Journal of Chromatography B, 798 (2003) 17

Characterization of human transferrin glycoforms by capillary

electrophoresis and electrospray ionization mass spectrometry
V. Sanz-Nebot
a,
, P. Gonzlez
a
, I. Toro
a
, A. Ribes
b
, J. Barbosa
a
a
Department of Analytical Chemistry, University of Barcelona, Av. Diagonal 647, 08028 Barcelona, Spain
b
Institut de Bioqumica Clnica, 08028 Barcelona, Spain
Received 2 August 2001; received in revised form 18 June 2002; accepted 28 June 2002
Abstract
Carbohydrate Decient Glycoprotein Syndrome (CDGS) is an inherited metabolic disease affecting all parts of the body. The biochemical
diagnosis of this syndrome is based on the presence of a special marker in blood, Carbohydrate Decient Transferrin (CDT), which is also a
marker of chronic alcohol abuse. CDT is characterized by abnormal glycoforms of serum transferrin (Tf). In the present study, electrophoretic
separation of human serumtransferrin glycoforms was carried out using a bare fused-silica capillary and the glycoforms present in commercial
Tf were baseline separated. The limit of detection (LOD) of human Tf was around the nmol concentration range. The LOD of the trisialo-
and disialo-Tf, expressed as percentages of the tetrasialo-Tf peak area, were 0.5% for trisialo-Tf and 0.4% for disialo-Tf, and these values
were appropriate for CDGS diagnosis. Moreover, Tf glycoforms were characterized using mass spectrometry (MS). The method was applied
to the analysis of normal and pathological serum samples, after dilution. The results obtained suggest a way of making a rapid and simple
CDGS diagnosis.
2003 Elsevier B.V. All rights reserved.
Keywords: Carbohydrate decient glycoprotein syndrome; Glycoforms; Transferrin
1. Introduction
Glycoproteins contain carbohydrate chains covalently
attached to their peptide structure by glycosidic linkages.
There is a polymorphism associated with type, number and
position of carbohydrate chains. This kind of diversity is
known as microheterogeneity and the different species gen-
erated are called glycoforms [1]. Whole glycoforms found
for a particular glycoprotein are related to the species and
tissues involved. The glycoform range associated with a
single glycoprotein may change during normal tissue devel-
opment or as a result of pathological processes [2]. This is
the case for carbohydrate decient glycoprotein syndrome
(CDGS). Patients with this syndrome present hypoglycosy-
lation of different plasmatic glycoproteins, transferrin being
one of these.

Corresponding author. Tel.: +34-93-402-9083;

fax: +34-93-402-1233.
E-mail address: [email protected] (V. Sanz-Nebot).
Transferrin (Tf) is the main iron-transporting glycopro-
tein in plasma [3]. It has a polypeptidic chain of 679 amino
acids and two N-glycosylation sites at Asn
413
and Asn
611
(asparagine at positions 413 and 611, respectively) [2]. Over
6% of its molecular mass is accounted for by two sugar
chains that may be mainly bi-antennary [4]. Each chain ter-
minates with sialic acid residues (SA) which give a net neg-
ative charge to the protein.
The sialic acid content affects the electrophoretic be-
haviour of transferrin, making it possible to separate differ-
ent sialoforms with the right technique [2]. Approximately
85% of Tf in normal serum is in the tetrasialo form, with
two N-linked disialylated biantennary sugar chains, known
as S
4
. Other sialoforms such as pentasialo-, hexasialo- and
trisialo-transferrin, known as S
5
, S
6
and S
3
, respectively, are
present at low concentrations in healthy serum. The isoelec-
tric point (pI) of transferrin glycoforms (Tf-glycoforms) is
in the range of 56.
Carbohydrate-decient transferrin (CDT) is typically con-
sidered to have less than three sialic acid residues. An in-
crease in the di-, mono- and asialoforms (i.e. CDT), named
1570-0232/$ see front matter 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S1570-0232(02)00442-7
2 V. Sanz-Nebot et al. / J. Chromatogr. B 798 (2003) 17
S
2
, S
1
and S
0
, respectively, has been noted only with the
above mentioned syndrome (CDGS) and chronic alcoholism
[2,4]. For historical reasons, the diagnosis of CDGS is based
upon the glycoform pattern observed for transferrin [2].
CDGS diagnosis is based on identication and quantitation
of CDT, mainly of S
2
and S
0
, in which one or both N-linked
disialylated biantennary sugar chains are missing, respec-
tively [2,4,5].
Until recently, the analytical determination of CDT was
based on isoelectric focusing (IEF) combined with immuno-
xation, zone immunoelectrophoresis or Western blotting,
as well as chromatographic techniques (i.e. anion-exchange
chromatography or chromatofocusing) followed by im-
munoassays. These multistep techniques are clearly too
complex and time consuming for routine application and are
unreliable due to the complicated and manual procedures
involved [6,7].
Capillary electrophoresis (CE) is a relative new tool for
the analysis of biologically active compounds [8]. As this
technique can improve the resolution of complex mixtures
under non-denaturing conditions, based on the differences
in charge-to-mass ratios, it has several advantages over con-
ventional techniques. However, electrophoretic separation
of proteins in uncoated fused-silica capillaries is more prob-
lematic, due mainly to protein interactions either with the
inner capillary wall or with other proteins. This problem has
been circumvented through the use of chemical additives
that coat the wall in order to reduce proteinwall interac-
tions [3,9]. It has also been shown that buffer composition
plays a role in separation of the different species. This is the
case for sodium borate buffer, which has been proposed to
play a key role in separation by complexing with the diols
of specic carbohydrate moieties on different glycoproteins
[1,6,7,10,11]. CE has been used to separate the isoforms
of different glycoproteins [1,8,10,1217] Tf being one of
them [6,7,18,19]. The sialic acid residues are deprotonated
at basic pH and this generates different charge/mass ra-
tios for each glycoform. This means that glycoforms can
be separated by CE due to their different electrophoretic
mobilities.
Mass spectrometry (MS) has also been widely used in
biochemical analysis [2028]. The development of ioniza-
tion methods, such as electrospray (ES), has enabled it to be
used for characterizing polar substances and biomolecules
of molecular mass up to 100 000 Da. The ability of ES to
produce multiply-charged ions, such as polyprotonated pep-
tides and proteins, has generated a wide and interesting eld
of research. Thus, mass spectrometry with electrospray ion-
ization is a useful tool in characterizing Tf-glycoforms.
In this work, an electrophoretic method was developed
in order to separate glycoforms in a recombinant Tf. The
method was applied to the analysis of normal and patholog-
ical serum samples, after dilution. Moreover, Tf glycoforms
were characterized using mass spectrometry (MS). The re-
sults obtained permit us to propose a method for a rapid and
simple CDGS diagnosis.
2. Experimental
2.1. Chemicals and reagents
Human transferrin (partially saturated, min. 98%
pure) was purchased from Sigma (Madrid, Spain). Neu-
raminidase (from Vibrio Cholerae, 2.1 U/ml), sodium
azide, l-glutamic acid and 1,4-butanediol of HPLC-grade
were purchased from Fluka (Madrid, Spain). Acetonitrile,
methanol, 2-propanol, ethyleneglycol, sodium tetraborate,
boric acid, sodium hydrogencarbonate, hydrochloric acid,
acetone, triuoroacetic acid and sodium hydroxide were
obtained from Merck (Darmstadt, Germany). Heptauo-
robutyric acid, ethanolamine, hexamethonium bromide and
1,4-diaminobutane (DAB) were purchased from Aldrich
(Madrid, Spain). Tris, ethylenediamine and triethanolamine
were obtained from J.T. Baker (Deventer, Holland). Iron
chloride was purchased from Panreac (Barcelona, Spain).
Water, with a resistivity of 18.2 M cm, was obtained using
a Milli-Q water purication system (Millipore, Molsheim,
France).
2.2. Instrumental equipment
2.2.1. Capillary electrophoresis
All CE experiments were performed on a P/ACE System
5500 (Beckman Instruments, Palo Alto, CA, USA) equipped
with an autosampler, automatic injector and a photodiode
array detector. The distance between the detection window
and the end of the capillary was 7 cm. Electropherograms
were recorded using a computer program(P/ACE Station 1.0
with interface Golden System) supplied by Beckman. A 57
cm50 m I.D. uncoated fused-silica capillary (Polymicro
Technologies, Phoenix, AZ, USA) was inserted in a capillary
cartridge thermostated to 25

C (0.1

C). Samples were

injected hydrodynamically at 0.5 p.s.i. (515 s). Experiments
were conducted under normal polarity applying different
separation voltages (1030 kV). Detection of Tf-glycoforms
was performed at 200 nm.
2.2.2. pH measurements
pH measurements were carried out with a Crison 2002
potentiometer (Crison Instruments, Barcelona, Spain),
equipped with a Ross electrode 8102 (Orion Research,
Boston, MA, USA).
2.2.3. Mass spectrometry
A VG Platform mass spectrometer (Fisons Instruments,
Manchester, UK) equipped with a pneumatically assisted
atmospheric pressure ionization (API) source was used.
The instrument was periodically calibrated using a solution
of bovine serum albumin (BSA). For the direct injection of
transferrin solutions an injection loop of 5 l with a ow-rate
of 10 l/min of mobile phase was used. The data acquisition
was made in Multi Channel Analysis (MCA) mode. With
this kind of acquisition, intensity data for each acquired scan
V. Sanz-Nebot et al. / J. Chromatogr. B 798 (2003) 17 3
is added to the accumulated data of previous scans, and only
one spectrum, obtained in full scan mode (m/z 16003000),
is registered for every injection. Instrument and data analysis
were carried out using MassLynx application software from
Micromass (Manchester, UK). Data obtained were analyzed
with the MaxEnt algorithm of MassLynx. The MaxEnt
algorithm works iteratively to nd the simplest molecular
mass spectra from multiply charged electrospray spectra.
2.3. Electrolyte solutions
2.3.1. Capillary electrophoresis
A pH range of 79.5 was covered using two types of run-
ning buffer solutions: 25 mM sodium tetraborate, adjusting
until the desired pH with 100 mM boric acid; Trisglutamate
buffer (50 mM Tris and 50 mM monosodium glutamate),
adjusting the pH with concentrated sodium hydroxide. For
CE experiments, Tf was dissolved in water (1 g/l) and
saturated with iron, adding 18 l of 10 mM FeCl
3
and 25 l
of 500 mM NaHCO
3
, in order to avoid heterogeneity in the
charge of the individual Tf glycoforms due to incomplete
iron saturation. The sample also contained acetone at 5%
(v/v) as electroosmotic ow (EOF) marker [29,30]. Samples
and running electrolytes were passed through a 0.45-m ny-
lon lter (MSI).
2.3.2. Mass spectrometry
For mass spectrometry experiments, Tf was dissolved in
water (0.44 g/l). The solvent used for these experiments
was a mixture of water/acetonitrile (50:50 v/v) containing
0.25% (v/v) triuoroacetic acid (TFAA) or 0.1% (v/v) hep-
tauorobutyric acid (HFBA). Samples and running elec-
trolytes were passed through a 0.45-m nylon lter (MSI).
2.4. Capillary treatment
All capillary rinses were performed at high pressure (20
p.s.i.). New capillaries were activated by ushing them for
20 min with 1 M NaOH followed by water for 20 min and
nally conditioned with running buffer for 60 min. Between
days or after a change of buffer, the capillary was condi-
tioned by rinsing successively for 20 min with 0.1 M NaOH,
20 min with water and 60 min with running buffer. Finally,
the system was conditioned by applying a constant voltage
for about 10 min, until a steady baseline was obtained. It
was empirically demonstrated that this nal step accelerates
capillary equilibration. Between runs, the capillary was suc-
cessively rinsed with 30 s of water and 1 min of running
buffer. It was stored overnight lled with working buffer
electrolyte.
2.5. Experimental procedure
2.5.1. Capillary electrophoresis
Different capillary electrophoretic separations were car-
ried out in order to optimize the Tf-glycoforms separation.
Table 1
Additives tested in the Transferrin CE analysis
Additive Concentration
Hydroxylamines
Ethanolamine 0.1% (w/v)
Triethanolamine 50 mM
Alkyldiamines
Ethylenediamine 50 mM
1,4-Diaminobutane (DAB) 1 mM
,-bis-quaternary ammonium
Hexamethonium bromide 0.75 mM
Organic solvents
Acetonitrile 30% (v/v)
Methanol 30% (v/v)
2-Propanol 15% (v/v)
1,4-Butanediol 15% (v/v)
Ethyleneglycol 15% (v/v)
The glycoforms were identied by the elution order and
named as S
0
, S
1
, S
2
, S
3
and S
4
according to the number of
contained sialic acid. A pH range of 79.5 was covered test-
ing two types of buffer solutions: borate and Trisglutamate
and different additives to buffer solution were tested. These
modiers are shown in Table 1. Different capillary dimen-
sions (inner diameters of 50 and 75 m, and lengths of 57,
77 and 107 cm) were also tested.
2.5.2. Mass spectrometry
For mass spectrometry experiments, the solvent used con-
sisted of a mixture of water/acetonitrile (50:50 v/v) and the
ES response was optimized by adding different acids, such
as triuoroacetic acid (TFAA) and heptauorobutyric acid
(HFBA), at different concentrations (0.050.25%). In order
to optimize the ES-MS source and analyzer parameters, an
aqueous Tf solution (0.4 g/l) was introduced directly into
the ES source at a ow-rate of 10 l/min. The optimal val-
ues for sample cone voltage and source temperature were:
75 V and 120

C, respectively, when 0.25% TFAA was used

as additive; 110 V and 60

C when 0.1% HFBA was used

as additive.
2.6. Neuraminidase treatment
As the most interesting sialoforms in CDGS studies
present fewer sialic acid residues (S
2
and S
0
) than the main
glycoform (S
4
) present in healthy serum, Tf was incubated
with the neuraminidase enzyme in order to split off sialic
acid residues and to obtain the different sialoforms with
lower sialic acid content.
2.6.1. Capillary electrophoresis
For capillary electrophoresis experiments, a 10-l volume
of enzyme solution (2.1 U/mg in aqueous solution, pH5.5)
was incubated with 1 ml of Tf solution (1 g/l) at room
temperature. The reaction was followed at different times
by direct injections of aliquots, until the peak attributed
to asialo-Tf appeared. The sialoforms of Tf generated by
4 V. Sanz-Nebot et al. / J. Chromatogr. B 798 (2003) 17
neuraminidase were labelled S

3
, S

2
, S

1
and S

0
, respectively,
according to the number of sialic acid residues they con-
tained, in order to distinguish them from the S
3
, S
2
, S
1
and
S
0
glycoforms, which have lost the complete carbohydrate
chain. The S

i
sialoforms differ in n292 Da, n being the
number of sialic acids split off by the neuraminidase.
2.6.2. Mass spectrometry
For mass spectrometry experiments, 1 ml of Tf (4 g/l)
was incubated with 40 l of enzyme. The reaction was fol-
lowed by direct injection of different aliquots to MS taken at
several times. A 15-l sample aliquot was acidied with 5
l of the eluent described in Section 2.5.2 prior to injection.
2.7. Analysis of serum samples
Normal serum samples from healthy individuals and
pathological serum samples from patients with the CDG
Syndrome were supplied by the Institut de Bioqumica
Clnica of the University of Barcelona. Serum samples
were saturated with iron by incubation with 10 mM FeCl
3
and 500 mM NaHCO
3
for 30 min, and diluted 1/101/3 in
water before injection. Normal Tf contents of human serum
ranges from 2 to 3 g/l.
3. Results and discussion
3.1. Capillary electrophoretic separation of Tf-glycoforms
In the present study, Tf isoforms were separated according
to the different mobility of the Tf-sialoforms, which differ
in the number of sialic acid residues and, consequently, in
the number of negative charges. Under basic CZE separation
conditions, the high electroosmotic ow generated pushes
the negatively-charged Tf isoforms to the detector (cathodic
end). Several factors, such as type and pH of running buffer,
type and concentration of buffer additives, length and inner
diameter of the capillary, and other instrumental parameters
(e.g. applied voltage, injection time) have been optimized in
order to achieve a good resolution of the different sialoforms.
The parameters with most inuence, among those indi-
cated above, were the pH of running buffer and the type and
concentration of buffer additives. Despite obtaining similar
results with both assayed buffers, borate and Trisglutamate
(Fig. 1), we chose the borate buffer because of the role of
borate anions in glycoform separations [6,7,10,11,31] and
the fact that borates high UV transparency allows higher
sensitivity detection [6]. Buffer pH ranged from 7 to 9.5 and
a loss of resolution was observed working at extreme pH
values (Fig. 1). Therefore, an optimal value of 8.3giving
the best resolutionwas chosen.
In order to reduce protein adsorption onto the fused-silica
capillary walls, different additives were tested (Table 1).
Hexamethonium bromide and 1,4-diaminobutane (DAB)
Fig. 1. Electropherograms at 15 kV of 1 g/l Tf solution (injection time:
5 s), using Trisglutamate and borate buffers, at different pH values.
gave the best results (Fig. 2). We chose DAB because
the baseline obtained was cleaner and simpler than that
obtained with hexamethonium bromide. Previously, other
workers have also found DAB to be the most suitable addi-
tive in the separation of transferrin glycoforms in uncoated
fused silica capillaries [6,7]. The positively charged amino
groups of DAB interact with the negatively charged free
silanol groups on the capillary wall, and thereby reducing
protein adsorption, decreasing EOF and, therefore, increas-
ing the migration time of the sample. These factors allow
an enhanced resolution of the different glycoforms. After
testing different DAB concentrations, from 0.5 to 4 mM,
in the separation borate buffer, the optimal concentration
of DAB was taken to be 1 mM. The use of higher DAB
concentrations failed to improve resolution and led to an
Fig. 2. Electropherograms at 15 kV of 1 g/l Tf solution (injection
time: 5 s) using different mobile phase additives.
V. Sanz-Nebot et al. / J. Chromatogr. B 798 (2003) 17 5
Fig. 3. Electropherogram of 1 g/l Tf solution obtained with the follow-
ing optimized conditions: 15 s injection time, 15 kV separation voltage,
100 mM borate containing 1 mM DAB at pH 8.3 as separation electrolyte,
25

C temperature, detection at 200 nm, normal polarity.

increase in analysis time. Using these optimal conditions,
a good resolution of Tf sialoforms was achieved in less
than 18 min with a bare capillary of 50 m I.D. and 57 cm
length, as can be seen in Fig. 3.
In order to validate the proposed method, the quality pa-
rameters were calculated [20]. Analytical precision of the
proposed method was studied by calculating run-to-run and
day-to-day reproducibility of migration times and peak ar-
eas of disialo-, trisialo- and tetrasialo-Tf, using a standard Tf
solution of 1 g/l. The results are shown in Table 2. Pre-
cision of migration times and peak areas were satisfactory.
Tf glycoforms with a higher number of sialic acid residues
than tetrasialo-Tf were not considered in the present study,
because of their minor relevance in CDGS.
With the proposed CE method, the limit of detection
(LOD, signal-to-noise ratio of 3) and the limit of quanti-
Table 2
Repeatability and reproducibility of migration times (min) and absolute
peak area of S
2
, S
3
and S
4
with the proposed CE method
Intra-day (n =5) Day-to-day (n =12)
Mean RSD (%) Mean RSD (%)
Migration time (min)
Disialo-Tf 15.5 1.0 15.4 1.6
Trisialo-Tf 15.8 1.1 15.7 1.7
Tetrasialo-Tf 16.4 1.1 16.2 1.7
Absolute peak area
a
Disialo-Tf 6621 4.8 5952 5.2
Trisialo-Tf 98 637 4.3 91 915 5.5
Tetrasialo-Tf 694 156 5.3 654 769 6.3
a
Arbitrary units.
Table 3
Calibration equations and linear ranges of the different Tf-glycoforms
present in commercial transferrin
Glycoform Equation Regression
coefcient,
R
Number of
considered
points
Linear
range
(ng/l)
S
4
y =793x1984 0.999 12 51000
S
3
y =112x55 0.999 6 351000
S
2
y =13x897 0.996 5 2001000
In the calibration equations, x represents the commercial Tf concentration
expressed in ng/l and y represents the peak area.
cation (LOQ, signal-to-noise ratio of 10) of human Tf were
about 4 and 25 ng/l, respectively. The LOD and LOQ
of the trisialo- and disialo-Tf, expressed as percentages
of the tetrasialo-Tf peak area of healthy serum, were 0.5
and 2.1% for S
3
, respectively, and 0.4 and 1.2% for S
2
,
respectively. That means that the total Tf concentrations
required to detect S3 and S2 sialoforms are 35 and 200
ng/l, respectively. For quantication purposes, the total Tf
concentrations required are 125 ng/l for S
3
and 550 ng/l
for S
2
. According to the percentage concentrations reported
by Martensson et al. [32] of about 2 and 6% for S
2
and S
3
,
respectively in healthy serum, the sensitivity achieved with
the present method is perfectly adequate for the determina-
tion of disialo- and trisialo-Tf in serum samples.
Quantitation was performed by external calibration, plot-
ting peak area versus concentration of commercial Tf in-
jected into the electrophoretic system. Calibration equations
and linear ranges of the different sialoforms are shown in
Table 3 (correlation coefcients >0.999).
The commercial Tf was treated with neuraminidase en-
zyme for increasing times. Neuraminidase only split off the
sialic acid residues from Tf, not the carbohydrate chain
[4,6,19]. However, the generated sialoforms, known as S

i
,
are frequently used as models in separation studies of CDGS
Tf-glycoforms because they present the same electrophoretic
behaviour based on the different negative charge. The elec-
tropherograms obtained by injection of different reaction
aliquots show a good resolution of the Tf sialoforms gen-
erated (Fig. 4). Therefore, the proposed method should also
be applicable to the separation of CDGS glycoforms.
Normal and pathological serum samples were saturated
with iron, diluted and injected into the CE system using the
proposed method. Fig. 5 shows comparisons between injec-
tion of unextracted sera from a control subject and from a
subject with CDG Syndrome. The Tf isoforms can be iden-
tied on the basis of the correspondence of migration time
with that of standard Tf (Fig. 3). The CE elution prole of
normal serum shows high concentrations of tetrasialotrans-
ferrin and low amounts of disialo-, trisialo-, and pentasialo-
transferrin. In the CE elution prole of CDGS serum(Fig. 5),
transferrin glycoforms with di- and asialylated carbohydrate
chains are increased. This prole corresponds to CDGS type
I serum, which typically shows dramatic reductions in S
4
6 V. Sanz-Nebot et al. / J. Chromatogr. B 798 (2003) 17
Fig. 4. Electropherograms obtained by injection of different aliquots of
solution obtained after digestion with neuraminidase at different reaction
times. Experimental conditions are as in Fig. 3.
and dramatic increases in S
0
and S
2
sialylated forms [19,33].
In fact, the increased appearance of less sialylated transfer-
rins, especially of disialotransferrin, is the specic marker
for CDG Syndrome [34]. These results suggest that the pro-
posed CE method can be used to resolve Tf glycoforms in
serum, suggesting a rapid diagnostic test for the Carbohy-
drate Decient Glycoprotein Syndromes group of diseases.
3.2. Characterization of Tf-glycoforms by ES-MS
Fig. 6 shows the electrospray mass spectra of a standard
solution of Tf (4 g/l) obtained at the optimal mass spec-
trometer conditions. Applying the MaxEnt algorithm to the
Fig. 6. Electrospray mass spectra of Tf (4 g/l) obtained with a mobile phase of acetonitrilewater (50:50, v/v) containing: (a) 0.25% TFAA and (b)
0.10% HFBA.
Fig. 5. Electropherograms obtained by injection of diluted serum sam-
ples from healthy individuals and from CDGS patients. Albumin and im-
munoglobulins are also labeled. Experimental conditions are as in Fig. 3.
spectral data, gave a molecular masses for the S
4
glyco-
form of 79 529 and 79 525 with the solvent composed of
water/acetonitrile (50:50 v/v) with 0.10% HFBA and 0.25%
TFAA, respectively. These molecular mass results are in
agreement with the theoretical value for the Tf-glycoform
S
4
, the main sialoform in serum, with a molecular mass of
79 570 Da [35].
In order to characterize the sialoforms generated by neu-
raminidase treatment, aliquots obtained at different reaction
times were directly injected into the electrospray ionization
mass spectrometry system. The cluster of peaks obtained
in the spectra was different at every injection, showing that
transformation of Tf was taking place, new Tf-sialoforms
V. Sanz-Nebot et al. / J. Chromatogr. B 798 (2003) 17 7
Table 4
Molecular masses obtained by MaxEnt algorithm for the different sialo-
forms of Tf generated by addition of neuraminidase using both solvents
considered: acetonitrilewater (50:50, v/v) with 0.25% TFAA (solvent A)
and acetonitrilewater (50:50, v/v) with 0.10% HFBA (solvent B)
Glycoform Molecular masses (Da)
Solvent A Solvent B
S
4
79 525 13 79 529 13
S

3
79 232 13 79 237 23
S

2
78 964 15 78 945 10
S

1
78 659 18 78 661 14
S

0
78 367 10 79 376 11
The standard deviation is also indicated.
being generated. The molecular masses of the different sialo-
forms generated were obtained by applying the MaxEnt al-
gorithm and are summarized in Table 4. The results obtained
are consistent with the molecular masses expected after con-
secutive loss of different sialic acid residues. Both solvents
considered prove to be useful for characterizing the differ-
ent Tf sialoforms generated, as there are no signicant dif-
ferences between the results obtained with either of them
(Table 4).
These results are simply a rst step in our ultimate goal:
the development of a CEMS method for the analysis of Tf
sialoforms. For this purpose, the modication of the CE con-
ditions to increase the compatibility with the MS detection
and the improvement of the MS sensitivity in order to obtain
suitable spectra at lower concentrations will be necessary.
Even at this early stage, the results obtained here demon-
strate the potential of mass spectrometry with electrospray as
an ionization source for the characterization of glycoforms
with small mass differences (292 Da) and, consequently,
the proposed method should also be applicable to glyco-
forms present in CDGS patients serum samples, which dif-
fer in about 2200 Da, because of the loss of a biantennary
sugar chain.
4. Conclusions
In this study, a CE method has been proposed for sepa-
rating human serum commercial Tf-glycoforms, a good res-
olution being achieved for the most relevant sialoforms of
CDGS, those containing fewer than four sialic acid residues.
The separation is carried out in less than 18 min without the
need for complex sample preparation. The precision, sensi-
tivity and linearity of this method enable a satisfactory quan-
titation of samples. The proposed method was applied to
control and CDGS sera samples and different proles were
obtained which can be used for diagnosis purposes. More-
over, characterization of all Tf-glycoforms, using a neu-
raminidase treatment, was achieved through ES-MS. The
CE method developed enables a quick and simple CDGS
diagnosis to be made.
Acknowledgements
Financial support of this project by DGICYT (Project
PB98-1174) is gratefully acknowledged. P. Gonzlez also
thanks the University of Barcelona for a F.I. grant.
References
[1] S. Kishino, K. Miyazaki, J. Chromatogr. B 699 (1997) 371.
[2] G. Keir, B.G. Winchester, P. Clayton, Ann. Clin. Biochem. 36 (1999)
20.
[3] R.P. Oda, J.P. Landers, Electrophoresis 17 (1996) 431.
[4] R.P. Oda, R. Prasad, R.L. Stout, D. Cofn, W.P. Patton, D.L. Kraft,
J.F. OBrien, J.P. Landers, Electrophoresis 18 (1997) 1819.
[5] K. Yamashita, T. Ohkura, H. Ideo, K. Ohno, M. Kanai, J. Biochem.
114 (1993) 766.
[6] F. Crivellente, G. Fracasso, R. Valentini, G. Manetto, A.P. Riviera,
F. Tagliaro, J. Chromatogr. B 739 (2000) 81.
[7] B.C. Giordano, M. Muza, A. Trout, J.P. Landers, J. Chromatogr. A
742 (2000) 79.
[8] V. Sanz-Nebot, F. Benavente, I. Toro, J. Barbosa, Electrophoresis 22
(2001) 4333.
[9] R.P. Oda, B.J. Madden, T.C. Spelsberg, J.P. Landers, J. Chromatogr.
A 680 (1994) 85.
[10] M.E. Legaz, M.M. Pedrosa, J. Chromatogr. A 719 (1996) 159.
[11] J.P. Landers, R.P. Oda, B.J. Madden, T.C. Spelsberg, Anal. Biochem.
205 (1992) 115.
[12] A. Pantazaki, M. Taverna, C. Vidal-Madjar, Anal. Chim. Acta 383
(1999) 137.
[13] M. Taverna, N.T. Tran, T. Merry, E. Horvarth, D. Ferrier, Elec-
trophoresis 19 (1998) 2572.
[14] K. Kakehi, S. Honda, J. Chromatogr. A 720 (1996) 377.
[15] E. Watson, F. Yao, Anal. Biochem. 210 (1993) 389.
[16] A.D. Tran, S. Park, P.J. Lisi, O.T. Huynh, R.R. Ryall, P.A. Lane, J.
Chromatogr. 542 (1991) 459.
[17] S.L. Wu, Anal. Biochem. 253 (1997) 85.
[18] R. Prasad, R.L. Stout, D. Cofn, J. Smith, Electrophoresis 18 (1997)
1814.
[19] O. Iourin, T.S. Mattu, N. Mian, G. Keir, B. Winchester, R.A. Dwek,
P.M. Rudd, Glycoconjug. J. 13 (1996) 1031.
[20] I. Toro, V. Sanz-Nebot, J. Barbosa, J. Chromatogr. A 893 (2000) 95.
[21] V. Sanz-Nebot, I. Toro, J. Barbosa, J. Chromatogr. A 846 (1999) 25.
[22] V. Sanz-Nebot, F. Benavente, A. Castillo, J. Barbosa, J. Chromatogr.
A 889 (2000) 119.
[23] R.S. Rush, P.L. Derby, D.M. Smith, C. Merry, G. Rogers, M.F.
Rohde, V. Katta, Anal. Chem. 67 (1995) 1442.
[24] V. Sanz-Nebot, I. Toro, R. Bergs, J. Segura, J. Barbosa, J. Mass
Spectrom. 36 (2001) 652.
[25] J.F. Kelly, S.J. Locke, L. Ramaley, P. Thibault, J. Chromatogr. A
720 (1996) 409.
[26] V. Sanz-Nebot, I. Toro, A. Garcs, J. Barbosa, Rapid Commun. Mass
Spectrom. 13 (1999) 2341.
[27] C.G. Huber, A. Premstaller, G. Kleindienst, J. Chromatogr. A 849
(1999) 175.
[28] E. Gelp, J. Chromatogr. A 703 (1995) 59.
[29] J.P. Landers, Handbook of Capillary Electrophoresis, 2nd ed., CRC
Press, Boca Raton, FL, 1997.
[30] D.R. Baker, Capillary Electrophoresis, Wiley, New York, 1995.
[31] J. Liu, O. Shiruta, M. Novotny, Anal. Chem. 63 (1991) 413.
[32] O. Martensson, A. Harlin, R. Brandt, K. Seppa, P. Sillanaukee,
Alcohol. Clin. Exp. Res. 21 (1997) 1710.
[33] H. Carchon, E. Van Schaftingen, G. Matthijs, J. Jaeken, Biochim.
Biophys. Acta 1455 (1999) 155.
[34] Y. Wada, A. Nishikawa, N. Okamoto, K. Inui, H. Tsukamoto, S.
Okada, N. Taniguchi, Biochem. Biophys. Res. Commun. 189 (1992)
832.