Identification of a novel α(1→6) mannopyranosyltransferase MptB from Corynebacterium glutamicum by deletion of a conserved gene, NCgl1505, affords a lipomannan- and lipoarabinomannan-deficient mutant

Mycobacterium tuberculosis and Corynebacterium glutamicum share a similar cell wall structure and orthologous enzymes involved in cell wall assembly. Herein, we have studied C. glutamicum NCgl1505, the orthologue of putative glycosyltransferases Rv1459c from M. tuberculosis and MSMEG3120 from Mycobacterium smegmatis. Deletion of NCgl1505 resulted in the absence of lipomannan (Cg-LM-A), lipoarabinomannan (Cg-LAM) and a multi-mannosylated polymer (Cg-LM-B) based on a 1,2-di-O-C16/C18:1-(α-D-glucopyranosyluronic acid)-(1→3)-glycerol (GlcAGroAc2) anchor, while syntheses of triacylated-phosphatidyl-myo-inositol dimannoside (Ac1PIM2) and Man1GlcAGroAc2 were still abundant in whole cells. Cell-free incubation of C. glutamicum membranes with GDP-[14C]Man established that C. glutamicum synthesized a novel α(1→6)-linked linear form of Cg-LM-A and Cg-LM-B from Ac1PIM2 and Man1GlcAGroAc2 respectively. Furthermore, deletion of NCgl1505 also led to the absence of in vitro synthesized linear Cg-LM-A and Cg-LM-B, demonstrating that NCgl1505 was involved in core α(1→6) mannan biosynthesis of Cg-LM-A and Cg-LM-B, extending Ac1PI[14C]M2 and [14C]Man1GlcAGroAc2 primers respectively. Use of the acceptor α-D-Manp-(1→6)-α-D-Manp-O-C8 in an in vitro cell-free assay confirmed NCgl1505 as an α(1→6) mannopyranosyltransferase, now termed MptB. While Rv1459c and MSMEG3120 demonstrated similar in vitroα(1→6) mannopyranosyltransferase activity, deletion of the Rv1459c homologue in M. smegmatis did not result in loss of mycobacterial LM/LAM, indicating a functional redundancy for this enzyme in mycobacteria.


Introduction
The taxon Corynebacterineae belongs to the Actinomycetes family which includes human pathogens, such as Mycobacterium tuberculosis, Mycobacterium leprae and Corynebacterium diphtheriae, the causal agents of tuberculosis, leprosy and diphtheria respectively (Coyle and Lipsky, 1990;Bloom and Murray, 1992). Some animal pathogens, for instance, Corynebacterium pseudotuberculosis and Corynebacterium matruchotii (Coyle and Lipsky, 1990;Funke et al., 1997;Stackebrandt et al., 1997), also belong to the Corynebacterianeae. In addition, the family member Corynebacterium glutamicum is widely used for the industrial production of amino acids (Eggeling and Bott, 2005). These bacilli share a unique cell wall ultra-structure that is composed of a mycolylarabinogalactan-peptidoglycan (mAGP) complex (Daffé et al., 1990;McNeil et al., 1990;1991;Besra et al., 1995;Brennan, 2003;Dover et al., 2004). The esterified mycolates of the mAGP complex are considered to be packed side by side and are intercalated by lipids and glycolipids. This combined lipid structure gives rise to an asymmetric bilayer critical for the survival of these organisms (Minnikin et al., 2002).
In this study, we have examined the function of C. glutamicum NCgl1505, and its orthologous genes Rv1459c of M. tuberculosis and MSMEG3120 of M. smegmatis encoding a putative GT-C glycosyltransferase. The NCgl1505 gene and its orthologues based on the results described below have been designated as mptB, as an acronym for mannopyranosyltransferase B. Null mutants of C. glutamicum together with in vitro cellfree assays established that NCgl1505 is a key a(1→6) mannosyltransferase involved in the initiation of core mannanbiosynthesisofCg-LM-AandCg-LM-BfromCorynebacterineae extending Ac1PIM2 and Man1GlcAGroAc2 respectively. In addition, the M. tuberculosis orthologue Rv1459c and M. smegmatis MSMEG3120 demonstrated a(1→6) mannosyltransferase activity in a membranebased in vitro assay when utilizing a C. glutamicumD-mptBDmptA double mutant complemented with either plasmid-encoded Rv1459c or MSMEG3120. Finally, using a M. smegmatis null mutant of MSMEG3120, we also demonstrate that the mycobacterial orthologue of NCgl1505 is functionally redundant.

Genome locus and structural features of Rv1459c/NCgl1505
Glycosyltransferases belonging to the GT-C superfamily have been shown by us (Alderwick et al., 2005;Alderwick et al., 2006b;Mishra et al., 2007;Seidel et al., 2007) and others Kaur et al., 2006;Morita et al., 2006) to play important roles in the biosynthesis of the cell wall heteropolysaccharides arabinogalactan (AG), LM-A, LM-B and LAM in Corynebacterinaeae. Our attention was recently drawn to a putative glycosyltransferase encoded by M. tuberculosis Rv1459c and C. glutamicum NCgl1505, which are members of the GT-C family of glycosyltransferases.
Orthologues of these genes are present in all Mycobacterium and Corynebacterium species as well as the sequenced Nocardia farcinica IFM 10152 and Rhodococcus sp. RHA1 strains ( Fig. 2A). In addition, this gene is retained in M. leprae, supporting the hypothesis that NCgl1505 encodes for a protein possessing a vital function inherent to this group of bacteria.
The glycosyltransferase encoded by NCgl1505 is a polytopic membrane protein, which is comprised of 558 amino acid (aa) residues, and is predicted to encode 15 hydrophobic segments (HSs) (Fig. 2B). Rv1459c constitutes 591 aa, with the additional length mostly due to an extended loop between HSs 7 and 8. This loop extension is not present in Mycobacterium paratuberculosis or M. smegmatis. It contains a number of repeated Pro and Arg residues, and similarly highly charged repeat sequences are found in loop regions of other transporters, without having a specific function (Eng et al., 1998;Vrljic et al., 1999). The sequence identity of the orthologues NCgl1505 and Rv1459c is 37% (52% similarity) and can therefore be considered very high. The strongest conserved regions are found in loops connecting HSs and adjacent regions with intermediate hydrophobicity, like those between HSs 3-4, HSs 7-8 and HSs 13-14 (Fig. 2B). Within the highest conserved regions; five of the six fully conserved acidic Asp and Glu residues are located, given as D and E in Fig. 2B, which are known to play important roles as general bases and nucleophiles in enzyme catalysis. They are also retained in the MptB orthologue in N. farcinica IFM 10152 and Rhodococcus sp. RHA1 are therefore likely to be involved in catalysis, or in interactions with the sugar donor or acceptor (Liu and Mushegian, 2003). Interestingly, among the glycosyltransferases of M. tuberculosis and C. glutamicum previously identified (Alderwick et al., 2006a;Dinadayala et al., 2006;Kaur et al., 2006;Morita et al., 2006;Seidel et al., 2007), NCgl1505 and Rv1459c possess the highest identities to the recently identified mannosyltransferase MptA Mishra et al., 2007) and, based on the results described below, the NCgl1505 gene and its orthologues have been designated as MptB.

Construction and growth of C. glutamicumDmptB and complemented strains
In order to delete mptB in C. glutamicum, the nonreplicative plasmid pK19mobsacBDmptB was constructed carrying sequences adjacent to Cg-mptB. Using this vector, C. glutamicum was transformed to kanamycin resistance, indicating integration of the vector into the genome by homologous recombination (Fig. 2C). The sacB gene enables for selection of loss of vector in a second homologous recombination event, which can Identification of a novel a(1,6) mannopyranosyltransferase 1597 result either in the original wild-type genomic organization or in clones deleted of Cg-mptB. Ninety clones exhibiting the desired phenotype of vector loss (kanamycinsensitive, sucrose-resistant) were analysed by PCR, but only one single colony was found to have Cg-mptB excised, whereas the others resulted in a wild-type genotype. The low number of recombinant knockouts indicates that the loss of Cg-mptB is apparently a disadvantage for cell viability, similar to that of previously observed mutants with altered mycolate (Gande et al., 2004) or arabinogalactan biosynthesis (Alderwick et al., 2006b). The resulting clone was subsequently termed C. glutamic-umDmptB and confirmed by PCR with different primer pairs to have Cg-mptB deleted, whereas controls with C. glutamicum wild type resulted in the expected larger amplification product (Fig. 2C).
In liquid culture, growth of C. glutamicumDmptB was very poor. Only when rich brain heart infusion (BHI) medium was used was a growth rate of 0.13 h -1 obtained ( Fig. 2D) in comparison with wild-type C. glutamicum growth rate of 0.31 h -1 (Mishra et al., 2007) and, on the same medium supplemented with 500 mM sorbitol (BHIS), the growth rate was 0.51 h -1 , which is still lower than that of the wild type on this medium (0.70 h -1 ). C. glutamicumDmptB was transformed with pVWEx-Cg-mptB and the resultant complemented strain exhibited a growth rate of 0.66 h -1 , almost superimposable to that of the wild type in BHIS medium. A. The locus in the bacteria analysed consists of mptB which has in C. glutamicum the locus tag NCgl1505 and in M. tuberculosis Rv1459c. sufR encodes a transcriptional regulator in front of an operon of the SUF machinery of [Fe-S] cluster synthesis (Huet et al., 2006). The genomic region displayed encompasses 7 kb, and orthologous genes are highlighted accordingly. Nocardia farcina, Nocardia farcina IFM 10152; Rhodococcus, Rhodococcus sp. strain RHA1. B. MptB is a hydrophobic protein predicted to span the membrane 15 times and the transmembrane helices are numbered accordingly. The lower part of the figure shows the degree of conservation of the orthologues given in A as analysed by the DIALIGN method (Morgenstern, 2004). Also shown is the approximate position of the fully conserved aspartyl (D) and glutamyl (E) residues. C. Strategy to delete Cg-mptB using the deletion vector pK19mobsacBDmptB. This vector carries 18 nucleotides of the 5′ end of Cg-mptB and 36 nucleotides of its 3′ end, thereby enabling the in-frame deletion of almost the entire Cg-mptB gene. The arrows marked PA and PB locate the primers used for the PCR analysis to confirm the absence of Cg-mptB. Distances are not drawn to scale. The results of the PCR analysis with the primer pair PA/PB are shown on the right. Amplification products obtained from the wild type ( Lyophilized cells were extracted using petroleum-ether and methanolic saline to initially recover apolar lipids. Further processing of the methanolic extract afforded the polar lipid fraction which was examined by twodimensional thin-layer chromatography (2D-TLC). In both the wild-type C. glutamicum and C. glutamicumD-mptB, Ac1PIM2 and Man1GlcAGroAc2  were visualized either by a-naphthol/sulphuric acid (specific for sugars), 5% ethanolic molybdophosphoric acid (general lipid stain) (Fig. S1) or Dittmer and Lester reagent (specific for phospholipids). In both C. glutamicum and C. glutamicumDmptB, no products could be observed which correspond to higher PIMs (i.e. Ac1PIM3 through to Ac1PIM6) or higher mannose variants of Man1GlcAGroAc2 Mishra et al., 2008). The presence of only Ac1PIM2 and Man1GlcAGroAc2, and the inability to synthesize Cg-LAM, Cg-LM-A and Cg-LM-B by C. glutamicumD-mptB (as shown below) demonstrated that MptB is involved in the early steps of a(1→6) mannan core biosynthesis by extending the substrates Ac1PIM2 and Man1GlcAGroAc2.
Analysis of lipoglycans from C. glutamicum, C. glutamicumDmptB and C. glutamicumDmptB pVWEx-Cg-mptB Lipoglycans were extracted by refluxing delipidated cells in ethanol, followed by hot-phenol extraction, protease digestion and dialysis to remove impurities. The extracted lipoglycans were examined initially on 15% SDS-PAGE (Fig. 3A). Extracts from wild-type C. glutamicum showed the presence of Cg-LAM, Cg-LM-A and Cg-LM-B with the latter product based on previous results comigrating with Cg-LM-A Mishra et al., 2008), while all of these lipoglycans were absent from C. glutamicumDmptB. Complementation of C. glutamicumD-mptB by transformation with plasmid pVWEx-Cg-mptB restored the wild-type phenotype (Fig. 3A). In addition, transformation of C. glutamicumDmptB with plasmid pVWEx-Cg-mptA failed to restore the wild-type phenotype (data not shown).

Construction and growth of C. glutamicumDmptADmptB and complemented strains
As a result of the similarity of MptB with MptA, we wanted to exclude any possible interferences and constructed a strains analysed using SDS-PAGE and visualized using a Pro-Q emerald glycoprotein stain (Invitrogen) specific for carbohydrates. A. Lipolglycans extracted from C. glutamicum, C. glutamicumDmptB and C. glutamicumDmptB pVWEx-Cg-mptB. The major bands represented by Cg-LAM, Cg-LM-A and Cg-LM-B are indicated. B. C. glutamicum, C. glutamicumDmptB, C. glutamicumDmptA, C. glutamicumDmptBDmptA, C. glutamicumDmptBDmptA pVWEx-Cg-mptB and C. glutamicumDmptBDmptA pVWEx-Cg-mptA. The truncated version of Cg-LM-A/B is indicated as Cg-t-LM-A/B (Mishra et al., 2007). The four major bands represent glycoproteins of 180, 82, 42 and 18 kDa respectively. strain of C. glutamicum deficient in mptB and mptA. For this purpose, C. glutamicumDmptB was transformed with plasmid pK19mobsacBDmptA (Mishra et al., 2007) and processed as described in Experimental procedures to afford the double mutant, C. glutamicumDmptBDmptA. Analysis of this strain showed that its growth characteristics were very similar to C. glutamicumDmptB (data not shown). For further analysis, C. glutamicumDmptADmptB was transformed with plasmid-encoded Cg-mptB, Cg-mptA, Mt-mptB and Ms-mptB.
Analysis of lipoglycans from C. glutamicumDmptBDmptA, C. glutamicumDmptBDmptA pVWEx-Cg-mptB and C. glutamicumDmptBDmptA pVWEx-Cg-mptA In addition to MptB, C. glutamicum possesses the known a(1→6) mannosyltransferase MptA, which is involved in the later stages of core mannan biosynthesis (Mishra et al., 2007) and, as a result, we wanted to study the in situ specificity of these glycosyltransferases. For this purpose, lipoglycans were extracted from C. glutamicum-DmptBDmptA, and from the same strain carrying either pVWEx-Cg-mptB or pVWEx-Cg-mptA and analyzed by 15% SDS-PAGE (Fig. 3B). Extracts from C. glutamicum-DmptBDmptA indicated that, as expected, no lipoglycans were present, whereas the presence of pVWEx-Cg-mptB resulted in formation of a truncated (Cg-t) version of Cg-LM-A and Cg-LM-B (Mishra et al., 2007;. However, lipoglycan extracts from C. glutamicumDmptB-DmptA carrying pVWEx-Cg-mptA were identical to that of C. glutamicumDmptBDmptA, indicating that MptA fails to substitute for MptB in the double mutant. As pVWEx-Cg-mptA results in functional MptA (Mishra et al., 2007), this result shows that MptA is unable to substitute in vivo for MptB. Therefore, both MptA and MptB are distinct and MptB is involved in the initial steps of Cg-LAM, Cg-LM-A and Cg-LM-B biosynthesis, prior to MptA. Furthermore, analysis of C. glutamicumDmptBDmptA carrying either pVWEx-Mt-mptB or pVWEx-Ms-mptB resulted in a complete lack of lipoglycan biosynthesis (data not shown), indicating that Mt-MptB and Ms-MptB do not function in vivo as the initial a(1→6) mannosyltransferase probably because of an inability to extend Ac 1PIM2 and Man1GlcAGroAc2 by mannose residues as shown below through in vitro chase experiments.
In vitro analysis of a(1,6) mannosyltransferase activity using C. glutamicumDmptB, C. glutamicumDmptBDmptA and complemented strains Initial attempts to develop an in vitro assay using either purified recombinant-expressed MptB, or Escherichia coli membranes harbouring the protein, have thus far proved unsuccessful. Alternatively, we assessed the capacity of membrane preparations from C. glutamicum and its recombinant strains to catalyse a(1→6) mannosyltransferase activity in a previously defined acceptor assay utilizing the neoglycolipid acceptor a-D-Manp-(1→6)-a-D-Manp-O-C 8 and C50-PP[ 14 C]M as a sugar donor (Brown et al., 2001) (Fig. 5A). The TLC autoradiography of products from in vitro assays when assayed with wild-type C. glutamicum resulted in the formation of product X, a trisaccharide a-D-  (Fig. 5B). These products comigrated on TLC autoradiography with the corresponding products previously chemically characterized and prepared using mycobacterial membranes, and were cleaved by acetolysis, demonstrating that they were a(1→6)-linked [ 14 C]Man products ( Fig. 5B and C) (Brown et al., 1997;. The intensity of the major product X, a trisaccharide a-D-[ 14 C]Manp-(1→6)-a-D-Manp-(1→6)-a-D-Manp-O-C8, was consistently slightly reduced in the case of C. glutamicumDmptB (89 217 Ϯ 4269 c.p.m.) in comparison with wild-type C. glutamicum (92 325 Ϯ 5017 c.p.m.) (Fig. 5B). This reduction in activity corresponded to the residual a(1→6) mannosyltransferase activity observed in C. glutamicumDmptA (2053 Ϯ 604 c.p.m.) (Fig. 5B) (Mishra et al., 2007). These results suggested the presence of two a(1→6) mannosyltransferase activities utilizing this neoglycolipid acceptor, catalysed by MptA and MptB, with the former more efficiently utilizing the neoglycolipid acceptor as a substrate. Assays containing membrane preparations from C. glutamicumDmptBDmptA showed no product formation on TLC, indicating a complete abrogation of both a(1→6) mannopyranosyltrans-ferase activities from C. glutamicum (Fig. 5B). Analysis of the double mutant with pVWEx-Cg-mptB revealed a significant but weak band (2682 Ϯ 940 c.p.m.) corresponding to product X on TLC analysis; however, when complemented with pVWEx-Cg-mptA, a similar phenotype to that of C. glutamicumDmptB could be observed (80 614 Ϯ 4135 c.p.m. for X), although at a slower transfer rate. The data confirmed that NCgl1505 is an a(1→6) mannopyranosyltransferase; however, the specific a(1→6) mannopyranosyltransferase activity is much lower in comparison with MptA, under the assay conditions utilizing the neoglycolipid acceptor.

In vitro and mutational analysis of the mycobacterial MptB
To study the function of the mycobacterial MptB, we transformed the C. glutamicumDmptBDmptA double mutant with a plasmid containing either M. tuberculosis Rv1459c (pVWEx-Mt-mptB) or M. smegmatis MSMEG3120 B. TLC analysis of products obtained in a cell-free assay for detecting a(1→6) mannosyltransferase activity with membranes prepared from M. smegmatis, C. glutamicum, C. glutamicumDmptB, C. glutamicumDmptA, C. glutamicumDmptBDmptA, C. glutamicumDmptBDmptA pVWEx-Cg-mptB and C. glutamicumDmptBDmptA pVWEx-Cg-mptA. C. TLC autoradiography of reaction products X and Y prepared with M. smegmatis and C. glutamicum membranes and subjected to acetolysis as described in the Experimental procedures (Brown et al., 1997). D. TLC analysis of products obtained in a cell-free assay for detecting a(1→6) mannosyltransferase activity with membranes prepared from C. glutamicumDmptADmptB, C. glutamicumDmptADmptB pVWEx-Mt-mptB and C. glutamicumDmptADmptB pVWEx-Ms-mptB. Assays were performed using the synthetic a-D-Manp-(1→6)-a-D-Manp-O-C8 neoglycolipid acceptor in a cell-free assay as described (Brown et al., 2001). The products of the assay were re-suspended in n-butanol before scintillation counting. The incorporation of [ 14 C]Manp was determined by subtracting counts present in control assays (incubations in the absence of acceptor), which were typically less than 100 c.p.m. per assay. The remaining labelled material was subjected to TLC using silica gel plates (5735 silca gel 60F254, Merck) developed in CHCl3:CH3OH:H2O; NH4OH (65:25:3.6:0.5, v/v/v/v) and the products visualized by phosphorimaging (Kodak K Screen). The results represent triplicate assays in three independent experiments. A schematic representation of the reaction is showed in (A) and the products X and Y are indicated by arrows.
(pVWEx-Ms-mptB). Membrane preparations of these strains restored in vitro a(1→6) mannopyranosyltransferase activity (Fig. 5D) by formation of the trisaccharide product X (Mt-MptB, 3159 Ϯ 456 c.p.m. and Ms-MptB, 2949 Ϯ 378 c.p.m.) to a similar level to that of the isogenic strain with pVWEx-Cg-mptB (Fig. 5B), showing that the M. tuberculosis and M. smegmatis gene could restore activity in an in vitro cell-free assay with the C. glutamicum double mutant. We then generated a null mutant of M. smegmatis mc 2 155 MSMEG3120 (homologue of Rv1459c) using specialized transduction (Fig. 6A), and analysed total lipids and lipoglycans in the mutant strain. Surprisingly, the mutant strain DMSMEG3120 had a total lipid profile iden-tical to the parental wild-type strain M. smegmatis mc 2 155 (TLC system designed to separate PIMs and other phospholipids is shown in Fig. 6B) and also synthesized LM and LAM (Fig. 6C). These results suggested that MSMEG3120, unlike its corynebacterial counterpart, was redundant and it was likely that another a-mannosyltransferase compensated for the loss of its function in the DMSMEG3120 mutant.

Discussion
Over the past decade, much research has been carried out on the mechanisms and genetics of mycobacterial cell wall carbohydrate biosynthesis, particularly the formation of the essential AG (Daffe et al., 1993;Besra et al., 1995;Belanger et al., 1996;Kremer et al., 2001;Alderwick et al., 2005;2006a,b;Berg et al., 2007;Seidel et al., 2007) and the immunomodulatory heteropolysaccharides LM and LAM (Schaeffer et al., 1999;Kordulakova et al., 2002;Kremer et al., 2002;Zhang et al., 2003;Dinadayala et al., 2006;Kaur et al., 2006;Mishra et al., 2007). An archetypal biosynthetic pathway is now emerging for the formation of these important macromolecules, which predominantly include enzymes from the GT-A, B and C superfamily of glycosyltransferases (Liu and Mushegian, 2003) (Fig. 1). PimA, PimB, PimB′ and PimC, all of which are GT-A/B glycosyltransferases, have been shown to be involved in PIM biosynthesis, which serves as a substrate for LM/LAM extension and maturation (Schaeffer et al., 1999;Kordulakova et al., 2002;Kremer et al., 2002;Lea-Smith et al., 2008;Mishra et al., 2008). We and others recently identified the GT-C glycosyltransferase MptA as an a(1→6) mannosyltransferase involved in intermediate LM biosynthesis, specifically in distal a(1→6) core LM formation Mishra et al., 2007). Apart from a core a(1→6) mannan backbone, a(1→2) mannose residues punctuate LM, and the GT-C glycosyltransferase Rv2181 has been identified to be responsible for some, if not all, of these branched mannose residues . At some point, LM is further glycosylated by other GT-C glycosyltransferases, such as EmbC for the biosynthesis of LAM (Zhang et al., 2003) and then mannosecapped Appelmelk et al., 2007). In this study, we have characterized the role of a putative glycosyltransferase (NCgl1505) belonging to the GT-C superfamily of glycosyltransferases (Liu and Mushegian, 2003) by virtue of genomic deletion in C. glutamicum. We present MptB as a PPM-dependent a(1→6) mannosyltransferase, involved in early stages of proximal a(1→6) core Cg-LM-A and Cg-LM-B biosynthesis in C. glutamicum (Fig. 7).
Our initial in vivo and in vitro studies of PIM and Man1GlcAGroAc2 biosynthesis in C. glutamicumDmptB highlighted no apparent change in lipid profiles, compared with those from wild-type C. glutamicum (Figs S1 and 4A). It is reasonable to conclude from the data that MptB is not involved in either early PIM or Man1GlcAGroAc2 Identification of a novel a(1,6) mannopyranosyltransferase 1605 biosynthesis. This was not surprising as these early biosynthetic steps are completely unique to enzymes belonging to the GT-A/B glycosyltransferase family, which utilize GDP-mannose as a substrate (Liu and Mushegian, 2003). Assays utilizing membrane preparations from C. glutamicum and C. glutamicumDmptB indicated that there was no further accumulation of higher mannosylated versions of PIMs and Man1GlcAGroAc2. The lack of higher mannosylated versions in C. glutamicum suggests that the next committed step in lipoglycan biosynthesis stems from Ac1PIM2 and Man1GlcAGroAc2 and that this is catalysed by Cg-MptB.
As a result of absence of MptB, C. glutamicumDmptB is unable to synthesize Cg-LAM, Cg-LM-A and Cg-LM-B in vivo, which is in contrast to our earlier studies on MptA, where a truncated Cg-LM-A and Cg-LM-B species was synthesized (Mishra et al., 2007). In C. glutamicum, we now also present in vitro evidence that Ac1PIM2 and Man1GlcAGroAc2 are acceptors for Cg-MptB, the first GT-C a-mannosyltransferase committed to Cg-LM-A and Cg-LM-B biosynthesis. This is supported by in vitro in situ chase experiments elongating the Ac1PI[ 14 C]M2 and [ 14 C]Man1GlcAGroAc2 primers by the sugar donor C50-PPM. These crucial observations, together with the presence of Ac1PIM2 and Man1GlcAGroAc2, completely support our hypothesis that Cg-MptB mannosylates Ac1PIM2 and Man1GlcAGroAc2. Our previous experiments on glycosyltransferase activities in membranes prepared from C. glutamicumDmptA identified a residual a(1→6) mannosyltransferase activity (Mishra et al., 2007). This a-mannosyltransferase activity can now be attributed to the presence of MptB as, upon its deletion in C. glutamicum, a partial depletion in a(1→6) mannosyltransferase activity is observed and a complete loss of activity is found upon deletion of both Cg-mptA and Cg-mptB. These data together with the in vivo analyses identify MptB as a bona fide a(1→6) mannosyltransferase. Interestingly, a(1→6) mannan extension is more complex in Mycobacterium based on the evidence that Mt-MptB and Ms-MptB fail to complement the C. glutamicumDmptB mutant and suggests a slightly different substrate specificity of the MptB orthologues of M. tuberculosis and M. smegmatis. Although, clearly a(1→6) mannosyltransferase(s) based on in vitro data, studies are currently underway exploring heterologous protein expression systems for Mt-MptB and Ms-MptB in combination with a variety of substrates in a revised in vitro assay format.
Given the high degree of homology between the C. glutamicum and mycobacterial orthologues of MptB and the similar organization of neighbouring genes in the two genera, we expected deletion of M. smegmatis mptB (MSMEG3120) to have the same effect as that in C. glutamicum. However, surprisingly, the M. smegmatis mptB mutant still synthesised LM and LAM, indicating that another, yet unidentified, a-mannosyltransferase could substitute for MptB in the mutant M. smegmatis strain. It has been previously shown that a high degree of functional redundancy exists in key enzymes involved in mycobacterial cell wall assembly, for instance, PimB/ PimB′ and MgtA (Schaeffer et al., 1999;Tatituri et al., 2007;Lea-Smith et al., 2008;Mishra et al., 2008), PimC , and EmbA and EmbB (Berg et al., 2007) in PIM/LM/LAM and AG biosynthesis, and the antigen 85 complex in mycolic acid biosynthesis (Puech et al., 2002). In this particular case, the C. glutamicum mutant study enabled the assignment of function to the GT-C glycosyltransferase NCgl1505, which would have otherwise not been possible if similar studies would have concentrated solely on mycobacterial species.
Interestingly, the mechanism of how Ac1PIM2 traverses the cytoplasmic membrane remains poorly understood. Bioinformatic inspection of the locus surrounding MptB has highlighted two possible candidates for potential flippases. Downstream of the putative glycosyltransferase Rv1459c, three conserved genes are located in all Corynebacterinaeae and the expression of the four-gene locus in C. glutamicum is translationally coupled (Wang et al., 2006). This presents strong evidence for a functional coupling of the putative glycosyltransferase Rv1459c with Rv1458c, Rv1457c and Rv1456c. The latter genes encode for two ABC transporter integral membrane proteins, with Rv1458c encoding for an ATP-dependent binding protein. Applying structure prediction comparisons and hidden Markov models (Soding et al., 2005), Rv1458c exhibits remote structural similarities to sugarbinding proteins of ABC carriers, such as the sugarbinding protein of Pyrococcus horikoshii or the maltose/ maltodextrin-binding protein MALK of E. coli (Lu et al., 2005). Rv1457c encodes a permease component of an ABC-2-type transporter, characteristically involved in catalysing the export of drugs and carbohydrates (Reizer et al., 1992). As transmembrane channels of ABC-2-type transporters are either homo-or heterooligomers and Rv1456c has features of a transporter protein, it is plausible to suggest that the membrane channel coupled to the glycosyltransferase might be a heterooligomer made up of Rv1457c and Rv1456c. In a previous study, Wang et al. (2006) proposed that one or more of the proteins encoded by the orthologues of Rv1456c-Rv1459c gene locus in C. matruchotii was involved in mycolic acid transport. A transposon mutant with an insertion in the cluster had an altered mycolic acid profile. However, in light of the evidence described in this work, this change in mycolylation may be an indirect effect as a result of the loss of Cg-LAM and Cg-LM-A/B. Further examination of this gene locus is required for characterization of potential roles in mycolic acid and glycolipid transport across the membrane bilayer.

Bacterial strains and growth conditions
Corynebacterium glutamicum ATCC 13032 (referred to the remainder of the text as C. glutamicum) and E. coli DH5amcr were grown in Luria-Bertani broth (Difco) at 30°C and 37°C respectively. The recombinant strains generated in this study were grown on rich BHI medium (Difco), and the salt medium CGXII used for C. glutamicum as described (Eggeling and Bott, 2005). Kanamycin and ampicillin were used at a concentration of 50 mg ml -1 . Samples for lipid analysis were prepared by harvesting cells at an OD of 10-15, followed by a saline wash and freeze drying. M. smegmatis strains were grown in Tryptic Soy Broth (TSB; Difco) containing 0.05% Tween80 (TSBT). Solid media were made by adding 1.5% agar to the above-mentioned broths. The concentrations of antibiotics used for M. smegmatis were 100 mg ml -1 for hygromycin and 20 mg ml -1 for kanamycin. M. tuberculosis H37Rv DNA was obtained from the NIH Tuberculosis Research Materials and Vaccine Testing Contract at Colorado State University. All other chemicals were of reagent grade and obtained from Sigma-Aldrich.

Construction of plasmids and strains
The genes analysed were the orthologues of Rv1459c and NCgl1505 from M. tuberculosis and C. glutamicum, respectively, termed mptB. The vectors made were pVWEx-Mt-mptB, pVWEx-Ms-mptB, pVWEx-Cg-mptB, pET-Mt-mptB, pET-Cg-mptB and pK19mobsacBDmptB. To construct the deletion vector pK19mobsacBDmptB, cross-over PCR was applied with primer pairs AB (A, CGTTAAGCTTCCAAAGGTAACCTT ATTTATGCTGGCCACAGG; B, CCCATCCACTAAACTTAAA CACGATGCGCGGCAAAGT) and CD (C, TGTTTAAGTTT AGTGGATGGGGAGTTTGAGGCGGAATCC; D, GCATGGA TCCGCGGTAAAACCTTCGCACATTTCAATG) (all primers in 5′-3′ direction) and C. glutamicum genomic DNA as template. Both amplified products were used in a second PCR with primer pairs AD to generate a 597 bp fragment consisting of sequences adjacent to Cg-mptB, which was ligated with HindIII-BamHI-cleaved pK19mobsacB.
All plasmids were confirmed by sequencing. The chromosomal deletion of Cg-mptB was performed as described previously using two rounds of positive selection (Schafer et al., 1994), and its successful deletion was verified by use of primer pair AB and the additional primer pair LM (L, GCGCGTATCACCGTCTCCGGTGTG; M, GCTGTTGGC CACCTGACAGACGTCG). Because of the similarity of MptB with MptA, C. glutamicumDmptB was transformed together with pK19mobsacBDmptA (Mishra et al., 2007) to yield the double mutant C. glutamicumDmptBDmptA. Plasmids pVWEx-Mt-mptB, pVWEx-Ms-mptB and pVWEx-Cg-mptB were introduced into C. glutamicumDmptB and C. glutamicumDmptBDmptA by electroporation with selection to kanamycin resistance (25 mg ml -1 ).
To generate an allelic recombination substrate to replace MSMEG3120 with hyg, approximately 1 kb of upstream and downstream flanking sequences were PCR-amplified from M. smegmatis mc 2 155 genomic DNA using the primer pairs MS3120LL (ttt-ttt-ttc-cat-aaa-ttg gAT-TGT-GAC-GGA-ATT-CGT-CCG-ACG-GT) and MS3120LR (ttt-ttt-ttc-cat-ttc-ttg-gAT GCC-CTG-ACC-GAT-CCA-CAG-GAA), and MS3120RL (tttttt-ttc-cat-aga-ttg-gTG-TTC-CAG-ATC-GTC-ATG-GCA-ACC-CT) and MS3120RR (ttt-ttt-ttc-cat-ctt-ttg-gAT-GAT-CAC-GAT-GCG-ATC-GGC-GAG-TT) respectively. The PCR products consisted of a 682 bp upstream DNA fragment (including the last 15 bp coding sequence of MSMEG3121, 89 bp intergenic sequence and the first 578 bp of MSMEG3120) and a 813 bp downstream DNA fragment (including the last 160 bp of MSMEG3120 and the first 655 bp of MEMEG3119). Following restriction digestion of the primer-introduced Van91I sites (shown in lower case), the PCR fragments were cloned into Van91I-digested p0004S to yield the knockout plasmid pDMSMEG3120 which was then packaged into the temperature-sensitive mycobacteriophage phAE159 as described previously (Bardarov et al., 2002), to create a recombinant phage, which was then used to transduce wildtype M. smegmatis mc 2 155 to generate the DMSMEG3120 deletion mutant which was confirmed by Southern blot and PCR analysis (data not shown).

Lipid extraction and analysis
Polar lipids and apolar lipids were extracted as described previously (Dobson et al., 1985). Briefly, 6 g of dried cells of wild-type, mutant and complemented strains of C. glutamicum or M. smegmatis were mixed thoroughly using the biphasic mixture of methanolic saline (220 ml containing 20 ml of 0.3% NaCl and 200 ml of CH3OH) and petroleum ether (220 ml) for 2 h. The upper petroleum-ether layer containing apolar lipids were separated following centrifugation. The lower methanolic saline extract was further extracted using petroleum ether (220 ml), mixed and centrifuged. The two upper petroleum-ether fractions were combined and dried. Polar lipids were extracted by the addition of CHCl3/ CH3OH/0.3% NaCl (260 ml, 9:10:3, v/v/v) added to the lower Identification of a novel a(1,6) mannopyranosyltransferase 1607 methanolic saline phase and stirred for 4 h. The mixture was filtered and the filter cake re-extracted twice with CHCl3/ CH3OH/0.3% NaCl (85 ml, 5:10:4, v/v/v). CHCl3 (145 ml) and 0.3% NaCl (145 ml) were added to the combined filtrates and stirred for 1 h. The mixture was allowed to settle, and the lower layer containing the polar lipids recovered and dried. The polar lipid extract was examined by 2D-TLC on aluminum-backed plates of silica gel 60 F254 (Merck 5554), using CHCl3/CH3OH/ H2O (60:30:6, v/v/v) in the first direction and CHCl3/CH3COOH/CH3OH/ H2O (40:25:3:6, v/v/v/v) in the second direction. C. glutamicum glycolipids were visualized by either spraying plates with a-naphthol/sulphuric acid or 5% ethanolic molybdophosphoric acid followed by gentle charring of plates. Identification of phospholipids was carried out using the Dittmer and Lester reagent as described .

Extraction and purification of lipoglycans
Lipoglycans from C. glutamicum and M. smegmatis strains were extracted as described previously (Nigou et al., 1997;Ludwiczak et al., 2002). Briefly, delipidated cells were re-suspended in deionized water and disrupted by probe sonication (MSE Soniprep 150, 12 mm amplitude, 60 s on, 90 s off for 10 cycles, on ice). Ethanol extraction was carried out by mixing C2H5OH/H2O (100 ml, 1:1, v/v) to the cell suspension and refluxing at 68°C, for 12 h intervals, followed by centrifugation and recovery of the supernatant. This C2H5OH/ H2O extraction process was repeated five times and the combined supernatants dried. The dried supernatant was then treated with phenol/H2O (80%, w/w) at 70°C for 1 h followed by dialysis using a 1500 MWCO membrane (Spectrapore) against deionized water. The retentate was dried, re-suspended in water and treated sequentially digested with a-amylase, DNase, RNase chymotrypsin and trypsin, and the lipoglycan recovered following extensive dialysis using a 1500 MWCO membrane (Spectrapore) against deionized water (Nigou et al., 1999). The lipoglycans were monitored on 15% SDS-PAGE using either a silver stain utilizing periodic acid and silver nitrate (Hunter et al., 1986) or a Pro-Q emerald glycoprotein stain (Invitrogen).

Extraction and analysis of [ 14 C]PIMs from M. smegmatis strains
Mycobacterium smegmatis cultures (5 ml) were grown in TSB and metabolically labelled using 1 mCi ml -1 [1,2-14 C]acetate (50-62 mCi mmol -1 , GE Healthcare, Amersham Bioscience) at an OD600 of 0.4 and cultures grown for a further 4 h at 37°C with gentle shaking. Cells were harvested by centrifugation, washed once with PBS and a small-scale apolar and polar lipid extraction performed according to the methods of Dobson et al. (1985). The polar lipid extracts were re-suspended in CHCl3:CH3OH (2:1) and crude lipid (50 000 c.p.m.) applied to the corners of 6.6 ¥ 6.6 cm pieces of Merck 5554 aluminium-backed TLC plates. The plates were developed using CHCl3:CH3OH:H2O (60:30:6, v/v/v) in the first direction and CHCl3:CH3COOH:CH3OH:H2O (40:25:3:6, v/v/v/v) in the second direction to separate [ 14 C]labelled PIMs. Lipids were visualized by autoradiography by overnight exposure of Kodak X-Omat AR film to the TLC plates to reveal labelled lipids and compared with known standards.

Preparation of enzymatically active membranes and cell envelope fraction
Mycobacterium smegmatis and C. glutamicum strains used in this study were cultured to the mid-logarithmic growth phase in 1 l BHIS medium supplemented with kanamycin (25 mg ml -1 ) and IPTG (0.2 mM) where appropriate. Cells were harvested by centrifugation, re-suspended in 20 ml of buffer A (50 mM MOPS pH 7.9, 5 mM b-mercaptoethanol and 5 mM MgCl2) and lysed immediately by sonication (60 s on, 90 s off for a total of 10 cycles). The lysate was clarified by centrifugation at 27 000 g (4°C, 30 min) and membranes were deposited by centrifugation of the supernatant at 100 000 g (4°C, 90 min). The membranes were resuspended in buffer A to a final protein concentration of 20 mg ml -1 . The 27 000 g pellet was re-suspended in 10 ml of buffer A and 15 ml of Percoll (Pharmacia, Sweden), and centrifuged at 27 000 g for 60 min at 4°C. The particulate, upper diffuse band, containing both cell walls and membranes, was removed, collected by centrifugation, washed three times in buffer A, and finally re-suspended in 1 ml of buffer A. The final concentration of this Percoll-60 cell envelope fraction (P-60) was 20 mg ml -1 .
In vitro analysis of a(1,6) mannosyltransferase activity The neoglycolipid acceptors a-D-Manp-(1→6)-a-D-Manp-O-C8 (stored in C2H5OH) and C50-PP[ 14 C]M (stored in CHCl3/ CH3OH, 2:1, v/v), prepared as described , were separated into aliquots into 1.5 ml eppendorf tubes to a final concentration of 2 mM and 0.25 mCi (0.305 Ci mmol -1 ) respectively, and dried under nitrogen. IgePal CA-630 (8 ml, Sigma Aldrich) was added and the tubes sonicated to re-suspend the lipid-linked components, and the remaining assay components in a final volume of 80 ml were added, which included: 1 mM ATP, 1 mM NADP, and membrane protein (1 mg) from either C. glutamicum, C. glut-amicumDmptB, C. glutamicumDmptA, C. glutamicumDmptB pVWEx-Cg-mptB, C. glutamicumDmptBDmptA, C. glutamic-umDmptBDmptA pVWEx-Cg-mptB, C. glutamicumDmptBD-mptA pVWEx-Cg-mptA, C. glutamicumDmptBDmptA pVWEx-Mt-mptB and C. glutamicumDmptBDmptA pVWEx-Ms-mptB. Assays were incubated at 37°C for 1 h and then quenched by the addition of CHCl3/CH3OH (533 ml, 1:1, v/v). The reaction mixtures were then centrifuged at 27 000 g for 15 min at 4°C, the supernatant removed and dried under nitrogen. The residue was re-suspended in C2H5OH/H2O (700 ml, 1:1, v/v) and loaded onto a 1 ml SepPak strong anion exchange cartridge (Supleco) pre-equilibrated with C2H5OH/H2O (1:1, v/v). The column was washed with 2 ml of C2H5OH, and the eluate collected, dried and partitioned between the two phases arising from a mixture of n-butanol (3 ml) and water (3 ml). The resulting organic phase was recovered after centrifugation at 3500 g, and the aqueous phase again extracted twice with 3 ml of water-saturated butanol. The pooled extracts were back-washed twice with n-butanol-saturated water (3 ml). The n-butanol fraction was dried and re-suspended in 200 ml of n-butanol. The extracted radiolabelled material was quantified by liquid scintillation counting using 10% of the labelled material and 5 ml of EcoScintA (National Diagnostics, Atlanta, GA). The incorporation of [ 14 C]Manp was determined by subtracting counts present in control assays (incubations in the absence of acceptor), which were typically less than 100 c.p.m. per assay. The remaining labelled material was subjected to TLC using silica gel plates (5735 silca gel 60F254, Merck) developed in CHCl3:CH3OH:H2O:NH4OH (65:25:3.6:0.5, v/v/v/v) and the products visualized by phosphorimaging (Kodak K Screen).