INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Farnesyltransferase inhibitors (FTIs) are a novel class of antitumor agents whose development was based upon the discovery that posttranslational prenylation is required for the oncogenic properties of Ras (reviewed in references 17, 18, 40, and 56). Protein prenylation involves C-terminal addition of C₁₅ (farnesyl) or C₂₀ (geranylgeranyl) isoprenoids, each of them intermediates in cholesterol biosynthesis. Protein prenylation reactions are carried out by one of three enzymes in the cell: farnesyltransferase (FT), geranylgeranyltransferase type I (GGT-I), or geranylgeranyltransferase type II (GGT-II; Rab GGT). FT and GGT-I are related heterodimeric enzymes that share a common subunit. They mediate prenylation of members of the Ras and Rho subfamilies of the Ras superfamily of small GTPases that include C-terminal CAAX prenylation motifs. GGT-II is an enzyme that is unrelated to the FT and GGT-I. It mediates geranylgeranylation of members of the Rab subfamily of Ras superfamily small GTPases through a reaction that is mechanistically distinct from the reactions catalyzed by FT or GGT-I (7, 71). Prenylation facilitates association with cellular membranes and mediates protein-protein interactions (71). Geranylgeranylation is the predominant type of prenylation in cells. It is unclear why two types of prenylation occur, but there are examples in which protein function can be altered by switching prenylation type (11, 34).

Farnesylation of Ras proteins is the crucial modification for oncogenicity (28). Therefore, compounds which specifically inhibit FT were sought as a strategy to block Ras function and suppress the growth of Ras-dependent tumors while leaving cellular geranylgeranylation intact (17). While the ultimate clinical potential of this strategy has yet to be assessed, proof-of-principle cell culture and animal experiments have firmly established the ability of FTIs to effectively reverse Ras-dependent cell transformation and to impede tumorigenesis (27, 30, 32, 44, 46, 49, 57, 62). In particular, studies with Ras oncomice models have offered dramatic examples of tumor regression in the absence of detectable toxic side effects, indicating that FTIs pinpoint a specific feature of neoplastic pathophysiology (4, 31, 43).

Notably, recent investigations into the biological mechanisms that underlie FTI treatment have raised questions about their exact mode of action (reviewed in reference 38). While it is quite clear that FTIs act by specifically inhibiting FT activity, it is much less clear that inhibiting Ras farnesylation is essential for the drugs' antitransforming effects. First, the kinetics of phenotypic reversion of Ras transformation are too rapid to be explained by loss of the function of Ras through inhibition of its farnesylation. Reversion is largely complete within 24 h of cell treatment (49), even though Ras has a half-life of ~24 h (60) and is only partially depleted by the time reversion is complete (49). Second, soluble forms of oncogenic Ras generated in drug-treated cells do not accumulate to steady-state levels which are sufficient to interfere with prenylated Ras (49) and, in any case, only the Ras L61 but not the Ras V12 mutant allele used in published experimental models can exert dominant negative effects (19). Third, FTIs can inhibit the anchorage-independent growth of cells transformed with oncogenic Ras proteins engineered to function independently of farnesylation, due to N-myristylation or geranylgeranylation (10, 37, 47a). Similarly, K-Ras-transformed cells are susceptible to growth inhibition, despite the fact that FTIs do not inhibit K-Ras prenylation due to geranylgeranylation of K-Ras by GGT-I when FT activity is blocked in cells (41, 53, 67). Lastly, there is no correlation between the susceptibility of human tumor cell lines to growth inhibition by FTIs and their Ras status (57). Thus, the biological susceptibility to FTIs can be separated to a significant degree from Ras inhibition. These investigations have stimulated efforts to identify farnesylated proteins other than Ras whose functional alteration is germane to the drugs' antitransforming mechanism (9, 38).

Our previous work in this area led us to suggest an alternate model for drug action termed the FTI-Rho hypothesis (38). The FTI-Rho hypothesis proposes that the antitransforming effects of FTIs are mediated at least in part through altering the function of farnesylated Rho proteins, including RhoB (34, 37, 39, 49, 50). Rho proteins are a family of small GTPases that are required for Ras transformation (29, 50, 51) and that regulate cytoskeletal actin, focal adhesion formation, cell adhesion signaling, and transcription (reviewed in references 63, 64, and 66). RhoB is closely related to RhoA, but it is differently localized, regulated, and prenylated. RhoB is localized in endosomes where it could conceivably participate in receptor-mediated endocytosis events where Rho has been implicated (33, 55). RhoB is short-lived and part of the immediate-early genetic response to v-Src and epidermal growth factor that may contribute to regulating cell cycle progression (25, 37, 70). However, RhoB also has cell cycle inhibitory roles suggested by its upregulation by UV irradiation and the stress response-associated kinases p38 and JNK (15, 16) and by its ability to govern transforming growth factor  (TGF-)-regulated transcription (14).

RhoB is unique among prenylated proteins in that it exists normally in vivo in two populations that are either farnesylated or geranylgeranylated (RhoB-F or RhoB-GG) (1). FT is responsible for the generation of RhoB-F, whereas GGT-I is responsible for the generation of RhoB-GG (34). RhoB has a half-life of only ~2 h (37), so the steady-state levels of RhoB-F decrease rapidly in FTI-treated cells. However, there is a simultaneous elevation in the steady-state levels of RhoB-GG, and a great increase in the ratio of RhoB-GG to RhoB-F in cells, because newly synthesized RhoB still serves as a substrate for GGT-I. The increased levels of RhoB-GG therefore represent a biochemical gain-of-function effect of FTI treatment. Interestingly, this shift in RhoB prenylation was associated with a loss of RhoB's ability to promote cell proliferation (34). This observation implied that the functions of RhoB-F and RhoB-GG might be dissimilar and that cell proliferation may be influenced by altering the ratio of RhoB-F to RhoB-GG. In this study, we tested the hypothesis that elevation of RhoB-GG could mimic the ability of FTIs to specifically inhibit the growth of Ras-transformed cells. In support of this hypothesis, we found that RhoB-GG was not only inhibitory but also caused a phenotypic reversion which was similar in its features to that produced by FTI treatment. The conclusions of this study indicate that FTIs act in part through gain-of-function effects. Our findings suggest that Rho regulates oncogenic Ras biology and offers a Ras-independent mechanism for how FTIs are able to broadly inhibit the growth of human tumor cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Plasmid constructions. An epitope-tagged RhoB polypeptide (HA-RhoB-GG) that is preferentially geranylgeranylated in cells was constructed in the following manner. A HindIII-BstXI restriction fragment was excised from cytomegalovirus (CMV) HA-RhoB-WT (34) to provide the RhoB N terminus, including an influenza hemagglutinin (HA) epitope. A BstXI-EcoRI restriction fragment was excised from RhoB-A^CT (2) to provide the remainder of RhoB and a modified C terminus which substituted the 13 C-terminal residues of RhoA in place of the 16 C-terminal residues of RhoB. The 13 residues derived from the RhoA C terminus include the CAAX box sequences responsible for prenyltransferase specificity (7), and they are sufficient for directing geranylgeranylation of RhoB-A^CT in cells (2). The recombinant product, designated HA-RhoB-GG, was generated by ligation of the restriction fragments into the HindIII-EcoRI site of pcDNA3.1zeo (Invitrogen) to generate the plasmid zeoCMV-HA-RhoB-GG. Similar expression vectors for HA-RhoB-S, an epitope-tagged polypeptide that is unprenylated due to a C193S mutation in the RhoB CAAX box, HA-RhoB^V14-S, an activated version that includes a V14 mutation, and HA-RhoA have been described (34).

Cell culture. The cell lines used in this study were derived from Rat1/ras, a clonal Rat1 fibroblast line generated by transformation with v-H-ras (30, 49). Rat1/ras cells were cultured in Dulbecco modified Eagle medium (DMEM; Gibco) containing 10% fetal bovine serum (Atlanta Biological) and 10 U of penicillin-streptomycin (Mediatech) per ml. Where indicated, the FT-specific inhibitor L-744,832 (31) was added to cultures to a final concentration of 10 µM. Rat1/Ras cell lines stably expressing HA-RhoB-GG were obtained by modified calcium phosphate transfection of zeoCMV-HA-RhoB-GG, selection in 200 µg of zeosin (Invitrogen) per ml, ring cloning of zeosin-resistant colonies, and expansion into mass culture. Expression was verified by Western analysis with the anti-HA antibody 12CA5 (BABCO, Inc.). Control cell lines harboring empty vector or expressing the unprenylated HA-RhoB-S or HA-RhoB^V14-S mutants (34) were derived similarly with zeocin resistance vectors (Invitrogen).

MTT assay. Cell growth was measured by MTT [3-(4,5-diethylthiazoly-2-yl)-2,5-diphenyltetrazolium bromide] assay (6). Briefly, cells were seeded at 500 cells per well in 96-well culture plates in quadruplicate. At various points, medium was removed, and cells were incubated with 180 µl of RPMI 1640 containing 5% fetal calf serum (FCS), 0.25 mg of MTT (Sigma) per ml at 37°C for 4 h, followed by solubilization with 20% sodium dodecyl sulfate-50% dimethyl formamide in water for another 4 h or overnight at 37°C. The absorbance of each well was measured with a microplate reader (Rainbow reader) at 595- and 655-nm dual wavelengths. The viable cell number is proportional to the absorbance.

Soft agar assay. One milliliter of 0.5% NuSieve agarose (FMC Biochemicals) in DMEM containing 10% FCS was used to coat the bottom of each well in 6-well culture dishes. After hardening, 10⁴ cells were suspended in 1 ml of 0.3% NuSieve agarose, DMEM, and 10% FCS solution and plated onto the bottom layer. The cell solution was allowed to set 30 min at room temperature before moving it to 37°C. Where indicated, L-744,832 was added to top and bottom agar mixtures to a final concentration of 10 µM. Colonies formed in the soft agarose culture were photographed 10 to 16 days later with an Olympus microscope with a 35-mm camera attachment.

Actin immunofluorescence. Cells were seeded onto coverslips in six-well dishes and treated the next day for 48 h with 10 µM L-744,832 or carrier. Cells were fixed and stained with fluorescein-phalloidin (Molecular Probes) as described previously (49). Photographs of stained cells were generated on a Leitz immunofluorescence microscope.

Apoptosis analysis. For apoptosis analysis, 1 × 10⁶ to 2 × 10⁶ cells were seeded onto polyHEMA-coated dishes. After 12 to 14 h of plating, 10 µM FTI was added as described earlier (39). After 48 h of FTI treatment, cells were collected, trypsinized, washed with phosphate-buffered saline (PBS), and fixed in 70% ethanol. The cells were then stained in PBS containing 5 µg of propidium iodide per ml, 10 µg of RNase A per ml, and 0.1% glucose. Flow cytometry was performed by using an EPIC/XL cell analyzer (Coulter).

Western analysis. At appropriate times, cells were washed in cold PBS and harvested in Nonidet P-40 lysis buffer containing phenylmethylsulfonyl fluoride, pepstatin, and leupeptin (20a). Cellular protein was quantitated by Bradford assay, and 40 µg was fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and electrophoretically transferred onto nitrocellulose membranes (Amersham). Blots was probed with 2.5 µg of anti-HA antibody 12CA5 (BABCO, Inc.) per ml followed by an anti-mouse immunoglobulin G horseradish peroxidase-conjugated secondary antibody (Boehringer Mannheim) at a 1:10,000 dilution by standard protocols. A chemiluminescence kit (Pierce) was used to detect the antibody complex according to the protocol recommended by the vendor. Western analysis of p21^WAF1 was performed similarly under the same conditions with the C19 anti-p21 antibody as suggested by the vendor (Santa Cruz Biotechnology).

Northern analysis. Total cytoplasmic RNA was isolated and subjected to Northern blotting and hybridization as described previously (48). Probes were generated by random-primed labeling of RhoB and RhoA cDNAs (34) with [-³²P]dCTP (NEN).

Transactivation assay. NIH 3T3 cells were maintained in DMEM containing 10% calf serum, 1 mM of sodium pyruvate, and 10 U of penicillin-streptomycin (Mediatech) per ml. Cells were transfected for 20 h in six-well dishes by modified calcium phosphate transfection (8) with 3 µg of total DNA including 0.2 µg of CMV-RasV12 (42), 0.4 µg of the p21 promoter reporter WWP-luc (13), 0.2 µg of CMV-gal (to normalize for transfection efficiency), and 2.2 µg of pBS+ as plasmid filler. Where indicated, 0.2 µg of CMV-HA-RhoB-GG or HA-RhoA and 2 µg of pBS+ was included. Cells were washed and refed with standard growth medium and 10 µM FTI L-744,832 was added where indicated (31). After an additional 24-h incubation, cells were harvested and extracts were prepared and assayed for -galactosidase and luciferase activities with a commercial kit under conditions recommended by the vendor (Promega). Relative luciferase activity was normalized to -galactosidase activity in each trial before the data were plotted.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

RhoB-GG induces morphological reversion and actin stress fiber formation in transformed cells. Previous work suggested a role for alteration of RhoB in the mechanism by which FTIs induce phenotypic reversion, loss of anchorage-independent growth capacity, and apoptosis (34, 37, 39, 49, 50). Since FTI treatment leads to an elevation of RhoB-GG in cells and a concomitant loss of growth-promoting activity (34), we investigated the possibility that the gain of RhoB-GG may be sufficient to mediate cell growth inhibition or other drug responses. This was performed by expressing in cells a RhoB-RhoA chimera in which the 16 C-terminal residues of RhoB, which direct both farnesylation and geranylgeranylation, were replaced with the 13 C-terminal residues of RhoA, which directs only geranylgeranylation (1, 34). The localization of this construction, termed RhoB-GG, is more similar to RhoB in FTI-treated cells or to the strictly geranylgeranylated RhoA protein (2, 37). An epitope-tagged version of this chimera, HA-RhoB-GG, was stably expressed in normal Rat1 fibroblasts or in Rat1/ras, an H-Ras-transformed Rat1 derivative (30, 49). Cell lines expressing either only vector sequences or HA-RhoB-S or HA-RhoB^V14-S, two unprenylated CAAX box mutants, were also generated as controls. Four sets of cell lines each for Rat1 and Rat1/ras were each derived by transfection of these vectors and are referred to as Rat1/B-GG, Rat1/B-S, Rat1/BV-S, and Rat1/vect and as Ras/B-GG, Ras/B-S, Ras/BV-S, and Ras/vect, respectively. Subsequent analyses demonstrated consistent phenotypes among the clones in each set, arguing against significant clonal variation. In addition, cell clones expressing the unprenylated RhoB-S or RhoB-VS proteins were identical in all tests. Therefore, a subset of cell lines generated in this study which are representative is presented. Exogenous expression of various RhoB mutants was confirmed in Rat1 or Rat1/ras derivatives by Western analysis with an anti-HA epitope antibody (Fig. 1A). Northern analysis indicated that the expression of the exogenous RhoB gene in the cell lines was a fewfold higher than the endogenous gene but not in gross excess (Fig. 1B).

View larger version (60K): [in this window] [in a new window]   FIG. 1.   HA-RhoB-GG expression. (A) Western blot analysis. Extracts from different stable Rat1 and Rat1/ras transfectants were immunoblotted with the anti-HA antibody 12CA5. A total of 40 µg of cellular protein was loaded into each lane of the gels. Expression and morphology of the lines examined in each set were highly similar. (B) Northern blot analysis. Total cytoplasmic RNA for randomly selected cell lines from each set generated was fractionated by Northern analysis and hybridized to RhoB and RhoA cDNA probes.

View larger version (112K): [in this window] [in a new window]   FIG. 2.   RhoB-GG reverts the Ras-transformed phenotype. Photomicrographs of various Rat1/ras and Rat1 derivatives in monolayer culture in the presence or absence of FTI are shown (see text for nomenclature). Ras/BV-S cells exhibited the same morphology and reversion response as Ras/B-S cells (not shown). Where indicated, cells were treated for 40 h with FTI before photography.

View larger version (96K): [in this window] [in a new window]   FIG. 3.   RhoB-GG activates actin stress fibers in Ras-transformed cells. Cells were treated with FTI or carrier, and actin was visualized by phalloidin staining and fluorescence microscopy as described in Materials and Methods.

RhoB-GG selectively inhibits the growth of transformed cells. The first indication that RhoB-GG could inhibit Ras transformation was that fewer colonies emerged after transfection of Rat1/ras and selection for stably transfected cells. The efficiency of colony formation by HA-RhoB-GG vector was decreased ~2-fold in Rat1/ras cells relative to vector, but no similar decrease was observed in Rat1 cells (data not shown). To determine explicitly whether RhoB-GG inhibited Rat1/ras growth, we compared the anchorage-dependent growth and the anchorage-independent growth of the cell lines obtained. We also compared the growth of cells in the presence or absence of FTI to determine whether RhoB-GG might mimic the growth-inhibitory effects of FTI. MTT assay was used to monitor proliferation in monolayer culture of replicate cultures of Ras/B-GG and Ras/vect lines (see Materials and Methods). We observed that Ras/B-GG cells grew ~50% more slowly than Ras/vect cells, with an average doubling time of ~24 h compared to ~16 h for Ras/vect cells (Fig. 4A). A similar experiment comparing the flat Rat1/vect and Rat1/B-GG cell lines indicated that RhoB-GG had only limited effects on normal fibroblast growth (Fig. 4B). Thus, RhoB-GG inhibited the growth of transformed cells relatively specifically. This resembled the effects of FTI treatment, which reduces the growth rate of Rat1/ras cells in monolayer culture substantially but has only limited effects on Rat1 cells cultured under the same conditions (49). We tested whether RhoB-GG and FTI treatment produced a similar reduction in monolayer growth rate. Consistent with previous observations (49), FTIs reduced the growth of Ras/vect cells in a way similar to that displayed by untreated Ras/B-GG cells (Fig. 5A). FTI treatment slightly potentiated the inhibition of cell growth by RhoB-GG but this effect was minimal. FTIs also slightly potentiated the effects of RhoB-GG on the growth of Rat1 cells (Fig. 5B). The slight effects of FTIs observed in Ras/B-GG or Rat1/B-GG cells could be explained by effects on the endogenous RhoB genes in those cells. In addition, the effects of RhoB-GG and FTI on normal Rat1 cells were similar in degree to those observed previously (49). Taken together, the results indicated that gain of RhoB-GG by FTI was sufficient to mediate selective inhibition of anchorage-dependent growth of Ras-transformed cells.

View larger version (12K): [in this window] [in a new window]   FIG. 4.   RhoB-GG inhibits the proliferation of Ras-transformed cells. (A) Growth curve of Ras/vect and Ras/B-GG cells. Cell growth was measured with the MTT assay. Three representative cell lines containing vector (solid symbols) or RhoB-GG (open symbols) were tested. Each point represents the average of four individual measurements. (B) Growth curve of Rat1/vect and Rat1/B-GG cells. Two representative cell lines containing vector (open symbols) or RhoB-GG (solid symbols) were tested as described above.

View larger version (14K): [in this window] [in a new window]   FIG. 5.   RhoB-GG mimics the selective inhibitory effects of FTI treatment. (A) FTI treatment versus RhoB-GG expression in Ras-transformed cells. Growth curves were determined as before. FTI was added where indicated and replenished every 48 h. A representative comparison between Ras/vect and Ras/B-GG cells is shown; similar results were obtained with other cell lines. (B) FTI treatment versus RhoB-GG expression in normal fibroblasts. Growth curves were determined as described above, and FTI was added as before where indicated. A representative comparison between Rat1/vect and Rat1/B-GG cells is shown; similar results were obtained with another cell line.

View larger version (107K): [in this window] [in a new window]   FIG. 6.   RhoB-GG inhibits anchorage-independent growth. 104 Cells from two Ras/vect and Ras/B-GG lines each were seeded at 104 cells per well in six-well dishes in soft agar culture. A control trial confirming FTI suppression of Rat1/ras parental cells was also performed as described elsewhere (30). Colonies were scored after 2 to 3 weeks. A photograph of representative colony formation in wells for each cell line tested is shown.

RhoB-GG does not promote apoptosis of Ras-transformed cells denied substratum adhesion. FTIs are normally cytostatic to Ras-transformed cells but are cytotoxic if cells are cultured on polyHEMA, a nonadherent substrate that sensitizes them to anoikis (39). To test whether RhoB-GG could mimic the ability of FTIs to promote anoikis, we compared the sensitivity of Ras/vect and Ras/B-GG cells to undergo apoptosis when cultured on polyHEMA. Apoptosis was measured by the detection of sub-G₁ phase DNA by flow cytometry, an assay which has been validated previously in this cell system (39). Interestingly, both Ras/vect and Ras/B-GG cells survived in polyHEMA culture, indicating that oncogenic Ras could promote survival in both cases and that FTI addition to either type of cell caused apoptosis within 24 h in a way similar to that observed with parental Rat1/ras cells (Fig. 7). Since it has been demonstrated previously that farnesyl-independent RhoB (myristylated RhoB) can inhibit FTI-induced death in Rat1/ras cells (39), this result suggested that FTI-induced apoptosis occurred through a distinct mechanism that was separable from the growth inhibitory and morphological effects mediated by RhoB-GG and where loss of RhoB-F was implicated. We concluded that gain of RhoB-GG by FTI was insufficient to sensitize Ras-transformed cells to anoikis.

View larger version (24K): [in this window] [in a new window]   FIG. 7.   RhoB-GG does not induce apoptosis on polyHEMA. Cells were seeded into 10-cm dishes coated with polyHEMA at 106 cells per well. Where indicated FTI was added, and 48 h later the cell suspensions were collected. Cells were trypsinized briefly to disperse clumps, pelleted, fixed in ice-cold ethanol, and processed for flow cytometry as described previously (54).

RhoB-GG mimics the ability of FTIs to induce expression of the CKI p21^WAF1. The cell cycle kinase inhibitors (CKIs) are key focal points for negative regulation of cell cycle progression. Since gain of RhoB-GG was sufficient to inhibit the proliferation of Ras-transformed cells, one might predict that one or more CKIs might be elevated in Ras/B-GG cells. Some human tumor cell lines which are susceptible to growth inhibition by FTIs exhibit induction of the CKI p21^WAF1 (58). Therefore, we investigated whether FTIs could increase p21^WAF1 expression in Rat1/ras cells and whether RhoB-GG might act similarly. Western analysis of cell extracts was performed to monitor steady-state levels of p21^WAF1. We found that p21^WAF1 was undetectable in all Ras/vect cell lines examined but was significantly elevated in all Ras/B-GG cell lines examined (Fig. 8A). The levels of induction were robust, being similar in degree to the levels stimulated by the activation of wild-type p53 in a control E1A-immortalized BRK cell line (12). A similar but transient elevation of p21^WAF1 occurred in Rat1/ras or Ras/vect cells treated with FTI. After drug addition, p21^WAF1 was first detected at 12 to 24 h (Fig. 8B), a period that is consistent with the kinetics of RhoB alteration, growth inhibition, and morphological reversion in this system (49). The relative increase elicited by FTI treatment was similar to that detected in Ras/B-GG cells, arguing that induction by the latter was not due to an overexpression artifact. Since p53 can activate expression of p21^WAF1 and since it was conceivable that RhoB-GG gain stressed cells such that p53 was stabilized, we examined the same extracts by immunoblotting with a p53 antibody. p53 was not detected, however, arguing against the possibility that p21^WAF1 was elevated by p53 activation. Notably, p21^WAF1 was not elevated in Rat1/B-GG cells (Fig. 8C). This argued that p21^WAF1 elevation was a synthetic effect produced by the combination of Ras activation and RhoB-GG elevation and that the induction of p21^WAF1 was associated with growth inhibition. Interestingly, Ras/B-S and Ras/BV-S cells expressing unprenylated wild-type or activated RhoB also exhibited elevation of p21^WAF1 (Fig. 8C). This implied that in the presence of oncogenic Ras RhoB does not have to be prenylated to activate p21^WAF1, a result reminiscent of previous demonstrations that prenylation is dispensible for RhoB to activate SRF (35). It also implied that induction of p21^WAF1 by FTIs may be a necessary but perhaps not a sufficient cause of phenotypic reversion, a conclusion consistent with the results of a recent study of human tumor cells in which p21^WAF1 induction is not consistently correlated with FTI-induced growth inhibition (58). To determine whether FTIs and RhoB-GG upregulated p21^WAF1 expression at the level of transcription, we performed a set of transient promoter activation experiments in NIH 3T3 cells with a luciferase reporter plasmid. Consistent with a transcriptional mechanism of activation, FTI and RhoB-GG each activated the p21^WAF1 reporter when introduced into cells with activated Ras (Fig. 8D). The activation by FTI was slightly more robust in these experiments, but RhoB-GG clearly mimicked the action of the drug. In support of the possibility that RhoB-GG had properties similar to the geranylgeranylated RhoA protein, p21^WAF1 transcription was activated by RhoA when it was cotransfected with activated Ras (data not shown). We concluded that RhoB-GG was sufficient to mediate transcriptional activation of p21^WAF1 by FTI in the presence of activated Ras.

View larger version (22K): [in this window] [in a new window]   FIG. 8.   RhoB-GG is sufficient to mediate activation of p21WAF1 by FTIs. (A) Constitutive elevation of p21WAF1 in Ras/B-GG cells. Cell extracts were prepared from the cell lines indicated and analyzed by immunoblotting with a p21WAF1 antibody. BRK An1 is a control cell line that contains a temperature-sensitive p53 gene that is mutant at 38°C. When cells are shifted to 32°C, the mutant assumes a wild-type configuration and function that leads to p21 induction. A single Ras/vect line lacking expression is representative of all lines examined. (B) Transient induction of p21WAF1 in Ras-transformed cells after FTI treatment. Cell extracts were prepared at the times indicated after addition of FTI to Rat1/ras cells and analyzed as described above. The cell extract from a Ras/B-GG cell line was included to illustrate the similar levels of expression in FTI-treated Rat1/ras cells and Ras/B-GG cells. (C) p21WAF1 is not induced in Rat1 cells expressing RhoB-GG but is induced in Rat1/ras cells expressing unprenylated RhoB-S or RhoBV-S. Cell extracts were prepared and analyzed as described above. RhoBV-S is the same as RhoB-S except that the former also includes an activating mutation (V14). An extract prepared from a Ras/B-GG cell line provides a positive control (right lane). (D) FTIs activate p21WAF1 at the level of transcription. NIH 3T3 cells were transiently transfected with an oncogenic Ras vector (42) and the p21WAF1 reporter plasmid WWP-luc (13), along with vectors expressing no insert or HA-RhoB-GG. Where indicated, cells were treated for 24 h with FTI before being processed for relative luciferase activity as described in Materials and Methods. The standard error was computed from three trials.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

In this study, we demonstrated that gain-of-function effects on RhoB elicited by FTIs are sufficient to mediate phenotypic reversion and cell cycle inhibition, two major effects of FTIs on transformed cells. Unlike FTIs, RhoB-GG did not sensitize Ras-transformed cells to anoikis, suggesting that this effect is mechanistically distinct and may instead involve loss of function of farnesylated RhoB or other proteins. These observations are important because they provide the first evidence that FTIs may act not only by causing loss of function but also by inducing gain of function of proteins that are normally farnesylated in cells. A caveat to this study is that the engineered RhoB-GG molecule is not exactly identical to RhoB-GG found in FTI-treated cells. We believe this caveat is small, however, because of the brevity of the C-terminal replacement (only ~10 amino acids are different) and because of other studies which have also implicated functional alteration of RhoB in the drug mechanism (34, 37, 39, 50). The extent of the phenotypes induced by RhoB-GG suggests that the antitransforming effects of FTIs that are mediated through gain of function may be rather broad. Our findings also provide a mechanism to explain how FTIs can inhibit the growth of neoplastically transformed cells that lack Ras mutations or that are dependent on K-Ras or N-Ras, which each remain active in the presence of FTIs due to their ability to be alternately geranylgeranylated in FTI-treated cells (26, 41, 43, 53, 67). K-Ras is the predominant Ras oncogene in human cancer but unlike RhoB and H-Ras (11) geranylgeranylation of either is normal or oncogenic K-Ras does not change its function (8a).

We previously proposed an alternate model for FTI action, termed the FTI-Rho hypothesis, to explain the biological effects of the drugs which cannot be attributed directly to functional inhibition of Ras (34). The FTI-Rho hypothesis, which emerged from studies of a role for RhoB alteration in the drug mechanism (34, 37, 39), proposes that inhibition or alteration of farnesylated Rho functions are crucial steps for the antitransforming properties of FTIs. This model was developed to address anomalies evident during the earliest phases of FTI research which argued that Ras inhibition could not be the sole basis for FTI action (49). Additional anomalies and difficulties with the original Ras-based model for FTI action emerged subsequently with evidence that the growth of neoplastically transformed cells can be blocked even if FTIs are ineffectual at blocking K-Ras farnesylation and function (37, 43, 57, 61). Animal experiments which further illustrate Ras-independent effects include xenograft experiments employing K-Ras- and N-Ras-transformed cells (32) or human tumor cells (46, 61), N-Ras oncomouse experiments (43), and oncomouse experiments where cell cycle inhibition represents the major mechanism for drug action (4). When taken together, the existing results indicate that Ras inhibition may be sufficient but is not necessary for FTIs to reverse or inhibit malignant cell growth.

The FTI-Rho hypothesis offers a new viewpoint to interpret cellular responses where loss of function in oncogenic Ras cannot explain the response. The loss-of-function and gain-of-function arms of the mechanism are distinguished in this model (Fig. 9). Gain of function through RhoB-GG elevation is proposed to mediate phenotypic effects and cell cycle inhibition. RhoB-GG did not potentiate anoikis (adhesion deprival-induced apoptosis), so loss of function through loss of RhoB-F, H-Ras, or other farnesylated proteins would be implicated in this response, a conclusion in support of the results of previous observations (39).

View larger version (8K): [in this window] [in a new window]   FIG. 9.   Gain of function and loss of function of RhoB mediate different antitransforming effects of FTI treatment. Cells treated with FTI undergo loss of RhoB-F and gain of RhoB-GG. Loss of RhoB-F is associated with anoikis in Ras-transformed cells, whereas gain of RhoB-GG is sufficient to mediate p21 activation and cell cycle inhibition. p21 activation may be required for RhoB-GG to direct phenotypic reversion and loss of anchorage independence. Other effector mechanisms are also required, since p21 elevation by itself may be necessary but not sufficient for the FTI response (see Results).

Elevation of RhoB-GG in Ras-transformed cells caused dramatic morphological reversion to a flat phenotype characteristic of normal fibroblasts. The reverted phenotype closely resembled that produced by FTI treatment of Ras-transformed cells. Reversion was associated with a shift in cytoskeletal actin structure, such that there was a loss of membrane ruffles associated with transformation to an increase in actin stress fibers characteristic of normal cells. RhoB-GG did not affect the morphology of normal cells but slightly increased stress fiber formation in those cells, also mimicking FTI treatment (49). While the exact mechanism(s) underlying this phenomenon is not yet clear, the ability of RhoB-GG to induce stress fiber formation addresses an apparent anomaly in the FTI-Rho hypothesis. Rho proteins induce stress fiber formation, so if FTIs act in part by altering RhoB, as the model proposed, one might have been expected FTIs to disperse rather than induce stress fibers in Ras-transformed cells. FTI treatment produces the opposite effect, however, such that stress fiber formation increases (44, 49). The observation that RhoB-GG can induce stress fibers addresses this anomaly. RhoB-GG has certain RhoA-like features (being not only structurally related but rendered more similar to RhoA by geranylgeranylation), so one consequence of FTI treatment is to shift RhoB to a RhoA-like localization and therefore elevate a RhoA-like function in cells. Based on accepted definitions of RhoA function, this elevation would be predicted to increase stress fiber formation (51, 52). This interpretation is simple and consistent with the effects of FTIs on RhoB function, cellular morphology, and cytoskeletal actin (34, 37, 49, 51).

Cell growth inhibition by RhoB-GG or by FTIs correlated with increased expression of p21^WAF1. Transformed cells but not normal cells were susceptible to RhoB-GG, similar to the effects of FTI treatment (49). Increased expression of p21^WAF1 in Ras/B-GG cells may be part of the mechanism through which RhoB-GG may inhibit cell cycle progression because of the role of p21^WAF1 in inhibiting cell cycle progression in many cells (21). Notably, p21^WAF1 was elevated in Ras/B-S cells which did not exhibit evidence of phenotypic reversion or growth inhibition, implying that although p21^WAF1 may be necessary it was not sufficient to mediate these responses to FTI. This finding is consistent with observations in human tumor cells, where p21^WAF1 elevation by FTIs has been seen but not correlated with FTI growth inhibition (58), as well as with previous observations that RhoB-S can stimulate gene expression by activating SRF (35). The mechanism of p21^WAF1 activation was p53 independent insofar as RhoB-GG did not stabilize p53 and elevate its level; moreover, RhoB-GG was still able to activate a p21^WAF1 promoter in which the p53 binding site was mutated. Identification of RhoB-GG as a potential mediator of the inhibitory effects of FTIs on cell cycle progression does not immediately shed light on why normal and transformed cells should respond so differently. It is tempting to speculate that morphological and/or cytoskeletal effects of RhoB-GG related to cell adhesion are important, because Rho proteins have been implicated in integrin-dependent cell growth control (3, 5, 22, 23, 65) and because reversion indirectly affects cell cycle regulation by returning integrin-mediated adhesion dependence. Reengagement of an appropriate adhesion signaling program could underlie the response of transformed cells, since normal cells already have such a program in placeas an adhesion checkpoint controland do not reengage substratum after FTI treatment and RhoB-GG elevation.

Several interpretations of our results can be considered in light of the inhibitory properties of oncogenic Ras in primary cells (59). Rho is necessary for Ras transformation (29, 50, 51), so one interpretation of our results is that RhoB-GG interferes with a Rho effector function that is required for Ras transformation. We believe that differences in the accessibility of RhoB-GG to transformation-associated Rho effectors could be important. Loss-of-function and gain-of-function scenarios can be imagined for how RhoB-GG might interfere with such functions. First, RhoB-GG may be localized away from certain RhoB effectors whose action is required for transformation. This concept can be illustrated by considering the RhoB effector PKN/PRK1, which is localized on endosomes (45) like RhoB and which is a likely effector for RhoB; since RhoB-GG is not on endosomes it may poorly interact with PKN/PRK1 in cells (2, 37). In this illustration RhoB-GG would represent a loss of function with regard to PKN/PRK1 signaling (whose role in transformation remains to be determined however). Second, since RhoB-GG localization overlaps with RhoA, RhoB-GG may acquire the ability to bind to certain RhoA effectors that are not normally accessible to RhoB. In this case, it is conceivable that RhoB-GG may competitively interfere with the operation of certain effectors of RhoA if RhoB-GG does not mimic RhoA function exactly. In this illustration RhoB-GG would represent a gain of function that dominantly inhibits certain functions of RhoA that may be associated with Ras transformation. This latter case offers an interpretation of our findings which are consistent with the findings of a recent report that suggests that Rho facilitates Ras transformation and that loss of Rho unleashes the growth-inhibitory properties of oncogenic Ras (47). If RhoB-GG disrupts or interferes with certain Rho effector functions, these functions themselves might even be nonphysiological in the sense that the effectors are themselves either mislocalized or dysfunctional in transformed cells (a possibility since cell attachment and actin cytoskeletal regulation are subverted in cancer cells).

It is possible that RhoB-GG has a gain of function directly related to growth inhibition. RhoB has been linked to some types of growth-inhibitory stimuli (14, 15, 16), so RhoB-GG may act through some physiological RhoB effectors. We favor the interpretation that RhoB-GG actions are based on altered cell localization, which is the simplest interpretation, but we cannot rule out the possibility that certain effectors distinguish RhoB-GG and RhoB-F such that differential biochemical specificities rather than localization changes underlie the effects of RhoB-GG (some RhoB-binding proteins have been identified which distinguish prenylation status [36, 69]). While implicated in inhibitory responses to UV irradiation and TGF- (14, 15, 16), RhoB has not been linked previously to p21^WAF1 regulation. In addition to connections between activated Ras and p21^WAF1 (this study and reference 47), activated Raf has been reported to cause cell cycle arrest in normal 3T3 fibroblasts through a p21^WAF1-dependent mechanism (68). Our findings would support a role for RhoB in mediating certain antiproliferative effects of oncogenic Ras but tend to suggest that p21^WAF1 may be necessary but perhaps not sufficient to mediate FTI effects.

The cell cycle inhibitory effects of RhoB-GG may be important to the FTI mechanism in human tumors because cytostatic rather than cytotoxic effects appear to predominate in xenograft models (20, 24, 46, 61, 62). At the pathophysiological level, clinical cancer can be defined by the ability of a cell to survive, proliferate, and invade in the absence of normal adhesion signals. Such signals normally act to stringently govern cell division so that anoikis occurs in their absence. Invasive and metastatic cancers are already substantially resistant to anoikis. If the apoptotic properties of FTIs are based on anoikis mechanisms (39), then one might predict that only premalignant or early-stage malignancies will exhibit apoptosis and show regression, whereas late-stage cancers of the type which are most often seen in the clinical setting will not. In support of this likelihood, while transformed cells are sensitive to anoikis when treated with FTIs or RhoB-GG, we have not seen a similar anoikis sensitivity in any human tumor cell lines tested (unpublished observations). Since these tumor cell lines retain cell cycle inhibitory responses to FTIs, human tumor xenograft models may actually be better models for predicting clinical responses than oncomouse models, which display an apoptotic response to FTIs.