Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

PDGF is a potent mitogen and chemotactic factor for a variety of mesenchymal cells such as fibroblasts, vascular smooth muscle cells, glomerular mesangial cells, and brain glial cells. It has been implicated in a wide variety of pathological conditions, including atherosclerosis, restenosis, glomerulonephritis, liver cirrhosis, pulmonary fibrosis, and cancer (Wilcox et al., 1988; Heldin and Westermark, 1990; Iida et al., 1991; Johnson et al., 1992; Wong et al., 1994; Yagi et al., 1998; Rice et al., 1999). PDGF is a disulfide-linked dimer of two related polypeptide chains, designated A and B, which are assembled as heterodimers (PDGF-AB) or homodimers (PDGF-AA and PDGF-BB) (Heldin et al., 1992; Fretto et al., 1993; Herren et al., 1993). PDGF exerts its biological activity by binding to structurally similar - or -PDGFRs, and inducing receptor dimerization. The receptor binding specificity for the PDGF isoforms dictates that PDGF-AA induces only / receptor dimers, PDGF-AB induces / and / receptor dimers, and PDGF-BB induces all three receptor dimer combinations (Hart et al., 1988; Heidaran et al., 1991; Heldin, 1992). Once dimerized, the PDGFR undergoes transphosphorylation on tyrosine that creates the sites for physical interaction with a number of proteins that contain a Src homology two domain (Claesson-Welsh, 1994). These phosphorylation-dependent interactions are essential for the activation of intracellular signaling pathways that mediate changes in gene expression, cell migration, and proliferation that in some instances lead to tissue fibrosis or cancer. Therefore, small-molecule PDGFR kinase inhibitors could have a broad therapeutic application. Here we describe the identification of CT52923, a piperazinyl quinazoline that is a novel potent and orally active PDGFR kinase inhibitor. This compound demonstrates remarkable specificity for blocking PDGF-induced receptor phosphorylation, cell migration, and cell proliferation.

Restenosis is the major limiting complication of interventional procedures for improving blood flow through obstructed arteries and small caliber arterial grafts. Using animal models of restenosis, we and others have demonstrated that PDGF plays a major role in the vascular response to injury (Ferns et al., 1991; Sirois et al., 1997; Bilder et al., 1999; Giese et al., 1999). Therefore, the rat carotid artery injury model was used to demonstrate the ability of CT52923 to block in vivo PDGFR signaling and significantly reduce neointima formation. These results imply that this approach could have therapeutic potential for the prevention of clinical restenosis.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. CT52923 was synthesized as described (Matsuno et al., 1998). Its chemical structure is shown in Fig. 1. A stock solution of 3 mM CT52923 was prepared in dimethyl sulfoxide and stored at 20°C. Dilutions for all assays were made fresh prior to use.

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Chemical structure of CT52923. CT52923 is 4-(6,7-dimethoxy-4-quinazolinyl)-N-(3,4-methylenedioxybenzyl)-1-piperazinethiocarboxamide (molecular formula is C23H25N5O4S, mol.wt. = 467.5).

Expression, Purification, and Enzymatic Characterization of Recombinant PDGFR Kinase Domain. A construct encoding residues 572 to 1027 of cytoplasmic domain of human PDGFR was cloned into a baculovirus expression vector, pBlueBAcHis2 (Invitrogen, San Diego, CA). Production of PDGFR kinase protein in infected Sf21 insect cell was done according to manufacturer's standard protocol. The PDGFR kinase protein was then purified to near homogeneity by sequential chromatographic methods using Ni-NTA agarose (Qiagen, Valencia, CA), Q-Sepharose, and S-Sepharose (Amersham Pharmacia Biotech, Piscataway, NJ) ion exchange columns. Subsequently, the protein was concentrated to a final level of 200 µg/ml for in vitro kinase assay.

Isolation of Receptor cDNAs, Construction of Chimeric Receptors, and Expression of Encoded Proteins in CHO Cells. Chimeric receptor constructs encoding the PDGFR extracellular and transmembrane domains fused to the cytoplasmic domain of CSF-1R, c-Kit, Flt3, and VEGFR-2 were generated. The PDGFR cDNA was isolated from a human placental cDNA library (Escobedo et al., 1988). The cDNAs encoding VEGFR-2 and Flt3 were obtained by RT-PCR using poly(A)⁺ mRNA of human umbilical vein endothelial cells and THP-1 human monocytes as the template, respectively. The cDNAs encoding human CSF-1R and mouse c-Kit were kindly provided by Dr. Mohammad Heidaran of the National Institutes of Health (Bethesda, MD) (Yu et al., 1994). Using standard molecular biology techniques, codons 1 to 556 of the PDGFR were fused to codons 538 to 972 of CSF-1R, codons 544 to 975 of c-Kit, codons 563 to 993 of Flt3, or codons 787 to 1356 of VEGFR-2. The human FGFR-1 IIIc isoform cDNA was obtained by RT-PCR using total RNA of KATO-III cell as a template. FGFR-1 and each of the chimeric receptor constructs were sequenced by automated dye-terminator cycle sequencing using Ampli-Taq FS and inserted into the mammalian expression vector pBJ-1 (Lokker et al., 1997).

Receptor Autophosphorylation Assays. CHO cell lines expressing wild-type PDGFR or PDGFR/c-Kit, PDGFR/CSF-1R, PDGFR/Flt3, and PDGFR/VEGFR-2 chimeric receptors were grown to confluency in 96-well microtiter plates under standard tissue culture conditions, followed by serum starvation for 16 h. Phosphorylation assays were performed as previously described with minor modifications (Lokker et al., 1997). Briefly, quiescent cells were incubated at 37°C with increasing concentrations of CT52923 (0.01-30 µM) for 30 min followed by the addition of 8 nM PDGF-BB for 10 min. Cells were lysed in 100 mM Tris, pH 7.5, 750 mM NaCl, 0.5% Triton X-100, 10 mM sodium pyrophosphate, 50 mM NaF, 10 µg/ml aprotinin, 10 µg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium vanadate, and the lysate was cleared by centrifugation at 15,000g for 5 min. Clarified lysates were transferred into a second microtiter plate in which the wells were previously coated with 500 ng/well of 1B5B11 anti-PDGFR mAb (Lokker et al., 1997), and then incubated for 2 h at room temperature. After washing three times with binding buffer (0.3% gelatin, 25 mM HEPES, pH 7.5, 100 mM NaCl, 0.01% Tween 20), 250 ng/ml of rabbit polyclonal anti-phosphotyrosine antibody (Transduction Laboratories, Lexington, KY) was added and plates were incubated at 37°C for 60 min. Subsequently, each well was washed three times with binding buffer and incubated with 1 µg/ml of horseradish peroxidase-conjugated anti-rabbit antibody (Roche Molecular Biochemicals, Indianapolis, IN) at 37°C for 60 min. Wells were washed prior to adding 2,2'-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (Sigma, St. Louis, MO), and the rate of substrate formation was monitored at 650 nm.

Phosphorylation Assays for Cytoplasmic Protein Kinases. Purified Src, PKA, and PKC kinase phosphorylation of a specific substrate peptide was assayed in the absence or presence of CT52923 (0.01-30 µM) according to the protocol provided by the manufacturer (Upstate Biotechnology).

[³H]Thymidine Incorporation Assay. NIH/3T3 cells (1 × 10⁴) were plated in each well of microtiter plates and grown in Dulbecco's modified Eagle's medium containing 10% calf serum for 7 days. Cells were stimulated with 0.6 nM PDGF-BB or 0.28 nM bFGF in the presence of increasing concentrations of CT52923 (0.01-30 µM). After incubation for 17 h at 37°C, 2 µCi/well of [H³]thymidine was added and the incubation was continued for another 5 h. Cells were then washed twice with 200 µl of phosphate-buffered saline, twice with 200 µl of 5% trichloroacetic acid, lysed by the addition of 100 µl of 50 mM NaOH, 0.01% SDS, and 150 µl of scintillation fluid was added to each well. The amount of incorporated [³H]thymidine was determined by scintillation counting. In the absence of CT52923, PDGF-BB caused an 8.5-fold increase in [³H]thymidine incorporation where as for bFGF the increase was 26.2-fold.

Cell Migration Assay. Costar transwell plates containing membranes with pore size of 5 µm were used to measure the migration of rat A10 smooth muscle cells. Cells (8 × 10⁴) were plated in top chamber while PDGF-BB (4 nM) or bFGF (1.7 nM) was added in bottom chamber and various concentrations of CT52923 (0.01-30 µM) were added to both top and bottom chambers. Following overnight incubation at 37°C, cells remaining on the top side of the membrane were wiped off by using a cotton swab. Cells that had migrated through the pore and attached to the bottom side of the membrane were removed with 0.5% trypsin, 5.3 mM EDTA (Invitrogen) and counted. In the absence of inhibitor, PDGF-BB and bFGF induced an 8.0- and 32.9-fold increase in cell migration, respectively.

Rat Carotid Artery Balloon Angioplasty. Rat carotid balloon catheter denudation was performed by the standard protocol (Ferns et al., 1991). Briefly, Lewis rats (200-300 g) were anesthesized with ketamine (75 mg/kg), xylazine (8 mg/kg), and acepromazine (91.5 mg/kg). A cervical midline incision was made, the left common carotid artery was isolated, and a 2F Fogarty embolectomy catheter was inserted through the external branch into the carotid artery to the aortic arch. The balloon was withdrawn slowly while inflated and the procedure repeated three more times. After ballooning, the side branch was ligated and the incision was closed.

Administration of CT52923 to Rats. CT52923 was suspended in 0.5% methyl cellulose and given by oral gavage twice daily for a total dose of 10, 30, 60, and 100 mg/kg/day while the control group received vehicle alone. Dosing was started just prior to balloon injury and continued until 14 days postinjury when the vessels were harvested. There were eight animals per group except for the 60-mg/kg/day group, which had seven animals.

Tissue Harvest and Morphometric Analysis. Carotid arteries were harvested fresh 14 days postinjury; each artery was divided into four equal segments, which were fixed overnight in 10% formalin solution and paraffin embedded. Paraffin cross sections from the four segments of each artery were stained with Mayer's hematoxylin-eosin and morphometric measurements were performed using a Zeiss Axioshop microscope coupled to a Sony charge-coupled device/red green blue color video camera. Average medial and neointimal areas were determined for each artery using KS300 image analysis software (Carl Zeiss Inc., Thornwood, NY).

Statistical Analysis. The medial and neointimal areas were expressed as mean ± S.E.M. Statistical comparisons were determined using a one-tailed Student's t test. Data were considered to be different if p < 0.05 was observed.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Mechanism of PDGFR Kinase Inhibition by CT52923. The nature of the PDGFR kinase inhibition by CT52923 was studied using purified receptor kinase domain expressed in insect cells. Enzyme inhibition analysis of PDGFR autophosphorylation revealed that CT52923 was competitive with the ATP substrate. We observed a definite decrease of CT52923 potency with increasing concentration of ATP (Fig. 2). The IC₅₀ values of CT52923 measured with 5, 50, and 500 µM ATP were 0.017, 0.4, and 0.65 µM, respectively. The transformed data presented as a Dixon plot indicated that the K_i was approximately 3 nM (Fig. 2, inset).

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Biochemical mechanism of action of CT52923. PDGFR kinase was expressed using a baculovirus system and purified. In vitro kinase reaction was performed to measure the inhibition activity of CT52923 on PDGFR kinase autophosphorylation in the presence of 5 µM ATP (), 50 µM ATP (), and 500 µM ATP (). Points represent the mean of triplicate determinations. The curve was fit by using a sigmoidal dose response nonlinear regression analysis algorithm (Prism software; GraphPad Software Inc., San Diego, CA). Data were transformed to a 1/v versus [inhibitor] Dixon plot where velocity is reported in micromoles per minute. The Ki value was estimated from the intersection of the lines extrapolated to abscissa.

Specificity and Potency of PDGFR Kinase Inhibitor CT52923. Since CT52923 and nearly all other kinase inhibitors described to date compete with the common substrate ATP for enzyme binding, achieving a high degree of specificity could be problematic (Levitzki and Gazit, 1995). Other than the PDGFRs, the most likely enzymes to be inhibited by CT52923 are the closely related family members c-Kit, CSF-1R, and Flt3. These proteins share 60 to 70% amino acid sequence identity with PDGFR in the kinase domains (Qiu et al., 1988). To accurately assess the relative kinase inhibitory activity of CT52923 against all PDGFR family members, we generated cDNAs that encode chimeric receptors of the extracellular and transmembrane domains of PDGFR fused to the intracellular domain of each of the other family members (see Materials and Methods). The wild-type PDGFR and each of the chimeric cDNAs were stably transfected into CHO cells and clones expressing 25,000 to 50,000 recombinant receptors/cell were isolated. Each cell line was stimulated with a saturating concentration of PDGF-BB (8 nM) in the presence of increasing concentrations of CT52923 and the cell lysates were prepared. To measure the level of receptor autophosphorylation, lysates were subjected to analysis with a previously described two-site enzyme-linked immunosorbent assay that uses an anti-PDGFR capturing antibody followed by detection with anti-phosphotyrosine antibody (Lokker et al., 1997).

CT52923 Specifically Inhibits PDGF-Induced Mitogenesis and Chemotaxis. To further evaluate the specificity of CT52923, its ability to inhibit cell proliferation and migration in response to PDGF-BB or bFGF was measured because these two growth factors have been shown to induce very similar signaling pathways, changes in gene expression, and cellular responses in mesenchymal cells (Fambrough et al., 1999). For cell proliferation, NIH3T3 cells were stimulated with PDGF-BB or bFGF in the presence of increasing concentrations of CT52923. The average fold increase of thymidine incorporation induced by PDGF-BB or bFGF in the absence of CT52923 was 8.5- and 26.2-fold, respectively (data not shown). As shown in Table 2, CT52923 inhibited PDGF-induced thymidine incorporation with an IC₅₀ of 280 nM, whereas the IC₅₀ for bFGF was 46-fold higher at 13 µM (p < 0.001).

Orally Administered CT52923 Inhibits PDGF-Mediated Response to Vascular Injury. Using animal models of restenosis, we and others have shown that PDGF signaling plays a major role in the formation of neointima (Ferns et al., 1991; Sirois et al., 1997; Bilder et al., 1999; Giese et al., 1999). Therefore, we have used the standard rat carotid artery balloon angioplasty model to evaluate the ability of CT52923 to inhibit PDGFR signaling in vivo as measured by a reduction in neointima formation. Lewis rats were dosed by oral gavage with vehicle or CT52923 at 5, 15, 30, and 50 mg/kg twice daily. Dosing was started just prior to injury and continued until 14 days postinjury when the animals were sacrificed. Table 3 shows that treatment with increasing doses of CT52923 had little or no effect on medial areas, but it caused a significant dose-dependent reduction in neointimal areas, which reached a maximum of 30% reduction (p < 0.01) at the highest dose. Alternatively, measurement of intima/media ratios to normalize for any variation in vessel size gave a comparable dose-dependent reduction reaching 27% at the highest dose (Table 3). To ensure that adequate dosing had been achieved, plasma samples were collected 6 h after the last dose and CT52923 levels were measured. The CT52923 plasma concentrations were 0, 2.7, 4.0, and 8.8 µM for the 5-, 15-, 30-, and 50-mg/kg doses, respectively (data not shown). In the presence of rat plasma, CT52923 inhibits PDGFR phosphorylation in cultured cells with an IC₅₀ around 2 µM, indicating that a significant level of inhibitory activity was maintained for at least 12 h each day at the three highest doses.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We have identified a novel piperazinyl quinazoline, CT52923, that is a potent and specific PDGFR kinase inhibitor. When CT52923 was tested against the closely related PDGFR family of kinases, PDGFRs and c-Kit were inhibited with an IC₅₀ of 100 to 200 nM, while 30- to >300-fold higher concentrations were required to inhibit Flt3 and CSF-1R, respectively. Other receptor tyrosine kinases, cytoplasmic tyrosine kinases, serine/threonine kinases, or members of the MAPK cascades were not significantly inhibited at 100- to 1000-fold higher concentrations than those required to inhibit the PDGFR. CT52923 also demonstrates specificity for inhibition of cellular responses. PDGF-induced smooth muscle cell migration or fibroblast proliferation was shown to be blocked by CT52923 with an IC₅₀ of 64 and 280 nM, respectively, whereas 50- to 100-fold higher concentrations were required to inhibit these responses when induced with FGF. These studies indicate that CT52923 does not inhibit kinases critical for cell cycle progression, DNA replication, or the reorganization of cytoskeletal proteins involved in cell movement.

The specificity profile of CT52923 demonstrates it is a highly PDGFR-specific inhibitor. To date, several other potent and selective PDGFR kinase inhibitors have been reported, including phenylaminopyrimidine derivative STI571 (Carroll et al., 1997), quinoline derivative Ki6896 (Yagi et al., 1997), quinoxaline derivative RPR101511A (Bilder et al., 1999), and indolocarbazole analog 3744W (Spacey et al., 1998). Among them, only STI571 was tested against a similarly broad panel of kinases (Carroll et al., 1997). Like CT52923, STI571 inhibits PDGFR and c-Kit, but not Flt3 or CSF-1R. However, STI571 inhibits Abl kinase whereas CT52923 does not, indicating the latter has a greater level of specificity for the PDGFR.

Over the past several years it has become increasingly evident that PDGF plays an important role in a wide variety of diseases. PDGFR signaling leading to pathological complications can be mediated by several different mechanisms. On one extreme, chromosomal translocations within the PDGFR gene have been identified that cause constitutive receptor activation leading to the development of chronic myelomonocytic leukemia (Golub et al., 1994). More subtle alterations in PDGFR signaling can occur through the establishment of an autocrine loop due to inappropriate PDGFR expression, which has been shown to occur routinely in human tumors of mesenchymal origin, especially glioblastoma (Hermanson et al., 1992; Guha et al., 1995). In these cases, blockade of PDGFR signaling using antisense oligonucleotides, anti-PDGF antibodies, dominant negative PDGFR, or tyrosine kinase inhibitors has been shown to inhibit tumorgenicity (Nitta and Sato, 1994; Strawn et al., 1994; Todo et al., 1996; Kilic et al., 2000). We have found that CT52923 blocks autocrine phosphorylation of PDGFR in glioblastoma cell lines and prevents xenograft tumor formation by these cells in nude mice (N. A. Lokker, C. Sullivan, J. C. Yu, J. O'Hare, S. Hollenbach, and N. A. Giese, unpublished data).

In addition to these instances of aberrant PDGFR signaling, an exaggeration of the normal PDGF-mediated healing process can occur due to chronic injury or inflammation that leads to pathological tissue fibrosis. Examples of such diseases that involve excessive PDGFR signaling include glomerulonephritis, liver cirrhosis, pulmonary fibrosis, atherosclerosis, and transplant vasculopathy (Wilcox et al., 1988; Heldin and Westermark, 1990; Iida et al., 1991; Johnson et al., 1992; Wong et al., 1994; Yagi et al., 1998; Rice et al., 1999; Sihvola et al., 1999). In each of these cases there is a preexistent insult to tissues by chemicals, viruses, immune complexes, or inflammatory cells that causes a dramatic up-regulation of PDGFR expression by resident mesenchymal cells and the local release of PDGF from endothelial cells, platelets, and macrophages (Wilcox et al., 1988; Iida et al., 1991; Wong et al., 1994). Evidence that PDGFR antagonism may be an effective strategy for treating these fibrotic diseases has been suggested by animal studies. A significant inhibition of disease progression was observed in models of pulmonary fibrosis (Rice et al., 1999), transplant atherosclerosis (Sihvola et al., 1999), and glomerulonephritis (Yagi et al., 1998) when animals were treated parenterally with a PDGFR kinase inhibitor.

Another example of a pathological response to injury is restenosis, the major complication limiting the usefulness of interventional procedures for improving blood flow through obstructed coronary arteries. It is well recognized that these procedures, especially stent placement, cause injury to the vessel wall, inducing the smooth muscle cell migration and proliferation that results in reocclusion for 20 to 30% of cases (Serruys et al., 1994). Since PDGF is the most potent mitogen and chemotactant known for smooth muscle cells its role in the response to vascular injury has been studied extensively. We have recently demonstrated that 2A1E2, a neutralizing monoclonal antibody directed against the PDGFR, could effectively block neointima formation following carotid endarterectomy, femoral artery balloon angioplasty, or stent placement in primates (Giese et al., 1999; data not shown). A major role for PDGF in the response to vascular injury is further supported by studies in lower species in which inhibition of PDGF signaling by neutralizing antibodies to PDGF, antisense to PDGFR, or PDGFR kinase inhibitors blocked neointima formation following balloon angioplasty (Ferns et al., 1991; Sirois et al., 1997; Bilder et al., 1999; Giese et al., 1999). Therefore, the ability of CT52923 to block PDGFR signaling in vivo was demonstrated using the rat carotid artery balloon injury model. Oral administration of CT52923 resulted in a significant dose-dependent reduction in neointima formation that reached 30% at the highest dose. These results are very comparable to the 50% reduction in neointima formation we observed due to PDGFR blockade in the baboon following femoral artery balloon injury (Giese et al., 1999).

Kinase inhibition represents a novel mechanism-based approach to selectively block signaling pathways that are known to mediate disease processes. Clinically, there is evidence that this strategy will prove to be successful for the treatment of chronic myelogenous leukemia. This disease involves the Philadelphia chromosome translocation leading to expression of the Bcr/Abl protein that is a constitutive active tyrosine kinase (Sawyers, 1999). STI571, a phenylaminopyrimidine inhibitor of Bcr/Abl kinase, has achieved complete remission in all 31 patients treated with 300 mg/day (Schindler et al., 2000). It is noteworthy that no major toxic effects have been observed in patients that have received the drug for up to 18 months in spite of the fact that STI571 also inhibits c-Kit and the PDGFRs (Schindler et al., 2000). CT52923 displays the same specificity except it does not inhibit Abl kinase, suggesting a good safety profile is likely.

In summary, PDGFR antagonism by CT52923 has therapeutic potential for the prevention of restenosis as demonstrated in this study. This strategy can also be extended to the treatment of other diseases that are mediated by PDGFR.