was modulating the response of the mice to theparasite.
11. S. Mellouk, S. J. Green, C. A. Nacy, S. L. Hoffman,J. Immunol. 146, 3971 (1991).
12. A. Nussler et al., Eur. J. Immunol 21, 227 (1991).13. K. A. Rockett, M. M. Awburn, W. B. Cowden, I. A.
Clark, Infect. Immun. 59, 3280 (1991).14. M. A. Marletta, P. S. Yoon, R. lyengar, C. D. Leaf,
J. S. Wishnok, Biochemistry27, 8706 (1988); D. J.Stuehr and C. F. Nathan, J. Exp. Med. 169,1543(1989).
15. R. G. Knowles and S. Moncada, Trends Biochem.Sci. 17, 399 (1992).
16. F. Y. Liew etal., Eur. J. Immunol. 21, 2489 (1991).17. T. R. Mosmann and R. L. Coftman, Immunol.
Today 8, 223 (1987).18. D. W. Taylor, in Malaria: Host Responses to Infec-
tion, M. M. Stevenson, Ed. (CRC Press, BocaRaton, FL, 1990), pp. 1-35.
19. S. A. McLean, C. D. Pearson, R. S. Phillips, Exp.Parasitol. 54, 296 (1982).
20. M. M. Stevenson and E. Skamene, Infect. Immun.54, 600 (1986).
21. K. A. Rockett, M. M. Awburn, B. B. Aggarwal, W.B. Cowden, I. A. Clark, ibid. 60, 3725 (1992).
22. Mice (six per group) were CD4-depleted (8).Twelve days after the last antibody treatment, theywere injected intravenously with 4 x 107 cells permouse, or not reconstituted, and then infectedintravenously immediately with 105 P. c. chabaudipRBC. The preparation of the T cell clone has
been described in detail (6). We enriched nalivesplenic T cells for CD4+ cells using a mouse T cellseparation kit (Pierce). Some of the mice recon-stituted with TH1 or TH2 were injected intraperito-neally with L-NMMA (250 mg/kg per day) dis-solved in sterile PBS. Courses of infection andnitrate levels in the sera were monitored as de-scribed (4, 5). Data shown are pooled from threeseparate experiments.
23. We assayed parasite-specific antibody in the se-rum by an indirect fluorescent antibody test [S. A.McLean, C. D. Pearson, R. S. Phillips, Exp. Para-sitol. 54, 213 (1982)], using trophozoite-schizont-infected RBC as the target antigen. Isotype-spe-cific antibodies were obtained from serum sepa-rated by affinity chromatography on proteinA-Sepharose CL-4B into fractions containingIgGl, IgG2a, IgG2b, or IgG3 [I. Seppald, H.Sarvar, F. Peterfy, 0. Makela, Scand. J. Immunol.14, 335 (1981)]. Immunoglobulin M was isolatedby Sepharose 6B gel filtration from nonbindingmaterial that contained a pool of IgA and IgE. Weconfirmed the purity of each fraction by doubleradial immunodiffusion, using antibodies specificto each 1g isotype.
24. We thank H. Hodson for L-NMMA and D-NMMAand E. Galloway and D. McLaughlin for technicalsupport. We thank The Wellcome Trust for finan-cial support.
24 December 1992; accepted 4 May 1993
Selective Inhibition of ras-DependentTransformation by a Farnesyltransferase Inhibitor
Nancy E. Kohl, Scott D. Mosser, S. Jane deSolms,Elizabeth A. Giuliani, David L. Pompliano, Samuel L. Graham,
Robert L. Smith, Edward M. Scolnick, Allen Oliff,*Jackson B. Gibbs
To acquire transforming potential, the precursor of the Ras oncoprotein must undergofarnesylation of the cysteine residue located in a carboxyl-terminal tetrapeptide. Inhibitorsof the enzyme that catalyzes this modification, farnesyl protein transferase (FPTase), havetherefore been suggested as anticancer agents for tumors in which Ras contributes totransformation. The tetrapeptide analog L-731,735 is a potent and selective inhibitor ofFPTase in vitro. A prodrug of this compound, L-731,734, inhibited Ras processing in cellstransformed with v-ras. L-731,734 decreased the ability of v-ras-transformed cells to formcolonies in soft agar but had no effect on the efficiency of colony formation of cellstransformed by either the v-raf or v-mos oncogenes. The results demonstrate selectiveinhibition of ras-dependent cell transformation with a synthetic organic inhibitor of FPTase.
The mammalian ras genes encode guano-sine triphosphate (GTP)-binding proteinsthat can acquire the potential to transformmammalian cells as a result of point muta-tions in codons 12, 13, or 61 (1). Mutated,oncogenic forms of ras are frequently foundin many human cancers, most notably inmore than 50% of colon and pancreaticcarcinomas (1, 2). These observations in-
N. E. Kohl, S. D. Mosser, D. L. Pompliano, E. M.Scolnick, A. Oliff, J. B. Gibbs, Department of CancerResearch, Merck Research Laboratories, West Point,PA 19486.S. J. deSolms, E. A. Giuliani, S. L. Graham, R. L. Smith,Department of Medicinal Chemistry, Merck ResearchLaboratories, West Point, PA 19486.*To whom correspondence should be addressed.
1934
dicate that Ras functions in the pathogen-esis of human cancers and emphasize thepotential broad utility of anticancer agentsdirected against ras-induced cell transfor-mation.
Ras is synthesized as a cytosolic precur-sor that ultimately localizes to the cytoplas-mic face of the plasma membrane after a
series of posttranslational modifications (3).The first and obligatory step in this series isthe addition of a farnesyl moiety to thecysteine residue of the COOH-terminalCAAX motif (C, cysteine; A, usually ali-phatic residue; X, any other amino acid) ina reaction catalyzed by farnesyl proteintransferase (FPTase). This modification isessential for Ras function, as demonstrated
SCIENCE * VOL. 260 * 25 JUNE 1993
by the inability of Ras mutants lacking theCOOH-terminal cysteine to be famesyl-ated, to localize to the plasma membrane,and to transform mammalian cells in cul-ture (4, 5). Moreover, strains of Saccharo-myces cerevisiae having a mutation inRAM1, a gene that encodes one of thestructural polypeptides of the yeast FPTase,are resistant to the biological effects ofoncogenic ras (6). The subsequent post-translational modifications-cleavage ofthe AAX residues, carboxyl methylation ofthe farnesylated cysteine, and palmitoyla-tion of cysteines located upstream of theCAAX motif-are not obligatory for Rasmembrane association or cell-transformingactivity (5, 7). Thus, FPTase appears to bean appropriate biochemical target for thedevelopment of inhibitors of posttransla-tional processing of Ras that might beexpected to interfere with Ras-mediatedcellular transformation.
The substrates of the farnesylation reac-tion, famesyl diphosphate (FPP) and poly-peptides containing a CAAX motif, can beused as a starting point for the design ofFPTase inhibitors. Several analogs of FPPare potent and selective inhibitors ofFPTase in vitro (8, 9), and one has shownactivity in cells (9). The CAAX tetrapep-tide is the minimal sequence required forthe interaction of Ras with FPTase (10,11). Thus, tetrapeptides with amino acidsequences identical to the COOH-terminalsequences of protein substrates for FPTasecompete with Ras for farnesylation by act-ing as alternative substrates (12, 13).CAAX derivatives have also been identifiedthat are not substrates for farnesylation andtherefore behave as pure inhibitors ofFPTase (8, 13, 14).
Although other cellular proteins besidesRas are in vivo substrates for farnesylation,most isoprenylated proteins are modified bythe 20-carbon geranylgeranyl moiety (15).Two classes of enzymes that catalyze theaddition of a geranylgeranyl group to pro-teins have been identified in mammaliancells. One class consists of geranylgeranylprotein transferase-I (GGPTase-I), whichmodifies proteins having a COOH-terminalCAAX sequence where X is a Leu residue(12, 16). The other class consists of one ormore enzymes (for example, GGPTase-II)that catalyze the modification of proteinsterminating in Cys-Cys or Cys-X-Cys (12,17). Comparison of the activity of variousCAAX tetrapeptides suggests that it may bepossible to design a specific inhibitor ofFPTase that does not affect the GGPTases.For example, CAAX tetrapeptides termi-nating in Ser or Met are potent inhibitorsof FPTase but are less active as inhibitors ofGGPTase-I or GOPTase-II (12). Con-versely, CAAX tetrapeptides terminating inLeu, which act as effective inhibitors of
%88. QE 100 RIORM
on
Sep
tem
ber
27, 2
012
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om
GGPTase-I, are relatively poor inhibitors ofFPTase (12, 18).
To obtain selective antagonists ofFPTase, we examined CAAX tetrapeptidesas models for the development of inhibi-tors. The COOH-terminal tetrapeptide ofhuman K4A-Ras, CIM, was used as atemplate to design the tetrapeptide analogsL-731,734, which is N-{2(S)-[2(R)-amino-3-mercaptopropylamino]-3 (S )-methyl-pentyl}isoleucyl-homoserine lactone, andL-731,735, which is N-{2(S )-[2(R)-amino-3-mercaptopropylaminoJ-3 (S ) -methyl-pentyl}isoleucyl-homoserine (Fig. 1). Thesecompounds differ from CIIM in two respects.First, in both compounds, the two NH2-terminal peptide bonds were reduced. Sec-ond, either homoserine lactone or homo-serine was substituted for methionine. TheNH2-terminal peptide bonds were reducedin order to confer resistance to hydrolysis byaminopeptidases, enzymes commonly foundin mammalian cell extracts (19). Homo-
L-731,734
H X
HH
H2N - H
- 0 >
HO
L-731.735
Fig. 1. Structure of L-731,734, which is N-{2(S)-[2(R)-amino-3-mercaptopropylamino]-3(S)-methylpentyl}isoleucyl-homoserine lac-tone, and L-731,735, which is N{2(S)-[2(R)-ami-no-3-mercaptopropylamino]-3(S)-methylpentyl}isoleucyl-homoserine.
Table 1. Selective inhibition of farnesyl proteintransferase. Prenyl protein transferase assayswere done essentially as described (12) andcontained the following: FPTase, 100 nM[3H]FPP and 650 nM E. col/iproduced Ras-CVLS; GGPTase-l, 100 nM [3H]GGPP and 500nM E. co/-produced Ras-CAIL (I, lie);GGPTase-ll, 100 nM [3H]GGPP and 500 nM E.col-produced YPT1. All enzymes were partiallypurified from bovine brain cytosol as described(12). FPTase values are the average of fourseparate determinations + SEM. GGPTase val-ues represent three determinations. IC., 50%inhibitory concentration.
Com- IC. (nM)pound FPTase GGPTase-l GGPTase-ll
L-731,734 282 ±41 >100,000 >100,000L-731,735 18 6 >100,000 >100,000
serine was anticipated to be tolerated as areplacement for methionine because of itsstructural similarity to serine, which is foundin the X position of the Ha-Ras (20) and S.cerevisiae RAS2 (21) CAAX motifs. Al-though this compound (homoserine in the Xposition) was anticipated to be an effectiveinhibitor in vitro, the charge on the COOH-terminal carboxylate was perceived as a pos-sible impediment to the entry of CAAXtetrapeptides into whole cells. Cyclization ofhomoserine to a lactone might facilitate cellpenetrability by masking the anionic carbox-ylate and increasing the solubility of thecompound in lipid.
L-731,735 exhibited potent inhibitionof FPTase in vitro. In assays containing[3Hjfamesyl diphosphate and Ras producedin Escherichia coli (Ras-CVLS: V, Val; L,Leu; S, Ser), 50% inhibition (IC50) ofpartially purified FPTase from bovine brainwas observed at a concentration of 18 nML-731,735 (Table 1). The related lactonecompound, L-731,734, was less potent ininhibiting FPTase (IC50 = 282 nM). Theseresults suggest that the COOH-terminalcarboxylate is an important determinant ofintrinsic FPTase inhibitory potency and areconsistent with our observation that amida-tion of the COOH-terminal carboxyl groupof the tetrapeptide CVLS similarly de-creased FPTase inhibitory activity (8).When evaluated as inhibitors of homoge-neous recombinant human FPTase, IC50values for L-731,734 and L-731,735 werecomparable to those observed with the bo-vine enzyme (22). L-731,735 is a nonsub-strate inhibitor that is competitive withrespect to Ras [inhibition constant (K,) =
20 ± 6 nM] and noncompetitive with
respect to FPP in the FPTase reaction.Although L-731,735 was a potent inhibitorof FPTase, it was a less effective (>5000times) inhibitor of the type I and type IIGGPTases (Table 1).
NIH 3T3 cells transformed by v-ras wereused to evaluate the effect of these com-pounds on the posttranslational processingof Ras in intact cells. The cells were incu-bated in the presence of the indicatedcompound for 24 hours and were labeledwith [35Slmethionine during the final 20hours. Ras was immunoprecipitated fromdetergent lysates of cell extracts with themonoclonal antibody to Ras, Y13-259 (23).Viral Ha-Ras contains a threonine at aminoacid position 59 that is a substrate forautophosphorylation (24). The viral Ras inthe transfected cells consists of a mixture ofphosphorylated and unphosphorylated pro-tein with approximately 25% being phos-phorylated (25). Farnesylation and phos-phorylation both alter the migration of viralHa-Ras during SDS-polyacrylamide gelelectrophoresis (SDS-PAGE) (Fig. 2A). Inthe presence of lovastatin, a compound thatblocks Ras processing in cells by inhibitinga rate-limiting step in the isoprenoid bio-synthetic pathway (5, 26, 27), the phos-phorylated and unphosphorylated forms ofRas migrated more slowly (Fig. 2A), indi-cating a lack of posttranslational process-ing. A similar pattern of migration wasobserved for cells incubated in the presenceof 250 pM L-731,734 (Fig. 2A), demon-strating that this compound similarlyblocked processing of Ras in cells. Furtherinhibition of Ras processing was not ob-served at higher concentrations ofL-731,734 (up to 1 mM) (28). Titration of
Fig. 2. Inhibition of Ras pro- A L-731,734 B L-731,735cessing by L-731,734. Meta-bolic labeling and immuno- I =L =Lprecipitation of Ras protein 0 0° 8 o 0 o 8gwas done essentially as de- 5' '0 (- cm
scribed (26). Briefly, NIH3T3 cells transformed by vi-ral H-ras were incubatedwith the indicated concen-tration of L-731,734 (A) or -45 -45L-731,735 (B) (dissolved inmethanol; the final concen-tration of methanol in the as-
-29say was 0.1 %) for 4 hours, at UPwhich time fresh compound p -29was added together with UP[35S]methionine (133 ,uCi/ml) p -
(Amersham). After incuba- -18 p23 4tion for another 20 hours,Ras was immunoprecipi- 1 2 3 4 5
tated from detergent lysatesof cell extracts with monoclonal antibody to Ras Y13-259 (10 ,g), resolved by SDS-PAGE (13%gels), and detected by fluorography. MeOH, 0.1% methanol; Lovastatin, lovastatin (15 [iM);Peptide, competition of the antibody with an excess of the peptide spanning the epitope (Ras60-76; -100-fold molar excess); P, processed, unphosphorylated Ras; U, unprocessed, unphos-phorylated Ras; PP, processed, phosphorylated Ras; and UP, unprocessed, phosphorylated Ras.Molecular sizes of the protein standards are indicated on the right (in kilodaltons).
1935SCIENCE * VOL. 260 0 25 JUNE 1993
on
Sep
tem
ber
27, 2
012
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om
L-731,734 in this assay indicated that thecompound inhibited Ras processing with anIC50 of -100 FM (Fig. 2A). L-731,734 (upto 1 mM) did not inhibit the processing ofa chimeric Ras protein (Ras-CVLL) that isprenylated by a geranylgeranyl moiety (28,29). This result indicates the specificity ofL-731,734 for the FPTase reaction in cells.
To demonstrate that the change in Rasmobility on SDS-PAGE correlated with aninability of Ras to localize to the plasmamembrane, we subjected v-ras cells treatedwith solvent alone or L-731,734 to cellfractionation. Most of the Ras in the sol-vent-treated cells was associated with themembrane, whereas the unprocessed Ras incells treated with L-731,734 was foundexclusively in the soluble fraction (28).Furthermore, we were unable to radiolabelthe soluble Ras with [3Hjmevalonate (30),which serves as a metabolic precursor ofFPP in these cells, suggesting thatL-731,734 blocked at least one step in thebiochemical pathway leading from meval-onate to famesylation of Ras. These resultsindicate that L-731,734 inhibited Ras far-nesylation in vivo.
In contrast to L-731,734, the pattern ofmigration of Ras in cells incubated in thepresence of L-731,735 (250 p.M) closelyresembled that seen with the solvent alone(Fig. 2B). The ability of L-731,734 toefficiently inhibit Ras processing in cellsappears, therefore, to be dependent on thepresence of the lactone, which was incor-porated into the molecule to enhance cellpenetrability. Indeed, using 3H-labeledCAAX tetrapeptide analogs similar toL-731,734 and L-731,735, we found thatthe homoserine lactone form accumulatedin v-ras cells whereas the homoserine formof the compound did not (31). Given themoderate potency of L-731,734 relative tothat of L-731,735 in vitro, it seems likelythat the active form of the inhibitor is
generated in vivo after penetration of thedrug into the cells. This process may in-volve cleavage of the lactone by intracellu-lar esterases to yield L-73 1,735. L-73 1,735is approximately 5000 times more potent atblocking FPTase activity in vitro (IC50 =18 nM) than L-731,734 is at blocking Rasprocessing in cells (IC50 = 100 1xM). Thisdifference may result from inefficient cellpenetration of L-731,734 or incompletehydrolysis of the lactone (or both). Chem-ical or metabolic instability of either formof the compound may also limit activity incell culture.
To determine the biological conse-quences of FPTase inhibition, we assayedthe effect of L-731,734 on the anchorage-independent growth of Ratl cells trans-formed with either a v-ras, v-raf, or v-mosoncogene. Cells transformed by v-Raf andv-Mos were included in the analysis toevaluate the specificity of L-731,734 forRas-induced cell transformation. NeitherRaf nor Mos requires farnesylation toachieve full biological activity. Moreover,Raf and Mos appear to act independently ofRas because dominant negative Ras mu-tants (29, 32) and neutralizing antibodiesagainst Ras (33) do not interfere with Raf-or Mos-induced cell transformation. In thepresence of 1 mM L-731,734, the v-ras-transformed cells did not form the multiple,large colonies that grew in the presence ofsolvent alone (Fig. 3). The minute coloniesobserved for these cells in the presence ofL-731,734 were identical to those seen forthe untransformed parental Ratl cells plat-ed under the same conditions (28). Theinhibition of Ras transformation byL-731,734 was dose-dependent (Fig. 3). Incontrast, 1 mM L-731,734 had no effect onthe anchorage-independent growth of Ratlcells transformed by either v-raf or v-mos(Fig. 3). Thus, the biological effects ofL-731,734 in these experiments appear to
Fig. 3. Inhibition of growth v-ras v-raf v-mosof ras-transformed cells insoft agar by L-731,734.Rati cells transformed MeOHwith either v-ras, v-raf, orv-mos were seeded at a 0 . -density of 1 x 104 cellsper plate (35 mm in diam-eter) in a 0.3% top aga- L-731,734._rose layer in medium A (100 gM)i;(Dulbecco's modified Ea-gle's medium supple-mented with 10% fetal bo-vine serum) over a bottom L-731,734agarose layer (0.6%). (1 mM)Both layers contained0.1% methanol or the indi-cated concentration of L-731,734 (dissolved in methanol at 1000 times the final concentrationused in the assay). The cells were fed twice weekly with 0.5 ml of medium A containing 0.1%methanol or the indicated concentration of L-731,734. Photomicrographs were taken 16 daysafter the cultures were seeded.
be specific for ras-transformed cells. More-over, the resistance of cells transformedwith v-raf and v-mos to the effects ofL-731,734 suggests that the effect seen withthe v-ras-transformed cells is not due togeneral cytotoxicity. Compounds related toL-731,734 that inhibited the anchorage-independent growth of Ras- but not Raf-transformed cells also inhibited the anchor-age-independent growth of cells trans-formed with v-src (31). Although Src doesnot require prenylation to achieve biologi-cal activity, its ability to transform cells hasbeen shown to proceed by a ras-dependentpathway (33). This result is consistent withthe specificity of our compounds for the Raspathway. Furthermore, these same com-pounds caused growth arrest and morpho-logical reversion of Ras- but not Raf-trans-formed cells in monolayer culture (34).
In this report, we have shown that arationally designed, cell-active inhibitor ofFPTase specifically inhibited Ras-depen-dent transformation. This demonstration ofspecificity is important because other fame-sylated cellular proteins, including lamin B(35), which are critical components of nor-mal cell physiology, appear to be substratesfor the same prenylation enzyme that mod-ifies Ras (1 1). The specificity of L-73 1,734for Ras-transformed cells contrasts with thelack of specificity observed with lovastatin.Although lovastatin and L-73 1,734 aresimilar in their ability to block Ras process-ing, lovastatin also inhibits the growth ofcells transformed by v-raf (26). Apparent-ly, lovastatin's growth-inhibitory effectsinvolve multiple mevalonate-dependentpathways that are required for all cellsto grow and divide. The specificity ob-served with L-731,734 likely results frominhibition of a single enzyme that is di-rectly responsible for promoting Ras biolog-ical activity. In this regard, neither L-731,734nor L-731,735 (50 ,uM) had any effect on theconversion of mevalonate into squalene, anintermediate in the cholesterol biosynthesispathway, in an in vitro assay (36).
Studies on the posttranslational modifi-cations of Ras have suggested that pharma-cological inhibition of FPTase would blockras functions. Our findings provide experi-mental evidence that FPTase is an appro-priate biochemical target to inhibit ras-dependent cell transformation. L-731,734is now a prototype for compounds thattarget a specific biochemical reaction thatgoverns cell growth and mitogenesis.
REFERENCES AND NOTES
1. M. Barbacid, Annu. Rev. Biochem. 56, 779(1987).
2. J. L. Bos, Cancer Res. 49, 4682 (1989).3. J. B. Gibbs, Cell 65, 1 (1991); A. D. Cox and C. J.
Der, Crit. Rev. Oncogenesis 3, 365 (1992).
SCIENCE * VOL. 260 * 25 JUNE 19931936
on
Sep
tem
ber
27, 2
012
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om
4. B. M. Willumsen, K. Norris, A. G. Papageorge, N.L. Hubbert, D. R. Lowy, EMBO J. 3, 2581 (1984);J. H. Jackson et al., Proc. Nati. Acad. Sci. U.S.A.87, 3042 (1990).
5. J. F. Hancock, A. I. Magee, J. E. Childs, C. J.Marshall, Cell57, 1167 (1989).
6. S. Powers etal., ibid. 47, 413 (1986); A. Fujiyama,K. Matsumoto, F. Tamanoi, EMBO J. 6, 223(1987).
7. K. Kato et al., Proc. Nati. Acad. Sci. U.S.A. 89,6403 (1992).
8. D. L. Pompliano et al., Biochemistry 31, 3800(1992).
9. J. B. Gibbs etal., J. Biol. Chem. 268, 7617 (1993).10. M. D. Schaber et al., ibid. 265, 14701 (1990).11. Y. Reiss, J. L. Goldstein, M. C. Seabra, P. J.
Casey, M. S. Brown, Cell62, 81 (1990).12. S. L. Moores et al., J. Biol. Chem. 266, 14603
(1991).13. J. L. Goldstein, M. S. Brown, S. J. Stradley, Y.
Reiss, L. M. Gierasch, ibid., p. 15575.14. M. S. Brown, J. L. Goldstein, K. J. Paris, J. P.
Burnier, J. C. Marsters, Jr., Proc. Nati. Acad. Sci.U.S.A. 89, 8313 (1992).
15. H. C. Rilling, E. Breunger, W. W. Epstein, P. F.Crain, Science 247, 318 (1990); C. C. Farnsworth,M. H. Gelb, J. A. Glomset, ibid., p. 320.
16. P. J. Casey, J. A. Thissen, J. F. Moomaw, Proc.
NatI. Acad. Sci. U.S.A. 88,8631 (1991); A. Joly, G.Popjak, P. A. Edwards, J. Biol. Chem. 266,13495(1991); M. C. Seabra, Y. Reiss, P. J. Casey, M. S.Brown, J. L. Goldstein, Cell 65, 429 (1991); K.Yokoyama, G. W. Goodwin, F. Ghomashchi, J. A.Glomset, M. H. Gelb, Proc. NatI. Acad. Sci. U.S.A.88, 5302 (1991); Y. Yoshida et al., Biochem.Biophys. Res. Commun. 175, 720 (1991).
17. M. C. Seabra, J. L. Goldstein, T. C. Sudhof, M. S.Brown, J. Biol. Chem. 267, 14497 (1992); H.Horiuchi et al., ibid. 266, 16981 (1991).
18. Y. Reiss, S. J. Stradley, L. M. Gierasch, M. S.Brown, J. L. Goldstein, Proc. Nati. Acad. Sci.U.S.A. 88, 732 (1991).
19. D. Heimbrook and J. B. Gibbs, unpublished data.20. D. J. Capon, E. Y. Chen, A. D. Levinson, P. H.
Seeburg, D. V. Goeddel, Nature 302, 33 (1983).21. R. Dhar, A. Nieto, R. Koller, D. DeFeo-Jones, E. M.
Scolnick, Nucleic Acids Res. 12, 3611 (1984); S.Powers et al., Cell 36, 607 (1984).
22. S. D. Mosser, J. B. Gibbs, C. A. Omer, unpub-lished data.
23. M. E. Furth, L. J. Davis, B. Fleurdelys, E. M.Scolnick, J. Virol. 43, 294 (1982).
24. T. Y. Shih, P. E. Stokes, G. W. Smythers, R. Dhar,S. Oroszlan, J. Biol. Chem. 257,11767 (1982).
25. J. B. Gibbs, R. W. Ellis, E. M. Scolnick, Proc. Nati.Acad. Sci. U.S.A. 81, 2674 (1984).
26. J. E. DeClue, W. C. Vass, A. G. Papageorge, D. R.Lowy, B. M. Willumsen, Cancer Res. 51, 712(1991).
27. M. Sinensky, L. A. Beck, S. Leonard, R. Evans, J.Biol. Chem. 265,19937 (1990).
28. N. E. Kohl, unpublished data.29. A. D. Cox, M. M. Hisaka, J. E. Buss, C. J. Der, Mol.
Cell. Biol. 12, 2606 (1992).30. N. E. Kohl and J. B. Gibbs, unpublished data.31. F. R. Wilson and N. E. Kohl, unpublished data.32. L. A. Feig and G. M. Cooper, Mol. Cell. Biol. 8,
3235 (1988); D. W. Stacey et al., Oncogene 6,2297 (1991).
33. M. R. Smith, S. J. DeGudicibus, D. W. Stacey,Nature 320, 540 (1986).
34. G. C. Prendergast and N. E. Kohl, unpublisheddata.
35. C. C. Farnsworth, S. L. Wolda, M. H. Gelb, J. A.Glomset, J. Biol. Chem. 264, 20422 (1989).
36. J. Karkas, J. Bergstrom, J. B. Gibbs, unpublisheddata.
37. We thank C. Omer and G. Prendergast for criti-cally reading the manuscript; R. Bock, S. Moores,E. Rands, T. Thomas, and F. Wilson for technicalassistance; D. Lowy for the vHa-ras plasmid; andG. Vande-Woude for the v-mos plasmid.
30 March 1993; accepted 24 May 1993
Benzodiazepine Peptidomimetics: Potent Inhibitorsof Ras Farnesylation in Animal Cells
Guy L. James, Joseph L. Goldstein, Michael S. Brown,Thomas E. Rawson, Todd C. Somers, Robert S. McDowell,
Craig W. Crowley, Brian K. Lucas, Arthur D. Levinson,James C. Marsters, Jr.
Oncogenic Ras proteins transform animal cells to a malignant phenotype only whenmodified by farnesyl residues attached to cysteines near their carboxyl termini. The far-nesyltransferase that catalyzes this reaction recognizes tetrapeptides of the sequenceCAAX, where C is cysteine, A is an aliphatic amino acid, and X is a carboxyl-terminalmethionine or serine. Replacement of the two aliphatic residues with a benzodiazepine-based mimic ofa peptide turn generated potent inhibitors offarnesyltransferase [50 percentinhibitory concentration (1C50) < 1 nM]. Unlike tetrapeptides, the benzodiazepine pepti-domimetics enter cells and block attachment of farnesyl to Ras, nuclear lamins, and severalother proteins. At micromolar concentrations, these inhibitors restored a normal growthpattern to Ras-transformed cells. The benzodiazepine peptidomimetics may be useful inthe design of treatments for tumors in which oncogenic Ras proteins contribute to abnor-mal growth, such as that of the colon, lung, and pancreas.
Oncogenic Ras proteins are causally impli-cated in certain human malignancies (1).Poised at the inner surface of the plasmamembrane, Ras proteins normally respondto growth stimuli like epidermal and plate-let-derived growth factors by exchangingguanosine triphosphate (GTP) for constitu-tively bound guanosine diphosphate(GDP), thereby triggering cell division.The signal is terminated when the Ras
G. L. James, J. L. Goldstein, M. S. Brown, Departmentof Molecular Genetics, University of Texas Southwest-ern Medical Center, Dallas, TX 75235.T. E. Rawson, T. C. Somers, R. S. McDowell, C. W.Crowley, B. K. Lucas, A. D. Levinson, J. C. Marsters,Jr., Departments of Bioorganic Chemistry and CellGenetics, Genentech Inc., South San Francisco, CA94080.
protein hydrolyzes its bound GTP to GDPin a reaction that is stimulated by a guano-
sine triphosphatase (GTPase) activatingprotein (GAP) (2). About 50% of humancolon carcinomas and 90% of pancreaticcarcinomas produce mutant Ras proteinsthat bind GTP but cannot hydrolyze it (1).The mutant proteins are constitutively ac-
tive, and this constant signal, coupled withother regulatory abnormalities, leads to ma-
lignant transformation.The function of normal and oncogenic
Ras proteins is absolutely dependent on theposttranslational attachment of a 15-carbonisoprenoid moiety, famesyl, through a
thioether linkage to a cysteine near theCOOH-terminus of the protein (3). This
SCIENCE * VOL. 260 * 25 JUNE 1993
modification is catalyzed by a heterodimericZn2+-dependent enzyme designated CAAXfamesyltransferase (4, 5). The enzyme usesfamesyl pyrophosphate as a donor and at-taches a famesyl group to the cysteine resi-due at the fourth position from the COOH-terminus of various proteins, including allfour Ras proteins, nuclear lamins A and B,skeletal muscle phosphorylase kinase, andthree retinal proteins (the y subunit of trans-ducin, the a subunit of cyclic guanosinemonophosphate phosphodiesterase, and rho-dopsin kinase) (6). The COOH-termini of allof these substrates share the tetrapeptide se-quence CAAX, where C is cysteine, A standsfor aliphatic residues, and X is methionine orserine (7). This CAAX motif appears to bethe sole recognition site for the enzyme;hence, addition of CAAX sequences to theCOOH-termini of other proteins rendersthem substrates for farnesylation (8). More-over, in vitro the enzyme attaches a famesylgroup to tetrapeptides that conform to theCAAX consensus (7, 9).
In the cell, farnesylation is the first stepin a sequence of modifications that rendersthe COOH-terminus of the Ras proteinhydrophobic. Farnesylation is followed byproteolytic removal of the terminal threeamino acids and methylation of the freeCOOH group on the farnesylated cysteine(3). These reactions are necessary for Ras tobecome attached to the inner surface of theplasma membrane.
Oncogenic Ras proteins lose their trans-forming ability when farnesylation is pre-vented, either by mutation of the CAAXsequence or by blocking synthesis of famesylpyrophosphate with an inhibitor of 3-hy-droxy-3-methylglutaryl coenzyme A reduc-tase (3, 10). Under some circumstances thecytosolic nonprenylated form of Ras may
1937
11 1: l,l,1,11
on
Sep
tem
ber
27, 2
012
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om
