JBC INTERFERin siRNA transfection reagent

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/jbc.M305427200 on June 26, 2003

J. Biol. Chem., Vol. 278, Issue 38, 36637-36651, September 19, 2003
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
278/38/36637    most recent
M305427200v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Guérardel, Y.
Right arrow Articles by Kremer, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Guérardel, Y.
Right arrow Articles by Kremer, L.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Lipomannan and Lipoarabinomannan from a Clinical Isolate of Mycobacterium kansasii

NOVEL STRUCTURAL FEATURES AND APOPTOSIS-INDUCING PROPERTIES*

Yann Guérardel {ddagger} §, Emmanuel Maes {ddagger}, Volker Briken ¶ ||, Frédéric Chirat {ddagger}, Yves Leroy {ddagger}, Camille Locht **, Gérard Strecker {ddagger} and Laurent Kremer ** {ddagger}{ddagger}

From the {ddagger}Laboratoire de Glycobiologie Structurale et Fonctionnelle, CNRS UMR8576, Université des Sciences et Technologies de Lille, F-59655 Villeneuve d'Ascq Cedex, France, the Department of Microbiology and Immunology, Albert Einstein College of Medicine, Bronx, New York 10461, and **Laboratoire des Mécanismes Moléculaires de la Pathogénie Microbienne, INSERM U447, IBL, Institut Pasteur de Lille, 1 Rue Pr. Calmette, BP245–59019 Lille Cedex, France

Received for publication, May 23, 2003 , and in revised form, June 25, 2003.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
Although Mycobacterium kansasii has emerged as an important pathogen frequently encountered in immunocompromised patients, little is known about the mechanisms of M. kansasii pathogenicity. Lipoarabinomannan (LAM), a major mycobacterial cell wall lipoglycan, is an important virulence factor for many mycobacteria, as it modulates the host immune response. Therefore, the detailed structures of the of M. kansasii LAM (KanLAM), as well as of its biosynthetic precursor lipomannan (KanLM), were determined in a clinical strain isolated from a human immunodeficiency virus-positive patient. Structural analyses revealed that these lipoglycans possess important differences as compared with those from other mycobacterial species. KanLAM carries a mannooligosaccharide cap but is devoid of the inositol phosphate cap present in Mycobacterium smegmatis. Characterization of the mannan core of KanLM and KanLAM demonstrated the following occurrences: 1) {alpha}1,2-oligo-mannopyranosyl side chains, contrasting with the single mannopyranosyl residues substituting the mannan core in all the other structures reported so far; and 2) 5-methylthiopentose residues that were described to substitute the arabinan moiety from Mycobacterium tuberculosis LAM. With respect to the arabinan domain of KanLAM, succinyl groups were found to substitute the C-3 position on 5-arabinofuranosyl residues, reported to be linked to the C-2 of the 3,5-arabinofuranose in Mycobacterium bovis bacillus calmette-guerin LAM. Because M. kansasii has been reported to induce apoptosis, we examined the possibility of the M. kansasii lipoglycans to induce apoptosis of THP-1 cells. Our results indicate that, in contrast to KanLAM, KanLM was a potent apoptosis-inducing factor. This work underlines the diversity of LAM structures among various pathogenic mycobacterial species and also provides evidence of LM being a potential virulence factor in M. kansasii infections by inducing apoptosis.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
Nontuberculous mycobacteria (NTM)1 are ubiquitous organisms that have been increasingly implicated in pulmonary and nonpulmonary diseases (1). The insidious and indolent nature of these diseases and the lack of well defined diagnostic criteria have long contributed to the scarcity of knowledge of pulmonary NTM infection. Mycobacterium kansasii, also called the "yellow bacillus," is one of the most frequent NTM mycobacterial pathogens isolated from clinical specimens (2). Annual rates of infection in the general population have been in the range of 0.5 to 1 per 100,000. However, significant geographical variability is observed. An increase in M. kansasii disease has been observed because the onset of the AIDS epidemic (3). In the United States, the rates of disseminated disease among the human immunodeficiency virus-infected subjects reach 138 per 100,000 (4). It has been postulated that the natural habitat of M. kansasii is water (5). M. kansasii causes pulmonary disease similar to tuberculosis in immunocompetent patients and pulmonary, extrapulmonary, or disseminated disease in patients with immunodeficiencies (6–8). However, little is known about the virulence factors of M. kansasii and the mechanisms developed by these bacteria to invade and survive within the infected host.

Lipoarabinomannans (LAMs) are important virulence factors for tuberculous mycobacteria and are complex lipoglycans restricted to the Mycobacterium genus and found in the envelope of all mycobacterial species (9). LAMs are composed of three structural domains: two homopolysaccharides, the D-mannan and the D-arabinan, constitute the carbohydrate backbone; the mannan core is terminated by the GPI anchor at its reducing end, and the arabinan domain is capped by either mannosyl (10, 11) or phosphoinositol residues (12, 13). Depending on the capping motif, LAMs can be classified into three classes. Mannosylated LAMs (ManLAM), characterized by the presence of mannosyl caps, are found in Mycobacterium tuberculosis, Mycobacterium leprae, Mycobacterium bovis BCG, and Mycobacterium avium. Phosphoinositol-capped LAMs (PILAMs), characterized by the absence of manno-oligosaccharide caps and the presence of phosphoinositol caps, have been isolated from nonpathogenic species, such as Mycobacterium smegmatis. We have recently isolated and characterized LAM from Mycobacterium chelonae, named CheLAM, that is devoid of both mannose and phosphoinositol caps, therefore representing the new family of AraLAM molecules (14).

Mycobacterial LAMs in their various forms have been implicated in a wide array of biological functions (9, 15). These include inhibition of T-cell proliferation (16, 17) and of macrophage microbicidal activity (18), neutralization of the cytotoxic oxygen free radicals (19), and evocation of CD1-restricted T-cell responses (20). ManLAM has been shown to inhibit IL-12 production by human dendritic cells in vitro and to modulate M. tuberculosis-induced macrophage apoptosis (15, 21). Furthermore, through its manno-oligosaccharide caps, ManLAM signaling was found to depend on binding to macrophage mannose receptors (22–24) and to DC-SIGN on dendritic cells (25). However, it is noteworthy that subtle differences in the LAM structures dictate various functional properties. PILAMs are pro-inflammatory molecules capable of stimulating the production of tumor necrosis factor-{alpha} and IL-12, and the inability of M. smegmatis to survive within activated macrophages correlates with the pro-inflammatory responses to PILAMs. On the other hand, ManLAMs are anti-inflammatory molecules, inhibiting the production of tumor necrosis factor-{alpha} and IL-12 by human macrophages, and the capacity of M. tuberculosis, M. bovis BCG, or M. avium to survive and multiply within macrophages is related to the anti-inflammatory effect of their ManLAMs (15).

In this study, we determined the fine structures of LM and LAM from a clinical isolate of M. kansasii and investigated their ability to modulate apoptosis in human macrophages.


   EXPERIMENTAL PROCEDURES
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
RESULTS
 DISCUSSION
 REFERENCES
 
Strain and Culture Conditions—M. kansasii PHRI 901 was isolated from a human immunodeficiency virus-positive patient. This strain was grown on Middlebrook 7H11 agar plates supplemented with 10% oleic acid/albumin/dextrose/catalase enrichment (Difco) or under shaking in liquid Sauton medium at 37 °C.

Purification of KanPIM, KanLM, and KanLAM—Extraction and purification of KanLM and KanLAM was based on the Triton X-114 phase partitioning and separation by gel filtration chromatography on a Sephacryl S-200 column as reported earlier for the purification of CheLAM (14). Purity of the eluted fractions was assessed by 13% SDS-PAGE and staining for carbohydrates (26). The endotoxin content of the purified lipoglycans was measured in a chromogenic Limulus lysate assay (BioWhittaker). The LMs and LAMs preparations contained less than 20 pg of lipopolysaccharide/10 µg of LMs or LAMs.

Phosphatidyl-myo-inositol mannosides (PIMs) were obtained from Tris deoxycholate-insoluble lipid fractions after KanLM and KanLAM purification. This material was extracted with chloroform/methanol/water (50:50:5) and applied onto a silica gel column (KG60 0.063–0.200 mm, Merck) irrigated with chloroform and then eluted with chloroform/methanol/water by increasing the methanol/water proportions. Lipoglycans, including lipo-oligosaccharides and PIMs, were eluted at a proportion of 50:50:5. PIMs were finally separated on a DEAE-cellulose (acetate form) column irrigated with chloroform/methanol/ammonium acetate by increasing proportions of methanol/ammonium acetate. Fractions containing purified PIMs were eluted with chloroform/methanol (1:2) containing 0.05 M ammonium acetate. Purity of the samples was assessed by GC/MS analysis. 1H homonuclear and 13C-1H heteronuclear HMQC NMR experiments were used to typify this fraction as PIM2.

NMR Analysis—Prior to NMR spectroscopic analysis, LM (15 mg) and LAM (5 mg) were repeatedly exchanged in 2H2O (99.97% purity, Euriso-top, CEA Saclay, France) with intermediate freeze-drying and then dissolved in 250 µlofMe2SO-d6 (Euriso-top). Chemical shifts were expressed in ppm downfield from the signal of the methyl group of Me2SO-d6 ({delta}1H/trimethylsilyl = 2.52 ppm, {delta}13C/trimethylsilyl = 40.98 ppm at 343 K). The samples were analyzed in 200 x 5-mm BMS-005-B Shigemi® tubes on a Bruker ASX-400 spectrometer (Centre d'Analyses RMN, Villeneuve d'Ascq, France) (1H, 400.33; 13C, 100.66, 31P, 162.5 MHz) equipped with a double resonance (1H/X) Broad Band Inverse Z-gradient probe head. All NMR data were recorded without sample spinning.

The one-dimensional proton 1H spectrum was measured using a 90° tipping angle for the pulse and 1.5 s as a recycle delay between each of 32 acquisitions of 2.4 s. The spectral width of 4006 Hz was collected in 16,384 complex data points. The one-dimensional 13C was recorded using a spectral width of 20,161 Hz, and 32,768 data points were collected to obtain a FID resolution of 0.6 Hz per point. The 31P spectra of both compounds were acquired with a spectral width of 16,233 Hz collected in 16,384 data points. Both experiments were recorded using a composite pulse decoupling during acquisition using the GARP sequence at the carbon or phosphorus frequency (27). An exponential transformation (line broadening factor = 5 for 13Cand3Hzfor 31P) was applied prior to processing the data points in the frequency domain.

Two-dimensional homonuclear (1H-1H) spectra (COSY, ROESY, and TOCSY) were measured using standard Bruker pulse programs. ROESY spectra were acquired with various mixing times (50, 100, 200, and 400 ms) and acquired in the States mode according to Bax and Davis (28), whereas both COSY and relayed COSY were acquired in the magnitude calculation mode. Moreover, the two-dimensional TOCSY spectrum was recorded using an MLEV-17 mixing sequence of 120 ms. The spin lock field strength corresponded to a 90° pulse width of 35 µs. The spectral width was 4000 Hz in both dimensions. 512 spectra of 4096 data points with 32 scans per t1 increment were recorded giving a spectral resolution of 0.9 Hz/point in F2 and ~8 Hz/point in F1. Heteronuclear data (1H-13C and 1H-31P) were obtained with standard Bruker pulse sequences, such as HMQC (inv4tp), HMQC-HOHAHA (inv4mltp), and HMBC (inv4lrnd). HMQC and HMQC-HOHAHA were acquired by the phase time proportional increment (TPPI) method, whereas HMBC was recorded in the magnitude mode calculation. All parameters (pulse widths, pulse powers, and delays) were optimized for each experiment.

Chemical Cleavages—To study the structure of the capping motifs, LAM was subjected to mild hydrolysis in 0.1 M trifluoroacetic acid at 80 °C for 1 h. The resulting hydrolysate was either reduced and permethylated prior to analysis by GC or fractionated by gel filtration on a P4 column (Bio-Rad) using acetic acid 0.5% (v/v) as the elution buffer. Elution was monitored by TLC analysis, and the oligosaccharidyl-containing fractions were subjected to 2-aminopyridine derivatization.

The nature of the residue substituting the mannan core was investigated by subjecting 1 mg of LM to mild acetolysis in a mixture of acetic anhydride/acetic acid/sulfuric acid (10:10:1) at 40 °C for 3 h. The reaction was quenched by the addition of water. Acetolysis products were then extracted twice with chloroform before analysis by MALDI and by GC/MS.

Gas Chromatography Techniques—Monosaccharides were analyzed as alditol-acetate derivatives. Lipoglycans were hydrolyzed in 4 M trifluoroacetic acid for 4 h at 100 °C and reduced with NaBH4 in 0.05 M NH4OH for 4 h. The reduction was stopped by dropwise addition of acetic acid until the pH reached 6.0, and borate salts were co-distilled by repetitive evaporation in dry methanol. Per-acetylation was performed in acetic anhydride at 100 °C for 2 h, and the derivatives were analyzed by GC on a BPX70 12-m x 0.22-mm inner diameter column (chrompack). Linkage analysis of monosaccharides was carried out by two steps of per-methylation (29), followed by derivatization with acetyl groups with acetic anhydride. Oligosaccharides generated by either mild hydrolysis or acetolysis were reduced using NaBH4 in 0.05 M NH4OH for 4 h and per-methylated prior analysis by GC/MS on a WCOT fused silica 30-m x 0.25-mm inner diameter column (chrompack).

Acylglycerols were released by cleaving the phosphodiester bond by acetolysis. Briefly, 100 µg of lipoglycans were treated with 400 µl of anhydrous acetic acid/acetic anhydride, 3:2 (v/v), at 110 °C for 12 h. Acetylated acylglycerols were extracted by cyclohexane and analyzed by GC/MS on a WCOT-fused silica 30-m x 0.25-mm inner diameter column (chrompack).

2-Aminopyridine Derivatization—Oligosaccharides were dissolved in 20 µl of a 3 M 2-aminopyridine solution in acetic acid and incubated at 90 °C for 1 h. 70 µl of a 26 M borane solution was freshly prepared in acetic acid/water (8:5; v/v) and added before incubation of the mixture at 80 °C for 35 min. The reaction was stopped by increasing the pH to 9 with ammonia, and 500 µl of water was added. Excess of 2-aminopyridine was removed by 10 successive extractions with chloroform. Derivatized oligosaccharides were further purified on a graphite column, and the resulting tagged oligosaccharides were separated by reverse phase HPLC on a 5 µm ODS Zorbax column (4.4 x 250, DuPont Institute, Paris) using a solution of water/acetonitrile (100:0; v/v for 10 min and then 100:0 to 80:20 in 40 min). These components were then subjected to ES/MS and NMR analyses.

Matrix-assisted Laser Desorption Ionization-Time of Flight MS—The molecular mass of the per-acetylated oligosaccharides was measured by MALDI-MS on a Vision 2000 time-of-flight instrument (Finnigan Mat) equipped with a 337-nm UV laser; 1 µl of sample at a concentration of 100 pmol/µl was mixed with an equal volume of matrix (2,5-dihydroxy-benzoic acid dissolved in methanol/water, 70:30) and allowed to crystallize.

Electrospray Mass Spectrometry—All MS measurements were carried out in positive ion mode on a triple quadrupole instrument (Micromass Ltd., Altrincham, UK) fitted with an atmospheric pressure ionization electrospray source. A mixture of polypropylene glycol was used to calibrate the quadrupole mass spectrometer. The samples were dissolved in methanol/water (50:50) at a concentration of 10 pmol/µl and infused using the nanoflow probe at 50 nl/min. Quadrupole was scanned from 200 to 2000 Da with a scan duration of 3 s and a scan delay of 0.1 s. The samples were sprayed using 1.4-kV needle voltage, and the declustering (cone) was typically set at 70 V. For collision-induced dissociation experiments, the pressure of argon in the cell was set at 4.10–3 mbar, and the collision energy was set to values ranging from 25 to 75 eV.

Apoptosis Assay—The human promonocytic THP-1 cell line was grown in DMEM (Invitrogen) supplemented with 10% fetal calf serum, 2mM L-glutamine, and 2 x 10–5 M {beta}-mercaptoethanol in an atmosphere of 10% CO2 at 37 °C. The cells were distributed in 24-well plates at 1 x 106 cells/well, and phorbol 12-myristate 13-acetate (PMA; Sigma) was added (20 ng/ml final) for 20 h to differentiate the THP-1 cells into adherent macrophages. The cells were then washed once with phosphate-buffered saline, and 500 µl of medium containing 5% human serum was added to each well. The purified glycoconjugates were added at the following final concentrations to duplicate wells: KanLM (10 µg/ml), KanLAM (20 µg/ml), KanPIM2 (10 µg/ml), and PILAM from a fast-growing Mycobacterium sp. (20 µg/ml). After incubation for 24 h, the apoptosis induced by the lipids was quantified by staining the cells with annexin V-Alexa 488 conjugates and propidium iodide (Molecular Probes), as described by the manufacturer. Flow cytometry analysis was performed, and the apoptotic cells were defined as the annexin V-positive and propidium iodide-negative population. Experiments were performed three times in duplicates.

Western Blot Analysis—Immunoblotting was performed as described previously (14), except that the membrane was probed overnight with the antigenic factor 1 serum from the Candida Check (1:100 dilution; Iatron Laboratories, Japan) to test for the presence of Man{alpha}1–2-linked side chains. After washing, the membrane was subsequently incubated with anti-rabbit antibodies conjugated to alkaline phosphatase (1:5000 dilution; Roche Applied Science).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
DISCUSSION
 REFERENCES
 
Considering the biosynthetic filiation between LM and LAM, and in order to ease attribution of the NMR signals, both lipoglycans were studied simultaneously. Due to the greater simplicity of LM, the structure of the mannan and the GPI domains was deciphered on LM (termed KanLM), whereas the structure of the arabinan domain and the cap was studied separately on LAM (termed KanLAM).

Structural Analysis of the Mannan Domain
Quantitative analysis of the alditol-acetate derivatives from KanLM and KanLAM led to Ara/Man/myo-inositol molar ratios of 4:38:1 and 65:56:1, respectively. Detection of arabinose residues in LM, which is generally devoid of arabinose, suggested a slight contamination of KanLM in the KanLAM preparation. These arabinose units were identified as t-Ara, 2-Ara, 5-Ara, and 3,5-Ara by methylation analysis. Subsequent NMR analyses showed that the arabinose units were identical in KanLM and KanLAM samples, indicating that these residues detected in the KanLM sample, estimated to about 4% (mol/mol), arose from a contamination by KanLAM. Analysis of partially methylated alditol-acetate derivatives revealed the presence of three major types of mannose units in KanLM, identified by their retention times and fragmentation patterns as t-Manp, 6-Manp, and 2,6-Manp. Small amounts of 2-Manp were also identified in the KanLM sample. The relative proportions of this residue (Table I) indicated that it originated from KanLM and not from a KanLAM contamination.


View this table:
[in this window]
[in a new window]
  		TABLE I
Methylation analysis of KanLM and KanLAM

The values of the molar ratio were obtained from peak area integration of the GC analysis and corrected by the use of effective carbon-response factors.

 

The exact arrangement of the monosaccharide units was further studied by NMR. The parameters (summarized in Table II) of the mannan domain from KanLM were assigned using one-dimensional 1H and 13C, as well as two-dimensional 1H-1H homonuclear and two-dimensional 1H-13C heteronuclear NMR. Based on these parameters, spin systems of major t-{alpha}-Manp, 6-{alpha}-Manp, and 2,6-{alpha}-Manp units were unambiguously identified and labeled as the IV, VI, and VIII units, respectively (30). Due to a strong overlapping of 2,6-{alpha}-Manp and 2-{alpha}-Manp signals, parameters from 2-{alpha}-Manp could not be unambiguously established. As seen in the anomeric region of the 1H homonuclear and 13C-1H heteronuclear NMR spectra (Fig. 1b), the VI unit showed a marked heterogeneity that led to the identification of two distinct 6-{alpha}-Man residues labeled VI1 and VI2 because of their H-1 chemical shifts at {delta} 4.70 and {delta} 4.78 ppm, respectively. This suggests that both 6-{alpha}-D-Manp units present distinct arrangements within the mannan core of KanLM. HMBC experiments unambiguously demonstrated that the mannan domain of KanLM is composed of a linear backbone of {alpha}1,6-linked {alpha}-D-Manp residues, partially substituted in the C-2 position mostly by {alpha}-D-Manp residues (Fig. 2). In particular, the occurrence of an extra-residual 3JH-C correlation between IV H-1 and deshielded C-2 at {delta} 78.4 ppm established that all t-Man units (IV) substituted 2,6-Man units (VIII), or 2-Man units as suggested by the methylation analysis, in the C-2 positions. Then VIII H-1/VIII C-6, 6', and VIII H-1/VI C-6,6' correlations at {delta} 67.3 and 66.5 ppm, respectively, established that the VIII units substituted both VIII and VI units in the C-6 positions. Interestingly, H-1 of the VI1 and VI2 units showed distinct 3JH-C correlations with VIII C-6 at {delta} 67.0 ppm and VI C-6 at {delta} 66.5 ppm, respectively. These distinct positions within the mannan backbone may in part account for the differences observed in the VI1 and VI2 H-1 chemical shifts. Furthermore, the slight differences of the chemical shifts found between C-6 positions of the VIII residues ({delta} 67.0 and 67.3) revealed the occurrence of two distinct populations of 2,6-Man residues, only differing in their C-6 parameters, VIII1 and VIII2, depending on the nature of the substituting unit in the C-6 position, 6-Man or 2,6-Man residues.


View this table:
[in this window]
[in a new window]
  		TABLE II
KanLM 1H and 13C chemical shifts in Me2SO-d6 at 343 K n.d., not determined.

 


View larger version (38K):
[in this window]
[in a new window]
  		FIG. 1.
Anomer regions ({delta}1H: 4.5–5.3, and {delta}13C, 98–110) of 13C-1H HMQC heteronuclear spectra: (a) KanPIM, (b) KanLM, and (c) KanLAM in Me2SO-d6. * indicates Ara residues.

 


View larger version (43K):
[in this window]
[in a new window]
  		FIG. 2.
Details of HMBC ({delta}1H: 4.5–5.3, and {delta}13C: 58–92) (a), 13C-1H heteronuclear HMQC ({delta}1H: 3.0–5.2, and {delta}13C 58–92) (b), and of 1H-1H ROESY ({delta}1H: 3.0–5.2 and 5.05–5.30) spectra from KanLM in Me2SO-d6(c). VIII H-1/VI C-6,6' correlation is indicated although it is not apparent on the HMBC spectrum due to the small quantities and the processing used. • indicates arabinosyl signal, and * indicates non-sugar signal.

 

As described above, the identification of small amounts of 2-Man residues in KanLM suggested the occurrence of few {alpha}-1,2 extended oligomannosyl motifs substituting the {alpha}1,6-linked mannan core. In order to investigate this issue, KanLM was submitted to acetolysis that specifically cleaves the {alpha}1,6-Man linkages. The resulting oligosaccharides were then analyzed by MALDI-MS. As shown in Fig. 3a, ions at m/z 989, 1005, 1277, and 1293, likely corresponding to [M + Na]+ and [M + K]+ adducts of per-acetylated Hex3 and Hex4 oligosaccharides, were observed. To establish whether these oligosaccharides originate from the presence of di- and trimannosidic side chains or from an incomplete acetolysis of the {alpha}(1–6) linkages, the acetolysis products were reduced, per-methylated, and analyzed by GC/MS at high temperature in the chemical ionization and EI modes. GC/chemical ionization-MS revealed the occurrence of three compounds of distinct retention times presenting [M + NH4] ions at m/z 699, which typified them as per-methylated reduced trisaccharides Hex3 (data not shown). From fragmentation patterns in EI/MS, it was demonstrated that two of them presented a reduced hexose residue substituted in the C-6 position (data not shown). This result demonstrated that these two oligosaccharides resulted from incomplete acetolysis of {alpha}(1–6) linkages of the mannan chain and presumably presented the following sequences: Man({alpha}1–2)Man({alpha}1–6)Man-ol and Man({alpha}1–6)Man({alpha}1–6)Manol. As shown in Fig. 3b, the EI/MS spectrum of the third compound showed [M-CHOMe-CH2OMe], [M-(CHOMe)2-CH2OMe], and [(CHOMe)3-CH2OMe] ions at m/z 586, 542, and 177, respectively, all characteristic of a terminal reduced hexose substituted in the C-2 position. Considering the specificity of acetolysis for the {alpha}-1,6 cleavage, and in accordance with the detection of low amounts of 2-Man in KanLM, it is likely that this oligosaccharide presented the sequence Man (1–2)Man(1–2)Man(1-. In either case, the observation of a trisaccharidic acetolysis fragment presenting at its terminal reducing extremity a hexose substituted in the C-2 position strongly suggested the occurrence of a low proportion of disaccharidic side chains in the mannan domain. This hypothesis was also supported by the observation of a tetrasaccharide fragment in minute amounts, which presented a reduced hexose residue substituted in the C-6 position by a single hexose residue and in the C-2 position by two hexose residues (data not shown).



View larger version (36K):
[in this window]
[in a new window]
  		FIG. 3.
a, detail of the MALDI-MS spectrum of acetolysis products from KanLM. b and c, spectra in EI/MS and fragmentation pattern of reduced, per-methylated tri-saccharide released by acetolysis of KanLM. Ions at m/z 177, 542, and 586 indicate that the Hex-ol residue is substituted in the C-2 position. Ion at m/z 500 is characteristic of a J1 type fragmentation that occurred in the terminal non-reducing position. As reported by others (56), the J1 fragment denotes the presence of a D-hexopyranosyl residue methylated in the C-3 position. d, immunoblotting of various mycobacterial LMs for the presence of Man{alpha}1,2-linked side chains. Left panel, ~0.3 µg of purified LM from M. kansasii, M. chelonae, and M. smegmatis were subjected to SDS-PAGE prior to treatment with periodic acid and staining with silver nitrate. Right panel, immunoblot analysis using the Candida factor 1 antiserum.

 

We next used commercial anti-Candida oligo-mannosylated determinants antisera (Candida Check, Iatron Laboratories, Tokyo, Japan) to directly detect the presence of Man(1–2)Man(1- side chains on the mannan core of KanLM. Immunoblotting experiments (Fig. 3d) showed that the antigenic factor 1, specific for Man{alpha}1-(2Man [PDB] {alpha}1)1–3 determinants (31), reacted strongly with KanLM and moderately with LM from M. smegmatis, known to possess Man{alpha}1,2-linked side chains, but not with LM from M. chelonae, containing Man{alpha}1,3-linked side chains (14), consistent with the presence of Man{alpha}1,2-linked oligomannosyl motifs in KanLM. Altogether, these data arising from composition analysis, structural analysis of acetolysis fragments, and immunoblotting experiments using Man{alpha}1-(2Man [PDB] {alpha}1)1–3-specific antibodies indicate the substitution of the mannan domain of KanLM by a low number of oligomannosyl side chains, likely consisting of a Man({alpha}1–2)Man({alpha}1- sequence, along with a majority of single mannose residues.

The occurrence of such motifs in the mannan domain of KanLAM could not be definitively established due to the strong overlapping effect of the oligomannosyl cap (see below) substituting the terminal, non-reducing end of its arabinan domain.

Structural Analysis of the Phosphatidyl-myo-inositol Anchor
The structure of the GPI anchor on KanLM was first analyzed by one-dimensional 31P and two-dimensional 31P-1H HMQC NMR. One-dimensional 31P spectra revealed the presence of two intense signals at {delta} 1.6 and 1.9 ppm, tentatively attributed to phosphorus atoms of distinct GPI anchors, as confirmed by two-dimensional NMR (data not shown). 31P-1H HMQC NMR established that both phosphorus atoms were linked to inositol in position C-1, and glycerol in position C-3, in agreement with Nigou et al. (32). 1H parameters of glycerols were attributed by the 31P-1H HMQC-HOHAHA parameters, indicating that the glycerols deriving from both phosphorus atoms displayed similar parameters (Table II). Their deshielded H-2 proton typified them as 1,2-di-acylated glycerols (33). In contrast, the 1H NMR parameters of the two myo-inositols showed slightly different values. In particular, their H-3 chemical shifts differed by 1.6 ppm at {delta} 71.3 and 72.9 ppm, in agreement with an acylation position in C-3 of one of the inositol residues, as described for LAMs in other mycobacterial species (14, 32, 34). These results demonstrate that the phosphorus atom at {delta} 1.6 ppm was part of a GPI anchor containing a di-acylated glycerol and a mono-acylated myo-inositol, whereas the phosphorus atom at {delta} 1.9 ppm was part of a GPI anchor containing a di-acylated glycerol and a non-acylated myo-inositol. By using the nomenclature proposed by Puzo and co-workers (35), we typified the phosphorus atom at {delta} 1.6 ppm as P1 and the phosphorus atom at {delta} 1.9 ppm as P3. Integration of both signals showed that P3 was the major form (P3/P1 ratio of 9:1).

The nature of the acyl groups associated with acylglycerol units was investigated by the release of intact acetylated acylglycerols upon acetolysis and analysis by GC/MS in the electronic impact mode. In agreement with the NMR data, this methodology confirmed that the KanLM anchor contains exclusively di-acylglycerol. Only two peaks presenting identical EI/MS spectra dominated the di-acylglycerol region of the total ion count gas chromatogram (data not shown). The nature of the substituting fatty acids was easily established by the m/z ions at 239 and 281, typifying hexadecanoic (C16) and nonadecanoic (C19) acids, respectively. This attribution was confirmed by the occurrence of fragment ions resulting from the loss of each acyl appendage: [M – C16] ion at m/z 397, [M – C19] ion at m/z 355, and [M –CH3COO] ion at m/z 592. On this basis, the two peaks were attributed to the two possible arrangements of fatty acids on the C-1 and C-2 positions of the glycerol. These experiments established that the GPI anchor of KanLM was substituted exclusively by 1/2-hexadecanoyl-1/2-non-adecanoyl-di-acylglycerol.

The glycosylation patterns of the inositols associated to P1 and P3 were defined in KanLM by NOESY (Fig. 4). Starting from the 1H signals of both inositols on the TOCSY spectrum, we observed in NOESY intense NOE contacts between H-6 Ins-P3, H-6 Ins-P1, H-2 Ins-P3, and H-2 Ins-P1 with four presumably mannose anomeric signals at {delta} 4.98, 4.99, 5.12 and 5.19 ppm, respectively. This suggests that Ins-P1 and Ins-P3 were equally glycosylated in the C-2 and the C-6 positions. From the signal at {delta} 4.98 ppm NOE contacts of decreasing intensities with Ins-P3 H-5, H-4 and H-3 were also observed, whereas only contact between the signal at {delta} 4.99 ppm and Ins-P1 H-5 could be observed due to the small quantity of Ins-P1. Similarly, low intensity NOE contacts with Ins-P3 H-1, H-3, H-4, and H-5 were observed from the signal at {delta} 5.12 and with Ins-P1 H-1 from the signal at {delta} 5.19 ppm, confirming their attachment to either Ins C-2 or C-6. The NOESY spectra also showed other contacts attributed to H-2 of each four units. The partial attribution of their spin systems from the TOCSY spectrum (H-2, -3, and -4 for units associated to Ins-P3 and H-2 and -3 for those associated to Ins-P1) confirmed that the inositol-substituting units were {alpha}-Manp, which were labeled Man1(P1) and Man1(P3) for the units linked to the C-2 position and Man2(P1) and Man2(P3) for the units linked in the C-6 position. These data are in total agreement with previous studies based on the GPI anchors of PIMs, LM, and LAM where inositol was invariably substituted at the C-2 position by a single {alpha}-D-Manp residue and at the C-6 position by the mannan core (35–37). A similar methodology was used to study the structure of the GPI anchor of KanLAM. The results from these experiments were indistinguishable from those obtained for KanLM (Table III), demonstrating that KanLM and KanLAM share identical GPI anchors.



View larger version (51K):
[in this window]
[in a new window]
  		FIG. 4.
Details of 1H-1H homonuclear TOCSY ({delta}1H: 3.0–5.3 and 3.0–3.3) (a), TOCSY ({delta}1H: 3.0–5.3 and 4.95–5.3) (b), and NOESY ({delta}1H: 3.0–5.3 and 4.95–5.3) (c) spectra of KanLM in Me2SO-d6 showing the connectivities between Man-1/Man-2 residues and H-6 and H-2 of inositol.

 

View this table:
[in this window]
[in a new window]
  		TABLE III
KanLAM 1H and 13C chemical shifts in Me2SO-d6 at 343 K n.d., not determined.

 

Structural Analysis of the Arabinan Domain
Methylation analysis revealed that the arabinan domain of KanLAM is composed of 3,5-Ara, 5-Ara, 2-Ara, and t-Ara units (Table I). By comparing the 13C-1H heteronuclear NMR spectra from KanLM and KanLAM (Fig. 1), and in agreement with previous studies (14, 34), the anomeric protons of all arabinose types could be identified and attributed by GC, and their spin systems could be determined. The attributed 1H and 13C NMR parameters from KanLAM are summarized in Table III. As observed previously for mycobacterial LAMs, some Ara units with identical spin systems presented multiple H-1 signals that led to the attribution of one 3,5-{alpha}-D-Araf unit (I), three 5-{alpha}-D-Araf units (II1 to II3), two 2-{alpha}-D-Araf units (III1 and III2), and a single t-{beta}-D-Araf unit (V). Such a multiplicity of signals was attributed to the different positions that each unit may take within the arabinan core (10). Altogether, the NMR parameters derived from the arabinan domain of KanLAM, including 3JH-C correlations determined by HMBC, showed no marked discrepancy with those derived from other mycobacterial species such as M. bovis BCG (35), M. tuberculosis H37Rv (34), M. smegmatis (13), and M. chelonae (14), suggesting that all these species share similar arabinan domains. Briefly, the extra residual correlation V H-1/III C-2 at {delta} 88.4–88.9 ppm established that {beta}-Ara units substitute 2-{alpha}-Ara units, whereas the III1 H-1/I C-5 and III2 H-1/I C-3 correlations at {delta} 67.2 and {delta} 83.1 ppm, respectively, established that 2-{alpha}-Ara units substitute 3,5-{alpha}-Ara units in both the C-3 and the C-5 positions.

Structural Analysis of the Cap
Composition analysis established that KanLAM contains more mannose residues than KanLM (see above). Methylation analysis also indicated that substantially more 2-Man residues occur in KanLAM than in KanLM (average of 7.2 and 1.5 2-Man residues per KanLAM and KanLM molecule, respectively), suggesting the presence of an additional mannan domain in KanLAM compared with KanLM. In agreement with the literature, the anomeric proton and carbon of the 2-{alpha}-D-Manp residue (VII) were attributed on the anomeric region of the 1H-13C heteronuclear spectrum at {delta} 4.85/99.6 ppm (Fig. 1). The complete NMR parameters of this residue were then determined by multirelayed COSY and TOCSY and found to perfectly match those described previously for 2-{alpha}-D-Manp residues of mannose-capped LAM from M. tuberculosis (34). A clear 3JH-C correlation was observed on the HMBC spectrum between H-1 VII and a signal at {delta} 70.5 ppm (Fig. 5) corresponding to the C-5' of {beta}-Araf V unit, which is in agreement with the presence of {alpha}-1,2-substituted oligomannosyl caps at the non-reductive extremities of the arabinan domain of KanLAM.



View larger version (51K):
[in this window]
[in a new window]
  		FIG. 5.
Details of 13C-1H heteronuclear HMQC ({delta}1H: 3.0–5.2, and {delta}13C: 58–92) (a) and HMBC ({delta}1H: 4.5–5.3, and {delta}13C: 58–92) (b), from KanLAM in Me2SO-d6.

 

To define more precisely the actual status of capping in KanLAM, the capping motifs were cleaved from the arabinan side chains by mild hydrolysis. The released oligosaccharides were reduced, per-methylated, and analyzed by GC/EI/MS. Four methylated oligosaccharides were identified in the hydrolysate (data not shown). The major species were identified by their fragmentation patterns as Hex-Pent-ol, Hex-Hex-Pent-ol, and Hex-Hex-Hex-Pent-ol, indicating that mono-, di-, and trimannosyl units substituted the arabinan domain. The last species was attributed to Hex-Hex-ol and was probably generated by the nonspecific cleavage of either the mannan core or di- and tri-mannosylated capping motifs. As reported earlier for M. tuberculosis and M. bovis BCG, the mannose cap of KanLAM consists mainly of di-mannosylated residues, whereas tri-mannosylated caps were barely observable. The EI/MS fragmentation pattern of the mono- and di-mannosyl motifs allowed us to confirm that the arabinose residue was substituted in the C-5 position by the observation of the ions resulting from the RCHOMe-RCHOMe cleavages within the arabinitol residue. However, this did not allow us to unambiguously discriminate between {alpha}1,2-linked and {alpha}1,6-linked mannose residues. At most, we observed from the Man-Man-Ara-ol motif a fragment ion at m/z 456 corresponding to MeO+ = CH-O-Man-O-Ara-ol, which indicated that the Man residue attached to the Ara-ol residue was not substituted in the C-3 position. Thus, the KanLAM hydrolysate was derivatized by 2-aminopyridine and fractionated by gel filtration in order to isolate oligomannosyl-capping motifs. The oligosaccharidic fraction was further subjected to reverse phase HPLC to purify each tagged oligosaccharide generated by mild hydrolysis. Four peaks, I to IV, were sequentially collected and subjected to electrospray mass spectrometry (ES/MS) in the positive mode and then to NMR 1H analysis. In agreement with the above GC/MS analysis, the two first peaks, I and II, consisted of Hex-Pen-2-AP and Hex-Hex-2-AP, as demonstrated in ES-MS by their ions [M + H]+ at m/z 391 and 421, respectively, and in ES-MS2 by their fragmentation ions [M-Man + H]+ at m/z 229 for Hex-Pent-2-AP and 259 for Hex-Hex-2-AP, respectively. On this basis, and in accordance with the above results, these two compounds were assigned as Man-Ara-2-AP and Man-Man-2-AP. Similarly, peak III was assigned to the oligosaccharide Man-Man-Ara-2-AP because of its [M + H]+ ion at m/z 553 in ES-MS, and its [M – Man + H]+ and [M – 2Man [PDB] + Na]+ fragment ions in ES-MS2 at m/z 391 and 229, respectively (Fig. 6a). The nature of the mannose linkages of this last compound was then investigated by two-dimensional 1H-1H NMR. As shown by the COSY-90 spectrum in D2O (Fig. 6c), this compound exhibited only two sets of NMR H-1/H-2 parameters at {delta} 5.113/3.968 ppm and {delta} 5.020/4.060 ppm attributed to 2-{alpha}-Manp and t-{alpha}-Manp, respectively, in accordance with previous studies (10). The shielded H-2 2-{alpha}-Manp chemical shift value ({delta} 3.968 ppm compared with {delta} 4.060 ppm for native molecule) was explained by the shielding effect of the aglycone residue in reducing position. This attribution was confirmed by the observation of internal and external NOE effects arising from 2-{alpha}-Manp residue on the ROESY spectrum (data not shown). A fourth minor compound, peak IV, presented a [M + H]+ ion at m/z 625 in ES-MS and generated in ES-MS2 two major fragment ions [M – Hex + H]+ and [M – 2Hex [PDB] + H]+ at m/z 463 and 301, respectively, suggesting the occurrence of a di-mannosyl motif in the non-reducing position (data not shown). In accordance with these data, the 1H NMR experiments demonstrated the presence of a Manp({alpha}1–2)Manp motif (data not shown). The [M – 2Hex [PDB] + H]+ ion at m/z 301 did not correspond to Pent-2-AP nor to Hex-2-AP, and the exact nature of the compound in the reducing position tagged by 2-AP could not be directly attributed by NMR due to the low intensity of the signal. However, the ES-MS2 analysis also generated a minor ion [M – 2Hex [PDB] + H]+ minus 90 at m/z 211 corresponding to Pent-2-AP minus H2O. This suggested the occurrence of a Pent-2-AP further modified with an unknown component. This ion at m/z 211 was also observed after ES-MS3 of fragment ion at m/z 301 (data not shown). After per-methylation, peak IV presented a mass increment of 154 Da to m/z 779, corresponding to 11 methyl groups. These were tentatively assigned as follows: 7 substituting the dimanosyl unit in the terminal position; 2 substituting amine group of aminopyridine to form a quaternary ammonium group; and 2 on the remaining hydroxyls groups of the pentose in the reducing position. In accordance with the results obtained with native oligosaccharides, the ES-MS2 fragmentation of the per-methylated oligosaccharide generated [M – Hex + H]+ and [M – 2Hex [PDB] + H]+ ions at m/z 561 and 357, as well as a very intense ion corresponding to [M – 2Hex [PDB] + H]+ minus 90 at m/z 267 (Fig. 6b). Thus, the comparison of the fragmentation patterns of native and methylated oligosaccharides established the presence of a modified pentose residue in its terminal position, characterized by the presence of an alkali stable and methylation-insensitive substituting group. The average molecular masses of the native and per-methylated modified pentose residue fit, among others, with both thio-butyl or –ether-linked O-(CH2)2-O-CH2-CH3 substituents. However, our analyses did not enable us to differentiate between these two hypotheses.



View larger version (22K):
[in this window]
[in a new window]
  		FIG. 6.
Analysis of KanLAM mild hydrolysate after gel filtration, 2-AP derivatization, and reverse phase HPLC fractionation. ES/MS-MS spectra of peak III (a) and per-methylated peak IV (b). c, detail of 1H-1H homonuclear COSY 90 in D2O of peak III.

 

Altogether, these data show that KanLAM belongs to the ManLAM class of molecules, due to the presence of {alpha}1,2-linked mono-, di-, and tri-mannosylated units at the non-reducing ends of its arabinan domain. Unexpectedly, they suggested also the occurrence of minor amounts of a modified pentose residue substituted by a {alpha}1,2-linked di-mannosylated unit, which may represent an original modification of the oligomannosyl capping. Deciphering the exact structure of this novel monosaccharide will require large scale purification of the compound.

Structural Analysis of Various Substitutions
Substitution by 5-Methylthiopentose (5-TMP) Residues— Careful analysis of the anomeric region of the 1H-13C heteronuclear spectrum of KanLM revealed the presence of an unattributed signal at {delta} 5.23–5.24/104.2 ppm (Fig. 1b). Examination of the 1H signals by one- and two-dimensional NMR disclosed the presence of two distinct signals with closely related parameters. The similarities of these parameters suggest that the two signals originated from identical components in slightly different environments. The entire spin system of this unit was established by two-dimensional multirelayed COSYs (Fig. 7) and TOCSY 1H-1H (data not shown). The chemical shifts of the corresponding carbons were attributed by the 1H-13C heteronuclear and HMBC spectra (Fig. 8) and are summarized Table II. The identification of only five ring carbons and corresponding protons revealed this residue as a pentose. Furthermore, the important downfield chemical shift of C-4 suggests a pentofuranoside unit. However, the high field chemical shift of the C-5 from about {delta} 65–75 ppm for a C-5 –CH2OH of a pentofuranose to {delta} 34.1 ppm suggested that this carbon is substituted by an element different from oxygen. The H-5/5' signals clearly appeared as a quadruplet, establishing the absence of coupling. Altogether, these data rule out the possibility of the presence of a substituting phosphorus atom. Starting from H-5/5' on the HMBC spectrum (Fig. 8), in addition to typical 3JH-C and 2JH-C correlations with C-3 and C-4, respectively, we observed a correlation with a carbon at {delta} 16.6 ppm. The comparison with standard chemical shifts from the literature revealed that CH2-{gamma} and –S-CH3 from methionine present average 1H/13C chemical shifts of {delta} 2.47/32.13 and 1.84/17 ppm in D2O. These values are in agreement with the H-5/C-5 values measured in Me2SO-d6 of the pentofuranoside unit with an additional carbon observed in HMBC, respectively. On this basis, the signal observed in HMBC at {delta} 16.6 ppm was interpreted as a 3JH-C correlation between H-5/5' and carbon atom of substituting –S-CH3 group. Altogether, these data suggest that the additional pentafuranose unit observed in LM is a 5-TMP. Furthermore, the ring carbon and proton parameters matched with those from 5-TMP, previously observed in minute amounts in M. tuberculosis (38).



View larger version (44K):
[in this window]
[in a new window]
  		FIG. 7.
1H-1H homonuclear COSY 90 (in black) and relayed R1 COSY (in red) spectra of KanLAM in Me2SO-d6. Only spin systems from succinylated arabinose (Ara) and 5-methylthiopentose (5-TMP) are assigned.

 


View larger version (24K):
[in this window]
[in a new window]
  		FIG. 8.
Details of HMBC spectrum in Me2SO-d6 starting from H-1 TMP ({delta}1H: 5–5.3, and {delta}13C: 10–90) (a) and H-5/5' TMP ({delta}1H: 2.4–2.9, and {delta}13C: 10–90) (b) and of HMQC spectrum ({delta}1H: 2.4–2.9, and {delta}13C: 20–50) (c) from KanLAM that show the entire 1H/13C spin system of 5-TMP residue.

 

As shown in Fig. 1, the anomeric proton signals of 5-TMP occurred both in KanLM and KanLAM, but not in KanPIM, suggesting that 5-TMP is linked to the mannan domain of both KanLM and KanLAM. However, due to the slight contamination of KanLM by KanLAM, we could not exclude the possibility that the 5-TMP would originate from the KanLAM contamination. A visual examination of the KanLM spectra showed that the 5-TMP signal intensity was comparable with that of the most abundant contaminating arabinose type (I) and higher than that of the other arabinose units, thus ruling out the possibility that 5-TMP would be associated with KanLAM. This was confirmed by the integration of the 1H signals on one-dimensional 1H spectra of the 5-TMP H-1 and of the well individualized Gro H-1s. Because both KanLM and KanLAM are composed of a single Gro H-1, the calculation of the ratio 5-TMP H-1/Gro H-1 established that KanLM and KanLAM contain averages of 0.9 5-TMP residue per molecule and 0.5 5-TMP residue per molecule, respectively. The lower value of KanLAM was confirmed by the calculation of the 5-TMP H-5/Gro H-1 ratio. Altogether, these data suggest that the 5-TMP is linked to KanLM and KanLAM, presumably via the mannan domain.

The HMBC spectra of 5-TMP H-1 (Figs. 2 and 8) present a set of four correlations, three of which were attributed to intra-residual 2JH-C and 3JH-C correlations with their own C-2, C-3, and C-4 at {delta} 78, 76.4, and 80 ppm, respectively. The fourth signal was assigned to an extra-residual correlation with a carbon easily identified on the 1H-13C heteronuclear spectrum because of its distinctive 13C/1H chemical shift at {delta} 75.5/3.67 ppm (Fig. 2). The linkage of 5-TMP to this carbon was confirmed by ROESY based on the observation of a NOE contact between 5-TMP H-1 and a proton at {delta} 3.67 ppm (data not shown). A second NOE contact at {delta} 4.917/4.898 ppm was attributed to its own H-2, whereas a third contact at {delta} 3.723/3.693 ppm remains unattributed. The chemical shift of the linkage carbon at {delta} 75.5 ppm rules out the possibility that the 5-TMP is linked to an Ara residue. This is consistent with our previous observations, suggesting that the 5-TMP is linked to the mannan domain of KanLM. Similarly, this carbon was much too far downfield-shifted to be a C-6 mannose-substituted, and its associated proton was much too high field-shifted to be associated with a C-2 mannose-substituted (34, 39, 40). We therefore assume that the 5-TMP unit is linked to a mannose residue, either to the C-3 or to the C-4 position. 13C/1H chemical shifts of the unsubstituted mannose C-3 and C-4 of KanLM and KanLAM in Me2SO-d6 were observed at {delta} 71.8/3.58 and 68.2/3.47 ppm, respectively. Considering that a substitution of the C-3 carbon has typically very little influence on the H-3 shift value, we cannot conclusively state that the carbon observed at {delta} 75.5/3.67 ppm is substituted by the mannose C-3 or by the mannose C-4. As shown on the HMBC and HMQC spectra of KanLAM, an identical correlation with a carbon at {delta} 75.5/3.67 ppm was observed from 5-TMP H-1, indicating that the 5-TMP residues in KanLM and in KanLAM share identical locations. Altogether, these data demonstrate that KanLM and KanLAM are substituted within their respective mannan domain by a 5-TMP residue on either the C-3 or the C-4 position of a mannose unit.

Substitution by Succinic Acid—The presence of succinic acid in KanLM and KanLAM was assessed by standard GC. Ester-linked substituents were released from both compounds by methanolysis and analyzed by GC. This revealed that only KanLAM contains significant amounts of succinic acid, evaluated because of the use of adipic acid as internal standard to an average of two succinic acid per KanLAM molecule (data not shown).

The 13C-1H heteronuclear HMQC spectrum of KanLAM in Me2SO-d6 revealed the presence of a broad signal at {delta} 2.5–2.6/30.5 ppm (data not shown), which was tentatively assigned to the two distinct methylene groups of succinic acid, according to Delmas et al. (41). However, our data differ from this previous study in that carbons from both groups exhibited identical chemical shifts. This was attributed to the effect of different solvents (Me2SO-d6 versus D2O), which has been subsequently confirmed by repeating the experiment in D2O (data not shown). Experiments in Me2SO-d6 did not allow us to differentiate the two methylenic protons, whereas the protons of the two methylene groups in the D2O experiment were unambiguously assigned at {delta} 2.51 and 2.66 ppm (41). From the signal at {delta} 2.51 ppm, the 13C-1H heteronuclear HMBC parameters in D2O revealed a 3JH-C correlation at {delta} 176.06 ppm and a 2JH-C correlation at {delta} 181.7 ppm attributed to carbons of ester and carboxyl groups, respectively. Conversely, the proton at {delta} 2.66 ppm established a 3JH-C correlation with a carboxyl group carbon at {delta} 181.7 ppm and 2JH-C correlation with an ester group at {delta} 176.06 ppm. These data indicate the attribution of protons at {delta} 2.51 and 2.66 ppm as methylenic protons of succinic acid.

From the ester group carbon at {delta} 176.06 ppm, no additional correlation was detected in D2O, which prevented us from localizing the ester position on the KanLAM molecule in these experimental conditions. Despite the fact that the methylene protons were not resolved in Me2SO-d6, HMBC experiments showed that they correlated with two intense signals at {delta} 172.4 and 173.6 ppm, assigned by comparison with data obtained in D2O to ester and carboxyl groups, respectively. From the signal at {delta} 172.4 ppm, a single additional correlation was observed with a proton at {delta} 4.83 ppm. On the 13C-1H heteronuclear HMQC spectrum, we observed a corresponding signal at {delta} 80.5/4.84 ppm that was tentatively attributed to a non-anomeric carbon bearing the succinyl group, in accordance with a previous study (41). This demonstrated that the succinyl-substituted carbon was up-shifted to {delta} 82.2 ppm. From the proton at {delta} 4.84 ppm, the complete proton spin system of the monosaccharide was assigned by the 1H-1H homonuclear COSY 90, single and double relayed COSY. On this basis, the proton at {delta} 4.84 ppm was unambiguously attributed as a H-3 of a C-5 substituted Araf residue. On the 1H-1H homonuclear COSY 90 spectrum, the proton at {delta} 4.84 ppm correlated with two protons at {delta} 4.02 and 4.15 ppm attributed to H-2 and H-4, respectively (Fig. 7). The absence of correlation with an anomeric proton in the COSY 90 experiment established that the proton at {delta} 4.84 ppm was not a H-2. This was further confirmed by the appearance of an additional correlation with an anomeric proton at {delta} 4.96 ppm in single relayed COSY, which suggested that the proton at {delta} 4.84 ppm was a H-3. The nature of the H-4 at {delta} 4.13 ppm was confirmed by the observation in COSY 90 of crosspeaks with H-5 and H-5' at {delta} 3.75 and 3.59 ppm, respectively. 1H parameters of H-5/H-5' established that Araf residue was substituted in the C-5 position. Furthermore, H-5' correlated with the proton at {delta} 4.84 ppm in single relayed COSY, but not in COSY 90, supporting that this proton was a H-3. Thus, the 1H NMR parameters of this residue were attributed as follows: H-1, {delta} 4.96 ppm; H-2, 4.01; H-3, 4.84; H-4, 4.13; and H-5/H-5', 3.75/3.59. These results demonstrate that succinate residues substitute some 5-Araf residues of the arabinan core of KanLAM in the C-3 position.

Apoptosis-inducing Properties of KanLM
An increasing body of evidence suggests that evasion of infection-induced apoptosis of macrophages by virulent mycobacteria represents a mechanism to circumvent the defenses of the host and to perpetuate a favorable environment for their intracellular replication. Keane et al. (42) investigated the relative induction of alveolar macrophage apoptosis by virulent and attenuated mycobacteria and showed that, although virulent strains of M. tuberculosis failed to induce apoptosis, more attenuated strains, including M. kansasii, caused a significant increase of apoptosis over control. However, although various studies (15, 43–45) have proposed the participation of ManLAM in the modulation of apoptosis in M. tuberculosis-infected macrophages, the possibility that LM may also modulate apoptosis has never been reported. In this study, we investigated whether KanPIM2, KanLM or KanLAM may be involved in the apoptosis-inducing activity of M. kansasii. The human promonocytic THP-1 cell line was used because it responds similarly to primary human alveolar macrophages to the induction of apoptosis by mycobacteria (46). THP-1 cells were differentiated into adherent macrophages using PMA for 24 h and stimulated with the glycoconjugates for another 24 h. Induction of apoptosis was quantified using annexin V and propidium iodide staining combined with flow cytometry analysis to monitor the number of apoptotic cells. Fig. 9 shows that neither KanPIM2 nor KanLAM induced increased apoptosis compared with non-treated cells. Unexpectedly, KanLM was reproducibly found to trigger THP-1 apoptosis to levels similar to that induced by PILAM. Similar results were obtained using CheLM and CheLAM (data not shown).



View larger version (12K):
[in this window]
[in a new window]
  		FIG. 9.
Apoptosis-inducing activity of the purified M. kansasii glycoconjugates. Following differentiation with PMA, THP-1 cells were stimulated for 24 h with KanPIM2 (10 µg/ml), KanLM (10 µg/ml), or KanLAM (20 µg/ml), as well as with PILAM from a fast-growing mycobacterial species (20 µg/ml). After labeling with annexin V and propidium iodide, flow cytometry analysis was used to score the number of cells undergoing apoptosis (annexin V-positive and propidium iodide-negative). The results are expressed as duplicates of one representative experiment.

 


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
REFERENCES
 
LM and LAM are major mycobacterial cell wall components, and numerous reviews (9, 15, 47–49) have highlighted the involvement of LAM in the pathogenesis of tuberculosis. The main features of the structure of LM and LAM were determined in the 1990s, based on the foundations of some early structural work (50, 51). Extensive studies performed on LAM from M. smegmatis, M. tuberculosis, or from M. bovis BCG have shown that important biological effects depend on the degree and chemical structure of capping functions and other substituents on the arabinan moiety of LAM. A major breakthrough was the discovery that ManLAM from slow-growing species differs from PILAM from fast-growing species including M. smegmatis in its ability to induce cytokine production (14, 35, 52, 53). PILAM is considered as being a stronger cytokine inducer than ManLAM, suggesting that capping has evolved in pathogenic species as part of their survival strategy within the host. We have recently characterized the structure of LAM from M. chelonae (CheLAM) and have shown that it belongs to a new family of AraLAM molecules, because it is devoid of both the mannose cap and of the phosphoinositol cap (14). In addition, like ManLAM, CheLAM is unable to trigger pro-inflammatory cytokine secretion by human macrophages, suggesting that the phosphoinositol capping represents a major cytokine-inducing component of PILAMs (14). Up to now, LAM structures of only a limited panel of mycobacterial species have been determined. To increase our understanding of the functions of this lipoglycan, new structural information is required. In this study, we undertook the structural determination of the LAM from a clinical isolate of M. kansasii with the goal to highlight new sources of structural variability.

The analysis of the GPI anchors of KanLM and KanLAM showed that they are identical, consisting of phosphatidyl-myo-inositol substituted exclusively by 1/2-palmitoyl-1/2-tuberculostearoyl fatty acids. This structure is similar to that of ManLAM from M. bovis BCG, which contains mainly 1/2-palmitoylglycerol, 1/2-tuberculostearoylglycerol, 1,2-dipalmitoylglycerol, and 1-tuberculostearoyl-2-palmitoylglycerol (35). However, this structure contrasts with the extraordinary diversity of fatty acids linked to the GPI anchor from CheLAM (14). Besides sharing a similar overall structure with the ManLAM class of molecules, we show here that KanLAM presents novel structural features that differ from those of ManLAMs from M. tuberculosis or M. bovis BCG.

Both KanLM and KanLAM possess an unusual mannan core characterized by the presence of different oligomannosyl motifs substituting the {alpha}1,6-linked mannan core, presumably in the C-2 position. MALDI-MS analysis of the oligosaccharides released by complete acetolysis of the {alpha}1,6-Manp linkages of KanLM demonstrated the presence of mono-, di-, and tri-mannosidic branches. This contrasts with the structures of the mannan domain of all mycobacterial LAM molecules studied so far, generally characterized by the presence of unique {alpha}1,2-linked mono-Manp substitutions, with the exception of CheLAM which carries {alpha}1,3-linked mono-Manp side chains (14). The presence of oligomannosyl branches substituting the mannan core was confirmed by immunoblotting analysis using antibodies used for the diagnosis of Candida infections and known to recognize Man{alpha}1-(2Man [PDB] {alpha}1)1–3 determinants. This specific antiserum may also represent a useful tool for the rapid detection of LM structures carrying {alpha}1,2-linked oligomannosyl motifs substituting the {alpha}1,6-linked mannan core. Although no potential role has been attributed to these oligomannosyl side chains, it remains possible that they are involved in various functions of KanLM.

We show here that, in contrast to KanLAM and KanPIM2, KanLM is potent in inducing apoptosis in THP-1 cells. In addition, LMs, but not LAMs, were found to be strong pro-inflammatory cytokine inducers in THP-1 cells (57), and PIM2 failed to induce pro-inflammatory cytokine secretion in macrophages (57). These observations strongly suggest that the presence of only two Manp residues bound to the GPI anchor is not sufficient to trigger a pro-inflammatory response nor an apoptosis-inducing activity. This supports the hypothesis that structures with a higher number of Manp residues are necessary to elicit these effects. Further work will be required to investigate whether the Manp residues substituting the mannan core may also participate in the biological properties mediated by KanLM. Although several studies have focused on the apoptosis-modulating activities of LAMs (15),