Oncogene (2009) 28, 4105–4115 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE p210BcrÀAbl desensitizes Cdc42 GTPase signaling for SDF-1a-directed migration in chronic myeloid leukemia cells

Y-C Chang1,3, S-C Tien1, H-F Tien2, H Zhang3, GM Bokoch3 and Z-F Chang1

1Institute of Biochemistry and Molecular Biology, College of Medicine, National Taiwan University, Taipei, Taiwan; 2Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan and 3Department of Immunology and Microbial Science, and Department of Cell Biology, The Scripps Research Institute, La Jolla, CA, USA

Chronic myeloid leukemia (CML) is a lethal hematolo- between the breakpoint-cluster region (bcr) gene on gical disorder caused by the p210BcrÀAbl oncogene. Previous chromosomes 22 and the Abelson leukemia (c-abl) gene studies have suggested that p210BcrÀAbl transformation on chromosome 9 (Rowley, 1990; Sawyers, 1999). contributes to homing and retention defects, typical of Cellular Abl kinase is negatively regulated by its immature myeloid cells in CML, by attenuating chemo- amino-terminal region binding to the carboxyl-terminal tactic response to stromal-derived factor-1a (SDF-1a). As tyrosine kinase domain (Nagar et al., 2003), whereas the Rho family GTPases are key regulators of the cytoske- fused p210BcrÀAbl sequence disrupts the autoinhibitory leton and have been previously found to interact with structure of Abl, leading to a constitutive protein p210BcrÀAbl, this study aimed to determine whether tyrosine kinase (PTK) activity (Deininger et al., 2000; p210BcrÀAbl signaling affects SDF-1a chemotaxis through Ren, 2005). Compelling evidence has shown that Rho GTPase signaling. We found that SDF-1a stimulated tyrosine kinase of p210BcrÀAbl is required to cause Cdc42 GTPase activation in myeloid progenitor 32D, but transformation with uncontrolled proliferation, anti- not in p210BcrÀAbl-transformed (32Dp210) cells. In fact, the apoptosis and decreased adhesion to the bone marrow basal level of active Cdc42 was elevated in 32Dp210 cells (BM) stroma (Deininger et al., 2000; Salesse and and mononuclear cells isolated from bone marrow of Verfaillie, 2002; Wertheim et al., 2002), resulting in CML patients. Inhibition of p210BcrÀAbl kinase activity CML-like disease (Daley et al., 1990; Pear et al., 1998). decreased basal Cdc42 activity and restored SDF-1a- CML cells circulate in large numbers in the blood at induced Cdc42 and migration responses. Transduction of virtually all stages of differentiation, an indicator of active Tat-Cdc42V12 abolished this reconstituted chemo- uncoupling cell differentiation from the ability to leave tactic response. As Cdc42 is particularly important in the BM (Pear et al., 1998). p210BcrÀAbl when expressed in cytoskeletal remodeling and directional sensing, these distinct hematopoietic cell lines, Ba/F3, 32D and Mo7e, results suggest that sustained activation of Cdc42 GTPase significantly impairs their response in the directional through p210BcrÀAbl tyrosine kinase signaling in CML cells migration assay to stromal-derived factor-1a (SDF-1a) contributes to defects in SDF-1a-chemotactic response (Salgia et al., 1999), which is a chemokine constitutively due to desensitization of the actin polarization signal secreted by reticular marrow stromal cells. Similarly, required for directional migration. human CD34 þ progenitor cells expressing p210BcrÀAbl Oncogene (2009) 28, 4105–4115; doi:10.1038/onc.2009.260; also show defects in SDF-1a responsiveness (Ramaraj published online 31 August 2009 et al., 2004). As SDF-1a is the major chemotactic agent that signals through chemokine (CXC motif) receptor 4 Keywords: chronic myeloid leukemia (CML); (CXCR4), a seven transmembrane G-protein-coupled p210BcrÀAbl; Cdc42, chemotaxis; stromal-derived factor- receptor, to induce directional migration, and to attract 1a (SDF-1a) stem and progenitor cells to the BM space (Burger and Burkle, 2007), the reduction of SDF-1a responsiveness in p210BcrÀAbl-transformed cells reflects the homing and retention defects of immature myeloid cells in CML Introduction peripheral blood. However, the mechanism(s) respon- sible for p210BcrÀAbl-mediated changes in chemotactic Human chronic myeloid leukemia (CML) is a malignant migration remain unknown. hematopoietic disease characterized by the Philadelphia Small GTPases of the Rho family, including Rho, Rac chromosome (Ph) due to reciprocal rearrangement and Cdc42, have essential functions in cell migration by transducing signals for cytoskeleton remodeling (Raftopoulou and Hall, 2004; Vicente-Manzanares and Correspondence: Z-F Chang, Correspondence: Institute of Biochem- Sanchez-Madrid, 2004). In migrating cells, Cdc42 and istry and Molecular Biology, National Taiwan University, College of Rac1 establish F-actin-rich finger-like membrane extru- Medicine, No. 1 Section 1 Jen-Ai Road, Taipei, Taiwan, R.O.C. sions (filopodia) and surface protrusions (lamellipodia), E-mail: [email protected] Received 19 February 2009; revised 3 July 2009; accepted 28 July 2009; respectively, in the anterior for protrusive activity, whereas published online 31 August 2009 RhoA-ROCK mediates cortical myosin formation to p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4106 generate retraction force at the posterior of the cell, unchanged in response to SDF-1a. It should be noted moving cells forward (Niggli, 2003; Van Haastert and that basal levels of active Cdc42 and Rac1 were higher in Devreotes, 2004). Chemotaxis is a process in which cells 32Dp210 than in 32D cells with equivalent level of sense gradients of chemoattractants to move toward CXCR4 expression (Figure 1c). higher concentrations of chemoattractants. In neutro- Next, we determined whether the tyrosine kinase phils, chemoattractant-ligated receptors trigger the activity of p210BcrÀAbl is necessary for the upregulation of Gbg-p21 activated kinase (PAK) 1/PAK-interacting ex- Cdc42 and Rac1 in transformed 32Dp210 cells. Cells change factor-a/Cdc42 pathway, which is essential for were treated with Gleevec to inhibit p210BcrÀAbl kinase the localized polymerization of F-actin and the exten- (Deininger et al., 2005) before carrying out Cdc42/Rac1 sion of a leading edge in a gradient-orientated polariza- activity assays. Gleevec treatment markedly decreased tion (Li et al., 2003; Macara, 2004). For SDF-1a- Cdc42 and Rac1 activities in 32Dp210 cells, but had mediated T cell migration, both Rac1 and Cdc42 little effect in parental 32D cells (Figures 2a and b). This activation are required for the chemotactic response suggests that deregulated kinase activity of p210BcrÀAbl (Volinsky et al., 2006). In accordance, the loss of was responsible for the elevation of Cdc42/Rac1 activity interaction between Cdc42 and its downstream effector in 32Dp210 cells. Correlated with Cdc42/Rac1 activa- Wiskott–Aldrich syndrome protein contributes to ab- tion in response to SDF-1a, the parental 32D cells normal SDF-1a chemotaxis in T-lymphocytes of Wis- showed a very clear response to SDF-1a for directed kott–Aldrich syndrome patients (Haddad et al., 2001). migration in the transwell assay, and Gleevec exposure Moreover, homing of hematopoietic progenitor cells to did not alter this response (Figure 2c). In contrast, the BM is also impaired by pre-treatment with lethal 32Dp210 cells, as expected, showed little response to toxin from Clostridium sordellii, which inactivates Rho SDF-1a in directional migration, whereas Gleevec family GTPases, Cdc42 and Rac1, in mice (Genth et al., treatment significantly restored their capacity in SDF- 1996), whereas RhoA inactivation by Rho inhibitor, 1a-directed migration. Thus, there is a reciprocal C2I-C3, does not alter the homing of hematopoietic relationship between basal level of Cdc42/Rac1 activity progenitor cells in mice (Gottig et al., 2006). Herein, we and SDF-1a response in this CML progenitor model. raised the question whether p210BcrÀAbl impairs SDF-1a- Furthermore, we used K562 cells, a p210BcrÀAbl mediated chemotaxis through Cdc42 and/or Rac1 positive myeloid cell line established from a CML pathway. patient at the stage of blast crisis, to test tyrosine kinase- The molecular link between p210BcrÀAbl and Rho dependent activation of Cdc42 and Rac1 GTPases by GTPases has been shown in a number of reports. For p210BcrÀAbl. Results showed that blocking p210BcrÀAbl example, p210BcrÀAbl activates RhoA through its Dbl- kinase activity with Gleevec in K562 cells reduced homology domain in the Bcr region, and its PTK endogenous Cdc42 activity markedly and Rac1 activity located in the Abl region phosphorylates Vav1 for Rac1 to a lesser extent (Figures 2a and b). Furthermore, the activation (Daubon et al., 2008). Vav1, known as a repression of Cdc42 activity by Gleevec treatment proto-oncogene, is predominantly expressed in hemato- seemed to be time- and dosage-dependent in K562 cells poietic cells (Matsuguchi et al., 1995; Hornstein et al., (data not shown), suggesting that p210BcrÀAbl kinase 2004), and tyrosine phosphorylation of Vav1 by activity is required for sustained Cdc42 activation. p210BcrÀAbl in its acidic domain relieves its autoinhibition Owing to the lack of CXCR4 expression, K562 cells by exposing the Dbl-homology domain for interaction did not show SDF-1a response (data not shown). with GDP–Rac1, thereby catalyzing GTP exchange (Aghazadeh et al., 2000; Miletic et al., 2003). Despite the functional implication of p210BcrÀAbl in regulation of Effect of inhibiting p210BcrÀAbl PTK on Cdc42 and Rac1 Rho GTPases, we know little about how p210BcrÀAbl activities, and SDF-1a chemotaxis in Ph( þ ) primary influences chemotactic migration. In this study, we cells from CML patients present evidence that constitutive activation of Cdc42 by To extend these findings to a more clinically relevant p210BcrÀAbl signal is involved in the loss of SDF-1a- condition, we isolated primary mononuclear cells from induced directional migration in CML cells. BM cells of two Ph( þ ) CML patients and treated these cells with Gleevec in serum-containing and serum- free medium for Cdc42/Rac activity assays. Cdc42 Results activity was decreased by more than 30% in both serum-containing and serum-free media by inhibiting p210BcrÀAbl expression alters SDF-1a-induced Cdc42 p210BcrÀAbl kinase (Figure 3a). Meanwhile, endogenous and Rac1 activation Rac1 activity was also reduced by Gleevec treatment in We first compared Cdc42/Rac1 activation after SDF-1a Ph( þ ) cells (Figure 3b). We further used Boyden stimulation in murine myeloid progenitor 32D and chamber assay to examine the effect of Gleevec p210BcrÀAbl-transformed 32D (32Dp210) cells. As ex- treatment on SDF-1a-dependent directional migration pected, 32D cells showed a marked increase in GTP in cells isolated from BM of these two Ph( þ ) patients bound form of Cdc42 and Rac1 on treatment with SDF- (Figure 3c). Consistently, Gleevec treatment was able 1a for 5 min (Figures 1a and b). As a contrast in to increase the migration capacity of the patient CML 32Dp210 cells, however, there was a significant decrease cells toward SDF-1a. Again, these results showed that in Cdc42–GTP level and Rac1–GTP level remained the decrease in Cdc42 and Rac1 activities due to the

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4107 inhibition of tyrosine kinase of p210BcrÀAbl is closely SDF-1a-induced chemotaxis. SDF-1a induced dynamic associated with the increase of SDF-1a-directed migra- finger-like outward projections in 32D cells (Figure 4a) tion in CML cells. and Gleevec treatment had no effect on the changes (Figure 4b and Supplementary movie 1). In contrast, 32Dp210 cells showed little response to SDF-1a Cdc42 GTPase regulates SDF-1a-induced response stimulation (Figure 4b and Supplementary movie 2). We further carried out time-lapse recordings to compare In agreement with data from the transwell assay, the morphological changes of 32D and 32Dp210 cells in Gleevec treatment conferred 32Dp210 cells the dynamic SDF-1a response as seen in 32D cells. Given that Cdc42 regulates cell polarity and Rac works in membrane protrusion of migrating cells by promoting actin polymerization for directional migration, we tranfected 32D cells with GFP-tagged Cdc42N17, Cdc42V12, Rac1N17 or Rac1V12, corresponding to dominant negative and constitutively active forms of Cdc42 and Rac1, respectively, to test their role in SDF-1a- stimulated finger-like protrusions (Figure 4c and Sup- plementary movie 3). We found that overexpression of Cdc42V12 or Cdc42N17, but not Rac1V12 or Rac1N17, was able to diminish SDF-1a-induced finger-like pro- trusions. Clearly, SDF-1a-induced finger-like protru- sions require a tight regulation of Cdc42. We further tested whether Gleevec treatment is able to restore Cdc42 activation by SDF-1a in 32Dp210 cells. The results showed that Cdc42 activity was stimulated by SDF-1a in Gleevec-treated 32Dp210 cells (Figure 4d), which is consistent with the restored finger-like projec- tion (Figure 4b) and migratory response (Figure 2c).

Enforced activation of Cdc42 perturbs restored SDF-1a chemotaxis of Gleevec-treated CML cells The aforementioned results showed that elevated steady- state level of Cdc42 activity in CML cells was inversely correlated with their capacity in SDF-1a-induced direc- tional migration and that deregulation of Cdc42 blocked SDF-1a response. Accordingly, we hypothesized that too much Cdc42 activation in Ph( þ ) cells would contribute to the defect in the SDF-1a-mediated chemotaxis. To validate this hypothesis, we then

Figure 1 Stromal-derived factor-1a (SDF-1a) differentially activates Cdc42 and Rac1 GTPases in 32D and 32Dp210 cells. (a and b) 32D and 32Dp210 cells deprived of serum for 18 h were treated with or without murine SDF-1a (500 ng/ml) for 5 min for measuring the amounts of Cdc42–GTP (a) and Rac1–GTP (b)by GST–PBD (glutathione S-transferase–Rac/Cdc42 binding do- main of p21 activated kinase 1 protein) pull-down assays. The western blot analysis was performed using Cdc42 and Rac1 antibodies. The top panel shows the western blot analysis for Cdc42 (a) and Rac1 (b) in the GTP-bound form that associates with GST–PBD and 8% of total input. By densitometric scanning, active Rho GTPase was measured and its percentage relative to total Rho GTPase was calculated. Average values from three individual experiments are shown below the western blot data. The intensity ratio of Rho GTPase in GTP-bound form to total Rho GTPase in lysate of 32D cells without SDF-1a treatment was set to 1, and the relative fold of active Rho GTPase for each condition is shown in the lower panel as means±s.d. (n ¼ 3) (‘*’ indicates Po0.05; ‘**’ indicates Po0.01). (c) The expression of chemokine (CXC motif) receptor 4 (CXCR4) in 32D and 32Dp210 cells was detected by western blot using anti- CXCR4 antibody. Actin is shown as a loading control.

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4108 transduced the constitutively active form of Tat- directed migration and Gleevec treatment had no effect Cdc42V12 into freshly isolated mononuclear cells from on their SDF-1a responsiveness (Figure 5a). Addition of leukemia patient that were verified to be Ph chromo- Tat-Cdc42V12 increased random migration, and abro- some negative (Ph(À)) for the transwell assays. As gated SDF-1a-directed responsiveness. As for mono- expected, Ph(À) leukemia cells were capable of SDF-1a- nuclear cells isolated from BM of four different Ph( þ ) CML patients, these CML cells consistently showed a defect in SDF-1a-stimulated directional migration and Gleevec treatment restored their SDF-1a responsiveness (Figure 5b). Similar to Ph(À) cells, addition of Tat- Cdc42V12 increased random motility of primary Ph( þ ) cells in all cases and abolished SDF-1a-responsiveness in Gleevec-treated Ph( þ ) CML cells, supporting our idea that sustained activation of Cdc42 is able to impair SDF-1a-responsive directional migration.

Vav1 mediates p210BcrÀAbl-induced Cdc42 activation It has been reported that Vav1 serves as a direct PTK substrate of p210BcrÀAbl (Matsuguchi et al., 1995) and forms a complex with the Abl portion of p210BcrÀAbl for Rac1 stimulation (Bassermann et al., 2002; Daubon et al., 2008). In addition, phosphorylation and activa- tion of Vav1, induced by SDF-1a, controls the chemotactic response in human primary lymphocytes (Vicente-Manzanares et al., 2005). To verify the relationship between deregulated kinase activity of p210BcrÀAbl and Vav1 activation for Cdc42/Rac1 upre- gulation, we first analyzed the phosphorylation status of immunoprecipitated Vav1 in 32D, 32Dp210 and K562 cells in response to Gleevec treatment. In agreement, the extent of Vav1 phosphorylation was clearly higher in 32Dp210 and K562 cells, and Gleevec treatment abolished the p210BcrÀAbl-promoted phosphorylation signal of Vav1 (Figure 6a). We then transfected 32D cells with an expression vector of Vav1 deletion mutant (D1–186), which is a constitutively active Vav1, inde- pendent of phosphorylation control (Lopez-Lago et al.,

Figure 2 Inhibition of p210BcrÀAbl tyrosine kinase inactivates Cdc42 and Rac1 GTPases and restores SDF-1a chemotaxis. (a and b) Murine 32D and 32Dp210 cells were maintained in serum-free medium for 12 h and then treated with or without Gleevec (2 mM) for additional 6 h. Human K562 cells were incubated in serum-free medium in the presence or absence of Gleevec (5 mM) for 18 h. Cells were then collected for GST–PBD pull-down assays. The representative amounts of active and total Cdc42 (a) and Rac1 (b) are shown in the top panel. The intensity ratio of Rho GTPase in GTP-bound form to total Rho GTPase in lysate of 32D/K562 cells without Gleevec treatment was set to 1, and the relative fold of active Rho GTPase for each condition is shown in the lower panel as means±s.d. (n ¼ 3) (‘*’ indicates Po0.05; ‘**’ indicates Po0.01). (c) 32D and 32Dp210 cells treated as described above were then re-suspended in fresh medium and seeded in triplicate into the upper chamber of the transwell. The lower chamber of the transwell was filled with serum-free medium with or without murine SDF-1a (100 ng/ml). After 4 h, transmigrated cells recovered from the lower chamber were counted. The percentage of migration was calculated by dividing the transmigrated number by the total cells in the initial suspension (105). Data are represented as mean±s.d. of six transwell experiments from two independent treatments (n ¼ 6) (‘*’ indicates Po0.05; ‘**’ indicates Po0.01).

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4109

Figure 3 Inhibition of p210BcrÀAbl tyrosine kinase decreases Cdc42/Rac1 activities and restores chemotaxis in primary Ph( þ ) cells. (a and b) Primary mononuclear cells isolated from BM samples were maintained in serum-containing or serum-free RPMI 1640 medium treated with or without Gleevec (5 mM) for 18 h. Representative CML cells from Ph( þ )1 and Ph( þ )2 patients were collected and lysed for Cdc42/Rac1 activity assays. The top panel shows the GTP-bound Cdc42 (a) and Rac1 (b), and the total amount of GTPases in 10% of total cell extract from Ph( þ )1 patient. The intensity ratio of Rho GTPase in GTP-bound form to total Rho GTPase in the lysate of primary cells in serum-containing medium without treatment was set to 1, and the relative fold of active Rho GTPase for each condition is shown from Ph( þ )1 (n ¼ 1) and Ph( þ )2 (n ¼ 2) patients in the lower panel. (c) CML cells from Ph( þ )1 and Ph( þ )2 patients treated in serum-free medium as described above were re-suspended in fresh serum-free medium and then added to the upper chamber of the transwell in triplicate. The lower chamber of the transwell was filled with serum-free medium with or without human SDF-1a (200 ng/ml). After 4 h of incubation, transmigrated cells recovered from the lower chamber were counted as described in Materials and methods section. Data are represented as mean±s.d. of six transwell assays from two independent treatment experiments (‘**’ indicates Po0.01).

2000). The results showed that overexpression of Discussion Vav1(D1–186) was sufficient to activate Cdc42 in 32D cells (Figure 6b). Conversely, Cdc42 activity was Rho family GTPases, RhoA, Rac1 and Cdc42, have a decreased in 32Dp210 cells when Vav1 expression level pivotal role in cytoskeleton rearrangement during was decreased by Vav1-specific shRNA transfection directional migration. However, whether or not (Figure 6c), indicating the essential role of Vav1 in the p210BcrÀAbl transformation disturbs SDF-1a response activation of Cdc42 in these transformed cells. through these Rho GTPases remains undefined. In this

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4110

Figure 4 SDF-1a-induced finger-like projection is abolished by p210BcrÀAbl-mediated deregulation of Cdc42. (a and b) 32D and 32Dp210 cells deprived of serum for 18 h were re-suspended in serum-free medium and plated onto fibronectin-coated glass slide assembled in the POC-R open cultivation chamber. Cells were then treated with murine SDF-1a (25 ng/ml) for time-lapse recording at 15-s intervals and were imaged using  100 DIC objective for 20 min. The morphology of representative 32D cells treated with or without SDF-1a was shown as indicated in (a). The total numbers of finger-like projection from 80 frames were counted by MetaMorph software and the projection numbers are shown as the mean±s.d. of five cells from at least three independent treatments (‘**’ indicates Po0.01) (b). (c) 32D cells were transfected with green fluorescent protein (GFP) or GFP-tagged dominant active or negative Cdc42 and Rac1 plasmids. Transfected cells were re-suspended in serum-free medium for 18 h and SDF-1a-induced finger-like projections were observed and quantified as described above (n ¼ 5). (d) 32Dp210 cells deprived of serum for 15 h were then treated with or without Gleevec (2 mM) for 3 h. Cells were then stimulated with or without murine SDF-1a (500 ng/ml) for 5 min for measuring the amounts of Cdc42–GTP by GST—PBD pull-down assays and the western blot using Cdc42 antibody. The representative western blot in top panel shows the Cdc42 in the GTP-bound form that associates with GST–PBD and 7% of total input. By densitometric scanning, the intensity ratio of Cdc42 in GTP-bound form to total Cdc42 in lysate of 32Dp210 cells with Gleevec treatment was set to 1, and the relative fold of active Cdc42 for each condition is shown in the lower panel as means±s.d. of three individual experiments (‘*’ indicates Po0.05).

study, we provide new insights into the interplay hematopoietic cells in BM might sense the SDF-1a between p210BcrÀAbl and Cdc42 in the regulation of the signal to initiate receptor-mediated Cdc42 and Rac1 chemotactic migration of CML progenitors. Although activation, resulting in cytoskeletal polarization as Cdc42 activation is required for cell motility, here our manifested by the increase in filopodia- and lamellipo- results suggest that sustained activation of Cdc42 by dia-like structures to orientate cell movement and p210BcrÀAbl signaling in CML cells disrupts SDF-1a- homing. One possible mechanism contributory to the induced polarization signal, offering a possible explana- deficiency of SDF-1a responsiveness in p210BcrÀAbl- tion for premature circulation of undifferentiated transformed and CML cells could be the loss of CXCR4 myeloid cells without BM retention in CML disease. function or expression in these cells. It has been shown The chemoattractant, SDF-1a, and its receptor, that p210BcrÀAbl utilizes its tyrosine kinase activity to CXCR4, are essential for myeloid and lymphoid inhibit SDF-1a chemotactic response through down- development. Defects in SDF-1a responsiveness in regulation of CXCR4 mRNA expression in blastic- p210BcrÀAbl-transformed cells, therefore, have certain phase CML cells and MO7e cells expressing high level of functions in the pathogenesis of CML. Theoretically, p210BcrÀAbl (Geay et al., 2005). However, our data

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4111

Figure 5 Sustained activation of Cdc42 signaling impairs SDF-1a chemotaxis. (a and b) Isolated mononuclear cells from Ph(À)(a) and Ph( þ )(b) BM were treated with or without Gleevec in serum-free RPMI 1640 medium for 18 h. Cells were then re-suspended in fresh serum-free medium with or without Tat-Cdc42V12 (20 mg/ml) for 10 min and then seeded in the transwells in triplicate. The lower chamber of the transwell was filled with serum-free medium with or without human SDF-1a (200 ng/ml). After 4 h of incubation, transmigrated cells recovered from the lower chamber were counted as described in Materials and methods. Data are represented as mean±s.d. of six transwell assays from two independent treatment experiments (n ¼ 6) (‘*’ indicates Po0.05; ‘**’ indicates Po0.01).

indicate that neither p210BcrÀAbl transformation in the lated directional migration. Unlike 32D progenitor cells 32D progenitor model nor p210BcrÀAbl inhibition by in which Cdc42 and Rac1 activities were increased by Gleevec in CML cells from chronic-phase patients SDF-1a stimulation, p210BcrÀAbl-transformed 32Dp210 had any effect on the modulation of CXCR4 protein cells showed no change in Rac1 activity, but an expression. Moreover, p210BcrÀAbl inhibition was able unexpected decline in Cdc42 activity. At present, we to restore SDF-1a response, suggesting the intact do not know why Cdc42 activity was specifically function of CXCR4 in p210BcrÀAbl transformation. decreased in 32Dp210 cells after SDF-1a treatment. Another possibility is related to the defective signaling Nevertheless, our results, at least, indicated the SDF-1a- in either SDF-1a-induced Cdc42 or Rac activation in induced activation of Cdc42 and Rac1 is defective after CML cells. As a matter of fact, Cdc42 and Rac1 were p210BcrÀAbl transformation. constitutively activated in CML cells. Inhibition of RhoA, Rac1 and Cdc42 GTPases have been shown to p210BcrÀAbl by Gleevec markedly decreased Cdc42 be activated by p210BcrÀAbl transformation through its activity with a concurrent increase of SDF-1a-stimu- Dbl-homology domain of Bcr and its target Vav1 for the

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4112

Figure 6 The involvement of Vav1 in p210BcrÀAbl -mediated Cdc42/Rac1 GTPase activation. (a) 32D, 32Dp210 and K562 cells treated with or without Gleevec as described in Figure 2 were lysed and subjected to immunoprecipitation for Vav1. Immunoprecipitates were separated by SDS–PAGE, immunoblotted with antibody against P-Tyr (PY), and then re-probed with Vav1 antibody. (b) 32D cells transfected with GFP or GFP–Vav1(D1–186) were maintained in culture medium for 8 h. Transfection efficiency was higher than 60% in all experiments examined by Olympus AX-70 fluorescence microscopy by detecting GFP-positive population. Cells were then subjected to GST–PBD pull-down assays. Rho GTPases in GTP-bound form and in total lysates were analyzed by western blot with anti-Cdc42 antibody and re-probing with anti-Rac1 antibody. By densitometric scanning, the relative fold of active to total Rho GTPase in each lysates was determined from three independent experiments (mean±s.d.) shown in the left panel. Expression levels of GFP and GFP–Vav1(D1–186) were confirmed by western blot using anti-GFP and anti-Vav1 antibodies in the right panel. Actin is shown as a loading control. The molecular weight of GFP–Vav1(D1–186) is the same as endogenous Vav1, and is also detected by Vav1 antibody (c) 32Dp210 cells transfected with control LacZ or Vav1-specific shRNA plasmid were maintained in serum-containing culture medium for 48 h. Cells were then subjected to GST–PBD pull-down assays for the detection of Cdc42 and Rac1 activities. The left panel shows the representative blot and quantification data of Cdc42 and Rac1 activities from six independent assays. The right panel shows representative Vav1 knockdown efficiency in total lysates using anti-Vav1 antibody. Actin is shown as a loading control. (d) Proposed model of p210BcrÀAbl in chemotaxis. p210BcrÀAbl deregulates Cdc42 GTPase through phosphorylation and activation of Vav1, which abrogates SDF-1a-induced Cdc42 activation, thus blocking chemotaxis in chronic myeloid leukemia (CML).

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4113 guanine nucleotide exchange factor (GEF) activity in purified by sequential Ni–NTA agarose pull-down assay, hematopoietic BaF3 cells (Harnois et al., 2003; Daubon Imidazole elution and phosphate buffered saline dialysis as et al., 2008). However, RhoA activity was almost previously described (Chang et al., 2006). undetectable in BM-derived Ph( þ ) primary cells (data not shown). Although p210BcrÀAbl phosphorylates Vav1 to Cell lines and primary cells activate Rac1 in 293 cells (Bassermann et al.,2002),we Human erythroleukemia cell line K562 was maintained in foundlessalterationofRac1ascomparedwithCdc42 RPMI 1640 (Gibco-BRL, Eggenstein, Germany) supplemented activity in K562 cells after Gleevec treatment. As for with 10% heat-inactivated fetal bovine serum (Hyclone, 32Dp210-transformed cells, inhibition of p210BcrÀAbl also Logan, UT, USA), 2 mML-glutamine, 100 U/ml penicillin and 100 U/ml streptomycin. The murine WEHI and abolished Vav1 phosphorylation but with both Cdc42 and p210BcrÀAbl-transformed 32Dp210 cells were kindly provided Rac1 activities being suppressed. Thus, the correlation by Dr RB Arlinghaus (The University of Texas MD Anderson BcrÀAbl between p210 -dependent phosphorylation of Vav1, Cancer Center, Houston, TX, USA.) and cultured in RPMI and Cdc42 and Rac1 activation varies in different cell 1640 medium supplemented with 10% heat-inactivated fetal types. As decreasing the expression of Vav1 by shRNA bovine serum, 2 mML-glutamine, 100 U/ml penicillin and transfection downregulated Cdc42/Rac1 activities, acti- 100 U/ml streptomycin, 1.5 g/l sodium bicarbonate, 10 mM vation of Vav1 by p210BcrÀAbl signaling contributed to HEPES and 1 mM sodium pyruvate. Mouse 32D (32Dcl3) the elevation of Cdc42/Rac1 activities in 32Dp210- cells were purchased from American Type Culture Collection transformed cells. Consistent with what was observed in and maintained in the same medium used for transformed 32Dp210 cells, Gleevec treatment in CML cells from 32Dp210 cell line with 25% WEHI-cultured conditioned medium as a source of interleukin-3. Ph( þ ) chronic phase patients decreased Cdc42 and Fresh primary leukemia samples without any advanced Rac1 activities and restored SDF-1a-directed migra- chemotherapy were collected from BM of Ph( þ )(n ¼ 4) and BcrÀAbl tion, confirming the contribution of p210 PTK Ph(À)(n ¼ 1) patients during medical evaluations. Informed activity to deregulation of these actin polymerization consent was approved by the Institutional Review Board. signals. Mononuclear cells were isolated using Ficoll-Paque PRE- Noticeably, the SDF-1a-induced finger-like protru- MIUM (Amersham Biosciences, Uppsala, Sweden) density sions were more sensitive to deregulation of Cdc42 gradient separation and were maintained in the same RPMI instead of Rac1. Interestingly, both p210BcrÀAbl transfor- 1640 medium used for K562 cells. mation and SDF-1a stimulation resulted in about a twofold increase in Cdc42 GTPase in 32D cells. Endogenous Rho GTPase activity assay Conversely, inhibition of p210BcrÀAbl in 32Dp210 cells Activation of Rho GTPases is manifested by their capacity in resulted in about twofold decrease in Cdc42 activity and binding the interacting domain present in their corresponding restored Cdc42 activation in response to SDF-1a. effector protein. In this study, the GST–PBD (GST, glutathione S-transferase; PBD, Rac/Cdc42 binding domain Likely, a tight control of Cdc42 activation magnitude of p21 activated kinase 1 protein) pull-down assays were used is crucial for the cytoskeletal changes in chemotaxis and to detect cellular GTP-bound Rac1 and Cdc42, respectively BcrÀAbl is interfered by kinase function of p210 . In support (Stofega et al., 2006). In brief, cells were lysed in a buffer of this view, transduction of Tat-Cdc42V12 in CML containing 50 mM Tris–HCl (pH 7.2), 1% (v/v) Triton X-100, cells from patients was sufficient to abrogate the Gleevec 0.5% sodium deoxycholate, 0.1% SDS, 500 mM NaCl, 10 mM effect on restoring SDF-1a-directed migration. Particu- MgCl2,1mM phenylmethanesulphonyl fluoride (PMSF) and larly, it is noted that random migration of all myeloid 100 Â dilution of protease inhibitor cocktail (Sigma, St Louis, cells was increased by Tat-Cdc42V12, suggesting that MO, USA). The supernatants of the lysates were incubated at deregulated Cdc42 activation promotes random motility 4 1C for 1.5 h with GST–PBD-coupled glutathione–sepharose and impairs chemotaxis. beads. The beads were washed three times with buffer containing 50 mM Tris–HCl (pH 7.2), 1% (v/v) Triton X-100, Considering the essential role of Cdc42 regulation in 150 mM NaCl, 10 mM MgCl2, 0.1 mM PMSF and 1000 Â actin polymerization and cell polarity, here we proposed dilution of protease inhibitor cocktail. The amounts of total BcrÀAbl that oncogenic p210 transformation deregulates and GTP-bound Rho GTPases were detected by western Cdc42 activation to desensitize SDF-1a-induced polar- blotting with an antibody against Rac1 (Millipore/Upstate, ized signal required for directional migration, thereby Billerica, MA, USA) or Cdc42 (Millipore/Upstate or Santa contributing to the homing and retention defects seen in Cruz Biotechnology, Santa Cruz, CA, USA). CML (Figure 6d). Transwell chemotaxis assay The assays were carried out by using 8-mm pore filter (Transwell, 24-well plate; Corning Costar, Lowell, MA, Materials and methods USA) for indicated conditions. Murine 32D and 32Dp210 cells were maintained in migration medium (serum-free RPMI Chemicals and Tat-Cdc42V12 protein 1640 medium) for 12 h and then treated with or without Bcr–Abl tyrosine kinase inhibitor Gleevec (also named as Gleevec (2 mM) for additional 6 h. Primary human BM cells STI571, imatinib mesylate and Glivec) (Novartis AG, were pre-incubated in migration medium with or without 5 Basel, Switzerland) was dissolved in ddH2O. Recombinant Gleevec (5 mM) for 18 h. Cells (10 ) were then re-suspended in chemokine human or murine SDF-1a (Cytolab/PeproTech, migration medium and placed in the upper chamber of the Rocky Hill, NJ, USA) were dissolved in phosphate buffered transwell in triplicate. The lower chamber of the transwell was saline containing 0.1% bovine serum albumin. Recombinant filled with 600 ml migration medium with or without murine Tat-Cdc42V12 protein was produced in Escherichia coli and SDF-1a (100 ng/ml), or with or without human SDF-1a

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4114 (200 ng/ml). After 4 h of incubation, the upper chamber was Vav1 immunoprecipitation carefully removed, and cells that had migrated to the bottom Cells were collected in ice-cold lysis buffer (50 mM Tris–HCl chamber were re-suspended and manually counted using a (pH 7.5), 150 mM NaCl, 5 mM MgCl2, 1% Triton X-100, 1 mM hemocytometer. The percentage of migrated cells was calcu- EGTA, 1 mM EDTA, 40 mM NaF, 2 mM Na3VO4,1mM PMSF lated as: (number of migrated cells/105) Â 100% and the results and 100 Â dilution of protease inhibitor cocktail) (Wilkinson were obtained from two independent treatment assays and et al., 2006). Supernatants from lysates (1 mg of total protein) shown as mean±s.d. (n ¼ 6). were incubated with the anti-Vav1 antibody or normal rabbit IgG for 2 h at 4 1C and protein A–agarose were added for Transient electroporation transfection additional 0.5 h. Immune complexes were washed three times DNA transfection in 32D or 32Dp210 cells was carried out in lysis buffer. The immunoprecipitates were subjected to using MicroPorator MP-100 (Digital Bio Technology, Seoul, SDS–PAGE immunoblot analysis using the anti-Vav1 (Santa Korea). A total of 5 Â 106 cells were suspended in 100 ml Cruz Biotechnology) and anti-P-Tyr (clone 4G10, Millipore/ Resuspension buffer with 10 mg transfection DNA and trans- Upstate) antibodies. ferred to a Gold-Tip, followed by electroporation in electrolytic buffer at 1350 V per 30 ms. Cells were then maintained in Conflict of Interest cultured medium without antibiotics for at least 8 h. The authors declare no conflict of interest. Differential interference contrast (DIC) microscopy for live-cell imaging Live-cell imaging was carried out using MetaMorph 7.1 Acknowledgements software to control a digital imaging system coupled to an inverted microscope (Axiovert 200M; Zeiss, Jena, Germany). We are grateful to Dr RB Arlinghaus (Department of A high-sensitivity CoolSNAP HQ CCD camera acquired Molecular Pathology, The University of Texas MD Anderson images using the MetaMorph software. Live-cell imaging Cancer Center, Houston, Texas, USA) for providing 32Dp210 for cells cultured on the POC-R2 open cultivation chamber and WEHI cells. We thank the staff of the Second Core Lab, (PeCon, Erbach, Germany) was carried out by 100 Â DIC Department of Medical Research, National Taiwan University PlanApo oil objective (NA 1.3). Time lapse DIC video Hospital for technical support during this study. This study recordings were acquired every 15 s for 20 min (80 continuous is supported by grant NSC 96-3112-B-002-006 and NSC 97- frames in total) on murine SDF-1a (25 ng/ml) stimulation. 3112-B-002-026 from the National Science Council, Taiwan.

References

Aghazadeh B, Lowry WE, Huang XY, Rosen MK. (2000). Structural Gottig S, Mobest D, Ruster B, Grace B, Winter S, Seifried E et al basis for relief of autoinhibition of the Dbl homology domain of (2006). Role of the monomeric GTPase Rho in hematopoietic proto-oncogene Vav by tyrosine phosphorylation. Cell 102: 625–633. progenitor cell migration and transplantation. Eur J Immunol 36: Bassermann F, Jahn T, Miething C, Seipel P, Bai RY, Coutinho S et al 180–189. (2002). Association of Bcr-Abl with the proto-oncogene Vav is Haddad E, Zugaza JL, Louache F, Debili N, Crouin C, Schwarz K implicated in activation of the Rac-1 pathway. J Biol Chem 277: et al (2001). The interaction between Cdc42 and WASP is 12437–12445. required for SDF-1-induced T-lymphocyte chemotaxis. Blood 97: Burger JA, Burkle A. (2007). The CXCR4 chemokine receptor in acute 33–38. and chronic leukaemia: a marrow homing receptor and potential Harnois T, Constantin B, Rioux A, Grenioux E, Kitzis A, therapeutic target. Br J Haematol 137: 288–296. Bourmeyster N. (2003). Differential interaction and activation of Chang YC, Lee HH, Chen YJ, Bokoch GM, Chang ZF. (2006). Rho family GTPases by p210bcr–abl and p190bcr–abl. Oncogene Contribution of guanine exchange factor H1 in phorbol ester- 22: 6445–6454. induced apoptosis. Cell Death Differ 13: 2023–2032. Hornstein I, Alcover A, Katzav S. (2004). Vav proteins, masters of the Daley GQ, Van Etten RA, Baltimore D. (1990). Induction of chronic world of cytoskeleton organization. Cell Signal 16: 1–11. myelogenous leukemia in mice by the P210bcr/abl gene of the Li Z, Hannigan M, Mo Z, Liu B, Lu W, Wu Y et al (2003). Directional Philadelphia chromosome. Science 247: 824–830. sensing requires G beta gamma-mediated PAK1 and PIX alpha- Daubon T, Chasseriau J, Ali AE, Rivet J, Kitzis A, Constantin B et al dependent activation of Cdc42. Cell 114: 215–227. (2008). Differential motility of p190bcr-abl- and p210bcr-abl- Lopez-Lago M, Lee H, Cruz C, Movilla N, Bustelo XR. (2000). expressing cells: respective roles of Vav and Bcr-Abl GEFs. Tyrosine phosphorylation mediates both activation and Oncogene 27: 2673–2685. downmodulation of the biological activity of Vav. Mol Cell Biol Deininger M, Buchdunger E, Druker BJ. (2005). The development of 20: 1678–1691. imatinib as a therapeutic agent for chronic myeloid leukemia. Blood Macara IG. (2004). Parsing the polarity code. Nat Rev Mol Cell Biol 5: 105: 2640–2653. 220–231. Deininger MW, Goldman JM, Melo JV. (2000). The molecular biology Matsuguchi T, Inhorn RC, Carlesso N, Xu G, Druker B, Griffin JD. of chronic myeloid leukemia. Blood 96: 3343–3356. (1995). Tyrosine phosphorylation of p95Vav in myeloid cells is Geay JF, Buet D, Zhang Y, Foudi A, Jarrier P, Berthebaud M et al regulated by GM-CSF, IL-3 and steel factor and is constitutively (2005). p210BCR-ABL inhibits SDF-1 chemotactic response via increased by p210BCR/ABL. EMBO J 14: 257–265. alteration of CXCR4 signaling and down-regulation of CXCR4 Miletic AV, Swat M, Fujikawa K, Swat W. (2003). Cytoskeletal expression. Cancer Res 65: 2676–2683. remodeling in lymphocyte activation. Curr Opin Immunol 15: Genth H, Hofmann F, Selzer J, Rex G, Aktories K, Just I. (1996). 261–268. Difference in protein substrate specificity between hemorrhagic Nagar B, Hantschel O, Young MA, Scheffzek K, Veach D, Bornmann toxin and lethal toxin from Clostridium sordellii. Biochem Biophys W et al (2003). Structural basis for the autoinhibition of c-Abl Res Commun 229: 370–374. tyrosine kinase. Cell 112: 859–871.

Oncogene p210BcrÀAbl interrupts Cdc42-mediated chemotaxis Y-C Chang et al 4115 Niggli V. (2003). Signaling to migration in neutrophils: importance of Stofega M, DerMardirossian C, Bokoch GM. (2006). Affinity-based localized pathways. Int J Biochem Cell Biol 35: 1619–1638. assay of Rho guanosine triphosphatase activation. Methods Mol Pear WS, Miller JP, Xu L, Pui JC, Soffer B, Quackenbush RC et al Biol 332: 269–279. (1998). Efficient and rapid induction of a chronic myelogenous Van Haastert PJ, Devreotes PN. (2004). Chemotaxis: signalling the leukemia-like myeloproliferative disease in mice receiving P210 bcr/ way forward. Nat Rev Mol Cell Biol 5: 626–634. abl-transduced bone marrow. Blood 92: 3780–3792. Vicente-Manzanares M, Cruz-Adalia A, Martin-Cofreces NB, Raftopoulou M, Hall A. (2004). Cell migration: Rho GTPases lead the Cabrero JR, Dosil M, Alvarado-Sanchez B et al (2005). Control way. Dev Biol 265: 23–32. of lymphocyte shape and the chemotactic response by the GTP Ramaraj P, Singh H, Niu N, Chu S, Holtz M, Yee JK et al (2004). exchange factor Vav. Blood 105: 3026–3034. Effect of mutational inactivation of tyrosine kinase activity on Vicente-Manzanares M, Sanchez-Madrid F. (2004). Role of BCR/ABL-induced abnormalities in cell growth and adhesion in the cytoskeleton during leukocyte responses. Nat Rev Immunol 4: human hematopoietic progenitors. Cancer Res 64: 5322–5331. 110–122. Ren R. (2005). Mechanisms of BCR-ABL in the pathogenesis of Volinsky N, Gantman A, Yablonski D. (2006). A Pak- and chronic myelogenous leukaemia. Nat Rev Cancer 5: 172–183. Pix-dependent branch of the SDF-1alpha signalling pathway Rowley JD. (1990). The Philadelphia chromosome translocation. A mediates T cell chemotaxis across restrictive barriers. Biochem paradigm for understanding leukemia. Cancer 65: 2178–2184. J 397: 213–222. Salesse S, Verfaillie CM. (2002). BCR/ABL: from molecular mechan- Wertheim JA, Miller JP, Xu L, He Y, Pear WS. (2002). The biology of isms of leukemia induction to treatment of chronic myelogenous chronic myelogenous leukemia:mouse models and cell adhesion. leukemia. Oncogene 21: 8547–8559. Oncogene 21: 8612–8628. Salgia R, Quackenbush E, Lin J, Souchkova N, Sattler M, Ewaniuk Wilkinson B, Koenigsknecht-Talboo J, Grommes C, Lee CY, Land- DS et al (1999). The BCR/ABL oncogene alters the chemotactic reth G. (2006). Fibrillar beta-amyloid-stimulated intracellular response to stromal-derived factor-1alpha. Blood 94: 4233–4246. signaling cascades require Vav for induction of respiratory burst Sawyers CL. (1999). Chronic myeloid leukemia. N Engl J Med 340: and phagocytosis in monocytes and microglia. J Biol Chem 281: 1330–1340. 20842–20850.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene