High-Performance Liquid Chromatography-Mass Spectrometry of Glycosphingolipids: II. Application to Neutral Glycolipids and Monosialogangliosides 1

Transcription

1 J. Biochem. 108, (1990) High-Performance Liquid Chromatography-Mass Spectrometry of Glycosphingolipids: II. Application to Neutral Glycolipids and Monosialogangliosides 1 Minoru Suzuki, Tamio Yamakawa, and Akemi Suzuki Department of Membrane Biochemistry, The Tokyo Metropolitan Institute of Medical Science, Bunkyo-ku, Tokyo 113 Received for publication, March 30, 1990 We previously reported a method of high-performance liquid chromatography-fast atom bombardment mass spectrometry (HPLC/FAB/MS) for the structural characterization of molecular species of G1cCer and IV319Gal-Gb,Cer [M. Suzuki et al. (1989) J. Biochem. 105, ]. In this paper, we report a modification of this HPLC/FAB/MS method, which was used for the separation and characterization of neutral glycosphingolipids (G1cCer, LacCer, Gb3 Cer, Gb4 Cer, and IV3aGa1NAc-Gb4 Cer) and monosialogangliosides [GM3(NeuAc or NeuGc), GM2 (NeuAc or NeuGc), and GM1 (NeuAc or NeuGc)]. Mixtures of the purified neutral glycolipids and monosialogangliosides were subjected to HPLC on a silica gel column, with programmed elution with isopropanol-n-hexane-water, with or without ammonium hydroxide. In order to obtain mass spectra and mass chromatograms of individual components, effluent from the HPLC column was mixed with a methanol solution of triethanolamine, which was used as the matrix for the FAB ionization, and one-thirtieth of the effluent mixture was introduced into a mass spectrometer through a frit interface. A mixture of the five neutral glycolipids, 5 jug of each, gave five peaks on a mass chromatogram obtained by monitoring of the corresponding major pseudo-molecular ions. A mixture of the six monosialogangliosides, 5ƒÊg of each, gave six peaks on a mass chromatogram obtained by monitoring of the major pseudo-molecular ions, indicating that GM3, GM2, and GM1 were clearly separated, and that separation due to differences in sialic acid species was also achieved. In the mass spectra of the neutral glycolipids and monosialogangliosides, pseudo-molecular ions and fragment ions due to the elimination of sugar moieties were clearly detected. We applied this method to the analysis of a mixture of neutral glycolipids isolated from human erythrocytes and that of a mixture of monosialogangliosides isolated from the liver of a mouse, GlcCer, LacCer, Gb3Cer, LccCer, Gb4 Cer, and nlc4 Cer and GM3(NeuAc), GM3(NeuGc), GM2(NeuGc), and GM1(NeuGc) being detected, respectively. This method is of great advantage for the structural characterization of neutral glycolipids and monosialogangliosides, especially when they are in mixtures and/or in small amounts. The method will also be quite promising for other neutral glycolipids and gangliosides with the establishment of suitable separation conditions on HPLC. Glycosphingolipids are components of mammalian cell membranes. They are located in the outer leaflet of the lipid bilayer and their carbohydrate moieties extend out into the extracellular space. Glycolipids, thus, can play roles as antigens, receptors for toxins, viruses, and bacteria, and modulators of membrane receptors, as annular lipids (3, 4). 1 This work was supported in part by Grants-in-Aid from the Ministry of Education, Science and Culture of Japan, and a grant from the Mitsubishi Foundation. Abbreviations: The nomenclature used for neutral glycosphingolipids follows recent recommendations (1). G1cCer, glucosylceramide; LacCer, lactosylceramide; Gb4 Cer, globotriaosylceramide; Lc3 Cer, lactotriaosylceramide; Gb4 Cer, globotetraosylceramide; nlc4 Cer, neolactotetraosylceramide; IV3 ƒà Gal-Gb4 Cer, galactosylglobotetraosylceramide; IV3 ƒ Ga1NAc-Gb4 Cer, Forssman glycolipid. The nomenclature used for gangliosides is based on the system of Svennerholm (2). The sialic acid species of gangliosides are indicated in parentheses; NeuAc, N-acetylneuraminic acid; NeuGc, N-glycolylneuraminic acid. HPLC FAB MS, high-performance liquid chromatography-fast atom bombardment mass spectrometry; Hex, hexose. In order to elucidate the physiological roles of glycolipids, much effort has been put into the isolation and characteriza tion of glycolipids, which are usually minor components in culture cells or isolated homogeneous cells, except for neural cells. So, we believe that the development of analytical methods with which it is possible to detect quite small amounts of glycolipids would be quite important for and contribute much to further investigation of the physiological roles of glycolipids. High-performance liquid chromatography (HPLC) is commonly used for the separation of neutral glycolipids (5-13) and gangliosides (14-25), and is the best analytical method at present, together with thin-layer chromatog raphy. Fast atom bombardment mass spectrometry (FAB/ MS) of glycolipids has become an indispensable means of obtaining information as to molecular weights, carbo hydrate species, carbohydrate sequences, and ceramide structures (26-35). In particular, negative ion FAB mass spectrometry (36-41) and secondary ion mass spectrom- 92 J. Biochem.

2 HPLC/MS of Neutral Glycolipids and Monosialogangliosides 93 etry (42-44) are quite useful for structural characterization, because they give simple mass spectra showing pseudo-molecular ions and fragment ions due to the cleavage of glycosidic bonds. The combination of HPLC and FAB/MS (HPLC/FAB/ MS) has been successfully applied to the structural anal yses of bile acids (45, 46), oligosaccharides (47, 48), and peptides (49). Recently, we reported that a HPLC/FAB/ MS system including a reversed-phase column had been successfully applied to the structural characterization of the molecular species of underivatized neutral glycolipids, GlcCer and IV3 ƒà Gal-Gb4 Cer (50). This system, however, could not be applied to the separation of glycolipids due to differences in their carbohydrate structures. In this paper, we report that a modified system, which includes a silica gel column, has been successfully applied to the separation and structural characterization of mixtures of neutral glycolipids and monosialogangliosides. We also report two examples of the application of this system, one for the analysis of neutral glycolipids isolated from human erythrocyte stroma and the other for that of monosialogangliosides isolated from the liver of a mouse. MATERIALS AND METHODS Materials-Triethanolamine was purchased from Tokyo Kasei Kogyo (Tokyo), and HPLC-grade isopropanol, n- hexane, water, and methanol from Wako Pure Chemical Industries (Tokyo). The solvents used for the extraction of neutral glycolipids and monosialogangliosides were dis tilled before use. GlcCer purified from the spleen of a patient with Gaucher's disease was kindly supplied by Dr. M. Oshima, Clinical Research Institute, National Medical Center, Tokyo. LacCer, Gb3Cer and Gb4Cer were purified from human erythrocyte stroma (51), and IV3ƒ Ga1NAc-Gb4Cer from sheep erythrocyte stroma (52) in our laboratory. A mixture of these neutral glycolipids, which was subjected to HPLC/FAB/MS analysis, was prepared, containing 5ƒÊg/ ƒê l of each component. Human erythrocytes give two LacCer bands on thin-layer chromatography (TLC) and in this study, the lower LacCer band was purified and used as the standard for further analysis. Thus, the purified LacCer used in this study contained ceramide consisting of C16:0 fatty acid and d18:1 sphingosine. GM3(NeuAc) was prepared from dog erythrocytes (53), GM2(NeuAc) from the brain of a patient with Tay-Sachs disease (54) and GM1(NeuAc) from bovine brain. GM3 (NeuGc), GM2(NeuGc), and GM1(NeuGc) were prepared from the livers of ICR mice, as described previously (55). A mixture of these six gangliosides, which was subjected to HPLC/FAB/MS analysis, was prepared and it contained 5 ƒê g/ƒêl of each ganglioside. HPLC and HPLC/FAB/MS-HPLC was performed with a model 120A (Applied Biosystems, Calif., U.S.A.) on a silica gel column (Aquasil-SS-B, Senshu Scientific, Tokyo; 1mm i.d. ~50mm), with programmed gradient elution with either isopropanol-n-hexane-water, from 20 : 80:0.5 (v/v) to 55:40:5 (v/v) in 30min, for the analysis of neutral glycolipids or isopropanol-n-hexane-5 N ammonium hydroxide-water, from 110:80:10:0 (v/v) to 160:20:15:5 (v/v) in 10min, for that of monosialogangliosides, at the flow rate of 100lƒÊ per min. The tempera ture of the column oven was maintained at 40 Ž. As a matrix, 3% triethanolamine in methanol was introduced into the effluent from the HPLC column at the flow rate of 50ƒÊl per min with a HPLC pump (Flow System 3100, Senshu Scientific). The effluent mixture was split with a splitter (JEOL, Tokyo) and one-thirtieth of the effluent mixture was introduced into a mass spectrometer through a frit interface (JEOL). Thus, one-thirtieth of the injected glycolipids was introduced into the mass spectrometer and the rest was discarded via a drain tube. Mass Spectrometer-Negative-ion mass spectra were obtained by a JMS-HX110 double focusing mass spectrometer equipped with a FAB ion source and a JMA-DA5000 data system (JEOL). The accelerating voltage was -6.0 kv for the analysis of neutral glycolipids and -7.0kV for that of monosialogangliosides, and the primary beam for bombardment was 6.0keV Xe KThe ion source tempera ture was maintained at 30 Ž. Liquid nitrogen was introduced into the trap in an oil diffusion pump in order to keep the vacuum as high as possible. Mass chromatography was performed to repeat the detection of the ions from m/z 300 to 2,000 at intervals of 6 s. Preparation of Neutral Glycosphingolipids from Human Erythrocytes and Monosialogangliosides from Mouse Liver for Application of the HPLC/FAB/MS Method-Dried stroma of human erythrocytes (210g) was subjected to lipid extraction with chloroform-methanol (2:1 and 1:1, v/v) (53). Lipids in lower phase obtained on partitioning according to Folch et al. (56) were applied to a DEAE- Sephadex column (acetate form, Pharmacia Fine Chemicals, Uppsala; 1.8cm i.d. ~80cm) and the neutral fraction was recovered in the flow-through fraction. The neutral lipids were acetylated and then subjected to latrobeads (Iatron Laboratories, Tokyo) column chromatography by a slight modification of the method of Saito and Hakomori (52). The fraction containing neutral glycolipids was deacetylated and recovered in the lower phase on Foich's partitioning. An aliquot of this fraction, equivalent to 21 mg of dry stroma, was subjected to HPLC/FAB/MS. Monosialogangliosides were prepared from lipids ex tracted from the liver (1.01g, wet weight) of a B10.AQR mouse of 8weeks of age. The lipids were extracted twice with 15ml of chloroform-methanol-water (4:8:3, v/v) and then the upper phase was obtained by the method of Svennerholm and Fredman (57). The upper phase lipids were applied to a DEAE-Toyopearl column (acetate form, Tosoh, Tokyo; 1.0cm i.d. ~3cm) and monosialogangliosides were recovered with chloroform-methanol-0.02m potassium acetate (30:60:8, v/v). After alkaline treatment with 0.1M potassium hydroxide in methanol, the monosialogangliosides were separated from the salt by chromatography on a Sephadex LH-20 column (Pharmacia Fine Chemicals; 1.0cm i.d. ~48cm). One-twentieth of the monosialogangliosides was subjected to HPLC/FAB/MS. RESULTS Mass Chromatograms of Neutral Glycosphingolipids -Figure 1 shows mass chromatograms obtained on the injection of a mixture containing 5ƒÊg each of the neutral glycolipids: GlcCer, LacCer, Gb3Cer, Gb4Cer, and IV3ƒ - Ga1NAc-Gb4Cer. In these mass chromatograms, the peaks selected as to the pseudo-molecular ions ([M-H]-) of the Vol. 108, No. 1, 1990

3 94 M. Suzuki et al. major molecular species were clearly observed at m/z 810 for G1cCer containing C24:0 fatty acid and d18:1 sphingosine (58), 860 for LacCer containing C16:0 and d18:1 (59), 1,134 for Gb3Cer containing C24:0 and d18:1 (60), 1,337 for Gb,Cer containing C24:0 and d18:1 (60), and 1,538 for IV3aGa1NAc-Gb4Cer containing C24:1 and d18: 1 (61, 62). The five neutral glycolipids were well separated within 30min. Mass Spectra of Neutral GlycosphingolipidsFigure 2 shows the mass spectra recorded for the tops of peaks A to E in Fig. 1. In the mass spectrum of GlcCer, as shown in Fig. 2A, seven pseudo-molecular ions were clearly detected at m/z 810, 808, 796, 782, 754, 726, and 698, corresponding to the seven molecular species due to fatty acids we reported in a previous paper (50). The major two, at m/z 782 and 810, produced the ions at m/z 620 for ceramide consisting of C22:0 and d18:1 and at 648 for that consisting of C24:0 and d18:1, respectively. In the mass spectrum of LacCer, as shown in Fig. 2B, a pseudo-molecular ion exhibiting strong intensity was detected at m/z 860. Fragment ions due to the successive elimination of Gal and Gal-Glc from the molecule were clearly observed at m/z 698 and 536, respectively. The ceramide responsible for the ion at m/z 536 is composed of C16:0 and d18:1. In Fig. 2C for Gb3Cer, the pseudomolecular ion and the ions due to the successive elimination of Gal, Gal-Gal, and Gal-Gal-Gle were detected at mf z 1,134, 972, 810, and 648, respectively. The ceramide responsible for the ion at m/z 648 is composed of C24: 0 and d18: 1. In Fig. 2D for Gb, Cer, the pseudo-molecular ion and the ions due to the elimination of Ga1NAc, GaINAc-Gal, Ga1NAc-Gal-Gal, and GalNAc-Gal-Gal-Glc were detected at m/z 1,337, 1,134, 972, 810, and 648, respectively. The ceramide responsible for the ion at m/z 648 is composed of C24:0 and d18:1. In Fig. 2E for IV3ƒ GalNAc-Gb4Cer, the pseudo-molecular ion and the fragment ions due to the elimination of GaINAc, Ga1NAc-Ga1NAc, Ga1NAc-Gal- NAc-Gal, Ga1NAc-GaINAc-Gal-Gal, and GalNAc- GalNAc-Gal-Gal-Glc from the molecule were found at m/z 1,538, 1,335, 1,132, 970, 808, and 646, respectively. The ceramide responsible for the ion at m/z 646 is composed of C24:1 and d18:1. In Fig. 2, C, D, and E, the pseudomolecular ions for the molecular species containing C22- fatty acids were also observed. These mass spectra are comparable to those obtained on FAB/MS with a direct inlet system in terms of the successful detection of pseudo-molecular ions and fragment ions. Application to the Analysis of Neutral Glycosphingolipids Isolated from Human Erythrocytes-In the mass chromatograms shown in Fig. 3, peaks B, C, D, and G, at m/ z 972, 860, 1,134, and 1,337, are possibly due to the pseudo-molecular ions of LacCer containing C24:0 and d18:1, and C16:0 and d18:1 (59), Gb3Cer containing C24: 0 and d18:1, and Gb,Cer containing C24:0 and d18:1, respectively. The small peaks, A and E, also detected are considered to be due to the pseudo-molecular ions of GicCer containing C24:0 and d18:1 and LccCer containing C24:0 and d18:1 (60), respectively. Peak F, with a similar retention time to that of Gb4 Cer and responsible for the ion at m/z 1,175, was identified as the fragment ion of nlc,cer (60) due to the elimination of galactose. The peak for Fig. 1. Mass chromatograms of the pseudo-molecular ions of the major molecular species of GleCer (A), LacCer (B), Gb3Cer (C), Gb4Cer (D), and IV3ƒ Ga1NAc-Gb4Cer (E). A mixture contain ing 5ƒÊg each of the neutral glycosphingolipids was injected into the HPLC and separated on a silica gel column with gradient elution, as described under "MATERIALS AND METHODS." RIC, reconstructed ion chromatogram. Fig. 2. Mass spectra of neutral glycolipids, GlcCer (A), LacCer (B), Gb3Cer (C), Gb4Cer (D), and IV3ƒ GaINAc-Gb4Cer (E), obtained for the tops of peaks A to E in Fig. 1. J. Biochem.

4 HPLC/MS of Neutral Glycolipids and Monosialogangliosid es 95 Fig. 3. Mass chromatograms of the pseudo-molecular ions of the molecular species of neutral glycolipids. Glycolipids amount ing to 21mg of dry stroma of human erythrocytes were injected. Characterization of each of peaks from A to G is indicated in the text. Fig. 5. Mass spectra of monosialogangliosides containing N- acetylneuraminic acid, GM3 (A), GM2 (C), and GM1 (E), ob tained for the tops of peaks A, C, and E in Fig. 4. Fig. 4. Mass chromatograms of the pseudo-molecular ions of the major molecular species of monosialogangliosides, GM3 (NeuAc) (A), GM3(NeuGc) (B), GM2(NeuAc) (C), GM2(NeuGc) (D), GM1(NeuAc) (E), and GM1(NeuGc) (F), obtained on the injection of a mixture containing 5ƒÊg of each. nlc4cer is not separated from that of Gb4Cer with this HPLC system, one major reason for which being that there is quite a large amount of Gb4 Cer compared with that of nlc4 Cer. Mass spectra of neutral glycolipids other than Gb4Cer were not obtained in this analysis, for which an aliquot of crude glycolipids equivalent to 21mg dry stroma was injected. The reason for this is also the presence of the large amount of Gb4 Cer. However, the components of which the occurrence has been reported, were detected by monitoring of their pseudo-molecular ions, indicating that this method is quite promising for application to neutral glycolipids isolated from a small number of cells or a homogeneous Fig. 6. Mass spectra of monosialogangliosides containing N- glycolylneuraminic acid, GM3 (B), GM2 (D), and GM1 (F), obtained for the tops of peaks B, D, and F in Fig. 4. population of isolated cells. Mass Chromatograms of Monosialogangliosides-Figure 4 shows mass chromatograms obtained on the injection of a mixture of 5pg each of the monosialogangliosides, GM3 (NeuAc), GM3(NeuGc), GM2(NeuAc), GM2(NeuGc), GM1 (NeuAc), and GM 1(NeuGc). In these mass chromatograms, the peaks selected as to the pseudo-molecular ions were clearly detected at m/z 1,261 for GM3(NeuAc) containing C24:1 and d18:1, 1,279 for GM3(NeuGc) containing C24:0 and d18:1, 1,382 for GM2(NeuAc) containing C18:0 and d18:1, 1,482 for GM2(NeuGc) containing 024:0 and d18:1, Vol. 108, No. 1, 1990

5 96 M. Suzuki et al. Fig. 7. Mass chromatograms of the major pseudo-molecular ions of monosialogangliosides from the liver of a mouse. Monosialogangliosides equivalent to 51mg of mouse liver were injected. Identification of each of peaks from A to H is described in the text. 1,544 for GM1(NeuAc) containing C18:0 and d18:1, and 1,644 for GM1(NeuGc) containing C24:0 and d18:1. Although in the reconstructed ion chromatogram (RIC) the peaks of these monosialogangliosides were not completely separated, the peaks of individual gangliosides were clearly separated in the mass chromatograms obtained on the detection of pseudo-molecular ions. It should be noted that the separation of GM3, GM2, and GM1 was complete and, in addition, that the gangliosides were also separated due to differences in their sialic acid species. Mass Spectra of GM3, GM2, and GM1 Containing N- Acetylneuraminic Acid-Figure 5 shows the mass spectra of GM3(NeuAc), GM2(NeuAc), and GM1(NeuAc). As shown in Fig. 5A for GM3(NeuAc), pseudo-molecular ions were clearly observed at m/z 1,261, 1,263, and 1,277, which were identified as GM3(NeuAc) containing C24:1 and d18:1, C24:0 and d18:1, and C25:0 and d18:1, respec tively (63). In Fig. 5C for GM2(NeuAc), pseudo-molecular ions were clearly observed at m/z 1,382 and 1,410, which were identified as GM2(NeuAc) containing C18:0 and d18: 1 and C18:0 and d20:1, respectively (21, 64). In Fig. 5E for GM1(NeuAc), pseudo- molecular ions were clearly observed at m/z 1,544 and 1,572, which were identified as GM1(NeuAc) containing C18:0 and d18:1 and C18:0 and d20:1, respectively (21, 65). Ions due to the elimination of N-acetylneuraminic acid from the molecule were detected at m/z 970, 972, and 986 in Fig. 5A for GM3(NeuAc), 1,091 and 1,119 in Fig. 5C for GM2(NeuAc), and 1,253 and 1,281 in Fig. 5E for GM1(NeuAc). Ions due to the successive elimination of sugar moieties were also detected in these mass spectra. The oligosaccharide-derived ions produced on the cleavage between II-Gal and I-Glc were clearly observed at m/z 468 and 470 for GM3(NeuAc), 671 and 673 for GM2(NeuAc), and 833 and 835 for GM1(NeuAc). These mass spectra are comparable with those obtained on FAB/MS with a direct inlet system (data not shown). Mass Spectra of GM3, GM2, and GM1 Containing N- Glycolylneuraminic Acid-The mass spectra of GM3(Neu- Gc), GM2(NeuGc), and GM1(NeuGc) (38) are shown in Fig. 6. In Fig. 6B, the pseudo-molecular ions of GM3(Neu- Fig. 8. Mass spectra of monosialogangliosides, GM3(NeuAc) (A), GM3(NeuGc) (B), GM2(NeuGc) (C), and GM1(NeuGc) (D), obtained for the tops of peaks A, C, E, and G in Fig. 7. Gc) were clearly observed at m/z 1,251, 1,277, and 1,279, which were identified as GM3(NeuGc) containing C22:0 and d18:1, C24:1 and d18:1, and C24:0 and d18:1, respectively. In Fig. 6D, the pseudo-molecular ions of GM2(Neu- Gc) were clearly detected at m/z 1,454, 1,480, and 1,482, which were identified as GM2(NeuGc) containing C22:0 and d18:1, C24:1 and d18:1, and C24:0 and d18:1, respectively. In Fig. 6F, the pseudo-molecular ions of GM1(Neu- Gc) were detected at m/z 1,616, 1,642, and 1,644, being identified as GM1(NeuGc) containing C22:0 and d18:1, C24:1 and d18:1, and C24:0 and d18:1, respectively. Ions due to the elimination of N-glycolylneuraminic acid from the molecule were detected at m/z 944, 970, and 972 in Fig. 6B for GM3(NeuGc), 1,147, 1,173, and 1,175 in Fig. 6D for GM2(NeuGc), and 1,309, 1,335, and 1,337 in Fig. 6F for GM1(NeuGc). Ions due to the successive elimination of sugar moieties were also detected in these mass spectra. The oligosaccharide-derived ions produced on cleavage between II-Gal and I-Glc were clearly observed at m/z 484 and 486 for GM3(NeuGc), 687 and 689 for GM2(NeuGc), and 849 and 851 for GM1(NeuGc). In the case of these oligosaccharide-derived ions, the 2 mass unit difference is due to the loss of 2H from the ions with higher numbers of mass units (34). These oligosaccharide ions were 16 atomic mass units higher than those of GM3, GM2, and GM1 containing N-acetylneuraminic acid. This difference corresponds to the structural difference between N-glycolyl and N-acetyl groups in the sialic acids. Application to the Analysis of Monosialogangliosides from the Liver of a B10.AQR Mouse-As shown in Fig. 7, eight peaks were obtained on monitoring of pseudo-molecular ions, at m/z 1,261, 1,263, 1,277, 1,279, 1,480, 1,482, J. Biochem.

6 HPLC/MS of Neutral Glycolipids and Monosialoganglios ides 97 1,642, and 1,644. The retention times of peaks A and B, C and D, E and F, and G and H were the same as those of GM3(NeuAc), GM3(NeuGc), GM2(NeuGc), and GM1 (NeuGc), respectively. Figure 8 shows the mass spectra of peaks A, C, E, and G in Fig. 7. In Fig. 8A, pseudo-molecular ions were observed at m/z 1,235, 1,261, and 1,263. Ions due to the elimination of NeuAc, NeuAc-Hex, and NeuAc- Hex-Hex from the molecule were observed at m/z 944, 970, and 972, 782, 808, and 810, and 620, 646, and 648, respectively. Furthermore, the ions of oligosaccharides due to the cleavage between II-Gal and I-Glc were observed at m/z 468 and 470. From these results, this ganglioside was identified as GM3(NeuAc). Fragment ions and major pseudo-molecular ions detected in the mass spectra of B, C, and D in Fig. 8 were identical with those of B, D, and F in Fig. 6, respectively. The small differences in the molecular species compositions between the corresponding gangliosides, as recognized in Figs. 6 and 8, are considered to be due to the differences in the purification procedures. DISCUSSION Recently, Kushi et al. reported the results of analysis of the molecular species of neutral glycosphingolipids, GlcCer, LacCer, and Gg3Cer, using HPLC-atmospheric pressure ionization mass spectrometry and a reversed-phase column for HPLC (66). We also reported the structural characterization and the separation of molecular species of GlcCer and IV3f3Gal-Gb,Cer, using HPLC/FAB/MS, glycerol as a matrix and a reversed-phasee column (50). With both systems, glycolipids were separated into molecular species on the basis of the differences in their ceramide moieties, but not separated on the basis of the differences in their carbohydrate structures. Thus, it was necessary to develop a new HPLC/MS system with which glycolipids can be separated into components on the basis of their sugar moieties. With our HPLC/FAB/MS system, there was another drawback, that is, fragment ions due to the elimi nation of hexoses or oligosaccharides from glycolipid molecules with rather long chain carbohydrates can hardly be obtained, even if large amounts of the glycolipids (about 5ƒÊg) are injected. This was due to the bad selection of a matrix. Triethanolamine, instead of glycerol, is a suitable matrix for such glycolipids (27), but could not be used because column packings are not stable at the high ph of triethanolamine. Thus, we tried to modify the HPLC/FAB/ MS system so that glycolipids can be separated due to differences in their carbohydrate structures and so that any matrix can be used. In the case of this modified system, a matrix solution is introduced into the effluent from a HPLC column after the effluent has flowed through the UV detector. Thus, UV detection at 205nm is possible without disturbance due to the UV absorption of the matrix. When two mixtures, one containing 5 u g each of the neutral glycolipids, GlcCer, LacCer, Gb3Cer, Gb4Cer, and IV3ƒ - GalNAc-Gb4Cer, and the other containing 5 u g each of the monosialogangliosides, GM3(NeuAc or NeuGc), GM2 (NeuAc or NeuGc), and GM1(NeuAc or NeuGc), were analyzed with this improved system, each glycolipid was separated on mass chromatography with detection of the major pseudo-molecular ions. And in all the mass spectra, pseudo-molecular ions and fragment ions due to the loss of sugar moieties were clearly detected, as shown in Figs. 2, 5, and 6. These mass spectra are quite comparable to those obtained on FAB/MS with a direct inlet system. As shown in Fig. 3, when a mixture of neutral glycolipids, equivalent to those obtained from 21mg of dry stroma of human erythrocytes, was subjected to HPLC/FAB/MS, the mass spectra of the neutral glycolipids other than Gb4Cer could not be obtained. However, peaks of GlcCer, LacCer, Gb3Cer, Lc3Cer, Gb4Cer, and nlc4cer were de tected in the mass chromatograms when they were monitored as to the pseudo-molecular ions. In these mass chromatograms, two pseudo-molecular ions of LacCer could be detected at m/z 860 (C16:0-d18:1) and 972 (C24: 0-d18:1), but the ion at m/z 972 was not detected when the purified LacCer was analyzed, as shown in Fig. 1. This difference was due to the purification procedures used and to that a molecular species containing C24:0 and d18:1 was not included in the purified LacCer. Peak F is judged to be GlcNAc-Gal-Glc-Cer, which is derived from nlc4cer. The numbers of atomic mass units of the pseudo-molecular ions of major molecular species of nlc4 Cer are the same as those of Gb4Cer, and the retention time of peak F was the same as that of Gb4Cer. Therefore, under the HPLC conditions used, nlc4cer was not separated from Gb4Cer. This HPLC/FAB/MS system was also successfully applied to the separation and characterization of monosialogangliosides from the liver of a mouse, as shown in Figs. 7 and 8. With this system, after the effluent had been mixed with the matrix solution, only 3.3% of the mixture, in other words only 3.3% of the injected materials, was introduced into the mass spectrometer. Therefore, when 5ƒÊg of glycolipid was injected, it was calculated that 167ƒÊg was analyzed by the mass spectrometer, its pseudo-molecular ion and also its mass spectrum being given. These results indicate that it would be promising to establish a sensitive method for the detection and characterization of neutral glycolipids and gangliosides. In order to apply this method to the quantitative determination of glycolipids, it is necessary to prepare glycolipids which are labeled with stable isotopes and which can be used as internal standards. Our next target is to develop this system for the quantita tive determination of glycolipids. For the sensitive detec tion of glycolipids, thethin-layer chromatography-mass spectrometry developed by Kushi et al. (67) and the thin-layer chromatography-immunostaining (TLC-IS) de veloped by Magnani et al. (68) are notable, the latter greatly contributing to the progress in glycolipid biochemistry. We believe that HPLC/FAB/MS will become a method comparable to TLC-IS as to sensitivity and will be applied to various kinds of glycolipids in the future. This paper describes a HPLC/FAB/MS system including a silica gel column, which has been successfully applied to the separation and characterization of neutral glycosphingolipids with one to five sugars, and also GM3, GM2, and GMl containing N-acetylneuraminic acid or N-glycolylneuraminic acid. We wish to thank Dr. M. Oshima, Clinical Research Institute, National Medical Center, for providing the glucosylceramide from the spleen of a patient with Gaucher's disease. We also thank Dr. K. Watanabe, Shigei Medical Research Institute, for his valuable advice. REFERENCES 1. IUPAC-IUB Commission on Biochemical Nomenclature, The vol. 108, No. 1, 1990

7 98 M. Suzuki et al. Nomenclature of Lipids (1978) J. Lipid Res. 19, Svennerholm, L. (1963) J. Neurochem. 10, Yamakawa, T. & Nagai, Y. (1978) Trends Biochem. Sci. 3, Hakomori, S. (1981) Annu. Rev. Biochem. 50, Evans, J.E. & McCluer, R.H. (1972) Biochim. Biophys. Acta 270, Uliman, M.D. & McCluer, R.H. (1977) J. Lipid Res. 18, Ullman, M.D. & McCluer, R.H. (1978) J. Lipid Res. 19, Yamazaki, T., Suzuki, A., Handa, S., & Yamakawa, T. (1979) J. Biochem. 86, Suzuki, A., Kundu, S.K., & Marcus, D.M. (1980) J. Lipid Res. 21, Schwarting, G.A. & Gajewski, A. (1983) J. Biol. Chem. 258, Kaye, E.M. & Ullman, M.D. (1984) Anal. Biochem. 138, McCluer, R.H. & Gross, S.K. (1985) J. Lipid Res. 26, Ullman, M.D. & McCluer, R.H. (1987) Methods Enzymol. 138, Nakabayashi, H., Iwamori, M., & Nagai, Y. (1984) J. Biochem. 96, Ullman, M.D. & McCluer, R.H. (1985) J. Lipid Res. 26, Miyazaki, K., Okamura, N., Kishimoto, Y., & Lee, Y.C. (1986) Biochem. J. 235, Watanabe, K. & Arao, Y. (1981) J. Lipid Res. 22, Watanabe, K. & Tomono, Y. (1984) Anal. Biochem. 139, Kannagi, R., Watanabe, K., & Hakomori, S. (1987) Methods Enzymol. 138, Kundu, S.K. & Scott, D.D. (1982) J. Chromatogr. 232, Handa, S. & Kushi, Y. (1982) Adv. Exp. Med. Biol. 152, Gazzotti, G., Sonnino, S., & Ghidoni, R. (1985) J. Chromatogr. 348, Ando, S., Waki, H., & Kon, K. (1987) J. Chromatogr. 405, Ando, S., Waki, H., & Kon, K. (1987) J. Chromatogr. 408, Rassi, Z.E., Horvath, C., Yu, R.K., & Ariga, T. (1989) J Chromatogr. 488, Rinehart, K.L., Jr. (1982) Science 218, Arita, M., Iwamori, M., Higuchi, T., & Nagai, Y. (1983) J. Biochem. 93, Hemling, M.E., Yu, R.K., Sedgwick, R.D., & Rinehart, K.L., Jr. (1984) Biochemistry 23, Sweeley, C.C. & Nunez, H.A. (1985) Anna. Rev. Biochem. 54, Rosner, H., Rahmann, H., Reuter, G., Schauer, R., Peter- Katalinic, J., & Egge, H. (1985) Biol. Chem. Hoppe-Seyler 366, Hara, A. & Taketomi, T. (1986) J. Biochem. 100, Ohashi, Y., Iwamori, M., Ogawa, T., & Nagai, Y. (1987) Biochemistry 26, Muthing, J., Egge, H., Kniep, B., & Muhlradt, P.F. (1987) Eur. J. Biochem. 163, Egge, H. & Peter- Katalinic, J. (1987) Mass Spectrom. Rev. 6, Pohlentz, G., Klein, D., Schmitz, D., Schwarzmann, G., Peter- Katalinic, J., & Sandhoff, K. (1988) Biol. Chem. Hoppe-Seyler 369, Arita, M., Iwamori, M., Higuchi, T., & Nagai, Y. (1983) J. Biochem. 94, Domon, B. & Costello, C.E. (1988) Biochemistry 27, Suzuki, M., Nakamura, K., Hashimoto, Y., Suzuki, A., & Yamakawa, T. (1986) Carbohydr. Res. 151, Sekine, M., Suzuki, M., Inagaki, F., Suzuki, A., & Yamakawa, T. (1987) J. Biochem. 101, Nakamura, K., Suzuki, M., Inagaki, F., Yamakawa, T., & Suzuki, A. (1987) J. Biochem. 101, Sekine, M., Nakamura, K., Suzuki, M., Inagaki, F., Yamakawa, T., & Suzuki, A. (1988) J. Biochem. 103, Kushi, Y., Handa, S., &Ishizuka, I. (1985) J. Biochem. 97, Nagai, K., Roberts, D.D., Toita, T., Matsumoto, H., Kushi, Y., Handa, S., & Ishizuka, I. (1989) J. Biochem. 106, Nagai, K., Roberts, D.D., Toita, T., Matsumoto, H., Kushi, Y., Handa, S., & Ishizuka, I. (1989) J. Biol. Chem. 264, Ito, Y., Takeuchi, T., & Ishii, D. (1985) J. Chromatogr. 346, Ito, Y., Takeuchi, T., & Ishii, D. (1986) J. Chromatogr. 358, Santikarn, S., Her, G.R., & Reinhold, V.N. (1987) J. Carbohydr. Chem. 6, Boulenguer, P., Leroy, Y., Alonso, J.M., Montreuil, J., Ricart, G., Colbert, C., Duquet, D., Dewaele, C., & Fournet, B. (1988) Anal. Biochem. 168, Caprioli, R.M., Moore, W.T., Dague, B., & Martin, M. (1988) J. Chromatogr. 443, Suzuki, M., Sekine, M., Yamakawa, T., & Suzuki, A. (1989) J. Biochem. 105, Ando, S. & Yamakawa, T. (1973) J. Biochem. 73, Saito, T. & Hakomori, S. (1971) J. Lipid Res. 12, Handa, S. & Yamakawa, T. (1964) Jpn. J. Exp. Med. 34, Makita, A. & Yamakawa, T. (1963) Jpn. J. Exp. Med. 33, Nakamura, K., Hashimoto, Y., Suzuki, M., Suzuki, A., & Yamakawa, T. (1984) J. Biochem. 96, Folch, J., Lees, M., & Sloane-Stanley, G.H. (1957) J. Biol. Chem. 226, Svennerholm, L. & Fredman, P. (1980) Biochim. Biophys. Acta 617, Samuelsson, K. & Samuelsson, B. (1969) Biochem. Biophys. Res. Commun. 37, Suzuki, A., Handa, S., & Yamakawa, T. (1977) J. Biochem. 82, Ando, S., Kon, K., Isobe, M., Nagai, Y., & Yamakawa, T. (1976) J. Biochem. 79, Ando, S. & Yamakawa, T. (1970) Chem. Phys. Lipids 5, Siddiqui, B. & Hakomori, S. (1971) J. Biol. Chem. 246, Yasue, S., Handa, S., Miyagawa, S., Inoue, J., Hasegawa, A., & Yamakawa, T. (1978) J. Biochem. 83, Taketomi, T., Kawamura, N., & Okano, A. (1971) Jpn. J. Exp. Med. 41, Yohe, H.C., Roark, D.E., & Rosenberg, A. (1976) J. Biol. Chem. 251, Kushi, Y., Rokukawa, C., Numajiri, Y., Kato, Y., & Handa, S. (1989) Anal. Biochem. 182, Kushi, Y., Rokukawa, C., & Handa, S. (1988) Anal. Biochem. 175, Magnani, J., Ball, E.D., Fanger, M.W., Hakomori, S., & Ginsburg, V. (1984) Arch. Biochem. Biophys. 233, J. Biochem.