Glycobiology vol. 14 no. 5 pp. 431±441, 2004
DOI: 10.1093/glycob/cwh034
Advance Access publication on January 21, 2004
Detailed structural features of glycan chains derived from a1-acid glycoproteins of
several different animals: the presence of hypersialylated, O-acetylated sialic acids
but not disialyl residues
Miyako Nakano2, Kazuaki Kakehi1,2, Men-Hwei Tsai3,
and Yuan C. Lee4
2
Faculty of Pharmaceutical Sciences, Kinki University, Kowakae 3-4-1,
Higashi-Osaka 577±8502, Japan; 3Ambryx Biotechnology Inc., 5603D
Foxwood Drive, Oak Park, CA 91377; and 4Biology Department, Johns
Hopkins University, North Charles Street, Baltimore, MD 21218
Received on September 22, 2003; revised on October 26, 2003;
accepted on November 14, 2003
We analyzed carbohydrate chains of human, bovine, sheep,
and rat a1-acid glycoprotein (AGP) and found that carbohydrate chains of AGP of different animals showed quite distinct variations. Human AGP is a highly negatively charged
acidic glycoprotein (pKa ˆ 2.6; isoelectic point ˆ 2.7) with a
molecular weight of approximately 37,000 when examined by
matrix-assisted laser-desorption/ionization time-of-flight
mass spectrometry (MALDI-TOF MS) and contains di-,
tri-, and tetraantennary carbohydrate chains. Some of the
tri- and tetraantennary carbohydrate chains are substituted
with a fucose residue (sialyl Lewis x type structure). In sheep
AGP, mono- and disialo-diantennary carbohydrate chains
were abundant. Tri- and tetrasialo-triantennary carbohydrate
chains were also present as minor oligosaccharides, and some
of the sialic acid residues were substituted with N-glycolylneuraminic acid. In rat AGP, very complex mixtures of
disialo-carbohydrate chains were observed. Complexity of
the disialo-oligosaccharides was due to the presence of N,
O-acetylneuraminic acids. Triantennary carbohydrate chains
carrying N,O-acetylneuraminic acid were also observed as
minor component oligosaccharides. We found some novel
carbohydrate chains containing both N-acetylneuraminic
acid and N-glycolylneuraminic acid in bovine AGP. Interestingly, triantennary carbohydrate chains were hardly detected
in bovine AGP, but diantennary carbohydrate chains with trior tetrasialyl residues were abundant. Furthermore the major
sialic acid in these carbohydrate chains was N-glycolylneuraminic acid. It should be noted that these sialic acids are
attached to multiple sites of the core oligosaccharide and are
not present as disialyl groups.
Key words: a1-acid glycoprotein/HPLC/
N-acetylneuraminic acid/N-glycolylneuraminic acid/
MALDI-TOF MS
1
To whom correspondence should be addressed; e-mail:
[email protected]
Introduction
a1-Acid glycoprotein (AGP), a highly heterogeneous
glycoprotein, is an acute-phase protein. AGP is produced
mainly in the liver (Sarcione et al., 1967), but extrahepatic
synthesis has also been reported (Sorensson et al., 1999).
Serum concentrations of AGP in physiological conditions
are about 1 g/L in human and increase severalfold during
acute-phase reactions, cancer, pregnancy and some other
diseases (Fournier et al., 2000). Although AGP is an abundant protein, its physiological significance is not fully
understood. Involvement of AGP in nonspecific resistance
to a Gram-negative infection (Hochepied et al., 2000) and
stabilization effect on erythrocyte membrane (Matsumoto
et al., 2003) have been reported.
Human AGP has a molecular mass of 41±43 kDa when
examined by sodium dodecyl sulfate±polyacrylamide gel
electrophoresis, and contains ~45% carbohydrates (Schmid
et al., 1977). Carbohydrate chains attach to the protein core
as the complex-type of five N-linked glycans (Fournet et al.,
1978). Various forms of di-, tri-, and tetraantennary carbohydrate chains contribute to the human AGP glycan complexity. Significant heterogeneity of the oligosaccharides
also exists due to variation in the linkage of N-acetylneuraminic acid (NeuAc) to galactose (Gal) and due to
the presence of fucose (Fuc) residues (Treuheit et al.,
1992). Micromolar quantities of asialo-carbohydrate chains
were isolated from human AGP, and their structures were
confirmed using a combination of proton nuclear magnetic
resonance and matrix-assisted laser-desorption/ionization
time-of-flight mass spectrometry (MALDI-TOF MS)
(Stubbs et al., 1997).
In human AGP, marked changes in glycoforms are
observed during acute-phase reactions. The changes comprise alterations in branching pattern as revealed by reactivity with Concanavalin A and the fucose-binding lectin
(Elliott et al., 1997). The expression of a sialyl Lewis
x{sLex, NeuAca2-3Galb1-4(Fuca1-3)GlcNAc-} portion in
the carbohydrate chains of human AGP molecules has been
a challenging target as related to inflammation. De Graaf
et al. (1993) found a direct relationship between the reactivity of human AGP to the fucose-specific binding lectin
(Aleuria auranita) and staining of human AGP by anti-sLex
monoclonal antibody under healthy and disease conditions.
Thus analysis of sialo- and asialo-oligosaccharides of AGP
as well as its glycoforms is important for understanding the
biological roles of AGP (Kakehi et al., 2002; Sei et al.,
2002).
Glycosylation is one of the most important posttranslational events for proteins, affecting their functions in health
Glycobiology vol. 14 no. 5 # Oxford University Press 2004; all rights reserved.
431
M. Nakano et al.
and disease, playing significant roles in various information
traffics for intracellular and intercellular biological events.
Recently, Fukui et al. (2002) proposed a microarray method
for the determination of protein±carbohydrate interactions.
We also proposed a glycomics approach for interaction
analysis between protein and carbohydrate using a set of
carbohydrates as library by capillary affinity electrophoresis (Nakajima et al., 2003). This method affords simultaneous determination of (1) carbohydrate chains, (2) binding
specificity of the constituent carbohydrate chains, and
(3) kinetic data, such as the association constant of each
carbohydrate. During the course of the studies, we found
that a mixture of carbohydrate chains derived from human
AGP is one of the best libraries of carbohydrates, because
human AGP contains typical di-, tri-, and tetraantennary
complex-type N-glycans as already described.
As an extension of the work for establishing glycan
libraries for glycomics studies, we have been analyzing
carbohydrate chains of many glycoprotein samples obtainable from commercial sources and found that AGP samples
from human, bovine, sheep, and rat sera showed distinct
variations in carbohydrate chains. Especially bovine AGP
contains quite unique and novel sialic acid±containing
diantennary oligosaccharides
composed
of only
N-glycolylneuraminic acid.
Results
MALDI-TOF MS measurement of bovine fetuin and
AGP samples from human, bovine, sheep, and rat sera
In the present study we used commercial samples of AGP
derived from human, sheep, bovine, and rat. Carbohydrate
chains of fetuin from bovine fetal sera have been extensively
studied by many research groups (Green et al., 1988;
Hayase et al., 1992; Townsend et al., 1986). Therefore, we
employed fetuin (Sigma, St. Louis, MO) as a reference
glycoprotein. As shown in Figure 1, bovine fetuin showed
a broad peak at 48 kDa, and AGP samples showed a broad
peak between m/z 30,000 and m/z 40,000. AGP samples
from bovine and sheep showed molecular ions at
m/z 33,800 and m/z 33,400, respectively. On the other
hand, AGP samples from human and rat showed molecular
ions at m/z 37,000 and m/z 36,900, respectively.
It has been shown that human AGP contains multiantennary carbohydrate chains and showed high molecular
masses by MALDI-TOF MS. Relatively low molecular
masses of bovine and sheep indicate that these proteins
have diantennary oligosaccharides as the major carbohydrate
chains (see following discussion).
Analysis of carbohydrate chains of fetuin from
bovine sera
Carbohydrate chains were released from the peptides with
N-glycoamidase F after digestion of the protein with a
combination of trypsin and chymotrypsin (Kawasaki et al.,
2003; Nakano et al., 2003), and were labeled with
2-aminobenzoic acid (2AA) and analyzed by highperformance liquid chromatography (HPLC) using an
amino column (Figure 2).
432
Fig. 1. MALDI-TOF MS spectra of bovine fetuin and AGP samples
from human, bovine, sheep, and rat sera.
By comparing the data with those reported by Anumula
and Dhume (1998), we could assign major peaks and
confirm the structures by MALDI-TOF MS measurement.
A list of the structures of carbohydrate chains and their
abbreviations are summarized in Figure 3 and Table I.
For example, the triantennary oligosaccharide containing
three N-acetylneuraminic acid (NeuAc) residues is represented as 3-NNN (peak no. 19 in Table I): The first numeral
indicates the branch numbers (triantennary in this case),
and NNN indicates the presence of three NeuAc residues. If
N-glycolylnueraminic acid is present, ``G'' instead of ``N'' is
used. In the similar manner, F and Pla indicate the presence
of fucose and polylactosamine residues, respectively.
In the present mode of separation using an amino-bonded
polymer-based stationary phase, the carbohydrate chains
were eluted according to their charges based on the number
of sialic acid residues, and positional isomers of sialic acid
residues (a-2,3 versus a-2,6) are also resolved. Two earliest
eluting groups (2±0 and 3±0; 1 and 2) appearing at ~27 min
(Figure 2) were due to diantennary and triantennary carbohydrate chains carrying no sialic acids. The peaks of the
second group at around 40 min were mixtures of monosialo
tri- and diantennary carbohydrate chains [3-N and 2-N
(3 and 4), respectively], and those of the third group at
around 52 min were disialo tri- and diantennary carbohydrate chains (3-NN and two 2-NN peaks; 11 and 13).
We found two disialo-diantennary carbohydrate chains,
Hypersialylated oligosaccharides of a1-acid glycoprotein
Fig. 2. Analysis of carbohydrate chains of bovine fetuin. 0SA, 1SA, 2SA,
3SA, and 4SA indicate that asialo-, monosialo-, disialo-, trisialo-, and
tetrasialo-oligosaccharides are observed in these regions. The numbers
and oligosaccharide structures are summarized in Table I and Figure 3.
In the box, the region between 85 min and 180 min is enlarged.
apparent isomers in NeuAca2-3/6Gal linkages (Green et al.,
1988), which we could not distinguish by MS technique.
The most abundant group of peaks at around 63 min was
trisialo-triantennary carbohydrate chains. Two major
peaks as well as a small peak (3-NNN; 19) were the positional isomers of a2-3- and a2-6-linked NeuAc residues
(Green et al., 1988). Furthermore, we found another trisialo-triantennary carbohydrate chains (3-NNG; 20) at a
later elution position, in which one of the sialic acids is
substituted with N-glycolylneuraminic acid (NeuGc). At
around 82 min, triantennary chains containing four
NeuAc residues (3-NNNN; 30) were found (Townsend
et al., 1986). Furthermore we found some other minor
sialo-carbohydrate chains eluted later. A pentasialo-tetraantennary carbohydrate chain (4-NNNNN; 35) was
observed at 99 min. We also found three pentasialo-triantennary carbohydrate chains (3-NNNNN, 3-NNNNG, and
3-NNNGG; 41, 42, and 43) at ~127 min. Interestingly, these
carbohydrate chains contained one or two NeuGc residues.
We observed MALDI-TOF mass spectra of these oligosaccharides after separation of each peak. The molecular
ions of oligosaccharides which we observed are shown in
Table I. MALDI-TOF mass spectra of some unique carbohydrate chains in bovine fetuin that were observed later are
shown in Figure 4 (a-series).
A pentasialo-tetraantennary carbohydrate chain (35) at
99 min was confirmed by the molecular ion at m/z 3949. We
also found a group of pentasialo-triantennary carbohydrate
chains at ~127 min. The most intense peak was that of
a triantennary chain having five NeuAc residues (41;
m/z 3583). The later eluting peaks (42 and 43) showed the
ions at m/z 3599 and m/z 3615, respectively. These molecular ions were 16 and 32 mass units higher than m/z 3583
(3-NNNNN; 41), indicating that these peaks were 3-NNNNG
and 3-NNNGG (i.e., triantennary carbohydrate chains
Fig. 3. List of the structures for sialooligosaccharides found in AGP
samples from human, bovine, sheep, and rat sera.
substituted with four NeuAc and one NeuGc, and three
NeuAc and two NeuGc residues, respectively).
Analysis of carbohydrate chains from human, bovine,
sheep, and rat AGP
Based on the data on the analysis of bovine fetuin, we
analyzed carbohydrate chains of AGP samples from
433
M. Nakano et al.
Table I. List of molecular masses for sialooligosaccharides found in AGP
samples from human, bovine and sheep sera
Peak No.
Abbreviation
[M ‡ 2AA-H]ÿ
Structure
1
2±0
1762
a
2
3±0
2127
b
3
3-N
2418
c
4
2-N
2053
d
5
2-G
2069
6
4-N
2638
e
7
3-NF
2564
f
8
4-NF
2930
g
9
4-NN
3075
h
10
4-NNF
3221
i
11
3-NN
2709
j
12
3-NNF
2855
k
13
2-NN
2344
l
14
2-NG
2360
15
2-GG
2376
16
4-NNN
3366
m
17
4-NNNF
3512
n
18
3-NNNF
3147
o
p
19
3-NNN
3001
20
3-NNG
3017
21
3-NGG
3033
22
3-GGG
3049
23
2-NNN
2635
24
2-NNG
2651
25
2-NGG
2667
26
2-GGG
2683
27
4-NNNNPla
4022
r
28
4-NNNNF
3803
s
q
29
4-NNNN
3657
t
30
3-NNNN
3292
u
31
3-NNNG
3308
32
3-NNGG
3324
33
3-NGGG
3340
34
3-GGGG
3356
35
4-NNNNN
3949
v
36
2-NNNN
2927
w
37
2-NNNG
2943
38
2-NNGG
2959
39
2-NGGG
2975
40
2-GGGG
2991
41
3-NNNNN
3583
42
3-NNNNG
3599
43
3-NNNGG
3615
44
3-NNGGG
3631
45
3-NGGGG
3647
46
3-GGGGG
3663
x
For the peak numbers, see Figures 2 and 5. The structures are summarized
in Figure 3. N, G, F, and Pla mean N-acetylneuraminic acid, N-glycolyl
neuraminic acid, fucose, and polylactosamine, respectively.
434
human, bovine, sheep and rat sera. The results are shown
in Figure 5.
The regions where neutral (0SA), mono- (1SA), di- (2SA),
tri- (3SA), and tetra-sialo (4SA) carbohydrate chains were
eluted are also indicated in the figure. As reported
previously by several groups (Kakehi et al., 2002; Sei et al.,
2002; Stubbs et al., 1997), human AGP contains trisialocarbohydrate chains as the major group, but di- and
tetrasialo-carbohydrate chains are also present. Monoand disialo-carbohydrate chains, observed at ~40 min and
52 min, respectively, were common to all animals. Trisialoand tetrasialo-carbohydrate chains in human AGP were
observed at around 65 min and 75 min, respectively.
However, the counterparts from bovine and sheep AGP
were obviously observed later. The major group in
sheep AGP was disialo-oligosaccharides, but mono- and
tri-sialo-oligosaccharides were also found. On the contrary,
rat AGP showed much more complex chromatogram
indicating that sialic acids of the oligosaccharides were
modified with N,O-acetyl groups.
Human AGP contains di-, tri-, and tetraantennary carbohydrate chains, and some of the tri- and tetraantennary
carbohydrate chains are substituted with a fucose residue
to form the sialyl Lewis x±type structure. We confirmed
identities of several peaks in human AGP as well as those
also found in fetuin using MALDI-TOF MS measurements
(see Figure 4, b-series). In the region (around 40 min) where
monosialo-carbohydrate chains were observed, monosialodiantennary carbohydrate chains (3-N, 2-N, 4-N, 3-NF,
and 4-NF; 3, 4, 6, 7, and 8) were detected. Five disialocarbohydrate chains (4-NN, 4-NNF, 3-NN, 3-NNF, and 2NN; 9, 10, 11, 12, and 13) were observed at ~52 min.
Trisialo-carbohydrate chains were observed at ~63 min.
Two trisialo-tri- and -tetraantennary carbohydrate chains
containing sialyl Lewis x structure (4-NNNF and 3-NNNF;
17 and 18) were observed, and the structures were confirmed by the molecular ions at m/z 3512 and m/z 3147,
respectively. Tetrasialo-carbohydrate chains were observed
at ~75 min. Tetrasialo-tetraantennary carbohydrate chain
carrying polylactosamine structure (4-NNNNPla; 27) and
tetrasialo-tetraantennary carbohydrate chains containing
sialyl Lewis x structure (4-NNNNF; 28) were identified by
the ions at m/z 4022 and m/z 3803, respectively.
We found that bovine AGP contained surprisingly unique
carbohydrate chains that were quite different from those of
human AGP. As shown in Figure 5c, abundant peaks
derived from disialo-diantennary carbohydrate chains
were observed at around 55 min. Diantennary carbohydrate
chain (2-NN; 13) with two NeuAc residues was observed
first. A carbohydrate chain (2-NG; 14) with both NeuAc
and NeuGc residues was more abundantly present and
observed slightly later than 13. The largest peak (15)
observed last in this region was clearly due to the carbohydrate chain substituted with two NeuGc residues.
MALDI-TOF MS spectra of these unique carbohydrate
chains are also shown in Figure 4 (c-series). 2-NG and 2GG (14 and 15; m/z 2360 and m/z 2376, respectively) were
easily identified by increase of 16 and 32 mass units to that
of 2-NN (13; m/z 2344). Another abundant group of peaks
between 70 and 80 min are diantennary carbohydrate
chains having trisialo structures (23±26). The carbohydrate
Hypersialylated oligosaccharides of a1-acid glycoprotein
Fig. 4. MALDI-TOF MS spectra of some unique carbohydrate chains found in bovine fetuin and AGP samples from human, bovine and rat. a, b, c,
and e mean bovine fetuin, human, bovine, and rat AGP, respectively. The numbers correspond to peak numbers in Table I and Table II.
chain (23) with three NeuAc residues was eluted the earliest,
the last peak being that substituted with three NeuGc residues (26). Characteristic molecular ions for 2-NNN (23), 2NNG (24), 2-NGG (25), and 2-GGG (26) were observed at
m/z 2635, m/z 2651, m/z 2667, and m/z 2683, respectively, as
expected. Furthermore, we found tetrasialo-diantennary
carbohydrate chains (36±40) between 95 min and 111 min.
These hypersialylated carbohydrate chains were also
435
M. Nakano et al.
Fig. 5. Comparison of carbohydrate chains in AGP samples among animals. b, human; c, bovine; d, sheep; e, rat. 0SA, 1SA, 2SA, 3SA, and 4SA indicate
that asialo-, monosialo-, disialo-, trisialo-, and tetrasialo-oligosaccharides are observed in these regions. The numbers and oligosaccharide structures
are summarized in Figure 3, Table I, and Table II. In the boxes, the region between 70 min and 170 min is enlarged.
confirmed by their molecular ions observed by MALDITOF MS. Although triantennary carbohydrate chains were
also detected in bovine AGP in trace amounts, they also
contained tetra- and pentasialyl residues of NeuAc and
NeuGc in different ratios. The presence of the triantennary
oligosaccharide carrying five NeuGc residues (46) was especially unique. MS data of some unique triantennary carbohydrate chains were also shown in Figure 4 (c-series).
436
From the accumulated data for fetuin, human, and
bovine AGPs, we tentatively assigned the major peaks of
carbohydrate chains derived from sheep AGP. It is interesting that monosialo-oligosaccharides in sheep AGP were
abundantly present. Triantennary carbohydrate chains
(19, 20, and 30±32) were much less as compared with
those of human AGP. Hypersialylated oligosaccharides
such as trisialo- and tetrasialo-diantennary oligosaccharides
Hypersialylated oligosaccharides of a1-acid glycoprotein
were also observed, and the diantennary-oligosaccharide (2GG; 15) with only NeuGc was also observed. But tri- and
tetrasialo-diantennary oligosaccharides with only NeuGc
residue were not observed.
Rat AGP contained a quite complex mixture of oligosaccharides, as shown in Figure 5e. In the present mode of
separation using a polymer-based amino column, the
separation of oligosaccharides is performed based on a
number of structural factors including (1) anionic charge
(main factor), (2) the number of antenna, (3) monosaccharide composition, and (4) modification on sialic acid
residues (O-substitution, NeuAc, and NeuGc) and the linkage position of sialic acid residues. The largest peak (49)
observed at ~50 min showed the molecular ion at 2386 that
is 42 mass units higher than the disialo-diantennary
oligosaccharide having 2 NeuAc residues (see Figure 4e).
This indicates that one of the NeuAc residues is substituted
with NeuAc having an O-acetyl group. Hydrolysis of the
oligosaccharide mixture (Figure 5e) in 10 mM trifluoracetic
acid (TFA) at 80 C for 1 h to remove sialic acids afforded
asialo-di- and -triantennary oligosaccharides (1 and 2
respectively in Table I; data not shown). We also observed
molecular ions of other peaks as shown in Table II and
found that many peaks contained acetyl groups. A few
examples are shown in Figure 4 (e-series). These data
indicate that the complex chromatogram of rat AGP oligosaccharides was due to sialic acids extensively modified with
N- and O-acyl groups, such as acetyl and glycolyl groups.
These data were well correlated with our previous data on
Table II. List of molecular masses for sialooligosaccharides found
in rat AGP
Abbreviation
[M ‡ 2AA-H]ÿ
1
2±0
1762
2
3±0
2127
4
2-N
2053
47
2-N-Ac
2095
48
2-NN-Ac2
2428
49
2-NN-Ac
2386
13
2-NN
2344
50
3-NNN-Ac3
3127
51
3-NNN-Ac2
3085
52
3-NNN-Ac
3043
19
3-NNN
3001
53
2-NNN-Ac3
2761
54
2-NNN-Ac2
2719
55
2-NNN-Ac
2677
23
2-NNN
2635
56
3-NNNN-Ac4
3460
57
3-NNNN-Ac3
3418
58
3-NNNN-Ac2
3376
59
3-NNNN-Ac
3334
30
3-NNNN
3292
Peak no.
For the peak numbers, see Figure 5e. Ac means acetyl group.
the analysis of sialic acids in various rat tissues (Morimoto
et al., 2001).
Analysis of oligosialyl units in bovine AGP
Tri- and tetrasialylated diantennary carbohydrate chains
found in bovine AGP have two possible structures. One is
the presence of the oligosaccharides which are substituted
with extra sialic acids at outer GlcNAc residues as observed
for NeuAca2-6Galb1-3(NeuAca2-6)GlcNAc branch in
fetuin triantennary chains. The other possibility is the
oligosialyl structure, and extra sialic acids bind the outermost sialic acid residues to form disialyl residues. Sato et al.
(1999) reported a method to detect oligosialyl residues
released by mild acid hydrolysis using 0.01 M TFA
followed by fluorometric labeling with 1,2-diamino-4,
5-methylenedioxybenzene (DMB).
We employed NeuAca2-8 NeuAc as model. Although
partial acid hydrolysis of NeuAca2-8NeuAc was observed
during hydrolysis even under mild conditions, the sialic acid
dimer was present as the major component even after
hydrolysis and derivatization with DMB. The dimer was
observed at 9 min using octadecyl silica column according
to the procedure reported previously (Morimoto et al.,
2001). The peak was observed earlier than that of NeuAc
(10.5 min). On the contrary, we could not observe any peaks
in this region other than NeuAc (10.5 min) and NeuGc
(7.7 min) when bovine AGP and oligosaccharide 38 were
analyzed after mild hydrolysis (data not shown).
We employed another approach to determine whether
bovine AGP contains oligosialyl residues or not. Sato et al.
(1998) also proposed a method to examine the presence of
oligosialyl group in glycoproteins using periodate oxidation
method. As shown in Figure 6, sialic acids at the nonreducing termini are oxidized with periodate to form a C-7 monosaccharide. On the contrary, the inner sialic acid, of which
hydroxyl group of the C8 position is involved in glycosidic
linkage to other sialic acid, is not oxidized with periodate.
Thus mild hydrolysis of the oligosaccharides affords a C-7
monosaccharide from the outermost sialic acid and intact
sialic acids from the inner part.
In this article, we treated a portion of bovine AGP with
periodate followed by reduction with sodium borohydride.
After hydrolysis of the modified AGP, the constituent sialic
acids were analyzed (Figure 6, route b). We also analyzed
sialic acids from bovine AGP after direct hydrolysis without
periodate oxidation (Figure 6, route a). The results were
shown in Figure 7.
Direct hydrolysis of bovine AGP gave two peaks at 7.7
min and 10.5 min due to N-glycolylneuraminic acid and
N-acetylneuraminic acid (dotted line in Figure 7). On the
contrary, periodate oxidation of bovine AGP showed two
products at 8.5 min and 11.5 min (solid line in Figure 7).
These two peaks were observed slightly later than those
observed after direct hydrolysis of bovine AGP, and
identified as C7(Neu5Gc)-DMB and C7(Neu5Ac)-DMB,
respectively, by comparison with the data reported by
Sato et al. (1998). These results obtained by mild acid
hydrolysis and periodate oxidation method clearly indicate that sialic acids in bovine AGP are not present as
disialyl residues.
437
M. Nakano et al.
Fig. 6. Principle to detect disialyl residues in bovine AGP. Direct acid hydrolysis allows determination of compositions of sialic acids (route a). By
hydrolysis after oxidation with periodate, the outermost sialic acids produce C-7 analogs. On the contrary, the inner sialic acids are not oxidized with
periodate, and produce intact sialic acids (route b).
Discussion
Fig. 7. Detection of disialyl residues in bovine AGP. The dotted line is
a chromatogram after direct hydrolysis of bovine AGP. The solid line is
a chromatogram after periodate oxidation treatment followed by
hydrolysis.
438
Except for human AGP, animal AGPs have received little
or no attention with regard to their glycans. We analyzed
oligosaccharides of AGP samples from human, bovine,
sheep, and rat and compared their patterns after labeling
with 2AA (Anumula and Dhume, 1998). The fluorescently
labeled sialic acid±containing oligosaccharides were
superbly resolved based on the number of sialic acids, and
isomers of oligosaccharides having sialic acids in different
positions were also resolved. In combination with MALDITOF MS measurement, we could identify the structures of
sialo-oligosaccharides after collection of the peaks, albeit
the positions where sialic acids were attached were not
determined.
AGP samples from human, bovine, sheep, and rat sera
manifested quite different sialo-oligosaccharides patterns.
Human AGP contained di-, tri-, and tetraantennary oligosaccharides, and some of the tri- and tetraantennary
oligosaccharides included a fucose residue to form sialyl
Hypersialylated oligosaccharides of a1-acid glycoprotein
Lewis x structures, as reported previously (Sei et al., 2002;
Stubbs et al., 1997). Rat AGP contained diantennary oligosaccharides as major oligosaccharides. An extremely complex pattern of the chromatogram indicates that rat AGP
contains highly acylated oligosaccharides as also shown
from MALDI-TOF MS measurement of the major peaks.
On the contrary, bovine AGP contained sialo-diantennary
oligosaccharides almost exclusively. Furthermore compositions of sialic acids were quite unique, and novel diantennary
chains having two NeuGc residues as well as two NeuAc
and both NeuAc and NeuGc residues were found abundantly. In addition, we found hypersialylated diantennary
oligosaccharides that contained three or four NeuAc and
NeuGc residues in various ratios.
We eliminated the possibility of the presence of disialyl
linkages by two different approaches using partial acid
hydrolysis and periodate oxidation. The results indicate
that each sialic acid is attached to different positions (i.e.,
Gal residues of nonreducing termini and GlcNAc residues
of Gal-GlcNAc branches). To the best of our knowledge,
there have been no reports on sialyl transferase regulating
biosynthesis of such hypersialylated oligosaccharides, and
further studies on their biosynthesis will be required.
Regulation of NeuGc biosynthesis is known in developing pig small intestine (Malykh et al., 2003). Sialic acids in
bovine fetal and adult tissues were analyzed, and NeuGc
was abundantly present in all bovine tissues (Schauer et al.,
1991). In bovine fetuin, oligosaccharides containing only
NeuAc were present almost exclusively as shown in the
present data (see Figure 2). We also found that NeuGc
was present in various digestive organs of mice and rats in
different ratios to NeuAc (Morimoto et al., 2001). There are
also many reports on adult animals having a higher proportion of NeuGc than young animals. AGP and fetuin are
produced in liver. At fetal and newborn stages, fetuin is
abundantly present in bovine sera. However, AGP is one
of the major acidic proteins at adult stages in mammals, and
fetuin is hardly detected. Establishment of the relationship
between fetal fetuin and AGP in adult stage will be a
challenging target for understanding biological regulation
of these proteins and their carbohydrate chains.
Materials and methods
Materials
Bovine fetuin was obtained from Gibco (Invitrogene,
Nihon-bashi, Chuo-ku, Tokyo). AGP samples of human,
bovine, sheep, and rat were from Sigma (St. Louis, MO).
NeuGc, a-chymotrypsin, and bicine were also purchased
from Sigma. TPCK-treated trypsin was from Worthington
(Lakewood, NJ). Sephadex LH-20 was from Amersham
Bioscience (Uppsala, Sweden). 2AA and sodium cyanoborohydride for fluorescent labeling of oligosaccharides
were from Tokyo Kasei (Chuo-ku, Tokyo). Peptide-N4(acetyl-b-D-glucosaminyl)asparagine amidase (N-glycoamidase F, E.C. 3.2.2.18) was from Roche Molecular
Biochemicals (Minato-ku, Tokyo). NeuAc was donated by
Drs. Tsukada and Ohta (Marukin-Bio, Uji, Kyoto, Japan).
Water purified with a Milli-Q purification system (Millipore, Shinagawa-ku, Tokyo) after double distillation of
deionized water was used for preparation of the eluent for
HPLC. Other reagents were of the highest grade commercially available.
Analysis of carbohydrate chains released from AGP
samples with N-glycoamidase F
Carbohydrate chains were released from the protein after
digesting with a mixture of trypsin and chymotrypsin as
reported recently (Nakano et al., 2003). Briefly, a sample
of AGP from animal sera (500 mg) was dissolved in 20 mM
bicine buffer (pH 8.0, 50 ml) and the solution was kept at
100 C for 10 min. After cooling, trypsin (5 mg) in bicine
buffer (5 ml) and chymotrypsin (5 mg) in the same buffer
(5 ml) were added to the mixture, and the mixture was incubated at 37 C overnight. After the mixture was kept on a
boiling water bath for 10 min, N-glycoamidase F (0.5 U, 1 ml)
was added to the mixture and kept at 37 C for 8 h, and the
mixture was again kept on a boiling water bath for 10 min.
Carbohydrate chains in the mixture thus obtained were
directly labeled with 2AA according to the method reported
previously (Anumula and Dhume, 1998). To the enzyme
reaction mixture, was added a solution (200 ml) of 2AA and
sodium cyanoborohydride, freshly prepared by dissolution
of both compounds (30 mg each) in methanol (1 ml) containing 4% sodium acetate and 2% boric acid. The mixture
was kept at 80 C for 1 h. After cooling, the solution was
applied to a column of Sephadex LH-20 (1 30 cm) equilibrated with 50% aqueous methanol. Earlier eluting fractions showing fluorescence at 410 nm with irradiating at
335-nm light were collected and evaporated to dryness. The
residue was dissolved in water (100 ml), a portion (10 ml) was
analyzed by HPLC, and the peaks were collected for MS
measurement.
HPLC of the fluorescent labeled carbohydrate chains
HPLC was performed with a Jasco apparatus equipped
with two PU-980 pumps and a Jasco FP-920 fluorescence
detector. Separation was done at 50 C with a polymerbased Asahi Shodex NH2P-50 4E column (Showa Denko,
Tokyo; 4.6 250 mm) using a linear gradient formed by
2% acetic acid in acetonitrile (solvent A) and 5% acetic acid
in water containing 3% triethylamine (solvent B). The column was initially equilibrated and eluted with 70% solvent
A for 2 min, at which point solvent B was increased to 95%
over 80 min and kept at this composition for further 100
min. The flow rate was 1.0 ml/min throughout the analysis.
Detection was performed by fluorometry with lex ˆ 350 nm
and lem ˆ 425 nm.
MALDI-TOF MS
MALDI-TOF mass spectra of AGP samples and the fluorescent labeled oligosaccharides were measured on a Voyager DE-PRO apparatus (PE Biosystems, Framingham,
MA). A nitrogen laser was used to irradiate samples at
337 nm, and an average of 50 shots was taken. The instrument was operated in linear mode using positive polarity
for proteins and negative polarity for oligosaccharides,
respectively, at an accelerating voltage of 20 kV. Samples
(~10 pmol, 0.5 ml each) were applied to a polished stainless
439
M. Nakano et al.
steel target, to which was added a solution (0.5 ml) of 2,5dihydroxybenzoic acid in a mixture of methanol±water
(1:1). The mixture was dried in atmosphere by keeping it
at room temperature for several minutes.
Detection of oligosialyl units
Bovine AGP or a fluorescent labeled oligosaccharide
collected by HPLC was dried and hydrolyzed in 0.01 N TFA
(20 ml) at 50 C for 1 h to release disialyl unit, and the
solution was directly derivatized with DMB (Dojin)
(Morimoto et al., 2001; Sato et al., 1999). In brief, to the
hydrolyzed solution was added 100 ml of 7 mM DMB solution containing 5.0 mM TFA, 1 M 2-mercaptoetanol and
18 mM sodium hydrosulfite. The mixture was incubated at
50 C for 2 h. In the similar manner, Neu5Ac, Neu5Gc, and
Neu5Aca2-8Neu5Ac (donated by Drs. Tsukada and Ohta
of Marukin-Bio, Kyoto) were derivatized with DMB and
used as the standard samples. A portion of the reaction
mixture was analyzed on an octadecyl silica column
(YMC-Pack ODS-A, 4.6 mm ID, 150 mm length, YMC
Co., Kyoto, Japan) using a Shimadzu SLC10A HPLC
apparatus with a Jasco FP-920 fluorometer at lem 448 nm
and lex 373 nm. The elution was performed in isocratic
mode using a mixture of methanol-acetonitrile-water
(14:2:84, v/v) at a flow rate of 0.9 ml/min at 40 C. At this
condition, NeuAc, NeuGc, and Neu5Aca2-8Neu5Ac were
observed at 10.5 min, 7.7 min, and 9.0 min, respectively.
Another approach to detect disialyl residues was performed according to the method reported by Sato et al.
Briefly, an aqueous solution (10 ml) of bovine AGP
(200 mg) was mixed with 50 mM sodium metaperiodata in
50 mM actate buffer (pH 5.0, 10 ml), and the mixture was
kept at room temperature for 10 min in the dark. After
addition of an aqueous solution of 10% ethyleglycol (10 ml)
followed by incubation of the mixture for further 15 min at
room temperature. 1 M sodium borohydride in saturated
sodium bicarbonate solution (10 ml) was added to the mixture and kept at room temperature for 15 min. To remove
the reagents, the reaction mixture was diluted with 300 ml
water and filtered through an ultrafiltration tube (molecular cutoff 10,000; Millipore). After washing the retentate
with water three times, the retentate was collected with
10 mM TFA (40 ml) and was kept for 1 h at 80 C to release
sialic acids by hydrolysis. A portion (20 ml) of the mixture
was derivatized with DMB and analyzed by HPLC in the
same manner as described.
Abbreviations
2AA, 2-aminobenzoic acid; AGP, a1-acid glycoprotein;
DMB, 1,2-diamino-4,5-methylenedioxybenzene; HPLC,
high-performance liquid chromatography; MALDI-TOF
MS, matrix-assisted laser-desorption/ionization time-of-flight
mass spectrometry; TFA, trifluoroacetic acid.
References
Anumula, K.R. and Dhume, S.T. (1998) High resolution and high
sensitivity methods for oligosaccharide mapping and characterization
by normal phase high performance liquid chromatography following
440
derivatization with highly fluorescent anthranilic acid. Glycobiology, 8,
685±694.
De Graaf, T.W., Van der Stelt, M.E., Anbergen, M.G., and van Dijk, W.
(1993) Inflammation-induced expression of sialyl Lewis X-containing
glycan structures on alpha1-acid glycoprotein (orosomucoid) in human
sera. J. Exp. Med., 177, 657±666.
Elliott, M.A., Elliott, H.G., Gallagher, K., McGuire, J., Field, M., and
Smith, K.D. (1997) Investigation into the concanavalin A reactivity,
fucosylation and oligosaccharide microheterogeneity of alpha1-acid
glycoprotein expressed in the sera of patients with rheumatoid
arthritis. J. Chromatogr. B, 688, 229±237.
Fournet, B., Montreuil, J., Strecker, G., Dorland, L., Haverkamp, J.,
Vliegenthardt, J.F.G., Binetter, J.P., and Schmid, K. (1978) Determination of the primary structures of 16 asialo-carbohydrate units
derived from human plasma alpha 1-acid glycoprotein by 360 MHz 1H
NMR spectroscopy and permethylation analysis. Biochemistry, 17,
5206±5214.
Fournier, T., Medjoubi, N.N., and Porquert, D. (2000) Alpha-1-acid
glycoprotein. Biochim. Biophys. Acta, 1482, 157±171.
Fukui, S., Feizi, T., Galstian, C., Lawson, A.M., and Chai, W. (2002)
Oligosaccharide microarrays for high-throughput detection and
specificity assignments of carbohydrate-protein interactions. Nature
Biotechnol., 20, 1011±1017.
Green, E.D., Adelt, G., Baenziger, J.U., Wilson, S., and Halbeek, H.V.
(1988) The asparagine-linked oligosaccharides on bovine fetuin.
Structural analysis of N-glycanase-released oligosaccharides by
500 MHz H1-NMR spectroscopy. J. Biol. Chem., 263, 18253±18268.
Hayase, T., Rice, K.G., Dziegielenwska, K.M., Kuhlenschmidt, M.,
Reilly, T., and Lee, Y.C. (1992) Comparison of N-glycosides of
fetuins from different species and human alpha2-HS-glycoprotein.
Biochemistry, 31, 4915±4921.
Hochepied, T., Van Molle, W., Berger, F.G., Baumann, H., and Libert, C.
(2000) Involvement of the acute phase protein alpha1-acid glycoprotein in nonspecific resistance to a lethal gram-negative infection. J. Biol.
Chem., 275, 14903±14909.
Kakehi, K., Kinoshita, M., and Nakano, M. (2002) Analysis of
glycoproteins and the oligosaccharides thereof by high-performance
capillary electrophoresisÐsignificance in regulatory studies on biopharmaceutical products. Biomed. Chromatogr., 16, 103±115.
Kawasaki, N., Itoh, S., Ohta, M., and Hayakawa, T. (2003) Microanalysis
of N-linked oligosaccharides in a glycoprotein by capillary liquid
chromatography/mass spectrometry and liquid chromatography/tandem mass spectrometry. Anal. Biochem., 316, 15±22.
Malykh, Y.N., King, T.P., Logan, E., Kelly, D., Schauer, R., and Shaw,
L. (2003) Regulation of N-glycolylneuraminic acid biosynthesis in
developing pig small intestine. Biochem. J., 370, 601±607.
Matsumoto, K., Nishi, K., Tokutomi, Y., Irie, T., Suenaga, A., and
Otagiri, M. (2003) Effects of alpha 1-acid glycoprotein on erythrocyte
deformability and membrane stabilization. Biol. Pharm. Bull., 26,
123±126.
Morimoto, N., Nakano, M., Kinoshita, M., Kawabata, A., Morita, M.,
Oda, Y., Kuroda, R., and Kakehi, K. (2001) Specific distribution of
sialic acids in animal tissues as examined by LC-ESI-MS after
derivatization with 1,2-diamino-4,5-methylenedioxybenzene. Anal.
Chem., 73, 5422±5428.
Nakajima, K., Oda, Y., Kinoshita, M., and Kakehi, K. (2003) Capillary
affinity electrophoresis for the screening of post-translational
modification of proteins with carbohydrates. J. Proteome Res., 2,
81±88.
Nakano, M., Kakehi, K., and Lee, Y.C. (2003) Sample clean-up method
for analysis of complex-type N-glycans released from glycopeptides.
J. Chromatogr. A, 1005, 13±21.
Sarcione, E.J. (1967) Hepatic synthesis and secretory release of plasma
alpha-2 (acute phase)-globulin appearing in malignancy. Cancer Res.,
27, 2025±2033.
Sato, C., Inoue, S., Matsuda, T., and Kitajima, K. (1998) Development of
a highly sensitive chemical method for detecting a2-8-linked oligo/
polysialic acid residues in glycoprotein blotted on the membrane. Anal.
Biochem., 261, 191±197.
Sato, C., Inoue, S., Matsuda, T., and Kitajima, K. (1999) Fluorescentassisted detection of oligosyalyl units in glycoconjugates. Anal.
Biochem., 266, 102±109.
Hypersialylated oligosaccharides of a1-acid glycoprotein
Schauer, R., Stoll, S., and Reuter, G. (1991) Differences in the amount
of N-acetyl- and N-glycolyl-neuraminic acid, as well as O-acylated
sialic acids, of fetal and adult bovine tissues. Carbohydr. Res., 213,
353±359.
Schmid, K., Nimberg, R.B., Kimura, A., Yamaguchi, H., and Binette, J.P.
(1977) The carbohydrate units of human plasma alpha1-acid
glycoprein. Biochim. Biophys. Acta, 492, 291±302.
Sei, K., Nakano, M., Kinoshita, M., Masuko, T., and Kakehi, K. (2002)
Collection of alpha1-acid glycoprotein molecular species by capillary
electrophoresis and the analysis of their molecular mass and
carbohydrate chains. J. Chromatogr. A, 985, 273±281.
Sorensson, J., Matejka, G.L., Ohlson, M., and Haraldsson, B. (1999)
Human endothelial cells produce orosomucoid, an important
component of the capillary barrier. Am. J. Physiol., 276,
H530±H534.
Stubbs, H.J., Shia, M.A., and Rice, K.G. (1997) Preparative purification
of tetraantennary oligosaccharides from human asialyl orosomucoid.
Anal. Biochem., 247, 357±365.
Townsend, R.R., Hardy, M.R., Wong, T.C., and Lee, Y.C. (1986) Binding
of N-linked bovine fetuin glycopeptides to isolated rabbit hepatocytes:
Gal/GalNAc hepatic lectin discrimination between Gal beta(1,4)
GlcNAc and Gal beta(1,3)GlcNAc in a triantennary structure.
Biochemistry, 25, 5716±5725.
Treuheit, M.J., Costello, C.E., and Halsall, H.B. (1992) Analysis of the
five glycosylation sites of human alpha 1-acid glycoprotein. Biochem.
J., 283, 105±112.
441