Rac1 and Rhog Promote Cell Survival by the Activation of PI3K and Akt, Independently of Their Ability to Stimulate JNK and NF-Kb

Rac1 and RhoG promote cell survival by the activation of PI3K and Akt, independently of their ability to stimulate JNK and NF-kB

Cristina Murga1,2, Muriel Zohar1, Hidemi Teramoto*,1 and J Silvio Gutkind1

1Oral and Pharyngeal Cancer Branch, National Institute of Dental Research, National Institutes of Health, Bethesda, Maryland, MD 20892-4330, USA

Small GTPases of the Rho family play a central role in Introduction cellular processes that involve the reorganization of the actin-based cytoskeleton. Rho-related GTPases, which Apoptosis, or programmed cell death, is a tightly include Rac and Cdc42, can also regulate gene expression regulated process by which cells initiate a sequence of often through the activation of kinase cascades leading to events ultimately destined to cause cellular suicide enhanced activity of stress activated protein kinases (Hengartner, 2000). A wide variety of cell surface (SAPKs), including JNK and p38 MAP kinases. As SAPKs receptors, ligands, physical insults, and intracellular- are implicated in programmed cell death, these observations triggered events can trigger the activation of the suggest that Rho GTPases may promote the initiation of the apoptotic program (Hengartner, 2000). This process apoptotic process. However, recent reports suggest that involves a number of morphological and biochemical Rho GTPases can have either a protective or a pro- changes that require ®nely tuned control mechanisms. apoptotic role, depending on the particular cellular context. In particular, the role of intracellular signaling In an eort to explore the molecular mechanisms underlying molecules in the initiation and regulation of apoptosis these divergent biological activities, we asked whether there has been intensively investigated in recent years. For was indeed a correlation between the ability to induce example, cumulative evidence has implicated the stress- SAPKs and apoptosis by Rho family members. We found activated protein kinases (SAPKs), such as JNK and that although constitutively activated (Q61L) mutants of p38 MAP kinase, in the regulation of cell death (Chen Rac1, Cdc42, and RhoG, a Rac1 related GTPase of and Tan, 2000; Davis, 2000). The key role of JNK in unknown function, potently induce JNK in COS 7 cells, programmed cell death has been recently supported by none of these GTPases could induce apoptosis, nor enhance gene disruption experiments targeting each of the three uv-induced cell death. In contrast, Rac1 and RhoG known JNK isoforms (Tournier et al., 2000). Com- eciently protected cells from uv-induced apoptosis. bined mutation of the two most ubiquitously expressed Furthermore, we provide evidence that Rac1 and RhoG isoforms of JNK, JNK1 and JNK2, causes embryonic can activate both apoptotic and anti-apoptotic pathways. death associated with defects in neural apoptosis Whereas the former is mediated through JNK, the latter is (Kuan et al., 1999; Sabapathy et al., 1999). Further- independent on the transcriptional activation of NF-kB, a more, studies performed in embryo ®broblasts derived pro-survival pathway, but results from the direct interaction from these mice have con®rmed that JNK is central to of these GTPases with phosphatidylinositol 3-kinase the regulation of cell death in response to stress stimuli, (PI3K) and the stimulation of Akt. Together, these ®ndings such as those elicited by uv-irradiation, protein indicate that members of the Rho family of small GTP- synthesis inhibitors and DNA alkylating agents, but binding proteins can provoke the concomitant stimulation is dispensable in apoptotic signals emanating from cell of two counteracting signaling pathways, and that their death receptors such as Fas (Davis, 2000). Recent balance ultimately determines the ability of these GTPases studies have also revealed that JNK modulates a to promote cell survival or death. number of apoptotic eectors, including Bcl-2, Bcl-xL, Oncogene (2002) 21, 207 ± 216. DOI: 10.1038/sj/onc/ c-myc, p53, cytochrome c release, and Fas-L expression 1205036 levels (see Chen and Tan, 2000) or (Davis, 2000 for a review). However, how cell surface receptors stimulate Keywords: RhoG; PI3K; Akt; cell survival JNK to promote apoptosis is still unclear. Small GTPases of the Rho family have been traditionally linked to the regulation of the cellular skeleton by promoting the remodeling of actin-contain- *Correspondence: H Teramoto, Oral and Pharyngeal Cancer Branch, National Institute of Dental Research, National Institutes ing cytostructures (Hall, 1998). More recent evidence of Health, 9000 Rockville Pike, Building 30, Room 211 Bethesda, has revealed that Rho-related GTPases, which include MD 20892-4330, USA Rho, Rac, and Cdc42, also regulate gene expression, 2Current address: Centro de BiologõÂ a Molecular, Facultad de primarily through the activation of kinase cascades Ciencias, C-V, Universidad Autonoma de Madrid, Campus de resulting in the potent activation of JNK and members Cantoblanco, Madrid-28049, Spain Received 21 May 2001; revised 25 September 2001; accepted 9 of the p38 family of MAP kinases (Coso et al., 1995; October 2001 Zhang et al., 1995). The ability to stimulate stress Activation of Akt and cell survival by Rac-related GTPases C Murga et al 208 activated protein kinases also suggests that these Rho forms of Ras and representative Rho-family members, GTPases can promote the initiation of apoptotic including RhoA, RhoG, Rac1, and Cdc42 (Figure 1). programs mediated by JNK and p38, thus resulting Cotransfection with HA-tagged forms of each MAP in cell death. However, paradoxical results suggest that kinase allowed us to assess the eect of these small G Rho GTPases can both trigger cell death or promote proteins on dierent MAP kinase pathways in this survival, and that these seemingly con¯icting responses cellular system. As expected, MAP kinase activity was are heavily dependent on the cellular context (Bazenet potently stimulated by an activated Ras, but Rho- et al., 1998; Bobak et al., 1997; Boehm et al., 1999; family members did not appear to have any eect Kimmelman et al., 2000; Lores et al., 1997; Subauste et (Figure 1). In contrast, the catalytic activity of JNK al., 2000; Yang et al., 2000). The molecular mechan- was markedly increased by Rac1, RhoG and Cdc42 isms underlying this strikingly distinct biological coexpression, while p38 activity was only slightly activity are still poorly understood. activated in response to Rac1 and RhoG (Figure 1). In this study, we asked whether a direct correlation This response was particularly modest, as osmotic exists between the ability to induce stress activated stress or uv light increased the activity of p38 more protein kinases and apoptosis by representative members of the Rho family of GTPases, RhoA, Rac1, and Cdc42, as well as RhoG. RhoG is a GTPase highly related to Rac1, which has been reported to cause morphological and cytoskeletal changes in ®broblasts (Blangy et al., 2000; Gauthier- Rouviere et al., 1998) and to induce neurite outgrowth in PC12 cells (Katoh et al., 2000), but whose normal function is still unknown. When expressed in COS 7 cells, we found that constitutively activated mutants of Rac1, Cdc42, and RhoG potently induce JNK, but surprisingly, none of these small GTPases induced apoptosis in these cells, nor did they enhance the ability of a physical stress, uv, to induce cell death. In contrast, two of these GTPases, Rac1 and RhoG, protected eectively cells from uv-induced apoptosis. In fact, we observed that Rac1 and RhoG could activate both apoptotic and anti-apoptotic pathways, the former through JNK and the latter through their direct interaction with phosphatidylinositol 3-kinase (PI3K) and the consequent stimulation of Akt. Together, our results suggest that the ®ne balance between these counteracting signaling pathways elicited by Rho-related GTP-binding proteins can ultimately determine whether these GTPases promote cell survival or death.

Results

Small G proteins of the Rho family have been implicated in a variety of cellular functions ranging from cytoskeletal changes to cell growth, gene Figure 1 Activated forms of small GTPases stimulate MAP expression and cell cycle progression (Hall, 1998). kinase pathways in COS 7 cells. (A) COS 7 cells were grown in They perform these diverse biological tasks by means 10 cm plates and transfected with 1 mg of expression vectors for of the selective activation of speci®c downstream HA-ERK2, HA-JNK1 or HA-p38a together with pCEFL-AU5- Ras V12, RhoA QL, RhoG QL, Rac1 QL, and Cdc42 QL eector molecules, which include those acting on (0.5 mg) or the same amount of pCEFL-GFP as a control. Cells distinct MAP kinase signaling modules (Coso et al., were serum-starved and total cell lysates were immunoprecipitated 1995; Zhang et al., 1995). In particular, the ability to with anti-HA antibodies. In vitro kinase reactions were carried activate MAP kinases has been proposed to participate out as described in Materials and methods. After electrophoretic separation of the labeled substrates, proteins were transferred to a in life versus death decisions: while MAP kinase PVDF membrane where the incorporation of 32P-ATP was promotes cell survival, JNK and p38 MAP kinase visualized by autoradiography. The amount of kinase protein elicit pro-apoptotic responses in a number of cellular present in each immunoprecipitate was assessed by Western blot systems (Bonni et al., 1999). To begin addressing the using an anti-HA antibody. The migration of radiolabeled correlation between the activation of dierent MAP substrates and HA-tagged kinases is indicated by arrows. Shown is one representative experiment out of three performed. (B) kinases by small GTPases and cell survival or death, Expression of the indicated AU5 tagged molecules in total cell we expressed in COS 7 cells epitope-tagged, activated extracts was con®rmed by Western blot analyses

Oncogene Activation of Akt and cell survival by Rac-related GTPases C Murga et al 209 than 10-fold (Chiariello et al., 2000, and data not GTPases. In particular, the activation of the NF-kB shown). transcription factor, which is often stimulated by small As JNK activation promotes cell death in a variety GTPases (Perona et al., 1997), constitutes a well- of cellular settings (see Chen and Tan, 2000 or Davis, established anti-apoptotic route (Pahl, 1999). We ®rst 2000 for a review), and Cdc42, Rac1, and RhoG con®rmed that Ras, RhoA, Rac1, and Cdc42 could activated JNK potently, we next explored the possibi- stimulate NF-kB dependent expression potently, as lity that expression of these GTPases may be sucient judged by the use of a reporter assay where luciferase to promote apoptosis in these cells. For these expression is driven by ®ve kB-responsive elements experiments, COS 7 cells were transfected with acti- (Figure 4A). Of interest, we observed that RhoG is one vated forms of Ras, RhoA, Rac1, RhoG and Cdc42, of the most potent inducers of this response. and analysed for the percentage of AU5-positive cells We then explored the potential role of NF-kB in the (immunostained in red) that were positive for apoptosis survival pathway elicited by these GTPases. For these by the TUNEL assay (stained in green), as previously experiments, we took advantage of the availability of described (Murga et al., 1998). Surprisingly, we could an IkB mutant (h-IkBa S32G S36A) that cannot be not see detectable dierences in cell death after phosphorylated, is not degraded, and thus acts as a overexpression of Rho family proteins, although potent dominant interfering protein for NF-kB signal- apoptosis was clear in FLICE-transfected cells used ing (Brown et al., 1997). First, we examined whether as a positive control (Figure 2A). As small G protein expression of this IkB mutant would interfere eec- overexpression was not sucient to promote apoptosis tively with the transcriptional activity of NF-kB in our per se, we then examined whether the activated forms cellular system. As shown in Figure 4A, IkB S32G of these proteins might help increase the apoptotic S36A markedly suppressed NF-kB stimulation by all response to other stimuli, such as that caused by small G proteins. Under similar conditions, however, exposure to uv-radiation. In each case, the percentage this IkB mutant did not diminish the percentage of of apoptotic cells in the background of untransfected dead cells in response to uv treatment nor prevented cells was used as an internal control. As shown in the protective eect of Ras, Rac1, and RhoG. These Figure 2b, 30 ± 40% of the cells were clearly apoptotic results strongly argue against a role for NF-kB after uv-treatment. When apoptosis was scored in signaling in the survival pathway utilized by these AU5-positive cells, we found that this percentage did small GTPases. On the contrary, inhibition of NF-kB not augment, but rather decreased upon expression of enhanced the limited anti-apoptotic activity of Cdc42. activated forms of Ras, RhoG, and Rac1. No The nature of the NF-kB-dependent pro-apoptotic signi®cant changes were observed upon RhoA over- pathway elicited by this particular GTPase is at the expression, while Cdc42 caused a partial protection present unknown, and under current investigation. that was highly variable among individual experiments. A substantial body of evidence now supports a Altogether, these results suggest that despite activating widespread role for the serine/threonine kinase Akt/ the JNK pathway potently, activated Rac1 and RhoG PKB in delivering survival signals (Chan et al., 1999). do not induce apoptosis, but may instead protect from This serine/threonine kinase is potently stimulated by uv-induced cell death. many growth factor receptors through the activation of This observation led us to question if indeed JNK phosphatidylinositol 3'-kinase (PI3K), which phosphor- activation was part of the apoptotic pathway elicited ylates PIP2 in the 3' position of the inositol ring and by uv in our cellular system. To address this issue, we causes the consequent recruitment of Akt to the plasma took advantage of the discovery of JIP-1, a scaolding membrane and its phosphorylation by PIP3 regulated protein for the JNK pathway that can act as a potent kinases, including PDK1 and a PDK2 activity (Chan et functional inhibitor of JNK-mediated responses when al., 1999). In order to investigate whether small overexpressed (Dickens et al., 1997; Whitmarsh et al., GTPases can act on the Akt pathway, we co- 1998). In fact, we observed that overexpression of JIP- transfected an epitope-tagged Akt construct together 1 eectively inhibits the apoptotic response to uv- with the active Ras and Rho-family members and irradiation (Figure 3). It also enhanced nearly twofold performed immune-complex kinase assays. As shown the protective activity of Ras, RhoG and Rac1. Of in Figure 5, activated forms of Ras, Rac1, and RhoG interest, JIP-1 did not protect cells from uv-induced potently stimulated Akt in this system. However, apoptosis when co-expressed with RhoA, thus suggest- neither RhoA nor Cdc42 were able to stimulate Akt ing that inhibition of JNK is not enough to decrease phosphotransferase activity even when many dierent uv-induced apoptosis in the presence of active RhoA. conditions were examined (Figure 5 and data not In contrast, Cdc42 in combination with JIP-1 became a shown). As PI3K-dependent and PI3K-independent potent inhibitor of apoptosis, reducing the apoptotic mechanisms of activation of Akt have been described response to uv to nearly 70% when compared to cells (Franke et al., 1995; Konishi et al., 1997; Sable et al., transfected with JIP-1 alone. Together, these results 1997), we next asked whether the observed activation suggest that the protective eect of certain GTPases of Akt by Ras, Rac1, and RhoG was mediated by can be enhanced upon blockade of the JNK pathway. PI3Ks. As shown in Figure 5, pre-incubation of cells As JNK activation promotes rather than prevents with a potent PI3K inhibitor, wortmannin, completely apoptosis in these cells, we next decided to explore the blocked small G protein-mediated stimulation of Akt. nature of the survival pathway elicited by small We can conclude that, in these cells, Ras, Rac1, and

Oncogene Activation of Akt and cell survival by Rac-related GTPases C Murga et al 210

Figure 2 Eect of overexpression of the activated forms of G proteins in cell survival and in apoptosis induced by uv-radiation. (A) COS 7 cells were grown on coverslips in 6-well plates and transfected using Lipofectamine PLUS reagent with pCEFL-AU5-Ras V12, RhoA QL, RhoG QL, Rac1 QL, and Cdc42 QL (1 mg), and the same amount of pCEFL ± FLICE or pCEFL-b-gal as positive and negative controls, respectively. Cells were starved for 24 h, ®xed and subjected to immuno¯uorescence analysis and TUNEL labeling. The percentage of apoptosis was determined by counting Rhodamine-positive cells (transfected) that were also TUNEL positive as judged by FITC staining (apoptotic). More than 200 cells were counted from several randomly chosen ®elds. Results from three independent experiments are shown as the mean+s.e.m. (B) After transfection with the AU5-tagged forms of Ras V12, RhoA QL, RhoG QL, Rac1 QL and Cdc42 QL (1 mg per well), cells were subjected to uv-irradiation (60 Jul/m2) and serum starved for 15 h. Cells were then ®xed and stained and the percentage of apoptotic cells was scored by counting Rhodamine-positive cells (transfected) or control cells that were visualized by DAPI (untransfected). Data represent the mean+s.e.m. of 3 ± 5 independent

Oncogene Activation of Akt and cell survival by Rac-related GTPases C Murga et al 211 RhoG can stimulate Akt potently in a PI3K-dependent these proteins with a kinase defective mutant of Akt, manner. that can act as dominant interfering molecule for Akt Ras-family proteins can form physical complexes with PI3Ks thereby enhancing their catalytic activity (Rodriguez-Viciana et al., 1994; Rubio et al., 1997). To explore further the possibility that this association represents the mechanism by which other GTPases stimulate PI3K in our system, we performed a PI3K assay in AU5-immunoprecipitates from cells cotransfected with active small G proteins and the catalytic and regulatory subunits of PI3Ka, p110a and p85a. As expected, a remarkable amount of PI3K activity was found in the RasV12 immunoprecipitates (Figure 6). A substantial amount of PI3K activity was also recovered in RhoG and Rac1-containing immunoprecipitates, while a more limited activity was found bound to Cdc42. This association to PI3K was speci®c for Rac-related GTPases since no detectable activity was bound to other small G proteins such as RhoA. Thus, the ability of Rac1 and RhoG to form physical complexes with PI3K may provide the likely mechanism by which these GTPases stimulate the Akt pathway. In order to explore whether the PI3K/Akt pathway plays a role in the survival response induced by Ras and Rho GTPases in these cells, we co-transfected

Figure 4 Eect of the NF-kB pathway on cell survival mediated by activated mutants of small GTPases. (A) COS 7 cells were grown in 6-well plates and transfected as described above with 0.25 mg of pNF-kB-LUC, 0.25 mg of pCEFL-b-gal and expression plasmids for the active forms of small GTP-binding proteins (0.5 mg of each construct per well) or the same amount of pCEFL ± GFP as a control. Fifty ng of pMT2-IkB S32G S36A (IkB DN) or the same amount of a GFP-expressing plasmid (control) were included in the transfection mix. After overnight starvation, cells were lysed and the luciferase activity was measured in cell extracts. The amount of b-galactosidase activity present in each sample was used to normalize for transfection Figure 3 Eect of overexpression of a JNK inhibitory protein eciency. Data are represented as fold induction of the normal- (JIP-1) on small GTPase-mediated survival from uv-induced ized luciferase activity with respect to control cells transfected apoptosis. One mg of pcDNA3-JIP-1 was cotransfected together with GFP. Bars show the mean values+s.e.m. of three separate with AU5-tagged active forms of the dierent small G proteins experiments. (B) The eect of inhibiting the NF-kB pathway was using the same amount of pcDNA3-b-gal as a control. The evaluated by comparing the apoptosis induced by uv-irradiation following day cells were irradiated with uv-light (80 Jul/m2) and in cells expressing the mutant IkB S32G S36A (IkB DN) or not left in serum-free medium for 6 h. Apoptotic cells were counted as (control). After transfection, apoptosis was quanti®ed as described for Figure 2. Bars represent the mean+s.e.m. of three described in Figure 2. Data shown represent the mean+s.e.m. independent experiments of three independent experiments

experiments where more than 200 cells were counted. (C) Representative ®elds of uv-treated cells expressing the indicated small GTPases were photographed under uv-light using an Axiophot2 ¯uorescence microscope. Rhodamine-staining corresponds to AU5- positive cells (open arrowhead) and FITC staining indicates TUNEL-positive cells (full arrowhead)

Oncogene Activation of Akt and cell survival by Rac-related GTPases C Murga et al 212

Figure 6 A PI3K activity associates with small GTP-binding proteins in COS 7 cells. COS 7 cells were transfected with expression constructs for PI3Ka (pCEV-p110a and pCDNA3- p85a,1mg each per 10 cm dish) and vectors containing the AU5- tagged constitutively active forms of RasV12, RhoA QL, RhoG QL, Rac1 QL or Cdc42 QL (2 mg) or pCEFL ± GFP as a control. Cells were lysed in a buer containing 10 mM MgCl2, immunoprecipitated with anti-AU5 antibodies and assayed for the presence of a PI3K activity using phosphatidylinositol as a substrate as described in Materials and methods. The chromato- 32 Figure 5 Activated forms of small GTPases promote PI3K- graphic mobility of the resulting P-labeled phosphatidylinositol- dependent activation of Akt. COS 7 cells were transfected with 3-phosphate (PI-3P) is indicated. The autoradiogram corresponds expression plasmids for AU5-tagged activated forms of small to a representative experiment that was repeated two additional GTP-binding proteins together with a plasmid expressing a HA- times tagged Akt (pCEFL-HA-Akt, 1 mg per 10 cm dish). Cells were starved overnight and treated for 30 min prior to lysis with 50 nM wortmannin (wortmannin) or were left untreated (control). Akt activity was determined in the anti-HA-immunoprecipitates as described in Materials and methods. The amount of kinase 2000). Furthermore, the functional activity of Ras is protein present in each sample was assessed by Western blot now known to be required for mitogenic signaling by analysis using anti-HA antibodies. The migration of HA-Akt and most growth factor receptors (Ise et al., 2000). In its phosphorylated substrate (H2B) is indicated by arrows. Films were quanti®ed using NIH image 1.61 software to produce a bar contrast, the diversity of biological functions per- graph representing the mean+s.e.m. Shown are autoradiograms formed by small GTPases of the Rho family is just corresponding to a representative experiment that was repeated beginning to be fully appreciated. In particular, the three times with nearly identical results role of Rho-like proteins in the regulation of the actin containing cytostructures and their dynamic changes during cell migration and adhesion, as well as their role in cellular growth and transformation has been recently (Murga et al., 1998). As seen in Figure 7, this Akt characterized in great detail (Hall, 1998). However, mutant slightly increased the number of apoptotic cells other important functions performed by these GTP- after uv-irradiation in the case of control and Ras binding proteins are much less understood. For transfected cells, and nearly abolished the protective example, con¯icting results have been reported regard- eect of Rac1, and RhoG. Although this inhibitory ing the role of small GTPases in the regulation of cell mutant of Akt did not aect the percentage of survival or death (see Gomez et al., 1998 and apoptotic cells in RhoA transfectants, it reverted the references therein). poor protective activity of Cdc42, which now displayed In this study, we have explored the molecular a limited pro-apoptotic eect. In sum, co-transfection mechanisms underlying these distinct biological activ- with dominant negative Akt completely abrogated the ities using a highly transfectable and well characterized survival eect mediated by small GTPases, thus model cellular system. indicating that the protective signals emanating from For this study, COS 7 cells were transfected with Ras and Rac-related proteins are dependent on the activated mutants of Ras and Rho GTPases, and activation of the PI3K/Akt pathway. found that none of these GTP-binding proteins stimulate cell death or enhance the apoptotic response to physical insults such as uv-irradiation, in spite of the Discussion ability of three of these GTPases, Cdc42, Rac1, and the highly related RhoG, to stimulate JNK, a pro- A key role for Ras in cell growth control was predicted apoptotic pathway in many cell types (Chen and Tan, from the frequent occurrence of activating mutations in 2000; Davis, 2000). In fact, uv-induced apoptosis in a large variety of human cancers (Minamoto et al., COS 7 cells is JNK-dependent, as blockade of the

Oncogene Activation of Akt and cell survival by Rac-related GTPases C Murga et al 213 speci®c for Rac-related GTPases since RhoA did not protect cells from uv-induced death, but instead it helped revert the protective eect of blocking JNK function. Thus, Rho may activate signaling pathways that, although do not induce apoptosis per se in these cells, may nonetheless circumvent the requirement of JNK for the activation of cell death programs by uv. Which of the many Rho-initiated pathways participate in this pro-apoptotic process is unknown at the present, and will be the subject of future investigation. Activation of PI3K by some membrane-localized receptors leads to the rapid production of 3'-phosphoi- nositides, which allows the recruitment of Akt to the membrane and its activation by upstream Akt-kinases (Chan et al., 1999). In turn, Akt initiates several anti- apoptotic pathways including inactivation of caspases, BAD phosphorylation, and the regulation of the expression of Bcl-2 family members, which eventually leads to survival responses in many dierent cellular systems (Chan et al., 1999). The activation of Akt by Rac-related proteins appears to involve the direct association of Rac1 and RhoG with PI3Ks. This eect is somewhat surprising, as it has been often observed that activation of PI3K results in the activation of guanine-nucleotide exchange factors acting on Rac (Han et al., 1998), and Rac1 has been shown to mediate the cytoskeletal changes provoked by PI3K activation (Reif et al., 1996). Accordingly, Rac proteins would be expected to act downstream rather than upstream from PI3Ks. However, early reports sug- Figure 7 Activated forms of small GTP-binding proteins gested that GTP-loaded Rac1 and Cdc42 could promote Akt-dependent survival. (A) Two mg of pCEFL-Akt enhance the activity of PI3Ks in vitro (Bokoch et al., K179M (Akt K179M) or -b-gal (control) were cotransfected 1996; Tolias et al., 1995). Thus, similar to the better together with AU5-tagged active forms of the dierent small understood process by which Ras stimulate PI3K GTPases into COS 7 cells grown on coverslips. Fifteen hours after transfection, cells were irradiated with uv-light (80 Jul/m2) (Rodriguez-Viciana et al., 1994; Rubio et al., 1997), and left in serum-free medium for 6 h, ®xed and subjected to the Rac branch of Rho-related GTPases, including TUNEL analysis. Apoptotic cells were counted as described for Rac1, Rac2, and RhoG, may be able to associate with Figure 2. Bars represent the mean+s.e.m. of three independent PI3Ks and cause their activation. On the other hand, experiments. (B) Members of the Rho family of small GTP- whereas Ras binds the catalytic subunit of PI3Ks, binding proteins can trigger the concomitant stimulation of two counteracting signaling pathways, and their balance probably p110a, p110b,p110g, and p110d, through a conserved determines the ability of these GTPases to promote cell survival Ras-binding region (Wymann and Pirola, 1998), no or death candidate Rac-binding site can be observed in these sequences. Instead, the p85 regulatory subunit exhibits a BCR domain that may provide the structural basis for the interaction between Rac1 and RhoG and PI3Ks functional activity of JNK by the expression of JIP-1 (Wymann and Pirola, 1998). Of interest, physical (Dickens et al., 1997; Whitmarsh et al., 1998), inhibited association may not be sucient to promote PI3K the initiation of cell death programs upon uv- activity in vivo, as Cdc42 also binds PI3K but does not irradiation. However, Rac1 and RhoG eectively result in the activation of PI3K-dependent events, such protected cells from the apoptotic response to uv, as Akt stimulation. and this eect was enhanced by the functional Biochemical analysis and the use of dominant interfer- blockade of the JNK pathway. These data indicated ing mutants revealed that the protective eect of Rac1 and that Rac1 and RhoG can promote the activation of RhoG is mediated by the stimulation of Akt, and is survival pathways in these cells, and that the JNK independent of NF-kB activation. Similar eects were cascade does not mediate this response. Cdc42, which observed for activated Rac2 (not shown). Thus, we stimulated JNK more potently, did not prevent cells propose that Rac-family proteins, while activating from undergoing apoptosis, but it exerted a limited potently both JNK and Akt kinases, initiate Akt- protective eect after JNK blockade. Thus, Cdc42 may dependent (but NF-kB-independent) survival mechan- be able to activate survival pathways weakly, but this isms that are prevalent over the apoptotic signals biological eect can only be unmasked after preventing mediated by JNK in these cells (see Figure 7). In line the pro-apoptotic eect of JNK. These eects are with this hypothesis, activation of RhoA and Cdc42 did

Oncogene Activation of Akt and cell survival by Rac-related GTPases C Murga et al 214 not protect cells from apoptosis, and that correlated with PLUS (Life Technologies) following the manufacturer's a lack of Akt activation by these GTPases. Conceptually, instructions. one may argue that, in dierent cellular systems, divergences in this balance between JNK and activation DNA constructs Akt achieved by small GTPases could help explain The previously described cDNAs for human RhoA, Rac1, con¯icting results obtained regarding their role in the Rac 2, RhoG and Cdc42Hs (referred in here as Cdc42) GTP- regulation of apoptosis. One can also reconcile the fact binding proteins were kindly provided by Dr Toru Miki that despite potent JNK activation, Rac can also (Horii et al., 1994). Their respective constitutively activated transduce anti-apoptotic eects by considering that Akt mutants were obtained as described (Coso et al., 1995), by survival signals could be stronger, more persistent in time replacing a glutamine corresponding to codon 61 of Rac1 or, simply dominant over those pro-apoptotic pathways and Cdc42 for a leucine residue, or the corresponding amino regulated by JNK. Alternatively, tissue speci®c mechan- acid in RhoG, Rac2 and RhoA (QL mutants). For the isms can also exist that help regulate this balance. For expression of an activated H-Ras (referred in here as Ras), as example, Akt-dependent inhibition of the JNK MAP RasV12 mutant was similarly engineered. Subsequently, the kinase module has been described in neurons (Shimoke et coding DNA fragments were subcloned in frame into the BglII and EcoRI site of the vector pCEFL-AU5 and veri®ed al., 1999), PC12 cells (Levresse et al., 2000) and in by sequencing. Plasmids expressing an epitope-tagged Akt endothelial cells (Madge and Pober, 2000), thus suggest- (pCEFL-HA-Akt) and a dead kinase Akt form (pCEFL-HA- ing the existence of a direct cross-talk between JNK and Akt K179M), JNK1 (pCEFL-HA-JNK1), p38a (pCEFL-HA- Akt pathways in certain cell types. p38a), ERK2 (pCEFL-HA-ERK2); p110a (pCEV29-p110a); In line with our present results, whereas early work p85a (pCDNA3-p85); JIP-1 (pCDNA3-JIP1); b-galactosidase indicated that small G proteins of the Rho family do not (pCDNA3-bgal) and GFP (pCEFL-GFP) have been pre- play a role in Akt regulation (Klippel et al., 1996), more viously described (Chiariello et al., 2000; Murga et al., 2000). recent studies have shown that Rac1 and 2 can indeed A construct containing a mutant IkB (pMT2T h-IkBa S32G activate Akt in cells of hematopoietic origin, such as S36A), as well as a plasmid encoding FLICE (caspase-8) that Jurkat and BaF3 cells (Genot et al., 2000; Nishida et al., was further engineered into pCEFL (pCEFL ± FLICE), were kindly provided by Dr Uri Siebenlist and Dr Michael 1999). In T cells, Rac activation of Akt participates in the Lenardo, respectively. A reporter plasmid containing ®ve formation of membrane spikes (Genot et al., 2000), and consensus kB-responsive sites followed by the luciferase gene in BaF3 cells in the viability upon IL-3 removal (Nishida was obtained from Stratagene. et al., 1999). Furthermore, in Rac2-de®cient mast cells, the survival response to growth factors was diminished, Akt and MAP kinase assays and that correlated with a decreased activation of Akt MAP kinase and Akt activity in cells transfected with and enhanced BAD/Bcl-XL expression (Yang et al., 2000). However, in these studies, no evidence was plasmids for an HA-tagged ERK, p38, JNK or Akt were provided of a direct functional link between Rac determined as described (Chiariello et al., 2000; Murga et al., activation and the stimulation of anti-apoptotic path- 2000) using myelin basic protein (Sigma). GST-cjun79 or histone 2B (Roche Molecular Biochemicals) as substrates, ways through Akt. Thus, our present ®ndings that all respectively. When necessary, autoradiograms were scanned tested Rac family-members, Rac1, Rac2, and RhoG, and subsequently quanti®ed using NIH Image 1.61 software. stimulate PI3Ks and promote Akt activation, provides a biochemical mechanism by which these small G proteins mediate cell survival, and expands the repertoire of the PI3-kinase assay biological activities performed by Rac proteins. As Rho After transfection of COS 7 cells with constructs for the and Cdc42 did not stimulate Akt and cell survival, our expression of small G proteins together with p110a and p85, results also provide clear evidence of the distinct nature cells were allowed to recover for 24 h and then serum-starved of the biological functions regulated by each of the overnight. Total cellular lysates were then immunoprecipi- branches of the Rho-superfamily of Ras-related small tated with anti-AU5 (Covance, Richmond, CA, USA) and an in vitro PI3-kinase reaction was performed using phosphati- GTPases. The ability of the Rac/RhoG-PI3K-Akt dyl-inositol as a substrate as previously described (Gutkind et pathway to promote cell survival indicates a broader al., 1990), except all buers were supplemented with 10 mM role for Rac-related GTPases in the regulation of cell life MgCl2 to preserve interactions between G proteins and PI3K. or death decisions, and the contribution of this small 32P-labeled phospholipids were separated by thin layer GTP-binding proteins in cell fate determination by chromatography onto LK6D plates (Whatman, Clifton, NJ, apoptosis during organ development, tissue homeostasis, USA), and visualized by autoradiography. immune repertoire selection and cancer will surely become an active ®eld of future investigation. Western blots and immunoprecipitations Cells were lysed in Akt lysis buer (Murga et al., 1998) and total cellular lysates or immunoprecipitates were analysed by Western blot using mouse anti-HA (HA.11, 1 : 1000, Materials and methods Covance, Richmond, CA, USA), for Akt, ERK and JNK or mouse anti-AU5 (1 : 1000, Covance, Richmond, CA, USA) Cell lines and transfection for small GTPases. Immunocomplexes were visualized by COS 7 cells were cultured in DMEM supplemented with 10% enhanced chemiluminescence (ECL Plus, Amersham Pharma- fetal bovine serum. Cells were transfected by Lipofectamine cia Biotech Inc., Piscataway, NJ, USA) using secondary

Oncogene Activation of Akt and cell survival by Rac-related GTPases C Murga et al 215 antibodies coupled to horseradish peroxidase from Cappel DNA was then visualized by terminal deoxynucleotidyltrans- Laboratories (Durham, NC, USA). When necessary, an anti- ferase-mediated dUTP-FITC nick end labeling (TUNEL) mouse IgG Fc-speci®c secondary antibody coupled to HRP using a kit from Roche Molecular Biochemicals, following (Sigma, 1 : 10 000) was used to avoid cross-reactions with the the manufacturer's instructions, except the reaction was light chain of the immunoglobulin proteins. carried out at room temperature instead of 378C. The frequency of apoptosis was scored by counting several hundred Rhodamine-stained (transfected) cells from at least Luciferase assay 10 dierent ®elds per coverslip, and examining them for COS 7 cells were transfected with Lipofectamine PLUS FITC-staining (TUNEL positive) under uv light in an (Life Technologies) with dierent expression plasmids for Axiophot2 ¯uorescence microscope (Zeiss, Jena, Germany). small G proteins together with pcDNA3-b-gal, a plasmid When required, pictures were taken at 406 magni®cation expressing the enzyme b-galactosidase, and pNF-kB-LUC, using a SPOT digital camera and software (Diagnostic the reporter plasmid expressing the luciferase gene under Instruments, Sterling Heights, MI, USA) and analysed using the control of ®ve consensus kB-responsive elements. When Adobe Photoshop. required, pMT2-IkB S32G S36A or the same amount of GFP plasmid as a control were included in the transfection mix. After overnight starvation, cells were lysed in reporter lysis buer (Promega) and the luciferase activity in cell extracts was measured using a luciferase assay system (Promega). b-galactosidase activity present in each sample Abbreviations was assayed by a colorimetric method with ONPG as a DMSO, dimethyl sulfoxide; FITC, ¯uorescein isotiocya- substrate, and was used to normalize for transfection nate; GFP, green ¯uorescent protein; HA, hemagglutinin eciency. epitope tag; HRP, horseradish peroxidase; H2B, histone 2B; JNK, c-jun N-terminal kinase; MAP kinase, mitogen Apoptosis assay activated protein kinase; MBP, myelin basic protein; ONPG, o-nitrophenyl b-D-galactopyranoside; PBS-phos- COS 7 cells were grown on coverslips and transfected with phate buered saline; PI3K, phosphatidylinositol-3-OH plasmids encoding small G proteins or b-galactosidase as a kinase; PVDF, polyvinylidene ¯uoride; s.e.m., standard control, using Lipofectamine PLUS. Protection from uv- error of the mean induced apoptosis was performed essentially as described (Murga et al., 1998) with few modi®cations. Brie¯y, 15 h after transfection, cells were subjected to uv-irradiation (60 ± 80 Jul/m2, uv-Stratalinker 1800, Stratagene) and subsequently Acknowledgments serum-starved for 6 ± 15 h in DMEM. Cells were then ®xed in We are especially grateful to C DeGra and E Powell for 4% paraformaldehyde and permeabilized with 0.01% Triton excellent secretarial assistance. We thank Dr Siebenlist for X-100. Transfected cells were identi®ed by immunostaining providing the construct for h-IkBa S32G S36A, Dr for small GTPases or b-galactosidase expression with a Lenardo for the vector containing FLICE and Dr Davis mouse anti-AU5 (Covance, 1 : 100) or anti-b-galactosidase for the kind gift of JIP-1 expression plasmids. Thanks to antibody (Promega, 1 : 100) followed by a Rhodamine- Dr Montaner and Dr Servitja for their critical reading of coupled secondary antibody (Sigma, 1 : 200). Fragmented the manuscript.

References

Bazenet CE, Mota MA and Rubin LL. (1998). Proc. Natl. Dickens M, Rogers JS, Cavanagh J, Raitano A, Xia Z, Acad. Sci. USA, 95, 3984 ± 3989. Halpern JR, Greenberg ME, Sawyers CL and Davis RJ. Blangy AE, Vignal E, Schmidt S, Debant A, Gauthier- (1997). Science, 277, 693 ± 696. Rouviere C and Fort P. (2000). J. Cell Sci., 113, 729 ± 739. Franke TF, Yang SI, Chan TO, Datta K, Kazlauskas A, Bobak D, Moorman J, Guanzon A, Gilmer L and Hahn C. Morrison D, Kaplan DR and Tsichlis PN. (1995). Cell, 81, (1997). Oncogene, 15, 2179 ± 2189. 727 ± 736. Boehm JE, Chaika OV and Lewis RE. (1999). J. Biol. Chem., Gauthier-Rouviere C, Vignal E, Meriane M, Roux P and 274, 28632 ± 28636. Montcourier P. (1998). Mol. Cell. Biol., 9, 1379 ± 1394. Bokoch GM, Vlahos CJ, Wang Y, Knaus UG and Traynor- Genot EM, Arrieumerlou C, Ku G, Burgering BM, Weiss A Kaplan AE. (1996). Biochem. J., 315, 775 ± 779. and Kramer IM. (2000). Mol. Cell. Biol., 20, 5469 ± 5478. BonniA,BrunetA,WestAE,DattaSR,TakasuMAand Gomez J, Martinez AC, Gonzalez A and Rebollo A. (1998). Greenberg ME. (1999). Science, 286, 1358 ± 1362. Immunol. Cell Biol., 76, 125 ± 134. Brown K, Franzoso G, Baldi L, Carlson L, Mills L, Lin YC, Gutkind JS, Lacal PM and Robbins KC. (1990). Mol. Cell. Gerstberger S and Siebenlist U. (1997). Mol. Cell. Biol., Biol., 10, 3806 ± 3809. 17, 3021 ± 3027. Hall A. (1998). Science, 279, 509 ± 514. Chan TO, Rittenhouse SE and Tsichlis PN. (1999). Annu. HanJ,Luby-PhelpsK,DasB,ShuX,XiaY,MostellerRD, Rev. Biochem., 68, 965 ± 1014. Krishna UM, Falck JR, White MA and Broek D. (1998). Chen YR and Tan TH. (2000). Int. J. Oncol., 16, 651 ± 662. Science, 279, 558 ± 560. Chiariello M, Marinissen MJ and Gutkind JS. (2000). Mol. Hengartner MO. (2000). Nature, 407, 770 ± 776. Cell. Biol., 20, 1747 ± 1758. Horii Y, Beeler JF, Sakaguchi K, Tachibana M and Miki T. Coso OA, Chiariello M, Yu JC, Teramoto H, Crespo P, Xu (1994). EMBO J., 13, 4776 ± 4786. N, Miki T and Gutkind JS. (1995). Cell, 81, 1137 ± 1146. Davis RJ. (2000). Cell, 103, 239 ± 252.

Oncogene Activation of Akt and cell survival by Rac-related GTPases C Murga et al 216 Ise K, Nakamura K, Nakao K, Shimizu S, Harada H, Ichise Reif K, Nobes CD, Thomas G, Hall A and Cantrell DA. T, Miyoshi J, Gondo Y, Ishikawa T, Aiba A and Katsuki (1996). Curr. Biol., 6, 1445 ± 1455. M. (2000). Oncogene, 19, 2951 ± 2956. Rodriguez-Viciana P, Warne PH, Dhand R, Vanhaesebroeck KatohH,YasuH,YamaguchiY,AokiJ,FujitaH,MoriK B, Gout I and Fry MJ. (1994). Nature, 370, 527 ± 532. and Negishi M. (2000). Mol. Cell. Biol., 20, 7378 ± 7387. Rubio I, Rodriguez-Viciana P, Downward J and Wetzker R. Kimmelman AC, Osada M and Chan AM. (2000). Oncogene, (1997). Biochem. J., 326, 891 ± 895. 19, 2014 ± 2022. Sabapathy K, Jochum W, Hochedlinger K, Chang L, Karin Klippel A, Reinhard C, Kavanaugh WM, Apell G, Escobedo M and Wagner EF. (1999). Mech. Dev., 89, 115 ± 124. MA and Williams LT. (1996). Mol. Cell. Biol., 16, 4117 ± Sable CL, Filippa N, Hemmings B and Van OE. (1997). 4127. FEBS Lett., 409, 253 ± 257. Konishi H, Matsuzaki H, Tanaka M, Takemura Y, Kuroda ShimokeK,YamagishiS,YamadaM,IkeuchiTand S, Ono Y and Kikkawa U. (1997). FEBS Lett., 410, 493 ± Hatanaka H. (1999). Brain Res. Dev. Brain Res., 112, 498. 245 ± 253. Kuan CY, Yang DD, Samanta Roy DR, Davis RJ, Rakic P Subauste MC, Von Herrath M, Benard V, Chamberlain CE, and Flavell RA. (1999). Neuron, 22, 667 ± 676. Chuang TH, Chu K, Bokoch GM and Hahn KM. (2000). Levresse V, Butter®eld L, Zentrich E and Heasley LE. J. Biol. Chem., 275, 9725 ± 9733. (2000). J. Neurosci. Res., 62, 799 ± 808. Tolias KF, Cantley LC and Carpenter CL. (1995). J. Biol. Lores P, Morin L, Luna R and Gacon G. (1997). Oncogene, Chem., 270, 17656 ± 17659. 15, 601 ± 605. Tournier C, Hess P, Yang DD, Xu J, Turner TK, Nimnual A, Madge LA and Pober JS. (2000). J. Biol. Chem., 275, 15458 ± Bar-Sagi D, Jones SN, Flavell RA and Davis RJ. (2000). 15465. Science, 288, 870 ± 874. Minamoto T, Mai M and Ronai Z. (2000). Cancer Detect. Whitmarsh AJ, Cavanagh J, Tournier C, Yasuda J and Davis Prev., 24, 1 ± 12. RJ. (1998). Science, 281, 1671 ± 1674. Murga C, Fukuhara S and Gutkind JS. (2000). J. Biol. Wymann MP and Pirola L. (1998). Biochim. Biophys. Acta., Chem., 275, 12069 ± 12073. 1436, 127 ± 150. MurgaC,LaguingeL,WetzkerR,CuadradoAandGutkind Yang FC, Kapur R, King AJ, Tao W, Kim C, Borneo J, JS. (1998). J. Biol. Chem., 273, 19080 ± 19085. Breese R, Marshall M, Dinauer MC and Williams DA. Nishida K, Kaziro Y and Satoh T. (1999). Oncogene, 18, (2000). Immunity, 12, 557 ± 568. 407 ± 415. Zhang S, Han J, Sells MA, Cherno J, Knaus UG, Ulevitch Pahl HL. (1999). Oncogene, 18, 6853 ± 6866. RJ and Bokoch GM. (1995). J. Biol. Chem., 270, 23934 ± Perona R, Montaner S, Saniger L, Sanchez-Perez I, Bravo R 23936. and Lacal JC. (1997). Genes Dev., 11, 463 ± 475.

Oncogene