Mutational Analysis of Conserved Residues of the β-Subunit of Human Farnesyl:Protein Transferase*

The roles of 11 conserved amino acids of the β-subunit of human farnesyl:protein transferase (FTase) were examined by performing kinetic and biochemical analyses of site-directed mutants. This biochemical information along with the x-ray crystal structure of rat FTase indicates that residues His-248, Arg-291, Lys-294, and Trp-303 are involved with binding and utilization of the substrate farnesyl diphosphate. Our data confirm structural evidence that amino acids Cys-299, Asp-297, and His-362 are ligands for the essential Zn2+ ion and suggest that Asp-359 may also play a role in Zn2+ binding. Additionally, we demonstrate that Arg-202 is important for binding the essential C-terminal carboxylate of the protein substrate.

Farnesyl:protein transferase (FTase) 1 catalyzes thioether bond formation between farnesyl, from farnesyl diphosphate (FPP), and the sulfur atom of a cysteine residue near the C terminus of its protein substrate (1,2). The protein substrate cysteine residue is located within a C-terminal motif called a CAAX box in which C is the cysteine that is S-farnesylated, A is often an aliphatic amino acid, and X is methionine, serine, glutamine, cysteine, or alanine (1,3,4). S-Farnesylation is the first step by which a number of proteins, including all forms of the Ras proto-oncoprotein, are post-translationally lipid-modified, facilitating membrane association and in some cases protein-protein interaction (2). Membrane association is required for Ras function, thus inhibitors of FTase have been proposed as antitumor agents (5). FTase inhibitors (FTIs) have been shown to inhibit anchorage-independent growth of ras-transformed rodent fibroblasts and human tumor cell lines (6 -8). In addition, FTIs have shown antitumor effects in rodents (6,9).
FTase kinetically proceeds through an ordered, sequential mechanism in which FPP is the first substrate bound (10,11). Subsequent protein substrate binding to FTase depends on an enzyme-bound Zn 2ϩ ion (12,13). Recent spectral data indicate that upon formation of the FTase⅐FPP⅐CAAX ternary complex, the cysteine thiol of the CAAX sequence interacts directly with Zn 2ϩ , possibly forming a thiolate anion (14). In addition to the enzyme-bound Zn 2ϩ , FTase activity requires millimolar Mg 2ϩ (12).
Two additional prenyl:protein transferases (PTases) have been described that catalyze reactions similar to that of FTase. Geranylgeranyl:protein transferase type I (GGTase-I) geranylgeranylates protein substrates with a C-terminal CAAX motif in which X is usually leucine (3,15). Geranylgeranyl: protein transferase type II (GGTase-II, also known as Rab geranylgeranyltransferase) geranylgeranylates both cysteine residues in protein substrates with Cys-Cys, Cys-Xaa-Cys, and Cys-Cys-Xaa-Xaa C-terminal motifs (3,16). PTases have been found in yeast, mammals, and plants and appear to be ubiquitously expressed in eukaryotes (2,17).
The catalytic moiety of all PTases purified to date are ␣␤ heterodimers, although GGTase-II requires a third subunit for protein substrate presentation (15,16,18). The ␣-subunits of FTase and GGTase-I are identical and show similarity to the corresponding subunit of GGTase-II (19 -21). The ␤-subunits for each of these enzymes are distinct, but have overall amino acid similarity of approximately 30% (20,21). In cross-linking studies, substrate proteins and photoactivatable analogs of FPP and CAAX peptides have been shown to bind to the ␤subunit of FTase (18,22,23). Similarly, photoactivatable geranylgeranyl diphosphate analogs can be cross-linked to the ␤subunit of GGTase-I (24,25). The recently published x-ray crystal structure of rat FTase (Ͼ93% amino acid sequence identity to human FTase) identified three conserved amino acids of the ␤-subunit as ligands for the essential Zn 2ϩ ion (26). Thus the ␤-subunit appears to play key roles in substrate binding and presumably in catalysis.
In this communication we have examined the role of the human FTase ␤-subunit by analyzing site-directed mutants of amino acids Arg-202, His-248, Cys-254, Arg-291, Lys-294, Asp-297, Cys-299, Tyr-300, Trp-303, Asp-359, and His-362. These residues are conserved among the ␤-subunits of all PTase enzymes identified to date and are near the essential Zn 2ϩ ion in rat FTase (26). We present biochemical evidence indicating the involvement of these amino acids in substrate interaction, zinc binding, and catalysis. Utilizing these results and the rat FTase crystal structure, we propose that ␤-subunit residues Arg-291, Lys-294, Trp-303, and possibly His-248 are involved in FPP binding. Our data suggest that ␤-subunit residue Arg-202 interacts with the C-terminal carboxylate of the CAAX substrate. We confirm that ␤-subunit residues Asp-297, Cys-299, and His-362 are Zn 2ϩ ligands and present data indicating that Asp-359 may also have an effect on Zn 2ϩ binding. * 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.
ʈ To whom correspondence should be addressed:  (22) except that it has a bacteriophage T7 promoter driving expression of the two subunits of FTase. The coding sequences for the ␤and ␣-subunits of human FTase were cloned in tandem into BamHI ϩ HindIII cleaved pT5T downstream of the T7 promoter (27). The sequence of the insert is shown in Sequence 1. Recombinant, human FTase made from this strain has the epitope Glu-Glu-Phe attached to the C terminus of the ␤-subunit to facilitate purification (22).
Site-directed mutations were introduced into pRD517 using the polymerase chain reaction (28). Mutationally silent restriction endonuclease sites were also introduced during the mutagenesis enabling us to screen clones for the presence of the new site. The amino acid changes made are abbreviated using the following notation. "␤R202A" indicates that ␤-subunit residue arginine 202 was changed to alanine. The corresponding heterodimeric human FTase is "␣␤R202A." All mutated regions were sequenced to ensure that only the intended changes were made. Bacterial growth and induction of FTase synthesis with isopropyl-1-thio-␤-D-galactopyranoside was performed as described (29). Purification of the recombinant FTases was similar to that described previously (29). Briefly, E. coli were lysed by sonication, and the debris was pelleted. The soluble protein was applied to a YL1/2 antibody column to bind to the Glu-Glu-Phe C-terminal epitope on the FTase ␤-subunit. Specifically bound protein was eluted at neutral pH with 5 mM Asp-Phe (Sigma). Approximately a 10-fold molar excess of ZnCl 2 was added to the enzyme before final purification by chromatography on Mono Q (Pharmacia Biotech Inc.). FTase was the only major absorbance (A 280 nm ) peak seen eluting from the Mono Q column. The final enzyme preparations were Ͼ70% pure as determined by SDS-polyacrylamide gel electrophoresis and subsequent Coomassie Blue staining. Since this purification method does not require following the enzyme activity, we could isolate mutant enzymes with little residual catalytic activity.
FPP Binding Studies-The dissociation constant for FPP was measured in 50 mM HEPES, pH 7.5, 5 mM MgCl 2 , 5 mM DTT, 10 M ZnCl 2 , 0.2% (w/v) N-octyl-␤-D-glucopyranoside utilizing a SPA (31,32). FTase was biotinylated using a biotinylation kit as described by the manufacturer. Biotinylated FTase (2-10 nM) was incubated with [ 3 H]FPP (0.27-1640 nM) and 5 g of neutral density streptavidin SPA beads for 30 min at room temperature after which the bound substrate was measured in a TopCount scintillation counter (Packard Instrument Co.). Scatchard plots (bound versus bound/free) of the resultant data were employed to determine the apparent K d [K d(app) ] values using the program k⅐CAT.
Zn 2ϩ Binding-Zinc binding to FTase was analyzed using a modification of the method of Fu et al. (33). 20 g of FTase in 100 l of buffer A (20 mM Tris-HCl, pH 8.0, 1 mM DTT, 100 mM NaCl, and 25 M EDTA) was incubated at 30°C for 1 h and then overnight on ice to deplete the enzyme of bound Zn 2ϩ . The Zn 2ϩ -depleted FTase was buffer-exchanged into buffer B (40 mM Tris-HCl, pH 8.5, 1 mM DTT, 1 M EDTA, 100 mM NaCl) using Bio-Spin-6 columns, and the protein concentration of the eluate was determined. 3.6 g of FTase was brought up to 18 l in buffer B, and 3 l of 143 M ZnCl 2 (either unlabeled ZnCl 2 in 0.1 N HCl or 65 ZnCl 2 ) was added, and the solution was incubated at room temperature for 3 h and then on ice overnight. Two l of 220 M FPP was added to the samples containing labeled 65 ZnCl 2 , and 2 l of 22 M [ 3 H]FPP was added to those containing unlabeled ZnCl 2 . After incubating at room temperature for 15 min, the samples were electrophoresed in nondenaturing 6% polyacrylamide gels using TBE as running buffer. After electrophoresis the gels were either washed 15-30 min in TBE running buffer ( 65 ZnCl 2 samples) or soaked in ENTENSIFY TM (samples containing [ 3 H]farnesyl diphosphate). The gels were dried between cellulose sheets and exposed to x-ray film.

Mutagenesis of the Human
FTase ␤-Subunit-Alignment of the ␤-subunits of PTases from various species identified a number of conserved amino acids (Fig. 1). Here we present the investigation of 11 conserved residues whose side chains contained aromatic rings or heteroatoms (Table I). These included residues that were shown to be near the essential Zn 2ϩ ion in the recently published crystal structure of rat FTase (26). To analyze the function of these amino acids, each was individually changed to alanine. Alanine was chosen because it has a small side chain and generally does not affect the secondary structure of a protein. E. coli expression strains were made for these mutated FTases. All except ␣␤Y300A were expressed at levels similar to wild type enzyme and upon purification were found to be ␣␤ heterodimers (data not shown). Further analysis of ␣␤Y300A FTase suggested that this enzyme was unstable; therefore, in its place we made ␣␤Y300F which was expressed well and was stable. We purified the 11 mutant FTases and have biochemically characterized them to elucidate the potential role(s) of these residues.
Summary of Kinetic Parameters and FPP Binding of Mutant Human FTases-Kinetic parameters (K mFPP , K mRas-CVLS , and k CAT ) and the apparent dissociation constant for FPP (FPP K d(app) ) for each enzyme were determined (Table I). Catalytic activity (k CAT ) of 5 of the 11 mutant human FTases was deficient relative to wild type enzyme. FTase ␣␤C299A was nearly inactive, whereas the catalytic activities of ␣␤D297A, ␣␤Y300F, ␣␤D359A, and ␣␤H362A were 5-100-fold lower than that of wild type FTase. Catalytic activities of the other mutant FTases were similar or somewhat higher than for wild type enzyme.
The crystal structure of rat FTase shows that amino acids Asp-␤297, Cys-␤299, and His-␤362 are ligands for the essential Zn 2ϩ ion; thus mutations in these residues would be expected to inhibit enzyme activity (26). The protein substrate K m values for ␣␤D297A and ␣␤H362A were elevated approximately 7-8fold relative to wild type FTase, but the FPP K m and K d (app) values were similar to wild type (Table I) Zn 2ϩ Binding by Mutant FTases-FTase tightly binds a single Zn 2ϩ ion that is essential for activity (12,13). Typical amino acid ligands for Zn 2ϩ include Asp, Glu, His, and Cys. We tested nine of the mutant enzymes for binding to radioactive 65 Zn 2ϩ using a native gel system described previously (33). The mutants tested included all 6 in which His, Cys, and Asp had been mutated (no Glu residues were mutated). As a control for enzyme integrity, we also examined, in parallel, [ 3 H]FPP binding in the same gel system. The results indicated that four of the mutant FTases, ␣␤D297A, ␣␤C299A, ␣␤D359A, and ␣␤H362A, were defective in Zn 2ϩ binding but retained the ability to bind [ 3 H]FPP (Fig. 2). Three of these amino acids, Asp-␤297, Cys-␤299, and His-␤362, appear as ligands in the crystal structure of rat FTase (26), with Cys-␤299 of rat FTase previously being shown to be a Zn 2ϩ ligand by biochemical and molecular techniques (33). Asp-␤359, although near the Zn 2ϩ ion, did not appear to be a ligand in the rat FTase structure.  (41) and rat GGTase-II (Rat GGII) (21), were aligned using the program CLUSTAL (42). Conserved residues (*) and similar residues (⅐) are noted. The numbers are for the amino acid residues in the ␤-subunit of human FTase. The conserved amino acids analyzed in this study are in bold. Only the most conserved, central regions are shown.

TABLE I Biochemical properties of mutant human FTases
The kinetic parameters were determined as described under "Experimental Procedures" (n ϭ 3-20). The kinetic parameters shown in Table I were determined in the presence of 10 M ZnCl 2 , thus mutant FTases that have a Zn 2ϩ binding defect may have had that defect partially masked. Since the method used to purify mutant and wild type FTases involved adding excess ZnCl 2 to the enzyme preparation before the final ion exchange step, enzymes that tightly bound Zn 2ϩ should have retained this metal ion and not require added Zn 2ϩ for activity. We, therefore, assayed the enzyme activity of wild type and several mutant FTases adding either 100 M EDTA (ϪZnCl 2 ), to scavenge residual free Zn 2ϩ , or 10 M ZnCl 2 (ϩZnCl 2 ). FTase ␣␤C299A was not tested since it is essentially inactive even in the presence of added ZnCl 2 ( Table I). The catalytic rates of wild type, ␣␤H248A, and ␣␤C254A FTases were found to be independent of added ZnCl 2 ( Table II). The activity of mutant FTases ␣␤D297A, ␣␤D359A, and ␣␤H362A, however, showed 10-fold or greater dependence on exogenously added ZnCl 2 , with the activity in the absence of ZnCl 2 being at the limit of detection (ϳ10 Ϫ5 mol of FPP incorporated/mol of FTase/s).
Interaction of Arg-␤202 with Farnesyl:Protein Transferase Inhibitors-FTase ␣␤R202A was found to have a K m value for Ras-CVLS Ͼ400-fold higher than that for wild type FTase with no defect in FPP K m or K d(app) ( Table I). An explanation for this may be that the negatively charged carboxylate of the CAAX substrate interacts with the positively charged guanidino side chain of the arginine residue. To test this hypothesis, we looked at the ability of four CAAX-competitive FTase inhibitors (FTI) to inhibit the activity of this mutant enzyme versus wild type ( Fig. 3 and Table III). The peptide FTI, CVFM, and the peptidomimetic FTI, L-739,750, both contain a cysteine thiol and a free C-terminal carboxylate. L-739,787 is similar to L-739,750 except that the C-terminal methionine sulfone has been reduced to methioninol. The FTI, L-745,631, is a tripeptide mimetic that contains a free thiol, but no carboxylate. All four inhibitors are members of FTI classes that have been shown to be competitive with the CAAX substrate, but not competitive with FPP (30,34,35). IC 50 values were determined for the four compounds under K m conditions for the respective enzymes (Table III). Under these conditions, the IC 50 values should be approximately twice the K i value for a competitive inhibitor (36). The two carboxylate-containing inhibitors, CVFM and L-739,750, had 100 -200-fold higher IC 50 values for ␣␤R202A as compared with wild type FTase. The IC 50 value for L-739,787, in which the carboxylate has been reduced to an alcohol, was less than 4-fold higher for ␣␤R202A as compared with wild type FTase. This indicates, at most, a slight difference in binding of L-739,787 to the two enzymes. The IC 50 values for the non-carboxylate FTI, L-745,631, were within 2-fold of one another indicating that it binds to both enzymes with similar affinity. Altogether these data are consistent with Arg-␤202 interacting with the free carboxylate of the CAAX competitive FTIs and by inference with the carboxylate of the CAAX substrate.

DISCUSSION
In this communication we have examined the roles of 11 conserved residues of the ␤-subunit of human FTase. From the crystal structure of the closely related rat FTase, all of these residues are in the vicinity of the essential Zn 2ϩ ion (26). Mutations in a number of these conserved residues affected the kinetic and biochemical properties of the enzyme (Table I). Mutations in Asp-297, Cys-299, Tyr-300, Asp-359, and His-362 of the FTase ␤-subunit decreased the catalytic efficiency 5-1000-fold.
Protein substrate K m values for human FTase ␣␤R202A were much higher than for wild type enzyme, but this mutant showed no defect in FPP binding or utilization (Table I). FTase ␣␤R202A was 100 -200-fold less sensitive than wild type enzyme to protein substrate competitive FTIs that contain a carboxylate while being similarly sensitive to protein substrate competitive FTIs lacking a free carboxylate (Table III). The structure of rat FTase shows that Arg-␤202 is near Tyr-␤361 and that both are in the vicinity of the Zn 2ϩ ion (26). Amino acid changes of Tyr-␤362 of yeast FTase, which is homologous to Tyr-␤361 of rat and human FTase, affects the X residue specificity of the CAAX substrate (37). The X residue of the protein substrate is the C-terminal amino acid containing a free carboxylate. Altogether, these data strongly suggest that Arg-␤202 interacts with the C-terminal carboxylate of the X residue of CAAX substrates. The interaction of the negatively Wild type  Table III.  (Table I). The evidence for an FPP binding defect for ␣␤W303A was less convincing as the K d(app) value was only 5-fold higher than wild type. Since FPP binds before the protein substrate (10,11), it may be that the elevated protein substrate K m values for these three mutant FTases are caused by the defect in FPP binding. In the crystal structure of rat FTase, Trp-␤303 is in a hydrophobic pocket that was proposed to be where the farnesyl chain binds (26). Arg-␤291 and Lys-␤294 are near the surface of this pocket. In the rat FTase structure paper it was postulated that Arg-␤202 might be the amino acid that interacts with the negatively charged phosphates of FPP. We believe, however, that our data with these mutant FTases indicate that Arg-␤291 and Lys-␤294 are likely to be residues that interact with these phosphates and that Arg-␤202 interacts with the CAAX carboxylate. An additional mutant FTase that had a defect in FPP interaction (K d(app) and K m ) was ␣␤H248A (Table I). His-␤248 is a residue near the Zn 2ϩ ion and nearby Arg-␤291 and Lys-␤294 in rat FTase (26).
The crystal structure of rat FTase identified Asp-␤297, Cys-␤299, and His-␤362 as Zn 2ϩ ligands (26). ␤C299A has also been shown to be a Zn 2ϩ ligand by biochemical and molecular techniques (33). Changing any one of these three residues or Asp-␤359 to alanine inhibited binding of 65 Zn 2ϩ (Fig. 2). All four mutant enzymes bound FPP with K d(app) values similar to wild type FTase (Table I). The ability of these mutants to bind FPP is consistent with previous results indicating that isoprenoid binding is Zn 2ϩ -independent (12). The residual catalytic activity of ␣␤D297A, ␣␤D359A, and ␣␤H362A (␣␤C299A was essentially inactive) was shown to be dependent on added Zn 2ϩ (Table II). While in the rat FTase crystal structure Asp-␤359 is too far away from the Zn 2ϩ ion to be a ligand, these data suggest that it may have an effect on Zn 2ϩ binding. Whether the effect is direct or indirect is presently unclear.
Besides mutations in residues that affected Zn 2ϩ ion binding, the only other mutant FTase that decreased k CAT was ␣␤Y300F (Table I). The k CAT of this enzyme was reduced about 5-fold relative to wild type enzyme. In the crystal structure of rat FTase, Tyr-␤300 appears to be close to the Zn 2ϩ ion (26). While mutation of this tyrosine to phenylalanine may simply cause a localized effect on the position of the two substrates, it may also indicate a catalytic role for this residue. In glutathione S-transferases the hydroxyl of a conserved tyrosine residue has been shown both structurally and biochemically to stabilize the catalytic thiolate anion (38,39). We are currently testing whether the proximity of Tyr-␤300 to the Zn 2ϩ indicates a similar role for this tyrosine in FTase.
The experiments presented here shed light on the roles of 11 conserved amino acids of the ␤-subunit of FTase in substrate binding, catalysis, and zinc binding. Additional studies will be needed to further define enzyme-substrate interactions and substrate specificity. While it appears that the essential Zn 2ϩ ion is involved in catalysis, it is not clear exactly how this occurs or what other contributions the enzyme might make to catalysis. Further structural, molecular, and biochemical studies should permit the answering of these questions.