Basic Fibroblast Growth Factor Stimulates Activation of Rac1 through a p85 βPIX Phosphorylation-dependent Pathway*

In a previous study (Shin, E. Y., Shin, K. S., Lee, C. S., Woo, K. N., Quan, S. H., Soung, N. K., Kim, Y. G., Cha, C. I., Kim, S. R., Park, D., Bokoch, G. M., and Kim, E. G. (2002) J. Biol. Chem. 277, 44417–44430) we reported that phosphorylation of p85 βPIX, a guanine nucleotide exchange factor (GEF) for Rac1/Cdc42, is a signal for translocation of the PIX complex to neuronal growth cones and is associated with basic fibroblast growth factor (bFGF)-induced neurite outgrowth. However, the issue of whether p85 βPIX phosphorylation affects GEF activity on Rac1/Cdc42 is yet to be explored. Here we show that Rac1 activation occurs in a p85 βPIX phosphorylation-dependent manner. A GST-PBD binding assay reveals that Rac1 is activated in a dose- and time-dependent manner in PC12 cells in response to bFGF. Inhibition of ERK or PAK2, the kinases upstream of p85 βPIX in the bFGF signaling, prevents Rac1 activation, suggesting that phosphorylation of p85 βPIX functions upstream of Rac1 activation. To directly address this issue, transfection studies with wild-type and mutant p85 βPIX (S525A/T526A, a non-phosphorylatable form) were performed. Expression of mutant PIX markedly inhibits both bFGF- and nerve growth factor (NGF)-induced activation of Rac1, indicating that phosphorylation of p85 βPIX is responsible for activation of this G protein. Both wild-type and mutant p85 βPIX displaying negative GEF activity (L238R/L239S) are similarly recruited to growth cones, suggesting that Rac1 activation is not essential for translocation of the PIX complex (PAK2-p85 βPIX-Rac1). However, expression of mutant p85 βPIX (L238R/L239S) results in retraction of the pre-existing neurites. Our results provide evidence that bFGF- and NGF-induced phosphorylation of p85 βPIX mediates Rac1 activation, which in turn regulates cytoskeletal reorganization at growth cones, but not translocation of the PIX complex.

The dynamics of the actin cytoskeleton is regulated by Rho family G proteins. To date, the most extensively characterized members of this family include Rac1, Cdc42, and RhoA (1). It is well established that in fibroblast cells, Rac1, Cdc42, and RhoA induce formation of lamellipodia and membrane ruffle, and filopodia and stress fiber, respectively (2). These G proteins regulate cytoskeletal rearrangement in a similar fashion in neuronal cells such as PC12, a model system in which the molecular mechanism of neurite outgrowth has been extensively investigated. However, their activation at neuronal growth cones exerts opposing effects on neurite outgrowth. Specifically, RhoA promotes growth cone collapse resulting in neurite retraction, whereas Rac1 and Cdc42 stimulate neurite outgrowth through the formation of lamellipodia and filopodia, respectively (3). Along the same lines expression of dominantnegative Rac1 or Cdc42 blocks nerve growth factor (NGF) 1induced neurite extension (4).
Rho GTPases cycle between inactive GDP-bound and active GTP-bound forms. Interconversion between these two forms is regulated by guanine nucleotide exchange factors (GEF), GTPase-activating proteins (GAP), and guanine nucleotide dissociation inhibitors (GDI). GEFs of the Dbl family stimulate activation of Rho GTPases by catalyzing GDP/GTP exchange of these G proteins (5)(6)(7). All members of this family contain the Dbl homology (DH) domain, which is responsible for catalytic activity. GEF proteins are activated in various ways, including phosphorylation by protein kinases (6 -8). Considerable evidence shows that p85 ␤PIX, a member of Dbl family GEF, is phosphorylated in response to extracellular stimuli (9 -11). However, the issue of whether this phosphorylation is associated with Rac1/Cdc42 activation remains to be resolved.
p85 ␤PIX functions as a specific GEF for Rac1/Cdc42, both in vitro and in vivo (9). The protein interacts with p21-activated kinase (PAK) and a group of multidomain proteins displaying Arf-GAP activity, specifically, Cat-1/Git1 (12), p95PKL (13), and p95-APP1 (14), through its SH3 and Git-binding domains (GBD), respectively. The LZ domain of p85 ␤PIX facilitates dimer formation (15). X-ray crystallographic analysis of the isolated DH domain suggests that Rac1 is also a component of the p85 ␤PIX complex (16). It seems thus likely that p85 ␤PIX makes a huge complex interacting with PAK2, p95 family member (presumably p95PKL in PC12 cells) and Rac1. Rac1 is considered a critical regulator in the membrane translocation of cellular complexes in a wide range of processes. Integrininduced recruitment of the PAK-PIX-PKL complex to focal adhesion for proper cell adhesion (17) or targeting of the complex for cell migration (18) is dependent on Rac1/Cdc42 activation. Rac1 is additionally a component of the NADPH oxidase complex, and its activity is required for the translocation of cytosolic oxidase subunits as well as NADPH oxidase activation (19). However, the issue of whether Rac1 activity is essential for targeting the multimeric p85 ␤PIX complex (p85 ␤PIX-PAK2-p95PKL-Rac1) in response to bFGF or NGF for neurite outgrowth remains to be determined.
In a previous study, we reported that bFGF, a potent stimulator of neurite outgrowth in PC12 cells, induces phosphorylation of p85 ␤PIX via the Ras/ERK/PAK2 pathway and that its translocation to growth cones is dependent on phosphorylation (11). Since one of the activation mechanisms of GEF involves phosphorylation, we postulate that bFGF-induced phosphorylation of p85 ␤PIX may result in activation of Rac1. Here, we show that phosphorylation of p85 ␤PIX is directly associated with Rac1 activation in the bFGF signaling. Basic FGF stimulates Rac1 activation in a p85 ␤PIX phosphorylation-dependent manner. However, p85 ␤PIX-mediated Rac1 activation is not essential for translocation of the p85 ␤PIX complex including Rac1, suggesting that the phosphorylated protein regulates activation of Rac1 and translocation of the complex via two distinct signaling pathways in bFGF-induced neurite outgrowth. Our results provide an exemplar case of the mechanism by which Ras and Rac1, key small G proteins in cell regulation, are linked in a linear array for neurite outgrowth.
Transient Transfection-PC12FW cells were seeded on 60-mm culture dishes or 20 g/ml poly-L-lysine-coated cover slips. A mixture of 5 g DNA and 5 l LipofectAMINE 2000 was added to culture dishes according to the manufacturer's instructions. After 24 -48 h, transfected cells expressing green fluorescent protein (GFP) were monitored with a fluorescence microscope (Olympus, CK-40).
Translocation of p85 ␤PIX and Its Effect on Neurite Outgrowth-To analyze translocation of p85 ␤PIX, PC12FW cells were cultured on poly-L-lysine-coated coverslips, induced to differentiate in culture media containing 10 ng/ml bFGF or 100 ng/ml NGF and 2% FBS for 48 h, and transfected with pEGFP-p85 ␤PIX using LipofectAMINE 2000. Transfected cells were incubated in DMEM containing 2% FBS in the presence or absence of 10 ng/ml bFGF or 100 ng/ml NGF for 24 -48 h. Cells were washed three times with phosphate-buffered saline (PBS), fixed in 4% paraformaldehyde/PBS for 15 min and permeabilized with 0.1% Triton X-100, 10% FBS, and 1% bovine serum albumin in PBS for 5 min. After fixation, coverslips were washed twice in PBS and mounted onto a glass slide with gelvatol. Fluorescence was visualized with a laser confocal microscope (MRC-1024, Bio-Rad Laboratories, Richmond, CA). To analyze the effect of translocated p85 ␤PIX on neurite outgrowth, PC12FW cells were induced to differentiate as described above. Cells were transfected with pEGFP-WT-PIX (wild-type) or pEGFP-MT-PIX (GEF-negative form, L238R/L239S). Six hours after transfection, culture medium was replaced with complete medium containing 10 ng/ml bFGF. Transfected cells were monitored by phase contrast and fluorescence microscopy with an inverted fluorescence microscope for 48 h.
Actin Staining-PC12FW cells were cultured on poly-L-lysine-coated cover slips and induced to differentiate with 10 ng/ml bFGF or with 100 ng/ml NGF as described in the above section. Cells were washed three times with PBS, fixed in 4% paraformaldehyde/PBS for 15 min and permeabilized with 0.1% Triton X-100, 10% FBS, and 1% bovine serum albumin in PBS for 5 min. After three rinses in PBS, cells were incubated in PBS containing 1% bovine serum albumin and 0.2 M phalloidin-TRITC for 30 min, washed three times in PBS and mounted onto a glass slide with gelvatol. Fluorescence was visualized with a laser confocal microscope.
Neurite Retraction Assay-Neurite retraction assay was conduced as previously described (20) with a slight modification. Briefly, PC12FW cells were differentiated for 24 h and then transiently transfected with plasmids encoding GFP fusion proteins of GST and wild-type or mutant p85 ␤PIX. After 6 h, cells were further differentiated with bFGF containing medium. Recording of changes in neurite length under a fluorescence microscope has begun at 24 h following transfection and lasted until 48 h. Cells with neurite length around two times their cell body length were initially selected.
Measurement of Cell Viability-Viability of cultured cells in dishes was assessed by a trypan blue exclusion test. Briefly, PC12FW cells were induced to differentiate for the indicated times and transfected with plasmids encoding GFP fusion proteins of wild-type or mutant p85 ␤PIX. Changes in neurite extension or retraction were monitored under a fluorescence microscope for 24 h (from 24 to 48 h following transfection). Cells were then stained with 0.4% trypan blue (Invitrogen) for 15 min, followed by washing two times with PBS.
Immunoprecipitation and Immunoblotting-Cells were washed twice with PBS and lysed in lysis buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 100 mM NaF, 10% glycerol, 1% Triton X-100, 200 M orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml leupeptin) for 1 h at 4°C. Proteins were immunoprecipitated with an appropriate antibody for 3 h at 4°C. Immunoprecipitates were collected by adding protein G Sepharose, and washed five times with lysis buffer and twice with PBS. Samples were fractionated by SDS-PAGE, and transferred to a polyvinylidene difluoride membrane in a Tris-glycinemethanol buffer (25 mM Tris base, 200 mM glycine, 20% methanol). Membranes were blocked with 3% skimmed milk in PBS for 1 h, incubated with primary antibodies for 1 h at room temperature, and washed three times (10 min each) with PBS containing 0.1% Tween-20. Membranes were blotted with secondary horseradish peroxidase-conjugated antibodies for 1 h at RT. After five washes with PBS and 0.1% Tween-20, signals were detected using enhanced chemiluminescence (ECL) reagent (Amersham Biosciences, Piscataway, NJ). In some cases, membranes were stripped and reprobed with different antibodies.
Statistical Analysis-Student's paired t test was applied for statistical analysis of neurite retraction assay using The SAS System for Windows (release 8.01) and the statistical significance was set at p Ͻ 0.05.

RESULTS
p85 ␤PIX Constitutively Binds to Rac1 through Its DH Domain-To investigate the molecular mechanism by which bFGF stimulates activation of Rac1, we employed a PC12FW model system in which the FGF receptor-1 is overexpressed (11). It is suggested that ␤PIX forms a stable ternary complex with activated Cdc42 and PAK3 (22). X-ray crystallographic analysis of the interaction between purified ␤PIX and G protein additionally suggested that the DH domain possesses binding capability in addition to catalytic activity on the G protein (16). To determine whether Rac1 binds to p85 ␤PIX in a complex cellular context, we employed immobilized GST or GST-Rac1 bound to glutathione-Sepharose beads as bait to precipitate endogenous p85 ␤PIX (Fig. 1A). No specific binding was observed between immobilized GST and p85 ␤PIX (Fig. 1A, lanes 1 and 2). However, GST-Rac1 strongly bound to p85 ␤PIX irrespective of bFGF stimulation (lanes 3 and 4). To further delineate the region of p85 ␤PIX responsible for Rac1 binding, cells were transfected with plasmids encoding the various domains of p85 ␤PIX (Fig. 1C), and lysates were affinity-precipitated with immobilized GST or GST-Rac1 (Fig. 1B). The various domains of p85 ␤PIX were expressed at similar levels (middle). In the control lanes GST did not precipitate DH-PH and SH3 domains (lanes 1 and 2, top). In contrast, GST-Rac1 specifically interacted with full-length p85 ␤PIX, DH and DH-PH domains (lanes 3, 5, and 7), but not with SH3 and PH domains (lanes 4 and 6). GBD and LZ domains did not bind GST-Rac1 (not shown). The data confirm that the DH domain of p85 ␤PIX is sufficient for Rac1 binding. Moreover, the inter- p85 ␤PIX Phosphorylation-dependent Activation of Rac1 actions between p85 ␤PIX and Rac1 are constitutive.
Basic FGF Stimulates Activation of Rac1 in a Dose-and Time-dependent Manner-Upon stimulation by agonists, Rho GTPases are converted to active GTP-bound forms, which specifically bind to and activate downstream effectors, such as PAK and Rho kinase (22). Therefore, the binding site of PAK, p21-binding domain (PBD), has been expressed as a fusion protein of GST and been employed to determine the in vivo activity of Rac1 (21,23). In this conventional GST-PBD pulldown assay GST-PBD is immobilized to glutathione-Sepharose beads and used to precipitate active Rac1. Here to specifically monitor Rac1 activation in the PIX complex, we made a slight modification of the conventional GST-PBD pull-down assay by using soluble GST-PBD instead of immobilized one. The principle of this assay, which we call GST-PBD binding assay, is depicted in Fig. 2A. Because the p85 ␤PIX protein forms a complex with PAK2 (9, 11) and Rac1 (Fig. 1), we first retrieved this complex by precipitation with anti-PAK2 antibody. This pre-clearance step seemed to eliminate background interfering proteins in the lysate, which results in enhancement of a signal-to-noise ratio. We next incubated the immunoprecipitate with soluble GST-PBD and detected GST-PBD bound to active Rac1 by immunoblotting with anti-GST antibody. PC12FW cells were starved and stimulated with the indicated concentrations of bFGF for 1 h (Fig. 2B). Lysates were immunoprecipitated with anti-PAK2 antibody to retrieve the PIX complex, and activated GTP-bound Rac1 in this complex was determined using soluble GST-PBD as described above. Activated Rac1, evaluated by the accumulation of GST-PBD over the untreated control, is initially observed at 5 ng/ml bFGF. Rac1 activity increases in proportion to bFGF concentrations up to 30 ng/ml. To monitor a time-course of Rac1 activation, cells were stimulated with bFGF (10 ng/ml) for the indicated times (Fig. 2C). After bFGF stimulation, Rac1 activation is apparent at 30 min and reaches a peak at 1 h, followed by a gradual decrease. However, the activity maintains at relatively higher levels, even at 4 h. These results indicate that Rac1 is activated by bFGF in a dose-and time-dependent manner.
ERK and PAK2 Function Upstream of bFGF-induced Rac1 Activation-Specific G proteins are activated via the phosphorylation of GEF (9 -10). We speculated that bFGF-induced phosphorylation of p85 ␤PIX might also operate for Rac1 activation. In a previous study, we showed that ERK and PAK2 are the upstream kinases for p85 ␤PIX phosphorylation (11). Accordingly, we examined whether these two kinases act as upstream regulators in bFGF-induced Rac1 activation. For this purpose, we used PD98059, a specific inhibitor of MEK, or a plasmid encoding a PAK-inhibitory domain (PID) (11) to block PAK2 activity. Cells were starved and treated with bFGF for 1 h. Lysates were processed as in Fig. 2. Basic FGF induced a ϳ4-fold increase in Rac1 activity (Fig. 3A, lane 2), consistent with the result from  4 and 5). These results strongly suggest that ERK/PAK2-mediated phosphorylation of p85 ␤PIX is closely associated with FIG. 2. Rac1 is activated in response to bFGF. A, schematic diagram of Rac1 assay. PC12FW cells were stimulated with bFGF and immune complex of PAK2-p85 ␤PIX-Rac1 was retrieved with anti-PAK2 antibody. This complex was then incubated with soluble GST-PBD, which can bind to an activated form of Rac1-GTP. Bound GST-PBD was detected by immunoblotting with anti-GST antibody. B and C, cells were starved for 24 h in the presence of 1.5 g/ml doxycycline and were stimulated with increasing concentrations of bFGF for 1 h (B) or with 10 ng/ml bFGF for the indicated times (C). Cells were lysed and equal amounts of proteins were immunoprecipitated with anti-PAK2 antibody. Immunoprecipitates were preincubated with 100 M GTP␥S at 37°C for 20 min and further incubated with soluble GST-PBD at 4°C for 2 h to detect activated Rac1. Beads were washed, resolved on 12% SDS-PAGE and immunoblotted with anti-GST antibody for Rac1 activation. The same blot was reprobed with anti-PAK2, -p85 ␤PIX, or -Rac1 antibody for equal precipitation and loading. Rac1 activation. It is well established that PAK2 interacts with p85 ␤PIX, and that two residues of PAK2, specifically, proline 185 and arginine 186, are critical in this association (9,11). To further determine whether PAK2 and p85 ␤PIX are involved in bFGF-induced Rac1 activation, we employed a PAK2 mutant (PAK2-P185A/R186A) that does not bind p85 ␤PIX. Cells were transfected with a plasmid encoding GFP wild-type or GFP mutant PAK2 and stimulated with bFGF. In cells expressing wild-type PAK2, bFGF induced strong Rac1 activation, as expected (Fig. 3B, lane 2). However, in cells expressing mutant PAK2, no significant Rac1 activation was observed (lanes 3 and  4), supporting the theory that PAK2-mediated phosphorylation of p85 ␤PIX regulates Rac1 activity.
Phosphorylation of p85 ␤PIX Mediates bFGF-induced Rac1 Activation-The above data (Fig. 3), although indirect, strongly suggest that p85 ␤PIX phosphorylation functions upstream of Rac1 activation. This theory was directly investigated using a non-phosphorylatable form of p85 ␤PIX. Serine 525 and threonine 526 of p85 ␤PIX have been identified as major phosphorylation sites following bFGF stimulation (11). Replacement of these two residues with alanine abolished the biological activities of p85 ␤PIX, such as translocation of the PIX complex and neurite outgrowth. We hypothesized that mutant p85 ␤PIX (S525A/T526A) is not capable of mediating bFGF-induced Rac1 activation. To determine whether this is the case, we employed two different assays to detect Rac1 activation, specifically, GST-PBD binding and GDP/GTP exchange assays. Cells were transfected with a plasmid encoding wild-type or mutant p85 ␤PIX, and stimulated with bFGF for 1 h. Lysates were immunoprecipitated with either anti-PAK2 antibody as in Figs. 2 and 3 (Fig. 4A, left) or anti-GFP antibody (Fig. 4A, right). Next, GST-PBD was added to detect the activated GTP-bound form of Rac1. Consistent with the result from Fig. 3A, bFGF treatment induced strong activation of Rac1 (Fig. 4A, lanes 1 and 2, left). Expression of wild-type p85 ␤PIX also resulted in appearance of a strong GST-PBD band, indicating the presence of activated Rac1 in immunoprecipitates with anti-PAK2 (Fig. 4A, lane 4,  left) or anti-GFP antibody (lane 2, right). However, in cells expressing mutant p85 ␤PIX, GST-PBD band intensities did not increase in response to bFGF (Fig. 4A, lane 6, left and lane  4, right). To confirm that phosphorylation of p85 ␤PIX affects its GEF activity on Rac1, GDP/GTP exchange activity of p85 ␤PIX was determined (Fig. 4B). Cells were processed as mentioned above, and anti-GFP immunoprecipitates were incubated with [ 35 S]GTP␥S in the presence of either GST (control) or GST-Rac1 (Fig. 4B). Irrespective of bFGF treatment or transfection with wild-type or mutant PIX, no significant changes in GEF activity on GST were detectible. Basic FGF induced a 1.8-fold increase in GDP/GTP exchange activity on Rac1 in cells expressing wild-type p85 ␤PIX. In contrast, nonphosphorylatable PIX mutant blocked bFGF-induced increase in the exchange activity. These results collectively indicate that phosphorylation of p85 ␤PIX regulates Rac1 activity in the bFGF signaling.
NGF Also Stimulates Rac1 Activation in a p85 ␤PIX Phosphorylation dependent Manner-NGF induces differentiation of PC12 cells through persistent ERK activation (24). Since we showed that p85 ␤PIX phosphorylation is mediated via the Ras/ERK cascade in the bFGF signaling (11), we reasoned that NGF treatment should result in phosphorylation of p85 ␤PIX. We thus tested whether NGF also induces phosphorylation of p85 ␤PIX, and whether this phosphorylation is dependent on ERK activation. Cells were pretreated with increasing concentrations of U0126, a specific inhibitor of both MEK1 and MEK2, prior to stimulation with NGF (100 ng/ml). NGF induced strong activation of ERK (Fig. 5A, middle) and phosphrylation of p85 ␤PIX (top), as determined by appearance of an upper slowly migrating band due to retarded mobility on electrophoresis. Comparison of intensity between the upper phosphorylated and the lower unphosphorylated forms showed that more than 90% of p85 ␤PIX are phosphorylated in response to NGF. U0126 pretreatment markedly inhibited NGF-induced p85 ␤PIX phosphorylation (Fig. 5A). Intensity of two forms was almost equal even at 5 M U0126 (lane 3), implying that ERK is responsible for p85 ␤PIX phosphorylation. These results prompted us to determine whether NGF-induced p85 ␤PIX phosphorylation also mediates Rac1 activation, as seen with bFGF. For this purpose cells were stimulated with NGF and a GST-PBD binding assay was performed as in the right column of Fig. 4A. Expression of wild-type p85 ␤PIX caused accumulation of enhanced GST-PBD band, implying that activated Rac1 is present in the GFP-p85 ␤PIX complex precipitated with anti-GFP antibody (Fig. 5B, lane 2). However, in cells expressing mutant p85 ␤PIX, intensity of GST-PBD band was not enhanced by NGF treatment (Fig. 5B, lane 4). These results suggest that NGF-induced Rac1 activation is also p85 ␤PIX phosphorylation-dependent, lending further support to our proposal that p85 ␤PIX phosphorylation is indeed a mechanism of Rac1 activation.
p85 ␤PIX-mediated Rac1 Activation Is Not Necessary for Translocation of the p85 ␤PIX Complex-Previously we showed that bFGF-induced translocation of the PAK2-p85 ␤PIX-p95PKL complex requires phosphorylation of p85 ␤PIX (11). Since p85 ␤PIX phosphorylation mediates Rac1 activation, it is interesting to determine whether Rac1 activity is involved in targeting of this complex. We thus examined whether Rac1 co-precipitates with mutant p85 ␤PIX (L238R/L239S) that does not exhibit guanine nucleotide exchange activity. PC12FW cells were transfected with a plasmid encoding either wild-type or mutant PIX, and lysates were precipitated with immobilized GST (control) or GST-Rac1 (Fig. 6A). No specific binding occurred between GST and p85 ␤PIX (lanes 1 and 2). Both wildtype and mutant PIX specifically bound to GST-Rac1, indicating that mutation of two leucine residues (Leu-238 and Leu-239) in the DH domain that abolishes GEF activity does not affect interactions between p85 ␤PIX and Rac1. Based on this constitutive nature of the binding between p85 ␤PIX and Rac1 (Figs. 1 and 6A), we hypothesized that Rac1 may move as a component of the mutant p85 ␤PIX (L238R/L239S) complex similar to wild-type p85 ␤PIX in response to bFGF. To test this idea, we determined whether mutant PIX targets the similar sites where wild-type p85 ␤PIX localizes. Cells were stimulated FIG. 4. Rac1 activation is linked to phosphorylation at S525/T526 of p85 ␤PIX. A, GST-PBD binding assay for Rac1 activation. Cells were transfected with pEGFP-WT-PIX or pEGFP-MT-PIX (mutagenized non-phosphorylatable form, S525A/T526A) and stimulated with 10 ng/ml bFGF for 1 h. Lysates were immunoprecipitated with anti-PAK2 (left) or anti-GFP (right) antibodies. GST-PBD binding assay was conducted as described in the legend to Fig. 2. Expression as GFP fusion proteins of wild-type or mutant p85 ␤PIX was evaluated by probing with anti-GFP antibody (left, bottom and right, top). Shown here are representative data from two independent experiments. B, guanine nucleotide exchange assay for Rac1 activation. Cells were transiently transfected with pEGFP-WT-PIX (wild-type) or pEGFP-MT-PIX (mutant) and stimulated with 10 ng/ml bFGF for 1 h. Lysates were immunoprecipitated with anti-GFP antibody. To determine the extent of GDP/GTP exchange by wild-type or mutant p85 ␤PIX, immunoprecipitates were incubated with 3 g of purified GST-Rac1 and 1 Ci of [ 35 S]GTP␥S at 30°C for 30 min. Bound [ 35 S]GTP␥S was determined by liquid scintillation counting (left). CPMs from the reactions using GST or GST-Rac1 as a substrate in the absence of bFGF stimulation were arbitrarily set as the control value, 1. Relative increase in GDP/GTP exchange compared with control is depicted in the graph. The histogram represents the combined data from three independent experiments performed in duplicate, and the error bars indicate the S.E. To determine expression levels of wild-type or mutant p85 ␤PIX, lysates from transfected cells were subjected to Western blotting with anti-GFP antibody (right).
to differentiate with bFGF, and transfected with a plasmid encoding either GFP wild-type or GFP mutant p85 ␤PIX. Localization of the complex was visualized with laser confocal microscopy (Fig. 6B). In the absence of bFGF, no significant translocation of the PIX complex was observed in either wildtype or mutant PIX-expressing cells (a-c and g-i). To determine the changes in actin cytoskeleton remodeling in response to bFGF, cells were stained with phalloidin. Actin was distributed diffusely in the cytoplasm in the absence of bFGF (b and h). When cells were treated with bFGF, actin was specifically concentrated at growth cones (e and k), indicating that actin dynamics is tightly regulated, concomitant with formation of growth cones. Wild-type p85 ␤PIX was targeted to bFGF-induced actin structures (lamellipodia at growth cones) (d-f). Interestingly, mutant p85 ␤PIX protein was also observed in areas where actin staining was apparent, although not as concentrated as wild-type protein (j-l). These results suggest that translocation of the PIX complex may occur independently of Rac1 activation. We further examined the localization of Rac1 and p85 ␤PIX using laser confocal microscopy (Fig. 6C). In the absence of bFGF stimulation, diffuse fluorescence of p85 ␤PIX (green) and Rac1 (red) was observed in the cytoplasm (a-c). Consistent with this finding, no growth cone-like structures were detected in these cells. In contrast, treatment of cells with bFGF resulted in specific concentration of both p85 ␤PIX and Rac1 at the growth cones (d-i). It is evident from the merged pictures of Rac1 and wild-type (f), and Rac1 and mutant p85 ␤PIX (i) that the proteins target the similar areas at growth cones (arrowheads). To determine whether NGF also induces similar responses, the above experiments were conducted except that bFGF was replaced by NGF. Basically NGF-induced responses were analogous to those as seen with bFGF (data not shown). These results confirm that Rac1 in conjunction with either wild-type or mutant p85 ␤PIX that does not possess GEF activity moves as a unit, supporting the conclusion that p85 ␤PIX-mediated Rac1 activation is not involved in translocation of the PIX complex.
The role of bFGF-induced Rac1 activation remains to be determined. The time course of Rac1 activation reveals that the process begins at least at 30 min following bFGF stimulation (Fig. 2C). Moreover, Rac1 co-localizes with p85 ␤PIX. Based on these findings, we speculate that Rac1 is initially translocated, not as an active participant, but as a passive passenger, to the sites where its activation occurs, and then exerts its effect on actin remodeling. Since at the fixed time points we observed translocation of the p85 ␤PIX complex per se in Fig. 6, B and C, the data may not represent the functional consequence of the PIX regulated-Rac1 activity. We therefore traced the fate of neurites expressing either wild-type or mutant p85 ␤PIX over a longer period (until 48 h) following transfection (Fig. 7). Cells expressing GFP fusion proteins of either wild-type or mutant p85 ␤PIX were identified using fluorescence microscopy (Fig.  7A, a-c and g-i) and neurite retraction was monitored for 2 days by phase contrast microscopy (Fig. 7A, d-f and j-l). Although the intensity of fluorescence faded with time, target cells were easily detected using additional information on the spatial arrangement of surrounding untransfected cells. In this experiment we use an irrelevant control, GST. The reason we selected GST as a control is that GST has been widely used as a fusion partner for expression of foreign proteins in the mammalian system. The neurites of cells expressing GFP-GST showed a little retraction to 82.7 Ϯ 6.3% (n ϭ 54) of their initial length at 48 h (Fig. 7B). This may reflect the effect of GFP or GST. Otherwise, this may be due to innate tendency to retract in cells bearing already extended neurites. Neurite extension was observed in some cells expressing wild-type p85 ␤PIX (Fig.  7A, d-f), but varied depending on the expression levels. At least no significant shortening of established neurites occurred due to expression of wild-type p85 ␤PIX, which showed 96.6 Ϯ 7.8% (n ϭ 59) at 48 h compared with their initial length (Fig. 7B). This is consistent with a previous result that showed a promoting effect on neurite extension of p85 ␤PIX during differentiation of PC12 cells (11). In contrast, cells expressing mutant (GEF-negative) p85 ␤PIX consistently exhibited neurite retraction, which resulted in 83.7 Ϯ 6.7% and 52.9 Ϯ 5.2% (n ϭ 59) of their initial length at 36 h and at 48 h, respectively. Some cells showed a near collapse of neurites at the end of monitoring, as illustrated in Fig. 7A (j-l). To rule out the possibility that this might occur due to deleterious effect of mutant (GEF-negative) p85 ␤PIX, a trypan blue exclusion test was conducted after termination of monitoring. No cells expressing mutant p85 ␤PIX were stained with trypan blue (data not shown), suggesting that mutant p85 ␤PIX itself does not have any general toxicity to cells. Notably, an untransfected cell (asterisk above the mutant PIX-expressing cells, Fig. 7A) shows similar morphological changes as wild-type PIX-expressing cells.
FIG. 5. NGF-induced Rac1 activation is also p85 ␤PIX phosphorylation-dependent. A, ERK mediates NGF-induced p85 ␤PIX phosphorylation. PC12FW cells were untreated or pretreated with the indicated concentrations of U0126 and then stimulated with NGF (100 ng/ml) for 30 min. Lysates were subjected to SDS-PAGE, and immunoblotting was conducted with anti-p85 ␤PIX (top), anti-phosphospecific ERK (middle), and anti-total ERK (bottom) antibodies. B, cells were transfected with pEGFP-WT-PIX or pEGFP-MT-PIX (mutagenized non-phosphorylatable form, S525A/T526A) and stimulated with 100 ng/ml NGF for 1 h. Lysates were immunoprecipitated with anti-GFP antibody. GST-PBD binding assay was conducted as described in Fig. 2. Expression as GFP fusion proteins of wild-type or mutant p85 ␤PIX was evaluated by probing with anti-GFP antibody (top). Representative data from two independent experiments are shown here.
FIG. 6. Re-localization and co-localization of wild-type and catalytically inactive mutant p85 ␤PIX (L238R/ L239S) is induced by bFGF. A, both wild-type and GEF negative mutant, p85 ␤PIX (L238R/L239S), bind to Rac1. PC12FW cells were transiently transfected with pEGFP-WT-PIX and pEGFP-MT-PIX (L238R/L239S). Cells were lysed and equal amounts of proteins were incubated with GST-Sepharose or GST-Rac1-Sepharose beads for 2 h at 20°C. Beads were washed, separated on 9% SDS-PAGE and immunoblotted with anti-GFP (top) or GST (bottom) antibody, respectively. B, both wild-type and mutant p85 ␤PIX translocate to lamellipodia at growth cones upon bFGF stimulation. PC12FW cells were allowed to differentiate for 48 h prior to transfection with a plasmid encoding either wild-type or mutant p85 ␤PIX (L238R/L239S). Twentyfour hours after transfection, cells were left untreated (a-c and g-i) or treated (d-f and j-l) with 10 ng/ml bFGF for 24 h at 37°C. Cells were fixed with paraformaldehyde and stained with phalloidin-TRITC to visualize rearrangement of the actin cytoskeleton (middle). Images from GFP (top) and actin staining are merged (bottom). C, both wild-type and mutant p85 ␤PIX co-localize to lamellipodia at growth cones with Rac1 upon bFGF stimulation. PC12FW cells were allowed to differentiate for 48 h, and co-transfected with pEGFP-WT-PIX plus pDsRed-Rac1 encoding Rac1 of red fluorescence (a-f) or pEGFP-MT-PIX (L238R/L239S) plus pDsRed-Rac1 (g-i). Twenty-four hours after transfection, cells were left untreated (a-c) or treated (d-i) with 10 ng/ml bFGF for 24 h at 37°C. After fixation with paraformaldehyde, co-localization of fluorescence was determined with laser confocal microscopy (pEGFP-PIX, top; pDsRed-Rac1, middle; merge, bottom). Representative images from more than three independent experiments are shown here. Scale bar, 10 m.

DISCUSSION
In the present study, we show for the first time that bFGFand NGF-induced phosphorylation of p85 ␤PIX mediates activation of Rac1. Firstly, the time-course of Rac1 activation correlates with that of p85 ␤PIX phosphorylation. Both Rac1 activation, as determined by GTP-bound forms of Rac1 (Fig. 2), and p85 ␤PIX phosphorylation (11) peak at 1 h following bFGF stimulation. Thereafter Rac1 activity and p85 ␤PIX phosphorylation gradually decrease, but remain at significantly higher levels up to 4 h. However, the time-course of Rac1 activation is slow in comparison with that of Madin-Darby canine kidney (MDCK) epithelial cells stimulated by hepatocyte growth factor (HGF) (25), NIH 3T3 cells stimulated by PDGF (23) and Swiss 3T3 cells stimulated by EGF, PDGF, or insulin (26,27). Upon stimulation of MDCK cells with HGF, Rac1 activation reached a maximum at 15 min. Peak Rac1 activity was achieved even earlier in PDGF-stimulated NIH 3T3 cells or EGF-, PDGF-, or insulin-stimulated Swiss 3T3 cells. Immediate and transient activation of Rac1 may be sufficient for growth factor-induced responses, such as spreading and dissociation of MDCK cells and membrane ruffling in Swiss 3T3 cells. However, in the case of long term processes such as neurite outgrowth, slow and persistent activation of Rac1 seems biologically more relevant. Secondly, activation of Rac1 is blocked upon inhibition of either ERK or PAK2 (Fig. 3), the upstream kinases for p85 ␤PIX phosphorylation in the bFGF signaling pathway (11), suggesting that these enzymes also function upstream of Rac1. Finally, expression of the non-phosphorylatable form of p85 ␤PIX  -c and g-i). Neurite retraction or extension in these cells was monitored for 1 day (from 24 to 48 h following transfection) under a phase contrast microscope (d-f and j-l). In images from j-l, an untransfected cell (asterisk) is marked for comparison. Magnification, ϫ200. B, effect of wild-type and mutant p85 ␤PIX on neurite retraction. Relative changes in neurite length after bFGF stimulation is expressed as mean Ϯ S.E. The data shown here are from five independent experiments. In each experiment, more than ten transfected cells whose initial neurite length is around twice the cell body length were tracked for each construct.
(S525A/T526A) prevents both bFGF-and NGF-induced accumulation of active GTP-bound forms and GDP/GTP exchange activities of Rac1 as well (Figs. 4 and 5). Collectively, our results strongly support the theory that p85 ␤PIX phosphorylation functions upstream of Rac1. However, Koh et al. (10) reported that PAK-mediated phosphorylation of PIX is independent of the exchange activity of PIX. This discrepancy may result from the different experimental conditions used by the two groups. In our study the native PIX complex, PAK2-PIX-(p95PKL)-Rac1, was immunoprecipitated in PC12 cells with or without bFGF stimulation. It is likely that following PAK2mediated PIX phosphorylation, phosphorylation-induced structural changes promote GEF activity of PIX. In contrast, Koh et al. (10) achieved PIX phosphorylation using constitutively active GST-PAK in vitro, which may phosphorylate residues other than the major sites, Ser-525 and Thr-526. In this case, a sum of these phosphorylation effects may not result in alteration in the GEF activity of PIX. An alternate possibility is that PIX in the control tube without GST-PAK is already phosphorylated, i.e. activated under culture conditions following transfection of the appropriate plasmids. Thus, elevated PIX activity in the control may mask the stimulatory effect of GST-PAKinduced PIX phosphorylation on GEF activity. In fact, all PIX isoforms (␣PIX, ␤1PIX, ␤2PIX) in the controls exhibit a linear increase in GEF activity on GST-Rac1 over the assay period of 40 min.
Accumulating evidence indicates that multiple pathways mediate growth factor-induced activation of Rac1 in various cellular responses. Upon stimulation with growth factors, their receptors with tyrosine kinase (RTK) activities are oligomerized, which trigger complex intracellular signaling cascades via phosphorylation. Regarding the Rac-activating pathways downstream of RTK, initial studies revealed Ras-dependent (RTK 3 Ras 3 phosphatidylinositol 3-OH kinase (PI3-K) 3 GEF 3 Rac) (27)(28)(29) and -independent (RTK 3PI3-K 3 GEF 3 Rac) ones in which PI3-K plays a major role as a common mediator. Activation of ␣PIX provides an example of Ras-independent pathways in which the p85 regulatory subunit of PI3-K or NCK, an adaptor molecule, directly links RTK to ␣PIX (30). Rac-GEF (such as Vav and Sos-1) is regulated by PI3-K in a Ras-dependent manner (31,32). Activation of these GEFs is mostly achieved via direct binding of phosphatidylinositol 3,4,5-phosphate (PIP 3 ), the product of PI3-K, to the PH domain or indirectly by phosphorylation (for instance, tyrosine phosphorylation of Vav increases GEF activity) (33). We previously showed that bFGF stimulates phosphorylation of p85 ␤PIX via the Ras 3 MEK 3 ERK 3 PAK2 pathway (11). Based on the above data and the present finding that phosphorylation of p85 ␤PIX is an activation mechanism for GEF activity, we propose a novel signaling pathway in which Ras cross-talks with Rac1 via the Ras 3 MEK 3 ERK 3 PAK2 3 p85 ␤PIX pathway. Thus Ras-mediated activation of Rac1 bifurcates at Ras into two distinct signaling pathways, specifically, the known Ras 3 PI3-K pathway and the Ras 3 ERK cascade. In view of the importance of the latter cascade in cell proliferation and differentiation, our data are functionally relevant and may implicate extension to other biological responses in which PAK2 (a bridge between ERK and p85 ␤PIX) takes part. The repertoire of upstream regulators of Rac is increasing with recent advances in this field. The involvement of phospholipase C-␥ in macrophage colony-stimulating factor-induced activation of MG5 microglial cells (34), CaMK II and protein kinase C in PDGF-stimulated NIH 3T3 cells (23), and Src in PDGFinduced cleavage of Alzheimer's amyloid precursor protein in C6 cells (35) has been reported, although additional information on the activation mechanisms is required. The evident complexity of Rac signaling may reflect the fact that Rac-mediated cellular events are critical, and that to satisfactorily meet the requirements under the diverse cellular context (i.e. diverse cell types or agonists), Rac activity should be precisely regulated with temporal and spatial coordination.
GEFs of the Dbl family are activated in various ways (5,6). One of the activation mechanisms involves phosphorylation of specific tyrosine or serine/threonine residues. In the basal state, GEF activity is suppressed by intramolecular interactions, for example, between N or C terminus and the catalytic DH domain. Thus the removal of either N-terminal sequences from Dbl, Vav, and Tiam 1 (36 -38) or C-terminal sequences from p115RhoGEF and Lbc (39,40) leads to constitutive activation of these GEF proteins. Similarly, ligand-induced phosphorylation may relieve the structural constraints of the DH domain imposed by these inhibitory domains, resulting in stimulation of catalytic activity. Recent NMR spectroscopy data on Vav clearly show that this is the case (41). Tyr-174 located in an inhibitory helix of Vav plays a key role in blocking access of G protein substrates by binding to the DH domain. Among a number of conserved residues (CR1-3), CR1 (E206) is critical for this interaction. Phosphorylation of Tyr-174 in Vav hinders autoinhibitory binding to CR1, providing new interactions between the DH domain and G proteins for GDP/GTP exchange. Conversely, in vitro mutagenesis of this tyrosine to phenyalanine results in impairment of GEF activity (42,43). Phosphorylation of Ser-525 and Thr-526 in p85 ␤PIX may lead to similar results. Interestingly, p85 ␤PIX contains an Asn at the position corresponding to CR1 in other GEF proteins, suggesting an activation mechanism distinct from that for Vav. An alternate possibility is that interactions between Ser-525/Thr-526 and CR3 (Lys or Arg), which is conserved in p85 ␤PIX, are functionally more important. Adding to the complexity, Ser-525/Thr-526 are located in the GB domain, which is the binding site for p95 proteins such as p95PKL in PC12 cells (data not shown). Therefore, it is possible that due to steric hindrance from p95PKL, phosphorylated Ser-525/Thr-526 act indirectly via inducing conformational changes rather than direct binding to the DH domain. Additionally, the issue of whether the adjacent PH domain that works cooperatively to activate the DH domain subsequent to lipid binding (31) and the SH3 domain that tightly binds PAK2 have any effect on GEF activity following phosphorylation of Ser-525/Thr-526 remains to be resolved.
The Rho family GTPases, including RhoA, Rac1, and Cdc42, function in a variety of cellular processes. Translocation of a complex composed of the G protein, GEF, and an effector protein(s) to appropriate subcellular locations is essential for activity. However, the precise mechanism of translocation of this complex, including the necessity of G protein activity per se remains to be elucidated. A well known role of Rac GTPases is the regulation of neutrophil function in response to various inflammatory mediators (19). Rac is a key component of the NADPH oxidase complex, which generates large quantities of superoxide for killing and digestion of pathogens. An initial event in neutrophil activation is NADPH oxidase assembly, which requires the translocation of cytosolic components, p40 phox , p47 phox , and p67 phox , and combining with two integral components, gp91 phox and p22 phox . Recent evidence indicates that Rac-GTP is necessary for these steps, in particular, inducing translocation of p47 phox (44) whose phosphorylation at multiple serine residues is necessary to uncover the binding sites for p22 phox (45,46) and functioning as a subunit in the assembly of an active NADPH oxidase complex. Interestingly, it is assumed that the relevant target of Rac-GTP for translocation of p47 phox is PAK, which is also a potential kinase for p47 phox (47). Integrin-mediated cell adhesion also requires translocation of active Rac-GTP. The downstream kinase, PAK, is lo-cated at the membrane (48). Rac activity is regulated by CrkII, an adaptor protein solely composed of the SH2 and SH3 domains. Tyrosine phosphorylation of this protein at Y221 is crucial. Thus, CrkII-Y221F is defective in activating Rac signaling, including membrane localization (49). Surprisingly, Rac-GEF activity is mediated through the Dock 180-ELMO complex, which has no obvious catalytic domain analogous to the conventional DH domain (50). Rac1 and Cdc42 promote neurite outgrowth through formation of lamellipodia and filopodia at growth cones of the extending neurite tips. Our study shows that mutant p85 ␤PIX without GEF activity also translocates to the site where wild-type p85 ␤PIX is localized (Fig. 6). During the monitoring period of 48 h, however, the mutanttransfected cells exhibit neurite retraction, while expression of wild-type p85 ␤PIX prevents neurites from retraction (Fig. 7). These results suggest that translocation of the complex containing Rac1 may occur independent of or precede Rac1 activation. In this scenario, we can envisage the opposite roles of wild-type p85 ␤PIX and mutant p85 ␤PIX following translocation, i.e. mutant p85 ␤PIX induces neurite retraction by inhibiting endogenous PIX activity at growth cones. This is consistent with the finding that dissociation of GDI and membrane translocation of Rac-GDP is a prerequisite for the efficient activation of Rac by DH-PH of Tiam1 GEF (51). We thus propose a dual role for p85 ␤PIX phosphorylation in promoting neurite outgrowth. Phosphorylation of p85 ␤PIX induces conformational changes in the p85 ␤PIX complex, which in turn somehow activate the translocation process (11). On the other hand, phosphorylation of p85 ␤PIX mediates Rac1 activation, which results in the regulation of actin cytoskeletal reorganization to form growth cones for neurite extension.
In conclusion, we provide evidence that phosphorylation of p85 ␤PIX functions as an upstream signal of Rac1 activation. However, this activation does not mediate translocation of the PIX complex. Further studies on the events subsequent to p85 ␤PIX phosphorylation should elucidate the molecular mechanisms by which GEF activity of p85 ␤PIX is up-regulated and the PIX complex is targeted to its proper subcellular location in relation to bFGF and NGF-induced neurite outgrowth.