Rho family GTPases cooperate with p53 deletion to promote ... - Nature

Oncogene (2004) 23, 5577–5585

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

Rho family GTPases cooperate with p53 deletion to promote primary mouse embryonic fibroblast cell invasion Fukun Guo1 and Yi Zheng*,1 1 Division of Experimental Hematology, Children’s Hospital Research Foundation, University of Cincinnati, Cincinnati, OH 45229, USA

The Rho family GTPases Rac1, RhoA and Cdc42 function as molecular switches that transduce intracellular signals regulating multiple cell functions including gene expression, adhesion, migration and invasion. p53 and its regulator p19Arf, on the other hand, are tumor suppressors that are critical in regulating cell cycle progression and apoptosis. Previously, we have demonstrated that the Rho proteins contribute to the cell proliferation, gene transcription and migration phenotypes unleashed by p19Arf or p53 deletion in primary mouse embryo fibroblasts (MEFs). To further investigate their functional interaction in the present study, we have examined the involvement of Rho signaling pathways in p53-mediated cell invasion. We found that in primary MEFs (1) p53 or p19Arf deficiency led to a marked increase in the number of focal adhesion plaques and fibronectin production, and RhoA, Rac1 and Cdc42 contribute to the p53- and p19Arfmediated focal adhesion regulation, but not fibronectin synthesis; (2) although endogenous Rac1 activity was required for the p19Arf or p53 deficiency-induced migration phenotype, hyperactive Rho GTPases could not further enhance cell migration, rather they suppressed cell–cell adhesion of p53/ MEFs; (3) expression of the active mutant of RhoA, Rac1 or Cdc42, but not Ras, promoted an invasion phenotype of p53/, not p19Arf/, cells; (4) although ROCK activation can partially recapitulate Rho-induced invasion phenotype, multiple pathways regulated by RhoA, in addition to ROCK, are required to fully cooperate with p53 deficiency to promote cell invasion; and (5) extracellular proteases produced by the active RhoA-transduced cells are also required for the invasion phenotype of p53/ cells. Combined with our previous observations, these results strongly suggest that mitogenic activation of Rho family GTPases can cooperate with p53 deficiency to promote primary cell invasion as well as transformation and that multiple signaling components regulated by the Rho proteins are involved in these processes. Oncogene (2004) 23, 5577–5585. doi:10.1038/sj.onc.1207752 Published online 3 May 2004 Keywords: Rho GTPases; p53; cell migration; invasion

*Correspondence: Y Zheng; E-mail: [email protected] Received 6 January 2004; revised 9 March 2004; accepted 23 March 2004; published online 3 May 2004

Introduction Rho family GTPases are intracellular signal transducers involved in the regulation of cell proliferation, gene induction, survival, actin cytoskeleton and motility (Van Aelst and D’Souza-Schorey, 1997; Etienne-Manneville and Hall, 2002). Many mitogenic signals, including those from growth factor receptors and integrins, can activate the Rho family members Rac1, RhoA and Cdc42 by promoting the exchange of bound GDP for GTP (Zheng, 2001), resulting in an active conformation to interact with an array of effector targets to elicit specific cellular effects (Zohn et al., 1998; Bishop and Hall, 2000). A wealth of evidence has implicated Rho GTPases in many aspects of tumor development (Aznar and Lacal, 2001; Boettner and Van Aelst, 2002). In addition to the recognized roles in mediating cell growth transformation by many oncogene products (Zohn et al., 1998; Bar-Sagi and Hall, 2000), RhoA, Rac1 and Cdc42 are intimately associated with morphological transformation of tumor cells and have been linked directly to tumor cell migration and invasion through their ability to regulate actin cytoskeleton, cell–extracellular matrix adhesion and cell–cell interaction (Kaibuchi et al., 1999; Schwartz and Shattil, 2000; Ridley, 2001). The p53 cell cycle inhibitor and its regulators, including p19ARF, are well established tumor suppressors that are components of a complex signaling network critical to tumor suppression (Sherr, 2001; Hahn and Weinberg, 2002). Deletion or mutation in p53 or its regulators occurs in many tumor cases and correlates with the onset of a wide spectrum of cancers. p53 is a key transcription factor essential for the response to cellular stress from DNA damage, hypoxia and oncogene activation, and when activated, can trigger cell cycle arrest or apoptosis (Levine, 1997), whereas p19ARF may serve as a sensor to oncogenic insult to stabilize p53 by sequestering Mdm2, a negative regulator of p53 activity (Sherr, 2001). The p19ARF–p53 tumor suppressor pathway therefore has been thought to be primarily involved in monitoring proliferation signals to prevent cells from uncontrolled growth. Indeed, excess of mitogenic signals that activate Ras or Rho GTPase pathways can cooperate with p19ARF or p53 deficiency to promote growth transformation (Kamijo et al., 1997; Serrano et al., 1997; Guo and Zheng, 2004). On the other hand, accumulating evidence also suggests a

Cooperativity of Rho GTPases and p53 defect in invasion F Guo and Y Zheng

5578

functional linkage between p53 and cell morphological regulation and motility modulation. For example, p53 has been shown to regulate the transcription of a number of genes involved in cell morphology and/or movement, including that of smooth muscle a-actin (Comer et al., 1998), collagens IIa1 and VIa1 (Sun et al., 1999), VEGF (Mukhopadhyay et al., 1995), metalloproteinase-1 (Zhao et al., 2000) and fibronectin (Iotsova and Stehelin, 1996). Overexpression of p53 seemed to cause a partial reversion of Ras-induced morphological transformation (Gloushankova et al., 1997) and to inhibit cell migration (Alexandrova et al., 2000), and the spontaneous tumors produced by p53 gene targeting are often highly invasive (Schmitt et al., 2002). A recent study also suggests that p53 may have a role in Cdc42mediated filopodia formation and cell polarization (Gadea et al., 2002) and we have shown that the p19Arf–p53 pathway negatively modulates PI3 kinase and Rac1 GTPase activities and regulates actin cytoskeleton and cell migration through the PI3 kinase-Rho GTPase signaling module (Guo et al., 2003a; Guo and Zheng, 2004). Here, we provide further evidence that the p53 tumor suppressor pathway is functionally connected to the Rho GTPase pathways in regulating the cell–extracellular matrix and cell–cell adhesions and the invasion properties. By using primary mouse embryonic fibroblasts deficient in p19Arf or p53, we show that RhoA, Rac1 and Cdc42 contribute to the p19Arf- and p53mediated focal adhesion regulation and that hyperactive Rho proteins, but not Ras, promote an invasive phenotype of p53/ cells. We have also mapped the downstream pathways regulated by RhoA, ROCK and extracellular proteases, among others, as important mediators of the invasion phenotype of p53/ cells. Our results support the hypothesis that malfunction of the same set of gene products can confer proliferative selection advantage as well as empower cell invasion and identify the Rho GTPases as a key set of cellular signal transducers mediating both processes. Results We have previously shown that deletion of p19Arf or p53 tumor suppressor gene resulted in actin cytoskeleton rearrangement and a significant increase in cell migration rate, a phenotype that requires Rho GTPase activities (Guo et al., 2003a; Guo and Zheng, 2004). Since focal adhesion complex and extracellular components such as fibronectin may play important roles in cell migration and they are intimately associated with Rho GTPase activities (Schwartz and Shattil, 2000; Ridley, 2001), we examined the effect of p19Arf or p53 deletion on the focal complex formation and fibronectin production and the possible involvement of Rho GTPases in these processes. Removal of p19Arf or p53 led to a drastic increase in the number of focal contact as revealed by antivinculin immunostaining, and the focal adhesion plaques appeared to be more widely distributed throughout the cell body in p19Arf/ and p53/ Oncogene

cells compared with those found typically at the edges of wild type mouse embryo fibroblast (MEFs) (Figure 1a). This effect was reversible by reconstitution of p19Arf or p53 into the mutant cells (Figure 1a). Concomitantly, fibronectin production in p19Arf/ and p53/ cells was increased significantly compared with wild-type MEFs, and this effect was also reversible by p19Arf or p53 reintroduction (Figure 1b). These results indicate that p19Arf and p53 regulate cell focal adhesion and extracellular matrix composition. Dominant-negative mutants of the Rho GTPases, Rac1N17, RhoAN19 and Cdc42N17, could reduce the number of focal adhesion plaques of the p19Arf/ and p53/ cells and change their distribution to a pattern of more sparse contacts at the cell edges to different extents (Figure 2). Each Rho protein mutant appeared to induce a distinct focal contact pattern in the cells with RhoAN19 and Rac1N17 showing relatively minor effects on the p19Arf/ and p53/ cells, respectively, reflecting the unique contribution of each to the modulation of cell adhesion complex formation in p19Arf/ and p53/ cells. As among the three Rho GTPases only dominantnegative Rac1 had a significant impact on the migration rate (Guo et al., 2003a), these results further suggest that focal adhesion is not required for cell migration. Consistent with the previous report that p53 can directly regulate fibronectin gene transcription (Iotsova and Stehelin, 1996), the dominant-negative Rho proteins had no significant effect on fibronectin production of the p19Arf/ and p53/ cells (Figure 2c; data not shown), suggesting that Rac1, RhoA and Cdc42 are not involved in the regulation of fibronectin production by p19Arf and p53. Despite the regulatory effect of p19Arf or p53 deletion on extracellular matrix composition (e.g. increased fibronectin), the addition of recombinant fibronectin and/or conditional culture medium derived from the p19Arf/ or p53/ cells to wild-type MEFs was insufficient to promote cell migration (data not shown), suggesting that even if the upregulation of fibronectin and other extracellular components by p19Arf and p53 defects might play a role in the cell movement, they are not sufficient to account for the migration phenotype of these cells. Indeed, we have shown previously that the modulation of intracellular pathways by the tumor suppressor deletion, including upregulation of PI3 kinase and Rac1 activities (Guo et al., 2003a; Guo and Zheng, 2004), is required for the migration phenotype. Next, we investigated whether further elevation of Rho GTPase activities in p19Arf/ or p53/ cells might modulate the migration phenotype and alter their invasive capability. To this end, we have examined the p19Arf/ and p53/ cells expressing the Rac1L28, RhoAL30 or Cdc42L28 fast cycling mutant, which possesses increased intrinsic rate of GDP/GTP exchange mimicking mitogenic stimulation (Lin et al., 1999), for the migration, cell–cell adhesion and invasion properties. Despite their growth- and transformation-promoting abilities in these cells, the Rac1, RhoA and Cdc42 mutants could not further enhance the migration rate of p19Arf/ or p53/ cells under the experimental


5579

Figure 1 p19Arf and p53 modulate focal adhesion complex and fibronectin production. The effects of deletion of p19Arf or p53 on focal adhesion complex (a) and fibronectin production (b) of primary MEF cells were examined by immunofluorescence or Western analysis. A total of 2 104 of wild-type (WT) or knockout/reconstituted MEF cells were cultured overnight on coverglass. Subconfluent cells were starved in a medium containing 0.5% serum for 12 h. After fixation in 3.7% formaldehyde, cells were permeabilized with 0.1% Triton X-100 and stained for vinculin by a monoclonal anti-vinculin antibody (a) or for fibronectin with an anti-fibronectin antibody (b). The anti-fibronectin blots were revealed by enhanced chemiluminescence of secondary antibody

conditions (Figure 3a; data not shown). Interestingly, each of the active RhoA, Rac1 and Cdc42 mutants could cause a significant suppression of cell–cell adhesion of p53-deficient cells in a cell aggregation assay (Figure 3b). These results suggest that the enhancement of the Rho protein activities caused by p19Arf or p53 deletion (Guo et al., 2003a) is sufficient to prime the cells for the optimal migration rate, but further elevation of the GTPase activities may modulate both cell–extracellular matrix adhesion and cell–cell adhesion. Deletion of p53 alone or introduction of the active Rho GTPase mutants to the wild-type MEFs was insufficient to induce cell invasion, but introduction of the fast cycling Rho GTPase mutants to the p53/ cells effectively promoted an invasion phenotype (Figure 4). The active RhoA mutant, RhoAL30, in particular, appeared to be 2–3-fold more potent than that of Rac1 or Cdc42 in promoting the invasion phenotype when expressed at a similar level as Rac1 or Cdc42 (Figure 4). Interestingly, although the expression of the fast cycling

RhoA, Rac1 or Cdc42 mutant could promote transformation of p19Arf/ cells (Guo and Zheng, 2004), they were unable to induce invasion in these cells. Moreover, oncogenic Ras was ineffective in promoting cell invasion under the conditions at which the active Rho protein mutants were effective despite its potent transforming activity in these cells (Figure 4; data not shown). These results suggest that Ras-independent mitogenic activation of Rho GTPases and p53 deletion may cooperatively promote cell invasion, and that the malfunction of the same set of gene products, Rho GTPases, can confer both proliferative selection advantage and invasive power in a primary cell setting. To examine the contribution to the invasion phenotype by signaling pathways regulated by Rho GTPases, we next focused on the effector pathways of RhoA, since the fast cycling mutant of RhoA, RhoAL30, appeared to be the most potent of all three Rho GTPase mutants in promoting p53/ cell invasion (Figure 4). As ROCK, one of the major RhoA effectors, was found to be Oncogene


5580

Figure 2 Effect of dominant-negative Rho GTPases on p19Arf- and p53-mediated focal adhesion production. The (HA)3-tagged, dominant-negative mutants of Rac1, RhoA or Cdc42 were introduced into p19Arf/ or p53/ MEFs by retroviral induction. The expression of the Rho proteins were detected by anti-HA Western blotting (a). The MEFs were stained with anti-vinculin antibody to reveal the focal adhesion plaques (b). Fibronectin production was detected by anti-fibronectin immunostaining (c)

significantly upregulated in the mRNA level in the RhoA-transduced p53/ cells in a gene array assay (our unpublished observation) and has been associated with cell invasion (Sahai and Marshall, 2003), we examined if it is involved in the invasion phenotype of p53/ cells that express RhoAL30. Inhibition of ROCK activity by the inhibitor Y-27632 partially reduced the invasive potential of RhoAL30-expressing p53/ cells, whereas overexpression of ROCKI in p53/ cells was able to Oncogene

stimulate invasion albeit to a less extent than RhoAL30 (Figure 5a). These results suggest that the ROCK pathway is necessary for RhoA to fully cooperate with p53 deficiency to promote invasion. To confirm the contribution of ROCK to RhoA-mediated invasion and to assess the involvement of additional effector pathways downstream of RhoA, we further tested a set of effector-domain mutants of RhoA for their ability to induce invasion of p53/ primary MEFs. As previously


5581

Figure 4 Active RhoA, Rac1 and Cdc42 mutants cooperate with p53 deletion to promote primary MEF invasion. (a) The active RhoAL30, Rac1L28 or Cdc42L28 promotes an invasion phenotype of p53/ MEFs. The cells were placed in an invasion chamber for 24 h at 371C. Cells invaded through the Matrigel matrix and 8 mm membrane pores were visualized by Giemsa staining. (b) The invasion assays were carried out in triplet and the number of cells invaded through the Matrigel matrix and membrane pores were normalized to that of the wild-type MEFs expressing GFP only. The effects of the Rho protein mutants were compared with that of oncogenic RasV12 mutant in p53/ MEFs or in p19Arf/ MEFs. Data are representative of three independent measurements

Figure 3 Hyperactivation of Rho GTPases did not further promote p19Arf/ or p53/ cell migration, but inhibited cell–cell adhesion in p53/ cells. GFP, wild-type p53, the fast cycling mutant of Rac1, RhoA or Cdc42, or the dominant-negative mutant of Rac1 protein together with GFP were introduced into wild-type or p53/ MEFs by retroviral induction. The migration distances of the cells were compared with that of the wild-type MEFs in a wound-healing assay (a). A cell aggregation assay was carried out to compare directly the cell–cell adhesion property in the presence or absence of 2 mM EGTA (b). The basal cell aggregations in the EGTA-treated samples were subtracted from the data without EGTA treatment

characterized in vitro (Fujisawa et al., 1998; Sahai et al., 1998) and depicted in Figure 5b, RhoA-F39V is defective in PKN binding but retains ROCK binding, RhoA-E40L is defective for ROCK recognition, RhoAE40T retains ROCK and PKN binding but is defective in Kinectin and mDia binding, and RhoA-Y42C is selectively defective in PKN binding. Consistent with the partial loss of invasion activity in the case of ROCK

inhibitor-treated cells, RhoAL30-E40L that is defective in ROCK binding suffered partial loss of invasion activity (Figure 5c). The p53/ cells expressing the RhoA mutants that have lost PKN-binding ability (RhoA-F39V and RhoA-Y42C) and the RhoA mutant that retains ROCK and PKN binding but is defective for Kinectin and mDia, and possibly also other effector pathways (RhoAL63-E40T) also partially lost invasion capability (Figure 5c), suggesting that PKN and other effectors may also contribute to the invasiveness. These RhoA mutants are biologically active since they either retained transforming activity in p53/ MEFs (Guo and Zheng, 2004) or actin stress fiber-inducing ability in NIH 3T3 cells (Sahai et al., 1998). Therefore, the RhoA– ROCK pathway is critical for the RhoA-induced invasion of primary p53/ MEFs, corroborating with the previous findings in human cancer cells (Sahai and Marshall, 2003). Moreover, additional effector pathways other than ROCK emanated from active RhoA, Oncogene


5582

Figure 6 Effect of extracellular PIs on RhoAL30-promoted p53/ MEF invasion. The p53/ cells transduced with RhoAL30 were assayed for their invasive activities in the Matrigel matrix in the presence or absence of the individual PI or a PI cocktail (20 mM GM6001, 1.2 mM aprotinin, 14.5 mM leupeptin, 64.5 mM bestatin, 22.5 mM E64 and/or 14.5 mM pepstatin), the ROCK inhibitor Y27632 (20 mM) or the combination of Y27632 and PI

Figure 5 Multiple pathways regulated by RhoA are required to synergize with p53 deletion to induce cell invasion. (a) RhoA– ROCK pathway is necessary but not sufficient to fully cooperate with p53 deficiency to promote cell invasion. Invasion activities of p53/ MEFs transduced with the RhoAL30 mutant or ROCK1 in the presence or absence Y27632 (20 mM) were determined as described in Figure 4. (b) A set of effector-domain mutants of RhoA allows selective uncoupling of RhoA with ROCK, PKN or other effectors. Single point mutants made in the switch I domain of RhoA impair or retain specific effector binding as depicted. (c) The effects of the effector-domain mutations on RhoA-induced invasion of p53/ cells were compared with that induced by active RhoA. Data shown are representative of two independent experiments

including PKN, also appear to be involved in the RhoAmediated invasion of p53/ cells. These results provide clues on the contribution of RhoA-regulated signaling network to the invasion phenotype of a primary cell system. It has been suggested recently that extracellular proteases, including matrix metalloproteases, are involved in the Rho GTPase-mediated tumor cell motility and invasion (Sahai and Marshall, 2003). Inhibition of proteolysis appeared to work synergistically with ROCK inhibition to reduce certain types of tumor cell invasion but not others (Sahai and Marshall, 2003). To determine if extracellular proteases are involved in RhoA-induced invasion of cells of the p53/ genetic background, we assayed the invasive activity of RhoAL30-expressing p53/ MEFs in the presence or Oncogene

absence of individual protease inhibitor (PI) (GM6001, aprotinin, leupeptin, bestatin, E64 or pepstatin), or a PI cocktail that contained all of them with or without the ROCK inhibitor, Y27632. As shown in Figure 6, while each component of the PI had no or marginal effect on the RhoAL30-promoted invasion of p53/ MEFs, PI was able to inhibit the invasive activity of the cells by B50%. Application of a combined dose of PI with Y27632 yielded a similar degree of inhibition as administrating PI alone, suggesting that pericellular proteolysis is partly regulated by the ROCK pathway and is required for the invasive activity. These results further implicate extracellular proteases released by the active RhoA mutant-transduced cells as important effectors for the invasion phenotype of p53/ cells.

Discussion In the previous studies, by genetic disruption of the tumor suppressor genes, we have shown that p19ARF and p53 have a profound effect on cell actin cytoskeleton and motility (Guo et al., 2003a). We found that the actins of the p19Arf/ and p53/ cells were rearranged to an intense cortical structure due to significantly elevated level of endogenous PI3 kinase and Rho GTPase activities, and both PI3 kinase and Rac1 were required for the p19Arf- and p53-regulated migration (Guo et al., 2003a; Guo and Zheng, 2004). We have also shown that Rho family GTPases Rac1, RhoA and Cdc42 contribute to the p19Arf- and p53-regulated gene transcription and cell growth, and that activation of these Rho GTPases can cooperate with p19Arf or p53 defect to promote hyperproliferation and transformation (Guo and Zheng, 2004). In the present studies, we have further determined the relationship between Rho


5583

GTPase pathways and the p19Arf–p53 tumor suppressor pathway in cell–extracellular matrix and cell–cell adhesion and fibronectin production that are likely related to cell migration or invasion. Consistent with a previous report of a fibroblast cell line (Alexandrova et al., 2000), we found that p19Arf and p53 regulated cell focal adhesion and extracellular matrix composition in primary MEFs (Figure 1). Moreover, we showed that Rac1, RhoA and Cdc42 contributed to the modulation of cell adhesion complex in p19Arf/ and p53/ cells (Figure 2). Although RhoA and Cdc42 are important contributors to focal adhesion, they are not required for cell migration, consistent with the results of a previous study in a fibroblast cell line (Nobes and Hall, 1999). The implication is that defect in p19Arf or p53 might stimulate the primary cell movement through the upregulation of fibronectin and/or other extracellular matrix components, but not through modulation of focal adhesion. Our results obtained using the genetically modified primary MEFs conceptually contradict a recent report (Sablina et al., 2003), suggesting that p53 expression is involved in promoting rather than suppressing cell movement in a number of tumor cell lines. This might reflect the differences in the cell types or the methods used to downregulate/upregulate p53 expression. Although either p53 deficiency or activation of members of the Rho family could cause an increase in cell motility (Alexandrova et al., 2000; Ridley, 2001; Guo et al., 2003a), defect in the tumor suppressor pathway or elevation of Rho protein activity alone apparently was insufficient to induce invasion of primary MEFs. Rather, a combination of p53 deficiency and Rho GTPase hyperactivation enabled the cells to gain an invasive phenotype (Figure 4). This correlated with the observed decrease in cell–cell adhesion (Figure 3b), which may also contribute to the invasion phenotype. It is therefore possible that mitogenic insults or oncogenic signals that activate Rho proteins independent of p53 network might work cooperatively with p53 deficiency to turn the migration advantage of the p53/cells to an invasive capability (Figure 7). Interestingly, the active mutants of RhoA, Rac1 and Cdc42 were unable to induce p19Arf/ MEFs to become

Figure 7 A model depicting a possible relationship between the p53 tumor suppressor pathway and Rho GTPase signaling module in the regulation of MEF cell invasion. In addition to malfunction in check point and apoptosis control, deficiency of p53 may alter actin cytoskeleton organization, focal adhesion property and extracellular matrix production that contribute to cell migration by upregulating Rho GTPases, Rac1 in particular. Mitogenic insults or oncogenic signals that cause hyperactivation of RhoA, Rac1 or Cdc42 can further modulate cell–matrix and cell–cell adhesions and extracellular matrix protein generation and cooperates with p53 deficiency to promote cell invasion

invasive, nor did oncogenic Ras in either p19Arf/ or p53/ MEFs, despite heir ability to induce growth transformation of these cells (Kamijo et al., 1997; Guo and Zheng, 2004). These results highlight a p19Arfindependent function of p53 and suggest that unique or quantitatively different signals of the p19Arf/ vs p53/ cells and of Rho GTPase vs Ras are involved in promoting the invasion phenotype. They also caution that increased migration and fibronectin production (e.g. p19Arf/ MEFs) are not necessarily predictive of the cell invasive property in the context of Rho GTPase activation. The conditional cultural medium from the p53/ cells expressing the active Rho GTPase mutants was insufficient to induce invasion of wild-type or p53/ MEFs (our unpublished observation). Thus, we expect that the abilities of individual Rho GTPases to regulate both intracellular and extracellular changes may combine to cooperate with the p53-regulated elements to promote invasion. To this end, we have investigated the involvement of RhoA-downstream effectors in the invasion phenotype of active RhoA by utilizing ROCKI and the ROCK inhibitor, Y27632, as well as a set of the effectordomain mutants of RhoA that impair or retain coupling with specific effectors in the invasion assays (Figure 5). Our results confirm an important role of the ROCK pathway in RhoA-mediated invasion of p53/ cells, meanwhile also implicate that multiple effectors are involved in permitting the optimal effect of RhoA. In addition, we found that pericellular proteolysis also play an important role in Rho-induced cell invasion (Figure 6). It will be of particular interests to further investigate how RhoA–ROCK pathway might regulate the extracellular proteases and other extracellular matrix production and/or activation in contributing to the invasion in the future studies. Our results present a seemingly paradoxical relationship between Rho GTPase-regulated pathways and the p53 tumor suppressor pathway. On the one hand, the basal activities of Rho GTPases, Rac1 in particular, elevated by defects in the p53 pathway, are required components of the p53-regulated cell migration and proliferation. On the other, mitogenic activation of Rho proteins can cooperate with p53 deficiency to promote cell invasion and transformation, suggesting that the events leading to Rho hyperactivation may serve as ‘second hits’ in tumor cell metastasis as well as in tumor induction. The observed cooperativities further indicate that components of the Rho protein pathways can deliver additional signals to the cells that suffer p53 pathway defects, enabling them to gain advantages in both movement and proliferation. It is worth noting that RhoC, a closely related Rho protein to RhoA, has been found to be associated with the invasion phenotype. Our current studies utilizing dominant-negative and fast cycling RhoA mutants cannot distinguish the contribution of RhoA from that of RhoC, which may also turn on the ROCK pathway. Given the increasing evidence of overexpression, hyperactivation or rearrangement of multiple Rho GTPases in human cancers including colon, breast, lung, myeloma and head and Oncogene


5584

neck squamous carcinomas (Suwa et al., 1998; Fritz et al., 1999; Clark et al., 2000; Mira et al., 2000; Preudhomme et al., 2000; Schnelzer et al., 2000), it is an attractive proposal that individual Rho GTPase activity and the associated pathways might serve as prognosis markers or useful anticancer therapeutic targets (Sahai and Marshall, 2002). Therefore, identifying the Rho protein and p53-regulated signals that cooperate to enhance changes in cell invasion and proliferation may have important implications in current cancer research. Moreover, in the debate of whether there exist invasion/ metastasis-specific genes (Bernards and Weinberg, 2002), our results support the hypothesis that malfunction of the same set of gene products can confer proliferative selection advantage as well as empower cell invasion and identify the Rho GTPases as a key set of cellular signal transducers capable of mediating both processes. In summary, in this study, we demonstrate that Rho family GTPases Rac1, RhoA and Cdc42 contribute to the p19Arf- and p53-regulated focal complex formation, but not fibronectin production. Significantly, we show that active Rac1, RhoA and Cdc42 mutants can modulate cell–cell adhesion of the p53/ MEFs and cooperate with p53 deletion to promote cell invasion. Moreover, multiple pathways regulated by RhoA including RhoA–ROCK and extracellular proteases are required to synergize with p53 defect to manifest the invasion phenotype. These results help establish yet another important functional connection of Rho GTPases with the p53 pathway, defects of which occur in many human cancer cases.

passage 5) were infected with the respective retroviruses and harvested 48–72 h postinfection. The EGFP-positive cells were isolated by fluorescence-activated cell sorting. Immunoblotting Whole-cell lysates were prepared by extraction of the MEF cells by the lysis buffer containing 20 mM Tris-HCl (pH 7.6), 100 mM NaCl, 10 mM MgCl2, 1% Triton X-100, 0.2% sodium deoxycholate, 2 mM phenylmethylsulfonyl fluoride, 10 mg/ml leupeptin, 10 mg/ml aprotinin and 0.5 mM dithiothreitol, for 30 min. Protein contents in the whole-cell lysates were normalized by the Bradford method. The lysates were separated by 10% SDS–polyacrylamide gel electrophoresis. The ectopic expression of the RhoA effector-binding defective mutants were probed by using an anti-HA antibody (Boehringer Mannheim). Immunofluorescence and cell imaging Primary MEFs were seeded onto coverslips at 3 104 cells/slip density. The cells were serum starved for 12 h prior to fixation. The focal adhesion plaques were visualized by antivinculin (Upstate Biotechnology, Inc.) staining followed by rhodamineconjugated anti-mouse secondary antibody labeling and the fibronectin production was revealed by anti-fibronectin (Sigma) immunostaining or immunoblotting. Stained cells were analysed by using a conventional or a confocal fluorescence microscope (Zeiss). Cell migration assay

Materials and methods

The wound-healing assays to determine semiquantitatively the cell migration rates were performed following a scratch of the confluent monolayer of cells as described (Guo et al., 2003a; Guo and Zheng, 2004). The migration distances of cells were monitored at various times and were measured 9 h after the introduction of the wound under low serum (0.5%) conditions.

DNA constructs

Cell aggregation assay

The Rac1, RhoA and Cdc42 dominant-negative mutants (Rac1N17, RhoAN19 and Cdc42N17) and fast cycling mutants (Rac1L28, RhoAL30 and Cdc42L28) and RhoA effectordomain mutants in the fast cycling backbone (RhoAL30-V39, RhoAL30-L40, RhoAL30-T40 and RhoAL30-C42) were generated by site-directed mutagenesis based on oligonucleotidemediated polymerase chain reaction (Guo et al., 2003a; Guo and Zheng, 2004). For retroviral expression, cDNAs encoding the dominant-negative and fast cycling forms of Rac1, RhoA and Cdc42, oncogenic H-Ras (RasV12), the effector-domain mutants of RhoA and ROCKI were ligated into the BamHI and EcoRI sites in frame with the nucleotides encoding (HA)3tag at the 50 -end of the retroviral vector MIEG3 that expresses EGFP bicistronically. The construct expressing p19Arf or p53 were described previously (Zindy et al., 1998).

Cell aggregation assays were performed as described previously (Guo et al., 2003a; Guo and Zheng, 2004). Briefly, subconfluent cells were washed with PBS and detached with a HCMF buffer (150 mM NaCl, 0.6 mM Na2HPO4, 10 mM glucose and 10 mM HEPES, pH 7.4) containing 0.02% Trypsin and 2 mM CaCl2. Single-cell suspensions were prepared, washed and resuspended in HCMF buffer containing 2 mM CaCl2 at a concentration of 105 cells/ml. A measure of 300-ml aliquots of cell suspension were added to the wells of 24-well plates precoated with 1% BSA and incubated for different times at 371C with 80 r.p.m. agitation. Cell aggregation was measured using an inverted phase-contrast microscope by counting the number of cells in aggregates (three or more cells) in each field over a total of five randomly chosen fields, calculating the mean and standard error of these measurements and expressing the mean relative to the mean total number of cells per field as percent aggregation. Calciumdependent aggregation was calculated by subtracting values obtained from aggregation assays in the presence of EGTA (2 mM) from the values of the total number of aggregating cells.

Cell culture and retroviral induction Primary wild-type, p53/ and p19Arf/ MEFs were derived from mouse embryos as described before (Zindy et al., 1998; C57BL/6x129/sv background) and were cultured in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum, 2 mM glutamine, 0.1 mM nonessential amino acids, 55 mM b-mercaptoethanol and 10 mg/ml gentamycin. Recombinant retroviruses were produced using the Phoenix cell packaging system (Liliental et al., 2000). Low passage primary MEFs (less than Oncogene

Cell invasion assay Cell invasion assays were performed using 6.4 mm Biocoat Matrigel invasion chambers with 8.0 mm pore size PET


5585 membrane (Becton-Dickinson) according to the manufacturer’s instruction. Briefly, 2.5 104 cells were suspended in 0.5 ml of the culture medium and added to the upper chamber. Fetal bovine serum (10%) (Sigma) in the culture medium was used as a chemoattractant in the lower chamber. The cells were incubated for 24 h in humidified incubator at 371C in 10% CO2. The cells that traversed the Matrigel matrix and the 8 mm membrane pores and spread to the lower surface of the

membrane filters were stained with Giemsa stain (Sigma) and quantified by counting under a microscope. Acknowledgements We thank Dr Martine Roussel (St Jude Children’s Research Hospital) for the gifts of MEFs and Arf cDNA. This work was supported in part by a grant from National Institutes of Health (CA105117).

References Alexandrova A, Ivanov A, Chumakov P and Vasiliev J. (2000). Oncogene, 19, 5826–5830. Aznar S and Lacal JC. (2001). Cancer Lett., 165, 1–10. Bar-Sagi B and Hall A. (2000). Cell, 103, 227–238. Bernards R and Weinberg RA. (2002). Nature, 418, 823. Bishop AL and Hall A. (2000). Biochem. J., 348, 241–255. Boettner B and Van Aelst L. (2002). Gene, 286, 155–174. Clark EA, Golub TR, Lander ES and Hynes RO. (2000). Nature, 406, 532–535. Comer KA, Dennis PA, Armstrong L, Catino JJ, Kastan MB and Kumar CC. (1998). Oncogene, 16, 1299–1308. Etienne-Manneville S and Hall A. (2002). Nature, 420, 629–635. Fritz G, Just I and Kaina B. (1999). Int. J. Cancer, 81, 682–687. Fujisawa K, Madaule P, Ishizaki T, Watanabe G, Bito H, Saito Y, Hall A and Narumiya S. (1998). J. Biol. Chem., 273, 18943–18949. Gadea G, Lapasset L, Gauthier-Rouviere C and Roux P. (2002). EMBO J., 21, 2373–2382. Gloushankova N, Ossovskaya V, Vasiliev J, Chumakov P and Kopnin B. (1997). Oncogene, 15, 2985–2989. Guo F and Zheng Y. (2004). Mol. Cell. Biol., 24, 1426–1428. Guo F, Gao Y, Wang L and Zheng Y. (2003a). J. Biol. Chem., 278, 14414–14419. Hahn WC and Weinberg RA. (2002). Nat. Rev. Cancer, 2, 331–341. Iotsova V and Stehelin D. (1996). Cell Growth Differ., 5, 629–634. Kaibuchi K, Kuroda S, Fukata M and Nakagawa M. (1999). Curr. Opin. Cell Biol., 11, 591–596. Kamijo T, Zindy F, Roussel MF, Quelle DE, Downing JR, Ashmun RA, Grosveld G and Sherr CJ. (1997). Cell, 91, 649–659. Levine AJ. (1997). Cell, 88, 323–331. Liliental J, Moon SY, Lesche R, Mamillapalli R, Gavrilova N, Zheng Y, Sun H and Wu H. (2000). Curr. Biol., 10, 401–404. Lin R, Cerione RA and Manor D. (1999). J. Biol. Chem., 274, 23633–23641. Mira JP, Benard V, Groffen J, Sanders LC and Knaus UG. (2000). Proc. Natl. Acad. Sci. USA, 97, 185–189.

Mukhopadhyay D, Tsiokas L and Sukhatme VP. (1995). Cancer Res., 55, 6161–6165. Nobes C and Hall A. (1999). J. Cell Biol., 144, 1235–1245. Preudhomme C, Roumier C, Hildebrand MP, DalleryPrudhomme E, Lantoine D, Lai JL, Daudignon A, Adenis C, Bauters F, Fenaux P, Kerckaert JP and GaliegueZouitina S. (2000). Oncogene, 19, 2023–2032. Ridley AJ. (2001). J. Cell Sci., 114, 2713–2722. Sablina AA, Chumakov PM and Kopnin BP. (2003). J. Biol. Chem., 278, 27362–27371. Sahai E, Alberts AS and Treisman R. (1998). EMBO J., 17, 1350–1361. Sahai E and Marshall CJ. (2002). Nat. Rev. Cancer, 2, 133–142. Sahai E and Marshall CJ. (2003). Nat. Cell Biol., 5, 711–719. Schmitt CA, Fridman JS, Yang M, Baranov E, Hoffman RM and Lowe SW. (2002). Cancer Cell, 1, 289–298. Schnelzer A, Prechtel D, Knaus U, Dehne K, Gerhard M, Graeff H, Harbeck N, Schmitt M and Lengyel E. (2000). Oncogene, 19, 3013–3020. Schwartz MA and Shattil SJ. (2000). Trends Biochem. Sci., 25, 388–391. Serrano M, Lin AW, McCurrach ME, Beach D and Lowe SW. (1997). Cell, 88, 593–602. Sherr CJ. (2001). Nat. Rev. Mol. Cell. Biol., 2, 731–737. Sun Y, Sun Y, Wenger L, Rutter JL, Brinckerhoff CE and Cheung HS. (1999). J. Biol. Chem., 274, 11535–11540. Suwa H, Ohshio G, Imamura T, Watanabe G, Arii S, Imamura M, Narumiya S, Hiai H and Fukumoto M. (1998). Br. J. Cancer, 77, 147–152. Van Aelst L and D’Souza-Schorey C. (1997). Genes Dev., 11, 2295–2322. Zhao R, Gish K, Murphy M, Yin Y, Notterman D, Hoffman WH, Tom E, Mack DH and Levine AJ. (2000). Genes Dev., 14, 981–993. Zheng Y. (2001). Trends Biochem. Sci., 26, 724–732. Zindy F, Eischen CM, Randle DH, Kamijo T, Cleveland JL, Sherr CJ and Roussel MF. (1998). Genes Dev., 12, 2424–2433. Zohn IM, Campbell SL, Khosravi-Far R, Rossman KL and Der CJ. (1998). Oncogene, 17, 1415–1438.

Oncogene