Inhibition of c-Abl tyrosine kinase activity by filamentous actin.

The catalytic activity of c-Abl tyrosine kinase is reduced in fibroblasts that are detached from the extracellular matrix. We report here that a deletion of the extreme C terminus of c-Abl (DeltaF-actin c-Abl) can partially restore kinase activity to c-Abl from detached cells. Because the extreme C terminus of c-Abl contains a consensus F-actin binding motif, we investigated the effect of F-actin on c-Abl tyrosine kinase activity. We found that F-actin can inhibit the kinase activity of purified c-Abl protein. Mutations of the extreme C-terminal region of c-Abl disrupted both the binding of c-Abl to F-actin and the inhibition of c-Abl by F-actin. Mutations of the SH3, SH2, and DNA binding domains did not abolish the inhibition of c-Abl kinase by F-actin. Catalytic domain substitutions that affect the regulation of c-Abl by the retinoblastoma protein or the ataxia telangiectasia-mutated kinase also did not abolish the inhibition of c-Abl by F-actin. Interestingly, among these c-Abl mutants, only the DeltaF-actin c-Abl retained kinase activity in detached cells. Taken together, the data suggest that F-actin is an inhibitor of the c-Abl tyrosine kinase and that this inhibition contributes in part to the reduced Abl kinase activity in detached cells.

The c-Abl tyrosine kinase is ubiquitously expressed in mammalian cells and participates in the transduction of diverse physiological signals. In proliferating cells, c-Abl is found in the cytoplasm and the nucleus, in a dynamic equilibrium between nuclear import and export (1). The nuclear c-Abl tyrosine kinase is cell cycle-regulated. Following the G 1 /S transition, the nuclear c-Abl tyrosine kinase can also be activated by DNA damage (2)(3)(4) and has been implicated in the regulation of apoptosis in response to DNA damage through the activation of the p53-related transcription factor p73 (5)(6)(7). Conversely, the cytoplasmic c-Abl tyrosine kinase is not cell cycle-regulated (8), nor is it activated by DNA damage (4). However, the cytoplasmic c-Abl is regulated by cell adhesion (9). Detachment of fibroblasts from the extracellular matrix (ECM) 1 leads to a reduction in the overall activity of c-Abl. Reattachment to ECM proteins, such as fibronectin, causes a rapid reactivation of c-Abl kinase, and this higher level of c-Abl kinase activity is maintained in attached cells (9).
Like the Src family of tyrosine kinases, c-Abl contains an SH3 and an SH2 Src homology domain, as well as a linker that connects the SH2 domain to the catalytic domain. The SH2 linker of c-Abl contains a proline sequence that can interact intramolecularly with the SH3 domain thereby anchoring the catalytic domain in an inactive conformation (10). Deletion of the c-Abl SH3 domain or point mutations within the SH2 linker results in an increased c-Abl kinase activity in vivo, unmasking the transforming potential of the c-Abl kinase (10 -12). Clearly, the SH3 domain is an important regulatory region in c-Abl, but because the expression of ⌬SH3-Abl requires levels at least 10-fold higher than the endogenous c-Abl to transform (12,13) and only a subset of fibroblasts expressing the activated Abl become transformed (14), SH3-independent mechanisms likely contribute to the oncogenic conversion of c-Abl. Indeed, several SH3-independent mechanisms have been reported to regulate c-Abl (2,8,(15)(16)(17)(18)(19). However, it is unclear how each Abl domain may contribute to c-Abl kinase regulation by the physiological stimuli reported to activate c-Abl in vivo (20).
Immediately C-terminal to the catalytic domain of c-Abl is an extended region not present in the Src family members. This C-terminal region consists of a proline-rich sequence that binds several adapter proteins (17,(21)(22)(23) followed by a DNA binding domain (24) and finally a region that interacts with both Gactin and F-actin (25,26). The F-actin binding function is mapped to the extreme C terminus of c-Abl, which contains a consensus F-actin binding motif found in proteins such as ␣-actinin and fimbrin (26). The C-terminal functional domains have not been linked to the regulation of c-Abl tyrosine kinase activity in previous studies.
We have found that F-actin inhibits c-Abl activity in vitro and that the inhibition requires the C-terminal actin binding region of c-Abl. Consistent with this, the ⌬F-actin c-Abl mutant has increased kinase activity when expressed in cells. Moreover, detached cells retained nearly full Abl kinase activity with the deletion of the F-actin binding region, whereas deletion or substitution of other regions of c-Abl had no effect on the decrease in kinase activity observed in detached cells.

EXPERIMENTAL PROCEDURES
Plasmids-The c-Abl cDNA was murine type IV. The following deletions or substitutions of wild type c-Abl have been previously described: kinsae-defective c-Abl, ⌬SH3, AS1, AS2, S465E, ⌬DNA (2,8,12,(27)(28)(29). Two ⌬F-actin c-Abl DNA constructs were generated by polymerase chain reaction-based mutagenesis using Pfu DNA polymerase (Stratagene). One is deletion of amino acids 1111-1142 and the other is a deletion of amino acids 1139 -1142; both were constructed by inserting STOP codons immediately following residue 1111 or 1139, respectively. Both deletions have been previously shown to disrupt Abl binding to F-actin (25). All constructs used for transfection contained a FLAG epitope tag, with the 8-amino acid FLAG sequence, flanked by three glycine residues on either side, being inserted internally between the DNA and actin binding domains of c-Abl at the unique SalI site. The FLAG tag did not affect adhesion regulation of c-Abl kinase as determined by side by side analysis with either endogenous c-Abl or transfected c-Abl without a FLAG sequence insert (data not shown and Ref. 30). For expression, the various c-Abl DNA constructs were cloned into the MSCV-hph retroviral expression plasmid (31).
Cell Culture and Transfection-The stable Abl-deficient fibroblast cell line AblϪ/Ϫ was a generous gift from Dr. Rubio Ren. AblϪ/Ϫ cells and c-Abl reconstituted AblϪ/Ϫ cells were maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 10% bovine calf serum, L-glutamine, and penicillin-streptomycin. 10T1/2 cells and 293T Bosc cells were maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 15% fetal bovine serum, L-glutamine, and penicillin-streptomycin. Stable cell lines expressing individual c-Abl constructs were obtained through retroviral mediated gene transfer using retrovirus produced from transfected Bosc cells (31). Infected AblϪ/Ϫ cells were selected with hygromycin for 7-12 days, and a polyclonal population of hygromycin-resistant cells was obtained and used for subsequent analysis. Monoclonal populations of cells were selected from the polyclonal pools using cloning cylinders and then subsequently performing single cell dilutions to ensure that each cell line was established from a single cell. The expression levels of c-Abl decreased over time and were comparable to endogenous levels found in NIH 3T3 cells after 8 -12 weeks in culture. Cells used for the adhesion regulation studies contained levels of c-Abl no greater than 4-fold 3T3 endogenous levels.
Fibronectin Stimulation, Immunoprecipitation, and Kinase Assay-Detached cells were prepared by serum starvation (0.1% fetal bovine serum, 24 h) followed by trypsinization and then held in suspension for 40 -60 min in serum-free Dulbecco's modified Eagle's medium containing 500 g/ml soybean trypsin inhibitor and 0.1% BSA (9). Attached cells were prepared by replating the detached cells onto fibronectincoated dishes (50 g/ml fibronectin) for 40 -60 min. Detached cells were collected by centrifugation and attached cells by scraping in ice-cold Triton lysis buffer (1% Triton, 25 mM Hepes, pH 7.4, 150 mM NaCl, 3 mM EGTA, 3 mM EDTA, and 10% glycerol supplemented with protease inhibitors: aprotinin, leupeptin, phenanthroline, benzamidine HCl, and phenylmethylsulfonyl fluoride, and phosphatase inhibitors: sodium orthovanadate, ␤-glycerophosphate). Lysates were sonicated briefly and then centrifuged at 10,000 ϫ g for 7 min at 4°C. Under these conditions, ϳ85% of c-Abl and 86% of actin were recovered in the supernatant of lysates from both attached and detached cells. Much higher g forces (100,000 ϫ g) were required to pellet the F-actin, which was not crosslinked in networks (see "F-actin Polymerization and F-actin Binding Assay"). Immunoprecipitations were carried out for 2-4 h at 4°C using 500 g of cell lysate and antibody coupled to either protein A-Sepharose (anti-K12, Santa Cruz) or protein G-Sepharose (anti-8E9 (Pharmingen), anti-Ab-3 (Oncogene Sciences), and anti-FLAG (Sigma) monoclonal antibodies). Other monoclonal antibodies used for Western blotting included anti-pTyr 4G10 (Upstate Biotechnology) and anti-actin (Sigma).
For kinase assays, immunoprecipitations were washed three times with lysis buffer and two times with kinase buffer. The final Sepharose pellet was resuspended in buffer containing 20 mM Hepes, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol, and 1-4 g of substrate. The addition of 20 M ATP and 50 Ci of [␥-32 P]ATP initiated the kinase reactions. After 30 -40 min of shaking at room temperature, reactions were stopped by the addition of boiling SDS sample buffer. SDS-polyacrylamide gel electrophoresis was performed (4 -15% gradient or 7.5 or 10% polyacrylamide gels), and then proteins were transferred onto polyvinylidene difluoride membranes. Membranes were stained with Amido Black to reveal substrate position and then exposed to film to reveal radioactivity incorporated into substrate. The quantity of 32 P incorporation into substrate was then determined by PhosphorImager analysis using kinase-defective c-Abl as a negative control. The same membranes were reactivated with methanol, and Western analysis with the 8E9, or Ab-3 c-Abl antibody was performed to determine the quantity of c-Abl in each kinase reaction. (Although K12 and FLAG antibodies recognize all of our Abl proteins in the immunoprecipitates, they did not recognize Abl on immunoblots.) Since the epitope for 8E9 is not present in AS1 Abl and the epitope for Ab-3 is not present in ⌬F-actin protein we first did Western analysis with the 8E9 antibody then stripped the blots and subsequently used Ab-3 as the detecting antibody. To calculate the amount of Abl protein per lane, we used the kinase-defective and wild type c-Abl protein levels as internal controls to set the ECL signal as equal between Westerns performed with different antibodies. c-Abl concentration was quantified using a charge-coupled device camera and software provided by Alpha Inotech. The specific activity of c-Abl was calculated by dividing the substrate 32 P incorporation (Phos-phorImager) by the amount of c-Abl protein in each immunoprecipitation (c-Abl Western). To be sure the signal we detected in our kinase reactions was due to phosphotyrosine, membranes were reactivated with methanol then soaked in 1 M KOH for 2 h at 55°C to reduce any background serine phosphorylation. Membranes were then re-exposed and compared with the original exposure. Using this method, we confirmed that the signal was exclusively due to phosphotyrosine. For P81 experiments the kinase reaction was performed under the same conditions with 4 l of the purified c-Abl in place of the immunoprecipitated c-Abl. To stabilize low concentrations of purified protein, we included 5-10 M BSA in all kinase reactions containing purified c-Abl. (BSA was not phosphorylated by c-Abl.) Reactions were terminated by pipetting onto P81 cation exchange paper and immediately dropping into 50 mM NaCl solution. P81 paper was washed for 15 min in 0.1% phosphoric acid with two changes of buffer and then rinsed in 50 mM NaCl. 32 P incorporation was determined by scintillation counting. For purified c-Abl kinase reactions it was determined that P81 analysis gave identical results as the SDS-polyacrylamide gel electrophoresis/Western analysis described above.
Protein Purification-FLAG-c-Abl was purified using the anti-M2FLAG resin (Sigma). 100 million cells were lysed in 20 ml of high salt Triton lysis buffer (750 mM NaCl) and briefly sonicated on ice. Lysate was rotated for 30 min at 4°C, and detergent-insoluble proteins were removed by high speed ultracentrifugation (100,000 ϫ g, 4°C, 15 min). Cleared supernatant was rotated for 2-3 h with 1 ml of anti-FLAG resin. Resin was washed with 30 ml of high salt lysis buffer then 5 ml of TBS, pH 8. Bound proteins were eluted with 200 g/ml FLAG peptide in TBS containing 20% glycerol and 1 mM dithiothreitol. Peak fractions were aliquoted and stored at Ϫ80°C. The FLAG peptide did not affect the binding of Abl to F-actin or the inhibition of Abl by F-actin as determined by dialyzing the FLAG peptide from purified Abl. To stabilize Abl at low concentrations, we added the 1 M BSA directly to the Abl preparation. The BSA buffer did not affect purified c-Abl activity after dilution into kinase reactions.
GST-tagged proteins were purified as previously described: GST-CTD (CTD, last 384 amino acids of murine C-terminal repeat domain of RNA polymerase II), GST-CrkCTD (Crk-CTD, 218 -224 of Crk fused to the last 31 amino acids of CTD) (32,33). Purified proteins were dialyzed against phosphate-buffered saline, 20% glycerol overnight and then aliquoted and stored at Ϫ80°C. F-actin was purified to homogeneity from rabbit skeletal muscle by Drs. Harry Higgs and Laurent Blanchoin (Dr. Tom Pollard's laboratory). Fibronectin was purified by Dr. Mark Renshaw as previously described (9).
F-actin Polymerization and F-actin Binding Assay-F-actin polymerization was carried out according to Pollard laboratory recommendations (34,35). Purified actin was polymerized at 25 M at 25°C for 1 h. The final buffer (G buffer with 1ϫ KMEI) containing the F-actin consisted of: 0.2 mM ATP, 0.1 mM CaCl 2 , and 0.5 mM dithiothreitol, 50 mM KCl, 1 mM MgCl 2 , 1 mM EGTA, and 10 mM Tris (pH 7.5). Buffers containing equivalent amounts of BSA were used as controls. To prevent quenching of the kinase reactions with excess cold ATP from the polymerization buffer, the polymerized F-actin and control BSA buffer were passed over a Dowex AG column, previously equilibrated to pH 7.5. (We also performed kinase assays by adding equivalent ATP to the control BSA buffer instead of passing over the Dowex AG resin. This method required longer exposure of autoradiograms but gave similar results.) Typically, the F-actin was diluted 10-fold or more into the kinase reactions or binding reactions to give a final concentration of 50 nM to 3 M. For reactions containing less than 0.3 M actin, phalloidin (Sigma) was added to the polymerized F-actin at a concentration equal to the actin before dilution. Phalloidin did not affect the binding of Abl to F-actin.
For F-actin sedimentation assays, we used 1ϫ polymerization buffer containing 10 M BSA as the binding buffer. Abl was incubated with F-actin for 15 min at room temperature. Centrifugation was performed in a Beckman TLA-100 rotor for 20 min at 100,000 ϫ g to pellet the F-actin. Because we observed some nonspecific binding of c-Abl to the polycarbonate centrifuge tubes, we incubated all tubes with the BSA buffer overnight. Pellets were washed in 1ϫ KMEI before solubilization and analysis by SDS-polyacrylamide gel electrophoresis.

RESULTS
The Inactivation of c-Abl Kinase by Cell Detachment Requires the F-actin Binding Domain-We have previously demonstrated that c-Abl kinase activity is reduced when fibroblasts are detached from the ECM (9). To investigate the regulation of c-Abl tyrosine kinase by cell adhesion, we examined a panel of c-Abl mutants for their activity in attached and detached fibroblasts. Each c-Abl protein was expressed in Abl-deficient (AblϪ/Ϫ) 3T3 cells established from AblϪ/Ϫ mice using retroviral mediated gene transfer. This gave a level of c-Abl expression approximately equal to that of endogenous c-Abl after selection (see "Experimental Procedures"). c-Abl was immunoprecipitated with the K12 antibody (directed against amino acids 521-531, Santa Cruz Biotechnology, Inc.) from either attached or detached cells to determine kinase activity. The kinase activity of wild type c-Abl when reconstituted into AblϪ/Ϫ background was regulated by the adherent status of these cells (Fig. 1). Using GST fusions of the C-terminal repeat domain of RNA polymerase II (CTD) or Crk-CTD as a specific substrate (5,32,36), we observed that the activity of c-Abl isolated from attached cells was 4 -5-fold higher than that from detached cells. The observed kinase activity toward CTD or Crk-CTD was specific to c-Abl, because a KD c-Abl mutant showed only background activity that was the same in attached or detached cells. The 4 -5-fold reduction in c-Abl activity caused by cell detachment is similar to the regulation we have previously reported for the endogenous Abl found in fibroblast cells (9) and is also consistent with a previous report in which the anti-FLAG antibody was used to immunoprecipitate FLAGtagged c-Abl and paxillin was utilized as the substrate for c-Abl kinase (30). The presence of the inserted FLAG tag within these constructs did not affect the activity or regulation by adhesion, consistent with these previous findings (30). The activity of each construct in attached and detached cells was consistent (relative to KD c-Abl) regardless of whether we used polyclonal or monoclonal populations of cells selected for Abl expression. These results demonstrated that the adhesion regulation of c-Abl activity could be reconstituted in AblϪ/Ϫ cells and thus allowed a series of c-Abl mutants to be tested in this system.
As summarized in Fig. 1, we examined three deletion mutants (⌬SH3, ⌬DNA, and ⌬F-actin) and four substitution mutants (KD, S465E, AS1, and AS2) for their kinase activity in attached and detached cells. The c-Abl kinase activity for the kinase-defective c-Abl immunoprecipitated from detached cells was independently set at 1, and the values summarized in Fig.  1 are the mean Ϯ S.D. from at least three independent experiments. The activity of five of the seven c-Abl mutants, ⌬SH3, AS1, AS2, S465E, and ⌬DNA, was regulated by cell adhesion, with kinase activity falling 4 -5-fold upon cell detachment. The activity of the ⌬SH3 mutant was assayed in polyclonal cells that were not transformed. Stable or transient expression of the ⌬SH3 mutant in the AblϪ/Ϫ cells did not result in transformation because the level of expression was not high enough (12,13) and because the AblϪ/Ϫ cells are predominantly "N" 3T3 cells, which cannot be transformed by Abl (14,37). In these cells ⌬SH3 c-Abl had kinase activity comparable to that of wild type c-Abl, and this activity was reduced upon cell detachment. The AS1 c-Abl mutant is a substitution of the SH2 domain with the analogous SH2 region in c-Src. The AS2 c-Abl mutant is a substitution of the c-Abl ATP binding lobe with the analogous region in c-Src, which thereby prevents inhibition by the retinoblastoma (RB) protein (15). Both of these substitutions resulted in a c-Abl that showed reduced activity upon cell detachment similar to wild type c-Abl. The substitution of Ser 465 , the phosphorylation site for ataxia telangiectasia mutated kinase, to either glutamic acid (S465E) or alanine (S465A), which abolishes regulation by DNA damage (2, 5), also did not affect the regulation by adhesion. Likewise, deletion of two of the three high mobility group-like boxes that comprise the DNA binding domain (24) did not affect the regulation of c-Abl activity by cell adhesion. Thus, the adhesion regulation of c-Abl kinase occurs through a novel mechanism that is distinct from the previously identified regulation of c-Abl kinase through its SH3 domain (11,12), through RB binding (8,15), or through the ataxia telangiectasia mutated kinase-dependent phosphorylation of Ser 465 (2).
The c-Abl mutant defective in F-actin binding (⌬F-actin) was significantly different from the wild type c-Abl in its response to cell detachment. Deletion of the 32 C-terminal amino acids, which contain an F-actin binding consensus sequence (25,26), had two apparent effects on c-Abl. First, the kinase activity of ⌬F-actin c-Abl was 2.7-fold higher in attached cells and 6.5-fold c-Abl Purified from Attached or Detached Cells Has Similar Kinase Activity-Because the F-actin binding domain of c-Abl has been reported to associate with actin, we investigated whether the association of c-Abl with F-actin could modify c-Abl activity. We used anti-FLAG immune affinity chromatography to purify FLAG-tagged c-Abl from attached or detached cells (Fig. 2). Stringent lysis conditions and high salt washing were used to purify c-Abl free from most, if not all, interacting proteins as determined by silver staining (Fig. 2A). In contrast to immunoprecipitated c-Abl from the same cell lines, the purified c-Abl eluted from the FLAG resin phosphorylated CTD to a high stoichiometry generating the CTD o form that is phosphorylated to a stoichiometry of 40 mol of phosphotyrosine per mol of CTD (29,32). The kinase activity of purified c-Abl was linear with time beyond 60 min determined by using scintillation counting to quantify kinase activity (Fig. 2C). Significantly, c-Abl purified from attached or detached cells had the same c-Abl kinase activity (Fig. 2, B, compare odd-versus  even-numbered lanes, and C). Additionally, we found that soluble purified c-Abl had higher kinase activity than immunoprecipitated (partially purified) c-Abl (Fig. 2D). This is consistent with the notion that c-Abl purification eliminates the difference in activity in attached versus detached cells.
F-actin Inhibits c-Abl Kinase-To test whether actin could regulate c-Abl tyrosine kinase activity, we added purified Factin (rabbit skeletal muscle) to our purified c-Abl preparations and measured the resulting kinase activity. The addition of F-actin caused a dose-dependent inhibition of c-Abl kinase (Fig.  3, A and B). We observed this for both substrate phosphorylation (Fig. 3) and phosphorylation of the c-Abl protein itself (Fig.   4). Maximal inhibition of the c-Abl kinase was observed at 2.5 M F-actin (Fig. 3B), whereas addition of BSA to concentrations as high as 30 M did not inhibit the c-Abl kinase activity (Fig.  3A, lane 1). F-actin inhibited c-Abl from either attached cells or detached cells to a similar extent when purified under stringent conditions (data not shown). The activity of c-Abl was decreased 8-fold by 3 M F-actin, and the K i of F-actin for inhibition of c-Abl was 0.5 M (Fig. 3B). We obtained similar results by terminating kinase reactions by blotting onto P81 paper (see "Experimental Procedures"). The kinase activity of our purified Abl was linear beyond 60 min, and we observed inhibition of activity by F-actin at various time points within this linear range (Fig. 3C).
We were concerned that the inhibition by F-actin might be the result of F-actin competing with the substrate for binding to Abl so we performed two additional experiments. First, we examined the effect of F-actin on c-Abl autophosphorylation. Autophosphorylation was also inhibited by F-actin (Fig. 4). Second, we examined whether the inhibition by F-actin was dependent on substrate concentration, which would indicate competitive inhibition. We varied the CTD concentration in kinase reactions while holding the F-actin (or BSA) concentration at 3 M (Fig. 5, A and B). The purified c-Abl had a K m ϭ 0.25 M toward CTD. This value is consistent with the previously determined K m using c-Abl immunoprecipitated from fibroblasts (29,32). In the presence of F-actin the K m toward CTD was not significantly increased. However, the kinase activity toward CTD was inhibited 3-fold or greater over the range of 0.3 to 9 M CTD concentrations (Fig. 5). These data indicate that the catalytic activity of c-Abl is compromised significantly by F-actin regardless of the substrate concentration.
F-actin Does Not Inhibit ⌬F-actin c-Abl-To explore the specificity of the F-actin inhibition, we purified KD, wild type, AS1, AS2, ⌬DNA, ⌬SH3, and ⌬F-actin c-Abl mutants (Fig. 1) and assayed each purified protein for its sensitivity to F-actin. We calculated the specific activity of each purified Abl protein by normalizing the 32 P incorporation to the level of Abl protein present in each kinase reaction and then set the value for the KD c-Abl as 1 (Fig. 6A). The 32 P incorporation was determined using a PhosphorImager, and the concentration of Abl protein in each reaction was determined by Western analysis using anti-Abl antibodies as described under "Experimental Procedures." The purified wild type, AS2, AS1, ⌬DNA, and ⌬F-actin  5 ml each). B, the tyrosine kinase activity of purified c-Abl was determined for c-Abl isolated from either adherent or nonadherent cells using GST-CTD as a substrate. The GST-CTD was phosphorylated to high stoichiometry resulting in an apparent mobility shift from the CTD a to CTD o as previously described (32). proteins showed similar specific activities in the absence of F-actin. Only the purified ⌬SH3 c-Abl showed an increased catalytic activity. The activity of the wild type, AS2, AS1, and ⌬DNA c-Abl was reduced by F-actin to a similar extent. The ⌬SH3 c-Abl was also inhibited by F-actin, although the extent of inhibition was less than that of the wild type c-Abl. The ⌬F-actin c-Abl, by contrast, was not inhibited by F-actin (Fig.  6A, lanes 13 and 14).
Previous experiments have demonstrated that the F-actin binding domain from c-Abl co-sediments with F-actin using high speed centrifugation (25,26). We were able to repeat those findings with our purified full-length c-Abl proteins (Fig. 6B). Purified c-Abl was incubated with increasing concentrations of purified F-actin, and then each binding reaction was then centrifuged to pellet the F-actin and the c-Abl associated with the F-actin. Binding reactions were analyzed for c-Abl and F-actin protein by Western blot or Amido Black staining of the same blot. We found that increasing amounts of c-Abl (input ϳ50 nmol) co-sedimented with increasing concentrations of F-actin (Fig. 6B, top panel). In the absence of F-actin the majority (80%) of the c-Abl was recovered in the supernatant, whereas in the presence of 2 M F-actin the majority (95%) of the c-Abl was recovered in the F-actin-containing pellet (data not shown). As a control, the ⌬F-actin c-Abl did not co-sediment with F-actin (Fig. 6B, middle panel). Unfortunately, nonspecific binding of low concentrations of the full-length Abl to centrifuge tubes prevented an accurate analysis of the stoichiometry of binding. Therefore, to investigate the stoichiometry of c-Abl binding to F-actin, we purified GST-(actin binding domain) using glutathione-Sepharose. The Abl portion of the GST protein was cleaved off of the Sepharose with thrombin and tested for binding to F-actin in this sedimentation assay. Similar to the full-length protein, the F-actin binding domain co-sedimented with F-actin (Fig. 6B, lower panels). Using these conditions, we observed that increasing amounts of c-Abl could be co-sedimented with increasing concentrations of F-actin. At the K i for actin-induced inhibition of c-Abl (0.5 M F-actin), we observed near saturation binding of c-Abl to F-actin (Fig. 6C). An approximate 2:1 molar ratio of F-actin to c-Abl binding was observed when 50% of the c-Abl was bound (0.25 nmol of c-Abl bound 0.45 nmol of F-actin). We could achieve a maximum of 95% binding at 1 M F-actin. The stronger binding affinity we observed relative to previous investigations may be attributable to their use of purified proteins containing bulky GST tags (25,26). Taken together, these results indicate that F-actin binds directly to c-Abl to inhibit its kinase activity. Since ⌬Factin c-Abl fails to bind to F-actin, its activity is not affected by F-actin.

DISCUSSION
Multiple Levels of Regulation of c-Abl Kinase-We have identified a new mechanism for c-Abl kinase regulation distinct from those described previously. The evidence that the F-actin inhibition of c-Abl kinase activity is a distinct mechanism of regulation is as follows. First, the nuclear c-Abl activity is regulated during cell cycle progression. Activity of c-Abl is increased during S due to the release of RB from c-Abl (15). Because c-Abl AS2, which is not inhibited by RB (8), remains sensitive to F-actin, the conformational regulation mediated by F-actin is not dependent on RB. Second, DNA damage can activate the nuclear c-Abl through the ataxia telangiectasia mutated kinase-dependent phosphorylation of a serine residue (Ser 465 ) within the c-Abl kinase domain (2). In vivo phosphotryptic peptide mapping data and analysis of S465A and S465E c-Abl mutants 2 argue against the involvement of phosphorylation in the adhesion-dependent regulation of c-Abl. Third, PDGF has been shown to activate c-Abl through Src-dependent tyrosine phosphorylation of Tyr 412 within the kinase domain of c-Abl (16). However, we found that in the c-Abl-reconstituted AblϪ/Ϫ fibroblasts, c-Abl was not tyrosine-phosphorylated (determined by both phosphoamino acid analysis of immunoprecipitated Abl and anti-phosphotyrosine Western blotting of immunoprecipitated Abl). Further, mutation of Tyr 412 to phenylalanine resulted in an Abl protein that could still be further inactivated by cell detachment (data not shown). Finally, the SH3 domain has been proposed to play an inhibitory role in c-Abl kinase regulation through both intra-and intermolecular interactions (10,20,38). However, we found that deletion of the SH3 domain did not affect the adhesion-dependent regulation of c-Abl as long as the cells remain untrans- formed. Thus, the F-actin-mediated regulation of c-Abl appears to be an additional mechanism by which c-Abl kinase activity can be modulated.
We find it interesting that the binding of F-actin to the extreme C terminus of Abl can affect the catalytic function of Abl since these two domains are separated by over 500 amino acids linearly. This suggests that c-Abl may be folded into an inactive conformation by F-actin through binding simultaneously to the C terminus and to the active site of c-Abl. Alternatively, F-actin binding to the C terminus may simply impose a change in the conformation of the c-Abl active site. This would not be unprecedented as such intramolecular interactions have been described for the Src family of tyrosine kinases. For example, the intramolecular interactions between the C-terminal tyrosine-phosphorylated tail and the SH2 domain of Src and also the intramolecular interaction between the SH2 linker and the SH3 domain of Src anchor the catalytic domain of Src in an inactive conformation. This is supported not only by the crystal structure of Src (39,40) but also by several biochemical analyses (for review, see Refs. 41 and 42). Although the SH2-linker of c-Abl has been implicated to participate in intramolecular interactions similar to that of the Src SH2-linker (10), there is not the equivalent of a tyrosine-phosphorylated tail in c-Abl, suggesting that another mechanism may participate in folding the Abl kinase into an inactive conformation. In this regard, it is noteworthy that deletion (but not substitution) of the SH2 domain of c-Abl results in loss of sensitivity to inhibition by F-actin. 3 We are currently investigating how the SH2 domain might be involved in the F-actinmediated inhibition of c-Abl kinase.
Biological Implications-The concentration of F-actin in cells is high relative to other cellular proteins. It has been estimated that the concentration of F-actin in non-muscle cells is on the order of 100 M, and the intracellular F-actin content is only decreased by ϳ50% upon cell detachment (43). 4 Since the K i for F-actin inhibition of Abl is 0.5 M, there is sufficient F-actin to inhibit the c-Abl kinase in both attached and detached cells. We have found an interesting correlation between the inhibition by F-actin and the reduction in Abl kinase activity in detached cells. The only c-Abl mutant that has a high catalytic activity in detached cells is the ⌬F-actin c-Abl (Fig. 1), which is also the only c-Abl mutant we have found to be insensitive to F-actin-mediated inhibition (Fig. 6). This correlation suggests that F-actin-mediated inhibition is required to reduce c-Abl activity in detached cells. However, given the relatively high cellular concentration of F-actin, its effect on the c-Abl activity cannot be through a simple concentration-dependent mass action. It is possible that the binding of c-Abl to F-actin is actively suppressed in attached cells despite the higher overall F-actin concentration. This could be through the sequestration of c-Abl away from F-actin or a direct inhibition of the F-actin binding activity of c-Abl. Because c-Abl purified from attached cells can be inhibited by F-actin in vitro, stable modifications of c-Abl are not likely to be a mechanism for increased c-Abl

FIG. 6. The F-actin binding domain of c-Abl is required for inhibition by F-actin.
A, the indicated c-Abl constructs were purified using stringent conditions, and each was assayed for activity in the absence or presence of 3 M F-actin. Specific activity of each construct was normalized to protein content (lower panel) as described under "Experimental Procedures" and is indicated below the autoradiogram (top panel). Because the 8E9 antibody does not recognize AS1 Abl and Ab-3 antibody does not recognize ⌬F-actin Abl, we used Ab-3 antibody to detect Abl proteins in lanes 1-12. After stripping, Ab-3 was used to detect Abl protein (lanes 13 and 14). ECL signals were normalized between the two blotting antibodies to quantify protein content for each Abl (see "Experimental Procedures"). The ⌬DNA Abl (222-amino acid deletion, 110 kDa), ⌬SH3 Abl (50-amino acid deletion, 140 kDa), and ⌬F-actin (32-amino acid deletion, 140 kDa) had increased mobility relative to other Abl constructs (145 kDa). B, increasing quantities of F-actin (lanes 1-6, 50 nM, 100 nM, 200 nM, 500 nM, 1 M, and 2 M, respectively) were incubated with purified c-Abl, purified ⌬F-actin c-Abl, or the F-actin binding domain (F-Actin BD) of c-Abl. The F-actin was sedimented, and the quantity of c-Abl that sedimented with the F-actin was determined by anti-Abl immunoblotting. The amount of Abl included in each binding reaction is shown in lane 7. C, the binding of the F-actin domain of c-Abl to F-actin is graphed as a function of F-actin concentration. WB, Western blot. kinase activity in attached cells. Alternatively, the binding of c-Abl to F-actin may not be actively suppressed in attached cells, but the inhibitory effect of F-actin on c-Abl kinase activity might be neutralized in attached cells.
We have found that ⌬F-actin c-Abl has higher activity than wild type c-Abl in the immune complex kinase assay. However, in the solution kinase assay ⌬F-actin c-Abl and wild type c-Abl have similar activity. The comparable catalytic activity observed with purified wild type and ⌬F-actin c-Abl indicates that the differences observed in the immune complex kinase assay might be due to other co-immunoprecipitated proteins. Consistently, wild type c-Abl purified from detached or attached cells had similar kinase activity, whereas wild type c-Abl immunoprecipitated from detached or attached cells had differing kinase activity (Fig. 2). In detached cells, the reduction of c-Abl kinase activity requires the F-actin binding function, as demonstrated by the results presented in this report. However, even with the ⌬F-actin c-Abl, adhesion can cause a further increase in kinase activity (Fig. 1). These data suggest that otherfactorsbesidesF-actinarelikelytoaccountfortheadhesiondependent differences in c-Abl activity. An important goal of our future research will be to determine the other factors that modulate c-Abl activity upon cell adhesion.
Deletion of only 32 amino acids at the C terminus of c-Abl can lead to loss of regulation of c-Abl coincident with loss of F-actin binding. We have also found that the deletion of the last 4 amino acids of c-Abl produced similar effects. This mutant also does not bind to F-actin, is not inhibited by F-actin, and remains active in detached cells. Previous reports have indicated that deletion of the C-terminal half of c-Abl may disrupt c-Abl regulation (17,44). However, the specific mechanism was unclear. It will be interesting to investigate whether these previous observations are linked to F-actin inhibition of c-Abl. We have recently obtained preliminary data consistent with our model that F-actin can associate with and inhibit c-Abl in vivo. Latrunculin A, which disrupts the F-actin cytoskeleton, increases c-Abl activity in detached cells. We have also found that that there is more actin associated with c-Abl in detached cells than in attached cells and less F-actin associated with Abl in latrunculin-treated cells. 2 In addition, Dr. Tony Koleske's laboratory has also found that the Abl-related gene kinase Arg can also bind to and be inhibited by F-actin. 5 This suggests that F-actin inhibition of the Abl tyrosine kinase may be a general feature of the Abl family consistent with the proposed role for Abl and Arg in processes involving the actin cytoskeleton (45,46).
The c-Abl tyrosine kinase has been implicated in the regulation of the F-actin cytoskeleton. The cytoplasmic c-Abl is involved in physiological responses that entail the organization and reorganization of the actin cytoskeleton. For example, c-Abl likely plays a significant role in neurulation, axon path finding, and cell migration (46). A previous study has suggested that the C-terminal region of c-Abl, when isolated, can affect the F-actin polymerization rate and bundling in vitro (25). Our results demonstrating that F-actin can inhibit c-Abl kinase activity suggest a dynamic interplay between c-Abl and Factin. This interplay may be essential to the role of c-Abl in the regulation of F-actin cytoskeleton in response to adhesion and other physiological signals.