Optic nerve integrity is required for light to affect retina ganglion cell gangliosides

The labeling of retina ganglion cell and optic tectum gangliosides after an intraocular injection ofN-[³H]acetylmannosamine ([³H]ManNAc) is higher in chickens exposed to light than in those maintained in darkness (1,2). In the present work we studied whether the signal for the higher labeling of ganglion cells in light originates in the photoreceptor layer or comes from the nerve terminal. For this purpose the labeling of ganglion cell gangliosides was determined in light and dark in chickens with one optic nerve severed. The results showed that the effect of light occurred only in the eye normally connected to the optic tectum. In the eye with its optic nerve severed, no difference was observed between the labeling of gangliosides in animals in light and dark, having both groups the labeling values of the normal eyes exposed to light. The results indicate that the information that decreases labeling in darkness or accelerates it in light originates in the nerve terminal.Special Issue dedicated to Prof. Edwardo De Robertis.     

Short term labeling of proteins,gangliosides and glycoproteins of the optic tract of chickens exposed to light or darkness

In contrast with previous findings of the labeling of the glycosidic moieties of the gangliosides and glycoproteins in chickens injected with $\text{N-[}{}^3\text{H}\text{]}\text{acetylmannosamine}$, the labeling of the ganglion cell layer and optic tectum proteins of chicks exposed to light after an intraocular injection of [³H]proline showed no differences with those of their counterpart chickens that remained in darkness. The same failure in finding a difference was met when the cytosolic or the particulate proteins or the acid soluble fraction in the retina were compared.Cycloheximide and puromycin inhibited the labeling of retina and optic tectum proteins, gangliosides and glycoproteins in both illumination conditions. Since the labelings in the optic tectum appeared more inhibited than those in the retina ganglion cell layer it was concluded that cycloheximide and puromycin, besides the synthesis of those compounds, also inhibit their axonal transport.On the basis of these contrasting results the working hypothesis is advanced that light stimulation enhances the activity of the Golgi apparatus but not (or less) that of the polyribosomes.     

Entry of Newly Synthesized Gangliosides into Myelin

Ganglioside synthesis and transport to myelin was studied in brainstem slices prepared from 19-21-day-old rats. The slices were incubated for up to 2 h in the presence of [3H]glucosamine to label primarily the hexosamine portion of complex gangliosides. The amount of radioactivity incorporated into gangliosides during slice incubations was only 10-15% of the amount of the label incorporated during in vivo labeling of brainstem gangliosides using equivalent amounts of [3H]glucosamine. Among individual gangliosides this inhibition was greater for the more complex gangliosides. When labeled gangliosides were isolated from homogenate and myelin fractions prepared from brain slices, the complex total gangliosides of both fractions showed a lag in labeling kinetics but with a lower specific radioactivity for the myelin fraction, reflecting the larger pool size and slower turnover rate exhibited by myelin components. Chase experiments showed that more complex gangliosides in homogenate exhibited almost no effect of chase after 30 min. Addition of the Golgi-disrupting agent monensin to slice incubations inhibited the labeling of all gangliosides except GM3, GM2, and GD3, and transport to myelin of all complex gangliosides except GM2. These results show that a monensin-sensitive mode of transport is responsible for the translocation of most newly synthesized gangliosides into myelin.     

Incorporation of N-Acetylmannosamine into Rat Brain Subcellular Gangliosides: Effect of Pentylenetetrazol-Induced Convulsions on Brain Gangliosides1

Abstract: The labeling pattern of the major individual gangliosides from the microsomal and synaptosomal fractions of rat brain was determined following intracerebral injection of the radioactive sialic acid precursor, N-acetylmannosamine. Microsomal gangliosides initially had a higher specific radioactivity than synaptosomal gangliosides, with both fractions reaching similar specific radioactivities 18 h after precursor injection. In both subcellular fractions, the polysialogangliosides G_T1b and G_Q1b were initially more highly labeled than all other gangliosides. With the establishment of the labeling pattern, the effect of the convulsant pentylenetetrazol on brain gangliosides was examined in detail. Significant decreases in radioactive label were noted in the polysialogangliosides, G_T1b and G_Q1b, from the synaptosomal and microsomal fractions of the convulsed animals. The decreases may be due to activation of the membrane-bound neuraminidase present with the gangliosides in neuronal tissue. Prior to experimentation, a methodology was developed to insure quantitative isolation of small amounts of ganglioside free of other lipids and water-soluble contaminants. Combination of this isolation procedure with quantitative densitometry of thin-layer chromatograms permits accurate distributional analyses for individual gangliosides. In applications involving radioactive gangliosides, the method allows the determination of both radioactivity and sialic acid distributions from the same thin-layer chromatogram.     

N-azidoacetylmannosamine-mediated chemical tagging of gangliosides

Peracetylated N-alpha-azidoacetylmannosamine (Ac(4)ManNAz) is metabolized by cells to CMP-azidosialic acid. It has been demonstrated previously that in this way azidosialic acid-containing glycoproteins are formed that can be labeled on the cell surface by a modified Staudinger ligation. Here, we first demonstrate that the same procedure also results in the formation of azidosialic acid-containing gangliosides. Deoxymannojirimycin, an inhibitor of N-glycan processing in proteins, decreases the total cell surface labeling in Jurkat cells by approximately 25%. Inhibition of ganglioside biosynthesis with N-[5-(adamantan-1-yl-methoxy)-pentyl]1-deoxynojirimycin reduces cell surface labeling by approximately 75%. In conclusion, exposure of cells to Ac(4)ManNAz allows in vivo chemical tagging of gangliosides.     

Synthesis and Translocation of Gangliosides and Glycoproteins During Urethane Anesthesia

In this work, we have studied (a) the contents of gangliosides, glycoproteins, and phospholipids of the vesicle and plasma membrane fractions from brains of anesthetized and control rats and chickens and (b) the labeling of gangliosides and glycoproteins in the retina ganglion cell layer and optic tectum of urethane-anesthetized and control chickens after intraocular injection of a labeled N-acetylneuraminic acid precursor and the distribution of the label after subcellular fractionation. We found an increase in the content of gangliosides relative to protein in the vesicle fraction of both anesthetized rats and chickens relative to their controls. Other values were not affected by anesthesia. These results do not reflect a faster synthesis of gangliosides stimulated by urethane, because their rate of labeling was diminished in anesthetized animals. During the 4-h period after the animals were injected intraocularly with the radioactive precursor, the highest values of ganglioside-specific radioactivity were found in the vesicle fraction of control and anesthetized animals; at longer intervals, the specific radioactivity of the vesicle and plasma membrane fractions became rather similar. These data are in accordance with previous studies from this laboratory suggesting that the synthesis of the carbohydrate chain of gangliosides is regulated by the physiological demands made by the neurotransmitting system.     

Synthesis of Retinal Gangliosides During Chick Embryonic Development

Abstract: Embryonic retina cells incorporated radioactivity from D-[6-³H]glucosamine into gangliosides in vitro. The incorporation was higher in retinas from younger embryos. The pattern of labeling of individual gangliosides of the retina changed gradually from a predominant labeling of gangliosides running chromatographically as GD₃ (nomenclature of Svennerholm) and GM₃ in retinas from 8-day-old embryos to a predominant labeling of those running as GD_Ia and GT, in retinas from 13–18-day-old embryos and newly hatched chicks. The shift in the pattern of labeling correlated with a temporary increase of about sixfold of the activity of UDP-N-acetyl-galactosamine:GM₃ N-acetylgalactosaminyltransferase occurring between days 8 and 14 of embryonic development and with a regular increase of the activity of the UDP-galactose:GM₂ galactosyltransferase occurring from day 8 until hatching. The activities of the CMP-NeuAc:lactosylceramide-, CMP-NeuAc:GM_3-, and CMP-NeuAc:GM₁-sialosyltransferases in the retinas of newly hatched chicks were 40, 20, and 40%(in comparison with the corresponding activities determined in retinas of the 8-day-old embryo.     

Gangliosides associated with microsomal subfractions of brain: Comparison with synaptic plasma membranes

To study ganglioside distribution within subcellular components and test the hypothesis that they are localized at the nerve ending, microsomes and synaptic plasma membranes were isolated from young adult rat brains and compared with respect to ganglioside composition. These were shown to be heterogeneous preparations by fractionation on a discontinuous sucrose gradient into subfractions which had differing ganglioside concentrations. The highest ganglioside concentrations occurred in membranes banding at the 0.8M/1.0M and 1.0M/1.3M interfaces for both microsomes and synaptic plasma membranes. These subfractions had closely similar ganglioside concentrations and pattern distributions. In addition, the kinetics of ganglioside labeling following administration of [³H]-glucosamine were similar for the two preparations. The fact that microsomal subfractions representing heterogeneous mixtures of brain cell membranes showed close similarity to synaptosomal plasma membranes argues against localization of gangliosides at the nerve ending. These results, together with other lines of evidence, support the concept that gangliosides are distributed over large portions of the neuron (and perhaps other brain cells). Data concerning the labeling of gangliosides in different microsomal subfractions indicated a movement of label over time from the more dense to the less dense membranes, as was also noted for the glycoproteins in the same subfractions. Specific radioactivity of the gangliosides increased relative to that of the glycoproteins with time.     

Anabolic sialosylation of gangliosides in situ in rat brain cortical slices

Radiolabeling of the sialic acid residues of gangliosides was examined in thin slices of rat brain cerebral cortex incubated under physiologic conditions in the presence of either [14C]N-acetyl-mannosamine (ManNAc) or cytidine 5'-monophosphoryl-[14C]N-acetyl-neuraminic acid (CMP-NeuAc). CMP-NeuAc is the direct donor substrate in the transfer of sialic acid to gangliosides by sialosyl transferases (SATs), including ectosialosyl transferases at the cell surface. ManNAc must be internalized by the neural cells (neuronal or glial) where it serves as an obligate precursor for the biosynthesis of the NeuAc moiety of intracellular CMP-NeuAc, via multiple reactions in the cytosol and nucleus. When exogenous [14C]ManNAc was supplied, there appeared to be a 2-h lag period before label was incorporated measurably into ganglioside sialic acid. That was followed by rapid ganglioside labeling continuing up to 6 h. There was high incorporation into ganglioside GM1. Labeling by ManNAc was inhibited by monensin, a monovalent cationophore that blocks anabolic transport in medial and trans Golgi. Extracellular CMP-NeuAc was not internalized by the cells. CMP-[14C]NeuAc labeling of gangliosides had no lag period, reached a maximum within 2 h, and then began to level. The label distribution among gangliosides was high in GD3, but quite low in GM1. CMP-NeuAc labeling was not inhibited by 10(-7) M monensin. These findings support a model in which ManNAc labels gangliosides by an intracellular route involving monensin-sensitive, Golgi-associated SATs. In this intracellular system, the major labeled products are gangliosides of the gangliotetraosyl series (GM1, GD1a, etc.).(ABSTRACT TRUNCATED AT 250 WORDS)     

Content and accessibility of sialic acid on the surface of rat hepatocytes during primary culture

The content and accessibility of terminal sialic acid and galactose residues of rat hepatocytes in primary culture were determined by in situ labeling using either periodate or sialidase/galactose oxidase treatment followed by sodium borotritiide reduction. Rat erythrocytes which were used for comparison showed a strongly enhanced tritium incorporation into galactose after sialidase treatment. In contrast, with freshly prepared rat hepatocytes only a small amount of galactose labeling was achieved after sialidase treatment. The amount of galactose labeled following sialidase treatment increased with time in culture up to day 6 and roughly paralleled the increase of the total sialic acid content. Major changes of sialic acid-containing glycoconjugates were restricted to the gangliosides. There was a transient drop in surface labeling of ganglioside-associated sialic acid on the first day in culture. The specific radioactivity of the in situ-tritiated ganglioside-sialic acid also fell by 50% in this period. Between day 2 and 4, there was an increase in gangliosidesialic acid labeling but the specific radioactivity of the sialic acid remained constant. This indicates that newly synthesized gangliosides but not the preexisting ones were accessible to periodate oxidation. The data allow conclusions about turnover and topology of the sialic acid-containing glycolipids.     

pH-dependent changes of ganglioside biosynthesis in neuronal cell culture

Ganglioside biosynthesis was studied in primary cultured murine cerebellar cells after labeling with [14C]galactose. A shift in biosynthesis from "a"-series to "b"-series gangliosides was observed after lowering the pH of the culture medium from 7.4 to 6.2; this effect was fully reversible on changing back to pH 7.4. The observed regulatory effect of pH is in accordance with a recent model of ganglioside biosynthesis. Sialyltransferase II (ST II), the first enzyme for biosynthesis of "b"-series gangliosides, is more active at pH 6.2 than Gal-NAc-transferase, the first enzyme for synthesis of "a"-series gangliosides, which is more active than sialyltransferase II at pH 7.4.     

The role of gangliosides in the interaction of human chorionic gonadotropin and cholera toxin with murine Leydig tumor cells

A clonal line of murine Leydig tumor cells (MLTC-1) bound both human chorionic gonadotropin (hCG) and cholera toxin (CT) with high affinity and accumulated cyclic AMP in response to either effector. The major cellular ganglioside was GM3 with small amounts of GM2, GM1, and GD1a. The gangliosides became labeled when the cells were grown in medium containing [3H] galactose or were exposed to galactose oxidase or NaIO4 followed by NaB3H4. CT specifically protected GM1 from surface labeling whereas hCG did not protect any gangliosides from being labeled. When the cells were exposed to sialidase, surface GD1a was eliminated, and GM1 increased with a corresponding increase in CT binding. When sialidase-treated cells were first incubated with the B component of CT, binding and action of CT was blocked. The cells, however, retained their ability to bind and respond to hCG. Addition of purified gangliosides to the medium effectively inhibited the binding and action of CT but not hCG. The cells incorporated the exogenous gangliosides and exhibited increased binding of and responsiveness to CT but not hCG. Both hCG- and CT-receptor complexes were extracted from the cells with nonionic detergent and analyzed by sucrose gradient centrifugation. The hCG-receptor complex had an apparent molecular weight of 190,000 whereas the CT-receptor complex sedimented only slightly faster than CT itself. MLTC-1 gangliosides were separated on thin layer chromatograms which were overlayed with either iodinated CT or hCG. The toxin bound to a ganglioside corresponding to GM1 whereas the hormone did not bind to any of the gangliosides. When the cells were incubated overnight with hCG, they lost their hCG receptors but exhibited an increase in CT binding and gangliosides. Our results indicate that GM1 is the specific receptor for CT whereas gangliosides are not involved in the binding and action of hCG.     

GM1 Ganglioside Treatment of PC 12 Cells Stimulates Ganglioside, Glycolipid, and Lipid, but Not Glycoprotein Synthesis Independently from the Effects of Nerve Growth Factor

The incorporation of radioactive precursors into gangliosides and other glycolipids, glycoproteins, and total lipids has been studied in rat pheochromocytoma PC12 cells. Starting with the same PC12 cell pool, cultures displaying different degrees of neuritic expression in response to nerve growth factor (NGF) and combinations of serum ganglioside GM1 were produced. Attempts were then made to correlate neuritic regulation with biochemical performances of these cells. NGF stimulates the incorporation of [3H]galactose into gangliosides and other glycolipids and glycoproteins and [14C]acetate into total lipids, regardless of the serum concentration. NGF both increased their initial labeling rates and promoted additional and more extensive labeling from culture day 4 onward. Unexpectedly, exogenous GM1 also elicited an increase in ganglioside labeling as well as that of the other lipid classes, but not of glycoproteins. The GM1-induced increase was evident at higher serum concentrations (1%) regardless of the presence or absence of NGF, but not apparent in low (0.15%) serum. Serum levels themselves did not affect labeling patterns in the absence of NGF and GM1. GM1-induced stimulation of labeling reflects an increase in the synthetic activities of the cells, and not increased precursor uptake or reduced product degradation. For all constituents stimulated by GM1, concurrent treatment with NGF produces cumulative effects, suggesting independent mechanisms of action by the two molecules.     

The organization of gangliosides and other lipid components in synaptosomal plasma membranes and modifying effects of calcium ion

Synaptosomes were prepared from bovine brain by zonal rotor sucrose density centrifugation. While a major fraction of lipid-bound sialic acid is included uniformly within the synaptosomal distribution profile, the sialoglycoproteins and some gangliosides do not follow this pattern, Exposure to extrasynaptosomal calcium results in alterations in the surface labeling properties of some gangliosides and membrane plasmalogens, suggesting that extrasynaptic Ca²⁺ may influence the conformation of complex lipids in synaptic plasma membranes. The level of intrinsic membrane-associated sialidase activity that liberates sialic acid from these sialoglycoconjugates parallels the synaptosomal buoyant density distribution profile, supporting a view that this enzyme resides in synaptosomal membranes in close association with a sialolipid substrate.     

Expression of Mouse Sialic Acid on Gangliosides of a Human Glioma Grown as a Xenograft in SCID Mice

Ganglioside sialic acid content was examined in the U87-MG human glioma grown as cultured cells and as a xenograft in severe combined immunodeficiency (SCID) mice. The cultured cells and the xenograft possessed N-glycolylneuraminic acid (NeuGc)-containing gangliosides, despite the inability of human cells to synthesize NeuGc. Human cells express only N-acetylneuraminic acid (NeuAc)-containing gangliosides, whereas mouse cells express both NeuAc- and NeuGc-containing gangliosides. Small amounts of NeuGc ganglioside sialic acid (2-3% of total ganglioside sialic acid) were detected in the cultured cells, whereas large amounts (66% of total ganglioside sialic acid) were detected in the xenograft. The NeuGc in gangliosides of the cultured cells was derived from gangliosides in the fetal bovine serum of the culture medium, whereas that in the U87-MG xenograft was derived from gangliosides of the SCID host. The chromatographic distribution of U87-MG gangliosides differed markedly between the in vitro and in vivo growth environments. The neutral glycosphingolipids in the U87-MG cells consisted largely of glucosylceramide, galactosylceramide, and lactosylceramide, and their distribution also differed in the two growth environments. Asialo-GM1 (Gg4Cer) was not present in the cultured tumor cells but was expressed in the xenograft, suggesting an origin from infiltrating cells (macrophages) from the SCID host. The infiltration of mouse host cells and the expression of mouse sialic acid on human tumor cell glycoconjugates may alter the biochemical and immunogenic properties of xenografts.     

Formation of Ganglioside GD 1b-Lactone in Rat Brain from Intracisternally Administered GD 1b

The presence of ganglioside GD1b, in lactone form GD1b-L, was ascertained in rat brain. The possible formation of GD1b-L from GD1b in brain was explored by the intracisternal injection of GD1b, 3H-labelled at the level of the terminal galactose. This was followed by recognition of the radioactive gangliosides formed at different times (1, 3, and 7 days) after injection. Whereas at 0 time after injection the only radioactive ganglioside was GD1b, after 1, 3, and 7 days other radioactive gangliosides were also found, thus indicating GD1b penetration into the brain tissue, followed by metabolic processing. Besides GD1b, the following radioactive gangliosides were recognized: GM1 and GM2, derived from GD1b degradation; GT1b, formed by the direct sialylation of GD1b; and GD1b-L, produced by metabolic lactonization. The radioactivity carried by GD1b-L was maximal 3 days after injection; its time course was different from that of the other gangliosides, suggesting that the process of lactonization is separate from that of both degradation and glycosylation. Under the same experimental conditions, some radioactive gangliosides also appeared in the liver, although in much smaller amounts than in brain. Radioactive GD1b-L could not be detected in liver, thus indicating that metabolic lactonization is a tissue- or organ-specific process.     

Modulation of Ganglioside Biosynthesis in Primary Cultured Neurons

Murine cerebellar cells were pulse labeled with [14C]galactose, and the incorporation of radioactivity into gangliosides and neutral glycosphingolipids was examined under different experimental conditions. In the presence of drugs affecting intracellular membrane flow, as well as at 15 degrees C, labeled GlcCer was found to accumulate in the cells, whereas the labeling of higher glycosphingolipids and gangliosides was reduced. Monensin and modulators of the cytoskeleton effectively blocked biosynthesis of the complex gangliosides GM1, GD1a, GD1b, GT1b, and GQ1b, whereas incorporation of radioactivity into neutral glycosphingolipids, such as glucosylceramide and lactosylceramide, as well as GM3, GM2, and GD3 was either increased or unaltered. As monensin has been reported to interfere with the flow of molecules from the cis to the trans stacks of the Golgi apparatus, this result highlights at least one subcompartmentalization of ganglioside biosynthesis within the Golgi system. Inhibitors of energy metabolism affected, predominantly, the biosynthesis of the b-series gangliosides, whereas a reduced temperature (15 degrees C) more effectively blocked incorporation of radiolabel into the a-series gangliosides, a result suggesting the importance of GM3, as the principal branching point, for the regulation of ganglioside biosynthesis.     

Effects of Brefeldin A on Synthesis and Intracellular Transport of Ganglioside GT3 by Chick Embryo Retina Cells

Abstract: Ganglioside GT3 is the precursor of c-series gangliosides. It is synthesized by sialylation of GD3 and is expressed in nervous tissue of birds and mammals at early stages of development. In this study we examined the sub-Golgi location of GT3 synthesis and the mechanism of its transport from the site of synthesis to the plasma membrane in chicken embryo retina cells in culture. Neural retina cells from 10-day-old chick embryo were cultured with [³H]galactose in the absence (control cells) or in the presence of 1 µg/ml brefeldin A (BFA). At the end of the labeling period, the fraction of labeled gangliosides transported to the plasma membrane was determined. For this, cells were treated with C . perfringens neuraminidase in conditions to desialylate only those gangliosides that were transported to the plasma membrane and consequently accessible to the enzyme. After neuraminidase treatment of cells, gangliosides were isolated, purified, and the pattern of radioactivity analyzed by HPTLC-fluorography. It was found that BFA blocked the synthesis of complex gangliosides without affecting the synthesis of GM3, GD3, and GT3. Furthermore, in BFA-treated cells, GM3, GD3, and GT3 were protected from the action of added neuraminidase, indicating an intracellular localization and, hence, an inhibition of their transport to the plasma membrane. The results indicate that synthesis of the first intermediates of a-, b-, and c- series gangliosides occurs in a proximal Golgi compartment and that the proximal Golgi-synthesized gangliosides (GM3, GD3, and GT3) use a transport mechanism that is dependent on ADP ribosylation factor and coatomer proteins.     

Age-Related Changes in the Ceramide Composition of the Major Gangliosides Present in Rat Brain Subcellular Fractions Enriched in Plasma Membranes of Neuronal and Myelin Origin

Abstract: Age-related changes of the ceramide composition of gangliosides were studied in the synaptosomal and myelin fractions from rat brain, carrying plasma membranes of neuronal and glial origin, respectively. The five major gangliosides (GM1, GD1 a, GD1 b, GT1 b, and GQ1 b) present in these fractions were separated and quantitated by normal-phase HPLC. Each ganglioside was then fractionated by reverse-phase HPLC into the molecular species carrying a single long-chain base (LCB). The largely preponderant LCBs in the synaptosomal and myelin fractions were the C18:1 and C20:1. The content of C20.1 LCB, generally low at 1 month, increased with age in all analyzed gangliosides and in all subcellular fractions and was greater in the "b series" than in the "a series" gangliosides. Remarkably, GM1 was the only ganglioside where the proportion of LCB 20:1 was higher in the synaptosomal fraction than in the myelin fraction. The fatty acid composition of the C18:1 or C20:1 LCB species of the different gangliosides in the synaptosomal and myelin fractions did not undergo appreciable changes with age. Stearic acid was largely predominant in all the gangliosides of the synaptosomal fraction, more in the C18:1 than in the C20:1 LCB species (80–90% vs. 60–70%). The gangliosides of the myelin fraction were characterized by a lower content of 18:0 and a much higher content of 16:0 and 18:1 fatty acids than those of the synaptosomal fraction. Thus, the ceramide composition is different in the gangliosides of neuronal and myelin origin and appears to be subjected to an age-related control.     

The adhesion protein TAG-1 has a ganglioside environment in the sphingolipid-enriched membrane domains of neuronal cells in culture

We studied the interactions between gangliosides and proteins at the exoplasmic surface of the sphingolipid-enriched membrane domains by ganglioside photolabeling combined with cell surface biotin labeling. After cell photolabeling with radioactive photoactivable derivatives of GM3, GM1 and GD1b gangliosides, followed by cell surface biotin labeling, sphingolipid-enriched domains were prepared and immunoprecipitated with streptavidin-coupled beads, under experimental conditions preserving the integrity of the lipid domain. About 50% of the total radioactivity linked to proteins was associated with acylated tubulin, about 10% with a 135-kDa protein present as a series of species with pI ranging from 6.5 to 8.0, about 5% with a protein of about 70 kDa and with pI near to 6.5. By immunoprecipitation with streptavidin-coupled beads under conditions disrupting the integrity of the lipid domain, the 135 kDa protein was recovered in the immunoprecipitate, that did not contain tubulin. Thus, the 135 kDa protein has an exoplasmic domain, and it was then identified as the GPI-anchored neural cell adhesion molecule TAG-1. Remarkably, TAG-1 was cross-linked in a similar extent by the photoactivated ganglioside GM3, GM1 and GD1b. The three gangliosides bear different oligosaccharide chains, suggesting that the ganglioside/TAG-1 interaction is not specifically associated with the ganglioside oligosaccharide structure.