Elucidation of the deficiency in two yeast coenzyme Q mutants. Characterization of the structural gene encoding hexaprenyl pyrophosphate synthetase.

The assembly of a polyisoprenoid side chain and its transfer to para-hydroxybenzoate are the first two steps of coenzyme Q biosynthesis. In yeast these reactions are catalyzed by hexaprenyl pyrophosphate synthetase and PHB:polyprenyltransferase, respectively. We have screened nine complementation groups of yeast coenzyme Q mutants for the activities of these two enzymes and found two strains deficient in either activity. The strain deficient in hexaprenyl pyrophosphate synthetase activity, C296-LH3, is complemented by the plasmid pG3/T1. When C296-LH3 was transformed with a shuttle vector containing a 2,187-base pair fragment from the genomic insert of pG3/T1, both glycerol growth and hexaprenyl pyrophosphate synthetase activity were restored. The activity of the latter enzyme was higher than that seen in wild-type yeast. The increase in activity could be attributed to a gene dosage effect of the multi-copy plasmid. A 1,419-base pair open reading frame encoding a 52,560-dalton protein was found on the genomic fragment. The size of the RNA transcript and the location of transcriptional initiation indicate that the entire open reading frame is contained within the mRNA. Comparison of the hexaprenyl pyrophosphate synthetase amino acid sequence with amino acid sequences from the related enzyme farnesyl pyrophosphate synthetase show the presence of three highly conserved domains. Within two of the domains is an aspartate-rich motif found invariantly in the amino acid sequences of farnesyl pyrophosphate synthetase from three species and the hexaprenyl pyrophosphate synthetase amino acid sequence reported here. These aspartic acid motifs may comprise binding sites for the allylic and homoallylic substrates. The hydrophobicity profiles of the hexaprenyl pyrophosphate synthetase sequence and the farnesyl pyrophosphate synthetase sequence from rat appear similar. Furthermore, the hydrophobicity correlation coefficient of the comparison of these two sequences indicate with a high degree of confidence (p less than 0.001) that the two proteins will fold into similar three-dimensional structures.

The assembly of a polyisoprenoid side chain and its transfer to para-hydroxybenzoate are the first two steps of coenzyme Q biosynthesis.
In yeast these reactions are catalyzed by hexaprenyl pyrophosphate synthetase and PHB:polyprenyltransferase, respectively. We have screened nine complementation groups of yeast coenzyme Q mutants for the activities of these two enzymes and found two strains deficient in either activity.
The synthesis of a variety of products including sterols, dolichols, and coenzyme Q. Interestingly, the branches of the pathway leading to these three products each contain polyprenyl synthetase enzymes which catalyze a l'-4 condensation reaction between 5 carbon isoprene units (for review see Poulter and Rilling, 1981). The most active and certainly the best understood of these enzymes is farnesyl pyrophosphate synthetase. This enzyme catalyzes the truns-addition of two molecules of isopentenyl pyrophosphate onto dimethylallyl pyrophosphate to form farnesyl pyrophosphate.
The polyprenyl synthetase of dolichol biosynthesis catalyzes the cis-addition of between 8 and 20 isopentenyl pyrophosphate molecules onto farnesyl pyrophosphate.
In contrast, the polyprenyl synthetase of coenzyme Q biosynthesis catalyzes the formation of all transpolyprenyl pyrophosphates generally ranging in length of between 6 and 10 isoprene units depending on the species (Olson and Rudney, 1983).
Coenzyme Q functions as a lipid-soluble electron transporter between lipoprotein complexes of the mitochondrial respiratory chain. As originally proposed (Green, 1966), coenzyme Q is thought to shuttle electrons from NADH dehydrogenase (Complex I) and succinate dehydrogenase (Complex II) to reduce the bc, complex (Complex III). The rate of electron transfer between the dehydrogenases and the bcl complex was shown to be limited by coenzyme Q diffusion which is in turn determined by the length of the isoprenoid side-chain (Schneider et al., 1982). Perturbations in the mitochondrial membranes by detergent (Nishino and Rudney, 1977) or by viral infection (Casey and Bliznakov, 1973) were shown to alter the length of the isoprenoid side-chain of the coenzyme Q molecule. As with quinone molecules from other electron transport systems, coenzyme Q homologues are classified on the basis of the number of isoprene units in their polyisoprenoid moieties. The particular homologue present in an individual is thought to be a species-specific phenomenon but tissue-specific differences have been noted (Nazir and Magar, 1964). Although heterologous coenzyme Q homologues can function in uitro, changes in the specificity of the polyprenyl synthetase could in theory alter electrochemical processes.
Although normally acquired through dietary intake, the major source of coenzyme Q is derived from de nouo synthesis and at least one human disease has been attributed to a defect in its biosynthesis (Ogasahara et al., 1989). In contrast to higher eukaryotes, yeast can grow aerobically by oxidative phosphorylation or anaerobically by the fermentation of glucose. This has enabled investigators to isolate mutant strains which are deficient in specific metabolic steps. Tzagoloff et ~1. (1975Tzagoloff et ~1. ( , 1990 isolated nine separate complementation groups of yeast which were defective in coenzyme Q. Presumably, these strains are deficient in specific steps of coenzyme 13158 Hexaprenyl Pyrophosphate Synthetase & biosynthesis. In this paper we report the identification of the deficiency in two of the nine complementation groups of coenzyme Q mutants. In addition, we describe the isolation and characterization of the structural gene encoding hexaprenyl pyrophosphate synthetase, the first committed enzyme of coenzyme Q biosynthesis.

EXPERIMENTAL PROCEDURES
Materials-The wild-type yeast strain D273-lOB/Al (a m&S) and the set of nine coenzyme Q mutants strains (Tzagoloff et al., 1975(Tzagoloff et al., , 1990 Lang et al. (1977). Yeast were grown to mid to late log phase in YPD media containing 1% yeast extract, 2% bacto-peptone, and 2% a-D-glucose at 30 "C!. All subsequent steps were carried out at 4 "C. The cells were first pelleted by centrifugation at 2,500 x g for 1 min and washed once with H*O. Cells were resedimented at 2,500 X g for 1 min and resuspended in buffer A (0.4 M sorbitol, 0.5 M mannitol, and 20 mM Tris-HCl, pH 7.4). The cells were centrifuged again at 2,500 X g for 1 min and resuspended in an equal volume of buffer B (buffer A diluted 2:l (v/v) with water). Sterile glass beads were added to the cells at a ratio of 4:l (v/v) to the packed cell volume. The cells were disrupted by vigorously shaking the tubes 2 min at room temperature.
The glass beads and particulate matter were removed by centrifugation at 100 x g for 1 min and the supernatant fraction saved. The pellet was washed with buffer B and the supernatant fractions pooled. The supernatant was centrifuged for 1 min at 100 x g, accelerated to 400 X g for 4 min, and then accelerated to 3,500 rpm for 5 min.  (Dieckmann and Tzagoloff, 1983). The strain C296-LH3 (a leu2 his3 pet) was grown to an ODW = 1.0 and harvested by centrifugation at 2,500 X g for 3 min. Cells were resuspended into Hz0 and centrifuged as above. The cell pellet was resuspended into TSB (0.1 M Tris sulfate, pH 9.4, 30 mM P-mercaptoethanol) and incubated at 30 "C for 15 min. Cells were washed once in 1 M sorbitol and resuspended into 10 ml of YPSG (1% yeast extract, 2% Bacto-peptone, 1.0 M sorbitol, and 1% glucose).
Oxalyticase (Enzogenetics) was added to a concentration of 3 units/ODsW and allowed to digest for 20 min at 30 "C. The resulting spheroplasts were washed twice in 1 M sorbitol and once in CaS (1.0 M sorbitol and 10 mM CaClz), and finally resuspended in CaS at 50 ODW/ml. 1-5 pg of DNA was added to each loo-n1 aliquot of the spheroplasted yeast cells. 0.5 ml of PEG solution (20% polyethylene glycol3350, 10 mM CaC12, and 10 mM Tris-HCl, pH 7,4) was added to the cells and let stand 5 min.  (Hill et al., 1986) to generate YEpG3 by standard methods (Maniatis et al., 1982). To determine the minimal region of the insert necessary for complementation, restriction enzymes whose sites were present once in the polylinker region and once in the insert were used to digest YEpG3.
The resulting plasmids were then re-ligated and tested for their ability to complement. DNA sequence was determined by the dideoxynucleotide chain termination method (Sanger et al., 1977)  The ratio of the observed ra value to the standard deviation of the mean FH after 10 such randomizations and alignments was then used in a t test for significance.

RESULTS
Tzagoloff and co-workers (1975,1990) isolated nine individual complementation groups of yeast that would not grow on glycerol unless exogenous coenzyme Q2 was added to the media (Table I). Presumably, these strains are defective in coenzyme Q biosynthesis.
Our attempts to directly measure the activity of the polyprenyl synthetic enzyme of the pathway by looking for the synthesis of polyprenyl pyrophosphates were unsuccessful. Therefore, the activities of hexaprenyl pyrophosphate synthetase and PHB:polyprenyltransferase were determined by the formation of 3-polyprenyl-4-hydroxybenzoates (Casey and Threlfall, 1978). This method of ana- lyzing both enzyme activities is facilitated by the apparent lack of specificity of PHB:polyprenyltransferase (Daves et al., 1967;Nishino and Rudney, 1977). Hence, in the absence of endogenous polyprenyl synthesis, PHB:polyprenyltransferase will alkylate PHB with any available allylic prenyl pyrophosphate.
In Vitro Synthesis of 3-Polyprenyl-4-hydroxybenzoates by Sonicated Mitochmzdria- Casey and Threlfall (1978) described the formation of 3-diprenyl-4-hydroxybenzoate and 3-hexaprenyl-4-hydroxybenzoate in vitro under a variety of conditions. The authors identified the products by differential isotope labeling as well as exhaustive chemical derivitizations. To confirm the formation of the same products in our assay system we carried out the experiment shown in Fig. 1 with wild-type yeast (D273-lOB/Al) mitochondria.

Screening the Yeast Coenzyme Q Mutants for HPS and PHB:Polyprenyltransferase
Activities-Seven of the nine coenzyme Q mutant complementation groups showed a normal ability to synthesize 3-hexaprenyl-4-hydroxybenzoate (data not shown). Two complementation groups, however, were deficient in either HPS or PHB:polyprenyltransferase activity. Wild-type mitochondria ( Fig. 2A) synthesize both 3-diprenyl-4-hydroxybenzoate and 3-hexaprenyl-4-hydroxybenzoate under the standard assay conditions (33 mM MC). The strain C33 was unable to synthesize any substituted hydroxybenzoates (Fig. 2B). This reflects an absence of any detectable PHB:polyprenyltransferase activity. In contrast, C296 synthesizes only 3-diprenyl-4-hydroxybenzoate reflecting an absence of HPS activity (Fig. 2C). Since the HPS enzyme is nonfunctional, the strain C296 is unable to elongate geranyl pyrophosphate. Therefore, under these conditions the enzyme PHB:polyprenyltransferase transfers the only available allylic pyrophosphate (geranyl pyrophosphate) to PHB to form 3-diprenyl-4-hydroxybenzoate. This situation is analogous to the omission of IPP from the assay with wild-type mitochondria (Fig. 1D). Based on these results we have named the two loci which are required for these reactions coql and coq2 to reflect that they are implicated in the first and second steps of coenzyme Q biosynthesis, respectively. Thus, C296 which is deficient in HPS carries the coql-I allele and C33 which is deficient in PHB:polyprenyl transferase carries the coq2-1 allele.
Restoration of HPS Activity in C296-LH3-Transformation of C296-LH3 with the plasmid pG3/Tl restores growth on glycerol-containing media, implying that the strain has a competent electron transport chain and thus, functional coenzyme Q. To determine whether the ability of C296-LH3 harboring pG3/Tl to grow on glycerol was accompanied by wildtype HPS activity, assays to detect the formation of 3-hexaprenyl-4-hydroxybenzoate were performed. Under the standard assay conditions wild-type mitochondrial membrane-rich fractions synthesize both 3-diprenyl-4-hydroxybenzoate and 3-hexaprenyl-4-hydroxybenzoate and serve as standards ( Fig.  3A  4-hydroxybenzoate in vitro (Fig. 3B). Alternatively, when C296-LH3 is complemented by pGS/Tl to restore growth on glycerol, HPS activity is also restored as indicated by the formation of 3-hexaprenyl-4-hydroxybenzoate (Fig. 3C). The amount of radioactivity associated with the 3-hexaprenyl-4hydroxybenzoate peak was approximately 2-fold higher in mitochondrial fractions from C296-LH3 harboring the pG3/ Tl plasmid compared with that from wild-type yeast (data not shown). This is probably due to a gene dosage effect since pG3/Tl is a multi-copy plasmid containing the 2~ circle origin of replication. Sequence of the HPS Gene-To locate the gene within the 4.5-kb genomic fragment contained in the plasmid YEpG3, complementation analyses were performed. The smallest region of DNA capable of complementing C296-LH3 was found to be a 2,187-bp S&I-BamHI restriction fragment. This region was sequenced in both directions by the strategy shown in Fig. 4. Sequence analyses revealed the presence of a 1,419-bp open reading frame coding for a 473-amino acid protein with a calculated molecular weight of 52,560 daltons (Fig. 5) Nucleotide sequence is shown from the 2187-base pair SstI-BarnHI fragment. A poly(dA. dT) region, two TATA sequences and two polyadenylation signals are underlined. Two start-sites of transcription are indicated by asterisks.
analyses where total yeast RNA from C296-LH3 and wildtype yeast was hybridized to the original 4.5-kb genomic fragment revealed a single RNA band approximately 1.5 kb in size (data not shown). To map the site of transcriptional initiation Sl nuclease protection and primer extension analyses were performed (Fig. 6). Extension of the primer Ul (see Fig. 4) resulted in the synthesis of two products, 248 and 253 bases in length (Fig. 6). When the 340-base Sl probe was hybridized to yeast poly(A+) RNA two distinct bands are protected which are the same size as those generated from the primer extension experiment (Fig. 6). These two methods corroborate each other since the probe used in the primer extension experiment was the same as was used to synthesize the Sl probe. Taken together these results indicate there are two major start-sites of transcription located 30 and 35 base pairs upstream of the ATG signal (Fig. 5).
Amino Acid Comparison of Yeast HPS and Farnesyl Pyrophosphate Synthetase-Alignments of amino acid sequences between yeast HPS and rat FPS were carried out by the method of Needleman and Wunsch (1970). The comparison revealed 22% of the amino acids were identical and 45% were conservative substitutions. Although the global sequence identity was not significant, three domains showing an impressive sequence identity were apparent. Fig. 7 shows an alignment of the three domains with the yeast HPS and FPS from rat (Clarke et al., 1987;Teruya et al., 1990), human (Sheares et al., 1989;Wilkin et al., 1990) and yeast (Anderson et al., 1989). The schematic diagram indicates the location of the domains within the polypeptides as well as the location of cysteine residues. The HPS amino acid sequence contains 5 cysteine residues while each FPS sequence contains 6 cysteines. Only 3 of the cysteines are conserved in FPS between the three species and their locations are indicated in Fig. 7.
Structural Comparisons-Initial hydrophobicity comparisons between the yeast HPS and rat FPS showed similar profiles especially in the carboxyl-terminal 150 amino acids (Fig. 8). Since proteins which fold into similar three-dimensional structures have highly correlated hydrophobicities we sought to determine the degree of similarity of hydrophobicities between the yeast HPS and rat FPS amino acid sequences. The hydrophobicity correlation coefficient, rH, is calculated by comparing the hydrophobicities of each pair of amino acids from two aligned sequences (Sweet and Eisenberg, 1983). Table II shows the calculated  rH values from  alignments of the yeast HPS and the rat FPS sequence as well as the randomized yeast HPS sequence aligned with the rat FPS amino acid sequence. The rH value of 0.3 or greater from two properly aligned sequences indicates similar threedimensional structures. Alignment of the HPS and FPS amino acid sequences produces an rH value of 0.3138 or 0.3055 depending on the hydrophobicity scale. This indicates that the two polypeptides will fold into similar structures. Align-PE  &-,  RNA  ,AjG,C,T,  I IAIGICITI~~'I~I'I~~'I   , : .
ments of ten randomized HPS amino acid sequences with FPS results in rH mean values of 0.2492 or 0.2043 which, as expected, are not significant.
In a further test of significance the difference between the observed and randomized rH values was found to have a confidence level of greater than 99.9%. DISCUSSION We have analyzed a set of yeast coenzyme Q mutants and found two strains defective in either the first or the second step of the coenzyme Q biosynt,hetic pathway. The strain C296 is deficient. in hexaprenyl pyrophosphate synthetase activity and C33 was found to be deficient in PHB:polyprenyltransferase activity. These findings are based on the observation that neither strain can synthesize 3-hexaprenyl-4-hydroxybenzoate, a naturally occurring intermediate of coenzyme Q biosynthesis in yeast (Winrow and Rudney, 1969;Olson and Rudney, 1983). In our in vitro assay system C296 synthesizes 3-diprenyl-4-hydroxybenzoate because it lacks functional HPS activity and is unable to elongate exogenous geranyl pyrophosphate by the classic 1'.4 condensation reaction. Therefore, geranyl pyrophosphate is the only allylic pyrophosphate available for PHB:polyprenyltransferase to add onto PHB. The strain C33 is unable to form any prenylated hydroxybenzoates despite the presence of allylic pyrophosphates since it, is deficient in PHB:polyprenyltransferase.
The inability to detect 3-triprenyl-4-hydroxybenzoate in our assay system is of interest since several groups (Momose and Rudney, 1972;Poulter and Rilling, 1981)" have detected FPS activity in mitochondria.
This mitochondrial activity represents only a small fraction of the cytosolic FPS activity ' M. Ashhy and P Edwards, unpublished data and may result from cytosolic contamination despite repeated washing of the mitochondria.
In experiments from which we based our own assay system, Casey and Threlfall (1978) showed that dimethylallyl pyrophosphate, geranyl pyrophosphate, or farnesyl pyrophosphate served about equally well as allylic primers for the synthesis of 3-hexaprenyl-4-hydroxybenzoate. These authors showed that the formation of 3triprenyl-4-hydroxybenzoate was detected only when farnesyl pyrophosphate was added exogenously. Furthermore, when mit.ochondria were incubated with farnesyl pyrophosphate at 33 mM MgCl,, the sole product observed was 3-triprenyl-4hydroxybenzoate indicating that when farnesyl pyrophosphate is present (albeit at a saturating concentration) it will serve as a substrate for the PHB:polyprenyltransferase enzyme. Our experiments (Figs. l-3) corroborate those of Casey and Threlfall (1978) in that the formation of triprenyl hydroxybenzoate was never observed when mitochondria were incubated with isopentenyl pyrophosphate (IPP), geranyl pyrophosphate, and PHB at high or low Mg'+ concentrations. There is a significant amount of FPS activity in our mitochondrial preparations relative to the rate of synthesis of polyprenyl hydroxybenzoates (data not shown). The inability to observe the synthesis of triprenyl hydroxybenzoate, especially in the HPS mutant C296, suggests that the HPS and PHB:polyprenyltransferase reactions may be coupled. This would enable these two enzymes to synthesize a single chain length polyprenyl hydroxybenzoate despite the presence of other available polyprenyl pyrophosphates and the inherent lack of specificity of PHB:polyprenyltransferase.
The structural gene encoding HPS was isolated from the plasmid pGB/Tl which was able to complement C296-LHn and restore growth on glycerol. The gene was located on a 2187-base pair SstI-BamHI fragment of genomic DNA on the plasmid. When this region of DNA was transformed into C296-LHZi on a multi-copy plasmid the yeast acquired a glyc-erol+ phenotype and regained HPS activity to a higher level than that observed in the wild-type strain D273-lOB/Al. This probably resulted from a gene dosage effect and provided evidence that the SstI-RamHI fragment contained the structural gene for HPS.
The sequence of the 2,187-base pair SstI-BamHI fragment revealed an open reading frame I>419 base pairs in length coding for a 473-amino acid protein with a calculated molecular mass of 52,560 daltons. Approximately 150 base pairs upstream of the initiator ATG site is a poly(dA.dT) region which has been reported to be a common feature in the promoters of constituitively expressed yeast genes (Struhl, 1985). Also present in the sequence are two TATA-like sequences located 29 and 34 base pairs upstream of the transcriptional initiation site. In replacement and deletion studies on transcriptional initiation of the yeast HIS4 gene (Nagawa and Fink, 1985) and the CYCl gene (Hahn et al., 1985) it was reported that initiation takes place at particular initiation sequences which are located within a window of 60-120 nucleotides downstream of the TATA element. The two TATAlike sequences in the HPS 5' region appear to be located too close to the transcription start-site and whether they are functional in uiuo remains to be determined. TATA elements are not an absolute requirement of yeast promoters. For example, the 7'KP3 promoter lacks the conserved TATAAA sequence and inst.ead contains a GCN4 binding site 28 nucleotides upstream of the mRNA start-site (Chen and Struhl, 1989). In the 3'.untranslated region of the HPS gene two polyadenylation signals are located 15 and 29 base pairs downstream from the termination codon TAA (Fig. 3) Dayhoff et al. (1979). b Hydrophobicity scale from Sweet and Eisenberg (1983).
DNA capable of complementing C296-LHS, identified only a single mRNA species approximately 1.5 kb in length (data not shown). The RNA mapping studies confirmed that the 1419-base pair open reading frame of HPS (Fig. 5) is in fact transcribed and present in the mRNA population. Polyprenyl synthetases have not received much attention and little is known about them in contrast to the cholesterogenie enzyme FPS. In all species examined, the polyprenyl synthetase involved in coenzyme Q synthesis is found to be associated with the inner mitochondrial membrane (Pennock and Threlfall, 1983). A survey of the amino acid sequence of HPS does not reveal any apparent transmembrane sequences and suggests that HPS may be peripherally associated with the inner mitochondrial membrane. The amino-terminal region of the HPS amino acid sequence does contain what appears to be a typical mitochondrial leader sequence as judged by the preponderance of positively charged and hydroxylated residues, a lack of acidic residues and a tendency to form cu-helices (Hart1 et al., 1989). However, the aminoterminal region of the sequence does not contain the highly conserved three-amino acid motif present in most mitochon-drial leader sequences which are processed in two steps (Hendrick et al., 1989). Comparison of the yeast HPS amino acid sequence with the rat FPS sequence did not reveal any major global similarities. Although the two enzymes carry out the identical condensation reactions differing only in the chain length of the final product, we expected several differences since FPS exists as a dimer and is located in the cytoplasm while HPS is membrane associated in mitochondria. The comparison did show, however, the presence of three highly conserved domains which were also present in the amino acid sequence of FPS from human and yeast (Fig. 7). A closer examination revealed the presence of two aspartate-rich sequences present in domains I and II of all four polyprenyl synthetase sequences (Fig. 7). The consensus sequence (I, L, or V)XDDXXD, where X can be any amino acid, was used to search the NBRF protein data base to look for similar sequences. Out of approximately 12,500 sequences we found the consensus sequence to occur in about 1% of the protein sequences or less than once in every 25,000 amino acids. We were not able to find any proteins which contained repeats of the consensus sequence other than the prenyl synthetases mentioned above. Based on the protein data base searches about one in every 10,000 proteins would be predicted to contain a repeat of the consensus sequence. The occurrence of a repeat of the aspartate-rich consensus sequence in all prenyl synthetase enzymes reported to date implies a functional and perhaps phylogenetic relationship exists between these proteins. A common feature between HPS and FPS is the binding to both the homoallylic isopentenyl pyrophosphate and to an elongating allylic polyprenyl pyrophosphate.
In binding experiments on FPS, King and Rilling (1977) reported evidence which strongly suggests that both the homoallylic and allylic substrates bind as their respective magnesium salts. We propose that the conserved aspartate repeats are involved in facilitating the binding of the substrates by forming magnesium salt bridges between the substrates and the catalytic site. There would be precedence for this type of ionic interaction. Brenner (1987) found a short conserved region containing an invariant pair of aspartic acid residues between the two ATP-binding enzymes aminoglycoside phosphotransferase and viomycin phosphotransferase.
The author postulated that the aspartates were involved in forming salt bridges through the magnesium of the phosphate groups of ATP. Additionally, aspartic acid 66 of phospholipase A, was found to control the binding of calcium (Van den Bergh et al., 1989). If the aspartate motifs found in prenyl synthetases are involved in the binding of substrates then it would be easy to envision the hydrophobic amino acid of the consensus se-quence stabilizing binding by interacting with the hydrocarbon tails of the allylic and homoallylic substrates. In addition to prenyl synthetase enzymes other enzymes which prenylate nonisoprene substrates have been isolated. These prenyltransferase enzymes are involved in tRNA modification (Najarian et al.,198'7) and cytokinin biosynthesis (Goldberg et al., 1984;Strabala et al., 1989). A search through these sequences for the presence of the aspartate consensus sequence was unsuccessful. This is not altogether surprising, though, since these enzymes must accommodate different electrophilic substrates which undoubtably impose different structural constraints on the catalytic sites. Of interest, however, was the finding that the yeast MOD5 amino acid sequence, an enzyme which prenylates tRNAs with dimethylallyl pyrophosphate (Najarian et al., 1987), contains 5 of the 7 invariant residues of Domain II, including the 3 aspartic acids (PEPLFQRLDDRVD). The common substrate of prenyl synthetases and non-isoprene prenylating enzymes is allylic pyrophosphates. This may suggest that Domain II comprises an allylic binding site.
The number of polyprenyl synthetase enzymes in vertebrates represents only a small fraction of that found in plants, fungi, and bacteria. In this respect, the identification of peptide domains critical to the l'-4 condensation reaction may facilitate the isolation of other members of this diverse family.