Platelet-derived Growth Factor (PDGF) Receptor-α Activates c-Jun NH2-terminal Kinase-1 and Antagonizes PDGF Receptor-β-induced Phenotypic Transformation

Platelet-derived growth factor (PDGF) is a potent mitogen for mesenchymal cells. The PDGF B-chain (c-sisproto-oncogene) homodimer (PDGF BB) and v-sis, its viral counterpart, activate both α- and β-receptor subunits (α-PDGFR and β-PDGFR) and mediate anchorage-independent growth in NIH3T3 cells. In contrast, the PDGF A chain homodimer (PDGF AA) activates α-PDGFR only and fails to induce phenotypic transformation. In the present study, we investigated α- and β-PDGFR specific signaling pathways that are responsible for the differences between the transforming ability of PDGF AA and BB. To study PDGF BB activation of β-PDGFR, we established NIH3T3 clones in which α-PDGFR signaling is inhibited by a dominant-negative α-PDGFR, or an antisense construct of α-PDGFR. Here, we demonstrate that β-PDGFR activation alone is sufficient for PDGF BB-mediated anchorage-independent cell growth. More importantly, inhibition of α-PDGFR signaling enhanced PDGF BB-mediated phenotypic transformation, suggesting that α-PDGFR antagonizes β-PDGFR-induced transformation. While both α- and β-receptors effectively activate ERKs, α-PDGFR, but not β-PDGFR, activates stress-activated protein kinase-1/c-Jun NH2-terminal kinase-1 (JNK-1). Inhibition of JNK-1 activity using a dominant-negative JNK-1 mutant markedly enhanced PDGF BB-mediated anchorage-independent cell growth, demonstrating an antagonistic role for JNK-1 in PDGF-induced transformation. Consistently, overexpression of wild-type JNK-1 reduced PDGF BB-mediated transformation. Taken together, the present study showed that α- and β-PDGFRs differentially regulate Ras-mitogen-activated protein kinase pathways critical for regulation of cell transformation, and transformation suppressing activity of α-PDGFR involves JNK-1 activation.

array of cellular responses including cell proliferation, transformation, migration, and survival of mesenchymal cells (reviewed in Refs. 1 and 2). PDGF exists in the form of three dimeric polypeptides: the homodimers PDGF AA and BB and the heterodimer PDGF AB. The discovery that the oncogene product (p28 v-sis ) of the simian sarcoma virus (SSV) is 92% homologous with the nonglycosylated PDGF B chain established a causative role for growth factors in malignancy (1,(3)(4)(5). Consistently, PDGF B-chain (c-sis) gene overexpression or exogenous treatment with PDGF BB homodimer induces phenotypic transformation of fibroblasts as effectively as the v-sis gene (6). NIH 3T3 cells have been widely used as a model to study PDGF-induced phenotypic transformation, including anchorage-independent cell growth. In contrast to PDGF BB, PDGF AA fails to induce transformation of NIH 3T3 cells, although PDGF AA and BB are equally potent mitogens (7)(8)(9). Two structurally similar protein-tyrosine kinase receptor subunits (␣-PDGFR and ␤-PDGFR) have been identified. PDGF BB binds to both of these receptors, while PDGF AA effectively binds only to ␣-PDGFR (10 -12). Dimerization and autophosphorylation of PDGFR occur upon receptor-ligand interaction. Differential binding of initial signaling molecules to phosphorylated PDGFRs is thought to mediate overlapping but distinct ␣and ␤-PDGFRs-induced signaling pathways.
Mitogen-activated protein kinase (MAPK) family members are among the most critical signaling molecules for PDGF responses. PDGF activates extracellular signal-regulated kinases (ERKs), members of the MAPK family. ERKs are essential for growth factor-mediated mitogenic responses in various cell types (13)(14)(15)(16)(17). PDGF also activates other members of the MAPK family, including stress-activated protein kinase-1/c-Jun NH 2 -terminal kinase (JNK) and stress-activated protein kinase-2 (p38) (18 -20). Increasing evidence suggests that protein-tyrosine kinase receptors, including PDGFR and epidermal growth factor receptor regulate cell death as well as cell growth (21,22). Constitutive activation of these receptors has been shown to cause growth arrest and apoptosis in some cell lines (21,(23)(24)(25)(26). At present, it is unclear how tyrosine kinase receptor-mediated signaling regulates different cellular responses such as cell proliferation, transformation, growth arrest, and apoptosis.
Since PDGF BB activates both ␣and ␤-PDGFRs and PDGF AA activates only ␣-PDGFR, we have raised two important questions aimed at understanding the difference between the transforming ability of PDGF AA and BB. First, is ␤-PDGFR alone sufficient for PDGF BB-mediated anchorage-independent cell growth, or is activation of both receptors required? Second, what signaling molecules are differentially regulated by ␣and ␤-PDGFRs that cause distinct ␣and ␤-PDGFRsinduced cellular responses?
To address these questions, we have established NIH3T3 clones in which ␣-PDGFR activation is inhibited by its dominant-negative mutant, or ␣-PDGFR expression is down-regulated using an antisense construct. Using these clones, PDGF BB activation of ␤-PDGFR was investigated. PDGF AA activation of ␣-PDGFR and PDGF BB activation of both receptors was studied in control NIH 3T3 cells.
DNA Constructs and Transfection-A dominant negative mutant of ␣-PDGFR was constructed as follows. The murine ␣-PDGFR cDNA (HBXK construct provided by Dr. Eisenbach) was digested with BamHI, and the 1.9-kb fragment was ligated into the BamHI site of pcDNAI/Neo expression vector (Invitrogen, Carlsbad, CA). The orientation was determined by DNA sequencing. The construct utilized an in-frame TAG in the pcDNAI/Neo polylinker as a translation stop codon. Antisense ␣-PDGFR was constructed with the same 1.9-kb BamHI fragment inserted inversely into the BamHI site of pcDNAI/Neo. These constructs were then transfected into NIH 3T3 cells using Lipofectin (Life Technologies, Inc.). Transfectants were selected on the basis of neomycin-resistant phenotype in the presence of 400 g/ml G418 (Life Technologies, Inc.), and individual clones were isolated.
Immunoblot Analysis-Cells were lysed in SDS lysis buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS), and protein concentration was determined by BCA protein assay kit from Pierce. Equal amounts of protein in each sample were separated by SDS-PAGE and transferred to a nitrocellulose membrane. Membranes were subjected to 1 h of blocking with 5% nonfat milk in TTBS (0.02% Tween 20, 10 mM Tris-HCl, pH 7.4, 150 mM NaCl), followed by incubation with primary antibodies in TTBS for 1 h. After three washes with TTBS, the blot was incubated with the appropriate horseradish peroxidase-conjugated secondary antibody. The antigen was detected using the ECL detection system (Pierce) according to the manufacturer's instruction.
Assay for Growth in Soft Agar-Soft agar assay was performed as described previously (9) in six-well plates using a 3-ml basal layer of 0.6% agarose in Dulbecco's modified Eagle's medium/F-12 medium supplemented as described above. Five thousand cells in 0.35% agarose containing various concentration of PDGF AA or BB were plated on top of the basal agarose layer in each well. Fresh top agarose containing PDGF was overlaid every other day. After 1-3 weeks, positive colonies were photographed under 400ϫ magnification. All of the colonies were stained using Giemsa solution (Sigma). Colonies bigger than 0.2 or 0.5 mm were counted.

RESULTS
Inhibition of ␣-PDGFR Signaling in NIH3T3 Cells-In order to examine the effect of ␤-PDGFR activation alone in PDGFmediated anchorage-independent cell growth, we established NIH3T3 cell clones in which ␣-PDGFR signaling is inhibited. One approach was to prevent autoactivation of ␣-PDGFR using a dominant-negative mutant of ␣-PDGFR (DN ␣-PDGFR) that contains the extracellular and transmembrane domains, but lacks the cytoplasmic kinase domains (Fig. 1). The DN ␣-PDGFR protein was expected to dimerize with wild-type ␣-PDGFR upon PDGF AA binding, but be unable to autophosphorylate the wild-type ␣-PDGFR, and therefore prevent ␣-PDGFR-mediated signal transduction. We selected DN ␣-PDGFR-transfected NIH3T3 clones (DN clones) that express truncated ␣-PDGFR mRNA (ϳ2 kb) by Northern blot analysis (data not shown). To identify the DN clones in which ␣-PDGFR autoactivation is inhibited, the ␣-PDGFR protein was immunoprecipitated with an anti-␣-PDGFR Ab and the active form was detected by immunoblot analysis using an anti-phosphotyrosine Ab. While ␣-PDGFR was autophosphorylated in the control cells following PDGF AA treatment, the active form of ␣-PDGFR was undetectable in DN clones 9 and 16 (Fig. 2). This showed that DN ␣-PDGFR successfully prevented PDGF AAmediated dimerization and activation of wild-type ␣-PDGFR in DN9 and DN16. Of note, the truncated DN ␣-PDGFR protein was not detected by immunoblot analysis, since anti-␣-PDGFR Ab recognized the COOH terminus of ␣-PDGFR. The second approach to inhibit the ␣-PDGFR signaling was to down-regu- late ␣-PDGFR expression using an antisense construct of ␣-PDGFR (AS ␣-PDGFR) (Fig. 1). The level of ␣-PDGFR expression was significantly down-regulated in AS clones 4 and 6, as determined by immunoblot analysis (Fig. 2). Hereafter, we present data obtained mostly using DN16 and AS6 to avoid redundancy.
To confirm that ␣-PDGFR signaling is inhibited in DN and AS clones, PDGF AA-activation of ERK was examined. As shown in Fig. 3, active ERK-2 was readily detected in the control NIH3T3 cells 7 min after exposure to 5 ng/ml PDGF AA. In contrast, PDGF AA-induced ERK-2 activation was significantly inhibited in DN and AS clones, showing that ␣-PDGFR signaling is down-regulated in these cells (Fig. 3).
To ensure that ␤-PDGFR signaling is not significantly altered in DN and AS clones, the expression level and activation of ␤-PDGFR was examined. While PDGF AA induces ␣-PDGFR homodimerization only, PDGF BB induces homodimerization of ␣␣and ␤␤-PDGFRs and heterodimerization of ␣␤-PDGFR. Thus, if DN ␣-PDGFR levels are too high, ␤-PDGFR activation can also be disturbed by DN ␣-PDGFR. The efficiency of PDGF BB-induced tyrosine phosphorylation of ␤-PDGFR in DN and AS clones was similar to that in control NIH3T3 cells (Fig. 4A). In contrast, PDGF BB-induced ␣-PDGFR phosphorylation occurred only in the control NIH3T3 cells, but not in DN or AS cells (Fig. 4B). This showed that DN ␣-PDGFR inhibited dimerization and activation of ␣-PDGFR without significant alteration of ␤-PDGFR activation.
An antisense ␣-PDGFR construct contained cDNA encoding extracellular, transmembrane, and juxtamembrane domains of ␣-PDGFR in reverse orientation. Although ␣and ␤-PDGFRs are closely related molecules, the antisense transcript of ␣-PDGFR should not interfere with ␤-PDGFR expression, since the nucleotide sequence homology is relatively low, especially in the extracellular domain. Indeed, Fig. 4A showed that ␤-PDGFR expression was not altered by AS ␣-PDGFR. The ␤-PDGFR protein and activation levels were comparable in the control NIH3T3, DN, and AS clones as determined by immunoblot analysis (Fig. 4A). To further ensure that ␤-PDGFR signaling is intact in DN and AS clones, PDGF BB activation of ERK-2 was examined. As shown in Fig. 4C, 5 ng/ml PDGF BB activated ERK-2 in DN and AS clones as efficiently as in the control NIH3T3 cells.
Inhibition of ␣-PDGFR Signaling Enhances PDGF BB-mediated Phenotypic Transformation of NIH3T3 Cells-We examined whether PDGF BB-mediated anchorage-independent cell growth requires activation of both ␣and ␤-PDGFR, or if ␤-PDGFR alone is sufficient. Soft agar assay was performed to compare the efficiencies of PDGF BB-induced colony formation among the control NIH 3T3, DN, and AS cells. PDGF BB activation of ␤-PDGFR alone in DN and AS cells was sufficient to induce anchorage-independent cell growth (Fig. 5A). Surprisingly, the efficiency of PDGF BB-induced colony formation was significantly higher in DN and AS cells than in the control NIH3T3 cells (Fig. 5B), suggesting that inhibition of ␣-PDGFR signaling further enhanced PDGF BB-induced phenotypic transformation. This suggests that ␣-PDGFR may antagonize PDGF BB-induced transforming activity.
␣-PDGFR Is Critical for PDGF Activation of JNK-1-Accumulating evidence implies that the balance between mitogenic (such as ERKs) and stress-induced (such as JNKs) signaling molecules downstream of Ras are critical for growth factorinduced cellular responses (28 -32). ERK, a MAPK family member known to be critical for cell proliferation, was activated either by ␣or ␤-PDGFR (Figs. 3 and 4C), in agreement with the previous observation that PDGF AA and BB are equally potent mitogens (7)(8)(9). In contrast to ERK, JNK activity is associated with cell cycle arrest or cell death following the loss of cell anchorage (33)(34)(35). Recent studies showed that JNK activates caspases (cysteine proteases that initiate apoptotic cell death) and caspases further activate JNK, suggesting a positive feedback loop between JNK and caspases leading to cell death (33,34). We next asked if the ability of ␣-PDGFR to antagonize anchorage-independent cell growth is associated with its ability to activate the JNK pathway. To this end, we examined ␣and ␤-PDGFR-induced activation of JNKs. Both PDGF AA and BB effectively activated JNK-1 in the control NIH3T3 cells, while they had little effect on JNK-2 (data not shown). The kinetics and dose responses of PDGF AA-or BBinduced JNK-1 activation were similar. Maximal JNK-1 induction occurred within 30 min with 50 ng/ml PDGF AA or BB (data not shown). PDGF AA-mediated JNK-1 activation was significantly inhibited in DN and AS clones as expected (Fig.  6A). Importantly, PDGF BB-induced JNK-1 activation was also drastically inhibited in these clones. This showed that ␤-PDGFR alone is not sufficient, and ␣-PDGFR signaling is critical for maximum induction of JNK-1 activity by PDGF. To ensure that lack of PDGF-induced JNK-1 activity was not due to intrinsic incapability to activate JNK-1 in these DN and AS clones, bFGF-activated JNK-1 was examined. As shown in Fig.  6B, active JNK-1 was readily detectable in DN and AS clones following bFGF treatment, as in the control cells. significantly higher in JNK-APF-transfected NIH 3T3 cells than in the control vector-transfected cells (Fig. 7A). When PDGF activation of JNK-1 was examined, both PDGF AA and BB failed to activate JNK-1 in JNK1-APF-transfected NIH3T3 cells (Fig. 7B), demonstrating a dominant negative activity of mutant JNK as reported previously (36). PDGF BB-induced phenotypic transformation was markedly enhanced in JNK1-APF-transfected cells compared with the control cells (Fig. 8), indicating that JNK-1 negatively regulates PDGF BB-induced transformation. To further confirm this, we next examined the effect of enhanced JNK-1 activity on PDGF BB-induced phenotypic transformation. To this end, we introduced wild-type JNK-1 (provided by Dr. R. Davis) into NIH 3T3 cells, and JNK-1 overexpression in these cells was confirmed by immunoblot analysis (Fig. 9A). PDGF BB activation of JNK-1 was significantly enhanced in JNK-1-transfected cells (Fig. 9B), and the efficiency for PDGF BB-induced anchorage-independent cell growth was significantly reduced when JNK-1 activity was enhanced (Fig. 9C). Taken together, the present study demonstrated that JNK-1 plays a critical role for PDGF regulation of cell transformation, and lack of JNK-1 activation in the absence of ␣-PDGFR enhances PDGF BB-induced phenotypic transformation in NIH3T3 cells. DISCUSSION Critical functions of PDGF isoforms and their receptor subunits during embryogenesis have been well studied using knock-out mice deficient in PDGF A, PDGF B, ␣-receptor, or ␤-receptor gene (37)(38)(39)(40). However, embryonic mortality of these knock-out mice does not allow studies of PDGF isoforms and their receptors in physiological and pathological processes in adults. Such processes include wound healing, inflammation, and proliferative diseases such as atherosclerosis, fibrosis, and tumorigenesis (reviewed in Refs. 1 and 2). In vitro, primarily PDGF BB has been used to study PDGF-induced signaling pathways, which binds both ␣and ␤-receptors (reviewed in Ref.

JNK-1 Regulates PDGF BB-mediated Phenotypic Transformation of NIH3T3 Cells-To
2). The ␣or ␤-PDGFR specific signaling pathway was investigated by introducing each receptor subunit into cells lacking endogenous PDGFRs (10,41,42). These studies helped identify signaling molecules that bind to each PDGFR subunit and reveal their roles in some PDGFRs-mediated cellular processes (43)(44)(45)(46)(47)(48)(49). However, it is now well recognized that PDGFmediated cellular responses vary among cell types. These variations are most likely due to innate differences in available signaling molecules, making it critical to investigate PDGFRmediated pathways in cell types that express PDGFRs and contain the requisite signaling molecules for diverse PDGFinduced cellular responses.
Using NIH3T3 cells that are highly responsive to PDGF, we studied the differential roles of ␣and ␤-PDGFRs in PDGF BB-induced anchorage-independent cell growth and activation of signaling molecules. Here, we report a transformation-suppressing activity of ␣-PDGFR in PDGF BB-induced transfor-mation through JNK-1 induction. Both PDGF AA and BB activated JNK-1 in NIH 3T3 cells without noticeable JNK-2 activation. JNKs are often constitutively activated in apoptotic cells (34) and also during transformation processes induced by growth factors, virus, or oncogene products (50 -54). JNK isoforms appear to mediate different cellular responses. The JNK-2 isoform mediates EGF-induced transformation of human A549 lung carcinoma cells (53). In contrast to the transforming activity of JNK-2, suppression of JNK-1 activity by dominant negative JNK-1 (JNK1-APF) enhanced arsenite-induced cell transformation of mouse epithelium (54), suggesting that JNK-1 transduces transformation-suppressing activity. JNK-1-mediated negative regulation of cell growth/survival was also suggested in another study (34). Following loss of cell-substrata interactions, JNK activity increases, followed by cell cycle arrest or apoptosis (34). Consistently, we found that JNK-1 down-regulation by either inhibition of ␣-PDGFR signaling or using a dominant negative JNK-1 mutant drastically increased anchorage-independent growth efficiency in NIH 3T3 cells in response to PDGF BB.
PDGF activation of ␣-PDGFR does not induce phenotypic transformation in murine fibroblasts (9). Interestingly, however, we previously showed that PDGF AA induces anchorageindependent cell growth of normal rat kidney (NRK) fibroblast cells that overexpress Bcl-2, an anti-apoptotic gene product (9). Bcl-2 was shown to inhibit JNK-1 activation and to prevent cell death following loss of cell anchorage (34). These studies (9,34), together with our present results, provided the basis for our working model for PDGF regulation of transformation pathway as diagrammed in Fig. 10. Activation of ␣-PDGFR may transduce both positive and negative signaling for cell transformation, while ␤-PDGFR mainly induces positive signaling for cell transformation. PDGF BB activation of both receptors shifts the balance of signaling to favor the transformation pathway, while PDGF AA activation of ␣-PDGFR alone does not. When ␣-PDGFR-mediated negative signaling is inhibited (e.g. by Bcl-2), ␣-PDGFR activation can result in phenotypic transformation. The present study clearly demonstrated that ␤-PDGFR activation alone is sufficient to induce phenotypic transformation of murine fibroblasts, and that ␣-PDGFR signaling down-regulates ␤-PDGFR-induced transformation through JNK-1 activation. However, it should be noted that PDGF BB activates JNK-1 as effectively as PDGF AA, and that PDGF BB induces phenotypic transformation in the presence of active JNK-1. In our working model (Fig. 10), we hypothesize that ␣-PDGFR, through activation of JNK-1, serves as a negative regulator for PDGF BB-induced phenotypic transformation.
At present, it is unclear how ␣and ␤-PDGFRs differentially activate JNK-1, a member of MAPK family. Tyrosine kinase growth factor receptors activate a protein kinase cascade that leads to MAPK activation by a complex mechanism involving the SH2/3 proteins, Grb2, Sos, and Ras (55). Several MAPK kinase kinases have been identified including c-Raf, c-Mos, and MEKK (56). Among them, MEKK-1 was shown to activate the JNK pathway (33,57). While active Ras is sufficient for ERKs activation, phosphatidylinositol 3-kinase and Rac1 are apparently required for maximum induction of JNK activity (58 -60). Both ␣and ␤-PDGFRs can activate Ras and phosphatidylinositol 3-kinase pathways, yet the level and duration of the activation may differ between ␣and ␤-PDGFR. This is suggested by the observation that the GTPase-activating protein of Ras, a negative regulator of Ras, preferentially binds to the ␤-PDGFR (61,62). The subtle differences in the activities of the initial signaling molecules are likely to trigger different biochemical cascades leading to different cellular responses. Consistently, it was shown that prolonged activation of ERK induces PC12 cell differentiation (63) and constitutive activation of Raf-1 at the cytoplasmic membrane induces apoptotic cell death (64), whereas transient activation of these signaling molecules results in cell growth (63). The cell lines that we have generated should provide powerful tools to investigate the Ras-MAPK pathways differentially regulated by ␣and ␤-PDGFRs.