Conversion of product specificity of archaebacterial geranylgeranyl-diphosphate synthase. Identification of essential amino acid residues for chain length determination of prenyltransferase reaction.

Prenyltransferases catalyze the consecutive condensation of isopentenyl diphosphate with allylic diphosphates to produce prenyl diphosphates whose chain lengths are absolutely determined by each enzyme. To investigate the mechanism of the consecutive reaction and the determination of the ultimate chain length, a random mutational approach was planned. A geranylgeranyl-diphosphate synthase gene from Sulfolobus acidocaldarius was randomly mutagenized by NaNO2 treatment to construct a library of mutated geranylgeranyl-diphosphate synthase genes on a yeast expression vector. The library was screened for suppression of a pet phenotype of yeast C296-LH3, which is deficient in hexaprenyl-diphosphate synthase. Five mutants that could grow on a YEPG plate, which contained only glycerol as an energy source instead of glucose, were selected from ~1,400 mutants. All selected mutated enzymes catalyzed the formation of polyprenyl diphosphates with prenyl chains longer than geranylgeranyl diphosphate. Especially mutants 1, 3, and 5 showed the strongest elongation activity to produce large amounts of geranylfarnesyl diphosphate with a concomitant amount of hexaprenyl diphosphate. Sequence analysis revealed that each mutant contained a few amino acid substitutions and that the mutation of Phe-77, which is located on the fifth amino acid upstream from the first aspartate-rich consensus motif, is the most effective for elongating the ultimate product. Amino acid alignment of known prenyltransferases around this position and our previous observations on farnesyl-diphosphate synthase (Ohnuma, S.-i., Nakazawa, T., Hemmi, H., Hallberg, A.-M., Koyama, T., Ogura, K., and Nishino, T. (1996) J. Biol. Chem. 271, 10087-10095) clearly indicate that the amino acid at the position of all prenyltransferases must regulate the chain elongation.

Prenyltransferases, which are essential enzymes in isoprenoid biosynthesis, catalyze the consecutive condensation of isopentenyl diphosphate (IPP) 1 with allylic diphosphates to syn-thesize linear prenyl diphosphates with various chain lengths. These enzymes are classified according to the products with the longest chain length and the geometry of the double bonds that are formed by the condensations. So far, a number of prenyltransferases have been found in various organisms and characterized (1). For example, farnesyl-diphosphate synthase, which is a key enzyme in the biosynthesis of steroids, prenylquinones, farnesylated proteins, and dolichols, catalyzes the condensations of IPP with dimethylallyl diphosphate (DMAPP; C 5 ) and with geranyl diphosphate (GPP; C 10 ) to give farnesyl diphosphate (FPP; C 15 ) as an ultimate product. Geranylgeranyl-diphosphate (GGPP; C 20 ) synthase, whose product is a precursor of carotenoids, geranylgeranylated proteins, chlorophylls, and ether-linked lipids of archaebacteria, utilizes DMAPP, GPP, and FPP as allylic substrates to give an amphipathic molecule containing four isoprene units, GGPP (Fig.  1). Hexaprenyl-diphosphate (HPP) synthase catalyzes the consecutive condensation of IPP with E stereochemistry to produce a C 30 compound (Fig. 1). Although these enzymes catalyze similar condensation reactions, they do not catalyze the condensation beyond the limit of the chain length of the product determined by their own specificities. Also, it was recently reported that these enzymes evolutionarily diverged from a common ancestor (2). Why does the condensation stop at a proper step that is determined by each enzyme? How have prenyltransferases evolved?
It has been reported that the product chain length of each enzyme is modulated by in vitro reaction conditions (3)(4)(5)(6)(7). In addition, several lines of evidence have been accumulated for the divergence of prenyl chain length distribution in living cells. The prenyl chain length of respiratory quinone is altered by viral infection (8) and differs from tissue to tissue (9). The dolichyl chain length in rat liver also changes on carcinogenesis (10) or aging (11,12). On the basis of in vitro examination, Ohnuma et al. (13) and Sagami and co-workers (14) have suggested that these phenomena reflect the level of IPP and metal ions in the living cells. However, based on these pieces of information, the mechanism of the termination of the consecutive condensation has not been elucidated in terms of the molecular structure and functions of prenyltransferases.
During the past few years, the deduced amino acid sequences of FPP synthase (15)(16)(17)(18)(19), GGPP synthase (2, 20 -23), HPP synthase (24), heptaprenyl-diphosphate synthase (25), and octaprenyl-diphosphate synthase (26) have been reported. Comparison of the primary structures has revealed that these enzymes have several conserved domains, including two aspartate-rich motifs, DDXXD, where X encodes any amino acid (2,18), and raises the question of what differences in amino acids determine the product specificity. It has been reported that a natural mutant of yeast FPP synthase, in which lysine at position 197 is replaced by glutamic acid, catalyzes the synthesis of FPP with a large amount of the intermediate chain product, GPP (27). However, the ultimate chain length was not altered by the mutation. Although site-directed mutagenesis studies have been carried out by several groups (26, 28 -31), all previous studies have focused on the mechanism of single condensation of IPP. Therefore, which amino acid residues are essential for the determination of the chain length of the ultimate product is unknown.
Recently, Ohnuma et al. succeeded in converting FPP synthase to GGPP synthase using random mutagenesis and expression screening of mutated enzyme that possessed GGPP synthase activity. They also showed that the substitution of only a single amino acid could bring about the conversion (32). However, the system of the expression screening used carotenoid synthesis from GGPP in vivo. Therefore, it cannot be applied to making mutated enzymes that catalyze the formation of polyprenyl diphosphates longer than GGPP. To obtain further information about amino acid residues involved in the chain length determination and evolutionary events, we have tried to convert Sulfolobus acidocaldarius GGPP synthase, which is thought to have emerged at a quite early stage of evolution, to a prenyltransferase that yields a product with a prenyl chain longer than GGPP using random chemical mutagenesis and a complement system of a yeast mutant (Fig. 1). Saccharomyces cerevisiae C296-LH 3 is deficient in HPP synthase, so it cannot produce coenzyme Q 6 , and consequently, the growth of the mutant is limited by the availability of fermentable substrates (33,34). If mutated GGPP synthase has HPP synthase activity, the transformed yeast is expected to be able to grow on a YEPG plate, which does not contain any fermentable substrates. This paper reports the isolation of mutated GGPP synthases producing longer polyprenyl diphosphates and the identification of the amino acid residues that are important for chain length determination.
DNA Sequence Analysis-All nucleotide sequences encoding the mutated GGPP synthases were determined by the dideoxy chain termination method using a Model 373A DNA sequencer (Perkin Elmer). DNA and deduced amino acid sequences were analyzed and compared with those of the wild-type enzyme using MacMollyTetra genetic information processing software.
Construction of Plasmids-New HindIII sites were introduced by polymerase chain reaction-mediated mutagenesis in the 5Ј-upstream region and 3Ј-downstream region of the GGPP synthase gene from S. acidocaldarius using synthetic primers (5Ј-AAGAGAAGCTTATGAGT-TACTTTGAC-3Ј and 5Ј-GATACAAGCTTTATTTTCTCC-3Ј). The polymerase chain reaction-amplified product was digested with HindIII, and the 1,002-base pair fragment was ligated into the HindIII site in pBluescriptII KS ϩ to give pBS-GGPS. Plasmid Y-crtE was digested with HindIII to cut out the crtE fragment, and then the resulting fragment that contained the phosphoglycerol kinase promoter and terminator was self-ligated to give the yeast expression vector Y-PGK. Plasmid pBS-GGPS was digested with HindIII, and then the fragment of the GGPP synthase gene was ligated with HindIII-cut dephosphorylated Y-PGK to construct Y-GGPS.
Random Mutagenesis of GGPP Synthase Gene-Introduction of random mutations in the coding region of the GGPP synthase gene was carried out using NaNO 2 as described by Myers et al. (35) with some modifications. Single-stranded DNA was isolated from Escherichia coli cells containing pBS-GGPS with a helper phage, M13KO7. The singlestranded DNA was treated with 1 M NaNO 2 for 60 min, and then the mutagenized single-stranded DNAs were purified. The complementary DNA was synthesized by the action of avian myeloblastosis virus reverse transcriptase using a DNA primer (5Ј-CCCCCCTCGAGGTC-GACGGTATCGATAA-3Ј) that was bound to the region of the T7 promoter. The fragments of mutated GGPP synthase genes were excised from the plasmids by HindIII and ligated with HindIII-cut dephosphorylated Y-PGK. The resultant plasmids were introduced into E. coli XL1-Blue to make a library. The plasmids were isolated from the library for further experiments.
Preparation of Mutated GGPP Synthase from Yeast-Preparation of a crude extract from yeast cells was essentially carried out by the spheroplast method described by Itoh et al. (36). S. cerevisiae C296-LH 3

FIG. 1. Biosynthesis of ubiquinone-6 in yeast.
Wild-type S. cerevisiae has three kinds of E-polyprenyl-diphosphate synthases (FPP synthase, GGPP synthase, and HPP synthase). The steps that the three enzymes catalyze are indicated by thick black arrows. Wild-type GGPP synthase from S. acidocaldarius catalyzes three consecutive condensations of IPP to produce GGPP with significant amounts of intermediates (white arrows). The S. acidocaldarius GGPP synthase gene was randomly mutagenized, and then the mutagenized genes were introduced into S. cerevisiae C296-LH 3 , which is deficient in HPP synthase. Complementary mutated genes, which are expected to show HPP synthase activity, were isolated. OPP, OP 2 O 6 3Ϫ . cells harboring mutated GGPP synthase genes were grown in leu Ϫ medium at 30°C for 4 days. The cells (ϳ400 mg) were collected by centrifugation and washed once with 800 l of buffer A (50 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 50 mM dithiothreitol, and 1 M sorbitol). The cells were resuspended in 1.2 ml of buffer B (50 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 3 mM dithiothreitol, and 1 M sorbitol) and were incubated with 0.8 mg of zymolyase 20T (Seikagaku Corp.) for 1 h at 30°C. The resulting spheroplasts were washed three times with buffer B and then resuspended in 1 ml of buffer C (50 mM KH 2 PO 4 /KOH, pH 5.8, 10 mM 2-mercaptoethanol, and 1 mM EDTA). The cell suspension was sonicated 10 times for 10 s at 2-min intervals with a Branson sonifier at the maximum output in an ice bath. The lysate was incubated at 55°C for 1 h and then centrifuged at 10,000 ϫ g for 10 min. The supernatant was used to assay for prenyltransferase activity. Construction of Plasmids for Expression of Mutated GGPP Synthase in E. coli-Plasmids (Y-GGPSmut1, -2, -3, -4, and -5) were digested with HindIII, and then the DNA fragments encoding mutated GGPP synthases were recovered. The fragments were ligated with HindIII-cut dephosphorylated pBluescriptII KS ϩ to construct pBS-GGPSmut1, -2, -3, -4, and -5.
Preparation of Mutated GGPP Synthase from E. coli-E. coli XL1-Blue was transformed with each of the plasmids (pBS-GGPSmut1, -2, -3, -4, and -5) and cultured according to the methods described previously (37). The cells were harvested and disrupted by sonication in 50 mM KH 2 PO 4 /KOH, pH 5.8, 10 mM 2-mercaptoethanol, and 1 mM EDTA. The homogenate was heated at 55°C for 60 min and then centrifuged at 100,000 ϫ g for 10 min. The supernatant was used as a crude enzyme to assay for prenyltransferase activity. The expression level of each mutated GGPP synthase was confirmed by SDS-polyacrylamide gel electrophoresis (12.5%) with Coomassie Brilliant Blue staining.
Measurement of Prenyltransferase Activity-The assay mixture contained, in a final volume of 1 ml, 25 nmol of [1-14 C]IPP (37 GBq/mol), 25 nmol of the indicated allylic substrate (DMAPP, GPP, all-(E)-FPP, all-(E)-GGPP), 5 mol of MgCl 2 , 10 mol of phosphate buffer, pH 5.8, and a suitable amount of enzyme. This mixture was incubated at 55°C for 60 min, and the reaction was stopped by chilling quickly in an ice bath. The mixture was shaken with 3.5 ml of 1-butanol that had been saturated with H 2 O. The 1-butanol layer was washed with water saturated with NaCl, and radioactivity in the butanol layer was determined with a liquid scintillation counter. One unit of enzyme activity is defined as the activity required to incorporate 1 nmol of [1-14 C]IPP into expanded prenyl diphosphates extracted into the 1-butanol layer.
Product Analysis-After the enzymatic reaction at 55°C with [1-14 C]IPP, the polyprenyl diphosphates were extracted with 1-butanol, and then the 1-butanol was evaporated under a N 2 stream. The resulting polyprenyl diphosphates were treated with acid phosphatase ac-cording to the method of Fujii et al. (38). The hydrolysates were extracted with pentane and analyzed by reversed-phase thin-layer chromatography using LKC-18 developed with acetone/H 2 O (9:1) and by normal-phase thin-layer chromatography using Kieselgel 60 developed with benzene/ethyl acetate (9:1). Authentic standard alcohols were visualized with iodine vapor, and the distribution of radioactivity in the products was detected with a Fuji BioImage BAS2000 analyzer. powerful method for identifying important amino acid residues. The yeast strain C296-LH 3 was used for screening mutated prenyltransferases whose products had prenyl chains longer than GGPP. C296-LH 3 is deficient in HPP synthase and does not produce coenzyme Q 6 with a hexaprenyl group. Consequently, C296-LH 3 forms colonies smaller than those of the wild type when grown on a medium containing a small amount of glucose (D plate) (Fig. 2, B and E) and cannot grow on a medium containing only a nonfermentable substrate as a carbon source such as glycerol (YEPG plate) (Fig. 2F) because coenzyme Q 6 is necessary for nonfermentable glycosylation. Before screening, we examined the effect of expression of wildtype GGPP synthase from S. acidocaldarius on C296-LH 3 . The yeast cells expressing wild-type GGPP synthase, C296-LH 3 /Y-GGPS (Fig. 2K), formed colonies with a size between that of C296-LH 3 /YEpG3⌬SpH (Fig. 2H), which has the yeast HPP synthase gene, and that of C296-LH 3 without any plasmids (Fig. 2E) on a D plate containing a slight amount of glucose as a carbon source. However, C296-LH 3 /Y-GGPS can hardly grow on a YEPG plate containing only glycerol as a carbon source (Fig. 2L). These data might indicate that S. acidocaldarius GGPP synthase could partially complement the deficiency of yeast HPP synthase. However, C296-LH 3 /Y-GGPS apparently formed smaller colonies compared with C296-LH 3 / YEpG3⌬SpH, and the growth characteristics of C296-LH 3 /Y-GGPS on a YEPG plate were quite different from those of the wild type (Fig. 2, L and I). Therefore, we were convinced that this screening system was useful.

Effect of Wild-type GGPP
Random Mutagenesis of GGPP Synthase Gene and Isolation of Mutants That Suppress pet Phenotype-A random chemical mutagenesis strategy using a single-stranded DNA was used to introduce mutations in the entire coding region of the GGPP synthase gene from S. acidocaldarius. Single-stranded DNA derived from pBS-GGPS was subjected to chemical mutagenesis with 1 M NaNO 2 for 60 min according to the methods described by Myers et al. (35), and then a plasmid library that can express mutated GGPP synthase genes in yeast was constructed. C296-LH 3 cells were transformed with the library plasmids, and then the mutants were isolated that could grow on a YEPG plate and formed colonies larger than those of C296-LH 3 on a D plate as a result of the expression of functional coenzyme Q. Five mutants (mutants 1-5) were isolated from ϳ1,400 recombinants. Fig. 2 (M-O) shows the results of mutant 1. Mutant 1 can grow on a YEPG plate and forms colonies larger than those without the plasmid on a D plate. The colony size of mutant 1 on a D plate is apparently larger than that of the mutant containing Y-GGPS. The other four mutants also gave similar results.

Identification of Prenyltransferase Activity and Effect of Mutations-
The yeast cells containing Y-GGPSmut1, -2, -3, -4, and -5, which were obtained from the corresponding mutants, were cultured, and then their cell homogenates were prepared. The homogenates were heated at 55°C for 1 h in order to denature the enzymes derived from the host cells, and then the denatured proteins were precipitated. Using the supernatants, we analyzed the reaction products of the mutated GGPP synthases (Fig. 3). When GGPP was used as an allylic substrate, all mutants gave geranylfarnesyl diphosphate (GFPP; C 25 ), which is one isoprene unit longer than GGPP. On the other hand, the wild-type enzyme gave no detectable amount of products longer than GGPP. When FPP was used as an allylic substrate, the ratio of GGPP to GFPP depended on the mutants (Fig. 3A). To analyze the mutated enzymes more precisely, they were overproduced in E. coli. The genes of the mutated GGPP synthases were integrated into pBluescriptII, and E. coli XL1-Blue cells were transformed with the plasmids (pBS-GGPSmut1, -2, -3, -4, and -5) and cultured. The crude enzyme solutions were prepared from the cultures and heated at 55°C for 60 min prior to the enzyme assay. The specific activities were determined by measuring the radioactivity in the 1-butanol-extractable material after the reaction using DMAPP, GPP, all-(E)-FPP, and all-(E)-GGPP as allylic substrates, and the reaction products of the mutated enzymes were analyzed ( Fig. 4 and Table I). All mutated enzymes gave polyprenyl diphosphates with chain lengths longer than GGPP. The specific activity of each mutant for DMAPP, GPP, or FPP was nearly similar to that of the wild-type enzyme for the same allylic substrate. The five mutated and wild-type enzymes prefer DMAPP to the other three allylic substrates. However, the activity for GGPP and the distribution of products obtained from various allylic substrates were different in each case. Mutant 1 produced GFPP and GGPP as the main products when DMAPP, GPP, or FPP was used as an allylic substrate. A small amount of HPP was also formed when DMAPP was used. The distribution of products depended on the allylic substrates used. In the case of mutant 2, GGPP was the major product, and the amount of GFPP was ϳ10% of the total amount of products. HPP could not be detected. Mutants 3 and 5 are similar in that they show strong GFPP synthase activities and weak HPP synthase activities. Also, these mutants produced a large amount of FPP when DMAPP or GPP was used as an allylic substrate. The amount of GPP was only different between mutants 3 and 5 when DMAPP was used. In mutant 4, the major product was GGPP, and the amount of GFPP was ϳ15% of the total amount of products. When GPP was used, FPP was effectively produced. This might reflect the competition of GPP and FPP that was formed by a single condensation and already bound to the enzyme. Such substrate competition was observed in all mutants and the wild-type enzyme, although the magnitude differed in each case.
Determination of Deduced Amino Acid Sequences of Mutants- Fig. 5 represents the substitutions of amino acids deduced from the nucleotide sequences of the mutated GGPP synthases. All selected mutants had a few mutations in the nucleic acid sequences, although the chemical treatment used here was slightly weaker than that reported previously (35). Mutant 1 had four coding mutations resulting in F77S, M85I, R199K, and D312N. Mutant 2 contained two nucleic acid changes, in which one substitution resulted in an amino acid alteration (F118L). Mutant 3 also contained two nucleic acid changes, in which one substitution resulted in an amino acid alteration (F77S). Mutant 4 contained two coding mutations (F77L and V99M) and two silent mutations. Mutant 5 possessed two coding mutations (F77S and Y101H). In these mutations, nonaromatic amino acid residues tended to substitute for aromatic amino acid residues. Especially phenylalanine at position 77 was found to be replaced in four of the five mutants.

DISCUSSION
In this paper, we have tried to convert GGPP synthase from S. acidocaldarius to a polyprenyl-diphosphate synthase that catalyzes condensation beyond GGPP using random mutagenesis and phenotypic screening. An essential element of this strategy is an efficient identification of mutants of interest among a larger population of variants. For this purpose, we used a suppression system using the yeast mutant C296-LH 3 , which is deficient in HPP synthase. It is known that the polyprenyl-diphosphate synthase involved in coenzyme Q synthesis is found to be associated with the inner mitochondrial membrane (39). S. acidocaldarius GGPP synthase has neither an apparent transmembrane sequence nor what appears to be a typical sequence for transport into mitochondria, which contains positively charged and hydroxylated amino acids. Therefore, before screening, we were worried about whether or not the mutated GGPP synthase expressed in the cytosol of yeast could suppress the pet phenotype even if the enzyme had HPP synthase activity. However, success in our screening clearly shows that the overexpression of the mutated GGPP synthase in the cytosol complements the synthesis of functional coenzyme Q in mitochondria. However, it remains obscure whether the mutated GGPP synthase is transported into the mitochondria and then produces polyprenyl diphosphate in the mitochondria or whether the polyprenyl diphosphate synthesized in the cytosol moves into the mitochondria.
We obtained five mutants from 1,400 clones of the library and determined the deduced amino acid residues changed and the reaction products. These mutants showed different properties, as summarized below.

TABLE I Product distribution of mutated GGPP synthase reactions using various allylic diphosphates
After the reaction of 25 nmol of [1-14 C]IPP (37 GBq/mol) and 25 nmol of the indicated allylic diphosphate with 100 g of the GGPP synthases, the radioactivity in the 1-butanol-extractable material was determined. The products were then hydrolyzed with acid phosphatase, and the hydrolyzed alcohols were separated by reversed-phase LKC-18 TLC as described under "Experimental Procedures." Radioactivities in every product were determined. Mutant 1 contained alterations of four amino acids (F77S, M85I, R199K, and D312N). The replacement of phenylalanine at position 77 seems to be essential for elongation of the ultimate product because mutant 3, which has a single mutation of F77S, catalyzes the condensation beyond GGPP. This enzyme produced similar amounts of GFPP and GGPP as main products when DMAPP or FPP was used as a priming substrate. A small amount of HPP was formed when DMAPP was used. Mutant 2 contained one amino acid alteration (F118L). This enzyme produced GFPP from any primers used, but the main product was GGPP. Mutant 3 contained one amino acid alteration (F77S) and showed a strong GFPP synthase activity and a weak HPP synthase activity. Mutant 4 contained alterations of two amino acids (F77L and V99M) and showed GFPP synthase activity. Mutant 5 had alterations of two amino acids (F77S and Y101H). The product distribution was similar to that in mutant 3 except that the amount of GPP from DMAPP was greater than that in mutant 3.
As described above, the in vitro product analysis of the mutated enzymes showed that the major product was GFPP and that overexpression of wild-type GGPP synthase partially suppressed the phenotype of the yeast mutant C296-LH 3 . Although the lengths of the prenyl chain of coenzyme Q in the mutants have not been analyzed, it is conceivable that the length of the side chain of ubiquinone is determined by the specificity of polyprenyl-diphosphate synthase, not by 4-hydroxybenzoate polyprenyltransferase, and that coenzyme Q 5 can play a physiological function in place of coenzyme Q 6 in yeast. It has been reported that the change in side chain length alters the diffusion of coenzyme Q, which is concerned with the rate of electron transfer from NADH dehydrogenase and succinate dehydrogenase to the bc 1 complex (40). So far, viral infection (8), treatment with detergent (41), and tissue differences (9) have been reported to bring about some change in the side chain length of coenzyme Q. Therefore, it would be quite interesting to determine the length of the side chain of coenzyme Q in these mutants and the rate of electron transfer in the HPP synthase-deficient strains with the mutated GGPP synthase genes.
The success in the mutagenic conversion of GGPP synthase to the polyprenyl-diphosphate synthase that catalyzes the condensation beyond GGPP led us to elucidate that the two amino acids, Phe-77 and Phe-118, play essential roles in chain length determination. Especially Phe-77, whose change to serine (mu-tants 1, 3, and 5) or leucine (mutant 4) was shown to cause the change in the ultimate product, is remarkable. Also, the fact that four out of five mutants have a similar mutation at the same position confirms the significance of the amino acid on chain length determination. During the past few years, the structural genes for several prenyltransferases have been identified and characterized. Prenyltransferases have two conserved DDXX(XX)D aspartate-rich motifs, which are thought to be binding sites for the diphosphate moieties of IPP and allylic substrates. Phe-77 is located five amino acids upstream from the first DDXXD motif (22). Comparison of the amino acid sequences of prenyltransferases around position 77 shows a unique profile (Fig. 6). All FPP synthases and GGPP synthases originating from archaebacteria possess an aromatic amino acid, phenylalanine or tyrosine, at the position five amino acids upstream from the first DDXXD motif. On the other hand, prenyltransferases that catalyze the condensation beyond GGPP and GGPP synthases originating from organisms other than archaebacteria have a nonaromatic amino acid at the corresponding position. Our recent study with Bacillus stearothermophilus FPP synthase has shown that the substitution of histidine for tyrosine at position 81, which corresponds to position 77 of S. acidocaldarius GGPP synthase, brings about the conversion of FPP synthase to GGPP synthase (32). Both findings clearly demonstrate that the amino acid at this position is crucial for the determination of the ultimate chain length. It FIG. 5. Amino acid substitutions deduced from nucleotide sequences encoding mutated GGPP synthases. In mutant 1, the nucleotide changes were T to C at position 230, T to C at position 231 (F77S), G to A at position 255 (M85I), G to A at position 596 (R199K), and G to A at position 934 (D312N). In mutant 2, the nucleotide changes were G to A at position 330 (silent) and T to C at position 352 (F118L). In mutant 3, the nucleotide changes were T to C at position 230 (F77S) and G to A at position 771 (silent). In mutant 4, the nucleotide changes were T to C at position 229 (F77L), G to A at position 295 (V99M), T to C at position 426 (silent), and T to C at position 585 (silent). In mutant 5, the nucleotide changes were G to A at position 129 (silent), T to C at position 230 (F77S), and T to C at position 301 (Y101H). Arrowheads correspond to the positions of the two DDXXD motifs.
FIG. 6. Comparison of amino acid sequences around first aspartate-rich motif of several prenyltransferases. A comparison of the amino acid sequences that were previously reported was carried out. Amino acid sequences identical to those of S. acidocaldarius GGPP synthase are boxed. Numbers indicate residue positions in the primary sequences. The arrowhead corresponds to position 77 of S. acidocaldarius GGPP synthase, and aromatic amino acids at this position are seems likely that phenylalanine or tyrosine directly interferes with the final product so that the chain elongation cannot continue any more. Recently, Chen et al. (2) compared the amino acid sequences of all-(E)-prenyl-diphosphate synthases and proposed a phylogenetic tree. In their paper, they postulated that the bifunctional archaebacterial GGPP synthase, which produces FPP as an intermediate, is an ancient enzyme and that HPP synthase was first separated from the bifunctional GGPP synthase before segregation into three kingdoms, Archaea, Eubacteria, and Eucarya. After segregation into the three kingdoms, separation between FPP synthase and GGPP synthase occurred. Our finding clearly supports this evolutionary scenario.
Although we have compared the present mutations of GGPP synthase with the mutations of FPP synthase previously reported (32) and have pointed out the common importance of the aromatic amino acid at the position five amino acids before the first DDXXD motif, the mutated GGPP synthases do not show any decrease in the total reaction rate or any decrease in activity for DMAPP. These phenomena have been observed in the mutated FPP synthase. It is unclear whether or not the phenomena observed in the mutated FPP synthase directly relate to the determination of chain length. We have been analyzing the mutated GGPP synthases and the mutated FPP synthases in order to understand the common mechanism of chain length determination.