Mena [mammalian Ena (Enabled)]/VASP (vasodilator-stimulated phosphoprotein) proteins are the homologues of Drosophila Ena. In Drosophila, Ena is a substrate of the tyrosine kinase DAbl (Drosophila Abl). However, the link between Abl and the Mena/VASP family is not fully understood in mammals. We previously reported that Abi-1 (Abl interactor 1) promotes phosphorylation of Mena and BCAP (B-cell adaptor for phosphoinositide 3-kinase) by bridging the interaction between c-Abl and the substrate. In the present study we have identified VASP, another member of the Mena/VASP family, as an Abi-1-bridged substrate of Abl. VASP is phosphorylated by Abl when Abi-1 is co-expressed. We also found that VASP interacted with Abi-1 both in vitro and in vivo. VASP was tyrosine-phosphorylated in Bcr-Abl-positive leukaemic cells in an Abi-1-dependent manner. Co-expression of c-Abl and Abi-1 or the phosphomimetic Y39D mutation in VASP resulted in less accumulation of VASP at focal adhesions. VASP Y39D had a reduced affinity to the proline-rich region of zyxin. Interestingly, overexpression of both phosphomimetic and unphosphorylated forms of VASP, but not wild-type VASP, impaired adhesion of K562 cells to fibronectin. These results suggest that the phosphorylation and dephosphorylation cycle of VASP by the Abi-1-bridged mechanism regulates association of VASP with focal adhesions, which may regulate adhesion of Bcr-Abl-transformed leukaemic cells.
The protein tyrosine kinase c-Abl was initially identified as a homologue of v-Abl which is encoded by A-MuLV (Abelson murine leukaemia virus) [1,2]. c-Abl contributes to many cellular events such as cell growth, stress response, apoptosis, cytoskeleton reorganization and transformation [3–5]. Chromosomal translocation generates a chimaeric Bcr-Abl protein which has an increased kinase activity and causes CML (chronic myelogenous leukaemia) [6–9]. Although autoinhibitory interactions between the N-terminal domain structures regulate the kinase activity of c-Abl [10–12], much remains unknown about how the activity and the substrate specificity of c-Abl are controlled in the cell. Elucidation of the regulatory mechanism of c-Abl is crucial to understand the signal transduction pathways downstream of the normal and oncogenic Abl kinases.
In Drosophila, the reduction of gene dosage of Ena (Enabled)  and Abi (Abl interactor)  suppresses the lethal phenotype of the loss-of-function mutation of DAbl (Drosophila Abl), suggesting that Ena and Abi-1 function as genetic antagonists of DAbl. In contrast, phosphorylation of Ena by Abl is necessary to rescue the lethal phenotype of the loss-of-function mutation of Ena . Phosphorylation of Ena is stimulated by overexpression of Abi . These findings indicate that Abi and Ena function co-operatively with DAbl. These discrepancies may be explained by a model in which accumulation of the unphosphorylated Ena results in the defects in Drosophila axonogenesis . Alternatively, accumulation of Ena defective in the phosphorylation and dephosphorylation cycle may be toxic to axonogenesis. The biochemical mechanism of how phosphorylation of Ena functions during the development of Drosophila remains largely unknown.
The Mena (mammalian Ena)/VASP (vasodilator-stimulated phosphoprotein) family of proteins are the homologue of Ena in mammals. The Mena/VASP family consists of Mena, VASP and EVL (Ena-VASP-like) [17,18]. VASP is a substrate of cyclic-nucleotide-dependent serine/threonine protein kinases PKA (protein kinase A) and PKG (protein kinase B) . Mena/VASP proteins localize to the tips of filopodia and lamellipodia and at focal adhesions [20,21]. The Mena/VASP family share conserved domain organization consisting of the N-terminal EVH (Ena/VASP homology) 1 domain, the central PRD (proline-rich domain) and the C-terminal EVH2 domain. In one study, cell adhesion-dependent co-immunoprecipitation of VASP and Abl was reported . However, the biochemical mechanism by which Abl regulates the function of VASP also remains to be elucidated in mammals.
Our previous studies showed that Abi-1 mediates the interaction between c-Abl and Mena  or BCAP [B-cell adaptor for PI3K (phosphoinositide 3-kinase)] . These Abi-1-mediated interactions promote the phosphorylation of Mena and BCAP by c-Abl. Others also reported that Cdc2 (cell division cycle 2)  and WAVE2 (Wiskott–Aldrich syndrome protein family member 2) [26,27] are Abi-1-dependent substrates of c-Abl. In addition, phosphorylation of Abi-1 by c-Abl enhances phosphorylation of Mena through the activation c-Abl kinase . Therefore Abi-1 might function as a unique adaptor protein which bridges the interaction of c-Abl to its various substrates. The Abi family consist of Abi-1, Abi-2  and Abi-3 (also known as NESH)  in mammals. Among the three isoforms Abi-1 and Abi-2, but not Abi-3, promote the phosphorylation of Mena and WAVE2 by c-Abl . Although the overexpression of a unphosphorylated form of WAVE2 impairs PDGF (platelet-derived growth factor)-induced membrane ruffling and microspike formation in fibroblast cells , the physiological significance of tyrosine phosphorylation of Mena remains unknown.
In CML, leukaemic cells infiltrate into liver, kidney and spleen, which results in the dysfunction of these organs. These phenomena are considered to arise from the altered leukaemic cell adhesion to the extracellular matrix or bone marrow stroma through abnormal integrin functions [32,33]. Expression of Bcr-Abl often rises before progression to accelerated phase or blast crisis in CML [34,35]. In an in vitro model using a Bcr-Abl-transformed 32D myeloblast-like cell line, the adhesion capacity to fibronectin increases with the rise of the amount of expression of Bcr-Abl . The accumulating evidence thus implies that the progression of CML may closely be related to the alteration of the cell adhesion of leukaemic cells.
Abi-1-mediated signal transduction pathway is also critical for cell-adhesion-related pathogenesis in disease models of Bcr-Abl-induced CML. Expression of an Abi-1 mutant defective in binding to Bcr-Abl reduces the adhesion to fibronectin in Bcr-Abl-transformed Ba/F3 cells , an immortalized pro-B-cell line. Knock down of Abi-1 in Bcr-Abl-transformed cells diminishes the leukaemic cell expansion and enlargement of spleen in vivo, leading to the prolonged survival in NOD (non-obese diabetic)/SCID (severe combined immunodeficiency) mice . These observations, together with the findings of Abi-1-mediated phosphorylation of various proteins, raise the possibility that Abi-1 might contribute to progression of CML by mediating the phosphorylation of cytoskeletal regulatory molecules by Bcr-Abl.
In the present study, we tested whether another member of the Mena/VASP family is phosphorylated by Abl in a manner similar to Mena. We found that VASP is tyrosine-phosphorylated in leukaemic cells in an Abi-1-dependent manner. The results of the present study are consistent with a phosphoproteome study which reported a list of phosphoproteins in several CML cell lines . Cdc2 and BCAP, the known Abi-1-dependent substrates of Abl, are listed as tyrosine-phosphorylated proteins. In addition, VASP is listed among the proteins commonly tyrosine-phosphorylated in Bcr-Abl-transformed CML cell lines. We further identified the domain structures in Abi-1 and VASP required for the interaction of the two proteins. Moreover, we show evidence that tyrosine phosphorylation of VASP regulates its association with a focal-adhesion protein, zyxin, and plays a role in leukaemic cell adhesion. The present study thus identifies VASP as a substrate for Abi-1-mediated tyrosine phosphorylation by Bcr-Abl and c-Abl, and reveals a role of tyrosine phosphorylation in VASP.
The coding regions of c-Abl and Bcr-Abl (p210) cDNAs were introduced into the pcDNA3 (Invitrogen) mammalian expression vector. The expression vector for c-Abl with mRFP1 (monomer red fluorescent protein 1) at the C-terminus (c-Abl–mRFP1) has been described previously . The expression vector for the c-Arg (Abl-related gene) , pFLAG-CMV-6c-Abi-1, -2, -3 and deletion mutants of Abi-1 [23,31], have been described previously. The coding region of Abi-1 cDNA linked with StrepII and His8 tags were subcloned into pCXGFP  to generate pCXGFP-SH-Abi-1. Xenopus Abi-1 cDNA (Open Biosystems, BC081178) was fused with the mitochondria targeting signal  and subcloned into the CMV (cytomegalovirus) promoter-driven mRFP1 fusion expression vector . Human VASP cDNA was amplified by PCR from Marathon-ready human bone marrow cDNA (Clontech), and subcloned into pGXGFP  to generate a mammalian expression vector for VASP fused with a GST (glutathione transferase) tag. An expression vector for StrepII–His–VASP was also constructed. Deletion mutants encoding VASP EVH1 (amino acids 1–161), VASP ΔEVH1 (amino acids 111–380), ΔPRD (amino acids 1–116, 225–380), EVH2 (amino acids 225–380) and CC (coiled-coil) (amino acids 336–380) were generated using PCR and subcloned into pGXGFP. Site-directed mutagenesis was carried out using a Quikchange® site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. EGFP [enhanced GFP (green fluorescent protein)] cDNA was subcloned into the BamHI sites of pCX4puro, generating pCX4puro-GFP. Full-length Abi-1, VASP and its mutants were then subcloned into pCX4puro-GFP. GFP-tagged Xenopus VASP (Open Biosystems, BC072836) has been described previously . cDNAs encoding human Abi-1, human VASP and human vinculin (Open Biosystems, BC039174) were fused with mPlum cDNA and subcloned into expression vectors. Mutations, deletions and reading frames in all plasmids were verified by nucleotide sequencing.
Western blot analysis
Methods used for transfection and Western blot analysis have been described previously . In some experiments, the band intensity was quantified by chemiluminescence signals using ECL (enhanced chemiluminescence) plus (GE Healthcare) and LAS4000 (GE Healthcare). The following antibodies were used in Western blot analysis: anti-c-Abl (8E9, Pharmingen), anti-Arg (C-20, Santa Cruz Biotechnology), anti-Abi-1 (1G9) , rabbit anti-VASP (9A2, Cell Signaling Technology), mouse anti-VASP (IE273, Alexis Biochemicals), anti-GST (B-14, Santa Cruz Biotechnology), anti-FLAG M2 (Sigma), anti-phosphotyrosine (4G10, Millipore), anti-β-actin (AC-74, Sigma) and anti-GFP antibody (598, MBL).
HEK (human embryonic kidney)-293T cells, Plat-E cells and NIH 3T3 cells were maintained in DMEM (Dulbecco's modified Eagle's medium) containing 10% FBS (fetal bovine serum). K562 cells and Meg-01 cells were maintained in RPMI-1640 medium. CHO (Chinese-hamster ovary) cells were maintained in CHO-SFM II (Gibco) supplemented with 0.5 μg/ml amphotericin B (Promo Cell). Xenopus laevis XTC cells were maintained in 70% Leibovitz's L15 medium containing 10% FBS in the dark.
Analysis of tyrosine phosphorylation and co-immunoprecipitation of endogenous VASP
Immunoprecipitation of VASP was performed using the Crosslink IP Kit (Pierce) according to the manufacturer's instructions. Briefly, K562 cells and Meg-01 cells were harvested and lysed in lysis buffer [1% Triton X-100, 25 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1 mM sodium vanadate and protease inhibitor cocktail (Nakalai Tesque)]. The lysates were clarified by centrifugation at 12000 g for 15 min at 4°C and the supernatant containing 1 mg of cell lysates was first mixed with non-cross-linked agarose beads. The supernatant was then immunoprecipitated with the beads cross-linked with the anti-VASP antibody (IE273) or mouse normal IgG (sc-2025, Santa Cruz Biotechnology) at 4°C overnight or for 2 h. Bound proteins were eluted from the beads with elution buffer supplied in the kit and then these were subjected to Western blotting. Can Get Signal Immunoreaction Enhancer Solution (Toyobo) was used to enhance signals for the detection of phosphorylation of endogenous VASP.
In some experiments, siRNA (small interfering RNA) targeted to Abi-1 (Flexitube SI02655338, Qiagen) and VASP (Flexitube SI00051352 and SI02664193, Qiagen) at 100 nM was introduced into K562 cells using the Neon transfection system (Invitrogen). Imatinib was purchased from LC Laboratories.
Protein expression and purification
The coding region of full-length VASP cDNA was subcloned into pET30a (Clontech) to express VASP tagged with His6 at the N-terminus. Full-length Abi-1 and the proline-rich motif of human zyxin encoding amino acids 59–140 (zyxin FP4) were cloned into pGEX6P-1. BL21 DE3 cells (Invitrogen) were used as a host. Purified proteins were dialysed against a buffer containing 20 mM Tris/HCl (pH 7.5), 150 mM NaCl and 1 mM DTT (dithiothreitol). The protein concentration was measured by densitometric analysis of the CBB (Coomassie Brilliant Blue) stained gel using BSA (Sigma) as a standard.
Interaction between VASP and the proline-rich motif of zyxin
Wild-type His–VASP or VASP Y39D (50 pmol) was mixed with 200 pmol of GST or increasing amounts of GST–zyxin FP4 (50, 100 and 200 μmol) in 300 μl of binding buffer [20 mM Tris/HCl (pH 7.5), 300 mM NaCl, 5% glycerol and 0.1% Triton X-100]. After 1 h incubation at 4°C, bound proteins were precipitated using GSH beads and washed three times with binding buffer. Precipitated proteins were separated by SDS/PAGE and stained with CBB. Densitometric analysis was carried out using ImageJ software. The areas for background were placed in each lane.
Protein–protein interactions using AlphaScreen assay
Tyrosine-phosphorylated or unphosphorylated VASP was purified in HEK-293T cells as follows. StrepII-His-tagged VASP (SH–VASP) and FLAG–Abi-1 were expressed in cells either with c-Abl or KD (kinase-deficient) c-Abl. At 2 days after transfection, the cell lysates were collected and SH–VASP was affinity-purified using Strep-Tactin Sepharose (IBA).
StrepII-His-tagged VASP (2.5 pmol) was mixed with 0.5 μg of nickel chelate donor beads (PerkinElmer) in 10 μl of binding buffer [25 mM Hepes/NaOH (pH 7.5), 150 mM NaCl, 0.1% Triton X-100, 0.1% BSA and 0.5 mM sodium orthovanadate]. Separately, 5 ng of a rabbit anti-GST antibody (ab9085, Abcam) and various concentrations of GST–Abi-1 (0–0.25 pmol) were mixed with 0.5 μg of Protein A donor beads in 15 μl of binding buffer. The mixtures were incubated for 30 min at 22°C. Donor and acceptor beads were then mixed in a 384-well Optiplate (PerkinElmer) and further incubated for 1 h at 22°C. The emission signal was monitored using an Envision luminescence detector (PerkinElmer).
Production of recombinant retrovirus
Recombinant retrovirus was generated from Plat-E cells . pGP (TaKaRa Bio) encoding the viral components gag and pol, and pE-Eco (TaKaRa Bio) encoding env, were co-transfected with the series of pCX retrovirus vectors encoding VASP or Abi-1 into Plat-E cells. At 24 h after transfection, the medium was changed and further cultivated for 24 h. The culture supernatant containing recombinant retrovirus was exposed to target cells with the range 0.1–0.3 MOI (multiplicity of infection).
Generation of K562 cells and NIH 3T3 cells stably expressing VASP and its mutants
For susceptiblity to retrovirus infection, EcoR (ecotropic retrovirus receptor) was stably transduced into K562 cells. pCX4bsr-EcoR was transiently transfected into K562 cells by DMRIE-C reagent (Invitrogen). At 1 day after transfection, K562 cells were transduced with the retroviral vector pCX4bsr-EcoR and selected on 5 μg/ml blasticidin S (Calbiochem) for 1 week, yielding K562-EcoR cells. For measurement of the growth rate, cells were seeded on to a 24 well plate at a density of 1×105 cells per well. The growth rate and cell adhesion to fibronectin were not affected by the expression of EcoR.
Recombinant retroviruses were generated by transfecting Plat-E cells with the plasmid pCX4puro-GFP-VASP WT (wild-type), pCX4puro-GFP-VASP Y39F and pCX4puro-GFP-VASP Y39D and selected on puromycin (Calbiochem) for 1 week. We kept these cells without clonal selection.
Generation of CHO cells stably expressing StrepII-His-tagged Abi-1
CHO-S EcoR cells, which are susceptible to retrovirus infection because of the expression of EcoR, have been established previously . CHO-S EcoR cells were infected with the retrovirus encoding SH-Abi-1-IRES (internal ribosome entry site)-GFP and cultured for 1 week. GFP-positive cells were sorted by flow cytometry (FACS Vantage SE; BD Biosciences) and cultured for a further week. More than 98% of sorted cells were positive for GFP by flow cytometry (FACSCalibur; BD Biosciences).
CHO-S cells expressing SH–Abi-1 were cultured in 100 ml of CHO-SFMII. Cells were collected by centrifugation (200 g for 2 min at 23°C) and washed three times with PBS, and lysed in lysis buffer [1% Triton X-100, 25 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1 mM DTT, 1 mM sodium vanadate, 20 μg/ml aprotinin and 1 mM PMSF]. Lysate was centrifuged at 12000 g for 20 min at 4°C, and then filtrated through a 0.22 μm chromatodisk. Then 500 μl of lysates containing 5 mg of protein were applied to a Superose 6 10/300 GL gel-filtration column (GE healthcare) equilibrated with a buffer [25 mM Tris/HCl (pH 7.5), 150 mM NaCl and 1 mM DTT] at a flow rate of 0.5 ml/min. The relative molecular mass of the elution profile was estimated by comparison with molecular mass standards (Sigma). The following molecular mass standards were used: thyroglobulin (669 kDa), apoferritin (440 kDa), β-amylase (200 kDa), bovine γ-globulin (150 kDa) and BSA (66 kDa).
For gel-filtration analysis of the Abi-1-associated protein, SH–Abi-1 was first collected using Ni–Sepharose 6 Fast Flow (GE Healthcare) from 50 mg of total proteins in lysis buffer containing 15 mM imidazol for 30 min at 4°C under gentle rotation. After incubation, beads were washed three times with lysis buffer and proteins were eluted in elution buffer [25 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1 mM DTT and 500 mM imidazol] and applied to a Superose 6 column at a flow rate of 0.5 ml/min.
Live cell imaging
Observation of Xenopus laevis XTC cells was carried out as described previously [39,43,45]. Briefly, prior to observation, XTC cells were trypsinized and seeded on to a poly-L-lysine (1 mg/ml)-coated glass coverslip mounted on a flow cell in 70% Leibovitz's L15 medium (Invitrogen) without serum for 30 min. The flow cell was placed on to the stage of an Olympus BX51 microscope equipped with a Cascade II:512 cooled charge-coupled device camera (Roper Scientific). Fluorescence images were acquired at 21–23°C using Metamorph software (Molecular Devices).
Localization of VASP in NIH 3T3 fibroblasts
NIH 3T3 cells were seeded on to fibronectin (25 μg/ml)-coated coverslips and incubated. Plasmids encoding c-Abl, FLAG–Abi-1 and mPlum–vinculin were transiently transfected and further incubated for 24 h. Cells were fixed with 3.7% PFA (paraformaldehyde) in a cytoskeleton buffer [CB; 10 mM Mes (pH 6.1), 90 mM KCl, 3 mM MgCl2, 2 mM EGTA and 0.16 M sucrose] for 20 min at room temperature (23°C). Fluorescence images were acquired using an Olympus BX51 epifluorescence microscope with a PlanApo 60× 1.45 numerical aperture, oil-immersion objective lens.
Quantification of VASP at focal adhesions
The focal adhesion area was determined using Metamorph as described below (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/441/bj4410889add.htm). (i) Focal adhesion areas were marked with mPlum–vinculin. The mPlum–vinculin image was first processed using the open-close command in the morphology filter with parameters of circle, 5 pixel diameter, to obtain a blurred image. (ii) The blurred image was then subtracted from the original image using the arithmetic command. (iii) The image was further processed with the binary operation command to obtain binarized images. (iv) Pixel noise was erased using the Erode command and the following parameters: neighbourhood 8 pixels, repeat count 1. (v) Non-specific signals, which appeared around the contour of nuclei, were manually erased. (vi) Each pixel intensity in the original GFP–VASP image was multiplied by that in the binary mPlum–vinculin image (v). This yields the image with the original fluorescence intensity only of the focal adhesion area. The ratio of GFP–VASP in focal adhesions and entire cell areas was then calculated.
Flat-bottomed 96-well microtitre plates were coated with 10 μg/ml fibronectin (Sigma) or 1% BSA for 2 h at room temperature. The wells were washed twice with RPMI 1640 medium (Nikken Bio) before use. K562 cells were seeded at a density of 5×104 cells per well in 100 μl of RPMI 1640 medium, and then incubated for 0.5, 1, 4 and 16 h at 37°C. Unattached cells were removed by washing with RPMI 1640 three times using an automatic dispenser (EDR-24LS; BioTec). In each wash step, the medium was aspirated from wells at a flow rate of 200 μl/s and refilled with 200 μl of the medium at a flow rate of 50 μl/s. Residual attached cells were fixed with 4% PFA in PBS and were stained with 0.1% Crystal Violet for 10 min. Crystal Violet was extracted with 1% SDS in water overnight, and the absorbance was measured at 595 nm using an Envision microplate reader.
Abi-1 and Abi-2 increase tyrosine phosphorylation of VASP by Bcr-Abl and c-Abl
We previously reported that Mena is phosphorylated by c-Abl in a manner dependent on the co-expression of Abi-1 . Mass spectrometric analysis has identified VASP, but not Mena, as a commonly tyrosine-phosphorylated protein in several Bcr-Abl-transformed leukaemia cell lines . Abi-1 and its interaction with Bcr-Abl play an important role in Bcr-Abl-induced transformation [36,37,46]. These findings prompted us to examine tyrosine phosphorylation of VASP by c-Abl or Bcr-Abl, and the contribution of Abi-1 to this phosphorylation. We overexpressed Abi-1, together with VASP and c-Abl, and observed the state of tyrosine phosphorylation in VASP. Phosphorylation of VASP was not detectable when co-expressed with c-Abl or FLAG–Abi-1 alone (Figure 1A, lanes 2 and 3). However, phosphorylation of VASP was observed when co-expressed with a combination of c-Abl and Abi-1 (Figure 1A, lane 4).
Abi-1 and Abi-2 increase tyrosine phosphorylation of VASP by Bcr-Abl and c-Abl
We also analysed whether Abi-1 promotes tyrosine phosphorylation of VASP by Bcr-Abl. Expression of Bcr-Abl alone weakly induced phosphorylation of VASP, whereas co-expression of Abi-1 with Bcr-Abl markedly increased tyrosine phosphorylation of VASP (Figure 1A, lanes 5 and 6). Activation of PKA induced a bandshift of VASP without inhibiting tyrosine phosphorylation of VASP by Bcr-Abl (Supplementary Figure S2 at http://www.BiochemJ.org/bj/441/bj4410889add.htm). We next examined whether VASP is phosphorylated by Arg, another member of the Abl family . c-Arg co-expressed with Abi-1 failed to phosphorylate VASP to a comparable degree with c-Abl with Abi-1 (Figure 1B), indicating that VASP is preferentially phosphorylated by Abl in the presence of Abi-1. We next examined the effect of other members of the Abi family proteins on phosphorylation of VASP by Abl. Abi-1 and Abi-2, but not Abi-3, promoted the phosphorylation of VASP by c-Abl, as well as by Bcr-Abl (Figure 1C). These results indicate that Abi-1 and Abi-2 mediate tyrosine phosphorylation of VASP by Bcr-Abl and c-Abl, as is the case with the phosphorylation of Mena or WAVE2 by c-Abl [23,31].
Domain analysis of the interaction between VASP and Abi-1
Our previous study reported that Abi-1 increases tyrosine phosphorylation of Mena  and BCAP  by interacting with both Abl and the substrate. Binding of Abi and the substrate is important for phosphorylation by c-Abl. Therefore we examined whether or not VASP binds Abi-1. GST or GST–VASP was co-expressed with FLAG–Abi-1 in HEK-293T cells, and the association of FLAG–Abi-1 with GST–VASP was examined. Co-precipitation of FLAG–Abi-1 with GST–VASP, but not GST, was observed (Figure 2B, lanes 1 and 2). To analyse the binding region of VASP to Abi-1, we created deletion mutants of VASP (Figure 2A). Abi-1 interacted with two distinct regions of VASP, the EVH1 and the CC domains. Abi-1 was precipitated more efficiently by CC domain-containing mutants (ΔEVH1, ΔPRD, EVH2 and CC) than by the EVH1 domain (EVH1) (Figure 2B, lanes 3–7). These results indicate that the CC domain of VASP is the major interaction site with Abi-1, whereas the EVH1 domain also contributes to the efficient interaction. Abi-1 was recently identified as the binding partner of the EVH1 domain of VASP . This previous study concluded that Abi-1 does not interact with the CC domain of VASP, on the basis of negative results in the yeast two-hybrid assay. Presumably, their fusion construct of Gal-AD and the EVH2 domain did not work properly in the two-hybrid assay. They also tested an in vitro interaction. Their in vitro assay did not include the EVH2 domain alone, and its results did not exclude the CC domain of VASP for the interaction site with Abi-1.
Binding domain between VASP and Abi-1
Next, we examined the binding domain in Abi-1. Abi-1 contains the SH3 (Src homology 3) domain and a PP (poly-proline) region near the C-terminus (Figure 2C). Full-length FLAG–Abi-1 and Abi-1ΔSH3 lacking amino acids 391–451 interacted with GST–VASP, whereas Abi-1ΔPP lacking amino acids 336–361 did not (Figure 2D). The level of tyrosine phosphorylation of GST–VASP co-expressed with FLAG–Abi-1ΔPP was lower than that co-expressed with full-length FLAG–Abi-1 (Figure 2E). These results indicate that the PP region of Abi-1 is required for the interaction with VASP and the promotion of phosphorylation of VASP by c-Abl.
VASP directly binds Abi-1
We next examined the direct interaction between VASP and Abi-1 and the effect of the tyrosine-phosphorylation of VASP in this interaction. Tyrosine-phosphorylated and unphosphorylated SH–VASP were purified from the HEK-293T cells (Figure 3A). GST-tagged Abi-1 was purified from bacteria (Figure 3A). The direct interaction between Abi-1 and VASP was examined using AlphaScreen (PerkinElmer), which utilizes two different types of beads called donor beads and acceptor beads. SH–VASP (2.5 pmol) and GST-Abi-1 (0–0.25 pmol) were immobilized on nickel chelate donor beads and anti-GST antibody-coated Protein A acceptor beads respectively. Then donor and acceptor beads were mixed and incubated for 1 h at 22°C. The AlphaScreen signal between GST–Abi-1 and SH–VASP was increased in a manner dependent on the amount of GST–Abi-1, indicating that VASP directly binds to Abi-1 (Figure 3B). Tyrosine phosphorylation decreased the interaction of VASP with Abi-1 (Figure 3B).
Direct interaction between Abi-1 and VASP
in vitro and in vivo
Interaction of VASP with Abi-1
The interaction of VASP and Abi-1 was further confirmed by live-cell imaging in Xenopus laevis XTC cells. After cell spreading on poly-L-lysine-coated coverslips, c-Abl associated most strongly with the tip of the actin network in lamellipodia  where c-Abl and VASP co-localized (Figure 3C and Supplementary Movie S1 at http://www.BiochemJ.org/bj/441/bj4410889add.htm). Abi-1 also co-localized with VASP at the tip of lamellipodia (Figure 3C and Supplementary Movie S2 at http://www.BiochemJ.org/bj/441/bj4410889add.htm). To further confirm the interaction between Abi-1 and VASP in vivo, we employed a strategy to overexpress Abi-1 on the surface of mitochondria. mRFP1-tagged Xenopus Abi-1 was fused to a mitochondria targeting signal (Abi-1-mito) derived from Listeria ActA protein , and co-expressed with GFP-tagged Xenopus VASP. When co-expressed with Abi-1-mito, a fraction of VASP was recruited to mitochondria and co-localized with Abi-1-mito (Figure 3D). VASP was found to associate and move with Abi-1-mito on the mitochondria surface (Figure 3E and Supplementary Movie S3 at http://www.BiochemJ.org/bj/441/bj4410889add.htm). Furthermore, endogenous VASP was co-precipitated with Abi-1 in K562 cells (Figure 7C). These results confirm that Abi-1 interacts with VASP in vivo.
Co-fractionation of VASP and c-Abl with Abi-1 at a molecular mass of approximately 600 kDa
We next examined whether VASP might form a ternary complex with c-Abl and Abi-1 using gel-filtration chromatography. We noticed that the cellular amount of Abi-1 appeared to be tightly controlled, as the amount of retrovirus-mediated stably expressed exogenous SH–Abi-1 did not exceed the amount of endogenous Abi-1 in CHO cells (Figure 4A, lanes 1 and 2). As previously reported [28,49], stable expression of tagged Abi-1 eliminated the expression of endogenous Abi-1. From these observations, we speculate that stable expression of SH–Abi-1 replaced the majority of endogenous Abi-1 in CHO cells. The elution profile of gel-filtration analysis in the total lysate (Figure 4B) revealed that peak fractions for each protein corresponded to the molecular mass of approximately 400 kDa for c-Abl, 600 kDa and 450 kDa for Abi-1, and 400 kDa for VASP. To detect the possible ternary complex of Abl, Abi-1 and VASP, we purified the proteins associating with His-tagged Abi-1 using Ni–Sepharose from 50 mg of CHO cell lysate. Gel-filtration analysis of affinity-purified SH–Abi-1 complex displayed an additional peak at a molecular mass of 600 kDa in the elution profile of VASP and c-Abl at which three proteins co-migrated (Figure 4C). These results indicate that VASP may form a ternary complex with c-Abl and Abi-1.
Gel-filtration analysis of c-Abl, Abi-1 and VASP
Tyr39 in VASP is phosphorylated by Bcr-Abl and c-Abl
In mouse Mena, Tyr296 is phosphorylated by c-Abl . However, a corresponding tyrosine residue is not conserved in VASP. Therefore we sought the phosphorylation site of VASP. VASP contains four tyrosine residues. We created GST-tagged VASP mutants by substituting each tyrosine residue with phenylalanine (Figure 5A). Each GST–VASP mutant was co-expressed with Bcr-Abl and Abi-1 in HEK-293T cells, and tyrosine phosphorylation of GST–VASP was examined by Western blotting. Y16F, Y72F and Y341F mutants were phosphorylated at almost the same level as WT VASP. On the other hand, the Y39F mutant was not phosphorylated (Figure 5B). To further confirm phosphorylation at Tyr39, we generated VASP mutants in which either three or all of these tyrosine residues were substituted by phenylalanine (Figure 5A). Phosphorylation of 16Y (Y39F/Y72F/Y341F), 72Y (Y16F/Y39F/Y341F), 341Y (Y16F/Y39F/Y72F) and All-F (Y16F/Y39F/Y72F/Y341F) mutants was barely observed, whereas the 39Y (Y16F/Y72F/Y341F) mutant was phosphorylated to the same extent as WT VASP (Figure 5C). When co-expressed with c-Abl and FLAG–Abi-1, Y39F was not phosphorylated, but 39Y and WT VASP were phosphorylated to an equal extent, as was the case with Bcr-Abl (Figure 5D). These results indicate that Tyr39 is a major phosphorylation site in VASP by Bcr-Abl and c-Abl.
Identification of the phosphorylation site in VASP
Tyrosine phosphorylation decreases localization of VASP at focal adhesions
To understand the function of phosphorylation in VASP, we next examined the localization of VASP in NIH 3T3 cells. We established cell lines expressing VASP WT and the Y39F mutant. We also created a VASP mutant in which Tyr39 was substituted by aspartic acid (Y39D), which is expected to act as a phosphomimetic mutation. VASP WT and its mutants were tagged with GFP and stably transduced into NIH 3T3 cells by the retrovirus vector. GFP–VASP WT and its mutants were almost equally expressed (Figure 6A). As described previously , VASP WT was localized at focal adhesions (Figure 6B). Y39F mutation did not alter the localization of VASP at focal adhesions. However, Y39D was diffusively distributed in the cytoplasm and localized at focal adhesions to a lesser extent than WT (Figure 6B).
Phosphorylation of Tyr39 inhibits accumulation of VASP at focal adhesions and reduces its affinity to the proline-rich motif of zyxin
We next examined the localization of VASP WT or Y39F in NIH 3T3 cells overexpressing c-Abl and Abi-1 (Figure 6C). mPlum–vinculin was used as the focal-adhesion marker and the intensity ratio of GFP–VASP at focal adhesions and in the whole cell was quantified (Figure 6C and Supplementary Figure S1 at http://www.BiochemJ.org/bj/441/bj4410889add.htm). Co-expression of a combination of c-Abl and FLAG–Abi-1 reduced the association of VASP to focal adhesions in VASP WT-expressing cells, but not in VASP Y39F-expressing cells (Figure 6D). These results suggest that Abi-1-dependent phosphorylation by Abl inhibits the association of VASP with focal adhesions.
The phosphomimetic form of VASP has a reduced affinity to the FP4 motif of zyxin
VASP is recruited to focal adhesions by interacting with the FP4 motif of zyxin [50,51]. The EVH1 domain of VASP acts as the binding module to the FP4 motif . Since Tyr39 is positioned in the EVH1 domain of VASP, we postulated that phosphorylation of VASP might affect the interaction with zyxin. To test this possibility, an in vitro binding assay was performed using the zyxin-derived FP4 motif tagged with GST (GST–zyxin FP4). His-tagged VASP and its phosphomimetic Y39D mutant were mixed with GST or different amounts of GST–zyxin FP4. While VASP WT was co-precipitated with GST–zyxin FP4 in a dose-dependent manner, the VASP Y39D mutant was less efficiently precipitated with GST–zyxin FP4 than VASP WT (Figure 6E). These results suggest that tyrosine phosphorylation of VASP reduces its affinity to the FP4 motif of zyxin. The reduced affinity of tyrosine-phosphorylated VASP and zyxin may cause delocalization of VASP at focal adhesions.
VASP is tyrosine-phosphorylated in Bcr-Abl-transformed leukaemic cells in an Abi-1-dependent manner
Tyrosine phosphorylation of endogenous VASP was observed in K562 erythroblastic cells and Meg-01 megakaryoblastic leukaemic cells, and this phosphorylation was reduced when cells were treated with imatinib, a selective Abl kinase inhibitor (Figure 7A). Depletion of Abi-1 from K562 cells reduces tyrosine phosphorylation of VASP (Figure 7B). Abi-1, c-Abl and Bcr-Abl were immunoprecipitated with anti-VASP antibodies in K562 cells (Figure 7C). These results suggest that VASP is phosphorylated by Bcr-Abl in leukaemic cells and phosphorylation of VASP is regulated by Abi-1.
VASP Y39F and Y39D impair cell adhesion to fibronectin in K562 cells
VASP Y39D and Y39F impair adhesion of leukaemic cells to fibronectin
We next tested whether tyrosine phosphorylation of VASP contributes to the ability of cell adhesion of K562 cells, because depletion of Abi-1 or disruption of the interaction between Bcr-Abl and Abi-1 impairs cell adhesion and migration [36,37]. GFP-tagged VASP WT, the unphosphorylated Y39F mutant, and the phosphomimetic Y39D mutant were stably transduced into K562 cells by retrovirus infection. The expression levels of exogenous GFP–VASP variants were higher than endogenous VASP (Figure 7C). Co-precipitation of endogenous VASP with GFP–VASP was observed in cells expressing GFP–VASP WT and its mutants (Figure 7C). Since VASP forms a tetramer , we postulated that overexpressed GFP–VASP has a dominant effect over the function of endogenous VASP. GFP–VASP WT was tyrosine-phosphorylated in K562 cells, whereas the phosphorylation of the Y39F and Y39D mutant was barely observed (Figure 7C), confirming that Tyr39 is the major phosphorylation site in VASP. The growth rate of each cell line was not altered by the expression of each protein (Figure 7D). Expression of GFP–VASP WT did not significantly alter the number of adherent cells compared with GFP-expressing cells, whereas the expression of Y39F and Y39D significantly reduced the number of K562 cells adherent to fibronectin (Figure 7E). We also tested the function of VASP and its mutants in VASP-depleted K562 cells. Similar to the overexpression analysis (Figure 7E), the Y39F or Y39D mutant reduced adhesion compared with GFP–VASP WT-expressing cells (Supplementary Figure S3C and S3D at http://www.BiochemJ.org/bj/441/bj4410889add.htm). As described previously [54,55], however, down-regulation of VASP abnormally increased cell adhesion in K562 cells (Supplementary Figure S3A and S3B). Since the cell adhesion was reduced to the same extent as control cells by re-expression of GFP–VASP WT in VASP-depleted K562 cells, we believe that GFP–VASP WT complements the function of endogenous VASP. Reduced adhesion of the cells re-expressing the Y39F or Y39D mutant indicates a crucial role of Tyr39 phosphorylation in the VASP regulation of cell adhesion.
The Y39D mutation impairs association of VASP at focal adhesions by reducing its affinity to the FP4 motif of zyxin (Figures 6B and 6D), whereas Y39F remains associated with focal adhesions even in the cells overexpressing c-Abl and Abi-1 (Figure 6C). Therefore not only removal of VASP from focal adhesions, but also constitutive association of VASP with focal adhesions, may down-regulate adhesion of K562 cells. Taken together, these results suggest that the phosphorylation and dephosphorylation cycle on Tyr39 in VASP might contribute to effecient adhesion of Bcr-Abl-positive leukaemia cells.
In the present study, we have found that Abi-1 bridges the interaction between Abl and VASP, and promotes phosphorylation of VASP by Abl. The kinase activity of c-Abl is inhibited by intramolecular interactions [10,56]. c-Abl is composed of several domains, including the NCAP motif, the SH3 and the SH2 domains, the proline-rich region, the DNA-binding region and the F-actin-binding domain. Phosphorylation of Tyr412 and Tyr245 stimulates the catalytic activity of c-Abl . In autoinhibited c-Abl, its SH3 domain interacts with the PP motif in the SH2-kinase linker. Disruption of this interaction by the mutation of P242E/P249E, which acts similarly to phosphorylation of Tyr245, highly activates the kinase [10,58]. The kinase activity of c-Abl is also regulated by its binding molecules. Abi-1 binds to the SH3 domain and the C-terminal proline-rich region of c-Abl [28,59]. In addition, phosphorylation of Abi-1 by c-Abl enhances its interaction with c-Abl and up-regulates the catalytic activity of c-Abl . Abi-1 thus binds and stabilizes the activated form of c-Abl released from autoinhibitory interactions. In addition, Abi-1 contributes to recognition of the substrate by c-Abl. Our previous findings suggest that Abi-1 binds to several substrates of c-Abl, including Mena , BCAP  and WAVE-2 . Abi-1 presumably bridges the interaction between c-Abl and the substrate. Our present study has revealed VASP as another example of Abi-1-bridged substrates of Abl. VASP co-localized with mitochondria-targeted Abi-1 in XTC cells. Gel-filtration analysis showed co-fractionation of VASP, c-Abl and Abi-1 at approximately 600 kDa. In addition to the previously identified EVH1 domain , we identified the CC domain of VASP as another binding site for Abi-1. The PP region of Abi-1 is critical for this interaction, which plays a crucial role in promoting phosphorylation of VASP by Abl. Tyrosine phosphorylation of endogenous VASP was diminished by depletion of Abi-1 from K562 cells. Therefore our present study supports the notion that Abi-1 not only works as a co-activator of Abl, but also serves as a unique regulatory molecule that bridges the interaction between Abl and the substrate.
In Drosophila, loss-of-function mutation of Abi shows a lethal phenotype . Interestingly, like Ena, a reduction in the gene dosage of Abi rescues the lethal phenotype of the homozygous loss-of-function mutation of DAbl in Drosophila [13,14]. In addition, heterozygous Abi mutation suppresses the overgrowth phenotype of synaptic neuromuscular junctions in homozygous Abl mutations . These observations suggest that Abi and Ena have an analogous antagonistic role against DAbl in Drosophila axonogenesis.
On the other hand, DAbl is required for the proper function of Ena. WT Ena rescues the lethal phenotype of the loss-of-function mutation in Ena by 86%, whereas the mutant lacking the phosphorylation sites restores viability only by 53% . Therefore phosphorylation of Ena by DAbl is required for Drosophila axonogenesis.
Abi enhances the phosphorylation of Ena by DAbl in Drosophila , and also shows a co-operative genetic interaction with DAbl. As mentioned above, lethality of the loss-of-function mutation of DAbl was rescued by the Abi homozygous mutation. In contrast, the heterozygous mutation of Abl increased the lethality of the loss-of-function mutation of Abi . Complex genetic interactions may thus be present between Abl and Ena or Abi in Drosophila development, although the underlying biochemical mechanism has remained unknown.
We have identified Tyr39 as the phosphorylation site of VASP. The same tyrosine residue was indicated by phosphoproteome analysis in leukaemic cells . So what might be the function of tyrosine phosphorylation in VASP? Decreased accumulation of VASP at focal adhesions was indicated in NIH 3T3 cells by co-expression of c-Abl and Abi-1, and the phosphomimetic VASP Y39D mutant is not localized at focal adhesions. The VASP Y39D mutant has a reduced affinity to the FP4 motif of zyxin in vitro. In Drosophila, substitution of Ala97 to valine (A97V) in Ena leads to embryonic lethality. The A97V mutant of Ena is defective in binding to zyxin and is not localized to focal adhesions , indicating that the interaction between zyxin and Ena has a critical role in Drosophila development. On the other hand, zyxin-null fibroblasts show reduced accumulation of Ena/VASP proteins at focal adhesions . These observations indicate that VASP is recruited to focal adhesions by interacting with zyxin. Our present study has revealed that Abl-catalysed phosphorylation inhibits the interaction between VASP and zyxin, and down-regulates the localization of VASP at focal adhesions. This mechanism may play an important role in leukaemic cell adhesion.
The role of the Abi-1-mediated signalling pathway has been extensively studied in Bcr-Abl-induced leukaemogenesis [36,37,46]. On the other hand, VASP is a tyrosine-phosphorylated protein in several Bcr-Abl-transformed cell lines . We also observed tyrosine phosphorylation of VASP in K562 and Meg-01 cells that are derived from CML patients in blast crisis phase, and the phosphorylation was reduced in the depletion of Abi-1. VASP is a focal adhesion protein which regulates cell adhesion [54,55] and migration . These observations prompted us to test whether Abi-1-bridged phosphorylation of VASP by Bcr-Abl might contribute to leukaemic cell adhesion. In NIH 3T3 cells, phosphomimetic VASP Y39D is not localized at focal adhesions, whereas WT and Y39F are localized at focal adhesions. These results suggest that Tyr39 phosphorylation in VASP blocks its association to focal adhesions. On the other hand, expression of both VASP Y39D and Y39F, but not WT, reduced the adhesion of K562 cells to fibronectin. Therefore not only phosphorylation, but also dephosphorylation of VASP at Tyr39 may be required for the enhanced adhesion of K562 cells. Since VASP functions as a tetramer , VASP mutants form a complex with endogenous VASP and may affect its function. From these observations we conclude that the phosphorylation and dephosphorylation cycle in VASP may contribute to the promoting effect on cell adhesion exerted by Bcr-Abl in leukaemic cells.
In addition, our previous study using high-resolution live-cell imaging of a c-Abl–GFP probe found that treatment of imatinib, a selective Abl inhibitor, induced translocation of c-Abl to the leading edge of lamellipodia . Several lines of evidence from mutation analysis on c-Abl indicate that imatinib induces the release of autoinhibition of c-Abl and promotes the binding of c-Abl to its putative partners . We have recently identified Abi-1 as the binding partner responsible for imatinib-induced cell edge translocation of c-Abl (Y. Yuan, M. Maruoka and N. Watanabe, unpublished work). VASP localizes to focal adhesions and to the tips of lamellipodia and filopodia . Others have also reported that VASP and Abi-1 partially co-localize at the tip of lamellipodia in human platelets when cells are stimulated with adenosine diphosphate or thrombin . Therefore c-Abl translocates to the tip of lamellipodia, where VASP is enriched, with the help of Abi-1. Further characterization of the molecular behaviour of VASP and its mutants in live cells may provide a clue to understanding the role of tyrosine phosphorylation in VASP.
In summary, we have identified VASP as an Abi-1-bridged substrate of Bcr-Abl and c-Abl. The results of the present study have revealed a new signal transduction mechanism involving the Abi-1-bridged phosphorylation of the substrate and, for the first time, provided evidence of the regulation of adhesion in Bcr-Abl-transformed leukaemic cells by such a bridged phosphorylation mechanism. The tyrosine phosphorylation and dephosphorylation cycle in VASP may play a critical role in cell–matrix attachment of Bcr-Abl-transformed leukaemic cells. Clarification of the molecular mechanisms by which the tyrosine phosphorylation and dephosphorylation cycle in VASP contributes to leukaemic cell adhesion is a subject of future investigation.
B-cell adaptor for PI3K (phosphoinositide 3-kinase)
Coomassie Brilliant Blue
cell division cycle 2
chronic myelogenous leukaemia
ecotropic retrovirus receptor
enhanced green fluorescent protein
fetal bovine serum
green fluorescent protein
human embryonic kidney
monomer red fluorescent protein 1
protein kinase A
Src homology 3
small interfering RNA
Wiskott–Aldrich syndrome protein family member 2
Masahiro Maruoka, Mizuho Sato, Yunfeng Yuan, Masayoshi Ichiba and Ryosuke Fujii performed the experiments and analysed the data. Takuya Ogawa, Norihiro Ishida-Kitagawa, Tatsuo Takeya and Naoki Watanabe designed the experimental methods. Masahiro Maruoka and Naoki Watanabe wrote the paper.
We thank Dr Frank B. Gertler (Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA, U.S.A.) for the plasmid encoding the mitochondria-targeting motif of Listeria ActA. We also thank Jun Suzuki for critical reading of the paper prior to submission. The present study was initially supervised by Tomoyuki Shishido. We are grateful to Tomoyuki Shishido and Katsuko Tani for their support in early phase of this project.
This work was supported, in part, by the Funding Program for Next Generation World-Leading Researchers [grant number LS013 (to N.W.)], and by grants from the Human Frontier Science Program (to N.W.) and the Takeda Science Foundation (to N.W.). M.M. was a Japan Society for the Promotion of Science research fellow.
These authors contributed equally to this work.