Identification of a cysteine residue essential for activity of protein farnesyltransferase. Cys299 is exposed only upon removal of zinc from the enzyme.

Protein farnesyltransferase (FTase) is a zinc metalloenzyme that performs a post-translational modification on many proteins that is critical for their function. The importance of cysteine residues in FTase activity was investigated using cysteine-specific reagents. Zinc-depleted FTase (apo-FTase), but not the holoenzyme, was completely inactivated by treatment with N-ethylmaleimide (NEM). Similar effects were detected after treatment of the enzyme with iodoacetamide. The addition of zinc to apo-FTase protects it from inactivation by NEM. These findings indicated the presence of specific cysteine residue(s), potentially located at the zinc binding site, that are required for FTase activity. We performed a selective labeling strategy whereby the cysteine residues exposed upon removal of zinc from the enzyme were modified with [3H]NEM. The enzyme so modified was digested with trypsin, and four labeled peptides were identified and sequenced, one peptide being the major site of labeling and the remaining three labeled to lesser extents. The major labeled peptide contained a radiolabeled cysteine residue, Cys299, that is in the β subunit of FTase and is conserved in all known protein prenyltransferases. This cysteine residue was changed to both alanine and serine by site-directed mutagenesis, and the mutant proteins were produced in Escherichia coli and purified. While both mutant proteins retained the ability to bind farnesyl diphosphate, they were found to have lost essentially all catalytic activity and ability to bind zinc. These results indicate that the Cys299 in the β subunit of FTase plays a critical role in catalysis by the enzyme and is likely to be one of the residues that directly coordinate the zinc atom in this enzyme.

Protein farnesyltransferase (FTase) 1 catalyzes the transfer of the 15-carbon isoprenoid from farnesyl diphosphate (FPP) to a conserved cysteine residue of protein acceptors containing the carboxyl-terminal CaaX motif, where C is cysteine, a is generally an aliphatic amino acid, and X is methionine, serine, glutamine, or alanine (1,2). A number of farnesylated proteins have been identified in eukaryotes, including Ras proteins, nuclear lamins, and at least two ␥ subunits of heterotrimeric G proteins (3)(4)(5). Through the hydrophobic nature of the farnesyl group, this modification mediates protein-membrane, and possibly protein-protein, interactions that play important roles in proper subcellular localization and function of these proteins, most of which are involved in signal transduction.
The properties of FTase are similar to those of a related enzyme, protein geranylgeranyltransferase Type I (GGTase I), which also recognizes and prenylates proteins with the carboxyl-terminal CaaX motif. However, GGTase I transfers the geranylgeranyl isoprenoid and prefers proteins with leucine at the X position of the CaaX motif (1,2). Known substrates for GGTase I include many Ras-related proteins, such as Rac and Rho, and most ␥ subunits of heterotrimeric G proteins (3,4). Since both FTase and GGTase I recognize proteins and short peptides containing appropriate CaaX motifs at their carboxyl terminus, they are also termed CaaX prenyltransferases (6). Geranylgeranyltransferase type II (GGTase II, also known as Rab geranylgeranyltransferase), on the other hand, catalyzes geranylgeranylation of proteins that terminate in CC or CXC motifs at their carboxyl terminus (6). Target proteins of GGTase II belong to the Rab protein family, members of which are involved in protein secretion and endocytosis (2,4).
Purified FTase is a heterodimer consisting of 48-kDa ␣ and 45-kDa ␤ subunit polypeptides (7). Like FTase, GGTase I is also composed of two nonidentical subunits, these being a 48-kDa ␣ and a 43-kDa ␤ subunit (8). cDNA clones for the subunits of both enzymes have been isolated from rat and human cDNA libraries (9 -12). Immunological evidence and sequence comparison showed that these two enzymes share the same ␣ subunit and that significant sequence similarity is found between the two ␤ subunits (12,13). Since both enzymes show specificity toward their protein and isoprenoid substrates, the unique ␤ subunits of these enzymes are thought to be responsible for the recognition and binding of both substrates. Consistent with this idea are studies showing that photoreactive isoprenoid analogues cross-link to the ␤ subunits of both FTase and GGTase I (11,14,15), and that both protein and peptide substrates are cross-linked to the ␤ subunit of FTase (16,17).
Both FTase and GGTase I are zinc metalloenzymes (8,16). Atomic absorption analysis of purified recombinant enzymes * This work was supported by National Institutes of Health Grant GM46372 and the Council for Tobacco Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Established Investigator of the American Heart Association. To whom correspondence should be addressed: Dept. of Molecular Cancer Biology, Duke University Medical Center, Research Dr., C303 LSRC, Durham, NC 27710-3686. Tel.: 919-613-8613; Fax: 919-613-8642; Email: pjc@galactose.mc.duke.edu. 1 The abbreviations used are: FTase, protein farnesyltransferase; GGTase, protein geranylgeranyltransferase; G protein, GTP-binding regulatory protein; FPP, farnesyl diphosphate; GGPP, geranylgeranyl diphosphate; CaaX, a sequence motif of proteins consisting of an invariant Cys residue fourth from the C terminus; Ras-CVLL, Ha-Ras protein with a Leu-for-Ser substitution at the C terminus; NEM, N-ethylmaleimide; DTT, dithiothreitol; PAGE, polyacrylamide gel electrophoresis; indicates that they contain one zinc atom per protein dimer (18,19). Prolonged dialysis against chelating reagents such as EDTA inactivates the enzymes in a fashion completely reversible by the addition of zinc, suggesting that the zinc is involved in catalytic activity of the enzyme (8,16,18,19). Additional evidence in this regard comes from cross-linking and direct binding studies that show that binding of the protein substrates to the enzymes is zinc-dependent (16,20). However, high affinity binding of isoprenoid substrates to zinc-depleted enzymes is similar to that of the native enzymes, indicating that the zinc is not required for binding of these substrates (16,20). Metal substitution studies show that cadmium can functionally substitute for zinc in both FTase and GGTase I (15,20,21), but the cadmium-substituted enzymes show altered substrate specificity (20,21). These results indicate that the metal atom in these enzymes is important in their interaction with substrates.
While it is clear that the zinc in CaaX prenyltransferases is critical for their activity, the precise role that this metal plays is not yet known. The zinc could be acting catalytically, for example by increasing the nucleophilicity of the cysteine residue in the CaaX motif of protein substrate and/or activating the diphosphate leaving group, or structurally, perhaps by stabilizing the heterodimeric form of the enzyme. In order to clarify the role of the zinc atom in FTase, we first set out to identify residues that might serve as the ligands of the metal. A common metal ligand in many zinc enzymes is the thiol group of cysteine residues (22,23). We used thiol modifying reagents to identify cysteine residues potentially important in FTase function. Through the use of selective chemical modification in conjunction with peptide sequencing and analysis by site-directed mutagenesis, we herein report the identification of an essential cysteine residue, Cys 299 in the ␤ subunit of FTase, that is critical for enzyme activity and, additionally, may serve as a zinc ligand in the enzyme.

EXPERIMENTAL PROCEDURES
Determination of Enzymatic Activity and Isoprenoid Binding-FTase activity was determined by quantitating the amount of 3 H transferred from [ 3 H]FPP into the Ha-Ras protein as described previously (24). The standard reaction mixture contained the following components in a final volume of 50 l: 50 mM Tris⅐HCl, pH 7.7, 20 mM KCl, 5 mM MgCl 2 , 5 M ZnCl 2 , 2 mM dithiothreitol (DTT), 8 M Ha-Ras, 1 M [ 3 H]FPP (8 Ci/mmol), and 50 ng of FTase. Assays were conducted for 15 min at 37°C, and the amount of trichloroacetic acid-precipitable [ 3 H]farnesylated Ha-Ras was determined by filter binding assay. GGTase I activity was also determined as described previously (8) using a reaction mixture that was the same as for FTase except that the protein and isoprenoid substrates were Ras-CVLL (Ha-Ras containing a Leu-for-Ser substitution at the C terminus) and [ (20). To compare the FPP binding ability of NEM-modified and unmodified FTases, both enzymes (ϳ13 nM) were incubated with increasing concentrations of [ 3 H]FPP (0 -80 nM) in 50 mM Tris⅐HCl, pH 7.7, 100 mM KCl, 5 M ZnCl 2 , 0.2% octyl ␤-glucopyranoside, 2 mM DTT in a final volume of 50 l at 0°C for 15 min, and bound FPP was determined as above.
Preparation of Native and Metal-depleted FTase and GGTase I-Except for the mutagenesis studies, which utilize bacterially produced enzymes (see below), the FTase and GGTase I used in this study were recombinant proteins produced by infection of Sf9 cells with recombinant baculoviruses and purified from cell extracts as described previously (18,19); both enzymes were Ͼ95% pure as judged by SDS-PAGE analysis. To prepare the apo forms of the two enzymes, the appropriate enzyme (about 1 mg) was dialyzed first for 24 h at 4°C against 1 liter of 20 mM Tris⅐HCl, pH 7.7, 1 mM DTT, and 5 mM EDTA and then for another 24 h against the same buffer except that the EDTA concentration was reduced to 1 mM. Removal of zinc from the dialyzed proteins was confirmed by assessing the zinc dependence of their activities (8).
Inactivation of FTase and GGTase I by Sulfhydryl-specific Reagents-Native and zinc-depleted FTase were chromatographed through G-50 Sephadex spin columns that had been equilibrated at 4°C with 20 mM Tris⅐HCl, pH 7.7, containing 5 M ZnCl 2 (for native enzyme) or 20 mM Tris⅐HCl, pH 7.7, containing 0.1 mM EDTA (for zinc-depleted enzyme), to remove DTT. The DTT-free enzymes were then diluted to a concentration of 14 M with 20 mM Tris⅐HCl, pH 7.7, in the presence or absence of 67 M FPP. After a 5-min preincubation at room temperature, the inactivation reactions were initiated by the addition of either 250 M N-ethylmaleimide (NEM) or 2 mM iodoacetamide. At various times, aliquots of the reaction mixture were withdrawn, and reactions were stopped by the addition of 10 mM DTT. For the zero time point, NEM or iodoacetamide was preincubated with DTT for 1 min before adding to the aliquot of the preincubated enzyme solution. Treatment of GGTase I by NEM was performed by the same procedure as that of To produce peptides for sequence analysis, lyophilized [ 3 H]NEMlabeled FTase (ϳ1 nmol) was dissolved in 8 M urea prepared in 0.4 M ammonium bicarbonate buffer, pH 8.0, carboxymethylated with iodoacetamide, diluted 4-fold with water, and digested with trypsin as described (25). Peptides were isolated by HPLC using a 250 ϫ 46-mm Vydac C18 reversed-phase column. The column was equilibrated with 10% acetonitrile and 0.1% trifluoroacetic acid. After a 5-min wash, a linear gradient was established to 60% acetonitrile over 75 min, and then to 100% acetonitrile in 10 min, followed by a 15-min wash with 100% acetonitrile; the trifluoroacetic acid was kept at a constant 0.1% throughout. Selected peaks were further purified by rechromatography on the same column but using a second solvent system. In this system, the column was equilibrated with 8% acetonitrile and 0.1% ammonium acetate, pH 5.8. After a 5-min wash, a linear gradient was established to 80% acetonitrile over 55 min, followed by a 15-min wash with 80% acetonitrile; the ammonium acetate was kept constant throughout. The eluent was monitored for absorbance at 214 nm, and 1-min (0.5-ml) fractions were collected. Samples (25 l) of the fractions were removed for radioactivity determination by scintillation spectroscopy. Fractions containing isolated peptides were pooled, solvent was reduced under reduced pressure, and the peptides were subjected to sequence analysis. One of the peptides purified by HPLC (designated under "Results" as that from peak III) was further subjected to cyanogen bromide cleavage in 70% formic acid as described (26) prior to rechromatography in the acetonitrile/ammonium acetate system.
Bacterial Expression System for FTase Production-To express heterodimeric rat FTase in Escherichia coli, the ␣ and ␤ subunits of FTase were subcloned into two different expression vectors in which expression of both cDNAs was controlled by T7 promoters. Restriction sites for NdeI and NcoI were generated at the start codons of the ␣ and ␤ subunits, respectively, by PCR. For the ␣ subunit, PCR primers designated FT␣-N (5Ј-AAAAGAATTCATATGGCGGCCACTGAGGGGGTC-GGGGA-3Ј) and FT␣-B (5Ј-CTGTCCCTGTACAAGACATAGGTGGG--3Ј), based on amino-terminal and internal sequences of the ␣ subunit of rat FTase, respectively, were used in a PCR reaction with 12.5 ng of plasmid pUC13-FTA encoding the ␣ subunit of rat FTase. The resulting PCR product was digested with EcoRI-BsrGI and ligated back into the EcoRI-BsrGI-digested pUC13-FTA. For the ␤ subunit, PCR primers designated FT␤-N (5Ј-AAAAAAGCTTCCATGGCTTCTTCGACTTCCT-TCACCTATT-3Ј) and FT␤-B (5Ј-CTGTGCAGGATCCAGTAGCAGAGC-CA-3Ј) based on amino-terminal and internal sequences of the ␤ subunit of rat FTase, respectively, were used in a PCR reaction with 12.5 ng of plasmid pUC13-FTB encoding the ␤ subunit of rat FTase. The resulting PCR product was digested with HindIII-BamHI and ligated back into the HindIII-BamHI-digested pUC13-FTB. These resulting plasmids, designated as pUC13-PCRFT␣ and pUC13-PCRFT␤ for the ␣ and ␤ subunits of FTase, respectively, were confirmed by DNA sequencing. A NdeI-HindIII fragment from pUC13-PCRFT␣ that contained the coding region of the ␣ subunit of FTase was ligated into a NdeI-HindIII-digested pET28a vector (Novagen, Madison, WI), and a NcoI fragment from pUC13-PCRFT␤ that contained the coding region of the ␤ subunit of FTase was ligated into a NcoI-digested pAlter-Ex2 vector (Promega, Madison, WI). The resulting plasmids, designated as pET28a-FT␣ and pAlter-Ex2-FT␤, were co-transformed into BL21(DE3) E. coli cells; expression of the pET28a-FT␣ plasmid yielded the ␣ subunit of FTase with an N-terminal extension containing a hexahistidine motif.
A 3-liter culture of E. coli BL21(DE3) cells harboring the appropriate FTase plasmids was grown to an optical density of 0.6, and FTase expression was induced by the addition of 0.4 mM isopropyl-␤-D-thiogalactoside; 0.5 mM ZnSO 4 was also added at this time to ensure availability of zinc. After a 4-h induction at 37°C, cells were harvested by centrifugation and washed once with phosphate-buffered saline. The cell pellet was resuspended in 40 ml of buffer containing 20 mM Tris⅐HCl, pH 7.7, 5 mM imidazole, 1 mM ␤-mercaptoethanol, and a protease inhibitor mix (27), and the cells were lysed by two passes through a French press. The sample was diluted to 80 ml, NaCl was added to a final concentration of 300 mM, and the extract was centrifuged at 30,000 ϫ g for 1 h at 4°C. The supernatant was subjected to Ni 2ϩ -NTA affinity chromatography under the conditions recommended by the manufacturer (Qiagen, Chatsworth, CA). The resin was eluted with a 5-150 mM imidazole gradient in 20 mM Tris⅐HCl, pH 7.7, 300 mM NaCl, 1 mM ␤-mercaptoethanol, and the protease inhibitor mix. Peak fractions, analyzed by assessing their FTase activity, were pooled and dialyzed twice for 1 h each against 1 liter each of 50 mM Tris⅐HCl, pH 7.7, 5 M ZnCl 2 , and 1 mM DTT. The dialyzed enzyme was applied to a Q-HP 16/10 high resolution anion exchange column (Pharmacia LKB Biotechnology Inc., Uppsala, Sweden) equilibrated in buffer containing 50 mM Tris⅐HCl, pH 7.7, 5 M ZnCl 2 , and 1 mM DTT. The column was washed with the equilibration buffer and then eluted with a linear gradient to 500 mM NaCl in the same buffer. Peak fractions were pooled and concentrated with buffer exchange into the starting buffer using a CentriPrep 30 concentrator (Amicon, Beverly, MA) to a protein concentration of Ͼ1 mg/ml, flash-frozen in aliquots, and stored at Ϫ80°C until use. For the mutant enzymes, the purification procedure was the same except that the Mono Q HR 5/5 column (Pharmacia) was used instead of the Q-HP column, and peak fractions were determined by immunoblot analysis and FPP binding rather than by enzymatic activity.
Determination of Zinc Binding Activity-An aliquot of each of the three purified mutant FTases (ϳ20 pmol, based on FPP binding) was diluted to 60 l with 20 mM Tris⅐HCl, pH 7.7, 1 mM DTT, and 10 M EDTA. Samples were chromatographed through G-50 Sephadex spin columns that had been equilibrated at 4°C with 20 mM Tris⅐HCl, pH 7.7, 1 mM DTT, 100 mM NaCl, and 1 M EDTA and incubated at 30°C for 1 h. A 14 M solution of 65 ZnCl 2 (30 Ci/mmol), prepared by diluting the stock reagent with 200 mM Tris⅐HCl, pH 8.5, was added to each of the enzyme samples so that the final concentration of 65 ZnCl 2 was 2 M. Following an incubation for 3 h at room temperature and overnight at 0°C, the incubation mixtures were supplemented with FPP to a final concentration of 5 M (to prevent dissociation of bound zinc) and held for 15 min at room temperature. One-fifth of each reaction mixture was loaded onto a 10% continuous native polyacrylamide gel prepared in 90 mM Tris⅐HCl (pH 8.2), 80 mM boric acid, and 2.5 mM EDTA, and electrophoresis was performed in 90 mM Tris⅐HCl, pH 8.2, 80 mM boric acid, and 2.5 mM EDTA for 1 h at 200 V. The gel was washed with electrophoresis buffer for 15 min, wrapped in cellophane, and exposed to film.
Miscellaneous Methods-Site-specific mutations were introduced into FTase using the Altered Sites Mutagenesis system (Promega, Madison, WI). Three mutant enzymes were produced, these being FT␣C341A with a mutation in the ␣ subunit of FTase and FT␤C299A and FT␤C299S with mutations in their ␤ subunits. Each of the mutant cDNAs was mapped with restriction enzymes, and the mutation was confirmed by DNA sequence analysis. Each was then subcloned into the expression vector pET28a (for the ␣ subunit mutant) or pAlter-Ex2 (for the ␤ subunit mutants). Standard molecular biology methods for DNA sequencing and manipulation were used (28). DNA sequencing was performed using the Sequenase version 2.0 DNA sequencing kit (U.S. Biochemical Corp.), except that the DNA plasmid pUC13-PCRFT␣ was sequenced at the University of North Carolina (Chapel Hill) Automated DNA Sequencing Facility on a model 373A DNA Sequencer (Applied Biosystems, Foster City, CA) using the Taq DyeDeoxy TM Terminator Cycle Sequencing Kit (Applied Biosystems). Radiolabeled peptides were sequenced by automated Edman degradation on an amino acid sequencer (Applied Biosystems) by C. R. Moomaw at the University of Texas Southwestern Medical Center (Dallas). SDS-PAGE and immunoblot analysis were performed as described previously (18). Antiserum p538, a polyclonal rabbit antibody directed against a peptide corresponding to residues 359 -370 of rat FTase-␣, was used to detect the ␣ subunit of FTase; antiserum p121, a polyclonal rabbit antibody directed against holo-FTase, was used to detect the ␤ subunit of FTase. Both antisera were affinity-purified on Protein A-Sepharose prior to use. Protein concentration was routinely analyzed by the Bradford method using a commercial dye (Bio-Rad) and bovine serum albumin as standard.
Materials-Recombinant Ha-Ras and Ras-CVLL proteins were purified from bacterial expression systems as described (24)

Inactivation of Zinc-depleted FTase by Sulfhydryl-specific
Reagents-To determine whether cysteine residues play an important role in FTase activity, cysteine-specific alkylating reagents were tested for their ability to inactivate the enzyme. Treatment of zinc-depleted (apo-) FTase with 250 M NEM resulted in a loss of 90% of its activity within 2 min and almost total loss of activity after 10 min. In contrast, the holoenzyme was quite resistant to inactivation by the same treatment (Fig.  1A). The addition of Zn 2ϩ to apo-FTase completely protected it from inactivation by NEM (Fig. 1B). Interestingly, preincubation of apo-FTase with FPP, the isoprenoid substrate of FTase, was also found to protect against NEM inactivation (Fig. 1A). When the other CaaX prenyltransferase, GGTase I, was treated with 250 M NEM, the apoenzyme was inactivated in a time-dependent manner similar to that of apo-FTase, and preincubation of apo-GGTase I with GGPP also prevented inactivation of the enzyme (data not shown).
To confirm these findings, we repeated these studies with another cysteine-specific alkylating reagent, iodoacetamide. Again, treatment of apo-FTase with iodoacetamide resulted in essentially complete inactivation in a time-dependent fashion, whereas the holoenzyme retained its activity even after a 1-h modification (Fig. 1C). As observed with NEM inactivation, inclusion of FPP prevented inactivation of apo-FTase by iodoacetamide (data not shown). Taken together, these results indicated the presence of specific cysteine residue(s) required for the activity of the CaaX prenyltransferases and, furthermore, that such cysteine residue(s) might be ligands that coordinate the zinc atom. We therefore designed a strategy, using FTase as the model enzyme, to identify the cysteine residue(s) involved and assess their influence on enzyme activity.
Radiolabeling of Essential Cysteine Residues in FTase-The strategy developed to identify the cysteine residues in FTase responsible for the above observed effects involved selective labeling of the enzyme with [ 3 H]NEM. In a preliminary assess-ment of the viability of the approach, we examined whether the NEM modification of FTase was a homogenous event. We first assessed whether NEM-modified holo-FTase that was subsequently subjected to zinc depletion behaved in a similar manner as unmodified FTase. We found that the NEM-modified enzyme was catalytically inactive after zinc depletion and that the activity could be completely restored by the addition of zinc to the enzyme (data not shown). These findings indicated that holo-NEM-modified FTase was quite similar to its unmodified counterpart, the only difference detected being a slightly reduced catalytic ability of the NEM-modified enzyme (see Fig.  1A). Since this reduction in activity is not seen with the enzyme modified by iodoacetamide (Fig. 1, compare A and C), it is likely that the reduced activity of the NEM-modified enzyme compared with that modified by iodoacetamide is due to some type of steric hindrance by the larger size of the former alkylating agent. In support of this assessment that NEM modification of holo-FTase did not significantly affect the protein's structure or function, the FPP binding ability of both holo-FTase and holo-NEM-modified FTase was found to be essentially identical; analysis of the binding data showed that the stoichiometry of FPP binding by NEM-modified FTase was only ϳ10% less than that of the unmodified enzyme (data not shown).
The results described above suggested a selective radiolabeling strategy that we felt would lead to the identification of the critical cysteine residue(s) required for FTase activity. Native FTase was first treated with unlabeled NEM to alkylate accessible cysteine residues, and then the enzyme so modified was treated with EDTA to remove zinc. This apo-NEM-modified FTase was then further treated with [ 3 H]NEM to label the cysteine residues exposed upon the removal of zinc. As a control, NEM-modified FTase not subjected to EDTA treatment was also treated with [ 3 H]NEM. A diagram summarizing this selective labeling scheme is shown in Fig. 2. The stoichiometry of [ 3 H]NEM incorporation into FTase subjected to these two treatments was determined by gel slice assay. For holo-FTase, approximately 0.1 and 0.3 mol of [ 3 H]NEM/mol of protein was incorporated into the ␣ and ␤ subunits, respectively, while incorporation into apo-FTase was substantially greater, with stoichiometries of 1.0 and 3.6 mol of [ 3 H]NEM/mol of protein into the ␣ and ␤ subunits, respectively (Fig. 3). Since the total incorporation of radiolabel into holo-FTase was only one-tenth of that into apo-FTase, it seemed likely that there were indeed specific cysteine residues protected by zinc in the enzyme.
The modified enzymes were subjected to trypsin digestion, and the peptides produced were isolated by reverse-phase HPLC. The comparison of the elution patterns of the peptides derived from the two samples is shown in Fig. 4, A and B, and revealed that the chromatograms for both holo-and apo-FTase were essentially indistinguishable. However, analysis of the radioactivity in the profile revealed that four major radioactive peaks were obtained from apo-FTase but not from holo-FTase (Fig. 4C). Since this profile was still somewhat complex, the material eluting in each of these peaks was collected and further separated by reverse-phase HPLC using a second solvent system. Rechromatography of the material in peak IV (the major radiolabeled species) yielded one major radiolabeled product that eluted from the second chromatographic step as a single, well defined peak of both absorbance and radioactivity (Fig. 5, A and B). Sequencing of this peptide yielded a radiolabeled cysteine derivative that was released at cycle 5 (Fig. 5C). This labeled peptide, whose sequence is shown in Fig. 5C, is in the ␤ subunit of FTase. The cysteine residue in this peptide is conserved in all known FTases and is also conserved in the other two known protein prenyltransferases, GGTase I and GGTase II (Table I).
In all, four 3 H-labeled polypeptides, i.e. one from each of the peaks in the first HPLC profile, were obtained and identified by sequence analysis (Table I). Only one of the additional three peptides appeared interesting. This peptide, whose sequence is shown in Table I, was isolated from the peak I material in the original chromatography. The labeled cysteine residue identified was in the ␣ subunit of FTase in a position that is conserved in mammalian FTase and GGTase I ␣ subunits, but not in their S. cerevisiae counterparts. The amount of labeling in this peptide was about 40% of that in the major peptide isolated from peak IV. We were also able to identify labeled cysteine residues in the radiolabeled peptides purified from the material in peak II and peak III, and these were all in the ␤ subunit of the enzyme. However, these residues were not conserved among CaaX prenyltransferases, and additionally, the amounts of their labeling were much lower (ϳ10 -30%) than that of Cys 299 in the ␤ subunit. The sequences of the FTase peptides containing cysteine residues labeled in this strategy and the comparison of their sequences with the corresponding peptides present in other protein prenyltransferases are summarized in Table I.
Site-directed Mutagenesis of Cysteine Residues-To determine if either or both of the two conserved cysteine residues identified through the selective labeling approach (i.e. Cys 299 in the ␤ subunit and Cys 341 in the ␣ subunit) were in fact essential for FTase activity, site-directed mutagenesis and analysis of the mutant enzymes were undertaken. Cys 299 , the major [ 3 H]NEM-labeled cysteine residue that is in the ␤ subunit of FTase, was changed to both alanine and serine; these mutants were designated FT␤C299A and FT␤C299S, respectively. Cys 341 in the ␣ subunit was changed to alanine to produce mutant FT␣C341A. These three mutant proteins were produced by expression in E. coli and purified by chromatography over the Ni 2ϩ -NTA affinity and Mono Q resins as described under "Experimental Procedures." Fractions obtained from the  Fig. 2 were subjected to SDS-PAGE, and the ␣ and ␤ subunits were visualized by staining with Coomassie Blue. The region of the gel corresponding to each subunit was excised and subjected to an overnight digestion in an aqueous solution containing 20% H 2 O 2 and 20% HClO 4 at 60°C, and the radioactivity in each sample was determined. The results from three independent experiments were pooled, and means and standard deviations are shown.

FIG. 4. Analysis of the tryptic digests of [ 3 H]NEM-labeled FTase preparations. [ 3 H]NEM-labeled apo-FTase (1 nmol, panels A
and C) and holo-FTase (0.5 nmol, panels B and C), obtained from the labeling approach summarized in Fig. 2, were subjected to trypsin digestion, and resultant digests were processed by reverse-phase HPLC using the acetonitrile/trifluoroacetic acid gradient (dashed line) described under "Experimental Procedures." The eluent was monitored at 214 nm to detect peptides produced (panels A and B), and aliquots of collected fractions were analyzed for radioactivity (panel C). The four major peaks of radioactivity derived from the digest of apo-FTase (E) are designated I-IV; negligible radioactivity was associated with the digest of the holoenzyme (q).
Mono Q chromatographic step for all three mutant enzymes were assayed both for FTase activity and FPP binding, and the results are shown in Fig. 6. The ␣ subunit mutant, FT␣C341A, exhibited properties essentially indistinguishable from the wild-type enzyme and retained both enzymatic activity and the ability to bind FPP. In contrast, both of the proteins that contained the Cys 299 mutations in the ␤ subunits lost essentially all of their catalytic activity. However, in spite of their inability to catalyze farnesyl transfer, both mutant enzymes retained the ability to bind FPP.
To directly compare the activity of each mutant to the wildtype enzyme, the latter was produced in E. coli and purified in a similar manner. As with the mutant enzyme, the wild-type protein eluted from the anion exchange column in a peak centered at 0.3 M NaCl (data not shown). For each of the four enzymes (i.e. the wild-type and the three mutants), the peak fraction from the anion exchange column was selected for analysis of enzyme levels and for catalytic and FPP binding activities. Immunoblot analysis confirmed that both the ␣ and ␤ subunits were present in all four purified enzymes (Fig. 7). The ␣ subunits of the enzymes contained the His-tag fusion at their N termini and hence migrated at an apparent molecular mass of 57 kDa in this system. This analysis indicated that all four enzymes were produced in E. coli as heterodimers, since the enzyme could be purified through the use of the affinity tag on the ␣ subunit, and the ␤ subunit could be co-purified. Additionally, the fact that all four enzymes retained the ability to bind FPP indicated that all of these proteins had folded correctly. A comparison of the enzymatic and FPP binding activities of each of the mutant enzymes to wild-type FTase is shown in Table II. The turnover number determined for the wild-type enzyme of 2.6 min Ϫ1 agrees well with previous studies on this enzyme (11,21). As noted above, the FT␣C341A variant displayed similar enzymatic activities as the wild-type enzyme; in fact, this quantitative analysis indicated that it has a slightly higher turnover number, although this is probably not significant. We were able to detect very low levels of enzymatic activity in mutant FT␤C299A, about 2% that of the wild-type enzyme. No residual enzymatic activity in mutant FT␤C299S could be detected. Additionally, increasing the zinc concentration in the assay buffer did not restore FTase activity to either of the two enzymes containing the Cys 299 mutations (data not shown).
The results detailed above strongly suggested that the loss in activity of the two ␤-Cys 299 mutants of FTase was due to an inability to coordinate the zinc atom required for activity of this enzyme. To directly examine this, we developed a 65 Zn binding assay for the enzyme (see "Experimental Procedures"). Important points in this assay are the binding of FPP to the enzymes after the zinc exchange reaction to prevent dissociation of bound metal and the use of a native gel system in which the FTase dimer migrates as an intact species. 2 The results of this analysis, shown in Fig. 8, clearly indicate that the FT␣C341A mutant retains the ability to bind zinc but that the two enzymes containing the Cys 299 mutations had lost this ability. Thus, the loss of FTase activity in the two FTase forms containing the Cys 299 mutations correlates with a loss of zinc binding by these enzymes.  4C) were pooled, lyophilized, and subjected to chromatography by reverse-phase HPLC using the acetonitrile/ammonium acetate gradient (dashed line) described under "Experimental Procedures." The eluent was monitored at 214 nm to detect peptides (A), and aliquots of collected fractions were analyzed for radioactivity (B). C, the radiolabeled peptide that eluted at 53 min was subjected to sequence analysis by automated Edman degradation (identified residues are listed in single letter code), and an aliquot from each cycle was also analyzed for radioactivity (closed bars).

TABLE I Sequence analysis of [ 3 H]NEM-labeled peptides derived from apo-FTase and sequence comparison with other protein prenyltransferases
The underlined cysteine residues are those that incorporated label from [ 3 H]NEM. Sequence alignment of the ␤ subunits is adapted from Omer et al. (11) and Zhang et al. (12); sequence alignment of the ␣ subunits is adapted from Andres et al. (31).

DISCUSSION
We have used a selective chemical modification strategy to identify a cysteine residue in FTase that is critical for catalytic activity of the enzyme. This strategy was designed to identify cysteine residues that we felt would be involved in zinc coordination in this metalloenzyme and was initiated upon finding that cysteine-specific alkylating agents inactivated FTase only after zinc was removed from the enzyme. This predominant residue labeled in this strategy was Cys 299 in the ␤ subunit of the enzyme, and this residue was found to be conserved in all known protein prenyltransferases. Indeed, replacement of Cys 299 with either serine or alanine resulted in production of an altered enzyme that had lost essentially all catalytic activity. The finding that these two mutant forms of FTase were still produced as heterodimers and retained normal binding of FPP in spite of the loss in enzymatic function indicated that Cys 299 is truly essential for FTase activity and not simply for correct folding of the enzyme. Additionally, since labeling of Cys 299 by [ 3 H]NEM only occurred after removal of the zinc and both Cys 299 mutants lost the ability to bind this metal, we feel it is likely that this residue is directly involved in binding of zinc by the enzyme and that it is quite possibly one of the metal ligands.
Although several other cysteine residues were identified through this labeling strategy, they were modified to a much lesser degree than Cys 299 and were also less conserved in the protein prenyltransferase family. Nonetheless, we did select one of these, that being Cys 341 in the ␣ subunit of the enzyme, for mutagenesis, since its labeling intensity was second only to Cys 299 in the ␤ subunit and it was conserved among mammalian, although not yeast, CaaX prenyltransferases. However, expression of the FT␣C341A mutant produced an enzyme that was fully active in both FPP binding and catalytic activity, indicating that this is not an essential residue in the enzyme. It , and FT␤C299S (C) were produced by bacterial expression, and extracts subjected to affinity chromatography on immobilized nickel resin as described under "Experimental Procedures." The peak fractions from the affinity chromatography step for each enzyme were pooled and subjected to anion-exchange chromatography on Mono-Q resin; proteins were eluted with an increasing gradient of NaCl (dashed line). Aliquots of each fraction were assayed for FTase activity (q) and FPP binding (f) as described under "Experimental Procedures. "   FIG. 7. Analysis of the final preparations obtained from purification of wild-type and mutant FTases produced in E. coli. An aliquot from the peak fraction obtained from the anion-exchange chromatographic step of each of the three mutant enzymes (see Fig. 6) as well as the wild-type enzyme (WT, see "Experimental Procedures") was subjected to SDS-PAGE. The proteins were transferred to nitrocellulose and subjected to immunoblot analysis using a mixture of anti-␣ (p538 at 1.5 g/ml) and anti-␤ (p121 at 1.0 g/ml) antisera as described under "Experimental Procedures." The migration positions of the ␣ and ␤ subunits are indicated. FTases. An aliquot from the peak fraction obtained from the anion exchange chromatographic step of each of the three mutant enzymes (see Fig. 6) was incubated in the presence of 1 M EDTA at 30°C for 1 h to promote the dissociation of zinc from the enzyme. Following incubation with 2 M 65 ZnCl 2 and processing as described under "Experimental Procedures," the proteins were subjected to gel electrophoresis using the nondissociating system described under "Experimental Procedures." The gel was exposed to film for 3 days at 4°C with an intensifying screen. The migration position of FTase in this gel system is indicated. seems possible that its labeling is due to some type of conformational change in FTase that occurs upon NEM modification of Cys 299 that might result in an increased accessibility of this and other cysteine residues for modification. In this regard, the ability to incorporate [ 3 H]NEM into Cys 364 and Cys 366 in the ␤ subunit after zinc removal may be significant, although the incorporation stoichiometry was quite low. Specifically, it is intriguing that these two Cys residues are juxtaposed to a histidine residue, His 362 , that is conserved in all ␤ subunits of the protein prenyltransferase family. Moreover, we have found that diethylpyrocarbonate, a histidine-specific modification reagent, can inactivate apo-FTase much more rapidly than holo-FTase. 3 This can be viewed as preliminary evidence for the existence of a histidine residue involved in zinc coordination; histidine residues are in fact the most common zinc ligands found in zinc metalloenzymes (22). Hence, His 362 in the ␤ subunit may also be one of the zinc ligands in FTase and other protein prenyltransferases.
In this thiol-specific modification study, binding of isoprenoid substrates to both FTase and GGTase I protected their apoenzyme forms from inactivation. This observation is consistent with the finding that binding of GGPP to GGTase I prevented the inactivation of the enzyme by metal chelating reagents (20). It seems likely that the binding site of the isoprenoid substrate is close to that of the zinc atom in these enzymes such that binding of this substrate physically blocks access of the alkylating reagent to the zinc binding site. Alternatively, binding of the isoprenoid substrate could induce a conformational change in the enzymes that results in a shielding of the zinc from molecules as large as the alkylating agents employed. In this regard, a conformational change in FTase upon FPP binding has been observed by fluorescence analysis (29).
The exact function of the critical cysteine residue in FTase-␤ identified in this study, Cys 299 -FT␤, in catalysis by the enzyme is not yet known. Cysteine residues have been identified as essential components of the active site of many different types of enzymes (30). In addition to their role as metal ligands in many metalloenzymes (22), they can also function directly in catalysis both as nucleophiles and in a general acid/base capacity, since the thiol group can be readily protonated or deprotonated under physiological conditions (30). However, as noted above, our results support the idea that this residue is involved in the coordination of the zinc atom in FTase. If this is the case, the loss in activity observed in the FT␤C299A and FT␤C299S mutants may be due primarily to a reduced binding of zinc in the mutant enzymes. While this hypothesis is the one that is most consistent with the data, we cannot yet exclude the possibility that Cys 299 is not actually a zinc ligand but rather is adjacent to the metal, where it plays one of the other above noted roles in catalytic function of the enzyme. A formal answer to this question will await detailed structural information on the enzyme; such information will be required to design mechanistic studies that would provide details of the precise role of this cysteine residue in FTase function.