Identification of the minimal functional unit in the low density lipoprotein receptor-related protein for binding the receptor-associated protein (RAP). A conserved acidic residue in the complement-type repeats is important for recognition of RAP.

The low density lipoprotein receptor-related protein (LRP), a member of the low density lipoprotein receptor family, mediates the internalization of a diverse set of ligands. The ligand binding sites are located in different regions of clusters consisting of approximately 40 residues, cysteine-rich complement-type repeats (CRs). The 39-40-kDa receptor-associated protein, a folding chaperone/escort protein required for efficient transport of functional LRP to the cell surface, is an antagonist of all identified ligands. To analyze the multisite inhibition by RAP in ligand binding of LRP, we have used an Escherichia coli expression system to produce fragments of the entire second ligand binding cluster of LRP (CR3-10). By ligand affinity chromatography and surface plasmon resonance analysis, we show that RAP binds to all two-repeat modules except CR910. CR10 differs from other repeats in cluster II by not containing a surface-exposed conserved acidic residue between Cys(IV) and Cys(V). By site-directed mutagenesis and ligand competition analysis, we provide evidence for a crucial importance of this conserved residue for RAP binding. We provide experimental evidence showing that two adjacent complement-type repeats, both containing a conserved acidic residue, represent a minimal unit required for efficient binding to RAP.

The normal processing of LRP and megalin requires the presence of RAP (12), a 39 -40-kDa protein (13) that appears to consist of three homologous domains (14 -17) of which domain 1 has been shown to consist of a three-helix bundle (18). RAP interacts with all LDLR-like receptors and is a universal antagonist for all receptor/ligand interactions. RAP domains 1 and 3 (RAPd3) are both receptor-binding (15,19), but only domain 3 is sufficient to mimic the chaperone-like functions of RAP in cells (20,21). RAP domain 2 is a substrate for cAMPdependent protein kinase (22) but has only a very low affinity for LRP and megalin compared with RAP domains 1 and 3 (23).
The ectodomain of LDLR members contains clustered complement-type repeats and epidermal growth factor precursor homology domains, consisting of multiple copies of cysteinerich epidermal growth factor-like repeats and regions with a 6-fold YWTD consensus sequence, the latter suggested to form a compact ␤-propeller domain (24). CRs consist of ϳ40 amino acids of which six are cysteines forming three disulfide bridges identified in all known repeats connecting Cys I -Cys III , Cys II -Cys V , and Cys IV -Cys VI (25)(26)(27). 2 The three-dimensional structure has been solved for three LDLR modules (27)(28)(29) and one LRP module (30), revealing an octahedral calcium cage that is formed by four conserved acidic residues plus two nearby carbonyl oxygens (27,31).
A multitude of ligands are binding to these complement-type repeats, some exhibiting cross-competition such as, for example, tissue-type plasminogen activator and transformed ␣ 2macroglobulin (␣ 2 M*), while others do not. This suggests a variety of individual binding sites. So far, no reports have succeeded in identifying single residues as important for the ligand specificity of any LDLR-like receptor. Recently, Rong et al. (32) were able to substitute the fourth ligand binding repeat (LB4) of LDLR into the avian receptor for subgroup A Rous sarcoma virus (the Tva receptor), instead of the single naturally occurring complement-type repeat. They succeeded with * The present work was supported by Danish Biotechnology Program Grant 9502045. 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.
ʈ To whom correspondence should be addressed. Tel.: 45 86 20 20 00; Fax: 45 86 18 01 85; E-mail: hct@biobase.dk. 1 The abbreviations used are: LDLR, low density lipoprotein receptor; VLDLR, very low density lipoprotein receptor; RAP, receptor-associated protein; LRP, ␣ 2 -macroglobulin receptor/LDLR-related protein; CR, LRP complement-type repeat; apoER2, apolipoprotein E receptor 2; RAPd3, RAP residues 216 -323; ␣ 2 M*, transformed ␣ 2 -macroglobulin; LB, LDLR ligand binding repeat; PCR, polymerase chain reaction; U-CRxy, fusion protein containing ubiquitin and complement-type repeats x and y; PAGE, polyacrylamide gel electrophoresis; HPLC, high pressure liquid chromatography; SPR, surface plasmon resonance; HBS, Hepes-buffered saline. only a few mutations converting LB4 residues to residues originally present in the Tva receptor repeat to obtain a chimeric receptor that functioned indistinguishably from the native Tva receptor. This engineered module has three mutations: Ala-19 LB4 3 Leu Tva , Asp-23 LB4 3 His Tva , and ⌬(Pro-34 LB4 -Gln-35 LB4 /Arg-36 LB4 3 Gly Tva . Solving the crystal structure of LB5 of LDLR revealed that the residue corresponding to Asp-23 LB4 is located on the surface of the molecule with its backbone carbonyl group in a rigid position coordinating a calcium ion (27). The solution structure of CR8 from LRP also showed an aspartic acid residue located at the surface at this position (30). This is interesting, since ligand binding has previously been suggested to be dependent on charged residues, but the conserved acidic SDE-cluster, formerly speculated to bind positive charged ligands, is buried in the interior of the module and is in fact coordinating Ca 2ϩ .
The present study was undertaken to define and characterize the minimal functional unit in LRP binding RAP. Our molecular dissection of the second ligand binding cluster of CR modules of LRP and expression in Escherichia coli now delineate a minimal two-repeat RAP-binding unit and demonstrate the importance of a conserved acidic residue in each RAPbinding CR for the LRP high affinity binding of RAP.
Each set of primers was designed to introduce BamHI and HindIII restriction sites, a sequence encoding a factor X a cleavage site and a stop codon at appropriate sites in the amplified products. PCR amplification products were digested with BamHI and HindIII and cloned into BamHI-HindIII-cut E. coli T7 expression vector pT7H6Ubi (15). Fusion proteins encoded by these vector constructs, pT7H6UbiFXCRxy, contain a hexahistidine affinity tag followed by residues 2-76 of human ubiquitin and a factor X a recognition sequence preceding the receptor fragment, CRxy.
Mutations were performed using the Quickchange kit (Stratagene, La Jolla, CA) and pT7H6UbiFXCR34, pT7H6UbiFXCR45, or pT7H6-UbiFXCR56 where appropriate as template for the mutagenesis. As template for the tandem mutant U-CR56D958N,D999N, pT7H6Ubi-FXCR56D958N was constructed first and then used as template in a second mutagenesis cycle. Mutagenesis primers were from DNA Technology A/S (Aarhus, Denmark). The initial melting temperature of all primers used for mutagenesis was designed to be at least 62°C, and mutation sites were located in the central region of the primers. All constructs and mutations were verified by DNA sequencing using the Thermo Sequenase TM II dye terminator cycle sequencing kit (Amersham Pharmacia Biotech).
The RAPd3 expression construct described by Ellgaard et al. (15) was modified to include three additional N-terminal residues (RAP residues 216 -218) and an upstream methionine residue to allow removal of the N-terminal fusion partner by CNBr cleavage. Briefly, the extended RAPd3 construct was generated using the oligodeoxynucleotides 5Ј-GGC GGA TCC ATG GCT GAG TTC GAG GAG CC-3Ј and 5Ј-CAG CCA ACT CAG CTT CCT TTC GGG C-3Ј as primers and the pT7H6FX-␣ 2 MRAP expression plasmid (33) as template in the PCR. The final expression vector, pT7H6UbiMetRAPd3, was obtained by inserting the BamHI-HindIII-digested PCR fragment into the pT7H6Ubi vector (15).
Protein Expression and Affinity Binding Analysis-Recombinant RAP (33) and RAPd3 (15) were produced as described previously and were immobilized on Sepharose CL-6B (Amersham Pharmacia Biotech), which had been activated with 1,1Ј-carbonyldiimidazole (34). Recombinant U-CRxy protein was expressed in E. coli essentially as described (35), and the protein preparation was applied to a Ni 2ϩactivated nitrilotriacetic acid-Sepharose column (36) and washed with 8 M urea, 500 mM NaCl, 50 mM Tris-HCl, pH 8.0, 10 mM 2-mercaptoethanol (buffer A) until stable A 280 of the effluent. The protein was eluted from the Ni 2ϩ -nitrilotriacetic acid-Sepharose with buffer A supplemented with EDTA to 10 mM and loaded to a Q-Sepharose column (Amersham Pharmacia Biotech) after desalting into 8 M urea, 10 mM NaCl, 50 mM sodium acetate, pH 5.0, 10 mM 2-mercaptoethanol. Elution was performed with a NaCl gradient, and when necessary the protein eluate was concentrated by ultrafiltration on a 10-kDa cut-off filter (Amicon). Refolding was carried out by dialysis against 50 mM Tris-HCl, pH 8.0, 500 mM NaCl, 20 mM CaCl 2 , 3 mM reduced glutathione, and 0.3 mM oxidized glutathione for 24 h at 4°C.
The mixtures of correctly and incorrectly folded U-CRxy fusion proteins were loaded on a RAPd3 column in MB buffer (140 mM NaCl, 10 mM HEPES, pH 7.4, 1 mM MgCl 2 , 2 mM CaCl 2 ) supplemented with CaCl 2 to a final concentration of 12 mM, at 0.5 ml/min at 4°C. Bound proteins were eluted with 100 mM glycine-HCl, pH 3.0, 150 mM NaCl, 20 mM EDTA. Non-RAP-binding and RAPd3-binding derivatives were loaded on a Superdex 75 (Prep grade) 10/30 column (Amersham Pharmacia Biotech), equilibrated in MB buffer and eluted at room temperature at a flow rate of 0.5 ml/min. Pure monomeric fractions eluted at peak position relative to a total column volume of 0.66 (ϳ14.5 ml).
Where appropriate, the U-CRxy was gel-filtrated into 50 mM Tris-HCl, pH 8.0, 500 mM NaCl, 5 mM CaCl 2 and liberated from the ubiquitin fusion partner by cleavage with the endoproteinase FX a using a 1:50 (w/w) enzyme/substrate ratio and an incubation time of 4 -9 h at 4°C. Uncleaved material, fusion partner, and factor X a were removed by passage through Ni 2ϩ -nitrilotriacetic acid and factor X a inhibitor Sepharoses (Protein Engineering Technology ApS, Aarhus, Denmark).
Protein Quantification and Examination of Structural Integrity of U-CRxy Derivatives-Extinction coefficients (⑀ 280 , M Ϫ1 cm Ϫ1 ) used for protein quantification were estimated using the ExPaZy server facilities (on the World Wide Web) to be 15380 (U-CR34, U-CR78), 14660 (U-CR45, U-CR56), 15940 (U-CR67), 13380 (U-CR89), and 8410 (U-CR910). The ⑀ 280 predicted for CR34 was verified by amino acid analysis. Aliquots of isolated protein were analyzed by nonreducing SDS-PAGE, followed by either Coomassie staining, where staining intensities for U-CRxy confirmed that the use of ⑀ 280 for the proteins was justifiable, or 45 Ca blotting, essentially as in Ref. 37. Briefly, identical samples were analyzed by SDS-PAGE or transferred to nitrocellulose membrane; washed three times with 10 mM Tris-HCl, pH 8.0, 60 mM KCl, 5 mM MgCl 2 ; and incubated for 10 min in the same buffer containing 45 CaCl 2 (1 mCi/liter) before washing with 50% ethanol, drying, and exposure for 24 h before analysis with a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).
Disulfide Pattern Analysis: Generation of Peptides-Approximately 250 g of recombinant CR56 were solubilized in 5% formic acid and digested with pepsin using a 1:50 (w/w) enzyme/substrate ratio and an incubation time of 16 h. Following lyophilization, the digest was solu-bilized in 100 mM NH 4 HCO 3 , pH 8.0, and subjected to further proteolysis with trypsin using a 1:50 (w/w) enzyme/substrate ratio and an incubation time of 6 h. Finally, the peptide mixture was subjected to digestion with chymotrypsin using the same conditions as for trypsin. All digestions were at 37°C. Following digestion and lyophilization, peptides were separated by reverse-phase high pressure liquid chromatography (HPLC) on a Vydac C 18 column using an Amersham Pharmacia Biotech system. The peptides were separated in 0.1% trifluoroacetic acid and eluted with a stepwise linear gradient of acetonitrile (B) developed over 50 min (0 -5 min, 0% B; 5-40 min, 0 -50% B; 40 -50 min, 50 -95% B) at a flow rate of 0.85 ml/min. The column was operated at 40°C, and peptides were detected in the effluent by monitoring absorbance at 226 nm.
Sequence Analysis-Edman degradation was performed on an Applied Biosystems 477A sequencer equipped with an on-line HPLC. For sample loading, isolated peptides (20 -200 pmol) were pipetted onto polybrene-coated glass filters.
Mass Spectrometry-Mass spectra were acquired with a Bruker BI-FLEX matrix-assisted laser desorption/ionization time-of-flight instrument (Bruker-Franzen, Bremen, Germany) equipped with a 1-m flight tube, a reflector, a 337-nm nitrogen laser, and a 500-MHz digitizer. Thin film matrix surfaces were prepared using the fast evaporation technique (38) from ␣-cyano-4-hydroxycinnamic acid (Sigma) dissolved in acetone/water (99:1) to 30 g/l. A 0.5-l volume of analyte (0.1-10 pmol/l) was deposited on the matrix surface and allowed to dry onto the crystals. Spectra were obtained by averaging 20 -50 single-shot spectra and calibrated internally by co-crystallizing small amounts of angiotensin II (Sigma) and adrenocorticotropic hormone, fragment 18 -39 (Sigma), with the analyte and by using the calibration constants of well known matrix ions.
Surface Plasmon Resonance (SPR) Analysis-U-CRxy fusion proteins were immobilized on CM5 BIAcore sensor chips using the Amine Coupling Kit as described by the manufacturer (BIAcore, Sweden). After chip activation by the injection of 0.2 M N-ethyl-N-(3-dimethylaminopropyl)carbodiimide and 0.05 M N-hydroxysuccimide, purified U-CRxy proteins in 25 mM sodium acetate, pH 5.0, 150 mM NaCl, 10 mM CaCl 2 were diluted to a concentration of 10 g/ml by the addition of 10 mM sodium acetate, pH 4.5, and passed through the BIAcore flow cell at a rate of 5 l/min. After coupling of proteins, BIAcore chips were capped by exposure to 1 M ethanolamine, pH 8.5. Total protein coupling yields for U-CR34, U-CR56, and U-CR78 were 64/25, 64/47, and 57/17 fmol/ mm 2 (Chip SM2310D1-98/SM2901E2-99), and for U-CR45, U-CR67, and U-CR89 yields were 33, 40, and 26 fmol/mm 2 (Chip SM2903E1-99), respectively. Analyte proteins were desalted into Ca-HBS buffer (HBS buffer with calcium added to 1.5 mM), and protein binding analysis was performed at a flow rate of 5 l/min. Before loading of the protein sample, the chip was equilibrated in Ca-HBS buffer, which also was used as running buffer. Aliquots of 40 l of protein sample were injected using the KINJECT option, and regeneration of the sensor chip was performed using 1.6 M glycine-HCl, pH 3.0. Kinetic parameters were determined by using the BIAevaluation program version 3.0 (BIAcore, Sweden).
U-CRxy Competition Analysis-Labeling of RAP using the chloramine-T method (39) and ligand competition analysis (15) were essentially as described previously. Human LRP was purified from placenta essentially as described (40). Briefly, microtiter wells (Nunc Maxisorp, Denmark) were coated with approximately 1 g/ml LRP by incubation for 16 h at 4°C in 50 mM NaHCO 3 , pH 9.6. After blocking with 5% bovine serum albumin for 2 h at room temperature, the wells were washed three times with MB buffer before incubation with 125 I-RAP (ϳ7000 cpm/well) and 500 nM or 1 M U-CRxy and derivatives (all proteins were in MB buffer) was performed for 16 h at 4°C in MB buffer supplemented with 2% bovine serum albumin. Following washes with MB buffer, bound radioactivity was released by adding 10% SDS. Nonspecific binding of tracer to wells coated with LRP and inhibited with 10 mM EDTA was determined and subtracted from the values determined in the binding experiments. Fig. 1 outlines the structural architecture of LRP and the dissection of the second ligand binding cluster of complement-type repeats (cluster II) into single-or double-repeat fragments produced in E. coli cells. An identified high affinity RAP binding site in LRP (41) has previously been located to cluster II (42), but none of the tested single repeats (CR3, CR4, CR5, and CR6) displayed affinity for RAP or RAPd3 affinity columns after in vitro refolding. We subsequently generated a complete set of overlapping two-domain derivatives of cluster II, representing CR pairs CR34, CR45, CR56, CR67, CR78, CR89, and CR910. Except for the last construct, all bound the RAP and RAPd3 affinity matrices used for purification of the constructs.

Identification of a Minimal RAP Binding LRP Unit-
Each RAP affinity-purified derivative was recovered as a pure homogeneous product, producing a well defined single band by SDS-PAGE analysis of nonreduced samples ( Fig. 2A). Analysis of crude and affinity-purified receptor fragment preparations by nonreducing SDS-PAGE (Fig. 2B) showed that the well defined single bands corresponding to the affinity-purified species (U-CR34, U-CR45, U-CR56, U-CR67, U-CR78, and U-CR89) were abundant components in the crude refolding mixture. A similar prominent (i.e. probably correctly folded) component was also present in the crude U-CR910 refolding product, suggestive of efficient refolding of this receptor fragment also.
In order to provide further evidence for correct folding we analyzed the disulfide-bridging pattern in one of the refolded, factor X a -processed, and affinity-purified two-domain products, CR56. A set of disulfide-bridged peptides from digested constructs was isolated by reverse-phase HPLC, characterized by partial sequencing and mass spectrometry, and found compatible with the expected three Cys I -Cys III , Cys II -Cys V , and Cys IV -Cys VI disulfide bridges of both CR5 and CR6 (a data summary is presented in Table I).
To examine the ligand binding properties of affinity-purified two-repeat receptor fragments by SPR analysis, we prepared biosensor chips with immobilized fusion proteins U-CR34, U-CR45, U-CR56, U-CR67, U-CR78, and U-CR89 on the surface. SPR analysis of the interaction of RAP and RAPd3 with immobilized two-repeat fragments showed that RAP (Fig. 3A) and RAPd3 (Fig. 3B) both bound strongly to U-CR34, U-CR45, U-CR56, U-CR67, U-CR78, and U-CR89. Experiments made with two separately prepared sensor chips coated with different surface densities of CR pairs produced virtually identical results. For RAP binding, fitting of the recorded sensorgrams to a simple one-site model generated estimated K d values in the range of 1-5 nM for each two-repeat protein.
Given that U-CR34, U-CR45, U-CR56, U-CR67, U-CR78, and U-CR89 were all able to bind RAP when immobilized on biosensor chips, we conducted solid-phase competition analysis to examine whether they showed any difference in ability to compete with the binding between immobilized LRP and 125 I-RAP. As seen from the results shown in Fig. 4A, U-CR34, U-CR45, U-CR56, U-CR67, U-CR78, and U-CR89 were all able to compete the binding between native LRP and 125 I-RAP.
Tentative Identification from a Sequence Alignment of a Conserved Acidic Residue in LRP-An alignment of the sequences FIG. 1. Schematic representation of LRP and molecular dissection of the second cluster of complement-type repeats. Singlemodule and overlapping two-module fragments of cluster II were expressed as ubiquitin-fused proteins in an E. coli system. of the eight repeats from LRP cluster II (CR3-10) and of those of the seven repeats of the ligand binding domain of LDLR (LB1-7) is shown in Fig. 5A. A striking difference is the conservation of a surface-exposed acidic residue located at the center position between Cys IV and Cys V . This acidic residue is conserved in most repeats of LRP cluster II, but only in few repeats of the LDLR domain. An acidic residue at this position was identified in the study by Rong et al. (Asp-23 LB4 ) as important for ligand binding (32). Notably, the only repeat in the LRP cluster that lacks the negative charged residue is CR10, which was the only repeat found here not to be involved in binding to RAP. Accordingly, we directed our experimental efforts toward investigations of the contribution of this conserved aspartic acid residue to high affinity RAP binding.
Isolation, Refolding, and RAP Binding of Asp-mutated Derivatives-The expression levels of mutant proteins were virtually identical to those obtained for the wild-type two-domain fragments. However, the purification step employing RAP affinity chromatography, which was very efficient in isolating refolded U-CR34, U-CR45, and U-CR56, was less useful for the mutant proteins. U-CR34D876N, U-CR34D917N, U-CR56D958N, and U-CR56D999N exhibited decreased affinity for RAP and RAPd3 immobilized on Sepharose, and U-CR45D917N and U-CR45D958N did not bind at all. Instead of using RAP affinity chromatography, the U-CR45 mutant derivatives were purified by gel filtration. A data summary is given in Table II.
The decreased RAP affinity was apparently not a result of

TABLE I Verification of authentic disulfide topology in recombinant CR56
A set of disulfide-bridged peptides, accounting for the assignment of all three authentic disulfide bridges, Cys I -Cys III , Cys II -Cys V , and Cys IV -Cys VI in each domain, was isolated by reverse-phase HPLC after sequential incubation of CR56 two-domain protein with pepsin, trypsin, and chymotrypsin. The sequence location of peptides was assigned by amino terminal sequencing and matrix-assisted laser desorption/ionization time-of-flight mass spectrometric analysis. misfolding, since nonreducing SDS-PAGE analysis (Fig. 6A) revealed similar migration patterns for mutants and native two-domain proteins, suggesting the presence of fully oxidized mutant proteins with the authentic disulfide-bridge pattern. Furthermore, all mutant protein products were found to bind Ca 2ϩ in 45-calcium blots with the same efficiency as the nonmutated protein (Fig. 6B). This strongly suggests that correct folding had been achieved, since calcium binding is dependent on a native conformation folding (27,31). That calcium binding was not the result of one remaining functional site in the nonmutated module of a two-module protein was inferred from the results obtained for U-CR56D958N,D999N (Fig. 6B, lane  10). Here both conserved acidic residues (Asp-958 and Asp-999) in U-CR56 were replaced, each located at the critical position in the individual repeats, and calcium binding was found to be as strong as for native U-CR56. The U-CR45 derivatives showed weaker calcium binding relative to U-CR34 and U-CR56 (lanes 4 -6 versus lanes 1-3 and 7-10). This was also true for wild-type U-CR45 and was therefore not indicative of improper folding of the mutant proteins.
The decreased RAP affinity was further confirmed by solidstate competition analysis of the mutant derivatives (Fig. 4B). When 1 M U-CR34 was used as competitor, 33% of added 125 I-RAP bound to immobilized LRP as compared with wells without any competitor. In contrast, 55 or 85% binding was achieved in the presence of 1 M U-CR34D876N or U-CR34D917N, respectively. A similar effect was observed for U-CR45 and U-CR56 where RAP binding to LRP was also inhibited less efficiently by Asp 3 Asn mutant proteins. The data are summarized in Table II. DISCUSSION The present study defines two complement-type repeats as the minimal unit of LRP required for the high affinity binding to RAP. Instead of only one site responsible for RAP recognition, as previously suggested (43,44), we demonstrate multisite binding of RAP to LRP cluster II. Furthermore, simultaneous binding of RAP to two adjacent CR modules is suggested, since impaired RAP binding is not located to only one repeat but is dependent of residues in both repeats in the two-repeat proteins.
From these results, it is tempting to suggest a model of RAP binding to members of the LDLR family, where binding is mediated by the interaction between one RAP domain and one two-CR unit. This model implies the existence of a high number of binding sites on LRP in agreement with previous reports  (20). Furthermore, because RAP contains at least two high affinity domains, RAP may cross-link regions of LRP in various ways. Fig. 7 is a putative model of the complex binding between RAP and LRP. The presence of multiple binding sites provides an explanation of the unique antagonizing effect of RAP and also explains the fact that ␣ 2 M* affinity chromatographic purification of LRP with associated RAP is possible, despite RAP being an inhibitor of ␣ 2 M* binding.
We also describe a preference for an acidic residue in the second cluster of ligand binding modules from LRP, identified by the alignment of the repeats with those from LDLR. The acidic residue is not buried inside the module as are other acidic residues, previously suggested to be involved in ligand recognition but now known to coordinate calcium binding. The present work considers only residues Asp-876, Asp-917, Asp-958, and Asp-999 present in the four amino-terminal repeats (CR3-CR6) of cluster II, but in view of the equivalent binding of RAP also to tandem repeats comprising CR7, CR8, and CR9, we predict a similar role of the conserved aspartic acid residues in these repeats (Asp-1037, Asp-1085, and Asp-1128).
According to an alignment of sequences of repeats from other human members of the LDLR family (not shown) a negative charged residue at the center position between Cys IV and Cys V is present at a high frequency in other RAP binding receptors as well. It is therefore tempting to speculate whether the number of modules containing the conserved acidic residue correlates with efficient binding of RAP in general. In fact, such a tendency seems apparent when comparing the RAP binding properties of various LDLR-like receptors. A schematic representation of the LDLR family is shown in Fig. 8.
Obermoeller et al. (20) demonstrated efficient RAP binding of the second and the fourth cluster and low affinity binding to the third cluster, whereas no binding was observed to the first cluster of CR modules from the LRP. This observation is in good correlation with the number of acidic residues being 1(2), 7(8), 4(10), and 8(11) 3 (see Fig. 8). Further detailed investigation showed that whereas both the amino-and the carboxylterminal halves of cluster II and cluster IV were able to bind immobilized RAP on nitrocellulose membranes, only the carboxyl-terminal part of cluster III showed affinity for RAP (20). This also supports the present hypothesis, since only one repeat in the amino-terminal part contains an acidic residue at the position in contrast to 3(5) repeats in the carboxyl-terminal end.
Furthermore, only low affinity interaction between RAP and LDLR was reported by Medh et al. (45), in agreement with 3(7) CR modules in the LDLR ligand binding domain harboring a negative charge at the center position between Cys IV and Cys V .
Finally, the VLDLR, apoER2, and SorLA-1, having 6(8), 6(7),  The identification of multiple RAP binding sites, presented by only two complement-type repeats, suggests the possible binding of multiple RAP molecules to each LRP molecule. This gives a plausible explanation of why RAP is co-purifying with LRP when RAP affinity chromatography is used (I). We have previously shown that in addition to RAPd3, also RAP domains 1 and 2 bind RAP, although with weaker affinity than RAPd3. This gives RAP the ability to cross-link repeats from different clusters, resulting in a possible conformational change (II). Note that RAP is shown as a three-domain protein, but an alternative four-domain architecture has also been suggested (17). and 5(11) Asp/Glu CR modules, respectively, bind strongly to RAP (46 -48). It should be noted that various splice variants of both apoER2 and VLDLR have been identified (47,49), and the reported RAP binding properties of those show no discrepancies with our hypothesis. The major variant of apoER2 in human brain lacking CR4 -6 of the normally seven repeats present shows efficient RAP recognition (47), where the remaining four repeats all contain the negatively charged residue. The binding was found to be weaker than for the sevenrepeat protein, in agreement with the loss of two repeats containing Asp.
Several groups have carried out detailed analysis of the RAP binding properties of VLDLR. Savonen et al. (21) reported that RAP interacts as well with CR1-3 as with CR1-5, but not with CR6 -8. These results are in agreement with another study by Rettenberger et al. (50) demonstrating impaired RAP binding to a naturally occurring VLDLR variant lacking the third CR module, which contains a negatively charged residue. An analysis by Mikhailenko et al. (51) suggested that the RAP binding site of VLDLR is located within the amino-terminal four class A repeats and also suggested that the most amino-terminal CR is especially important for RAP binding.
It should be noted that the ligand binding repeat also is present in several proteins not exhibiting other homologies with the LDLR family. In these proteins, there is no preference for an acidic residue (e.g. the two novel discovered serine proteases, Matriptase (52), containing four modules with a tryptophan, a valine, and two lysine residues at the critical position, and TMPRSS2 (53), with a single class A repeat with a valine residue). No RAP binding has been reported for these molecules.
In conclusion, the present molecular dissection of LRP cluster II has resulted in the identification of a range of different double CR domains as independent RAP-binding units. This implies a multiplicity of RAP binding sites in LRP. The presence of a partly conserved acidic residue among clusters of CR repeats is reported to enhance RAP affinity, but solving the structure of CR pairs and RAP domains will provide further information on the molecular interactions. FIG. 8. Schematic representation of the putative RAP-binding acidic residues in various repeats from the human low density lipoprotein receptor family. Distribution of acidic residues in clusters of ligand binding repeats from members of the human LDLR family is shown. D and E represent aspartic acid and glutamic acid residues, respectively, at the position shown by an asterisk in Fig. 5, obtained from alignment of sequences of all of the repeats under the same conditions as employed in the alignment shown in Fig. 5. Triangles represent modules not present in the major splice variant of VLDLR (CR3) and ApoER2 (CR4 -6) in the brain. Removing these modules containing an acidic residue has been reported to decrease the affinity for RAP (47,50). RAP binding has been observed for ApoER2, LDLR, and SorLA-1 as well as for constructs comprising CR3-6, CR7-10, CR16 -20, CR21-26, and CR27-31 of LRP and CR1-3, CR1-5, and CR1-8 of VLDLR.