Geranylgeraniol Potentiates Lovastatin Inhibition of Oncogenic H-Ras Processing and Signaling While Preventing Cytotoxicity

Geranylgeraniol potentiates lovastatin inhibition of oncogenic H-Ras processing and signaling while preventing cytotoxicity

Terence F McGuire and SaõÈ d M Sebti*

Department of Pharmacology, School of Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania 15261, USA

Oncogenic H-Ras requires farnesylation for its trans- catalyzes the condensation of FPP and isopentenyl forming activity. Lovastatin inhibits both protein pyrophosphate to form GGPP. GGPP and FPP are farnesylation and geranylgeranylation by decreasing utilized by geranylgeranyltransferases (GGTases I and cellular pools of farnesylpyrophosphate (FPP) and II), and farnesyltransferase (FTase), respectively, for geranylgeranylpyrophosphate (GGPP), respectively. Use posttranslational isoprenylation of proteins on carbox- of lovastatin as a chemotherapeutic agent has been yl terminal cysteine residues (reviewed in GruÈ nler et al., precluded by its signi®cant cytotoxic eects. In this 1994; Maltese, 1990; Casey, 1992). FTase and GGTase report, we describe a novel approach utilizing a I prenylate proteins with carboxyl termini that end combination of lovastatin and geranylgeraniol (GGOH) with a CAAX box where C=cysteine, A=aliphatic, to potentiate the ability of lovastatin to block oncogenic and X=any amino acid. FTase prefers X as a serine or H-Ras signaling and concomitantly rescue lovastatin methionine whereas GGTase I prefers X as a leucine or toxicity. GGOH co-treatment with lovastatin enhances isoleucine. GGTase II prenylates proteins that end in inhibition of oncogenic H-Ras processing and constitutive XXCC and XCX where X is any amino acid. For activation of mitogen-activated protein kinase (MAPK), several proteins, isoprenylation is essential for proper and preserves the processing of geranylgeranyltransfer- intracellular localization and biological function (Holtz ase (GGTase) I and GGTase II protein substrates. et al., 1989; Fukada et al., 1990; Der and Cox, 1991; Moreover, co-treatment with GGOH signi®cantly (15- Hori et al., 1991; Inglese et al., 1992). In contrast to fold) attenuates the cytotoxic eects of lovastatin as well FPP, GGPP is currently known to be utilized only for as prevents lovastatin-induced cell rounding. These protein geranylgeranylation and possibly ubiquinone results demonstrate that GGOH potentiates the anti- biosynthesis (ViganoÁ et al., 1995). oncogenic/anti-signaling activity of lovastatin while Geranylgeranylated proteins and farnesylated pro- antagonizing its cytotoxicity. These opposing eects teins appear to comprise distinct, but overlapping sets are due to a GGOH metabolite that serves simulta- of proteins, with the former being greater in number neously as a potent inhibitor for farneslyltransferase as than the latter (Epstein et al., 1990; Farnsworth et al., well as a substrate for GGTases I and II. 1990). Many of these proteins have been shown to play essential roles in signal transduction pathways Keywords: Ras inhibition; lovastatin; geranylgeraniol and some have been implicated in malignant transformation. For example, the geranylgeranylated guanine nucleotide-binding proteins Rho and Rac have recently been shown to be critical players in Introduction regulating not only the organization of the actin cytoskeleton (Nobes and Hall, 1995) but also the

The mevalonic acid (MVA) pathway is responsible for progression of the cell cycle through G1 (Olson et al., the biosynthesis of cholesterol and isoprenoid inter- 1995). In addition, Ras, another family of guanine mediates such as geranylgeranylpyrophosphate nucleotide-binding proteins, control normal cell (GGPP) and farnesylpyrophosphate (FPP). Two sites growth (Mulcahy et al., 1985) and dierentiation in the MVA pathway have been cited to be of (Bar-Sagi and Feramisco, 1985) and, when mutated, particular importance: the synthesis of MVA by can produce malignant transformation (Reddy et al., hydroxymethylglutaryl-coenzyme A (HMG-CoA) re- 1982). The Ras family of proteins serve as transducers ductase, an early step thought to be the major point of of extracellular signals from receptor tyrosine kinases regulation, and the so-called `branch-point' of FPP to the nucleus (McCormick, 1993). Their stimulation metabolism (Brown and Goldstein, 1980; Sabine, 1983; by these receptors results in the activation of several reviewed in GruÈ nler et al., 1994). FPP is the last growth-related pathways including a cascade of common intermediate in the pathway and is the mitogen-activated protein kinases (MAPKs) such as substrate for a number of dierent enzymes that Raf, MEK and ERK (McCormick, 1993), the latter of catalyze committed steps in branching pathways which can translocate to the nucleus and regulate the leading to the biosynthesis of cholesterol, ubiquinone, activity of some transcription factors. In some human dolichol, as well as isoprenylated proteins and hemes. cancers, Ras is GTP-locked and constitutively GGPP synthase, one of the branch-point enzymes, activates the MAPK cascade. The ability of Ras to cause cancer requires its attachment to the plasma Correspondence: SM Sebti membrane which is mediated by prenylation (Der and *Present address: H. Lee Mott Cancer Center & Dept. Biochem. Cox, 1991; Kato et al., 1992). The prenylation of Ras Mol. Biol. Univ. South Florida 12902 Magnolia Dr, Tampa, FL in vitro exhibits a preference for farnesylation (James 33612, USA This work was supported by NIH grant CA-55823. et al., 1995). Furthermore, although the prenylation of Received 13 May 1996; revised 12 September 1996; accepted KB-Ras in vivo is, at present, uncertain (Casey et al., 12 September 1996 1989; Lerner et al., 1995), H-Ras has been shown to GGOH/lovastatin: selective inhibition of H-Ras signaling TF McGuire and SM Sebti 306 require farnesylation for its cancer-causing activity a (Hancock et al., 1989; Seabra et al., 1991; Lerner et 12345678 al., 1995). Moreover, in nude mice, inhibitors of U Ras P FTase are eective at suppressing the growth of tumor MAPK–P cells possessing oncogenic H- or K-Ras (Sun et al., MAPK 1995). Lovastatin, a potent competitive inhibitor of HMG- CoA reductase (Alberts, 1988), signi®cantly reduces not b 1234 5678 91011121314 1516 only the biosynthesis of the end-product cholesterol, for which it is used clinically (Illingworth and Bacon, Ras 1989), but also depletes the intracellular pools of Rap1A GGPP and FPP, resulting in the inhibition of protein geranylgeranylation and protein farnesylation (Leonard Figure 1 GGOH potentiates lovastatin-induced inhibition of H- et al., 1990). We have recently shown that lovastatin Ras-CVLS processing and signaling while rescuing that of H-Ras- disrupts early signaling events such as tyrosine CVLL. (a), NIH3T3 cells transformed with either H-Ras- CVLS(61L) (lanes 1 ± 4) or H-Ras-CVLL(61L) (lanes 5 ± 8) were phosphorylation levels of the PDGF receptor and its treated two consecutive days with vehicle (lanes 1 and 5), association with PI-3-kinase (McGuire et al., 1993). lovastatin (20 mM on day 1 and 10 mM on day 2) (lanes 2, 4, 6 Moreover, lovastatin arrests cultured cells predomi- and 8), and 20 mM GGOH (lanes 3, 4, 7 and 8). Cells were lysed and equivalent amounts of protein electrophoresed and immuno- nantly in the G1 phase of the cell cycle and produces a characteristic rounded morphology (Quesney-Huneeus blotted with either anti-Ras antibody (Y13-238) or anti-MAPK as described under Materials and methods. The slower-migrating, et al., 1979; Sinensky and Logel, 1985; Fenton et al., unprocessed form (U) and the processed form (P) of Ras as well 1992). Lovastatin has also been found to inhibit tumor as the inactive and activated forms of MAPK (MAPK and growth of cells expressing oncogenic H-Ras in nude MAPK-P, respectively) are indicated. (b) NIH3T3 cells trans- mice, but at doses that blocked tumor growth, formed with H-Ras-CVLS(61L) were treated two consecutive lovastatin was extremely toxic and some animals died days with vehicle (lanes 1 ± 4) or with lovastatin at 1 mM (lanes 5 ± 8), 5 mM (lanes 9 ± 12), or 15 mM (lanes 13 ± 16) in conjuction with (Sebti et al., 1991). It is not known whether inhibition GGOH at either 0 mM (lanes 1, 5, 9 and 13), 7.5 mM (lanes 2, 6, 10 of protein farnesylation and/or protein geranylgerany- and 14), 15 mM (lanes 3, 7, 11 and 15), or 30 mM (lanes 4, 8, 12 lation, or the reduction in levels of some other end- and 16). Cells were lysed and equivalent amounts of protein product of the MVA pathway, are responsible for electrophoresed and immunoblotted with anti-Ras antibody (Y13- 238). The identical membane was then reprobed with anti-Rap1A lovastatin's toxicity. antibody. Data are representative of three and four independent In this report, we describe the use of lovastatin experiments for a and b, respectively treatment of cells in conjunction with geranylgeraniol (GGOH) as a novel approach for selectively inhibiting oncogenic H-Ras processing and signaling while rescuing cells from lovastatin toxicity. Our mechanistic processed form of Ras. Oncogenic H-Ras constitutively investigations lead to a model where GGOH is activates MAPK and, therefore, both the active converted to a metabolite which concomitantly (hyperphos-phorylated) and inactive (hypophosphory- synergizes with lovastatin to inhibit protein farnesyla- lated) forms of MAPK are observed in control cells tion and which serves as a substrate for geranylger- (Figure 1a, lanes 1 and 5). Treatment of cells with anyltransferases to reverse inhibition of protein GGOH alone had no eect on the processing or signaling geranylgeranylation. of either H-Ras-CVLS or H-Ras-CVLL (Figure 1a, lanes 3 and 7). Lovastatin inhibited the processing of both H- Ras-CVLS and H-Ras-CVLL and inhibited the activation of MAPK slightly in H-Ras-CVLS-transformed Results cells and completely in H-Ras-CVLL-transformed cells (Figure 1a, lanes 2 and 6). GGOH enhanced greatly the GGOH potentiates the ability of lovastatin to inhibit ability of lovastatin to inhibit Ras-CVLS farnesylation oncogenic H-Ras processing and constitutive activation while it restored Ras-CVLL geranylgeranylation (Figure of MAPK 1a, lanes 4 and 8). Furthermore, in the presence of The eects of GGOH on lovastatin inhibition of protein GGOH, lovastatin blocked oncogenic activation of farnesylation were investigated in NIH3T3 cells trans- MAPK in Ras-CVLS-transformed cells, whereas, in formed with an oncogenic, GTP-locked mutant of H- Ras-CVLL-transformed cells, lovastatin was not able to Ras-CVLS (an FTase substrate). NIH3T3 cells trans- inhibit MAPK activation (Figure 1a). The ability of formed with oncogenic H-Ras-CVLL (mutated at its GGOH to enhance the inhibition of oncogenic H-Ras- CAAX box to become a GGTase I substrate) were CVLS processing was concentration-dependent as well treated in parallel and served as a control exhibiting as synergistic. As shown in Figure 1b, GGOH treatment processing that is geranylgeranylation-dependent. Cells alone (0 ± 30 mM; lanes 1 ± 4) as well as 1 mM were treated with lovastatin and GGOH alone or in lovastatin alone (lane 5) did not inhibit H-Ras combination, and the extent of oncogenic H-Ras processing. However, co-treatment of cells with 1 mM processing, as well as its ability to stimulate constitu- lovastatin and 7.5 mM GGOH (Figure 1b, lane 6) tively MAPK, were assessed. After 2 days of treatment, achieved detectable inhibition of H-Ras processing; the cells were harvested and lysed and the lysate proteins signi®cant inhibition was observed with the same subsequently separated by SDS ± PAGE and immuno- concentration of lovastatin and 30 mM GGOH (lane 8). blotted with anti-Ras or anti-MAPK antibodies as While treatment of cells with 5 mM and 15 mM l described under Materials and methods. Cells treated ovastatin produced a partial inhibition of H-Ras with vehicle or GGOH alone exhibited only the processing, co-treatment with 7.5 mM GGOH resulted GGOH/lovastatin: selective inhibition of H-Ras signaling TF McGuire and SM Sebti 307 in a much greater inhibition (Figure 1b, lanes 9, 10, and A 13, 14). Co-treatment of cells with lovastatin/GGOH (15 mM/30 mM) was observed to completely inhibit H- Ras-CVLS processing (Figure 1b, lane 16). When the same nitrocellulose membrane was reprobed with an a b antibody against Rap1A, the processing of this endogenous geranylgeranylated protein was observed to be signi®cantly inhibited at 5 mM and 15 mM lovastatin (Figure 1b, lanes 9 and 13), but was restored upon co- treatment with GGOH at concentrations as low as 7.5 mM (lanes 10 and 14). Quantitative anlaysis of three independent experiments demonstrated that while 15 mM c d lovastatin inhibited H-Ras processing 54% and Rap1A processing 93%, co-treatment with 15 mM GGOH further reduced levels of processed oncogenic H-Ras (80% inhibition) while processing of Rap1A was completely restored (data not shown). Statistical B analysis of the data obtained from these experiments demonstrated that the enhanced inhibition of H-ras processing eected by lovastatin/GGOH co-treatment is statistically dierent from that achieved by using lovastatin alone (P50.01 using a Student's two-tailed paired t-test). Hence, by employing this co-treatment with cells that express constitutively activated H-Ras, the processing and signaling of this oncoprotein can be potently inhibited without aecting the processing/ function of geranylgeranylated proteins.

Geranylgeraniol (GGOH) rescues lovastatin-induced cell rounding, cytoxicity and inhibition of processing of geranylgeranylated proteins Although lovastatin at high enough concentrations (415 mM) could be used to achieve signi®cant inhibition Figure 2 GGOH prevents lovastatin-induced cell rounding and cytotoxicity in NIH3T3 cells transfected with oncogenic H-Ras- of oncogenic H-Ras processing and signaling, this also CVLS. (A) NIH3T3 cells transformed with H-Ras-CVLS(12R) produced signi®cant cell rounding and cytotoxicity. We were plated and then were treated two consecutive days with next determined the extent to which GGOH might either vehicle (a), 40 mM GGOH (b), 20 mM lovastatin (c), or both alleviate lovastatin-induced cytotoxicity in H-Ras- 40 mM GGOH and 20 mM lovastatin (d). (B) cells were plated into transformed NIH3T3 cells. Figure 2A illustrates the 96-well plates (10 000 cells/well) and the following day were treated with various concentrations of lovastatin (0 ± 300 mM)in characteristic rounded cell morphology produced by the presence and absence of 15 mM GGOH. After 4 days, medium treatment of cells with lovastatin alone as well as the was removed from cells and replaced with fresh medium complete prevention of this change by co-treatment with containing 1 mg/ml MTT for 3 h. The number of viable cells GGOH. The ability of GGOH to attenuate lovastatin was determined by the MTT assay described under Materials and methods. Data are representatiave of two independent experi- cytoxicity was assessed by plating cells in 96-well plates ments and then treating with various concentrations of lovastatin alone or in combination with GGOH as described under Materials and methods. Figure 2B shows that while lovastatin produced cytoxicity in H- processing and function of these proteins, like that of

Ras-transformed 3T3 cells with an IC50 of 8 mM, co- Rap1A, are preserved in cells co-treated with lovastatin treatment with GGOH (15 mM) signi®cantly attenuated and GGOH. Although the ability of GGOH to allow lovastatin cytotoxicity, increasing the IC50 by 15-fold. recovery of processing for GGTase protein substrates The ability of GGOH to alleviate lovastatin cytotoxicity in lovastatin-treated cells might be expected and was could be due to a replenishing of either protein implied by previous work (Crick et al., 1994), geranylgeranylation or ubiquinone (ViganoÁ et al., 1995; demonstration of this had not been previously Fenton et al., 1992). The ability of ubiquinone to reverse reported. To further investigate the eect of GGOH lovastatin toxicity was directly tested in the identical on the processing of GGTase protein substrates in system described above for GGOH and was found to lovastatin-treated cells, cells were treated with lovasta- have no eect on preventing lovastatin toxicity (data not tin and GGOH alone and in combination and the shown). Thus, the inhibition of processing/function of processing of Rap1A, RhoB (a substrate for GGTase I protein substrates for GGTases may be the primary reported to be geranylgeranylated or farnesylated cause for lovastatin cytotoxicity. (Adamson et al., 1992)) and Rab5 (a substrate for Rho and Rac proteins, both of which are GGTase II) was assessed. As shown in Figure 3, the geranylgeranylated by GGTase I, have been demon- inhibition by lovastatin of the processing of these three strated to regulate cytoskeleton organization and to geranylgeranylated proteins was reversed in cells co- play a pivotal role in maintaining cellular shape (Nobes treated with lovastatin and GGOH. However, the and Hall, 1995). Thus, it would seem likely that the processing of H-Ras was further inhibited. GGOH/lovastatin: selective inhibition of H-Ras signaling TF McGuire and SM Sebti 308 1234Blot: Finally, to assess the eect of GGOH co-treatment on the processing of all GGTase I protein substrates in NIH3T3 cells we used a novel approach. Cells were Ras treated with lovastatin (50 mM) and GGOH (25 mM) alone or in combination, harvested and cytosolic fractions prepared and used as a source of protein substrates in an in vitro GGTase I assay as described under Materials and methods. As shown in Figure 4, the Rap1A cytosolic fractions obtained from cells treated with vehicle or GGOH alone contained no unprocessed proteins capable of serving as substrates in the in vitro assay (lanes 1 and 2). However, the cytosolic fractions from lovastatin-treated NIH3T3 cells contained several RhoB protein substrates for GGTase I, as demonstrated by the pro®le of tritium-labeled geranylgeranylated proteins ranging from 21 ± 29 kDa that was detected using this assay (Figure 4, lane 3). When the assay was performed on the cytosolic fractions from NIH3T3 cells co-treated with lovastatin and GGOH, all of the protein substrates Rab5 for GGTase I (detected in the cytosolic fractions from lovastatin-treated cells) were observed to be signi®cantly reduced (Figure 4, lane 4). In longer exposures, a few Figure 3 GGOH rescues lovastatin-induced inhibition of geranylgeranylated protein processing. The identical cells shown other bands of higher and lower molecular weight in Figure 2A were harvested, lysed, and whole cell lysate protein appeared in the lane containing cytosol from lovasta- (50 mg) was electrophoresed and immunoblotted for Ras, Rap1A, tin-treated cells, but these also were not detected in the RhoB, and Rab5 using appropriate antibodies as described in cytosol from cells co-treated with lovastatin and GGOH Materials and methods. Lanes 1 ± 4 represent cells treated with (data not shown). Thus, this novel approach demon- vehicle, 40 mM GGOH, 20 mM lovastatin, or both 40 mM GGOH and 20 mM lovastatin, respectively strates that GGOH is converted to a form capable of rescuing the processing of virtually all proteins that are geranylgerantylated by GGTase I in the cell.

3 12345 Metabolic labeling of proteins using [ H]GGOH In order to con®rm that GGOH was being used as a —112 metabolic source for protein prenylation in mouse ®broblasts and to investigate the metabolic fate of —84 GGOH in the cell, H-Ras-CVLS-transformed NIH3T3 cells were treated with [3H]GGOH and 20 mM lovastatin for various lengths of time and the cellular lipids and the delipidated proteins subsequently analysed (see Materials and methods). As shown in Figure 5, a pattern of labeled —53 proteins (MW range of 21 ± 29 kDa) resembling that obtained for small G proteins was observed. Optimal metabolic labeling of protein was found to occur when unlabeled GGOH was added such that the ®nal concentration was increased from 0.6 mM ([3H]-GGOH —35 alone) to 5.0 mM (Figure 5; compare lanes 1 and 2), and when cells were pre-treated overnight with 20 mM lovastatin prior to metabolic labeling (Figure 5; compare lanes 2 and 3). These results suggest that GGOH is being —29 metabolically converted by mouse ®broblasts to an activated form and used for protein geranylgeranylation. In order to assess whether this form is GGPP itself, thin-layer chromatogaphy (TLC) was performed on the lipid extracts of cells labeled for various lengths of time with [3H]GGOH as described under Materials and —21 methods. Figure 6 shows that [3H]GGOH and [3H]GGPP were well-separated in the TLC system used (lanes 1 and 2, respectively). Although signi®cant Figure 4 Eect of GGOH on lovastatin-induced accumulation of 3 GGTase I protein substrates in the cytosol. The cytosolic (60S) metabolism of [ H]GGOH is observed over the time fraction was obtained from NIH3T3 cells treated with vehicle course, no labeled metabolite corresponding to (lane 1), 25 mM GGOH (lanes 2 and 4) and 50 mM lovastatin [3H]GGPP was observed (Figure 6, lanes 4 ± 8). How- (lanes 3 and 4) and used as the source of protein substrates for an ever, after 24 h, at least four labeled, GGOH-derived in vitro geranylgeranylation assay as described under Materials products, migrating between GGPP and GGOH, were and methods. Equivalent amounts of protein (160 mg) were used for each sample. Recombinant, non-prenylated H-Ras-CVLL detected but their identity is unknown. It is unlikely that, (2.5 mg; lane 5) was used as a positive control had [3H]GGPP been formed within the cells, that GGOH/lovastatin: selective inhibition of H-Ras signaling TF McGuire and SM Sebti

309 123 ¨ SF

112—

84— GGOH ¨

53— 35— GGPP ¨ ¨ Origin 12345678 Figure 6 Thin-layer chromatography of lipid extracts from cells 29— labeled with [3H]GGOH. Lipid extracts of cells labeled with [1-3H]GGOH for various lengths of time (10, 30 min, 1, 2, and 24 h, lanes 4 ± 8, respectively) were prepared as described under Materials and methods, 100 000 c.p.m. of each spotted onto a silica gel plate, and the samples run as described under Materials and methods. Lanes 1 and 2 demonstrate the migration of [3H]GGOH and [3H]GGPP, respectively. Lane 3 shows [3H]GGPP that was incubated under the conditions of the 21— extracts (CHCl3/CH3OH, 2 : 1). Data are representative of three independent experiments Figure 5 [3H]GGOH metabolic labeling of proteins. NIH3T3 cells transfected with H-Ras-CVLS(61L) were metabolically 3 labeled for 21 ± 22 h with 0.6 mM [ H]GGOH (30 mCi/ml) in the presence of 20 mM lovastatin and in the absence (lane 1) or the farnesylation and restoration of protein geranylgeranyla- presence (lanes 2 ± 3) of 4.4 mM unlabeled GGOH. In addition, the eect of pre-treating cells overnight with 20 mM lovastatin prior to tion, respectively, are responsible for these two opposing metabolic labeling was also assessed (lane 3; see Materials and eects of GGOH. The fact that GGOH had no apparent methods). Equivalent proportions of each sample were applied to eect on protein prenylation in the absence of lovastatin SDS ± PAGE. Lanes 1 ± 3 contained 8100 c.p.m., 27 900 c.p.m. (Figures 1, 3 and 4) strongly suggests that it acts to and 40 000 c.p.m., respectively. Fluorography was performed on the dried gel using a 9-day exposure at 7808C. Data are replenish a metabolite that inhibits FTase and that is a representative of four independent experiments substrate for GGTases I and II. While the identity of the metabolite is not known, our results suggest that a GGOH derivative other than GGPP was formed and was responsible for restoring geranylgeranylation and enzymatic breakdown would occur in the organic solvents further inhibiting farnesylation. Furthermore, it is

(CH3OH and CHC13/CH2OH, 2 : 1) used for extraction of interesting to note that this metabolite can be used as a the cellular lipids (see Materials and methods). Controls substrate by two dierent GGTases, as indicated by the indicated that no breakdown of [3H]GGPP occurs during preservation of the processing for speci®c protein sample handling and preparation (Figure 6, lane 3). Since substrates (Rap1A and RhoB for GGTase I and Rab5 a signi®cant pool of intracellular GGPP would be for GGTase II; see Figure 3). In addition, it was expected to be required to inhibit FTase (and, therefore, previously demonstrated that when [3H]GGOH metabo- H-Ras processing) within lovastatin-treated cells (see lically-labeled proteins were subjected to Pronase E Discussion below), this should have been easily detected digestion, a labeled product that was chromatographi- in our system. Hence, these data suggest that the GGOH- cally identical to geranylgeranyl-cysteine was released derived metabolite that is responsible for both the (Crick et al., 1994). By achieving the selective restoration restoration of protein geranylgeranylation, as well as for of protein geranylgeranylation, not only could cell further inhibition of H-Ras processing, is not GGPP. morphology be maintained at the same phenotype as Further investigation is required to isolate and identify control cells (Figure 2A), but much of the cytotoxic eect the active GGOH metabolite. of lovastatin could be alleviated (Figure 2B). The maintenance of cell shape as well as cell cycle progression

through G1 might be expected since these cellular Discussion processes have recently been shown to be regulated by geranylgeranylated proteins (Rho and Rac; see Nobes The results presented herein clearly demonstrate that and Hall, 1995; and Olson et al., 1995). Hence, taking all GGOH potentiates the anti-oncogenic/anti-signaling of these results together, GGOH potentiates the activity of lovastatin while antagonizing its cytotoxic inhibitory eect of lovastatin on protein farnesylation eects. The data suggest that inhibition of H-Ras while preserving the cell from the deleterious eects of a GGOH/lovastatin: selective inhibition of H-Ras signaling TF McGuire and SM Sebti 310 block on protein geranylgeranylation. These results, Materials and Methods therefore, raise the possibility of using a combination of lovastatin/GGOH as a chemotherapeutic regimen against Cell culture Ras-dependent human tumors. Further investigation of NIH3T3 mouse ®broblasts (ATCC) were maintained in the eects of this co-treatment on human tumor growth in Dulbecco's modi®ed Eagles medium (DMEM) supplemen- nude mice is needed to assess this possibility. ted with 10% fetal bovine serum and 1% Pen-Strep (Life Based on the results presented here, we have Technologiess Inc.). NIH3T3 cells transfected with the proposed a model depicting the mechanism by which oncogene encoding H-Ras-CVLS(12R) or H-Ras- GGOH confers to lovastatin a selective and enhanced CVLS(61L) or with the empty vector (pZIPneo) were ability to inhibit protein farnesylation and subsequent kind gifts from Dr Channing Der and Dr Adrienne Cox MAPK activation (Figure 7). As discussed above, (University of North Carolina, Chapel Hill) and were maintained in DMEM supplemented with 10% calf serum, GGOH is suggested to be taken up and converted by 1% Pen-Strep and 400 mg/ml G418. cells to an activated form (designated GGX in Figure 7) which serves as the geranylgeranyl donor for protein prenylation catalyzed by GGTases I and II (only Immunoblotting to assess protein processing and MAP kinase GGTase I is depicted in Figure 7). Moreover, GGX is activation proposed to competitively inhibit FPP from the active Cells were seeded into 100 mm plates (8.06105 ±1.06106/ site of FTase, further inhibiting oncogenic H-Ras plate) on day 0 such that the following day they were 50 ± processing. Previous work demonstrated that GGPP 70% con¯uent. Cells were treated twice (days 1 and 2) with can bind to the active site of FTase with a similar vehicle (10 mM DTT in DMSO) or concentrations of anity to that of FPP but cannot be used as a lovastatin and/or GGOH as indicated in the ®gure legends. On day 3, cells were harvested in ice cold PBS, pH 7.5, substrate for protein prenylation (Reiss et al., 1992). pelleted and lysed in 50 mM HEPES, pH 7.5, 10 mM NaCl, Moreover, we have observed in vitro that GGPP 1% Triton X-100, 10% glycerol, 5 mMMgCl2,1mMEGTA, inhibits FTase with an IC of 5 mM (data not 50 25 mg/ml leupeptin, 2 mM Na3VO4,10mg/ml soybean trypsin shown). Thus, in the presence of lovastatin, endogen- inhibitor, 10 mg/ml aprotinin and 6.4 mg/ml phosphatase ous pools of FPP are low and GGX can act to potently substrate. After clearing the lysates (14 000 r.p.m., 48C, inhibit FTase (Figure 7): in the absence of lovastatin, 10 min), equivalent amounts of protein were applied to SDS- no inhibition of H-Ras processing is observed since the polyacrylamide gels (12.5% for protein processing; 15% for levels of GGX formed are not high enough to compete MAP kinase), separated by electrophoresis, and subse- out endogenous FPP from the FTase active site quently transferred to nitrocellulose ®lters. Filters were (Figures 1 and 3). Therefore, the novel approach blocked with 5% non-fat dry milk in PBS, 0.1% Tween-20 (PBS-T) and then probed with either anti-Ras (Y13-238 or described in this manuscript is a powerful tool for Y13-259, ATCC), anti-Rap1A (Santa Cruz Biotechnology, selectively blocking oncogenic H-Ras signaling while Santa Cruz, CA), anti-RhoB (Santa Cruz Biotechnology, sparing cells from lovastatin cytotoxicity due to Santa Cruz, CA), anti-Rab5 (Transduction Laboratories, inhibition of protein geranylgeranylation. Lexington, KY), or anti-MAP kinase (erk2,UBI,Lake Placid, NY) antibodies in 3% non-fat dry milk in PBS-T. Positive antibody reactions were detected using appropriate horseradish peroxidase-conjugated antibodies (Oncogene Science and Jackson ImmunoResearch Laboratories Inc., West Grove, PA) and an enhanced chemiluminescence detection system (ECL, Amersham Corp.).

[3H] GGOH metabolic labeling of cellular proteins and thin- layer chromatography of lipid extracts NIH3T3 cells transfected with H-Ras-CVLS(61L) were seeded into 12-well plates (200 000 cells/well) and incubated the following day (at 90 ± 100% con¯uency) with tritiated GGOH in a manner similar to that described by Crick et al. (1994). Cells were incubated with 500 ml DMEM medium containing 30 mCi/ml [1-3H]GGOH (50 ± 60 Ci/mmole; 0.6 mM GGOH, ®nal), 5% calf serum and 20 mM lovastatin. As noted in the legend to Figure 5, some cells were treated with additional unlabeled GGOH which increased the ®nal GGOH concentration to 5 mM but decreased the speci®c activity by 8.7-fold. In addition, some cells were pre-treated overnight with 20 mM lovastatin prior to metabolic labeling. After 21 ± 22 h at 378C, the labeling medium was removed and the cells were washed once with 1.5 ml ice-cold PBS, pH 7.5 and harvested using two 1.0 ml aliquots of ice-cold PBS, pH 7.5. The cells were Figure 7 A proposed model depicting the mechanism by which spun down and the pellets were disrupted in ice-cold GGOH acts, in the presence of lovastatin, to perserve protein CH3OH. CH3OH extracts were removed from the protein geranylgeranylation and concomitantly potentiate inhibition of pellets and the pellets were extracted twice more with protein farnesylation. Dashed lines indicate decreased FPP and CHCl /CH OH (2 : 1). The extracts for each sample were GGPP biosynthesis due to the inhibition of HMG-CoA reductase 3 3 (and, thus, the MVA pathway) by lovastatin. GGOH is taken up then combined, concentrated under a stream of N2,and by cells and is converted to a metabolite GGX that serves as an 100 000 c.p.m. of each was spotted on Kieselgel 60 silica inhibitor of FTase and as a substrate for GGTases (only GGTase gel plates (EM Separations, Gibbstown, NJ). The spotted I shown) lipid extracts were chromatographed using a n-propanol/ GGOH/lovastatin: selective inhibition of H-Ras signaling TF McGuire and SM Sebti 311

5N NH4OH/H2O (6 : 1 : 1) solvent system and the labeled sample buer (2X). Samples were electropheresed on 12.5% products detected by ¯uorography using EN3Hance SDS-polyacrylamide gels and the gels processed for (DuPont NEN, Boston, MA). The delipidated protein ¯uorography as described above for [3H]GGOH metabolic pellets were air dried and subsequently dissolved in 90 ml labeling. SDS ± PAGE sample buer. For each sample, an aliquot (5 ml) was used to determine the amount of tritium MTT cytotoxicity assay incoroporated into protein and 41 ml was loaded onto a 12.5% SDS-polyacrylamide gel. After electrophoretically NIH3T3 cells transfected with either the H-Ras-CVLS(61L) separating the labeled proteins, the gels were ®xed with oncogene or the empty vector were seeded into 96-well plates

CH3OH/CH3CO2H/H2O (9 : 2 : 9), treated with ENTENSI- (10 000/well) and the following day treated (at 50 ± 70% FY (DuPont NEN, Boston, MA), dried down on con¯uency) with increasing concentrations of lovastatin (0 ± Whatmam paper, and exposed to X-ray ®lm for 9 days at 300 mM) in the presence or the absence of either 15 mM 7808C. GGOH or 30 mM ubiquinone (Sigma, St. Louis, MO). After 4 days, the medium was replaced with 100 mlofMTT(1mg/ ml) in DMEM. After 3 h incubation at 378C, the In vitro GGTase I Assay of 60S Proteins tetrazolium/formazan reaction was stopped and the uptake NIH3T3 cells were seeded into 100 mm plates (3.86106 cells/ of MTT assessed by replacing the medium with 100 ml plate) and were treated the following day (about 70% DMSO, shaking the plate for 5 min to solubilize all the dye, con¯uency) with either vehicle, 25 mM GGOH, 50 mM and measuring the absorbance at 492 nm with a Titertek lovastatin, or 50 mM lovastatin and 25 mM GGOH. After Multiskan spectrophotometer (Flow Laboratories, 40 h, the cells were washed, harvested in ice-cold PBS, McClean, VA). pH 7.5, and the cell pellet volume (PV) estimated. Cells were disrupted by sonication in 1.46PV of 50 mM TRIS, pH 7.5, 1mMEDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF, 25mg/ ml leupeptin, and 10 mg/ml aprotinin and the cytosolic (60S) fraction prepared by centrifugation at 25 000 r.p.m. for 1 h Abbreviations at 48C using a Beckman SWTI55 swinging bucket rotor. FTase, farnesyltransferase, GGTase, geranylgeranyltrans- Equivalent amounts of cytosolic protein (160 mg) were ferase; FPP, farnesylpyrophosphate; GGPP, geranylgera- incubated in 50 mM TRIS, pH 7.5, 50 mM ZnCl2,20mM nylpyrophosphate; GGOH, geranylgeraniol; HMG-CoA,

KCl, 3 mM MgCl2,1mM DTT with exogenously added hydroxymethylglutaryl-coenzyme A; SDS ± PAGE, sodium human GGTase I (Mono Q-puri®ed from human Burkitt dodecyl sulfate polyacrylamide gel electrophoresis; MAPK, lymphoma (Daudi) cells). Prenylation of 60S proteins was mitogen-activated protein kinase; DTT, dithiothreitol; started by addition of 4.8 mCi [3H]GGPP and the reaction DMSO, dimethylsulfoxide; PMSF, phenylmethylsulfonyl allowedtoproceedfor45±60minat378C. Total reaction ¯uoride; MVA, mevalonic acid; PBS, phosphate-buered volume was 55.4 ml. The reaction was stopped by brie¯y saline; CAAX, C=cysteine, A=aliphatic, X=any amino placing tubes on ice and then adding 25mlSDS±PAGE acid.

References

Adamson P, Marshall CJ, Hall A and Tilbrook PA. (1992). J. Inglese J, Glickman JF, Lorenz W, Caron M and Lefkowitz Biol. Chem., 267, 20033 ± 20038. RJ. (1992). J. Biol. Chem., 267, 1422 ± 1425. Alberts AW. (1988). Am. J. Cardiol., 62, 10J ± 15J. James GL, Goldstein JL and Brown MS. (1995). J. Biol. Bar-Sagi D and Feramisco JR. (1985). Cell, 42, 841 ± 848. Chem., 270, 6221 ± 6226. Brown MS and Goldstein JL. (1980). J. Lipid Res., 21, 505 ± Kato K, Cox AD, Hisaka MM, Graham SM, Buss JE, Der 517. CJ. (1992). Proc. Natl. Acad. Sci. USA., 89, 6403 ± 6407. Casey PJ. (1992). J. Lipid Res., 33, 1731 ± 1740. Leonard S, Beck L, Sinensky M. (1990). J. Biol. Chem., 265, Casey PJ, Solski PA, Der CJ and Buss JE. (1989). Proc. Natl. 5157 ± 5160. Acad. Sci. USA, 86, 8323 ± 8327. Lerner EC, Qian, Y, Blaskovich MA, Fossum RD, Vogt A, Crick DC, Waechter CJ and Andres DA. (1994). Biochem. Sun J, Cox AD, Der CJ, Hamilton AD and Sebti SM. Biophys. Res. Comm., 205, 955 ± 961. (1995). J. Biol. Chem., 270, 26802 ± 26806. Der CJ and Cox AD. (1991). Cancer Cells, 3, 331 ± 340. Lerner E, Qian Y, Hamilton AD and Sebti SM. (1995). J. Epstein WW, Lever DC and Rilling HC. (1990). Proc. Natl. Biol. Chem., 270, 26770 ± 26773. Acad. Sci. USA, 87, 7352 ± 7354. Maltese WA. (1990). FASEB J., 4, 3319 ± 3328. Farnsworth CC, Gelb MH and Glomset JA. (1990). Science, McCormick F. (1993). Nature, 363, 15 ± 16. 247, 320 ± 322. McGuire TF, Corey S and Sebti SM. (1993). J. Biol. Chem., Fenton RG, Kung H-F, Longo DL and Smith MR. (1992). J. 268, 22227 ± 22230. Cell Biol., 117, 347 ± 356. Mulcahy LS, Smith MR and Stacey DW. (1985). Nature, Fukada Y, Takao T, Ohguro H, Yoshizawa T, Akino T and 313, 241 ± 243. Shimonishi Y. (1990). Nature, 346, 658 ± 660. Nobes CD and Hall A. (1995). Cell, 81, 53 ± 62. GruÈ nler J, Ericsson J and Dallner G. (1994). Biochim. Olson MF, Ashworth A and Hall A. (1995). Science, 269, Biophys. Acta., 1212, 259 ± 277. 1270 ± 1272. Hancock JF, Magee AI, Childs JE and Marshall CJ. (1989). Quesney-Huneeus V, Wiley MH and Siperstein MD. (1979). Cell, 57, 1167 ± 1177. Proc.Natl.Acad.Sci.USA,76, 5056 ± 5060. Holtz D, Tanaka RA, Hartwig J and McKeon F. (1989). Reddy EP, Reynolds RK, Santos E and Barbacid M. (1982). Cell, 59, 969 ± 977. Nature, 300, 149 ± 152. Hori Y, Kikuchi A, Isomura M, Katayama M, Miura Y, Reiss Y, Brown MS and Goldstein JL. (1992). J. Biol. Chem., Fujioka H, Kaibuchi K and Takai Y. (1991). Oncogene, 6, 267, 6403 ± 6408. 515 ± 522. Sabine JR. (1983). CRC Press, Boca Raton FL. Illingworth DR and Bacon S. (1989). Arterioscleriosis, 9, I- Seabra MC, Reiss Y, Casey PJ, Brown MS and Goldstein JL. 121 ± I-134. (1991). Cell, 65, 429 ± 434. GGOH/lovastatin: selective inhibition of H-Ras signaling TF McGuire and SM Sebti 312 Sebti SM, Tkalcevic GT and Jani JP. (1991). Cancer Comm., ViganoÁ T, Hernandez A, Corsini A, Granata A, Belloni P, 3, 141 ± 147. Fumagalli R, Paoletti R and Folco G. (1995). Eur. J. Sinensky M and Logel J. (1985). Proc. Natl. Acad. Sci. USA, Pharmacol., 291, 201 ± 203. 82, 3257 ± 3261. Sun J, Qian Y, Hamilton AD and Sebti SM. (1995). Cancer Res., 55, 4243 ± 4247.