G(i)-dependent activation of c-Jun N-terminal kinase in human embryonal kidney 293 cells.

Heterotrimeric G proteins stimulate the activities of two stress-activated protein kinases, c-Jun N-terminal kinase (JNK) and p38 mitogen-activated protein kinase in mammalian cells. In this study, we examined whether alpha subunits of G(i) family activate JNK using transient expression system in human embryonal kidney 293 cells. Constitutively activated mutants of Galpha(i1), Galpha(i2), and Galpha(i3) increased JNK activity. In contrast, constitutively activated Galpha(o) and Galpha(z) mutants did not stimulate JNK activity. To examine the mechanism of JNK activation by Galpha(i), kinase-deficient mutants of mitogen-activated protein kinase kinase 4 (MKK4) and 7 (MKK7), which are known to be JNK activators, were transfected into the cells. However, Galpha(i)-induced JNK activation was not blocked effectively by kinase-deficient MKK4 and MKK7. In addition, activated Galpha(i) mutant failed to stimulate MKK4 and MKK7 activities. Furthermore, JNK activation by Galpha(i) was inhibited by dominant-negative Rho and Cdc42 and tyrosine kinase inhibitors, but not dominant-negative Rac and phosphatidylinositol 3-kinase inhibitors. These results indicate that Galpha(i) regulates JNK activity dependent on small GTPases Rho and Cdc42 and on tyrosine kinase but not on MKK4 and MKK7.

Several lines of evidence suggest that G protein-coupled receptors stimulate the ERK pathway through some G protein subunits in various cells (9,10). It is likely that G i -dependent ERK activation is mediated primarily by G␤␥ (9,10). G␤␥ directly activates phosphatidylinositol 3-kinase ␥ and ␤-adrenergic receptor kinase 1, resulting in an increase of activities of Src family tyrosine kinases (9 -14). Tyrosine-phosphorylated Shc permits the translocation of the Grb2-Sos complex to plasma membranes, leading to the promotion of the GDP-GTP exchange on Ras. Ras regulates the activity of Raf, which induces the activation of MEK1 and -2 and subsequently ERK.
Some G protein-coupled receptors are able to activate JNK in certain types of cells (9). JNK is activated by agonist stimulation of G i -coupled receptors in NIH3T3, Rat-1, and COS-7 cells (15)(16)(17). It has been reported that JNK activation by G i -coupled m2 muscarinic acetylcholine receptor is mediated mainly by G␤␥ in COS-7 cells (17). Small GTPases Ras and Rac and phosphatidylinositol 3-kinase ␥ are involved in this G␤␥-induced JNK activation (17,18). In the course of studying G i -dependent JNK activation in human embryonal kidney (HEK) 293 cells, we found that its activation was mediated by both G␣ i and G␤␥. Here we show that constitutively activated G␣ i2 mutant as well as G␤␥ (19) stimulates JNK activity. Furthermore, we investigate whether JNK kinases, Rho family GTPases, tyrosine kinases, and phosphatidylinositol 3-kinases participate in this signal transduction pathway.

EXPERIMENTAL PROCEDURES
Materials-Mastoparan and pertussis toxin were purchased from Calbiochem-Novabiochem Co. and Kaken Pharmaceutical Co., respectively. Tyrosine kinase inhibitors PP1/AG1872 and PP2/AG1879 were kindly provided by A. Levitzki (Hebrew University). Phosphatidylinositol 3-kinase inhibitors wortmannin and LY294002 were purchased from Calbiochem-Novabiochem Co. and BIOMOL, respectively. Mouse monoclonal antibodies M2, 12CA5, and 9E10 against FLAG, HA, and Myc epitopes were obtained from Eastman Kodak Co., Roche Molecular Biochemicals, and Babco, respectively. Mouse monoclonal antibody B-14 against Schistosoma japonicum GST was purchased from Santa Cruz Biotechnology, Inc. Rabbit polyclonal antibodies T-20 and 06-238 against G␤ were obtained from Santa Cruz Biotechnology and Upstate Biotechnology, Inc., respectively. Rabbit polyclonal antibodies AS/7, EC/2, and GC/2 against G␣ i1/2 , G␣ i3 , and G␣ o , respectively, were purchased from NEN Life Science Products, Inc. Rabbit polyclonal antibodies against G␣ i1/2 and G␣ z (X264) were generously provided by T. Asano (Aichi Human Service Center) and P. C. Sternweis (University of Texas Southwestern Medical Center), respectively. Rabbit polyclonal anti-Csk antibody C-20 was purchased from Santa Cruz Biotechnology. Goat anti-mouse and anti-rabbit IgG antibodies conjugated with horseradish peroxidase were obtained from NEN Life Science Products.
Transfection-Plasmid DNAs were transfected into HEK 293 cells by the calcium phosphate precipitation method. The final amount of the transfected DNA for a 60-mm dish was adjusted to 25 g by empty vector, pCMV. Three g of SR␣-HA-JNK1, SR␣-HA-ERK2, pCMV-GST-MKK4, pCMV-GST-MKK7, or pCMV-GST-MKK7␤ was cotransfected with 3 g of pCMV-Myc-␤ARK1ct, 10 g of each G␣ wild type or QL mutant plasmid, 5 g of pCMV-G␤ 1 , and 5 g of pCMV-G␥ 2 , 10 g of pCMV-FLAG-MKK4K95R, 10 g of pCMV-FLAG-MKK7K63R, 10 g of dominant negative Rho family plasmid, 10 g of pEF-BOS-C3 toxin, 10 g of pCMV-Myc-Pak1CRIB, or 3 g of pCMV-Csk. The medium was replaced 24 h after transfection, and the cells were starved in the serum-free medium containing 1 mg/ml bovine serum albumin (Nacalai) for 24 h.
GST-c-Jun was eluted with column buffer B containing 10 mM glutathione. Hexahistidine-JNK was dialyzed against column buffer B containing 200 mM NaCl and stored at Ϫ80°C until use. Under these storage conditions, hexahistidine-JNK retained the catalytic activity within at least 6 months. The eluate of GST-c-Jun and Trx-c-Jun was dialyzed against column buffer B and stored at Ϫ80°C until use.
Kinase Assays-After 24 h of serum starvation, the cells transfected with SR␣-HA-JNK1, SR␣-HA-ERK2, pCMV-GST-MKK4, pCMV-GST-MKK7, or pCMV-GST-MKK7␤ were lysed in 600 ml of lysis buffer A (20 mM HEPES-NaOH (pH 7.5), 3 mM MgCl 2 , 100 mM NaCl, 1 mM dithiothreitol, 1 mM phenylmethanesulfonyl fluoride, 1 g/ml leupeptin, 1 mM EGTA, 1 mM Na 3 VO 4 , 10 mM NaF, 20 mM ␤-glycerophosphate, and 0.5% Nonidet P-40) on ice. The lysates were centrifuged at 14,000 rpm for 10 min at 4°C. For JNK and ERK assay, aliquots (500 g) of the supernatants were mixed with protein A-Sepharose CL-4B (Amersham Pharmacia Biotech) preabsorbed with a mouse anti-HA antibody for 12 h at 4°C. The immune complexes were washed twice with lysis buffer A and twice with reaction buffer A (20 mM HEPES-NaOH (pH 7.5), 10 mM MgCl 2 , 0.1 mM phenylmethanesulfonyl fluoride, 0.1 g/ml leupeptin, 0.1 mM EGTA, 10 mM Na 3 VO 4 , and 2 mM ␤-glycerophosphate) and incubated in 30 l of reaction buffer A containing 3 g of GST-c-Jun for JNK assay or 5 g of myelin basic protein for ERK assay, 20 M ATP, and 5 Ci of [␥-32 P]ATP (Amersham Pharmacia Biotech) at 30°C for 10 min. For MKK4, MKK7, or MKK7␤ assay, aliquots (500 g) of the supernatants were mixed with glutathione-Sepharose 4B for 12 h at 4°C and centrifuged. The precipitate was washed with lysis buffer A and with reaction buffer A and was incubated in 30 l of reaction buffer A containing 2 g of hexahistidine-JNK, 10 g of Trx-c-Jun, 20 M ATP, and 5 Ci of [␥-32 P]ATP at 30°C for 20 min. The reaction was stopped by adding 10 l of 4 ϫ Laemmli sample buffer and boiling, and the sample was subjected to SDS-polyacrylamide gel electrophoresis. The radioactivity incorporated into GST-c-Jun, Trx-c-Jun, and myelin basic protein was measured by an imaging analyzer (FUJI BAS 2000) and detected by autoradiography.
Immunoprecipitation and Immunoblotting-Aliquots (250 g) of cell lysates were mixed with protein A-Sepharose CL-4B preabsorbed with a mouse anti-Myc antibody for 12 h at 4°C. The immune complexes were precipitated by centrifugation and washed four times with lysis buffer A. Aliquots of cell lysates and immune complexes were boiled in Laemmli sample buffer. The boiled samples were separated by SDS- polyacrylamide gel electrophoresis, and the proteins were transferred to nitrocellulose membranes (BA85; Schneider & Schnell). After the membranes were blocked with phosphate-buffered saline containing 0.1% Tween 20 and 5 mg/ml bovine serum albumin, the separated proteins were immunoblotted with various antibodies. The bound antibodies were detected using anti-rabbit or mouse IgG antibody conjugated with horseradish peroxidase.

Mastoparan-induced JNK Activation Is Mediated by both G␣ i and G␤␥ in HEK 293
Cells-We introduced plasmids encoding HA-tagged JNK with various cDNAs into HEK 293 cells. Using anti-HA antibody, the epitope-tagged JNK was immunoprecipitated from lysates of the transfected cells. The JNK activity was assessed as the radioactivity incorporated into recombinant GST-c-Jun. The expression level of HA-JNK was confirmed by immunoblotting in each experiment to compare the transfection efficiency (see Figs. 1-7). To examine G i -dependent JNK activation, we searched an endogenous G icoupled receptor in HEK 293 cells. However, we could not find a candidate for the receptor suitable to this study. Then we transfected a plasmid encoding G i -coupled m2 muscarinic acetylcholine receptor into the cells, but JNK was activated only weakly with an agonist stimulation. The reason may be that the ectopic expression level of m2 muscarinic acetylcholine receptor is not so high (24). As shown in Fig. 1A, mastoparan, a peptide that directly activates G proteins by a mimic of G protein-coupled receptor (31), stimulated the activity of JNK by approximately 4-fold.
Next, we investigated whether G i is involved in mastoparaninduced JNK activation. The cells were treated with pertussis toxin, which ADP-ribosylates G i /G o and inhibits the coupling of G i /G o with the receptors, for 24 h before the addition of mastoparan (Fig. 1A). The activation was almost completely inhibited by the pretreatment of pertussis toxin. Since there is no G o in HEK 293 cells as described below, the inhibition by pertussis toxin indicated that mastoparan increases JNK activity via G i .
To determine whether mastoparan-induced JNK activation is mediated by G␣ i and/or G␤␥, a plasmid encoding carboxylterminal peptide of ␤-adrenergic receptor kinase 1 (␤ARK1) was cotransfected. It has been shown that ␤ARK1ct associates with G␤␥ and inhibits G␤␥-mediated ERK and JNK activations by a G protein-coupled receptor (11,17). Mastoparan-induced JNK activation was reduced approximately 50% by cotransfection of ␤ARK1ct (Fig. 1B), suggesting that the JNK activation is mediated by both G␣ i and G␤␥ in HEK 293 cells.
G␣ i2 Q205L Stimulates the Activity of JNK in HEK 293 Cells-We explored which ␣ subunit of the G i family increases JNK activity. As shown in Fig. 2A, constitutively activated mutants of G␣ i1 , G␣ i2 , and G␣ i3 stimulated JNK activity by approximately 3.5-, 5-, and 3.5-fold, respectively. On the other hand, constitutively activated G␣ o and G␣ z mutants did not stimulate JNK activity. The activation of JNK by G␣i2Q205L was comparable with that by G␤␥ (Fig. 2A). The expression of endogenous G␣ o and G␣ z in HEK 293 cells was not detected by immunoblotting ( Fig. 2A).
To confirm that the inhibitory effect of ␤ARK1ct on mastoparan-induced JNK activation results from the sequestration of G␤␥, ␤ARK1ct was cotransfected with G␣i2Q205L or G␤␥ (Fig. 2B). The activation of JNK by G␤␥, but not G␣i2Q205L, was blocked completely by cotransfection of ␤ARK1ct.
Mastoparan-induced JNK Activation Is Dependent Partially on MKK4 and MKK7-To clarify whether mastoparan activates JNK through two JNK kinases, MKK4 and MKK7, we cotransfected a plasmid of MKK4K95R or MKK7K63R. MKK4K95R and MKK7K63R are kinase-deficient mutants that sequester upstream components such as MEKK1. Mastoparan-induced JNK activation was suppressed partially by cotransfection of MKK4K95R or MKK7K63R (Fig. 3, A and B).
Next, we analyzed the activities of MKK4 and MKK7. The cells were cotransfected with a plasmid encoding GST-fused MKK4 or MKK7. GST-MKK4 or -MKK7 was precipitated from lysates of the transfected cells using glutathione-Sepharose 4B and incubated with recombinant JNK and c-Jun, and the radioactivity incorporated into c-Jun was measured. Mastoparan activated only weakly MKK4 and MKK7 (Fig. 3, C and D). This result, considered together with results of kinase-deficient mutants, indicates that JNK activation by mastoparan is dependent partially on MKK4 and MKK7.
G␣ i2 Q205L Fails to Activate MKK4 and MKK7-We reported previously that G␤␥ activates JNK mainly through MKK4 and to a lesser extent through MKK7 (19). Because JNK activation by mastoparan was dependent partially on MKK4 and MKK7, we considered that MKK4 and MKK7 might be involved in JNK activation mediated by G␣ i as well as that by G␤␥. However, cotransfection of MKK4K95R failed to attenuate G␣ i2 Q205L-induced JNK activation (Fig. 4A). Furthermore, G␣ i2 Q205L failed to activate MKK4 (Fig. 4C). Although cotransfection of MKK7K63R inhibited partially G␣ i2 Q205L-induced JNK activation, MKK7 activity was not stimulated by G␣ i2 Q205L (Fig. 4, B and D). A human MKK7 gene appears to generate some alternative splicing forms. A MKK7␤ isoform has 86 extra amino acid residues at the N terminus of MKK7 (26) and is expressed mainly in HEK 293 cells (data not shown). We thought that G␣ i might activate a MKK7␤ isoform. However, G␣ i2 Q205L failed to stimulate MKK7␤ activity (Fig. 4E). In contrast, G␤␥ increased the activities of MKK4, MKK7, and MKK7␤ by approximately 6-, 2.5-, and 2-fold, respectively (Fig.  4, C-E). These results indicate that G␣ i2 regulates JNK activity through MKK4-and MKK7-independent pathway. On the other hand, G␤␥ activates JNK through MKK4-and MKK7-dependent pathways (19). It is likely that mastoparan induces MKK4 and MKK7 activation through G␤␥ but not G␣ i . G␣ i2 -mediated JNK Activation Is Dependent on Rho and Cdc42 but Not on Rac-Rho family GTPases have been shown to be implicated in the activation of JNK by various stimuli (32). In addition, JNK activation by G␤␥ depends on Rho family GTPases (17)(18)(19). To test the possibility that Rho family GTPases are involved in the pathway from G␣ i to JNK, we cotransfected each plasmid of dominant-negative Rho family GTPases. Mastoparan-induced JNK activation was inhibited by RhoT19N and Cdc42T17N but not RacT17N (Fig. 5, A-C). In addition, JNK activation by G␣ i2 Q205L was also blocked by the dominant negative mutants of Rho and Cdc42 (Fig. 5, D-F), indicating that G␣ i2 regulates JNK activity through Rho and Cdc42 in HEK 293 cells. To confirm that Rho and Cdc42 are involved in G␣ i2 Q205L-induced JNK activation, we utilized C3 toxin and Pak1CRIB. C3 toxin ADP-ribosylates Rho and inhibits the Rho functions (28). Pak1CRIB is associated with active Rac or Cdc42 and inhibits the interaction of active Rac or Cdc42 with the effectors (29,32). As shown in Fig. 5, G and H, JNK activation by G␣ i2 Q205L was blocked by cotransfection of C3 toxin or Pak1CRIB. These results suggest that Rho and Cdc42 participate in the JNK pathway from G␣ i .
Effect of Tyrosine Kinase Inhibitors on G␣ i -mediated JNK Activation-To investigate whether tyrosine kinases are involved in G i -mediated activation of JNK, we used PP1 and PP2, which are tyrosine kinase inhibitors preferential for Src family tyrosine kinases (33), and Csk, which is a negative regulator of Src family tyrosine kinases (30). As shown in Fig. 6A, mastoparan-induced JNK activation was attenuated by treatment with PP1. Moreover, the activation of JNK by G␣ i2 Q205L was inhibited by PP1 and PP2 in a dose-dependent manner (Fig. 6, B and C). Next, a plasmid encoding cDNA of Csk was transfected. As shown in Fig. 6, D and E, cotransfection of Csk suppressed the JNK activation by G␣ i2 Q205L, but not by G␤␥. We further examined whether G␣ i2 Q205L-induced JNK activation is inhibited by dominant-negative Fyn. Cotransfection of dominant-negative Fyn resulted in a 60% decrease of JNK activation by G␣ i2 Q205L (data not shown). These results indicate the involvement of Src family tyrosine kinases in G␣ imediated JNK activation.
Effect of Phosphatidylinositol 3-Kinase Inhibitors on G␣ imediated JNK Activation-It has been shown that G␣ i1 directly activates phosphatidylinositol 3-kinase ␥ in vitro (34), and overexpression of this enzyme in the cells induces the activation of JNK (18). To examine the involvement of phosphatidylinositol 3-kinase in the JNK pathway from G␣ i , the transfected cells were treated with wortmannin and LY294002, phosphatidylinositol 3-kinase inhibitors (13,18). Wortmannin inhibited mastoparan-induced activation of ERK, but not JNK, as shown in Fig. 7, A-D. Furthermore, neither wortmannin nor LY294002 inhibited G␣ i2 Q205L-induced JNK activation (Fig.  7, E and F), indicating that phosphatidylinositol 3-kinase is not implicated in G␣ i -mediated JNK activation in HEK 293 cells. DISCUSSION JNK activation induced by constitutively activated mutant of G␣ i2 has been shown in two systems including G␣ i2 Q205Loverexpressing mice (35) and Rat-1 cells inducibly expressing G␣ i2 Q205L (36), although the mechanism by which G␣ i2 regulates JNK activity remained to be characterized. In the present study, we first found that mastoparan-induced JNK activation was dependent on G i and mediated by both G␣ i and G␤␥ in HEK 293 cells. In addition, constitutively activated mutants of G␣ i1 , G␣ i2 , and G␣ i3 induced JNK activation. But JNK was not stimulated by activated G␣ o and G␣ z mutants. Moreover, G␣ i2 appeared to regulate JNK activity through a MKK4-and MKK7-independent pathway. Furthermore, JNK activation by G␣ i2 was dependent on small GTPases Rho and Cdc42 and Src family tyrosine kinases but not on phosphatidylinositol 3-kinase.
Two JNK kinase genes, MKK4 and MKK7, have been cloned to date. We showed previously that G␤␥ activates JNK via mainly MKK4 and to a lesser extent MKK7 (19). In contrast, JNK activation by G␣ i2 Q205L was not inhibited effectively by cotransfection of kinase-deficient MKK4 and MKK7 (Fig. 4). Moreover, transfection of G␣ i2 Q205L into the cells failed to stimulate the activities of MKK4, MKK7, and MKK7␤, an alternative splicing form of MKK7 (Fig. 4). It has been reported that there is a JNK kinase other than MKK4 and MKK7 at the level of the fractionation by column chromatography with lysates of stress-stimulated cells (37). G␣ i2 may regulate JNK activity through an unidentified JNK kinase.
A Rac and Cdc42-binding region is found in MAPKKKs including MEKK1, MEKK4, MLK2, and MLK3. MEKK1 activates primarily JNK pathway and appears to mediate JNK activation by Rac or Cdc42 (38). It has been shown that MEKK1 and ASK1 mediate JNK activation induced by G␣ 12 and G␣ 13 in COS-7 and HEK 293 cells (39). Additionally, Collins et al. (40) have reported that G␣ 12 -induced JNK activation partially involves Cdc42 in HEK 293 cells. It appears that a Cdc42-MEKK1 signaling unit functions upstream of JNK. Because G␣ i2 -induced JNK activation depended on Cdc42 (Fig. 5), MEKK1 might be a candidate of MAPKKK in the JNK pathway from G␣ i . JNK activation by G␣ i depended on Rho (Fig. 5). Although a MAPKKK regulated by Rho has not yet been identified, it is expected that there may exist a MAPK cascade linking Rho with a transcription factor SRF in c-fos promoter activation induced by lysophosphatidic acid, a G i -coupled receptor ligand (41).
G␣ i2 Q205L-induced JNK activation was inhibited by tyrosine kinase inhibitors PP1 and PP2 in a dose-dependent manner (Fig. 6). These inhibitors preferentially inhibit Src family tyrosine kinases, and the IC 50 value of PP2 for the inhibition of Src is 15 M in intact cells. 2 Furthermore, cotransfection of Csk attenuated G␣ i2 Q205L-induced JNK activation (Fig. 6). Src family tyrosine kinases are likely to be involved in the signaling pathway from G␣ i to JNK.
We reported previously that G␤␥ stimulates MKK4 activity in a Rho-and Cdc42-dependent manner (19). In contrast, G␣ i2 Q205L failed to activate MKK4, but activated JNK in a Rho-and Cdc42-dependent manner. This difference may be due to the difference of downstream signaling components of G␣ i and G␤␥. It must be noted that JNK activation by G␣ i2 Q205L, but not by G␤␥, was reduced by Csk (Fig. 6) and dominantnegative Fyn. These results suggest that the relationships among tyrosine kinases, Rho, and Cdc42 may be more complex than a single sequential cascade. Fig. 8 shows a proposed pathway from G i to JNK. In the present study, we used mastoparan as an activator of G i . JNK activation by mastoparan, as well as G␣ i2 Q205L and G␤␥ (19), was almost completely inhibited by dominant negative mutants of Rho and Cdc42 (Fig. 5). Moreover, mastoparan-induced JNK activation was blocked partially by Src family inhibitor PP1 and the kinase-deficient mutant of MKK4 (Figs. 3 and 6). These results are consistent with the model in which G␣ i and G␤␥ participate equivalently in the signaling pathway from G i to JNK.
Activating mutations of G␣ i2 , which were denoted as the gip2 oncogene, have been found in some types of tumor including ovarian sex cord stromal tumors, adrenal cortex tumors, and nonfunctional pituitary tumors (42). In addition, ectopic expression of the gip2 oncogene has been shown to cause oncogenic transformation of Rat-1 fibroblast cells. However, it appears that the gip2 oncogene fails to transform other fibroblast 2 A. Levitzki, personal communication. cells such as NIH3T3 (42). Recently, we reported that conditional expression of active G␣ i2 mutant in Rat-1 cells induces the colony formation on soft agar (36). We found that JNK is activated with the expression of active G␣ i2 mutant in the cells (36). On the other hand, JNK activity was not stimulated in NIH3T3 cells expressing active G␣ i2 mutant (43). Further elucidation of the G␣ i2 -JNK pathway may clarify the relationship between the activation of JNK and the mechanism of oncogenic transformation by the gip2 oncogene.