MS Approach for Characterization of the Fatty Acid Distribution on

Anal. Chem. 2006, 78, 8543-8548

MS/MS Approach for Characterization of the Fatty Acid Distribution on Mycobacterial Phosphatidyl-myo-inositol Mannosides Martine Gilleron,*,† Buko Lindner,‡ and Germain Puzo†

Institut de Pharmacologie et de Biologie Structurale du Centre National de la Recherche Scientifique, 205 route de Narbonne, 31077 Toulouse Cedex 4, France, and Division of Biophysics, Research Center Borstel, D-23845 Borstel, Germany

Phosphatidyl-myo-inositol mannosides (PIM) are not only important structural components of the mycobacterial envelope but also are major non-peptidic antigens of the host innate and acquired immune responses. Indeed, they are ligands of TLR-2 and they activate CD1-restricted T lymphocytes. In addition, PIM constitute the basic structure of the lipidic anchor of two lipoglycans, lipomannans and lipoarabinomannans, which are important immunomodulators in the course of tuberculosis. The fatty acyl substituents present on PIM molecules play a crucial role for both their physical properties and biological activities. PIM contain four acylation sites, two on the glycerol, one on a mannose, and one on the myo-inositol units. We propose here an analytical procedure, based on mass spectrometry, to determine the structure of the fatty acids present on each of these different acylation sites. We show that the nature of the fatty acids located on both positions of glycerol can be deduced from IRMPD analysis of negative precursor ions from native PIM species, while the fatty acids located on myo-inositol and mannose units can be identified by MALDI-TOF CID MS of protonated and cationized molecular ions. Thus, the combination of MS/MS data obtained from positive and negative pseudomolecular ions generated by ESI or MALDI appears as a powerful approach for the structural characterization of the PIM acyl form structure.

Mycobacterium tuberculosis envelope is characterized by extraordinary high lipid content and molecular diversity. Indeed lipids constitute up to 60% of the bacilli dry weight as compared to ∼20% for the lipid-rich envelope of Gram-negative organisms.1-3 Accordingly, ∼250 genes within the M. tuberculosis genome are * To whom correspondence should be addressed. E-mail: Martine.Gilleron@ ipbs.fr. Fax: 33 5 61 17 59 94. † Centre National de la Recherche Scientifique. ‡ Research Center Borstel. (1) Puzo, G. Ann. Institut Pasteur 1993, 4, 225-238. (2) Daffe, M.; Draper, P. Adv. Microb. Physiol. 1998, 39, 131-203. (3) Brennan, P. J.; Nikaido, H. Annu. Rev. Biochem. 1995, 64, 29-63. 10.1021/ac061574a CCC: $33.50 Published on Web 11/15/2006

© 2006 American Chemical Society

involved in lipid metabolism.4 Among these lipids, phosphatidylmyo-inositol mannosides (PIM), lipomannans (LM), and lipoarabinomannans (LAM) share a mannosyl-phosphatidyl-myo-inositol (MPI) anchor based on a sn-glycero-3-phospho-(1-D-myo-inositol) glycosylated in both positions 2 and 6 of the myo-inositol (Ins) by one R-D-mannopyranose (R-D-Manp) unit. (PIM is used to describe the global family of PIM that carries one to four fatty acids and one to six mannose residues. In PIMY, y refers to the number of mannose residues. AcPIM and Ac2PIM refer to tri-and tetraacylated form of PIM, respectively.) Four acylation sites are defined, two on the glycerol (Gro) moiety, one on the C3 of the Ins and one on the C6 of the R-Manp residue linked at position 2 of the Ins. This anchor has been shown to be predominantly acylated by palmitic, stearic, and tuberculostearic acids. The structures of PIM, LM, and LAM are now well established (see reviews),5-7 and synthetic PIM8-10 and LM11 are even now available. These lipoglycans play a role in the ability of mycobacteria to persist and replicate inside the macrophages. Indeed, they are described as immunoregulators of the proinflammatory cytokine secretion.5 M. tuberculosis mannosylated LAM (ManLAM) alter innate immunity response by inhibiting the production of IL-12 and TNF-R from activated macrophages and dendritic cells (DCs).12 Moreover, ManLAM inhibit phagosome/lysosome fusion13 and are also putative ligands allowing the entrance of the (4) Cole, S. T.; Brosch, R.; Parkhill, J.; Garnier, T.; Churcher, C.; Harris, D.; Gordon, S. V.; Eiglmeier, K.; Gas, S.; Barry, C. E., 3rd; Tekaia, F.; Badcock, K.; Basham, D.; Brown, D.; Chillingworth, T.; Connor, R.; Davies, R.; Devlin, K.; Feltwell, T.; Gentles, S.; Hamlin, N.; Holroyd, S.; Hornsby, T.; Jagels, K.; Barrell, B. G.; et al. Nature 1998, 393, 537-544. (5) Nigou, J.; Gilleron, M.; Puzo, G. Biochimie 2003, 85, 153-166. (6) Briken, V.; Porcelli, S. A.; Besra, G. S.; Kremer, L. Mol. Microbiol. 2004, 53, 391-403. (7) Gilleron, M.; Jackson, M.; Nigou, J.; Puzo, G. In The mycobacterial cell envelope : an overview; Daffe´, M., Reyrat, J. M., Eds.; Vol. submitted. (8) Stadelmaier, A.; Schmidt, R. R. Carbohydr. Res. 2003, 338, 2557-2569. (9) Ainge, G. D.; Hudson, J.; Larsen, D. S.; Painter, G. F.; Gill, G. S.; Harper, J. L. Bioorg. Med. Chem. 2006, 14, 5632-5642. (10) Liu, X.; Stocker, B. L.; Seeberger, P. H. J. Am. Chem. Soc. 2006, 128, 36383648. (11) Jayaprakash, K. N.; Lu, J.; Fraser-Reid, B. Angew. Chem., Int. Ed. 2005, 44, 5894-5898. (12) Nigou, J.; Zelle-Rieser, C.; Gilleron, M.; Thurnher, M.; Puzo, G. J. Immunol. 2001, 166, 7477-7485. (13) Fratti, R. A.; Chua, J.; Vergne, I.; Deretic, V. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 5437-5442.

Analytical Chemistry, Vol. 78, No. 24, December 15, 2006 8543

bacilli inside DCs through DC-SIGN.14,15 LM and PIM are agonists of Toll-like receptor 2 (TLR-2).16,17 In addition, PIM stimulate unconventional RβT cells in the context of the cluster of differentiation 1 (CD1) proteins.18,19 All these activities are abrogated with the loss of the fatty acyl appendages. Even more interesting, a modulation of these activities has been found according to the acylation degree of the molecules. For instance, tri- and tetracylated LM, but not monoand diacylated forms, induce a TLR-2-dependent production of proinflammatory cytokines.17 Acylation on a specific site might be important in the context of PIM presentation by CD1 proteins.19 Indeed, we observed that only some particular PIM2 acyl forms are able to stimulate T-cells, meaning that CD1b proteins can discriminate PIM2 according to their extent and most likely their site of acylation (Gilleron, Puzo, and De Libero, unpublished results). The crystal structures of CD1b proteins have revealed that their antigen-binding groove is composed of four hydrophobic pockets, but only two of them are used for the loading of the antigens via their lipidic moiety. We presently do not know what are the fatty acids, on the PIM molecule, required for the presentation of the PIM-derived antigenic structure to T-cells. Altogether, these data highlight the importance of both the acylation degree and the site of acylation. Up to now, different strategies based on MS have been developed to investigate the structure of the fatty acids esterifying the different sites of the MPI anchor. Khoo et al.20 successfully used a procedure based on Prehm methylation followed by FABMS analysis. Then, Kremer et al.21 developed a method based on FAB-MS analysis in positive mode of O-peracetylated PIM, allowing the chemical characterization of the AcPIM3 acyl form. However, both methods require derivatization steps, and analysis of PIM acyl form mixture would necessitate their purification. To overcome these problems, another approach based on MS/MS analysis of PIM pseudomolecular ions obtained by electrospray ionization (ESI) in negative mode22 was proposed. This method allowed the characterization of the fatty acyl appendages located on the Gro but was unsuccessful in determining the structure of the fatty acid residues located on the Ins and Manp residues. Then, an alternative approach was developed, consisting in ESI-MS analysis of the acetolysis reaction products of each PIM2 acyl form.22 Acetolysis cleaves the phosphoglycerol linkage without (14) Tailleux, L.; Maeda, N.; Nigou, J.; Gicquel, B.; Neyrolles, O. Trends Microbiol. 2003, 11, 259-263. (15) Geijtenbeek, T. B.; Van Vliet, S. J.; Koppel, E. A.; Sanchez-Hernandez, M.; Vandenbroucke-Grauls, C. M.; Appelmelk, B.; Van Kooyk, Y. J. Exp. Med. 2003, 197, 7-17. (16) Quesniaux, V. J.; Nicolle, D. M.; Torres, D.; Kremer, L.; Guerardel, Y.; Nigou, J.; Puzo, G.; Erard, F.; Ryffel, B. J. Immunol. 2004, 172, 4425-4434. (17) Gilleron, M.; Nigou, J.; Nicolle, D.; Quesniaux, V.; Puzo, G. Chem. Biol. 2006, 13, 39-47. (18) Sieling, P. A.; Chatterjee, D.; Porcelli, S. A.; Prigozy, T. I.; Mazzaccaro, R. J.; Soriano, T.; Bloom, B. R.; Brenner, M. B.; Kronenberg, M.; Brennan, P. J.; et al. Science 1995, 269, 227-230. (19) De la Salle, H.; Mariotti, S.; Angenieux, C.; Gilleron, M.; Garcia-Alles, L. F.; Malm, D.; Berg, T.; Paoletti, S.; Maitre, B.; Mourey, L.; Salamero, J.; Cazenave, J. P.; Hanau, D.; Mori, L.; Puzo, G.; De Libero, G. Science 2005, 310, 1321-1324. (20) Khoo, K. H.; Dell, A.; Morris, H. R.; Brennan, P. J.; Chatterjee, D. Glycobiology 1995, 5, 117-127. (21) Kremer, L.; Gurcha, S. S.; Bifani, P.; Hitchen, P. G.; Baulard, A.; Morris, H. R.; Dell, A.; Brennan, P. J.; Besra, G. S. Biochem. J. 2002, 363, 437-447. (22) Gilleron, M.; Ronet, C.; Mempel, M.; Monsarrat, B.; Gachelin, G.; Puzo, G. J. Biol. Chem. 2001, 276, 34896-34904.

8544

Analytical Chemistry, Vol. 78, No. 24, December 15, 2006

altering the fatty acid esters, leading to two entities: the dimannosyl-inositol phosphate moiety (Man2-Ins-P) (observed in negative ion mode as [M - H]- ions) and the acyl glycerol residue (analyzed in positive mode in the presence of ammonium acetate as [M + NH4]+ ions). The drawback of these two methods was that no diagnostic fragment ions were observed to accurately establish the structure of the fatty acid located on each position of the glycerol moiety or on the Manp versus the Ins. For another class of bacterial glycolipids containing lipid A, infrared multiphoton dissociation (IRMPD) and collision-induced dissociation (CID) MS was used with Fourier transform (FT)-MS in the positive-ion mode to determine the distribution of fatty acids linked to different sugar units. Lipid A consists of a bisphosphorylated β-1,6-linked GlcN disaccharide backbone, which carries hydroxy and acyloxyacyl fatty acids in ester and amide linkages. Abundant B-type fragment ions resulting from the cleavage of the glycosidic bond were observed only in the positive-ion mode. These fragment ions allowed the determination of the number of fatty acids linked to the nonreducing and the reducing GlcN.23,24 Since the potential acylation sites on the MPI anchor are known from nuclear magnetic resonance (NMR) data, we propose here a new powerful analytical approach to unambiguously define the acylation degree and the structure of the fatty acid residues present on each position from a combination of MS analyses of complex mixtures of native PIM. To achieve this goal, we have exploited the advantage of sensitivity offered by matrix-assisted laser desorption/ionization (MALDI) and ESI ionization modes and the capacity of MS/MS techniques to generate diagnostic fragment ions from both positive and negative pseudomolecular ions. EXPERIMENTAL SECTION PIM2 Fraction. The PIM-containing lipidic extract was obtained through purification of the phenolic glycolipids from Mycobacterium bovis bacillus Calmette Gue´rin 1173P2 (BCG, the Pasteur strain) as described elsewhere. An acetone-insoluble, phospholipids-containing lipid extract was prepared22 and applied to a QMA-Spherosil M (BioSepra SA, Villeneuve-la-Garenne, France) column, which was first irrigated with chloroform, chloroform/methanol (1:1, v/v), and methanol in order to elute neutral compounds. Phospholipids were eluted using ammonium acetate-containing organic solvents. The 0.1 M ammonium acetate in chloroform/methanol (1:2, v/v) allowed elution of 750 mg of phospholipids, enriched in phosphatidyl-myo-inositol dimannosides (PIM2). Repeated lyophilizations were necessary to eliminate ammonium acetate salts. Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry. Analysis by MALDI-TOF mass spectrometer was carried out on a 4700 Proteomics Analyzer (with TOF/TOF optics, Applied Biosystems, Voyager DE-STR, Framingham, MA) using the reflectron mode. Ionization was effected by irradiation with a Nd:YAG laser (355 nm) operating by pulses of 500 ps with a frequency of 200 Hz. PIM were analyzed in the negative- or positive-ion modes, as specified in the text. Spectra from 2500 to 5000 laser shots were summed to obtain the final spectrum. The (23) Schwudke, D.; Linscheid, M.; Strauch, E.; Appel, B.; Zahringer, U.; Moll, H.; Muller, M.; Brecker, L.; Gronow, S.; Lindner, B. J. Biol. Chem. 2003, 278, 27502-27512. (24) Kondakova, A. N.; Lindner, B. Eur. J. Mass Spectrom. 2005, 11, 535-548.

2-(4-hydroxyphenylazo)benzoic acid (HABA) or 2,5-dihydroxybenzoic acid (DHB) matrixes (Sigma) were used at a concentration of ∼10 mg/mL in water or ethanol/water (1:1, v/v). Typically, 4 µL of PIM (5-10 µg) in HCCl3/CH3OH/H2O, 60:35:8, v/v/v, and 4 µL of the matrix solution were mixed with a micropipet and 0.3 µL of the mixture was deposited on the target. The measurements were internally calibrated at two points with PIM. Cation-Exchange Resin. The resin was conditioned in this way: it was first put in a 1 N NaOH solution, vortexed, and rinsed with several steps of centrifugation until a neutral pH was reached. Then the resin was put in 1 N HCl, vortexed, and again rinsed with several steps of centrifugation until a neutral pH was reached. The 4 µL of PIM (5-10 µg) in HCCl3/CH3OH/H2O, 60:35:8, v/v/ v, and 4 µL of the matrix solution were mixed together with some beads with a micropipet and 0.3 µL of the mixture was deposited on the target. CID MS/MS. CID gas type is atmosphere, and the gas pressure was set to medium. Electrospray Ionization Fourier Transform Mass Spectrometry. FT-MS was performed in the negative- and positiveion modes using an APEX II instrument (Bruker Daltonics, Billerica, MA) equipped with an actively shielded 7-T magnet and an (nano) ESI source. Mass spectra were acquired using standard experimental sequences as provided by the manufacturer. For the negative-ion spectra, samples (∼20 ng·µL-1) were dissolved in a 50:50:0.001 (v/v/v) mixture of 2-propanol, water, and triethylamine (pH 8.5). For the positive-ion mode, a 50:50:0.03 (v/v/v) mixture of 2-propanol, water, and 30 mM ammonium acetate adjusted with acetic acid to pH 4.5 was used. The samples were sprayed at a flow rate of 2 µL·min-1. Capillary entrance voltage was set to 3.8 kV and drying gas temperature to 150 °C. The instrument was externally calibrated with appropriate standards. Infrared Multiphoton Dissociation. The unfocused beam of a 25-W, 10.6-µm CO2 laser (Synrad) was directed through the center of the trap. The duration of laser irradiation was adapted to generate optimal fragmentation and varied between 10 and 80 ms. Fragment ions were detected after a delay of 0.5 ms. If the laser beam did not hit the ions due to excited ion motion, they were cooled down by a short pulse of argon. Argon was also used as collision gas for sustained off-resonance irradiation collision induced dissociation (SORI-CID) at a peak pressure of ∼10-6 mbar. An average of at least 20 transients composed of 1M data points. RESULTS AND DISCUSSION Phosphatidyl-myo-inositol mannosides constitute a group of phospholipids ubiquitously found in the mycobacterial cell envelope. M. bovis BCG 1173P2 (the Pasteur strain), Mycobacterium smegmatis ATCC-607, and M. tuberculosis H37Rv ATCC-27294 were found to mainly contain two PIM families, the dimmanosylated (PIM2) and the hexamannosylated (PIM6) ones.25 PIM1, PIM3, PIM4, and PIM5 were always observed in very small amounts, suggesting that they are biosynthetic intermediates. These glyco forms exist as different acylated entities, the tri- and tetra-acylated molecules being the most abundant. Figure 1B illustrates the structure of PIM2 with the potential acylation sites, R1-R4. These (25) Gilleron, M.; Quesniaux, V. F.; Puzo, G. J. Biol. Chem. 2003, 278, 2988029889.

Figure 1. MALDI-TOF-MS analysis of M. bovis BCG PIM2 in negative-ion mode (A) and structure of PIM2 (B). PIM2 can be monoup to tetraacylated (R1, R2, R3, and R4 correspond to fatty acyl substituents).

sites were clearly elucidated by NMR study as being both positions of Gro, the C3 of the Ins unit and the C6 of the Manp linked to the C2 of the Ins unit. A combination of NMR and MS analyses has established that the two fatty acid residues of PIM2, mainly 2C16 and C16/C19, are predominantly located on Gro. The triacylated form of PIM2 (AcPIM2) shows a major acyl form containing two C16 and one C19, located on both positions of the Gro and on the Manp unit. The tetraacylated form, Ac2PIM2, was present predominantly as three populations, bearing either 3C16/ 1C19, 2C16/2C19, or to a lesser extent 2C16/1C18/1C19. The MS results have led to the conclusion that the Gro is preferentially acylated by C16/C19, the Manp unit by one C16, even if a weak variation exits.22 In contrast, the nature of the fatty acid present on the Ins appeared variable, being C16, C18, or C19. However, this conclusion did not arise from the direct ESI-MS analysis of the acetolysis reaction products of the purified PIM2 species22 but was rather a deduction of the comparative analysis of the different PIM2 acyl forms, bearing or not a fatty acyl group on the Ins. Indeed, no diagnostic ion fragments could be observed allowing discriminating between acyl appendages located on the Manp or the Ins. In the present study, to tentatively get these informative diagnostic ions, we investigated different MS ionization modes (MALDI and ESI) and used MS/MS fragmentation from both positive and negative pseudomolecular ions. MS/MS Fragmentation Pathway of Negative PIM2 Pseudomolecular Ions. We first took advantage of the phosphate residue present on the PIM molecule to obtain ESI (not shown) and MALDI (using HABA as a matrix) (Figure 1A) mass spectra in the negative mode. Both spectra were similar and exclusively composed of peaks assigned to single-charge deprotonated molecular ions [M - H]-, typifying the different PIM2 acyl forms. The most intense peaks were assigned to monoacylated form of PIM2 (lyso-PIM2) at m/z 895.4 (1C16) and 937.4 (1C19), diacylated form of PIM2 (PIM2) at m/z 1133.6 (2C16) and 1175.7 (1C16/1C19), triacylated form of PIM2 (AcPIM2) at m/z 1413.9 (2C16/1C19), and tetraacylated form of PIM2 (Ac2PIM2) at m/z 1652.1 (3C16/1C19), 1680.2 (2C16/1C18/1C19), and 1694.2 (2C16/2C19). As expected, both ionization methods revealed the same PIM2 acyl forms, indicating that ESI and MALDI are suitable ionization methods for PIM acyl form analysis. Then, the location of the different acyl appendages was investigated by MS/MS experiments. Analytical Chemistry, Vol. 78, No. 24, December 15, 2006

8545

Figure 2. Negative ion ESI-IRMPD-MS (A) and MALDI-TOF/TOF CID MS (B) spectra of Ac2PIM2 (2C16 and 2C19) and fragmentation pathways (C). In italics, a nonassigned ion (m/z 588.9).

Figure 2A shows the infrared multiphoton dissociation Fourier transform mass spectrum (IRMPD FT-MS) of the negative precursor ion of Ac2PIM2 (2C16, 2C19) at m/z 1694.2 generated by ESI. In the high-mass range, fragment ions at m/z 1082.6 arose from the fragmentation of the ester linkage between the phosphate and the Gro with the transfer of one proton, the negative charge being located on the phosphate. The following structure, Ac2Man2Ins-P (Figure 2C), was proposed for these ions, resulting from the loss of Ac2Gro. The mass of Ac2Manp2-Ins-P revealed that it bears one C16 and one C19. Consequently, the two remaining fatty acid residues (one C16 and one C19) are located on the Gro. As expected, this ion fragment was shifted to m/z 802.3 in the IRMPD spectra of the precursor ions of AcPIM2 (2C16/1C19) at m/z 1413.9 (not shown), in agreement with a Gro acylated by one C16 and one C19 and Manp2-Ins-P bearing one C16. Fragment ions that retain the Gro moiety are also present in the spectrum at m/z 153.0, 433.3, and 851.6. The fragment ions at m/z 153.0 were assigned to anhydroGro-P. The fragment ions at m/z 433.3, assigned to monoacylated (C19) anhydroGro-P, were more informative as they allowed the relative positioning of the fatty acids on the Gro. Indeed, it is well-established that the fatty acid located on C2 of Gro is more easily lost than that on the C1.26 Consequently, it can be proposed that the C19 is located at position 1 of the Gro, while the C16 is on C2, in contrast with what we previously suggested.22 Finally, the fragment ions at m/z 851.6 were assigned to Ac2Ins-P-Gro (with 1C16/1C19). However, there was no evidence to locate the C16 either on the Gro or on the Ins. These fragment ions at m/z 851.6 were absent in the CID TOFMS spectrum (Figure 2B) while the ions at m/z 153.0 and 433.3 were present. The latter ions were observed whatever the PIM2 acyl form analyzed (excepted the mono(C16)-acylated form of PIM2) revealing that the C19 systematically acylated the C1 of the Gro. Finally, fragments ions at m/z 255.2 and 297.3 were assigned to the carboxylate fatty acids C16 and C19, respectively. All these ions (except the m/z 851.6) are also observed in the CID MS (26) Hsu, F. F.; Turk, J. J. Am. Soc. Mass Spectrom. 2000, 11, 986-999.

8546 Analytical Chemistry, Vol. 78, No. 24, December 15, 2006

Figure 3. Full-scan positive-ion MALDI-TOF mass spectra of PIM2 using DHB matrix without (A) or with ion-exchange beads (B). In both spectra, the non-underlined series of peaks correspond to Na+ cationized PIM2 (PIM2Na + Na)+. The underlined series of peaks correspond in (A) to K+-cationized PIM2 (PIM2Na + K)+ and in (B) to protonated molecular species (PIM2Na + H)+.

spectrum of MALDI-TOF precursor ions (Figure 2B). This latter spectrum shows fragment ions at m/z 1438.0 and 1395.9 arising from the loss of C16 and C19 fatty acids, which were missing in the IRMPD FT-MS spectrum (Figure 2A). In addition, it is noticeable that MALDI-TOF MS/MS experiments (Figure 2B) generated more fragment ions than IRMPD FT-MS experiments in agreement with the collision energy. In summary, whatever the dissociation process used (CID or IRMPD), the MS/MS fragmentation pathway allowed us to determine the relative localization of the fatty acids on both positions of the Gro. However, these experiments were not appropriate to distinguish the fatty acid residues located on the Ins or the Manp. Then, this structural information was investigated by CID or IRMPD MS/MS experiments on positive precursor ions. MS/MS Fragmentation Pathway of Positive PIM2 Pseudomolecular Ions. PIM2 were analyzed by MALDI-TOF in positive mode, using HABA as a matrix (Figure 3A). Two series of peaks were observed, one corresponding to cationized sodiated PIM2 (PIM2Na + Na)+ and the other one cationized by K+ instead of Na+ (PIM2Na+K)+. For example, the peaks at m/z 1459.9 and 1475.9 correspond to the peak at m/z 1413.9 in negative mode with shifts of 46 (2Na) or 62 (Na + K) mass units, respectively. The most intense peaks were assigned to the cationized sodiated PIM2 (PIM2Na + Na)+ of monoacylated form of PIM2 (lyso-PIM2) at m/z 941.4 (C16), diacylated form of PIM2 (PIM2) at m/z 1179.6 (2C16) and 1221.7 (1C16/1C19), triacylated form of PIM2 (AcPIM2) at m/z 1459.9 (2C16/1C19), and tetraacylated form of PIM2 (Ac2PIM2) at m/z 1698.1 (3C16/1C19) and 1740.2 (2C16/2C19). A second series of peaks, separated from the previous one by 16 mass units, corresponded to (PIM2Na + K)+ of monoacylated form of PIM2 (lyso-PIM2) at m/z 957.3 (C16), diacylated form of PIM2 (PIM2) at m/z 1195.6 (2C16) and 1237.6 (1C16/1C19), triacylated

Figure 4. Positive ion MALDI-TOF/TOF CID MS spectrum of the cationized sodiated precursor ion (PIM2Na + Na)+ of the tetraacylated form of PIM2 (Ac2PIM2) (2C16 and 2C19) at m/z 1740.2 (A) and CID fragmentation pathways (B).

Figure 5. Positive ion MALDI-TOF/TOF CID MS spectrum of the protonated precursor ion (PIM2Na + H)+ of the tetraacylated form of PIM2 (Ac2-PIM2) (2C16 and 2C19) at m/z 1718.2 (A) and CID fragmentation pathways (B).

form of PIM2 (AcPIM2) at m/z 1475.9 (2C16/1C19), and tetraacylated form of PIM2 (Ac2PIM2) at m/z 1714.1 (3C16/1C19) and 1756.1 (2C16/2C19). It has been reported that the fragmentation pathway of the cationized carbohydrate ions is different from that of the protonated molecular species. Indeed, protonated carbohydrates fragment more easily than the metal-cationized species. PIM2 acyl forms were treated with H+ form cation-exchange resin beads in order to generate protonated molecular species by exchanging Na+ or K+ by H+. The resulting MALDI-TOF mass spectrum, recorded using DHB as matrix, was still very complex (Figure 3B). Besides the major peaks previously assigned to cationized sodiated PIM2 (PIM2Na + Na)+, a new set of peaks were observed, separated from the previous one by 22 mass units and corresponding to protonated molecular species (PIM2Na + H)+ of monoacylated form of PIM2 (lyso-PIM2) at m/z 919.4 (C16), diacylated form of PIM2 (PIM2) at m/z 1157.6 (2C16) and 1199.7 (1C16/1C19), triacylated form of PIM2 (AcPIM2) at m/z 1437.9 (2C16/1C19), and tetraacylated form of PIM2 (Ac2PIM2) at m/z 1676.1 (3C16/1C19) and 1718.2 (2C16/2C19). In summary, according to the sample preparation, three types of pseudomolecular ions could be obtained, either cationized by Na+, by K+, or by H+, and these different types of pseudomolecular ions were analyzed by MS/MS to tentatively generate diagnostic fragment ions allowing us to determine the location of the different acyl appendages. Figure 4A shows the CID MS spectrum of the cationized precursor ion (Ac2PIM2Na + Na)+ at m/z 1740.2 with 2C16 and 2C19 as acyl appendages. Figure 4B summarizes the fragmentation pathways. In the high-mass range, the CID mass spectrum is dominated by fragment ions at m/z 1483.9 and 1441.9 arising from the loss of C16 and C19 fatty acids, respectively. All the other fragment ions arise from the cleavage of the phosphate linkage, leading to fragment ions containing either the P-Gro or the ManIns-P moieties (Figure 4B). The major fragment ions A at m/z 1147.6 were assigned to (Ac2Man2-Ins-PNa + Na)+. This structure was supported by the analysis of the MS/MS spectrum of the precursor ion (AcPIM2Na + Na)+ at m/z 1459.9 (2 C16, 1 C19)

where the fragment ions at m/z 1147.6 were downshifted at m/z 867.3 (not shown). We next postulated that the fragment ions at m/z 967.5 (Ax) and 729.3 (Ay) arise from the (Ac2Man2-Ins-PNa + Na)+ ions by the loss of Manp or acylated C16-Manp, respectively. So, the mass value of the fragment ions Ay at m/z 729.3 (Ac1Man2-Ins-PNa + Na)+ appear as diagnostic fragment ions to positively identify the structure of the fatty acid located on the Manp unit. Fragment ions Ax and Ay were shifted as expected to m/z 687.2 and 449.0 when the precursor ion (AcPIM2Na + Na)+ at m/z 1459.9 (2C16, 1C19) was analyzed (not shown). Moreover, these diagnostic ions Ax and Ay were shifted at m/z 925.4 and 687.2 when the precursor ion (Ac2PIM2Na + Na)+ at m/z 1698.1 (3 C16, 1 C19) was analyzed (not shown), indicating again the loss of Manp or acylated C16-Manp respectively. Taken together, these data reveal that the Manp is selectively acylated by a C16 acyl appendage while indicating by deduction that the Ins can be acylated by one C16, one C18, or one C19 acyl appendage. Finally, the fragment ions B at m/z 735.5 were assigned to (Ac2Gro-P + 2Na)+ in which the acyl appendages are one C16 and one C19. These fragment ions can lose one C16 to give the ion Bs at m/z 479.3 or one C19 to give the ion Bt at m/z 437.2. In this case, the loss of a fatty acid is not restricted to a specific position. So, this method is not suitable to determine the relative distribution of the fatty acyl appendages on the Gro. The remaining fragment ions at m/z 142.9 were assigned to (PNa + Na)+, and these fragment ions can lose one molecule of water to give the ion at m/z 124.9. In conclusion, a main advantage of MS/MS spectra on cationized molecular ions is the presence of the diagnostic fragment ions Ay that allow identifying the fatty acyl appendage located on the Manp unit. Figure 5A shows the MS/MS mass spectrum of protonated (Ac2PIM2Na + H)+ (2C16 and 2C19) at m/z 1718.2. The fragmentation pathways (Figure 5B) drastically differ from those of the cationized species (Figure 4B), resulting in a simpler CID mass spectrum. The high-mass range is dominated by two peaks at m/z 1556.1 and 1317.9 resulting from the loss of anhydroManp (fragment C) Analytical Chemistry, Vol. 78, No. 24, December 15, 2006

8547

and monoacylated (C16) anhydroManp (fragment D). Then, the presence of the fragment ions D at m/z 1317.9 characterize again the presence of a C16 acyl appendage on the Manp unit. Similarly, the fragment ions D at m/z 1275.8 generated from the protonated molecular ion of Ac2PIM2 at m/z 1676.1 (3C16, 1C19) arise also from the loss of monoacylated (C16) anhydroManp. All these data indicate that Manp unit is selectively acylated by a C16. The middle mass range is characterized by a peak at m/z 1125.6, arising from the cleavage of the phosphate bond and leading to the fragment ions A (Ac2Man2-Ins-PNa + H)+. Again, this fragment can lose anhydroManp or monoacylated (C16) anhydroManp, yielding the fragment ions Av and Aw at m/z 963.6 (Ac2Man-Ins-PNa + H)+ and 725.3 (Ac1Man-Ins-PNa + H)+, respectively. Then, fragment ions Az at m/z 563.3 assigned to (Ac1Ins-PNa + H)+ may arise from (Ac2Man-Ins-PNa + H)+ by the loss of Ac-anhydroManp or from (Ac1Man-Ins-PNa + H)+ by the loss of anhydroManp. Ions Az at m/z 563.3 typify the presence of a C19 acyl appendage on the Ins unit. The assignment of these fragment ions is supported by the fact that it is shifted to m/z 521.3 or 283.1 in the cases of the precursor ions of Ac2PIM2 (3C16, 1C19) at m/z 1676.1 or of AcPIM2 (2C16, 1C19) at 1437.9, respectively. Ions Az at m/z 521.3 typify a C16 acyl appendage on the Ins unit whereas ions Az at m/z 283.1 indicate a nonacylated Ins. Thus, fragment ions Az (Ac1Ins-PNa + H)+ appear as diagnostic ions for identifying the fatty acyl group located on the Ins ring. This type of fragment ions was not observed when negative or cationized molecular precursor ions were analyzed. ESI FT-MS provided in the positive-ion mode the same molecular ion species as observed with MALDI-TOF-MS. However, IRMPD-MS of protonated PIM2Na revealed, according to fragmentations described on Figure 5B, the fragment ions C and Av but not the fragments D, Aw, and Az. It is worth mentioning that using SORI-CID MS, again only C and Av fragments were observed. A possible reason for these differences to the TOF MS/ MS experiments may be that the lifetime of the intermediates is too short to be detected with FT-MS.

8548

Analytical Chemistry, Vol. 78, No. 24, December 15, 2006

CONCLUSIONS The aim of the present study was to establish MS strategies to determine the relative distribution of fatty acyl appendages on the different acylation sites for each PIM2 acyl form. MS/MS experiments by IRMPD of negative precursor ions allowed locating the acyl groups on each position of the Gro, but were unsuccessful in assigning their relative distribution on the Manp versus Ins. However, MS/MS by CID on protonated molecular ions appeared to be a powerful approach to locate the fatty acyl appendages on both Manp and Ins residues. Using an analytical approach including MS/MS of negative and protonated molecular ions, we are now able to identify the fatty acyl appendages esterifying the different residues, directly on native PIM mixtures. Taken together, these data confirm that, in triacylated form of PIM2 (AcPIM2), the third fatty acid residue, assigned to C16, is exclusively located on the Manp unit. Moreover, the analysis of the different tetraacylated forms (Ac2PIM2) shows that Manp is selectively acylated by C16 while Ins can be acylated by different fatty acyl appendage (C16, C18, or C19). The combined MS approach developed here should constitute in the future a very useful, fast, and sensitive procedure to analyze PIM fractions without any chemical derivatization or deep purification steps. ACKNOWLEDGMENT M.G. and G.P. were funded by the Agence Nationale de la Recherche (ANR-05-MIIM-006-02 and ANR-05-MIIM-038-04), by the Centre National de la Recherche Scientifique (CNRS). and are supported by the NIH (5R01AI64798-2) and the European Union (Cluster for a tuberculosis vaccine, LSHP-CT-2003-503367). We gratefully acknowledge Dr. B. Monsarrat (IPBS, CNRS, Toulouse) for MALDI-TOF MS facilities. The MALDI-TOF spectrometer was financed by the Re´gion Midi-Pyre´ne´es and the Ge´nopoˆle Toulouse Midi Pyre´ne´es. Received for review August 23, 2006. Accepted September 29, 2006. AC061574A