RAPID COMMUNICATIONS IN MASS SPECTROMETRY

Rapid Commun. Mass Spectrom. 2004; 18: 29832988

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rcm.1718
Site-specic carbohydrate proling of human transferrin
by nano-ow liquid chromatography/electrospray
ionization mass spectrometry
Yoshinori Satomi
1
, Yasutsugu Shimonishi
1{
, Toshiharu Hase
2
and Toshifumi Takao
1
*
1
Laboratory of Protein Proling and Functional Proteomics, Institute for Protein Research, Osaka University, Suita, Osaka 565-0871, Japan
2
Division of Enzymology, Institute for Protein Research, Osaka University, Suita, Osaka 565-0871, Japan
Received 12 September 2004; Revised 11 October 2004; Accepted 11 October 2004
Glycopeptides derived from a lysylendopeptidase digest of commercially available human trans-
ferrin were analyzed by nano-ow liquid chromatography/electrospray ionization mass spectrome-
try (LC/ESI-MS), which permitted the carbohydrate proles at Asn
432
and Asn
630
to be determined.
Both are located in a well-known motif for N-glycosylation, Asn-Xaa-Ser/Thr. The contents of the
carbohydrates at each site were signicantly different from each other, and consisted of a variety of
minor types of oligosaccharides in addition to the major one, a biantennary complex-type oligosac-
charide. Nano-ow ESI tandem mass spectrometry (MS/MS) of the glycopeptides (Cys421-Lys433
and Ile619-Lys646) containing these two sites yielded predominantly ions originating from the non-
reducing termini (oxonium ions) and reducing terminus, resulting from cleavage of the glycosidic
bonds of the carbohydrate moieties; this permitted the structural read-out of a small minority of
the carbohydrate moieties. In particular, the observation of oxonium ions at m/z 512.2 and 803.2
is useful for probing outer non-reducing terminal fucosylation, which represented carbohydrate
structures consisting of Hex, dHex, and HexNAc, and NeuNAc, Hex, dHex, and HexNAc, respec-
tively, from which the Lewis X structure (Galb1-4(Fuca1-3)GlcNAc) was readily deduced. Moreover,
fucosylation at the reducing-terminal GlcNAc (Fuca1-6GlcNAc) specically occurred at Asn
630
, as
demonstrated by treatment of the glycopeptides with a1-3/4-L-fucosidase. Copyright # 2004 John
Wiley & Sons, Ltd.
The importance of structural analyses of sugar chains of gly-
coconjugates has signicantly increased since their func-
tional properties have become better understood. However,
the characterization of oligosaccharide structures of glyco-
proteins remains somewhat incomplete owing to the com-
plexities of such structures, as represented by heterogeneity
andbranching structures, and difculties associated with the
exhaustive detection of all components.
Mass spectrometry (MS) in conjunction with electrospray
ionization (ESI) or matrix-assisted laser desorption/ioniza-
tion (MALDI) has been successfully used for the structural
analysis of carbohydrate moieties of glycoconjugates such as
those linked to a glycoprotein, and is capable of revealing not
only the molecular mass of an intact sugar chain, but also the
arrangement of the constituent sugar units. The increased
capabilities of MS for the structural analysis of glycoconju-
gates should permit an overall proling of carbohydrate
moieties ona glycoprotein.
19
However, the detection of very
minor components in a complex mixture of glycopeptides,
such as those produced by the enzymatic digestion of a
glycoprotein, continues to pose a challenge.
In this study we report on the use of nano-ow liquid
chromatography (LC)ESI-MS and MS/MS to analyze glyco-
peptides derived from a commercially available human
transferrin, the major carbohydrate moieties of which are
well knowntobe biantennarycomplex-type oligosaccharides
locatedat Asn
432
andAsn
630
, bothof whichare present within
the consensus sequence (Asn-Xaa-Ser/Thr) of N-glycosyla-
tion.
10,11
During the course of the carbohydrate proling via
the analysis of the glycopeptides, we detected several minor
glycosylation types that were specic for each site. Further-
more, ESI-MS/MS and a1-3/4-L-fucosidase treatment of
the glycopeptides revealed an antenna-fucosylated form, a
core-fucosylated form, or both, at each site; both of these
fucosylated forms have been identied in transferrin derived
from human amniotic uid
12
or in hepatoma-derived
transferrin as aberrant glycosylations.
13
EXPERIMENTAL
Materials
Human transferrin (T2252) was purchased from Sigma Che-
mical Co. (St. Louis, MO, USA). Achromobacter lyticus protease
Copyright # 2004 John Wiley & Sons, Ltd.
*Correspondence to: T. Takao, Laboratory of Protein Proling and
Functional Proteomics, Institute for Protein Research, Osaka
University, Yamadaoka 3-2, Suita, Osaka 565-0871, Japan.
E-mail: tak@protein.osaka-u.ac.jp
{
Current address: Nagahama Institute of Bio-Science and Tech-
nology, Nagahama, Shiga 526-0829, Japan.
Contract/grant sponsor: Ministry of Education, Culture,
Sports, Science and Technology of Japan; contract/grant num-
ber: 15GS0320.
Contract/grant sponsor: Ministry of Health, Labour and
Welfare of Japan.
I (lysylendopeptidase (LEP)) was obtained from Wako Pure
Chemical Industries (Osaka, Japan). Neuraminidase (siali-
dase) and a1-3/4-L-fucosidase were purchased from New
England Biolabs (Massachusetts, USA) and Takara Biomedi-
cals (Tokyo, Japan), respectively. All reagents were of analy-
tical grade and were used without further purication.
Preparation of glycopeptides
S-Carboxymethylated human transferrin was prepared as
described elsewhere,
9
and digested with LEP (E/S 1:100)
at 378C for 10 h.
Exoglycosidase treatments
The solution of the LEP digest obtained above was lyophi-
lized, redissolved in 10 mM citrate buffer (pH 6.0), and trea-
ted with neuraminidase (Clostridium perfringens) followed by
a1-3/4-L-fucosidase (Streptocyces sp.142), with enzyme/sub-
strate (E/S) ratios of 50 and 5 munit/5pmol, by incubating at
378C for 2 and 6 h, respectively.
Nano-ow LC/ESI-MS
The LEP peptides (ca. 2 pmol) were separated by using a
C
18
Pepmap column (0.075 100 mm, Dionex, Idstein,
Germany), using a Ultimate nano-LCsystem(Dionex). Alin-
ear gradient with solvent A (0.5% formic acid in water) and
solvent B (0.5% formic acid in acetonitrile) was used for the
separation. The peptides were eluted by increasing solvent
B from 5 to 25% over 55 min and then 25 to 50% over 20 min
at a ow rate of 300 nL/min. The eluates were monitored
at 214 and 280 nm, and continuously introduced into a
nano-ow ESI source. The ions produced were analyzed by
a Q-TOF mass spectrometer (Micromass, Manchester, UK),
a hybrid quadrupole-orthogonal acceleration tandem mass
spectrometer. Mass calibration was performed using cluster
ions derived from NaI.
Low-energy collision-induced dissociation (CID) MS/MS
was carried out as described elsewhere.
9
Briey, a precursor
ion was fragmented in a collision cell using argon as the
collision gas and an appropriate collision energy (2035 eV).
Figure 1. (a) Total ion current chromatogram of the LEP digest of S-carboxymethylated
human transferrin (2pmol). Glycopeptides containing the well-known glycosylation sites
(Asn
432
and Asn
630
) and a minor glycosylation site (Asn
491
)
9
were eluted in fractions 2, 3,
and 1, respectively. ESI-MS spectra of fractions 1 (b), 2 (c) and 3 (d) in the typical mass
ranges, and the expanded mass range of d (e), giving the quadruply and quintuply charged
ions of various glycoforms (Table 1) of the glycopeptides (Ile
490
-Lys
508
, Cys
421
-Lys
433
and
Ile
619
-Lys
646
), respectively. Each of these multiply charged ions was subjected to low-energy
CID MS/MS (see Figs. 2 and 3).
Copyright # 2004 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2004; 18: 29832988
2984 Y. Satomi et al.
MS/MS data were processed by the maximum entropy data
enhancement program, MaxEnt 3
TM
(Micromass), which is
capable of deconvoluting a spectrum with peaks in a variety
of charge states, thus producing a simplied spectrum
consisting of only monoisotopic peaks independent of charge
state.
RESULTS AND DISCUSSION
In order to obtain site-specic proles of the carbohydrates at
Asn
432
and Asn
630
of human transferrin, carboxymethylated
transferrin, after treatment with LEP, was subjected to nano-
ow LC/ESI-MS (Fig. 1(a)). The glycopeptides containing
Asn
432
and Asn
630
, both of which were assumed to be fully
glycosylated on the basis of the total ion current chromato-
gram, were eluted in peak fractions 2 and 3, respectively; gly-
copeptides containing Asn
491
, which we recently found to be
present at a level of ca. 2 mol % in the same preparation of
transferrin,
9
were detected as a minor component in peak
fraction 1. The molecular masses observed for these glyco-
peptides could be used to reveal the modes of glycosylation
at Asn
432
and Asn
630
, and also at Asn
491
. Moreover, the peak
intensities of each ion species reected their relative abun-
dances (Fig. 1), since the ionization efciencies of glycopep-
tides differing only in the carbohydrate moieties are
negligibly affected by differences in these moieties. As a
result, the glycosylation sites Asn
432
and Asn
630
were found
to be glycosylated with several types of oligosaccharides,
including the abundant biantennary complex-type, while
the minor glycosylation site Asn
491
was detected only as
occupied by the biantennary complex-type (Table 1).
The carbohydrate structures and their relevance to the
biological activity of human transferrin have been exten-
sively studied in relation to several diseases,
13,14
uid
specicity,
12
etc. Yamashita et al. reported an increase in
highly branched sugar chains, including core- and antenna-
fucosylated structures, in hepatoma transferrin as aberrant
glycosylation in comparison with those from normal indivi-
duals or a commercial source
13
(Table 1). van Rooijen et al.
demonstrated a sialyl Le
X
element on the antennae of the
sugar chains derived from amniotic uid transferrin using a
nuclear magnetic resonance method.
12
However, little information is available concerning the
glycosylationmodes at eachglycosylationsite. As depictedin
Table 1, the glycosylation forms obtained in the present
study, irrespective of non-reducing terminal sialic acids,
have been reported previously. However, in the present
work, a di-fucosylated biantennary form, a relatively minor
fraction (0.3% of the total carbohydrates at Asn
630
), was
detected for the rst time in human transferrin (Fig. 1(e),
Table 1). ESI-MS/MS analysis of this glycopeptide gave ions
at m/z 3693.7 and 512.2 (data not shown), which can be
attributed to the core- and antenna-fucosylated form (V),
respectively (see below), although the occurrence of di-
fucosylation in the two antennae cannot be completely ruled
out. It is noteworthy that, since this analysis allowed a site-
specic carbohydrate proling (at Asn
432
, Asn
630
and
Asn
491
), the data in Fig. 1 suggest a signicant difference in
the abundances of a mono-fucosylated biantennary form
between Asn
432
(1.6%, III) and Asn
630
(9.7%, a sum of III and
IV) (Table 1). Fucosylation on a complex-type oligosacchar-
ide sugar chain is known to take place at the reducing
Table 1. Relative abundance (%) of the glycoforms identied at Asn
432
, Asn
491
and Asn
630
of human transferrin
a
a
The relative abundance of each glycoformat Asn
491
, Asn
432
and Asn
630
was calculated, based on the signal intensities of the corresponding gly-
copeptides obtained in Figs. 1(b), 1(c) and 1(d), respectively. The relative abundances reported by Yamashita et al.
13
and van Rooijen et al.
12
were
obtainedusingthe total free oligosaccharides (asialoforms) liberatedfromcommercial, normal individuals andhepatocellular carcinoma patient-
derived transferrin, and amniotic uid derived transferrin, respectively.
b
Carbohydrate structures are represented by the symbol notation: solid stars, NeuNAc; solid circles, Gal; solid squares, GlcNAc; open circles,
Man; solid triangles, Fuc. The molecular masses (M) of each glycopeptide were obtained, based on the observed m/z in Figs. 1(b), 1(c) and 1(d):
I, 3681.5; II, 3390.4; III, 3827.5; IV and V, n.d.; VI, 4046.6; VII, n.d. (Cys
421
-Lys
433
); I, 5548.3; II, 5257.2; III and IV, 5694.4; V, 5840.3; VI, 5913.4; VII,
6059.3 (Ile
619
-Lys
646
).
Site-specic carbohydrate proling of human transferrin 2985
Copyright # 2004 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2004; 18: 29832988
terminus (Fuca1-6GlcNAc) and/or proximal to the outer
non-reducing termini (Fuca1-2Gal or Fuca1-4GlcNAc or
Fuca1-3GlcNAc).
15,16
(Abbreviations used here are: Neu-
NAc, N-acetylneuraminic acid; Hex, hexose; HexNAc, N-
acetylhexosamine; dHex, deoxyhexose.)
In order to further characterize the mono-fucosylated
biantennary forms linked to Asn
432
and Asn
630
, the corre-
sponding glycopeptides were directly subjected to low-
energy CID ESI-MS/MS. Oxonium ions in the low-mass
region and reducing-terminal ions in the high-mass region,
which retained positive charges in the peptide portion, were
observed as major ion species. The reducing-terminal ions,
formed by truncation of the non-reducing terminal sugar
units, readily revealed the relative alignment of the consti-
tuent sugars as well as the branching structure in the non-
reducing termini (see the schemes in Figs. 2 and 3). In
Figure 2. Low-energy CID ESI-MS/MS spectrum from [M4H]
4
at m/z 957.9 of the glycopeptide
(Cys
421
-Lys
433
) eluted in fraction 2 (see Fig. 1). Multiply charged ions in the raw data were
transformed to singly charged ones using MaxEnt3
TM
. The arrows ( and !) indicate the
arrangement of the constituent sugar units from the non-reducing termini based on the reducing
terminal ions and oxonium ions, respectively, which were produced upon the cleavage of glycosidic
bonds. One of the possible assignments of the observed masses (monoisotopic) as the mono-
fucosylated biantennary oligosaccharide is depicted.
2986 Y. Satomi et al.
Copyright # 2004 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2004; 18: 29832988
particular, the reducing-terminal ion at m/z 3693.7 observed
for the Asn
630
-containing glycopeptide couldbe assignedto a
core-fucosylated form (Fuca1-6GlcNAc-peptide) (Fig. 3),
whereas the corresponding ion for the Asn
432
-glycopeptide
was barely observed. These results indicate that reducing-
terminal fucosylation at GlcNAc at Asn
630
was favored over
that at Asn
432
. Meanwhile, the oxonium ions observed in
Figs. 2 and 3 provided a signature of the non-reducing
terminal carbohydrate structures of both glycopeptides. In
particular, the oxonium ions at m/z 512.2 (the sum of Hex,
dHex, HexNAc)
8
and 803.2 (the sumof NeuNAc, Hex, dHex,
HexNAc),
7
indicative of an antenna-fucosylated form, i.e. a
Figure 3. Low-energy CID ESI-MS/MS spectrum from [M5H]
5
at m/z 1139.5 of the
glycopeptide (Ile
619
-Lys
646
) eluted in fraction 3 (see Fig. 1). Multiply charged ions in the raw data
were transformed using MaxEnt3
TM
. The arrows ( and !) indicate the arrangement of the
constituent sugar units from the non-reducing termini based on the reducing terminal ions and
oxonium ions, respectively, which were produced upon the cleavage of glycosidic bonds. One of the
possible assignments of the observed masses (monoisotopic) as the mono-fucosylated biantennary
oligosaccharide is depicted. The assignments of the oxonium ions proximal to the outer non-
reducing terminus are depicted in the scheme in Fig. 2.
Site-specic carbohydrate proling of human transferrin 2987
Copyright # 2004 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2004; 18: 29832988
Le
X
element (Galb1-4(Fuca1-3)GlcNAc) by analogy, were
observed for both glycopeptides, which have been pre-
viously reported for human amniotic uid derived transfer-
rin
12
and hepatoma transferrin as aberrant glycosylation.
13
As a result, the mono-fucosylated biantennary oligosacchar-
ide (9.7% of the carbohydrates at Asn
630
) linked to Asn
630
consists of an antenna-fucosylated form (III) corresponding
to a Le
X
element as well as a core-fucosylatedform(IV), while
another mono-fucosylated form at Asn
432
, which was
obtained with much lower relative abundance (1.6% of the
carbohydrates at Asn
432
), can be attributed to only a Le
X
element (III).
Since ion intensities of the product ions observed in a MS/
MS spectrum do not reect the populations of the corre-
sponding fragments, those of the ions at m/z 512.2 and 3693.7
observed in the same spectrum (Fig. 3) for the mono-
fucosylated form at Asn
630
, which were the signatures of an
antenna-fucosylated form (III) and core-fucosylated form
(IV), respectively, could not provide information on the
relative ratio of these two forms. We then compared the LC/
ESI-MS spectra of LEPdigests before andafter treatment with
a1-3/4-L-fucosidase, following neuraminidase treatment
(see Experimental). The Fuc residues in the glycopeptides
containing Le
X
elements were removed fromthe antennae by
this treatment, resulting in a shift to lower m/z by 146 Da
(Fuc). As a result, 2.8% of 9.7% of the monofucosylated form
at Asn
630
turnedout tocorrespondtothe Le
X
-containingform
(III), the remainder (6.9%) of which was a core-fucosylated
form (IV) that had been resistant to a1-3/4-L-fucosidase
treatment (Table 1). The relative abundance of the core-
fucosylated form (IV), calculated on the basis of the total
carbohydrate content, was ca. 3.5%, a value that is close to
that obtained (4.6%) for the commercial transferrin in the
previous study,
13
although the Le
X
-containing form(III) was
not obtained (Table 1). The signal corresponding to the di-
fucosylated form (V), observed with much lower abundance
(Fig. 1(e)), disappeared and was likely to shift to that of the
mono-fucosylated form (IV) after the a1-3/4-L-fucosidase
treatment.
CONCLUSIONS
Glycopeptides were successfully analyzedby nano-owLC/
ESI-MS and MS/MS, providing relevant structural data on
the carbohydrates in a site-specic manner, even for the min-
or sugar chains. The peptide portion, which retains a positive
charge ina positive-ionmode measurement, permits not only
the highly sensitive detection of glycopeptides by nano-ow
LC/ESI-MS (10 fmol of transferrin was required to detect the
mono-fucosylatedbiantennary forms reportedhere), but also
the efcient detection using low-energy CID of reducing-
terminal ions formedby truncationof the non-reducing term-
inal sugar units, even though the peptide portion fragments
poorly. In addition, this analysis of glycopeptides, which
requires a simple preparation process, should be valid for
protecting sialic acids at the non-reducing termini frombeing
cleaved; such residues are labile with respect to various che-
mical treatments that are typically used in the preparation of
free or chemically derivatized oligosaccharides. Further-
more, a signicant difference in the glycosylation modes
between Asn
432
and Asn
630
was clearly revealed. The com-
mercial transferrin used in this study was a pooled material,
but was derived from human serum, and its carbohydrate
content might vary frombatch-to-batch. However, the results
herein demonstrate the potential for characterization of the
site-specic heterogeneity of glycosylation in human trans-
ferrin that could be involved in some biological events, and
the utility of the nano-owLC/ESI-MS technique for probing
such heterogeneity of glycosylation in a complex mixture.
Acknowledgements
This study was supported by a Grant-in-Aid for Creative
Scientic Research (15GS0320 to T. T.) from the Ministry of
Education, Culture, Sports, Science andTechnology of Japan,
and a project on Research on Proteomics from the Ministry
of Health, Labour and Welfare of Japan.
REFERENCES
1. Huddleston MJ, Bean MF, Carr SA. Anal. Chem. 1993; 65:
877.
2. Schindler PA, Settineri CA, Collet X, Fielding CJ, Burlin-
game AL. Protein Sci. 1995; 4: 791.
3. Stimson E, Hope J, Chong A, Burlingame AL. Biochemistry
1999; 38: 4885.
4. Sullivan B, Addona TA, Carr SA. Anal. Chem. 2004; 76: 3112.
5. Demelbauer UM, Zehl M, Plematl A, Allmaier G, Rizzi A.
Rapid Commun. Mass Spectrom. 2004; 18: 1575.
6. Wuhrer M, Hokke CH, Deelder AM. Rapid Commun. Mass
Spectrom. 2004; 18: 1741.
7. Dage JL, Ackermann BL, Halsall HB. Glycobiology 1998; 8:
755.
8. Mo W, Sakamoto H, Nishikawa A, Kagi N, Langridge JI,
Shimonishi Y, Takao T. Anal. Chem. 1999; 71: 4100.
9. Satomi Y, Shimonishi Y, Takao T. FEBS Lett. 2004; 576: 51.
10. Spik G, Bayard B, Fournet B, Strecker G, Bouquelet S,
Montreuil J. FEBS Lett. 1975; 50: 296.
11. Dorland L, Haverkamp J, Schut BL, Vliegenthart JF. FEBS
Lett. 1977; 77: 15.
12. van Rooijen JJ, Jeschke U, Kamerling JP, Vliegenthart JF.
Glycobiology 1998; 8: 1064.
13. Yamashita K, Koide N, Endo T, Iwaki Y, Kobata A. J. Biol.
Chem. 1989; 264: 2415.
14. Coddeville B, Carchon H, Jaeken J, Briand G, Spik G.
Glycoconjugate J. 1998; 15: 265.
15. Guile GR, Harvey DJ, ODonnell N, Powell AK, Hunter AP,
Zamze S, Fernandes DL, Dwek RA, Wing DR. Eur. J.
Biochem. 1998; 258: 623.
16. Dupuy F, Germot A, Marenda M, Oriol R, Blancher A,
Julien R, Maftah A. Mol. Biol. Evol. 2002; 19: 815.
2988 Y. Satomi et al.
Copyright # 2004 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2004; 18: 29832988