A Regulatory Role for p38δ MAPK in Keratinocyte Differentiation

p38 MAPK isoforms are important in the regulation of a variety of cellular processes. Among the four described p38 isoforms, p38α, β, and δ are expressed in keratinocytes (Dashti, S. R., Efimova, T., and Eckert, R. L. (2001) J. Biol. Chem. 276, 8059–8063). However, very little is known about how individual p38 isoforms regulate keratinocyte function. In the present study, we use okadaic acid (OA) as a tool to study the role of p38 MAPKs as regulators of keratinocyte differentiation. We demonstrate that OA activates p38δ but not other p38 isoforms. p38δ activation is increased as early as 0.5 h after OA addition, and activity is maximal at 8 and 24 h. ERK1 and ERK2 activity are reduced on an identical time course. We show that p38δ forms a complex with ERK1/2, and overexpression of p38δ inhibits ERK1/2 activity without reducing ERK1/2 level. Thus, p38δ may directly suppress ERK1/2 activity. Additional studies show that p38δ is expressed in the epidermis, suggesting a role for p38δ in regulating differentiation. To evaluate its function, we show that increased p38δ activity is associated with increased levels of AP1 and CAATT enhancer binding protein factors, increased binding of these factors to the involucrin (hINV) promoter, and increased expression. Moreover, these responses are maintained in the presence of SB203580, an agent that inhibits p38α and β, further suggesting a central role for the p38δ isoform. Dominant-negative p38 also inhibits these responses. These unique observations suggest that p38δ is the major p38 isoform driving suprabasal hINV gene expression and that p38δ directly regulates ERK1/2 activity via formation of a p38δ-ERK1/2 complex.

. However, very little is known about how individual p38 isoforms regulate keratinocyte function. In the present study, we use okadaic acid (OA) as a tool to study the role of p38 MAPKs as regulators of keratinocyte differentiation. We demonstrate that OA activates p38␦ but not other p38 isoforms. p38␦ activation is increased as early as 0.5 h after OA addition, and activity is maximal at 8 and 24 h. ERK1 and ERK2 activity are reduced on an identical time course. We show that p38␦ forms a complex with ERK1/2, and overexpression of p38␦ inhibits ERK1/2 activity without reducing ERK1/2 level. Thus, p38␦ may directly suppress ERK1/2 activity. Additional studies show that p38␦ is expressed in the epidermis, suggesting a role for p38␦ in regulating differentiation. To evaluate its function, we show that increased p38␦ activity is associated with increased levels of AP1 and CAATT enhancer binding protein factors, increased binding of these factors to the involucrin (hINV) promoter, and increased expression. Moreover, these responses are maintained in the presence of SB203580, an agent that inhibits p38␣ and ␤, further suggesting a central role for the p38␦ isoform. Dominant-negative p38 also inhibits these responses. These unique observations suggest that p38␦ is the major p38 isoform driving suprabasal hINV gene expression and that p38␦ directly regulates ERK1/2 activity via formation of a p38␦-ERK1/2 complex.
Mitogen-activated protein kinases (MAPK), 1 including p38, ERK1/2, and JNK1/2, are important regulators of cell function (1,2). The p38 MAPK family includes the p38␣, ␤, ␦, and ␥ isoforms (3). p38␣, ␤, and ␦ are expressed in keratinocytes (4), but only limited information is available regarding the function of each individual isoform. p38␦ is a recently discovered p38 isoform that has not been extensively studied, and very few physiological functions have been identified. Thus, although p38␦ is abundant in keratinocytes (4), the physiological function of p38␦ in keratinocytes has not been defined.
In the present study, we provide new information regarding the role of p38␦ in keratinocytes. We use okadaic acid (OA), a potent inhibitor of serine/threonine protein phosphatases 2A, 1, and 3 (5), as a tool to stimulate p38 activation. By binding to the catalytic subunits of protein phosphatase 2A, OA suppresses enzyme activity (6). Protein phosphatase 2A is a predominant serine/threonine phosphatase present in cultured human epidermal keratinocytes (7), and inhibition of phosphatase activity shifts the balance between phosphorylation and dephosphorylation in favor of the former, which leads to an accumulation of hyperphosphorylated proteins and ultimately results in signaling cascade activation and altered gene expression (8 -10).
Our studies show that OA treatment of keratinocytes increases p38 MAPK activity because of a selective activation of p38␦. This activation coincides with a simultaneous decrease in ERK1/2 activity. A recent study suggests that p38␣ and ERK1/2 kinases can associate in HeLa cells (11); however, this interaction has not been noted in other systems. Our study provides a first example that p38␦ is co-precipitated with another MAPK, ERK1/2. This suggests that p38␦ may directly interact with ERK1/2 or interact via other proteins to influence ERK1/2 activity and vice versa. Our studies also suggest that p38␦ activation is physiologically important, because differentiation-associated downstream responses are activated following p38␦ activation. These findings point to a unique interaction of p38␦ with ERK1/2 and also suggest that p38␦ is an important mediator of differentiation-associated gene activation.

MATERIALS AND METHODS
Reagents-Keratinocyte serum-free medium, gentamicin, trypsin, and Hanks' balanced salt solution were purchased from Invitrogen. OA and SB203580 were obtained from Calbiochem. Phorbol ester (12-O-tetradecanoylphorbol-13-acetate (TPA)) and dimethyl sulfoxide were purchased from Sigma. The pGL2-Basic plasmid and the chemiluminescent luciferase assay system were purchased from Promega. [␥-32 P]ATP was obtained from PerkinElmer Life Sciences. The human involucrin-specific polyclonal antibody was generated by injecting rabbits with recombinant human involucrin (12). Rabbit polyclonal antibodies specific for AP1 transcription factor family members c-Jun (SC-1694X), JunB (SC-46X), JunD (SC-74X), Fra-1 (SC-605X), Fra-2 (SC-171-X), c-Fos, and Pan-Fos (SC-253X), as well as rabbit polyclonal antibodies selective for C/EBP transcription factor family members C/EBP␣ (SC14AAX) and C/EBP␤ (SC-150X) were from Santa Cruz Biotechnology. Goat polyclonal antibody specific for p38␦ (SAPK4) (SC-7585) was obtained from Santa Cruz Biotechnology. Anti-FLAG M2 mouse monoclonal antibody (F-3165) was purchased * This work was supported by National Institutes of Health Grant AR39750 to the Skin Diseases Research Center of Northeast Ohio and other grants from the National Institutes of Health (to R. L. E.). 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  from Sigma. Protein A/G-agarose was obtained from Santa Cruz Biotechnology.
Plasmids-The structure of the hINV promoter construct pINV-241, which consists of nucleotides Ϫ241/Ϫ7 of the hINV promoter, linked to the luciferase reporter gene in pGL2-Basic, has been previously described (13). All of the positions are defined relative to the hINV gene transcription start site (13,14). Dominant-negative p38 MAPK and dominant-negative MEK3 were generously provided by Dr. Roger Davis (15,16).
Immunoblot Analysis-Total cell extracts or nuclear extracts were prepared from cultured human epidermal keratinocytes as described (18,19). Protein concentration of the samples was determined using Bradford Bio-Rad protein assay. Equal quantities of protein were electrophoresed on a 10% denaturing SDS-polyacrylamide gels and transferred to nitrocellulose. The blots were blocked and then incubated with an indicated primary antibody, washed, and exposed to an appropriate horseradish peroxidase-conjugated secondary antibody. Secondary antibody binding was detected using a chemiluminescent detection system (Amersham Biosciences).
In Vitro Kinase Assays-The activities of MAP kinases were measured using nonisotopic p44/42 (ERK1/2), p38 MAPK, and JNK/SAPK assay systems (New England BioLabs) (4,19,20). Briefly, keratinocyte total cell lysates were prepared under nondenaturing conditions. Protein concentration was determined, and equal amounts of total protein were used per each in vitro kinase assay sample. Active (phosphorylated) ERK1/2 and p38 kinases were selectively immunoprecipitated from cell lysates using immobilized, dual phospho-specific monoclonal antibodies. JNK/SAPK activity was selectively precipitated from cell lysates using c-Jun fusion protein glutathione-Sepharose beads. Precipitated kinases were then allowed to phosphorylate their major substrate proteins (Elk-1 for ERK1/2, ATF-2 for p38, and c-Jun for JNK/ SAPK) in a kinase reaction performed in the presence of ATP. Phosphorylation of the substrate proteins was measured by immunoblot using the corresponding phospho-specific antibody. The activity of adenovirus transduced FLAG epitope-tagged p38 MAPK isoforms was assessed in the similar fashion using anti-FLAG M2 mouse monoclonal antibody (Sigma; F3165) to immunoprecipitate specific p38 isoforms. The expression of individual p38 MAPK isoforms at a comparable level was confirmed by immunoblot using anti-FLAG antibody. To measure the activity of the endogenous p38␦ isoform, p38␦-specific antibody (SC-7585) was used to selectively immunoprecipitate this kinase followed by a kinase assay performed as described above.
Tissue Culture, Transient Transfection, and Luciferase Assay-Third passage normal human foreskin keratinocytes were transfected when ϳ50% confluent with 2 g of involucrin promoter-luciferase reporter construct, pINV-241, in the presence of 6 l of FuGENE 6 transfection reagent/dish (Roche Applied Science), according to the manufacturer's instructions. At 24 h post-transfection, the keratinocytes were treated with keratinocyte serum-free medium in the presence or absence of TPA (50 ng/ml), OA (10 nM), or the two reagents combined. For co-transfection experiments, the cells were transfected with 2 g of hINV promoter reporter plasmids and 2 g of empty control vector or vector encoding dominant-negative p38 in the presence of 12 l of FuGENE 6/dish. After 24 h, the keratinocytes were treated with 10 nM OA. After an additional 24 h of incubation, the cells were harvested, and the luciferase activity was measured as previously described, using Promega luciferase assay kit and a Berthold luminometer (13,19). All of the assays were performed in triplicate, and each experiment was repeated three times. The luciferase activity was normalized per g of protein.
The transfection efficiency was monitored using a green fluorescent protein expression plasmid (Clontech) (21).
Reverse Transcription-PCR Analysis of hINV mRNA Level-Total RNA was prepared from cultured keratinocytes and treated for 24 h FIG. 1. Okadaic acid activates p38 MAPK and JNK1/2 and inhibits ERK1/2 in keratinocytes. A, kinase activity of p38, JNK1/2, and ERK1/2. Cultured human epidermal keratinocytes were treated with 100 nM OA for 24 h. The cells were then lysed, total cell extracts were prepared, and active forms of p38, JNK, or ERK1/2 MAP kinases were immunoprecipitated from samples containing equal amount (200 g) of protein, using immobilized monoclonal phospho-p38 antibody, c-Jun fusion protein glutathione-Sepharose beads, or immobilized monoclonal phospho-ERK1/2 antibody, respectively. The ability of each immunoprecipitated kinase to phosphorylate the appropriate substrate was analyzed by immunoblot using antibodies specific to phospho-ATF-2, phospho-c-Jun, and phospho-Elk1, respectively. B, p38, JNK, and ERK1/2 total protein levels were measured by immunoblot using specific antibodies. The ␤-actin blot is included as indicator of equal protein loading. C, kinetics of OA-induced changes in activity of p38, JNK1/2, and ERK1/2 in keratinocytes. Cultured cells were treated with OA for the indicated time periods. After that the cells were lysed, and total cell extracts were prepared. p38 and JNK1/2 activities were measured exactly as described for A. ERK1/2 activity was assessed by immunoblot using phospho-specific ERK1/2 antibody that recognizes only phosphorylated activated form of ERK1/2 protein. In a parallel experiment, ERK1/2 activity was measured by kinase assay with similar results (not shown). The total ERK1/2 protein level was assayed by immunoblot. These experiments were repeated a minimum of three times with similar results. with 0 or 100 nM OA, using RNeasy RNA isolation kit (Qiagen; 75142). RNA was quantified by spectrophotometry prior to PCR analysis. Ribosomal 18 S RNA levels were monitored as an internal control using specific primers. The hINV upstream and downstream PCR primers are 5Ј-CTCCACCAAAGCCTCTGC and 5Ј-CTGCTTAAGCTGCTGCTC. The expected size of the results PCR product is 380 bp. Reverse transcription-PCR was performed using Titan One Tube reverse transcription-PCR system (Roche Applied Science; 1855476). Briefly, each PCR reaction mixture (25 l) contained 1.5 mM MgCl 2 , 20 M of each primer, 1 g of RNA, 1 l of the enzyme blend, which consisted of reverse transcriptase avian myeloblastosis virus, Taq DNA polymerase, proofreading polymerase, and optimized reaction buffer. The reaction buffer contained 0.2 mM of dATP, dCTP, dGTP, and dTTP and 5 mM dithiothreitol. The reverse transcription step was performed at 55°C for 30 min. The complex was denatured at 94 C for 1 cycle and amplified for 35 cycles (denaturation, 94°C for 10 s; annealing, 55°C for 30 s; and elongation, 68°C for 4 min).

RESULTS
OA Regulation of MAPK Activity-We treated keratinocytes with 100 nM OA for 24 h and then monitored the effects on MAPK activity. To monitor p38 MAPK activity, active (phosphorylated) p38 MAPK was immunoprecipitated using an antibody that reacts with the phosphorylated form of all p38 MAPK isoforms (␣, ␤, ␥, and ␦). The ability of the precipitated activated p38 MAPK to phosphorylate its substrate protein, ATF-2, was then examined by kinase assay. As shown in Fig.  1A, OA treatment markedly increases p38 MAPK activity. Assay of JNK1/2 activity, based on ability of to JNK1/2 to phosphorylate c-Jun, revealed a slight increase. ERK1/2 activity is detected in cultured keratinocytes in the absence of OA stimulation. However, in contrast to p38 MAPK and JNK1/2 activities, ERK1/2 activity (P-Elk1 formation) is suppressed by OA treatment. To confirm that these changes are due to altered kinase activity and not to changes in kinase level, we prepared extracts and directly measured p38, JNK1/2, and ERK1/2 levels by immunoblot. Fig. 1B demonstrates that the levels of these enzymes are not changed by OA treatment.
Because the extent and rate of MAPK activation is thought to influence subsequent downstream events, we monitored the time course of kinase activity following OA treatment. Keratinocytes were treated with 100 nM OA for times ranging from 0 min to 24 h. To measure p38 activity, keratinocytes were harvested at each time point, and phospho-p38 was precipitated. We then monitored the ability of the precipitated kinase to phosphorylate ATF-2. As shown in Fig. 1C (top panel), p38 activity is detected by 4 h and is maximally increased by 8 h. To differentiate the effects of OA on ERK1 versus ERK2 activity, we prepared cell extracts and assayed for the presence of phosphorylated (active) ERK1/2. The upper ERK1/2 panel in Fig. 1C shows that P-ERK levels decrease beginning at 4 h and that both ERK1 and ERK2 are inhibited. The lower panel shows that there are no changes in total ERK1 or ERK2 concentrations during the time course. The JNK1/2 panel in Fig. 1C shows that JNK1/2 activity is transiently increased at 4 h and then returns to a level that is slightly elevated over untreated cells.
Differential Regulation of p38 Isoforms by OA-The above study indicates that p38 activity is increased by OA treatment.
FIG. 2. p38␦ is the major okadaic acid-responsive p38 MAPK isoform in keratinocytes. A, to measure the effect of OA treatment on the kinase activity of individual p38 isoforms, keratinocytes were infected with EV or FLAG epitope-tagged p38␣, ␤, ␥, or ␦ isoforms. After 24 h, keratinocytes were treated with 100 nM OA for the indicated time intervals. The cells were then lysed, total cell extracts were prepared, and the FLAG-p38 isoforms were immunoprecipitated using anti-FLAG monoclonal antibody. The activity of each immunoprecipitated isoform was determined based on the ability of the kinase to phosphorylate its substrate protein ATF-2. Phosphorylation of ATF-2 was measured by immunoblot using a phospho-ATF-2 antibody. B, the expression of individual p38 MAPK isoforms was confirmed by immunoblot, at t ϭ 0 and 24 h after the start of OA treatment, using anti-FLAG antibody. The ␤-actin blot is included as indicator of equal protein loading. C, endogenous p38␦ activity (top panel) and protein levels (middle panel) were measured in cells treated with 100 nM OA for the specified time periods. After that cells were lysed, and endogenous p38␦ protein was immunoprecipitated using p38␦-specific antibody followed by kinase assay using ATF-2 as a substrate. Phosphorylation of ATF-2 was measured by immunoblot using a phospho-ATF-2 antibody (P-ATF2). p38␦ protein expression was analyzed by immunoblot using p38␦-specific antibody (p38␦). The ␤-actin blot is included as an control to assure equal loading. D, SB203580 does not suppress OA-induced activation of p38 MAPK. Keratinocytes were pretreated with or without 2 M SB203580 for 30 min prior to administration of 100 nM OA for 24 h. The cells were then lysed, and p38 activity was measured by kinase assay with ATF-2 as a substrate. The ␤-actin level was monitored as a loading control. E, keratinocytes were infected with 15 MOI of adenovirus EV or adenovirus encoding p38␦. After 48 h, the cells were harvested and assayed for ERK1/2 levels and activity as described for Fig. 1C. The ␤-actin levels were monitored to normalize gel loading.
However, p38 is expressed as four isozymes, p38␣, ␤, ␦, and ␥ (22)(23)(24)(25)(26), that can produce different responses in cells. To identify the responsive p38 isoform, the cells were infected with adenoviruses encoding FLAG-tagged p38␣, ␤, ␦, or ␥ and equilibrated for 24 h prior to the start of treatment with 100 nM OA. The cell extracts were prepared at time points from 0 min to 24 h after the start of OA treatment, and each adenovirusdelivered p38 isoform was immunoprecipitated using an anti-FLAG antibody. The precipitated kinase was then assayed for the ability to phosphorylate ATF-2. As shown in Fig. 2A, p38␣ activity is modestly increased at 0.5-24 h, and p38␥ activity is elevated at 24 h. In contrast, p38␤ activity is not regulated. The p38␦ isoform exhibits the most dramatic increase in activity. Activity begins to increase at 30 min and is maximal at 8 -24 h. This result suggests that p38␦ is the major OA-responsive p38 isoform. Fig. 2B confirms that expression of each FLAG-tagged p38 isoform is stable over the 24-h OA treatment time course. This was monitored by isolating FLAG-p38␣-, ␤-, ␦-, or ␥-expressing cells at the beginning of OA treatment (t ϭ 0) and at t ϭ 24 h and measuring the level by anti-FLAG immunoblot. ␤-Actin levels were monitored to assure equal loading.
The above results suggest that p38␦ is the major OA-responsive p38 isoform. To confirm that the endogenous p38␦ isoform is activated, keratinocytes were treated with 100 nM OA for 0, 6, 16, and 24 h. Endogenous p38␦ was precipitated using a p38␦-specific antibody, and the precipitated kinase was evaluated for the ability to phosphorylate ATF-2. Fig. 2C (top panel) shows that endogenous p38␦ is activated by OA treatment. The middle panel displays an immunoblot showing that the increased activity is not due to an increase on p38␦ levels. A ␤-actin blot is included to normalize for differences in sample loading.
SB203580 is an inhibitor of the p38␣ and p38␤ isoforms but does not inhibit p38␦ (25,27). As an additional method of confirming that p38␦ is the isoform that transmits the OA signal downstream, we treated keratinocytes with OA in the presence or absence of 2 M SB203580. After 24 h, the extracts were prepared, total activated p38 (phosphorylated p38␣, ␤, ␦, and ␥) was precipitated using anti-phospho-p38, and the ability of the precipitated p38 kinase to phosphorylate ATF-2 was determined. As shown in Fig. 2D, OA increases p38 activity, but this increase is not inhibited by SB203580, further confirming that the responsive isoform is p38␦. Although p38␥ is activated at 24 h ( Fig. 2A), our previous studies show that p38␥ is not expressed in cultured keratinocytes (4).
The above results suggest that increased p38␦ activity is associated with reduced ERK1/2 activity. Because overexpression of p38␦ is known to increase the amount of active p38␦ (4), we tested whether p38␦ expression, in the absence of OA treatment, reduces ERK1/2 activity. As shown in Fig. 2E, p38␦ expression causes a marked reduction in both ERK1 and ERK2 (P-ERK1 and P-ERK2) activity, as compared withy cells infected with empty vector (EV).
p38␦ Associates with ERK1/2 Kinase-One possible explanation for the p38␦-dependent reduction in ERK1/2 activity is an interaction of p38␦ with ERK1/2. A recent study using HeLa cells identified an ERK1/2-p38␣ complex and suggested that the complex may be physiologically important (11). However, no interaction of this kind has been identified in other systems or with other p38 MAPK members. We therefore assessed whether ERK1/2 and p38␦ form a complex in keratinocytes. Keratinocytes were infected with p38␦-encoding adenovirus and then treated with OA for 0 -4 h. The extracts were prepared, and ERK1/2 was immunoprecipitated. The ERK1/2 immunoblot in Fig. 3A confirms ERK1/2 immunoprecipitation. The p38␦ immunoblot demonstrates that p38␦ is co-immuno-precipitated with ERK1/2. To confirm the specificity of the ERK1/2 precipitation, we performed a parallel precipitation with anti-IgG. As expected, no ERK1/2 or p38␦ precipitation was observed (Fig. 3B). We also performed the inverse precipitation experiment. FLAG-tagged p38␦ (FLAG-p38␦) was expressed in keratinocytes and then immunoprecipitated using anti-FLAG. The left panel in Fig. 3C shows that the anti-FLAG reagent precipitates FLAG-p38␦. The right panel shows that endogenous ERK1/2 is co-precipitated with p38␦.
The above results were obtained using expressed MAPK FIG. 3. p38␦ and ERK1/2 form a complex. A, keratinocytes were infected with 15 MOI of adenovirus EV or adenovirus encoding p38␦. After 48 h, the cells were treated as indicated with 100 nM OA. The extracts were then prepared, and equal quantities of extract (200 g of protein) were immunoprecipitated (IP) with ERK1/2-conjugated agarose beads (Santa Cruz). Precipitated material was electrophoresed on a 10% gel and then immunoblotted (IB) with the indicated antibody. The arrows indicate the migration of p38␦ and ERK1/2, and the asterisks indicate antibody reactivity with IgG. B, extracts from A were immunoprecipitated with anti-IgG as a control for nonspecific binding and then blotted with the indicated antibodies. C, keratinocytes were infected with 15 MOI of adenovirus encoding FLAG-p38␦. After 48 h, the extracts were prepared, and FLAG-p38␦ was immunoprecipitated using anti-FLAG M2 reagent. The sample was electrophoresed and then blotted with anti-p38␦ or anti-ERK1/2. D, keratinocytes were treated for 2 h with or without 100 nM OA. The extracts were then prepared, and equivalent quantities of extract (200 g of protein) were immunoprecipitated with ERK1/2-conjugated agarose beads or anti-IgG (with protein A/G-agarose). The proteins were separated on a 10% polyacrylamide gel and then immunoblotted as indicated.
p38␦ Regulates Keratinocyte Differentiation proteins. We next confirmed that the endogenous kinases also associate. We prepared cell extracts and precipitated with an ERK1/2-specific antibody. The results (Fig. 3D) show that endogenous ERK1/2 is precipitated (lower panels) and confirm that p38␦ is co-immunoprecipitated with ERK1/2 (upper panels). As a control, we show that nonspecific IgG did not precipitate these kinases (right panels). The asterisk in each panel indicates antibody interaction with the IgG derived from the immunoprecipitation.
p38␦ Expression in Human Epidermis-The above results suggest that p38␦ is functionally important in cultured keratinocytes. However, although p38␦ has been detected in developing mouse epidermis (28), evidence demonstrating that p38␦ is expressed in human epidermis has not been presented. We therefore prepared extracts of adult human epidermis for detection of p38␦ by immunoblot. Fig. 4A compares p38␦ obtained from epidermal extracts with authentic p38␦ obtained from cultured keratinocytes infected with adenovirus producing authentic p38␦. This experiment clearly shows that p38␦ is expressed in epidermis. Fig. 4B confirms expression in all epidermal layers.
OA Regulation of AP1 and C/EBP Transcription Factor Expression-The above studies suggest that p38␦ is activated by OA and also suggests that it may have a physiologic role, because it is expressed in human epidermis. Our next effort was to identify downstream responses associated with p38␦ activation. C/EBP and AP1 transcription factors are known to be important in regulation of keratinocyte differentiation (13, 18, 19, 29 -36). We therefore examined the effects of OA treatment on AP1 and C/EBP factor expression. The cells were treated with 100 nM OA for 24 h, and the nuclear extracts were prepared for immunoblot with AP1 and C/EBP factor-specific antibodies. Fig. 5A shows that OA treatment increases c-Jun, JunB, JunD, Fra-1, and Fra-2 levels and also increases C/EBP␣ and C/EBP␤ levels. We next examined the time course of the OA-dependent increase. The cells were treated for 0.5-24 h with 100 nM OA, and C/EBP␣, C/EBP␤, and Fra-1 levels were monitored by immunoblot. As shown in Fig. 5B, the C/EBP␣, C/EBP␤, and Fra-1 levels are optimally increased by 8 h after initiation of treatment. Thus, the time course of increase is remarkably consistent with that observed for p38␦ activation as shown in Fig. 2. To confirm that the response was not mediated by p38␣ or p38␤, we treated cells with OA in the presence or absence of SB203580 for 24 h and then assayed C/EBP␤ and Fra-1 levels by immunoblot. As shown in Fig. 5C, SB203580 does not inhibit the response, a result that is consistent with the regulation being p38␦-dependent. It is important to note that additional exposure of the film reveals expression of each of the AP1 and C/EBP factors in untreated cells. These exposures are not shown, because they result in overexposure of the OA-treated lanes.
OA Regulates Differentiation-associated Gene Expression-Involucrin is a marker of differentiation that is regulated in an AP1-and C/EBP transcription factor-dependent manner (13,18,19,36). We have used it as an end point of OA and p38␦ action. Keratinocytes were transfected with pINV-241, a plasmid that encodes the proximal involucrin promoter linked to luciferase (13). Fig. 6A shows that both OA and TPA treatment increases hINV promoter activity and that the activation is FIG. 4. p38␦ is expressed in adult human epidermis. A, extracts prepared from human epidermis (EPI) were electrophoresed on a denaturing and reducing 8% acrylamide gel, transferred to nitrocellulose, and blotted with p38␦-specific antibody. As a control, extract prepared from keratinocytes infected for 48 h with FLAG-p38␦-encoding adenovirus was electrophoresed in the adjacent lane (KER). ␤-Actin was detected on a parallel blot as a control. Binding of the primary antibodies was detected using an appropriate secondary antibody. B, human epidermal sections were stained with anti-IgG as a control or anti-p38␦. Binding to the primary antibodies was visualized using horseradishconjugated peroxidase and chemiluminescent detection reagent. The epidermis is indicated by the brackets in each panel.

FIG. 5. AP1 and C/EBP transcription factor expression is regulated by OA.
A, keratinocytes were treated with 100 nM OA for 24 h (AP1 factors) or 8 h (C/EBP factors). Nuclear extracts were then prepared and assayed for AP1 and C/EBP protein expression by immunoblot using antibodies against specific AP1 and C/EBP factors. Gel scanning of longer film exposures that show the bands in the untreated group indicate that the major c-Jun, JunB, JunD, Fra-1, Fra-2, C/EBP␣, and C/EBP␤ bands are increased ϳ6-, 4-, 8-, 12-, 4-, 8-, and 6-fold, respectively. B, time course of OA-dependent activation of C/EBP and AP1 factor expression. The cells were treated with 100 nM OA for the specified time periods followed by nuclear extract preparation and immunoblot analysis using antibodies against individual AP1 and C/EBP proteins. C, SB203580 does not suppress OA-dependent increase in AP1 and C/EBP expression. Keratinocytes were pretreated with or without 2 M SB203580 for 30 min prior to administration of 100 nM OA for 24 h. Nuclear extracts were prepared, and immunoblot was performed using antibodies against Fra-1 and C/EBP␤ proteins. enhanced by OA/TPA co-treatment. This regulation is mediated via an AP1-dependent mechanism, because mutation of the functionally important AP1-1 site (13), in construct pINV-241(AP1-1m), results in a loss of the regulation. In addition, the OA-dependent regulation is lost when the functionally important C/EBP site (18) is inactivated by mutation. To determine whether OA treatment and the concomitant p38␦ activa-tion translates into enhanced binding of AP1 and C/EBP factors to their respective binding sites on the hINV promoter, we performed gel mobility shift studies using oligonucleotides encoding the hINV promoter AP1-1 and C/EBP sites (13,18). The level of AP1 factor binding is markedly increased in cells that have been treated with 100 nM OA (Fig. 6B, left panel). In addition, the shifted band migrates with a unique mobility   FIG. 6. Involucrin promoter expression is regulated by okadaic acid. A, hINV promoter activity is increased by OA treatment. Cultured human epidermal keratinocytes were transfected with the 2 g of pINV-241, pINV-241(AP1-1m), or pINV-241(C/EBPm) reporter plasmid. Twenty-four hours later the cells were treated with keratinocyte serum-free medium in the presence or absence of 50 ng/ml TPA, 10 nM OA, or the two reagents combined. After an additional 24 h of incubation, the cells were harvested, and the total cell extracts were prepared and assayed for luciferase activity. B, AP1 factor DNA binding activity is increased by OA. The nuclear extracts were prepared from keratinocytes treated for 24 h with or without 100 nM OA. To measure AP1 binding to DNA, nuclear extract from untreated or OA-treated cells (2 g of protein/sample) was incubated with double-stranded 32 P-labeled oligonucleotide encoding the hINV promoter AP1 and electrophoresed (left panel). The specificity of binding was demonstrated by incubating extract from OA-treated cells with radioinert oligonucleotide encoding the intact AP1-1 site, a mutant AP1-1 site, or an authentic Sp1 site at a 50-fold molar excess. FP indicates free probe migration, and NE indicates nuclear extract. C, gel mobility supershift analysis. Extract (2 g of protein) prepared from OA-treated cells was treated with AP1-1-P 32 and the indicated antibody (or no antibody) and then electrophoresed as outlined above. D, C/EBP factor DNA binding activity is increased by OA. Nuclear extracts prepared from keratinocytes treated for 24 h with or without OA (left panel) were incubated with double-stranded oligonucleotides, 32 P-C/EBP, encoding the hINV C/EBP site, and then electrophoresed. To assess the specificity of binding, extract from OA-treated cells was incubated with double-stranded 32 P-C/EBP in the presence of a 50-fold molar excess of radioinert C/EBP or C/EBPm. To assess C/EBP␣ binding, anti-C/EBP␣ was added to the extract/probe mixture prior to electrophoresis. compared with the band that is observed in untreated extracts. As shown in the right panel of Fig. 6B, this band can be competed by addition of a 50-fold excess of the homologous DNA fragment but not by a fragment containing an AP1 site mutation or an double-stranded oligonucleotides encoding and Sp1 site. Fig. 6C characterizes individual AP1 factor binding to the AP1-1 region using extracts prepared from OA-treated cells. Antibodies specific for Fra-1, Fra-2, JunB, and JunD produce supershifted bands, and a Fos antibody, Pan-Fos, which detects all Fos factors, efficiently reduces binding.
In addition, as shown in Fig. 6D (left panel), C/EBP factor binding to the C/EBP site is markedly enhanced by OA treat-ment. Moreover, this binding is specific because it is competed by homologous radioinert C/EBP oligonucleotides but not by the same nucleotide in which the C/EBP site is mutated (C/ EBPm). In addition, some C/EBP␣ is present in this complex, as evidenced by the C/EBP␣ antibody-dependent supershift.
OA Regulation of Endogenous hINV Gene Expression-To be physiologically relevant, OA-dependent p38␦ activation should regulate expression of the endogenous hINV gene. To test this, we treated keratinocytes for 24 h with 100 nM OA and prepared total cell extracts for immunodetection of hINV (12). Fig. 7A (left panel) shows that OA treatment increases hINV protein level. A ␤-actin blot is included as an indicator to normalize protein loading. The right panel in Fig. 7A shows that hINV mRNA levels change in parallel. Gel scanning reveal that hINV level increase by 4 -6-fold in several experiments. To provide additional evidence for a p38␦ role in this regulation, we treated with SB203580, a pharmacological agent that inhibits p38␣ and ␤ but does not inhibit p38␦ (25,37). Fig. 7B shows that the increase in endogenous hINV gene expression is not inhibited by this agent, suggesting that the regulation is mediated by p38␦. To confirm this, Fig. 7C shows that expression of dominant-negative p38 inhibits the OA-dependent increase in promoter activity, confirming that the targeted knockout of p38 activity can blunt the response. In addition, dominant-negative MEK3 also antagonized the OA-dependent increase.

Regulation of p38 MAPK Activity in Keratinocytes-
The cells respond to extracellular signals by generating intracellular instructions that coordinate cellular responses. The MAPK cascades are among the best characterized of these intracellular signaling pathways (2). These cascades consist of a three-kinase module that includes a MEK kinase that activates a MEK that activates a MAPK (2). Three distinct MAP kinase families have been identified in mammalian cells, including mitogenresponsive ERKs, stress-responsive JNK/SAPKs, and p38 MAP kinases. Activated MAPKs translocate to the nucleus where they phosphorylate and activate transcription factors and other target proteins (2, 39). The total cell extracts were then prepared and assayed for hINV protein expression by immunoblot. In parallel cultures, mRNA was prepared, and the hINV mRNA level was monitored by reverse transcription-PCR. Ribosomal 18 S RNA level was monitored in parallel to assure equal gel loading. B, OA-dependent increase in hINV protein expression is not blocked by pretreatment with SB203580. Keratinocytes were pretreated with or without the indicated concentrations of SB203580 for 30 min prior to administration of 100 nM OA for 24 h. The nuclear extracts were prepared, and immunoblot was performed using antibody against hINV. ␤-Actin level was monitored as a loading control. C, dominant-negative p38 (dnp38) and dominant-negative MEK3 (dnMEK3) inhibit the OA-dependent activation of gene expression. The keratinocytes were transfected with 2 g of pINV-241 and hINV promoter reporter plasmid (13) in the presence of 2 g of EV or vector encoding dnp38 or dnMEK3 and treated with 10 nM OA for 24 h. After an additional 24 h, the cells were harvested, and the extracts were assayed for luciferase activity. OA treatment inhibits phosphatase activity and indirectly activates signal transduction kinase activity. This results in MEK3 activation, and MEK3 phosphorylates and activates p38␦. p38␦ and ERK1/2 exist as a preformed complex in keratinocytes. p38␦ activation results in a simultaneous inactivation of ERK1/2 activity via an unknown mechanism that could involve direct interaction between ERK1/2 and p38␦ or interaction via an intermediary protein(s). ERK1/2 inactivation in turn reduces the ability of the cell to survive and proliferate. p38␦ triggers downstream signals that increase transcription factor expression. The transcription factors in turn interact with the hINV promoter to increase hINV promoter activity.

p38␦ Regulates Keratinocyte Differentiation
There are four known p38 MAPK isoforms: ␣, ␤, ␦, and ␥ (22-25) (26). Three of these isoforms are expressed in keratinocytes (␣, ␤, and ␦) (4), and recent studies suggest they function as regulators of keratinocyte differentiation (4,19,20). However, little information is available regarding the role of individual p38 isoforms. In the present study, okadaic acid, a phosphatase inhibitor, is used to activate MAPK signaling and study p38 function. Our studies show that OA increases p38 kinase and JNK kinase activity and decreases ERK1/2 activity. The increase in JNK1/2 activity is transient and was not studied in the present report. In contrast, the increase in total p38 activity is substantial and sustained; activity increases at 4 h, and maximal levels are observed at 8 and 24 h. A detailed examination of individual p38 isoforms reveals that only p38␦ is activated by OA stimulation. Parallel studies using pharmacologic inhibitors and dominant-negative p38 kinase further suggest that p38␦ mediates the OA-dependent regulation. These findings are consistent with reports in other systems indicating that p38 isoforms are selectively activated in response to particular stimuli (40 -44). We also evaluated the ability of the MEK kinase, MEK3, to regulate OA-and p38␦dependent responses. This kinase has been reported to modulate p38 activity in other systems (22,40). Studies using dnMEK3 indicate that MEK3 activity is required for OA-dependent regulation, suggesting that MEK3 is the kinase that functions immediately upstream of p38␦ (Fig. 8). This result suggests that okadaic acid activates a cascade that includes MEK3 and p38␦. Additional studies (not shown) indicate that protein kinase C activity is not required for the OA-dependent regulation. This is expected, because OA, which increases the level of active phosphorylated kinase, activates the signaling cascade downstream of protein kinase C.
A p38␦-ERK1/2 Complex in Keratinocytes-The kinetics of increase in p38 activity is associated with a corresponding change in ERK1/2 activity. An important feature of this regulation is the inverse relationship between p38 and ERK1/2 activity. An important question is the mechanism that achieves this inverse regulation. Our results show that p38␦ and ERK1/2 form a complex in keratinocytes. The presence of this complex was confirmed by reciprocal co-immunoprecipitation using expressed p38 MAPK as bait and also by demonstrating an interaction between endogenous ERK1/2 and p38␦. Moreover, when p38␦ was overexpressed, a decrease in ERK1/2 activity was observed in the absence of any change in ERK1/2 level. These are the first findings suggesting an interaction between ERK1/2 and p38␦ in keratinocytes or any other cell type. This unique finding suggests that p38␦, by association with ERK1/2, may directly influence ERK kinase activity. Whether this involves direct interaction or an intermediary protein will require further study. The only other known example of this type of interaction is a p38␣-ERK1/2 complex that has been described in HeLa cells. This complex forms upon treatment with stimulating agent (11). This interaction appears to block ERK1/2 phosphorylation/activation by upstream MEK1/2 kinase (11). The complex we describe differs in two ways. First, it involves p38␦ and not p38␣, and second, ERK1/2 and p38␦ appear to be constitutively associated (i.e. association is not stimulus-dependent). Our study and the study by Zhang et al. (11) suggest that ERK and p38 MAPKs may have motifs designed to facilitate interaction (40). Additional studies will be necessary to identify the interaction motifs. Based on these results, we propose a simple model in which OA activates MEK3, and MEK3 activates p38␦. Then p38␦, either by covalently modifying ERK1/2, by making it inaccessible to upstream kinases (MEKs), or by causing a conformation change in ERK1/2, causes its inactivation. It should be emphasized that this influence may be conferred by direct interaction or via interaction with another protein or proteins. Additional studies show that p38␦ activation increases transcription factor level and activity, which in turn increases hINV gene expression (Fig. 7). This includes an increase in Fra-1, Fra-2, JunB, and JunD binding to the hINV AP1-1 site and increased C/EBP␣ binding to the C/EBP site. Although these studies provide some clues regarding the composition of these OA-stimulated transcriptional complexes, additional studies will be required to identify other participants. More studies are also required to understand the mechanism whereby p38␦ activation alters nuclear transcription factor level and activity. It is well established that MAPKs can translocate to the nucleus and directly or indirectly regulate transcription factors (45). This regulation could alter the AP1 and C/EBP factor level (46) and/or the phosphorylation state (47,48). Additional studies will be needed to provide this information.
Expression of p38␦ in Human Epidermis and Regulation of Downstream Responses-The above results suggest that p38␦ can be activated by OA in cultured keratinocytes. However, it was not clear that this enzyme has a physiological role in vivo, because it has not been demonstrated in human epidermis. We show by immunoblot of epidermal extracts and immunohistological staining of skin sections that p38␦ is expressed in skin. Moreover, staining was observed throughout the epidermis. At present, we do not know whether the enzyme is differentially activated (phosphorylated) in specific layers. It is possible that p38␦ activation is controlled by the presence of its activator. For example, protein kinase C, which plays a key role in regulation of terminal keratinocyte differentiation, is expressed in the suprabasal epidermal layers (49) and activates p38␦ (50). Hence, the differentiation-specific expression of upstream activators may ensure that p38␦ is activated only in the suprabasal layers. However, the availability of downstream targets may also be a point of regulation. For example, C/EBP␣ expression is restricted to the upper suprabasal epidermal layers (31,33), and AP1 factor expression is also differentiationdependent (51). Thus, p38␦ may be active throughout the epidermal layers, but differentiation-associated responses may only be engaged when the appropriate downstream target is present.
In cultured keratinocytes p38␦ clearly activates involucrin gene expression. Involucrin is a differentiation marker that is expressed in the spinous and granular epidermal layers in vivo (52,53). Involucrin gene expression is regulated by AP1 and C/EBP transcription factors (13,18,19,36,38). Our results demonstrate that OA-dependent p38␦ activation is associated with increased AP1 and C/EBP transcription factor level, increased binding of AP1 and C/EBP factors to the appropriate sites on the hINV promoter, and increased hINV promoter activity. These responses are not inhibited by SB203580, an inhibitor of ␣ and ␤ isoforms of p38 MAPK. Thus, these studies suggest that p38␦ activates hINV gene expression. Moreover, this regulation requires the presence of intact AP1 and C/EBP sites within the hINV promoter, suggesting that the OA-and p38␦-associated increase in AP1 and C/EBP activity is responsible for the regulation. Moreover, expression of the endogenous hINV gene is also increased, suggesting that the promoter response is physiologically meaningful. Based on these studies, we propose that OA activates differentiation via a signaling cascade that includes MEK3 and p38␦. This cascade in turn regulates AP1 and C/EBP transcription factor function to regulate hINV gene expression. A novel component of this response involves a p38␦-dependent decrease in ERK1/2 activity.