INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

Mycobacterium tuberculosis replicates within the hostile environment of the mammalian macrophage whose microbicidal products are generally insufficient to kill this highly resistant bacillus. This intrinsic resistance to antibacterial substances is due, in large part, to the impermeable nature of the mycobacterial cell wall (1-3). The mycobacterial cell wall consists of three covalently attached polymers: the peptidoglycan, the arabinogalactan, and the mycolic acids. The very low fluidity of the hydrophobic domain of this structure significantly reduces the rate at which hydrophobic substances are taken in and thereby potentiates the toxicity of such substances. This low fluidity is directly attributable to the structure of the mycolic acids, which comprise a large proportion of the cell wall mass (4). Because of the essential nature of this structure to the intracellular life of M. tuberculosis, its biosynthesis and assembly offer critical potential targets for chemotherapeutic intervention.

Mycolic acids are -alkyl--hydroxy fatty acids of exceptional length and complexity, which range up to 80 carbons in total chain length and include various functional groups such as cis or trans cyclopropanes or olefins, -methyl methyl ethers, or -methyl ketones. Because of their structural complexity and issues of overlap with enzymes responsible for synthesizing short-chain fatty acids, in vitro systems for studying their biosynthesis have been difficult to develop. We have identified a family of six homologous enzymes by heterologous expression in the saprophytic mycobacterial species Mycobacterium smegmatis that are involved in generating structural diversity among these molecules by the apparent addition of a methyl group from S-adenosyl-L-methionine (SAM)¹ (5-8). The mechanism of methylation by these enzymes has been proposed to involve the formation of an intermediate carbocation, which can then react to form a variety of different chemical structures depending upon the active site configuration of the particular enzyme resulting in the observed structural diversity of mycolic acids.

Although heterologous expression studies have proven useful for identifying the genes encoding the enzymes involved in mycolic acid modification, they have not resolved critical issues regarding the substrate for methyl(ene) group transfer, including lipid chain length during SAM addition and the head group or acyl carrier moiety during methyl addition. Previous reports of radiomethyl group addition from SAM to very long-chain (~48-56 carbons) mycolate precursors by cell-free supernatants of M. tuberculosis H37Ra suggested that development of a cell-free system based upon combining recombinant enzyme produced in Escherichia coli with soluble acceptor fractions from various mycobacterial species would prove successful (9). In addition, we have recently demonstrated that the long meromycolic acid chain is synthesized on an acyl carrier protein designated AcpM by a novel type II fatty acid synthase system (10). To understand the biosynthetic details of meromycolate modification and the relationship of this system to the AcpM-utilizing type II fatty acid synthase system, we developed cell-free systems for studying the transfer of a methyl group by the meromycolate methyl transferase enzyme family.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Materials and Strains-- M. tuberculosis strain H37Rv (ATCC 27294), M. smegmatis mc²155 (provided by William R. Jacobs, Albert Einstein College of Medicine, New York) and the various recombinants were grown at 37 °C in Middlebrook 7H9 medium with albumin/dextrose/catalase supplement (ADC) containing, where appropriated, kanamycin (25 µg/ml) (Sigma) or hygromycin 50 µg/ml) (Calbiochem). E. coli strain DH5 and JM109 (Life Technologies, Inc.) were used for routine DNA manipulations and grown in LB broth with, when appropriate, kanamycin or hygromycin (200 µg/ml) or ampicillin (Sigma) (50 µg/ml). The electroporation-competent cells of E. coli strain K38 (ATCC 35049) transformed with T7 DNA polymerase-containing plasmid pGP1-2 (11) were prepared by growing at 30 °C in LB with kanamycin to A_600 nm 0.4-0.6. After 1 h in ice, the 500-ml culture was centrifuged and washed twice by prechilled distilled water and once in 10% cold glycerol. Cells were resuspended in 1.5 ml of 10% glycerol, and its aliquots were frozen in dry ice-ethanol and kept at 80 °C.

Vectors and Constructs-- PCR primers mma4.1 (5'-gcgcgccatATGACGAGAATGGCCGAGAAACCG) and mma4.2 (5'-ccgcgggctcttccgcaGGCCGCGGCACCCGGCTTGAGG) were used to amplify the mma4 open reading frame (primer sequence shared by mma4 is shown in uppercase). These primers introduce an NdeI site at the start of the open reading frame and a SapI site after the last codon. The 929-base pair PCR product was cloned in pCR-Blunt (Invitrogen Corp., San Diego, CA). The NdeI-SapI fragment was then removed from this plasmid and inserted into the same sites of pTYB1 (New England Biolabs, Inc., Beverly, MA) to create the expression plasmid pBGS25. The E. coli strain ER2566 transformed with pBGS25 was grown at 37 °C in LB containing 100 µg/ml ampicillin to an OD_650 nm of 1.2, then induced with 0.3 mM isopropyl--D-thiogalactopyranoside for 3 h at 30 °C. The predicted molecular mass of the MMAS-4-intein-chitin binding domain fusion protein produced by pBGS25 is 92 kDa. The open reading frames of genes coding for the methyltransferases CMAS-1, CMAS-2, MMAS-1, MMAS-2, and MMAS-3 were amplified from the corresponding cosmids by PCR with Pwo DNA polymerase (Boehringer Mannheim) using the following pairs of primers: cma1, 5'-GCGGTACCATGCCCGACGAGCTGAAG-3', 5'-GGAAGCTTGGTGTGCATTGGTAGTCA-3'; cma2, 5'-AGAGAATTCATGCCGATGCAACGGTGG-3', 5'-CGCAGGCTTTTATTTGACCAGAGTG-3'; mma1, 5'-AGAGGTACCATGGCCAAGCTGAGACC-3', 5'-CGCGAATTCGAGCTACTTGGTCATGG-3'; mma2, 5'-AGAGGTACCATGCCTCGAGCATGCG-3', 5'-CGCGAATTCCTACTTCGCCAGCGTG-3'; mma3, 5'-AGAGGTACCATGTCTGATAACTCAACG-3', 5'-GCGGAATTCGTCTACTTGGCCAGCGTG-3'. The PCR products were digested with the proper restriction endonucleases (KpnI and HindIII for cma1, EcoRI and HindIII for cma2, KpnI and EcoRI for mmas) and ligated with pRSET-B (Invitrogen, Carlsbad, CA) also cut with those enzymes. After transformation in either DH5 or JM109, plasmids with correct inserts were selected and sequenced. They were then used to transform strain K38(pGP1-2) (11) for expressions as His₆ tag fusion proteins using thermal induction. For expression in M. smegmatis, pMH29-cma1, pMH29-cam2, pMV206-mma1, pMV206-mma2, pMV206-mma3, and pMV206-mma4 were constructed as described previously (6, 7). Restriction endonucleases, DNA-modifying enzymes, and T4 DNA ligase were purchased from New England Biolabs. Plasmids or DNA fragments were purified and isolated with Qiagen plasmid kit, QIAquick nucleotide removal kit, QIAquick PCR purification kit, or gel extraction kit (Qiagen Inc., Chatsworth, CA).

Cell-free Assay for Methyltransferase Activity-- Crude cell lysates were prepared from 1 liter of M. smegmatis cells grown to an OD_650 nm of 0.5-1.0. The cell pellets were washed with 100 ml of cold Buffer 1 (50 mM potassium phosphate, pH 7.0, 1 mM dithiothreitol, 1 mM EDTA) and centrifuged at 12,000 × g at 4 °C for 10 min. The cells were resuspended in 10 ml of Buffer 1 and removed to a bead beater chamber with 10 g of glass beads (0.1 mm in diameter) (Biospec Products Inc., Bartlesville, OK). The cells were lysed by beadbeating twice for 1 min with an intervening 3-min cooling on ice. The cell lysate was centrifuged twice at 12,000 × g at 4 °C for 10 min and then (where indicated) the lysate was heat-inactivated at 90 °C for 10 min. In most assays, the substrate-containing lysate was used on the same day. In some cases frozen heat-treated substrate was used. Activity from frozen substrate was comparable with that of fresh lysates over 1-2 weeks. Overproduction of recombinant methyltransferases in K38(pGP1-2) was induced by growing cells (1:40 dilution from an overnight grown culture) in 500 ml of LB broth at 30 °C to an A₅₉₀ of 0.4. The culture was then incubated with shaking at 42 °C for 30 min followed by 37 °C for 90 min with shaking. E. coli cells were washed and lysed as described for mycobacterial cells. Equal volumes of substrate (mycobacterial supernatant) and enzyme (E. coli supernatant) were mixed in a glass vial and incubated with [³H]SAM (American Radiolabeled Chemicals, Inc., St. Louis, MO) (275 µCi of 72.4 Ci/mmol per 10-ml reaction) at 37 °C for 1 h. The lipids were saponified in 10% tetrabutylammonium hydroxide at 100 °C overnight. Methylation with iodomethane was as described previously (6). The methyl esters were separated by an acetonitrile to dioxane gradient by C18 reverse phase HPLC, and radioactivity was monitored by a -RAM inline scintillation detector (In/Us System Inc., Fairfield, NJ) (12, 13). Columns were calibrated using methyl esters prepared from commercially available saturated fatty acids (Aldrich). For the cell-free assays the following gradient was used: 0-2 min 100% acetonitrile; 2-52 min linear gradient to 100% dioxane, 52-62 min hold at 100% dioxane. The extended gradient shown in Fig. 7 was 0-2 min 75/25 acetonitrile/dioxane, 2-62 min linear gradient to 25/75 acetonitrile/dioxane, 62-65 min to 100% dioxane. In inhibition experiments, the substrates were incubated with 1 mg/ml S-adenosyl-L-homocysteine (Sigma) or sinefungin (from Bill Baker, PathoGenesis Corp., Seattle, WA) (17) or rabbit anti-AcpM polyclonal antisera at 37 °C for 30 min before the addition of methyltransferase and [methyl-³H]SAM.

Partial Purification and Identification of the Substrate Complex for Methyltransferase-- Ammonium sulfate was added to 20% saturation to the mycobacterial lysate, and the precipitates were resuspended and dialyzed against distilled water at 4 °C overnight. The substrate was further purified by Resource Q anion exchange using a 0-1 M NaCl gradient. The sensitivity of the substrates (or products) to proteinase treatment and alkaline conditions was tested by incubating the reaction solution with trypsin at 37 °C followed by the addition of an excess of soybean trypsin inhibitor or by dialysis against 0.01 N NaOH for 2 h, followed by Folch extraction (2:1 chloroform:methanol and aqueous partition) to produce an organic and an aqueous phase. The radioactivity of methylated lipids of each phase was measured by HPLC. A procedure described previously (10) was modified and used to detect whether AcpM is the carrier of the substrate for methyltransferase. The aqueous phase from the cell-free reaction was lyophilized and then resuspended with 10:10:3 (chloroform:methanol:water). The soluble materials were loaded on a silica gel TLC plate that was developed with butanol:water:acetic acid (5:3:2). Pools of materials with different R_F value were scraped from the plate and hydrolyzed with 20% tetrabutylammonium hydroxide.

Other Procedures-- The K_m value of MMAS-2 was measured by the competitive inhibition of the incorporation of [methyl-³H]SAM with increasing concentrations of cold SAM. The corrected rates for the reactions were used to generate a Lineweaver-Burk plot for estimation of the kinetic constants (14). Optimal reaction pH and the affect of pH on the distribution of radioactivity in organic and aqueous phases were determined by making the crude lysate in Buffer 1 with different pH values. For all of these quantitative assays, the peaks corresponding to all long-chain products were integrated and converted to total counts per 1-h reaction. Immunogold staining of ultrathin sections was performed with anti-CMAS1 sera as described previously (15).

Meromycolic Acid Preparation-- M. tuberculosis strain H37Rv was grown at 37 °C to an OD of 1.0 at 650 nm in 7H9 broth with Tween 80 and ADC. The organisms were harvested by centrifugation at 10,000 × g for 15 min. The resulting cell pellet was suspended in an equal volume of 40% tetrabutylammonium hydroxide and incubated at 100 °C overnight. The suspension was then cooled to room temperature, and an equal volume of 5% iodomethane in dichloromethane was added. This suspension was mixed on a rotator for 5 h at room temperature and then allowed to sit until phase separation was complete. The resulting upper phase was discarded. The lower organic phase was washed with an equal volume of water, then 0.1 N hydrochloric acid, then water again and transferred to a 30-ml Corex centrifuge tube and dried under nitrogen at 70 °C. The resulting material was solubilized in 10 ml of toluene and 5 ml of acetonitrile, and the full-length mycolic acids were precipitated by the further addition of 10 ml of acetonitrile (16). The 25-ml toluene/acetonitrile solution was chilled to 20 °C for 2 h and then centrifuged at 12,000 × g for 1 h. The supernatant was discarded, and the toluene/acetonitrile precipitation of the mycolic acids was repeated. The precipitated mycolic acids were then subjected to class separation by preparative silica gel TLC with a mobile phase of hexane:ethyl acetate (95:5). ¹H NMR and mass spectral analysis of the isolated classes of mycolic acids were in accord with that reported previously (7).

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Overproduction of the Six Meromycolyl Methyltransferases in E. coli and M. smegmatis-- We had previously implicated six different enzymes involved in modification reactions of mycolic acids by heterologous expression in M. smegmatis and M. tuberculosis. CMAS-1 introduces the distal cis-cyclopropane (5), CMAS-2 introduces the proximal cis-cyclopropane (6), MMAS-1 converts a cis- to a trans-olefin with the introduction of an allylic methyl branch (8), and MMAS-2, MMAS-3, and MMAS-4 together produce the cis-cyclopropyl methoxymycolic acids (7). To assess the functions of these enzymes in vitro, we cloned each gene into one of two E. coli expression vectors, which included either a hexahistidine fusion (His₆ tag) at the N terminus or a chitin-binding domain and intein-containing cleavage sequence at the C terminus (Fig. 1). Robust expression of each of these genes was evident with the production of a fusion protein of the appropriate molecular weight as shown. The identity of the fusion proteins of CMAS-1, MMAS-1, MMAS-2, MMAS-3, and MMAS-4 was confirmed by Western blotting using polyclonal antisera directed to affinity-purified peptide-elicited antisera to a CMAS-1 sequence common to several members of this family of enzymes (data not shown). Expression of cma1 containing an N-terminal His₆ tag in M. smegmatis and analysis of the mycolic acids produced by this strain revealed the presence of the distally cyclopropanated, proximally -methylated trans-olefinic mycolic acid characterized previously (5), demonstrating that the presence of the fusion peptide did not adversely affect enzyme activity (data not shown).

View larger version (92K): [in this window] [in a new window]   Fig. 1.   Overproduction of mycolate methyltransferases in E. coli. Coomassie Brilliant Blue-stained 12.5% polacrylamide gel electrophoresis of protein lysates prepared from E. coli strain K38(pGP1-2) containing the pRSETB hexahistidine fusion proteins with CMAS-1 (lane 1, uninduced; lane 2, induced), CMAS-2 (lane 3), MMAS-1 (lane 4), MMAS-2 (lane 5), MMAS-3 (lane 6). The latter samples are all shown induced only; uninduced samples all appeared similar to lane 1. Lane 7 contains MMAS-4 fused to the chitin binding domain and intein sequence produced in E. coli strain ER2566. Arrowheads point to proteins of the correct predicted mass CMAS-1 (37.1 kDa), CMAS-2 (41.7 kDa), MMAS-1 (37.6 kDa), MMAS-2 (40.2 kDa), MMAS-3 (37.9 kDa), and MMAS-4 (92 kDa). M, molecular mass standards in kilodaltons.

In Vitro Methyltransferase Activity of the Recombinant Enzymes-- To evaluate various in vitro assay conditions, we first reproduced the transfer of tritium from [methyl-³H]SAM to long-chain fatty acids in lysates of M. tuberculosis, which had been previously reported by Takayama and co-workers (9). Using similar conditions we then labeled clarified wild type M. smegmatis lysates and discovered that, following saponification and methyl ester formation, no long-chain components contained radiolabel. Instead, we observed by reversed phase radio-HPLC an intense peak centering at about 22 min (Fig. 2A). Presumably this product corresponds to tuberculostearic acid, the product of methylation of oleic acid (18). This activity could be completely eliminated by heating the M. smegmatis extract to 90 °C for 10 min (Fig. 2B). Soluble E. coli extracts containing any of the methyltransferase enzymes showed no specific labeling of lipid-containing materials (Fig. 2C), but when combined with heat-treated (Fig. 2D) or non-heat-treated (Fig. 2E) lysates from M. smegmatis the recombinant proteins catalyzed the incorporation of a tritium label from SAM into very long-chain lipid components with HPLC retention times between 35 and 45 min. The temperature and duration of the heat treatment were critical to maintaining substrate activity; treatment of the M. smegmatis lysate at 100 °C for 20 min prior to addition of the recombinant methyltransferase completed eliminated activity (data not shown).

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Definition of the cell-free methyltransferase assay. All traces represent reversed phase radio-HPLC analysis of methyl esters prepared from 1-h incubations of the indicated lysate with [methyl-3H]SAM followed by saponification, extraction of lipids, and methyl ester formation. A, M. smegmatis strain mc2155 crude cell lysate (clarified at 10,000 × g). B, M. smegmatis crude cell lysate heated at 90 °C for 10 min. C, E. coli (pRSETB_mma2) crude cell lysate also clarified at 10,000 × g. D, lysates from B and C mixed together. E, lysates from A and C mixed together. F, lysates from B and C preincubated for 30 min with 1 mg/ml S-adenosyl-L-homocysteine before adding [methyl-3H]SAM. In this figure and in Fig. 3, the gradient used shows the following retention times for saturated fatty acid methyl esters: C20:0 (14.8 min), C24:0 (22.2 min), C28:0 (28.8 min), C30:0 (31.6 min).

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Cell-free activity from various methyl transfer enzymes. All traces represent reversed phase radio-HPLC analysis of saponified methyl esters of fatty acids as in Fig. 2. In each case a lysate from the indicated recombinant E. coli strain(s) from Fig. 1 was mixed with a heat-inactivated M. smegmatis lysate prepared as in Fig. 2B. A, control E. coli lysate; B, CMAS-1; C, CMAS-2; D, MMAS-2; E, MMAS-4; F, MMAS-4 + MMAS-3.

Inhibition and Characterization of the in Vitro Methyl Transfer Reaction-- In vitro methyl(ene) transfer could be entirely inhibited by low micromolar concentrations of S-adenosyl-L-homocysteine, the demethylated analog of SAM (Fig. 2F). This level of inhibition was similar to that seen with other methyltransferases (19) and was also observed with another SAM analog, sinefungin (20). Treatment of whole cells of either M. tuberculosis or various recombinant M. smegmatis strains with up to millimolar concentrations of these inhibitors did not affect the mycolic acids produced nor did they inhibit growth.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   pH optimization and stability and product distribution of the cell-free methyltransferase assay. A, pH versus total product formed by MMAS-2 containing lysates from integration of radio-HPLC tracings expressed as counts/min. Each pH point contained an identical amount of substrate containing material and an identical amount of enzyme. B, aqueous associated counts when the product of the reaction run at the indicated pH was partitioned between chloroform:methanol (2:1) and water. C, organic phase-associated counts from the same extraction, note the lower scale.

Cellular Localization of the Mycolyl Methyltransferase Activity-- The enzyme produced in M. smegmatis appeared to be entirely soluble when analyzed by Western blotting of soluble lysates and solubilized membrane fractions using antipeptide antisera that reacted with either CMAS-1 or CMAS-2 (data not shown). The substrate for methyl transfer also appeared to be soluble, although some activity could be observed with resuspended membrane fractions. Immunogold localization of the methyltransferases with peptide-based antisera to CMAS-1, which cross-reacts with several of the protein family members, revealed a pronounced bias for membrane association when the proteins were present at natural levels in MTB (Table I). This bias disappeared when the proteins were overproduced in M. smegmatis, suggesting that, like cyclopropane fatty acid synthase from E. coli, these enzymes are predominantly cytoplasmic with a transient, peripheral association with membrane systems (21).

Identification of the Lipid Chain Length during Methyl Transfer-- The cell-free methyl(en)ation reaction allowed us to address the question of the length of the lipid during this enzymatic transformation. Simple calibration of the HPLC column revealed a linear relationship between chain length and retention time as has been described by previous authors (12, 13). Interpolating from the longest commercially available saturated fatty acid (30 carbons in length), it was possible to estimate that the lipid labeled by MMAS-2 in the cell-free system with a retention time of 45 min possessed a minimum of 40-42 carbons. The introduction of a single unsaturation into the lipid would have the effect of reducing the retention time (and thus the apparent chain length) by reversed phase HPLC depending upon the position of the unsaturated group (9). Full-length meromycolate from M. tuberculosis would be expected to have the equivalent of two such desaturations as well as possible methyl ethers or ketones making the precise retention times difficult to predict. Full-length mycolic acid, with an extra secondary hydroxy group and an -alkyl branch, elutes from this column at about 50 min (Fig. 5A).

View larger version (10K): [in this window] [in a new window]   Fig. 5.   Synthesis and HPLC analysis of the meromycolic acid from the methoxymycolate of M. tuberculosis H37Rv. A, HPLC analysis with evaporative light-scattering detection of purified methoxymycolate. The same gradient as used for the cell-free methyltransferase assay was used for A and B. B, HPLC analysis of the meromycolic acid derived from the methoxymycolate series shown in A. C, synthetic scheme for the pyrolytic cleavage of the methoxymycolate and oxidation to produce the meromycolic acid shown in B.

View larger version (14K): [in this window] [in a new window]   Fig. 6.   Analysis of the fine structure of the radioactive products formed by a proximally cyclopropanating and a distally cyclopropanating enzyme. A, radio-HPLC trace of an expanded gradient showing the fine structure of products produced from the reaction of CMAS-1 operating on the crude heat-inactivated substrate material from M. smegmatis. B, the same gradient showing the more limited distribution and longer length of the products formed by the action of MMAS-2 on the same pool of substrate. The gradient used shows the following retention times for saturated fatty acid methyl esters: C20:0 (8.0 min), C24:0 (13.8 min), C28:0 (22.8 min), C30:0 (28.1 min).

AcpM Is the Lipid Carrier during Methyl Transfer-- Free fatty acids of even moderate length are relatively insoluble in aqueous systems and fatty acyl-CoA esters of greater than about 20 carbons are also highly insoluble. Despite this, the substrates for these methyl transfer reactions appear to be well behaved soluble molecules. To understand the nature of the carrier molecule that allows these lipids to remain soluble, we conducted a series of preliminary trials to ascertain the identity of this substance. As noted above the carrier was sensitive to heat at 100 °C for 20 min but was resistant to 90 °C for 10 min. We also found that preincubation of the substrate with trypsin followed by the addition of excess trypsin inhibitor (but not mock-incubations) destroyed the ability to incorporate [³H]SAM into such acceptors. In addition, labeling with [³H]SAM followed by treatment with trypsin followed by chloroform:methanol (2:1) to water partition resulted in all of the radiolabel partitioning into the organic phase, whereas control samples without added trypsin remained in the aqueous phase. The substrate could be precipitated by 20% ammonium sulfate treatment and retained activity following resuspension. The substrate could also be partially purified by ResourceQ anion exchange fast protein liquid chromatography. The extreme base sensitivity noted above was consistent with a thioester linkage, and the above results strongly suggested that the carrier molecule was proteinaceous in nature.

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Inhibition of the cell-free methyltransferase activity by anti-AcpM antisera. Soluble clarified lysates prepared from M. tuberculosis H37Rv producing only the normal amounts of endogenous methyltransferases when incubated with [methyl-3H]SAM in the absence (A) or presence (B) of affinity-purified antisera directed to the C terminus of the AcpM polypeptide. Incubations with irrelevant antisera (anti-KatG, anti-Acr) did not show any inhibition of the reaction (not shown).

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

Our previous identification of a family of methyltransferases whose heterologous expression resulted in the modification of full-length mycolic acids allowed the implication of the function of these enzymes but left open many important questions regarding the details of the biosynthetic pathway to these important cell wall constituents (3). Elucidating these details is essential if these reactions are to be exploited as potential chemotherapeutic targets. In addition, overproduction of active enzymes was essential for structural studies of these enzymes to evaluate the hypothesis that a limited number of active site residues determine product outcome and produce chemical diversity among mycolic acids. As a first step all six of these enzymes have been overproduced in E. coli as fusion proteins, and an in vitro assay has been developed allowing us to demonstrate that these enzymes retain catalytic activity as fusion proteins.

In vitro methyl transfer from radiomethyl-SAM to an acceptor molecule isolated from the mycobacterial cytoplasm has allowed us to identify both the lipid length of the meromycolate chain during modification as well as the carrier for mero chain synthesis. Fig. 8 shows a working hypothesis which integrates this information with the information from an analysis of the mechanism of action of isoniazid (10). In this model primers for the type II fatty acid synthase system, which produces mycolic acids, are produced by the type I fatty acid synthase. These primers are predominantly palmitate (C_16:0) and hexacosanoate (C_26:0). These are then transferred, presumably via coenzyme A esters, to the acyl carrier protein AcpM. The various acyl-AcpM molecules are elongated using malonyl-CoA by the type II system components, which include KasA, KasB, InhA, and MabA (23). The full 56 carbons of the mero acid do not appear to be completely synthesized before they are modified. Instead, modification parallels synthesis and occurs while the acids are attached to the ACP. These studies do not distinguish the point at which desaturation occurs to the growing chain and whether desaturation occurs in a parallel fashion or to a defined substrate that is then elongated by the type II system. In these heat-inactivated extracts the components of the type II system are not capable of elongation, therefore the length of the product formed reflects the length of the substrate used. This has the interesting implication that the various proximally specific and distally specific enzymes are measuring the mero chain from the -end or, alternatively, that they measure from the thioester end but have a very wide substrate specificity.

View larger version (14K): [in this window] [in a new window]   Fig. 8.   Parallel synthesis and modification of the -meromycolic acid. This scheme shows the AcpM-bound growing meromycolic acid chain that is being elongated by the KasA/B-containing type II fatty acid synthase system and illustrates the wider range of chain lengths available to CMAS-1 than to CMAS-2 for use in the cyclopropanation reaction.

The fact that the in vitro reaction can be inhibited strongly by antibodies specific for AcpM and that the acceptor has the expected physical properties of an acyl-AcpM strongly suggests that the growing mero acid remains attached to this ACP during modification and synthesis. Further evidence for this fact comes from an inspection of potential desaturases in the mycobacterial genome (24) in which there appears no clear homologs of a FabA-like dehydrase that produces cis-olefinic fatty acids. Instead the genome reveals several potential aerobic mixed function desaturases homologous to plant enzymes (which act on ACP-bound lipids) but not homologous to vertebrate enzymes (which act on CoA esters).

These results also confirm the work of Takayama and co-workers who identified families of monounsaturated, diunsaturated, and cyclopropanated potential meromycolate intermediates ranging from 26-56 carbons in length (9, 12, 13, 25, 26). The nature of the carrier molecule during the condensing reaction has not been identified (27), although we have argued that this is likely to be a simple CoA ester (3). The ability to specifically label the mero acid product of these methyltransferases will also allow the details of the condensation step to be explored.

Several reports have appeared of a particulate system capable of synthesizing intact mycolates from radio-acetate (although importantly not from malonate) (28-31). These reports have suggested that all of the enzymes necessary for mycolic acid production are associated with a macroparticulate system following partial purification through a gradient of Percoll. It is difficult to assess the relevance of this system precisely because: 1) incorporation of acetate but not malonate or malonyl-CoA would be unusual for an fatty acid synthase system, 2) in our preparations the meromycolic acid modification enzymes do not fractionate into this particulate fraction,² 3) in our hands such fractions contain small numbers of intact or nearly intact organisms that incorporate acetate very efficiently, and (4) it has not been possible to inhibit such acetate-dependent activity by cell-impermeable inhibitors such as sinefungin or S-adenosyl-L-homocysteine. The results presented here support the proposal that mero acid synthesis occurs largely in the cell cytoplasm, whereas the growing fatty acyl chain is covalently attached to AcpM via a thioester linkage.

This assay has also allowed us to demonstrate, in principle, that potential inhibitors can be assessed directly against the appropriate reaction components, because inhibition could be observed by known inhibitors of similar enzymes such as S-adenosyl-L-homocysteine and sinefungin. Reformatting this assay for high-throughput screening of potential inhibitors will greatly facilitate the identification of novel chemotherapeutics for use in the treatment of tuberculosis and related mycobacterial diseases. In addition, rational drug design using structural information will be possible following structure elucidation using these recombinant enzyme overproduction systems.