Identification of a Short (C15) ChainZ-Isoprenyl Diphosphate Synthase and a Homologous Long (C50) Chain Isoprenyl Diphosphate Synthase inMycobacterium tuberculosis *

We report the cloning, overexpression, and partial characterization of two unique Z-isoprenyl diphosphate synthase homologs from Mycobacterium tuberculosis. The first enzyme, Rv1086, adds one isoprene unit to ω,E-geranyl diphosphate. The product, ω,E,Z-farnesyl diphosphate, is the putative substrate of the second enzyme, Rv2361c. This enzyme adds seven more isoprene units to ω,E,Z-farnesyl diphosphate and releases decaprenyl diphosphate. Both open reading frames were cloned from the M. tuberculosis H37Rv genome and overexpressed in M. smegmatis. Membrane and cytosol fractions from wild type and the two recombinant strains were assayed for [14C]isopentenyl diphosphate incorporation into isoprenyl diphosphates in the presence of various allylic isoprenyl diphosphate acceptors. Membrane fractions of recombinant cells overexpressing Rv2361c incubated with farnesyl diphosphate showed a 10-fold increase of [14C]isopentenyl diphosphate incorporation into decaprenyl diphosphate. Membrane fractions of recombinant cells overexpressing Rv1086 incubated with geranyl diphosphate showed a 5-fold increase of [14C]isopentenyl diphosphate incorporation into farnesyl diphosphate. Analysis of the stereochemistry revealed that all of the overexpressed farnesyl diphosphate was in the ω,E,Z-configuration. This is the first description of a short chain isoprenyl diphosphate synthase that generates products with Z-stereochemistry. Previously, all known short chain isoprenyl diphosphate synthases catalyze the synthesis of products withE-stereochemistry.

The isoprenoid compounds are chemically diverse, with over 23,000 compounds currently characterized (1). Representative members of this family of compounds (cholesterol, quinones, carotenoids, polyprenyl phosphates, and rubber) display diversity in structure as well as function. Polyprenol phosphate (Pol-P) 1 is intimately involved in prokaryotic cell wall biosynthesis. In fact, evidence suggests that the rate of bacterial cell wall synthesis in vivo could be regulated by Pol-P levels (2)(3)(4)(5). Eubacteria usually contain a single Pol-P molecule (C 55 ) composed of 11 isoprene units: ,E,E,polyZ-undecaprenyl-P. 2 Mycobacteria are exceptions to this rule. Takayama et al. demonstrated that M. smegmatis possesses two unique Pol-P: 1) a heptaprenyl phosphate (C 35 ) with four saturated, two E, and one Z double bond and 2) a decaprenyl phosphate (C 50 ) with one E and eight Z double bonds (6). In 1998, Wolucka and de Hoffman (7) isolated a form of heptaprenyl phosphate that contained four saturated and three Z-isoprene units from M. smegmatis. All of these Pol-P molecules were isolated as mannosyl-1-phosphorylpolyaprenols (Pol-P-Man), and probably have roles in mannan and arabinomannan synthesis (8). M. tuberculosis appears to be more typical of other eubacteria than M. smegmatis, since it appears to contain a single Pol-P molecule, decaprenyl phosphate, whose stereochemistry has not yet been determined (9).
We have implicated Pol-P-Man in the biosynthesis of mycobacterial lipomannan and lipoarabinomannan; Pol-P-Man is the direct donor of mannose to the phosphatidyl-myo-inositol oligomannosides to give rise to phosphatidyl-myo-inositol oligomannoside-containing lipomannan and lipoarabinomannan (10). In addition, heptaprenyl phosphate is found in the form of mycolyl-6-␣-D-mannosyl-1-phosphorylheptaprenol and may function to carry mature mycolic acids across the plasma membrane (11). Arabinosyl-1-phosphoryldecaprenol is donor of the arabinofuranosyl residue in the arabinan of arabinogalactan, arabinomannan, and lipoarabinomannan (12). The disaccharide linker unit that bridges the arabinogalactan to the peptidoglycan is also formed while attached to Pol-P, which acts as the template for the synthesis of the entire mycolylarabinogalactan-linker unit complex (8). Pol-P (probably the decaprenyl phosphate) also has a role in mycobacterial peptidoglycan synthesis (5,13). Despite its obvious importance, the genetics and biochemistry of Pol-P synthesis have not been investigated in mycobacteria.
All known isoprenoids share a common biosynthetic mechanism beginning with the condensation of two five carbon precursors, isopentenyl diphosphate (IPP) and dimethylallyl diphosphate, to form geranyl diphosphate (GPP, C 10 ) (14). This and subsequent additions of IPP to the growing allylic diphosphate (to form farnesyl diphosphate (FPP, C 15 ) and geranylgeranyl diphosphate (GGPP, C 20 ), etc.) are catalyzed by a family of enzymes known as isoprenyl diphosphate synthases. The sequential addition of IPP to allylic diphosphate precursors continues until a physiologically relevant chain length is reached. At this point, the molecule can be dephosphorylated to form Pol-P.
Only a small fraction of the enzymes involved in isoprenoid chain elongation have been studied, and genetic information is available for only a subset of these (1). Amino acid alignments of these enzymes generated by Chen et al. in 1994 (15) and again by Kellogg and Poulter in 1997 (16) have defined five conserved regions, including two aspartic acid-rich (DD(XX) 2 D motifs. However, the resulting consensus sequences have to date been useful in identifying only those enzymes catalyzing the formation of E double bonds. Enzymes responsible for catalyzing the formation of Z double bonds are believed to be of a different family. Recently, a publication described the molecular cloning, expression, and purification of undecaprenyl diphosphate synthase, an enzyme from Micrococcus luteus that catalyzes Z-polyprenyl chain elongation of the allylic substrate ,E,E-FPP (17), the first such enzyme to have its amino acid sequence determined. As anticipated, it had no homology to any of the E-isoprenyl diphosphate synthases. Apfel et al. (18) published an alignment of 28 putative undecaprenyl diphosphate synthase homologs and demonstrated long chain prenyltransferase activity for three of them (Escherichia coli, Haemophilus influenzae, Streptococcus pneumoniae). Also, the Z-isoprenyl diphosphate synthase (dolichol synthase) from Saccharomyces cervisiae has been identified (19,20).
The determination of the complete genome sequence for Mycobacterium tuberculosis H37Rv has allowed for systematic exploration of gene function in this organism (21). We have identified two open reading frames in the M. tuberculosis H37Rv genome, Rv1086 and Rv2361c, whose predicted protein products have homology to the undecaprenyl diphosphate proteins identified by Shimizu et al. and Apfel et al. and homology to the yeast dolichol synthase (17)(18)(19)(20). The open reading frames of the putative M. tuberculosis Z-isoprenyl diphosphate synthases were cloned and overexpressed in an attempt to map the pathway for the synthesis of Pol-P.

EXPERIMENTAL PROCEDURES
Materials-[ 14 C]Isopentenyl diphosphate ([ 14 C]IPP, 55 mCi/mmol) was purchased from Amersham Pharmacia Biotech. Potato acid phosphatase (grade 2) was purchased from Roche Molecular Biochemicals. Kanamycin, farnesol (mixed stereoisomers), ,E,E-farnesol, ,E-geraniol, ,E,E,E-geranylgeraniol were purchased from Sigma. ,E,E-Farnesyl diphosphate and ,E-geranyl diphosphate were synthesized as described by Davisson et al. (22). Authentic prenols of various chain lengths were purchased from the Institute of Biochemistry and Biophysics, Polish Academy of Sciences (Warsaw, Poland). Kieselgel 60 F254 TLC plates were from EM Science, and Baker Si-Reverse Phase C 18 TLC plates were from J. T. Baker Inc. Restriction enzymes NdeI and HindIII, LB agar and LB broth, DNTPs, and T4 DNA ligase were purchased from Life Technologies, Inc. QIAprep Spin Miniprep Kit was purchased from Qiagen Inc. Escherichia coli competent cells (XL-1 Blue) were purchased from Stratagene. Vent Polymerase was purchased from New England Biolabs, Inc.
Subcellular Fractionation-Wild type and recombinant (containing pVV16 empty vector or pVV16 with insert Rv1086 or Rv2361c) M. smegmatis strains were grown to mid-log phase in LB broth (20 g/ml kanamycin for recombinant strains). Cells were harvested by centrifugation, washed with a 0.9% saline solution, and centrifuged again. The resulting pellet was resuspended in buffer containing 50 mM MOPS (pH 7.9), 0.25 M sucrose, 10 mM MgCl 2 , and 5 mM 2-mercaptoethanol. The cells were disrupted by probe sonication on ice with a Sanyo Soniprep 150 (10 cycles of 60 s on and 90 s off). The resulting suspension was then centrifuged at 15,000 ϫ g for 15 min. The pellet was discarded, and the supernatant was centrifuged at 200,000 ϫ g for 1 h in a Beckman 70.1 TI ultracentrifuge rotor. The supernatant (cytosol) was divided into aliquots and frozen at Ϫ70°C until used. The pellet (membranes) was resuspended in homogenization buffer, divided into aliquots, and frozen at Ϫ70°C until used. The protein concentrations of the fractions were estimated using a BCA protein assay kit (Pierce).
In Vitro Prenyltransferase Assays-Isoprenyl diphosphate synthase activity was assayed in mixtures containing 50 mM MOPS (pH 7.9), 10 mM sodium orthovanadate, 5 mM MgCl 2 , 2.5 mM dithiothreitol, 0.3% Triton X-100, 100 M allylic diphosphate, 30 M [ 14 C]IPP, and 75 g of protein in a final volume of 50 l. After incubating at 37°C for 20 min, the reaction was stopped by the addition of 1 ml of water saturated with NaCl. Radiolabeled products were extracted with butanol saturated with water, and an aliquot was taken for liquid scintillation spectrometry. Reactions were assayed under conditions where they were linear for both time and protein concentration. The radiolabeled products were characterized by TLC after treatment with potato acid phosphatase.
Enzymatic Treatment of Reaction Products-Samples to be dephosphorylated by potato acid phosphatase (in order to determine the length of the prenyl chain and the stereochemistry of the isoprene units) were dried under a stream of nitrogen. The samples were dissolved in 5 ml of buffer containing 100 mM sodium acetate (pH 4.8), 0.1% Triton X-100, and 60% methanol. After a brief bath sonication, 20 units of potato acid phosphatase were added, and the mixture was incubated at 25°C overnight. Dephosphorylated products were extracted 3 times with 1 ml of n-hexane. The pooled extracts were washed with 1 ml of water, and the extracts were evaporated under nitrogen. The samples were dissolved in 200 l of chloroform/methanol (2:1, v/v), and an aliquot was taken for liquid scintillation spectrometry. A second aliquot was spotted on TLC plates for analysis.
Analysis of Radiolabeled Product Chain Length-Analysis of the chain length of the dephosphorylated products was accomplished by spotting the radioactive compounds on reverse phase C 18 TLC plates. The plates were then developed in methanol/acetone (8:2, v/v). Radioactive spots were located with a Bioscan System 200 Imaging Scanner (Bioscan Inc.) and by autoradiography. Standard polyprenols were located with an anisaldehyde spray reagent (23).
Analysis of Radiolabeled Product Stereochemistry-The radioactive spots derived from enzymatically labeled, dephosphorylated products that were identified as farnesol were scraped from the TLC plates and extracted from the gel with two 5-ml portions of chloroform/methanol (2:1, v/v). The extracts were pooled and dried under nitrogen. The radiolabeled compounds were dissolved in chloroform/methanol (2:1, v/v) containing authentic, nonradioactive ,E,E-farnesol or mixed stereoisomers (,E,E and ,E,Z) of farnesol; the stereochemistry of the radiolabeled products were determined by spotting on silica gel 60 TLC plates, which were developed with toluene/ethyl acetate (7:3, v/v). Radioactive spots were located with the Bioscan System 200 Imaging Scanner and by autoradiography. Standard polyprenols were located with an anisaldehyde spray reagent (23).
Cloning of Isoprenyl Diphosphate Synthases-Open reading frames Rv1086 and Rv2361c in the M. tuberculosis H37Rv genome (21) were identified as potential isoprenyl diphosphate synthases based on homology to M. luteus undecaprenyl diphosphate synthase (17). Advanced BLAST searches were performed online at the National Center for Biotechnology Information. 3 Alignments were created using Multalin 4 and Genedoc. 5 The following primers were designed to amplify open reading frame Rv1086 from H37Rv genomic DNA: 5Ј-GGTACATATG-GAGATCATCCCGCCG-3Ј and 5Ј-CGTCCTGCGAAGCTTTCACCTGC-CG-3Ј. Primers were also designed to amplify open reading frame Rv2361c from H37Rv genomic DNA: (5Ј-ATATCATATGGCTAGGGAT-GCACGG-3Ј and 5Ј-CCGGTAGGAAGCTTCTAGGCGCTC-3Ј). NdeI and HindIII restriction endonuclease sites (nucleotides underlined in above primer sequences) were engineered into the N-terminal and C-terminal primers, respectively. Polymerase chain reaction was performed on a Perkin-Elmer GeneAmp 2400 PCR System using Vent DNA Polymerase. The polymerase chain reaction products were digested with NdeI and HindIII and ligated into the mycobacterial expression vector pVV16 (a gift from Dr. Varalakshmi Vissa, Colorado State University), which had been previously digested with the same enzymes. The ligation mixture was electroporated into E. coli competent cells (XL-1 Blue). Cells containing plasmid were selected on LB agar containing kanamycin at a concentration of 20 g/ml. Purified plasmids were subjected to restriction and sequence analysis. The resulting constructs, named pVV-Rv1086 and pVV-Rv2361c, were electroporated into M. smegmatis mc 2 155. In addition, pVV16 (without insert) was electroporated into M. smegmatis for use as a control. For electroporation, M. smegmatis cells were grown to late log phase in LB broth; washed seven times with ice-cold, sterile 10% glycerol; and frozen at Ϫ70°C in 50-l aliquots until used. Electroporation was performed at 2.5 V, 800 ohms, 25 microfarads. Cells were recovered in LB broth for 90 min and plated on LB agar with kanamycin (20 g/ml). A single 3 The National Center for Biotechnology Information can be found on the World Wide Web. 4 Multalin alignments can be generated on the World Wide Web. 5 Genedoc is freely distributed on the World Wide Web.
colony was chosen to start liquid cultures in LB broth with kanamycin (20 g/ml).

Recombinants Rv1086 and Rv2361c Increase [ 14 C]IPP Incorporation into Butanol-extractable Material-Rv1086 and
Rv2361c were cloned into pVV16. In this mycobacterial specific expression vector, the cloned genes are constitutively expressed under the control of the heat shock promoter HSP60. Protein (membrane or cytosol) from wild type, empty vector or recombinant strains was assayed for [ 14 C]IPP incorporation into butanol-extractable material in the presence of ,E-GPP or ,E,E-FPP as the reaction primer. The presence of pVV16 (empty vector) in bacteria and the required kanamycin in the growth medium had no effect on the expression of isoprenyl diphosphate synthases (data not shown). Both Rv1086 and Rv2361c cytosolic assays primed with ,E-GPP showed an increase of [ 14 C]IPP incorporation into butanol-extractable material when compared with the wild type cytosolic assays (Table  I). However, only the Rv2361c cytosolic assay primed with ,E,E-FPP showed an increase in [ 14 C]IPP incorporation. Assays of membrane protein revealed similar results (Table I). The Rv2361c recombinant membrane protein was able to utilize both primers more effectively than wild type membrane protein. The Rv1086 recombinant membrane protein was able to utilize more [ 14 C]IPP than wild type membrane protein when primed with ,E-GPP but not when primed with ,E,E-FPP. Protein fractions derived from the strains expressing Rv1086 or Rv2361c had increased isoprenyl diphosphate synthase activity when compared with the corresponding wild type protein fractions.

TLC Analysis Reveals That the Rv1086 and Rv2361c Recombinants Have Increased [ 14 C]IPP Incorporation into Farnesyl
Diphosphate and Decaprenyl Diphosphate, Respectively-The products of the above assays were subjected to enzymatic dephosphorylation for analysis of chain length by TLC. Butanolextracted reaction products were dephosphorylated, and equal amounts of radioactivity were loaded onto reverse-phase TLC plates. Figs. 1 and 2 show the migration of [ 14 C]IPP-labeled products from the cytosol and membrane reactions, respectively. Wild type cytosol primed with ,E-GPP (Fig. 1A) primarily produces geranylgeranyl diphosphate (specific activity ϭ 173 pmol/mg/min). The smaller amounts of FPP (specific activity ϭ 69 pmol/mg/min) and heptaprenyl diphosphate (specific activity ϭ 27 pmol/mg/min) may be due to imperfect fractionation of membrane and cytosolic proteins as these products are seen at approximately 2-3-fold higher concentration in the membrane assays (compare Figs. 1A and 2A). Wild type cytosol when primed with ,E,E-FPP (Fig. 1B), once again primarily produced geranylgeranyl diphosphate (specific activity ϭ 169 pmol/mg/min). Rv2361c cytosolic assays primed with ,E-GPP and ,E,E-FPP (Fig. 2, C and D) showed an increase in the synthesis of decaprenyl diphosphate (specific activity ϭ 141 and 126 pmol/mg/min, respectively), an activity that was almost undetectable in wild type cytosol. Products corresponding to the calculated migrations of octaprenol and nonaprenol are also present. Rv1086 cytosolic assays primed with ,E-GPP (Fig. 1E) showed a 10-fold increase of [ 14 C]IPP incorporation into FPP (specific activity ϭ 714 pmol/mg/min) versus wild type (specific activity ϭ 69 pmol/mg/min), whereas Rv1086 cytosolic assays primed with ,E,E-FPP (Fig. 1F) showed 3-fold decrease in FPP synthesis (but had approximately equal heptaprenyl diphosphate synthesis) when compared with the wild type.  When the products from the wild type membrane assays primed with ,E-GPP ( Fig. 2A) are compared with the recombinant membrane assays primed with ,E-GPP (Fig. 2, C and  E), there is a significant increase in [ 14 C]IPP incorporation into decaprenyl diphosphate caused by the expression of Rv2361c (specific activity ϭ 443 pmol/mg/min) and into FPP caused by the expression of Rv1086 (specific activity ϭ 701 pmol/mg/min). In the membrane assays primed with ,E,E-FPP (Fig. 2, B, D,  and F), there is a 10-fold increase of [ 14 C]IPP incorporation into decaprenyl diphosphate by the Rv2361c recombinant (specific activity ϭ 205 pmol/mg/min) compared with the corresponding wild type assay. As expected from the results in Table I, the Rv1086 membrane assay primed with ,E,E-FPP did not reveal any increased incorporation when compared with the corresponding wild type assay.
Rv1086 Catalyzes Synthesis of ,E,Z-Farnesyl Diphosphate-The farnesyl diphosphate created by adding one molecule of [ 14 C]IPP to ,E-GPP can have two possible stereochemistries, ,E,E-FPP or ,E,Z-FPP. It is possible to separate these stereoisomers by silica gel TLC (Fig. 3, lane 2). The farnesol produced from enzymatically dephosphorylating the products of the Rv1086 membrane assay primed with ,E-GPP (Fig. 2E) was recovered from reverse-phase TLC plate and loaded onto silica gel 60 TLC plates. The recovered farnesol migrated along with cold standard ,E,Z-farnesol, which runs ahead of cold standard ,E,E-farnesol (Fig. 3, lanes 2 and 3). The farnesol produced from enzymatically dephosphorylating the products of the Rv1086 cytosolic assay primed with ,E-GPP was also in the ,E,Z-configuration (data not shown). DISCUSSION The M. tuberculosis open reading frame Rv1086 encodes an ,E,Z-farnesyl diphosphate synthase (Z-FPPS). This is the first report of an amino acid sequence for a short chain Z-isoprenyl diphosphate synthase. Previously, the family of Z-isoprenyl diphosphate synthases contained only undecaprenyl diphosphate synthases, dolichol synthases, and rubber synthase.
Based on the relative lengthy nature of their products (C 55 or greater) and the common Z-stereochemistry catalyzed, members of the Z-isoprenyl diphosphate synthase family became synonymous with "long chain" isoprenyl diphosphate synthases. It is now clear that short chain Z-isoprenyl diphosphate synthases exist. However, with the amount of amino acid sequence information available at this time, short chain Z-isoprenyl diphosphate synthases remain indistinguishable from the long chain Z-isoprenyl diphosphate synthases in genomic data bases. The fact that the cloned Z-FPPS activity was equally distributed between the cytosolic and the membrane fractions confirms our earlier observations that M. tuberculosis cytosolic and membrane fractions contain nearly equal amounts of Z-FPPS activity, whereas the Z-FPPS activity of M. smegmatis was preferentially localized to the membrane fraction. 6 The M. tuberculosis open reading frame Rv2361c encodes a decaprenyl diphosphate (DecaPP) synthase. If the M. tuberculosis DecaPP synthase produces a product stereochemically identical to the DecaPP synthase from M. smegmatis, then its presumed allylic diphosphate substrate in vivo would be ,E,Z-FPP, and each molecule of IPP would be added with Z-stereoconfiguration to yield ,E,polyZ-decaprenyl phosphate (6). Our assays showed that Rv2361c was able to use the allylic primers ,E-GPP and ,E,E-FPP. ,E-GPP may not be used directly for DecaPP synthesis but instead may be used for ,E,Z-FPP synthesis by the background wild type M. smegmatis enzymes. ,E,Z-FPP is then used for De-caPP synthesis. This would explain the decrease in the amount of FPP seen in the Rv2361c membrane assay (Fig. 2C) when compared with the wild type membrane assay ( Fig. 2A). On the other hand, ,E,E-FPP was also a functional substrate for DecaPP synthesis. Precedence for this lack of absolute substrate specificity has been demonstrated in vitro with other isoprenyl diphosphate synthases (24)(25)(26). The stereochemistry of each isoprene addition by DecaPP synthase has not been established. However, based on amino acid sequence homology between Rv2361c and the known Z-isoprenyl diphosphate synthases, it is fair to assume that each isoprene addition would be in the Z-stereoconfiguration.
The first Z-isoprenyl diphosphate synthase to have its amino acid sequence reported, undecaprenyl diphosphate synthase 6 D. Crick, submitted for publication. (17), had no homology to the E-isoprenyl diphosphate synthases (15,16). It did however contain an aspartate rich DD(XX) 2 D motif reminiscent of the E-isoprenyl diphosphate synthases. Shimizu et al. (17) suggested that this motif may represent the diphosphate binding site. The signature motif (Fig. 4, amino acids 168 -172 in MICLU) is not conserved among the M. tuberculosis Z-isoprenyl diphosphate synthases described here or any of the other known Z-isoprenyl diphosphate synthases. Therefore, its occurrence in M. luteus undecaprenyl diphosphate synthase is probably coincidental and not related to enzymatic activity.
We have generated an alignment of M. tuberculosis open reading frames Rv2361c, Rv1086 along with all of the biochemically assayed Z-isoprenyl diphosphate synthases known at this time (undecaprenyl diphosphate synthases from M. luteus, E. coli, H. influenzae, and S. pneumoniae and dolichol synthase from S. cervisiae (Fig. 4)). This alignment demonstrates that there are several regions of amino acid sequence conservation among the Z-isoprenyl diphosphate synthases. The amino acid sequence conservation can be roughly distributed into four regions, which we have designated A-D (to avoid confusion with the five conserved regions (I-V) in E-isoprenyl diphosphate synthases). Region B contains a high degree of residues with charged and uncharged polar side chains, which could hypothetically aid in binding of the diphosphate moiety in the active site. Region B also contains conserved amino acids with large aromatic side chains such as tryptophan and phenylalanine (Fig. 4, amino acid positions 121 and 126). However, the Rv1086 (Z-FPPS) sequence contains leucine at both of these positions (the relevance of this particular observation is unknown).
The amino acid sequences of Rv1086 and Rv2361c were also analyzed for secondary structure using various computer programs (TMpred, 7 TMAP, 8 TopPred2, 9 and SOSUI 10 ). None of the programs reported a membrane-spanning region in Rv1086; however, TMpred and TMAP both predicted a single membrane-spanning region (amino acids 92-120) in Rv2361c. The preferred model from the TMpred report suggested that the amino terminus of the protein exists outside the membrane, amino acids 92-120 span the membrane, and the remaining carboxyl terminus of the protein is on the interior of the membrane. This model is consistent with our observations in that DecaPP synthase activity is found in the membrane 7 Tmpred can be found on the World Wide Web. 8 TMAP can be found on the World Wide Web. 9 TopPred2 can be found on the World Wide Web. 10 SOSUI can be found on the World Wide Web. fraction of M. tuberculosis H37Rv, and the subcellular location of Rv1086 is more ambiguous as activity can be found in both the cytosolic and membrane fractions. 6 Until more information about the active sites and crystal structures of Z-isoprenyl diphosphate synthases is gathered, it will not be possible to predict the chain length of the products of these enzymes based on amino acid sequence alone. Recently, crystallization and preliminary x-ray diffraction studies have been reported for the M. luteus undecaprenyl diphosphate synthase (27). It would be of great interest to determine the quaternary structures of the mycobacterial enzymes, since they could reveal structural requirements for chain length determination. We have initiated purification of these enzymes.