UVA Induces Ser 381 Phosphorylation of p90 RSK /MAPKAP-K1 via ERK and JNK Pathways*

UVA exposure plays an important role in the etiology of skin cancer. The family of p90-kDa ribosomal S6 kinases (p90 RSK /MAPKAP-K1) are activated via phosphorylation. In this study, results show that UVA-induced phosphorylation of p90 RSK at Ser 381 through ERKs and JNKs, but not p38 kinase pathways. We provide evidence that UVA-induced p90 RSK phosphorylation and kinase activity were time- and dose-dependent. Both PD98059 and a dominant negative mutant of ERK2 blocked ERKs and p90 RSK Ser 381 phosphorylation, as well as p90 RSK activity. A dominant negative mutant of p38 kinase blocked UVA-induced phosphorylation of p38 kinase, but had no effect on UVA-induced Ser 381 phosphorylation of p90 RSK or kinase activity. UVA-induced p90 RSK phosphorylation and kinase activity were markedly attenuated in JnK 1 2 / 2 and JnK 2 2 / 2 cells. A dominant negative mutant of JNK 1 inhibited UVA-induced JNKs and p90 RSK phosphorylation and kinase activity, but had no effect on ERKs phosphorylation. PD169316, a novel inhibitor of JNKs and p38 kinase, inhibited phosphorylation

The incidence of nonmelanoma and melanoma skin cancers has been increasing for several decades in most parts of the world (1-3), but mainly in populations of European origin (3,4). Approximately 90% of nonmelanoma skin cancers are thought to be caused by ultraviolet (UV) 1 exposure (5)(6)(7). The UV part of the solar electromagnetic spectrum is divided into UVC (200 -290 nm), UVB (290 -320 nm), and UVA (320 -400 nm) (5,8), and UVA is also subdivided into UVA2 (320 -340 nm) and UVA1 (340 -400 nm) (9,10). Although UVC radiation can induce skin cancers by damaging DNA (11), UVC is not pertinent to sunlight-induced human carcinogenesis because UVC is completely absorbed by the earth's stratospheric ozone layer and does not reach the surface of the earth (9). On the other hand, although UVB is only absorbed partially by the ozone layer (5,6,9) and can induce skin cancer by generating DNA damage (11), the proportion of UVB in the solar UV is small. Therefore, the risk of UVB-induced acute and chronic damage, including skin cancer (4,8,11), is diminished and can be blocked by limiting sunlight exposure during midday hours, wearing protective clothing, and using sunscreens (4,8). On the other hand, UVA is not absorbed by the ozone layer and thus the amount of UVA radiation reaching the earth's surface is ϳ20 times greater than that of UVB (5,8). Therefore, UVA may be a major component of the solar UV radiation contributing to skin cancer. Results from epidemiological (4,8) and animal (5,6,8,12) studies support the concept that recreational UV exposure may play an important role in the etiology of human skin cancer. UV-induced signal transduction pathways may be a significant component in the mechanism of UV-induced carcinogenesis (5,6,9,13). However, most recent reports focus on UVC-or UVB-induced signal transduction (13,14) and little is known regarding pathways induced by UVA.
Extracellular signals have been shown to activate mitogenactivated protein kinase (MAPK) cascades including extracellular signal-regulated kinases (ERKs), c-Jun NH 2 -terminal kinases (JNKs), and p38 kinase (p38) (15). Among the first substrates of ERKs are the family of 90-kDa ribosomal S6 kinases (p90 RSK , also known as MAPKAP-K1) (16). The MAP-KAP-K1 family is activated via phosphorylation and shown to be ubiquitous and versatile mediators of signal transduction (17,18). These signaling molecules are the focus of much interest due to their potent ability to be phosphorylated by activation of ERKs (18) and their involvement in regulation of various cellular functions (16,19). Initially, p90 RSK was isolated from Xenopus and identified on the basis of its ability to phosphorylate the 31-kDa protein S6 that is a component of the 40 S ribosomal subunit in vitro (20,21). As a signal-transducing serine/threonine kinase, p90 RSK is an important member of a growing subfamily of MAPKs-activated protein kinases (16,18,22) that contain two distinct kinase catalytic domains in a single polypeptide chain. The two domains are the aminoterminal kinase domain (NTD) and the carboxyl-terminal kinase domain (CTD) (16,22). With regard to primary structure, the NTD of p90 RSK is most closely related to p70 S6 kinase (p70 S6K ) (16,23). It was shown to phosphorylate exogenous substrates for p90 RSK , including the cAMP response elementbinding protein (24), c-Fos (25), and the estrogen receptor (26). These substrates suggest that p90 RSK may play a role in transcriptionalregulation.TheCTDofp90 RSK isrelatedtocamodulindependent protein kinases and is most similar to phosphorylase kinase (16). The NTD may also be activated by phosphorylation of the CTD (16). ERKs were shown to phosphorylate and activate p90 RSK in vivo (27,29). To date, six phosphorylation sites have been identified, three of which are phosphorylated by MAPKs in vitro (28). MAPKs-catalyzed phosphorylation of Ser 364 and Thr 574 is critical for activation of the NTD and CTD, respectively, and the phosphorylation of Ser 381 catalyzed by the CTD is also important for activation of the NTD (28,29). The Ser 381 site is located on a linker region between NTD and CTD (16,30). Recent studies suggest that phosphorylation of Ser 381 creates a docking site for PDK1 (31).
In mammalian cells, p90 RSK is activated in response to a broad range of cellular perturbations (16,32,33), including oncogenic transformation (32), insulin (17,32), growth factors (33), phorbol esters (16,33), growth hormone, changes in cAMP levels, heat shock, ionizing radiation, and T cell receptor activation (16,17,32,33). The p90 RSK signal transduction pathways are involved in cell growth, proliferation, differentiation, and apoptotic death (16). In this study, we demonstrated that p90 RSK is phosphorylated and activated when cells are exposed to UVA irradiation. To examine the potential dependence of UVA-induced phosphorylation of p90 RSK on MAPK pathways, we employed dominant negative mutants of ERK2, JNK1, and p38 kinase, knockout JnK 1 and JnK 2 cells, and an inhibitor of ERKs and a novel inhibitor of JNKs and p38 kinase. From these studies, we conclude that ERKs and JNKs, but not p38 kinase, are involved in UVA-induced p90 RSK activation and phosphorylation at Ser 381 .
UVA Irradiation of Cells-The UVA source used was a Philips TL100w/10R system from Ultraviolet Resources International (Lakewood, OH). It consists of a Magnetek transformer number 799-XLH-TC-P, 120 volts 60 hertz, and six bulbs each six feet long. UVA irradiation filtered through about 6 mm of plate glass, eliminating UVB and UVC light at all wavelengths below 320 nm, was performed on cultured cells in the UVA box with two ventilation fans installed to eliminate thermal stimulation. These adjustments were necessary because the normal UVA lamps also produce a small amount of UVB and UVC.
Phosphorylation of ERKs, JNKs, and p38 -Immunoblot analysis for phosphorylated proteins of ERKs, JNKs, and p38 kinase was carried out using the phospho-specific MAPK antibodies as reported previously (34 -37). STAT3 was used as an internal control to determine equal loading of protein. The antibody-bound protein complexes were detected by Western immunoblotting using a chemiluminescent detection system (ECL, New England BioLabs, Inc.). Some transfer membranes were washed with stripping buffer (7 M guanadine hydrochloride, 50 mM glycine, pH 10.8, 0.05 mM EDTA, 0.1 M KCl, and 20 mM ␤-mercaptoethanol) and reprobed with other primary phospho-specific or nonphospho-specific antibodies.
Analysis of p90 RSK Phosphorylation with Phospho-specific Antibodies-Cells (5 ϫ 10 5 ) were seeded into 100-mm dishes and cultured for 24 to 48 h. After the cells reached 80-90% confluence, the Cl 41, DNM-ERK2, DNM-JNK1, or DNM-p38 cells were starved for 24 to 48 h in MEM containing 0.1% FBS, 2 mM L-glutamine, and 25 g/ml gentamicin. After treatment with UVA or kinase inhibitors as indicated (prior to UVA irradiation), the cells were washed once with ice-cold phosphate-buffered saline and lysed in 200 l of SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% (w/v) SDS, 10% (v/v) glycerol, 50 mM dithiothreitol, and 0.1% bromphenol blue). The lysed samples were scraped into 1.5-ml tubes and sonicated for 5 to 10 s. Samples containing an equal amount of protein (Bio-Rad protein assay, Bio-Rad) were loaded into each lane of an 8% SDS-polyacrylamide gel for electrophoresis and subsequently transferred onto Immobilon-p transfer membrane (Millipore, Danbury, CT). The phosphorylated p90 RSK protein was selectively detected by Western immunoblotting using a chemiluminescent detection system and a phospho-specific antibody against phosphorylation of p90 RSK at Ser 381 or Thr 360 /Ser 364 (28,29). However, p90 RSK phosphorylation shows the total levels of the homologous phosphorylated sites of RSK1, RSK2, and RSK3.
Assay for p90 RSK Activity-p90 RSK activity was measured by an immune complex kinase assay using an S6 peptide AKRRRLSSLRA as a substrate according to the procedure recommended in the S6 kinase assay kit (Upstate Biotechnology, Inc.) (38,39). Briefly, cell lysates were prepared from JB6 Cl 41 cells or JB6 Cl 41 cells with DNM-JNK1, DNM-p38, or DNM-ERK2 grown in 100-mm dishes. After starving by replacing medium with 0.1% FBS/MEM, the cells were irradiated with UVA (160 kJ/m 2 ). The cells were harvested at the times indicated and lysed in 300 l of buffer A (20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% (v/v) Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerol phosphate, 1 mM Na 3 VO 4 , 1 g/ml leupeptin, 10 g/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride). The cell lysates were centrifuged at 17,000 ϫ g for 5 min at 4°C. The supernatant fractions containing equal amounts of protein were incubated with anti-p90 RSK1 (MAPKAP-K1a) antibody at 4°C overnight, and then for an additional 4 h with protein-A/G plus Sepharose (Santa Cruz Biotechnology, Inc.). After washing four times with phosphate-buffered saline, the immunoprecipitates were incubated at 30°C for 10 min in a mixture of the following: 20 l of assay dilution buffer (ADB: 20 mM MOPS, pH 7.2, 25 mM ␤-glycerol phosphate, 5 mM EGTA, 1 mM Na 3 VO 4 , and 1 mM dithiothreitol), 10 l of substrate mixture (S6 peptide in ADB), 10 l of inhibitor mixture (20 M PKC inhibitor peptide, 2 M protein kinase A inhibitor peptide, and 20 M compound R24571 in ADB), and 10 l of [␥-32 P]ATP (1 Ci/l; Amersham Pharmacia Biotech, Inc.). To stop the reaction, samples were spotted onto a numbered P81 paper square and washed three times (5 min each) with 0.75% phosphoric acid and once (3 min) with acetone. Each sample paper was transferred into a scintillation vial and counted in a ␤-scintillation counter. At the same time, immunoprecipitates with normal non-immune serum instead of p90 RSK antibody were used as background controls. After subtraction of background from each sample, the UVAstimulated p90 RSK activity was normalized to unstimulated controls and is shown as fold change.
Co-immunoprecipitation of Proteins-JB6 Cl 41 cell lysates were prepared as described above. Following exposure to UVA (160 kJ/m 2 ), cells were harvested at 15 or 30 min and lysed for 5 min in buffer A. Supernatant fractions were incubated at 4°C overnight with normal rabbit serum as a non-immune IgG control or with antibodies against ERKs, JNKs, p38 kinase, or p90 RSK1 (MAPKAP-K1a) and for an additional 4 h with protein-A/G plus Sepharose (4°C). Samples were then washed four times with phosphate-buffered saline and the final pellet resuspended in 3 ϫ SDS sample buffer. The immunoprecipitated proteins were analyzed by using Western immunoblotting as recommended by New England BioLabs, Inc. (40,41). Immunoprecipitates of ERKs, JNKs, or p38 kinase were incubated with anti-phospho-p90 RSK (Ser 381 ) as the primary antibody, whereas immunoprecipitates of p90 RSK1 (MAPKAP-K1a) were incubated with phospho-specific ERKs, JNKs, or p38 kinase as the primary antibody.
Preparation and Analysis JnKϩ/ϩ of JnK 1 Ϫ/Ϫ, and JnK 2 Ϫ/Ϫ Primary Embryo Fibroblasts-Embryo fibroblasts from normal JnK 1 Ϫ/Ϫ and JNK 2 Ϫ/Ϫ knockout mice were isolated and prepared according to the procedure of Loo and Cotman (42). Cells were established in culture in DMEM supplemented with 10% FBS, 2 mM Lglutamine, 100 units/ml penicillin, and 100 g/ml streptomycin in a humidified atmosphere of 5% CO 2 at 37°C. For analysis of protein phosphorylation and kinase activity, the cells were starved by replacing growth medium with serum-free DMEM for 24 h at which time they were exposed to UVA. The cells were lysed with SDS sample buffer and protein concentration in the cell lysates was determined (Bio-Rad assay). Equal amounts of protein were loaded onto an 8% SDS-polyacrylamide gel and phosphorylated and nonphosphorylated proteins were determined by Western blotting analysis. Kinase activity of p90 RSK in these cells was performed as described above.
In Vitro Assay for p90 RSK Phosphorylation and Activation (29,43)-JB6 Cl 41 cell lysates were cultured in 100-mm dishes and starved for 24 h and co-immunoprecipitation experiments with p90 RSK1 (MAPKAP-K1a) were performed as described above. Samples containing immunoprecipitated p90 RSK1 (MAPKAP-K1a) were incubated at 30°C for 60 min with activated ERK2, JNK2, or p38 kinase (10 ng/ml) (Upstate Biotechnology, Inc.) in kinase buffer (50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM EGTA, 1 mM dithiothreitol, 5 mM ATP, and 0.01% Brij 35). At the same time, kinase activity of ERK2, JNK2, or p38 kinase was tested by incubating with kinase substrates, Elk1, or ATF2 fusion proteins (1 mg/ml) (New England BioLabs, Inc.). Reactions were stopped by adding SDS sample buffer. The phosphorylation of kinase protein substrates and immunoprecipitated p90 RSK protein was analyzed by using SDS-polyacrylamide gel electrophoresis, Western blotting, and a chemiluminescent detection system (Amersham Pharmacia Biotech, Inc.). Total lysates taken directly from Cl 41 cells that were irradiated with UVA (160 kJ/m 2 ) were used as an internal control. To further analyze whether p90 RSK is activated by MAPKs in vitro, samples containing immunoprecipitated p90 RSK were incubated at 30°C for 30 min with S6 peptide plus active ERK1, ERK2, JNK1, JNK2, or p38 kinase (10 ng/ml, Upstate Biotechnology, Inc.) and p90 RSK kinase activity was determined as described above. At the same time, incubations of S6 peptide with MAPKs were used as internal controls.
Data Analysis-Some data were analyzed using the Image-QuaNT TM Microsoft System (Molecular Dynamics, Sunnyvale, CA). This system calculates the intensity of bands in Western blots.
Statistical Analysis-Significant differences in p90 RSK S6 kinase activity were determined by using Student's t test.

RESULTS
Phosphorylation of p90 RSK at Ser 381 Is Induced by UVA, TPA, and EGF-TPA is a known tumor promoter and acts during tumor promotion and progression (44,45). EGF is a well described growth factor having tumor promoting action (45). TPA and EGF were reported to activate p90 RSK via phosphorylation (16,17,39,40) and were used here as positive controls for comparison with UVA for stimulation of p90 RSK phosphorylation. Our results show that like TPA and EGF, UVA induced phosphorylation of p90 RSK (Fig. 1A). In contrast to the non-irradiated control, UVA at a higher dose (160 kJ/m 2 ) induced a 4.7-fold increase in phosphorylation of p90 RSK , which was 2.8 times that of either TPA or EGF (Fig. 1B).
UVA-induced Ser 381 Phosphorylation of p90 RSK Is Dose-and Time-dependent-As shown in Figs. 1 and 2A, UVA-induced phosphorylation of p90 RSK was dose-and time-dependent.
Phosphorylation of p90 RSK was observed 15 min following a low dose (80 kJ/m 2 ) of UVA exposure and the phosphorylation decreased by 30 min and remained low (Fig. 2, A and B). On the other hand, a higher dose (160 kJ/m 2 ) of UVA resulted in a peak phosphorylation of p90 RSK at 30 min and decreased thereafter (Fig. 2, A and B).
Activation of p90 RSK by UVA Is also Time-dependent-To study whether phosphorylation of p90 RSK reflects p90 RSK S6 kinase activity, we investigated the effect of UVA on p90 RSK S6 kinase activity specifically. As shown in Fig. 2C, p90 RSK S6 kinase was activated by UVA (160 kJ/m 2 ) in a time-dependent manner. The activity peaked at 30 min following UVA and was about 2 times higher than the unstimulated control value. These data indicate that phosphorylation ( Fig. 2A) of p90 RSK correlated well with p90 RSK S6 kinase (Fig. 2C).
PD98059 Inhibits UVA-induced Ser 381 Phosphorylation of p90 RSK Coinciding with Inhibition of ERKs Phosphorylation-PD98059, an inhibitor for MAP kinase kinase (MEK1) (40,46), was used in experiments to study the role of MEK1/ERKs in the UVA-induced Ser 381 phosphorylation of p90 RSK . PD98059 markedly inhibited both UVA-induced phosphorylation of ERKS and p90 RSK at Ser 381 (Fig. 3, A and B). At the same time, PD98059 significantly blocked p90 RSK S6 kinase activation by UVA (Fig. 3E). In contrast, PD98059 was less effective in inhibiting UVA-induced phosphorylation of JNKs and p38 kinase (Fig. 3, C and D). Additionally, p90 RSK S6 kinase activation was also blocked by LY294002 (Fig. 3E), a PI 3-kinase inhibitor, which was used here as a positive control. These data suggest that Ser 381 phosphorylation and activation of p90 RSK may be dependent on ERKs.
DNM-ERK2 Blocks UVA-induced Phosphorylation of Both p90 RSK at Ser 381 and ERKs-The concept that PD98059 is a MEK1-specific inhibitor was challenged by the report of Kamakura et al. (47), in which PD98059 was shown to inhibit ERK5 activation in response to EGF, oxidant, and osmotic stresses. Therefore, to further confirm the role of ERKs in UVA-induced phosphorylation of p90 RSK at Ser 381 , a dominant negative ERK2 mutant (37) was used in this experiment. UVAinduced phosphorylation of p90 RSK at Ser 381 (Fig. 4, A and D) and ERKs (Fig. 4, B and D) was inhibited and p90 RSK S6 kinase activity was also significantly blocked in DNM-ERKs cells (Fig.  4E). However, phosphorylation of JNKs was only weakly affected and no effect on p38 phosphorylation was observed (Fig.  4C). Additionally, the basal expression level of p90 RSK in DNM-ERK2 cells was slightly decreased as compared with control Cl 41 cells (Fig. 4, A and B). These data confirmed that ERKs play an important role in the activation of p90 RSK via phosphorylation at Ser 381 and in the expression of the basal level of p90 RSK .

DNM-p38 Kinase Does Not Block Phosphorylation of p90 RSK
at Ser 381 -The JB6 cell line expressing DNM-p38 established in our laboratory (36) was used to study the role of p38 kinase in the UVA-induced phosphorylation of p90 RSK Ser 381 . We found that DNM-p38 did not inhibit the phosphorylation of p90 RSK Ser 381 or ERKs (Fig. 5, A and B), but inhibited phosphorylation of p38 kinase induced by UVA (Fig. 5B). DNM-p38 also had no significant inhibitory effect on p90 RSK kinase activity when the cells were stimulated by UVA (Fig. 5D). These data suggested that the Ser 381 site in p90 RSK is phosphorylated through the ERKs pathway, but not through the p38 kinase pathway in UVA-irradiated JB6 cells.
Basal and Phosphorylated p90 RSK Is Blocked in Knockout JNK 1 Ϫ/Ϫ and JNK 2 Ϫ/Ϫ Cells-To study whether the JNK pathway plays a role in the activation and basal expression of p90 RSK , we prepared primary embryo fibroblasts from wild-type JnK (JnKϩ/ϩ) and knockout JnK (JnK 1 Ϫ/Ϫ or JnK 2 Ϫ/Ϫ) mice. Our results showed that basal and phosphorylated levels of p90 RSK at Ser 381 and Thr 360 /Ser 364 (Fig. 6A) and JNKs (Fig. 6B), as well as p90 RSK S6 kinase activity (Fig.  6D), were lower in JnK 1 Ϫ/Ϫ and JnK 2 Ϫ/Ϫ cells than those in control JnKϩ/ϩ cells. However, basal and phosphorylated ERKs (Fig. 6C) were not significantly changed in JnK 1 Ϫ/Ϫ and JnK 2 Ϫ/Ϫ cells. These data suggested that the JNK pathway is involved in basal expression of p90 RSK , probably having an indirect effect on UVA-induced phosphorylation and activation of p90 RSK . Thus, whether UVA-stimulated activation of p90 RSK Ser 381 requires the JNK pathway was further tested with DNM-JNK1 and PD169316.
DNM-JNK1 Inhibits UVA-induced Phosphorylation of p90 RSK and JNKs-To further confirm that UVA-induced phosphorylation of p90 RSK occurs through the JNK pathway in vivo, a dominant negative mutant of JNK 1 (34) was used in this experiment. Our results showed that UVA-induced phosphorylation of p90 RSK at Ser 381 (Fig. 7, A and D) and JNKs (Fig.  7B), as well as p90 RSK S6 kinase activity (Fig. 7E), were inhib-ited in DNM-JNK1 cells, but inhibition of ERKs phosphorylation was not observed (Fig. 7C). These data indicated that the JNK pathway may play a role in UVA-induced phosphorylation and activation of p90 RSK .
PD169316 but Not SB202190 Blocks UVA-induced Ser 381 Phosphorylation and Activation of p90 RSK -The findings above (Fig. 5) suggest that phosphorylation and activation of p90 RSK appear to occur through a p38 kinase independent pathway. This idea was further supported by evidence showing that, SB202190, a selective inhibitor of p38 kinase (48), only inhibited phosphorylation of p38 kinase, but not phosphorylation of JNKs, ERKs (Fig. 8B), or p90 RSK at Ser 381 (Fig. 8A). Another compound, PD169316, is confirmed to be a novel inhibitor of JNKs and p38 kinase (49,50) and it completely blocked UVAinduced phosphorylation of JNKs, p38 kinase (Fig. 8B), and p90 RSK Ser 381 (Fig. 8A), but not that of ERKs (Fig. 8B). Our data also showed that p90 RSK S6 kinase activity was blocked markedly by PD169316, but not by SB202190 (Fig. 8C). These data further indicate that UVA-induced phosphorylation and activation of p90 RSK are mediated through the JNKs, but not the p38 kinase pathway.
p90 RSK Co-immunoprecipitates with ERKs or JNKs, but Not p38 Kinase-The non-phosphorylated-p90 RSK protein-antibody complex was co-immunoprecipitated strongly with anti-phospho-ERKs and weakly with anti-phospho-JNKs, but not at all with anti-phospho-p38 kinase as determined by Western immunoblotting analysis (Fig. 9A). Inversely, a UVA-induced phosphorylated p90 RSK was co-immunoprecipitated weakly with the immunoprecipitates containing nonphospho-ERKs (Fig. 9B) and -JNKs (Fig. 9C) antibodies, but not with those containing nonphospho-p38 kinase antibody (Fig. 9D). At the same time, immunoprecipitates with normal non-immune IgG The phosphorylated p90 RSK , ERKs, JNKs, and p38 kinase proteins, as well as p90 RSK S6 kinase activity, were measured as described under "Experimental Procedures." The sample membrane was stripped and reprobed with different antibodies. Nonphosphorylated STAT3 was used as an internal control. This is one of three independent similar experiments. The figure shows that DNM-ERK2 blocks UVA-induced changes in basal and phosphorylated p90 RSK protein (A) and ERKs phosphorylation (B). C, shows DNM-ERK2 weakly inhibits phosphorylation of JNKs and has no effect on p38 phosphorylation. D, shows that inhibition of p90 RSK by DNM-ERK2 parallels its inhibition of ERKs. Each value is the relative ratio of UVA-treated sample to the unstimulated Cl 41 control (value of 1) determined by analysis of Western blots using the Image-QuaNT TM Microsoft System and represents the mean from three independent experiments (D). E, data are presented as the mean and standard deviation from three independent experiments performed in duplicate and results show that DNM-ERK2 significantly blocks p90 RSK activation by UVA (160 kJ/m 2 ) compared with corresponding control cells (**, p Ͻ 0.01).  4. The p90 RSK and p38 proteins and their phosphorylated proteins, as well as phosphorylated ERKs and p90 RSK S6 kinase activity were analyzed as described under "Experimental Procedures." Nonphosphorylated STAT3 was used as an internal control. The sample membrane was stripped and reprobed with different antibodies. This is one of three similar experiments. A, shows that DNM-p38 did not block UVA-induced phosphorylation of p90 RSK . B, shows that DNM-p38 blocked UVA-induced phosphorylation of p38 kinase, but not ERKs (C). D, data are presented as the mean and standard deviation from three independent experiments performed in duplicate and show that DNM-p38 does not significantly inhibit p90 RSK activation by UVA at 160 kJ/m 2 compared with corresponding control cells (p Ͼ 0.10). serum, a background control, did not react with the above mentioned antibodies (Fig. 9, A-D). These data indicated that a possible interaction occurs between p90 RSK and ERKs or JNKs, but not p38 kinase. Whereas the ERK-docking site in p90 RSK has been identified by Smith et al. (51), the JNK-docking site on p90 RSK remains to be determined.
Phosphorylation of p90 RSK Ser 381 by Activated ERK2 and JNK2, but Not p38 Kinase in Vitro-To further study the role of MAP kinases in the phosphorylation of p90 RSK , we performed kinase assays by incubating p90 RSK protein with activated ERK2, JNK2, or p38 kinases in vitro. As shown in Fig.  10, Elk1, a substrate was for ERK2, phosphorylated by ERK2 (Fig. 10A) and ATF2, a substrate of JNK and p38 kinase, was phosphorylated by JNK2 and p38 kinase (Fig. 10B). The p90 RSK Ser 381 was phosphorylated by activated ERK2 (Fig.  10C) and JNK2 (Fig. 10, C and D), but not by p38 kinase (Fig.  10D). Furthermore, immunoprecipitated p90 RSK was activated in vitro by active ERK1, ERK2, JNK1, and JNK2, but not p38 kinase (Fig. 10E). At the same time, the S6 peptide was not phosphorylated by MAPKs (Fig. 10F). These data further suggested that UVA-induced phosphorylation of p90 RSK Ser 381 occurs through activation of ERKs and JNKs, but not p38 kinase. DISCUSSION Two major signaling pathways regulating cell function that are activated by various stimuli (16, 52, 53, 54) include: 1) the phosphatidylinositol 3-kinase (PI 3-kinase) and protein kinase B/Akt pathway which lead to cell survival (55) and 2) the Ras-ERK pathway, which is important in cell division and differentiation (56). The p90 RSK is known to be a downstream kinase of the Raf-MEK-ERK protein kinase cascade and it contains two kinase catalytic domains, the NTD and CTD (16,22). The CTD of p90 RSK kinase is activated by ERKs, which leads to activation of NTD kinase (16). Thus, p90 RSK represents a continuation of the ERKs cascade with two additional protein kinase activities (16). However, purified p90 RSK that had been deactivated by treatment with phosphatase in vitro could only be activated partially by incubation with active ERKs (18,28,57). This suggests that besides ERKs-dependent pathways, p90 RSK may also be activated by ERKs-independent pathways.
Recently, full activation of the NTD kinase of p90 RSK was shown to require cooperation of three kinases: ERKs (58,59), the CTD kinase of p90 RSK (16), and PDK1 (3-phosphoinositidedependent protein kinase-1, a newly identified downstream kinase of PI 3-kinase). Ser 221 in the activation loop of the NTD of p90 RSK is known to be phosphorylated by PDK1 and this phosphorylation is proven to be essential for the activation of all p90 RSK isoforms (29, 60 -62). Furthermore, Ser 221 phosphorylation leading to activation of the NTD kinase was shown to require Ser 381 phosphorylation in the linker region of p90 RSK  Fig. 4. The phosphorylated p90 RSK , JNKs, and ERKs proteins, as well as p90 RSK S6 kinase activity, were determined as described under "Experimental Procedures." The sample membrane was stripped and reprobed with different antibodies. This is one of three independent similar experiments. A, shows that DNM-JNK1 inhibited UVA-induced dose-dependent phosphorylation of p90 RSK Ser 381 (A) and JNKs (B), but not ERKs (C), 15 min after UVA irradiation. D, shows that DNM-JNK1 inhibited time-dependent phosphorylation of p90 RSK Ser 381 following UVA irradiation at 160 kJ/m 2 . E, data are presented as the mean and standard deviation of six assay samples from three independent experiments and show that DNM-JNK1 significantly suppressed UVA (160 kJ/m 2 )-stimulated p90 RSK S6 kinase activity compared with corresponding control cells (*, p Ͻ 0.05). (28,29). This notion is further supported by the results of Frödin et al. (31) who reported that Ser 381 phosphorylation may create a docking site that recruits and activates PDK1 and induce a conformational change of p90 RSK that facilitates PDK1-mediated Ser 221 phosphorylation contributing to p90 RSK activation (31). These previous studies suggest that Ser 381 phosphorylation is an important rate-limiting step of p90 RSK activation. In our study, Ser 381 phosphorylation is shown to be induced by UVA irradiation and correlates with UVA-stimulated p90 RSK S6 kinase activity. However, little is known regarding whether ERK-independent pathways are involved in Ser 381 phosphorylation of p90 RSK . Like Ser 221 , Ser 381 phosphorylation may also be regulated through the PI 3-kinase pathway, inasmuch as UVA-stimulated p90 RSK activity and Ser 381 phosphorylation are blocked by a PI 3-kinase inhibitor LY294002 ( Fig. 3E and data not shown). Moxham et al. (43) indicated that JNKs may be involved in activation and phosphorylation of RSK3, a p90 RSK isoform. We, therefore, examined whether the ERKs and JNKs pathways are involved in UVA-stimulated Ser 381 phosphorylation and activation of p90 RSK .
ERKs are known to be upstream kinases of p90 RSK (15,16,18,28,51,57). But whether Ser 381 phosphorylation occurs via activation of ERKs is not clear. Data from our present study showed that ERKs phosphorylation (Fig. 5) and activation (data not shown) were induced by UVA irradiation and involved in UVA-stimulated Ser 381 phosphorylation and activa-tion of p90 RSK . Consistent with results of the studies showing that p90 RSK is phosphorylated by ERKs in response to a variety of stimuli (16,18,28,63), we provide evidence that UVAinduced phosphorylation of p90 RSK Ser 381 , as well as its kinase activity, is completely blocked by MEK1 inhibitor, PD98059, and a DNM-ERK2, and the blockage was paralleled with inhibition of ERKs by PD98059 and DNM-ERK2. In addition, PD98059 was less effective in inhibiting UVA-induced phosphorylation of JNKs and p38 kinase, which appears to agree with the suggestion of Smith et al. (51) and Cheng and Feldman (64) that JNKs and p38 kinase are implicated in p90 RSK regulation. However, DNM-ERK2 also suppressed phosphorylation of JNKs, but not p38 kinase (Fig. 4C), suggesting that JNKs, like ERKs, but not p38 kinase, may be involved in activation and phosphorylation of p90 RSK in vivo.
Recently, p38 kinase was identified as an upstream kinase of p90 RSK -related kinases including PLPK (65), RSK-B (66), and MSK1 (40). Another p90 RSK family member, MAPKAP kinase-2, was also shown to be an in vivo substrate of p38 kinase and to be mediated by the mammalian target of rapamycin pathway (68). However, our data showed that p38 kinase may not be involved in UVA-stimulated p90 RSK activity and Ser 381 phosphorylation. We provide evidence that a selective p38 kinase inhibitor, SB202190, and DNM-p38 had no effect on p90 RSK Ser 381 phosphorylation and its kinase activity induced by UVA. Although the studies of Lian et al. (46) and Horstmann et al. (67) suggested that the p38 kinase inhibitor-sensitive pathway was involved in activation of S6 kinases, our data suggest that UVA-induced Ser 381 phosphorylation may not be dependent on p38 kinase. The discrepancy may be related to cell type and the kind of stimuli.
Recently, RSK3 was reported to be activated by JNKs both in FIG. 8. PD169316 but not SB202190 completely blocks UVAinduced phosphorylation and activation of p90 RSK . JB6 Cl 41 cells were treated as described in the legend to Fig. 3. The cell samples were harvested 30 min after UVA irradiation. Phosphorylated p90 RSK and its S6 kinase activity, as well as phosphorylated JNKs, p38 kinase, and ERKs, were analyzed as described under "Experimental Procedures." The sample membrane was stripped and reprobed with different antibodies. This is one of three independent similar experiments. The figure shows that PD169316, a novel inhibitor of JNKs and p38 kinase (49,50), completely blocks UVA-induced phosphorylation of p90 RSK Ser 381 (A), JNKs and p38 kinase, but not ERKs (B). SB202190 at 0.5 M only inhibited phosphorylation of p38 kinase, but did not inhibit phosphorylation of p90 RSK Ser 381 (A), JNKs or ERKs (B). C, data are presented as the mean and standard deviation of six assay samples from three independent experiments and the results show that UVAstimulated p90 RSK activation was significantly attenuated by pretreatment with PD169316 (**, p Ͻ 0.01), but not with SB202190 (p Ͼ 0.10) prior to UVA irradiation compared with corresponding treatment with UVA only.
FIG. 9. p90 RSK co-immunoprecipitates with ERKs or JNKs, but not p38 kinase. JB6 Cl 41 cells were cultured for 24 h in 100-mm dishes and then starved for 48 h. Fifteen or 30 min following UVA irradiation (160 kJ/m 2 ), the cells were harvested and lysed in 300 l of immunoprecipitation buffer A. Nonirradiated cells were used as a negative control. Co-immunoprecipitation and subsequent Western immunoblotting analysis were performed as described under "Experimental Procedures." The upper three panels (A) show that nonphosphorylated p90 RSK proteins co-immunoprecipitate with phosphorylated ERKs and JNKs, but not p38 kinases. The lower three panels show that nonphosphorylated ERKs (B) and JNKs (C), but not p38 kinase (D) proteins co-immunoprecipitate weakly with phosphorylated p90 RSK . Additionally, non-immune IgG controls (IP: serum) were negative. P/C indicates total cell lysates as an internal positive control. IP, immunoprecipitation; WB, Western blotting.
vivo and in vitro (43). Here, we also found that activation of JNKs may be involved in UVA-stimulated p90 RSK Ser 381 phosphorylation and kinase activity. We provide evidence that the basal expression and possibly UVA-stimulated phosphorylation of p90 RSK , as well as its kinase activity, were abrogated in JnK 1 Ϫ/Ϫ and JnK 2 Ϫ/Ϫ cells compared with wild-type JnKϩ/ϩ cells. Furthermore, a DNM-JNK1 almost completely blocked UVA-induced Ser381 phosphorylation and activation of p90 RSK and the same blockage also occurred following pretreatment with PD169316, a novel inhibitor of JNKs and p38 kinase (49,50), whereas SB202190 had no effect. No change in ERKs phosphorylation was observed in these experiments. Importantly, the observed effects of PD169316 and SB202190 were not due to their absorption by UV (data not shown). These data indicated that Ser 381 phosphorylation and activation of p90 RSK may occur via activation of the JNKs pathway in vivo.
Moreover, ERKs, but not JNKs or p38 kinase, is shown to interact with the COOH-terminal tails of three p90 RSK isoenzymes (RSK1, RSK2, and RSK3) (51), indicating that a docking site for ERKs, but not for JNKs, is located in the COOHterminal tail of p90 RSK . However, we observed that JNKs, like ERKs, but not p38 kinase, co-immunoprecipitated with p90 RSK following UVA exposure, suggesting that a binding site of JNKs is most likely located with another region of p90 RSK . This notion was supported further by the evidence from our in vitro studies showing that Ser 381 was phosphorylated by active JNK2 and ERK2, but not p38 kinase. Furthermore, S6 kinase activity in the immunoprecipitated p90 RSK was activated by JNKs and ERKs, but not p38 kinase in vitro (Fig. 10E), consistent with the suggestions of Moxham et al. (43) that JNKs are required for p90 RSK activation. Taken together, our results strongly support the hypothesis that activation and phospho-FIG. 10. Activated ERK2 and JNK2, but not p38 kinase phosphorylate and activate p90 RSK in vitro. Cell lysates were prepared from Cl 41 cells cultured in 100-mm dishes and starved for 24 h. Immunoprecipitation with anti-p90 RSK1 antibody, in vitro assay for p90 RSK phosphorylation and subsequent Western immunoblotting analysis, as well as, in vitro assay for activation of immunoprecipitated p90 RSK1 by MAPKs, were performed as described under "Experimental Procedures." Activity of ERKs, JNKs, and p38 kinase was tested by incubation with their substrates, fusion protein Elk1 (for ERK2) or ATF2 (for JNK2 and p38 kinase). Total lysates from SDS lysis of Cl 41 cells were used as internal controls in the in vitro p90 RSK phosphorylation assays. Incubation of immunoprecipitated p90 RSK with no MAPKs was used as a negative control in the in vitro p90 RSK activation assays. At the same time, incubations of S6 peptide with MAPKs, but without immunoprecipitated p90 RSK , were performed as internal control experiments. The figure shows that ERK2 phosphorylates Elk1 (A), JNK and p38 kinase both phosphorylate ATF2 (B), and ERK2 and JNK2 (C), but not p38 kinase (D), phosphorylate immunoprecipitated p90 RSK proteins at Ser 381 . The lower panel (E) shows that the S6 kinase activity found in immunoprecipitates of p90 RSK that were incubated with active ERK1, ERK2, JNK1, or JNK2 in vitro was significantly greater (*, p Ͻ 0.05; **, p Ͻ 0.01) than the S6 kinase activity of p90 RSK in the absence of MAPKs, but the p90 RSK activity induced by incubating with p38 kinase is not different from that by incubating with no p38 kinase (p Ͼ 0.10). F, shows that in the absence of p90 RSK , incubation of the S6 peptide with each of the MAPKs in vitro did not induce phosphorylation that was significantly different from the control level (p Ͼ 0.10). Data are presented as the mean and standard deviation of five assay samples from three independent experiments (E and F). rylation of p90 RSK Ser381 by UVA is mediated ERKs and JNKs and not by p38 kinase.