Structural Differences in Lipomannans from Pathogenic and Nonpathogenic Mycobacteria That Impact CD1b-restricted T Cell Responses*

Mannosylated molecules on the Mycobacterium tuberculosis surface are important determinants in the immunopathogenesis of tuberculosis. To date, much attention has been paid to mannose-capped lipoarabinomannan, which mediates phagocytosis and intracellular trafficking of M. tuberculosis by engaging the macrophage mannose receptor and subsequently binds to intracellular CD1b molecules for presentation to T cells. Another important mannosylated lipoglycan on the M. tuberculosis surface is lipomannan (LM). Comparative structural detail of the LMs from virulent and avirulent strains is limited as is knowledge regarding their differential capacity to be recognized by the adaptive immune response. Here, we purified LM from the avirulent M. smegmatis and the virulent M. tuberculosis H37Rv, performed a comparative structural biochemical analysis, and addressed their ability to stimulate CD1b-restricted T cell clones. We found that M. tuberculosis H37Rv produces a large neutral LM (TB-LM); in contrast, M. smegmatis produces a smaller linear acidic LM (SmegLM) with a high succinate content. Correspondingly, TB-LM was not as efficiently presented to CD1b-restricted T cells as SmegLM. Thus, here we correlate the structure-function relationships for LMs with CD1b-restricted T cell responses and provide evidence that the structural features of TB-LM contribute to its diminished T cell responsiveness.

The Mycobacterium tuberculosis surface is particularly rich in mannose-containing biomolecules, including mannosecapped lipoarabinomannan (ManLAM), 2 the related lipoman-nan (LM), phosphatidyl-myo-inositol mannosides (PIMs), arabinomannan, mannan, and mannoglycoproteins (reviewed in Ref. 1). Standard laboratory strains of virulent M. tuberculosis (i.e. H 37 R v and Erdman) contain ManLAM and more higher order PIMs in their cell wall than nonpathogenic mycobacterial strains (2)(3)(4). ManLAM, LM, and PIMs deserve special attention because they have been shown to directly influence the immune response to M. tuberculosis infection (reviewed in Ref. 5). ManLAM and higher order PIMs direct mycobacterial phagosome formation via their association with the mannose receptor (MR) thereby defining a unique pathway exploited by M. tuberculosis to survive within the host (reviewed in Ref. 5). Conversely, lower order PIMs do not associate with the MR but enhance early endosomal fusion with the mycobacterial phagosome, a process thought to aid in M. tuberculosis intracellular survival (reviewed in Ref. 5).
Mycobacterial LM is an extremely heterogeneous population of lipoglycan molecules with two well defined domains, a mannosyl-phosphatidyl-myo-inositol (MPI) anchor and a mannan polymer. The intrinsic heterogeneity of the population resides in the length and branching of the mannan polymer, the number of fatty acids present in the MPI anchor, and the presence or absence of other undetermined acyl groups. Specifically, mannan structure of LM consists of a linear ␣-(136)-linked mannopyranosyl backbone that is linked to position 6 of the inositol of its MPI anchor. LMs are considered multimannosylated forms of PIMs because both types of molecules share a common MPI anchor (6). Differences in the degree of acylation and nature of the fatty acids present in the LM MPI anchor have been studied in detail in a few mycobacterial species (reviewed in Ref. 6). Palmitic and tuberculostearic acids are always present in the C-1-and C-2-position of the glycerol. In addition, a third and fourth fatty acid of variable nature may be present in the C-6-position of the (132)-linked mannose and the C-3-position of the myo-inositol (6), respectively.
Structural differences in LM populations among some pathogenic mycobacterial strains exist (6). Overall, LM molecules have been shown to regulate cytokine, oxidant, and T cell responses (1). LM populations from M. tuberculosis H 37 R v associate with DC-SIGN but not with the MR and induce apoptosis and IL-12 production (1). BCG LM populations induce a pro-inflammatory response through TLR2 depending on the * This work was supported, in whole or in part, by National Institutes of Health Grants AI052458 and AI33004 (to L. S. S.), AR040312 (to R. L. M. and P. A. S.), and AI073856. This work was also supported by a Parker B. Francis Fellowship (to J. B. T.). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Fig. S1. 1  number of fatty acids per molecule (1). To what extent structural differences in a particular LM molecule dictate its immune effects is still unknown. Several studies (7,8) have shown that ManLAM molecules and their structural variants can be recognized by T cells in the context of CD1, although the precise motifs involved in this recognition have yet to be delineated. Early studies showed that a CD4/CD8 double-negative T cell clone (LDN4) derived from a skin lesion of a leprosy patient recognized ManLAM from Mycobacterium leprae but not ManLAM from M. tuberculosis H 37 R v or M. tuberculosis Erdman, in the context of human CD1b molecules (7). Conversely, LMs from all of the mycobacterial species tested were found to be less effective in activating the T cell clone, lending support to the possibility that T cell proliferation is primarily through some component of the D-arabinan of ManLAM. In contrast, another human CD1-restricted T cell clone (BDN4), derived from the peripheral blood of a healthy donor, recognized ManLAMs from all three of the mycobacterial strain/species noted above, as well as LM and higher order but not lower order PIMs. Further studies showed that those CD1b molecules had specificity for the fatty acid components present in ManLAM, LM, and PIMs (9). ManLAM treated with mild alkali results in a decrease in its ability to suppress in vitro antigen-induced T cell proliferation, IFN-␥ activation of macrophages, and cytokine secretion highlighting the importance of the alkali-labile residues (10). To date, fatty acids and succinates are the only alkali-labile residues that have been identified in ManLAM, although other undefined acyl groups may be present (6).
Here, we characterize the LM population structural properties that correlate with CD1b-restricted T cell activation. We purified LM populations from the avirulent Mycobacterium smegmatis (SmegLM) and virulent M. tuberculosis H 37 R v (TB-LM). Detailed structural analyses of both LM populations show that SmegLM molecules are heavily acylated in their MPI anchor by fatty acids and their mannan domain by succinates. In addition, the SmegLM population has a smaller but less branched mannan domain with more arabinose residues per molecule than TB-LM. Therefore, the TB-LM population consists of a large neutral LM with relatively more branched mannan structures and fewer arabinose residues. We determined that TB-LM is a less potent activator of CD1b-restricted T cell responses as compared with SmegLM, providing evidence for a correlation between LM structure and activation of T cell responses.
Growth Conditions of M. tuberculosis Strains-M. smegmatis and M. tuberculosis H 37 R v strains were grown on 7H11 plates containing oleic acid, albumin, dextrose, and catalase enrich-ment supplement for 2 and 14 days, respectively, as described previously (3).
SmegLM, TB-LM, TB-PIM, and LepLAM Purifications-SmegLM and TB-LM purifications were performed as we reported previously (4). The pure LM populations (at 1 g/l sample buffer) were analyzed by 15% SDS-PAGE followed by periodic acid silver staining. Endotoxin levels were Ͻ18 pg of endotoxin/1 mg of sample. Pure RvPIMs and LepLAM used in this study have been previously reported (4).
MALDI-MS Analyses of LMs-Analyses by MALDI-TOF MS were carried out on a Bruker Daltonic Reflex III (Bruker Daltonic, MA) mass spectrometer. LMs (0.5 l of 10 g/l) were mixed with 0.5 l of matrix solution (2,5-dihydroxybenzoic acid (10 g/l) in a mixture of water/ethanol (1:1, v/v) and 0.1% trifluoroacetic acid) and air-dried. MALDI-TOF spectra were acquired in negative linear mode detection between 10,000 and 25,000 m/z and using a 300-ns time delay with a grid voltage of 80% of full accelerating voltage (25 kV) and guide wire voltage of 0.15% as we reported previously (3,4).
Monosaccharide, Fatty Acid, and Succinate Composition-TB-LM and SmegLM were analyzed for their monosaccharide, fatty acid, and succinate composition using proper internal standards as we described previously (4). Briefly, for monosaccharide composition, LMs were hydrolyzed with 2 M trifluoroacetic acid, converted to alditol acetates using scyllo-inositol as internal standard, and analyzed by GC. GC of alditol acetates was performed on a ThermoQuest Trace Gas Chromatograph 2000 connected to a GCQ/Polaris MS detector (ThermoQuest, Austin, TX) at an initial temperature of 50°C for 1 min, increasing to 170°C at 30°C/min and finally to 270°C at 5°C/min. Experiments were done three times in duplicate.
Methanolysis of LM samples (4 nmol) followed by trimethylsilylation-generated fatty acid methyl esters for quantification by GC/MS (heptadecanoic acid (17:0)) was used as an internal standard. Derivatives were dissolved in hexanes prior injection on a DB-5 column (10 m ϫ 0.18 mm internal diameter, 0.18-m film thickness, J&W Scientist, Folsom, CA) at an initial temperature of 60°C for 1 min, increasing to 130°C at 30°C/min and finally to 280°C at 5°C/min, as we described previously (4).
Succinates were quantified by GC/MS of the octyl succinates using 4 nmol of LM after octanolysis with 3 M acetyl chloride in 1-octanol (99.98%, 100 l) at 120°C for 30 min. The derivatives were dissolved in hexanes prior to injection on a DB-5 column at an initial temperature of 60°C held for 1 min followed by an increase of temperature to 330°C at the rate of 30°C/min, as we described previously (4).
Linkage Analysis of LMs-For linkage analyses, LMs were permethylated by the NaOH/dimethyl sulfoxide slurry method (13), and alditol acetates were prepared as described above. Quantification was performed by GC/MS using selected ion chromatograms as we described previously (3,4).
NMR Spectrometry of LMs-One-and two-dimensional P 1 H NMR and two-dimensional 1P H-P 13P C heteronuclear single quantum correlation (HSQC) NMR spectra were obtained on a Bruker 600-MHz NMR spectrometer using the supplied Bruker pulse sequences. LM samples were dissolved at 15 mg/ml in 100% DB 2B O, analytically quantified by GC analyses, and lyophilized several times prior to experiments. HSQC data were acquired with a 7-kHz window for proton in F2 and a 15-kHz window for carbon in F1 with a total cycle time of 1.65 s between transients as we previously reported (3,4). Adiabatic decoupling was performed to carbon during proton acquisition. The final resolution was 3.5 Hz/point in F2 and 15 Hz/point in F1.
T Cell Lines and Proliferation Assays-T cell lines were derived from the blood of healthy, purified protein derivativepositive donors (B8 and B2) and a leprosy patient (c100-3) as described previously (7,16). For measurement of antigen-specific proliferation, T cells (1 ϫ 10 4 ) were cultured with varying numbers (usually 1 ϫ 10 4 ) of irradiated (5000 rads) HLA-DRmatched or heterologous CD1 ϩ APCs in culture medium (0.2 ml) in the presence or absence of bacterial antigens for 3 days in microtiter wells (in triplicate) at 37°C in a 7% CO 2 incubator. Antigen concentrations were used in the range of 0.01 to 10 g/ml. Because of their high lipid content, PIMs were prepared as we described previously (17). Briefly, lyophilized PIMs were dissolved in chloroform/methanol (2:1, v/v) the amount required for each experiment was removed and dried down, and then a PIM suspension was created in RPMI 1640 medium containing 10% FCS. The PIM suspension was sonicated in a water bath to disrupt micelles and then added to the culture well. Cells were pulsed with [ 3 H]thymidine (1 Ci/well, ICN Biomedicals Inc, Costa Mesa, CA) and harvested 4 -6 h later for liquid scintillation counting. To determine the role of TLR2 in the CD1b-restricted T cell response to LM, neutralizing Abs were added 30 min prior to the addition of T cells. Cytokine release from T cells was measured by ELISA after stimulation with CD1 ϩ APCs and antigen or media for 24 h. IFN-␥ ELISA (Pharmingen, San Diego) was performed according to the manufacturer's instructions.
Statistical Analysis-Statistical analyses were performed using GraphPad Prism version 4.01 (available on line).

RESULTS
Size and Molecular Composition of LMs from M. smegmatis and M. tuberculosis-SmegLM and TB-LM populations were extracted and purified by size exclusion chromatography (3,4). Analysis by 15% SDS-PAGE showed that SmegLM is a heterogeneous population of molecules (i.e. broad, poorly defined band on the gel) that had a greater electrophoretic mobility than TB-LM suggesting a smaller size (Fig. 1A). This size difference was also observed for the phosphatidyl-myo-inositol capped LAM (PILAM) derived from M. smegmatis versus Man-LAM derived from M. tuberculosis H 37 R v (Fig. 1A), suggesting that the LM mannan domain is shorter in SmegLM. To confirm this, we analyzed the overall size of both LM populations by MALDI-TOF MS. The negative MALDI-TOF MS spectra con-firmed the larger size of the TB-LM population showing a broad spectrum of masses with the larger unresolved peak at m/z 8,996.05 consistent with a molecular mass of ϳ9.0 kDa; in contrast, similar to a previous report (18), analysis of the SmegLM population revealed a smaller average molecular mass with the larger unresolved peak at m/z 6,432.50, and thus a molecular mass of ϳ6.4 kDa. LM monosaccharide composition analysis by GC showed an Ara/Man ratio of 0.062 and 0.192 for TB-LM and SmegLM, respectively, suggesting that SmegLM has a smaller mannan domain or contains more arabinose residues (Araf). To address this question, we determined the neutral sugar composition based on one inositol/molecule of LM. TB-LM yielded a ratio of 1:21:1 (Ara/Man/inositol) compared with SmegLM with a ratio of 2:11:1 (Fig. 1B) indicating that SmegLM had both a smaller mannan domain and more Ara residues per molecule. Our results are in agreement with previous studies, where the sugar composition of TB-LM and SmegLM was reported to be 21:1 (Man/inositol) (19) and 15:1 (18), respectively. Thus, the absolute neutral sugar composition is in accordance with the pattern observed by SDS-PAGE and MALDI-TOF MS analyses in support of an overall smaller size for SmegLM.
SmegLM and TB-LM fatty acid composition was also determined by GC/MS from their methyl ester derivatives after methanolysis followed by trimethylsilylation (3). The m/z values for palmitic acid (16:0, m/z 270) and tuberculostearic acid (TBST, m/z 312) were identified in both samples. Stearic acid (18:0, m/z 298) and traces of oleic acid (18:1, m/z 296), arachidic acid (20:0), behenic acid (22:0), and lignoceric acid (24:0) were also detected in both samples. Quantification of major fatty acid methyl esters observed indicated that the overall TB-LM population contains less fatty acids than the SmegLM population (Table 1), where in 4 nmol of TB-LM and SmegLM there were 10.6 and 16.7 nmol of fatty acids, respectively. In addition, the fatty acid ratios for SmegLM versus TB-LM populations were 1.7:1 for 16:0, 2.75:1 for 18:0, 3:1 for 18:1, and 0.83:1 for TBST. Thus, a plausible explanation for these results is that the SmegLM population contains more tri-acylated and tetra-acy- lated forms when compared with the TB-LM population analyzed. Although many permutations are plausible, our results suggest that for 100 molecules of SmegLM, all of them are triacylated with 16:0, 18:0, and TBST. Of these, ϳ70 molecules could contain an additional 16:0, ϳ10 molecules an additional 18:0, and ϳ30 molecules an additional 18:1. Conversely, for every 100 molecules of TB-LM, all of them are di-acylated with 16:0 and TBST. Of these, ϳ40 molecules could contain an additional 18:0, ϳ10 molecules an additional 18:1, and ϳ20 molecules an additional TBST (Table 1). We next evaluated whether the high degree of acylation observed in SmegLM is limited to only fatty acids. Previous studies showed that this is not the case for ManLAM, where ManLAM molecules also contain succinates, malates, and glutamates (6). GC/MS analyses of the octyl ester derivatives of succinates obtained through octanolysis of intact LM led to the identification of a peak that co-eluted with authentic dioctyl succinate (4). Integration of dioctyl succinate ([M ϩ H] ϩ , m/z 343 with the corresponding fragmentation at m/z 157 and 213) with respect to neutral sugar indicated that there is an average of ϳ5 succinyl residues in SmegLM and ϳ1 succinyl residue in TB-LM. Malates and glutamates were not detected in either LM type studied. Thus, in general, the SmegLM population is significantly more acylated when compared with the TB-LM population, and succinates are a major distinguishing feature (Table 1).
Methylation Analysis of SmegLM and TB-LM-To assess the exact composition of the mannan domain of SmegLM and TB-LM populations, we performed linkage analysis of their permethylated alditol acetates by GC/MS ( Fig. 2A) (3). The bar graph in Fig. 2B reflects the linkage composition of both LMs based on areas for each peak detected by GC/MS. From these results, we concluded first that SmegLM has a smaller mannan domain than TB-LM (sum of t-Man ϩ 2,6-Man ϩ 6-Man). The observed increase in 2,6-Man in the TB-LM population indicates that it is more branched and complex than the SmegLM population. This was further corroborated by the direct relationship observed between the degree of branching and the significant increase of t-Man present in TB-LM. Second, we observed the presence of t-Ara and 5-Ara in both LMs, but SmegLM contained significantly more of these saccharides in accordance with our neutral sugar analysis. Overall, our linkage analysis results indicate a smaller and less branched form of LM in the SmegLM population when compared with the TB-LM population.
Structural Features of SmegLM and TB-LM-To further corroborate the exact chemical composition present in both LM populations, they were characterized and quantified by onedimensional 1 H and two-dimensional 1 H-13 C HSQC NMR (Fig.   3) using acetone as an internal standard. The assignment of the resonances in a one-dimensional 1 H NMR spectrum for LM was based on the presence of two intense peaks in the anomeric region (␦5.2-4.9) (supplemental Fig. S1, A and B), which were identified as t-Manp (␦5.01) and 2,6-Manp (␦5.10) (Fig. 3, A and  B). These resonances were more intense in the TB-LM population when compared with SmegLM population indicating that the TB-LM population is more branched consistent with our linkage analysis (Fig. 2B). High amounts of fatty acids in the SmegLM population (supplemental Fig. S1B) were directly attributed to the resonances between ␦1.5 and 0.6 ppm. In TB-LM, these resonances (-CH 2 at ␦1.3 ppm and -CH 3 at ␦0.85 ppm) were much smaller when compared with those  10  4  1  12  8  17  11  3  10  47  In 100 nmol b  100  40  10  120  80  170  110  30  100  470 a Values for each experiment are nanomoles of fatty acid/nmol myo-inositol for fatty acids (n ϭ 4) and nanomoles of succinates (Suc)/nmol of myo-inositol (n ϭ 2) considering a mass of 6.4 kDa for SmegLM and a mass of 9.0 kDa for TB-LM. b Estimated amounts are based on the data using 4 nmol of LM.  OCTOBER Fig. S1A) indicating that the TB-LM population contains fewer fatty acids per molecule. Overall, a direct comparison of the anomeric and fatty acid resonance regions on the 1 H NMR spectra indicates that the SmegLM population has more fatty acids and a smaller mannan domain than TB-LM (supplemental Fig. S1, A and B). Resonances overlapping in the anomeric region were better resolved by two-dimensional 1 H, 13 C HSQC NMR experiments (Fig. 3, A  and B). As highlighted in our previous reports, the backbone of the mannan domain of LM is composed of a linear ␣-(136)-Manp (i.e. 6-Manp), which can be substituted at most of its C-2 positions (becoming 2,6-Manp) by a single mannose unit (t-Manp). Linkage analysis identified the presence of t-Manp, 6-Manp, 2,6-Manp, 5-Araf, and traces of t-Araf in both LM populations. By comparing the 1 H, 13 C HSQC NMR spectra from TB-LM and SmegLM populations, and in agreement with our previous work (9, 10), the spin system for all residues present in LM could be identified and quantified ( Table 2). The SmegLM population demonstrated an overall reduction in its spin systems related to its mannan domain (Fig. 3B) when compared with the TB-LM population (Fig. 3A). The level of branching present in both LMs was determined by calculating the 2,6-␣-Manp/(2,6-␣-Manp ϩ 6-␣-Manp) ratio (10) from the H-1 signal integration values. Results showed a degree of branching of 68.7 and 82.5% for SmegLM and TB-LM populations, respectively. These data are in agreement with the values obtained by the methylation analyses (i.e. ϳ9.2% increase in branching in the TB-LM population when compared with the  Table 2. Acetone (20 nmol) was used as an internal standard.

LMs from Pathogenic and Nonpathogenic Mycobacteria
SmegLM population). A decrease (1.2-fold) in the spin system corresponding to 6-␣-Manp (␦98.83, ␦4.88) was also observed for TB-LM population. Thus, direct comparison of SmegLM with TB-LM shows that the TB-LM population has a larger mannan domain with a higher degree of branching.
To evaluate the degree of succinylation present in both LMs (as reported previously for Mycobacterium bovis Bacillus Calmette-Guérin ManLAM (11)), we identified two intense broad triplets between ␦2.95 and ␦2.77 in SmegLM population and only one in the TB-LM population (␦2.79), which were attributed to the methylene of the succinyl groups located in the C-2 position of the few 5-Araf residues present in the LM molecule (supplemental Fig. S1, A and B). These resonances as well as the resonances corresponding to the fatty acid were removed upon chemical deacylation, indicating their attachment via covalent ester bonds (9). Two-dimensional 1 H and 13 C HSQC NMR (Fig. 3, C and D, and Table 2) confirmed the presence of succinates. The SmegLM population had ϳ4.5-fold more succinates than the TB-LM population (generally consistent with the increase observed by GC/MS of the dioctyl succinates, see Table 1). These two distinct set of carbons, with four crosspeaks at ␦35.16, ␦35.39, ␦35.40, and ␦35.52 correlating with protons at ␦2.79, ␦2.89, ␦2.92, and ␦2.81 present in SmegLM, and attributed to the nonequivalent methylene groups of the succinyl derivatives, were very well separated. We interpret these data to indicate that one of the cross-peaks is situated in the arabinose residues of SmegLM, most likely in the C2-position of the 5-Araf (as reported previously for the succinates present in the arabinan of ManLAM) and the other one may be in the mannan domain.
CD1b-restricted T Cell Proliferation Induced by LM-Based on their overall structural differences, we next evaluated whether TB-LM and SmegLM populations differentially stimulate CD1b-restricted T cell responses. We hypothesized that a larger number of succinyl moieties in the SmegLM population would be important in stimulating T cell proliferation based on our previous study demonstrating CD1b-restricted T cell responses to M. leprae ManLAM (LepLAM) (4). In that study, we found that the abundant succinyl substituents (up to eight succinates per molecule) were important in enhancing T cell proliferation compared with M. tuberculosis H 37 R v ManLAM (RvLAM had only two succinates per molecule) (4). The importance of succinates in CD1b-restricted T cell activation was further demonstrated by the observation that a heavily succi-nylated ManLAM obtained from an M. tuberculosis clinical isolate was also capable of stimulating T cells and triggering their proliferation (4). Here, we studied three M. tuberculosis LAMreactive CD1b-restricted T cell clones, obtained from patients infected with either M. tuberculosis or M. leprae, in terms of their response to the two different LM populations. The three different CD1b-restricted T cell clones responded in a dose-dependent manner to increasing amounts of SmegLM and TB-LM, both in terms of proliferation and IFN-␥ production (Fig. 4A). Strikingly, despite the donor to donor variation seen in the magnitude and kinetics of the response related to the APCs, the SmegLM population was a more potent inducer of T cell activation with respect to CD1b-restricted T cell responses as compared with the TB-LM population at every dose of antigen tested. For example, SmegLM at 0.01 g/ml (equal to 1.5 nM) stimulated greater CD1-restricted T cell proliferation and activation than TB-LM at 0.1 g/ml (equal to 11 nM). Among the acyl moieties in the LMs, it is difficult to differentiate between the fatty acids in the MPI anchor and the succinyl groups because chemical deacylation will remove all acyl groups indiscriminately. Thus, to address the role of the fatty acids in the MPI anchor in CD1-restricted T cell presentation, we elected to study purified species of M. tuberculosis H 37 R v PIMs (RvPIMs) with different degrees of acylation. PIMs contain the same MPI anchor as found in LM and LAM (1) and, importantly, lack succinates. Our results (Fig. 4B) show that all PIMs failed to induce T cell proliferation regardless of the number of fatty acids present (2)(3)(4) and the length of the mannan domain present in the PIMs (2-6 mannose residues). Comparison of these molecules by their molecular mass shows that even by using more molecules of PIMs (300 -800 nM, depending on the PIM species studied) versus SmegLM and TB-LM (155 and 111 nM, respectively), they were not capable of stimulating CD1b-restricted T cell responses. These results suggest that the MPI anchor of LM (the same anchor as present in PIMs) is not directly responsible for T cell activation. Together, the results provide evidence that a common feature of LepLAM and SmegLM but not TB-LM and RvLAM is the presence of a large number of succinates. The fact that two separate CD1b-restricted T cell clones demonstrate potent proliferation in response to SmegLM and LepLAM but weaker proliferation to TB-LM (Fig. 4C) and RvLAM points toward the functional significance of the succinates. Thus, a greater acylation together with the linear and shorter mannan chain in SmegLM may result in a particular three-dimensional structure important for T cell interaction. All of these structural parameters, together with the degree of succinylation, could represent the prerequisite for the T cell responses.
One possible explanation for the differential capacity of the different LM populations to activate CD1b-restricted T cells is their ability to act as TLR2 ligands (1). However, under conditions in which the response of two CD1b-restricted T cell clones to the SmegLM population was significantly blocked in the presence of an anti-CD1b mAb, the effect of anti-TLR2 mAb was small and not significant (Fig. 5A). We next attempted to "rescue" the weaker response of the CD1b-restricted T cell clones to the TB-LM population by adding a TLR2 ligand, the M. tuberculosis 19-kDa lipopeptide, to the cultures to enhance

Anomeric resonances of TB-LM and SmegLM populations
Signal volumes were determined from 1 H-13 C HSQC NMR experiments consecutively performed using the same Bruker 600 MHz NMR spectrometer. APC activity. However, the addition of the TLR2 ligand at a concentration known to induce cell activation did not enhance the response of the CD1b-restricted T cell clones to the TB-LM population to the level induced by the SmegLM population (Fig. 5B). Therefore, the less potent T cell stimulatory activity of the TB-LM population was not due to its diminished capacity to activate TLR2 responses. We therefore conclude that the diminished capacity of the TB-LM population as compared with the SmegLM population to activate CD1b-restricted T cell responses is due to the paucity of succinate residues along with other structural motifs, which contribute to antigen uptake, processing, and/or presentation to T cells.

DISCUSSION
In this study, we purified and structurally characterized LM populations from the nonpathogenic M. smegmatis and from the virulent strain of M. tuberculosis H 37 R v . We demonstrate that the SmegLM population has a less branched and shorter mannan domain that is heavily substituted with acylated groups (i.e. succinates) and contains more arabinose residues than the TB-LM population (Fig. 6). Furthermore, we show that the glycosylphosphatidylinositol anchor of SmegLM versus TB-LM contains more fatty acids per molecule. Our neutral sugar data are in agreement with previous reports describing mycobacterial mannosyltransferase functions in the biosynthesis of LAM, LM, and PIMs (18,19). Finally, we show that the TB-LM population induces limited CD1b-restricted T cell activation; in contrast, the SmegLM population is a potent inducer of CD1brestricted T cell responses, including proliferation and IFN-␥ production.
The presence of succinates in the hydrophilic domain of mycobacterial LAM has been described previously (4,20). For LM, members of the genera Micrococci possess an acidic LM population (21) in which 10 -25% of the mannosyl residues are substituted with ester-linked succinates (22). A neutral LM Notice that SmegLM at 0.01 g/ml (1.5 nM) stimulated greater CD1-restricted T cell proliferation and IFN-␥ production compared with TB-LM at 0.1 g/ml (11 nM). B, CD1b-restricted T cell proliferation was measured in response to RvPIMs (total M. tuberculosis H 37 R v PIMs mixture); Ac 1 PIM 2 and Ac 2 PIM 2 , Ac 1 PIM 5 , and Ac 2 PIM 5 and Ac 1 PIM 6 and Ac 2 PIM 6 (triand tetra-acylated phosphatidyl-myo-inositol di-, penta-, and hexa-mannosides, respectively). LM and PIMs were used at 3 g/ml (clones B2 and B8) and at 10 g/ml (clone c100-3) with similar results. C, CD1b-restricted T cell responses to LepLAM, SmegLM, and TB-LM. LMs and LAM were used at 10 g/ml. Data shown are representative of two or more independent experiments with each T cell clone. Significant differences between data points compared (see text) are indicated with a bracket and p value. population, lacking succinyl residues, has been isolated from other members of the Micrococci genera (23) and from the Mycobacterium genera (i.e. M. bovis BCG, Mycobacterium chelonae, and Mycobacterium kansasii) (24 -26). To what extent the succinates contribute to the amphiphilic nature of LM, its membrane location, and net negative charge and how these succinates may influence the immune response to LM, specifically in CD1b Ag presentation, are currently unresolved issues.
CD1 represents a family of glycoproteins expressed on the surface of various human APCs (reviewed in Ref. 27). They are related to MHC class I molecules and are involved in the presentation of lipid-anchored antigens to T cells (27). Microbial antigens from pathogenic mycobacteria, such as trehalose monomycolates, sulfolipids, mannosyl phosphomycoketides, and PIMs are known to be presented by CD1 molecules to human T cells (28,29). These CD1b antigens differ in their structures; however, all of them are characterized by a rigid hydrophilic domain bound to two aliphatic hydrocarbon chains (30). It has been proposed that CD1b contains a hydrophobic groove that accommodates the two aliphatic carbon chains (i.e. fatty acids) placing the rigid hydrophilic domain on the CD1 protein surface allowing for Ag recognition by the T cell antigen receptor (30).
In this context, CD1 molecules are capable of presenting mycobacterial glycolipids containing fatty acids of diverse length (27,31); however the exact mechanism(s) enabling this presentation is still unknown. Previous studies indicated that the presence of at least two fatty acids is important; however, it is still debatable if the length and/or type of fatty acid is important for CD1-restricted T cell recognition. Recognition is rather attributed to the specific glycolipid hydrophilic domain (30), especially when it is composed of only one carbohydrate, where the simple difference in the orientation of a single hydroxyl group of a hexose comprising the hydrophilic domain (e.g. glucose versus mannose) can completely abrogate the glycolipid recognition by CD1b-restricted T cells (reviewed in Ref. 32). Conversely, recent studies focused on glycolipids with large chains showed that the fatty acid length is also important (31,33). However, little is still known regarding the structural determinants of short chain glycolipids and lipoglycans. In this context, several studies using mycobacterial lipoglycans support the importance of the hydrophilic domain in T cell recognition (4,7). This is the case for ManLAM and LM, where the structural complexity of their hydrophilic domain, defined as a rigid, linear, and, in some cases, large structure with multiple branches (34), allows us to speculate on additional factors (i.e. charge) to stabilize them within the CD1 groove for Ag presentation.
In this context, recent studies support two different mechanisms for Ag presentation as follows: the intracellular processing of antigens for loading onto CD1b (35) and a pH-dependent binding of the whole antigen (36). Regarding the latter, CD1bantigen binding has been found to be optimal at acidic pH, where the CD1b-ligand interaction involves denaturing of the ␣-helix structures present at the entrance of the CD1 groove (9). Thus, our results suggest that the acidic nature of the SmegLM population (i.e. ϳ5 succinates per molecule) may aid in the unfolding of the ␣-helix structures revealing the hydrophobic binding site within the groove, which favors antigen loading. Furthermore, antigenic ligands for CD1b have been shown to bind in such a way that their hydrophobic acyl domains are stabilized by nonpolar van der Waals interactions with nonpolar amino acids that line the interior of the CD1 groove (37). In the case of the SmegLM population, the high content of succinates may reduce the pH enough to protonate amino acids in the CD1 binding groove or it may generate the formation of new hydrogen bonds thereby stabilizing SmegLM molecules for presentation to T cells (38). Thus, apart from the potential importance of the fatty acids present in the MPI anchor of the LMs, succinates may directly participate in stabilizing the molecule within the CD1 binding groove allowing for the antigenic epitopes to be highly exposed to the T cell receptor. Other structural aspects of SmegLM may also contribute to its favorable spatial conformation for T cell presentation (e.g. overall size and amphiphilicity, including the degree of branching).
Importantly, our current findings in comparing the TB-LM to SmegLM populations can be related to previous work by Chatterjee and co-workers (4) with LepLAM. LepLAM is also a smaller, less complex molecule containing a shorter mannan domain and a shorter and linear arabinan domain when compared with ManLAM from M. tuberculosis H 37 R v . Importantly, LepLAM is an acidic molecule, heavily acylated with multiple succinates (4). LepLAM like SmegLM is capable of stimulating CD1b-restricted T cells (4,7). Thus, LepLAM and SmegLM share in common smaller, less complex linear carbohydrate domains, and notably, both are heavily succinylated.
M. tuberculosis ManLAM has been shown to be loaded onto CD1b via the MR, a process that appears to involve the MPI anchor, which helps to establish the proper spatial configura- tion for binding to the MR and then to CD1b (8). Our recent studies showed that the TB-LM population does not associate with the MR (2) as defined in our tissue culture assays, and this may be one explanation for why the TB-LM population has a limited capacity to stimulate CD1b-restricted T cells. It is possible that the TB-LM population simply has limited affinity for the MR. However, in one instance, even at escalating doses of antigen, the TB-LM population was not as effective as the SmegLM population in activating the CD1b-restricted T cell clone, making lower affinity for the MR a less likely possibility.
It is conceivable that the structural features of TB-LM have evolved to diminish its antigenicity and contribute to the ability of M. tuberculosis to avoid detection by the adaptive T cell response (39). Future studies are required to determine whether the mechanism by which TB-LM is a less potent stimulator of CD1b-restricted T cells is related to antigen uptake, processing, binding to CD1b, or affinity for the T cell receptor. Nevertheless, we speculate that the diminished ability of the TB-LM population as compared with the SmegLM population to activate CD1b-restricted IFN-␥ production is expected to contribute to a weaker cell-mediated immune response to the pathogen and perhaps contributes to the pathogenesis of tuberculosis.