B-Raf Regulation of Integrin α4β1-mediated Resistance to Shear Stress through Changes in Cell Spreading and Cytoskeletal Association in T Cells*

Background: Contribution of MAPK members to integrin adhesion is not established. Results: Inhibition of B-Raf function or expression selectively regulates integrin α4β1 in T cells. Conclusion: B-Raf is a signaling component for integrin α4β1 cytoskeletal association, cell spreading, and adhesion. Significance: This novel association of B-Raf and integrin α4β1 suggests new therapeutic targets in T cells and indicates potential off-target effects of sorafenib. The regulation of integrin-mediated adhesion is of vital importance to adaptive and innate immunity. Integrins are versatile proteins and mediate T cell migration and trafficking by binding to extracellular matrix or other cells as well as initiating intracellular signaling cascades promoting survival or activation. The MAPK pathway is known to be downstream from integrins and to regulate survival, differentiation, and motility. However, secondary roles for canonical MAPK pathway members are being discovered. We show that chemical inhibition of RAF by sorafenib or shRNA-mediated knockdown of B-Raf reduces T cell resistance to shear stress to α4β1 integrin ligands vascular cell adhesion molecule 1 (VCAM-1) and fibronectin, whereas inhibition of MEK/ERK by U0126 had no effect. Microscopy showed that RAF inhibition leads to significant inhibition of T cell spreading on VCAM-1. The association of α4β1 integrin with the actin cytoskeleton was shown to be dependent on B-Raf activity or expression, whereas α4β1 integrin affinity for soluble VCAM-1 was not. These effects were shown to be specific for α4β1 integrin and not other integrins, such as α5β1 or LFA-1, or a variety of membrane proteins. We demonstrate a novel role for B-Raf in the selective regulation of α4β1 integrin-mediated adhesion.

of the role of MAPK in the regulation of lymphocyte adhesion or the direct regulation of integrin activity is still limited.
We set out to study secondary roles of the MAPK pathway members as regulators of integrin function in T cells. Inhibition of RAF but not MEK/ERK reduces the adhesion of Jurkat T cells to VCAM-1 under shear stress, demonstrating the independence of this function from downstream signaling of the MAPK pathway. shRNA-mediated knockdown of B-Raf reproduced the effect of chemical RAF inhibition, confirming a role for B-Raf in lymphocyte adhesion.
Generation of Inducible B-Raf Knockdown Cells-Knockdown cells were generated using shRNA-pTRIPZ clones (Open Biosystems, Huntsville, AL). Lentivirus was generated by Lipofectamine 2000 transfection of the packaging cell line, 293T-METR, with packaging plasmids containing p⌬R8.91, CMV-pVSVG, and either scrambled non-silencing controls or a BRAF targeting sequence. Viral supernatants were collected at 48 h and concentrated by ultracentrifugation. Jurkat cells (1 ϫ 10 6 ) were transduced and then selected in medium containing puromycin (1 g/ml). After 1 week, cells were divided into a no treatment group and a doxycycline (1 g/ml)-treated group. After 72 h, doxycycline-induced fluorescent red (TurboRFP) cells were purified with a FACS Aria IIu high-speed cell sorter (BD Biosciences) and cultured in complete medium with doxycycline.
Parallel Plate Flow Detachment Assay-The detachment assay was performed as described (26). In brief, human FN (5 g/ml), VCAM-1 (10 g/ml), or mAbs (1 g/ml) were immobilized to plastic slides, washed with PBS, blocked with 2% BSA in PBS, and assembled to a parallel plate flow chamber. Cells (4 ϫ 10 6 ) in running buffer (10 mM Tris, 103 mM NaCl, 24 mM NaHCO 3 , 5.5 mM glucose, 5.4 mM KCl, and 0.2% BSA, pH 7.4) were injected into the flow chamber and allowed to settle on the slide for 10 min. A computer-controlled syringe pump (Harvard Apparatus) was used to apply an increasing linear gradient of fluid flow to the cells for 300 s and recorded by digital microscopy. Shear stress calculations were determined every 50 s, and the shear stress in dynes/cm 2 was defined as (6Q)/(wh 2 ), where is the viscosity of the medium (0.007), Q is the flow rate in cm 3 /s, w is the width of the chamber (0.3175 cm), and h is the height of the chamber (0.0254 cm).
Bright Field Microscopy-Human VCAM-1 was immobilized to 6-well non-tissue culture-treated plates (Falcon), washed with PBS, and blocked with 2% BSA in PBS. Cells (1 ϫ 10 6 ) in complete medium were added, incubated at 37°C for 10 min, and then fixed with 2% paraformaldehyde in PBS for 20 min at room temperature. Images were captured at ϫ20 magnification using a Nikon Diaphot-TMD microscope, equipped with a VI-470 CCD video camera (Optronics Engineering). Images were analyzed using Slidebook software (version 5.0) to distinguish spread cells from non-spread cells by creating a mask of spread cells and counting all cells that were larger or smaller than the threshold.
Super-resolution Immunofluorescence-Human VCAM-1 (10 g/ml) was immobilized to glass coverslips, washed with PBS, and blocked with 2% BSA in PBS. Cells (5 ϫ 10 5 ) in complete medium were added and incubated at 37°C for 10 min and then fixed with 2% paraformaldehyde in PBS for 20 min at room temperature. Cells were permeabilized by adding saponin to a concentration of 0.1% for 30 min at room temperature. Cells were washed three times with PBS, 2% BSA, 0.1% saponin, stained for total B-Raf (AlexaFluor 647) and ␤1 integrin (Alexa-Fluor 488), and mounted to slides using Prolong Gold anti-fade reagent (Invitrogen). Images were acquired at room temperature using the OMX Blaze V4 structured illumination microscope (Applied Precision) with a ϫ100 numerical aperture 1.40 objective lens, two EM-CCD Photometrics Evolve 512 cameras, and DeltaVision OMX acquisition software. The images were reconstructed and rotated in three dimensions by 90°, and the height of cells was measured using the softWoRx software (version 6.0 beta 19). The image stacks were then transferred to either Slidebook software (version 5.0) to measure the area of contact of the cell with the glass coverslip or Imaris Bitplane software (version 7.6.1) to measure the colocalization of ␤1 integrin and B-Raf. The colocalization was quantified from the reconstructed three-dimensional image using the spot detection function for absolute fluorescence of both ␤1 integrin and B-Raf channels. Spots were generated with a 200-nm maximum xy diameter and a 500-nm maximum z diameter, identifying between 2000 and 15,000 spots for each channel per reconstructed image. Then the spots-to-spots colocalization function was used to identify all spots within 300 nm of spots from the other channel.
Soluble VCAM-1 Binding Assay-The soluble VCAM-1 binding assay was modified from a previous procedure (27). In brief, cells (1 ϫ 10 6 ) in 100 l of serum-free medium were incubated with human VCAM-1-Fc (10 g/ml) at 37°C for 10 min. The cells were then diluted and fixed by adding 2 ml of RPMI 1640 with 2% paraformaldehyde for 20 min at room temperature. The cells were washed twice with 2% BSA in PBS and incubated with AlexaFluor 488-conjugated rabbit anti-mouse for 20 min at room temperature. The cells were then washed and analyzed by flow cytometry using a FACSCalibur flow cytometer (BD Biosciences).
Cytoskeletal Stabilization Assay-The quantification of integrin-cytoskeleton attachment was modified from a previous procedure (26 -28). Cells (2 ϫ 10 6 ) in 100 l of complete medium were incubated with mAb (1 g/ml) at 4°C for 30 min, and then either they were left untreated or AlexaFluor-conjugated rabbit anti-mouse was added at 4°C for 30 min. The cells were incubated at either 4 or 37°C for 10 min. The cells were then washed and resuspended in cytoskeletal stabilizing buffer (50 mM NaCl, 2 mM MgCl, 0.22 mM EGTA, 13 mM Tris, 1 mM PMSF, 10 mM iodacetamide, and 2% FBS, pH 8.0) with or without 0.1% Nonidet P-40. After 5 min at room temperature, 1 ml of cytoskeletal stabilizing buffer was added, and cells were immediately pelleted and fixed with 2% paraformaldehyde in PBS for 20 min at room temperature. The cells were then washed three times in PBS, and the amount of remaining bound mAb was determined by flow cytometry using a FACSCalibur (BD Biosciences).

RAF Inhibition, but Not ERK Inhibition, Leads to Decreased
Adhesion of T Cells to Fibronectin-The parallel plate flow assay was used to measure adhesion and investigate the role of ERK signaling in integrin-mediated adhesion to human FN under conditions of shear stress. Two inhibitors of the ERK pathway were used, U0126, a MEK inhibitor, and sorafenib, a RAF inhibitor. Jurkat cell adhesion to FN was unchanged by 1 h of 1 M U0126 treatment (Fig. 1A). The U0126-and vehicle (DMSO)treated cells show very similar rates of detachment, with ϳ49% of the initial cells remaining at 45 dynes/cm 2 (maximum shear) (Fig. 1B). The phosphorylation of ERK was inhibited in these cells (Fig. 1C). In contrast to MEK/ERK inhibition, adhesion to FN was significantly reduced by 50 nM sorafenib when compared with vehicle control (MeOH) (Fig. 1D). The cells pretreated with vehicle show 47.5% of cells remaining at maximum shear, whereas the cells pretreated with sorafenib show an increased rate of detachment with only 11.9% of cells remaining, a 75% inhibition (Fig. 1E). The phosphorylation of B-Raf was inhibited by 50 nM sorafenib (Fig. 1F). These results suggest that RAF, and not MEK/ERK, contributes to T cell resistance to shear stress after adhesion to FN.
RAF Inhibition Decreases ␣4␤1 Integrin-mediated Adhesion to VCAM-1-Both ␣5␤1 and ␣4␤1 integrin bind FN, so to address the specificity of the adhesion, we tested adhesion to VCAM-1, which is a ligand for ␣4␤1 and not ␣5␤1. Adhesion to VCAM-1 was unchanged by 1 h of 1 M U0126 treatment ( Fig. 2A). The U0126-and vehicle (DMSO)-treated cells show very similar rates of detachment to VCAM-1, with ϳ40% of cells remaining at maximum shear (Fig. 2B). The phosphorylation of ERK was inhibited in these cells (Fig. 2C). However, adhesion to VCAM-1 was reduced by 50 nM sorafenib (Fig. 2D). The cells pretreated with either vehicle or 10 nM sorafenib show very similar rates of detachment to VCAM-1, with ϳ65% of the initial cells remaining at maximum shear, whereas the cells pretreated with 50 nM sorafenib show an increased rate of detachment with only 29.1% of the initial cells remaining, a 57% inhibition (Fig. 1E). The phosphorylation of B-Raf was inhibited at the 50 nM concentration of sorafenib (Fig. 2F). Sorafenib is most specific for Raf-1 (6 nM IC 50 ) and B-Raf (22 nM IC 50 ) (29), and combined, these data suggest that B-Raf phosphorylation contributes to ␣4␤1 integrin-mediated adhesion to VCAM-1.  4B) and FN (Fig. 4C). The control cells show very similar rates of detachment to VCAM-1, with ϳ55% of the initial cells remaining at maximum shear, whereas the KDϩD cells show an increased rate of detachment with only 4.9% of the initial cells remaining (Fig.  4D). Similarly, with FN, the control cells show equal rates of detachment, with ϳ36% of the initial cells remaining at maximum shear, whereas the KDϩD cells show an increased rate of detachment, with only 7.4% of the initial cells remaining (Fig.  4E). These results demonstrate that knockdown of B-Raf leads to decreased adhesion to ␣4␤1 integrin ligands.

B-Raf Is Unessential for T Cell Proliferation or ␣4 and ␤1
Integrin Expression-A reduction in viability or ␣4␤1 integrin expression of B-Raf knockdown cells could account for decreased adhesion. To confirm that B-Raf knockdown did not produce a defect in cell viability, the proliferation of the transduced or control cells cultured with or without doxycycline was measured for 10 days and found to remain unchanged (Fig. 5A). To confirm that B-Raf knockdown does not lead to reduced surface expression of integrin subunits, ␣4 or ␤1 integrin subunits were measured by flow cytometry and were unchanged (Fig. 5B). Thus, the decreased adhesion observed under conditions of B-Raf knockdown is not due to reduced viability or ␣4␤1 integrin expression.
␣4␤1 Integrin Affinity Is Not Affected by B-Raf Knockdown or Sorafenib-Another mechanism of regulating the adhesion strength of integrins is the modulation of binding affinity. To test the role of B-Raf in the regulation of ␣4␤1 integrin affinity, the binding affinity for soluble VCAM-1 was measured and found to be unchanged by pretreatment with 50 nM sorafenib (Fig. 5C) or by B-Raf knockdown (Fig. 5D). Cells were incubated with Mn 2ϩ as a positive control for maximal integrin affinity, and the binding affinity for VCAM-1 of cells incubated with 1 mM Mn 2ϩ was increased over the cells that did not receive Mn 2ϩ but unchanged by B-Raf knockdown or sorafenib, indi-cating that the ability of the integrin to achieve a maximal affinity conformation is unaffected. These results indicate that B-Raf is unessential to ␣4␤1 integrin affinity for soluble VCAM-1, and the reduced adhesion is not due to affinity regulation of ␣4␤1.
Sorafenib or B-Raf Knockdown Inhibits Cell Spreading after Adhesion to VCAM-1-Bright field microscopy was used to investigate the effects of RAF inhibition on cell morphology after adhesion to VCAM-1. Images were captured and quanti-fied after 10 min of adhesion to VCAM-1 (Fig. 7A). An average of 84.4% of MeOH vehicle-treated cells were spread, compared with 9% of cells treated with 50 nM sorafenib (Fig. 7B). Similarly, an average of 77.3% of shRNA vector control cells (VCϩD) were spread, compared with 7.9% of B-Raf knockdown cells (KDϩD) (Fig. 7C). These results demonstrate that B-Raf activity or expression is required for efficient ␣4␤1 integrin-driven cell spreading.

Sorafenib Inhibits Cell Spreading and ␤1 Integrin Colocalization with B-Raf after Adhesion to VCAM-1-Super-resolution
microscopy was used to investigate RAF inhibition on cell morphology and localization of ␤1 integrin and B-Raf after 10-min adhesion to VCAM-1. Cell spreading after pretreatment with 50 nM sorafenib was measured using the height of cells (Fig. 8A) and the area of cellular contact with the glass coverslip (Fig. 8B) as indicators of spreading. The average height of cells pretreated with vehicle control (MeOH) was 5.5 m, whereas the average height of cells pretreated with sorafenib was 10.2 m (Fig. 8C). The average area of cellular contact with VCAM-1 of control cells was 512 m 2 , whereas the average with sorafenib was 110 m 2 , a 78.4% reduction (Fig. 8D). The colocalization of ␤1 integrin and B-Raf was quantified from reconstructed threedimensional images using the spots generated from absolute fluorescence (Fig. 8E). The control cells had 31.6% of ␤1 integrin colocalize with B-Raf, and the cells pretreated with sorafenib had only 13.3% of ␤1 integrin colocalize with B-Raf, a 58% reduction (Fig. 8F). A trend was observed in the decreased amount of B-Raf to colocalize with ␤1 integrin in sorafenibtreated cells, but it failed to reach significance (Fig. 8G). These results demonstrate that sorafenib inhibits cell spreading on VCAM-1 and reduces ␤1 integrin colocalization with B-Raf after adhesion to VCAM-1.
Sorafenib Prevents ␤1 Integrin Association with the Cytoskeleton-Integrin association with the actin cytoskeleton increases post-ligand binding and is necessary for focal adhesion formation and cellular resistance to shear stress. It has been shown that antibody cross-linking of ␣4 integrin on Jurkat cells leads to a significant increase in the amount of anti-␣4 mAb able to resist solubilization by a non-ionic detergent, and this is interpreted to result from an increase in ␣4 integrin association with the cytoskeleton (30). The association of ␤1 integrin with the cytoskeleton after activation by cross-linking of an anti-␤1 mAb was measured. Cells pretreated for 1 h with 50 nM sorafenib show significantly reduced ␤1 integrin association with the cytoskeleton after activation at 37°C (5.2%) compared with control (23.6%), a 78% reduction (Fig. 9A). These results show that RAF inhibition reduces ␤1 integrin association with the cytoskeleton and indicate a role for B-Raf phosphorylation in this pathway.
B-Raf Is Essential for ␤1 Integrin Association with the Cytoskeleton-The association of ␤1 integrin with the cytoskeleton was measured in B-Raf knockdown or shRNA vector control cells pretreated with 50 nM sorafenib or vehicle control. Consistent with results in Fig. 9A, shRNA vector control cells pretreated for 1 h with sorafenib show significantly reduced ␤1 integrin association with the cytoskeleton after activation at 37°C (4.2%) compared with vehicle control (20.5%), a 79.6% reduction (Fig. 9B). In comparison with vector control cells (20.5%), B-Raf knockdown cells show significantly reduced ␤1 integrin association with the cytoskeleton (6%) after activation at 37°C, a 71.8% reduction. These results indicate that B-Raf is essential for ␤1 integrin association with the cytoskeleton.

DISCUSSION
This work has demonstrated a novel role for B-Raf in the direct regulation of the ␣4␤1 integrin in lymphocytes. Both the chemical inhibition and knockdown of B-Raf lead to decreased resistance to shear stress of T cells after adhesion to ␣4␤1 ligands and decreased ␣4␤1 association with the cytoskeleton after mAb cross-linking. These effects were both specific for ␣4␤1 integrin and independent from affinity regulation or downstream MEK/ERK activity. We have also shown that sorafenib inhibits ␣4␤1 integrin-driven cell spreading and ␤1 integrin colocalization with B-Raf. The increased height and reduced area of cellular contact with VCAM-1 of cells treated with sorafenib suggest that these cells are experiencing greater shear stress distributed over a smaller area of anchorage, probably contributing to decreased adhesion in our laminar flow system. Cell adhesion has been shown to be dependent on the dynamics of the actin cytoskeleton and cell spreading, promoting adhesion strengthening and migration, in part by providing a more streamlined shape for the cell to reduce the shear stress imposed by laminar flow (6,31,32).
At present, it is not clear on the molecular level how B-Raf regulates ␣4␤1 integrin function. Chemokine binding to G protein-coupled receptors induces the activation of phospholipase C and calcium signaling, leading to the rapid up-regulation of ␣4␤1 and LFA-1 integrin affinity (33,34). Downstream of calcium signaling, Rap1 from the Ras family of small GTPases plays an important role in ␤2 integrin affinity regulation and adhesion, but this has not been established for ␣4␤1 integrin (4,35). It has been shown that Rap1 mediates phorbol-ester (phorbol 12-myristate 13-acetate)-stimulated adhesion to FN, but Rap1 does not mediate SDF-1␣-stimulated affinity up-regulation of ␣4␤1 integrin for VCAM-1 or adhesion to VCAM-1 after SDF-1␣ or phorbol 12-myristate 13-acetate stimulation (36,37). However, Rap1 specifically activates B-Raf, not Raf-1, and precedence for RAF family members in adhesion regulation was provided by studies demonstrating that H-Ras activation of Raf-1 suppressed integrin activation in CHO cells, but there were conflicting results concerning whether the suppression was independent of ERK (38 -42). Therefore, a secondary role for B-Raf in the direct regulation of integrin-mediated adhesion was feasible.
Identification of the specific B-Raf-containing complex and binding partners of this mechanism will be important to the understanding of lymphocyte adhesion. Our results indicate that B-Raf is specifically regulating the ␣4␤1 integrin and not LFA-1 or a variety of other membrane proteins. Integrin cytoplasmic tails binding to integrin-associated proteins mediate the events following receptor occupancy, such as adhesion strengthening and outside-in signaling. It has been shown that talin regulates cytoskeletal association, resistance to shear stress, and affinity of both ␣4␤1 and LFA-1 (43,44). However, it is established that different integrin heterodimers expressed by the same cells can utilize distinct signaling components and downstream effectors. For example, paxillin binding to ␣4 integrin cytoplasmic domains is known to regulate cytoskeletal association and resistance to shear stress but not affinity of ␣4␤1, whereas paxillin does not bind to ␣L of LFA-1 (5,30,45). In addition to possible Rap1 interaction, novel binding partners of B-Raf in T cells have been found (46). Of interest, B-Raf was shown to interact with both Dock2 (dedicator of cytokinesis protein 2), a guanine nucleotide exchange factor known to specifically activate Rac, and IQGAP1 (Ras GTPase-activatinglike protein 1), a scaffolding protein known to interact with actin, Rac1, calmodulin, and Src (46 -48). Alternatively, B-Raf regulates cytoskeletal dynamics in melanoma cells by mediating cross-talk with the Rho/ROCK pathway through Rnd3 and Integrin ␣4␤1 Regulation of T Cell Adhesion by B-Raf AUGUSST 15, 2014 • VOLUME 289 • NUMBER 33 in fibroblasts through the ROCKII/LIMK/cofilin pathway (19,49). Also, whereas an ERK-paxillin complex has since been identified in human adenocarcinoma cells, it was previously shown that Ras-induced serine phosphorylation of paxillin was mediated by induced expression of an activated B-Raf construct in a variety of cell types (17,50). However, we found that sorafenib or B-Raf knockdown did not affect the induced phosphorylation of paxillin (Tyr-118) after adhesion to VCAM-1 (data not shown), a residue essential for actin cytoskeleton-dependent cell spreading and motility in lymphocytes (51-53), whereas other studies using adherent cell lines have identified Raf-1 association with vimentin, myosin phosphatase, and the Rho-effector Rok-␣, raising the possibility of similar interactions for B-Raf (54 -56). Altogether, and given the inherent differences between how adherent cell types and lymphocytes regulate adhesion, more studies are required to explore the molecular details of how B-Raf and ␣4␤1 integrin interact in T cells.
The activities of all Raf isoforms are subject to complex regulation but have been shown to be dependent on Ras activity for the initiation of Raf activation. The activation of both A-Raf and Raf-1 requires phosphorylation on the B-Raf corresponding residue Ser-455, but this residue has been shown to be constitutively phosphorylated on B-Raf in many cell types (57,58). Therefore, B-Raf activation is solely dependent on Ras activity and phosphorylation of Thr-598/Ser-601 (11). Constitutive activation of MAPK signaling by mutant forms of B-Raf (i.e. V600E) is observed broadly in solid cancers of multiple primary sites (59,60). Leukemias of lymphoid origin express B-Raf, and other than the recent discovery of hairy cell leukemia, mutations of the BRAF gene are very rare, suggesting a fundamental and conserved role in T cell leukemia and possibly normal T cell function (61)(62)(63)(64)(65)(66)(67). Although there are few studies that focus on the importance of B-Raf to T cell physiology, it has been shown that MAPK signaling during T cell development progression beyond the CD4-CD8 double positive stage requires B-Raf, and rescue experiments with B-Raf can restore proliferation through MAPK signaling in anergic T cells (40,67).
Specific inhibitors of mutant B-Raf were approved for treatment of melanoma in 2011, whereas general RAF inhibitors have been clinically used since 2007 (68 -70). Chemical inhibition of MEK or RAF results in the inhibition of ERK and is sufficient to stop proliferation of many cancer cells (71). Sorafenib is a multikinase inhibitor once higher concentrations have been reached, but it is most specific for C-Raf (6 nM IC 50 ) and B-Raf (22 nM IC 50 ) and is still used in the clinic for the treatment of advanced renal cell carcinoma or hepatocellular carcinoma (29). The findings presented in this work should cause a reevaluation of clinical use of sorafenib having potential off-target effects on T cells (72)(73)(74). The effects of sorafenib on ␣4␤1 integrin could impact T cell migration and homing to sites of inflammation or potentially effector functions (75)(76)(77). This is significant because the selectivity of this interaction may provide a therapeutic target for ␣4 integrin-related diseases, such as asthma, multiple sclerosis, rheumatoid arthritis, inflammatory bowl disease, and certain leukemias and lymphomas (78). Given that we have focused on sorafenib inhibition of B-Raf Ser-445 and the rarity of B-Raf mutations in T cells, this work also suggests that specific V600E B-Raf inhibitors may have fewer off-target effects on T cells (79).
We propose a novel secondary role for B-Raf in the up-regulation of cytoskeletal association of ␣4␤1 and cell spreading mediated by the ␣4␤1 integrin to regulate resistance to shear stress. Interestingly, this effect is independent from downstream MEK/ERK signaling, and within the scope of our studies, this association is unique to ␣4␤1 integrin and not to other integrins or unrelated surface proteins. Both ␣4␤1 integrin and B-Raf play important roles in human diseases, and understanding the mechanisms of their functions will provide important insights into the adaptive immune response and design of therapeutic strategies.