Effect of site-directed mutagenesis of conserved aspartate and arginine residues upon farnesyl diphosphate synthase activity.

All polyprenyl synthases catalyze the condensation of the allylic substrate, isopentenyl diphosphate, with a specific homoallylic diphosphate substrate. Polyprenyl synthases from Homo sapiens, Ratus rattus, Escherichia coli, Saccharomyces cerevisiae, Neurospora crassa, and Erwinia herbicola contain two conserved "aspartate-rich domains" (Ashby, M.N., and Edwards, P.A. (1992) J. Biol. Chem. 267, 4128-4136). In order to determine the importance of these domains in catalysis, the conserved aspartates or arginines in domains I and II of rat farnesyl diphosphate synthase were individually mutated to glutamate or lysine, respectively. The putative "active site" arginine (Brems, D.N., Breunger, E., and Rilling, H. C. (1981) Biochemistry 20, 3711-3718) was mutated to lysine. Each mutant enzyme was overexpressed in E. coli and purified to apparent homogeneity. Detailed kinetic analyses of the wild type and mutant enzymes indicated that mutagenesis of Asp104, Asp107, Arg112, Arg113, and Asp243 resulted in a decreased Vmax of approximately 1000-fold compared to wild type. However, no significant change in the Km values for either the isopentenyl diphosphate or geranyl diphosphate substrate were observed. The results strongly suggest that these amino acids, and to a lesser extent Asp244, are involved in either the condensation of isopentenyl diphosphate and geranyl diphosphate to form farnesyl diphosphate and/or the release of the farnesyl diphosphate product from farnesyl diphosphate synthase. The conservation of these amino acid residues in different enzymes from several species suggests that these domains play a similar role in other polyprenyl synthases.

Comparison of the amino acid sequences of a number of prenyltransferases (5) revealed the presence of three conserved regions. Two of these regions have been referred to as domains I and I1 and are characterized by an aspartate-rich motif which is flanked by other conserved residues (see Fig. 1). The third region of homology is similar to an active site peptide proposed from the studies of Brems et al. (13) on avian FPP synthase. Since all prenyltransferases and synthases share the common feature of binding at least one prenyl diphosphate substrate, it was hypothesized that the conserved domains I and I1 might be involved in substrate binding (5). In order to begin testing this hypothesis a previous study was performed (14) in which two of the conserved aspartate residues in domain I1 of rat FPP synthase were separately changed to glutamate residues. The results of this analysis suggested that the first aspartate in domain I1 was involved in the catalytic reactions of the enzyme (14).
We report here an extension of the initial study in which we have created point mutants of rat FPP synthase in both conserved domains I and 11. In addition, we mutated the rat FPP synthase arginine at position 192 to a lysine residue in order to examine the effect(s) of a conservative mutation within the proposed active site peptide (13).
The wild type and mutant proteins were expressed in Escherichia coli, purified by ion exchange chromatography and the kinetic parameters were determined. The results of this study imply that several of the conserved residues of domains I and I1 play an essential role in the catalytic reactions of FPP synthase since the mutations described greatly reduced the enzyme's specific activity. Analysis of Proteins-Proteins were analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) using the discontinuous buffer system of Laemmli (16). Following electrophoresis the gels were either stained with Coomassie Brilliant Blue R or were transferred to Amersham Hybond ECL nitrocellulose membranes for 16 h a t 20 V in transfer buffer (20 mM Tris, 150 mM glycine, 20% ( d v ) methanol). Following transfer the membrane was air-dried, rinsed with distilled water, soaked in 15% hydrogen peroxide for 15 min, and blocked with Trisbuffered saline (TBS), pH 7.0, containing 0.5% Tween 20 and 5% nonfat dry milk for 1 h. The membrane was washed to remove excess milk, incubated with affinity purified anti-FPP synthase antibodies (see below) in TBS, 0.5% Tween 20 for 2 h, washed three times with TBS, 0.3% Tween 20 for 10 min each, and incubated with the secondary antibody, anti-rabbit IgG coupled to horseradish peroxidase in TBS, 0.3% Tween 20 for 1 h. The membrane was finally washed with TBS, 0.3% Tween 20 (3 x 5 min washes) and TBS, 0.1% Tween 20 (3 x 5-min washes). The membrane was then soaked in the Amersham ECL detection reagents ( l : l , v/v) for 1 min and exposed to x-ray film for 1-6 s. Anti-FPP synthase antisera for Western analysis was generated by injection of rabbits with purified recombinant wild type FPP synthase. The antisera was purified by passing it over an affinity column consisting of purified wild type FPP synthase coupled to Tresyl-activated agarose (Schleicher & Schuell) according to the manufacturers instructions. Purified anti-FPP synthase antibodies were eluted from the affinity column with 100 mM glycine, pH 2.0, and the identity confirmed by Western analysis. Protein estimations were determined by the method of Bradford (17) using bovine serum albumin (Sigma) as a reference protein.

EXPERIMENTAL
Recombinant DNA Reagents"T4 DNA ligase, polynucleotide kinase, and restriction enzymes used in the construction of plasmids were obtained from Life Technologies Inc. Preparation of plasmid DNA, restriction enzyme digests, and agarose gel electrophoresis were performed by standard methods as described by Maniatis et al. (18). Sequence analysis of plasmids was performed by the dideoxy chain termination method of Sanger et al. (19) using the Sequenase Version 2.0 sequencing kit from U. S. Biochemical Corp. Specific oligonucleotide internal prim-ers were designed to obtain sequence throughout the FPP synthase gene.
Construction of the Reading Frame Cassette a n d the Prokaryotic Expression Vector for FPP Synthase cDNA-The strategy used in the construction of the reading frame cassette for FPP synthase is explained fully in previous publications (14,20).
Strains, Media, and lkansformations-E. coli strain XL1-Blue Site-directed Mutagenesis of FPP Synthase-cDNA mutants for rat FPP synthase were generated by the oligonucleotide-directed in vitro mutagenesis system from Amersham Corp. Single strand wild type FPP synthase, used as a template in the mutagenesis reaction, was generated with VCS-M13 interference resistant helper phage (Stratagene) infection of XL1-Blue cells which contained the wild type FPP synthase gene inserted into the Bluescript vector. The resulting sense single strand DNA template was isolated and purified by standard methods (18) and sequenced to confirm identity. The synthetic anti-sense oligonucleotides designed to produce the desired point mutations (with the changes to the wild type sequence underlined and in bold type) were as follows:   TARI.E I Purification of wild type and mutant rat FPP synthase expressed in E. coli Wild type and mutant FPP synthase proteins were expressed in E. coli, purified, and assayed as described under "Experimental Procedures." Typically FPP synthase protein accounted for 2040% of the total soluble E. coli protein and was purified to 98% homogeneity following Mono Q FPLC ion exchange chromatography. Following mutagenesis, transformants were screened by dideoxy sequencing to confirm the mutation. The mutant cDNA was excised from the Bluescript vector with two restriction enzymes, NdeI and SalI. This produced a 1.07-kilobase pair fragment which was ligated into the expression vector pARC360N and used to transform JMlOl cells.

FPPS (Rratus) M G E F F Q I Q D D Y L D L F G D P S V T G K FPPS ( E c o l f ) I G L A F Q V Q D D I L D V V G D T A T L G K
Purification of Recombinant FPP Synthase-A stationary phase culture (1 ml) was inoculated into 500 ml of M9+CAGM media and grown a t 37 "C with aeration until the Asoo nm was 0.5-0.6 (typically 6-8 h). The bacterial cells were harvested by centrifugation (4,000 rpm, 10 min). The resulting paste was resuspended in 5 ml of buffer A (IO m potassium phosphate, pH 7.0, 10 m M 2-mercaptoethanol, 1 mM EDTA). The cells were disrupted by three cycles of sonication for 30 s each with 1 min cooling on ice in between. Undisrupted cells and cell membranes were separated from the soluble material by centrifugation (15,000 xg, 30 min, 4 "C) and the total protein content and FPP synthase activity of the extract were determined. The protein precipitating between 30 and 65% ammonium sulfate was collected, dissolved in 2 ml of buffer A, and dialyzed a t 4 "C for 18 h against two 2-liter changes of buffer A. The dialyzed material was centrifuged (15,000 x g, 10 min, 4 "C) and the protein concentration estimated before diluting to a final protein concentration of approximately 10 mg/ml. The sample was pumped onto a 15 x 2.5-cm DE52 column (at a flow rate of 1 mumin) which had been equilibrated in buffer A. The column was washed with bufferAunti1 the absorbance a t 280 nm of the eluate had returned to baseline (typically 300 ml). The bound protein was eluted with 200 ml of buffer B (90 mM potassium phosphate, pH 7.0, 10 m M 2-mercaptoethanol, 1 mM EDTA), and precipitated by the addition of ammonium sulfate to a final concentration of 70%. The protein was dissolved in 2 ml of FPLC buffer (50 m M Tris-HCI, pH 7.0, 1 m M EDTA, 1 m M dithiothreitol, 0.01 M NaCI) and dialyzed for 18 h at 4 "C against two 2-liter changes of FPLC buffer. The dialyzed DE52 purified material was chromatographed on a Mono Q 5/5 HR ion-exchange column with an FPLC system (Pharmacia LKB Biotechnology Inc.). The column was equilibrated with FPLC buffer a t a flow rate of 1 ml/min and the sample was applied to the column via a 2-ml injection loop. Following sample application the column was washed with 25 ml of FPLC buffer followed by a 75-ml linear gradient from 0.01 to 1 M NaCl in the same buffer. Fractions (1 ml) were collected and assayed for FPP synthase activity.
In Vitro Assay of Farnesyl Diphosphate Synthase-FPP synthase was assayed essentially as described by Holloway  specific activity 4 or 50.7 Ci/mol was used dependent upon the specific activity of the various mutant proteins. Assays were carried out in quadruplet or triplicate and the variation between the multiple values was less than 10%. Each analysis was performed on a t least three separate occasions with essentially similar results. In all cases less than 10% of the substrate was converted into product. A separate control assay which contained no GPP was performed for each assay utilizing a different concentration of [1-l4C1IPP.

RESULTS AND DISCUSSION
The alignment of the conserved amino acid residues in domains I and I1 and the active site peptide of prenyltransferases from several different species is summarized in Fig. 1. All of the enzymes share the common feature of binding both allylic and homoallylic diphosphate substrates.
In order to test the hypothesis that the conserved amino acids might play a role in enzymatic activity we systematically mutated several of the conserved residues. The mutations were designed so that the charge of the residue in question remained constant; thus, an aspartate was changed to a glutamate and an arginine was changed to a lysine. The residues mutated in domain I were the aspartate at positions 103, 104, and 107 and the arginines at positions 112 and 113 according to the numbering of the rat FPP synthase protein. In domain I1 aspartate 244 was changed to a glutamate and in the active site peptide the arginine at position 192 was changed to a lysine.
Following oligonucleotide site-directed mutagenesis of the wild type gene each plasmid was sequenced to confirm that the desired mutation was present. Next, either the wild type or mutated gene fragments were inserted into the bacterial expression vector pARC360N which was then used to transform E. coli strain JM101. The wild type or mutant enzymes were expressed in E. coli after growth on minimal media and purified as described under "Experimental Procedures." Table  I summarizes the purification of the wild type and mutant FPP synthase enzymes as monitored by the specific activity at each step in the purification. Fig. 2A shows a Coomassie Brilliant Blue-stained SDS-PAGE analysis of the purification steps using the expression of the wild type protein as an example. As can be seen from Fig. 2  JMlOl which had been transformed with the expression vector alone (lane 1 ). An apparent subunit molecular mass of rat liver FPP synthase of 39 kDa as seen here is consistent with previous studies (14) and with the predicted molecular mass as calculated from the cDNA sequence (24,26,27). Lanes 2-6 of Fig. 2A show the three-step purification of the expressed wild type enzyme while lanes [7][8][9][10][11][12] show the expressed mutant proteins obtained after the final step of purification. The purified proteins appear homogeneous. Immunoblot analysis of an identical gel with affinity purified anti-FPP synthase antibodies confirmed the identity of the homogeneous band (Fig. B ) .
The purified proteins were used in multiple in vitro enzyme assays to determine the K, values for the IPP and GPP substrates and the V, , values for both the wild type and mutant FPP synthase proteins. Fig. 3A shows the double reciprocal plots of the initial velocities versus varied substrate concentration with IPP being the varied substrate a t constant GPP concentrations. These data were used to determine the Km(Ipp) values summarized in Table 11. Fig. 3B shows the double reciprocal plots of initial velocities versus varied substrate concentration with GPP being the varied substrate at constant IPP concentrations. These data were used to determine the K,(Gpp) values summarized in Table 11. In order to determine the V , , values of the enzymes which are also summarized in Table 11, the intercept on the y axis of the Lineweaver-Burk plots were -continued replotted against the reciprocal of the constant substrate concentration. These straight line plots (data not shown) intercepted the y axis at a value corresponding to the reciprocal of the V, , value (28,29).
The kinetic constants for the wild type rat FPP synthase given in Table I1 are in agreement with the previous analysis (14) and are similar to the values previously reported for the human, avian, and pig liver (23, 30-321, E. coli (33) and Saccharomyces cerevisiae (34) FPP synthases. The results of the point mutations, summarized in Table 11, indicate that the mutant enzymes fall into two groups, 1)   where As discussed previously (14), FPP synthase obeys an ordered sequential mechanism for the synthesis of FPP from IPP and GPP (35,36) for which there are several rate constants. The most likely interpretation for the reduced V, , value for the mutant A~p'~~-Glu was that once both substrates had bound, the rate of chemical conversion of GPP and IPP to FPP (k5) is reduced, or that the rate of release of FPP from the enzyme (kc) is much reduced (14). The decreased V, , in the mutant Asp243-Glu was most likely a result of a decrease in k5 such that kg, rather than k6 became rate-limiting (14). The product release step (k6) is rate-limiting in the wild type enzyme (35,36). However, we previously concluded that the mutant A~p'~~-Glu was unlikely to decrease V, , as a result of a further decrease in k6 since such a change would require compensatory changes in k3 and k4 that would in turn result in a reduced affinity for IPP and an increased affinity for the product FPP (14). Such a scenario appears unlikely.
Since several of the mutants analyzed here also exhibited a greatly reduced V, , value in the absence of changes in Km(lpp) or K,(Gpp), we propose a similar interpretation, i.e. these mutations decrease kg, the rate of the chemical step. Furthermore, since VmJKm(Gpp) = k l when k5 << k6 (141, then the mutations, Asplo4-Glu, A~p '~~-G l u , Argl"-Lys, and Arg113-Lys, that result in significant decreases in V, , but no change in K,(~pp) will also have decreased values of kl, the rate constant for addition of GPP to the enzyme.
KmCIPp) can be simplified to k4/k3, when k5 << k6 (14). Thus the finding that Km(Ipp) does not change for the mutations Asplo4-G1u, Asplo7-G1u, Arg112-Lys, and Arg113-Lys (Table 11) indicates that these mutations are unlikely to affect k4, the dissociation of IPP from the E.GPP.IPP complex. The results are consistent with a n important role for these four amino acids in kg, the catalytic step that results in the formation of the E.PPi.FPP complex.
One of the mutants, A~p'~~-Glu, showed a V, , value that was only 7-fold lower than wild type (Table 11). We hypothesize that one of the two rate constants for product formationhelease were less affected by this mutation.
When either of the 2 conserved arginine residues in domain I (Arg"' and Argil3) were mutated to lysine residues there was approximately a 1000-fold reduction in V, , with little or no change in the K , values for either substrate (Table 11). Previous studies have reported the inactivation of pig liver FPP synthase by the arginine modifying agent phenylglyoxal (37). Bernard and Popjak (37) proposed that the biphasic nature of this observed inactivation was due to two important arginine residues per subunit that were both accessible to phenylglyoxal. These residues were postulated as being important in the binding of either the allylic or homoallylic substrates (37). In the current study, separate mutation of the 2 domain I arginine residues resulted in a relatively inactive enzyme, we conclude that these residues are important for the catalytic reactions of FPP synthase. Since mutation of either domain I arginine residues resulted in a 1000-fold decrease in V, , while mutation of the "active site" arginine (Arglg2) resulted in only a 65% decrease in V, , (Table 11), it appears possible that phenylglyoxal modification of the 2 domain I1 arginines was responsible for the observed enzyme inactivation in the original study by Bernard and Popjak (37).
The term "active site peptide" of prenyltransferases (summarized in Fig. 1) arose from an elegant series of experiments carried out in the laboratory of Rilling (13,38). These workers synthesized a radiolabeled photoactivatable analogue of IPP, ~-azidophenyl[l-~H]ethyl pyrophosphate, which they incubated with FPP synthase purified from chicken liver. Following photoactivation a 30-amino acid CNBr fragment was found to contain 80% of the total radiolabel and upon Edman degradation of this peptide 16 of the 30 residues appeared to be modified (13). The two most extensively labeled being an arginine and an alanine (13). It was not known if this arginine was also modified by phenylglyoxal with the resulting loss in enzyme activity (37).
In order to determine the importance of this conserved active site arginine residue, FPP synthase containing a conservative mutation (Arglg2-Lys) was generated and purified. The V, , and Km(Ipp) values of the Arglg2-Lys FPP synthase mutant were approximately 65% of the wild type values, while the K,(GPP) remained unchanged (Table 11). These apparent V, , and Km(Ipp) values are not significantly different from wild type values. These studies indicate that Argl" does not play an important role in catalytic activity. Further studies to define a role for Arglg2 might involve less conservative mutations, whereby the charge on the residue is altered to neutrality or negativity.
In conclusion, in the current report we tested the hypothesis that the conserved "aspartate-rich domains I and 11" of prenyltransferases were involved in catalysis of the allylic and homoallylic substrates of these enzymes. These studies demonstrate for the first time that a number of the conserved aspartate and arginine residues in rat FPP synthase domains I and I1 are critical for efficient enzyme catalysis, since conservative mutations result in a 1000-fold decrease in V, , , but no change in the K, for IPP and GPP. These studies, together with the previous report (14), indicate that domains I and I1 are most likely critically involved in either the chemical conversion of IPP and GPP to FPP andor the release of the FPP product from the enzyme. Domains I and I1 in other polyprenyl synthases are likely to play a similar role.