Rac activation by integrins is essential for cell spreading, migration, growth and survival. Based mainly on overexpression of dominant-negative mutants, RhoG has been proposed to mediate integrin-dependent Rac activation upstream of ELMO and Dock180. RhoG-knockout mice, however, display no significant developmental or functional abnormalities. To clarify the role of RhoG in integrin-mediated signaling, we developed a RhoG-specific antibody, which, together with shRNA-mediated knockdown, allowed analysis of the endogenous protein. Despite dramatic effects of dominant-negative constructs, nearly complete RhoG depletion did not substantially inhibit cell adhesion, spreading, migration or Rac activation. Additionally, RhoG was not detectably activated by adhesion to fibronectin. Using Rac1–/– cells, we found that constitutively active RhoG induced membrane ruffling via both Rac-dependent and -independent pathways. Additionally, endogenous RhoG was important for Rac-independent cell migration. However, RhoG did not significantly contribute to cell spreading even in these cells. These data therefore clarify the role of RhoG in integrin signaling and cell motility.
Rho family small GTPases are crucial mediators of both extracellular and internal cues on actin assembly and organization, cell cycle progression and survival (Bustelo et al., 2007; Fritz and Kaina, 2006; Hall, 2005; Nakaya et al., 2006). These GTPases function as molecular switches that cycle between GDP- and GTP-bound states. They are activated by guanine nucleotide exchange factors (GEFs), which displace GDP from the nucleotide-binding site to allow GTP binding (Meller et al., 2005; Rossman et al., 2005). Active, GTP-bound Rho GTPases bind effector proteins to trigger actin assembly and other processes (Bishop and Hall, 2000). Three Rho family members, Rac1, Cdc42 and RhoA, have been extensively characterized as regulators of cytoskeletal organization and cell function; the roles of other Rho proteins in actin regulation are less well understood.
RhoG (Arhg) is a Rho family GTPase most closely related to Rac1, Rac2 and Rac3. Overexpression of constitutively active RhoG induces dorsal and peripheral membrane ruffles (Gauthier-Rouviere et al., 1998). These processes are inhibited by dominant-negative (DN) Rac1, suggesting that RhoG regulates the actin cytoskeleton upstream of Rac (Blangy et al., 2000; Gauthier-Rouviere et al., 1998). Another study, however, has suggested that RhoG might regulate actin independently of Rac, because short-time treatment with a PAK-binding domain (PBD)-containing peptide that inhibits Rac did not prevent RhoG-induced ruffles (Wennerberg et al., 2002).
Further insight into mechanisms of RhoG action came from the finding that it can directly bind and activate ELMO, a partner for the unconventional Rac GEF Dock180 (Katoh and Negishi, 2003). This study provided evidence that constitutively active RhoG (RhoG G12V) stimulated endogenous Rac through ELMO and Dock180, leading to morphology changes such as ruffling in HeLa cells and neurite outgrowth in PC12 cells. The results suggested that the ELMO-Dock180-Rac pathway mediates the effects of active RhoG on the actin cytoskeleton. This study also reported that DN mutants of RhoG, ELMO and Dock180 inhibited Rac activation and cell spreading induced by adhesion to fibronectin. These data led authors to conclude that RhoG is crucial for integrin-mediated Rac activation and cell spreading. They also reported that a HeLa cell clone in which RhoG was suppressed by expression of short hairpin (sh)RNA exhibited slower cell migration and decreased spreading on fibronectin (Katoh et al., 2006).
Most of the components that are crucial for integrin-mediated spreading and cell migration, including Rac1, RhoA, Cdc42, paxillin, focal adhesion kinase and vinculin, are early embryonic lethal when deleted in mice (Chen et al., 2000; Hagel et al., 2002; Ilic et al., 1995; Sugihara et al., 1998; Wang and Zheng, 2007; Xu et al., 1998). However, RhoG-knockout mice were born at a frequency consistent with Mendelian inheritance, appeared healthy and reproduced normally (Vigorito et al., 2004), indicating that RhoG is dispensable for normal development and physiology in mice. Thus, there is a discrepancy between in vitro studies, in which RhoG was proposed to be an important mediator of integrin function, and analyses in vivo.
Recent studies suggest that DN mutants of Rho GTPases have effects that are more severe than those of gene deletion. For example, complete deletion of Rac1 in conditional-knockout fibroblasts, which do not express Rac2 and Rac3, only moderately inhibited cell spreading and migration (Vidali et al., 2006). Double knockout of Rac1 and Rac2 in macrophages that do not express Rac3 did not prevent migration or chemotaxis (Wheeler et al., 2006). These data suggest that some of the previously reported effects of DN mutants of Rho family members might be due to inhibition of other closely related Rho family members.
These results prompted us to address the function of RhoG using shRNA-mediated knockdowns, together with the development of an antibody that allows analysis of endogenous RhoG. Our data indicate that RhoG is not a crucial intermediate in integrin-dependent cell spreading and signaling. However, results with conditional Rac-knockout fibroblasts show that RhoG contributes to membrane ruffling and cell migration in the absence of Rac.
DN RhoG and ELMO inhibit spreading of MEF cells
To confirm published reports that RhoG inhibitory constructs block cell spreading on fibronectin, we transiently transfected normal mouse embryo fibroblasts (MEFs) with GFP-tagged DN RhoG (GFP-RhoG-T17N) and GFP-ELMOT625. ELMOT625 is a fragment that lacks the C-terminal region required for binding Dock180 (deBakker et al., 2004). Both constructs dramatically reduced spreading of MEFs on fibronectin, as measured by the area of phalloidin-stained GFP-positive cells at different times after the replating, whereas expression of GFP alone had no effect (Fig. 1). Expression of wild-type (WT) RhoG or WT ELMO did not inhibit cell spreading (our unpublished data).
Development of a RhoG-specific monoclonal antibody
Because no antibody was available that specifically recognizes endogenous RhoG, we raised a monoclonal antibody against the C-terminal hypervariable region of RhoG (aa161-180). Because of sequence similarity to Rac, the clones were screened against bacterially purified GST-RhoG and GST-Rac1 by ELISA. One RhoG-specific clone was identified. We confirmed that the antibody did not recognize bacterially purified Rac1, RhoA or Cdc42 in a dot-blot assay (data not shown). On western blots, the antibody recognized a single protein at around 18 kDa (Fig. 2). As shown below, this band was reduced by RhoG shRNA. The antibody also recognized exogenously expressed GFP-RhoG but not GFP-Rac1 or Cdc42.
Knockdown of endogenous RhoG does not inhibit cell spreading
Next, we generated an shRNA expression vector that targets mouse RhoG (pSuperRhoG). The targeted sequence is unique to RhoG, as determined by BLAST searching of the GenBank database. Transient transfection of MEFs resulted in approximately 93% knockdown of RhoG without any effect on Rac1 (Fig. 3A). To evaluate the requirement of RhoG in cell spreading, MEFs were co-transfected with either pSuper, pSuperRhoG shRNA or pSuperRac shRNA, together with a small amount of GFP plasmid to mark transfected cells (GFP vector:shRNA plasmid, 1:40). After 72 hours, the cells were detached and replated on fibronectin. Expression of GFP had no effect on cell spreading compared to untransfected cells (data not shown). Surprisingly, RhoG knockdown also had little effect, whereas knockdown of Rac substantially inhibited the rate of cell spreading (Fig. 3A). Similar results were obtained with NIH3T3 cells (data not shown). The transfection efficiency in these experiments, determined using high levels of GFP vector, was at most 94-96%. Because the efficiency of RhoG knockdown was 93%, the main source of the remaining RhoG protein must be from non-transfected cells. To visualize cells with the highest knockdown, we co-transfected GFP at a low ratio of GFP vector to shRNA plasmid (1:40), leading to detectable expression of GFP in only 40-45% of the cells. RhoG depletion in GFP-positive cells should therefore be nearly complete.
Because Dock180 was proposed to mediate adhesion-dependent Rac activation downstream of RhoG (Katoh et al., 2006; Katoh and Negishi, 2003), we also knocked down this protein. Dock180 shRNA inhibited cell spreading less efficiently than Rac shRNA, but, unlike RhoG, inhibition of spreading was statistically significant (Fig. 3B). These results suggest that the strong inhibition of spreading by RhoG-T17N and ELMOT625 mutants is non-specific and that RhoG is not essential for spreading of MEFs.
A previous study reported that a HeLa-derived clonal cell line in which RhoG was knocked down spread slowly and migrated poorly (Katoh et al., 2006). To test whether differences in cell type account for the discrepancy, we carried out transient RhoG knockdowns in HeLa cells. An shRNA targeting human RhoG caused loss of 82% of endogenous RhoG in transiently transfected HeLa cells. Although parallel knockdown of Rac strongly inhibited HeLa cell spreading, knockdown of RhoG had no effect (Fig. 3C). Therefore, HeLa cells do not appear to differ from MEFs under these conditions.
We also considered the possibility that RhoG might have effects on cell adhesion to the extracellular matrix that could account for the published results. We therefore measured cell adhesion to fibronectin over a range of concentrations. Cell attachment was dependent on the concentration of fibronectin (Fig. 3D), but RhoG knockdown again had no detectable effect.
Knockdown of RhoG does not substantially inhibit cell migration
We next tested effects of RhoG knockdown on cell migration using time-lapse imaging of randomly migrating MEFs. pSuper or the shRNA vectors were co-transfected with GFP as described above, movies recorded and the velocity and directionality of GFP-positive cells were analyzed. Knockdown of Rac mildly but significantly decreased the velocity of random cell migration (Fig. 4A). Knockdown of RhoG, however, failed to cause a statistically significant change in migration velocity. Migration directionality was not substantially affected by Rac or RhoG knockdowns (Fig. 4B).
Role of RhoG in Rac activation by integrins
Rac is activated upon cell adhesion to fibronectin (del Pozo et al., 2000; Price et al., 1998). Several Rac GEFs, including Dock180, VAV2 and βPIX, have been implicated in Rac activation by integrins (Jones and Katan, 2007; Kiyokawa et al., 1998; Marignani and Carpenter, 2001; ten Klooster et al., 2006). To study the role of RhoG in this pathway, we measured Rac activation in RhoG-knockdown cells after adhesion to fibronectin. Serum-starved cells were kept in suspension for 3 hours and then replated on fibronectin-coated plates for 10 minutes, the point of maximal Rac activity. Replating of control cells stimulated Rac activity approximately 2.5-fold (Fig. 5A). RhoG knockdown decreased baseline Rac activity in suspended cells by 40% but did not prevent Rac activation upon replating. After subtracting the baseline activity in suspended cells, RhoG knockdown inhibited adhesion-dependent Rac stimulation by only 9%, an effect that was not statistically significant (P=0.46). Thus, RhoG mediates some integrin-independent pathways of Rac activation but does not contribute significantly to the integrin-dependent increment.
As a control, we investigated the involvement of Dock180. Although Dock180 knockdown was less efficient than RhoG, it lowered the baseline Rac activation in suspended cells by 55% and inhibited the adhesion-dependent increase by 60% (Fig. 5B). These data suggest that Dock180 makes a major contribution to Rac activation by integrins in mouse fibroblasts, whereas the contribution of RhoG is not significant. However, RhoG and Dock180 might both be involved in integrin-independent Rac activation in suspended cells.
Activation of endogenous RhoG
To test whether the activity of endogenous RhoG is regulated by adhesion, we established a RhoG pulldown assay using GST-ELMO protein on agarose beads as the effector domain. GTPγS and GDP loading of endogenous RhoG in cell lysates was used to develop optimal conditions for the pulldown assay. GTPγS-loaded RhoG specifically bound to GST-ELMO (Fig. 6A). Specific binding of active RhoG was also confirmed using overexpressed active and negative mutants (our unpublished data). As a positive control for activation of endogenous RhoG, we expressed a fragment of Trio that contained the GEF1 domain specific for RhoG and Rac (Estrach et al., 2002). Expression of the Trio GEF1 domain activated endogenous RhoG by 8.5-fold (Fig. 6B). However, when RhoG activity was measured upon replating on fibronectin, no activation was detected at any time point (Fig. 6B). By contrast, Rac activity was consistently elevated approximately threefold in the same time period after replating (our unpublished data). We conclude that adhesion to fibronectin does not activate RhoG under these conditions.
RhoG affects actin through Rac-dependent and -independent pathways
Depletion of Rac1 by RNAi (this study) or by conditional knockout (Vidali et al., 2006; Wheeler et al., 2006) does not completely abolish cell migration, suggesting that other Rho family proteins can substitute for Rac in this process. Because RhoG is the closest Rac homolog expressed in fibroblasts, we considered that RhoG might drive cell migration independently of Rac. To test this idea, we used fibroblasts from conditional Rac-knockout mice (Vidali et al., 2006). Treatment of floxed Rac1 fibroblasts with Cre adenovirus caused a 98% reduction in Rac1 protein after 6 days (Fig. 7A).
As a first step, GFP–RhoG-Q61L was expressed in Rac1–/– cells or control cells, and the induction of peripheral and dorsal ruffles was analyzed by phalloidin staining. In control and Rac-deficient cells expressing GFP alone, baseline ruffling was, respectively, either very low or virtually undetectable (our unpublished data). When membrane ruffling was quantified in cells showing low, medium or high RhoG-Q61L expression using GFP as a marker, high levels of RhoG-Q61L induced ruffles in both control and Rac-deficient cells (Fig. 7B,C). By contrast, in cells with lower RhoG-Q61L expression, ruffling was induced in control but very poorly in Rac1–/– cells. These results suggest that RhoG-Q61L affects the actin cytoskeleton by at least two distinct pathways: a Rac-dependent pathway that functions at low RhoG levels and a Rac-independent pathway that requires higher levels of RhoG activation or expression.
Because of the sequence homology between RhoG and Rac, we considered that RhoG might trigger Rac-independent actin polymerization through a known Rac effector. The Rac effectors PAK and IRSp53 can mediate effects on actin but were reported to not bind RhoG (Wennerberg et al., 2002). We therefore tested whether SRA1, which mediates the activation of the WAVE complex by Rac, can interact with GDP- or GTPγS-bound RhoG (supplementary material Fig. S1). These assays detected only weak and nucleotide-independent binding of SRA1 to RhoG. Positive controls showed strong, nucleotide-dependent binding of both SRA1 to Rac and RhoG to ELMO1. Thus, Rac-independent RhoG-induced actin rearrangements do not occur through known Rac-effector pathways.
RhoG mediates Rac1-independent migration
Because we found that RhoG can induce actin rearrangement in the absence of Rac, we next tested whether this pathway is relevant to cell spreading and migration. For these experiments, we combined the Rac1 knockout with knockdown of RhoG. Initial experiments showed that deletion of Rac1 greatly reduced the efficiency of transfection with RhoG shRNA vectors or oligonucleotides (our unpublished data). We therefore generated stable RhoG shRNA clones derived from the floxed Rac MEFs. Two clones (3 and 6), which showed 97% and 91% depletion of RhoG, respectively, were isolated. The parental floxed Rac1 MEF cells and a line expressing only pSuper were used as controls. These cells were then infected with the GFP Cre adenovirus to delete Rac1. GFP-only adenovirus was used as a control. This protocol generated cells in which both Rac1 and RhoG were suppressed either individually or together (Fig. 8A).
When random cell migration was assayed, deletion of Rac1 in parental cells reduced migration speed by 32% (Fig. 8B). Migration speed in RhoG-deficient clones treated with the control virus was comparable to that of parental cells, consistent with our data from transient RhoG knockdown in Fig. 4A. However, knockout of Rac1 in RhoG-deficient clones reduced velocity by nearly 70% in both clones. These results suggest that RhoG is important for migration in the absence of Rac. To confirm the specificity of RhoG-shRNA effects, we performed a rescue experiment with shRNA-resistant human GFP-RhoG. Expression of human RhoG in RhoG- and Rac-deficient MEFs increased migration velocity nearly to the level of control cells, whereas expression of control vector alone had no effect (Fig. 8C). These results show that RhoG makes an important contribution to cell migration in the absence of Rac.
Effects of Rac1 and RhoG on cell spreading were also analyzed in these cells. Because the total cell size varies somewhat between clones, spread area was normalized to cell size in completely round cells. As previously reported (Vidali et al., 2006), deletion of Rac1 substantially decreased the rate and extent of spreading (Fig. 8D). Knockdown of RhoG, however, had little effect in either control or Rac1–/– cells. Thus, even in the absence of Rac1, RhoG does not detectably contribute to integrin-mediated cell spreading.
It is well accepted that integrin-derived signals regulate many cellular processes, including growth, survival and cytoskeletal dynamics (Cordes, 2006; Delon and Brown, 2007; Reddig and Juliano, 2005; Schwartz and Assoian, 2001). Rho GTPases are important components of this signaling, and regulate actin polymerization and organization through multiple pathways. These pathways include activation of the Arp2/3 complex by Rac and Cdc42, activation of formin proteins through Rho, and activation of myosin phosphorylation through ROCK and Rho. Rac was proposed to be the main regulator of lamellipodial protrusions that mediate cell migration and spreading (Hall, 2005); Rac is also important for cell proliferation and resistance to apoptosis (Ruggieri et al., 2001; Vidali et al., 2006). The mechanism by which integrins trigger Rac activation is of great interest because of its relevance to abnormal cell migration and survival in pathologies, such as cancer, and developmental and immune disorders.
Recent publications suggested that RhoG is a crucial mediator of integrin signaling to Rac during cell spreading and migration. This conclusion was based on dramatic inhibition of adhesion-induced cell spreading and migration by DN mutants of RhoG and its downstream partner ELMO, and on the behavior of a RhoG-knockdown clone in HeLa cells. By contrast, we found that efficient knockdown of RhoG did not substantially affect cell spreading or migration. We also found that RhoG was not detectably activated by adhesion to fibronectin and did not contribute significantly to activation of Rac1 during cell adhesion to fibronectin.
The inconsistencies between the effects of shRNA-mediated RhoG depletion and effects of DN RhoG mutants are most likely to be due to non-specific effects of the DN mutants. DN mutants of Rho proteins such as RhoG-T17N might bind and sequester multifunctional GEFs that normally interact with other GTPases. Indeed, Trio, Kallirin, VAV1, VAV2, VAV3 and PLEKHG6 are all able to interact with both Rac and RhoG (Abe et al., 2000; Blangy et al., 2000; D'Angelo et al., 2007; Movilla and Bustelo, 1999; Rabiner et al., 2005; Schuebel et al., 1998; Tybulewicz et al., 2003; Wennerberg et al., 2002). Furthermore, the observation that the effects of constitutively active RhoG-Q61L were reduced by RhoG-T17N (Wennerberg et al., 2002) supports the notion that this DN has broader specificity. Similar findings of non-specific effects by DN constructs have been reported for other Rho family GTPases (Wang and Zheng, 2007). Similarly, overexpressed fragments of ELMO might interact with other target proteins.
The reason for reduced spreading and migration in the stable RhoG-knockdown HeLa clone (Katoh et al., 2006) is less clear. We noticed, however, that HeLa cells are remarkably heterogeneous in their rate of spreading on fibronectin. It is therefore not surprising that individual clones would spread at different rates. Testing additional RhoG-deficient clones would help to resolve the discrepancy.
Whereas physiological signals upstream of RhoG are only beginning to be identified (see below), the effects of its active mutant on the actin cytoskeleton are well documented (Gauthier-Rouviere et al., 1998; Katoh et al., 2006; Katoh and Negishi, 2003; Katoh et al., 2000). There is, however, a disagreement regarding the mechanism of actin regulation by RhoG. Whereas some studies indicated that active RhoG induced the formation of ruffles via Rac (Gauthier-Rouviere et al., 1998; Katoh and Negishi, 2003), others suggested a parallel, Rac-independent pathway (Prieto-Sanchez and Bustelo, 2003; Wennerberg et al., 2002). These investigations tested the effects of DN Rac on the phenotype induced by active RhoG. To address these issues, we used conditional Rac-knockout cells, an approach that should be more specific than overexpression of DN constructs (Vidali et al., 2006). We found that RhoG can affect the cytoskeleton through both Rac-dependent and -independent pathways. The Rac-dependent pathway was evident at low expression of active RhoG, whereas the Rac-independent pathway required higher expression. The functionality of a Rac-independent pathway for induction of actin rearrangements by RhoG is supported by a study in Salmonella Typhimurium (Patel and Galán, 2006). Salmonella independently activates Rac through bacterial GEF SopE and RhoG through the bacterial protein SopB. RhoG activation by SopB is mediated by SGEF, a host GEF for RhoG. Although a SopE-deficient Salmonella strain could not activate Rac, it was still able to induce SopB-dependent actin rearrangements leading to SGEF- and RhoG-mediated bacterial uptake. Furthermore, SopB could function also in the absence of Rac, suggesting that RhoG can contribute to bacterial uptake independently of Rac.
However, the mechanism of RhoG-induced actin rearrangements in the absence of Rac is still not clear. Rac induces actin polymerization by activation of the WAVE complex, which in turn leads to Arp2/3-complex activation (Ridley et al., 2003). The Rac effectors IRSp53 and SRA1 and its homolog PIR121 (CYFIP2) have been reported to mediate the activation of the WAVE complex by Rac (Ridley, 2006). Rac can also affect actin polymerization through its effector PAK (Jaffe and Hall, 2005). In spite of sequence similarity between RhoG and Rac, neither IRSp53 nor PAK was found to interact with RhoG (Wennerberg et al., 2002). We considered the possibility that SRA1 might mediate the effect of RhoG on actin. However, we found that, in contrast to Rac, only weak, nucleotide-independent binding of RhoG to SRA1 could be detected. Thus, despite their close homology, RhoG does not appear to affect actin through known Rac effectors. The mechanism of Rac-independent actin regulation by RhoG will require further investigation.
In contrast to conclusions based on DN Rac mutants, conditional knockout of Rac1 in fibroblasts (Vidali et al., 2006) and double-knockout of Rac1 and Rac2 in macrophages (Wheeler et al., 2006) showed that Rac is not absolutely essential for migration. Our data demonstrate that RhoG is an important mediator of this Rac-independent cell migration in fibroblasts. It therefore seems plausible that RhoG might mediate migration in response to specific stimuli or under specific conditions; however, its physiological role in migration remains unknown. We did find that serum stimulation induced mild and transient activation of RhoG (our unpublished data); however, the effects were weak enough that it was difficult to obtain a statistically significant effect. RhoG might also be regulated by signals derived from cellular events involved in cytokinesis or membrane trafficking.
While this work was in progress, publications suggested that RhoG participates in phagocytosis (deBakker et al., 2004; Nakaya et al., 2006), endothelial cup formation (van Buul et al., 2007) and macropinocytosis (D'Angelo et al., 2007; Ellerbroek et al., 2004). It is interesting that these functions, at least in culture, involve localized actin rearrangements at the dorsal membrane. A restricted spectrum of RhoG-dependent functions is more consistent with the lack of developmental abnormalities in RhoG-knockout mice, as opposed to the model claiming a non-redundant role of RhoG for developmentally essential processes such as integrin-mediated signaling and migration.
Materials and Methods
The shRNAs for RhoG, Dock180 and Rac1 were expressed using the pSuper vector (OligoEngine, Seattle, WA). The RNAi vectors targeting RhoG and Dock180 were designed and constructed according to the manufacturer's guidelines. Three to five shRNA sequences for each protein were tested by western blot and the most effective sequences were subsequently used. Mouse RhoG RNAi targets nucleotides 348-367 of RhoG mRNA (5′-CGTCTTCGTCATCTGTTTC-3′). For generation of stable RhoG-knockdown MEF cell lines, this sequence was subcloned into pSuper Retro Puro vector. Human RhoG RNAi targets nucleotides 484-502 (5′-CAGGATGGTGTCAAGGAAG-3′). Mouse Dock180 RNAi targets nucleotides 1525-1543 (5′-GCGATTGGAGCACGTGATT-3′). The RNAi vector for human and mouse Rac1 was constructed using the previously reported sequence targeting nucleotides 552-570 (5′-GAGGAAGAGAAAAUGCCUG-3′) (Noritake et al., 2004).
ELMOT625 and GST-ELMO1 constructs, and human GFP-RhoG WT, Q61L and T17N constructs were obtained from Kodi Ravichandran (deBakker et al., 2004). GST-Rac1 has been previously described (del Pozo et al., 2000). GST-RhoG was generated by subcloning WT RhoG into the pGEX-4T vector using BamH1 and EcoR1 restriction sites. A fragment containing the first GEF domain of Trio was constructed in pEF flag vector (Meller et al., 2004) by amplification of the corresponding sequence from full-length human Trio with the following primers: forward 5′-ATTGGTACCCGGGAGAACAGGGTATTGC-3′ and reverse 5′-GAAGAATTCTTAGCTTTTCACGAGTTCCTCAATTG-3′.
Cell culture and transfections
MEFs and HeLa cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum, 4 mM glutamine, penicillin and streptomycin at 37°C with 5% CO2. Cells were transiently transfected using a BioRad Gene Pulser electroporator and 4 mm electroporation cuvettes. MEFs were electroporated at 300 V and 550 μF; HeLa cells were electroporated at 250 V and 900 μF. 6×106 cells were suspended in 400 μl of PBS. For knockdowns, 40 μg of shRNA plasmid and 1 μg of pMaxGFP (Amaxa, Koln, Germany) were used. Cells were assayed 72 hours after transfection. For expression of GFP-RhoG, GFP-Rac1, GFP-ELMO1 or GFP alone, cells were electroporated with 5 μg of the corresponding plasmid with addition of 35 μg of salmon sperm DNA and assayed 24-48 hours after transfection.
MEFs in which the first exon of Rac1 was flanked by loxP sites from conditional Rac1-knockout mice were obtained from the laboratory of David Kwiatkowski (Vidali et al., 2006). To delete Rac1, Cre was introduced by infection with 6×103 PFU/cell of Cre-GFP or GFP-only adenovirus (University of Iowa Gene Transfer Vector Core Facility, Iowa City, IA). Cells were assayed 6 days later and knockout efficiency was confirmed for each experiment. For expression of GFP-Rac1 or GFP-RhoG after Rac1 knockout, cells 5 days after infection with Cre-GFP or GFP were transfected with the corresponding plasmids using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). The transfections were done according to the manufacturer's instructions and cells were used 24 hours later.
To establish RhoG-deficient stable clones, Rac1 conditional-knockout fibroblasts were transfected with pSuper Retro Puro RhoG-shRNA or pSuper Retro Puro alone, then subjected to puromycin selection (3 μg/ml). Single clones were isolated using colony rings (VWR) and tested for RhoG expression. Two clones with the most-efficient RhoG knockdown were used in the study.
Antibodies and western blot analysis
The monoclonal antibody to RhoG was raised against a peptide CQQDGVKEVFAEAVRAVLNPT (RhoG aa161-180 plus an N-terminal cysteine) at the UVA hybridoma facility. The peptide conjugated to KLH via the cysteine was injected into A/J mice. After two immunizations, the sera were tested by ELISA using bacterially purified RhoG. The mouse with the highest titer was then immunized by intra-splenic injection, and 4 days later the spleen cells were collected and fused with the Sp2/0 myeloma cell line. The culture supernatant of the resulting hybrid cells was screened by ELISA against bacterially purified RhoG and Rac proteins. The RhoG-specific clone (1F3 B3 E5) was selected for further use. The antibody was purified using protein G.
Rac1 monoclonal antibody was from Upstate Biotechnology (Lake Placid, NY) and Dock180 polyclonal antibody was from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal anti-tubulin, clone E7 from the Developmental Studies Hybridoma Bank (Iowa City, IA), was grown in the UVA hybridoma facility. Anti-mouse HRP-conjugated secondary antibody was from the Jackson Laboratory (Bar Harbor, ME) and anti-goat HRP-conjugated secondary antibody was from Santa Cruz.
For western blot analysis, cells were lysed in buffer containing 1% Triton, 150 mM NaCl, 50 mM Tris pH 7.4 and protease inhibitor cocktail (Sigma-Aldrich, St Louis, MO). The lysates were clarified by centrifugation at 1.6×104 g for 10 minutes at 4°C and protein concentration in the supernatants determined using BioRad protein assay reagent (BioRad Laboratories, Hercules, CA). For the analysis of RhoG expression, 30 μg of cell lysate was separated by SDS PAGE on 13% gels. Lysates (20 and 100 μg, respectively) were used for Rac and Dock180 immunoblots. Proteins were transferred to PVDF membrane (Millipore, Bedford, MA) and analyzed by immunoblotting.
RhoG and Rac pulldown assays
To measure Rac and RhoG activity, cells were serum-starved overnight and detached with trypsin. Trypsin was inactivated with soybean trypsin inhibitor (Sigma) and the cells kept in suspension for 1 hour in serum-free medium containing 0.5% BSA and 0.4% methyl cellulose (to reduce clumping). Cells in suspension were sedimented at low speed, washed in ice-cold PBS and frozen at –80°C. The rest of the cells were replated in serum-free medium on dishes coated with fibronectin at 10 μg/ml. At the indicated times, cells were rinsed twice in ice cold PBS and frozen at –80°C.
Rac activity was determined using the pulldown assay with the PAK-binding domain (PBD) fused to GST, as described previously (del Pozo et al., 2000). RhoG activity was determined using a pulldown assay with GST-ELMO1 expressed in DH5α Escherichia coli strain. Bacterial pellets were suspended in buffer containing 1% Triton, 150 mM NaCl and 50 mM Tris pH 7.4, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and broken by sonication. After centrifugation at 1.6×104 g for 30 minutes at 4°C, the supernatants were supplemented with 20% glycerol, aliquoted and kept at –80°C. GST-ELMO was purified by incubation with glutathione beads (GE Healthcare Bio-Sciences, Uppsala, Sweden) for 1 hour followed by three washes in cold lysis buffer. Cell lysates were incubated with GST-ELMO beads for 40 minutes at 4°C, washed three times with lysis buffer and bound proteins were eluted with Laemmli buffer. The eluates and the total cell lysates were separated by SDS-PAGE and probed for RhoG.
For nucleotide loading of endogenous RhoG, cell lysates were incubated with 2 mM EDTA and 20 μM GTPγS or GDP for 5 minutes at 30°C. MgCl2 was then added to a final concentration of 10 mM and RhoG activity was assayed as described above.
Cell spreading, adhesion, ruffle quantification and migration
For spreading assays, cells were starved in serum-free DMEM containing 0.5% BSA for 3 hours and detached with trypsin. The trypsin was inactivated with soybean trypsin inhibitor, and the cells were washed and kept in suspension for 30 minutes in serum-free medium. Cells were then replated in serum-free medium on fibronectin-coated coverslips. At the indicated times, cells were fixed with 2% formaldehyde, permeabilized with 0.2% Triton X-100 and stained with Alexa-Fluor-568-conjugated phalloidin. Digital fluorescent images of randomly selected fields were acquired with Inovision software using a Nikon Diaphot microscope with 10× objective and CoolSNAP HQ camera (Roper Scientific, Tucson, AZ). Spread area was measured using the threshold function of ImageJ software (NIH).
For quantification of membrane ruffles, cells transfected with GFP–RhoG-Q61L or GFP alone for 24 hours were replated on slides coated with fibronectin (2 μg/ml). After 4 hours, they were fixed and stained with Alexa-Fluor-568–phalloidin. GFP and phalloidin fluorescent images were acquired using a 60× oil objective. The same settings were used for acquisition of all of the GFP images. Ruffles were identified by phalloidin staining that colocalized with GFP–RhoG-Q61L at the cell periphery or dorsal surface. To quantify the area of ruffles, we used the fact that RhoG-GFP fluorescence was significantly brighter in the ruffles than in the rest of the cell periphery. Fluorescence intensity from GFP-RhoG was thresholded and the area of each ruffle quantified using the shareholding function of ImageJ software. Perinuclear fluorescence of GFP-RhoG did not colocalize with phalloidin staining and, therefore, was not included in ruffle area. Total GFP–RhoG-Q61L fluorescence intensity for each cell was also recorded.
For adhesion assays, cells were plated for 10 minutes in wells coated with the indicated concentrations of fibronectin or poly-L-lysine (pLL). The wells then were gently washed with PBS and cells were stained with 0.25% crystal violet solution for 10 minutes, washed three times with water and allowed to dry. Crystal violet was eluted with 33% glacial acid and the OD was measured at 570 nm. The values were translated into cell number using calibration standards.
For random-migration assays, cells were plated in tissue-culture plastic dishes coated with 2 μg/ml fibronectin. After 24 hours, the cells were placed on a humidified microscope stage with 5% CO2 at 37°C in culture medium and phase-contrast images taken every 5 minutes for 10 hours. Cell migration was analyzed using ImageJ software.
We thank Kodi Ravichandran for providing ELMO constructs and helpful suggestions; Amy Bouton, Alan Rick Horwitz, Dorothy Schafer and Nahum Meller for helpful discussions; and Kostas Moissoglu, Nagaraj Balasubramanian, Brenton Hoffman and Brian Hall for assistance with microscopy and analysis. This work was supported by National Institutes of Health grant RO1 GM47214.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/121/12/1981/DC1
- Accepted March 29, 2008.
- © The Company of Biologists Limited 2008