Structural Role of Extracellular Domain 1 of (cid:97) -Platelet-derived Growth Factor (PDGF) Receptor for PDGF-AA and PDGF-BB Binding*

The purpose of this study was to bacterially express, purify, and refold combinations of the extracellular im-munoglobulin (Ig)-like domains (2–3, 1–3, and 1–5) of the human (cid:97) -platelet-derived growth factor receptor ( (cid:97) PDGFR) to characterize molecular interactions with its ligand, platelet-derived growth factor (PDGF). The far UV circular dichroism spectroscopy of the (cid:97) -PDGFR extracellular domains (ECDs) revealed a predominantly (cid:98) -sheet protein, with a structure consistent with folded Ig-like domains. The addition of PDGF-BB to these ECD types changed the conformation of all three types with a decrease in mean residue ellipticity in the following rank order: 1–5 (cid:53) 1–3 (cid:62) 2–3. In striking contrast, addition of PDGF-AA to these ECD types markedly changed the conformation of ECD 2–3, by an increased mean residue ellipticity but no changes were observed for ECDs 1–3 and 1–5. PDGF-AA bound to the immobilized ECD types 2–3, 1–3, and 1–5 at concentrations of 20, 11, and 7.5 n M , respectively. In six-histidine tag (Qiagen). DNA sequencing confirmed the authenticity of the cloned inserts. The recombinant vectors were transfected into competent M15 Escherichia coli cells carrying the plasmid pREP4. Protein Expression, Refolding, and Purification— Protein expression and nickel-affinity purification was performed as described previously (19). The three domain types (2–3, 1–3, and 1–5) were refolded by dialyzing into 2 M urea buffer (100 m M phosphate), pH 7.4, containing 1 m M EDTA, 150 m M NaCl, and 0.1% (cid:98) -mercaptoethanol. Subsequently, the proteins were dialyzed into refolding buffer consisting of 100 m M phosphate, 150 m M NaCl, 2 m M reduced glutathione, 1 m M oxidized glutathione, 1 m M EDTA at pH 7.4. The refolded proteins were then dialyzed into 100 m M phosphate buffer, pH 7.4, to remove the reduced and oxidized glutathione. The refolded domains were further purified by FPLC (Mono Q) (Pharmacia) and analyzed by 4–20% SDS-PAGE (Bio-Rad) under reducing and nonreducing conditions. Solid-phase Binding Assays— Direct immobilization of the (cid:97) -PDGFR extracellular domains (ECDs) 2–3, 1–3, and 1–5 were performed by overnight incubation in immunosorbent 96-well plates (Falcon 3912) at room temperature using 80 ng of purified protein in 50 (cid:109) l of PBS in 0.02% sodium azide. Binding studies were performed with the monoclonal antibody to ECD 2 of the (cid:97) -PDGFR (mAb (cid:97) R1), PDGF-AA and -BB. The bound PDGF-AA or -BB were detected by anti-PDGF antibodies to the COOH-terminal region (20). The NH 2 -terminal SH2 do- main of GTPase-activating protein was also directly immobilized and used as controls. For competition n as Ligand Binding Assay— ECD 1–3 or ECD 2–3 was labeled by the chloramine-T method (specific activity 8 4 PDGF-AA or PDGF-BB (40 n M ) were immobilized by overnight incubation in immunosorbent 96-well plates (Falcon 3912) at temperature of Immobilized were washed by m m and incubated in the presence of 5 ng/ml labeled preincubated or PDGF-BB 125 and 125 I-ECD was quantitated and analyzed Far UV Circular Dichorism Spectroscopy— The (cid:97) -PDGFR ECD 2–3, 1–3, and 1–5, for circular dichroism spectroscopy were prepared in PBS at a protein concentration of 0.15 mg/ml. The protein concentrations of the stock solutions, used in CD measurements, were determined spec- trophotometrically (23) and the spectra scaled according to standard CD spectra for immunoglobulins. The spectra were recorded using a 1-mm demountable “strain free” quartz cuvette on a JASCO J-500C spectropolarimeter in the wavelength range of 260–195 nm at room temperature. The following settings were used: bandwidth, 1 nm; time constant, 2.0 s; step resolution, 0.1 nm; scan speed, millidegree/min; millidegree/cm.

The purpose of this study was to bacterially express, purify, and refold combinations of the extracellular immunoglobulin (Ig)-like domains (2-3, 1-3, and 1-5) of the human ␣-platelet-derived growth factor receptor (␣PDGFR) to characterize molecular interactions with its ligand, platelet-derived growth factor (PDGF). The far UV circular dichroism spectroscopy of the ␣-PDGFR extracellular domains (ECDs) revealed a predominantly ␤-sheet protein, with a structure consistent with folded Ig-like domains. The addition of PDGF-BB to these ECD types changed the conformation of all three types with a decrease in mean residue ellipticity in the following rank order: 1-5 ‫؍‬ 1-3 > 2-3. In striking contrast, addition of PDGF-AA to these ECD types markedly changed the conformation of ECD 2-3, by an increased mean residue ellipticity but no changes were observed for ECDs 1-3 and 1-5. PDGF-AA bound to the immobilized ECD types 2-3, 1-3, and 1-5 at concentrations of 20, 11, and 7.5 nM, respectively. In contrast, PDGF-BB bound the ECD types 2-3, 1-3, and 1-5 at concentrations of 3, 3, and 2.2 nM, respectively. Scatchard analysis of binding studies using labeled ECDs indicated that PDGF-BB bound ECD 1-3 and ECD 2-3 with K D values of 74 and 72 nM, respectively. While, PDGF-AA bound ECD 1-3 and ECD 2-3 with K D values of 33 and 87 nM, respectively. Therefore, our results indicated that the loss of ECD 1 impaired the binding affinity of ␣PDGFR ECD 1-3 toward PDGF-AA without having a similar effect on PDGF-BB binding. Together all of our data suggest that ECD 1 is differentially required for proper orientation of PDGF-AA but not PDGF-BB binding determinant within ECDs 2 and 3.
Platelet-derived growth factor (PDGF) 1 is a potent serum mitogen, promoting the growth of mesenchymal cells (for a review, see Ref. 1). PDGF exists as a disulfide-linked dimer (M r 28,000), composed of two homologous polypeptide chains designated A and B (2)(3)(4). Both homodimers (AA and BB) and a heterodimer (AB) have been isolated from serum (5) and bind with high affinities to either or both cell surface-glycosylated receptors designated as ␣ (6) and ␤ (7) of 180 and 185 kDa, respectively. Although both receptors are highly homologous, the three PDGF isoforms bind the ␣-PDGFR (8) while the ␤-PDGFR primarily binds the PDGF-BB form (4).
The PDGFRs are members of the receptor tyrosine kinases, which possesses five immunoglobulin-like extracellular domains, a single transmembrane spanning motif, and a split intracellular tyrosine kinase domain (7). Ligand binding leads to receptor dimerization, which activates the tyrosine kinase leading to trans-autophosphorylation of the receptor (9, 10). The tyrosine-phosphorylated receptor now becomes a target for binding Src homology region 2 (SH2) domains of a number of signaling molecules, which include phosphatidylinositol-3 kinase, GTPase-activating protein, phospholipase C␥ (PLC␥), Src, Grb2, Nck, and the tyrosine phosphatase Syp (for a review, see Ref. 11), which activate a number of inter-linked downstream signaling pathways.
A number of other growth factor receptors have been shown to dimerize in the the presence of ligand. These include the soluble extracellular domain of the epidermal growth factor receptor (12), the human growth hormone receptor (13), and the ␣- (14) and ␤-PDGFR (15). For the PDGFR, both groups expressed all five immunoglobulin-like domains in Chinese hamster ovary cells (15) and in baculovirus-infected insect cells (14). These glycosylated receptor extracellular domains were shown to exist as dimers but when deglycosylated appeared as monomers (15). These soluble receptors dimerized in the presence of added PDGF (14,15), confirming the conclusion that receptor dimerization is essential for tyrosine kinase activation.
The expression of reciprocal chimeric ␣/␤-PDGFRs in an interleukin-3-dependent cell line (32D), showed that Ig-like domains 2-3 of the ␣-PDGFR must contain the major high affinity determinants for PDGF-AA binding (16 -18). Furthermore, deletion within the second Ig-like loop of the ␣-PDGFR resulted in a marked decrease in binding PDGF-AA but not PDGF-BB (17). To further evaluate the structural roles of Iglike domains 2 and 3 in PDGF binding, we expressed and purified combinations of the ␣-PDGFR ECDs (2-3, 1-3, and 1-5) in quantities suitable for structure-function studies.

EXPERIMENTAL PROCEDURES
Construction of Expression Vectors-The DNA representing the human ␣-PDGFR extracellular domains 2-3 (ECD 2-3 101-341 ), 1-3 (ECD 1-3 24 -341 ), and 1-5 (ECD 1-5 24 -524 ) (6) were synthesized by the polymerase chain reaction method with a BamHI and HindIII site at the 5Ј and 3Ј ends, respectively. The polymerase chain reaction products were cloned into the pQE9 plasmid (type 4) possessing an NH 2 -terminal * 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.
§ Fogarty International Fellow. ‡ ‡ To whom correspondence should be addressed.  1 The abbreviations used are: PDGF, Platelet derived growth factor; ␣-PDGFR, ␣-platelet derived growth factor receptor; ECD, extracellular domain; PAGE, polyacrylamide gel electrophoresis; PBS, phosphatebuffered saline; mAb, monoclonal antibody. six-histidine tag (Qiagen). DNA sequencing confirmed the authenticity of the cloned inserts. The recombinant vectors were transfected into competent M15 Escherichia coli cells carrying the plasmid pREP4.
Solid-phase Binding Assays-Direct immobilization of the ␣-PDGFR extracellular domains (ECDs) 2-3, 1-3, and 1-5 were performed by overnight incubation in immunosorbent 96-well plates (Falcon 3912) at room temperature using 80 ng of purified protein in 50 l of PBS in 0.02% sodium azide. Binding studies were performed with the monoclonal antibody to ECD 2 of the ␣-PDGFR (mAb ␣R1), PDGF-AA and -BB. The bound PDGF-AA or -BB were detected by anti-PDGF antibodies to the COOH-terminal region (20). The NH 2 -terminal SH2 domain of GTPase-activating protein was also directly immobilized and used as controls. For competition experiments, suramin was added at increasing concentrations in the presence of PDGF-AA (10 nM) or -BB (10 nM). Immunodetection was followed as described previously (21).
Ligand Binding Assay-ECD 1-3 or ECD 2-3 was labeled by the chloramine-T method (specific activity 8 ϫ 10 4 cpm/ng). PDGF-AA or PDGF-BB (40 nM) were immobilized by overnight incubation in immunosorbent 96-well plates (Falcon 3912) at room temperature in 50 l of PBS in 0.02% sodium azide. Immobilized PDGF ligands were washed by ice-cold HEPES binding buffer (25 mM HEPES, 150 mM NaCl, 5 mM KCl, 1.2 mM MgSO 4 , and 0.1% bovine serum albumin, pH 7.5) once and incubated in the presence of 5 ng/ml labeled ECD preincubated in the presence of increasing concentrations of unlabeled ECD, PDGF-AA, or PDGF-BB for 2 h at 16°C. Free 125 I-ECD was removed by washing wells three times with ice-cold HEPES binding buffer and bound 125 I-ECD was quantitated and analyzed by the method of Scatchard.
Far UV Circular Dichorism Spectroscopy-The ␣-PDGFR ECD 2-3, 1-3, and 1-5, for circular dichroism spectroscopy were prepared in PBS at a protein concentration of 0.15 mg/ml. The protein concentrations of the stock solutions, used in CD measurements, were determined spectrophotometrically (23) and the spectra scaled according to standard CD spectra for immunoglobulins. The spectra were recorded using a 1-mm demountable "strain free" quartz cuvette on a JASCO J-500C spectropolarimeter in the wavelength range of 260 -195 nm at room temperature. The following settings were used: bandwidth, 1 nm; time constant, 2.0 s; step resolution, 0.1 nm; scan speed, 10 millidegree/min; sensitivity, 1 millidegree/cm. Each spectrum represents an average of 4 scans with the baseline subtracted.
To each of the domain types, recombinant human PDGF-AA, PDGF-BB, and EGF (UBI) at a concentration of 0.15 mg/ml was added and the spectra were recorded as described above. The spectra for the growth factors alone were also measured as described above. In order to measure conformational changes assuming 1:1 stoichiometry, the combined spectrum of the ligand and receptor, added together, digitally, was compared to the spectrum of the complex as described by Timm et al. (24).

RESULTS
Expression, Refolding, and Purification of the ␣-PDGFR ECDs 2-3, 1-3, and 1-5-In order to characterize the molecular interactions between the ␣-PDGFR extracellular domains 2-3, 1-3, and 1-5 and PDGF-AA or -BB, the above domain combinations were cloned and expressed in bacteria (Fig. 1A). Each ECD combination was purified to about 90% by Ni 2ϩ chelate affinity chromatography and analyzed by 4 -20% SDS-PAGE under reducing conditions. As shown in Fig. 1B, the apparent molecular masses for domains 2-3, 1-3, and 1-5 are 32, 46, and 70 kDa, respectively, predicting that they exist as soluble monomers under reducing and denaturing conditions. However, under nonreducing but denaturing conditions, domains 2-3 exist as a monomer while domains 1-3 and 1-5 exist as oligomers (data not shown). This suggests that domain 1 possesses the determinants that may mediate the receptor oligomerization of ECD 1-3 and ECD 1-5 by interdisulfide bonds.
␣-PDGFR ECDs 2-3, 1-3, and 1-5 Are Recognized by Receptor-specific Monoclonal Antibody-A neutralizing mAb ␣R1 directed against the ␣-PDGFR, bound the native receptor with high affinity (25). The epitope recognized by the mAb ␣R1 was shown to be denatured under reducing conditions (18). Furthermore, the major ␣PDGFR epitope binding to mAb ␣R1 was localized to Ig-like domain 2 (18). These findings indicated that high affinity binding of mAb ␣R1 to ␣PDGFR requires proper folding of determinants within Ig-like domain 2. Therefore, bacterially expressed domains 2-3, 1-3, and 1-5 were analyzed for proper folding by solid phase binding assays with the monoclonal antibody of the ␣-PDGFR (mAb ␣R1). Results shown in Fig. 2 indicated binding to the ECD 1-5, ECD 1-3 with a half-maximal value of around 0.4 nM while ECD 2-3 showed half-maximal binding at concentration of about 0.6 nM. This confirmed that all three receptor ECD types were correctly refolded, however, the maximal binding of ECDs 1-3 and 1-5 was estimated to be approximately 2-fold higher than that of ECD 2-3. Therefore, our results suggest that ECD 1 may be indirectly involved in binding of mAb ␣R1.
CD Spectroscopic Studies of PDGF-AA and -BB-induced Conformational Changes in the Three ECD Types-Far-UV CD spectroscopy was performed to investigate the folding of the three ECD types, to determine whether they possessed Ig-like structures and to study ligand-induced conformational changes. The CD spectra of the purified ECD 2-3, 1-3, and 1-5 FIG. 1. A, a schematic 2. ECD 2-3 (f), 1-3 (q), and 1-5 ( ) folding assessed by a monoclonal antibody to ␣-PDGFR ECD 2 (mAb ␣R1). The three ECD types were immobilized and probed with increasing concentrations of mAb ␣R1. The signals were detected with goat anti-mouse Fc conjugated to alkaline phosphatase. The results are represented as absorbance at 405 nm against mAb ␣R1 concentration.
indicated domains composed essentially of ␤-sheet (trough at 215 nm) similar to that observed for Ig-like domains and also confirmed that they were correctly folded. The calculated secondary structure for the 3 ECD types are 45-56% ␤-sheet and 44 -56% turns (Table I) (28) . Fig. 3, A-C, show the far-UV CD spectra of the the ECDs and their recorded spectra of a 1:1 mixture with the indicated PDGF ligands. Moreover, a calculated spectra for a 1:1 mixture of ECD and PDGF ligand is indicated. Fig. 3D, shows the spectrum of PDGF-AA or -BB which has a calculated ␤-sheet content of 60%. The addition of PDGF-AA at an equimolar ratio to the three ECD types showed a significant conformational change for domains 2-3 (increased mean residue ellipticity) in which the ␤-sheet content increased from 56 to 100% but not for domains 1-3 and 1-5 (Table I).
However, the addition of EGF to these ECD types did not show conformational changes, eliminating the possibility of nonspecific binding and conformational changes (data not shown).
Binding of PDGF-AA and -BB to ECDs 2-3, 1-3, and 1-5-The wild type receptor as well as the eukaryotic expression of all five Ig-like domains have shown high affinity binding to both PDGF-AA and -BB (14,15,26,27). To investigate binding of PDGF-AA and -BB to each ECD type, we developed a solid phase binding assay in which the ECDs were immobilized. As shown in Fig. 5A, the immobilized ECD 1-3 and ECD 1-5 each bound PDGF -AA at a concentration of 2-20 nM. However, the binding capacity of ECD 2-3 was 50% less as compared to ECD 1-3 and ECD 1-5. This is very similar to the differential observed for mAb ␣R1 binding to these ECDs. PDGF-BB bound to all three ECD types at concentrations between 2.0 and 3.0 nM (Fig. 5B), indicating that ECD 2-3 possess the major determinants for PDGF-BB binding. We next quantitated the affinity of ECD 1-3 and ECD 2-3 for each PDGF ligand. Accordingly, the level of 125 I-ECD 1-3 or 125 I-ECD 2-3 binding to the immobilized PDGF-AA or PDGF-BB was determined in the presence of increasing concentrations of unlabeled homologous ECDs. The results were then analyzed by the method of Scatchard using a one ligand/one class of binding site model. As shown in Fig. 6A, Scatchard analysis of 125 I-ECD 1-3 binding to PDGF-BB measured high affinity binding sites with a K D of 74 nM. A similar analysis of 125 I-ECD 2-3 binding to PDGF-BB measured a similar number of binding sites with a K D value of 72 nM (Fig. 6B). Furthermore, as shown in Fig. 6C, Scatchard analyses of 125 I-ECD 1-3 binding to PDGF-AA estimated high affinity binding sites with a K D value of 33 nM. In striking contrast, 125 I-ECD 2-3 reproducibly showed a similar number of binding sites with a K D value of 87 nM (Fig. 6D). Thus deletion of ECD 1 impaired the affinity of ECD 2-3 for PDGF-AA binding by more than 2-fold, without having a comparable effect on PDGF-BB binding. Under these conditions, the binding of 125 I-ECD 1-3 or 2-3 was also competed by increasing concentrations of unlabeled PDGF-AA or PDGF-BB with similar kinetics as shown with unlabeled ECDs (data not shown). These results are consistent with the finding that the soluble ␤-PDGFR receptors have K D values of 30 -100 nM, but when immobilized provided a value of 0.4 nM (27). Furthermore, ␤-PDGFR ECD 1-3 was expressed in NIH 3T3 cells as a Fc fusion protein bound to PDGF-BB with an affinity of 1.5 nM. The ␣-PDGFR ECD 1-3 competed with the ␤-PDGFR ECD 1-3 for PDGF-BB binding with a half-maximal value of 10 nM (21), thus confirming that the recombinant ␣-PDGFR ECD types

FIG. 3. The far-UV CD of the uncomplexed (A) ECD 2-3 (thin line) (B) ECD 1-3 (thin line) (C) ECD 1-5 (thin line), and (D) PDGF (AA or BB) (thin line) recorded at a protein concentration of 0.15 mg/ml in PBS.
A-C also shows the spectrum of PDGF-AA complexed ECD types (dash lines) and the combined spectrum of the PDGF⅐ECD complexes calculated (dotted lines).

TABLE I Calculated conformational changes in the ECD types upon addition of
PDGF-AA or PDGF-BB The percentage of secondary structure was calculated using the Contin program (28) (␤, ␤-sheet content; t; turn content).

Structure-function Studies on the ␣-PDGFR Extracellular Domains
shown here compare well with those expressed in eukaryotic systems. Furthermore, our findings suggest that the extent of complex formation between PDGF and ECDs is at least 78% in an equimolar mixture since the K D value of ECD for PDGF-AA or PDGF-BB is not greater than 80 nM. Thus, our data indicates that most ECDs exist in complexes with PDGF under the experimental conditions used for CD spectroscopy.
To further demonstrate the specificity of the binding interaction, we utilized suramin as a potent PDGF antagonist. Competition studies demonstrated that suramin inhibited PDGF-AA or -BB binding to the ECD types with half-maximal values in the range 1.25-2.5 M (Fig. 7, A and B). This assay provides means to test potential agonists or antagonists as described previously (21,22). DISCUSSION In this report we describe bacterial expression, purification, and refolding of the ␣-PDGFR ECDs 2-3, 1-3, and 1-5 to evaluate structural features that contribute to PDGF binding.
The nativeness of ECDs was confirmed by 1) CD spectroscopy and 2) a monoclonal antibody (mAb ␣R1) which recognizes a nonlinear epitope within ECD 2. In vitro binding studies using an immunoadsorbant assay indicated that immobilized ECD types 2-3, 1-3, and 1-5 each bound to PDGF-BB at very similar concentrations (2-3 nM). However, ECD 2-3, 1-3, and 1-5 bound PDGF-AA at concentrations of 20, 11, and 7.5 nM, respectively. Scatchard analyses of binding studies using labeled ECD 1-3 and ECD 2-3 cross-competed with unlabeled ECD types or PDGF ligands, indicated that deletion of ECD 1 from ECD 1-3 impaired PDGF-AA binding by more than 2-fold without affecting PDGF-BB binding. Therefore, our results suggest that ␣PDGFR extracellular domain 1 is differentially required for high affinity interaction with PDGF-AA but not PDGF-BB.
Previously, it was shown that an ␣/␤ chimeric PDGFR, in which domain 1 of the ␣-PDGFR was substituted for domain 1 of the ␤-PDGFR, bound to PDGF-AA with high affinity. Another ␣/␤ chimeric PDGFR in which ECD 1-3 of the ␤-PDGFR was substituted for the ECD 1-3 of the ␣-PDGFR also bound to PDGF-AA with high affinity (16). In addition, a carboxyl-terminal deletion mutants encoding the first two Ig-like domains (␣R  ) and an internal deletion mutants lacking Ig-like loop 3 (␣R 235-290 ) could not bind to PDGF-AA (18). Moreover, an internal deletion mutant lacking Ig-like loop 2 (␣R 150 -189 ) did not affect PDGF-BB binding but affected PDGF-AA binding (17). Taken together these results suggest that ECD 1 of the ␣-PDGFR is not directly involved in PDGF-AA binding. Instead ECD 1 is likely to be required for correctly orienting the other domains with respect to each other, so that the high affinity binding determinants are positioned close to each other. In contrast, for PDGF-BB, it appears that the determinants on ECD 2-3 are in the correct orientation, as indicated by high affinity binding to PDGF-BB.
We have also examined the physical characteristics of the ␣-PDGFR ECD types in the absence and presence of PDGF. Our data indicate that PDGF induces distinct conformational changes in the three ECD types. PDGF-AA significantly changed the conformation of ECD 2-3 by increasing the mean residue ellipticity but did not change the conformation of ECDs 1-3 or 1-5. It is postulated that the orientation of ECD 2-3, with respect to each other by domain 1, may be essential for correct exposure of the binding determinants. This observation is reflected in the direct binding assay where the efficiency of ECD 2-3 for PDGF-AA binding is enhanced in the presence of ECD 1. In contrast, PDGF-BB binding changed the conformation all three domain types with the following rank order: 1-5 ϭ 1-3 Ͼ 2-3 by decreasing the mean residue ellipticity. Since direct binding of the three ECD types to PDGF-BB are very similar, the conformational change observed here is likely to be due to an induced fit mechanism.
The crystal structure and mutagenesis studies performed on PDGF-BB has allowed the identification of three surface loops, two at one end of the molecule (loops 1 and 3) and the third (loop 2) at the opposite end of the elongated twisted ␤-sheet monomer. The dyad axis relating the two monomers brings loops 1 and 3 of one monomer close to loop 2 of the symmetry related monomer (29 -31). Since, PDGF-BB and -AA have a high degree of sequence similarity they are predicted to possess similar three-dimensional structures. The sequences within the same three loops in PDGF-AA, when compared with those identified for PDGF-BB, have changed from basic to hydroxy amino acids. These major sequence differences between PDGF-AA and -BB loop regions may contribute to the different affinities and mode of interaction with the ␣-PDGFR observed in this study.
Taken together, all of our findings are consistent with the proposed model depicted in Fig. 8, A and B, demonstrating the molecular interaction of PDGF-AA and PDGF-BB with the ECD 1-3 and ECD 2-3. According to this model, during the refolding process, the determinants within ECD 1 induce conformational changes within ECD 2 required for a tight interaction with PDGF-AA but not with PDGF-BB. As shown in Fig.  8A the requirement of ECD 1-induced changes in ECD 2 and 3 for PDGF-AA binding is further necessitated by the distinct physiochemical properties of binding determinants within PDGF-AA in comparison to PDGF-BB (see discussion above). Fig. 8B also illustrates how the loss of determinants within ECD 1 leads to lack of conformational change within ECD 2 and 3 necessary for tight PDGF-AA bindings. The physical consequence of this effect is PDGF-AA-induced conformational changes in ECD 2 and 3 (measured by CD and reflected by the reduced affinity) to allow such binding. In contrast the binding of PDGF-BB to ECD 2-3 is not affected since the binding site for this ligand is in the correct orientation.
In conclusion, we show that for PDGF-AA binding, ECD 2 and 3 are necessary and that ECD 1 orients these domains with respect to each other so that binding determinants are correctly positioned in three-dimensional space. However, for PDGF-BB, ECD 2-3 are correctly oriented for high affinity binding. The availability of large amounts of purified ␣-PDGFR ECDs will allow us to define the molecular interactions with PDGF in detail using x-ray crystallography. Furthermore, since PDGF and its receptor are implicated in a number of disease states including cancer, the immunosorbent assay system developed will help screen and identify potential agonists and antagonists of the ␣-PDGFR that could be of therapeutic value.