Artigo Acesso aberto Revisado por pares

Structural Analysis of the Mannan Region of Lipoarabinomannan from Mycobacterium bovis BCG.

1995; Elsevier BV; Volume: 270; Issue: 25 Linguagem: Inglês

10.1074/jbc.270.25.15012

ISSN

1083-351X

Autores

Anne Venisse, Michel Rivière, Joseph Vercauteren, Germain Puzo,

Tópico(s)

Glycosylation and Glycoproteins Research

Resumo

Lipoarabinomannan (LAM) is a major antigen of mycobacterial cell walls, involved in host-Mycobacterium interactions. In a previous work, LAM from the vaccine strain, Mycobacterium bovis BCG, was found to exhibit mannooligosaccharides at its arabinan nonreducing ends (ManLAM). The present report concerns the mannan core structure of this ManLAM. After partial hydrolysis of ManLAM, two populations of mannans (Ma1 and Ma2) were obtained by gel filtration chromatography. Their structural features were defined by means of two-dimensional homo- and heteronuclear (1H-13C) NMR sequences and methylation analysis. They were both found to be composed of an α-(1 → 6)-linked mannan backbone with α-(1 → 2)-Manp-linked side chains. They are highly branched, and Ma2 presents a higher frequency of branching than Ma1. Moreover, chemical analysis indicates that only Ma1 is phosphorylated. By a two-dimensional heteronuclear 1H-31P total correlation experiment, the phosphate was found to be involved in a phosphodiester bond between inositol C-1 and glycerol C-3. Then, the molecular mass of mannan was established by mass spectrometry, which revealed a molecular mass of 3517 Da for the major molecular species of Ma1. Likewise, analysis of unfractionated mannans showed the occurrence of other, quantitatively minor molecular species, endowed with two phosphates.This study clearly indicates that the mannan region of M. bovis BCG ManLAM exists as a heterogeneous population of molecules whose structures differ in their degree of glycosylation, level of branching, and phosphorylation state. The hypothesis that the relative abundance of these different molecules modulates the biological functions of LAM is discussed. Lipoarabinomannan (LAM) is a major antigen of mycobacterial cell walls, involved in host-Mycobacterium interactions. In a previous work, LAM from the vaccine strain, Mycobacterium bovis BCG, was found to exhibit mannooligosaccharides at its arabinan nonreducing ends (ManLAM). The present report concerns the mannan core structure of this ManLAM. After partial hydrolysis of ManLAM, two populations of mannans (Ma1 and Ma2) were obtained by gel filtration chromatography. Their structural features were defined by means of two-dimensional homo- and heteronuclear (1H-13C) NMR sequences and methylation analysis. They were both found to be composed of an α-(1 → 6)-linked mannan backbone with α-(1 → 2)-Manp-linked side chains. They are highly branched, and Ma2 presents a higher frequency of branching than Ma1. Moreover, chemical analysis indicates that only Ma1 is phosphorylated. By a two-dimensional heteronuclear 1H-31P total correlation experiment, the phosphate was found to be involved in a phosphodiester bond between inositol C-1 and glycerol C-3. Then, the molecular mass of mannan was established by mass spectrometry, which revealed a molecular mass of 3517 Da for the major molecular species of Ma1. Likewise, analysis of unfractionated mannans showed the occurrence of other, quantitatively minor molecular species, endowed with two phosphates. This study clearly indicates that the mannan region of M. bovis BCG ManLAM exists as a heterogeneous population of molecules whose structures differ in their degree of glycosylation, level of branching, and phosphorylation state. The hypothesis that the relative abundance of these different molecules modulates the biological functions of LAM is discussed. INTRODUCTIONMycobacterium tuberculosis infects one-third of the world's population, and tuberculosis remains the largest cause of mortality in the world from a single infectious agent(1Bloom B.R. Murray G.J. Science. 1992; 257: 1055-1064Crossref PubMed Scopus (1236) Google Scholar). This emphasizes the need to understand the interaction of M. tuberculosis with host phagocytic cells and the immune system. However, little is known about the molecular basis of the pathogenesis of tuberculosis.Lipoarabinomannan (LAM)1( 1The abbreviations used are: LAMlipoarabinomannanManLAMLAM with mannosyl units capping the arabinan endsAraLAMLAM with arabinofuranosyl terminiLMlipomannanPIMsphosphatidylinositol mannosidesBCGBacillus Calmette GuérinTMStrimethylsilylationHMBCheteronuclear multiple bond connectivity spectroscopyHMQCheteronuclear multiple quantum correlation spectroscopyHOHAHAhomonuclear Hartmann-Hahn spectroscopyROESYrotating frame nuclear Overhauser spectroscopyHPLChigh performance liquid chromatographyLSIMSliquid secondary ion mass spectrometryInsinositolGroglyceroltterminal.) is considered as a major antigen of mycobacterial cell walls, widely distributed within Mycobacterium species(2Daniel T.M. Kubica G.P. Wayne L.G. The Mycobacteria: A Sourcebook. Marcel Dekker Inc., New York1984: 417-465Google Scholar). LAM from M. tuberculosis has been demonstrated to exhibit a wide variety of biological activities such as suppression of T lymphocyte proliferation (3), inhibition of g-interferon-mediated activation of murine macrophages(4Sibley L.D. Hunter S.W. Brennan P.J. Krahenbuhl J.L. Infect. Immun. 1988; 56: 1232-1236Crossref PubMed Google Scholar), and immunomodulation of a large array of macrophage cytokines(5Adams L.B. Fukutomi Y. Krahenbuhl J.L. Infect. Immun. 1993; 61: 4173-4181Crossref PubMed Google Scholar, 6Barnes P.F. Chatterjee D. Abrams J.S. Lu S. Wang E. Yamamura M. Brennan P.J. Modlin R.L. J. Immunol. 1992; 149: 541-547PubMed Google Scholar, 7Roach T.I.A. Barton C.H. Chatterjee D. Blackwell J.M. J. Immunol. 1993; 150: 1886-1896PubMed Google Scholar).LAM from M. tuberculosis is composed of a mannan core linked to a linear arabinan chain to which oligoarabinosyl side chains are attached(8Misaki A. Azuma I. Yamamura Y. J. Biochem. (Tokyo). 1977; 82: 1759-1770Crossref PubMed Scopus (60) Google Scholar). It was established that most of these side chains of the LAM from the virulent strain (Erdman) of M. tuberculosis are capped with either mono-, di-, or trimannosyl residues. This LAM is termed mannosylated and named ManLAM(9Chatterjee D. Lowell K. Rivoire B. McNeil M. Brennan P.J. J. Biol. Chem. 1992; 267: 6234-6239Abstract Full Text PDF PubMed Google Scholar, 10Chatterjee D. Khoo K.H. McNeil M. Dell A. Morris H.R. Brennan P.J. Glycobiology. 1993; 3: 497-506Crossref PubMed Scopus (78) Google Scholar). This capping is missing from the LAM called AraLAM isolated from a rapidly growing strain of Mycobacterium, initially described as the avirulent strain of M. tuberculosis H37Ra(11Chatterjee D. Bozic C. McNeil M. Brennan P.J. J. Biol. Chem. 1991; 266: 6960-9652Abstract Full Text PDF Google Scholar). Moreover, the reducing end of this AraLAM mannan core was found to be linked to a phosphatidyl-myo-inositol anchor (12Hunter S.W. Brennan P.J. J. Biol. Chem. 1990; 265: 9272-9279Abstract Full Text PDF PubMed Google Scholar, 13Chatterjee D. Hunter S.W. McNeil M. Brennan P.J. J. Biol. Chem. 1992; 267: 6228-6233Abstract Full Text PDF PubMed Google Scholar) similar to the mycobacterial phosphatidyl-myo-inositol mannoside (PIM) structure.Some of the biological activities of LAM were shown to be related to characteristic structural features of these complex molecules. LAM treatment with mild alkali results in a decrease of its ability to suppress in vitro antigen-induced proliferation, g-interferon activation of macrophages(4Sibley L.D. Hunter S.W. Brennan P.J. Krahenbuhl J.L. Infect. Immun. 1988; 56: 1232-1236Crossref PubMed Google Scholar), and cytokine release(6Barnes P.F. Chatterjee D. Abrams J.S. Lu S. Wang E. Yamamura M. Brennan P.J. Modlin R.L. J. Immunol. 1992; 149: 541-547PubMed Google Scholar). Acyl groups such as the fatty acids of the anchor but also the short fatty acids previously described (14Weber P.L. Gray G.R. Carbohydr. Res. 1979; 74: 259-278Crossref PubMed Scopus (20) Google Scholar, 15Leopold K. Fischer W. Anal. Biochem. 1993; 208: 57-64Crossref PubMed Scopus (35) Google Scholar) seem to be crucial to these biological properties. The Man-capping at the LAM nonreducing termini also results in significant functional differences in vitro which could have important consequences in the pathogenicity of M. tuberculosis. AraLAM in vitro rapidly triggers tumor necrosis factor release from infected murine macrophages, thereby enhancing their bacteriostatic activity, whereas ManLAM from the virulent strain elicits low levels of tumor necrosis factor that would be too low to restrain the parasite's growth(5Adams L.B. Fukutomi Y. Krahenbuhl J.L. Infect. Immun. 1993; 61: 4173-4181Crossref PubMed Google Scholar, 7Roach T.I.A. Barton C.H. Chatterjee D. Blackwell J.M. J. Immunol. 1993; 150: 1886-1896PubMed Google Scholar, 16Chatterjee D. Roberts A.D. Lowell K. Brennan P.J. Orme I.M. Infect. Immun. 1992; 60: 1249-1253Crossref PubMed Google Scholar, 17Bradbury M.G. Moreno C. Clin. Exp. Immunol. 1993; 94: 57-63Crossref PubMed Scopus (28) Google Scholar).We previously reported the global structure of LAM from Mycobacterium bovis Bacille Calmette-Guérin (BCG) strain Pasteur, used throughout the world as a vaccine against tuberculosis(18Venisse A. Berjeaud J.M. Chaurand P. Gilleron M. Puzo G. J. Biol. Chem. 1993; 268: 12401-12411Abstract Full Text PDF PubMed Google Scholar). We demonstrated that this LAM shares structural features with LAM from a virulent strain (H37Rv) of M. tuberculosis, since it is also capped by mannopyranosyl residues. Despite these data, ManLAM could still be considered as a virulence factor involved in mycobacterial invasion and/or survival in host cells. A more detailed structural definition of M. bovis BCG ManLAM was prevented by the considerable molecular heterogeneity of the fraction that was outlined during its molecular mass determination by matrix-assisted UV-laser desorption/ionization mass spectrometry. The present study was undertaken in order to define a more accurate structure of the mannan core of the ManLAM from M. bovis BCG and, more particularly, to establish the presence of a phosphatidyl-myo-inositol anchor at the reducing end of this molecule. Through the two-dimensional homo- and heteronuclear scalar coupling NMR and mass spectrometry analysis of the mannan, we unequivocally assess the presence of one phosphate group on an oligosaccharide of 20 glycosidic units. Moreover, the molecular heterogeneity of this fraction was shown to be size- and charge-dependent, especially by the occurrence of non-, mono-, and even diphosphorylated populations. Minor structural units causing this heterogeneity may be at the origin of important functional differences.In view of the dependence of biological functions on structural difference, the implications of these structural observations in the role of ManLAM as a virulence factor will be discussed.MATERIALS AND METHODSMild Acid HydrolysisThe dManLAM from M. bovis BCG strain Pasteur was purified as described previously (18Venisse A. Berjeaud J.M. Chaurand P. Gilleron M. Puzo G. J. Biol. Chem. 1993; 268: 12401-12411Abstract Full Text PDF PubMed Google Scholar) and was treated with 0.1 N HCl, 100°C, 15 min in order to specifically hydrolyze the arabinan chains. The hydrolysate was applied on a Bio-Gel P-4 or P-6 eluted with water, and samples were assayed by GC after hydrolysis and TMS derivatization.Separation of Mannans by HPLCMannan fractions obtained after Bio-Gel P-6 were separated by anion exchange HPLC using an analytical Carbopac PA1 (Dionex) column at neutral pH. Elution was performed using a gradient of sodium acetate (ACS Grade, Merck) in deionized water (18 MW) with a flow rate of 1 ml/min. Separation was monitored by pulsed amperometric detection (PAD II) (Dionex Corp.) after NaOH post-column addition. Using a two-split system, only 15% of the product was directed to the detector after addition of 300 μl min-1 of NaOH solution (500 mM), and the remaining 85% were collected. HPLC was conducted on a Gilson (Gilson, France) gradient system.MethylationMannans were O-methylated three times according to a modified (18Venisse A. Berjeaud J.M. Chaurand P. Gilleron M. Puzo G. J. Biol. Chem. 1993; 268: 12401-12411Abstract Full Text PDF PubMed Google Scholar) procedure of Ciucanu and Kerek(19Ciucanu I. Kerek F. Carbohydr. Res. 1984; 131: 209-217Crossref Scopus (3170) Google Scholar). Per-O-methylated products were hydrolyzed with 2 N trifluoroacetic acid at 110°C for 3 h, reduced with NaBH4, and peracetylated. The alditol acetates of methylated sugars were identified by GC/MS. Glycosidic linkage analysis was carried out according to Sweet et al.(20Sweet D.P. Shapiro R.H. Albersheim P. Carbohydr. Res. 1975; 40: 967-974Google Scholar).Chemical AnalysisQuantitative carbohydrate determinations were routinely carried out by GC analysis after hydrolysis (2 N trifluoroacetic acid at 110°C for 2 h) and TMS derivatization(18Venisse A. Berjeaud J.M. Chaurand P. Gilleron M. Puzo G. J. Biol. Chem. 1993; 268: 12401-12411Abstract Full Text PDF PubMed Google Scholar). myo-Inositol content was measured following acid hydrolysis (6 N HCl, 110°C, 18 h) and TMS derivatization. These conditions were proved to improve inositol recovery from inositol-1-P(21Roberts W.L. Kim B.H. Rosenberry T.L. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 7817-7821Crossref PubMed Scopus (96) Google Scholar). Glucitol was added as an internal standard prior to hydrolysis. Total phosphorus was determined after perchloric acid hydrolysis according the Bartlett procedure(22Bartlett G.R. J. Biol. Chem. 1959; 234: 466-468Abstract Full Text PDF PubMed Google Scholar).GC and GC/MSRoutine gas chromatography (GC) was performed on alditol acetates and TMS derivatives of monosaccharides using a Girdel series 30 equipped with an OV-1 (0.3-μm film thickness, Spiral, Dijon, France) fused silica capillary column (25-m length ∙ 0.22-mm inside diameter). A temperature program from 100 to 280°C at a speed of 3°C/min was used.GC/MS was performed on a Hewlett-Packard MS engine using both EI and CI ionization modes.Mass Spectrometry LSIMSLSIMS was performed on a ZAB 2E mass spectrometer (VG analytical). Spectra were generated by a 35-kV cesium ion beam with an acceleration voltage of 8 kV. Thioglycerol and (thioglycerol + 1% acetic acid) were used as matrix for negative and positive mode spectra, respectively.NMR SpectroscopyNMR spectra were recorded on a Bruker AMX-500 spectrometer equipped with an Aspect X32 computer. Samples were repeatedly lyophilized in D2O for hydroxyl deuterium exchange and dissolved in D2O (Spin et Techniques, Paris, France, 99.96% purity) at a concentration of 2 mg/ml in a 200 ∙ 5 mm 535-PP NMR tube. All spectra were recorded at 303 K.Power-gated 1H-decoupled phosphorus-31 (31P) spectra (202 MHz) was recorded with 25-kHz spectral width, a 17-μs 90° pulse, and 3 s of relaxation time. The 16K acquired complex points were zero-filled to 64K and processed after exponential multiplication (line broadening = 5 Hz). Phosphoric acid (85%) was used as the external standard (dP = 0.0).The homonuclear Hartmann-Hahn (HOHAHA) (24Bax A. Davis D.G. J. Magn. Res. 1985; 65: 355-360Google Scholar) and rotating frame NOE (ROESY) (25Bothner-By A.A. Stephens R.L. Lee J. Warren C.D. Jeanloz R.W. J. Am. Chem. Soc. 1984; 106: 811-813Crossref Scopus (1962) Google Scholar, 26Bax A. Davis D.G. J. Magnetic Res. 1985; 63: 207-213Google Scholar, 27Rance M. Sorensen O.W. Bodenhausen G. Wagner G. Ernst R.R. Wüthrich K. Biochem. Biophys. Res. Commun. 1983; 117: 479-485Crossref PubMed Scopus (2594) Google Scholar) experiments were recorded without sample spinning using the standard pulse sequences supplied by Bruker, and data were acquired in the phase-sensitive mode using the time-proportional phase increment method(23Marion D. Wüthrich K. Biochem. Biophys. Res. Commun. 1983; 113: 967-974Crossref PubMed Scopus (3516) Google Scholar).Heteronuclear correlation 1H {13C} HMQC (28), 1H {13C} HMBC (31Bax A. Summers M.F. J. Am. Chem. Soc. 1986; 108: 2093-2094Crossref Scopus (3261) Google Scholar) spectra were recorded in the proton-detected mode with a Bruker 5-mm 1H-broad band tunable probe with reversal geometry. The 1H {31P} HMQC-HOHAHA was obtained according to Lerner's pulse sequence(30Lerner L. Bax A. J. Magnetic Res. 1986; 69: 375-380Google Scholar), and a GARP sequence (29Shaka A.J. Barker P.B. Freeman R. J. Magnetic Res. 1985; 64: 547-552Google Scholar) was used for 31P decoupling during acquisition.RESULTSChromatographic and Chemical AnalysisAfter an acid-catalyzed partial depolymerization of the deacylated molecule (dManLAM) we previously demonstrated that LAM from M. bovis BCG is mannosylated at its nonreducing ends(18Venisse A. Berjeaud J.M. Chaurand P. Gilleron M. Puzo G. J. Biol. Chem. 1993; 268: 12401-12411Abstract Full Text PDF PubMed Google Scholar). The ManLAM hydrolysate (0.1 N HCl, 15 min, 100°C) was applied to a Bio-Gel P4 column (data not shown) to separate the mono- and oligosaccharides from the nonhydrolyzed mannan (Ma). These mono- and oligosaccharides, arising from the arabinan domain, and the mannan core were identified from their monosaccharide composition routinely established by GC analysis after hydrolysis and TMS derivatization. The mannan core, eluted in the void volume, was predominantly composed of mannosyl units (Man) as expected and also of arabinosyl (Ara), inositol (Ins), and glycerol (Gro) units in small amounts. Moreover, 31P NMR analysis of Ma revealed one signal at 0.75 ppm, the chemical shift of which is not affected by pH in a range between pD = 6.8 and pD = 9.8, thus strongly suggesting the presence of a phosphodiester group.The Ma core was analyzed by liquid secondary ion mass spectrometry (LSIMS) in both negative and positive modes (Fig. 1, A and B). In the high mass range, the negative mode-mass spectrum (Fig. 1A) is dominated by the peak at m/z 3037.9. Two other signals were observed at m/z 3015.9 and m/z 2999.8 distant from the base peak by 22 and 38 atomic mass units, respectively. These mass differences agree with the presence of sodium and potassium, suggesting that these three peaks characterize one molecular species. Thus, the m/z 3037.9 pseudomolecular ion contained one Na+ and one K+ leading to the following structure (M + Na + K - 3H)-. In order to support this ion assignment, Ma was analyzed in negative mode-LSIMS with a matrix doped with KI (data not shown). As expected, the peak at m/z 3037.9 was missing, and the mass spectrum was dominated by one signal at m/z 3015.9 assigned to (M + K - 2H)-. The positive mode mass spectrum of Ma (Fig. 1B) showed abundant cationized molecular ions (M + Na + K - H)- at m/z 3039.6, supporting the previous interpretations. From all these data, the average molecular weight of M was established at 2978.4 Da. From this and from the Ma composition determined by GC analysis, we can propose that M, the major molecular species detected in Ma, was composed of 16 (Man, Ins), one Ara, one Gro, and two phosphates.Moreover, LSIMS analysis revealed two other sets of peaks down-shifted from M by 162 and 324 atomic mass units indicating the presence of two other molecular species, in lesser abundance, containing, respectively, one and two Man (or Ins) less than M. More interestingly, the high mass range of the positive mode-LSIMS spectrum shows two other sets of pseudomolecular ions at m/z 3561.5, 3539.4 and m/z 3399.3, 3377.2. The analogous ions were missing in the negative mode spectrum. Moreover, the mass difference between these ions and those assigned to M cannot be explained only by a different number of monosaccharide units.In order to fractionate the various molecular species revealed by the LSIMS analysis, the ManLAM hydrolysate was chromatographed on a Bio-Gel P6 column monitored by refractory index (Fig. 2). Fractions 1 and 2, namely Ma1 and Ma2, respectively, were of interest since they predominantly contained mannosyl residues. The remaining oligosaccharidic fractions, 3 to 6, showed higher Ara/Man ratios and arose from the nonreducing termini of the arabinan domain(18Venisse A. Berjeaud J.M. Chaurand P. Gilleron M. Puzo G. J. Biol. Chem. 1993; 268: 12401-12411Abstract Full Text PDF PubMed Google Scholar).Figure 2:Bio-Gel P-6 chromatography of the dManLAM hydrolysate (0.1 N HCl, 15 min, 100°C) eluted with water. Fractions 1 to 6 were analyzed by GC after hydrolysis and TMS derivatization. Samples 1 and 2 contained a high percentage of mannosyl residues (called Ma1 and Ma2), while samples 3-6 contained a higher Ara/Man ratio.View Large Image Figure ViewerDownload Hi-res image Download (PPT)The molecular homogeneity of Ma1 and Ma2 was then checked by ion exchange chromatography with an analytical Carbopac PA1 column using an HPLC device monitored by an amperometric detector (Fig. 3). Gradient of sodium acetate in water was used as solvents, and sodium hydroxide was added post-column for amperometry detection. 80% of the Ma1 fraction was eluted with a retention time of 4.5 min (peak II) during the isocratic elution with water while most of the Ma2 appeared as a single peak (I) in the void volume (3 min). Small quantities of materials were detected with the sodium acetate gradient, between 15 and 20 min. The separation on Carbopac at neutral pH was dependent on the charge-to-mass ratio of the oligosaccharides. These chromatographic studies revealed the relative homogeneity of each fraction and suggested that fraction Ma1 was predominantly composed of a more strongly charged mannan population than Ma2.Figure 3:Ion exchange HPLC profile (analytical column Carbopac PA1) of Ma1 (A) and Ma2 (B). The column was eluted for 10 min with water then with a gradient of 0.5 M sodium acetate in water for 20 min. The glycosidic samples were detected by amperometry as explained under "Materials and Methods." In these conditions, a standard galactose sample was eluted in the void volume at 3 min while glucuronic acid and xylose-1-P were eluted in the gradient at 19 min and 25 min, respectively.View Large Image Figure ViewerDownload Hi-res image Download (PPT)GC quantitative analysis established that Ma1 and Ma2 were composed by Aras and Mans in an approximate Ara/Man molar ratio of 1:20 and 2:20, respectively. In the same way, the ratio Ins/Man was estimated at 1:20 for Ma1 and 1:80 for Ma2. Finally, the phosphorus assay of each fraction indicated 0.7 mol of phosphorus/mol of Ma1 and 0.15 mol of phosphorus/mol of Ma2 (with an approximate molecular mass of 3000 Da for mannans).Methylation data (Table I) obtained with Ma1 and Ma2 were in agreement with those previously reported for the ManLAM (18Venisse A. Berjeaud J.M. Chaurand P. Gilleron M. Puzo G. J. Biol. Chem. 1993; 268: 12401-12411Abstract Full Text PDF PubMed Google Scholar) where only mannopyranosyl and arabinofuranosyl ring forms were observed. Ma1 showed the presence of t-Manp and 2,6-linked Manp (40 and 35%, respectively) at twice the proportion of 6-linked Manp (16%). This latter unbranched Manp was found in lower proportions in Ma2 (9%), being 4-fold less abundant than 2,6-linked Manp (34%). Small amounts of t-Araf, 5-linked Araf, and 2-linked Manp residues were also present in both mannans. These data indicate that both mannans are highly branched.Table I:Methylation analysis of Ma1 and Ma2 from M. bovis BCG dManLAM Open table in a new tab Both fractions Ma1 and Ma2 were then analyzed by LSIMS, one-dimensional and two-dimensional homo- and heteronuclear 1H-13C and 1H-31P NMR.LSIMS AnalysisThe high mass range of the positive mode-LSIMS spectrum of Ma1 (Fig. 4A) showed an intense peak at m/z 3539.3. The presence of two other signals of lower intensity at m/z 3517.5 and m/z 3555.9 separated by −22 atomic mass units and +16 atomic mass units, respectively, from the base peak, suggested that the signal at m/z 3539.3 corresponds to the cationized molecular ion (M1 + Na)+. Thus, the proton and potassium-containing molecular ions were assigned to the peaks at m/z 3517.5 and m/z 3555.9, respectively. Moreover, these assignments were supported by the negative mode mass spectrum (Fig. 4B), showing an intense peak at m/z 3515.5 attributed to the pseudomolecular ion (M1 - H)-. From these values, the M1 average molecular mass was determined at 3517.0 Da. From the molecular mass of M1 and the chemical composition of Ma1, we can propose that M1 is composed of 18 Manps, 2 Arafs, 1 Ins, 1 Gro, and 1 phosphate group. A second set of ions was observed in lower abundance in both spectra and characterized another molecular species in Ma1 containing one Manp less.Figure 4:LSIMS analysis of Ma1. A, positive mode LSIMS spectrum of Ma1 in a matrix of thioglycerol, 10% acetic acid. B, negative mode LSIMS spectrum of Ma1 in thioglycerol. Both spectra showed a second set of ions in lower abundance with a mass difference of −162 atomic mass units from the base peak.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Despite our efforts, the Ma2 LSIMS analysis was unsuccessful. Positive and negative mass spectra were devoid of pseudomolecular ions in the expected mass range. Nevertheless, the absence of Ma2 pseudomolecular ions can be explained by the absence of phosphate groups in this molecule.We noted the absence, in the Ma1 and Ma2 LSIMS spectra, of the abundant pseudomolecular ion at m/z 3039.9 observed in the positive LSIMS spectrum of Ma and assigned to a mannan core with two phosphate groups.In order to define the Ma1 and Ma2 structures more precisely, they were analyzed by one-dimensional and two-dimensional NMR spectroscopy.NMR Studies of Ma1 and Ma2The Ma1 and Ma2 proton anomeric resonance regions (Fig. 5) are dominated by two signals at 5.06 and 5.14 ppm (denoted B and D). In both spectra, the resonance at 5.14 ppm appeared to form a complex signal with two near-resonances at 5.11 (C) and 5.15 ppm. Two weaker superimposed signals at 4.915 and 4.93 ppm (denoted A) were also common to both mannans. However, they differed in their relative integration value, which was lower in the case of Ma2 (reported in Fig. 5). The downfield resonance at 5.20 ppm (F) was also observed in both spectra but was of lower intensity in Ma2. It can be noticed that a weak signal at 5.18 ppm (E) was only present in Ma2.Figure 5:Expanded region of the one-dimensional 1H NMR spectra (d 4.3 to 5.2) of Ma1 (A) and Ma2 (B) anomeric protons are labeled A-F, as shown in Table I. The relative integration values are reported. Spectra were recorded over a spectral width of 5005 Hz using a 7-μs 90° pulse, 1.5 s of recycle delay and 1536 acquisitions of 1.63 s. Flame ionization detectors (16K) were zero-filled to 64K and multiplied by a exponential function (line broadening = 0.5 Hz) prior to Fourier transformation.View Large Image Figure ViewerDownload Hi-res image Download (PPT)These anomeric protons and carbons were routinely assigned from the HMQC spectrum (not shown). The remaining protons and carbons of each spin system which characterize the units involved in Ma1 and Ma2 were partially attributed from the COSY (not shown), HOHAHA, HMQC, and HMBC spectra. The assignment of signals was also based on the chemical shifts of analogous compound NMR studies(18Venisse A. Berjeaud J.M. Chaurand P. Gilleron M. Puzo G. J. Biol. Chem. 1993; 268: 12401-12411Abstract Full Text PDF PubMed Google Scholar, 32Allerhand A. Berman E. J. Am. Chem. Soc. 1984; 106: 2400-2412Crossref Scopus (32) Google Scholar, 33Kogan G. Pavliak V. Masler L. Carbohydr. Res. 1988; 172: 242-253Crossref Scopus (57) Google Scholar). These values are summarized in Tables II and III.Table II:Assignment of some protons and carbons of both Ma1 and Ma2 glycosidic units, based upon the interpretation of HMQC, HMBC, and HOHAHA spectra and data from literature. Chemical shifts are reported in parts per million and were measured in D2O. Open table in a new tab Table III:Assignment of inositol protons based upon their chemical shift and multiplicity and data from literature Open table in a new tab Starting from the downfield anomeric resonances D at 5.14 and 5.15 ppm, the cross-peak in the HMQC spectrum showed a correlation with an upfield anomeric carbon resonance at 100.89 ppm. The C-1 chemical shift of an α-D-Manp is described to be shifted upfield (Δ∙ = 2 ppm) while its H-1 is deshielded (Δd = 0.2 ppm) upon 2-O-substitution(34Cohen R.E. Ballou C.E. Biochemistry. 1980; 19: 4345-4358Crossref PubMed Scopus (117) Google Scholar). Thus, both the anomeric carbon and the D protons were attributed to a 2-O-linked α-Manp. This assignment was confirmed by the HMQC and HOHAHA spectra (Fig. 6) allowing the attribution of the C-2 resonance at 81.45 ppm and the H-2 resonances at 4.045 and 4.055 ppm. This downfield chemical shift of the C-2 resonance (Δd = 7 ppm) typified a 2-O-linked α-Manp. Moreover, the C-5 resonance was localized at 74.05 ppm owing to the (C-5,H-1) correlation observed by HMBC (Fig. 7). This chemical shift is shifted upfield from its normal range (Δd = 2 ppm) () and is in agreement with glycosylation in C-6. Thus, this spin system was attributed to the 2,6-di-O-linked α-Manp unit characterized by methylation analysis.Figure 6:Partial two-dimensional HOHAHA spectrum (d 3.6-4.2/d 4.85-5.23)

Referência(s)