BCR-ABL–induced adhesion defects are tyrosine kinase–independent

P210^BCR-ABL is a 210-kd nonreceptor tyrosine kinase that is the product of the t(9;22) balanced translocation between the bcr gene on chromosome 22 and theabl gene on chromosome 9.1 The translocation results in the formation of the Philadelphia chromosome (Ph), which is the diagnostic marker for chronic myelogenous leukemia (CML), a disease characterized by abnormal accumulation of hematopoietic cells in the bone marrow, elevated peripheral white blood cell counts, and defective progenitor-stromal adhesion in vitro.2,3 There is intense interest in the observation that P210^BCR-ABLexpression in cell lines and bone marrow progenitor cells leads to defective cell binding to fibronectin and stromal cells because abnormal adhesion in transformed cells may contribute to loss of contact inhibition and proliferation in the bone marrow of patients with CML.3-5 However, the mechanism by which P210^BCR-ABL alters cellular adhesion is poorly understood.

P210^BCR-ABL is primarily found in the cytoplasm and colocalizes with actin through a C-terminal actin-binding domain.6,7 It is unknown if P210^BCR-ABLdirectly modifies actin or recruits signaling molecules to the cytoskeleton. However, transformed cells exhibit increased motility and enhanced membrane ruffling and proteins of the focal adhesion complex, such as focal adhesion kinase and paxillin, are phosphorylated in P210^BCR-ABL cells.8,9 Together, these observations suggest that P210^BCR-ABL induces abnormal cell-substratum binding that is likely involved in the pathogenesis of CML. One possible mechanism is that P210^BCR-ABL regulates an inside-out signaling pathway that stimulates integrin-mediated adhesion in a manner similar to growth factors, such as interleukin 3 (IL-3).10 The tyrosine kinase activity is critical for the oncogenic effects of P210^BCR-ABL because expression of P210^BCR-ABL mutants with either temperature-sensitive tyrosine kinase activity or an inactive tyrosine kinase domain are not oncogenic in vitro or in vivo.11,12Accordingly, the P210^BCR-ABL constitutively activated tyrosine kinase is a plausible focal point for inside-out signaling.

Differentiation of the hematopoietic compartment from stem cells through progenitors is a tightly regulated process that relies on specific interactions between stromal and hematopoietic cells for proper cues governing proliferation and differentiation.13,14 In the case of CML, defective progenitor-stromal interactions have been documented using both bone marrow from CML patient donors3,4 and transformed cell lines5,15; yet these studies have yielded equivocal results. It is unclear if these disparate findings are due to methodologic differences or the nature of the cells assayed. Most adhesion studies are limited to panning-style assays that use an uncharacterized and inconsistent washing technique to detach cells. To overcome the inherent variability of panning-style assays to measure adhesion, we adapted a quantitative adhesion assay that relies on fluid flow to remove bound cells using a measurable detachment force.16 This assay provides a quantitative analysis of cell-ligand interactions, and thus provides the capability to measure small changes in adhesion.17 The cell detachment device has previously been used to examine the role of α₅β₁ integrins in facilitating adhesion between the K562 erythroleukemia cell line and fibronectin to show that antibody-mediated integrin activation stimulates adhesion and increases the force needed to detach adherent cells.18 Radial flow detachment assays using a similar hydrodynamic environment have shown that neutrophils bind more tightly to fibrinogen than to kininogen19 and that osteoblasts adhere to biomaterial implants coated with Arg-Gly-Asp (RGD) sequences.20 

To investigate the role of P210^BCR-ABL in cellular adhesion, we used the cell detachment device to measure adhesive changes in 32D cells following transduction by P210^BCR-ABL–expressing retroviruses. The 32D cells expressing P210^BCR-ABL were 1.7 times more adhesive to fibronectin than cells transduced with a control vector. We then used either a tyrosine kinase dead P210^BCR-ABL mutant with a Lys1176Arg (K1176R) amino acid substitution12 or wild-type P210^BCR-ABL cells treated with the specific P210^BCR-ABL tyrosine kinase inhibitor, STI571 (imatinib mesylate, Gleevec),21 to evaluate the role of tyrosine kinase activity in regulating cellular adhesion. The importance of the P210^BCR-ABL catalytic activity in contributing to CML is underscored by recent studies indicating that the P210^BCR-ABL tyrosine kinase inhibitor STI571 normalizes white blood cell counts in patients with CML.22 We found that neither the tyrosine kinase inactive mutant nor treatment of wild-type P210^BCR-ABL with STI571 could normalize the increased adhesion observed on expression of P210^BCR-ABL. We also found that localization of P210^BCR-ABL to F-actin filaments is independent of tyrosine kinase activity. These findings suggest that both altered adhesion to fibronectin and actin localization are tyrosine kinase-independent activities of P210^BCR-ABL. These tyrosine kinase-independent properties of P210^BCR-ABL may synergize with its tyrosine kinase-dependent activity to induce the transformed phenotype.

Full-length P210^BCR-ABL (Pear et al)23and the kinase dead K1176R mutant12 were ligated into pK1, an MSCV-based vector that is identical to MigR1 except that an IRES-puromycin resistance gene has been inserted in place of the IRES-GFP (kind gift from Karen Ehrmann and Stephen Emerson, University of Pennsylvania). High-titer retroviral supernatants were generated.23 BCR-ABL constructs were transduced into 32D cells24 by spinoculation.25 BCR-ABL or control pK1 cells were selected with puromycin (1.5 μg/mL) and maintained in 32D media (RPMI media, 10% fetal bovine serum, 1% penicillin/streptomycin, and 1% glutamate) containing 10% WEHI as a source of IL-3. Meg-01 cells26(CRL-2021, American Type Culture Collection, Rockville, MD), a Ph⁺ megakaryoblastic cell line, were maintained in Meg-01 media (Dulbecco modified Eagle medium [DMEM] supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin, and 1% glutamate).

Fifteen to 17 hours prior to assessing adhesion, pK1 and/or BCR-ABL–expressing 32D cells were diluted into 32D medium containing 1% WEHI with or without 10 μM STI571, as indicated. Meg-01 cells were treated with 10 μM STI571 for 3.5 hours or an equal volume of the drug carrier (phosphate-buffered saline [PBS]) for 3.5 hours or 1.5 hours prior to assaying binding. Cells were washed 2 times in PBS and resuspended in adhesion buffer (24 mM Tris, 137 mM NaCl, 1 mM MgCl₂, 3 mM KCl, 2 mM glucose) at 1 × 10⁶cells/mL. In some control experiments, Meg-01 cells were treated with 2.125 mM RGD peptide (catalogue no. 03-34-0035, Calbiochem, San Diego, CA) for 10 minutes prior to seeding on fibronectin substrates. Adhesion assays were carried out in the absence of growth factors and were completed within 2 hours of washing. Controls showed that adhesion levels did not change during the period in which multiple samples were assayed (data not shown).

Circular glass coverslips (Fisher, Pittsburgh, PA) were adsorbed with 10 μg/mL fibronectin (Becton Dickinson, San Jose, CA) and blocked with 1% bovine serum albumin (BSA). The cell detachment device was operated as described.16 Coverslips were analyzed by recording the adherent fraction of cells at various radial distances from the center. The shear stress, τ, that is imparted on the cells is calculated by the following equation: τ = 0.800 r (ρμω3)^1/2where τ is proportional to r, the radial distance; ρ and μ are the fluid density and dynamic viscosity, respectively; and ω is the angular velocity of the spinning disk. The convective fluid flow that is established by a spinning disk has been characterized,27,28 and the fraction of beads or cells under shear flow that remain bound to an adhesive surface is known to follow a sigmoid-shaped profile.17,29,30 Adhesion profiles for this study were prepared16 and were accepted forR² ≥ 0.80, with the exception of runs using matrices consisting of BSA alone or when Meg-01 cells were treated with the RGD peptide due to the few cells that remain attached in the absence of fibronectin. The critical shear stress, τ₅₀, was measured as the surface shear stress, τ, at an adherent fraction of 0.50. Determination of significance between mean values of critical shear stress, τ₅₀, was determined using a 2-tailed Student t test with significance defined asP ≤ .05. Data are presented as mean ± SD.

Western blots were performed on lysates from 32D cells expressing BCR-ABL or pK1 that were incubated overnight in 32D media with 10% WEHI with or without decreasing concentrations of STI571. Meg-01 cells were treated with 10 μM STI571 or an equal volume of PBS alone for 3 hours. Equal volumes of lysates were assayed on 7% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gels to detect P210^BCR-ABL, 12.5% SDS-PAGE gels to detect grb2, or 10% SDS-PAGE gels to detect tyrosine phosphorylated proteins. Proteins were transferred from gels to nitrocellulose membranes and were blotted with an anti-abl antibody (24-11, Santa Cruz Biotechnology, Santa Cruz, CA), an antiphosphotyrosine antibody (PY20, Pharmingen, San Diego, CA), or an anti-grb2 antibody (81, Transduction Laboratories, Lexington, KY). Anti-α₄ (R-2, Pharmingen) and anti-α₅ (5H10-27, Pharmingen) antibodies were used to assess integrin expression by flow cytometry. All samples and controls were run in triplicate. The mean fluorescence index (MFI) was calculated as the difference in geometric mean fluorescence between experimental samples stained with primary and secondary antibodies and negative control samples stained with the secondary antibody alone; this value was normalized to the geometric mean fluorescence of the negative control to give the MFI.

NIH3T3 or 32D cells were transduced with retroviral constructs encoding P210^BCR-ABL, K1176R, or the pK1 control plasmid and selected for 4 days with puromycin. NIH3T3 cells were cultured overnight on fibronectin coverslips with or without 10 μM STI571 as indicated. 32D cells were plated onto fibronectin coverslips at a concentration of 1 × 10⁵ cells/mL by following the same procedure outlined above for adhesion assays. Cells were fixed with 2% paraformaldehyde, permeablized with 0.1% Triton-X, and blocked with 1% denatured BSA (NIH3T3 cells) or 5% normal goat serum diluted in 1% denatured BSA (32D cells). Coverslips were stained with 5 μg/mL anti-abl antibody (8E9, Pharmingen). Bound anti-abl antibody was revealed with goat antimouse Texas red antibody (NIH3T3 cells) or goat antimouse Cy3 antibody (32D cells, Jackson ImmunoResearch, West Grove, PA). Actin was stained with 0.1 μg/mL fluorescein isothiocyanate–phalloidin (Sigma, St Louis, MO) and the nucleus was revealed with 4′,6-diamidino-2-phenylindole (DAPI) staining (1:10 000 dilution). Cells were imaged using confocal microscopy at the Bioengineering Confocal and Multiphoton Imaging Facility at the University of Pennsylvania.

The cell detachment device presents a 2-dimensional matrix of fibronectin and BSA to a monolayer of cells seeded at a constant surface density throughout the coverslip (data not shown). This well-characterized surface is mounted onto a rotating circular stage that creates convective fluid flow within the chamber that dislodges bound cells (Figure 1A). The detachment force is calculated (equation in “Materials and methods”) from the properties of the buffer, the dimensions of the coverslip, and the rotational speed that is measured by a digital tachometer. The detachment force is zero at the center of the matrix and increases linearly to reach a maximum near the circumference, allowing for a range of detachment forces to be tested during a single experiment. Accordingly, few cells remained near the edge due to the higher force at this location than closer to the center (Figure 1B).

The fraction of cells that remain on the surface after a spin (normalized to the count at the center where r = τ = 0) is plotted versus the shear stress applied to detach bound cells at each radial position. Each data point indicates the degree of cell-fibronectin binding (Figure 1C). These data trace a sigmoid-shaped profile that allows a curve to be fitted to the observed data.16 Adhesion of cells to BSA-coated coverslips, representing the level of nonspecific binding, is minimal and is just below or at the detection threshold of the device (Figure 1C). Addition of fibronectin to the matrix significantly increases cell adhesion as indicated by the higher fraction of cells remaining attached to the matrix at every given shear stress, effectively shifting the adhesion profile to the right (Figure 1C). Addition of fibronectin-blocking antibodies to the monolayer or treatment of 32D cells with integrin-blocking antibodies or RGD-containing peptides significantly reduces or completely abrogates cell binding (J.A.W., manuscript in preparation). This shows that the interaction between 32D cells and the adhesive matrix is specific for fibronectin-integrin linkages.

The advantage of the spinning disk adhesion assay is its ability to quantitatively measure the strength by which cells bind fibronectin over a range of defined forces. Accordingly, we asked whether expression of P210^BCR-ABL influenced cell adhesion to fibronectin compared to pK1 control cells expressing an empty vector. In the presence of fibronectin, expression of P210^BCR-ABLin 32D cells increased the fraction of bound cells at nearly all shear stresses tested compared to pK1 control cells, indicating that expression of P210^BCR-ABL led to increased cell adhesion (Figure 2A). The critical shear stress, τ₅₀, (ie, adhesion strength) is the force required to dislodge half of the cells and is a measurement of the strength of association between cells and fibronectin.29,30 The τ₅₀ is averaged for a set of identical experiments and can be compared over different conditions, such as with and without P210^BCR-ABL expression (Figure 2B). Expression of P210^BCR-ABL led to an increase in the average critical shear stress, τ₅₀, from 35.1 dyne/cm² to 59.6 dyne/cm², indicating that adhesion increased approximately 1.7-fold. We also observed nearly a 2-fold increase in adhesion when P210^BCR-ABL is expressed from a retroviral construct coexpressing green fluorescent protein (data not shown).

The finding that P210^BCR-ABL expression leads to increased cell adhesion in our system may potentially be explained by inside-out signaling initiated by the P210^BCR-ABL tyrosine kinase. To assess the role of tyrosine kinase activity on cell adhesion to fibronectin, a P210^BCR-ABL mutant lacking tyrosine kinase activity, K1176R, was transduced into 32D cells. Similar to wild-type P210^BCR-ABL 32D cells, K1176R 32D cells bound more tightly to fibronectin than control pK1 cells (Figure3A). Moreover, we found that the critical shear stress of a series of spins using 2 separate populations of K1176R-expressing cells were not statistically different from P210^BCR-ABL (P = .14, P = .89, not shown), but were significantly elevated compared to the control pK1 cells (P < .001, P = .003, Figure3B).

One possible mechanism by which P210^BCR-ABL increases adhesion is by influencing integrin expression. The α₄β₁ and α₅β₁integrins are the major cell surface receptors for fibronectin and bind to different, conserved sequences on the macromolecule.31,32 Expression of α₄β₁ and α₅β₁integrins on P210^BCR-ABL, K1176R, and pK1 control cells was measured by flow cytometry. An increase in α₄β₁ expression was observed; however, there was a simultaneous decrease in α₅β₁expression (Figure 3C). The changes in expression of α₄β₁ and α₅β₁integrins were similar in the presence or absence of the tyrosine kinase activity of P210^BCR-ABL (Figure 3C). In parallel studies using α₄- and α₅-blocking antibodies, we found that anti-α₄ had minimal affects on the magnitude of P210^BCR-ABL induced adhesion, whereas anti-α₅ significantly diminished the P210^BCR-ABL–induced increase in adhesion (J.A.W., manuscript in preparation). These results suggest that the changes in α₄β₁- and α₅β₁-integrin expression are unlikely to account for the nearly 2-fold increase in fibronectin binding induced by P210^BCR-ABL expression.

To confirm our previous finding using the tyrosine kinase–inactive K1176R mutant, we treated P210^BCR-ABL–expressing cells with STI571 as an additional method to evaluate the contribution of kinase activity on adhesion. STI571 is a specific inhibitor of the P210^BCR-ABL tyrosine kinase by nature of the unique adenosine triphosphate-binding cleft in the catalytic domain.21,33 Although STI571 is effective at inhibiting the kinase of the platelet-derived growth factor receptor and the c-kit receptor,34 we do not expect these receptors to play a role in our assay because their ligands are absent from the adhesion buffer.

P210^BCR-ABL expression dramatically increases phosphotyrosine levels in hematopoietic cells lines leading to cell transformation and IL-3 independence.35 We confirmed that K1176R expression or treatment of P210^BCR-ABL cells with STI571 abrogated both the kinase activity and transforming ability of P210^BCR-ABL. Specifically, the K1176R mutant or overnight treatment of P210^BCR-ABL–expressing cells with 10 μM STI571 or less reduced the total cellular phosphotyrosine level in a dose-dependent manner, even though P210^BCR-ABL expression was unchanged (Figure 4A). Furthermore, expression of K1176R was also comparable to untreated or STI571-treated wild-type P210^BCR-ABL in our system (Figure 4A). K1176R failed to rescue 32D cells from IL-3 withdrawal (data not shown) and treatment with STI571 reversed the growth factor independence of wild-type P210^BCR-ABL–expressing 32D cells leading to apoptosis in the absence of IL-3 (Figure 4B). However, STI571-treated P210^BCR-ABL cells survived in 10% WEHI medium containing IL-3, showing that STI571 was not cytotoxic at the concentrations used in this study.

As an independent method to confirm our finding that K1176R exhibits increased adhesion in the absence of tyrosine kinase activity, we treated P210^BCR-ABL–expressing 32D cells with STI571. We compared the adhesion of P210^BCR-ABL cells that were untreated or treated with STI571 overnight at the same conditions used to evaluate proliferation and tyrosine kinase activity. We found that attenuation of catalytic activity not only failed to decrease cell adhesion compared to untreated cells, but may have slightly increased the ability of these cells to bind fibronectin (Figure5). Taken together, these results suggest that the kinase activity is not a major contributor to the increased cell adhesion to fibronectin brought about by P210^BCR-ABLexpression.

To determine whether human Ph⁺ cells also exhibit increased adhesion to fibronectin, we investigated the role of P210^BCR-ABL kinase activity in modulating adhesion in Meg-01 cells. The Meg-01 cell line was derived from the bone marrow of a 55-year-old man in the blast crisis phase of CML.26Treatment of these cells with 10 μM STI571 led to a marked reduction in cell viability by 48 hours following treatment initiation (Figure6A). Similar to expression of wild-type P210^BCR-ABL in 32D cells, STI571 treatment did not effect expression of P210^BCR-ABL but led to a reduction in the whole cell phosphotyrosine levels compared to Meg-01 cells treated with PBS (Figure 6B). Attenuation of tyrosine kinase activity was observed as early as 3 hours after treatment of Meg-01 cells. At this time, binding of kinase attenuated Meg-01 cells to fibronectin was assessed and was found to be unchanged (P = .81) when compared to cells treated with an equal volume of PBS.

We sought to determine other characteristics of P210^BCR-ABL that may be kinase-independent. Because adhesion to fibronectin did not correlate with kinase activity, we asked if the kinase activity of P210^BCR-ABL affects its subcellular localization in NIH3T3 fibroblasts or 32D cells when detected by confocal microscopy. Like wild-type P210^BCR-ABL, abrogation of the kinase activity by either the K1176R point mutation or STI571 treatment did not alter the cytoplasmic localization of P210^BCR-ABL in either cell line (Figure 7A,B). In NIH3T3 cells, P210^BCR-ABL was found along membrane ruffles, actin filaments, and strongly stained close to the cellular membrane in the presence or absence of kinase activity regardless of the method by which it was attenuated (Figure 7A). The 32D cells remained rounded and did not spread on fibronectin during the 15-minute incubation period used in both the immunohistochemistry analysis and the adhesion assays. P210^BCR-ABL and actin both displayed a cortical staining pattern along the membrane of 32D cells, and this pattern did not vary in the presence or absence of tyrosine kinase activity (Figure7B).

To investigate the role of P210^BCR-ABL in modulating cell adhesion to fibronectin, a major component of the bone marrow extracellular matrix,36 we used a highly sensitive and quantitative adhesion assay to measure the hydrodynamic forces used to detach bound cells. This device has advantages over conventional plate and wash assays by directly measuring the strength by which a population of cells is anchored to the extracellular matrix. The standardized detachment forces used in this system eliminate operator bias and variability in washing techniques that may occur in studies using panning-style adhesion assays. In contrast to long-term culture-initiating cell assays,37 or similar adhesion assays that rely on secondary colony formation to enumerate adherent and nonadherent progenitor cells in a heterogeneous population of primary bone marrow,3 our adhesion apparatus directly measures the magnitude of binding between a homogeneous population of cells and fibronectin. Due to the unique geometry of the detachment device, we assay a range of shear stresses during each spin and observe changes in adhesion throughout the entire range of forces that are tested. The cell detachment device generates a range of detachment forces (0-100 dyne/cm²) on a single fibronectin-coated coverslip, allowing our data to be fitted to known models of cell adhesion,16,17,29,30,38 and recapitulates shear stresses commonly found throughout the human body. It should be noted that the precise hydrodynamic environment of the bone marrow is unknown; however, it is likely that some of the forces generated by the detachment device exceed forces found in the bone marrow. Nevertheless, we observe differences in adhesion between P210^BCR-ABL–expressing 32D cells and control cells at all shear stresses assayed, suggesting that increased adhesion induced by P210^BCR-ABL occurs at physiologic forces.

We expressed P210^BCR-ABL in the myeloblastic cell line, 32D, which affords a homogenous population of cells to study changes in adhesion in the predominant cell type affected by CML. Using this model system, we found that P210^BCR-ABL expression led to a 1.7-fold increase in adhesion. Consistent with our findings, there have been 2 other reports demonstrating that P210^BCR-ABLpromotes increased adhesion to fibronectin.5,15 Our results are also consistent with studies showing that adhesion between P210^BCR-ABL–expressing cells and fibronectin promotes cell proliferation, cell cycle progression, and protects cells from DNA damage-induced apoptosis.15,39 In this scenario, increased adhesion may both promote the accumulation of progenitor cells and enhance their survival, 2 features associated with the pathogenesis of CML.

P210^BCR-ABL phosphorylates paxillin, focal adhesion kinase, and other members of the focal adhesion complex, suggesting a link between the catalytic activity of P210^BCR-ABL and regulation of integrin-mediated adhesion.9 To assess whether the tyrosine kinase activity of P210^BCR-ABLinitiates inside-out signaling leading to increased adhesion, we tested the binding of 32D cells transduced with both the K1176R mutant or wild-type P210^BCR-ABL cells treated with STI571 at concentrations shown to attenuate tyrosine kinase activity, reverse 32D cell transformation, and kill human leukemia cells over a period of days. Using these 2 independent methods to attenuate kinase activity in 32D cells transduced with P210^BCR-ABL, we found that P210^BCR-ABL–mediated increased adhesion remains intact in the absence of tyrosine kinase activity. Adhesion to fibronectin was also shown to be independent of tyrosine kinase activity in a second cell line, Meg-01, a Ph⁺ cell line derived from a patient in the blast crisis phase of CML. Together, these results support our hypothesis that altered adhesion due to P210^BCR-ABLexpression is kinase-independent. Consistent with our results describing tyrosine kinase–independent functions of P210^BCR-ABL, a kinase inactive mutant of Drosophila abl rescued pupal lethality and defects in eye development of flies lacking abl function.40 Also in support of our findings, P210^BCR-ABL has been reported to localize to F-actin in a noncatalytic manner.7,41 Furthermore, expression of a kinase dead c-src mutant restored cell spreading of src^−/− fibroblasts on fibronectin.42Together, these findings suggest an important functional role for the tyrosine-kinase independent activities of P210^BCR-ABL.

Two previous studies found an inverse relationship between P210^BCR-ABL–induced transformation and adhesion. p3T3 fibroblasts, which become anchorage independent on transformation by P210^BCR-ABL, return to the adherent layer of cells after treatment with STI571.43 Bhatia and colleagues found that treatment of CML bone marrow with tryphostin AG957, a tyrosine kinase inhibitor, enriched adherent cells with progenitors that populated colony-forming cell assays, suggesting that P210^BCR-ABLdecreased adhesion in a kinase-dependent manner.44 In contrast to these assays, our rotating disk system directly measures the magnitude of cell-fibronectin binding, which is independent of the transformation status of the cells and does not require a read-out step that may be sensitive to processes that are known to require tyrosine kinase activity such as proliferation or differentiation.

One significant implication of our results is that altered binding to fibronectin is likely to be a direct consequence of P210^BCR-ABL expression and not a side effect of cellular transformation because increased adhesion does not correlate with cell transformation in our system. Two functions of P210^BCR-ABL, actin binding and activation of Ras signaling, are potential mechanisms by which P210^BCR-ABL may modulate adhesion. P210^BCR-ABL is found in the cytoplasm, colocalized with F-actin filaments,6 and cannot be removed from F-actin filaments by treatment with either 10 μM STI571 alone or by complete abrogation of kinase activity by the K1176R mutant. Using confocal microscopy we show that P210^BCR-ABL is found along membrane ruffles, actin filaments, and close to the cellular membrane irrespective of its kinase activity or the method by which the tyrosine kinase activity was attenuated. This suggests that F-actin localization is kinase independent, and a recent report has shown that the Abl-related gene-1 (Arg-1) also binds F-actin even after deletion of its kinase domain.45 An intact actin cytoskeleton is required for cell adhesion in our system because treatment with cytochalasin D severely attenuates binding of 32D cells and other cells to fibronectin (data not shown).46 One possibility is that P210^BCR-ABL may modify the actin cytoskeleton and influence binding. Alternatively, the role of P210^BCR-ABL as a docking protein to recruit downstream molecules to actin, rather than its tyrosine kinase activity, may contribute to defective adhesion. Transgenic mice expressing P190^BCR-ABL with a deletion of the C-terminal actin-binding domain survive up to 3 times longer and develop hematologic disease with lower frequency compared to wild-type P190^BCR-ABL transgenic mice.47 As shown in our immunohistochemistry studies, the association of P210^BCR-ABL with actin is independent of its kinase activity. Thus, cell adhesion and proper subcellular localization of P210^BCR-ABL may be prerequisites for P210^BCR-ABL tyrosine kinase to initiate CML.

Our results are surprising in light of the wealth of evidence showing that the kinase activity of P210^BCR-ABL is critical to disease initiation and in vitro cell transformation. Thus, it is unlikely that elevated adhesion due to P210^BCR-ABLexpression alone can initiate the cellular changes that lead to CML. The K1176R mutant retains the ability to modulate adhesion to fibronectin exhibited with wt P210^BCR-ABL. However, mice receiving bone marrow transduced with the K1176R mutant fail to develop a CML-like disease under conditions that would normally support disease development with wild-type P210^BCR-ABL.12 This raises the question of which functions of P210^BCR-ABL are required for CML induction. To this end, murine retroviral transduction models of CML have been particularly insightful. Although tyrosine kinase activity is required, mutants that retain tyrosine kinase activity but lack either the oligomerization domain or grb2-binding site are markedly inhibited in their ability to induce CML in murine models.48-50 This suggests that the activated tyrosine kinase activity of P210^BCR-ABL is insufficient for CML induction. Furthermore, we have found that even though the oligomerization domain of P210^BCR-ABL is unnecessary for tyrosine kinase activity, it is required to maintain actin colocalization and to obtain wild-type levels of binding to fibronectin (J.A.W., manuscript in preparation). One possibility is that tyrosine kinase activity, signaling through grb2 binding, and tyrosine kinase-independent regulation of oligomerization effecting actin localization and cell adhesion, are all required for efficient CML induction in murine models. Utilization of the cell detachment device on additional P210^BCR-ABL mutants and primary CML bone marrow should allow us to investigate this and other possibilities.

We thank Martin Carroll, Alan Gewirtz, Ruibao Ren, and members of the Pear Lab for critical reading of the manuscript and helpful suggestions. We are particularly grateful to Andrés Garcı́a and Laura Lynch for their assistance with the spinning disk system and Ruibao Ren and Karen Ehrmann for providing useful reagents.