Association of Bcr-Abl with the Proto-oncogene Vav Is Implicated in Activation of the Rac-1 Pathway*

Vav is a guanine nucleotide exchange factor for the Rho/Rac family predominantly expressed in hematopoietic cells and implicated in cell proliferation and cytoskeletal organization. The oncogenic tyrosine kinase Bcr-Abl has been shown to activate Rac-1, which is important for Bcr-Abl induced leukemogenesis. Previous studies by Matsuguchi et al. (Matsuguchi, T., Inhorn, R. C., Carlesso, N., Xu, G., Druker, B., and Griffin, J. D. (1995)EMBO J. 14, 257–265) describe enhanced phosphorylation of Vav in Bcr-Abl-expressing Mo7e cells yet fail to demonstrate association of the two proteins. Here, we report the identification of a direct complex between Vav and Bcr-Abl in yeast, in vitroand in vivo. Furthermore, we show tyrosine phosphorylation of Vav by Bcr-Abl. Mutational analysis revealed that the SH2 domain and the C-terminal SH3 domain as well as a tetraproline motif directly adjacent to the N-terminal SH3 domain of Vav are important for establishing this phosphotyrosine dependent interaction. Activation of Rac-1 by Bcr-Abl was abrogated by co-expression of the Vav C terminus encoding the SH3-SH2-SH3 domains as a dominantnegative construct. Bcr-Abl transduced primary bone marrow from Vav knock-out mice showed reduced proliferation in a culture cell transformation assay compared with wild-type bone marrow. These results suggest, that Bcr-Abl utilizes Vav as a guanine nucleotide exchange factor to activate Rac-1 in a process that involves a folding mechanism of the Vav C terminus. Given the importance of Rac-1 activation for Bcr-Abl-mediated leukemogenesis, this mechanism may be crucial for the molecular pathogenesis of chronic myeloid leukemia and of importance for other signal transduction pathways leading to the activation of Rac-1.

The proto-oncogene p95 Vav-1 is expressed predominantly in hematopoietic cells (1). The protein consists of several conserved domains, among them a calponin homology domain, a Dbl homology domain, a pleckstrin homology domain, and a cysteine-rich region, as well as an Src homology 2 (SH2) 1 do-main framed by two SH3 domains, suggesting a role for Vav in multiple signal transduction pathways (2,3). Deletion of the N-terminal region encoding the calponin homology domain results in oncogenic activation of Vav (1). Vav has been shown to be important for the transduction of signals from the T-cell receptor; Vav knock-out animals have defects in T cell development, and their T-cells proliferate poorly and produce little IL-2 in response to T-cell receptor stimulation (4 -7). In addition, Vav-deficient T-cells have specific defects in T-cell receptor-induced calcium fluxes and in the activation of the extracellular signal-related kinase (ERK), mitogen-activated protein kinase (MAPK), and NF-B pathways (8). Vav is phosphorylated in response to T-cell receptor stimulation and to cytokines and growth factors like IL-3, SCF, platelet-derived growth factor, or granulocyte/macrophage colony-stimulating factor, suggesting that it is also involved in the signal transduction pathways activated by these receptors (9 -12). Through this mechanism, the SH2 domain of Vav mediates association with the tyrosine kinase receptor itself or with cytoplasmic tyrosine kinases like Lyn, JAK, or ZAP-70 (3,9,11,13). Vav has been shown to be a GDP/GTP exchange factor (GEF) for members of the Rho family of GTPases (14). It has been demonstrated that tyrosine phosphorylation of Vav in vitro and in vivo by the tyrosine kinase Lck can activate its Rac GEF activity (14). The GEF activity of Vav is encoded by the Dbl homology domain. In addition to tyrosine phosphorylation, products of phosphatidylinositol 3Ј-kinase activation have been proposed to contribute to the GEF activation of Vav via the PH domain (15). Rac activation leads to a wide array of cellular responses including membrane ruffling, JNK (c-Jun N-terminal kinase) activation, and transcriptional activation from Rho/Rac-responsive elements such as NF-AT and NF-B.
Both the SH2 and SH3 domains of Vav have been shown to bind to a whole array of signaling molecules. Among them, as mentioned above, are tyrosine kinases such as Lck, JAK-2, and ZAP-70, which bind to the SH2 domain, and adapter molecules like Cbl, Grb2, and SLB70, binding to the SH3 domain. In addition, heterogeneous ribonucleoprotein (hnRNP), the focal adhesion protein zyxin, Ku-70, hnRNP-c, and hSiah2 have been identified as binding partners of Vav (reviewed in Ref. 3).
The chimeric oncogene Bcr-Abl is crucial for the pathogenesis of chronic myeloid leukemia (CML) and a subset of acute lymphoblastic leukemia (16). The dysregulated tyrosine kinase activity of Bcr-Abl has been shown responsible for the transformation of fibroblasts and hematopoietic cells and for the induction of a CML-like disease in mice (17)(18)(19)(20). Bcr-Abl contains several autophosphorylation sites, which can serve as docking sites for SH2 domain-containing signaling molecules, some of which may be critical for the transformation ability of Bcr-Abl. One of the best studied and defined interactions is the binding of Grb2 to tyrosine 177 in the Bcr part of Bcr-Abl, which links Bcr-Abl to Sos/Ras pathway (21,22). In addition, a multitude of other adapter proteins and substrates of Bcr-Abl have been identified such as Grb10, Grb4, SHC, Crk, Crkl, phosphatidylinositol 3-kinase, and STAT (signal transducers and activators of transcription), which are thought to be involved in the molecular pathogenesis of CML (23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33). Thus, screening for Bcr-Abl-interacting proteins seems an efficient method of identifying molecules important for Bcr-Abl-induced leukemia.
Recently it has been shown that Bcr-Abl expression in Mo7e cells results in constitutive phosphorylation of Vav (11). However, the authors could not detect coprecipitation of Vav and Bcr-Abl. Instead it was reported that Vav associates and is phosphorylated by JAK-2 in these cells (11). In this paper, in contrast to the previous report, we demonstrate that Vav interacts directly with Bcr-Abl. This interaction is dependent on the tyrosine kinase activity of Bcr-Abl and an unusual binding mechanism involving a proline-rich region and the two SH3 domains framing the SH2 domain of Vav. We present evidence that complex formation of Bcr-Abl and Vav is necessary for activation of the Rac pathway by Bcr-Abl, which has been shown to be crucial for Bcr-Abl-mediated leukemogenesis (34).

MATERIALS AND METHODS
Yeast Two-hybrid System-The DNA for Bcr-Abl⌬Sal was cloned into the yeast expression vector BTM116 (35). A cDNA library derived from K562 cells was used (CLONTECH, Heidelberg, Germany) as the prey. Various Bcr-Abl mutants were used in the screen as described previously (23,24).
Plasmid Construction-For in vitro translation and expression in 293 cells, the cDNAs of Bcr-Abl WT, Bcr-Abl mutants, F46Wt, and the F46 mutants were cloned in pcDNA 3.1, a mammalian expression vector containing the cytomegalovirus promoter and the SV40 origin of replication (Invitrogen). Cloning of the corresponding human F46/Vav mutant into the full-length murine Vav c-DNA generated VavR696L and VavAA⌬CSH3. Expression of these chimerical constructs was performed using the pCMV Tag II vector (Invitrogen). Tyrosine to phenylalanine mutants Bcr-Abl/115F, Bcr/177F-Abl, Bcr/246F-Abl, Bcr/246F-Abl/449F, Bcr/246F-Abl/342F, 449F, Bcr/246F-Abl/353F, 449F, and Bcr/ 177F-Abl/393 were generated by site-directed mutagenesis using overlapping oligonucleotides and Pfu DNA polymerase (Stratagene, Heidelberg, Germany). The numbers correspond to the human Bcr and murine c-Abl sequence. The Bcr-Abl temperature-sensitive mutant (TS-Bcr-Abl) was generated by subcloning a temperature-sensitive v-Abl mutant (DP) into Bcr-Abl (23). The Bcr/1-509-Abl, Bcr/1-242-Abl, and Bcr/1-63-Abl constructs have been described previously (4). Point mutations of the F46/Vav cDNA were conducted using the QuikChange mutagenesis kit following the instructions of the supplier (Stratagene), and deletion mutants were made by PCR using oligonucleotides containing EcoRI and XhoI restriction sites and standard cloning procedures. All constructs derived from site-directed mutagenesis and PCR were verified by automatic sequence analysis.
Cell Culture and Transfections-293 cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% FCS and penicillin/streptomycin (200 units of penicillin/ml and 200 g of streptomycin/ml). K562, Jurkat, Mo7e, and Mo7e/p210 cells were maintained in RPMI 1640 medium (Invitrogen) with 10% fetal calf serum (Seromed, Berlin, Germany) and penicillin/streptomycin. The medium for Mo7e was supplemented with granulocyte/macrophage colony-stimulating factor (R&D Systems DPC Bierman GmbH, Wiesbaden, Germany). SCF stimulation of Mo7e cells was performed at a final concentration of 200 ng/ml SCF at 37°C for 5 min and subsequent resuspension in ice-cold phosphate-buffered saline. Transfections were performed with the DOTAP transfection reagent (Roche Molecular Biochemicals).
GST Fusion Proteins and Pull-down Assay-The cDNAs of F46/Vav and F46/Vav mutants were cloned in-frame into the vector pGEX KG to make GST (glutathione S-transferase) fusion proteins. WT and mutant Bcr-Abl proteins were in vitro translated and 35 S-radiolabeled using the TNT system (Promega, Heidelberg, Germany). The translation mix was diluted to a final concentration of 25 mM Tris⅐HCl (pH 7.4), 10 mM MgCl 2 , 1 mM dithiothreitol, and 100 M cold ATP and incubated for 30 min at 4°C to allow for autophosphorylation of the translated proteins. Reactions were stopped by dilution to a final concentration of 10 mM Tris⅐HCl (pH 7.4), 5 mM EDTA, 130 mM NaCl, 1% Triton, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml each phenantroline, aprotinin, leupeptin, and pepstatin (lysis buffer). GST fusion proteins were added and incubated for 1 h at 4°C. Protein complexes were collected on glutathione-agarose beads (Amersham Biosciences, Inc.), washed thoroughly with NETN buffer (100 mM NaCl, 1 mM EDTA, 50 mM Tris⅐HCl (pH 7.4), 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 5 mM benzamidine), and subjected to SDS-PAGE. In vitro translated proteins were visualized by autoradiography. For binding experiments with cell extracts, 1 ϫ 10 7 cells were solubilized in lysis buffer, precleared with glutathione-beads, and incubated with the GST fusion protein for 3 h at 4°C. Protein complexes were collected on glutathione-beads, washed thoroughly with lysis buffer or NETN buffer, and subjected to SDS-PAGE. Immunoblotting was performed with the antibodies indicated.
In Vitro Kinase Assay-Bcr-Abl and kinase-defective Bcr-Abl were in vitro translated using the TNT system (Promega). The translation mix was diluted to a final concentration of 10 mM Tris⅐HCl (pH 7.4), 5 mM EDTA, 130 mM NaCl, 1% Triton, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 , and 10 g/ml each phenantroline, aprotinin, leupeptin, and pepstatin. Immunoprecipitation was performed with 2 g of anti-Abl (Ab-3). The immune complexes were collected on protein A-Sepharose beads, washed twice each with lysis buffer containing 500 mM NaCl, lysis buffer plus 100 mM NaCl, and lysis buffer alone and twice with kinase buffer containing 25 mM Tris⅐ HCl (pH 7.4), 10 mM MgCl 2 , and 1 mM dithiothreitol. The pellet was resuspended in 20 l of kinase buffer with 1 g of GST-F46/Vav as a substrate. Reactions were started by the addition of 15 M cold ATP and 10 Ci of [␥-32 P]ATP and incubated for 30 min at 30°C. Reactions were stopped by the addition of Laemmli buffer, heated at 95°C for 10 min, and resolved by SDS-PAGE.
Rac-1 Activity-293 cells (3 ϫ 10 5 ) were cotransfected in 60-mm dishes using the DOTAP transfection reagent (Roche Molecular Biochemicals) with 3 g of plasmid encoding Rac-1, 2.5 g of the plasmid expressing WT-Bcr-Abl, KD-Bcr-Abl, or the control vector, and 5 g of the dominant-negative construct Vav/F46 or Vav/F46AA⌬CSH3 or the control vector. 65 h post-transfection, cells were starved in FCS-free media and harvested 12 h later. Cell lysates were precleared with glutathione-beads and incubated with equal amounts of GST-PAK for 1 h at 4°C. Samples were separated by SDS-PAGE, and proteins were detected with a Rac-1 specific antibody (Santa Cruz). To normalize transfection efficiencies, lysates were analyzed by immunoblotting with anti-Rac-1 (Santa Cruz), anti-Abl 8E9 (BD PharMingen), and anti-Xpress ® (Invitrogen).
Generation of Retrovirus Stocks-Retroviral supernatant was produced by transiently transfecting the Phoenix ecotropic producer line (Gary Nolan, Stanford, CA) with the Mig-p210 or Mig vectors (W. Pear, Philadelphia, PA) as described previously (37). Medium was changed at 36 h after transfection and collected after 48 h. Titers of ϳ1 ϫ 10 5 -1 ϫ 10 6 helper virus-free viral particles were obtained as measured by fluorescence-activated cell sorter titer of EGFP-positive infected Rat-1 cells.
Bone Marrow Transformation Assay-Wild-type and VavϪ/Ϫ mice (4) were treated with 150 mg/kg 5-fluorouracil 4 days prior to bone marrow harvest. BM cells flushed from tibia and femur were preincubated for 1 day in bone marrow medium (IMDM, Invitrogen) supplemented with 30% FCS (Seromed), 1% bovine serum albumin, 1% glutamine, 0,5% penicillin-streptomycin, 100 M 2-mercaptoethanol, and 1 M hydrocortisone) 50 ng/ml mouse SCF, 8 ng/ml mouse IL-3, and 12 ng/ml human IL-6 (all from R&D Systems). After 1 day of preincubation, 1 ϫ 10 6 bone marrow cells were coincubated with 3 ml of viral supernatant and 1 ml of new medium in the presence of the above mentioned growth factors and 4 g/ml polybrene. New supernatant was added after 24 h. 3 days after infection, cells were washed in phosphatebuffered saline. Transformation of bone marrow cells in liquid culture was determined as described previously (39). Briefly, 1 ϫ 10 5 infected cells were plated in bone marrow medium in 12-well plates in a volume of 2 ml. Nonadherent cells were counted after 9 days.

A Vav Fragment Encoding the C-terminal SH3-SH2-SH3
Domain Interacts Specifically with Bcr-Abl in a Phosphotyrosinedependent Manner-To isolate signaling intermediates important for the oncogenic signal transduction of Bcr-Abl, a yeast two-hybrid screen was performed using Bcr-(1-509)-Abl/Lex A as a bait and a cDNA library obtained from a cell line established from a CML patient in blast crisis of the disease (K562). An identical 800-bp clone was isolated four times, which encoded the C-terminal SH3-SH2-SH3 region of p95 Vav (F46/ Vav, amino acids 560 -845) ( Table I). Co-expression of kinaseactive Bcr-Abl and F46/Vav resulted in histidine autotrophy and growth of the yeast cells, indicating interaction of Bcr-Abl with F46/Vav (Fig. 1A). The interaction of Vav/F46 and Bcr-Abl was specific, as the control protein lamin showed no binding (Fig. 1A). Yeast cells co-expressing F46/Vav and a tyrosine kinase-defective mutant of Bcr-Abl (KDBcr-Abl) failed to grow, indicating that Bcr-Abl autophosphorylation is required for this interaction (Fig. 1A). Binding of F46/Vav to Bcr-Abl in yeast suggested a direct interaction between Vav and Bcr-Abl. To confirm a direct association of Bcr-Abl and Vav/F46, we tested the ability of GST-Vav/F46 to bind to in vitro translated Bcr-Abl. Fig. 1B shows that Vav/F46 is able to directly bind to Bcr-Abl in vitro. Moreover, we found two constructs carrying internal deletions in the Bcr part of Bcr-Abl (BCR/1-242-Abl and BCR/1-63-Abl) to bind to GST-Vav/F46 (data not shown). Bcr/1-63-Abl is missing the autophosphorylation site Tyr 177 responsible for the binding of Bcr-Abl to Grb2, indicating that an Abl phosphorylation site other than Tyr 177 is responsible for this phosphotyrosine-dependent interaction. These results together with the data obtained in yeast suggested that the Vav SH2 domain binds to an autophosphorylation site located in the Abl part of Bcr-Abl, which is different from Tyr 177 . Concordant with this hypothesis, complex formation of Vav could also be demonstrated with v-Abl (Fig. 2D). One tyrosine residue within the Abl portion of Bcr-Abl resembles, at least in part, the predicted consensus site for the Vav SH2 domain, YMEP, which is tyrosine 115 with the sequence -YITP- (40). In an attempt to identify the tyrosine residue in Abl that is responsible for complex formation with Vav, a series of tyrosine to phenylalanine mutations and combinations thereof were introduced into Bcr-Abl. These mutants, including Abl Tyr 115 , Abl Tyr 393 , the major Abl autophosphorylation site, and the Grb2 binding site, Bcr Tyr 177 , were tested for their ability to bind to the Vav/F46 fragment in a GST pull-down assay (Abl numbers refer to the Abl Ia isoform). All of the tested tyrosine-to-phenylalanine mutants retained the ability to bind Vav/F46 (data not shown). Thus, the complex between Vav and Bcr-Abl may involve multiple tyrosine residues or involve other binding motifs and intermediate proteins (for example, c-Cbl). Alternatively the binding may be dependent on a so-called "open" conformation that can only be adopted by kinase-active Bcr-Abl, as shown by the paper of Schindler et al. (41).
Only Kinase-active Bcr-Abl Forms a Complex with Vav in Vivo-To demonstrate the complex of Vav and Bcr-Abl in vivo and to determine the role of tyrosine kinase activity, a temperature-sensitive (TS) mutant of Bcr-Abl was transiently expressed in 293 cells. These cells express endogenous p95 Vav. Bcr-Abl was expressed to comparable levels at the restrictive and permissive temperature in these cells ( Fig. 2A, upper panel). Shifting to the permissive temperature (32°C) for 3 h Equal amounts of GST fusion proteins of the represented Vav/F46 mutants were tested for their binding affinity to Bcr-Abl from K562 lysates of 5 ϫ 10 6 cells in a GST pull-down assay as described in Fig. 4. The amino acid numbers of the F46/Vav fragments are indicated. ϩϩϩ, strong binding; (ϩ), very weak binding; Ϫ, no binding. R696L is an SH2 mutant defective in phosphotyrosine binding. AASH3-SH2-SH3 is a proline-to-alanine mutant of four prolines (amino acids 607-610). Dbl, Dbl homology; PH, pleckstrin homology; SH, Src homology; p-rich, proline-rich. induced tyrosine kinase activity of TSBcr-Abl as anticipated ( Fig. 2A, lower panel). Cells kept at the restrictive (39°C) or shifted to the permissive temperature (32°C) were analyzed by an anti-Vav immunoprecipitation (Fig. 2B). Equal amounts of Vav were precipitated at both temperatures (Fig. 2B, upper  panel). The phosphotyrosine blot shows a highly phosphorylated protein coprecipitating with Vav only at the permissive temperature (Fig. 2B, middle panel). This protein was proven to be TSBcr-Abl by probing the blot with an anti-Abl antibody (Fig. 2B, lower panel). Only trace amounts of TSBcr-Abl were precipitated at the restrictive temperature, probably because of some leaky kinase activity at 39°C. The same results were obtained in a GST pull-down assay with GST-F46/Vav and TS Bcr-Abl (data not shown). Thus, Bcr-Abl and Vav are complexed in vivo in a phosphotyrosine-dependent manner, confirming the data obtained in yeast and in vitro.
Vav Forms a Complex with Bcr-Abl in the CML Cell Lines K562 and Lama-To demonstrate the Vav⅐Bcr-Abl complex in CML cells, co-immunoprecipitation experiments were performed using two cell lines established from patients with Bcr-Abl-positive CML and a hematopoietic cell line engineered to overexpress Bcr-Abl. The anti-Vav antibody was able to precipitate a complex of Vav and Bcr-Abl in all three cell lines (Fig. 2C). This complex formation is specific because no Bcr-Abl protein was detected using an unrelated rabbit antiserum (Fig.  2C). Similar complex formation of endogenous Vav and Bcr-Abl was observed in other cell lines expressing Bcr-Abl (data not shown).
Vav Forms a Complex with v-Abl-If complex formation of Bcr-Abl and Vav is mediated by an autophosphorylation site located in the Abl part of Bcr-Abl, a kinase-active Abl construct should also show complex formation with Vav. To test this theory we performed co-immunoprecipitation experiments with cells overexpressing Bcr-Abl, v-Abl (v-Abl is activated by deletion of the Abl SH3 domain), and Flag-tagged Vav. Both Bcr-Abl and v-Abl were co-immunoprecipitated using an antibody recognizing tagged Vav (Fig. 2D), supporting the notion that the Bcr-Abl⅐Vav complex is mediated by the Abl part of Bcr-Abl.
Vav Serves As a Direct Substrate for the Bcr-Abl Tyrosine Kinase-The isolated C-terminal tail of Vav (F46) contains several potential phosphorylation sites, among which are sites matching the consensus Abl substrate phosphorylation site (42). To determine whether complex formation between Bcr-Abl and Vav leads to phosphorylation of Vav by Bcr-Abl, we expressed the tagged C-terminal tail of Vav (F46/Vav) together with Bcr-Abl in 293 cells. An immunoprecipitation using an anti-tag antibody was performed and analyzed by anti-phosphotyrosine immunoblotting (Fig. 3A). Only two phosphorylated bands were detectable representing the immunoprecipitated Vav tail itself and complexed Bcr-Abl. To determine the phosphorylation status of endogenous Vav in response to Bcr-Abl in hematopoietic cells, we immunoprecipitated equal amounts of Vav from parental Mo7e cells and Mo7e cells overexpressing Bcr-Abl (Fig. 3B). SCF stimulation in parental Mo7e cells leads to tyrosine phosphorylation of Vav (Fig. 3B,  upper panel, second lane from right). Mo7e cells expressing Bcr-Abl show significantly enhanced levels of Vav phosphorylation compared with parental Mo7e cells maximally stimulated with SCF (Fig. 3B, compare the two rightmost lanes). Again, Bcr-Abl coprecipitated with Vav together with a phosphorylated ϳ50-kDa protein of unknown identity. The resulting complexes were specific because no phosphorylated protein could be detected using a nonspecific rabbit antiserum (Fig. 3B, left half). To reconstitute this kinase reaction, in vitro purified Vav and purified kinase-active and kinase-defective Bcr-Abl were used in an in vitro kinase reaction (Fig. 3C). Kinase-active but not kinase-defective Bcr-Abl incorporated 32 P into F46/Vav (Fig. 3C, left panel). Thus, the C-terminal tail of Vav serves as a direct substrate of Bcr-Abl in vitro.

SH2-dependent Binding of Bcr-Abl Requires Both the Proline-rich Region Upstream of the N-terminal SH3 Domain and the C-terminal SH3
Domain-To define the domains of Vav which are required for the phosphotyrosine-dependent binding to Bcr-Abl different mutants of F46/Vav were tested for their ability to form a complex with Bcr-Abl from lysates of K562 cells. Vav/F46R696L carries a point mutation in the SH2 domain replacing the arginine of the central FLVR motif with leucine, resulting in a nonfunctional SH2 domain. As anticipated, this mutant is no longer able to interact with Bcr-Abl (Fig. 4A, upper panel). However, surprisingly, the isolated Vav SH2 domain failed to bind Bcr-Abl (Fig. 4B, upper panel). Further binding studies utilizing different truncated mutants of F46/Vav as shown in Table I could delineate the region minimally required for binding to the SH3-SH2-SH3 formation including a short 4-amino acid proline repeat upstream of the N-terminal SH3 domain of Vav (amino acids 607-610). This proline repeat was absolutely essential for the complex formation of Bcr-Abl and Vav because a mutant in which the tetraproline motif was replaced by alanines no longer bound Bcr-Abl (AASH3-SH2-SH3, Table I). In addition, deletion of the Cterminal SH3 domains rendered F46/Vav equally unable to bind Bcr-Abl (PPSH3-SH2, Table I). These results indicated that in addition to the Vav SH2 domain, the C-terminal SH3 domain and the proline repeat play an equally important role. The Abl SH3 domain is not able to bind to the Vav proline-rich region, thus excluding an Abl SH3-mediated binding to Vav (data not shown). However, the isolated C-terminal SH3 domain of Vav is capable to bind the F46/Vav fragment (F46⌬SH3) in a pull-down assay (Fig. 4C). This binding is mediated by the proline repeat, because a F46/Vav mutant in which the prolines have been replaced by alanines (F46AA⌬SH3) no longer was complexed by the Vav SH3 domain (Fig. 4C). This finding supports the hypothesis of an intramolecular folding mechanism involving the C-terminal SH3 domain and the short proline-rich domain, which enables the SH2 domain to develop its full binding potential. Similarly stronger binding of Vav to ZAP-70 and to other tyrosine phosphorylated proteins required the proline-rich region upstream of the N-terminal SH3 domain (data not shown). Thus, the general binding capacity of the Vav SH2 domain is influenced by the adjacent SH3 domains and the proline repeat.
Mutational Analysis of the in Vivo Complex Formation between Vav and Bcr-Abl-To determine whether the SH3 domains and the proline repeat are also required for the in vivo complex formation between Vav and Bcr-Abl, we introduced the mutations into full-length Vav. Co-expression of WT Vav with kinase-active and -inactive Bcr-Abl confirmed the phosphotyrosine dependence of the interaction (Fig. 5, first two  lanes). Mutation of the Vav-SH2 domain (VavR696L) or deletion of the C-terminal SH3 domain and mutation of the prolines (VavAA⌬SH3) resulted in abrogation of the complex formation between Bcr-Abl and Vav in vivo (Fig. 5, last two lanes).
The C-terminal Vav Tail Has a Dominant-negative Effect on Bcr-Abl-induced Rac Activation-Bcr-Abl has been reported to activate Rac (34). Vav is a GEF for Rac. Therefore, it is possible that Bcr-Abl-induced Rac activation is mediated by binding and phosphorylation of Vav through Bcr-Abl. To test this hypothesis we employed a pull-down assay for GTP-bound Rac using the p21-binding domain of p21-activated kinase 1, which has been shown to specifically bind Rac-GTP (43). With this assay the amount of precipitated GTP-Rac in a cell can easily be quantified by anti-Rac immunoblotting.
Bcr-Abl expression in 293 cells leads to Rac activation (Fig.  6). Co-expression of the C-terminal Vav fragment resulted in inhibition of Bcr-Abl induced Rac activation, suggesting a dominant-negative function of the Vav tail. Coexpression of a Cterminal fragment incapable to bind Bcr-Abl had no dominant-negative effect. These data indicate that Bcr-Abl-induced Rac activation is mediated by the Bcr-Abl⅐Vav complex and probably by Vav phosphorylation because a kinase-defective Bcr-Abl could not activate Rac (Fig. 6) (34).
Bcr-Abl Utilizes Vav to Transform Primary Bone Marrow Cells-Because the isolated C-terminal tail of Vav showed dominant-negative activity on Bcr-Abl-induced Rac activation and Rac activation has been shown to be important for the oncogenic potential of Bcr-Abl, we wished to determine the importance of Vav for Bcr-Abl-induced mitogenicity. We first investigated whether Vav1 overexpression in fibroblasts could contribute to the transformation potential of Bcr-Abl in NIH3T3 fibroblasts. The transformation property was scored by anchorage-independent growth in soft agar. In the absence of Bcr-Abl, neither the control vector nor Vav gave rise to morphologically transformed foci. Retroviral infection of Bcr-Abl induced outgrowth of soft agar colonies in both the control vector and two different Vav-expressing stable NIH3T3 cell lines. No significant difference in colony number and size was observed between these entities (Fig. 7). Next we used bone marrow from wild-type and Vav knock-out mice, which was retrovirally transduced with Bcr-Abl, to perform a bone marrow transformation assay as previously described (39). Primary cells were retrovirally transformed with Bcr-Abl, and the outgrowth of cells was measured. When analyzed for EGFP fluorescence, the liquid cultures of Bcr-Abl-transformed bone marrow cells of both the Vav-deficient mice as well as the wild-type control mice showed a much higher percentage of fluorescent cells than the mock-EGFP-infected bone marrow cultures (data not shown). Bcr-Abl induces the outgrowth of transformed cells in wild-type bone marrow (Fig. 8, left panel). However, in Vav knock-out bone marrow, Bcr-Abl-induced bone marrow growth is severely impaired (Fig. 8). Both control wild-type and Vav knock-out bone marrow showed comparable growth after the addition of IL-3 (Fig. 8). Thus, although Bcr-Abl also mediates a proliferative advantage in Vav-deficient bone marrow cells, efficient transformation of primary bone marrow cells by Bcr-Abl requires Vav. DISCUSSION We report here that p95 Vav interacts directly with Bcr-Abl in yeast, both in vitro and in vivo. This interaction was identified in a yeast two-hybrid screen using Bcr-Abl as bait and a cDNA library from the CML cell line K562 as prey. An 800-bp clone was isolated that interacted with Bcr-Abl and encoded for the C-terminal SH3-SH2-SH3 region of p95 Vav. Deletion mutants show that the interaction occurs at an Abl autophosphorylation site within Bcr-Abl. Although the Abl sequence within Bcr-Abl contained a consensus binding site for the Vav SH2 domain, mutation of that site and combinations thereof did not abolish binding suggesting that additional domains may be involved. An Abl-SH2-pTyr interaction contributing to the complex is possible because the Vav C terminus contains several Abl-SH2 consensus binding sites and is heavily phosphorylated (44). A Vav GST fusion protein encoding this region was shown to bind strongly to Bcr-Abl but not c-Abl using whole cell lysates of Bcr-Abl-expressing K562 cells. This probably reflects the poor autophosphorylation status of c-Abl in normally cycling cells. A recent study solved the crystal structure of the Abl kinase domain binding to the inhibitor STI571 (41). Whereas kinase-active Abl displays an open conformation of the kinase domain, a pseudo-substrate loop is observed in kinase-inactive Abl. Thus, binding of Vav to Bcr-Abl may depend on a conformation that can only be adopted by the active Bcr-Abl.
Co-immunoprecipitation studies clearly showed that Vav forms a complex with Bcr-Abl in the CML cell line K562 and LAMA cells, as well as in Bcr-Abl-expressing Mo7e/p210 cells, and is constitutively phosphorylated. A previous paper also reported Vav phosphorylation in Mo7E cells transfected with Bcr-Abl (11); in this paper, however, the authors did not detect complex formation of Bcr-Abl and Vav. Therefore, the authors (11) speculated that Vav may not be a direct substrate of Bcr-Abl but is phosphorylated by other tyrosine kinases in these cells. Here, we clearly show that Bcr-Abl and Vav form a complex in different experimental systems including in vitro purified proteins, yeast cells, and mammalian cells. We also show complex formation of Vav and Bcr-Abl by coprecipitating Bcr-Abl with a Vav specific antibody. However, we were not able to coprecipitate Vav with a Bcr-Abl-specific antibody. This may be due to the fact that the Abl antibody interferes with the binding site for Vav on Bcr-Abl. In addition, the Vav antibody used in the present study detects the native protein very well but the denatured protein only poorly. Thus, the differences between their study (11) and ours may be because of the Vav antibody used to coprecipitate Bcr-Abl.
We show that this interaction is strictly dependent upon the tyrosine kinase activity of Bcr-Abl. Concurring with the yeast data showing a tyrosine-dependent interaction of the Vav fragment with Bcr-Abl, a TS mutant of Bcr-Abl showed detectable complex formation with and phosphorylation of Vav only at the permissive temperature for the TSBcr-Abl tyrosine kinase. We identified the region that mediates this interaction as the Cterminal SH3-SH2-SH3 domain of Vav. Even though mutational studies showed that the interaction is entirely phospho- tyrosine-dependent and requires the intact SH2 domain of Vav, the C-terminal SH3 domain as well as a proline repeat next to the N-terminal SH3 domain of Vav are required for complex formation between Vav and Bcr-Abl; the Vav SH2 domain alone was not able to bind to Bcr-Abl. The deletion of the C-terminal SH3 domain as well as the replacement of the four adjacent prolines with alanine completely abolished the phosphotyrosine-dependent binding of the C-terminal Vav fragment. The C-terminal SH3 domain or the N-terminal SH3 domain alone does not bind to Bcr-Abl.
Additionally, we found that the C-terminal SH3 domain binds with high affinity to the proline-rich stretch adjacent to the N-terminal SH3 domain. Therefore a model is possible in which an intramolecular folding mechanism involves the Cterminal SH3 domain and the short spanning proline-rich domain next to the N-terminal SH3 domain within Vav. This intramolecular folding then enables the SH2 domain to exert its full binding potential, leading to complex formation between the C terminus of Vav and Bcr-Abl. Alternatively, our data would also support a model in which a head-to-tail dimerization process of two Vav molecules results in better accessibility of the Vav SH2 domain. However, by running protein gels under native conditions no experimental evidence of Vav dimerization could be obtained. 2 It is very likely that many direct phosphotyrosine-mediated interactions that involve Vav have been undetected because the binding ability of the single SH2 domain, often used for binding studies, is greatly reduced. In a pull-down experiment we found a panel of tyrosine-phosphorylated proteins precipitating only with the pp-SH2-SH3-SH3 fragment but not with the isolated SH2 domain (data not shown). In addition, one of the known Vav-SH2-binding proteins was precipitated with much higher affinity by the Cterminal Vav fragment than by the isolated SH2 domain. A 3-4-fold increase in binding to ZAP-70 from CD3-stimulated Jurkat cells was observed when both the proline-rich region adjacent to the SH3N and the SH3C were present compared with the binding of the SH2 domain alone (data not shown). In fact, other proteins have been reported recently to bind to Vav in a phosphotyrosine-dependent manner requiring the SH3-SH2-SH3 configuration (45,46). However, these authors did not explore the contribution of the proline-rich stretch in Vav to this complex.
We show that Vav itself and the Vav fragment are phosphorylated in cells expressing Bcr-Abl. Phosphorylation of Vav strictly depends on the activity of the Bcr-Abl tyrosine kinase as demonstrated by use of a temperature-sensitive mutant of Bcr-Abl. In vitro kinase reactions with purified Bcr-Abl and Vav proteins indicate that Vav is a direct substrate of the Bcr-Abl kinase. The N-terminal Vav fragment contains several potential tyrosine phosphorylation sites, among them a tyrosine residue matching the predicted Abl substrate consensus site (42). Alternatively, Vav could also be phosphorylated by a different kinase in which activation strictly depends on kinaseactive Bcr-Abl. For example the Src family kinases were shown to be able to phosphorylate Vav and to be activated by Bcr-Abl (47).
Vav phosphorylation by LCK has been shown to activate its GEF activity. Much effort has been made to identify the tyrosine residues in Vav that are responsible for the activation of its GEF activity. Most phosphorylation sites characterized so far are located in the N-terminal part of Vav. Tyr 174 is phosphorylated by Lck in vitro and in vivo and is suggested to be important for the activation of Vav GEF activity. However, it was recently demonstrated that mutation of this site and mutation of additional tyrosine residues in the N terminus leads to oncogenic activation rather than inactivation of Vav (38). Thus, the mechanism by which Vav is activated by phosphorylation is still enigmatic and may involve additional phosphorylation sites including sites present in the C terminus. It will be interesting to determine the Vav tyrosine residues phosphorylated in Bcr-Abl-expressing cells and mutational studies thereof.
Rac activation by Bcr-Abl was completely abrogated by coexpression of the Vav C terminus encoding the SH3-SH2-SH3 domains as a dominant-negative construct. However, expression of a construct lacking the C-terminal SH3 domain and bearing the proline to alanine mutation had no effect on Rac activation. Thus, Bcr-Abl utilizes Vav to activate Rac-1. Rac activation has been shown to be crucial for Bcr-Abl mediated leukemogenesis (34). Bcr-Abl transforms cell lines of hematopoietic origin more effectively than for example fibroblasts. Vav is a GEF for Rac and is predominantly expressed in hematopoietic cells. Hence it is attractive to speculate that Bcr-Abl utilizes Vav to activate Rac-1 and therefore transforms hematopoietic cells more efficiently. This notion is supported by our data employing Vav knock-out bone marrow. To elucidate the contribution of Vav to Bcr-Abl induced leukemia it will be interesting to determine the leukemic potential of mice reconstituted with Bcr-Abl transduced WT compared with Ϫ/ϪVav bone marrow.