Relative Glycosyltransferase Levels Determine Ganglioside Patterns1

(CANCER RESEARCH 52, 5725-5731, October 15, 1992] Glycosylation Pathways in the Biosynthesis of Gangliosides in Melanoma and Neuroblastoma Cells: Relative Glycosyltransferase Levels Determine Ganglioside Patterns1

Shutian RÃºanand Kenneth O. Lloyd2

Memorial Sloan-Kettering Cancer Center, New York, New York 10021

ABSTRACT The pathways leading to the biosynthesis of the various major gangliosides of mammalian cells are quite well understood. In order to elucidate some of the factors that determine the charac Specifically, three major pathways (designated a, b, and c) are teristic expression of gangliosides in malignant melanoma and neuro considered to lead to a, b, and c series gangliosides respectively blastoma the levels of ganglioside synthases (glycosyltransferases) were (Fig. 1). GM3 synthase, GD3 synthase, and possibly GT3 syn determined in a panel of cell lines from those tumors that exhibited a wide range of ganglioside composition. Sialyltransferases (GM.v (;,,,. thase play key roles in determining the prominence of each <â€¢!,,â€ž.andCub synthases), /V-acetylgalactosaminyltransferases (GM2 pathway (12-15). Less information is available, however, on the and G|>2 synthases), and galactosyltransferase (G|\n and GUM,syn factors that determine the overall ganglioside expression exhib thases) were analyzed in crude membrane preparations from these cells. ited by a particular cell type. In an initial attempt to analyze The results confirmed the importance of GM, and G,(, synthases in factors that control the characteristic ganglioside patterns in determining the prominence of the a (G\i.. to GI ,.,) or b (G|M to G(_,n,) melanoma and neuroblastoma tumors we have correlated gan biosynthetic pathways. The overall ganglioside composition in cells was glioside synthase levels with ganglioside expression in a number found to be dependent on the relative levels of specific enzymes acting of melanoma and neuroblastoma cell lines that exhibit a spec sequentially or in competing pathways. In general, the pattern and levels trum of different ganglioside patterns. The results show that the of transferases correlated with the actual ganglioside content of the cell line, although several important discrepancies were noted. For example, major factors in determining the ganglioside phenotype of a cell in cell lines containing high amounts of Gm ganglioside, the level of the are the relative ratios of the specific glycosyltransferases and the preceding enzyme in the pathway (<;,,, synthase) was unexpectedly low. levels of acceptors which together interact to produce the final Thus, the high (,,,,:<;,,, ratios characteristic of most neuroblastomas ganglioside profile. result from low levels of G|M synthase as well as high levels of G|>2 synthase. In other cell lines, (.,.1 synthase was completely absent, re sulting in the synthesis of <>\i>.but not (.,,> by yV-acetylgalactosami- MATERIALS AND METHODS nyltransferase I, as would be expected. It was concluded that different Cell Lines and Cell Culture. The derivation and culture conditions glycosyltransferases play key roles in determining glycolipid expression of all human melanoma cell lines used in this study have been described in different cell types. (16). Mouse melanoma cell line B78 is a subline of the B16 line (17). JB-RH was obtained from Dr. Jane Berkelhammer (University of Mis INTRODUCTION souri, Columbia, MO) and subcloned (18). Human neuroblastoma cell lines SH-SY5Y, SH-EP1, BE(2)C, and SK-N-MC were kindly provided Gangliosides (sialylated glycosphingolipids) have been found by Dr. June Biedler (Memorial Sloan-Kettering Cancer Center, New to be significant antigens in tumors of neural crest origin, i.e., York, NY); others were maintained in our laboratory. The neuroblas toma cell lines were cultured in a mixture of Eagle's minimum essential melanomas, astrocytomas, and neuroblastomas (reviewed in Refs. 1-3). For example, normal melanocytes express GM33 as medium with nonessential amino acids and Ham's nutrient mixture F-12 supplemented with 10% heat-inactivated fetal bovine serum. Cells their major gangliosides, whereas malignant melanomas syn were passaged by incubation with 0.02% EDTA and 0.12% trypsin in thesize GD.I, as well as GM1, as their major gangliosides (4, 5). phosphate-buffered saline. All cells were harvested in phosphate-buff A few melanomas also synthesize small amounts of more com ered saline and used immediately for glycolipid extraction and cell plex gangliosides, e.g., GM2 and GD2 (6). In neuroblastomas, on membrane preparation or stored at -80Â°C before use. the other hand, GM2 and GD2 are the major gangliosides in Chemicals and Reagents. Gangliosides (GMJ, GM2, GMI, Gma, most cell lines and tumors (7, 8). These gangliosides are recog GDib, GTibi and Lac-Cer) were purchased from Sigma Chemical Co. nized by many of the mouse (2) and human (1) monoclonal (St. Louis, MO). Gr>.iwas obtained from Calbiochem (San Diego, CA). antibodies that have been developed in recent years. Mouse Their purity was checked by TLC, as described below. Gr>2was isolated monoclonal antibodies to Gnu (9) and Gn2 (10) have been used from human neuroblastoma tissue and kindly provided by Dr. F. Hell in therapy trials in the treatment of melanoma and neuroblas ing (Memorial Sloan-Kettering Cancer Center). CMP-[4,5,6,7,8,9- toma, respectively. Moreover, some gangliosides are immuno- 14C]NeuAc (250 mCi/mmol), UDP-Ar-acetyl-n-[l-1H (N)]galactosamine (8.3 mCi/mmol), UDP-/V-acetyl-D-[l-14C]galactosamine (55 genic in humans and form the basis of vaccines for active im- mCi/mmol), and UDP-[4,5,-1H (N)jgalactose (43.3 Ci/mmol) were pur munotherapy in melanoma patients (11). chased from New England Nuclear (Boston, MA). Phenylmethylsulfo- nylfluoride, 2-mercaptoethanol, Triton CF-54, MgCl2, and MnCl2 were from the Sigma Chemical Co. Cacodylic acid (sodium salt), chlo Received 4/30/92; accepted 8/6/92. roform, and methanol were from Fisher Scientific (Pittsburgh, PA). All The costs of publication of this article were defrayed in part by the payment of other chemicals and reagents used were of analytical grade quality. page charges. This article must therefore be hereby marked advertisement in accord ance with 18 U.S.C. Section 1734 solely to indicate this fact. Extraction and Isolation of Glycolipids from Cells. Methods for the 1Supported by grants from the National Cancer Institute (CA08478 and extraction and isolation of acidic glycolipids were previously described CA21445). (19). Briefly, the total lipids were extracted sequentially from cell pellet 2 To whom requests for reprints should be addressed, at Sloan-Kettering Insti in 30 volumes of 2:1, 1:1, and 1:2 chloroform:methanol. A glycolipid tute, Section 6155. 1275 York Avenue, New York, NY 10021. 3 The abbreviations used are: Lac-Cer, lactosylccramide; TLC, thin-layer chro- fraction was subsequently separated from other lipids by Florosil chromatography; SAT, sialyltransferase. Ganglioside nomenclature is according to matography of the acetylated sample (20). After deacetylation and de Svennerholm (34) salting, the acid glycolipids and neutral glycolipids were separated by 5725

LacCer tions for GD.Isynthase were based on earlier work in our laboratory (5). Assay conditions for GM, synthase were optimized for Lac-Cer and Sialyltransferase I CMP-NeuAc concentrations using membranes from SK-MEL-28 cells; (GM3 synthase) the apparent Kms for these substrates were determined to be 30 and 480 MM,respectively. Assay conditions for GM2 and GD2 synthases were Sialyltransferase optimized for GM.,, GD.i, and UDP-GalNAc concentrations using GM3 GD3 membranes from IMR32 cells; the apparent Afmswere 48 MM(GM.i)and (GD3 synthase) 71 MM(UDP-GalNAc) for GM2 synthase and 42 MM(GD3) for GD2

GalNActransferase I synthase. GM1, Gmb, GDla, and GTlb synthases were assayed under (GM2/GD2 synthase) conditions described previously (23-29). First, the indicated concentra tion of glycolipid acceptor and sugar donor were dried under a gentle GM2 GD2 stream of nitrogen. The mixture was redissolved in an aqueous cacodylate buffer solution containing the detergent and metal ion at the opti mal pH. The reaction was started at 37Â°Cbythe addition of 200 Mgof Galactosyltransferase membrane protein. After 1, 2, or 3 h, 2 ml of water were added, and the reaction mixture was then passed several times through C|8 Sep-Pak GM1 GDlb cartridges that had been pretreated according to the manufacturer's instruction. The cartridge was washed with 25 ml of water to remove Sialyltransferase IV nonlipid components. The bound glycolipids were sequentially eluted with 2 ml of methanol and 3 ml of chloroform:methanol (2:1) and pooled. A portion of the eluted sample was dried and counted by liquid GDla GTlb scintillation spectrometry after the addition of Liquiscint (National Diagnostics, Mannville, NJ). The remainder of the sample was ana lyzed by thin-layer chromatography. For quantitation of the products the plates were scanned with a Berthold linear analyzer (Wildbad, Ger a pathway b pathway many). Enzyme activity was calculated from the ratios of the products Fig. 1. Pathways leading to the biosynthesis of gangliosides (modified from and the total radioactivity of the samples. The radioactive products on Refs. 13 and 14). Pathway c is omitted for simplicity. the TLC plates were also detected by fluorography as described below. Thin-Layer Chromatograph}- and Fluorography. Samples were ap DEAE-Sephadex A-50 (Pharmacia, Inc.. Piscataway. NJ) chromatog- plied to silica gel 60 high-performance TLC plates (EM Science, Ger raphy (21). The ganglioside fraction was desalted with a Sep-Pak C]8 many) and chromatographed in a solvent system of chloroformtmeth- cartridge (Waters; Millipore Co., Milford, MA). The gangliosides were anol:0.2% aqueous CaCl2 (55:45:10). The plates were sprayed with identified by thin-layer chromatography and fluorography (5). resorcinol/HCl for the analysis of glycolipids extracted from cell lines. Membrane Preparation. A crude membrane fraction from mela The ganglioside levels were quantitated after scanning the plates on a noma or neuroblastoma cells was used as the source of enzymes and was Shimadzu TLC scanner (model GS-930) by comparison with standard prepared as follows. Cells were suspended in chilled phosphate-buffered amounts of gangliosides. For radioactive samples the plates were saline containing 1 m\i phenylmethylsulfonylfluoride and 10 imi treated with Enlighting spray (New England Nuclear, Boston, MA) and 2-mercaptoethanol and disrupted by using a nitrogen cavitation appa exposed to X-ray film (XAR-5; Kodak, Rochester, NY). ratus (Artisan Industries, Waltham, MA). Nuclei were removed from the suspension by low-speed centrifugation. The supernatant was re- centrifuged at 100,000 x g (Servali, 625 rotor) for l h at 4Â°C.The RESULTS resulting pellet was resuspended in 0.1 Msodium cacodylate-HCl buffer, pH 6.4-7.35, depending on the ganglioside synthase to be assayed. Ganglioside Composition of Melanoma and Neuroblastoma Protein concentration of the membrane suspension was determined by Cell Lines. Six melanoma cell lines were selected for analysis Lowry's method (22). that were representative of the range of ganglioside patterns Glycosyltransferase Assays. The standard conditions used for the found in melanoma cells. The ganglioside profiles of these six assay of ganglioside synthases are summarized in Table 1. The condi typical melanoma cell lines are shown in Fig. JA. The identi-

Tablc 1 Assay conditions for glycolipid synthases"

SynthaseAcceptorMMDonorMMdpm/nmolMetal

D.I260'HJGalNAc 500CMP-I

500 400 1.000 1.000 9.2MgCI25ISO200506.0501-2GDSGM.I325l4C|NeuAcx IO3MgCI25ISO200506.0501-2GMIGMJ325UDP-I5MnCI2 x IO4MnCI2 1 xIO4MnCl2 IO3MgCl24.6 x

ion DMDetergent 102002001007.2502-3GDIG102002001007.2502-3GMIGM2300GoibGo2250UDP-l'HJGal152502001007.0502MnCl2152502001007.3502GDUGMI260G-ribGDIb220CMP-|'4C)NeuAc101502001006.6SO1-2MgCl2102002001006.4501-2

TritonMgProtein. CF54.

MgCacodylate

MMpHReactivebuffer,

M'Reactivevolume,

time, hGMÃŽLac-Cer " CDP-choline (10 mvi) was added to the GM2. GD2, and GDiÂ»assays and 2-mercaptoethanol (10 m\t) to the GT|b assay. 5726

Melanoma of GM2 and GD2 (MeWo cells), and (e) cells with GM3, GM2i A Grxi, and GD2 and traces of GD,a or GTlb (SK-MEL-31 and SK-MEL-37 cells). All of the melanoma cell lines express GM3 as a major ganglioside (24-100% of total gangliosides). GD3 GM3 was the other major ganglioside detected in human melanoma GM2 cell lines (19-42.5%). GM2 and GD2 (10-44% and 3-33%, re spectively) were found in some of the cell lines and except for GD3 JB-RH cells were always presented together. Trace amounts of Gma and GTlb (5% and 4%) were detected in two cell lines GD2 (SK-MEL-31 and SK-MEL-37). The ganglioside patterns of seven neuroblastoma cell lines were analyzed, and the ratios of the individual gangliosides were determined (Table 2; Fig. 2B). The identity of the gangli osides (except GTlb) was confirmed by TLC immunostaining (data not shown). Compared to melanomas, the ganglioside Neuroblastoma pattern in neuroblastomas was relatively complicated. Seven gangliosides, GM3, GM2, GD3, GD2, GM,, GDla, and GTIb, were found in these cell lines. Specifically, GM3 and GM2 were found GM3 in all the neuroblastomas examined, with the content varying from 5.7% to 56.2% and 16.4% to 39.5%, respectively. Four GM2 cell lines, SK-N-MC, SH-EP-1, SH-SY5Y, and SMS-SAN, ex GMl pressed GM1 (3.5-25.7%) and GmÂ¡1(8.3-30.2%). The concur GD3 rent presentation of GD3 (4.2-25.3%), GD2 (16.8-34.9%), and GOla G-rib (6.5-14%) was found in four cell lines (LA-N-1, IMR-32, GD2 BE(2)C, and SMS-SAN). Based on these data, it is possible to GDlb GTlb classify these neuroblastoma cell lines into three groups based on their ganglioside expression. The first group expresses GMi, GM2, GMI, and GDla gangliosides (SH-EP1, SH-SY5Y, and SK-N-MC); the second group of cells contain GM3, GM2, GD3, Fig. 2. A, thin-layer chromatogram of gangliosides extracted from six mela GD2, and GTlb gangliosides (LA-N-1, IMR-32, and BE(2)C); noma cell lines B16 (Lane 1), JB-RH (Lane 2), SK-MEL-28 (Lane 3), MeWo (Lane 4). SK-MEL-31 (Lane 5), and SK-MEL-37 (Lane 6). B, ganglioside pat and the third group had the most complex pattern of ganglio terns of human neuroblastoma cell lines: Lane 1, SK-N-MC; Lane 2, SH-EP1; sides and contained GM3, GM2, GD3, GD2, GMI, GDla, and Lane 3, SH-SY5Y; Lane 4, SMS-SAN; Lane 5. LA-N-1; Lane 6, IMR-32; Lane GTlb (SMS-SAN). 7, BE(2)C. Lane S, ganglioside standards. The samples were developed on a high-performance TLC plate in a solvent system of chloroform:methanol:0.2% Levels of Ganglioside Synthases in Melanoma and Neuro- aqueous CaClj (55:45:10). Gangliosides were visualized by spraying with resor- blastoma Cells. To probe the relationship between glycolipid cinol-HCI reagent. expression and enzyme level in the biosynthesis of gangliosides, the levels of eight ganglioside synthases were determined in the fication of the individual gangliosides was confirmed by melanoma and neuroblastoma cell lines described above. The TLC immunostaining (data not shown). The proportions of results are summarized in Table 3. individual gangliosides were assessed densitometrically (Table High levels of GM3 synthase were detected in all the mela 2). Five types of ganglioside patterns could be found in these noma cell lines and in one neuroblastoma cell line (SH-EP-1). cells. They were: (a) cells containing only GM3 (B16); (b) cells Relatively lower levels of GN13synthase were observed in neu with GM.I and GM2 (JB-RH cells); (c) cells with GM3 and roblastoma than in melanoma cell lines. In GD3 synthesis, two GD3 (SK-MEL-28 cells); (d) cells with GM3 and GD3 and traces melanoma cell lines, B16 and JB-RH, did not show any activity,

Table 2 Ganglioside composition of melanoma and neuroblastoma cells (%)*Cell Ganglioside composition

lineMelanomas designation"11224443331GM3100

B16 JB-RH 56.2 SK-MEL-28MeWoSK-MEL-37SK-MEL-31NeuroblastomasSK-N-MCSH-EP60.044.524.028.556.038.413.45.721.027.710.4GM243.810.024.516.517.435.230.628.234.439.56.4GMI18.314.825.73.5CDU8.311.630.217.9GDÃ•40.042.519.020.04.26.49.525.3GD!3.023.033.026.431.716.834.9GTib2.04.014.06.55.512.9

1SH-SY 5YSMS-SANIMR-32BE(2)CLA-N-1Group

' See Fig. 4. * As percentage of total ganglioside content determined by TLC and resorcinol-HCl reagent as outlined in "Materials and Methods.' 5727

Table 3 Levels of ganglioside synthase activity in melanoma and neurohlastoma cell lines synthases"Cell Ganglioside

lineMelanomaB16JB

RHSK-MEL-28MeWoSK-MEL-37SK-MEL-31NeuroblastomaSK-N-MCSH

EPISH-SY5YSMS-SANLA-N-1IMR-32BE(2)CGMJ2777104426317433771315065427402641951012818540GDJ00428512321015070360811295GlM2029405317123355489161034782905224GDZ0139068105131581765361261363885GMI00075008213036227217200Goib1230019800000638300GDI.1215801413112614548078210746Gl-lb2201731320328710166148363179158

' Enzyme activity unit: pmol donor incorporated into product/mg membrane protein/h. Average of two experiments (SEs were less than 5%).

Table 4 Products of ganglioside synthesis using mixed acceptors and donors (%)Cell Product of ganglioside synthesis lineMelanomaJB-RHSK-MEL-28SK-MEL-31NeuroblastomaIMR-32SMS-SANSH-SY5YAcceptors"Lac-CerLac-Cer,

GMJLac-Cer. GDILac-CerLac-Cer.GM,,

GM.iLac-Cer, GD1Lac-CerLac-Cer,GMi.

GMjLac-Cer, GDjLac-CerLac-Cer,GMJ,

Gâ€ž.,Lac-Cer, GDJLac-CerLac-Cer,GMJ.

GMiLac-Cer, GD3Lac-CerLac-Cer.GMJ.

GMjLac-Cer, GMJ. GDjGM.I91.271.871.052.037.637.379.940.836.892.947.534.892.825.16.888.036.519.8GDJ00048.062.462.716.035.542.20trÂ»tr0trtr000GM28.828.217.40002.522.512.54.647.624.7073.832.79.062.820.8Go20011.60001.61.28.52.54.940.57.21.160.53.00.749.3 " Concentration of acceptors in assay: Lac-Cer. 500 P.M:GMI. 325 Â»IM:DJ, 260 IIM.pH: 7.2. * Trace, below 0. 1% of total.

whereas the four other melanoma lines had moderate or high cell lines. Three melanomas, SK-MEL-28, SK-MEL-31, and levels of activity. Fresh melanoma tumors also had high levels SK-MEL-37, and one neuroblastoma, LAN-1, expressed of Gn> synthase as well as GM, synthase (data not shown). GD3 moderate amounts of both activities. Three neuroblastomas, synthesis was absent or low in all the neuroblastoma cell lines, IMR-32, SH-SY5Y, and SMS-SAN, contained low levels of except LAN-1. GM2 synthase and GI)2 synthase activities in GTib synthase and traces of GDla synthase. both melanoma and neuroblastoma cell lines were correlated. Ganglioside Synthesis in the Presence of Mixed Donors and Thus, two melanoma cell lines, B16 and SK-MEL-28, demon Acceptors. Since the total pattern of ganglioside expression in strated no GM2 and Gi>2 synthase activity. Melanoma cell lines cells is probably governed by both the total complement of JB-RH, SK-MEL-31, and SK-MEL-37 had moderate levels of synthases present as well as the levels of acceptors and donors, GM2/Gr>2 synthase activity. Neuroblastoma cell lines could be we attempted to partially reproduce this situation in vitro by divided into two groups based on their levels of this enzyme: examining the products produced from two radioactive donors One group (4/7 neuroblastomas) had approximately 9- to in the presence of mixtures of glycolipid acceptors. Specifically, 12-fold higher GM2/GD2 synthesis than the other group cell extracts were incubated with mixtures of CMP-[14C]NeuAc (3/7 neuroblastomas). Low levels of GMi and GDih synthesis and UDP-[14C]GalNAc as donors and with Lac-Cer; Lac-Cer were measured in one melanoma cell line (MeWo), while others and GM,; or Lac-Cer, GM3, and GD1 as acceptors in separate (5/6 melanomas) had no detectable activities. Of the seven neu experiments and the products examined (Table 4; Fig. 3). It was roblastomas, two of them, LA-N-1 and SMS-SAN, expressed noted that, in general, the formation of radioactive products moderate levels of GM, and Gmh synthases; three of them, was dependent on the addition of exogenous acceptor. For in SH-SY5Y, SH-EP1, and SK-N-MC, had only GM1 synthase stance, radioactive GM2 was efficiently formed by SK-MEL-31 activity. CM, and GDth synthesis was not detected in the other membranes only when exogenous GMj was added. Likewise, two neuroblastoma cell lines, IMR-32 and BE(2)C. GDJa and radioactive GD2 was synthesized by the three neuroblastoma Crib synthase levels varied considerably among individual cell membranes only when exogenous GDi acceptor was added. 5728

SK-Mel-28 SK-Mel-31 JBRH be accurately measured; (c) the three neuroblastoma cell lines studied produced very low levels of GD3 ganglioside, even when A provided with GM3 as substrate, whereas the two human melanoma cell lines studied (SK-MEL-28 and MeWo) pro GM3 duced high levels of GD3. GD3 synthase is the dominant enzyme GM2 in the two melanoma cell lines, whereas it is minimally active in the three neuroblastoma cell lines. This result is remarkable, GD3 considering that the neuroblastoma cell lines produce high lev els of GD2 ganglioside, the synthesis of which requires GD3 as an intermediate. GD2

DISCUSSION A major difference between human melanoma cells and most neuroblastoma cells is the high expression of GD3 in the former and its absence or low level in the latter cell type. As reported previously (5), and confirmed in this study, all human mela noma cell lines examined have high levels of GD3 synthase (SAT-2). This is also the case in fresh melanoma tumors (data IMR-32 SMS-SAN SH-SY-5Y not shown). The two mouse melanoma cell lines studied lack B GD3 and have no detectable GB3 synthase levels (Table 3). With the exception of LA-N-1 cells, all of the neuroblastoma cell GM3 lines examined have very low levels of this enzyme. All mela GM2 noma and most neuroblastoma cell lines studied have relatively high values of GM3 synthase (SAT-1). Most neuroblastoma cell lines and two of the melanoma cell GD3 lines studied express GM2 and GD2 as major gangliosides. These cell lines examined had moderate or high levels of GM2/ GD2 synthase, as would be expected (Table 3). Enzyme levels in GD2 the neuroblastoma cell lines were approximately 3-fold higher than those in the two human melanoma cell lines. Since GD3 is -â€”.â€”¿,. Â«â€¢Â» the immediate precursor of GD2 it might be expected that all ^^* ^^W ^^Â» ^^W WW^ cell lines expressing high levels of GD2 ganglioside would also have high levels of GD3 synthase (to provide sufficient levels of acceptor for GD2 synthase). Interestingly, three of the four neu roblastoma cell lines with high GD2 levels had very low GD3 Fig. 3. Ganglioside synthesis in melanoma (A) and neuroblastoma (A) cell lines in the presence of mixed acceptors and donors. Ganglioside synthesis was carried synthase levels (Table 3). It appears that the characteristic gan out using CMP [l4C]NeuAc and UDP [14C]GalNAc as the donors and three glioside pattern of neuroblastoma cells (GD2 Â» GD3) is deter different glycolipid acceptor(s): Lane I, Lac-Cer; Lane 2, Lac-Cer and GM}; Lane 3, Lac-Cer, GM3. and GD.I. The products were identified by thin-layer chroma- mined by low levels of GD3 synthase as well as high levels of tography and autoradiography as described in "Materials and Methods." Go2 synthase. The two melanoma cell lines that express rea sonable levels of GD2 also have high levels of GD3, and this is reflected in their higher GD3 synthase levels. These results from the standard assays were strongly reinforced by carrying out In the case of SK-MEL-28, however, high levels of GD3 were simultaneous assays for GD3 and GD2 synthesis using mixed formed even when only Lac-Cer was added exogenously. The donors and acceptors (Table 4). In the presence of Lac-Cer and GM3 as acceptors, SK-MEL-31 cells synthesized appreciable results from these experiments indicate that the levels of en dogenous glycolipids in the membrane preparations were amounts of GD3 (35.5% of total ganglioside) and little GD2 generally not high enough to ensure that the radioactive prod (1.2%). With the addition of GD3, the level of GD3 synthesized uct produced from the previous acceptor was efficiently con remained high, and the level of GD2 synthesized increased verted to second product. The results reinforced three points slightly (to 8.5%). This was in contrast to the neuroblastoma that were noticed in the previous experiments using the stan cells, which synthesized only trace levels of Gr>3(in the presence dard assays: (a) GM2 and GD2 synthase activities were concom- of GM3 acceptor) but high levels of GD2 (in the presence of GD3 itantly expressed. For all cell lines, except SK-MEL-28 and B78 acceptor). It can be concluded that in these cells the low (in which this enzyme is not present), the combination of GM3 amounts of GD3 produced are efficiently converted to GD2 by and GD3 as acceptors resulted in the production of both GM2 high levels of GD2 synthase, whereas in melanoma cells, the and GD2. This was the case even in JB-RH cells, which express higher levels of GD3 synthase and proportionately low levels GM2 but not GD2 ganglioside (Fig. 2); (b) GM2 and/or GD2 of GD2 synthase ensure that appreciable levels of GD3 leave synthesis was much higher in the neuroblastoma cell lines ex the Golgi and are expressed at the cell surface. The possibility amined than in the melanoma cell lines examined. Thus, even that the high levels of GD2, concomitant with low levels of in SK-MEL-31 cells which produce moderate amounts of GM2 GD3, observed in neuroblastoma cells result from the direct and GD2, the ratio of GM2 synthase to GD3 synthase was 1.0:1. conversion of GM2 to GD2 by a novel sialyltransferase was 59, whereas in neuroblastoma the ratio was higher than could refuted by our inability to detect such an enzyme activity in 5729

GROUP 1 GROUP 2 GROUPS GROUP 4

Lac-Cer Lac-Cer Lac-Cer Lac-Cer

GM3 GD3 GM3 GD3 GM3 GM3

GM2 GD2 GM2 GM2

GD1b GM1 GD1b GM1

GT1b GT1b GDta

Examples: Melanoma cell line Melanoma cell lines: Neuroblastoma cell Neuroblastoma cell SK-MEL-28 and SK-MEL-31 and -37 lines: SMS-SAN, IMR-32 lines: SH-SY5Y, SH-EP1 most melanoma tumors Neuroblastoma cell line and BE(2)C and SK-N-MC LA-N-1 Fig. 4. Pathways leading to the synthesis of gangliosides in melanoma and neuroblastoma cell lines deduced from ganglioside and glycosyltransferase levels. Filled arrow, high levels; open arrow, medium levels: thin arrow, low levels of enzyme and/or product.

SK-SH5Y and SMS-SAN cells.4 This finding again confirms 4). In agreement with this result, these three cell lines had no the complete separation of a and b pathways as in other cell detectable or very low levels of GD3 synthase (the first enzyme types (14, 15). in the b pathway). Whether or not these three lines represent a Another interesting point is the differential expression of particular clinical subgroup of neuroblastomas is the subject of G|vi2 and GD2 gangliosides in the various cell lines and their ongoing studies. relationship to the levels of the corresponding synthase. In our In summary, the ganglioside patterns observed in the mela studies it is noticeable that GM2 synthase activity was always noma and neuroblastoma cell lines studied can be primarily concomitantly expressed with GD2 synthase in both melanoma explained on the basis of the levels of the various glycosyltrans- and neuroblastoma cells (when the appropriate acceptor gly- ferases present in the cells and the availability of substrates on colipid was supplied). Enzymatic studies (15) with rat liver which these enzymes act. Modulations in these parameters re preparations have shown that GM2 and GD2 synthesis is carried sult in the spectrum of different patterns observed in these out by the same enzyme (GalNAc transferase I). Recent studies various cell lines. Mouse melanoma cell lines B78 and JB-RH on the expression of the cloned complementary DNA for Gal have relatively simple patterns and possess only GM3 synthase NAc transferase I have confirmed this conclusion (30). Despite and GM3 and GM2/GD2 synthases, respectively. In the other cell the presence of high levels of GalNAc transferase I in eight of lines more complex arrays of gangliosides are produced, and the cell lines examined this did not invariably result in the their profiles are governed by the relative ratios of the various expression of GD2 ganglioside. Thus, mouse melanoma cell line JB-RH and the neuroblastoma cell lines SK-N-MC, SH-EP1, transferases, with particular transferases playing key roles in and SK-SY5Y all expressed GM2 but not GD2. This pattern of the different cell lines (Fig. 4). The results reinforce the impor tance of SAT-I (GM3 synthase) and SAT-II (GD3 synthase) in expression can be explained by the absence of SAT-II in these governing ganglioside synthesis and, in particular, guiding cells and the resulting lack of GD3, the immediate precursor of the pattern of synthesis toward the a or b pathways (12-15). Go2- These studies, however, also demonstrate another level of con Although four of the seven neuroblastoma cell lines studied had ganglioside patterns rich in GD2, which is typical of most trol. Thus, in some instances low levels of an intermediate neuroblastomas (7, 8, 31), three of the lines had a different transferase, together with high levels of a transferase occurring pattern. These cell lines (SK-N-MC, SH-EP1, and SH-SY5Y) later in the pathway, result in a ganglioside profile in which the had GM2 as the predominant ganglioside and very little or more complex ganglioside predominates. This is most clearly no GD2 (Fig. 2B). GM3, GMn and GDla were the other major seen by comparing the synthetic pathways in group 2 and group gangliosides present. These cell lines clearly synthesize gangli 3 cells (Fig. 4). In group 2 cell lines, GD3 synthase levels are osides of the a pathway rather than the b pathway (Figs. 1 and high, and even in the presence of reasonable levels of GD2 synthase, GD3 is detected as the major disialoganglioside in the cells. In contrast, in group 3 cells (in which the b pathway 4 S. RÃºanand K. O. Lloyd, unpublished results. also predominates), GD3 synthase levels are very low. Thus 5730

Downloaded from cancerres.aacrjournals.org on September 24, 2021. © 1992 American Association for Cancer Research. GANGLIOSIDK BIOSVNTHKTIC PATHWAYS in the presence of moderate or high levels of GD2 synthase, Science (Washington DC). 194: 906-915. 1976. 14. Capullo. R.. Maccioni, H. J.. Arce. A., and Cumar, F. A. Biosynthesis of most of the Gni formed is apparently converted to GD2,result brain gangliosides. Adv. Exp. Mcd. Biol., 71: 27-44, 1971. ing in the observed ganglioside profile. In group 4 cells, there 15. Pohlentz, G., Klein, D., Schwartzmann. G.. Schmilz, D., and Sandoff, K. is even less or no synthesis of Gn, and pathway a products Both GA2. GMi. and Gi>2synthascs and GM Â¡b.G|>|.,and G, ,b synthases are single enzymes in Golgi vesicles from rat liver. Proc. Nati. Acad. Sci. USA, predominate. The possibility of whether the glycolipid patterns Â«5:7044-7048. 1988. are also influenced by the action of cellular glycosidaseswas not 16. Carey. T. E.. Takahashi. T.. Resnick. L. A.. Oettgen, H. F.. and Old, L. J. Cell surface antigens of human malignant melanoma. I. Mixed hemadsorp- examined. tion assays for humoral immunity to cultured autologous melanoma cells. From this and other studies it is apparent that the factors that Proc. Nati. Acad. Sci. USA, 73: 3278-3282. 1976. govern the aberrant synthesis of gangliosides and other carbo 17. Graf, L., Kaplan. P.. and Silagi. S. Efficient DNA-mediated transfer of selectable genes and unselccted sequences into differentiated and undiffer- hydrate antigens in tumor cells and the characteristic expres entiated mouse melanoma clones. Somatic Cell Mol. Genet., 10: 139-151, sion of different structures in different tumors are quite com 1984. plex. Further progress in this area will come from studies on 18. Natoli, E. J., Jr., Livingston, P. O., Pukel, C. S., Lloyd, K. O., Wiegandt, H.. Szalay, J., Oettgen, H. F.. and Old, L. J. A murine monoclonal antibody purified enzymes (32, 33), cloning of the genes coding for the detecting /V-acctyl- and .V-glycolyl-GMj: characterization of cell surface re transferases (30), and an analysis of factors that control their activity. Cancer Res.. 46: 4116-4120, 1986. transcription and translation. 19. Furukawa, K.. Clausen. H.. Hakomori. S., Sakamoto. J., Look, K., Lund- blad. A., Mattes, M. J.. and Lloyd, K.O. Analysis of the specificity of five murine anti-blood group A monoclonal antibodies, including one that ACKNOWLEDGMENTS identifies type 3 and type 4 A determinants. Biochemistry, 24: 7820-7826, 1985. 20. Saito, T.. and Hakomori. S-I. Quantitative isolation of total glycosphingolip We thank Dr. June Biedler, Dr. Marian Myers, and Barbara Spen ids from animal cells. J. Lipid Res.. 12: 257-259. 1971. gler for helpful advice and for providing the neuroblastoma cell lines. 21. Ledecn, R. W., and Yu, R. K. Gangliosides: structure, isolation, and analysis. Methods Enzymol., 83: 139-191. 1982. 22. Lowry, O. H., Rosebrough. N. J., Farr. A. L., and Randall. R. J. Protein REFERENCES measurement with the Polin phenol reagent. J. Biol. Chem.. 193: 265-275, 1. Lloyd. K. O., and Old, L. J. Human monoclonal antibodies to glycolipids and 1951. other carbohydrate antigens: dissection of the humoral immune response in 23. Basu, M., De, T., Das, K. K., Kyle, J. W., Chon, H-C., Schaeper, R. J.. and cancer patients. Cancer Res.. 49: 3445-3451. 1989. Basu, S. Glycolipids. Methods Enzymol., 138: 575-607, 1987. 2. Furukawa, K.. and Lloyd, K. O. Gangliosides in melanoma. In: S. Ferrone 24. Keenan, T., Morre, D., and Basu, S. Concentration of glycosphingolipid (ed.). Human Melanoma. Immunology. Diagnostics, and Therapy, pp. 15- glycosyltransferases in Golgi apparatus from rat liver. J. Biol. Chem., 249: 30. Heidelberg: Springer-Verlag, 1989. 310-315, 1974. 3. Hakomori, S. Bifunctional role of glycosphingolipids. Modulators for trans- 25. Iber. H., Kaufmann. R.. Pohlentz. G., Schwarzmann. G., and Sandhoff, K. membrane signaling and mediators for cellular interactions. J. Biol. Chem., Identity of GAI, GM|U and Gi>ib synthase in Golgi vesicles from rat liver. 265: 18713-18716, 1990. FEBS Lett.. 248: 18-22. 1989. 4. Carubia, J. M., Yu, R. K.. MÃ¡cala,L. J., Kirkwood. J. M., and Varga, J. M. 26. Senn, H-J., Manke, C.. Dieter. P., Tran-Thi. T-A.. Fitzke. E., Gerok. W.. and Gangliosides of normal and neoplastic human melanocytcs. Biochem. Bio- Decker, K. Ganglioside biosynthesis in rat liver: different distribution of phys. Res. Commun., 120: 500-504. 1984. ganglioside synthases in hepatocytes. Kupffer cells, and sinusoidal endothe- 5. Thampoe. I. J., Furukawa. K., Vcllve. E., and Lloyd. K. O. Sialyltransferase lial cells. Arch. Biochem. Biophys., 27Â«:161-167, 1990. levels and ganglioside expression in melanoma and other cultured human 27. TrincherÃ , M., Pirovano. B., and Ghidoni, R. Sub-Golgi distribution in rat cancer cells. Cancer Res.. 49: 6258-6264, 1989. liver of CMP-NeuAc GMi and CMP-NeuAc:GTib Â«2,8sialyltransferases and 6. Tsuchida, T.. Ravindrananth, M. H., Saxton. R. E., and Irie, R. F. Gangli comparison with the distribution of the other glycosyltransfcrase activities osides of human melanoma: Altered expression in viro and in vitro. Cancer involved in ganglioside biosynthesis. J. Biol. Chem., 265: 18242-18247, Res., 47: 1278-1281. 1987. 1990. 7. Wu, Z-L.. Schwartz, E., Seeger, R., and Ladisch. S. Expression of G|,2 28. Iber. H.. Van Echten. G., and Sandhoff. K. Substrate specificity of Â«2â€”3- ganglioside by untreated primary human ncuroblastomas. Cancer Res., 46: sialyltransferascs in ganglioside biosynthesis of rat liver Golgi. Eur. J. Bio 440-443, 1986. chem., 195: 115-120, 1991. 8. Sasaki, H., Momoi, T., Yamanaka. C., Yorifuji, T.. Kaji, M.. and Mikawa. H. 29. Cumar. F. A., Fishman. P. H., and Brady. R. O. Analogous reactions for the Changes in the ganglioside composition of human ncuroblastoma cells under biosynthesis of monosialo- and disialo-gangliosides in brain. J. Biol. Chem., different growth conditions. Int. J. Cancer, 47: 742-745, 1991. 246:5075-5084. 1971. 9. Houghton. A. N., Mintzer, D., Cordon Cardo, C., Welt, S., Fliegel. B.. 30. Nagata, Y., Yamashiro, S., Yodoi. J., Lloyd, K. O., Shiku. H.. and Furukawa, Vadhan, S.. Carswell, E.. Melamed, M. R., Oettgen. H. F., and Old, L. J. K. Expression cloning of 0-1,4-W-acetylgalactosaminyltransferase cDNAs Mouse monoclonal IgG3 antibody detecting GUI ganglioside: a phase I trial that determine the expression of GM2 and G[)2 gangliosides. J. Biol. Chem., in patients with malignant melanoma. Proc. Nati. Acad. Sci. USA. 82:1242- 267: 12082-12089, 1992. 1246, 1985. 31. Schulz. G., Cheresh, D. A., Varki, N. M., Yu, A., Staffileno, L. K., and 10. Cheung, N. K., Medof, M. E., and Munn, D. Immunothcrapy with GD2 Reisfeld, R. A. Detection of ganglioside ( t,,. in tumor tissues and sera of specific monoclonal antibodies. Prog. Clin. Biol. Res.. 277: 619-632, neuroblastoma patients. Cancer Res., 44: 5914-5920, 1984. 1988. 32. Melkerson-Watson. L. J., and Sweeley, C. C. Purification to apparent ho 11. Livingston, P. O., Ritter, G., Srivastava, P.. Padavan. M.. Calves, M. J., mogeneity by immunoaffinity chromatography and partial characterization Oettgen, H. F., el al. Characterization of IgG and IgM antibodies induced in of the GMJ ganglioside-forming enzyme. CMP-sialic acid:lactosylceramide melanoma patients by immunization with purified Sphingolipid Metabolism, pp. 193-213. New York: Pergamon Press. Biophys. Res. Commun.. 766: 387-393, 1990. 1966. 34. Svennerholm, L. Chromatographie separation of human brain gangliosides. 13. I Â¡simian.P. H., and Brady, S. Biosynthesis and function of gangliosides. J. Neurochcm.. 10: 613-623, 1963.

5731

Downloaded from cancerres.aacrjournals.org on September 24, 2021. © 1992 American Association for Cancer Research. Glycosylation Pathways in the Biosynthesis of Gangliosides in Melanoma and Neuroblastoma Cells: Relative Glycosyltransferase Levels Determine Ganglioside Patterns

Shutian Ruan and Kenneth O. Lloyd

Cancer Res 1992;52:5725-5731.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/52/20/5725

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/52/20/5725. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.