Neurochemical Research

, Volume 35, Issue 12, pp 1867–1874

In Search of a Solution to the Sphinx-Like Riddle of GM1


    • Department of Neurology and NeurosciencesNew Jersey Medical School, UMDNJ
  • Gusheng Wu
    • Department of Neurology and NeurosciencesNew Jersey Medical School, UMDNJ

DOI: 10.1007/s11064-010-0286-0

Cite this article as:
Ledeen, R.W. & Wu, G. Neurochem Res (2010) 35: 1867. doi:10.1007/s11064-010-0286-0


Among the many glycoconjugates contributing to the sugar code, gangliosides have drawn special attention owing to their predominance as the major sialoglycoconjugate category within the nervous system. However, their occurrence, albeit at lower levels, appears ubiquitous in vertebrate cells and even some invertebrate tissues. Now that over 100 gangliosides have been structurally characterized, their diverse physiological functions constitute a remaining enigma. This has been especially true of GM1, for which a surprising array of functions has already been revealed. Our current research has focused on two areas of GM1 function: (a) signaling induced in neural and immune cells by cross-linking of GM1 in the plasma membrane that leads to activation of TRPC5 (transient receptor potiential, canonical form 5) channels, a process important in neuritogenesis and autoimmune suppression; (b) activation by GM1 of a sodium-calcium exchanger (NCX) in the inner membrane of the nuclear envelope (NE) with resulting modulation of nuclear and cellular calcium. The latter has a role in maintaining neuronal viability, loss of which renders neurons vulnerable to Ca2+ overload. Pathological manifestations in mutant mice and their cultured neurons lacking GM1 have shown dramatic rescue with a membrane permeable derivative of GM1 that enters the nucleus and restores NCX activity. Nuclear function of GM1 is related to the presence of neuraminidase in the NE, an enzyme that generates GM1 through hydrolysis of GD1a. A different isoform of this enzyme was found in each of the two membranes of the NE.


GM1GangliosideNeuraminidaseCalcium regulationCross-linking of GM1TRPC5 channelsGM1 in the nuclear envelopeSodium-calcium exchanger



B subunit of cholera toxin


Endoplasmic reticulum








Sodium-calcium exchanger


Nuclear envelope


Effector T cell


Regulatory T cell


Transient receptor potential, isoform 5 of canonical subgroup


Acknowledging the special significance of the 35th anniversary of this journal is an occasion for reflection by those of us who were privileged to be active in the field at that time. It was a landmark event in that it contributed significantly to emergence of neurochemistry as a mature and recognized discipline. Neurochemical Research soon came to provide an important vehicle for publication of new findings in this burgeoning field, and will likely remain so in the future.

The true beginning of neurochemistry is generally attributed to J.L.W. Thudichum, known to his peers as “chemist of the brain”, who succeeded in isolating and characterizing a number of sphingolipids from human brains. This included sphingosine itself, which he named to signify the chemical puzzle it presented. Among the more complex sphingolipids subsequently discovered were gangliosides, defined as glycosphingolipids containing one or more sialic acids in the oligosaccharide chain. Pioneering efforts in elucidating such structures have been credited to the groups of Ernst Klenk and Tamio Yamakawa, employing brain and erythrocytes as tissue source, respectively. A ganglioside of primary interest to many research groups including ours has been GM1, so designated by Svennerholm who demonstrated thin-layer chromatographic separation [1]. This coincided in time with structural elucidation by Kuhn and Wiegandt [2]. Our entry to the field consisted of comparison of GM1 in normal brain to that in two neurological disorders, showing identity of structure (Fig. 1) despite major differences in levels of expression [3]. This along with other members of the GM1 (gangliotetraose) family was shown to undergo rapid axonal transport in reaching its destination in the axonal and nerve ending membranes [4]. A widely held view at that time was that neuronal gangliosides existed and functioned primarily at the nerve ending, their presence in neuronal perikarya reflecting their site of synthesis prior to translocation. However, data then coming to light on synaptic density and ganglioside concentration in isolated CNS synaptosomes permitted calculations that suggested only a modest percentage of neuronal gangliosides actually occur at the nerve ending [5]. In due course evidence accumulated from many labs that gangliosides in fact occur over the entire neuronal surface and at a number of intracellular loci as well. Moreover, neurons are by no means unique in this respect as gangliosides appear to be ubiquitous, their presence having been detected in every type of vertebrate cell examined to date. Some invertebrate tissues also express gangliosides but of entirely different structures. All told over 100 gangliosides have been isolated and characterized to date, based on oligosaccharide structure [6], and the number would grow substantially if hydrophobic variations were included. GM1 has gained recognition as a prominent constituent of the cholesterol-enriched microdomains (rafts) of plasma membranes [7] where, along with other sphingolipids, they are thought to contribute to “glycosignaling domains” involved in cell adhesion and a variety of signal induction reactions [8].
Fig. 1

Structure of GM1, GD1a and LIGA-20. Additional gangliotetraose gangliosides are formed by attaching N-acetylneuraminic acid to R1 and/or R2 position(s) in GD1a. LIGA-20 is a semisynthetic analog of GM1 in which the long chain fatty acid of the ceramide unit is replaced by dichloroacetyl [52]

Whereas sphingosine posed an enigma to Thudichum in relation to isolation and characterization, gangliosides today present a somewhat different sphinx-like challenge to neurochemists in relation to physiological function. The efforts of many laboratories in recent years have registered significant progress toward unraveling that mystery. Our studies have focused on the function of GM1 and in recent years we have become impressed with the need to consider intracellular loci in addition to the plasma membrane to obtain a more comprehensive picture. The variety of roles revealed for GM1 has been noteworthy, prominent among these being Ca2+ regulation by a diversity of regulatory mechanisms [9, 10]. As outlined below, these pertain to such processes as neuronal differentiation and immune cell regulation.

GM1 and Neuronal Development

Studies on neuroblastoma cell differentiation, generally an induced process, revealed that elevation of cell surface GM1 could cause significant neurite outgrowth. This could be achieved by exogenous application of GM1 (or other gangliosides) [11] but in more robust manner by application of exogenous neuraminidase (N’ase), an enzyme that converts oligosialogangliosides to GM1 [12]. Similar outgrowth was achieved by Miyagi and coworkers through upregulation of a ganglioside-specific N’ase in Neuro2a cells [13]. This kind of induced differentiation was accompanied by Ca2+ influx and indeed depended on elevation of intracellular Ca2+ [12]. A specific effect for GM1, as opposed to alteration of other sialoglycoconjugates by this enzyme, was suggested in the observation that both neuritogenesis and Ca2+ influx could be blocked by the B subunit of cholera toxin (CtxB) [12], a ligand of relatively high affinity and specificity for GM1. A point of some interest was that not all neuroblastoma cells behaved like Neuro2a, some experiencing enhanced Ca2+ influx and neuritogenesis when exposed to CtxB following N’ase treatment [14]. Although such influx was initially suggested to occur via L-type voltage regulated Ca2+ channels, subsequent work revealed such channels to be voltage independent [15] and to possess the properties of TRPC5 [16], isoform 5 of the canonical subgroup of mammalian genes homologous to the transient receptor potencial family in Drosophila [17, 18].

Reaction of CtxB as above proved to involve cross-linking of plasma membrane GM1, facilitated by the five subunits within the CtxB multivalent ligand and also N’ase which elevates cell surface GM1. That study provided the first indication of the mechanism by which TRPC5 Ca2+ channels are activated. It also revealed that GM1 affects this change through a signaling sequence beginning with cross-linking of α5β1 integrin with which GM1 is associated in the membrane (Fig. 2A). This result demonstrated that while GM1 is intimately associated with α5β1 integrin, it is not directly associated with TRPC5. Activation of this channel by GM1 cross-linking occurred “at a distance” through a signaling sequence involving autophosphorylation of focal adhesion kinase followed by activation of phospholipase Cγ and phosphoinositide-3 kinase. In addition to undifferentiated neuroblastoma cells, this mechanism was observed in primary neurons such as cerebellar granular neurons prior to differentiation; TRPC5 was detected at 2 days in vitro in such cells, a stage corresponding to retained capacity for CtxB-stimulated Ca2+ influx that was lost following differentiation. TRPC5 was shown to be down regulated following this induced differentiation in both cerebellar granule neurons and NG108-15 cells (Fig. 2B). Neuritogenesis of primary neurons was accelerated by CtxB and suppressed by TRPC5 siRNA (Fig. 3). The crucial role of GM1 was indicated with neurons from GM1-null mice, in which growth of axons was significantly retarded (Fig. 3). While such differentiation induced by CtxB might appear artificial, the validity of the GM1 cross-linking phenomenon was vindicated by the observation that galectin-1 (Gal-1), a homodimeric lectin produced by astrocytes, induces GM1 cross-linking and stimulates neuritogenesis in a manner similar to CtxB [19]. This is consistent with the description of Gal-1 as a major receptor for GM1 in human neuroblastoma cells [20]. It is of interest that GM1-Gal-1 interaction of similar nature occurs as a key feature in regulatory T cell (Treg) suppression of autoreactive effector T cells (Teffs) (see below).
Fig. 2

A Colocalization of GM1 with α5 and β1 integrins in neuronal plasma membranes. Undifferentiated NG108-15 cells treated with N’ase were incubated with CtxB–FITC in Ca2+ free buffer at 4°C for 15 min, then at 37°C for an additional 15 min to cross-link GM1. The cells were fixed, then immuno-stained for TRPC5, α5- or β1 integrin (second antibodies linked to Texas Red). Confocal images showed GM1 (a, d, g) in the plasma membrane was enhanced at loci that suggested sprouting. TRPC5 (b) was expressed throughout the cell body with a small portion in the plasma membrane that was distinct from GM1 (c). In contrast, α5 integrin (e) and β1 integrin (h) were recruited into same membrane regions as GM1 after cross-linking (f, i). B Expression of GM1 and TRPC5 in primary cerebellar granule neurons (CGN) and NG108-15 cells (NG). CGNs were grown 2- and 5 DIV (ad), and NGs 6 DIV in differentiating medium (eh). Both cells were co-stained with CtxB–FITC and anti-TRPC5 antibody (secondary antibody linked to Texas Red) as shown. TRPC5 in less differentiated CGN (b) and NG108-15 (f) cells was abundant in cell bodies and depleted following differentiation. GM1 was expressed in membrane and soma and increased with differentiation. Adapted from Wu et al. [16] with permission from J. Neuroscience
Fig. 3

Effect of CtxB on axon formation in cerebellar granule neurons cultured from normal and ganglio-series null (KO) mice. Cultures prepared from normal (a, b, e–g) and KO (c, d) mice were grown for 48 h (3 DIV) in the presence (b, d–g) or absence (a, c) of CtxB. Cells in e and g were treated with TRPC5-specific siRNA. Phase images are shown of living cells (a–e). Fluorescent images with SMI-31 mAb are shown in f, g. Retarded spontaneous axon growth was seen in KO cells compared with normal cells (c vs. a). Axon formation was enhanced by CtxB in normal but not KO cells (b vs. d) and inhibited by TRPC5-specific siRNA (e vs. a, g vs. f). Adapted from Wu et al. [16] with permission from J. Neuroscience

GM1 Cross-Linking by Galectin-1 in Autoimmune Suppression

It has been recognized for some time that GM1 is present in cells of the immune system and is intimately involved in regulatory mechanisms, but mechanistic details have been lacking. Some aspects of such involvement are now coming to light with revelation of their role in Treg-Teff interaction [21]. This relates to their presence in autoreactive T cells that escaped thymic deletion and that lead to autoimmune disorders when not effectively suppressed. Such suppression is known to depend to a large extent on the Treg population that constitutively expresses CD4, CD25, and the Foxp3 forkhead box transcription factor [2224]. A characteristic of Tregs is dramatic elevation and release of Gal-1 upon T cell receptor activation [21, 24] and, as mentioned, the homodimeric property of Gal-1 enables it to induce cross-linking of GM1 in the membrane of Teffs. It is such GM1-Gal-1 interaction that suppresses Teff proliferation, such suppression being further enhanced by significant upregulation of GM1 in Teffs following T cell receptor activation [21, 25]. Galectin-1 is known to react with glycoproteins as well as GM1 in various cell types [26, 27], whereas our evidence indicated preferential interaction with GM1 in the case of Teffs [21].

In vivo support for the above mechanism involving GM1 cross-linking came in the demonstration that experimental autoimmune encephalitis was strongly inhibited by both CtxB and Gal-1 [21], confirming earlier studies (in the case of Gal-1) [28, 29]. This parallels similar findings with other autoimmune conditions such as type 1 diabetes, which was inhibited in the NOD mouse by both Gal-1 [30] and CtxB [31]. Those findings are consistent with the general phenomenon of immune modulation by the cholera-like enterotoxins [32]. The proposed role for GM1 provides a rationale for its potency in suppressing disease onset in a variety of animal models of autoimmunity and explains why GM1-null mice proved highly susceptible to EAE [21]. The latter study thus provided evidence suggesting a similar signaling sequence as revealed for neuroblastroma cells, based on intimate association of GM1 with α5β1 (as well as α4β1) integrin as prelude to TRPC5 Ca2+ channel activation.

Nuclear GM1 as Regulator of Calcium Homeostasis

Considering the plethora of information acquired over many years on the structure, metabolism and function of gangliosides in the plasma membrane, it is perhaps surprising that we are only at an early stage of similar understanding about gangliosides of the nucleus. Earlier studies indicated they are indeed present in this organelle, the nuclear envelope (NE) being the primary locus [3335]. GM1 was clearly demonstrated at this site in our study employing CtxB linked to horseradish peroxidase as cytochemical indicator [36]. Its presence in the nucleus of neural cells was confirmed and amplified in a developmental study of rat brain [37]. Since the NE consists of a double membrane, we attempted a more specific localization utilizing a procedure involving mild treatment with sodium citrate solution to selectively remove and isolate the outer membrane [38]. The inner membrane was obtained from the resulting nucleus remnant. Both membranes were found to contain GM1 along with its ganglioside precursor, GD1a (Fig. 1 for structure) as the principle gangliosides [39]. A concurrent finding was that GM1 existed in tight association with a Na+/Ca2+ exchanger (NCX), a protein not previously reported as a NE constituent. Such tight binding was not seen between GM1 and plasma membrane NCX, although there was an indication of looser association. Further probing of this high affinity interaction revealed it as essential for efficient functioning of NCX [39].

Further localization of nuclear NCX, employing the above mentioned procedure [38], established the inner membrane of the NE as the locus of the NCX/GM1 complex. Thus it is strategically situated to mediate Ca2+ transfer from nucleoplasm to NE lumen, and such transfer was demonstrated in vitro with 45Ca2+ and isolated nuclei [39]. Since Ca2+ transfer by this NCX is driven by a Na+ gradient, this required elevation of Na+ in the NE which was accomplished by pre-incubating the nuclei in Na+-containing medium in the presence of appropriate ionophores. Such Na+ transfer is believed to occur naturally by a Na+/K+-ATPase reported to occur in the inner membrane of the NE and to create high intra-lumenal Na+ concentration [40]. Additional evidence for NCX-mediated Ca2+ transfer to the NE was obtained with living cells employing cameleon-fluorescent Ca2+ indicators genetically targeted to the NE/endoplasmic reticulum (ER) and nucleoplasm [41, 42]. Using various cell lines we demonstrated that cells containing both GM1 and NCX in the NE transferred Ca2+ readily from nucleoplasm to the NE and contiguous ER network, in contrast to cells lacking either molecule which transferred little or no Ca2+ [43]. Because of facile Ca2+ movement from cytosol to nucleoplasm via nuclear pore complexes, those results suggested nuclear NCX/GM1 as alternative mechanism to the SERCA pump for cytosol to ER transfer of Ca2+. Several isoforms of NCX are known to exist [44, 45] and we have speculated that specific forms with greater abundance of positively-charged amino acids are the ones that predominate in the NE [46].

Immunoprecipitation of NCX indicated that in addition to GM1, GD1a also occurs in association, although this ganglioside did not potentiate the exchanger. We now have evidence that GD1a can serve as metabolic reserve for GM1 in the NE, undergoing conversion to the latter by N’ase present in the NE. This enzyme was originally detected in the intact NE with activity toward gangliosides [47], and was subsequently shown by our group to occur in both membranes of the NE [48]. Immunohistochemical analysis of the separated membranes indicated occurrence of Neu3 in the inner- and Neu1 in the outer membrane.

Cytoprotective function of nuclear NCX/GM1 complex

It was reasonable to postulate that the NCX/GM1 complex serves a cytoprotective role in shielding the nucleus against prolonged elevation of cytosolic Ca2+, which would equilibrate with nucleoplasm via the nuclear pore complexes. Calcium in excess is well known to promote apoptosis [49], the nucleus with its Ca2+ activated nucleases and proteases being particularly vulnerable. This protective function was demonstrated in vivo using a knockout mouse lacking the GM1 family of gangliosides [50; GalNAcT−/−]. Such mice, when administered kainic acid, developed temporal lobe seizures of significantly greater severity and duration than normal mice [51]. Attempted rescue with GM1 (intraperitoneal administration) gave limited results but was highly effective with LIGA-20, a membrane permeable analog of GM1 developed by Costa and coworkers [52; Fig. 1 for structure]. This enhanced efficacy correlated with the ability of LIGA-20 to cross the blood brain barrier, enter neurons, insert into the NE, and activate the subnormally active NCX of the NE (Fig. 4). LIGA-20, with identical oligosaccharide structure as GM1, thus served as functional replacement for the latter. Additional evidence came in study of neuronal cell cultures from the above knockout mouse wherein LIGA-20 restored Ca2+ homeostasis and effectively rescued neurons otherwise vulnerable to high K+ and excitotoxic levels of glutamate [53].
Fig. 4

Influence of nuclear NCX/GM1 complex on seizure activity. a Persistence of seizure activity several hours after kainate (KA) injection in knockout (KO) mouse lacking GM1 family (erect, “piano player” posture) compared to wild type (WT) and heterozygous (HT) mice that recovered within 3–4 h. b Young adult mice, 13–15 weeks of age, were injected IP with kainate (KA) at 25 or 30 mg/kg; highest seizure score attained by each group within 3 h post-injection is indicated. KO mice were significantly more susceptible to KA-induced seizures, as indicated in higher severity and mortality. c Rescue comparison of GM1 and LIGA-20: Mice were injected IP with GM1 or LIGA-20 2 h before KA (25 mg/kg). GM1 (30 mg/kg) marginally attenuated seizures in KO mice, whereas LIGA-20 (2.5 mg/kg) reduced seizure activity to 3.5 h, similar to WT. d Potentiation of NCX by LIGA-20 in nuclei from KO brain. Brain nuclei were isolated and Na-dependent 45Ca2+ uptake into NE lumen determined. Exchange activity of KO nuclei was significantly less than that of WT, but was elevated to near normal level in nuclei from KO mice treated with LIGA-20; GM1 had no effect (not shown). Citric acid-treated nuclei which were unable to sequester 45Ca2+ in NE lumen due to removal of outer membrane of NE showed no NCX activity. Adapted from Wu et al. [51] with permission from J. Neuroscience


Understanding the sugar code, purportedly more complex in some respects than the genetic code, is today a major undertaking in biology [54]. GM1 is a miniscule part of nature’s carbohydrate complexity but is already known to mediate a surprising array of functional roles in numerous cell types. Analogous to the nuclear NCX/GM1 complex, GM1 in the plasma membrane binds tightly to the NGF receptor (Trk A) and regulates receptor function [55]. Another plasma membrane role is that of opioid receptor modulator, GM1 effecting conversion from inhibitory to excitatory mode [56]. Intracellular roles for GM1 are drawing more attention, as in the discovery of its association with α-synuclein that promotes α-helical structure and prevents pathological fibrillation [57], a hallmark of Parkinson’s disease. GM1 present in mitochondria-associated ER membranes was described as influencing Ca2+ flux between the ER and mitochondria [58], consistent with an earlier study showing GM1 inhibition of Ca2+-ATPase in sarcoplasmic reticulum membrane [59]. Several other enzymes are known to be inhibited by GM1 but non-specifically and at relatively high concentrations, thus questioning physiological significance in those cases. GM1 is one of the very few sialoglycoconjugates in nature whose sialic acid is resistant to co-localized N’ase, thereby providing a mechanism for its elevation through hydrolysis of associated oligosialogangliosides (principally GD1a). The fact that this relatively small molecule can fulfill a large and growing list of regulatory functions poses a question as to its general attributes that facilitate such functional diversity. While much progress has been made, some miles would seem to remain on the road to more complete solution of this sphinx-like riddle.


Supported by grants from the NIH and NMSS.

Copyright information

© Springer Science+Business Media, LLC 2010