bPix-enhanced p38 Activation by Cdc42/Rac/PAK/MKK3/6-mediated Pathway IMPLICATION IN THE REGULATION OF MEMBRANE RUFFLING*

bPix (PAK-interacting exchange factor) is a recently identified guanine nucleotide exchange factor for Rho family small G protein Cdc42/Rac. The protein interacts with p21-activated protein kinase (PAK) through its SH3 domain. We examined the effect of bPix on MAP kinase signaling and cytoskeletal rearrangement in NIH3T3 fibroblast cells. Overexpression of bPix enhanced the activation of p38 in the absence of other stimuli and also induced translocation of p38 to the nucleus. This bPix-induced p38 activation was blocked by coexpression of dominant-negative Cdc42/Rac or kinase-inactive PAK, indicating that the effect of bPix on p38 is exerted through the Cdc42/Rac-PAK pathway and requires PAK kinase activity. The essential role of bPix in growth factor-stimulated p38 activation was evidenced by the blocking of platelet-derived growth factor-induced p38 activation in the cells expressing bPix SH3m (W43K) and bPix DHm (L238R,L239R). In addition, SB203580, a p38 inhibitor, and kinase-inactive p38 (T180A,Y182F) blocked membrane ruffling induced by bPix, suggesting that p38 might be involved in mediating bPixinduced membrane ruffling. The results in this study suggest that bPix might have a role in nuclear signaling, as well as in actin cytoskeleton regulation, and that some part of these cellular functions is possibly mediated by p38 MAP kinase.

␤Pix (PAK-interacting exchange factor) is a recently identified guanine nucleotide exchange factor for Rho family small G protein Cdc42/Rac. The protein interacts with p21-activated protein kinase (PAK) through its SH3 domain. We examined the effect of ␤Pix on MAP kinase signaling and cytoskeletal rearrangement in NIH3T3 fibroblast cells. Overexpression of ␤Pix enhanced the activation of p38 in the absence of other stimuli and also induced translocation of p38 to the nucleus. This ␤Pix-induced p38 activation was blocked by coexpression of dominant-negative Cdc42/Rac or kinase-inactive PAK, indicating that the effect of ␤Pix on p38 is exerted through the Cdc42/Rac-PAK pathway and requires PAK kinase activity. The essential role of ␤Pix in growth factor-stimulated p38 activation was evidenced by the blocking of platelet-derived growth factor-induced p38 activation in the cells expressing ␤Pix SH3m (W43K) and ␤Pix DHm (L238R,L239R). In addition, SB203580, a p38 inhibitor, and kinase-inactive p38 (T180A,Y182F) blocked membrane ruffling induced by ␤Pix, suggesting that p38 might be involved in mediating ␤Pixinduced membrane ruffling. The results in this study suggest that ␤Pix might have a role in nuclear signaling, as well as in actin cytoskeleton regulation, and that some part of these cellular functions is possibly mediated by p38 MAP kinase.
The Rho family GTPases have emerged as key regulators that mediate extracellular signaling pathways leading to the formation of polarized actin-containing structures such as stress fibers, membrane ruffles, lamellipodia, and filopodia. Besides changes in cytoskeletal architecture, these GTPases mediate diverse biological events, including stimulation of DNA synthesis, cellular transformation, and signaling to the nucleus (for recent reviews, see Refs. 1 and 2). Although direct association of serine/threonine kinases or signaling proteins with the individual GTPases is critical for downstream effects, the precise mechanism by which these GTPases regulate actin cytoskeleton and exhibit these biological activities remains to be determined (3,4). Among these effectors, the binding partner of Cdc42 and Rac, p21-activated kinase (PAK) 1 activity has been linked to Cdc42 and Rac-mediated cytoskeletal changes. The important roles that PAK plays as an effector of Cdc42/Rac signaling have been established from genetic and biochemical studies in yeasts and mammalian cells (reviewed in Refs. 5 and 6). The repertoire of signaling pathways that are responsive to activation by Rho family members is rapidly expanding, and many of these have shown apparent multiplicity of PAK-mediated signaling pathways (7).
Recently, the PAK-interacting exchange factor (Pix/COOL/ p85SPR) family was cloned by several groups, including our laboratory (8 -11). The Pix family has Dbl homology and a flanking PH domain, which are conserved in all of the guanine exchange factors for Rho GTPases. Pix can induce membrane ruffling, with an associated activation of Rac1 (9); however, the extent and selectivity between Pix isoforms to activate Cdc42 and Rac are not clear (10,12). Pix interacts tightly with the regulatory N terminus of PAK via its SH3 domain. As shown by Daniels et al. (12), ␣Pix stimulates PAK kinase activity through exchange factor-dependent and -independent mechanisms. Additionally, the guanine nucleotide exchange activity was positively modulated by PAK binding. However, the exact role of Pix in regulating PAK localization and function in the actin rearrangement remains to be clarified. Pix was included in a protein complex containing paxillin, PAK, Nck, and p95PKL (paxillin-kinase linker), a member of ARF-GAP family, and this suggests a possibility for the integration of the ARF and Rho family signal transduction at the cytoskeleton (13,14). It was also reported that the guanine nucleotide exchange activity of Pix could be activated by direct association of the p85 regulatory subunit of phosphatidylinositol 3-kinase (15). Interaction of Pix with a variety of signaling proteins suggests that Pix might have an important role in mediating the effects of extracellular signals, e.g. growth factors, extracellular matrix, stress, and cytokine, to cytoskeletal rearrangement.
It was reported that Cdc42 and Rac also regulate the JNK/ SAPK and p38 signaling pathway and the ability of PAK to regulate these same MAP kinases pathways (16 -18). There are several reports showing that Dbl-related proteins, the activators of Rho GTPases, have roles in the regulation of several biological activities, including transformation, metastasis, cy-toskeletal reorganization, and transcriptional activation (7). However, relatively little was known of the exact roles of these GEFs in the MAP kinase pathway, whereas Cdc42 and Rac were known as potent SAPK and p38 activators (19 -21). Despite the coincidence of two signalings leading to actin reorganization and MAP kinase activity by Cdc42/Rac/PAK in many cases, there were few reports about cross-talk between these two signalings (22).
In the present study, we demonstrate that the overexpression of ␤Pix in NIH3T3 fibroblast cells induces activation of p38 through a Cdc42/Rac-PAK-MKK3/6-mediated pathway, and this activation of p38 is necessary for the formation of membrane ruffles induced by ␤Pix overexpression.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-All materials for culture were purchased from Life Technologies, Inc. NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% bovine calf serum and maintained in 10% CO 2 at 37°C. Transient transfection of cells with mammalian expression vectors was performed using Lipo-fectAMINE PLUS according to manufacturer's instructions.
Mismatches are indicated by lowercase letters. The mutation was verified by automatic DNA sequencing (Applied Biosystems).
Cell Lysis, Immunoprecipitation, Immunoblotting, and Immune Complex Kinase Assay-To measure the activities of p38 and Erk, cells on a 100-mm dish were transfected with pcDNA3FLAG-p38 or -Erk together with other mammalian expression vectors of interest. At 40 h after transfection, cells were washed three times with ice-cold PBS and lysed in immunoprecipitation buffer (1% Triton X-100, 50 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM sodium vanadate, 10 M leupeptin, and 1.5 M pepstatin). Lysates were clarified by centrifugation at 15,000 ϫ g for 10 min. Protein concentration in the supernatant was determined using Bio-Rad protein assay reagent (Bio-Rad). Clarified lysates of 500 g of total protein were incubated with 4 After visualization of phosphorylated ATF2 or MBP by autoradiography (first panels), the same membranes were probed with anti-FLAG M2 antibody (third panels) and then stained with Coomassie Brilliant Blue to normalize the amount of total ATF2 or MBP (second panels). The fourth panels show the amount of transfected Myc-␤Pix in 20 g of total cell lysate using anti-Myc 9E10 antibodies. Data shown are representative of three independent experiments. MBP (for Erk kinase assay) and incubated at 30°C for 30 min. Reactions were stopped by the addition of SDS-sample buffer (100 mM Tris, pH 6.8, 2% SDS, 10% glycerol, 5% ␤-mercaptoethanol, 0.25% bromphenol blue). After SDS-polyacrylamide gel electrophoresis, proteins were transferred onto a polyvinylidine difluoride membrane (Millipore), and the membrane was subjected to autoradiography. The same membrane was then washed with 0.1% Triton X-100 containing PBS and immunoblotted using anti-FLAG M2 antibody. After visualization of kinase proteins by ECL reagent (Amersham Pharmacia Biotech), the membrane was stained with Coomassie Brilliant Blue to normalize the amount of substrates. To visualize active MKK3/6 protein, phosphorylation-specific MKK3/6-specific antibody (500:1) and MKK3 antibody (500:1) were used, respectively (New England Biolabs, Inc.). Horseradish peroxidase-conjugated secondary antibodies were purchased from Jackson ImmunoResearch Laboratories, Inc.

RESULTS AND DISCUSSION
The effect of ␤Pix on MAP kinases was examined using NIH3T3 fibroblast cells that were double transfected with FLAG-tagged MAP kinase and Myc-tagged ␤Pix. Transfectants were subject to immune complex kinase assay and phosphorylation of each substrate was visualized by autoradiography. As shown in Fig. 1A, ␤Pix augmented p38 activity about 3.5 times. We observed similar results in the immunoblotting analysis using phosphorylation-specific p38 antibody (data not shown). ␤Pix had no effect on Erk kinase activity in several separated experiments (Fig. 1B), and JNK activity was slightly increased upon ␤Pix overexpression (data not shown). Under these conditions, strong activation of Erk kinase by PDGF was observed. Next, we examined the effect of ␤Pix overexpession on the p38 activation in a single cell level using phosphorylation-specific p38 antibody. In ␤Pix-overexpressing cells, we could observe increased phospho-p38 signal, especially in the nucleus, compared with untransfected cells in the same field that had no or very weak signal (Fig. 2, upper panels). The treatment of SB203580, a p38-specific inhibitor, blocked ␤Pix-induced p38 activation (Fig. 2, lower panels).
We next generated two mutants of ␤Pix, ␤Pix SH3m (W43K) and ␤Pix DHm (L238R,L239S), whose SH3 domain and Dblhomology domain were changed, respectively. Coimmunoprecipitation assay and glutathione S-transferase-pulldown assay showed that ␤Pix SH3m cannot bind to PAK (data not shown). ␤Pix DHm has been reported to have no exchange activity,  Fig. 1. B, after 24 h of transfection, media were changed with serum-free DMEM, and cells were cultured in serum-free media for 16 h. PDGF was treated at 20 ng/ml for 15 min, and immune complex kinase assays were performed as in A. The fourth panels show the amount of transfected FLAG-␤Pix and FLAG-p38 in 20 g of total cell lysate using anti-FLAG M2 antibody. Data shown are representative of three to four independent experiments. while wild-type ␤Pix acted as a GEF for Rac1 in vivo (9). Unlike wild-type ␤Pix, overexpression of these two mutants failed to activate p38 (Fig. 3), indicating that PAK or other SH3 domainbinding proteins and GEF activity of ␤Pix are essential for transmission of the extracellular signaling to p38 MAP kinase activation. p38 has been reported to be activated by several extracellular stimuli, such as growth factors, cytokines, and stresses (25,26). We also examined the involvement of ␤Pix in PDGF-induced p38 activation. In ␤Pix-expressing cells, p38 activity was increased about four times upon PDGF treatment (Fig. 3B, lanes Vector and WT). However, in cells expressing ␤Pix SH3m or ␤PIX DHm, PDGF-induced p38 activation was not evident (Fig. 3B, lanes W43K and L238R,L239S), suggesting that ␤Pix might mediate p38 activation by PDGF in vivo. However, it should be noted that in either case of ␤Pix SH3m or DHm, expression of mutant ␤Pix still induced membrane ruffling (data not shown). The extent of ruffling was not much different from that of wild-type ␤Pix in case of ␤Pix SH3m and reduced by 20 -30% in case of ␤Pix DHm. In contrast, expres-sion of the ␤Pix SH3m-DHm double mutant failed to induce membrane ruffling and also did block PDGF-induced membrane ruffling (data not shown). Results suggest that disruptions of both PAK interaction (SH3) domain and GEF activity are necessary to block ␤Pix-induced membrane ruffling. Blocking the function of either domain was not sufficient to inhibit membrane ruffling. This could be explained by GEF-dependent and -independent PAK activation by Pix (12) or by the role of PAK as a upstream activator of Rac (27). On the other hand, inhibition of p38 by either ␤Pix SH3m or DHm (Fig. 3) indicates that both PAK interaction (SH3) domain and GEF activity are required for ␤Pix-induced p38 activation. ␤Pix-induced membrane ruffling in the case of ␤Pix SH3m and DHm despite p38 inhibition is probably due to incomplete inhibition of p38 by either mutant. Although ␤Pix SH3m and DHm could block p38 activation by PDGF upstream, the basal p38 activity could concert with other Rac (and/or PAK) effector(s) to induce membrane ruffling.
Rho family GTPases regulate JNK and p38, as appeared in the previous reports that the dominant-negative form of Cdc42 or Rac blocked p38 activity and transfection of active Cdc42 or Rac caused a large enhancement of p38 activity (16 -18). To elucidate the hierarchy of Cdc42/Rac and ␤Pix, the effect of dominant-negative (T17N) Cdc42 or Rac1 on the p38 activation by ␤Pix were examined. T17N mutants of Cdc42 and Rac1 completely inhibited ␤Pix-enhanced p38 activation as expected (data not shown), indicating that Cdc42 and Rac1 act downstream of ␤Pix, possibly being activated by the GEF activity of ␤Pix (9).
The PAK family of protein kinases has been suggested as a main target of the Cdc42/Rac1, and Pix tightly binds to a proline-rich motif of PAK and regulates its kinase activity (9,12). But it is also suggested that PAK could act upstream of Rac1 mediating lamellipodia formation through interaction with Pix (27). To address the role of PAK in ␤Pix-induced p38 activation, we utilized wild-type PAK1 or the K299R mutant of PAK1 whose kinase activity was eliminated. Immune complex kinase assay after triple transfection of FLAG-tagged p38, HAtagged wild-type, or mutant PAK1 and Myc-tagged ␤Pix showed that kinase-inactive PAK1 blocked p38 activation induced by ␤Pix (Fig. 4A). To determine whether the blocking of ␤Pix-induced p38 activation by kinase-inactive PAK1 was due to loss of kinase activity or due to dominant-negative effects of the expression of mutant PAK1 that could sequester interacting proteins, we examined ␤Pix-induced p38 activation in cells expressing the PID, which can block endogenous PAK kinase activity (Fig. 4B). Expression of PID abolished ␤Pix-induced p38 activation. Under the same conditions, inactive PID (PID L107F) did not block ␤Pix-induced p38 activation. The results suggest that PAK kinase activity is essential for ␤Pix-induced p38 activation. We next investigated the effect of ␤Pix on MAP kinase kinases, MKK3 and MKK6, which are known to phosphorylate and activate p38 MAP kinase, using phosphorylation-specific MKK3/6 antibody. ␤Pix overexpression led to a large enhancement of MKK3/6 activity (Fig. 5).
It has been reported that PDGF stimulation of some adherent cells leads to an induction of membrane ruffles in a Rac1 activity-dependent pathway (27). When we examined the effect of ␤Pix overexpression on actin structure by double immunofluorescence assay, we found that actin structure modified by ␤Pix overexpression mimicked the PDGF-induced membrane ruffles (Fig. 6). 2 In porcine aortic endothelial cells, PDGF treatment enhanced chemotaxis in addition to ruffling, and these effects were blocked by SB203580, suggesting that PDGF-induced membrane ruffling and chemotaxis are dependent on p38 activity (25). In this regard, to test whether ␤Pix-induced membrane ruffle is also mediated by p38 activity or not, we examined the effect of a p38-specific inhibitor, SB203580. As shown in Fig. 6A, under the ␤Pix-transfected condition, control cells formed enhanced membrane ruffles, whereas 10 M SB203580-treated cells failed to make membrane ruffles. In addition, the localization of FLAG-tagged ␤Pix was cytoplasmic in the SB203580-treated cells, in sharp contrast with the membrane and edge-localized FLAG-␤Pix in control cells. We observed similar results in the cells expressing kinase-inactive p38 (T180A,Y182F) (Fig. 6B, p38AF/␤Pix).
There is now increasing evidence showing the involvement of p38 in actin cytoskeleton regulation and/or cell migration (23). PDGF activates p38 through a Ras-dependent pathway that is important for actin reorganization and cell migration in porcine aortic endothelial cells (25). Activation of p38 MAP kinase pathway by growth factors and inflammatory cytokines regulates smooth muscle cell migration (24). In this report, p38 activates MAP kinase-activated protein kinase-2 and -3 (MAP-KAPK2 and -3), and these kinases phosphorylate various substrates including heat shock protein 27 (HSP27). Phosphorylated HSP27 appears to modulate the polymerization of actin and is proposed to play a role in actin cytoskeletal remodeling. In our experiments, SB203580, at the concentration physiologically blocking p38 activity, inhibited ␤Pix-induced membrane ruffling. It seems likely that p38 has a role in mediating Pix/ PAK/Cdc42/Rac signaling to the cytoskeletal rearrangement. It is reported that activation of LIM-kinase by PAK1 couples Cdc42/Rac GTPase signaling to actin cytoskeletal dynamics (28,29). LIM-kinase is known to catalyze phosphorylation of cofilin, thereby inactivating actin depolymerizing activity and leading to accumulation of actin filaments (30,31). We cannot rule out the possibility that ␤Pix also acts downstream of p38, where it is proximal to actin structure regulation (Pix-Cdc42/ Rac/PAK-LIM kinase-cofilin-actin) and p38 activation is a prerequisite for Pix/PAK/Cdc42/Rac action in actin regulation, rather than downstream of ␤Pix.
In many cases, the actin reorganization and MAP kinase activation arose from the same extracellular stimuli, such as the ligands of several receptor tyrosine kinases, and were mediated by the same cytoplasmic signaling components during the signal transduction pathway (16,19). Cdc42 and Rac have been implicated in JNK/SAPK and p38, and to a lesser degree, Erk kinase activation (16,17,32). We tested the involvement of ␤Pix in the MAP kinase pathway and found that ␤Pix activates p38 significantly (Fig. 1). It is well established that growth factor stimulation of Ras leads to the activation of the Raf/ MEK/Erk pathway, and Ras is also involved in signaling to the cytoskeleton (25). In recent reports, Erk can be activated by Rho family proteins, and Cdc42/Rac selectively interacts and activates MEK kinases (32,33). In our experiment, ␤Pix did not activate Erk kinase activity, excluding the possibility that ␤Pix acts upstream of the Erk-mediated mitogenic signaling pathway in NIH3T3 fibroblast cells. It is still unclear whether ␤Pix has a role in the downstream of Ras.
Zhang et al. (18) proposed that Cdc42 and Rac might regulate p38 activity through the downstream effector, PAK1. In our experiment, kinase-inactive PAK1 blocked p38 activity, supporting the results of others (Refs. 17 and 18 and Fig. 4). It is possible that PAK-Pix interaction enhances both kinase activity of PAK and the GEF activity of Pix, as suggested by others (5,9). ␤Pix might activate PAK, MKK3/6 and p38 in two ways: by activating Cdc42/Rac1 (and effector PAK) through its GEF activity and by interacting directly with the PAK regulatory N terminus. The complexity of the signaling involving ␤Pix might arise from the dual property of ␤Pix, which interacts with the Cdc42/Rac downstream effector, PAK, and also acts as a GEF of Cdc42/Rac1. We propose a model of the Pix/PAK/Rho GTPase-mediated p38 activation signaling pathway that is shown in Fig. 7. ␤Pix might be activated by extracellular stimuli such as PDGF, then activate Cdc42 and Rac through its GEF activity. PAK can be activated by Cdc42/Rac binding and also by the interaction with ␤Pix SH3 domain. PAK activates MAPK kinase, MKK3 and/or MKK6, leading to p38 activation. p38 activation has a role in actin cytoskeletal change, such as lamellipodia formation and membrane ruffling. It remains to be resolved what is the upstream activator of ␤Pix. Yoshii et al. (15) reported that ␣Pix is activated by interaction with phosphatidylinositol 3-kinase and we also have observed the association between ␤Pix and phosphatidylinositol 3-kinase in vivo. 3 We are now investigating other upstream regulators of ␤Pix and extracellular stimuli that activate ␤Pix. Our present study suggests that signaling pathways through ␤Pix, in addition to Cdc42/Rac and PAK, are orchestrated to yield a coordinated response of cytoskeletal and nuclear events to extracellular stimuli.