Journal of Chromatography A, 1214 (2008) 157–164

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Simultaneous analysis by capillary electrophoresis of five amyloid peptides as

potential biomarkers of Alzheimer’s disease
Romain Verpillot a , Markus Otto b , Hans Klafki c , Myriam Taverna a,∗
a
Univ Paris-Sud, JE 2495, Protéines et Nanotechnologies en Sciences Séparatives, Faculté de Pharmacie, 92296 Châtenay-Malabry, France
b
University of Ulm, Department of Neurology, Steinhövelstrasse 1, 89075 Ulm, Germany
c
University of Duisburg-Essen, Department of Psychiatry and Psychotherapy, Virchowstrasse 174, D-45147 Essen, Germany

a r t i c l e i n f o a b s t r a c t

Article history: We report here a CE method for the separation and quantitation of five amyloid peptides (A␤1-42, 1-40,
Received 25 July 2008 1-39, 1-38, and 1-37) considered as potential biomarkers of Alzheimer’s disease. These amyloid peptides
Received in revised form 9 October 2008 have very similar structures. Sample preparation and storage conditions are critical parameters to ensure
Accepted 14 October 2008
their solubility and to avoid the aggregation process in particular for A␤1-42. Their solubility was found
Available online 18 October 2008
fully dependent on the NH4 OH concentration that was employed initially to dissolve the lyophilized
amyloid peptides. Conditions to achieve a full separation of these peptides were found using a dynamic
Keywords:
coating with 1,4-diaminobutane (DAB). The linear decrease of their electrophoretic mobility highlighted
Amyloid peptides
Capillary electrophoresis
an ion-pairing phenomenon between the peptides and DAB. The optimal background electrolyte was a
Dynamic coating 40 mM borate buffer, pH 9 containing 3 mM of DAB. Under these conditions, resolutions ranged from 1.3 to
Alzheimer’s disease 2.4 with theoretical plates reaching 300,000. Under the retained conditions, we showed that adsorption of
CSF peptides to silica was negligible (recovery over 94.5%) and depletion effect of the background electrolyte
was overcome. The method was finally validated in terms of linearity and repeatability and the limits of
detection for the five A␤ peptides were estimated. The inter-day repeatability of the migration times was
very satisfactory with RSDs less than 1.55%. The RSDs of the peak areas were below 5%. With this CE–UV
method, limits of detection of the peptides ranged from 300 to 500 nM. We finally demonstrated that this
method can be applied to real biological samples such as CSF.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction In the brains of AD patients, aggregated A␤ peptides form the

Alzheimer’s disease (AD) affects approximately 24 millions peo-
ple today and the prevalence in the group of individuals older than
65 years doubles every 5.1 years [1–3]. The preclinical period prob-
ably starts as early 20–30 years before the first clinical symptoms
appear [4].
␤-Amyloid peptides (A␤ peptides) are generated from ␤-
amyloid precursor protein (APP) by two sequential proteolytic
cleavages by the so-called ␤- and ␥-secretases.
A␤ peptides are produced during normal cellular metabolism
[5,6], and they are normal soluble constituents of biological fluids,
such as cerebrospinal fluid (CSF), blood and urine [7–10]. Matrix-
assisted laser desorption ionization time-of-life mass spectrometry
(MALDI-TOF-MS) performed on immunoprecipitated A␤ peptides
revealed the existence of several A␤ peptides of different lengths
in human CSF and blood samples [11–13].
∗ Corresponding author. Fax: +33 146835944.
E-mail address: myriam.taverna@u-psud.fr (M. Taverna).
major proteinaceous constituent of the amyloid fibrils deposited
in senile plaques and in cerebrovascular amyloid [6,14,15]. In
human CSF, five different A␤ peptides, A␤1-37, 1-38, 1-39, 1-
40 and 1-42 are regularly present, and their relative abundances
were reported to be significantly correlated with each other
[16]. A selective reduction of A␤ peptides ending at residue
42 in CSF from AD patients was reported in several studies,
while total A␤ levels were found to be in the normal range
[17].
The analysis of CSF for multiple biomarkers and, in particular
the combined and simultaneous quantitation of the amyloid quin-
tet (A␤1-42, 1-40, 1-39, 1-38 and 1-37), represents a promising
approach in search for improved early and differential dementia
diagnosis.
Capillary zone electrophoresis (CZE) is a powerful method that
has already proven to be efficient and fully quantitative for pro-
teins and peptide analysis [18–22]. Varesio et al. reported, in 2002,
a CE–MS approach to characterize A␤1-40 but the method was
not sensitive enough to reach the usual A␤1-40 concentration in
plasma or serum [23].

0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2008.10.051
158 R. Verpillot et al. / J. Chromatogr. A 1214 (2008) 157–164
Table 1
Primary structure of the A␤ amyloid quintet. The primary sequences are the same
except the N or C terminal truncated parts.
A␤1-42: H2 N-DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIA-OH
A␤1-40: H2 N-DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVV-OH
A␤1-39: H2 N-DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGV-OH
A␤1-38: H2 N-DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGG-OH
A␤1-37: H2 N-DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVG-OH
In two recent papers, CE was used for monitoring the aggrega-
tion of A␤1-40. Sabella et al. [24] have proposed, for the first time,
one CZE–UV for the monitoring of the kinetic of A␤1-42 and A␤1-40
aggregation. Very recently Kato et al. [25] developed a CE-laser-
induced fluorescence (LIF) detection method to analyse aggregated
forms of A␤1-42. This method was based on the use of thioflavine
which specifically labels ␤-sheet structures.
To our best knowledge, CZE has never been exploited to analyse
simultaneously the main soluble A␤-peptide variants A␤1-37, 1-38,
1-39, 1-40 and 1-42 that are present in human CSF and blood. The
separation of these A␤ peptides is quite challenging as all of those
starting at Asp1 exhibit the same charge [theoretical isoelectric
point (pI) is 5.21] and differ only by their carboxy termini (Table 1).
Therefore, the electrophoretic mobility difference between these
species depends exclusively on their hydrodynamic volume which
is very similar for all five peptides but also dependent on the sur-
rounding medium. In addition, several of these peptides are prone
to aggregation and a careful selection of both the sample and back-
ground electrolyte (BGE) media is a key requirement to achieve
their separation as soluble and monomeric species.
To meet these requirements, we propose in this work a novel CZE
method to achieve a full quantitative separation of the five main
A␤ peptides that are found in human CSF. For modulating both,
the electroosmotic flow (EOF) amplitude and the electrophoretic
behavior of amyloid peptides, an alkylamine was used. Alky-
lamines belong to one class of dynamic coatings for fused-silica
capillaries which are used to reduce EOF and protein adsorp-
tion [18–22,26–28]. At alkaline pH values, they strongly interact
with the silanol groups of the capillary wall and can thereby
improve protein and peptide separations. To our best knowledge,
1,4-diaminobutane (DAB) has not been yet exploited for the sepa-
ration of acidic peptides under alkaline conditions.
In the present work, a full separation of the five A␤ peptides by
CE is presented. Particular attention was paid to the solubility of the
peptides under the experimental conditions and to possible adsorp-
tion of the peptides to the dynamically modified capillary surface,
both of which are key factors for achieving maximum repeatability.
Finally the method was validated and applied to a real sample of
CSF.
2. Materials and methods
2.1. Chemicals, reagents and samples
Amyloid peptides were purchased from Anaspec (Le Perrey en
Yvelines, France), Sigma (St. Louis, MO, USA) and Rpeptide (Bogart,
GA, USA). The CSF sample analysed in this study was obtained from
a patient attending the outpatient memory clinic in 2007. Collec-
tion and analysis of the CSF sample were approved by the Ethics
Committee in Ulm. CSF was taken, aliquoted within 2 h and stored
at −80 ◦ C.
Sodium hydroxide (1 M) was obtained from VWR (Fontenay-
sous Bois, France). Boric acid (H3 BO3 ) ammonium hydroxide 28.1%
(m/v), thiourea and DAB were obtained from Sigma. All buffers were
prepared with deionised water and were filtered through a 0.22 ␮m
membrane before use.
2.2. Apparatus and material
For the very preliminary experiments, a P/ACE 2100 instrument
(Beckman Instruments, Fullerton, CA, USA) was employed with
ultraviolet (UV) detection ( = 214 nm). Uncoated capillaries were
from Phymep (Paris, France). Data acquisition and instrument con-
trol were carried out using P/ACE station 1.21 software. Then, all
the studies have been performed using a Beckman Coulter PA 800
ProteomLab equipped with a UV and photodiode array detection
(DAD). Data acquisition and instrument control were carried out
using Karat 7.0 software. Water used in all experiments was puri-
fied using a Direct-Q 3 UV purification system (Millipore, Milford,
MA, USA) with resistivity higher than 18 M/cm. The pH values of
buffer solutions were adjusted with InoLab WTW series pH 730
pH meter. Buffer ionic strength (I) calculations were performed
using the Phoebus program (Analis, Suarlée, Belgium) using correc-
tion of activity coefficients according to the Gütelberg and Davies
equation.
2.3. Methods
2.3.1. Dissolution and storage of the peptide samples
Dissolution medium and storage conditions are critical param-
eters for amyloid peptides which are prone to aggregation or
fibrillogenesis. Composition of the sample medium, pH and tem-
perature of storage, need to be carefully controlled to avoid
these phenomena. A␤1-42 was dissolved in 0.16% (m/v) ammo-
nium hydroxide aqueous solution. All other amyloid peptides were
dissolved in 0.10% (m/v) ammonium hydroxide. All the peptide
samples, which were received in lyophilized form, were individu-
ally dissolved in aqueous solution of ammonium hydroxide to reach
a concentration of 2 mg/ml. The solutions were then divided into
several aliquots and individually stored at −20 ◦ C. Each aliquot was
thawed only once immediately prior to analyses. When preparing
mixtures of these peptide solutions, care was taken to keep an over-
all ammonium hydroxide concentration above 0.004% (m/v) after
dilution.
2.3.2. Capillary electrophoresis conditions
For the very preliminary studies performed on a P/ACE 2100
Beckman, the capillary had an I.D. of 50 ␮m, 57 and 50 cm total
and effective lengths, respectively. When employing the PA 800
instrument, the uncoated capillary had an I.D. of 50 ␮m I.D. and
57 cm total length (47 cm effective length). During all the optimi-
sation process, new capillaries were pretreated applying pressures
of 138 kPa to the capillary inlet and using the following sequence:
0.1 M NaOH for 5 min, 1 M NaOH for 5 min and then deionised
water for 5 min. The in-between-runs rinsing cycle was carried out
by pumping through the capillary deionised water, 0.1 M NaOH,
deionised water, and running buffer for 2 min each. These con-
ditions (pretreatment and in-between-runs rinsing procedure)
were improved during the course of the optimisation to improve
reproducibility. Finally, the following conditions were adopted.
After pretreatment with successively 0.1 M NaOH, 1 M NaOH, and
deionised water for 5 min each, five successive cycles were carried
out to stabilise the silica surface and the EOF before the first analy-
sis: one cycle entailed one rinsing with the running buffer for 10 min
times, then voltage application for 15 min at 30 kV, followed by one
rinse with deionised water for 10 min and 0.1 M NaOH for 5 min.
The in-between-runs rinsing was carried out by pumping through
the capillary deionised water for 10 min, 0.1 M NaOH for 5 min, and
finally the running buffer for 10 min. The samples were introduced
into the capillary by hydrodynamic injection under 3.4 kPa. The
capillary was thermostated at 25 ◦ C and the samples were main-
tained at 6 ◦ C by the storage sample module of the PA 800. The
 R. Verpillot et al. / J. Chromatogr. A 1214 (2008) 157–164 159

peptides were detected by UV detection at 214 nm with PACE 2100, with a centrifugal evaporator Jouan RC 10-22 (St. Herblain, France),
and either using UV detection (214 nm) or DAD (190 nm) with PA under vacuum, at 241 × g and at room temperature.
800. For the BGE, stock solutions of borate buffer were prepared
using boric acid (H3 BO3 ) at different ionic strengths from 40 to
3. Results and discussion
150 mM. Sodium hydroxide solution was added in respective pro-
portion according to Phoebus software to yield buffers from pH 7
3.1. Optimisation of the experimental conditions
until 10.5. 100 mM stock solutions of DAB were prepared freshly
by dissolving DAB in deionised water. The running buffers with the
3.1.1. pH and ionic strength of the BGE
desired concentrations of DAB, pH, and I values were prepared daily.
To optimize the pH of the BGE a borate buffer was employed.
The separations were carried out at 30 kV with positive polarity at
This buffer was previously reported to be suitable for the analysis
the inlet using borate buffer pH 9 at I 40 mM and containing 3 mM
of A␤1-42 [24] and, it exhibits both low conductivity and good UV
of DAB. The running electrolyte was renewed after every run.
transparency. We have estimated the resolution between A␤1-42
and A␤1-40 in this alkaline buffer as a function of the pH (from 7
2.3.3. Validation of the method
to 10.5) with a constant borate concentration of 40 mM.
Validation of the method was performed on a PA 800. Standard
As shown in Fig. 1A, the resolution increased as the BGE pH was
solutions containing the five peptides at equal concentrations were
raised from 7 to 11. The best separation was achieved at pH 10–11
employed to evaluate the linearity of the method. The concentra-
when the peptides carry the highest negative charge. In addition,
tions tested ranged from 0.133 to 8.31 × 10−3 mg/ml. The analyses
we observed more reproducible peak areas under these conditions
were carried out in triplicate for each concentration level and the
that we ascribed to a better solubility of the amyloid peptides.
calibration curves were calculated by means of the least-square
We have defined pH 10.5 as the optimum and fixed this value to
method. The limit of detection (LOD) was estimated as the peptide
study the influence of I on the resolution. Concentrations ranging
concentration giving a peak height equal to three times the baseline
from 35 to 150 mM of borate were tested. Fig. 1B shows a positive
noise. To reach the highest LOD, the injection time was increased to
effect of I on the resolution. This observation was attributed to both
20 s with a pressure at 3.4 kPa. Analyses under optimized conditions
a staking effect and to closer absolute values of the EOF and peptide
of a mixture of the five peptides at a concentration 0.033 mg/ml
electrophoretic mobilities (ep ) in the counter electroosmotic flow
were repeated five times within the same day to evaluate the intra-
separation. The best resolution, attained without producing joule
day precision and five times over five consecutive days to obtain the
effect, was achieved at I 100 mM.
inter-day precision of the method, both were expressed as relative
It turned out that these optimum conditions in terms of pH and
standard deviation (RSD).
ionic strength were not synergistic. We kept, however, 100 mM
borate buffer at pH 10.0 for the subsequent experiments. Dur-
2.3.4. Recovery studies
Possible adsorption of the peptides to the capillary wall was esti-
mated by recovery studies according to a method described in detail
by Tran et al. [29] and which was adapted from Yeung and Lucy [30].
The recovery studies were performed on the PA 800 instrument. The
recovery of the different amyloid peptides was determined by com-
paring the peak areas obtained after their injection and separation
from the long (30 cm) and the short (10 cm) portions of the same
capillary. To avoid synergistic aggregation processes between the
peptides, recovery studies were performed on the peptides (A␤1-
42, 1-40, 1-39 and 1-37) analysed individually.
The recovery percentage (X) after analysis on a 20 cm effective
length capillary was then deduced from the following equation:
 A 2
L
X (%) = 100 × (1)
AS
where AL and AS are the peak area of peptides injected at the inlet
and the outlet of the capillary, respectively.

2.3.5. CSF sample pretreatment

A human CSF sample with a total protein content of 143 mg/l
was employed to evaluate the compatibility of the method with
real samples. This concentration was determined by using a neph-
elometric method. The BN prospect nephelomer was from DADE
(Behring, Germany) and a total amount of 100 ␮l was used for this
analysis [31]. The CSF was stored at −80 or −20 ◦ C before use. The
CSF was diluted in deionised water threefold and then analysed
directly.
The desalted and concentrated CSF samples were prepared as
follows: 500 ␮l of CSF was desalted on a Nanosep Centrifugal Device
with a 3 kDa cutoff membrane (Pall Life Science, St. Germain-en-
Laye, France) using an Eppendorf centrifuge 5415D (Sartrouville, Fig. 1. Resolution of amyloid peptides A␤1-42 and A␤1-40 versus BGE pH (A) and
ionic strength (B) using an uncoated capillary. Conditions for (A): running buffer,
France). The sample volume was kept constant during the centrifu-
borate buffer at 40 mM (I); applied voltage, 16 kV and for (B) borate buffer at pH 10.5;
gation by repeated additions of water (3925 × g, five cycles of 10 min applied voltage, 16 kV. See Section 2.3.2 for other conditions. Analyses performed
at room temperature). Then, the sample was concentrated 33-fold with P/ACE 2100.
160 R. Verpillot et al. / J. Chromatogr. A 1214 (2008) 157–164

ing the course of the optimisation, we observed, at certain times,
irreproducible profiles which were presumably related to peptide
aggregation or peptide instability. These factors can have an impact
on the Rs evaluation; therefore we have analysed the stability of the
peptides in solution as a function of time.
3.1.2. Optimisation of the storage conditions
The physicochemical properties including conformation,
hydrophobicity, solubility and tendency to aggregate of the five
A␤ peptides studied here are slightly different. Among these five
peptides, A␤1-42 is the least soluble and the one most prone to
aggregation in aqueous solution. Sabella et al. [24] and Kato et al.
[25] monitored the fibrillogenesis of A␤1-42 and A␤1-40 in vitro by
CE methods. For our purposes, however, it was important to avoid
aggregation during the CE analysis, to ensure good repeatability
of peak areas and migration times. To maintain a good solubility
of A␤1-42, addition of NH4 OH in the samples is recommended
by various suppliers. We observed that the solubility was fully
dependent on the NH4 OH concentration that was employed
initially to dissolve the lyophilized amyloid peptides.
A␤1-42 and A␤1-40 from Sigma were mixed and dissolved in
deionised water to reach a concentration of 0.033 mg/ml of each
peptide and 0.004% of ammonium hydroxide. These aqueous sam-
ples were analysed repeatedly over a period of 48 h with storage at
6 ◦ C between individual runs. Under these conditions, we observed
a dramatic decrease of A␤1-42 and A␤1-40 peak areas, in particular
after 24 h (Fig. 2A). Moreover this decrease of peak area was more
pronounced in the case of A␤1-42 which is presumably due to a
higher rate of aggregation compared to that of A␤1-40. This trend
was accompanied by the appearance of a late migrating peak which
could be attributed, to oligomeric or fibril species according to pre-
vious work [24,25]. As the profile remained relatively stable during
the first 24 h, we deduced that, the samples should be thawed only
once and stored at 6 ◦ C for not more than 24 h.

Fig. 2. CE monitoring of the stability of peptides A␤1-42 and A␤1-40 over 48 h (A) and from different suppliers Sigma, Anaspec and Rpeptide (B). t = 0, t = 24 h and t = 48 h are
the electropherogramms obtained with fresh sample at t = 0, with sample stored for 24 and 48 h under the same conditions. Conditions for (A): borate buffer at 100 mM (I)
and pH 10. Concentration of A␤ 0.033 mg/ml. Conditions for (B): same buffer with the addition of 3 mM of DAB in BGE. Analyses were performed with a PA 800, three times
for each supplier. Other conditions as in Fig. 2.
 R. Verpillot et al. / J. Chromatogr. A 1214 (2008) 157–164
We have also compared the CE profiles obtained from mixtures
of A␤1-40 and 1-42 obtained from different suppliers. A␤ peptides
purchased from Rpeptide were generated by recombinant strategy
while those from Anaspec and Sigma were produced by chemical
peptide synthesis. All the peptides were provided with the same
counter ion (TFA). Although identical concentrations (according to
the manufacturer information) and the same dissolution protocol
were employed for all of these samples, the peaks areas after CE
separation were quite different (Fig. 2B). The analysis of the Sigma
peptides in triplicates (Fig. 2B) indicated that the observed differ-
ences did not arise from a lack of repeatability of the CE method
but rather from differences in the solubility, purity or stability of
the different peptides in solution. In particular two major features
may be deduced from this figure: first, it appears that, the ratio of
A␤1-42 to A␤1-40 (equimolar concentration) as judged from the
peak areas was more favorable in the case of the Sigma samples
than with those from Rpeptide. A␤1-42 from Rpeptide was proba-
bly less soluble or already partially aggregated under the conditions
employed for the analysis (which includes sample pretreatment
and CE separation). Second, according to the chromatograms shown
in Fig. 2B, it appears that the peptides from Rpeptide and Sigma are
more pure than the one from Anaspec presenting additional minor
peaks. Impurities in the peptide preparations may have a signif-
icant impact on the spontaneous aggregation of peptides as they
might function as seeds for aggregation.
3.1.3. Optimisation of DAB concentration
In the initial experiments, the EOF was too high to achieve a
full separation of A␤1-42 and 1-40 in this counter electroosmosis
mode. In this context, the addition of alkylamine was envisaged to
decrease the EOF amplitude. Moreover, we wanted to evaluate what
was the benefit of adding an alkylamine as a potential ion-pairing
agent for the peptides. Indeed A␤1-42, 1-40, 1-39, 1-38, and 1-37
carry exactly the same charge (same pI) and differ only in their
carboxytermini (Table 1). Therefore the electrophoretic mobility of
the different species depends exclusively on their respective hydro-
dynamic volume which could be modulated through variable ion
pairing.
We added DAB into the BGE (I = 100 mM) at different concentra-
tions and estimated its impact on the resolution of the A␤1-42 and
A␤1-40 peaks. The study was performed at two different BGE pH
values. The best resolution (Rs = 1.56) was obtained with a DAB con-
centration between 2 and 3 mM and with a pH of 10 (Fig. 3). When
the concentration of DAB was increased to 10 mM, the migration
Fig. 3. Resolution of A␤1-42 and A␤1-40 versus DAB concentrations. BGE: borate
buffer at 40 mM (I) at pH 10 () or pH 9 (). Other conditions as in Fig. 1.
161
Fig. 4. DAB concentration versus electrophoretic mobility: identification of ion pair.
The decrease of ion pair is the same for A␤1-42 () and A␤1-40 (). BGE: pH 10,
40 mM. Other conditions as in Fig. 2.
time was substantially longer and the resolution at pH 10 became
quite poor. We speculate that the pairing between DAB and the
peptides at high DAB concentrations may promote the adsorption
of the peptides by the formation of bridges between peptides and
silica via the di-cationic DAB. Finally, with 2 mM of DAB in the BGE,
the EOF was decreased by 2.4-fold.
The ability of DAB to bind the peptides by ion-pairing effect was
further investigated in order to eventually fine-tune their ep . This
kind of ion pairing is favored at low BGE ionic strength and alkaline
pH. The minimum ionic strength was fixed at 40 mM for this study,
in order to preserve some stacking effect.
When calculating the electrophoretic mobilities of the two pep-
tides, we observed a slight linear decrease of their ep (Fig. 4) with
increased DAB concentrations confirming the ion pairing between
the negatively charged peptides and DAB. As the slopes were simi-
lar for both peptides, we concluded that this ion pairing could not
be helpful for improving the separation and to further discriminate
these peptides on the basis of ep . On the basis of the findings pre-
sented so far, we have further optimized the separation conditions
in the presence of DAB while keeping a significant EOF to maintain
an efficient separation. In particular I, pH and DAB concentrations
are interrelated parameters which were found to be critical for opti-
mum resolution of A␤ peptides. We have therefore re-optimized
the BGE conditions focusing, in particular, on these three critical
parameters and exploiting the understanding gained from the pre-
liminary study. As displayed in Fig. 5, the simultaneous and fine
Fig. 5. CE separation of the quintet of amyloid peptides using dynamic coating.
Running buffer: borate buffer at 40 mM (I), pH 9, 3 mM of DAB at 30 kV. See Section
2.3.2 for other conditions.
162 R. Verpillot et al. / J. Chromatogr. A 1214 (2008) 157–164

Table 2
Recovery studies and reproducibility of migration times and peak areas under normal CE and large reservoir conditions (n = 5).

Peptides Peak area RSD (%) Migration time RSD (%) Recovery (%)

Normal CE Large volume buffer Normal CE Large volume buffer Normal CE Large volume buffer

A␤1-42 10.6 1.55 0.6 0.3 41 94.5

A␤1-40 4.8 2.2 0.7 0.5 62 95
A␤1-39 1.6 – 0.5 – 88 –
A␤1-37 0.4 – 1.1 – 98 –

adjustment of both EOF and ep with these three factors enabled
the full separation of A␤1-42, 1-40, 1-39, 1-38, and 1-37. The best
resolution was observed at pH 9, I 40 mM, and 3 mM of DAB with
values ranging from 2.4 to 1.3. Under these conditions and accord-
ing to the peptides, the number of theoretical plates ranged from
291,000 to 302,000.

3.2. Recovery and repeatability studies

Possible adsorption of the peptides to the capillary wall was esti-

mated by recovery studies according to a procedure described in
a previous work [29] and summarized in Section 2.3.4. To avoid
the interaction between different A␤ peptides during the analy-
sis process, each of the five peptides was analysed individually. The
recovery for each peptide was determined from five consecutive CE
runs and values are reported in Table 2. For the most hydrophobic
peptides A␤1-42 and 1-40, the recoveries were poor (41 and 62%
for A␤1-42 and 1-40, respectively, while it was almost 100% for
A␤1-37). In view of these findings, it was obvious that the method
had to be further optimized to achieve an accurate and quantitative
method.
To better understand why the recoveries for the more hydropho-
bic A␤ peptides were low, we have systematically measured current
intensities and migration times during successive CE separation
performed on the same capillary. As shown in Fig. 6 and for con-
ditions I (successive analyses under the same BGE) we observed a
decrease of migration times and current intensities after just sev-
eral consecutive runs. The EOF was the major cause of the migration
Fig. 6. Evolution of Tm of A␤1-40 (A) and current intensities (B) from successive runs
time shift. We hypothesized that the instability of electroosmotic
performed under different conditions of buffer replenish. Capillary length: 30 cm
flow from run to run could be related either to a DAB depletion total length (20 cm effective length). Conditions I, I and I : one replenish of the BGE
from the electrolyte upon voltage application or to the adsorption vial at the 1st analysis then the same BGE is kept for five consecutive analyses; con-
of the peptide to the capillary walls. To test for these hypotheses, ditions II: replenish of the BGE vial between each analysis; III large volume reservoir
we monitored the migration times during 25 subsequent analyses used to perform five successive analyses. Other conditions as in Fig. 5.

without BGE replenishment (conditions I), with BGE replenishment

after five consecutive runs (conditions I and I ), with replenish-
ment after each run (conditions II) and with the use of larger BGE
reservoir (without replenishment, conditions III). As shown in Fig. 7,
the replenishment of BGE before each analysis limited at least par-
tially the shift in migration time in subsequent runs. Finally, the
use of large BGE reservoir (25 ml) leads to a further stabilization of
both, the EOF and the current intensity in subsequent runs. At the
same time, a dramatic improvement of the recoveries of A␤1-42
and A␤1-40 (from 41 to 94.5% and from 62 to 95% for 1-42 and 1-
40, respectively) was achieved under these optimized experimental
conditions (BGE replenishment after each run or use of a large reser-
voir). In addition shift in migration time was reduced (intra-day RSD
for migration times decreased from 0.6 to 0.3%) and the peak areas
were stabilised (the intra-day RSD for the peak areas dropped from
10.6 to 1.5 for A␤1-40) (Table 2). These results clearly indicate that
depletion of DAB from the BGE was responsible for the instability of
the EOF and the resulting shifts in migration time that we observed
initially. Importantly, these features could be prevented by using a
Fig. 7. Electrophoregramm of a CSF sample from a patient suffering from AD (a). The
25 ml reservoir. Peptide adsorption to the capillary walls could not same CSF spiked with 3 ␮M of the quintet amyloid peptides (b). Electropherogramm
be incriminated, at this stage, as the EOF instability was observed of the desalted and concentrated CSF by 33-fold (d) and spiked with 2.5 ␮M of A␤1-
even when a neutral (non-proteinaceous compound) was analysed 40 (c); profile of the CSF spiked with A␤1-40 confirmed the assignment of the peaks.
 R. Verpillot et al. / J. Chromatogr. A 1214 (2008) 157–164 163

Table 3
Comparison of different between-run rinsing and capillary preconditionning procedures on the EOF intensity, RSD (n = 5) of EO and amplitude, and on the RSD of migration
times. All the preconditioning cycles follow a classical pretreatment: 5 min 0.1 M NaOH; 5 min 1 M NaOH; 5 min water.

Procedure Preconditioning cycles Electroosmotic flow Migration time RSD (%)

−9 2 −1 −1
Amplitude (× 10 m V s ) RSD (%) A␤1-42 A␤1-37

A Storage of the capillary under water over the night 0.53 1.77 4.26 4.64
B Storage of the capillary under 0.1 M NaOH over the night 0.54 1.29 2.56 2.74
C 1× (10 min BGE, 15 min BGE at 30 kV, 10 min water, 10 min 0.1 M 0.49 0.9 1.59 1.68
NaOH) and storage of the capillary under 0.1 M NaOH over the night
D 5× (10 min BGE, 15 min BGE at 30 kV, 10 min water, 10 min 0.1 M 0.49 1.34 3.73 3.82
NaOH) and storage of the capillary under water over the night
E 5× (10 min BGE, 15 min BGE at 30 kV, 10 min water, 10 min 0.1 M 0.50 0.22 0.31 0.32
NaOH) and storage of the capillary under 0.1 M NaOH over the night

(results not shown). However a possible alteration of the dynamic
coating during the depletion effect probably occurred. This evo-
lution of the surface state (after only a few successive analyses)
generated an irreversible trend, maintained upon replenishment
and abolished only with the use of large buffer reservoirs.
As a next step in the development of a reproducible CE method,
we examined in-depth, the impact of preconditioning on the homo-
geneity and stability of the capillary wall. For this purpose, a fresh
capillary was used. The different pretreatments and experimental
conditions that were tested are indicated in Table 3. Relatively sta-
ble migration times and EOF (as indicated by the respective RSDs)
were achieved with a preconditioning protocol including a cap-
illary equilibration under successive voltage applications. Under
the optimized conditions, the intra-day migration time RSD was
substantially improved from 5.21 to 0.31%.
3.3. Validation of the method
The method was finally validated in terms of linearity and
repeatability and finally, the limits of detection for the five A␤ pep-
tides were estimated. The assays were repeated five times within 1
day and once per day during 5 days to estimate the intra-day and
inter-day repeatabilities. For all of the five peptides investigated,
the RSDs of migration times of five runs performed within 1 day
were less than 0.32% and those of peak area under 4.22% (intra-day
precision).
With regard to the inter-day repeatability, the variance was
slightly higher for the migration times but still smaller than 1.55%.
The RSDs of the peak areas were below 3.6% for all the peptides
except for A␤1-42 (the RSD value reached 5.0%). As discussed
above, A␤1-42 is particularly sensitive to storage conditions. Lin-
earity detection was assessed by triplicate analyses of peptide
solutions containing A␤ concentrations ranging from 0.133 to
8.31 × 10−3 mg/ml. Calibration curves for the five peptides were cal-
culated by the least-square method. The determination coefficients
ranged from 0.998 to 0.996 showing a very satisfactory linearity of
response compatible with a fully quantitative method. LOD values,
evaluated as the signal to noise ratio equal to 3 and for injection
time of 20 s, ranged from 300 to 500 nM depending on the peptide
under investigation.
each peptide) and analysed under identical conditions (Fig. 7b). The
sensitivity of the UV detection (see above) is not sufficient for mea-
suring the reported concentration of total A␤ peptides in human
CSF, which is approximately 12–13 ng/ml [16]. Therefore, it was
not surprising that direct CE analysis of the diluted CSF sample did
not show a clear signal for any of the five A␤ peptides of interest
(Fig. 7a). CE analysis of a sample of the diluted CSF spiked with 3 ␮M
of each one of the five A␤ peptides resulted in the expected pattern
of five A␤ peaks. It therefore appears that there were no compo-
nents in the diluted CSF sample that substantially interfered with
the A␤ separation and detection. Fig. 7c and d shows CE separations
of desalted and 30-fold concentrated human CSF with (Fig. 7c) and
without (Fig. 7d) the addition of synthetic A␤1-40 at a concentra-
tion of 2.5 ␮M. The added synthetic A␤1-40 peptide was detected as
a clear peak after ∼11.1 min migration time (Fig. 7c). A smaller peak
with a similar migration time was observed in the concentrated
CSF sample without spiking (Fig. 7d), indicating that this method
may be applicable to separate and monitor soluble A␤ peptides in
human CSF samples, provided that the detection sensitivity can be
improved.
4. Conclusion
In this work, a fully quantitative and simultaneous analysis of
the five soluble A␤ peptides regularly found in human CSF is pre-
sented. The novel method is based on a CZE method combined
with dynamic coating. For achieving good resolution, satisfactory
repeatability and high recoveries, we optimized BGE conditions,
preconditioning of the capillary and in-between-run rinsings. Fur-
thermore, measures were taken to minimize the BGE depletion
effect and to modulate the EOF. The ion pairing between peptides
and DAB as well as the interaction of DAB with the capillary surface
were found to play a crucial role in the separation performance. We
finally demonstrated that this method can in principle be applied
to authentic biological samples such as CSF. At present, the detec-
tion sensitivity is not sufficient to measure the A␤ peptides directly
in CSF without a preconcentration step. Current and future efforts
are focusing on improving sensitivity either by coupling CE to LIF
detection or by employing on line preconcentration methods.

3.4. Application to a real CSF sample from AD patient Acknowledgement

To evaluate the suitability of our optimized CE procedure for the This work was supported by an European grant NeuroTAS
analysis of authentic A␤ peptides in human CSF, we used one CSF (STREP-FP6).
sample which was obtained from a patient presenting AD symp-
toms. References
An aliquot of the CSF sample was diluted threefold with water
and this diluted sample was subjected directly to CE analysis [1] J.M. Schott, J. Kennedy, N.C. Fox, Curr. Opin. Neurol. (2006) 552.
[2] N. Andreasen, K. Blennow, Clin. Neurol. Neurosurg. 107 (2005) 165.
(Fig. 7a). For comparison, a second aliquot of the same diluted CSF [3] A.F. Jorm, A.E. Korten, A.S. Henderson, Acta. Psychiatr. Scand. 20 (1987) 736.
was spiked with A␤ peptides 1-37, 1-38, 1-39, 1-40, 1-42 (3 ␮M of [4] K. Blennow, NeuroRx 1 (2004) 213.
164 R. Verpillot et al. / J. Chromatogr. A 1214 (2008) 157–164

[5] C. Haass, M.G. Schlossmacher, A.Y. Hung, C. Vlgo-Pelfrey, A. Mellon, B.L. [17] P. Lewczuk, J. Wiltfang, Proteomics 8 (2008) 1292.
Ostaszewski, I. Lieberburg, E.H. Koo, D.B. Schenk, D.B. Teplow, D.J. Selkoe, Nature [18] C.A. Lucy, A.M. MacDonald, M.D. Glucev, J. Chromatogr. A 1184 (2008) 81.
353 (1992) 322. [19] V. Dolnik, Electrophoresis 29 (2008) 143.
[6] M. Shoji, T.E. Golde, J. Ghiso, T.T. Cheung, S. Estus, L.M. Shaffer, X.D. Cai, D.M. [20] V. Dolnik, Electrophoresis 27 (2006) 126.
McKay, R. Tintner, B. Frangione, S.G. Younkin, Science 258 (1992) 126. [21] V. Kasicka, Electrophoresis 29 (2008) 179.
[7] P. Seubert, C. Vigo-Pelfrey, F. Esch, M. Lee, H. Dovey, D. Davis, S. Sinha, M. Schioss- [22] V. Kasicka, Electrophoresis 27 (2006) 142.
macher, J. Whaley, C. Swindlehurst, R. McCormack, R. Wolfert, D. Selkoe, I. [23] E. Varesio, S. Rudaz, K.H. Krause, J.L. Veuthey, J. Chromatogr. A 974 (2002) 135.
Lieberburg, D. Schenk, Nature 359 (1992) 325. [24] S. Sabella, M. Quaglia, C. Lanni, M. Racchi, S. Govoni, G. Caccialanza, A. Calligaro,
[8] S. Wittke, H. Mischak, M. Walden, W. Kloch, T. Rädler, K. Wiedemann, Elec- V. Belloti, E.D. Lorenzy, Electrophoresis 25 (2004) 3186.
trophoresis 26 (2005) 1476. [25] M. Kato, H. Kinoshita, M. Enokita, Y. Hori, T. Hashimoto, T. Iwatsubo, T. Toyo’oka,
[9] J. Hernandez-borges, C. Neusüss, A. Cifuentes, M. Pelzing, Electrophoresis 25 Anal. Chem. 79 (2007) 4887.
(2005) 2257. [26] A. Eckhardt, I. Miksik, Z. Deyl, J. Charvatova, J. Chromatogr. A 1051 (2004)
[10] R. Aebersold, M. Mann, Nature 422 (2003) 198. 111.
[11] C. Vigo-Pelfrey, D. Lee, P. Keim, J. Neurochem. 61 (1993) 1965. [27] D. Corradini, G. Cannarsa, E. Fabbri, C. Corradini, J. Chromatogr. A 709 (1995)
[12] A. Asami-Odaka, Y. Ishibashi, T. Kikuchi, C. Kitada, N. Suzuki, Biochemistry 34 127.
(1995) 10272. [28] M. Taverna, A. Baillet, D. Biou, M. Schülter, R. Werner, D. Ferrier, Electrophoresis
[13] J.Z. Wang, I. Grundke-Iqbal, K. Iqbal, Nat. Med. 2 (1996) 871. 13 (1992) 359.
[14] C.L. Masters, G. Simms, N.A. Weinman, G. Multhaup, B.L. McDonald, K. [29] N.T. Tran, M. Taverna, L. Miccoli, J.F. Angulo, Electrophoresis 26 (2005) 3105.
Beyreuther, Proc. Natl. Acad. Sci. U.S.A. 82 (1985) 4245. [30] K.K.C. Yeung, C.A. Lucy, Anal. Chem. 69 (1997) 3435.
[15] G.G. Glenner, C.W. Wong, Biochem. Biophys. Res. Commun. 120 (1984) 885. [31] H. Reiber, E.J. Thompson, G. Grimsley, G. Bernardi, P. Adam, S. Monteiro de
[16] J. Wiltang, H. Esselmann, M. Bibl, A. Smirnov, M. Otto, S. Paul, B. Schmidt, H.W. Almeida, P. Fredman, G. Keir, M. Lammers, R. Liblau, M. Menna-Barreto, M.J. Sá, E.
Klafki, M. Maler, T. Dyrks, M. Bienert, M. Beyermann, E. Rüther, J. Kornhuber, J. Seres, C.J. Sindic, A. Teelken, C. Trendelenburg, M. Trojano, M.P. Van Antwerpen,
Neurochem. 81 (2002) 481. M.M. Verbeek, Clin. Chem. Lab. Med. 41 (2003) 331.