Characterization of Tumor-associated Ganglio-N-triaosylceramide in Mouse Lymphoma and the Dependency of Its Exposure and Antigenicity on the Sialosyl Residues of a Second Glycoconjugate”

Ganglio-N-triaosylceramide (GalNAc/31+4Galfil+ 4Glc/3141Cer), a tumor-associated marker for L5178 cells, was previously reported to separate on thin layer chromatography into three distinct bands (bands a, b, and c). The present paper describes the characteriza- tion of these bands and the factor that determines the degree of glycolipid exposure at the cell surface and its antigenicity. 1) The resolution of ganglio-N-triaosyl- ceramide into three bands was found to be due to molecules having different fatty acid compositions. Band a contained nervonic (C24:l) and lignoceric (C24:O) acids, band b contained palmitic acid (C16:0), and band c contained a-hydroxypalmitic acid. 2) Sur- face labeling of L5178c127 cells with galactose oxi-daselsodium borotritide, followed by fluorography of the isolated glycolipids, revealed that all three bands were exposed on the surface of the cell. However, treatment of cells with sialidase before treatment with galactose oxidase resulted in a 10-fold increase of label incorporated into ganglio-N-triaosylceramide. Since no sialylated form of ganglio-N-triaosylceramide was detected on these cells, and no change in the chemical amount of this glycolipid could be detected, the in- crease were for determination of radioactivity and determination of protein by the fluorescamine assay ml of cell suspension was extracted with chloroform/methanol (21) and extract equivalent to 2 X lo6 cells (163 pg of protein) was applied to high performance TLC plate. Regions on the plate that corresponded to GgOsesCer and GgOseJer (located by fluorography) were scraped and counted for radioactivity. The chemical quantity was analyzed by high performance TLC after reaction with the orcinol/ sulfuric acid reaction according to the method as previously described (45). Both activity and chemical quantity are the values from 2 lo6 chemical quantity GgOseSCer summation of three and c.

Ganglio-N-triaosylceramide (GalNAc/31+4Galfil+ 4Glc/3141Cer), a tumor-associated marker for L5178 cells, was previously reported to separate on thin layer chromatography into three distinct bands (bands a, b, and c). The present paper describes the characterization of these bands and the factor that determines the degree of glycolipid exposure at the cell surface and its antigenicity. 1) The resolution of ganglio-N-triaosylceramide into three bands was found to be due to molecules having different fatty acid compositions. Band a contained nervonic (C24:l) and lignoceric (C24:O) acids, band b contained palmitic acid (C16:0), and band c contained a-hydroxypalmitic acid. 2) Surface labeling of L5178c127 cells with galactose oxidaselsodium borotritide, followed by fluorography of the isolated glycolipids, revealed that all three bands were exposed on the surface of the cell. However, treatment of cells with sialidase before treatment with galactose oxidase resulted in a 10-fold increase of label incorporated into ganglio-N-triaosylceramide. Since no sialylated form of ganglio-N-triaosylceramide was detected on these cells, and no change in the chemical amount of this glycolipid could be detected, the increase of label into this molecule was due to the exposure by sialidase of a normally cryptic glycolipid. The exposure of ganglio-N-triaosylceramide after sialidase treatment was also reflected by the increased sensitivity of these cells to monoclonal antibodies to the glycolipid and complement after enzyme treatment. Thus, the results provide clear evidence that crypticity, as well as antigenicity, of a membrane glycolipid is determined by the degree of sialylation in a second membrane glycoconjugate.
Among various functional notions assigned to glycolipids (1,2), their role as cell type-specific markers and antigens is the most conclusive (3,4). A great deal of current interest has been aroused by tumor-associated glycolipid antigens (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17), which can result from a change of glycolipid composition and metabolism associated with oncogenic transformation (1, 2). Thus, a tumor-associated glycolipid can be found chemically * This investigation was supported by National Institutes of Health Grants CA20026 and GM23100 and American Cancer Society Grant BC9K. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. in tumor cells, which is absent or present in small quantity in normal cells. One important factor which affects expression of a glycolipid antigen at the cell surface is its degree of crypticity associated with the change of glycolipid organization in the membrane. The degree of glycolipid exposure, as determined by cell surface labeling or immunochemical reactivity, is affected by factors (lectins and proteases) that cause changes in the protein structure and distribution of other membrane components (18,19,44).
Ganglio-N-triaosylceramide accumulates in Kirsten virustransformed 3T3 cells and tumors derived therefrom in BALB/c mice (11). Similarly, this glycolipid is found in mouse lymphoma L5178c127, a tumor of DBA/2 origin (20). Since this glycolipid was virtually absent or present in small quantities in various tissues and organs of these mice, the glycolipid can be regarded as an excellent model for a tumor-associated glycolipid marker. GgOse3Cer' has been successfully utilized as a target for the antibody-dependent, avidin-mediated killing of cells by biotinyl neocarzinostatin (21) and a monoclonal IgG3 antibody directed to this marker was used successfully for the serotherapy of the L5178c127 tumor in DBA/2 mice (12). This paper describes the complete chemical characterization of this glycolipid and the factors affecting its exposure and antigenic reactivity on the cell surface. The difference between bands a, b, and c became evident upon examination of the "ceramide window" of the mass spectrum and analysis of the fatty acid composition. All three direct probe mass spectra had an ion at m/e 253 suggesting the presence of C:18 4-sphingenine as the long chain base. The differences between the mass spectra of bands a, b, and c ( Fig. 2) were thus interpreted as being due to differences in the fatty acid composition of GgOse3Cer from each band. Band a representing GgOse3Cer containing nervonic acid (C24:l) and lignoceric acid (C240), band b representing GgOseSCer containing palmitic acid (160), and band c representing GgOse3Cer containing a-hydroxypalmitic acid (hC160) would be consistent with the ions detected. Gas chromatography-mass spectrometry of fatty acid methyl esters before and after trimethylsilylation (30) conclusively demonstrated that bands a and b contained normal fatty acids whereas band c had a-hydroxypalmitic acid? These results confirmed the predictions made from the direct probe mass spectral analysis and firmly established the difference in TLC migration of bands a, b, and c as due to differences in the fatty acid content of the GgOse3Cer molecules. The relative amount of each fatty acid present in each band is summarized in Table I. Other Glycolipids Present in W178c127 Cells-Bands d and e (Fig. 1) were earlier identified as ganglio-N-tetraosylceramide (GgOseJer) (21).

MATERIALS AND METHODS~
Small quantities of gangliosides were present on these cells. Three resorcinol staining bands that migrated close to each Analysis of the trimethylsilylated methyl glycosides by gas-liquid chromatography revealed an additional peak on the chromatogram for band c that was not seen with bands a and b. This peak eluted just after N-acetylgalactosamine. The mass spectrum of this peak was identical with that of hydroxypalmitate. Moreover, co-injection of band c fatty acid methyl esters and methyl glycosides intensified this peak, confirming this observation. Hydroxypalmitate had partitioned into both the methanol (55%) and the hexane (45%) phases. This observation emphasized one advantage of direct probe mass spectrometry over degradative methods of structural analysis: the entire molecule is examined intact, precluding the loss of components following hydrolysis. other in the TLC region between GM, and GD,. were observed no bands corresponding to GM3 or GM, were found. One preliminary means of distinguishing between types of gangliosides is by exo-and endoglycosidase digestion. Vibrio cholera sialidase will cleave terminal sialic acid from glycolipid, while GM, is not affected by this treatment (34). Endo-@-galactosidase recognizes a lactosamine structure (R-GlcNAc~1+3Ga1~1+4Glc+R) and cleaves the glycolipid at the internal galactose (36). Thus, sialylparagloboside is digested to glucosylceramide with this enzyme, while GM, is not affected. Endo-&galactosidase had no effect on L517&127 gangliosides under conditions that digested sialylparagloboside to glucosylceramide (data not shown). Digestion of L517&127 gangliosides with V. cholera sialidase under conditions that digested sialic acid from sialylparagloboside   -, less than 1% of total fatty acid. resulted in the appearance of two neutral glycolipids that migrated on TLC with spots d and e (GgOse4Cer) (Fig. 3). Weak acid hydrolysis of L5178 gangliosides yielded neutral glycolipids d and e as well (results not shown). These results suggested that all the gangliosides present on L5178c127 cells had the gangliotetrasaccharide core and that among these were some that bore terminal sialic acids (labile to sialidase). Such a ganglioside, GMlb, has been obtained in trace amounts from human erythrocyte stroma (37), and the presence of sialidase-susceptible gangliosides in murine lymphocytes has been reported (38,39). L517&127 cells seem to contain this ganglioside also. Thus, all glycolipids present on L5178c127 cells were derived from the ganglio family of glycolipids and were represented by glucosylceramide, lactosylceramide, GgOsenCer, GgOse4Cer, and gangliosides, including GMlb. The alternate pathway for ganglioside biosynthesis (35,(40)(41)(42) that would result in the formation of these compounds is in contrast to the established pathway for GM, synthesis (for review, see Ref. 2).
Surface Exposure of L.5178~127 Glycolipid and the Degree of Exposure and Antigenicity are Defined by the Sialosyl Residue at a Second Clycoconjugate-Glycolipid bands a, b, c, d, and e were readily labeled by treatment of cells with galactose oxidase (Fig. 4, lane 2). No label was associated with these bands by treatment of cells with sodium borotritide alone (Fig. 4, lane 1 ) . Clearly, all three gangliotriaosylceramide bands were exposed on the surface of the cell.
Treatment of cells with sialidase prior to galactose oxidase resulted in a substantial increase in label that was chloroform/ methanol-soluble. As demonstrated in the experiment summarized in Table 11, galactose oxidase treatment of cells followed by reduction with NaB3H, resulted in the extraction of 198 cpm/pg of protein by the lipid solvent. If the cells were first treated with sialidase, this value was increased to 1283 cpm/pg of protein.
Since we had shown that there were gangliosides present that yielded GgOse4Cer after treatment with sialidase ( Fig.  3), we initially predicted that the increased label would appear in this neutral glycolipid. As expected, label associated with GgOse4Cer was increased 24-fold (Table 11) after sialidase treatment of cells, and the amount of chemically detectable GgOse4Cer was increased (Table 11).

TABLE I1
Labeling ojglycolipids in L5178c127 ceUs and their chemical quantity 2 X 10' cells were labeled with NaB3H4 after enzyme treatment. Labeled cells were suspended in 0.54 ml of NaCl/Pi and 0.01-ml aliquots were taken for determination of radioactivity and for determination of protein by the fluorescamine assay (43). 0.5 ml of the cell suspension was extracted with chloroform/methanol (21) and extract equivalent to 2 X lo6 cells (163 pg of protein) was applied to high performance TLC plate. Regions on the plate that corresponded to GgOsesCer and GgOseJer (located by fluorography) were scraped and counted for radioactivity. The chemical quantity was analyzed by high performance TLC after reaction with the orcinol/ sulfuric acid reaction according to the method as previously described (45). Both activity and chemical quantity are the values from 2 X lo6 cells. The chemical quantity of GgOseSCer is the summation of three spots, a, b, and c. Spots a and b were estimated to be 0.4 pg; spot c was 0.1 pg. result. Label associated with GgOse3Cer (bands a, b, and C) was enhanced 11-fold (Table I1 and Fig. 4,?lane 3), with no detectable change in the chemical amount of these bands (Table 11), and since no sialylated forms of this glycolipid were found on these cells (Fig. 3), the increase in label was ascribed to the increased exposure of GgOse3Cer after sialidase treatment. This phenomenon was also revealed by the enhanced reactivity of sialidase-treated cells to monoclonal antibodies directed to GgOse3Cer (Fig. 5). Cells treated with sialidase (Fig. 5, open symbols) were 10 times more sensitive to complement-mediated lysis than untreated cells (Fig. 5 , closed symbols). It is interesting to note that treatment of cells with trypsin prior to treatment with galactose oxidase had no effect on the exposure of GgOsesCer or GgOse4Cer (Table 11).

DISCUSSION
The mouse lymphoma tumor-associated glycolipid, Gg-Ose3Cer, was resolved by TLC into three distinct bands (a, b, and c). By direct probe mass spectrometry of the intact permethylated glycolipids and analysis of the glycolipid fatty acids of each band by gas chromatography-mass spectrometry, these differences were found to be in the fatty acid composition of the molecules. Band a was GgOse3Cer which contained nervonic (C241) and lignoceric (C24:O) acids, band b contained palmitic acid (C16:0), and band c contained a-hydroxypalmitic acid. Since band a showed the same mobility as an authentic GgOse3Cer isolated from guinea pig erythrocytes, band b had the same mobility as authentic globo-N-tetraosylceramide from human erythrocytes, and band c coincided with lactoneo-N-tetraosylceramide, identification of glycolipids by migration on TLC alone is misleading.
By cell surface labeling with galactose oxidase/sodium borotritide, we were able to show that all three bands were exposed on the surface of the cell and potentially available for interaction with antibodies directed towards the carbohydrate determinant expressed on these molecules (20). We observed, unexpectedly, that after sialidase treatment, the amount of label incorporated into GgOse3Cer by this technique increased 11-fold. No change in the chemical levels of this glycolipid could be detected. Since the only sialidasesensitive gangliosides present on these cells had the Gg-Ose4Cer core (GMlb), the increased label incorporated into GgOse3Cer was due to the exposure of normally cryptic glycolipid to the activity of galactose oxidase. The increase exposure of GgOse3Cer was also revealed by monoclonal antibodies directed to this glycolipid. Cells treated with sialidase were 6-10 times more sensitive to complement-mediated lysis with this antibody than were untreated L517&127 cells. Clearly, the antigenicity of these cells was influenced by the sialylation of other membrane components.
This study revealed two aspects about the antigenicity of glycolipids that are important in understanding the potential of these molecules as tumor markers.
1) The reactivity of L5178c127 cells with antibodies directed to GgOse3Cer was the result of the expression on these cells of a complex family of ganglio-N-triaosylceramides. The heterogeneity of ceramides containing the same carbohydrate moiety raises the intriguing possibility that the specific activity of these compounds with antibody may differ. Perhaps GgOse3Cer with a-hydroxylpalmitic acid is more exposed or exposed in a different orientation in the cell membrane, and is hence more reactive with antibody.
2) The organization and topography of the membrane are important parameters in defining the exposure and subsequent antigenicity of these molecules. Several examples of the importance of membrane organization to glycolipid antigepicity exist in the literature. Human fetal erythrocytes, for example, are much more reactive to anti-globo-N-tetraosylceramide, even though the chemical level of this glycolipid in fetal and adult erythrocytes is nearly identical (18). Human adult erythrocytes become reactive to anti-globo-N-tetraosylceramide only after sialidase or protease digestion (18). Globo-N-tetraosylceramide can, however, be labeled equally well by the galactose oxidase-sodium borotritide procedure, with or without treatment of protease or sialidase, suggesting that it is exposed on the surface of adult erythrocytes, but hidden from the view of the larger antibody molecules (19). The reactivity of globo-N-tetraosylceramide with galactose oxidase at the surface of hamster fibroblast NIL cell was twice enhanced when cells were pretreated with a small quantity of concanavalin A or Ricinus communis lectin (19). The present study indicates a much more drastic change of crypticity of glycolipid induced by sialidase, and so far, this is the clearest evidence that the crypticity and the antigenicity of a glycolipid are affected by a second membrane component.
This study clearly established the molecular basis for the ganglio-N-triaosylceramide antigencity of this T-cell lymphoma. This analysis also enabled us to better understand the glycolipids present on normal murine lymphoid cells, where sufficient numbers of cells for structural analysis are difficult to obtain and information about the structure of glycosphingolipids in these cells is scarce.