INHIBITION OF HUMAN TUMOR CELL GROWTH IN VITROAND IN VIVOBY A SPECIFIC INHIBITOR OF HUMAN FARNESYLTRANSFERASE: BIM-46068Gregoire P. PREVOST1*, Anne PRADINES2, Isabelle VIOSSAT1, Marie-Christine BREZAK1, Karine MIQUEL2, Marie-Odile LONCHAMPT1,Philip KASPRZYK3, Gilles FAVRE2, Bernadette PIGNOL1, Christine LE BRETON4, Jesse DONG3 and Barry MORGAN3

1Institut Henri Beaufour, Les Ulis, France2Institut Claudius Regaud, Toulouse, France3Biomeasure, Milford, MA, USA4Expansia, Aramon, France

Oncogenic mutations of the ras gene leading to constitutiveactivation of downstream effectors have been detected in alarge spectrum of human cancers (pancreas, thyroid, colonand NSCLC). Membrane anchorage of Ras required forfunctional activity in signal transduction is facilitated bypost-translational modifications resulting in covalent attach-ment of a farnesyl group to the cysteine in the C-terminalCAAX motif. This attachment is mediated by farnesyltrans-ferase (FTase). Here, we report a novel series of potentFTase inhibitors, where the tetrapeptide CAAX motif hasbeen modified by incorporation of a thiazolidine carboxylicacid moiety followed by reduction of the 1st and 2nd peptidebonds to a secondary and tertiary amine, respectively. TheC-terminal carboxylate was converted to esters for improvedcellular penetration. These compounds showed specific inhi-bition of purified human FTase enzyme, inhibition of prolifera-tion in vitro in a large spectrum of human tumor cell lines andinhibition of growth of human tumor xenografts in athymicnude mice. In addition, in regard to a panel of cell lines, usingthe Compare analysis to determine the Pearson coefficientcorrelation, the anti-proliferative spectrum of BIM-46068 hasbeen shown to be distinct from the profile of typical chemo-therapeutic agents. Int. J. Cancer 83:283–287, 1999.r 1999 Wiley-Liss, Inc.

The mutated species of Ras protein exist in an active GTP-boundstate without any requirement for external growth factor stimuli totransmit a proliferation signal. Oncogenicras has been detected ina large spectrum of human cancers, including pancreatic tumors(90%), thyroid carcinoma (50%) and colon tumors (50%). Mem-brane association of Ras protein is required for its biologicalfunctions. The critical step of Ras post-translational modificationsis mediated by the cytosolic enzyme farnesyltransferase (FTase),allowing farnesylation on the sulfur of the CAAX motif cysteine ofRas protein. Interest in inhibiting FTase in tumor cells has beenreinforced by 2 facts: FTase activity is higher in carcinomascompared with normal tissues (Khanet al., 1996) and over-expression of FTase amplifies growth factor–mediated cell growthand tumorigenicity (Nagaseet al., 1999). Farnesylation can beblocked by suppression of the prenyl poolvia inhibition ofHMG-CoA reductase, the key enzyme of isoprenoid biosynthesis(Xu et al.,1996). However, this approach is limited due to the largenumber of geranylgeranylated proteins that are altered.

To obtain a selective blockade of farnesylation, several potentinhibitors of FTase (FTIs) have been described, including naturalproducts and synthetic compounds (Sebti and Hamilton, 1997).Such FTIs are potent inhibitors of tumor cell growthin vitro as wellasin vivo (Liu et al.,1998). Although inhibition of the prenylationof H- and N-Ras has been noted with FTIs in tumor cells, blockadeof K-Ras processing appears to be resistant to these compounds(Jameset al., 1996). The Ftase-related enzyme geranylgeranyltransferase I (GGTase) can add a geranylgeranyl moiety to K-Raswhile preserving its membrane anchorage. Geranylgeranylation ofK-Ras has also been observed in whole cells in the presence ofFTIs. Both FTase and GGTase I inhibitors are required forinhibiting oncogenic K-Ras prenylation, but each alone is sufficientto suppress human tumor growth in nude mouse xenografts (Sunet

al., 1998). This observation suggests the blockade of additionalsubstrates of FTase, such as RhoB (Duet al.,1999).

We describe here a novel CAAX-based pseudopeptide, BIM-46068,which induces specific inhibition of purified FTase enzyme, growthinhibition of human tumor cellsin vitro and growth inhibition of amutated K-rashuman tumor xenografted into nude mice.

MATERIAL AND METHODS

MaterialBIM-46068 and BIM-46050 were synthesized at Biomeasure

(Milford, MA). The synthetic FTI B-581 was purchased fromBachem (N1390; Voisin le Bretonneux, France). Expansia (Ar-amon, France) provided FTI-277 and GGTI-286. The chemothera-peutic drugs doxorubicin, etoposide and daunorubicin were ob-tained from Sigma (Saint-Quentin Fallavier, France). SN38 waskindly provided by Dr. O. Lavergne (IHB, Les Ulis, France) andCis-platyl by Laboratoire Roger Bellon (Neuilly sur Seine, France).All compounds were stored and used according to the manufactur-ers’ recommendations.

Prenyltransferase assaysFTase activity was assayed by [3H]-farnesylation of recombinant

human H-Ras protein (Biomol, Plymouth Meeting, PA), using amicroplate method. The incubation mixture (25µl) contained 50mM Tris-HCl (pH 7.5), 5 mM dithiothreitol, 20µM ZnCl2, 40 mMMgCl2, 0.6µM [3H]-farnesyl pyrophosphate (22.3 Ci /mmol)(NEN, Boston, MA), 4µM H-Ras and 10µg FTase from humanbrain cytosol (ABS Reagents, Wilmington, DE). Inhibitors wereadded in sufficient amounts of solvent and incubations started byadding FTase. After 60 min at 37°C, the reaction was stopped byadding 100µl of 10% HCl in ethanol and incubation run for 15 minat 37°C; 150µl of absolute ethanol were added, and the mixture wasfiltered on a Unifilter GF/B microplate (Packard, Rungis, France)and washed 6 times with ethanol. After adding 50µl of Microscint0, plates were counted with a Packard Top Count scintillationcounter. GGTase I activity was assayed by a similar method butsubstituting 4µM human recombinant H-Ras CVLL type (Biomol),0.6µM [3H]-geranylgeranyl-pyrophosphate (19.3 Ci /mmol) (NEN)and 100µg GGTase I from human brain (ABS Reagents).

Cell linesAll human cell lines except T24R were purchased from the

ATCC (Rockville, MA) and cultivated in DMEM with 10% heatedFCS, glutamine and penicillin/streptomycin at 37°C in a 5% CO2atmosphere. Dr. R. Kiss (University of Brussels, Belgium) kindlyprovided T24R cells. All products for culture were purchased fromSigma.

*Correspondence to: Institut Henri Beaufour, 5 Avenue du Canada,Les Ulis, F-91966, France. Fax:133–1-69 07 38 02.E-mail: gregoire.prevost@beaufour-ipsen.com

Received 9 April 1999; Revised 1 June 1999

Int. J. Cancer:83,283–287 (1999)

r 1999 Wiley-Liss, Inc.

Publication of the International Union Against CancerPublication de l’Union Internationale Contre le Cancer

In vitro cell proliferation assaysCells were seeded in 96-well plates at day 0 in DMEM

supplemented with 10% heated FCS, 50,000 units/l penicillin and50 mg/l streptomycin. At day 1, cells were treated for 96 hr withincreasing concentrations of drugs (up to 50µM for prenyltransfer-ase inhibitors and up to 10µM for chemotherapeutic agents). At theend of this period, cell proliferation was evaluated by a colorimet-ric assay based on cleavage of the tetrazolium salt WST1 bymitochondria dehydrogenases in viable cells, leading to formazanformation (Boehringer-Mannheim, Mannheim, Germany). Thenumber of seeded cells for each line was previously determined,allowing the control cells to be in growth phase at the end of theexperiment. These experiments were performed twice with 8determinations per tested concentration. For each compound,values falling in the linear part of the sigmoid curve were includedin a linear regression analysis and used to estimate the 50%inhibitory concentration (IC50).

Analysis with the COMPARE computer algorithmCompounds with a high correlation between their screening data

generally have a similar mechanism of action. Growth inhibitionwas measured in a 96-microwell plate using the colorimetric assayWST1 described above. The growth fraction relative to untreatedcells was measured 96 hr after introduction of the tested com-pounds at 10 concentrations. The COMPARE computer algorithm(Freijeet al.,1997) was used to reduce each experiment to an arrayof x IC50 values (x corresponds to the set of cell lines) in a pattern ofactivity for each compound. The Pearson correlation coefficientbetween 2 compounds measures the degree of similarity betweenthe activity profiles of these compounds within the set of cell linesused.

Western blot analysisWestern blot analysis of Ras and Rap1A was performed as

previously described (Miquelet al., 1997). Briefly, cell lysateswere separated onto SDS-polyacrylamide gels (15%) and trans-ferred to nitrocellulose membranes. Nitrocellulose filters wereprobed with either anti-H-Ras (Santa Cruz Biotechnology, SantaCruz, CA) or anti-Rap1A (Santa Cruz Biotechnology) antibodies,then with horseradish peroxidase–conjugated antibodies (Bio-Rad,Ivry sur Seine, France). Detection was performed with the ECLchemiluminescence system (Amersham, Les Ulis, France).

In vivo growth of human tumors xenografted in nude miceCells of the human pancreatic carcinoma line MiaPaCa-2 (53

106) were injected s.c. into the flank of female athymic NCr-nu/numice, 4 to 6 weeks old. Tumors were allowed to reach a volume of100 mm3. Once tumors were established, treatment was started,once daily for a total of 10 days. Tumor measurements and animalweights were monitored twice a week and animals followed for atleast 2 weeks after treatment had been completed. Any sign oftoxicity was monitored and recorded daily. Animal care was inaccordance with institutional guidelines.

RESULTS

Chemical structure of BIM-46068

Figure 1 shows the structure of a novel series of peptidomimeticFTIs. BIM-46050 and BIM-46068 possess a cyclic amino acid5,5-dimethylthiazolidine-4-carboxylic acid, which has been incor-porated at the A2 residue of the CA1A2X motif. In addition, the 1stand 2nd peptide bonds have been reduced to a secondary andtertiary amine, respectively. BIM-46068 is the methyl ester ofBIM-46050, and this transformation has been shown, in otherseries, to improve potency in cellular assays. Their chemicalsynthesis is extensively described in the patent application W098/00409 (submitted in 1996 by Dr. J. Dong, Boston, MA).

Specific inhibition of purified human FTase by BIM-46068The IC50 values for in vitro inhibition of human brain FTase

indicate that BIM-46050 and the ester form BIM-46068 are potent

inhibitors of the enzyme (Table I). Their potencies are in thenanomolar range and compare favorably with the compoundsB581, FTI-277 and L745,631. B581 is an analog of the tetrapeptideCys-Val-Phe-Met obtained by replacement of the amino-terminalamide bonds inhibiting processing of farnesylated proteins specifi-cally. FTI-277 is a methyl ester of FTI-276, reported as apreferential inhibitor of FTase over GGTase I (100-fold). L745,631is a 2-substituted piperazine reported to be a highly selectiveinhibitor of FTase over GGTase (2,000-fold). The selectivity ofBIM-46050 and BIM-46068 for FTase over GGTase is very similarfor both compounds (3,000-fold).

Specific inhibition of Ras processing in MiaPaCa-2 cancer cellsby BIM-46068

Inhibition of endogenous H-Ras protein farnesylation in wholecells treated by FTI produces a shift in Ras electrophoretic mobilityin SDS-PAGE. To investigate this parameter, MiaPaCa-2 cells weretreated in culture with increasing concentrations of BIM-46068 orFTI-277 for 48 hr. Cell proteins were then analyzed by Westernblotting and probed with an antibody that recognizes H-Ras. Figure2 (upper panel) shows a mobility shift of Ras protein from a fastmigrating form in the control (lane Cont) to a slow migrating formin cells treated by inhibitors (BIM-46068 and FTI-277), correspond-ing to the unprenylated form and to the farnesylated form of H-Ras,respectively. This result is similar to that observed with an inhibitorof HMG-CoA reductase, mevastatin (lane MEV). Similar resultswere obtained in another cell line, A549 (data not shown).BIM-46068 did not induce any modification in the migration ofRap1A, a geranylgeranylated protein, whereas mevastatin treat-ment induced a shift of the Rap1A band (Fig. 2, lower panel). Also,increasing concentrations of BIM-46068 to 30µM in cell line A549did not modify Rap1A (data not shown). These results demonstratethe selectivity of BIM-46068 for FTase over GGTase I.

In vitro growth inhibition of human tumor cell lines byBIM-46068

The anti-proliferative activity of BIM-46068 testedin vitro on apanel of human tumor cell lines is shown in Table II. In this set of

FIGURE 1 – Chemical structure of FTIs derived from the thiazolidinecarboxylic structure. BIM-46050 is the acidic form, with R5 H, andBIM-46068 is an ester form, with R5 CH3.

TABLE I – INHIBITORY EFFECTS OF FTIs ON PURIFIED HUMANPRENYLTRANSFERASE ACTIVITY

Enzyme assay IC50 (nM)1

FTase GGTase I

BIM-46050 13.3 (8.3–18.4) 50700 (34,200–67,100)BIM-46068 91.4 (83.4–99.4) 268000 (15,200–384,000)FTI-277 30.4 (26.2–34.5) 314.2 (305–323)GGTI-286 365.5 (329–401) 114.8 (106–123)B-581 25.8 (20.1–28.3) nd

1Values of IC50 represent the concentration required to inhibit 50% ofthe enzyme-catalyzed incorporation of labeled prenyl into recombinanthuman H-Ras with the wild-type CAAX box and the modified CAAXbox (CVLL). Assay results are reported as the mean of 2 experimentswith 95% CI in parentheses. nd, not determined.

284 PREVOSTET AL.

cell lines, BIM-46068 inhibited growth of all lines with IC50 valuesranging from 6 to 37µM. BIM-46068 inhibited the growth of cellsbearing the wild-typerasas well as cells with the different mutatedras (K-ras, N-ras or H-ras). In addition, BIM-46068 was aneffective inhibitor of the multidrug-resistant cell line T24R. TheIC50 ratio between the drug-resistant T24R cells and the drug-sensitive T24 cells was only 3- to 4-fold with BIM-46068compared to 9- to 23-fold with drugs such as doxorubicin,etoposide, SN38 and daunorubicin (data not shown). L745,631 wasa poor growth inhibitor of these cell lines, and both BIM-46050 andB-581 were inactive (data not shown).

Different spectrum of human tumor cell lines sensitive toBIM-46068 and to other chemotherapeutic agents

The differential activity of various drugs against several celllines can be analyzed by determining the Pearson correlationcoefficient using the COMPARE program. If this coefficientbetween 2 compounds is far from 1, those compounds may beconsidered as having distinct spectra of activity. However, nocut-off value can be calculated using statistical analysis to assumewhen the spectra are really distinct. For such a comparison, wehave generated a matrix with the IC50 values of prenyltransferaseinhibitors and chemotherapeutic agents obtained with a series ofcell lines. These values were derived from 2 independent experi-ments performed in octoplicate. Table III shows that the Pearsoncorrelation coefficients of BIM-46068 compared with each chemo-therapeutic agent were far from 1. This suggests that the cell linessensitive to BIM-46068 are distinct from those sensitive to theclinically used agents included in the study. Figure 3 shows anexample of such data obtained with 2 compounds (BIM-46068 anddoxorubicin) on 2 cell lines (DU145 and HL60). A higherconcentration of doxorubicin is required to reach IC50 values inHL60 than in DU145 cells (6- to 7-fold), whereas lower concentra-

tions of BIM-46068 are required to reach IC50 values in HL60 thanin DU145 cells (5- to 6-fold). The Pearson correlation coefficientbetween BIM-46068 (a selective FTI) and FTI-277 (bearing bothFTase and GGTase inhibitory activities) was not equal to 1, thedifference possibly reflecting the different enzyme selectivities.

In vivo growth inhibition of MiaPaCa-2 cells in athymic miceNude mice were implanted s.c. with MiaPaCa-2 cells (53 106),

a human pancreatic tumor cell line expressing mutated K-ras. Drugtreatment was initiated when the tumors reached a volume of 130mm

3and continued once a day, i.p., for 10 days. With this schedule,

BIM-46068 was shown to reduce tumor growth in a dose-dependent manner (Fig. 4). At day 34 post-implantation, the meantumor volume of the group treated with 30 mg/kg was smaller by50% than the mean tumor volume in control animals. At the highesttested dose (30 mg/kg i.p.), no toxicity (weight loss, death) wasnoted during the course of the experiment.

DISCUSSION

We show here that the novel compound BIM-46068 is a potentand specific inhibitor of purified human FTase. Its application incell culture inhibits growth of cells derived from a variety of humantumor types. Moreover, BIM-46068 has the ability to inhibit thefarnesylation of H-Ras in tumor cells, without any effect on theprenylation of Rap1A, a geranylgeranylated protein. Furthermore,in a model of xenografted tumors, BIM-46068 significantlyreduced the growth rate of a human pancreatic MiaPaCa-2 tumorbearing a K-Ras mutation, without any signs of toxicity.

No real difference was noted between the efficacyin vitro ofBIM-46068 on cell lines bearing wild-type or mutated ras (N-ras,H-ras and K-ras). This is in accordance with the data of Sepp-Lorenzinoet al. (1995) using L744,832, a pure FTI, but not withthe observations of Nagasuet al. (1995), who used B-956, which

FIGURE 2 – Specific inhibition of H-Ras processing in MiaPaca-2cells by BIM-46068. Total proteins of cells treated for 48 hr withincreasing concentrations of compounds, as indicated in the figure,were separated by SDS-PAGE, blotted on nitrocellulose membranesand immunodetected by an antibody directed against H-Ras (upperpanel) or an antibody directed against Rap1A (lower panel). cont,control; MEV, mevastatin.

TABLE II – INHIBITORY EFFECTS OF BIM-46068 ONIN VITROGROWTHOF TUMOR CELL LINES EXPRESSING DIFFERENTrasGENES

Cell lines TissuesCell growth assay IC50 (µM)1

ras status BIM-46068

HT29 Colon Wild-typeras 8.8 (8.6–9.1)MCF7 Breast Wild-typeras 19.4 (16.6–24.3)U87MG Glioma Wild-typeras 26 (24–27.9)HS766 Pancreas Wild-typeras 19 (18–20.1)A427 Lung Mutated Ki-ras 6.0 (4.5–8.8)MiaPaCa-2 Pancreas Mutated Ki-ras 8.5 (6.2–10.5)CFPAC Pancreas Mutated Ki-ras 22.6 (19.1–33.2)HL60 Leukemia Mutated N-ras 6.6 (5.0–11.8)T24 Bladder Mutated H-ras 10.7 (8.5–14.4)T24R Bladder Mutated H-ras 37 (30.9–40.1)A127 Glioma Unknown 16 (13.4–18.3)

1Results are reported as the mean of 2 ormore experiments performedin octoplicate with the lowest and highest IC50 values observed inindividual experiments.

TABLE III – COMPARISON OFIN VITROGROWTH INHIBITION BY BIM-46068WITH OTHER COMPOUNDS

Drugs Anti-proliferative activitiesPearson correlation

coefficient withBIM-460681

BIM-46068 FTI 1FTI-277 FTI 0.527SN38 Topoisomerase I inhibitor 0.162Etoposide Topoisomerase II inhibitor 0.155Cis-platyl Alkylator 0.003Doxorubicin Topoisomerase II inhibitor 20.038Daunorubicin Topoisomerase II inhibitor 20.150

1The Pearson correlation coefficient between BIM-46068 and otheragents was determined with the COMPARE computer algorithm usingvarious cell lines (DU145, MiaPaCa-2, A427, HL60, T24, HT29 andMCF7).

FIGURE 3 – Sensitivity to BIM-46068 or doxorubicin (Adriamycin)differs in 2 cell lines. Comparison of IC50 values showed that DU145 ismore sensitive than HL60 to doxorubicin, whereas HL60 is moresensitive than DU145 to BIM-46068.

285ANTI-TUMORAL ACTIVITIES OF FTI BIM-46068

showed greater potency against cells bearing H-ras. Several reports(Whyteet al.,1997) illustrating the resistance of K-Ras to FTIs andthe ability of K-Ras to be geranylgeranylated suggest that com-pounds combining the ability to block both GGTase I and FTaseshould be required for inhibiting tumor cell growth. However, wehave shown that a highly selective FTI inhibits the growth of cellswith a K-RAS mutationin vitro as well asin vivo. Indeed, Lerneretal. (1997) reported that inhibition of mutated K-Ras processing wasnot required to inhibit cell proliferationin vitro and in vivo. Thisobservation supports the strategic option for developing specificFTIs for the control of human tumors.

Total inhibition of H-Ras processing by BIM-46068 was achievedin MiaPaCa-2 as well as other cell lines (data not shown) atconcentrations as low as 100 nM. This is in accordance with therange of concentrations used to obtain complete inhibition ofpurified FTase (Table I). However, despite such potent effects onH-Ras processing at nanomolar concentrations, the concentrationsrequired to observe anti-proliferative effects on cells were in themicromolar range. Such a discrepancy between the loss of Rasprotein activity and the reduction of cell growth in human cells

suggests that functional Ras protein is not the only parameterinvolved in cell proliferation control. Other prenylated proteins,such as Rho family members or the oncogenePTP-CAAX(Catesetal., 1996), have been proposed as critical targets of FTIs butwithout consensus being reached. Experiments using other agentswhich specifically block theras oncogene (Jansenet al., 1995),such as anti-ras ribozymes delivered by retroviral gene transfer (Liet al., 1996), H-ras or K-ras anti-sense nucleotides (Kitaet al.,1999), however, exhibit successful tumor inhibition.

Manumycin, a natural FTI, exhibits toxicity at a concentrationeffective for pancreatic tumor growth inhibition (Itoet al., 1996).Here, using BIM-46068, no toxicity in nude mice was noticed atthe dose leading to pancreatic tumor inhibition. Although the rateof tumor growth was reduced, no tumor regression was observed.At present, none of the FTIs has induced tumor regression inanimal models xenografted with human tumors. However, tumorregression has been observed withras transgenic models. Prophy-lactic treatment of N-ras transgenic animals completely preventstumor formation. Apoptosis induction by FTIs might explain thetumor regression, but the mechanism of such induction remainsunclear because some FTIs are not inducers of apoptosis insensitive human tumor cells. FTI-277 induces apoptosis in the lungcancer K-ras mutated A427 but not in other K-ras mutated celllines (MiaPaca-2) or the lung cancer line A549 (data not shown).As well, BIM-46068 induces apoptosis in A427, but not in A549,cells (data not shown). The molecular mechanisms leading to theapoptosis must be investigated in detail to identify the tumorphenotypes allowing apoptosis induction by FTI.

A major factor involved in chemoresistance in cancer treatmentis attributed to the expression of multidrug-resistant proteins (MRP,LRP and MDR). In this report, we have observed an anti-proliferative effect of BIM-46068 on cells expressing differentmultidrug-resistant proteins and activated Ras (T24R, A549 andHL60). In addition, Ras activation has been associated with theinduction of chemoresistance and radioresistance (Bernhardet al.,1996). These observations favor the use of BIM-46068 as anadjuvant treatment for tumors with such forms of chemo- orradioresistance.

Our results with BIM-46068 support the concept that this FTI,alone or in combination with other cancer therapies, might beeffective against human tumor growth.

ACKNOWLEDGEMENTS

We thank Expansia for the synthesis of FTI-277 and GGTI-286.We thank Drs. F. Thomas and P. de Cremoux for helpful discus-sions. We thank Mr. J. Marsais and Ms. J. Schulz for assistance.

REFERENCES

BERNHARD, E.J., KAO, G., COX, A.D., SEBTI, S.M., HAMILTON , A.D.,MUSCHEL, R.J. and MCKENNA, W.G., The farnesyltransferase inhibitorFTI-277 radiosensitizes H-ras-transformed rat embryo fibroblasts.CancerRes.,56,1727–1730 (1996).

CATES, C.A., MICHAEL, R.L., STAYROOK, K.R., HARVEY, K.A., BURKE, Y.D.,RANDALL , S.K., CROWELL, P.L. and CROWELL, D.N., Prenylation ofoncogenic human PTP(CAAX) protein tyrosine phosphatases.CancerLett.,110,49–55 (1996).

DU, W., LEBOWITZ, P.F. and PRENDERGAST, G.C., Cell growth inhibition byfarnesyltransferase inhibitors is mediated by gain of geranylgeranylatedRhoB.Mol. cell. Biol.,19, 1831–1840 (1999).

FREIJE, J.M., LAWRENCE, J.A., HOLLINGSHEAD, M.G., DE LA ROSA, A.,NARAYANAN , V., GREVER, M., SAUSVILLE, E.A., PAULL , K. and STEEG, P.S.,Identification of compounds with preferential inhibitory activity againstlow-Nm23-expressing human breast carcinoma and melanoma cell lines.Nature (Med.),3, 395–401 (1997).

ITO, T., KAWATA , S., TAMURA, S., IGURA, T., NAGASE, T., MIYAGAWA , J.I.,YAMAZAKI , E., ISHIGURO, H. and MATASUZAWA, Y., Suppression of humanpancreatic cancer growth in BALB/c nude mice by manumycin, a farnesyl:protein transferase inhibitor.Jpn. J. Cancer Res.,87,113–116 (1996).

JAMES, G., GOLDSTEIN, J.L. and BROWN, M.S., Resistance of K-RasBV12proteins to farnesyltransferase inhibitors in Rat1 cells.Proc. nat. Acad. Sci.(Wash.),93,4454–4458 (1996).

JANSEN, B., WADL, H., INOUE, S.A., TRULZSCH, B., SELZER, E., DUCHENE, M.,EICHLER, H.G., WOLFF, K. and PEHAMBERGER, H., Phosphorothioate oligo-nucleotides reduce melanoma growth in a SCID-hu mouse model by anonantisense mechanism.Antisense Res. Dev.,5, 271–277 (1995).

KHAN, S.G., DUMMER, R., SIDDIQUI, J., BICKERS, D.R., AGARWAL, R. andMUKHTAR, H., Farnesyltransferase activity and mRNA expression in humanskin basal cell carcinomas.Biochem. biophys. Res. Comm.,220,795–801(1996).

KITA, K., SAITO, S., MORIOKA, C.Y. and WATANABE, A., Growth inhibitionof human pancreatic cancer cell lines by anti-sense oligonucleotidesspecific to mutated K-rasgenes.Int. J. Cancer,80,553–558 (1999).

LERNER, E.C., ZHANG, T.T., KNOWLES, D.B., QIAN, Y., HAMILTON , A.D. andSEBTI, S.M., Inhibition of the prenylation of K-Ras, but not H- or N-Ras, ishighly resistant to CAAX peptidomimetics and requires both a farnesyltrans-ferase and a geranylgeranyltransferase I inhibitor in human tumor cell lines.Oncogene,15,1283–1288 (1997).

FIGURE 4 – Growth inhibition of human pancreatic tumor Mia-Paca-2 cells in nude mice by BIM-46068. Cells fromin vitro culturewere injected s.c. at day 0, and when the tumor volume reached 100mm3 (day 11), treatment was started and applied i.p. for 10 days.

286 PREVOSTET AL.

LI, M., LONIAL, H., CITARELLA , R., LINDH, D., COLINA, L. and KRAMER, R.,Tumor inhibitory activity of anti-ras ribozymes delivered by retroviral genetransfer.Cancer Gene Ther.,3, 221–229 (1996).LIU, M. and 24 OTHERS, Antitumor activity of SCH 66336, an orallybioavailable tricyclic inhibitor of farnesyl protein transferase, in humantumor xenograft models and wap-ras transgenic mice.Cancer Res.,58,4947–4956 (1998).MIQUEL, K., PRADINES, A., SUN, J., QIAN, Y., HAMILTON , A.D., SEBTI, S.M.and FAVRE, G., GGTI-298 induces G0-G1 block and apoptosis whereasFTI-277 causes G2-M enrichment in A549 cells.Cancer Res.,57, 1846–1850 (1997).NAGASE, T., KAWATA , S., NAKAJIMA , H., TAMURA, S., YAMASAKI , E., FUKUI,K., YAMAMOTO, K., MIYAGAWA , J., MATSUMURA, I., MATSUDA, Y. andMATSUZAWA, Y., Effect of farnesyltransferase overexpression on cellgrowth and transformation.Int. J. Cancer,80,126–133 (1999).NAGASU, T., YOSHIMATSU, K., ROWELL, C., LEWIS, M.D. and GARCIA, A.M.,Inhibition of human tumor xenograft growth by treatment with the farnesyltransferase inhibitor B956.Cancer Res.,55,5310–5314 (1995).

SEBTI, S. and HAMILTON , A.D., Inhibitors of prenyl transferases.Curr. Opin.Oncol.,9, 557–561 (1997).SEPP-LORENZINO, L., MA, Z., RANDS, E., KOHL, N.E., GIBBS, J.B., OLIFF, A.and ROSEN, N., A peptidomimetic inhibitor of farnesyl:protein transferaseblocks the anchorage-dependent and -independent growth of human tumorcell lines.Cancer Res.,55,5302–5309 (1995).SUN, J., QIAN, Y., HAMILTON , A.D. and SEBTI, S.M., Both farnesyltransfer-ase and geranylgeranyltransferase I inhibitors are required for inhibition ofoncogenic K-Ras prenylation but each alone is sufficient to suppress humantumor growth in nude mouse xenografts.Oncogene,16,1467–1473 (1998).WHYTE, D.B., KIRSCHMEIER, P., HOCKENBERRY, T.N., NUNEZ-OLIVA , I.,JAMES, L., CATINO, J.J., BISHOP, W.R. and PAI, J.K., K- and N-Ras aregeranylgeranylated in cells treated with farnesyl protein transferase inhibi-tors.J. biol. Chem.,272,14459–14464 (1997).XU, X.Q., MCGUIRE, T.F., BLASKOVICH, M.A., SEBTI, S.M. and ROMERO, G.,Lovastatin inhibits the stimulation of mitogen-activated protein kinase byinsulin in HIRcB fibroblasts.Arch. Biochem. Biophys.,326, 233–237(1996).

287ANTI-TUMORAL ACTIVITIES OF FTI BIM-46068