds to LRP with a K D of 6 nM (13,22). The objective of the present investigation was to gain insight into how these two regions of RAP interact with LRP and function to modulate its ligand binding properties.

EXPERIMENTAL PROCEDURES
Proteins and Antibodies-LRP was isolated from human placenta as described by Ashcom et al. (8). Human RAP was expressed in bacteria as fusion proteins with glutathione S-transferase and was cleaved and purified as described previously (19). Monoclonal antibody 8G1 has been described previously (23). Rabbit anti-RAP IgG (R438) was prepared against recombinant human RAP, and the IgG fraction was purified. RAP mutants were prepared using the ExSite TM PCR-based site-directed mutagenesis kit from Stratagene (24). Full-length RAP in the pGEX-2T vector was used as the template for the PCR. The D4 RAP mutant (206 -323) inserts were prepared by PCR using the following primers: 5Ј-GACCGCGGATCCAGGGTCAGCCACCAG-3Ј and 5Ј-CA-GAATTCTCAGAGTTCGTTGTGCCGAGC-3Ј. A denaturing temperature of 94°C was used with an annealing temperature of 74°C. The inserts were cleaved with the restriction enzymes BamH1 and EcoR1 and ligated into the pGEX-2T vector. All plasmids were sequenced in entirety. Protein concentrations in all experiments were determined spectrophotometrically using the following absorption coefficients Random Mutagenesis of the D3 of RAP-To identify amino acid residues within the carboxyl-terminal domain of RAP (D3) that are critical for its binding to LRP, a library of random RAP mutants was constructed. The library contained clones with randomly occurring mutations within the sequence of RAP encoding for residues 206 -323. For construction of the RAP random mutant library, the D3 coding sequence (9) was amplified by error-prone PCR as described by Lawrence et al. (25). Sequence analysis of 24 clones revealed a mutation frequency of 1:92 bp (ϳ3.8 per clone). To identify clones that were deficient in LRP binding, a receptor ligand blotting protocol was utilized. In a typical assay, 160 unique clones were screened for LRP binding by expressing the mutated D3 as a fusion with glutathione S-transferase. Bacterial extracts containing the expressed protein were analyzed for antigen by immunoblot analysis and for LRP binding by receptor blot analysis. The majority (ϳ77%) of the clones were completely negative for LRP binding, likely due to folding mutations introduced into RAP. Approximately 16% of the clones were strongly positive for LRP binding, whereas, interestingly, ϳ10% of the clones were weakly positive for LRP binding as revealed by receptor blot analysis. The weakly positive clones were selected for further analysis, because we reasoned that these molecules were defective in LRP binding but were still appropriately folded. Sequence analysis of all of these clones revealed that most clones had multiple mutations as expected from the mutation frequency.
Receptor Blotting-5 g of RAP was subjected to SDS-PAGE on 4 -20% Tris-glycine gradient gels. Following electrophoresis, proteins were transferred overnight to nitrocellulose and were visualized with Ponceau stain. After blocking with 20 mM Tris, 150 mM NaCl, pH 7.5, containing 3% milk for 1 h, the blot was incubated with 14 nM LRP in TBS containing 3% milk, 0.05% Tween 20 and 5 mM CaCl 2 for 2 h at room temperature and then washed and incubated with 1 g/ml 8G1 for 45 min. The blot was washed and developed using Bio-Rad rabbit anti-mouse horseradish peroxidase conjugate.
Ligand Blot-10 ng of LRP was subjected to SDS-PAGE using 4 -20% Tris-glycine gels. Proteins were then transferred to nitrocellulose for 3 h on 75 V at 0°C, and the nitrocellulose was then blocked with TBS containing 3% milk for 1 h. The blot was incubated with 1 or 10 nM RAP, RAP K256A, or RAP K270E in TBS containing 3% milk, 0.05% Tween, and 5 mM CaCl 2 overnight at 4°C. The blot was washed and incubated with 1 g/ml R438 anti-RAP IgG for 1 h. The blot was then washed and developed using Bio-Rad goat anti-rabbit horseradish peroxidase conjugate.
Solid-phase Binding Assay-Microtiter wells were coated with human LRP (4 g/ml in TBS, pH 7.5, 100 l/well) overnight and then blocked with 300 l of 3% BSA in TBS. Wells were washed three times with TBS containing 3 mg/ml BSA, 0.05% Tween, and 5 mM CaCl 2 . 100 l of 125 I-labeled RAP, uPA⅐PAI-1 complex, or ␣ 2 M* in wash buffer were then added to the wells in the absence or presence of competitor as indicated in the figure legends. Direct binding of D3 and D3 mutants (K256A and K270E) was measured by adding 100 l of 125 I-labeled fragment in wash buffer at concentrations indicated in the figure legend. The binding was carried out overnight at 4°C. Following incubation, the microtiter wells were washed and counted.
Heparin Binding-The fluorescence anisotropy of 0.1 M fluoresceinlabeled heparin in TBS was measured while a concentrated solution of D3 (WT), D3 K256A, or D3 K270E was added continuously with a motorized syringe controlled by the same computer controlling the fluorometer. Measurements were made at 25°C with an SLM-8000C spectrofluorometer in the T format using excitation and emission wavelengths of 493 and 524 nm, respectively. Anisotropy (A), as a function of titrant concentration, was fit to a single class of equivalent binding sites using the following equation, being the free concentration of D3, ⌬A max being the maximum anisotropy change occurring at saturating concentrations of titrant, and K D being the apparent dissociation constant of the heparin⅐protein complex. Because the concentration of fluorescein-heparin was low compared with the range of concentrations of D3, the concentration of free protein was taken as the total.
Heparin-Sepharose Affinity Chromatography-Approximately 73 g of D3 and its mutants were injected onto a 1.7-ml heparin-Sepharose column equilibrated with TBS, pH 7.4, at a flow rate of 1 ml/min controlled by an Amersham Biosciences fast-protein liquid chromatography system. A linear gradient of NaCl from 0.15 to 1 M was applied to elute bound proteins. Elution was monitored using a Shimadzu 535 fluorescence monitor to detect intrinsic fluorescence at 340 nm using an excitation wavelength of 280 nm.
Calorimetric Measurements-Differential scanning calorimetry (DSC) measurements were made with a DASM-4M calorimeter (26) at a scan rate of 1°C/min essentially as described previously (13). Protein concentrations varied from 3 to 4 mg/ml in 20 mM Gly, pH 8.7, with 0.25 M guanidinium chloride. Under these conditions, the endotherms were completely reversible (13). The DSC curves were corrected for the instrumental baseline obtained by heating the solvent. Deconvolution analysis was performed as described previously (26,27). Melting temperature (T m ) and enthalpies were determined from the DSC curves using the same software.
CD Measurements-Circular dichroism (CD) spectra were recorded on a Jasco J-715 spectropolarimeter with a Peltier PFD-350S unit for temperature control. Proteins were dissolved at a concentration of 0.28 mg/ml in phosphate buffer (10 mM, pH 8.65). A 1-mm path length cell was used, and CD spectra data points were recorded every 0.5 nm for the wavelength range 300 -180 nm at 25°C with 20 nm/min scans and a 2-s response time. Four scans were accumulated per spectrum.
Cell Uptake Assays-Cellular internalization assays were generally conducted as described previously (7). Human WI-38 fibroblasts were seeded into 12-well culture dishes (5 ϫ 10 4 cells per well) and grown in Dulbecco's modified Eagle's medium supplemented with 10% bovine calf serum and penicillin/streptomycin for 2 days. Cells were washed and incubated in assay media (Dulbecco's modified Eagle's medium containing 1.5% BSA, 1% nutradoma, and 20 mM HEPES, pH 7.5). Assay media containing 3 nM 125 I-labeled ␣ 2 M* and 200 nM RAP, RAP mutants, or RAP fragments was added to the corresponding wells and incubated for 3 h at 37°C. Following incubation the cells were washed with phosphate-buffered saline and detached from plastic using trypsin (0.5 mg/ml), proteinase K (0.5 mg/ml), and EDTA (5 mM) containing buffer. Internalized 125 I-␣ 2 M* was defined as radioactivity associated with the cell pellet. The cell numbers for each experimental condition were measured in parallel wells that did not contain radioactivity.

RESULTS
Random Mutagenesis of the D3-To identify amino acid residues within D3 that are critical for its binding to LRP, a library of random D3 mutants was constructed. This library was screened for deficiency in LRP binding using a receptor ligand blotting protocol. Those clones with impaired binding in this assay were selected and sequenced. As expected from the mutation frequency of the library, most clones had multiple mutations. Of interest, we found that the mutation frequency was significantly higher at three lysine residues located within D3: Lys-256, Lys-270, and Lys-306. One mutant that displayed impaired LRP binding had a single mutation in which Lys-270 was converted to glutamic acid. D3 is known to contain an important LRP binding site (13,28), and because prior studies have implicated charged residues in LRP binding (29), the current results implicate a role for these residues in binding to LRP. To test this hypothesis, we introduced the individual K256A, K270E, and K306A mutations into RAP and subjected the purified mutant molecules to further analysis.
Binding Analysis of RAP Point Mutants to LRP- Fig. 1 shows a receptor blot analysis measuring the binding of LRP to wild-type RAP and the three mutant RAP molecules following SDS-PAGE and transfer to nitrocellulose. It is apparent that both the K256A mutant and the K270E mutant have impaired LRP binding in this assay. Although the protein load on the gel for the K256A mutant was slightly lower in this experiment than the other proteins, in several repeats of this experiment, the K256A mutant binding was always deficient in binding LRP. In contrast, to the K256A and K270E mutants, the K306A mutant appeared normal in its binding to LRP. Thus, for the remainder of this study, experiments focused on the K256A and K270E mutants.
The direct binding of RAP and the K256A and K270E RAP mutants to purified LRP was also measured by ligand blotting approaches in which various concentrations of RAP were incubated with LRP following SDS-PAGE and transfer to nitrocellulose. The results ( Fig. 2A) demonstrate that even at low RAP concentrations (1 nM) wild-type RAP binds to LRP, whereas neither of the two mutant RAP molecules bind very effectively to LRP at this concentration. The mutant RAP molecules do bind to LRP, but even at higher concentrations (10 nM) the extent of binding is somewhat less than that of wild-type RAP (Fig. 2B).
We next measured the ability of RAP molecules containing these individual mutations to inhibit the binding of 125 I-RAP to purified LRP using a solid-phase binding assay in which LRP was coated to the surface of microtiter wells. The results of this experiment (Fig. 3) reveal that wild-type RAP competes for 125 I-labeled RAP binding with a K I,app of 3.3 nM. Both the K256A and K270E mutants showed a significant defect in their ability to inhibit the binding of 125 I-labeled RAP to LRP, with K I,app values of 86 and 94 nM, respectively. These results indicate that RAP molecules containing point mutations at residues 256 and 270 are defective in competing for the binding of 125 I-RAP to LRP.
Analysis of D3 Point Mutants-To determine if mutations at Lys-256 and Lys-270 alter the binding of D3 to LRP, we expressed and purified recombinant D3 containing the K256A and K270E point mutations. The binding of the wild-type and mutant D3 to purified LRP was assessed by solid-phase assays. The results (Fig. 4A) demonstrated that in contrast to wild-type D3, mutant D3 molecules containing the K270E mutation failed to bind to LRP, whereas only weak binding of the K256A mutant to LRP is apparent. Together, these results reveal that mutation at Lys-256 and Lys-270 impact binding of the carboxyl-terminal domain of RAP to LRP, implying that these residues contribute significantly to this interaction. D3 is also known to bind to heparin (29), and thus we performed titrations to measure the effect of these mutations on heparin binding. Wild-type D3 bound heparin with a K D of 15.4 M, whereas the K256A mutant showed a diminished affinity for heparin and the K270E mutants failed to bind (Fig. 4B). To further examine this, these mutant molecules were subjected to affinity chromatography on heparin-Sepharose (Fig. 4C). Wildtype D3 bound to the column and was eluted at 0.4 M NaCl. In contrast, the K256A mutant had markedly reduced affinity for  heparin-Sepharose and was retarded by the column, whereas the K270E mutant did not bind to the column at all. Thus, not only do mutations at Lys-256 and Lys-270 alter the ability of this fragment to bind LRP, but they also alter the interaction of the carboxyl-terminal portion of RAP with heparin.
Characterization of the Folding of D3 Point Mutants-The loss of binding activity in D3 associated with the Lys-256 3 Ala or Lys-270 3 Glu mutations could result from a general unfolding of this domain induced by the amino acid changes, because this region is known to be relatively unstable (13). Thus, we compared the unfolding properties of these two mutant proteins with wild-type D3 by differential scanning calorimetry (Fig. 5A). As reported earlier (13), D3 unfolds as a single two-state transition with an enthalpy of 45.3 kcal/mol and a T m of 41.5°C. The endotherm of the K256A mutant was virtually identical to wild-type D3 and revealed that the K256A mutation unfolded with a enthalpy of 43.2 kcal/mol and a T m of 43.1°C, confirming that this mutation does not alter the stability or unfolding properties of the domain. Interestingly, the K270E mutant was actually stabilized when compared with wild-type D3 and unfolded with an enthalpy of 59.5 kcal/mol and a T m of 49.1°C. Lys-270 is predicted to occur in a large ␣-helix, and it is possible that changing lysine to glutamic acid replaces a charge repulsion that occurs between two helical regions with a charge attraction leading to stabilization. Re-gardless of the mechanism, these two mutations in the RAP carboxyl-terminal domain do not lead to destabilization or unfolding of this region.
Previous studies have shown that D3 exhibits a CD spectrum typical for ␣-helical proteins that is abolished when the temperature is raised (13). We therefore compared the CD-spectra of wild-type D3 domain with the mutant molecules and the results, shown in Fig. 5B, reveal that both mutant D3 molecules contain extensive ␣-helical content, confirming that the secondary structure of these mutant molecules is not significantly altered. Together with the data of Fig. 5A, the results indicate that these mutations in the carboxyl-terminal domain of RAP do not lead to general unfolding of the structure nor to a decrease in the helical content of the molecule.
The K256A and K270E Mutations in RAP Effect Its Ability to Inhibit ␣ 2 M* Binding to LRP-A major function of RAP is to inhibit binding of all known ligands to LRP. We therefore measured the ability of the mutant RAP molecules to inhibit two ligands whose binding sites on LRP have been mapped: uPA⅐PAI-1 complexes and ␣ 2 M*. Like wild-type RAP, both the K256A and K270E mutants were potent inhibitors of 125 Ilabeled uPA⅐PAI-1 binding to LRP (Fig. 6A). Although differences in the dose-response curves were noted, importantly, at saturating amounts of competitor, uPA⅐PAI-1 binding was completely inhibited by all RAP molecules. In contrast to the data obtained for inhibition of uPA⅐PAI-1 binding, both the K256A and K270E mutants were unable to completely inhibit the binding of ␣ 2 M* to LRP and only reduced the binding to about 50% at saturating concentrations (Fig.  6B), indicating that mutations at residues 256 and 270 generate a RAP molecule altered in its ability to antagonize ␣ 2 M* binding to LRP.
We also examined the effect of these mutants on the ability of WI-38 fibroblasts to internalize 125 I-labeled ␣ 2 M* (Fig. 7). At a concentration of 200 nM, RAP completely blocked the internalization of 3 nM 125 I-labeled ␣ 2 M*. In contrast, both the K256A and K270E mutants reduced the amount of 125 I-labeled ␣ 2 M* internalized but were unable to completely block it, even at saturating concentrations of inhibitor. The magnitude of the effect was similar to that seen with saturating amounts of a fragment of RAP, which contains the amino-terminal LRP binding site (RAP 1-164 ) or with D1-D2 (data not shown). Curiously, D3 consistently increased the amount of 125 I-labeled ␣ 2 M* that was internalized, suggesting that its binding to LRP alters the conformation of the receptor, resulting in increased affinity for ␣ 2 M*.
Inhibition of Mutant RAP Binding to LRP by the Amino-and Carboxyl-terminal Domains of RAP-To gain insight into the relationship between the amino-and carboxyl-terminal LRP binding sites on RAP, competition experiments were performed to determine if these two RAP fragments are capable of competing with one another for binding to LRP. The results indicate that RAP and both RAP fragments were effective competitors for the binding of 125 I-labeled RAP 1-164 to LRP (Fig. 8A). In contrast, only RAP and D3 were able to compete for the binding of 125 I-labeled D3 to LRP (Fig. 8B). The inability of RAP  to block the binding of D3 to LRP reveals that these two regions of RAP bind to distinct sites on LRP. Thus, the binding of D3 to LRP must induce a conformational change in LRP that blocks binding of the amino-terminal RAP domain to its site on LRP but at the same time increases binding of ␣ 2 M* to LRP.
Mutation of either Lys-256 or Lys-270 impaired binding of the carboxyl-terminal region of RAP to LRP. Thus RAP molecules containing these mutations are likely to interact with LRP primarily via sites on the amino-terminal region of the molecule. To test this, we therefore examined the ability of D1-D2 and D3 to compete for the binding of wild-type and mutant RAP to LRP (Fig. 9A). D1-D2 does not effectively compete for the binding of wild-type RAP to LRP. In contrast, D1-D2 inhibits the binding of the mutant RAP molecules to LRP. We also examined the ability of D3 to compete with RAP and RAP mutants for LRP binding. The results (Fig. 9B) reveal that D3 was ϳ10-fold more effective as a competitor for the binding of mutant RAPs to LRP than the wild-type RAP, consistent with the ability of this RAP domain to block binding of the amino-terminal domain of RAP to LRP. DISCUSSION RAP is an endoplasmic-resident protein that associates tightly to newly synthesized LRP, gp330/megalin, and very low density lipoprotein receptor and prevents them from binding endogenously produced ligands (15)(16)(17). Recent studies reveal that RAP also acts in concert with MESD, another ER resident protein, to facilitate the export of LRP5 and LRP6 from the ER (30). The importance of RAP was revealed when the gene was deleted in mice (31), and it was found that the functional levels of LRP in the liver and brain were significantly reduced. Al- though the reason for this is not entirely clear, it appears that RAP is required to prevent receptor aggregation suggesting that it may assist in protein folding (16). Furthermore, association of LRP with certain ligands in the ER, which occurs in the absence of RAP, leads to degradation of the receptor, thereby reducing the amount of LRP on the cell surface (31). Mechanisms by which RAP prevents ligands from binding to LDL receptor family members are not fully understood at this time. This is an exceptionally complex problem, because LRP binds to 30 or more structurally distinct ligands. The objective of this study was to gain insight into mechanisms by which RAP so effectively antagonizes the binding of all known ligands to members of the LDL receptor family using LRP as the model system.
In the current investigation, we employed a random mutagenesis approach to identify critical residues in the carboxylterminal domain of RAP (D3) that are important for its interaction with LRP. This domain contains one of the two sites on RAP that bind to LRP (12,13,32,33) and was chosen because the isolated domain can completely inhibit RAP binding to LRP (12) and can prevent some, but not all, ligands from binding to LRP (33). Furthermore, transfected U87 cells are unable to secrete soluble LRP mini-receptors unless RAP or D3 is coexpressed (28). A second LRP binding site on RAP is located at the amino-terminal portion of RAP within D1-D2. In contrast to D3, a fragment containing this site is unable to compete for RAP binding to LRP, and it does not promote secretion of soluble fragments of LRP but can partially inhibit ligands like ␣ 2 M* from binding to LRP. The results of the current investigation give insight into how these two domains function in intact RAP to modulate LRP binding.
We first identified critical residues within D3 that are important for LRP binding. The results from the current study reveal that alteration of lysine 256 or lysine 270 abolishes binding of the isolated carboxyl-terminal RAP domain to LRP. It is somewhat surprising that single point mutations have such a dramatic effect on the ability of D3 to bind to LRP and to heparin. The RAP D3 domain is unusual in that it unfolds at a low temperature (T m ϭ 43°C) indicating that it may be highly flexible at physiological temperatures. Secondary structure analysis of D3 predicts a high content of ␣-helix with a large helical region from residues 237 to 277 suggesting that Lys-256 and Lys-270 are likely to occur within a helical region. The mutations do not lead to instability of the domain as revealed by differential scanning calorimetric analysis. In fact, one of the mutations (Lys-270 3 Glu) increased the stability of D3, increasing its unfolding T m by 6°C. It is possible that the mutations induce a more rigid structure in this RAP domain, which may dramatically impact its binding to LRP. Isothermal titration calorimetry binding analysis revealed that association of the carboxyl-terminal domain of RAP with two complementtype repeats from cluster II of LRP is primarily driven by entropic contributions (34). Furthermore, surface-plasmon resonance binding data are consistent with a model in which this domain of RAP undergoes a conformational change upon binding to LRP (34). A more rigid structure within RAP D3 may not accommodate these necessary conformational changes upon association with LRP.
When we introduced these mutations into full-length RAP, mutant RAP molecules were generated that are defective in their ability to compete for RAP binding and for ␣ 2 M* binding. Interestingly, the mutant RAP molecules were still effective in inhibiting the binding of uPA⅐PAI-1 complexes to LRP. Together, these data give insight into the mechanisms by which RAP modulates ligand binding by LRP and suggest that both D1-D2 and D3 domains are required for inducing conformational changes in LRP that reduce ligand binding (Fig. 10). The conformation change induced in LRP as a result of RAP binding must be initiated by D3. This is based on evidence in the current study indicating that the binding of isolated D3 to LRP actually increases the binding of ␣ 2 M* to LRP. It seems likely that the conformational change in LRP upon D3 binding alters the relationship between ligand binding clusters in LRP. Increased binding of ␣ 2 M* in the presence of D3 is consistent with a rearrangement of cluster I and cluster II to facilitate ␣ 2 M* binding (Fig. 10C).
The results in the current study reveal that mutation of lysine residues in D3 impact not only LRP binding properties but also the heparin binding properties of D3. Previous studies have noted that mutations in RAP that lead to a loss of heparin binding also lead to a loss of LRP binding (29). These investigators found that mutation of basic amino acids within two  RAP is known to bind to at least three sites on LRP one each in cluster II, III, and IV. For simplicity, only binding of RAP to cluster II is depicted. Regions on LRP within cluster I and II that are responsible for the binding of ␣ 2 M* and uPA⅐PAI-1 complexes are shown (B). Note that uPA⅐PAI-1 also binds repeats in cluster IV (6), but for simplicity, this is not shown. RAP interacts with LRP via its D3 domain and induces a conformational change in the molecule, which is depicted by a change in the shape of repeats in cluster II. This change is speculated to alter the relationship of clusters I and II (A). However, when D3 alone is present, the conformational change is not complete, and only ligands that recognize repeats in cluster II are blocked from binding. However, the binding of ligands, such as ␣ 2 M*, that recognize cluster I and a portion of cluster II is actually enhanced (C). clusters of basic residues (Arg-203 to Arg-206 along with Arg-282 to Lys-289) reduced binding of RAP to both LRP and heparin, suggesting that overlapping motifs within RAP are required for both LRP binding and heparin binding. Together with our study, it is apparent that basic residues within the carboxyl-terminal domain of RAP seem critical for binding of this portion of the molecule to LRP. A complete understanding of how RAP interacts with this receptor, however, will require solving the three-dimensional structure of the receptor⅐ ligand complex.
In summary, we have identified critical residues on the carboxyl-terminal domain of RAP important for the binding of this region to LRP. By introducing these mutations into full-length RAP, we have generated a molecule that is defective in preventing ligands, such as ␣ 2 M*, from binding to LRP but is still very effective in antagonizing the binding of uPA⅐PAI-1. The data reveal that interaction of multiple regions of RAP with LRP are important for the potent effect this molecule has on ligand binding to this receptor, and impairment of just one of these sites significantly impacts its ability to modulate ligand binding to LRP.