The Lipoprotein LpqW Is Essential for the Mannosylation of Periplasmic Glycolipids in Corynebacteria*

Background: LpqW regulates synthesis of mycobacterial cell wall lipoglycans via unknown mechanisms. Results: A Corynebacterium glutamicum lpqW mutant has a global defect in lipoglycan synthesis, and LpqW is functionally linked to the mannosyltransferase MptB. Conclusion: LpqW activates lipoglycan synthesis pathways in C. glutamicum by directly regulating MptB. Significance: These results highlight the regulatory role of lipoproteins in glycolipid biosynthesis. Phosphatidylinositol mannosides (PIM), lipomannan (LM), and lipoarabinomannan (LAM) are essential components of the cell wall and plasma membrane of mycobacteria, including the human pathogen Mycobacterium tuberculosis, as well as the related Corynebacterineae. We have previously shown that the lipoprotein, LpqW, regulates PIM and LM/LAM biosynthesis in mycobacteria. Here, we provide direct evidence that LpqW regulates the activity of key mannosyltransferases in the periplasmic leaflet of the cell membrane. Inactivation of the Corynebacterium glutamicum lpqW ortholog, NCgl1054, resulted in a slow growth phenotype and a global defect in lipoglycan biosynthesis. The NCgl1054 mutant lacked LAMs and was defective in the elongation of the major PIM species, AcPIM2, as well as a second glycolipid, termed Gl-X (mannose-α1–4-glucuronic acid-α1-diacylglycerol), which function as membrane anchors for LM-A and LM-B, respectively. Elongation of AcPIM2 and Gl-X was found to be dependent on expression of polyprenol phosphomannose (ppMan) synthase. However, the ΔNCgl1054 mutant synthesized normal levels of ppMan, indicating that LpqW is not required for synthesis of this donor. A spontaneous suppressor strain was isolated in which lipoglycan synthesis in the ΔNCgl1054 mutant was partially restored. Genome-wide sequencing indicated that a single amino acid substitution within the ppMan-dependent mannosyltransferase MptB could bypass the need for LpqW. Further evidence of an interaction is provided by the observation that MptB activity in cell-free extracts was significantly reduced in the absence of LpqW. Collectively, our results suggest that LpqW may directly activate MptB, highlighting the role of lipoproteins in regulating key cell wall biosynthetic pathways in these bacteria.

Bacteria of the Corynebacterineae, a suborder of the Actinobacteria, include a number of important human pathogens, including Mycobacterium tuberculosis, Mycobacterium leprae, and Corynebacterium diphtheriae (1). These bacteria synthesize a distinctive multilaminate cell wall composed of peptidoglycan, complex polysaccharides, and both covalently linked and free glycolipids and lipoglycans (2). The structure and hydrophobic properties of the mycobacterial cell wall contribute to the intrinsic resistance of these bacteria to an array of host microbiocidal processes, many antibiotics, and sterilization conditions (3). Many of the cell wall components of pathogenic mycobacteria species are essential for pathogenesis and in vitro growth, hampering efforts to characterize the function of individual genes in their assembly (4 -7). In contrast, a number of non-pathogenic Corynebacterineae such as Corynebacterium glutamicum can tolerate the loss of major cell wall components, making them useful model systems for delineating processes involved in the assembly of core cell wall structures (8 -18).
All Corynebacterineae synthesize a family of glycolipids termed phosphatidyl-myo-inositol mannosides (PIMs) 3 that are important components of both the cell membrane and outer wall layers. Polar PIM species can also function as membrane anchors for the lipomannans (LM) and lipoarabinomannans (LAMs). These lipoglycans are essential for both the viability and in vivo survival of pathogenic mycobacterial species and have been shown to have potent immunomodulatory properties (19 -22). Many of the steps of PIM/LM/LAM biosynthesis have been elucidated (see Fig. 1, and recently reviewed in Refs. 7 and 15). In M. tuberculosis, PIM synthesis is initiated by the mannosylation of phosphatidylinositol by PimA (Rv2610c) to form PIM1 (23,24). Next, O-6-mannosylation of the myo-* This work was supported by the Australian Research Council Centre of inositol ring is performed by PimBЈ (Rv2188c) (16,24) resulting in the formation of PIM2. Both enzymes are cytoplasmic ␣-mannosyltransferases that require GDP-mannose (GDP-Man) as the sugar donor. PIM2 accumulates in the cell envelope mainly in its acylated forms AcPIM2 and Ac 2 PIM2, the former produced by the acyltransferase Rv2611c (25). Acylated PIM2 species can be further mannosylated to form more polar PIMs (Ac 1/2 PIM4-Ac 1/2 PIM6) and their hyperglycosylated forms (LM and LAM) (26). These reactions are performed by a number of glycosyltransferases that require a lipid sugar donor, C35/C50-polyprenylphosphomannose (ppMan) (26 -31). In mycobacteria, Ac 1/2 PIM4 is proposed to be a branch point leading to the synthesis of polar PIMs and LM/LAM, respectively. PimE (Rv1159) has been shown to elongate AcPIM4 with one or more ␣1-2-linked mannose residues to form AcPIM6 (31). Alternatively, a subpopulation of AcPIM4 is extended with chains of ␣1-6 linked mannose to form LM that is further modified with a number of single ␣1-2-mannose side chains (12,32,33). Deletion of the C. glutamicum gene encoding the ppMan-dependent mannosyltransferase, MptB, prevents initial AcPIM2 elongation, indicating that this mannosyltransferase can prime the synthesis of the LM ␣1-6-mannan backbone (17). Interestingly, loss of MptB in Mycobacterium smegmatis has little effect on LM synthesis, suggesting redundancy in this priming step in more distantly related mycobacteria (17). The short lipomannans generated by MptB are further elongated and side chain-modified by related ppMandependent mannosyltransferases MptA (Rv2174) (12,33) and MptC (Rv2181), respectively (7,32,35,36). Finally, LM is converted to mature LAM following the addition of arabinose units by EmbC (Rv3793), AftC (Rv2673), AftD (Rv0236c), and unidentified ␣1-5 arabinofuranosyltransferases (5,15,29,37).
In addition to the conserved phosphatidylinositol 3 PIM 3 LM 3 LAM pathway, some corynebacteria utilize a second pathway of lipoglycan biosynthesis in which a subpopulation of LM lipoglycans is assembled on a glucopyranosyluronic acid diacylglycerol (Gl-A, GlcADAG) glycolipid anchor (see Fig. 1) (16,18,38). In this pathway, Gl-A is first mannosylated by MgtA (NCgl0452 in C. glutamicum) forming mannosyl-glucuronic acid diacylglycerol (Gl-X, Man-GlcA-DAG), which is extended with further ␣1-6 (backbone) and ␣1-2 (side chain) linked mannose residues to produce a distinct population of LM molecules (38). In species of corynebacteria that have this second pathway, the phosphatidylinositol-anchored LM pool is termed LM-A, whereas the Gl-A-anchored LM is termed LM-B (39). The extension of Gl-X and PIM species involves common ppMan-dependent mannosyltransferases. In particular, deletion of the C. glutamicum gene encoding MptB (NCgl1505) results in the accumulation of AcPIM2 and Gl-X, and a block in both LM-A and LM-B biosynthesis (17).
LpqW is a putative lipoprotein that is highly conserved in members of Corynebacterineae. The M. tuberculosis lpqW gene (Rv1166) is essential for viability (40), whereas loss of the M. smegmatis ortholog (MSMEG_5130) results in a defect in LM/LAM biosynthesis (41)(42)(43). An M. smegmatis lpqW transposon mutant had a reduced capacity to synthesize LM/LAM but had a normal spectrum of polar PIMs (41). This mutant was unstable and readily accumulated secondary mutations in pimE, which resulted in a block in synthesis of polar PIMs and restored synthesis of LM and LAM (43). The crystal structure of M. smegmatis LpqW contains a putative AcPIM4 binding pocket, raising the possibility that LpqW may function as a glycolipid chaperone, regulating access of AcPIM4 to either PimE and/or the elongating ppMan-dependent mannosyltransferases (42,44).
To further investigate the function of LpqW here, we have deleted the lpqW ortholog, NCgl1054, in C. glutamicum. C. glutamicum is an excellent experimental system for investigating the potential regulatory function of LpqW in LM/LAM synthesis as (i) it is more accepting of loss of cell wall components than Mycobacterium spp., (ii) it lacks polar PIMs and a pimE ortholog (43), simplifying analysis of its role in LM versus PIM biosynthesis, and (iii) this species synthesizes both LM-A and LM-B, allowing dissection of the role of LpqW in regulating different classes of glycolipid anchor into the common LM pathway (Fig. 1). We show here that deletion of NCgl1054 results in a global defect in the elongation of both AcPIM2 and Gl-X. Although similar to the biochemical defect induced by disruption of ppMan synthesis, we show that LpqW is not required for synthesis of this donor. Significantly, loss of LpqW can be bypassed by substitution of a single amino acid residue in the ppMan-dependent mannosyltransferase MptB, leading to reactivation of lipoglycan pathways in the NCgl1054 mutant. Loss of LpqW was also associated with loss of MptB activity in a cell-free cell envelope assay. Collectively, our findings strongly suggest that LpqW directly regulates the activity of MptB thereby controlling the elongation of both the AcPIM2 and Gl-X membrane anchors.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Culture Conditions, Transformation, and Genetic Manipulation-Escherichia coli DH5␣ strain was grown in Luria-Bertani (LB) medium at 37°C with aeration. Early PIM intermediates are assembled on the cytoplasmic leaflet of the cell membrane by GDP-Man-dependent mannosyltransferases before being transported ("flipped") to the periplasmic leaflet where they are further elongated by a family of ppMan-dependent mannosyltransferases, that includes MptB, to form LM-A. LM-A is further modified by a second family of arabinosyltransferases to form LAM. The ppMan donor is synthesized by the cytoplasmically orientated enzyme, Ppm1, and must also be transported to the periplasmic leaflet. An alternative family of lipomannans, termed LM-B, are assembled on a structurally distinct glycolipid anchor that contains the core structure Gl-A. Gl-A is initially extended with a single mannose residue by the cytoplasmically orientated enzyme MgtA and subsequently elongated by MptB and other ppMan-dependent mannosyltransferases to form LM-B. PI, phosphatidylinositol.
C. glutamicum ATCC 13032 was grown in brain heart infusion (BHI) medium (Oxoid) or LBHIS (LB, brain heart infusion, sorbitol) (45) at 30°C with aeration. When necessary, ampicillin was added to a final concentration of 100 g ml Ϫ1 and kanamycin at 50 g ml Ϫ1 . E. coli plasmid DNA was isolated from 10 ml of an overnight culture using the High Pure plasmid isolation kit (Roche) and C. glutamicum genomic DNA was extracted from ϳ0.5 g of cells using the Illustra DNA extraction kit (GE Healthcare), according to the manufacturer's instructions. When necessary, DNA was purified using an UltraClean 15 DNA purification kit (Mo Bio). The concentration and purity of DNA was assessed using a NanoDrop ND-1000 spectrophotometer (Nanodrop Technologies). PCR reactions were performed in a PTC-200 thermal cycler (MJ Research) using Taq polymerase (Roche) or ProofStart DNA polymerase (Qiagen). Initial denaturation of template DNA was done at 95°C for 5 min, followed by 35 cycles of denaturation at 94°C for 1 min, a 1-min primer-specific annealing step, and a 1-min per kb extension step at 72°C. The program included a 10-min final extension step at 72°C. PCR products were purified by extraction from 1% agarose gels using an UltraClean 15 DNA purification kit (MoBio). Endonucleases, T4 ligase, polynucleotide kinase system, and alkaline phosphatase were obtained from New England Biolabs and used according to the manufacturer's instructions.
Bioinformatic Identification and Analysis of Corynebacterial LpqW-The corynebacterial ortholog of mycobacterial LpqW was found using the tBLASTn (46) algorithm on the NCBI website using the protein sequences of Rv1166 and MSMEG_5130. The Genome Region Comparison tool located on JCVI website was used to visualize synteny in the region and to identify orthologs in related species.
Construction of C. glutamicum ⌬NCgl1054 and Complementation Studies-The ortholog of lpqW was deleted using a twostep recombination strategy previously used successfully in our laboratory (11,16). A 2.4-kb fragment containing the entire NCgl1054 gene was amplified using ProofStart DNA polymerase (Qiagen) and the primers NCgl1054For (5Ј-GCTC-TAGACGTATTCCTGCTCGTGGCCTG) and NCgl1054Rev (5Ј-CCCAAGCTTATGCGTTGCTCGCCGGCTGC) and cloned into the XbaI/HindIII sites (underlined) of pUC19 (47). A 1.1-kb EcoRI fragment internal to the gene was excised, the remaining plasmid was religated, and the deletion cassette was subcloned into XbaI/HindIII-digested pK18mobsacB, a suicide plasmid in C. glutamicum (48), which contains kanamycin and sucrose selection markers. The resultant plasmid, pK18mobsacB:⌬NCgl1054, was sequenced then electroporated into electrocompetent C. glutamicum cells, prepared as described previously (45), using an ECM 630 electroporator (BTX). Clones resulting from single homologous recombination events were selected on kanamycin. These were grown overnight without antibiotic selection and then serially diluted and plated onto LBHIS plates containing 10% sucrose to select for a second crossover event. Small, sucrose-resistant and kanamycin-sensitive colonies were screened by PCR using NCgl1054For and NCgl1054Rev primers. Southern blot hybridization (see below) was used to confirm the initial single crossover and NCgl1054 deletion strains.
To complement the ⌬NCgl1054 strain, the entire NCgl1054 gene together with 113 bp of upstream sequence was PCRamplified using primers NCgl1054CompF (5Ј-ATATGGAT-CCAAGGAGATATAGATTTGGGGGTGAGAATAAGGTT) and NCgl1054CompR (5Ј-ATATGAGCTCAGATCATTCTT-CAACATCGT) and cloned into BamHI/SacI sites (underlined) of pUC19, followed by subcloning into the unique PvuII site of pSM22 (49), which contains the corynebacterial origin of replication repA and kanamycin resistance gene aphA3. A sequenced complementation plasmid (pSM22:NCgl1054) and pSM22 control plasmid were then electroporated into the C. glutamicum ⌬NCgl1054 deletion strain, followed by selection on kanamycin-supplemented BHI plates.
Southern Hybridization-For Southern blot analysis, 2 g of genomic DNA was digested with selected restriction enzymes under optimal conditions for 16 h. To ensure complete digestion, additional units of endonuclease were added, and incubation was continued for another 3 h. Purified samples, and digoxygenin-labeled, HindIII-digested DNA markers, were separated on a 1% agarose gel followed by depurination, denaturation, neutralization, and capillary transfer onto a nylon membrane. The membrane was then hybridized at 65°C with a gene-specific probe prepared by digoxygenin labeling a PCR product obtained using primers NCgl1054CompF/ NCgl1054CompR for the ⌬NCgl1054 strain.
Compositional Cell Wall Analysis-Wild-type (WT) and mutant cells were harvested by centrifugation at logarithmic growth phase (A 600 of between 1 and 3) from 100 ml of culture in BHI. Pellets were freeze-dried, weighed, and stored at Ϫ20°C until needed. Cells were delipidated by agitation in 6 ml of chloroform:methanol (2:1, v/v) for 2 h at room temperature twice followed by extraction in chloroform:methanol:water (1:2:0.8, v/v). Supernatants were collected, dried under a N 2 stream, and 1-butanol:water partitioned as follows. The dried lipid fraction was resuspended in 400 l of water-saturated 1-butanol and 200 l of 1-butanol-saturated water followed by vortexing and brief centrifugation. The top (organic) phase was collected, and the process was repeated. The final extract was dried under vacuum and resuspended in 5 l/0.01 g (dry weight) of watersaturated 1-butanol.
The delipidated cell pellet was then subjected to ethanol reflux by suspending in 5 ml of 50% (v/v) ethanol and incubating at 100°C for 2 h, with occasional vortex mixing. Samples were centrifuged at 300 ϫ g for 5 min, the supernatant was collected, and the procedure was repeated two more times. Pellets were used for corynomycolic acid extraction. The supernatant was pooled, and ethanol was blown off under a stream of N 2 followed by freeze-drying. The resultant samples were resuspended in 400 l of water containing 0.02% (w/v) CaCl 2 and 10 units of proteinase K followed by incubation at 37°C for 2 h. The digest was diluted with 50% propan-1-ol (50 l) and 1 M ammonium acetate (50 l) and loaded onto a column of octyl-Sepharose (1 ml) equilibrated in 5% 1-propanol and 50 mM ammonium acetate. After washing the column, lipoglycans were eluted with 30, 40, 50, and 60% 1-propanol (1-ml volumes) and carbohydrate-containing fractions identified by spotting 5-l aliquots on high-performance thin layer chromatography (HPTLC) sheets and stained with orcinol-H 2 SO 4 (50). Lipoglycan-containing fractions (in 30% to 40% propan-1-ol fractions) were pooled, dried under vacuum, and resuspended in 30% (v/v) propan-1-ol for further analysis.
Polyacrylamide Gel Electrophoresis of LM/LAM-Purified lipoglycan fractions were mixed with PAGE sample buffer and incubated at 100°C for 5 min. Samples were then separated on a 15% polyacrylamide gel followed by fixing for 45 min in 100 ml of 40% (v/v) methanol and 10% (v/v) acetic acid, incubation for 10 min in 100 ml of 40% (v/v) methanol, 10% (v/v) acetic acid and 0.7% (v/v) periodic acid, followed by a 10-min wash in 100 ml of 5% (v/v) methanol and 7% (v/v) acetic acid, and 5 min in 2.5% (v/v) glutaraldehyde. The gel was then rinsed in pure water four times and incubated for 10 min with DTT solution followed by staining with the SilverSnap Kit (Pierce). The staining reaction was stopped by washing the gel in 5% (v/v) acetic acid solution. Sugar content was also quantified by GC-MS after methanolysis and trimethylsilyl derivatization (50).
Mass Spectrometric Analysis-HPTLC-purified lipids were resuspended in 1-butanol and diluted in 10 mM ammonium formate in methanol (1:10, v/v). Samples were flow injected (0.2 ml/min) with the solvent system H 2 O/MeOH/THF (7:10:33, v/v/v) into the electrospray source of an Agilent 6460 triplequadrupole LC-MS (Agilent Technologies). A capillary voltage of 4000 V, gas temperature of 300°C, and flow rate of 7 liters/min were used. The sheath gas temperature and gas flow was set to 220°C and 7 liters/min, respectively. The nebulizer was set at 15 psi. Total ion scan was performed in positive mode at fragmentator voltage 120 V. For MS/MS experiments, Gl-X [M ϩ NH 4 ] ϩ m/z ϭ 950.1, m/z ϭ 1112.9, and m/z ϭ 1274.9 are frag-mented with a fragmentator voltage of 135 V or 220 V and collision energy of 55 eV.
Metabolic Radiolabeling of Whole Cells-Mid-log phase cultures were harvested by centrifugation, gently resuspended in prewarmed PBS (800 l), and incubated at 30°C for 30 min. Cells were pulse labeled by adding [2-3 H]-mannose (50 Ci/ml; PerkinElmer Life Sciences). After 5 min (30°C), an aliquot of sample (200 l) was removed, centrifuged at 14,000 ϫ g, and the supernatant was frozen in liquid nitrogen. The remaining sample was centrifuged (14,000 ϫ g, 30 s), the supernatant was discarded, and cell pellets were gently resuspended in 1500 l of prewarmed BHI. Sample aliquots (500 l) were removed and frozen after 1-, 5-, and 30-min incubation in unlabeled medium (chase). Lipids were extracted from the frozen pellets and analyzed by HPTLC as described above. Labeled species were detected either with a TLC linear analyzer (Berthold) or after coating HPTLC sheets with EN 3 HANCE Spray (PerkinElmer Life Sciences) and exposure to Kodak BioMax MR film (Sigma-Aldrich) at Ϫ80°C.
Mannosidase Treatment-Samples were dried under vacuum, resuspended in 2 l of 1% taurodeoxycholate, and incubated with 20 l of jack-bean ␣-mannosidase (Sigma) in 0.1 M sodium acetate, pH 5.0, at 37°C. An additional 20 l of mannosidase was added after 24 h, and samples were incubated for another 24 h. Lipids were extracted with 80 l of chloroform: methanol (2:1, v/v) and recovered by butanol partitioning.
Cell Envelope Cell-free Assay-Wild-type and mutant strains were grown in BHI medium at 30°C to mid-log phase (A 600 ϭ 4 -6), and cell pellets were washed twice in 25 mM HEPES-NaOH (pH 7.4) prior to being suspended at 0.15 g wet pellet/ml in lysis buffer containing 50 mM HEPES-NaOH (pH 7.4), 2 mM EGTA, and protease inhibitor mixture (Roche Diagnostics). The cell suspension was sonicated on ice with a tapered microtip 1 ⁄ 4 using an Ultrasibuc processor (750 W, Cole-Palmer; 1-s pulse, 1-s pause for a total of 10 min at 40% amplitude). The resultant lysate was centrifuged twice at 1000 ϫ g, (4°C, 10 min) to remove cellular debris. The supernatant was ultracentrifuged at 100,000 ϫ g at 4°C for 1 h using a Beckman coulter TLA 120.2 rotor. The resultant pellet, containing both cell wall and cell membrane, was resuspended in a one-tenth volume of the lysis buffer and recovered as the cell envelope fraction (26). Protein concentration was measured using Bradford reagent (Bio-Rad).
The cell envelope fraction (0.5 mg protein) was supplemented with 5 mM MgCl 2 and 5 mM ␤-mercaptoethanol and preincubated for 5 min at 30°C prior to addition of GDP-[ 3 H]mannose (230,000 cpm; ϳ2 Ci) and further incubated for 180 min. The reaction was stopped by adding a 20 times volume of chloroform/methanol 2:1 (v/v). Total lipid was extracted as described above.
Genome Sequencing and Bioinformatics Analyses-Whole genome sequencing of C. glutamicum isolates was achieved using Ion Torrent semi-conductor sequencing on the PGM platform, with a single 316 chip with 100 bp chemistry for each isolate, following the manufacturer's instructions (Invitrogen). Resulting sequence reads were mapped to a C. glutamicum reference genome (accession no. NC_003450.3) in the NCBI collection, using an in-house Python utility called Nesoni, which uses SHRiMP2 (51). A global variant analysis was then performed and allelic variability at any nucleotide position was tallied to generate a list of differences for each genome compared with the reference. Sequence reads have been deposited in the NCBI Sequence Read Archive under accession no. SRA057127.

RESULTS
Identification of Corynebacterial lpqW-An ortholog of the mycobacterial lpqW gene was identified in C. glutamicum ATCC 13032 using a tblastn search of LpqW protein sequences from M. tuberculosis H37Rv (Rv1166) and M. smegmatis mc 2 155 (MSMEG_5130). The best match was NCgl1054, which encodes a protein 502-residues-long, sharing 21.2% identity and 32.4% similarity with mycobacterial LpqW. Syntenic analysis revealed that corynebacterial lpqW is found in the same genetic context as in other analyzed mycobacterial genomes (Fig. 2), further supporting NCgl1054 as the most likely ortholog of lpqW in C. glutamicum.
Inactivation of the NCgl1054 Gene-A two-step recombination strategy was used to inactivate NCgl1054. Deletion of a 1.1-kb internal fragment was achieved using the suicide vector pK18mobsacB (49), carrying a kanamycin resistance gene (aph) and Bacillus subtilis sacB gene conferring sensitivity to sucrose (Fig. 3A). Successful deletion of NCgl1054 was detected by PCR (data not shown) and then confirmed by Southern blotting analyses (Fig. 3B).
Growth Characteristics and Complementation of C. glutamicum ⌬NCgl1054-The mutant was observed to form relatively small colonies (ϳ1 mm) on BHI plates after 24 h incubation at 30°C. A slow growth phenotype was also observed in liquid BHI (Fig. 4). The NCgl1054 mutant grows at about half the rate of the WT parent (doubling time of ϳ100 min) under these culture conditions. To complement the mutation, ⌬NCgl1054 was transformed with pSM22:NCgl1054, a plasmid carrying an intact NCgl1054 gene with 113 bp of the upstream sequence that was expected to include the native promoter. Unmodified pSM22 was also introduced into ⌬NCgl1054 as a control. Introduction of pSM22:NCgl1054 but not pSM22 restored colony size and growth rate to nearly WT levels (Fig. 4), demonstrating that the growth defect was due to deletion of the NCgl1054 gene.
Cell Wall Analyses of C. glutamicum ⌬NCgl1054 and Complementation Strains-Cell wall components of mutant and complemented strains were analyzed and compared with the parental WT strain C. glutamicum ATCC 13032. No changes were found in the levels or composition of mycolic acids or cell wall arabinogalactan (data not shown). However, the mutant was shown to have an altered glycolipid profile when total lipid extracts were analyzed by HPTLC (Fig. 5A). First, the mutant strain accumulated higher levels of AcPIM2 and Gl-X. These species run as a doublet in the solvent system used in Fig. 5A but could be resolved in an alternative solvent system, revealing accumulation of both species relative to trehalose corynomycolates that serve as internal controls (supplemental Fig. 1). These species co-migrated with authentic standards, and their identities were confirmed by LC-MS/MS analysis. Specifically, the HPTLC band assigned as AcPIM-2/Gl-X contained a glycolipid species with a [M ϩ NH 4 ] ϩ ion of m/z 950.1, corresponding to Gl-X (Man-GlcA-DAG). MS/MS of this ion produced a fragment ion at m/z 577.4 corresponding to a diacylglycerol (C34:1 DAG) species after neutral loss of a Hex-HexA disaccharide. Secondly, WT bacteria expressed two additional glycolipid species, termed glycolipid A and B, which were not detected in ⌬NCgl1054. LC-MS analysis of HPTLC fractions enriched for glycolipids A and B gave [M ϩ NH 4 ] ϩ ions of m/z 1112.9 and 1274.9, which correspond to ammonium adducted molecular ions for Hex 2 HexA-DAG and Hex 3 HexA-DAG, respectively (Fig. 5B). Both species contained the same DAG moiety (34:1) as Gl-X, because MS/MS of these ions gave fragment ions at 577.6/577.3 (Fig. 5B). These species, designated Gl-Y and Gl-Z, were metabolically labeled with [ 14 C]acetate and were sensitive to Jack bean ␣-mannosidase digestion (Fig. 5C), suggesting that they correspond to Man 2 GlcA-DAG and Man 3 GlcA-DAG, respectively. The absence of Gl-Y and Gl-Z species in the mutant indicated that NCgl1054 is required for the elongation of Gl-X with mannose residues.
PIM species have been shown to function as membrane anchors for LM-A and LAM lipoglycans, whereas mannosylated forms of Gl-X function as membrane anchors for LM-B (18). To further investigate the consequences of LpqW disruption on lipoglycan biosynthesis, total lipoglycans were purified from WT, ⌬NCgl1054, and the complementation strains and analyzed by PAGE and GC-MS (Fig. 6). As expected, WT bacteria produced two lipoglycan populations corresponding to LM (LM-A and LM-B) and LAM (Fig. 6, A and B). In contrast, the ⌬NCgl1054 strain and ⌬NCgl1054 bearing a control plasmid lacked both lipoglycan classes. LM/LAM biosynthesis was completely restored in the ⌬NCgl1054 strain carrying a func- tional copy of the gene. These results suggest that loss of LpqW results in a global lipoglycan defect arising from loss of synthesis of mannosylated Gl-X and polar PIM glycolipids that are required for cell wall lipoglycan biosynthesis.

Synthesis of Polar LM-B Precursors Is Dependent on ppMan
Synthase-The synthesis of Gl-A and Gl-X is thought to occur on the cytoplasmic face of the cell membrane (Fig. 1). In particular, the conversion of Gl-A to Gl-X is catalyzed by the GDP-Man-dependent mannosyltransferase, MgtA (Fig. 1). Whether Gl-X is further elongated to Gl-Y or Gl-Z by MgtA or other GDP-Man-dependent mannosyltransferase or by ppMan-dependent mannosyltransferases, such as those involved in LM biosynthesis (Fig. 1) has not been defined (12,17,31,32). To investigate whether loss of ppMan synthase also leads to disruption of mannosylated Gl-X biosynthesis, we generated a C. glutamicum mutant lacking Ppm1 (27, 52) by deleting a 748-bp internal fragment of the ppm1 gene using the same twostep recombination strategy that we used to create the ⌬NCgl1054 strain. The integration event and disruption of the gene was detected by PCR analysis using ppm1-specific primers (data not shown) and then confirmed by Southern hybridization (Fig. 7). The resultant strain, termed ⌬ppm1, formed small colonies on agar plates and had a reduced growth rate comparable with the ⌬NCgl1054 mutant (Fig. 4), suggesting that those strains had similar fitness.
WT, ⌬NCgl1054, and ⌬ppm1 strains were pulse labeled with 3 H-mannose for 5 min and label-chased over 30 min. Samples were taken with rapid quenching, and purified lipids were analyzed by HPTLC (Fig. 8). [ 3 H]Mannose was rapidly incorporated into ppMan and a spectrum of PIM species (PIM1, PIM2, AcPIM2) in wild type bacteria. Subsequently, label was chased out of ppMan and into mannosylated Gl-X species. Although label in PIM1 and AcPIM1 was also rapidly chased, label in AcPIM2 was only partially chased (Fig. 8), consistent with this species being a metabolic end-product as well as a precursor for LM-A/LAM synthesis. In marked contrast, label was only incorporated into PIM1, AcPIM1 and AcPIM2 in the ⌬ppm1 mutant (Fig. 8, middle panel). The mannosylated Gl-X species, Gl-Y and Gl-Z were not detected, and only label in PIM1 species was chased. The same PIM species were labeled in the showing the arrangement of genes in the NCgl1054 region of wild-type C. glutamicum. Below, the fragment amplified for cloning into pK18mobsacB Xba/ HindII sites is shown with the EcoRI sites used to delete the 1.1-kb fragment internal to NCgl1054. Above, SalI digestion sites, sizes of expected bands on Southern blot, and position of the probe used are shown. NCgl1054 and surrounding sequences were amplified by PCR and cloned into pUC18, then a 1.1-kb EcoRI internal fragment of the gene was removed, and the deletion cassette was subcloned into a suicide plasmid pK18mobsacB carrying the sacB gene. Kanamycin-sensitive and sucrose-resistant clones were tested for the second crossover event by PCR and Southern blot. B, Southern blot analysis of SalI-digested DNA of C. glutamicum WT, a single crossover strain and the NCgl1054 mutant. SalI endonuclease has a restriction site within the deleted section of the gene; expected band sizes are shown in A. Lanes: Markers, digoxygenin-labeled DNA standards digested with HindIII; WT, wildtype C. glutamicum ATCC 13032; sco, single cross-over strain of C. glutamicum carrying a copy of pK18mobsacB:NCgl1054 integrated into the NCgl1054 locus; ⌬NCgl1054, C. glutamicum NCgl1054 deletion mutant. WT and NCgl1054 mutant strains were inoculated in BHI and bacterial growth at 30°C monitored using A 600 over 24 h. Results from three independent cultures were plotted, and error bars represent S.E. Growth rate and colony size were fully restored in the cells carrying pSM22:NCgl1054, a shuttle plasmid containing the NCgl1054 gene and upstream sequences expected to contain its native promoter. DECEMBER 14, 2012 • VOLUME 287 • NUMBER 51

JOURNAL OF BIOLOGICAL CHEMISTRY 42731
⌬NCgl1054 strain, indicating a defect in the synthesis of ppMan-dependent PIM and Gl-X species. Significantly, the ⌬NCgl1054 mutant synthesized similar levels of ppMan to wild type bacteria, although the rate of chase of label out of ppMan, as well as PIM1 and AcPIM1, was dramatically slowed in the ⌬NCgl1054 strain (Fig. 8, far right panel). Collectively, these results provide further evidence that LpqW is required for the elongation of both Gl-X and AcPIM2 species. Although these reactions are dependent on the presence of ppMan, LpqW is not required for the synthesis of this mannose donor, suggesting that it regulates Gl-X/AcPIM2 elongation via an independent mechanism.
A Mechanism for Bypassing lpqW in C. glutamicum-We have previously shown that secondary mutations in the  Labels on x axis are as follows: WT, C. glutamicum wild-type; ⌬NCgl1054, mutant lacking 1.1-kb internal fragment of NCgl1054; ⌬NCgl1054ϩpSM22, mutant complemented with empty pSM22 vector; ⌬NCgl1054ϩpSM22:NCgl1054, mutant complemented with a functional copy of NCgl1054; ⌬ppm1, ppm1 deletion mutant. B, PAGE analysis of LM/LAM fraction extracted from various strains. Polar glycolipids were extracted from delipidated cells by ethanol reflux and purified on octyl-Sepharose column, followed by proteinase K treatment. Samples were separated on 12% PAGE and stained with SilverSNAP kit. Strains are labeled as in A. Std., protein molecular weight standard (Invitrogen). NCgl1054 mutant fails to synthesize detectable levels of lipoglycans, whereas complementation with a functional copy on the gene restored lipoglycan synthesis to near wild-type levels. FIGURE 7. Construction of a ⌬ppm1 mutant of C. glutamicum. A, NCgl1423 (ppm1), encoding one component of the polyprenol-mannose synthase, was disrupted by homologous recombination between the locus and pK18mobsacB:⌬ppm1, which carries ϳ1-kb sequences (gray blocks) flanking a 748-bp fragment internal to ppm1. B, Southern hybdridization analysis of the ⌬ppm1 mutant. Genomic DNA of wild-type C. glutamicum, a single crossover strain (sco), and the ⌬ppm1 mutant was digested with BamHI and subjected to Southern blotting using a ppm1-specific probe (dashed line). Expected bands were 4.8 kb for the wild-type and 4.0 kb for the mutant. Markers, digoxygenin-labeled DNA standards digested with HindIII; WT, wild-type C. glutamicum ATCC 13032; ⌬ppm1, C. glutamicum ppm1 deletion mutant; sco, single crossover strain of C. glutamicum carrying a copy of plasmid pK18mobsacB:⌬ppm1 integrated at the ppm1 locus. DECEMBER 14, 2012 • VOLUME 287 • NUMBER 51

JOURNAL OF BIOLOGICAL CHEMISTRY 42733
M. smegmatis pimE gene can restore LM/LAM synthesis following disruption of lpqW in this species (43). Although C. glutamicum lacks a PimE ortholog, we were interested to determine whether spontaneous suppressor mutants could arise in the ⌬NCgl1054 strain leading to restored LM synthesis and growth rate. To promote the out-growth of suppressor strains, ⌬NCgl1054 bacteria were diluted 1:10 in 10-ml BHI broths at 30°C with shaking for 24 or 48 h for a total of 10 passages. When aliquots from each passage were plated on BHI agar, putative suppressor strains (based on wild type colony size) were identified after passages 4 -6. Analysis of one of these colonies, collected after four passages, revealed partial restoration of lipoglycan synthesis (Fig. 9A). Specifically, this suppressor/bypass strain expressed appreciable level of LM and LAM as well as wild type levels of polar glycolipid species (Fig. 9B). Whole genome sequencing and comparison to the parental ⌬NCgl1054 strain identified a single base pair change (A to G) within the mptB gene (NCgl1505), encoding the ppMan-dependent mannosyltransferase, MptB. This mutation is predicted to produce a tyrosine to cysteine substitution at residue 507 of MptB that is located in the middle of the second last predicted transmembrane domain of the polytopic membrane protein.
These data suggested that LpqW may either regulate the activity of MptB or the access of glycolipid precursors to MptB and that the suppressor mutation removes this requirement. To distinguish between these possibilities, cell-free assays were performed using cell envelope extracts from WT cells, the ⌬NCgl1054 mutant and the bypass mutant. Following incubation with GDP-[ 3 H]mannose, MptB activity was inferred by the appearance of labeled Gl-Y/Gl-Z. If LpqW is involved in regulating substrate accessibility to MptB, we would predict that loss of LpqW would have little effect on MptB activity in the absence of compartmentalization. If, on the other hand, it is directly involved in regulating MptB activity, loss of LpqW should result in loss of activity in the mutant and restoration in the bypass strain. As shown in Fig. 9C, Gl-Y/Gl-Z biosynthesis was detected in cell envelope preparations of WT and the bypass mutant but was absent in the ⌬NCgl1054 mutant preparation. These findings support a role for LpqW in the activation of MptB and indicate that the Y to C substitution in the bypass mutant renders MptB partially independent of LpqW activation.

DISCUSSION
We have previously shown that the lipoprotein LpqW has a role in the biosynthesis of hyperglycosylated PIM species such as LM and LAM in mycobacteria. Specifically, disruption of lpqW in M. smegmatis leads to reduced LM/LAM synthesis without affecting the synthesis of PIMs (41). This mutant is rapidly overgrown by suppressor strains in which LM/LAM synthesis is restored at the expense of polar PIM biosynthesis due to mutations in the ppMan-dependent mannosyltransferase, PimE (43). Based on these findings, we hypothesized that LpqW functions to regulate either polar PIM and/or LM/LAM biosynthesis. In this study, we extend these observations to show that LpqW has a direct role in regulating the elongation of diverse glycolipid anchors by ppMan-dependent mannosyltransferases.
We investigated the function of the lpqW ortholog in C. glutamicum because of the following: 1) the organism is more permissive to loss of cell wall components than Mycobacteria spp, increasing the likelihood that the knock-out would have a more stable phenotype, 2) the organism has a relatively simple PIM profile, composed primarily of PIM2/AcPIM2 and lacks a recognizable PimE ortholog, and 3) the organism accumulates a second family of glycolipids that are structurally distinct from the PIMs, allowing analysis of the selectivity of LpqW function. The C. glutamicum lpqW ortholog exhibits significant sequence similarity (ϳ32%) to the M. tuberculosis and M. smegmatis lpqW genes and is located in the same genetic context. Interestingly, five of the six residues proposed to constitute a binding pocket in the M. smegmatis LpqW (42) are conserved in the C. glutamicum protein, despite its reduced size and relatively low sequence similarity with a mycobacterial ortholog. In the ppm1 mutant, labeling is only incorporated into early PIM and Gl-X species. Gl-X species modified with mannose were not detected. In the NCgl1054 mutant, ppMan was strongly labeled during the pulse, whereas labeling was slowly chased from this species. No labeling was detected in the polar Gl-X species.
Disruption of NCgl1054 resulted in a slow growth phenotype that was complemented by episomal expression of NCgl1054. A similar slow growth phenotype was observed following transposon-disruption of M. smegmatis lpqW (41), supporting the notion that this protein has similar functions within the Corynebacterium genus. Biochemical analysis of the mutant indicated that loss of NCgl1054 was associated with a block in the elongation of the major PIM species, AcPIM2, to form LM-A (the precursor to LAM), as well as the elongation of Gl-X to form LM-B. Specifically, HPTLC analysis and LC-MS/MS revealed that the mutant lacked two mannosylated Gl-X species that corresponded to Gl-Y (Man 2 GlcA.DAG) and Gl-Z (Man 3 GlcA.DAG). These analyses suggest that LpqW is involved in regulating a process or enzyme that is common to the synthesis of structurally distinct PIM and Gl-glycolipids, rather than just regulating fluxes in the PIM pathway.
The biochemical phenotype generated by disruption of NCgl1054 was very similar to that generated by deletion of the C. glutamicum gene encoding ppMan synthase, ppm1. Specifically both mutants exhibited a defect in the elongation of AcPIM2 and Gl-X to more polar glycolipid species and LM/LAM. These analyses suggest that Gl-X, as well as AcPIM2, are initially assembled on the cytoplasmic leaflet of the cell membrane and subsequently flipped to the periplasmic side of the membrane where they are elongated by ppMan-dependent mannosyltransferases. ppMan is synthesized on the cytoplasmic leaflet of the cell membrane (Fig. 1), and LpqW could, in principal, be required for the synthesis of this mannose donor FIGURE 9. Analysis of a suppressor mutant of C. glutamicum ⌬NCgl1054 with the capacity to produce LM-A/LAM. A, PAGE analysis of LM/LAM fraction extracted from various strains. Lipoglycans were purified on an octyl-Sepharose column, proteinase K-treated, separated by 12% PAGE, and detected using a SilverSNAP kit. WT, C. glutamicum wild-type; ⌬NCgl1054, lpqW mutant; bypass mutant, a ⌬NCgl1054 derivative with partially restored lipoglycan synthesis; Std., protein molecular weight standard (Invitrogen). B, HPTLC analysis of C. glutamicum total lipid extracts. Glycolipids were visualized using orcinol/H 2 SO 4 . Arrow indicates a modified PIM species that is found exclusively in the bypass mutant. C, cell-free assays using cell envelope fractions from WT C. glutamicum, ⌬NCgl1054, and the bypass mutant. Cell envelopes were incubated with GDP-[ 3 H]mannose for 3 h and labeled lipids analyzed by HPTLC and visualized by fluorography. The mobility of [ 14 C]acetate-labeled C. glutamicum lipids is provided as standard. D, a model for the mode of action of LpqW in C. glutamicum. By analogy with the E. coli Lpo proteins, LpqW may regulate the activity of MptB that initiates the elongation of AcPIM2 and Gl-X with ␣1-6-linked mannose chains leading to the formation of LM-A/LAM and LM-B, respectively. Disruption of NCgl1054 leads to inactivation of MptB and a global defect in LM/LAM biosynthesis. In the bypass mutant derived from the ⌬NCgl1054 strain, a Y to C mutation at residue 507 of MptB causes a conformational change that partially activates MptB in the absence of LpqW. This strain can now synthesize LM/LAM in the absence of functional LpqW activation. and/or its transport to the periplasmic leaflet of the cell membrane. However, we show here that ppMan synthesis is normal in the NCgl1054 mutant. This is consistent with our previous analysis of the M. smegmatis ⌬lpqW mutant and derived suppressor strains that synthesize normal levels of polar PIMs or LM/LAM, respectively, indicating ongoing synthesis of ppMan in both the initial and derived suppressor strains. Collectively, these results suggest that LpqW does not play a role in regulating either the synthesis or periplasmic transport of ppMan.
M. smegmatis mutants lacking LpqW initially exhibit a defect in LM/LAM biosynthesis but are rapidly overgrown by suppressor/bypass mutants when cultivated in minimal medium (41). These strains have restored LM/LAM synthesis at the expense of polar PIM biosynthesis. Genetic analysis of these suppressor mutants demonstrated that the restoration of LM/LAM synthesis was dependent on mutation of PimE, a ppMan-dependent mannosyltransferase that is involved in converting AcPIM4 to AcPIM5 and AcPIM6. C. glutamicum lacks an ortholog of PimE and does not accumulate polar PIM species. Nonetheless, suppressor strains were readily generated after 4 -6 passages when the C. glutamicum ⌬NCgl1054 mutant was incubated in BHI broth. In common with the M. smegmatis LpqW suppressor strains, the ⌬NCgl1054 suppressor strains exhibited restored LM/LAM biosynthesis. Intriguingly, whole genome sequencing of one of these suppressor strains revealed a single base mutation corresponding to one of the transmembrane domains of the ppMan-dependent mannosyltransferase, MptB. The C. glutamicum MptB has previously been shown to catalyze the elongation of both AcPIM2 and Gl-X, and a C. glutamicum mutant strain lacking MptB has the same biochemical phenotype as ⌬NCgl1054, i.e. an accumulation of AcPIM2/Gl-X and loss of all downstream LM and LAM lipoglycans (17). Collectively, our observations suggest that LpqW regulates MptB function, either directly or through a, as yet uncharacterized, complex.
Despite our evidence for direct regulatory effects on MptB function, our findings could not entirely rule out an alternative scenario in which LpqW functions to enhance the rate of transmembrane flipping and/or regulates the transport of AcPIM2/ Gl-X precursors to MptB. Indeed, LpqW shares overall structural similarity to a large family of substrate-binding proteins that are commonly associated with ABC transporters in the cell membrane. The latter have been shown to translocate diverse metabolites including cell wall components (53). Substratebinding proteins are thought to initially bind transporter substrates in the periplasmic space and to be responsible for the substrate specificity of the cognate transporter (54,55). Although substrate-binding proteins are generally involved in import processes, it is possible that LpqW could associate with a flippase or ABC transporter involved in the transmembrane externalization of AcPIM2 or Gl-X. Such a role is consistent with the presence of a possible AcPIM-binding cleft in the LpqW crystal structure (42). Apart from regulating fluxes into different enzyme complexes, a glycolipid chaperone might also be required to regulate the movement of intermediates between distinct subdomains of the cell membrane (52). However, results from our cell free assays did not support such a transport/substrate accessibility role for LpqW because the loss of compartmentalization did not restore Gl-Y/Gl-Z synthesis in cell envelopes from the ⌬NCgl1054 strain, despite synthesis of these species in cell envelopes extracted from WT and the bypass mutant.
Having obtained evidence against a transport/substrate accessibility function for LpqW, we strongly favor our original hypothesis that LpqW regulates the activity of the AcPIM2/ Gl-X mannosyltransferase MptB in C. glutamicum (Fig. 9D). Such a model would be consistent with recent studies showing that E. coli lipoproteins have a role in regulating enzymes involved in peptidoglycan biosynthesis (56 -58). The E. coli lipoproteins, LpoA and LpoB, are N-terminally anchored into the outer membrane bilayer but can bind to and stimulate the enzymatic activity of the transpeptidases that are involved in the assembly of new peptidoglycan. LpqW, similar to LpoA/B, is predicted to be secreted into the periplasmic space, although its precise location in the cell wall has not been defined. However, it is conceivable that it is anchored into the outer membrane of C. glutamicum or Mycobacterium spp that are rich in mycolic acids and, in the case of C. glutamicum, cardiolipin (59). In addition to regulating the activities of key enzymes involved in cell wall assembly, outer membrane lipoproteins that span the periplasmic space may also be involved in mediating the insertion of (lipo)polysaccharides into the outer membrane. Our proposal that MptB and LpqW may directly interact in vivo would provide the first example of a cell wall polymerase/lipoprotein activator pair in mycobacteria.