The RhoA Activator GEF-H1/Lfc Is a Transforming Growth Factor-β Target Gene and Effector That Regulates α-Smooth Muscle Actin Expression and Cell Migration

Reagents, Cell Culture, and Treatments

Recombinant human TGF-β1 was from (PeproTech Rocky Hill, NJ). SB431542, actinomycin D, and cycloheximide were from Sigma Chemical (Poole, Dorset, United Kingdom). RPE cells were isolated from porcine eyes (Lee et al., 2001) and used at passage 1. For TGF-β1 experiments, cells were plated at 3 × 10⁴ cells/cm², serum starved (0.5% fetal bovine serum) for 24 h, and then stimulated by adding 10 ng/ml TGF-β1 for the indicated times. For inhibitor studies, cells were preincubated with 10 μM SB431542, 50 ng/ml actinomycin D, or 10 μg/ml cycloheximide for 1 h and then treated with TGF-β1 in the continuous presence of the inhibitor for the specified time. For spontaneous transdifferentiation, RPE cells (passage 1) were plated at 0.5 × 10⁴ cells/cm²; in some experiments, cells were cultured in the presence of SB431542 (10 μM; 14 d). The human keratinocyte cell lines HaCaT-TR (stably expressing the tetrcycline [Tet] repressor) and HaCaT-TR-S4 (stably expressing the Tet-inducible Smad4 small interfering RNA (siRNA) in addition to the Tet repressor) have been characterized previously (Levy and Hill, 2005). For siRNA induction, the cells were grown for 48 h in the presence of tetracycline (2 μg/ml) and then stimulated by adding 2 ng/ml TGF-β1 for 4 d in the continuous presence of tetracycline. Madin-Darby canine kidney (MDCK) cells allowing the conditional depletion of GEF-H1 were described previously (Benais-Pont et al., 2003).

Immunoblotting and Immunofluorescence Microscopy

Total cell extracts were prepared in SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer and western blotting was performed using standard procedures. For immunofluorescence, cells were fixed in ice-cold methanol and processed for immunostaining as described previously (Benais-Pont et al., 2003). Photographs were obtained with an LSM510 confocal microscope (Carl Zeiss, Jena, Germany) using a 63× objective, and the manufacturer's image acquisition software. Brightness and contrast of the images were adjusted with Photoshop (Adobe Systems Mountain View, CA). Antibodies used were as follows: zona occludens-1 (Benais-Pont et al., 2003), occludin (mouse; Zymed Laboratories, South San Francisco, CA), cingulin (rabbit; Zymed Laboratories), GEF-H1 (Benais-Pont et al., 2003), α-SMA (1A4; Sigma Chemical), α-tubulin (1A2; Kreis, 1987), Smad4 (B8; Santa Cruz Biotechnology, Santa Cruz, CA), Slug (Santa Cruz Biotechnology), myosin-IIA (rabbit [Sigma Chemical] and mouse, 3/36), fibronectin (monoclonal F0791; Sigma Chemical) myosin light chain phosphates (MYPT1) and phosphorylated (T696) MYPT1 (Millipore, Billerica, MA). Secondary antibodies conjugated to horseradish peroxidase, fluorescein isothiocyanate, and cyanine 3 were from Jackson ImmunoResearch Laboratories (West Grove, PA). IRDye-680- and IRDye 80-conjugated secondary antibodies were from Li-COR Biosciences (Lincoln, NE) and were used in combination with an Odyssey fluorescence reader.

Immunocytochemistry

On approval of the ethics committee of the local health authority (REC 05/Q0504/17), eyes consented for research were obtained from Moorfields Hospital Eye Bank (London, United Kingdom). Nine evisceration specimens and one enucleation specimen were examined. In all eviscerations, there was relatively extensive disorganization of intraocular contents, generally with at least partial retinal detachment and changes of proliferative vitreoretinopathy. Trauma was the most common underlying pathology with two postinfection cases and one patient with retinopathy of prematurity. The enucleation was carried out because of a choroidal malignant melanoma. Routine, buffered Formalin-fixed, paraffin-embedded sections were cut at 4 to 5 μm in thickness. Hybridoma supernatant with anti-GEF-H1 monoclonal antibody (mAb) was diluted 1 in 3 and incubated overnight followed by washing and alkaline phosphatase-conjugated secondary antibody. Immunoreaction product was visualized using a red alkaline phosphatase-based technique and an Autostainer (Dako, Ely, Cambridgeshire, United Kingdom) after pretreatment in citrate buffer, pH 6.0, in a Pascal pressure cooker (Dako) according to manufacturer's instructions.

Semiquantitative Reverse Transcription-Polymerase Chain Reaction (RT-PCR)

Total RNA was reverse transcribed with specific anti-sense primers using avian myeloblastosis virus reverse transcriptase (Promega, Madison, WI) for 1 h. PCR was carried out in the exponential phase (25 cycles) to allow comparison of PCR product levels. This was achieved by performing initial reactions with different amounts of template to determine optimal amounts of input. For reverse transcription, total RNA (0.5 μg in 15-μl RT reaction) was incubated for 1h at 50°C for GEF-H1 primer 5′-acatctgtcatcagcagga-3′. For PCR 1 μl of RT reaction was used: primers 5′-TTCTCATCACCCAGTTCTCA-3′ (forward) and 5′-acatctgtcatcagcagga-3′ (reverse) and an annealing temperature of 56°C. Other primers used were myosin-IIA (forward 5′-AAGCTGCAGGAGATGGA GGGC-3′; reverse 5′-AAAAAAGAATTCCGGCCTGGAGCT CCTCCTCTTT-3′) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (forward 5′-ATCACTGCCACCCAGAAGAC-3′; reverse ATGAGGTCCACCACCCTGTT-3′).

Microarray Analysis

RPE cells were incubated for 3 d in the absence or presence of TGF-β1 (10 ng/ml), and RNA was isolated. Three samples for each condition were obtained and the RNA quality analyzed using a Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA). cDNA and subsequent cRNA were prepared as described previously (Chambers et al., 2003) and then hybridized to porcine GeneChip arrays according to Affymetrix (Santa Clara, CA) standard protocols (http://www.affymetrix.com) at the University College London, Institute of Child Health Gene Microarray Centre (London, United Kingdom). Labeled GeneChips were scanned, using a confocal argon ion laser (Agilent Technologies). The data were analyzed using Gene Spring 7.2 software (Agilent Technologies). Genes were excluded if the signal strength did not significantly exceed background values and if expression did not reach a threshold value for reliable detection (based on the relaxed Affymetrix MAS 5.0 probability of detection (p ≤ 0.1) in each of the samples (Seo et al., 2004).

Reporter Gene Assays

RPE cells were transfected using Lipofectamine 2000 (Invitrogen, Paisley, United Kingdom) with the indicated reporter promoter constructs driving firefly luciferase expression and an expression construct for GEF-H1 (pCB6-GEF-H1) or empty vector (pCB6), a reference promoter driving Renilla luciferase was used to normalized the data. Reporters genes used were as follows: serum response element (SRE) (SRE containing promoter; Clontech, Mountain View, CA), α-SMA-fl (full-length α-SMA promoter), α-SMA-155 (155-base pair α-SMA promoter construct), and α-SMA-BmAm (155-base pair α-SMA promoter construct with mutated SRE elements; Liu et al., 2003). After 26 h, firefly and Renilla luciferase were measured. Where indicated, cells were incubated with 0.5 μM TAT-C3 transferase, a membrane permeable C3 transferase (Coleman et al., 2001) for 24 h.

Transfection of siRNAs and Determination of Active RhoA

HaCaT cells were plated into 12-well plates, to determine Rho activation, or 24-well plates, to analyze protein expression. Cells were then transfected with nontargeting control siRNA pools or siRNAs specific for GEF-H1, Snail, and Slug (Thermo Fisher Scientific, Waltham, MA; and Dharmacon RNA Technologies, Lafayette, CO), using Interferin transfection reagent (Polyplus Transfection, Calne, Wilts, United Kingdom), and a total final siRNA concentration was 100 nM (Steed et al., 2009). Twenty-four hours after the transfection, TGF-β1 (20 ng/ml) was added, and the cells were analyzed after another 3 d of culture. The cells were then rinsed with phosphate-buffered saline (PBS) and lysed in SDS-PAGE sample buffer for protein analysis, or levels of active RhoA were measured with the G-LISA assay kit (Cytoskeleton, Denver, CO). The assay was performed as instructed by the manufacturer but avoiding the freezing step in all samples.

Inhibition of α-SMA Expression by DN-GEF-H1

RPE cells were transiently transfected with pcDNA4/TO-CTD-VSV construct (DN-GEF-H1 (Aijaz et al., 2005) using Lipofectamine 2000 and then treated with TGF-β1 for 3 d. Samples were fixed and stained for α-SMA, vesicular stomatitis virus (VSV; transfected cells), and DNA. Random fields were photographed and the percentages of α-SMA positive cells in the control (VSV-negative) and DN-GEF-H1 expressing (VSV-positive) cells were calculated (a total of 600 cells were counted for each condition, shown are means ± 1 SD of 3 determinations). For the lentiviral constructs, either the CTD-VSV sequence or the VSV epitope cassette were cloned into the lentiviral pHR'IN plasmid (Bainbridge et al., 2001), giving rise to LNT-DN-GEF-H1 or LNT-control, respectively. RPE cells were infected with control (LNT-VSV) or DN-GEF-H1 (LNT-VSV-DN-GEF-H1) lentivirus at multiplicity of infection (MOI) 100 and stimulated with TGF-β1 for 3 d (experiments were performed in triplicates). Cells lysates were analyzed for α-SMA and fibronectin expression; α-tubulin was used as loading control.

Wound-Healing and Morphological Assays

The Electric Cell-Substrate Impedance Sensing (ECIS) Model 1600R (Applied BioPhysics, Troy, NY) was used to monitor cell migration. RPE cells were plated in ECIS electrode array (8W1E) (Applied BioPhysics). The following day, they were infected with LNT-control or LN T-DN-GEF-H1 at MOI 100 and 24 h later treated with TGF-β1 (50 ng/ml) for 2 d, and electrical wounds were inflicted as described previously (Keese et al., 2004). In another type of wound-healing assays, manual wounds were inflicted with a pipette yellow tip, pictures were then taken after 0, 16, 24, 48 and 72 h, and the wound area was measured. The wound areas were normalized to the ones obtained at 0 h that are referred as 1 and all other areas were then expressed as fractions of the initial wound. For MDCK cells, the same wounding and impendence assay was used. GEF-H1 depletion was induced for 4 d before the wounding assay using tetracycline and was confirmed in parallel experiments as described previously (Benais-Pont et al., 2003).

To follow morphological changes, RPE cells confluent for 3 wk were split and plated on glass coverslips coated with fibronectin (6 × 10⁴ cells/cm²), cells were infected with LNT-control or LN T-DN-GEF-H1 at 200 MOI, and reached confluence after 1 d. Monolayer detachment/contraction was then followed for up to 5 d. Quantification of monolayer detachment/contraction was performed by measuring cell-free areas using ImageJ (National Institutes of Health, Bethesda, MD.

TGF-β–induced Disorganization of Cell–Cell Adhesion Correlates with Up-Regulation of GEF-H1

We used primary porcine RPE cells as a model to analyze TGF-β signaling because they form well-differentiated monolayers in culture and respond to TGF-β (Lee et al., 2001; Ablonczy and Crosson, 2007). As expected, addition of TGF-β stimulated dissolution of cell–cell adhesion structures, such as adherens and tight junctions, correlating with altered cell morphology and reduced expression of junctional proteins, such as ZO-1 and occludin (Figure 1, A–F).

RhoA is a key player in the control of the actin cytoskeleton, cell–cell adhesion and gene expression (Fujita and Braga, 2005; Hall, 2005; Posern and Treisman, 2006; Heasman and Ridley, 2008; Nelson, 2008). To identify the Rho activators that transmit the TGF-β stimulus, we performed a genome-wide expression analysis using microarrays. Total RNA was isolated from triplicate samples of control and TGF-β–treated RPE cells and used to probe Affymetrix porcine arrays. GEF-H1 was the only detectable Rho exchange factor that was up-regulated in response to TGF-β1 (Table 1), suggesting that induction of GEF-H1 expression is likely to be of functional relevance for TGF-β-induced responses in RPE cells.

GEF-H1/Lfc is a guanine nucleotide exchange factor for RhoA (Benais-Pont et al., 2003; Aijaz et al., 2005; Birkenfeld et al., 2008). In contrast to permanent cell lines, GEF-H1 is expressed at very low levels in primary cultures of differentiated RPE cells (Figure 1G), similar to the levels previously reported for adult epithelial tissues (Ryan et al., 2005). Stimulation with TGF-β1, however, up-regulated GEF-H1 expression (Figure 1G). Similar results were obtained with GEF-H1 antibodies recognizing different epitopes (data not shown). Immunofluorescence also revealed increased expression of GEF-H1 and accumulation in areas of cell protrusions (Figure 1D). Up-regulation of GEF-H1 not only correlated with increased expression of α-SMA (Figure 1G) but also enhanced phosphorylation of myosin phosphatase (Figure 1I), suggesting increased activity of the Rho–Rho kinase pathway. Thus, up-regulation of GEF-H1 by TGF-β correlates with activation of Rho signaling and α-SMA expression.

We next analyzed the importance of GEF-H1 for the activation of RhoA in response to TGF-β treatment. We used the HaCaT cells, a keratinocyte cell line, for this purpose as we could down-regulate GEF-H1 in these cells effectively with commercially available siRNAs, and control siRNAs did not cause nonspecific effects as in primary porcine RPE cells. Figure 2A shows that TGF-β treatment also resulted in increased GEF-H1 expression in HaCaT cells. Up-regulation was inhibited by transfection of GEF-H1-specific but not nontargeting control siRNAs (Figure 2B). When the levels of active RhoA was measured in identically treated cells, we found that active RhoA levels increased in response to TGF-β and that this was inhibited if up-regulation of GEF-H1 was blocked by RNA interference. These data thus indicate that the increased levels of GEF-H1 expression contribute to Rho activation in response to TGF-β.

TGF-β Transcriptionally Up-Regulates GEF-H1 Expression through a Smad4-dependent Pathway

In response to TGF-β, activated Smad2 and Smad3 form complexes with Smad4 and accumulate in the nucleus, where they regulate expression of TGF-β target genes (Ross and Hill, 2008). Therefore, we next analyzed whether the TGF-β–induced up-regulation of GEF-H1 at the protein level and the increased mRNA levels observed by microarray analysis were due to changes at the transcriptional level and whether up-regulation depended on Smad4.

Figure 3A shows that increased GEF-H1 mRNA levels in response to TGF-β were also observed if analyzed by semiquantitative RT-PCR instead of microarrays: in both types of assays, an approximately twofold up-regulation of GEF-H1 mRNA was observed. To determine whether transcription was required for up-regulation of GEF-H1 protein, we treated the cells with actinomycin D or cycloheximide. Both drugs inhibited induction of GEF-H1, indicating that transcription is required (Figure 3B).

We next tested whether up-regulation of GEF-H1 involves the canonical Smad pathway (Derynck and Zhang, 2003). Treatment of RPE cells with the ALK5 kinase inhibitor SB431542 abrogated GEF-H1 expression (Figure 3C), indicating that TGF-β type I receptor kinase activity is necessary for GEF-H1 expression. To test involvement of the Smad pathway directly, we used again HaCaT cells that stably express a tetracycline-inducible shRNA targeting Smad4 (HaCaT-TR-S4 cells) (Levy and Hill, 2005). TGF-β1 induced the expression of GEF-H1 in control HaCaT cells, and Smad4 depletion inhibited TGF-β1 induced GEF-H1 up-regulation (Figure 3D), revealing that GEF-H1 induction by TGF-β1 requires Smad4. These observations thus indicate that up-regulation of GEF-H1 involves activation of the TGF-β type I receptor kinase and the Smad pathway.

Microarray analysis previously identified two populations of TGF-target genes: Smad-dependent and -independent genes (Levy and Hill, 2005). Because the Smad-dependence groups TGF-β–responsive genes are into different functional groups, the observed Smad-dependence for GEF-H1 suggests that it may function in Smad-dependent processes such as cell migration. Certain other TGF-β–stimulated processes, such as EMT, are Smad-independent and require the up-regulation of other transcription factors, such as Snail and Slug (Levy and Hill, 2005). Hence, we tested whether Snail and Slug are involved in GEF-H1 up-regulation by transfecting cells with siRNAs targeting the two transcription factors before TGF-β stimulation. Figure 3E shows that up-regulation of GEF-H1 was not prevented by down-regulation of Slug. We were not able to detect Snail in HaCaT cells using two different antibodies, suggesting that Snail does not become up-regulated in these cells. This is in agreement with previous observations (Levy and Hill, 2005). Thus, TGF-β1 induced GEF-H1 expression is smad4 dependent and does not require up-regulation of Slug.

Primary RPE cells in culture transdifferentiate into myofibroblast-like cells not only when treated with TGF-β but also spontaneously when plated at low density (Grisanti and Guidry, 1995; Lee et al., 2001; Wiencke et al., 2003). Transdifferentiated RPE cells up-regulate both α-SMA and GEF-H1, supporting a myofibroblast-like phenotype (Figure 1G). Strikingly, treatment of low-density cultures with the ALK5 inhibitor prevented morphological changes (Figure 3F) as well as α-SMA and GEF-H1 up-regulation (Figure 3G), further supporting the correlation between TGF-β signaling and expression of α-SMA and GEF-H1.

TGF-β modulates cellular phenotypes not only by regulating α-SMA expression but also of nonmuscle myosin isoforms (Sinha et al., 2004; Obara et al., 2005). Therefore, we examined whether TGF-β stimulation affects myosin-IIA expression. Indeed, TGF-β increased myosin-IIA expression with similar kinetics as expression of GEF-H1 (Figure 4A). Although the ALK5 kinase inhibitor abrogated myosin-IIA up-regulation (Figure 4A), depletion of Smad4 only partially counteracted the increase (Figure 4B) and mRNA levels did not significantly change in response to TGF-β (Figure 4C), indicating posttranscriptional regulation by a mechanism at least partially distinct from the one that targets GEF-H1.

GEF-H1 Regulates α-SMA Expression Induced by TGF-β1

Induction of α-SMA expression has been suggested to be of functional relevance for the pathologies of retinopathies such as retinal detachments or proliferative vitreoretinal disorders (Grisanti and Guidry, 1995). Transcription of α-SMA is regulated by actin reorganization induced by Rho activation through serum response factor (SRF) (Hill et al., 1995; Wamhoff et al., 2006). Because GEF-H1 is an activator of RhoA and can activate an SRE-specific reporter gene construct in MDCK (Aijaz et al., 2005) and RPE cells (Supplemental Figure 1), we next asked whether GEF-H1 stimulates α-SMA expression in response to TGF-β.

We first used a reporter gene assay to determine whether GEF-H1 is able to stimulate transcription of the α-SMA promoter and, if yes, whether this involves the SREs, the binding sites of the transcription factor SRF (Mack and Owens, 1999; Miano et al., 2007). Figure 5A shows that cotransfection of GEF-H1 stimulated the full-length promoter (α-SMA-fl) and a shorter promoter (α-SMA-155) in a manner that depended on the two SRE elements (α-155-BmAm) (Figure 5B). As expected, the short promoter responded more strongly to Rho activation as it lacks the repressing upstream region of the promoter. Inhibition of Rho with C3 transferase counteracted stimulation of the α-SMA promoter by GEF-H1 (Figure 5C), confirming the Rho dependence. Thus, GEF-H1 regulates α-SMA promoter activity in an SRE- and Rho-dependent manner.

We next tested whether GEF-H1 and Rho signaling regulate α-SMA expression during TGF-β stimulation. First, we incubated control and TGF-β–treated RPE cells with membrane-permeable C3 transferase. Figure 6A shows that this resulted in an efficient repression of α-SMA induction, indicating up-regulation requires Rho signaling.

We next used RNA interference to down-regulate GEF-H1 expression in RPE cells. However, various control siRNAs already repressed α-SMA levels, indicating an unspecific effect of siRNAs in RPE cells. Therefore, we made use of a dominant-negative (DN) construct containing the C-terminal domain (CTD) of GEF-H1and a C-terminal VSV epitope as a tag (DN-GEF-H1). SRE reporter assays confirmed that DN-GEF-H1 is able to suppress SRE-driven transcription (Supplemental Figure 2).

We next used a transient transfection assay to determine whether DN-GEF-H1 is able to counteract α-SMA-induction by TGF-β1. Double immunofluorescence revealed that most DN-GEF-H1–expressing cells failed to up-regulate α-SMA (Figure 6B). Quantification demonstrated that only 25% of the DN-GEF-H1–expressing cells were positive for α-SMA, whereas 60% of the control cells expressed the EMT marker (Figure 6C). For biochemical quantification, we repeated the experiment with lentiviral vectors to transduce RPE cells with DN-GEF-H1 (LNT-DN-GEF-H1) or a control lentivirus (LNT-control) and then stimulated with TGF-β1. Immunoblot analysis showed that LNT-DN-GEF-H1 transduction resulted in a 2.7-fold decrease in the α-SMA expression compared with LNT-control (Figure 6D). If the samples were probed for expression of fibronectin, a TGF-β target gene that is up-regulated in a Smad-independent manner (Tsuchida et al., 2003), no inhibition of up-regulation was observed. These results show that expression of DN-GEF-H1 counteracts the TGF-β1–induced increase in α-SMA expression.

Treatment of cultures plated at low-density (0.5 × 10⁴ cells/cm²) with the ALK5 inhibitor prevented morphological changes (Figure 3F) as well as α-SMA and GEF-H1 up-regulation (Figure 3G). ALK5 inhibitor also prevented the generation of gaps and monolayer detachment and contraction of older primary cultures (>3 wk) that were plated at high density (6 × 10⁴ cells/cm²) on fibronectin after they had reached confluence, indicating that it was also caused by endogenous TGF-β production (Figure 7A). Because inhibition of GEF-H1 counteracts up-regulation of α-SMA expression, we next tested whether it also inhibits monolayer contraction and cell detachment.

Figure 7B shows that monolayers formed by RPE cells infected with a control lentivirus (LNT-control) started to detach and contract, whereas those infected with a virus encoding DN-GEF-H1 (LNT-DN-GEF-H1) did not. Quantification of such images confirmed that expression of dominant-negative GEF-H1 counteracted the appearance of cell-free areas even after 4 d of culture (Figure 7C). Thus, these results indicate that GEF-H1 drives morphological changes such as cell contraction and detachment induced by TGF-β1. Because contraction and detachment were measured by quantification of cell-free areas (Figure 7), further analysis will be necessary to identify whether cell-free areas are due to contraction only or also to reduced adhesion and/or increased cell death.

GEF-H1 Is Up-Regulated in RPE from Patients with Disorganized Retina and Pigment Epithelium

TGF-β signaling and expression of α-SMA have been related to the ability of RPE cells to form periretinal membranes and are thought to contribute to retinal detachments in proliferative vitreoretinopathy (PVR) and in response to trauma (Fuchs et al., 1991; Saika et al., 2004; Zheng et al., 2004). Therefore, we next studied the expression of GEF-H1 in eye sections from patients with retinal detachments due to different types of insults.

In control RPE cells, there was little or no immunoreactivity for GEF-H1 (Figure 8A1), confirming the observations we made in nonstimulated primary porcine cultures and further supporting the conclusion that expression of high levels of GEF-H1 requires a stimulus in most adult tissues. In contrast, in pathological specimens, there was consistent GEF-H1 immunoreactivity in subsets of RPE cells that had migrated away from their normal location between photoreceptor outer segments and Bruch's membrane in nine of the ten investigated samples (Figure 8). There were four pigmented RPE cell phenotypes associated with this staining: migratory cells that remained configured as a monolayer, RPE cells around blood vessels, individual migratory cells, or apex to apex islands of RPE cells (Fig. 8, A2–A5). RPE cells were identified on the basis of intense pigmentation and a side-to-side arrangement typical of epithelia except for when arranged as individual cells. Furthermore, their cytoarchitecture was generally cuboidal or polygonal rather than rounded, as would be expected for macrophages that had engulfed uveal pigment. Clusters of CD68 expressing macrophages were, however, identified and they were also strongly immunoreactive (data not shown). These observations indicate that up-regulation of GEF-H1 occurs in response to ocular insults and can be observed in migratory RPE cells in vivo.

GEF-H1 Regulates Cell Migration

The observed up-regulation of GEF-H1 in migratory pigmented RPE cells in vivo, suggests that the exchange factor plays a role in TGF-β-stimulated migration, a process that involves RhoA activation and that is thought to be one of the underlying reasons for failure of retinal detachment surgery due to PVR (Kon et al., 1999; Kim et al., 2006). As TGF-β-induced migration is abolished after silencing of Smad4 in HaCaT cells (Levy and Hill, 2005) as is up-regulation of GEF-H1 (Figure 3D), we next tested whether GEF-H1 contributes to TGF-β-induced RPE migration using manual and electrical wound-healing assays.

RPE cells were infected with LNT-DN-GEF-H1 or LNT-control, pre-stimulated with TGF-β and then wounded either manually or with a strong electrical field. Wound closure was then followed microscopically or by measuring impedance of the monolayer. Figure 8, B and C, shows that expression of DN-GEF-H1 impaired wound closure in both assays. This indicates that the exchange factor indeed regulates TGF-β-induced RPE cell migration.

To test the importance of GEF-H1 for cell migration with a different cell type, we used spontaneously immortalized MDCK cells that constitutively express high levels of the exchange factor (Benais-Pont et al., 2003; Aijaz et al., 2005). We took advantage of previously generated cell lines that permit the tetracycline-induced depletion of GEF-H1 by RNA interference (Benais-Pont et al., 2003; Aijaz et al., 2005). Figure 8D shows that depletion of GEF-H1 resulted in a strong retardation of wound closure in the electrical wound healing assay. Visual inspection of the slides confirmed that the failure in wound healing was due to reduced migration of cells into the induced wound as compared with control RNA interference cells. Thus, GEF-H1 regulates migration of different epithelial cell types and may be of general importance for epithelial migration.