Activation of STAT3 by Gαs Distinctively Requires Protein Kinase A, JNK, and Phosphatidylinositol 3-Kinase*

Signal transducer and activator of transcription 3 (STAT3) can be stimulated by several Gs-coupled receptors, but the precise mechanism of action has not yet been elucidated. We therefore examined the ability of GαsQ226L (GαsQL), a constitutively active mutant of Gαs, to stimulate STAT3 Tyr705 and Ser727 phosphorylations in human embryonic kidney 293 cells. Apart from GαsQL, the stimulation of Gαs by cholera toxin or β2-adrenergic receptor and the activation of adenylyl cyclase by forskolin, (Sp)-cAMP, or dibutyryl-cAMP all promoted both STAT3 Tyr705 and Ser727 phosphorylations. Moreover, the removal of Gαs by RNA interference significantly reduced the β2-adrenergic receptor-mediated STAT3 phosphorylations, denoting its capacity to regulate STAT3 activation by a G protein-coupled receptor. The possible downstream signaling molecules involved were assessed by using specific inhibitors and dominant negative mutants. Induction of STAT3 Tyr705 and Ser727 phosphorylations by GαsQL was suppressed by inhibition of protein kinase A, Janus kinase 2/3, Rac1, c-Jun N-terminal kinase (JNK), or phosphatidylinositol 3-kinase, and a similar profile was observed in response to β2-adrenergic receptor stimulation. In contrast to the Gα16-mediated regulation of STAT3 in HEK 293 cells (Lo, R. K., Cheung, H., and Wong, Y. H. (2003) J. Biol. Chem. 278, 52154–52165), the Gαs-mediated responses, including STAT3-driven luciferase activation, were resistant to inhibition of phospholipase Cβ. Surprisingly, Gαs-mediated phosphorylation at Tyr705, but not at Ser727, was resistant to inhibition of c-Src, Raf-1, and MEK1/2 as well as to the expression of dominant negative Ras. Therefore, as with other Gα-mediated activations of STAT3, the stimulatory signal arising from Gαs is transduced via multiple signaling pathways. However, unlike the mechanisms employed by Gαi and Gα14/16, Gαs distinctively requires protein kinase A, JNK, and phosphatidylinositol 3-kinase for STAT3 activation.

By virtue of their linkage to the superfamily of seven-transmembrane receptors, the heterotrimeric G proteins are critical players in the regulation of cellular functions ranging from cell proliferation to differentiation. There is increasing evidence to suggest that malfunctions in G protein signaling may be associated with various disease states such as bacterial infections (Vibrio cholera and Bordetella pertussis), pseudohypoparathyroidism, McCune-Albright syndrome, cancers, and night blindness (1). All four classes of G proteins (G s , G i , G q , and G 12 ) possess the ability to stimulate mitogenesis and induce neoplastic growth (2)(3)(4)(5)(6)(7)(8), as illustrated by the expression of constitutively active mutants of the G␣ subunits in different cell types. Attempts to elucidate the mechanisms by which G proteins transduce proliferative signals have revealed many new conduits for G protein signaling. These new pathways provide linkages to oncogenes, kinases, and transcription factors. It has now become apparent that many G protein-coupled receptors (GPCRs) 3 can modulate the activities of mitogen-activated protein kinases (MAPKs), thereby allowing them to regulate cell proliferation and differentiation (9). Regulation of MAPKs by G proteins proceeds via complex signaling networks involving oncogenes such as Src tyrosine kinases (10) and the monomeric GTPases Ras and Rac1 (11,12). Transcription factors such as signal transducers and activators of transcription (STATs) have also been shown to participate in the transduction of G proteinmediated proliferative signals (7).
STATs are key players in mitogenesis because all seven members of STATs are associated with major types of cancer (13). Although STATs are typically stimulated by cytokines or growth factor receptors via Janus kinases (JAKs) (14), at least 20 GPCRs have now been shown to possess the ability to activate STATs in a variety of cell types (15)(16)(17)(18)(19)(20). STAT3 is one of the most widely studied STAT proteins, and its activation apparently involves multiple pathways. Both viral Src (v-Src) and cel-lular Src (c-Src) tyrosine kinases can activate STAT3 in mammalian fibroblasts (21,22). Moreover, MAPKs can phosphorylate Ser 727 on STAT3 to modulate its transcriptional activity (23), whereas activation of p38 MAPK and c-Jun N-terminal kinase (JNK) is thought to be required for v-Src activation of STAT3 (24). These and other signaling molecules have been similarly implicated in the activation of STAT3 by G proteins. Neoplastic transformation of NIH-3T3 cells by constitutively active G␣ o is mediated via STAT3 in an Src-dependent manner (7), although expression of a dominant negative mutant of G␣ i2 in the same cell type inhibits Src kinase activity and Tyr 705 phosphorylation of STAT3 (25). The G␣ q family members, G␣ 14 and G␣ 16 , also employ c-Src for the activation of STAT3 (26,27). Unlike G␣ o/i2 -mediated stimulatory signals, G␣ 14/16 additionally requires the Ras/Raf/MEK/ERK and PLC␤/PKC/ CaMKII pathways for STAT3 activation.
Among the GPCRs that are known to be capable of activating STAT3, most are linked to either G i/o or G q , with only a handful coupled to G s . Although activation of the G s -coupled human thyrotropin receptor and the prostacyclin receptor (hIP) can lead to STAT3 phosphorylation in FRTL-5 (28) and human erythroleukemia (HEL) (29) cells, respectively, both receptors also possess the ability to signal via G q proteins (29,30). This raises a concern as to whether the G s protein can truly activate the STAT3 pathway because G q proteins such as G 14 and G 16 are known activators of STAT3 (26,27). In FRTL-5 cells, elevation of intracellular cAMP activates STAT3, but inhibition of protein kinase A (PKA) does not affect thyrotropin receptormediated STAT3 activation (28). In contrast, cytokine-triggered STAT3 Tyr 705 phosphorylation and DNA binding are inhibited by cAMP in human mononuclear cells (31). Moreover, isoproterenol-induced phosphorylation of STAT3 in cultured cardiomyocytes can be markedly enhanced by the phosphodiesterase inhibitor amrinone, indicating that cAMP is critically involved in ␤-adrenergic receptor (␤-AR)-mediated STAT3 activation (32). However, this G s -induced cAMP signal may facilitate JAK/STAT3 signaling indirectly through the induction of cytokine expression such as interleukin-6 (33,34). The issue of whether G s can indeed directly stimulate STAT3 activity remains controversial.
We have recently demonstrated that the G s /G q -coupled hIP is capable of activating STAT3 in HEL cells, and the signal propagation appears to require JNK (29). Because G 16 -mediated activation of STAT3 in HEL cells, as well as in HEK 293 cells, is independent of JNK signaling (26), we hypothesized that this difference is because of G s -mediated signaling. In this study, we examined the ability of G␣ s to activate STAT3 by expressing the constitutively active G␣ s QL or stimulating the G s -coupled ␤ 2 AR in HEK 293 cells and characterizing the molecular components involved in the signal transduction.

EXPERIMENTAL PROCEDURES
Materials-The cDNAs of human G proteins were obtained from Guthrie Research Institute (Sayre, PA). Cell culture reagents, including Lipofectamine PLUS and Lipofectamine 2000 reagents, were obtained from Invitrogen. All kinase inhibitors, their negative analogues, and other cAMP analogues were ordered from Calbiochem (Darmstadt, Germany). Isoprotere-nol, salbutamol, carbachol, ICI 118-551, epidermal growth factor (EGF), and cholera toxin (CTX) were purchased from Sigma. Pertussis toxin (PTX) was obtained from List Biological Laboratories (Campbell, CA). G␣ s and Ras-GRF1 antisera were from Santa Cruz Biotechnology (Santa Cruz, CA), and other antisera were purchased from Cell Signaling (Beverly, MA). Nitrocellulose membrane and ECL kit were ordered from Bio-Rad and Amersham Biosciences, respectively. The Stealth Select RNAi for G␣ s and Stealth RNAi negative control were products of Invitrogen, and the RNAi for Ras-GFR1 was purchased from Santa Cruz Biotechnology. The luciferase reporter gene, pSTAT3-TA-luc, was from Clontech (Palo Alto, CA). The luciferase substrate and its lysis buffer were purchased from Roche Diagnostics (Mannheim, Germany).
Cell Culture and Transfection-Human embryonic kidney (HEK) 293 cells were purchased from the American Type Culture Collection (CRL-1573, Manassas, VA) and were grown in Eagle's minimum essential medium at 5% CO 2 , 37°C with 10% fetal bovine serum (FBS), 50 units/ml penicillin, and 50 g/ml streptomycin. For Western blotting analysis, HEK 293 cells were seeded on 6-well plates at a density of 5 ϫ 10 5 cells/well and were cultured in the growth medium for 18 -24 h prior to transfection. They were co-transfected with various cDNAs using Lipofectamine PLUS reagents. The transfection mixtures in serum-free Opti-MEM (600 l/well) contained 3 l of both PLUS and Lipofectamine reagents and 500 ng of each G protein cDNA. For the experiments with other signaling molecules, 250 ng of each construct was added. 3 h after transfection, 300 l of Opti-MEM containing 30% FBS was added into the wells. For G␣ s knockdown experiments, HEK 293 cells were seeded into 6-well plates at 3 ϫ 10 5 cells/well and again cultured in growth medium overnight. They were transfected and maintained in serum-free Opti-MEM containing 80 pmol of Stealth Select RNAi targeting G␣ s , RNAi targeting Ras-GRF1, or Stealth RNAi Negative Control Med GC and 5 l Lipofectamine 2000 per well for 24 h. For luciferase assays, HEK 293 cells were seeded into 96-well white microplates designed for luminescent work at 1.5 ϫ 10 4 cells/well and were cultured in minimum essential medium overnight. Cells were transiently transfected using Lipofectamine PLUS reagents (26). The transfection mixtures in serum-free Opti-MEM (100 l/well) contained 0.2 l of both PLUS and Lipofectamine reagents, 10 ng of cDNAs encoding G proteins or the control vector cDNAs, and 100 ng of pSTAT3-TA-luc. For the experiments with extra signaling molecules, 5 ng of each construct cDNA was added. After 3 h, 50 l of Opti-MEM containing 30% FBS was added into the wells and cultured overnight.
Western Blotting Analysis-30 h after transfection, HEK 293 cells were serum-starved overnight. Prior to cell lysis, incubation with different kinase inhibitors for 30 min, if applicable, was performed. An extra 30 min of agonist exposure was performed for receptor-induced STAT3 activations. For G␣ s knockdown experiments, transfectants were directly challenged with different ligands for 30 min. Cells were then lysed in 150 l of lysis buffer and then gently shaken on ice for 30 min. Supernatants were collected by centrifugation at 16,000 ϫ g for 5 min. Clarified lysates were resolved on 12% SDS-polyacrylamide gels and transferred to nitrocellulose membrane (35). STAT3, phospho-STAT3-Tyr 705 , phospho-STAT3-Ser 727 , and G␣ s were detected by specific primary antisera and horseradish peroxidase-conjugated secondary antisera. The immunoblots were visualized by chemiluminescence with the ECL kit, and the images detected in x-ray films were quantified by densitometric scanning using the Eagle Eye II still video system (Stratagene, La Jolla, CA).
Luciferase Assay-Transfectants were grown in culture medium for 30 h and then maintained in serumfree medium. Where indicated, cells were treated with different kinase inhibitors overnight in serum-free medium. Cell lysates were analyzed as described previously (26).

Activation of G␣ s Induces STAT3
Activations in HEK 293 Cells-We have previously demonstrated that hIP is coupled to both G s and G q for signal transduction (36,37), and G s may also participate in mediating the hIP-induced STAT3 phosphorylations in HEL cells (29). Hence, we began our study by determining whether a constitutively active mutant of G␣ s (G␣ s QL) possesses the ability to stimulate STAT3 phosphorylation and activation. HEK 293 cells were transiently transfected with pcDNA1, G␣ s , or G␣ s QL. Total cell lysates prepared from the transfected cells were probed with anti-phospho-STAT3-Tyr 705 , antiphospho-STAT3-Ser 727 , and anti-STAT3 antisera. Expression of either G␣ s or G␣ s QL in HEK 293 cells did not affect the expression of total STAT3 as compared with the vector control (Fig. 1A). Anti-phospho-STAT3-Tyr 705 and anti-phospho-STAT3-Ser 727 antisera revealed very low levels of basal STAT3 phosphorylation at either Tyr 705 or Ser 727 sites in the vector control and G␣ s -expressing cells. In contrast, expression of G␣ s QL led to a detectable increase in both Tyr 705 and Ser 727 phosphorylation of STAT3 (Fig.  1A). Pretreating the transfectants with PKA inhibitors, 10 M H-89 or 100 M (R p )-cAMP, significantly inhibited G␣ s QL-induced STAT3 Tyr 705 and Ser 727 phosphorylations. Examination of the STAT3-driven transcriptional activity was also performed using a luciferase reporter gene assay. In HEK 293 cells transiently co-transfected with pSTAT3-TA-luc, G␣ s QL significantly induced STAT3 transcriptional activation as com- pared with the corresponding controls (Fig. 1B). Inhibition of PKA by H-89 or (R p )-cAMP completely abrogated the G␣ s QLinduced luciferase activity, indicating that G␣ s /cAMP is able to mediate STAT3 phosphorylations and transcriptional activation.
To further confirm the specific involvement of G␣ s /cAMP in the activation of STAT3, we utilized G␣ s -activating CTX, adenylyl cyclase (AC)-stimulating forskolin (Fsk), and cAMP analogues (Bt 2 cAMP and (S p )-cAMP; Fig. 1C). Pretreating HEK 293 cells for 4 h with CTX to directly stimulate G␣ s by ADP-ribosylation (38) significantly induced both STAT3 Tyr 705 and Ser 727 phosphorylations (Fig. 1C). Stimulation of AC by Fsk and the applications of cAMP analogues (Bt 2 cAMP and (S p )-cAMP) also increased the STAT3 Tyr 705 and Ser 727 phosphorylations, although using (R p )-cAMP did not induce any significant change in STAT3 phosphorylations. Thus, activation of the G␣ s pathway could indeed result in STAT3 phosphorylation and transcriptional activation.
G␣ s QL Induces STAT3 Phosphorylations via MEK1/2, JNK, c-Src, and PI3K in HEK 293 Cells-A number of signaling molecules have been shown to be required for G␣-induced activation of STAT3 (7,26,27). They include c-Src (for G␣ o QL, G␣ 14 QL, and G␣ 16 QL), ERK, and PLC␤/PKC/CaMKII (for G␣ 14 QL and G␣ 16 QL). Hence, we asked if G␣ s QL-mediated STAT3 phosphorylations utilize the same signaling intermediates as other constitutively active G␣ subunits. PLC␤/PKC/CaMKII signaling cascade is one of the critical pathways regulating G␣ 14 -and G␣ 16 -mediated STAT3 activations (26, 27). Because G␣ s is unable to modulate the PLC␤ cascade, inhibiting PLC␤ should have no effect on G␣ s QLmediated STAT3 phosphorylations. HEK 293 cells were transfected with pcDNA1, G␣ s , or G␣ s QL and treated with different kinase inhibitors for 30 min (Figs. 1D, 2, 3, and 5). As expected, treatment of G␣ s QLtransfected cells with U73122 or U73343 (10 M) did not affect STAT3 phosphorylations (Fig. 1D), suggesting the lack of involvement of PLC␤ in G␣ s QL-mediated STAT3 activation.
ERK has been shown to be a central component in G␣ 16 QL-induced activation of STAT3 (26). To study the requirement of ERK in the G␣ s QL-induced STAT3 phosphorylations, we used 10 M U0126 to inhibit MEK1/2, the upstream regulators of ERK. Suppression of MEK1/2 activity by U0126 significantly blocked the Ser 727 , but not the Tyr 705 , phosphorylation of STAT3 ( Fig.  2A). This is in agreement with the differential attenuation of hIP-induced STAT3 Ser 727 phosphorylation by U0126 in HEL cells (29). Apart from ERK, the possible involvement of the other two MAPKs was also examined. Blockade of JNK and p38 MAPK by SP600125 (30 M) and SB202190 (10 M), respectively, produced very different results ( Fig. 2A). Inhibition of  Fig. 1A. Numerical values shown above the immunoreactive bands represent relative intensities of STAT3 phosphorylations expressed as a ratio of the basal level with dimethyl sulfoxide (set as 1.0). *, G␣ s QL-induced STAT3 phosphorylations were significantly higher than the basal value (one-way ANOVA with Dunnett's post-tests, p Ͻ 0.05). Data shown represent the mean Ϯ S.E. from three separate experiments, and the immunoblots shown represent one of three sets; two other sets yielded similar results.
JNK suppressed both Tyr 705 and Ser 727 STAT3 phosphorylations, whereas p38 MAPK inhibition was unable to alter any G␣ s QL-induced STAT3 activations. To address the specificity of these kinase inhibitors, we analyzed the MEK/ERK and JNK/ c-Jun activations using various antibodies following the treatment of cells with different kinase inhibitors. As shown in Fig.  2B, only the application of U0126 led to the attenuation of ERK phosphorylation, whereas SP600125 and the negative controls (U0124 and SP-ve) were ineffective. Inhibition of JNK by SP600125 suppressed the phosphorylation of c-Jun. The use of U0126, U0124, and SP-ve did not affect the phosphorylations of JNK and c-Jun (Fig. 2B).
Further studies on different monomeric GTPases associated with the MAPK cascades were performed by overexpressing their dominant negative mutants (Fig.  2C). The ability of G␣ s QL to stimulate STAT3 phosphorylations was unaffected by the presence of wild type monomeric GTPases. Consistent with the results obtained with U0126 ( Fig. 2A), inhibition of the Ras/Raf/MEK pathway by overexpressing the dominant negative Ras (RasDN) inhibited Ser 727 STAT3 phosphorylation but failed to attenuate the Tyr 705 STAT3 phosphorylation ( Fig. 2C). Along with STAT3, G␣ s QL-induced ERK phosphorylations were reduced as well, although the reductions were only partial ( Fig. 2D), thus indicating that ERK activation can be attained via other pathways besides Ras.
Inhibition of the Rac1/JNK pathway by overexpression of Rac1 negative mutant (Rac1DN) completely abolished G␣ s QL-induced STAT3 phosphorylations at both Tyr 705 and Ser 727 (Fig. 2C) where wild type Rac1 had no effect STAT3 phosphorylation. Overexpression of Rac1DN markedly reduced the phosphorylations of JNK as well as c-Jun (Fig. 2D). G␣ s QL-induced ERK phosphorylation was also partially inhibited by Rac1DN, denoting that Rac1 can regulate both JNK and ERK activities (Fig. 2D). As a control, we also examined monomeric RhoA GTPase on STAT3 regulations. Neither the expression of wild type RhoA nor dominant negative RhoA (RhoADN) elicited any inhibitory effect on G␣ s QL-induced STAT3 phosphorylations. Next we tested the involvement of c-Src because this tyrosine kinase is required for G␣ 16 QL-induced STAT3 phosphorylations (29). As shown in Fig. 3A, inhibitors of c-Src (PP1 and PP2) were capable of reducing G␣ s QL-induced STAT3 Ser 727 phosphorylation. This reduction was also observed when the dominant negative mutant of c-Src (c-SrcDN) was overexpressed ( Fig. 3B), whereas wild type c-Src neither affected the STAT3 expression nor the G␣ s QL-induced STAT3 Tyr 705 phosphorylation. However, unlike STAT3 Ser 727 phosphorylation, STAT3 Tyr 705 phosphorylation was insensitive to inhibi-  (Fig. 3A). However, JAK2/3 did not display any modulating capacity on G␣ s QL-induced ERK and JNK activations because neither AG490 nor WHI-P131 showed any effect on their phosphorylations (Fig. 3E). None of the inactive analogues of c-Src (PP3) and JAKs (WHI-P258 or WHI-ve) had any effect on STAT3 phosphorylations (Fig. 3A).
Because c-Src is capable of regulating both ERK and JNK (Fig.  3, C and D), we asked whether the MAPK signaling cascades can reciprocally modulate the c-Src/JAK2 pathway. We first examined the effect of monomeric Ras on c-Src phosphorylation. In Fig. 4A, overexpression of RasDN did not attenuate the G␣ s QLinduced c-Src phosphorylation. Further studies were performed using transfectants treated with U0126, SP600125, or their corresponding inactive analogues. G␣ s QL-induced c-Src and JAK2 phosphorylations were unaffected by U0126-medi-ated ERK inhibition, whereas inhibiting JNK by SP600125 was accompanied by a prominent suppression of JAK2, but not c-Src, phosphorylation (Fig. 4B). As a control, their inactive analogues failed to demonstrate any effects on the phosphorylations of these two tyrosine kinases. In addition, to confirm the regulatory capacity of JNK on JAK2 phosphorylation, we introduced the constitutively active mutant of JNKK (JNKK-CA; serving as an upstream activator of JNK signaling cascade) into HEK 293 cells (Fig. 4C). The presence of JNKK-CA significantly induced JNK as well as JAK2 phosphorylations. In contrast, JNKK-CA failed to promote either ERK or c-Src phosphorylation. As a control, overexpressing MEK1 (upstream kinase of ERK) was only able to stimulate ERK phosphorylation, but not JNK, c-Src, or JAK2. These data illustrated that both ERK and JNK signaling pathways failed to modulate c-Src phosphorylation because they are located downstream of c-Src (Fig. 3, C and  D). However, JNK, but not ERK, appeared to mediate the G␣ s QL-induced JAK2 phosphorylation leading to possible STAT3 Tyr 705 phosphorylation.
PI3K has been shown to mediate wortmannin-sensitive activation of MAPKs by GPCRs (39). Thus, PI3K appears to be a prime candidate for signal integration in the activation of STAT3 by G␣ s QL. G␣ s QL-induced STAT3 Tyr 705 and Ser 727 phosphorylations were both attenuated upon suppression of PI3K by its inhibitors, wortmannin (100 nM) and LY294002 (10 M; Fig. 5A). Similar suppressions were observed when the  dominant negative mutant of PI3K (PI3KDN) was transiently co-transfected into HEK 293 cells with G␣ s QL (Fig. 5B). Overexpression of wild type PI3K had no effect on STAT3 activations, whereas the overexpression of PI3KDN completely abrogated the G␣ s QL-induced STAT3 responses. In control experiments, application of PI3K inhibitors (wortmannin and LY294002) or overexpression of PI3KDN effectively removed Akt phosphorylation (data not shown). Suppression of PI3K activity weakly inhibited the ability of G␣ s QL to induce ERK phosphorylation (Fig. 5, A and B), suggesting that the regulation of STAT3 by PI3K may in part be mediated via ERK.
Effects of Various Inhibitors on G␣ s QL-induced STAT3-dependent Luciferase Activity-We have previously shown that G␣ 16 QL is fully capable of inducing STAT3-driven luciferase expression and that this response is c-Src-and MEK1/2-dependent but does not require the participation of PI3K (26). To investigate if activated G␣ s can similarly regulate STAT3-dependent transcriptional activity, HEK 293 cells were transiently transfected with pSTAT3-TA-luc, pcDNA1, G␣ s , or G␣ s QL and treated with different kinase inhibitors. As described earlier, G␣ s QL significantly induced STAT3-driven luciferase expression (Fig. 1B). The G␣ s QL-induced luciferase activity was, however, almost totally suppressed in the presence of 10 M H-89, 30 M SP600125, 100 M AG490, 100 g/ml WHI-P131, 100 nM wortmannin, or 10 M LY294002 (Fig. 6A). Inhibition of STAT3-driven luciferase activity by these agents corroborated with their ability to block G␣ s QL-induced STAT3 Tyr 705 and Ser 727 phosphorylations (Figs. 1-3 and 5). Given that STAT3 Tyr 705 phosphorylation alone is sufficient to drive STAT3 transcriptional activity (14), inhibition of Ser 727 phosphorylation by the inhibitors of c-Src and MEK should not completely suppress G␣ s QL-induced STAT3-driven luciferase activity. As predicted, G␣ s QL-induced STAT3 transcriptional activation was significantly but only partially (Ͼ65%) attenuated by 25 M PP1, 25 M PP2, 10 M Raf-1 kinase inhibitor, or 10 M U0126 (Fig. 6A). None of the corresponding inactive analogues, including inactive U0124 (10 M), PP3 (25 M), LY303511 (10 M), and WHI-P258 (10 M) showed any inhibitory effects.
The different dominant negative mutants of various signaling molecules, including the three small GTPases (Ras, Rac1, and RhoA), PI3K, and c-Src, were also examined for their effects on G␣ s QL-induced STAT3-driven luciferase expression (Fig. 6, B and C). None of the wild type proteins elicited any effects on G␣ s QL-induced STAT3-driven luciferase expression as compared with the vector control (pcDNA1). In line with the previous findings (Figs. 2 and 5), both Rac1DN (Fig. 6B) and PI3KDN (Fig. 6C) were capable of almost completely attenuating the G␣ s QL-induced STAT3-driven luciferase expression. RasDN (Fig. 6B) and c-SrcDN (Fig. 6C) only partially reduced the G␣ s QL-mediated luciferase responses. Finally, RhoADN was unable to suppress the G␣ s QL-induced STAT3-driven luciferase expression (Fig. 6B).
Pivotal Role of PKA in G␣ s -mediated STAT3 Activation-Because the cAMP/PKA pathway is immediately downstream of G␣ s , we further analyzed the effects of inhibiting PKA by H-89 on the various signaling molecules. Cell lysates from H-89treated cells were resolved and analyzed for the activations of c-Src, JAK2, ERK, and JNK (Fig. 7A). In agreement with Fig. 1, overexpression of G␣ s QL or the application of Bt 2 cAMP led to both STAT3 Tyr 705 and Ser 727 phosphorylations, and the STAT3 responses were abolished upon inhibition of PKA by H-89. H-89 also attenuated the G␣ s QL-and Bt 2 cAMP-induced activations of c-Src, JAK2, ERK, and JNK (Fig. 7A). These results suggest that PKA is probably upstream of these signaling molecules.
Recently, an endogenously expressed GEF, named Ras-GRF1, has been shown to be activated by G s -coupled serotonin 5-HT 7 receptor via PKA, and its activation leads to the stimu- , 100 M AG490, 100 g/ml WHI-P131 (WHI), or 100 g/ml WHI-P258 (WHI-ve) for overnight. B and C, HEK 293 cells, along with various G proteins and the STAT3 luciferase reporter, were co-transfected with different constructs of signaling molecules, including Rac1, Rac1, or RhoA (B) and c-Src, PI3K, or control vector (C). Transfectants were serum-starved overnight, and cell lysates were collected for luciferase evaluation as described in Fig. 1B. Data shown represent the mean Ϯ S.E. from four separate experiments performed in triplicate. *, STAT3 transcriptional activations were significantly higher than the pcDNA1 control without G␣ s QL (one-way ANOVA with Dunnett's post-tests, p Ͻ 0.05). #, Inhibitors or dominant negative mutants significantly suppressed the G␣ s QL-induced luciferase response (one-way ANOVA with Dunnett's post-tests, p Ͻ 0.05).
lation of the Ras/Raf/MEK/ERK cascade in HEK 293 cells (40). Interestingly, this GEF can also activate Rac1 via a separate structural domain (41). To test the involvement of Ras-GRF1 in G␣ s -mediated STAT3 activations, we employed the gene silencing technique on this GEF. As shown in Fig. 7B, significant reduction (ϳ70%) of the endogenous Ras-GRF1 was observed when 80 pmol of siRas-GRF1 was introduced into HEK 293 cells for 2 days; no significant change in the Ras-GFR1 expression level was detected in the controls (no addition or control siRNA; Fig. 7B). In cells subjected to siRas-GRF1, the G␣ s QL-induced STAT3 Tyr 705 and Ser 727 phosphorylations were reduced by ϳ60%. Likewise, G␣ s QLinduced phosphorylations of ERK and JNK were attenuated by ϳ50% upon Ras-GRF1 knockdown. In contrast, G␣ s QLinduced activations of c-Src and JAK2 were unaffected by the knockdown of Ras-GRF1. Similar findings were also obtained for Bt 2 cAMP-induced STAT3 phosphorylations (data not shown). Taken together, these results indicate that Ras-GRF1 plays a significant role in modulating ERK and JNK cascades but not for the c-Src/JAK pathway. G s -coupled ␤ 2 AR-mediated STAT3 Activations Also Require MEK1/2, JNK, PI3K, and c-Src-Because G␣ s QL stimulated the phosphorylation and activation of STAT3, we then sought to investigate whether the G s -coupled ␤ 2 AR is also capable of stimulating STAT3. HEK 293 cells are known to endogenously express ␤ 2 AR, and application of isoproterenol, a selective ␤-AR agonist, leads to an increase in AC activity (42). As shown in Fig. 8A, weak levels of STAT3 Tyr 705 and Ser 727 phosphorylations were observed in HEK 293 cells incubated with 10 M isoproterenol for 10 min or more. The magnitude of the isoproterenol-induced STAT3 phosphorylations (ϳ2-fold) was similar to that obtained with G␣ s QL (Fig. 1A), and the phosphorylations were sustained up to 120 min; the level of phosphorylated STAT3 gradually decreased thereafter and returned to basal level at 4 h after isoproterenol challenge (data not shown). As isoproterenol is a nonselective agonist for both ␤ 1and ␤ 2 ARs, we employed the highly selective agonist for ␤ 2 AR (salbutamol) to confirm the involvement of ␤ 2 AR. Exposure of HEK 293 to salbutamol (10 M) for 30 min significantly induced both Tyr 705 and Ser 727 STAT3 phosphorylations with the magnitude of stimulation comparable with that obtained with isoproterenol (Fig. 8B). Furthermore, isoproterenol-induced STAT3 phosphorylations were effectively blocked in the presence of a selective antagonist of ␤ 2 AR. As shown in Fig. 8B, ICI 115-881 (1 mM) treatment did not have any effect on STAT3 stimulations, although it has been reported to elevate basal ERK activity (43). Co-treatment of ICI 115-881 with isoproterenol completely abrogated the isoproterenol-induced STAT3 responses.
In light of the capacity of ␤ 2 AR to signal via G i proteins (44) and the ability of G␣ i to promote STAT3 activations (25), we next asked whether the G␣ i signals contribute to isoproterenol-induced STAT3. To eliminate the possible coupling of ␤ 2 AR to G␣ i proteins, HEK 293 cells were pretreated with PTX (100 ng/ml) overnight. The PTX pretreatment had no effect on basal levels nor did it suppress the isoproterenol-induced STAT3 phosphorylations (Fig. 8B). This suggests that the ␤ 2 AR-mediated STAT3 phosphorylations are primarily dependent on G s signaling.
To confirm this postulation, we introduced G␣ s -targeting siRNA into HEK 293 cells (Fig. 9A). Introduction of siG␣ s into HEK 293 cells substantially reduced the G␣ s expression level (Ͼ90%) as compared with parental HEK 293 cells (Fig. 9A). As reported in HeLa cells (45), compensatory increase of other G␣ subunits was not detectable (data not shown). In contrast, the expression of G␣ s was unaffected in cells transfected with the control siRNA. When HEK 293 cells were transfected with control siRNA, isoproterenol and salbutamol remained capable of promoting STAT3 phosphorylations at both Tyr 705 and Ser 727 (Fig. 9B, Ctrl siRNA). Moreover, activation of G q -coupled muscarinic receptors and EGF receptor by carbachol (100 M) and EGF (50 ng/ml), respectively, also resulted in STAT3 phosphorylations. However, reduction of G␣ s expression by siRNA completely inhibited the ␤ 2 AR-induced STAT3 activations, whereas the CCh-induced and EGF-induced STAT3 stimulations remained intact (Fig. 9B). These results demonstrate that knockdown of G␣ s by siRNA specifically affected the ␤ 2 ARinduced STAT3 phosphorylations without altering STAT3 signaling elicited by other non-G␣ s -dependent pathways. Cell lysates were resolved and analyzed for STAT3, ERK, JNK, JAK2, and c-Src phosphorylations as described in Fig. 1A. B, HEK 293 cells were transfected with pcDNA1, G␣ s , or G␣ s QL in the absence or presence of the siRNA of Ras-GRF1 (siRas-GRF1) or control siRNA (Ctrl siRNA). Effective knockdown of Ras-GRF1 was confirmed by an anti-Ras-GRF1 antiserum (lowest panel). Cell lysates were resolved and analyzed for STAT3, ERK, JNK, JAK2, and c-Src phosphorylations as well as total expression of Ras-GRF1 as described in Fig. 1A. Numerical values shown above the immunoreactive bands represent relative intensities of STAT3 phosphorylations expressed as a ratio of the basal level with vector control (set as 1.0). Data shown represent the mean Ϯ S.E. from three separate experiments, and the immunoblots shown represent one of three sets; two other sets yielded similar results. NOVEMBER 24, 2006 • VOLUME 281 • NUMBER 47

JOURNAL OF BIOLOGICAL CHEMISTRY 35819
The siRNA studies suggest that the regulation of STAT3 by ␤ 2 AR is mainly mediated via G␣ s -dependent pathways. If so, ␤ 2 AR-induced STAT3 phosphorylations should exhibit the same sensitivity to various kinase inhibitors as G␣ s QL (Figs. 1D,  2, 3, and 5). Indeed, inhibition of the MEK1/2 pathway by Raf-1 inhibitor and U0126 (Fig. 10A) as well as the attenuation of c-Src by PP1 and PP2 (Fig. 10B) significantly attenuated the isoproterenol-induced Ser 727 STAT3 phosphorylation, whereas the Tyr 705 STAT3 phosphorylation was not affected. Inhibitions of JNK by SP600125 (Fig. 10A), JAKs by AG490 or WHI-P121 (Fig. 10B), and PI3K by wortmannin or LY294002 (Fig. 10C) completely abrogated both Ser 727 and Tyr 705 STAT3 phosphorylations induced by isoproterenol. Specific inhibitors against p38 MAPK (SB202190; Fig. 10A) and PLC␤ (U73122; Fig. 10D) were unable to modify the patterns of ␤ 2 AR-induced STAT3 activations. Additionally, none of the inactive analogues tested (U0124, SP-ve, SB-ve, PP3, WHI-ve, and U73343) had any suppressive effect on isoproterenol-induced STAT3 phosphorylations. In summary, the patterns of STAT3 phosphorylations generated by the isoproterenol-induced ␤ 2 AR activation were identical to those elicited by the expression of G␣ s QL in HEK 293 cells.

DISCUSSION
In recent years, it has become increasingly clear that G protein signaling can modulate gene transcription and thus affect FIGURE 8. Isoproterenol is capable of activating Tyr 705 and Ser 727 STAT3 phosphorylations in a time-dependent and PTX-insensitive manner mediated through ␤ 2 AR. A, HEK 293 cells were challenged by isoproterenol (10 M) for different time intervals. B, to test the specificity of ␤ 2 AR to induce STAT3 activations by isoproterenol, HEK 293 cells were either stimulated with salbutamol (Sal, 10 M) or co-treated with ICI 118551 (ICI, 1 mM) and isoproterenol (Iso, 10 M) for 30 min. To eliminate the involvement of G i proteins, HEK 293 cells were pretreated with PTX (100 ng/ml) overnight before the 30-min isoproterenol exposure. Cell lysates were immunoblotted for STAT3 phosphorylations and analyzed as described in Fig. 1A. Numerical values shown above the immunoreactive bands represent relative intensities of agonistinduced STAT3 phosphorylations expressed as a ratio of the basal level (set as 1.0). *, agonist-induced STAT3 phosphorylations were significantly higher than the basal value (one-way ANOVA with Dunnett's post-tests, p Ͻ 0.05). Data shown represent the mean Ϯ S.E. from three separate experiments, and the immunoblots shown represent one of three sets; two other sets yielded similar results. Cell lysates were immunoblotted for STAT3 phosphorylations and analyzed as described in Fig. 1A. Numerical values shown above the immunoreactive bands represent relative intensities of agonist-induced STAT3 phosphorylations expressed as a ratio of the basal level (set as 1.0). *, agonist-induced STAT3 phosphorylations were significantly higher than the basal value (one-way ANOVA with Dunnett's post-tests, p Ͻ 0.05). Data shown represent the mean Ϯ S.E. from three separate experiments, and the immunoblots shown represent one of three sets; two other sets yielded similar results. numerous biological processes. Among the many transcription factors identified to date, the STAT proteins are the most well established in terms of their regulation by G proteins. Members of the G i/o , G q , and G 12 families have been reported to stimulate STAT3 Tyr 705 and Ser 727 phosphorylations through several common signaling intermediates such as Src and MAPKs (26, 27,46,47). Despite the fact that a number of G s -coupled receptors have been reported to activate STAT3 (28,29,48), very little is known with regard to their mechanism of action. Moreover, because many GPCRs can activate multiple G proteins simultaneously, there remains a possibility of G␣ s -independent activation of STAT3 (e.g. via G␤␥ or other G␣ subunits) by these receptors. In this study, we have unequivocally demonstrated that activated G␣ s is indeed fully capable of stimulating STAT3 phosphorylations as well as STAT3-driven gene expression. This conclusion is based on several lines of evidence. 1) Expression of G␣ s QL in HEK 293 cells led to STAT3 phosphorylations and transcriptional activation. 2) Manipulation of G␣ s -dependent signals can stimulate or inhibit STAT3 activity accordingly. 3) Stimulation of endogenous G s -coupled ␤ 2 AR resulted in STAT3 activation, and the sensitivities of STAT3 responses to specific inhibitors were identical to those obtained with G␣ s QL. 4) Knockdown of G␣ s expression by RNAi effectively abolished the receptor-induced STAT3 responses. More interestingly, our study revealed that the G␣ s -mediated STAT3 regulations involve a complex mechanism uniquely requiring PKA, PI3K, and JNK.
Previous studies on the regulation of STAT3 by members of the G␣ q proteins (26, 27) have indicated that the PLC␤/PKC/CaMKII cascade plays a crucial role in stimulating STAT3. Because the primary signal generated upon activation of G␣ s is cAMP instead of inositol 1,4,5-trisphosphate/Ca 2ϩ , it is hardly surprising that G␣ s -mediated STAT3 phosphorylations employ components of the cAMP pathway (Fig. 1A) and are refractory to the inhibition of PLC␤ (Figs. 1D,  6A, and 10D). We have used a number of approaches to verify that activation of the AC/PKA pathway can lead to STAT3 Tyr 705 and Ser 727 phosphorylations as well as its transcriptional activation in HEK 293 cells. These include stimulation of endogenous G␣ s by CTX (Fig. 1C) or ␤ 2 AR (Fig. 8A), direct activation of AC by Fsk, mimicking cAMP actions with Bt 2 cAMP and (S p )-cAMP (Fig. 1C), and inhibiting PKA by H-89 and (R p )-cAMP (Fig. 1A). These results illustrate that the G␣ s /AC/PKA pathway is indeed capable of activating STAT3 and may provide the molecular basis for the regulation of STAT3 by G s -coupled receptors such as luteinizing hormone receptor (49) and ␤-AR (32).
Extensive studies have been performed on G protein-mediated STAT3 regulations where the involvement of ERK and c-Src/JAK signaling pathways are evident (26,27). In G␣ s -mediated STAT3 regulation, PKA is likely to be involved in the activation of STAT3 because inhibition of PKA suppressed the phosphorylation of STAT3 at both Tyr 705 and Ser 727 (Fig. 1). Indeed, PKA appears to play a pivotal role in STAT3 regulation because of its ability to modulate various signaling cascades. Blockade of PKA resulted in the attenuation of ERK, JNK, c-Src, and JAK2 phosphorylations (Fig. 7A), signifying that PKA is a key control for several downstream cascades important for G␣ s -induced STAT3 regulations.
Despite their primary linkage to different effector pathways, G␣ s and G␣ 16 share some common features in activating STAT3. As in G␣ 16 -induced STAT3 activation, the participation of c-Src/JAK pathway also appears to be important in G␣ smediated STAT3 activations. G␣ s has been reported to stimulate the activity of c-Src (50), whereas in terms of signaling intermediates, c-Src and JAKs are well known regulators of STAT3 (51). It is therefore conceivable that c-Src/JAK acts as a site for the integration of signals from GPCRs to STAT3. Our findings support the participation of c-Src and JAK in G␣ sinduced STAT3 phosphorylations because the application of selective inhibitors of c-Src (PP1 and PP2) and JAK2/3 (AG490 and WHI-P131) (Fig. 3A) as well as the expression of c-SrcDN ( Fig. 3B and 6C) suppress these responses. We further delineated that c-Src appears to be upstream of JAK2, as illustrated by the ability of c-Src inhibitors (Fig. 3C) or overexpression of c-SrcDN (Fig. 3D) to inhibit G␣ s QL-induced JAK2 phosphorylation. Yet, unlike C5a/G␣ 16 -induced responses (26), the G␣ s QL-or isoproterenol-induced STAT3 Tyr 705 phosphorylation was not affected upon inhibition of c-Src (Fig. 3, A and B,  and 10B). The insensitivity of G␣ s -linked STAT3 Tyr 705 phosphorylation to c-Src inhibitors suggests that activation of G s may trigger an alternative route for the phosphorylation of STAT3 at Tyr 705 by a tyrosine kinase other than c-Src. Indeed, Etk, a Bruton tyrosine kinase, was reported to directly regulate STAT3 activation through G␣ q (52), whereas the cAMP/PKA cascade is able to modulate spleen tyrosine kinase (53). Nevertheless, much remain to be elucidated, especially the identity and specificity of JAK isoforms. Apart from JAK2/3 (as illustrated in the present study), other JAKs may also be involved in the regulation of STAT3.
A large body of evidence has illustrated that PKA can activate MAPKs (54) such as ERKs which are crucial in STAT3 regulations. The importance of MEK/ERK in the regulation of the Ser 727 STAT3 phosphorylation has long been suggested (23), and this regulation appears to be equally critical in G proteinmediated STAT3 activation (26,27,46). At least two different routes are available for the activation of MEK/ERK by PKA (Fig.  11). First, MEK/ERK activations by PKA can be mediated through a small G protein named Rap1 (55). Rap1 is stimulated by PKA and consequently leads to the excitation of Raf-1 kinase that is directly upstream of the MEK/ERK pathway. Another route to ERK activation involves Ras. Its involvement was demonstrated by the overexpression of RasDN that attenuated G␣ s QL-mediated ERK phosphorylations (Fig. 2D). As observed for the G s -coupled serotonin 5-HT 7 receptor (40), the activation of PKA promotes Ras activation and subsequent MEK/ ERK stimulation via the recruitment of a GEF named Ras-GRF1. Our findings confirm the involvement of the MEK/ERK pathway in G s -mediated STAT3 responses. However, unlike the G 16 -mediated STAT3 stimulations (26), the G␣ s QL-or isoproterenol-induced STAT3 Tyr 705 phosphorylation is insensitive to inhibition of MEK1/2 by U0126 ( Figs. 2A and 10A) or to blockade by RasDN (Fig. 2C). The Ras/Raf/MEK/ERK pathway thus appears to selectively modulate the Ser 727 site on STAT3. Because the precise mechanism by which MEK/ERK mediates STAT3 Tyr 705 phosphorylation has not been elucidated, it is difficult to envisage how G␣ s and G␣ 16 signals can produce different degrees of dependence on MEK/ERK. Rac1 represents another monomeric GTPase that can regulate STAT3 signaling. This monomeric GTPase is known to activate JNK through mitogen-activated kinase kinase 4/7 (MKK4/7; like MEK in ERK pathway) (56); indeed, overexpression of Rac1DN significantly attenuated the activity of JNK as illustrated by the suppression of c-Jun phosphorylation (Fig.  2D). As compared with Rac1-dependent but JNK-independent G␣ 16 -induced STAT3 activations, remarkably, our results show that both Rac1 and its downstream JNK signaling cascade play significant regulatory roles in G s -mediated STAT3 phosphorylations (Fig. 2, A and C, and 10A) and transcriptional acti- vation (Fig. 6, A and B); expression of the Rac1DN as well as inhibition of JNK by SP600125 abolished the G␣ s QL-or isoproterenol-induced STAT3 phosphorylations. Rac1 is able to directly bind STAT3. The expression of the activated Rac1 stimulated STAT3 phosphorylation at both Tyr 705 and Ser 727 residues (57); nevertheless, it remains to be determined whether this activation is mediated through direct phosphorylation because indirect activation of STAT3 by Rac1 has also been suggested (58). The participation of JNKs in the modulation of Ser 727 STAT3 phosphorylation upon cytokine induction has been demonstrated previously (59), whereas STAT3 Tyr 705 phosphorylation can be regulated through JAK2 as the expression of either activated Rac1 (60) or constitutively active JNKK (JNKK-CA) is sufficient to drive JAK2 stimulation (Fig. 3C). This specific requirement of JNKs in G s -mediated STAT3 activations, but not in G 16 -mediated responses, might result from the distinctive activation of the AC/PKA pathway (Fig. 11). The ability of the G s -coupled parathyroid hormone receptor to stimulate Rac1 in opossum kidney cells was documented more than a decade ago (61), whereas Rac1 regulation by PKA required for the growth factor-stimulated migration of carcinoma cells was also demonstrated (62). Yet the mechanism has only recently been elucidated to also require the Ras-GRF1. Interestingly, this Ras-activating GEF contains two separate structural domains for stimulating Ras and Rac1 (41). Indeed, the ability of Rac1DN to inhibit STAT3 phosphorylations at both Tyr 705 and Ser 727 (Fig. 2C) points to the involvement of Rac1 in G␣ s QL-induced STAT3 activation. However, it should be noted that G␣ 16 QL-induced STAT3 activations require Rac1 but not JNK (26). It is not clear as to why JNK is distinctively required for G␣ s QL-induced STAT3 activation, when Rac1 is apparently activated in both G␣ s QL-and G␣ 16 QL-induced responses.
Recently, the role of Ras-GRF1 has been suggested to be crucial in regulating monomeric GTPases through different structural domains (41). In line with the established studies that both Ras (40) and Rac1 (63) are modulated by this GEF, ERK and JNK (downstream of Ras and Rac1, respectively) were inhibited upon the knockdown of endogenous Ras-GRF1 in HEK 293 cells by the RNAi technique (Fig. 7B). Furthermore, c-Src also participates in the Ras-GRF1 regulation in that its phosphorylation by c-Src is required for Rac1 induction (63). Hence, as Ras-GRF1 is located downstream of c-Src, it is not surprising that its reduction had negligible effect on c-Src and JAK2 phosphorylations (Fig. 7B). It is also important to note that only a partial inhibition of G␣ s QL-induced STAT3 phosphorylations was observed upon the introduction of siRas-GRF1. This may be due to either the incomplete removal of Ras-GRF1 by RNAi or the involvement of other parallel signaling cascades such as the Ras-GRF1-independent c-Src/JAK pathway (Fig. 11).
Another noteworthy observation relates to the apparent involvement of PI3K in G␣ s -mediated STAT3 phosphorylation in HEK 293 cells. The G s -induced Tyr 705 and Ser 727 STAT3 phosphorylations and transcriptional activity are abrogated upon inhibition of PI3K by wortmannin or LY294002 (Figs. 5A, 6A, and 10C), or by overexpression of PI3KDN ( Fig. 5B and 6C). This is distinctively different from those of G␣ 16 QL-(26), G␣ 14 QL- (27), or G i -mediated (46) STAT3 phosphorylations, wherein the same approaches failed to implicate PI3K. PI3K has been shown to play a critical role in mediating STAT3 Ser 727 phosphorylation (64), possibly via the cross-talk between the ERK signaling cascade as blockade of PI3K by wortmannin, LY294002, or PI3KDN produced partial but significant attenuation of the phosphorylated ERKs (Fig. 5). Other possible involvement, for instance the downstream effectors Akt/ mTOR of PI3K, has also been suggested (65). However, little is known as to how PI3K regulates STAT3 phosphorylation at Tyr 705 site. Although it is generally accepted that PI3K can be activated by G␤␥ subunits (39), our results illustrate that the expression of G␣ s QL alone is fully capable of eliciting PI3Ksensitive STAT3 phosphorylations. In fact, there is considerable literature available to support a linkage between AC/PKA and PI3K. First, PI3K can be stimulated by cAMP-induced Ras activation demonstrated in oxidized phospholipid-induced endothelial cell/monocyte binding (66). Second, stimulation of PKA by G s -coupled thyrotropin receptor has been shown to promote the formation of the PI3K-Ras complex and the subsequent activation of PI3K in thyroid FRTL-5 cells (67). In neutrophils, the cytosolic regulator of AC that binds PI3K products via a pleckstrin homology domain is essential for chemoattractant-mediated activation of AC (68). Finally, PI3K can serve as a scaffolding protein and contribute to the control of cAMP levels in the cardiac system (69). Collectively, these studies illustrate that elaborate cross-talk interactions exist between the AC/PKA and PI3K pathways, some of which may form part of the circuitry in G s -mediated activation of STAT3. Additional studies are required to discern their functional relationship.
A notable feature of G s -induced changes in STAT3 phosphorylations is the relative resistance of Tyr 705 to suppression by the kinase inhibitors. It is generally believed that Tyr 705 phosphorylation is required for STAT3 transcriptional activity, whereas Ser 727 phosphorylation enhances transcriptional activity (14). Hence, blockade of Ras, c-Src, and MEK signals would be expected to partially inhibit G␣ s QL-induced STAT3 transcriptional activity (Fig. 6), as opposed to complete suppressions seen with G␣ 16 QL-or G␣ 14 QL-induced activity (26, 27). Our results do conform to such a postulation. In depth experimentation and analysis are required to fully appreciate the significance of differential phosphorylation of STAT3 on its transcriptional activity.
Based on the known signaling properties and abilities of various signaling components to regulate their respective targets, we have generated a mechanistic model for G s -mediated STAT3 phosphorylation (Fig. 11). Given that the production of cAMP and stimulation of PKA are direct consequences of G␣ s activation, they may serve as a point of divergence for the regulation of other signaling cascades that are pertinent for the stimulation of STAT3 by G s . The involvement of Ras-GFR1 can lead to the stimulation of MEK/ERK and JNK cascades via Ras and Rac1. A number of signaling molecules are also capable of stimulating ERK and can thus induce STAT3 Ser 727 phosphorylation; they include Rac1 (Fig. 2D) (12), c-Src (Fig. 3), and PI3K ( Fig. 4) (70). JAK2-mediated STAT3 Tyr 705 phosphorylation, on the other hand, can be elicited via c-Src or JNK (Fig. 3,  B and C). It should also be noted that the proposed models are simplistic views of the actual signaling networks because paral-lel and alternative pathways may well be involved, for instance the regulation of Ras by Rac1 (71) and the release of G␤␥ to modulate Src activity (72) and PI3K (39). In summary, this study has highlighted the role of G s in regulating STAT3 activity and provided a rudimentary network for the possible participating signaling. However, a number of details remain to be determined. The most outstanding question lies in what other tyrosine kinase participates in the phosphorylation of STAT3 Tyr 705 and provides alternative routes for stimulation when the MEK and c-Src pathways are blocked. Because such an alternative pathway does not appear to be engaged in G q -and G imediated STAT3 activation (both STAT3 Tyr 705 and Ser 727 phosphorylations are abolished upon inhibition of MEK and c-Src (26, 27, 46)), presumably it is evoked by the AC/PKA signals. A search for novel signaling molecules along the AC/PKA pathway may provide some useful clues. Nevertheless, this study has provided conclusive evidence that G s can indeed transduce GPCR signals to regulate STAT3 activity.