Substrate Binding Is Required for Release of Product from Mammalian Protein Farnesyltransferase*

Protein farnesyltransferase (FTase) catalyzes the modification by a farnesyl lipid of Ras and several other key proteins involved in cellular regulation. Previous studies on this important enzyme have indicated that product dissociation is the rate-limiting step in cataly- sis. A detailed examination of this has now been performed, and the results provide surprising insights into the mechanism of the enzyme. Examination of the binding of a farnesylated peptide product to free enzyme revealed a binding affinity of (cid:59) 1 (cid:109) M . However, analysis of the product release step under single turnover condi- tions led to the surprising observation that the peptide product did not dissociate from the enzyme unless addi- tional substrate was provided. Once additional substrate was provided, the enzyme released the farnesy- lated peptide product with rates comparable with that of overall catalysis by FTase. Additionally, stable FTase- farnesylated product complexes were formed using Ras proteins as substrates, and these complexes also require additional substrate for product release. These data have major implications in both our understanding of overall mechanism of this enzyme and in design of inhibitors against this therapeutic target. Farnesylation in each was determined. Formation and Analysis of Enzyme-Product Complexes Substrates— The FTase-FPP complex was formed and isolated as men- tioned above. To form the product complex with H 6 -K-Ras or H 6 -H-Ras substrates, the FTase-FPP complex (250 n M ) was mixed with 50 n M H 6 -K-Ras or H 6 -H-Ras in Buffer C as above. To examine the dissocia- tion of product from this complex, either FPP or buffer (20 m M Tris-Cl, pH 8.0, 0.2% (cid:98) -octylglucoside) was added to give a total final concen- tration of 275 (cid:109) M of FPP in 10 (cid:109) l. The reaction mixture was incubated at 37 °C for 30 min, and then cytochrome c (30 (cid:109) g) was added as a carrier protein. A 50% slurry of nickel resin (10 (cid:109) l) equilibrated in Buffer A containing 250 m M NaCl was added, and the mixture was rocked for 30 min at 4 °C. The nickel resin was then isolated by pellet-ing in a microfuge for 30 s. The resin pellet was washed five times with Buffer A containing 250 m M NaCl, and bound proteins were eluted by incubation for 5 min at room temperature with a solution of 500 m M imidazole in the same buffer. The eluates were removed to a new tube containing SDS sample buffer, and the proteins were resolved on a 14% SDS-PAGE gel, followed by transfer to a polyvinylidene difluoride filter. These samples were then subjected to immunoblot using anti-FTase or anti-Ras antiserum (31). Visualization was by the alkaline phosphatase method using a commercial kit (Promega).

Protein farnesyltransferase (FTase) catalyzes the modification by a farnesyl lipid of Ras and several other key proteins involved in cellular regulation. Previous studies on this important enzyme have indicated that product dissociation is the rate-limiting step in catalysis. A detailed examination of this has now been performed, and the results provide surprising insights into the mechanism of the enzyme. Examination of the binding of a farnesylated peptide product to free enzyme revealed a binding affinity of ϳ1 M. However, analysis of the product release step under single turnover conditions led to the surprising observation that the peptide product did not dissociate from the enzyme unless additional substrate was provided. Once additional substrate was provided, the enzyme released the farnesylated peptide product with rates comparable with that of overall catalysis by FTase. Additionally, stable FTasefarnesylated product complexes were formed using Ras proteins as substrates, and these complexes also require additional substrate for product release. These data have major implications in both our understanding of overall mechanism of this enzyme and in design of inhibitors against this therapeutic target.
Protein farnesyltransferase (FTase) 1 catalyzes the S-farnesylation of a number of key cellular regulatory proteins. Farnesylation is directed by a C-terminal CAAX motif, where C is cysteine, A is usually an aliphatic residue, and X is typically methionine, serine, glutamine, or alanine (1,2). The farnesyl lipid is attached to the substrate protein via a thioether linkage to the cysteine residue using farnesyl diphosphate (FPP) as the prenyl donor. Among the substrates for FTase are the Ras family of proto-oncogenes, several ␥ subunits of heterotrimeric G proteins, and nuclear lamins (1,3). Farnesylation of these proteins is required for their proper membrane localization and activity. In the case of oncogenic forms of Ras proteins, the finding that farnesylation is required for expression of their transforming activity has led to FTase becoming an important target for anticancer drug design (4). Both in cell culture (5,6) and in animal models (7), specific inhibitors of FTase have been shown to reverse the oncogenic phenotype induced by mutationally activated Ras.
FTase has been purified to homogeneity from both rat and bovine brain by affinity purification on immobilized CAAX peptide substrates (8,9). The enzyme is a Zn 2ϩ metalloenzyme that consists of ␣ and ␤ subunits that migrate on SDS-PAGE with apparent molecular masses of 48 and 46 kDa, respectively (8). Both subunits of the enzyme have been cloned (10 -12), and their co-expression in either Sf9 (13) or E. coli (14) results in production of quantities of the enzyme required for detailed biochemical and structural analyses. Cross-linking experiments have provided strong evidence that the ␤ subunit is involved in recognition of both the isoprenoid and protein substrates (15)(16)(17), although there is also evidence that the ␣ subunit may participate (15,18). In addition to its bound Zn 2ϩ , FTase also requires Mg 2ϩ for activity. The Zn 2ϩ is involved coordinating the thiol of the peptide substrate in the ternary complex of enzyme-isoprenoid-peptide substrate (19) and thus is presumed to play a direct role in catalysis. The role of Mg 2ϩ is not yet known.
Steady-state kinetic studies indicate that mammalian FTase can bind either FPP or protein substrate independently, but product formation requires that the enzyme bind FPP first, giving FTase an ordered sequential mechanism (20 -22). The overall k cat under steady-state conditions is a relatively sluggish 1-3 min Ϫ1 for the mammalian enzymes (8,23), with product dissociation being rate determining in catalysis (22). The rate of the chemical step has been directly determined to be 17 s Ϫ1 through spectroscopic studies using enzyme containing a Co 2ϩ -for-Zn 2ϩ substitution (19). This spectroscopic study also revealed that the sulfur atom of the product thioether remains coordinated to the metal atom, an observation that may in part explain the slow release of product in steady-state turnover.
To gain better insight into the product dissociation step in the mechanism of mammalian FTase, we have performed an examination of the binding and the release of both peptide-and protein-derived products. The results of the study have major implications in regard to the mechanism of mammalian FTase and design of inhibitors targeting this enzyme.

EXPERIMENTAL PROCEDURES
Materials-The isoprenoid substrates FPP and its 3 H-labeled counterpart ([ 3 H]FPP) were purchased from American Radiolabeled Company (St. Louis, MO). Peptides were synthesized by solid-state methods and purified by reverse-phase HPLC as described (24). Sephadex resins were obtained from Pharmacia Biotech Inc., and the immobilized nickel resin was from Qiagen. Recombinant rat FTase was produced in Sf9 cells and purified as described (13). The His-tagged K-Ras (H 6 -K-Ras) and H-Ras (H 6 -H-Ras) were produced by expression of the appropriate * This work was supported by National Institutes of Health Grant GM46372 (to P. J. C.) and Fellowship GM18069 (to W. R. T.). 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.
¶ cDNAs, both gifts of Guy James (Southwestern Medical Center, Dallas, TX), in bacterial expression systems and purification as described (25). Radiolabeled prenylated peptides were produced enzymatically using FTase and [ 3 H]FPP substrates and purified as described (24). Unlabeled farnesyl-CVIM was produced by chemical farnesylation of the peptide using farnesyl bromide, and the product was purified by HPLC (26). Silica G60 TLC plates were from EM Separations.
Synthesis and Purification of Farnesylated K-Ras-Bacterially produced H 6 K-Ras was enzymatically farnesylated as described above for radiolabeled peptide farnesylation in Buffer A (20 mM Tris-Cl, pH 8.0, 20 mM KCl, 5 mM MgCl 2 , 5 M ZnCl 2 , 2 mM DTT) containing 0.1% Lubrol. Following incubation, the reaction mixture was applied to a 1-ml column of heptylamine-Sepharose (27) equilibrated in Buffer B (20 mM Hepes, pH 7.5, 1 mM EDTA, 1 mM DTT) containing 0.1% Lubrol, and then the column was washed extensively with the same buffer. Farnesylated H 6 -K-Ras remains bound to the column, whereas unmodified protein flows through. The modified protein was eluted in the same buffer containing 2% sodium cholate as the detergent instead of Lubrol. The cholate eluate was diluted 10-fold with the initial column buffer, applied to a 1-ml column of S-Sepharose (Pharmacia), and once again washed extensively with Buffer B containing 0.1% Lubrol. Farnesylated H 6 -K-Ras was eluted with this buffer containing 600 mM NaCl, flash frozen in aliquots, and stored at Ϫ80°C until use.
Equilibrium Binding Analysis-The method used was that of Hummel and Dreyer (28). Lyophilized FTase (177 g) was dissolved in 50 mM Tris-Cl, pH 8.0, 100 mM KCl, 5 mM MgCl 2 , 5 M ZnCl 2 , 1 mM DTT containing 0.2% ␤-octylglucoside and 500 nM [ 3 H]farnesyl-CVIM peptide (0.1 Ci/mmol; [ 3 H]f-CVIM), and applied to a 0.7 ϫ 16-cm Sephadex G-25 gel filtration column. The column was developed at room temperature using the same buffer containing the radiolabeled, modified peptide. Fractions of 250 l were collected, and the radioactivity in each was determined. The fractions corresponding to the void volume, i.e. the elution position of the enzyme, were used to determine the amount of enzyme-bound [ 3 H]f-CVIM; the baseline was determined by averaging several fractions of the eluent prior to this peak. The K d was calculated from the following equation (28): Formation and Analysis of Enzyme-Product Complexes Using Peptide Substrates-The FTase-FPP complex was made as described (22). Briefly, the enzyme (50 g) and [ 3 H]FPP (1 nmol; 15 Ci/mmol) were mixed with Buffer A containing 0.2% ␤-octylglucoside in a total volume of 100 l and incubated for 15 min at room temperature. Following the incubation, the FTase-FPP complex was separated from free FPP by Sephadex G-50 spin chromatography (29) on a 1-ml column equilibrated with Buffer C (50 mM Tris-Cl, pH 8.0, 100 mM KCl, 5 mM MgCl 2 , 0.2% ␤-octylglucoside). Comparison of the FTase (by protein determination) and FPP (by radioactivity determination) in the eluted product showed the stoichiometry of FPP binding to be nearly 100%. A stoichiometric amount of CVIM peptide was added to the FTase-[ 3 H]FPP complex, and the mixture was incubated for 15 min at room temperature in a total volume of 100 l in Buffer C. The reaction mixture was once again spin chromatographed and assayed as above for both protein and bound product.
To determine the dissociation rate of the product, the FTase-product complex was diluted to a concentration of 1.5 M with Buffer A containing 0.2% ␤-octylglucoside. An aliquot of this mixture (2 l) was added to a tube containing 1 l of a solution of FPP, peptide, or prenylated peptide (see figure legends for exact concentrations). Following incubation at the time and temperature conditions indicated in the appropriate figure legend, 1.5 l of the reaction was withdrawn and added to 125 l of cold Buffer C with the addition of 10 g of cytochrome c as a carrier protein and immediately applied to a 1-ml Sephadex G-25 spin chromatography column equilibrated with Buffer C. The amount of product remaining bound to FTase was determined by quantitation of radioactivity.
k cat Determination-Briefly, FTase activity was determined at saturating FPP (2 M) and increasing concentrations of the tetrapetide CVIM (0.05-4 M) at 10°C in a 10-min assay. Product formation was determined using a thin layer chromatography assay (30), in which the product spots were scraped, and the radioactivity in each was determined.
Formation and Analysis of Enzyme-Product Complexes Using Protein Substrates-The FTase-FPP complex was formed and isolated as mentioned above. To form the product complex with H 6 -K-Ras or H 6 -H-Ras substrates, the FTase-FPP complex (250 nM) was mixed with 50 nM H 6 -K-Ras or H 6 -H-Ras in Buffer C as above. To examine the dissociation of product from this complex, either FPP or buffer (20 mM Tris-Cl, pH 8.0, 0.2% ␤-octylglucoside) was added to give a total final concentration of 275 M of FPP in 10 l. The reaction mixture was incubated at 37°C for 30 min, and then cytochrome c (30 g) was added as a carrier protein. A 50% slurry of nickel resin (10 l) equilibrated in Buffer A containing 250 mM NaCl was added, and the mixture was rocked for 30 min at 4°C. The nickel resin was then isolated by pelleting in a microfuge for 30 s. The resin pellet was washed five times with Buffer A containing 250 mM NaCl, and bound proteins were eluted by incubation for 5 min at room temperature with a solution of 500 mM imidazole in the same buffer. The eluates were removed to a new tube containing SDS sample buffer, and the proteins were resolved on a 14% SDS-PAGE gel, followed by transfer to a polyvinylidene difluoride filter. These samples were then subjected to immunoblot using anti-FTase or anti-Ras antiserum (31). Visualization was by the alkaline phosphatase method using a commercial kit (Promega).

RESULTS AND DISCUSSION
As noted in the Introduction, kinetic analyses have shown that the rate-limiting step in FTase is the release of product from mammalian FTase, with a k cat of ϳ2 min Ϫ1 at 30°C. Although this result would seem to imply a high affinity binding of product to the enzyme, previous steady-state analysis of FTase indicated that a farnesylated peptide product was a very poor competitor of the reaction with a K i of ϳ5 M (20). Because steady-state kinetics are an indirect measure of affinity, we chose to directly determine the binding affinity of a farnesylated peptide to FTase. The method used was that of Hummel and Dreyer (28), which employs equilibrium gel filtration. A graphical depiction of this method is shown in Fig. 1. A Sephadex G-25 column is equilibrated with buffer containing radiolabeled ligand (shaded solution), and then the binding protein (i.e. FTase) is applied to the column in a small volume at a concentration that is much higher than that of the ligand in the buffer. As the enzyme moves through the column it binds ligand, with the result that a peak of radioactivity is observed in the void volume where the enzyme-ligand complex elutes (stippled black band); this peak is followed by a "trough" in the profile where the buffer that has been depleted of radiolabeled ligand (white solution) emerges. Equilibration of the binding protein and ligand within the column is indicated by the resolution of the peak and trough in the elution profile (28). From this type of profile it is possible to extract an equilibrium binding constant of the enzyme-ligand complex.
We Fig. 1. Analysis of such profiles (see "Experimental Procedures") yielded a K d for the interaction of 0.78 M (range of 0.5-1.0 M). Furthermore, addition of excess FPP or peptide substrate to the running buffer of the column did not affect the binding constant (not shown).

performed this type of analysis on the binding of [ 3 H]farnesylated-CVIM ([ 3 H]f-CVIM) to FTase; a typical profile is shown in
We next sought to directly examine the dissociation of product formed on the enzyme during catalysis. To investigate this step in the catalytic process, the FTase-FPP complex was prepared and isolated, and then a stoichiometric amount of a tetrapeptide substrate was added. Previous studies have shown that the reaction occurs quite rapidly under these conditions (k ϭ 2 ϫ 10 5 M Ϫ1 s Ϫ1 ) (22). Formation of the FTaseproduct complex under these single-turnover conditions allowed a direct examination of product dissociation rate, which was determined by rapid separation of the complex from free product on Sephadex G-25 spin chromatography columns. The results of this analysis, shown in Fig. 2A, revealed that there was no appreciable dissociation of product from the enzyme even after 10 min of incubation. This inability to detect product release was not simply due to its release and re-binding, be-cause addition of a large excess of unlabeled product to the reaction mixture after product formation but prior to the separation procedure did not result in any exchange with the radiolabeled product formed on the enzyme. Surprisingly, however, addition of excess peptide or isoprenoid to the reaction did trigger product release ( Fig. 2A). FPP was slightly more efficient than peptide substrate in this regard; k off values for product release were 0.13 and 0.08 min Ϫ1 , respectively, in the presence of the two substrates. Because these experiments were performed at 10°C, we determined the k cat under the same conditions. The results of this analysis, shown in Fig. 2B, gave a turnover number of 0.11 min Ϫ1 , a value in close agreement with the product release under the same conditions (see above). From these data, we conclude that FTase must bind an additional substrate molecule before it can release its product; the implications of this finding are discussed below.
We then asked the question of whether the requirement of substrate binding for product release also applied to modification of authentic protein substrates by FTase. To assess this, we developed an affinity co-precipitation method to examine the formation and dissociation of the enzyme-product complex (see "Experimental Procedures"). The FTase-FPP was prepared as before (except that unlabeled FPP was used), and the com-plex was then incubated with a substoichiometric amount of His-tagged Ras substrates. Following catalysis, the reaction mixture was incubated with a resin of immobilized nickel to precipitate the Ras and the associated enzyme. Proteins bound to the nickel resin were eluted with imidazole and analyzed by SDS-PAGE gel followed by immunoblotting using antisera directed against both FTase and Ras. Formation of a stable complex between FTase and Ras is thus detected by the appearance of the enzyme in the affinity precipitate.
The results of this type of product dissociation analysis using protein substrates are shown in Fig. 3. The analysis was initially performed with His-tagged K-Ras as the substrate and clearly indicated that FTase and H 6 -K-Ras do in fact form a stable complex under conditions where the K-Ras is subject to farnesylation (Fig. 3A, lane 3). As seen with the experiments using the peptide substrate of FTase, addition of excess FPP resulted in product release as demonstrated by the absence of FTase in the affinity precipitate under these conditions (Fig.  3A, lane 4). Additionally, FTase did not stably bind the nickel resin either in the absence or the presence of H 6 -K-Ras nor when FPP was omitted from the incubation (Fig. 3A, lanes 1  and 2). Furthermore, incubation of the product (farnesylated-H 6 -K-Ras) did not result in formation of a stable complex that could be precipitated with the nickel resin (not shown), demonstrating once again that only product formed on the enzyme binds in such a stable fashion.
Because all of the studies to this point were performed either with K-Ras or peptides encompassing the C terminus of K-Ras, we also examined a distinct substrate of FTase to determine whether this inability to release product was a general property of FTase or whether it was something unique to the properties of K-Ras as a substrate. For these studies, we selected His-tagged H-Ras as the protein substrate; this protein is quite different from K-Ras in that its CAAX sequence is CVLS and it does not contain the polybasic region just upstream of the CAAX box that is found in K-Ras. This combination of Ser as the X residue and the absence of the polybasic region results in a 10-fold higher K m for H-Ras as a substrate for FTase as compared with K-Ras (25). Addition of the H 6 -H-Ras to the FTase-FPP complex did indeed result in formation of a FTaseproduct complex that could be precipitated with the nickel resin (Fig. 3B, lane 3). Again, addition of FPP prior to the affinity precipitation resulted in release of the product by the enzyme, as demonstrated by the inability to detect FTase in the precipitate under these conditions (Fig. 3B, lane 4). The control experiments again confirmed that no stable complex was formed under conditions where FTase-product complex could not be formed (Fig. 3B, lanes 1 and 2).
What is the significance of the finding that FTase does not release its product until there is additional substrate present for it to bind? From a mechanistic viewpoint, these data suggest the presence of two distinct binding conformations for product on the enzyme. One of these conformations, to which product binds relatively weakly with a K d of around 1 M, exists on the free enzyme, and the second is a conformation formed during the catalytic process and from which product dissociates so slowly that no appreciable off-rate can be detected without additional substrate being present. The conversion to the conformational state that allows product release would not result from the catalytic step itself but would require encountering the additional substrate molecule, most likely FPP, given the high affinity of the enzyme for this substrate. Such a conversion, presumably involving some sort of conformational change such as that suggested by earlier fluorescence studies (22), would explain the finding that exogeneously added product binds only weakly, because it would only "see" the lower affinity conformation, and the energy barrier to go from this state to the high affinity one may be to high to be overcome.
Although it is not yet clear what the physiological significance of this property of FTase is, there are several intriguing possibilities. The first and the one we consider most likely is that in vivo FTase remains bound to the product till it encounters a specific site, e.g. a membrane compartment where FPP is located, to which the enzyme delivers its product (Fig. 4). This type of mechanism has been proposed for a related prenyltransferase, the type II geranylgeranyltransferase, which modifies specific GTP-binding proteins of the Rab family that are involved in intracellular membrane trafficking. Available data indicate that the complex of protein geranylgeranyltransferase type II and its product remains intact until it encounters the correct intracellular membrane to which the protein is targeted (32,33). In a similar fashion, FTase itself may act as an escort protein to deliver farnesylated protein to the intracellular membrane compartment where subsequent processing (i.e. proteolytic removal of the -AAX and methylation of the farnesylcysteine (2, 34)) occur. Another possibility that could account for the product remaining bound to FTase is that bound product serves to protect the enzyme. Free FTase may be labile, so that it is adventitious for the enzyme to have either substrate (i.e. FPP) or farnesylated protein bound to it.
An additional point of significance of these findings is that understanding the substrate requirement for product release by FTase provides insights into the design of inhibitors of this  1 and 2) or as the enzyme-FPP complex (lanes 3 and 4) were incubated with (lanes 2, 3, and 4) or without (lane 1) 50 nM H 6 -K-Ras. In the experiment in lane 4, additional FPP (275 M) was added during the initial incubation. The incubation was carried out at 37°C for 30 min, at which point 30 g of cytochrome c was added as a carrier protein. Immobilized nickel resin (10 l) was then added, and the incubation mixture was rocked for 30 min at 4°C. The resin was pelleted and then washed as described under "Experimental Procedures," and bound proteins eluted with imidazole buffer. The eluate was subjected to SDS-PAGE and immunoblot analysis using antisera directed against both FTase (upper strip) and Ras (lower strip). B, analysis with H-Ras substrate. The experiment was performed as described for A except that H 6 -H-Ras was used as the substrate protein instead of H 6 -K-Ras. For both panels, the standard (Std) lanes contain the amount of FTase that could be theoretically precipitated assuming stoichiometric conversion to product and the amount of Ras protein present in the reactions, respectively.
FIG. 4. Kinetic scheme for FTase. A simplified version of the overall kinetic scheme for FTase is shown. For simplicity, FPP binding to the free enzyme is shown as a single step, although it is best described as a two-step process (22). In this scheme, however, FPP binding to free enzyme is considered of minimal importance in vivo because the enzyme cycles directly to the E-FPP complex after catalyzing the farnesylation of a protein substrate (a Ras protein in the case depicted here). This direct conversion of the FTase-product complex to the FTase-FPP complex is the hypothesized consequence of FPP-triggered release of the farnesylated Ras protein (Ras-S[C15]). See text for further details. critical processing step. One can certainly envision the design of compounds capable of interacting with the product binding site in a very tight fashion. Elucidation of the structural basis of this interaction could lead to the design of highly effective product-based inhibitors. Another significant implication of the findings that FPP binding by FTase results in formation of a relatively stable FTase-FPP complex and that the enzyme does not release product until an additional substrate molecule is encountered is that the enzyme in vivo probably never exists for any appreciable period of time as a free (i.e. unliganded) species (see Fig. 4). Thus, it seems likely that the many types of FTase inhibitors that are under development as therapeutic agents are actually targeting the E-FPP complex rather than the free enzyme; such a realization could provide for strategies to optimize design of even more effective compounds. Additionally, if FTase is involved in the delivery of its product to another cellular protein or membrane (see above), defining this process could identify new targets for design of agents that can block subcellular trafficking and thus perturb the activities of specific products of the enzyme, e.g. oncogenic Ras proteins.