Evidence for a catalytic role of zinc in protein farnesyltransferase. Spectroscopy of Co2+-farnesyltransferase indicates metal coordination of the substrate thiolate.

Protein farnesyltransferase (FTase) is a zinc metalloenzyme that catalyzes the addition of a farnesyl isoprenoid to a conserved cysteine in peptide or protein substrates. We have substituted the essential Zn2+ in FTase with Co2+ to investigate the function of the metal polyhedron using optical absorption spectroscopy. The catalytic activity of FTase is unchanged by the substitution of cobalt for zinc. The absorption spectrum of Co2+-FTase displays a thiolate-Co2+ charge transfer band (epsilon320 = 1030 M(-1) cm(-1)) consistent with the coordination of one cysteine side chain and also ligand field bands (epsilon560 = 140 M(-1) cm(-1)) indicative of a pentacoordinate or distorted tetrahedral metal geometry. Most importantly, the ligand-metal charge transfer band displays an increased intensity (epsilon320 = 1830 M(-1) cm(-1)) in the ternary complex of FTase x isoprenoid x peptide substrate indicative of the formation of a second Co2+-thiolate bond as cobalt coordinates the thiolate of the peptide substrate. A similar increase in the ligand-metal charge transfer band in a product complex indicates that the sulfur atom of the farnesylated peptide also coordinates the metal. Transient kinetics demonstrate that thiolate-cobalt metal coordination also occurs in an active FTase x FPP x peptide substrate complex and that the rate constant for the chemical step is 17 s(-1). These data provide evidence that the zinc ion plays an important catalytic role in FTase, most likely by activation of the cysteine thiol of the protein substrate for nucleophilic attack on the isoprenoid.

Protein farnesyltransferase (FTase) is a zinc metalloenzyme that catalyzes the addition of a farnesyl isoprenoid to a conserved cysteine in peptide or protein substrates. We have substituted the essential Zn 2؉ in FTase with Co 2؉ to investigate the function of the metal polyhedron using optical absorption spectroscopy. The catalytic activity of FTase is unchanged by the substitution of cobalt for zinc. The absorption spectrum of Co 2؉ -FTase displays a thiolate-Co 2؉ charge transfer band (⑀ 320 ‫؍‬ 1030 M ؊1 cm ؊1 ) consistent with the coordination of one cysteine side chain and also ligand field bands (⑀ 560 ‫؍‬ 140 M ؊1 cm ؊1 ) indicative of a pentacoordinate or distorted tetrahedral metal geometry. Most importantly, the ligand-metal charge transfer band displays an increased intensity (⑀ 320 ‫؍‬ 1830 M ؊1 cm ؊1 ) in the ternary complex of FTase⅐isoprenoid⅐peptide substrate indicative of the formation of a second Co 2؉ -thiolate bond as cobalt coordinates the thiolate of the peptide substrate. A similar increase in the ligand-metal charge transfer band in a product complex indicates that the sulfur atom of the farnesylated peptide also coordinates the metal. Transient kinetics demonstrate that thiolatecobalt metal coordination also occurs in an active FTase⅐FPP⅐peptide substrate complex and that the rate constant for the chemical step is 17 s ؊1 . These data provide evidence that the zinc ion plays an important catalytic role in FTase, most likely by activation of the cysteine thiol of the protein substrate for nucleophilic attack on the isoprenoid.
Protein farnesyltransferase (FTase) 1 catalyzes the transfer of the farnesyl group of farnesyl pyrophosphate (FPP) to a conserved cysteine residue of a protein substrate, including Ras proteins, nuclear lamins, and several proteins involved in cell signaling (1)(2)(3)(4). Protein farnesylation mediates membrane association and, possibly, interactions with other proteins essential for the localization and function of these proteins (3,4). One example in this regard is the requirement of farnesylation for the cell transforming ability of oncogenic Ras proteins (5); this result has stimulated an intense search for FTase inhibitors as potential anticancer drugs (6,7). An increased understanding of the molecular mechanism of FTase should enhance the rational design of such FTase inhibitors.
FTase is a metalloenzyme that contains one zinc ion per ␣/␤ heterodimer that is essential for optimal activity (8,9). Crosslinking and direct binding studies indicate that the zinc ion is required for the binding of protein but not isoprenoid substrates (8). Additionally, FTase containing Cd 2ϩ substituted for Zn 2ϩ has essentially normal catalytic activity, demonstrating that other metal ions can functionally substitute for the zinc (10). Chemical modification and site-directed mutagenesis studies have identified a conserved cysteine residue of FTase, Cys 299 , in the ␤ subunit, as important for catalytic activity and zinc binding, suggesting that the thiolate of this residue may directly coordinate the zinc ion (11).
Although it is clear that the zinc ion in FTase is critical for activity, the precise function of the metal, particularly the question of whether the primary role of the zinc ion is structural or catalytic, is not yet known. Proposed catalytic functions for the zinc ion in FTase include increasing the nucleophilicity of the cysteine residue of the protein substrate (3,8,11,12) and/or activating the diphosphate leaving group (11,13,14). Here we investigate the metal coordination polyhedron in FTase by substituting Co 2ϩ for Zn 2ϩ , which does not change the catalytic activity of the enzyme. This substitution provides a useful spectroscopic probe of the composition and geometric arrangement of the ligands around the metal ion (15,16), and the spectral data obtained indicate that the metal ion coordinates the thiolate of the peptide substrate in the presence of bound FPP.
Preparation of Metal-free Substrates-Recombinant H-Ras was purified from a bacterial expression system as described (1,17). Residual metals in the H-Ras preparation were removed by dialysis against Metal Reconstitution Experiments-FTase was reconstituted with different divalent metals (Zn 2ϩ , Co 2ϩ , Cd 2ϩ , or Ni 2ϩ ) by incubating the apo-enzyme (150 g/ml) with 60 M of the respective metal salt (atomic absorption grade, Aldrich) in Buffer 1 and 50 M EDTA for 5-10 min on ice before dilution into the assay mixture. For reactions containing dithiothreitol (DTT), the reducing agent (1 mM) was added to the apoenzyme prior to the addition of the metal. * This work was supported by National Institutes of Health Grants GM40602 (to C. A. F.) and GM46372 (to P. J. C.) and by funds from the Council for Tobacco Research (to P. J. C.). 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.  (19) were purchased from Calbiochem and Bachem, respectively. The substrate peptide, sequence TKCVIM, was synthesized on an Applied Biosystems synthesizer and purified as described (9). Co 2ϩ -FTase was prepared by mixing apo-FTase (40 -60 M) with excess CoCl 2 (90 -120 M) in Buffer 1, 5 mM MgCl 2 , and 50 M EDTA at room temperature. A sample of Zn 2ϩ -FTase was prepared for the reference cuvette by substituting ZnCl 2 in the incubation. The various complexes were formed by adding a 1:1 stoichiometry of the inhibitor (I a or I b ) and/or substrate (FPP or TKCVIM peptide) to both the sample and reference cuvette. Optical absorption spectra of Co 2ϩ -substituted FTase complexes were recorded on a Uvikon 9410 UV/VIS double beam spectrophotometer using a 120-l cuvette. Each spectrum was measured 2-3 times and then averaged. The calculated extinction coefficients for each spectrum were reproducible to within 10%.
Transient Absorption Spectrum-The transient absorbance at 340 nm was measured on a KinTek stopped flow spectrophotometer either by mixing 50 M Co 2ϩ -FTase⅐FPP binary complex (1:1 stoichiometry of enzyme to FPP) with 10 M peptide substrate, TKCVIM, to form an active ternary complex or by mixing 50 M peptide substrate, TKCVIM, with 10 M Co 2ϩ -E⅐I a to form an inactive ternary complex in Buffer 1, 5 mM MgCl 2 , and 50 M EDTA at 23 Ϯ 1°C. The absorbance transient was fit to an equation describing either a single first-order exponential (inactive ternary complex) or two consecutive first-order reactions (active ternary complex) (20) by nonlinear regression using the Marquat-Levenberg method provided in the KinTek software package (21).
Miscellaneous Procedures-All solutions were prepared with deionized water (18 M⍀) in plasticware presoaked in 5 mM EDTA followed by deionized water. Protein concentrations were determined by the Coomassie Blue binding method using a commercial kit (Bio-Rad).

RESULTS AND DISCUSSION
Co 2ϩ -FTase Is Catalytically Active-Previous metal substitution experiments indicated that Zn 2ϩ and Cd 2ϩ restore activity to apo-FTase, whereas Ni 2ϩ , Mn 2ϩ , Hg 2ϩ , and Co 2ϩ do not (10). Because Co 2ϩ can generally substitute for Zn 2ϩ in other zinc metalloenzymes (15), we retested the activity of the Co 2ϩ -FTase in an assay with TCEP substituted for DTT. (DTT catalyzes the formation of Co 3ϩ from Co 2ϩ in the presence of oxygen while TCEP reacts more slowly (22,23).) Using these conditions, the activity of Co 2ϩ -FTase is comparable with that of Zn 2ϩ -FTase (Fig. 2). This restoration of activity by Co 2ϩ is not due to trace contamination of Zn 2ϩ in the assay solutions for the following reasons. First, the high concentration of FTase employed in this reconstitution experiment (0.16 M) made it very unlikely that Zn 2ϩ -FTase could form quantitatively in the assay. Second, the addition of 1 mM DTT to apo-FTase before the addition of metal had no effect on the activity of Zn 2ϩ -FTase but abolished 75% of the activity in the presence of Co 2ϩ (Fig. 2), and third, the addition of Ni 2ϩ , which binds EDTA with an affinity greater than either Zn 2ϩ or Co 2ϩ (24), does not restore the activity of apo-FTase (Fig. 2).
Absorption Spectra of Co 2ϩ -FTase-Substitution of Co 2ϩ into the zinc binding site of proteins provides a useful spectroscopic probe of the composition and geometry of the metal polyhedron (15,16). Because the high catalytic activity of Co 2ϩ -FTase indicates that the metal binding site in the enzyme is not significantly disturbed by this replacement, spectral information obtained from Co 2ϩ -FTase is both structurally and catalytically relevant. Indeed, the optical absorption spectrum of Co 2ϩ -substituted FTase (Fig. 3A, solid line) displays characteristic features of the spectra of many Co 2ϩ -substituted zinc enzymes (15,16). The low wavelength absorbance (below 450 nm) is assigned to a ligand-metal charge transfer (LMCT) band indicative of sulfur-to-cobalt charge transfer resulting from thiolate coordination (16). Furthermore, the intensity of the absorption shoulder at 320 nm (⑀ 320 ϭ 1030 M Ϫ1 cm Ϫ1 ) is consistent with one thiolate ligand; the extinction coefficient of LMCT bands are normally 900-1300 M Ϫ1 cm Ϫ1 per cobalt-thiolate bond (25). Consistent with this assignment, an essential cysteine in the ␤ subunit of FTase, Cys299, has been identified that exhibits the characteristics of a zinc ligand (11). In addition, the intensity of the ligand field absorption bands that are observed in the low energy region of the spectrum (⑀ 560 ϭ 140 M Ϫ1 cm Ϫ1 ; ⑀ 635 ϭ 100 M Ϫ1 cm Ϫ1 ) is indicative of a pentacoordinate or distorted tetrahedral metal geometry (16). Removal of the magnesium ion from the solution did not alter the absorption spectrum (data not shown).
Absorption Spectra of Co 2ϩ -FTase Binary Complexes-The addition of FPP to Co 2ϩ -FTase to form the binary complex has little influence on the optical spectrum (Fig. 3A, dotted line), except that the extinction coefficient of the ligand field bands decreases (⑀ 560 ϭ 100 M Ϫ1 cm Ϫ1 ). This diminution suggests a slight alteration in the geometry of the metal polyhedron, perhaps caused by a protein conformational change or interaction of the pyrophosphate group of FPP with Co 2ϩ . Additionally, the spectrum of Co 2ϩ -FTase with a bound FPP analogue (compound I a , see Fig. 1) is virtually identical to the spectrum of E⅐FPP (Fig. 3, A and B), suggesting that FPP and I a bind to the Catalytic Metal in FTase 21 enzyme in a similar manner. Finally, the shape of the spectrum of Co 2ϩ -FTase in the presence of the peptide substrate TKCVIM is indistinguishable from that of Co 2ϩ -FTase alone, although the extinction coefficients of both the ligand field and the LMCT bands increase slightly upon the addition of peptide (Ͻ20%; data not shown). The marginal influence of peptide binding on the absorption spectrum suggests that cobalt does not coordinate the thiolate of the peptide substrate in the binary complex. Absorption Spectra of Ternary Complexes of Co 2ϩ -FTase-To probe the environment of the metal binding site in the ternary complex, we used the aforementioned FPP analogue (compound I a ) that binds to FTase but is not utilized in catalysis to form an inactive but stable ternary complex mimic. The spectrum of the Co 2ϩ -FTase⅐I a ⅐TKCVIM ternary complex (Fig. 3B) exhibits quite striking differences from the absorption spectrum of either the E⅐I a or E⅐TKCVIM binary complex. Most importantly, the intensity of the LMCT band at 320 nm essentially doubles (⑀ 320 ϭ 1830 M Ϫ1 cm Ϫ1 ), indicating the formation of a second Co 2ϩ -thiolate bond, consistent with coordination of the thiolate of the peptide substrate with Co 2ϩ in this ternary complex. Furthermore, the increased intensity of the ligand field absorption bands at higher wavelengths in the E⅐I a ⅐TKCVIM complex compared with those in the E⅐I a binary complex is also consistent with additional coordinating thiolates (15,(27)(28)(29). Finally, the increased extinction coefficient of the d-d transition maximum at 635 nm (⑀ 635 ϭ 310 M Ϫ1 cm Ϫ1 ) is typical of a cobalt binding site with tetrahedral geometry (30). The observed extinction coefficients at both 320 and 640 nm increase linearly with the concentration of added peptide before saturating at a peptide concentration equal to the enzyme concentration (Fig.  3C), providing compelling evidence that (i) the peptide binds stoichiometrically to the E⅐I a complex and (ii) the increased extinction coefficient observed upon addition of peptide reflects the spectrum of the E⅐I a ⅐TKCVIM complex and not the absorption of nonspecific Co 2ϩ ⅐TKCVIM complexes.
We also formed the ternary complex using authentic FPP and an analog of a peptide substrate, B581 (compound I b , see Fig. 1), that cannot be used in catalysis. The spectrum of the resultant Co 2ϩ -FTase⅐FPP⅐I b ternary complex retains the significant spectral features of the Co 2ϩ -FTase⅐I a ⅐TKCVIM complex described above, including the increase in the LMCT band (⑀ 320 ϭ 1900 M Ϫ1 cm Ϫ1 ) and changes in the shape and extinction coefficients of the ligand field absorption bands (⑀ 560 ϭ 150 M Ϫ1 cm Ϫ1 ; ⑀ 610 ϭ 170 M Ϫ1 cm Ϫ1 ; ⑀ 660 ϭ 150 M Ϫ1 cm Ϫ1 ) (data not shown). This similarity suggests that the observed spectral properties are general features of Co 2ϩ -FTase ternary complexes rather than unique to the inactive E⅐I a ⅐TKCVIM complex. Taken together, these spectral data provide direct evidence that the cysteine thiol(ate) of the peptide substrate directly coordinates Co 2ϩ in these ternary complexes.
Absorption Spectrum of the Co 2ϩ -FTase Product Complex-Because the equilibrium constant for the formation of farnesylated peptide is very large (26), the reaction of FTase⅐FPP with one equivalent of the peptide substrate should produce a stoichiometric amount of product bound to FTase. Furthermore, under these conditions the enzyme should not turnover or release the farnesylated peptide. We therefore used this method to prepare the Co 2ϩ -FTase⅐farnesylated-peptide product complex and recorded its spectrum (Fig. 3A). Examination of the spectrum of the product complex reveals features that are distinct from those of either the binary or ternary complexes described above. The LMCT band in the product complex displays two distinctive shoulders at 320 and 370 nm (⑀ 320 ϭ 1600 M Ϫ1 cm Ϫ1 and ⑀ 370 ϭ 650 M Ϫ1 cm Ϫ1 ). However, as observed in the ternary complexes, the extinction coefficient at 320 nm in the product complex is significantly greater than

Catalytic Metal in FTase 22
that in the Co 2ϩ -FTase⅐FPP binary complex, indicative of additional sulfur-cobalt coordination in the product complex. On the other hand, the shape and extinction coefficient (⑀ 560 ϭ 130 M Ϫ1 cm Ϫ1 ) of the ligand field absorption band resembles the Co 2ϩ -E⅐FPP binary complex rather than the ternary complexes. These spectral changes also increase linearly upon the addition of limiting TKCVIM to the E⅐FPP complex and saturate when one equivalent of the peptide is added (data not shown), indicating that the spectral changes are due to an interaction with Co 2ϩ -FTase. This is a surprising result because metal coordination by a thioether in a zinc metalloenzyme has only been implicated in one previous case, the DNA repair protein Ada (12,31,32). However, an interaction between a thioether sulfur and Co 2ϩ has been observed in Co 2ϩsubstituted azurin (33), and ligand-metal charge transfer has been detected between a thioether sulfur and Co 3ϩ in model compounds (34). The observed spectral changes in the product complex suggest that the thioether moiety of the farnesylated peptide product interacts with Co 2ϩ in the enzyme⅐product complex.
Transient Absorbance of Active and Inactive Ternary Complex-To verify that the increase in the LMCT band observed in inactive ternary complexes also occurs in an active ternary complex, we measured the transient absorbance at 340 nm after mixing excess E⅐FPP with TKCVIM (Fig. 4). The transient increase in absorbance caused by formation of an E⅐FPP⅐TKCVIM complex followed by an exponential decay due to formation of the product complex is exactly the behavior predicted from the static spectra (Fig. 3, A and B). These data clearly demonstrate that the peptide thiol also interacts with the metal ion in the active ternary complex. A two-exponential fit of these data indicates that the rate constant for formation of the ternary complex is 60 s Ϫ1 , consistent with a second-order rate constant of 1 ϫ 10 6 M Ϫ1 s Ϫ1 , and the rate constant for product formation is 17 s Ϫ1 (Fig. 4). A single exponential increase was observed when excess TKCVIM was mixed with E⅐I a to form an inactive ternary complex with a second-order rate constant of 1 ϫ 10 6 M Ϫ1 s Ϫ1 (data not shown). Our data are compatible with previous transient kinetic measurements where the rate constant of the chemical step and the secondorder rate constant in Zn 2ϩ -FTase were determined as Ͼ12 s Ϫ1 and 2 ϫ 10 5 M Ϫ1 s Ϫ1 , respectively (26).
In conclusion, these spectral studies clearly indicate that the metal polyhedron of FTase has either pentacoordinate or distorted tetrahedral geometry and contains one cysteine ligand. Furthermore, these data demonstrate that both the peptide substrate (in the ternary complex) and the farnesylated peptide product bind FTase in such a way that the Co 2ϩ directly coordinates the sulfur atom. Minimally, this places the divalent metal ion at the active site of FTase and demonstrates the presence of an open coordination sphere, most likely due to the presence of at least one water molecule in the metal coordination polyhedron. These features strongly argue that the zinc ion in FTase plays a catalytic rather than a structural role (30). In fact, these spectral data are completely consistent with the properties of the metal site in the Ada protein, where the bound zinc activates the thiol of a cysteine residue for nucleophilic attack on the carbon of an S p -methylphosphotriester (12,31,32). Furthermore, catalysis of S-alkylation by Zn 2ϩ coordination of thiol compounds has also been implicated in cobalaminindependent methionine synthase (35) and several chemical reactions (36 -38). Therefore, we propose a minimal scheme of the molecular mechanism of FTase ( Fig. 5) in which the zinc ion in FTase activates the thiol of the peptide substrate for nucleophilic attack. Studies investigating the rate constant of the chemical step in metal-substituted FTase should further illuminate the function of the zinc ion and the catalytic mechanism of FTase.