Characterization of a Major Neutral Glycolipid in PC12 Cells as II13Gala-globotriaosylceramide by the Method for Determining Glycosphingolipid Saccharide Sequence with Endoglycoceramidase*

Neutral glycolipids in PC12 cells were examined. A major neutral glycosphingolipid, isolated from a chlo-roform/methanol extract of the cells, was found to contain only galactose and glucose at a ratio of 3: 1 and identified as ceramide tetrahexoside by fast atom bom-bardment (FAB) mass spectrometry. Its saccharide sequence was determined by a new method developed here using endoglycoceramidase (Ito, M., and Yamagata, T. (1986) J. Biol. Chern. 261, 14278-14282). The glycosphingolipid was digested with endoglycoceramidase to produce oligosaccharide which was sub- sequently pyridylaminated. The fluorescence-labeled oligosaccharide was digested with a series of specific exoglycosidases and fractionated by high performance liquid chromatography. The 2-aminopyridyl oligosaccharide was hydrolyzed by a-galactosidase to give a 2- aminopyridyl oligosaccharide which was identified as 2-aminopyridyl lactose

The PC12 pheochromocytoma cell line is a clone of a rat adrenal medullary tumor cell (1) and has been found to serve well as a model for the study of neuronal differentiation. Since there is increasing evidence indicating the participation of glycolipids in cell differentiation (2), we considered it worthwhile to direct our attention to glycolipids in PC12 cells. Several reports are presently available on glycolipids in PC12 cells. Margolis et al. (3,4) reported a thin layer chromatographic pattern of gangliosides and the presence of fucosylgangliosides of PC12 cells. Schwarting et al. (5) has recently reported minor neutral glycolipids, fucosyl-and galactosylglobotetraosylceramide. In consideration of these data, we at-tempted to determine the relationship between glycolipid expression and cell differentiation in PC12 cells. In so doing, the data obtained indicated glycolipid compositions differing from that maintained by previous reports (3)(4)(5).
This report presents the structure of a major glycolipid (designated Gal-Gba),l whose presence in PC12 cells has so far not been reported in previous papers. A new method for determining saccharide sequences of glycosphingolipids using endoglycoceramidase (6), a novel enzyme cleaving their saccharide-ceramide linkage, is also described. A preliminary report of this study was presented at the IXth International Symposium on Glycoconjugates in Lille (7).

EXPERIMENTAL PROCEDURES
Cell Culture"PC12 cells (kindly provided by Dr. T. Amano of this Institute) were cultured as described (1) in RPMI-1640 medium supplemented with 5% fetal calf serum and 5% heat-inactivated horse serum.
Extraction and Fractionation of Glycolipids from PC12 Cells-The cells were extracted with chloroform/methanol (2:l and 1:2) sequentially. The extracts were combined and then chromatographed on a DEAE-Sephadex A-25 column equilibrated in chloroform/methanol/ water (30608) as previously described (8) so as to separate neutral glycolipids from gangliosides. This neutral glycolipid fraction was purified by HPLC (Gilson, France) with an Iatrobeads (Iatron Laboratories, Japan) column. Glycolipids were eluted in chloroform/ methanol gradients ranging from 9010 to 65:35. Individual glycolipids were further purified on high performance thin layer chromatography (HPTLC) plates (E. Merck AG, Federal Republic of Germany) developed in chloroform/methanol/O.O2% aqueous CaClz (65:25:4).
Labeling Experiment-In some experiments, cells were labeled with [1-3H]Gal (6.9 Ci/mmol, Amersham) at a concentration of 10 pCi/ml for 24 h. Neutral glycolipids were prepared from the chloroform/ methanol extract of the cells as described above and analyzed by HPTLC. Glycolipid bands were detected by fluorography for 40 days at -80 "C using Kodak X-Omat film after soaking the plate in diethyl ether solution of 2,5-diphenyloxazole.
Saccharide Composition Analysis-Saccharide compositions of glycolipids were analyzed by the modified method of Honda et al. (9). Briefly, a glycolipid sample was hydrolyzed in 2.5 N trifluoroacetic acid for 5 h at 100 'C. The solvent was evaporated, and the residue was redissolved with 50 pl of water. An aliquot of the solution was injected into an HPLC system (Gilson, France) equipped with a CDR-10 column (Mitsubishi Kasei Corp., Japan). The separated monosaccharides were postlabeled with 1% 2-cyanoacetamide and detected by a voltametry detector (VMD501, Yanako, Japan).
Fatty Acid Analysis-Glycolipids were treated with 5% HCl in methanol at 80 "C for 11 h, and the fatty acid methyl esters thus produced were analyzed by gas-liquid chromatography (GLC). GLC was carried out with a GC-5A chromatograph and CBP-10 capillary column (Shimadzu, Japan). The fatty acid methyl ester mixture used The abbreviations used are: Gal-Gba, II13Gala-globotriaosylceramide; HPTLC, high performance thin layer chromatography; HPLC, high performance liquid chromatography; CTH, ceramide trihexoside; Gbr, globotetraosylceramide; GA1, asialogangliotetraosylceramide.

60'
Japan) was used. A glycolipid sample dissolved in chloroform/methanol (2:l) was mixed with triethanolamine and 1,1,3,3-tetramethylureaonasamplehold-40er, and the analysis was performed by detecting negative ions according to Arita et al. (12). Partial Degradation of Gal-Gb3-Gal-Gbs (about 1 mg) was incubated with a-galactosidase (from green coffee, Sigma, 250 milliunits) in 0.1 M acetate buffer (300 ~1 , pH 6.0) containing sodium taurodeoxycholate (1 mg/ml) at 37 "C for 10 h. The digest was extracted with chloroform, and the components thus obtained were separated by preparative TLC using HPTLC plates. Glycolipids were located by 12 vapor. The remaining Gal-Gbs was recovered and digested with agalactosidase again. The digest was fractionated in the same manner as above. The resulting CTH was collected and subjected to 'H NMR spectroscopic analysis and then to methylation analysis. Determination of the Saccharide Sequence of Gal-Gbs-Gal-Gb3 (50 nmol) was digested with endoglycoceramidase (total, 12 milliunits) prepared as described below in 0.1 M acetate buffer (pH 6.0, containing 1 mg/ml sodium taurodeoxycholate, 50 pl) for 3 days at 37 "C. Chloroform/methanol(2:1,1 ml) was added to the digest. The solution was shaken and centrifuged at 2000 rpm for 10 min. The upper phase containing the resulting oligosaccharide was removed, lyophilized, and pyridylaminated according to the method of Hase et al. (13). The 2-aminopyridyl oligosaccharide was purified by Sephadex G-15 column chromatography (13) and further purified by HPLC in the same manner for analysis of 2-aminopyridyl oligosaccharides as described below. Two-fifths of the 2-aminopyridyl oligosaccharide were used to determine the saccharide sequence of Gal-Gb3. The 2-aminopyridyl oligosaccharide was dissolved in 10 pl of 0.05 M acetate buffer (pH 5.5) and digested with a-galactosidase (3.3 milliunits, from green coffee, Sigma) for 1 day at 37 "C. It was further digested with pgalactosidase (0.5 unit, from Escherichia coli, Sigma) for 1 day at 37 "C. These reactions were monitored by HPLC analysis, using a Trirotar VI HPLC system (JASCO, Japan), TSK gel NHz60 column (4.6 X 260 mm, Tosoh, Japan), and FP-210 fluorescence spectrometer (JASCO, Japan), under the following conditions: elution, CH&N/ Hz0 linear gradient ranging from 7525 to 2080; flow rate, 0.5 ml/ min; column temperature, 43 "C. Detection was conducted by a fluorescence spectrometer with excitation at 310 nm and emission at 400 nm.
2-Aminopyridyl oligosaccharides derived from glucose, lactose, CTH, and globotetraosylceramide (Gb,) (both from porcine erythrocytes) used as the standard markers were prepared in the same manner as described above.
Preparation of Endoglycoceramidme-Endoglycoceramidase was isolated from the culture filtrate of Rhodococcus sp. G-74-2 as described previously (6). Endoglycoceramidase was further purified by DEAE-and octyl-Sepharose column chromatography in order to remove contaminating oligosaccharides at the final stage of enzyme preparation.

RESULTS
Glycolipid Composition of PC12 Cells-PC12 cells were extracted with chloroform/methanol(2:1 and 1:2), and the components obtained were developed on HPTLC. Fig. lA shows clearly the appearance of phospholipids and cholesterol. In addition, a major band of a glycolipid, judging from its characteristic color, is present nearly at the position of the Gb, isolated from porcine erythrocytes. Although this glycolipid (tentatively named Gal-Gbs) was predominant on the TLC plate (Fig. IA), another faint spot representing another glycolipid was also noted at the position of the marker, asialogangliotetraosylceramide (GA1). Because of its scanty amount, this glycolipid is not dealt with in this paper. The occurrence of the glycolipid, Gal-Gba, was confirmed by the labeling experiment of PC12 cells with 1-[3H]Gal. PC12 cells metabolically labeled with 1-[3H]Gal were extracted with chloroform/methanol and processed in the same manner as the unlabeled cells to prepare the neutral glycolipid fraction. It was developed on HPTLC, and glycolipids were detected by fluorography (Fig. 1B). A pair of bands (due to the heterogeneity of the ceramide portion), one clear and the other weak, was observed, its position coinciding with that of the unlabeled Gal-Gb3. To characterize the main glycolipid of PC12 cells, Gal-Gb3 was isolated from unlabeled PC12 cells and analyzed for detailed structure. Isolation and Characterization of Gal-Gba-Chloroform/ methanol extracts of PC12 cells were fractionated by HPLC equipped with an Iatrobeads column. The eluates were monitored by TLC, and the fractions containing Gal-Gba were combined. Gal-Gba was further purified by preparative HPTLC. From about 1 g (wet weight) of cells, 0.1 mg of Gal-Gb3 was obtained. When analyzed for saccharide composition, only galactose and glucose in the ratio of 3:l were found present. Considering this and its mobility on HPTLC, the new glycolipid (Gal-Gb3) may be reasonably concluded to be ceramide tetrahexoside. This was confirmed by FAB-mass spectrometry (Fig. 2) and 'H NMR spectroscopy (Fig. 3). The Determination of Saccharide Sequence of Gal-Gba-Gal-Gb3 was digested with endoglycoceramidase (6) followed by pyridylaminating oligosaccharide thus obtained (13). The 2-aminopyridyl oligosaccharide was digested with a series of exoglycosidases. The digests were separated by HPLC, and the results are shown in Fig. 4. The 2-aminopyridyl oligosaccha- ride was susceptible to a-galactosidase, but not to B-galactosidase or glucosidases. 2-Aminopyridyl trisaccharide transiently appeared (Fig. 4C) with a-galactosidase digestion, and, after that, 2-aminopyridyl lactose was generated in the expense of the trisaccharide (Fig. 40). The resulting 2-aminopyridyl lactose was only digestible with P-galactosidase, yielding 2-aminopyridyl glucose (Fig. 4E). These results revealed that Gal-Gb3 oligosaccharide has the following structure: Gala-Gala-Gal@-Glc. Based on this along with the information on its 'H NMR spectroscopic analysis, Gal-Gb3 should be considered Gala-Gala-Gal@-Glcp-Cer.
To decide which is applicable, CTH was prepared by the controlled hydrolysis of the terminal Gal from Gal-Gb3 with a-galactosidase, and its structure was analyzed by 'H NMR spectroscopy and methylation analysis. The 'H NMR spectrum of the resulting CTH is shown in Fig. 5A and compared with that of the standard CTH (Galal-4Gal~l-4Glc~l-lCer) prepared from human erythrocytes (Fig. 5B). The former was precisely identical with the latter except for several peaks due to contaminants. It should be noted that the chemical shifts of the signals of anomeric protons in Galal-3 have been reported to be present at a somewhat higher field than that in Galal-4 (14)(15)(16). It is thus quite clear that the linkage of the terminal galactose of this CTH is Galal-4. Methylation analysis of the CTH derived from Gal-Gb3 also supports this conclusion. The peaks of partially methylated alditol acetates due to nonreducing end Gal, 4-substituted Gal, and 4-substituted Glc were observed nearly in the same amounts, and no peak due to 3-substituted Gal was detected. The results of these two experiments provide an adequate basis for concluding Gal-Gb3 to include the structure of the standard CTH.

DISCUSSION
Our present data show the structure of the major glycolipid in PC12 cells to be Galal-3Galal-4Gal@1-4Glc@l-1Cer. A glycolipid having this structure was previously reported by Angstrom et al. (17), although apparently without sufficient evidence to substantiate its structure. They isolated the glycolipid from rat small intestine as a minor component and postulated its structure to be Galal-3Galal-4Gal~l-4Glc~l-lCer, based on NMR measurement. The saccharide linkages of the nonreducing end and penultimate a-galactosyl residues (Galal-3Galal-4. . .) of the glycolipid were only empirically assigned by chemical shifts of the H-1 protons of permethylated glycolipid in NMR spectroscopy (17). In contrast, our conclusion is based unambiguously on linkage analyses on the degradation product (CTH) of Gal-Gb3.
The PC12 cells used in the present study are a good source of the glycolipid with the unusual structure, Galal-3Galal-4Gal@l-4Glc@l-lCer, since it is the major glycolipid in PC12 cells that can be used as an immunogen to obtain new classes of monoclonal antibodies as reported recently (18).
Very slight amounts of glucosylceramide, CTH, and an unidentified glycolipid migrating nearly at GA1 on an HPTLC plate were found in PC12 cells along with Gal-Gb3. Interestingly, the glycolipid compositions of PC12 cells determined by us do not agree with the previous report of Schwarting et al. (5), who showed Gb4 (GalNAc~l-3Galal-4Gal@l-4Glc@l-1Cer) to be most abundant. It is known that the nature of PC12 cells transforms in part during repeated subcultures without artificial treatment by a mutagen and some subclones have been established (17)(18)(19)(20)(21)(22). This discrepancy may possibly arise from differences in the subclones of the PC12 cells used. Differences in glycolipid compositions may also be due to the particular experimental conditions used. Our PC12 cells were cultured as a monolayer, but cells were grown in a spinner culture by Schwarting et al. (5). It is clear that we have subclones of PC12 cells whose glycolipid compositions are mutually quite different. Taking into consideration recent information on the participation of glycolipids in cell differentiation, especially in that of PC12 cells (23-25), these subclones can be used to resolve the relationships between cell differentiation and membrane glycolipids. We recently reported the presence of a glucose polymer in the same PC12 cells used in this study (26). Mutual glycolipid compositions are quite different from each other, and the presence of the glucose polymer is so unique to PC12 cells that a search for the glucose polymer in the PC12 strain of Schwarting et al. (5) may yield significant results.
We have developed a new method in which the specificity of endoglycoceramidase (6) is used for the determination of the saccharide sequence of PC12 glycolipids. For this purpose, endoglycoceramidase was further purified from contaminating

3-0 3 -4 pprn
oligosaccharides in the previous preparations. When a glycolipid is digested with endoglycoceramidase, the intact oligosaccharide and ceramide are obtained at the same time (6). The resulting reducing end of oligosaccharide can be labeled with 2-aminopyridine (13). A 2-aminopyridyl oligosaccharide thus obtained can be detected by a fluorophotometer at high sensitivity. The saccharide sequence can be determined by digestion with a series of specific exoglycosidases. The saccharide sequence of the new glycolipid from PC12 cells was determined only with 20 nmol of the sample in the present study. The amount of exoglycosidases to digest the material must be made as little as possible in this highly sensitive analysis of saccharide sequence since fluorescent contaminants are present in enzyme samples. Another advantage of