Glycosphingolipid domains on cell plasma membrane

Bioscience Reports, Vol. 19, No. 3, 1999

Glycosphingolipid Domains on Cell Plasma Membrane

Maurizio Sorice,1 Tina Garofalo,1 Roberta Misasi,1 Vincenza Dolo,2Giuseppe Lucania,1 Tiziana Sansolini,3 Isabella Parolini,4Massimo Sargiacomo,4 Maria Rosaria Torrisi,1 and Antonio Pavan2,5

In this study we analyzed by immunofluorescence, laser confocal microscopy, immunoelec-tron microscopy and label fracture technique the ganglioside distribution on the plasmamembrane of several different cell types: human peripheral blood lymphocytes (PBL), Molt-4 lymphoid cells, and NIH 3T3 fibroblasts, which mainly express monosialogangliosideGM3, and murine NS20Y neuroblastoma cells, which have been shown to express a highamount of monosialoganglioside GM2. Our observations showed an uneven distributioiof both GM3 and GM2 on the plasma membrane of all cells, confirming the existence ofganglioside-enriched microdomains on the cell surface. Interestingly, in lymphoid cells theclustered immunolabeling appeared localized over both the microvillous and the nonvillousportions of the membrane. Similarly, in cells growing in monolayer, the clusters were dis-tributed on both central and peripheral regions of the cell surface. Therefore, glycosphingo-lipid clusters do not appear confined to specific areas of the plasma membrane, implyinggeneral functions of these domains, which, as structural components of a cell membranemultimolecular signaling complex, may be involved in cell activation and adhesion, signaltransduction and, when associated to caveolae, in endocytosis of specific molecules.

KEY WORDS: Glycosphingolipids; plasma membrane; GM3; immunocytochemistry.

ABBREVIATIONS: PBL, peripheral blood lymphocytes; LDTI, low density Triton-insoluble; PBS, phosphate-buffered saline; FITC, fluorescein isothiocyanate; mAb,monoclonal antibody.

INTRODUCTION

Gangliosides are sialic acid containing glycosphingolipids (GSLs) inserted into thelipid bilayer of the plasma membrane through their ceramide moiety(Hakomori, 1993) and expressed mainly at the cell surface. Gangliosides may func-tion as specific binding sites for bacterial toxins and viruses (Fishman, 1982), andare implicated in membrane receptor modulation (Garofalo et al., 1988), control ofthe cell cycle (Usuki et al., 1988), cell adhesion and antigen recognition (Whisler

1Dipartimento di Medicina Sperimentale e Patologia, Universita di Roma "La Sapienza", Viale ReginaElena 324, 00161 Roma, Italy. E-mail: [email protected]

2Dipartimento di Medicina Sperimentale, Universita di L'Aquila, Via Vetoio Coppito 2, 67100 L'Aquila,Italy.

3Istituto Dermatologico San Gallicano, IRCCS, Roma, Italy.4Dipartimento di Ematologia, Oncologia e Virologia, Istituto Superiore di Sanita, Roma, Italy.5To whom correspondence should be addressed.

197

0144-8463/99/0600-0197$16.00/0 © 1999 Plenum Publishing Corporation

198 Sorice et al.

et al., 1980). Gangliosides form clusters on the cell plasma membrane where theyare assembled together with sphingomyelin, cholesterol, glycosylphosphatidyl-inositol and signal transducer molecules (Simons and Ikonen, 1997). However, GSLsmay self-assemble independently of membrane proteins or cholesterol (Hakomoriet al., 1998). Anti-GSLs antibodies, lectins (Tillack et al., 1982) and cholera toxin(Thompson et al., 1985) have been used to determine the localization and distri-bution of glycosphingolipids on artificial liposomes or on the cell plasma mem-branes. Clusters of monosialoganglioside GM3 can be observed on the plasmamembrane of human skin fibroblasts (Rosner et al., 1990), mouse melanoma B16cells (Nores et al., 1987) and human peripheral blood lymphocytes (PBL) (Sorice etal., 1997). Additionally, GSL-enriched membrane microdomains, resistant to solubi-lization with Triton X-100 (Brown and Rose, 1992), are found closely associatedwith caveolae, the non-coated plasma membrane invaginations, and in membranesubfractions, distinct from caveolae, associated with GPI-anchored proteins and sig-nal transduction proteins (Stefanova et al., 1991; Sorice et al., 1997).

In this report we used anti-GM3 (GMR6) and anti-GM2 (GMB28) monoclonalantibodies (Kotani et al., 1992), to illustrate by confocal laser scanning immuno-fluorescence microscopy and immunoelectron microscopy the clustered distributionof GM3 and GM2 on the plasma membrane of different cell types. Additionally, weanalyzed the cell surface assembly of GSLs using the label-fracture technique, whichcombines freeze-fracture with immunogold labeling and provides high resolutionsurface views of freeze-fractured plasma membrane. This method has been exten-sively applied to relate the distribution and dynamics of immunogold labeled surfacemolecules to that of intramembrane structural components revealed by freeze-frac-ture (Pavan et al., 1989a; 1989b; 1992a; 1992b). Finally, we also investigated theGM3 composition of low density Triton insoluble (LDTI) fractions isolated fromPBL.

MATERIALS AND METHODS

Cells

Human PBL were isolated from fresh heparinized blood by Lymphoprep(Nycomed AS Pharma Diagnostic Div., Oslo, Norway) density-gradient centrifug-ation and washed three times in phosphate-buffered saline (PBS) pH7.4. Molt-4lymphoid cells (Minowada et al., 1972) were cultured in RPMI 1640 medium (FlowLaboratories, Irvine, Scotland) supplemented with 10% fetal calf serum (FCS)(Flow Laboratories) and 2% L-glutamine. Murine NS20Y neuroblastoma cells andNIH 3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM,Bio Whittaker, Walkersville, Maryland, USA), containing 10% FCS plus 100 units/ml penicillin, l00 mg/ml streptomycin and 250 pg/ml fungizone, at 37°C underhumidified 5% CO2 atmosphere.

Confocal Laser Scanning Immunofluorescence Microscopy

Murine neuroblastoma NS20Y cells were fixed in 4% formaldehyde in PBS for2 hr at 4°C. After washing three times in PBS, cells were, alternatively, incubated

Membrane Lipid Domains 199

with anti-GM3 (GMR6) or anti-GM2 (GMB28) mAb for 1 hr at 4°C, followed bygoat anti-mouse IgM (jU-chain specific), (1:10 in PBS, for 1 hr at 4°C). (Sigma Chem.Co., St Louis, MO, USA). Scanning confocal images were acquired through con-focal laser scanning microscope (Sarastro 2000, Molecular Dynamics) adapted to aNIKON OPTIPHOT microscope (objective PLAN-APO 60/1.4 oil) and equippedwith argon ion laser (25 mW). FITC was excited at 488 nm laser power was set at0.5 mW. Images were collected at 512x512 pixel with voxel dimensions 0.08mm(lateral), 0.7 mm (axial). After having been processed with routines for noise filtering,serial optical sections were assembled in Look-Depth mode. Acquisition and pro-cessing were carried out using Image Space software (Molecular Dynamics).

Immunoelectron Microscopy

Thin sections: Molt-4 lymphoid, NIH 3T3 and NS20Y cells were fixed in for-maldehyde (4% in PBS for 2 hr at 4°C) and incubated with anti-GM3 (GMR6) oranti-GM2 (GMB28) mAbs for 1 hr at 4°C, followed by goat anti-mouse IgM (ju-chain specific) (Sigma Chem.), (1:10 in PBS, for 1 h at 4°C). After a second fixationin glutaraldehyde (1% in PBS for 1 hr at 4°C), cells were extensively washed andlabeled with colloidal gold (18 nm, prepared by the citrate method) conjugated withprotein A (Pharmacia Fine Chemicals, Uppsala, Sweden) for 3 hr at 4°C. Controlexperiments were performed omitting the monoclonal antibody from the immuno-labeling procedure. All samples were postfixed in osmium tetroxide 1% in Veronalacetate buffer, pH 7.4, for 2 hr at 4°C, stained with uranyl acetate (5 mg/ml), dehy-drated in acetone, and embedded in Epon 812. Thin sections were examinedunstained or poststained with uranyl acetate and lead hydroxide, with a transmissionelectron microscope CM 100 TEM Philips.

Label-fracture: For the label-fracture procedure, Molt-4 cells were fixed with4% formaldehyde in PBS for 2 hr at 4°C and immunogold labeled as above. Toinduce the displaying and clustering of intramembrane particles (IMPs) we treatedunfixed cells with 30% glycerol for 1 hr at 37°C (Pavan et al., 1989a; 1989b; 1992a;1992b) before fixation with glutaraldehyde (1% in PBS for 1 hr at 4°C) and goldimmunolabeling as above. All immunogold labeled samples were impregnated with30% glycerol, frozen, freeze-fractured in a freeze-fracture device (-105°C,10~6mmHG) and replicated by evaporation from a platinum/carbon gun. The rep-licas were floated into distilled water, picked by adhesion to form var-coated gridsand observed with Philips TEM CM100. Stereo pairs were made with +6° tilt.

Isolation of Low-Density Triton-Insoluble Complexes (LDTI)

5 x 108 PBL were washed and resuspended in 2 ml Mes buffered saline (MBS)(25 mM mes, pH 6.5, 0.15 M NaCl) containing 1% Triton X-100 and 1 mMphenylmethylsulfonyl fluoride and homogenized with 10 strokes of a Dounce homo-genizer, adjusted to 40% sucrose and placed at the bottom of an ultracentrifuge tubeas reported (Sargiacomo et al., 1993). A 5-30% linear sucrose gradient was thenplaced above the lysate, and centrifuged at 39,000 rpm for 16 hr at 4°C in a SW41rotor (Beckman Inst., Palo Alto, CA). The visible band migrating at 20% sucrose

200 Sorice et al.

was harvested and washed twice with MBS at 14,000 rpm for 30 min at 4°C. 5 x 108

cells (5-7 mg total protein), yielded 15-17 ^g of LDTI complexes, i.e., 0.26% ofthe initial homogenate. For negative staining in immunoelectron microscopy, freshsuspension of LDTI in PBS was applied to collodion-coated grids and fixed withformaldehyde 2% in PBS for 15 min at room temperature. The LDTI preparationwas immunolabeled with anti-GM3 mAb in PBS containing 1% BSA for 1 hr in ahumidified chamber. After washing, the sample was incubated with gold-conjugatedanti-mouse IgM, l0 nm (Sigma Chem.) for 1 hr. The sample was then negativelystained with 1% phosphotungstic acid, brought to pH 7.0 with NaOH and examinedunder Philips TEM CM 100.

RESULTS

Immunofluorescence Analysis

Immunofluorescence experiments were performed on neuroblastoma NS20Ycells using anti-GM2 GMB28 mAb (Fig. 1A, B), or anti-GM3 GMR6 mAb (Fig.1C). Our results, using confocal laser scanning microscope, showed that on NS20Ycells, immunostaining of both gangliosides appeared uneven and punctate over thecell plasma membrane. To exclude the possibility that the clustered distribution

Fig. 1. Scanning confocal microscopic analysis of GM2 (A, B) and GM3 (C) immunolabelingon plasma membrane of NS20Y neuroblastoma cells. Immunolabeling appears uneven and punc-tate over the cell surfaces. (D) Immunogold labeling of GM3 on NS20Y cell surface showing anintense and uneven distribution over the cell plasma membranes. A-C, x 1.000; bar, 4^m.D, x 14.000; bar, 0.5 jm.

Membrane Lipid Domains 201

might result from cross-linking by secondary antibody of only partially immobilizedganglioside molecules, formaldehyde-fixed NS20Y cells were incubated with GMB28mAb and then fixed with formaldehyde before the addition of the second antibody(Butor et al., 1991; Sorice et al., 1997). In this experimental procedure, our obser-vations confirmed that the immunofluorescence clustered signal represents the nativedistribution of ganglioside molecules over the cell plasma membrane (not shown).

Immunoelectron Microscopic Analysis

To investigate the distribution of ganglioside molecules on cell plasma mem-brane, we analyzed, by immunoelectron microscopy, different cell types. We selectedfor our study murine neuroblastoma NS20Y cells, (Fig. 1D) where GM2 is the mainganglioside constituent (Misasi et al., 1996) and PBL (Fig. 2A), Molt-4 lymphoid(Fig. 2B) and NIH 3T3 cells (Fig. 2C) in which GM3 represents the major ganglio-side component of the plasma membrane (Minowada et al., 1972; Meivar-Levy etal., 1977). Immunoelectron microscopic observations of PBL and Molt-4 cells,showed, that the surface distribution of GM3 was not uniform. In fact, the goldimmunolabeling appeared in clusters of variable size and localized either over themicrovilli (not shown) or over the non-villous portions of the membrane (Fig. 2A,B, arrowheads). Only a very low number of isolated gold particles could be observed.Our analysis of GSLs distribution was extended to murine NIH 3T3 fibroblasts andNS20Y neuroblastoma cells, expressing high amount of GM3 and GM2 moleculesrespectively, as examples of cells growing in monolayer. Also in these cells, GM3(Fig. 2C, arrowheads) or GM2 (Fig. 1D) immunolabeling appeared clustered andequally distributed over the central regions of the plasma membrane and on theperipheral portions of the cell surface. Totally similar findings were observed whenthe cells were fixed with glutaraldehyde immediately after the incubation with anti-GM3 or anti-GM2 mAbs and before the secondary antibody (not shown), confirmingthat the immunogold clustering represents the native distribution of the gangliosidemolecules over the cell plasma membrane.

Label-Fracture Analysis

Label-fracture was generally applied to relate the distribution and dynamics ofimmunogold labeled surface molecules over the cell surface with that of structuralcomponents on the exoplasmic faces of the freeze-fractured plasma membranes, suchas intramembrane particles (IMPs), taken to represent integral membrane proteins(Pavan et al., 1989a; 1989b; 1992a; 1992b). We decided here to use this method toeventually localize the GM3 molecules on the smooth areas of the freeze-fracturedplasma membrane corresponding to the lipid portion of the membrane. Ultra-structural observation of the label-fracture images of GM3 immunolabeled Molt-4cells revealed a dense and uniform distribution of IMPs (Fig. 3). Again, the immuno-gold particles were mostly distributed over the cell surfaces in clusters correspondingto ganglioside domains. Stereo views only partially indicated that these gold clusterswere localized on the smooth areas between the intramembrane particles. To bettervisualize this localization, we incubated unfixed Molt-4 cells in a glycerol solution,

202 Sorice et al.

Fig. 2. (A) Immunogold labeling of GM3 on the lymphocyte plasma membranes. The surfacedistribution of the immunogold particles is not uniform. Small clusters are localized over non-villous portions (arrowheads) of the plasma membranes. (B) Immunogold labeling of GM3 onformaldehyde pre-fixed Molt-4 cells is more intense than that observed on lymphocytes. Goldparticles are non-uniformly distributed and reveal the presence of large clusters of GM3 over thecell surface. (C) Immunogold labeling of GM3 on formaldehyde pre-fixed NIH 3T3 cells. Also incells growing in monolayer, gold particles are non-uniformly distributed revealing the presence ofclusters of GM3 molecules on both central and peripheral regions of the cell surface. A, x 40.000;B, x 25,000; C, x 27,000; bar, 0.5 /mi.

a non-physiological treatment used to experimentally induce a generalized aggre-gation of the IMPs and to leave clearly demarcated smooth, particle-free areas corre-sponding to the lipid portions of the membrane. Label-fracture of glycerol treatedcells showed a decreased density of anti-GM3 labeling, because of a loss of anti-genicity caused by the glutaraldehyde fixation; however, gold particles, and thereforeGM3 molecules, were exclusively confined to the particle-free areas of the freeze-fractured membranes (Fig. 4).

Membrane Lipid Domains 203

Fig. 3. Immunogold labeling of GM3 in freeze-fractured Molt-4 lymphoid cells as seenin label-fracture (stereo view). IMPs are uniformly distributed over the freeze-fracturedplasma membranes. Gold particles are mostly in clusters corresponding to gangliosidedomains, which appear localized on the smooth areas between the intramembraneparticles. x 35.000; bar, 1 /an.

Ganglioside Distribution on LDTI Plasma Membrane Domains

We investigated the GM3 composition of LDTI fractions of lymphocyte plasmamembrane by negative staining of immunolabeled fresh preparation of LDTI com-plexes. The application of heavy metal salt solution dries down support film,resulting in an electron-translucent membrane domain on a dark surrounding back-ground. The LDTI fraction immunogold labeled with anti-GM3 mAb indicated theexistence of GM3-enriched domains on these membrane fragments (Fig. 5).

DISCUSSION

Clustering the GSLs may occur in all cell plasma membranes determining theformation of GSL microdomains involved in antigen recognition, cell adhesion,(Whisler et al., 1980) and binding of signal transduction molecules (Gouy et al.,1995). Since gangliosides show cell type-specific expression patterns (Hakomori,1990), in this study we analyzed the ganglioside distribution on the plasma mem-brane of several different cell types: human PBL, Molt-4 lymphoid cells, and NIH

204 Sorice et al.

Fig. 4. (A-C) Immunogold labeling of GM3 in freeze-fractured glycerol-treatedMolt-4 lymphoid cells as seen in label-fracture. Immunogold labeling shows thatGM3 molecules are exclusively confined to the smooth and particle-free areasbetween the aggregated IMPs of the exoplasmic fracture face. x 60.000;bar, 0.2 fim.

3T3 fibroblasts, which mainly express GM3, and murine NS20Y neuroblastomacells, which have been shown to express a high amount of GM2. Our immunoelec-tron microscopic observations showed an uneven distribution of GM3 and GM2 onthe plasma membrane of all cells analyzed. Since very few isolated gold particleswere present, our results confirm the existence of ganglioside-enriched microdomainson the cell surface. These observations are consistent with previously reportedthermodynamic results, showing that gangliosides form clusters when their concen-tration is higher than a critical value and may be responsible for an increase ofthe ion permeation across the membrane and for an enhanced polarizability of itshydrophilic region (Cametti et al., 1992). Interestingly, the clustered immunolabelingappeared localized over both the microvillous and the non-villous portions of themembrane in lymphoid cells, and distributed on both central and peripheral regionsof the cell surface in cells growing in monolayer. These results suggest that general

Membrane Lipid Domains 205

Fig. 5. Electron microscopy negative staining of LDTI complexes. Immunogold label-ing of GM3 on LDTI complexes. The presence of gold particles on the membranefractions identifies domains enriched in GM3. x 30.000; bar, 0.5 ^m.

functions of these GSL domains may not be confined to specific areas of the plasmamembrane.

Label-fracture technique, which combines freeze-fracture with immunogoldlabeling, was used here to relate the distribution of GM3 molecules with that ofintramembrane structural components. These results show that, in freeze-fracturedMolt-4 cells, the gold immunolabeling is restricted to the smooth portions of theexoplasmic fracture-face, whereas the IMPs, aggregated by glycerol treatment,appear totally unlabeled. In our previous reports, label-fracture was used to addressthe relationship between IMPs and the surface distribution and dynamics of CD4,CD3, CD8, HLAI and LFA-1 antigens in lymphocyte plasma membrane and inepithelial transfected cells (Pavan et al., 1989a; 1989b; 1992a; 1992b). We showedthat, after glycerol-induced aggregation of IMPs, the lymphocyte antigens wereexclusively confined to the IMPs aggregates, indicating that transmembrane pro-teins, upon fracture, give rise to IMPs and suggesting that lipid molecules corre-spond to the smooth, particle free areas in between contoured particle aggregates ofthe freeze-fractured membrane. Here we showed for the first time, that immunolab-eled ganglioside molecules are strictly confined to these smooth areas between theparticle aggregates.

The immunolabeled GM3 molecules detected in our LDTI membrane fractionisolated from lymphocytes, confirms the statement that these GSL-enriched domainshave some functional connection with transducer molecules. In fact, it has beenreported the presence of c-Src family kinases p53/56lyn, CD4, Rho and FAK in

206 Sorice et al.

detergent insoluble membrane fractions (Hakomori et al., 1998; Parolini et al., 1996).Similarly, caveolae, invaginations of plasma membrane thought to be involved inendocytosis and signal transduction (Simons and Ikonen, 1997), are enriched in cav-eolin, glycosphingolipids, cholesterol and sphingomyelin and are associated to deter-gent-insoluble membraneous fraction. However, we showed the presence of GM3clusters in cells that have been shown to lack caveolin, confirming the hypothesisthat GM3-enriched membrane subfractions and caveolae are functionally distinct,although their composition is partially overlapping.

GSL domains are structurally and functionally associated with membrane-bound proteins. Specific binding of GM3 to EGF receptor tyrosine kinase (Bremeret al., 1986), and of GM1 to Trk A kinase (Mutoh et al., 1995) indicates a specificrole of these gangliosides as modulators of tyrosine kinase activities. In addition, werevealed colocalization and complex formation between prosaposin, a neurotrophicfactor involved in extracellular signal-regulated protein kinase (ERK) phosphoryl-ation and cell differentiation (O'Brien et al., 1994; Campana et al., 1996), and GM3on NS20Y cell plasma membrane (Misasi et al., 1998). In human PBL, we recentlyreported a GM3-CD4 interaction, showing that GM3, CD4 and 1ck are selectivelyrecovered in the same domains of the plasma membrane (Sorice et al., 1997). Inaddition, we demonstrated that addition of exogenous GM3 to lymphocyte plasmamembrane selectively induces redistribution, clustering and internalization via endo-cytic pits and vesicles of CD4 molecules (Sorice et al., 1995), and that this processis dependent on PKC activation, CD4 phosphorylation and CD4:p56Ick dissociation(Garofalo et al., 1998).

In conclusion, our results confirm the existence of ganglioside plasma mem-brane microdomains functionally capable, as structural components of a cell mem-brane multimolecular signaling complex, to regulate cell activation and adhesion,signal transduction and, when associated to caveolae, endocytosis of specificmolecules.

ACKNOWLEDGEMENTS

This work was supported by grants from Associazione Italiana per la Ricercasul Cancro (AIRC) Italy, Ministero dell'Universita e della Ricerca Scientifica(MURST) Italy, Consiglio Nazionale delle Ricerche (CNR) target project on Bio-technology Italy, and Ministero del Lavoro n. 792, Italy.

REFERENCES

Bremer, E. G., Schlessinger, J., and Hakomori, S. (1986) Ganglioside-mediated modulation on cellgrowth: specific effects of GM3 on tyrosine phosphorylation of the epidermal growth factor receptor.J. Biol. Chem. 261:2434-2440.

Brown, D. A. and Rose, J. K. (1992) Sorting of GPI-anchored proteins to glycolipid-enriched membranesubdomains during transport to the cell surface. Cell 68:533-544.

Butor, C., Stelzer, E. H. K., Sonnenberg, A., and Davoust, J. (1991) Apical and basal Forssmann antigenin MDCK II cells: a morphological and quantitative study. Eur. J. Cell Biol. 56:269-285.

Cametti, C. et al. (1992) Alteration of the passive electrical properties of lymphocyte membranes inducedby GM1 and GM3 glycolipids. Biochim. Biophys. Acta 1111:197-203.

Membrane Lipid Domains 207

Campana, W. M., Hiraiwa, M., Addison, K. C., and O'Brien, J. S. (1996) Induction of MAPK phos-phorylation by prosaposin and prosaptide in PC12 cells. Biochem. Biophys. Res. Commun. 229:706-712.

Fishman, P. H. (1982) Role of membrane gangliosides in the binding and action of bacterial toxins. J.Membr. Biol. 69:85-97.

Garofalo, T. et al. (1998) A novel mechanism of CD4 down-modulation induced by monosialogangliosideGM3. J. Biol. Chem. 273:35153-35160.

Gouy, H., Debre, P., and Bismuth, G. (1995) Triggering of a sustained calcium response through a p56lck-dependent pathway by exogenous ganglioside GM1 in human T lymphocytes. J. Immunol.155:5160-5166.

Hakomori, S. (1990) Bifunctional role of glycosphingolipids. J. Biol. Chem. 265:18713-18716.Hakomori, S. (1993) Structure and function of sphingoglycolipids in transmembrane signaling and cell-

cell interactions. Biochem. Soc. Trans. 21:583-595.Hakomori, S., Handa, K., Iwabuchi, K., Yamamura, S., and Prinetti, A. (1998) New insights in glyco-

sphingolipid function: "glycosignaling domain", a cell surface assembly of glycosphingolipids withsignal transducer molecules, involved in cell adhesion coupled with signaling. Glycobiology 8:xi-xix.

Kotani, M., Ozawa, H., Kawashima, I., Ando, S., and Tai, T. (1992) Generation of one set of monoclonalantibodies specific for a-pathway ganglio-series gangliosides. Biochim. Biophys. Acta 1117:97-103.

Meivar-Levy, I., Sabanay, H., Bershadsky, A. D., and Futerman, A. H. (1997) The role of sphingolipidsin the maintenance of fibroblast morphology. J. Biol. Chem. 272:1558-1564.

Minowada, J., Ohnuma, T., and Moore, G. E. (1972) Rosette-forming human lymphoid cell lines. I:Establishment and evidence for origin of thymus-derived lymphocytes. J. Natl. Cancer Inst. 49:891-895.

Misasi, R. et al. (1966) Prosaposin and prosaptide, a peptide from prosaposin, induce an increase inganglioside content on NS20Y neuroblastoma cells. Glycoconjugate J. 13:195-202.

Misasi, R. et al. (1998) Colocalization and complex formation between prosaposin and monosialo-ganglioside GM3 in neural cells. J. Neurochem. 71:2313-2321.

Mutoh, T., Tokuda, A., Miyadai, T., Hamaguchi, M., and Fujiki, N. (1995) Ganglioside GM1 binds tothe Trk protein and regulates receptor functions. Proc. Natl. Acad. Sci. USA 92:5087-5091.

Nores, G. A., Dohi, T., Taniguchi, M., and Hakomori, S. (1987) Density-dependent recognition of cellsurface GM3 by a certain anti-melanoma antibody, and GM3 lactone as a possible immunogen:requirements for tumor-associated antigen and immunogen. J. Immunol. 139:3171-3176.

O'Brien, J. S., Carson, G. S., Seo, H. C., Hiraiwa, M., and Kishimoto, T. (1994) Identification of prosa-posin as a neurotrophic. Proc. Natl. Acad. Sci. USA 91:9593-9596.

Parolini, I., Sargiacomo, M., Lisanti, M. P., and Peschle, C. (1996) Signal transduction and glyco-phosphatidylinositol-linked proteins (LYN, LCK, CD4, CD45, G proteins, and CD55) selectivelylocalize in Triton-insoluble plasma membrane domains of human leukemic cell lines and normalgranulocytes. Blood 87:3783-3794.

Pavan, A., Mancini, P., Frati, L., Torrisi, M. R., and Pinto da Silva, P. (1989a) Molecular cytochemistryof CD3 and CD4 antigens in human lymphocytes as studied by label-fracture and by fracture-label.Biochim. Biophys. Acta 978:158-168.

Pavan, A., Mancini, P., Cirone, M., Frati, L., Torrisi, M. R., and Pinto da Silva, P. (1989b) Cappingof HLA antigens in human lymphocytes as followed by immunogold label-fracture. J. Histochem.Cylochem. 37:1489-1496.

Pavan, A. et al. (1992a) Dynamics of transmembrane proteins during Sindbis virus budding. J. Cell Sci.102:149-155.

Pavan, A., Lucania, G., Sansolini, T., Frati, L., and Torrisi, M. R. (1992b) Patching and capping ofLFA-1 molecules on human lymphocytes. Histochemistry 98:253-258.

Rosner, H., Greis, C. H., and Rodemann, H. P. (1990) Density-dependent expression of ganglioside GM3by human skin fibroblasts in an all-or-none fashion, as a possible modulator of cell growth in vitro.Exp. Cell Res. 190:161-169.

Sargiacomo, M., Sudol, M., Tang, Z. L., and Lisanti, M. P. (1993) Signal transducing molecules andglycosylphosphatidylinositol-linked proteins form a caveolin-rich insoluble complex in MDCK cells.J. Cell. Biol. 122:789-807.

Simons, K. and Ikonen, E. (1997) Functional rafts in cell membranes. Nature 387:569-572.

208 Sorice et al.

Sorice, M. et al. (1997) Evidence for the existence of ganglioside-enriched plasma membrane domains inhuman peripheral lymphocytes. J. Lip. Res. 38:969-980.

Sorice, M. et al. (1995) Monosialoganglioside GM3 induces CD4 internalization in human peripheralblood T lymphocytes. Scand. J. Immunol. 41:148-156.

Stefanova, I. and Horejsi, V. (1991) Association of the CD59 and CD55 cell surface glycoproteins withother membrane molecules. J. Immunol. 147:1587-1592.

Thompson, T. E., Allietta, M., Brown, R. E., Johnson, M. L., and Tillack, T. W. (1985) Organizationof ganglioside GM1 in phosphatidylcholine bilayers. Biochim. Biophys. Acta 817:229-237.

Tillack, T. W., Wong, M., Allietta, M., and Thompson, T. E. (1982) Organization of the glycosphingo-lipid asialo-GMl in phosphatidylcholine bilayer. Biochim. Biophys. Acta 691:261-273.

Usuki, S., Hoops, P., and Sweely, C. C. (1988) Growth control of human foreskin fibroblasts and inhi-bition of extracellular sialidase activity by 2-deoxy-2,3-dehydro-N-acetylneuraminic acid. J. Biol.Chem. 263:10595-10599.

Whisler, R. L. and Yates, A. J. (1980) Regulation of lymphocyte responses by human gangliosides:characteristics of inhibitory effects and the induction of impaired activation. J. Immunol. 125:2106-2111.