Gleevec, an Abl Family Inhibitor, Produces a Profound Change in Cell Shape and Migration

The issue of how contractility and adhesion are related to cell shape and migration pattern remains largely unresolved. In this paper we report that Gleevec (Imatinib), an Abl family kinase inhibitor, produces a profound change in the shape and migration of rat bladder tumor cells (NBTII) plated on collagen-coated substrates. Cells treated with Gleevec adopt a highly spread D-shape and migrate more rapidly with greater persistence. Accompanying this more spread state is an increase in integrin-mediated adhesion coupled with increases in the size and number of discrete adhesions. In addition, both total internal reflection fluorescence microscopy (TIRFM) and interference reflection microscopy (IRM) revealed a band of small punctate adhesions with rapid turnover near the cell leading margin. These changes led to an increase in global cell-substrate adhesion strength, as assessed by laminar flow experiments. Gleevec-treated cells have greater RhoA activity which, via myosin activation, led to an increase in the magnitude of total traction force applied to the substrate. These chemical and physical alterations upon Gleevec treatment produce the dramatic change in morphology and migration that is observed.


Introduction
The study of cell migration is essential for understanding a variety of processes including wound repair, immune response and tissue homeostasis; importantly, aberrant cell migration can result in various pathologies [1,2,3]. However, the relationship between cytoskeletal dynamics, including actin network growth, contractility, and adhesion, to cell shape and migration remains incompletely understood.
Abl family tyrosine kinases are ubiquitous non-receptor tyrosine kinases (NRTKs) involved in signal transduction [4,5,6]. They can interact with other cellular components through multiple functional domains for filamentous and globular actin binding, as well as through binding phosphorylated tyrosines (SH2), polyproline rich regions (SH3), DNA (Abl), and microtubules (Abl Related Gene (Arg)) [7,8]. Abl family tyrosine kinases have also been found to regulate cell migration [8,9]. In some cases, Abl family kinases have been reported to promote actin polymerization and migration [10] as well as filopodia formation during cell spreading [11,12]. By contrast, in other studies Abl was found to restrain lamellipodia extension [13,14] or inhibit initial cell attachment to the substrate [15]. Abl family kinases have been suggested to regulate cell adhesion size and stress fiber formation [16]; Li and Pendergast recently reported that the Abl family member Arg, could disrupt CrkII-C3G complex formation to reduce b1-integrin related adhesion formation [17]. Thus, a complete understanding of how Abl family kinases regulate cell migration is lacking [8,9].
In this study, we report that Gleevec (also called Imatinib/ STI571), an Abl family kinase inhibitor that is used as a chemotherapeutic agent for leukemia, produces a profound change in the shape and migration of the rat Nara bladder tumor (NBT-II) cells plated on collagen-coated substrates. Within 20 min of Gleevec treatment the majority of NBT-II cells develop a new D-shaped morphology and start migrating more rapidly and with greater persistence. The new morphology is characterized by stronger cellsubstrate adhesion and an increase in the size and number of discrete adhesions which at the leading margin turnover more rapidly. RhoA activity in Gleevec-treated cells was increased which, via myosin activation, led to an increase in the magnitude of total traction forces applied to the substrate. Upon Gleevec treatment, these chemical and physical alterations combined to produce the dramatic change in morphology and migration.

Treatment with Gleevec induces a D-shaped morphology in NBTII cells
The morphology of a migrating cell is related to cell migration modes. NBTII is a rat-derived carcinoma cell line [18]. A normal cultured NBTII cell shows typical epithelial morphology; however, when NBTII cells were cultured on type I collagen-coated plastic cell culture dishes for 4-12 h, they acquired a polarized shape and migrate individually, exhibiting an epithelial to mesenchymal transition (EMT) [19,20,21,22]). During our experiments, we observed that NBTII cells on collagen had medium-sized lamellae (Marked with ''LM'') and lamellipodia (Marked with ''LP''), some filopodia (Marked with ''FP'') dynamically formed at the leading edge of the cell, and multiple retraction fibers (Marked with ''RF'') formed at the trailing edge of the cell. ( Figure 1A, Movie S1). Figure 1B shows NBTII cells cultured on type I collagen for 8 h and then treated with 20 mM Gleevec, an inhibitor of the Abl family of NRTK (Novartis, Stein, Switzerland) for 30 minutes [23,24]). Within about 10 minutes of Gleevec addition, a profound change in cell morphology can be observed (Movie S2). These kinetics can be seen by the change in area of Gleevec-treated cells occurring after about 5 minutes ( Figure S1). Cells began to form lamellipodial protrusions, which usually merged into a single, intact lamella facing the migration direction (Movie S3). This cell morphology may persist for over 8 hours. After Gleevec treatment, cells had reduced numbers of both filopodia and retraction fibers. The actin and microtubule cytoskeleton of NBTII cells or Gleevectreated NBTII cells differed somewhat ( Figure 1C and 1D); particularly noticeable were the number of f-actin rich retraction fibers in the control cells. Gleevec-treated NBTII cells had about a 75% increase in migration speed compared with control NBTII cells ( Figure 1E) and maintained their direction significantly better than control NBTII cells ( Figure 1F) (Movie S4).
Gleevec treatment induced a pronounced change in cell morphology when compared to control cells ( Figure 1 and Movies S1, S2, S3). To better determine the changes in cell morphology, we used four parameters as defined in Figure 2B: (1) Aspect ratio, the ratio of the widest dimension of the cell in the direction perpendicular to the direction of migration divided by the longest dimension of the cell in the direction of cell migration; (2) Nuclear aspect ratio, the aspect ratio of the cell nuclear region; (3) Area ratio, the ratio of the total cell area to the nuclear area; and (4) Retraction fiber length, the population average of the sum of the length of all retraction fibers in one individual cell divided by the same parameter calculated for the control cells. The results of our analysis revealed that cells treated with Gleevec had significantly increased aspect ratio, nuclear ratio, and area ratio, while having a reduced retraction fiber length ratio ( Figure 2A). Interestingly, unlike most known mesenchymal migrating cells, which are typically elongated in the direction of migration, Gleevec-treated cells were elongated in the direction perpendicular to their movement and showed visual similarity to fish or amphibian keratocytes [25,26,27]

Both Gleevec concentration and substrate adhesiveness affect NBTII cell migration
To investigate the NBTII cell dose response for Gleevec concentration we determined migration speed and persistence for NBTII cells treated with different concentrations of Gleevec. Cells were plated on substrates coated with 10 mg/ml collagen. The concentration of Gleevec employed to inhibit Abl family kinase activity was in the range of 0.25 mM to 50 mM. The average cell migration speed reached a maximum (,2 mm/min), when a 20 mM concentration of Gleevec was used. For lower Gleevec concentrations (0.25 mM, 1 mM), cells did not show a significant speed increase. The highest concentration of Gleevec (50 mM) actually caused cell migration speed to decrease ( Figure 3A). The ability of NBTII cells to migrate persistently in one direction was also highest after treatment with 20 mM of Gleevec ( Figure 3B). images of the f-actin (Rhodamine-phalloidin, Red) and microtubules (alpha-tubulin antibody, Green) in the control (C) and Gleevec-treated cells (D). (E and F) Box and whisker plots of the average cell migration speed (E) and directional persistence (F) for the control group (N = 60) and NBT-II cells treated with 20 mM Gleevec (N = 60). Standard deviations are indicated by the box sizes; maximum and minimum data values are indicated by the extent of the whiskers. The bar and the square inside the box are the median and mean value respectively. Gleevec-treated NBTII cells migrate significantly faster and are more persistent in their directionality (* p,0.001, by students t-test). Scale bars are 20 mm. doi:10.1371/journal.pone.0052233.g001 To investigate how the substrate adhesiveness influences NBTII cell migration, we tested different concentrations of collagen for substrate coating. With a higher collagen coating concentration, more collagen would absorb to the substrate providing more integrin binding sites, thereby presumably increasing cell-substratum adhesions. For these experiments we used control NBTII cells and cells treated with a 20 mM concentration of Gleevec. For control NBTII cells, when the collagen coating concentration was increased from 1 mg/ml to 100 mg/ml, both the migration speed ( Figure 3C) and persistence ( Figure 3D) increased. For the Gleevec-treated NBTII cells, the speed of migration was greatest on the substrates with medium and higher adhesivity (10 and 100 mg/ml collagen). The largest difference in speed and persistence between control and Gleevec-treated NBTII cells occurred at a 20 mM of Gleevec concentration on 10 mg/ml of collagen-coated substrates.
Gleevec-treated NBTII cells are more adherent to their substrate than control cells The highly spread lamellae of Gleevec-treated NBT II cells suggested that they had become more adhesive. Therefore, we investigated cell adhesion strength using a laminar flow system reported previously [28,29]. Basically, by varying the flow rate, the system generates various shear stresses on cells attached in the flow channel. When applied shear stress exceeds the global cell adhesion strength, cells will detach from the substrate ( Figure 4A). Images showing the cells attached before and after flow application were recorded and cell numbers were counted (Materials and Methods, Figure S2). For our experiments, we tested 100 dynes/cm 2 , 200 dynes/cm 2 and 253 dynes/cm 2 values of shear stress ( Figure S2G). We found that a shear stress of 200 dynes/cm 2 is the most appropriate for estimation of the relative NBTII cell adhesion strength. A shear stress of the 100 dynes/cm 2 was too weak to affect cell attachment and a shear stress of 253 dynes/cm 2 quickly removed most of the attached cells. We applied 200 dynes/cm 2 shear stress for 1 min to the control and Gleevec-treated NBTII cells which were plated four hours before experiment on 10 mg/ml collagen. The fraction of the remaining adherent cells was significantly larger for the cells treated with Gleevec ( Figure 4A). This result indicates that the global adhesion strength between cells and collagen substrates was increased after inhibition of Abl family kinases. By plotting the fraction of adherent cells remaining after application of shear stress vs. the applied shear stress, we could estimate that critical shear stress at which 50% of the cells detached increased by about 10% (from 214 to 236 dynes/cm 2 ) when cells were treated with Gleevec ( Figure S2G).
We further transfected NBTII cells with GFP-Paxillin to indicate adhesions and employed total internal reflection fluorescence microscopy (TIRFM) to monitor GFP-Paxillin localization on the ventral surface of the cell. Adhesions were automatically tracked and measured by an algorithm developed in a previous report (Materials and Methods) [30]. Compared to control NBTII cells ( Figure 4B), Gleevec-treated cells ( Figure 4C) had an increased number of adhesions at their leading edge and wings; in addition, the total adhesion number increased by ,25% ( Figure 4D) and the total adhesion area increased by ,70% ( Figure 4E). To assess the extent of detachment by bond breakage versus cohesive failure due to membrane rupture, we examined cells after application of sheer stress using confocal microscopy as shown in Figure S2H-J. This figure shows that NBT-II cell detachment occurred predominantly at the level of integrin and other adhesion bonds to the matrix coated substratum as opposed to membrane rupture around the adhesion sites. We reached this conclusion because the number of observations of membrane fragments or remnant focal adhesions ( Figure S2J) were few undetectable indicating that most cells detached by breaking substrate adhesion bonds.
Punctate adhesions are present at the leading edge of Gleevec-treated D-shape NBT-II cells TIRFM and Interference Reflection Microscopy were combined to capture time-lapse images of adhesions in migrating NBTII cells. The darker regions in interference reflection image are usually considered regions which are closer to the substrate [31]. In Figure S3A and S3B, the dark regions in interference reflection images are generally co-localized with the GFP-Paxillin regions in the TIRF image, indicating that those dark regions and dots are actually cell adhesions. The images of the leading edge of control and Gleevec-treated NBTII cells are shown in Figures 5A to 5D and 5E to 5H, respectively. In interference reflection images, the small punctate adhesions were only observed in the leading edge of Gleevec-treated cells ( Figure 5E and 5F) (Movie S6), but not in control cells ( Figure 5A and 5B) (Movie S5). As shown in the TIRF images, compared with control cells ( Figure 5C and 5D) (Movie S7), D-shaped NBTII cells had a larger amount of dotted GFP-Paxillin at their leading edge ( Figure 5G and 5H) (Movie S8). We measured the intensity of GFP-Paxillin along the sample lines indicated in Figure 5C and 5G. For each cell, lines crossing the leading edge were summed together and normalized (Materials and Methods). We found that D-shaped NBTII cells had significantly increased intensity of GFP-paxillin fluorescence signal in the vicinity of the leading edge ( Figure 5I). Adhesion turnover in the leading edge of a Gleevectreated cell was imaged and shown in Figure S3C or Movie S8. Figures S3D-G are TIRF images of EGFP-Paxillin in Gleevectreated cells, showing a rim of punctate adhesions at the leading margin as a common feature. The size distribution of punctate adhesions in Gleevec-treated NBTII cells had a peak at ,350 nm in diameter, and the average area of punctate was 0.1 mm 2 . The dimensions of many punctate adhesions are close to the diffraction limit of the microscope, and some punctates may be even smaller in dimension. Observation of the TIRF movies suggested that

NBTII cell migration behavior depends on substrate adhesiveness and Gleevec concentration. A) and B) Graphs depicting the average cell migration speed (A) and persistence (B) of NBT-II cells from the control group (no inhibitor) and NBT-II cells treated with different
concentrations of Abl family kinase inhibitor (Gleevec). Cells were cultured on 10 mg/ml collagen coated substrates. Gleevec concentration has significant effect to both cell migration speed and persistence (ANOVA, p,0.05). Cells treated with 5 mM or 20 mM Abl kinase inhibitor migrated significantly than control cells, while 20 mM and 50 mM group also migrated more persistently than control cells. The significance between control and each Gleevec treated groups was tested by one-way ANOVA followed by Bonferroni's post hoc test, (*) p,0.05. C) and D) Graphs depicting the average cell migration speed (C) and persistence (D) of NBT-II cells on substrates coated with 1 mg/ml, 10 mg/ml, and 100 mg/ml collagen. The average migration speed and persistence of control and Gleevec cells are presented with gray and black bars, respectively. Significance of differences in migration speed and persistence between control and Gleevec treated cells were marked in the figure (* p,0.001, by students t-test). For all panels (A-D), results are calculated from more than 30 cells in each group. The error bars indicate standard deviations. doi:10.1371/journal.pone.0052233.g003 these punctate adhesions near the leading edge turned over very rapidly; indeed, intensity analysis [32] indicated an average lifetime of ,70 s; by contrast, adhesions at the cell wings are much more stable with lifetimes in excess of 5 minutes ( Figure  S3H-I).
b1 integrin-containing cell adhesions are important for maintaining D-shape morphology and migration status Next, we asked whether integrins were important for the Gleevec induced-phenotype. To competitively disrupt Integrincollagen binding we employed two different agents: an RGDcontaining peptide (Gly-Arg-Gly-Asp-Thr-Pro) [33] and direct b1 integrin blocking antibodies, because b1 integrin is known to bind type I collagen [34,35,36]. We found that either 1 mg/ml of b1 integrin blocking antibody or 100 mg/ml RGD-containing peptide (G5646, SIGMA) quickly decreased cell migration speed ( Figure 6A) and reverted the Gleevec-treated cell morphology approximately back to control NBTII cells ( Figure 6B-D).

Gleevec induces changes in actin cytoskeleton, p-MLC localization and traction
The filamentous actin and active myosin were detected by fluorescent phalloidin and phosphorylated myosin light chain antibodies, respectively. The distribution of active myosin in control and Gleevec-treated NBTII cells differs. In the case of the control cells, active myosin is localized around the cell nucleus so that some appears near the central part of the trailing edge, with little near the leading edge ( Figure 7A-C). By contrast, the distribution of active myosin II in Gleevec-treated cells is distributed further away from the nucleus and the rear margin of the cell, mainly at the backside of cell wings, much like keratocytes ( Figure 7D-F). Since the active myosin status is associated with cell contractility, we compared the traction distribution between the two groups utilizing an elastic substrate methodology [37,38,39,40,41,42]. Figure 7G-H show the colorcoded magnitudes of the bead displacements mapping for control and Gleevec-treated cells, respectively. The white line drawn indicates the outline of the cell. The insets are the phase image of a control (7G) or a Gleevec-treated cell (7H). Figure 7I-J shows the calculated constrained traction map for control and Gleevectreated cells with the insets showing magnified traction maps at the cell wing regions. Tractions are much higher in the wings of the Gleevec-treated cells than control cells ( Figure 7J) where immunofluorescence labeling by anti-p-MLC antibody indicates a concentrated region of active myosin proximate to the high traction regions ( Figure 7F). The Gleevec-treated cells generated a remarkable almost four-fold greater total traction force than control NBTII cells ( Figure 7K).

Effects of RhoA family GTPases on D-shaped NBTII cell migration
The rapid response of NBTII cells to Gleevec suggests that the inhibition of the Abl-family kinases is altering active signaling pathways [8,43] as opposed to affecting transcriptional regulation. Because of the importance of RhoA GTPase in cell shape and migration [44,45,46], we measured its activity before and after Gleevec treatment. Using a RhoA activity pull down assay, we found that RhoA activity significantly increased when Abl-family kinases were inhibited ( Figure 8A). Cells treated with Gleevec for one hour have nearly doubled RhoA activity when compared to control cells ( Figure 8B).
Because of the significant increase in active RhoA in Gleevectreated cells, we asked whether RhoA and its downstream effector ROCK were important in the Gleevec-induced phenotype by introducing the RhoA inhibitor (C3) and the Rho kinase inhibitor (Y27632) to NBTII cells previously treated with Gleevec ( Figure 9A-C, respectively). We found that adding either C3 ( Figure 9B) or Y27632 ( Figure 9C) to Gleevec-treated cells resulted in a significantly increased number of retraction fibers and more rounded nuclei compared to cells treated with Gleevec only ( Figure 9A). In addition, both of these reagents reduced migration speed ( Figure 9D) and persistence ( Figure 9E). The nuclear aspect ratio and total retraction fiber length were calculated and compared to cells treated with Gleevec only or control cells ( Figure 9F-G). C3 or Y27632 treated cells in the presence of Gleevec exhibit similar nuclear aspect ratios and retraction fiber parameters as the control group, suggesting that inhibition of RhoA/ROCK activity in Gleevec treated cells strongly rescues control cell phenotype. Compared with the significant change in nuclear aspect ratio and retraction fibers, the whole cell aspect and area ratios for the Gleevec + C3 treatment group or for the Gleevec + Y27632 group are partially rescued. (Figure S4). We also observed that while Gleevec + C3 or Gleevec + Y27632 treated cells produce extended lamellae these lamellae often fragmented during migration. Because C3 and Y27632 in effect rescue the control phenotype, these results indicate that RhoA and its downstream effector ROCK are required for the Gleevec-induced NBTII cell phenotype.

Discussion
Here, we show that inhibition of Abl family kinase activity with Gleevec produced a rapid and remarkable change in cell morphology and migration in which cells spread out a thin, extended lamella and migrated faster and with more persistence with some similarities to fish and amphibian keratocyte migration [25,27]. In addition, this rapidly spreading, very thin lamella is similar to the rapid and extensive, ''pancake'' spreading of fibroblasts derived from Abl null mice [15]. Associated with the Gleevec phenotype was an increase in RhoA activity, increased global cell adhesion strength, a pronounced change in adhesion patterns and an increase in total traction applied to the substrate.

Regulatory mechanisms involved in the Gleevec-induced change in morphology and migration
Abl family kinases inhibit cell adhesion formation. Abl family kinases have been reported to be located at cell adhesions [7,47]. They are correctly positioned to regulate the reorganization of the cytoskeleton at sites of membrane protrusion and at focal adhesions where integrins are engaged. In 10T1/2 fibroblasts, during the initial 20-30 minutes of fibronectin stimulation, when c-Abl activity is the highest, the nuclear pool of c-Abl re-localizes transiently to focal adhesions [7,47]. This transient re-localization also occurs in NIH3T3 cells, where a fraction of the cellular Abl associates with the focal adhesion proteins, paxillin and Grb2 [48,49]. Abl family kinases have also been reported to reduce initial cell attachment to the substrate. On fibronectin, fibroblasts derived from Abl-null mouse embryos spread faster than their wild-type counterparts, while restoration of Abl expression in the Abl-null fibroblasts reduced the rate of spreading [15]. Kain and Klemke provided evidence that Abl family kinases negatively regulate cell migration by uncoupling CAS-Crk complexes [13]. Li and Pendergast recently reported that Arg could disrupt CrkII-C3G complex formation to reduce b1-integrin related adhesion formation [17]. These reports indicate that Abl family kinases negatively regulate cell adhesion, thus supporting our observations that Abl family kinase inhibition results in a more adhesive and motile phenotype. It is important to note that Gleevec has been reported to have inhibitory effects on other signaling pathways involving PDGF-R and c-kit that also impact the cytoskeleton [51] and therefore, potentially, cell migration. Cells with inhibited PDGF-R or c-Kit pathways exhibit reductions in migration or membrane protrusions [46,52,53,54] opposite to the effects reported here; this suggests that Gleevec inhibition of the c-kit and PDGF-R pathways is probably not the major factor for the profound NBT-II cell morphology transformation. Nevertheless, while Gleevec effects on Abl family kinase activity and cell adhesive behavior as well as on RhoA activity have been established [17,50], it is well to keep in mind potential 'off-target' effects on other regulatory pathways.

Changes in the adhesive behavior of Gleevec-treated NBTII cells
The D-shaped NBTII cells have a band of punctate dot-like adhesions in the vicinity of their leading edge that appear different from known adhesions in mesenchymal-type migrating cells. The area of these adhesions is quite small (,0.10 mm 2 ) compared to normal focal adhesions (,1 mm 2 ), and their turnover as estimated from observation of TIRFM movies is ,1 min, compared to .5 min for focal adhesions. These punctate adhesions are similar in size to nascent focal adhesions [55,56,57]; but they rarely Figure 6. Blocking of integrin related adhesion dramatically inhibits the migration speed of Gleevec-treated NBT-II cells. A) Migration speed of Gleevec-treated cells incubated with either beta1-integrin blocking antibody (1 mg/ml) or an RGD-containing peptide (100 mg/ml). Migration speed was measured 30 minutes after addition of beta1-Integrin blocking antibody or RGD containing peptide (n.10 for each group). Panels B-D) are DIC images of an NBT-II cell treated with Gleevec (20 mM) migrating on a 10 mg/ml collagen-coated substrate, and then treated with 1 mg/ml RGD containing peptide. Images before addition of RGD containing, 2 minutes after and 5 minutes after addition of the RGD peptide are shown. Scale bars are 20 mm. Error bars indicate standard deviations. The statistical significance of difference between control cells with and without RGD, and difference between Gleevec treated cells with and without integrin blocking antibody or RGD is indicated by an (*. p,0.05) as evaluated by one-way ANOVA followed by Bonferroni's post hoc test. doi:10.1371/journal.pone.0052233.g006 (,1%) matured to larger adhesions as many nascent adhesions do [57].
Often, an abundance of retraction fibers at the trailing edge of a cell is taken as evidence for strong adhesion in this region. However, at the rear of Gleevec-treated cells, in spite of greater global adhesion strength, there are fewer retraction fibers than in control cells. What might be the reason for this observation? A potential explanation is found in the fact that the trailing edge tractions of Gleevec-treated cells were significantly stronger than in control cells. These tractions may effectively break all adhesions in the rear of the cell, even those in that normally result in retraction fiber formation.
Our results taken as a whole indicate Abl family kinases play an important role in the regulation of cell adhesion and migration in that their inhibition produces a profound change in adhesions, morphology and cell migration. A fully integrated, quantitative view of inhibition of how these ubiquitous kinases produce these changes remains a challenge for the future. For immuno-staining, NBTII cells were fixed by using paraformaldehyde solution [4% (w/v) in PBS, pH 7.4] for 20 minutes at 25uC. Cells were then permeabilized with PBS containing 0.05% Triton-X-100 for 5 minutes at 25uC. Fluorescence labeling was carried out by treating with primary antibodies, washing with medium and then treating with fluorescent secondary antibodies followed by washing. Imaging was done as described below.

Cell culture and transfection
NBT-II cells were acquired from the ATCC (Manassas, VA) and maintained in DMEM/F-12 medium (Gibco, Grand Island, NY) containing 10% FBS, 100 units/ml penicillin and 100 mg/ml streptomycin. The EGFP-Paxillin-b and EGFP-Vinculin construct were generated by subcloning DNA fragments expressing wildtype paxillin and wild type vinculin into a pEGFP-C1 vector (Clontech, Mountain View, CA). NBT-II cells were transfected using the Lipofectamine Plus transfection reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Cells stably expressing either EGFP-Paxillin-b or EGFP-Vinculin were obtained by sorting for EGFP positive cells after G418 selection in the UNC Flow Cytometry Facility.

Cell migration, surface coating and drug treatment
For the experiments imaging the migration of NBT-II cells, glass bottom Petri dishes (35 mm) (MatTek) were coated by incubating with 10 mg/ml type I collagen (BD Biosciences, Bedford, MA) overnight at 4uC. NBT-II cells were treated with trypsin and resuspended in DMEM/F12 medium (GIBCO) containing 10% fetal bovine serum, plated at low density on the dishes, and cultured for 4-12 h in a CO 2 incubator. Cells were incubated with either no inhibitor, or various concentrations of the 20 mM Abl family inhibitor GleevecH (Novartis, Basel, Switzerland) by adding the inhibitor to culture media which was mixed by gently pipetting up and down. The cells were incubated for 30 minutes prior to imaging and imaging was performed in the continued presence of the inhibitor. 5 mM ROCK inhibitor(Y-27632, Sigma, MO) or 2 mg/ml of Rho inhibitor, C3 transferase (Cytoskeleton Inc., Denver, CO), was used to inhibit ROCK or Rho activity in Gleevec pre-treated NBTII cells for 30 minutes. 100 mg/ml RGD-containing peptide (Gly-Arg-Gly-Asp-Thr-Pro) (G5646, SIGMA), or 1 mg/ml anti-Mouse/Rat CD29 (Biosciences Pharmingen, CA) was used to block integrin related cell adhesion. Cell migration status was studied after one hour of incubation with these inhibitors.

Assay for active RhoA GTPases
Active RhoA pulldown experiments were done as described previously [58]. For active RhoA pulldown cells were lysed in 300 ml 50 mM Tris, pH 7.4, 10 mM MgCl2, 500 mM NaCl, 1% Triton X-100, 0.1% SDS, 0.5% deozycholate, 1 mM PMSF, and 10 mg/ml each of aprotinin and leupeptin. Lysates were cleared by centrifugation at 14,000 g for 5 min. Supernatants were rotated cells. The bar graph indicates NBTII cells treated with Gleevec generate considerably more total traction force than control NBTII cells. Error bars are standard deviation. Bar = 20 mm. 8 cells were examined for each case. Control and Gleevec-treated cells are significantly different in total cell traction force (* p,0.01, by student's t-test). doi:10.1371/journal.pone.0052233.g007 for 20 minutes with 30-50 mg GST-RBD conjugated to glutathione-Sepharose beads (GE Healthcare). Beads were washed with in ml 50 mM Tris, pH 7.4, 10 mM MgCl2, 150 mM NaCl, 1% Triton X-100, 1 mM PMSF, and 10 mg/ml each of aprotinin and leupeptin. Active RhoA and total RhoA levels were analyzed by SDS-PAGE. Gel intensity results were quantified by analyzing inverted images using ImageJ. The signal from protein bands was measured after background subtraction and the intensity of each image was then normalized according to the average signal of loading control band.

Cell imaging
Differential Interference Contrast, Total Internal Reflection Fluorescence (TIRF) and epi-fluoresence imaging were carried out on a dual-channel Olympus IX81 inverted microscope equipped with a 606, oil immersion, 1.45 NA objective. Interference Reflection Microscopy imaging was performed using a 1006, oil immersion, 1.65 NA objective. Objective (660) style TIRFM was performed on the Olympus IX81 with the Olympus TIR option. Images were captured using an air-cooled SensiCam QE CCD camera (Cooke Corp., Romulus, MI) driven by Metamorph (Molecular Devices/Meta Imaging, Downingtown, PA). Confocal imaging was performed with an inverted Olympus FV1000 equipped with a live cell chamber and a 6061.42 N.A. oil immersion objective. Migration of single cells was tracked for durations between 5 minutes to 2 hours; time-lapse images were taken with the intervals of 1 second (for morphology changes), Figure 9. RhoA/ROCK activity is important for the Gleevec phenotype. A), B) and C) are DIC images of NBT-II cell migration status in the presence of 20 uM Gleevec only, both 20 uM Gleevec and ROCK inhibitor(5 uM Y-27632), or both 20 uM Gleevec and RhoA inhibitor (C3, 1 ug/ml), respectively. These panels show that RhoA inhibition after Gleevec treatment increases the number of retraction fibers and produces more rounded nuclei. Scale bars are 20 mm. Panels D) to G) are cell migration speed, cell migration persistence, nuclear aspect ratio and cell retraction fiber ratio, respectively. In each figure, the four bars represent the control group, the 20 uM Gleevec-treated group, the 5 mM Y-27632 +20 mM Gleevec-treated group, and the 1 mg/ml C3 +20 mM Gleevec-treated group, respectively. Error bars indicate standard deviations. At least 15 cells were measured for each group. The significance of the difference between control and other treated groups was evaluated by one-way ANOVA followed by Bonferroni's post hoc test, and marked by (*), p,0.05;. doi:10.1371/journal.pone.0052233.g009 5 seconds (for rapid adhesion turnover) or 60 seconds (for cell migration and long lifetime adhesion turnover), as indicated.

Measurement of cell adhesion strength
A flow system designed to produce laminar shear stress on attached NBTII cells consisted of a flow cell, a dual syringe pump (Harvard Apparatus, Holliston, MA), 5% CO2 percolated media reservoir, and a pulse dampener (Cole Parmer Instrument Company, Chicago, IL) as previously described in detail [28,29]. Briefly, a two-piece, top and bottom plate, anodized aluminum flow cell (1267.5 cm) was constructed with plastic inlet and outlet tubes. The Nunc SlideFlask (Thermo) on which NBT-II cells were plated was placed in the middle of the flow cell bottom plate with a 0.27-mm thick silicone gasket placed underneath it. This brought the coverslip to the same height as the top of the bottom plate. Shear stress was calculated using the following equation: where m is the viscosity of 0.0086 g.cm/s, Q is the flow rate in mL/s, w is the width of the flow channel (1.7 cm), and h is the height of the flow channel (0.027 cm). The applied shear stress ranged from 100 to 253 dynes/cm 2 . Cells were cultured for 4 hours on 10 mg/ml pre-coated collagen Nunc SlideFlask (Thermo) substrates; and labeled with 1:1000 Cell tracker orange (Invitrogen, Carlsbad, CA) for 10 minutes 30 minutes before flow experiments. All flow experiments were performed at 37uC for 1 minute with PBS containing calcium and magnesium. In each experiment, multiple (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) images were taken showing cells in different regions of the flow chamber, before and after shear stress was applied. The cell attached to substrate before and after shear stress were counted, expressed as the percent of adherent cells remaining, and averaged over multiple experiments. Data were reported as mean 6 SEM.

Traction force microscopy
Preparation of polyacrylamide substrates and experimental imaging has been described previously [42]. Briefly, the fabrication of polyacrylamide substrate involves following three steps: 1) Silanization the glass substrate with 0.5% silane for 20 minutes, 2) Use 0.5% glutaraldehyde treat previous substrate for 40 minutes, 3) polymerization with 6% polyacrylamide and 1% bis-acrylamide in 10 mM HEPES buffer with rhodamine-dextran beads and NHS-acrylate. Polymerization is initiated with the addition of 0.05 g/ml APS. The elastic modulus of our substrate was measured to be approximately 48 kPa [42]. Elastic substrates after polymerization were stored at 4uC in PBS.
Tractions were calculated using the method of Butler et al (2002) in which particle imaging velocimetry is employed to measure the displacement of small windows that contain a number of beads.

Data quantification and calculation
The movement of individual cells was analyzed with Metamorph software and ImageJ. Statistical analysis was performed using unpaired student's t-tests. Statistical significance was indicated by * in the figure, and the p value was defined in each figure legend. One-way ANOVA with the Bonferroni post hoc test was used (as indicated in the figures) to compare the differences in NBT-II cell migration speed, persistence or other cellular morphology parameters, with values of P,0.05 sufficient to reject the null hypothesis for all analyses.
Speed and persistence of migration. To get the position of the cell, we manually tracked the geometric center of the cell nucleus. Although cell boundary morphology changed considerably during migration because of protrusion, the nucleus was usually centrosymmetric and could be employed as a marker for cell location. The cell position and cell migration speed was tracked and calculated using the ImageJ plugins: ''Manual Tracking Plug-in'' (http://rsbweb.nih.gov/ij/plugins/track/ track.html). For cell migration persistence, we employed a conventional definition: the ratio of the net distanced traveled to the total distance traveled. The net distance traveled is the magnitude of the vector between the starting point and end point of the cell trajectory over an hour and the total distance traveled taken as the sum of net distances traveled over twelve 5 minutes intervals in that hour in order to capture the changes in direction that occur. Thus, the highest persistence would have a value of one, representing unidirectional migration.
Aspect ratio measurement and retraction fiber counting. The outline of the cell was manually obtained using Photoshop to trace the edge of each cell, based on DIC images. The dimension of the cell along the cell migration direction or perpendicular to cell migration direction was then measured from the cell outline with a Matlab program (The direction of cell migration was determined using ImageJ and manual tracking plugins as described above.). The aspect ratio was then calculated using cell dimension perpendicular to its migration direction divided by cell dimension along its migration direction. Retraction fibers could be counted manually because they were generally well-defined in the images.

Segmentation of adhesions in TIRF
images. Segmentation of adhesions in TIRF images: Timelapse TIRF images of EGFP-Paxillin were analyzed using previously published methods [59] [30]. First, the distribution of high-pass filtered pixel intensities was determined for each cell. Adhesions were segmented by selecting pixels at least two standard deviations away from the mean. Next, connected components labeling was used to identify the adhesions and any object less than two pixels in size was discarded. The average number of adhesions and the total area of the adhesions per image were calculated for each cell. T-tests were used to test for statistical differences between these measurements. Lifetime of adhesions. ImageJ was used to measure changes in fluorescent intensity of individual adhesions over time in cells expressing fluorescent-tagged adhesion proteins. A perimeter was drawn around the punctate or wing adhesions at the point where intensity was a maximum; average pixel intensity was calculated from the pixel intensities within that perimeter as function of time. Background and photobleaching corrections were applied to obtain true intensities of the adhesions. Assembly and disassembly rates were plotted and calculated using Microsoft Excel (Microsoft Corporation) or Origin 6.1(OriginLab) using a previously published method (Huang et al., 2009) (Choi et al., 2008 [32]. Quantification of collective adhesion profile. Multiple lines (n = 4 for each cell, 12 cells in each group) were drawn along the cell migration direction from outside to the interior of the cell (line length = 200 pixels) with the center of the line is positioned manually at leading edge. The fluorescent signal along each line was measured using PlotProfile function in ImageJ. The EGFP-Paxillin fluorescence intensity in different cells was not uniform. In order to better compare the profile of collective adhesion intensities between control and Gleevec-treated cells, background (S bgd ) was subtracted from the raw fluorescence intensity profile (for each pixel S(n))and then these profiles were normalized. We used following formulae to get the normalized relative signal intensity (S rsi-Line (n)) for the n th pixel (n = 1 to 200) along the line: where S avg represents average pixel intensity averaged along the entire line. Movie S1 Control NBT-II cell migration. Time-lapse DIC microscopy recording of a control NBT-II cell migrating on collagen substrate (10 mg/ml). Movie demonstrates that NBT-II cells on collagen have medium-sized lamellae and lamellipodia, the multiple dynamic filopodia at the leading edge, and multiple retraction fibers at the rear end of the cell. Movie was recorded with 606objective at 10 s intervals and plays back at 6 frames per second (fps).

(AVI)
Movie S2 NBT-II cells in response to Gleevec treatment. Time-lapse DIC microscopy recording of a NBT-II cell migrating on collagen substrate (10 mg/ml) and its response to Gleevec treatment (20 mM). Gleevec (20 mM) was added at time 00:00. Movie shows a quick transformation of cell morphology, including fast lamellipodia formation and changes of cell nucleus shape. Movie was recorded with 606 objective at 10 s intervals and plays back at 6 fps.
Movie S3 Gleevec-treated NBT-II cell migration. Timelapse DIC microscopy recording of two Gleevec-treated (20 mM) NBT-II cells migrating on collagen substrate (10 mg/ml). Note that after Gleevec treatment NBT-II cells had an intact lamella facing the migration direction and markedly reduced numbers of both filopodia and retraction fibers. Movie was recorded with 606 objective at 10 s intervals and plays back at 6 fps. (AVI) Movie S4 Special migration status and persistence in multiple Gleevec-treated NBT-II cell migration. Timelapse DIC microscopy recording of multiple Gleevec-treated (20 mM) NBT-II cells migrating on collagen substrate (10 mg/ml). NBT-II Cells were imaged immediate after Gleevec treatment. Note that NBT-II cells after Gleevec treatment have a big lamella facing the migration direction. The migration speed and persistence were significantly increased after Gleevec treatment. Movie was recorded with 206 objective at 5 min intervals and plays back at 3 fps.