Mapping the charged residues in the second immunoglobulin-like domain of the vascular endothelial growth factor/placenta growth factor receptor Flt-1 required for binding and structural stability.

Flt-1 is one of two receptor tyrosine kinases through which the angiogenic factor vascular endothelial growth factor (VEGF) functions. Placenta growth factor (PlGF) is an additional ligand for Flt-1. The second immunoglobulin-like domain in the extracellular domain of Flt-1 has previously been identified as the region containing the critical ligand-binding determinants. We analyzed the contribution of charged residues within the first three domains of Flt-1 to ligand binding by alanine-scanning mutagenesis. Domain 2 residues Arg159, Glu208 and His223-Arg224 (together) affect both VEGF and PlGF binding, while Glu137, Lys171, His223, and Arg224 affect PlGF but not VEGF. Several charged residues, especially Asp187, are important in maintaining the structural integrity of domain 2. In addition, some residues in domain 3 contribute to binding (Asp231) or provide for additional discrimination between ligands (Arg280-Asp283).

Flt-1 is one of two receptor tyrosine kinases through which the angiogenic factor vascular endothelial growth factor (VEGF) functions. Placenta growth factor (PlGF) is an additional ligand for Flt-1. The second immunoglobulin-like domain in the extracellular domain of Flt-1 has previously been identified as the region containing the critical ligand-binding determinants. We analyzed the contribution of charged residues within the first three domains of Flt-1 to ligand binding by alanine-scanning mutagenesis. Domain 2 residues Arg 159 , Glu 208 and His 223 -Arg 224 (together) affect both VEGF and PlGF binding, while Glu 137 , Lys 171 , His 223 , and Arg 224 affect PlGF but not VEGF. Several charged residues, especially Asp 187 , are important in maintaining the structural integrity of domain 2. In addition, some residues in domain 3 contribute to binding (Asp 231 ) or provide for additional discrimination between ligands (Arg 280 -Asp 283 ).
Vascular endothelial growth factor (VEGF) 1 is a major effector of vasculogenesis and angiogenesis (1)(2)(3). The pivotal role it serves is emphasized by the fact that loss of a single allele results in embryonic lethality (4,5). VEGF activities are mediated by the endothelial cell-specific receptor tyrosine kinases, Flk-1/KDR and Flt-1 (2,6,7). Although structurally similar, the two receptors appear to invoke different signal transduction pathways, as indicated by the distinctive phenotypes associated with inactivation of each receptor gene (8,9). Mouse embryos null for Flk-1 died in utero at day 8.5 and displayed a lack of endothelial cell precursors (8). Mice lacking Flt-1 also died in utero at day 8.5. However, although endothelial cells were present, they failed to organize into normal vascular channels (9).
Most of the work in the field has focused on the Flk-1/KDR receptor due to its strong VEGF-induced tyrosine autophosphorylation and its role in mitogenesis (10 -13). Less is understood about the role of Flt-1. Although Flt-1 does exhibit VEGFinduced tyrosine phosphorylation, the signal is very weak compared with that seen with Flk-1/KDR (12,14), and Flt-1 has not been shown to be directly involved in endothelial cell mitogen-esis. However, Flt-1 also binds a VEGF-like molecule placenta growth factor (PlGF), which does not bind to Flk-1/KDR (15,16). PlGF has minimal activity in endothelial cell growth assays and vascular permeability assays (16,17), consistent with the concept that activation of Flk-1/KDR is required for both functions. However, in monocytes that express Flt-1, but not Flk-1/KDR, stimulation by either VEGF or PlGF leads to the production of tissue factor and chemotaxis (18,19). Recently, naturally occurring PlGF/VEGF heterodimers have been identified (17,20). Consistent with the ability of VEGF but not PlGF to bind to Flk-1/KDR, it has been shown that the PlGF/ VEGF heterodimers can bind Flk-1/KDR (17). Cao et al. (21) also demonstrated that PlGF/VEGF heterodimers were approximately 50% less potent than VEGF homodimers in mitogenic assays but were equally potent in stimulating cell migration. In vivo mouse corneal neovascularization assays revealed that the PlGF/VEGF heterodimers could induce new vessel growth with similar vessel lengths as VEGF homodimers, although vessel density was reduced (21). Keyt et al. (22) have shown that a plasmin-cleaved product of VEGF 165 , consisting of the first 110 NH 2 -terminal amino acids, is able to bind both Flk-1/KDR and Flt-1, suggesting that the binding determinants for both receptors reside within that region. Further analysis identified several charged residues that appeared to be associated with receptor binding (13). Residues Arg 82 , Lys 84 , and His 86 were found to be critical for binding to Flk-1/KDR, while negatively charged amino acids Asp 63 , Glu 64 and Glu 67 , were associated with Flt-1 binding (13).
Recently, we and others demonstrated that of the seven Ig-like domains within the extracellular domain (ECD) of Flt-1, the second Ig-like domain is critical for ligand binding (23)(24)(25). In the context of a recombinant chimeric receptor, the second domain of Flt-1 was sufficient for either VEGF or PlGF to bind and initiate a signal transduction cascade (23). We also demonstrated that a soluble receptor immunoadhesin (26) possessing only the first three ECD domains was sufficient to bind VEGF with the same affinity as the full-length ECD (23). In addition, PlGF can compete with VEGF for binding to constructs possessing the first three domains of Flt-1, further supporting the concept that PlGF and VEGF share similar contact sites on the receptor (23)(24)(25).
To better understand VEGF and PlGF action, analysis of the Flt-1 ligand-binding site would be useful. In the present study, using alanine-scanning mutagenesis, we examine the role of charged residues within the first three domains of Flt-1 in ligand binding. We demonstrate that at least one charged residue within the second domain serves a critical role in maintaining proper conformation of the domain and that only two of the charged residues within the second domain appear to be directly involved in ligand interaction. Furthermore, we examine the contribution of both the second and third domains of Flt-1 in distinguishing VEGF from PlGF binding.

Site-directed Mutagenesis and Expression of Flt(1-3)IgG Variants-
Mutants were created via oligonucleotide-directed mutagenesis performed according to the method of Kunkel (27). A plasmid vector, pHEBO23 (28), containing cDNA encoding the signal peptide and first three Ig-like domains of Flt-1 fused to the human heavy chain IgG␥1 (23) was used for mutagenesis and expression. Variants were screened and verified by sequence analysis using the Dye Terminator Cycle Sequencing Ready Reaction kit (Perkin Elmer) followed by application of the Sequencher 3.01 program (Gene Codes Corp.). Double-stranded DNA encoding each selected variant was prepared using the Qiagen DNA purification kit (catalog number 12163, QIAGEN Inc.) and used for transfection into CEN4 cells via calcium phosphate precipitation. Forty-eight hours post-transfection, conditioned medium was collected from transiently expressing cells, and the concentration of soluble receptor was determined by an enzyme-linked immunosorbent assay detecting human Fc-IgG.
Preparation of 125 I-Labeled VEGF 165 , PlGF 152 , and Rabbit Antimouse IgG-Radiolabeling of Escherichia coli-derived recombinant human VEGF 165 , recombinant human PlGF 152 purified from conditioned medium of stably expressing CEN4 cells, and purified rabbit antimouse IgG (Cappel) was performed using a modification of the chloramine T-catalyzed iodination reaction (29). A typical reaction consisted of 10 g of VEGF, 3 g of PlGF, or 10 g of rabbit anti-mouse IgG in phosphate-buffered saline containing 0.01% Tween 20 added to a halfvolume of 1 M Tris-HCl, pH 7.5, with 0.01% Tween 20, followed by the addition of 0.5 mCi of sodium iodide-125 (NEN Life Science Products). The reaction was catalyzed with 10 g of chloramine T (Aldrich), resuspended in 100 mM sodium phosphate buffer, pH 7.5, and incubated at room temperature for 1 min. Following the termination of the reaction by the addition of 20 g of sodium metabisulfate (Aldrich) dissolved in sodium phosphate buffer, the radiolabeled protein was isolated on a PD-10 Sephadex G-25 column (Pharmacia Biotech Inc.) that was preequilibrated with phosphate-buffered saline containing 0.5% bovine serum albumin and 0.01% Tween 20. The average specific activity of the iodinated recombinant human VEGF 165 and recombinant human PlGF 152 was 29.01 ϫ 10 6 cpm/g and 54.65 ϫ 10 6 cpm/g, respectively. Rabbit anti-mouse IgG labeled with a specific activity of 29.5 ϫ 10 6 cpm/g.
Soluble Receptor Binding Assays-Assays were performed as described previously (23) with a few modifications. In brief, 96-well breakaway immunoabsorbent assay plates (Nunc) were coated overnight at 4°C with 1 g/well rabbit IgG anti-human Fc-IgG (Cappel) in 50 mM sodium carbonate buffer, pH 9.6. Prior to the addition of 100 l of binding reaction per well, the wells were blocked with phosphatebuffered saline containing 10% fetal bovine serum (buffer B) at room temperature for 1 h. Binding reactions consisted of 10 ng/ml soluble receptor-IgG, either 125 I-labeled recombinant human VEGF 165 or 125 Ilabeled recombinant human PlGF 152 (ϳ9000 cpm/100-l reaction), and cold competitor protein where indicated. The reactions were assembled and allowed to equilibrate overnight at 4°C prior to loading into the coated wells. Incubation in the wells proceeded either for 4 h at 25°C or overnight at 4°C. Following removal of the supernatant from the wells and several washes with buffer B, the amount of bound protein was determined by counting individual wells in a ␥-counter. In addition, the unbound radiolabeled proteins in the supernatant was also determined. The data were analyzed and reported as percentage relative to the amount bound to wild-type Flt(1-3)IgG. For competition studies, a four-parameter nonlinear curve-fitting program (Kaleidagraph, Synergy Software) was used to determine the point of inflection.
Antibody Binding-Monoclonal antibodies were generated against the Flt-1 ECD region (Hybridoma Group, Genentech, Inc.). For binding, 1 ng of a receptor-IgG variant was incubated with 2 ng of monoclonal antibody (mAb) and 125 I-labeled rabbit anti-mouse IgG (ϳ6000 cpm/ 100-l reaction) in buffer B at 4°C overnight. The reactions were then placed in immunoabsorbent assay plates that had been treated as described above, and incubation continued to proceed either for 4 h at 25°C or overnight at 4°C. After the removal of unbound supernatant and several washes with buffer B, the amount of monoclonal antibody bound to receptor-IgG variant was determined by counting the amount of radiolabeled material bound to the well. The results were then normalized to that bound to the wild-type Flt(1-3)IgG.
Molecular Modeling of the Flt-1 Binding Domain-The model of human Flt-1 domain 2 (Flt-1d2) was based on the crystal structure of telokin (30). The sequence alignment of human Flt-1 domain 2, as well as human KDR domain 2, with telokin is shown in Fig. 1. Telokin was transformed into Flt-1d2 in three steps. First, all residues in the ␤-strands were changed to the Flt-1d2 sequence using the INSIGHT-II program (Biosym Technologies, San Diego, CA). If possible, conformations of Flt-1d2 side chains were kept similar to those of telokin; otherwise, they were based on INSIGHT-II rotamer libraries, packing, and hydrogen bond considerations. Second, loop structures connecting the ␤-strands were gleaned from a search of loops of similar size from the Protein Data Bank (31) using the loop search algorithm in the INSIGHT-II program. Finally, the relatively large segment connecting ␤-strands C and E (20 residues long; Phe 172 -Gly 191 ) was modeled in several conformations by a combination of loop searching and manual adjustment.
The Flt-1d2 models were subjected to energy minimization using the DISCOVER program (Biosym Technologies, San Diego, CA). The all atom AMBER force field (32) as provided in the DISCOVER program was used for all calculations, employing a 14-Å cutoff for nonbonded interactions, a linear dielectric (e ϭ 4.0 ϫ r), and 1-4 atom-atom interactions (i.e. atoms connected by two intervening covalently bonded atoms) scaled by a 0.5 factor. Prior to minimization, hydrogen atoms were added to the structure using INSIGHT-II, and positions of hydrogens on Ser, Thr, and Tyr side chains were checked visually for proper alignment in hydrogen bonds, if present. Energy minimization was performed in three stages. In stage one, 500 cycles of steepest descents minimization was employed with 53 nitrogen donor-oxygen acceptor atom pairs constrained to be within hydrogen bonding distance (2.90 Å; force constant of 50 kcal/Å). The latter were included to retain the ␤-strand nature of the immunoglobulin-like domain. In stage two, conjugate-gradient minimization was employed for 2000 iterations, and the hydrogen bond constraints were retained. In stage three, another 3000 cycles of conjugate-gradient minimization were employed with the hydrogen bond constraints released.

Alignment of VEGF and PlGF
Sequences-The sequences of mature, secreted human VEGF 165 and PlGF 152 were aligned based upon the 10 conserved cysteine residues (Fig. 2). PlGF shares 47% sequence identity with VEGF within the first 110 residues and 63% similarity overall. Two of the negatively DRd2 and -3), and telokin (Protein Data Bank code 1TLK). ␤-strands in telokin are underlined. Buried side chains are denoted by a dot above the residue; partially buried side chains are identified by an asterisk above the residue. Residue numbering is provided for Flt-1 domain 2 based on the precursor form of the receptor. charged residues previously identified as being important in the interaction of VEGF with Flt-1, Asp 63 and Glu 64 , are conserved in PlGF, whereas a histidine replaces Glu 67 , which has also been implicated in VEGF binding to Flt-1 (13). It is also noteworthy that positively charged VEGF residues identified as being involved in Flk-1/KDR binding (Arg 82 , Lys 84 , and His 86 ) are not conserved in PlGF (13,33).

FIG. 1. Alignment of human Flt-1 domains 2 and 3 (humFltd2 and -3), human KDR domains 2 and 3 (humK-
Charge-to-Alanine Mutagenesis within the First Three Domains of Flt-1-Based on the involvement of charged residues in the binding site on VEGF, we opted to examine the influence of charged residues within the first three domains of Flt-1 on ligand binding. Thirty-five mutants of the Flt(1-3)IgG soluble receptor were generated where 1-5 neighboring charged amino acids were replaced with alanine(s) ( Table I). This analysis focused on both acidic and basic amino acids (Asp, Glu, Lys, Arg, and His). Each of the resulting mutant receptors was then tested for its ability to bind either 125 I-VEGF or 125 I-PlGF relative to wild-type Flt(1-3)IgG. The overall ligand binding profiles of the variant receptors were very similar ( Table I). The receptors that bound the least amount of ligand (AS170, AS180, and AS187) all had mutations occurring within the second domain.
To determine whether these three alanine scan variants were conformationally impaired and thus unable to bind ligand, a set of five monoclonal antibodies that recognize Flt(1-3)IgG was utilized. Two of these antibodies were mapped to domain 1 (mAbs 3 and 12), and the remaining three antibodies were mapped to domain 2 (mAbs 1, 6, and 8) using Flt-1 domain deletion mutants and Flt-1/KDR domain swap chimeras described previously (23). Only mAb 6 is able to fully neutralize VEGF binding in soluble receptor-binding assays (Fig. 3). The entire set of antibodies bound poorly to the alanine scan variant AS170, implying that this receptor was incorrectly folded and thus unable to bind either ligand (Table I). Alanine scan variants AS180 and AS187 may also have structural problems as evidenced by at least one of the individual alanine mutants comprising these multiple mutants binding poorly to the panel of antibodies (see below).
Analysis of the Effects of Individual Charge-to-Alanine Mutations-Since several of the multiple alanine mutations resulted in an impaired conformation, we further analyzed domain 2 by separating some of the clusters of mutations and generating variants of Flt(1-3)IgG, which had single residue charge-to-alanine changes. The variants were tested for their ability to bind both VEGF and PlGF. The single residue mutations can be separated into three classes: those that had minimal or no effect on binding (D175A, K182A, R183A, R189A, K190A, and K200A), those that affected binding of both VEGF and PlGF, and those that affect binding of PlGF but not VEGF (Table I).
Although several individual charge-to-alanine mutants did exhibit decreases in the amount of both VEGF and PlGF bound (R159A, K170A, D187A, and E208A), only D187A showed a significant reduction in bound ligand ( To determine whether the reduced binding of the D187A mutation could be attributed to the lack of a negative charge or the difference in size of the residue, Asp 187 was changed to either glutamic acid (D187E), which retains the negative charge on the side chain, or to asparagine (D187N), which eliminates the charge but maintains the general size and polarity of the residue. Although both alterations resulted in an increase in ligand binding over the alanine mutation (D187A), neither substitution fully restored the binding observed with the wild-type Flt(1-3)IgG (Table I). Hence, both the negative charge and size of the side chain appear to be important at residue 187 for the receptor to maintain its ability to bind either ligand. Upon analysis with the set of monoclonal antibodies, we observed a significant decrease in binding by the antibodies that recognize domain 2 and a more modest reduction in binding by antibodies recognizing domain 1 (Table I). Interestingly, all three mutants affected antibody binding to domain 1 equivalently, whereas they differed in their effect on antibodies that map to domain 2. These results suggest that residue 187 is critical for maintaining the structure of domain 2 and/or the integrity of the domain 1/domain 2 interface.
Of the other single charge-to-alanine mutations that affected binding of VEGF and PlGF, K170A bound poorly only to the antibodies mapped to domain 2, and R159A and E208A showed reduced binding only to mAb 8 and could thus be part of the mAb 8 epitope on domain 2 (Table I). In the model, the Lys 170 side chain forms a salt bridge with the Glu 208 side chain. Disruption of the salt bridge might effect a conformational change, but the difference in the K170A and E208A antibodybinding patterns suggests that disruption of the salt bridge, as predicted in the model, does not result in an impaired conformation. Alternatively, the aliphatic segment of the Lys 170 side chain could be required for maintenance of the conformation, which would be removed by replacement with an alanine.
Several single charge-to-alanine mutants affected binding of PlGF but not VEGF: E137A, E141A, K171A, D180A, E201A, H223A, R224A (Table I). D180A and E201A did not bind well to One double alanine mutation that affected the amount of ligand binding but did not affect binding of the anti-Flt-1 antibodies was H223A/R224A (AS223 , Table I). When these two residues were assessed individually, the H223A and R224A mutants did not affect antibody binding and showed less pronounced reduction in bound VEGF than did the double mutant (Table I). Although the number of counts of 125 I-VEGF bound by each of the variants was similar, the possibility existed that the mutations affected the affinity of binding. Therefore, com- The reduction in affinity appeared additive, since both sites mutated together (AS223) further decreased the IC 50 mutant: wild type ratio to 0.48 Ϯ 0.06. Interestingly, the picture was very different for PlGF binding. The H223A and R224A mutants resulted in a more drastic reduction in the amount of PlGF bound, a decrease of 45 and 70%, respectively, compared with VEGF (Table I). The double mutation (AS223) further diminished the amount of bound PlGF by 90%.
Hence, ignoring mutants that are likely structurally compromised, in domain 2 we are left with Arg 159 , Glu 208 , and His 223 / Arg 224 , which affect both VEGF and PlGF binding, and Glu 137 , Lys 171 , His 223 , and Arg 224 which affect PlGF but not VEGF.

Role of Domain 3 in
Binding-Although domain 2 may be the primary site of binding for VEGF and PlGF (23)(24)(25), amino acids located in domain 3 that are near domain 2 might also influence ligand binding. Examination of the ligand binding profiles of the alanine scan mutants in domain 3 highlights two mutants: AS231 and AS279 (Table I). Asp 231 should be at the N-terminal end of ␤-strand A in domain 3 (Fig. 1) near the domain 2 interface. This mutation decreased the amount bound of both VEGF and PlGF by about 50% but did not perturb binding to the antibodies directed against domain 2 (Table I). AS279 includes residues 279 -283 of Flt-1 domain 3 and resulted in a 30% reduction in the amount of VEGF binding; PlGF appeared to be more sensitive to the charge-to-alanine changes, since the amount of ligand bound decreased by 60% (Table I). Interestingly, the antibody screening showed that domain 1 was not affected by the alterations, but there was a slight increase in the amount of binding by the antibodies that recognized domain 2 (Table I). When residues 279 -283 were instead changed to the sequence found in the KDR receptor, R280D/R281L/I282K/D283T (F280 -283K), the difference between VEGF and PlGF binding by this mutant receptor was further enhanced compared with AS279 (where all charges were changed to alanine) ( Table I). Although the amount of VEGF bound was comparable with wild type, there was a slight reduction in the binding affinity for VEGF as evidenced by the mutant:wild type IC 50 ratio of 0.68 Ϯ 0.16.
Hence, in domain 3, Asp 231 and residues 280 -283 may play a role in ligand binding. Furthermore, the difference in PlGF binding by the mutant in which residues 279 -283 were changed to alanine (AS279) compared with the mutant in which residues 280 -283 were changed to the KDR sequence (F280 -283K) suggests that this segment in KDR plays a role in preventing PlGF from binding to the KDR receptor while still allowing VEGF to bind KDR.

DISCUSSION
Examination of the charge-to-alanine scans emphasized that the domain 2 of Flt-1 is important for ligand binding, in agreement with previous work (23)(24)(25). Residues Arg 159 , Glu 208 , and His 223 /Arg 224 affect both VEGF and PlGF binding, while Glu 137 , Lys 171 , His 223 , and Arg 224 affect PlGF but not VEGF. Most of these residues (except Arg 159 ) are located on the side of  domain 2 comprising ␤-strands C, F, and G (Fig. 4). The ␤-sheet consisting of strands C, F, and G also contains many exposed hydrophobic residues (Ile 145 , Phe 172 , Leu 204 , and Leu 221 ) (Fig.  4) that were not evaluated in this study. Another patch of hydrophobic residues (Val 155 , Ile 185 , and Ile 193 ) are located adjacent to Asp 187 (Fig. 4).
Several charged residues were ascertained to play a role in maintaining the structural integrity of domain 2, as determined by monoclonal antibody binding. Interactions between an antibody and antigen (in this case a receptor) are dependent on shape complementarity of the interacting surfaces as well as direct contact with the epitope site (36,37). Disruption of either the epitope site or a receptor's structural conformation can lead to loss of antibody binding (38 -40). Utilizing several antibodies that recognize different epitopes may help in distinguishing between these possibilities (36 -40).
Asp 187 may be required to maintain the conformation of the loop between ␤-strands D and E; the binding of monoclonal antibodies, which recognize either the first or second domains of Flt-1, was reduced when either the negative charge or side chain size was altered ( Table I). The role of other residues, e.g. Lys 170 and Glu 201 , was less pronounced than that of Asp 187 , since the binding of the anti-Flt-1 antibodies was reduced much less than for the D187A mutant (Table I).
Two important residues in domain 2 are His 223 and Arg 224 . When these residues were changed to alanine, either individually or together, the resulting variants did not show any significant structural changes as determined by antibody binding (Table I). There was, however, a dramatic difference in receptor binding between VEGF and PlGF. PlGF binding was much more sensitive than VEGF to the removal of the positive charges in Flt-1 (Table I). Two other residues in domain 2, Glu 137 , distant from His 223 and Arg 224 (Fig. 4), and Lys 171 also discriminate between VEGF and PlGF (Table I).
Residues in domain 3 also contribute to the interaction. D231A affected binding of both VEGF and PlGF, while residues 280 -283 discriminated between the two ligands. The latter, when changed to the sequence of human KDR receptor, affected PlGF binding more than VEGF binding, suggesting that this segment of amino acids may contribute in preventing PlGF from binding to KDR.
The crystal structure of VEGF has recently been reported (33,41). VEGF forms an antiparallel homodimer covalently linked by two disulfide bridges. Hence, each homodimer contains two binding sites located at each end of the dimer, and each binding site has residues contributed by both monomers: Ile 43 , Ile 46 , Gln 79 , Ile 83 , Lys 84 , and Pro 85 from one monomer and Phe 17 and Glu 64 from the second monomer (33). The presence of five hydrophobic residues in the binding site on VEGF implies that part of the binding site on the receptors also contains hydrophobic residues. Hydrophobic residues were not evaluated in the present study, but several hydrophobic residues lie among His 223 /Arg 224 , Lys 171 , and Glu 137 , residues determined to be important for ligand binding and discrimination.
Although crystal structure information on PlGF is not available, based on the similarities between VEGF and PlGF (41), one might predict that a PlGF/VEGF heterodimer would also form in an antiparallel fashion with receptor binding sites at each end of the dimer. Speculating that a PlGF/VEGF heterodimer could bind Flt-1, one might hypothesize that mutations affecting binding to both homodimers (R159A, E208A, and H223A/R224A) would also affect binding to the heterodimer. It is more difficult to predict how mutations that affect PlGF homodimer binding but not VEGF binding (E137A, K171A, H223A, and R224A) would bind a heterodimer. Screening of the receptor mutants generated in this study with a PlGF/VEGF heterodimer will provide more information as to how a heterodimer may interact with Flt-1. In addition, alanine-scanning mutagenesis of PlGF will help define critical residues within that molecule involved in receptor binding.
Since VEGF binds to both Flt-1 and KDR, the Flt-1 residues involved in binding VEGF might be expected to have an identical or similar character in the KDR receptor. Comparison of the amino acid sequences of the two receptors, however, shows that this is not the case. While Glu 208 and Asp 231 are conserved, Arg 159 is a leucine in KDR and His 223 and Arg 224 are both valine in KDR (Fig. 1). Of the Flt-1 residues that discriminate between VEGF and PlGF, Glu 137 is an alanine in KDR and Lys 171 is an arginine. Even more interesting is that several of the amino acids that help maintain structure are conserved in both Flt-1 and KDR: Asp 187 , Asp 180 , and Glu 141 (aspartic acid in KDR). This suggests that domain 2 of Flt-1 and KDR are structurally similar but that the orientation of VEGF when it binds to domain 2 differs between Flt-1 and KDR. To determine if such a difference does occur, the crystal structure of KDR/ VEGF must be compared with the crystal structure of the complex of Flt-1 domain 2 with VEGF; only the latter structure has been solved (42).