Functional Interaction between the c-Abl and Arg Protein-tyrosine Kinases in the Oxidative Stress Response*

The Abl family of mammalian nonreceptor tyrosine kinases consists of c-Abl and Arg. Recent work has shown that c-Abl and Arg are activated in the cellular response to oxidative stress. The present studies demonstrate that reactive oxygen species (ROS) induce the formation of c-Abl and Arg heterodimers. The results show that the c-Abl SH3 domain binds directly to a pro-line-rich site (amino acids 567–576) in the Arg C-termi-nal region. Formation of c-Abl (cid:1) Arg heterodimers also involves direct binding of the Arg Src homology 3 domain to the C-terminal region of c-Abl. The results further demonstrate that the interaction between c-Abl and Arg involves c-Abl-mediated phosphorylation of Arg. The functional significance of the c-Abl-Arg interaction is supported by the demonstration that both c-Abl and Arg are required for ROS-induced apoptosis. These findings indicate that ROS induce c-Abl (cid:1) Arg heterodimers and that both c-Abl and Arg are necessary as effectors in the apoptotic response to oxidative stress. The mammalian c-Abl and Arg nonreceptor tyrosine kinases are expressed widely in adult tissues (1–3). The N-terminal regions of c-Abl and Arg share (cid:1) 90% identity and, as found in members

The mammalian c-Abl and Arg nonreceptor tyrosine kinases are expressed widely in adult tissues (1)(2)(3). The N-terminal regions of c-Abl and Arg share ϳ90% identity and, as found in members of the Src family, contain tandem Src homology 3 (SH3), 1 SH2, and tyrosine kinase (SH1) domains. Following the kinase domain, the next 135 amino acids of both proteins contain three conserved PXXP motifs that can serve as binding sites for SH3 domains (4,5). The C-terminal regions of c-Abl and Arg share 29% identity and differ from other nonreceptor tyrosine kinases by the presence of globular and filamentous actin binding domains (6). In addition, the C-terminal region of c-Abl differs from that of Arg by the presence of a nuclear localization signal (7) and DNA binding sequences (8). In concert with these structural differences, c-Abl is expressed in both the nucleus and cytoplasm, whereas Arg has been found predominantly in the cytoplasm (5,9).
The available evidence supports a role for c-Abl and Arg in regulating cytoskeletal dynamics. Mammalian c-abl and arg exhibit structural conservation with genes in the sea urchin (E-abl), fruit fly (D-abl), and nematode (N-abl) (10,11). D-abl is expressed in neuronal axons (12) and functions in control of the axonal cytoskeleton (13). Other studies have demonstrated that D-Abl interacts with the Notch transmembrane receptor to regulate axon extension (14). Mice with targeted disruption of the c-abl gene are born runted and exhibit head and eye abnormalities (14). Mice deficient in Arg develop normally and exhibit behavioral abnormalities (9). Moreover, embryos deficient in both c-Abl and Arg die before 11 days postcoitus with defects in neurulation (9). The finding that neuroepithelial cells from c-abl Ϫ/Ϫ arg Ϫ/Ϫ mice have an altered actin cytoskeleton has supported involvement of c-Abl and Arg in the regulation of actin microfilaments (9). Further support for interactions between c-Abl and the actin cytoskeleton has been obtained from the demonstration that clustering of integrins and thereby docking of actin stress fibers is associated with stimulation of c-Abl activity (15).
Other insights into a functional role for c-Abl have been derived from the findings that overexpression of c-Abl in fibroblasts induces cell cycle arrest (16,17). Growth suppression is dependent on the nuclear localization sequences, an intact SH2 domain, and tyrosine kinase activity (17). Expression of c-Abl in Schizosaccharomyces pombe similarly induces growth arrest by a mechanism dependent on the c-Abl kinase function (18). In mammalian cells, c-Abl-dependent growth arrest is mediated in part by interactions with p53 and the induction of p21 (19,20). Nuclear c-Abl also associates with the DNA-dependent protein kinase complex (21,22) and with the product of the gene mutated in ataxia telangiectasia (23,24). Activation of these serine/threonine kinases in the response of cells to DNA damage is associated with induction of c-Abl activity (21,(23)(24)(25)(26). Activation of c-Abl contributes to DNA damage-induced apoptosis by mechanisms in part dependent on p53 and its homolog p73 (27)(28)(29)(30)(31). In contrast to the involvement of nuclear c-Abl in DNA damage-induced signaling, there is no known role for Arg in the cellular response to genotoxic stress.
Recent work has shown that the cytoplasmic forms of c-Abl and Arg are activated in the response of cells to oxidative stress. Normal cellular metabolism is associated with the production of reactive oxygen species (ROS) and, as a consequence, damage to DNA and proteins (32,33). Cytoplasmic c-Abl is activated in response to ROS production by a mechanism that depends on interactions with protein kinase C␦ (34 -36). Activation of cytoplasmic c-Abl by ROS is associated with targeting of c-Abl to mitochondria, release of cytochrome c, and induction of cell death (36,37). Other studies have shown that Arg is activated by oxidative stress and that this response involves Arg-mediated phosphorylation of the proapoptotic Siva-1 protein (38). The finding that ROS-induced apoptosis is attenuated in arg Ϫ/Ϫ cells has supported a role for Arg in the cell death response to oxidative stress (38). Thus, the available evidence indicates that cytoplasmic c-Abl and Arg are both functional in the oxidative stress response.
The present studies demonstrate that c-Abl forms heterodimers with Arg in the response to oxidative stress. The functional significance of these findings is supported by the demonstration that both c-Abl and Arg are required for ROSinduced apoptosis.
Vectors-FLAG-tagged c-Abl, Arg, and their truncated variants were expressed by cloning into the pcDNA3.1-based FLAG vector. His/Exptagged constructs were prepared by cloning into pcDNA4HisMAX (Invitrogen). Myc-tagged constructs were prepared by cloning into the pCMV-Myc vector (Clontech). Retroviruses expressing FLAG-c-Abl(K-R) or FLAG-Arg(K-R) were prepared by cloning into a MSCVbased retroviral vector (MSCVpuro; Clontech).
Binding Assays-Glutathione S-transferase (GST) fusion proteins were prepared as described (28,38). Cell lysates were incubated with 5 g of GST or GST fusion proteins conjugated to Sepharose 4B beads for 2 h at 4°C. The adsorbates were washed with lysis buffer and then subjected to immunoblotting with anti-FLAG. An aliquot of the total lysate (2% v/v) was included as a control. For direct binding assays, purified GST fusion proteins were incubated with in vitro translated 35 S-labeled proteins. The adsorbates were analyzed by SDS-PAGE and autoradiography.
Surface Plasmon Resonance-Recombinant GST-c-Abl was coupled to a carboxymethyl dextran sensor chip CM5 in the presence of EDC, ethanolamine HCl, and N-hydroxysuccinimide. Binding assays were performed in a Biocore 1000 (BIAcore AB, Uppsala, Sweden). Purified recombinant Arg (9) diluted in HBS buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM EDTA and 0.005% surfactant P20) was applied to the chip at a flow rate of 10 l/min at 35°C. BIAevaluation software 3.0 and the 1:1 Langmuir binding model were used to assess binding kinetics.
Kinase Assays-Lysates from c-abl Ϫ/Ϫ cells expressing FLAG-Arg were subjected to immunoprecipitation with anti-FLAG. The protein complexes were washed, normalized by immunoblotting, heat inactivated at 80°C for 20 min, and then resuspended in kinase buffer (20 mM HEPES, pH 7.5, 75 mM KCl, 10 mM MgCl 2 , 10 mM MnCl 2 ) containing 2.5 Ci of [␥-32 P]ATP and kinase-active Abl (New England Biolabs) for 30 min at 30°C. The reaction products were analyzed by SDS-PAGE and autoradiography.
Apoptosis Assays-The DNA content was assessed by staining ethanol-fixed and citrate buffer-permeabilized cells with propidium iodide and monitoring by a fluorescence-activated cell sorter (BD Biosciences). The numbers of cells with sub-G 1 DNA were determined with a MOD-FIT LT program.

ROS Induce c-Abl⅐Arg Heterodimers-
To determine whether c-Abl associates with Arg in vivo, lysates from MCF-7 cells were subjected to immunoprecipitation with anti-c-Abl or, as a control, mouse IgG. Immunoblot analysis of the precipitates with anti-Arg demonstrated a low level of c-Abl⅐Arg complexes (Fig. 1A). Significantly, treatment of the cells with 40 and 160 M H 2 O 2 was associated with detectable increases in c-Abl⅐Arg complexes (Fig. 1A). Exposure to 640 M H 2 O 2 , however, re-sulted in an association between c-Abl and Arg which was comparable with that found in control cells (Fig. 1A). Based on the total amount of Arg in lysates subjected to immunoprecipitation, less than 1% of the Arg protein was complexed with c-Abl in control cells. Treatment with 40 M H 2 O 2 increased the formation of c-Abl⅐Arg complexes 3.2-fold such that ϳ3% of the Arg protein was complexed with c-Abl. Although total cellular levels of Arg and c-Abl were similar in these cells, the fraction of c-Abl complexed to Arg was also ϳ3% in response to H 2 O 2 . Similar findings were obtained in MEFs treated with H 2 O 2 (Fig. 1B). The association between c-Abl and Arg was increased after exposure to 10, 40, and 160 M H 2 O 2 , whereas treatment with 640 M H 2 O 2 had little effect (Fig. 1B). To determine whether c-Abl and Arg interact in response to other inducers of oxidative stress, wild-type MEFs were treated with menadione, a redox-cycling agent that increases ROS generation (39). Like H 2 O 2 , treatment with 25 M menadione increased the formation of c-Abl⅐Arg complexes, whereas exposure to higher concentrations had less of an effect (Fig. 1C, left). Similar results were obtained when cells were treated with 20 ng/ml TNF-␣ to induce an endogenous oxidative stress response (40,41) (Fig.  1C, right). As a control, anti-c-Abl immunoprecipitates from c-abl Ϫ/Ϫ , and arg Ϫ/Ϫ MEFs showed no detectable signals when probed with anti-Arg (Fig. 1D, left). As additional controls, anti-c-Abl reacted specifically with c-Abl, whereas anti-Arg reacted specifically with Arg in immunoblot analyses of arg Ϫ/Ϫ and c-abl Ϫ/Ϫ MEFs (Fig. 1D, right).
To extend these findings, 293 cells were transfected to express Myc-tagged c-Abl (Myc-c-Abl) and FLAG-tagged Arg (FLAG-Arg). Immunoblot analysis of anti-Myc immunoprecipitates with anti-FLAG demonstrated detection of complexes containing c-Abl and Arg ( Fig. 2A). In the reciprocal experiment, immunoblot analysis of anti-FLAG immunoprecipitates with anti-Myc provided further support for the association of c-Abl and Arg in cells (Fig. 2B). To define the kinetics of the interaction between c-Abl and Arg, the parameters for binding of Arg were determined using GST-c-Abl immobilized to the sensor chip in a BIAcore. Arg bound to c-Abl with a dissociation constant (K d ) of 0.05 M (Fig. 2C). These findings demonstrate that c-Abl binds to Arg in the response to oxidative stress and that the interaction is direct.
c-Abl SH3 Interacts with the Arg C Terminus-To define the interaction between c-Abl and Arg, lysates from cells expressing FLAG-Arg were incubated with GST or GST fusion proteins containing the c-Abl SH3 or SH2 domains. The results show that FLAG-Arg associates with GST-c-Abl SH3 and not GSTc-Abl SH2 (Fig. 3A). To assess whether the interaction between c-Abl SH3 and Arg is direct, the GST fusion proteins were incubated with in vitro translated 35 S-FLAG-Arg. The finding that GST-c-Abl SH3 associates with 35 S-FLAG-Arg supported a direct interaction (Fig. 3B). Moreover, the finding that there is no detectable binding of 35 S-FLAG-Arg with the c-Abl SH2 domain supported specificity of the c-Abl SH3-Arg interaction (Fig. 3B). Other studies were performed with 35 S-labeled Arg(1-501) or Arg(532-1182) to define regions of Arg responsible for binding to c-Abl SH3. The results demonstrate that GST-c-Abl SH3 binds to Arg(532-1182) and not Arg(1-501) (Fig. 3C). Arg(532-1182) contains six proline-rich sequences that could function as binding sites for the c-Abl SH3 domain (Fig. 3D). To define the Arg site(s) responsible for the c-Abl SH3 interaction, FLAG-Arg proteins mutated at each of the PXXP sequences were incubated with GST-c-Abl SH3. Analysis of the adsorbates with anti-FLAG demonstrated that binding of c-Abl SH3 is decreased, but not completely abrogated, with the Arg(P570A/P573A) mutant (Fig. 3D). By contrast, the other Arg mutants had little if any effect on c-Abl SH3 binding (Fig.  3D). These findings demonstrate that the c-Abl SH3 domain interacts, at least in part, with the Arg proline-rich site at amino acids 567-576.
To assess binding in vivo, lysates from 293 cells expressing FLAG-Arg(1-501) or FLAG-Arg(532-1182) were subjected to immunoprecipitation with anti-c-Abl. Immunoblot analysis of the precipitates with anti-c-FLAG demonstrated that c-Abl associates with Arg(1-501) (Fig. 4A, left) and Arg(532-1182) (Fig. 4A, right). In the reciprocal analysis, coexpression of Arg and c-Abl(1-486) or c-Abl(487-1130) demonstrated that Arg associates with the c-Abl N-and C-terminal regions (Fig. 4B). In studies performed with 293 cells expressing Exp-c-Abl(1-486) and Arg(532-1182), immunoblot analysis of anti-Exp immunoprecipitates with anti-Arg also demonstrated association of the c-Abl N-terminal and the Arg C-terminal regions (Fig.  4C). By contrast, when anti-FLAG immunoprecipitates from 293 cells expressing FLAG-Arg(1-501) and Exp-c-Abl(1-486) were subjected to immunoblotting with anti-Exp, there was no detectable interaction between these N-terminal regions of c-Abl and Arg (data not shown). Moreover, in similar experiments performed on 293 cells expressing FLAG-Arg(532-1182) and Exp-c-Abl(487-1130), there was no apparent association of these c-Abl and Arg C-terminal regions (data not shown).
Taken together with the in vitro binding data, the results demonstrate that the c-Abl SH3 domain interacts with the Arg C-terminal region and that the c-Abl C-terminal region interacts with the Arg N-terminal region.
The Arg SH3 Domain Binds Directly to the c-Abl C Terminus-To determine whether the Arg SH3 domain binds to c-Abl, GST-Arg SH3 was incubated with in vitro translated 35 S-labeled FLAG-c-Abl. The results demonstrate that Arg SH3 binds directly to c-Abl (Fig. 5A). To extend this finding, we generated a c-Abl C-terminal (amino acids 487-1130) fragment. The results demonstrate that GST-Arg SH3 binds to 35 S-labeled c-Abl(487-1130) (Fig. 5B). These findings indicate that, in addition to binding of c-Abl SH3 to the Arg C terminus, a direct interaction between Arg SH3 and the c-Abl C-terminal region contributes to the formation of c-Abl⅐Arg complexes (Fig. 5C).
c-Abl-mediated Phosphorylation of Arg-To determine whether Arg is a substrate for c-Abl, in vitro translated/heatinactivated Arg was incubated with a 45-kDa kinase-active Abl and [␥-32 P]ATP. Analysis of the reaction products demonstrated that Abl phosphorylates the 145-kDa Arg protein (Fig.  6A). To assess tyrosine phosphorylation of Arg in vivo, a kinase-inactive FLAG-Arg(K-R) mutant was expressed in 293 cells. Immunoblot analysis of anti-FLAG immunoprecipitates with anti-Tyr(P) demonstrated that ectopically expressed Arg is constitutively phosphorylated on tyrosine (Fig. 6B). To confirm that Arg is phosphorylated by c-Abl in cells, c-abl Ϫ/Ϫ and c-abl ϩ cells were infected with a retrovirus expressing FLAG-Arg(K-R). Immunoblot analysis of anti-FLAG immunoprecipitates with anti-Tyr(P) showed that tyrosine phosphorylation of FLAG-Arg(K-R) is substantially higher in c-abl ϩ compared with c-abl Ϫ/Ϫ cells (Fig. 6C). These findings indicate that Arg is phosphorylated, at least in large part, by a c-Abl-dependent mechanism.
c-Abl and Arg Are Required for ROS-induced Apoptosis-To assess involvement of c-Abl and Arg in the response of cells to ROS, cells were studied for H 2 O 2 -induced apoptosis. Compared with MCF-7 cells expressing the empty vector, treatment of MCF-7/c-Abl(K-R) cells with 250 M H 2 O 2 resulted in an attenuated apoptotic response (Fig. 7A). Similar findings were obtained in MCF-7 cells stably expressing Arg(K-R) (Fig. 7A). In studies of MEFs, c-abl Ϫ/Ϫ cells exhibited little if any apoptosis in response to treatment with 40 or 250 M H 2 O 2 (Fig. 7B). The finding that stable expression of c-Abl in the c-abl Ϫ/Ϫ cells reconstitutes the apoptotic response to ROS demonstrates dependence on c-Abl (Fig. 7B). The arg Ϫ/Ϫ cells were also less sensitive to ROS-induced apoptosis compared with arg ϩ cells (Fig. 7B). The results further demonstrate that compared with MCF-7 cells expressing the empty vector, menadione-induced apoptosis is attenuated in MCF-7/c-Abl(K-R) and MCF/ Arg(K-R) cells (Fig. 7C). Similar results were obtained when these cells were treated with TNF-␣ (Fig. 7C). By contrast, apoptosis induced by stabilization of microtubules with pacli- taxel was unaffected by a expression of c-Abl(K-R) or Arg(K-R) (Fig. 7C). These findings collectively support a model in which both c-Abl and Arg are required for the apoptotic response to oxidative stress. DISCUSSION c-Abl Interacts with Arg-Recent findings that c-Abl and Arg are both activated in the cellular response to oxidative stress suggested that these related proteins may share similar functions (36,38). The present studies were thus performed to determine whether c-Abl and Arg interact in ROS-induced signaling. The results of coimmunoprecipitation studies dem-onstrate that endogenous c-Abl and Arg associate in the response to oxidative stress. These findings were confirmed by showing the association of ectopically expressed c-Abl and Arg in 293 cells. The results further demonstrate that c-Abl and Arg interact directly. The in vitro findings support direct binding of the c-Abl SH3 domain to a proline-rich site (amino acids 567-576) in the Arg C-terminal region. By contrast, there was no detectable binding of the c-Abl SH3 domain to the Arg N-terminal region. In concert with these results, expression of c-Abl(1-486) and Arg(572-1182) confirmed that the c-Abl Nterminal region associates with the Arg C-terminal region in cells. In addition, binding of c-Abl to the Arg(P570A/P573A) mutant was decreased compared with that found for wild-type Arg. The results also demonstrate that the Arg SH3 domain associates with the c-Abl C-terminal region. In concert with an interaction between c-Abl and Arg, we demonstrate that Arg functions as an in vitro substrate for c-Abl phosphorylation. Moreover, the results show that tyrosine phosphorylation of Arg in cells expressing c-Abl is substantially higher than that found in c-abl Ϫ/Ϫ cells. These findings demonstrate that c-Abl forms heterodimers with Arg in vivo by mechanisms involving intermolecular binding of the respective SH3 domains and C-terminal regions (Fig. 5C) and that Arg is phosphorylated by a c-Abl-dependent mechanism.
Induction of c-Abl⅐Arg Heterodimers in the Oxidative Stress Response-Certain insights into the involvement of c-Abl in the response of cells to oxidative stress came from the finding that ROS induce tyrosine phosphorylation and activation of protein kinase C␦ (34,35). In an apparent feedback mechanism, protein kinase C␦ activates c-Abl, and, in turn, c-Abl phosphorylates protein kinase C␦ on Tyr-512 (35,36). Activation of c-Abl is associated with targeting to mitochondria, release of mitochondrial cytochrome c, and induction of apoptosis (36,37). Arg is also activated in the oxidative stress response and contributes to the induction of apoptosis by interacting with the proapoptotic Siva-1 protein (38). The present results demonstrate that oxidative stress induces the formation of c-Abl⅐Arg heterodimers. Moreover, the findings show that binding of c-Abl and Arg is dependent on the concentration of H 2 O 2 . Thus, treatment with 10 to greater than 160 M H 2 O 2 was associated with increases in c-Abl⅐Arg heterodimers, whereas treatment with 640 M H 2 O 2 resulted in binding of c-Abl and Arg at a level found in control cells. The finding that menadione and TNF-␣ also induce the formation of c-Abl⅐Arg heterodimers is in concert with the effects of these agents on redox cycling and ROS generation (39,41,42). Thus, c-Abl and Arg form heterodimers in response to diverse agents that induce oxidative stress.
Regulation of the Apoptotic Response to Oxidative Stress by c-Abl and Arg-ROS have been implicated in the regulation of both mitogenic and apoptotic signaling pathways. Mitogenic signals induced by growth factors or activated Ras are mediated through ROS production (42,43). Other work has indicated that ROS induce topoisomerase II-mediated cleavage of chromosomal DNA and thereby apoptosis (44). The p66 shc adaptor protein (45) and the p85 subunit of phosphatidylinositol 3-kinase (46) have also been implicated in the apoptotic response to oxidative stress. Moreover, p53-induced apoptosis is mediated by ROS-dependent mechanisms (44,47,48). The present results provide support for involvement of both c-Abl and Arg in the apoptotic response to oxidative stress. Stable expression of either kinase-inactive c-Abl(K-R) or Arg(K-R) blocked H 2 O 2 -induced apoptosis of MCF-7 cells. In concert with these findings, the apoptotic response of MEFs to H 2 O 2 -induced oxidative stress was attenuated by targeted disruption of either c-abl or arg. Menadione-and TNF-␣-induced apoptosis was also attenuated by expression of c-Abl(K-R) or Arg(K-R). The present results also demonstrate that, in contrast to oxidative stress, both c-Abl and Arg are dispensable for paclitaxelinduced apoptosis. These findings thus provide the first evidence that c-Abl and Arg form heterodimers and that both c-Abl and Arg are required for the apoptotic response to oxidative stress.