Ataxia telangiectasia mutated proteins, MAPKs, and RSK2 are involved in the phosphorylation of STAT3.

Phosphorylation at Ser(727) is known to be required for complete activation of STAT3 by diverse stimuli including UV irradiation, but the kinase(s) responsible for phosphorylating STAT3 (Ser(727)) is still not well discerned. In the present study, we observed that activation of ATM is required for a UVA-stimulated increase in Ser(727) phosphorylation of STAT3 as well as in activation and phosphorylation of p90 ribosomal protein S6 kinases (RSKs). Moreover, UVA-stimulated activation of upstream kinases, such as c-Jun N-terminal kinases (JNKs) and ERKs, involved in mediating phosphorylation of RSKs and STAT3 was defective or delayed in ATM-deficient cells. Furthermore, we provide evidence that RSK2-deficient cells were defective for UV-induced Ser(727) phosphorylation of STAT3, and the defect was restored after ectopic expression of transfected full-length RSK2. In vitro experiments showed that active RSK2 and JNK1 induce the phosphorylation of STAT3 precipitates from immunoprecipitation but not from glutathione S-transferase (GST) pull-down. Interestingly, the GST fusion STAT3 proteins mixed together with STAT3 immunoprecipitates can be phosphorylated by JNK. However, the in vitro phosphorylation of STAT3 was reduced by the GST-STAT3 beta protein, a dominant negative form of STAT3. Taken together, our results demonstrate that the STAT3 phosphorylation at Ser(727) is triggered by active RSK2 or JNK1 in the presence of a downstream kinase or a cofactor, and thereby the intracellular phosphorylation process is stimulated through a signaling pathway involving ATM, MAPKs, RSK2, and an as yet unidentified kinase or cofactor. Additionally, RSK2-mediated phosphorylation of STAT3 (Ser(727)) was further determined to be required for basal and UVA-stimulated STAT3 transcriptional activities.

Signal transducer and activator of transcription 3 (STAT3; also called STAT3␣) 1 was shown to be highly homologous to STAT1 and was first identified as an interleukin-6-activated transcription factor that transduces signals to the nucleus and activates expression of many targeted genes (reviewed in Refs. [1][2][3]. STAT3 has since been shown to play several important roles in controlling diverse cellular processes such as growth, development (reviewed in Refs. [3][4][5][6], cell transformation, and oncogenesis (reviewed in Refs. 7 and 8). In these processes, STAT3 signaling is activated by cytokines and growth factors, and the activation is well documented to occur through Tyr 705 phosphorylation by Janus kinase family members or receptor or nonreceptor tyrosine kinases (1,4,6). Further, maximal activation of STAT3 signaling was demonstrated to require modulation of Ser 727 phosphorylation as well (9 -12) (reviewed in Ref. 13). Interestingly, STAT3 signaling was also activated by various stresses, including UV, hyperosmolarity, and osmotic shock (14 -16), and the activation was triggered by Ser 727 phosphorylation but independent of Tyr 705 phosphorylation (17)(18)(19)(20)(21). These findings, therefore, demonstrate that STAT3 is a converging point for numerous intracellular signaling pathways and that phosphorylation at Ser 727 is essential for STAT3 signaling activation. However, the kinases responsible for phosphorylating STAT3 (Ser 727 ) are not yet well elucidated. Hence, identification of the STAT3 kinases will help in understanding the mechanism of STAT3 activation.
Our recent work showed that MAPKs were required in mediating STAT3 (Ser 727 ) phosphorylation via MSK1 in the cellular response to UVA (320 -400 nm), and in vitro kinase assays indicated that MSK1 might be a potential kinase in-volved in phosphorylation of STAT3 (Ser 727 ) (29). However, the UVA-stimulated phosphorylation of STAT3 (Ser 727 ) was not completely abolished in cell lines expressing a kinase-dead mutant of MSK1, suggesting the involvement of other downstream kinases (e.g. RSKs) of MAPKs in the phosphorylation process. Furthermore, RSKs and STAT3, as well as their upstream MAPKs, were reported to be concomitantly activated and to play a role in controlling cell development, growth, and survival (33), suggesting that a certain association existed between MAPKs/RSKs and STAT3. Here, we find that RSKs are also required for phosphorylation of STAT3 (Ser 727 ) following exposure of cells to UVA or UVC (200 -290 nm). Moreover, UVA-stimulated but not UVC-stimulated activation and phosphorylation (Ser 380 ) of RSKs as well as STAT3 (Ser 727 ) phosphorylation are modulated by activation of ATM (ataxia telangiectasia, mutated) kinase. Taken together, our results suggest that RSKs may be potential upstream kinases for modulating phosphorylation of STAT3, and the process may be regulated by ATM. Interestingly, further in vitro kinase assays provide evidence that active RSKs or JNK1 only in the presence of a downstream kinase or cofactor may induce phosphorylation at the C-terminal Ser 727 and/or non-Ser 727 residues of STAT3␣, and the phosphorylation is diminished by STAT3␤, a dominant-negative variant of STAT3␣ (34). In addition, a role of RSKs and ATM in regulation of STAT1 phosphorylation is also investigated.

EXPERIMENTAL PROCEDURES
Amplification of the Wild-type STAT3␣ and STAT3␤ cDNA-The cDNA fragments of the STAT3␣ and STAT3␤ including ORF (2310 and 2166 bp, respectively) were amplified by the PCR with primers 5Ј-CGT GTC GAC TGG CCC AAT GGA ATC AGC T-3Ј (SalI site underlined) and 5Ј-TCG ACG CGT CGT TCT CAG CTC CTC ACA TG-3Ј (MluI site underlined) and cloned into pACT vector (Promega, Madison, WI) via SalI and MluI restriction sites. The PCR-amplified sequences of STAT3␣ and STAT3␤ were verified by comparison of restriction fragment length and DNA sequence analysis. The primer synthesis and DNA sequencing were performed by Sigma.
Construction and Mutagenesis of GST Fusion STAT3 Vectors-The STAT3␣ encoding a fragment digested with SalI/EcoRV from the pACT was cloned into SalI/SmaI sites of the pUC19 vector. To introduce the point mutation in the position of the 705(Y705F) or the 727(S727A) residue, a pair of sense and antisense primers for STAT3␣Y705F (5Ј-CG CTG CCC CAT TCC TGA AGA CCA AG-3Ј and 5Ј-CT TGG TCT TCA GGA ATG GGG CAG CG-3Ј) and another pair of the primers for STAT3␣S727A (5Ј-C CTG CCG ATG GCC CCC CGC-3Ј and 5Ј-GCG GGG GGC CAT CGG CAG G-3Ј) (underlined nucleotides indicate the codon coding a mutated amino acid) were synthesized, respectively (Sigma). Subsequently, the STAT3␣Y705F or STAT3␣S727A nucleotide substitution was accomplished using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and then confirmed by DNA sequence analysis (Sigma). Further, the STAT3␣Y705F and STAT3␣S727A were ligated into the SalI/MluI sites of the pGEX-5X-C modified multicloning sites from the pGEX-5X-1 vector. In addition, wild type STAT3-␣ and STAT3-␤ were directly introduced into the SalI/MluI sites of the pGEX-5X-C vector. The above-mentioned restriction endonucleases and ligases were purchased from New England BioLabs, Inc. (Beverly, MA).
GST Fusion STAT3 Protein Expression and Pull-down-The pGEX-5X-C plasmids encoding the wild-type full-length STAT3␣ or STAT3␤ as well as the STAT3␣ point mutants (STAT3␣Y705A or STAT3␣S727A) as glutathione S-transferase (GST) fusion protein sources were used to transform a DH5␣-competent Escherichia coli strain (Invitrogen). After induction of protein expression with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside (Sigma) for 4 h, the bacteria were resuspended in a lysis buffer containing 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 1% (v/v) Triton X-100, 10 g/ml aprotinin and leupeptin, and 100 M Na 3 VO 4 ) and then were further disrupted by the addition of 0.1 volume of 10% mg/ml lysozyme (Sigma) and subsequent sonication. Following centrifugation at 10,000 ϫ g for 20 min, the supernatant fraction containing the induced proteins was incubated with a 50% slurry of pretreated glutathione-Sepharose 4B (Amersham Biosciences). After washing twice with the above-mentioned lysis buffer and an additional two times with kinase buffer (described below), the beads were subjected to SDSpolyacrylamide gel electrophoresis followed by Western blotting with a GST or STAT3 antibody to determine expression of the GST fusion STAT3 proteins. In subsequent in vitro kinase reactions, eluates of the beads with 20 mM reduced GSH (Roche Molecular Biochemicals) were used as enzymatic substrates. Control experiments were performed with GST-Sepharose beads generated by expression of GST alone, using the empty pGEX-5X-C vector.
Cell Lines and Cell Culture-GM09621 is a human lymphoblast line containing wild-type RKS2 genes and therefore was used as the internal control cells; GM03317 is a lymphoblast line from a patient with Coffin-Lowry syndrome that is linked to RSK2 mutation (RSK2 Ϫ ) (35). The two cell lines were bought from the NIGMS Human Genetic Cell Repository, Coriell Institute for Medical Research (Camden, NJ) and were cultured in RPMI 1640 medium (Invitrogen) supplemented with 15% heat-inactivated FBS (Gemini Bio-Products, Inc., Calabasas, CA), 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin in a 37°C humidified atmosphere of 5% CO 2 . Human ataxia telangiectasia (AT) fibroblast lines stably transfected with an empty mammalian expression vector, pEBS7 (Atm Ϫ ), or a construct, pEBS7-YZ5, containing recombinant full-length ATM (Atm ϩ ) (a generous gift from Dr. Yosef Shiloh) (36,37) was maintained in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% FBS, 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 100 g/ml of hygromycin B (Sigma) for selection. In addition, mouse epidermal tumor promotion-sensitive JB6 Cl 41 cells were grown in 5% FBS with Eagle's minimum essential medium (BioWhittaker, Inc., Walkersville, MD) (29,38,39).
UV Sources and Cell Treatments-To diminish the basal level of protein phosphorylation or protein kinase activity, the experimental cell lines were starved for 36 -48 h in the above mentioned RPMI 1640 containing 0.5% FBS for lymphoblast lines or in serum-free Dulbecco's modified Eagle's medium for AT fibroblast lines prior to treatments with UVA or UVC irradiation. Nonirradiated cell samples were used as negative controls. For the detailed description of UVA or UVC sources, see our previous reports (38,40).
Transfection of RSK2 Ϫ Cells-An RSK2 cell line, GM03317, from a Coffin-Lowry syndrome patient was exponentially grown in the RPMI 1640 medium containing 15% FBS and then was transfected with an empty vector pcDNA3.1(ϩ) (3.0 g) (Invitrogen) or co-transfected with 0.3 g of the empty vector and 2.7 g of a construct expressing a wild-type full-length RSK2 enzyme (a generous gift from Dr. Morten Frodin (41)) according to the modified protocol recommended by Qiagen Inc. (Valencia, CA). Briefly, 2 ϫ 10 6 cells were washed twice with phosphate-buffered saline, resuspended in 1 ml of fresh RPMI 1640 containing 1% phosphate-buffered saline, and then seeded into each well of a six-well plate. Three micrograms of the plasmid DNA were mixed in 96 l of serum/antibiotic-free RPMI 1640 medium, and then 15 ml of SuperFect Reagent was added to the DNA solution. After incubation at 4°C for 10 min, the transfection complexes gently mixed with 400 l of RPMI 1640 containing 1% FBS and antibiotics were dropped into the prewashed cells. After incubation for 3-5 h, some cells were observed to die. Then the transfected cells were grown for 3 days in the normal complete medium containing 15% FBS and subsequently were selected in a 250 g/ml G418 RPMI medium.
STAT3-mediated Luciferase Activity Assays-The above mentioned pGEX-5X-C vectors expressing wild type STAT3␣ and STAT3␤ and point-mutated STAT3 were amplified by PCR with the STAT3 upstream and downstream primers: 5Ј-CCC AAG CTT GGG ATG GCC CAA TGG AAT CAG CTA-3Ј (HindIII site underlined) and 5Ј-TCC CCC CGG GGG ACG TTC TCA GCT CCT CAC ATG GG-3Ј (SmaI site underlined). These PCR products were digested with HindIII/SmaI and then subcloned into the HindIII/EcoRV sites of the pcDNA3.1(ϩ) vector (Invitrogen). Cy1luc (VH3) vector, a STAT-responsive luciferase reporter containing the 180-base pair cytokine response element (CyRE) region of the human VIP gene (12), was truncated out with KpnI/PstI (187 bp). Then the truncated Cylluc reporter was reacted with Klenow and dNTP and ligated with a T4 ligase, and it is designated here as ⌬Cy1luc, an internal control. Within the CyRE, two motifs have been identified to bind the transcriptional complexes made up of STAT dimers and AP-1 (12). Thus, the truncated ⌬Cy1luc still contains a partial STAT binding motif. The above mentioned restriction endonucleases and ligases were purchased from New England BioLabs, Inc. (Beverly, MA). Subsequently, these STAT3 vectors, Cy1luc, or ⌬Cy1luc reporters were or were not transfected or co-transfected into wild-type cells or RSK2-deficient cells according to the procedure as described above. The transfected cells were cultured for 24 h in normal media and then starved for 24 h with serum-free media. The cells were harvested 4 h after irradiation with UVA (80 kJ/m 2 ) and destroyed in 100 l of lysis buffer containing 0.1 M potassium phosphate, pH 7.8, 1% Triton X-100, 1 mM dithiothreitol, 2 mM EDTA. The STAT3-mediated luciferase activity in the supernatant fraction of samples was measured by Luminoskan Ascent (Labsystems Inc. Franklin, MA). Results of the luciferase activity are expressed in relative luminescence units.
Western Blot Analysis of Protein Phosphorylation-Equal numbers of experimental cells (1 ϫ 10 6 to 1.5 ϫ 10 6 ) were cultured for 12-24 h in 100-mm dishes. The cells were starved for 48 h and then irradiated with UVA or UVC. At the indicated times after irradiation, the cells were harvested and washed once with ice-cold phosphate-buffered saline. Then the cell samples were disrupted in 200 l of radioimmune precipitation buffer (1ϫ phosphate-buffered saline, 1% (v/v) Nonidet P-40, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS, and the inhibitors added before use, 10 g/ml phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 100 M Na 3 VO 4 ). The cell lysates were clarified by microcentrifuging at 13,000 rpm, and the supernatant fractions were saved. The samples containing equal amounts of proteins (modified Lowry's method; Sigma) in an equal volume of radioimmune precipitation buffer were diluted with 3ϫ SDS sample buffer (187.5 mM Tris-HCl (pH 6.8), 6% (w/v) SDS, 30% (v/v) glycerol, 150 mM dithiothreitol, and 0.3% (w/v) bromphenol blue). Then samples were subjected to separation by 8% SDS-PAGE followed by Western blot analysis according to the methods reported previously (29, 38 -40). Antibodies against a specifically phosphorylated RSK (Ser 380 ), STAT3 (Ser 727 ), ERKs, JNKs, or p38 kinase, as well as against total STAT3, STAT1, ERKs, JNKs, p38 kinase, GST, or ␤-actin were from Cell Signaling, Inc. (Beverly, MA). Antibodies to a specific phospho-STAT1 (Ser 727 ) and total RSK1 or RSK2 were purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). The total RSK3 antibody (C-20) was from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The levels of total protein were used as internal controls to verify basal level of protein expression or equal amounts of protein loading. In addition, some Western blots were analyzed, and areas were calculated using the ImageQuant Microsoft system.
In Vitro Assay for STAT3 Phosphorylation by Protein Kinases-After starvation for 48 h in Eagle's minimum essential medium and 0.1% FBS, JB6 Cl 41 cells were disrupted and then subjected to immunoprecipitation with a STAT3 antibody according to the procedure reported previously (29,39). The STAT3 proteins from immunoprecipitation or GST pull-down or the mixture of both were incubated at 30°C for 1 h with active RSK1, RSK2, ERK2, or JNK1 (Upstate Biotechnology) in kinase buffer (50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 1 mM EGTA, 1 mM dithiothreitol, and 0.01% (v/v) Brij 35) containing 0.5 mM ATP or 10 Ci of [␥-32 P]ATP. Subsequently, phosphorylation of STAT3 at Ser 727 and total STAT3 proteins were detected by Western blot analysis with specific antibodies (29) or by autoradiography (42).
Assay for RSK1 and RSK2 Activity-After treatments of the experimental cells with UVA or UVC, the lysates were subjected to immunoprecipitation with a RSK1 or RSK2 antibody, followed by RSK1 or RSK2 activity assays that were performed according to the methods recommended by Upstate Biotechnology or as reported previously (39). A peptide, AKRRRLSSLRA, in the S6 kinase assay kit (Upstate Biotechnology) was used as a substrate for the RSK1 immune complexes, whereas another peptide KKRNRTLTK served as a substrate for RSK2 (Upstate Biotechnology). After the reactions, the radioactive phosphate incorporated into the indicated substrate peptide in the samples was quantified with a ␤-scintillation counter. Then the stimulated activity of RSK1 or RSK2 was normalized to the nonirradiated control value after subtraction of background and is expressed as -fold change compared with controls.
MAPK Activity Assay-ERKs, JNKs, and p38 kinase activities were measured by immune complex kinase assays (39). In brief, the lysates from UV-irradiated cells were subjected to immunoprecipitation with an antibody to ERKs, JNKs, or p38 kinase. Then the immunoprecipitates of ERKs, JNKs, or p38 kinase were incubated for 20 min at 30°C, respectively, with Elk1, c-Jun, or ATF2 fusion protein as substrates (Cell Signaling). The radioactive phosphate incorporated into the reactive substrate was quantified as described above.
Statistical Analysis-Significant differences between stimulated activity of RSK1, RSK2, ERKs, JNKs, or p38 kinase and their corresponding controls, as well as STAT3-mediated luciferase activities, were determined using the Student's t test.

RESULTS
ATM Kinase Activation Is Involved in Regulating UVA-stimulated Phosphorylation of STAT3 and STAT1-ATM kinase is known to be a member of the phosphatidylinositol 3-kinaserelated protein kinase family and plays a vital role in controlling cell development, growth, and survival (reviewed in Refs. [43][44][45]. Our recent studies indicated that ATM kinase is activated by irradiation with UVA but not UVC, and the activation is involved in UVA-induced signaling (42). Here, our data showed that UVA-induced Ser 727 phosphorylation of STAT3 as well as STAT1 (Fig. 1A) was differentially reduced in ATMdeficient cells (ATM Ϫ ) compared with control cells ectopically expressing wild-type full-length ATM kinase (ATM ϩ ). On the other hand, UVC-stimulated Ser 727 phosphorylation of STAT3 and STAT1 (Fig. 1B) were not significantly changed in either ATM-deficient or ATM-expressing cells. Importantly, the deficiency of ATM kinase did not influence total levels of STAT3 or STAT1 expression (Fig. 1, A and B). These results suggest that ATM activation may be involved in the regulation of phosphorylation (Ser 727 ) of STAT3 or STAT1 in the cellular response to UVA but not UVC. In addition, the partial reduction in the UVA-stimulated phosphorylation of STAT3 or STAT1 (Fig. 1A) indicates a requirement for activation of other signaling pathways involved in the process.
UVA-induced Activation and Phosphorylation of MAPKs Are Modulated by ATM Activation-Our previous studies have indicated that MAPKs are upstream kinases involved in UVAstimulated signaling activation of STAT3 (29). ATM kinase has been shown to be located upstream of multiple signal transduction pathways (reviewed in Ref. 44). However, whether ATM kinase has an effect on activation and phosphorylation of MAPKs is unclear. Here, our further experiments revealed that UVA-stimulated phosphorylation and activation of JNKs (Fig.  2, A and C) were markedly prevented in ATM-deficient cells Then, the cell lysates were subjected to Western blot analysis to detect phosphorylation of STAT3 or STAT1 at Ser 727 , as well as total levels of STAT3 or STAT1. In addition, the intensity of bands showing phosphorylated levels of STAT3 or STAT1 was quantified. These data are representative of at least three similar independent experiments. compared with ATM-expressing control cells. Moreover, UVAinduced phosphorylation and activity of ERKs (Fig. 2, A and D) and p38 kinase (Fig. 2, A and E) were diminished and/or delayed in ATM-deficient cells. Conversely, ATM deficiency had no inhibitory effect on UVC-stimulated phosphorylation of ERKs, JNKs, or p38 kinase (Fig. 2B), and no change in total levels of ERKs, JNKs, or p38 kinase (Fig. 2, A and B) was observed in the two cell lines. Overall, these data suggest an involvement of ATM kinase in the regulation of the MAPK cascades contributing to activation of STAT3 signaling by UVA. But how the activated ATM kinase mediates phosphorylation and activation of MAPKs remains to be further investigated.
UVA-induced but Not UVC-induced Activation and Phosphorylation of RSKs Are Mediated by ATM Kinase Activity-Our recently reported data indicated that activation of JNKs and ERKs are required in UVA-induced phosphorylation and activation of RSKs (39). Further, the UVA activation of JNKs and ERKs was initiated by ATM kinase as described above. However, whether ATM plays a role in regulating RSKs activation is as yet unknown. To evaluate the role of ATM, we assayed stimulated levels of phosphorylation (Ser 380 ) of RSKs as well as activity of RSK1 or RSK2 in two UV-irradiated AT cell lines. The results showed that UVA-induced phosphorylation of RSKs (Ser 380 ) was markedly prevented (Fig. 3A) in ATM-deficient cells compared with control cells, but the phosphorylation by UVC was not different between the two cell lines (Fig. 3B). In addition, no change of total levels of RSK1 or RSK2 expression (Fig. 3, A and B) was detected in the two cell lines. The phosphorylation of RSKs (residue 380) may indirectly reflect total levels of RSK activity, based on the fact that the specificphospho-RSK (Ser 380 ) antibody is known to cross-react with the homologous sequence of RSK1, RSK2, and RSK3 (39) (see the Cell Signaling site on the World Wide Web at www.cellsignal. com). Furthermore, both RSK1 and RSK2 activity stimulated by UVA (Fig. 3, C and E), but not UVC (Fig. 3, D and E), were significantly (p Ͻ 0.05) diminished in ATM-deficient cells (ATM Ϫ ) compared with ATM-expressing control cells (ATM ϩ ). Thus, these data suggest an involvement of ATM kinase in modulating activation and phosphorylation of RSK1 and RSK2 in the intracellular response to UVA but not UVC.
RSK2 Is Required in UV-stimulated Phosphorylation of STAT3 but not STAT1-As described above, phosphorylation and activation of STAT3, RSKs, and MAPKs were triggered by UVA-activated ATM kinase (Figs. 1-3). The UVA-stimulated MAPKs were shown to be upstream kinases responsible for the intracellular phosphorylation or activation of STAT3 (29) and RSKs (39). However, phosphorylation of the STAT3 immunoprecipitates was induced directly by active JNKs only, and not ERKs or p38 kinase in vitro (29), although MAPKs are postulated to be a potential kinase for catalyzing phosphorylation of STAT3 (Ser 727 ) (9,11). These observations, therefore, suggest that a downstream effector such as RSK may be involved in mediation of ATM-initiated phosphorylation of STAT3. To examine the suggestion, we used two cell lines expressing normal  A and B), as well as total levels of ERKs, JNKs, or p38 kinase (A and B). These data represent at least three similar independent experiments. Furthermore, the lysates from UVA-irradiated cells were subjected to immunoprecipitation and subsequent kinase activity assays for JNKs (C), ERKs (D), or p38 kinase (E). The UVA-stimulated MAPK activity is expressed as a fold change after subtraction of background and normalization to the non-irradiated control cells (value of 1). Each column and bar represents the mean and standard deviation from three independent experiments performed in duplicate (C, D, and E). UVA-induced JNKs (C), ERKs (D) or p38 kinase (E) activity was significantly (*, p Ͻ 0.05) reduced and delayed in ATM-deficient cells compared with corresponding ATM-expressing cells.
human wild-type RSK2 or a mutant RSK2 gene (RSK2 Ϫ ), leading to Coffin-Lowry syndrome (35). Western blot analysis of intracellular protein phosphorylation showed that RSK2 Ϫ cells are defective for Ser 727 phosphorylation of STAT3 stimulated by UVA or UVC (Fig. 4, A and B). On the other hand, STAT1 (Ser 727 ) phosphorylation by UVA or UVC was unaffected or, conversely, increased to a modest level in RSK2 Ϫ cells (Fig. 4,  A and B). Additionally, no change in total levels of STAT3 or STAT1 expression (Fig. 4, A and B) was observed in the two cell lines.
In addition, the RSK2 Ϫ cell line from a Coffin-Lowry syndrome patient was also confirmed here. As expected, these cells had a significantly lower expression level of RSK2 compared with the wild-type cell line from a normal human subject (Fig.  4C). However, expression of RSK1, RSK3, or ␤-actin was not different between the two cell lines (Fig. 4C). Moreover, UVAstimulated phosphorylation of RSK (Ser 380 ) was apparently prevented in RSK2 Ϫ cells (Fig. 4C). These results, together with our previously reported data (39), may indirectly reflect a marked decrease in total levels of RSK activity in the response of the cell to UVA. Indeed, further studies revealed that RSK2 activity stimulated by UVA or UVC was defective in RSK2 Ϫ cells (Fig. 4E), but a modest compensatory increase in RSK1 activity was observed when the RSK2 Ϫ cells were exposed to UVA or UVC irradiation (Fig. 4D). Overall, our results suggest that RSK2 may be required for UVA-and UVC-stimulated phosphorylation of STAT3 (Ser 727 ), whereas the phosphorylation of STAT1 appears to occur through both RSK2-independent and -dependent pathways.

Phosphorylation of STAT3 (Ser 727 ) Is Restored by Ectopic Expression of Full-length RSK2 in RSK2
Ϫ Cells-Here, we further assess a requirement for RSK2 in phosphorylation of STAT3. As expected, a significant ectopic expression of the wild-type RSK2 protein (Fig. 5A) was determined in the RSK2 Ϫ cells transfected with a construct containing a human wildtype full-length RSK2 (RSK2-wt) (41), compared with control RSK2 Ϫ cells transfected with an empty vector. Further, a marked restoration in the UVA-stimulated phosphorylation of STAT3 (Ser 727 ) was determined to correspond to the ectopic expression of RSK2-wt (Fig. 5B). On the other hand, no induction of the STAT3 phosphorylation was observed in the original RSK2 Ϫ cells or the RSK2 Ϫ cells transfected with an empty vector only (Fig. 5B). Additionally, total STAT3 levels were unaffected by the transfection (Fig. 5, A and B). These data further indicate an involvement of RSK2 in UVA-stimulated phosphorylation of STAT3 (Ser 727 ).
Phosphorylation of the STAT3 Immunoprecipitates Is Induced Indirectly by RSKs or JNK1 in Vitro-Our recent report (29) indicated that Ser 727 phosphorylation of STAT3 immunoprecipitates (IP-STAT3) from JB6 Cl 41 cell lysates was induced in vitro by active MSK1 as well as JNKs but not by ERKs or p38 kinase. Further studies showed an involvement of ERKs/p38 kinase-mediated MSK1 in the intracellular phosphorylation of STAT3 (29). Here, we performed further in vitro kinase assays followed by Western blotting with a phosphowith corresponding specific antibodies. In addition, the intensity of bands showing phosphorylated levels of RSKs was quantified. These data are representative of at least four similar independent experiments. Furthermore, the cell lysates were subjected to immunoprecipitation followed by kinase activity assays for RSK1 (C and D) or RSK2 (E). After subtraction of background, the UV-stimulated activity was normalized to the non-irradiated control cells (value of 1) and is expressed as a fold change. Each column and bar represents the mean and standard deviation from three independent experiments performed in duplicate (C, D, and E). UVA-induced RSK1 (C) or RSK2 (E) activity was significantly (*, p Ͻ 0.05) inhibited in ATMcells compared with corresponding ATM ϩ cells.  (A and B), as well as total levels of RSK1 or RSK2 (A and B), in the cell lysates was detected by Western blotting Ser 727 antibody (Fig. 6, A and F) or with radioisotope [␥-32 P]ATP labeling (Fig. 6, B, D, and E). The results showed that phosphorylation (Ser 727 ) of intact STAT3 immunoprecipitates (IP-STAT3) was significantly triggered by active RSK2, similar to the positive control JNK1 (Fig. 6A). On the other hand, a lower level of phosphorylation was initiated by active RSK1 (Fig. 6A), and total levels of STAT3 were unaffected (Fig.  6A). Moreover, autoradiograph analysis revealed a high phosphorylation level of IP-STAT3 proteins by activated JNK1, RSK1, or RSK2 but not ERK2 as a negative control (Fig. 6B). The discrepancies in the degrees of detectable phosphorylation in the above mentioned two different assays appear to be related to phosphorylation at an additional residue, besides Ser 727 , of STAT3.
To further identify a kinase directly responsible for catalyzing phosphorylation of STAT3 (Ser 727 ), we established a kJ/m 2 ) (A, C, D, and E) or UVC (60 J/m 2 ) (B, D, and E). Then, the cells were harvested at the indicated times (A, B, D, and E), or at 30 min (C) and the cell lysates were resolved by 8% SDS-PAGE. Then, the total and phosphorylated (Ser 727 ) levels of STAT3, or STAT1 (A and B), as well as phosphorylated levels of RSKs (Ser 380 ), and total levels of RSK1, RSK2, RSK3 or ␤ -actin (C) were assayed by Western blotting with corresponding specific antibodies. In addition, the intensity of bands showing phosphorylated levels of STAT3 or STAT1 was quantified. These data are representative of at least five independent experiments. In addition, RSK2 Ϫ cells were further characterized by using functional kinase activity assays. The cell lysates harvested at the indicated times (D and E) were subjected to immunoprecipitation with a RSK1 (D) or RSK2 (E) antibody followed by RSK1 or RSK2 activity assays. After subtraction of background, the UV-stimulated activity was normalized to the nonirradiated wild-type control (value of 1) and is expressed as a fold change. Each column and bar represents the mean and S.D. of three independent experiments performed in duplicate (D and E). UV-induced RSK2 activity was significantly (*, p Ͻ 0.05) blocked in RSK2 Ϫ cells compared with corresponding wild-type cells (E).  A and B) were starved and then irradiated with UVA as described for Fig. 4. Subsequently, the cell lysates were subjected to Western blot analysis with a specific antibody against Ser 727 phosphorylation or total STAT3 (B), as well as total RSK2 (A). In addition, a nonspecific band was presented as an internal control (A). These data are representative of at least two independent experiments. GST fusion STAT3 protein expression system to pull down a wild-type full-length STAT3␣ (GST-STAT3␣wt) or a dominantnegative STAT3␣ variant (STAT3␤wt) (34) as well as STAT3␣ point-mutated at the Ser 727 or Tyr 705 residue (named STAT3␣S727A or STAT3␣Y705F, respectively). These constructs were confirmed by sequencing (data not shown). The expressed STAT3 proteins from the GST pull-down were primarily characterized by Western blotting with a STAT3 or GST antibody, and the results were as expected (Fig. 6C). Subsequently, our protein kinase assays using the pull-down precipitate as an enzymatic substrate showed that GST-STAT3␣wt, GST-STAT3␤wt, GST-STAT3␣S727A, GST-STAT3␣Y705F, or GST alone was not phosphorylated directly by active JNK1, RSK1, or RSK2 in vitro (data not shown). However, phosphorylation of GST-STAT3␣wt, but not GST-STAT3␤wt, in a mixture with IP-STAT3 was stimulated by JNK1, but not RSK1 or RSK2 (Fig. 6D). Interestingly, a significant increase in the JNK1-stimulated phosphorylation of GST-STAT3␣S727A or GST-STAT␣Y705F in the presence of IP-STAT3 was observed (Fig. 6D). Conversely, the GST-STAT3␣S727A or GST-STAT3␣Y705F phosphorylation was undetectable in the RSK1 or RSK2 kinase reactions (Fig. 6D). Furthermore, phosphorylation of IP-STAT3 by RSK1, RSK2, or JNK1 was shown as described above (Fig. 6, A and B) but was diminished by the addition of GST-STAT3␤wt (Fig. 6D) and was also to different degrees affected by the addition of GST-STAT3␣S727A or GST-STAT3␣Y705F (Fig. 6D). As expected, Western blotting with a phosphospecific antibody showed that Ser 727 phosphorylation in JNK1-induced GST-STAT3␣wt or GST-STAT3␣Y705F (Fig.  6E) in the presence of IP-STAT3 was detectable, but Ser 727 phosphorylation in GST-STAT3␤ or GST-STAT3␣S727A was not observed (Fig. 6E). These observations, together with previously reported data (29), demonstrate that the antibody specifically recognized Ser 727 phosphorylation of STAT3. The addition of GST fusion STAT3 protein substrates also interfered with Ser 727 phosphorylation of IP-STAT3 by JNK1 (Fig. 6E). Overall, the results of these in vitro kinase assays suggest that JNK1 only in the presence of a cofactor may catalyze phosphorylation at the C-terminal Ser 727 and/or non-Ser 727 residues of STAT3, and the Ser 727 /non-Ser 727 phosphorylation may be also stimulated indirectly by RSK2 as well as RSK1 in cooperation with a RSK downstream kinase or/and a cofactor binding to IP-STAT3 precipitates.
Dependence of STAT3-mediated Luciferase Activity on RSK2-Here, to examine the functional significance of STAT3 serine phosphorylation by related kinases, we used a STATresponsive luciferase Cy1luc reporter containing the CyRE region of the human VIP gene (12). Two motifs within the CyRE of the Cy1luc reporter have been determined to bind transcriptional complexes that are made up of STAT dimers and AP-1 "Experimental Procedures." (A) After incubation of IP-STAT3 with active RSK1, RSK2, or JNK1 at the indicated doses, the reactive products were analyzed by Western blotting with a specific antibody to detect total and Ser 727 phosphorylation levels of STAT3. Additionally, the intensity of the blots representing phosphorylated STAT3 was calculated. (B) The kinase reactions of JNK1, RSK1, RSK2, or ERK2 with IP-STAT3 as a substrate were carried out in a kinase buffer containing ␥-32 pATP. Then, the reactive products were subjected to 8% PAGE resolution followed by autoradiography. (C) The indicated GST-fusion STAT3 proteins were assayed by Western blotting with a specific STAT3 or GST antibody. (D and E) IP-STAT3 proteins were mixed together with each of the indicated GST-fusion STAT3 proteins and the mixtures were used as substrates and then were reacted with RSK1, RSK2, or JNK1 at the indicated doses. Subsequently, the kinase reactive products were determined by radioisotope ␥-32 pATP (D) or a phospho-specific STAT3 (Ser 727 ) antibody (E). In addition, a nonspecific IgG blot was used as an internal control (E). These data are representative of at least three independent experiments.  A, B, D, and E) or a GST pull-down assay (GST-STAT3) (C, D, and E) were prepared as described under (12). On the other hand, ⌬Cy1luc reporter contains a partial motif of STAT binding. Our studies showed that phosphorylation of STAT3 and STAT1, but not other STATs (data not shown), was stimulated by UVA and that the intracellular phosphorylation of STAT3, but not STAT1, was markedly dependent on RSK2 activation by UVA. These findings thus suggest that changes in Cy1luc luciferase activity in RSK2 cells may reflect a specific STAT3-dependent transcriptional activation. Experimental results revealed that basal and UVA-stimulated luciferase activities of STAT3 were significantly induced in wild-type RSK2 cells transfected with the Cy1luc reporter (Fig. 7A). Additionally, a modest level of luciferase activity in wild-type RSK2 cells transfected with the ⌬Cy1luc reporter containing only an incomplete STAT binding motif was determined compared with the original Cy1luc reporter (Fig. 7A). On the other hand, both basal and stimulated luciferase activities were defective (Fig. 7A) in corresponding RSK2-deficient cells transfected with Cy1luc or ⌬Cy1luc. These observations further indicate that RSK2-mediated Ser 727 phosphorylation of STAT3 is required in the functional regulation of STAT3 transcriptional activation.
Furthermore, we performed co-transfection of wild-type RSK2 cells with ⌬Cy1luc or Cy1luc reporters and STAT3␣ or STAT3␤ vectors to further assess STAT3-dependent transcriptional activation. Results showed that overexpression of STAT3␣ in wild-type RSK2 cells partially enhanced UVA-stimulated luciferase activity of Cy1luc or ⌬Cy1luc reporters (Fig.  7B) compared with cells with only the reporters transfected. These stimulated luciferase activities were, however, partially suppressed by expression of STAT3␤, a dominant negative form containing the Tyr 705 but not the Ser 727 site (Fig. 7B). However, basal levels of luciferase activities were not significantly affected by STAT3␤ expression. Overall, these data indicate that Ser 727 phosphorylation of STAT3 is required in the functional regulation of STAT3-mediated transcriptional activation in the UVA response and that the intracellular phosphorylation process may occur through a pathway involving ATM, MAPKs, and/or RSK2.

DISCUSSION
The most information regarding regulation of STAT3 or STAT1 signaling is concentrated heavily on tyrosine phosphorylation by Janus kinases or other tyrosine kinases, leading to the idea that tyrosine phosphorylation is indispensable for the activation process of STAT3 or STAT1 (reviewed in Refs. [1][2][3]. This idea, however, is challenged by recent findings revealing the existence of tyrosine phosphorylation-independent activation mechanisms of STAT3 or STAT1 (17)(18)(19)(20)(21). In fact, serine phosphorylation (Ser 727 ) was shown to be essential for activation of STAT3 (9 -16) or STAT1 (46,47), in particular, following stimulation of cells with diverse stresses, including UVA (29) or UVC (14,46) irradiation. These findings thus demonstrate that multiple signaling pathways may regulate STAT3 or STAT1 signaling activation by Ser 727 phosphorylation. However, the upstream kinases responsible for phosphorylating Ser 727 in STAT3 or STAT1 are not fully identified. MAPKs, including ERKs, JNKs, and p38 kinase, were postulated to phosphorylate STAT3 or STAT1 (Ser 727 ) (9)(10)(11)(12) and were shown to be involved in the induction of Ser 727 phosphorylation (22)(23)(24)(25), but no direct evidence was presented. Recently, we reported that MSK1 is a potential upstream kinase required for the Ser 727 phosphorylation contributing to UVA-induced STAT3 signaling activation (29). However, the UVA-induced Ser 727 phosphorylation of STAT3 was not totally prevented by blocking MSK1 activity by pretreatment with H89, a potential inhibitor of MSK1, or expression of a MSK1 kinase-dead mutant (29), suggesting that the intracellular phosphorylation process may also occur through a MSK1-independent pathway (e.g. RSK).
RSKs are another downstream serine-threonine kinase family of ERKs and PDK1 and are involved in the signaling regulation of some transcription factors to execute a variety of cellular programs, such as embryonic development, cell growth, proliferation, and differentiation (39,48) (reviewed in Ref. 49). Indeed, these cellular processes also require functional regulation of STAT3 (reviewed in Ref. 50) or STAT1 (51) in addition to MAPKs (e.g. ERK2) (52). Furthermore, STAT3 has been shown to be an oncogene (53) and is activated concomitant with MAPKs and RSKs following stimulation of some oncogenes by UVA or UVC (29,33,39) (reviewed in Refs. 7, 8, 49, and 54). Therefore, these findings suggest that MAPK-mediated RSKs may be involved in the regulation of the STAT3 signaling pathway. To explore the relationship, we here used a cell line expressing a mutant RSK2 gene (RSK2 Ϫ ), which leads to Coffin-Lowry syndrome. The RSK2 Ϫ cells are confirmed to be defective for basal RSK2 expression and UVA/UVC-induced RSK2 activity as well as UVA-stimulated phosphorylation (Ser 380 ) of total RSKs. Further, we present evidence showing a significant decrease in UVA/UVC-stimulated Ser 727 phosphorylation of STAT3 in RSK2-deficient cells compared with the The experimental cells were cultured for 24 h in normal complete media and then starved for 24 h in serum-free media. The cells were or were not irradiated with UVA (80 kJ/m 2 ) and harvested 4 h after irradiation and then placed for 1 h in 100 l of lysis buffer. The lysates were subsequently subjected to assays for STAT3-mediated luciferase acitivity as described under "Experimental Procedures." The results of the luciferase activity are expressed in relative luminescence units (RLU). Each column and bar represents the mean and standard deviation of two independent experiments performed in triplicate. * indicates a significant difference (p Ͻ 0.05) from corresponding controls. wild-type control cells. Importantly, UVA-stimulated phosphorylation of STAT3 (Ser 727 ) was restored after the transfected RSK-deficient cells ectopically expressed wild-type full-length RSK2. Moreover, in vitro phosphorylation of immunoprecipitated STAT3 complexes was induced by active RSK2. On the other hand, RSK2 deficiency did not cause a decrease in the intracellular STAT1 phosphorylation (Ser 727 ) stimulated by UVA/UVC. In addition, basal levels of STAT3 or STAT1 expression were unaffected by deficiency of RSK2. Taken together, these observations indicate that activation of RSK2 may be required for phosphorylation of STAT3 (Ser 727 ) in the intracellular response to UVA/UVC irradiation, whereas STAT1 phosphorylation (Ser 727 ) appears to occur principally through a RSK2-independent pathway (e.g. RSK1 or MSK1).
ATM has been shown to be a protein kinase of the phosphatidylinositol 3-kinase family and to be involved in multiple signal transduction pathways that regulate diverse cellular functions (reviewed in Refs. [43][44][45]. Recently, we reported that ATM kinase was activated by UVA, but not UVC, and the activation was required for UVA activation of p53 and JNK signaling pathways leading to apoptosis (42). In the present study, we provide evidence showing that UVA-stimulated, but not UVCstimulated, activity of RSK1 and RSK2 as well as phosphorylation of RSKs (Ser 380 ), STAT3 (Ser 727 ), and STAT1 (Ser 727 ), were differentially prevented in ATM-deficient cells compared with ATM-expressing control cells. Furthermore, the defective kinase activation of ATM caused a reduction and/or delay in UVA-stimulated phosphorylation and activity of ERKs, JNKs, and p38 kinase but had no effect on the response to UVC. But basal protein expression of the above mentioned signal molecules, including RSK1, RSK2, STAT1, STAT3, ERKs, JNKs, and p38 kinase, was not influenced by deficiency of ATM kinase. Moreover, RSKs have been shown to be activated and phosphorylated by active PDK1 (54), ERKs, and JNKs but not p38 kinase (39). However, phosphorylation (Ser 727 ) of STAT3 immunoprecipitates was induced by active JNKs but not ERKs or p38 kinase, and the UVA-stimulated intracellular phosphorylation of STAT3 was not mediated by the phosphatidylinositol 3-kinase/PDK1 pathway (29). Overall, these findings suggest a model for UVA-induced STAT3 phosphorylation (Ser 727 ) (Fig. 8). The model indicates that RSK2 is a potential upstream kinase required for STAT3 phosphorylation (Ser 727 ), and ATM kinase may play a regulatory role in UVA-induced phosphorylation of STAT3 through a pathway involving ERKs, JNKs, and RSK2. In addition, the regulation of ATM kinase may also occur via an ERK/p38 kinase-dependent MSK1 pathway. However, whether ATM kinase plays a direct role in the phosphorylation of RSK2 or STAT3 will be further determined. On the other hand, our studies also suggest that UVC-stimulated STAT3 phosphorylation (Ser 727 ) may be triggered by an ATMindependent JNK/RSK pathway.
Ser 727 is shown to be located within a potential MAPK consensus (PMS727P) motif of the C-terminal transcriptional domain in STAT3 (9,11,12), predicting that MAPKs are direct kinases for catalyzing Ser 727 phosphorylation. Here, to further determine a role for RSK2 or JNK1 in the phosphorylation catalysis, we prepared GST-fusion wild-type or mutant STAT3 proteins as enzymatic substrates for in vitro kinase assays. Phosphorylation of the kinase reactive products was determined by both Western blotting with a phosphor-Ser 727 antibody and radioisotope labeling. We present evidence showing that active JNK1 may stimulate phosphorylation of the STAT3 proteins from immunoprecipitation, but not of GST-fusion STAT3␣wt from the pull-down assay. But, GST-fusion STAT3␣wt combined with immunoprecipitated STAT proteins was phosphorylated by JNK1. Moreover, Cao's laboratory (30) reported that JNKs can phosphorylate Ser 727 in the STAT3 protein, which is a recombinant C-terminal peptide-deleted bigger STAT3 fragment, but not a wild-type full-length STAT3 (personal correspondence). These observations indicate that a cofactor in the STAT3 immunoprecipitates may be required for JNK1-catalyzed phosphorylation of STAT3 and the action occurs probably by modulation of the molecular conformation of the kinase-substrate docking interaction.
In a mixture of STAT3 proteins from both immunoprecipitation and GST pull-down, GST-STAT3␤wt, a dominant negative STAT3 variant (34), was not phosphorylated by JNK1, suggesting that the C-terminal 55 amino acid sequence is essential for both total and Ser 727 phosphorylation of STAT3. Importantly, radioisotope assays showed that JNK1 can induce total phosphorylation of a STAT3 point mutated at the Y705 (GST-STAT3␣Y705F) or Ser 727 site (GST-STAT3␣S727A) mixed with IP-STAT3 precipitates, whereas Western blotting revealed that JNK1 only stimulates Ser 727 phosphorylation of GST-STAT3␣Y705F, but not GST-STAT3␣S727A. Moreover, our recent report (29) showed that Ser 727 phosphorylation of IP-STAT3 is suppressed by preincubation of the phospho-STAT3 (Ser 727 ) antibody with a Ser 727 phospho-specific blocking peptide (sc-7993P), but not a Tyr 705 phospho-specific blocking pep-FIG. 8. A hypothetical model for RSK-mediated phosphorylation of STAT3 (Ser 727 ) induced by UVA or UVC irradiation. ATM or other kinases such as ATR (ATM-and Rad3-related) that lie at the beginning of multiple signal transduction pathways are activated by UVA or UVC irradiation. Subsequently, MAPKs including JNKs and ERKs are stimulated and contribute to RSK-mediated STAT3 signaling activation. RSKs or JNKs only in the presence of a downstream kinase and/or a cofactor may act as upstream kinases responsible for phosphorylation of STAT3 at Ser 727 , as well as at an unidentified non-Ser 727 residue. STAT3␤, a dominant negative variant, has not yet reported to be phosphorylated by so far known kinases and but inhibits induction of phosphorylation of STAT3. The arrows or bars indicate activation or inhibition, respectively. The broken arrow or question markers represent unidentified or unknown. The p, pS727, or pS/T indicates phosphorylation, STAT3 phosphorylation at Ser 727 or at a non-Ser 727 serinethreonine site, respectively.