Effects of Farnesylcysteine Analogs on Protein Carboxyl Methylation and Signal Transduction*

Several proteins associated with signal transduction in eukaryotes are carboxyl methylated at COOH-ter-minal S-farnesylcysteine residues. These include members of the Ras superfamily and y-subunits of hetero- trimeric G-proteins. The enzymes that catalyze the carboxyl methylation reaction also methylate small molecules such aa N-acetyl-S-tran8,trane-farnesyl-L-cysteine (AFC). AFC inhibits carboxyl methylation of p21- and related proteins both in vitro and in vivo. Saturating concentrations of AFC cause a >SO% inhi- bition of chemotactic responses of mouse peritoneal macrophages. Our results suggest that carboxyl meth- ylation may play a role in the regulation of receptor-mediated signal transduction processes in eukaryotic cells.


Saturating concentrations of AFC cause a >SO% inhibition of chemotactic responses of mouse peritoneal macrophages. Our results suggest that carboxyl methylation may play a role in the regulation of receptormediated signal transduction processes in eukaryotic cells.
In bacteria, the activities of membrane receptor-transducer proteins are regulated by carboxyl methylation at glutamate residues (Springer et al., 1979;Ninfa et ul., 1991;Stock et al., 1991). Recently, several important classes of signal transduction proteins in eukaryotes have been shown to be carboxyl methylated. These include members of the Ras superfamily of guanine nucleotide-binding proteins (Clarke et al., 1988;Gutierrez et al., 1989;Stimmel et al., 1990), the y-subunits of heterotrimeric G-proteins (Fukada et al., 1990;Fung et ul., 1990;Yamane et al., 1990), the a-subunit of cGMP phosphodiesterase from retinal rods (Ong et al., 1989), and a class of fungal mating pheromones (Sakagami et al., 1981;Ishibashi et al., 1984;Anderegg et al., 1988). This raises the possibility that carboxyl methylation may play a central role in the regulation of stimulus-response coupling in eukaryotic cells.
Eukaryotic signal transduction proteins are carboxyl methylated at modified cysteine residues that are produced by a series of post-translational modification events (Clarke et al., 1988;Hancock et ul., 1989). The nascent proteins have a cysteine residue 4 amino acids from the COOH terminus within a characteristic sequence that has been termed a CAAX tail (Barbacid, 1987). A polyisoprenoid group is coupled to this cysteine through a thioether bond, the 3 residues that follow the cysteine are proteolytically cleaved, and the resultant COOH-terminal carboxyl group is methyl esterified.
8 Supported by fellowship GM-08309 from the National Institutes of Health.
Here we report the effects of a specific inhibitor of the COOH-terminal S-farnesylcysteine methyltransferase. Evidence is presented that the enzyme that methylates Ras will also methylate small molecule analogs of the modified COOHterminal amino acid, S-farnesylcysteine. One of these analogs, N-acetyl-S-trans,trans-farnesyl-L-cysteine (AFC),' has a high affinity for the methyltransferase (KM = 20 p~) and functions as a specific inhibitor of the COOH-terminal S-farnesylcysteine methyltransferase both in cell-free extracts and in living cells. The availability of a specific inhibitor for this type of protein carboxyl methylation in eukaryotic cells has allowed us to begin to directly assess the physiological significance of this reaction. Our results indicate that in vertebrate cells, as in bacteria, protein carboxyl methylation plays a role in receptor-mediated signal transduction processes.

MATERIALS AND METHODS
S-Substituted Cysteine Compounds-S-Substituted cysteine compounds were prepared from the chloride or bromide of the desired substituent in a manner analogous to our previously reported syntheses of S-farnesylcysteine derivatives . The products were characterized by NMR, mass spectroscopy, and HPLC.
Tissue Extracts-Extracts from mouse tissues were prepared at 4 "C by homogenization with a Teflon-glass homogenizer in 5 ml/g, wet weight, tissue of 100 mM Tris-C1, 1 mM EDTA, 1 mM DTT, pH 7.9 (buffer A). Bovine brain extracts were prepared from whole brains that were homogenized in a blender with 3 ml of buffer A/g, wet weight, tissue, followed by filtration through cheesecloth. Extracts of S. cereuisiae and E. coli were prepared similarly, except for the cell disruption procedure. Bacteria were lysed by sonication in buffer A, and yeast cells were broken by vortexing with glass beads in buffer A with 0.1 mM phenylmethylsulfonyl fluoride (buffer B). Rat tissue extracts were prepared with a Teflon-glass homogenizer in 9 volumes of 320 mM sucrose, 7.5 mM Tris-C1, 1 mM CaC12, 1 mM DTT, 0.5 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, pH 7.4 (buffer C).
Rat tissue extracts were centrifuged for 10 min at 800 X g. The pellets were washed once in buffer C, then resuspended in 320 mM sucrose, 7.5 mM Tris-C1, 1 mM DTT, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, pH 7.4 (buffer D), to yield the crude nuclear bined, an additional 1 mM EDTA added, then centrifuged for 20 min fractions. The postnuclear supernatants and washings were comat 16,000 X g. The pellets were washed once in buffer D, then resuspended in buffer D to yield the crude mitochondrial fractions. The postmitochondrial supernatants and washings were combined then centrifuged for 90 min at 100,000 X g. The supernatants constituted the soluble fractions. The high speed pellets were washed, then resuspended in buffer D to produce the crude microsomal fractions. All resuspensions were performed using a Teflon-glass homogenizer. Bovine brain fractions were prepared in the same way, except buffer A was used in each step. With yeast, buffer B was used throughout, and the 16,000 X g fractionation step was omitted.
Protein concentrations were determined by the method of Bradford (1976) using bovine serum albumin as a standard. To determine the concentration of bovine serum albumin, the extinction coefficient at 280 nm for a 1 mg/ml solution was taken to be 0.66 (Tanford and Roberts, 1952).
Protein Carboxyl Methylation Assay-To determine the level of carboxyl methylated proteins from dried SDS-PAGE gels, lanes were cut into 3-mm slices that were each placed into 1.5-ml open-topped polyethylene microcentrifuge tubes, and 75 p1 of 1 M NaOH were added to hydrolyze 3H-protein methyl esters to [3H]methanol. The amount of [3H]methanol released was measured by the methanol vapor-phase diffusion procedure (Stock et al., 1984).

RESULTS
S-Farnesylcysteine Carboxyl Methyltransferase Actiuity-Cell-free extracts from a wide variety of eukaryotic sources catalyze the transfer of methyl groups from S-adenosyl-Lmethionine (AdoMet) to AFC. By following the radioactivity from [3H]AdoMet we were able to identify by reverse phase HPLC the product of the methylation reaction as the acarboxyl methyl ester derivative of AFC . Significant methyltransferase activity was detected in all eukaryotic cells and tissues that were examined, including S. cereuisiae ( Table I). The highest levels were found in mammalian brain and the lowest in blood. No methyltransferase activity was detected in E. coli.
An analysis of the subcellular distribution of AFC methyltransferase activity in brain, liver, and heart indicates that the enzyme is tightly associated with membranes (Table 11). Although significant levels were detected in all membrane fractions, the highest specific activity was found in the crude microsomal fractions. The total activity was associated primarily with the material that sedimented with the crude nuclear fractions. This correlates with the location of nuclear lamin B, a protein that is reversibly methylated at a COOHterminal S-farnesylcysteine residue in a cell cycle-dependent manner (Chelsky et al., 1987).
Mutant strains of S. cereuisiae with defects in STE14 are deficient in the ability to methylate a-factor and other peptides with a COOH-terminal S-farnesylcysteine (Marr et al., Distribution of S-farnesylcysteine methyltransferase activity Cell extracts, prepared as described under "Materials and Methods," were incubated with AFC (100 pM) and [3H]AdoMet (10 pM; 8000 cpm/pmol) at 37 "C in buffer A in a total volume of 50 pl for 25 min. The level of t3H]AFC methyl ester was determined by the heptane extraction method described previously   Subcellular localization of S-farnesylcysteine methyltransferase activity Cell extracts, prepared as described under "Materials and Methods," were incubated with AFC (100 p~) and [3H]AdoMet (10 pM; 6000 cpm/pmol) in buffer A at 37 "C in a total volume of 67 pl for 25 min. The level of [3H]AFC methyl ester was determined by the heptane extraction method described previously . The level of protein in each incubation mixture was 15 pg. The crude nuclear fraction was prepared from the pellet that precipitated as a result of centrifugation of the tissue homogenate for 10 min at 800 X g. The crude mitochondrial fraction was isolated as a 16,000 X g pellet, and the crude microsomal fraction as a 100,000 X g pellet.

Source Fraction
Methyltransferase Total Overall specific activity protein activity  Hrycyna and Clarke, 1990). A stel4 mutant was also found to be totally deficient in AFC methylation (Table 111).
In yeast, as in mammalian tissues, S-farnesylcysteine methyltransferase activity is tightly associated with membrane fractions (Table 111). Under the conditions used, AFC methyl ester was produced at a constant rate for 25 min. The rate increased proportionately with protein level. The activity in both mouse and bovine S-Farnesylcysteine methyltransferase activity in S. cerevisiae Cell extracts, prepared as described under "Materials and Methods," were incubated with AFC (100 pM) and [3H]AdoMet (10 pM; 8000 cpm/pmol) in buffer B at 37 "C in a total volume of 50 pl for 25 min. The level of t3H]AFC methyl ester was determined by the heptane extraction method described previously . The level of protein in the incubation mixtures was adjusted to be 73 pg for the membrane fractions and 80 pg for the soluble fractions. Each membrane fraction was prepared from the pellet that precipitated as a result of centrifugation of the postnuclear supernatant for 90 min at 100,000 X g. The specificity of the mouse brain methyltransferase was investigated using a series of AFC analogs (Table IV). A reduction in the size of the prenyl group from farnesyl to geranyl or dimethylallyl resulted in a dramatic loss of reactivity. With each isoprene unit removed, the K M increased roughly 50-fold. At a 10-fold molar excess, these compounds had little effect on the rate of carboxyl methylation of AFC. The enzyme is also specific for the stereochemistry at the acarbon with D-AFC being a much less effective substrate than L-AFC. The free amino acid, S-farnesylcysteine, with an unmodified a-amino group, is also a poor substrate with a V,,,.,/KM ratio approximately 2% of that exhibited by AFC.

Strain
Thus, the specificity profile of the AFC methyltransferase is consistent with what one might expect for an enzyme that modifies farnesylcysteine groups at the COOH termini of proteins. Inhibition of Endogenous Protein Carboxyl Methylation-When extracts prepared from mammalian tissues were incubated with [3H]AdoMet and separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), a variety of proteins were carboxyl methylated (Fig. 1). In brain homogenates, for instance, there are major carboxyl methylated species with apparent molecular weights of 18,000-26,000, 36,000, 45,000, and 70,000. The lower molecular weight species correspond to members of the Ras superfamily (Maltese et al., 1990), the 70-kDa molecular mass component to nuclear lamin B (Chelsky et al., 1987), and the 45-kDa molecular mass species to the principal methylated degradation product of nuclear lamin B (Chelsky et al., 1987). Proteins at these molecular weights are also labeled when vertebrate cells are incubated with [3H] mevalonate, a precursor of isoprenoids (Schmidt et al., 1987;Beck et al., 1988;Wolda and Glomset, 1988;Maltese et al., 1990), and many of these proteins have been shown to have COOH-terminal S-farnesylcysteine residues (Farnsworth et al., 1989;Casey et al., 1989;Buss et al., 1991). We also detect a major peak of protein carboxyl methylation with an apparent molecular mass of 36 kDa. Mevalonate-labeled proteins of this size have not been reported. The identity of the 36-kDa methylated protein remains to be established.
Carboxyl methylation of the 70,45, and some of the 18-26 kDa molecular mass proteins are inhibited by AFC (Fig. 2). Thus, AFC inhibits the methyltransferase activities that modify proteins that are known to be modified at COOH-terminal S-farnesylcysteine residues. In contrast, methylation of the 36-kDa species is not affected by AFC. We have identified a cytolsolic methyltransferase activity that catalyzes the carboxyl methylation of the 36-kDa protein, and this activity is clearly distinct from the membrane-associated S-farnesylcysteine methyltransferase.2 Thus, AFC provides a useful means to distinguish at least two major classes of eukaryotic protein carboxyl methyltransferases. The precise nature of the methylated residues in the 36-kDa protein remain to be established.
Inhibition of p21" Carboxyl Methylation in a Reconstituted Post-translational Modification System-We examined the methylation of Ras in a reconstituted post-translational modification system. Unmodified p21Ki"".2B was overproduced in E. coli and purified. This protein is an efficient substrate of the farnesyltransferase that has recently been characterized by Reiss et al. (1990), and we have identified a membraneassociated protease in rat brain that cleaves the COOHterminal 3 amino acids from farnesylated p21K"r"-2B.2 In the experiment reported in  (Table IV), is a poor inhibitor of p21Ki~r"~2B carboxyl methylation. D-AFC (lane 4 ) , which has a KM nearly as low as AFC (Table IV), also significantly inhibits carboxyl methylation of p21Ki""-2B. Although it is a good inhibitor, D-AFC is not methylated as well as AFC (Table IV). This feature may prove useful to the design of pharmacologically relevant inhibitors of S-farnesylcysteine methylation. Recently, AFC has also been reported to inhibit the in uitro carboxyl methylation of other CAAX tail proteins, including platelet Rap1 (Huzoor-Akbar et al., 1991) and the y-subunit of transducin in retinal rod outer segments (Perez-Sala et al., 1991). Inhibition of p21" Carboxyl Methylation in Transformed Rat Embryo Fibroblasts-AFC is a relatively hydrophobic amino acid derivative that might be expected to move across cell membranes by simple diffusion or through an amino acid or peptide transport system. To explore this possibility we examined the effects of AFC on the methylation of p21Ha"" in transformed fibroblasts. Levels of p2lHn"" methylation were examined in cells grown for 3 h with [meth~l-~HImethionine in the presence or absence of AFC. Under these conditions, 100 pM AFC caused a 60-70% decrease in p2lHa-" carboxyl methylation (Fig. 4). Mature p21Ha-" contains 4 methionine residues (Barbacid, 1987) so that under steady state labeling conditions the incorporation of 1 methyl ester per p21 monomer would give a 1:4 ratio of 3H-methyl esters to [3H]methionine. In the absence of inhibitor, a ratio of 1:6 was obtained after the 3-h incubation. This indicates a stoichiometry of 0.7 methyl groups/molecule. In the presence of AFC, this value was reduced to 0.2-0.3. AFC had no detectable effect on [3H]methionine incorporation into p2lHa-" protein.
Despite its inhibitory effect on carboxyl methylation, AFC had little effect on rates of proliferation of these p2lH"'"" transformed cells.
Inhibition of Macrophage Chemotaxis with AFC-General methylation inhibitors have been previously shown to block P21Ki-rm-2B was used as a substrate for the methyltransferase.
P21Ki-".ZB was greatly inhibited as compared to the control J. B. Stock, unpublished results.

Specificity of the S-farnesylcysteine methyltransferme activity
The indicated cysteine derivatives were incubated at 37 "C with [3H]AdoMet (10 PM, specific activity: 2500 cpm/ pmol) and mouse brain homogenate (300 mg of protein) in buffer A for various times up to 25 min. Carboxyl methylation of the derivatives was assayed by our previously reported HPLC or heptane extraction methods (Volker et al., 1990). Initial rates were determined at 6 concentrations of each substrate, except for FC and ABC where 2 concentrations were used. ND, not determined. Two 30-pl aliquots were removed and suhjected to SDS-PAGE according to the method of Laemmli (1970). The methyl esterification of proteins was assayed hy the methanol vapor diffusion procedure as descrihed under "Materials and Methods." The protein concentrations (mg/ml) in each reaction were: hrain, 2.8; testis, 2.6; thymus, 1.6; heart, 3.1; small intestine, 1.5; spleen, 3.5; kidney, 6.6; hlood, 6.1; liver, 7.7; lung. 2.8; thigh muscle, 0.6. Levels of protein carhoxyl methylation in liver, lung. and thigh muscle were not significantly above background. macrophage chemotaxis (Snyderman, 1985).

Substrates
To determine the potential role of S-farnesylcysteine carboxyl methylation in this response pathway, we incuhated mouse peritoneal macrophages with various concentrations of AFC and measured their response toward endotoxin-activated mouse serum. AFC was found to dramatically reduce the chemotactic response, even a t 10 p M , while a t a 20-fold higher concentration, AGC had only a slight effect (Fig. 5 ) . The inhibitory effect of AFC was reversed if the cells were washed free of the drug prior to the assay. AFC did not affect cell motility, morphology, or viability. Moreover, AFC had no effect on the ability of the cells to migrate in a gradient of the phorbol ester, phorbol 12-myristate 13-acetate. Phorbol 12-myristate 13acetate is known to activate protein kinase C (Snyderman, 1985), thus obviating the requirement for diacylglycerol, the second messenger generated by phospholipase C-catalyzed hydrolysis in the chemotactic response pathway (Kikkawa and Nishizuka, 1986;Kaibuchi et al., 1985). Thus, AFC interferes with a step in signal transduction that precedes protein kinase C activation.

DISCUSSION
We previously identified an activity in eukaryotic cells that methylates the n-carboxyl group of the small molecular weight S-farnesylcysteine derivative, AFC (Volker et al., 1990). Here we show that this is the same activity that catalyzes the methyl esterification of proteins such as Ras at COOH-terminal S-farnesylcysteine residues. The following lines of evidence support this contention: (a) the methyl acceptor activities of small molecular weight cysteine derivatives fit the specificity one would expect for a Ras methyltransferase in that the farnesyl group provides an essential recognition MnCI2, 1 mM DTT, pH 7.0. ' To this reaction was added 1 pl (7 p g l of a 100,000 x R rat hrain crude microsomal fraction that was isolnted in 20 mM Tris-CI, 1 mM DTT, pH 7.0. This was the source of hoth S-farnesylcysteine peptidase and S-farnesvlcvsteine carhoxvl methyltransferase. After a 1-h incuhation at X7 "C. [ 'HIAdoMt-t nnd vnrious AFC analogs were added and the incuhntion continued 20 min longer. The methylation reaction was stopped by the addition of 5 X protein gel loading huffer antl heated to 90 'C for : i min. The extracts were separated hv SDS-PAGE nnd visualized hy fluorography. 'The following compounds were included in the methylation reactions: lnnc 1, 100 mM Tris-CI, 1 mM DTT, pH 7.0; lnnr 2. 100 p~ AFC [nnv 3 , 100 p M AGC; lane 4, 1 0 0 p M r>-AFC. h n c . 5 was the same as lonc I except that p21'" was omitted from the reaction mixture. element, and an amide linkage to the n-amino group is preferred over the free amino acid; ( h ) a S. carmiqiaP mutant that is defective in its ability to methylate a-factor and other polypeptides a t COOH-terminal S-farnesylcysteines is also defective in AFC methyltransferase activity; and ( c ) AFC functions as a specific inhihitor of the methylation of proteins such as p21"" that are known to he carboxyl methvlated at COOH-terminal S-farnesylcysteine residues.
Proteins and peptides translated with CAAX tails are suhject to a series of post-translational modifications that, includes thioether linkage of a farnesyl group to the cysteine, FIG. 4. Effect of AFC on the carboxyl methylation of p21"""" in transformed rat embryo fibroblasts. Cells co-transfected with p21"""" and p53 (5 X 10" cells on 100-mm plates) were labeled with 0.5 mCi of [."'S]methionine (ICN trans lahel, 70% methionine, 2.1 X 10'' cpm/pmol) or 1.0 mCi of ["Hjmethionine (Du Pont-New England Nuclear, 9.3 X 10' cpm/pmol) in the ahsence or presence of 100 p M AFC in 5 ml of Ihlhecco's modified Eagle's medium that contained 2% fetal hovine senlm on 100-mm plates. After a 3-h incuhation at 37 "C, the cells were harvested, washed, lysed, and p21"" ' =* was immunoprecipitated with Y13-259 antihody (Oncogene Science). Half of each immunoprecipitate was suhjected to SDS-PAGE. Total protein was assayed hy autoradiography of the dried ."S-laheled gel (top). The percentage of Ras that was carhoxyl methylated was det.ermined hy analysis of hase-lahile methyl groups with the methanol vapor diffusion method relative to total met.hyl groups incorporated per gel slice. The detailed protocols for these procedures have been previously descrihed (Clarke rf ol., 1988). cleavage of the residues distal to this modified group, and carboxyl methylation of the exposed S-farnesylcysteine a- carboxyl (Miyakawa et al., 1985;Clarke et al., 1988;Hancock et al., 1989). A mammalian activity that catalyzes the transfer of a farnesyl group from farnesyl pyrophosphate to the CAAX tail cysteine in p2lrn' proteins has been purified and characterized (Reiss et al., 1990;Schaber et al., 1990;Manne et al., 1990;Reiss et al., 1991). The enzyme also farnesylates a peptide that contains only the last 6 amino acids of unprocessed human p21K"""-2" (Reiss et al., 1990). Moreover, tetrapeptides that consist solely of the COOH-terminal CAAX motif are potent inhibitors of farnesylation, and the inclusion of more upstream amino acids does not dramatically increase inhibition.
An activity in rat liver has been characterized that carboxyl methylates an S-farnesylcysteine peptide with an analogous sequence to that which is upstream of the CAAX tail of the Drosophila Drasl protein (Stephenson and Clarke, 1990). The apparent KM for this substrate is 2.2 p~. The affinity of the mouse brain methyltransferase for AFC is only an order of magnitude less (20 p~) .
This result indicates that the principal recognition element for the carboxyl methyltransferase is simply a COOH-terminal S-farnesylcysteine. Thus, the farnesyltransferase and the carboxyl methyltransferase each have little specificity for protein substituents upstream from the CAAX tail.
In addition to the S-farnesylcysteine methyltransferase, which is membrane-associated, there is a soluble methyltransferase activity that methylates a 36-kDa protein. Throughout a wide range of tissues, methylation is more prevalent a t 36 kDa than for species that are modified a t S-farnesylcysteine residues. We are currently in the process of determining the hv a method adapted from Aksamit ef al. (1981). The lower chamher was filled with a 1:lOO dilution of control senlm or endotoxin-activated m o~~s e senlm (Hoetcher and Meltzer, 1975) in Iil'MI I640 (GIHCO). The filled lower chamher was then overlaid with a fresh 10-pm thick polycarhonate memhrane with 5-pm pores (Nucleopore; Horwitz and (iarrett. 1971) such that no air huhhles were trapped underneath. Filter retainers that contain the upper chamber were then attached. PI.:("s. 6 X 10'' cells, 200 p l ) , were added to the upper well immediatelv after they were prepared as descrihed under "Materials and Methods." After a 4-h incuhation a t 37 "C in a humidified atmosphere. enriched to 5% C02, the memhrane was removed. fixed in methanol, and stained with Wright-Giemsa stain (Camco Quik Stain 11; American Scientific Products). The chemotactic response was determined with the aid of a light microscope hy counting and averaging the numher of cells in 100 X power fields that had migrated across the filter to the side that faced the lower well ( h h r and Snvderman, 1981). A .
I'EC's were prepared in media that contained AFC (O), A M : (0). or neit,her inhihitor (9). The chemotactic response va1rlt.s are the averages of seven experiments with three memhranes per experiment. An average of 5 cells WAS ohserved for the Chemotaxis of P I X ' S toward control serum. This hackground was suhtracted from each value ohtained where endotoxin-activated mouse serum was used.
'The error bars represent the standard errors associated with each set of measurements. A 100% response was equivalent to 383 cells. H. I'W's were prepared in media with (W) or without AFC K l ) . Phorhol myristate acetate (20 nM) was addrd to the lower well. nature of this modification. Our preliminary results indicate that the carboxyl methyl group is associated with an amino acid that is more hydrophobic than an S-farnesvlcysteine methyl ester. A detailed analysis of the total pool of prenylated proteins in tissue homogenates indicates that a 20-carbon isoprenoid, geranylgeranyl, is actually a more common cysteine adduct than the 15-carbon farnesyl group Rilling et al., 1990). It seems unlikely, however, that the methylated residue in the 36-kDa protein is a COOHterminal geranylgeranylcysteine. Preliminary results suggest that the same carboxyl methyltransferase that modifies COOH-terminal farnesylcysteine residues also methylates proteins that have COOH-terminal geranylgeranvlcvsteine residues.' In addition, AFC appears to inhibit the methylation of proteins with a COOH-terminal geranylgeranylcysteine. AFC has been shown to inhibit Rapl methylation in human platelets (Huzoor-Akbar et al., 19911, and RaplA expressed in baculovirus-infected insect cells has been shown to be geranylgeranylated (Buss et al., 1991). Moreover, we have observed that AFC inhibits the carboxyl methylation of 20-23-kDa proteins in human neutrophils including one that is immunoprecipitated by polyclonal antisera to Rapl and Rap2:' A set of 20-23-kDa proteins in a mouse macrophage ' C. Volker and J. R. Stock FIG. 6. S-Farnesylcysteine targets in eukaryotes. Specific signal transduction components are reversibly methylated at COOH-terminal S-farnesylcysteine residues. These proteins may have two distinct classes of regulatory targets, those that interact with the proteins in their methyated state (M-targets) and those that interact with the demethylated proteins (D-targets). Changes in levels of methylation would regulate the relative outputs from these two types of effector systems.

Membrane
cell line were reported to be carboxyl methylated by a soluble methyltransferase activity (Backlund and Aksamit, 1988), but recent results indicate that the substrates are soluble, not the methyltransferase.6 Several distinct functions have been proposed for S-farnesylcysteine carboxyl methylation. In the case of nuclear lamin B, it has been suggested that methylation and demethylation play a role in the regulation of nuclear lamina assembly (Chelsky et al., 1987;Chelsky et al., 1989). In the case of S. cereuisiae, a-factor pheromone binding to a membrane receptor of the rhodopsin/@-adrenergic superfamily mediates the activity of a heterotrimeric G-protein (Whiteway et al., 1989). a-Factor is much more active in eliciting a mating response when its COOH-terminal S-farnesylcysteine residue is methylated, and it has been proposed that demethylation provides a mechanism for pheromone inactivation (Anderegg et al., 1988).
The functions for methyl esterification of proteins in the Ras and the G, families remain to be established. These proteins interact with other proteins to direct an array of cellular processes. Activating mutations in ras proto-oncogenes can cause uncontrolled growth in some cell types, while activating mutations in homologous guanine nucleotide-binding proteins, such as RaplA, can antagonize these effects (Buss et al., 1991). It has been estimated that each mammalian cell may have from 30 to 100 different members of the Ras superfamily, each responding to a different spectrum of regulatory inputs, and interacting with a different group of effector activities (Chardin, 1988). Because of this diversity, the physiological significance of S-farnesylcysteine carboxyl methylation may be quite complex.
The discovery that mammalian S-farnesylcysteine methyltransferase activities are specifically inhibited by AFC provides a means to begin to probe the functions of this modification. Our results indicate that inhibition of p21Ha""S methylation does not significantly affect growth or transformation of p21H"-" transformed mammalian cell lines. These findings are consistent with the observation that a methyltransferase defect in yeast, STE14, does not preclude the function of the yeast RAS proteins. This does not mean that carboxyl methylation of S-farnesylcysteine residues is generally without import, however. The chemotactic response of mouse peritoneal macrophages is dramatically inhibited by AFC. The effect of AFC is consistent with a defect in the signal transduction pathway. Heterotrimeric G-proteins are integral players in P. S. Backlund, personal communication. this pathway, and all G, subunits that have been investigated are prenylated and methyl esterified at COOH-terminal cysteines. Besides an effect on G,, the possibility should not be discounted that AFC may act through a member of the Ras superfamily with an important role in eukaryotic chemotaxis. Despite a lack of detailed information concerning a specific role for carboxyl methylation in eukaryotic signal transduction one can begin to formulate a general mechanism (Fig. 6). Farnesylation directs signal transduction proteins to regulatory targets in the membrane Casey et al., 1989;Schafer et al., 1989). It seems likely that these membrane targets fall into two classes: those that prefer a demethylated acidic COOH terminus (D-targets), and those that prefer a methylated COOH terminus ("targets). Increases in the level of methylation of a signal transduction protein would increase its interaction with "targets and decrease its interaction with D-targets. Decreases in methylation would have an opposing effect. According to this view, the a-factor receptor is an "target since it specifically interacts with methylated a-factor. On the other hand, the Sfarnesylcysteine methyltransferase provides an example of a D-target, since it is a membrane protein that preferentially interacts with demethylated signal transduction proteins. A consideration of this model suggests that another possible locus for the inhibition of signal transduction by AFC could be a direct competition for COOH-terminal S-farnesylcysteine receptor sites. In effect, inhibition of the methyltransferase may represent only one example of competition of AFC for an S-farnesylcysteine recognition site in a target protein.