The small GTPase R-Ras regulates organization of actin and drives membrane protrusions through the activity of PLCϵ

Cell adhesion to the extracellular matrix is a complex phenomenon that occurs in stages. Adhesion starts with the interaction of integrins with the extracellular matrix (ECM), followed by active cell spreading and culminates in the contraction of the actomyosin cytoskeleton. Following contact with the ECM, cells dynamically remodel the actin cytoskeleton (Jones et al., 1998; van Kooyk and Figdor, 2000; van Kooyk et al., 1999), which drives membrane extension during cell spreading (Chen et al., 2003; Engler et al., 2004; Wakatsuki et al., 2001) and continues until the cell has reached its maximum contact area with the substratum. Cells spread by multiple mechanisms, in part dominated by either lamellipodia- or filopodia-based membrane extensions, depending on the cell type, fate and environment. Most of the events involved in the structural regulation of membrane protrusions have been well characterized (Pantaloni et al., 2001; Pollard and Borisy, 2003; Small et al., 2002). Lamellipodia are composed of short actin filaments that are highly branched owing to the activity of capping and severing proteins (Pollard and Borisy, 2003; Small et al., 2002). Filopodia are composed of long, unbranched, parallel bundles of actin filaments and require a decrease in capping and severing protein activity and the anti-capping activity of Mena/VASP (Bear et al., 2002; Mejillano et al., 2004; Svitkina et al., 2003). Although the actin-binding proteins regulating lamellipodial and filopodial protrusions have been well studied, much less is known about the upstream signaling pathways that regulate these events.

Initiation of cell adhesion and spreading requires an increase in cytosolic Ca²⁺ concentration [Ca²⁺]_c (Pettit and Hallett, 1996a; Pettit and Hallett, 1996b; Pettit and Hallett, 1998; van Kooyk and Figdor, 2000). The Ca²⁺ rise is mediated through the release of intracellular Ca²⁺ stores by inositol triphosphate (IP₃) produced as a result of the activity of phospholipase C (PLC); as well as an influx of extracellular Ca²⁺ through Ca²⁺ channels at the plasma membrane (McNamee et al., 1993; McNamee et al., 1996). The activation of PLC has been shown to initiate cell spreading in multiple cell types (Jones et al., 2005; Pettit and Hallett, 1998), and initiates membrane protrusions during carcinoma cell migration (Mouneimne et al., 2004). These effects of PLC activity are potentially carried out in two ways. First, the mobilization of intracellular Ca²⁺ stores by IP₃ production can activate capping proteins (Arora et al., 2003; Arora et al., 2004; McGough et al., 2003). Second, by hydrolyzing phosphatidylinositol 4,5-bisphosphate (PIP₂), PLC also relieves the inhibitory effect of PIP₂ on capping and severing proteins (Sechi and Wehland, 2000; Takenawa and Itoh, 2001).

R-Ras, a member of the Ras subfamily of small GTPases, promotes cell adhesion, integrin activation, spreading, migration, cell polarity, axonal outgrowth and phagocytosis (Berrier et al., 2000; Ivins et al., 2000; Jeong et al., 2005; Keely et al., 1999; Kwong, 2003; Self et al., 2001; Sethi et al., 1999; Wozniak et al., 2005; Zhang et al., 1996) through mechanisms not fully known. These cellular events require the tight regulation of actin dynamics, suggesting that R-Ras signaling contributes to the modulation of the actin cytoskeleton. Furthermore, many of these events are driven by persistent protrusions mediated by lamellipodia formation. We found that R-Ras activation drives persistent membrane protrusion and promotes cell spreading, whereas knockdown of endogenous R-Ras inhibits protrusion and spreading. Moreover, expression of constitutively active R-Ras changes the dynamics of membrane protrusion to a lamellipodium-based mode of spreading, whereas dominant-negative R-Ras enhances filopodial-based protrusions. This switch was associated with a longer protrusive process mediated by a strong ruffling lamellipod. The formation of the ruffling lamellipod was blocked by PLC inhibitors and by the Ca²⁺ chelator BAPTA. We find evidence that PLCϵ is a novel R-Ras effector, as it co-precipitates with R-Ras and is activated by R-Ras. Furthermore, knockdown of PLCϵ with siRNA inhibited the ability of R-Ras cells to form the ruffling lamellipod. Overall these data suggest that R-Ras activity sustains protrusive motility during cell spreading by regulating the actin cytoskeleton through the activity of PLCϵ.

R-Ras potently enhances cell adhesion and subsequent events of adhesion-based signaling (Kwong, 2003; Zhang et al., 1996). To better understand mechanisms by which R-Ras increases cell adhesion, the immediate events that occur during initial ECM contact and spreading of cells were determined using total internal reflection fluorescence (TIRF) microscopy (Dubin-Thaler et al., 2004; Giannone et al., 2004). TIRF allowed us to study the dynamics of cell spreading at the region of contact between the cell and the extracellular matrix by exploiting the high contrast between contact and non-contact areas. Cell adhesion was induced by centrifugation and the kinetics of cell spreading were observed during the first 40 minutes of adhesion. MCF10A cells overexpressing constitutively active R-Ras (R-Ras^38V) spread to a greater extent than control cells. On average, control cells reached a maximum spread area of 350 μm², whereas their R-Ras counterpart reached a spread area of 800 μm² during the same time period (Fig. 1A).

Differences in cell spreading kinetics could be explained by a switch in the mechanism of cell spreading when R-Ras was activated. In general, control cells spread by extending filopodia-like structures, with minimal protrusive activity as shown by the uniform intensity of TIRF images during cell spreading (Fig. 1B and Movie 1 in supplementary material). This indicated that this cell was making stable contact with the ECM even as the contact area of the cell with the ECM was increasing (Fig. 1B). Contrary to filopodial extensions, R-Ras^38V-expressing cells had a smoother leading edge (Fig. 1B and supplementary material Movie 2). Interestingly, R-Ras^38V cells had a stronger TIRF signal at the center of the cell surrounded by a lower intensity at the cell leading edge. Thus, unlike the cell center, the protrusive edge of these cells did not make strong contact with the ECM as the cell was spreading. The intensity became even after 500-700 seconds, suggesting that the cells protruded their leading edge for this period of time before they could stabilize these protrusions.

Subsequent mathematical analysis (Dubin-Thaler et al., 2004), of TIRF images demonstrated that control cells initiated spreading with a brief global burst (3 minutes) of protrusive activity. This was followed by alternating retractive and protrusive events (displayed in Fig. 1C as cool and warm colors, respectively) for the next 5-15 minutes and ended with a decline in membrane activity (>15 minutes). These protrusive stages correlated with the increase in cell surface area (Fig. 1C, top panel). The kinetics of cell spreading were altered in R-Ras^38V-expressing cells. The protrusive stage was longer, lasting about 15 minutes on average (Fig. 1C). The timing of this protrusive phase correlated with the time at which the leading edge of R-Ras cells had little interaction with the substratum as determined by TIRF. The second phase of spreading started with a decrease in the edge extension events and a burst in cell edge retraction events (phase 2; Fig. 1C). Strong contact with the ECM correlated with the appearance of retraction events at this time. Thus, the duration of the initial protrusive activity set the spreading behavior of the cell, and this duration was increased in cells expressing active R-Ras. In summary, TIRF analysis of membrane dynamics in R-Ras^38V cells indicates that the major difference in the spreading of control and R-Ras cells is defined by the duration of the initial protrusive events that lasted 3 minutes in control cells and 15 minutes in R-Ras cells.

The observation that the duration of the initial protrusive events in R-Ras^38V-expressing cells is fivefold higher than control cells prompted us to investigate the underlying mechanism of that increase. Control and R-Ras cells were fixed and stained with TRITC-phalloidin to determine the structure of the actin cytoskeleton during cell spreading. Upon adhesion to collagen I or human fibronectin, control cells spread with a combination of filopodia and lamellipodia (Fig. 2A). These cells acquired a polarized phenotype within 45 minutes (data not shown). By contrast, cells expressing R-Ras^38V, spread faster and exhibited a prominent peripheral actin network reminiscent of thick lamellipodia or ruffles, referred to hereafter as a `ruffling lamellipod' (Fig. 2A). At the end of spreading (>90 minutes), R-Ras^38V cells formed strong stress fibers and large focal adhesions, but did not acquire a polarized morphology (data not shown). To determine whether the effects of R-Ras on actin organization were specifically integrin dependent, cells were plated on poly-L-lysine (Bailly et al., 1998; Dubin-Thaler et al., 2004; Giannone et al., 2004). Here too, control and R-Ras^38V cells organized their actin cytoskeletons differently (Fig. 2A, bottom panels). Control cells formed extensive filopodia upon adhesion, whereas cells expressing R-Ras^38V did not form filopodia, but had the same ruffling lamellipod. Thus activation of R-Ras changed the organization of the actin cytoskeleton in an integrin-independent manner. Contrary to activated R-Ras, expression of dominant-negative R-Ras^41A in MCF10A cells enhanced filopodia formation (arrows in Fig. 2B).

To further examine the role of R-Ras in the regulation of membrane protrusions, wild-type, dominant-negative and constitutively active GFP-R-Ras were transiently transfected into Cos7 cells, and spreading on fibronectin was determined. These cells were used in order to determine effects of R-Ras under transient transfection conditions, rather than the stable cell lines that are usually required for both MCF10A and T47D breast cells, which are not as readily transfected. Cells were double labeled with phalloidin to determine the localization of R-Ras relative to the actin cytoskeleton, and three phenotypes identified (A, B and C). Cells expressing GFP-R-Ras^wt often had polar membrane protrusions, and localized wild-type R-Ras to the protrusive leading edge (phenotype A, Fig. 2C,D). Consistent with results for MCF10A cells, GFP-R-Ras^41A enhanced the number of cells with prominent filopodia (phenotype B, Fig. 2C,D) and did not localize to the plasma membrane. By contrast, cells expressing GFP-R-Ras^38V often lost cell polarity and exhibited a strong ruffling lamellipod (phenotype C, Fig. 2C,D). Notably, like GFP-R-Ras^wt, GFP-R-Ras^38V localized to membrane ruffles. These observations strengthen our hypothesis that activation of R-Ras drives cell spreading by enhancing membrane protrusions that are lamellipodial-based. Moreover, these results suggest that R-Ras activity is a negative regulator of filopodial formation.

Previously, we reported that constitutive activation of R-Ras inhibited random cell migration of T47D breast epithelial cells by inhibiting cell polarity and membrane protrusion (Wozniak et al., 2005). To reconcile those results with the enhanced membrane protrusion and spreading resulting from R-Ras activation noted in Fig. 1, we determined whether expression of R-Ras^38V would affect the random migration of MCF10A cells. Control and R-Ras^38V cells were plated on fibronectin for 90 minutes and their ability to migrate was monitored by time-lapse video microscopy. Within that time frame, control cells actively migrated, whereas cells expressing R-Ras^38V showed greatly reduced migration (Movies 3 and 4 in supplementary material). Moreover, R-Ras cells had reduced ability to polarize front to rear, consistent with the lack of polarization also noted during cell spreading. Thus, consistent with our previous results in a different cell line, cells expressing active R-Ras^38V show impaired migration due in part to a defect in cell polarity (Wozniak et al., 2005). Overall, these data suggest that the mechanism by which R-Ras regulates membrane protrusions during spreading is distinct from the mechanism by which R-Ras modulates protrusions during migration.

To further determine whether endogenous R-Ras plays a role in cell spreading, MCF10A cells were transfected with siRNA directed against R-Ras, which caused a dramatic reduction in R-Ras expression levels (Fig. 3A). Knockdown of R-Ras caused a significant (>50%) decrease in the spreading of MCF10A cells, as determined by the spread cell area (Fig. 3B). Knockdown of R-Ras by RNAi in T47D cells also inhibited cell spreading (not shown), consistent with previous results demonstrating knockdown of R-Ras by RNAi in these cells inhibited membrane protrusion and cell migration (Wozniak et al., 2005).

The activation of PLC has been shown to initiate cell spreading and membrane protrusions in multiple cell types (Jones et al., 2005; Mouneimne et al., 2004; Pettit and Hallett, 1998). Based on the observations made in Figs 1, 2, 3, we hypothesized that the formation of the ruffling lamellipod in cells expressing R-Ras^38V was potentially mediated by the activation of PLC pathways. To test the role of PLC in R-Ras^38V signaling, cells were pretreated with U73122, a synthetic PLC inhibitor. Note that this is a broad-specificity PLC inhibitor, so it will not discriminate between PLC isoforms. Treatment of cells with 1 μM U73122 inhibited the ability of control cells to adhere and spread on fibronectin (Fig. 4A-C). In R-Ras^38V cells, the same concentration of U73122 had minimal effect on adhesion (Fig. 4A), but these cells could not spread well and more importantly, they could not form the ruffling lamellipod (Fig. 4B,C). However, 10 μM U73122 reduced adhesion of R-Ras-expressing cells to that of control cells (Fig. 4A). These data suggest that the formation of a ruffling lamellipod in R-Ras^38V cells is dependent on the activity of PLC.

PLC activity can potentially regulate protrusion via the activation of PKC downstream of DAG, and/or via the mobilization of intracellular Ca²⁺ that will regulate actin capping and severing proteins. Because cell spreading in epithelial cells is regulated by the mobilization of intracellular Ca²⁺ stores (Spoonster et al., 1997), and R-Ras activity has been linked to Ca²⁺ regulation (Koopman et al., 2003), we tested the effects of Ca²⁺ depletion on R-Ras-enhanced cell spreading. Cells in suspension were pretreated with 10 μM BAPTA-AM, a chelator of cytoplasmic Ca²⁺, for 20-30 minutes at 37°C. Treatment of both control and R-Ras^38V cells with BAPTA-AM inhibited cell adhesion to fibronectin (Fig. 4A). The inhibitory effect of BAPTA on cell adhesion could be lessened by reducing the incubation time, allowing us to assess the effect of BAPTA-AM on cell spreading. BAPTA-AM inhibited the formation of the ruffling lamellipod in R-Ras cells and restored the formation of filopodia (Fig. 5A). Moreover, BAPTA-AM significantly diminished cell spreading (Fig. 5B). Since BAPTA-AM could inhibit adhesion (Fig. 4A), it was important to rule out effects on cell spreading that were secondary to adhesion; thus, cells were pre-plated on fibronectin for 20 minutes before the addition of BAPTA-AM. Chelating intracellular Ca²⁺ after the initiation of cell spreading also abolished the ability of these cells to form the ruffling lamellipod (Fig. 5A) suggesting that cytoplasmic Ca²⁺ plays an important role in actin organization and cell spreading separate from mediating initial adhesion.

To verify that the effects of BAPTA-AM were specific to intracellular, external Ca²⁺ was chelated by incubating cells with EGTA. Surprisingly, EGTA did not affect either lamellipod formation or spreading of R-Ras^38V cells (Fig. 5A). However, EGTA did inhibit adhesion of control cells and delayed the onset of adhesion in R-Ras cells (Fig. 4A). Because integrin-mediated adhesion is divalent cation-dependent, again cells were allowed to attach prior to treatment with EGTA to rule out effects on adhesion. Both pre-adherent control and R-Ras-expressing cells were unaffected by treatment with EGTA. Thus external Ca²⁺ is essential for the initiation of adhesion, but not for subsequent spreading, which is driven by cytosolic Ca²⁺.

Chronic activation of PLC depletes internal Ca²⁺ stores (Parekh, 2003). Because PLC may be activated downstream of R-Ras, we determined the ability of R-Ras^38V cells to mobilize internal Ca²⁺ stores. Control and R-Ras^38V-expressing cells were treated with thapsigargin (Tg), a selective inhibitor of the ER Ca²⁺-ATPase (SERCA), which induces a transient flux in cytosolic Ca²⁺ concentration by preventing the active re-uptake of Ca²⁺ back into the continuously leaky ER lumen. When 4 μM thapsigargin was applied to control cells in suspension, a strong Ca²⁺ flux was observed (Fig. 5C). However, the same concentration of Tg had no effect on the Ca²⁺ concentration in R-Ras^38V cells (Fig. 5C) (Koopman et al., 2003). These experiments were repeated either in Ca²⁺-free medium or with 1 mM calcium chloride, with the same results (data not shown). R-Ras^38V-expressing cells were largely unresponsive to every stimulus tested, including the Ca²⁺ ionophore, ionomycin (Fig. 5C), forskolin, bradykinin, 8-bromo-CPT-cAMP and db-cAMP (data not shown). These results suggest that chronic signaling due to expression of R-Ras^38V depletes internal Ca²⁺ stores, consistent with a model in which constitutive activation of R-Ras leads to over activation of PLC pathways and chronic release of Ca²⁺. Moreover, these results support the hypothesis that the mobilization of Ca²⁺ downstream of PLC activity is required for the initiation of cell spreading in breast epithelial cells and that R-Ras activation prolongs the activity of PLC in the initiation of cell protrusion.

Phosphoinositide-specific phospholipases are a large family of enzymes that differ in structure and tissue distribution, but have the same function. In mammalian cells there are 11 PLC isozymes grouped in four different classes (PLCβ, -δ, -ϵ and -γ) that are regulated by distinct mechanisms. The function of PLCγ has been linked to integrin signaling (Keely and Parise, 1996). However, tyrosine phosphorylation of PLCγ was not enhanced in cells expressing R-Ras^38V, suggesting that R-Ras does not activate PLCγ (data not shown).

PLCϵ, the most recently cloned member of the PLC family, is activated by members of the Ras family of small GTPases and by increases in cytosolic Ca²⁺ concentration (Czyzyk et al., 2003; Kelley et al., 2001; Kelley et al., 2004; Song et al., 2001; Song et al., 2002). To investigate a link between R-Ras and PLCϵ, their interaction in cells was evaluated. R-Ras^WT or R-Ras^38V was co-expressed with Flag-PLCϵ in Cos7 cells, and PLCϵ was immunoprecipitated with an anti-Flag antibody. We found that PLCϵ co-immunoprecipitated R-Ras^WT and R-Ras^38V (Fig. 6A, lanes 2 and 3). These results suggest that PLCϵ and R-Ras can interact in cells. Notably more R-Ras^38V was co-immunoprecipitated (Fig. 6A, lane 3), suggesting a preference for the active GTP-bound form of R-Ras.

GST-pulldown assays were used to investigate whether R-Ras interacts with the Ras-binding domain of PLCϵ (RA2). R-Ras^38V expressed in MCF10A cells binds the RA2 domain of PLCϵ (GST-PLCϵ-RA2) (Fig. 6B). Endogenous R-Ras also bound to GST-PLCϵ-RA2 (Fig. 6C,D). Activation of endogenous R-Ras, with 8-CPT-2Me-cAMP slightly enhanced binding to PLCϵ-RA2 (Fig. 6C). However, activation of endogenous R-Ras with fibronectin did not significantly enhance binding to PLCϵ-RA2 (Fig. 6D). Similar results were found for the Ras-binding domain of Raf, which robustly bound to endogenous R-Ras even in the absence of additional stimulation, suggesting that a pool of active R-Ras exists in cells.

The ability of R-Ras to activate PLCϵ was determined with a PLC activity assay. Flag-PLCϵ was co-expressed with either an empty pCMV5 vector (cmv), dominant-negative R-Ras^41A, or constitutively active R-Ras^38V. Forty-eight hours post-transfection, cells were assayed for PLC activity (Fig. 6E). Cells that were co-transfected with R-Ras^38V and PLCϵ showed a high PLC activity when compared with cells that expressed either an empty vector or dominant-negative R-Ras (Fig. 6E). In this case, dominant-negative R-Ras did not further diminish the already low baseline PLC activity (Fig. 6E). Immunoblotting confirmed that these results were not due to expression differences, as we noted equal expression of R-Ras^41A or R-Ras^38V (Fig. 6F). Thus, activation of R-Ras enhances PLCϵ activity in cells.

To further demonstrate that PLCϵ is downstream of R-Ras signaling, PLCϵ function was blocked using siRNA directed against PLCϵ. PLCϵ protein was knocked down to about 50% (Fig. 7A). In this cell population, there was a decrease in the mean area of spreading, with more cells having a smaller area (Fig. 7B). However, the difference was not statistically significant, perhaps reflecting that the knockdown was partial. Notably, knocking down PLCϵ altered the actin cytoskeleton and the phenotype of the cells, and inhibited the formation of the large ruffling lamellipod during cell spreading (Fig. 7C, phenotype C). Although about 50% of R-Ras^38V-expressing cells were characterized by a strong ruffling lamellipod, this was inhibited to less than 5% upon treatment with PLCϵ siRNA (phenotype C, Fig. 7D). Some cells treated with siRNA for PLCϵ acquired filopodia (Fig. 7C, phenotype B) and resembled cells treated with the Ca²⁺ chelator, BAPTA-AM. This phenotype represented about 5% of the population compared with 0% in the absence of PLCϵ siRNA (phenotype B, Fig. 7D). Additionally, a new phenotype was noted, characterized by the extension of multiple cell protrusions, which represented about 80% of R-Ras^38V-expressing cells treated with PLCϵ siRNA (D in Fig. 7C,D). Overall these data suggest that the effects of R-Ras on the actin cytoskeleton and the formation of the ruffling lamellipod are due in part to PLCϵ.

R-Ras increases integrin-mediated adhesion and many cellular processes that are dependent on integrin function, but the molecular mechanisms involved in these events have remained elusive. Our data here and previously (Wozniak et al., 2005) demonstrate an important role for R-Ras in the regulation of the actin cytoskeleton and membrane protrusion during both cell spreading and migration. Knocking down endogenous R-Ras by RNAi inhibits membrane protrusion, cell spreading (Fig. 3) and cell migration (Wozniak et al., 2005). Proper regulation of R-Ras activity appears to regulate the actin cytoskeleton and determine the balance between the formation of lamellipodia versus filopodia (Fig. 2). Misregulation of R-Ras by the expression of dominant-negative R-Ras^41A enhances filopodia formation, whereas expression of constitutively active R-Ras^38V inhibits filopodia and results instead in a strong ruffling lamellipod. The formation of this lamellipod in cells expressing constitutively active R-Ras^38V increases the persistence of leading-edge protrusion to enhance cell spreading. Further analysis demonstrated that R-Ras regulates the cytoskeleton in part through PLCϵ (Figs 6 and 7). Inhibiting phospholipase activity by pharmacologic inhibition, siRNA or depleting cytosolic Ca²⁺ with BAPTA abolished the ability of R-Ras cells to form the ruffling lamellipod and diminished cell spreading. Moreover, R-Ras co-immunoprecipitates with a novel isoform of PLC, PLCϵ and increases its phospholipase activity, suggesting a functional link between R-Ras activation and PLCϵ.

Based on these results, we propose the following model (Fig. 8): R-Ras is activated by integrin-mediated adhesion (Jeong et al., 2005; Wozniak et al., 2005) and directly binds to and activates PLCϵ, which in turn, leads to Ca²⁺ mobilization from intracellular stores via IP₃-signaling pathways. The subsequent increase in cytosolic Ca²⁺ concentration locally activates actin-binding proteins, such as capping proteins (Arora et al., 2003; Arora et al., 2004; McGough et al., 2003) and regulates membrane/vesicle trafficking (Bader et al., 2004), all of which will promote the formation of the ruffling lamellipod. The resulting lamellipod exerts outward forces on the plasma membrane leading to an increase in cell spreading. The finding that R-Ras localizes to membrane microdomains (Hansen et al., 2003) and is found on the ruffling lamellipod (Fig. 2) (Kwong, 2003) is consistent with a role in membrane protrusion via PLCϵ. We further propose that regulation of the actin cytoskeleton and enhanced cell spreading leads to enhanced integrin clustering and adhesion, and the subsequent formation of strong focal adhesions. Indeed, R-Ras enhances focal adhesion formation and integrin-mediated signaling events (Kwong, 2003). Moreover, signaling events leading to remodeling of the actin cytoskeleton regulate integrin activation and function, presumably in part through the clustering of integrins, which strengthens the adhesion process (Kucik et al., 1996; Lub et al., 1997; Sampath et al., 1998; van Kooyk and Figdor, 2000; Yauch et al., 1997).

Our results suggest that phospholipase Cϵ is a novel effector of R-Ras. R-Ras could be co-precipitated with PLCϵ under conditions in which a known R-Ras effector, Raf, could also be co-precipitated. When co-expressed in cells, constitutively active R-Ras associated with PLCϵ to a greater extent than wild-type R-Ras, and activation of R-Ras activated PLCϵ, both consistent with an effector role for PLCϵ downstream of R-Ras. Consistent with our findings, it has been previously noted that activated R-Ras2/TC21 also activates PLCϵ (Kelley et al., 2004). R-Ras was specifically pulled down with the Ras-association domain, RA2, which binds to H-Ras and Rap1 (Kelley et al., 2001; Song et al., 2001). A role for PLCϵ as an R-Ras effector is further supported by the finding that R-Ras point mutants in the effector-binding loop that eliminate the effects of R-Ras on focal adhesion formation and signaling (Kwong, 2003) are the same point mutations that eliminate H-Ras binding to PLCϵ (Kelley et al., 2001; Song et al., 2001).

Our results suggest that PLC regulates membrane protrusion and cell spreading. PLC inhibition blocked cell spreading and membrane protrusions in control and R-Ras^38V-expressing cells. Our data support recent observations that initiation of protrusion in carcinoma cells is driven by the activation of PLC (Jones et al., 2005; Mouneimne et al., 2004). Various PLC isoforms have been implicated as key players in early integrin-mediated adhesion (Jones et al., 2005; Keely and Parise, 1996). Our results suggest that the PLCϵ isoform also regulates early events of spreading and adhesion. Once activated, PLC cleaves phosphoinositol 4,5-bisphosphate (PIP₂). By cleaving PIP₂, PLC activity is thought to decrease membrane tension, allowing the cell to expand its plasma membrane (Raucher and Sheetz, 2000; Raucher and Sheetz, 2001; Raucher et al., 2000). Furthermore, the hydrolysis of PIP₂ relieves the inhibitory effects of PIP₂ on gelsolin and cofilin, thereby promoting actin polymerization (reviewed by Yin and Janmey, 2003). The cleavage products of PIP₂ - IP₃ and diacylglycerol (DAG) - also have regulatory effects on cell spreading. DAG activates PKC pathways and the MARCKs proteins, all of which have been shown to promote cell spreading (Yue et al., 2000a; Yue et al., 2000b; Zhou and Li, 2000). Previously we have shown that PKC pathways were partially involved in the promotion of haptotatic migration by R-Ras^38V (Keely et al., 1999).

IP₃, on the other hand, mobilizes internal Ca²⁺ stores. Recently, Itano et al. showed that endoplasmic and nuclear Ca²⁺ stores are released during cell spreading (Itano et al., 2003). Moreover, thapsagargin treatment to release Ca²⁺ stores enhances the formation of lamellipodia in epithelial cells (Price et al., 2003). ER Ca²⁺ levels are misregulated in cells overexpressing R-Ras^38V (Fig. 5C) (Koopman et al., 2003), suggesting that R-Ras activation does indeed cause the release of Ca²⁺ from ER stores. This rise in intracellular Ca²⁺ is necessary for R-Ras to induce the ruffling lamellipod, as formation of the lamellipod was inhibited by cells treated with the intracellular Ca²⁺ chelator, BAPTA-AM (Fig. 5A). We propose that the resulting rise in Ca²⁺ concentration activates Ca²⁺-dependent actin-remodeling proteins, such as gelsolin and other capping proteins (Lader et al., 1999; Lin et al., 2000; Sun et al., 1999). Indeed, we find that formation of the ruffling lamellipod observed in R-Ras^38V-expressing cells depends on the activity of gelsolin (A.S.A.-N. and P.J.K., unpublished data).

The cAMP-Epac-Rap pathway has been shown to directly activate PLCϵ to mobilize intracellular Ca²⁺ and promote cell adhesion (Jin et al., 2001; Rangarajan et al., 2003; Schmidt et al., 2001; Song et al., 2002; Stope et al., 2004). Recently, R-Ras was also shown to be activated by cAMP (Cole et al., 2003) (A.S.A.-N. and P.J.K., unpublished results), suggesting that it may be in this cAMP-Epac pathway. In addition to Ras-association domains, PLCϵ contains a Cdc25 domain that functions as an exchange factor for H-Ras and Rap1 (Jin et al., 2001; Lopez et al., 2001). Whether PLCϵ is not only an effector, but also an exchange factor for R-Ras, and mediates R-Ras activation by cAMP remains to be determined. However, PLCϵ could serve as an exchange factor by which R-Ras activates Rap1, as some of the effects of R-Ras on adhesion and phagocytosis are dependent on Rap1 (Bos et al., 2003; Caron et al., 2000; Ohba et al., 2001; Self et al., 2001; Song et al., 2001; Song et al., 2002). In light of this new evidence, we propose that R-Ras1, R-Ras2 and Rap1 have redundant functions in the regulation of integrin signaling through the activation of a common PLCϵ-Ca²⁺-integrin pathway.

Membrane protrusion is driven by actin polymerization, myosin-based retractions and actin depolymerization events. Whether protrusion takes the form of filopodia, lamellipodia or both is regulated by several actin binding proteins including capping protein, ADF/cofilin, WASp/Scar proteins, Mena/VASP, and Arp2/3, which determine the degree of branching of the actin network (Bear et al., 2002; Mallavarapu and Mitchison, 1999; Mejillano et al., 2004; Pollard and Borisy, 2003; Small et al., 2002; Vignjevic et al., 2003). TIRF analysis shows that R-Ras^38V cells exhibit long-lived extension events and minimal retraction events to yield robust spreading and a significant increase in cell surface area. Exactly how R-Ras^38V signaling opposes the retraction events is not clear at this moment. Our data show that actin polymerization is greatly enhanced in R-Ras^38V cells, as shown by the ruffling lamellipod (Fig. 3A) and by biochemical analysis demonstrating a substantial increase in the insoluble actin fraction (A.S.A.-N. and P.J.K., unpublished data). Moreover, many actin-binding proteins involved in protrusive extensions were enhanced in the actin insoluble fraction and exhibited strong colocalization with F-actin at the protruding lamellipod (A.S.A.-N. and P.J.K., unpublished data).

Recent reports (Jeong et al., 2005; Wozniak et al., 2005) demonstrate that the activation of R-Ras inhibits membrane protrusion during cell migration, whereas here we report that R-Ras activation enhances protrusion during cell spreading (Fig. 1), but inhibits random cell migration (Movies 1-4 in supplementary material). Thus, there appears to be a difference in the mechanism of cell spreading versus cell migration. During cell spreading, R-Ras enhances leading-edge protrusions that lead to enhanced adhesion. Once a cell has become stably attached, this enhanced adhesion could prevent further protrusions (Cox et al., 2001). For cell migration, cells require short-lived extension and retraction events, which result in a net movement of the leading and trailing edge of the cell (Dunn et al., 1997). The inability of R-Ras^38V cells to move suggests that misregulating membrane protrusions in any way would hamper the ability of these cells to migrate. Cell migration is dependent on the ability of a cell to polarize and provide directionality to membrane protrusions. Misregulation of R-Ras by expression of R-Ras^38V eliminates cell polarity (Fig. 3) (Wozniak et al., 2005), which is needed for cell migration, but apparently not for cell spreading.

Our data suggest that the effects of R-Ras^38V on cell adhesion and focal adhesion formation occur secondarily to its effects on membrane protrusion during cell spreading. Focal adhesion formation does not occur until well after the cells have spread, consistent with the idea that a cell must achieve a given size in order to generate the internal forces necessary for focal adhesion formation (Chen et al., 2003; Tan et al., 2003). In support of this, we find that inhibitors of Rho/ROCK contractility block the formation of large focal adhesions (Wozniak et al., 2005), but have no effect on the spreading and formation of the dense actin network noted here (data not shown), consistent with the idea that focal adhesion formation is subsequent to cell spreading. Therefore, the combined effects of R-Ras on increasing cell-ECM contact area through cell spreading, and regulating the organization of the actin cytoskeleton probably contribute to an increase in integrin avidity and cell adhesion.

The formation of the lamellipodia in R-Ras^38V cells was not associated with an increase in Rac activation, nor was it mediated by PI3K (our unpublished observations). Thus, our work hints at a novel Rac-independent pathway in the regulation of lamellipodia formation that instead involves R-Ras proteins. So far, Rab5, a small G-protein involved in the regulation of vesicle fusion and trafficking, has been shown to induce lamellipodia in a PI3K-Rac independent manner (Spaargaren et al., 1994). The novel mechanism mediated by Rab5 was strongly correlated to its ability to promote endocytosis. Whether R-Ras could regulate vesicle trafficking is unclear at this point, but the strong ruffles suggest that this is a possibility.

In summary, R-Ras is an important regulator of membrane protrusion and the actin cytoskeleton during both cell spreading and cell migration. Enhanced cell spreading is driven by the formation of a strong ruffling lamellipod due in part to the action of PLCϵ and the release of intracellular Ca²⁺ downstream of R-Ras signaling. Moreover, misregulation of R-Ras disrupts cell polarity and migration, suggesting that R-Ras-mediated activation of PLC pathways is normally regulated in a spatial and temporal manner.

MCF10A cells were infected with pZIP retrovirus or pZIP containing R-Ras in which glycine 38 was mutated to valine (R-Ras^38V), and selected in G418 as pools of stable transfectants using protocols previously described (Clark et al., 1996). Expression of R-Ras^38V was verified by immunoblotting using anti-R-Ras antibody (Santa Cruz Biochemicals, Santa Cruz, CA).

Cells in suspension were loaded with 5 μM calcein-AM (Molecular Probes) for 20 minutes and centrifuged at low velocity. Cells were imaged with a 20× water-immersion objective for calcein using an upright Olympus BX-50 microscope using 568 nM fluorescent excitation light from a Melles Griot krypton-ion laser. Images were recorded with a Coolsnap FX CCD camera. Individual fluorescent TIRF images of spreading cells were processed as described elsewhere using a custom MATLAB program in which the outer cell perimeter was fitted with a contour (Dubin-Thaler et al., 2004; Giannone et al., 2004). Further analysis was performed as described (Dubin-Thaler et al., 2004). For the adhesion assay, control and R-Ras^38V cells were loaded with 20 μg/ml Calcein-AM for 20 minutes, centrifuged and resuspended in medium pretreated with the indicated pharmacological agent (DMSO, EGTA, BAPTA and U73122 PLC inhibitor) for 30 minutes, followed by adhesion on 96-well plates coated with 20 μg/ml fibronectin for 15 or 50 minutes. Cells were washed three times with PBS and adherent cells were evaluated with a fluorescent plate reader.

MCF10A cells were detached from tissue-culture flasks using 1 mg/ml TPCK trypsin (Sigma). Trypsin was inactivated using 1 mg/ml soy bean trypsin inhibitor (Sigma), and washed into serum-free DMEM F12 media supplemented with 5 mg/ml fatty-acid-free BSA for about 20 minutes. If treated with inhibitors [LY294002 (Alexis Biochemicals), BAPTA-AM (Molecular probes), U73122 (Sigma), EGTA (Sigma)], cells were pretreated with inhibitor for 20-30 minutes at 37°C in suspension. Cells were then plated on plastic dishes or glass coverslips coated with 30 μg/ml collagen I (BD Bioscience), 20 μg/ml human fibronectin (BD bioscience), or 0.01% poly-L-lysine (Sigma) for the indicated time points. Following adhesion, cells were rinsed with PBS and fixed for 15 minutes in 4% paraformaldehyde at room temperature. Formaldehyde was quenched with 0.15 M glycine in PBS. Cells were permeabilized with 0.2% Triton X-100 in PBS, blocked with 1% donkey serum (Jackson ImmunoResearch) and BSA (Fisher Scientific) in PBS and detected with the antibodies of interest. Coverslips were mounted using Prolong antifade mounting medium (Molecular Probes). Microscopy was performed using a Nikon Eclipse TE300 inverted microscope with a Coolsnap FX CCD camera (Roper Scientific). Images were collected and three-dimensional (3D) deconvolution was performed using Inovision software (Durham, NC). Time-lapse images were acquired with E-See Inovision Software (Inovision, Raleigh, NC) as previously described (Wozniak et al., 2005), with one image taken every 60 seconds for 100 minutes.

Digital photographs were captured for random fields from each experiment, and the area of 20 or more individual cells was quantified by tracing the perimeter of the cell, and using Image J software to calculate area, perimeter, and circularity. Data for cell area was further analyzed using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA) to perform statistical analysis using two-tailed unpaired t-tests.

MCF10A cells were seeded into 12-well plates and grown and maintained at 37°C in 10% CO₂ atmosphere in DMEM/F-12 medium containing 5% horse serum. Cells were transfected with dominant-negative R-Ras construct (R-Ras^41A) in a pCMV5 vector and GFP vector (green lantern) using Mirus TransIT-LT1 transfection reagent. 48 hours post transfection, cells were detached and cell spreading assays performed on fibronectin coated coverslips. MCF10A cells were seeded into six-well plates and grown and maintained at 37°C in a 10% CO₂ atmosphere in DMEM/F-12 medium containing 5% horse serum. Cells were transfected with a pool of three PLCϵ siRNAs from Ambion (GCA-CAU-ACU-GUC-AGA-CGA-Att and UUC-GUC-UGA-CAG-UAU-GUG-Ctt; GGU-GAU-AGC-UUU-UGU-AGG-Att and UCC-UAC-AAA-AGC-UAU-CAC-Ctg; GGG-ACU-AAU-AAU-GUC-AUU-UGA-Att and UUC-AAA-UGA-CAU-UAG-UCC-Ctg) using siQuest reagent (Mirus). 72 hours post transfection, cells were lysed and PLCϵ protein levels were assessed by western blotting and fluorescence microscopy using an unpurified anti-PLCϵ antibody (Kelley et al., 2004). For siRNA approaches, MCF10A cells were transfected with siRNA directed against R-Ras or against PLCϵ (each designed as a pool obtained from Dharmacon) using siQuest reagent (Mirus). T47D cells were transfected with siRNA directed against R-Ras as previously described (Wozniak et al., 2005).

Cos7 cells were transiently transfected with pCMV-PLCϵ, pCMV-R-Ras and pCMV-R-Ras^38V constructs were lysed in ice-cold modified RIPA buffer containing 1% NP-40 or 1% Triton X-100 (Keely and Parise, 1996; Keely et al., 1999). PLCϵ was immunoprecipitated by incubating Flag antibody and gamma Sepharose beads with a clarified cell lysate overnight while rotating at 4°C. Immunoprecipitates were separated on acrylamide gels, transferred to PVDF membrane and immunoblotted with the appropriate antibodies. For whole-cell lysate analysis, cells were stimulated by adhesion on fibronectin and lysed with SDS-Laemmli buffer. Lysates were separated on acrylamide gels, transferred to PVDF membrane and immunoblotted with the appropriate antibodies. To measure Cdc42 and Rac activation, cells were stimulated by adhesion on fibronectin (20 μg/ml) and collagen I (30 μg/ml) for the indicated time. Cells were lysed in a modified RIPA buffer containing 1% Nonidet P-40 (MP Biomedicals). After centrifugation, supernatant was incubated with Raf-RBD coupled to glutathione-Sepharose beads. Bound Rac and Cdc42 were detected by immunoblotting with respective antibodies (Transduction Labs). To measure R-Ras and PLCϵ interaction, cells were stimulated by adhesion on fibronectin (20 μg/ml) for 45 minutes and 8-CPT-2Me-cAMP (200 μM) for 30 minutes. Cells were lysed in ice-cold modified RIPA buffer containing 1% Nonidet P-40 (MP Biomedicals) or 1% Triton X-100 (Keely et al., 1999). After centrifugation, the clarified lysate was incubated with GST, GST-Raf-RBD and GST-RA2 (Kelley et al., 2001; Kelley et al., 2004) for 2 hours while rotating at 4°C. Bound R-Ras was detected on immunoblot with an anti-R-Ras antibody (BD-Bioscience-Pharmigen).

Cos7 cells were seeded on six-well plates and transfected with 3 μg plasmid encoding PLCϵ (Song et al., 2001) and 1 μg of dominant-negative R-Ras (R-Ras^41A) or activated R-Ras (R-Ras^38V), using Mirus TransIT-LT1 transfection reagent (Mirus). For PLC assay measurements (Kelley et al., 2001; Kelley et al., 2004), 24 hours post-transfection, Cos7 cells were labeled with 4 μCi myo-D-[2-3H]-inositol (Perkin-Elmer) in inositol- and serum-free Dulbecco's modified Eagle's medium for 24 hours. Cell labeling with myo-D-[2-3H]-inositol was stopped by removing the incubation medium and quenching cells with 1.5 ml of 60 mM perchloric acid plus 0.2 mg/ml InsP6 on ice for 30 minutes. The acidified supernatant was transferred to a polypropylene tube and neutralized with 650 μl of 1 M K₂CO₃ plus 5 mM EDTA. The samples were kept on ice overnight. Total inositol phosphates were purified by gravity-fed column chromatography using AG 1-X8 200-400 μm mesh, formate form (Bio-Rad), and quantified by liquid scintillation counting (Shears, 1997). The expression of R-Ras and PLC constructs were determined in parallel plates without radioactivity by western blotting and were not significantly different within each experiment (Clark et al., 1996).

MCF10A cells were detached with TPCK trypsin (1 mg/ml in PBS (Sigma, St Louis, MO). Trypsin was neutralized with soy bean trypsin inhibitor (1 mg/ml in PBS) (Sigma) and loaded with 2 μg/ml of indo1-AM (Molecular Probes) for 20 minutes at 37°C with occasional shaking. Intracellular Ca²⁺ concentrations were determined by flow cytometry following stimulation for 6-10 minutes with 4 μM thapsigargin (Sigma), 1 μM CaCl₂ or 60 μM ionomycin (Calbiochem).

The authors thank Grant Kelley for providing antibodies and reagents for PLCϵ, David Inman for expert technical assistance, Chunhua Song for helpful discussion, and Erik Peden, Scott Gehler and Matt Conklin for comments on the manuscript. This work was supported by grants from the NIH/NCI (CA076537) and the Milwaukee Research Foundation (both to P.J.K.).