The mycolic acid reductase Rv2509 has distinct structural motifs and is essential for growth in slow‐growing mycobacteria

Abstract The final step in mycolic acid biosynthesis in Mycobacterium tuberculosis is catalysed by mycolyl reductase encoded by the Rv2509 gene. Sequence analysis and homology modelling indicate that Rv2509 belongs to the short‐chain fatty acid dehydrogenase/reductase (SDR) family, but with some distinct features that warrant its classification as belonging to a novel family of short‐chain dehydrogenases. In particular, the predicted structure revealed a unique α‐helical C‐terminal region which we demonstrated to be essential for Rv2509 function, though this region did not seem to play any role in protein stabilisation or oligomerisation. We also show that unlike the M. smegmatis homologue which was not essential for growth, Rv2509 was an essential gene in slow‐growing mycobacteria. A knockdown strain of the BCG2529 gene, the Rv2509 homologue in Mycobacterium bovis BCG, was unable to grow following the conditional depletion of BCG2529. This conditional depletion also led to a reduction of mature mycolic acid production and accumulation of intermediates derived from 3‐oxo‐mycolate precursors. Our studies demonstrate novel features of the mycolyl reductase Rv2509 and outline its role in mycobacterial growth, highlighting its potential as a new target for therapies.

array of enzymes that includes a mammalian-like Fatty Acid Synthase I (FAS-I), and a multienzyme complex of Fatty Acid Synthase-II (FAS-II) (Marrakchi et al., 2014). One of the late stages of mycolic acid biosynthesis involves the Claisen condensation of a FAS-II-derived long meromycolate chain (C 42 -C 62 ) with a short FAS-I-derived fatty acid (C 24 -C 26 ) to yield α-alkyl, β-keto fatty acid intermediate, catalysed by the polyketide synthase Pks13 (Gande et al., 2004;Portevin et al., 2004). The final step in the formation of a mature mycolic acid involves the enzymatic reduction of the β-keto group of the product of Pks13 to produce α-alkyl, β-hydroxy fatty acids (mature mycolic acid). In M. tuberculosis, the reductase required for this final step is encoded by Rv2509 (Bhatt, Brown, Singh, Minnikin, & Besra, 2008;Lea-Smith et al., 2007). Initial studies in Corynebacterium glutamicum, a species that can survive the loss of mycolic acids, indicated that deletion of cmrA (NCgl2385), the Rv2509 orthologue, resulted in a strain that accumulated the α-alkyl, β-keto fatty acid precursor instead of mature mycolic acids (Lea-Smith et al., 2007). Mycolic acid biosynthesis genes are essential in mycobacteria; however, surprisingly, we were able to generate a viable mutant of MSMEG4722, the homologue of Rv2509 in the fast-growing Mycobacterium smegmatis (Bhatt et al., 2008). Similar to the C. glutamicum NCgl2385 mutant, the M. smegmatis MSMEG4722 mutant produced α-alkyl, β-keto fatty acyl precursors of mycolic acids (3-oxo-mycolic acid precursors) which were transported and esterified to arabinogalactan in the cell wall (Bhatt et al., 2008). While the precise knowledge of the substrate for Rv2509 is lacking, functional analysis of the preceding enzyme in the biosynthetic pathway, Pks13, indicated that, while this polyketide synthase primarily catalyses the formation of the α-alkyl, β-keto fatty acyl precursor, it also contains enzymatic motifs that facilitate the release of the nascent fatty acyl chain and its subsequent transfer to a trehalose residue to produce a trehalose residue esterified with the α-alkyl, β-keto fatty acyl precursor of mycolic acid (Gavalda et al., 2014). These findings suggest that the trehalose-bound mono 3-oxo-mycolic acid precursor may likely be the substrate for Rv2509, which catalyses its conversion to TMM (Figure 1). In this study, we set out to do a detailed study of the amino acid sequence of Rv2509 with the aim of deducing its structure and functional connections, which led to the identification of series of unique sequences, and by proxy, novel structural features. We then probed the role of one distinct structural feature of the mycobacterial mycolyl reductase in vitro by deletion analysis. We also probed the essentiality of Rv2509 for the growth and viability of the slow-growing M. tuberculosis complex using Mycobacterium bovis BCG.

| Homologues of M. tuberculosis Rv2509 are found across mycolic acid-producing species
While mycolic acids are produced by several species belonging to the suborder Corynebacterineae, the biosynthesis of mycolates has been studied in detail exclusively in the genera Mycobacterium and Corynebacterium. While the two genera share a common set of enzymes required for mycolic acid biosynthesis, select differences also exist. For example, while the meromycolate chains are synthesised by a Type-II fatty acid synthase complex in species of Mycobacterium, the same function is carried out by a mammalian-like Type-I fatty acid synthase in Corynebacterium species (Radmacher et al., 2005).
Similarly, differences exist in the transport of mycolic acids between the two genera (Varela et al., 2012;Yang et al., 2014). To investigate whether the reduction of premycolates by an exclusive mycolyl reductase, was an enzymatic process conserved across other mycolic acid-producing species, we used the amino acid sequence of Rv2509 as a source template for a BLASTp search of genome sequences of other known mycolic acid-producing bacteria. Homologues of Rv2509 were found across the mycolate-producing genera (Figure 2).
A phylogenetic analysis of Rv2509, its homologues in other mycobacteria and in select mycolic acid-producing species revealed that mycolyl reductases from Corynebacterial species branched early (Supporting information Figure S1). Corynebacteria produce relatively short chain mycolic acids (Radmacher et al., 2005). To test whether mycolyl reductases across the mycolata had evolved specificities to accommodate differing chain lengths, we transformed the M. smegmatis ΔMSMEG4722 strain with a plasmid-containing C.
glutamicum mycolyl reductase gene NCgl2385 to generate the strain ΔMSMEG4722-CNCgl2385. Mature mycolic acid production was restored in the transformed strain, indicating that NCgl2385 could reduce M. smegmatis 3-oxo-mycolic acid precursors and consequently fully complement the mutant M. smegmatis strain. This outcome suggested that mycolyl reductases across the mycolata likely did not evolve specificities for differing chain lengths ( Figure 3).

| Predicted structure of M. tuberculosis Rv2509
We attempted to deduce possible functional properties of Rv2509 using a comparative sequence and structural analysis with proteins of known function. Sequence analysis of Rv2509 using the NCBI conserved domain database (Marchler-Bauer et al., 2015) categorises the protein as a part of the DltE family of short-chain dehydrogenases (cluster COG0300), which are in turn members of the cl27753 short-chain dehydrogenase/reductase (SDR) superfamily of NAD(P)-dependent oxidoreductases. This is a diverse group, containing over 80,000 unique sequences (Fujisawa, Nagata, & Misono, 2003). However, despite this diversity, all SDR-superfamilies display the classical Rossmann fold typical of NAD + /NADP + -binding proteins and hence can be classified as belonging to the cl21454 family of topologically and structurally related proteins.
Next, we screened the peptide sequence of Rv2509 against the PDB database to find structural templates for homology modelling. The serine dehydrogenase YdfG (UniProt ID P39831; PDB ID 3ASU/3ASV) (Yamazawa, Nakajima, Mushiake, Yoshimoto, & Ito, 2011) had the highest identity score; however, structural analysis revealed several gaps in the pairwise alignment and the C-terminal region was shorter than that observed in Rv2509, which necessitated additional template input. Therefore, the structures of the four highest-scoring experimental SDRs (3ASU/3ASV, 1XG5, 4X54 and 4BMV) were used in parallel as a basis for comparative homology modelling of Rv2509 (Figure 4) using the I-TASSER package (Yang et al., 2015) with specific template selection enabled. Notably, while, three of the templates yielded models with roughly equivalent (confidence) C-scores, the one based on the SDR from Sphingobium yanoikuyae (4BMV.pdb; GenBank ID ACB78183.1) stood out with the highest C-score of 0.6, indicating a high quality of prediction.
As visualised by the structural alignment ( Figure 2) which shows the predicted secondary structure elements of Rv2509 aligned to its Corynebacterial homologues, the core structure of the Rv2509  and was limited in terms of sequences used. We, therefore, sought to expand this initial analysis by performing a systematic analysis of the Rv2509 orthologues from a wide range of Corynebacterineae to represent all known mycolate-producing genera.
We initiated these studies by analysing the cofactor binding site. SDRs can be divided into two large families, "Classical" SDRs which are around 250 amino acid residues long and "Extended" SDRs which contain around 350 residues (Jornvall et al., 1995), exhibiting differences in the arrangement of glycine residues within the coenzyme binding sites. Coenzyme preference-predictions based on the strong conservation of the signature residues (Persson et al., 2003) classified the Rv2509 into the group of the NADP(H)-binding SDRs, consistent with its mycolyl reductase function. However, it was difficult to categorically allocate Rv2509 into a specific group of SDRs as the predicted NADP-binding site of Rv2509 (G 16 XXQ 19 N 20 I 21 G 22 ) displays a unique and novel variation of the (T)-G-X-X-X-G-I-G motif described by Brakoulias and Jackson (2004) (Figure 2). The variation seems to be restricted to the long-chain mycolates-producing species of Mycobacteriaceae, as short-chain mycolates-producing Corynebacteria retain the more established G-X-X-X-G-I-G motif. (Figure 2). Key residue F I G U R E 2 A structural alignment of the predicted Rv2509 structure from Mycobacterium tuberculosis, with annotated secondary structure elements (top) with the sequences of the Mycobacterial and Corynebacterial homologues highlighting the unique sequence motifs of the Rv2509 subfamily of SDRs. The GxxQNIG motif along with RR pair identifies a novel NADP(H) binding consensus, limited to Mycobacteriaceae (highlighted in yellow). GxxSGIG is the wider consensus seen in the more remotely related Corynebacteria. Similarly, the unique Rossmann core-stabilisation signature CANAGT reported here is also restricted to the Mycobacteriaceae (highlighted in yellow) and reverts to the more common (C)NNAG(I/F) in the more remote homologues. Active site residues N-G-S-Y-K in alpha-helix 4 and alphahelix 5 are preserved throughout the family; however, a unique additional tyrosine residue can be seen to be present in Mycobacteriaceae (highlighted in yellow). The C-terminal extension is also uniquely conserved across the wider Corynebacterial species positions in the NADP(H) site suggest that Rv2509 is a cP2 member of the "Classic" SDR family, according to the classification system of Persson et al., (Brakoulias & Jackson, 2004), which is The active site in SDRs is formed by a tight cleft, lined by α-helix 4, β-strand 4 and α-helix 5 (Filling et al., 2002;Oppermann et al., 1997). Rv2509 exhibits the "classical" SDR motif Yx[AS][ST]K in α-helix 5 (Y157 and K161 being the central catalytic residues), while the GxggxgSS/T motif in β-strand 4 (G137, S144) and the central residue of α-helix 4, N116 ( Figure 2) are also instantly recognisable. In addition, it is possible that some contact with the substrate may be provided by the β-strand 4.
Rather intriguingly, the consensus residues "NNAG" involved in the stabilisation of the central β-sheet of the Rossmann fold in the "Classical" SDRs and located at the core of the protein are not fully conserved in Rv2509 but take the form of a novel "ANAG" sequence (see Figure 2). Furthermore, this novel sequence is paired with a preceding Cys residue, which is also correlated with the presence of a C-terminal Thr (in position 91 in M. tuberculosis). This unique "CANAGT" signature makes Rv2509 orthologues in Mycobacteriaceae instantly recognisable ( Figure 2).
In SDR proteins, the substrate-binding loop links the sixth β-sheet with the seventh α-helix and is one of the larger loops. In the predicted structure of Rv2509 this loop spans roughly from P 187 to S 214 and appears largely conserved within the mycolic ac- The most intriguing finding upon the analysis of the Rv2509 orthologues was that all of them seem to present a C-terminal domain which seems to be highly specific to the Corynebacterineae. It proved also the most difficult to model reliably, as in the majority of the available SDR structural templates, the C-terminal region, appears to be either truncated or disorganized. ConSurf mapping using the multiple sequence alignment shown in Figure 2 and multiple sequence alignment based on 150 nonredundant sequences of remote SDR homologues revealed that the C-terminus of Rv2509 is uniquely a feature of the subfamily of Corynebacteriales (Supporting information Figure S3). Both pairwise and multiple sequence analyses indicated that the C-terminal region of the 4BMV. pdb is closely related to Rv2509 and thus can be tentatively used as a template for that region. This was further corroborated that by comparative modelling (Song et al., 2013) and pure ab-initio modelling (Raman et al., 2009), as implemented in the NewRobetta server (http://new.robet ta.org). To validate this hypothesis, we conducted additional ab initio secondary structure predictions of Rv2509 (236-268) using JPred4 (Drozdetskiy, Cole, Procter, & Barton, 2015) and PSIPRED (Buchan & Jones, 2019;Jones, 1999).

Both of these bioinformatics tools supported a high alpha-helical
propensity of the C-terminal region of Rv2509, suggesting that F I G U R E 3 Complementation of the M. smegmatis ΔMSMEG4722 mutant with the Corynebacterial mycolyl reductase. TLC analysis of 14 C-labelled mycolic acid methyl esters (MAMEs) isolated from different strains. α; α MAMEs, α′; α′ MAMEs, e; epoxy MAMEs, O; origin, D, degradation products derived from 3-oxo-mycolate precursors. Solvent system; petroleum ether:acetone (95:5, v:v) 4BMV as a reliable template for the region (Supporting information Figure S3). Indeed, the C-terminus of the 4BMV.pdb structure, similarly to Rv2509, appears to be extended and is well-structured, presenting an α-helical organisation.

| The extended C-terminal of Rv2509 is essential for the function
Mature mycolic acid production could be restored in the M. smegmatis ΔMSMEG4722 mutant on complementation with the M. tuberculosis orthologue Rv2509 (Bhatt et al., 2008). The mutant strain thus provided us with a means of identifying residues and domains required for Rv2509 function, by assessing the ability of mutated constructs of plasmid-borne Rv2509 to complement the ΔMSMEG4722 strain by restoring mature mycolic acid biosynthesis. To test if the extended C-terminal of Rv2509 was required for function, we generated truncated versions of Rv2509 cloned in the replicative shuttle plasmid pMV261 (Stover et al., 1991). Individual plasmids were then introduced by electroporation into the M. smegmatis ΔMSMEG4722 mutant and the transformed strain analysed for the restoration of mature mycolic acid biosynthesis. Deletion of 14 amino acid residues (Ala 255 -Ser 268 ) from F I G U R E 4 Visualisation of the predicted structure of Mycobacterium tuberculosis Rv2509 monomer. Two different orientations are given; sideview and top-down view towards the coenzyme binding site. The protein is coloured like a rainbow from blue (N-terminus) to red (C-terminus) and the main secondary structure elements are labelled, as well as the predicted C-terminal extension domain is circled in a red oval. The location of the key sequence and structural motifs discussed in the text is also provided the C-terminal of Rv2509 resulted in a complete loss of function as the strains transformed with the truncated construct failed to restore mature mycolic acid biosynthesis in the ΔMSMEG4722 strain ( Figure 5; ΔMSMEG4722-CRvD1). A construct with a shorter truncation (Tyr 262 -Ser 268 deleted) also failed to complement the ΔMSMEG4722 strain ( Figure 5; ΔMSMEG4722-CRvD2). Expression of Rv2509 and truncated constructs could be detected by RT-PCR confirming that the loss of Rv2509 function was linked to the deletion of the C-terminal residues (Supporting information Figure S4). These results demonstrated that the C-terminus of Rv2509 spanning terminal residues Ala 255 -Ser 268 was critical for function. The findings suggested an essential role for the unique, C-terminal extension in the activity of Rv2509.
To further query whether this C-terminal extension played a role in either protein stability, protecting Rv2509 from degradation, or in oligomerisation, we expressed N-terminal His-tagged versions of Rv2509 and Rv2509-D1 (Rv2509 with Ala 255 -Ser 268 deleted) in E.
coli. Western blot analysis of native polyacrylamide gels showed that there were no patterns of degradation detected in Rv2509-D1, indicating that the loss of function due to the deletion of this region was not related to protein stability (Supporting information Figure S5).
Furthermore, we did not observe additional slower migrating bands, indicating dimerisation (or oligomerisation) of full-length or truncated Rv2509. Furthermore, cross-linking Rv2509 with glutaraldehyde did not reveal additional slower migrating bands, suggesting that Rv2509 did not oligomerise in solution (Supporting information Figure S5).

| Rv2509 is an essential mycobacterial gene
Given the nonessentiality of the mycolyl reductase gene in M. smegmatis and prediction of transposon-site hybridisation screens of a slowgrowing M. tuberculosis Rv2509 mutant, we aimed to generate a null mutant of Rv2509 in M. tuberculosis using Specialised Transduction (Bardarov et al., 2002). Such a mutant would also allow us to test the role of the mycolyl reductase in virulence using macrophage and mouse models of infection. However, repeated attempts failed to generate transductants, suggesting that Rv2509 was essential for growth in M. tuberculosis. Indeed, subsequent transposon mutagenesis screens utilising deep sequencing predicted Rv2509 to be essential for in vitro growth of M. tuberculosis (Griffin et al., 2011). We then utilised a pro- The addition of ATc resulted in the loss of growth of BCG::P Tet -BCG2529 in broth, indicating that the expression of the gene encoding the mycolyl reductase was essential for the in vitro growth of slow-growing mycobacteria (Figure 6a). Furthermore, while concentrated cultures of BCG::P Tet -BCG2529 showed confluent growth on 7H10 agar plates, no confluent growth was observed on plates containing ATc (Figure 6b). We did observe a few scattered colonies on the ATc-containing plates and these are likely to be suppressors allowing leaky expression of P Tet driven BCG2529. These results demonstrated that the mycolyl reductase was essential for the growth and viability of slow-growing mycobacteria.

| Conditional depletion of the mycolyl reductase results in the loss of mature mycolic acids
The M. smegmatis mycolyl reductase mutant was unable to synthesise fully mature mycolic acids and instead produced 3-oxo-mycolate precursors that were found esterified to the cell wall arabinogalactan and to trehalose (Bhatt et al., 2008). These labile precursors gave rise to palmitone derivatives following alkali treatment, which were visualised on Thin Layer Chromatography (TLC) plates migrating close to the F I G U R E 5 TLC analysis of 14 C-labelled MAMEs isolated from the strains of M. smegmatis ΔMSMEG4722 mutant transformed with the different deletion constructs of Rv2509. α; α MAMEs, α′; α′ MAMEs, e; epoxy MAMEs, O; origin, D, degradation products derived from 3-oxo-mycolate precursors. Solvent system; petroleum ether:acetone (95:5, v:v) solvent front in a solvent system used for separating methyl esters of mycolic acids and fatty acids (FAMEs and MAMEs) (Bhatt et al., 2008).
To test the effects of conditional depletion of the mycolyl reductase we labelled the cultures of BCG::P Tet -BCG2529 with 14 [C]-acetic acid following grown in the presence and absence of ATc, and extracted

MAMEs from the cultures. TLC analysis of the extracted MAMES
showed a decrease in the levels of α and keto MAMES and an accumulation of the degradation products derived from 3-oxo-mycolate precursors close to the solvent front (Figure 7). These results showed that, as in M. smegmatis, loss of the mycolyl reductase function led to the loss of mature mycolic acid production and accumulation of the premature 3-oxo-mycolate precursors in slow grow mycobacteria.

| D ISCUSS I ON
As summarised above, the predicted structure of the Rv2509 displays a Rossmann fold with seven parallel β-strands forming a central beta-sheet, which is sandwiched between two layers of alpha-helices-three on each side (Figure 4). While largely consistent with the "classical" family of SDRs (Persson et al., 2003), in Mycobacteriaceae,

the NADP(H)-binding motif (G-x-x-Q-N-I-G), alongside the core stabilisation consensus (C-A-N-A-G-T) is significantly divergent from
the consensus, to instantly identify this group of SDRs, and in our view is distinct enough to be separated into a distinct family.
While we currently lack knowledge about the exact mode of cofactor and ligand binding within the Rv2509 family, some information could be inferred from the structurally related FabG family of β-ketoacyl reductases which are classified as "Complex family SDRs" (Persson et al., 2003). Given that the structural overlay of Rv2509 models with existing ligand-bound FabG members yields an RMSD of around 1 Å over the C-alpha backbone atoms, and that FabG share common active site residues with Rv2509 it is tempting to suggest that these structures share common ligand-binding and NADP(H) orientations towards the active site residues with that observed in the aceto-acyl-CoAs. Furthermore, the predicted 3D structure of Rv2509 also revealed a unique α-helical C-terminal extension, which we subsequently showed to be essential for function-constructs with deletions in the C-terminal failed to complement (restore mycolate biosynthesis) in the M. smegmatis mutant.
The role of this C-terminal domain is difficult to assign with certainty; however, in related SDRs similar C-terminal extensions play a number of functions. In the NADP(H)-dependent tetrameric serine-dehydrogenase YdfG from E. coli (3ASV.pdb) (Yamazawa et al., 2011), the C-terminal region protrudes from the main core of the protein and is required for tetramer formation, while it also plays a role in the substrate binding via stabilisation of the substrate loop. It is also notable that in FabG family of related SDRs discussed above also organise as tetrameric assemblies with the participation of their C-terminal domains (Javidpour et al., 2014). Furthermore, in the mycobacterial FabG4 dimer, Arg146 and Arg445 of one protomer interact with the C-terminus of the second protomer and play an essential role in the substrate association and catalysis (Dutta, Bhattacharyya, Roychowdhury, Biswas, & Das, 2013). However, our experiments with purified full-length and truncated Rv2509 do not seem to suggest that the mycolyl reductase forms oligomers.
While it isn't in itself surprising that Rv2509 presents a con-  broth or on 7H10 agar plates. E. coli strains were grown in LB broth or LB agar. Hygromycin was used at a concentration of 100 mg/ml for E. coli and mycobacterial strains. Kanamycin was used at a concentration of 25 mg/ml and 50 mg/ml for mycobacteria and E. coli respectively. Recombinant mycobacterial strains, phages and plasmid constructs used in this study are listed and described in Table 1.

| Expression, purification and analysis of Rv2509 and its C-terminal truncated derivative
The reaction was halted with 1M Tris pH 8 and subsequently run on SDS-PAGE for visualisation using SYPRO-ruby (Molecular Probes).

| Conditional depletion of the Rv2509 homologue BCG2529 in M. bovis BCG-Pasteur
For establishing the regulated expression of the BCG2529 gene, a synthetic gene cassette (hyg-Pmyc1-4XtetO) comprising a hygromycin-resistance gene and the Pmyc1 promoter from M. smegmatis engineered to contain four tetO operator sites (Korte et al., 2016) was inserted immediately upstream of the BCG2529 start codon in M. bovis BCG-Pasteur. Targeted gene knock-in was achieved by specialised transduction employing temperature-sensitive mycobacteriophages essentially as described previously (Korte et al., 2016). Briefly, for the generation of an allelic exchange construct for site-specific insertion  Figure S6).
For achieving the controlled gene expression of the Mb2537 gene, a synthetic gene (rev-tetR) derived from Tn10 tetR encoding a mutated TetR protein with reversed binding affinity to tetO sites upon the binding of tetracycline (Klotzsche, Ehrt, & Schnappinger, 2009) was heterologously expressed in the knock-in mutant by electroporation of the episomal E. coli-mycobacterium shuttle plasmid pMV261::rev-tetR-RBS-D providing constitutive rev-tetR gene expression from the HSP60 promoter in mycobacteria using solid medium containing 50 mg/L of hygromycin and 20 mg/L of kanamycin for selection (Famulla et al., 2016). This yielded the conditional mutant BCG-Pasteur c-BCG2529-4XtetO pMV261::rev-tetR-RBS-D (referred to as BCG::P Tet -BCG2529) allowing silencing of the BCG2529 gene in the presence of anhydrotetracycline (ATc).
For silencing experiments, precultures of the BCG::P Tet -BCG2529 strain were first grown in liquid medium containing 0.3 µg/ml of ATc, before these precultures were used to inoculate the test cultures containing 0-5 µg/ml of ATc. Growth in 96-well microtitre plates was quantified after incubation at 37°C for five days by adding 10% of 100 µg/ml of resazurin solution. After further incubation at ambient temperature for 16 hr, cells were fixed for 30 min by formalin addition (5%, v/v, final concentration) and fluorescence was quantified using a microplate reader (excitation 540 nm, emission 590 nm). For the growth on 7H10 agar, nonpermissive conditions were obtained by adding ATc to a final concentration of 5 µg/ml.

| Functional complementation studies
Deletion constructs for Rv2509 were generated by PCR amplification of selected regions of Rv2509 using pMV261-Rv2509 as a template, incorporating a premature stop codon in the reverse primer. The plasmid constructs generated in this manner were called pMV261-Rv2509D1 and pMV261-Rv2509D2, encoding a truncated protein missing the C-terminal residues Ala 255 -Ser 268 and Tyr 262 -Ser 268 respectively. The C. glutamicum Rv2509 orthologue, NCgl2385 was PCR amplified using C. glutamicum genomic DNA as a template and the PCR product was cloned into pMV261 using primer incorporated BamHI and EcoRI sites generating the plasmid pMV261-NCgl2385. Plasmid clones were electroporated into the M. smegmatis ΔMSMEG4722 mutant strain using previously described protocols (Snapper et al., 1990). Kanamycin-resistant transformants were cultured in TSB and FAMEs and MAMEs were extracted from the cell pellets and separated by TLC as previously described (Vilcheze & Jacobs, 2007).

| RT-PCR of Rv2509 transcripts
M. smegmatis strains were grown to mid-log phase and RNA was extracted by acid phenol and chloroform extraction. The purified RNA was treated with DNase and converted to cDNA, which was used as a template for quantitative PCR. PCR products were run on a 1% agarose gel to detect the expression of Rv2509. The gene sigA was monitored as a constitutively expressed gene. The primers used for PCR amplification were Rv2509RT-f (5′-ACAAGTACCGCGTCACGGTC -3′) and Rv2509RT-r (5′-AAGTCCGGCACC AGCTTCTC-3′) for the amplification of nondeleted sections of Rv2509.

ACK N OWLED G M ENTS
CC is funded by a BBSRC PhD studentship provided by the BBSRC