Roles for Rho/ROCK and Vinculin in Parietal Endoderm Migration

Abstract

The first cell migration event in the mouse embryo is the movement of parietal endoderm cells from the surface of the inner cell mass facing the blastocoel cavity to line the inner surface of the trophectoderm. F9 embryoid bodies provide an in vitro model for this event. They have an inner core of undifferentiated stem cells surrounded by an outer visceral endoderm layer. When plated on a laminin coated substrate, visceral endoderm transitions to parietal endoderm and migrates onto the dish, away from the attached embryoid body. We now show that this outgrowth contains abundant focal complexes and focal adhesions, as well as lamellipodia and filopodia. Treatment with the ROCK inhibitor Y-27632 promotes a 2-fold increase in outgrowth, and a transition from focal adhesions and associated stress fibers, to focal complexes and a decrease in stress fibers. ROCK inhibition also leads to an increase in lamellipodia. Inhibition of RhoA by transfection of a vector encoding C3 transferase, direct administration of the C3 enzyme, or transfection of a vector encoding p190 Rho GTPase Activating Protein also promotes outgrowth and an apparent transition from focal adhesions to focal complexes. Parietal endoderm outgrowth generated using vinculin-deficient F9 stem cells migrates 2-fold further than wild type cultures, but this outgrowth retains the morphology of wild type parietal endoderm, including focal adhesions and stress fibers. Addition of Y-27632 to vinculin-null outgrowth cultures further stimulates migration an additional 2-fold, supporting the conclusion that Rho/ROCK and vinculin regulate parietal endoderm outgrowth by distinct pathways.

Keywords:

• Rho
• ROCK
• vinculin
• cell migration
• parietal endoderm

INTRODUCTION

Cell migration is a multi-step process that involves lamellipodial extension at the leading edge, adhesion to an underlying substrate, and contraction at the rear. This cycle includes the turnover of adhesion sites, formed at the front and broken at the rear, to facilitate forward movement [1, 2, 3, 4]. Initial adhesions contain actin and talin, and develop into focal complexes, short aggregates localized just behind the lamellipodial leading edge and marked by the recruitment of vinculin [5, 6]. Focal complexes are transient structures that contain a variety of structural and signaling proteins including integrins, vinculin, talin, paxillin, and focal adhesion kinase (FAK), and may mature into the larger more stable focal adhesions [7], which include the focal complex components and additional proteins, such as zyxin [7, 8, 9, 10].

Vinculin plays a key role in cell migration. It contains binding sites for focal adhesion proteins talin and α-actinin at its amino-terminus [11, 12, 13], and paxillin and actin binding sites at its carboxy-terminus [14, 15, 16, 17]. It can interact directly with the Arp2/3 complex, providing a link between integrin and the adhesion complex to the machinery that promotes lamellipodial extension [18, 19].

Cell migration is also regulated by Rho family GTPases. In broad strokes, Rac and Cdc42 regulate the formation of smaller focal complexes associated with protruding lamellipodia and filopodia respectively, and Rho regulates the formation of focal adhesions and associated actin stress fibers, as well as the contraction of the rear of the cell [20]. Subcellular localization, of activators and inhibitors of these GTPases, for example to the leading lamellipodial edge, contributes to the cell's migratory state [21].

RhoA is the best studied Rho family member. As with all small GTPases, RhoA is activated by guanine nucleotide exchange factors (GEFs) that exchange GTP for GDP, and inhibited by GTPase activating proteins (GAPs) that promote hydrolysis of bound GTP to GDP. p190-RhoGAP regulates a variety of actin-dependent events, including axon branch stability [22] and cell migration [23]. Rho activity can also be inhibited experimentally by the addition of C3 transferase, which ADP-ribosylates and inactivates Rho [24, 25, 26].

In the context of focal adhesion formation and cell migration, Rho acts primarily through its downstream effector the serine/threonine kinase ROCK (Rho-associated kinase) that has two isoforms, ROKα /ROCK-II [27] and ROKβ /ROCK-1/p160ROCK [28]. The C-terminal portion of the central coiled coil domain contains the Rho-binding region that specifically interacts with Rho-GTP, which activates the kinase. ROCK phosphorylates two distinct targets that both, indirectly and directly, modulate myosin-II regulatory light chain (MLC) phosphorylation. The first target, myosin light chain phosphatase, is phosphorylated on its myosin binding subunit at a threonine (Thr-697) and serine residue (ser-854), thereby inactivating the phosphatase and keeping MLC phosphorylated and active [29, 30]. ROCK directly phosphorylates its second target, MLC, at serine-19 and threonine-18, leading to actin-dependent myosin ATPase activity [31, 32, 33]. Constitutively active forms of Rho or ROCK enhance MLC phosphorylation and promote formation of stress fibers and focal adhesions [34, 35]. ROCK has other targets that effect cell migration as well, including members of the ERM family of proteins (ezrin, radixin, and moesin) [36] and cofilin [37]. Cofilin plays a key role in actin assembly at the lamellipodial leading edge [38]. In addition, cofilin-mediated depolymerization and severing of F-actin can regulate tail retraction and limit membrane protrusions in leukocytes [37]. Cofilin activity is inhibited by phosphorylation at ser-3 by LIM-kinases 1 and 2, which are substrates of ROCK [39, 40]. Thus ROCK promotes phosphorylation and inactivation of cofilin by activating LIM-kinases.

Y-27632 is a cell-permeable compound that inhibits ROCK selectively, 20–30 times more potently than other Rho-dependent kinases [41]. Y-27632 contains a pyridine moiety and competes with ATP at the catalytic domain of ROCK [42]. Y-27632 treatment decreases levels of MLC phosphorylation, but the ultimate effect on cellular behavior differs in a cell-type specific manner. In cells that have stress fibers and associated focal adhesions, such as fibroblasts, ROCK inhibition increases cell migration, at least in part, by inhibiting cell substrate adhesion [43]. Fibroblasts treated with Y-27632 show significantly enhanced motility, along with increased migratory structures such as lamellipodia and filopodia [43]. In other less adherent cell types that lack focal adhesions, such as macrophages and neutrophils, inhibition of the Rho pathway inhibits cell migration, likely due to inhibition of contraction at the cell rear necessary for migration. In migratory leukocytes, Y-27632 induced ROCK inhibition leads to increased membrane protrusions that interfere with productive migration [37]. These studies demonstrate that the effect of ROCK inhibition on cell migration may depend on how polarized and adherent the cells are.

F9 teratocarcinoma stem cells provide a unique system for studying the migration of parietal endoderm (PE), the first motile cell type in the mouse embryo. F9 cells can differentiate directly into PE if cultured in monolayer and treated with retinoic acid (RA) and cAMP [44]. During the stem cell to PE transition, β1 integrin-containing focal adhesions form [45]. When cultured in suspension and treated with RA alone, F9 cells form embryoid bodies, consisting of an undifferentiated stem cell core with an outer layer of visceral endoderm (VE) [46]. When these embryoid bodies are plated on an extra-cellular matrix (ECM) coated substrate, VE transitions to PE when it contacts the substrate and migrates away from the embryoid body, mimicking the earliest migration event in the mouse embryo, the migration of PE along the inner surface of the trophectoderm [47, 48, 49].

To evaluate the role of vinculin in cell attachment and migration, vinculin deficient F9 stem cell lines were created by gene targeting [50]. These vinculin -/- F9 stem cells have a more rounded morphology and exhibit significantly decreased adhesive abilities, although they are still able to form focal adhesions that contain α -actinin, talin and paxillin [51]. The vinculin-deficient F9 stem cells migrate 2.4 times faster than wild type F9 cells, and recent experiments suggest that this is due to an increased interaction between FAK and paxillin in the absence of vinculin [52]. The ability of the FAK-paxillin interaction to promote migration is dependent upon ERK activation [53, 54]. The vinculin null cells retain their ability to differentiate into PE, which also exhibit increased migratory abilities [50]. We further investigate the migration of vinculin-null PE here.

We have begun to characterize the migration of F9-derived PE, using the embryoid body outgrowth system. Prior studies indicated that outgrowth on laminin substrates is mediated predominantly by the α6β1 integrin heterodimer [55]. We now show that these outgrowths contain abundant focal adhesions and focal complexes. Treatment with the ROCK inhibitor Y-27362 inhibits phosphorylation of myosin light chain phosphatase and cofilin, and promotes PE outgrowth. Untreated outgrowth contains abundant focal adhesions associated with actin stress fibers, whereas treated outgrowth shows decreased levels of focal adhesions and increased focal complexes, and a reduction in filamentous actin. We also inhibited RhoA directly upstream of ROCK using C3-transferase, which ADP-ribosylates and thereby inhibits Rho signaling, by transient transfection of a C3-transferase-encoding vector, or addition of the C3-transferase enzyme directly to lipofectamine permeabilized outgrowth cells. Rho activity was also inhibited by transient transfection of a p190-RhoGAP-encoding construct. These conditions also promote PE outgrowth and a loss of focal adhesions in both F9 and vinculin-deficient F9 cultures. Outgrowth derived from vinculin-null F9 cells also shows more extensive migration than wild type cultures, and Y-27632 treatment increases outgrowth another two-fold. These data support a role for Rho-ROCK and vinculin in inhibiting the migration of PE cells.

METHODS

Cell Culture

Wild type F9 teratocarcinoma cells were grown on tissue culture dishes coated with gelatin in a modified DMEM media with 10% calf serum (Hyclone) supplemented with pen/strep and glutamine (Gibco). The vin -/- F9 cells, the generous gift of Eileen Adamson, were grown on tissue culture dishes coated with gelatin in a modified DMEM media with 5% hyclone calf serum and 5% fetal calf serum (Atlanta Biologicals), supplemented with pen/strep and glutamine (Gibco). To form embryoid bodies, wild type F9 or vin-/- F9 cells were plated in suspension culture and treated with 7.5× 10⁻⁷ M RA for 7–8 days, until an outer layer of visceral endoderm formed. To obtain a monolayer culture of PE, F9 stem cells were plated on gelatin at a low density and treated with RA and .125 mg/ml cAMP for five days.

Antibodies, ECM Proteins and Reagents

The mouse anti-vinculin antibody was obtained from Sigma. Mouse anti-human α-actinin monoclonal antibody was obtained from Chemicon International. Polyclonal rabbit anti-paxillin pY31 phosphospecific antibody was obtained from Biosource International. The monoclonal anti-FAK antibody and the anti-phosphotyrosine antibody (clone PY20) were obtained from Transduction Laboratories. Polyclonal anti-phospho-Histone H3 (mitosis marker) was obtained from Upstate Biotechnology. Hoechst nuclear marker was obtained from Molecular Probes. Troma-1 is a monoclonal antibody developed by investigators and was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. Polyclonal rabbit anti-Myosin Phosphatase antibody was obtained from Covance. For immunofluorescence, FITC-conjugated anti-mouse IgG was obtained from Cappel. Rhodamine-conjugated goat anti-rabbit IgG was obtained from Alexa Molecular Probes Inc. Goat anti-rabbit Alexa 488 conjugated anti-GFP was obtained from Molecular Probes. Alexa 647 CY5 (infrared)- conjugated goat anti-mouse IgG was obtained from Molecular Probes. Laminin-1 was obtained from Sigma. Goat anti-rabbit IgG HRP-conjugated secondary antibody was obtained from Sigma. Rhodamine-conjugated phalloidin was obtained from molecular probes. ROCK inhibitor Y-27632 and Rho inhibitor Exoenzyme C3 were obtained from Calbiochem Inc. For transient transfections, Lipofectamine 2000, Opti-MEM media, and the empty vector pcDNA3.1/Zeo were obtained from Invitrogen. The empty vector pEGFP-N1 was obtained from Clontech.

Immunofluorescence

After allowing PE to migrate on laminin coated coverslips for 48–72 h, the cells were fixed for 12 min in 3.7% formaldehyde and then washed twice with PBS and permeabilized with 0.5% Triton-X 100. Cells were then washed twice in PBS and blocked for 30–45 min at 37°C in PBS supplemented with 1–2% BSA. Cells were then washed with PBS/BSA and incubated for 1 h at 37°C with primary antibodies at the following dilutions: 1:50 mouse anti-vinculin and mouse anti-FAK; 1:100 rabbit anti-paxillin, mouse anti-phosphotyrosine (clone PY20), mouse anti-α -actinin, rabbit anti-phospho-histone H3. Troma-1 supernatant was used at a 1:10 dilution. Alexa 488 conjugated anti-GFP was used at a 1:2000 dilution and incubated overnight at 4°C. Cells were then washed five times in PBS/BSA for five min. Cells were then incubated with secondary antibodies for 1 h at room temperature at the following concentrations: FITC conjugated goat anti-mouse IgG (1:50), alexa-fluor Rhodamine conjugated goat anti-rabbit IgG (1:1,000), or Alexa 647 CY5 goat anti-mouse IgG (1:1,000). Cells were then subjected to five, five min washes with PBS. Cells were then mounted and examined using a Nikon fluorescence microscope, with images captured using a SPOT digital camera and Adobe Photoshop software.

When rhodamine conjugated phalloidin was used in conjunction with antibodies, it was added after three washes following incubation with secondary antibodies at a dilution of 1:100. After a 30-min incubation at room temperature, three more five minute washes were performed, and cells were mounted and viewed as described previously.

When Hoechst nuclear staining was performed, it was added at a 1:10,000 dilution after three washes following incubation with secondary antibodies, or rhodamine conjugated phalloidin. After a 10-min incubation at room temperaure, three more five minute washes were performed, and cells were mounted and viewed as described previously.

Parietal Endoderm Outgrowth Migration Assay and Measurement of Lamellipodia

After 7–8 days in suspension culture treated with retinoic acid, embryoid bodies were plated on laminin coated glass coverslips (30 μg/ml, incubated for 2–3 h at 37°C). ROCK inhibitor Y-27632 was added at a concentration of 10 μg/ml to half of the dishes. 24 h after plating, phase contrast pictures were taken of a minimum of five embryoid bodies and their associated PE outgrowth. The distance from the center of the embryoid body to the outer edge of its outgrowth was measured in eight different directions. At these same points, the distance from the center of the embryoid body to the outside of the embryoid body itself was measured. These two values were subtracted, providing the distance the PE outgrowth migrated from the edge of the embryoid body. These distances were averaged for five embryoid bodies under each condition (F9 untreated, F9 treated with 10 μg/ml Y-27632, F9 vin-/- untreated, and F9 vin-/- treated with 10 μ g/ml Y-27632) and used as one data point. The experiments were repeated in triplicate. To quantify numbers of the lamellipodia or filopodia under the different conditions, the number of lamellipodia or filopodia per cell in the outermost ring of the outgrowth was determined and then divided by the total cell number in the outer ring of each outgrowth. For lamellipodia, data shown are the average from the outgrowth of five embryoid bodies. Lamellipodia are identified in phalloidin-stained cells as broad protrusions and filopodia as thin extensions containing a single fiber of actin. The cell number was determined by Hoechst staining.

Western Blotting

Monolayer-derived cells were used for this analysis. Some of these cultures were treated with 10 μ g/ml Y-27632. Cells were scraped in ice cold PBS supplemented with a protease inhibitor cocktail (1 μg/ml/leupeptin, 1 μ g/ml pepstatin, 1 μ g/ml aprotinin, 1 μg/ml benzamidine) and 2 mM PMSF. Stem cells and monolayer PE were also scraped under the same conditions. The resultant cell suspensions were centrifuged at 4,000 rpm for 3 min. The cell pellet was then lysed in buffer containing 50 mM Tris pH 7.4, 150 mM NaCl, 1 mMCaCl₂ 0.1% SDS, 0.5% sodium deoxycholate, 1% NP-40, and 2 mM PMSF in the presence of the protease inhibitor cocktail (see above), for 30 min on ice. Lysates were precleared by centrifugation at 13,000 rpm for five minutes. A Biorad (Bradford) assay was then performed to normalize protein concentrations of the lysates. Samples were solubilized in 2X Laemmli buffer, boiled for five minutes, and the proteins were resolved by SDS-PAGE. Proteins resolved by SDS-PAGE were electrophoretically transferred to an Immobilon-P membrane (Millipore) and the membrane was blocked for one h at room temperature with TBS containing 0.3% Tween-20 (TBST) and 3% dry milk. The blots were incubated overnight rocking at 4°C with either rabbit anti-myosin light chain phosphatase two h at room temperature with rabbit anti-phosphorylated myosin light chain phosphatase (1:1,000), rabbit anti-myosin light chain phosphatase (1:1,000), or mouse anti-human α -actinin (1:5,000) diluted in TBST with 3% milk. The blots were then washed five times in TBST, and incubated an additional hour with HRP-conjugated secondary antibodies. After five washes in TBST the proteins were visualized with chemiluminesence.

Lipofectamine-Mediated Introduction of C3 Transferase Construct and Enzyme and Introduction of p190 Rho GAP Construct

Day 7 embryoid bodies were plated on laminin coated glass coverslips (30 μ g/ml laminin for three h at 37°C) in 1 ml of Opti-MEM media for 24 h. For C3 transferase, 1.6 μ g of the pcDNA3.1/zeo empty vector and the C3 transferase-encoding vector (generous gift of Rebecca Worthylake, Burridge laboratory) were incubated with 4 μ l of Lipofectamine and 200 μ l Opti-MEM for 20 min at room temperature (Invitrogen's protocol). 800 μ l of Opti-MEM was added to the transfection reagents to bring the total volume to 1 ml. For the exoenzyme C3 transferase, 2.5 μ g of the purified enzyme was incubated with 5 μ l of Lipofectamine 2000 [24] and 200 μ l Opti-MEM for 20 min at room temperature. 800 μl of Opti-MEM was added to bring the total volume to 1 ml. The old media was aspirated off the F9 or F9 vin^−/− outgrowth, and transfection reagents were added. All controls were incubated in Opti-MEM media. 24 h post transfection, the Opti-MEM media was replaced with modified DMEM appropriate for the cell line. The same protocol was followed for transient transfection of the p190 Rho GAP encoding vector [23], except the empty vector control was pEGFP-N1. Transfection with pEGFP-N1 resulted in a 25–30% transfection efficiency after 48 h, based upon the percent Hoechst positive cells expressing high levels of GFP, but faint fluorescence could be observed in almost all outgrowth cells. 48 h post transfection, phase contrast photos were taken (a minimum of five, with an average of seven). Outgrowth distance from the phase contrast pictures were measured in eight directions using Adobe Photoshop as described more in detail in Parietal Endoderm Outgrowth Migration Assay. All experiments were repeated in triplicate per treatment per cell type.

RESULTS

Adhesions and Lamellipodia in PE Outgrowth

In previous studies we characterized the emergence of robust focal adhesions upon differentiation of F9 stem cells in monolayer culture into PE [45]. We also demonstrated that visceral endoderm cells on the surface of F9-derived embryoid bodies produce migratory PE when plated on an adhesive substrate [48]. To limit the number of integrin interactions that promote PE outgrowth, laminin was used as the ECM substrate based on prior studies that demonstrated reliance solely upon α 6β 1 for migration on this substrate [55]. A phase contrast image of PE outgrowth is shown in Figure 1A. The PE outgrowth cells contain extensive focal adhesions, visualized by the expression of vinculin (Figure 1B), traditionally used as a focal adhesion marker [56]. In the embryoid body shown, PE outgrowth containing vinculin staining focal adhesions is restricted to the right side of the embryoid body, a pattern of outgrowth sometimes observed perhaps due to incomplete visceral endoderm differentiation surrounding the embryoid body.

Figure 1 PE outgrowth contains focal adhesions, focal complexes, filopodia, and lamellipodia. PE outgrowth was observed 24 hr after embryoid body attachment. Phase contrast image of outgrowth associated with an embryoid body (A). Reconstruction of confocal images of vinculin-containing focal adhesions in entire PE outgrowth associated with a single embryoid body (B). Localization of vinculin, α -actinin, FAK, or paxillin to focal adhesions and focal complexes (C, D, E, and F). Double labeling for PY20 and paxillin in apparent focal complexes just behind the lamellipodial leading edge (G, H and I). J and K (blow-up of boxed region in J) show α -actinin distribution at lamellipodial leading edge (red arrow head), focal complexes (white arrow), focal adhesions (white arrow head), or filopodia (red arrow).

In addition to vinculin, which is present predominantly in large focal adhesions (Figures 1B and C), PE contains α -actinin, associated with a variety of adhesion and migratory structures, including focal adhesions, focal complexes (located just behind the leading edge of lamellipodia), the outer lamellipodial edge, and filopodial extensions (Figures 1D, J, and K). These observations are consistent with other studies examining α -actinin's diverse subcellular localization [57, 58, 59]. FAK is found in large focal adhesions (Figure 1E), whereas a phosphospecific form of paxillin localizes to both large focal adhesions and smaller, apparent focal complexes (Figure 1F). Phosphotyrosine expression, examined using the PY20 antibody, colocalizes with paxillin, in focal complexes just behind the lamellipodial leading edge (Figures 1 G–I), suggesting that active phosphorylation events occur in these adhesive sites. These data demonstrate that PE outgrowth is rich in focal adhesions and focal complexes that contain structural and signaling components.

Effects of ROCK and Rho Inhibition

To determine the role of Rho-ROCK signaling in the migration of PE, we treated PE outgrowths with the ROCK inhibitor Y-27632. At the concentration of inhibitor used (10 μ g/ml), PE monolayers showed decreased levels of phosphorylated myosin light chain phosphatase, a downstream target of ROCK, relative to untreated cultures (Figure 2A), demonstrating the effectiveness of Y-27632 treatment. In addition, cofilin, an indirect downstream target of ROCK, showed decreased phosphorylation levels following Y-27632 treatment (data not shown). Treatment of PE outgrowths with Y-27632 resulted in a dramatic increase in the extent of PE migration (Figures 2B and C). Quantifying migration distances from three separate experiments revealed that the Y-27632 treated outgrowths migrated 2.4 times farther than control outgrowths in the same time period (Figure 2D). These measurements were taken 24 h after plating embryoid bodies on laminin, but the increase in migration could be seen at up to 72 h after plating (data not shown). Outgrowth in Y-27632-treated cultures displayed an increase in the number of lamellipodia per cell in the outer ring of the outgrowth (Figure 2E). Filopodia also appear more numerous under these conditions.

Figure 2 ROCK inhibition decreases PE outgrowth. A, PE monolayers treated with the ROCK inhibitor Y-27632 have decreased levels of phosphorylated myosin light chain phosphatase (A). Phase contrast images of PE outgrowth after 24 hr untreated (B) of treated with Y-27632 (C). Outgrowth distance is approximately 2-fold higher with Y-27632 treatment (D). The number of lamellipodia per outer ring PE outgrowth cell is increased somewhat by Y-27632 treatment, particularly at 24 hr (E). Vinculin and α -actinin localization in untreated and treated outgrowth cultures suggests a transition from focal adhesions to focal complexes (F-I). Y-27632 treatment disrupts actin stress fibers (J, K).

The increased outgrowth observed following Y-27632 treatment is not attributable to higher levels of proliferation since the percent of phosphohistone H3 labeled cells, a measure of cells in M phase, is similar under both conditions, 3.85 % in untreated and 3.67 % in Y-27632-treated outgrowths. These data, demonstrating a dramatic increase in the extent of PE migration with ROCK inhibition, are consistent with previous studies examining the effects of ROCK inhibition on fibroblasts [43].

We examined the adhesion and migratory structures in the Y-27632-treated PE outgrowth. Vinculin and α -actinin localized predominantly to large focal adhesions in untreated outgrowth (Figures 2F and H), whereas in Y-27632 treated outgrowth, vinculin and α -actinin were found in smaller, apparent focal complexes (Figures 2G and I). Paxillin and PY20 also localized predominantly to smaller, apparent focal complexes in Y-27632 treated outgrowths (data not shown). These observations indicate that Y-27632 treated outgrowths have reduced levels of large, stable focal adhesions as a consequence of ROCK inhibition. In addition, untreated outgrowth contains abundant focal adhesions associated with phalloidin-labeled stress fibers (Figure 2J), whereas treatment with Y-27632 results in the dissolution of these stress fibers and associated focal adhesions and an apparent increase in focal complexes (Figure 2K).

Effects of C3 Transferase and p190RhoGAP

To verify the role of Rho-ROCK in PE migration, we inhibited RhoA directly, upstream of ROCK, using C3-transferase, which ADP-ribosylates and thereby inhibits Rho signaling. C3-transferase activity was delivered to outgrowth cells by two routes; transient transfection of C3-transferase-encoding vector, or addition of the C3-transferase enzyme [24] directly to lipofectamine permeabilized outgrowth cells. Either of these conditions resulted in an increase in the distance migrated, 48 h post transfection or enzyme treatment, over untreated cultures or those incubated with an empty vector (Figure 3A). Examination of the outgrowth exposed to C3-transferase indicated a breakdown in actin stress fibers and a transition from focal adhesions to focal complexes (Figure 3B), as observed with Y-27632 treatment. There was also an increase in filopodia following addition of C3 transferase (Figure 3B).

Figure 3 C3 transferase and p190 Rho GAP promote PE outgrowth. The extent of outgrowth was promoted when PE outgrowth was transfected with a vector encoding C3 transferase, or by addition of the C3 transferase enzyme to lipofectamine permeabilized cells (A). Addition of C3 transferase promotes a transition from focal adhesions to focal complexes and an increase in filopodia formation (B). The extent of outgrowth was promoted by transfection with a vector expressing p190 Rho GAP (C). Addition of P190-Rho GAP promotes a focal adhesion to focal complex transition, and the formation of filopodia (D).

Rho-GAPs are naturally occurring Rho regulators that act by promoting the GTPase activity of Rho-GTP, resulting in production of Rho-GDP at the expense of Rho-GTP. Transient transfection of a p190-RhoGAP-encoding construct [23] increased outgrowth distance relative to untreated controls and outgrowths transfected with empty vector (Figure 3C). Again, inhibition of Rho led to a decrease in actin stress fibers and focal adhesions, and an increase in focal complexes and filopodia (Figure 3D), with control outgrowths averaging 0.7 filopodia per outer ring cell and p190-RhoGAP transfected outgrowth averaging greater than 1.7 filopodia per outer ring cell.

F9 Vin-/- PE Outgrowth

To determine the role of vinculin in PE migration, we generated embryoid bodies and outgrowth cultures using a vinculin -/- F9 stem cell line [50]. The morphology of the vin-/- PE outgrowth differed from wild-type in that individual elongated cells were readily visible and the outgrowth was less sheet like. Despite this altered morphology, we established that this outgrowth maintained PE identity based upon the expression of TROMA-1, an intermediate filament protein characteristic of this cell type (data not shown). We observed a 2-fold increase in the extent of PE migration from F9 vin-/- embryoid bodies when compared to wild-type F9 PE, (Figures 4A and B). These results are consistent with previous findings that F9 vin -/- stem cells migrate 2.4 times faster than wild type F9 stem cells [50]. To verify that the increased migration was due to the absence of vinculin and not another attribute of the cell line, we used F9 vin-/- cells that had been rescued by transfection with an expression vector encoding mouse vinculin. When outgrowth cultures were generated from these cells, the extent of PE migration was similar to that observed in wild-type cultures (data not shown), demonstrating that the increased motility seen in the F9 vin-/- cells is due to the loss of vinculin.

We compared the adhesive structures observed in wild-type F9 and F9 vin-/- outgrowth cultures. Despite the absence of vinculin, there were abundant focal adhesions in vinculin-null outgrowth. α -actinin (Figure 4F), as well as other proteins, localized to large focal adhesions in F9 vin-/- PE outgrowth. This finding is consistent with previous studies demonstrating that F9 vin-/- stem cells can form focal adhesions with morphological characteristics similar to those found in wild-type cells [51]. The increase in the extent of vin -/- PE migration cannot simply be due to a loss of focal adhesions, suggesting a role for vinculin in cell migration outside of its contribution to focal adhesion integrity.

Figure 4 Loss of vinculin promotes PE outgrowth, and ROCK inhibition of these cultures promotes an additional increase in outgrowth. Phase contrast images of wild type (A), vinculin-null (B), and Y-27632-treated vinculin-null PE outgrowth (C). Loss of vinculin promotes an approximately 2-fold increase in PE outgrowth and addition of Y-27632 increases the extent of outgrowth an additional 2-fold in these cultures (D). Based upon α-actinin localization, vinculin-null outgrowth retains robust focal adhesions (E, F) that are absent upon treatment with Y-27632 (G). Colocalization of α-actinin or paxillin with phalloidin demonstrates the loss of stress fibers as well as focal adhesions upon treatment of vinculin-null outgrowth (H and J) with Y-27632 (I and K).

Upon treatment with Y-27632, the extent of F9 vin -/- PE outgrowth increased dramatically to approximately 2-fold the distance of untreated vin-/outgrowth and 4-fold the distance of control F9 outgrowth (Figures 4A–D). There was a modest increase in lamellipodia and a striking increase in filopodia with inhibitor treatment (Figure 4G I, K). The altered morphology of Y27632-treated vin -/- outgrowth suggested that the cells may no longer be PE. The continued expression of the intermediate filament PE marker recognized by the TROMA-1 antibody by these cells suggests they are PE (data not shown). Phalloidin staining and expression of paxillin and α -actinin are consistent with a loss of stress fibers and focal adhesions, and an increase in focal complexes upon Y-27632 treatment (Figure 4E–K). These data suggest that the loss of focal adhesions as a consequence of ROCK inhibition occurs in both wild-type F9 and F9 vin -/- outgrowth. Delivery of C3-transferase, via transfection of a C3-transferase encoding vector or direct uptake of the enzyme by lipofectamine treated cells or transfection with a p190Rho-GAP-encoding vector also promoted an increase in the extent of outgrowth in vin -/- PE (data not shown).

DISCUSSION

Rho-ROCK and Cell Migration

The F9 embryoid body outgrowth system provides an in vitro model for examining the migration of PE, the first motile cell type in the mammalian embryo. As cells migrate directionally, away from the embryoid body core, they display structural characteristics of migratory cells, including focal complexes, focal adhesions, lamellipodia, and filopodia (Figure 1). To begin to identify the molecular signals regulating PE outgrowth, we looked first at the role of Rho GTPase and its downstream effector ROCK. The ROCK inhibitor Y-27632 promotes migration of PE cells, accompanied by an apparent increase in focal complexes and lamellipodia and a decrease in focal adhesions and stress fibers (Figure 2). Interfering upstream of ROCK by inhibiting Rho activity directly, either by the action of C3 transferase or p190-RhoGAP, has the same effect on cell migration and outgrowth adhesion structures as Y-27632 (Figure 3), suggesting that for PE migration, Rho's effects are mediated via ROCK and not other downstream effectors.

Clues to the mechanism by which Rho/ROCK may inhibit PE outgrowth come from examining ROCK's substrates. ROCK regulates phosphorylation and activation of MLC by at least two routes: it phosphorylates MLC, and myosin light chain phosphatase [33]. Both actions result in a higher level of MLC phosphorylation, and therefore increased myosin II function. There are multiple myosin regulated contractile events involved in cell motility. In some cell types, contraction of the tail at the rear of the cell is an essential myosin-dependent step in migration [60, 61, 3]. This provides an explanation for why certain cell types show decreased motility upon ROCK inhibition, in the absence of tail contraction the cell cannot move forward [37]. For cell types like PE, this contraction at the back of the cell may be less essential for migration than other ROCK/myosin II-mediated events, perhaps because the cells migrate more as a sheet than as single cells. Myosin-II-dependent contractility is also involved in maintaining stable stress fibers and focal adhesions [62, 54]. It is now well established that focal adhesion turnover is required for cell migration [63, 64]. The more large focal adhesions a cell has linked to actin stress fibers, the more difficult it is for that cell to move. Inhibiting this MLC-based activity would promote cell migration.

A role for myosin II at the cell periphery has also been proposed. Myosin II appears to restrict membrane protrusions to the site of migration [65], promoting directed cell movement. Data also suggest that initial assembly of vinculin and zyxin-containing adhesions just behind the leading edge requires myosin II and formation of these attachment sites is essential for the production of membrane protrusions [65]. Inhibition of these MLC functions should block outgrowth by interfering with MLC phosphorylation, yet we see a clear promotion of migration when ROCK is inhibited. An explanation for this apparent discrepancy comes from observations of where in the cell MLC modulators are active [65]. In addition to ROCK, myosin light chain kinase (MLCK) also phosphorylates MLC [33]. Recent studies examining the effect of ROCK or MLCK inhibitors on MLC phosphorylation noted that ROCK inhibition blocks MLC phosphorylation at the center of the cell, where it stabilizes focal adhesions and stress fibers, but not at the periphery, while MLCK inhibition blocks phosphorylation at the periphery and not the center of the cell [65]. ROCK inhibited cells were unable to assemble focal adhesions, and MLCK inhibited cells maintained focal adhesions at the cell center [65]. Consistent with these observations, fibroblasts used in this study showed increased migration when treated with the ROCK inhibitor and decreased migration when treated with a MLCK inhibitor. We see a similar impact of treatment with these two inhibitors on PE outgrowth (Figure 2 and Hong and Grabel, unpublished observations). This division of labor for ROCK and MLCK may not occur in all cell types. For example, in leukocytes, ROCK activity limits membrane protrusions at the periphery, and in this cell type Y-27632 treatment inhibits migration [37].

Cofilin is an indirect downstream target of ROCK that also plays a role in cell migration. ROCK phosphorylates and activates LIM-Kinase, which phosphorylates and inactivates cofilin. Cofilin severs and depolymerizes actin filaments and is involved in Arp 2/3-based actin polymerization that promotes membrane protrusion [66, 38]. ROCK action inhibits this activity, which would stabilize existing actin filaments and slow migration. Y-27632 treatment therefore activates cofilin, promoting membrane protrusion and cell migration. Our observation that cofilin phosphorylation is decreased in PE following Y-27632 treatment suggests that cofilin activation may contribute to the increased levels of migration observed under this condition.

Vinculin and Cell Migration

The absence of vinculin also promotes migration, but based upon the continued presence of focal adhesions and stress fibers in vinculin-null outgrowth, the molecular basis of this enhancement is distinct from ROCK inhibition. The observation that vinculin-deficient F9 stem cells have focal adhesions comparable to wild type cells [50], suggests they compensate for the absence of vinculin by increasing the targeting of other proteins, such as talin, to the focal adhesion, demonstrating that vinculin is not essential for focal adhesion formation. The mechanism promoting increased outgrowth in these cultures is likely due to the recently described role of vinculin in preventing the interaction between paxillin and FAK [52]. Association between paxillin and FAK promotes phosphorylation of both proteins, and mediates two downstream events key to cell migration, focal adhesion turnover, and Rac activation, via PI3-K [53, 67]. A Y to F point mutation at residue 822 within the hinge-tail fragment of vinculin is sufficient to eliminate the ability of vinculin to bind to paxillin and interfere with FAK-paxillin binding [52]. F9 stem cells deficient for vinculin, or expressing the Y822F mutation, migrate more extensively than wild type cells. The enhanced migration of vinculin deficient stem cells relies upon ERK activity [52]. Paxillin can apparently provide a scaffold for the binding of the MAP Kinase cascade components Raf, MEK, and ERK, and paxillin itself is a substrate for ERK phosphorylation [53]. This phosphorylation event promotes the association of FAK with paxillin, and the activation of FAK in focal adhesion turnover. Vinculin's association with paxillin may block one or more of paxillin's interactions.

The work presented here demonstrates a role for the Rho signaling pathway in PE migration. In addition, this study, along with previously published reports, demonstrates the complexity of Rho regulation of the various actin-based events that mediate cell migration. Despite the conservation of downstream targets of Rho, there is temporal, and spatial specificity to their action, which can alter their overall contribution to cell migration in different cell types. The F9 PE outgrowth system described here provides an in vitro model for examining the regulation of the embryo's first migration event. Future studies will unravel how the action of the Rho family GTPases, and other signaling cascades, are coordinated in time and space to promote the directed outgrowth of PE.

ACKNOWLEDGEMENTS

We thank Eileen Adamson for generously supplying us with vin-/- and rescued F9 stem cells and Keith Burridge for providing us with C3 transferase-encoding and p190 Rho-GAP-encoding vectors. This work was supported by NIH grant R15 CA090305 (to L.G.).