Investigating the Function of the Putative Mycolic Acid Methyltransferase UmaA

Mycolic acids are major and specific lipid components of the cell envelope of mycobacteria that include the causative agents of tuberculosis and leprosy, Mycobacterium tuberculosis and Mycobacterium leprae, respectively. Subtle structural variations that are known to be crucial for both their virulence and the permeability of their cell envelope occur in mycolic acids. Among these are the introduction of cyclopropyl groups and methyl branches by mycolic acid S-adenosylmethionine-dependent methyltransferases (MA-MTs). While the functions of seven of the M. tuberculosis MA-MTs have been either established or strongly presumed nothing is known of the roles of the remaining umaA gene product and those of M. smegmatis MA-MTs. Mutants of the M. tuberculosis umaA gene and its putative M. smegmatis orthologue, MSMEG0913, were created. The lipid extracts of the resulting mutants were analyzed in detail using a combination of analytical techniques such as matrix-assisted laser desorption/ionization time-of-flight mass spectrometry and proton nuclear magnetic resonance spectroscopy, and chemical degradation methods. The M. smegmatis mutants no longer synthesized subtypes of mycolates containing a methyl branch adjacent to either trans cyclopropyl group or trans double bond at the “proximal” position of both α- and epoxy-mycolates. Complementation with MSMEG0913, but not with umaA, fully restored the wild-type phenotype in M. smegmatis. Consistently, no modification was observed in the structures of mycolic acids produced by the M. tuberculosis umaA mutant. These data proved that despite their synteny and high similarity umaA and MSMEG0913 are not functionally orthologous.

Mycolic acids, ␣-branched ␤-hydroxylated long chain fatty acids, represent major and very specific constituents of the lipid-rich envelope of mycobacteria such as Mycobacterium tuberculosis, the causative agent of tuberculosis, and play an important structural role in the envelope architecture (1)(2)(3)(4). For instance, mycolic acids bound to the cell wall arabinogalactan are organized with other lipids to form an outer barrier with an extremely low fluidity that confers an exceptional low permeability to mycobacteria and may explain their intrinsic resistance to many antibiotics (5). Other natural constituents containing mycolic acids, such as trehalose mycolates, are known to be implicated in numerous biological functions related both to the physiology and virulence of M. tuberculosis (3). Furthermore, their biosynthesis is the only clearly identified target inhibited by the major antitubercular drug, isoniazid (2,3,6). Therefore, deciphering the biosynthesis pathway leading to mycolates still represents a major objective.
Mycolic acids are found in mycobacterial species as a mixture of structurally related molecules that differ primarily from one another by the nature of the chemical groups at the so-called "proximal" and "distal" positions of their main (meromycolic) chain (Fig. 1). In M. tuberculosis and phylogenetically close species, as well as in several slow growing mycobacteria, three types of mycolates commonly occur (7,8). The least polar ␣-mycolates of M. tuberculosis consist of C 76 -82 fatty acids (9) and contain two cis cyclopropyl groups. The more polar mycolates are composed of C 80 -89 (9) and contain both a cis or trans (with a methyl group on the vicinal carbon atom) cyclopropyl group at the proximal position, and a methoxy-or keto-group (with a methyl group on the vicinal carbon atom) at the distal position (Fig. 1A). Mutations resulting in either the loss or the modification of these chemical functions profoundly alter the perme-ability of the cell envelope to solutes and severely affect the virulence and pathogenicity of the mutant strains in experimental infections (10 -14). Accordingly, the enzymatic systems that introduce the chemical modifications in the mycolic acid chain, mycolic acid S-adenosylmethionine-dependent methyltransferases (MA-MTs), 4 merit special attention as these dis-crete structural changes may be of crucial biological importance. In M. smegmatis and other rapid growing mycobacterial species, which are generally non pathogens, cis and trans double bonds are found in place of cyclopropyl groups in both ␣-mycolates and related oxygenated molecules (Fig. 1B).
M. tuberculosis has eight predicted MA-MTs (cmaA1-2, mmaA1-4, pcaA, and umaA) (15). These map on the genome with mmaA1-4 together, cmaA1-2 separately and pcaA and umaA together and convergently transcribed. Of the eight MA-MT genes, four have been unambiguously characterized through gene inactivation. Characterization of the resulting mutant strains has shown that the cmaA2 gene encodes a methyltransferase required for the synthesis of trans cyclopropyl at the proximal position of methoxy-and keto-mycolates in M. tuberculosis (16) whereas the mmaA2 gene catalyzes the introduction of a cis cyclopropyl group at both the distal position of ␣-mycolates and proximal position of methoxy-and keto-mycolates (17). The mmaA4 gene (18), also called hma, has been shown to be necessary for the synthesis of both keto-and methoxy-mycolic acids (10) whereas the umaA2 (Rv0470c) gene has been shown to encode the methyltransferase that is responsible for the cyclopropanation of the double bound at the proximal position of ␣-mycolates in M. tuberculosis and, accordingly, renamed pcaA (12). PcaA has been shown to be required for both the cording of the tubercle bacillus and its long term virulence in mice (12) and is absent from the genome of the non pathogenic M. smegmatis. For the remaining candidates, the cmaA1, mmaA1, and mmaA3 genes from M. tuberculosis have been overexpressed in a surrogate host, M. smegmatis, and the three putative MA-MTs have been shown to function as methyltransferases (18,19) although subsequent inactivation of cmaA1 in M. tuberculosis has led to no phenotypical change (17). In contrast, nothing is known on the reaction catalyzed by the remaining umaA (Mt-umaA) gene product.
M. smegmatis has at least seven MA-MT paralogues; two genes are present in the mmaA1-4 region, and one in the umaA-pcaA region in the same orientation as Mt-umaA. The cmaA1 and cmaA2 regions are not present in M. smegmatis although other genes in these regions are conserved. M. smegmatis has four additional paralogues. All seven proteins are similar (73-77% similarity) and the conserved gene arrangement in some loci suggests that some of the M. tuberculosis and M. smegmatis genes may be or may have been orthologues. Mt-umaA is most similar to M. smegmatis MSMEG0913 in terms of homology (85% similarity) and is also syntenic. The present study was undertaken in order to determine whether the two genes are orthologues and to investigate their functions.
Isolation of a M. smegmatis MSMEG0913 Mutants-A library of Tn611 (20) mutants of M. smegmatis was screened using inverse PCR and amplification of the chromosomal sequence flanking the site of insertion with the Tn611-specific oligonucleotide primers (Supplemental Table S1) on a genomic DNA template that had been digested with RsaI and then selfligated. One mutant, Myco69, had a Tn611 insertion 117-bp upstream of the MSMEG0913 start codon.
A mutant of M. smegmatis where the MSMEG0913 open reading frame (ORF) was disrupted was made by inserting a kan resistance gene into the ORF. A 2.5-kb fragment containing MSMEG0913 was PCR-amplified from M. smegmatis mc 2 155 using oligonucleotide primers 436 and 437 (Supplemental Table S1) and the PCR product was cloned into pBLUNTII-TOPO (Invitrogen). The insert was excised from this plasmid with Xba1 and cloned into the Xba1 site of the sacB-str suicide vector, pHBJ428 (21). The resulting plasmid, pHBJ474 was linearized with Nco1 within the MSMEG0913 ORF and a kan resistance cassette was inserted to disrupt the ORF. The kan cassette was in the opposite orientation to the MSMEG0913 ORF. The resulting plasmid, pHBJ489, was transformed into M. smegmatis and transformants were plated onto Middlebrook 7H10 agar containing either kan or str. A single crossover strain, Myco734 was isolated and grown in Middlebrook 7H9 broth with no antibiotics for 3 days. The cultures were spread on Middlebrook 7H10 agar containing kan and sucrose (10% w/v). A mutant, Myco735, with a double crossover (kan R str S ) was isolated. Disruption of MSMEG0913 in Myco735 was confirmed by Southern blot analysis. Both MSMEG0913 mutants had the same colony morphology and growth characteristics as M. smegmatis mc 2 155 when grown on Middlebrook media.
Construction of a M. tuberculosis umaA Mutant-Mutagenesis was carried out essentially as described previously (22). The coding sequence of M. tuberculosis H37Rv umaA with flanking DNA, 978-bp upstream and 977-bp downstream of the gene was amplified by PCR using primers umaA1 and umaA2 (Supplemental Table S1) in a reaction with 10% v/v Me 2 SO. The PCR product was phosphorylated using polynucleotide kinase and cloned into the PmlI site of plasmid p2NIL, producing pFM173. The umaA1 gene was eliminated by inverse PCR, using umaAR1 and umaAR2 (Supplemental Table S1) in a reaction with 5% v/v Me 2 SO. The PCR product consisting of the vector and the regions originally flanking umaA was phosphorylated using polynucleotide kinase followed by re-ligation (pFM183). Following insertion of a gene cassette carrying lacZ and sacB from pGOAL19 into the vector PacI site to produce pFM194, the DNA was introduced into M. tuberculosis H 37 Rv by electroporation. Cells carrying single-crossovers were isolated by selection for blue hyg R kan R colonies on Middlebrook 7H11 agar containing OADC supplement (Difco). One single crossover colony was plated onto agar containing 2% w/v sucrose to isolate bacteria with a double crossover event.
Mutant strains were selected by PCR and confirmed by Southern analysis (data not shown). One of these (FAME89) was analyzed further.
Complementation of the M. smegmatis MSMEG0913 Mutant-The mutant, Myco735, was complemented with MSMEG0913, which had been amplified by PCR using primers 420 and 421 (Supplemental Table S1) and cloned as a BamH1/ HindIII fragment into pVV16-creating strain Myco740. Myco735 with a vector control was named Myco742. This construct contained the entire MSMEG0913 gene, and part of each flanking gene. Mt-umaA and Mt-mmaA1 genes were each PCR-amplified (primer 604 and 605 and 612 and 613, respectively (Supplemental Table S1) and cloned into pCR-BluntII-TOPO then subcloned individually as a BamH1-HindIII fragments into pVV16, which had been digested with the same enzymes producing pHBJ616 (Mt-umaA) and pHBJ624 (Mt-mmaA1). A second Mt-umaA complementation construct was made in a version of pVV16 in which the hsp60 promoter had been removed by Xba1 digestion. The cloned fragment of M. tuberculosis DNA contained the 5Ј-end of Mt-uma through to 1-kb upstream of the gene, including the natural promoter. All complementation constructs expressed recombinant proteins with C-terminal hexahistidine epitopes. The expression of the proteins in M. smegmatis strains was confirmed by Western blotting of cell lysates with anti-hexahistidine monoclonal antibodies (data not shown).
Purification of Mycolic Acids-Whole cells or bacterial residues obtained after lipid extraction with organic solvents (10) were saponified by a mixture of 40% KOH and methoxyethanol (1:7, v/v) at 110°C for 3 h in a screw-capped tube. After acidification, fatty acids were extracted with diethyl ether and methylated with an ethereal solution of diazomethane (7). The mycolate patterns of the strains were determined by analytical thin-layer chromatography (TLC) on Silica Gel 60 (Silica Gel 60 Macherey-Nagel) using either eluent A (dichloromethane) or eluent C (petroleum ether/diethyl ether; 9:1, v/v, five runs). Revelation of lipid spots was performed by spraying the plates with molybdophosphoric acid (10% in ethanol), followed by charring. The different classes of mycolates were separated by chromatography on a Florisil column irrigated with increasing concentrations of diethyl ether (0, 10, 20, 30, and 50%, v/v) in petroleum ether and purification was achieved by preparative TLC using eluent A (9). To search for the presence of ethylenic compounds, the various purified mycolate types were analyzed and fractionated on AgNO 3 -impregnated silica gel TLC plates developed with eluent A (CH 2 Cl 2 ). The various purified types and subtypes of mycolates were quantified by weighing.
Degradative Techniques-Cleavage of double bonds was performed by permanganate-periodate oxidation (23) at 30°C in tertiary butanol as solvent. The acids obtained from the oxidative cleavages were methylated and purified on preparative TLC with dichloromethane as eluent.
Miscellaneous Analytical Techniques-1 H NMR spectra of purified mycolic acid methyl esters were obtained in CDCl3 (100% D) using a Bruker AMX-500 spectrometer at 298 K. Chemical shifts values (in p.p.m.) were relative to the internal CHCl3 resonance (at 7.27 ppm).

Function of the Putative Mycolic Acid Methyltransferase UmaA
MALDI-MS detection of purified samples in reflectron mode was performed on a 4700 Proteomics Analyzer (Applied Biosystems, Voyager DE-STR, Framingham, MA) equipped with a Nd :YAG laser (355 nm) operating by pulses of 500 ps with a frequency of 200 Hz. 2500 shots were accumulated in positive ion mode and MS data were acquired using the instrument default calibration. Mycolates samples were dissolved in chloroform, at the concentration of 1 mM, and were directly spotted onto the target plate as 0.5-l droplets, followed by the addition of 0.5 l of matrix solution. Samples were allowed to crystallize at room temperature. The matrix used was 2,5-dihydroxybenzoic acid (10 mg/ml) in CHCl 3 /CH 3 OH (1:1, v/v).
Gas chromatography (GC) of fatty acid methyl esters derived either from the saponification of whole cells or from degradation of purified mycolates was performed on a Hewlett-Packard 5890 series II apparatus equipped with an OV1 capillary column (0.30 mm ϫ 25 m) using helium gas. The temperature separation program involved an increase from 100 to 300°C at the rate of 5°C/min, followed by 10 min at 300°C.

M. smegmatis MSMEG0913 Mutants Have Subtle Modifications in the Structures of
Mycolates-Lipids of M. smegmatis mc 2 155 and mutants Myco69 and Myco735 were analyzed as described. All the strains contained the same types of major extractable lipids, which included triacylglycerol, glycerol mycolate, trehalose mono-and di-mycolates, and phospholipids (data not shown). M. smegmatis mc 2 155, Myco69, and Myco735, had comparable contents in extractable lipids (20% of the bacterial dry mass) and in wall-linked mycolates (9% of the bacterial residue dry mass). They also showed identical C 16 -C 24 fatty acid methyl ester profiles on GC and types of mycolates on silica gel TLC, namely ␣-, ␣Јand epoxy-mycolates ( Fig. 2A). To determine whether the inactivation of the MSMEG0913 resulted in the absence methyl branches in the mycolates of the mutant strains, the various types of mycolates of the parental and mutant strains of M. smegmatis were purified and comparatively analyzed by MALDI-TOF mass spectrometry (9) and 1 H NMR spectroscopy (24). The mass and NMR spectra of purified mycolates of the Myco69 and Myco735 were superimposable but different from those of the mc 2 155. As a consequence, only the data from Myco735 will be presented hereafter.
Although the mass spectra of the ␣-mycolates of the wild-type and the mutant contained pseudomolecular ion (MϩNa) ϩ peaks assignable to C 74 to C 82 ␣-mycolic acid methyl esters the major peaks were observed at 1160 and 1188 m/z (corresponding to C 77 and C 79, respectively) in the mass spectra of mc 2 155 whereas they were seen at 1146, 1174 m/z (corresponding to C 76 and C 78 ), respectively, in the mass spectra of the mutant strains (Fig. 3A). Similarly, the major pseudomolecular ion peaks assignable to epoxymycolates were observed at 1204 and 1218 m/z (corresponding to C 79 and C 80 , respectively) in the mass spectra of mc 2 155 whereas they were seen at 1176 m/z (C 77 ) and 1204 m/z (C 79 ) in the mass spectra of the mutant strains (Fig. 3B). These differences were independent from both the temperature and the phase of growth of the strains and suggested that modifications have occurred in the structures of both ␣and epoxy-mycolates of Myco69 and Myco735. In contrast, comparable intense pseudomolecular ion peaks attributable to C 62 and C 64 ␣Ј-mycolic acid methyl esters, at 951.92 and 979.95 m/z, were observed in the mass spectra of the purified mycolates from all strains (data not shown), confirming the absence of methyl branch in ␣Ј-mycolates (3).
Purification and Characterization of the Subtypes of ␣-Mycolates from M. smegmatis-To further structurally characterize the mycolates from the mutant strains and those of the parent the various purified mycolates were further analyzed by AgNO 3 -impregnated TLC (Fig. 2B), a layer that is known to have an affinity for cis-ethylenic bonds. A single spot was detected for ␣Ј-mycolates from mc 2 155 and the mutants, as expected (data not shown). In contrast, four major ␣-mycolate spots, called ␣ 1 to ␣ 4 and a minor ␣ 5 form, were detected and isolated by preparative AgNO 3 -impregnated TLC from mc 2 155 whereas three of them, ␣ 1 , ␣ 3 , and ␣ 5 were observed in the mutants (Fig. 2B). The ␣ 1 -mycolates, present in all the strains, exhibited the same mobility on AgNO 3 -impregnated TLC as the dicyclopropanated mycolic acid methyl esters from the control strain while the ␣ 2 -and ␣ 3 -mycolates had a migration consistent with the presence of one cyclopropyl group and one double bond of trans and cis configuration, respectively. The more polar ␣ 4 -and ␣ 5 -mycolates may correspond to diethylenic molecules. Analysis of the MALDI-TOF mass spectra of the various purified compounds showed that the ␣ 1 -mycolates from mc 2 155 were composed of C 78 , C 80 , C 81 , and C 83 molecules. In contrast, the ␣ 1 -mycolates from the mutant Myco735 were devoid of molecules containing odd number of carbon (Table 1). ␣ 2 -mycolates from mc 2 155, a subclass of mycolates that were absent from the mutants, consisted of C 78 , C 80 , and C 82 constituents whereas ␣ 3 -and ␣ 4 -mycolates were mainly composed of C 77 and C 79 molecules (Table 1). ␣ 5 -mycolates that was abundantly produced by the mutant consisted of C 76 , C 78 , and C 80 . To discriminate between the various structural possibilities for the different ␣-mycolates, especially in terms of contents in cyclopropyl groups and/or double bonds, the puri-fied subclasses of ␣-mycolates were examined by NMR spectroscopy.
Analysis of the different types of purified mycolates by 1 H NMR showed that all the spectra contained the huge signal attributed to the isolated methylene proton resonances (at 1.29 ppm, broad signal), signals due to terminal methyl groups (at 0.85 ppm, triplet), methoxyl from the carboxyl methyl ester (at 3.71 ppm, singlet), methine located at position C-2 (at 2.50 ppm, multiplet). In addition, the spectrum of ␣ 1 -mycolates from mc 2 155 (Fig. 4a)  showing that these mycolates were composed of molecules containing both odd and even numbers of carbon. It could thus be deduced that the ␣ 1 -mycolates of mc 2 155 were composed of two types of dicyclopropanated mycolates: (i) molecules with an even number of carbon containing only cis cyclopropyl rings and (ii) compounds with an odd number of carbon containing one cis and one trans cyclopropyl groups. As trans cyclopropyl rings are known to have an adjacent methyl branch (3), the latter type of mycolates probably corresponds to the longest C 81 and C 83 molecules (Table 1). In agreement with this hypothesis, the 1 H NMR spectrum of ␣ 1 -mycolates from the mutant strains contained signals attributable to resonances of cis-cyclopropyl protons with only traces of trans molecules (Fig. 4b). This indicated  The amounts of the different subtypes of mycolates were determined by weighing the purified compounds; the amounts of ␣1 and E1 from the mc 2 155 strain were arbitrary fixed to 1. b The number of carbon for the major component is indicated in bold. c cis and trans indicated the configuration of double bonds (DB) and cyclopropyl (CP) groups in the different subtypes of ␣and epoxymycolates (E). that they consisted almost exclusively of dicyclopropanated mycolates with cis configuration and with an even number of carbon, C 80 and C 82 molecules ( Table 1).
The 1 H NMR spectrum of ␣ 2 -mycolates from mc 2 155 (Fig.  4c) contained signals attributable to resonances of cis-cyclopropyl protons and trans double bond protons (␦ 5.32) whereas that of ␣ 3 -mycolates from both strains contained signals assignable to resonances of cis double bond and cis cyclopropyl protons (Fig. 4d). Consistently, ␣ 2 -mycolates exhibited an even number of carbon (Table 1), as trans double bonds in mycolates are known to have a methyl branch at the adjacent position (3). The 1 H NMR spectrum of ␣ 4 -mycolates that also typified the parental strain was devoid of resonances of cyclopropyl protons but contained resonances attributed to both cis and trans ethylenic protons (Fig. 4e). Similarly, the 1 H NMR spectrum of ␣ 5 -mycolates, which were prominent in the mutant strain, was also devoid of resonances of cyclopropyl protons but, in contrast to that of ␣ 4 -mycolates from mc 2 155, contained resonances attributed to only cis ethylenic protons (Fig. 4f). This observation was consistent with the even number of carbons of ␣ 5 -mycolates (Table 1) and the absence of a methyl branch in mycolates of the mutant, as cis double bonds are known to be devoid of a methyl branch at the adjacent position (3). It was thus concluded that the disruption of the MSMEG0913 of M. smegmatis resulted in the production of ␣-mycolates with cis cyclopropyl groups and/or cis double bonds and the disap-pearance of both trans cyclopropyl groups (found in ␣ 1 -mycolates) and trans double bonds (found in ␣ 2 -and ␣ 4 -mycolates) of the parental strain (Table 1).
Purification and Analysis of the Subtypes of Epoxymycolates-The various classes of epoxymycolates from the mc 2 155 and Myco735 were analyzed by AgNO 3 -impregnated TLC (Fig. 2C). This showed that mc 2 155 contained three major spots, called E 1 to E 3 , whereas the mutants contained only two well visible mycolate spots, E 1 and E 3 . The E 1 -mycolates from the parental strain consisted of C 78 , C 80 , C 81 , C 82 , and C 83 molecules whereas those of the mutants were composed of molecules containing an even number of carbon only (Table 1). E 2 -mycolates that typified mc 2 155 consisted in C 78 and C 80 . E 3 -mycolates from the parental and mutant strains were composed in majority of C 77 and C 79 molecules.
Analysis of the purified subclasses of epoxymycolates by 1 H NMR (Fig. 5) showed that all the spectra contained, in addition to the common signals shared by all the mycolates (see above), the signal resonance assigned to the methine protons of a trans epoxy ring at 2.65 ppm (25)(26)(27), the resonances of cis-epoxymycolates being expected at 2.90 ppm (24). The resonances of the protons of the methyl branch adjacent to the trans epoxy group were also observed as a doublet at 1.10 ppm (25)(26)(27).
The 1 H NMR spectrum of E 1 -mycolates from mc 2 155 contained also signal resonances of both cis-and trans-cyclopropyl protons and devoid of resonances of double bond protons (Fig.  5a); the corresponding 1 H NMR spectrum of the mutants contained signal resonances of cis cyclopropyl protons with only traces of trans molecules (Fig. 5b). This observation indicated that mc 2 155 produced two populations of trans-epoxy-monocyclopropanated mycolates: (i) C 78 and C 80 cis cyclopropanated molecules and C 81 and C 83 trans cyclopropanated substances ( Table 1). The spectrum of E 2 -mycolates from mc 2 155 contained no signal resonances attributable to cyclopropyl protons (Fig. 5c); instead, it showed signals attributable to resonances of  Fig. 2B. a and b, spectra of ␣1 from the wild type and mutant strains, respectively; ␣2 (c) and ␣4 (e) were purified from the wild-type mc 2 155 strain, as they were absent from the mutant, whereas ␣5 (f) was purified from the mutant and was weakly produced by the wildtype strain. The spectra of the ␣3 (d) from the two strains were superimposable and only that from the wild-type strain is shown. Spectra were recorded in CDCl 3 at 500 MHz. Chemical shift values are expressed in ppm relative to the internal CHCl 3 (7.2 ppm). The inset corresponds to the expansion of the region of methyl proton resonances (around 1 ppm).  Fig. 2C. a and b, spectra of E1 from the wild type and mutant strains, respectively; E2 (c) was obtained from the wild-type strain and absent from the mutant strain, whereas E3 (d) was present in both strains, and their spectra were superimposable. Spectra were recorded and calibrated as in Fig. 4. The insets correspond to the expansion of the region of methyl proton resonances (around 1 ppm).
trans double bond protons (␦ 5.32); in addition, signals of the proton resonances of the methyl branch adjacent to trans double bonds were observed at 0.94 ppm (Fig. 5c). The 1 H NMR spectrum of E 3 -mycolates from all the strains contained signals assignable to resonances of cis double bond protons (at 5.34 ppm) and devoid of resonances of protons of both cyclopropyl and methyl branch adjacent to double bond (Fig. 5d). It was thus concluded that the disruption of the MSMEG0913 resulted in the production of cyclopropanated and ethylenic epoxymycolates and the disappearance of the methyl branch that occurs at the proximal position of trans cyclopropyl ring (found in a fraction of E 1 -mycolates) and trans double bond (found in E 2 -mycolates) from the parental strain.
Localization of Double Bonds in Mycolates-Data on the location of the methyl branches adjacent to trans double bonds in mycolates, the production of which was abolished in the Myco69 and Myco735 mutants, were obtained by oxidative cleavage (23), a method known to cleave ethylenic bonds of mycolates and to yield mono-and/or di-carboxylic acids (11,24). When the ␣ 5 -mycolates from Myco735 were oxidized and the resulting fatty acids esterified, purified, and analyzed by TLC, three spots were detected; their migrations corresponded to those of esters of non-hydroxylated fatty acid (e.g. C 18 ), of -carboxylic acid (e.g. C 14 ) and of hydroxylated -carboxylic acid (data not shown). GC and GC/MS analyses of the nonhydroxylated fatty acid methyl esters identified them as heptadecanoïc, nonadecanoïc, and eicosanoic acid methyl esters, establishing the values of m 1 (Fig. 1 and Table S2) as 15, 17, and 19. Similarly, analysis of the -carboxylic acid methyl esters by GC and GC/MS identified an octadeca-dioic (-carboxylic acid) methyl ester as the major diester, establishing the values of m 2 as 14. The remaining long-chain hydroxylated diester cleavage products of mycolates were purified and analyzed by MALDI-TOF mass spectrometry, comparatively to the -carboxylic acid methyl ester of similar structure isolated from M. phlei (9). Analysis of the MALDI-TOF mass spectra of the long-chain diacid methyl esters showed pseudomolecular peaks at 703, 731, and 759 m/z. Pyrolytic cleavage of these hydroxylated -carboxylic acid methyl esters in the GC apparatus liberated a tetracosanoic acid methyl ester, establishing the value of m 4 as 21 and allowing the determination of m 3 as 15, 17, and 19.
To determine the location of the double bond in monocyclopropanated compounds, the ␣ 2 -mycolates from mc 2 155 were oxidized and the resulting products and derivatives were analyzed. TLC analysis showed the presence of two spots whose migrations corresponded to those of very long chain (ϾC 30 ) non-hydroxylated fatty ester and hydroxylated -carboxylic acid ester (data not shown). This observation allowed the location of the cyclopropyl ring at the distal position of mycolates. Further analyses of the oxidative products were performed by MALDI-TOF mass spectrometry. The data derived from these various chemical analyses allowed the structural determination of the mycolates from the parental mc 2 155 and umaA mutant strains of M. smegmatis (Supplemental Table S2).
Complementation of M. smegmatis Mutants with MSMEG0913-To prove the function of MSMEG0913 complementation experiments were needed, as transformation of mutants with the wild-type gene are expected to restore the parental phenotype. Accordingly, M. smegmatis mutant Myco735 was complemented with the wild-type gene from M. smegmatis mc 2 155, and the transformants were analyzed.
Analysis of the various types of mycolates from the complemented strain by AgNO 3 -impregnated TLC (Fig. 2, B and C) showed that the production of the mc 2 155 subtypes of mycolates, i.e. ␣ 1 to ␣ 5 and E 1 to E 3 , was restored. Expectedly, transformation of Myco735 strain with the empty vector pVV16, as a control for complementation did not restore the parental phenotype (data not shown). Examination of the MALDI-TOF mass spectra of the subtypes of mycolates of the complemented strain confirmed the identity of mycolates from the complemented strain with those of the wild-type strain (data not shown).
Inactivation of umaA in M. tuberculosis Did Not Affect the Mycolic Acid Composition-M. tuberculosis CDC1551 has a frameshift mutation in umaA; therefore it is a naturally occurring umaA mutant. Analysis of its mycolates by MALDI-TOF mass spectrometry showed no significant differences, compared with those of M. tuberculosis H 37 Rv (data not shown), suggesting that umaA is not critical for mycolic acid synthesis in this strain. Because there could be other changes in CDC1551 that account for the results observed, a umaA-deleted mutant strain of H 37 Rv was constructed (FAME89). This mutant exhibited similar content in extractable lipids (20 -21% of the bacterial dry mass) and in wall-linked mycolates (10% of the bacterial residue dry mass) to M. tuberculosis H37Rv. No obvious difference was seen between the two strains in terms of major extractable lipids detected by TLC; these included phthiocerol dimycocerosates, sulfatides, triacylglycerol, glycerol mycolate, trehalose monomycolate, trehalose dimycolates, and phospholipids (data not shown). They also showed identical C 16 -C 26 fatty acid methyl ester profiles on GC and types of mycolates on silica gel TLC, namely ␣-, methoxy-and ketomycolates (Fig. 2D). To determine whether the inactivation of Mt-umaA resulted in the absence of cyclopropyl groups and/or methyl branches in the mycolates of the mutant strains, the total fatty methyl esters of FAME89 and H 37 Rv were comparatively analyzed by MALDI-TOF mass spectrometry. The mass spectra of the lipid compounds from the mutant strain were superimposable to those of the wild-type H 37 Rv strain, notably in the region of 1100 -1300 m/z, which corresponds to the masses of mycolates (9). The major pseudomolecular ion (MϩNa) ϩ peaks assignable to ␣-mycolates were observed at 1146, 1174, and 1202 m/z; these corresponded to C 76 , C 78 , and C 80 , respectively. Two other series of peaks were attributable to methoxy-(major peaks at m/z 1262 for C 83 , 1290 for C 85 , and 1318 for C 87 ) and keto-mycolates (major peaks at m/z 1246 for C 82 , 1274 for C 84 , 1302 for C 86 , 1316 for C 87 , and 1330 for C 88 ) were observed in the MALDI-TOF mass spectra of the fatty acid methyl esters from both strains. The lack of modification of mycolate structures in the mutant was confirmed by AgNO 3impregnated TLC. Cells from the H 37 Rv and FAME89 harvested in either early or late logarithmic phase exhibited identical lipid patterns, as judged by GC, TLC and mass spectrometry (data not shown). It was thus concluded that the inactivation of the umaA gene in M. tuberculosis did not result in a significant change in the structures of the mycolates of the resulting mutant strain.
Complementation of M. smegmatis Myco735 with Mt-umaA and Mt-mmaA1-Possible reasons to explain the lack of effect of the deletion of Mt-umaA from M. tuberculosis are (i) Mt-umaA is not expressed in axenic culture, (ii) Mt-umaA is redundant and another MA-MT can compensate for it, or (iii) Mt-umaA is not an orthologue of MSMEG0913. The first possibility is unlikely as transcripts of Mt-umaA have been detected under the same conditions used here, and the polypeptide has been identified in the M. tuberculosis proteome. Redundancy is a possibility; MSMEG0913 is most similar in sequence to umaA and mmaA1 of M. tuberculosis, but there is still a degree of divergence. To test hypotheses ii and iii, we cloned Mt-umaA and Mt-mmaA1 into pVV16, and transformed them into M. smegmatis mc 2 155 and Myco735 to test if either could complement the MSMEG0913 mutation. Expression of the M. tuberculosis genes in both M. smegmatis strains was confirmed by Western blotting of the C-terminal hexahistidine epitopes. Analysis of lipids showed that neither Mt-umaA nor Mt-mmaA1 complemented Myco735. The result for Mt-umaA was surprising. Closer examination of the Mt-umaA sequence showed that there are two possible translational start codons. The codon atg2 is the annotated start of an 861-bp gene (Tuberculist), and the annotation is reasonable given the occurrence of a putative ribosome binding site just upstream. The alternative start, atg1 would add 6 amino acids to the N-terminal of UmaA. Another version of Mt-umaA was cloned to include the extra amino acids. The construct was introduced into M. smegmatis mc 2 155 and Myco735, expression was confirmed by Western blotting, and the lipids were analyzed as before; however, there was no functional complementation. It was thus concluded that MSMEG0913 is not the orthologue of Mt-umaA.

DISCUSSION
The lipids of M. smegmatis mc 2 155 were comparatively analyzed with those two MSMEG9013 mutants of M. smegmatis, Myco735 and Myco69. Application of MALDI-TOF mass spectrometry, 1 H NMR spectroscopy and chemical degradation techniques to the analysis of the various purified subclasses of mycolates produced by the M. smegmatis mutants revealed new structural features of mycolates from M. smegmatis despite the numerous studies previously dedicated to these compounds since 1960 (28 -30). For instance, this is the first report of the existence of trans cyclopronated ␣-mycolates in the mycobacterial world. More importantly, the series of structural analyses established the function of the MSMEG0913 protein as an enzyme that adds a methyl branch to the vicinal position of both a cis double bond and cyclopropyl group located at the proximal position of both ␣and epoxy-mycolic acids to yield trans homologues. The mutants lacked all semblances of trans double bound in their mycolates while producing normal amounts of both cis mono-and di-ethylenic ␣-mycolates, di-cis-cyclopropanated ␣-mycolates and monoethylenic and mono-cis-cyclopropanated epoxymycolates. This mycolate phenotype was specific to the absence of the MSMEG0913 gene because complementation with the wild-type gene fully restored the occurrence of both trans double bond and cyclopropane in both ␣and epoxy-mycolates.
The M. smegmatis results contrasted with those obtained from analysis of M. tuberculosis Mt-umaA mutants, CDC1551 and FAME89 which had the same mycolate phenotypes as the control strain, M. tuberculosis H 37 Rv. No differences were observed between the lipid patterns of the mutants and those of the control strain, in terms of amounts of extractable and cell wall-linked lipids and in structures of these compounds and their constituents implying that the umaA gene is not essential for the production of the methyl branches and cyclopropyl rings in mycolates in M. tuberculosis. The redundancy of Mt-umaA appears to be shared by M. leprae that produces mycolates exhibiting the same structures as those of M. tuberculosis, i.e. dicyclopropanated ␣-, and (cis and trans) monocyclopropanated keto-and methoxy-mycolates (31,32); however, the M. leprae umaA gene is predicted to be a pseudogene (33). It may be simply that another MA-MT can catalyze the reactions performed by the M. smegmatis MSMEG0913. However, despite their synteny and high homology, the failure of Mt-umaA or Mt-mmaA1 to complement the M. smegmatis mutants showed that the genes are no longer functionally orthologous.
McAdam et al. (34) showed that Tn disruption of Mt-umaA resulted in hypervirulence of M. tuberculosis in SCID mice, proving that the gene is functional in H37Rv, and inferred that altered mycolates had a role in the phenotype. Our data showing that Mt-umaA deletion had no detectable effect on mycolates in axenic culture suggest that the role of Mt-umaA may depend on environmental factors that occur in vivo. The situation with Mt-umaA may be comparable to that of Fbps of M. tuberculosis. Although the inactivation of fbpA affected the multiplication of the mutant in macrophages (35), no differences were seen in terms of lipid content and profile between the parental and the fbpA mutant (36). Interestingly, the fbpA was shown to complement the deficiency of the fbpC mutant (36), establishing its function. In other mycobacterial species and/or models of infection, the lack of production of the UmaA protein may result in an attenuation phenotype, as reported for M. paratuberculosis (37). Further studies are thus warranted to elucidate the precise function of UmaA in species other than M. smegmatis.