Roles of the Rac1 and Rac3 Gtpases in Human Tumor Cell Invasion

Amanda Y Chan1, Salvatore J Coniglio2, Ya-yu Chuang1, David Michaelson3,4, Ulla GKnaus 5, Mark R Philips3,4 and Marc Symons*,1,2,6

1Institute for Medical Research at North Shore-LIJ, 350 Community Drive, Manhasset, NY 11030, USA; 2Department of Anatomy and Structural Biology, Albert Einstein College of Medicine, Bronx, NY 10461, USA; 3Department of Medicine, New York University School of Medicine, New York, NY 10016, USA; 4Department of Cell Biology, New York University School of Medicine, New York, NY 10016, USA; 5Department of Immunology, The Scripps Research Institute, La Jolla, CA 92037, USA; 6Department of Surgery, North Shore-Long Island Jewish Research Institute, Manhasset, NY 11030, USA

Members of the Rho familyof small GTPases have been Burridge and Wennerberg, 2004). Importantly, work shown to be involved in tumorigenesis and metastasis. from a large number of laboratories has shown that the Currently, most of the available information on the function Rho family members Rac, Cdc42 and Rho are essential of Rho proteins in malignant transformation is based on the for cell transformation (Khosravi-Far et al., 1995; Qiu use of dominant-negative mutants of these GTPases. The et al., 1995a, b, 1997; Lin et al., 1997), tumor cell specificityof these dominant-negative mutants is limited invasion (Keely et al., 1997; Shaw et al., 1997) and however. In this study, we used small interfering RNA metastasis (Bouzahzah et al., 2001); reviewed in Sahai E directed against either Rac1 or Rac3 to reduce their and Marshall (2002). expression specifically. In line with observations using Rho GTPases essentially function as switches, they dominant-negative Rac1 in other cell types, we show that are ‘on’ in the GTP-bound state and ‘off’ in the GDP- RNA interference-mediated depletion of Rac1 strongly bound state. The nucleotide state of Rho proteins is inhibits lamellipodia formation, cell migration and invasion controlled by three classes of regulatory proteins in SNB19 glioblastoma cells. Surprisinglyhowever, Rac1 (Symons and Settleman, 2000). Guanine nucleotide depletion has a much smaller inhibitoryeffect on SNB19 exchange factors (GEFs) catalyse the exchange of cell proliferation and survival. Interestingly, whereas GDP for GTP, thereby activating the protein. GTPase depletion of Rac3 stronglyinhibits SNB19 cell invasion, activating proteins (GAPs) promote the intrinsic GTP it does not affect lamellipodia formation and has only hydrolysing activity of Rho proteins, leading to their minor effects on cell migration and proliferation. Similar inactivation. Guanine nucleotide dissociation inhibitors results were obtained in BT549 breast carcinoma cells. (GDIs) preferentially bind to GDP-bound GTPases and Thus, functional analysis of Rac1 and Rac3 using RNA prevent spontaneous and GEF-catalysed release of interference reveals a critical role for these GTPases in the nucleotide, thereby maintaining the GTPases in the invasive behavior of glioma and breast carcinoma cells. inactive state. Oncogene (2005) 24, 7821–7829. doi:10.1038/sj.onc.1208909; There are three Rac proteins in the human genome: published online 18 July 2005 Rac1, 2 and 3. These proteins are highly homologous and differ in their transcriptional regulation and tissue Keywords: Rac1; Rac3; glioma; invasion; migration; distribution. Rac1 is ubiquitously expressed, Rac2 is RNA interference hematopoietically specific, whereas Rac3 is highly enriched in the brain and expressed at lower levels in a wide range of tissues (Haataja et al., 1997). Thus far, studies aimed at determining the functions Introduction of Rho proteins have largely relied on the use of dominant-negative versions of these GTPases. These are The Rho family constitutes a subgroup of the Ras point mutants that have low affinity for guanine superfamily of small GTPases. In humans, the Rho nucleotides and therefore are nonfunctional, but display family comprises 20 members that regulate a large a high affinity for the respective GEFs (Feig, 1999). variety of different functions, including the organization However, dominant-negative GTPases successfully com- of the actin cytoskeleton, cell migration, mitogenesis pete with their endogenous counterparts and therefore and cell survival (Van Aelst and D’Souza-Schorey, 1997; inhibit their activation. An important caveat of dominant-negative mutants of Rho GTPases is that GEFs often catalyse nucleotide exchange on a number of *Correspondence: M Symons, Center for Oncology and Cell Biology, different Rho family members (Zheng, 2001), thereby North Shore-Long Island Jewish Research Institute, 350 Community Dr., Manhasset, NY 11030, USA; E-mail: [email protected] limiting the specificity of the dominant-negative ver- Received 16 December 2004; revised 2 June 2005; accepted 3 June 2005; sions. For example, expression of the dominant-negative published online 18 July 2005 mutant Rac1-N17 has been shown to interfere with the Roles of Rac GTPases in human tumor cell invasion AY Chan et al 7822 activation of RhoA by the Dbl GEF (Debreceni et al., 2004). The limited specificity of dominant-negatives is particularly problematic for the functional analysis of highly similar GTPases such as Rac1 and Rac3, because we expect that most GEFs would not discriminate between these GTPases (Karnoub et al., 2001). Therefore, to analyse the specific functions of Rac1 and Rac3 in tumor cell growth and invasion, we have used RNA interference (RNAi) (Denli and Hannon, 2003; Dykxhoorn et al., 2003) to inhibit their expression specifically. A recent study has indicated that Rac family proteins play a critical role in the survival of glioma cells (Senger et al., 2002) and we therefore focused on this system. Expression of the dominant-negative Rac1-N17 mutant has been shown to inhibit dramatically the survival of glioma cells, but not of primary astrocytes, the cell type from which glioma originates (Senger et al., 2002). In this study, we used transient transfection of Rac1- and Rac3-directed small interfering RNA (siRNA) to inhibit specifically the expression of these proteins in glioblastoma cells and examined the effects of depleting Rac1 or Rac3 on functions that are thought to be mediated by Rac proteins, that is, lamellipodia formation, cell migration, invasion, survival and proliferation.

Results Figure 1 Characterization of siRNA-mediated depletion of Rac proteins. (a) Rac1 and Rac3 mRNA levels were determined by Characterization of siRNA-mediated depletion of Rac quantitative PCR in SNB19 cells transfected with siRNA targeting luciferase (control) or the indicated siRNAs targeting either Rac1 proteins or Rac3. The mRNA expression level of Rac1 and Rac3 in Rac1- To examine the specificity of RNAi-mediated depletion depleted cells was normalized to the respective Rac1 and Rac3 mRNA expression levels in control cells transfected with siRNA of Rac1 and Rac3 we used SNB19, a grade IV directed against luciferase. Shown is the mean (7range) of two glioblastoma cell line, and determined the effect of independent experiments. (b) Rac protein levels were determined by transfecting siRNA oligonucleotide duplexes directed Western blotting using an antibody that recognizes both Rac1 and against either Rac1 or Rac3 on the mRNA levels of Rac3. Tubulin levels are shown as loading control Rac1 and Rac3 by quantitative PCR. Two independent Rac1-directed siRNA oligonucleotide duplexes, target- and the inhibition is sustained for up to 6 days after ing either the ORF (Rac1–1) or the 30-UTR region transfection (data not shown). The strong inhibition in (Rac1–2), had a similar inhibitory effect on Rac1 total Rac protein levels caused by depletion of Rac1, mRNA expression, with no significant effect on Rac3 together with the observation that depletion of Rac3 has mRNA levels (Figure 1a). Conversely, both Rac3- no significant effect on total Rac protein levels directed siRNA duplexes inhibited Rac3 expression, (Figure 1b), indicates that Rac3 protein levels are but not Rac1. Whereas the siRNA that targets the relatively low. This is supported by the observation that 30-UTR region (Rac3–2) was similar in potency to the the levels of Rac3 mRNA in SNB19 cells, as estimated Rac1-directed siRNAs, the Rac3 ORF-targeted oligo by quantitative PCR, are 20–30-fold lower than those of was less effective. Rac1 (data not shown). Currently, there are no good commercial antibodies available that are specific for either Rac1 or Rac3. We Depletion of Rac1 but not Rac3 inhibits lamellipodia therefore examined the effect of the respective siRNA formation in SNB19 cells duplexes on total Rac protein using an antibody that recognizes both Rac1 and Rac3. Transfection of Rac1- Expression of a dominant-negative mutant of Rac1 has directed siRNA strongly reduced Rac1 protein expres- been shown to inhibit the formation of lamellipodia sion in comparison to luciferase-directed control siRNA induced by a variety of stimuli (Ridley et al., 1992). (Figure 1b). We verified that an additional control oligo Lamellipodia are thin, actin-rich veil-like extensions that with a GC content of 42%, similar to that of Rac1–1 protrude from the cell body and are thought to play a (38% GC) and Rac1–2 (43% GC) does not affect Rac role in cell migration (Small et al., 2002). To determine expression (data not shown). We routinely achieved the roles of Rac1 and Rac3 in lamellipodia formation, 70–80% inhibition in protein expression. Maximum we examined the effects of transfecting control and inhibition is achieved within 3–4 days of transfection Rac1- and Rac3-directed siRNA on serum-induced

Oncogene Roles of Rac GTPases in human tumor cell invasion AY Chan et al 7823 lamellipodia extension in SNB19 cells. Stimulation with formation, as did expression of dominant-negative serum produced numerous lamellipodia in control and Rac1. Moreover, expression of constitutively active Rac3-depleted cells, but failed to do so in Rac1-depleted mutants of either Rac1 or Rac3 strongly stimulated cells (Figure 2a and b), indicating that Rac1, but not lamellipodia formation, as has been observed in Rac3 mediates lamellipodia formation. ﬁbroblasts (Joyce and Cox, 2003). These results indicate Interestingly, although depletion of Rac3 did not that neither the dominant-negative nor constitutively affect lamellipodia formation, transient expression of active versions of Rac3 accurately report on Rac3 dominant-negative Rac3 strongly inhibited lamellipodia function.

Figure 2 Depletion of Rac1 but not Rac3 inhibits serum-induced lamellipodium extension in SNB19 cells. (a) Fluorescence micrographs showing Texas Red-conjugated phalloidin staining of SNB19 cells transfected with siRNA directed against luciferase (control), Rac1 or Rac3 or plasmids expressing Rac1-N17 or Rac3-N17. Transfected cells were serum-starved overnight and stimulated with 5% serum for 5 min. Scale bar represents 10 mm. (b) Percentage of lamellipodia formation per cell was quantiﬁed as described in Materials and methods. Cells transfected with siRNA or with dominant-negative GTPases were treated as in (a). Cells transfected with constitutively active GTPases were serum-starved and not further stimulated. Shown are the mean values (7s.e.m.) for approximately 40 cells per condition

Oncogene Roles of Rac GTPases in human tumor cell invasion AY Chan et al 7824 Roles of Rac1 and Rac3 in SNB19 cell migration and transwell matrigel invasion assay (Repesh, 1989). Inter- invasion estingly, siRNA-mediated depletion of Rac1 or Rac3 resulted in similar inhibition of SNB19 cell invasion, by Since Rac proteins have been implicated in the regula- approximately 80% (Figure 4). The inhibitory effect of et al tion of cell migration (Anand-Apte ., 1997; Allen Rac3 on invasion is proportional to the efficiency of the et al., 1998; Nobes and Hall, 1999), we also investigated oligo. Oligo Rac3–2, which is similar in efficiency to that the effect of depleting Rac1 and Rac3 on cell migration of the two Rac1 oligos (Figure 1) causes a similar extent using a monolayer wound healing assay. Quantification of inhibition, whereas the less efficient oligo Rac3–1 of the rate of cell movement over the first 16 h showed inhibits less well. We also note that the invasion assay that depletion of Rac1 inhibits migration by about 80%, was performed with serum present both in the top and whereas depletion of Rac3 inhibits migration only by bottom wells of the transwell chamber, implying that 20% (Figure 3). depletion of Rac1 or Rac3 inhibits invasion proper and To examine the roles of Rac1 and Rac3 in cell that this inhibition is not caused by a potential invasion, we determined the effect of depleting Rac1 or inhibitory effect of Rac1 or Rac3 depletion on Rac3 on the invasive properties of SNB19 cells using a chemotaxis. These results therefore indicate an important role for both Rac1 and Rac3 in SNB19 cell invasion. We also verified that an additional control oligo (42% GC) does not affect SNB19 cell invasion (data not shown). In addition, we showed that siRNA- mediated depletion of Rac1 strongly inhibits invasive behavior of another grade IV glioblastoma cell line, U87MG(Supplementary Information, Figure S1B).

Effects of depletion of Rac1 and Rac3 on SNB19 cell proliferation and survival Expression of dominant-negative Rac1 has been shown to strongly inhibit the survival of glioma cell lines in normal growth conditions (Senger et al., 2002). To determine whether depletion of either Rac1 or Rac3 has similar effects on SNB19 cells, we ﬁrst examined the effects of Rac1- and Rac3-direced siRNA on SNB19 cell proliferation using sulforhodamine B, a colorimetric assay (Skehan et al., 1990). Depletion of either Rac1 or Rac3 causes a small, but reproducible decrease in cell growth in the presence of 10% serum (Figure 5a). These results imply that the dramatic inhibition in cell invasion caused by depletion of Rac1 or Rac3 is not due to a decrease in cell growth. We observed a much more

Figure 3 Depletion of Rac1, and to a lesser extent Rac3, inhibits cell migration. (a) Phase contrast micrographs. SNB19 cells were transfected with the indicated siRNA. (b) Kinetics of cell migration. SNB19 cells were transfected with luciferase-directed Figure 4 Depletion of either Rac1 or Rac3 inhibits SNB19 cell siRNA (solid squares), Rac1–1 (open circles) or Rac3–2 (open invasion. SNB19 cells were transfected with the indicated siRNA diamonds). The migration distance was quantiﬁed as described in and cell invasion through Matrigel was quantiﬁed as described in Materials and methods. Shown are the mean values (7s.e.m.) of Materials and methods. The number of invading cells in Rac1- or eight measurements for each time point and condition. Data shown Rac3-depleted cells was normalized to that of control cells. Shown are representative of three experiments is the mean (7s.e.m.) of four independent experiments

Oncogene Roles of Rac GTPases in human tumor cell invasion AY Chan et al 7825

Figure 6 Depletion of Rac1 or Rac3 inhibits BT549 cell invasion. (a) Western blot showing inhibition of Rac expression by Rac1- directed siRNA. Tubulin is shown as loading control. (b) Effect of Rac1- and Rac3-directed siRNA inhibits BT549 invasion through Matrigel. Cell invasion was quantified as described in Materials and methods. The number of invading cells in Rac1- or Rac3- depleted cells was normalized to that of control cells. Shown is the mean (7range) of two transfections plated in two wells each. (c) Effects of Rac1- and Rac3-directed siRNA on BT549 cell proliferation in the presence of serum. Solid squares, luciferase siRNA; open circles, Rac1–1 siRNA and open diamonds, Rac3–2 siRNA. Cell growth was quantified using SRB staining. Shown is the mean (7s.e.m., which is smaller than the size of the squares) of three samples. Data shown are representative of two experiments Figure 5 Effects of depletion of Rac1 or Rac3 on cell proliferation and survival. Effects of Rac1 or Rac3 depletion on cell prolifera- of Rac1 does not affect cell proliferation of U87MG tion in the presence of serum (a) or absence of serum (b). Cells were cells (Supplementary Information, Figure S1C). transfected with siRNA directed against luciferase as control (solid squares), Rac1 (open circles) or Rac3 (open diamonds). Data in (a) for the Rac1- and Rac3-directed siRNAs are largely coincident. Functional analysis of Rac1 and Rac3 in BT549 breast Cell growth was quantified using SRB staining. Shown is the mean carcinoma cells (7s.e.m.) of six wells. Data shown are representative of six experiments for (a) and two experiments for (b). (c) Effect of Rac1 To extend our analysis to carcinoma cells, we examined depletion on cell survival in the absence of serum. Cell apoptosis the effects of depleting either Rac1 or Rac3 on invasion was measured using an ELISA assay that quantifies histone- associated DNA fragments. Shown are the mean values (7s.e.m.) and cell proliferation of BT549 human breast carcinoma of four samples comprising two independent transfections cells. Similar to the results obtained in glioblastoma cells, depletion of Rac1 or Rac3 has a strong inhibitory effect on BT549 invasion (Figure 6b) with a minor effect pronounced inhibitory effect of depleting either Rac1 or on cell proliferation (Figure 6c). Rac3 on the growth of SNB19 cells in the absence of serum (Figure 5b). Determination of cell death by an ELISA assay that quantifies histone-associated DNA Discussion fragments indicates that this inhibitory effect is not caused by an increase in apoptosis however (Figure 5c). This study is the first characterization of the specific We also demonstrated that siRNA-mediated depletion functions of Rac1 and Rac3 in tumor cells. We found

Oncogene Roles of Rac GTPases in human tumor cell invasion AY Chan et al 7826 that siRNA-mediated depletion of either Rac1 or Rac3 and it is likely that the precise localization of GTPases has a strong inhibitory effect on the invasive behavior of contributes to their functions (Symons and Settleman, glioblastoma and breast carcinoma cells, whereas cell 2000; Gulli and Peter, 2001; Filippi et al., 2004). In proliferation is strongly inhibited only in the absence of addition, although Rac1 and Rac3 are likely to share serum. These results suggest critical roles for these Rac most of their effectors ((Haataja et al., 1997, 2002; Chan GTPases in tumor cell invasion. et al., unpublished observations), the hypervariable This work is also the first to report significant domain may also contribute to effector binding selec- differences between the roles of Rac1 and Rac3 in the tivity. This is supported by the observation that the organization of the actin cytoskeleton and cell migra- Rac3-specific effector CID binds to the C-terminal tion. Our results (Figure 2) show that RNAi-mediated domain of Rac3 but not of Rac1 (Haataja et al., 2002). depletion of Rac1 strongly inhibits lamellipodia forma- The strong inhibitory effect of depleting Rac3 on tion, but depletion of Rac3 does not. In addition, tumor cell invasion is quite remarkable in view of the depletion of Rac1 strongly inhibits monolayer wound fact that Rac3 protein only makes up a very small healing, whereas depletion of Rac3 only has a minor fraction of that of total Rac (Figure 1b). One expla- effect. nation for these observations is that the small comple- Notably, in contrast to the effect of RNAi-mediated ment of Rac3 is constitutively active in the tumor cell depletion of Rac3, transient expression of a dominant- lines that we have examined, as was found in highly negative version of Rac3 strongly inhibits lamelli- proliferative breast carcinoma cell lines (Mira et al., podia formation, to the same extent as caused by 2000). dominant-negative Rac1. This nonspecific effect of In agreement with observations that used expression dominant-negative Rac3 is not surprising, because of of dominant-negative Rac1 to interfere with the activa- the high degree of similarity of Rac1 and Rac3 (Haataja tion of Rac-like GTPases, we found that depletion of et al., 1997). Rac1 in SNB19 cells strongly inhibits lamellipodia More surprising is the observation that expression of formation and cell migration. These results provide constitutively active Rac3 strongly stimulates lamelli- evidence for a critical role for Rac1 in lamellipodia podia formation (Figure 2; Joyce and Cox, 2003), in formation and cell migration. Rac GTPases are thought view of our observation that RNAi-mediated depletion to regulate lamellipodia formation and cell migration in of Rac3 does not affect lamellipodia production. One large part by stimulating actin polymerization (Small explanation of this discrepancy could be that Rac1 and et al., 2002; Ridley et al., 2003). Inroads have been made Rac3 display a distinct intracellular localization and that toward the understanding of the signaling mechanisms the respective overexpressed versions are somehow that mediate these functions downstream of Rac1 mislocalized. In support of this scenario, we found that (Burridge and Wennerberg, 2004; Raftopoulou and Hall, whereas GFP-Rac1, when expressed in excess of the 2004). Rac1-stimulated actin polymerization is mediated binding capacity of RhoGDIa, predominantly targets to by a multiprotein complex that includes WAVE, which the plasma membrane, GFP-Rac3 shows prominent in turn stimulates Arp2/3-dependent actin nucleation endosomal and nuclear envelope localization in addition (Eden et al., 2002), whereas the inhibitory effect of Rac1 to plasma membrane targeting (Supplementary Infor- on myosin contractility is thought to be mediated by the mation, Figure S2). GFP-Rac2 predominantly localizes Rac effector PAK (Bokoch, 2003). to endomembranes, including the nuclear envelope, Notably, the inhibitory effect of Rac1 depletion on endosomes and the peripheral endoplasmic reticulum cell migration in SNB19 cells is in contrast with findings (Michaelson et al., 2001; Supplementary Information, in hematopoietic cells. Thus, neutrophils obtained from Figure S2). mice with a conditional Rac1 allele show essentially Our observations that the effects of expression of normal chemotaxis toward fMLP, whereas Rac2À/À dominant-negative or constitutively active mutants of neutrophils are significantly inhibited in chemotaxis (Gu Rac3 do not reflect the functions of Rac3 as indicated by et al., 2003). Furthermore, Rac1À/À bone marrow RNAi-mediated depletion indicates that the usefulness of macrophages also display normal migratory behavior these tools is limited. Thus, these findings have important (Wells et al., 2004). The migratory behavior of a colon implications for the study of Rho family GTPases and carcinoma cell line also has been found to be regulated beyond. We emphasize that salient results reported in this in a Rac-independent fashion (O’Connor et al., 2000). paper were reproduced using two independent siRNA These observations suggest that the role of Rac1 in cell oligos for each of the Rac isoforms, in addition to two migration is cell type-dependent and it will therefore be control oligos, thereby strongly diminishing the possibi- of interest to extend our studies to a larger panel of lity that the observed effects could be due to off-target tumor cell lines. action of the respective siRNAs (Jackson et al., 2003; It is likely that Rac1 regulates cell invasion by Saxena et al., 2003; Scacheri et al., 2004). stimulating signaling pathways in addition to those that The marked differences between the functions of Rac1 control the actin cytoskeleton. For example, Rac and Rac3 are intriguing because these two Rac proteins proteins have been shown to control the expression share 92% amino-acid identity, differing mainly in their and activity of a number of different metalloproteinases C-terminal hypervariable region. The hypervariable involved in extracellular matrix remodeling, including region of Rho GTPases is an important determinant MMP1, MMP2 and MMP14 (Kheradmand et al., 1998; of their subcellular localization (Michaelson et al., 2001) Zhuge and Xu, 2001). In addition, we recently showed

Oncogene Roles of Rac GTPases in human tumor cell invasion AY Chan et al 7827 that the phosphatidylinositol phosphatase synaptojanin siRNA was carried out using either Oligofectamine (Invitro- 2 is a Rac effector that is also necessary for efficient cell gen) or Lipofectamine 2000 (Invitrogen). For the Oligofecta- migration and invasion (Chuang et al., 2004), although mine transfections, 1 day before transfection, cells were 5 the molecular mechanisms whereby synaptojanin 2 trypsinized and plated on a six-well plate at 1 Â 10 cells/well mediates these functions remain to be clarified. Much in 1 ml of DMEM 10% serum without antibiotics. In all, 1.5 ml of siRNA (to yield a final concentration 25 nM) diluted in less is known about how Rac3 regulates cell invasion. 100 ml of serum-free DMEM and 4 ml of Oligofectamine diluted We did find however that as for Rac1, depletion of Rac3 in 16 ml of serum-free DMEM was preincubated for 7 min. The inhibits MMP1 production (Supplementary Informa- two mixtures were combined and incubated for 25 min at room tion, Figure S3). temperature for complex formation. After addition of 76 mlof We were surprised to find that although depletion of serum-free DMEM, the entire mixture (200 ml) was added to either Rac1 or Rac3 strongly inhibits the invasive each well, containing a final volume of 1.2 ml. For the behavior of tumor cells, it has no significant effect on Lipofectamine 2000 transfections, cells were plated on a cell survival. In contrast, expression of a dominant- six-well plate at 1.7 Â 105 cells/well in 2.5 ml of DMEM negative version of Rac1 has a very strong inhibitory 10% serum without antibiotics. Transfections were carried effect on cell survival in a number of different cell lines out as soon as the cells were fully spread. In all, 3 ml of siRNA (to yield a final concentration of 20 nM) diluted in 250 ml (Coniglio et al., 2001; Ruggieri et al., 2001; Senger et al., of DMEM and 4 ml of Lipofectamine 2000 diluted in 250 mlof 2002). Thus, it appears that there are qualitative DMEM was preincubated for 5 min. The two mixtures were differences between the effects of Rac1-directed siRNA combined and incubated for 20 min at room temperature for and dominant-negative Rac1. One possible explanation complex formation. The entire mixture (500 ml) was added to for this discrepancy is that dominant-negative Rac1, in each well, to a final volume of 3 ml. Cells were assayed 3–4 addition to inhibiting endogenous Rac GTPases, in- days after transfection. Rac and tubulin expression was hibits other Rho GTPases, whereas RNAi does not. determined by Western blot analysis, respectively, using Another possibility, however, is that overexpression of monoclonal antibodies against Rac and b3-tubulin (Upstate dominant-negative Rac1 at high enough levels can Biotechnology, Lake Placid, NY, USA). completely abolish the activity of Rac proteins (and For transient transfection of plasmids, SNB19 cells were plated on laminin-coated coverslips (BD Biosciences) and potentially of additional Rho family members) and that transfected with 0.4 mg of Myc-tagged pRK5 Rac3-V12 or the relatively small amount of Rac1 that remains in the pRK5 Rac3-N17 (Mira et al., 2000), pEXV Rac1-N17 (Qiu Rac1-directed siRNA-treated cells performs essential et al., 1995a) or T7-tagged pCGT Rac1-V12 (Joneson et al., functions such as cell survival. 1996) using Effectene transfection reagent (Qiagen) following In conclusion, the functional characterization of Rac1 the protocol provided by the manufacturer for a 24-well plate. and Rac3 in tumor cell lines using RNAi revealed At 24 h after transfection, Rac1-expressing cells were identified critical roles for these GTPases in the invasive behavior by immunofluorescence with either anti-Myc (Roche) or anti- of glioma and breast carcinoma cells. The use of RNAi T7 (Novagen) antibodies. to specifically deplete Rho family members in tumor cells is likely to yield important new information about Quantitative PCR the signaling mechanisms that contribute to malignant Total RNA was isolated using RNeasy (Qiagen, Valencia, CA, transformation. USA) from SNB19 cells 3 days after transfection with 20 nM of siRNA in a six-well plate. cDNA synthesis was performed using 1 mg of total RNA primed with oligo(dT) as described in Materials and methods ThermoscriptTM RT–PCR System (Invitrogen). Quantitative PCR was then performed using the ABI TaqMan system siRNA preparations (Rac1 forward primer: AAGAGAAAATGCCTGCTGTTG siRNA oligonucleotide duplexes specific for Rac1 or Rac3 TAA, reverse primer: GCGTACAAAGGTTCCAAGGG; were designed according to Elbashir et al. (2001). 21-nt RNAs Rac1 Probe: TET-TGTCTCAGCCCCTCGTTCTTGGTCC- were purchased from Dharmacon (Lafayette, CO, USA) in TAMRA. Rac3 forward primer: GGGAAGACATGCTT deprotected and desalted form. The siRNA sequences used GCTGATC, reverse primer: CCTCCTGACCCGCTGTGT; are: Rac1–1, corresponding to bp 439–459 after the start Rac3 Probe: TET-CTACACGACCAACGCCTTCCCC- codon of the Rac1 gene (50-AAGGAGATTGGTGCTG TAMRA). Expression of b-actin mRNA was used as internal TAAAA-30), Rac1–2, corresponding to bp 616–636 after the control. The mRNA expression level of Rac1 and Rac3 in start codon of the Rac1 gene (50-AACCTTTGTA Rac1- or Rac3-depleted cells was normalized to the respective CGCTTTGCTCA-30), Rac3–1, corresponding to bp 550–570 Rac1 and Rac3 mRNA expression levels in control cells after the start codon of the Rac3 gene (50-AAGCCGGG transfected with siRNA directed against luciferase. GAAGAAGUGCACC-30), Rac3–2, corresponding to bp 701– 0 721 after the start codon of the Rac3 gene (5 -AAGCATGGG Immunofluorescence GATGAGGCTGGG-30) and luciferase, corresponding to bp 291–309 after the start codon of the GL2 luciferase gene Cells that were transfected with either Rac1-, Rac3- or (50-AACGTACGCGGAATACTTCGA-30). siRNA duplex luciferase-directed siRNA for 24 h were trypsinized and plated formation was performed as described in Elbashir et al. (2001). on laminin-coated coverslips overnight. After 16 h of serum starvation, cells were stimulated with 5% serum-containing medium for 5 min. Subsequently, the cells were washed with Cell culture and transfections PBS, fixed in 4% formaldehyde/PBS, permeabilized with 0.1% Cells were grown at 371C in DMEM supplemented with 10% Triton X-100 dissolved in PBS and incubated with FITC- FCS and penicillin/streptomycin. Transient transfection of conjugated phalloidin (Molecular Probes) to stain for F-actin.

Oncogene Roles of Rac GTPases in human tumor cell invasion AY Chan et al 7828 Processed coverslips were mounted in 75% Vectashield cells were allowed to recover for 15 min. Images were collected mounting medium (Vector Laboratory). Images were collected for each siRNA treatment at various times after wounding for using an IX70 Olympus inverted microscope equipped with a the same eight spots, localized on the underside of the dish by a Â 60 (1.4 NA) objective, an Orca II cooled CCD camera marker. Cell migration distance was determined by measuring (Hamamatsu) and ESee (Inovision, Raleigh, NC, USA) image the width of the wound divided by two and subtracting this analysis software. value from the initial half-width of the wound. The rate of cell movement was calculated as the slope of the migration Quantification of lamellipodia formation distance over an 18-h time period. Cells were treated as described under ‘Immunofluorescence’. For each experimental condition, images were taken in a Sulforhodamine B assay random fashion. Lamellipodia were traced using ESee image Cell proliferation was measured using the sulforhodamine B analysis software. For each cell, the total length of the colorimetric assay (Skehan et al., 1990). Briefly, 1 day after lamellipodia was divided by the circumference of the cell to siRNA transfection, cells were seeded at 2 Â 103 cells/well in a give the percentage of lamellipodia per cell. 96-well microtiter plate. At various times, cells were fixed in 10% trichloroacetic acid for 1 h at 41C, rinsed and sub- Invasion assay sequently stained for 30 min at room temperature with 0.2% SRB dissolved in 1% acetic acid, followed by air drying. The Invasion was assayed by measuring cell invasion through bound dye was solubilized in 100 mlof10mM unbuffered Tris Matrigel Invasion Chambers (Becton Dickinson, MA, USA). base for 30 min and the OD was read at 490 nm in an ELISA At 2 days after transfection with siRNA, cells were placed in plate reader. the upper chamber in whole medium (DMEM and 10% FCS), 4 Â 104 cells for SNB19 and 1 Â 105 cells for BT549 and U87MGcells. 500 ml of whole medium was added to the Apoptosis assay bottom chamber. After 24 h of incubation at 371C, cells on the The level of DNA fragmentation was quantified using the Cell upper surface of the filter were wiped off with a Q-tip and the Death ELISA PLUS kit (quantifying histone-associated DNA filter was fixed in 4% formaldehyde/PBS. After staining with fragments) using the protocol suggested by the manufacturer crystal violet, all cells on the bottom of the chamber were (Roche). counted using an IX70 Olympus inverted microscope. Acknowledgements Monolayer wound healing assay We thank Dr L Van Aelst (Cold Spring Harbor) for the pCGT At 2 days after transfection with siRNA, 2 Â 105 cells were Rac1-V12 plasmid and SNB19 and U87MGcells. We also like detached and plated in a 12-well plate. After overnight to thank N Rusk and A Valster for critical reading of the incubation, two parallel wounds of approximately 400 mm manuscript. This work was supported by National Institutes of were made using a p1000 pipette tip. After rinsing with PBS, Health Grant CA87567.

References

Allen WE, Zicha D, Ridley AJ and Jones GE. (1998). J. Cell. Gu Y, Filippi MD, Cancelas JA, Siefring JE, Williams EP, Biol., 141, 1147–1157. Jasti AC, Harris CE, Lee AW, Prabhakar R, Atkinson SJ, Anand-Apte B, Zetter BR, Viswanathan A, Qiu RG, Chen J, Kwiatkowski DJ and Williams DA. (2003). Science, 302, Ruggieri R and Symons M. (1997). J. Biol. Chem., 272, 445–449. 30688–30692. Gulli MP and Peter M. (2001). Genes Dev., 15, 365–379. Bokoch GM. (2003). Annu. Rev. Biochem., 72, 743–781. Haataja L, Groffen J and Heisterkamp N. (1997). J. Biol. Bouzahzah B, Albanese C, Ahmed F, Pixley F, Lisanti MP, Chem., 272, 20384–20388. Segall JD, Condeelis J, Joyce D, Minden A, Der CJ, Haataja L, Kaartinen V, Groffen J and Heisterkamp N. Chan A, Symons M and Pestell RG. (2001). Mol. Med., 7, (2002). J. Biol. Chem., 277, 8321–8328. 816–830. Jackson AL, Bartz SR, Schelter J, Kobayashi SV, Burchard J, Burridge K and Wennerberg K. (2004). Cell, 116, 167–179. Mao M, Li B, Cavet Gand Linsley PS. (2003). Nat. Chuang YY, Tran NL, Rusk N, Nakada M, Berens ME and Biotechnol., 21, 635–637. Symons M. (2004). Cancer Res., 64, 8271–8275. Joneson T, White MA, Wigler MH and Bar-Sagi D. (1996). Coniglio SJ, Jou TS and Symons M. (2001). J. Biol. Chem., Science, 271, 810–812. 276, 28113–28120. Joyce PL and Cox AD. (2003). Cancer Res., 63, 7959–7967. Debreceni B, Gao Y, Guo F, Zhu K, Jia B and Zheng Y. Karnoub AE, Worthylake DK, Rossman KL, Pruitt WM, (2004). J. Biol. Chem., 279, 3777–3786. Campbell SL, Sondek J and Der CJ. (2001). Nat. Struct. Denli AM and Hannon GJ. (2003). Trends Biochem. Sci., 28, Biol., 8, 1037–1041. 196–201. Keely PJ, Westwick JK, Whitehead IP, Der CJ and Parise LV. Dykxhoorn DM, Novina CD and Sharp PA. (2003). Nat. Rev. (1997). Nature, 390, 632–636. Mol. Cell. Biol., 4, 457–467. Kheradmand F, Werner E, Tremble P, Symons M and Werb Eden S, Rohatgi R, Podtelejnikov AV, Mann M and Z. (1998). Science, 280, 898–902. Kirschner MW. (2002). Nature, 418, 790–793. Khosravi-Far R, Solski PA, Clark GJ, Kinch MS and Der CJ. Elbashir SM, Harborth J, Lendeckel W, Yalcin A, Weber K (1995). Mol. Cell. Biol., 15, 6443–6453. and Tuschl T. (2001). Nature, 411, 494–498. Lin R, Bagrodia S, Cerione R and Manor D. (1997). Curr. Feig LA. (1999). Nat. Cell. Biol., 1, E25–E27. Biol., 7, 794–797. Filippi MD, Harris CE, Meller J, Gu Y, Zheng Y and Michaelson D, Silletti J, Murphy G, D’Eustachio P, Rush M Williams DA. (2004). Nat. Immunol., 5, 744–751. and Philips MR. (2001). J. Cell. Biol., 152, 111–126.

Oncogene Roles of Rac GTPases in human tumor cell invasion AY Chan et al 7829 Mira JP, Benard V, Groffen J, Sanders LC and Knaus UG. Scacheri PC, Rozenblatt-Rosen O, Caplen NJ, Wolfsberg TG, (2000). Proc. Natl. Acad. Sci. USA, 97, 185–189. Umayam L, Lee JC, Hughes CM, Shanmugam KS, Nobes CD and Hall A. (1999). J. Cell. Biol., 144, 1235–1244. Bhattacharjee A, Meyerson M and Collins FS. (2004). Proc. O’Connor KL, Nguyen BK and Mercurio AM. (2000). J. Cell. Natl. Acad. Sci. USA, 101, 1892–1897. Biol., 148, 253–258. Senger DL, Tudan C, Guiot MC, Mazzoni IE, Molenkamp G, Qiu RG, Abo A, McCormick F and Symons M. (1997). Mol. LeBlanc R, Antel J, Olivier A, Snipes GJ and Kaplan DR. Cell. Biol., 17, 3449–3458. (2002). Cancer Res., 62, 2131–2140. Qiu RG, Chen J, Kirn D, McCormick F and Symons M. Shaw LM, Rabinovitz I, Wang HH, Toker A and Mercurio (1995a). Nature, 374, 457–459. AM. (1997). Cell, 91, 949–960. Qiu RG, Chen J, McCormick F and Symons M. (1995b). Proc. Skehan P, Storeng R, Scudiero D, Monks A, McMahon J, Natl. Acad. Sci. USA, 92, 11781–11785. Vistica D, Warren JT, Bokesch H, Kenney S and Boyd MR. Raftopoulou M and Hall A. (2004). Dev. Biol., 265, 23–32. (1990). J. Natl. Cancer Inst., 82, 1107–1112. Repesh LA. (1989). Invasion Metastasis, 9, 192–208. Small JV, Stradal T, Vignal E and Rottner K. (2002). Trends Ridley AJ, Paterson HF, Johnston CL, Diekmann D and Hall Cell Biol., 12, 112–120. A. (1992). Cell, 70, 401–410. Symons M and Settleman J. (2000). Trends Cell Biol., 10, Ridley AJ, Schwartz MA, Burridge K, Firtel RA, Ginsberg 415–419. MH, Borisy G, Parsons JT and Horwitz AR. (2003). Science, Van Aelst L and D’Souza-Schorey C. (1997). Genes Dev., 11, 302, 1704–1709. 2295–2322. Ruggieri R, Chuang YY and Symons M. (2001). Mol. Med., 7, Wells CM, Walmsley M, Ooi S, Tybulewicz V and Ridley AJ. 293–300. (2004). J. Cell. Sci., 117, 1259–1268. Sahai E and Marshall CJ. (2002). Nat. Rev. Cancer, 2, 133–142. Zheng Y. (2001). Trends Biochem. Sci., 26, 724–732. Saxena S, Jonsson ZO and Dutta A. (2003). J. Biol. Chem., Zhuge Y and Xu J. (2001). J. Biol. Chem., 276, 278, 44312–44319. 16248–16256.

Supplemenatry Information accompanies the paper on Oncogene website (http://www.nature.com/onc)

Oncogene