Glycogen Synthase Kinase 3- and Extracellular Signal-Regulated Kinase-Dependent Phosphorylation of Paxillin Regulates Cytoskeletal Rearrangement

Abstract

Paxillin is a 68-kDa focal adhesion-associated protein that plays an important role in controlling cell spreading and migration. Phosphorylation of paxillin regulates its biological activity and thus has warranted investigation. Serine 126 and serine 130 were previously identified as two major extracellular signal-regulated kinase (ERK)-dependent phosphorylation sites in Raf-transformed fibroblasts. Here serine 126 is identified as a phosphorylation site induced by lipopolysaccharide (LPS) stimulation of RAW264.7 cells. A number of other stimuli, including adhesion and colony-stimulating factor, induce serine 126 phosphorylation in RAW264.7 cells, and nerve growth factor (NGF) treatment induces serine 126 phosphorylation in PC12 cells. The kinase responsible for phosphorylation of this site is identified as glycogen synthase kinase 3 (GSK-3). Interestingly, this GSK-3-dependent phosphorylation is regulated via an ERK-dependent priming mechanism, i.e., phosphorylation of serine 130. Phosphorylation of S126/S130 was required to promote spreading in paxillin null cells, and LPS-induced spreading of RAW264.7 cells was inhibited by expression of the paxillin S126A/S130A mutant. Furthermore, this mutant also retarded NGF-induced PC12 cell neurite outgrowth. Hence, phosphorylation of paxillin on serines 126 and 130, which is mediated by an ERK/GSK-3 dual-kinase mechanism, plays an important role in cytoskeletal rearrangement.

Paxillin is a 68-kDa focal adhesion-associated protein that functions as a scaffolding protein assembling signaling molecules into complex downstream of integrins (6, 34). It plays an important role in regulating cell spreading and migration. The paxillin knockout mouse exhibits embryonic lethality, which suggests that paxillin plays an essential role in development (18). Paxillin contains five LD motifs in the N-terminal half of the molecule. These peptide motifs mediate protein-protein interactions and bind a number of proteins, including focal adhesion kinase (FAK) and vinculin (6, 40). Four LIM domains are found in the C-terminal half of paxillin, two of which are required for the discrete localization of paxillin to focal adhesions (3).

Multiple stimuli induce phosphorylation of paxillin, including growth factors, integrin-dependent cell adhesion to extracellular matrix, and other ligands (6, 34). Two major tyrosine phosphorylation sites, Y31 and Y118, have been identified in the N-terminal half of paxillin (35). Phosphorylation of these sites modulates docking of SH2 domain-containing proteins, such as CRK, and is important for regulation of cell motility (32, 43). In addition to tyrosine phosphorylation sites, serine and threonine phosphorylation sites have been identified in paxillin. Serine residues 188 and 190 are phosphorylated following integrin ligation (1). Threonines 398 and 403 in LIM2 and serines 457 and 481 in LIM3 are phosphorylated following cell adhesion and stimulation with angiotensin II (4, 5). Phosphorylation of these LIM domain residues regulates focal adhesion localization of paxillin and/or cell adhesion to fibronectin. Though the upstream kinases responsible for phosphorylation of many of these sites remain unidentified, several kinases have been shown to directly phosphorylate paxillin. Jun N-terminal protein kinase phosphorylates threonine 178, and phosphorylation of this site functions in the regulation of cell migration (22). Two kinases, p38 mitogen-activated protein kinase and extracellular signal-regulated kinase (ERK), have been reported to phosphorylate serine 83 in murine/rat paxillin (21, 23). This site is not precisely conserved in human paxillin, but p38 phosphorylates a similar sequence at serine 85 in the human homologue. P38-dependent phosphorylation of this site regulates neurite outgrowth in PC12 cells, and ERK-dependent phosphorylation of the site regulates epithelial morphogenesis. Two additional serine phosphorylation sites in the N-terminal domain of paxillin, serines 126 and 130, were identified in Raf-transformed cells, and phosphorylation is apparently mediated by the Raf-mitogen-activated protein kinase/ERK kinase (MEK)-ERK pathway (42). However, it is unclear whether ERK directly phosphorylates these two sites, and the function of phosphorylation of these sites has not been determined.

Glycogen synthase kinase 3 (GSK-3) was first identified as the enzyme that phosphorylates and regulates glycogen synthase (14). The two isoforms, GSK-3α and GSK-3β, share high similarity in structure but are not redundant in function (13). GSK-3 is now known to phosphorylate a broad range of substrates and control many processes in addition to glycogen metabolism. GSK-3 plays a key role in regulating the Wnt signaling pathway and the control of cell proliferation (31). GSK-3 has also been suggested to regulate microtubule stability through phosphorylation of three microtubule/tubulin-associated proteins, Tau, microtubule-associated protein 1B, and collapsin response mediator protein 2 (17, 19, 44). Regulation of GSK3 activity via regulation of microtubule dynamics is believed to play an important role in the regulation of neuronal cell axon polarity (45). GSK-3 has also been suggested to control actin cytoskeleton rearrangement, since it can regulate formation of long lamellipodia in human keratinocytes (27).

Unlike many other kinases, GSK-3 is constitutively active in cells, and initiation of downstream signaling is not modulated by activation of the kinase but by modification of the substrate, resulting in its interaction with GSK-3 (2). GSK-3 prefers a primed substrate, which has been previously phosphorylated by a priming kinase, and the priming phosphorylation increases the efficiency of substrate phosphorylation of most GSK-3 substrates by 100- to 1,000-fold (15). In several instances the detailed mechanism of GSK-3 substrate phosphorylation has been elucidated. For example, casein kinase 2 is required to prime glycogen synthase to promote the sequential multisite phosphorylation by GSK-3 (15), and casein kinase 1 was identified as a priming kinase promoting GSK-3-mediated β-catenin phosphorylation (29).

Here we identify paxillin, a focal adhesion-associated protein, as a GSK-3 substrate. Serine 126 is identified as a phosphorylation site induced by lipopolysaccharide (LPS) stimulation of RAW264.7 cells. Phosphorylation of this site is regulated by ERK but is directly mediated by GSK-3. LPS-induced cell spreading was partially inhibited in cells expressing the paxillin S126A/S130A mutant, and this mutant was defective for promoting fibroblast spreading on fibronectin. Furthermore, we found ERK/GSK-3-mediated phosphorylation of paxillin is also involved in nerve growth factor (NGF)-induced PC12 cell neurite outgrowth. These data suggest that phosphorylation of paxillin at serine residues 126 and 130 plays an important role in the control of cytoskeleton rearrangements and provides insight into the molecular mechanism via which GSK-3 controls remodeling of the actin cytoskeleton.

MATERIALS AND METHODS

Cells.

RAW264.7 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS). HEK293T cells were maintained in DMEM-F12 containing 10% FBS. PC12 cells were maintained as described previously (21). Primary murine peritoneal macrophages, isolated following thioglycolate injection (a kind gift from Glenn Matsushima), were maintained in RPMI 1640 containing 5% FBS for 1 week prior to LPS stimulation. Paxillin null cells were a generous gift from Sheila Thomas and were maintained in DMEM containing 15% FBS. In some experiments, RAW264.7 cells were starved in DMEM without serum overnight and then stimulated with LPS (1 μg/ml) (Sigma, St. Louis, MO), phorbol myristate acetate (PMA) (100 ng/ml) (Calbiochem, San Diego, CA), or macrophage colony-stimulating factor (macrophage CSF) (1.32 nM) (PeproTech, Rocky Hill, New Jersey). PC12 cells were starved in DMEM containing 1% FBS for 6 h. The cells were then stimulated with 100 ng/ml NGF (Calbiochem, San Diego, CA). In some experiments cells were pretreated with the indicated doses of LiCl, U0126 (Calbiochem, San Diego, CA), GSK-3 inhibitor IX (Calbiochem, San Diego, CA), or GSK-3 inhibitor I (Biosource, Camarillo, CA) for 1 h before stimulation. In some experiments PP2 (Calbiochem, San Diego, CA) was added 20 min before stimulation.

Molecular biology.

The MEK constructs were gifts from Channing Der. The enhanced green fluorescent protein (EGFP)-paxillin β plasmid was a gift from Ken Jacobson (22). EGFP-paxillin-derived point mutations were created by PCR using the Quik-Change mutation kit (Stratagene, La Jolla, CA). Sequence analysis was performed on each mutant to verify the intended point mutations and that no unintended mutations were present. These analyses were performed in the University of North Carolina at Chapel Hill Genome Analysis Facility on a model 3730 DNA analyzer (Perkin-Elmer, Applied Biosystems Division) using the ABI PRISM dye terminator cycle sequencing ready reaction kit with AmpliTaq DNA polymerase, FS (Perkin-Elmer, Applied Biosystems Division). The control small interfering RNA (siRNA) and siRNA targeting GSK-3β (SMARTpool) were from Dharmacon (Lafayette, CO).

Stable and transient transfection.

RAW264.7, PC 12, HEK293T, and paxillin null cells were transfected using Lipofectamine Plus according to the manufacturer's instructions (Invitrogen, Carlsbad, CA). To establish stable transfectants, RAW264.7 cells were incubated in fresh medium containing 10% FBS for 48 h and then selected with 400 μg/ml G418. After 10 days, surviving cells were cultured in medium containing 10% FBS with 200 μg/ml G418. GFP-expressing cells were further enriched by fluorescence-activated cell sorting. Despite this double-selection procedure, the resulting population was heterogeneous, containing both GFP-positive and -negative cells. Further, EGFP-paxillin expression was lost upon passaging. For this reason, studies with these cells were restricted to biochemical studies examining regulation of phosphorylation of the exogenous EGFP-paxillin constructs and single-cell biological assays where expression of the EGFP-paxillin constructs could be validated by fluorescence microscopy. Stable populations of paxillin null cells reexpressing wild-type or S126A/S130A paxillin were established by infection with pBABE retroviral vectors encoding these constructs, followed by selection with puromycin. RAW264.7 cells were transfected with 75 nM siRNAs using TransitTKO according to the manufacturer's instructions (Mirus, Madison, WI). After 24 h, the cells were starved in serum-free DMEM overnight and then stimulated with LPS.

Cell spreading assay.

Paxillin null cells and variants reexpressing wild-type and mutant paxillin were serum starved overnight, trypsinized, and plated onto fibronectin-coated dishes in serum-free medium. After various times, the cells photographed and the relative area of individual cells (>50 per experiment) was determined using Image J software. RAW264.7 cells expressing wild-type or mutant EGFP paxillin fusion proteins were plated on 35-mm petri dishes, cultured overnight, and then starved in serum-free medium for 12 h. The cells were then stimulated with 1 μg/ml LPS and cultured for 3 h. Cells were examined using a Zeiss Axiovert 200 microscope. Transfected cells were identified as green cells by fluorescence microscopy, and the morphology of the transfected cells was scored by phase-contrast microscopy. Round phase bright, refractile cells were scored as unspread. Phase dark and nonrefractile cells and cells that had obviously become elongated (exhibiting a ratio of length to width of greater than 2:1) were defined as spread. At least 100 green cells were scored in each experiment.

Neurite outgrowth assay.

The PC12 cell neurite outgrowth assay was performed as described previously (21). In brief, PC12 cells transiently transfected with constructs expressing wild-type or mutant EGFP paxillin fusion proteins were plated on 35-mm petri dishes precoated with 10 μg/ml collagen I (BD Biosciences, San Jose, CA) and cultured overnight. The cells were starved in DMEM medium containing 1% FBS for 6 h. The cells were then stimulated with 100 ng/ml NGF (Calbiochem, San Diego, CA) and cultured for 36 h at 37°C. Cells were examined using a Zeiss axiovert 200 microscope. Transfected cells were identified as green cells by fluorescence microscopy, and the morphology of the transfected cells was scored by phase-contrast microscopy. More than 100 green cells were examined in each experiment, and the length of neurites was scored as described previously (12).

Protein purification and in vitro phosphorylation assay.

The expression and purification of the N-terminal glutathione S-transferase (GST) fusion proteins, GST-N-C3 and GST-N1-C1A, were performed as described previously (39). For in vitro kinase assays, GST fusion proteins were washed twice with reaction buffer (25 mM HEPES, pH 7.5, 10 mM MgCl₂, 1 mM EDTA, 0.1 mM dithiothreitol) and divided into aliquots. Two micrograms of substrate was incubated in reaction buffer containing 40 μM ATP with ERK1 (UBI, Lake Placid, NY) and/or active GSK-3 (Biosource, Camarillo, CA) at 30°C for 1 h. The reactions were stopped by the addition of sample buffer. The samples were boiled and analyzed by Western blotting. To further examine phosphorylation by ERK, 2 μg of GST-N1-C1A was incubated in reaction buffer containing 40 μM ATP (2 μCi [γ-³²P]ATP) with ERK1 at 30°C. The reactions were stopped by the addition of sample buffer. The samples were boiled and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The gel was stained with Coomassie blue before autoradiography.

Cell lysis, protein analysis, and immunoprecipitation.

Cells were lysed in ice-cold modified radioimmunoprecipitation assay buffer (37). Lysates were clarified, and protein concentrations were determined using the bicinchoninic acid assay (Pierce, Rockford, IL). For immunoprecipitations, the paxillin antibody (BD Biosciences, San Jose, CA) or the PS126 antibody was incubated with 500 μg of cell lysate at 4°C for 1 h or 16 h. Immune complexes were precipitated at 4°C for 1 h with protein A-Sepharose beads (Sigma, St. Louis, MO). For immunoprecipitations using the paxillin monoclonal antibody, the beads were coated with AffiniPure rabbit antimouse immunoglobulin G (Jackson ImmunoResearch Labs, West Grove, PA). Immune complexes were washed twice with ice-cold lysis buffer and once with ice-cold PBS. Beads were resuspended in sample buffer and boiled to elute the proteins, and the samples were analyzed by Western blotting. The paxillin phospho-specific PS126, PY31, PY118, and pERK antibodies were from Biosource (Camarillo, CA). The ERK and ERK2 antibodies were from Santa Cruz Biotech (Santa Cruz, CA). The PS83 antibody was a generous gift from Shuta Ishibe and Lloyd Cantley (Yale University).

Immunofluorescence.

Paxillin null fibroblasts were transiently transfected with plasmids encoding GFP-paxillin fusion proteins and cultured overnight. The cells were then trypsinized, held in suspension for 45 min, and plated onto fibronectin-coated coverslips for 60 min prior to fixation. PC12 cells were plated on collagen-coated coverslips and stimulated with NGF for various times prior to fixation. Cells were fixed in 3.7% formaldehyde and permeabilized with 0.5% Triton X-100. Serine 126 phosphorylation was detected using PS126 and a rhodamine- or fluorescein-conjugated antirabbit antibody (Molecular Probes, Eugene, OR) as described previously (9). The cells were visualized using a Nipkow-type spinning disk confocal scan head attached to an IX81 inverted microscope (Olympus Dulles, VA) equipped with a ×60 1.45 numerical-aperture objective and a charge-coupled-device camera, and controlled by AQM Advance 6 software.

RESULTS

LPS induces paxillin phosphorylation in macrophages.

Paxillin, a major focal adhesion-associated protein, becomes tyrosine phosphorylated upon LPS stimulation of human monocytes and other macrophage cell lines (41). To examine phosphorylation of paxillin in RAW264.7 cells, the cells were serum starved overnight and then stimulated with 1 μg/ml LPS for 60 min prior to lysis. Paxillin was immunoprecipitated from cell lysates and the immune complexes analyzed by Western blotting. The most striking observation from this experiment was the evident retardation of the relative mobility of paxillin following LPS stimulation (Fig. 1A), which suggests that paxillin may be phosphorylated following LPS stimulation. To determine if the LPS-induced shift in gel mobility of paxillin was in fact due to phosphorylation, paxillin immune complexes from LPS-stimulated RAW264.7 cells were incubated with calf intestine alkaline phosphatase for half an hour. Phosphatase treatment decreased the apparent molecular weight of paxillin to the original level and lower, suggesting that the LPS-induced molecular weight shift was due to phosphorylation and that basal levels of phosphorylation of paxillin also produced a small shift in the apparent molecular weight of the protein. To directly examine tyrosine phosphorylation of paxillin, site-specific antibodies were used to probe Western blots of cell lysates from LPS-stimulated and unstimulated RAW264.7 cells. The two major tyrosine phosphorylation sites of paxillin, tyrosine 31 and tyrosine 118, were phosphorylated in response to LPS stimulation (Fig. 1B). In addition to tyrosine phosphorylation, it was also likely that LPS induced serine/threonine phosphorylation of paxillin. In particular, serine 126 was an excellent candidate as an LPS-induced site of phosphorylation, since phosphorylation of this site has been correlated with a shift in the electrophoretic mobility of paxillin (42). A phosphorylation-site-specific antibody recognizing PS126 was used to probe lysates of stimulated and unstimulated cells. Whereas paxillin from unstimulated cells was weakly recognized by the PS126 antibody, LPS stimulation induced a dramatic increase in signal (Fig. 1B). Phosphorylation of serine 83 was also induced in response to LPS. In order to confirm the specificity of the PS126 antibody, its ability to recognize wild-type paxillin and an S126A mutant of paxillin was examined. RAW264.7 cell line derivatives stably expressing EGFP-paxillin or EGFP-paxillin S126A were established by drug selection using G418. GFP-expressing cells were further enriched by fluorescence-activated cell sorting. These two cell populations were challenged with LPS, and cell lysates were blotted with the PS126 antibody (Fig. 1C). As observed with endogenous paxillin, the wild-type EGFP-paxillin was recognized weakly by this antibody in unstimulated cells and the signal was profoundly increased upon LPS stimulation. In contrast, EGFP-paxillin S126A was not recognized by the PS126 antibody under either set of conditions. These results demonstrate that this antibody specifically recognizes paxillin that is phosphorylated on serine 126. To establish the generality of the observation that serine 126 phosphorylation was induced by LPS, primary cultures of peritoneal macrophages and two additional macrophage cell lines, J774 and Bac1, were also analyzed. After LPS stimulation, serine 126 was phosphorylated in each of these cell types (Fig. 1D), which suggests that this phosphorylation event is a general response to LPS signaling in macrophages. To estimate the stoichiometry of phosphorylation at serine 126, the PS126 antibody was used to immunoprecipitate paxillin following LPS stimulation of RAW264.7 cells. The amount of paxillin recovered in the immune complex and the amount remaining in the supernatant were examined by Western blotting. From these studies it is estimated that 30 to 40% of paxillin is phosphorylated on serine 126 following LPS stimulation.

FIG. 1.

LPS induces paxillin tyrosine and serine phosphorylation. (A) RAW264.7 cells were serum starved overnight (lane 1) and then stimulated with 1 μg/ml LPS for 1 h (lanes 2, 3) prior to lysis. Paxillin was immunoprecipitated from 500 μg of cell lysate. Half of the immune complex from the LPS-stimulated sample was incubated with calf intestine alkaline phosphatase for half an hour (lane 3). The immune complexes were then analyzed by Western blotting. (B) Cell lysates from unstimulated (lane 1) and LPS-stimulated (lane 2) RAW264.7 cells were Western blotted using phospho-specific (top four panels) or paxillin (bottom panel) antibody. (C) RAW264.7 cell line derivatives stably expressing EGFP-paxillin (lanes 1, 2) or EGFP-paxillin S126A (lanes 3, 4) were serum starved (lanes 1, 3) or starved and then stimulated with LPS (lanes 2, 4). Cell lysates were blotted with the PS126 (top) or paxillin (bottom) antibody. The positions of the exogenous EGFP-paxillin and endogenous paxillin are indicated. (D) Primary peritoneal macrophages (lanes 1 to 6) were stimulated with LPS (1 μg/ml) for the indicated times prior to lysis. J774 cells (lanes 7, 8) and Bac1 cells (lanes 9, 10) were starved (lanes 7, 9) or stimulated with LPS (1 μg/ml) (lanes 8, 10) for 1 h prior to lysis. Lysates were blotted with PS126 or paxillin antibody. (E) Lysates from RAW264.7 cells stimulated with LPS were analyzed. Fifty micrograms of lysate was directly analyzed (lane 1) or 500 μg of lysate was immunoprecipitated using PS126. Ten percent of the immune complex (IP) (lane 2) or supernatant (lane 3) was analyzed by Western blotting for paxillin.

Serine 126 phosphorylation is ERK dependent.

Serine 126 and serine 130 were first identified as paxillin phosphorylation sites in Raf-transformed NIH 3T3 cells (42). In this scenario, additional evidence demonstrated that this phosphorylation event occurred downstream of the Raf-MEK-ERK pathway. It is well established that ERK is activated in macrophages after LPS stimulation (38). To begin to address the role of ERK in LPS-induced paxillin phosphorylation, serine 126 phosphorylation was temporally compared with ERK activation following LPS stimulation of RAW264.7 cells. ERK was strongly activated at 40 and 60 min following LPS stimulation, as assessed by blotting with a phosphorylation-specific ERK antibody (Fig. 2A). Phosphorylation of serine 126 was increased after 40 min and more dramatically increased following 60 min of stimulation. Thus, the kinetics of ERK activation is consistent with a role in controlling phosphorylation of paxillin at serine 126. Furthermore, treatment of cells with two different MEK inhibitors, PD98095 and U0126, efficiently blocked the LPS-induced phosphorylation of paxillin on serine 126 (Fig. 2B and C). These findings demonstrate that LPS-induced activation of ERK is required for the induction of paxillin phosphorylation on serine 126. To determine if activation of the ERK signaling pathway is sufficient to induce serine 126 phosphorylation, wild-type MEK1 or constitutively active MEK1 were transiently transfected into 293T cells. The latter induced activation of ERK and phosphorylation of paxillin at serine 126 (Fig. 2D). These results demonstrate that serine 126 phosphorylation is regulated by ERK signaling.

FIG. 2.

LPS-induced paxillin serine 126 phosphorylation is ERK dependent. (A) RAW264.7 cells were serum starved overnight (lane 1) and then stimulated with 1 μg/ml LPS for the indicated time (lanes 2-4). Cell lysates were blotted with the PS126, paxillin, pERK, and ERK2 antibodies. (B) RAW264.7 cells were serum starved overnight (lane 1) and then pretreated with the MEK inhibitor PD98095 at the dose shown for 1 h. Cells were stimulated with LPS (1 μg/ml) for 1 h. Cell lysates were blotted with PS126, paxillin, pERK, or ERK antibody. (C) RAW264.7 cells were treated as in panel B, except that MEK was inhibited with 20 μM U0126. (D) Wild-type or constitutively activated MEK was transiently expressed in 293 cells, and lysates were blotted with the PS126, paxillin, or pERK antibody.

Serine and tyrosine phosphorylation of paxillin are independent.

Recently serine 83 of paxillin was identified as a direct ERK phosphorylation site (23). An intriguing mechanism regulating ERK-dependent phosphorylation of serine 83 has been proposed. Phosphorylation of paxillin at tyrosine 118 creates a docking site for ERK, which recruits ERK into complex, facilitating serine 83 phosphorylation (24). In order to investigate whether a similar mechanism is used in the regulation of serine 126 phosphorylation in response to LPS in macrophages, a paxillin mutant was analyzed. RAW264.7 cells were transfected with EGFP-paxillin or EGFP-paxillin Y31F/Y118F, a mutant in which the two major tyrosine phosphorylation sites of paxillin have been substituted with phenylalanines. The cells were challenged with LPS, and then lysates were blotted with the PS126 antibody. LPS induced phosphorylation of the mutant on serine 126 (Fig. 3A). While there was an apparent increase in serine 126 phosphorylation of the mutant relative to the wild type, this was likely due to the increased expression level of the mutant. Notably, the EGFP-paxillin Y31F/Y118F mutant was not defective for serine 126 phosphorylation. Furthermore, treatment with PP2, a Src kinase inhibitor, potently blocked the LPS-induced phosphorylation of tyrosine 118 but had little effect on serine 126 phosphorylation (Fig. 3B). These findings demonstrate that serine 126 phosphorylation occurs independently of phosphorylation at tyrosine 118. Given the proximity of serine 126 to tyrosine 118, it was also of interest to determine if serine 126 phosphorylation could affect phosphorylation at tyrosine 118. This was tested using a pharmacological approach. U0126 blocked serine 126 phosphorylation without affecting phosphorylation at tyrosine 118 (Fig. 3C). These data suggest that LPS-induced paxillin phosphorylation at serine 126 and tyrosine 118 are independent and are controlled by separate mechanisms. It further suggests that the mechanisms regulating ERK-mediated serine 83 and serine 126 phosphorylation are distinct.

FIG. 3.

Paxillin serine and tyrosine phosphorylation are regulated by separate pathways. (A) RAW264.7 cell line derivatives transiently expressing EGFP-paxillin (lanes 1, 2) or EGFP-paxillin Y31F/Y118F (lanes 3, 4) were stimulated with LPS (lanes 2, 4) and cell lysates blotted with the paxillin or PS126 antibody. The positions of the exogenous EGFP-paxillin and endogenous paxillin are indicated. (B) RAW264.7 cells were serum starved overnight (lane 1) and then pretreated with the Src inhibitor PP2 at the doses shown for 20 min. Cells were stimulated with LPS (1 μg/ml) for 1 h. Cell lysates were blotted with PS126, PY118, paxillin, pERK, or ERK antibody. (C) RAW264.7 cells were serum starved overnight (lane 1) and then were pretreated with U0126 at the dose shown for 1 h. Cells were stimulated with LPS (1 μg/ml) for 1 h. Cell lysates were blotted with the paxillin or PY118 antibody.

Paxillin serine 126 phosphorylation can be induced by different stimuli.

Since paxillin phosphorylation at serine 126 is mediated by the ERK pathway, which plays a key role in many signaling pathways, it was of interest to determine whether ERK-mediated paxillin phosphorylation at serine 126 was specific for LPS signaling or occurred in response to other stimuli that activate ERK. ERK is activated upon integrin-dependent cell adhesion to extracellular matrix proteins (36). Further, paxillin was shown to become serine phosphorylated in response to adhesion to extracellular matrix in macrophages (11). To examine cell adhesion-dependent phosphorylation, RAW264.7 cells were trypsinized, incubated in suspension at 37°C for half an hour, and then replated on fibronectin for half an hour at 37°C prior to lysis. Phosphorylation at serine 126 decreased when cells were held in suspension. Upon cell adhesion, phosphorylation of paxillin at serine 126 was induced (Fig. 4A). Serine 126 was also phosphorylated in RAW264.7 cells following protein kinase C activation by stimulation with PMA (Fig. 4B). Treatment of RAW264.7 cells with CSF, an important cytokine controlling macrophage function, also modestly induced phosphorylation of serine 126 (Fig. 4C). Phosphorylation of serine 126 is also likely induced by growth factors that signal via receptor tyrosine kinases (42). Interestingly, the phosphorylation level of S126 was elevated following NGF stimulation in PC12 cells (Fig. 4D). Each of these stimuli activates ERK, and in each case inhibition of this signaling pathway with MEK inhibitors blocks phosphorylation of serine 126 in response to the stimulus. These findings suggest that serine 126 phosphorylation is a general response to activation of the ERK pathway.

FIG. 4.

Paxillin serine 126 phosphorylation is induced by multiple stimuli. (A) RAW264.7 cells were trypsinized, incubated in suspension at 37°C for half an hour (lane 2), and then replated on fibronectin-coated plates (50 μg/ml) for half an hour (lane 3) at 37°C prior to lysis. Cell lysates were blotted with paxillin or PS126 antibody. (B and C) RAW264.7 cells were serum starved overnight (lane 1) and then were stimulated with PMA (100 ng/ml) (lane 2 in panel B), LPS (1 μg/ml) (lane 2 in panel C), or CSF (1.32 nM) (lane 3 in panel C) for 1 h. Cell lysates were blotted with paxillin, PS126, or pERK antibody. (D) PC12 cells were starved in DMEM containing 1% FBS for 6 h, treated with 20 μM U0126 for 1 h (lane 3), and stimulated with 100 ng/ml NGF for 2 h (lanes 2, 3). Cell lysates were blotted with PS126, paxillin, or pERK antibody.

Phosphorylation of serine 126 is abolished by GSK-3 inhibitors.

Though the data demonstrate that residue 126 is a major serine phosphorylation site in paxillin that is dependent upon ERK activation, there is no evidence that this site is directly phosphorylated by ERK. Further, the sequence flanking serine 126 does not conform to a consensus ERK phosphorylation site. Note, however, that serine 130 has also been identified as a paxillin phosphorylation site in Raf-transformed NIH 3T3 cells (42). Upon phosphorylation of serine 130, serine 126 resembles a consensus GSK-3 phosphorylation site, SXXX(P)S. It therefore appeared likely that paxillin was phosphorylated by GSK-3 at serine 126. To examine whether serine 126 phosphorylation was dependent upon GSK-3, RAW264.7 cells were incubated with LiCl, which is an inhibitor of GSK-3, for 1 h prior to simulation with LPS. Phosphorylation at serine 126 was blocked by LiCl in a dose-dependent manner (Fig. 5A). In addition to LiCl, two other pharmacological inhibitors of GSK-3, GSK-3 inhibitor IX and GSK-3 inhibitor I, were tested. Both efficiently inhibited paxillin phosphorylation at serine 126 in RAW264.7 cells in response to LPS (Fig. 5B and C). Further, LiCl also blocked serine 126 phosphorylation in NGF-stimulated PC12 cells (Fig. 5D). It is interesting that ERK activation is unattenuated in all samples treated with GSK-3 inhibitors. To further test the role of GSK-3β in regulating serine 126 phosphorylation, an siRNA approach was used. RAW264.7 cells were transfected with a control siRNA or a pool of siRNAs targeting GSK-3β. Western blotting demonstrated knockdown of GSK-3β but not GSK-3α expression (Fig. 5E). Transfected cells were stimulated with LPS, and inhibition of GSK-3β reduced the induction of serine 126 phosphorylation on paxillin but had no effect upon ERK activation (Fig. 5E). The GSK-3β-independent phosphorylation of serine 126 is likely mediated by GSK-3α. Although serine 126 phosphorylation downstream of multiple stimuli is ERK dependent, it is also GSK-3 dependent, and GSK-3 appears to operate downstream of ERK. These findings are consistent with the hypothesis that ERK phosphorylates paxillin at serine 130, priming paxillin for subsequent phosphorylation by GSK-3 at serine 126.

FIG. 5.

Phosphorylation of serine 126 is abolished by GSK-3 inhibitors. (A, B, and C) RAW264.7 cells were serum starved overnight (lane 1) and then pretreated with LiCl, GSK-3 inhibitor IX, or GSK-3 inhibitor I at the dose indicated for 1 h. Cells were stimulated with LPS (1 μg/ml) for 1 h. Cell lysates were blotted with the indicated antibodies. (D) PC12 cells were starved in DMEM containing 1% FBS for 6 h, treated with the indicated drug for 1 h, and stimulated with 100 ng/ml NGF for 2 h. Cell lysates were blotted with the indicated antibodies. (E) RAW264.7 cells were transfected with the control or GSK-3β siRNAs, starved, and stimulated with LPS. Lysates were blotted for PS126, paxillin, pERK, or GSK-3α/β.

Phosphorylation of paxillin is mediated by an ERK/GSK-3 dual-kinase mechanism.

In order to examine whether paxillin can be directly phosphorylated by GSK-3 in vitro, the N-terminal half of paxillin was expressed in Escherichia coli as a GST fusion protein, called GST-N-C3. Recombinant GSK-3β was incubated with GST-N-C3 in kinase reaction buffer containing ATP, and the phosphorylation of serine 126 was monitored by Western blotting with a phospho-specific antibody. There was little evidence of serine 126 phosphorylation on GST-N-C3 following incubation with GSK-3β (Fig. 6A). This again is consistent with the hypothesis that this GSK-3 site in paxillin is primed by phosphorylation by another kinase. ERK was a candidate for this kinase, since phosphorylation of 126 in vivo was ERK dependent and the sequence around serine 130 resembles an ERK phosphorylation site. ERK failed to directly phosphorylate serine 126. To test whether ERK functions as a priming kinase for GSK-3-dependent phosphorylation of paxillin, GST-N-C3 was preincubated with ERK in vitro, prior to incubation with GSK-3β in the in vitro kinase reaction, and phosphorylation was measured by blotting with PS126. Under these conditions, serine 126 phosphorylation was dramatically increased (Fig. 6A). To confirm that GSK-3-dependent phosphorylation in vitro was dependent upon serine 130 phosphorylation, a GST-N-C3 variant with a substitution of alanine for serine 130 was also analyzed. While GST-N-C3 was phosphorylated at serine 126 when incubated in vitro with ERK and GSK-3, GST-N-C3/S130A was not (Fig. 6B), demonstrating the requirement of the priming site for GSK-3-dependent phosphorylation in vitro. To further determine if ERK could phosphorylate serine 130 in vitro, recombinant ERK was incubated with GST-N1-C1A in kinase reaction buffer containing [γ-³²P]ATP. Wild-type GST-N1-C1A was efficiently phosphorylated by ERK in vitro, whereas the S130A variant of GST-N1-C1A was marked defective for phosphorylation, exhibiting approximately 50% of the level of phosphorylation of the wild-type protein (Fig. 6C). Further, the paxillin S130A mutant was expressed transiently as an EGFP fusion protein in PC12 cells, and phosphorylation of serine 126 in response to NGF stimulation was measured by Western blotting. While wild-type EGFP-paxillin was phosphorylated on serine 126 upon NGF stimulation, the S130A mutant was not phosphorylated in response to NGF (Fig. 6D). These data suggest that paxillin is phosphorylated by ERK at serine 130, both in vitro and in vivo, and that this phosphorylation event primes paxillin for phosphorylation by GSK-3 on serine 126 (Fig. 6E).

FIG. 6.

Phosphorylation of paxillin is mediated by ERK/GSK-3 dual-kinase mechanism. (A) Recombinant GST-N-C3 was incubated with active GSK-3 and ERK1, alone or in combination, in 50 μl of kinase reaction buffer at 30°C for 60 min. Samples were resolved by SDS-PAGE and immunoblotted using the PS126 or paxillin antibody. (B) GST-N1-C1A was incubated with ERK in kinase reaction buffer containing [γ-³²P]ATP and the samples resolved by SDS-PAGE. The autoradiograph and Coomassie-stained gel are shown. (C) The wild-type GST-N-C3 or GST-N-C3 variant containing an alanine substitution for serine 130 was treated as for panel A. (D) PC12 cells transiently expressing EGFP-paxillin (lanes 1, 2) or EGFP-paxillin S130A (lanes 3, 4) were challenged with NGF (lanes 2, 4), and cell lysates were blotted with the paxillin or PS126 antibody. The positions of the exogenous EGFP-paxillin and endogenous paxillin are shown. (E) A model for phosphorylation of paxillin by the ERK/GSK-3 dual-kinase mechanism is shown.

Localization of paxillin phosphorylated on serine 126.

Paxillin null fibroblasts were used for the initial studies. The cells were transiently transfected with EGFP-paxillin or EGFP-paxillin S126A and then fixed and stained with the PS126 antibody (Fig. 7A). Cells expressing the wild-type and mutant EGFP fusion proteins were identified by fluorescence microscopy. The cells expressing wild-type EGFP-paxillin stained positive with the PS126 antibody, exhibiting nuclear, cytoplasmic, and focal adhesion staining. Since the untransfected cells in the population also exhibited nuclear staining, this is apparently nonspecific (data not shown). Besides this background staining, the PS126 staining was similar to the EGFP-paxillin localization. The localization of the mutant protein was similar to that of EGFP-paxillin; however, cells expressing EGFP-paxillin S126A exhibited only the background nuclear staining with the PS126 antibody. These data suggest that PS126 exhibits sufficient specificity in immunofluorescence experiments that it can be used to determine the localization of paxillin that is phosphorylated on serine 126.

FIG. 7.

Subcellular localization of the serine 126-phosphorylated form of paxillin in fibroblasts. (A) Paxillin null fibroblasts were transiently transfected with EGFP-paxillin (top panels) or the EGFP-paxillin S126A mutant (bottom panels). Twenty-four hours later, the cells were trypsinized, held in suspension for 45 min and then plated onto fibronectin-coated coverslips for 60 min in serum-free medium. Localization of the protein was visualized by fluorescence microscopy (left panels). Localization of paxillin that was phosphorylated on serine 126 was determined by immunofluorescence using the PS126 antibody (right panels). (B) PC12 cells were serum starved (left panel) and stimulated with NGF (100 ng/ml) for 3 h (middle panel) or 30 h (right panel). Cells were fixed and immunostained using the PS126 antibody.

NGF-stimulated PC12 cells were also fixed and stained with the PS126 antibody (Fig. 7B). PS126 staining was dramatically increased after NGF stimulation and was located at the tips of developing neurites several hours after NGF treatment. After 30 h of NGF stimulation, long neurites had grown and PS126 staining was enriched near the tips of these neurites. These studies demonstrate the localization of paxillin that is phosphorylated on serine 126 at dynamic sites of cytoskeleton remodeling and suggest the potential role of serine 126 and serine 130 phosphorylation of paxillin in controlling NGF-induced neurite outgrowth.

ERK/GSK-3-mediated phosphorylation of paxillin is involved in cell spreading.

To address the function of phosphorylation of paxillin at residues 126 and 130, paxillin null fibroblasts stably expressing wild-type paxillin or the S126A/S130A mutant were established. The pBABE retroviral vector, engineered to encode the paxillin constructs, was used to infect the paxillin null cells, and the infected cells were selected using puromycin. Western blotting demonstrated expression of the wild-type and mutant proteins (Fig. 8A). Further, the wild-type protein was phosphorylated on serine 126 in cells growing in culture, whereas the mutant was not (Fig. 8A). Paxillin null cells exhibit a defect in cell spreading on fibronectin that can be rescued by reexpression of wild-type paxillin (18). To quantify spreading, cells were serum starved overnight and then plated on fibronectin and photographed at various times. The area of spread cells was determined using Image J software. The null cells and wild-type paxillin reexpressers increased in area over time, but at each time point the average area of paxillin-expressing cells was approximately 25% greater than the average area of the null cells (Fig. 8B). In contrast, cells expressing the S126A/S130A mutant spread slower than wild-type-expressing cells, exceeding the average area of the null cells only after 60 min on fibronectin. Thus, phosphorylation of these sites is required for the paxillin-dependent spreading of fibroblasts on fibronectin.

FIG. 8.

ERK/GSK-3-mediated phosphorylation of paxillin is involved in cell spreading. (A) Lysates from paxillin null cells or cells reexpressing wild-type or S126A/S130A paxillin were analyzed by Western blotting for paxillin (top panel) or PS126 (bottom panel). (B) Cells were serum starved overnight and then plated on fibronectin. At various times, the area of spreading cells was determined using Image J software (>50 cells per experiment) and the average from three experiments plotted. The data at each time point were analyzed using one-way analysis of variance (P < 0.0001 at each time) and the Dunnett post test (for wild-type-expressing cells, P < 0.01 at each time; for S126A/S130A cells, P < 0.05 only at 60 min). (C) RAW264.7 cells or derivatives stably expressing EGFP-paxillin or EGFP-paxillin S126A/S130A were challenged with 1 μg/ml LPS as described. Cell spreading was recorded after 3 h of LPS stimulation. Phase-contrast images (left panels) and fluorescent images (right panels) of cells transfected with EGFP-paxillin (top panels) or EGFP-paxillin S126A/S130A (bottom panels) are shown. (D) GFP-negative, untransfected cells and GFP-positive cells expressing exogenous paxillin were scored as spread or unspread, and percentages of cells spread were analyzed. Greater than 100 GFP-positive cells were scored in each experiment. The data were plotted as the means ± standard errors from three experiments. The P value is less than 0.05 (unpaired student t test).

LPS induces distinct changes in cell morphology and actin organization in monocytes and macrophages. Serum-starved RAW264.7 cells are round and phase bright when examined by phase-contrast microscopy, whereas many LPS-stimulated cells are phase dark and spread or exhibit an elongated phenotype (data not shown). To begin to address the role of serine 126 phosphorylation in the control of cell spreading, MEK and GSK-3 were inhibited with U0126 and LiCl, respectively. LPS-stimulated RAW264.7 cell spreading was inhibited by both compounds (data not shown). To directly explore the role of paxillin in regulation of this phenotype, RAW264.7 cells were stably transfected with EGFP-paxillin or EGFP-paxillin S126A/S130A, and the cells were stimulated with LPS to induce spreading. Transfected cells were identified as green cells by fluorescence microscopy and the morphologies of both the transfected and untransfected cells scored by phase-contrast microscopy (Fig. 8C). Approximately 25% of the cells expressing EGFP-paxillin exhibited a spread/elongated morphology (Fig. 8D). This frequency of spreading was very similar to that observed for untransfected cells in the same culture. In contrast, cells expressing EGFP-paxillin S126A/S130A exhibited only 10% cell spreading/elongation following LPS stimulation. This mutant is apparently functioning in a dominant-negative fashion to inhibit the action of endogenous wild-type paxillin, suggesting that phosphorylation at serines 126 and 130 is important for optimal spreading. Therefore, ERK/GSK-3-mediated phosphorylation of paxillin is involved in the control of LPS-induced RAW264.7 cell spreading.

ERK/GSK-3-mediated phosphorylation of paxillin is involved in NGF-induced PC12 cell neurite outgrowth.

Both GSK-3 and ERK are required for NGF-induced neuronal cell neurite outgrowth (7, 33). To assess the role of ERK/GSK-3-mediated paxillin phosphorylation in neurite extension, PC12 cells were transiently transfected with EGFP-paxillin or EGFP-paxillin S126A/S130A, and the NGF-induced neurite extension of these cells was examined after 36 h. Transfected cells were identified as green cells by fluorescence microscopy, and neurite extension was examined by phase-contrast microscopy (Fig. 9A). Approximately 36% of the PC12 cells expressing EGFP-paxillin exhibited neurites longer than two cell bodies, and 29% of cells produced neurites longer than three cell bodies (Fig. 9B). Neurite extension in nontransfected cells in the same culture exhibited a similar morphological response to NGF. In contrast, cells expressing EGFP-paxillin S126A/S130A exhibited a retardation in neurite extension. Only 23% of the cells expressing EGFP-paxillin S126A/S130A had neurites longer than two cell bodies, and 16% of the cells had neurites longer than three cell bodies. Cells transiently transfected with EGFP-paxillin S126D/S130D exhibited NGF-induced neurite extension similar to that observed in untransfected and EGFP-paxillin-transfected cells. These findings suggest that phosphorylation of paxillin at serine residues 126 and 130 is involved in the control of NGF-induced neurite extension.

FIG. 9.

ERK/GSK-3-mediated phosphorylation of paxillin is involve in NGF-induced neurite outgrowth. (A) PC12 cells transiently expressing EGFP-paxillin or EGFP-paxillin S126A/S130A were plated on collagen-coated petri dishes and treated with 100 ng/ml NGF. Neurite outgrowth was recorded after 36 h of NGF stimulation. Phase-contrast images (left panels) and fluorescent images (right panels) of cells transfected with EGFP-paxillin (top panels) or EGFP-paxillin S126A/S130A (bottom panels) are shown. (B) Quantitative analysis of neurite outgrowth from three independent experiments is shown. In each experiment, more than 100 GFP-negative, untransfected cells (light-gray bars) and GFP-positive cells from the EGFP-paxillin (white bars), EGFP-paxillin S126A/S130A (black bars), or EGFP-paxillin S126D/S130D (dark gray bars) transfected populations were scored. The data was plotted as the means ± standard errors from three experiments. In each group the wild-type and S126A/S130A data were analyzed using one-way analysis of variance and the Dunnett post test (for each: ANOVA, P < 0.0002; Dunnett, P < 0.01).

DISCUSSION

GSK-3 is a key regulatory component of a large number of cellular processes, and aberrant control of GSK-3-regulated pathways plays a role in a number of human diseases, such as diabetes, Alzheimer's disease, and cancer (13, 26, 31). More than 40 proteins are phosphorylated by GSK-3, and these substrates include metabolic proteins, cytoskeleton proteins, and transcription factors (13). Here we identify a new GSK-3 substrate, the focal adhesion-associated protein paxillin. GSK-3 phosphorylates paxillin at serine 126 and requires priming by phosphorylation of serine 130. Our findings suggest that GSK-3-mediated serine 126 phosphorylation is a general response to activation of the ERK pathway and is controlled by ERK-dependent priming. Using a mutant of paxillin that cannot be phosphorylated at serines 126 and 130 as a tool to explore the function of these phosphorylation events, we provide evidence that phosphorylation of these sites controls cell spreading and neurite extension in macrophage and neuronal cell lines, respectively.

Paxillin phosphorylation at serine 126 is ERK dependent, but the site is not a direct ERK phosphorylation site, in contrast to serine 83, which is directly phosphorylated by ERK (23, 28). Interestingly, phosphorylation of serine 83 depends upon phosphorylation of paxillin at tyrosine 118, which creates a docking site for ERK and consequently phosphorylation of paxillin at serine 83 (24). The ERK-dependent phosphorylation of serine 126/130 is mediated by a distinct mechanism. Pharmacological inhibitors that block paxillin tyrosine phosphorylation did not block serine 126 phosphorylation. Further, a paxillin mutant with phenylalanine substitutions for the major sites of tyrosine phosphorylation, Y31F/Y118F, was phosphorylated at serine 126 following LPS stimulation. These data suggest that there are two different ERK-mediated paxillin phosphorylation events and that these occur via two different mechanisms, one of which is dependent upon paxillin tyrosine phosphorylation (serine 83) and the other independent (serine 126/130). Thus, tyrosine phosphorylation provides an intriguing mechanism to direct site-selective phosphorylation of paxillin by ERK.

ERK has been identified as a priming kinase for GSK-3-mediated paxillin phosphorylation at serine 126. In addition to paxillin, this ERK/GSK-3 dual-kinase mechanism of phosphorylation has been reported for heat shock transcription factor 1 and results in the inactivation of heat shock transcription factor 1 (8, 20). Another ERK-dependent mechanism of GSK-3 activation has also been reported, in which ERK can operate through a downstream kinase to activate GSK-3 (16). However, this mechanism is unlikely to control ERK/GSK-3-mediated paxillin phosphorylation at serine 126 for several reasons. First, on its own the active form of GSK-3 cannot phosphorylate serine 126 of paxillin in vitro. Second, expression of constitutively active GSK-3 cannot induce serine 126 phosphorylation in vivo (data not shown). Third, disruption of the priming site, serine 130, can inhibit serine 126 phosphorylation both in vivo and in vitro. These results suggest that paxillin is phosphorylated by an ERK/GSK-3 dual-kinase mechanism and that the major regulatory event is substrate priming for GSK-3 phosphorylation by ERK.

Why do cells use such a complicated regulatory mechanism to control GSK-3-mediated substrate phosphorylation? The requirement for the precise temporal and spatial regulation of downstream signaling may be the answer to this question. Signaling events might be transduced only under local conditions where the priming kinase is activated and where GSK-3 remains active due to the absence of inhibitory signals regulating GSK-3 serine phosphorylation. In macrophages, LPS induces phosphatidylinositol 3-kinase/AKT activation, resulting in the phosphorylation-dependent inactivation of GSK-3. This results in the nuclear accumulation of β-catenin and transcription of genes regulated by the β-catenin/Lef complex (30). In contrast, GSK-3-mediated paxillin phosphorylation is stimulated by ERK activation via a priming mechanism upon LPS stimulation. Thus, two different downstream signaling pathways controlled by LPS have opposing effects on GSK-3 signaling, one impairing β-catenin phosphorylation and the other promoting paxillin phosphorylation. Similarly, in neuronal cells, both activation and inhibition of GSK-3-regulated signaling processes are apparently required to promote neuronal cell polarity and axon extension. The dephosphorylation of two GSK-3 substrates, APC and collapsin response mediator protein 2, is required for axon extension, and concomitantly the phosphorylation of microtubule-associated protein 1B by GSK-3 is required (10, 17, 44, 46). We have now shown that paxillin is phosphorylated by GSK-3 downstream of NGF stimulation and that phosphorylation of paxillin is required for efficient neurite outgrowth. These paradoxical observations can be reconciled if populations of active and inactive GSK-3 are spatially segregated within the cell.

Paxillin plays an important role in neurite outgrowth, since a number of mutants act in a dominant-negative fashion to retard this process. Expression of a paxillin variant lacking the LD4 motif inhibits the neurite extension of PC12 cells upon EGF stimulation (25). Expression of the paxillin p38 mitogen-activated protein kinase phosphorylation site mutant, S85A, also strongly inhibits the neurite outgrowth of PC12 cells following NGF stimulation (21). It has been reported that both GSK-3 and ERK activities are required for neurite outgrowth (7, 33), and phosphorylation of paxillin at serines 126 and 130 via GSK-3 and ERK is also important for NGF-induced neurite extension in PC12 cells. Collectively these data suggest that multiple sites of phosphorylation on paxillin, which are regulated by different kinases, may all function in the control of neuronal cell polarity. This is an intriguing mechanism where multiple signaling pathways may converge on the same substrate to regulate a biological response.

How ERK/GSK-3-mediated phosphorylation of paxillin modulates cytoskeleton rearrangement remains to be elucidated. Since serines 126 and 130 are in proximity to the LD2 motif, it seems likely that phosphorylation regulates binding to other proteins. However, our results and other published data suggest phosphorylation of paxillin at these sites does not appreciably alter the binding of paxillin to any proteins that are known to dock to LD2 (data not shown) (42). It is also possible that paxillin phosphorylation is involved in control of the activity of paxillin-associated signaling molecules. In paxillin null cells, the associated tyrosine kinase FAK exhibits a defect in tyrosine phosphorylation, suggesting a role for paxillin in controlling its activity (18). However, null cells reexpressing wild-type paxillin or the S126A/S130 mutant exhibit similar levels of FAK phosphorylation (data not shown). Perhaps paxillin phosphorylation modulates the activities of other associated enzymes, including ArfGAP, PKL/Git1, and the Rac exchange factor, Cool-1/Pix. Interestingly, phosphorylation of serine 83 of paxillin has been linked to the control of Rac activity in epithelial cells (23). Investigation of the mechanism of control of downstream signaling pathways by serine 126/130 phosphorylation of paxillin will be the focus of future investigations.