Activation of the Epidermal Growth Factor Receptor Signal Transduction Pathway Stimulates Tyrosine Phosphorylation of Protein Kinase C

Abstract The expression of an oncogenic ras gene in epidermal keratinocytes stimulates the tyrosine phosphorylation of protein kinase C and inhibits its enzymatic activity (Denning, M. F., Dlugosz, A. A., Howett, M. K., and Yuspa, S. H.(1993) J. Biol. Chem. 268, 26079-26081). Keratinocytes expressing an activated ras gene secrete transforming growth factor α (TGFα) and have an altered response to differentiation signals involving protein kinase C (PKC). Because the neoplastic phenotype of v-ras expressing keratinocytes can be partially mimicked in vitro by chronic treatment with TGFα and the G protein activator aluminum fluoride (AlF), we determined if TGFα or AlF could induce tyrosine phosphorylation of PKC. Treatment of primary keratinocyte cultures for 4 days with TGFα induced tyrosine phosphorylation of PKC, whereas AlF only slightly stimulated PKC tyrosine phosphorylation. The PKC that was tyrosine-phosphorylated in response to TGFα had reduced activity compared with the nontyrosine-phosphorylated PKC. Treatment of keratinocytes expressing a normal epidermal growth factor receptor (EGFR) with TGFα or epidermal growth factor for 5 min induced PKC tyrosine phosphorylation. This acute epidermal growth factor treatment did not induce tyrosine phosphorylation of PKC in keratinocytes isolated from waved-2 mice that have a defective epidermal growth factor receptor. In addition, the level of PKC tyrosine phosphorylation in v-ras-transduced keratinocytes from EGFR null mice was substantially lower than in v-ras transduced wild type cells, suggesting that activation of the EGFR is important for PKC tyrosine phosphorylation in ras transformation. However, purified EGFR did not phosphorylate recombinant PKC in vitro, whereas members of the Src family (c-Src, c-Fyn) and membrane preparations from keratinocytes did. Furthermore, clearing c-Src or c-Fyn from keratinocyte membrane lysates decreased PKC tyrosine phosphorylation, and c-Src and c-Fyn isolated from keratinocytes treated with TGFα had increased kinase activity. Acute or chronic treatment with TGFα did not induce significant PKC translocation in contrast to the phorbol ester 12-O-tetradecanoylphorbol-13-acetate, which induced both translocation and tyrosine phosphorylation of PKC. This suggests that TGFα-induced tyrosine phosphorylation of PKC results from the activation of a tyrosine kinase rather than physical association of PKC with a membrane-anchored tyrosine kinase. Taken together, these results indicate that PKC activity is inhibited by tyrosine phosphorylation in response to EGFR-mediated signaling and activation of a member of the Src kinase family may be the proximal tyrosine kinase acting on PKC in keratinocytes.

The expression of an oncogenic ras Ha gene in epidermal keratinocytes stimulates the tyrosine phosphorylation of protein kinase C ␦ and inhibits its enzymatic activity ( Keratinocytes expressing an activated ras Ha gene secrete transforming growth factor ␣ (TGF␣) and have an altered response to differentiation signals involving protein kinase C (PKC). Because the neoplastic phenotype of v-ras Ha expressing keratinocytes can be partially mimicked in vitro by chronic treatment with TGF␣ and the G protein activator aluminum fluoride (AlF 4 Ϫ ), we determined if TGF␣ or AlF 4 Ϫ could induce tyrosine phosphorylation of PKC␦. Treatment of primary keratinocyte cultures for 4 days with TGF␣ induced tyrosine phosphorylation of PKC␦, whereas AlF 4 Ϫ only slightly stimulated PKC␦ tyrosine phosphorylation. The PKC␦ that was tyrosine-phosphorylated in response to TGF␣ had reduced activity compared with the nontyrosinephosphorylated PKC␦. Treatment of keratinocytes expressing a normal epidermal growth factor receptor (EGFR) with TGF␣ or epidermal growth factor for 5 min induced PKC␦ tyrosine phosphorylation. This acute epidermal growth factor treatment did not induce tyrosine phosphorylation of PKC␦ in keratinocytes isolated from waved-2 mice that have a defective epidermal growth factor receptor. In addition, the level of PKC␦ tyrosine phosphorylation in v-ras Ha -transduced keratinocytes from EGFR null mice was substantially lower than in v-ras Ha transduced wild type cells, suggesting that activation of the EGFR is important for PKC␦ tyrosine phosphorylation in ras transformation. However, purified EGFR did not phosphorylate recombinant PKC␦ in vitro, whereas members of the Src family (c-Src, c-Fyn) and membrane preparations from keratinocytes did. Furthermore, clearing c-Src or c-Fyn from keratinocyte membrane lysates decreased PKC␦ tyrosine phosphorylation, and c-Src and c-Fyn isolated from keratinocytes treated with TGF␣ had increased kinase activity. Acute or chronic treatment with TGF␣ did not induce significant PKC␦ translocation in contrast to the phorbol ester 12-O-tetradecanoylphorbol-13-acetate, which induced both translocation and tyrosine phosphorylation of PKC␦. This suggests that TGF␣-induced tyrosine phosphorylation of PKC␦ results from the activation of a tyrosine kinase rather than physical association of PKC␦ with a membrane-anchored tyrosine kinase. Taken together, these results indicate that PKC␦ activity is inhibited by tyrosine phosphorylation in response to EGFR-mediated signaling and activation of a member of the Src kinase family may be the proximal tyrosine kinase acting on PKC␦ in keratinocytes.
The protein kinase C (PKC) 1 family of serine/threonine protein kinases is a central component of phospholipase-coupled growth factor receptor signaling pathways (1). The enzymatic activities of several PKC isoforms are regulated by the allosteric activators diacylglycerol and Ca 2ϩ , which are elevated in response to growth factor receptor activation (2). The G protein Ras is also a component of the mitogen-activated protein kinase signaling pathways for growth factors such as epidermal growth factor (EGF) and platelet derived growth factor (PDGF). In addition, PKC␣ can directly affect the mitogenactivated protein kinase pathway by phosphorylating and stimulating the autokinase activity of Raf-1 (3). Ras can act either as a regulator or effector of PKC function, but the molecular mechanisms involved are unclear (4 -8).
In epidermal keratinocytes, Ras can have both positive and negative effects on PKC signaling. Neoplastic mouse keratinocytes expressing an activated ras Ha allele have increases in phosphatidylinositol turnover, diacylglycerol levels, and calcium-dependent PKC activity (9,10). Activation of PKC␣, the only calcium-dependent PKC isoform expressed in keratinocytes, is correlated with the expression of granular layer differentiation markers in normal keratinocytes and in v-ras Ha expressing keratinocytes, where granular layer differentiation markers are up-regulated (10,11). Keratinocytes expressing a v-ras Ha oncogene also have decreased Ca 2ϩ -independent PKC activity and have a block in their ability to commit to Ca 2ϩ and TPA-induced terminal differentiation (10,(12)(13)(14). The block in differentiation response may be due to the tyrosine phosphorylation and inactivation of PKC␦ because the kinase inhibitor staurosporine blocks PKC␦ tyrosine phosphorylation and induces terminal differentiation of neoplastic keratinocytes (5).
In addition, v-ras Ha transformation of keratinocytes up-regulates epidermal growth factor receptor (EGFR) ligand expression and induces the secretion of transforming growth factor ␣ (TGF␣) (15)(16)(17). Several studies have indicated that increased secretion of TGF␣ can substitute for ras Ha mutations in the altered response to differentiation signals (17) and can initiate skin carcinogenesis (18,19).
In this study, we further characterized the signal transduction pathways regulating PKC␦ tyrosine phosphorylation in murine keratinocytes. We demonstrate that TGF␣ is able to induce tyrosine phosphorylation of PKC␦ resulting in inhibition of its activity. Although tyrosine phosphorylation of PKC␦ in intact cells requires a functional EGFR, it is not directly mediated by EGFR and may involve a member of the Src kinase family. Tyrosine phosphorylation of PKC␦ by the EGFR pathway demonstrates a perturbation of PKC signaling by TGF␣ that may contribute to the process of neoplastic transformation in epithelial cells.

EXPERIMENTAL PROCEDURES
Materials-Purified EGFR was from Promega, and Src and Fyn kinases were purchased from Upstate Biotechnology Inc. PKC isozymes were produced in SF9 insect cells using a baculovirus expression system as described previously (20) and were kindly provided by Drs. Marcelo G. Kazanietz and Peter M. Blumberg of the Laboratory of Cellular Carcinogenesis and Tumor Promotion at the National Cancer Institute.
Cell Culture-Primary keratinocytes were isolated from newborn BALB/c, waved-2 mice, and EGFR ϩ/ϩ or Ϫ/Ϫ newborn mice (21). The waved-2 mice were obtained from Jackson Laboratories. The phenotype of the wa-2/wa-2 mice were distinguished from the ϩ/ϩ and ϩ/wa-2 mice by the curly whiskers observed in the wa-2/wa-2 newborn mice (22). The keratinocytes were cultured in Eagle's minimal essential medium containing 8% Chelex-treated (Bio-Rad) fetal bovine serum with the final Ca 2ϩ concentration adjusted to 0.05 mM as described (23). The EGFR genotype of newborn mice was determined by polymerase chain reaction amplification of tail DNA as described previously (21). Primary keratinocytes from the EGFR ϩ/ϩ and Ϫ/Ϫ mice (21) were cultured in 0.05 mM Ca 2ϩ -containing medium with 1 ng/ml keratinocyte growth factor for 2-3 days to stimulate proliferation of the EGFR Ϫ/Ϫ keratinocytes. After reaching approximately 80% confluence, the cells were cultured without keratinocyte growth factor for 3-4 days before harvesting. For the introduction of a v-ras Ha gene into cells, keratinocytes were exposed to 4 g/ml polybrene in the absence or the presence of a replication-defective retrovirus containing the v-ras Ha gene and cultured 4 -6 days before use (24).
Immunoprecipitations were performed as described previously (5). Briefly, the cells were washed with phosphate-buffered saline and scraped into immunoprecipitation lysis buffer, and equal amounts of protein were immunoprecipitated with 20 l of protein A/G PLUS-Agarose (Santa Cruz Biotechnology) and either 2.5 g anti-phosphotyrosine antibody or 0.5-1 l of anti-PKC␦ antibody. The c-Src and c-Fyn immunoprecipitations were performed with 2 g of antibody from Santa Cruz Biotechnology, Inc. (sc-19 and sc-16 respectively) and 50 l protein A-Sepharose from Sigma. For some phosphotyrosine immunoprecipitations, an agarose-conjugated anti-phosphotyrosine monoclonal antibody from Upstate Biotechnology Inc. was used. For immunoblotting, the immunoprecipitates were washed three times with RIPA buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, and 1% sodium deoxycholate) and boiled in 20 l of SDS sample buffer (25).
In Vitro Kinase Assays-Tyrosine kinase reactions were performed in a volume of 20 -30 l containing 50 mM Tris, pH 7.5, 10 mM MgCl 2 , 2 mM MnCl 2 , 100 M phenylmethylsulfonyl fluoride, 100 M NaVO 3 , and 1 mM NaF. PKC isozymes from baculovirus-infected Sf-9 cells were included as substrates for either Src or Fyn kinase (10 units) or keratinocyte membranes (25 g of protein). The keratinocyte membranes were prepared from normal primary keratinocytes by scraping the cells into hypotonic lysis buffer (1 mM HEPES, pH 7.5, 5 mM MgCl 2 , and 100 M phenylmethylsulfonyl fluoride) and spinning at approximately 50 ϫ g for 5 min at 4°C. The supernatant was then spun in a microcentrifuge for 20 min, and the membrane pellet resuspended in assay buffer and used as a source of tyrosine kinase. The reactions were started by adding 25 M ATP and incubated for 10 min at 30°C. Tyrosine phosphorylation was assessed by immunoblotting with an anti-phosphotyrosine antibody. For the Src and Fyn immunoprecipitation kinase assays, the immunoprecipitates were washed twice with immunoprecipitation lysis buffer, twice with assay buffer (20 mM HEPES, pH 7.4, 3 mM MnCl 2 , and 5 mM MgCl 2 ), and incubated in 30 l of assay buffer containing 10 M ATP for 10 min at 30°C. The Sepharose beads were washed once in immunoprecipitation lysis buffer and analyzed by immunoblotting with an anti-phosphotyrosine antibody.
PKC assays of tyrosine-phosphorylated and nontyrosine-phosphorylated PKC␦ were performed as described previously with minor modifications (5). Primary keratinocytes were cultured in 10 ng/ml TGF␣ for 4 days to induce the tyrosine phosphorylation of a fraction (ϳ50%) of PKC␦. Lysates were prepared (n ϭ 3) and immunoprecipitated with an anti-phosphotyrosine antibody for 2.5 h. The phosphotyrosine-cleared lysates were taken for immunoprecipitation of nontyrosine-phosphorylated PKC␦. Both phosphorylated and nonphosphorylated samples were immunoprecipitated for PKC␦ overnight in the presence of 30 mM p-nitrophenylphosphate to release phosphotyrosine containing proteins from the anti-phosphotyrosine antibody. The immunoprecipitates were washed once with immunoprecipitation lysis buffer and twice in 50 mM Tris-HCl, pH 7.4, and resuspended in 25 l of Tris-HCl and 5 mM 2-mercaptoethanol. PKC assays were performed on the immunoprecipitated PKC␦ in the presence or the absence of 1 M TPA as described by Nakadate et al. (26) except that the PKC substrate peptide [ser 25 ] (Life Technologies, Inc.) was the phosphate acceptor, and 20%/80% phosphatidylserine/phosphatidylcholine vesicles were the phospholipid. After the enzyme reaction, the agarose beads were spun, and 30 l of supernatant was spotted for scintillation counting. The PKC␦ remaining on the beads was boiled in SDS sample buffer and analyzed by anti-PKC␦ immunoblot to correct for differences in the amount of PKC␦ in the assay.

TGF␣ Induces Tyrosine Phosphorylation of PKC␦ in Normal
Keratinocytes-Because treatment of primary keratinocyte cultures with TGF␣ and the nonspecific G protein activator AlF 4 Ϫ elicits many phenotypic changes characteristic of v-ras Ha transformation (9), we examined whether the tyrosine phosphorylation of PKC␦ was also induced by these pharmacological agents. Culturing primary keratinocytes in the presence of 10 ng/ml TGF␣ for 4 days induced tyrosine phosphorylation of PKC␦ (27-fold), whereas a 4-day AlF 4 Ϫ (1 M AlCl 3 and 1 mM NaF) treatment induced tyrosine phosphorylation only slightly (Fig. 1A). The combined treatment of TGF␣ and AlF 4 Ϫ resulted in a more than additive increase in PKC␦ tyrosine phosphorylation (48-fold).
Because TGF␣ was a more effective inducer of PKC␦ tyrosine phosphorylation than AlF 4 Ϫ , we further analyzed TGF␣-induced PKC␦ tyrosine phosphorylation. The kinetics of TGF␣induced PKC␦ tyrosine phosphorylation were biphasic (Fig.  1B). Tyrosine phosphorylation of PKC␦ was increased approximately 8-fold after a 5-min TGF␣ treatment (10 ng/ml), returned toward basal levels by 1-2 h, and increased again from 4 and 6 h (9-11-fold), remaining elevated (4-fold) even after 24 h of TGF␣ treatment. The total levels of PKC␦ did not change appreciably during this time. Acute and chronic treatment with EGF also induced PKC␦ tyrosine phosphorylation (see Fig. 3A).
TGF␣-induced Tyrosine Phosphorylation of PKC␦ Inhibits Its Activity-The tyrosine-phosphorylated form of PKC␦ isolated from ras Ha expressing keratinocytes is not activated by the phorbol ester TPA (5). To determine the effect of TGF␣-induced tyrosine phosphorylation on PKC␦ enzymatic activity, we performed assays on tyrosine-phosphorylated and nontyrosinephosphorylated PKC␦ isolated from keratinocytes treated for 4 days with 10 ng/ml TGF␣ (Fig. 2). Similar to the results for v-ras Ha , the tyrosine-phosphorylated PKC␦ from TGF␣-treated cells had much lower constitutive activity, and TPA caused only a modest increase in activity. In fact, the specific activity of the tyrosine-phosphorylated PKC␦ assayed in the presence of 1 M TPA was substantially lower than the basal specific activity of the nontyrosine-phosphorylated PKC␦.
Tyrosine Phosphorylation of PKC␦ Requires a Functional EGF Receptor-To determine the contribution of the EGFR to PKC␦ modification, keratinocytes were isolated from waved-2 mice that have a mutated EGFR with decreased tyrosine kinase activity (22). Fig. 3A shows that tyrosine phosphorylation of PKC␦ was induced by v-ras Ha transduction and acute (5 min) or chronic (4 day) treatment with EGF in keratinocytes from BALB/c and phenotypically normal waved-2 mice (ϩ/ϩ and ϩ/wa-2). However, acute EGF treatment (5 min) did not induce tyrosine phosphorylation of PKC␦ in the waved-2 keratinocytes (wa-2/wa-2), whereas chronic treatment or v-ras Ha did. This is consistent with the defective but not inactive EGFR previously documented in the waved-2 strain (22).
Tyrosine phosphorylation of PKC␦ was also assessed in keratinocytes isolated from mice harboring a disrupted EGFR gene (21). The immunoblots in Fig. 3B show that PKC␦ was tyrosine-phosphorylated to a similar extent in both EGFR ϩ/ϩ and Ϫ/Ϫ keratinocytes (1200 and 1700% increase, respectively) in response to acute TPA treatment. In contrast, v-ras Ha transduction was less effective at inducing PKC␦ tyrosine phosphorylation in the EGFR Ϫ/Ϫ keratinocytes than in the ϩ/ϩ cells of the same genetic background (60 and 250% increase, respectively). There was no difference in the amount of total PKC␦ between the EGFR ϩ/ϩ and Ϫ/Ϫ keratinocytes. These genetic data strongly support the involvement of the EGFR in the v-ras Ha -induced PKC␦ tyrosine phosphorylation.
In Vitro Tyrosine Phosphorylation of PKC␦ by Src Family Kinases but Not EGFR-The ability of EGF or TGF␣ to rapidly induce the tyrosine phosphorylation of PKC␦ suggests that the EGFR directly phosphorylates PKC␦. To test this hypothesis, we performed in vitro phosphorylation reactions using EGFR, c-Src, c-Fyn, or a membrane fraction from keratinocytes as the tyrosine kinase and recombinant PKC␦ as the phosphate acceptor (Fig. 4A). As detected by phosphotyrosine immunoblotting, purified EGFR did not tyrosine phosphorylate PKC␦ either in the total reaction mixture or in the PKC␦ immunoprecipitate. The EGFR did autophosphorylate, indicating that it was enzymatically active. The tyrosine-phosphorylated band at the M r of the EGFR in the PKC␦ immunoprecipitates has never been detected in PKC␦ immunoprecipitates from cell lysates and may associate with PKC␦ only under these in vitro conditions. As shown in the PKC␦ immunoprecipitate for c-Src and in the total blot for c-Fyn, both kinases were able to tyrosine phosphorylate PKC␦, consistent with previous work demonstrating the phosphorylation of PKC␦ by c-Src and c-Fyn in vitro (27,28). A crude membrane fraction from normal keratinocytes also mediated tyrosine phosphorylation of PKC␦ (see PKC␦ IP in Fig. 4A). Although the tyrosine-phosphorylated form of PKC␦ has been previously localized to the membrane fraction (5,29), the tyrosine-phosphorylated PKC␦ in the membrane reaction mixture is not endogenous PKC␦ formed in vivo because the amount of tyrosine-phosphorylated PKC␦ is very low in normal keratinocytes (see Figs. 1, 3, and 5).
To determine if either c-Src or c-Fyn was present in the keratinocyte membranes and was phosphorylating PKC␦, we immunoprecipitated c-Src or c-Fyn from a solubilized membrane fraction and tested if this cleared lysate could phosphorylate PKC␦ on tyrosine. Fig. 4B shows that removal of c-Src or c-Fyn from the membrane lysate significantly decreased the PKC␦ tyrosine kinase activity compared with the nonspecific rabbit IgG cleared lysate. The tyrosine kinase activity of c-Src and c-Fyn immunoprecipitated from keratinocytes treated with TGF␣ for 5 min or 5 days is shown in Fig. 4C. The autophosphorylation and heterophosphorylation of multiple proteins by both c-Src and c-Fyn was increased by acute and chronic TGF␣ treatment. Kinase activity was determined by densitometry of the phosphotyrosine immunoblots. C-Src kinase activity increased 1.42 Ϯ 0.35-fold and 1.66 Ϯ 0.34-fold (mean Ϯ standard Ϫ ) for 4 days, and lysates (700 g protein) were immunoprecipitated with an anti-phosphotyrosine antibody. The immunoprecipitates were immunoblotted with an anti-PKC␦ antibody. In B, 10 ng/ml of TGF␣ was added to cultures of primary keratinocytes for the indicated times, and lysates (550 g of protein) were prepared for PKC␦ immunoprecipitation. The immunoprecipitates were immunoblotted against an anti-phosphotyrosine antibody and restained with an anti-PKC␦ antibody. Similar results were obtained in one additional experiment.

FIG. 2. Enzymatic activity of tyrosine-phosphorylated PKC␦.
Primary keratinocytes were cultured for 4 days in medium containing 10 ng/ml TGF␣, and the lysates (n ϭ 3) were immunoprecipitated sequentially with an anti-phosphotyrosine antibody followed by an anti-PKC␦ antibody. PKC activity was assayed in the tyrosine-phosphorylated and nonphosphorylated fractions in the presence or the absence of 1 M TPA. For calculating PKC␦-specific activities, the amount of PKC␦ in each assay tube was normalized by immunoblotting with an anti-PKC␦ antibody. Similar results were obtained in two additional experiments. deviation, n ϭ 2) after acute and chronic TGF␣ treatment. The effect of TGF␣ on c-Fyn kinase activity was greater, increasing 2.05 Ϯ 0.35-fold and 3.26 Ϯ 0.50-fold after acute and chronic TGF␣ (n ϭ 3). Immunoprecipitation with nonspecific IgG recovered no detectable tyrosine kinase activity (Fig. 4C). In addition, the recovery of tyrosine kinase activity in c-Src and c-Fyn immunoprecipitates was completely blocked by adding an excess of the peptide that the antibodies were raised against (data not shown). PKC␦ was not present in c-Src and c-Fyn immunoprecipitates when assayed by immunoblot (data not shown). These results indicate that keratinocyte c-Src and c-Fyn can phosphorylate PKC␦ and that both of these tyrosine kinases are activated in response to TGF␣ treatment of keratinocytes.
Translocation Is Not Required for Tyrosine Phosphorylation of PKC␦-Treatment of cells with certain growth factors is able to cause translocation of PKC isozymes, including PKC␦ (29 -31). Because translocation is associated with PKC␦ tyrosine phosphorylation (29,31,32), we measured the subcellular distribution of PKC␦ following TGF␣ treatment. Fig. 5A shows that there was no significant increase in the proportion of particulate PKC␦ following a 2-min or 4-day exposure to 10 ng/ml TGF␣. Both of these treatments induced tyrosine phosphorylation of PKC␦ as shown by the phosphotyrosine immunoprecipitation and PKC␦ immunoblot in Fig. 5A. As a positive control for translocation, the phorbol ester TPA was used to translocate PKC␦ (Fig. 5B). TPA induced both PKC␦ translo-cation and tyrosine phosphorylation as described previously (27,29,33). These results indicate that translocation of PKC␦ is not a prerequisite for TGF␣-induced PKC␦ tyrosine phosphorylation.
In Vitro Tyrosine Phosphorylation of Other PKC Isoforms-We examined the ability of c-Src and c-Fyn to tyrosine phosphorylate recombinant PKCs ␣, ␤1, ␥, ␦, ⑀, , and in vitro. Analysis of the total kinase reaction mixtures in Fig. 6A indicated that PKC␣, ␤1, and ␥ were not appreciably phosphorylated by either c-Src or c-Fyn but tyrosine-phosphorylated proteins corresponding to the M r of PKC␦, ⑀, , and were detected. To verify tyrosine phosphorylation of PKC␦, ⑀, , and , phosphotyrosine-containing proteins were immunoprecipi-

FIG. 3. Mutant alleles of the EGFR inhibit PKC␦ tyrosine phosphorylation.
A, primary keratinocytes were isolated from mice having a wild type EGFR (BALB/c, ϩ/ϩ and ϩ/wa-2) and from waved-2 mice homozygous for the mutant EGFR (wa-2/wa-2). The keratinocytes were cultured and infected with v-ras Ha for 5 days (v-ras), treated with 10 ng/ml EGF for 4 days (EGF 4D), or 5 min (EGF 5 M), and PKC␦ was immunoprecipitated from cellular lysates (750 g of protein). The immunoprecipitates were immunoblotted with an anti-phosphotyrosine antibody (Anti p-Tyr) and anti-PKC␦ antibody (Anti PKC␦). Similar results were obtained in two additional experiments. B, primary keratinocytes were isolated from mice having a wild type EGFR (ϩ/ϩ) and from EGFR null mice (Ϫ/Ϫ). The keratinocytes were initially cultured in 1 ng/ml keratinocyte growth factor for 3 days, switched to unsupplemented medium, and either treated with 100 nM TPA for 10 min (TPA) or infected with v-ras Ha for 4 days (ras). Cellular lysates were immunoprecipitated with anti-phosphotyrosine antibody (p-Tyr IP), and gel separated total lysates (Total) were immunoblotted for the detection of total PKC␦ levels with the anti-PKC␦ antibody. A similar result was observed in one additional experiment.

FIG. 4. Tyrosine phosphorylation of PKC␦ by Src family kinases.
In A, recombinant PKC␦ from baculovirus-infected Sf-9 insect cells was incubated alone or with purified EGFR, c-Src, c-Fyn, or a membrane fraction from untreated normal keratinocytes in tyrosine kinase assay buffer as indicated. Total reaction mixtures and PKC␦ immunoprecipitated from the reactions were analyzed for phosphotyrosine by immunoblotting with anti-phosphotyrosine antibody. In B, 10 g of membrane protein from untreated keratinocytes were incubated with the indicated clearing antibody followed by protein A-Sepharose, and the cleared lysate was added to recombinant PKC␦ for the tyrosine kinase assay. The reaction mixtures were immunoprecipitated for phosphotyrosine and stained with an anti-PKC␦ antibody. In C, primary keratinocytes were treated with 10 ng/ml TGF␣ for 5 min (5M) or 5 days (5D) as indicated, and cellular lysates immunoprecipitated with nonspecific rabbit IgG, c-Src, or c-Fyn antibodies. The immunoprecipitates were incubated with tyrosine kinase assay buffer ("Experimental Procedures"), and the phosphorylated proteins were analyzed by immunoblotting with an anti-phosphotyrosine antibody. Kinase refers to the band corresponding to either Src or Fyn kinase. IP, immunoprecipitation. tated from the kinase assays and the immunoprecipitates immunoblotted for the indicated PKC isozyme. Fig. 6B confirms that c-Src and c-Fyn phosphorylate PKC␦. The phosphotyrosine immunoprecipitations revealed that PKC⑀, , and were phosphorylated by c-Src and to a lesser extent by c-Fyn. These results suggest that PKC⑀, , and may also be tyrosinephosphorylated under conditions where Src family kinases are activated in cells. However, in three independent experiments, we were unable to detect tyrosine phosphorylation of other PKC isoforms in TGF␣-treated cells,demonstrating the speci-ficity of this phosphorylation for the ␦ isoform of PKC in vivo (data not shown). DISCUSSION In this report, we demonstrate that TGF␣ treatment induces tyrosine phosphorylation of PKC␦ and inhibits its activity. In view of the high level of TGF␣ produced by transformed mouse keratinocytes (15,16), these results provide a mechanism for the tyrosine phosphorylation of PKC␦ induced by oncogenic ras (5). Studies using keratinocytes from mice with mutant alleles of EGFR revealed that the EGFR is required for TGF␣-induced tyrosine phosphorylation of PKC␦, but biochemical analysis indicates that the EGFR does not phosphorylate PKC␦ directly. We demonstrate that c-Src and c-Fyn kinases become activated after TGF␣ treatment, and PKC␦, as well as PKC⑀, , and , are substrates for c-Src and c-Fyn kinases in vitro.
Activation of Src family kinases in response to EGFR-ligand interaction has been reported previously in other cell types (34,35). Several possible mechanisms have been proposed for the activation of Src family tyrosine kinase by EGFR. The Src SH2 domain can bind to activated, tyrosine-phosphorylated EGFR (36,37), and Src itself has been shown to phosphorylate the EGFR and create a Src SH2 docking site (35). C-Src can also bind to tyrosine-phosphorylated Neu, which can be transphosphorylated by activated EGFR (36,38). Finally, signaling molecules downstream from the EGFR such as p21 ras Gap and Raf-1 associate with Src family members, illustrating that direct association with the EGFR is not necessary for Src activation (39,40).
The mutant mouse strain waved-2 harbors a point mutation in the EGFR, which decreases its tyrosine kinase activity Ͼ90% (22), consistent with the lack of PKC␦ tyrosine phosphorylation in waved-2 keratinocytes after acute EGF treatment. In contrast, v-ras Ha keratinocytes from waved-2 mice have tyrosine-phosphorylated PKC␦, suggesting that EGFR ligands secreted by transformed keratinocytes are responsible for the PKC␦ tyrosine phosphorylation, just as chronic (4 days) treatment with EGF is capable of inducing PKC␦ phosphorylation in waved-2 keratinocytes.
The role of EGFR ligands in v-ras Ha -stimulated PKC␦ tyrosine phosphorylation was strongly supported by experiments using EGFR deficient keratinocytes (Fig. 3B). The increase in PKC␦ tyrosine phosphorylation in the EGFR Ϫ/Ϫ v-ras Ha keratinocytes was only 35% Ϯ 17% (mean Ϯ standard deviation, n ϭ 2) of that induced in the EGFR ϩ/ϩ v-ras Ha keratinocytes of the same strain. In the same experiments, the induction of PKC␦ tyrosine phosphorylation by TPA was almost identical in the EGFR ϩ/ϩ and Ϫ/Ϫ keratinocytes (18.9-fold and 18.4-fold, respectively), indicating that the EGFR kinase is not essential for PKC␦ tyrosine phosphorylation, that the EGFR null keratinocytes are competent to respond to other external stimuli, and that TPA targets PKC␦ through an independent pathway. These studies establish the critical role for EGFR signal transduction in altering PKC␦ tyrosine phosphorylation and subsequent enzymatic activity in Ras-transformed keratinocytes. Nevertheless, transduction of EGFR Ϫ/Ϫ keratinocytes with v-ras Ha induced a small increase in tyrosine phosphorylation of PKC␦, indicating that other pathways may contribute to Rasinduced PKC␦ tyrosine phosphorylation. For example, stimulation of phosphatidylinositol turnover by AlF 4 Ϫ slightly elevated PKC␦ tyrosine phosphorylation in normal keratinocytes (Fig. 1A), and inositol phosphate metabolism is up-regulated in v-ras Ha -transformed keratinocytes (9,41).
Tyrosine phosphorylation of PKC␦ has been reported in the promyeloid cell line 32D and NIH-3T3 fibroblasts treated with the tumor promoter TPA or PDGF (27,29). PKC␦ tyrosine phosphorylation may also play a role in the IgE receptor sig- FIG. 5. Subcellular distribution of PKC␦ after TGF␣ or TPA treatment. In A, keratinocytes were treated with 10 ng/ml TGF␣ for either 2 min or 4 days and either fractionated into soluble and particulate fractions or lysates prepared and immunoprecipitated with phosphotyrosine antibody. The fractionated and immunoprecipitated samples were immunoblotted with anti-PKC␦ antibody. In B, keratinocytes were treated with 100 nM TPA for 10 min, and PKC␦ translocation and tyrosine phosphorylation were assessed as described for A. IP, immunoprecipitation.
FIG. 6. In vitro tyrosine phosphorylation of PKC isozymes. In A, recombinant PKC isozymes from Sf-9 cells were incubated with either c-Src or c-Fyn kinase as described under "Experimental Procedures," and the total reaction mixtures were analyzed for phosphotyrosine by immunoblotting. To identify the location of each PKC isozyme, the blot was stripped and restained for individual PKC isoforms. Similar results were obtained in two additional experiments. In B, total tyrosine kinase reaction mixtures of the indicated PKC isozyme either alone or with c-Src or c-Fyn were immunoprecipitated with an antiphosphotyrosine antibody. The immunoprecipitates were immunoblotted with PKC antibodies specific for PKC␦, ⑀, , or as indicated.
naling pathway of mast cells (32) and in saliva production by parotid acinar cells in response to substance P or the muscarinic agonist carbachol (31). However, the published effects of tyrosine phosphorylation on PKC␦ activity are ambiguous. Tyrosine-phosphorylated PKC␦ isolated from ras Ha -transformed (5) and TGF␣-treated keratinocytes (Fig. 2) has reduced activity. Li et al. observed increased PKC activity in the membrane fraction of 32D/PDGF-␤R/PKC-␦ or NIH-3T3/PKC-␦ cells treated with either TPA or PDGF to increase the level of membrane-associated tyrosine-phosphorylated PKC␦ (29). However, this activity may be due to nontyrosine-phosphorylated PKC␦ or other PKC isozymes that translocate to the membrane fraction after TPA treatment. In vitro tyrosine phosphorylation of PKC␦ by Fyn, insulin receptor, or PDGF receptor also resulted in a Ͻ2-fold increase in PKC activity (27), but the relevance to tyrosine phosphorylation in intact cells is unknown because differences in substrates or assay conditions can influence the activity of PKC. For example, the activity of tyrosine-phosphorylated PKC␦ in response to activation of the IgE receptor showed decreased activity toward its physiological substrate, the Fc⑀RI␥ chain, and increased activity toward myelin basic protein (32). Li et al. have also generated a mutant PKC␦ that was constitutively phosphorylated on tyrosine and catalytically inactive, further supporting an association between tyrosine phosphorylation and inhibition of PKC␦ activity (42). PKC␦ can be tyrosine-phosphorylated on more than one site, and distinct phosphorylation sites may regulate activity differently (43). 2 To date, most experiments where the tyrosinephosphorylated PKC␦ is isolated from intact cells support a role for tyrosine phosphorylation in the inhibition of PKC␦ activity (5,32,42).
Translocation of PKC␦ in response to EGF is observed in some cell types (44), but not in others ( Fig. 5 and Ref. 31), and our results indicate that translocation is not a prerequisite for tyrosine phosphorylation. In unstimulated mouse keratinocytes, 30 -50% of the PKC␦ is localized to the particulate fraction (11,33). Furthermore, as shown in Fig. 4 (A and B), c-Src and c-Fyn constitute the major PKC␦ tyrosine kinase activity in the membrane fraction of keratinocytes where tyrosinephosphorylated PKC␦ is localized (5,29). Thus, a pool of particulate-associated PKC␦ exists in keratinocytes that can become tyrosine-phosphorylated upon activation of the appropriate kinase.
Specialized functions for individual PKC isozymes have been identified in several cell types (45). PKC␦ is involved in myeloid differentiation (46) and secretion in basophilic RBL-2H3 cells (47). PKC␦ also regulates cell cycle progression in CHO cells (48) and growth arrest in fibroblasts (49,50). In primary mouse keratinocytes, PKC␦ translocates in response to Ca 2ϩ -induced differentiation (11). Moreover, TPA-induced keratinocyte differentiation is inhibited by concentrations of bryostatin 1 (10 -1000 nM) that protect PKC␦ from down-regulation (33). Thus, activation of PKC␦ may be important for commitment to keratinocyte terminal differentiation. Both v-ras Ha and EGFR ligands modify keratinocyte differentiation in vitro, and EGFR activation reproduces a subset of phenotypic alterations characteristic of neoplastic v-ras Ha keratinocytes (9,14,17). Therefore, the common target of PKC␦ tyrosine phosphorylation for v-ras Ha and EGFR ligands may be relevant to the phenotypic alterations in epidermal neoplasia.
In addition to PKC␦, c-Src and c-Fyn phosphorylate other PKC isoforms (⑀, , and ) in vitro, but we have not detected tyrosine phosphorylation of any isoforms except PKC␦ in intact cells. The in vivo substrate specificity of c-Src and c-Fyn may be determined by the subcellular distribution of the kinases and substrates as well as direct physical associations. Thus, PKC isoforms such as PKC and , which are readily phosphorylated in vitro by c-Src and c-Fyn, may not be accessible to the appropriate tyrosine kinase in living cells. As can be seen from Fig.  4C, c-Src and c-Fyn have a different pattern of associated proteins, and these could influence their substrate specificity.
This report defines a novel connection between tyrosine kinase signaling and the PKC family of enzymes, which may have important functional consequences for epithelial growth, differentiation, and carcinogenesis. The EGFR and PKC pathways are two major signaling systems for epidermal keratinocytes. A better understanding of these signal transduction pathways and cross-talk between the different kinase cascades provides new insight for the design of drugs to treat diseases involving this cell type.