Genetic Basis for the Synthesis of the Immunomodulatory Mannose Caps of Lipoarabinomannan in Mycobacterium tuberculosis*

Lipoarabinomannan (LAM) is a high molecular weight, heterogenous lipoglycan present in abundant quantities in Mycobacterium tuberculosis and many other actinomycetes. In M. tuberculosis, the non-reducing arabinan termini of the LAM are capped with α1→2 mannose residues; in some other species, the arabinan of LAM is not capped or is capped with inositol phosphate. The nature and extent of this capping plays an important role in disease pathogenesis. MT1671 in M. tuberculosis CDC1551 was identified as a glycosyltransferase that could be involved in LAM capping. To determine the function of this protein a mutant strain of M. tuberculosis CDC1551 was studied, in which MT1671 was disrupted by transposition. SDS-PAGE analysis showed that the LAM of the mutant strain migrated more rapidly than that of the wild type and did not react with concanavalin A as did wild-type LAM. Structural analysis using NMR, gas chromatography/mass spectrometry, endoarabinanase digestion, Dionex high pH anion exchange chromatography, and matrix-assisted laser desorption ionization-time-of-flight mass spectrometry demonstrated that the LAM of the mutant strain was devoid of mannose capping. Since an ortholog of MT1671 is not present in Mycobacterium smegmatis mc2155, a recombinant strain was constructed that expressed this protein. Analysis revealed that the LAM of the recombinant strain was larger than that of the wild type, had gained concanavalin A reactivity, and that the arabinan termini were capped with a single mannose residue. Thus, MT1671 is the mannosyltransferase involved in deposition of the first of the mannose residues on the non-reducing arabinan termini and the basis of much of the interaction between the tubercle bacillus and the host cell.

Lipoarabinomannan (LAM) 2 is a high molecular weight amphipathic lipoglycan, which makes up one of the major components of the cell wall of mycobacteria and exhibits a wide spectrum of immunomodulatory effects. Its structure is complex and heterogeneous with three distinct structural domains, including a phosphatidylinositol anchor (PI anchor), a branched mannan, and a branched arabinan (Fig. 1). The PI anchor is composed of a myo-inositol phosphoryl diacylglycerol substituted at the 2 position with a single mannopyranose (Manp) and at the 6 position with the mannan; this structure is identical to those found in the mycobacterial phosphatidylinositolmannosides and lipomannan (1,2), which are thought to be biosynthetic precursors of LAM (3). The mannan core is linked to the 6 position of the myoinositol residue of the PI anchor and consists of a linear ␣(136) Manp chain with varying degrees of ␣(132) Manp branching with single Manp residues (3). This structure is conserved in all mycobacterial species studied with two exceptions (4,5). The arabinan component consists of linear stretches of ␣(135) arabinofuranose (Araf) residues with some ␣(133) branching. Two distinct non-reducing termini are generated with a ␤-D-Araf-(132)-␣-D-Araf disaccharide linked to the 5 position of ␣-D-Araf-(135)-␣-D-Araf-(13 or the 3 and 5 positions of ␣-D-Araf-(135)-␣-D-Araf (13, resulting in the formation of well characterized Ara 4 and Ara 6 motifs, respectively, when digested with an endoarabinanase obtained from a Cellulomonas species ( Fig. 1; Refs. 3 and 6). In slow growing mycobacteria such as Mycobacterium tuberculosis, Mycobacterium leprae, Mycobacterium bovis, Mycobacterium avium, and Mycobacterium kansasii (2,5,(7)(8)(9)(10)(11)(12) a portion of the non-reducing termini of the Araf chains is capped, to varying degrees, with short ␣(132) Manp chains consisting of one to three residues ( Fig. 1), thus the molecule is termed ManLAM. The situation is different in rapidly growing mycobacteria. LAM isolated from Mycobacterium smegmatis and an unidentified Mycobacterium sp. is largely uncapped with a small fraction being capped with inositol phosphate (PILAM) (7,13) and in Mycobacterium chelonae, no modification exists on the arabinosyl termini (AraLAM) (4).
LAM has been implicated in a plethora of biological functions; typically ManLAM is thought to be anti-inflammatory, while PILAM is thought to be pro-inflammatory. ManLAM has * This work was supported by National Institutes of Health/NIAID Grants AI49151 (to D. C. C.), AI64798 and AI18357 (to P. J. B.), and AI37139 (to D. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed. been implicated in inhibition of phagosomal maturation, apoptosis, and interferon-␥ signaling in macrophages and interleukin-12 secretion of dendritic cells (reviewed in Refs. 3, 14, and 15). It has also been suggested that the type of LAM capping is a major structural feature in determining how the immune system is modulated (14), and a recent publication suggests that dendritic cell-specific intercellular adhesion molecule-3-grabbing non-integrin (DC-SIGN) may act as a pattern recognition receptor and discriminate between Mycobacterium species through selective recognition of the Manp caps on LAM molecules (16). Thus, LAM structure is generally considered to be a crucial factor in mycobacterial pathogenesis. Despite the level of understanding of the structure of LAM and the variety of immunomodulatory effects it mediates, nothing was known of the enzymes involved in capping of Man-LAM. However, the facts that decaprenylphosphorylarabinofuranose is the only known donor of the Araf residues in the mycobacterial cell wall (17), and synthesis of polar phosphatidylinositolmannosides and linear LM can be inhibited by amphomycin (18,19), suggested that the glycosyltransferases (GTs) involved in the later steps of ManLAM synthesis likely utilize prenylphosphorylglycoses as sugar donors. This idea was supported by the fact that in Corynebacterium glutamicum disruption of polyprenylphosphorylmannose synthase completely obviates lipoglycan synthesis (20). Therefore, the enzymes involved in ManLAM capping could have structural motifs similar to those reported in GTs that use dolichylphosphorylglycose as sugar donors in eukaryotes (21). These eukaryotic GTs have been classified as members of a superfamily of integral membrane GTs (GT-C), that have modified DxD signatures in the first extracellular loop (22). Iterative searches of sequence data bases, motif extraction, structural comparison, and analysis of completely sequenced genomes indicate that members of the GT-C superfamily have limited phylogenetic distribution in that representatives were found in all sequenced eukaryotic genomes, were absent from archaea, and were rare in other prokaryotes with the exception of mycobacteria (22). Approximately 40 putative members of the GT-C superfamily were identified in the genomes of M. tuberculosis H 37 Rv, M. tuberculosis CDC1551, M. leprae, and M. smegmatis (22). Of these putative GTs 16 were identified in M. tuberculosis H 37 Rv; these included putative arabinosyltransferases, the mycobacterial Emb proteins, and Rv1002c, which was subsequently shown to catalyze the initial Manp transfer in mycobacterial protein mannosylation (23). It was hypothesized that the mannosyltransferases (MTs) involved in capping of ManLAM would be among these 16 putative GTs and would have orthologs in M. leprae, M. avium, and M. bovis but would not have an ortholog in M. smegmatis. Based on this hypothesis MT1671 from M. tuberculosis was identified as a gene encoding a potential GT involved in LAM capping. Therefore, LAM was isolated and structurally characterized from a transposition mutant in which the gene encoding the protein was disrupted and from a recombinant strain of M. smegmatis that expresses the protein.

EXPERIMENTAL PROCEDURES
Materials-All chemical reagents were of the highest grade from Sigma unless otherwise specified. M. smegmatis mc 2 155 was obtained from the American Type Culture Collection. M. tuberculosis CDC1551 (⌬MT1671) strain was obtained by inactivation of the MT1671 gene in a M. tuberculosis ⌬SigF background by transposition (24,25); these strains along with the parental strain, M. tuberculosis CDC1551, were generously provided by Dr. W. R. Bishai (Center for Tuberculosis Research, The Johns Hopkins University School of Medicine, Baltimore, MD). The insertion of the kanamycin cassette in MT1671 gene was checked by PCR (using primers MT1671f (5Ј-ATG TTG CTG TGC AAG GCT-3Ј) and MT1671r (5Ј-TTA CCG CGT TGA CTT GAC-3Ј) specific to the MT1671 gene; TSPf (5Ј-CGC TTC CTC GTG CTT TAC GGT ATC G-3Ј) and TPSr (5Ј-CCC GAA AAG TGC CAC CTA AAT TGT AAG CG-3Ј) specific to the transposon).
Construction of Recombinant M. smegmatis mc 2 155-The 1671-bp open reading frame of MT1671 from M. tuberculosis CDC1551 was amplified from genomic DNA by PCR using primers (MT1671f (5Ј-TTT TTT CAT ATG CAT GCG AGT CGT CCC G-3Ј) and MT1671r (5Ј-TTT TTT AAG CTT ACC GCG TTG ACT TGA CCA C-3Ј)) engineered to include NdeI and HindIII restriction sites (underlined), respectively. The PCR product was cloned into the vector pGEM (Promega) for sequence confirmation and subsequently ligated into pVV16 derived from pMV261 (26) after digestion with NdeI and HindIII to create plasmid pVV16-MT1671. M. smegma- tis mc 2 155 was then transformed with pVV16 or pVV16-MT1671 by electroporation.
Extraction of LAM and LM-A quick method was used when LAM and LM were extracted from small samples of mycobacteria (50 mg) as described previously (27). Briefly, a mixture of chloroform/methanol/water (10:10:3) was added to the cell pellet and incubated 30 min at 55°C. The sample was centrifuged, and the organic solvent was removed. Water and phenol saturated with phosphate-buffered saline (1:1) were added to the pellet and then incubated at 80°C for 2 h. Chloroform was added, and the sample was centrifuged. The supernatant containing LAM and LM was dialyzed against water overnight, and the LAM and LM were analyzed by SDS-PAGE.
When LAM and LM were extracted from large quantities of M. tuberculosis for structural analysis the cell pellets from 5liter cultures were delipidated by serial extractions of 2:1 chloroform/methanol and 10:10:3 chloroform/methanol/water. Subsequently, LAM and LM were extracted essentially as described (8). Wet, delipidated cells were resuspended in breaking buffer (phosphate-buffered saline containing 8% Triton X-114 (Sigma), pepstatin, phenylmethylsulfonyl fluoride, leupeptin, DNase, and RNase). Cells were then disrupted using a French pressure cell. The resulting suspension was centrifuged at 2000 ϫ g for 10 min, and the pellet was discarded. The supernatant was rocked overnight at 4°C and then centrifuged at 27,000 ϫ g for 15 min. The resulting 27,000 ϫ g pellet was resuspended in breaking buffer and centrifuged as before. The combined supernatant was placed at 37°C to generate a biphase and centrifuged at 27,000 ϫ g for 15 min at room temperature. The aqueous layer and the detergent layer were back extracted twice, and 9 volumes of cold 95% ethanol was added to the combined detergent layers and incubated at Ϫ20°C overnight. The ethanol precipitate was collected, dried, resuspended in water at 50 mg/ml, digested with 1 mg/ml Pronase (Roche Applied Science), and dialyzed against water.
Extraction of LAM and LM from large quantities of M. smegmatis for structural analysis was done as described previously (28). Cultures (10 liters) in late log phase were harvested by centrifugation. The cells were delipidated using organic solvents (29) and subjected to several freeze-thaw cycles before mechanical disruption by sonication. The resulting suspension was refluxed in 50% ethanol three times. The extracts were pooled and debris removed by centrifugation. The solvent was evaporated, and the sample was resuspended in water at ϳ50 mg/ml and digested with 1 mg/ml Proteinase K (Invitrogen). After dialysis the LAM/LM fractions from either M. tuberculosis CDC1551 or M. smegmatis were further purified by size fractionation and analyzed by SDS-PAGE and Western or lectin blotting.
Size Fractionation, SDS-PAGE, and Blotting-HPLC size fractionation was performed on a Rainin SD 200 series LC system fitted with a Sephacryl S-200 HiPrep 16/60 column in tandem with a HiPrep 16/60 Sephacryl S-100 column (Amersham Biosciences) at a flow rate of 1 ml/min (29). SDS-PAGE (6% stacking gel and 15% resolving gel) followed by periodic acid-Schiff base (PAS) staining (9) was used to monitor the elution profile of the fractions containing LAM and LM, which were then pooled and dialyzed. Pooled fractions were re-analyzed by SDS-PAGE to check for purity prior to detailed analysis. Samples were also analyzed by Western blot using monoclonal antibody CS-35 (provided by National Institutes of Health/NIAID contract AI-25469) or lectin blot with concanavalin A (ConA) conjugated to peroxidase (Sigma). LAM and LM were electroeluted from 15% SDS-PAGE to Protran nitrocellulose membranes (Whatman Schleicher and Schuell Bioscience), which were then blocked and incubated with CS-35 (9) or conjugated ConA. In the case of the Western blots, membranes were incubated with the secondary antibody (anti-mouse IgG coupled to alkaline phosphatase), and color was developed using 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (BCIP/ NBT) substrate (Sigma). In the case of the lectin blots, compounds were visualized using the 4-chloro-1-naphthol/3,3Јdiaminobenzidine, tetrahydrochloride (CN/DAB) substrate kit according to manufacture's instructions (Pierce).
Monosaccharide Composition and Linkage Analysis-LAM samples were hydrolyzed with 2 M trifluoroacetic acid, converted to alditol acetates, and analyzed by GC using scyllo-inositol as an internal standard (30). GC analysis of the alditol acetates was performed on an Hewlett Packard gas chromatograph model 5890 fitted with a SP 2380 column (30 m, 0.25-m film thickness, 0.25-mm inner diameter; Supelco) using a temperature gradient of 50°C for 1 min, 30°C/min to 170°C, and then 4°C/min to 260°C.
For linkage analysis, LAM was permethylated using the NaOH/dimethyl sulfoxide slurry method (31), hydrolyzed with 2 M trifluoroacetic acid, and acetylated (30). GC/MS of the partially methylated alditol acetates was carried out using a ThermoQuest Trace gas chromatograph 2000 (ThermoQuest) connected to a GCQ/Polaris MS mass detector (ThermoQuest). Sample was dissolved in chloroform prior to injection on a DB-5 fused silica capillary column (10 m, 0.18-m film thickness, and 0.18-mm inner diameter (J&W Scientific)) at an initial temperature of 50°C, which was held for 1 min. The temperature was then increased to 180°C over 20 min and then to 250°C over 8 min.
NMR Spectroscopy-Spectra were acquired after several lyophilizations in D 2 O of 4 mg/0.6 ml in 100% D 2 O. Twodimensional 1 H-13 C heteronuclear single quantum correlation spectroscopy (HSQC) NMR spectra were acquired on a Varian Inova 500-MHz NMR spectrometer using the supplied Varian pulse sequences. The HSQC data were acquired with a 7-kHz window for proton in F2 and a 15-kHz window for carbon in F1. The total recycle time was 1.65 s between transients. Adiabatic decoupling was applied to carbon during proton acquisition. Pulsed field gradients were used throughout for artifact suppression but were not used for coherence selection. The data set consisted of 1000 complex points in t2 by 256 complex points in t1 using States-TPPI (time proportional phase incrementation). Forward linear prediction was used for resolution enhancement to expand t1 to 512 complex points. A cosine-squared weighting function and zero filling were applied to both t1 and t2 prior to the Fourier transformation. The final resolution was 3.5 Hz/point in F2 and 15 Hz/point in F1.
Endoarabinanase Digestion and Analysis by HPAEC-LAM (20 g) was incubated for 16 h at 37°C with an endoarabinanase isolated from Cellulomonas gelida (3,6). An aliquot of the digestion mixtures containing both the mannan core and the released oligosaccharides was analyzed directly by Dionex analytical HPAEC performed on a Dionex liquid chromatography system fitted with a Dionex Carbopac PA-1 column. The oligosaccharides were detected with a pulse-amperometric detector (PAD-II) (Dionex). The remaining mixtures were peracetylated as described above and analyzed by MALDI-TOF mass spectrometry.
Matrix-assisted Laser Desorption Ionization-Time-of-Flight Mass Spectrometry-The peracetylated oligosaccharides (10 g/l) or the aqueous solutions of the native LAM (10 g/l) were mixed with an equal volume of matrix (2,5-dihydroxybenzoic acid dissolved in 10 mg/ml acetonitrile/water, 50:50, 0.1% trifluoroacetic acid) prior to analysis and the molecular mass was measured in negative ion mode by MALDI-TOF on a Bruker Ultraflex TOF/TOF mass spectrometer (Bruker Daltonics).
Other Techniques-Protein sequences were obtained from the National Institute for Biotechnology Information (NCBI) world wide web site. BLAST searches were performed at the NCBI site or the TB Structural Genomics Consortium web site. Amino acid sequence alignments were performed using the MultAlin (32) interface on the L'institut National de la Recherché Agronomique Chemin de Borde-Rouge-Auzeville web site. Standard molecular biology techniques were done as described previously (33).  (Fig. 2B). MT1671 is predicted to be a 60.3-kDa protein with an isoelectric point at pH 10.66. The protein is also predicted to have 10 (TMHMM 2.0, PHDhtm) or 11 (SOSUI) transmembrane helices. The program PHDhtm was utilized by Oriol et al. (21) to predict the secondary structure of GTs, which use dolichylphosphorylmonosaccharides as the donor substrate, and when used to predict the secondary structure of MT1671, a model was generated in which there were 10 transmembrane helices with the first loop on the extracellular face of the membrane. This loop contains an Asp residue and a Glu residue (DE motif) that are consistent with conserved amino acids in the first, extracellular loop of proteins in the ␣ 2/6 MT superfamily (21) and is similar to that predicted for Rv1002c, an enzyme responsible for the addition of a Manp residue to protein acceptors in M. tuberculosis H 37 Rv (23).

Identification of MT1671 as a GT Potentially Involved in
Characterization of LAM from M. tuberculosis CDC1551 with a Disrupted Copy of MT1671-Since a M. tuberculosis CDC1551 transposon mutant in which MT1671 was disrupted had already been constructed in a ⌬SigF background (24, 25), and MT1671 had 100% identity with Rv1635c, LAM from this mutant was subjected to structural analysis. LAM was purified from equal amounts of M. tuberculosis CDC1551 and ⌬MT1671 and accounted for 0.03% of dry weight in M. tuberculosis ⌬MT1671 but made up 0.3% of dry weight in M. tuberculosis CDC1551; thus, a 10-fold reduction in LAM content could be attributed to the disruption of MT1671. When LAM from each strain was analyzed by SDS-PAGE (Fig. 3A) it was apparent that ⌬MT1671 LAM migrated more rapidly than that extracted from either wild-type M. tuberculosis CDC1551 or the parental strain M. tuberculosis CDC1551 ⌬SigF (⌬SigF), indicative of a smaller size. Indeed, MALDI-TOF mass spectrometry confirmed that the average mass of the heterogenous LAM from ⌬MT1671 was smaller (m/z ϭ 22 kDa) than that from wild-type bacteria (m/z ϭ 25 kDa). The extracted LAM and LM were also analyzed by SDS-PAGE and visualized by Western or lectin blot (Fig. 3, B and C). ManLAM from M. tuberculosis CDC1551 and ⌬SigF both react to CS-35 (a monoclonal antibody that recognizes the nonreducing arabinan termini of LAM (34)) and ConA (a lectin that recognizes terminal Manp residues). The LAM isolated from ⌬MT1671 also reacted with CS-35 but did not react with ConA, even though LM isolated from both strains did react with ConA. In addition, PILAM, isolated from M. smegmatis mc 2 155 and devoid of Manp caps, did not react with ConA (Fig. 3F). These results strongly suggested that LAM from the ⌬MT1671 strain was not Manp capped.
Construction of a Recombinant Strain of M. smegmatis mc 2 155 Expressing MT1671-Preliminary results using the CDC1551 transposon mutant clearly suggested that MT1671 and its orthologs are involved in Manp capping of LAM in slow growing mycobacteria. However, since there are likely to be multiple enzymatic activities associated with LAM, capping it was hypothesized that expression of one of these proteins in a Mycobacterium sp. devoid of ManLAM could shed light on the exact function. Therefore, to elucidate the specific role of the MT1671 protein, M. smegmatis mc 2 155 was transformed with an expression plasmid harboring MT1671 from M. tuberculosis CDC1551, pVV16-MT1671, or empty vector, pVV16. Transformation with either plasmid did not alter bacterial growth in liquid culture or colony morphology relative to the wild-type strain (data not shown).
Characterization of LAM from M. smegmatis mc 2 155 Expressing MT1671-Transformation with pVV16-MT1671 reduced the amount of LAM in M. smegmatis harvested in late log-phase by about 3-fold, i.e. 0.25% of the dry weight of M. smegmatis mc 2 155 and 0.08% of M. smegmatis mc 2 155-pVV16-MT167 could be accounted for by LAM. Extracted LAM and LM were analyzed by SDS-PAGE and visualized by PAS staining, Western blot using CS-35, or lectin blot with ConA (Fig. 3,  D-F). The most striking result came from the ConA blot (Fig.  3F). The LAM isolated from M. smegmatis mc 2 155 does not react with ConA; however, it became ConA positive when M. smegmatis was transformed with MT1671. This result clearly indicated that one or more Manp residues had likely been added to the PILAM normally synthesized by the fast growing Mycobacterium sp. Consistent with the addition of extra Manp residues, the LAM of the recombinant strain mc 2 155-pVV16-MT1671 migrated more slowly than the LAM of the wild-type strain indicative of increased size. This putative size differential was confirmed by MALDI-TOF mass spectrometry. LAM isolated from the recombinant strain mc 2 155-pVV16-MT1671 (m/z ϭ 22 kDa) was significantly larger than LAM isolated from the wild-type strain (m/z ϭ 19.5 kDa); therefore, the average mass of the heterogeneous population of LAM molecules had increased by 2.5 kDa, a value that was consistent with the 3-kDa decrease in size observed for the MT1671 disruption mutant, suggesting that as many as 18 -20 glycosyl residues could be involved. This number is consistent with the facts that LAM exists in a mixed population of linear and branched molecules, and there can be up to three Manp residues capping multiple terminal ␤-D-Araf residues per molecule (Fig. 1) as well as published estimates of the number of Manp residues capping LAM from M. tuberculosis Erdman (35), making a 2-3-kDa change in mass due to Manp capping plausible. Thus, all of the results suggest that the proteins under study are involved in capping of LAM. However, the data did not shed light on where the glycosylation occurred or how many glycosyl residues were added.
NMR Analysis of LAM Variants-Purified LAM samples were analyzed by two-dimensional 1 H-13 C HSQC experiments; resonances for anomeric regions were assigned by referring to a body of published NMR data on the structure of LAM (2,13,36,37) and endoarabinanase-generated fragments (38). Initial analysis of the heterogeneous population of LAM molecules from wild-type CDC1551 showed no marked discrepancies from published NMR spectra of LAM from M. tuberculosis H 37 Rv (37), M. tuberculosis CSU20 (37), M. leprae (37), M. bovis BCG (2), or M. kansasii (5). Therefore, chemical shifts for Man-LAM isolated from CDC1551 were used as a standard for simplification in presenting the data.
Based on the literature referred to above, evidence for the presence or absence of Manp capping was sought from the 13 C resonance at ␦ 101.1 ppm, correlating to the anomeric protons at ␦ 5.17 of 2-␣-Manp from the Manp caps and anomeric protons at ␦ 5.14 for the core 2,6-␣-Manp residues. In addition, overlapping anomeric carbon signals at ␦ 105.1 ppm correlated to proton signals at ␦ 5.07 ppm were assigned to terminal-␣-Manp (t-␣-Manp) residues belonging to both the Manp caps and mannan core. The glycosyl residue composition and peak volumes are summarized in Fig. 4.
The volume of the cross-peaks for 2-␣-Manp and t-␣-Manp generated from LAM samples isolated from ⌬MT1671 were markedly reduced, 9.5-and 5.9-fold respectively, relative to those from wildtype CDC1551 LAM. These results suggested that the extent of Manp capping had been greatly reduced or eliminated. Conversely, the volume of the cross-peaks for 2-␣-Manp and t-␣-Manp generated from LAM samples isolated from M. smegmatis expressing ⌴⌻1671 were somewhat increased, 1.4-and 1.3-fold, respectively, relative to those from wild-type M. smegmatis. Thus, it appeared that there was a gain of function, Manp capping, in M. smegmatis that was transformed with pVV16-MT1671. Unfortunately, the exact degree of Manp capping cannot unambiguously be determined from these experiments due to the strong overlapping of the resonances from the 2-␣-Manp and 2,6-␣-Manp residues as well as resonances from the t-␣-Manp residues located in the mannan core with those on the cap.
The NMR spectra show two additional cross-peaks that appear to be affected by the deletion of MT1671. One of these peaks, with 13 C resonance at ␦ 105.2 ppm, correlating to the anomeric protons at ␦ 5.42, is characteristic of 5-deoxy-5-methyl-5-thio-␣-xylofuranose (MTX (37,38)). The fact that this peak is not present in the NMR spectrum shown in Fig. 4B is not surprising, in that this compound has unambiguously been shown to be linked to a capping Manp in LAM from M. tuberculosis (38). Thus, if MT1671 is involved in Manp capping, and LAM from ⌬MT1671 is either devoid of or greatly reduced in Manp capping, a reduction in MTX would also be expected. Another cross-peak with 13 C resonance at ␦ 103.6 ppm, correlating to the anomeric protons at ␦ 5.31 ppm in Fig. 4A is also missing from the spectrum shown in Fig. 4B. This signal is unassigned, but it is possible that it is associated with an as yet undescribed modification associated with the caps of ManLAM.
Linkage and Endoarabinanase Analysis-Linkage of the sugars in the LAM samples was determined by GC/MS analysis of alditol acetates derived from partially per-O-methylated LAM from the various strains (data not shown). Results corroborated the NMR results in that the quantity of 2-Manp was reduced to the limits of detection in ⌬MT1671-LAM. In addition, the t-Manp was also slightly reduced from an abundance of 17.8% to 14.6% in the ⌬MT1671-LAM. The only observable difference in the linkage analyses of LAM isolated from the recombinant M. smegmatis strain, and the wild type was an increase in the relative abundance of t-Manp residues from 6.9 to 15.6%.
To further analyze the extent of Manp capping, LAM isolated from the bacterial strains was digested by endoarabinanase from C. gelida (3,6,37), and the products were separated by Dionex HPAEC (Fig. 5), and digested material was also peracetylated and analyzed by MALDI-TOF mass spectrometry to confirm the structures shown in Fig. 5. The HPAEC profiles for M. tuberculosis CDC1551 (Fig. 5A) (8,37). However, the profile of the digested ⌬MT1671-LAM (Fig. 5B) most closely resembles the LAM of M. smegmatis mc 2 155 (Fig. 5C), which has no Manp capping (the small amount of inositol phosphate-capped material is not Function of MT1671-Despite the lack of a formal direct enzymatic assay, the data presented here strongly indicate that MT1671 and its orthologs are mannosyltransferases and, more specifically, catalyze the addition of the first Manp residue of the cap structure to the t-␤-Araf of the arabinan of both Ara 4 and Ara 6 motifs in LAM. Thus, the prediction that MT1671 was a putative GT and member of the GT-C superfamily (22) appears to be correct. In the absence of a direct enzymatic assay and experimental information regarding the topological orientation in the membrane, the nature of the Manp donor is unknown. However, given the classification in the GT-C superfamily of GTs it seems likely that the Manp donor is decaprenylphosphorylmannose. To facilitate a direct enzymatic assay a ␤-D-Araf-(132)-␣-D-Araf-(135)-␣-D-Araf-(135)-␣-D-Araf-aglycon acceptor molecule is currently being synthesized. With this molecule in hand it should be possible to unambiguously characterize the activity of this enzyme.
Since M. smegmatis mc 2 155 transformed with pVV16-MT1671 produces a LAM that has the terminal Araf residues capped with a single Manp residue (similar to that found in M. avium (8)) it is clear that MT1671 is only involved in the first step of the Manp capping. Therefore, at least one more enzyme is required to complete the Manp caps of ManLAM in M. tuberculosis. This protein must be a ␣132 mannosyltransferase, and it should be possible to identify using a similar approach to the one taken in this manuscript. The biological significance of the second or third Manp residue of the cap of ManLAM, or for that matter the cap itself has not FIGURE 4. Comparative partial two-dimensional NMR spectra of LAM. The NMR 1 H-13 C HSQC spectra of LAM from M. tuberculosis CDC1551 (A) and M. tuberculosis CDC1515-⌬MT1671 (B) were acquired in D 2 O. Only the expanded anomeric regions are shown. The relative intensities of the peak volumes were measured, and the data are presented in C. Signal volumes were integrated and normalized to the sum of the overlapping signals from t-␤-araf and 5-␤-araf. The peaks annotated with a question mark have not been assigned. MTX indicates a cross-peak characteristic of 5-deoxy-5-methyl-5-thio-␣-xylofuranose. *, the 6-␣-Manp peak in B is very weak due to the small quantity of LAM-⌬MT1671 available for analysis. NF, not found.
been defined in precise molecular terms. An early hypothesis linking Manp cap to virulence was weakened by the observation that LAM from all strains of M. tuberculosis and M. bovis, including attenuated ones, studied were Manp capped regardless of virulence status (3,7). Recently, several groups have utilized LAM isolated from organisms with varying types of capping with, and without, chemical and or enzymatic modifications in efforts to define the importance of the Manp cap in mycobacterial pathogenesis (16,39,40). These experiments led to the suggestions that: 1) the ultimate response of the host to the pathogen may depend on the LM or LAM in the mycobacterial envelope (40); 2) the caps of ManLAM are likely not necessary to modulate the induction of IL-12 and apotosis (39); and 3) that DC-SIGN may act as a pattern recognition receptor and discriminate between mycobacterium species through selective recognition of the Manp caps on LAM molecules (16). In addition, it has been shown that the selective removal of Manp caps from ManLAM completely abrogates the ability of LAM to bind to phagocyte mannose receptors (41). Thus, it is likely that the presence of one, two, or three Manp residues in the cap has as profound biological consequences in vivo as they do in vitro. To date it has not been possible to test this hypothesis, but with the identification of the function of MT1671 we now have the ability to generate slow growing Mycobacterium strains lacking Manp caps and fast growing strains with Manp caps, which can be used in detailed immunological studies aimed at defining the precise role of Man-LAM caps in the pathogenesis of M. tuberculosis in vivo.