Indian Journal of Experimental Biology Vol. 38, April2000, pp. 301-312

Review Article

Immunology of gangliosides

Mepur H Ravindranath*, Alexandra M Gonzales, Kevin Nishimoto, Wai-Yin Tam, Daniel Soh & Donald L Morton

Laboratory of Glycoimmunotherapy, John Wayne Cancer Institute at Saint John's Health Center, 2200 Santa Monica Blvd, Santa Monica, CA 90404, USA

Phone: 310-449-5263; FAX: 310-449-5259; E-mail: ravi@ jwci.org.

This review discusses the immunology of gangliosides from the perspective of tumor, neuronal and general immunol­ogy. Antiganglioside antibodies in human sera are invariably IgM and are found in healthy individuals. Their titers decline with age. Persistent high titer of IgM is associated with several diseases, particularly neuropathies. Membrane-bound gan­gliosides are important tumor-associated antigens and targets for immune attack. Cells enriched with gangliosides can be used as cancer vaccines. Efficacy of these vaccines depends on the viability of whole cells, integrity of the cell membranes, adjuvants and topography of the tumor-associated antigens. The role of antiganglioside IgM is to eliminate the immunosup­pressive gangliosides shed from tissues during ageing, degeneration of neural and extraneural tissues, and tumor growth and necrosis . In addition, in vitro observations with human and murine monoclonal antibodies suggest that they are capable of complement dependent cytotoxicity and apoptosis.

Textbooks of basic or clinical immunology focus on protein and peptide antigens. Reference to carbohy­drate antigens is minimal and often restricted to bac­terial polysaccharides and blood group antigens. Car­bohydrate antigens constitute the glycocalyx of all cells (including T and B cells), and their cell-surface distribution makes them attractive targets for immune recognition. There is little information regarding the ability of carbohydrate antigens to elicit a direct T­cell response, although several peptide epitopes mimic carbohydrate epitopes with respect to antibody and .lectin recognition. The immune responses to car­bohydrate antigens differ from responses to peptides and proteins because carbohydrate antigens do not require T-cell help and hence are considered T-cell independent. The structure of anticarbohydrate IgM antibodies may differ from that of conventional pen­tameric IgM antibodies that have a J-chain . This re­view discusses the immunology of gangliosides from the perspective of tumor immunotherapy, although many of the tenets also extend to neuronal or general immunology.

Gangliosides: A family of carbohydrate antigens­Klenk1 coined the term ganglioside for glycosphino­lipids containing sialic acids. Gangliosides are a fam­ily of carbohydrate antigens that are overexpressed in nervous tissues and malignant extraneural tissues. They represent an important class of specific recog-

*Correspondent author: Phone: 31 0-449-5263; FAX : 310-449-5259; E-Mail : ravi @jwci.org.

nition molecules on the surface of all vertebrate cells . Gangliosides may not be evenly distributed on the cell surface, but rather exist in clusters. Structure, nomenclature and tissue distribution of the ganglia­sides have been reviewedv. Gangliosides are ampho­philic molecules with atomic mass units (AMU) ranging from 1300 to 2500. They have a hydrophilic head group of two or more sugars (glucose and/or galactose, neuraminic acid [sialic acid] with or with­out N-acetyl galactosamine or N-acetyl glucosamine), and a hydrophobic tail group of ceramide (sphingo­sine and a long chain fatty acid). Sugar chain elonga­tion of gangliosides involves a series of gene-specific glycosyltransferases4

'5

. The ganglioside head group oligosaccharide chain extends up to 2.5 nm from the membrane/water interface6

. Its orientation on the cell surface depends on the nature of the tai I groups, which are capable of rapid lateral and rotational dif­fusion. The chain length of fatty acids, the number of double bonds and hydroxylation of fatty acids m;:ty differ for a particular ganglioside in a cell. In aqueous media, the molecules aggregate and form irregular micelles. In ethanol, micelle formation requires a higher critical micellar concentration (CMC). There fore, ethanol suspension is employed to coat the mi­crometer plates in an enzyme -linked immunosorbent assay (ELISA) for measuring antiganglioside anti­bodies7. During drying in vacuo, CMC increases, causing the attachment of the tail group to the poly­styrene plate.

302 INDIAN J EXP BIOL, APRIL 2000

Specificity of antigenic determinants of ganglia­sides-The antigenic determinants of gangliosides, like those of other carbohydrate antigens, are chains of sugars (such as glucose, galactose, N-acetylneu­raminic acid, N-acetylgalactosamine, N-acetylgluco­samine and fucose). The specific epitope varies with the nature of these sugars and their glycosidic link­ages (a and ~ ; e.g., Gala1 ,3Gal; Gala1,4Gal ; Gal~ I ,3Gal ; Gal~ I ,4Gal; Gal~ I ,3GicNAc; Gal~ I , 4GicNAc; NeuAca2,3Gal ; NeuAca2,6Gal ; and NeuAca2,8NeuAc). Other families of carbohydrate antigens may share ganglioside antigenic epitopes. They include polysaccharide antigens (e.g., polysialic acids associated with neural cell adhesion molecule), proteoglycans, glycoproteins and glycolipid antigens. Polysi alic acids, an example of polysaccharide anti­gens consist of more than hundred sialic acids in fetal neuronal tissues. After birth and in adults, the number of sialic acids markedly decreases. However, after neoplastic transformation or dedifferentiation (neuro­blastoma), the number of sialic acids once again ap­proximates that of the fetal condition , befitting the definition of an oncofetal antigen . A murine mono­clonal antibody (Mab 3F8) developed against NeuAca2,8NeuAc residues of di sialoganglioside GD2 of neuroblastoma cell s, binds to polysia lyl resi­dues (NeuAca2,8NeuAc NeuAca2,8NeuAc). Glyco­protein antigens usually contain multiple sugar chai ns with different structures.

Based on the structure of their linkage to polypep­tide backbone, the glycoprotein antigens can be clas­sified into two groups-0-linked and N-linked. Sugar chains attached to the polypeptide by 0-glycos idic linkage from GalNAc to hydroxyl (-OH) of serine or threonine have been arbitrarily called muci n-type sugar chai ns. A relatively common feature of mucin­type sugars is a Gal~ I ,3GaiNAc-disaccharide core. Two murine monoclonal IgM antibodies, 2A3D2 and 2D II E2, developed with a ganglioside mixture pre­pared from human hepatocellular carcinoma cells as the immunogen, detected the ganglioside antigens, particularly ganglioside GM2, in carcinoma cells and bound to an 0-linked sialoglycoprotei n related to a rare blood group antigen called Cad8

. Similarly, a murine antibody (KM696) spec ific for ganglios ide GM2 (GaiNAc~ I ,4 [NeuAca2,3]Gal~ I , 4Gic~ I , lceramide) may recognize a Cad-like blood group antigen (Hanai N, personal communication). Anti­bodies to carbohydrate determinant (s ialyl Lewis•) of

a ganglioside ( 121 SLE [IgM], Neomarkers, Union City , CA; 1116-NS-19-9 [IgG I], Signet Laboratories, Dedham, MA; M8073022 [IgG I], Fitzgerald, Con­cord, MA; ZY -C09 [IgG I], Zymed, San Francisco, CA; MED-CLA 143, Accurate Chemical and Scien­tific Corporation, Westbury, NY; NS 19-9, Interna­tional CIS, Cedex, France) are directed against the sugar residues of mucin-type glycoproteins belonging to MUC family , such as MUC-1. In the second major class of glycoprotein antigens (N-linked), the sugar chains are attached to polypeptide by a N-glycosidic linkage from GlcNAc to amide nitrogen of asparagine and have been called asparagine-linked glycoproteins. Structural variations in 0 -linked and N-l inked hetero­glycan moieties have been reviewed9

.

Glycolipid antigens include several families of neutral glycolipids, fucolipids, and sulfolipids . Structure and distribution of some of the fucolipid antigens found in normal and malignant human tis­sues have been reviewed3

. Acidic glycol ipids also include those containing uronic acid and sulphate groups. Sulphated glycolipids or su lphatides are found in several normal and malignant tissues 10

. Gly­colipids containing neutral and acid heteroglycans are associated with several genetic abnormalities3

. The ganglia-compounds of glycol ipids are most common and are abundant in most normal and mal ignant hu­man tissues. Specific immune recognit ion of ganglio­side epitopes requires absence of cross-reactivity with other acidic or neutral glycolipid antigens.

Ganglioside epitopes--There is remarkable diver­si ty in ganglioside antigenic ep itopes. An epitope is the part of an antigen forming contacts with an anti­bod/ 1• The complementary combining site (CCS) of an antibody is the paratope. The upper limit of epi­tope size is determined by the variable region CCS of the paratope. Measuring the displacing activity of different oligosaccharides shows that maximum physical capacity of CCS is six sugar residues for anticarbohydrate antibodies and four amino acids for antipeptide anti bodies. X-ray crystallographic studies of oligosaccharide-antibody complexes have con­firmed these estimates. The contact areas invol ve 255 A of sugar residues and 304 A of paratope. Fifteen amino acids of the antibody establish 90 van der Waal forces and nine hydrogen bonds 12

. Antigen-antibody interactions may also in volve salt bridges . The struc­ture of some antibody-antigen complexes suggests rigid binding of both. The hydrophilic head group of

RAVINDRANATH et al.: IMMUNOLOGY OF GANGL! OS IDES 303

gangliosides contains antigenic determinant of CCS of paratope.

Distribution of gangliosides in human tissues­Gangliosides are present at low levels in extraneural tissues and in abundance in nervous tissues 13

·14

• GM3 constitutes more than 90% of the total gangliosides in normal cells such as melanocytes 15

, while the level of other gangliosides (e.g., GM2, GD3, GD2) ranges from 1 to 10% of total gangliosides . Structures of some of the most common gangliosides in normal and malignant human tissues are presented in Table I. Figure 1 depicts the changing pattern of gangliosides during progression of melanoma, based on a detailed study of the gangliosides of tumor biopsies and cell 1. 16- 18 I d . f G INA . . mes . ncrease expressiOn o a c-contammg gangliosides such as GM2 and GD2 in cells cultured from tissue biopsy specimens suggests enhanced ac­tivity of ~I ,4 GalNAc-transferase, an enzyme in­volved in the final step in synthesis of these ganglia-

sides 16• We have recently reviewed the overexpres­

sion of gangliosides in other human cancers and some of the common ganglioside antigens shared by solid tumors 19

'20

.

Why T cells do not recognize ganglioside anti­gens-T cell recognition requires antigen to be proc­essed and presented by an antigen-presenting cell, such as macrophage or a dendritic cell. Both proc­essing and presentation of peptide epitopes to major histocompatibility complex (MHC) can be affected by the presence of sugar residues in epitopes 21

• In addi­tion, inability of oligosaccharides to bind into groove of MHC molecules makes gangliosides unfit for di­rect and exclusive recognition by conventional T cells22

·23

. However, human T cells (a/~) may be stimulated by nonpeptide antigens presented by the nonclassical MHC molecule (CDl), which is distantly related to MHC class I molecules24

. The human CD I b isotype can present lipoglycans, namely lipoarabino-

Table 1-The structure of gangliosides commonly found in normal human tissues and cancer tissues

Glycolipid Structure

GM I GM2

GM3

GD3

GD2

GD 1h

GT1h

Galp I ,3GaiNAcp I ,4(a2,3NeuAc)Gaip I , I Cer GalNAcp I ,4(NeuAca2,3)Gaip I , I Cer NeuAca2,3Gaip I , I Cer NeuAca2,8NeuAca2,3Galp I , I Cer

GaiNAcp I ,4(NeuAca2,8NeuAca2,3)Galp I , I Cer Galp I ,3GalNAcp I ,4(NeuAca2,8NeuAca2,3)Galp I , I Cer NeuAca2,3Galp I ,3GaiNAcp I ,4(NeuAca2,8NeuAca2,3)Galp I, I Cer

Melanocyte Differences observed in tumor biopsies

19:1 1:1 1:2<

GM3 D GM2 (@ill

GD2 • GD3 •

A rare melanoma

Atomic Mass Unit

1547 1385 1236 1545 1694 1838 2144

1:19

Changes observed in culture Fig. !-Profi le of gangliosides in human tumor biopsy pecimens and cell lines. Pie charts depict the changes and evolution of ganglia­sides during neopl astic transformati on of normal melanocytes during radial and vertical migration and metastas is. The ratio of GM3:GD3 is also indicated under each pie chart. Culture conditions favour overexpression of GaiNAc-containing gangliosides, probably due to overexpression of Gal Ac-transferase. [Reprinted from Ravindranath et a/., Prog Brain Res, I 0 I ( 1994) 251]

304 INDIAN J EXP BIOL, APRIL 2000

mannan and mycolic acid derived from mycobacterial cell walls, to T cells (a/~)25 '26 . Nature of interaction between lipoglycan and CD I b is not clear. These findings have important implications for immune rec­ognition of membrane-bound gangliosides.

T-cell independent antibody response to ganglia­sides-Protein or peptide antigens are processed by intracellular enzymes and MHC molecules, which transport antigen fragments to the cell surface and present them to T cells as well as B cells, to generate primary and secondary antibody responses . MHC­restriction of antigen presentation is characteristic of peptjde antigens . The antibody responses to carbohy­drate antigens follow a different pattern . Carbohy­drate antigens are not presented in the context of MHC molecules; they stimulate primary but not sec­ondary B-cell responses, and do not require T-cell help. Hence they are considered T-independent anti­gens. Animal studies have shown that glyco-antigens can trigger B cells to produce antibody (IgM) in T­cell deficient mice, indicating that they are T-cell in­dependent27 .

There are two classes of T-independent antigens. T-independent-1 antigens at high concentrations stimulate polyclonal proliferation of B cells, and at low doses stimulate division of B cell s that secrete heteroglycan-specific antibodies. Response to these glycan antigens occurs in cultures rigorously depleted of T cells, but can be modified by other cell types such as monocytes or natural killer cells. Secondary

1000

a:: 800 lLJ t-

600 t-ct u 400 en lLJ c 200 en

0 0

... ;. :::: · .. ·: ,

.. . ~ .;_ . . ,;·, . . ,.

. ' ·' .. ..

200 400 600 800 1000

FORWARD SCATTER

exposure to these antigens does not result in acceler­ated kinetics and isotype switching typical of memory responses mediated by T cells. The second class ofT­independent antigens is T-independent-2. These glyco-antigens stimulate antigen-specific responses in athymic nude mice, but do not stimulate polyclonal proliferation at high do es. In vitro responses may require interferon-y, which can also be provided by natural killer cells. Interferon-y may induce these T­independent-2 antigen-stimulated B cells to switch isotype production to lgG3 in animal models. Gan­glioside antigens also fail to induce a memory re­sponse in humans, consequently antiganglioside anti­bodies are invariably IgM in human, although anti­ganglioside IgG antibodies are found in mice and rabbits .

Antiganglioside lgM may differ from •conventional /gM-Structure of anticarbohydrate lgM antibodies in patient sera may be different from that of conven­tional pentameric IgM. Recent re.search indicates that IgM also can exist as a hexamer, heptamer or other polymer28. Furthermore, unlike conventional IgM, polymeric lgMs may not have J-chain29·30. The poly­meric IgM without J-chain seems to fix complement 20-fold more efficiently than conventional pentameric IgM31

• Persistent IgM antibodies in patients with motor-neuron diseases are directed against carbohy­drate residues of glycolipids32. Factors contributing to production of polymeric a J-chainless IgM are far from clear. However, evidence suggests that IgM an-

1000

> 103 800

102 600

400 10'

200

10 1 102 103 10~

CD19 POSIT iVITY

Fig. 2-Two subpopulations of B lymphocytes identified based on expression of pan-T cell antigen CDS . CDI 9 is a common marker for B cells. These B cell s may be CDS positive or negative. Flow cytometric profile of Ficoll-separated PBL obtained from normal individu­als, stained with nuorescein-conjugated ant i-human Leu 12 (CDI9+) (lgGI) and phycoerythrin-conjugated anti-human Leu I (CD5+) (lgG2a) monoclonal antibodies, and analyzed on a dual laser FACScan now cytometer (Becton-Di ckinson) equipped with a 15 mW 488 nm air-cooled argon ion laser-(A) A representative forward-angle light scatter (FSC) and side-angle light scatter (SSC) dot plot gate generated from PBL The gate is restricted to a distinct population of small lymphocytes (FSC 280-480). Large mononuclear cells com­prising lymphoblasts and monocytes (FSC>520) were excluded. Red blood cells and debris were also excl uded by the FSC threshold setting; (B) FACS profile of CD5+/CD 19+ (B I) and CDS-/CD 19+(82) lymphocytes (Total: I 0 ,000 events) [Modified from Ravindranath eta! .. Cell Mol Life Sci, 53 ( 1997) 750]

RAVINDRANATH et al. : IMMUNOLOGY OF GANGLIOSIDES 305

Table 2-Literature correlating anti -ganglioside antibody levels in di ffe rent kinds of human di seases

Anti-Ganglios ide lsotype Di sease Authors Source

Neuropathies

GMI lgM/IgG Guillain-Barre syndrome Mata et al., J Clin Apheresis 1998; 13: 155 GMI Miller-Fisher/B ickerstaff s encephaliti s Fargas et al.. Pediatr Neurol 1998; 18:73

GMI lgM/IgG Peri pheral nervous system disorders Utex-von Allmen Eur euro l 1998; 39: I 03 et al.

GMI Demyelinating polyneuropathy Bech e/ a/., Ann Neurol 1998; 43:72

GMI lgM Multi foca l motor neuropathy Pestonk et a/. eurology 1997: 49: 1289

GMI lgM Guillain-Barre syndrome Bech e/ a/. J Neuroimmunol 1997 : 72:59

GMI Acute immune polyneuropathies Nevo et a/. J In fec t Di s 1997; 176: s l54

GMI lgM Mul tiple sc leros is Endo eta /. Ann NY Acad Sci 1984; 436:213

GMI· Rheumatologic disorders Zeballos et al. J Clin Lab Anal 1994; 378 :84

GMI Infl ammatory demyelinating neuropathy Tsukaguchi et a/. Rinsho Shinkeigaku 1997; 37:228

GMI lgA Di arrhoea with Guillain-Barre syndrome Koga eta/. Rinsho Shinkeigaku 1997; 37:848

GMI Alzhei mer's disease Schott et a/. Psychi atry Res 1996; 3 1 :251

GMI Multi foca l motor neuropathies Younes-Chennoufi et a/. Rev Neurol (Pari s) 1996; 152:375

GMI lgM!IgG Rheum. arth rit is+peripheral neuropathy Salih et a/. GR J Rheumatol 1996;35:725

GMI Multi foca l motor neuropathies Pouget et a/. Rev eurol 1996: 152:370

GMI lgG Chronic asymmetric LMN syndromes Kornberg et a/. Ann Neurol 1994; 36:455

GMI lgM!IgG Cryptogenic parti al epilepsies Barolomie et a/. Epilelpsia 1996;37 :922

GMI lgG Acute motor axonal neuropathy Hafer-Macko era/. Ann Neurol 1996;40:635

GMI lgG Unilateral atrophy of the distal upper limb Kaji , Rinsho Shinkeigaku 1994:34:1272

GMI lgM Inflammatory demyelinating neuropathy Bech et a/. Clin Chem 1994;40: 133 1

GMI Fibromyalgia Klein et a/. Clin lnvestig 1994;72:541

GM I Neuromuscular diseases Adams eta/. J euroimmunol 199 1; 140:223

GMI lgM Multifoca l motor neuropathy Bab et a/. J Neuroimmunol 1989;25 : 1989

GMI lgG/lgM Multi foca l motor neuropathy Pestonk er a/. Ann Neurol 1988; 24: ,73

GMI lgM!IgG Epilepsy Karpiak er a/. J neuroi mmunol 1982;3: 15

GMl b lgG GB syndrome/ Mycopl asma pneumoni ae Kitazawa era/. J Neurol Sci 1998 ; 156:99

GDlb Motor neuron disease (MND) Fi shmen era/. Arch Neurol 48:11 88

GM2 lgM GB syndrome/cytomegalovirus in fec tion Khalili-Shirazi el a/. J euro l Neurosurg Psychi atry 1999: 66:376

GM 3 Atherosclerotic patients Golovanova e/ a/. Clin Chim Acta 1998;272: 197

GDl a lgG Acute motor axonal neuropathy Ho eta/. Ann Neurol 1999:45:1 68

GDl a lgG Progressive multi ple sc lerosis Acarin et a/. Acta eurol Scand 1996;93:99

GDl a lgM!IgG Peripheral motor syndromes Carpo et a/. Ann Neurol 1996;39:539

GDl a Guill ain-Barre syndrome Kusunoki et al. Ann Neurol 1994; 35:570

GDl a lgG Guillain-Barre syndrome Yuki et a/. Muscle Nerve 1992; 15:889

GTlb Guillain-Barre syndrome Hitoshi er a/. J Neuroimmunol 1996; 66:95

GQib Guillain-Barre & Miller-Fisher syndrome Goffette era/. Acta Nelll·ol Belg 199R ; 98 :322

GQib lgG Miller-Fi sher syndrome Chida et a/. J Neurol Neurosurg Psychi atry 1998; 64:399

GQl b lgG Miller-Fi sher syndrome/ C. jejuni infection Ohtsuka et al. Ophtalmology 1998; I 05: 1281

GQib lgM Dysimmune neuropathies Carpo et a/. J Neurol Sci 1998; 155: 186

GQib lgM Sensory atax ia Tagawa er a/. J Neuroimmunol 1997;75:1 96

GQlb/GTl a Miller-Fisher syndrome Neisser et a/. In fec t Immunity, 1997; 65:4038

GQlb/GTl a lgG Oropharyngeal pal sy O'Leary et a/. J Neruol Neurosurg Psychiatry 1996 ; 6 1 :649

GMI/GDl b Myeloradiculoneuropathy Niwa er a/. Rinsho Shinkeigaku 1997; 37:26

GMIIGDlb Motor neuron disease Shy e/ al. Ann Neurol 1989; 25:511

(Con/d.)

306 INDIAN J EXP BIOL, APRIL 2000

Tabl e 2-Literature correlating anti-ganglioside antibody levels in diffe rent kinds of human di scases-Contd.

Anti-Ganglioside lsotype Di sease Authors Source

Neuropathies

GMI/GDlb lgM Lower motor neuron disease Thomas et a/. J Neuroimmunol 1989; 23 :1 67

GMI/GDl b Motor neuron disease Nardelli et a/: Ann Neura l 1982; 23: 174

GMI/GDlb lgM Neuromuscular disease/ paraproteinemias Ito et a/. Ann NY Acad Sc i 1988; 540:258

GMI /GDlb lgM Motor neuron diseases Latov et a/. Neurology 1988; 38:763

GDl b/GTlb/GDl a lgM/l gG Guillain-Barre syndrome llyas et a/. Ann Neura l 1988; 23:440

GDlb/AGMI lgM Biclonal garnrnopathy llyas eta/. J Neurochern 1988; 51 :851

GA I/GMI lgA/lgG Chronic Chagas' Disease patients Avila et a/. Am J Trop Med Hyg 1998; 58:338

AGMI/GMI lgG GB syndrome/ C. jejuni infection Ogino eta/. J Neuroimmu no l 1995; 58:77

AsG MI/GMI/LMI Proxi mal Lower motor neuron syndrome Yuki eta/. Rinsho Shinkeigaku 1993; 33:89 1

GDlb/GD3 IgG Acute atax ic neuropathy Willison et a/. Neurology 1994; 44:2395

Ganglios ides lgM Paraproteinemi a ll yas et a/. Proc Nat! Acad Ac i USA, 1984: 8 1:1 225

Ganglios ides lnOamrnatory neuropathies Quarles et a/. Chern Phys Lipids 1986; 42:235

GM 2/GMlb/GDl a lgM Paraproteinemi a and peripheral neuropathy llyas et a/. J Bioi Chern 1988: 263:4369

GMI/GDl a lgM Amyotrophic lateral sclerosis Pes tronk et a/. Ann Neural 1989; 25:98

GTlb/GTl a lgG Myasthenia gravis & Eaton-Lambert Synd. Kusunoki et a/. J Neur Sci 1988; 87:61

GTlb/GTl a Paraproteinemi a! polyradiculoneuropathy Baba eta /. Acta Neura l Scand 1985; 72: 2 18

GTl a!GMib IgG Guill ain-Barre syndrome Koga eta/. J Neura l Neurosurg Psychiatry 1999;66: 5 13

Cancers

GM 2 lgM/lgG Human melanoma Livings ton et a/. Semin Oneal 1998; 25:636

GM2 lgM Human melanoma Li vings ton et a/. Ca ncer lmrnunol lmmnother 1997; 43 :324

GM2 lgM Human melanoma Nishimaka et a/. Can Res 1996; 53:5666

GM2 lgM Human melanoma Livingston eta/. J Clin Oneal 1994; 12: I 036

GM2 lgM/lgG Human melanoma Livingston et a/. Cancer Res 1989; 49:7045

GM3 lgM Non-Hodgkin lymphomas Roelckd eta/. Scand J lmmunol 1978; 8:1 79

GD2 lgM Human melanoma Hoon er al. Cancer Res 1993: 53:5244

GD2 lgM Human melanoma lrie era/. Proc Nalt Acad Aci USA 1986: 83 :8694

GD2 lgM/lgG Human melanoma Tai et al. , J Nat! Cancer lnst 1984; 73:627

GD2 lgM Human melanoma Katana eta/. J Clin Oneal 1984; 15:119

GD3 lgM Human melanoma Yao eta/. Clin Cancer Res 1999; 5:77

GD3 Human melanoma Ritter et a/. lnt J Cancer 1995 : 62:668

GM I/GDl b Gastric cancer Konstandoulakis et a/. H ybridoma 1998; 17:43 1

GM2/GD2/GD3/GM 3 lgG/IgM Human melanoma Mortan et a /. Bio Func Gang 1994; 19:25 1

GD2/GD3 lgG Human melanoma Sela et al. J Nat! Cancer lnst 1989; 8 1: 1489

Infec tions

GMI Polyneuropathy/C. jejuni/H .pylori infec tion Neva Y. et a/. J Infect Di s 1997; 176: s l54

GM2 lgM HIV infection Wu X. et a/. J lmmunol 1999; 162:533

GM2 lgM Acute cytomegalovirus infec tion Yuki N. er a/. J Neura l Sci 1998; 154: 14

GM2 lgM GB syndrome/ cytomegalovirus infec tion lrie S. er a!. J Neuroirnmunol 1996: 68: 19

GM 2 Amoebiasis Sorice M. er a/. Parasote lmmunol 1996; 18:1 33

Asialo-GMI Vacc ini a virus in fec tion Karupiah G. er a/. lmrnunol Cell Cio 1990; 68:343

Asialo-GMI Rotavirus infect ion Willoughby et a/. J Viral 1990 64, 4830

GM3 HIV in fec tion Soric er a/. J Acquir Immu ne Defi c Syndr Hum Retroviral 1996; 12:1 22

(Colli d.

RAVINDRANATH eta/. : IMMUNOLOGY OF GANGLIOSIDES 307

Table 2-Li terature correlating anti-ganglioside antibody levels in different kinds of human diseases-Con/d.

Anti-Ganglioside lsotype Disease Authors Source

Neuropathies

GM 3 AIDS patients Griggi et al.

GDi a IgG/IgA Acute-motor-axonal-neuropathy/C. enteritis Lugaresi et a/.

Scand J lmmunol 1994; 40:77

J Neuro1 Sci 1997; 147:193

Asialo-GMI lgG Homosex ual men with AIDS Witkin et al.

GMI/GM2/GTib Mycoplasma pneumoniae mcningoencepha- Komatsu eta/.

Clin Exp lmmunol 1983; 54:346

Pediatr Neurol 19 liti s and cerebelliti s

GMI/GDib/AsM IG lgM Neuroborreliosis (Lyme disease)

Other Diseases

GT3 Diabetes mellitus

GM 2 lgG Diabetes

GM2 Type I diabetes onset

GM 3 IgM/IgG Abortion

GMI/GT1b Stroke

Garcia eta/.

Gillard et a /.

Mi sasi eta/.

Dotta eta/.

Ozaki era/.

Hsieh era/.

J Neurol Sci 1993; 117:206

J lmmunol 1989; 142:3826

Diabetes Metab Rev 1997; 13 :163

J Autoimmun 1997; 10:585

Am J Reprod lmmunol 1995;33:234 Kao Hsiung I Hsueh Ko Hsueh Tsa Chih 1998; 14:68

3.0 n.c -Reson:imi n.c -with LS-SI ~ 1:50

- 1:100 mssa 1 :2oo

GMJ

2.5 GM2 , .. ,

E 2.0 c:: GOO

0 GDla 0') 1.5

~ GDZ

<( GDTh

1.0 Sit. Ml4 M24 Sit. Ml4 M24 totll totU totll totll

0.5

0.0 aGM1 GM1 GD1 a GT1 b GD1 b GM3 GM2 GD3 GD2

ELISA-with LSS-81

Fig. 3-Monospecificity of human monoclonal anti-GM2 lgM antibody (L55-81 ) is documented by ELISA and by TLC immunostain­ing. The insert shows gangli oside profile of human melanoma cell lines M 14 and M24 as well as immunostai ning of thin layer chroma­togram with monoclonal antibody. ELISA documents that monoclonal antibody does not recogni ze other gangliosides [Modified from Nishinaka, Ravindranath & lrie, Cancer Res, 56 (1996) 5666]

tibodies directed aga inst carbohydrate epitopes of glycolipids may be secreted by a separate subclass of B cells, which also express aT-cell marker: (CD5)32

.

Two subclasses .of B cells (CD 19+/CDS+; CD 19+/CDS-) and the ir relationship are shown in Figure 2.

Antiganglioside lgM antibodies-the natural anti­bodies-B cells, in general, and those specifically

producing antiganglioside IgM antibodies are found in healthy individual s. IgM antibodies appear at least five days after birth , probably due to exposure to ex­ogenous antigens including mother's milk. During the first 40 days after her child' s birth, a nursing mother's milk contains GD3 , a ganglioside highly prevalent in human melanoma; after that, the prevalent gang I io­side in breast milk changes to GM3, suggesting that a newborn is orally and repeatedly immunized with

308 INDIAN 1 EXP BIOL, APRIL 2000

GD3 (ref. 33). In the absence of endoceramidases in the human gut system, the smaller molecul ar size of gangli os ides (AMU < 2500; Table I) is absorbed and probably exposed directly to gut-associa ted lymphoid organs. Our observations show that antibodies to all ganglios ides are fo und in hea lthy individuals but their levels dec line with age34

. A low level of antiganglio­side IgM antibodies in normal sera may be of phys iological signi ficance with respect to neutral iz ing and clearing from the circulati on pathogens express­ing gangli osides or gangliosides shed during tissue destruction and growth . This physiologica l ro le would explain the hi gher level of anti gangli os ide IgM in health y young individuals (age < 50) than in older indi viduals (age> 50). It is not surpri sing that dec line in titer of antigangli oside JgM is assoc iated with older age, since older individuals are more susceptible to infec ti on and cancer. Nutritional strategies to aug­ment anti gangli os ide IgM in older ind iv iduals deserve attent ion. Ri ch sources of gangli osides include milk, buttermilk and yogurt (curd). Ganglios ide GD3 and its deri vatives such as 0 -Ac GD3 are purified from · bovine buttermil k35

.

Therefore, the emergence of anti ganglios ide IgM is brought about by selec ting and stimulati ng natural autoant ibody produc ing cell s that are normall y mini ­mall y activated or anergic36

. Presence of natural anti ­ganglios ide IgM in humans suggests that gangli osides are immunogenic in man. Immunogenicity of ganglia­sides is con firmed by immortali zing B ce ll s with Ep­stein-Barr virus and develop ing human monoclonal antibodies37

. Monospec ificity of human monoc lonal ant ibodies against spec ific carbohydrate ep itopes is confirmed using sensiti ve ELISA teclmiques38

. Figure 3 illustrates monospecificity of one such antibody puri fied from culture supernatan ts of EBY­transformed lymphoblasts obtained from a breast can­cer patient. These human monoc lonal JgM antibodies directed against cancer-assoc iated ganglios ides have immense therapeutic potential.

Antiganglioside antibodies as surrogate markers of diseases-We have extensively surveyed the litera­tu re on the prevalence of antigangli oside IgM anti­bodies in human di seases. Table 2 summarizes the associati on between a particular spec ies of antigan­gli oside lgM antibody and human di sease. Some of the anti ganglios idc antibodies such as anti -GM I and anti-GQ I b IgM are found in high titers in patients wi th neurologica l di sorders such as polyneuropathy, amyolaterosclerosis and Guilli an-Barre syndrome.

These antibodies might cause degeneration of myelin and other neuronal components. They may also play a role in eliminating gangli os ides shed from the degen­erat ing ti ssues, since ganglios ides are known to be . . Immunosuppress ive.

Antiganglioside lgM may eiiminate circulating gangliosides-Gangli os ides shed from a growing tu­mor can suppress express ion of CD4 on human helper T lymphocytes, express ion of MHC class I and class II genes, and mitogen-induced lymphoblas tic trans-f . 19-4 1 Th I . d T ormation· . ey can a so 111 uce -suppressor acti vity. Circul ating gangli os ides can reduce the nu mber of macrophages and CD4+ and CD8+ T lym­phocytes, and they can inhi bit interleukin-2 depend­ent lymphocyte proliferation by prevent ing the inter­acti on of thi s cytokine with its ce ll-surface recep-

42 41 0 b . . tors · · . ur recent o servat1 on on apparent mverse correlati on between serum level of total gangli os ides (> 25 mg/dl ) and survival in pancreatic carcinoma pati ents may be related to increased tumor burden and may refl ect the immunosuppressive effect of circu­lat ing ganglios ides44

. If gangli os ide-i nduced immuno­suppress ion is linked with poor prognos is, then clear­ance of tumor-deri ved gangli os ides from the circu la­tion may be of clinica l benefit fo r patients with unre­sectable disease. Our resul ts indicate that antigangli o­side antibod ies should be augmented in patients with a large tu mor burden. One of the major functions of a cli nica ll y effec ti ve cancer vacc ine wi ll be to augment ant icarbohydrate antibodies so that they can remove immuno-uppressive gangli osides from the blood.

Capability of antiganglioside lgM for complement dependent cytotoxicity (CDC) and apoptosis-Tumor regression was documented afte r injecting an anti ­GM2 human monoclonal antibody (L-55) into mela­noma lesions. Hi stologica l secti on displayed tumor degeneration, fi bros is, free melanin and some degree of lymphocyte and macrophage in fi ltrati on45

. An in vitro chromium release assay was performed to test the CDC capabiliti es of the same monoclonal anti ­body against cultured melanoma cell s in the presence of human complement36

. When melanoma cell s (M 14) were exposed to antibody in the presence of human complement, significant 51Cr release acti vity was observed in the range of 0.2-1 20 11g/ml antibody. In the absence of complement, the human anti-GM2 IgM did not show any CDC acti vity even at I 20 )..lg/ml. An irrelevant human IgM monoclonal anti ­body had no CDC activity against M 14 ce lls at the same concentration of the antibody. A humanized

RAVINDRANATH era/.: IMMUNOLOGY OF GANGLIOSIDES 309

murine monoclonal anti-GM2 l aM caused apoptosi s 46 . 0

of ~umor cells , sugges t1ng that antiganglioside IgM ant1bod1es are capable of human complement­mediated cytotoxicity as well as apoptotic activity.

Augmentation of immuncgenicity of gangliosides­Antibodies to carbohydrate antigens can be raised in vivo using appropriate carriers and adj uvants. Anti­genic determinants may also be influenced by other colloidal molecules associated with carbohydrate an­tigens such as peptides (proteoglycan), proteins (gly­coprotein) and sphingolipids (glycolipid) . Immuno­genicity of carbohydrate antigens may be decreased or increased based on the assoc iated moiety such as proteins and lipids. Immunogenicity of carbohydrates can be increased by chemically linking ol igosaccha­rides to peptides and proteins . Antibody response is augmented by admixing gangliosides with fore ign carrier proteins such as pig serum47'48

, serum albu-. 4'>.so h I I . s1 f . mm , uman eryt 1rocyte g ycoprotem , ore1gn

h s2 . f . I eryt rocytes· or a m1xture o menmgococca outer membrane prote ins, cationized bovine serum albumin, multiple antigenic peptides, polylysine and keyhole limpet hemocyanin5-' . Use of bacterial carriers such as Salmonella minnesota or Mycobacterium bovis aug­mented the antibody response against gangliosides in human and in murine studies5~. Ganglioside antigens presented in the contex t of an intact membrane elicit a better immune re ponse than do so luble anti­gens55'57. Immunogenic ity of carbohyd rate-protei n­adjuvant complexes and its c linical relevance in can-

. h b . d56 cer pat1ents as een rev1ewe .

Membrane-hound gangliosides elicit a better im­mune response than do soluble gangliosides-Immu ­nogenicity of gangliosides is assessed by vaccinating animal models with cells rich in gangliosides , lipo­somes containing gangli osides, or purified ganglia­sides conjugated to adjuvants. A nontoxic adjuvant called monophosphoryl lipid A (MPL) was conju­gated with viable B 16 mouse melanoma cells and used to vaccinate C57BL/6j and BALB/c mice55·57

We monitored the conjugation of MPL using anti­MPL monoclonal antibody (Mab 8A I : Centacor). The mice were immunized with one of these immuno­gens-( I) B 16 cells without adj uvant ; (2) MPL­incorporated B 16 cell s; (3) MPL alone; (4) GM3 alone; (5) GM3-admixed with MPL; and (6) MPL and GM3 incorporated to single-unit liposomes. The level of antibodies reacting to GM3 was measured by ELISA, since the major ganglioside of B 16 melanoma cells is GM3 . Prevaccine and postvaccine sera were

compared for antibody response. Antibodies to aan­gliosides were found in low levels in normal 1~ce . MPL conjugated to natural (viable whole cell) or arti­ficial (single-unit liposome) aanolioside-containino

0 0 0

membranes augmented antiganglioside antibody re-sponse I 000-fold more than did multiple large doses of gangliosides

58"59

, gangliosides linked to whole bacteria Salmonella and Mycobacteria54

·60 or ganglia­

sides mixed with complete Freund's adjuvant51 or Freund's adjuvant together with pertuss is vaccine52 or 3000 kDa KLH

53'61

. In our study, a higher antigan­glioside antibody response was obtained with 2 im­munizations than with 4-24 immunizations57. Purified gangliosides or gangliosides coupled to MPL in mi­cellar form did not have this effect, even when the amount of purified gangliosides (expressed as lipid­bound sialic acids) was more than 2.5 to I 0-fold greater than that found on cell membranes. A very high antiganglioside IgM response was induced by only 2 low doses of MPL-incorporated viable whole cells (with 600 ng of ganglioside) . This response is far superior to that observed after attaching MPL to ganglioside-containing liposomes, particularly con­sidering the 27-fold hi gher level of gangl iosides in liposomes. Our experiment documents a strategy to augment immunogenicity of the membrane-bound gangliosides of viable whole cells.

We have further confirmed the superiority of the immune response to membrane-bound gangliosides using a guinea pi g modd'2. In this system, we vacci­nated the anima ls with whole melanoma (hu man) cells or soluble lysates produced from the same cells and administered with or wi thout Baci lle Calmette­Guerin (BCG). Sera from vaccinated animals were tested for antibody response agai nst purified ganglio­s ide antigens. Whole cell s admixed with BCG elicited a better an tiganglioside IgM response than did soluble lysate vaccine admixed with BCG, indicating th:Jt the whole cell vaccine may be superior to soluble or ly­sate vaccine because it facilitates better immune rec­ognition of cell-surface carbohydrate an tigens. Addi­tion of BCG augmented the antibody response, con­firming that an exogenous adjuvant may immunopo­tentiate antigens better in the presence of an intact cell membrane .

/mmwzogenicity of ganglioside GD3-Melanoma cells express GD3 in abundance. We have docu­mented that during progress ion of human melanoma, GM3:GD3 ratio changes in favour of GD3

17• GD3

may be uniformly distributed on the cell surface or

310 INDIAN J EXP BIOL, APR IL 2000

anti-asialoGM1 • 1\~anorm patients[AJCC St Ill & IV] [n =44]

lml!!!!J Nooml and Healthy individuals (n =93)

p<0.001

p<0.001

anti-GD1b p<0.01

0 10 20 30 40 50 Incidence of darinant species of antilxx:ly (in%)

Fig. 4-Proli le of serum antiganglioside lgM antibod ies in healthy individuals differs from that in patients with stage Ill melanoma. No lgG antibodies to gangliosides were observed in healthy individuals or cancer patients. The most common antiganglioside antibody is anti-GO I b lgM in healthy individuals and anti -GD3 lgM in melanoma patients. This difference may renee! overexpression of GD3 by human melanoma; shedding of GD3 from the tumor cell into the circu lation would augment lgM response. Antiganglioside lgM antib~dy profile of cancer patients should vary depending on the most prevalent ganglioside of tumor type.

occur as dense aggregates. Our studies on antigan­glioside antibody response after immunization with BCG-admixed viab le whole cells c learly document that antibodies are elicited not on ly against GM2, G02 and 0-acetyl G03, but also against G03 (Ref. 63) .. However, Livingston's group60 has reported that none of their immunization protocols favoured pro­duction of anti-G03 antibodies in recip ients of puri­fied vaccines . This group has used a variety of immu­nopotentiating agents. Noteworthy are KLH and QS-21 . Immunization of melanoma patients with G03-KLH conjugate vaccine mixed with QS-21 failed to e licit anti-G03 antibodies. However, our studies indi­cate that immunogenicity of G03 depends on its den­sity and functional expression on a viable cell mem­brane. In addition, an appropriate adjuvant is needed to present ganglioside to antibody-producing cells. The adjuvant is critical to facilitate recognition of poorly immunogenic gangliosides.

Immun e recognition vs enzyme recognition of disialyl gangliosides-Presentation of antigen in the context of an intact membrane may be the key to effi­cient immunogenicity. Some of our enzymatic studies suggest that orientation of terminal dis ialyl residues of G03 could be important for immune recognition. We have done a series of experiments to cleave a2,8 ketosidic linkages of the terminal sialic acids with neuraminidase, since a2,8 ketosidic linkages in gly-

col ipids are resistant to periodic acid ox idat ion64. We

have found that a2,8 ketosidic linkage of G03 could not be dissociated by prolonged treatment in resorci­nol-HCI or by weak or strong acid hydrolysis . Inter­estingly, a2,8 ketosidic linkages of GO I b were disso­ciated by weak ( <0.1 N) acid hydrolysis. When we treated disialyl residues of G03 , G02 and GO I b with neuraminidase in the presence of detergents and cal­cium, we found that c leavage of 2,8 glycosidic link­age was most efficient in GO I b and progressively less so in G02 and G03.

Susceptibility of a2,8 ketosidic linkages of G02 and GO I b to Vibrio cholerae neuraminidase in the presence of calcium and detergent is due to exposure of dis ialy l residues for recognition by the enzyme. In the presence of calcium and detergents, the enzyme recog­nizes NeuAc a2,8NeuAc residues better in GO I b than in G02. In the absence of these two agents, recognition by the enzyme is equally affected. Enzyme recognition of NeuAc a2,8NeuAc residues of GD3 is significantly lower. GaiNAc and Gai-GaiNAc residues linked to sialyl-lactase facilitate enzyme-recognizable orienta­tion of NeuAc a2,8NeuAc of G02 and GD I b. The orientation of the terminal sialyl residues in GD3 may be different between free (purified) and membrane­bound G03. An adjuvant such as MPL may effectively expose disialyl residues of GD3, thereby facilitating immune recognition of GD3.

RAVINDRANATH eta/.: IMMUNOLOGY OF GANGLIOSIDES 311

Augmentation of antiganglioside antibody with prolonged survival in a preclinical model­C57BLI6J mice immunized three times with MPL­incorporated B 16 cells have a significantly slower rate of tumor growth and a significantly longer survival than mice immunized with B 16 alone or MPL alone56

. In addition, we have documented that augmentation of anti-GM3 IgM titers after immunization with tumor cells plus BCG or MPL is correlated with retarded tumor growth65

. Takahashi et a/. 66 have shown that intraperitoneal administration of GM3 in B 16 melanoma-bearing mice significantly augmented melanoma growth, suggesting that GM3 shed from tumor cells may promote tumor growth by suppressing immune surveillance. In support of this observation, that GM3 preferentially suppressed the generation and activity of cytotoxic lymphocytes in tumor-bearing mice, indicating that melanoma­derived gangliosides in the circulation may impede antitumor functions of the immune system. Our experiments indicate that MPL-B 16 vaccine augments anti-GM3 IgM response. High level of anti-GM3 lgM antibodies may clear GM3 from the circulation and thereby reverse GM3-induced immunosuppression, restoring immunocompetence and facilitating tumor regression. Failure of tumor vaccine trials may not be due to inefficiency of adjuvant but instead to nonfunctional expressiOn of tumor-associated antigens on viable tumor cells .

Clinical significance of serum antiganglioside I gM titers-The most prevalent antiganglioside IgM is anticGD I b in healthy individuals and anti-GD3 in patients with stage III melanoma (Fig. 4)34

. GD3 is the major melanoma-associated ganglioside 17

·18

·67

.

Although we have not studied the sera of patients with earlier stages of melanoma, it appears that tu­mor-derived gangliosides may induce specific aug­mentation of antibodies . If so, then serum antigan­glioside IgM titers may serve as a surrogate marker of melanoma. We are also examining this possibility in other human cancers. Patients with pancreatic adeno­carcinoma have high serum titers of anti-GM2 and anti-GD I b lgM antibodies44

.

Conclusion Gangliosides, unique carbohydrate antigens, are

overexpressed on the surface of tumor cells and may be shed into the circulation. They do not elicit a T­cell mediated immune response. Antiganglioside an­tibodies in human sera are invariably IgM, suggesting

that B cells producing antiganglioside IgM may not undergo class-switching. Antiganglioside IgM are found in healthy individuals but their titers may de­cline with age. Persistent IgM is associated with sev­eral diseases, particularly neuropathies. Membrane­bound gangliosides are important tumor-associated antigens and targets for immune attack. Although ganglioside antigens occur in normal tissues, they are poorly recognized by the immune system because they are either hidden (cryptic) or expressed in a sub­optimal density on the cell surface. Our studies reveal that incorporation of a bacterial adjuvant such as MPL onto the cell surface (during infection) or during preparation of cellular vaccines can allow effective recognition of cell-surface gangliosides by the im­mune system. Again, the immune response is humoral not cellular.

Our experiments with MPL suggest that failure of some clinical vaccine trials is due to an ineffective adjuvant. Efficient adjuvanticity depends on the vi­ability of whole cells, integrity of the cell membrane, and topography of tumor-associated antigens. Incor­porating adjuvant onto membrane of tumor cell is crucial to augment immune response to gangliosides . The first step in this immune response is activation of pre-existing B cells that are primed to produce IgM antibodies against gangliosides. Antiganglioside IgM antibodies are naturally occurring autoantibodies that clear immunosuppressive gangliosides shed from tis­sues during aging, during degeneration of neural and extraneual tissues, and during tumor growth . An ef­fective cancer vaccine should therefore stimulate an antiganglioside IgM response that will reverse tumor­induced immunosuppression and restore immuno­competence.

Acknowledgement

We wish to thank Ms. Gwen Berry for editing the manuscript. Supported by grant CA 12582 from the National Cancer Institute and by funding from the Wayne and Gladys Valley Foundation, Oakland, CA.

References I Klenk E, Z Physiol Chem, 267 ( 1942) 128. 2 Ando S, Neurochem lnt, 5 ( 1983) 507 . 3 Weigandt H, Glycolipid (Elsevier, Amsterdam) 1985, 31 4. 4 Basu M, De T, Das K K, Kyle J W, Chon H C, Schaeper R J

& Basu , Methods Enzymol, 138 ( 1987) 575 . 5 Kleene R & Berger E G, Biochem Biophys Acta , 1154 ( 1993)

283. 6 Lee P T , Ketis N V, Barber K R & Grant C W M, Biochim

Biophys Acta , 60 I ( 1980) 302.

312 INDIAN J EXP BIOL, APRIL 2000

7 Ravidranath M H, Ravi ndranath R M H, Morton D L & 36 A vrameas S, Nawral allloantibodies: Their physiologival Graves M C, 1 lmmunol Methods, 169 ( 1994) 257. role and regulatory significance, (CRC Press, Florida) 1993 ,

8 Hiraiwa N, Tsyyuoka K, Li Y, Tanaka M, Seno T, Okubo Y, I. Fukuda Y, lmura H & Kannagi R, Cancer Res, 50 (1990) 37 lri e R F, Sze L L & Saxton R E, Proc Nat/ Acad Sci USA, 79 5497. (1982) 5666.

9 Berger E, Karnerling J P, Kobata A, Pau lson J C & Vliegen- 38 Nishinaka Y, Ravi ndranath M H & lrie R F, Cancer Res, 56 thart J F, Experientia, 38 ( 1982) 11 29. ( 1996) 5666.

10 Vos J P, Lopez-Cardozo M & Cadell a B M, Biochem Bio- 39 Valentino L, Moss T, Olson E, Wang H J, Elashoff R, physActa, 12 11 (1994) 125. Ladisch S, Blood, 75 (1990) 1564.

II Jerne N K, Ann Rev Microbial, 14 ( 1960) 34 1. 40 Offner H, Thieme T Vandenbark, A A, 1 lmmunol, 139 12 Ri ch R R, Fleisher T A, Schwartz B D, Shearer W T & Stro- ( 1987) 3295.

ber W, Clinical Immunology, Principles and Practice, vols I 41 MassaPT,1ExpMed, !78(1993) 1357. and II, (CV Mosby, StLouis) 1996. 42 Robb R J, 1 lmmunol, 136 ( 1986) 971.

13 Svennerholrn L, 1 Lipid Res, 5 (1964) 145. 43 Chu W K & Sharorn F J, Immunology, 79 ( 1993) I 0. 14 Livingston P 0, Ritter G, Oettgern H F & Old L J, Ganglia- 44 Chu K, Ravindranath M H. Gonzales A. Nishimoto K, Tarn

sides and Cancer, (VCH Publi shers, New York) 1989, 293. W Y, Soh D, Bilchik A, Katopodi s N & Morton D L, Cancer 15 Carubia J M, Yu RK, Macala L J, Kirkwood J M & Varga J (I n press) .

M, Biochem Biophys Res Commun, 120 ( 1984) 500. 45 lrie R F, Matsuk i T & Morton D L, Lancet, ( 1989) 786. 16 Tsuchida T, Ravindranath, M H, Saxton , R E & lrie, R F, 46 Nakamura K, Hanibuchi M, Yano S, Tanaka Y, Fujino I,

Cancer Res, 47 ( 1987) 1278. Inoue M, Takezawa T, Sh itara K & Hanai N, Cancer Res, 59 17 Ravindranath M H, Tsuchida T, Morton D L & lrie R F, ( 1999) 5323.

Cancer, 67 ( 199 1) 3029. 47 Rapport M M & Graf L, Prog Allergy, 13 ( 1969) 273. 18 Ravindranath M H & lri e, R F, Malignant Melanoma: Bioi- 48 Sherwin A L, Lowden I A & Wolfe J S, Can 1 Biochem, 42

ogy, Diagnosis and Th erapy (Kiu wer Academic Publishers , ( 1964) 1640. Boston) 1988, 17. 49 Pascal T A, Saifer A & Git lin J, Proc Soc Exp Bioi Med, 121

19 Ravindranath M H & Morton D L, Cancer In vest, 15 ( 1997) (1966) 739. 491. 50 Koscielak J, Hakomori Y & Jeanloz R W, lmmunochemistJ)',

20 Ravindranath MH, Bilchik A J, Chu K, Soh D, Shen P & 5 ( 1968) 441. Morton D L, Proc Am Assoc Cancer Res, 40 ( 1999) 557. 5 1 Naiki M, Marcus D & Ledeen R, 1 lmmunol, 90 ( 1974) 372.

21 Carbone F R & Gleeson P A, Clycobiology, 7 ( 1997) 735. 52 Yokoyama M, Trams E G & Brady R 0, 1 lmmunol, 90

22 Hard ing C V, Kihlberg J, Elofsson M, Magnusson G & Un-( 1963) 372.

53 Helling F M, Claves M, Shang Y, Oettgen H F & Livington anueE R, 11mmunol, 15 1 (1993)2419.

P 0, Ann NY A cad Sci, 690 ( 1993) 396. 23 lshi oka G Y, Lamont A G, Thomson D, Bulbow N, Gas ta F

54 Li vingston P 0 , Calves M J & Natoli E J, 1 lmmww l, 138 C A, Sette A & Grey H M, 1 lmmunol, 148 ( 1992) 2446.

( 1987) 1524. 24 Bendelac A, Science, 269 ( 1995) 185.

55 Ravindranath M H, Brazeau S M & Morton D L, Experien-25 Beckman EM, Porcell SA, Morita C T, Behar S M, Furiong

tia, 50 (1994) 648. S T & Brenner M B, Nature, 372 (1994) 691.

56 Livingston P 0, Zhang S & Lloyd K 0 , Cancer lmmunol 26 Sieling P A, Chatterjee D, Porcelli S A, Prigozy T I, Maz-

lmmunother, 45 ( 1997) I. zacearo R J. Soriano T, Bloom B R, Brenner M B, Kornen-

57 Ravindranath M H, Morton, D L & lrie R F, 1 Autoimmun, 7 berg M, Brennan P J & Modlin R L, Science, 269 ( 1995)

(1994) 803. 227.

58 Bogoch S, Nature, 183 ( 1960) 392. 27 Freirner M L, Mcintosh K, Adams R A, Alvi ng C R &

59 Shy M E, Evans V A & Lubin F D, Ann Neural, 25 ( 1989) Draehrnan D B, 1 Autoimmun , 6 ( 1993) 281.

511. 28 Brewer J W, Randall T D, Park house M E & Corl ey R B,

60 Li vi ngston P 0 , Natoli E J, Calves N J, Stockert E, Oettgen lmmunol Today, 15 (1994) 165 . H F & Old L J, Proc Nat / A cad Sci USA , 84 ( 1987) 291 I.

29 Davis AC, Raux K H & Shulamn M J, Eur 1 lmmww/, 18 6 1 Livingston P O, lmmunol Rev. 145 ( 1995) 147.

(1988) I 00 I. 30 Randall T D, King L B & Corl ey R B, Eur 1 lmmuno/, 20 62 Ravindranath M H, Bauer PM, Amiri A A, Miri

( 1990) 1971. S M, Ke lley M C, Jones R C, & Morton Dl 31 Randall T D, Parkjhouse R M & Corley R B, Proc Na t/ Acad AntiCancer Drugs, 8 ( 1997) 217 .

USA 89 ( 1992) 962. 63 Ravindranath M H, Morton D L & lrie R F, Cancer Res, 49 32 Ravindranath R H M, Ravindranath M H & Graves M C, ( 1989) 3891.

Cell Mol Life Sci, 53 ( 1997) 750. 64 Mali akal M A, Ravindranath M H, lri e R F & Morton D L, 33 Takarnizawa K, lwamori M, Mutai M & Nagai Y, Biochem Clycoconj 1, II (1994) 97.

Biophys Acta, 879 ( 1986) 73. 65 Ravindranath M H & Kelley M C, Jones R C, Arniri A A, 34 Ravindranath M H, Hsueh E, Jones R C & Morton D L, Proc Bauer PM, Morton D L, lnt 1 Cancer, 75 ( 1998) 117.

Amer Assoc Cancer Res, 89 ( 1998) 369. 66 Takahashi K, Ono K, Hi rabayash i Y & Taniguchi, M. 1 lm -35 Ritter G, Boosfeld E, Markstein E. Yu R K, Ren S, Stallcup l/lttno/, )40 ( 1988) 3244.

W B, Oettgen H F, Old L J & Li vingston P 0, Cancer Res, 67 Ravi ndranath M H, Morton D L & lrie R F, Cancer Res, 49 50 (1990) 1403. ( 1989) 3891.