Biosynthesis and expression of gangliosides during differentiation of chick embryo retina cells in vitro

Experimental Guillain-Barre syndrome induced by immunization with gangliosides: Keyhole limpet hemocyanin is required for disease triggering

An experimental model of Guillain-Barré Syndrome has been established in recent years. Rabbits develop disease upon immunization with a single dose of an emulsion containing bovine brain gangliosides, KLH and complete Freund's adjuvant. Within a period of four to ten weeks after immunization, they began to produce anti-ganglioside IgG-antibodies first, and to show clinical signs of neuropathy afterwards. In addition to gangliosides, KLH is a requirement for antibody production and disease triggering. Although KLH is commonly used as an immunological carrier protein, an anti-KLH-specific immune response was necessary for induction of both events. KLH is a glycoprotein carrying most of the immunogenicity in its glycan moiety. Between 20% to 80% of anti-ganglioside IgG-antibodies present in sick rabbit sera cross-reacted with KLH, indicating that both immune responses are related. The terminal Gal-ß(1,3)-GalNAc glycan (present in gangliosides and KLH) is proposed as "key" antigenic determinant involved in inducing the anti-ganglioside immune response. These results are discussed in the context of the "binding site drift" hypothesis.

Defective GM3 synthesis in Cog2 null mutant CHO cells associates to mislocalization of lactosylceramide sialyltransferase in the Golgi complex

The conserved oligomeric Golgi (COG) complex is a eight subunit (COG1 to 8) tethering complex involved in the retrograde trafficking of multiple Golgi processing proteins. Here we studied the glycolipid synthesis status in ldlC cells, a Cog2 null mutant CHO cell line. Biochemical studies revealed a block in the coupling between LacCer and GM3 synthesis, resulting in decreased levels of GM3 in these cells. Uncoupling was not attributable to decreased activity of the glycosyltransferase that uses LacCer as acceptor substrate (SialT1). Rather, immunocytochemical experiments evidenced a mislocalization of SialT1 as consequence of the lack of Cog2 in these cells. Co-immunoprecipitation experiments disclose a Cog2 mediated interaction of SialT1 with the COG complex member Cog1. Results indicate that cycling of some Golgi glycolipid glycosyltransferases depends on the participation of the COG complex and that deficiencies in COG complex subunits, by altering their traffic and localization, affect glycolipid composition.

Anti-GM1 antibodies as a model of the immune response to self-glycans

Glycans are a class of molecules with high structural variability, frequently found in the plasma membrane facing the extracellular space. Because of these characteristics, glycans are often considered as recognition molecules involved in cell social functions, and as targets of pathogenic factors. Induction of anti-glycan antibodies is one of the early events in immunological defense against bacteria that colonize the body. Because of this natural infection, antibodies recognizing a variety of bacterial glycans are found in sera of adult humans and animals. The immune response to glycans is restricted by self-tolerance, and no antibodies to self-glycans should exist in normal subjects. However, antibodies recognizing structures closely related to self-glycans do exist, and can lead to production of harmful anti-self antibodies. Normal human sera contain low-affinity anti-GM1 IgM-antibodies. Similar antibodies with higher affinity or different isotype are found in some neuropathy patients. Two hypotheses have been developed to explain the origin of disease-associated anti-GM1 antibodies. According to the "molecular mimicry" hypothesis, similarity between GM1 and Campylobacter jejuni lipopolysaccharide carrying a GM1-like glycan is the cause of Guillain-Barré syndrome associated with anti-GM1 IgG-antibodies. According to the "binding site drift" hypothesis, IgM-antibodies associated with disease originate through changes in the binding site of normally occurring anti-GM1 antibodies. We now present an "integrated" hypothesis, combining the "mimicry" and "drift" concepts, which satisfactorily explains most of the published data on anti-GM1 antibodies.

High affinity of anti-GM1 antibodies is associated with disease onset in experimental neuropathy

High antibody affinity has been proposed as a disease determinant factor in neuropathies associated with anti-GM1 antibodies. An experimental model of Guillain-Barré syndrome, induced by immunization of rabbits with bovine brain gangliosides or GM1, was described recently (Yuki et al. [2001] Ann. Neurol. 49:712-720). We searched plasma from these rabbits, taken at disease onset and 1 or 2 weeks prior to onset, for the presence of high-affinity anti-GM1 IgG antibodies. Affinity was estimated by soluble antigen binding inhibition. High-affinity antibodies (binding inhibition by 10(-9) M GM1) were detected at disease onset but not before. No such difference was found for other antibody parameters such as titer, fine specificity, and population distribution. These findings support the proposed role of high affinity as an important factor in disease induction by anti-GM1 antibodies.

Unusual presence of anti-GM1 IgG-antibodies in a healthy individual, and their possible involvement in the origin of disease-associated anti-GM1 antibodies

Anti-GM1 antibodies of the IgG isotype are found in serum from patients with Guillain-Barré syndrome. In normal human sera, anti-GM1 IgM-antibodies are commonly present, but their IgG counterpart has not been previously demonstrated. During routine screening, we found a normal human serum with a high titer of anti-GM1 IgG-antibodies (IgG1 subclass). Affinity estimation by soluble antigen-binding inhibition indicated that they are low-affinity antibodies with IC50 values between one and two orders of magnitude higher than those of anti-GM1 IgG-antibodies from Guillain-Barré patients. Various antibody parameters remained fairly constant for 1 year, in additional serum samples taken at 4-month intervals. Such anti-GM1 IgG1-antibodies were not detected in > 100 other normal serum samples tested, indicating a very low frequency in the general population. The low affinity of these unusually present antibodies could explain the absence of disease, despite their relatively high titer. The significance of this finding in the origin of disease-associated anti-GM1 antibodies is discussed.

Early variations of the disialoganglioside GD3 in chicken embryonic brain support its role in cell migration

In the present study a primary culture system of chicken embryo brain neurons was used in the early period of chicken brain development from day 6 until day 8, which was shown to be a suitable model of neuritogenesis, cell migration and reaggregation. Dissociated chicken optic tectum cells from embryonic stage 31 were cultured on polylysine-coated dishes under serum-free conditions up to 3 days. Freshly dissociated neurons developed short processes, which contacted one another and formed fasciculated bundles. Cell somata migrated along the neurite bundles, similar to migrating neurons in vivo, forming three-dimensional tissue-like clusters. This system was used to study the possible functions of the disialoganglioside GD3 for these neuronal differentiation steps. GD3 represents the predominant ganglioside of embryonic neurons before neuritogenesis in vitro and in vivo. Its biosynthesis is followed during day 6 until day 8 of embryonic brain development. Incubation of dissociated neurons with the monoclonal antibody R-24, recognising the GD3 on the cell surface, led to a total blocking of neurite outgrowth. Accordingly, neither cell migration nor reaggregation could be found. These results indicate that the disialoganglioside GD3 plays a central role in neuronal differentiation and development in the embryonic chicken brain.

Isolation of total-RNA from formalin-fixed rat retina

The aim of this project was to establish a method for the purification of total-RNA from fixed rat-retina. Two different established methods were used for RNA purification, and successful isolation was verified with RT-PCR for amplification of beta-actin (two different product-lengths) and subsequent gel-electrophoresis. Total-RNA was successfully isolated from fixed rat-retina. The house keeping gene, beta-actin could be detected after fixing the retina either with 1% formalin or with 4% paraformaldehyde (PFA). Hexamer-primer based RT-PCR gave better results than the oligo-d(T)-primer based RT-PCR method. Both the 698 and 225 bp beta-actin-fragments could be successfully amplified, where amplification of the latter was more efficient. This approach shows that tissue fixation prior to RNA-isolation facilitates the rapid isolation of undamaged RNAs in tissues such as the retina, which are known to yield low levels of RNA and are vulnerable to RNases.

Neurotrophic factors and depolarization do not enhance viability and process regrowth on a purified set of chick embryo retinal ganglion cells

As retinal ganglion cells (RGCs) develop, the amount of membrane glycoprotein Thy-1 increases, consequently, the RGC binding to Thy-1 antibody enlarges. This phenomenon sustained Thy-1 panning purification of two sets: loose bound cells (LBCs); and tight bound cells (TBCs). Thy-1, neurofilament and growth associated protein-43 characterized both sets as RGCs, but the expression of RA4 antigen and gangliotetraosylgangliosides distinguished LBCs as immature and TBCs as mature. The cell composition in 4-day TBC cultures remained unchanged as immunocharacterization indicated. These cells survived and regrew their processes in plain medium, however, trophic factors (TFs) and depolarization did not enhance their development. This peculiar behavior might be due to their neural maturity, whereas in LBC cultures, TFs produced positive effects.

The origin of anti-GM1 antibodies in neuropathies: the "binding site drift" hypothesis

Elevated titers of serum antibodies against GM1-ganglioside are associated with a variety of autoimmune neuropathies. The origin of these autoantibodies is still unknown, although there is evidence that they are produced by CD5+ B-lymphocytes and that antigen mimicry is involved. Anti-GM, IgM-antibodies in the normal human immunological repertoire are low affinity antibodies that cross-react with other glycoconjugates carrying Gal beta1-3GalNAc and probably do not have GM1-mediated biological activity. Other anti-GM1 IgM-antibodies with higher affinity and/or different fine specificity are present in patients with motor syndromes. Based on our studies of structural requirement for binding, we hypothesize that disease-associated anti-GM1 antibodies originate at random by mutations affecting the binding site of naturally-occurring ones. The hypothesis is conceptually similar to the established phenomenon of "genetic drift" in species evolutionary biology and is therefore termed "binding site drift".

High affinity as a disease determinant factor in anti-GM(1) antibodies: comparative characterization of experimentally induced vs. disease-associated antibodies

Elevated titers of serum anti-GM(1) antibodies of IgG isotype are found frequently in patients with Guillain-Barré syndrome. Much evidence indicates that these autoantibodies are involved in disease progression, but their exact function and the mechanism of their appearance are still unclear. In an attempt to reproduce "ganglioside syndrome", the experimental model of neuropathy developed by Nagai et al. (Neurosci. Lett. 2 (1976) 107), rabbits were intensively immunized with GM(1) in complete Freund adjuvant (CFA). High titers of anti-GM(1) antibodies were produced, with class switch and affinity maturation indicating an elaborate immune response. Unexpectedly, the rabbits did not show any clinical symptoms of neuropathy. Relatively affinities of both IgM and IgG antibodies were significantly lower than those of similar antibodies from neuropathy patients. These results suggest the existence of a threshold value above which affinity of anti-GM(1) antibodies becomes an important factor in disease induction. The absence of neuropathy symptoms in rabbits may be explained by absence of these high-affinity anti-GM(1) antibodies.

Specific ganglioside changes in extraneural tissues of adult rats with hypothyroidism

Adults rats with hypothyroidism were prepared by administration of 6-propyl-2-thiouracil (PTU) or methimazole, and the tissues were examined for their gangliosides through methods including glycolipid-overlay techniques. Normal thyroid tissue contained GM3, GD3, and GD1a as the major gangliosides, with GM1, GD1b, GT1b, and GQ1b in lesser amounts. The goitrous tissue of PTU-induced hypothyroid rats had higher concentrations of GM1 and GD1a with a concomitant decrease of GM3. The amount of GT3 in thyroid tissue was increased in hypothyroid animals. While normal liver tissue had a complex ganglioside pattern with a- and b-series gangliosides, the PTU-induced hypothyroid tissue showed a simpler ganglioside profile that consisted mainly of a-series gangliosides with almost undetectable amounts of b-series gangliosides. The expression of c-series gangliosides was suppressed in the hypothyroid liver tissue. Heart tissue had higher contents of GM3 and GT3 than control. No apparent change was observed in the compositions of major and c-series gangliosides in other extraneural tissues (i.e., kidney, lung, spleen, thymus, pancreas, testis, skeletal muscle, and eye lenses), and neural tissues (i.e., cerebrum and cerebellum) from PTU-induced hypothyroid rats. The ganglioside changes of thyroid, liver, and heart tissues were reproduced in corresponding tissues of methimazole-induced hypothyroid rats. These results suggest that hypothyroid conditions affect the biosynthesis and expression of gangliosides in specific tissue and cell types.

Ganglioside expression during differentiation of chick retinal cells in vitro

The neural retina has been widely used to study the developmental patterns of ganglioside metabolism. Recent findings about in vitro differentiating chick embryo retina cells showed that: a) GD3 and GD1a ganglioside patterns undergo the most dramatic changes; b) when the cells emit neurites, GD3 ganglioside and a group of complex gangliotetraosylgangliosides (GTOG) are transiently coexpressed; c) synchronized developmental phenomena are dissociated by anti-GM1 antibodies; d) GD3 remains as a major ganglioside in differentiated neurons, though it is almost not immunoexpressed; e) GTOG affect antibody binding to GD3; f) the content of gangliosides involved in neural differentiation modifies their immunostain localization on cell membrane; g) after exogenous GTOG uptake, immature neurons mimic GD3 immunofluorescent localization of mature cells; h) a subset of purified retinal ganglion cells express GTOG characteristic of mature neurons.

Survival and process regrowth of purified chick retinal ganglion cells cultured in a growth factor lacking medium at low density. Modulation by extracellular matrix proteins

Panning-purified retinal ganglion cells (RGCs) were cultured at low density in a chemical-defined growth factor (GF)-lacking medium on substrate of different extracellular matrix (ECM) proteins. The process regrowth under these severe conditions were evaluated by morphometric measurements and by cell ELISA (CELISA) performed for neurofilaments regardless of their phosphorylated state (NF-CELISA), or for phosphorylated neurofilaments (PNF-CELISA), to respectively assess process regrowth or axonal development. The development obtained in cultures performed on laminin was taken as standard to refer the other substrata. The cellular content of Thy-1 required for panning purification as well as the gangliotetraosylganglioside (GTOG) expression and the lack of the immunolabeling of the RA4 antigen strongly suggest that the purified RGCs were mature neurons. About 80% of the 7-day-old embryo (E7)-RGCs survived 4 days in culture on any substrate, including polylysine. Conversely, E10-RGCs in about 75% of cultures on polylysine did not survive for 4 days. E7-RGCs developed better on thrombospondin and vitronectin. E10-RGCs cultured on vitronectin grew better than on laminin; on thrombospondin and collagen, E10-RGCs grew like on laminin and on fibronectin they had a poor development. The values of PNF-CELISA obtained on vitronectin, collagen and fibronectin on E7-RGC cultures were significantly higher than on laminin, which are in agreement with the longer processes observed. The flavoridin disintegrin caused a dose-response inhibition on E7-RGC cultures on thrombospondin but not on laminin, suggesting on process regrowth, the integrin-thrombospondin interaction(s) are significantly involved, while on laminin, it is the non-integrin receptor(s) which are significant involved.

Interaction of cholera toxin and Escherichia coli heat-labile enterotoxin with glycoconjugates from rabbit intestinal brush border membranes: relationship with ABH blood group determinants

The capacity of cholera toxin (CT) and type I heat-labile enterotoxin produced by Escherichia coli isolated from human intestine (LTh) to interact with glycoconjugates bearing ABH blood group determinants from rabbit intestinal brush border membranes (BBM) was studied. On the basis of the type of intestinal compounds related to the human ABH blood group antigens, rabbits were classified as AB or H. Toxin binding to the intestinal glycolipids and glycoproteins depends on the blood group determinant borne by the glycoconjugate and on the analyzed toxin. LTh was capable of interacting preferentially with several blood group A- and B-active BBM glycolipids compared to those isolated from animals lacking these antigens (H rabbits). Also, LTh preferably bound to several BBM glycoproteins from AB rabbit intestines compared to those from H ones. One of these glycoproteins, the sucrase-isomaltase complex (EC 3.2.1.48-10) isolated from AB and H rabbits showed the same differential LTh binding. Conversely, CT practically did not recognize either blood group A-, B-, or H-active glycolipids and glycoproteins. These results may be relevant for carrying out in vivo experiments in rabbits in order to disclose the role of ABH active-glycoconjugates in the secretory response induced by LTh in rabbit intestine.

GD3 and GM2 synthase activities in rat testes during the period of sexual development

Activities of two key enzymes of gangliosides biosynthesis were determined in rat testes during development. GD3 synthase activity was low and showed small variations with age. GM2 synthase activity increased 10-fold in testes from 10- to 30-d-old animals, showing a maximum activity at 30 d, followed by a small decrease until 45 d and then a constant activity up to adulthood. These developmental changes in the activity of both glycosyltransferases were related to the increasing complexity in the ganglioside pattern observed in rats testes during the period of sexual development.

Anti-GM1 ganglioside IgM-antibodies present in human plasma: affinity and biological activity changes in a patient with neuropathy

Low affinity anti-GM1 IgM-antibodies are part of the normal repertoire of human plasma antibodies (Mizutamari et al.: J Neuroimmunol 50:215-220, 1994), a fact that is against the pathological role proposed for them in autoimmune diseases. Here we present evidence that these low affinity IgM-antibodies are devoid of complement-mediated lytic activity to GM1-liposomes, suggesting that they should not be considered harmful. In contrast to the absence in normal individuals, in the plasma of a patient with sensory polyneuropathy we detected high affinity anti-GM1 IgM-antibodies. Concomitant with the presence of these high affinity anti-GM1 IgM-antibodies, the patient plasma is capable of producing complement-mediated lysis of GM1-liposomes. These results suggest that an increase in the affinity of the naturally existing anti-GM1 antibodies could be the trigger that switches them from non-harmful to pathological.

Survival and differentiation of purified embryonic chick retinal ganglion cells cultured at low density in a chemically defined medium

Both, a tailored chemically defined nutrient medium (BP5) and a sandwich culture sustain the survival for more than a week and allow the differentiation of embryonic chick retinal ganglion cells (RGCs) seeded at low density. Purification of RGCs from 7-11-day old embryos was accomplished by panning using specific anti-chicken Thy-1 antibodies immobilized in plaques. Yield of RGCs was less than 1% of the calculated number of these cells in the used retinas. This result agrees with the scarce expression of Thy-1 in immature retina; accordingly, the most mature RGCs are those probably selected by the panning. This assumption obtained support on the expression of gangliotetraoxylgangliosides (GTOG), that characterize the differentiated retinal neurons. Thus, the outgrowth of processes observed in cultured cells, might imply axonal regeneration in mature neurons. This manageable RGC culture method approaches a system for studying the in vitro trophic factors and substrata which affect axonal regrowth in central nervous system cells.

Expression of beta 1-4 N-acetylgalactosaminyltransferase gene in the developing rat brain and retina: mRNA, protein immunoreactivity and enzyme activity

The developmental pattern of expression of the UDP-GalNAc:GM3 N-acetylgalactosaminyltransferase (GalNAc-T) gene was examined in the rat brain and retina. A GalNAc-T cDNA cloned from a rat olfactory bulb cDNA library was used as a probe for Northern blot and in situ hybridization experiments and a rabbit polyclonal antibody to rat GalNAc-T peptide was used for Western blot analysis. In Northern blot experiments, a single approximately 3 kb transcript was detected both in brain and retina. In brain, the abundance of this transcript increased from E15 to PN1-5 and then declined while, in retina, it increased steadily from PN1 to PN13-24. The developmental trends of GalNAc-T mRNA expression, GalNAc-T immunoreactive protein and GalNAc-T activity were comparable in brain. In retina, however, GalNAc-T activity and GalNAc-T peptide immunoreactivity followed developmental patterns that were similar between them and different from that of the specific mRNA. Results suggest that post-transcriptional controls of the GalNAc-T gene expression operate in the rat CNS, which are particularly evident in retina. The expression of the GalNAc-T gene in glial and neuronal cells was examined in rat retina cell cultures by in situ hybridization. The GalNAc-T mRNA was abundant in GM1+/GD3+ neurons and almost absent in the flat, GM1-/GD3+ Müller glia-derived cells.

Light exposure stimulates the activity of ganglioside glycosyltransferases of retina ganglion cells

In chicks submitted to light stimulation, the synthesis of gangliosides of the retina ganglion cell increases with respect to chicks maintained in the dark. In an attempt to elucidate if the activation of glycosyltransferases participates in the establishment of these light-dark differences detected in vivo, we examined the activity of a key ganglioside glycosyltransferase, the GalNAc-T (N-acetylgalactosaminyltransferase) that converts GM3 to GM2, in the retina ganglion cells isolated from light and dark exposed chicks. We found that GalNAc-T and other glycosyltransferases are active in these ganglion cell preparations; the kinetic parameters for GalNAc-T were similar to those previously reported for chick retina. The other glycosyltransferase activities assayed were the galactosyltransferase (Gal-T2) that converts GM2 to GM1 and the N-acetylneuraminyltransferase (Sialyl-T1) that converts lactosylceramide to GM3. The three glycosyltransferase activities were higher in the ganglion cell preparations obtained from chicks exposed to light compared to those maintained in the dark. For the GalNAc-T activity, the differences disappear when the cell preparations are sonicated or if the assays are carried out in the presence of detergents or if the end product of the reaction is added to the incubates. The results indicate that the activation of the glycosyltransferases is part of the phenomenon required for cells to achieve the precise rate of synthesis of gangliosides needed in vivo.

Compartmental organization of the synthesis of GM3, GD3, and GM2 in golgi membranes from neural retina cells

The relationship among lactosylceramide-(LacCer), GD3- and GM2-synthases and between the two last transferases and their common GM3 acceptor was investigated in intact Golgi membrane from chick embryo neural retina cells at early (8-days) and late (14 days) stages of the embryonic development. [3H]Gal was incorporated into endogenous glucosylceramide by incubation of Golgi membranes with UDP-[3H]Gal. Conversion of the synthesized [3H]Gal-LacCer into GM3, and of the latter into GD3, GM2 and GD2 was examined after a second incubation step with unlabeled CMP-NeuAc and/or UDP-GalNAc. With CMP-NeuAc, most [3H]Gal-LacCer was converted into GM3 in either 8- or 14- day membranes. However, while about 90% of GM3 was converted into GD3 in 8-day membranes, only about 25% followed this route in 14-day membranes. With CMP-NeuAc and UDP-GalNAc, about 90% of GM3 was used for synthesis of GM2 in 14-day membranes, while in 8-day membranes about 80% followed the route to GD3, and a part to GD2. Performing the second incubation step in the presence of increasing detergent concentrations showed that conversion of GM3 to GM2 was inhibited at concentrations lower than those required for inhibition of LacCer to GM3 conversion. Taken together, results indicate that transfer steps leading to synthesis of GM3, GD3, GM2 and GD2 from LacCer are functionally coupled in the Golgi membranes, and that GD3- and GM2-synthases compete in a common compartment for using a fraction of GM3 as substrate. In this competition, the relative activities of the transferases and their relative saturation with the respective donor sugar nucleotides, are important factors influencing conversion of GM3 toward either GD3 or GM2.

Identification of an endogenous inhibitor of the UDP-N-acetylgalactosamine: GM3, N-acetylgalactosaminyl transferase as apolipoprotein A1

A previously described inhibitor of the UDP-N-acetylgalactosamine: GM3, N-acetylgalactosaminyltransferase (GalNAc-T) (Quiroga et al., 1, 2), was purified from chicken blood serum by a new procedure. When subjected to SDS-PAGE, two major polypeptides of 27 and 70 kDa were observed. When tested in vitro, only the 27 kDa polypeptide inhibited the GalNAc-T. When added to chick cerebral embryonic neurons in culture, both polypeptides inhibited neuritogenesis. Both the 27 kDa and the 70 kDa fractions were present in the cells at 3 h following their addition to the cultures; both polypeptides had aneuritogenic activity and both inhibited the incorporation of [3H]-galactose into the cell gangliosides modifying their labeling pattern to a similar extent. Sequencing of the amino terminal end of the polypeptides showed that 18 and 9 amino acids from, respectively, the 27 and the 70 kDa polypeptides, were 100% homologues with the corresponding region of chick apolipoprotein Al (apo Al). After addition to cells in culture, no interconversion between the two polypeptides was detected after up to 20 h in culture. A monoclonal antibody that recognizes only the 70 kDa polypeptide, blocks its aneuritogenic effect without modifying that of the 27 kDa fraction. It is concluded that the endogenous inhibitor of GalNAc-T is apo Al.

Age-related changes of the ganglioside long-chain base composition in rat cerebellum

The ganglioside mixture from the cerebellum of young, 6 month old and two years old rats, was fractionated by reversed phase high performance liquid chromatography, each ganglioside homogeneous in the oligosaccharide chain as well as in the long-chain base being subsequently quantified. Two long-chain bases, LCB, were components of the five major gangliosides GM1, GD1a, GD1b, GT1b and GQ1b, these being the C18:1 LCB and C20:1 LCB. The content of C20:1 ganglioside molecular species was lower than that of the C18:1 one. In very young animals, day 8, the C20:1 ganglioside species represented about 8% of the total ganglioside content, then they progressively increased and reached, at 2 years, about 42% of the total. C18:1 GD1a and C18:1 GT1b, were the major species in young animals and reached their highest content at day 29, being 1.45 and 1.28 nmol/mg protein, respectively. The content of these two species decreased in adult and old animals and at two years it was 0.71 and 0.82 nmol/mg protein, respectively.

Complex gangliosides affect GD3 accessibility to antibody in developing neuronal cells

Ganglioside expression of embryonic chick retina cells developed in vitro was analyzed by indirect immunofluorescence. Immature neurons were GD3 positive cells and the labeling was chiefly distributed all over their cell membrane. Mature neurons became GD3 negative and expressed complex gangliosides of the a- and b-pathways; nevertheless, the content of GD3 accounted for approximately 40% of the total gangliosides in these cells. Neuraminidase hydrolysis pointed out that GD3 was located in membrane of differentiated cells. The frequency of cells with the GD3 immunostain localized in restricted area of membrane of undifferentiated neurons increased significantly after adding a mixture of bovine brain gangliosides (largely complex gangliosides). Antibody binding to immobilized GD3 showed a dose-dependent inhibition by adding a mixture of bovine brain gangliosides, GM1, GD1a or asialo-GM1. Glycosphingolipids with shorter oligosaccharide chains, as cerebrosides or sulfatides, did not affect this binding. These results suggest that, concomitant with the accretion of content of complex gangliosides, a rearrangement in the membrane would occur, which progressively masks GD3 to its antibody. This rearrangement might affect putative ganglioside functions involved in neuronal differentiation.

Functional coupling of glycosyl transfer steps for synthesis of gangliosides in Golgi membranes from neural retina cells

The synthesis of the oligosaccharide of gangliosides is carried out in the Golgi complex by successive sugar transfers to proper glycolipid acceptors. To examine how the product of one glycosylation step couples with the next transfer step, the endogenous gangliosides of Golgi membranes from 14-day-old chick embryo retina were labeled from CMP-[3H]NeuAc or UDP-[3H]GalNAc or UDP-[3H]Gal in conditions which do not allow vesicular intercompartmental transport. After saturation of the endogenous acceptor capacity, labeling was mostly in the immediate acceptors of the corresponding labeled sugars. However, some labeled intermediates progressed to more glycosylated gangliosides if the membranes were incubated in a second step in the presence of the necessary unlabeled sugar nucleotides. This was particularly evident in the case of membranes incubated with UDP-[3H]Gal, in which most of the [3H]Gal-labeled lactosylceramide synthesized in the first step was converted to GM3 and GD3, or to GM2 or to GD1a in a second incubation step in the presence of unlabeled CMP-NeuAc alone, or together with UDP-GalNAc, or together with UDP-Gal plus UDP-GalNAc, respectively. Conversion was time dependent and dilution-independent. Since prior reports using brefeldin A indicate that transfer steps catalyzed by GalNAc-T, Gal-T2, and Sial-T4 localize in the trans-Golgi network (TGN), our results lead to the following major conclusions: (a) transfer steps catalyzed by GalNAc-T, Gal-T2, and Sial-T4 colocalize and are functionally coupled in the TGN; (b) proximal Golgi Gal-T1, Sial-T1, and Sial-T2, and their corresponding glycolipid acceptors, extend their presence to the TGN, and (c), GalNAc-T and Sial-T2 compete for a common pool of acceptor GM3 in the synthesis of GM2 and GD3.

Gangliosides raise the intracellular Ca2+ level in different cell types

Total gangliosides from bovine brain at micromolar concentration induce intracellular Ca2+ increments in a temperature, time and dose dependent manner when assayed with suspensions of rat macrophages, rat and chicken neurons, human erythrocytes and liposomes, loaded with the fluorescent Ca2+ indicator FURA 2. The effect was independent on the endogenous ganglioside composition of the cells and in the case of neurons it was also independent on the differentiation state. Gangliosides do not induce the release of Ca2+ from inner stores. These findings indicate that the reported inhibition of arachidonic acid release (Bressler, J., et al., (1994) Life Sci., 54, 49-60) and anti-inflammatory properties of gangliosides (Correa, S.G. et al., (1991) Eur. J. Pharmacol. 199, 93-98) are not due to impairments of Ca2+ flux. The results also suggest the possibility that the well-known neurotrophic effect produced by gangliosides on undifferentiated neurons in culture may be due to subtoxic cytosolic Ca2+ increments.

Activities of five different sialyltransferases in fish and rat brains

To investigate the role of sialyltransferases in the metabolism of brain gangliosides, we examined activities of five different sialyltransferases (GM3-, GD3-, GT3-, GD1a-, and GT1a-synthase) using total membrane preparations from cichlid fish and Sprague-Dawley rat brains, and analyzed the relationship between the enzyme activities and the ganglioside compositions. The patterns of sialyltransferase activities in fish and rat brains differed from each other. In fish brain, the GM3-synthase activity was lower than GD3-synthase activity, whereas the opposite relationship was observed in rat brain. The GT3-synthase reaction with fish brain membranes produced radiolabeled GM3, GD3, and a ganglioside that was identified as GT3 based on mobility on TLC using two different solvent systems. No GT3-synthase activity was detected in rat brain. The GD1a- and GT1a-synthase activities in fish brain were higher than those in rat brain. Although GT1a was a single radiolabeled ganglioside in fish GT1a-synthase reaction, this ganglioside could not be detected in rat brain. The ratios of GM3-, GD3-, GT3-, GD1a-, and GT1a-synthase activities in fish and rat brain were 23:31:4:28:14 and 61:21:0:18:0, respectively. Ganglioside analysis showed that fish brain was enriched with c-series gangliosides including GT3 and polysialo-species, whereas a- and b-series gangliosides were major components in rat brain. These results suggest that the species-specific expression of gangliosides in brain tissues may be regulated, at least in part, at the level of sialyltransferase activities.

Differential interaction of Escherichia coli heat-labile toxin and cholera toxin with pig intestinal brush border glycoproteins depending on their ABH and related blood group antigenic determinants

The ability of glycoproteins from pig intestinal brush border membranes (BBM) to bind cholera toxin (CT) or heat-labile toxins from strains of Escherichia coli isolated from human (LTh) or pig (LTp) intestines was studied. Glycoproteins capable of binding the toxins are also recognized by antibodies or lectins specific for ABO(H) blood group and related antigens. Pigs expressing A, H, or I antigenic determinants were used for comparison. The toxin-binding capacity of a glycoprotein depends on the toxin type and the blood group epitope borne by the glycoprotein. LTh and LTp preferably bound to several blood group A-active glycoproteins rather than H-active glycoproteins. By contrast, CT practically did not recognize either blood group A- or blood group H-active glycoproteins, while glycoproteins from pigs expressing I antigenic determinants were able to interact with LTh, LTp, and CT. LTh, LTp, or CT glycoprotein binding was selectively inhibited by specific lectins or monosaccharides. Affinity purification of the toxin binding brush border glycoproteins on the basis of their blood group reactivity suggests that such glycoproteins are hydrolytic enzymes. BBM from A+ pigs contain about 27 times more LTh binding sites, in addition to those recognized by CT, than an equivalent membrane preparation from H+ pigs. The present findings may help clarify some previous unclear results on LTh binding to intestinal BBM glycoproteins obtained by use of animals not typed by their ABO(H) blood group phenotype.

In vitro modulation of changes in ganglioside patterns of differentiating neurons in the presence of an anti-GM1 antibody

Retinal cells from 7-day-old chicken embryos were cultured in the presence of a polyclonal anti-GM1 antibody, at low and high density in a "sandwich cell culture". Cells that were about 80% neurofilament positive at all times, changed their morphology and emitted processes as controls. By examining immunocytochemical expression of gangliosides, cells cultured in the presence of the antibody maintained GD3 expression longer than controls, albeit the expression of the gangliotetraosylgangliosides (GTOG) was not considerably affected. This leads to an extension of the transient period in which differentiating cells coexpressed both types of gangliosides (GD3 and GTOG). At 3-4 days in vitro the relative synthesis of GD3 was about 30% higher and that of GD1a about 40% lower than in controls, indicating a delay in the shift of the synthesis pattern. Nevertheless, the pattern of ganglioside composition resembled at 4 days in vitro. Results indicate that the anti-GM1 antibody may modulate the expression and synthesis of gangliosides without a detectable decrease in neuritogenesis. Considering that the emission of neurites occurs in coexpressing GD3 and GTOG neurons, it is suggested that neuritogenesis could be irrespective of losing the GD3 expression.

Regulation of ganglioside composition and synthesis is different in developing chick retinal pigment epithelium and neural retina

We examined the immunocytochemical expression of GM3 and GD3 in 3-day-old chick embryo retinal pigment epithelium (RPE) and neural retina (NR). We also compared the composition of gangliosides and the activities of key ganglioside glycosyltransferases of the RPE and NR of 8-, 12-, and 15-day old embryos. The immunocytochemical studies in 3-day-old embryos showed heavy expression of GM3 and GD3 at the inner and outer layers of the optic vesicle that are the precursors of the RPE and NR, respectively. The compositional and enzymatic studies showed pronounced differences between RPE and NR of 8-day and older embryos. HPTLC showed that at 8 days the major species were GM3 and GD3 in RPE and GD3 and GT3 in NR. As development proceeded, GD3 decreased in both tissues, GM3 became the major ganglioside in RPE, and ganglio-series gangliosides (mainly GD1a) became the major species in NR. At 15 days the major species were GD1a in NR and GM3 in RPE. Enzyme determinations showed that whereas in RPE from 12-day-old embryos GM2 synthase was under the limit of detection and GD3 synthase activity was about sixfold lower than GM3 synthase, in NR the activities of GM3 and GD3 synthases were similar and both six- to ninefold lower than GM2 synthase. These results evidence a markedly different modulation of the ganglioside glycosylating system in cells of a common origin that through distinct differentiation pathways originate two closely related tissues of the optic system.(ABSTRACT TRUNCATED AT 250 WORDS)

Adult rat retina interneurons synthesize GD3: GD3 expression by these cells is regulated by cell-cell interactions

GD3, a ganglioside of the lactosyl series, is prevalent in rat retina neuronal cells. We studied here whether rat retina neurons synthesize their own surface GD3 or if they acquire it from Müller glia cells. We analyzed the activity of GD3 synthase and the in vivo labeling of gangliosides from N-[3H]acetylmannosamine in adult rat retinas after selective destruction of Müller glia cells with the gliotoxic alpha-D,L-aminoadipate (AAA). Immunostaining of rat retina sections and western blot analysis with an antivimentin antibody confirmed the gliotoxic effect of AAA. Neither GD3 synthase activity nor the in vivo labeling of GD3 and other gangliosides was significantly affected by AAA, indicating that neuronal cells synthesize their own GD3. We next analyzed the regulation of the expression of GD3 by these neurons in culture. About 80% of freshly dissociated cells from retina of 4-day-old rats (R4) immunoexpress surface GD3. After 3 days in dispersed cell culture conditions, GD3 expression was under the limit of detection in 80% of neuronal cells, indicating a failure of these cells to maintain the expression of surface GD3 in these experimental conditions. Most flat Müller glia-derived cells present in these cultures were GD3 positive. Surface GD3 was detected in approximately 60% of neuronal cells dissociated from R4 tissue that was developed in vitro as an organ culture for 3 days. Likewise, approximately 50% of neurites that had grown out from R4 retinal explants within 3 days in culture and whose neuronal character was indicated by immunoexpression of growth-associated protein GAP-43 were GD3 positive. These findings suggest that the tissue organization and/or specific interactions modulate GD3 expression in neuronal cells. Under dispersed-cell culture conditions, c-pathway gangliosides (GQ1c and GT1c), which are built up from the sialylation of GD3 and later completion of the oligosaccharide backbone, were detected in approximately 60% of neuronal cells, suggesting a maintenance of production of GD3 as an intermediate for gangliotetraosyl gangliosides.

Gangliosides and neuronal differentiation

Using the GD3-specific mAb R24 we demonstrate by immunohistochemistry that the first embryonic cells of chicken expressing GD3 represent heavily proliferating cells of mesodermal origin (mesenchymal stem and endothelial cells). At this developmental stage (E1-1.5) neuroectodermal cells of the forming neural tube are not stained by R24 or any other available anti-ganglioside antibodies. These cells of the neural tube start to express GD3 at around E1.5 in parallel with increasing proliferative activity. Likewise proliferating and migrating neuronal crest derivates as well as undifferentiated retinal cells, the forming lens and otic placodes increasingly express GD3 in an organ-specific pattern following the spatiotemporal increase in mitotic activity. Immunostaining of GD1b (mAb D21b) or c-pathway polysialogangliosides (mAb Q211) is not obtained before E2.5, is nervous tissue specific and restricted to "new-born" neurons, which start to migrate and form first neurites. This striking change in ganglioside synthesis and expression also occurs in primary cell cultures (after or without previous Q211-mediated complement kill of neurons) during differentiation of mitotic progenitor cells to neurons (neurogenesis). In cell culture, the fluorescence staining is evenly distributed over the whole neuronal surface including filopodia at the growth cones. Monensin (10(-8) M) prevents expression of GD1b and c-polysialogangliosides and simultaneously differentiation of neuronal morphology (neurogenesis). The presence of exogenous gangliosides from bovine brain leads to a decrease of the monensin effect or even abolishes it.

Immunohistochemical localisation of monoclonal antibody R 24-recognized ganglioside Glac2 in early chick embryos

The spatio-temporal cellular expression and biosynthesis of ganglioside Glac2 was investigated in early chick embryogenesis. For demonstration of embryonic Glac2-biosynthesis, chick embryos of stage 0 and of stages 4-5 were incubated in vitro in the presence of radioactive sugar precursors. It was found that chick embryos synthesize Glac2 as early as at the blastula stage as well as at the gastrula stage, both within the area pellucida and the area opaca. In contrast to the biosynthetical findings immunohistochemical staining of the chick embryos at various stages by aid of the mouse monoclonal antibody (mAb) R 24, specific for the immunoepitope NeuAc alpha, 8NeuAc alpha, 3Gal beta less than, as present on the ganglioside Glac2, revealed a spatio-temporal cellular pattern of expression of this ganglioside in early chick embryos. Immunohistochemical staining of the chick embryo at stage 0 shows that all cells of the embryo, the extraembryonic epiblast and the yolk endoderm included, are mAb R 24-positive. At the intermediate streak stage (stage 3), the cranial part of the deep layer, the so-called endophyll, is strongly mAb R 24-positive, whereas at the end of gastrulation (stage 5), mAb R 24-recognized epitopes appear to be restricted to a narrow band of deep-layer cells in the endophyllic crescent and to the yolk endoderm of the area opaca. At this stage, no labelling by the antibody is observed in cell layers of the future embryo. The beginning of neurulation (stage 7) is characterized by the expression of the mAb R 24-recognized epitope in the notochord, whilst the deep layer in the cranial part of the neural fold still expresses this epitope. No ecto- or mesodermal structures are stained by the antibody at this developmental stage. During further development (stage 12 and 13), mAb R 24-reactivity is restricted to the cranial part of the embryo with a preferential staining of cells of endodermal origin. At these stages, the notochord expresses mAb R 24 binding sites only in its cranial region. The spatial and temporal correlation between the presence of mAb R 24-recognized epitopes and the morphogenetic positioning of tissues may be indicative for a possible role of the ganglioside Glac2 in corresponding cellular interactions.

Developmental changes in gangliosides in cultured cerebellar granule neurons

The content and composition of gangliosides in cultures enriched in granule neurones and in astrocytes from rat cerebellum (P6-8) showed marked differences: astrocytes contained less than 10% of the amount of granule neurones and the profile was dominated by simple gangliosides with lactosyl ceramide backbone, while gangliosides of the 'b' series, which constitute about 40% in nerve cells, were virtually undetectable. Granule cell maturation was accompanied by a 16-fold increase in the ganglioside content during the initial 8 days in a serum-supplemented medium (S+), reaching a plateau much earlier and at a higher level than observed in the cerebellum in vivo. Developmental changes were characterized, as in vivo, by a pronounced decrease in the GD3 proportion and an increase in the 'b' series of gangliosides. Compared with S+, adhesion among cells and fibres is different in a serum-free medium (S-), in which the rise in cellular ganglioside content was less (30%), but the developmental changes in ganglioside profile were similar. However, in cultures in S- only, GM3 was not detectable, while the distribution of GM1 and GD3 indicated that maturation is retarded relative to cells in S+. Surface exposure of gangliosides (studied by the periodate/[3H]borohydride method) was similar under the two culture conditions. There was an initial delay, especially in S-, in the insertion of gangliosides into the plasma membrane, while the labelling of GD3 (the dominant ganglioside of immature granule cells) was very low compared with all the other species throughout the whole cultivation time.

GD3 prevalence in adult rat retina correlates with the maintenance of a high GD3-/GM2-synthase activity ratio throughout development

Unlike neurons from avian retina and other regions of avian and mammalian brain, neurons from mammalian retina not only contain gangliosides of the gangliotetraosyl ceramide series but also maintain a prevalence of GD3, a ganglioside of the lactosylceramide series characteristic of proliferative neural cells, when they are fully differentiated. We show here that GD3 is prevalent at all developmental periods of the rat retina from birth [50% of total gangliosidic N-acetylneuraminic acid (NeuNAc)] to adult (30% of total gangliosidic NeuNAc). GD3-synthase specific activity increased about 1.5-fold from birth to day 7 and essentially plateaued thereafter. The GD3-/GM2-synthase specific activity ratio was compared in rat and chicken retina at early and late developmental stages. In chicken retina the ratio was about 0.7 at early (when GD3 is prevalent) and decreased to 0.07 at late (when GD1a is prevalent) developmental stages. In rat retina the ratio was about 13 and 6 at, respectively, early and late developmental stages. These findings suggest that the prevalence of GD3 and of other "b" pathway gangliosides in adult rat retina neurons could be due in part to the maintenance of a high GD3-/GM2-synthase activity ratio throughout development of the tissue.

Internalization of the inhibitor of the N-acetylgalactosaminyltransferase by chicken embryonic retina cells: reversibility of the inhibitor effects

Retina cells from 6-day-old chicken embryos were cultured in the presence of an 125I-labeled protein inhibitor of the UDP-N-acetylgalactosamine:GM3,N-acetylgalactosaminyltransferase. The cells were labeled and did not lose the incorporated radioactivity when treated with 0.125% trypsin or 1 M NaCl at 37 degrees C for 1 hr, indicating that the iodinated inhibitor was inside the cells. Immunostaining procedures using an anti-inhibitor antibody were applied to the cells cultured in the presence of the inhibitor after permeabilization of the cells. The inhibitor was found inside the round cells virtually devoid of neurites, but not in flat glial-like cells or in process-bearing neural cells. Also found was an apparent self-recovery effect of the cells for both the anti-neuritogenic effect and the modification of the pattern of labeled gangliosides produced by the inhibitor when the agent was withdrawn from the culture medium after the initial period of 20 hr. This recovery was clearly observed 72 hr after the removal of the inhibitor.

GD3 ganglioside is prevalent in fully differentiated neurons from rat retina

Adult mammalian retinas contain unusually high amounts of GD3, a ganglioside of the lactosylceramide series. In this respect, they differ from adult avian retina and other regions of the adult avian and mammalian brain, where GD3 is a minor ganglioside and gangliosides of the gangliotetraosylceramide series (GM1, GD1a, GD1b, GT1b) are the predominant ones. We compare here the ganglioside patterns of rat, human, horse, and guinea pig retinas, which are known to differ in the degree of vascularization and astrocytic cell content. All these retinas showed a prevalence of pathway "b" gangliosides over pathway "a" gangliosides but showed no correlation between GD3 content and the degree of vascularization and astrocytic cell content. Immunostaining of rat retina sections showed the presence of GD3 in the inner and outer plexiform layers and also in the ganglion cell and inner nuclear layers. About 60% of the cells dissociated from rat retina showed immuno-colocalization of GD3 and the neuronal marker class III beta tubulin isotype or cholera toxin binding. All morphologically identifiable glial Muller cells coexpress GD3 and gangliotetraosylgangliosides. GD3 was a minor ganglioside among these axonally transported by ganglion cells in rats and guinea pigs, suggesting that it is either not synthesized by ganglion cells or, if so, it is restricted to the cell soma and/or dendritic tree. Our results demonstrate that, unlike neurons from avian retina and other regions of avian and mammalian brain, neurons from mammalian retina not only contain gangliosides of the gangliotetraosylceramide series but also keep a prevalence of gangliosides of the lactosylceramide series (GD3) when they are fully differentiated.

Gangliosides in lesion-induced synaptogenesis: studies in the hippocampus of the rat brain

Changes in ganglioside composition, biosynthesis and individual distribution were studied in hippocampal regions after unilateral destruction of the entorhinal cortex. After 1 and 3 days postlesion (dpl), a decrease in ganglioside content was detected in area dentata (AD) and pyramidal cell regions CA1-CA3 (CA), both ipsilateral and contralateral to the lesion. By 5 dpl all the values had returned to control values, except in AD which showed a dramatic increase in total ganglioside content reaching a maximum at 12 dpl. By 30 dpl this area also showed control content. A significant increase in biosynthesis of gangliosides was observed at 5 and 8 dpl in the hippocampus ipsilateral to the lesion without changes in the contralateral counterpart. Individual ganglioside distribution showed a pronounced change in GM1 and GQ1b with small changes in the other major gangliosides. Significant differences were observed in the distribution of gangliosides between the two hippocampal regions studied in unoperated control animals. GD1a was more concentrated in AD, whereas GQ1b, GT1b and GD1b predominated in CA. The data presented here indicate that important modifications in ganglioside content as well as pattern occur in the deafferented hippocampus a phenomenon that could be related with the known effect of gangliosides on neuritogenesis observed in cell culture studies.

An endogenous inhibitor of N-acetylgalactosaminyltransferase inhibits retina neuron differentiation in culture

An inhibitor of N-acetylgalactosamine:GM3, N-acetylgalactosaminyltransferase (EC 2.4.1.92) from chicken blood serum, was tested for its activity on embryonic chicken neural retina in culture. The inhibitor did not change the cellular protein content of the cultures but produced a significant reduction of the labeling of gangliosides. The ratio of labeling of GD3 to GD1a increased from about 0.1 to about 0.8 in the cells cultured without or with the inhibitor, respectively. A striking effect of the inhibitor was seen on the morphology of the neurons, those cultured in its presence being practically devoid of neurites. Glial flat cells were apparently not affected.

Coexpression of lactosyl and gangliotetraosyl gangliosides in rat cerebellar radial glial cells in culture

The expression of gangliosides of the lactosylceramide (LC) and of the gangliotetraosylceramide (GTC) series on the surface of cells from rat embryonic cerebellar tissue was investigated by double-color indirect immunofluorescence. GD3 was assumed to be representative of LC and was detected using a specific monoclonal antibody. GM1 was assumed to be representative of GTC and was detected using the binding of cholera toxin followed by the binding of cholera toxin antibodies. The expression of polysialosylated GTC (polysialosyl-GTC) was detected using the cholera toxin-cholera toxin antibody experimental approach after conversion of polysialosyl-GTC to GM1 by treatment of the cells with neuraminidase. To distinguish the major neural cell types present in the cultures the expression of the following cell type-specific markers was investigated: neuron-specific enolase and microtubule-associated protein-2 (MAP-2) as probes for neuronal cells and the intermediate filament protein glial fibrillar acidic protein (GFAP) as a probe for astroglial cells. More than 80% of cells dissociated from cerebellar tissue of 15-day-old rat embryos (E15) are positive for the expression of GD3 and about 50% for the expression of GM1 and polysialosyl-GTC, but most are negative for the expression of neuron-specific enolase, MAP-2, and GFAP. After culturing for 4 days (E15 + 4) most cells that show characteristics of neuronal cells are positive for the expression of polysialosyl-GTC and "inactivate" the expression of GD3. Most cells with characteristics of radial and stellate glial cells are also positive for the expression of polysialosyl-GTC, but unlike neuron-like cells, they do not "inactivate" the expression of GD3.

In vivo and in vitro expression of gangliosides in chick retina Müeller cells

The expression of gangliosides of the lactosylceramide (LC) and of the gangliotetraosylceramide (GTC) series on the surface of cells from the chick neural retina was investigated by double-color indirect immunofluorescence. GD3 was assumed to be representative of LC and was detected using a specific monoclonal antibody. GM1 was assumed to be representative of GTC and was detected using the binding of cholera toxin followed by the binding of cholera toxin antibodies. The expression of polysialosylated GTC (polysialosyl-GTC) was detected using the cholera toxin-cholera toxin antibody experimental approach, after conversion of polysialosyl-GTC to GM1 by treatment of the cells with neuraminidase. In retinas from 6-day-old embryos (R6), most cells (approximately 80%) expressed GD3 but not GTC. After culturing for 7 days, (R6+7), the expression of GTC was found confined to neuron-like cells; flat cells derived from Müller cells expressed GD3 but were negative for GTC expression. On the other hand, postmitotic Müller cells obtained from 13-day-old embryo (R13) or 1-day-old hatched chick retina (RP1) expressed GD3, GM1, and polysialosyl-GTC but were unable to maintain the expression of these GTCs when kept in culture for several days. According to these results, retinal cells can be defined on the basis of their ganglioside expression as follows: (a) retinoblasts, by the expression of GD3; (b) postmitotic neuronal cells, by the expression of GTC; and (c) postmitotic Müller cells, by the expression of GD3 and GTC.(ABSTRACT TRUNCATED AT 250 WORDS)