Characterization of Ha-Ras, N-Ras, Ki-Ras4A, and Ki-Ras4B as in Vitro Substrates for Farnesyl Protein Transferase and Geranylgeranyl Protein Transferase Type I*

Ras proteins are small GTP-binding proteins which are critical for cell signaling and proliferation. Four Ras isoforms exist: Ha-Ras, N-Ras, Ki-Ras4A, and Ki-Ras4B. The carboxyl termini of all four isoforms are post-trans-lationally modified by farnesyl protein transferase (FPT). Prenylation is required for oncogenic Ras to transform cells. Recently, it was reported that Ki-Ras4B is also an in vitro substrate for the related enzyme geranylgeranyl protein transferase-1 (GGPT-1) (James, G. L., Goldstein, J. L., and Brown, M. S. (1995) J. Biol. Chem. 270, 6221–6226). In the current studies, we compared the four isoforms of Ras as substrates for FPT and GGPT-1. The affinity of FPT for Ki-Ras4B ( K m (cid:53) 30 n M ) is 10–20- fold higher than that for the other Ras isoforms. Consistent with this, when the different Ras isoforms are tested at equimolar concentrations, it requires 10–20-fold higher levels of C AAX -competitive compounds to inhibit Ki-Ras4B farnesylation. Additionally, we found that, as reported for Ki-Ras4B, N-Ras and Ki-Ras4A are also in vitro substrates for GGPT-1. Of the Ras isoforms, N-Ras is the highest affinity substrate for GGPT-1

Ras proteins are small GTP-binding proteins that play critical roles in cell signaling, differentiation, and proliferation (1). Ras signaling is regulated by a GDP-GTP cycle. Binding of GTP to Ras is required for its productive interaction with Raf-1 and other downstream effector proteins (2). Ras proteins are activated by nucleotide exchange factors such as SOS-1 which stimulate the exchange of GDP for GTP. The lifetime of activated Ras is limited by its intrinsic GTPase activity, which hydrolyzes GTP to GDP. GTPase-activating proteins, such as p120 Ras-GAP and NF-1, stimulate this activity and thereby facilitate inactivation of Ras proteins (2). Transforming mutations of Ras which decrease the rate of GTP hydrolysis result in its constitutive activation. Such oncogenic Ras mutations have been found in about 40% of human cancers and are thought to be a critical factor in the proliferation of these tumors (3). The activity of oncogenic Ras can be assessed by its ability to transform established rodent fibroblast cell lines leading to their growth in soft agar and tumorigenesis in nude mice (4).
Four isoforms of Ras exist: Ha-Ras, N-Ras, Ki-Ras4A, and Ki-Ras4B (2). They are products of three genes, with Ki-Ras4A and Ki-Ras4B being splice variants of the same gene. Their protein sequences are 80% identical with major differences residing in their carboxyl termini. Expression of the four Ras isoforms is tissue-specific in mouse (5). Moreover, oncogenic mutations of the different isoforms predominate in different tumors (3). For example, Ha-Ras mutations are found in carcinomas of the bladder, kidney, and thyroid; N-Ras mutations are found in myeloid and lymphoid disorders, liver carcinoma, and melanoma; whereas Ki-Ras mutations predominate in colon and pancreatic carcinoma. However, the functional differences of the four isoforms are not clear.
The carboxyl termini of all Ras proteins are modified by the isoprenoid farnesyl (6 -8). This modification is catalyzed by farnesyl protein transferase (FPT). 1 FPT transfers a farnesyl group from its prenyl donor, farnesyl diphosphate (FPP), to the cysteine residue of substrate proteins containing a carboxylterminal CAAX motif in which X is typically methionine, serine, or glutamine and A is an aliphatic residue (9). Besides Ras proteins, other FPT substrates include nuclear lamin A and B, the ␥ subunit of the retinal trimeric G protein transducin, rhodopsin kinase, the ␣ subunit of retinal cGMP phosphodiesterase, and PxF, a peroxisomal protein of unknown function (10,11). Short peptides encompassing the CAAX motif of these substrates are utilized as substrates by the enzyme (9,12). A related enzyme geranylgeranyl protein transferase-1 (GGPT-1) transfers a geranylgeranyl group from geranylgeranyl diphosphate (GGPP) to the cysteine residue of substrates containing a CAAX motif in which X is leucine. The substrates of GGPT-1 include RhoA, Rac-1, and the ␥ subunit of several heterotrimeric G proteins (13,14). Although FPT and GGPT-1 display substrate selectivity, there are a few exceptions to these general rules. The Ki-Ras4B protein can be a substrate for both FPT and GGPT-1 in vitro (15). In addition, the small GTPbinding protein, RhoB, can be either farnesylated or geranylgeranylated in vitro by GGPT-1 (16).
One role of farnesylation is to anchor Ras proteins to the membrane where they can participate in signal transduction. Abolishing isoprenylation disrupts Ras membrane association, thereby disrupting its function (7). It has been shown that prenylation is required for oncogenic Ras to transform cells (17,18). Thus, inhibiting farnesylation may be a route for controlling the growth of Ras-transformed tumor cells. This has made FPT a very attractive target for anti-tumor drug discovery. Numerous inhibitors have been developed (11,19), including FPP analogs (20), CAAX peptide analogs such as BZA-5B, L-731,734, B581, and Cys-AMBA-Met (21)(22)(23)(24), and bisubstrate analogs (25). SCH44342 belongs to a novel class of tricyclic inhibitors. It is entirely nonpeptidic and has no sulfhydryl function, but it is a competitive inhibitor versus the CAAX protein substrate (26).
Inhibitors of FPT have been demonstrated to inhibit Ras farnesylation in cell culture and reverse cellular transformation induced by oncogenic Ras (21,22,26). Furthermore, such inhibitors can block tumor formation by Ha-, Ki-, and N-Ras transformed cells in nude mice (27,28). Recently it was reported that several classes of FPT inhibitors block the anchorage-independent growth of human tumor cell lines; however, the sensitivity of a particular cell line to these inhibitors does not correlate to their Ras mutational status or their tissue of origin (29,30). Similar observations have been made with the tricyclic FPT inhibitors. 2 These observations raise questions as to the mechanism of this growth inhibition.
In the studies reported here, we examined for the first time all four full-length Ras isoforms as substrates for the prenyltransferases, FPT and GGPT-1. We found that the affinity of FPT for Ki-Ras4B is more than 20-fold higher than for the other forms of Ras. As a result, it requires 10 -20-fold higher concentrations of SCH44342 or other CAAX-competitive inhibitors to block farnesylation of Ki-Ras4B than Ha-Ras. In addition, we found that not only Ki-Ras4B but also N-Ras and Ki-Ras4A are substrates for GGPT-1. The different distribution of Ras isoforms and the ability of cells to alternatively process these proteins may explain in part the resistance of some tumor cell lines to FPT inhibitors and may also contribute to the lack of cytotoxicity associated with these compounds.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human FPT was expressed and purified (Ͼ95% pure) from the baculovirus/Sf9 cell expression system according to procedures described previously (31 (19.5 Ci/mmol) were obtained from DuPont NEN. BL21(DE3) and SURE strains of Escherichia coli were from Stratagene. Ultrapure Tris, magnesium chloride, and potassium chloride were obtained from Fluka. Zwittergent 3-14 was obtained from Sigma. Nickel affinity resin was from Qiagen. Protein concentrations were routinely determined using the Bradford method with a commercial dye preparation (Bio-Rad) using bovine serum albumin as the standard.
Peptide Synthesis-Sequences of the Ras carboxyl-terminal peptides used are listed in Table I. The peptides were assembled from a preloaded Wang resin on an ABI model 431A peptide synthesizer using FastMoc TM chemistry. The side chain protecting groups were tert-butyl for Ser, Thr, Asp, and Glu, tert-butyloxycarbonyl for Lys, and trityl for Cys and Gln. Solid-phase amino-terminal biotinylation after aminoterminal Fmoc (N-(9-fluorenyl)methoxycarbonyl) deprotection was performed manually, using D-biotin preactivated by O-(7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyl uronium hexafluorophosphate and 1-hydroxy-7-azabenzotriazole (PerSeptive Biosystems). The peptides were cleaved and deprotected by trifluoroacetic acid with scavengers (80% trifluoroacetic acid, 4% H 2 O, 4% phenol, 4% thioanisole, 4% ethanedithiol, 4% triisopropylsilane). The cleaved and deprotected peptides were separated from the resin by filtration and precipitated by anhydrous ethyl ether. The precipitated peptides were dissolved in H 2 O, rotary evaporated to remove the ether, and lyophilized. The crude peptides were finally purified by reversed phase high performance liquid chromatography, and the molecular weights were confirmed by mass spectroscopy.
Polymerase Chain Reaction Cloning and Expression of Ras Proteins-All Ras proteins were expressed as fusion proteins containing six histidine residues at the amino terminus. The coding region of human Ha-, N-, Ki-Ras4A, and Ki-Ras4B were amplified from appropriate cell lines by polymerase chain reaction and subcloned into QE vector to yield QE-Ras. The identity of all plasmids was confirmed by restriction mapping and DNA sequencing of the polymerase chain reaction-amplified fragments. Ha-Ras, Ha-Ras-CVLL, and N-Ras plasmids were transformed into SURE bacteria, while Ki-Ras4A and Ki-Ras4B were transformed into BL21 bacteria.
Purification of Ha-Ras, Ras-CVLL, and N-Ras Proteins-Six-histidine-tagged Ha-Ras, Ha-Ras-CVLL, and N-Ras proteins were purified from bacterial cells by nickel-affinity chromatography (Qiagen). All protein purification steps were performed at 4°C. Cells were lysed with a microfluidizer in 64 mM Tris, pH 7.5, 5 mM MgCl 2 , 1 mM ␤-mercaptoethanol (buffer A) with 4 mM benzamidine, 2 g/ml aprotinin, 2 g/ml soybean trypsin inhibitor, 0.7 g/ml pepstatin, 0.2 mM Pefabloc, and 0.5 g/ml leupeptin. Insoluble material was removed by centrifugation at 15,000 ϫ g for 20 min, and soluble protein was filtered (0.8 M), then loaded onto a nickel-affinity column, and washed with buffer A followed by a short wash with buffer A containing 1 M NaCl. Bound protein was eluted from the column with buffer A plus 100 mM imidazole. Imidazole was removed using a desalting column equilibrated with buffer A. Dithiothreitol was added to 1 mM, and the purified protein was stored in small aliquots at Ϫ80°C. Ha-Ras, Ha-Ras-CVLL, and N-Ras obtained from this procedure are about 90% pure.
Purification of Ki-Ras4A and Ki-Ras4B Proteins-Six-histidinetagged Ki-Ras4A and Ki-Ras4B proteins were purified by denaturing nickel-affinity chromatography followed by renaturation while bound to the column (32). Bacterial pellets were resuspended in denaturation buffer (6 M urea, 20 mM Tris, pH 8.0, 5 mM MgCl 2 , 30 M GDP, 1 mM ␤-mercaptoethanol, 5% (v/v) glycerol, 50 mM NaCl with a protease inhibitor mixture) and lysed using a microfluidizer. Insoluble material was removed by centrifugation as above, and soluble protein was filtered (0.8 M) and loaded onto a nickel-affinity column. The column was washed overnight with renaturation buffer (20 mM Tris, pH 8.0, 5 mM MgCl 2 , 50 mM NaCl, 5% (v/v) glycerol, 30 M GDP, and 1 mM ␤-mercaptoethanol). The column was then washed with renaturation buffer plus 1 mM NaCl, and bound proteins were eluted with renaturation buffer with 100 mM imidazole. Imidazole was removed and proteins were stored in small aliquots at Ϫ80°C. Ki-Ras4A and Ki-Ras4B obtained are about 90% pure.
Expression and Purification of Human GGPT-1-GGPT-1 and FPT share a common ␣ subunit, but have distinct ␤ subunits (33). To produce recombinant GGPT-1, a cDNA clone for the human FPT ␣ subunit was obtained from ATCC (ATCC63225). This cDNA was excised with BamHI and PpuMI, and the resulting fragment was ligated into the vector p2-BAC (Invitrogen) to generate the construct pA2BN. A cDNA clone for the ␤ subunit of rat GGPT-1 in a pGEM vector was obtained from Dr. Patrick Casey (Duke University Medical Center). This clone was humanized by mutating 10 rat residues to their human counterparts using the transformer site-directed mutagenesis kit (Clontech). The residues changed were: A2V, D7E, D91N, N110A, I133V, D143N, S198T, R355L, D371E, and S377T. The mutagenized sequence was confirmed by DNA sequencing. The humanized GGPT-1 ␤ cDNA was excised from pGEM by digesting first with NcoI, filling in with Klenow fragment, and digesting with EcoRI. The resulting fragment was ligated into pBluescript KSϩ to generate plasmid pB-GG. The cDNA for GGPT-1 ␤ subunit was excised from pB-GG by digesting with NotI and ApaI, and the resulting fragment was ligated into pA2BN to generate construct pGG-2B. This construct was used to produce recombinant baculovirus according to Summers and Smith (34). For GGPT-1 production, log phase Sf9 cells (2 ϫ 10 6 /ml) were infected at a multiplicity of infection of 1.5 in a 10-liter Biolafitte tank and cultured for 3 days at 28°C and 140 rpm in SF900-II medium (Life Technologies, Inc.). Cells were harvested by centrifugation at 12,000 ϫ g for 10 min at 4°C. The purification of recombinant GGPT-1 was essentially the same as described previously (35). The final enzyme was about 50% pure.
Determination of FPT and GGPT-1 Activity by Product Precipitation-Activity of FPT and GGPT-1 with protein substrates was determined as described previously using acid precipitation of the prenylated product (9,36).

RESULTS
Expression and Purification of Ras Proteins-All four Ras isoforms (Ha-Ras, N-Ras, Ki-Ras4A, and Ki-Ras4B) were expressed in E. coli as amino-terminal His 6 -tagged proteins and purified. The carboxyl-terminal sequences of each isoform are shown in Table I. The purified fusion proteins are about 90 -95% pure and migrate at 28 kDa in SDS-PAGE (Fig. 1, top  panel). The recombinant Ras proteins migrate slower than native Ras proteins due to the His-tag. To check the integrity of the CAAX box of these proteins, they were reacted with FPT in the presence of excess [ 3 H]FPP under conditions where the reaction could proceed to completion. Following reaction with FPT, Ha-Ras and N-Ras proteins shift to a higher electrophoretic mobility form upon SDS-PAGE, characteristic of the prenylated proteins. Essentially all of the Ha-Ras and N-Ras undergo this gel shift, indicating that they possess intact CAAX boxes and can be farnesylated. Under the SDS-PAGE conditions employed here, Ki-Ras4A and Ki-Ras4B do not undergo a gel mobility shift following their farnesylation. To examine the extent of their farnesylation, gels were exposed to x-ray film. The autoradiographic intensity of Ki-Ras4A and Ki-Ras4B are comparable to those of Ha-Ras and N-Ras (Fig. 1, lower panel), indicating that these proteins are also intact and can serve as substrates for FPT.
Farnesylation and Geranylgeranylation of Ras Proteins-All four Ras proteins serve as substrates for FPT (Fig. 2, A and B).  (Table II).
It was previously shown that Ki-Ras4B is an in vitro substrate for geranylgeranyl protein transferase-1 (GGPT-1) (15). We performed GGPT-1 assays to determine whether other Ras isoforms are also substrates for this enzyme. In agreement with previous reports, Ha-Ras is not a substrate for GGPT-1. However, like Ki-Ras4B, N-Ras and Ki-Ras4A are GGPT-1 substrates (Fig. 2C). K m and k cat values for these proteins were determined at saturating GGPP concentration. Among N-Ras and the two Ki-Ras isoforms, all of which terminate in methionine, N-Ras is the highest affinity GGPT-1 substrate (K m ϭ 2.1 M). This is close to the K m value for Ha-Ras-CVLL, a protein with the GGPT-1 consensus Leu in the X position of the CAAX box. The K m values for Ki-Ras4A and Ki-Ras4B are about 10-fold higher than that for Ha-Ras-CVLL. The k cat values for Ki-Ras4A and Ki-Ras4B are about 5-fold higher than those for N-Ras and Ha-Ras-CVLL. Overall, the catalytic efficiencies for the GGPT-1 reactions with N-Ras or either Ki-Ras isoform are similar and about 2-fold lower than that for Ha-Ras-CVLL (Table II). It is clear from these data that farnesylation of these substrates is the preferred reaction. In the case of Ki-Ras4B, the catalytic efficiency of its farnesylation reaction is about 140-fold higher than that of its reaction with GGPT-1, largely due to the affinity difference.
Carboxyl-terminal Peptides Can Account for Most, but Not All, of the Properties of Ras Proteins-The four Ras isoforms are highly homologous to each other, with most of the differences residing in the last 24 residues (2). To examine whether the carboxyl-terminal sequence differences account for the differences in substrate properties between these proteins, we prepared four biotinylated peptides comprising the 15 carboxylterminal residues of the various Ras proteins (Table I). These peptides were evaluated as substrates for FPT and GGPT-1 using an SPA assay. All four peptides are substrates for FPT (Fig. 3A). The apparent V max for the Ki-Ras4A peptide is 3-to 4-fold higher than those for the other Ras peptides, consistent with assays utilizing whole protein substrates. The highest affinity peptide substrate was that derived from the Ki-Ras4B

D D G T Q G C M G L P C V V M
Ki-Ras4A

E K T P G C V K I K K C I I M
Ki-Ras4B G K K K K K K S K T K C V I M Lanes 2, 4, 6, and 8 are Ha-Ras, N-Ras, Ki-Ras4A, and Ki-Ras4B, respectively, following their reaction with FPT. The gel was stained with Coomassie Blue. In the lower panel, the same SDS-PAGE gel was exposed to x-ray film.
sequence, again consistent with our whole protein studies.
These peptides were also tested as substrates for GGPT-1 (Fig. 3B). The Ki-Ras4A and Ki-Ras4B peptides are substrates for this enzyme. The GGPT-1 reaction with the Ki-Ras4B peptide proceeded with a higher affinity (about 3-fold) and a higher V max than the reaction with the Ki-Ras4A peptide. Surprisingly, the N-Ras peptide was not geranylgeranylated by GGPT-1, suggesting that upstream sequences present in the N-Ras protein may play a critical role in its recognition by GGPT-1.

Both the Polylysine Region and the Carboxyl-terminal Methionine Are Important for Geranylgeranylation of the Ki-Ras4B
Peptide-To understand the critical features enabling the Ki-Ras4B peptide to serve as a substrate for GGPT-1, we prepared carboxyl-terminal peptides of different lengths (Table I). Biotin-GKKKKKKSKTKCVIM is the longest peptide tested, biotin-KKSKTKCVIM contains four less amino-terminal lysine residues, and biotin-TKCVIM is further truncated. In biotin-TKCVIS the last residue is changed from methionine to serine. GKKKKKKSKTKCVIM is a very good substrate for GGPT-1 (Fig. 4). Removing four of the six contiguous lysine residues greatly decreases the affinity of GGPT-1 for this peptide. Further removal of the KKSK sequence had little effect on its utilization; however, changing methionine to serine completely abolished its ability to serve as a GGPT-1 substrate (Fig. 4). Therefore, both the lysine residues and the carboxyl-terminal methionine contribute to the utilization of Ki-Ras4B by GGPT-1. This conclusion is consistent with the results of James et al. (15) who addressed this question using chimeric Ha/Ki-Ras proteins.
Geranylgeranylation of Ki-Ras4B by FPT-We also tested the ability of the four full-length Ras proteins to be used as substrates for FPT and GGPT-1 with different isoprene donors. GGPT-1 did not transfer farnesyl from FPP to any of the Ras proteins. Similarly, FPT did not transfer geranylgeranyl from GGPP to Ha-Ras, N-Ras, or Ki-Ras4A (data not shown). However, FPT can catalyze the transfer of geranylgeranyl from GGPP to Ki-Ras4B (Fig. 5A). The K m and k cat values for this reaction were 1.4 M and 0.6 min Ϫ1 , respectively (Table II). This is similar to kinetic values reported by Pompliano et al.   (Fig. 5B). FPT can transfer geranylgeranyl from GGPP to the Ki-Ras4B peptide, but not to the Ha-Ras or N-Ras peptides. Weak activity was observed with the Ki-Ras4A peptide at the highest concentration tested. The V max for the Ki-Ras4B peptide is about one-half of that for the normal FPT reaction. Inhibition of Ras Farnesylation-SCH44342 is a tricyclic inhibitor of FPT, which competes with various farnesyl acceptors with a K i of 0.24 M. To evaluate its potency against the different isoforms of Ras, we measured its IC 50 using a constant substrate concentration for all four proteins. When Ras proteins were present at 1.2 M, the IC 50 values of SCH44342 for Ha-Ras, N-Ras, Ki-Ras4A, and Ki-Ras4B were 0.6 M, 0.5 M, 0.9 M, and 10 M, respectively (Fig. 6A). The IC 50 for Ki-Ras4B is 10 -20-fold higher than for the other Ras proteins, indicating that, as predicted for a competitive inhibitor, Ki-Ras4B farnesylation is more difficult to inhibit than the other Ras isoforms when present at equimolar concentration. These results are consistent with the affinity of the Ras isoforms for FPT.
To confirm this finding, we also measured the IC 50 of SCH44342 against different Ras peptides. When peptides were present at 0.3 M, the IC 50 of SCH44342 against Ha-Ras, N-Ras, Ki-Ras4A, and Ki-Ras4B peptides were 0.18 M, 0.4 M, 0.7 M, and 60 M, respectively (Fig. 6B). This striking difference between the Ki-Ras4B peptide and other Ras peptides is qualitatively similar to the results we obtained with the fulllength proteins. Similar observations were made with the tetrapeptide FPT inhibitor CVFM (data not shown). When the truncated Ki-Ras4B peptide (lacking the four amino-terminal lysine residues) was tested, the IC 50 of SCH44342 was similar to that for the Ha-Ras peptide, indicating that the polylysine region also increases the binding affinity to FPT (data not shown).
SCH44342 did not inhibit any of the GGPT-1-catalyzed reactions except at very high concentrations (IC 50 Ն 200 M). This is in contrast to the results of James et al. (15) who reported that BZA-5B inhibited the reaction of GGPT-1 with Ki-Ras4B but not with Ha-Ras-CVLL. Interestingly, the geranylgeranylation of Ki-Ras4B catalyzed by FPT is inhibited by SCH44342 more potently than farnesylation of Ki-Ras4B by FPT (data not shown). This most likely reflects the reduced affinity of FPT for Ki-Ras4B when geranylgeranyl diphosphate is the isoprene donor. DISCUSSION In this work we directly compared all four full-length Ras proteins as substrates for the two isoprenyl protein transferases, FPT and GGPT-1. As FPT substrates, the affinity for Ki-Ras4B is 20-fold greater than that for the other Ras proteins (Table II; Fig. 2, A and B). This affinity difference between Ki-Ras4B and the other isoforms is also evident in the experiments using FPT inhibitors. The IC 50 values of SCH44342 for Ha-Ras, N-Ras, and Ki-Ras4A are very similar, while the IC 50 value for Ki-Ras4B is 10 -20-fold higher (Fig. 6A). Under the assay conditions employed, a greater difference was observed for the various carboxyl-terminal peptide substrates derived from these proteins (Fig. 6B). This observation may be significant in the clinical development of FPT inhibitors. To completely inhibit the farnesylation of Ki-Ras4B, significantly higher FPT inhibitor concentrations will be required than those needed for inhibiting Ha-Ras processing.
Early studies of FPT specificity employing tetrapeptides indicated that the Ki-Ras CAAX peptide (CVIM) was a more potent inhibitor of FPT activity than the Ha-Ras CAAX peptide (CVLS) (12). Studies in which various CAAX boxes were introduced into the yeast Ras-1 protein also indicated that the Ki-Ras CAAX sequence supported a higher affinity reaction (38). In these experiments the chimeric protein terminating in CVIM had a K m value of approximately 0.14 M (39, 40). The authentic full-length Ki-Ras4B which we used has an apparent K m of 30 nM (Table II). Our data indicate that Ki-Ras4A and N-Ras behave more like Ha-Ras with respect to their affinity for FPT and sensitivity to competitive inhibitors.
The high affinity of FPT for Ki-Ras4B is not simply due to the presence of a carboxyl-terminal methionine, since N-Ras and Ki-Ras4A have 10-fold lower affinity despite having carboxylterminal methionines. It is likely that the stretch of lysine residues upstream of the Ki-Ras4B CAAX box, and not found in the other Ras isoforms, contributes to this high affinity. This is supported by the observation that Ki-Ras4B peptides lacking these lysines are inhibited by SCH44342 with IC 50 values The standard FPT SPA assay conditions were used except that various concentrations of SCH44342 or vehicle control were included. The concentration of all Ras peptides was 0.3 M, and the assays were conducted at 25°C for 20 min. The FPT activity in vehicle controls was designated as 100%. Symbols are the same as in A. similar to that of the Ha-Ras peptide. The higher affinity that we observe with the authentic, full-length Ki-Ras4B (versus the yeast Ras-CVIM construct) probably reflects the presence of this region. Similar observations were made by James et al. (15) who found that Ki-Ras4B is about a 50-fold higher affinity substrate for FPT than is Ha-Ras. However, in those studies the apparent K m values reported (10 M for Ha-Ras and 0.2 M for Ki-Ras4B) are unusually high compared with those we found in the current studies and those reported by Pompliano et al. (39). Through construction of chimeric proteins James et al. (15) observed that addition of either the Ki-Ras4B CAAX box or the polybasic domain onto Ha-Ras increased its affinity for FPT.
The other significant finding in this work is that in addition to being FPT substrates, N-Ras and Ki-Ras4A, like Ki-Ras4B, serve as in vitro GGPT-1 substrates (Fig. 2C), while Ha-Ras is only a FPT substrate. This is the first demonstration that N-Ras and Ki-Ras4A are GGPT-1 substrates despite the fact that these proteins terminate in methionine rather than the leucine present in most substrates for this enzyme. However, it is now clear that the substrate specificity of GGPT-1 is not restricted to proteins terminating in Leu or Phe. Surprisingly, the Ras isoform with the highest affinity for GGPT-1 was N-Ras ( Fig. 2C and Table II), having an affinity similar to that of a standard in vitro GGPT-1 substrate, Ha-Ras-CVLL (13,14). Although N-Ras, Ki-Ras4A, and Ki-Ras4B can be geranylgeranylated in vitro, the catalytic efficiencies (k cat /K m ) of the GGPT-1 reactions are much smaller than those for the FPTcatalyzed reactions, indicating that farnesylation of these proteins is preferred (Table II). Consistent with this, when the isoprene content of Ki-Ras was analyzed following mevalonate labeling of cells, only the farnesyl isoprene was detected (6). Recently, however, we observed that when COS monkey kidney cells or human colon carcinoma DLD-1 cells are treated with FPT inhibitors there is a switch in the isoprene content of N-Ras and Ki-Ras4B to geranylgeranyl. 3 Sixteen-residue peptides derived from the carboxyl termini of the various Ras proteins were also tested as GGPT-1 substrates. In agreement with the results with the full-length proteins, Ki-Ras4A and Ki-Ras4B peptides were geranylgeranylated. Surprisingly, however, the carboxyl-terminal peptide from N-Ras was not an in vitro GGPT-1 substrate (Fig. 3B). These results suggest that additional upstream residues present in N-Ras may influence its interaction with GGPT-1. Similarly, upstream sequences in Ki-Ras4A may also play a role since, although the 2 full-length Ki-Ras isoforms have similar affinities for GGPT-1, the Ki-Ras4A peptide is a lower affinity GGPT-1 substrate than the Ki-Ras4B peptide. The nature of these upstream sequences in N-Ras and Ki-Ras4A have not been determined. Recently, Maltese and co-workers (41) reported that upstream structural elements may also play a role in the recognition of G protein ␥ subunits by the prenyltransferases. Influence of upstream sequences on GGPT-1 substrate recognition was also suggested by the work of Cox et al. (42) who showed that while the CGLF motif found at the carboxyl terminus of G i ␣ does not support prenylation of this protein it does support prenylation when grafted onto the carboxyl terminus of Ha-Ras or Ki-Ras.
Since the properties of the Ki-Ras4B carboxyl-terminal peptide are very similar to the full-length protein, we further explored the structure/function relationship in this region. We found that removing the upstream polybasic residues dramatically reduced the ability of Ki-Ras4B to serve as a substrate for GGPT-1 (Fig. 4). Further, changing methionine to serine completely abolished its GGPT-1 substrate activity. The conclusion that both the carboxyl-terminal methionine and polybasic region contribute to the recognition of Ki-Ras4B by GGPT-1 is consistent with the conclusion reached using Ha-Ras/Ki-Ras chimeric proteins (15).
Ki-Ras4B was also unique among the Ras isoforms in its ability to be geranylgeranylated by FPT (Fig. 5). The k cat for geranylgeranylation of Ki-Ras4B by FPT is only about onethird that for its farnesylation. Similarly, the K m for Ki-Ras4B geranylgeranylation by FPT is about 46-fold larger than that for its farnesylation, indicating that the farnesylation reaction is greatly favored. The Ki-Ras carboxyl-terminal peptide GKKKKKKSKTKCVIM was also geranylgeranylated by FPT. Similar activity has been reported for the yeast Ras-1 protein containing a CVIM terminus (38). FPT can bind both isoprenoid diphosphates, but it has much higher affinity for FPP. The finding that the K m for Ki-Ras4B farnesylation is much smaller than that for its geranylgeranylation by FPT provides further evidence that isoprenoid diphosphate binding to FPT affects its binding affinity for protein acceptors.
FPT inhibitors, including the tricyclic compounds related to SCH44342, block anchorage-independent growth of a number of human tumor cell lines with a wide range of potency. The data reported here suggest that the type and level of Ras isoforms expressed in various cell types and the ability of cells to alternatively prenylate these proteins may contribute to such differences in sensitivity. We are currently examining the capacity of a variety of tumor cell lines which differ in sensitivity to FPT inhibitors to carry out the alternative prenylation reactions. Finally, as suggested by James et al. (15) the lack of cytotoxicity exhibited by FPT inhibitors may in part be explained by alternative prenylation of Ras isoforms and, perhaps, of other FPT substrates such as the nuclear lamins.