Inactivation of Corynebacterium glutamicum NCgl0452 and the Role of MgtA in the Biosynthesis of a Novel Mannosylated Glycolipid Involved in Lipomannan Biosynthesis*

Mycobacterium tuberculosis PimB has been demonstrated to catalyze the addition of a mannose residue from GDP-mannose to a monoacylated phosphatidyl-myo-inositol mannoside (Ac1PIM1) to generate Ac1PIM2. Herein, we describe the disruption of its probable orthologue Cg-pimB and the chemical analysis of glycolipids and lipoglycans isolated from wild type Corynebacterium glutamicum and the C. glutamicum::pimB mutant. Following a careful analysis, two related glycolipids, Gl-A and Gl-X, were found in the parent strain, but Gl-X was absent from the mutant. The biosynthesis of Gl-X was restored in the mutant by complementation with either Cg-pimB or Mt-pimB. Subsequent chemical analyses established Gl-X as 1,2-di-O-C16/C18:1-(α-d-mannopyranosyl)-(1→4)-(α-d-glucopyranosyluronic acid)-(1→3)-glycerol (ManGlcAGroAc2) and Gl-A as the precursor, GlcAGroAc2. In addition, C. glutamicum::pimB was still able to produce Ac1PIM2, suggesting that Cg-PimB catalyzes the synthesis of ManGlcAGroAc2 from GlcAGroAc2. Isolation of lipoglycans from C. glutamicum led to the identification of two related lipoglycans. The larger lipoglycan possessed a lipoarabinomannan-like structure, whereas the smaller lipoglycan was similar to lipomannan (LM). The absence of ManGlcA-GroAc2 in C. glutamicum::pimB led to a severe reduction in LM. These results suggested that ManGlcAGroAc2 was further extended to an LM-like molecule. Complementation of C. glutamicum::pimB with Cg-pimB and Mt-pimB led to the restoration of LM biosynthesis. As a result, Cg-PimB, which we have assigned as MgtA, is now clearly defined as a GDP-mannose-dependent α-mannosyltransferase from our in vitro analyses and is involved in the biosynthesis of ManGlcAGroAc2.

The Corynebacterianeae represent a distinct and unusual group within Gram-positive bacteria, with the most prominent members being the human pathogens Mycobacterium tuberculosis and Mycobacterium leprae (1). In addition, the human pathogen Corynebacterium diphtheriae is the causal agent of diphtheria, and serious economic losses occur from the infection of animals by corynebacteria, such as Corynebacterium pseudotuberculosis and Corynebacterium matruchotii (2,3). Furthermore, nonpathogenic bacteria belong to this taxon, such as Corynebacterium glutamicum, which is used in the industrial production of amino acids (4,5).
A common feature to all these bacteria is that they possess an unusual cell wall matrix composed of mycolic acids, arabinogalactan, and peptidoglycan that is often referred to as the mycolyl-arabinogalactan-peptidoglycan complex (6 -13). In addition, they also possess a similar array of cell wall-associated glycolipids, such as phosphatidyl-myo-inositol (PI) 6 mannosides (PIMs) and lipoglycans, termed lipomannan (LM) and lipoarabinomannan (LAM) (12, 14 -18).
Four major PIMs, mono-and diacyl dimannosides (Ac 1 PIM 2 and Ac 2 PIM 2 ) and mono-and diacyl hexamannosides (Ac 1 PIM 6 and Ac 2 PIM 6 ), usually accumulate (12) with intermediates occurring in low abundance. Furthermore, the characteristic mycobacterial lipoglycans, LAM and LM, are both multiglycosylated versions of PIMs. We initially proposed the biosynthetic pathway PI 3 PIM 3 LM 3 LAM (15), which is now largely supported by biochemical and genetic evidence (19 -22). PimA catalyzes the addition of Manp provided by GDP-mannose to the 2-position of the myo-inositol of PI to form PIM 1 (21), whereas PimB might be responsible for the addition of a second Manp to the 6-position to yield Ac 1 PIM 2 (19). PimC has been demonstrated to allow further mannosylation to Ac 1 PIM 3 (20), and more recently PimE has been shown to be involved in the biosynthesis of Ac 1 PIM 5 (23). It has been proposed that PIM 4 is the direct precursor of LM, characterized by a linear ␣(136)-linked mannan backbone linked with ␣(132) mannopyranose side chains generated through Rv2181 (24). LM is then further glycosylated with arabinan to produce LAM and finally "mannose-capped" to produce ManLAM, a process initiated by the capping enzyme encoded by Rv1635c (25).
In the present study, we have established that C. glutamicum possesses both PIMs and lipoglycans, which are reminiscent of M. tuberculosis products, suggesting that conserved biosynthetic machineries are present within these two bacteria. Interestingly, the polar lipid profile revealed a previously uncharacterized major glycolipid. Furthermore, disruption of pimB in C. glutamicum abolished the synthesis of this novel glycolipid, which had a subsequent profound effect on LM biosynthesis. In addition, complementation with either Cg-pimB or Mt-pimB led to the restoration of wild type glycolipid and lipoglycan biosynthesis.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-The strains and plasmids used are given in Table 1 (26 -28). C. glutamicum ATCC 13032 (wild type) and Escherichia coli DH5␣mcr were grown in Luria-Bertani broth (LB; Difco) at 30 and 37°C, respectively. For C. glutamicum, kanamycin was used at a concentration of 25 g/ml, and spectinomycin was used at a concentration of 250 g/ml where appropriate. The minimal medium used for C. glutamicum was CGXII, and mutants were selected on LBHIS (28). Samples for lipid analyses were prepared by harvesting cells grown either on BHIS for 9 h up to an A 600 nm of 7-8 or on CGXII for 17 h up to A 600 nm of 47-51. Cells were harvested by centrifugation, followed by saline washing and freeze drying. M. tuberculosis H37Rv DNA was obtained from Dr. J. T. Belisle and the NIH Tuberculosis Research Materials and Vaccine Testing Contract at Colorado State University. All other chemicals were reagent grade or better and obtained from Sigma.
Construction of Plasmids-The inactivation vector pK19mobsacB::pimB was made by amplification of a 359-bp internal fragment of C. glutamicum pimB using the primer pair p557up and p557low (Table 2). After purification and treatment with polynucleotide kinase, the fragment was ligated with SmaI-cleaved pK19mobsacB. To construct pEKEx3-Cg-pimB, chromosomal DNA of C. glutamicum together with primers p557up3 and p557low3 and KOD DNA polymerase were used to amplify Cg-pimB. The resulting 1232-bp fragment was cloned into the SmaI site of pUC18, from which the fragment was reisolated by digestion with ScaI/EcoRI to ligate it with SmaI/EcoRI-cleaved pEKEx3. Similarly, Mt-pimB was amplified using primers ppimBMtex and ppimbMtrev using M. tuberculosis chromosomal DNA. The resulting 1153-bp fragment was cloned into pUC18, subsequently excised as a SmaI/ EcoRI fragment, which was ligated with SmaI/EcoRI-cleaved pEKEx3. All cloned fragments were verified by nucleotide sequencing.
Construction of Strains-Cells of C. glutamicum were made competent as described (28) and transformed by electroporation with pK19mobsacB::pimB to kanamycin resistance, signifying the integration of the construct into the chromosome. Using the two different primer pairs prsp1/p557up2 and puni/ p557low2, respectively, the correct disruption of Cg-pimB integration was verified. Competent cells of one disruption mutant were chosen and transformed by electroporation with either pEKEx3-Cg-pimB or pEKEx3-Mt-pimB to spectinomycin resistance, and the plasmid integrity of the recombinant clones was confirmed in plasmid preparations.
Lipid Extraction and Analysis-Polar lipids and apolar lipids were initially extracted from 6 g of dry C. glutamicum cells according to the procedures of Dobson et al. (29) by stirring in 220 ml of methanolic saline (20 ml of 0.3% NaCl and 200 ml of CH 3 OH) and 220 ml of petroleum ether for 2 h. The cells were centrifuged at 3000 rpm for 5 min. The resulting biphasic solu-  Restriction enzyme recognition sites are in boldface type, and the ribosome binding site is in italic type.

Name Primer
tion was separated, and the upper layer containing apolar lipids was recovered. An additional 220 ml of petroleum ether was added, mixed, and harvested as described above. The two upper petroleum ether fractions were combined and dried under reduced pressure.
To extract polar lipids, 260 ml of CHCl 3 , CH 3 OH, 0.3% NaCl (9:10:3, v/v/v) was added to the lower aqueous CH 3 OH layer, and the solution was stirred for 4 h. This mixture was filtered, and the filter cake was re-extracted twice with 85 ml of CHCl 3 , CH 3 OH, 0.3% NaCl (5:10:4, v/v/v). CHCl 3 (145 ml) and 0.3% NaCl (145 ml) were added to the combined filtrates. This mixture was stirred for 1 h and allowed to settle, and the lower layer containing the polar lipids was recovered and dried under reduced pressure. The polar lipid extract was examined by twodimensional TLC on aluminum-backed plates of silica gel 60 F 254 (Merck 5554), using CHCl 3 , CH 3 (29). Glycolipids were visualized by spraying plates with ␣-naphthol/sulfuric acid followed by gentle charring of plates, by spraying with 5% ethanolic molybdophosphoric acid and charring, or by using a Dittmer and Lester reagent that is specific for phospholipids and glycophospholipids.
In Vitro Synthesis of Radiolabeled Mannolipids-C. glutamicum strains ATCC 13032 and the 13032::pimB mutant transformed with either pEKEx3, pEKEx3-Gg-pimB, or pEKEx3-Mt-pimB were cultured to the midlogarithmic growth phase in 2 liters of BHIS medium with appropriate antibiotics. Cells were harvested by centrifugation; resuspended in 20 ml of 50 mM MOPS, pH 7.9, 5 mM 2-mercaptoethanol, 5 mM MgCl 2 ; and lysed immediately by passing twice through a French pressure cell (aperture pressure change ϭ 3000 p.s.i.). The lysate was clarified by centrifugation at 27,000 ϫ g (4°C, 30 min), and membranes were deposited by centrifugation of the supernatant at 105,000 ϫ g (4°C, 90 min). The membrane pellet was gently resuspended in the above buffer to 250 l and supple-mented with CaCl 2 to 10 mM and 25 g of amphomycin in order to inhibit transfers of radiolabel to polyprenyl phosphate acceptors. This mixture was held at 37°C for 15 min prior to the addition of 0.25 Ci of GDP-Man (0.3054 Ci/mmol; GE Healthcare). The reaction was held at 37°C for a further 45 min before quenching with 4 ml of CHCl 3 /CH 3 OH/H 2 O (10:10:3, v/v/v). A biphase was formed by the addition of 1.75 ml of CHCl 3 and 0.55 ml of H 2 O. The lower organic phase was washed twice with 2 ml of CHCl 3 /CH 3 OH/H 2 O (3:47:48, v/v/v). The extract was dried, and radioactivity was quantified by liquid scintillation counting of a sample of the extract. Two-dimensional TLC analysis was carried out as above, and the mannolipids formed were visualized by phosphorimaging (Kodak K Screen).
Extraction and Purification of Lipoglycans-Lipoglycans were extracted from delipidated cells as previously described (31,32). Briefly, cells were broken by sonication (MSE Soniprep 150; 12-m amplitude, 60 s on, 90 s off for 10 cycles, on ice) and the cell debris was refluxed five times with 50% C 2 H 5 OH at 68°C for 12-h intervals. The cell debris was removed by centrifugation, and the supernatant containing lipoglycans, neutral glycans, and proteins dried. This dried extract was then treated with hot phenol-H 2 O. The aqueous phase was dialyzed and dried before separate and extensive treatments with ␣-amylase, DNase, and RNase chymotrypsin and trypsin. Finally, the lipoglycan fraction was dialyzed extensively against water.
The crude lipoglycan extract was resuspended in buffer A (50 mM ammonium acetate and 15% propan-1-ol) and subjected to octyl-Sepharose CL-4B hydrophobic interaction chromatography (2.5 ϫ 50 cm) as previously reported (33). The column was initially washed with 4 column volumes of buffer A to ensure removal of neutral glycans, followed by buffer B (50 mM ammonim acetate and 50% propan-1-ol). The eluate was collected and concentrated to ϳ1 ml and precipitated using 5 ml of C 2 H 5 OH, and the sample was dried using a Savant SpeedVac. The freeze-dried sample containing the retained material from the hydrophobic interaction column was then resuspended in buffer C (0.2 M NaCl, 0.25% sodium deoxycholate (w/v), 1 mM EDTA, and 10 mM Tris-HCl, pH 8) to a final concentration of 200 mg/ml. The sample was gently mixed and left to incubate for 48 h at room temperature. The sample was then loaded onto a 200 ml Sephacryl S-200 column previously equilibrated with buffer C. The sample was eluted with 400 ml of buffer C at a flow rate of 3 ml/h, collecting 1.5-ml fractions. The fractions were monitored by SDS-PAGE using either a silver stain utilizing periodic acid and silver nitrate (34) or a Pro-Q emerald glycoprotein stain (Invitrogen), and individual fractions were pooled and dialyzed extensively against buffer D (10 mM Tris-HCl, pH 8, 0.2 M NaCl, 1 mM EDTA) for 72 h with frequent changes of buffer. The samples were further dialyzed against deionized water for 48 h with frequent changes of water, lyophilized, and stored at Ϫ20°C.
Glycosyl Compositional and Linkage Analysis-Glycosyl compositional analysis was performed by either routine gas chromatography (GC) or capillary electrophoresis analysis as described previously (35). Glycosyl linkage analyses were performed as described previously (36). Briefly, per-O-methylated samples were hydrolyzed using 500 l of 2 M trifluoroacetic acid FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 at 110°C for 2 h, reduced using 350 l of a 10 mg/ml solution of NaBD 4 (1 M aqueous NH 4 OH/C 2 H 5 OH, 1:1, v/v), and per-Oacetylated using 300 l of acetic anhydride for 1 h at 110°C. The resulting alditol acetates were solubilized in cyclohexane before analysis by GC and gas chromatography/mass spectrometry (GC/MS) (37).

Identification of a Novel Mannosylated Glycolipid
GC analysis was performed using a Thermoquest Trace GC 2000 equipped with a flame ionization detector. Samples were separated using a temperature program as follows. Injector temperature was set at 50°C, held for 1 min, and then increased to 110°C at 20°C/min. The oven was held at 110°C and then ramped to 290°C at 8°C/min and held for 5 min to ensure that all of the products had eluted from the column. All of the data were collected and analyzed using Xcaliber (version 1.2) software.
Gl-A, Gl-X, and Ac 1 PIM 2 Matrix-assisted Laser Desorption Ionization-Time of Flight-Mass Spectrometry (MALDI-TOF-MS) Analyses-Analyses of PIMs were carried out on a Voyager DE-STR MALDI-TOF instrument (PerSeptive Biosystems, Framingham, MA) using the reflectron mode of detection as previously described (38). PIMs were analyzed by the instrument operating at 20 kV in the negative ion mode, using an extraction delay time set at 200 ns. Typically, spectra from 100 -250 laser shots were summed to obtain the final spectrum. All of the samples were prepared for MALDI-TOF-MS analyses using the on-probe sample clean-up procedure with cation exchange resin. The 2-(4-hydroxyphenylazo)-benzoic acid matrix was used at a concentration of 10 mg/ml in C 2 H 5 OH/H 2 O (1:1, v/v). Typically, 0.5 l of PIM sample (10 g) in a CHCl 3 /CH 3 OH/ H 2 O solution and 0.5 l of the matrix solution, containing 5-10 cation exchange beads, were deposited on the target, mixed with a micropipette, and dried under a gentle stream of warm air.
Analyses of the Gl-A and Gl-X were carried out on a 4700 Proteomics Analyzer (with TOF-TOF optics; Voyager DE-STR; Applied Biosystems, Framingham, MA) using the reflectron mode. Ionization was effected by irradiation with a Nd:YAG laser (355 nm) operating by pulses of 500 ps with a frequency of 200 Hz. Gl-A and Gl-X were analyzed in the positive ion mode. Spectra from 2500 to 5000 laser shots were summed to obtain the final spectrum. Typically, 0.3 l of Gl-X (at 10 mg/ml in CHCl 3 /CH 3 OH, 1:1, v/v) and 0.3 l of the matrix solution (2,5dihydrobenzoic acid at ϳ10 mg/ml in C 2 H 5 OH/H 2 O, 1:1, v/v) were deposited on the target. Collision-induced dissociation (CID)-MS/MS was performed using atmosphere as the gas type. The pressure was set to medium, and the collision energy was set to 1 kV.
Analysis of per-O-methylated Gl-X was performed using an Applied Biosystems 4800 MALDI-TOF/TOF in the positive mode. MS/MS was performed with the collision energy set to 1 kV with air as the collision gas. The per-O-methylated sample was dissolved in CH 3 OH, and 0.5-l aliquots were loaded onto the target plate with 0.5 l of the matrix 2,5-dihydrobenzoic acid at 10 mg/ml in CH 3 OH/H 2 O. Sequazyme peptide mass standards were used as external calibrants (Applied Biosystems).
NMR Spectroscopy-NMR spectra of lipoglycans were recorded on a Bruker DMX-500 equipped, with a double reso-nance ( 1 H/X)-BBi z-gradient probe head. All samples were exchanged in D 2 O (D, 99.97% from Euriso-top, Saint-Aubin, France), with intermediate lyophilization, and then dissolved in 0.5 ml of D 2 O and analyzed at 313 K. The 1 H and 13 C NMR chemical shifts were referenced relative to internal acetone at 2.225 and 34.00 ppm, respectively. All of the details concerning NMR sequences used and experimental procedures were described in previous studies (39,40).
Gl-X NMR spectra were recorded in d 6 -Me 2 SO. NMR spectra were recorded on a Bruker DRX500 operating at 500. 13 MHz for 1 H and 125.77 MHz for 13 C. The instrument was equipped with a 5-mm 1 H-{ 13 C}-{X}-triple broadband inverse triple resonance z-gradient probe head, and all spectra were run at 300 K. d 6 -Me 2 SO was used as lock, and the residual solvent was used as internal reference (2.49 ppm for 1 H and 39.5 ppm for 13 C). Data were acquired and processed using XWINNMR version 2.6 software on a Silicon Graphics work station. The 13 C NMR data were acquired with 1 H decoupling and NOE. Additionally and under the same conditions, a DEPT-135 experiment was performed to provide spectral editing information. All two-dimensional NMR data were acquired nonspinning. Data points (2048) were used in acquisition for the fast domain (F2), and 512 points were used in the incremented domain (F1). The F1 dimension was zero-filled to 1024 data points in processing. The gradient COSY-90 data were transformed using a magnitude calculation, and a nonshifted sine bell window function was used in both frequency domains. The HSQC, NOESY, and T-ROESY NMR experiments (all performed using pulsed field gradients) were acquired in phasesensitive mode using the time-proportional phase increment for the HSQC and NOESY experiments, whereas the Statestime-proportional phase increment was used for the T-ROESY experiment. In all three experiments, a Qsine window function shifted by /2 was used for both frequency domains in processing. GARP decoupling was employed in F1 for the HSQC experiment. A mixing period of 450 ms was employed in the NOESY experiment. A mixing period of 400 ms was used in the T-ROESY experiment. To minimize the HOHAHA contribution to the T-ROESY data, a spin lock field of 8.8 kHz was used, and the transmitter was offset by 8.8 kHz (during the spin lock period only). The spin lock field was also ramped using a Squar-eramp20 shaped pulse (described by 1000 points). In both NOESY and T-ROESY experiments, 24 transients/increment were employed.

RESULTS
Genomic Organization of the pimB Locus-Mt-pimB has been shown to encode an ␣-mannosyltransferase potentially involved in Ac 1 PIM 2 biosynthesis (19), and this gene is predicted to be essential in M. tuberculosis (41). However, the biosynthesis of PIMs remain unaffected upon disruption of Mt-pimB, 7 suggesting either a degree of redundancy or that Mt-pimB performed another function in M. tuberculosis. Mt-pimB lies within a cluster of genes involved in menaquinone biosynthesis (Fig. 1A). Directly upstream is a small open reading frame of unknown function and menD, and directly down-7 G. S. Besra and L. S. Schlesinger, unpublished data. stream is menH. The entire locus consisting of seven genes is syntenic in all sequenced Mycobacterium species and, in part, also in Corynebacterium, both genera belonging to the taxon Corynebacterianeae (Fig. 1A). The putative orthologue of Mt-pimB from C. glutamicum, Cg-pimB (NCgl0452), shares 49% sequence identity with the M. tuberculosis gene, and, since C. glutamicum can be regarded as a model organism of this taxon due to its archetypical genomic organization as manifested in a low number of gene duplications, its structural simplicity, and ease of handling (42), we decided to study the function of Mt-pimB in more detail using C. glutamicum.
Construction and Growth of the C. glutamicum 13032:: pimB Disruption Mutant-C. glutamicum ATCC13032 was transformed to kanamycin resistance with plasmid pK19mobsacB::pimB (Fig. 1B) using the method we have previously reported (43,44). Fourteen colonies were obtained from 1 g of plasmid DNA. The clones were analyzed via PCR, and all were found to have the vector integrated chromosomally, thus demonstrating disruption of the Cg-pimB coding sequence. One strain was chosen and termed 13032::pimB. This strain displayed no detectable change in phenotype in terms of growth rate or colony morphology (data not shown) and was transformed with either pEKEx3-Cg-pimB or pEKEx3-Mt-pimB as well as the unmodified vector, pEKEx3, as a control for further studies.
Chromatographic Analysis of Polar Lipids-Polar lipids were extracted from wild type C. glutamicum, and glycolipid profiles were recorded by two-dimensional TLC (Fig. 2). The faster The locus in other Mycobacterium species and Corynebacterium species is largely syntenic. par., paratuberculosis. B, construction of C. glutamicum 13032::pimB. Shown is Cg-pimB with its adjacent genes NCgl0451 and NCgl0453 and the strategy to disrupt Cg-pimB using vector pK19mobsacBpimB-int. This vector carries a 359-bp internal fragment of Cg-pimB of 1221 bp in size, thereby enabling homologous recombination with the wild type genome to generate C. glutamicum::pimB. The arrow pairs marked P1 (prsp1/p557up2) and P2 (puni/p557low2) locate the primers used for the PCR analysis to confirm the proper integration of vector pK19mobsacBpimB-int in C. glutamicum 13032::pimB and its absence in wild type C. glutamicum. The results of the PCR analysis are shown on the right, where P1 and P2 mark the result obtained with primers P1 and P2, respectively. Samples were applied pairwise with the amplificate obtained from C. glutamicum 13032::pimB applied in the left lane, exhibiting the expected PCR products of 442 and 603 bp, respectively, whereas these were absent in wild type C. glutamicum (WT). St marks the standard, and the arrowheads are located at 0.5, 1, and 1.5 kb, respectively.  (29). Glycolipids were visualized by spraying plates with ␣-naphthol/sulfuric acid, followed by gentle charring of the plates (A-D). E-H show in vitro labeling of polar lipids utilizing GDP-[ 14 C]Man, and products were analyzed using the same two-dimensional TLC system. The novel glycolipids Gl-X and Gl-A, and Ac 1 PIM 2 are highlighted. migrating lipids were confirmed as trehalose monocorynomycolate and glucose monocorynomycolate (data not shown). Interestingly, in the pattern shown in Fig. 2A, not all of the remaining glycolipids gave a positive response with the Dittmer-Lester lipid phosphate reagent. The predominant lipid phosphate spot, which was also carbohydrate-positive, corresponded to Ac 1 PIM 2 and was confirmed by negative ion mode MALDI-MS analyses due to the characteristic ions at m/z 1398 (M Ϫ H) Ϫ and in positive mode at m/z 1444 (M Ϫ H ϩ 2Na) ϩ (data not shown). The phosphorus-free glycolipid, indicated as Gl-X ( Fig. 2A) was unusual in terms of chromatographic mobility and staining properties. Analysis of the C. glutamicum 13032::pimB disruption mutant revealed that synthesis of Ac 1 PIM 2 was intact; however, it failed to produce Gl-X (Fig.  2B). Transformation with pEKEx3-Cg-pimB complemented the mutant phenotype and restored Gl-X biosynthesis (Fig. 2C). Transformation with pEKEx3-Mt-pimB also restored the wild type phenotype, suggesting that the product of Mt-pimB was able to complement the lesion in Gl-X biosynthesis in the mutant (Fig. 2D).
To further clarify the structure of Gl-X and the role of Cg-pimB, the crude polar lipid extract was fractionated using anion exchange chromatography on DEAE-cellulose using a stepwise gradient of increasing ammonium acetate concentration in CHCl 3 /CH 3 OH ranging from 1 mM to 500 mM. It was anticipated that since Gl-X was phosphorus-negative, this would provide a convenient purification protocol allowing the neutral glycolipid to elute from the column while retaining the contaminating phospholipids. Surprisingly, Gl-X was also retained on the DEAE column and was eluted with 15 mM ammonium acetate in CHCl 3 /CH 3 OH (2:1), just before Ac 1 PIM 2 , suggesting that it possesses an acidic group. Pooled fractions containing Gl-X were pooled, and the lipid was purified further by preparative TLC and analyzed by MALDI-TOF-MS, 1 H, 13 C two-dimensional COSY, and two-dimensional HSQC NMR.
Chemical Composition of Gl-X-Initial glycosyl compositional analysis using alditol acetates determined the presence of mannose by GC (data not shown). 1 H and 13 C NMR spectra recorded in d 6 -Me 2 SO were in agreement with a diglycosyl diacylated glycerol. Indeed, the 1 H-13 C HSQC NMR spectrum showed two anomeric resonances at ␦ H1/C1 4.98/100.3 (I 1 ) and 4.57/99.0 (II 1 ). The diacylated glycerol unit was identified by two-dimensional 1 H-1 H COSY NMR spectrum (Fig. 3A)  Gl-X was subsequently analyzed both in negative and positive mode MALDI-TOF-MS. Spectra were only obtained in positive mode, revealing a molecular ion at m/z 977 (Fig. 4A). Positive ion MALDI-TOF CID-MS/MS spectrum of the cationized sodiated precursor ion (M Ϫ H ϩ 2Na) ϩ of Gl-X revealed ions at m/z 721 and 695, corresponding to the loss of C 16 and C 18:1 fatty acids, respectively (Fig. 4B), which were also later confirmed by fatty acid methyl ester analysis by GC/MS (data not shown).
However, a simple Man-Man-GroAc 2 (containing no inositol) and both mannose units interconnected did not coincide with the deduced molecular weight (m/z 977), suggesting that we were not dealing with a simple Man-Man type structure but a Man-Y-GroAc 2 with Y possibly carrying an acidic function, explaining the retention of Gl-X on DEAE-cellulose. Indeed, the molecular ion at m/z 977 is in agreement with a Man-Y-Gro-C 16 /C 18:1 structure, where Y represents a hexosyl uronic acid in the (M Ϫ H ϩ 2Na) ϩ form. MALDI-TOF/TOF analysis of the per-O-methylated Gl-X glycan observed at m/z 579 (M ϩ Na) ϩ gave data rich in informative fragment ions (Fig. 4C). In particular, the data are indicative of Y being consistent with a hexuronic acid (Fig. 4C, inset). These assignments were substantiated by MS/MS analysis of the per-O-deuteriomethylated glycan observed at m/z 606 (M ϩ Na) ϩ . Furthermore, mild hydrolysis of the per-O-deuteriomethylated glycan with methanolic HCl resulted in a shift of 3 mass units to m/z 603, consistent with methyl exchange on the carboxylic group of hexuronic acid.
The linkage of the two different glycosyl units was established from 1 H-1 H ROESY, T-ROESY NMR, and 13 C NMR experiments (data not shown) that revealed that C-4 of the unit II at 78.6 ppm was shifted to low field away from the remaining resonances at 68 -74 ppm. H-1 of unit I (I 1 ) at 4.98 ppm showed an intense interresidue NOE with H-4 of unit II (II 4 ) at 3.59 ppm and weaker NOE with H-5 (II 5 ) and H-3 (II 3 ) of unit II at 3.53 and 3.48 ppm, respectively. Taken together, these data indicate that unit I is linked at O-4 of unit II. 1 H and 13 C resonances of spin system I correlate with a t-Manp unit, with an ␣-anomeric configuration suggested by the presence of an intense intraresidue contact between H-1 at 4.98 ppm (I 1 ) and H-2 at 3.64 ppm (I 2 ) and the absence of an intraresidue H-1/H-3 NOE contact.
Based on the assignment of 1 H-1 H COSY NMR and proton coupling constants, unit II (sugar Y) was shown to be ␣-D-glucopyranosyluronic acid as follows. The small J 1,2 coupling constant (3.2 Hz) of H-1 at 4.57 ppm (II 1 ) indicated an ␣-anomeric configuration. The large coupling constant of H-2 (dd) at 3.19 ppm (II 2 ) J 2,3 (9.6 Hz), H-4 (t) at 3.59 ppm (II 4 ) J 3,4 (9.6 Hz), H-5 (d) at 3.53 ppm (II 5 ) J 4,5 (9.6 Hz) correlates with a gluco-configuration of sugar Y. The connectivity of H-5 (II 5 ) to only one proton H-4 (II 4 ) in the two-dimensional COSY NMR spectra while a distinct signal for an extra carboxyl group at 171.6 in the 13 C NMR spectra is clearly visible indicating that sugar Y is uronic acid. In addition, H-1 of glucuronic acid (II 2 ) at 4.57 ppm showed an interresidue NOE contact with H-3 of the glycerol unit (III 3 ) at 3.45 ppm, demonstrating that unit II is linked at O-3 of the glycerol backbone. The complete assignment of resonances of both glycosyl residues is given in Table 3.
In Vitro Analysis of Mannosyltransferase Activity-In order to confirm the mannosyltransferase activity and acceptor specificity of both Cg-PimB and Mt-PimB, we prepared membrane fractions from wild type C. glutamicum, the 13032::pimB mutant, and transformants of the latter bearing pEKEx3-Cg-pimB and pEKEx3-Mt-pimB. Analysis of radiolabeled mannolipids, formed after introduction of GDP-[ 14 C]mannose by two-dimensional TLC (Fig. 2, E-H), revealed a clear difference between the profiles derived from the wild type strain and the 13032::pimB mutant. The upper spot of the dominant pair of glycolipids, corresponding to ManGlcAGroAc 2 , is clearly absent in the mutant profile. Complementation with Cg-pimB in trans, as expected, restored ManGlcAGroAc 2 biosynthesis, and, consistent with our extracted lipid profiles, complementation with Mt-pimB was also achieved. In all of the profiles, the second spot corresponding to Ac 1 PIM 2 was evident, the apparent increase in its abundance in the 13032::pimB mutant profile being due to the absence of ManGlcAGroAc 2 when equal radioactivity was loaded on the plates. The identities of these two lipids were confirmed by staining of the same TLC plates with the Dittmer-Lester lipid phosphate reagent (data not shown).
Chemical Analysis of C. glutamicum Lipoglycans-We examined extracts of wild type C. glutamicum and the C. glutamicum 13032::pimB disruption mutant for lipoglycans by SDS-PAGE followed by staining for carbohydrates with either silver nitrate or Pro-Q Emerald stain for glycoconjugates. Extracts of wild type C. glutamicum contained two closely migrating lipoglycans (Fig. 5). Interestingly, the lower molecular weight lipoglycan was significantly reduced in the C. glutamicum  FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4567 13032::pimB mutant (Fig. 5). A two-step purification protocol was performed to fractionate the lipoglycans from wild type C. glutamicum and the C. glutamicum 13032::pimB mutant, and fractions containing the lipoglycans were monitored by SDS-PAGE stained with either silver nitrate or Pro-Q Emerald staining.
Interestingly, in comparison with wild type C. glutamicum, the C. glutamicum 13032::pimB disruption mutant and analysis by SDS-PAGE using Pro-Q Emerald stain for glycoconjugates revealed that the upper lipoglycan appears unaffected (Fig. 5). This was confirmed by glycosyl compositional analysis and per-O-methylation analysis of Cg-pimB LAM (data not shown). However, the smaller lipoglycan, Cg-pimB-LM, was now barely detectable (Fig. 5). Glycosyl compositional analysis of the residual Cg-pimB-LM now revealed the presence of both mannose and inositol (47:1), and per-O-methylation analysis indicated the presence of t-Manp, 2-Manp, 6-Manp, and 2,6-Manp (data not shown). Furthermore, 1 H NMR spectra exhibited an anomeric region with the typical resonances corresponding to these different units (data not shown). The pattern was simpler than that observed for the wild type Cg-LM and corresponded to the profile typically observed for mycobacterial PI-based LM. Integration of the resonances indicated a mannan core with a reduced branching degree as compared with the wild type Cg-LM.
In summary, these results suggest that Cg-LM is most likely two components, a dominant Cg-LM based on the ManGlcA-GroAc 2 and a minor component akin to the characteristic mycobacterial PI-based LM.

DISCUSSION
Along with the genus Mycobacterium, species of Corynebacterium belong to a suprageneric actinomycete taxon termed Corynebacterianeae, which also includes Rhodococcus, Nocardia, and other closely related genera. In this study, we sought to establish the role of Cg-PimB and Mt-PimB and whether C. glutamicum possesses both PIMs and lipoglycans reminiscent of M. tuberculosis products, suggesting conserved biosynthetic machineries within these two bacteria. In M. tuberculosis, Mt-pimB (Rv0557) was shown earlier to encode an ␣-D-mannose-␣(136)-phosphatidyl-myo-inositol-monomannoside mannosyltransferase and to be involved in the formation of Ac 1 PIM 2   from GDP-mannose and Ac 1 PIM 1 (19). In this study, we attempted to disrupt Cg-pimB and examine the consequences of PIM and LAM biosynthesis in C. glutamicum.
These results raised several interesting questions and possibilities concerning the biochemical role and function of Cg-PimB and Mt-PimB, the immediate question being whether Cg-PimB represents a GDP-mannose-dependent ␣-mannosyltransferase or adds glucopyranosyluronic acid to a diacylated glycerol precursor. During our purification process of polar lipids, we isolated and characterized in the C. glutamicum 13032::pimB disruption mutant a second phosphorus-free gly-colipid, which we termed Gl-A (Fig. 2). Gl-A was subsequently analyzed both in negative and positive mode MALDI-MS. Spectra were only obtained in positive mode, revealing a molecular ion at m/z 815 (M Ϫ H ϩ 2Na) ϩ (Fig. 9), which is in agreement with previous reports of a 1,2-di-O-acyl-(␣-D-glucopyranosyluronic acid)-(133)-glycerol, which would represent the precursor of ManGlcAGroAc 2 as GlcAGroAc 2 (46). This would suggest that PimB represents a bona fide ␣-mannosyltransferase, which we have now reassigned as an ␣-mannosylglucopyranosyluronic acid-transferase A (MgtA), to a diacylated glucuronosyl glycerol involved in ManGlcAGroAc 2 biosynthesis, since the synthesis of GlcAGroAc 2 remained unaffected in the C. glutamicum 13032::pimB disruption mutant. This assignment is wholly compatible with our in vitro mannosylation data. Interestingly, we observed no consistent accumulation of Gl-A as might be expected, although the amounts of the lipid were increased in some cultures. Presumably, the regulation of this pathway is complex and will require careful study.
Glycosylated diacylglycerols are commonly found in Grampositive bacteria as well as in higher plants. It was previously established that these glycosylated diacylglycerols function as precursors/anchors for hyperglycosylated variants, such as the lipomannans, as found in the case of dimannosyl diacylglycerols in Micrococcus and lipoteichoic acids (47)(48)(49). Moreover, related di-and monoacylglycerols containing glucuronosyl residues have been well reported in Pseudomonas spp. (50 -52), Bacillus cereus T (53), and halotolerant bacteria (54 -57). In relation to Mycobacterium, the data are sparse in regards to the presence of glycosyl diacylglycerols, except for a few limited cases, such as a diglucosyl diacylglycerol (58). The presence of the uronic acid residue in ManGlcAGroAc 2 is interesting due to its rarity with regard to the published literature, especially with reference to Mycobacterium spp. In mycobacteria to date, a glucuronosyl diacylglycerol in Mycobacterium smegmatis (46), uronosyl-containing glycopeptidolipid in Mycobacterium  avium (59 -61), and a uronosyl-containing polar glycopeptidolipid in Mycobacterium habana (62) have been reported. This is the first description of its kind in Corynebacterianeae of a mannosylated variant of glucuronosyl diacylglycerol.
Our results also suggest that ManGlcAGroAc 2 also participates in the biosynthesis of a novel Cg-LM-like molecule. The above notion is based on two observations. First, when Man-GlcAGroAc 2 is not synthesized in the C. glutamicum 13032::pimB mutant, the production of Cg-LM is greatly reduced, but when complemented with either Cg-pimB, Cg-LM biosynthesis is restored (Fig. 5). Second, glycosyl compositional analysis of the residual Cg-pimB-LM revealed the presence of both mannose and inositol. Per-O-methylation analysis indicated the presence of t-Manp, 2,6-Manp, and 6-Manp. The pattern was simpler than that observed for the wild type Cg-LM and corresponded to the profile typically observed for mycobacterial PI-based LM. In summary, these results suggest that Cg-LM is most likely two components, a dominant Cg-LM based on the ManGlcAGroAc 2 and a minor component akin to the characteristic mycobacterial PI-based LM.
Although Cg-LM and Cg-LAM are consistent in belonging to the LM/LAM archetype, these lipoglycans are notably distinct from mycobacterial LAMs. In addition, given the potent immunomodulatory effects of PIMs/LM/LAM from M. tuberculosis, the possibility that similar products to Gl-X, Gl-A, and related lipoglycans could exist in M. tuberculosis is intriguing. We are currently developing methods to fractionate polar lipids and lipoglycans from M. tuberculosis in a search for such novel glycolipids. Linked to this is the question of the role of Mt-PimB. Our results clearly show that Cg-pimB plays no crucial role in PIM biosynthesis in C. glutamicum. However, since relaxed substrate specificities in microbial glycosyltransferases have been recorded on heterospecific expression (63), we must not overinterpret the ability of Mt-pimB to complement the lesion in ManGlcAGroAc 2 biosynthesis in our C. glutamicum 13032::pimB mutant. The possibility remains that the M. tuberculosis PimB does contribute to the biosynthesis of PIMs but merely has relaxed acceptor specificity and is able to utilize GlcAGroAc 2 . The clear definition of the substrate specificities of these enzymes will require their purification to homogeneity, and toward this we have attempted to construct a number of fusion proteins to afford a facile purification protocol but have yet to produce soluble recombinant protein. A protocol yielding active soluble protein is now being sought to address this point.