Src Family Kinases Phosphorylate the Bcr-Abl SH3-SH2 Region and Modulate Bcr-Abl Transforming Activity*

Bcr-Abl is the oncogenic protein-tyrosine kinase responsible for chronic myelogenous leukemia. Recently, we observed that inhibition of myeloid Src family kinase activity (e.g. Hck, Lyn, and Fyn) induces growth arrest and apoptosis in Bcr-Abl-transformed cells, suggesting that cell transformation by Bcr-Abl involves Src family kinases (Wilson, M. B., Schreiner, S. J., Choi, H. J., Kamens, J., and Smithgall, T. E. (2002) Oncogene 21, 8075-8088). Here, we report the unexpected observation that Hck, Lyn, and Fyn strongly phosphorylate the SH3-SH2 region of Bcr-Abl. Seven phosphorylation sites were identified by matrix-assisted laser desorption ionization time-of-flight mass spectrometry: Tyr89 and Tyr134 in the Abl-derived SH3 domain; Tyr147 in the SH3-SH2 connector; and Tyr158, Tyr191, Tyr204, and Tyr234 in the SH2 domain. SH3 domain Tyr89, the most prominent phosphorylation site in vitro, was strongly phosphorylated in chronic myelogenous leukemia cells in a Src family kinase-dependent manner. Substitution of the SH3-SH2 tyrosine phosphorylation sites with phenylalanine substantially reduced Bcr-Abl-mediated transformation of TF-1 myeloid cells to cytokine independence. The positions of these tyrosines in the crystal structure of the c-Abl core and the transformation defect of the corresponding Bcr-Abl mutants together suggest that phosphorylation of the SH3-SH2 region by Src family kinases impacts Bcr-Abl protein conformation and signaling.

The cytogenetic hallmark of chronic myelogenous leukemia (CML) 3 is the Philadelphia (Ph) chromosome (1), which results from a reciprocal translocation between the c-abl proto-oncogene on chromosome 9 and the bcr locus on chromosome 22 (2,3). The Ph translocation results in the expression of Bcr-Abl, a 210-kDa oncogenic fusion protein with constitutive proteintyrosine kinase activity. Bcr-Abl transforms hematopoietic cell lines (4) and bone marrow cells (5) in culture and produces a CML-like myeloproliferative disease in mice (6,7), directly implicating this chimeric protein-tyrosine kinase in disease onset.
Bcr-Abl phosphorylates a broad range of cellular proteins, affecting many signaling pathways linked to the growth, differentiation, and survival of hematopoietic progenitors (3). For example, the Grb2-Sos guanine nucleotide exchange factor complex binds to phospho-Tyr 177 in the Bcr-derived region, contributing to the activation of Ras (8,9). Bcr-Abl also activates Ras via Shc, an adapter protein that couples the receptors for many growth factors and cytokines to the Grb2-Sos complex (10). In addition, Bcr-Abl associates with and phosphorylates the CrkL adapter (11), stimulates phosphatidylinositol 3-kinase/Akt survival signaling by direct interaction with the p85 subunit of phosphatidylinositol 3-kinase (12), and activates several Stat transcription factors (13,14). All of these signaling pathways involve components with SH2 and SH3 domains and are dependent upon interactions with phosphotyrosine or polyproline docking sites, respectively.
Despite its intrinsic protein-tyrosine kinase activity, several studies have established that Bcr-Abl interacts with other tyrosine kinases, including members of the Fps/Fes (15), Jak (16), and Src (17)(18)(19)(20) kinase families. Hallek and co-workers (20) originally showed that Bcr-Abl forms complexes with the Src family members Hck and Lyn in several Ph-positive cell lines and is phosphorylated by Hck at Bcr Tyr 177 (19), which links Bcr-Abl to Grb2-Sos as described above (8). Interestingly, Hck activation does not require Bcr-Abl kinase activity (19), suggesting that Bcr-Abl may stimulate Src family kinases through displacement of inhibitory intramolecular interactions (21). This idea is supported by subsequent work showing that the SH3 and SH2 domains of Hck directly associate with Bcr-Abl (17). Furthermore, kinase-defective Hck blocks transformation of myeloid leukemia cells to cytokine independence by Bcr-Abl (17), whereas pharmacological inhibitors selective for Src family kinases induce apoptosis in the CML cell lines Meg-01 and K-562 (18). The effects of these compounds correlate with down-regulation of both Stat5 and Erk (extracellular signalregulated kinase) activation, suggesting that Src family kinases may couple Bcr-Abl to certain downstream signaling pathways (18).
Although Bcr-Abl interacts directly with Hck, Lyn, and other Src family kinases, the molecular mechanisms and functional consequences of binding in terms of Bcr-Abl signaling and oncogenesis are not understood. In this study, we demonstrate that Bcr-Abl binds multiple members of the Src family through the Abl-derived SH3-SH2 regulatory region. Unexpectedly, Hck, Lyn, and Fyn phosphorylated the Bcr-Abl SH3-SH2 region at multiple tyrosine residues. The most prominent of these is SH3 domain Tyr 89 , which was also phosphorylated in CML cells in a Src family kinase-dependent manner. Substitution of Tyr 89 and other SH3-SH2 phosphorylation sites with phenylalanine reduced Bcr-Abl transforming activity without altering phosphorylation of the Abl kinase domain activation loop. These data show that Src family kinase-mediated phosphorylation of tyrosine residues located in the SH3-SH2 region is critical for Bcr-Abl function, offering a possible mechanistic explanation for the remarkable efficacy of Src-selective and dual Src/ Abl inhibitors against CML (18,22,23).

MATERIALS AND METHODS
Binding Assays-Interaction of glutathione S-transferase (GST)-fused human c-Abl proteins with Src family kinases was conducted in Sf9 insect cells as described previously (17). For binding assays, Sf9 cells (2.5 ϫ 10 6 ) were co-infected with each of the GST-Abl baculoviruses (or GST as a negative control) and wild-type Hck, Lyn, or Fyn baculovirus. Forty-eight hours after infection, the cells were lysed in Hck lysis buffer (17), and GST fusion proteins were precipitated with glutathione-agarose beads. The precipitates were washed with radioimmune precipitation assay buffer as described previously (17), and bound proteins were eluted in SDS-PAGE sample buffer. Proteins were resolved on SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes, and associated Hck, Lyn, and Fyn were visualized by immunoblotting with kinase isoform-specific polyclonal antibodies (Santa Cruz Biotechnology, Inc.). The amount of precipitated GST-Abl fusion protein present in each reaction was determined by immunoblotting with anti-GST antibodies (Santa Cruz Biotechnology, Inc.). Equivalent expression of each Src family kinase was verified by immunoblotting the clarified cell lysates. Immunoreactive bands were visualized with alkaline phosphataseconjugated goat anti-rabbit secondary antibodies with 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium as colorimetric substrate (SouthernBiotech).
GST-Bcr-Abl Fusion Protein Purification and in Vitro Kinase Assay-The Bcr-Abl SH3 domain (Gly 57 -Ser 126 ), SH2 domain (Ser 121 -Thr 224 ), and SH3-SH2 coding regions were subcloned into the bacterial expression vector pGEX-2T, and the corresponding GST fusion proteins were expressed in Escherichia coli and purified with glutathione-agarose beads (17). The recombinant proteins were dialyzed overnight against 20 mM HEPES (pH 7.4) and 1 mM ␤-mercaptoethanol, and protein concentrations were determined by densitometry of Coomassie Blue-stained SDS-polyacrylamide gels relative to a bovine serum albumin standard. Each purified GST fusion protein or GST alone (380 pmol) was incubated with recombinant purified Hck-YEEI (24), Lyn (Upstate), or Fyn (Upstate). The molar ratio of substrate to kinase was at least 20:1 in each case. Phos-phorylation reactions (50 l) were conducted in 50 mM HEPES (pH 7.4) containing 10 mM MgCl 2 and 500 M ATP for 30 min at 30°C and quenched by freezing in a dry ice/ethanol bath. Aliquots of each reaction were analyzed by immunoblotting with anti-GST or anti-phosphotyrosine (PY99) monoclonal antibodies (Santa Cruz Biotechnology, Inc.).
Hexahistidine-Bcr-Abl Fusion Protein Purification-The Bcr-Abl SH3-SH2 coding region (Gly 57 -Thr 224 ; human c-Abl numbering) was subcloned into the pET-14b expression vector (Novagen) to allow addition of an N-terminal hexahistidine tag, yielding His 6 -Abl3ϩ2. BL21(DE3)pLysS bacterial cells were transformed with the vector, and His 6 -Abl3ϩ2 protein expression was induced with isopropyl ␤-D-thiogalactopyranoside. His 6 -Abl3ϩ2 was purified from clarified cell extracts using a HiTrap metal-chelating column charged with Ni 2ϩ (GE Healthcare), and phosphorylated with recombinant Hck, Lyn, and Fyn as described above for the GST-Abl3ϩ2 protein.
Full-length kinase-dead p210 Bcr-Abl was expressed in Sf9 insect cells and purified as follows. A hexahistidine tag was added to the C-terminal coding region of a kinase-defective mutant of Bcr-Abl (15) using a PCR-based approach and subcloned into the baculovirus transfer vector pVL1393 (BD Biosciences). A recombinant baculovirus was created using the transfer vector and BaculoGold DNA according to the supplier's protocol. Spinner cultures of Sf9 cells (1 liter, 2 ϫ 10 6 cells/ ml) were infected with 100 ml of high titer virus and harvested 72 h later. Cells were lysed in 20 mM Tris-HCl (pH 8.3) containing 10% glycerol, 5 mM ␤-mercaptoethanol, 20 mM imidazole, and 500 mM NaCl. The protein was purified in one step using a HiTrap metal-chelating column. The Bcr-Abl-positive fractions were identified by SDS-PAGE and pooled, and the concentration of Bcr-Abl was quantitated by densitometry of a Coomassie Blue-stained gel. Phosphorylation reactions with Src family kinases were conducted as described above for the GST-Abl fusion proteins and contained ϳ8 pmol of recombinant Bcr-Abl p210-KR.
Mass Spectrometry (MS)-Recombinant His 6 -Abl3ϩ2 and full-length kinase-dead Bcr-Abl (p210-KR) were phosphorylated in vitro with purified Src kinases, and aliquots of the phosphorylation reactions were digested overnight with trypsin at a 1:25 trypsin/protein ratio. The His 6 -Abl3ϩ2 peptides were mixed with matrix (␣-cyano-4-hydroxycinnamic acid), spotted onto a MALDI target, and analyzed directly by MALDI-TOF-MS (Applied Biosystems 4700 proteomics analyzer). The area under the isotope distribution for the unphosphorylated and phosphorylated forms of each tryptic peptide was determined and used to estimate the relative extent of phosphorylation.
Tryptic fragments derived from full-length Bcr-Abl p210-KR were separated by reverse-phase capillary high pressure liquid chromatography (Dionex Ultimate Nano-LC system). Column fractions were mixed on-line with MALDI matrix and automatically spotted onto a MALDI plate at 20-s intervals with a Dionex Probot robot (144 spots total). Peaks for phosphopeptides and their unphosphorylated counterparts were identified in the MALDI-TOF spectra based on their m/z ratios, and the sequence of each peptide was verified by tandem MS (MS/MS) analysis when possible. The accuracy of the m/z measurements was Ϯ0.1 Da. An internal calibrant consisting of a mixture of des-Arg-bradykinin, angiotensin I, Glu-fibrinopeptide B, and ACTH- (18 -39) was added to each MALDI spot prior to analysis.

Coexpression of Src Family Kinases and Bcr-Abl in Sf9
Cells-Sf9 insect cells were co-infected with full-length kinase-dead Bcr-Abl and Hck, Lyn, or Fyn baculovirus as described previously (25). Cells were lysed 48 h later in 1.0 ml of ice-cold radioimmune precipitation assay buffer, and Bcr-Abl was immunoprecipitated with 4 g of anti-c-Abl monoclonal antibody 8E9 (Pharmingen) and 25 l of protein G-Sepharose (50:50 (w/v) slurry; GE Healthcare) by rotation overnight at 4°C. Precipitated proteins were collected by centrifugation, washed with radioimmune precipitation assay buffer, and eluted in SDS-PAGE sample buffer. Proteins were resolved by SDS-PAGE; transferred to polyvinylidene difluoride membranes; and immunoblotted with antibodies specific for Bcr phospho-Tyr 177 (Cellular Signaling Technology), c-Abl phospho-Tyr 245 (Cellular Signaling Technology), c-Abl phospho-Tyr 412 (Cellular Signaling and BIOSOURCE), and c-Abl protein (Pharmingen).
Generation of Phospho-specific Antibodies to Abl Phospho-Tyr 89 -Phospho-specific antibodies were raised in rabbits against a phosphopeptide corresponding to the Abl Tyr 89 phosphorylation site. The sequence of the peptide antigen was LFVALpYDFVASGDN. The antiserum was first purified by protein A chromatography, and nonspecific antibodies were removed by passing the protein A eluate through a second column on which the corresponding unphosphorylated peptide was immobilized. The eluate from this column was further purified on a column containing the original phosphopeptide. To test antibody specificity, recombinant His 6 -Abl3ϩ2 protein was phosphorylated with Hck as described above, resolved by SDS-PAGE, and immunoblotted with the anti-Abl phospho-Tyr 89 antibody at 1:1000 dilution. This experiment was repeated with purified His 6 -Abl3ϩ2 protein in which Tyr 89 was changed to Phe. The antibody was also used to probe Bcr-Abl Tyr 89 phosphorylation in the CML cell lines Meg-01 and K-562 by immunoblotting as described previously (18). To demonstrate the requirement for Src family kinases in the phosphorylation event, cell lines were treated with the Src family kinase-selective inhibitor A-419259 overnight prior to lysis and immunoblotting as described (18,26). Src family kinase activity was determined by immunoblotting with the anti-Src phospho-Tyr 418 antibody (BIOSOURCE), which reacts with the phosphorylated activation loop of all members of the Src kinase family (26).
Mutagenesis of Bcr-Abl SH3-SH2 Phosphorylation Sites-A 1121-bp oligonucleotide spanning the SH3-SH2 region of Bcr-Abl containing seven tyrosine-to-phenylalanine substitutions corresponding to the phosphorylation sites shown in Fig. 2 (7YF mutant) and flanked by unique restriction sites was commercially synthesized (DNA 2.0 Inc.). The SH3-SH2 7YF oligonucleotide was swapped for the corresponding region of wildtype Bcr-Abl, and the resulting full-length Bcr-Abl 7YF coding region was subcloned into the retroviral vector pMSCV-neo (Clontech). The Y89F single mutant and a 6YF add-back mutant were created using the QuikChange site-directed mutagenesis kit (Stratagene). To create the 6YF mutant, the Bcr-Abl 7YF mutant was used as the template, and oligonucleotides encoding Tyr 89 were used to revert this site from Phe to Tyr. These mutants and wild-type Bcr-Abl were also subcloned into the pMSCV-neo retroviral vector.
Transformation of TF-1 Cells with Bcr-Abl Retroviruses-The human granulocyte-macrophage colony-stimulating factor (GM-CSF)-dependent myeloid leukemia cell line TF-1 (27) was obtained from American Type Culture Collection and grown in RPMI 1640 medium supplemented with 10% fetal bovine serum, 50 g/ml gentamycin, and 1 ng/ml recombinant human GM-CSF. To make retroviral stocks, 293T cells were cotransfected with each retroviral construct and an amphotropic packaging vector as described (28,29). TF-1 cells (10 6 ) were incubated with 5 ml of viral supernatant in the presence of 4 g/ml Polybrene and centrifuged at 2400 rpm for 3 h at room temperature to enhance infection (17,28). Cell populations were selected with 800 g/ml G418 for 10 -14 days. To analyze wildtype and mutant Bcr-Abl protein expression and phosphorylation status, 10 7 cells were collected by centrifugation, washed once with phosphate-buffered saline, and lysed in ice-cold radioimmune precipitation assay buffer supplemented with the protease inhibitors aprotinin (25 g/ml), leupeptin (25 g/ml), and phenylmethylsulfonyl fluoride (1 mM) and the phosphatase inhibitors NaF (10 mM) and Na 3 VO 4 (1 mM). Clarified lysates were resolved by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and blotted with antibodies specific for Abl protein and site-specific phosphorylation as described above. GM-CSF-independent cell proliferation was assessed using the CellTiter-Blue cell viability assay (Promega Corp.). Cells were washed and seeded in 5-ml cultures at 10 5 /ml in the absence of GM-CSF. For each time point, 100 l of each culture was withdrawn, mixed with 20 l of the assay reagent in a 96-well plate, and incubated at 37°C for 120 min prior to reading the fluorescence (excitation at 560 nm and emission at 590 nm) on a Molecular Dynamics SpectraMax Gemini XS fluorometric plate reader. All time points were assayed in triplicate, and the entire experiment was repeated three times.

Src Family Kinases Phosphorylate the Bcr-Abl SH3-SH2
Region at Multiple Tyrosines-Previously, we found that the Src family kinase Hck binds directly to the SH3-SH2 region, kinase domain, and C-terminal region of Bcr-Abl in vitro (17). In the present study, we extended these findings to include Lyn and Fyn, which are also present in Bcr-Abl-transformed myeloid progenitor cells. Each Src family member was coexpressed in Sf9 insect cells along with a series of GST fusion proteins encompassing the Abl-derived portion of Bcr-Abl (Fig. 1A). The fusion proteins were precipitated and analyzed for bound Src family kinases by immunoblotting. Hck, Lyn, and Fyn exhibited a nearly identical pattern of binding involving the SH3-SH2, kinase, and C-terminal regions of Abl (Fig. 1B). No association was observed with GST alone. This result shows that Src family kinases expressed in myeloid cells associate with Bcr-Abl by a common mechanism.
The immunoblots from the binding assays shown in Fig. 1 were then reprobed with anti-phosphotyrosine antibodies. Very strong phosphorylation of the Abl SH3-SH2 region was observed upon coexpression with Hck, Lyn, and Fyn (data not shown). To confirm this finding, we purified the GST-Abl3ϩ2 fusion protein and tested it as a substrate for recombinant Hck, Lyn, and Fyn in vitro. As shown in Fig. 1C, catalytic amounts of all three Src family kinases strongly phosphorylated the Bcr-Abl SH3-SH2 region under these conditions. Lower levels of phosphorylation were observed with equimolar amounts of GST-Abl SH2 and GST-Abl SH3 domain fusion proteins, suggesting that tethering of the domains is required for proper recognition and phosphotransfer by Src family kinases. No phosphorylation of GST was observed, indicating that the phosphorylation sites localize to the Abl-derived portion of each fusion protein.
We next identified the sites of Src family kinase-mediated Abl SH3-SH2 phosphorylation by MALDI-TOF-MS. To eliminate interference from the GST moiety, the SH3-SH2 region was re-expressed with a hexahistidine tag at the N terminus (His 6 -Abl3ϩ2), purified, and incubated in vitro with a catalytic amount of recombinant Hck, Lyn, or Fyn. As shown in Fig. 2A, all three Src kinases strongly phosphorylated the purified His 6 -Abl3ϩ2 protein.
Each phosphorylation reaction was digested overnight with trypsin, and the resulting peptides were analyzed by MALDI-TOF-MS. Of 10 possible tyrosines in the His 6 -Abl3ϩ2 protein, seven were reproducibly phosphorylated: Tyr 89 and Tyr 134 in the SH3 domain; Tyr 147 in the SH3-SH2 connector; and Tyr 158 , Tyr 191 , Tyr 204 , and Tyr 234 in the SH2 domain. The extent of phosphorylation of each peptide (estimated from the ratios of the peak intensities for the phosphopeptides and their unphosphorylated counterparts) is presented in Fig. 2A. Tyr 89 was phosphorylated to very high stoichiometry by all three Src family members. Interestingly, Hck phosphorylated several other sites to higher stoichiometry relative to Lyn or Fyn. Whether this represents true differences in substrate specificity among the Src kinase isoforms or simply differences in the specific activities of the purified kinase preparations will require further investigation. Note that Tyr 112 in the Abl SH3 domain and Tyr 186 and Tyr 193 in the SH2 domain were not detectably phosphorylated, indicative of selectivity for specific Tyr sites (data not shown).
The positions of these novel Src family kinase tyrosine phosphorylation sites within the crystal structure of the c-Abl core region (30) are shown in Fig. 2B. Intriguingly, four of these tyrosine residues (Tyr 89 and Tyr 134 in the SH3 domain and Tyr 158 and Tyr 191 in the SH2 domain) lie along the interface between the SH3-SH2 region and the kinase domain. Because c-Abl kinase activity is down-regulated in part through tight docking of the SH3-SH2 "clamp" onto the back of the kinase domain, phosphorylation of residues along this interface could influence kinase activity, even in the context of Bcr-Abl (see "Discussion"). Numbering is based on the human c-Abl sequence. B, the recombinant GST-Abl fusion proteins shown in A and GST alone as a negative control were coexpressed with Hck, Lyn, or Fyn in Sf9 insect cells. Fusion proteins were precipitated from clarified cell extracts using glutathione-agarose beads and washed, and associated Src family kinases were visualized by immunoblotting. The recovery of each GST fusion protein was verified by immunoblotting an aliquot of the precipitate with anti-GST antibodies (data not shown). Equivalent expression of each Src family member was verified by immunoblotting the cell lysates with Src family kinase-specific antibodies (not shown). C, recombinant GST-Abl SH3 domain, GST-Abl SH2 domain, and GST-Abl3ϩ2 fusion proteins and GST alone were purified from bacteria and incubated in the absence (No kinase) or presence of recombinant Hck, Lyn, or Fyn as described under "Materials and Methods." Aliquots of each reaction were separated by SDS-PAGE, and phosphotyrosine ( P-Tyr) content was determined by immunoblotting (upper panels). The positions of GST and each GST-Abl fusion protein are indicated on the right (arrows). Replicate filters were probed with anti-GST antibodies to confirm equal amounts of purified protein in each reaction (lower panels). Each experiment was repeated at least twice with comparable results.
Hck Phosphorylates Full-length Bcr-Abl in the SH3-SH2 Region-We next investigated whether Src family kinases phosphorylate the Abl SH3-SH2 region within the context of the full-length Bcr-Abl protein. Full-length kinase-dead p210 Bcr-Abl was expressed in Sf9 insect cells and purified by virtue of a hexahistidine tag on its C terminus. This form of Bcr-Abl cannot undergo autophosphorylation, allowing for clear characterization of Src family kinase-dependent phosphorylation events. Purified kinasedead Bcr-Abl (p210-KR) was incubated either alone or with a catalytic amount of recombinant Hck, and an aliquot of the phosphorylation reaction was analyzed by anti-phosphotyrosine immunoblotting. As shown in Fig. 3, p210-KR was strongly phosphorylated in the presence of Hck. The phosphorylated p210-KR protein was then digested with trypsin, and the resulting peptides were separated by liquid chromatography. Each column fraction was spotted onto a MALDI plate, and a total of 144 spectra were collected and analyzed. Using a peak picking routine, we were able to identify ions corresponding to six of the phosphotyrosine-containing peptides originally observed in the smaller His 6 -Abl3ϩ2 construct (Fig. 2). Fig. 3B presents representative spectra for the tryptic peptides containing Tyr 147 , a phosphorylation site found in the SH3-SH2 connector. The peak at 1225.59 Da corresponds to the Tyr 147 tryptic peptide (HSWYHGPVSR), and the peak at 1305.55 Da corresponds to its phosphorylated counterpart (HSWpYHGPVSR). Note the expected ϳ80-Da shift in molecular mass corresponding to the covalent addition of a phosphate group by Hck. The sequences of these peptides were confirmed in subsequent MS/MS experiments, which are presented in Fig. 4.   OCTOBER 13, 2006 • VOLUME 281 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 30911
The complete results of the MALDI-TOF-MS analysis of the Hck-phosphorylated, full-length Bcr-Abl p210-KR protein are summarized in Table 1. In addition to Tyr 147 , mass peaks corresponding to the Tyr 134 and Tyr 158 tryptic phosphopeptides (along with their unphosphorylated counterparts) were identi-fied and confirmed by MS/MS sequencing. A mass peak corresponding to the Tyr 191 tryptic phosphopeptide (VYHYR) could not be found, despite strong phosphorylation of this site by Hck in the His 6 -Abl3ϩ2 protein (Fig. 2). However, the unphosphorylated form of the Tyr 191 peptide was also absent, suggesting possible loss during chromatography due to the small size and hydrophobic nature of this peptide. Candidate peaks for both the phosphorylated and unphosphorylated peptides containing Tyr 89 , Tyr 204 , and Tyr 234 were detected, although low abundance prevented sequence confirmation by MS/MS analysis. The Tyr 89 peptide is much longer in the context of full-length Bcr-Abl relative to His 6 -Abl3ϩ2, where it is directly adjacent to the hexahistidine tag; this may account for its low abundance.
We also surveyed the full-length Bcr-Abl data set for phosphorylated tryptic peptides corresponding to reported sites of c-Abl and Bcr-Abl autophosphorylation and c-Abl transphosphorylation by Src family kinases, including Tyr 245 in the SH2-kinase linker, Tyr 412 in the kinase domain activation loop, and Tyr 177 in the Bcr region (Table 1). Tryptic phosphopeptides corresponding to Tyr 177 and Tyr 412 were detected (as were their unphosphorylated counterparts), and their amino acid sequences and phosphorylation states were confirmed by MS/MS sequencing. In contrast, the tryptic peptide containing Tyr 245 or its phosphorylated counterpart could not be located, despite strong reactivity of this site with a phospho-specific antibody (see below). This is likely due to the limitations  of separation and detection of the large number of Bcr-Abl peptides.
To complement the MS analysis of Bcr-Abl tyrosine phosphorylation, we also used an immunoblot strategy with phospho-specific antibodies. For these experiments, full-length kinase-dead Bcr-Abl was expressed alone or together with Hck, Lyn, or Fyn in Sf9cells.ActiveBcr-Ablwasalsoexpressedalonetogaugeautophosphorylation. Bcr-Abl proteins were then immunoprecipitated and probed with antibodies specific for phosphorylation of Bcr Tyr 177 , Tyr 245 , and Tyr 412 . As shown in Fig. 5A, strong phosphorylation of all three sites by Hck, Lyn, and Fyn was observed using this approach, suggesting that these sites are transphosphorylated by Src family kinases in full-length Bcr-Abl. Note that all three sites were also strongly autophosphorylated within kinase-active Bcr-Abl. For reference, the locations of Tyr 245 in the SH2-kinase linker and Tyr 412 in the activation loop are mapped onto the crystal structure of the c-Abl core in Fig. 5B.
Src Family Kinase-dependent Phosphorylation of the Bcr-Abl SH3 Domain in CML Cells-We next investigated whether phosphorylation of the Bcr-Abl SH3-SH2 region by Src family kinases occurs in the context of CML cells. For these experiments, we focused on Tyr 89 , the SH3 site most strongly phosphorylated by Hck, Lyn, and Fyn in vitro (Fig. 2). To probe Tyr 89 phosphorylation in cells, we raised a phospho-specific antibody against this site and validated it using the recombinant His 6 -Abl3ϩ2 protein originally used to map the phosphorylation sites by MS. As shown in Fig. 6A, the anti-Abl phospho-Tyr 89 antibody did not recognize the unphosphorylated His 6 -Abl3ϩ2 protein.
However, incubation of the protein with Hck prior to immunoblotting resulted in a strong signal with the antibody. The experiment was then repeated with a mutant form of His 6 -Abl3ϩ2 in which Tyr 89 was replaced with Phe (Y89F mutant). This mutant form of His 6 -Abl3ϩ2 did not react with the anti-Abl phospho- FIGURE 5. Src family kinases phosphorylate full-length Bcr-Abl in the SH2-kinase linker, activation loop, and Grb2-binding sites. A, full-length kinase-dead Bcr-Abl was expressed in Sf9 insect cells either alone (control (Con)) or together with Hck, Lyn, or Fyn (left panels). Active Bcr-Abl was expressed alone as a positive control (right panels). Bcr-Abl proteins were immunoprecipitated using an anti-Abl antibody and immunoblotted with phospho-specific antibodies for the Grb2-binding site (Bcr phospho-Tyr 177 ( pY177)), the Abl SH2-kinase linker (phospho-Tyr 245 ( pY245)), the activation loop tyrosine (phospho-Tyr 412 ( pY412)), and the Bcr-Abl protein. Each experiment was repeated at least twice with comparable results. B, the location of the linker (Tyr 245 ) and activation loop (Tyr 412 ) tyrosine residues are indicated on the crystal structure of the c-Abl core (30); domains are colored as described in the legend to Fig. 2. Bcr-Abl Tyr 177 is found in the N-terminal Bcr-derived portion of the protein and is not present in this structure. Tyr 89 antibody following incubation with Hck, indicating that the antibody is specific for Bcr-Abl phospho-Tyr 89 . A replicate filter with the same four samples was then probed with a general anti-phosphotyrosine antibody. In this case, immunoreactivity was reduced but not eliminated with the Y89F form of His 6 -Abl3ϩ2 following phosphorylation by Hck in comparison with the wild-type protein. This observation agrees with the MS showing that Tyr 89 is the preferred site of phosphorylation for Src family kinases in this protein (Fig. 2).
Using the anti-Abl phospho-Tyr 89 antibody, we next investigated the phosphorylation of Tyr 89 in the CML-derived cell lines K-562 and Meg-01. Our previous work has shown that these two cell lines are very sensitive to Src family kinase inhibitors in terms of growth arrest and programmed cell death (18). Fig. 6B shows that lysates from both CML cell lines exhibited a band of 210 kDa that reacted strongly with the anti-Abl phospho-Tyr 89 antibody, indicating that the SH3 domain of Bcr-Abl is phosphorylated at this site in cells. In contrast, the Ph-negative leukemia cell line TF-1 showed no immunoreactivity with the anti-Abl phospho-Tyr 89 antibody, consistent with the lack of Bcr-Abl expression in this cell line. As expected, control immunoblots showed expression of p210 Bcr-Abl protein in K-562 and Meg-01 cells, but not in TF-1 cells.
To implicate Src family kinases in the phosphorylation of the Bcr-Abl SH3 domain in CML cells, we employed the Src family kinase inhibitor A-419259 (31). Previous studies by our group have shown that this compound blocks Src family kinase activity in vitro in the low nanomolar range, but is at least 2 orders of magnitude less active against the Abl kinase domain (18,26). A-419259 also induces growth suppression and apoptotic cell death in both the K-562 and Meg-01 CML cell lines, but does not affect Ph-negative myeloid leukemia cells (18). To determine whether Src family kinase activity is required for phosphorylation of Bcr-Abl at Tyr 89 , Meg-01 cells were treated with a range of A-419259 concentrations, followed by immunoblotting with the anti-Abl phospho-Tyr 89 antibody. As shown in Fig. 6C, phosphorylation of Abl Tyr 89 was blocked in a concentration-dependent manner, with IC 50 ϳ 300 nM, and phosphorylation of this site was completely blocked at 1 M. This dose response for inhibition of Abl Tyr 89 phosphorylation closely paralleled suppression of Src family kinase autophosphorylation (Src phospho-Tyr 418 immunoblot), strongly implicating Src family kinases in the phosphorylation of the Bcr-Abl SH3 domain at Tyr 89 in CML cells. Control blots showed equivalent levels of Bcr-Abl protein in each sample. Similar results were obtained with K-562 cells (data not shown).
The data presented in Fig. 5 and Table 1 show that Hck phosphorylated Bcr-Abl at Bcr Tyr 177 , Abl Tyr 245 , and Abl Tyr 412 in vitro, all of which are known sites of Bcr-Abl phosphorylation. To evaluate whether Hck and other Src kinases contribute to phosphorylation of these sites in CML cells, lysates from the A-419259-treated Meg-01 cells were probed with phosphospecific antibodies directed against these sites. As shown in Fig.  6C, A-419259 treatment also substantially reduced the phosphorylation of these sites, suggesting that Src family kinases may have a major role in maintaining the active conformation of Bcr-Abl in CML cells (see "Discussion"). Very similar results were obtained with K-562 cells (data not shown).

Phosphorylation of Tyr 89 and Other Sites in the SH3-SH2
Region Is Required for Full Bcr-Abl Transforming Function-To determine whether tyrosine phosphorylation of the SH3-SH2 region is important for Bcr-Abl function, we engineered a mutant form of Bcr-Abl (7YF) in which phenylalanine replaced each of the seven SH3-SH2 tyrosine phosphorylation sites for Src family kinases shown in Fig. 2 and Table 1. The 7YF mutant was then compared with wild-type Bcr-Abl in terms of its ability to transform the human TF-1 myeloid cell line to cytokine independence. TF-1 cells require GM-CSF or interleukin-3 for growth and survival and undergo apoptosis following cytokine withdrawal (27); introduction of Bcr-Abl with a recombinant retrovirus reverses the cytokine dependence. TF-1 cells were infected with wild-type and 7YF mutant Bcr-Abl retroviruses, and cell growth in the absence of GM-CSF was measured over 3 days. Fig. 7A shows that the cytokine-independent proliferation of TF-1 cells expressing the 7YF mutant was reduced by Ͼ50% relative to cells expressing wild-type Bcr-Abl, providing evidence that phosphorylation of the Bcr-Abl SH3-SH2 region is required for full transforming activity.
Because Tyr 89 is most prominent among the Src family kinase phosphorylation sites in the Bcr-Abl SH3-SH2 region, we FIGURE 7. Tyrosine phosphorylation of the SH3-SH2 region is essential for full Bcr-Abl biological activity. The seven Bcr-Abl SH3-SH2 tyrosine phosphorylation sites for Src family kinases identified by MS were replaced with Phe in full-length Bcr-Abl (7YF mutant). Tyr 89 was restored in the context of the 7YF mutant to create mutant 6YF. A single point mutant of Tyr 89 was also created (Y89F). Each of these mutants, as well as wild-type (WT ) Bcr-Abl, were expressed in the human GM-CSF-dependent myeloid progenitor cell line TF-1 using recombinant retroviruses. A, GM-CSF-independent proliferation of each cell line was monitored using the CellTiter-Blue cell viability assay as described under "Materials and Methods." Cells infected with a green fluorescent protein retrovirus served as negative control (Con). The experiment was performed in triplicate, and the mean -fold increase in cell number Ϯ S.D. is shown at each time point. B, lysates from each of the cell lines shown in A were probed with phospho-specific antibodies against Bcr-Abl phospho-Tyr 89 ( pY89) and phospho-Tyr 412 ( pY412). Replicate membranes were also probed with a general anti-phosphotyrosine antibody ( pTyr) and with anti-Abl antibodies to control for Bcr-Abl expression. Blotting was performed on two separate infected cell populations and produced the same result in each case.
next investigated whether adding back this single phosphorylation site restores the biological activity of the 7YF mutant in the TF-1 cell transformation assay. This mutant, termed 6YF, showed transforming activity intermediate to that observed with wild-type Bcr-Abl and the 7YF mutant. In addition, we created a Bcr-Abl point mutant in which Tyr 89 alone was replaced with Phe (Y89F). This mutant also exhibited transforming activity intermediate to that of wild-type Bcr-Abl and the 7YF mutant. Taken together, these results suggest that, although phosphorylation of Tyr 89 is required for full transforming activity, the other tyrosine phosphorylation sites within the SH3-SH2 region are likely to contribute to Bcr-Abl function in CML cells.
Finally, we investigated the impact of SH3-SH2 phosphorylation site mutagenesis on Bcr-Abl at the biochemical level by immunoblotting lysates from each of the transformed TF-1 cell populations with phospho-specific antibodies. As shown in Fig.  7B, wild-type Bcr-Abl reacted strongly with the anti-Abl phospho-Tyr 89 antibody, indicating that this site is phosphorylated in TF-1 cells in a manner similar to that in the CML cell lines (Fig. 6B). The immunoreactivity of the anti-Abl phospho-Tyr 89 antibody was sharply reduced in TF-1 cells expressing the Bcr-Abl 7YF and Y89F mutants, but was completely restored with the Tyr 89 add-back mutant, 6YF. In contrast, phosphorylation of the Bcr-Abl activation loop (phospho-Tyr 412 immunoblot) and the overall Bcr-Abl phosphotyrosine content (phospho-Tyr immunoblot) were not remarkably changed between the wild-type and mutant Bcr-Abl proteins, supporting the idea that the Tyr-to-Phe substitutions in the SH3-SH2 region do not simply disrupt the folding of the Bcr-Abl protein. Notably, all four TF-1 cell populations expressed equivalent levels of Bcr-Abl protein (Abl immunoblot).

DISCUSSION
This work provides the first evidence for tyrosine phosphorylation of the Bcr-Abl SH3-SH2 regulatory region by Src family kinases both in vitro and in CML cells. The possible impact of these phosphorylation events on Bcr-Abl function requires further consideration of c-Abl structure and regulation. Although Bcr-Abl exhibits constitutive tyrosine kinase activity, Hantschel and Superti-Furga (32) have proposed that Bcr-Abl may retain some of the regulatory features observed in the recent x-ray crystal structures of the c-Abl core (see Fig. 2) (30,33). These structures show that, in the down-regulated conformation, the c-Abl SH3 domain engages the polyproline type II helix formed by the SH2-kinase linker in an intramolecular fashion, as is the case for c-Src and Hck (21). Unlike Src kinases, however, the SH2 domain docks onto the back of the C-terminal lobe of the Abl kinase domain. This interaction is stabilized by binding of the myristoylated N-terminal cap through a unique pocket in the kinase domain (Fig. 2). Residues linking the SH3 and SH2 domains form a rigid connector that dynamically couples the SH3 and SH2 domains, which together provide a regulatory clamp that allosterically holds the kinase domain in the closed inactive state (30). Although the regulatory impact of myristoylation is lacking in Bcr-Abl, some evidence suggests that SH3/linker interaction may be retained. For example, Smith et al. (34) showed that a transformation defect associated with mutations in the N-terminal coiled-coil region of Bcr-Abl can be complemented by mutations predicted to disrupt SH3/linker interaction, consistent with the idea that the SH3 domain still exerts some negative regulatory influence over Bcr-Abl tyrosine kinase activity. Phosphorylation of the SH3 domain at Tyr 89 or Tyr 134 by Src family kinases as described here may have a similar destabilizing effect, as these sites come in close proximity to the SH2-kinase linker. Indeed, phosphorylation of the Abl SH3-SH2 region by Hck caused dissociation of the SH2-kinase linker as determined by hydrogen exchange MS. 4 In addition to the SH3 sites, several of the other tyrosine residues phosphorylated by Src family kinases localize to the interface between the SH3-SH2 regulatory clamp and the large lobe of the c-Abl kinase domain (Fig. 2). Particularly interesting is the SH2 domain residue Tyr 158 . The aromatic rings of Tyr 158 and the kinase domain residue Tyr 361 stack together in the down-regulated c-Abl structure, and the hydroxyl group of Tyr 158 forms a hydrogen bond with the backbone carbonyl group of Asn 393 of the kinase domain (30). Interestingly, mutation of Tyr 158 to glutamate results in a nearly 4-fold stimulation of c-Abl kinase activity compared with the wild type (35). Phosphorylation of this site by Src family kinases may also disrupt kinase regulation of c-Abl and raises the possibility of a similar regulatory interaction in Bcr-Abl.
Other recent studies show that mutations outside of the catalytic domain can allosterically affect not only Bcr-Abl kinase activity but sensitivity to imatinib as well. For example, Azam et al. (36) used an unbiased random mutagenesis screen to uncover novel mutations in Bcr-Abl that confer imatinib resistance and mapped these residues onto the autoinhibited c-Abl structure. These authors uncovered a striking correlation between residues that impair c-Abl negative regulation and imatinib resistance of Bcr-Abl, again strongly suggesting that mechanisms governing c-Abl autoinhibition are retained in Bcr-Abl. These observations agree with previous work showing that imatinib favors the down-regulated conformation of the Abl catalytic cleft for binding (33,37). Notably, several of these imatinib resistance mutations localize to SH3, SH2, and kinase domain residues that contribute to the negative regulatory interface. In particular, Azam et al. (36) found that substitution of the Bcr-Abl SH3 domain residue Tyr 89 , which is strongly phosphorylated by Hck, Lyn, and Fyn in vitro (Fig. 2) and in CML cells (Fig. 6), resulted in Bcr-Abl imatinib resistance in four independent isolates. Phosphorylation of this site by Src family kinases may also stabilize the active conformation of Bcr-Abl and contribute to its transforming activity. The reduced transforming activity of the Y89F mutant in TF-1 cells also supports this view (Fig. 7).
In addition to novel sites in the SH3-SH2 region, we found that Hck, Lyn, and Fyn phosphorylate full-length Bcr-Abl at Tyr 177 in the Bcr region and at Tyr 245 and Tyr 412 in the Abl kinase domain (Fig. 5 and Table 1). Although phosphorylation of Bcr-Abl Tyr 245 by Src family kinases was readily detected both in vitro and in CML cells using a phospho-specific antibody (Figs. 5 and 6), we were unable to locate this phosphopeptide (or its unphosphorylated counterpart) by MS. This discrepancy most likely reflects technical limitations of the MS-based approach. Experiments with the selective inhibitor A-419259 strongly suggested that Src family kinases contribute to the phosphorylation of these sites in CML cells as well (Fig.  6). Previous work has shown that Hck phosphorylates Bcr Tyr 177 (19), creating a docking site for the Grb2-Sos guanine nucleotide exchanger for Ras (8,9). Tyr 245 and Tyr 412 have been established as sites of both c-Abl autophosphorylation (38) and transphosphorylation by Src family kinases (discussed below) (39,40), and phosphorylation of these sites strongly up-regulates c-Abl kinase activity (38).
Several studies have linked Src family kinases to c-Abl regulation and signaling, which have important implications for Bcr-Abl function and drug sensitivity. Plattner et al. (39) demonstrated that the kinase activity of c-Abl increases 10 -20-fold in the presence of constitutively active v-Src in mouse Ba/F3 hematopoietic cells and 10T1/2 fibroblasts. This increase in c-Abl kinase activity directly correlates with phosphorylation of c-Abl by Src or Fyn. Dorey et al. (40) showed that active c-Src can phosphorylate the c-Abl activation loop at Tyr 412 and enhance the ability of c-Abl to phosphorylate the downstream substrate c-Jun. A similar study provided evidence that c-Src phosphorylates c-Abl at both Tyr 412 and Tyr 245 in NIH 3T3 cells and that dual phosphorylation by c-Src is required for c-Abl function (41). Consistent with this report, Brasher and Van Etten (38) proposed that phosphorylation of Tyr 245 enhances c-Abl activity by disrupting SH3/linker interaction and demonstrated that dual phosphorylation of Tyr 245 and Tyr 412 is necessary for full catalytic activation. Finally, Hck can desensitize Abl to inhibition with imatinib by phosphorylation of Tyr 412 in vitro (37,42). These studies provide important precedents suggesting that Src family kinase-directed phosphorylation of Bcr-Abl may stabilize an active conformation of the kinase via transphosphorylation. Disruption of intramolecular interactions may also expose the SH3 and SH2 domains for target protein binding. The possible dependence of Bcr-Abl on Src family kinase-dependent phosphorylation documented here may help to explain the efficacy of dual Src/Abl kinase inhibitors currently in clinical trials as second-line agents for imatinib-resistant CML (23). Interestingly, up-regulation of Lyn and Hck expression has been reported in imatinib-resistant blasts from CML patients (43). Whether this event contributes directly to imatinib resistance through direct phosphorylation of Bcr-Abl will require further investigation.