Serine Phosphorylation and Negative Regulation of Stat3 by JNK*

STATs are activated by various cytokines and growth factors via tyrosine phosphorylation, which leads to sequential dimer formation, nuclear translocation, binding to specific DNA sequences, and regulation of gene expression. Recently, serine phosphorylation of Stat3 on Ser-727 by ERK has been identified in response to epidermal growth factor (EGF). Here, we report that Ser-727 phosphorylation of Stat3 can also be induced by JNK and activated either by stress or by its upstream kinase and that various stress treatments induce serine phosphorylation of Stat3 in the absence of tyrosine phosphorylation. Inhibitors of ERK and p38 did not inhibit UV-induced Stat3 serine phosphorylation, suggesting that neither of them is involved. We further demonstrate that JNK1, activated by its upstream kinase MKK7, negatively regulated the tyrosine phosphorylation and DNA binding and transcriptional activities of Stat3 stimulated by EGF. Correspondingly, pretreatment of cells with UV reduced the EGF-stimulated tyrosine phosphorylation and phosphotyrosine-dependent activities of Stat3. The inhibitory effect was not observed for Stat1. Our results suggest that Stat3 is a target of JNK that may regulate Stat3 activity via both Ser-727 phosphorylation-dependent and -independent mechanisms.

STATs 1 are activated by various cytokines and several growth factors and function as important biological links between the cell surface and the transcriptional events in the nucleus. Seven STAT genes have been identified, which contain a conserved structure of SH2 and a DNA-binding domain (reviewed in Ref. 1). Binding of cytokines to their respective receptors stimulates the Janus kinase family, which phosphorylates STAT proteins on a specific tyrosine residue (Tyr-701 in Stat1 and Tyr-705 in Stat3) at the COOH terminus. Homo-or heterodimers are formed between the phosphorylated tyrosine and its partner's SH2 domain. These dimers translocate into the nucleus and function as transcription factors by binding to their recognition sequences and regulating the target gene expression (reviewed in Refs. [1][2][3]. In addition to cytokines, growth factors such as EGF, platelet-derived growth factor, and colony-stimulating factor-1 also stimulate STAT tyrosine phosphorylation presumably through their intrinsic receptor tyrosine kinase activity (4 -7) or by the non-receptor tyrosine kinases such as Src (8,9).
Serine phosphorylation of STATs has also been demonstrated. A Pro-X-Ser-Pro sequence that is a recognition site of ERKs has been found at the COOH terminus of Stat1, Stat3, and Stat4, suggesting that ERK is involved in the phosphorylation of these STATs (10). ERKs are members of the mitogenactivated protein kinase (MAPK) family that are activated by growth factor stimulation and have been shown to play a role in cell proliferation and differentiation (11)(12)(13). It has been reported that ERK2 co-immunoprecipitated with Stat1␣ in response to interferon-␤ and was involved in the regulation of interferon-␤-induced gene expression (14). Recently, it has also been reported that ERKs phosphorylate Stat3 on Ser-727 in vitro as well as in vivo in response to EGF (15). However, the existence of serine/threonine kinases other than ERKs phosphorylating STATs on serine has also been suggested. For example, Stat1 is a relatively poor substrate for ERKs (15). In addition, although phosphorylation of Stat1 on Ser-727 is induced in response to interferon-␥, ERKs are not activated by interferon-␥ and therefore are unlikely to be involved in such phosphorylation (16). Moreover, serine phosphorylation of Stat3 by IL-6 stimulation has been shown to be ERK-independent (15), and the involvement of H-7-sensitive serine kinases has also been reported (17)(18)(19).
Two other subtypes of mammalian MAPKs that are activated by environmental stress and pro-inflammatory cytokines have been identified. JNKs, also known as stress-activated protein kinases, are activated by IL-1, tumor necrosis factor (TNF), UV radiation, and anisomycin (20 -22). JNKs bind to the amino terminus of c-Jun and phosphorylates it on Ser-63 and Ser-73 (23). The third group of MAPKs, p38, is activated by endotoxic lipopolysaccharide or hyperosmolarity (24). Although ERKs, JNKs, and the p38 kinase families are closely related due to their similar regulatory TXY motif for activity, they are distinguishable by unique (although sometimes overlapped) upstream activators and downstream substrates (25)(26)(27). We investigated whether Stat3 can be phosphorylated by stress and pro-inflammatory cytokines and examined which kinases are involved in such phosphorylation. We observed that Stat3 was phosphorylated on Ser-727 by TNF-␣ and various stress treatments. JNK1 activated either by UV or anisomycin or by its upstream kinase MEKK1 phosphorylated Stat3 in vitro. The major phosphorylation site was identified to be Ser-727. Stat3 can also be phosphorylated by cotransfection of JNK1 with MEKK1 in vivo. We further demonstrate that activation of JNK1 either by its upstream kinases or by UV treatment resulted in negative regulation of its activity.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-The glutathione S-transferase (GST)-Stat3 fusion protein, containing an almost full-length Stat3, was constructed as described previously (9). The point mutant GST-S1, in which Ser-727 of GST-Stat3 was replaced by Ala, was prepared using the polymerase chain reaction-based site-directed mutagenesis kit Ex-Site TM (Stratagene) following the manufacturer's instructions. The deletion mutant GST-C2 (containing amino acids 480 -770) was generated by digesting GST-Stat3 with XmnI/XhoI, isolating the fragment, and inserting it into pGEX-KG. The expression plasmid of Stat3, pRc/CMV-Stat3 (6), was obtained from Dr. J. E. Darnell, Jr. (Rockefeller University). Substitution of the phosphorylation site Ser-727 by Ala was performed with the QuikChange TM site-directed mutagenesis kit (Stratagene). The correct construction was confirmed by sequencing. The activated MEKK1 mutant (28) was provided by Dr. R. Janknecht (Mayo Foundation). pSR␣-HA-JNK1 (26) and kinase-deficient mutant JNK1 pcDNA3.FLAG.JNK1(APF) (21) were obtained from Drs. A. Whitmarsh and R. J. Davis (University of Massachusetts). MKK7 plasmids (29) expressing active MKK7 (pcDNA3.MKK7D) and the kinasedeficient mutant (pcDNA3.MKK7A) were provided by Dr. J. Han (Scripps Research Institute). The reporter plasmid pSIE-CAT for CAT assays was prepared by inserting three copies of the hSIE consensus sequence (TTCCCGTAA) upstream of a c-fos minimal promoter followed by the CAT gene in plasmid pFOSCAT⌬56 (30).
Immunoprecipitation/Western Blotting and Immune Complex Protein Kinase Assay-COS-1 cells transfected with expression plasmids were lysed in radioimmune precipitation assay buffer (150 mM NaCl, 50 mM Tris-HCl (pH 7.2), 1% deoxycholic acid, 1% Triton X-100, 0.25 mM EDTA (pH 8.0), and protease and phosphatase inhibitors (5 g/ml leupeptin, 5 g/ml aprotinin, 1 g/ml pepstatin A, 1 mM phenylmethylsulfonyl fluoride, 5 mM NaF, and 100 M sodium orthovanadate)). The cell lysates were incubated with an anti-JNK1 antibody (Santa Cruz Biotechnology) overnight at 4°C, followed by incubation with protein G PLUS/protein A-agarose (Oncogene Science Inc.) for 1 h. The immunoprecipitates were washed twice with radioimmune precipitation assay buffer and twice with cold phosphate-buffered saline and divided into two portions. One portion was subjected to Western blot analysis with an anti-phospho-JNK1 antibody (Santa Cruz Biotechnology). The other portion was subjected to in vitro kinase assay. Briefly, the immunoprecipitates were washed once with JNK kinase assay buffer containing 20 mM Hepes (pH 7.3), 20 mM MgCl 2, 20 mM ␤-glycerophosphate, 0.2 mM sodium orthovanadate, and 2 mM dithiothreitol. The GST-Stat3 fusion proteins used as substrates were produced and partially purified as described previously (9). Glutathione-Sepharose-bound GST-Stat3 was eluted by vortexing for 15 min at room temperature in an equal volume of 20 mM reduced glutathione, resuspended in 50 mM Tris-HCl (pH 8.0), concentrated using a Centriprep 10 (Amicon, Inc.), washed once in 20 mM Hepes (pH 7.3), and further concentrated using an ULTRA-FREE-MC filter unit (Millipore Corp.). Equal amounts of fusion proteins were used in the kinase assays. The fusion proteins were incubated with immunoprecipitated JNK1 in the kinase assay buffer in the presence of 5 or 10 Ci of [␥-32 P]ATP at 30°C for 15-30 min. The reaction mixture was boiled in Laemmli buffer, separated by 10% SDS-PAGE, transferred to nitrocellulose membrane, and exposed to x-ray film. The blots were also subjected to Amido Black staining to show the equal amount of GST fusion proteins used in each reaction. As for Western analysis of total cell lysates, equal amounts of lysates were separated by SDS-PAGE, transferred to a polyvinylidene difluoride, membrane, and blotted with the respective antibodies, including antiphospho-Ser-727 Stat3, anti-phospho-Tyr-705 Stat3 (both from New England Biolabs Inc.), and anti-Stat3 (Transduction Laboratories) antibodies.
[ 32 P]Orthophosphate Labeling-COS-1 cells were transfected and labeled with [ 32 P]orthophosphate at a final concentration of 1 mCi/ml for 4 h before harvesting. The cells were lysed, and the lysate was immunoprecipitated with an anti-Stat3 antibody (Santa Cruz Biotechnology). The immunoprecipitates were washed, and the proteins were separated by SDS-PAGE, transferred to nitrocellulose membrane, and exposed to x-ray film as described previously (31).
DNA Transfection, CAT Assay, and Mobility Shift DNA Binding Assay-COS-1 cells were grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum (Life Technologies, Inc.). Transfection of plasmids into COS-1 cells was performed by the calcium phosphate-DNA coprecipitation method (32). For CAT assays, the cells were cotransfected with 4 g of CAT-containing reporter plasmids, 3-5 g of expression plasmids, and 1 g of pCMV-␤-gal containing the bacterial ␤-galactosidase gene. The cells were lysed in 0.25 M Tris-HCl (pH 8.0) with three freeze-thaw cycles after 45 h of transfection. The lysate was spun, and the supernatant was collected and used for ␤-galactosidase activity and CAT assays. The pellets were resuspended in high salt buffer (20 mM Hepes (pH 7.9), 1 mM EDTA, 1 mM EGTA, 420 mM NaCl, 20% glycerol, 1 mM Na 4 P 2 O 7 , 1 mM Na 3 VO 4 , 20 mM NaF, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) and extracted as crude nuclear extract for DNA mobility shift assay. The amount of cytoplasmic extract used in each CAT assay was normalized with equivalent ␤-galactosidase activity. Acetylated and non-acetylated forms of [ 14 C]chloramphenicol were separated by thin-layer chromatography, followed by autoradiography and quantification using a Bio-Rad GS700 imaging densitometer. The mobility shift DNA binding assay was performed using hSIE as a probe under conditions described previously (9), except in the absence of salt in the binding buffer.

Stress Treatments Induce Serine Phosphorylation of Stat3 in
Vivo-Stat3 is activated by tyrosine phosphorylation on Tyr-705 in response to growth factors and cytokines. In addition, serine phosphorylation of Stat3 has been observed, and the major phosphorylation site is Ser-727. We investigated whether environmental stress or inflammatory cytokines can induce phosphorylation of Stat3. COS-1 cells, which express low levels of endogenous Stat3 (10), were transfected with Stat3 expression plasmid and treated with TNF-␣ and various stresses. Phosphorylation of Stat3 was examined with antibodies specifically recognizing either Ser-727-or Tyr-705-phosphorylated Stat3 protein in Western blot analysis. Fig. 1A (upper panel) shows that Stat3 Ser-727 phosphorylation was induced by UV, anisomycin, TNF-␣, and sodium arsenite and, to a weaker extent, by NaCl, okadaic acid, and lipopolysaccharide. In contrast, Tyr-705 phosphorylation of Stat3 was undetected in cells with these treatments (middle panel). EGF, as a positive control, induced both strong tyrosine and serine phospho- rylation of Stat3. An equal expression of Stat3 is shown in the lower panel. We also tested the induction of serine phosphorylation of endogenous Stat3 by stress in NIH 3T3 cells. TNF-␣ and various stress treatments, together with a positive control (platelet-derived growth factor), stimulated the serine phosphorylation of Stat3 (Fig. 1B). These results indicate that both endogenous and exogenous Stat3 can be phosphorylated on Ser-727 by stress treatments.
JNK1 Phosphorylates Stat3 on Ser-727 in Vitro-Since JNK/ stress-activated protein kinase is activated by stress and JNK1 is a major JNK, we next examined whether JNK1 was able to phosphorylate Stat3 in vitro. COS-1 cells were treated with UV, anisomycin, NaCl, or EGF; and the lysates were immunoprecipitated with an anti-JNK1 antibody, followed by an immune complex protein kinase assay using the GST-Stat3 fusion protein as a substrate. As shown in Fig. 2A, UV and anisomycin induced Stat3 phosphorylation by JNK1, whereas NaCl and EGF, which mainly activate p38 kinase and ERKs, respectively, did not stimulate Stat3 phosphorylation significantly.
MEKK1 phosphorylates the JNK upstream kinase SAPK/ ERK kinase-1, which in turn phosphorylates and activates JNK1 (28). To further confirm Stat3 phosphorylation by JNKs, we transfected COS-1 cells with JNK1 in the absence or presence of constitutively activated MEKK1 and performed the immune complex kinase assay. We observed that GST-Stat3 was strongly phosphorylated by MEKK1-activated JNK1, but not by JNK1 transfected alone or by endogenous JNK1 (Fig.  2B, upper panel). The activation of JNK1 by MEKK1 was confirmed by the strong phosphorylation of GST-c-Jun-(1-79), a physiological substrate of JNK1 (second panel, lane 3). This was also verified by the strong phosphorylation of JNK1 detected with an anti-phospho-JNK1 antibody that recognizes dual-phosphorylated JNK1 in Western blot analysis (third panel, lane 3). An equal expression of exogenous JNK1 in the presence or absence of MEKK1 was shown (lower panel). These data suggest that JNK1, activated by either stress or by its upstream kinase, phosphorylates Stat3 on Ser-727 in vitro.
We further tested whether Ser-727 was the only phosphorylation site of JNK1. A point mutant (GST-S1) in which Ser-727 was replaced by alanine and a deletion mutant (GST-C2) containing the COOH-terminal portion of Stat3 (amino acids 480 -770) were generated, and the phosphorylation by JNK1 was tested. As shown in Fig. 2C, MEKK1-activated JNK1 strongly phosphorylated GST-Stat3 and GST-C2, but failed to phosphorylate GST-S1. The amounts of the fusion proteins GST-Stat3 and GST-S1 used in the kinase assays were comparable (indicated by asterisks in the lower panel). These data suggest that Ser-727 is the only phosphorylation site of JNK1 in vitro. JNK1 Phosphorylates Stat3 in Vivo-Next, we examined whether JNK1 phosphorylated Stat3 in vivo. COS-1 cells were either transfected with Stat3 expression plasmid alone or cotransfected with JNK1 and MEKK1 and labeled with [ 32 P]orthophosphate. The lysates were immunoprecipitated with an anti-Stat3 antibody. A basal level of Stat3 phosphorylation was observed in cells transfected with Stat3 alone (Fig. 3,  lane 1), which was strongly enhanced in cells cotransfected with JNK1 and MEKK1 (lane 2). As a positive control, EGF also stimulated Stat3 phosphorylation (lane 3). These data indicate that Stat3 can be phosphorylated by JNK1 in vivo.
Effects of ERK and p38 Inhibitors on UV-or EGF-induced Stat3 Serine Phosphorylation-To ascertain the noninvolvement of other MAPK family members in the stress-induced Ser-727 phosphorylation of Stat3, the inhibitors of MEK1 (PD98059) (33) and p38 kinase (SB203580) (34) were used to pretreat cells, followed by UV or EGF treatment. The Ser-727 phosphorylation was analyzed. UV-induced phosphorylation of Stat3 was not affected by either inhibitor (Fig. 4, middle panel). In contrast, EGF-induced phosphorylation was inhibited by PD98059, but not by SB203580 (right panel). The basal level of phosphorylation in uninduced cells was slightly decreased by both inhibitors (left panel). These results suggest that whereas ERKs phosphorylate Stat3 by EGF stimulation, ERKs and p38 are unlikely to be involved in Stat3 serine phosphorylation induced by UV.
JNK1 Activated by MKK7 Negatively Regulates Tyrosine Phosphorylation and DNA Binding and Transcriptional Activities of Stat3-The tyrosine phosphorylation of Stat3 by growth factors and cytokines is a prerequisite for its dimerization, DNA binding, and transactivation, whereas the Ser-727 phosphorylation alone does not stimulate Stat3 DNA binding and transcriptional activities (6,35,36). We studied the role of JNK1 in Stat3 function by testing its effect on the DNA binding and transcriptional activities of Stat3 stimulated by EGF. MKK7 (JNK kinase-2) has recently been identified to be a specific upstream activator of JNK1 without affecting ERKs or p38 (29,(37)(38)(39)(40). COS-1 cells were transfected with Stat3 alone or were cotransfected with JNK1 and/or MKK7 expression plasmids and treated with EGF. The DNA binding activity was analyzed using hSIE, a high affinity binding site for Stat3, as a probe. As reported previously, EGF induced the activation of Stat3 and Stat1 to form three complexes with SIE: SIF-A (Stat3 homodimer), SIF-B (Stat1/Stat3 heterodimer), and SIF-C (Stat1 homodimer) (6). The DNA binding activity of Stat3 was not observed in the untreated cells transfected with Stat3 (Fig.  5A, lane 3), but was induced after EGF treatment (lane 4, SIF-A and SIF-B). SIF-A was not affected by constitutively activated MKK7, JNK1, or kinase-deficient mutant JNK1 Ϫ alone (lanes 5-7), but was almost completely destroyed by cotransfection of MKK7 and JNK1 together (lane 8). The DNA binding activity was largely restored by cotransfection of either mutant MKK7 and wild-type JNK1 (lane 10) or constitutively activated MKK7 and kinase-deficient JNK1 (lane 9). It has been previously reported that the formation of SIF-C by endogenous Stat1 can be detected after EGF stimulation in COS cells (41). As shown in Fig. 5A, SIF-C was also detected upon EGF treatment (lane 4), but was unaffected by cotransfection with wild-type or mutant MKK7 and/or JNK1 (lanes 5-10).
We next examined whether activated JNK1 also affected Stat3 transcriptional activity stimulated by EGF. A reporter plasmid containing three copies of hSIE followed by a CAT gene was cotransfected with Stat3 in the presence or absence of JNK1 and/or MKK7 expression plasmids, and the CAT activities were analyzed. As illustrated in Fig. 5B, CAT activity increased 10.4-fold after EGF stimulation, which slightly decreased in the presence of MKK7 (7.5-fold), but was unaffected by either wild-type or mutant JNK1 (10.3-and 11.8-fold, respectively). However, when Stat3 was cotransfected with MKK7 and JNK1 together, CAT activity was completely inhibited (1.4-fold), whereas such inhibition was not observed with cotransfection of MKK7 and mutant JNK1 or of mutant MKK7 and JNK1. These results are consistent with the DNA binding data, indicating that activated JNK1 suppresses both the DNA binding and transcriptional activities of Stat3.
Src has been shown to specifically stimulate the tyrosine phosphorylation and DNA binding activity of Stat3, but not of Stat1 (8,9). To confirm the repression of Stat3 activity by JNK1 described above, we performed similar transfection experiments in which EGF treatment was replaced by cotransfection with Src, and the DNA binding activity and transactivation of Stat3 stimulated by Src were examined. Similar repression by activated JNK1 was observed (data not shown).
Western blot analysis verified the activation and expression of JNK1 and Stat3 in these transfection experiments. As shown in Fig. 5C, we observed strong JNK1 phosphorylation only in the presence of MKK7, but not in its absence or in the presence of mutant MKK7 Ϫ . Notably, JNK1 Ϫ was not phosphorylated by cotransfection with MKK7 (upper panel), although an equivalent expression of JNK1 and JNK1 Ϫ was observed (second panel). The different apparent molecular masses of JNK1 and JNK1 Ϫ were due to the different constructions, as JNK1 contains three copies of HA epitope, whereas JNK1 Ϫ contains one copy of FLAG sequence (21,26). The expression of transfected Stat3 protein was comparable in all the samples (lower panel). However, a significant decrease in the tyrosine phosphorylation of Stat3 in cells cotransfected with MKK7 and JNK1 stimulated with EGF (third panel) was observed, which correlated well with its reduced DNA binding and transcriptional activities. This further indicates that the repression of Stat3 activity by JNK1 is due to a decrease in its tyrosine phosphorylation.
To confirm these results, we performed similar DNA binding and CAT assays with MEKK1 instead of MKK7. The results were essentially the same (data not shown), indicating that JNK1 activated by both upstream kinases negatively regulates Stat3 DNA binding and transcriptional activities stimulated by either EGF or Src.
UV Pretreatment Decreases Tyrosine Phosphorylation and DNA Binding and Transcriptional Activities of Stat3-To investigate a possible physiological role of JNK1 phosphorylation in Stat3 function, we examined whether stress affects Stat3 activity stimulated by EGF. COS-1 cells were transfected with wild-type Stat3 and pretreated with UV for various time points, followed by EGF treatment; and the DNA binding activity was measured. As shown in Fig. 6A, EGF induced the activation of transfected Stat3 and endogenous Stat1 to form SIF-A, SIF-B, and SIF-C (lane 3). Pretreatment of the cells with UV significantly decreased SIF-A formation. SIF-B was also reduced, whereas SIF-C was not significantly affected (lanes 4 -6). This indicates that UV specifically decreases the DNA binding activity of Stat3, but not of Stat1. The effect of UV pretreatment on the EGF-induced transcriptional activity of Stat3 was also tested in CAT assays and shown to be inhibitory (Fig. 6B). Finally, we analyzed the effect of UV treatment on Stat3 tyrosine phosphorylation. In agreement with the inhibition of the DNA binding and transcriptional activities, a decrease in the tyrosine phosphorylation of Stat3 by UV pretreatment was detected (Fig. 6C, upper panel, lanes 3-5), whereas the Ser-727 phosphorylation induced by EGF was not affected by UV pretreatment (middle panel), probably due to the strong Ser-727 phosphorylation induced by EGF stimulation. An equal expression of Stat3 was indicated (lower panel). These results suggest that pretreatment of UV negatively affects Stat3 tyrosine phosphorylation and the phosphotyrosine-dependent activities. This inhibitory effect could be due to the Ser-727 phosphorylation by UV-activated JNK1 that occurred prior to the tyrosine phosphorylation stimulated by EGF. Alternatively, the repression could also be independent of Ser-727 phosphorylation. The possible factors include a general toxic effect of UV irradiation or phosphorylation on other serine  1 and 2) or treated with EGF (100 ng/ml) for 15 min (lane 3). The lysates were immunoprecipitated with an anti-Stat3 antibody. The immunoprecipitates were resolved by SDS-PAGE and transferred to a membrane, followed by autoradiography.
site(s) that affects Stat3 tyrosine phosphorylation and activities (see details under "Discussion"). DISCUSSION In addition to tyrosine phosphorylation, Stat1 and Stat3 are also phosphorylated on serine in response to cytokines and growth factors. ERKs, the prototype of MAPKs, were the first identified Ser/Thr kinases that phosphorylate Stat3 on serine by EGF stimulation (10, 15, and 17). In this study, we investigated whether environmental stress can induce phosphorylation of Stat3 and elucidated which kinase(s) is likely to be involved. We demonstrated that various stress treatments  4 -10). Crude nuclear extracts were prepared, and 10 g was used for the mobility shift DNA binding assay with hSIE as a probe as described under "Experimental Procedures." FP (lane 1) indicates free probe. SIF-A, SIF-B, and SIF-C are indicated by arrows. B, COS-1 cells were transfected with empty vector (Ϫ) or with Stat3 (St3) alone or with other expression plasmids as indicated, together with the reporter plasmid pSIE-CAT and pCMV-␤-gal. The cells were either left untreated or treated with EGF for 6 h before harvesting. The amount of cell lysates used for CAT assays was normalized by ␤-galactosidase assay as described under "Experimental Procedures." Acetylated and non-acetylated forms of [ 14 C]chloramphenicol were separated by thin-layer chromatography, followed by autoradiography. A representative autoradiograph is shown in the upper panel. The CAT activities from three independent transfection experiments were quantified using a Bio-Rad GS700 imaging densitometer, and the average -fold induction is indicated at the top of each bar (lower panel). C, total cell lysates from the transfected cells described above were prepared and subjected to Western blot analysis to analyze the activation and expression of HA-JNK1, mutant FLAG-JNK1 (as indicated by JNK1 Ϫ ), and Stat3 with the respective antibodies as indicated. stimulate serine phosphorylation of both endogenous and exogenous Stat3 (Fig. 1); and JNK, a subtype of MAPKs, mediates the stress-dependent serine phosphorylation of Stat3 (Figs. 2 and 3). The site of phosphorylation of Stat3 by JNK1 was identified to be Ser-727 in vitro. Our data using the inhibitors of ERK and p38 pathways (Fig. 4) further support the specificity of Stat3 serine phosphorylation by JNK. These results demonstrate that JNK is the kinase that phosphorylates Stat3 in response to stress. Since phosphorylation of Stat3 on Ser-727 was also observed upon treatment of cells with sodium chloride, okadaic acid, and lipopolysaccharide, which stimulate p38 activity, we examined if p38 kinase, the other member of the MAPK family, could phosphorylate Stat3. We were not able to detect Ser-727 phosphorylation of the GST-Stat3 fusion protein by p38 activated either by stress or by cotransfection with its upstream kinase MKK3 in vitro (data not shown). However, the possibility of Stat3 phosphorylation by p38 in vivo cannot be excluded, and whether JNK is the only kinase family that is involved in Stat3 serine phosphorylation by various stress treatments remains to be determined. In addition to MAPKs, we recently identified protein kinase C ␦ to specifically associate with and to phosphorylate Stat3 in an IL-6-dependent manner (42). Together, these data indicate that Stat3 is a target for multiple Ser/Thr kinases that are activated by distinct extracellular stimuli and suggest that Stat3 may be functionally involved in diverse cellular processes.
It is generally accepted that the tyrosine phosphorylation of STATs is a prerequisite for their DNA binding and transactivation, although growth factors and cytokines induce phosphorylation of STATs on both tyrosine and serine (1-3). The question arising here is how does serine phosphorylation affect STAT activity? An initial report indicated that serine phosphorylation is required for the DNA binding of Stat3 in certain cell types (17). However, it was demonstrated later that phosphorylation on Ser-727 is not necessary for its DNA binding, but is required for the full transcriptional activity of Stat1 and Stat3 (10,36). On the other hand, a negative effect of Ser-727 phosphorylation on the tyrosine phosphorylation of Stat3 has also been suggested (15). We examined how JNK affects the DNA binding and transcriptional activities of Stat3 stimulated either by EGF or by Src and observed that JNK1, activated either by its upstream kinase MKK7 or by UV irradiation, inhibited the tyrosine phosphorylation and DNA binding and transcriptional activities of Stat3 in both cases (Figs. 5 and 6 and data not shown). Such repression is specific for Stat3 since Stat1 DNA activity was not inhibited (Fig.  5A). These results are in agreement with previous reports showing that ERK2, when activated by MEK1, represses the tyrosine phosphorylation and tyrosine phosphorylation-dependent activities of Stat3 stimulated by EGF or IL-6 (43,44). Furthermore, an inhibitory effect of protein kinase C␦ on the activity of Stat3 was also reported (42). These results suggest that activated MAPKs as well as other Ser/Thr kinases may negatively regulate STAT activity. These observations seem to be contradictory to the positive regulatory role of Ser-727 phosphorylation in STAT transcriptional activity (10). We attempted to address this question by further investigating the mechanisms of the repression. From our preliminary results, in agreement with the positive role of Ser-727 phosphorylation, we also observed a reduced transcriptional activity of the Stat3 mutant S727A stimulated by EGF. However, the DNA binding and transcriptional activities of FIG. 6. UV pretreatment decreases EGF-induced Stat3 tyrosine phosphorylation and DNA binding and transcriptional activities. A, COS-1 cells were transfected with Stat3 and left untreated (U) or treated with UV (120 J/m 2 ) for 15, 30, or 60 min, followed by EGF treatment (E; 100 ng/ml) for 15 min. Crude nuclear extracts were prepared, and 10 g was used for the mobility shift DNA binding assay with hSIE as a probe as described under "Experimental Procedures." FP (lane 1) indicates free probe. B, COS-1 cells were transfected and treated in the same manner as described for A, except that EGF treatment was for 6 h. CAT activity was analyzed, and a representative autoradiograph is shown in the upper panel. The average -fold induction indicated on top of each bar was from two independent transfections (lower panel). C, shown are the results from the inhibition of EGF-induced tyrosine phosphorylation of Stat3 by UV pretreatment. Total cell lysates from the transfected cells described above were prepared and subjected to Western blot analysis with antiphospho-Tyr-705 Stat3, anti-phospho-Ser-727 Stat3, or anti-Stat3 antibody as indicated. vec, empty vector. S727A were further inhibited by activated ERK or JNK (data not shown), suggesting that the repression is unlikely mediated by Ser-727 phosphorylation. This result is consistent with the report showing that repression of IL-6-stimulated Stat3 activity by ERK is independent of Ser-727 phosphorylation (44). From these data, we propose that ERK and JNK may have dual effects on Stat3 transcriptional activity, i.e. up-regulation by Ser-727 phosphorylation and down-regulation in a Ser-727-independent manner, a concerted contribution to the resultant regulation of Stat3 transactivation. These results also suggest a critical and complex role of the MAPK pathway in the regulation of STATs. Although the mechanisms of the negative regulation are still unknown, a few possibilities may be considered. First, activation of the MAPK pathways may negatively regulate the activity of the upstream tyrosine kinases such as EGF receptor, Src, and Janus kinases, which are involved in Stat3 tyrosine phosphorylation. Second, although we only detected phosphorylation of Ser-727 by JNK1 and ERK2 in vitro (Fig. 2C) (43), serine/threonine site(s) other than Ser-727 may be phosphorylated by these kinases in vivo (15). It is possible that phosphorylation on serine in DNA-binding domain of Stat3 may inhibit its DNA binding and transcriptional activities. Third, activated ERKs and JNKs may affect the activity of the specific inhibitors of Janus kinase/STAT pathways, namely the recently identified suppressor of cytokine signaling-1 or protein inhibitor of activated Stat3 (reviewed in Ref. 45). Finally, these kinases may modulate Stat3 activity by association since we observed a strong binding of Stat3 with activated ERK2 as well as protein kinase C-␦ (42,43).
In addition to stress, emerging evidence has shown that STATs are phosphorylated exclusively on serine in the absence of tyrosine phosphorylation. Examples include insulin-induced serine phosphorylation of Stat3 in 3T3L1 adipocytes (46) and steel factor-induced phosphorylation of Stat3 in human growth factor-dependent myeloid cell lines (47). Activation of protein kinase C by phorbol esters is also reported to induce phosphorylation of Stat3 on Ser-727 in T lymphocytes (19). These data indicate that serine phosphorylation alone may play a role in cellular regulation. Although the physiological role of serine phosphorylation in STAT function is still obscure, reports suggest that Stat3 may be involved in the regulation of differentiation in macrophages and pathogenesis of chronic lymphocytic leukemia (48,49). A recent report indicated that Stat1 regulates apoptosis induced by TNF-␣ by a novel signaling pathway in which phosphorylation on serine, but not tyrosine, may be involved (50). A challenge for further studies is to determine the physiological role of serine phosphorylation in STAT function.
JNK binds and phosphorylates some transcriptional activators. The best studied example is the c-Jun transactivator. The phosphorylation of c-Jun at Ser-63 and Ser-73 in the activation domain increases c-Jun transactivation (23). JNK also increases the transcriptional activity of the transactivators ATF-2 and Elk-1, a subfamily of ETS domain transcription factors (51,52). In this study, we have identified Stat3, another transcription factor, as a substrate of JNK and demonstrated that the regulation of Stat3 by JNK may be via a novel mechanism that involves Ser-727 phosphorylation-dependent and -independent mechanisms.