Signaling from G Protein-coupled Receptors to c-Jun Kinase Involves (cid:98)(cid:103) Subunits of Heterotrimeric G Proteins Acting on a Ras and Rac1-dependent Pathway*

Stimulation of a variety of cell surface receptors en-hances the enzymatic activity of mitogen-activated protein kinases (MAPKs). MAPKs have been classified in three subfamilies: extracellular signal-regulated kinases (ERKs), stress-activated protein kinases or c-Jun NH 2 -terminal kinases (SAPKs/JNKs), and p38 kinase. Whereas the pathway linking cell surface receptors to ERKs has been partially elucidated, the mechanism of activation of JNKs is still poorly understood. Recently, we have shown that stimulation of G protein-coupled receptors can effectively induce JNK in NIH 3T3 cells (Coso, O. A., Chiariello, M., Kalinec, G., Kyriakis, J. M., Woodgett, J., and Gutkind, J. S. (1995) J. Biol. Chem. 270, 5620–5624). In the present study, we have used the transient expression in COS-7 cells of m1 and m2 muscarinic receptors (mAChRs) as a model system to study the signaling pathway linking G protein-coupled receptors to JNK. We show that stimulation of either muscarinic receptor subtype leads to JNK activation; however, this effect was not mimicked by expression of activated forms of (cid:97) s , (cid:97) i2 , (cid:97) q , or (cid:97) 13 G protein (cid:97) subunits. In con- trast, overexpression of cells transfected with an expression vector for an epitope-tagged JNK (pcDNA3-HA-JNK) was determined upon immunoprecipitation with the anti-HA-specific monoclonal antibody 12CA5 (Babco) using bacte- rially expressed GST-ATF2(96) fusion protein as a substrate, as described previously (9). The products of the kinase reactions were fractionated in SDS-12% polyacrylamide electrophoresis gels, and ra-dioactivity incorporated into GST-ATF2(96) was determined with the use of a PhosphorImager system (Molecular Dynamics). Parallel immunoprecipitates were analyzed by Western blotting after SDS-polyacryl- amide gel electrophoresis, transferred to nitrocellulose, and immuno-blotted with the anti-HA monoclonal antibody. Immunocomplexes were visualized by enhanced chemiluminescence detection (Amersham Corp.) using goat anti-mouse serum coupled to horseradish peroxidase as a secondary antibody (Cappel).

turn, control the expression of genes that are essential for many cellular processes, including cell growth and differentiation (3). MAPKs have been classified in three subfamilies: extracellular signal regulated kinases (ERKs), including ERK1 and ERK2, also known as p44 mapk and p42 mapk , respectively; stress-activated protein kinases (SAPKs), also termed c-Jun NH 2 -terminal kinases (JNKs); and p38 kinase (2). ERKs phosphorylate and regulate the activity of certain enzymes, including phospholipase A 2 and p90 rsk , and nuclear proteins, such as the ternary complex factor p62 TCF or Elk-1 (4). The latter represents a critical event in controlling the expression of several genes, including c-fos (5). JNKs phosphorylate the aminoterminal transactivating domain of c-Jun and ATF2 (6), thereby increasing their transcriptional activity. p38 is the homologue of the Saccharomyces cerevisiae HOG1 gene, and its function is still unknown, although recently available information suggests that p38 might play a critical role in the inflammatory response (7).
Recent findings have helped to unveil the pathway linking cell surface receptors to ERKs (3). In contrast, the mechanism of activation of JNKs is still poorly understood. JNKs were shown to be activated by a variety of stimuli distinct from those that elevate the enzymatic activity of ERKs, including protein synthesis inhibitors, heat shock, changes in osmolarity, and ultraviolet irradiation (6,8). JNKs can be also activated by agents acting on cell surface receptors, such as tumor necrosis factor-␣, interleukin-1, or epidermal growth factor (8). Furthermore, available evidence suggest that whereas Ras controls the activation of ERKs, members of the Rho family of small GTPbinding proteins, Rac1 and Cdc42, regulate the activity of JNKs (9).
Recently, we have shown that stimulation of certain G protein-coupled receptors expressed in NIH 3T3 cells can effectively induce JNK activity, however, following a time course clearly distinct from that of ERK activation (1). In the present study, we have used the transient expression in COS-7 cells of m1 and m2 muscarinic receptors (mAChRs) as a model system to study the biochemical route connecting G protein-coupled receptors to an epitope-tagged JNK. We present evidence that signaling from m1 and m2 mAChRs to JNK involves ␤␥ subunits of heterotrimeric G proteins, acting on a Ras-and Rac1dependent pathway.

EXPERIMENTAL PROCEDURES
Cell Lines and Transfection-COS-7 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% of fetal bovine serum. Cells were transfected by the DEAE-dextran technique, adjusting the total amount of DNA to 5-10 g/plate with vector DNA (pcDNA3; Invitrogen), when necessary (9).
JNK Assay and Western Blots-JNK activity in lysates from COS-7 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. cells transfected with an expression vector for an epitope-tagged JNK (pcDNA3-HA-JNK) was determined upon immunoprecipitation with the anti-HA-specific monoclonal antibody 12CA5 (Babco) using bacterially expressed GST-ATF2(96) fusion protein as a substrate, as described previously (9). The products of the kinase reactions were fractionated in SDS-12% polyacrylamide electrophoresis gels, and radioactivity incorporated into GST-ATF2(96) was determined with the use of a PhosphorImager system (Molecular Dynamics). Parallel immunoprecipitates were analyzed by Western blotting after SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, and immunoblotted with the anti-HA monoclonal antibody. Immunocomplexes were visualized by enhanced chemiluminescence detection (Amersham Corp.) using goat anti-mouse serum coupled to horseradish peroxidase as a secondary antibody (Cappel).

RESULTS AND DISCUSSION
In order to study the mechanism controlling the activation of JNK by m1 and m2 G protein-coupled receptors, we coexpressed these receptors together with an epitope-tagged JNK (HA-JNK) in COS-7 cells. We observed that when cotransfected with the HA-JNK cDNA, both m1 and m2 were expressed at similar high levels (data not shown), and the epitope-tagged JNK was efficiently expressed and readily detectable upon immunoprecipitation with the anti-HA monoclonal antibody (9). In cells expressing either muscarinic receptor, the cholinergic agonist carbachol induced an increase in JNK activity, as judged by its in vitro phosphorylating activity using GST-ATF2(96) as a substrate (Fig. 1). Whether mediated by m1 or m2 receptors, induction of GST-ATF2(96) phosphorylation showed a peak at approximately 15 min after stimulation, being of 2-4-fold higher than the control with a slow decrease at later times (Fig. 1). Thus, JNK is activated by either m1 or m2 G protein-coupled receptors when expressed in COS-7 cells. Whereas m1 receptors are typical of those coupled through G proteins of the G q family to phospholipase C activation, m2 is known to couple through G i to a number of effector pathways, including to the inhibition of adenylyl cyclases (12). Interestingly, both m1 and m2 mAChRs appear to activate JNK irrespective of their G protein-coupling specificity.
As an approach to investigate which G proteins mediate the activation of JNK, we took advantage of the observation that the expression of GTPase-deficient, mutationally activated forms of G protein ␣ subunits can activate effector pathways by obviating the need for receptor stimulation (13). Thus, we coexpressed the epitope-tagged JNK together with GTPase-deficient mutants for G␣ s , G␣ i2 , G␣ q , and G␣ 13 , which are repre-sentative members for each of the four G␣ subunit families (14). Expression of each G protein ␣ subunit, when transfected into COS-7 cells, could be demonstrated by immunoblotting with subtype-specific antibodies (data not shown; Refs. 11 and 12). However, these activated mutants enhanced JNK activity only to a very limited extent, either when each G␣ subunit was expressed alone (Fig. 2) (15) or in all possible combinations (data not shown). Thus, ␣ subunits of heterotrimeric G proteins might not mediate JNK activation by G protein-coupled receptors.
When activated, receptors linked to G proteins catalyze the replacement of GDP by GTP-bound to the ␣ subunit and induce the dissociation of ␣-GTP from ␤␥ dimers. Although the ␣ subunits were thought to be solely responsible for coupling receptors to second-messenger-generating systems, recent work has established a critical role for ␤␥ dimers in signal transduction (10,14). Thus, the failure of mutationally activated G protein ␣ subunits to elevate JNK activity prompted us to explore whether ␤␥ dimers participate in signaling to JNK. We observed that, when cotransfected, ␤ 1 ␥ 2 subunits induce a remarkable increase in the phosphorylating activity of the epitope-tagged JNK, although expression of the HA-JNK was similar for each transfected cell population (Fig. 2). Similar results were obtained when ␤ 2 ␥ 2 subunits were expressed (data not shown). In contrast, JNK was poorly activated when coexpressed with ␤ 1 or ␥ 2 alone or when cotransfected with ␤ 1 and an altered form of the ␥ 2 subunit, designated ␥ 2 *, that lacks an isoprenylation signal and fails to associate to the plasma membrane (10) (Fig. 2). We conclude that membrane-bound ␤␥ subunits of heterotrimeric G proteins, but not G␣ subunits, can effectively stimulate JNK activity in COS-7 cells.
We next explored a role for ␤␥ complexes in JNK stimulation by mAChRs. To approach this question, we employed a chi- Overexpression of ␤␥ subunits of heterotrimeric G proteins results in stimulation of JNK activity. COS-7 cells were transfected with pcDNA3-HA-JNK (1 g/plate) together with pcDNA3 vector (control) or with an expression vector carrying cDNAs for the activated (QL) forms of the ␣ subunits of G s , G i2 , G q , and G 13 , or expressing ␤ 1 , ␥ 2 , or ␥ 2 * G protein subunits, alone or in combination, as indicated (2 g/plate in each case). Kinase reactions and Western blot (WB) analysis were performed in anti-HA immunoprecipitates from the corresponding lysates, as described under "Experimental Procedures." Autoradiograms correspond to representative experiments. 32 P-Labeled products as well as specific bands detected by the anti-HA antibody are indicated with an arrow. Data represent the mean Ϯ S.E. of four to five independent experiments, expressed as -fold increase with respect to vector-transfected cells (control). meric molecule combining the extracellular and transmembrane domain of CD8 fused to the carboxyl-terminal domain of ␤ARK, which includes the ␤␥-binding region (11). This chimeric molecule expresses the CD8 antigen at the cell surface, localizing the ␤ARK carboxyl-terminal domain to the inner face of the plasma membrane. The CD8-␤ARK chimera is expected to bind and sequester free ␤␥ complexes when dissociated from G␣ subunits upon receptor stimulation, thus blocking ␤␥-dependent pathways (11). As shown in Fig. 3, coexpression of CD8-␤ARK with the m1 or m2 mAChRs nearly abolished the activation of JNK in response to carbachol, while CD8 alone had no demonstrable effect. In contrast, JNK activation by stress-inducing agents such as anisomycin (9) was unaffected by CD8-␤ARK, demonstrating the specificity of this approach (Fig. 3). Taken together, these findings strongly suggest that signaling from m1 and m2 mAChRs to JNK is mediated by ␤␥ subunits of heterotrimeric G proteins.
We have recently shown that the Rho-related small GTPbinding proteins Rac1 and Cdc42 are integral components of signaling pathways linking certain cell surface receptors to JNK (9). Thus, we set out to investigate whether these small GTPases mediate JNK activation by G protein-coupled receptors, using as a tool the expression of dominant-negative mutants for Ras, RhoA, Rac1, and Cdc42. None of these dominantinhibitory small GTP-binding proteins affected the JNK activation in response to anisomycin (Fig. 4). In contrast, as shown in Fig. 4, coexpression of N17Ras and N17Rac1 prevented JNK activation by either mAChRs, or when induced by ␤␥ complexes, N17Cdc42 diminished JNK stimulation by m1 mAChRs, and the dominant-negative form of RhoA did not display any significant effect. Similarly, we have recently shown that JNK activation by EGF in COS-7 cells is also blocked by N17Ras, N17Rac1, and N17Cdc42 (9). Thus, these findings support a role for Ras and Rac1 or Cdc42 in linking both tyrosine kinase and G protein-coupled receptors to the JNK pathway.
Although Ras controls the activity of ERKs, recent available evidence suggests that Rac1 and/or Cdc42 regulate JNK acti-vation (9). Thus, distinct small GTP-binding proteins appear to link cell surface receptors with independent signaling pathways leading to the activation of each member of the MAPK superfamily. In this study, we present evidence supporting a role for ␤␥ subunits of heterotrimeric G proteins in communicating G protein-coupled receptors with the JNK pathway, acting on a Ras-and Rac1-dependent biochemical route. Furthermore, we and others have previously shown that ␤␥ complexes link this class of cell surface receptors to ERKs, in this case acting through Ras (10,16). Thus, taking these findings together we can postulate that ␤␥ heterodimers provide a link between heterotrimeric G proteins and small GTP-binding proteins.
The molecular basis for this interaction is still poorly defined (16). However, it is strikingly similar to that of the pathway linking the G protein-coupled pheromone receptors to MAPKrelated enzymes in the budding yeast S. cerevisiae. In this case, extracellular ligands (␣ or a factors) activate pheromone receptors which, in turn, induce the dissociation of a heterotrimeric G protein into ␣ (GPA1) and ␤␥ (Ste4, Ste18) subunits (see Ref. 17 for review). Free ␤␥ dimers then activate a serine-threonine kinase (Ste20), initiating activity from a linear cascade of kinases, including sequentially Ste11, Ste7, and the yeast MAPK homologues Fus3 and Kss1. Extensive search for molecules linking yeast ␤␥ complexes to Ste20 has led to the recent discovery that Cdc42 participates in Ste20 activation (18), and that Ste4 might directly bind and activate Cdc24, a nucleotide exchange factor for Cdc42 (19). Available data suggest that additional proteins might also be involved, including a protein designated Ste5, which plays a role as a platform or scaffold recruiting Ste11, Ste7, Kss1/Fus3 (20), and another small GTPbinding protein, the Ras homologue Rsr1 (21). The latter also appears to play an important role by binding Cdc24 and, probably, by positioning this guanine nucleotide exchange factor within the cell, thus allowing its interaction with Cdc42 and its targets (21). These observations represent an interesting ex- pcDNA3-HA-JNK was cotransfected with expression plasmids for m1 or m2 mAChRs (1 g/plate), and ␤ 1 and ␥ 2 subunits (2 g per plate), together with vector alone (vector) or plasmids expressing the CD8 receptor or a chimeric molecule CD8-␤ARK, as indicated. Cells were stimulated with carbachol (10 M) or anisomycin (10 g/ml) for 15 min or left untreated. Quantitation of JNK activity present in anti-HA immunoprecipitates was performed as in Fig. 1. Data represent the mean Ϯ S.E. of three independent experiments, expressed as -fold increase with respect to unstimulated cells.
FIG. 4. Effect of dominant-negative mutants for Ras, RhoA, Rac1, and Cdc42 on JNK activation. COS-7 cells were transfected with pcDNA3-HA-JNK and expression plasmids for m1 and m2 receptors (1 g/plate) or for ␤ 1 and ␥ 2 G protein subunits (2 g/plate), together with either pcDNA3 (vector) without insert or carrying cDNAs for N17Ras, N19RhoA, N17Rac1, or N17Cdc42 (1 g/plate), and cells were left untreated or stimulated with carbachol (10 M) or anisomycin (10 g/ml) for 15 min, as indicated. Cell lysates were processed as in Fig. 1. Data represent the mean Ϯ S.E. of four to five independent experiments, expressed as percent of activation with respect to the corresponding vector cotransfected control (100%). ample of convergence between Ras-like (Rsr1) and Rho-like (Cdc42) biochemical routes and might also provide a clue regarding the pathway connecting mammalian ␤␥ to JNK. Based upon these observations in yeast and our present findings, we can hypothesize that upon receptor stimulation ␤␥ dimers might recruit a yet to be identified guanine nucleotide exchange factor for Rac1 and/or Cdc42 and that Ras functioning may be necessary for the effective activation of these GTPbinding proteins and the consequent stimulation of JNK. We conclude that our present study might represent a biologically relevant example of a signal transduction pathway extraordinarily conserved from yeast to mammals and might provide an attractive model to elucidate the nature of those molecules linking heterotrimeric G proteins to small GTP-binding proteins.