The biosynthesis of cyclopropanated mycolic acids in Mycobacterium tuberculosis. Identification and functional analysis of CMAS-2.

The major mycolic acid produced by Mycobacterium tuberculosis contains two cis-cyclopropanes in the meromycolate chain. The gene whose product cyclopropanates the proximal double bond was cloned by homology to a putative cyclopropane synthase identified from the Mycobacterium leprae genome sequencing project. This gene, named cma2, was sequenced and found to be 52% identical to cma1 (which cyclopropanates the distal double bond) and 73% identical to the gene from M. leprae. Both cma genes were found to be restricted in distribution to pathogenic species of mycobacteria. Expression of cma2 in Mycobacterium smegmatis resulted in the cyclopropanation of the proximal double bond in the α series of mycolic acids. Coexpression of both cyclopropane synthases resulted in cyclopropanation of both centers, producing a molecule structurally similar to the M. tuberculosis α-dicyclopropyl mycolates. Differential scanning calorimetry of purified cell walls and mycolic acids demonstrated that cyclopropanation of the proximal position raised the observed transition temperature by 3°C. These results suggest that cyclopropanation contributes to the structural integrity of the cell wall complex.

An estimated 8 million persons develop tuberculosis each year, and over 30 million people are expected to die from the disease in this decade (1). Mycobacterium tuberculosis, the causative agent of tuberculosis, is an intracellular pathogen that establishes an infection in oxygen-rich alveolar macrophages of the lung (2). Mycolic acids are long chain ␣-alkyl-␤hydroxy fatty acids unique to mycobacteria and related taxa and represent major components of the cell wall (3). Mycolic acids are thought to contribute to both drug resistance and survival in the hostile intracellular environment of the macrophage by the formation of an impermeable asymmetric lipid bilayer (4). The biosynthetic pathway for these complex lipids is also thought to be the target for several clinically useful chemotherapeutics, including isoniazid (5). With the increasing incidence of multidrug-resistant bacilli, alternate chemotherapeutic targets are urgently needed.
Mycolic acids have been proposed to be biosynthesized via a diversion of normal fatty acid metabolism in which short chain fatty acids are extended and modified to form lipids of exceptional length (6). Mycobacterium smegmatis synthesizes three different series of ␣-mycolates (which lack oxygen functionalities in the meromycolate chain outside the ␤-hydroxy acid) shown in Fig. 1 (7). The ␣ 1 and ␣ 2 series are full-length mycolic acids extending to an average of 78 and 79 carbons, respectively (8). ␣ 1 contains two cis-olefins in the meromycolate chain, while ␣ 2 contains only a single cis-olefin and a transolefin with an adjacent methyl group. In addition to these three mycolates, M. smegmatis also produces a shorter ␣Ј mycolic acid, which is 64 carbons in length as well as a full-length epoxy mycolate (9). M. tuberculosis contains only one series of ␣-mycolic acids that averages 78 -80 carbons in length (4) (10,11). The tubercle bacilli also produces two oxygenated mycolic acid series, ketomycolates and methoxymycolates (not shown), which are generally of lower abundance than the ␣ series (12). Pathogenic mycobacteria cyclopropanate a majority of their mycolic acids, whereas in saprophytic organisms, this modification is unusual (3). The functions of the various classes of mycolic acids in each of these organisms is unknown. However, we have recently shown that cyclopropanation at the distal position confers increased resistance to in vitro killing by hydrogen peroxide (13). The present studies were initiated to expand our understanding of the relationship between mycolic acid structure and function. We have previously reported the identification of cma1, a gene whose protein product (cyclopropane mycolic acid synthase-1, CMAS-1) 1 catalyzed the introduction of a cyclopropane at the distal position in the meromycolate chain (13). In the course of these studies, we discovered that an unannotated related sequence had been deposited in Genbank as part of the Mycobacterium leprae genome sequencing project (accession number U00018) (14). In this paper we report that this M. leprae sequence represents a second cyclopropane synthase with a homolog in M. tuberculosis whose protein product functions distinctly from CMAS-1 to cyclopropanate the proximal cis-olefin in mycolic acid biosynthesis. (50 g/ml) (Calbiochem). Peroxide susceptibility measurements were conducted as described previously (13).
Cloning and Sequencing cma2-PCR products from M. leprae genomic DNA were used to screen an M. tuberculosis H37Ra cosmid library. PCR product 1 displayed high (57%) homology with the M. tuberculosis cma1 gene (GenBank accession number U27357), and the M. leprae cma2 gene. This product was 459 nucleotides in length and corresponded to nucleotides 22918 -23377 from the M. leprae cosmid U00018. The primers used to generate this probe were 5Ј-CACTAT-GCTGGGCGAATT-3Ј and 5Ј-GTTCGGGTGTGGTCTATTT-3Ј. PCR product 2 (768 nucleotides) was less homologous (46%), corresponded to nucleotides 23482-24250 of the same cosmid, and was generated with the following two primers: 5Ј-GCGACGCCGGATTC-3Ј and 5Ј-CG-GCTCGGAAGAGATTT-3Ј. Colony lifts were performed of Escherichia coli containing the cosmid DNA library from H37Ra in pYUB18 with GeneScreen Plus hybridization membranes (DuPont NEN), which were then hybridized to each PCR-generated probe separately according to the manufacturer's protocol. DNA sequencing was performed using the Sequenase 7-deaza-dGTP DNA sequencing kit (U. S. Biochemical Corp.) with synthetic universal and custom primers. Codon preferences were determined by reference to the published sequence of mycocerosic acid synthase (15) (GenBank accession number M95808).
Mycolic Acid Methyl Ester (MAME) Isolation and Purification-MAMEs were isolated and purified following basic methanolic hydrolysis as described previously (13). For NMR analysis, acetonitrile/toluene precipitation was followed by a 10-cm silica gel column in 9.5:0.5 hexanes/ethyl acetate. For radiolabeling, 50 Ci of sodium [1-14 C]acetate was added for several hours to growing cultures of M. smegmatis containing the appropriate construct and antibiotics before purifying MAMEs as above.
Two-dimensional TLC Procedure-Two-dimensional TLC analyses were performed by immersing 90% of a square silica gel 60 TLC plate (0.2 mm thickness, EM Separations, Gibbstown, NJ) into 5% aqueous silver nitrate (w/v). Following air-drying, these plates were activated as described by Kennerly (16). 14 C-Labeled samples were run in the first dimension along the narrow strip without silver impregnation by developing twice with 9.5:0.5 hexanes/ethyl acetate. The plates were then dried, turned 90°and run into the silver layer by developing 3 times with 85:15 petroleum ether/diethyl ether. Plates were then visualized, and the individual mycolate species were quantitated using a Phosphor-Imager (Molecular Dynamics).
Vectors and Constructs-pYUB18 was a gift of W. R. Jacobs, Albert Einstein College of Medicine, NY (17). pMV206 and pMV261 were provided by MedImmune, Inc., Gaithersburg, MD. pMH29 was derived from pMV206 by the addition of a synthetic regulatory region (termed mycobacterial optimal promoter, or MOP) consisting of promoter sequences from the BCGhsp70 gene, and an E. coli tac promoter with the E. coli -independent rrnAB T1T2 terminator (18) upstream of MOP. pMH29 was provided by C. K. Stover and M. J. Hickey (PathoGenesis Corp., Seattle, WA). pMV206_Hyg and pMH29_Hyg were constructed by removing the SpeI to NheI kanamycin resistance cassette and replacing it with a hygromycin resistance cassette consisting of a 1.3-kb BspHI to SmaI fragment from p16R1 (provided by Douglas Young, Wright Flemming Institute, London) (19). pYUB-cma1 contains the BamHI to PstI fragment with the cma1 open reading frame and upstream region as described previously (13) cloned into the BamHI site of pYUB18. pYUB-cma2 contains a 35-kb chromosomal fragment from M. tuberculosis H37Ra cloned at the BamHI site of pYUB18. pMH29_Hyg-cma1 contains the 1.5-kb BamHI to PstI fragment cloned into the same sites in pMH29_Hyg, resulting in expression from MOP. pMH29_Hyg-cma2 contains a 1.2-kb fragment constructed by digesting the 3.9-kb cma2-containing insert with NruI followed by ligation of XbaI linkers and digestion with both BamHI and XbaI. The gel-purified insert was ligated to BamHI, XbaI-digested pMH29_Hyg. pMV206_Hyg-  (13) and the amino acid sequence of cyclopropane fatty acid synthase from E. coli (24). The putative S-adenosyl-L-methionine binding domain (25) spans amino acids 171-179 (using cyclopropane fatty acid synthase numbering), and cysteine 354 has been proposed to have catalytic function (24). cma1ϩcma2 was constructed by cloning the 1.5-kb BamHI to PstI fragment containing cma1 into pMV206_Hyg restricted at the same sites. The cma1-containing vector was cut with BamHI and XbaI and a 1.7-kb BamHI to XbaI fragment containing cma2 was inserted.
Differential Scanning Calorimetry-Purified methyl mycolates (10.0 mg) were added to a 1-ml crucible along with 0.2 ml of phosphatebuffered saline (50 mM sodium phosphate, pH 7.5, 75 mM NaCl) and placed in a model 4110 differential scanning calorimeter (Calorimetry Sciences Corp., Provo, Utah). Initially, three crucibles were heated at 30°C/h from 10 to 60°C and cooled at the same rate to 10°C. This rapid cycle was followed by a slower heating and cooling cycle through the same temperature range at 10°C/h, with data collection at 10-s intervals. Cell wall material was not cycled through an initial melt, and data were collected at 10°C/h from 10 to 65°C on the first melt/downscan cycle.
M. smegmatis cell walls were prepared as follows: M. smegmatis (250 ml A 650 nm ϭ 1.0) culture was harvested by centrifugation and resuspended in 4 ml of 1 mM phenylmethylsulfonyl fluoride. This suspension was divided equally among eight 1.5-ml tubes containing 0.5 g of 0.1-mm glass beads. These were then placed in a Mini Beadbeater-8 (BioSpec Products, Bartlesville, OK) and lysed at maximum speed for 3 min. After briefly spinning at 5,000 ϫ g, the supernatants were removed, and fresh phenylmethylsulfonyl fluoride was added (0.4 ml/ tube). The pellets were again lysed for 3 min, and the beads were allowed to settle, after which the supernatants were combined (ϳ5 ml), cooled on ice, and 2.25 ml of 10% Triton X-114 (CalBiochem) was added. After mixing on ice for several minutes, cell walls were removed by centrifugation at 100,000 ϫ g for 1 h. These were washed again in fresh 2% Triton X-114 followed by an additional wash in phosphate-buffered saline to remove excess detergent. Monoclonal antibodies used to establish purity of these cell wall preparations in Western blots (Hsp60 (IT-13), the 19-kDa lipoprotein (IT-12), and the 32-kDa ␣-antigen (IT-49)) were provided by the UNDP/World Bank/World Health Organization Special Program for Research and Training in Tropical Diseases.

RESULTS
Cloning and Characterization of the cma2 Gene from M. tuberculosis-Two factors led us to conclude that cma1 and the unannotated cosmid sequence from M. leprae, although both clearly cyclopropane synthases, were not coding for homologous proteins with identical function. First, sequences surrounding the cma1-coding region were nonhomologous to the sequences surrounding the coding region of the M. leprae open reading frame. Second, the relatively low similarity of the two protein sequences (53% identity) seemed anomalous compared with the relatedness of other proteins between the two pathogens (for example, the RecA proteins share 92% identity (20) and the ahpC (alkylhydroperoxidase) genes share 85% identity between these two organisms (21)). Since both organisms have two centers that ultimately become cyclopropanated in their ␣-mycolate series, we postulated that the M. leprae homolog may represent a second enzyme that introduces the proximal cyclopropane.
To test this hypothesis, PCR primers were designed to amplify one region of high homology and one region of low homology between the two sequences. These two probes were used sequentially to screen colony lifts of E. coli tranformed with an M. tuberculosis cosmid library (13). Out of approximately 300 colonies, four clones were positive with the more homologous probe while only one of these was also positive with the less homologous probe. Cosmid DNA isolated from the positive clone was digested with BamHI and probed in a Southern blot with the same probes. The clone that reacted with both probes showed a strong band at 4.2 kb as well as weaker hybridization to the 7.2-kb fragment known to contain cma1 (13). Sequences homologous to cma2 were found by Southern blot analysis to be absent from the saprophytic M. smegmatis and present in pathogenic strains of Mycobacterium avium, Mycobacterium bovis BCG, and Mycobacterium marinum (data not shown).
Sequence Analysis of cma2-The 4.2-kb BamHI fragment containing the putative cma2 sequence was subcloned into pBluescript II (KSϩ). Restriction mapping and Southern anal-ysis yielded a 1.7-kb BamHI to XbaI fragment, which then was used to determine the nucleotide sequence of cma2. This sequence has been deposited in GenBank (accession number U34637). The GC content of the sequenced region was 61%, typical of mycobacterial DNA (22). In addition, the presence of the cma2 open reading frame was supported by (i) the presence of appropriate translational start and stop signals, (ii) the bias of GC distribution in the third codon position (23), and (iii) appropriate codon usage compared with other known mycobacterial genes (17).
A comparison of the deduced amino acid sequence of CMAS-2 (the cma2 gene product) with the other known cyclopropane synthases is shown in Fig. 2. As expected, the M. tuberculosis CMAS-2 sequence is more closely related to the M. leprae CMAS-2 sequence (73% identity) than either of the CMAS-2 are to CMAS-1 from M. tuberculosis (52% identity for each). The four proteins share the highest homology in the region corresponding to the N terminus of CMAS-1 (amino acids 14 -96 of CMAS-1). Interestingly, the three cyclopropane synthases (including cyclopropane fatty acid synthase from E. coli (24)) have variable N-terminal extensions, the significance of which is unknown. Functionally, one of the most important areas of homology shared by these four proteins spans amino acids 171-179 (using the cyclopropane fatty acid synthase numbering) and corresponds to a consensus motif of (V/L)L(E/ D)XGXGXG, which has been proposed to play a role in binding of the enzyme cofactor S-adenosyl-L-methionine (25). Cysteine 354 is also absolutely conserved and has been proposed to have catalytic function (24).
CMAS-2 Expression in M. smegmatis-In order to study the function of the cma2 gene product CMAS-2, the pYUB18 cosmid containing this gene was introduced into M. smegmatis, and MAMEs were purified as described under "Experimental Procedures." Analysis of the total MAMEs by 500 MHz 1 H NMR revealed the presence of resonances characteristic of cyclopropane ring hydrogens (␦ Ϫ0.33 ppm, multiplet; ␦ 0.56 ppm, multiplet; ␦ 0.64 ppm, broad multiplet). By integration of these resonances and comparison with the corresponding integration of the signal for the terminal methyl groups (␦ 0.88ppm), it was estimated that expression of cma2 in this system resulted in monocyclopropanation of 10% of the total mycolic acids. In this same analysis, control samples of wildtype M. smegmatis showed Ͻ2% cyclopropanation.
To confirm that this transformation was due to the putative cma2 sequence as well as to improve the extent of conversion, a 1.2-kb NruI to BamHI fragment containing the cma2 reading frame was subcloned into pMH29_Hyg. pMH29_Hyg is a derivative of pMV261 that contains a hygromycin resistance marker in place of kanamycin. Hygromycin selection has much lower background and allows much faster recovery times than kanamycin for transformed mycobacteria. 2 This vector also contains a synthetic MOP in place of the Hsp60 promoter region. CMAS-2 produced from this construct was capable of converting 25% of the total mycolates to the cyclopropanated type as determined by NMR.
Identification of the CMAS-2 Product and Co-expression of CMAS-1 and 2-For comparison purposes, cma1 was placed in pMH29_Hyg to give a CMAS-1-overproducing system. In addition, cma1 and cma2 were both cloned with several hundred nucleotides of upstream sequence into pMV206_Hyg, a derivative of pMV206 that has a hygromycin cassette replacing the kanamycin resistance gene. pMV206_Hyg is promoterless, allowing expression of both genes from their own promoter regions. MAMEs from M. smegmatis transformed with each of these constructs were prepared following labeling with [1-14 C]acetate and were analyzed by two-dimensional TLC. In this experiment, the first dimension was normal silica gel, and the second dimension was silica gel impregnated with silver ions. Such argentation TLC allows the selective retardation of components containing cis-olefins, while components with either trans double bonds or cyclopropanes are less affected or unaffected in their mobility (18). Analysis of wild-type M. smegmatis by this technique allowed the identification of all the major mycolates (Fig. 3A) whose structures are shown in Fig. 1. Introduction of cma1 in this system (Fig. 3B) results in production of a single spot of high mobility in the argentation dimension, which we have previously identified as a hybrid mycolate containing a distal cyclopropane and a proximal trans double bond with an ␣-methyl branch (13). pMH29_Hyg-cma2 also results in production of a new mycolate that is retarded more strongly than the cma1 product by silver ions (compare Fig. 3B, spot 1, with Fig. 3C, spot 2). This product (2) appears to result from conversion of the ␣ 1 -mycolate. In addition, the epoxy mycolate series (spot e in Fig. 3C) also appears to change retention time on argentation chromatography in a manner consistent with cyclopropanation at the proximal position to produce spot 3.
Introduction of both cma genes into M. smegmatis resulted in a pronounced change in the radio-TLC profile, which contained both the CMAS individual products as well as a unique MAME, which was unaffected by silver ion impregnation (Fig. 3D). This MAME exactly co-elutes with the major ␣-mycolic acid from M. tuberculosis (data not shown). To confirm these structural predictions, MAMEs were purified from 1 liter of M. smegmatis containing pMV206_Hyg-cma1ϩcma2 and separated by preparative argentation TLC. 500 MHz of 1 H NMR analysis of spot 4 (Fig. 4A) showed that this spot had no olefinic resonances but had cyclopropane resonances (␦ Ϫ0.33, 0.56, and 0.65 ppm). These were present in a 4:6 ratio with terminal methyl groups, indicating that this M. smegmatis mycolate corresponded to structure 4, which is the major mycolate from M. tuberculosis (Fig. 1). The mycolate corresponding to spot 1 in Fig. 3D showed 1 H NMR resonances (Fig. 4B) corresponding to a trans-olefin at ␦ 5.3 ppm (J ϭ 15 Hz) as well as a doublet corresponding to an ␣-methyl group at ␦ 0.93 ppm and cyclopropane resonances as Spectrum A shows the dicyclopropyl mycolate corresponding to structure 4 (Fig. 1), which represents the major ␣ mycolate from M. tuberculosis. By integration, the cyclopropane resonances (␦ Ϫ0.33, 0.56, 0.65 ppm) are present in a ratio of 4:6 with terminal methyl groups. Spectrum B shows the mycolic acid corresponding to spot 1 in Fig. 3D, which contains a trans double bond (␦ 5.36 ppm, J ϭ 15 Hz) and a single cyclopropane. By integration the olefinic and cyclopropyl protons represent two and four protons with respect to terminal methyl groups. Spectrum C shows the mycolic acid corresponding to spot 2 in Fig. 3D, which contains a single cis-olefin as well as a single cis-cyclopropane. This mycolate also lacks the ␣-methyl branch since there is no doublet at ␦ 0.95ppm. Spectra were recorded in deuterochloroform and are referenced to internal tetramethylsilane.
above. The olefinic and cyclopropane resonances integrated for two and four protons, respectively; thus this mycolate corresponds to the previously described structure containing a distal cyclopropane and a methyl-branched trans-olefin. Spot 2 in Fig.  3D showed 1 H NMR resonances (Fig. 4C) consistent with a cis-olefin at ␦ 5.36 ppm (J ϭ 10 Hz) as well as the cis-cyclopropane resonances. This mycolate also displayed no ␣-methyl branch and is produced by CMAS-2 alone, consistent with a proximally-cyclopropanated ␣ 1 -mycolic acid.
Radio-TLC data for M. smegmatis transformants containing cma genes were quantitated by PhosphorImaging analysis ( Table I). Introduction of CMAS-1 decreases the ␣ 2 series from which it is derived, while the ␣ 1 series decreases when CMAS-2 is present. Interestingly, the total amount of mycolate cyclopropanated at the distal position in the coexpressing construct (where cma1 is expressed from its own promoter) is twice that produced when the gene is expressed from the MOP promoter. The total amount of mycolic acids cyclopropanated at the proximal position is the same between the coexpressor and the overexpressor, suggesting that CMAS-1 activity is affected by the presence of CMAS-2 but not vice versa.
Effect of Cyclopropanation on Cell Wall Fluidity-To assess the relative contributions of these modifications on cell wall fluidity, we examined both purified methyl mycolic acids and purified M. smegmatis cell walls from the recombinant organisms by differential scanning calorimetry (DSC). Cell walls were prepared by a simple lysis and Triton TX-114 extraction procedure as described under "Experimental Procedures." When analyzed by SDS-polyacrylamide gel electrophoresis, such cell wall preparations showed simplified protein profiles similar to the reported profiles of Mycobacterium chelonei cell wall preparations (data not shown) (26). In addition, cell wall material prepared in this fashion was shown to be depleted of the cytoplasmic Hsp60 and plasma membrane-associated 19-kDa lipoprotein when analyzed by Western blotting using monoclonal antibodies. This material was also shown to be enriched for the 32-kDa mycobacterial ␣-antigen, which has been shown to be associated with the cell wall. When analyzed by DSC, cell wall preparations showed distinct thermal transitions or melting temperatures at 45-55°C (Fig. 5B). These transitions were fully reversible and could be observed through multiple cycles of up-and downscans. In addition, the same thermal transitions can be observed using whole organisms, although these were more difficult to interpret (data not shown). Notably, cell walls isolated from M. smegmatis ex-pressing CMAS-1 were indistinguishable from control isolates, but when CMAS-2 was expressed, either alone or in combination with CMAS-1, the transition temperature increased by 3°C. This suggests that a proximal cyclopropane has the effect of decreasing the fluidity of the cell wall. Similar transitions can be observed using purified methyl mycolates from each strain after an initial melting/cooling cycle (Fig. 5A). Transitions using purified mycolates occur at consistently lower temperatures than intact cell walls. Using purified MAMEs, a proximal cyclopropane also resulted in a higher thermal transition than in control MAMEs, although in this case, the distal cyclopropane lowered the observed temperature slightly. DISCUSSION In this work, we have used a homologous sequence from the M. leprae genome sequencing project to identify the protein involved in construction of the proximal cyclopropane from M. tuberculosis. This enzyme is the fourth identified member of a family of proteins to catalyze the transfer of a methylene group from S-adenosyl-L-methionine to the double bond of a fatty acid substrate. The three mycobacterial members of this family are closely related to one another, with the cma2 genes from M. leprae and M. tuberculosis being more closely related to one another (73% identity) than to the cma1 gene of M. tuberculosis (52% identity). Heterologous expression of cma2 in M. smeg-  matis results in a proportion of the ␣-mycolates becoming cyclopropanated at the proximal position. Expression of cma1 results in cyclopropanation at the distal position, while coexpression of both genes results in the production of a dicyclopropyl mycolate nearly identical to the major mycolic acid produced by M. tuberculosis.
The cyclopropanation of the epoxy mycolates by CMAS-2 in M. smegmatis suggests that the enzyme is either insensitive to substituents occurring toward the end of the chain or that CMAS-2 acts on a precursor meromycolate, which can become either cyclopropanated to form the dicyclopropyl mycolate or further oxidized to form the epoxy series. CMAS-2 activity is unchanged upon co-expression of both cyclopropane synthases with about 30% of the total mycolates cyclopropanated at the proximal position (Table I). Total CMAS-1 activity, however, increases upon coexpression from 30 to 50% cyclopropanation of the distal position. One interpretation of this result is that the distal cyclopropane is formed after the proximal cyclopropane with CMAS-1 preferentially recognizing the proximally cyclopropanated precursor as a substrate.
The biological significance of lipid cyclopropanation has been most extensively studied in E. coli; however, the lack of any dramatic phenotype associated with either cyclopropane fatty acid synthase null mutants or cyclopropane fatty acid synthase overexpressors has left the role cyclopropanation plays in cellular metabolism unclear (28,29). A large increase in the synthesis of cyclopropane-containing plasma membrane fatty acids has been shown to accompany the transition from log to stationary phase, which suggests that cyclopropanation offers some protective advantage to stationary cultures (27). E. coli, which have been grown on cyclopropane fatty acids, are more resistant to killing by hyperbaric oxygen treatment, suggesting that cyclopropanes do have a stabilizing or rigidifying effect on the membrane (30). This is confirmed by the increased susceptibility to killing by freezing observed in cyclopropane fatty acid synthase mutants of E. coli (29). It has also been shown by examining the 2 H NMR of specifically deuterated cyclopropane-containing lipids, that cyclopropanated membranes enhance stability by suppressing segmental mobility of hydrocarbon chains, thus providing increased rigidity with respect to external shock (31). These studies consistently support the position that cyclopropanation of membrane lipids, although a rather subtle modification, does contribute to increased structural integrity of membranes containing short chain fatty acids (32). In addition, cyclopropanation is intermediate in fluidity effects between the more fluid cis-olefin and the less fluid trans-olefin as measured by DSC (33).
Recent work on the structure of the mycobacterial cell wall suggests that the proximal cyclopropane lies at the boundary of what Minnikin (3,26) has referred to as the structural permeability barrier. A dramatic high temperature phase transition has recently been demonstrated to occur at 60°C in purified cell walls of M. chelonei by DSC (4). The temperature of this transition suggests that at physiologically relevant temperatures, much of the cell wall exists in a state of exceptionally low fluidity. Cyclopropanation of mycolic acids, in addition to rendering lipids less susceptible to peroxidation, may decrease the actual fluidity even more, thus contributing to the overall impermeability of the cell wall. We examined the effect of substitution of a cis-olefin with a cis-cyclopropane in mycolic acids on cell wall thermochemistry and showed, with either purified cell walls or MAMEs, that proximal cyclopropanation increased the observed temperature of the transition by approximately 3°C. The magnitude of this change seems quite reasonable since substitution of a cis-cyclopropane for a cis-olefin in the much shorter palmitoleate (C16:1), raises the observed temperature of phase transition by 15.6°C (33), and only about 30% of the mycolates are converted to the cyclopropanated form. The distal cyclopropane had no such effect, possibly reflecting the role of this cyclopropane in interacting with other lipids that form a less tightly associated region that is not observed by DSC of detergent-extracted cell walls. In fact, our M. smegmatis cell wall preparations gave significantly lower melting temperatures than purified cell walls from M. smegmatis prepared without detergent extraction 3 presumably due to the loss of ancillary lipids during the Triton X-114 extraction.
The impermeability of the mycobacterial cell wall is a hallmark of the organism. In the case of slow growing and pathogenic mycobacteria such as M. tuberculosis, it seems likely that high durability of mycolic acids would be essential, especially in the face of environmental and host-initiated oxidative stress in its intracellular habitat (13,16). Dicyclopropyl mycolic acids are the major species found in many slow growing and pathogenic strains of mycobacteria including M. avium, Mycobacterium kansasii, M. marianum, M. leprae, Mycobacterium paratuberculosis, and M. tuberculosis (3). In contrast fast growing saprophytic mycobacteria such as M. smegmatis, Mycobacterium phlei, and Mycobacterium chelonae appear to possess primarily diunsaturated mycolic acids with an abundance of cis-olefins (34). In the case of the distal cyclopropane, we have previously demonstrated that expression in M. smegmatis results in significant protection from hydrogen peroxide (13). In the case of the proximal cyclopropane, we have been unsuccessful in demonstrating a similar role in protection from oxidative stress (data not shown). This may be related to the largely internal and less accessible location of the proximal cyclopropane.
Cyclopropanation of fatty acids only occurs in a small number of related taxa of bacteria. Among mycobacteria, this modification is limited to the slow growing pathogens. Mammals do not cyclopropanate unsaturated lipids. Thus, enzymes catalyzing this unique modification constitute a viable target for the design of new chemotherapy against pathogenic mycobacteria, as well as providing the tools for understanding the biosynthesis, regulation, and function of these complex lipids.