Rgnef (p190RhoGEF) Knockout Inhibits RhoA Activity, Focal Adhesion Establishment, and Cell Motility Downstream of Integrins

Background Cell migration is a highly regulated process that involves the formation and turnover of cell-matrix contact sites termed focal adhesions. Rho-family GTPases are molecular switches that regulate actin and focal adhesion dynamics in cells. Guanine nucleotide exchange factors (GEFs) activate Rho-family GTPases. Rgnef (p190RhoGEF) is a ubiquitous 190 kDa GEF implicated in the control of colon carcinoma and fibroblast cell motility. Principal Findings Rgnef exon 24 floxed mice (Rgnefflox) were created and crossed with cytomegalovirus (CMV)-driven Cre recombinase transgenic mice to inactivate Rgnef expression in all tissues during early development. Heterozygous RgnefWT/flox (Cre+) crosses yielded normal Mendelian ratios at embryonic day 13.5, but Rgnefflox/flox (Cre+) mice numbers at 3 weeks of age were significantly less than expected. Rgnefflox/flox (Cre+) (Rgnef−/−) embryos and primary mouse embryo fibroblasts (MEFs) were isolated and verified to lack Rgnef protein expression. When compared to wildtype (WT) littermate MEFs, loss of Rgnef significantly inhibited haptotaxis migration, wound closure motility, focal adhesion number, and RhoA GTPase activation after fibronectin-integrin stimulation. In WT MEFs, Rgnef activation occurs within 60 minutes upon fibronectin plating of cells associated with RhoA activation. Rgnef−/− MEF phenotypes were rescued by epitope-tagged Rgnef re-expression. Conclusions Rgnef−/− MEF phenotypes were due to Rgnef loss and support an essential role for Rgnef in RhoA regulation downstream of integrins in control of cell migration.


Introduction
Directed cell migration is a physical process that requires regulated changes in cell shape and adhesion to the extracellular matrix (ECM) [1]. Sites of cell adhesion (termed focal adhesions, FAs) are mediated by integrins, transmembrane receptors that couple the ECM to the filamentous actin cytoskeleton [2]. The migration cycle begins with membrane protrusion, FA formation at the cell front, FA linkage to the actin cytoskeleton, the generation of traction and forward cell movement, followed by disassembly of FAs at the cell rear [3]. At FAs, integrins bind ECM proteins such as fibronectin (FN) and multi-protein signaling complexes form in association with integrin cytoplasmic domains that drive the migration cycle in part through regulation of Rhofamily GTPase activity [4].
Rho GTPases, including Cdc42, Rac1, RhoA, and RhoC are key effectors of cell migration and actin cytoskeletal dynamics that function as molecular switches cycling between an inactive GDPbound state and an active GTP-bound form that interacts with downstream targets [5]. Rho GTPases are activated by guanine nucleotide exchange factors (GEFs) that catalyze the exchange of GDP for GTP [6]. Rho GTPases return to an inactive state upon hydrolysis of GTP to GDP, a reaction enhanced by GTPaseactivating proteins (GAPs) [7]. Initial steps of integrin binding to FN and cell spreading are associated with transient RhoA inhibition followed by a more prolonged period of RhoA activation associated FA formation and the generation of cell tension [8]. Analysis of knockout fibroblasts revealed the importance of both focal adhesion kinase (FAK) and Src-family tyrosine kinases in promoting signals leading to transient RhoA inhibition downstream of integrins [9,10]. Integrin-stimulated Src and FAK tyrosine phosphorylation of p190RhoGAP is associated with elevated RhoGAP activity and the transient inhibition of RhoA needed for efficient cell motility-polarity [11,12,13,14]. Our understanding of GEFs involved in facilitating RhoA reactivation and FA formation upon FN adhesion remains incomplete.
There are at least 69 different proteins that comprise an extended GEF family [6,15]. These GEFs contain a conserved region first identified in a transforming gene from diffuse B-celllymphoma (Dbl), designated Dbl-homology (DH) [16,17]. Many GEFs also contain a pleckstrin homology (PH) domain, known to bind phosphorylated phosphoinositide lipids and promote mem-brane localization [18]. The GEF DH-PH module is the minimal unit promoting nucleotide exchange, but specificity for Rho-GTPase regulation is mediated by additional targeting interactions unique to different GEF proteins [19].
For integrin signaling, knockdown experiments have identified Lsc/p115RhoGEF, LARG, GEF-H1, and p190RhoGEF (Rgnef) as contributing to RhoA activation, actin stress fiber, and FA formation in response to cell adhesion to FN [20,21,22]. LARG and GEF-H1 have been linked to RhoA activation in response to mechanical forces on integrins [23]. Over-expression analyses have revealed partial co-localization of p115RhoGEF, LARG, and Rgnef with integrins at FAs [20,21]. Rgnef binds directly to FAK through a motif in the Rgnef C-terminal domain, a feature not shared with other GEFs [24]. FAK binding directs Rgnef localization to FAs within fibroblasts and this FAK-Rgnef linkage also functions to promote colon carcinoma motility, invasion, and tumor progression [25]. Thus, FAK associates with both GAPs and GEFs in the spatial and temporal control of RhoA regulation and cell motility [26].
Herein, we present results from the generation of an Rgnef knockout mouse. Loss of Rgnef expression does not prevent embryonic development, but numbers of Rgnef2/2 mice obtained at weaning were of smaller size and significantly less in numbers compared to expected Mendelian ratios from Rgnef+/2 crosses. Primary Rgnef2/2 mouse embryo fibroblasts (MEFs) exhibited reduced haptotaxis migration, wound closure motility, and FA numbers formed on FN. Affinity-binding assays to a nucleotide-free RhoA (G17A) mutant revealed Rgnef activation at 60 min after FN stimulation and Rgnef2/2 MEFs exhibited significantly reduced RhoA GTP binding at 60 to 120 min on FN compared to normal Rgnef+/+ MEFs. As Rgnef re-expression rescued Rgnef2/2 MEF phenotypic defects, these studies establish the importance of Rgnef in RhoA regulation, FA establishment, and cell migration downstream of integrins.

Generation of Rgnef knockout mice
Rgnef, originally termed p190RhoGEF [27], is comprised of Nterminal leucine-rich and cysteine-rich zinc-finger-like regions of unknown function, a central DH-PH domain catalytic region, a FAK interaction motif, and a C-terminal potential coiled-coil region that binds other proteins [28,29,30,31] (Fig. 1A). To determine the necessity of Rgnef expression in mouse development, a transgenic mouse model was created by homologous recombination whereby Rgnef exon 24 (coding for part of the DH domain) was flanked by loxP sequences (Rgnef flox ) (Fig. 1B, Fig.  S1). Initial crosses revealed that homozygous Rgnef flox/flox mice developed normally, were fertile, and displayed no detectable phenotype (data not shown).
At e13.5, Rgnef protein expression is maximally detected within the developing brain by immunohistochemical staining with antibodies to Rgnef ( Fig. 2A). This is consistent with findings of Rgnef protein expression within mouse brain tissue extracts at e13 and e19 [24]. Immunoblotting analyses of adult mouse tissue revealed the highest relative level of Rgnef protein expression in the brain, lung, ovary, and spleen (Fig. 2B). Low but detectable levels of Rgnef were found in heart, testes, and kidney. Upon overexposure, an Rgnef band was detected in liver but not in skeletal muscle. These results are consistent with a low ubiquitous pattern of Rgnef expression in most tissues with higher Rgnef levels present in specific organs.

Rgnef is required for normal MEF cell motility
To assess the necessity of Rgnef in cell function, primary MEFs were isolated from e13.5 Rgnef2/2 and Rgnef+/+ embryos ( Fig. 3 and Fig. S2). Rgnef mRNA (Fig. S2) and ,190 kDa Rgnef protein (Fig. 3A) were detected in MEFs from Rgnef+/+ but not Rgnef2/2 littermates. Although total FAK levels were slightly increased and verified by densitometry analyses from independently-derived cell lines, no changes were detected in the FAK paralog Pyk2, related GEF-H1/Lfc, or cytoskeletal protein paxillin expression (Fig. 3A). Generation of N-terminal directed polyclonal antibodies to Rgnef did not detect any truncated protein products in Rgnef+/+ or Rgnef2/2 MEFs (Fig. 3B). This is consistent with no Rgnef mRNA detected using 59 end-directed RT-PCR analyses (Fig. 1D). Taken together, these results support the conclusion that floxed deletion of Rgnef exon 24 results in the absence of Rgnef expression in embryos and primary MEFs.
Primary MEFs were immortalized via large T antigen expression to facilitate cell culture analyses. Since knockdown experiments MEFs [21] and human colon carcinoma cells [25] support the importance of Rgnef in promoting cell motility, Rgnef2/2 and Rgnef+/+ MEFs were analyzed by scratch wound (Figs. 3C and D, Videos S1 and S2) and FN-stimulated haptotaxis transwell (Fig. 3E) migration assays. In both assays, lack of Rgnef expression significantly inhibited cell movement.

Rgnef facilitates integrin-initiated FA establishment
Efficient cell migration requires precise FA assembly-disassembly and stress fiber for tension generation [3,4]. MEFs will spread and form FAs as detected by paxillin staining with integrated actin stress fibers by 60 to 90 min when replated onto FN-coated surfaces [2]. Rgnef+/+ and Rgnef2/2 MEFs were FN replated and analyzed for differences in FA numbers and size at 90 and 120 min (Figs. 4 and 5). Whereas Rgnef2/2 MEFs spread equally and exhibited no morphological differences to normal MEFs, loss of Rgnef resulted in significantly fewer FAs at 90 and 120 min (Figs. 4 and 5). Adhesion formation is a dynamic process that can result either in disassembly or maturation corresponding to an increase in FA size [1]. At 90 min on FN, median Rgnef+/+ MEF FA size was 40 pixels compared to Rgnef2/2 MEFs at 30 pixels as determined by Cell Profiler software analyses (Fig. 4C). At 120 min on FN, median Rgnef+/+ MEFs FA size increased to 50 pixels whereas Rgnef2/2 MEF FA size decreased to 20 pixels (Fig. 5C). The difference in Rgnef+/+ and Rgnef2/2 FA size at 120 min was significant and may contribute to the motility defects of Rgnef2/2 MEFs. Together, these results support the importance of Rgnef in promoting FA establishment and size after integrin binding to FN.

Rgnef activation parallels RhoA GTP binding upon FN stimulation
Replating experiments with Rgnef+/+ and Rgnef2/2 MEFs support the notion that Rgnef activity may influence FA dynamics. Since adhesion to FN triggers the activation of Lsc/p115 and LARG RhoGEFs within MEFs as determined by RhoGEF affinity binding to a nucleotide-free mutant (G17A) of RhoA [20], similar affinity pull-down experiments were performed to evaluate Rgnef activation upon MEF binding to FN (Fig. 6A). Compared to lysates of cells held in suspension or replated on FN for 30 min, Rgnef was significantly activated and bound to GST-RhoA G17A at 60, 90, and 120 min after FN stimulation (Fig. 6A). These results show that Rgnef is activated within 60 min upon cell binding to FN.

Rgnef promotes RhoA activation, increased FA numbers, and cell motility
To determine that Rgnef2/2 MEF phenotypes are linked to the loss of Rgnef expression, green fluorescent protein (GFP) or GFP-Rgnef were transiently re-expressed in Rgnef2/2 MEFs as a rescue strategy (Fig. 6D). GFP-Rgnef significantly increased RhoA GTP binding at 60 and 90 min after Rgnef2/2 MEF replating on FN compared to GFP alone (Fig. 6E). Whereas GFP was primarily distributed in the cytoplasm, a portion of GFP-Rgnef co-localized with paxillin at peripheral FAs when Rgnef2/ 2 MEFs were analyzed by confocal microscopy at 90 min after FN replating (Figs. 7A and B). GFP-Rgnef expression significantly increased the number but not size of FAs formed after 90 min on FN compared to GFP alone (Figs. 7C and D). In parallel, HAtagged Rgnef re-expression resulted in a 6-fold increase in Rgnef2/2 MEF haptotaxis motility on FN after 3 h compared to control-transfected cells (Fig. 7E). When analyzed by real time

Discussion
RhoA GTPase regulation in response to growth factor, G protein linked, and integrin receptor stimuli plays key roles in the formation and regulation of FAs at both leading and trailing edges of migrating cells [4]. Transient knockdown studies have implicated specific signaling pathways connected to the sub-family of RhoA, RhoB, and RhoC GTPases in the control of cell migration [33]. However, results from mouse genetic knockouts (KOs) support the notion of functional overlap within this RhoGTPase sub-family [34]. RhoB KO and RhoC KO mice have no major developmental defects and MEFs derived from these mice exhibit either minor alterations in actin stress fibers (RhoC) or reduced motility and b1 integrin expression (RhoB) [35,36]. Conditional KO of RhoA in keratinocytes showed that it is not essential for skin development or wound healing in vivo, but RhoA KO keratinocytes exhibit motility defects in culture [37]. In contrast, analyses of RhoA KO MEFs revealed redundant signaling roles with RhoB and RhoC in actomyosin regulation in culture [38]. Thus, understanding the factors of Rho GTPase sub-family regulation is complex and likely associated with the fact that GEFs outnumber Rho GTPases [6]. Moreover, spatiotemporal signal integration within cells is important in the control of cytoskeletal dynamics needed for efficient cell movement [4].
Here we provide the first characterization of an Rgnef KO mouse. From heterozygous crosses, Rgnef2/2 mice are found at expected Mendelian ratios at e13.5, but are born at lower than expected Mendelian frequency and exhibit an overall smaller size than Rgnef+/2 or Rgnef+/+ littermates. Gross analyses of Rgnef2/2 offspring did not reveal apparent abnormalities and this size difference was negligible by 6 to 8 weeks of age. We hypothesize there is an important role for Rgnef in mouse development, but that some type of partial redundancy or compensation may be occurring to lessen or bypass the developmental or physiological restriction point between e13.5 and birth. Although it remains unclear whether the Rgnef KO phenotype parallels RhoA inactivation, embryonic lethal phenotypes are uncommon in other RhoGEF KO mouse models ( Table 2). Except for KO of AKAP13 (Brx) that results in heart developmental defects [39] or Trio that results myofibril defects in late embryonic development [40], other RhoGEF KOs have either restricted hematopoietic or other non-lethal phenotypes ( Table 2). Despite transient knockdown studies implicating Lsc/ p115 RhoGEF and LARG in FN signaling to RhoA [20], mice from these KOs were viable and fertile with either a leukocyte homeostasis defect [41,42] or smooth muscle hypertension defects that did not necessarily involve integrin signaling connections [43]. Interestingly, as observed with Rgnef KO mice, KOs of the RhoA effector proteins ROCK1 or ROCK2 (Rho-associated protein kinases) also result in partial embryo lethality and birth of small pups [44,45]. ROCK2 loss was associated with late placental dysfunction and ROCK1 loss with cellular actomyosin bundling defects. Future studies of Rgnef KO embryos in utero between e13.5 and birth will be focused on identifying potential phenotypes as a means to link Rgnef to RhoA signaling in vivo.
Many of the restricted hematopoietic or neural defects associated with RhoGEF KOs are associated with alterations in cell movement (Table 2). Surprisingly, there are no reports characterizing cell motility defects RhoGEF KO MEFs in cell culture. It is established that initial MEF binding and spreading on FN is associated with transient RhoA inhibition followed by a more prolonged period of RhoA activation associated FA formation and the generation of cell tension [8,10]. When compared to wildtype littermate MEFs, Rgnef KO significantly inhibited haptotaxis migration, wound closure motility, FA   In WT MEFs, Rgnef activation occurs within 60 min upon FN stimulation and this parallels the time course of FN-induced RhoA activation. Importantly, Rgnef KO MEF phenotypes were rescued by Rgnef re-expression; demonstrating that this was a direct result of Rgnef loss. Moreover, our results with Rgnef KO MEFs are in complete agreement with previous studies using short-hairpin RNA interference to knockdown Rgnef expression [21] that resulted in fibroblasts that exhibited motility defects on FN with fewer FAs formed but no alterations in cell spreading.
With regard to the processes involved in FA formation and the control of cell migration, it is important to note that RhoA GTPase regulation is a cycle and that inhibition or constitutive activation of Rho GTPases result in cell migration defects. Too many or too few adhesions formed can prevent efficient cell movement [46]. Moreover, it is overly simplistic to equate decreased FA formation and size differences observed in Rgnef KO MEFs to the inhibition of cell migration. Future studies evaluating the kinetics of FA formation and turnover at both leading and trailing cell projections, studies evaluating the processes of FA maturation or tension generation, and potential differences in adhesome protein content may lead to further mechanistic insights associated with Rgnef KO MEF motility defects. Moreover, Rgnef contains a unique region that binds to FAK and is required for Rgnef localization to FAs [21,24]. It has been hypothesized that this Rgnef linkage to FAK is key to FNinduced RhoA regulation and future re-expression studies using Rgnef KO MEFs will serve as a powerful system to elucidate the molecular connections of this signaling pathway.
In addition to facilitating RhoA in MEFs, Rgnef expression is elevated as a function of colon cancer tumor grade and stage [25]. A complex of Rgnef, FAK, and paxillin promote colon carcinoma tumor cell motility and matrix degradation in vitro with corresponding increases in tumor growth and surrounding tissue invasion in vivo [25]. Cellular projections that promote matrix degradation are termed invadopodia and Rgnef localizes around these sites to activate RhoC in breast carcinoma cells [47]. Whether Rgnef signaling connections linked to tumor progression involve RhoA or RhoC remain unknown. As Rgnef KO mice are fertile, analyses of these mice crossed to spontaneous mouse tumor models will serve to elucidate the role of Rgnef in promoting tumor invasion and progression.

Ethics Statement
All procedures were approved by UCSD Institutional Animal Care and Use Committee and mice were maintained in accordance with Association for Assessment and Accreditation of Laboratory Animal Care-approved guidelines.  Mice Rgnef exon 24 floxed mice were created by homologous recombination (InGenious Targeting Laboratory, Stony Brook, NY) and the cloning strategy depicted (Fig. 1B). An 8.64 kb region (encompassing Rgnef exons 23-26, Ensembl release 65 -Dec 2011) from a C57BL/6 BAC clone (RPCI23 388E6) was used to construct the targeting vector in pGK-gb2 loxP/FRT Neo. The region was designed such that the short homology arm (SA) extends about 1.42 kb 59 to exon 24. The long homology arm (LA) ends 39 to exon 24 and is 6.77 kb long. The loxP flanked Neo cassette is inserted on the 59 side of exon 24 and the single loxP site is inserted at the 39 side of exon 24. The target region is 455 bp and includes exon 24 (Fig. S1A). Targeting vector was confirmed by restriction digest and sequencing after each modification. The targeting vector was subcloned into pSP72 prior to linearization (NotI) and electroporation into mouse BA1 (C57BL/6 x129/ SvEv) hybrid embryonic stem (ES) cells. The total size of the targeting construct (including vector backbone and Neo cassette) was 12.84 kb.
After G418 selection, surviving ES clones were expanded for PCR analysis to identify recombinants. Screening primers A1 and A2 were designed upstream of the short homology arm outside the 59 region used to generate the targeting construct. A list of PCR primers used is provided in Table 3. PCR reactions using A1 or A2 with the UNI primer (located within the Neo cassette) amplify 1.84 and 1.96 kb fragments, respectively. The control PCR reaction was performed using the internal targeting vector primers AT1 and AT2, which are located at the 59 and 39 ends, respectively, of the SA. This amplifies a product 1.17 kb in size. Individual clones from positive pooled samples were screened using the A2 and UNI primers. Five positive SA recombinant clones were identified by a 1.8 kb PCR fragment (Fig. S1B) and sequenced for integration using the OUT1 primer. Confirmation of integration of LA region was performed by PCR using LAN1 and Lox1 (Fig. S1B) or SDL2 and Lox1 primers and sequenced using the LAN1 and OUT1 primers to confirm the presence of the third LoxP site. A secondary confirmation of positive clones was performed by Southern Blotting analyses of DNA digested with StuI using a probe targeted against the 39 LA region (Fig. S1C). Two confirmed clones (252 and 362) were microinjected into C57BL/6 blastocysts. Resulting chimeras with a high percentage agouti coat color were mated to wild-type C57BL/6 mice to generate F1 heterozygous offspring that were genotyped using primers flanking the neomycin cassette insertion site.
For RT-PCR, poly-A RNA was isolated using RNeasy (Qiagen) and cDNA was generated using 1 mg RNA and the SuperScript III first-strand synthesis system (Invitrogen). PCR reaction conditions were performed using a BioRad S1000 thermal cycler and GoTaq green master mix (Promega). Primers used are shown in Table 3.

Immunohistochemistry
Embryos were fixed in 4% paraformaldehyde (4uC, overnight) and dehydrated in ethanol/water washes prior to paraffin embedding. Sections (10 mm) were floated onto SuperFrost Plus slides (Fisher) and dried overnight at 37uC. Prior to staining, slides were incubated at 60uC for 30 min, deparaffinized in xylene washes, and rehydrated in ethanol/water washes. Antigen retrieval (boiling for 10 min in 10 mM sodium citrate) and peroxidase quenching (0.3% hydrogen peroxide for 10 min) were performed. Sections were incubated in Blocking Buffer (PBS with 5% normal goat serum, 1% BSA, and 0.3% Triton X-100) for 60 min at room temperature and then incubated in anti-Rgnef antibodies (1:200 in Blocking Buffer) overnight at 4uC. Biotinylated goat-anti-rabbit IgG (1:300), Vectastain ABC elite, and diaminobenzidine (Vector Labs) were used to visualize Rgnef antibody binding. Images were acquired using an Olympus IX81 microscope (46 objective), an Infinity1 color CCD camera, and whole embryos images stitched together using Adobe Photoshop CS3 software.

Cell Migration
Haptotaxis. For transient transfection studies, cells were cotransfected with a pCDNA3-LacZ, pCDNA3-Rgnef, or pCDNA empty vector and evaluated after 36 h. MilliCell chambers (8 mm pores; Millipore) were coated on the membrane underside with FN (10 mg/ml) in Migration Medium for 2 h, washed with PBS, and air dried (30 min) prior to use. Cells were collected as described above, 10 5 cells in 0.3 ml were added to each MilliCell chamber, units were placed into 24-well plates containing 0.4 ml Migration Medium, and incubated for 3 h at 37uC. Cells on the lower membrane surface were fixed and visualized either by crystal violet staining or analyzed for b-galactosidase activity using X-gal as a substrate. Cells per microscopic field were counted (9 fields per chamber) and mean values were obtained from three individual chambers for each experimental point per assay.
Wound healing. Glass bottom dishes (MatTek) were coated with FN (1 mg/ml), cells were plated at a subconfluent density (75%), and after 24 h, cells were serum starved overnight, wounded with a pipette tip, washed with PBS, and incubated in growth media containing 0.5 mg/ml of mitomycin-C prior to imaging. Phase contrast images of cells were acquired every 15 min in a humidified 5% CO2 environment at 37uC using an Olympus IX51 microscope, XY-controlled stage with Z focus drive (Olympus), 106 objective (UPLFL, 0.30 NA), and an Orca-ER camera (Hamamatsu) controlled by Slidebook (v5.0) software. Wound closure percentage was calculated by the change in area between 0 and 8 h. Random migration. 1610 4 cells were seeded onto fibronectin (FN)-coated (1 mg/ml) 6-well plates in growth media for 8 h, then serum-starved overnight. Media was replaced with growth media, phase contrast and GFP fluorescent images were collected at 5-min intervals over 5 h with a 106lens on an automated stage (Olympus IX51) at 37uC with humidity and CO 2 regulation. Slidebook (v5.0) software was used to track cell trajectories by nuclear position over time, calculate distance traveled, and average speed of 34 cells per group.

Immunofluorescence
Cells replated for the indicated time on FN-coated (10 mg/ml) acid-washed glass coverslips (as above) were fixed in 3.7% paraformaldehyde (10 min), permeabilized with 0.1% Triton X-100 in PBS (10 min) and incubated in blocking buffer (2% BSA in PBS) for 1 h. Paxillin antibodies were diluted (1:300) in blocking buffer and incubated overnight at 4uC. Coverslips were washed in PBS, incubated with AlexaFluor-488 or -594 goat anti-mouse secondary antibodies, Texas Red-Phalloidin, 49-6-Diamidino-2phenylindole (DAPI) or Hoechst 33342 (Invitrogen) diluted in blocking buffer (30 min), and mounted using Vectashield (Vector Labs). Images were acquired sequentially using a mercury lamp source, multiband dichroic, single-band exciter, and single band emitter filter sets (Chroma) on dual filter wheels, an Olympus IX81 spinning disc confocal microscope at 606 (PlanApo, N.A. 1.42) and an Orca-AG camera (Hamamatsu) controlled by Slidebook (v5.0) software. Files were cropped, pseudocolored, and adjusted using Adobe Photoshop CS3. Adhesion size (pixels) and number within a cell were determined by analysis of paxillin staining using Cell Profiler (v2.0, Broad Institute) using a pipeline to threshold images and reduce background fluorescent staining in at least 10 cells per group or by using Image J (v1.4).

Statistical analyses
Data were analyzed using unpaired Students t-test or one-way ANOVA where indicated with GraphPad Prism (5.0 d). Significance was determined at a p value less than 0.05.  Table 3.(B) PCR confirmation of ES clone recombination. A1 and UNI primers were used to amplify a 1.8 kb sequence within the short arm (left). Lox and Lan1 primers were used to amplify a 1.1 kb band within the long arm (right). (C) Southern blot confirmation of ES clone recombination. StuI-digested DNA was electrophoretically-separated on a 0.8% agarose gel, transferred to nylon membrane, and hybridized with a probe generated by primers PB3 and PB4 to give a 19.1 kb band for wild type and a 9.4 kb band for the recombined allele. ES clones 252 and 362 were used for blastocyst injection. (TIF) Figure S2 MEFs were generated from Rgnef+/+ and Rgnef2/2 embryos. (A) Rgnef and Cre genotyping of primary normal (Rgnef+/+) and Rgnef2/2 MEFs isolated from e13.5 embryos and established in culture (B) Total RNA isolated from cells from primary Rgnef+/+ and Rgnef2/2 MEFs and samples analyzed by RT-PCR using primers to Rgnef and GAPDH. (TIF) Video S1 Rgnef+/+ MEF scratch wound motility. Primary normal (Rgnef+/+) MEFs were plated at a subconfluent density (75%) on glass bottom dishes (MatTek) coated with FN (1 mg/ml). After 24 h, cells were serum starved overnight, wounded with a pipette tip, washed with PBS, and incubated in growth media containing 0.5 mg/ml of mitomycin-C prior to phase contrast imaging every 15 min in a humidified 5% CO2 environment at 37uC using an Olympus IX51 microscope, XY-controlled stage with Z focus drive (Olympus), 106 objective (UPLFL, 0.30 NA), and an OrcaER camera (Hamamatsu) controlled by Slidebook (v5.0) software. Images were acquired every 15 minutes and assembled into 10 frames per second video spanning 8 h.