Role of MLK3-mediated Activation of p70 S6 Kinase in Rac1 Transformation*

The signaling pathways that mediate the transforming activity of the Rac1 GTPase remain to be determined. In the present study, we used effector domain mutants of the constitutively activated Rac(61L) mutant that display differential transforming activities and differential activation of downstream effector pathways to investigate the contribution of p70 S6 kinase (p70S6K) to Rac1 transformation and to decipher the signaling pathways leading from Rac1 to p70S6K. First, we found that Rac1 transforming activity could be dissociated from Rac1 activation of p70S6K. A weakly transforming Rac1 mutant retained the ability to activate p70S6K, whereas some potently transforming effector mutants were impaired in their ability to activate p70S6K. These data suggest that p70S6K is not necessary to promote full Rac1 transforming activity. We also found a strong correlation between the ability of the Rac(61L) effector mutants to activate p70S6K and their ability to activate the JNK mitogen-activated protein kinase. We found that the MLK3 serine/threonine kinase activated JNK and p70S6K, whereas activation of p70S6K by Rac(61L) was significantly inhibited by dominant-negative MLK3. Additionally, the ability of the Rac(61L) effector mutants to activate MLK3 correlated well with their ability to activate p70S6Kand JNK. Taken together, these results provide evidence that Rac1 coordinately activates p70S6K and JNK via MLK3 activation. Finally, we found that co-expression of wild type, but not kinase-dead, MLK3 significantly inhibited Rac1 transforming activity. These results suggest that MLK3 may be a negative regulator of the growth-promoting and transforming properties of Rac1.

The signaling pathways that mediate the transforming activity of the Rac1 GTPase remain to be determined. In the present study, we used effector domain mutants of the constitutively activated Rac(61L) mutant that display differential transforming activities and differential activation of downstream effector pathways to investigate the contribution of p70 S6 kinase (p70 S6K ) to Rac1 transformation and to decipher the signaling pathways leading from Rac1 to p70 S6K . First, we found that Rac1 transforming activity could be dissociated from Rac1 activation of p70 S6K . A weakly transforming Rac1 mutant retained the ability to activate p70 S6K , whereas some potently transforming effector mutants were impaired in their ability to activate p70 S6K . These data suggest that p70 S6K is not necessary to promote full Rac1 transforming activity. We also found a strong correlation between the ability of the Rac(61L) effector mutants to activate p70 S6K and their ability to activate the JNK mitogen-activated protein kinase. We found that the MLK3 serine/threonine kinase activated JNK and p70 S6K , whereas activation of p70 S6K by Rac(61L) was significantly inhibited by dominant-negative MLK3. Additionally, the ability of the Rac(61L) effector mutants to activate MLK3 correlated well with their ability to activate p70 S6K and JNK. Taken together, these results provide evidence that Rac1 coordinately activates p70 S6K and JNK via MLK3 activation. Finally, we found that co-expression of wild type, but not kinase-dead, MLK3 significantly inhibited Rac1 transforming activity. These results suggest that MLK3 may be a negative regulator of the growth-promoting and transforming properties of Rac1.
Rac1 is a member of the Rho family of small GTPases (1,2). Like other Rho family members, Rac1 cycles between active GTP-bound and inactive GDP-bound states and functions as a binary molecular switch that regulates a number of diverse cellular processes. Rac1 contributes to actin cytoskeleton reorganization by inducing the polymerization of monomeric actin, leading to lamellipodium formation and membrane ruffling (3). Rac1 can also regulate gene expression by activating various transcription factors (4). For example, Rac1 stimulates the activity of the c-Jun NH 2 -terminal kinase (JNK) 1 and p38 mitogen-activated protein kinases (MAPKs), which in turn activate the transcription factors c-Jun (via JNK) and ATF-2 (via JNK and p38) (1,2).
There is now considerable evidence that Rac1 plays an important role in cell proliferation and can promote growth transformation. It has recently been demonstrated that Rac1 is required for cell cycle progression (5) and that activated Rac1 stimulates increased expression of cyclin D1 (6) and the function of E2F transcription factors (7), which in turn are positive regulators of cell cycle progression. Furthermore, constitutively active mutants of Rac1 have been shown to promote tumorigenic transformation of rodent fibroblasts (8,9), invasion of T-cell lymphomas (10), and increased motility and invasiveness of breast carcinoma cells (11). In addition, Rac1 function is required for the transforming activity of Ras, Dbl family oncoproteins (e.g. Vav) (12), and other oncoproteins (e.g. Mas) (13). Thus, a critical role for aberrant Rac1 function in human oncogenesis has been proposed (2).
A critical aspect of understanding how Rac1 is able to regulate such diverse cellular processes and to cause transformation involves identifying the downstream effector targets and signaling pathways activated by Rac1. However, this has become a very complex question because of the growing number of effectors known to interact with Rac1 and the wide spectrum of signaling pathways associated with Rac1 activation (1, 2). Known or candidate effectors of Rac1 now number over 28 and include the PAK and mixed lineage kinase (MLK) serine/threonine kinases, p70 S6 kinase, phosphatidylinositol 3-kinase, POSH, and POR-1.
One powerful approach for establishing the contribution of specific effector-mediated pathways to specific Rac1 functions has been the use of effector domain mutants of Rac1 (6,14,15). The general aim of this approach has been to introduce amino acid substitutions into regions of Rac1 that are involved in effector interaction. This includes the core effector sequences of Rac1 (residues 25-45) as well as sequences immediately upstream of this region and sequences in the COOH terminus of Rac1 (2). We previously established such mutants and showed that they exhibit differential impairment in transforming activity as well as in activation of specific signaling pathways (6). These studies showed that Rac1 transforming activity did not correlate with the activation of any one specific signaling pathway. For example, we identified mutants that retained transforming activity, yet showed impaired ability to activate JNK, and mutants with impaired transforming activity that retained full ability to activate JNK. Thus, we were able to dissociate Rac1 activation of JNK, as well as p38, serum response factor (SRF), and lamellipodia formation from Rac1 transforming activity. Therefore, despite our considerable knowledge on Rac1 effectors and signaling pathways, the key effector pathway(s) by which Rac1 causes growth transformation still remains to be identified.
Shortly after our initial study, the 70-kDa S6 kinase (p70 S6K ) was described as a potential new effector of Rac1 function (16). p70 S6K is a serine/threonine kinase that phosphorylates the S6 ribosomal subunit (17,18). This leads to an increase in the rate of translation of the class of 5Ј-TOP mRNA transcripts, which encode critical components of the cellular translational apparatus, thus facilitating an increase in the overall rate of protein translation. Increased protein translation has been shown to contribute to cellular growth transformation. In addition, p70 S6K is activated by a variety of mitogenic stimuli (17,19) and antibody inhibition of p70 S6K can block cell cycle progression (20), suggesting that p70 S6K may play a role in cellular transformation. Recently, Martin and colleagues (21) demonstrated the importance of p70 S6K in Src oncoprotein transformation. Because it was shown that activated Rac1 can associate with p70 S6K in vitro and that activated Rac1 stimulates activation of p70 S6K (16), it is plausible that p70 S6K may be a key effector that mediates Rac1 transformation.
Because the contribution of p70 S6K to Rac1 transformation has not been determined, and the exact mechanism by which Rac1 activates p70 S6K remains unclear, we utilized Rac1 effector domain mutants to address these two issues. First, we found that Rac1 transforming potential did not correlate with the ability of Rac1 to activate p70 S6K , suggesting that p70 S6K activation is neither necessary nor sufficient for Rac1 transforming activity. Second, we determined that MLK3 can serve as an effector to facilitate the coordinate activation of p70 S6K and JNK by Rac1. Finally, we determined that MLK3 function antagonizes Rac1 transforming activity. Thus, our results exclude the importance of MLK3-mediated signaling in Rac1 transformation.

EXPERIMENTAL PROCEDURES
Molecular Constructs-The pCGN-hygro expression vectors encoding Rac(61L) and the associated effector domain mutants, and the pZIP-NeoSV(x)1 vector encoding activated Raf-22W, have been described and characterized previously (6,22). The pCDNA3 expression vectors encoding hemagglutinin epitope-tagged MLK3 and MLK3-KR were kindly provided by S. Gutkind and have been described previously (23). The Flag epitope-tagged expression vector for JNK1 and the bacterial expression vector glutathione S-transferase-Jun-  were kindly provided by M. Karin (24). The pCDNA3 expression vector encoding Flag epitope-tagged p70 S6 kinase was generated using PCR-mediated DNA amplification using the cDNA sequence of p70 S6 kinase in the PMT2-p70␣I expression vector (kindly provided by J. Avruch) (25). Briefly, PMT2-p70␣I plasmid DNA was digested with NotI and SpeI, and ligated with a NotI-SpeI-digested PCR fragment, prepared using PMT2-p70␣I as the template and using the forward primer (5Ј-TAC GCG GCC ATG GAC TAC AAG GAC GAC GAT GAC AAG GCG GCA GGA GTG TTT GAC ATA GAC-3Ј) encoding a NotI site followed by an initiator Met, sequence encoding the Flag epitope (DYKDDDDK), and succeeding NH 2 -terminal p70 sequence, and the reverse primer (5Ј-GCC CCC TTT ACC AAG TAC CCG-3Ј). The resulting Flag-p70 S6K construct was then excised from pMT2 and ligated into the EcoRI site in the pCDNA3 (Invitrogen) mammalian expression vector. The resulting construct was verified by double-stranded DNA sequence analysis.
Cell Culture and Transformation Assays-293 and NIH 3T3 cells were maintained in Dulbecco's modified Eagle's medium (high glucose) supplemented with 10% fetal bovine serum or newborn calf serum, respectively. 293 cells were transfected using LipofectAMINE reagent (Invitrogen) as described by the manufacturer. Thirty h after transfection, the medium was changed to Dulbecco's modified Eagle's medium containing 0.5% fetal bovine serum, and then after an 18-h incubation, lysates were prepared as described below. NIH 3T3 cells were transfected by calcium phosphate co-precipitation, and focus formation assays were performed as described previously (26). For cooperation assays with Raf-22W (an NH 2 -terminal truncated and constitutively activated mutant of human Raf-1), cells in 60-mm dishes were cotransfected with 10 ng of pZIP-raf-22W, 500 ng of pCGN-rac1(61L), and, in some cases, 1 g of either pCDNA3-MLK3 or pCDNA3-MLK3-KR. Cognate empty vectors of each plasmid were used as controls where appropriate. After 14 -16 days of growth, the number of foci on each plate was assessed, and then plates were stained with 0.2% crystal violet and photographed. For soft agar assays, NIH 3T3 cells stably expressing Rac(61L) and either MLK3 or MLK3-KR were generated by calcium phosphate transfection followed by selection for 14 days in growth medium supplemented with hygromycin (200 mg/ml) and G418 (400 mg/ml). Expression of the Rac1 and MLK3 mutant proteins was confirmed by Western blot analysis (data not shown). These cells were then seeded into 60-mm dishes in growth medium containing 0.3% soft agar, and colony formation was assessed after 18 days as described previously (26).
Immunoprecipitation and in Vitro Kinase Assays-The activity of p70 S6K was analyzed in 293 cells following transfection of plasmid DNAs encoding Flag epitope-tagged p70 S6K and various Rac1 mutants, MLK3, and SEK1. Cells were transfected in 60-mm dishes, maintained for 24 h in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, and then starved for 18 h. Lysates were then collected in 500 l of Tris/Triton lysis buffer (20 mM Tris, pH 7.5, 137 mM NaCl, 1% Triton X-100, 10% glycerol, 2 mM EDTA) containing protease and phosphatase inhibitors. Lysates were normalized to 250 g of total cellular protein in 500 l, and Flag-p70 S6K was immunoprecipitated with anti-Flag antibody and protein A-agarose (Santa Cruz Biotechnology). The immunoprecipitate/protein A-agarose beads were washed twice with Tris/Triton lysis buffer and twice with phosphatebuffered saline. Kinase assays were then performed in 20 mM Hepes (pH 7.3), 10 mM ␤-glycerophosphate, 1.5 mM EGTA, 0.1 mM Na 3 VO 4 , 1 mM dithiothreitol, 10 mM MgCl 2 , and 50 M [␥-32 P]ATP (3000 Ci/ mmol) plus 250 M S6 peptide substrate (RRLSSLRA) in a final reaction volume of 60 l. The assay was performed at 30°C for 20 min with constant shaking in an Eppendorf 5436 Thermomixer. The kinase reaction was terminated by the addition of 20 l of 100 mM EDTA (pH 7.0). The samples were centrifuged for 3 min at 13,000 ϫ g, and then 40 l of supernatant was spotted on Whatman P-81 paper. The papers were washed five times in 10% phosphoric acid, and the radioactivity incorporated into the bound S6 peptide was determined by liquid scintillation counting. Assays for MLK3 were performed in an identical fashion, except that MLK3 was immunoprecipitated with anti-MLK3 antibody (Santa Cruz Biotechnology) followed by protein A-agarose, and 4 g of myelin basic protein was used as the substrate. JNK activity was analyzed by immunoprecipitating Flag epitope-tagged JNK-1 from 293 cells and performing in vitro kinase assays with 2 g of glutathione S-transferase-Jun-(1-79) as the substrate, as described previously (27).
Luciferase Reporter Assays-Reporter plasmids where the luciferase gene was under the control of a minimal promoter that contained c-Jun or SRF-responsive DNA elements, which have been described previously, were used to evaluate the specificity of the dominant negative MLK3-KR and SEK-AL mutants (6). We found that co-expression of either dominant negative caused selective inhibition of c-Jun versus SRF activation (data not shown), verifying that the activity of these mutants are not the result of nonspecific inhibitory effects.
Western Blot Analysis-Cellular lysates containing ϳ25 g of total cellular protein (taken from samples normalized as described above) were boiled in 2ϫ sodium dodecyl sulfate (SDS) sample buffer, resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), and then transferred to polyvinylidene difluoride membranes (Immobilon-P, Millipore). Resolved proteins were visualized by probing the membranes with anti-Flag, anti-Rac1 (23A8, Upstate Biotechnology, Inc.), or anti-MLK3 primary antibodies and appropriate secondary antibodies, followed by enhanced chemiluminescence (Amersham Biosciences, Inc.) according to manufacturer instructions.

Rac1
Activation of p70 S6K Is Not Required for Transformation-Our previous studies found no correlation between the ability of Rac1 to activate various signaling pathways and to cause transformation of NIH 3T3 fibroblasts (6). Subsequent studies showed that activated Rac1 can associate with and stimulate p70 S6K activity (16). Because p70 S6K is an important translational regulator of proteins important for cell growth p70 S6 Kinase and Rac1 Transformation and is required for cell cycle progression, it is reasonable to predict that p70 S6K may be a key effector for Rac1 transformation. To evaluate this possibility, we determined whether the ability of the Rac1 mutants to activate p70 S6K correlated with their transforming activity. These analyses involved the characterization of mutants of the constitutively activated and transforming Rac(61L) mutant that harbored single amino acid substitutions in the core effector domain (spanning residues 25-45) (6).
We showed previously that the Rac(61L/40C) mutant retains potent transforming activity, whereas the Rac(61L/26D) mutant is greatly impaired in transforming activity (Table I) (6). Therefore, we assessed the ability of these Rac1 mutants to activate p70 S6K . To optimize for the sensitivity of our p70 S6K activation assays, we used transient expression assays in 293 cells, rather than NIH 3T3 cells that exhibited suboptimal transfection efficiencies, for these analyses. Cells were transfected with an expression construct encoding Flag epitopetagged p70 S6K along with expression constructs encoding the various effector domain mutants of activated Rac(61L). Flag-p70 S6K was then immunoprecipitated, and its activity was measured using an in vitro kinase assay. Expression of Rac(61L) caused a 3-fold activation of p70 S6K over basal levels, whereas expression of the strongly transforming Rac(61L/40C) mutant failed to cause significant activation of p70 S6K (Fig. 1). In contrast, the weakly transforming Rac(61L/26D) mutant caused activation of p70 S6K to levels comparable with that of the nonmutated Rac(61L) protein. Western blot analysis was performed to verify that comparable levels of Rac1 and Flag-p70 S6K were expressed in each experimental sample. These results show that Rac1 activation of p70 S6K does not correlate with transforming activity and suggest that Rac1 activation of p70 S6K is neither necessary nor sufficient for Rac1 transforming activity.
Rac1 Activates p70 S6K and JNK via a Common Signaling Pathway Mediated by MLK3-It has been shown previously that Rac(61L) causes activation of p70 S6K (16). However, although it was determined that p70 S6K and Rac1 could be found in a complex together, it was not determined whether this interaction was a result of direct binding or if it was instead mediated by other proteins, leaving unclear the exact mechanism by which Rac1 activates p70 S6K . Therefore, to further evaluate how Rac1 causes p70 S6K activation, we compared the ability of Rac(61L) effector domain mutants to activate p70 S6K with their ability to activate other signaling pathways. In particular, because we determined previously that the 40C mutant that was impaired in p70 S6K activity was impaired in JNK activation (6), we analyzed additional Rac1 mutants to determine whether Rac1 activation of p70 S6K and JNK shared the same structural requirements. These analyses included two additional mutants with missense mutations in the core effector region of Rac1 (31V and 43D). Additionally, we recently determined that missense mutations in other regions of Rac1 also cause impairment of effector function. 2 Three of these mutants (52L, 66A, and 148R) were also included in these analyses.
In our analyses of seven Rac(61L) mutants, we found that the ability of each mutant to activate p70 S6K correlated with its ability to activate JNK (Fig. 2). For example, the 26D, 31V, and 43D mutants retained the same activity as Rac(61L) to activate both JNK and p70 S6K , whereas the 148R mutant reproducibly stimulated a higher level of activity for both p70 S6K (Fig. 2B)  Table I. For p70 S6K activity, HEK 293 cells were transfected with 1 g of a plasmid expressing Flag epitope-tagged p70 S6K together with 0.5 g of plasmid DNA encoding the indicated Rac1 mutant or cognate empty vector (pCGN-hygro). Cells were serum-starved for 18 h before lysate preparation. Flag-p70 S6K was immunoprecipitated, and kinase activity was determined as described under "Experimental Procedures" and is expressed as -fold activation relative to empty vector. The levels of Rac1 and Flag-p70 S6K proteins present in each cell lysate were assessed by Western blot (WB) analysis with anti-Rac1 or anti-Flag antibodies as described under "Experimental Procedures." Data shown represent the average of three independent experiments performed in duplicate. and JNK 2 than did Rac(61L). In contrast, like Rac(61L/40C), the 52L and 66A mutants of Rac(61L) both showed impaired abilities to activate JNK as well as p70 S6K . When considered together, the ability of all seven mutants of Rac(61L) to activate p70 S6K and JNK to a similar extent strongly suggests that Rac1 utilizes a common effector to activate each of these kinases.
MLK3 May Be a Common Effector for Rac1 Activation of p70 S6K and JNK-We next wanted to determine whether Rac1 activated a common effector to activate p70 S6K and JNK. Previous studies had implicated PAK, MLK3, and POSH as possible effectors for mediating Rac1 activation of JNK (23, 28 -30). However, our previous analyses of effector domain mutants of Rac1 demonstrated that Rac1 activation of JNK could be dissociated from activation of PAK, indicating that PAK was not the effector for activating this MAPK cascade (6). Therefore, we focused our attention on MLK3. MLK3 is a ubiquitously expressed serine/threonine kinase, which contains several domains known to mediate protein-protein interactions, including a CRIB (Cdc42/Rac interactive binding) motif. One study determined that MLK3 binds to active Rac1 in vivo and that MLK3 mediates the activation of JNK by Rac1 (23), although a second study with Rac1 effector domain mutants dissociated MLK3 from JNK activation by Rac1 (30).
Previous analyses showed that overexpression of wild type MLK3 caused activation of JNK (23). Therefore, we first determined if MLK3 is an activator of p70 S6K . We found that transient overexpression of wild type MLK3 stimulated p70 S6K activity ϳ3-fold over basal levels, similar to the activation of p70 S6K stimulated by Rac(61L) (Fig. 3). We then determined if a kinase-dead dominant negative mutant of MLK3 (designated KR) could impair the ability of Rac(61L) to activate p70 S6K . Previous studies have shown that co-expression of MLK3-KR blocked Rac1 activation of JNK (23). We found that co-expression of MLK3-KR also completely blocked the ability of Rac(61L) to cause activation of p70 S6K (Fig. 3). The specificity of the inhibition seen with MLK3-KR was verified by determining that this dominant negative mutant selectively blocked Rac1 activation of a c-Jun-responsive reporter plasmid, when compared with a SRF-responsive reporter plasmid (data not shown). We showed previously that Rac1 activates c-Jun and SRF by distinct effector signaling pathways (6). These results suggest that MLK3 may serve as a common effector to facilitate Rac1 activation of both p70 S6K and JNK.
To further evaluate the possibility that MLK3 functions as a common effector for Rac1 activation of JNK and p70 S6K , we determined if the ability of the various Rac1 mutants to activate MLK3 correlated with their ability to activate p70 S6K and JNK. These analyses involved transient overexpression of each Rac1 mutant together with an expression construct encoding wild type MLK3 (23). MLK3 was then immunoprecipitated and analyzed for activation using an in vitro kinase assay. We found that there was a direct correlation between the ability of each mutant to activate p70 S6K , JNK, and MLK3 (Fig. 4). For example, like Rac(61L), the 43D and 148R mutants showed strong activation of MLK3, as well as activation of p70 S6K and JNK. In contrast, the 52L mutant, which showed impaired activation of p70 S6K and JNK, also showed impaired activation  Table I. HEK 293 cells were cotransfected with expression plasmids encoding Flag epitope-tagged p70 S6K (1 g) and Rac1 effector mutants (0.5 g) harboring mutations either within the core effector domain (A) or regions outside of the core effector domain (B), as indicated. p70 S6K activity was assessed as for Fig. 1 and is expressed as the percentage of activation relative to the level of activation stimulated by Rac(61L). Average activation of p70 S6K by Rac(61L) relative to empty vector was 3.2-fold in these assays. Western blot (WB) analysis was performed to assess the levels of Flag-p70 S6K and Rac1 in the lysates used for immunoprecipitation as described in Fig. 1. Data shown represent the average of three independent experiments performed in duplicate.
FIG. 3. MLK3 mediates activation of p70 S6K by Rac1. HEK 293 cells were transfected with 1 g of a plasmid expressing Flag epitopetagged p70 S6K together with plasmids encoding the indicated proteins (1 g of MLK3, MLK3-KR, or cognate empty vector (pCDNA3), or 0.5 g of Rac(61L). p70 S6K activity was assessed as for Fig. 1 and is expressed as -fold activation relative to empty vector. Western blot (WB) analysis was performed to assess the levels of Flag-p70 S6K in the lysates used for immunoprecipitation as described in Fig. 1. Data shown represent the average of three independent experiments performed in duplicate. of MLK3. In addition, the 40C and 66A mutants, which showed reduced abilities to activate both p70 S6K and JNK, also showed reduced abilities to activate MLK3. Only the 26D mutant showed some differences in its ability to activate these three signaling pathways. This mutant showed a slight reduction in its ability to activate JNK, but its ability to activate MLK3 and p70 S6K was unchanged. Nevertheless, when taken together with the data shown in Fig. 3, these analyses suggest that MLK3 can serve as a common effector to mediate Rac1 activation of JNK and p70 S6K .
Our finding that p70 S6K and JNK appear to lie on a common MLK3-mediated signaling pathway downstream of Rac1 prompted us to investigate the extent of overlap of these two pathways. The SEK serine/threonine kinase is the immediate upstream activator of JNK, and previous studies showed that SEK was required for the activation of JNK by both Rac1 and MLK3 (23,31). To determine whether Rac1 activation of p70 S6K was also dependent on SEK, we assessed the ability of a dominant negative mutant of SEK1 (designated SEK-AL) to inhibit the activation of p70 S6K and JNK by Rac1 and by MLK3. Co-expression of dominant negative SEK-AL caused a partial (30%) reduction in Rac(61L)-stimulated activation of both p70 S6K and JNK (Fig. 5), indicating an involvement of SEK in Rac1 activation of p70 S6K and JNK. Similarly, expression of dominant negative SEK-AL also reduced MLK3-stimulated activation of both p70 S6K and JNK by almost 50% (Fig. 5), whereas expression of wild type SEK had no effect (data not shown), suggesting that SEK mediates the activation of both kinases by MLK3. The specificity of the inhibition seen with SEK-AL was verified by determining that this dominant negative mutant selectively blocked Rac1 activation of a c-Junresponsive reporter plasmid, when compared with a SRF-responsive reporter plasmid (data not shown). Taken together, these results suggest that Rac1 uses a common signaling pathway, mediated by MLK3 and SEK, to stimulate the activity of both p70 S6K and JNK.
MLK3 Inhibits the Transforming Activity of Rac1-Although our analyses above indicated that MLK3/p70 S6K activation did not correlate with Rac1 transforming activity, the possibility remained that this pathway still contributed to transformation. Support for such a role comes from the observation that overexpression of wild type MLK3 can cause growth and morpho-logic transformation of NIH 3T3 mouse fibroblasts (32). Therefore, we used two transformation assays to determine whether MLK3 activity could modulate Rac1 transforming potential.
First, we and others have previously demonstrated that coexpression of activated Rac(61L) and an activated form of Raf-1 (the NH 2 -terminal truncated Raf-22W mutant) results in cooperative focus-formation activity in NIH 3T3 cells (8,9). As we have observed previously, cultures transfected with expression plasmid DNA encoding activated Rac(61L) or Raf-22W alone (Fig. 6) showed no or little focus-forming activity, whereas co-expression of activated Rac(61L) and Raf-22W showed significant focus-forming activity (ϳ50 foci/dish). We found that co-expression of wild type MLK3 reduced the focus-forming activity of Rac(61L)ϩRaf-22W by more than 60%, whereas expression of kinase-dead MLK3-KR resulted in only a slight inhibition (Fig. 6), suggesting that the kinase activity of MLK3 exerts a negative influence on the ability of Rac1 to transform cells.
The results shown in Fig. 6 may be caused by MLK3 inhibition of Raf, rather than Rac1, activity. Therefore, we also used a second approach to evaluate the ability of MLK3 to regulate Rac(61L) transforming activity. We showed previously that NIH 3T3 cells stably expressing activated Rac(61L) are able to grow in an anchorage-independent manner in soft agar (6). To determine whether MLK3 inhibited the ability of Rac1 to promote anchorage-independent cell growth, we performed soft agar assays using pooled populations of NIH 3T3 cells stably co-transfected with expression plasmids encoding Rac(61L) and vectors encoding MLK3, MLK3-KR, or empty expression vector. Mass populations of double drug-selected cells were then established for these analyses. Western blot analyses verified comparable levels of activated Rac(61L) protein expression in the different cell lines (data not shown). Consistent with the results of the focus assays, cells co-expressing Rac(61L) and wild type MLK3 were severely inhibited in their ability to form colonies in soft agar (Fig. 7). In contrast, cells co-expressing Rac(61L) and MLK3-KR showed essentially the same colony forming activity as cells expressing Rac(61L) alone. These results, when combined with those from the focus formation analyses, indicate that MLK3 may be a negative regulator of Rac1 transformation.  Table I, respectively. HEK 293 cells were cotransfected with expression plasmids encoding MLK3 (1 g) and Rac1 effector mutants (0.5 g) harboring mutations either within the core effector domain (A) or regions outside of the core effector domain (B), as indicated. Cells were serum-starved for 18 h before lysate preparation. MLK3 was immunoprecipitated, and kinase activity was determined as described under "Experimental Procedures" and is expressed as the percentage of activation relative to the level of activation stimulated by Rac(61L). Average activation of MLK3 by Rac(61L) relative to empty vector was 5.4fold in these assays. The levels of MLK3 and Rac1 proteins present in each cell lysate were assessed by Western blot (WB) analysis with anti-MLK3 or anti-Rac1 antibodies as described under "Experimental Procedures." Data shown represent the average of three independent experiments performed in duplicate.

DISCUSSION
Constitutively active mutants of Rac1 have been shown to promote tumorigenic transformation of rodent fibroblasts, invasion of T-cell lymphoma cells, and increased motility and invasiveness of breast carcinoma cells (2), suggesting that Rac1 may directly contribute to multiple aspects of tumorigenesis. Rac1 function is also required for the transforming activity of Ras and other oncoproteins. Although a multitude of Rac1 effectors and Rac1-stimulated signaling activities have been identified, which effectors and signaling activities are important for Rac1 transformation remain to be identified. Recently, p70 S6K was identified as a potential effector of Rac1 (16). Because significant evidence has linked p70 S6K function to cellular proliferation (17), p70 S6K may be an important effector of Rac1 transformation. In this study, we have used effector do-main mutants of Rac(61L) that display differential activation of downstream effector pathways to investigate the contribution of p70 S6K to Rac1 transformation and to decipher the signaling pathways that link Rac1 to p70 S6K . First, we found that the transforming activity of different Rac(61L) effector domain mutants did not correlate with their ability to activate p70 S6K . Second, we determined that MLK3 can serve as a common effector to mediate Rac1 activation of JNK and p70 S6K . Finally, co-expression analyses showed that MLK3 activity antagonized, rather than promoted, Rac1 transforming activity. Thus, we conclude that MLK3-mediated signaling pathways do not promote Rac1 transformation.
Our previous use of effector domain mutants of activated Rac1 determined that Rac1 transforming activity did not correlate with any one signaling activity studied. Our assessment FIG. 5. Activation of p70 S6K and JNK by MLK3 is dependent on SEK. HEK 293 cells were transfected with 1 g of a plasmid expressing Flag epitopetagged p70 S6K (A) or Flag epitope-tagged JNK (B), together with 1 g of the indicated expression constructs or cognate empty vectors. Cell lysates were prepared as described previously and assayed for either p70 S6K or JNK activity as described under "Experimental Procedures." Kinase activity is expressed as -fold activation relative to empty vector. The levels of Flag-p70 S6K and Flag-JNK proteins present in each cell lysate were assessed by Western blot (WB) analysis with anti-Flag antibody as described under "Experimental Procedures." Data shown represent the average of four independent experiments performed in duplicate. of a series of mutants that varied in transforming activity when assayed in NIH 3T3 cells determined that Rac1 activation of the JNK and p38 MAPKs, activation of the PAK serine/threonine kinase, SRF activation, or stimulation of transcription from the cyclin D1 promoter, was each alone dispensable for Rac1 transforming activity (6). In the present study, we extend these observations to show that p70 S6K activation failed to correlate with Rac1 transformation. Rac(61L) is potently transforming and stimulates a high level of p70 S6K activity. However, the Rac(61L/40C) mutant, which retained full transformation potential, failed to activate p70 S6K , whereas the weakly transforming Rac(61L/26D) mutant still retained the ability to fully activate p70 S6K . These results suggest that p70 S6K activation is neither necessary nor sufficient to promote Rac1 transformation.
One interpretation of these data is that the key signaling activity that mediates Rac1 transforming activity remains to be identified. Recently, Par-6 has been identified as a Rac1 effector that promotes Rac1 transformation (32), and it will be interesting to determine whether Rac1 activation of Par-6 will correlate directly with the transforming activity of our Rac1 effector domain mutants. Alternatively, another interpretation is that Rac1 transformation can be mediated by the activation of different combinations of signaling activities, with no one signaling activity being critical. Thus, Rac1 transforming potential will not be strictly associated with any one activity, yet that activity can still contribute to transformation. Evidence for this possibility comes from our analyses of Rac1 activation of the NF-B transcription factor. Our analyses of different Rac(61L) effector domain mutants determined that NF-B activation could also be dissociated from Rac1 transforming ac-tivity (33). However, we recently determined that inhibition of NFB function can block Rac1-mediated transformation. 2 Hence, it remains possible that p70 S6K activation, although dispensable for Rac1 transformation, may still contribute to the transforming actions of Rac1. Constitutively activated and dominant negative mutants of p70 S6K would provide useful reagents to address this possibility. However, although such mutants have been described, our preliminary analyses of these mutants did not indicate sufficiently detectable gain or loss of function to be confident that they were useful reagents for such analyses.
It was demonstrated previously that constitutively activated Rac1 stimulates p70 S6K activity (16). Although Rac1 and p70 S6K could be found complexed together in vitro, it was not determined whether this interaction was a result of direct binding or if it was instead mediated by other proteins. Thus, it remains possible that p70 S6K activation by Rac1 may be indirect and mediated by some other Rac1 effector. To address this question, we compared the ability of Rac1 effector domain mutants to activate p70 S6K with their ability to activate other known Rac1 signaling pathways. We found that the ability of each Rac1 mutant to activate p70 S6K correlated with its ability to activate JNK. Additionally, we observed that the ability of Rac1 effector mutants to activate MLK3 correlated directly with their ability to activate of JNK and p70 S6K . Finally, we found that overexpression of wild type MLK3 alone can activate JNK and p70 S6K . Taken together, these data strongly suggest that Rac1 utilizes a common pathway, mediated by MLK3, to coordinately regulate each of these kinases. At what point does the pathway diverge downstream of MLK3? Our observation that a dominant negative mutant of SEK1, the immediate upstream activator of JNK, could block Rac1 activation of both JNK and p70 S6K suggests that it may diverge at either the level of SEK or JNK. Whether SEK or JNK can cause direct activation of p70 S6K has not been determined.
A mechanism where p70 S6K and JNK are coordinately regulated is in contrast to a finding in the previous report (16) that, based on differential activation profiles by extracellular stimuli such as growth factors and sorbitol, p70 S6K and JNK are activated by independent pathways. However, we have previously observed that p70 S6K and JNK can be activated by many of the same stimuli, including angiotensin II and other calcium agonists, protein synthesis inhibitors such as anisomycin, and UV irradiation (34,35). Furthermore, there is evidence that growth factors, which potently activate p70 S6K , can also activate JNK in a Rac1-dependent fashion (31,36). Finally, it seems reasonable to expect that the cell would have a mechanism to coordinately regulate gene transcription (via JNK) and mRNA translation (via p70 S6K ) to effect an efficient, robust response to mitogenic or stress stimuli.
To date, the mechanism by which Rac1 activates JNK has also not been clearly defined, although several candidate effectors have been implicated. Several groups have reported that PAK is involved (28,29). However, our previous work with Rac1 effector domain mutants demonstrated that Rac1 activation of JNK was not mediated by PAK (6). Additional studies have shown that members of the MLK family of kinases are potent activators of JNK (37)(38)(39)(40), and at least two groups have provided evidence that MLK3 phosphorylates SEK, leading to JNK activation (40,41). It has been further demonstrated that MLK3 binds to Rac1 and mediates Rac1-stimulated activation of JNK (37). On the other hand, another study identified a Rac1 mutant that failed to bind to MLK3 but was still able to activate JNK, suggesting Rac1 can activate JNK in an MLK3independent manner (30). However, in the present study we have demonstrated that all seven of the Rac1 mutants inves- p70 S6 Kinase and Rac1 Transformation tigated are able to activate MLK3 and JNK to the same extent, providing evidence consistent with those studies that propose MLK3 as the effector by which Rac1 activates JNK.
A recent study reported that MLK3 exhibits transforming properties, as stable expression of MLK3 in NIH 3T3 cells resulted in morphologic transformation and promoted growth in soft agar (32). These results prompted us to determine whether MLK3 contributes to Rac1 transformation. Unexpectedly, our results indicate that MLK3 acts as a negative regulator of Rac1 transformation, as we found that overexpression of wild type MLK3 potently inhibited the transforming properties of Rac1, when tested in both focus formation and soft agar assays. In addition, in contrast to the previous study, activated MLK3 alone did not exhibit transforming properties in our hands, as NIH 3T3 cells stably expressing MLK3 failed to form colonies in soft agar (Fig. 7), nor did they appear morphologically transformed compared with untransfected cells (data not shown). The reason for this discrepancy is currently unknown. One possibility is that it may be caused by differential ability of the MLK3 constructs to activate the ERK MAPK pathway, because the authors of that study observed that MLK3 was able to stimulate ERK activity, whereas we and others have not observed MLK3-stimulated ERK activity (data not shown) (23,40,41). Another possibility may be the different susceptibility of different strains of NIH 3T3 mouse fibroblasts to transformation. We previously observed significant differences in the transforming potencies of effector domain mutants of activated Ras when assayed in two different strains of NIH 3T3 cells (42).
In summary, our studies demonstrate that Rac1 may coordinately stimulate activation of p70 S6K and JNK via MLK3 and that p70 S6K activation can be dissociated from Rac1 transforming activity. Questions that remain include whether MLK3independent mechanisms exist for Rac1 activation of p70 S6K and whether the SEK/JNK pathway promotes p70 S6K activation. Additionally, whether MLK3 can promote both growth promoting and growth inhibitory functions, depending on signal and/or cell context, will require further analyses. Finally, whether MLK3 serves a similar role for Cdc42, in causing the coordinate activation of p70 S6K and JNK, and in antagonizing Cdc42 transforming activity, will be interesting to determine. Although the key effector(s) that mediates Rac1 transformation remains to be established, what is clear is that Rac1 function requires the complex interplay of a multitude of effector-mediated pathways.