Capillary electrophoresis time-of-flight mass spectrometry for a confident elucidation of a glycopeptide map of recombinant human erythropoietin

Article abstract of DOI:10.1002/rcm.5114

Capillary electrophoresis coupled to orthogonal accelerated time-of-flight mass spectrometry (CE/TOFMS) was used for the analysis of O- and N-glycopeptides of recombinant human erythropoietin (rhEPO). O₁₂₆ and N₈₃ with a tetraantennary complex type glycan (N₈₃-4Ant) were selected as glycopeptide models to develop an optimum CE/TOFMS methodology capable of detecting and characterizing the wide variety of glycopeptides present in the glycoprotein digest. Glycopeptide adsorption in the inner surface of the fused-silica capillary was prevented after using a capillary conditioning of 1 M HAc between runs. On the other hand, different acidic conditions in the sheath liquid (SL) and in the background electrolyte (BGE) were tested with the aim of studying their influence in glycopeptide fragmentation. Finally, the fragmentor voltage value of the TOF-MS instrument was optimized to avoid the involuntary fragmentation of the native glycopeptides. Hence, the established method may be regarded as an excellent starting point to obtain reliable glycopeptide maps of complex glycoproteins such as rhEPO by CE/TOFMS. Copyright

Full text of DOI:10.1002/rcm.5114

Research Article
Received: 2 March 2011
Revised: 15 April 2011
Accepted: 22 May 2011
Published online in Wiley Online Library
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316
(wileyonlinelibrary.com) DOI: 10.1002/rcm.5114
Capillary electrophoresis time‐of‐flight mass spectrometry for a
confident elucidation of a glycopeptide map of recombinant
human erythropoietin
*
Estela Giménez , Raquel Ramos‐Hernan, Fernando Benavente, José Barbosa and
Victoria Sanz‐Nebot
Department of Analytical Chemistry, University of Barcelona, Diagonal 647, 08028 Barcelona, Spain
Capillary electrophoresis coupled to orthogonal accelerated time‐of‐flight mass spectrometry (CE/TOFMS) was used
for the analysis of O‐ and N‐glycopeptides of recombinant human erythropoietin (rhEPO). O126 and N83 with a
tetraantennary complex type glycan (N83‐4Ant) were selected as glycopeptide models to develop an optimum CE/
TOFMS methodology capable of detecting and characterizing the wide variety of glycopeptides present in the glyco-
protein digest. Glycopeptide adsorption in the inner surface of the fused‐silica capillary was prevented after using a
capillary conditioning of 1 M HAc between runs. On the other hand, different acidic conditions in the sheath liquid
(SL) and in the background electrolyte (BGE) were tested with the aim of studying their influence in glycopeptide
fragmentation. Finally, the fragmentor voltage value of the TOF‐MS instrument was optimized to avoid the involun-
tary fragmentation of the native glycopeptides. Hence, the established method may be regarded as an excellent start-
ing point to obtain reliable glycopeptide maps of complex glycoproteins such as rhEPO by CE/TOFMS. Copyright ©
2011 John Wiley & Sons, Ltd.
Glycosylation is one of the most common post‐translational
modifications in proteins. It has been estimated that at least
50% of the proteins in mammalian proteomes are glycosy-
lated. Glycoproteins consist of mixtures of glycosylated
variants, known as glycoforms, which share the same peptide
sequence but show different oligosaccharides at their char-
acteristic O‐ or N‐glycosylation sites. Glycoproteins play a
crucial role in recognition, signaling and adhesion processes
on the cell surfaces. Many of these functions are mediated
by the oligosaccharide chains. Moreover, O‐ and N‐glycan
chain structures of glycoproteins may be altered in many dis-
eases such as congenital disorders of glycosylation, alcohol-
ism or cancer.[1] Hence, detection and characterization of the
different oligosaccharides attached to a given glycoprotein
invokes great interest in many research fields such as bio-
technology, biochemistry, medicine and pharmacology. The
analysis of a glycoprotein carbohydrate microheterogeneity
is usually carried out at the glycan or glycopeptide level after
chemical or enzymatic digestion. In contrast to glycans, gly-
copeptides not only provide information about the structure
and composition of the oligossacharides, but also allow infor-
mation to be obtained about glycosylation sites and their
degree of occupation.
In recent years, capillary electrophoresis coupled to mass
spectrometry (CE/MS) has been used in glycoproteomics
for the analysis of intact glycoproteins, as well as for glyco-
peptides and glycans from protein digests.[2–10] CE offers
the possibility of separating these biomolecules according to
their mass‐to‐charge ratios (m/z) while MS detection enables
elucidation of their molecular masses and structures. Ortho-
gonal acceleration time‐of‐flight (oaTOF) mass spectrometers
are excellent MS analyzers for the study of glycopeptides,
especially those of high molecular mass, as they provide accu-
rate and high‐resolution mass measurements over wide
mass‐to‐charge acquisition ranges permitting the detection
of the multiply charged molecular ions expected for this kind
of complex analyte. However, the analysis of glycopeptides
by CE/MS is a challenging task. On the one hand, they are
difficult to ionize in positive ESI mode, especially if they con-
tain glycans with negatively charged sialic acid residues, or
ion suppression occurs because the separation from the other
peptides and glycopeptides in the digest is not complete.
On the other hand, glycopeptides may undergo fragmenta-
tion of labile groups such as sialic acids (SiA), hexoses (Hex),
deoxysugars such as fucose or hexose‐N‐acetylhexose units
(HexHexNAc) depending on the MS detection conditions.
This involuntary fragmentation is in general overlooked and
may mislead glycoproteomics researchers, leading to wrong
conclusions about glycopeptide composition of a certain
glycoprotein. For these reasons, new CE/MS methodology
needs to be established in order to improve glycopeptide ana-
lysis in terms of separation and MS detection. The goal may
be separation and detection of the maximum number of
* Correspondence to: E. Giménez, Department of Analytical
Chemistry, University of Barcelona, Diagonal 647, 08028
Barcelona, Spain.
E‐mail: estelagimenez@ub.edu
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316
Copyright © 2011 John Wiley & Sons, Ltd.

E. Giménez et al.
glycopeptides of a complex protein digest, while preventing
their fragmentation to obtain a reliable map for a confident
glycoprotein characterization.
4.5 mg NaCl, and 3.5 mg of Na2HPO4. The content of each
vial was dissolved in purified water to obtain a 1000 mg·L–1
solution of rhEPO. Excipients of low molecular mass were
removed from the rhEPO sample by passage through a
Microcon YM‐10 centrifugal filter from Millipore (Mr cut‐off
10 kDa, Bedford, MA, USA).[10] The filter was initially washed
with purified water for 10 min in a centrifuge at 13 000 g, and
the sample was centrifuged after that for 10 min under the
same centrifugal force. The residue was washed three times
for 10 min in the same way with an appropriate volume of
purified water. The final residue was recovered from the
upper reservoir by upside‐down centrifugation in a new vial
(3 min at 1000 g) and sufficient purified water was added to
adjust the rhEPO concentration to 1000 mg·L–1.
rhEPO was reduced, alkylated and immediately subjected
to enzymatic digestion. Briefly, 2.5 μL of 0.5 M DTT in
50 mM NH4HCO3 (pH 7.9) were added to an aliquot of
100 μL of the desalted 1000 mg·L–1 rhEPO solution. The mix-
ture was incubated in a water bath at 56 °C for 30 min and
then alkylated with 50 mM IAA for 30 min at room tempera-
ture in the dark (7 μL of 0.73 M IAA). Excess of low molecular
weight reagents was removed with Microcon YM‐10 centrifu-
gal filters as explained before. In this case, the final residue
was reconstituted in 100 μL of 50 mM NH4HCO3 (pH 7.9).
For trypsin digestion, an enzyme‐to‐protein ratio of 1:40 by
mass was added, and the mixture was carefully vortexed
and incubated at 37 °C in a water bath for 18 h.[10] Digestion
was stopped by heating for 5 min in boiling water and stored
at −20 °C until its use. rhEPO digest was analyzed by CE/
TOFMS injecting ~16 ng of digest for a single analysis. Diges-
tion reproducibility was evaluated preparing two rhEPO
digests from two different EPO BRP batches (lot2 and lot3).
No differences could be observed between digests and lots
by CE/TOFMS demonstrating the reproducibility of the
enzymatic digestion procedure.
In this work, with the aim of developing an optimal meth-
odology for the analysis of glycopeptides in enzymatic
protein digests by CE/TOFMS, recombinant human erythro-
poietin (rhEPO) has been selected as a model. rhEPO is a
biologically and therapeutically relevant glycoprotein.
Approximately 40% of its molecular mass (29 888 Da,
the most abundant glycoform[2]) is composed of carbohy-
drates, of which 17% correspond to sialic acids, mainly
N‐acetylneuraminic acid (Neu5Ac). The wide variety of
glycans are attached to three N‐glycosylation sites at Asn
24, 38 and 83 and one O‐glycosylation site at Ser 126.[11] The
complex peptide and glycopeptide mixture obtained after
enzymatic digestion is similar to those that can be expected
for other complex glycoproteins. Moreover, the analysis of
rhEPO glycopeptides gives rise to great interest in doping
analysis and quality control to find, in the future, discrimina-
tion points between the endogenous and the recombinant
erythropoietins. The study is focused on the O126 glycopep-
tide and also N83 with a tetraantennary complex type glycan
(N83‐4Ant) glycopeptide. Both glycopeptides show a great
number of structural differences and may be useful to
develop a general method for detection and characterization
of the wide variety of glycopeptides present in a glycoprotein
digest.
EXPERIMENTAL
Chemicals
All chemicals used in the preparation of buffers and solu-
tions were of analytical reagent grade. Isopropanol (iPrOH),
acetic acid (HAc, glacial), formic acid (HFor, 98–100%),
ammonia (25%), ammonium acetate (NH4Ac) and sodium
hydroxide were supplied by Merck (Darmstadt, Germany).
DL‐Dithiothreitol (DTT, ≥99%), iodoacetamide (IAA) and
ammonium hydrogen carbonate (NH4HCO3) were supplied
by Sigma‐Aldrich (Madrid, Spain). Trypsin (sequencing
grade modified) from Promega (Madison, WI, USA) was
obtained from Roche (Mannheim, Germany). Water with a
conductivity lower than 0.05 mS/cm was obtained using a
Milli‐Q water purification system from Millipore (Molsheim,
France).
ESI Low Concentration (ESI‐L) tuning mix was supplied by
Agilent Technologies (Waldbronn, Germany) for tuning and
calibration of the oaTOF mass spectrometer. The reference com-
pound solutions for internal mass recalibration of the TOF mass
spectrometer (2.5mM HP‐0921, hexakis(1H,1H,3H‐tetrafluoro-
propoxy)phosphazine; 0.2 mM HP‐1221, hexakis(1H,1H,4H‐
hexafluorobutyloxy)phosphazine; 0.2 mM HP‐1821, hexakis
(1H,1H,6H‐decafluorohexyloxy)phosphazine; 0.5 mM HP‐
2421, hexakis(1H,1H,8H‐tetradecafluoroctyoxy)phosphazine
in acetonitrile) were also supplied by Agilent Technologies.
CE/TOFMS
The CE/TOFMS experiments were performed in the HP3DCE
system coupled to a 6220 oaTOF LC/MS mass spectrometer
with an orthogonal G1603A sheath‐flow interface (Agilent
Technologies, Waldbronn, Germany).[12] The sheath liquid
was delivered at a flow rate of 3.3 μL·min–1 by a KD Scientific
100 series infusion pump. CE control and separation data
acquisition (e.g. voltage, temperature and current) were per-
formed using ChemStation software (Agilent Technologies)
that was running in combination with the MassHunter work-
station software (Agilent Technologies) for control, data
acquisition and processing of the oaTOF mass spectrometer.
The oaTOF mass spectrometer was tuned and calibrated fol-
lowing the manufacturer’s instructions. Once or twice per
day, a ’Quick Tune’ of the instrument was carried out in posi-
tive mode followed by a mass‐axis calibration to ensure accu-
rate mass assignments. Measurement parameters were tuned
for the analysis of O126 and N83‐4Ant glycopeptides paying
special attention to the fragmentor voltage value. The opti-
mum parameters were as follow: capillary voltage 4000 V,
drying gas (N2) temperature 200 °C, drying gas flow rate 4 L
min–1, nebulizer gas (N2) 7 psig, fragmentor voltage 190 V,
skimmer voltage 60 V, OCT 1 RF Vpp voltage 300 V.
Data were collected in profile (continuum) at 1 spectrum/s
(approx. 10 000 transients/spectrum) between m/z 100 and
rhEPO samples
rhEPO produced in a Chinese hamster ovary (CHO) cell line
was provided by the European Pharmacopoeia as a Biological
Reference Product (BRP‐lot2 and lot3). Each sample vial con-
tained 250 µg of EPO (a mixture of epoetin alpha and beta),
0.1 mg of Tween 20, 30 mg of trehalose, 3 mg of arginine,
wileyonlinelibrary.com/journal/rcm
Copyright © 2011 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316

Elucidation of a glycopeptide map of rhEPO
3200 working in the highest resolution mode (4 GHz). The
’Molecular Feature Extractor’ algorithm provided with the
MassHunter workstation software was used for the identifi-
cation of peptides and O126 and N83‐4Ant glycopeptides from
rhEPO digests.
(T) and Table 1 lists their corresponding amino acid position
in the native protein sequence, their theoretical and experi-
mental mass (Mtheo and Mexp, respectively). As expected,
separation did not significantly change from that obtained
by CE/ITMS.[10] Nevertheless, owing to the higher sensitivity
of this mass spectrometer, peptides 13 and 14 were detected
by CE/TOFMS increasing the coverage of the protein. Table 1
also shows the signal‐to‐noise ratio (S/N), as well as the rela-
tive standard deviations and average errors of the Mexp (RSD
Mexp and Av. error, respectively) for each detected peptide
with and without reference compounds in the SL. More accu-
rate and reproducible Mexp values were obtained using refer-
ences in the SL since the av. error and RSD Mexp values were
lower in most of the cases. However, S/N values significantly
decreased in the presence of references causing a total loss of
signal for peptide 16. These results contrast with those we
obtained for the analysis of neuropeptides in which S/N
values were not significantly different with and without refer-
ences in the SL.[12] Hence, depending on the complexity of the
sample to be analyzed, the use of references in the SL can be
inconvenient for the detection of the analytes of interest by
CE/TOFMS.
A Polymicro bare fused‐silica capillary of 70 cm total length
(LT) × 50 µm internal diameter (i.d.) x 360 µm outer diameter
(o.d.) supplied by Composite Metals Service (Worcester, UK)
was used for CE/TOF‐MS. New capillaries were activated
by flushing them sequentially with the following solutions
for 30 min each: 1 M NaOH, water and background electro-
lyte (BGE). Each capillary was conditioned every day by rin-
sing for 5 min with NaOH, 7 min with water and for 10 min
with BGE. Activation and conditioning procedures were per-
formed off‐line in order to avoid NaOH entering the mass
spectrometer. Electrophoretic separations were carried out at
25 °C under normal polarity (18 kV). Between runs two types
of capillary conditioning were tested: A rinse of BGE for 5 min
or rinsing the capillary for 1 min with water, 3 min with 1 M
HAc, 1 min with water and 5 min with BGE. Columns were
stored overnight filled with water.
Sheath liquids (SLs) of iPrOH/H2O (50:50 v/v) with differ-
ent amounts of HFor were tested: 0.01, 0.05 and 0.1% (v/v) of
HFor. When internal mass recalibration was used, a SL con-
taining HP‐0921, HP‐1221, HP‐1821 and HP‐2421 was pre-
pared. 0.1, 1.4, 5.4 and 18 μL of each reference compound
solution, respectively, were added to 25 mL of SL solution.
All signal intensities of the reference compounds were around
5000–15 000 counts as recommended by the manufacturer.[13]
The sheath liquids were degassed for 10 min by sonication
before use. For analysis of rhEPO digests, injection was per-
formed hydrodynamically at 50 mbar for 15 s and two BGEs
were used: an acidic BGE of 50 mM HAc and 50 mM HFor
(pH 2.2) and a basic BGE of 50 mM NH4Ac (pH 8.0). Before
CE/MS, all solutions were passed through a 0.45‐mm nylon
filter (MSI, Westboro, MA, USA). All samples were kept at
4 °C and stored at −20 °C when not in use for a long period.
pH measurements were performed with a Crison 2002
potentiometer and a Crison electrode 52–03 (Crison Instru-
ments, Barcelona, Spain). Centrifugation procedures were
carried out in a Mikro 20 centrifuge (Hettich, Tuttlingen,
Germany).
Analysis of O126 and N83‐4Ant glycopeptides
In addition to peptides, enzymatic digestion of glycoproteins
results in a heterogeneous mixture of glycopeptides which, in
general, implies increased difficulties related to separation,
detection and characterization. rhEPO (T) was analyzed by
CE/TOFMS detecting different glycoforms of N83, O126 and
N24‐N38 glycopeptides (data not shown). Glycopeptides
O126 and N83 with a tetraantennary complex type glycan
(N83‐4Ant) were selected as glycopeptide models to develop
an optimum CE/TOFMS method capable of detecting and
characterizing, in the future, the wide variety of glycopep-
tides present in the different recombinant and endogenous
erythropoietins. rhEPO (T) permitted the detection of various
glycopeptide glycoforms with different number of Neu5Ac
for both O126 and N83‐4Ant glycopeptides. Figure 2 shows
the sum of EIEs of the main detected glycoforms for O126
and N83‐4Ant glycopeptides with their corresponding mass
spectra (Figs. 2(a) and 2(b), respectively). Three glycoforms
of O126 glycopeptide showing from 0 to 2 Neu5Ac were
detected (Fig. 2(a), (ii, iii, iv)), whereas N83‐4Ant showed gly-
coforms from 2 to 4 Neu5Ac (Fig. 2(b), (ii, iii, iv)). In addition
to these main glycoforms, acetylation of the Neu5Ac was also
observed in the O126 glycopeptide. O126/1NeuAc showed
non‐ and mono‐acetylated glycoforms (see Fig. 2(a), (iii))
but we were not able to confirm the di‐acetylated glycoform
described by other authors, probably because its signal was
too low to be differentiated from the background.[14–16] More-
over, O126/2Neu5Ac showed non‐ and mono‐acetylated gly-
coforms (see Fig. 2(a), (iv)) but also di‐ and tri‐acetylated
glycoforms at a higher m/z range (data not shown). This con-
trasts with that described by Stübiger et al.[14] and Groleau
et al.[15] that only detected up to two acetylations of this
glycopeptide form. On the other hand, glycoforms showing
N‐glycolylneuraminic acid (Neu5Gc) instead of Neu5Ac were
also observed in O126 but in very low abundance (e.g. Fig. 2
(a), (iv)). N83‐4Ant mass spectra showed higher complexity
with great number of Na, K and NH4 adducts due to the
higher degree of glycosylation of this glycopeptide in
RESULTS AND DISCUSSION
Analysis of rhEPO peptides
In a previous study, peptides obtained by trypsin‐peptide: N‐
glycosidase F (PNGase) enzymatic digestion of rhEPO and
NESP (novel erythropoiesis stimulating protein) glycopro-
teins were analyzed by capillary electrophoresis ion trap
mass spectrometry (CE/ITMS) using an acidic background
electrolyte (BGE) of 50 mM HAc and 50 mM HFor (pH
2.2).[10] As oaTOF‐MS provides higher sensitivity, mass
accuracy and resolving power than ITMS,[12] we analyzed
rhEPO digested with trypsin (rhEPO (T)) by CE/TOFMS
with the aim of increasing the coverage of the protein
sequence obtained by CE/ITMS and also studying the dif-
ferent O‐ and N‐glycopeptides contained in the rhEPO (T)
sample (see next section). Figure 1 shows the extracted ion
electropherograms (EIEs) for the detected peptides of rhEPO
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316
Copyright © 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm

E. Giménez et al.
Peptides from rhEPO (T) (EIEs)
x106
4
2
1.8
1.6
1.4
1.2
1
8
3
9
5
0.8
0.6
0.4
0.2
0
10
12
7
11
15
2
1
14
13
6
16
6.2 6.6
7
7.4 7.8 8.2 8.6
9
9.4 9.8 10.2 10.6 11 11.4 11.8
Time (min)
Figure 1. Extracted ion electropherograms (EIEs) of peptides from rhEPO (T) by
CE/TOFMS. Peptide peak numbers are referred to their identification numbers
presented in Table 1. (Sample: 1000 mg·L–1 of digested rhEPO; injection: 15 s at
50 mbar; voltage: +18 kV; T: 25 °C BGE: 50 mM HAc and 50 mM HFor pH 2.2.
Fragmentor voltage 270 V).
Table 1. Peptides detected from rhEPO (T) by CE‐TOFMS with and without reference compounds in the SL
WITH REFERENCE (n = 6)
WITHOUT REFERENCE (n = 6)
S/N
Mexp (Da)b
RSD
Mexp (%) (ppm)d
Av. error S/N Mexp (Da)b
RSD
Av. error
Position
MTheo
(Da)b
Mexp (%) (ppm)d
Sequencea
1. APPR
1‐4
439,2543 1732,5
762,3694 1311,0
439,2540
762,3684
4,32 x10‐5
7,35 x10‐5
0,68
1,29
3891,3 439,2555 3,27 x10‐4
2,69
4,92
2. LICDSR‐
(Cys‐CAM)c
3. VLER
5‐10
2768,0 762,3732 8,89 x10‐5
8136,4 515,3072 2,84 x10‐4
11‐14
15‐20
46‐52
515,3067 4537,3
735,4167 8953,3
926,4650 2200,0
515,3065
735,4166
926,4644
4,82 x10‐5
1,70 x10‐4
1,23 x10‐4
4,96 x10‐5
0,42
1,03
3,08
4,38
7,95
4. YLLEAK
5. VNFYAWK
6. MEVGQQA-
VEVWQ
0,15 14086,1 735,4190 2,53 x10‐4
0,61
4,78
3190,2 926,4691 3,12 x10‐4
4686,7 2525,3110
1,08 x10‐4
54‐76 2525,3311 3162,0 2525,3190
GLALLSEAVLR
7. AVSGLR
8. SLTTLLR
9. ALGAQK
10. TITADTFR
11. LFR
12. VYSNFLR
13. GK
14. LK
98‐103 601,3547 5844,7
104‐110 802,4912 4537,4
111‐116 586,3438 4710,1
132‐139 923,4712 5329,1
141‐143 434,2641 4116,7
144‐150 897,4708 3623,8
151‐152 203,1270 646,1
153‐154 259,1896 1210,2
601,3546
802,4919
586,3437
923,4707
434,2640
897,4695
203,1270
259,1896
968,4367
3,71 x10‐5
5,15 x10‐5
3,24 x10‐5
5,68 x10‐5
4,70 x10‐5
7,34 x10‐5
1,38 x10‐5
3,79 x10‐5
1,43 x10‐5
0,19
9349,3 601,3548 2,86 x10‐4
0,17
4,20
0,54
2,17
1,84
4,57
4,55
0,93
1,51
0,83 10809,2 802,4946 7,55 x10‐5
0,17
6482,7 586,3441 3,46 x10‐4
0,60 10799,2 923,4732 2,86 x10‐4
0,27
1,47
0,17
0,06
1,86
6095,0 434,2649 2,72 x10‐4
4583,5 897,4749 8,19 x10‐5
326,8 203,1279 2,03 x10‐4
1250,4 259,1898 1,45 x10‐4
2916,8 968,4400 2,94 x10‐4
15. LYTGEACR‐ 155‐162 968,4385 1615,3
(Cys‐CAM)c
16. TGDR
163‐166 447,2077
‐
‐
‐
‐
59,9 447,1866 2,85 x 10‐4 47,20
aAA sequence of EPO from SwissProt.
bMTheo and Mexp monoisotopic molecular masses.
cCysteine was treated with iodoacetamide to form carbamidomethyl‐cysteine (Cys‐CAM).
dAverage error was calculated as: ∣(Mexp‐Mtheo) / Mtheo
∣
comparison with the O126 glycopeptide (see Fig. 2(b), (ii‐iv)).
Moreover, we were not able to detect acetylated Neu5Ac
and Neu5Gc glycoforms of N83‐4Ant probably because sensi-
tivity was not enough to detect these less abundant species.
Table 2 shows the most abundant glycoforms detected for
compounds in the SL. The Mtheo, Mexp, S/N values, RSD Mexp
and av. errors are also illustrated in Table 2. As happened
before for the rhEPO peptides (see previous section), glyco-
form S/N values decreased in the presence of reference com-
pounds. However, in contrast to the results obtained for the
peptides of the digest, now the presence of the reference
O
126 and N83‐4Ant glycopeptides with and without reference
wileyonlinelibrary.com/journal/rcm
Copyright © 2011 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316

Elucidation of a glycopeptide map of rhEPO
Figure 2. CE/TOFMS analysis of the main (a) O126 and (b) N83‐4Ant glycopeptide glycoforms.
(i) Sum of EIEs and (ii), (iii) and (iv) m/z spectra for the different glycoforms. (Sample: 1000 mg·L–1
of digested rhEPO; injection: 15 s at 50 mbar; voltage: +18 kV; T: 25 °C BGE: 50 mM HAc and
50 mM HFor pH 2.2. Fragmentor voltage 270 V).
compounds in the SL allowed slightly better Mexp repeatabil-
ity but had a detrimental effect on Mexp accuracy. Hence, the
presence of reference compounds impedes glycopeptide
detection without improving mass accuracy. For this reason,
we continued our investigation without using reference
compounds in the SL, and we recommend avoiding their
use for the analysis of glycopeptides or peptides at low
concentration.
As is well known, one of the main limitations of CE is the
adsorption of the analytes in the fused‐silica capillary inner
surface. This phenomenon is especially important in analytes
of peptide nature with a larger molecular mass such as glyco-
peptides, as they strongly interact with the negative charges
of the silanols of the capillary wall. This adsorption usually
provides low separation efficiency and poor reproducibility
in terms of migration time and peak area. The use of coated
capillaries is probably the best alternative to solve this nega-
tive effect.[17] However, coatings are usually incompatible
with MS detection since they bleed to some extent producing
contamination of the mass spectrometer.[3] In this work, with
the aim of solving glycopeptide adsorption on the capillary
wall avoiding the need to use coatings, two capillary washing
procedures between analyses were evaluated: washing the
capillary with BGE or with 1 M HAc before the BGE. Figure 3
shows, as an example, the results obtained with O126 glyco-
peptide glycoforms with both capillary washing procedures.
As can be observed in the sum of EIEs of Fig. 3(a), BGE wash-
ing was not enough to avoid carry‐over as one extra peak
was detected at 16 min, which corresponded to a mixture
of various O126 glycopeptide glycoforms (O126/0Neu5Ac,
O
126/1Neu5Ac, O126/2Neu5Ac). On the contrary, 1 M HAc
washing seemed to solve this problem, as the extra peak
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316
Copyright © 2011 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

E. Giménez et al.
Table 2. Main O126 and N83‐4Ant glycopeptide glycoforms detected from rhEPO (T)
WITH REFERENCE (n = 6)
Glycopeptide
WITHOUT REFERENCE (n = 6)
glycoforms
rhEPO (T)
digest
RSD
Mexp
(%)
RSD
Mexp
(%)
a
a
a
MTheo
(Da)
Mexp
Av. error
(ppm)b
Mexp
Av. error
(ppm)b
S/N
51,9
(Da)
S/N
45,6
(Da)
O126
O126
0Neu5Ac
O126
1Neu5Ac
O126
/
1829,8895
1829,8838
1,09
x10‐4
5,67
3,09
2,96
2,04
7,51
4,03
3,78
1829,8852
2,35
x10‐4
1,03
2,32
1,44
0,35
4,83
2,05
5,27
/
2120,9849 2077,1 2120,9786
2412,0803 2078,9 2412,0754
3096,3 2120,9818
5805,3 2412,0795
x10‐5
1,16
x10‐4
2,27
/
2Neu5Ac
N83‐4Ant/
2Neu5Ac
N83‐4Ant/
3Neu5Ac
N83‐4Ant/
4Neu5Ac
x10‐4
5,88
x10‐4
4,81
N83‐
4Ant
5439,3280
5730,4234
37,8
5439,2871
5730,4003
70,6
5439,3017
5730,4117
x10‐4
1,24
x10‐4
3,08
528,4
827,8
x10‐4
3,68
x10‐4
4,46
6021,5188 1411,0 6021,4961
2145,2 6021,4871
x10‐4
x10‐3
aMTheo and Mexp : Monoisotopic molecular masses.
bAverage error was calculated as: ∣(Mexp‐Mtheo) / Mtheo
∣
Figure 3. Sum of the EIEs for the CE/TOFMS analysis of the main O126 glycopeptide
glycoforms using two different capillary conditionings between analyses: (a)
BGE capillary washing and (b) 1 M HAc capillary washing before the BGE. (Sam-
ple: 1000 mg·L–1 of digested rhEPO; injection: 15 s at 50 mbar; voltage: +18 kV;
T: 25 °C BGE: 50 mM HAc and 50 mM HFor pH 2.2. Fragmentor voltage 270 V).
wileyonlinelibrary.com/journal/rcm
Copyright © 2011 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316

Elucidation of a glycopeptide map of rhEPO
Figure 4. EIEs of the main (a) O126 and (b) N83‐4Ant glycopeptide glycoforms by
CE/TOFMS. (Sample: 1000 mg/mL of protein digest; injection: 15 s at 50 mbar;
voltage: +18 kV; T: 25 °C BGE: 50 mM HAc and 50 mM HFor pH 2.2. Fragmentor
voltage 270 V).
completely disappeared (see Fig. 3(b)). Furthermore, %RSD of
as O126/1Neu5Ac (~12.5 min). Similarly, the glycoforms with
0Neu5Ac and 1Neu5Ac units were detected at the migration
time corresponding to O126/2Neu5Ac (~15.5 min). That frag-
mentation behavior was also observed for N83‐4Ant glyco-
peptide glycoforms (Fig. 4(b)). Thus, glycoforms with
3Neu5Ac and 2Neu5Ac were detected with N83‐4Ant/
4Neu5Ac, while the glycoform with 2Neu5Ac was detected
with N83‐4Ant/3Neu5Ac. Sugar fragmentation seemed to be
produced during the ionization process and not during the
electrophoretic separation, as the precursor and product ions
appeared at the same migration time regardless of their elec-
trophoretic mobilities. Following the same argument, other
product ions could be observed apart from those resulting
from Neu5Ac fragmentation. By way of example, Table 3
shows the detected product ions coming from O126/2Neu5Ac
migration times improved from ~3% to ~0.5% confirming
that 1 M HAc washing improved the repeatability of the
method. Consequently, this capillary conditioning was con-
sidered the most appropriate for the analysis of glycopeptides
by CE/TOFMS.
The established method allowed to detect O126 and N83‐
4Ant from rhEPO tryptic digests. Nevertheless, we noticed
these glycopeptides were susceptible to fragmentation. The
most predominant fragmentation was the loss of Neu5Ac
residues due to the lability of these sugar units. Figure 4
shows the EIEs of the main detected glycoforms of O126 and
N83‐4Ant glycopeptides (Figs. 4(a) and 4(b), respectively).
Each solid line corresponds to a different glycoform. As can
be observed in Fig. 4(a), due to fragmentation, a small propor-
tion of O126/0Neu5Ac appeared at the same migration time
and N83‐4Ant/4Neu5Ac precursor ions. In the case of O126
/
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316
Copyright © 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm

E. Giménez et al.
Table 3. Glycopeptide fragmentation observed by CE‐TOFMS for O126/2Neu5Ac and N83‐4Ant/4Neu5Ac.
Glycopeptide glycoform
(Parent Ion)
Observed Fragmentationa
(Product Ions)
O126
Pep+HexNAc+Hex+2Neu5Ac
Pep+HexNAc+Hex+Neu5Ac
Pep HexNAc+1Neu5Ac
Pep+HexNAc+Hex
Pep+HexNAc
Pep
N83‐4Ant
Pep+6HexNAc+7Hex+1Fuc+4Neu5Ac
Pep+6HexNAc+7Hex+1Fuc+3Neu5Ac
Pep+6HexNAc+7Hex+1Fuc+2Neu5Ac
Pep+5HexNAc+6Hex+1Fuc+3Neu5Ac
Pep+4HexNAc+5Hex+1Fuc+2Neu5Ac
Pep+1HexNAc+1Fuc
Pep+1HexNAc
Pep
aPep=peptide sequence, Neu5Ac=N‐acetylneuraminic acid, Hex=hexose, HexNAc=Nacetylhexosamine and Fuc=fucose.
a) O126/2Neu5Ac
Pep+HexNAc+Hex+2Neu5Ac
Parent Ion
Pep+HexNAc+Hex+1Neu5Ac
Pep+HexNAc+1Neu5Ac
Pep+HexNAc+Hex
Pep+HexNAc
100
90
80
70
60
50
40
30
20
10
0
Product Ions
Pep
S/N:
24408.1
S/N:
30488.7
S/N:
35654.1
S/N:
24488.5
S/N:
18692.9
S/N:
22863.3
160
190
215
240
270
325
b) N83-4Ant/4Neu5Ac
Pep+6HexNAc+7Hex+1Fuc+4Neu5Ac
Pep+6HexNAc+7Hex+1Fuc+3Neu5Ac
Parent Ion
100
Pep+6HexNAc+7Hex+1Fuc+2Neu5Ac
Pep+5HexNAc+6Hex+1Fuc+3Neu5Ac
Pep+4HexNAc+5Hex+1Fuc+2Neu5Ac
Pep+1HexNAc+1Fuc
90
Product Ions
S/N:
80
70
60
50
40
30
20
10
0
13648.4
Pep+1HexNAc
Pep
S/N:
16758.7
S/N:
15615.3
S/N:
17906.5
S/N:
S/N:
15178.2
7186.3
160
190
215
240
270
325
Fragmentor Voltage (V)
Figure 5. Percentage of peak area (%A) (n = 6) obtained for (a) O126/2Neu5Ac
and (b) N83‐4Ant/4Neu5Ac precursor ions and their corresponding product ions
at the different fragmentor voltage values by CE/TOFMS. Error bars show the
standard deviations (s). S/N values for each precursor ion are also indicated.
wileyonlinelibrary.com/journal/rcm
Copyright © 2011 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316

Elucidation of a glycopeptide map of rhEPO
2Neu5Ac, we observed all possible fragmentations: Desialy-
lations but also the loss of Hex and HexNAc units, leaving
O126 peptide free of sugars. Some authors described the pre-
sence of the non‐glycosylated O126 in rhEPO samples by
MALDI‐MS.[14] However, our results suggest the peptide
detected by Stübiger et al.[14] probably originated from the
fragmentation of the glycopeptides taking into account that
intact glycoproteins and glycopeptides may undergo frag-
mentation easily by MALDI‐MS depending on the detection
conditions.[8,18] As can be observed in Table 3, N83‐4Ant/
4Neu5Ac showed more product ions than O126/2Neu5Ac
due to the higher sugar content of this glycopeptide.
Tandem mass spectrometry is widely used for the charac-
terization of glycans and glycopeptides as fragmentation
patterns may reveal finer details of glycan structure to com-
pletely characterize the various post‐translational modi-
fications of a glycoprotein.[19–21] However, in our case,
glycopeptide fragmentation must be prevented as the aim
was the comprehensive detection of the intact glycopeptide
glycoforms of the protein digest to achieve a reliable charac-
terization of the glycoprotein. Hence, there can be no doubts
as to whether the detected glycoforms were present in the
native glycoprotein or came from the fragmentation of the
native glycopeptide glycoforms.
obtained for each precursor ion at the different fragmentor
voltages are also illustrated in Fig. 5. Although we observed
a slight fragmentation of O126/2Neu5Ac mainly due to the
loss of Neu5Ac residues, the abundance of the precursor ion
at low fragmentor voltage values (160–190 V) was the highest
(near 100%) (Figs. 5(a) and 5(b)). As the value increased, frag-
mentation of the precursor ion was promoted, increasing the
percentage of the product ions (see Fig. 5). This was also the
reason why in general S/N values of the precursor ion were
lower at higher fragmentor voltage values. The fragmentor
voltage 190 V was selected as the optimum fragmentor vol-
tage value for O126/2Neu5Ac since it provided higher S/N
values than 160 V and fragmentation was rather low in com-
parison with that obtained at 215 V. Regarding N83‐4Ant/
4Neu5Ac, no fragmentation was observed at 160–190 V (see
Fig. 5(b)) which confirmed the higher stability of N83‐4Ant
compared to O126 glycopeptides as other authors sug-
gested.[15] An optimal fragmentor voltage of 190 V was also
selected for N83‐4Ant/4Neu5Ac as it gave better S/N values
than 160 V. Hence, although N83‐4Ant/4Neu5Ac and O126
/
2Neu5Ac glycopeptides showed a great number of structural
differences, we were able to select an optimum fragmentor
voltage value that minimized fragmentation for both glyco-
peptides. These conditions may be useful to obtain a reliable
glycopeptide map of rhEPO and other glycoproteins.
One explanation for the observed fragmentation could be
the acidic conditions of the BGE and SL as, for example, sialic
acids (Neu5Ac or Neu5Gc) can be hydrolyzed under acidic
conditions. Using the acidic BGE of 50 mM HAc and 50 mM
HFor (pH 2.2), we tested different percentages of HFor in
the SL: 0.01, 0.05 and 0.1% v/v of HFor. The percentages of
peak area (%A), measured from the extracted ion electro-
pherogram (EIE), of each precursor and their corresponding
product ions were selected to evaluate the extent of
CONCLUSIONS
A CE/TOFMS method using a volatile BGE of 50 mM HAc
and 50 mM HFor at pH 2.2 was developed for the analysis
of glycopeptides present in rhEPO tryptic digests. A capillary
conditioning consisting in 1 M HAc washing was selected to
prevent carry over due to glycopeptide adsorption on the
capillary wall. It was demonstrated that the presence of refer-
ence compounds in the SL had a detrimental effect on the
detection of the peptides and glycopeptides of the protein
digest. Furthermore, we proved the importance of the se-
lection of an appropriate fragmentor voltage value for a
unambiguous glycopeptide detection and quantification by
CE/TOFMS. We were able to select the same optimum frag-
mentor voltage value for N83‐4Ant and O126 despite the great
structural differences between both glycopeptides. These
conditions may be regarded as an excellent starting point to
obtain reliable glycopeptide maps of complex glycoprotein
enzymatic digests by CE/TOFMS.
O
126/2Neu5Ac and N83‐4Ant/4Neu5Ac glycopeptide frag-
mentation. The percentage of peak area (%A) obtained for
the product ions of the glycopeptides remained practically
constant at the different percentages of HFor. The SL with
0.1% of HFor was discarded because it provided lower sensi-
tivity due to the increase in the background signal. We
selected 0.05% v/v of HFor as the optimum percentage since
migration time and S/N values were more reproducible than
with 0.01% v/v of HFor. As decreasing the acidity of the SL
did not avoid fragmentation, we explored the use of a basic
BGE (50 mM NH4Ac, pH 8). However, the results were worse
than those obtained with the acidic BGE. Glycopeptides were
poorly separated, sensitivity extremely decreased and frag-
mentation of O126/2Neu5Ac and N83‐4Ant/4Neu5Ac was
also observed.
Once we had investigated if the acidic conditions were
responsible for the glycopeptide fragmentation, we consid-
ered the influence of the selected TOF parameters. The frag-
mentor voltage is the MS parameter that accelerates the ions
of interest towards the detector. However, its value has been
described to be directly related to in‐source collision‐induced
dissociation (ISCID) of neuropeptides, pesticides and antibio-
tics.[12,22,23] Therefore, six fragmentor voltage values were
evaluated: 160, 190, 215, 240, 270 and 325 V. Figure 5 shows
the percentage of peak area (%A) of O126/2Neu5Ac and
N83‐4Ant/4Neu5Ac precursor ions and their corresponding
product ions at the different fragmentor voltage values (Figs. 5
(a) and 5(b), respectively). Error bars show the standard
deviations obtained for %A (n = 6). Moreover, S/N values
Acknowledgements
Part of this work was supported by the Spanish Ministry of
Science and Innovation (CTQ2008‐00507/BQU).
REFERENCES
[1] R. L. J. Graham, S. Hess. Mass spectrometry in the elucida-
tion of the glycoproteome of bacterial pathogens. Curr.
Proteomics 2010, 7, 57.
[2] E. Balaguer, U. M. Demelbauer, M. Pelzing, V. Sanz‐Nebot,
J. Barbosa, C. Neusüβ. Glycoform characterization of eryth-
ropoietin combining glycan and intact protein analysis by
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316
Copyright © 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm

E. Giménez et al.
capillary electrophoresis‐electrospray‐time‐of‐flight mass
spectrometry. Electrophoresis 2006, 27, 2638.
[13] Agilent 6200 Series TOF and 6500 Series Q‐TOF LC/MS
System, Concepts Guide, The Big Picture, (5th edn.), Agilent
Technologies, Santa Clara, CA, USA, 2008.
[14] G. Stübiger, M. Marchetti, R. Nagago, G. Grimm, G. Gmeiner,
C. Reichel, G. Allmaier. Characterization of N‐ and O‐
glycopeptides of recombinant human erythropoietins as
potential biomarkers for doping analysis by means of
microscale sample purification combined with MALDI‐
TOF and quadrupole IT/RTOF mass spectrometry. J. Sep.
Sci. 2005, 28, 1764.
[3] E. Giménez, F. Benavente, J. Barbosa, V. Sanz‐Nebot. Analy-
sis of intact erythropoietin and novel erythropoiesis stim-
ulating protein by capillary electrophoresis‐electrospray
ion‐trap mass spectrometry. Electrophoresis 2008, 29, 2161.
[4] S. Amon, A. D. Zamfir, A. Rizzi. Glycosylation analysis of
glycoproteins and proteoglycans using capillary electro-
phoresis‐mass spectrometry strategies. Electrophoresis 2008,
29, 2485.
[5] Z. El Rassi. Electrophoretic and electrochromatographic
separation of proteins in capillaries: an update covering
2007–2009. Electrophoresis 2010, 31, 174.
[6] A. Zamfir, J. Peter‐Katalinić. Capillary electrophoresis‐mass
spectrometry for glycoscreening in biomedical research.
Electrophoresis 2004, 25, 1949.
[7] L. R. Ruhaak, C. Huhn, W.‐J. Waterreus, A. R. de Boer,
C. Neusüβ, C. H. Hokke, A. M. Deelder, M. Wuhrer. Hydro-
philic interaction chromatography‐cased high‐throughput
sample preparation method for N‐glycan analysis from total
human plasma glycoproteins. Anal. Chem. 2008, 80, 6119.
[8] S. Amon, A. Plematl, A. Rizzi. Capillary zone electrophor-
esis of glycopeptides under controlled electroosmotic flow
conditions coupled to electrospray and matrix‐assisted laser
desorption/ionization mass spectrometry. Electrophoresis
2006, 27, 1209.
[15] P. E. Groleau, P. Desharnais, L. Cote, C. Ayotte. Low LC‐
MS/MS detection of glycopeptides released from pmol
levels of recombinant erythropoietin using nanoflow HPLC‐
chip electrospray ionization. J. Mass Spectrom. 2008, 43, 924.
[16] H. J. Boss, D. B. Watson, R. S. Rush. Peptide capillary zone
electrophoresis mass spectrometry of recombinant human
erythropoietin: An evaluation of the analytical method. Elec-
trophoresis 1998, 19, 2654.
[17] C. Huhn, R. Ramautar, M. Wuhrer, G. W. Somsen. Rele-
vance and use of capillary coatings in capillary electrophor-
esis–mass spectrometry. Anal. Bioanal. Chem. 2010, 396, 297.
[18] E. Giménez, F. Benavente, J. Barbosa, V. Sanz‐Nebot.
Towards a reliable molecular mass determination of intact
glycoproteins by matrix‐assisted laser desorption/ioniza-
tion time‐ of‐flight mass spectrometry. Rapid. Commun. Mass
Spectrom. 2007, 21, 2555.
[9] E. Balaguer, C. Neusüβ. Glycoprotein characterization
combining intact protein and glycan analysis by capillary
electrophoresis‐electrospray ionization‐mass spectrometry.
Anal. Chem. 2006, 78, 5384.
[10] E. Giménez, F. Benavente, C. de Bolós, E. Nicolás, J. Barbosa,
V. Sanz‐Nebot. Analysis of recombinant human erythro-
poietin and novel erythropoiesis stimulating protein digests
by immunoaffinity capillary electrophoresis–mass spectro-
metry. J. Chromatogr. A 2009, 1216, 2574.
[11] J. Segura, J. A. Pascual, R. Gutiérrez‐Gallego. Procedures for
monitoring recombinant erythropoietin and analogues in
doping control. Anal. Bioanal. Chem. 2007, 388, 1521.
[12] M. Borges‐Alvarez, F. Benavente, E. Giménez, J. Barbosa, V.
Sanz‐Nebot. Assessment of capillary electrophoresis TOF
MS for a confident identification of peptides. J. Sep. Sci.
2010, 33, 2489.
[19] Y. Mechref, M.V. Novotny. Glycomic analysis by capillary
electrophoresis–mass spectrometry. Mass Spectrom. Rev.
2009, 28, 207.
[20] J. Zaia. On‐line separations combined with MS for analysis
of glycosaminoglycans. Mass Spectrom. Rev. 2009, 28, 254.
[21] J. Zaia. Mass spectrometry of oligosaccharides. Mass Spec-
trom. Rev. 2004, 23, 161.
[22] A. Polettini, R. Gottardo, J. P. Pascali, F. Tagliaro. Implemen-
tation and performance evaluation of a database of chemical
formulas for the screening of pharmaco/toxicologically rele-
vant compounds in biological samples using electrospray
ionization‐time‐of‐flight mass spectrometry. Anal. Chem.
2008, 80, 3050.
[23] I. Ferrer, E. M. Thurman, A. R. Fernández‐Alba. Quantita-
tion and accurate mass analysis of pesticides in vegetables
by LC/TOF‐MS. Anal. Chem. 2005, 77, 2818.
wileyonlinelibrary.com/journal/rcm
Copyright © 2011 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2011, 25, 2307–2316