Gα12/13 Is Essential for Directed Cell Migration and Localized Rho-Dia1 Function*

Scratch-wound assays are frequently used to study directed cell migration, a process critical for embryogenesis, invasion, and tissue repair. The function and identity of trimeric G-proteins in cell behavior during wound healing is not known. Here we show that Gα12/13, but not Gαq/11 or Gαi, is indispensable for coordinated and directed cell migration. In mouse embryonic fibroblasts endogenous Rho activity is present at the rear of migrating cells but also at the leading edge, whereas it is undetectable at the cell front of Gα12/13-deficient mouse embryonic fibroblasts. Spatial activation of Rho at the wound edge can be stimulated by lysophosphatidic acid. Active Rho colocalizes with the diaphanous-related formin Dia1 at the cell front. Gα12/13-deficient cells lack Dia1 localization to the wound edge and are unable to form orientated, stable microtubules during wound healing. Knock down of Dia1 reveals its requirement for microtubule stabilization as well as polarized cell migration. Thus, we identified Gα12/13-proteins as essential components linking extracellular signals to localized Rho-Dia1 function during directed cell movement.

Directional cell migration is a fundamental mechanism in wound repair and embryogenesis. In collective cell migration modes such as during embryogenesis, wound healing or invasion, cells dynamically regulate their cytoskeleton and become polarized toward direction of movement (1,2). Many of these cytoskeletal changes are brought about by Rho-family GTPases, which are activated through cell surface receptors that in turn regulate specific guanine-nucleotide exchange factors (GEFs) (3). In wound healing assays with mammalian cells, the polarized morphology develops 1-6 h after injury of the cell monolayer and is characterized by formation of protrusions at the leading edge containing lamellipodia and filopodia, reorientation of the Golgi and the centrosome, as well as formation of stable and reoriented microtubules (3,4). Integrin signaling has been identified as a critical process for polarized cell migration in scratch-wound assays of cell monolayers (5). The potential role of heterotrimeric G-protein-dependent mechanisms during wound healing is less well understood. G i -proteins have been implicated as signaling intermediates particularly in cell polarization during chemoattractant-guided motility of Dictyostelium discoideum amoebae or leukocytes (6 -8). The G 12/13 family of G-proteins have been implicated in various cellular processes such as Rho-mediated organization of the cytoskeleton and subsequently the cell shape (9). G␣ 12/13 -proteins appear to exert essential functions as mice double-deficient for G␣ 12 and G␣ 13 die at embryonic day E 8.5 (9). Coexpression of dominant negative versions of G␣ 12 and G␣ 13 reduced chemotactic polarity of differentiated HL-60 cells, suggesting that G␣ 12/13 -proteins may function during directional cell migratory processes (8). However, their potential role for wounding induced cell migration and tissue repair has not been investigated.
Studies on wounding induced behavior of NIH3T3 fibroblasts revealed that addition of serum or lysophosphatidic acid (LPA) 3 promotes the formation of a subpopulation of orientated microtubules known as stable, detyrosinated microtubules (Glu-MTs) (10). Overexpression of dominant active versions of RhoA and the diaphanous-related formins have been shown to induce microtubule stabilization (11), but the upstream signaling components await to be defined. Although active mutants of RhoA or diaphanous-related formins can induce stable microtubule formation, it is not apparent whether they are indeed essential for this process. Rho-induced stabilization of Glu-MTs has been characterized as an early event in the generation of cellular asymmetry in wound healing assays and appears to precede the onset of cell migration (10,12). Nevertheless, regulation of RhoA activity during changes in cell-cell contact inhibition and directional migration such as in wound repair has not yet been analyzed directly. Although it has been suggested that Rho is involved in cell migration and is believed to be required for rear end retraction, its precise endogenous localization of activation such as in epithelial or mesenchymal cells is not known (13).
The present study was undertaken to investigate whether trimeric G-protein function is principally required for wound healing in mammalian cells. We used previously established MEF cell lines that are double deficient for either G␣ q/11 or G␣ 12/13 in addition to the usage of pertussis toxin to inactivate G i (14). Additionally, we set out to systematically analyze the downstream signaling of Rho-family GTPases and addressed the role of localized RhoA activation in wounding induced cell migration using a novel GFP-tagged Rho probe combined with an optimized in situ Rho assay approach that was previously reported (15,16). Our analysis provides evidence that G␣ 12/13 is absolutely essential for directional cell migration and wound healing involving Rho activity at the leading edge, which may control the formation of stable microtubules via Dia1.
Cell Culture, Wounding, and Live Cell Recording-MEF cell lines were generated from wild type as well as from embryos double deficient for G␣ q /G␣ 11 and G␣ 12 /G␣ 13 as described (17) and maintained in Dulbecco's modified Eagle's medium supplemented with 10% FBS. For wound healing assays, cells were seeded in 6-well plates, grown until confluent, and wounding was performed with a 10-l tip that was cut longitudinally. Live cell recordings were performed immediately after wounding for 18 h at 37°C using a Leica Microscope CTR MIC. Pictures were acquired every 30 min with a motor-controlled Leica DC 350 FX camera using the Leica software FW4000, which enables simultaneous recordings from multiple wells. For statistical analysis, the wound distance from each well was measured in duplicates at 3 randomly defined wound gap locations per frame recorded per experiment, and at least three independent scratch-wound experiments were used for calculations. To rescue G␣ 12/13 deficiency 15 ϫ 10-cm dishes of G␣ 12/13 -deficient MEFs were transfected with pEF-GFP, pCis-G ␣12 , and pCis-G ␣13 using Transfectin. Cells were sorted for GFP expression 15 h after transfection with a FACS Vantage SE (BD Biosciences) in 0.2% FBS in PBS and were grown until confluent before wounding.
Immunofluorescence-Cells were fixed in 4% paraformaldehyde, permeabilized in 0.3% Triton X-100, and stained in PBS containing 5% FBS as described (22). For visualization of Glu-microtubules, cells were fixed at Ϫ20°C in methanol or in 8% formaldehyde. Images were obtained with a cooled CCD Leica DC 350 F camera on a Leica DMIRE 2 microscope and Leica IM 50 software and were imported in Adobe Photoshop. Fluorescence intensities were quantified and plotted using Scion Image (Scion Corp.).
GTPase Activity Assays and Western Analysis-Biochemical analysis of scratch-induced GTPase activities was performed by making multiple wounds with an eight-channel pipette scratched across a 10-cm dish (38). Lysates from wounded MEFs or unwounded MEFs (time point 0) were subjected to either RhoA or Ras pull-down assays using immobilized GST-rhotekin-RBD or GST-Raf1-RBD respectively, as described (14,39). Rac and Cdc42 activity assays were performed in parallel to Ras or RhoA pull-downs by adding 1 g/ml biotinylated CRIB peptide to the lysis reaction. CRIB peptide-bound GTP-Rac or GTP-Cdc42 was precipitated after addition of streptavidin-agarose (Sigma-Aldrich) as described (19). Proteins were eluted with SDS-sample buffer and analyzed by 15% SDS-PAGE.
GFP-GTPase-binding Domain (GBD) Probes and in Situ GTPase Affinity Assay-The plasmid encoding for GFP-RBD was generated using standard procedures by amplifying the cDNA of the RBD of mouse rhotekin encoding amino acids 7-89 using the following primers, forward 5Ј-tccggtaccatgggtatcctggaggacctcaatatg-3Ј and reverse 5Ј-agatctggatccgtcttattagcctgtcttctccagcac-3Ј, introducing an NcoI and BamHI site, respectively. GFP-WASp-GBD was generated by amplifying the cDNA of the GTPase-binding domain of mouse Wiskott-Aldrich-Syndrome protein (WASp) encoding amino acids 201-310 using the primers forward 5Ј-tccggtaccatgggtatccagaaccctgacatcacg-3Ј and reverse 5Ј-agatctggatccgtcttattactcctggcgcctcatctc-3Ј, introducing an NcoI and BamHI site, respectively.
The PCR products were ligated into the H10GFPspacer-MCS1 vector. Bacterial fusion proteins were expressed in the presence of 1 mM isopropyl 1-thio-␤-D-galactopyranoside at 30°C for 4 h in Luria-Bertani medium. His 6 -tagged fusion proteins were purified under native conditions according to the manufacturer (The QIAexpressionist TM , Qiagen) on nickel-nitrilotriacetic acid-agarose (Qiagen). Proteins were eluted from beads with four incubations for 15 min at 4°C using elution buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, and 250 mM imidazole). Fusion proteins were dialyzed overnight in dialyzing buffer (25 mM Tris, 10 mM MgCl 2 , 50 mM NaCl, 5% glycerol, 1 mM dithiothreitol). GFP-RBD purifications were tested for in vitro activity by recoupling an aliquot to nickel-beads for pull-down experiments. For in situ Rho[GTP] affinity assays cells grown on glass coverslips were fixed with 4% paraformaldehyde for 10 min and permeabilized in 0.3% Triton/PBS. After blocking (5% FBS, 3% bovine serum albumin, 0.1 M glycine in PBS, 45 min) cells were incubated for 2 h at 4°C with soluble GFP or GFP-RBD (0.02 g/l) in 5% FBS/PBS. Samples were washed and mounted using MOWIOL. Images were obtained by fluorescence microscopy using a Leica DMIRE 2 microscope and a CCD camera (Leica, Germany).
RNA Interference-mediated Knockdown of Mouse Dia1-siRNA corresponding specifically to mDia1 RNA sequences were obtained from Qiagen and chosen according to Arakawa et al. (40). Transfection of MEFs with siRNA duplexes was efficiently performed using magnetassisted transfection (MATra) as recommended by the manufacturer (IBA). Briefly, 2 l of MATra-A reagent containing magnetic nanoparticles was complexed with siRNA (20 M, 6 l) in 6-well plates, and magnet force transfection was performed for 15 min on a magnet plate at 37°C and 10% CO 2 . Cells were analyzed 48 -72 h after siRNA transfection. For complementation assays RNA interference-resistant mDia1, mDia1-compl, containing two silent mutations at nucleotide positions 10 and 13 in the siRNA target sequence (A711G and C714A, mDia1 cDNA), was generated by overlap PCR using the primers 5Ј-cagcaggatccattgccctgaccagcagtag-3Ј, 5Ј-ctactgctggtcagggcaatggatcctgctg-3Ј and 5Ј-aaggaaaatggcggccgctggagccgtccggcgggg-3Ј, 5Ј-cggccgctcgagttagcttgcacggccaaccagctc-3Ј. The full-length mDia1 complementation mutant was inserted via NotI and XhoI into the pcDNA3-Flag vector. Mutations were confirmed by DNA sequencing.
DNA Microinjections-siRNA-treated wild type or G␣ 12/13 -deficient MEF monolayers were wounded for 60 min before cells at the wound edge were microinjected into nuclei using an Eppendorf Femtojet and Injectman micromanipulator attached to a Zeiss Axiovert 135 microscope. Pressure was adjusted to 75 hPa for 0.5 s using an Eppendorf femtotip needle, and plasmids were microinjected at 0.05 g/l. Wounding induced migration was continued for additional 2 h before fixation and immunofluorescence analysis.

G␣ 12/13 -proteins Are Essential for Directed Cell Movement during
Wound Healing-To test whether G-protein-mediated signaling pathways are principally required for wound healing in the presence of serum we compared MEF lines generated from wild type embryos or from embryos double deficient for either G␣ q /G␣ 11 or G␣ 12 /G␣ 13 , or MEFs treated with pertussis toxin to inhibit the G i family of G-proteins (14,17). MEF migration was initiated in the presence of 10% serum in a monolayer using the scratch-induced wound healing assay, and multiple time lap recordings were performed in parallel for 18 h. MEFs migrated into the wound as sheets to reform a tight monolayer within 7-8 h (250 -300-m wound width) (Fig. 1, A and C and supplemental movie 1). The average velocity was 0.84 Ϯ 0.1 m/min. Preincubation of MEFs with 100 ng/ml pertussis toxin only moderately reduced wound healing, indicating that G␣ i -proteins are not critically involved in mesenchymal cell migration ( Fig. 1A) in contrast to their role in leukocyte motility (8). G␣ q/11 -deficient fibroblasts required longer time periods as compared with wild type MEFs but also efficiently reformed an intact monolayer ( Fig. 1A and supplemental movie 2). Surprisingly, G␣ 12/13deficient cells completely lacked the ability to close the scratch wound ( Fig. 1, A and C and supplemental movie 3). This defect was not due to impaired proliferation over the monitored time period (18 h) because G␣ 12/13 -deficient cells proliferated at a slightly faster rate as wild type MEFs, or G␣ q/11 -deficient MEFs judged by cell countings performed in parallel (data not shown). The wound healing defect was confirmed using two different G␣ 12/13 -deficient MEF clones to exclude clonal variation.
To verify that G␣ 12/13 -proteins are indeed required for wound healing in MEF monolayers G␣ 12/13 -deficient fibroblasts were transfected with plasmids encoding G␣ 12 , G␣ 13 , and GFP, sorted for GFP-positive cells (3-7% transfection efficiency), and replated for wounding experiments. Reexpression of G␣ 12 and G␣ 13 restored the ability of G␣ 12/13deficient MEFs to efficiently close the gap within 7-8 h ( Fig. 1 and supplemental movie 4), clearly demonstrating that G␣ 12/13 are essential for wounding induced cell migration. Expression of either G␣ 12 or G␣ 13 alone only partially rescued wound healing of G␣ 12/13 -deficient MEFs (data not shown).
Cytoskeletal Characteristics in Wounded G␣ 12/13 -deficient MEF Monolayers-Cell migration requires coordinated changes in cytoskeletal dynamics, which govern the protrusion of actin filament (F-actin) and microtubule-containing lamellipodia and filopodia at the front, retraction at the back, and constant assembly and disassembly of focal adhesions (18). Hence, we analyzed the morphology of focal adhesions, F-actin, and microtubules of G␣ 12/13 -deficient MEFs after wounding. Visualization of vinculin revealed that the general appearance of focal adhesions in G␣ 12/13 -deficient cells was comparable with those of wild type MEFs over the entire wound-healing period ( Fig. 2A). Comparison of F-actin between wild type and G␣ 12/13 -deficient cells showed only moderate differences; leading edge G␣ 12/13 -deficient MEFs appeared to have slightly lower F-actin contents, as judged by phalloidin staining ( Fig. 2A). Strikingly although and in contrast to wild type MEFs, which adopted an elongated shape with long microtubule-containing protrusions toward the wound, G␣ 12/13 -deficient MEFs displayed a microtubule meshwork that was not orientated and appeared disorganized ( Fig.  2A). This phenotype was observed throughout scratch-induced migration, demonstrating that microtubule morphology and rearrangement were altered in cells lacking G␣ 12/13 . FIGURE 1. G␣ 12/13 is essential for wounding-induced cell migration. A, shown are statistics from wound distances in scratch-wound assays from 30-min interval live cell recordings of wild type MEFs without (WT) or with pertussis toxin pretreatment (WTϩPTX), G␣ q/11 -deficient MEFs, G␣ 12/13deficient MEFs, and G␣ 12/13 -deficient MEFs rescued by cotransfection with GFP, G␣ 12 , and G␣ 13 (Rescue) as indicated. For G␣ q/11 -deficient MEFs and G␣ 12/13 -deficient MEFs data were acquired and summarized from two different clonal cell lines each. Wound distance data are shown from at least four independently performed experiments. In each setup, four movies of four different cell lines or cell treatments were recorded simultaneously. B, cell extracts from G␣ 12/13 -deficient MEFs (G 12/13 -def.) and G␣ 12/13 -deficient MEFs cotransfected with GFP, G␣ 12 , and G␣ 13 (rescue) were immunoblotted for the indicated proteins. C, representative phase contrast images from 18-h live cell recordings of wounded wild type MEFs (upper panel), G␣ 12/13 -deficient MEFs (middle panel), or G␣ 12/13 -deficient MEFs that were FACSsorted for GFP after transfection with plasmids encoding for GFP, G␣ 12 , and G␣ 13 (2-g each/ 10-cm dish, 15 dishes/FACS sorting) (lower panel), at 15 min, 5 h, and 10 h after wounding. FIGURE 2. Cytoskeletal characteristics and Rho-GTPase signaling during directed migration of G␣ 12/13 -deficient MEFs. A, scratch-induced migration was initiated in confluent monolayers of wild type (wt) or G␣ 12/13 -deficient (G12/13-def.) MEFs for 4 h prior fixation and permeabilization. Cells were stained either for focal adhesions using vinculin antibodies, for F-actin using phalloidin, or for visualization of microtubules using ␣-tubulin antibodies as indicated. The direction of wound is indicated by a dotted white line. Scale bar, 10 m. B-E, for GTPase activity lysates from scratch-wounded MEF monolayers were subjected to pull-down assays as described under "Experimental Procedures." Times after wounding before cell lysis are indicated. Cdc42 (B) or Rac (C) pull-down assays were performed simultaneously to either Ras (D) or RhoA (E) assays by adding biotinylated CRIB peptide to the lysis reaction. Densitometric quantifications of 3-5 independently performed GTPase pull-down assays Ϯ S.E. are shown together with representative Western blots using the indicated antibodies. DECEMBER 23, 2005 • VOLUME 280 • NUMBER 51

G␣ 12/13 -dependent Cell Migration by Rho and Dia1
Role of Rho GTPases in G␣ 12/13 -mediated Wound Healing-Several trimeric G-proteins are known to regulate the functions of Rho GTPases that control F-actin, microtubule, and focal adhesion dynamics. Therefore, we analyzed scratch-induced activation of Cdc42, Rac, Ras, and RhoA in G␣ 12/13 -deficient MEFs under the applied wound assay conditions. Using immobilized PAK-CRIB peptides as a probe (19), we found that the activities of Cdc42 and Rac were not significantly reduced during migration of wild type or G␣ 12/13 -deficient MEFs (Fig. 2, B and C). We also directly assessed the involvement of Ras as it cooperates with RhoA to control focal adhesion turnover (20). Ras became rapidly activated upon wounding induced migration in both wild type and G␣ 12/13 -deficient monolayers (Fig. 2D). These data show that wounding induced Cdc42, Rac, or Ras activation does not depend on G␣ 12/13 .
To determine total RhoA activity we performed biochemical pulldown experiments using the Rho-binding domain (RBD) of rhotekin as a probe (21). Interestingly, wounding of MEF monolayers caused a down-regulation of total RhoA activity (Fig. 2E), demonstrating that total RhoA activity is inversely regulated as compared with Cdc42, Rac, or Ras (Fig. 2). However, low levels of active RhoA appeared to be maintained during wound healing in wild type MEFs (Fig. 2E). In contrast, RhoA activity levels were found to be significantly reduced or in some cases even undetectable in G␣ 12/13 -deficient MEFs during wounding (Fig. 2E). These results showed that wild type MEFs maintained higher levels of total active RhoA during migration, when compared with MEFs lacking G␣ 12/13 .

Specificity of a Recombinant Rho[GTP] Affinity
Probe-To be able to investigate endogenous RhoA activity in more detail we performed a modified in situ Rho[GTP] affinity assay (15,16) for which we have generated a His-tag-purified GFP-RBD fusion protein as a probe in which the Rho-binding-domain of rhotekin was coupled to GFP and produced recombinantly. To characterize the binding properties of this fusion protein we performed in vitro pull-down assays by recoupling purified GFP-RBD to nickel beads. Starved cells were serum stimulated for 5 min and endogenous active RhoA was efficiently precipitated by GFP-RBD as compared with untreated cells (Fig. 3A), clearly demonstrating the specificity of the probe for GTP-RhoA in vitro.
We then studied the ability of the GFP-RBD to detect active RhoA in situ. Incubation of fixed and permeabilized cells with GFP alone as a control resulted in a background signal labeling the nuclei (Fig. 3B). To control for the specificity of GFP-RBD for in situ fluorescence applications we transfected fibroblasts either with dominant negative RhoA (N19RhoA) or with a constitutively active mutant for RhoA (V14RhoA). Cells were transfected at subconfluency and serum-starved in 0.5% FBS. As expected, cells that expressed V14RhoA displayed excessive stress fiber formation and cell body contraction (Fig. 3C). As shown in Fig. 3 cells expressing V14RhoA displayed a robust increase in signal intensity for GFP-RBD, whereas cells expressing dominant negative N19RhoA did not (Fig. 3C). In contrast, cells expressing active V14RhoA that were incubated with GFP alone displayed a nonspecific background signal labeling the nucleus as compared with non-transfected cells or cells that were maintained in 10% serum conditions (Fig. 3B). To further verify that the GFP-RBD probe only recognizes the GTP-bound conformation of RhoA, fibroblasts were cotransfected with V14RhoA and C3 exoenzyme to inactivate Rho proteins (22). This resulted in a reduction of the GFP-RBD signal almost to background detection levels as compared with starved, non-transfected cells (Fig. 3C). We also transfected cells with the active form of Cdc42, V12Cdc42. Cells expressing V12Cdc42 were not detected by GFP-RBD again demonstrating the specificity of this probe for active Rho (Fig. 3C). In contrast and as an additional control, overexpression of V12Cdc42 could be strongly visualized by the addition of purified GFP-WASpGBD, a probe which we have generated to detect active Cdc42 (Fig. 3C).
Absent Rho Activity at the Leading Edge of Wounded Fibroblasts Lacking G␣ 12/13 -Because G␣ 12/13 -deficient MEFs had reduced levels of Rho activity we decided to visualize the regulation of this GTPase in a spatial manner. Hence, to study the localization of endogenous RhoA activity during cell migration we performed in situ Rho[GTP] affinity assays on scratch-wounded MEFs. Using GFP-RBD as a probe Rho activation could be visualized in the cytoplasm and tail regions of wounded wild type MEFs with signal intensities declining toward the lamellipodial front (Fig. 4, A and D). Interestingly, a distinct signal could be detected at the leading edge and along F-actincontaining protrusions (Fig. 4, A, B, and D), indicating that active Rho is present in these structures. Rho activation at the leading edge was readily observed within the first hour after wounding and slowly declined over the next 2-3 h (Fig. 4E), showing that frontal Rho activity is temporally regulated during wounding induced MEF migration. In contrast, in wounded G␣ 12/13 -deficient fibroblasts overall Rho activity appeared to be lower, and no significant increase of active Rho could be observed at the cell front over wounding times (Fig. 4), suggesting a specific role for Rho activity at the leading edge of migrating fibroblasts. Thus, we identified G␣ 12/13 -proteins as essential components for localized RhoA activation at the leading edge during wounding induced cell migration.
G-protein-regulated Localization of Active Rho and Dia1 to the Leading Edge-Next, we asked if serum or LPA stimulation could activate Rho at the wound edge of MEFs. Monolayers of wild type MEFs were serum-starved in 0.5% FBS, wounded, and stimulated with 100 nM LPA for 1 h before fixation and incubation with GFP-RBD. Serumstarved cells showed low signal intensities for GFP-RBD, whereas LPA-treated MEFs displayed a general increase in GFP-RBD staining (Fig. 5A). Additionally, ϳ61% of LPA-treated cells exhibited active Rho at the leading edge as compared with ϳ19% cells maintained under serum-free conditions (Fig. 5, A and B). LPA-stimulated localization of Rho activity to the wound edge was almost completely blocked by preincubation with the cell-permeable Rho inhibitor TAT-C3 (15) (Fig. 5, A and B). TAT-C3-treated cells consistently showed almost undetectable levels of GFP-RBD in the cytoplasm and in most cases displayed strongly reduced F-actin staining characteristic for inactivation of RhoA (Fig. 5A).
The presence of Rho activity at the leading edge led us to investigate the involvement of Rho effectors in wounding induced cell migration. Inhibition of Rho-kinase (ROCK) has been previously reported not to interfere with cell motility in fibroblast scratch-wound assays (23). We therefore investigated whether Dia1 and active RhoA colocalize to the leading edge of wounded wild type MEFs using immunostaining. Endogenous Dia1 was distributed diffusely throughout the cytoplasm, and no localization of Dia1 at the wound edge could be observed in cells under serum-free conditions (Fig. 5A). In contrast, in MEFs treated with LPA we readily detected Dia1 staining at the leading edge along with active Rho (Fig. 5, A and C). Similar results were observed when cells were stimulated with 10% FBS (not shown).
We also compared endogenous Dia1 localization between wild type and G␣ 12/13 -deficient MEFs under normal wound assay conditions and found that Dia1 was mostly absent from the leading edge of migrating cells lacking G␣ 12/13 apart from a moderate increase to ϳ24% cells that showed leading edge Dia1 at 2 h postwounding (Fig. 5, D and E), a fact which is also reflected by the correlative increase of frontal GFP-RBD signals in G␣ 12/13 -deficient MEFs (Fig. 4E). We next microinjected con-stitutively active V14RhoA into wounded G␣ 12/13 -deficient MEFs to restore Dia1 recruitment to the leading edge, but, because of increased stress fiber formation and shape change of V14RhoA-expressing cells we were unable to detect any specific Dia1 signal at the cell periphery (data not shown). Our data demonstrate that localization of active Rho and Dia1 to the leading edge of wounded fibroblasts is regulated dynamically and depends on G␣ 12/13 . Dia1 Is Required for Polarized Cell Migration-So far our analysis revealed that wounded G␣ 12/13 -deficient MEFs had overall reduced RhoA activity and undetectable levels of active Rho at the leading edge in addition to altered microtubule morphology. Additionally, Dia1 localized to the cell front of migrating wild type MEFs but not in cells lacking G␣ 12/13 . We hence speculated whether Dia1 is required for polarized cell movement. Therefore, we knocked down Dia1 expression by RNA interference prior to induction of scratch-wound assays (Fig.  6A). Analysis of wild type MEFs after Dia1 knock-down showed that scratch-induced cell migration was efficiently blocked (supplemental movies 5 and 6, and Fig. 6B), whereas preincubation with 5 M Y27632 to inactivate ROCK only moderately reduced wound healing by ϳ1 h (Fig. 6B). These data clearly show that Dia1 plays an essential role in polarized cell migration.
G␣ 12/13 and Dia1 Are Essential for the Formation of Orientated, Stable Microtubules-It has recently been shown that transient expression of active mutants of RhoA and its effector diaphanous, but not active ROCK, induces the formation of orientated, stable microtubules (24). Orientated Glu-MTs represent a subset of microtubules -affinity precipitation from MEF lysates using purified His-GFP-RBD recoupled to nickel beads with or without serum (fetal calf serum (FCS)) stimulation. Total expression of RhoA is shown (lysate). B, scratch-wounded wild type MEFs (1 h) were fixed and permeabilized before incubation with purified GFP as a control showing some background nuclear labeling. C, fibroblasts were transfected with plasmids encoding myc-tagged constitutively active RhoA (V14RhoA), myc-tagged constitutively active RhoA (V14RhoA) and C3 transferase (0.1 g), myc-tagged dominant-negative RhoA (N19RhoA), or myc-tagged constitutively active Cdc42 (V12Cdc42) as indicated. Cells were maintained at 0.5% FBS. 24 h after transfection cells were fixed with 4% paraformaldehyde and permeabilized, and in situ GTPase assays were performed using 0.02 g/l GFP-RBD, GFP-WASpGBD, or GFP alone (green). Colabeling was done using antimyc-TRITC antibodies (red) and Alexaா 350 phalloidin (blue). Merged green, red, and blue channels are shown in the panel on the right. Scale bars, 10 m.
at the leading edge with a long half-life (Ͼ1 h) and are thought to be important for the onset of directed cell migration (24). Thus, we decided to analyze Glu-MT formation in G␣ 12/13 -deficient MEFs. Comparing scratched monolayers of wild type and G␣ 12/13 -deficient MEFs for Glu-MTs using a specific antibody (25), we found that wild type MEFs started to form orientated Glu-MTs within the first 2 h after wounding, whereas in contrast G␣ 12/13 -deficient MEFs largely failed to do so (Fig. 6, C and D). Both, orientation toward the wound edge as well as overall formation of polarized Glu-MTs appeared to be diminished.
Although expression of active mutants of diaphanous-related formins are sufficient to promote Glu-MT formation in fibroblasts (11) we tested whether Dia1 is in fact necessary for stable microtubule formation using RNA interference. RNA interference against the RhoA effector mDia1 efficiently and almost completely blocked the formation of orientated Glu-MTs in wounded wild type MEFs (Fig. 6, C and E). Reexpression of an siRNA target sequence resistant mutant mDia1 (Dia1-compl, see "Experimental Procedures") using microinjection of wound edge MEFs silenced for mDia1 restored Glu-MT formation in these cells (Fig. 6F). Thus, Dia1 appears to be essential for the generation of stable microtubules, clearly indicating that this is a critical function exerted by this formin-related protein.
We then analyzed whether Dia1 is sufficient for Glu-MT formation in G␣ 12/13 -deficient MEFs. Therefore, we generated an active Dia1 mutant lacking the C-terminal diaphanous autoregulatory domain (DAD), known to interact with regions in the N-terminal RhoA-binding domain to maintain a dormant protein conformation (26). The ⌬DAD mutant was microinjected into G␣ 12/13 -deficient MEFs at the wound edge. Strikingly, introduction of Dia1-⌬DAD into wound edge cells lacking G␣ 12/13 rescued the formation of Glu-MTs that were polarized toward the leading edge (Fig. 6G), suggesting that G␣ 12/13 functions upstream of Dia1. To confirm that G␣ 12/13 is essential for microtubule stabilization we transfected G␣ 12/13 -deficient MEFs with plasmids encoding G␣ 12 , G␣ 13 , and GFP, for FACS sorting. GFP-sorted MEFs were replated, wounded, and analyzed for Glu-MT formation. As shown in Fig. 6H reintroduction of G␣ 12 and G␣ 13 restored scratch-induced formation of orientated Glu-MTs in G␣ 12/13 -deficient MEFs. Taken together, these results show that G␣ 12/13 -proteins are critical components for microtubule stabilization in migrating cells.

DISCUSSION
In this paper, we have identified G␣ 12/13 subunits as essential components for directed cell migration. We further showed that global RhoA activity becomes down-regulated during initiation of migration but that low levels may be maintained to ensure formation of Glu-MTs and hence efficient wound healing. Endogenous active Rho can be detected at the cell front of motile fibroblasts, and this depends on functional G␣ 12/13 -proteins. In parallel, the RhoA effector mDia1 is recruited to the wound edge and is required for directed cell movement during wound repair. G␣ 12/13 -proteins have been suggested to be able to interact with the family of RGS domain-containing RhoGEFs thereby transducing signals toward the activity of the small GTPase RhoA (27,28). Our results show that under normal serum conditions wild type MEFs maintained higher levels of overall RhoA activity during migration, when compared with MEFs lacking G␣ 12/13 , indicating an impaired responsiveness of G␣ 12/13deficient cells toward serum factors and/or agonists produced in response to wounding. It shall be interesting to elucidate whether certain RGS domain-containing RhoGEFs play a specific role in directed cell movement.
Our data reveal that total RhoA activity becomes down-regulated in fibroblasts in response to wounding thereby providing direct evidence for the longstanding hypothesis that overall RhoA activity must be sup-pressed during cell movement, possibly by specific RhoGAPs (29). Previous investigations demonstrated that constitutively elevated levels of active RhoA negatively modulate cell migration because of excessive formation of stress fibers and adhesion forces (30,31). However, a low level of total RhoA activity is clearly detectable throughout cell migration in agreement with the idea that RhoA activity is responsible for efficient rear end retraction (3).
We generated a probe to investigate spatial Rho activity in migrating cells. Indeed and as may be expected, endogenous Rho activity was present at the rear of polarized wild type MEFs, whereas it was reduced at the rear of migrating G␣ 12/13 -deficient MEFs, and these findings support the current model that RhoA and ROCK are locally involved in tail retraction (13), probably adding to the overall phenotype during G␣ 12/13 -deficient wound healing. However, random cell movement of G␣ 12/13 -deficient MEFs was comparable with wild type cells. 4 4 P. Goulimari and R. Grosse, unpublished observations. . D, wild type (wt) or G␣ 12/13 -deficient MEFs (G12/13-def.) were wounded and subjected for detection of microtubules using ␣-tubulin or of stable microtubules using Glu-MT antibodies as indicated. E, wild type MEFs were transfected with mock siRNA as a control (siRNA mock) or siRNA directed against mDia1 (siRNA Dia1). MEF monolayers were scratch-wounded for 2 h, fixed, and subsequently analyzed for microtubules (␣-tubulin) and for Glu-MTs as indicated. Merged images showing microtubules in green and Glu-MTs in red are indicated in the right panels. Shown are wound edges in 40ϫ magnification. F, wild type MEFs were transfected with siRNA directed against mDia1 (siRNA Dia1), and monolayers were scratchwounded for 60 min before microinjecting Flag-mDia1-complementation expression plasmids (Dia1-compl) into wound edge cells. 2 h later cells were stained for F-actin using Alexaா 350 phalloidin (blue), for mDia1-compl using Flag-M2 antibodies (green), and for Glu-MTs (red) as indicated. Merged green, red, and blue channels are shown in the panel on the right. G, G␣ 12/13 -deficient MEF monolayers were scratch-wounded for 60 min before microinjecting Flag-mDia1-⌬DAD expression plasmids (⌬DAD) into wound edge cells. 2 h later cells were stained for F-actin using Alexaா 350 phalloidin (blue), for mDia1-⌬DAD using Flag-M2 antibodies (green), and for Glu-MTs (red) as indicated. Merged green, red, and blue channels are shown in the panel on the right. H, G␣ 12/13 -deficient MEFs were FACS-sorted for GFP after transfection with plasmids (2-g each/10-cm dish; 10 dishes/FACS sorting) encoding for GFP alone or for GFP, G␣ 12 , and G␣ 13 as indicated. After reaching confluence MEF monolayers were wounded for 2 h prior to fixation and Glu-MT staining. The percentage of cells at wound margins containing orientated Glu-MTs were determined (n ϭ 3, mean Ϯ S.E.). Scale bars, 10 m.
In addition and to our surprise, we observed spatial Rho activity at the leading edge of motile cells, which depended on G␣ 12/13 -proteins. This strongly argues that Rho is involved in additional functions that regulate cell movement apart from cell tail retraction. Interestingly, Wang et al. (32) recently demonstrated that RhoA becomes degraded in filopodia of migrating cells by the E3 enzyme Smurf1, consistent with the previous hypothesis that Rho activity at the leading edge would rather interfere with cell migration (3). Nevertheless, this could also indicate that Rho function is tightly controlled in a spatiotemporal manner and constantly turned over at these structures to fine tune protrusive and adhesive activity at the front of migrating cells. We observed that endogenous active Rho clearly appears temporally at the leading edge of migrating cells within 1 h postwounding. This localization is not only dependent on G␣ 12/13 but can be stimulated by serum or LPA demonstrating a dynamic subcellular regulation of active Rho and its effector Dia1 by extracellular signals. Endogenous active Rho and Dia1 colocalize at the leading edge of migrating cells. We do not know whether Rho becomes locally activated or whether active Rho is subsequently recruited to the wound edge. Another persuasive idea is that upon local activation of RhoA, Dia1 may become localized to the leading edge of polarized cells and the fact that Dia1 is not efficiently localized to the cell front of G␣ 12/13 -deficient MEFs supports this hypothesis. By using GFP-RBD in our assay system we are not able to visualize real time dynamics of wound edge Rho activity or to distinguish between the three different Rho isoforms (RhoA, RhoB, RhoC). It is certainly an interesting question to pursue whether specific Rho isoforms become activated at different subcellular locations such as at the leading edge. Further studies are necessary to clarify this matter.
We found that microtubule morphology was altered in migrating G␣ 12/13 -deficient MEFs. The reduced RhoA activity in wounded G␣ 12/13deficient MEFs is likely to account for this, because the regulation of other investigated small GTPases, i.e. Ras, Rac, or Cdc42 during wound healing was essentially unaffected. In line with these observations wounded G␣ 12/13 -deficient MEFs failed to establish polarized, stable detyrosinated microtubules suggesting that heterotrimeric G␣ 12/13 -proteins are upstream of Rho-Dia1-dependent Glu-MT formation. Thus, G␣ 12/13 -proteins are essential components to ensure the formation and polarization of Glu-MTs. The inability of G␣ 12/13 -deficient MEFs to form orientated, stable microtubules likely represents an early and critical step underlying the wound-healing defect observed in the absence of G␣ 12/13 . However, additional RhoA-dependent mechanisms may be contributing to the overall phenotype. Although total RhoA activity is down-regulated during fibroblast migration we speculate that a subpopulation of active Rho is localized to the cell front to control Dia1-dependent Glu-MT formation. Also, because overall high levels of active RhoA favor signaling through ROCK low levels of total active RhoA may favor signaling through Dia (33), resulting in stable microtubule formation.
While this manuscript was in preparation Matsuda and colleagues using FRET imaging technique could show that cells transiently transfected with a YFP-RBD-RhoA-CFP fusion protein (Raichu-RhoA), displayed activation of this construct in membrane ruffles, indicating that Rho activity indeed plays a role at the leading edge of motile cells (34), although this technique does not allow for the study of endogenous Rho proteins.
In summary we found that G␣ 12/13 subunits are critically involved in wounding induced, directed cell movement. One may speculate that cells respond to wounding and/or changes in cell-to-cell contacts by local production of G-protein-coupled receptor agonists such as LPA (35), which act in an auto-/paracrine fashion to promote G␣ 12/13 -dependent local RhoA activation at the wound edge. Our results suggest that G␣ 12/13 functions upstream of Dia1 to control microtubule stabilization and further identified Dia1 to be necessary for polarized cell migration. Dia1 may have several roles in cell migration. Dia1 nucleates actin filaments and was shown to interact with profilin, VASP, and other cell cytoskeleton-regulating proteins (22,26,36,37). We found that Dia1 is indispensable for microtubule stabilization and cell migration, indicating that this is an important physiological function of the protein. In conclusion, our results clearly identify G␣ 12/13 as essential components to promote localized Rho activity at the leading edge that is likely to control Dia1 function required for microtubule stabilization and directed cell migration.