Developmental regulation of oligosialylation in zebrafish.

Abstract

Zebrafish appears as a relevant model for the functional study of glycoconjugates along vertebrate's development. Indeed, as a prelude to such studies, we have previously identified a vast array of potentially stage-specific glycoconjugates, which structures are reminiscent of glycosylation pathways common to all vertebrates. In the present study, we have focused on the identification and regulation of major protein and lipids associated alpha2-8-linked oligosialic acids motifs in the early development of zebrafish. By a combination of partial hydrolysis, anion exchange HPLC-FD and mass spectrometry, we demonstrated that glycoproteins and glycolipids differed by the extent and the nature of their substituting oligosialylated sequences. Furthermore, relative quantifications showed that alpha2-8-linked sialylation was differentially regulated in both families of glycoconjugates along development. Accordingly, we established that alpha2,8-sialyltransferase mRNA levels was directly correlated with changes of alpha2,8-sialylation status of glycolipids, but independent of those observed on major glycoproteins that appear to originate from the mother.

Full text

Page 1

Developmental regulation of oligosialylation in zebrafish
Lan-Yi Chang & Anne Harduin-Lepers & Ken Kitajima &
Chihiro Sato & Chang-Jen Huang & Kay-Hooi Khoo &
Yann Guérardel
Received: 21 April 2008 /Revised: 9 June 2008 /Accepted: 10 June 2008 / Published online: 14 August 2008
# Springer Science + Business Media, LLC 2008
Abstract Zebrafish appears as a relevant model for the
functional study of glycoconjugates along vertebrate’s
development. Indeed, as a prelude to such studies, we have
previously identified a vast array of potentially stage-
specific glycoconjugates, which structures are reminiscent
of glycosylation pathways common to all vertebrates. In the
present study, we have focused on the identification and
regulation of major protein and lipids associated α2-8-
linked oligosialic acids motifs in the early development of
zebrafish. By a combination of partial hydrolysis, anion
exchange HPLC-FD and mass spectrometry, we demon-
strated that glycoproteins and glycolipids differed by the
extent and the nature of their substituting oligosialylated
sequences. Furthermore, relative quantifications showed
that α2-8-linked sialylation was differentially regulated in
both families of glycoconjugates along development.
Accordingly, we established that α2,8-sialyltransferase
mRNA levels was directly correlated with changes of
α2,8-sialylation status of glycolipids, but independent of
those observed on major glycoproteins that appear to
originate from the mother.
Keywords α2,8-sialylation.α2,8-sialyltransferases.
Neu5Ac/Neu5Gc.Zebrafish.Early development.
Glycolipids.Glycoproteins
Introduction
Sialic acids constitute a vast family of heterogeneous
monosaccharides that are distributed throughout most of
the living organisms, from bacteria to vertebrates. About 40
different naturally occurring members have been described
so far [1]. Most of them occur in bound forms to
glycoproteins, glycolipids and free oligosaccharides as
terminal monosaccharide units. Internal α2-8-linked sialic
acids in the diasialosyl motif have been early recognized as
a common constituent of gangliosides and then of glyco-
proteins [2–6]. α2-8-polySia chain with a degree of
polymerization (DP) >7 was first identified in bacteria
soluble colominic acid prepared from culture medium [7].
Not until 1980 was it identified in animal kingdom
associated with a major sialoglycoprotein, the so called
polysialoglycoprotein (PSGP), isolated from the eggs of
rainbow trout [8]. Although ubiquitously identified in all
fish species examined so far, α2-8-polySia chains exhibited
a remarkable species specificity associated with differential
sequences, nature of constituent sialic acids and degree of
O-acetyl and O-lactyl substitutions [9]. Subsequently, α2-8-
Glycoconj J (2009) 26:247–261
DOI 10.1007/s10719-008-9161-5
Electronic supplementary material The online version of this article
(doi:10.1007/s10719-008-9161-5) contains supplementary material,
which is available to authorized users.
L.-Y. Chang:A. Harduin-Lepers:Y. Guérardel (*)
Unité de Glycobiologie Structurale et Fonctionnelle,
UMR CNRS 8576,
Université des Sciences et Technologies de Lille 1,
59655 Villeneuve d’Ascq, France
e-mail: yann.guerardel@univ-lille1.fr
L.-Y. Chang:C.-J. Huang:K.-H. Khoo
Institute of Biochemical Sciences, National Taiwan University,
Taipei 106, Taiwan
L.-Y. Chang:C.-J. Huang:K.-H. Khoo
Institute of Biological Chemistry, Academia Sinica, Nankang,
Taipei 11529, Taiwan
K. Kitajima:C. Sato
Laboratory of Animal Cell Function,
Bioscience and Biotechnology Center,
Department of Bioengineering Sciences,
Nagoya University, Nagoya, Japan

Page 2

polySia chain were shown to be associated with multiple
glycoproteins in eukaryotic organisms, including neural cell
adhesion molecule (N-CAM) [10], sodium channels [11],
CD-36 from human milk [12] and neuropilin [13].
The cellular sialic acid content is metabolically regulated
mainly by sialyltransferases and sialidases. α2,8-sialyltrans-
ferases (ST8Sia) are biosynthetic enzymes catalyzing the
transfer of sialic acid residues either to sialoglycoproteins or
sialoglycolipids (reviewed in [14–16]), while sialidases or
neuraminidases are glycohydrolytic enzymes that remove
sialic acid residues during sialoglycoconjugate degradation
[17]. Sialylated glycoconjugates are known to be tightly
regulated developmentally. In particular, gangliosides are
exclusively expressed in mice embryos from 7 days, when
neural crest appears, in agreement with their major local-
isation in central nervous system [18]. Then along brain
development, gangliosides pattern show drastic modifica-
tions, including a decrease of GM3 and GD3 concomitant to
an increase of GD1a and GT1b from mid-embryonic
development onward [19]. Similarly in rat, ganglioside
expression pattern was shown to shift from simple b-series
to complex gangliosides along brain development [20].
Polysialylated N-CAM exhibits also an exquisite spatio-
temporal regulation in the developing brain. Indeed, shortly
after its appearance in mouse brain at 8 days, N-CAM
become polysialylated with a peak expression of α2-8-
polySia during perinatal phase and a complete clearing
within the three weeks of post-natal brain development,
with the exception of sites of neuronal plasticity [21, 22].
In fish, polysialylation profile of PSGP is regulated along
oocytogenesis in ovary. Indeed, it was shown to shift from
a diSia-PSGP in earlier stages of oogenesis to a α2-8-
polySia-PSGP in later stages, based on the temporal
regulated expression of the α2,8-sialyltransferases ST8Sia
II and ST8Sia IV [23, 24]. However, with the exception of
N-CAM, little is known about the regulation and functions
of polySia sequences of various glycoconjugates along
embryogenesis in vertebrates.
The zebrafish, Danio rerio, has emerged in recent years
as an excellent model system to study the genetic under-
pinnings of vertebrate development. Recent profiling of
zebrafish embryos glycosylation established that this
organism synthesizes a wide variety of sialylated glyco-
conjugates, the expression of which is potentially regulated
along development [25]. Collected data has poised zebra-
fish as an ideal model to study the role of sialylation in
embryogenesis. The present report focuses on the identifi-
cation and regulation of α2-8-linked sialylated glycoconju-
gates in the early stages of zebrafish development through a
concerted approach combining structural, biochemical and
molecular biology analyses. In a first step, we established
with different structures of oligoSia sequences associated to
different types of glycoconjugates. Then, we evaluated
the implication of different glycogenes, including α2,8-
sialyltransferases (ST8Sia) and sialidases in the early
developmental regulation of the expression of α2-8-linked
sialic acids.
Materials and methods
Materials
The molecular biology kits RNeasy Midi and Plasmid
extraction were obtained from Qiagen (Chatsworth, CA, U.
S.A.), the TOPO TA cloning kit was from Invitrogen
(Cergy Pontoise, France), the NucleoSpin® RNA II kit was
from Macherey-Nagel (Düren, Germany). The oligonucleo-
tides were synthesized and purified by Eurogentec (Seraing,
Belgium), Sybr Green Brilliant Q-PCR master mix, eight-
well strip tubes and the MX-4000 Quantitative PCR System
were from Stratagene (La Jolla, CA, USA). The first strand
cDNA synthesis kit was from Amersham Pharmacia
Biotech (Little Chalfont, U.K.). The cDNA Kidney library
was kindly provided by L. Zon (ZFIN, Oregon). The
experion ARN Std Sens Analysis kit was from Biorad
(Marnes-la Coquette, France). Taq polymerase, 4-methyl-
umbelliferone (4-MU) and 2’-(4methylumbelliferyl)-α-d-N-
acetylneuraminic acid (4-MU-Neu5Ac) were from Sigma
(St Louis, MO, USA).
Sample collection
Zebrafish (D. rerio) were maintained at 28°C on a 14 h-
light/10 h-dark cycle. Embryos were incubated at 28°C and
different developmental stages were determined according
to the description in the Zebrafish Book [26].
Extraction and preparation of glycoconjugates
Embryos were suspended in 200 μl of water and homoge-
nised by sonication on ice. The resulting material was dried
and then sequentially extracted three times by chloroform/
methanol (2:1, v/v) and chloroform/methanol (1:2, v/v).
Supernatants from the extractions were pooled, dried and
subjected to a mild saponification in 0.1 M sodium
hydroxide in methanol at 37°C for 3 h, and then evaporated
to dryness. Sample was reconstituted in methanol/water
(1:1, v/v) and applied to a C18 Sep-Pak cartridge (Waters)
equilibrated in the same solvent system. After washing with
five volumes of methanol/water (1:1, v/v), glycosylcera-
mides were eluted by five volumes of methanol and five
volumes of chloroform/methanol (2:1, v/v).
248Glycoconj J (2009) 26:247–261

Page 3

Delipidated pellet from chloroform/methanol/water ex-
traction was re-suspended in a solution of 6 M guanidinium
chloride and 5 mM EDTA in 0.1 M Tris/HCl, pH 8, and
agitated for 4 h at 4°C. Dithiothreitol was then added to a
final concentration of 20 mM and incubated for 5 h at
37°C, followed by addition of iodoacetamide to a final
concentration of 50 mM and further incubated overnight in
the dark at room temperature. Reduced/alkylated sample
was dialysed against water at 4°C for 3 days and
lyophilized. The recovered protein samples were then
sequentially digested by TPCK treated trypsin for 5 h and
chymotrypsin overnight at 37°C, in 50 mM ammonium
bicarbonate buffer, pH 8.4. Crude peptide fraction was
separated from hydrophilic components on a C18 Sep-Pak
cartridge (Waters) equilibrated in 5% acetic acid by
extensive washing in the same solvent and eluted with a
step gradient of 20, 40 and 60% propan-1-ol in 5% acetic
acid. Pooled propan-1-ol fraction was dried and subjected
to N-glycosidase F (Roche) digestion in 50 mM ammonium
bicarbonate buffer pH 8.4, overnight at 37°C. The released
N-glycans were separated from peptides using the same
C18 Sep-Pak procedure as described above. To liberate O-
glycans, retained peptide fraction from C18 Sep-Pak was
submitted to alkaline reductive elimination in 100 mM
NaOH containing 1.0 M sodium borohydride at 37°C for
72 h. The reaction was stopped by addition of Dowex 50×
8 cation-exchange resin (25–50 mesh, H+form) at 4°C until
pH 6.5 and, after evaporation to dryness, boric acid was
distilled as methyl ester in the presence of methanol. Total
material was then submitted to cation-exchange chroma-
tography on a Dowex 50×2 column (200–400 mesh, H+
form) to remove residual peptides.
Chemical derivatization and MS analyses
Monosaccharide compositions were determined by gas
chromatography (GC)-mass spectrometry (MS) analysis as
alditol acetate derivatives. Briefly, glycan samples were
hydrolysed in 4 M trifluoroacetic acid (TFA) for 4 h at
100°C and then reduced with sodium borohydride in
0.05 M NaOH for 4 h. Reduction was stopped by drop
wise addition of acetic acid until pH 6 was reached and
borate salts were co-distilled by repetitive evaporation in
dry methanol. Peracetylation was performed in acetic
anhydride at 100°C for 2 h.
For MALDI-MS analyses, the glycan samples were
permethylated by NaOH in dimethyl sulfoxide, and then
extracted in chloroform and repeatedly washed with water.
MALDI-MS and MS/MS data were acquired on either a Q-
TOF Ultima MALDI instrument (Micromass) or a MALDI-
TOF/TOF system, the ABI 4700 Proteomic Analyzer,
exactly as described [27].
Analysis of oligo-sialylated sequences
In order to minimize internal fragmentation of polysialylated
sequences, sialylated glycan samples were directly coupled
to 1,2-diamino-4,5-methylenedioxybenzene (DMB) without
priormildhydrolysis[28]. Samples were incubated for 2.5 h
at 50°C in 50 μl of a DMB reagent solution (2.7 mM DMB,
9 mM sodium hydrosulfite, and 0.5 mM β-mercapto-
ethanol in 20 mM TFA). 10 μl of 1 M NaOH was then
added and the reaction mixtures further incubated in the
dark at room temperature for 1 h. Samples were stored at
4°C before analysis.
DMB-derivatized sialic acid oligomers were separated
on a HPLC apparatus fitted with either an anion exchanger
column, mono-Q (Amersham-Biosciences), or a CarboPac
PA-100 column (Dionex). For mono-Q column, the sample
was loaded and eluted with a flow rate of 0.5 ml/min with
20 mM Tris–HCl (pH 8.0), followed by a NaCl gradient (0–
10 min, 0 M; 10–60 min, 0 to 0.6 M; 60–65 min, 0.80 M)
in 20 mM Tris–HCl (pH 8.0). CarboPac column was eluted
at 1 ml/min with a concentration gradient of 2 to 32% of
1 M NaNO3 in water. In both systems, elution was
monitored by an on line fluorescence detector set at
wavelengths of 373 nm for excitation and 448 nm for
emission. Periodate oxidation and C7/C9 analyses for
oligosialyl linkage determination were performed essential-
ly as described by Sato et al. (1998) [33]. Briefly, samples
were dissolved in a mixture of 25 μl of 40 mM sodium
acetate buffer (pH 5.5) and 2 μl of 0.25 M sodium
metaperiodate and left at 0°C for 45 min in the dark. Five
microliters of 5% glycerol was then added and allowed to
react for another 40 min at 0°C, followed by 32 μl of 0.2 M
sodium borohydride in 0.2 M sodium borate buffer (pH 8.0)
and left overnight at 0°C. Finally, TFA was added to a final
concentration of 1 M and incubated at 80°C for 1 h before
subjected to DMB derivatization. To determine the chem-
ical nature of sialic acids, intact sialic acids were liberated
directly by mild hydrolysis in 0.01 N TFA at 50°C and
reacted with a volume of DMB reagent at 50°C for 2 h
30 min. The monomeric DMB-sialic acid derivatives were
separated isocratically on a C18 reverse phase HPLC
column (250×4.6 mm, 5 μm, Vydac) by a solvent mixture
of acetonitrile/methanol/ water (7:9:84) and identified by
referring to the elution positions of standard Neu5Ac and
Neu5Gc derivatives. For additional MS analysis, the DMB-
derivatives were separated instead with a gradient of
acetonitrile/methanol/water (7:9:84) mixture in water (0–
10 min, 10%; 10–40 min, 10 to 100%). Fluorescence-
detected peaks were individually collected on ice and
immediately freeze dried. Samples were then reconstituted
in 10 μl of water and analysed by nanoESI-MS and MS/MS
on an LCQ DK XP+ ion trap (Thermo Finnigan).
Glycoconj J (2009) 26:247–261249

Page 4

RNA extraction and cDNA synthesis
D. rerio unfertilized eggs and embryos (0, 6, 14, 24 and
36 hpf) kindly provided by the Thisses’ Lab were sampled
(200 embryos) and snap frozen in liquid nitrogen. Total
cellular RNA was extracted from embryos at various
developmental stages using the RNeasy Midi kit according
to the manufacturer’s instructions. Total RNA purity was
established by calculating the ratio of the absorbance readings
at 260 and 280 nm and quantified using the NanoDrop® ND-
100 spectrophotometer (NanoDrop Technologies, Wilming-
ton, DE, U.S.A.). The integrity and purity of the extracted
RNAwas also analyzed by means of gel electrophoresis on a
bioanalyzer (Experion, Biorad). Total RNA (1.2 µg) was
reverse transcribed using the first strand cDNA synthesis kit
in 33 µl following the manufacturer’s instruction. RNA
samples were tested for genomic DNA contamination by
PCR amplification of the zebrafish β-actin (GenBank
accession number AF025305; [29], using oligonucleotide
primers designed in two distinct exons (Sup. Table 1) and
aerosol contamination by including no template controls
(NTC). Another set of developmental stages cDNAs were
kindly prepared by H. Ahmed and G. Vasta according to [30].
Real-time PCR of ST8Sia genes during development
in zebrafish
Primers used for quantitative Q-PCR (Sup. Table 1) were
designed in the coding region of previously identified
zebrafish ST8Sia genes [14] using the Primer Premier
version 31.1 software (Primer Premier, Biosoft Internation-
al, Palo Alto, USA). Each primer pair was carefully
selected so to give rise to an amplified DNA fragment of
about 300 bp and such that their Tm values were very close
(around 51°C). The suitability of the primers for their
uniqueness to amplify a single PCR product was assured by
regular end-point PCR (Denaturation step at 94°C for 2 min
followed by 38 cycles at 95°C 1 min; 50°C 1 min; 72°C
1 min and an elongation step at 72°C for 10 min) using
cDNA kidney library provided by L. Zon. The amplified
products were subsequently run on an agarose gel, sub-
cloned in TOPO TA cloning vector and finally, fully
sequenced (Genoscreen, Lille). The TOPO plasmids con-
taining the amplified regions of the targeted genes were
amplified, purified and quantified by nanodrop and used for
the establishment of a standard curve for absolute quanti-
fication. Efficiency of target amplification for each primer
set (ST8Sia I, ST8Sia II, ST8Sia III, ST8Sia IV, ST8Sia V,
ST8Sia VI, Sup. Table 1) was optimized by real-time PCR
performed in a Stratagene MX4000 by trialing several final
primer concentrations. Each 25 µl Q-PCR master mix
contained 12.5 µl 2X Master Mix (Brilliant® SYBR®
Green Q-PCR Master Mix (Stratagene, CA)), 150 nM of
each primer, and 5 µl of diluted cDNA (equivalent to
100 ng total RNA) extracted from 0, 6, 14, 24 hpf embryos
andthe real-time quantitativePCRwhere thethermal cycling
program consisted of 10 min at 95°C followed by 45 cycles
of 30 s at 95°C, 1 min at 50°C and 30 s at 72°C and this
was followed by a melting step consisting of heating from
50°C to 95°C at an increment of 1°C per 30 s to check the
specificity of the amplified product. PCR for all the samples
were carried out in triplicate in eight-well strip tubes and
data were expressed as means +/− SD. The reactions were
quantified by selecting the amplification cycle when the
PCR product of interest was detected (threshold cycle, Ct).
Calibration curves were generated by ten-fold serial
dilution of Hind III linearized TOPO plasmids containing
the amplified regions of the targeted genes (from 2×10 5
copies to 2×10 1 copies). The same PCR master mix and
thermocycler conditions as described above were used and
plasmid standard curve equations were used to calculate the
absolute copy number of each gene. The amplification
efficiencies of each calibrator were found to be between
95.9% and 100.5%. We used absolute quantification relying
on the serial diluted DNA fragment with known concen-
tration, called calibrators, which were amplified from
cDNA of 24-hpf embryos with the same primers.
Sialidases assays
D. rerio embryos (0, 8, 24 and 48 hpf) and unfertilized eggs
were sampled (500 embryos) and snap frozen in liquid
nitrogen. The eggs were homogenized in 500 µl of water,
then different amount of total cell lysate corresponding to 1
to 60 eggs, were mixed with 0.2 mM 4-MU-Neu5Ac in
50 mM sodium acetate buffer (pH 3, 4, 5, 6 and 7) in a final
volume of 250 µl. Protein concentrations used were
determined using the micro BCA TM protein assay reagent
kit (Thermo Scientific Pierce, Rockford, USA). The
incubation was performed at 37°C. At 0.5, 1, 2 and 4 h,
30 µl of the reaction mixture was taken back, and the
reaction was quenched by adding 120 µl 0.5 M Na2CO3.
The released 4-methyl-umbeliferone (MUN) was measured
and quantified by fluorescence detector at 360 nm for
excitation, 460 nm for excitation. Sialidase activity was
calculated according to a MUN standard curve.
Results
Glycans from embryos contain oligosialic acid chains
Oligosialylation on N-glycans
We previously described in zebrafish embryos a family of
unusual di- and tri-antennary sialylated N-glycans along
250Glycoconj J (2009) 26:247–261

Page 5

with major ubiquitous oligomannosylated N-glycans [25].
They are characterized by the presence of Neu5Ac/Neu5Gc
monosialylated Lewis x motifs further substituted by a β4-
Gal residue. A 10% sialic acid content of the total N-glycan
fraction (Fig. 1) nonetheless implied that a much larger
proportion of sialylated N-glycans than that detectable by
MS might be present. In fact, after purification of sialylated
N-glycans by anion exchange chromatography, the propor-
tion of sialic acids in sialylated N-glycans increased sharply
up to 23% of total monosaccharides, which represents an
average of three to four sialic acid residues per N-glycan.
As demonstrated by reverse phase (RP)-HPLC analysis of
sialic acid-DMB derivatives, the Neu5Gc:Neu5Ac ratio,
which ranges between 2:1 and 4:1 depending on the sample
batch, was also somewhat inconsistent with a prevalence of
Neu5Ac over Neu5Gc implicated by MALDI-MS profiling
of the N-glycans. These discrepancies between the MS and
sialic acid composition data indicated that some additional
oligosialylated N-glycans may be refractory to MALDI-MS
detection.
To gain a better picture of the sialylation, the well
established DMB-tagging and HPLC analytical method
[31] was further employed to identify possible presence of
oligo- or polysialyl motifs. We first conducted structural
analyses of glycoproteins associated oligosialylation on
1 hpf embryos, then established that oligosialylation was
qualitatively identical in other developmental stages. The
N-glycan sample was incubated in the acidic derivatization
reaction mixtures without prior acidic liberation to mini-
mize internal fragmentation of polysialic acid chain. The
resulting tagged products were then separated on anion
exchange HPLC columns (MonoQ and CarboPac PA-100)
according to their degree of polymerization (DP) and
detected with a fluorescent detector (FD). Under the
experimental conditions employed, a monoQ column
permits a ready detection of polymeric sialic acid chains
from DP 2 up to DP 50 (data not shown), whereas the
CarboPac PA-100 column also allows detection of Neu5Ac
monomer. Since the FD response per mol of (Neu5Ac)n-
DMB remains constant for low DPs [32], integration of
peak areas therefore provides a good estimation of the
relative abundance of various Sian units. On MonoQ
column, total N-glycans fraction was found to yield at most
six peaks with retention times corresponding to Sia[(α2–8)
Sia]n-DMB standards of DP 2 to DP 7 (Fig. 2a). Identical
results were obtained with acidic N-glycans obtained after
purification by anion exchange chromatography (data not
shown). A sharp drop in the relative intensities of peaks
occurred from DP 2 onwards. On PA-100 column, it could
be estimated that DP 1 and DP 2 constitute 59% and 38%,
respectively, of the total content with higher oligosialyl
chains contributing to less than 3% in total (data not
shown). As expected, the observed peaks are sensitive to
the action of exoneuraminidase (data not shown). Closer
examination of the chromatograms showed that standard [-
8)Neu5Ac(α2-]nand [-8)Neu5Gc(α2-]nexhibited slightly
different retention times, in particular for DP 2 and DP 3
(Fig. 3a). Accordingly, chromatographic behaviours of
DMB-tagged oligoSia from N-glycans suggest that Sia2
and Sia3are exclusively composed of Neu5Gc residues. As
shown in Fig. 3c, standard Neu5Gc2peak co-migrates with
DP 2, whereas Neu5Ac2peak exhibits a clear time shift
compared to DP 2 (Fig. 3b).
To ascertain the identity of the major dimeric peak, the
same DMB derivative mixtures were subjected to RP-
HPLC in order to purify DP 2. The elution position of
dimeric sialic acid-DMB was inferred from standard
Neu5Ac1–3-DMB mixtures. A single major dimeric peak
was detected at a retention time similar to that of standard
Neu5Ac2-DMB (data not shown) and was collected for MS
and MS/MS analyses. As shown in Fig. 4a, ESI-MS
analysis of a standard Neu5Ac2-DMB afforded three
molecular ion signals in positive ion mode, corresponding
to [M+H]+, [M+Na]+and [M–H+2Na]+at m/z 717, 739
and 761, respectively. Further CID-MS/MS on the mono-
sodiated parent ion (Fig. 4b) yielded a major y ion at m/z
448 due to facile loss of the non-reducing terminal Neu5Ac
residue. In contrast, similar ESI-MS analysis on the
collected dimeric peak from the sample gave the
corresponding molecular ions at m/z 749, 771 and 793
(Fig. 4c), which differ from those afforded by Neu5Ac2-
DMB dimer by 32 mass units and are consistent with a
Neu5Gc2-DMB composition. This is supported by CID-
MS/MS on the candidate mono-sodiated parent ion at m/z
771 (Fig. 4d), which afforded a major y ion at m/z 464,
corresponding to loss of a non-reducing terminal Neu5Gc.
Further confirmation was then sought by referring to the
CID MS/MS spectrum of an authentic Neu5Gc2-DMB
standard which was found to co-elute with Neu5Ac2-DMB
0
10
20
30
40
50
60
Fuc GalManGlcNAcSialic
%
Fig. 1 Relative monosaccharide composition of N-glycans.
Monosaccharide composition of total (in grey) and acidic (in black)
N-glycans liberated from 1 hpf embryos were analysed by gas-
chromatography. Results are expressed in percentage of total
monosaccharides
Glycoconj J (2009) 26:247–261251

Page 6

standard at the same retention time under the HPLC
conditions employed. In contrast, a putative mono-sodiated
Neu5Ac1Neu5Gc1-DMB peak at m/z 755 did not afford
either a loss of Neu5Ac or Neu5Gc and was subsequent-
ly shown to be a prominent ESI-MS contaminant peak
commonly observed, when sample amount was low. Thus,
our innovative MS and MS/MS approaches have provided
unambiguous evidence for the presence of a Neu5Gc-
Neu5Gc dimer, and not a Neu5Ac2or Neu5Ac1Neu5Gc1
dimer, as a major oligosialyl motif on the N-glycans.
Exclusive presence of Neu5Gc in oligosialylated sequen-
ces was further assessed by mild periodate oxidation
followed by hydrolysis and DMB-labelling. Applied to
the sialylated N-glycans, it cleaves the non-substituted side
chains of Neu5Ac/Neu5Gc at the C7–C8bond, which are
identified by RP-HPLC as DMB-labelling C7/C9 ana-
logues [33]. This demonstrated that all 4 expected
products, namely C9(Neu5Gc)-DMB, C7(Neu5Gc)-DMB,
C9(Neu5Ac)-DMB and C7(Neu5Ac)-DMB could be
detected at increasing retention time (Fig. 5a), and
quantified as representing 28, 51, 2 and 18% of the total
sialic acid content, respectively on an N-glycan sample
with a Neu5Gc to Neu5Ac ratio of 4:1. Assuming the mild
periodate oxidation of sialic acid has proceeded to
completion and not hampered by any undetected non-
saccharide substitution on the side chain, the recovery of
intact C9Neu5Ac/Neu5Gc-DMB derivatives (about 30%
of total) is normally indicative of an internal α2,8-linked
sialyl motif since terminal or α2–9 linked sialic acids
would be cleaved to C7 analogues. Strikingly, there was 14
fold more α2,8-linked Neu5Gc than Neu5Ac which
suggests that where oligosialylation may occur, it prefer-
entially extends from Neu5Gc and not from Neu5Ac. This
conclusion is consistent with the observation that Neu5Gc
dimer and trimers are the major oligosialyl motifs.
Moreover, after subtracting the proportion of terminal
Neu5Gc that was α2–8 linked to internal Neu5Gc/Neu5Ac
in the dimer, the amount of the remaining C7-Neu5Gc
(21%) was roughly the same as that of C7-Neu5Ac (18%).
This figure is in good agreement with the MALDI-MS
analysis which detected complex type N-glycans with
antenna monosialylated by approximately equal amount
of terminal Neu5Gc and Neu5Ac [25].
10
10
20
20
30
30
40
40
50
50
60
60
70
70
Minutes
0
0
10
10
20
20
30
30
40
40
mVolts
mVolts
506070
1.0
1.1
1.2
mVolts
a
b
2
3
4
567
Fig. 2 Profiles of oligosialylated sequences on O- and N-glycans.
OligoSia sequences were released from glycans isolated from 1hpf
embryos, tagged with DMB and separated by HPLC-FD on an anion
exchange column. OligoSia profiles from (a) N-glycans including the
inset in upper panel, and (b) O-glycans. Peaks are labeled according
to the DP values as established by comparison with authentic
standards
0
10
20
30
mVolts
1020 3040506070
0
5
10
15
20
mVolts
0
10
20
30
40
mVolts
10
2030 405060 70
102030 40 50 6070 Minutes
NeuAc2
NeuAc3
NeuGc2
NeuGc3
a
b
c
Fig. 3 Identification of oligosialylation on N-glycans by anion
exchange DMB/HPLC-FD. Chromatographic profiles of co-injected
DMB-derivatized a [-8)Neu5Ac(α2-]nand [-8)Neu5Gc(α2-]nstand-
ards, b [-8)Neu5Ac(α2-]nstandard and oligo-Sia from zebrafish 1-hpf
embryos N-glycans, c [8)Neu5Gc(α2-]nstandard and oligo-Sia from
zebrafish 1 hpf embryos N-glycans, showing that the diSia (DP=2)
peak from zebrafish N-glycans co-migrates exclusively with Neu5Gc
(α2–8)Neu5Gc
252Glycoconj J (2009) 26:247–261

Page 7

Oligosialylation on O-glycans
Profiling of the glycosylation pattern of zebrafish embryos
also demonstrated the presence of prominent sialylated O-
linked glycans [25]. In contrast to N-glycans, disialylated
motifs on O-glycans could be directly identified by MS
analysis owing to lower molecular mass of O-glycans
compared with N-glycans. In particular, we identified
Neu5Gc–Neu5Gc as well as Neu5Ac–Neu5Gc motifs, but
could not observe Neu5Gc–Neu5Ac and Neu5Ac–Neu5Ac,
suggesting again the existence of exquisite specificity in
the synthesis of α2-8-sialylated epitopes. The extent of α2-
8-sialylation on O-glycans was evaluated using an exper-
imental approach identical to that for N-glycans and
showed very similar results. O-glycans are substituted by
oligosialylated motifs including up to seven residues, as
determined by DMB/HPLC-FD (Fig. 2b). As observed
for N-glycans, slight shifts in the retentions times compared
with Neu5Ac[(α2–8)Neu5Ac]n-DMB suggest the preva-
lence of Neu5Gc containing oligoSia over Neu5Ac
(Sup. Fig. 1). Separation of periodate oxidised compound
by RP-HPLC confirmed also the absence of internal α2-8-
linked Neu5Ac residues in the molecule, as observed on
N-glycans (Fig. 5b). However, O-glycans differed from N-
glycans by a lower C9(Neu5Gc) to C7(Neu5Gc) ratio which
suggests that the proportion of oligosialylation is lower in
O-glycans.
Collectively, the data presented show that both O- and
N-glycans are substituted by Neu5Gc containing oligosia-
lylated sequences which exhibit similar overall patterns.
Although Neu5Ac has been identified along Neu5Gc in
O- and N-glycans, it seems to be restricted to monosialy-
lated compounds or in non-reducing terminal position of
oligosialylated sequences.
Oligosialylation on glycolipids
Direct MALDI-MS-mapping of the acidic glycolipids
demonstrated the presence of oligosialylated glycolipids
substituted by up to five sialic acids (Fig. 6b). The major
tri-sialylated components were previously shown to be
substituted by a mixture of Neu5Ac and Neu5Gc residues
Fig. 4 Structural determination of the disialylated sequences on N-
glycans by CID-MS/MS of DMB-derivatives. a ESI-MS profile of a
Neu5Ac2-DMB standard isolated by C18 HPLC fractionation. The
most abundant molecular ion at m/z 739 corresponds to a mono-
sodiated species which was selected for CID MS/MS sequencing as
shown in (b). c ESI-MS analysis of the putative dimeric sialic acid-
DMB peak afforded by zebrafish embryonic N-glycans and similarly
isolated by C18 HPLC, followed by CID MS/MS on the most
abundant molecular ion at m/z 771 d which established its identity as
monosodiated Neu5Gc2-DMB derivative. In both MS/MS, loss of
204 u corresponds to loss of the common DMB moiety through
cleavage at C3–C4 of the derivatized, reducing end Neu5Ac/Neu5Gc,
as shown schematically
Glycoconj J (2009) 26:247–261 253

Page 8

in all possible combinations [25]. In contrast to N- and
O-glycans, the presence of polymerized Neu5Ac indicates
that no restriction seems to prevail in the synthesis of
oligosialylated motifs in glycolipids. Accordingly, charac-
terization of sialic acids by DMB/RP-HPLC demonstrated
the prevalence of a molar ratio of 1.5:1 for Neu5Ac:
Neu5Gc, indicating that sialylation patterns of glycolipids
differ from those of glycoproteins in which Neu5Gc
prevails over Neu5Ac. In agreement with direct observation
of sialylated glycolipids by MS, DMB/HPLC-FD analysis
shows the presence of oligosialylated motifs up to DP 6
(data not shown).
The sialic acid content changes along development
The presence of α2-8-sialylation on glycoproteins and
glycolipids was assessed along the development time-line
from 0 to 48 hpf. For O-glycans, presence of di-sialylated
glycans could be directly assessed by MS profiling. MALDI-
TOF analysis of permethylated glycans after separation of
mono- and disialylated compounds shows that previously
identified di-sialylated O-glycans can be observed exclusive-
ly in the earlier stages of developments (0 and 8 hpf) as
signals at m/z 1706 (Fucα1–3GalNAcβ1–4(Neu5Ac–
Neu5Gcα2–3)Galβ1–3GalNAc-itol) and at m/z 1736
(Fucα1–3GalNAcβ1–4(Neu5Gc–Neu5Gcα2–3)Galβ1–
3GalNAc-itol), but never in the later stages (Fig. 7).
Although mass spectrometry does not provide quantitative
information, it suggests a disappearance of oligosialylation
on O-glycans along development. DMB/HPLC-FD analysis
provided a semi-quantitative comparison of the oligosialyla-
tion content of total O-glycan fractions purified from
identical numbers of embryos at each development stage.
Each fraction presented a very similar pattern of (Sia)n-DMB
with 1<n<6–7, and thus did not show qualitative variation
in the extent of α2,8-sialylation. However, a clear decrease
in the quantity of each oligomer was observed as shown by
integration of chromatographic peaks for di-, tri and tetra-
sialyl components (Fig. 8a), confirming the rapid decrease of
oligosialylated O-glycans during embryonic development.
Indeed, after 8 hpf, quantity of disialylated motif dropped by
more than 60% and less than 5% of the initial di-sialylated
18002000 220024002600 280030003200 3400
m/z
0
100
%
1733
1763
1793
2095
2125
2155
2456
2486
254525742780 2905
2935
3266
32963326
Sia Lac-Cer
2
Sia Lac-Cer
3
0
100
%
1941
2145
*
2186
*
2361
23912432
2565
2606
2677
2810
2984
Hex Fuc -Cer
43
HexNAc Hex Fuc -Cer
154
HexNAc Hex Fuc -Cer
255
2473
2503
2923
2851
2637
a
b
Fig. 6 MALDI-MS analyses of
permethylated glycolipids from
zebrafish embryos. Glycolipid
profiles of high molecular mass
glycolipids from a 1 hpf and
b 48 hpf embryos. No sialylated
glycolipids were observed at 1 h
psf, whereas a complex mixture
of oligosialylated glycolipids
containing from 2 to 5 sialic
acids were detected at the later
stages. Symbols used: circle
Hex, square HexNAc, diamond
sialic acid, Cer ceramide
0
10
20
30
40
50
60
70
C9 NeuGc
C7 NeuGc
C9 NeuAc
C7 NeuAc
a
0
10
20
30
40
50
60
70
C9 NeuGc
C7 NeuGc
C9 NeuAc
C7 NeuAc
%
b
%
Fig. 5 Relative quantification of internal and external non-reducing
sialic acids in glycoproteins oligosialylated motifs. Periodate oxidized
Neu5Ac/Neu5Gc-DMB monomers from zebrafish embryonic glycans
were resolved on RP-HPLC to distinguish the respective C7/C9
products by referring to the elution positions of authentic standards.
Chromatographic profiles of a N-glycans and b O-glycans from 1 hpf
embryos. Results are expressed in percentage of total oxidized
derivatives and are representative of two independent experiments
254Glycoconj J (2009) 26:247–261

Page 9

motif could be observed prior to hatching (48 hpf). Similarly,
the content of tri-sialylated motifs was reduced by more than
20 fold in 48 h. Identical methodology was applied to N-
glycans and showed a similar trend of rapid clearance of
oligosialylation along development (Fig. 8b).
In contrast to glycoprotein glycosylation, several lines of
evidence demonstrated that oligosialylation in glycolipids
increases along development. First, direct observation of
sialylated glycolipids by MS was only possible in later stages
of development (24 and 48 hpf) as previously reported.
Indeed, MS profiling of total glycolipid from 0 and 8 hpf
embryos exclusively showed a complex pattern of neutral
fucosylated glycolipids (Table 1), but no sialylated com-
pounds (Fig. 6). Repeated attempts to purify acidic com-
pounds from early stages embryos failed to provide any
evidence for their presence. These results were confirmed by
comparing endoceramidase treated total glycolipid fractions
from embryos at 0 and 48 hpf. Both samples show overall
similar profiles characterized by complex mixtures of
0
20
40
60
80
100
0h8h 24h 48h
Sia 2
Sia 3
Sia 4
0
20
40
60
80
100
0h8h 24h 48h
Sia 2
Sia 3
Sia 4
0
20
40
60
80
100
0h8h24h48h
Sia 2
Sia 3
Sia 4
0
20
40
60
80
100
0h 8h24h 48h
NeuGc
NeuAc
a
b
c
d
%
%
%
%
Fig. 8 Relative quantification of oligosialylation along embryos
development. Proportions of Sia 2, Sia 3 and Sia 4 associated to a
O-glycans, b N-glycans and c glycolipids were compared from 0 to
48 hpf by anion exchange DMB/HPLC-FD. Relative quantities of
Neu5Ac and Neu5Gc from glycolipids were also followed along
development by RP-HPLC after total release of sialic acids (d).
Results are representative of three independent experiments
0h
8h
24h
48h
a
%
%
%
%
m/z
m/z
b
Fig. 7 MALDI-MS profiling of permethylated O-glycans from
zebrafish embryos. The presence of a mono-sialylated O-glycans
Fucα1–3GalNAcβ1–4(Neu5Acα2–3)Galβ1–3GalNAc-itol at m/z
1314 and Fucα1–3GalNAcβ1–4(Neu5Gcα2–3)Galβ1–3GalNAc-itol
at m/z 1344 and b di-sialylated O-glycans Fucα1–3GalNAcβ1–4
(Neu5Ac-Neu5Gcα2–3)Galβ1–3GalNAc-itol at m/z 1706 and
Fucα1–3GalNAcβ1–4(Neu5Gc-Neu5Gcα2–3)Galβ1–3GalNAc-itol
at m/z 1736 was checked by MALDI-MS analyses of the four
embryonic stages from 0 to 48 hpf
Glycoconj J (2009) 26:247–261255

Page 10

identical neutral glycans (Sup. Fig. 2). Later stage sample
shows the presence of additional major signals at m/z 838.6,
868.6, 1083.8 and 1113.8 attributed to Neu5Ac1Hex2,
Neu5Gc1Hex2, Neu5Ac1Hex2HN1and Neu5Gc1Hex2HN1,
respectively. Furthermore, careful analysis of MS spectra
reveals the presence of additional minor signals exclusively
in later stage at m/z 1199.9, 1229.9 and 1259.9 attributed to
Neu5Ac2Hex2, Neu5Ac1Neu5Gc1Hex2 and Neu5Gc2Hex2
as well as m/z 1145.0, 1475.1 and 1505.1 attributed to
Neu5Ac2Hex2HN1, Neu5Ac1Neu5Gc1Hex2HN1and Neu5-
Gc2Hex2HN1. The chemicalnatureof oligosialylatedmotifsin
endoceramidase treated glycolipids was confirmed by MS/MS
sequencing of their permethylated derivatives. Indeed, frag-
mentation of molecular ions at m/z 1199, 1229 and 1259
showed the presence of Neu5Ac2, Neu5Ac1Neu5Gc1 and
Neu5Gc2sequences owing to the presence of B/Y ion pairs at
m/z 760/463, 790/463 and 820/463, respectively (Sup. Fig. 3).
Furthermore, presence of both Y ions at m/z 825 and 855
established that compound Neu5Ac1Neu5Gc1Hex2 at m/z
1229 is a mixture of the two isobaric structures Neu5Gc–
Neu5Ac–Hex–Hex and Neu5Ac–Neu5Gc–Hex–Hex. In
agreement with MS analysis, profiling by DMB/HPLC-FD
demonstrated that oligosialylation increased along develop-
ment, with a sharp increase between 24 and 48 hpf (Fig. 8c).
Indeed, as compared with embryos at 0 hpf, quantity of diSia
increased by 3.5 fold and triSia by 17 fold in 48 hpf embryos.
Accordingly, the total amount of sialic acid at 48 hpf is 4 to 6
higher than in other stage (Fig. 8d).
These data clearly demonstrate that the amount of
oligosialylation in embryos varies along development.
Surprisingly, comparative profiling of glycoconjugates
established that the overall content of sialylation decreases
for glycoproteins along embryogenesis, but increases for
glycolipids.
Changes of α2-8-sialylation status of glycolipids is directly
correlated with α2,8-sialyltransferase mRNA levels,
whereas sialidase activities might be responsible
glycoproteins
To investigate whether ST8Sia genes may participate in the
di-, oligo- and polysialylation of major glycoconjugates
during the zebrafish development, we examined the mRNA
levels of the α2,8-sialyltransferase genes (ST8Sia I, ST8Sia
II, ST8Sia III, ST8Sia IV, ST8Sia V and ST8Sia VI) in
different developmental stages by real time PCR. Six
human ST8Sia orthologues were previously identified in
the zebrafish genome [14] and were molecularly cloned
from various zebrafish organs by RT-PCR and sequenced.
Since β-actin showed variable level of expression among
the various developmental stages with the same amount of
total cDNA (data not shown), we have chosen absolute
quantification to quantify ST8Sia. Except for ST8Sia II, all
the ST8Sia genes showed very low level of expression at
0 hpf (Fig. 9b). They all increase along embryogenesis
from 0 to 36 hpf but at very different rates (Fig. 9a). Indeed,
expression levels of ST8Sia I and ST8Sia VI sharply
increase in the first 14 h of development, whereas other
ST8Sia exhibited either more modest or delayed increase
(Fig. 9a). Then, ST8Sia I and ST8Sia III reach maximum
sustained expression levels at 14 hpf, whereas the others
show a gradual increase in expression level up to 36 hpf.
Interestingly, ST8Sia I and ST8Sia V genes, the human
orthologs of which are known to be involved in glycolipids
sialylation, are both dramatically up-regulated along
development by a factor of 300 and 165, respectively. This
result is in agreement with the observation of an increase of
sialylation associated to glycolipids. On the contrary, the
modestly increased expression levels of ST8Sia II and
ST8Sia IV, which are responsible for the biosynthesis of
polysialic acid, and of ST8Sia III also involved in the α2,8-
sialylation of glycoproteins are still slightly inconsistent
with the decreased oligosialylation status observed on
glycoproteins. More surprisingly is the 4000 fold increase
of ST8Sia VI gene expression that is not directly correlated
with an increase of Sia2 motifs synthesis on glycoproteins.
In order to check whether zebrafish sialidases might be
involved in these α2-8-sialoglycoprotein metabolism, we
performed sialidase assays using 4-methyl-umbelliferyl-
Neu5Ac (4MU-Neu5Ac) as a substrate at various pH for
various embryonic developmental stages (0, 8, 24, and
48 hpf) and unfertilized eggs. Our preliminary data showed
Table 1 Monosaccharidecompositionofneutralglycolipidscalculated
from MALDI-MS analysis of permethyl derivatives
m/z [M+Na]+
Composition
HexNAcHexdHex
1,941
2,145
2,186
2,361
2,391
2,432
2,473
2,503
2,565
2,606
2,637
2,677
2,810
2,851
2,923
2,984
0
0
1
1
1
2
3
3
1
2
2
3
2
3
4
2
4
5
4
4
5
4
3
4
5
4
5
4
5
4
4
5
3
3
3
4
3
3
3
2
4
4
4
3
4
4
3
5
256 Glycoconj J (2009) 26:247–261

Page 11

the existence of intense sialidase activities associated to
embryos. Survey of sialidase activities showed dramatic
differences of intensities depending on the pH and along
development (Sup. Fig. 4). The fact that the evolution of
total sialidase activities measured at different pH follows
different trends strongly suggest the presence of different
enzymes, as recently demonstrated by the identification of
several genes coding for potential sialidases in zebrafish
[34]. Indeed, the sialidase activities detected at pH 5
sharply decreased along the zebrafish development, whereas
sialidase activities detected at pH 4 increased slightly along
the zebrafish development. These data demonstrate a tight
regulation of sialidase activities along development and
suggest a possible involvement of sialidases in the α2-8-
sialylation status of sialoglycoproteins, which may explain
the discrepancies observed between the regulation of
ST8Sia genes involved in the glycoproteins biosynthesis
and the actual synthesis of oligosialylated motifs on
glycoproteins.
Discussion
A previous study of the glycosylation profiles of zebrafish
embryos demonstrated that this organism synthesizes a vast
panel of unusual sialylated glycoconjugates. Of particular
interest are the β4-galactosylated sialyl Lewis x, Galβ1–4
(Siaα2–3)Galβ1–4(Fucα1–3)GlcNAc, motif observed on
all complex N-glycans and Fucα1-3GalNAcβ1–4(Siaα2–3)
Galβ observed on O-glycans. The Galβ1–4Galβ1–
4GlcNAc motif now appears to be a common feature of
several fish glycoproteins, as described in Oryzias mela-
stigma, Tribolodon hakonensis and Oryzia latipes [35, 36],
but the α2–3Neu5Ac sialylated Galβ1–4Galβ1–4GlcNAc
motif observed in D. rerio was additionally identified
only in O. latipes. However, its Neu5Gc substituted
equivalent is specific to D. rerio so far. Similarly, the
Fucα1–3GalNAcβ1–4(Siaα2–3)Galβ motif has been ex-
clusively identified on the O-glycans from D. rerio
although a wide range of other O-glycan structures have
been described in fishes [37].
The present study extended further our knowledge of the
fine structures of sialylated glycans synthesized by zebra-
fish along embryonic development by focusing on the
α2-8-sialylation pattern (quality and quantity) of glycopro-
teins and glycolipids. Our data clearly established that both
glycoproteins and glycolipids were α2-8-sialylated. Sur-
prisingly, fine structural analysis demonstrated that the
major glycolipids and glycoproteins presented different
patterns of oligosialylation. Whilst the α2-8-sialylated
glycolipids may be substituted by mixtures of oligo
(Neu5Ac), oligo(Neu5Gc) and hybrid type oligo(Neu5Ac,
Neu5Gc) sequences, glycoproteins are mainly substituted
by oligo(Neu5Gc). Indeed, although Neu5Ac and Neu5Gc
4059.40 165.564.58 5.53 6.31303.74
36h
75.1813.00 1.555.77 0.97 92.78
24h
173.390.461.19 4.510.91 266.78
14h
1.001.001.00 1.00 1.00 1.00
0h
21.65 2.79 1.82 3.91 6.380.15
unf
ST8Sia VIST8Sia V ST8Sia IV ST8Sia IIIST8Sia IIST8Sia I
a
b
Fig. 9 Absolute quantification
of Dre ST8Sia genes expression
by real-time PCR. a “Fold of
increase” represents the relative
expression quantity of each
ST8Sia in different develop-
mental stages compared with
expression at 0 hpf. b The
different expression levels of all
the ST8Sia genes were analyzed
with cDNA from 0 hpf empty
square, 14 hpf
36 hpf
eggs and ovary filled
square by quantitative real-time
PCR. The absolute amount of
transcripts was quantified
according to the same DNA
fragment amplified and cloned
in the plasmid. Values are the
mean of triplicate points
, 24 hpf,
Glycoconj J (2009) 26:247–261 257

Page 12

are present in similar proportions on mono-sialylated motifs
of N-glycans [25], we have now demonstrated that only
Neu5Gc is further elongated by other sialic acids to form
oligo(Neu5Gc) sequences. Identical biosynthetic restric-
tions seem to prevail during the extension of oligosialic
acids associated with O-glycans that prevent formation of
oligo(Neu5Ac). These differences between glycoproteins
and glycolipids were also reflected in their respective
content of sialic acids: two to four times more Neu5Gc
than Neu5Ac in glycoproteins and about twice as much
Neu5Ac than Neu5Gc for glycolipids. As previously
reported, polysialic acid structures of fish egg glycoproteins
are exquisitely species specific, differing by their extent,
composition, acetylation and sequences [9]. However, to
our knowledge, zebrafish represents the only documented
example of the preferential use of Neu5Gc over Neu5Ac for
oligosialic acid elongation found on glycoproteins. This
suggests an α2,8-sialyltransferase activity dedicated to
α2-8-sialylation of glycoproteins, such as ST8Sia II,
ST8Sia III, ST8Sia IV or ST8Sia VI that would exhibit a
preference for CMP-Neu5Gc over CMP-Neu5Ac. In mam-
mals, ST8Sia II and ST8Sia IV are known to be involved
primarily in the polysialylation of the N-glycans of N-
CAM, ST8Sia III catalyzes the transfer of one to seven
sialic acid residues onto glycoproteins or glycolipids,
whereas ST8Sia VI catalyzes the transfer of a unique sialic
acid residue on the O-glycans of glycoproteins [16, 38]. In
this context, it is also interesting to note that the zebrafish
ST8Sia IV shows very low capability to transfer Neu5Ac
from CMP-Neu5Ac onto N-CAM compared to the zebra-
fish ST8Sia II [39]. This last observation might reflect a
preference of the zebrafish ST8Sia IV for CMP-Neu5Gc
over CMP-Neu5Ac. Also, it is noteworthy that the major
glycoprotein associated α2-8-sialyl motifs in zebrafish
exhibit a significantly lower DP compared to those
previously identified on the polysialylated glycoprotein
(PSGP) of other fish species. Indeed, whereas salmonids
eggs contain polysialyl units with chain length up to 20
residues [8, 37], the major sialylated glycoproteins of
zebrafish eggs are mainly substituted by diSia (DP=2)
motif as well as minute amounts of oligoSia (3<DP<6).
In addition, expression of oligosialylation on glycolipids
and glycoproteins is differentially regulated along embry-
onic development. Indeed, the extent of α2-8-sialylation on
major glycoproteins sharply decreases whereas that of
glycolipid increases along development. Surveys of sialy-
lation by MS analyses of intact and endoceramidase treated
glycolipids, as well as the quantification of sialic acids and
oligosialic acids, all indicated that significant sialylation
specifically and reproducibly appears around 24 hpf, which
strongly suggests that glycolipid associated α2-8-sialylation
is triggered during early development. Accordingly, we
failed to detect significant glycolipid associated sialylation
in mature ovaries before spawning (data not shown). It is
noteworthy that the complex neutral glycolipids observed
in all development stages are apparently not the substrates
for sialylation events occurring in later developmental
stages. MS analyses demonstrated that the sialylated
glycolipids of later stages comprised mainly Neu5Ac/
Neu5Gc substituted (Sia)1–4LacCer, GM2 and GD2 with
no sialylated equivalents of the fucosylated neutral glyco-
lipids (Table 1), suggesting that these two families of
compounds are independently synthesized.
The appearance of glycolipids associated α2-8-sialylation
in later stages of embryonic development positively corre-
lates with the up-regulation of genes coding for α2,8-
sialyltransferases ST8Sia I, ST8Sia III and ST8Sia V
(Fig. 9). The human recombinant enzymes have been
shown to be involved in the biosynthesis of gangliosides
GD3, GT3, GD1a, GT1b and GQ1c (reviewed in [16]) [40–
46]. The onset expression of these genes starts around
10 hpf and is essentially located in the developing brain
(Chang et al. 2008, this issue). Up-regulation of oligo-
sialylation along zebrafish development is in agreement
with previous observations made on Xenopus laevis
showing by in vitro assays that ST8Sia I (SAT-2) and
ST8Sia V (SAT-4) activities were dramatically increased
along the early development, with a maximum activity at
day 4 [47]. However, whereas quantities of both neutral
glycolipids and gangliosides sharply increase in X. laevis,
only gangliosides appear to be up-regulated in D. rerio
[48]. In contrast to glycolipids, evolution of glycoproteins
associated α2,8-sialylation does not correlate with the
temporal expression and the general increase of mRNAs
of ST8Sia II, ST8Sia IV and ST8Sia VI from 10 hpf along
the early stages of zebrafish development, therefore
implicating a different kind of regulation. These latter
ST8Sia genes were found to be expressed in the ovaries
(Fig. 9), suggesting that the α2-8-sialylated glycoproteins
of the zebrafish embryo detected at very early develop-
mental stages well before the onset zygotic expression of
these ST8Sia (around 10 hpf), might be inherited from the
mother. We hypothesize that these inherited α2-8-sialylated
glycoproteins could be subsequently degraded in the
embryos by endogenous sialidases.
As a first step towards substantiating this hypothesis, we
have essayed the sialidase activities with synthetic sub-
strates at various pH (3, 4, 5, 6, 7) in unfertilized eggs.
Since the higher and lower activities were obtained at pH 5
and 4 respectively, we then assayed the sialidase activities
at pH 4 and 5 for the various developmental stages of
interest. Our preliminary data showed that sialidase
activities found at pH 4 increased along zebrafish develop-
ment, whereas sialidase activities found at pH 5 decreased
(Sup. Fig. 4). The correlation between the detected
sialidases activities and glycoproteins oligosialylation along
258Glycoconj J (2009) 26:247–261

Page 13

embryonic development therefore remain equivocal and the
implicated zebrafish sialidases still need to be clearly
identified. A recent study of Manzoni et al. [34] reported
the identification of seven zebrafish sialidase genes homol-
ogous to three of the four known human genes (NEU1,
NEU2, NEU3, NEU4) The zebrafish neu3.1, neu3.3 and
neu4 were shown to be active towards gangliosides at very
low pH (2–3) and the corresponding genes were found to
be expressed differentially along the embryonic develop-
ment. On the other hand, an additional NEU3 orthologous
gene named neu3.2 has been described, which started to be
expressed at 24 hpf. The corresponding enzyme appears to
be soluble in the cytosol and active at higher pH (5.5).
However, fine enzymatic characterization of all the zebra-
fish sialidases still awaits studies.
Such an up-regulation of the ST8Sia II gene expression
along embryonic development of zebrafish has been
previously reported in the context of an increased synthesis
of polysialic acid chains (PSA) on the N-glycans of the
neural cell adhesion molecule (N-CAM), which reaches a
maximum around 27–40 hpf [49]. Our present study
focused instead on the global α2-8-sialylation status of
the different classes of glycoconjugates along embryonic
zebrafish development and has identified a rapid decrease
of glycoproteins associated α2-8-sialylation content. This
might be explained by the large quantities of PSGP
synthesized in the cortical alveoli during oogenesis in
fishes [24] compared to the natural low abundance and
restricted localization of PSA-N-CAM. It is most probable
that the N-CAM polysialylation pattern cannot be discrim-
inated from the one of PSGP or other polysialylated
glycoproteins by a global approach.
As a premise to the identification of other oligosialylated
glycoproteins in zebrafish fertilized eggs, we have assessed
the oligosialylation patterns of glycoproteins in different
compartments of the fertilized oocyte: embryo tissue,
chorion and perivitelline space. Semi-quantification analy-
sis by DMB-HPLC revealed that about 94% of the total
oligoSia in 1 hpf fertilized oocyte was associated with
soluble glycoconjugates in the perivitelline space, 5% in the
chorion and less than 1% in the embryonic tissue (data not
shown). Surprisingly, the extent of oligosialylation distrib-
uted within each of these fractions was very different. The
perivitelline space associated components show a very short
DP distribution dominated by DP 2, reminiscent of the one
observed on total glycoprotein fraction, whereas the
chorion and embryonic tissues exhibit more evenly distrib-
uted patterns with up to DP 10 (Sup. Fig. 5). Each fraction
was further differentiated by their sialic acids content, as
embryo associated glycoproteins was almost exclusively
composed of Neu5Gc (Neu5Gc/Neu5Ac 14:1), whereas
perivitelline space associated glycoproteins of Neu5Gc and
Neu5Ac in a 4:1 ratio (data not shown).
The presence of large quantities of oligosialylated soluble
glycoproteins in perivitelline space of zebrafish is in good
agreement with previous reports of the polysialylated
peptides originating from a fertilization induced proteolysis
of cortical alveoli PSGP in several other species of fish,
including trouts, salmonids and medaka fish [9, 24, 50].
Accordingly, we observed large quantities of protein
associated oligoSia chains in zebrafish mature ovaries that
exhibit a distribution pattern and a composition identical to
that of the soluble glycoproteins of fertilized oocytes (data
not shown). However, despite the postulated conservation
of this phenomenon among fishes and the large quantity of
excreted PSGP, the destiny and the function of these
compounds are still largely unknown. Altogether, our data
established that the vast majority of oligosialylation
observed in fertilized oocytes is synthesized prior to
embryogenesis in the mother ovary and then degraded
along embryonic development for a yet unknown purpose.
The observation of a protein associated oligosialylation
in embryonic tissue as early as 1 hpf, thus long before that
N-CAM associated PSA appears, strongly suggests the
presence of so far undescribed oligosialylated glycoproteins
in oocytes. Diversity of polysialylated components in fish
embryos increases the possible endogenous substrates for
zebrafish α2,8-sialyltransferases. Indeed, ST8Sia II and IV
were shown to be involved in the synthesis of polysialic
acid chains on both N-CAM N-glycans in zebrafish and on
PSGP O-glycans in rainbow trout, which suggest that these
enzymes might have multiple substrates acceptors in
various animal species [49, 51]. Altogether, the results
presented here establish the structural bases for the
investigation of the fine enzymatic specificities of the
different ST8Sia identified from zebrafish genome [14].
These studies are actually in progress by using the
endogenous molecules as acceptor substrates. Further work
is also needed to identify the protein carriers of the
oligosialylated motifs and to assess their localization within
the fertilized embryo. The collected data will enable a better
evaluation of the importance of sialyltransferases as well as
sialidases in the regulation of synthesis of sialylated motifs
along zebrafish embryogenesis.
Acknowledgement
National de la Recherche Scientifique), EGIDE, the Ministère des
Affaires Etrangères and PPF Bioinformatique de Lille, and Academia
Sinica, Taiwan. We acknowledge Dr. Estelle Garénaux and Dr.
Bernadette Coddeville for their help on mass spectrometry analyses
of derivatized oligosialic acids and Yi-Cyun CHEN for help in
preparation of zebrafish embryos. We acknowledge also the technical
assistance of Anne-Marie Mir, Beatrice Catieaux and Dr. Sandrine
Duvet, and Pr. Yu-Teh Li for helpful discussion. D. rerio embryos
were provided by Dr. Bernard and Christine Thisse (IGBMC,
Strasbourg) and a set of staged cDNA were kindly prepared by Hafiz
Hamed and Gerardo Vasta (Baltimore, USA) that were used for
quantification of sialyltransferase mRNA levels.
This research was supported by CNRS (Centre
Glycoconj J (2009) 26:247–261 259

Page 14

Abbreviations
Sialyltransferases nomenclature is according to Tsuji et al. [52],
gangliosides nomenclature is according to Svennerholm [53]. hpf|
hours post-fertilization;diSia|disialyl motif;triSia|trisalylmotif;DHB|
2,5-dihydroxybenzoic acid;DP|degree of polymerisation;GC|gas chro-
matography;Hex|hexose;HexNAc|N-acetyl hexosamine;HexNAcitol|
reduced N-acetyl hexosaminitol;LacCer|lactosylceramide;MALDI|ma-
trix-assisted laser-desorption ionization;MS|mass spectrometry;TFA|
trifluoroacetic acid;TOF|time-of-flight
References
1. Schauer, R., Kamerling, J.: Chemistry, biochemistry and biology of
sialic acids. In: Montreuil, J., Vliegenhart, J., Schachter, H. (eds.)
Glycoproteins II, pp. 243–372. Elsevier, Amsterdam (1997)
2. Finne, J., Krusius, T., Rauvala, H.: Occurrence of disialosyl
groups in glycoproteins. Biochem. Biophys. Res. Commun. 74,
405–410 (1977). doi:10.1016/0006-291X(77)90318-7
3. Handa, N., Handa, S.: The chemistry of lipids of posthemolytic
residue or stroma of erythrocytes. XIV. Chemical structure of
glycolipid of cat erythrocyte stroma. Jpn. J. Exp. Med. 35, 331–
341 (1965)
4. Handa, S., Burton, R.M.: Lipids of retina. I. Analysis of
gangliosides in beef retina by thin layer chromatography. Lipids
4, 205–208 (1969). doi:10.1007/BF02532630
5. Krusius, T., Finne, J., Karkkainen, J., Jarnefelt, J.: Neutral and
acidic glycopeptides in adult and developing rat brain. Biochim.
Biophys. Acta. 365, 80–92 (1974)
6. Kuhn, R., Wiegandt, H.: Die konstitution der ganglio-n-tetraose
und des gangliosids gl. Chem. Ber. 96, 866–880 (1963).
doi:10.1002/cber.19630960329
7. McGuire, E.J., Binkley, S.B.: The structure and chemistry of
colominic acid. Biochemistry 3, 247–251 (1964). doi:10.1021/
bi00890a017
8. Inoue, S., Iwasaki, M.: Characterization of a new type of
glycoprotein saccharides containing polysialosyl sequence. Bio-
chem. Biophys. Res. Commun. 93, 162–165 (1980). doi:10.1016/
S0006-291X(80)80260-9
9. Sato, C., Kitajima, K., Tazawa, I., Inoue, Y., Inoue, S., Troy 2nd,
F.A.: Structural diversity in the alpha 2®8-linked polysialic acid
chains in salmonid fish egg glycoproteins. Occurrence of poly
(Neu5Ac), poly(Neu5Gc), poly(Neu5Ac, Neu5Gc), poly(KDN),
and their partially acetylated forms. J. Biol. Chem. 268, 23675–
23684 (1993)
10. Finne, J.: Occurrence of unique polysialosyl carbohydrate units in
glycoproteins of developing brain. J. Biol. Chem. 257, 11966–
11970 (1982)
11. James, W.M., Agnew, W.S.: Multiple oligosaccharide chains in the
voltage-sensitive Na channel from electrophorus electricus: evidence
foralpha-2,8-linkedpolysialicacid.Biochem.Biophys.Res.Commun.
148, 817–826 (1987). doi:10.1016/0006-291X(87)90949-1
12. Yabe, U., Sato, C., Matsuda, T., Kitajima, K.: Polysialic acid in
human milk. CD36 is a new member of mammalian polysialic
acid-containing glycoprotein. J. Biol. Chem. 278, 13875–13880
(2003). doi:10.1074/jbc.M300458200
13. Curreli, S., Arany, Z., Gerardy-Schahn, R., Mann, D., Stamatos,
N.M.: Polysialylated neuropilin-2 is expressed on the surface of
human dendritic cells and modulates dendritic cell-T lymphocyte
interactions. J. Biol. Chem. 282, 30346–30356 (2007).
doi:10.1074/jbc.M702965200
14. Harduin-Lepers, A., Mollicone, R., Delannoy, P., Oriol, R.: The
animal sialyltransferases and sialyltransferase-related genes: a
phylogenetic approach. Glycobiology 15, 805–817 (2005).
doi:10.1093/glycob/cwi063
15. Harduin-Lepers, A., Recchi, M.A., Delannoy, P.: 1994, the year of
sialyltransferases. Glycobiology 5, 741–758 (1995). doi:10.1093/
glycob/5.8.741
16. Harduin-Lepers, A., Vallejo-Ruiz, V., Krzewinski-Recchi, M.A.,
Samyn-Petit, B., Julien, S., Delannoy, P.: The human sialyltrans-
ferase family. Biochimie 83, 727–737 (2001). doi:10.1016/S0300-
9084(01)01301-3
17. Saito, M., Yu, R.: Biochemistry and function of sialidases. In:
Rosenberg, A. (ed.) Biology of Sialic Acids, pp. 261–313.
Plenum, New York (1995)
18. Hakomori, S.I.: Structure and function of glycosphingolipids and
sphingolipids: Recollections and future trends, Biochim Biophys
Acta (2007).
19. Ngamukote, S., Yanagisawa, M., Ariga, T., Ando, S., Yu, R.K.:
Developmental changes of glycosphingolipids and expression of
glycogenes in mouse brains. J. Neurochem. 103, 2327–2341
(2007). doi:10.1111/j.1471-4159.2007.04910.x
20. Yu, R.K., Macala, L.J., Taki, T., Weinfield, H.M., Yu, F.S.:
Developmental changes in ganglioside composition and synthesis
in embryonic rat brain. J. Neurochem. 50, 1825–1829 (1988).
doi:10.1111/j.1471-4159.1988.tb02484.x
21. Hildebrandt,H.,Muhlenhoff,M.,Weinhold,B.,Gerardy-Schahn,R.:
Dissecting polysialic acid and NCAM functions in brain develop-
ment. J. Neurochem. 103(Suppl 1), 56–64 (2007). doi:10.1111/
j.1471-4159.2007.04716.x
22. Probstmeier, R., Bilz, A., Schneider-Schaulies, J.: Expression of
the neural cell adhesion molecule and polysialic acid during early
mouse embryogenesis. J. Neurosci. Res. 37, 324–335 (1994).
doi:10.1002/jnr.490370305
23. Asahina, S., Sato, C., Kitajima, K.: Developmental expression of a
sialyltransferase responsible for sialylation of cortical alveolus
glycoprotein during oogenesis in rainbow trout (Oncorhynchus
mykiss). J. Biochem. 136, 189–198 (2004). doi:10.1093/jb/
mvh106
24. Kitazume, S., Kitajima, K., Inoue, S., Inoue, Y., Troy 2nd, F.A.:
Developmental expression of trout egg polysialoglycoproteins and
the prerequisite alpha 2,6-, and alpha 2,8-sialyl and alpha 2,8-
polysialyltransferase activities required for their synthesis during
oogenesis. J. Biol. Chem. 269, 10330–10340 (1994)
25. Guérardel, Y., Chang, L.Y., Maes, E., Huang, C.J., Khoo, K.H.:
Glycomic survey mapping of zebrafish identifies unique sialyla-
tion pattern. Glycobiology 16, 244–257 (2006). doi:10.1093/
glycob/cwj062
26. Westerfield, M.: The zebrafish book. A guide for laboratory use of
zebrafish (Danio rerio), p. 385. University of Oregon Press,
Eugene (1995)
27. Yu, S.Y., Wu, S.W., Khoo, K.H.: Distinctive characteristics of
MALDI-Q/TOF and TOF/TOF tandem mass spectrometry for
sequencing of permethylated complex type N-glycans. Glycoconj.
J. 23, 355–369 (2006). doi:10.1007/s10719-006-8492-3
28. Sato, C., Inoue, S., Matsuda, T., Kitajima, K.: Fluorescent–
assisted detection of oligosialyl units in glycoconjugates. Anal.
Biochem. 266, 102–109 (1999). doi:10.1006/abio.1998.2921
29. Chen, W.Y., John, J.A., Lin, C.H., Chang, C.Y.: Molecular cloning
and developmental expression of zinc finger transcription factor
MTF-1 gene in zebrafish, Danio rerio. Biochem. Biophys. Res.
Commun. 291, 798–805 (2002). doi:10.1006/bbrc.2002.6517
30. Ahmed, H., Du, S.J., O'Leary, N., Vasta, G.R.: Biochemical and
molecular characterization of galectins from zebrafish (Danio
rerio): notochord-specific expression of a prototype galectin
during early embryogenesis. Glycobiology 14, 219–232 (2004).
doi:10.1093/glycob/cwh032
31. Inoue, S., Inoue, Y.: A challenge to the ultrasensitive chemical
method for the analysis of oligo- and polysialic acids at a
260Glycoconj J (2009) 26:247–261

Page 15

nanogram level of colominic acid and a milligram level of brain
tissues. Biochimie 83, 605–613 (2001). doi:10.1016/S0300-9084
(01)01307-4
32. Inoue, S., Lin, S.L., Lee, Y.C., Inoue, Y.: An ultrasensitive
chemical method for polysialic acid analysis. Glycobiology 11,
759–767 (2001). doi:10.1093/glycob/11.9.759
33. Sato, C., Inoue, S., Matsuda, T., Kitajima, K.: Development of a
highly sensitive chemical method for detecting alpha2®8-linked
oligo/polysialic acid residues in glycoproteins blotted on the
membrane. Anal. Biochem. 261, 191–197 (1998). doi:10.1006/
abio.1998.2718
34. Manzoni,M.,Colombi,P.,Papini,N.,Rubaga,L.,Tiso,N.,Preti,A.,
et al.: Molecular cloning and biochemical characterization of
sialidases from zebrafish (Danio rerio). Biochem. J. 408, 395–
406 (2007). doi:10.1042/BJ20070627
35. Inoue, S., Iwasaki, M., Ishii, K., Kitajima, K., Inoue, Y.: Isolation
and structures of glycoprotein-derived free sialooligosaccharides
from the unfertilized eggs of Tribolodon hakonensis, a dace.
Intracellular accumulation of a novel class of biantennary disialoo-
ligosaccharides. J. Biol. Chem. 264, 18520–18526 (1989)
36. Iwasaki, M., Seko, A., Kitajima, K., Inoue, Y., Inoue, S.: Fish egg
glycophosphoproteins have species-specific N-linked glycan units
previously found in a storage pool of free glycan chains. J. Biol.
Chem. 267, 24287–24296 (1992)
37. Inoue, S., Inoue, Y.: Fish glycoproteins. In: Montreuil, J.,
Vliegenthart, J., Schachter, H. (eds.) Glycoproteins II, pp. 143–
161. Elsevier, Amsterdam (1997)
38. Teintenier-Lelievre, M., Julien, S., Juliant, S., Guerardel, Y.,
Duonor-Cerutti, M., Delannoy, P., et al.: Molecular cloning and
expression of a human hST8Sia VI (α2,8-sialyltransferase)
responsible for the synthesis of the diSia motif on O-glycosyl-
proteins. Biochem. J. 392, 665–674 (2005). doi:10.1042/
BJ20051120
39. Marx, M., Rivera-Milla, E., Stummeyer, K., Gerardy-Schahn, R.,
Bastmeyer, M.: Divergent evolution of the vertebrate polysialyl-
transferase Stx and Pst genes revealed by fish-to-mammal
comparison. Dev. Biol. 306, 560–571 (2007). doi:10.1016/j.
ydbio.2007.03.032
40. Haraguchi, M., Yamashiro, S., Yamamoto, A., Furukawa, K.,
Takamiya, K., Lloyd, K.O., et al.: Isolation of GD3 synthase gene
by expression cloning of GM3 α2,8-sialyltransferase cDNA using
anti-GD2 monoclonal antibody. Proc. Natl. Acad. Sci. USA. 91,
10455–10459 (1994). doi:10.1073/pnas.91.22.10455
41. Nara, K., Watanabe, Y., Maruyama, K., Kasahara, K., Nagai, Y.,
Sanai, Y.: Expression cloning of a CMP-NeuAc: NeuAcα2–
3Galb1–4Glcb1–1¢Cer α2,8-sialyltransferase (GD3 synthase)
from human melanoma cells. Proc. Natl. Acad. Sci. USA. 91,
7952–7956 (1994). doi:10.1073/pnas.91.17.7952
42. Sasaki, K., Kurata, K., Kojima, N., Kurosawa, N., Ohta, S.,
Hanai, N., et al.: Expression cloning of a GM3-specific α2,8-
sialyltransferase (GD3 synthase). J. Biol. Chem. 269, 15950–
15956 (1994)
43. Daniotti, J.L., Rosales Fritz, V., Kunda,P., Nishi, T., Maccioni, H.J.:
Cloning, characterization and developmental expression of alpha2,8
sialyltransferase (GD3 synthase, ST8Sia I) gene in chick brain and
retina. Int. J. Dev. Neurosci. 15, 767–776 (1997). doi:10.1016/
S0736-5748(97)00027-0
44. Lee, Y.C., Kim, Y.J., Lee, K.Y., Kim, K.S., Kim, B.U., Kim, H.N.,
et al.: Cloning and expression of cDNA for a human Sia alpha
2,3Gal beta 1, 4GlcNA:alpha 2,8-sialyltransferase (hST8Sia III).
Arch. Biochem. Biophys. 360, 41–46 (1998). doi:10.1006/
abbi.1998.0909
45. Kim, Y.J., Kim, K.S., Do, S., Kim, C.H., Kim, S.K., Lee, Y.C.:
Molecular cloning and expression of human alpha2,8-sialyltrans-
ferase (hST8Sia V). Biochem. Biophys. Res. Commun. 235, 327–
330 (1997). doi:10.1006/bbrc.1997.6725
46. Kono, M., Yoshida, Y., Kojima, N., Tsuji, S.: Molecular cloning
and expression of a fifth type of α2,8-sialyltransferase (ST8Sia
V). Its substrate specificity is similar to that of SAT-V/III, which
synthesize GD1c, GT1a, GQ1b and GT3. J. Biol. Chem. 271,
29366–29371 (1996). doi:10.1074/jbc.271.46.29366
47. Gornati,R.,Basu,S.,Bernardini,G.,Rizzo,A.M.,Rossi,F.,Berra,B.:
Activities of glycolipid glycosyltransferases and sialidases during the
early development of Xenopus laevis. Mol. Cell. Biochem. 166, 117–
124 (1997). doi:10.1023/A:1006891414663
48. Gornati, R., Rizzo, A.M., Tong, X.W., Berra, B., Bernardini, G.:
Glycolipid patterns during Xenopus embryo development. Cell
Biol. Int. 19, 183–189 (1995). doi:10.1006/cbir.1995.1061
49. Marx, M., Rutishauser, U., Bastmeyer, M.: Dual function of
polysialic acid during zebrafish central nervous system develop-
ment. Development 128, 4949–4958 (2001)
50. Taguchi, T., Seko, A., Kitajima, K., Inoue, S., Iwamatsu, T.,
Khoo, K.H., et al.: Structural studies of a novel type of
tetraantennary sialoglycan unit in a carbohydrate-rich glycopep-
tide isolated from the fertilized eggs of Indian Medaka fish,
Oryzias melastigma. J. Biol. Chem. 268, 2353–2362 (1993)
51. Asahina, S., Sato, C., Matsuno, M., Matsuda, T., Colley, K.,
Kitajima, K.: Involvement of the α2,8-polysialyltransferases II/
STX and IV/PST in the biosynthesis of polysialic acid chains on
the O-linked glycoproteins in rainbow trout ovary. J. Biochem.
140, 687–701 (2006). doi:10.1093/jb/mvj200
52. Tsuji, S., Datta, A.K., Paulson, J.C.: Systematic nomenclature for
sialyltransferases. Glycobiology 6, v–vii (1996). doi:10.1093/
glycob/6.7.647
53. Svennerholm, L.: The gangliosides. J. Lipid Res. 5, 145–155
(1964)
Glycoconj J (2009) 26:247–261261