Dexamethasone-induced cellular tension requires a SGK1-stimulated Sec5–GEF-H1 interaction

Glucocorticoids mediate a wide array of physiological processes through their binding to the glucocorticoid receptor, which exerts biological effects through transcriptional processes or non-genomic actions. It is known that endogenous glucocorticoids are required for osteogenesis in mice (Bellows et al., 1987; Sher et al., 2004). Understanding the mechanism of glucocorticoid-receptor-mediated osteogenesis is important for developmental biology and regeneration medicine. Human bone-marrow-derived mesenchymal stem cells (hBMSCs) can be induced into osteoblast differentiation in tissue culture by dexamethasone, a man-made steroid ligand of the intracellular glucocorticoid receptor (Shih et al., 2011; Lo et al., 2012). Mesenchymal stem cell (MSC) differentiation into osteoblasts requires tension force through RhoA signaling, thereby promoting the gene expression of phenotypic osteoblast markers (McBeath et al., 2004; Engler et al., 2006). However, the mechanism by which RhoA signaling is activated through glucocorticoid receptor signaling is unclear. In this study, we investigated how dexamethasone acts to increase RhoA-dependent cellular tension.

Among RhoA activators, GEF-H1 (also known as ARHGEF2) is unique in its association with microtubules (MTs), where its activity in RhoA activation is inhibited by phosphorylation (Krendel et al., 2002; Meiri et al., 2009). Upon MT depolymerization, GEF-H1 is released and becomes activated. Therefore, GEF-H1 activates RhoA signaling in a spatiotemporal manner that is dependent on MT dynamics. In vascular endothelial cells, GEF-H1 has been shown to mediate mechanical-stretch-induced endothelial cell permeability owing to MT destabilization (Birukova et al., 2010). Using integrin-based stretching, it has been reported that GEF-H1 contributes to RhoA activation to control the physical force through a feedback mechanism (Guilluy et al., 2011). We have recently shown that dexamethasone is able to increase the amount of GEF-H1 in focal adhesion sites, where it interacts with myosin-II heavy chain B to confer stress-fiber polarization (Huang et al., 2014). In this study, we further determined the molecular mechanism of how dexamethasone stimulates the function of GEF-H1 to increase adhesion-mediated cellular tension.

A recent study has reported that GEF-H1 interacts with Sec5 (also known as EXOC2), a component of the exocyst, to increase exocytosis through RhoA activation (Pathak et al., 2012). Here, we have shown that treatment with dexamethasone stimulates the Sec5–GEF-H1 interaction, which is necessary to promote the localization of GEF-H1 to focal adhesions and to increase the extracellular deposition of fibronectin fibrils of hBMSCs. We further found that dexamethasone-induced expression of serum-glucocorticoid-induced protein kinase 1 (SGK1) is necessary and sufficient to increase the Sec5–GEF-H1 interaction. It has been reported previously that SGK1 increases MT depolymerization, through which neurite formation is promoted (Yang et al., 2006). In this study, we found that increased expression of SGK1 decreased the density of MT growth in the cell periphery. Given the sequestration of GEF-H1 through MT binding, we propose that glucocorticoid receptor activation upregulates SGK1 to cause MT instability, through which GEF-H1 is released from MTs to increase the interaction with Sec5 in the cell periphery. Because Sec5 is a component of the exocyst, we further propose that this pathway mediates GEF-H1 targeting to focal adhesion sites, which is essential for transducing inside–out signals in order to promote fibril formation through the binding of integrins to fibronectin, thereby regulating adhesion to generate cellular tension.

To determine the effects of dexamethasone on adhesion-mediated cell tension, we first tested the influence of dexamethasone on cell adhesion, stress fibers and spreading. The results showed that dexamethasone treatment expanded the hBMSC-3A6 cell-spreading area within 6 h, with concurrent increases in the intensity of F-actin staining and the number of focal adhesions (Fig. 1A,B). It is known that clustering of integrin α5β1 heterodimers (composed of the subunits ITGA5 and ITGB1) mediates hBMSC adhesion to fibronectin and spreading. The expression levels of ITGA5 and fibronectin were increased upon treatment with dexamethasone in a time-dependent manner (Fig. 1C), whereas the level of ITGB1 remained constant. Moreover, the extracellular deposition of fibronectin colocalized with active ITGB1, as indicated by staining with the 9EG7 antibody, on the cell surface, which became more pronounced after dexamethasone treatment, indicating a correlation of the synthesis and export of fibronectin to form adhesion complexes with fibrils on the cell surface (Fig. 1D). Furthermore, primary hBMSCs were plated onto compliant polyacrylamide gel that had been coated with fibronectin to measure intracellular tension. After incubation of cells in serum-free medium for 30 h, the cells were treated with and without 0.1 μM dexamethasone for 18 h, and we then measured the cell-spreading area and traction stress using traction-force microscopy (TFM) (Hur et al., 2009). The measurement of total intracellular traction stress and digital computation showed that dexamethasone treatment caused an enlargement of the cell-spreading area and a significant increase in net contractile moment, which indicates cellular contractile strength (Fig. 1E). Because the compliant gel used for this traction-force analysis is much softer than coverslips, the increase in cellular tension on gel as a result of dexamethasone treatment is slower than that on coverslips in stress fiber analyses.

We found that the induction of stress fibers after 6 h of dexamethasone treatment was abolished by either 1 h of co-treatment with Taxol, which stabilizes MTs (Fig. 2A). Because GEF-H1 is an MT-regulated RhoA activator, we further found that GEF-H1 knockdown decreased dexamethasone-induced F-actin formation (Fig. 2B). Thus, dexamethasone-induced cellular tension requires MT dynamics and GEF-H1–RhoA signaling. It is known that dexamethasone is also used to induce adipogenic differentiation of MSCs, which downregulates RhoA-mediated contractility. It has been shown that the inhibitory phosphorylation of GEF-H1 at residue S885 promotes the binding of 14-3-3, which then enhances localization of GEF-H1 to MTs (Zenke et al., 2004; Meiri et al., 2009). We found that treatment of hBMSCs with an adipogenesis-inducing cocktail markedly increased phosphorylation at residue S885 of GEF-H1, whereas an osteogenesis-inducing cocktail had no effect (Fig. 2C). It is possible that an additional agent, such as 3-isobutyl-1-methylxanthine (IBMX) – an activator of protein kinase A – in the adipogenesis cocktail is sufficient to increase S885 phosphorylation, thereby suppressing GEF-H1-dependent cellular tension. We then asked whether dexamethasone treatment stimulates GEF-H1 activity. However, the conventional GEF-H1 activity assay using GST-RhoA^G17A beads in a pulldown assay failed to reveal any alterations in GEF-H1 activity upon dexamethasone treatment (data not shown). It is possible that dexamethasone only increases the local activation of GEF-H1, which is not revealed in a total GEF-H1 activity assay. We then analyzed the subcellular changes in GEF-H1 activity by assessing its spatial interaction with RhoA. The signal in a proximity ligation assay (PLA) for the GEF-H1–RhoA interaction was significantly increased by dexamethasone treatment (Fig. 2D). Taxol treatment abolished this interaction signal, indicating the requirement of MT dynamics for GEF-H1–RhoA interaction. Knockdown of RhoA or GEF-H1 abolished the PLA signal, indicating the specificity of the interaction signal of PLA (Fig. S1A). Dexamethasone is a class of steroid hormone that binds to the glucocorticoid receptor and is then translocated to nucleus, where the glucocorticoid receptor regulates the transcription of specific genes. The dexamethasone-induced GEF-H1–RhoA signal and stress fiber formation were abolished by blocking either general transcription or glucocorticoid receptor function with actinomycin D and RU486, respectively (Fig. S1B), indicating the involvement of glucocorticoid-receptor-dependent gene transcription.

It is well established that the exocyst is involved in the adhesion process (Spiczka and Yeaman, 2008; Thapa et al., 2012) and that GEF-H1 interacts with the exocyst through a direct interaction with Sec5 (Pathak et al., 2012). We found that knockdown of Sec5 also reduced the amount of dexamethasone-induced F-actin (Fig. 3A). We then tested whether dexamethasone treatment increases the interaction of Sec5 with GEF-H1. By using co-immunoprecipitation, we were unable to detect the effect of dexamethasone on Sec5–GEF-H1 complex formation (data not shown). It is possible that their interaction is transient and spatially controlled, which cannot be detected by co-immunoprecipitation. We then analyzed their interaction at subcellular locations by performing the PLA. The results of the PLA clearly showed that dexamethasone treatment significantly increased the number of Sec5–GEF-H1 interactions (Fig. 3B). The specificity of the PLA was confirmed by using knockdown experiments (Fig. S2). Taxol co-treatment also abolished the dexamethasone-induced Sec5–GEF-H1 interaction signal (Fig. 3B).

To strengthen the in vivo findings of an interaction between GEF-H1 and Sec5 in response to dexamethasone treatment, we used an in vivo proximity protein labeling system, in which GEF-H1 is fused with APEX, an engineered ascorbate peroxidase that catalyzes biotin-phenol to short-lived phenoxyl radicals (<1 ms), dependent on H₂O₂ (Rhee et al., 2013). The radicals can then covalently react with electron-rich amino acids to biotinylate neighboring and interacting proteins. To this end, hBMSC-3A6 cells were transfected with cDNA-GEF-H1-GFP-APEX and treated with dexamethasone for 6 h. To trigger GEF-H1–GFP–APEX-mediated biotin labeling, cells were incubated in medium to which biotin-phenol had been added and then exposed to hydrogen peroxide for 5 min. After termination of labeling by using PBS containing an antioxidant, cells were lysed, and endogenous Sec5 that had been biotinylated by GEF-H1–GFP–APEX as a result of their interaction was enriched using streptavidin beads. The western blot analysis showed biotinylation of Sec5 by GEF-H1–GFP–APEX in an H₂O₂-dependent manner. Owing to the low transfection efficiency of hBMSC-3A6 cells, the basal signal for Sec5 pulldown was low. However, dexamethasone treatment clearly increased the pulldown amount of Sec5 (Fig. 3C).

It is known that GEF-H1 interacts with Sec5 through the sequence spanning amino acids 119–236 of GEF-H1. Therefore, overexpression of a polypeptide containing this sequence disrupts the interaction (Pathak et al., 2012). Therefore, we generated a cell-permeable hemagglutinin (HA)-tagged TAT-fusion polypeptide, TAT–HA–GEF-H1^{a.a.119–236}, which was delivered into the cell during dexamethasone treatment. Treatment with this peptide reduced both the PLA signal of GEF-H1 and Sec5, and the GEF-H1–GFP–APEX-mediated Sec5 biotinylation (Fig. 3D,E).

Consistently, treatment of cells with the TAT–HA–GEF-H1^{a.a.119–236} polypeptide inhibited the dexamethasone-induced stress fiber formation and GEF-H1–RhoA interaction signaling (Fig. 3F), indicating that the Sec5–GEF-H1 interaction is crucial for these responses to dexamethasone.

Delivery of TAT–HA–GEF-H1^{a.a.119–236} to cells attenuated dexamethasone-induced extracellular fibronectin fibril assembly and ITGA5 adhesion formation after 6 h (Fig. 4A,B), whereas the addition of TAT–HA–GEF-H1^{a.a.119–236} on its own had no effect (Fig. S3). Of note, the expression levels of fibronectin and ITGA5 were unaffected by TAT–HA–GEF-H1^{a.a.119–236} (Fig. 4C). Treating cells with TAT–HA–C3, an inhibitor of RhoA (Aepfelbacher et al., 1996), also did not affect dexamethasone-induced ITGA5 expression (Fig. 4D). Although this treatment did not significantly reduce the amount of ITGA5 on the cell surface, we noticed that dexamethasone-induced ITGA5 fibril formation on the cell surface was clearly diminished. These results suggest that the Sec5–GEF-H1 interaction is essential specifically for extracellular fibronectin fibril formation on the cell surface, which is a process associated with the clustering of active integrin.

GEF-H1 has been identified as a component in the focal adhesion complex and as having a functional interaction with myosin-II heavy chain B in order to regulate anisotropic stress-fiber orientation (Huang et al., 2014). The visualization of the distribution of enhanced green fluorescent protein (EGFP)–GEF-H1 by using total internal reflection fluorescence (TIRF) microscopy showed the dynamic accumulation of GEF-H1 at focal adhesion sites after dexamethasone treatment (Fig. 5A). Our western blot analysis of focal adhesion complexes, which had been isolated by using hydrodynamic force, showed that dexamethasone treatment caused a substantial increase in the amount of GEF-H1 in focal adhesion complexes (Fig. 5B,C), and this was accompanied by increases in ITGA5, paxillin, zyxin, vinculin and α-actinin. Co-treatment with Taxol prevented the increase of GEF-H1 in the focal adhesion fraction but had little effect on the increases of ITGA5, paxillin, zyxin, vinculin and α-actinin. These findings indicate that the increase of GEF-H1 in focal adhesions is uniquely sensitive to MT dynamics. We then tested whether the Sec5–GEF-H1 interaction plays an essential role in determining GEF-H1 distribution in focal adhesions in response to dexamethasone stimulation. To this end, we tested the effect of the TAT–HA–GEF-H1^{a.a.119–236} peptide on the focal adhesion targeting of GEF-H1, indicated by paxillin staining, and EGFP–GEF-H1 distribution during dexamethasone treatment. It should be mentioned that EGFP–GEF-H1 by itself caused enlarged focal adhesions at the cell periphery region. With TAT–HA–GEF-H1^{a.a.119–236} polypeptide co-treatment, the distance between the boundary of EGFP–GEF-H1 and the cell edge was clearly increased, and the colocalization of EGFP–GEF-H1 with focal adhesions was decreased (Fig. 5D,E). It is likely that the interaction with Sec5 is required for the trafficking of GEF-H1 to the cell periphery and to focal adhesions.

Next, we wanted to understand which glucocorticoid-receptor-regulated gene is responsible for the upregulation of the Sec5–GEF-H1 interaction signal. Aforementioned experiments suggested that MT instability is required for the dexamethasone-induced Sec5–GEF-H1 signal. We then searched for a glucocorticoid-receptor-regulated gene that destabilizes MTs. Serum- and glucocorticoid-induced protein kinase 1 (SGK1), which belongs to the AGC subfamily of Ser/Thr protein kinases, is rapidly induced by glucocorticoids (Webster et al., 1993). By using real-time (RT)-PCR and western blot analysis, we showed that dexamethasone treatment increased SGK1 mRNA and protein expression at 1 h (Fig. 6A). Knockdown of SGK1 abolished the dexamethasone-induced Sec5–GEF-H1 interaction signal and stress fiber formation (Fig. 6B).We co-treated cells with an SGK1 inhibitor, G650, which also strongly suppressed the dexamethasone-induced Sec5–GEF-H1 interaction signal, as well as stress fiber formation (Fig. 6C), indicating the contribution of its kinase activity. Consistently, inhibition of SGK prevented dexamethasone-induced extracellular fibronectin fibril assembly (Fig. 6D). By contrast, overexpression of SGK1 stimulated the interaction of GEF-H1 and Sec5, as dexamethasone treatment did (Fig. 6E). In conclusion, glucocorticoid-receptor-induced SGK1 upregulation is necessary and sufficient to increase the GEF-H1–Sec5 interaction signal. However, SGK1 overexpression on its own was unable to increase the levels of ITGA5, fibronectin on the cell surface and stress fibers (data not shown). This is reasonable because the levels of fibronectin and ITGA5 expression were not induced by SGK1.

We further investigated the role of MTs in the SGK1-induced GEF-H1–Sec5 interaction. Taxol treatment, which stabilizes MTs, prevented the SGK1-induced GEF-H1–Sec5 interaction signal (Fig. 7A), indicating the requirement of MT dynamics. MT depolymerization by using nocodazole treatment restored these signals in cells that had been co-treated with dexamethasone and G650 (Fig. S4A). Thus, MT depolymerization bypasses the requirement of SGK1 activity for the GEF-H1–Sec5 interaction and stress fiber formation. Accordingly, we speculate that the glucocorticoid-receptor-induced SGK1 regulates MT stability in MSCs, thereby increasing the localized release of GEF-H1 for interaction with Sec5. To test this hypothesis, we first treated cells with nocodazole to depolymerize MTs, after which nocodazole was washed out in order to observe the effect of glucocorticoid-receptor-induced SGK1 on MT re-growth. After nocodazole washout for 10 min, MT re-growth established a MT network pattern in control cells. As a comparison, dexamethasone-treated cells exhibited slower MT re-growth, which was reversed by either RU486-mediated glucocorticoid receptor inhibition or G650 co-treatment (Fig. 7B). Knockdown of SGK1 also rescued MT re-growth in dexamethasone-treated cells (Fig. S4B). Reciprocally, overexpression of EGFP–SGK1 by itself caused a lower amount of MT fiber spreading over the cell periphery (Fig. S4C). Taking these results together, we propose that dexamethasone stimulates glucocorticoid-receptor-mediated transcription of SGK1, which has a negative effect on MT polymerization and stability, thereby increasing the localized release of GEF-H1 for interaction with Sec5 and allowing targeting of GEF-H1 to focal adhesions. Together with glucocorticoid receptor-induced expression of ITGA5 and fibronectin, this is crucial for extracellular fibril formation, involving integrin and fibronectin, thus generating adhesion-mediated cellular tension (Fig. 7C).

Glucocorticoid receptor signaling has been long recognized as a chemical factor for inducing many important physiological processes. In this study, we investigated dexamethasone-induced cellular tension in bone MSCs. Our results highlight the link between glucocorticoid receptor signaling and the GEF-H1–exocyst pathway. Most importantly, we have identified SGK1 expression as the key player that mediates the dexamethasone-induced GEF-H1–exocyst pathway through control of MTs.

The exocyst complex is essential for tethering the trafficking vesicle to the plasma membrane before fusion for exocytosis (Pfeffer, 1999). It is known that RalA signaling controls exocytosis by activating Sec5 to stimulate exocyst complex formation (Moskalenko et al., 2002). Disruption of the RalA interaction with Sec5 inhibits protein targeting to the basolateral domain in polarized epithelial cells (Moskalenko et al., 2002). It has been shown that RalA regulates the interaction of Sec5 with paxillin to target the exocyst to the focal adhesion complex (Spiczka and Yeaman, 2008). Therefore, RalA-regulated exocyst targeting might play a crucial role in molecular targeting for the organization of polarity, adhesion and migration. It has been reported that RalA regulates the direct interaction between Sec5 and GEF-H1 in order to activate RhoA signaling for the assembly, stability and localization of the exocyst; these findings pinpoint the importance of GEF-H1 in exocyst-mediated exocytosis (Pathak et al., 2012). Here, our data add to the knowledge of the function of this pathway in bringing GEF-H1 to focal adhesion sites in order to establish the inside–out signal that is essential for fibril formation through the binding of integrin and fibronectin on the cell surface. Here, it should be emphasized that the expression levels of ITGA5 and fibronectin are increased by dexamethasone treatment, independent of the Sec5–GEF-H1 interaction and RhoA activation. However, disruption of this interaction inhibits fibronectin extracellular deposition without affecting the ITGA5 that is expressed on the cell surface or the secretion of fibronectin. Therefore, the Sec5–GEF-H1 interaction is not required for the export of ITGA5 and fibronectin, rather it is required for the deposition of fibronectin fibrils and integrin clustering.

It is well known that the cellular GEF-H1 involved in RhoA activation is mainly upregulated through MT disassembly. Similarly, the Sec5–GEF-H1 interaction is stimulated by treatment with nocodazole and inhibited by Taxol. Thus, the release of GEF-H1 upon MT disassembly is sufficient to activate the Sec5–GEF-H1–RhoA axis. In agreement with this notion, nocodazole treatment also induces osteoblast differentiation (Zhao et al., 2009). Then, the question is whether dexamethasone treatment stimulates MT disassembly, by which the GEF-H1–exocyst signal is promoted. By using immunofluorescence staining of MTs, we did not find a clear change in the pattern of MTs in dexamethasone-treated cells. However, from the nocodazole washout experiment, we did find that dexamethasone treatment caused a delay in MT re-growth and that this effect requires the function of SGK1. Consistent with this finding, SGK1 is also essential for the dexamethasone-induced GEF-H1–Sec5 interaction. Accordingly, we propose that glucocorticoid receptor signaling stimulates the expression of SGK1, the kinase function of which has a negative effect on MT growth. In turn, the amount of GEF-H1 that is free from MT sequestration is increased for interaction with Sec5. It remains an open question as to how the kinase function of SGK1 regulates MTs in hBMSCs. Overall, our data shed new light on the role of dexamethasone-induced SGK1 in facilitating the GEF-H1–exocyst interaction, which cooperates with the induction of fibronectin and ITGA5 expression to establish organization of adhesions for force generation.

Primary hBMSCs were isolated from bone marrow, as previously described, with Institutional Review Board approval and informed consent (Lee et al., 2004a,,b; Shih et al., 2011). These cells were maintained in MesenPRO RS™ medium (Gibco) with 2 mM of l-glutamine. The hBMSC-3A6 cell line was generated and immortalized, as described previously (Hung et al., 2004; Tsai et al., 2010). The cells were maintained in low glucose Dulbecco's modified Eagle's medium (DMEM) (Gibco) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin and 10 μg/ml streptomycin. hBMSC-3A6 cells were used for all transient transfections with expression vector and Turbofect (Thermo Scientific). HEK293T cells were co-transfected with 5 μg of pLKO or pLenti6 vectors encoding small hairpin (sh)RNAs against GEF-H1 or Sec5 together with lentivirus package plasmids containing 4 μg of pCMV delta 8.91 and 1 μg of pMD.G. Culture media were refreshed at 6 h post-transfection and lentivirus-containing media were collected and filtered for infection.

Elastic polyacrylamide gels with a Young's elastic modulus E of 21.5 kPa (10% acrylamide and 0.2% bis-acrylamide) were prepared based on the method developed by Wang and Pelham (1998). Red-fluorescent (excitation 580, emission 605) polystyrene beads (0.2 μm diameter, Invitrogen) were added to the polyacrylamide solution before polymerization. Gels were typically 30–40 μm thick and covalently cross-linked with fibronectin (fibronectin, 20 μg/ml) on the surface of gel. Cells were allowed to attach onto the fibronectin-conjugated substrate overnight in cell culture medium and then incubated with serum-free medium for treatment with dexamethasone. Cell outlines were determined from differential interference contrast microscopy (DIC) images for cell area analysis. The substrate deformation field was obtained from the lateral displacements of fluorescent beads embedded in the gel. The lateral traction stresses exerted by the cells were determined from the measured substrate deformation after solving the equation of static equilibrium for an elastic substrate (Del Alamo et al., 2007; Hur et al., 2009,, 2012). The net contractile moment (M) was calculated from the measured traction stresses by summing the diagonal components of the shear moment matrix in its principal form (M^rot). M=trace (M^rot)=M_xx^rot+M_yy^rot; where M_xx^rot and M_yy^rot represent the total contribution of cell-substrate contraction in the x and y directions (Butler et al., 2002). The procedure was performed using programs written in MATLAB (MathWorks, Natick, MA) (Lee et al., 2013). The finite analysis was performed with ABAQUS (Dassault Systèmes) (Del Alamo et al., 2007; Meili et al., 2010).

Cells were plated onto fibronectin-coated glass-coverslips. Cells were fixed with 4% paraformaldehyde in PBS for 15 min at 37°C. Cells were permeabilized with 0.3% Triton-X100 for 5 min and blocked with 5% normal goat serum for 1 h at room temperature. Cells were incubated with primary antibody overnight at 4°C and secondary antibody for 1 h at room temperature (anti-paxillin, 1:400, Becton Dickinson; anti-α-tubulin, 1:1000, Sigma-Aldrich; goat anti-mouse AlexaFluor568, 1:400; goat anti-mouse AlexaFluor633, 1:400; phalloidin–AlexaFluor488 or –AlexFluor568, 1:200, Invitrogen). For extracellular fibronectin and ITGA5 staining, cells were plated onto glass coverslips without fibronectin coating. Cells were fixed, blocked and stained as described above without permeabilization (anti-fibronectin, 1:400, GeneTex, GTX112794; ITGA5, clone mAb11, a gift from Kenneth Yamada, National Institutes of Health, Bethesda, MD). For MT quantification, cells were treated with 0.2 ng/ml nocodazole for 2 h. After being washed with ice-cold PBS twice, MTs were allowed to re-grow in serum-free medium for 10 min or 20 min at 37°C. Cells were incubated with CSK buffer containing 80 mM PIPES pH 6.8, 1 mM MgCl₂, 4 mM EGTA, 0.5% Triton X-100 and 10 nM Taxol for 30 s, and were then washed with PBS once. After fixation with cold methanol for 10 min at −20°C, cells were blocked and stained for α-tubulin, as described above. Images were obtained by using fluorescence microscopy (Carl Zeiss, AxioObserver A1) and confocal laser scanning microscopy (Carl Zeiss, LSM700). AxioVision Rel. 4.8 and ZEN 2009 light Edition (Carl Zeiss) instruments were used for image quantification and analysis. Statistical analyses were calculated by using unpaired two-tailed Student's t-test.

Serum-starved hBMSCs were co-transfected with EGFP–GEF-H1 and mApple–paxillin. Cells were re-plated onto a 10 μg/ml fibronectin-coated standard circular glass coverslip (0.17 mm, n=1.515), which was then placed into a live-cell imaging chamber. Dexamethasone (0.1 μM) was added into the culture after conditional equilibrium for 1 h. Time-lapse recording was performed by using a total internal reflection fluorescence (TIRF) microscopy (AxioObserver Z1, Carl Zeiss) with an acousto-optic tunable filter (AOTF) laser control and α Plan-Apochromat Oil DIC lens (100×, NA=1.46). Images were captured using an electron-multiplying CCD (EMCCD, Evolve, Photometrics) camera operated through AxioVision (Zeiss) image software.

Focal adhesion fractionation was achieved as described previously (Kuo et al., 2011). hBMSC-3A6 cells were plated onto 100-mm-dishes (around 80% confluence) coated with 10 μg/ml of fibronectin and were washed with PBS once, and then the cells were subjected to hypotonic shock in low-ionic-strength buffer with 2.5 mM tri-ethanolamine (TEA) (pH 7.0) for 3 min at room temperature. Hydrodynamic force was pulsed by using a Waterpik instrument (setting ‘3.5’, Interplak dental water jet WJ6RW, Conair), which was filled with PBS containing PMSF, NaF and Na₃VO₄. The focal adhesions that remained on the dish were harvested for analysis.

A PCR-amplified DNA fragment covering the GEF-H1^{a.a.119–236} sequence was ligated into the pTAT-HA vector by using the XhoI and EcoRI cutting sites. BL21 pLysS was transformed with pTAT-HA-GEF-H1^{a.a.119–236} and pTAT-HA-C3 plasmid for peptide purification. For TAT–HA–GEF-H1^{a.a.119–236} purification, BL21 pLysS were grown at 37°C to OD₆₀₀=0.6 followed by incubation with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) for 4 h at 30°C. Bacteria were centrifuged and lysed in lysis buffer containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 5 mM imidazole, 0.1% Triton X-100 and 6 M urea. After centrifugation, cell lysates were incubated with nickel beads (GE Healthcare) for 1 h at 4°C. Peptide-coated beads were washed with wash buffer containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl and 20 mM imidazole three times. Nickel-bead-bound peptides were eluted with a buffer containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 1 M imidazole and 6 M urea. The eluted solution was dialyzed with a 500-fold volume of PBS for 6–8 h twice and concentrated by using a Centricon instrument (Merck Millipore). Purification of TAT–HA–C3 was similar to that described above.

An in situ PLA kit was obtained from Sigma-Aldrich (catalog number DUO92008). Briefly, cells were fixed with 4% paraformaldehyde at 37°C for 15 min and were permeabilized in 0.3% Triton X-100 for 5 min. Following blocking, cells were stained using anti-RhoA (1:300, Santa Cruz, catalog number sc-418) or anti-Sec5 (1:600, Proteintech group, catalog number 66011) antibodies with an anti-GEF-H1 antibody (1:600, GeneTex, GTX125893) overnight at 4°C. After incubation with the primary antibodies, cells were washed with TBST with 0.1% Triton X-100 for 5 min twice and then incubated with the PLA anti-mouse PLUS probe (Sigma-Aldrich, catalog number DUO92001) and anti-rabbit MINUS probe (Sigma-Aldrich, catalog number DUO92005) for 1 h at 37°C. Cells were washed with TBST for 5 min twice and subjected to the ligation reaction for 30 min at 37°C, followed by the amplification reaction for 100 min at 37°C. Afterwards, Hoechst 33342 and AlexaFluor488-conjugated phalloidin (Invitrogen, A12379) were added for DNA and F-actin staining, respectively. Images were acquired by using fluorescence microscopy (Carl Zeiss, AxioObserver A1). The PLA foci were counted by using ImageJ. Over 50 cells were examined for each experiment.

APEX-mediated biotinylation was performed as described previously (Rhee et al., 2013). Cells were transfected with GFP–APEX and GEF-H1–GFP–APEX overnight. To initiate APEX-mediated biotinylation, 500 μM of biotin-phenol was added to the medium for 30 min before adding 5 mM hydrogen peroxide for 5 min. Cells were washed with quencher solution (10 mM sodium azide, 10 mM sodium ascorbate, 5 mM Trolox in PBS) and lysed with RIPA buffer (50 mM Tris HCl, pH 7.4, 150 mM NaCl, 0.1% SDS, 0.1% DOC, 1% Triton X-100, 1 mM PMSF, 10 mM sodium azide, 10 mM sodium ascorbate, 5 mM Trolox). Afterwards, cell lysates were dialyzed in a buffer (50 mM Tris HCl, pH 7.4, 150 mM NaCl, 0.1% SDS, 0.1% DOC, 1% Triton X-100) to reduce the biotin-phenol and were subjected to incubation with streptavidin beads for 1.5 h and analysis by western blotting.

Statistical significance was determined by using unpaired two-tailed Student's t-test.

We thank Eminy H. Y. Lee (Academia Sinica, Taipei, Taiwan) for providing the SGK1 expression vector and Kenneth Yamada (National Institutes of Health, Bethesda, MD) for providing the antibody against ITGA5 (mAb11).