Anal. Chem. 2001, 73, 5875-5885
Capillary Electrophoresis-Electrospray Mass
Spectrometry for the Characterization of
High-Mannose-Type N-Glycosylation and
Differential Oxidation in Glycoproteins by Charge
Reversal and Protease/Glycosidase Digestion
Tao Liu, Jia-Da Li, Rong Zeng, Xiao-Xia Shao, Ke-Yi Wang, and Qi-Chang Xia*
Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences,
320 Yue-Yang Road, Shanghai 200031, People’s Republic of China
The characterization of high-mannose-type N-glycosylation
by capillary electrophoresis-electrospray mass spectrometry (CE-ESI MS) was described. In addition to the use
of a cationic noncovalent capillary coating, strong acidic
buffer, and charge reversal to increase the glycoform
resolving power, N-glycosidase F (PNGase F) combined
with a basic protease and r-mannosidase combined with
an acidic protease were used to analyze the high-mannosetype N-glycosylation in ribonuclease B (RNase B) and in
a novel C-type lectin from the venom of Trimeresurus
stejnegeri (TSL). The structures of oligosaccharide, glycosylation sites, and glycoform distributions were determined simultaneously, and the differential oxidation of
Met residues in glycopeptides obtained from TSL protease
digestion was also characterized successfully by CE-MS/
MS. The results showed that the oligosaccharide attached
to RNase B has a structure of GlcNAc2Man5∼9, and that
attached to TSL has a structure of GlcNAc2Man5∼8. The
glycoform distributions in these glycoproteins are quite
different, with the GlcNAc2Man5 type predominant in
RNase B, and the GlcNAc2Man8 type, in TSL. This method
may be useful not only for the characterization of glycosylation sites and glycan structures, but also for the
determination of the relative abundance of individual
glycoforms.
Glycosylation has long been recognized as one of the most
important posttranslational events affecting the functions of
proteins in health and disease. The intracellular or intercellular
events can be regulated by modifying the glycoform distribution
of glycoproteins.1 Differential glycosylation is a major source of
protein microheterogeneity.2 However, studies on the structure
of glycans are much retarded in comparison with proteins and
DNAs. Full structural characterization of glycans requires the
defining of branching, linkages, and configurations and the
identification of sugar isomers. Furthermore, oligosaccharides can
be linked to Ser or Thr residues (O-glycosylation) or to Asn
* To whom correspondence should be addressed. Fax: (+86-21) 64333781.
E-mail: [email protected].
(1) Dwek, R. A. Science 1995, 269, 1234-1235.
(2) Dwek, R. A.; Edge, C. J.; Harvey, D. J.; Wormald, M. R.; Parekh, R. B. Annu.
Rev. Biochem. 1993, 62, 65-100.
10.1021/ac0106748 CCC: $20.00
Published on Web 11/08/2001
© 2001 American Chemical Society
residues (N-glycosylation), and glycoprotein can have different
oligosaccharides attached to any given possible sites, so the
analysis of glycoprotein is even more difficult. Fortunately,
N-glycosylation normally occurs at the Asn residue in a tripeptide
recognition signal Asn-X-Ser/Thr (where X is any amino acid
residue except Pro or Asp).3 The N-linked oligosaccharides can
be classified into high-mannose, complex, and hybrid types. They
all share a common core pentasaccharide structure composed of
GlcNAcβ1-4GlcNAcβ1-4ManR1-3Man(R1-6Man) but differ from each
other in appending a variable number of different saccharide
residues on the pentasaccharide core. High-mannose-type oligosaccharide can have two to nine additional mannose residues
linked to the pentasaccharide core by an R-glycosidic bond.3 Thus,
it is possible to utilize protease and glycosidase combined
digestion to fully define the primary structures of oligosaccharides
of this type.
Generally, glycoform analysis consists of the separation of
intact glycoprotein glycoforms, structural characterization of the
released glycan pool, and determination of glycosylation sites and
glycoform distribution. In the past decade, methodologies for
resolving glycoforms by capillary electrophoresis (CE) were
developing rapidly by using basic or borate buffers4-6 or strong
acidic buffers in zone electrophoresis mode7,8 or by using different
capillary coatings.8-10 Recently, Pacakova et al.11 reported the
effects of electrolyte modification and capillary coating on the
separation of glycoprotein isoforms by CE. On the other hand,
mass spectrometry (MS) has become a powerful tool for the
detection of oligosaccharides and glycoconjugates with the invention of “soft ionization” methods, such as electrospray ionization
(3) Kornfeld, R.; Kornfeld, S. Annu. Rev. Biochem. 1985, 54, 631-664.
(4) Watson, E.; Yao, F. Anal. Biochem. 1993, 210, 389-393.
(5) James, D. C.; Freedman, R. B.; Hoare, M.; Jenkins, N. Anal. Biochem. 1994,
222, 315-322.
(6) Che, F. Y.; Song, J. F.; Shao, X. X.; Wang, K. Y.; Xia, Q. C. J. Chromatogr.
A 1999, 849, 599-608.
(7) Wiktorowicz, J. E.; Cloburn, J. C. Electrophoresis 1990, 11, 769-773.
(8) Kelly, J. F.; Locke, S. J.; Ramaley, L.; Thibault, P. J. Chromatogr. A 1996,
720, 409-427.
(9) Katayama, H.; Ishihama, Y.; Asakawa, N. Anal. Chem. 1998, 70, 52725277.
(10) Yeung, B.; Porter, T. J.; Vath, J. E. Anal. Chem. 1997, 69, 2510-2516.
(11) Pacakova, V.; Hubena, S.; Ticha, M.; Mad, ra M.; Stulik, K. Electrophoresis
2001, 22, 459-463.
Analytical Chemistry, Vol. 73, No. 24, December 15, 2001 5875
(ESI) and matrix-assisted laser desorption/ionization (MALDI).
Oligosaccharides can be conveniently characterized by MS either
in underivatized form12-15 or in derivatized form.16-18 MS has also
been used to characterize glycosylation by peptide mass mapping
(PMM)19 or in-source fragmentation.20 In the proteome analysis
of glycoforms reported by Packer et al.,21 glycoproteins separated
by 2-dimsensional polyacrylamide gel electrophoresis can be
characterized in microscale. Reviews on glycoprotein structure
determination by MS have been published recently.22-26
On-line coupling of CE to MS (CE-MS) has been proved to
be an extremely useful tool for oligosaccharide and glycoprotein
characterization. Boss et al.27 reported on the analysis of recombinant human erythropoietin by capillary zone electrophoresis
mass spectrometry (CZE-MS). Alternatively, on-line capillary
isoelectric focusing-electrospray ionization mass spectrometry
(CIEF-ESI-MS) as a 2-dimensional separation system has been
applied to the high-resolution analysis of bovine serum apotransferrin glycoforms.28 In this work, we have used dynamic cationic
capillary coating, strong acidic buffer, and charge reversal for the
separation of glycoforms. According to the suitable pH range of
enzymes, N-glycosidase F (PNGase F) and trypsin (or Lys-C
protease) or R-mannosidase and Glu-C protease were used in pairs
for comparative peptide mass mapping. First, RNase B as a model
protein was examined, and then the method was applied to a novel
C-type lectin from the venom of Trimeresurus stejnegeri (TSL).
Interestingly, the process-induced differential oxidation of Met
residues within the TSL was also detected by this method.
EXPERIMENTAL SECTION
Materials. All chemicals used were of analytical grade if not
stated otherwise. Bovine pancreatic RNase B, endoproteinase LysC, R-mannosidase (from Jack Beans), hexadimethrine bromide
(Polybrene), dithiothreitol (DTT), iodoacetamide (IAA), formic
acid (>99%), and ammonium bicarbonate were purchased from
Sigma (St. Louis, MO); HPLC-grade methanol (MeOH) and
acetonitrile (ACN) were obtained from Fisher (Fair Lawn, NJ);
(12) Kelly, J.; Masoud, H.; Perry, M. B.; Richards J. C.; Thibault, P. Anal. Biochem.
1996, 233, 15-30.
(13) Bateman, K. P.; Banoub, J. H.; Thibault, P. Electrophoresis 1996, 17, 18181828.
(14) Mechref, Y.; Novotny, M. V. Anal. Chem. 1998, 70, 455-463.
(15) Weiskopf, A. S.; Vouros, P.; Harvey, D. J. Anal. Chem. 1998, 70, 44414447.
(16) Jeso, B. D.; Liguoro, D.; Ferranti, P.; Marinaccio, M.; Acquaviva, R.;
Formisano, S.; Consiglio, E. J. Biol. Chem. 1992, 267, 1938-1992.
(17) Okafo, G.; Burrow, L.; Carr, S. A.; Roberts, G. D.; Johnson, W.; Camilleri,
P. Anal. Chem. 1996, 68, 4424-4430.
(18) Charlwood, J.; Birrell, H.; Gribble, A.; Burdes, V.; Tolson, D.; Camilleri, P.
Anal. Chem. 2000, 72, 1453-1461.
(19) Tsarbopoulos, A.; Karas, M.; Strupat, K.; Pramanlk, B. N.; Nagabhushan,
T. L.; Hillenkamp, F. Anal. Chem. 1994, 66, 2062-2070.
(20) Mazsaroff, I.; Yu, W.; Kelley, B. D.; Vath, J. E. Anal. Chem. 1997, 69, 25172524.
(21) Packer, N. H.; Pawlak, A.; Kett, W. C.; Gooley, A. A.; Redmond, J. W.;
Williams, K. L. Electrophoresis 1997, 18, 452-460.
(22) Harvey, D. J. J. Chromatogr. A 1996, 720, 429-446.
(23) Reinhold, V. N.; Reinhold, B. B.; Chan, S. Methods. Enzymol. 1996, 271,
377-402.
(24) Burlingame, A. L. Curr. Opin. Biotechnol. 1996, 7, 4-10.
(25) Rudd, P. M.; Dwek, R. A. Curr. Opin. Biotechnol. 1997, 8, 488-497.
(26) Dell, A.; Morris, H. R. Science 2001, 291, 2351-2356.
(27) Boss, H. J.; Watson, D. B.; Rush, R. S. Electrophoresis 1998, 19, 26542664.
(28) Yang, L.; Tang, Q.; Harrata, A. K.; Lee, C. S. Anal. Biochem. 1996, 243,
140-149.
5876
Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
trifluoroacetic acid (TFA) was purchased from Merck (Schuchardt, Germany); Staphylococcus aureus protease (Glu-C) was
obtained from Pierce (Rockford, IL); PNGase F was purchased
from Roche (Mannheim, Germany); TPCK-trypsin was made in
our laboratory; and all water was distilled and deionized (18 Ω)
using a Milli-Q system from Millipore (Bedford, MA). TSL was
purified according to the method described by Liang et al.29
Protein Reduction and Alkylation. A 0.2-mg portion of
RNase B was dissolved in 80 µL of 30 mM NH4HCO3 buffer (pH
7.8), and 10 µL of 50 mM DTT solution was added. The reaction
mixture was incubated for 15 min at 50 °C and cooled to room
temperature. A 10-µL portion of 100 mM IAA solution was added,
and the reaction was continued for another 15 min at 37 °C in the
dark. In the case of TSL, the protein concentration was 1 mg/
mL, and 0.5 M DTT was added to a final concentration of 50 mM.
The mixture was reacted for 1 h at 50 °C. IAA (1 M) was added
at room temperature to a final concentration of 100 mM. The
reaction was continued for another 30 min at 37 °C in the dark.
Purification of S-Carboxyamidomethylated Glycoprotein.
Alkylated proteins were purified by a HP 1100 reversed-phase
HPLC (Palo Alto, CA) using a C8 column (RP-300, 5 µm, 30 mm
× 2.1 mm i.d., Applied Biosystems, Foster City, CA). The
conditions were as follows: A solution, 0.01% TFA in water; B
solution, 0.01% TFA in ACN; flow rate, 200 µL/min; gradient, 0∼2
min, 0% B solution; 2∼15 min, 0∼100% B solution.
Protease Digestion. A 100-µg poriton of alkylated protein was
dissolved in 100 µL of NH4HCO3 buffer (pH 7.8 for tryptic or Lys-C
digestion) or in 100 µL of ammonium acetate buffer (pH 4.0 for
Glu-C digestion). A 2.5-µL portion of protease solution (1 µg/µL)
was added, and the reaction mixture was incubated at 37 °C for
4 h, then another 2.5 µL of enzyme solution was added for
overnight digestion at 37 °C. The digests were kept at -20 °C.
Glycosidase Digestion. In PNGase F digestion, 10 µL of a 1
U/µL PNGase F solution was mixed with 90 µL of protease
digests, and the mixture was incubated at 37 °C for 4 h. In
R-mannosidase digestion, 10 µL of a 0.1 U/µL enzyme solution
(containing 3.0 M (NH4)2SO4, 0.1 M ZnSO4, pH 7.5) were mixed
with 90 µL of Glu-C digests, and the mixture was incubated at 37
°C for 6 h. The digests were kept at -20 °C.
CE-MS. All CE experiments were performed using an eCAP
uncoated fused-silica capillary (80 cm × 50 µm I. D., 360 µm O.
D.) with an eCAP user-assembled capillary cartridge (Beckman,
Fullerton, CA). Dynamic cationic coating was performed basically
according to the methods described by Wiktorowicz et al.7 and
Kelly et al.8 The capillary was reconditioned by flushing under
pressure (138 kPa) with 1 M NaOH (15 min), followed by rinsing
with water (5 min), 0.1 M HCl (5 min), and water again (5 min).
Then a coating solution, 5% (w/v) Polybrene and 2% (v/v) ethylene
glycol, was infused into the capillary for 5 min. Finally, the capillary
was rinsed with CE buffer (2 M formic acid) for 5 min. The capillary surface was recoated after each run using the same method.
Beckman P/ACE system 5500 was coupled on-line to a
Finnigan MAT LCQ electrospray ion trap mass spectrometer (San
Jose, CA) via a sheath-liquid interface supplied by Finnigan. The
LCQ was tuned by angiotensin (Sigma) and calibrated by infusing
an aqueous solution of Ultramark (Finnigan) at a rate of 2 µL/
min. The sheath liquid made up of 60/39/1 (v/v/v) MeOH/water/
(29) Liang, X. L.; Wang, K. Y. Acta Biochem. Biophys. Sin. 1993, 25, 515-521.
Figure 1. Summed MS spectra (A) and observed masses (B) of RNase B by flow injection. Hexn means the sugar chain attached to the
peptide backbone has a structure of HexNAc2Hexn.
formic acid was delivered to the probe tip at a rate of 2 µL/min
by a syringe pump on LCQ. No sheath gas or auxiliary gas was
used. The polyimine coating was burned off 2-3 mm from the
outlet end and was washed off with ethanol to ensure the electric
contact. The other end of the capillary (injection end) was
immersed in a reservoir containing 2 M formic acid buffer. To
minimize the influence of gravity flow, the whole CE apparatus
was placed on a height-adjustable platform, and care was taken
to ensure that the CE buffer reservoir and the electrospray end
of the capillary were at the same level. An ESI voltage of +4.25
kV was applied to the electrospray tip, and the net separation
voltage applied was -15 kV. The LCQ was operated with
automatic gain control (AGC) for all of the experiments unless
stated otherwise. In full scan mode, the mass analyzer was
scanned from 600 to 2000 u. The number of microscans was 3 for
all modes: full MS, zoom scan, and MS/MS. According to these
settings, the most intense ion in the spectrum was selected
automatically to perform a zoom scan, followed by a collisioninduced dissociation (CID) scan with an isolation width of 3 u
after scanning each range in full scan mode. The relative collision
energy was set to 45. Other operating parameters were set as
follows: heated capillary temperature, 180 °C; capillary voltage,
24 V; tube lens offset, -5 V.
To avoid bubble formation that may arise from the injection
of bicarbonate buffer in the capillary filled with 2 M formic acid,
samples were freeze-dried and redissolved in 20 µL 2 M formic
acid buffer. After centrifugation at 6000 rpm for 10 min, the
samples were injected under pressure (138 kPa) for 30 s.
RESULTS AND DISCUSSION
Comparative Mass Mapping of RNase B. The amino acid
sequence of RNase B is the same as that of RNase A. The
difference between these two proteins is that the Asn34 residue
in RNase B is glycosylated. Flow injection of an RNase B sample
(2 mg/mL) directly from the CE instrument under pressure of
138 kPa clearly revealed the existence of minor RNase A (Mr
13681.0) in commercial RNase B (Figure 1). The glycoprotein
RNase B shows obvious microheterogeneity, the observed molecular weights of its five glycoforms are 14898.0, 15061.0, 15223.0,
15386.0, and 15549.0. Considering the mass shifts between each
two adjacent glycoforms (162 Da) and those between glycoforms
and RNase A (from 1217 to 1868 Da), the structure of oligosaccharides attached to the peptide was assigned as HexNAc2Hex5∼9.
There is only one tripeptide sequon for N-glycosylation in the
amino acid sequence of RNase B: Asn34-Leu-Thr. The N-glycosylation sites in glycoprotein can be discovered by comparative
peptide mass mapping using Lys-C digestion. Typical CE-MS
profiles before and after PNGase F treatment shown in Figure 2
are very similar to each other. Table 1 summarizes the interpreted
results. Almost all of the expected peptides were found in the mass
Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
5877
Table 1. Peptide Mapping by Lys-C Digestion of RNase B by CE-MSa
peptide
residue
sequence
tr (min)
calc massd
obs massd
obs ions
L1
L2
L3e
L3′
L4
L5
1-7
8-31
32-37
32-37
42-61
62-91
92-98
99-104
105-124
KETAAAK
FERQHMDSSTSAASSSNYCbNQMMK
SRNcLTK
SRDLTK
PVNTFVHESLADVQAVCSQK
NGQTNCYQSYSTMSITDCRETGSSK
YPNCAYK
TTQANK
HIIVACEGNPYVPVHFDASV
29.94
10.88
9.89
30.21
8.77
4.66
717.8
2798.1
1934.7
718.8
2230.4
2876.1
916.0
661.7
2225.5
717.8
2797.5
1934.6
718.5
2229.6
2876.7
718.8+
933.53+
968.32+
719.5+
744.23+
959.93+
661.4
2223.8
662.4+
1112.92+
L6
L7
a
16.00
8.49
Peptide mass above 600 Da. b All cysteines were alkylated. c Glycosylation site. d Average mass. e Dominant sugar chain is HexNAc2Hex5.
Table 2. Peptide Mapping by Glu-C Digestion of RNase B by CE-MSa
f
peptide
residue
sequence
tr
(min)
calc massd
obs massd
obs ions
G1
G2e
G2′f
G3
G4
G5
1-9
10-49
10-49
50-86
87-111
112-124
KETAAAKFE
RQHMDSSTSAASSSNYCbNQMMKSRNcLTKDRCKPVNTFVHE
RQHMDSSTSAASSSNYCNQMMKSRNcLTKDRCKPVNTFVHE
SLADVQAVSQKNVACKNGQTNCYQSYSTMSITDCRE
TGSSKYPNCAYKTTQANKHIIVACE
GNPYVPVHFDASV
17.81
14.40
14.73
5.14
13.25
5.42
994.1
5922.1
5597.9
4275.7
2842.2
1401.5
993.6
5923.5
5598.5
4274.8
2842.2
1401.4
994.6+
1185.75+
1120.75+
1069.74+
948.43+
701.7+
a Peptide mass above 600 Da. b All cysteines were modified by IAA. c Glycosylation site. d Average mass. e Dominant sugar chain is HexNAc Hex .
2
5
Dominant sugar chain is HexNAc2Hex3.
Figure 3. Base peak and extracted ion chromatogram for glycopeptides resulting from Lys-C digestion of RNase B. Relative
abundance of doubly charged ions was listed for comparison.
Figure 2. Comparative PMM by Lys-C digestion of RNase B before
(A) and after (B) PNGase F treatment. Separation was carried out
by applying a net voltage of -15 kV using charge reversal.
spectra. Glycopeptide L3 was eluted at 9.9 min (Figure 2a). After
PNGase F treatment, L3 disappeared and another peak corre5878 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
sponding to its deglycosylated product appeared at 30.2 min
(Figure 2b). The separation of glycopeptides and their relative
abundance is shown by the extracted ion chromatogram (EIC):
HexNAc2Hex5/HexNAc2Hex6/HexNAc2Hex7/HexNAc2Hex8 is 6.3
× 107/3.6 × 107/1.0 × 107/8.0 × 106 (Figure 3). The relatively
good resolution of glycoforms allowed the mass spectrometer to
have enough time to perform zoom scan and MS/MS analysis
for each of the glycopeptides. Comparing the MS/MS spectra of
two adjacent glycopeptides clearly revealed that their structures
differed only in one mannose residue attached to the sugar chain
(Figure 4). Under the current MS conditions, the breakage mainly
occurred at the covalent bonds between sugar residues. The
number of sugar residues and their linkage could be deduced
from the ion pattern and the differences of CID spectra. After
Figure 4. MS/MS spectra of glycopeptides L3-HexNAc2Hex5 (A) and L3-HexNAc2Hex6 (B).
Figure 5. MS/MS spectra of glycopeptide G2′-HexNAc2Hex3.
PNGase F treatment, the glycopeptides were converted to peptide
SRD34LTK, which was also confirmed by its CID spectrum.
In high-mannose-type N-glycosylation, the mannose residues
were linked by R-glycosidic bonds, which makes it different from
other types of N-glycosylation. So if the mannose residues
appended to the pentasaccharide core can be removed by
R-mannosidase or if the structure of oligosaccharide left is
GlcNAc2Man1, the glycosylation can be assigned as high-mannose
type. Accordingly, RNase B was digested by Glu-C protease and
then by R-mannosidase. The results of comparative PMM were
summarized in Table 2. The sugar chains in glycopeptide G2 were
all converted to GlcNAc2Man3 after R-mannosidase treatment for
6 h (Figure 5). A longer period of R-mannosidase treatment (>48
h) could give rise to a structure of GlcNAc2Man1 (data not shown).
These results showed that the N-glycosylation in RNase B belongs
to the high-mannose type.
Comparative Mass Mapping of TSL. TSL is a C-type lectin
from the venom of Trimeresurus stejnegeri with galactose binding
Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
5879
Table 3. Peptide Mapping by Trypsin Digestion of TSL by CE-MSa
peptide
T1f
T1-O
T1-2O
T1′
T1′-O
T1′-2O
T2
T3
T3-O
T4
T5
T6
T7
T8
T9
residue
sequence
tr (min)
calc massd
obs massd
obs ions
1-16
1-16
1-16
1-16
1-16
1-16
17-22
23-32
23-32
34-45
46-61
63-73
76-84
85-94
95-101
104-117
118-126
127-134
SCbCTNcDSLPMNGMCYK
1.31
1.31
1.15
4.48
4.33
4.33
12.80
7.30
7.00
3641.4
3657.4
3673.4
1939.2
1955.2
1971.2
747.9
1344.5
1360.5
1472.7
1867.1
1299.5
1241.3
1301.4
902.0
1709.9
1165.3
1056.2
3641.1
3659.7
3675.6
1939.2
1955.4
1971.2
747.4
1344.2
1360.2
1214.73+
1220.93+
1226.23+
970.62+
978.72+
986.62+
748.4+
673.12+
681.12+
1867.2
1299.4
1241.2
1301.2
623.43+
650.72+
621.62+
651.62+
1709.4
855.72+
SCCTNcDSLPMNGMdCYK
SCCTNcDSLPMdNGMdCYK
SCCTDDSLPMNGMCYK
SCCTDDSLPMNGMdCYK
SCCTDDSLPMdNGMdCYK
IFDEPK
TWEDAEMFCR
TWEDAEMdFCR
YKPGCHLASFHR
LAESLDIAEYISDYHK
QAEVWIGLLDRK
DFSWEWTDR
SCTDYLNWDK
NQPDHYK
EFCVELVSLTGYHR
WNDQVCESK
NSFLCQCK
9.70
9.30
7.99
7.99
10.84
10.29
1055.4
1056.4+
a Peptide mass above 600 Da. b All cysteines were modified by IAA. c Glycosylation site. d Methionine was oxidated. e Average mass. f Dominant
sugar chain is HexNAc2Hex8.
activity. It has 135 amino acid residues, including 10 cysteine
residues, and displayed monomer, dimer, and trimer on nonreducing SDS-PAGE (data not shown), indicating that some
cysteines were involved in pairing of an interchain linkage. The
analysis of the amino acid sequence of TSL revealed that there is
also only one glycosylation site matching the tripeptide recognition
signal: Asn5-Asp-Ser.
Comparing the base peak (representing the most abundant
ion in a peak) of the separation of tryptic digest of TSL before
and after PNGase F treatment, it was found that the deglycosylated
peptide was eluted about 3 min later than the glycopeptides
(Figure 6), indicating that the attachment of the sugar chain had
a great influence on the electrophoretic mobility of the peptides.
Results in Table 3 show that most of the tryptic peptides were
detected by CE-MS. Interestingly, we also found peaks representing the differential oxidation of glycopeptide T1 (S1CCTNDSLPMNGMCYK16) and the Met oxidation of peptide T3
(T23WEDAEMFCR32). Figure 7 shows the summed MS spectrum
of glycopeptide T1. Although the peaks were not separated very
well because of the presence of differentially oxidized products,
it still could be seen clearly that T1 had four major glycoforms.
According to the mass shifts between adjacent peaks (162 Da),
and those between the glycoforms and peptide (1214 to 1702 Da),
the structure of sugar chains was assigned as HexNAc2Hex5∼8.
The corresponding mono-oxidized and double-oxidized products
of the most abundant glycopeptide (denominated as T1-HexNAc2Hex8-O and T1-HexNAc2Hex8-2O, respectively) were characterized
by MS/MS (Figure 8). Fragment ions from T1-HexNAc2Hex1 to
T1-HexNAc2Hex7 could be found in the CID spectra, indicating
that the breakage mainly occurred mainly between the mannose
residues. Closer inspection of these two CID spectra revealed that
the doubly charged fragment ions from T1-HexNAc2Hex8-2O were
(30) Liu, T.; Zeng, R.; Shao, X. X.; Xia, Q. C. Acta Biochem. Biophys. Sin. 1999,
31, 425-432.
(31) Liu, T.; Shao, X. X.; Zeng, R.; Xia, Q. C. J. Chromatogr. A 1999, 855, 695707.
(32) Hsu, Y. R.; Narhi, L. O.; Spahr, C.; Langley, K. E.; Lu, H. S. Protein Sci.
1996, 5, 1165-1173.
5880 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
Figure 6. Comparative PMM by trypsin digestion of TSL before
(A) and after (B) PNGase F treatment. Separation was carried out
by applying a net voltage of -15 kV using charge reversal.
8 Da larger than those from T1-HexNAc2Hex8-O, thus confirming
the differential Met oxidation in glycopeptides.
After PNGase F treatment, glycopeptides T1 were converted
to peptide T1′. The influence on the electrophoretic mobility of
peptide caused by differential oxidation was much smaller than
that by glycosylation. T1′ and its differential oxidation products
(T1′-O and T1′-2O) eluted at almost the same time (around 4.5
min), and T1′-2O is the most abundant component (Figure 9). It
is noteworthy that, although T1′, T1′-O, and T1′-2O were not well-
Figure 7. Summed MS spectra from the labeled peak T1 in the base peak in Figure 6A. Hexn means the sugar chain attached to the peptide
backbone has a structure of HexNAc2Hexn.
Figure 8. MS/MS spectra of the mono- and doubly oxidized products of glycopeptide T1-HexNAc2Hex8. (A) T1-HexNAc2Hex8-O; (B) T1HexNAc2Hex8-2O.
separated, the fast scan speed of MS still allowed their CID spectra
to be collected separately. It can be seen clearly from the CID
spectrum of T1′-O (SCCTDDSLPM10NGM13CYK) that the preferred oxidation site was Met13 instead of Met10 (Figure 10). The
oxidation that took place on peptide T3 was also confirmed by its
CID spectrum.
Comparative PMM was also performed by using Glu-C protease and R-mannosidase successively to further characterize the
type of N-glycosylation in TSL (Figure 11). The results are
summarized in Table 4. Two glycopeptides (G1 and G2) containing
the same glycosylation site were found by Glu-C protease cleavage.
Differential oxidation in these glycopeptides was also observed
Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
5881
Figure 9. Summed MS spectra from the labeled peak T1′ in the base peak in Figure 6B shows the doubly charged ions for the deglycosylated
peptide and its differential oxidation products.
Figure 10. MS/MS spectrum of the mono-oxidized and deglycosylated peptide. M indicates the oxidized Met residue.
in PMM using Glu-C protease, indicating strongly that it is a kind
of process-induced modification. Comparing the separation profiles
of digests before and after R-mannosidase treatment, the treated
glycopeptides were all eluted later because of the removal of
mannose residues (G1′, 8.4 min vs G1, 6.4 min; G2′, 6.5 min vs
G2, 4.7 min), while the other peptides remained at almost the
same positions. Although the separation of glycopeptides was not
very good because of the presence of differential oxidation
products with similar electrophoretic mobilities, the CID spectra
of some major components could still be collected separately, and
the sugar chain structure and Met oxidation in these glycopeptides
could be characterized. In glycopeptides G1 and G2, the dominant
sugar chain has a structure of HexNAc2Hex8 (observed ions are
1597.03+ and 1339.73+). For example, G1-HexNAc2Hex8 and its
5882 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
differential oxidation products (G1-HexNAc2Hex8-O, G1-HexNAc2Hex8-2O) were all identified by MS/MS (Figure 12). Under the
specific CID condition, breakage mainly occurred at the glycosidic
bonds instead of at the peptide bonds. Fragment ions from G1HexNAc2Hex7 to G1-HexNAc1 could be found in these spectra,
from which the sequence of the sugar chain could be deduced.
The differential oxidation was also confirmed by comparing these
CID spectra. It could be seen from Figure 13 that the sugar chain
structure was converted to HexNAc2Hex1∼3 (observed ions are
1218.23+, 1273.23+, 1326.03+, and their differential oxidation products) after R-mannosidase treatment (Figure 13b shows the
identification of the double-oxidized product G1-HexNAc2Hex32O), indicating that all hexoses appended to the pentasaccharide
core could be removed by R-mannosidase. So the glycosylation
Table 4. Peptide Mapping by Glu-C Digestion of TSL by CE-MSa
peptide
G1f
G1-O
G1-2O
G1′g
G1′-O
G1′-2O
G2f
G2-O
G2-2O
G2′g
G3
G4
G4-O
G5
G6
G7
G8
residue
sequence
tr (min)
calc massd
obs massd
obs ions
1-25
1-25
1-25
1-25
1-25
1-25
1-19
1-19
1-19
1-19
20-25
29-48
29-48
49-54
55-65
66-80
81-104
109-124
125-135
SCbCTNcDSLPMNGMCYKIFDEPKTWE
6.41
6.41
6.27
8.41
8.41
8.28
4.69
4.69
4.69
6.50
15.87
39.55
38.82
7.18
26.67
4787.7
4803.7
4819.7
3977.2
3993.2
4009.2
4016.7
4032.7
4048.7
3206.2
788.9
2509.0
2525.0
646.7
1409.5
1892.2
3115.3
1977.2
1418.7
4788.0
4802.4
4820.4
3975.0
3993.9
4008.3
4016.1
4034.4
4049.7
3206.4
788.5
2508.8
2524.8
646.3
1049.6
1597.03+
1601.83+
1607.83+
1326.03+
1332.33+
1337.13+
1339.73+
1345.83+
1350.93+
1604.22+
789.5+
628.24+
632.24+
647.3+
705.82+
1976.6
1418.4
989.32+
710.22+
SCCTNcDSLPMNGMdCYKIFDEPKTWE
SCCTNcDSLPMdNGMdCYKIFDEPKTWE
SCCTNcDSLPMNGMCYKIFDEPKTWE
SCCTNcDSLPMNGMdCYKIFDEPKTWE
SCCTNcDSLPMdNGMdCYKIFDEPKTWE
SCCTNcDSLPMNGMCYKIFD
SCCTNcDSLPMNGMdCYKIFD
SCCTNcDSLPMdNGMdCYKIFD
SCCTNcDSLPMNGMCYKIFD
EPKTWE
MFCRKYKPGCHLASFHRLAE
MdFCRKYKPGCHLASFHRLAE
SLDIAE
YISDYHKRQAE
VWIGLLDRKKDFSWE
WTDRSCTDYLNWDKNQPDHYKDKE
LVSLTGYHRWNDQVCE
SKNSFLCQCKF
12.48
18.65
a Peptide mass above 600 Da. b All cysteines were modified by IAA. c Glycosylation site. d Methionine was oxidated. e Average mass. f Dominant
sugar chain is HexNAc2Hex8. g Dominant sugar chain is HexNAc2Hex3.
Figure 11. Comparative PMM by Glu-C digestion of TSL before
(A) and after (B) R-mannosidase treatment. Separation was carried
out by applying a net voltage of -15 kV using charge reversal.
in TSL can be assigned as high-mannose-type N-glycosylation, and
the molecular formula of the sugar chains is GlcNAc2Man5∼8.
CONCLUSIONS
In this paper, a strategy of glycoprotein digestion by protease
combined with glycosidase to characterize the high-mannose-type
N-glycosylation by CE-ESI-MS is described. PNGase F can
remove all N-linked sugar chains from the peptide. Its optimal
pH range is 7∼8, so it can be used directly after trypsin or Lys-C
digestion to determine the N-glycosylation sites. R-Mannosidase
with an optimal pH range of 4∼5 can be used right after Glu-C
digestion to determine the linkage of the pentasaccharide core
with other sugar residues. Using this strategy to perform
comparative peptide mass mapping by CE-MS/MS, the highmannose-type N-glycosylation in RNase B and TSL were characterized rapidly and conveniently. The use of charge reversal can
improve the resolution of glycoforms when it is combined with
CE-MS/MS, and glycoforms can be separated and characterized
without any extra oligosaccharide release, derivatization, labeling,
or purification; thus, the whole analytical procedure is straightforward with minimal sample loss and contamination.
Sensitivity at the low picomole level can be achieved in this
study, although only a conventional sheath-flow interface with a
much higher flow rate, as compared to the microspray interface
and nanospray interface, was used. In our previous study, protein
identification and modification characterization could be accomplished at a low picomole level using this analytical system
and automatic database searching.30,31 A minimal sample concentration of 5 × 10-6 M still needs to be used, because no other
sample preconcentration techniques were employed. In comparison to traditional LC-MS, CE-MS has better sensitivity and a
similar limit of detection for sample concentration. In addition,
its detection limits can be further enhanced by using low flowrate interfaces and sample preconcentration techniques (e.g., solidphase microextration). Up to now, quantification is relatively hard
to carry out on this analytical system, because the electrospray
in CE-MS is hard to keep consistent using a coaxial sheath-flow
interface with a much lower flow rate than that in LC-MS, yet
the determination of the relative abundance of individual glycoforms still can be achieved by this technique (e.g., the relative
abundance of tryptic glycopeptides of RNase B: HexNAc2Hex5/
HexNAc2Hex6/HexNAc2Hex7/HexNAc2Hex8 is 53.8/30.8/8.5/
6.8%). Its ability of quantification may well be improved by using
more stable sheathless interfaces and internal calibrants. Relatively
Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
5883
Figure 12. MS/MS spectra of glycopeptide G1 and its differential oxidation products. (A) G1-HexNAc2Hex8, (B) G1-HexNAc2Hex8-O, and (C)
G1-HexNAc2Hex8-2O.
good reproducibility could be obtained when fresh coating was
used for each run. A standard deviation of retention time of ∼0.10.3 min was obtained routinely. The value could be enlarged to
∼0.2-0.4 min in comparative PMM, because the sample properties were changed slightly after glycosidase treatment (See
Figures 2, 6, and 11).
Met oxidation could happen when proteins were processed at
room temperature for a long time. It is very common in in-gel
protein digestion, although the mechanism of methionine oxidation is still not clear, but the possibility that it might happen in
the junction between CE and MS or in the ion source of MS can
5884 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
be eliminated. The PMMs of RNase B and cytochrome c were
carried out under the same conditions as TSL, but no Met
oxidation was found. It seems like the oxidation of methionine
residues may be related to the nature of Met-containing peptides
(e.g., its amino acid sequence and its local position in the
macromolecule). In vitro Met-oxidation experiments revealed that
the initial oxidation rates for the methionine residues of recombinant human stem cell factor were quite different.32
Analysis of process-induced protein modification can be a
challenge when it is coexisting with other kinds of posttranslational modifications. In this study, the differential Met oxidation
Figure 13. (A) Summed MS spectra from the labeled peak G1′ in the base peak in Figure 11B shows the triply charged ions for the
R-mannosidase-treated glycopeptide and their differential oxidation products. (B) MS/MS spectrum of a double-oxidized glycopeptide G1-HexNAc2Hex3-2O.
that occurred in the glycopeptides of TSL had added extra
microheterogeneity to the glycopeptides. Although the highly
heterogeneous peptides were not well-separated, the fast scan
speed and strong MS/MS capacity of the ion trap mass spectrometer allowed most of these heterogeneous peptides to be
identified. In particular, the preferred oxidation site in the doubleoxidized peptide was identified by interpreting the high-quality
MS/MS spectrum. Carefully controlling the relative CID energy
can make the breakage occur mainly at the glycosidic bonds
instead of at the peptide bonds, thus greatly facilitating the
interpretation of the sequence of oligosaccharides.
ACKNOWLEDGMENT
This work was supported by a grant from the “863” Biotechnology Project of China, no. 102-08-06-02, and by a grant for a
key project from the National Natural Science Foundation of
China, no. 39990600.
Received for review June 19, 2001. Accepted September
18, 2001.
AC0106748
Analytical Chemistry, Vol. 73, No. 24, December 15, 2001
5885