Developmental changes of glycosphingolipids and expression of glycogenes in mouse brains


Address correspondence and reprint requests to Robert K. Yu, PhD Med.Sc.D, Institute of Molecular Medicine and Genetics and Institute of Neuroscience, Medical College of Georgia, 1120 15th Street, Augusta, GA 30912, USA. E-mail:


Glycosphingolipids (GSLs) and their sialic acid-containing derivatives, gangliosides, are important cellular components and are abundant in the nervous system. They are known to undergo dramatic changes during brain development. However, knowledge on the mechanisms underlying their qualitative and qualitative changes is still fragmentary. In this investigation, we have provided a detailed study on the developmental changes of the expression patterns of GSLs, GM3, GM1, GD3, GD1a, GD2, GD1b, GT1b, GQ1b, A2B5 antigens (c-series gangliosides such as GT3 and GQ1c), Chol-1α (GT1aα and GQ1bα), glucosylceramide, galactosylceramide (O1 antigen), sulfatide (O4 antigen), stage-specific embryonic antigen-1 (Lewis x) glycolipids, and human natural killer-1 glycolipid (sulfoglucuronosyl paragloboside) in developing mouse brains [embryonic day 12 (E12) to adult]. In E12–E14 brains, GD3 was a predominant ganglioside. After E16, the concentrations of GD3 and GM3 markedly decreased, and the concentrations of a-series gangliosides, such as GD1a, increased. GT3, glucosylceramide, and stage-specific embryonic antigen-1 were expressed in embryonic brains. Human natural killer-1 glycolipid was expressed transiently in embryonic brains. On the other hand, Chol-1α, galactosylceramide, and sulfatide were exclusively found after birth. To provide a better understanding of the metabolic basis for these changes, we analyzed glycogene expression patterns in the developing brains and found that GSL expression is regulated primarily by glycosyltransferases, and not by glycosidases. In parallel studies using primary neural precursor cells in culture as a tool for studying developmental events, dramatic changes in ganglioside and glycosyltransferase gene expression were also detected in neurons induced to differentiate from neural precursor cells, including the expression of GD3, followed by up-regulation of complex a- and b-series gangliosides. These changes in cell culture systems resemble that occurring in brain. We conclude that the dramatic changes in GSL pattern and content can serve as useful markers in neural development and that these changes are regulated primarily at the level of glycosyltransferase gene expression.

Abbreviations used

basic fibroblast growth factor






cerebroside sulfotransferase or sulfatide synthase


embryonic day




fucosyltransferase 9








β1–4 N-acetylgalactosaminyltransferase












ceramide glucosyltransferase


glucuronic acid




human natural killer-1




microtubule-associated protein 2


lysosomal sialidase


cytosolic sialidase


plasma membrane sialidase


mitochondrial sialidase


neural precursor cell


post-natal day


sulfoglucuronosyl lactosaminyl paragloboside


sulfated glucuronyl paragloboside


stage-specific embryonic antigen-1





Glycosphingolipids (GSLs) are lipid molecules composed of ceramide (Cer) and one or more carbohydrate units (Hakomori 1990; Yu et al. 2007). GSLs containing one or more sialic acid residues in the carbohydrate moiety are referred to as gangliosides. Although the functional roles of GSLs, including gangliosides, are not fully elucidated yet, it has been suggested that these molecules play important functions as mediators of cell adhesion/signal transduction and cell type-specific marker molecules (Hakomori 1990, 2003; Ledeen and Wu 2002; Yu et al. 2004).

In vertebrate cells, GSLs, including gangliosides, are localized primarily on the plasma membrane and represent important constituents of cell-surface microdomains (known as GSL-enriched microdomains, caveolae, or lipid rafts) with cholesterol and sphingomyelin (Hakomori et al. 1998; Anderson 1998; Simons and Toomre 2000). These specialized microdomain structures have been suggested to be important for modulation of cell adhesion and signal transduction. GSLs also have been implicated to have functional relevance in mediating signal transduction and cell–cell recognition and adhesion in the microdomains. Furthermore, in the developing brain, gangliosides are presumed to modulate Cer-induced apoptosis and to maintain cellular survival and differentiation (Bieberich et al. 2001). It is also known that some serious diseases develop in animals or humans with aberrant GSL metabolism. For example, massive accumulation of certain GSLs, including gangliosides, in late endosomes and lysosomes by defects of the lysosomal glycosidases causes lysosomal storage diseases such as GM1 gangliosidosis, Tay–Sachs disease, Fabry disease, and other similar lysosomal storage disorders (Kolter and Sandhoff 2006). Simpson et al. (2004) found that human autosomal recessive infantile-onset symptomatic epilepsy syndrome is associated with a non-sense mutation of sialyltransferase-I (ST-I; GM3-synthase), a key enzyme for the synthesis of complex gangliosides.

Glycosphingolipids are found in virtually all vertebrate cells and bodily fluids but are particularly abundant in the nervous system. The expression patterns of GSLs, in particular gangliosides, are frequently and drastically changed during development (Yu et al. 1988; Yu 1994; Bouvier and Seyfried 1989). Thus, GSLs are considered to be useful as lineage specific and differentiation markers of cells. For example, O4 antigen (sulfatide), O1 antigen (galactosylceramide; GalCer), A2B5 antigens (c-series gangliosides), and GD3 have been used as markers of neural lineage cells (Yanagisawa and Yu 2007). However, the molecular mechanisms regulating the expression of GSLs and gangliosides during development is not fully elucidated, although most of glycogenes such as the genes for glycosyltransferase and glycosidase have been cloned in recent years (Zeng and Yu 2007).

In this study, we systematically analyzed expression patterns of GSLs, especially gangliosides (Fig. 1), in the developing mouse brains between embryonic day 12 (E12) and adult. We further analyzed the developmental expression of glycogenes, including glycosyltransferase and glycosidase genes (Fig. 1) that are expected to regulate the expression of GSLs.

Figure 1.

 Structures and metabolic pathways of GSLs. ARSA, arylsulfatase A; β-gal, lysosomal acid β-galactosidase; Cer, ceramide; CST, cerebroside sulfotransferase (sulfatide synthase); GALC, galactosylceramidase; GalNAc-T, N-acetylgalactosaminyltransferase I (GA2/GM2/GD2/GT2-synthase); GalT-I, galactosyltransferase I (lactosylceramide synthase); GalT-II, galactosyltransferase II (GA1/GM1/GD1b/GT1c-synthase); GalT-III, galactosyltransferase III (galactosylceramide synthase); GLCC, glucosylceramidase; GlcT, glucosyltransferase (glucosylceramide synthase); GM2A, GM2 activator protein; HEX, β-N-acetylhexosaminidase; SA, sialic acid; SAP, saposin; ST-I, sialyltransferase I (GM3-synthase); ST-II, sialyltransferase II (GD3-synthase); ST-III, sialyltransferase III (GT3-synthase); ST-IV; sialyltransferase IV (GM1b/GD1a/GT1b/GQ1c-synthase); ST-V, sialyltransferase V (GD1c/GT1a/GQ1b/GP1c-synthase); ST-VII, sialyltransferase VII (GD1α/GT1aα/GQ1bα/GP1cα-synthase).

Materials and methods

Western blot analysis

Proteins and GSLs were extracted from brains (telencephalons) of ICR mice. Mice used in this study were treated according to the guidelines of the Institutional Animal Care and Use Committee of the Medical College of Georgia. The telencephalons were carefully prepared from mice [E12, E14, E16, E18, post-natal day 1 (P1), P10, and adult] by dissecting in Hanks’ balanced salt solution using fine forceps under the stereo microscope.

For protein analysis, mouse brains were homogenized in a buffer containing 1% Triton X-100, 10 mmol/L Tris–HCl (pH 7.4), 150 mmol/L NaCl, 5 mmol/L EDTA, and a protease inhibitor cocktail (Sigma, St Louis, MO, USA) using a loose fitting homogenizer. After clearing the homogenates by centrifugation at 21 000 g for 10 min, the protein concentration was measured by a Protein Assay Kit (Bio-Rad Laboratories, Hercules, CA, USA). The homogenates containing the same amount of proteins were applied to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (4–20% polyacrylamide gel) and transferred to polyvinylidene difluoride membrane (Bio-Rad Laboratories). Western blot analysis was performed using monoclonal antibodies against nestin (BD Biosciences, San Jose, CA, USA), microtubule-associated protein 2 (MAP2) (Sigma), and β-actin (Sigma). Horseradish peroxidase-conjugated anti-mouse IgG antibody (Amersham Biosciences, Uppsala, Sweden) was used as the secondary antibody.

Glycosphingolipids purification

Extraction and purification of GSLs from mouse brains were performed based on a procedure developed previously in our laboratory (Ledeen and Yu 1982; Ariga et al. 1988). In brief, total lipids were extracted from 200 to 400 mg (wet weight) of mouse brains with chloroform/methanol (C/M; 1 : 2, 1 : 1, and 2 : 1, v/v, consecutively), and the combined extracts were adjusted to the ratio of C/M/water (C/M/W; 30 : 60 : 8, v/v/v). Neutral lipid, acidic lipid, and sulfated glucuronyl paragloboside (SGPG) fractions were separated by diethylaminoethyl Sephadex A-25 (Sigma, acetate form; 1.5 mL bed volume) column chromatography with C/M/W (30 : 60 : 8, v/v/v), C/M/0.8 mol/L sodium acetate in water (30 : 60 : 8, v/v/v) and C/M/1.6 mol/L sodium acetate in water (30 : 60 : 8, v/v/v), respectively (Ariga et al. 1987). After separation, each fraction was desalted using a Sep-Pak C-18 column (Waters Corporation, Milford, MA, USA) (Kubo and Hoshi 1985).

The total sialic acid content in each of the ganglioside fractions was estimated by the method of Svennerholm (1957) with some modifications. In brief, an aliquot of the ganglioside fraction was treated with 0.5 mL of the resorcinol–HCl reagent for 15 min at 100°C. After vigorously mixed with 0.75 mL of butyl acetate/t-butyl alcohol (85 : 15, v/v) and centrifuged at 400 g for 5 min, the absorbance of supernatants was measured using a spectrophotometer at 580 nm. Sialic acid concentrations were calculated based on a standard curve generated from known concentrations (1, 2, 3, 4, and 6 μg) of sialic acid.

TLC and TLC-immunostaining

Glycosphingolipids were separated on a high-performance TLC plate (silica gel 60; Merck Chemicals, Darmstadt, Germany) in a solvent system of C/M/0.2% CaCl2 in water (55 : 45 : 10, v/v/v) for gangliosides, C/M/W (65 : 25 : 3, v/v/v) for cerebrosides, or C/M/W (60 : 35 : 8, v/v/v) for other neutral GSLs. To efficiently separate glucosylceramide (GlcCer) and GalCer, samples were applied onto a TLC plate that had been impregnated with 1% sodium borate in advance. After development, gangliosides and neutral GSLs were visualized by spraying with the resorcinol–HCl and orcinol–H2SO4 reagents, respectively, followed by heating on a hot plate. Densitometric analysis was performed on a Windows computer using Scion Image, the Windows version of NIH Image developed by Scion Corporation (Frederick, MD, USA) available at

For TLC-immunostaining, the TLC plate after development was coated in n-hexane containing 0.4% polyisobutylmethacrylate (Sigma) for 1 min. After drying, the plate was incubated with a primary antibody diluted with 1% bovine serum albumin in phosphate-buffered saline at 20–25°C for 2 h. For primary antibodies, A2B5 (anti-c-series gangliosides monoclonal antibody, mouse IgM; Eisenbarth et al. 1979), GGR-41 (anti-Chol-1α monoclonal antibody, mouse IgG; Kusunoki et al. 1993), AK97 [anti-stage-specific embryonic antigen-1 (SSEA-1) monoclonal antibody, mouse IgM; Yanagisawa et al. 1999], or the serum from a patient with peripheral neuropathy and IgM monoclonal gammopathy (LT serum; anti-human natural killer-1 (HNK-1) human IgM; Ariga et al. 1987) were used. The TLC plate was then incubated with a horseradish peroxidase-conjugated secondary antibody directed to mouse IgM (Jackson ImmunoResearch, West Grove, PA, USA), mouse IgG (Amersham Biosciences), or human IgM (Amersham Biosciences) diluted with 1% bovine serum albumin in phosphate-buffered saline at 37°C for 1 h. GSL bands reacted with the antibodies were detected using the WesternLightning western blot chemiluminescence reagent plus (Perkin Elmer Life and Analytical Sciences, Boston, MA, USA). After the chemiluminescence reaction, the TLC plates were washed with chloroform and then sprayed with the resorcinol–HCl and orcinol–H2SO4 reagents for visualizing gangliosides or neutral GSLs, respectively.

Reverse transcription-polymerase chain reaction

Total RNA samples were isolated from mouse brains using the Trizol reagent (Invitrogen, Carlsbad, CA, USA). cDNAs were synthesized from the total RNAs as templates using Superscript III reverse transcriptase (Invitrogen). PCR was performed using JumpStart™ REDTaq® (Sigma) with the following settings: 94°C for 2 min; 26–32 cycles of 94°C for 20 s, 60°C for 20 s, and 72°C for 0.5–1 min; and 72°C for 2 min. To detect ST-II and Sox2, dimethylsulfoxide was added to the reaction mixture (final concentration, 10%). Primer sets used for PCR are shown in Table 1. PCR products were analyzed by agarose gel electrophoresis using 1.5–2% agarose gels containing SYBR Safe™ DNA Gel stain (Invitrogen).

Table 1.   Sequences of primer sets used for detection of glycogenes
AbbreviationEnzyme nameAccession numberReferenceSequence of primerPCR product
GlcTGlcCer-synthase; Ceramide glucosyltransferaseD89866Ichikawa et al. 19985′-TGCATTTCATGTCCATCATCTAC-3′
412 bp
GalT-ILacCer-synthase; UDP-Gal, GlcCer β1–4 galactosyltransferaseAF097158 5′-TCTACTTCATCTATGTGGCTCC-3′
321 bp
ST-I (ST3Gal-V)GM3-synthase; CMP-NeuAc, LacCer α2–3 sialyltransferaseAF119416Kapitonov et al. 19995′-TTTGGAGTCTGGCTCCTGTAC-3′
343 bp
ST-II (ST8Sia-I)GD3-synthase; CMP-NeuAc, GM3 α2–8 sialyltransferaseX84235Kono et al. 19965′-ATGCTAGCTCGGAAATTCCCG-3′
214 bp
ST-III (ST8Sia-III)GT3-synthase; CMP-NeuAc, GD3 α2–8 sialyltransferaseX80502Kono et al. 19965′-TCTTCACCACTCCCAAGTACG-3′
413 bp
GalNAcTGA2/GM2/GD2/GT2-synthase; UDP-GalNAc, LacCer/GM3/GD3/GT3
β1–4 N-acetylgalactosaminyltransferase
U18975Sango et al. 19955′-ATCAAGGAGCAAGTGGTGGAG-3′
217 bp
GalT-IIGA1/GM1/GD1b/GT1c-synthase; UDP-Gal, GA2/GM2/GD2/GT2
β1–3 galactosaminyltransferase
AF082504Daniotti et al. 19995′-TTGATTTCTAACTCTCATGCCTG-3′
390 bp
ST-IV (ST3Gal-II)GM1b/GD1a/GT1b/GQ1c-synthase; CMP-NeuAc, GA1/GM1/GD1b/GT1c
α2–3 sialyltransferase
X76989Lee et al. 19945′-CATGGCTACCTTGCCCTACC-3′
321 bp
ST-V/III (ST8Sia-V)GD1c/GT1a/GQ1b/GT3-synthase; CMP-NeuAc, GM1b/GD1a/GT1b/GD3
α2–8 sialyltransferase
X98014Kono et al. 19965′-AAGGAGATCAACAGCGCTGAC-3′
305 bp
ST-VII (ST6GalNAc-VI)GD1α/GT1aα/GQ1bα/GP1cα-synthase; CMP-NeuAc, GM1b/GD1a/GT1b/GQ1c
α2–6 sialyltransferase
AB035123Okajima et al. 20005′-GCGGTCAGCAGTGTTTGTGAT-3′
372 bp
GalT-IIIGalCer-synthase; UDP-Gal, Ceramide galactosyltransferaseX92122Bosio et al. 19965′-CTGCAGAGGTGGGTAAGTGG-3′
214 bp
CSTSulfatide synthase; Cerebroside sulfotransferaseBC026806 5′-TTTCCTATTGCTGCTGTACTCC-3′
315 bp
GlcAT-PUDP-glucuronyltransferase-PAB055781Yamamoto et al. 2002a5′-GGGTAATGAGGAGCTGTGGG-3′
392 bp
GlcAT-SUDP-glucuronyltransferase-SAB055902Imiya et al. 20025′-ATGAAGTCCGCGCTGTGCAG-3′
428 bp
HNK-1 STHNK-1 sulfotransferaseAF360543 5′-ATGTTCATGGTGGCCAGCAAG-3′
383 bp
FUT9α1–3 fucosyltransferase IXAB015426Kudo et al. 19985′-CAACATCCAAAGGCATTCTTCG-3′
416 bp
β-GalLysosomal acid β-galatosidaseM57734Nanba and Suzuki 19905′-TGATGTGGAGCATTTCATCCAG-3′
443 bp
HEXαβ-N-acetylhexosaminidase α-subunitX64331Beccari et al. 19925′-TTCCAGTTCCGGTACCATGTC-3′
458 bp
HEXββ-N-acetylhexosaminidase β-subunitY00964Bapat et al. 19885′-TACAAGAGACATCATGGCCCTG-3′
486 bp
Neu1Lysosomal sialidaseY11412Carrillo et al. 19975′-AAGTTCATCGCCATGAGGAGG-3′
470 bp
Neu2Cytosolic sialidaseAB028023Hasegawa et al. 20005′-GCTTACAGAATCCCTGCTCTG-3′
488 bp
Neu3Plasma membrane sialidaseAB026842Hasegawa et al. 20005′-ACTGATGGAGGCCACATTACC-3′
422 bp
Neu4Mitochondrial sialidaseNM_173772Comelli et al. 20035′-AGCACTCTGGTACCATCTTCC-3′
445 bp
GLCCGlucosylceramidase, GlucocerebrosidaseM24119O’Neill et al. 19895′-CGGTATCTTGGGCATATGGTG-3′
461 bp
GALCGalactosylceramidase, GalactocerebrosidaseD38557Sakai et al. 19965′-CCCTATCGTTCTGAAATACTGG-3′
431 bp
ARSAArylsulfatase AX73231Kreysing et al. 19945′-TGGACTACGGTTCACAGATTTC-3′
325 bp
PSAPProsaposin, Sphingolipid activator protein precursorS36200Tsuda et al. 19925′-TGTCCAAGACCCGAAGACATG-3′
418 bp
GM2AGM2 activator proteinL19526Bellachioma et al. 19935′-ATGAAGGAAAGGACCCTGCAG-3′
448 bp
SOX2 X94127Collignon et al. 19965′-ATGTATAACATGATGGAGACGGAGC-3′
960 bp
G3PDHGlyceraldehyde-3-phosphate dehydrogenaseM32599 5′-ACCACAGTCCATGCCATCAC-3′
452 bp

Neural precursor cell culture

Neural precursor cells (NPCs; neuroepithelial cells) were prepared from telencephalons of ICR mouse embryos (E14.5) as previously described (Nakashima et al. 1999a). The neuroepithelial cells were cultured in N2-supplemented Dulbecco’s modified Eagle medium/F12 medium containing 10 ng/mL of basic fibroblast growth factor (bFGF; Peprotech, Rocky Hill, NJ, USA) on dishes that had been pre-coated with poly-l-ornithine (Sigma) and bovine fibronectin (Sigma). Neuronal differentiation was induced by culturing the cell in the absence of bFGF for 10 days. Post-mitotic neurons were selected by suppressing proliferation of glial cells with 0.5 μmol/L of cytosine β-d-arabinoside (Sigma) based on a method by Mitoma et al. (1998).


Major ganglioside expression in the developing brain

It has been well known that the ganglioside composition of the nervous system undergoes remarkable changes during development. For instance, a shift from synthesis of the simplest gangliosides of the a- and b-series to synthesis of the more complex gangliosides during rat development has been reported (Yu et al. 1988) (Fig. 1). To confirm the shift during development, we analyzed the contents and composition of major gangliosides (GM3, GM1, GD3, GD1a, GD2, GD1b, GT1b, and GQ1b; Ganglioside nomenclature is based on that of Svennerholm 1963) in the developing mouse brain (E12, E14, E16, E18, P1, P10, and adult). In these brains, a drastic change was found in expression of marker proteins such as nestin (neural stem/precursor cell marker), MAP2ab (post-mitotic neuron marker), and MAP2c (immature neuron marker) (Fig. 2a). The ganglioside content in the mouse brain drastically increased as development proceeded (Fig. 2b). As with the contents, the composition of gangliosides also exhibited dramatic changes (Fig. 2c and d). In mouse brains at the mid-embryonic stages (E12 and E14) which expressed nestin, GD3 was predominant. In addition, GM3 and GT1b were also robustly expressed. After E16, the content of GD3 and GM3 markedly decreased, and in contrast the content of a-series gangliosides such as GD1a and GM1 increased. The latter was accompanied by an increase in MAP2ab expression (Fig. 2a). Interestingly, at all stages, the expression of GT1b was sustained.

Figure 2.

 Marker protein and ganglioside expression in developing mouse brains. (a) Nestin (neural stem/precursor cell marker), MAP2ab (post-mitotic neuron marker), MAP2c (immature neuron marker), and β-actin (control) expressed in mouse brains were analyzed by western blot. (b) Total sialic acid contents in the ganglioside fractions prepared from the mouse brains were quantified using the resorcinol–HCl reagent method of Svennerholm. (c) TLC of gangliosides prepared from mouse brains. Chloroform/methanol/0.2% CaCl2 in water (55 : 45 : 10, v/v/v) was used as the developing solvent system. Gangliosides developed on TLC plates were stained with the resorcinol–HCl reagent. Lane 1, bovine brain gangliosides (major: GM1, GD1a, GD1b, and GT1b); 2, E12 mouse brain gangliosides; 3, E14 mouse brain gangliosides; 4, E16 mouse brain gangliosides; 5, E18 mouse brain gangliosides; 6, P1 mouse brain gangliosides; 7, P10 mouse brain gangliosides; 8, adult mouse brain gangliosides; 9, authentic GM2 and GQ1b; 10, authentic GM3, GD3, and GD2. (d) The sialic acid content in each of the ganglioside fractions was quantified by densitometric analysis.

To elucidate the molecular mechanism underlying the ganglioside expression shift during development, we analyzed glycosyltransferase expression in developing brains. The ganglioside compositional change during brain development is considered to be strictly regulated by the activities of two key glycosyltransferases, ST-II (GD3-synthase) and N-acetylgalactosaminyltransferase (GalNAcT, GM2/GD2-synthase), which are at a branching point of the ganglioside biosynthetic pathways (Fig. 1). As shown in Fig. 3, most of the glycosyltransferases catalyzing ganglioside synthesis such as ceramide glucosyltransferase (GlcT), galactosyltransferase-I (GalT-I), ST-I, ST-III, GalT-II, and ST-IV showed no significant difference in expression pattern during development. Surprisingly, ST-II expression did not significantly decrease during the embryonic stage except in adulthood. In contrast, GalNAcT expression increased at the mid-embryonic stages in a similar manner as GD1a expression.

Figure 3.

 Glycosyltransferases expressed in developing mouse brains were analyzed by RT-PCR using specific primer sets. GlcAT-P, UDP-glucuronyltransferase-P; GlcAT-S, UDP-glucuronyltransferase-S; HNK-1, human natural killer-1; HNK-1 ST, HNK-1 sulfotransferase; FUT9, α1–3 fucosyltransferase IX. Sox2 (a transcription factor expressed in immature neural cells such as neural stem/precursor cells) and G3PDH (a ubiquitously expressed enzyme catalyzing a step of glycolysis) were detected as controls. −RT indicates negative controls without reverse transcription.

Minor ganglioside expression in developing mouse brains

Certain minor gangliosides are known to have interesting properties. However, sometimes it is difficult to detect them by TLC with chemical visualization because of their low amount or the overlapping Rf values with major gangliosides. Therefore, we analyzed the expression patterns of these gangliosides, including A2B5 antigens and Chol-1α, by TLC-immunostaining.

A2B5 antigens (specific for the c-series gangliosides including GT3, GT1c, and GQ1c) are well-known marker molecules of rat glial precursor cells, O-2A progenitor cells, which differentiate into type-2 astrocytes and oligodendrocytes (Raff et al. 1983; Zhang 2001). GT3 and a more complex ganglioside, possibly GQ1c, were detected in the developing mouse brains using A2B5 antibody (Fig. 4a and b). While GQ1c was sustained in brains between E12 and adult, the expression of GT3 significantly decreased during development, as reported by Hirabayashi et al. (1989). However, the expression of ST-V/III and ST-III, possible GT3 synthases, was not consistent with GT3 expression: ST-V/III and ST-III increased and was sustained during development, respectively (Fig. 3). In addition to ST-V/III (ST8Sia-V) and ST-III (ST8Sia-III), ST-II (ST8Sia-I) also has been reported to catalyze GT3 synthesis (Nakayama et al. 1996). GT3 is expected to be synthesized in a more complex manner and is catalyzed by multiple STs and probably GalNAcT (see Discussion).

Figure 4.

 Minor ganglioside expression in developing mouse brains. Gangliosides prepared from the developing mouse brains were subjected to TLC-immunostaining with A2B5 anti-c-series ganglioside antibody (a) and GGR-41 anti-Chol-1α (GT1aα and GQ1bα) antibody (c), and then visualized with the resorcinol–HCl reagent (b and d). Gangliosides for A2B5 staining were treated with NaOH before TLC.

Chol-1α is one of the cholinergic-specific antigens and has been suggested to play a role in cholinergic synaptic transmission (Ando et al. 2004). Two antigenic substances with Chol-1α activity have been identified as GT1aα (Ando et al. 1992) and GQ1bα (Hirabayashi et al. 1992), belonging to the α-series gangliosides having a sialic acid attached to an N-acetylgalactosamine residue (Nakamura et al. 1988). In mouse brains, these antigens were detected only in P10 and adult mouse brains, but not in embryonic brains (Fig. 4c and d). ST-VII (ST6GalNAc-VI), which catalyzes the synthesis of GT1aα and GQ1bα, has been cloned (Okajima et al. 2000). This enzyme has been shown to have the GT1aα- and GQ1bα-synthase activity, but the gene expression pattern during development has not been compared with those of GT1aα and GQ1bα. Using RT-PCR, the expression of this enzyme was found to increase in the P10 and adult mouse brains; this is consistent with the expression of GT1aα and GQ1bα. This finding supports the conclusion by Okajima et al. (2000) that the synthesis of GT1aα and GQ1bα is regulated by ST-VII.

Cerebroside expression in the developing brains

As described above, A2B5 antigens are well-known marker molecules of precursor cells of astrocytes and oligodendrocytes. As oligodendrocyte development proceeds, GSLs other than the A2B5 antigens emerge on the cell surface. These GSLs include sulfatide (O4 antigen; sulfated GalCer) and GalCer (O1 antigen), useful markers to define immature and mature oligodendrocytes, respectively (Zhang 2001). Next, we analyzed expression of these antigens and another cerebroside, GlcCer, in the developing brains. In mouse brains at the mid-embryonic stages (E12 and E14), GlcCer, but not GalCer or sulfatide, is predominantly expressed (Fig. 5a). The expression of GlcCer decreased during development, and then GalCer and sulfatide became predominant in adult mouse brains (Fig. 5a and b) as reported by Dasgupta et al. (1997). Expression of GlcT (GlcCer-synthase) and GalT-I (LacCer-synthase) is sustained, but expression of GalT-III (GalCer-synthase) and cerebroside sulfotransferase (or sulfatide synthase; CST) increased as development proceeds (Fig. 3). Therefore, it is thought that a shift of cerebroside expression to GalCer and sulfatide after birth is attributed to the increase of GalT-III expression.

Figure 5.

 Neutral and sulfated GSL expression in developing mouse brains. (a) Neutral GSLs prepared from the developing mouse brains were developed on a TLC plate that had been impregnated with 1% sodium borate. The developing solvent system was C/M/W (60 : 35 : 8, v/v/v). Neutral GSLs developed on TLC plates were stained with the orcinol–H2SO4 reagent. An asterisk indicates a non-specific band. (b) Acidic GSLs prepared from developing mouse brains were developed on a TLC plate and stained with the orcinol–H2SO4 reagent. (c–f) Neutral GSLs prepared from developing mouse brains were subjected to TLC-immunostaining with AK97 anti-SSEA-1 antibody (c) and LT serum (e), and then visualized with the orcinol–H2SO4 reagent (d and f).

SSEA-1 and HNK-1 glycolipid expression in the developing brains

Stage-specific embryonic antigen-1 is a carbohydrate antigen originally reported for undifferentiated, but not differentiated, mouse embryonic carcinoma cells (Solter and Knowles 1978, 1979; Knowles et al. 1980). Recently, this antigen has been used as a marker of immature cells such as mouse embryonic stem cells and neural stem/precursor cells (Muramatsu and Muramatsu 2004; Yanagisawa and Yu 2007). The carbohydrate epitope, Galβ1–4(Fucα1–3)GlcNAcβ-, is carried by proteoglycans, glycoproteins, and glycolipids. In this study, we analyzed the expression of the SSEA-1 glycolipids in the developing mouse brains. As analyzed by TLC-immunostaining, a neutral GSL reactive with anti-SSEA-1 antibody was found in embryonic brains (Fig. 5c and d). Based on the Rf value, this component is consistent with the following glycolipid structure: Galβ1–4(Fucα1–3)GlcNAcβ1–3Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ1–1′Cer. The expression of the SSEA-1 glycolipid decreased after birth as reported by Dasgupta et al. (1996). However, the expression of a key enzyme for SSEA-1 synthesis, fucosyltransferase 9 (FUT9), was sustained even after birth (Fig. 3). The reason why the expression of SSEA-1 and FUT9 is not consistent is unknown. One possibility is competition of the precursor, lactosaminyl paragloboside (Galβ1–4GlcNAcβ1–3Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ1–1′Cer), with other glycosyltransferases. For example, a GSL having the HNK-1 epitope, termed sulfoglucuronosyl lactosaminyl paragloboside (SGLPG), is known to be derived from lactosaminyl paragloboside. The HNK-1 antigen is a carbohydrate antigen whose structure has been established as HSO3-3GlcUAβ1–3Galβ1–4GlcNAc- (Chou et al. 1986; Ariga et al. 1987). This structure is carried not only by glycoproteins and proteoglycans, but also by GSLs such as SGPG (HSO3-3GlcAβ1–3Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ1–1′Cer) and SGLPG (HSO3-3GlcAβ1–3Galβ1–4GlcNAcβ1–3Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ1–1′Cer) (Chou et al. 1986; Ariga et al. 1987). In the developing mouse brains, SGPG, but not SGLPG, was transiently expressed at the mid-embryonic stage (Fig. 5e and f). This expression pattern is similar to that of SGPG in rat brains reported by Chou et al. (1991). Key enzymes catalyzing HNK-1 synthesis such as UDP-glucuronyltransferase (GlcAT) and HNK-1 sulfotransferase have been characterized. At least two GlcATs, GlcAT-P and GlcAT-S, are known. Because genetically modified mice deficient in GlcAT-P lack the GlcAT activity toward paragloboside (Yamamoto et al. 2002b), GlcAT-P is presumed to be the enzyme synthesizing SGPG and SGLPG. In mouse brains, however, GlcAT-P expression did not show significant developmental regulation. In contrast, GlcAT-S showed a transient expression pattern from E14 to E18 (Fig. 3). For this reason, GlcAT-S is presumed to be involved in the transient expression of SGPG in developing mouse embryonic brains. It also suggests that the expression patterns of GlcATs are not competitive with that of FUT9. The molecular mechanism underlying the decrease of SSEA-1 is still unknown.

Glycosidase gene expression in the developing brains

Glycosphingolipids are degraded primarily by glycosidases in the lysosomal compartment. A defect of the lysosomal glycosidases is the major cause underlying lysosomal storage diseases with massive accumulation of certain glycolipids, including gangliosides, in late endosomes and lysosomes. Therefore, not only the synthetic pathway but also the catabolic pathway is expected to affect ganglioside expression (Fig. 1). To evaluate the involvement of glycosidases in ganglioside expression during development, we analyzed the expression of several glycosidases, including lysosomal acid β-galactosidase, β-N-acetylhexosaminidase α- and β-subunits, Neu1 (lysosomal sialidase), Neu2 (cytosolic sialidase), Neu3 (plasma membrane sialidase), Neu4 (mitochondrial sialidase), glucosylceramidase, galactosylceramidase, and arylsulfatase A in mouse brains.

We found that glycosidases other than Neu2 and Neu4 were similarly expressed in normal mouse embryonic brains from E12 to birth (Fig. 6). Neu2 and Neu4 were robustly expressed after birth. These results suggest that glycosidases are not involved in ganglioside compositional changes in the embryonic brains, as we predicted earlier (Yu et al. 2004). However, the catabolism of GSLs is not only carried out by glycosidases but may involve the participation of co-factors, such as GM2 activator protein (for β-galactosidase and β-hexosaminidase) and saposin-A, -B, and -C (for β-galactosidase, β-hexosaminidase, sialidase, glucosylceramidase, galactosylceramidase, and arylsulfatase A). Analysis of expression of these co-factors revealed that GM2 activator protein and prosaposin (a precursor protein of saposin-A, -B, -C, and -D) also do not exhibit significant developmental changes in expression patterns (Fig. 6). Our data thus suggest that glycosyltransferases may regulate the amounts of GSLs but not their patterns in the developing mouse brains. However, there is a possibility that the contribution of glycosyltransferases to the GSL amounts and/or patterns might also be dependent on the mouse strain (see Discussion).

Figure 6.

 Glycosidases expressed in the developing mouse brains were analyzed by RT-PCR using specific primer sets. Neu1, lysosomal sialidase; Neu2, cytosolic sialidase; Neu3, plasma membrane sialidase; Neu4; mitochondrial sialidase; PSAP, prosaposin (sphingolipid activator protein precursor).

Ganglioside expression in mouse neural precursor cells and NPC-derived neurons

During development, neurons and glial cells (astrocytes and oligodendrocytes) are generated from common progenitor cells, the neural stem cells and the more restricted NPCs. The expression of GD2 and GD3 has been reported in mouse NPCs (Klassen et al. 2001; Yanagisawa et al. 2004, 2005). To confirm whether ganglioside expression patterns in NPCs change progressively during development, we analyzed gangliosides and glycosyltransferases in mouse embryonic NPCs and their progeny.

The NPCs used in this study were neuroepithelial cells prepared from telencephalons of ICR mouse embryos (E14) according to the method by Nakashima et al. (1999a). These cells are proliferative, express neural stem/precursor cell markers such as nestin, and are capable of differentiating into neurons or glial cells (Nakashima et al. 1999b, 2001; Yu and Yanagisawa 2007). In this investigation, post-mitotic neurons were derived from the NPCs by culturing in the absence of bFGF and selected with cytosine β-d-arabinoside. These derived neurons exhibited a neuronal morphology with long processes (Fig. 7a) and the increased expression of neuronal marker genes (data not shown). Analysis of the total lipid fractions prepared from the NPCs and neurons by TLC revealed that GD3 was a predominant ganglioside of the NPCs (Fig. 7b) as found in mid-embryonic mouse brains. After induction of neuronal differentiation, a drastic increase of complex gangliosides, especially GD1a, was found (Fig. 7b). Interestingly, an increase of GalNAcT gene expression was also detected in the neurons derived from NPCs, consistent with the ganglioside pattern change in developing mouse brains (Fig. 7c). This result suggests that increase of GalNAcT and GD1a found in developing brains is accompanied with neuronal differentiation.

Figure 7.

 Ganglioside expression in primary NPCs and neurons. (a) Phase views of mouse embryonic NPCs and neurons derived from NPCs by culturing in the absence of basic fibroblast growth factor for 10 days (Neurons). Neurons were treated with 0.5 μmol/L of cytosine β-d-arabinoside to suppress proliferation of glial cells. (b) Total lipid fractions prepared from NPCs and neurons derived from NPCs were analyzed by TLC. Gangliosides developed on a TLC plate were detected with the resorcinol–HCl reagent. Lane 1, bovine brain gangliosides (GM1, GD1a, GD1b, and GT1b); 2, authentic GD3; 3, total lipid fractions prepared from mouse NPCs; 4, total lipid fractions prepared from neurons derived from NPCs. (c) ST-II and GalNAcT expressed in NPCs and the neurons were analyzed by RT-PCR. Sox2 and G3PDH were detected as controls. −RT indicates negative controls without reverse transcription. (d) Expression of Chol-1α antigens in the neural precursor cells was analyzed by TLC-immunostaining. GSLs developed on a TLC plate were detected with the orcinol–H2SO4 reagent (left panel) or immunostained with GGR-41 anti-Chol-1α antibody (right panel). Lane 1, bovine brain gangliosides; 2, total lipid fractions prepared from mouse NPCs; 3, authentic GT1aα; 4, authentic GQ1b.

In addition to the major ganglioside, GD3, minor gangliosides expressed in the NPCs were analyzed by TLC-immunostaining. Although Chol-1α antigens, GT1aα and GQ1bα, were not detected in E14 mouse brains (Fig. 4c), NPCs prepared from E14 brains were found to express both gangliosides (Fig. 7d). This specific expression pattern of Chol-1α suggests the usefulness of this antigen as a marker of embryonic NPCs.


In this study, we analyzed expression of GSLs, including gangliosides, and glycogenes in mouse brains during development. The expression patterns of most of the GSLs changed drastically during development. For instance, GD3 and GM3 are predominantly expressed in mid-embryonic brains but their expression is markedly decreased during later development when a- and b-series gangliosides such as GD1a and GT1b are increased. Although GT3, GlcCer, HNK-1, and SSEA-1 glycolipids were expressed in embryonic brains, Chol-1α, GalCer, and sulfatide were found exclusively after birth. Analyses of the expression pattern of glycogenes, including genes encoding glycosyltransferases and glycosidases, revealed that the expression of certain, but not all, glycosyltransferases exhibited different patterns during development as did the GSLs. In some cases, the expression of glycosyltransferases such as ST-VII, GalT-III, and GlcAT-S exhibited consistent patterns with those of the GSL products such as Chol-1α, GalCer, and SGPG, respectively. In contrast, the expression patterns of glycosidases did not exhibit significant change during development. Therefore, it is thought that the GSL expression patterns in the developing mouse brains are regulated mainly by glycosyltransferases, but not by glycosidases. On the other hand, a previous study suggested that the GM1 amount increased in embryos of DBA/2 mice, exhibiting reduced β-galactosidase activity (Bouvier and Seyfried 1990). We do not have a satisfactory explanation as to this anomaly. As ganglioside patterns can vary from one strain of mouse to another (Seyfried et al. 1979), there is a distinct possibility that strain difference may contribute to the discrepancy.

In developing brains, the ganglioside expression patterns have been well known to shift from the simplest gangliosides belonging to the b-series to the more complex gangliosides (Yu et al. 1988). In this study, we could detect the expression shift of gangliosides; during development, GD3 drastically decreased, while GD1a gradually increased. However, the expression level of ST-II (GD3-synthase) did not exhibit a concomitant decrease during development, as reported by Ishii et al. (2007). On the other hand, the expression of GalNAcT (GD1a-synthase) was significantly increased, as reported by Yamamoto et al. (1996). The up-regulation of GD1a and GalNAcT during development was also reproduced in neurons derived from mouse primary NPCs. These results strongly suggest that a shift of ganglioside expression during development (GD3 to GD1a) is caused by induction of GalNAcT expression, but not by ST-II. In fact, although it is expected that all b-series gangliosides should be down-regulated if ST-II expression and/or activity actually decrease as development proceeds, mice at the late embryonic to the adult stages still abundantly expressed a b-series ganglioside, GT1b (Fig. 2). Furthermore, in GalNAcT knockout mice, which are expected to express only GM3 if ST-II is down-regulated during development, more GD3 than GM3 was reported to be expressed (Liu et al. 1999). Therefore, in mouse brains at the late embryonic to the adult stages, GD3 is thought to be persistently expressed but is converted immediately by GalNAcT to downstream gangliosides, such as GD1b and GT1b (Fig. 8). It is possible that the GalNAcT increase is also involved in GT3 expression. During development, GT3 expression decreased, although the expression of GT3 synthases (ST-V/III, ST-III, and ST-II) did not. On the other hand, expression of GQ1c, a downstream ganglioside of GT3, was sustained during development. Similar to GD3, GT3 might also be converted to downstream gangliosides such as GQ1c by the action of GalNAcT immediately after its synthesis in late embryonic, post-natal, and adult brains. So far, a human GalNAcT promoter has already been defined (Furukawa et al. 1996). It will be interesting to elucidate the molecular mechanism regulating GalNAcT expression at the transcriptional level during development, which is in progress.

Figure 8.

 A model of the molecular mechanism underlying a ganglioside expression shift during development. During development, GalNAcT becomes dominant when compared with ST-II (GD3-synthase). As a result, more GD1a than GD3 is synthesized in late embryonic, post-natal, and adult brains.

In addition to the shift in ganglioside expression, a shift in cerebroside expression during development is also similarly regulated. As reported by Dasgupta et al. (1997), cerebrosides, GlcCer, and GalCer, exhibit complementary expression patterns. While the expression of GlcT (GlcCer-synthase) is sustained, the expression of GalT-III (GalCer-synthase) is drastically increased after birth. Although the reason why GlcCer decreased in the absence of a concomitant decrease of GlcT or increase of GalT-I (LacCer-synthase) is unclear, it is expected that the exclusive expression of GalCer after birth is induced by the increased GalT-III. On the other hand, it has been reported that an increase of sulfatide and GM4 was found in mouse embryos treated with a GlcCer synthesis inhibitor, N-butyldeoxygalactonojirimycin (Brigande et al. 1998). Therefore, a low level of CST and GalT-III when compared with GlcT might be expressed in the mouse embryos, although no significant expression of CST and GalT-III was found in the brains of mouse embryos in this study. Furthermore, CST was found to be expressed earlier than GalT-III (GalCer-synthase), i.e. CST increases after the late embryonic stage whereas GalT-III increases after the postnatal stage. During oligodendrocyte differentiation, sulfatide first appears in immature oligodendrocytes followed by the appearance of its precursor GalCer, which appears in more mature oligodendrocytes (Zhang 2001). The expression patterns of sulfatide and GalCer in oligodendrocytes may be caused by the expression of CST and then GalT-III as shown in this current investigation.

As demonstrated in this study, GD3, GM3, and GlcCer are highly expressed in mouse brains at the mid-embryonic stages. A robust expression of GD3, GM3, and GlcCer is found not only in brains but also in whole bodies, heads, and neural tubes of E10–E12 embryos (Seyfried 1987; Bouvier and Seyfried 1989; Seyfried and Ariga 1992). It is not clear whether the ganglioside expression patterns in these tissues also drastically change during development, but glycogenes may generally regulate the expression patterns of GSLs in a similar manner. However, the expression of GSLs is not simply regulated by glycosyltransferases responsible for their synthesis. For instance, the expression of GT3 and SSEA-1 is not consistently explained by the activities of the synthases. Additional regulatory mechanisms must exist to account for the expression of these GSLs. In addition to the relative activities of glycosyltransferases with those of the competing enzymes, other levels of control should also be considered. These factors include gene transcription and post-translational modification of the glycosyltransferases, enzyme subcellular localization, and substrate availability. All these factors are important for regulating the expression of GSLs during development (Yu et al. 2004). With the advent of contemporary molecular and cellular biological studies of glycogenes, future studies should be directed toward a better understanding of how GSL expression is regulated. How do glycogenes respond to developmental and differentiation cues, such as growth factors and cytokines? How do the signaling pathways regulate glycogene activation and deactivation? More systematic studies, such as the use of NPC culture, as shown in this study, should facilitate a better understanding of the expression of GSLs and glycogenes.


The authors thank Ms Donna Li and Dr Somsankar Dasgupta (Institute of Molecular Medicine and Genetics, Medical College of Georgia, Augusta, GA, USA) for their technical help. We also acknowledge Dr Rhea Markowitz for her editorial assistance. This work was supported in part by grants from National Institutes of Health and a grant from the Children’s Medical Research Foundation.