An Antibody Reactive with Domain 4 of the Platelet-derived Growth Factor β Receptor Allows BB Binding while Inhibiting Proliferation by Impairing Receptor Dimerization*

A panel of murine monoclonal antibodies was generated against the extracellular domain of the human platelet-derived growth factor (PDGF) β receptor (PDGFRβ). These antibodies were assayed for both the ability to inhibit binding of PDGF BB to PDGFRβ+ cells as well as the capacity to inhibit PDGF BB-mediated mitogenesis. As expected, all antibodies that could prevent PDGF BB binding also inhibited mitogenesis. However one antibody (M4TS.11), with no detectable ability to inhibit PDGF BB binding, was a potent inhibitor of proliferation induced by PDGF BB. Further characterization indicated that M4TS.11 impaired PDGFRβ dimerization, revealing the mechanism by which it prevented PDGF BB-mediated mitogenesis. Using domain deletion mutants of the extracellular portion of PDGFRβ, the determinant recognized by this antibody was localized to the fourth extracellular domain of PDGFRβ, indicating that this domain, which is not involved in ligand binding, actively participates in receptor dimerization and signal transduction. The M4TS.11 antibody could also inhibit PDGF BB-mediated proliferation of responsive cells from both the baboon and the rabbit, indicating the determinant recognized by the antibody is not limited to humans and making it possible to use this antibody to evaluate the therapeutic benefit of interfering with PDGF in animal models of human disease.

Platelet-derived growth factor (PDGF) 1 is a mitogen and chemoattractant for cells of mesenchymal origin, such as fibroblasts, smooth muscle cells, and glial cells (1)(2)(3). PDGF is encoded by two genes, the products of which are designated A and B. The active PDGF molecule is a disulfide-linked dimer of these polypeptides and thus can exist in three forms: the homodimers AA or BB or the heterodimer AB (4).
Studies examining the interaction of PDGF with responsive cells have revealed the existence of two specific receptors designated ␣ and ␤ and encoded by separate genes. Each receptor type is composed of five extracellular immunoglobulin-like domains attached to an intracellular tyrosine kinase domain via a transmembrane segment (5). This structural organization is the prototype for the PDGF receptor family of protein-tyrosine kinases, which includes stem cell factor receptor, colony stimulating factor receptor, and Flk-2 (6).
The PDGF dimer stimulates responsive cells by cross-linking two receptor subunits (5,6). The different forms of PDGF exhibit different affinities for the two forms of the PDGF receptor. The PDGFR␣ can interact with all three forms of PDGF; PDGFR␤ can only interact with PDGF BB and AB. This pattern of reactivity dictates that PDGF AA can signal the cell only through homodimers of PDGFR␣, PDGF BB can signal the cell through homodimers of PDGFR␣ or PDGFR␤ or the heterodimer PDGFR␣⅐PDGFR␤, and PDGF AB can stimulate cells through either homodimers of PDGFR␣ or the PDGFR␣⅐ PDGFR␤ heterodimer (7).
As a potent mitogenic and chemotactic agent, PDGF has been implicated as a contributing factor in a number of pathologic conditions that involve the migration and proliferation of PDGF-responsive cells. Such conditions include arteriosclerosis (8), restenosis following coronary bypass surgery or balloon angioplasty (9), nephritis (10), scleroderma (11), and some neoplasias (12). Thus, interfering with the biologic activities of PDGF may be of therapeutic value for one or more of these conditions.
We have generated and characterized a panel of murine monoclonal antibodies against the extracellular portion of the PDGFR␤ and examined the ability of these antibodies to inhibit PDGF BB-specific binding and/or induction of mitogenesis. One of these antibodies exhibits no effect on PDGF BB binding to the receptor but does inhibit PDGF-mediated mitogenesis by impairing receptor dimerization. This antibody cross-reacts with PDGFR␤ from other species, making it an ideal candidate to study the therapeutic potential of an antibody PDGF antagonist in animal models of human disease.

EXPERIMENTAL PROCEDURES
Cell Lines and Cytokines-The Chinese hamster ovary cell line CHO/ dhFr Ϫ and the rabbit cornea cell line SIRC were obtained from the American Type Culture Collection. Baboon primary aortic smooth muscle cells (SMC) were kindly provided by J. Anderson and S. Hanson at Emory University (Atlanta, GA). Purified PDGF BB was obtained from Boehringer Mannheim.
Transfectants Producing Soluble and Membrane-bound PDGFR␤-A gene encoding the PDGFR␤, lacking the nucleotides encoding a portion of the 5Ј end, was obtained from the American Type Culture Collection. The missing portion of the gene, which included the secretion signal sequence, was constructed by oligonucleotide synthesis and used to assemble the complete PDGFR␤ gene with an XbaI site at each end. This fragment was inserted into the XbaI site of the plasmid pVk (13) and co-transfected together with plasmid pVgl (13), which contains a dhfr gene, into CHO/dhFr Ϫ cells using the calcium phosphate method essentially as described (14). Methotrexate-resistant transfectants expressing the PDGFR␤ were identified by indirect immunofluorescence using a commercially available antibody (Genzyme) and cloned by sin-* 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.
gle cell sorting using a FACStar PLUS (Becton Dickinson). The resulting cell line (CHO C4) was responsive to PDGF BB as evidenced by increased DNA synthesis and cell proliferation in the presence of the growth factor as demonstrated previously (15).
To produce a soluble form of PDGFR␤ (sPDGFR␤), the gene encoding the protein was truncated immediately before the transmembrane sequence, and a stop codon followed by an XbaI restriction endonuclease site was inserted. This gene fragment was inserted into the XbaI site of pVk. The resulting plasmid was cotransfected as above into CHO/dhFr Ϫ cells together with plasmid pVgl. Transfectants resistant to methotrexate were selected and used for production of sPDGFR␤ by growing to confluence in Dulbecco's modified Eagle's medium plus 10% fetal calf serum, then replacing this serum-containing medium with protein-free medium and incubating for an additional 72 h. The exhausted medium was harvested and passed over a wheat germ agglutinin-Sepharose column. The column was washed and eluted with N-acetyl-D-glucosamine. The eluant containing sPDGFR␤ was dialyzed and concentrated to approximately 300 g/ml. The purified protein was Ͼ95% pure based on SDS-polyacrylamide gel electrophoresis analysis.
Monoclonal Antibodies-A panel of monoclonal antibodies was generated against the extracellular portion of the human PDGFR␤ by immunizing outbred Swiss Webster mice with 50 g of purified sP-DGFR␤ in RIBI adjuvant (ImmunoChem Research). The mice received booster immunizations of 50 g of PDGFR␤ every 1-2 weeks. Mice were bled 1 week following each boost, and the sera were tested for reactivity with sPDGFR␤ by an ELISA. The mouse exhibiting the highest serum titer of anti-PDGFR␤ activity was sacrificed 3 days after receiving a final boost of 50 g of sPDGFR␤, and hybridomas were prepared. Hybridoma supernatants were tested for reactivity with sPDGFR␤ by an ELISA (see below). Hybridomas that exhibited reactivity were expanded and cloned. All monoclonal antibodies reactive with soluble PDGFR␤ were assayed for reactivity against cell surface-expressed PDGFR␤ on the surface of CHO C4 transfectants by flow cytometry. Only antibodies demonstrating reactivity with cell surface-expressed PDGFR␤ were further characterized.
ELISA-Immulon 1 96-well plates (Dynatech Laboratories Inc.) were coated with sPDGFR␤ by adding 100 l of a 0.5 g/ml solution of sPDGFR␤ in phosphate-buffered saline to each well. After an overnight incubation at 4 C°, 200 l of phosphate-buffered saline plus 1% bovine serum albumin and 0.5% Tween 20 were added to each well to block unoccupied protein binding sites. After a 1-h incubation at room temperature, the wells were washed three times with phosphate-buffered saline plus 1% Tween 20. Dilutions of purified antibody or antibodycontaining supernatants (100 l) were added to each well, and the plate was incubated for 1 h at room temperature, after which it was washed three times as described above. One hundred l of a 1 g/ml solution of horseradish peroxidase-conjugated GAMIgG (Tago, Inc.) was added to each well, and the plates were incubated for 1 additional h at room temperature, after which they were washed three times. One hundred l of peroxidase substrate (Bio-Rad) was added to each well, the plate was incubated for 15-60 min, and absorbance at 415 nm was determined.
Antibody Inhibition of Radiolabeled PDGF BB Binding to Cells-CHO C4 cells were harvested, washed twice with cold Dulbecco's modified Eagle's medium, and resuspended at 10 6 cells/ml in Dulbecco's modified Eagle's medium. The assay was carried out in triplicate by incubating 100 l of the cell suspension with either no antibody (to determine maximum binding), 5 g of the indicated antibody, or 100 ng of cold PDGF BB (to saturate specific binding sites and determine nonspecific binding) in 12 ϫ 75 mm polystyrene tubes for 15 min at 4 C°. Each sample received 1.0 ng of 125 I-labeled PDGF BB (Amersham Life Science, Inc.) and was incubated for an additional 60 min at 4 C°. Unbound 125 I-labeled PDGF BB was separated from that bound to the cell by layering the sample over a mixture of 80% dibutylphthalate, 20% olive oil in a Reagiergefä ␤ test tube (Sarstedt, Inc.) and microfuging briefly to pellet the cells through the oil mixture (16). The tubes were placed in dry ice to freeze the contents, the tip of the tube containing the cell pellet was cut off into a vial, and radioactivity was determined in a gamma counter.
Antibody Inhibition of PDGF BB-mediated Proliferation-Antibody neutralization of PDGF BB-mediated proliferation was assessed using CHO C4 cells. The amount of PDGF BB that induced 90% maximum proliferation (25-50 ng/ml) as measured by [ 3 H]thymidine incorporation was selected for use in the assay. All assays were carried out in triplicate or quadruplicate. The assay involved preparing a 96-well plate containing 50,000 CHO C4 cells/well in Ham's F-12 plus 10% fetal calf serum. After a 24-h incubation, the medium was replaced with Ham's F-12 plus 0.1% bovine serum albumin, and an additional 24-h incubation served to put the cells into a quiescent state. Varying concentrations of anti-PDGFR␤ or control antibodies were added. Following a 2-h incubation with antibody, PDGF BB was added. The cells were incubated overnight, then 1 Ci of [ 3 H]thymidine was added to each well. The cells were incubated for an additional 4 h then harvested using a PHD cell harvester (Cambridge Technology, Inc.). [ 3 H]Thymidine incorporation in each well was determined with a scintillation counter. The same assay was used to measure the inhibitory activity of the antibodies on baboon SMC and the rabbit cornea cell line SIRC.
Domain Deletion Mutants of PDGFR␤-Domain deletion mutants of PDGFR␤ were prepared by cloning the PDGFR␤ gene fragment (with various extracellular domains deleted) into the XbaI site of a version of pVgl that had the IgG1 constant region replaced with the human lambda constant region cDNA. This allowed production of fusion proteins that contained varying numbers of the PDGFR␤ extracellular domains with the human lambda light chain constant domain at the carboxyl terminus. The mutants included only the first, the first and second, the first through third, and the first through fourth extracellular domains of PDGFR␤ fused to the human lambda light chain constant domain. The lambda constant domain served as a tag and allowed the deletion mutants to be captured onto the surface of an ELISA plate. The five-domain extracellular portion of the human PDGFR␤ with no lambda constant domain (sPDGFR␤) served as the positive control. The domain deletion mutants were captured onto an ELISA plate using a goat anti-human lambda chain antiserum (Tago). Deletion mutant binding was confirmed using a horseradish peroxidase-conjugated antihuman lambda antiserum. The sPDGFR␤ (domains 1-5) was directly coated onto the plate as described above. Reactivity of the antibodies with the domain deletion mutants was determined by incubating antibody with the plate-bound deletion mutants for 1 h, washing away unbound antibody, and developing the assay with a horseradish peroxidase-conjugated GAMIgG.
Analysis of PDGFR␤ Dimerization Status-The dimerization status of PDGFR␤ on the surface of CHO C4 cells exposed to PDGF BB was carried out in a manner similar to that described previously (17).

RESULTS
Monoclonal Antibodies against PDGFR␤ Differ in the Ability to Prevent PDGF BB Binding and Proliferation-We generated a panel of monoclonal antibodies against the extracellular portion of the human PDGFR␤ and assayed the antibodies both for the ability to inhibit radiolabeled PDGF BB binding to the receptor as well as the capacity to inhibit PDGF BB-mediated cell proliferation. The antibodies examined fell into three categories. Fig. 1 presents the results for a representative example of each category. Several antibodies, such as M4TS.15, had little or no effect on PDGF BB binding and likewise did not inhibit PDGF-mediated proliferation. Another class of antibodies, represented by M4TS.22, inhibited both PDGF BB binding to PDGFR␤ and PDGF BB-mediated proliferation. A single antibody (M4TS.11) exhibited no inhibitory activity in the PDGF BB binding assay; however, it did display inhibition of PDGF BB-mediated proliferation. Despite the difference in the ability to inhibit binding of PDGF BB to cells expressing PDGFR␤, M4TS.11 and M4TS.22 were indistinguishable in their ability to inhibit PDGF BB-mediated proliferation (Fig. 1B).

M4TS.11 Prevents PDGF BB-mediated Proliferation by Impairing PDGFR␤ Dimerization-
The easily detectable ability of M4TS.11 to inhibit PDGF BB-mediated proliferation in the absence of the ability to block ligand interaction with receptor implied that the antibody was influencing an event post-ligand binding that was a prerequisite for the induction of the mitogenic signal. PDGFR␤ is a receptor tyrosine kinase whose activity is dependent upon ligand-mediated dimerization (5).
To determine if M4TS.11 influenced the ability of PDGF BB to induce receptor dimerization, we examined the status of PDGFR␤ on cells exposed to PDGF BB in the presence and absence of this antibody. CHO C4 cells were exposed to 125 Ilabeled PDGF BB after a preincubation with no antibody or with M4TS.11 or M4TS.15. The cells were then treated with bis(sulfosuccinimidyl) suberate (Pierce) to covalently cross-link the 125 I-PDGF BB⅐PDGFR␤ complex (17). These cells were lysed with Nonidet P-40, and lysate aliquots were subjected to SDS-polyacrylamide gel electrophoresis on a 6% gel. A 6% gel was selected to allow migration of all cross-linked complexes into the gel, and this was confirmed by the lack of radioactivity detected at the top of the gel as well as the ability to recover between 86 and 94% of the counts loaded from the gel lanes (data not shown). Density scans of autoradiographs of the gels indicated that radioactive PDGF BB predominantly migrated in the two areas of the gel that corresponded to the molecular weight of the 125 I-labeled PDGF BB cross-linked to one (monomer) or two (dimer) PDGFR␤ molecules (unbound 125 I-labeled PDGF BB runs off the gel). Cells preincubated with M4TS.11 before exposure to 125 I-labeled PDGF BB and cross-linking had approximately 50% that of the level of receptor dimer (with a corresponding 100% increase in receptor monomer) as compared with cells exposed to PDGF BB in the presence M4TS.15 or no antibody (Fig. 2).
Mapping Antibody Reactivity Using PDGFR␤ Domain Deletion Mutants-To identify the portion of the PDGFR␤ recognized by the M4TS antibodies, domain deletion mutants of the extracellular portion of PDGFR␤ were constructed and expressed in a soluble form. The deletion mutants included the first, first and second, first through third, and first through fourth extracellular domains of PDGFR␤ fused to the human lambda immunoglobulin constant domain (see "Experimental Procedures"). Each deletion mutant was tested for reactivity with M4TS.11, M4TS.15, or M4TS.22. This analysis revealed that M4TS.11 exhibited reactivity only when extracellular domain 4 of the PDGFR␤ was present (Fig. 3). M4TS.15 required the presence of extracellular domain 2, whereas reactivity with M4TS.22 was dependent on the presence of extracellular domain 3 (Fig. 3).
Reactivity of M4TS Antibodies with PDGFR␤ from Various Species-It was of interest to determine if the M4TS.11 and M4TS.22 antibodies could react with the PDGFR␤ from species other than human and if they would function as a PDGF BB antagonist for these species. PDGFR␤ expression has been reported for a variety of cell types derived from a number of mammalian species, and human PDGF BB can induce proliferation in these cells. Baboon SMC (18)  those that had little or no effect on ligand binding or PDGF BB-induced proliferation, 2) those that prevented interaction of the ligand with the receptor and thus prevented ligand-induced proliferation, and 3) a single antibody, M4TS.11, that had no detectable effect on ligand binding but was a potent inhibitor of ligand-induced proliferation.
The ability of M4TS.11 to inhibit mitogenesis in the absence of any detectable effect on PDGF BB binding to the receptor implied that the antibody was interfering with an event required for triggering mitogenesis but distinct from ligand binding. Ligand-induced dimerization of transmembrane tyrosine kinases such as PDGFR␤ is requisite for transmission of the mitogenic signal (6). Examination of the status of PDGFR␤ on cells preincubated with M4TS.11 and then exposed to PDGF BB revealed a decrease in the level of receptor dimer and a corresponding increase in receptor monomer as compared with the levels observed in cells preincubated with a control antibody (Fig. 2). This indicates that M4TS.11 impairs the ability of PDGF BB to induce receptor dimerization, a characteristic that makes it a PDGF BB antagonist that is equivalent in potency to an antibody such as M4TS.22 that directly inhibits PDGF BB binding.
Deletion mutants allowed us to map the determinant recognized by M4TS.11, M4TS.15, and M4TS.22 to the fourth, second, and third extracellular Ig-like domains, respectively. Domain 3 is required for M4TS.22 binding, indicating that the antibody recognizes a determinant that either resides in domain 3 or is composed of portions of domain 3 and domain(s) 1 and/or 2. The first three domains of PDGFR␤ are required to form the ligand binding site (20), thus M4TS.22 reacts with the receptor at a site near to, or possibly identical with, that part of PDGFR␤ that binds PDGF BB. This is consistent with the observation that M4TS.22 can inhibit PDGF BB-induced proliferation by preventing the interaction of the ligand with the receptor. M4TS.15, an antibody that has no effect on PDGF BB binding or proliferation, reacts with each of the deletion mutants that contains domain 2. Thus, despite the involvement of domain 2 in forming the ligand binding site, the portion recognized by M4TS.15 is spatially distinct from that which interacts with PDGF BB. The requirement of domain 4 for reactivity with M4TS.11 indicates that the determinant recognized by this antibody is distinct from the ligand binding portion of the PDGFR␤ formed by domains 1-3 and suggests that domain 4, despite being uninvolved in PDGF BB binding (20), does participate in transmitting the mitogenic signal to the cell. Each of these antibodies appears very different from the anti-PDGFR␤ monoclonal antibody 2A1E2, which inhibits ligand binding and mitogenesis but reacts with the fifth extracellular domain of PDGFR␤ (21).
The mechanism by which M4TS.11 inhibits PDGF BB-induced mitogenesis appears very similar to that of antibodies that have been described by Blechman et al. (17,22) against the stem cell factor receptor (SCFR). The SCFR, which is the product of the c-kit proto-oncogene, is a receptor protein kinase that, like PDGFR␤, transmits a signal when dimerized by the divalent ligand, SCF (23). Antibodies that bind to the fourth Ig-like domain of SCFR inhibit SCF-induced dimerization but have no effect on SCF binding (17,22). These findings, coupled with the inability of a domain-four deletion mutant of SCFR to dimerize in the presence of SCF, have led to the hypothesis that divalent ligand binding per se is insufficient for receptor dimerization but rather exposes an intrinsic dimerization site on the receptor that mediates subsequent dimerization and thus, signal transmission (17). The behavior of M4TS.11, which binds to the analogous domain four of the PDGFR␤, is consistent with the above hypothesis and supports the contention that the proposed mechanism may be a general feature of receptor tyrosine kinase signaling (17). In addition, the reactivity of M4TS.11 with PDGFR␤ from a wide variety of species indicates that the determinant recognized by the antibody is evolutionarily conserved. This is a characteristic expected of a portion of the molecule essential for proper function.
The unique ability of M4TS.11 to inhibit PDGF BB-mediated proliferation while allowing the interaction of this ligand with PDGFR␤ may make it an especially efficient PDGF BB antagonist in vivo, as local fluctuations in the concentration of PDGF BB should have no effect on the binding of antibody to the receptor. An antibody that inhibits PDGF BB-induced proliferation by blocking ligand binding could theoretically be displaced by high local concentrations of the ligand, possibly negating the inhibitory effect and making it a less effective antagonist than M4TS.11.
The involvement of PDGF in a variety of human disease conditions makes it an attractive target for therapy, and neutralizing the activity of PDGF with an antibody is a viable strategy. Advances over the past decade have made human treatment with monoclonal antibodies a feasible therapeutic approach. The major complication of the human anti-murine antibody response has largely been eliminated by "humanization" techniques, which transform a murine monoclonal antibody into a molecule that is indistinguishable from a human immunoglobulin (24). The excellent safety profile and extended half-life of such antibodies (25) make them ideal candidates for applications such as interfering with the PDGF system. We have successfully humanized both the M4TS.11 and M4TS. 22 antibodies and retained all the characteristics of the original murine versions, 3 thereby making these antibodies excellent candidates for potential development as therapeutic agents.