Complete Structural Elucidation of Triacylglycerols by Tandem Sector

Sep 12, 1998 - Use of [C18] Oleic Acid and Mass Isotopomer Distribution Analysis to Study Synthesis of Plasma Triglycerides In Vivo: Analytical and Ex...
0 downloads 8 Views 201KB Size
Download PDF
Anal. Chem. 1998, 70, 4417-4426

Complete Structural Elucidation of Triacylglycerols by Tandem Sector Mass Spectrometry Changfu Cheng and Michael L. Gross*

Department of Chemistry, Washington University, St. Louis, Missouri 63130 Ernst Pittenauer*

Federal Office & Research Center for Agriculture, Institute for Agricultural Ecology, Vienna, Austria

We developed a method to elucidate the complete structure of triacylglycerols by means of high-energy collisional activation tandem mass spectrometry (MS/MS). Both ESI- and FAB-produced [M + NH4]+ and [M + met.]+ ions (where met. ) Li, Na, and Cs) of triacylglycerols undergo charge-remote and charge-driven fragmentations. We emphasize the study of fragment ions from ESI-produced [M + NH4]+ and [M + Na]+ ions and FAB-produced [M + Na]+ ions. ESI-produced [M + NH4]+ ions fragment to produce four types of ions, [M + NH4 - RnCOONH4]+, [RnCO + 128]+, [RnCO + 74]+, and RnCO+ ions, from which the carbon number and the degree of unsaturation of each acyl group are obtained. In addition, three series of ions are produced by charge-remote fragmentations (CRFs), and analysis of their patterns gives the position and the number of double bonds on the acyl groups. Information about the position of each acyl group on the glycerol backbone, however, is not provided by collisionally activated dissociation of [M + NH4]+ ions. On the other hand, ESI- and FAB-produced [M + Na]+ ions fragment to form eight types of ions (named A-J ions) that, like those produced by CRF, are highly structurally informative. The absence of certain series members also carries useful structural information. Interpretation of these patterns enables one to obtain the number of carbons, degrees of unsaturation, and location of double bonds, as well as the positions of acyl groups on the glycerol backbone. Triacylglycerols (TAGs) play an important role in nutrition and other biological processes. They are the primary means of energy storage in animals and humans, and their hormonally controlled hydrolysis and oxidation release energy to meet the energygeneration needs of organisms.1 A major source of TAGs is seed oils, which are found as renewable agricultural raw materials; these materials have applications in technology and nutrition.2 * Corresponding author. Tel. 314-935-4814; Fax 314-935-7484; E-mail mgross@ wuchem.wustl.edu. (1) Mathews, C. K.; van Holde, K. E. in Biochemistry; The Benjamin/Cummings Publishing Co., Inc.: Redwood City, CA, 1990; pp 571-578. S0003-2700(98)00519-8 CCC: $15.00 Published on Web 09/12/1998

© 1998 American Chemical Society

Because triacylglycerols are important in biology, many analytical methods involving mass spectrometry (MS) have been used to determine their structures; the MS methods include electron ionization (EI),3-5 chemical ionization (CI),3,6-18 desorption chemical ionization (DCI),19-21 fast atom bombardment (FAB),22-24 field desorption (FD),25,26 thermospray (TSP),27,28 electrospray ionization (ESI),29-31 and atmospheric pressure (2) Barrett, L. W.; Sperling, L. H.; Murphy, C. J. J. Am. Oil Chem. Soc. 1993, 70, 523-534. (3) Murphy, R. C. In Handbook of Lipid Research 7: Mass Spectrometry of Lipids; Snyder F., Ed.; Plenum Press: New York, 1993; p 213. (4) de Mirbuker, M.; Blomberg, L. G.; Olsson, N. U.; Bergqvist, M.; Herslof, B. G.; Jacobs, F. A. Lipids 1992, 27, 436-441. (5) Kallio, H.; Laasko, P.; Huopalathi, R. Linko, R. R.; Oksman, P. Anal. Chem. 1989, 61, 698-700. (6) Johansson, A.; Laakso, P.; Kallio, H. Z. Lebensm. Unters. Forsch. 1997, 204, 308-315. (7) Evershed, R. P. J. Am. Soc. Mass Spectrom. 1996, 7, 350-361. (8) Manninen, P.; Laakso, P.; Kallio, H. Lipids 1995, 30, 665-671. (9) Manninen, P.; Laakso, P.; Kallio, H. J. Am. Oil Chem. Soc. 1995, 72, 10011008. (10) Kallio, H.; Rua, P. J. Am. Oil Chem. Soc. 1994, 71, 985-992. (11) Cheung, M.; Young, A. B.; Harrison, A. G. J. Am. Soc. Mass Spectrom. 1994, 5, 553-557. (12) Laakso, P.; Kallio, H. J. Am. Oil Chem. Soc. 1993, 70, 1173-1176. (13) Laakso, P.; Kallio, H. J. Am. Oil Chem. Soc. 1993, 70, 1161-1171. (14) Kallio, H.; Currie, G. Lipids 1993, 28, 207-215. (15) Currie, G.; Kallio, H. Lipids 1993, 28, 217-222. (16) Taylor, D. C.; Gibblin, E. M.; Reed, D. W.; Hogge, L. R.; Olson, D. J.; MacKenzie, S. L. J. Am. Oil Chem. Soc. 1995, 72, 305-308. (17) Huang, A. S.; Delano, G. M.; Pidel, A.; Janes; L. E.; Softly, B. J.; Templeman, G. J. J. Agric. Food Chem. 1994, 42, 453-460. (18) Huang, A. S.; Robinson, L. R.; Gursky, L. G.; Profita, R.; Sabidong, C. G. J. Agric. Food Chem. 1994, 42, 468-473. (19) Laakso, P.; Kallio, H. Lipids 1996, 31, 33-42. (20) Stroobant, V.; Rozenberg, R.; el Monier, B.; Deffense, E.; de Hoffmann, E. J. Am. Soc. Mass Spectrom. 1995, 6, 498-506. (21) Anderson, M. A.; Collier, L.; Dillipane, R.; Ayorinde, F. O. J. Am. Oil Chem. Soc. 1993, 70, 905-908. (22) Lamberto, M.; Saitta, M. J. Am. Oil Chem. Soc. 1995, 72, 867-871. (23) Hori, M.; Sahashi, Y.; Koike, S.; Yamaoka, R.; Sago, M. Anal. Sci. 1994, 10, 719-724. (24) Evans, C.; Traldi, P.; Bambigiotti-Alberti, M.; Gianelli, V.; Coran, S. A.; Vincieri, F. F. Biol. Mass Spectrom. 1991, 20, 351-356. (25) Lehmann, W. D.; Kessler, M. Biomed. Mass Spectrom. 1983, 10, 220-226. (26) Evans, N.; Games, D. E.; Harwood, J. L.; Jackson, A. H. Biochem. Soc. Trans. 1974, 2, 1091-1092. (27) Sundin, P.; Larsson, P.; Wese´n, C.; Odham, G. Biol. Mass Spectrom. 1992, 21, 633-641. (28) Kim, H.-Y.; Salem, N., Jr. Anal. Chem. 1987, 59, 722-726.

Analytical Chemistry, Vol. 70, No. 20, October 15, 1998 4417

chemical ionization (APCI).32-38 Although all these techniques are able to provide the molecular weight of the intact molecule and, when combined with tandem mass spectrometry, some information about the acyl groups, it has not been possible to perform a complete structural determination with mass spectrometry alone. A complete determination includes elucidating the structure of each acyl group (carbon number, degree of unsaturation, position of branches or double bonds, etc.) and locating the three acyl groups on the glycerol backbone (sn-1, sn-2, or sn-3). For the former, charge-remote fragmentations (CRFs) can play a role because this technique has proven very useful for the structural elucidation of a variety of long-chain molecules, including fatty acids and their esters, fatty alcohols, surfactants, phospholipids, and peptides.39-59 CRFs are gas-phase dissociations that are analogous to thermal processes,39 and they involve losses of CnH2n+1 and CnH2n+2 from precursor ions. There are two proposed mechanisms for reactions along an alkyl chain,40-49 but that subject is not within the scope of this article. In a recent example utilizing CRFs, the structural elucidation of sulfoquinovosyl, monogalactosyl, and digalactosyl diacylglycerols was carried out by Kim et al.;60 the structure of the sugar head and the two acyl groups (including double-bond positions) can be determined completely. (29) Duffin, K. L.; Henion, J. D.; Shieh, J. J. Anal. Chem. 1991, 63, 1781-1788. (30) Myher, J. J.; Kuksis, A.; Geher, K.; Park, P. W. Lipids 1996, 31, 207-215. (31) Pittenauer, E.; Aichinger, T.; de Hueber, K.; Bailer, J. Proceedings of the 44th ASMS Conference on Mass Spectrometry and Allied Topics, Portland, OR, May 12-16, 1996; p 928. (32) Neff, W. E.; Byrdwell, W. C. J. Am. Oil Chem. Soc. 1995, 72, 1185-1191. (33) Byrdwell, W. C.; Emken, E. A. Lipids 1995, 30, 173-175. (34) Neff, W. E.; Byrdwell, W. C. J. Liq. Chromatogr. 1995, 18, 4165-4181. (35) Byrdwell, W. C.; Neff, W. E. J. Liq. Chromatogr. 1996, 18, 2203-2225. (36) McIntyre, D.; Fischer, S. Proceedings of the 44th ASMS Conference on Mass Spectrometry and Allied Topics, Portland, OR, May 12-16, 1996; p 289. (37) Byrdwell, W. C.; Emken, E. A.; Odlof, R. O. Lipids 1996, 31, 919-935. (38) Mottram, H. R.; Evershed, R. P. Tetrahedron Lett. 1996, 37, 8593-8596. (39) Adams, J.; Gross, M. L. J. Am. Chem. Soc. 1989, 111, 435-440. (40) Jensen, N. J.; Tomer, K. B.; Gross, M. L. J. Am. Chem. Soc. 1985, 107, 1863-1868. (41) Cordero, N. N.; Wesdemiotis, C. Anal. Chem. 1994, 66, 861-866. (42) Contado, M. J.; Adams, N. J.; Jensen, N. J.; Gross, M. L. J. Am. Soc. Mass Spectrom. 1991, 2, 180-183. (43) Cheng, C.; Pittenauer, E.; Gross, M. L. submitted to J. Am. Soc. Mass Spectrom. 1998, 9, 840-844. (44) Claeys, M.; Van den Heuvel, H.; Claereboudt, J.; Corthout, J.; Pieters, L.; Vlietinck, A. J. Biol. Mass Spectrom. 1993, 22, 647-653. (45) Claeys, M.; Van den Heuvel, H. Biol. Mass Spectrom. 1994, 23, 20-26. (46) Claeys, M.; Nizigiyimana, L.; Van den Heuvel, H.; Derrick, P. J. Rapid Commun. Mass Spectrom. 1996, 10, 770-774. (47) Griffiths, W. J.; Yang, Y.; Lindgren, J. A.; Sjo¨vall, J. Rapid Commun. Mass Spectrom. 1996, 10, 21-28. (48) Wysocki, V. H.; Bier, M. E.; Cooks, R. G. Org. Mass Spectrom. 1988, 23, 627-633. (49) Wysocki, V. H.; Ross, M. M. Int. J. Mass Spectrom. Ion Processes 1991, 104, 179-211. (50) Gross, M. L. Int. J. Mass Spectrom Ion Processes 1992, 118/119, 137-165. (51) Jensen, N. J.; Gross, M. L. Mass Spectrom. Rev. 1987, 6, 497-536. (52) Adams, J. Mass Spectrom. Rev. 1990, 9, 141-186. (53) Tomer, K. B.; Crow, F. W.; Gross, M. L. J. Am. Chem. Soc. 1983, 105, 5487-5488. (54) Adams, J.; Gross, M. L. J. Am. Chem. Soc. 1986, 108, 6915-6921. (55) Adams, J.; Gross, M. L. Anal. Chem. 1987, 59, 1576-1582. (56) Jensen, N. J.; Tomer, K. B.; Gross, M. L. Anal. Chem. 1985, 57, 2018. (57) Deterding, L. J.; Gross, M. L. Anal. Chim. Acta 1987, 200, 431. (58) Crockett, J. S.; Gross, M. L.; Christie, W. W.; Holman, R. T. J. Am. Soc. Mass Spectrom. 1990, 1, 183-190. (59) Jensen, N. J.; Hass, G. W.; Gross, M. L. Org. Mass Spectrom. 1992, 27, 423-427. (60) Kim, Y. H.; Yoo, J. S.; Kim, M. S. J. Mass Spectrom. 1997, 32, 968-977.

4418 Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

So far, CRFs for determining the structures of triacylglycerols have not been systematically applied, although interesting results have been demonstrated by previous researchers.24,31 In this article, we report the fragmentations, including CRFs, of triacylglycerols charged with ammonium and alkali metal ions and formed under ESI and FAB conditions. Interpretation of the MS/MS data allows one to characterize completely the structure of these molecules; that is, their molecular weights, the carbon number and the degree of unsaturation of each acyl group, the position of double bonds along acyl groups, and the positions of acyl groups on the glycerol backbone. No effort, however, was made to determine chirality of TAGs by MS/MS. EXPERIMENTAL SECTION All synthetic triacylglycerols, including tristearoyl-, trilinolenoyl-, 1,2-dipalmitoyl-3-oleoyl-, 1,3-dipalmitoyl-2-oleoyl-, 1,2-dioleoyl3-stearoyl-, 1,3-dioleoyl-2-stearoyl-, and 1-palmitoyl-2-oleoyl-3stearoylglycerols, were purchased from Sigma (St. Louis, MO) and used without further purification. All matrix and solvent compounds, including 3-nitrobenzyl alcohol, NaI, NaOAc, NH4I, NH4OAc, CsI, LiI, CH3OH, and CHCl3, were obtained from either Sigma or Aldrich (Milwaukee, WI). ESI-MS/MS experiments were carried out with a ZAB-T foursector tandem mass spectrometer manufactured by VG Analytical (Manchester, UK), equipped with a VG electrosprary source (Micromass, Manchester, UK).61 A triacylglycerol and the appropriate salt (NH4OAc, NaOAc, etc.) were dissolved in CHCl3 and CH3OH (7:3 volume ratio). The final concentration of the triacylglycerol was 10 µM, and those of NH4OAc and NaOAc were 10 and 1 µM, respectively (low concentration of NaOAc was used because high concentration of sodium salt makes the signal unstable).29 The solution was infused in the continuous mode at a flow rate of 10 µL/min through a Harvard model 22 syringe pump (Harvard Apparatus, South Natick, MA). The spray needle was maintained at 8000 V, and the counter electrode (pepper pot) potential was 5000 V. The sampling cone, skimmer lens, skimmer, hexapole, and ring electrode were 4200, 4160, 4150, 4150, and 4120 V, respectively. Nitrogen was used separately as both bath and nebulizer gas with flow rates of 400 and 12 L/h, respectively. The bath gas temperature was maintained at 80 °C. Argon was used as the collision gas, and sufficient gas was added to attenuate the main beam by 60%. The collision cell was floated to 2 kV (resulting in ELAB ) 2 keV). The product ions were analyzed with MS2; approximately 40 15-s scans at a mass resolving power of ∼1000 (full width at half-maximum) were taken. ESI-linked scan (B/E ) constant) experiments were carried out with a Finnigan MAT 95S double-focusing instrument (Finnigan MAT, Bremen, Germany), fitted with a second-generation atmospheric pressure ionization (API) source. The sample solution (described above) was continuously infused with a Harvard Apparatus 22 syringe pump at a flow rate of 50 µL/min. The source conditions for producing abundant precursor ions were as follow: sheath gas pressure, 6 bar N2; temperature of the heated capillary, 250 °C; capillary exit voltage, -45 V; tube lens voltage, +55 V; skimmer voltage, -2.5 V; and rf octapole, -5 V. Scans of product ions formed by collisionally activated dissociation (CAD) (61) Gross, M. L. Tandem Mass Spectrometry: Multisector Magnetic Instruments. In Methods in Enzymology; McCloskey, J. A., Ed.; Academic Press: San Diego, CA, 1990; Vol. 193, pp 131-153.

Figure 1. CAD spectra of ESI-produced TAG [M + NH4]+ ions with identical acyl groups: (A) tristearoylglycerol (18:0/18:0/18:0, MW ) 890.8) and (B) trilinolenoylglycerol (18:3/18:3/18:3, MW ) 872.8). The position of a double bond is indicated by the “)” sign, whereas the peak corresponding to an allylic cleavage is labeled with “an”.

in the first field-free region were taken with a grounded gas collision cell (ELAB ) 4.75 keV) with sufficient argon added to attenuate the main beam by 70-80%. All data were acquired in the profile mode (scan speed 10 s/100 u) and represent an average of 20 scans. All FAB-MS/MS experiments were carried out with the foursector tandem instrument. In a typical experiment, a small amount (∼1 µg) of a triacylglycerol was mixed with approximately 1 µL of matrix (3-nitrobenzyl alcohol saturated with NaI) on the FAB tip and bombarded with a high-energy (∼25 keV) Cs+ ion beam. The ions were accelerated by 8 kV, selected by the first stage (MS1) at a mass resolving power of ∼1500, and collided with helium in the collision cell between MS1 and MS2 with a 70% main beam attenuation. The collision cell was floated to 4 kV (ELAB ) 4 keV). The product ions were detected with MS2 using approximately 30 15-s scans at a mass resolving power of ∼1000 (full width at half-maximum). To investigate the fragmentation of the first generation of product ions, ESI source CAD-MS/MS experiments were carried out with the four-sector mass spectrometer; MS/MS/MS experiments under FAB conditions were carried out with the Kratos triple-analyzer mass spectrometer,62 following procedures described elsewhere.63 RESULTS AND DISCUSSION Fragmentation of triacylglycerols under EI and CI conditions has been studied extensively by early researchers.3 EI of TAGs (62) Gross, M. L.; Chess, E. K.; Lyon, P. A.; Crow, F. W.; Evans, S.; Tudge, H. Int. J. Mass Spectrom. Ion Phys. 1982, 42, 243-254. (63) Burinsky, D. J.; Cooks, R. G.; Chess, E. K.; Gross, M. L. Anal. Chem. 1982, 54, 295-299.

produces abundant [M - RnCOO]+, [RnCO + 128]+, [RnCO + 74]+, and RCO+ ions, whereas CI of TAGs gives abundant [M + H - RnCOOH]+, [RnCO + 74]+, and RCO+ ions (where n ) 1, 2, and 3). Generation of these ions allows one to calculate the carbon number and the degree of unsaturation of acyl groups. Both methods, however, cannot locate the position of double bonds in acyl groups or distinguish the sn-1, sn-2, or sn-3 substituents on the glycerol backbone.3 In the presence of an ammonium or alkali metal salt, ESI and FAB of TAGs produce abundant [M + cat.]+ ions, where cat. ) Li, Na, Cs, or NH4. Duffin et al.29 investigated the fragmentations of ESI-produced [M + NH4]+ ions of TAGs under low-energy collisional activation (CA), and their results enable one to calculate readily the length and double-bond number of each acyl group. Detailed structural information on each acyl group and its linkage to the glycerol backbone, however, was not obtainable with lowenergy collisional activation. Upon high-energy activation, ESIand FAB-produced [M + cat.]+ ions of TAGs fragment through a number of reaction channels. In addition to the same ions that were produced by low-energy CA, more structurally diagnostic product ions are produced by high-energy CA, including ions formed by charge-remote fragmentations. Fragmentation of ESI-Produced TAG [M + NH4]+ Ions. Collisional activation of ESI-produced [M + NH4]+ ions from TAG gives product ions from both charge-remote and charge-driven chemistry. Interestingly, the majority of the CRF ions are formed by losses of CnH2n+1, indicating that the products are, at least initially, long-chain distonic ions (Figure 1).64,65 Our previous (64) Hammerum, S. Mass Spectrom. Rev. 1988, 7, 123-193.

Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

4419

Figure 2. CAD spectra of ESI-produced [M + NH4]+ ions of 1-palmitoyl-2-oleoyl-3-stearoylglycerol (16:0/18:1/18:0, MW ) 860.8) obtained by different instruments: (A) an MS/MS spectrum from a four-sector mass spectrometer and (B) a linked-scan spectrum from a two-sector mass spectrometer. The peak corresponding to a vinyl cleavage is labeled with “v”.

studies showed that the internal energy of the precursor ions determines the nature of the products formed by CRF.43 Nevertheless, the pattern of the peaks corresponding to the CRF ions provides sufficient information to locate the double bond. A vinyl cleavage at the side proximal to the charge produces an ion by loss of CnH2n-3, indicating the position of the double bond. In addition to ions formed by CRFs, CA of the [M + NH4]+ ions of TAG produces [M + NH4 - RnCOONH4]+, [RnCO + 128]+, [RnCO + 74]+, and RnCO+ ions, presumably by charge-driven processes. If the compositions we assigned are correct, these ions should be structurally diagnostic. To test, we obtained CAD spectra of [M + NH4]+ ions of TAGs with identical and different acyl groups. [M + NH4]+ Ions of a TAG Containing Identical Acyl Groups. Collisional activation of the [M + NH4]+ ions of a TAG containing identical acyl groups produces three series of CRF ions with identical m/z values and four ions with m/z values corresponding to compositions of [M + NH4 - RCOONH4]+, [RCO + 128]+, [RCO + 74]+, and RCO+. For example, CAD of the ESI-produced [M + NH4]+ ions of tristearoylglycerol (18:0/18:0/18:0) gives rise to [M + NH4 - RCOONH4]+ at m/z 607, [RCO + 128]+ at m/z 395, [RCO + 74]+ at m/z 341, and RCO+ at m/z 267 (Figure 1A), whereas that of trilinolenoylglycerol (18:3/18:3/18:3) gives corresponding ions at m/z 595, 389, 335, and 261, respectively (Figure 1B). In both cases, a typical pattern of CRF ions is generated, and one can readily see that the former TAG contains saturated acyl groups and the latter has three double bonds on each acyl group. The array of peaks corresponding to ions of CRFs of the (65) Stirk, K. M.; Kiminkinen, L. K. M.; Kenttamaa, H. I. Chem. Rev. 1992, 92, 1649-1665.

4420 Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

former shows no interruptions, whereas that of the latter contains three interruptions. Further, one can unmistakably locate the double bonds on the alkyl chain of the latter (they are at C9, C12, and C15) by identifying and interpreting the three ions formed by allylic cleavages, a1, a2, and a3, and the three gaps associated with those ions (Figure 1B). [M + NH4]+ Ions of a TAG Containing Three Different Acyl Groups. Expectedly, CAD of ESI-produced [M + NH4]+ ions of TAGs containing three different acyl groups produces three different members of each of the ion types: [M + NH4 - RnCOONH4]+, [RnCO + 128]+, [RnCO + 74]+, and RnCO+ ions (where n ) 1, 2, and 3). For example, CAD of ESI-produced [M + NH4]+ ions of 1-palmitoyl-2-oleoyl-3-stearoylglycerol (16:0/18: 1/18:0) produces [M + NH4 - RnCOONH4]+ ions at m/z 607, 579, and 577; [RnCO + 128]+ ions at m/z 395, 393, and 367; [RnCO + 74]+ ions at m/z 341, 339, and 313; and RnCO+ ions at m/z 267, 265, and 239 (Figure 2B). From the m/z values of these ions, one can readily identify that the intact molecule contains two saturated acyl groups (C16:0 and C18:0) and one monounsaturated (C18:1) acyl group. In addition, three series of CRF ions are formed, although they are often not distinguishable because they overlap with each other when the neutrals lost from the precursor ion are identical. But when a CRF ion is formed by vinyl cleavage at the side proximal to the charge (labeled with “v” in Figure 2), a doublet is produced, and the onset of the doublet allows one to conclude that the position of the double bond is at C9. Therefore, the structures of the three acyl groups are elucidated by combining the information obtained from both charge-remote and charge-driven ions (to be 16:0, 18:0, and 18: 1∆9 respectively). The linkage information of the three acyl

Figure 3. Comparison of CAD spectra of the [M + Na]+ ions of tristearoylglycerol (18:0/18:0/18:0, MW ) 890.8) with different ionization methods: (A) with FAB and (B) with ESI.

Scheme 1 groups on the glycerol backbone, however, cannot be deciphered because there is no apparent discrimination in the formation of ions in the same group; that is, all three ions of each type, [M + NH4 - RnCOONH4]+, [RnCO + 128]+, [RnCO + 74]+, or RnCO+, are produced with nearly the same abundance (Figure 2B). Although the ions that arise by loss of R2COONH4 (Figure 2A and B) are slightly less abundant than those from losses of R1COONH4 and R3COONH4, we are not certain that this is always true. [M + NH4]+ Ions of TAG Isomers. From the above discussion, we predict that CAD of [M + NH4]+ ions of positional TAG isomers would give similar product ions, making it impossible to distinguish them. We found this to be true for the fragmentations of the [M + NH4]+ ion of two pairs of TAG isomers [1,2-dioleoyl3-stearoylglycerol (18:1/18:1/18:0) and 1,3-dioleoyl-2-stearoylglycerol (18:1/18:0/18:1) as well as 1,2-dipalmitoyl-3-oleoylglycerol (16:0/16:0/18:1) and 1,3-dipalmitoyl-2-oleoylglycerol (16:0/18:1/ 16:0)]. Product-ion spectra (not shown) of the isomers are nearly identical. [M + NH4]+ Ions of TAGs Obtained by Linked Scanning. Figure 2B is a product-ion spectrum obtained by linked scanning (B/E ) constant) with a two-sector mass spectrometer. The quality of the spectrum shows that linked scanning serves well in the analysis of TAGs because it provides almost the same structural information as does tandem four-sector mass spectrometry (Figure 2A). Nevertheless, it is difficult to locate a double bond by interpreting the product-ion spectrum from linked scanning because CRF ions are no longer singlets, owing to the inability to select a single precursor ion. For example, the position of the double bond is easily located in Figure 2A, but not in Figure 2B.

Possible Mechanisms for the Formation of Product Ions from [M + NH4]+ Ions of TAGs. Unlike product ions from CRF, all the Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

4421

Figure 4. Comparison of CAD spectra of the [M + Na]+ ions of 1-palmitoyl-2-oleoyl-3-stearoylglycerol (16:0/18:1/18:0, MW ) 860.8) with different ionization methods: (A) with FAB and (B) with ESI.

four types of ions, [M + NH4 - RnCOONH4]+, [RnCO + 128]+, [RnCO + 74]+, and RnCO+, do not contain ammonia, indicating that they are formed by charge-driven chemistry. The initial position of the charge, which is presumably carried by the ammonium ion, must be close to the glycerol backbone, which is the region containing nonbonding electrons. We propose that all the reactions are initialized by the interaction of a Lewis base functional group with the positive charge. For example, interaction of the sn-2 acyl group with the ammonium ion may give rise to the [M + NH4 - R2COONH4]+ and the R2CO+ ions (Scheme 1, pathways a and b). In both cases, ammonia was lost as part of the neutral. The [M + NH4 - R2COONH4]+ ion, a stable oxonium ion, may serve as the intermediate for the formation of other ions. For example, if a 1,4-H2 elimination occurs near the oxonium ion, the [R3CO + 128]+ ion is produced. ESI source CAD to give [M + NH4 - RnCOONH4]+ followed by MS/MS proved that the decomposition of this first-generation product ion gives the [RnCO + 128]+ as the major secondary product (spectra not shown). On the other hand, a ketene loss from the sn-1 acyl group produces the [R3CO + 74]+ ion (Scheme 1), and the pathway is similar to that proposed by earlier researchers3 when the ion is generated by EI or CI. We could not obtain evidence for the latter decomposition because the signal-to-noise ratio of the production spectrum of the [M + NH4 - RnCOONH4]+ was not sufficiently high. Fragmentation of FAB- and ESI-Produced TAG [M + met.]+ Ions. FAB or ESI of TAG that is in solution with an appropriate alkali metal salt produces abundant [M + met.]+ ions (where met. ) Li, Na, and Cs). Evans et al.24 studied the fragmentations of FAB-produced [M + Na]+ ions and reported abundant [M + Na - RnCOOH]+ ions and other product ions 4422 Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

that may be formed by CRF, although no spectra were shown in their article. We will discuss the fragmentation of only [M + Na]+ ions in this article because CAD of other alkali-metal-cationized ions gives similar structural information. Product Ions from FAB- and ESI-Produced Precursors. Similar to fragmentations of ESI-produced [M + NH4]+ ions, ESI- and FAB-produced [M + Na]+ ions of TAGs also fragment through charge-remote and charge-driven pathways. Interestingly, the CRFs of FAB-produced [M + Na]+ ions of TAGs are principally CnH2n+2 losses, whereas those of ESI-produced [M + Na]+ ions are CnH2n+1 losses (Figure 3). We explained these phenomena in terms of differences in internal energy distributions of ions formed by the two ionization modes.43 The relative abundances of product ions are also quite different: under FAB conditions, the most abundant product ion is formed by loss of a fatty acid (corresponding to peak B in Figure 3A), whereas for ESI-produced precursors, the most abundant product ion is that formed by methyl radical loss (Figure 3B). Furthermore, product ions formed by charge-driven processes of ESI-produced [M + Na]+ and [M + NH4]+ precursors are different not only in relative abundances but also in chemical compositions. Fragmentation of [M + Na]+ ions yields more structurally diagnostic ions (labeled A-J in Figure 3) than that of [M + NH4]+ ions (compare Figure 1A with Figure 3B). Nomenclature and Proposed Structures of Product Ions from [M + Na]+. In addition to product ions formed by CRFs along the alkyl chains, there are eight types of other product ions (A-J ions) that are obtained at relatively high abundances by collisional activation of [M + Na]+ precursors (Figures 3 and 4 and summarized in Table 1). When a TAG contains identical acyl groups, ions of a given type will have the same m/z value, whereas

Table 1. Proposed Structures for A-J Ions

Scheme 2

when a TAG contains three different acyl groups, some types of ions will have three different m/z values. For example, the B ions from the [M + Na]+ ions of tristearoylglycerol (18:0/18:0/ 18:0) (Figure 3) have the same m/z value (629.5), whereas the B ions from the [M + Na]+ ions of 1-palmitoyl-2-oleoyl-3-stearoylglycerol (16:0/18:1/18:0) (Figure 4) have three different m/z values, reflecting the carbon number and degree of unsaturation of each acyl group. Ions of the same type are differentiated by the subscript in their names (Table 1), with each digit denoting the location of the acyl group that is retained in the fragment. For example, the designation A12 means that the ion is formed by loss of the sn-3 acyl group with retention of the sn-1 and sn-2 acyl groups. Some of these ions have more than one structure. For example, there are two possible structures for a B13 ion because the double bond could form between sn-1 and sn-2, or sn-2 and sn-3 carbons on the glycerol backbone (Scheme 2B).

Starting with the D ions, we note that formation of the fragment ions involves elimination of two acyl groups, and the structures of the ions are not proved at this point. The mathematical relationships of m/z values of A-J ions and that of the [M + Na]+ ion are listed in Table 2, and one can easily use the formulation to identify product ions from a given spectrum. For example, given that M123 of the [M + Na]+ of 1-palmitoyl-2-oleoyl-3stearoylglycerol (16:0/18:1/18:0) is 883.7 u and that R1COOH, R2COOH, and R3COOH are 256.2, 282.2, and 284.2 u, respectively, the B23, B13, and B12 ions can be found at m/z 627.5, 601.5, and 599.5, respectively, by the differences between M123 and RnCOOH (n ) 1, 2, 3; Figure 4A). [M + Na]+ Ions of a TAG That Contains Identical Acyl Groups. A very simple TAG is one in which the three acyl groups are identical. The CAD spectra of [M + Na]+ ions of these simple TAGs are straightforward to interpret. Ions of any given type have the same m/z value, even though the ion structures are different. For example, fragmentation of [M + Na]+ ions of tristearoylglycerol (18:0/18:0/18:0) (Figure 3) produces A ions at m/z 701.6, B ions at m/z 629.5, C ions at m/z 607.5, D ions at m/z 477.3, E ions at m/z 419.3, F ions at m/z 405.3, G ions at m/z 347.2, and J ions at m/z 333.2. Identification of the A-D ions, particularly Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

4423

Figure 5. Comparison of CAD spectra of FAB-produced [M + Na]+ ions of TAG isomers: (A) 1,3-dipalmitoyl-2-oleoylglycerol (16:0/18:1/16:0, MW ) 832.7) and (B) 1,2-dipalmitoyl-3-oleoylglycerol (16:0/16:0/18:1, MW ) 832.7). A peak corresponding to an allylic cleavage is labeled with “a”. Table 2. Formulas for Calculating m/z Values for Various Fragments from the [M + Na]+ Ion of Triacylglycerols (M123 ) TAG + Na]+ A12 ) M123 - R3COOH + 72 A13 ) M123 - R2COOH + 72 A23 ) M123 - R1COOH + 72 B12 ) M123 - R3COOH B13 ) M123 - R2COOH B23 ) M123 - R1COOH C12 ) M123 - R3COONa C13 ) M123 - R2COONa C23 ) M123 - R1COONa D1 ) M123 - R2COOH - R3COOH + 132 D2 ) M123 - R1COOH - R3COOH + 132 D3 ) M123 - R1COOH - R2COOH + 132 E1 ) M123 - R2COOH - R3COOH + 74 E3 ) M123 - R1COOH - R2COOH + 74 F1 ) M123 - R2COOH - R3COOH + 60 F3 ) M123 - R1COOH - R2COOH + 60 G1 ) M123 - R2COOH - R3COOH + 2 G3 ) M123 - R1COOH - R2COOH + 2 J2 ) M123 - R1COOH - R3COOH - 12

the B ions, allows one to calculate the carbon number and the degree of unsaturation of each acyl group. For example, the mass difference between the precursor and the B ions is 284 u (Figure 3), indicating that each acyl group has 18 carbons and no double bonds. This is verified by the peak pattern corresponding to ions formed via CRF where no interruption is observed. Therefore, the complete structure of the TAG is established to be 18:0/18: 0/18:0. 4424 Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

[M + Na]+ Ions of TAGs Containing Different Acyl Groups. Figure 4 contains the CAD spectra of FAB- and ESI-produced [M + Na]+ ions of 1-palmitoyl-2-oleoyl-3-stearoylglycerol (16:0/18:1/ 18:0). The nature of the acyl groups is easily established by analyzing ions formed from the FAB-produced precursors. There are three A ions (A12, A13, and A23), three B ions (B12, B13, B23), etc. produced with high abundances. The differences between the precursor ion (of m/z 883.7) and the three B ions (of m/z 599.5, 601.5, and 627.5, respectively) are 284.2, 282.2, and 256.2, respectively, indicating that there are two saturated (16:0 and 18: 0) and one monounsaturated (18:1) acyl groups in the TAG. Identification of the ion formed by allylic cleavage (labeled with “a” in Figure 4A) enables one to locate the position of the double bond (18:1∆9). Unfortunately, the A, B, and C ions generated from the ESI-produced precursor ions are relatively less abundant (Figure 4B), raising the difficulty of assigning peaks and lowering confidence when the TAG is unknown. It is expected that each of the A-J ions would split into three ions when all the acyl groups are different. In fact, we did detect three A, B, C, and D ions, but only two E ions, two F ions, two G ions, and one J ion, indicating that the formation of some ions is energetically or structurally favored whereas that of others is not. For example, the CAD of [M + Na]+ ions of 1-palmitoyl-2-oleoyl3-stearoylglycerol (Figure 4) produces E ions at m/z 391 and 419 (corresponding to E1 and E3), F ions at m/z 377 and 405 (corresponding to F1 and F3), G ions at m/z 319 and 347 (corresponding to G1 and G3), and the J ion at m/z 331 (corresponding to just J2). The discovery that the E2, F2, G2, J1, and J3 ions are not produced has been validated by examining the MS/

Figure 6. Comparison of CAD spectra of ESI-produced [M + Na]+ ions of TAG isomers: (A) 1,2-dioleoyl-3-stearoylglycerol (18:1/18:1/18:0, MW ) 886.6) and (B) 1,3-dioleoyl-2-stearoylglycerol (18:1/18:0/18:1, MW ) 886.8). A peak corresponding to a vinyl cleavage is labeled with “v”.

MS of more than 10 TAG samples introduced under both ESI and FAB conditions. On the basis of these results, we conclude that the J ion contains the acyl group at sn-2. Therefore, except for the chirality of the sn-2 carbon (when the sn-1 and sn-3 substituents are different), the complete structure of the intact molecule is readily obtained. The A-C ions give carbon numbers and degree of unsaturation (16:0, 18:1, and 18:0), the chargeremote allylic cleavage reveals the position of the double bond (at C9), and the E-J ions show the location of acyl groups on the glycerol backbone (the oleoyl is at the sn-2 position, whereas the palmitoyl and stearoyl are at the sn-1 and sn-3 positions). Distinction of TAG Isomers. The positional information provided by E, F, G, and J ions formed from [M + Na]+ ions can be used to distinguish TAG isomers (except for optical isomers). In the CAD spectra of the [M + Na]+ ions that were formed by FAB from 1,3-dipalmitoyl-2-oleoylglycerol (16:0/18:1/16:0) and 1,2dipalmitoyl-3-oleoylglycerol (16:0/16:0/18:1) (Figure 5), we see that A-D ions are produced with similar abundances and the E-J ions appear as different patterns, allowing one to conclude easily that the TAGs are isomers. For example, the former isomer produces single E, F, or G ions (of m/z 391, 377, and 319, respectively) because the two outer acyl groups are the same (Figure 5A), whereas the latter isomer gives two each E, F, and G ions because the two outer acyl groups are different (E ions are of m/z 391 and 417; F ions of m/z 377 and 403, and G ions of m/z 319 and 345) (Figure 5B). In addition, the former gives a J2 ion at m/z 331 and the latter at m/z 305. From these values, it is straightforward to conclude that the former isomer has an 18:1 acyl group and the latter a 16:0 acyl group at the sn-2 positions.

Figure 6 compares the CAD spectra of the ESI-produced [M + Na]+ ions of another set of isomers, 1,2-dioleoyl-3-stearoylglycerol (18:1/18:1/18:0) and 1,3-dioleoyl-2-stearoylglycerol (18: 1/18:0/18:1). As expected, the former gives two E ions (E1 at m/z 417 and E3 at m/z 419), two F ions (F1 at m/z 403 and F3 at m/z 405), and two G ions (G1 at m/z 345 and G3 at m/z 347), whereas the latter gives only one E ion (E1 and E3, at m/z 417), one F ion (F1 and F3, at m/z 403), and one G ion (G1 and G3 at m/z 345). Additionally, the former produces a J2 ion at m/z 331 and the latter at m/z 333. Therefore, one can not only distinguish the TAG isomers by MS/MS of the [M + Na]+ ions but also determine the carbon numbers, degree of unsaturation, and location of the three acyl groups by an analysis of the A-J ions. Possible Mechanisms for the Formation of A-J Ions. To study the precursor/product relationships of A-J ions, we investigated fragmentation of A, B, and C ions by utilizing FAB-MS/MS/MS. We found that the A ions fragment to produce the B and C ions (spectra not shown), but we also know from CA of the [M + Na]+ that the B and C ions come, at least in part, from the molecular ion. Further, no D, E, F, G, and J ions arise from the fragmentation of A, B, and C ions. Therefore, all A-J ions are likely formed directly from the [M + Na]+. The A ions, which may be considered as a member of the CRF family, contain a terminal double bond conjugated with a carbonyl (Scheme 2A). The high abundance of A ions is due to the stabilizing effect of the interaction of the R,β-conjugated system with the sodium ion. The A ions may serve as the intermediates for the D ions. For example, a McLafferty-type rearrangement on the sn-3 acyl group of the A13 ion would produce the D1 ion Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

4425

(Scheme 2A). The B and C ions are formed by the cleavage of an acyl group, and the difference between B and C is that the former ions have a composition of [M + Na - RCOOH]+, whereas the latter ions have a composition of [M + Na - RCOONa]+. The carbonyl on an acyl group plays a crucial role in the formation of B ions, probably by permitting a McLafferty-type rearrangement to occur (Scheme 2B). Formation of the C ions is similar to that of [M + NH4 - RCOONH4]+ (Scheme 1), except that NH4+ is replaced by Na+ (Scheme 2C, pathway a). The E, F, and G ions may form from the [M + Na]+ via intermediates that do not contain the sn-2 acyl group. For example, the loss of the R2COO radical from the precursor produces an intermediate with a five-membered ring, and the intermediate gives rise to E, F, and G ions by losing three different radicals (Scheme 2D). The J ions are formed differently from others because they involve a cleavage of the glycerol backbone. An interaction between the sn-1 and sn-3 acyl groups may facilitate formation of the J ions because only J2 ions are formed. The complete loss of the sn-1 acyl group may trigger the bond breakage between carbons sn-3 and sn-2, or vice versa, and form a double bond that is always substituted by the sn-2 acyl group (Scheme 2C, pathway b).

of double bonds if present. Using the CAD of TAG [M + Na]+ ions, we are able to elucidate the full structure of the TAG (except for the chirality at the sn-2 site). Although ESI-MS/MS with kiloelectronvolt collisional activation (laboratory frame) provides relatively less structural information, it has a great potential for analyzing samples of TAG mixtures because it is compatible with on-line separation methods (for example, HPLC). The successful combination of HPLC and MS/MS will make it possible to determine completely the components in mixtures of TAGs, provided the ion activation in MS/MS is high energy. Although this activation is usually done by four-sector instruments, it can also be accomplished by double-focusing instruments (product ions by linked scans as demonstrated in this article) or, possibly in the future, by time-of-flight (TOF) or tandem TOF instruments. We found it remarkable that the mechanism of CRFs of TAG [M + met.]+ ions varies so strongly with internal energy. Paradoxically, precursor ions with lower internal energy (by ESI) undergo high-energy reactions (homolytic cleavage),49 whereas those with higher internal energy (by FAB) react through the low-energy 1,4-H2 elimination process.40 We reported previously on this phenomenon,43 but a complete understanding must await further studies.

CONCLUSIONS In addition to undergoing charge-remote fragmentations, ESIand FAB-produced [M + cat.]+ ions of triacylglycerols fragment to produce characteristic ions. The CAD of ESI-produced [M + NH4]+ or [M + Na]+ ions and of FAB-produced [M + Na]+ ions of TAGs is the basis of an analytical tool to obtain detailed structural information of triacylglycerols, including carbon numbers, degree of unsaturation of each acyl group, and the position

ACKNOWLEDGMENT This research was supported by NIH Center for Research Resources (Grant No. 2P41RR00954).

4426 Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

Received for review May 11, 1998. Accepted July 23, 1998. AC9805192