The SH3 Domain Directs Acto-Myosin-Dependent Targeting of v-Src to Focal Adhesions via Phosphatidylinositol 3-Kinase

Abstract

The v-Src oncoprotein is translocated to integrin-linked focal adhesions, where its tyrosine kinase activity induces adhesion disruption and cell transformation. We previously demonstrated that the intracellular targeting of Src is dependent on the actin cytoskeleton, under the control of the Rho family of small G proteins. However, the assembly of v-Src into focal adhesions does not require its catalytic activity or myristylation-dependent membrane association. Here, we report that the SH3 domain is essential for the assembly of focal adhesions containing the oncoprotein by mediating a switch from a microtubule-dependent, perinuclear localization to actin-associated focal adhesions; furthermore, v-Src translocation to focal adhesions requires myosin activity, at least under normal conditions when the actin cytoskeleton is being dynamically regulated. Although the SH3 domain of v-Src is also necessary for its association with focal adhesion kinase (FAK), which is often considered a likely candidate mediator of focal adhesion targeting via its carboxy-terminal targeting sequence, we show here that binding to FAK is not essential for the targeting of v-Src to focal adhesions. The p85 regulatory subunit of phosphatidylinositol (PI) 3-kinase also associates with v-Src in an SH3-dependent manner, but in this case inhibition of PI 3-kinase activity suppressed assembly of focal adhesions containing the oncoprotein. Thus, the Src SH3 domain, which binds PI 3-kinase and which is necessary for activation of Akt downstream, is required for the actin-dependent targeting of v-Src to focal adhesions.

The dynamic regulation of the eukaryotic cell adhesion network and cytoskeleton controls cell shape, adhesive strength, and dependent physiological processes such as cell motility. In addition, adhesive interactions themselves can contribute to cytoskeletal organization and, reciprocally, the cellular cytoskeleton can influence the assembly and function of cell interactions, including those mediated by both integrins and cadherins (48). Thus, the interplay between the assembly-disassembly cycles of cellular adhesions and the cytoskeleton is both complex and crucial to the proper functioning of the cell. Although a great deal of information about the dynamic regulation of the adhesion and cytoskeletal network has been gathered over the past decade (48), including the important role played by the Rho family of GTP-binding proteins (27), we still lack an understanding of the link between the biochemical regulators and the biophysical changes that lead to altered cell structure.

Focal adhesions are specialized structures where cells contact the surrounding extracellular matrix (ECM) (6, 7, 33). They consist of clustered integrin heterodimers, structural or cytoskeleton-associated proteins that link the ECM, through the integrins, to the actin cytoskeleton, and proteins involved in intracellular signal transduction (reviewed in reference 3). While it is well accepted that focal adhesion structures are constantly being assembled, altered, and disassembled as cells move or respond to their extracellular environment, the mechanics of integrin clustering and focal adhesion assembly and how these events are tightly controlled by biochemical signals within the cell remain to be established. One particularly complex relationship is that between focal adhesion dynamics and tyrosine phosphorylation. On one hand, focal adhesions can clearly form in the absence of detectable tyrosine phosphorylation of their components (19, 23), while on the other, agents that stimulate tyrosine phosphorylation often promote focal adhesion formation (reviewed in reference 48). An explanation for these apparently paradoxical findings is that tyrosine phosphorylation at focal adhesions is required for the assembly of signaling complexes, mediated in part by SH2 domain-phosphotyrosine interactions, but that assembly of focal adhesion components into adhesive structures does not require tyrosine phosphorylation (reviewed in reference 48). Furthermore, for cells transformed by the v-Src tyrosine kinase there is abundant evidence that tyrosine phosphorylation of both structural and signaling proteins at focal adhesions is linked to adhesion disassembly and to disruption of the associated actin cytoskeleton; most likely, these effects are mediated by direct tyrosine phosphorylation of adhesion components (8, 19, 20, 24, 28, 42) rather than via altered gene expression (4, 21). These tyrosine kinase-induced changes result in the loss of normal control of cell adhesion and actin organization evident during oncogenic transformation and lead to deregulation of processes that are dependent on these cellular structures.

In order to understand further the regulation of focal adhesion assembly, we have studied the intracellular targeting of a v-Src protein that is temperature dependent (ts) for both focal adhesion targeting and transformation. This molecular tool allows the conditional assembly of focal adhesions in which v-Src is present and has enabled us to address the molecular determinants and features of the assembly process. Specifically, at the nonpermissive temperature, the ts LA29 v-Src protein locates around the perinuclear region of the cell (17, 18). Upon activation by a switch to the permissive temperature, v-Src colocalizes with actin stress fibers and is assembled into focal adhesions by a process that is dependent on the activity of Rho family proteins and the integrity of the actin cytoskeleton (17, 18). Other focal adhesion components, including focal adhesion kinase (FAK) and paxillin, colocalize with v-Src as discrete protein complexes in the cell interior at early times after v-Src activation and in focal adhesions at the cell periphery at later times, indicating that v-Src is coassembled with these proteins into peripheral adhesions (17).

To study the structural determinants of v-Src translocation, we used mutant derivatives of ts LA29 v-Src that were constitutively kinase inactive or myristylation defective to demonstrate that neither catalytic activity nor N-terminal myristylation is required for the assembly of v-Src into focal adhesions (17). However, both tyrosine kinase activity and myristylation-mediated membrane association are required for v-Src-induced focal adhesion turnover during transformation and cell motility (17). In this study, we demonstrate that inactive v-Src is apparently constrained in the perinuclear region of the cell by the microtubule network. Upon activation, this constraint is removed and v-Src associates with polymerized actin and is assembled into focal adhesions at the cell periphery, events that normally require the integrity of the SH3 domain of v-Src, its association with phosphatidylinositol (PI) 3-kinase, and myosin-induced bundling of actin filaments.

MATERIALS AND METHODS

Generation of CE cells expressing v-Src mutants and FAK.

Primary chicken embryo (CE) cells were routinely grown in Dulbecco's modified Eagle's medium supplemented with 5% newborn calf serum, 1% chick serum, and 10% tryptose phosphate. The generation of v-Src-expressing CE cells was similar to that described previously (17). Briefly, cultures were transfected with replication-competent avian retroviral RAV-src or RCAS-fak constructs (5 μg per 25-cm² flask) by the DOTAP method (Roche) and subcultured at the permissive temperature of 35°C until the cultures were uniformly infected and were expressing Src protein (judged by protein immunoblotting). The generation of retrovirus encoding ts LA29 v-Src has been described (51). Retrovirus encoding the kinase-defective (KD) variant of ts LA29 v-Src was generated by converting the ATP binding site at position 295 from Lys to Arg using PCR mutagenesis as described previously (17). A mutation that has previously been shown to inhibit SH3 domain-binding activity (Trp to Ala at position 118 of Src) (15) was introduced to ts LA29 v-Src using the mutant sense oligonucleotide 5′-GAA GGT GAC GCG TGG CTG GCT CA-3′ to produce ts LA29 v-Src-W118A. A Tyr-to-Phe mutation at position 397 of avian FAK was generated using the mutant sense oligonucleotide 5′-CAG AAA CAG ATG ACT TTG CAG AGA TAA TAG ATG-3′. To detect exogenous FAK, a myc tag sequence was introduced at the BclI site at position 3186 of the avian fak sequence using the double-stranded oligonucleotide encoding the c-Myc 9E10 epitope with BclI-compatible termini. RCAS-Pro2 FAK was myc tagged as described above. For experiments that required coexpression of both v-Src and myc-tagged FAK, v-Src was cloned into the replication-defective, neomycin-selectable avian retrovirus SFCV-LE (22). Cell cultures infected with retroviruses encoding ts v-Src mutants and/or myc-tagged FAK were grown either at the restrictive (41°C) or permissive (35°C) temperature and were buffered with 5% CO₂.

Generation of FAK^−/− cells expressing v-Src.

FAK^−/− cells that were also null for p53 (as described in reference 32) were cultured in Dulbecco modified Eagle medium supplemented with 1% nonessential amino acids, 10% fetal calf serum, and 10 μM 2-mercaptoethanol at 37°C. They were transiently transfected with the murine retrovirus FpGV-1 (11) containing wild-type PrA v-Src using Superfect (Qiagen), maintained at 37°C for 24 h, and then fixed and stained for v-Src as described above for CE cells.

Protein immunoprecipitation and immunoblotting.

For protein analyses, dishes of cells were washed with cold phosphate-buffered saline (PBS) and drained. For immunoprecipitation, monolayers were lysed in radioimmunoprecipitation assay (RIPA buffer) (50 mM Tris [pH 7.4], 150 mM NaCl, 5 mM EGTA, 0.1% sodium dodecyl sulfate [SDS], 1% NP40, 1% deoxycholate) with inhibitors (10 mM pyrophosphate, 0.5 mM sodium fluoride, 1 mM phenylmethylsulfonyl fluoride [PMSF], 10 μg of aprotinin [Sigma]/ml, 100 μM sodium vanadate, 10 μg of leupeptin [Sigma]/ml, 10 μg of benzamidine [Sigma]/ml), clarified by a high-speed spin at 4°C, and precleared with normal immunoglobulin G (IgG) and protein A-Sepharose (Sigma). Cell lysate (750 μg; measured by the Micro BCA protein assay kit [Pierce]) was immunoprecipitated with 2 μg of anti-Src monoclonal antibody (MAb) EC10 (Upstate Biotechnology) or 5 μl of anti-FAK carboxy-terminal peptide serum. Immune complexes were collected on either protein A-Sepharose, anti-mouse IgG-coated protein A-Sepharose, or anti-mouse IgG-conjugated agarose (Sigma); precipitated proteins were washed four times with RIPA buffer and once with 0.6 M lithium chloride, eluted using buffer containing a high level of SDS at high temperature, and separated by SDS–7.5% polyacrylamide gel electrophoresis (PAGE). v-Src and FAK or their associated proteins were detected by transferring immunoprecipitated proteins to nitrocellulose, blocking with 3% bovine serum albumin (Sigma) or 5% low fat milk in PBS–0.2% Tween 20 (Sigma), and probing with EC10 (Upstate Biotechnology) at 1:1,000, anti-FAK (Transduction Laboratories) at 1:500, anti-Myc horseradish peroxidase (HRP) conjugate (Invitrogen) at 1:2,000, anti-HSP-90 (Upstate Biotechnology) at 1:400, anti-p110^AFAP (Transduction Laboratories) at 1:250, or anti-p85 (Upstate Biotechnology) at 1:200. Detection was by incubation with HRP-conjugated secondary antibody, and visualization was by enhanced chemiluminescence (Amersham). To monitor protein expression in whole-cell lysates, 5 or 20 μg of total protein was separated by SDS-PAGE and immunoblotted using antibodies as described above, anti-Akt or anti-Akt phosphorylated at Ser-473 (anti-phospho-Akt-Ser-473) (New England Biolabs), both at 1:1,000, or anti-phospho-Src-Tyr-416 (Biosource Inc.) at 1:1,000.

Immunofluorescence.

Subconfluent cells were grown on glass coverslips at the nonpermissive temperature (41°C) and shifted to the permissive temperature (35°C) in some cases. Inhibitors were added to the cultures 1 h prior to the temperature shift (unless otherwise stated) at the following concentrations: BDM (Sigma), 10 mM; nocodazole (Sigma), 0.5 μg/ml; LY294002 (Calbiochem), 20 μM; ML-7 (Calbiochem), 25 μM. Cells were fixed at 4°C for 15 min with 3.7% formaldehyde, permeabilized with 0.5% Triton X-100, and incubated with primary antibodies at the following concentrations: 1:500 for anti-Src MAb EC10, 1:500 for anti-Src rabbit polyclonal antibody MW2 (51), 1:500 for anti-α-tubulin (Sigma), 1:200 for anti-Myc 9E10 (Roche), 1:200 for anti-myosin light chain (MLC) (Sigma), 1:200 for antipaxillin (Transduction Laboratories, and 1:500 for anti-FAK BC3 (Upstate Biotechnology). Antibody detection was by reaction with fluorescein isothiocyanate (FITC)- or Texas red-conjugated secondary antibodies (Jackson Laboratories) at 1:200 for 45 min at room temperature. As a control, cells were incubated with secondary antibody alone (not shown). To visualize actin stress fibers, cells were stained with FITC- or tetramethyl rhodamine isocyanate-conjugated phalloidin at 1 μg/ml (Sigma). Stained cells were visualized by optical sectioning with a confocal microscope (model MRC 600; Bio-Rad), and images were printed on a dye sublimation printer (Kodak).

Cytoskeletal protein preparations.

F-actin fractionation was carried out with minor modifications to a procedure described previously (44). Briefly, cell monolayers were washed with cold PBS, and cytoskeletons were extracted in CSK buffer (0.5% Triton X-100, 20 mM HEPES [pH 7.4], 50 mM NaCl, 1 mM EGTA, 1 mM PMSF, 10 μg of leupeptin/ml, 100 μM sodium vanadate, 5 μg of aprotinin/ml). The insoluble material was pelleted at 12,000 × g for 5 min at 4°C and washed twice with cold CSK buffer. The pellet was dissolved in 0.6 M KI–100 mM PIPES (piperazine-N,N′-bis(α-ethanesulfonic acid) (pH 6.5)–100 mM KCl–10 μg of leupeptin/ml–1 mM PMSF–100 μM sodium vanadate at 4°C for 30 min. Insoluble material was pelleted at 40,000 × g for 20 min at 4°C, and the supernatant was extensively dialyzed against 10 mM PIPES (pH 6.8)–0.5 mM EGTA–2 mM MgCl₂–1 mM PMSF–5 mM benzamidine–100 μM sodium vanadate. The resulting granular precipitate, enriched in F-actin, was collected by centrifugation at 12,000 × g for 5 min at 4°C. A further round of solubilization and dialysis was performed, and the precipitate was dissolved in SDS sample buffer. Preparations were normalized with respect to actin concentration by titration on SDS-PAGE minigels and immunoblotted using anti-Src EC10 as described above and antiactin (Sigma) at 1:500.

RESULTS

The SH3 domain of v-Src is required for actin association and assembly of v-Src into focal adhesions.

To determine the requirements of v-Src's intracellular targeting to focal adhesions, we previously mutated the catalytic site and the site of myristylation to show that neither kinase activity nor amino-terminal myristylation was required. Here, we have investigated the role of the SH3 domain by generating a mutant v-Src protein, ts LA29 v-Src-W118A, that inhibits SH3-binding interactions (15). This rendered the v-Src protein incapable of translocation to focal adhesions upon a shift to the permissive temperature (Fig. 1A, right). Compared to ts LA29 v-Src or its kinase-defective variant (Fig. 1A, left and center, respectively), the SH3 mutant v-Src protein remained cytoplasmic, even after prolonged times at the permissive temperature (Fig. 1A, right). Although v-Src was unable to assemble into focal adhesions in the absence of SH3 function, paxillin was still evident in residual focal adhesions of cells expressing the ts v-Src SH3 mutant protein (Fig. 1A, lower right). However, we noted that long-term expression of ts LA29 v-Src-W118A at the permissive temperature had a detrimental effect on cell growth and viability (not shown), implying that the presence of active Src kinase in the cytoplasm or the inability to assemble new focal adhesions in the presence of the v-Src SH3 mutant was toxic.

FIG. 1.

The SH3 domain of ts LA29 v-Src is required for translocation and actin association. (A) CE cells expressing ts LA29 v-Src, its KD variant ts LA29 v-Src-KD, or ts LA29 v-Src which has a point mutation in the SH3 domain (W118A) were maintained at 41°C or were shifted to 35°C for 2, 7, or 24 h prior to staining with anti-Src (and antipaxillin for cells after 24 h at 35°C) and a secondary antibody coupled to Texas red or FITC. Bars, 25 μm. (B) CE cells expressing ts LA29 v-Src or its kinase-defective variant, ts LA29 v-Src-KD, or the SH3 domain mutant, ts LA29 v-Src-W118A were maintained at 41°C or were shifted to 35°C for 30 or 60 min. Cytoskeletons were extracted and subjected to two rounds of partial depolymerization and repolymerization as described in Materials and Methods. After the final polymerization, preparations were solubilized and proteins were separated by SDS–10% PAGE and immunoblotted using anti-Src or antiactin as the probe. As a control (B, top left), a whole-cell lysate of ts v-Src-expressing cells grown at 35°C was immunoblotted with anti-Src. (C) Whole-cell lysates of ts LA29 v-Src, ts LA29 v-Src-KD-, or ts LA29 v-Src-W118A-expressing cells were immunoblotted using anti-Src-phospho-Tyr-416 (upper) or anti-Src (lower) as the probe. The positions of v-Src are indicated. CEF, CE fibroblasts.

Our previous work demonstrated colocalization of activated ts v-Src with the actin cytoskeleton (18). Therefore, we tested whether we could demonstrate a biochemical association between v-Src and cellular polymerized actin and whether this was influenced by SH3 domain integrity. Cytoskeletons were extracted from CE cells expressing ts LA29 v-Src at the restrictive temperature and after the shift to the permissive temperature and subjected to 2 cycles of partial depolymerization and polymerization by treatment with KI and dialysis, respectively (as described previously [44] and in Materials and Methods). At the restrictive temperature, ts LA29 v-Src was not visibly associated with polymerized actin (Fig. 1B, upper left). However, after the shift to the permissive temperature, association was readily detected by 30 min and continued to increase at 60 min (Fig. 1B, upper left). Consistent with the ability of a KD mutant of ts LA29 v-Src to translocate to focal adhesions (17), this mutant v-Src protein also associated with polymerized actin after the switch to the permissive temperature (Fig. 1B, upper middle). In contrast, although there was some detectable association of the v-Src-W118A SH3 mutant with polymerized actin at the restrictive temperature, this was not stimulated by the switch to the permissive temperature (Fig. 1B, upper right). The relative amount of cellular actin under each of these conditions is shown (Fig. 1B, lower). Thus, activation of ts LA29 v-Src induces an SH3-dependent increase in association with polymerized actin and translocation to focal adhesions at the termini of actin stress fibers. Since there was a small amount of the v-Src-W118A SH3 mutant apparently binding to actin, even at the restrictive temperature (Fig. 1B), it is formally possible that functional SH3 interactions might normally suppress the association of a small proportion of inactive v-Src with actin; however, while the v-Src proteins that translocate to focal adhesions showed a substantial increase in association with cellular actin after activation, the SH3 mutant did not.

To determine expression levels and states of activity of the ts v-Src mutants, we probed lysates with anti-Src (Fig. 1C, lower) or with an antiserum that we have previously used to monitor the activation-specific Src Tyr-416 residue in its phosphorylated form (43) (Fig. 1C, upper). These immunoblots confirmed that exogenous v-Src was expressed (no c-Src was detected at this exposure, as can be judged by the lack of signal in CE fibroblast lysate [Fig. 1C, lower]) and that both ts LA29 v-Src and the SH3 mutant, ts LA29 v-Src-W118A were enzymatically active after the shift to the permissive temperature (Fig. 1C, upper).

Integrin-induced assembly of new focal adhesions containing v-Src is SH3 dependent.

Since our work on the targeting of v-Src to focal adhesions has been carried out with adherent cells, we sought to determine whether v-Src was incorporated into newly assembling focal adhesions after integrin engagement. Thus, cells expressing ts v-Src were plated onto fibronectin and onto poly-l-lysine for comparison. We found that v-Src did not translocate to the cell periphery when cells were plated on poly-l-lysine (Fig. 2A, upper), indicating a requirement for integrin engagement induced after attachment to fibronectin (Fig. 2A, lower). The finding that the peripheral targeting of v-Src was integrin dependent was surprising, since activated v-Src is often regarded as a surrogate integrin signal that can function independently of cellular integrin function. That control of the peripheral targeting of v-Src was integrin dependent implies that the “inside-out” signaling required for focal adhesion assembly after integrin engagement is also necessary for the assembly of focal adhesions containing the oncoprotein in adherent cells. This further suggests that studying the requirements for the assembly of focal adhesions that are tracked by the presence of ts v-Src after the switch to the permissive temperature is relevant also to integrin-induced focal adhesion assembly. In this regard, we found that cells expressing the v-Src SH3 mutant, ts LA29 v-Src-W118A, were impaired at spreading on fibronectin, with apparently fewer focal adhesions formed (Fig. 2B, right). To monitor spreading on fibronectin with another focal adhesion protein, we double-stained cells expressing ts v-Src or the SH3 mutant with both anti-Src and antipaxillin. This showed that ts v-Src colocalized with paxillin during the formation of multiple focal adhesions as cells spread on fibronectin at the permissive temperature (Fig. 2C, upper). In contrast, the assembly of focal adhesions containing paxillin was impaired by the SH3 mutant of v-Src (Fig. 2C, lower); however, in cells expressing low levels of the SH3 mutant v-Src protein, some paxillin staining in small structures at the cell periphery was evident (Fig. 2C, lower right). Thus, overexpression of the SH3 mutant v-Src protein impairs focal adhesion assembly after cells have attached to fibronectin, indicating a need for SH3 adapter function for integrin-induced focal adhesion assembly, as well as for assembly of focal adhesions containing v-Src in adherent cells.

FIG. 2.

The peripheral targeting of v-Src is integrin dependent. (A) CE cells expressing ts LA29 v-Src that had been grown at 41°C were trypsinized, held in suspension, and plated onto poly-l-lysine- or fibronectin-coated dishes for 40 min at either 41 or 35°C. Fixed cells were stained with anti-Src (and a secondary antibody coupled to FITC). (B) CE cells expressing the SH3 domain mutant, ts LA29 v-Src-W118A, were plated onto fibronectin for 40 min at either 41 or 35°C. (C) CE cells expressing ts LA29 v-Src or the SH3 domain mutant, ts LA29 v-Src-W118A, were plated onto fibronectin for 40 min at 35°C. Fixed cells were stained with anti-Src and antipaxillin (and secondary antibodies coupled to Texas red or FITC). Broken arrows indicate small structures at all periphery; some paxillin staining is visible in the lower right panel. Bars, 25 μm.

Inactive v-Src is constrained in the perinuclear region by the microtubule network and switches to actin after activation.

In our previous studies with serum-deprived Swiss 3T3 and CE cells, we found that ts v-Src rapidly translocated from the perinuclear region to focal adhesions at the cell periphery in association with actin stress fibers. To further investigate the cytoskeletal associations during the peripheral targeting of v-Src, we costained CE cells at the nonpermissive temperature (41°C) with anti-α-tubulin and anti-Src. We found that, in their inactive state, v-Src and α-tubulin displayed similar levels of perinuclear staining (Fig. 3A). Furthermore, upon treatment of cells at the nonpermissive temperature with 0.5 μg of nocodazole/ml, v-Src was no longer perinuclear (Fig. 3B), implying that the tight retention of inactive v-Src in the perinuclear region was dependent on the integrity of the microtubule network.

FIG. 3.

Inactive v-Src colocalizes with regions of dense tubulin staining and switches to actin after activation. Confocal images are of CE cells expressing ts LA29 v-Src stained with anti-Src (and a secondary antibody coupled to Texas red) and anti-α-tubulin (and a secondary antibody coupled to FITC). v-Src was maintained in the inactive state at 41°C, and cells were either untreated (A) or treated with 0.5 μg of nocodazole/ml (B). Arrows (A), regions of colocalization of v-Src and tubulin. (C) Cells were stained with anti-Src (visualized as red) and anti-α-tubulin (visualized as green) after the switch to the permissive temperature (35°C) for 2 h. In the merged image, v-Src-staining focal adhesions at the cell periphery are indicated by arrows. (D) v-Src-containing complexes at the ends of MLC-staining filaments are shown in CE cells expressing ts LA29 v-Src that were shifted to 35°C for 1 h and double stained with both anti-Src (followed by a secondary antibody coupled to FITC) and anti-MLC (followed by a secondary antibody coupled to Texas red). (E) CE cells expressing ts LA29 v-Src were shifted to 35°C for 1 h and double stained with both anti-Src (followed by a secondary antibody coupled to FITC) and phalloidin-tetramethyl rhodamine isocyanate. Arrows, v-Src-containing complexes colocalizing with the ends of actin filaments. Bars, 25 μm.

As we showed previously (17, 18), some of the expressed v-Src was recruited to focal adhesions at the cell periphery upon activation by the shift to the permissive temperature (35°C) (Fig. 3C). Furthermore, these peripheral v-Src-containing focal adhesions were associated with actin stress fibers (17, 18) (Fig. 3E). Here we showed by double staining with anti-Src and anti-α-tubulin after 2 h at the permissive temperature that most of the adhesion structures containing Src no longer colocalized with dense regions of tubulin staining in the perinuclear region (Fig. 3C). Thus, upon the switch to the permissive temperature, v-Src is released from its microtubule-dependent localization in the perinuclear region of the cell, becomes associated with actin, and is assembled into focal adhesions at the ends of actin stress fibers. However, our experiments have not distinguished whether the switch to actin occurs in the perinuclear region or close to sites of formation of peripheral focal adhesions. Nonetheless, the formation of v-Src-containing focal adhesions is apparently independent of microtubules once assembly has been initiated by a shift to the permissive temperature (18).

Myosin ATPase activity is required for translocation of kinase-active v-Src, but not kinase-defective v-Src, to focal adhesions.

Since we have established that translocation of ts LA29 v-Src to the cell periphery requires release from the perinuclear region and association with actin mediated by the Src SH3 domain, we sought to refine our understanding of the determinants of actin organization that contribute to the assembly of focal adhesions containing v-Src. Our previous work demonstrated the involvement of the small G protein RhoA, mediated by its effects on actin organization (18). Here we have also addressed the role of myosin during conditions under which the actin cytoskeleton was being dynamically regulated or was stabilized by the dominant-negative action of overexpressed KD v-Src. First, we stained CE cells with an antibody to MLC and found that it localized along actin stress fibers (Fig. 3D). However, in contrast to the phalloidin-stained actin stress fibers that overlapped with v-Src at the stress fiber termini (visualized as yellow in the merged image in Fig. 3E), MLC was mostly excluded from the v-Src-containing focal adhesions (as judged by the predominantly green staining of v-Src in the merged image in Fig. 3D). Since the ATPase activity of myosin is regulated by phosphorylation of MLC by MLC kinase (MLCK) (1, 12), we tested whether the MLC associated with stress fibers was involved in v-Src translocation. For this, we used butanedione-2-monoxime (BDM), a myosin ATPase inhibitor (29) that blocks muscle contractility by slowing the rate of phosphate release from myosin after ATP hydrolysis (40) and also inhibits the ATPase activity of both nonmuscle myosin II and unconventional myosins (10). Treatment of CE cells expressing ts LA29 v-Src with 10 mM BDM at the time of the shift to the permissive temperature inhibited the translocation of v-Src to focal adhesions (Fig. 4A, lower right), although residual FAK-containing focal adhesions persisted in treated cells (Fig. 4A, lower left). In contrast, translocation of KD ts LA29 v-Src-KD to focal adhesions was unaffected by BDM treatment (Fig. 4B, lower right). Consistent with our previous findings, KD v-Src was localized in enlarged focal adhesions that were stabilized as a consequence of impaired turnover induced by the dominant-negative action of KD v-Src (17). In addition to BDM, an inhibitor of MLCK, ML-7 (47), had similar effects on v-Src translocation (Fig. 4C).

FIG. 4.

The myosin ATPase inhibitor BDM and the MLCK inhibitor ML7 block kinase-active v-Src translocation. CE cells expressing ts LA29 v-Src (A) or KD ts LA29 v-Src-KD (B) were maintained at 41°C or shifted to 35°C for 2 h in the presence of BDM and stained with anti-Src (followed by a secondary antibody coupled to FITC; right) and anti-FAK (followed by a secondary antibody coupled to Texas red; left). Arrows, enlarged peripheral adhesions. (C) Cells expressing ts LA29 v-Src (upper) or ts LA29 v-Src-KD (lower) after the shift to 35°C for 2 h in the absence (−) or presence (+) of ML7 were stained with anti-Src (followed by a secondary antibody coupled to FITC). Bars, 25 μm.

KD v-Src acts in a dominant-negative manner to inhibit actin remodeling.

To investigate whether the different sensitivities to BDM of kinase-active and KD v-Src were due to a difference in the stability of cross-linked actin filaments, we examined the actin cytoskeleton in v-Src-expressing, BDM-treated and untreated cells. At the restrictive temperature, actin stress fibers could be visualized in the absence of BDM (Fig. 5A, left) and residual stress fibers were also still evident in kinase-active v-Src-expressing cells after 2 h at the permissive temperature (Fig. 5A, top right). However, we consistently observed that the intensity of phalloidin-stained stress fibers was substantially greater in the KD v-Src-expressing cells at the permissive temperature (Fig. 5A, bottom right). After treatment with BDM for 2 h, there were few remaining stress fibers detected by phalloidin staining in cells expressing either mutant of v-Src at the restrictive temperature, although some cortical actin was still visible around the cell edges (Fig. 5B, left). However, at the permissive temperature, there was a substantial difference between kinase-active and KD v-Src-expressing cells, with the latter displaying much higher levels of phalloidin-staining actin even in the presence of BDM (Fig. 5B, right). These data imply that BDM was inhibiting the new formation of bundled actin filaments in cells in which the actin was being dynamically regulated, i.e., in cells expressing either v-Src mutant at the restrictive temperature or the kinase-active v-Src mutant at the permissive temperature. However, in KD v-Src-expressing cells at the permissive temperature, phalloidin-staining actin stress fibers were unaffected by treatment with BDM, indicating that the existing actin filaments were not being turned over and that cytoskeletal integrity was independent of the new formation of bundled actin filaments.

FIG. 5.

Actin stress fiber stability is unaffected by BDM in cells expressing KD ts LA29 v-Src at the permissive temperature. CE cells expressing either ts LA29 v-Src or KD ts LA29 v-Src-KD were maintained at 41°C or shifted to 35°C for 2 h and stained with phalloidin-FITC to visualize actin stress fibers. Cells were either untreated (− BDM; A) or treated with 10 mM BDM for 2 h at 41°C or during the 2-h period of the shift to 35°C (+ BDM; B).

Thus, the actin cytoskeleton was apparently stabilized as a consequence of the dominant-negative effect of KD ts LA29 v-Src-KD at the permissive temperature, and the requirement for myosin-dependent functions to maintain an organized cytoskeleton was lost. These findings imply that the role of myosin in the intracellular targeting of v-Src to focal adhesions is to maintain bundled actin stress fibers in a state that is permissive for translocation when the actin is constantly being remodeled. In contrast, when the actin cytoskeleton is artificially stabilized by expression of KD v-Src, myosin ATPase activity and the new formation of bundled actin stress fibers are no longer required. Consistent with the observed cytoskeletal disassembly in v-Src-transformed CE cells (16), these findings imply that activated Src normally antagonizes the pathway leading to myosin-induced cross-linking and organization of actin filaments during routine remodeling of the actin cytoskeleton.

Stabilization of the adhesion and cytoskeletal network permits translocation of KD v-Src into preexisting focal adhesions.

Since one effect of KD ts LA29 v-Src was to inhibit the turnover of bundled actin stress fibers and the adhesions at their termini, we reasoned that translocation of this v-Src mutant must be in association with existing stabilized actin and, therefore, into existing focal adhesions. Thus, we examined merged images of BDM-treated CE cells expressing KD v-Src which were double stained with anti-Src and anti-FAK (Fig. 6A) or anti-Src and antipaxillin (Fig. 6B). After 2 h at the permissive temperature, intracellular complexes containing both v-Src and FAK were evident (Fig. 6A, upper, yellow). At the cell periphery, v-Src-containing complexes were evident in the most intracellular region of enlarged focal adhesions (Fig. 6A, upper), which also contained FAK at the membrane-proximal portion of the peripheral adhesions (Fig. 6A, upper, red). Thus, KD v-Src that translocated to focal adhesions in the presence of BDM was assembled into preexisting focal adhesions that already contained FAK, as judged by the asymmetrical staining of the enlarged adhesions with anti-FAK and anti-Src. In contrast, complete colocalization of v-Src and FAK was evident in peripheral focal adhesions of cells expressing kinase-active v-Src after 2 h at the permissive temperature (visualized as yellow in Fig. 6A, lower), indicating that v-Src and FAK were normally coassembled into new focal adhesions when the actin-adhesion network was being dynamically regulated.

FIG. 6.

KD v-Src can assemble into preexisting focal adhesions. (A) Merged confocal images shown in Fig. 4B representing CE cells expressing KD ts LA29 v-Src-KD (upper) that were shifted to 35°C for 2 h in the presence of BDM (+ BDM) and stained with both anti-FAK (and a secondary antibody coupled to Texas red; visualized as red) and anti-Src (and a secondary antibody coupled to FITC; visualized as green). Grey arrow, intracellular complexes containing both FAK and v-Src (visualized as yellow); broken white arrows, enlarged peripheral adhesions containing FAK at the membrane-proximal region (visualized as red); solid white arrows, enlarged peripheral adhesions containing both FAK and v-Src at the membrane-distal region (visualized as yellow). Also shown are merged confocal images after dual staining of kinase-active ts LA29 v-Src-expressing cells (lower) that were shifted for 2 h to 35°C with anti-FAK and anti-Src. Colocalization of v-Src and FAK is visualized as yellow. (B) Merged confocal images after dual staining of KD ts LA29 v-Src-KD-expressing cells (upper) or kinase-active ts LA29 v-Src (lower) that were shifted for 2 h to 35°C with antipaxillin (and a secondary antibody coupled to Texas red) and anti-Src (and a secondary antibody coupled to FITC). Focal adhesions that stain with paxillin (visualized as red) in the membrane-proximal region (broken arrows) and both Src and paxillin (visualized as yellow) in the membrane-distal region (solid arrows) are shown. Bars, 25 μm.

A similar asymmetrical staining pattern was evident with v-Src and paxillin in KD v-Src-expressing cells at the permissive temperature. In this case, v-Src-containing complexes were present at the intracellular side of enlarged focal adhesions, with paxillin also present at the membrane-proximal region (Fig. 6B, upper). In contrast, kinase-active v-Src and paxillin completely colocalized in smaller focal adhesions at the cell periphery (Fig. 6B, lower). These data indicate that, when the actin cytoskeleton is artificially stabilized by the dominant-negative action of KD v-Src, translocation into preexisting focal adhesions occurs in association with stabilized stress fibers. Although the actin stabilization that occurs in KD v-Src-expressing cells is artificially induced, these findings are consistent with the notion that activated Src normally translocates via preformed, myosin-induced bundled actin stress fibers to focal adhesions forming at their termini.

FAK binds v-Src via the SH3 domain but is not a crucial Src targeting partner.

Having established that the SH3 domain directs the actomyosin-dependent targeting of v-Src to peripheral focal adhesions, we sought to determine the nature of SH3-binding proteins that mediate the translocation process. In our previous work, we demonstrated that ts v-Src and FAK colocalized, both in intracellular complexes containing focal adhesion proteins and in mature focal adhesions at later times after a shift to the permissive temperature (17). This, together with the fact that FAK contains a well-defined focal adhesion targeting sequence (30) and complexes with Src, at least in part, via proline-rich sequences in FAK (50), made it a good candidate for mediating the translocation of v-Src to focal adhesions. Furthermore, FAK associates with v-Src mutants that are permissive for translocation but defective for transformation (17). Thus, we tested whether the SH3 mutant of v-Src, ts LA29 v-Src-W118A, which was unable to translocate, could associate with FAK. Consistent with previous findings that mutation of a putative SH3-binding proline-rich region of FAK suppressed binding to Src in vivo (50), we found that FAK was not visibly coimmunoprecipitated with the W118A mutant of v-Src that was defective for translocation (Fig. 7A). This was in contrast to results for mutants that were permissive for translocation, including ts LA29 v-Src (Fig. 7A), its KD derivative ts LA29 v-Src-KD (shown in reference 17), and a myristylation-defective ts LA29 v-Src (shown in reference 17) which can associate with FAK.

FIG. 7.

FAK binding v-Src via the Src SH3 domain is not required for translocation. (A) v-Src was immunoprecipitated (IP) from lysates of cells expressing ts LA29 v-Src or ts LA29 v-Src-W118A (-W118A) at 41°C and after a shift to 35°C for 2 h. Precipitated proteins were immunoblotted using both anti-FAK and anti-Src as probes. The position of FAK coprecipitating with v-Src is indicated. Immunoprecipitated v-Src is also indicated. The positions of the 97- and 69-kDa molecular mass markers are shown. NI, nonimmune serum. (B) Lysates of CE cells expressing ts LA29 v-Src either alone (−) or together with wild-type myc-tagged FAK (wt) or a mutant in which Tyr-397 was replaced by Phe (397F) were immunoblotted using anti-Src, anti-myc, or anti-FAK as the probe. The exogenously expressed v-Src and FAK proteins are indicated. Endogenous FAK is not visible on the exposure shown for the anti-FAK blot, demonstrating the overexpression of exogenous retrovirus-encoded FAK proteins (endogenous FAK is visible on longer exposures [not shown]). (C) CE cells expressing both ts LA29 v-Src and myc-tagged wild-type FAK or 397F-FAK were maintained at 41°C or were shifted to 35°C for 2 h. Lysates were immunoprecipitated with anti-Src or anti-FAK and blotted using anti-myc as the probe. The position of myc-tagged FAK precipitated by anti-Src or anti-FAK is indicated. Anti-Src immunoprecipitates were also probed with anti-FAK and anti-Src. (D) CE cells expressing both ts LA29 v-Src and myc-tagged FAK-397F were maintained at 41°C or were shifted to 35°C for 2 h and were stained with anti-Src (followed by a secondary antibody coupled to Texas red) and anti-myc (followed by a secondary antibody coupled to FITC). Bars, 25 μm.

This correlation prompted us to further investigate whether FAK binding was important for v-Src translocation. We expressed a variant of FAK that is also unable to associate with v-Src as a result of mutation of FAK Tyr-397, which binds the Src SH2 domain (397F-FAK). We confirmed that both myc-tagged wild-type FAK and 397F-FAK were vastly overexpressed relative to endogenous FAK (Fig. 7B, middle and lower). Furthermore, although overexpressed wild-type FAK was coimmunoprecipitated with anti-Src, we established that neither exogenous 397F-FAK nor endogenous FAK was visibly complexed with v-Src (Fig. 7C), confirming the findings of others (13, 50, 53) and indicating that the overexpressed mutant FAK was acting as a dominant-negative inhibitor of v-Src-FAK complex formation. However, despite the inability of v-Src to bind FAK as a result of 397F-FAK overexpression, ts v-Src translocated normally to focal adhesions that also contained myc-tagged 397F-FAK after the switch to the permissive temperature (Fig. 7D). These data suggest that, although the association of v-Src and FAK required the Src SH3 domain, FAK was not the SH3-binding adapter protein that mediated actin association and translocation of v-Src to focal adhesions. We confirmed this interpretation by two other approaches. First, we overexpressed a FAK mutant (FAK-Pro2) that has the proline-rich Src-binding sequences mutated (50). We confirmed that this mutant suppressed association between v-Src and FAK in ts v-Src-expressing CE cells (Fig. 8A, immunoblot) but did not inhibit the translocation of v-Src to focal adhesions (Fig. 8A). Second, we expressed v-Src transiently in cells derived from FAK^−/− embryos (32) and could visualize v-Src in peripheral focal adhesion structures (Fig. 8B).

FIG. 8.

FAK is not essential for v-Src translocation. CE cells expressing both ts LA29 v-Src and the FAK-Pro2 mutant were maintained at 41°C or were shifted to 35°C for 2 h and stained with anti-Src (followed by a secondary antibody coupled to Texas red) and anti-myc (followed by a secondary antibody coupled to FITC). To determine whether FAK was binding Src, lysates of cells expressing wild-type (wt) FAK or FAK-Pro2 were immunoprecipitated (IP) with anti-Src, blotted, and probed with anti-Myc (lower). The position of FAK is indicated. (B) FAK^−/− cells were transiently transfected with the wild-type PrA strain of v-Src. Cells were fixed and stained with anti-Src (followed by a secondary antibody coupled to FITC). Arrows, focal adhesions containing v-Src. Also shown is an immunoblot using anti-FAK as the probe confirming FAK deficiency in the FAK^−/− cells we used. Bars, 25 μm.

The p85 regulatory subunit of PI 3-kinase binds v-Src in an SH3-dependent manner and is required for translocation.

Since the association of FAK with v-Src was apparently not mediating focal adhesion targeting, we tested the SH3 mutant v-Src protein, ts LA29 v-Src-W118A, for binding to other potential regulators of Src translocation. These included HSP90, the molecular chaperone that binds to v-Src and that is essential for its function (5), p110^AFAP, the actin filament-associated protein that acts as a determinant of actin integrity (20, 45), and p85, the regulatory subunit of PI 3-kinase that binds to v-Src and that is implicated in signaling to the cytoskeleton (25, 31, 34, 49, 52). We found that while HSP90 and p110^AFAP could be coprecipitated with either ts LA29 v-Src or its SH3 mutant derivative, ts LA29 v-Src-W118A (Fig. 9A), p85 could only be coimmunoprecipitated with v-Src when the SH3 domain was functional (Fig. 9B, left). We previously showed that the association between ts LA29 v-Src and p85 was temperature dependent (26), indicating that its binding to v-Src and also activation of its associated lipid kinase activity only occurred after a shift to the permissive temperature (26). In addition, we and others, have shown using glutathione S-transferase–SH3 and –SH2 fusion proteins that p85 binds to the v-Src SH3 domain in vitro (26, 39), although binding is modulated by the adjacent SH2 domain (26). Here, using the discriminating SH3 domain mutant of v-Src, ts LA29 v-Src-W118A, we have identified a requirement for the SH3 domain of v-Src for association with cellular p85. In keeping with the requirement of SH3-dependent association of v-Src with p85 for PI 3-kinase activation, we found that the SH3 domain mutant of ts LA29 v-Src was unable to stimulate Akt phosphorylation after a switch to the permissive temperature, as judged by reactivity with an antibody recognizing Akt Ser-473 in its phosphorylated form (2) (Fig. 9C). This was in contrast to ts LA29 v-Src, which induced robust stimulation of Akt phosphorylation after a shift to the permissive temperature (Fig. 9C). Thus, the SH3 domain mutant of ts v-Src that was unable to translocate to focal adhesions was also unable to associate with the regulatory subunit of PI 3-kinase or activate a downstream signaling protein that is dependent on the lipid products of PI 3-kinase. Our data clearly correlate the SH3 domain–PI 3-kinase functional interaction with v-Src's focal adhesion targeting; however, we previously found that although v-Src binds p85 via the SH3 domain, the binding interaction is modulated by the adjacent Src SH2 domain (26). This implies that the Src SH2 domain is also likely to contribute to the SH3-dependent targeting of v-Src to focal adhesions.

FIG. 9.

The SH3 mutant v-Src protein binds HSP90 and p110^AFAP but not the p85 regulatory subunit of PI 3-kinase. (A) Lysates of CE cells expressing ts LA29 v-Src or the SH3 mutant, ts LA29 v-Src-W118A, at 35°C were immunoprecipitated (IP) with anti-Src and immunoblotted using anti-HSP90, anti-p110^AFAP, or anti-Src as the probe. The positions of the individual immunoprecipitated proteins are shown. (B) Lysates of CE cells expressing ts LA29 v-Src or the SH3 mutant, ts LA29 v-Src-W118A, at 35°C were immunoprecipitated with anti-Src and immunoblotted using anti-p85 as the probe (left). For comparison, the expression of both anti-p85 reactive forms was examined by immunoblotting whole-cell lysates using anti-p85 as the probe (right). (C) Lysates of cells expressing ts LA29 v-Src or the SH3 mutant, ts LA29 v-Src-W118A, that were maintained at 41°C or shifted to 35°C for 30 or 60 min (in the absence [−] or presence [+] of LY294002) were immunoblotted using anti-Akt-phospho-Ser-473 (upper) or anti-Akt (lower) as the probe. The positions of phosphorylated Akt (phospho-Akt) and Akt are indicated.

The correlation between p85 binding and focal adhesion targeting described above prompted us to test the effect of a selective inhibitor of PI 3-kinase on the ability of ts v-Src to translocate to focal adhesions after a shift to the permissive temperature. We found that treatment of cells with LY294002 at a concentration that effectively blocked the activation of Akt by v-Src (Fig. 9C), suppressed the assembly of v-Src into focal adhesions (Fig. 10, right), although FAK persisted in residual focal adhesions (Fig. 10A, left). In some cells, we observed focal adhesion-like structures or complexes that remained in the interior region of the cell (Fig. 10A, lower right). These findings imply that the intracellular targeting of v-Src to focal adhesions requires the activity of cellular PI 3-kinase, which we have shown to bind to v-Src via the SH3 domain that is critical for the assembly of a focal adhesion containing the oncoprotein. Furthermore, the specific defect in cells treated with LY294002 might be in the peripheral targeting of focal adhesion protein complexes once these have formed.

FIG. 10.

The selective PI 3-kinase inhibitor LY294002 suppresses targeting of v-Src to focal adhesions. (A) Cells expressing ts LA29 v-Src at 30, 60, or 120 min after the switch to 35°C in the presence of LY294002 were fixed and stained with anti-FAK (followed by a secondary antibody coupled to Texas red) and anti-Src (followed by a secondary antibody coupled to FITC). The retention of v-Src in the cell interior is indicated (arrows). (B) Cells expressing ts LA29 v-Src either at 41°C or after 2 h at 35°C in the absence (−) or presence (+) of LY294002 were fixed and stained with phalloidin-FITC. Bars, 25 μm.

Since our previous work showed that the translocation of v-Src to peripheral focal adhesions required actin stress fibers that were under the control of RhoA (18), we examined the actin of cells that had been treated with the selective PI 3-kinase inhibitor LY294002. We found that the number of stress fibers in LY294002-treated cells was considerably reduced, although peripheral actin was clearly visible (Fig. 10B). Thus, the v-Src SH3-dependent association and activation of PI 3-kinase most likely serve to maintain the integrity of actin stress fibers that are required for the peripheral targeting of v-Src.

DISCUSSION

Despite the long-standing evidence that v-Src was located in the residual focal adhesions of transformed cells (37, 38, 41, 46), we still lack a detailed understanding of the mode of intracellular targeting of this oncogenic tyrosine kinase and its cellular counterparts. In addition to the assembly of the Src family kinases into cell-ECM adhesions, the general mechanics of focal adhesion assembly remain relatively obscure. Here we have used conditionally translocated ts v-Src mutants to define critical regulatory features associated with the assembly of focal adhesions containing the oncoprotein. Although our work with the conditional v-Src mutants provides considerably greater detail on the regulation of Src translocation, the analogy with previous reports on the subcellular localization of c-Src (discussed below) implies that these regulatory features are also relevant to the intracellular targeting of the endogenous Src family kinases.

In its inactive state, v-Src colocalizes with regions of dense tubulin staining in the perinuclear region and is diffusely located throughout the cell after treatment with nocodazole (Fig. 3). After activation, v-Src associates with polymerized actin in an SH3 domain-dependent manner and is translocated to focal adhesions at stress fiber termini (Fig. 3). These findings, together with our previous data (17), indicate that v-Src switches from microtubule-dependent retention in the perinuclear region to associate with actin stress fibers. Furthermore, this switch is independent of kinase activity and myristylation-mediated membrane association (17), although the latter is necessary for both perinuclear retention (17) and association with peripheral membranes. These features of v-Src translocation are likely to be conserved also for c-Src, as suggested by immunofluorescence experiments that demonstrated a substantial proportion of exogenously expressed c-Src is concentrated around the nucleus (35). Both exogenous and endogenous c-Src colocalizes with microtubules as well as markers of endosomes and the trans-Golgi and is present at the microtubule organizing center, an area of the cell that is rich in membranous organelles (35). Furthermore, when Tyr-527 of c-Src is not phosphorylated, so that Src is in the open and active conformation, c-Src is translocated to focal adhesions by a mechanism that requires only the amino-terminal portion of c-Src containing the homology domains (36). Thus, activation of the Src protein is accompanied by a switch between subcellular cytoskeletal networks, a switch that is an integral feature of its intracellular targeting.

Our previous work had also established that RhoA-mediated actin organization was essential for translocation of v-Src to focal adhesions (18). RhoA is known to induce MLC phosphorylation via Rho kinase or MLCK and, consequently, to stimulate myosin ATPase activity and actin filament cross-linking into stress fibers that are under high tension (48). We found that the RhoA requirement of v-Src translocation was reflected also in a requirement for myosin activity and the integrity of tensile actin stress fibers. Consistent with this, MLC was visibly associated with stress fibers in CE cells used in these studies (Fig. 3), and the myosin ATPase inhibitor BDM or the MLCK inhibitor ML7 blocked the translocation of kinase-active v-Src (Fig. 4). However, the translocation of kinase-defective v-Src and the integrity of phalloidin-staining stress fibers in cells expressing this mutant protein at the permissive temperature were insensitive to myosin inhibitors (Fig. 4 and 5). These findings indicate that the dominant-negative action of KD v-Src traps the actin cytoskeleton in a stable state that is permissive for translocation, without new assembly of cross-linked actin filaments. These data also imply that the primary role of myosin in v-Src translocation is to maintain the integrity of cellular actin stress fibers.

The finding that KD ts v-Src induces actin stability (Fig. 5) and the formation of enlarged focal adhesions (Fig. 4 and 6) (17) implies that focal adhesion size is determined by turnover that normally occurs during the routine assembly-disassembly cycle of the adhesion-cytoskeletal network. Furthermore, enlarged adhesions arise from continued assembly of focal adhesion protein complexes into preexisting adhesions that are not turning over and that are associated with stabilized actin. Consequently, cells that have enlarged stabilized focal adhesions are overadhered and exhibit impaired cell motility (17). These results further imply that, under these conditions, some focal adhesion components form a complex in the interior part of the cell and can be translocated to existing adhesions at the ends of stable actin stress fibers (Fig. 6). Together with the lack of a requirement for further myosin ATPase function under these conditions, this suggests that it is primarily the integrity of the actin stress fibers tethered into stabilized adhesions that directs the targeting of the focal adhesion protein complexes containing v-Src to these sites.

In keeping with the model for focal adhesion assembly proposed by Burridge and Chrzanowska-Wodnicka, i.e., that integrin clustering during adhesion formation is driven by intracellular activities, including RhoA- and myosin-induced contractility (6, 9), our data imply that intracellular stimuli that trigger the translocation of signaling proteins from the cell interior to focal adhesions also rely on the cytoskeletal organization induced by RhoA and myosin. Interestingly, one further interpretation of our results is that activated Src antagonizes the RhoA/myosin pathway, thus contributing to disassembly of contractile stress fibers during normal cytoskeletal remodeling or oncogenic transformation, and that the KD v-Src interferes with this process. In this way, the dominant-negative v-Src mutant alters the balance in favor of maintenance of contractile actin stress fibers. One potential candidate for Src-induced antagonism of the RhoA/myosin pathway is p190^RhoGAP, which is tyrosine phosphorylated and enzymatically activated in response to v-Src (14, 16).

Although the SH3 domain of v-Src is required for the association of FAK and although FAK contains a known focal adhesion targeting sequence, overexpression of mutant FAK proteins that inhibit the binding of v-Src to FAK does not interfere with translocation (Fig. 7 and 8). In addition, v-Src can localize in the focal adhesions of FAK^−/− cells (Fig. 8), indicating that FAK binding is not critical for v-Src translocation. By an investigation of other potential Src-binding proteins that are known to influence Src functionality or actin integrity, we found that HSP90 and p110^AFAP could bind to the SH3 mutant v-Src protein (Fig. 9). However, the association with the p85 regulatory subunit of PI 3-kinase and v-Src-induced Akt phosphorylation that is downstream of PI 3-kinase were SH3 dependent (Fig. 9), correlating with the SH3 dependence of v-Src's peripheral targeting. To distinguish between cause and effect, we used LY294002, the selective PI 3-kinase inhibitor, to demonstrate that the translocation of v-Src to peripheral focal adhesions required the activity of PI 3-kinase (Fig. 10). Furthermore, we demonstrated that treatment with LY294002 substantially reduced the number of actin stress fibers in treated cells (Fig. 10), indicating that the role of the v-Src SH3-dependent association and activation of PI 3-kinase is most likely to maintain actin stress fibers. The important role that we have identified for the Src SH3 domain and for its binding partner p85 might reflect a more general role for SH3 domains and PI 3-kinase in cytoskeleton-mediated peripheral targeting of other cellular proteins that contain SH3 domains and that associate with actin.

In conclusion, inactive Src is retained in the interior of the cell, most likely via myristylation-dependent endosomal membrane association and the microtubule network. Upon conformational activation, Src associates with polymerized actin and is assembled into focal adhesions at actin stress fiber termini in an SH3- and PI 3-kinase-dependent manner. Use of a KD, a dominant-negative mutant of Src that induces artificial stabilization of the adhesion-cytoskeletal network allowed us to deduce that myosin ATPase also plays an important role by maintaining actin filaments in a state that is permissive for Src translocation. Under these conditions, focal adhesion protein complexes that include v-Src assemble into preexisting focal adhesions. Thus, the intracellular targeting and catalytic activity of ts v-Src, and most likely also c-Src, are tightly and coordinately regulated by simultaneous release of constraints on the SH3 domain and kinase activity. By this means, inadvertent adhesion disruption as a result of inappropriate localization of kinase-active Src to focal adhesions is avoided.