Role for Rhob and PRK in the Suppression of Epithelial Cell Transformation by Farnesyltransferase Inhibitors

Role for RhoB and PRK in the suppression of epithelial cell transformation by farnesyltransferase inhibitors

Ping-Yao Zeng1,5, Neena Rane1,5, Wei Du2, Janaki Chintapalli3 and George C Prendergast3,4

1The Wistar Institute, Philadelphia PA, USA; 2Cancer Research Group, The DuPont Pharmaceuticals Company, Glenolden Laboratory, Glenolden PA, USA; 3The Lankenau Institute for Medical Research, Wynnewood PA, USA; 4Department of Pathology, Anatomy, and Cell Biology, Jefferson Medical School, Thomas Jefferson University, Philadelphia PA, USA

Recent genetic investigations have established that RhoB function of proteins other than Ras (Rowinsky et al., gain-of-function is sufficient to mediate the antitransform- 1999; Sebti and Hamilton, 2000; Prendergast and Oliff, ing effects of farnesyltransferase inhibitors (FTIs) in 2000; Prendergast, 2000a, 2001; Adjei, 2001; Cox, 2001). H-Ras-transformed fibroblast systems. In this study, we One candidate target that has emerged from studies of addressed the breadth and mechanism of RhoB action in the cellular response to FTI is RhoB. Investigations of epithelial cells transformed by oncoproteins which are this endosomal Rho protein have revealed an antionco- themselves insensitive to FTI inactivation. Rat intestinal genic function that is recruited in cells by FTI through a epithelial (RIE) cells transformed by activated K-Ras or gain-of-function mechanism (Prendergast, 2000a, 2001). Rac1 were highly sensitive to FTI-induced actin reorga- Genetic investigations have established that RhoB is nization and growth inhibition, despite the inability of FTI sufficient and/or necessary to mediate many facets of the to block prenylation of either K-Ras or Rac1. Ectopic FTI response displayed in mouse and rodent model expression of the geranylgeranylated RhoB isoform systems or in human cancer cells (Du et al., 1999a; Du elicited in cells by FTI treatment phenocopied these and Prendergast, 1999; Liu et al., 2000). These findings effects. Analysis of RhoB effector domain mutants pointed have been extended recently by the identification of to a role for PRK, a Rho effector kinase implicated in the RhoB as a negative modifier function in cancer cells, physiological function of RhoB in intracellular receptor which is required for the antineoplastic response to trafficking, and these findings were supported further by DNA damaging agents and paclitaxel (Liu et al., experiments in a fibroblast system. We propose that FTIs 2001a, b). Thus, the use of FTIs to probe neoplastic recruit the antioncogenic RhoB protein in the guise of pathophysiology has yielded two new concepts: the RhoB-GG to interfere with signaling by pro-oncogenic concept that Rho signaling affects chemotherapeutic Rho proteins, possibly by sequestering common exchange responses, and the concept that cancer chemotherapeu- factors or effectors such as PRK that are important for tics must recruit negative modifier functions to effec- cell transformation. tively target neoplastic cells. Oncogene (2003) 22, 1124–1134. doi:10.1038/sj.onc.1206181 Important remaining issues regarding RhoB in the FTI response include the following. First, the main Keywords: Ras; Rac; Rho; actin evidence for this link has been obtained in cells transformed by H-Ras, which is a relatively less important human oncogene than K-Ras. Is RhoB involved in the FTI response in K-ras-transformed Introduction cells? This question bears on the long-standing issue of how FTI manages to suppress K-ras transformation Farnesyltransferase inhibitors (FTIs) are a class of without blocking K-ras prenylation (because of the signal transduction inhibitors in cancer clinical trials ability of geranylgeranyltransferase-I (GGT-I) to pre- that are providing a useful new tool to study neoplastic nylate and maintain the function of K-ras in FTI-treated pathophysiology. FTIs were designed to attack Ras cells (Lerner et al., 1997; Rowell et al., 1997; Whyte et al., oncoproteins, the function of which depends on post- 1997). Second, what effector signaling pathways does translational modification by farnesyl isoprenoid. How- RhoB use to mediate its antiproliferative effects in ever, it has become increasingly clear and accepted that transformed cells? Work from Mellor and colleagues has the drugs’ antineoplastic properties can be traced to a defined a function for RhoB in intracellular receptor large degree to an alteration in the prenylation and trafficking (Gampel et al., 1999), which depends on interactions with the Rho effector kinase PRK (Mellor *Correspondence: GC Prendergast, Lankenau Institute for Medical et al., 1998; Flynn et al., 2000), but it is not known Research, 100 Lancaster Avenue, Wynnewood PA 19096, USA; whether this pathway is involved in the antiproliferative E-mail: [email protected], Website: www.limr.org actions of RhoB in transformed cells. This question also 5These authors contributed equally to this study. Received 14August 2002; revised 18 October 2002; accepted 23 bears on the relation between RhoB and its pro- October 2002 oncogenic relatives, the RhoA and RhoC proteins, FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1125 which can also interact with PRK as well as other Rho passaged 48 h after transfection in media containing 100 mg/ml effector proteins (because of the identical effector zeocin, ring cloned, and expanded as cell lines. Transgene domains on the RhoA/B/C proteins). Pro-oncogenic expression was confirmed in cell populations by Western Rho signals are required for transformation by Ras analysis. (Khosravi-Far et al., 1995; Qiu et al., 1995). Marshall and colleagues have proposed that this requirement is Western analysis based on the negation of cell cycle kinase inhibitors Cell extracts prepared in NP40 lysis buffer were analysed by (CDKIs) that are induced by oncogenic Ras (Olson standard methods using a commercial chemiluminescence kit et al., 1998). Because of the identity of the effector to develop Western blots (Pierce). K-RasV12 was detected domains in the RhoA/B/C proteins, the simplest way to with sc-30 mouse monoclonal antibody, Rac1 was detected interpret the antagonistic effects of RhoB on transfor- with sc-217 rabbit antisera, and endogenous RhoB was mation is to propose that it antagonizes signaling detected with sc-180 rabbit antisera (Santa Cruz Biochem- from the pro-oncogenic RhoA/C proteins (Prendergast, icals); HA-tagged RhoB transgene was detected with anti-HA 2001), perhaps by sequestering common Rho effector mouse monoclonal antibody 12CA5 (Roche Biochemicals); PRK2 was detected with anti-PRK2 antibody (Transduction molecules such as PRK kinases. A strong prediction Laboratories); myc epitope-tagged kinase-dead PRK2 was of this model is that FTI should block transformation detected with anti-myc monoclonal antibody 9E10; V5 by pro-oncogenic Rho proteins that are geranylgerany- epitope-tagged PRK2 transgene was detected using an anti- lated. Work in this study and elsewhere (van Golen et al., V5 antibody (Invitrogen). 2002) offers evidence to support this model, which suggests new insights into RhoB function and Cell morphology into its role in the response to FTI and other chemotherapeutics. Cells were seeded in a 10 cm tissue culture dish and photographed 24h later at Â 100 magnification under the conditions noted. In FTI experiments, dimethyl sulfoxide (DMSO) was added as vehicle in control trials to no more than Materials and methods 0.1% (v/v) of the culture media.

Cell culture Cell proliferation and viability assays Rat Intestinal Epithelial (RIE) cells were cultured in Dulbec- Anchorage-dependent cell growth in monolayer cultures was co’s modified Eagle medium (DMEM; Life technologies) measured by MTT (3-(4,5-dimethylthiazoly-2-yl)2,5-diphenyl- supplemented with 5% fetal bovine serum (FBS; Hyclone) tetrazolium bromide) assay (Roche) according to the manu- and antibiotics. Mouse embryo fibroblasts (Liu et al., 2000) facturer’s protocol. Briefly, cells were seeded at 2000 cells per and Rat1/ras cells (Kohl et al., 1993) were cultured in the same well in 96-well culture plates in quadruplicate. At appropriate media except being supplemented with 10% FBS. RIE cells times, 0.5 mg/ml MTT was added to each well and the 96-titer were transformed with an activated K-Ras-4B(G12V) cDNA plate was incubated at 371C in a humidified incubator in which expressed from a neo-tagged pZIP retroviral vector or with an 5% CO2 was maintained. After 4h 100 ml of solubilization activated Rac1(Q61L) cDNA expressed from a neo-tagged buffer (10% SDS in 0.01 m HCl) was added to each well and pcDNA3 vector (Invitrogen). Cells were passaged 48 h after incubated overnight in a humidified atmosphere before transfection in media containing 500 mg/ml G418 and trans- quantification. To monitor anchorage-independent cell pro- formed colonies were ring cloned and expanded as clonal cell liferation, colony formation assays in soft agar culture were lines. Transgene expression in transformed cell lines was performed as described previously (Du et al., 1999a). confirmed by Western analysis. Multiple clones tested for FTI Apoptosis was quantitated by flow cytometry using standard responsiveness yielded highly similar results; representative methods that have been described previously (Liu et al., 2000, data from a single randomly chosen K-Ras or Rac1 clone are 2001a). The proportion of apoptotic cells in a population was presented in the study. RIE/K-ras or RIE/Rac1 derivatives defined as the proportion of cells exhibiting sub-G1 phase that expressed RhoB-GG or RhoB-GG effector mutants were DNA. generated by infection with recombinant retroviruses. For these experiments, a hemagglutinin antigen epitope (HA)- tagged RhoB-GG cDNA that had been described previously Actin immunofluroscence (Du et al., 1999a) was used as a target for mutagenesis and Cells were seeded onto coverslips in six-well dishes and subcloned into the replication-incompetent retroviral vector treated 24h with 10 mm FTI or an equivalent volume of MSCVpac, which includes a puromycin selection marker vehicle. Cells were fixed and stained with fluorescein-phalloidin (Hawley et al., 1994). Effector mutants were generated in the (Molecular Probes) and photographed using indirect immuno- HA-RhoB-GG cDNA using the QuikChanget mutagenesis fluorescence microscopy as described previously (Prendergast kit (Strategene) according to the vendor’s protocol and verified et al., 1994). by DNA sequencing (oligonucleotide sequences used for plasmid construction are available upon request). Recombi- Transcription assay nant MSCVpac viruses were packaged by standard methods in Phoenix cells. After infection, cells were passaged in to A derivative of the SRE reporter construct 3D.ACAT that DMEM media containing 1–4 mg/ml puromycin and drug- employs luciferase instead of chloramphenicol acetyl transfer- resistant cell populations were pooled for analysis. Rat1/ras ase as the reporter gene (Liao et al., 1997) was used to probe cells were transfected with epitope-tagged wild-type or kinase- the effects of various RhoB effector mutations on SRE dead PRK2 expression vectors or with the pcDNA3 vector activation. Transcription assays were performed similarly as alone using the commercial CalPhos mammalian transfection described except using FuGene 6 reagent (Roche Biochem- kit, using the vendor’s instructions (Clontech). Cells were icals) in rhoBÀ/À mouse embryo fibroblasts (Liu et al., 2000).

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1126 The b-galactosidase reporter plasmid CMVÀb gal was used to normalize for transfection efﬁciency (Lebowitz et al., 1997b).

Results

Intestinal epithelial cells transformed by K-Ras or Rac1 are highly susceptible to FTI treatment: association with elevation of RhoB-GG RIE/K-ras cells were generated by transformation with the activated K-Ras mutant K-Ras4B-G12V. The response of these cells to farnesyltransferase inhibition was studied using the FTI L-744,832, a highly potent and specific compound that does not block the activity of cellular geranylgeranyltransferases (Kohl et al., 1995). Control experiments to examine the effects of FTI on untransformed RIE cells employed the matched vector cell line RIE/neo. FTI treatment rapidly elicited an elevation in the levels of the geranylgeranylated isoform of RhoB that is synthesized in drug-treated cells (RhoB- GG), consistent with previous observations (Lebowitz et al., 1997a). RhoB-GG induction was more pronounced in RIE/K-ras cells, but in both cell lines its elevation was detectable as early as 4h after drug treatment, preceding any apparent cellular response (Figure 1a). Within 12–16 h a dramatic shift in the morphology of drug-treated RIE/K-Ras cells became apparent (Figure 1b). The flattened and enlarged morphology observed was highly reminiscent of the phenotypic reversion displayed in H-Ras-transformed fibroblasts, and unexpectedly, owing to the reported ‘resistance’ of K-Ras-transformed fibroblasts to FTI treatment, within the same dose range. While less dramatic, the morphology of untransformed RIE/neo cells was altered similarly with the same kinetics and dose range (Figure 1b). Following work in fibroblast systems, which links phenotypic reversion to a RhoB- dependent reorganization of the actin cytoskeleton (Du et al., 1999a; Liu et al., 2000), we examined the organization of F-actin in drug-treated RIE cells. As Figure 1 Induction of RhoB-GG accompanies phenotypic rever- expected, K-Ras transformation was associated with a sion of K-Ras-transformed RIE cells by FTI. (a) Western analysis. complete disruption of the stress fiber network that Cells were treated for the times indicated with 10 mm FTI L-744,832 before harvest and extract preparation. Blots were probed with a characterized untransformed RIE cells (Figure 1c). rabbit polyclonal anti-RhoB antisera. (b) Morphology. Cells Similar to what is seen in fibroblast models (Prendergast seeded into monolayer culture were photographed 24h after et al., 1994), the morphological shift produced by FTI treatment with 10 mm FTI or vehicle control. (c) Actin organiza- was associated with the reinstatement of a stress fiber tion. Cells seeded onto coverslips were treated as above, fixed, and network in RIE/K-ras cells and with the stimulation of a processed for F-actin staining using FITC-conjugated phalloidin more robust stress fiber network in RIE/neo cells. RhoB-GG is sufficient and necessary to mediate FTI- proliferation of H-Ras-transformed fibroblasts (data induced actin reorganization in H-Ras-transformed not shown). By comparison, the division of RIE/neo fibroblasts (Du et al., 1999a; Liu et al., 2000); the cells was slowed, but not stopped, consistent with induction of RhoB-GG by FTI in RIE cells was previous demonstrations that FTIs more effectively consistent with a similar role here. target the growth of transformed cells. Under ancho- The proliferation of RIE/K-ras cells was also highly rage-independent conditions, where RIE/K-ras, but not sensitive to disruption by FTI treatment, despite the RIE/neo, cells were capable of cell division, a similar noted inability of FTI to block K-Ras prenylation and sensitivity of RIE/K-ras cells to growth suppression by therefore function. The proliferation of RIE/K-ras cells FTI treatment was demonstrated (Figure 2b). Consis- in monolayer culture was virtually halted by FTI tent with observations in K-Ras-transformed NRK treatment (Figure 2a). The IC50 for this effect was fibroblasts (Suzuki et al., 1998), RIE/K-ras cells were indistinguishable from that required to block the susceptible to FTI-induced apoptosis under low growth

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1127

Figure 2 K-ras-transformed RIE cells are highly sensitive to growth suppression and apoptosis by FTI. (a) Anchorage-dependent proliferation. Cell proliferation in monolayer culture was monitored by SRB assay. (b) Anchorage-independent growth. Colony formation in soft agar culture was documented by photomicrography 12 days after seeding 104 cells per well in six-well dishes. (c) Apoptosis assay. The percentage of sub-G1 phase cells as determined by flow cytometry was used to quantify the extent of apoptosis in cell populations treated 24h in monolayer culture with 10 mm FTI. The percentage in each panel notes the proportion of the population that scored in the sub-G1 phase factor conditions (data not shown). The dose required to RIE/K-ras cells (Figure 3d). However, under anchorage- suppress colony formation or induce apoptosis under independent conditions, growth suppression was more these different conditions (1–5 mm L-744,832) was in the robust and comparable to that seen in RIE/K-ras cells same range as that determined in assays using H-Ras- (Figure 3e). As before, the antiproliferative effects were transformed fibroblasts. These observations also achieved at dose ranges that were comparable to those contrasted with the reported ‘resistance’ of K-Ras- required to block the anchorage-independent growth of transformed fibroblasts to FTI treatment. We concluded H-Ras-transformed fibroblasts. We were unable to that, despite the well-documented inability of FTI to perform the apoptosis experiment, because RIE/rac1 block K-Ras prenylation, because of its alternate cells underwent apoptosis when deprived of serum geranylgeranylation in FTI-treated cells (Lerner et al., growth factors alone (data not shown). Nevertheless, 1997; Rowell et al., 1997; Whyte et al., 1997), RIE/K-ras these observations indicated that RIE1/rac1 cells were cells were fully susceptible to the antitransforming susceptible to the antitransforming effects of FTI, effects of FTI treatment. despite its inability to block the geranylgeranylation To explicitly test whether FTI could bypass a and therefore the function of the Rac1 oncoprotein. We geranylgeranylated oncoprotein to suppress cell trans- concluded that FTI could bypass geranylgeranylated formation, we investigated the response of Rac1- ‘driver’ oncoproteins to elicit reversion and suppress transformed cells to FTI treatment. Rac1 is solely proliferation in transformed cells. geranylgeranylated in cells, so FTI L-744,832 cannot affect the function of this oncoprotein. The activated RhoB-GG phenocopies the actin reorganization and mutant Rac1-Q61L was used to generate transformed growth suppression produced by farnesyltransferase RIE/rac1 cells. As before, FTI treatment rapidly elicited inhibition in transformed intestinal epithelial cells an elevation in the cellular levels of RhoB-GG within 4h of drug treatment, preceding any apparent cellular In H-Ras-transformed fibroblast model systems, the response (Figure 3a). The phenotype of RIE/rac1 cells RhoB-GG elicited by FTI is sufficient and necessary to was comparatively flat, relative to RIE/neo control cells, mediate actin reorganization and sufficient to suppress but pronounced membrane ruffles were apparent that proliferation (Du et al., 1999a; Liu et al., 2000). The gave the cells a hard-edged appearance. Within 12–16 h ability of FTI to elicit RhoB-GG in RIE cells prompted of drug treatment, a shift in the morphological us to confirm whether ectopic RhoB-GG was sufficient phenotype could be seen, with a loss of the pronounced to phenocopy the FTI response in these cells. RIE/K-ras membrane ruffles and a subtle flattening of cell shape or RIE/neo cell populations that stably expressed (Figure 3b). F-actin staining revealed the presence of RhoB-GG or vector sequences were generated by both ruffles (blurred in the figure) and stress fibers in infection with recombinant retroviruses (Figure 4a). untreated RIE/rac1 cells (Figure 3c). More pronounced RhoB-GG expression caused a slight flattening of RIE/ pseudopods were also apparent. Interestingly, FTI neo cells, increasing their size to a degree similar to FTI treatment diminished the ruffles and pseudopods but treatment (Figure 4b, top panels). RhoB-GG also had little overt effect on stress fiber organization caused a flattening of RIE/K-ras cells, but only to a (Figure 3c). The proliferation of RIE/rac1 cells in partial extent that did not fully mimic the effects of FTI monolayer culture was suppressed by FTI treatment, treatment (Figure 4b, bottom panels). The partial effect but under these conditions not as robust as produced in seen might reflect a dose response; FTI induces

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1128

Figure 3 Induction of RhoB-GG accompanies actin reorganization and growth suppression of Rac-transformed RIE cells by FTI. (a) Western analysis. Cells were treated 4h with 10 mm FTI L-744,832 before harvest and extract preparation. Blots were probed with anti- RhoB antisera as before. (b) Morphology and actin organization. Cells seeded into monolayer culture were photographed 24h after treatment with 10 mm FTI or vehicle control. For actin staining, cells seeded onto coverslips were treated and processed as before using FITC-conjugated phalloidin. (c) Anchorage-dependent proliferation. Cell proliferation in monolayer culture was monitored by SRB assay as before. (d) Anchorage-independent growth. Colony formation in soft agar culture was documented by photomicrography 14 days after seeding 104 cells per well in six-well dishes. (e) Apoptosis assay. The extent of apoptosis in cell populations treated 24h in monolayer culture with 10 mm FTI was determined as above. Serum deprivation by itself triggered apoptosis in RIE/rac cells and FTI accentuated this response slightly. The percentage in each panel notes the proportion of the population that scored in the sub-G1 phase

endogenous RhoB-GG to much higher levels in RIE/K- We concluded that the induction of RhoB-GG by FTI ras cells than RIE/neo cells (Figure 1a), so the dose of was a sufficient cause to mediate actin stress fiber ectopic RhoB-GG might reach the dose needed to formation and growth inhibition. induce the full morphological shift in RIE/neo cells, but not in RIE/K-ras cells. Alternately, RhoB-GG might Evidence of a role for the RhoB effector kinase PRK in not be sufficient to phenocopy the morphological shift the antiproliferative effects of farnesyltransferase fully. The latter interpretation is perhaps more appeal- inhibition in transformed cells ing, insofar as ectopic RhoB-GG could phenocopy the actin stress fiber induction to a more similar extent in The results above suggest that RhoB-GG indirectly both RIE/neo and RIE/K-ras cells (Figure 4c). This interferes with cell transformation by K-Ras or Rac1, result argues that the induction of endogenous RhoB- consistent with other evidence that RhoB has antionco- GG by FTI is sufficient to mediate actin reorganization. genic role in cells (Liu et al., 2001a, b). To probe the Ectopic RhoB-GG also effectively phenocopied the basis for its activity, we investigated the effect of growth suppression of RIE/K-ras cells by FTI, both in various effector domain mutations in RhoB-GG. monolayer culture (data not shown) and under ancho- RhoB is B90% identical to RhoA in its primary rage-independent conditions (Figure 4d). Similar anti- structure and, in particular, the effector domains proliferative effects of RhoB-GG were observed under of these small GTPases are identical. Thus, effector the same conditions in RIE/rac1 cells (data not shown). molecules that have been defined through interaction

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1129

Figure 4 RhoB-GG phenocopies actin reorganization and growth suppression by FTI. (a) Western analysis. Expression of ectopic HA-RhoB-GG in extracts from cells that were stably infected by the puromycin-resistant retroviral vectors indicated was conﬁrmed by blotting with anti-HA antibody, which recognizes the epitope tag on the transgene. (b) Morphology. Cells in monolayer culture were photographed 24h after seeding in the presence or absence of 10 mm FTI as indicated. (c) Actin organization. Cells seeded onto coverslips were treated as above and processed as before for actin staining with FITC-conjugated phalloidin. (c) Anchorage- independent growth. Colony formation in soft agar culture was documented by photomicrography 14days after seeding 10 4 cells per well in six-well dishes. Vector cells were treated where indicated with 10 mm FTI as before

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1130 Table 1 PRK interactions correlate with the ability of RhoB-GG to inhibit transformed cell growth and to activate SRE-dependent transcription. The actin, growth inhibition, and SRE activation results are summarized from Figures 5 and 6. The Rho effector binding data is adapted from Sahai et al. (1998); effector domain sequences in RhoA and RhoB are identical RhoB mutant Actin Growth Inhibition SRF activation PRK ROCK Rhophilin Kinectin Citron mDia mNET Wild-type RhoB-GG ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ T37Y ++ – – – – – +/À ––– F39A ++ – – – – – – – ++ – F39L + ++ ++ + + + – – ++ + F39V ++ – – – ++ – – – ++ ++ E40T ++ + ++ + ++ ++ – ++ ++ + Y42C ++ – – – ++ + + ++ ++ +

Figure 5 Effects of effector domain mutations on growth suppression and actin organization by RhoB-GG. (a) RIE/K-ras cell populations that were derived from infections with retroviral vectors expressing no insert (vector), RhoB-GG, or the RhoB-GG effector domain mutants indicated were seeded into soft agar culture. Colony formation was documented by photomicrography 12 days after seeding. (b) The same RIE/K-ras cell populations derived above were seeded onto coverslips and processed the next day for actin staining with FITC-conjugated phalloidin as before

with RhoA have in every case tested also interacted with The set of effector domain mutants tested discrimi- RhoB, as would be expected on the basis of structural nated between possible mechanisms of growth suppres- considerations. In a retroviral vector, we introduced into sion and SRE activation, but not actin reorganization. RhoB-GG six different effector domain mutations that All the effector mutants were able to induce actin stress have been shown in RhoA to affect interactions with fibers to a considerable degree in RIE/K-ras cells various Rho effector molecules to variable degrees (Figure 5a). These observations included the T37Y (Table 1). RIE/K-ras cell populations that expressed mutant that binds weakly to most Rho effector these mutant derivatives were then examined in the molecules (Sahai et al., 1998). This result argued that assays for actin organization and anchorage-indepen- interactions with several different effectors could trigger dent growth used above. Since the response of cells to actin reorganization, any one of which may be sufficient. FTI and RhoB is influenced by serum growth factors In contrast to this result, two mutants clearly discrimi- (Liu et al., 2000), we also tested the effector mutants for nated interactions with the Rho effector kinase PRK as their ability to activate transcription of the serum important for suppression of anchorage-dependent response element (SRE), a transcriptional target of growth and SRE activation (Figure 5b and Figure 6). RhoB (Lebowitz et al., 1997b). The results of these The patterns of activity for F39L and E40T, mutants experiments are summarized in Table 1 and shown in which retained the ability to bind PRK, linked Figures 5 and 6. productive interactions with RhoB-GG to its ability to

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1131 deprivation for FTI-induced apoptosis. Myr-PRK2-expressing cells were resistant to FTI-induced apoptosis. Myristylation relieves the requirement for Rho to mediate membrane recruitment. Therefore, the ability of myr- PRK2 to block apoptosis suggested that the proapoptotic effects of PRK2 in the FTI response were linked not only to its kinase activity, but also to its localization and/or movement in cells. This observation was notable in light of the evidence that FTI alters the localization as well as the level of RhoB-GG in cells, and that both changes might be needed for the FTI response (Adamson et al., 1992; Lebowitz et al., 1995; Du et al., 1999a). Taken together with the RIE results, these experiments supported a role for PRK activity and localization in the RhoB-mediated facets of the FTI response.

Figure 6 Effects of effector domain mutations on SRE-dependent Discussion transcriptional activation by RhoB-GG. Normalized luciferase activity was determined in cell extracts prepared 48 h after The results of this study offer direct evidence that RhoB transfection of NIH3T3 cells with an SRE-luciferase reporter plasmid and vectors expressing no insert (vector), RhoB-GG, or gain-of-function is a sufficient cause for FTI to bypass the RhoB-GG effector domain mutant indicated the action of farnesyl-independent oncoproteins in suppressing cell transformation. The work also strengthens the connection between RhoB and the FTI suppress proliferation and activate transcription response by confirming its relevance in cells of epithelial (Table 1). This connection was satisfying, insofar as origin, where actin-dependent polarity signals that the results were highly consistent with the evidence that impact transformation and proliferation processes likely PRK interaction is crucial for the physiological function occur through distinct mechanisms that are irrelevant in of RhoB in the intracellular trafficking of the EGF fibroblasts. Lastly, the results also support a role for receptor and perhaps other receptors (Mellor et al., interactions with the Rho effector protein PRK in 1998; Gampel et al., 1999). mediating RhoB-dependent facets of the FTI response. To further probe the relation of PRK to the FTI PRK is a critical effector for the physiological function response, we expressed it in Rat1/ras fibroblasts, an H- of RhoB in intracellular receptor trafficking, and our Ras-transformed cell system that has been widely used observations are consistent with a link between this for FTI studies (Kohl et al., 1993; Prendergast et al., function and the antiproliferative effects of RhoB. 1994; Lebowitz et al., 1995, 1997c; Du et al., 1999a ,b). K-ras-transformed RIE cells were highly susceptible For these experiments, we used PRK2 since this PRK to FTI treatment, lacking the relative ‘resistance’ isoform is expressed in Rat1/ras cells; both PRK1 and observed by some laboratories in K-ras-transformed PRK2 are recruited similarly by RhoB to endosomal fibroblasts (compared to H-Ras-transformed fibro- vesicles (Mellor et al., 1998; Gampel et al., 1999) (H blasts). Although this resistance has been ascribed to Mellor, pers. comm.). To probe possible effects of the the relatively higher avidity of K-Ras for FT, compared PRK kinase domain or membrane recruitment, we also to H-Ras, our observations suggest that the ‘resistance’ expressed a kinase-dead mutant (PRK2-KD) or an N- phenomenon is a cell-type-specific property unrelated to myristoylated derivative of PRK2 in Rat1/ras cells. the avidity of different Ras proteins for FT. Indeed, Several clones analysed for each Rat1/ras derivative since the oncogenic activity of K-Ras remains intact acted similarly; results from representative clones are when the protein is geranylgeranylated, as it is in FTI- presented here (Figure 7a). Interestingly, ectopic PRK2 treated cells, the ability of FTI to suppress K-ras relieved the requirement for serum deprival for FTI to transformation is more readily explained by effects on induce apoptosis in Rat1/ras cells. Under control non-Ras targets. Our observations rule in RhoB gain-of- conditions (i.e. monolayer culture in media containing function as a sufficient cause for explaining the 10% FBS), FTI treatment induces reversion and growth suppression of K-ras transformation by FTI. inhibition in Rat1/ras fibroblasts, whereas if serum Another illustration that FTI can bypass geranylger- growth factors are deprived, FTI treatment triggers an anylated oncoproteins to suppress transformation was apoptotic program (Du et al., 1999c). In contrast to provided by the demonstration that RIE/rac cells were vector control cells, which acted like the parental cells, susceptible to FTI. Rac1 is solely geranylgeranylated in PRK2-expressing cells underwent apoptosis in serum- cells, so FTI must act by indirectly blocking transforma- containing media as evidenced in phase micrographs or tion; the induction of RhoB-GG and its ability to flow cytometry (Figure 7b,c). The kinase activity of phenocopy FTI effects as in RIE/K-ras cells is PRK2 was implicated, because cells expressing the consistent with a role for RhoB in the drug response. kinase-dead PRK-KD mutant did not recapitulate the These observations have been corroborated and ex- effect, instead retaining the requirement of serum tended still further by the demonstration that FTI

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1132

Figure 7 Impact of PRK activity and membrane association on the FTI response of H-Ras-transformed ﬁbroblasts. (a) Western analysis. Expression of PRK2, PRK2-KD (myc epitope-tagged kinase-dead PRK2 mutant), or myr-PRK2 (V5 epitope-tagged N- myrisylated PRK) in extracts prepared from Rat1/ras cell derivatives was documented on Western blots probed with anti-PRK2, anti- myc 9E10, or anti-V5 antibodies, respectively. (b) Morphology. Rat1/ras cell clones indicated were incubated 8 h after seeding at 106 cells per 10 cm dish in DMEM containing 10% FBS or 0.1% FBS. The next day 10 mm FTI or an equivalent volume of DMSO vehicle (control) was added and 24h later cell morphology was documented by photomicrography. ( c) Flow cytometry. Rat1/ras cell clones indicated were treated as above and harvested for ﬂow cytometry 24h after FTI treatment

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1133 treatment both induces RhoB and suppresses RhoC- The most informative mutations in the RhoB effector dependent transformation of HMECs (van Golen et al., domain that were tested in this study were the F39L and 2002). This result has clinical implications, because F39V mutations, which differentially distinguish PRK RhoC overexpression occurs in >90% of inflammatory interaction (Table 1). RhoB-GG-mediated growth sup- breast cancers and ectopic RhoC recapitulates pheno- pression was maintained by F39L mutation, which type of these cancers (van Golen et al., 1999, 2000a, b). supports PRK interactions, but was abolished by F39V FTI was shown to inhibit the anchorage-independent mutation, which abolishes PRK interaction. Notably, growth, motility and invasion of RhoC-transformed the same mutations made in the pro-oncogenic RhoA cells, and ectopic RhoB was sufficient to phenocopy protein abolish its cell-transforming activity with Raf N- these effects (van Golen et al., 2002). Like Rac1, terminal truncates (Sahai et al., 1998; Zohar et al., RhoC is solely geranylgeranylated in cells, so these 1998). Thus, F39L mutation suppresses the transform- observations strongly reinforce the notion that FTIs ing activity of RhoA, whereas F39V supports this interfere with pro-oncogenic Rho signals by eliciting activity. Taken together, these results suggest that the RhoB-GG. antioncogenic function of RhoB is based on its We obtained genetic evidence supporting a role for competition with RhoA for PRK binding, insofar as PRK in RhoB-dependent facets of the FTI response. RhoB must bind PRK to suppress transformation and Effector mutations that abolished PRK interactions RhoA must bind PRK to drive transformation. The effectively limited the ability of RhoB-GG to suppress induction and relocalization of RhoB-GG caused by anchorage-independent growth or to activate SRE FTI treatment may sequester PRK, inhibiting its ability transcription. One implication is that antiproliferative to bind RhoA and mediate pro-oncogenic signals. This effects of RhoB may involve downstream transcriptional model is consistent with the Rat1/ras results, which events, a possibility supported by evidence that FTI- suggested that PRK localization or movement is induced reversion in Rat1/ras cells requires derepression important to the FTI response. It is also consistent with of the collagen Ia2 gene (Du et al., 1999b). Another the evidence that FTI elicits a relocalization of RhoB implication is that the antiproliferative and transcrip- (i.e. RhoB-GG) in drug-treated cells (Lebowitz et al., tional effects of RhoB are derivative of the physiological 1995) (H. Mellor, pers. comm.). As mentioned above, function of RhoB in intracellular receptor trafficking, Ras transformation requires pro-oncogenic Rho signals. which also depends upon PRK interaction (Mellor et al., Therefore, by eliciting RhoB-GG, FTI may interfere 1998; Gampel et al., 1999; Flynn et al., 2000). with pro-oncogenic Rho signals to block Ras transfor- Experiments in a fibroblast model system supported a mation (see Figure 8). This model can also offer an role for PRK in the FTI response. Elevation of PRK explanation for how RhoB-GG mediates the antagonis- relieved the serum deprival requirement for FTI-induced tic effects of FTI on RhoC transformation (van Golen apoptosis, in a manner dependent on the integrity of the PRK kinase domain, whereas elevation of myristylated PRK (which does not require Rho interaction for membrane localization) blocked FTI-induced apoptosis. The latter observation argued that PRK localization is important: the ability of PRK to block FTI-induced apoptosis when PRK was localized to membranes in a Rho-independent fashion argued that movement away from membrane-associated sites was integral to its participation in the FTI response. This interpretation of the results is consistent with the evidence that FTI both elevates and relocalizes RhoB in cells (Lebowitz et al., 1995). Although PRK2 has been reported to be cleaved and activated by caspases during apoptosis (Cryns et al., 1997; Takahashi et al., 1998), we did not observe this event. A link between PRK and FTI- induced apoptosis is potentially interesting based on reports of the ability of PRK to complex with RhoB and Figure 8 Model. RhoB is a short-lived protein with a rapid the Akt-regulatory enzyme PDK1 in certain epithelial synthesis rate. FTI blocks the ability to replace RhoB-F, which is and endothelial cells (Balendran et al., 1999; Flynn et al., turned over. Relative levels of the geranylgeranylated RhoB isoform 2000) (L. Benjamin, pers. comm.), the potential of increase as the farnesylated isoform is depleted (Figure 1a). This event is also associated with a change in subcellular localization cleaved PRK to modulate Akt activity (Koh et al., (Lebowitz et al., 1995). Elevation and movement of the antionco- 2000), and the ability of RhoB-GG to phenocopy genic RhoB protein may lead to competition for regulators or the inhibition of Akt activity by FTI in epithelial effectors that normally interact with the pro-oncogenic RhoA or and endothelial cells (Liu and Prendergast, 2000) RhoC proteins, such as PRK. Such competition may attenuate pro- (L Benjamin, pers. comm.). Further investigations may oncogenic Rho signals owing to sequesteration of certain effectors needed for cell transformation. In the model, RhoB accesses shed additional light on the signaling mechanisms that redundant effector molecules that can drive actin stress fiber relate PRK to the RhoB-dependent aspects of the FTI formation, separate from effectors that may be needed for response. transformation

Oncogene FTI suppresses epithelial transformation via RhoB and PRK P-Y Zeng et al 1134 et al., 2002), through the ability of RhoB-GG to suggested (Prendergast, 2000b; Prendergast and Rane, compete with RhoC for PRK interactions. The notion 2001). that RhoB antioncogenic activity is based on competition for effectors or nucleotide exchange factors (GEFs) of pro-oncogenic Rho proteins is consistent with Acknowledgements structural considerations and offers considerable expla- We thank Adrienne Cox for providing RIE cells, Channing Der for the human K-Ras4B-V12 cDNA, Alan Hall for Rac1- native and predictive power. The utility of FTI to Q61L cDNA, and Diane Sharp and Jeffrey Faust for interfere with Rho signaling through these mechanisms performing flow cytometry. NR was supported by a post- may find clinical utility in novel disease settings that doctoral research grant from the US Army Breast Cancer involve Rho, including inflammatory breast cancer, Research Program. This work was supported in part by NIH melanoma, and age-associated blindness as we have Grant CA82222 (GCP).

References

Adamson P, Paterson HF and Hall A. (1992). J. Cell Biol., Liu A-X, Du W, Liu J-P, Jessell TM and Prendergast GC. 119, 617–627. (2000). Mol. Cell. Biol., 20, 6105–6113. Adjei AA. (2001). J. Nat. Cancer Inst., 93, 1062–1074. Liu A-X and Prendergast GC. (2000). FEBS Lett., 481, Balendran A, Casamayor A, Deak M, Paterson A, Gaffney P, 205–208. Currie R, Downes CP and Alessi DR. (1999). Curr. Biol., 9, Liu, A-X, Rane N, Liu J-P, Prendergast GC. (2001b). Mol. 393–404. Cell. Biol., 21, 6906–6912. Cox AD. (2001). Drugs, 61, 723–732. Mellor J, Flynn P, Nobes CD, Hall A and Parker PJ. (1998). Cryns VL, Byun Y, Rana A, Mellor H, Lustig KD, Ghanem J. Biol. Chem., 273, 4811–4814. L, Parker PJ, Kirschner MW and Yuan J. (1997). J. Biol. Olson MF, Paterson HF and Marshall CJ. (1998). Nature, 394, Chem., 272, 29449–29453. 295–299. Du W, Lebowitz P and Prendergast GC. (1999a). Mol. Cell. Prendergast GC. (2000a). Curr. Opin. Cell Biol., 12, 166–173. Biol., 19, 1831–1840. Prendergast GC. (2000b). Lancet Oncol., 1, 73. Du W, Lebowitz PF and Prendergast GC. (1999b). Cancer Prendergast GC. (2001). Nature Rev. Cancer, 1, 162–168. Res., 59, 2059–2063. Prendergast GC, Davide JP, deSolms SJ, Giuliani E, Graham Du W, Liu A and Prendergast GC. (1999c). Cancer Res., 59, S, Gibbs JB, Oliff A and Kohl NE. (1994). Mol. Cell. Biol., 4808–4812. 14, 4193–4202. Du W and Prendergast GC. (1999). Cancer Res., 59, 5924– Prendergast GC and Oliff A. (2000). Sem. Cancer Biol., 10, 5928. 443–452. Flynn P, Mellor H, Casamassima A and Parker PJ. (2000). Prendergast GC and Rane N. (2001). Expert Opin. Investig. J. Biol. Chem., 275, 11064–11070. Drugs, 10, 2105–2116. Gampel A, Parker PJ and Mellor H. (1999). Curr. Biol., 9, Qiu RG, Chen J, McCormick F and Symons M. (1995). Proc. 955–958. Natl. Acad. Sci. USA, 92, 11781–11785. Hawley RG, Lieu FH, Fong AZ and Hawley TS. (1994). Gene Rowell CA, Kowalczyk JJ, Lewis MD and Garcia AM. (1997). Ther., 1, 136–138. J. Biol. Chem., 272, 14093–14097. Khosravi-Far R, Solski PA, Clark GJ, Kinch MS and Der CJ. Rowinsky EK, Windle JJ and Von Hoff DD. (1999). J. Clin. (1995). Mol. Cell. Biol., 15, 6443–6453. Oncol., 17, 3631–3652. Koh H, Lee KH, Kim D, Kim S, Kim JW and Chung J. Sahai E, Alberts AS and Treisman R. (1998). EMBO J., 17, (2000). J. Biol. Chem., 275, 34451–34458. 1350–1361. Kohl NE, Mosser SD, deSolms SJ, Giuliani EA, Pompliano Sebti SM and Hamilton AD. (2000). Oncogene, 19, 6584–6593. DL, Graham SL, Smith RL, Scolnick EM, Oliff A and Suzuki N, Urano J and Tamanoi F. (1998). Proc. Natl. Acad. Gibbs JB. (1993). Science, 260, 1934–1937. Sci. USA, 95, 15356–15361. Kohl NE, Omer CA, Conner MW, Anthony NJ, Davide JP, Takahashi M, Mukai H, Toshimori M, Miyamoto M and deSolms SJ, Giuliani EA, Gomez RP, Graham SL, Ono Y. (1998). Proc. Natl. Acad. Sci. USA, 95, 11566– Hamilton K, Handt LK, Hartman GD, Koblan KS, Kral 11571. AM, Miller PJ, Mosser SD, O’Neill TJ, Rands E, Schaber van Golen KL, Bao LW, DeVito MM, Wu Z-F, Prendergast MD, Gibbs JB and Oliff A. (1995). Nat. Med., 1, 792–797. GC and Merajver SD. (2002). Mol. Cancer Ther., 1, 575– Lebowitz P, Casey PJ, Prendergast GC and Thissen J. (1997a). 583. J. Biol. Chem., 272, 15591–15594. van Golen KL, Davies S, Wu ZF, Wnag Y, Bucana CD, Lebowitz P, Du W and Prendergast GC. (1997b). J. Biol. Root H, Chandrasekharappa S, Strawderman M, Ethier Chem., 272, 16093–16096. SP and Merajver SD. (1999). Clin. Cancer Res., 5, 2511– Lebowitz PF, Davide JP and Prendergast GC. (1995). Mol. 2519. Cell. Biol., 15, 6613–6622. van Golen KL, Wu Z-F, Qiao X, Bao L and Merajver SD. Lebowitz PF, Sakamuro D and Prendergast GC. (1997c). (2000a). Neoplasia, 2, 418–425. Cancer Res., 57, 708–713. van Golen KL, Wu Z-F, Qiao XT, Bao LW and Merajver SD. Lerner EC, Zhang TT, Knowles DB, Qian YM, Hamilton AD (2000b). Cancer Res., 60, 5832–5838. and Sebti SM. (1997). Oncogene, 15, 1283–1288. Whyte DB, Kirschmeier P, Hockenberry TN, Nunez-Olivia I, Liao J, Hodge C, Meyer D, Ho PS, Rosenspire K and James L, Catino JJ, Bishop WR and Pai JK. (1997). J. Biol. Schwartz J. (1997). J. Biol. Chem., 272, 25951–25958. Chem., 272, 14459–14464. Liu A-X, Cerniglia GJ, Bernhard EJ and Prendergast GC. Zohar M, Teramoto H, Katz B-Z, Yamada KM and Gutkind (2001a). Proc. Natl. Acad. Sci. USA, 98, 6192–6197. JS. (1998). Oncogene, 17, 991–998.

Oncogene