Analyst

View Article Online

PAPER View Journal

Magneto-immunocapture with on-bead

fluorescent labeling of amyloid-β peptides:
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

Cite this: DOI: 10.1039/c5an01179e

towards a microfluidized-bed-based operation
Thanh Duc Mai,a,b Iago Pereiro,c Mohamed Hiraoui,a,b Jean-Louis Viovy,c
Stéphanie Descroix,c Myriam Tavernaa,b and Claire Smadja*a,b

A new sample treatment approach for sensitive determination of three amyloid-β peptides (Aβ 1-42,
Aβ 1-40 and Aβ 1-38) with capillary electrophoresis coupled with laser induced fluorescent detection is
reported herein. These Aβ peptides are considered an important family of biomarkers in the cerebrospinal
fluid (CSF) for early diagnosis of Alzheimer’s disease (AD). Due to their extremely low abundance in CSF
(down to sub nM ranges), batch-wise preconcentration via magneto-immunocapture with enrichment
factors up to 100 was implemented. The Aβ peptides were first captured onto magnetic micro-beads.
Then, on-beads fluorescent labeling of the captured Aβ peptides were carried out to avoid the unwanted
presence of extra fluorescent dye in the eluent as in the case of in-solution labeling. Finally thermal
elution was performed and eluted labeled peptides were analyzed off line with CE-LIF. The Aβ-capturing
Received 12th June 2015, efficiencies of different commercially available antibodies grafted onto magnetic beads were tested. Aβ
Accepted 17th July 2015
peptides in CSF samples collected from AD’s patients and healthy persons (used as controls) were
DOI: 10.1039/c5an01179e measured and evaluated. As a proof of concept, the developed strategy was adapted into a miniaturized
www.rsc.org/analyst fluidized bed configuration that has the potential for coupling with a microchip separation system.

1. Introduction determination of Aβ 1-42 together with other biomarkers has

therefore been implemented for more precise discrimination
Amyloid β (Aβ) peptides having from 37 to 43 amino acids (AA) of AD from other NDs. Different proposed biomarker combi-
are naturally produced in our body fluids, mainly cerebro- nations for AD include Aβ 1-42 and Tau/p-Tau protein,5,6 Aβ
spinal fluid (CSF), via proteolysis of a larger protein known as 1-42/Aβ 1-40 ratio,6,7 Aβ 1-42, Aβ 1-40 and Aβ 1-38,8,9 the ratio
the amyloid precursor protein (APP).1 Among these Aβ iso- of Aβ 1-42/(Aβ 1-42 + Aβ 1-40 + Aβ 1-38),10 the ratio Aβ 1-40/42 and
forms, Aβ 1-42 and Aβ 1-40 are considered well-established and those between Aβ 1-37, Aβ 1-38 and Aβ 1-3911,12 and Aβ 1-42
internationally validated biomarkers for early diagnosis of Alz- together with Aβ 2-42.13 Esselmann et al. described in their
heimer’s disease (AD).2 CSF is a relevant biological fluid for patent the use of quantitative ratio of Aβ 1-42, Aβ 2-40 and Aβ
the monitoring of AD evolution as it reflects molecular events 2-42 for early diagnosis of AD.14
occurring in the brain thanks to its direct contact with the In the context of determination/quantification of different
extracellular brain space.3 In AD patients, the CSF levels of Aβ AD’s biomarkers in general and Aβ peptides in particular in
1-42 (<500 pg mL−1) are lower than those in normal people CSF, the most frequently employed analytical approaches are
(about 800 pg mL−1) due to its aggregation in the brain’s immunoassays,15–19 mass spectrometry16,20–23 and Western
senile plaques.2 On the other hand, low levels of Aβ 1-42 in blot.11,24–26 Like the two other techniques, the immunoassay
CSF may be also associated with other neurodegenerative dis- based ones (MSD, xMAP, ELISA) can now detect different AD’s
eases (NDs), notably dementia with Lewy body (DLB), Parkin- biomarkers at the same time as they can employ simul-
son’s disease (PDD) or Creutzfeld–Jacob disease (CJD).4 The taneously different antibodies specific for each biomarker.15,27
However, all these methods, albeit being well established and
having been used as reference techniques, are hardly converti-
a
Université Paris-Sud, Institut Galien Paris-Sud, 5 rue JB Clément, 92296 Châtenay- ble into miniaturized formats and therefore have little or no
Malabry, France. E-mail: [email protected]; Fax: +33 1-46-83-54-62
b
potential for production of point-of-care devices for early AD
CNRS, UMR 8612, 5 rue JB Clément, 92296 Châtenay-Malabry Cedex, France
c
Macromolecules and Microsystems in Biology and Medicine, Institut Curie, Centre
diagnosis purpose.
National de Recherche Scientifique, Université Pierre et Marie Curie, UMR 168, One alternative to the aforementioned techniques which
75005 Paris, France has a high potential for miniaturization, automation and

This journal is © The Royal Society of Chemistry 2015 Analyst

 View Article Online

Paper Analyst

integration while offering also peptide separation capabilities operation. The magnetic beads in this case were made circu-
is electrokinetic separation. Capillary zone electrophoresis lated inside the microchannel rather than kept stationary
(CZE), being the simplest mode of this technique, was applied during the passage of the sample flow for improvement of
successfully for separation and quantification of a mixture of contact between Aβ peptides and magnetic beads and to
different Aβ peptides down to 300–500 nM with UV detection28 increase the analysis throughput.
and 35 nM with laser-induced fluorescent detection (LIF),10
respectively. Downscaling of this approach to microchip plat-
forms has also been reported by our group with detectable 2. Experimental
levels of Aβ peptides close to 200 nM.9,29 These detection
2.1. Chemicals, reagents and samples
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

limits however were not sufficient for analyses of CSF samples

where concentrations of Aβ peptides are only at sub nM Amyloid peptides Aβ 1-38 and Aβ 1-40 were purchased from
ranges. Recently Wiltfang’s group presented a novel method Anaspec (Fremont, CA, USA) whereas Aβ 1-42 was obtained
based on capillary isoelectric focusing (CIEF) immunoassay from American peptide (Sunnyvale, CA, USA). Boric acid,
which was capable of detecting total Aβ in CSF after desalting/ formic acid, phosphate buffered saline (PBS 10×), bovine
buffer exchange.26 This method nevertheless is not applicable serum albumin (BSA), ammonium hydroxide 28.1% (m V−1),
to C-truncated peptides possessing the same isoelectric point sodium hydroxide, dimethyl sulfoxide (DMSO, 99.9% purity),
such as Aβ 1-38, Aβ 1-40 and Aβ 1-42. To compensate for the triethanolamine (TEA), IgG from murine serum (reagent grade,
insufficient sensitivity of Aβ peptides determination via ≥95%) and diaminobutane chloride (DAB) were provided by
electrokinetic techniques, inclusion of a forefront sample treat- Sigma (St. Louis, MO, United States). The Fluoprobe 488 NHS
ment module based on magnetic immuno-capture, so-called (FP-488) was purchased from Interchim (Montluçon, France)
immuno-precipitation (IP) has been proposed.10,26,29,30 This and was dissolved in DMSO to obtain aliquots of 10 mg mL−1
technique belongs to the sample treatment strategies that are which were then stored at −20 °C in the darkness. Dimethyl
based on bio-functionalized magnetic nanoparticles. They are pimelimidate dihydrochloride (DMP) was purchased from
particularly suitable for peptides pre-concentration due to Thermo Scientific (Rockford, USA). All buffers were prepared
their large surface area, biocompatibility and ease in manipu- with deionized water purified with a Direct-Q3 UV purification
lation.31 To further enhance the detection sensitivity, one system (Millipore, Milford, MA, USA).
option is to combine the magnetic pre-concentration step with Magnetic micro-particles (diameter of 2.8 µm) surface-
fluorescent chemical labeling to allow then LIF detection. bound with sheep anti-mouse IgG (Dynabeads M-280, 10 mg
Fluorescent derivatization of Aβ peptides was already carried mL−1) and different monoclonal anti-Aβ antibodies (6E10,
out off-line prior to the magnetic pre-treatment step.29,30 12F4 and 4G8) were purchased from Covance, Emeryville, CA.
However this labeling–capture–elution sequence was hardly The magnetic microbeads employed in our work were the
adaptable to the microchip platform due to at least two follow- same as those reported in ref 10. All CSF samples were taken
ing limitations: (1) the requirement of one additional micro- by the department of Neurology, university of Ulm (Ulm,
chamber containing no microbeads for fluorescent labeling of Germany), aliquoted and stored at −20 °C until use. The
Aβ peptides prior to their magneto-immunocapture and (2) the sampling procedure was detailed elsewhere.9,10 Our CSF
difficulty in well mixing the laminar flows of sample and fluoro- samples were provided by the University of Ulm. Their collec-
phore solutions in microchannels. In addition, the previous tion and analysis were approved by the Ethics Committee at
studies employed magnetic micro-beads that were kept station- the University of Ulm.
ary in the form of tiny columns inside the micro-chip. This
may not offer optimal contact between Aβ peptides and the 2.2. Materials and apparatus
beads’ surface which is needed for higher capture efficiency. All CE-LIF analyses were implemented with a Beckman Coulter
Herein we reported for the first time an integrated capture– PA 800 ProteomLab coupled with a LIF detection system (a
label–elution strategy that allows Aβ peptides enrichment via 3.5 mW argon-ion laser having an excitation wavelength of
magnetic-bead-based immunocapture followed by their in situ 488 nm and a 520 nm band-pass filter for collection of the
fluorescent-labeling directly on beads prior to CE-LIF oper- emitting beam light). Data acquisition and instrument control
ation. One of the advantages of this novel sample treatment were realized with the Karat 7.0 software. Bare fused silica
strategy is that the unwanted presence of extra fluorescent dye capillaries of 375 µm o.d. and 50 µm i.d. were purchased from
and fluorescent side products9,10 are avoided in the electro- Phymep (Paris, France). pH values of solutions were controlled
phoresis profile. In an effort to enhance IP efficiency, different with an inoLab WTW series pH 730 meter. Amicon ultra 0.5
antibodies with varying specificities toward several Aβ peptides centrifugal filters (3 kDa) were purchased from Millipore
were investigated. The optimized whole sample treatment Ireland (Cork, Ireland). Thermal operation was carried out
process developed in batchwise mode, including magnetic- with a GC oven (GC 5890, series II from Hewlett Packard, USA).
bead-based capturing, on-bead fluorescent labeling and The micro-fluidized bed employed in this work was adapted
thermal elution offered enrichment factors up to 100. The from the design reported recently.32,33 The chips were fabri-
batch-wise mode was then downscaled into a micro-fluidized cated from cyclic olefin copolymer (COC) material to withstand
bed for perspective coupling with microchip electrophoresis temperature increase during the elution step.34 Before the

Analyst This journal is © The Royal Society of Chemistry 2015

 View Article Online

Analyst Paper

loading of magnetic beads, the COC channel was coated with peptides, especially Aβ 1-42 and also to keep the same buffer
BSA 1% to prevent/minimize any possible accumulation of the composition for labeling, elution and separation. A neody-
beads onto the channel wall. More details on the setup and mium magnet (Adem-Mag MSV from Ademtech, Pessac,
operation of this micro-fluidized bed can be referred to ref. 32 France) was employed to retain the magnetic beads during
and 33. removal or addition of a suspension solution.
For magnetic immunocapture in batch, the suspension of
2.3. Methods antibodies-coated magnetic beads (10 mg mL−1) was vortexed
2.3.1 Peptides preparation and storage. Stock Aβ 1-42 was for 3 min for homogenization before withdrawal of 50 µL ali-
prepared in ammonium hydroxide 0.16% (m V−1) whereas quots. The solution was removed from the aliquot and a
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

other amyloid peptides were dissolved in ammonium hydrox- volume of 800 µL of either an STD solution ( prepared in PBS
ide 0.10% (m V−1). Aliquot solutions (10 µL) of individual pep- 1× and BSA 0.1%) or CSF sample was incubated with this 500
tides were prepared at a concentration of 2 mg mL−1 and µg of magnetic beads coated with the desired antibodies on a
subsequently lyophilized to remove all traces of ammonia. shaker at 4 °C for 15 hours. The beads were then washed once
These lyophilized aliquots were then stored at −20 °C until with PBS 1×/BSA 0.1% and twice with borate buffer ( pH 10.5,
use. For preparation of standard solutions (STDs), the lyo- IS 40 mM). Subsequently, 50 µL of borate buffer containing
philized Aβ peptides were diluted with borate buffer ( pH 10.5, FP-488 (0.2–1 mg mL−1) was poured into the washed beads
ionic strength IS 40 mM) to obtain desired concentrations. and the suspension was vigorously orbitally shaken for
These solutions were freshly prepared and stored at 4 °C for 5–60 min at room temperature. This optimization was
use within one day. implemented based on our previous work on in-solution
2.3.2 CE-LIF conditions. The procedure for in-solution fluorescent labeling of Aβ peptides.10 Several parameters were
labeling of Aβ peptides with FP-488, as well as details on changed upon optimization for on-beads fluorescent labeling,
CE-LIF conditions for analyses of the fluorescence-labeled Aβ including the shaking time (from 5 to 60 min) and the concen-
peptides can be found in our previous publication.10 To mini- tration of FP-488 (from 0.2 to 1 mg mL−1). The optimized on-
mize the adverse effect of CE-inherent migration time fluctu- beads labeling conditions (FP-488 at 0.6 mg mL−1, shaking
ation on peak identification and quantification, the migration- time of 30 min, see section 3.1.2) were employed throughout
time-based electropherograms of Aβ peptides after the capture– all capture and elution experiments. The beads were then
label–elution process were converted into electrophoretic- washed twice with borate buffer to remove all un-reacted
mobility-based ones. The peak of FP-488 nonspecifically FP-488, followed by elution of the labeled Aβ peptides bound
adsorbed on the beads surface during the labeling process, on beads. Different elution approaches were tested, including
which always appeared before those of Aβ peptides in the electro- chemical elution with borate buffer ( pH 10.5) or 0.1 M formic
pherograms, was employed as a reference for this profile acid ( pH 2.2) at ambient temperature and thermal elution at
correlation. 95 °C. The eluents were subsequently subjected to CE-LIF ana-
2.3.3 Magneto-immunoprecipitation with fluorescent lyses without further dilution. Note that the eluent volumes
labeling. Magnetic micro-particles (Dynabeads M-280, 10 mg were dependent on the desired enrichment factors. For the
in 1 mL) were separately coated with 40 µg of different mono- tested enrichment factors of 40 and 100, the used eluent
clonal anti-Aβ antibodies (i.e. 6E10, 12F4, 4G8 and IgG) accord- volumes were 20 µL and 8 µL respectively.
ing to the manufacturer protocol. Briefly, the Dynabeads For immunocapture with the fluidized bed platform,
M-280 (10 mg) and the antibodies (40 µg) of selected type were 50–100 µg of magnetic beads were injected in the microchip.
suspended in 1 mL of PBS 1× solution containing 0.1% The cross diameters of its input and output channels were 110
(m V−1) BSA, and were incubated over night on an orbital shaker µm and 100 µm, respectively. The microbeads – containing
(VXR basic Vibrax, Ika, Staufen, Germany) at 4 °C. The mixture chamber had a conic form that reached a diameter of 1800 µm
was then allowed to react with DMP for 1 h at room tempera- at its widest width. The total volume of the chamber was 0.46
ture under a basic condition (0.2 M TEA, pH 10.8) to crosslink µL whereas that occupied by the microbeads was 0.085 µL. 100
the bound antibodies to the surface of magnetic microbeads µL of an STD solution of Aβ peptides (60 nM) was injected in
via the covalent binding between this noncleavable imidoester the chamber with a flow-rate of 1 µL min−1. The beads were
and primary amines in the antibodies. The antibodies-bound then washed with 20 µL of borate buffer at a flow-rate of 1.5 µL
magnetic beads were subsequently washed with PBS 1× con- min−1, followed by passage of 15 µL of borate buffer contain-
taining BSA 0.1% and then re-suspended in PBS 1× containing ing FP-488 at 0.5 µL min−1. After this labeling step, the extra
BSA 0.1% and NaN3 0.02% for storage at 4 °C. These beads and unreacted fluorescent dye was washed out of the chamber
(10 mg mL−1) did not aggregate and performed well even after with a flow of borate buffer (2 µL min−1 for 20 min). For
several months of storage. Addition of any unnecessary com- thermal elution, the chip was heated to 70 °C once the flow
pound other than the reaction medium, including the manu- inside the chamber was completely stopped. The neodymium
facturer-recommended washing buffers citrate and EDTA, in magnet was always positioned at the bottom of the fluidized
any step of the capture–label–elution procedure (even for the bed to prevent the magnetic beads from escaping out of the
beads-washing purpose) was avoided to prevent any possible system. Upon conclusion of the elution step, the solution
triggering of aggregation of the easily-aggregable amyloid inside the fluidic chamber was pushed into a PEEK tubing

This journal is © The Royal Society of Chemistry 2015 Analyst

 View Article Online

Paper Analyst

(length of 30 cm, internal diameter of 250 µm) connected at

the output of the micro-fluidic chip. The solution inside this
reservoir tubing was then collected for subsequent CE-LIF
analyses.

3. Results and discussion

3.1. Magnetic-beads-based sample treatment optimization
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

The sample treatment operation can be divided into three

steps: (1) capture of the Aβ peptides onto the antibody-grafted
magnetic beads, (2) chemical fluorescent labeling of the Aβ
peptides by the fluoprobe FP488 and (3) elution of the pep-
tides from the immunosupport into a small and defined
eluent volume for analyte preconcentration and further electro-
kinetic separation. These steps can in principle be carried out
with different orders. We selected the capture–labeling–elution
sequence as it was expected to simultaneously satisfy efficient
Aβ peptides enrichment, removal of extra fluorophore and
possible translation from batchwise into micro-fluidized
bed format. Compared to the method of static interaction
where the antibodies are immobilized onto a planar surface,
the circulation of the magnetic beads in our approach is Fig. 1 CE-LIF electropherograms of a mixture of Aβ 1-42 (8 nM) and Aβ
expected to facilitate the dynamic interaction between the anti- 1-40 (15 nM) after magnetic-beads-based sample treatment (capture–
label–elution) obtained with different anti-Aβ antibodies. The batch-
bodies and the Aβ peptides, which in turn should improve the based capture and label conditions were detailed in the ‘Experimental’
immune-capture efficiency. This also is the case for sub- section. For elution, the magnetic beads were heated up to 95 °C for
sequent on-beads fluorescent labeling. In addition, the surface 5 min in the presence of borate buffer ( pH 10.5, IS 40 mM). Enrichment
area provided by a microbead of diameter D (Area = πD2) is factor of 40. CE-LIF conditions: fused silica capillary with effective
much larger than that of a square planar surface having the length (leff ) of 38.4 cm and total length (L) of 48.2 cm; BGE: borate
buffer ( pH 9.25, IS 40 mM) added with DAB 3.25 mM; LIF detection with
same length D (Area = D2). As a result, for the same antibody excitation wavelength (λ) of 488 nm.
immobilization efficiency, a higher density of antibodies is
expected for micro-beads, allowing better immuno-capture
and higher enrichment gains.
3.1.1. Capture of Aβ peptides onto magnetic beads. The inal concentration ratio of Aβ 1-42 (8 nM)/Aβ 1-40 (15 nM) was
capture of Aβ peptides onto the magnetic beads is based on preserved after the immunocapture process. The peaks
specific interaction between Aβ peptides with the antibodies obtained with 6E10 and 12F4 are much higher (approximately
grafted onto the surface of the beads.29,30,35,36 The binding 4 times) than those with 4G8. Clearly, 6E10 and 12F4 exhibit
yields of the antibodies to the magnetic microbeads and their the best performance in terms of capture of Aβ 1-40 and Aβ
specificity towards the targeted peptide epitope play a crucial 1-42 for the former and Aβ 1-42 for the latter respectively under
role in determining efficiency of the immune-capture process. the tested conditions. To our opinion, the poorer result with
Accordingly, three commercial monoclonal mouse antibodies 4G8 is possibly associated with: (1) low binding yields of anti-
from Covance, namely 6E10, 4G8 and 12F4, which are reactive bodies to the beads surface, leading to an unsatisfactory on-
to amino acid residues 1–16 (N-terminus), amino acid residues bead antibody density and (2) less or no accessibility of
17–24 and the C-terminus of beta amyloid, respectively were primary and/or Lysine’s amino groups which are needed for
tested. 12F4 is specific for the peptide isoform’s ending at the subsequent fluorescent labeling once the Aβ peptides are on-
42nd amino acid and was already used in a silicon-platform- bead-immobilized. The efficient binding of 4G8 immobilized
based high-sensitivity immunoassay for Aβ 1-42.17 Non specific onto a silicon microarray platform to underivatized Aβ 1-42
IgG from murine serum was employed to check for non and Aβ 1-39, as reported by Gagni et al. (see ref. 17), excludes
specific capture. Magnetic beads grafted with the different the possibility that the immobilized 4G8 antibody shows low
antibodies were subjected to the same sample treatment affinity for the concerned peptides.
process, i.e. capture–labeling–elution. Eluted fractions with the 3.1.2. On-bead fluorescent labeling of Aβ peptides. Con-
enrichment factor of 40 were directly analyzed with CE-LIF sidering the aforementioned preliminary results, the optimiz-
(Fig. 1). With the antibodies 6E10 and 4G8, Aβ 1-42 and Aβ ation of on-bead fluorescent labeling with FP488 was then
1-40 peaks could be detected whereas only that of Aβ 1-42 performed with 6E10 as the capture antibody. Borate buffer
appeared in the case of 12F4. In the case of 6E10, the area ( pH 10.5) was employed to provide a basic medium needed to
ratio of Aβ 1-40/Aβ 1-42 was 0.44, which means that the orig- produce mainly ditagged species formation. Ditagged species

Analyst This journal is © The Royal Society of Chemistry 2015

 View Article Online

Analyst Paper

display indeed higher fluorescent intensities and are better the beads during this labeling step. Furthermore, the elution
resolved by CE-LIF than monotagged species.10 The alkaline medium had to be optimized taking into consideration the
pH required for achievement of desirable di-tagged peptides subsequent CE-LIF analysis as it would significantly influence
still belongs to the pH range for pH-independent fluorescence the analytical performance in terms of Aβ peak shape and sen-
emission of FP-488 which is 4–10 according to the manufac- sitivity. This situation is reflected in Fig. 2 with electrophero-
turer. In our previous work reporting in-solution fluorescent grams of fluorescently labeled Aβ 1-40 prepared in different
labeling, it was found that a 5 min static incubation of Aβ pep- media. The Aβ 1-40 peptide was labeled by a 5 min static incu-
tides with 0.2 mg mL−1 FP-488 (molar ratio of FP-488 to Aβ bation of Aβ 1-40 with 0.2 mg mL−1 FP-488 prepared in DMSO
peptides more than 4) was sufficient for complete in-solution according to the procedure reported in our previous publi-
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

peptide labeling with Aβ concentrations ranging from some cation.10 High and sharp peaks were obtained with alkaline
dozen up to at least a thousand nM.9,10 The direct application media while an evident decrease in peak height was observed
of this in-solution labeling conditions however led to unsatis- with a shift of pH towards acidic conditions. Based on these
factory fluorescent tagging of Aβ peptides when they were considerations, different elution options were then investi-
bound on biofunctionalized beads. FP-488 is a fluorescent dye gated and their capacities of desorbing labeled Aβ out of the
that binds covalently to amino groups of the peptides. Each magnetic beads were compared (Fig. 3). Regardless of the
intact beta amyloid possesses three conjugable amino groups, medium used and its pH, the chemical elution at room temp-
i.e. one primary and two on Lysine’s residues. Nevertheless, erature did not lead to appearance of Aβ peaks in the electro-
they can become less accessible after the binding of Aβ pep- pherograms. Elution with ammonium hydroxide 0.16–1% or
tides on beads. In the case of 6E10, the primary amine of Aβ citric acid 100 mM was tested as well, but led to no detectable
peptides might partially lose its availability once 6E10 binds to peaks with CE-LIF. This confirms that the on-beads bound Aβ
their 1–16 amino acid residues. Another obstacle to efficient were not eluted by the basic medium ( pH 10.5) employed
on-bead Aβ labeling came from the employment of a basic during the fluorescent labeling process. Only thermal elution
medium ( pH 10.5). While this medium is expected to facilitate at 95 °C for 5 min allowed detecting Aβ peptides with CE-LIF.
ditagged labeling,10 it can also accelerate hydrolysis of the The heating is therefore critical in breaking the interaction
FP-488, which in turn could degrade the labeling efficiency between Aβ peptides and the antibodies. No observable ther-
over time. To overcome these problems, two optimizations mally induced modification of labeled Aβ peptides confor-
were performed. Firstly, a prolongation of incubation time to mation was found, as no electrophoretic mobilities shift was
30 min in combination with vigorous agitation was carried out
to favor the contact between FP-488 and on-beads bound Aβ
peptides. Secondly, an increase in FP-488 concentration was
implemented to compensate for the amount of FP-488 lost due
to hydrolysis in the basic medium and to maintain the reac-
tion towards stable conjugate formation over the prolonged
period. Accordingly, it was found that a shaking incubation
with an elevated FP-488 concentration of 0.6 mg mL−1 in
borate buffer at room temperature for 30 min offered the best
on-bead fluorescent labeling of Aβ peptides. The labeling reac-
tion occurred under the presence of DMSO at small amounts
(2–6% v/v) when FP-488 prepared in DMSO was added into the
mixture. Nevertheless, the effect of DMSO on the established
binding equilibrium between Aβ peptides and on-beads-
immobilized antibodies (if any) was considered insignificant
due to its small volumetric percentage. An efficient on-beads
capture of Aβ peptides under such presence of DMSO was
already reported in the label–capture–elution approach develo-
ped by Svobodova et al.30
3.1.3. Elution of the bound and labeled Aβ peptides.
Different approaches for elution of the captured Aβ right after
magnetic-immunoprecipitations were already reported, includ-
ing chemical elution with a large volume of ammonium
hydroxide,10,30 formic acid35,36 or thermal elution in the pres-
ence of a neutral buffer.26 While these methods exhibited sat-
Fig. 2 CE-LIF electropherograms of fluorescently labeled Aβ 1-40
isfactory performance in a two-step capture–elution procedure,
(1000 nM) prepared in different media. The Aβ 1-40 peptide was labeled
the inclusion of an on-bead fluorescent labeling process in by 5 min static incubation of Aβ 1-40 with 0.2 mg mL−1 FP-488 prepared
between may render the Aβ peptides elution more difficult due in DMSO according to the procedure reported in our previous publi-
to the hindrance of FP-488 dyes non-specifically adsorbed onto cation.10 SM: sample medium. CE conditions as in Fig. 1.

This journal is © The Royal Society of Chemistry 2015 Analyst

 View Article Online

Paper Analyst

elution medium in this thermal approach. In addition the use

of the same buffer compositions for thermal elution and for
the fluorescent labeling is expected to facilitate the subsequent
translation of this batchwise operation into micro-fluidic flui-
dized bed format. By employing the borate buffer, three con-
secutive steps, including fluorescent labeling, beads washing
and elution can share the same medium.

3.2. Separation and sensitive detection of Aβ peptides in

CSF samples
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

Excluding all washing steps, the optimized batchwise pro-

cedure of magnetic-bead-based sample treatment prior to
CE-LIF comprised 15 hours of magneto-immunocapture at
4 °C, 30 min of on-bead fluorescent labeling at room tempera-
ture and 5 min of thermal elution at 95 °C. With this protocol
using 6E10 antibody, the salient performance data were pre-
sented in Table 1. Calibration curves were acquired with satis-
factory linearity (correlation coefficients more than 0.96) for
the concentration ranges of 40–400 nM for Aβ 1-42 and 40–800
nM for Aβ 1-38 and Aβ 1-40 respectively. Over these ranges no
further increase in peak heights was observed. The reproduci-
bility of the inter-batch measurements of corrected peak
Fig. 3 CE-LIF electropherograms of the captured and labeled Aβ 1-42 areas and migration times was found to be about 12% and
(8 nM), Aβ 1-40 (15 nM) and Aβ 1-38 (15 nM) after chemical and thermal 1%, respectively. They deemed acceptable considering that
elutions. Capture was done with magnetic micro-beads coated with these RSD values are due to the accumulation of errors of all
either 6E10 or IgG (for blank control). CE conditions as in Fig. 1. (A)
operations, i.e. sample preconcentration, labeling, elution,
Capture with 6E10 and elution with formic acid 0.5% at room tempera-
ture for 15 min; (B) capture with 6E10 and elution with borate buffer ( pH
injection and separation. In the case of repeated CE-LIF
10.5, IS 40 mM) at room temperature for 15 min; (C) capture with 6E10 measurements of the same batch, the intra-batch RSD values
and elution with borate buffer ( pH 10.5, IS 40 mM) at 95 °C for 5 min for peak areas were improved to around 6%. An enrichment
and (D) capture with IgG and elution with borate buffer ( pH 10.5, IS factor of 100 (calculated from the sample to eluent volumes
40 mM) at 95 °C for 5 min (blank).
ratio) for the Aβ 1-38, Aβ 1-40 and Aβ 1-42 peptides could be
obtained (Table 1). The calculation of preconcentration gains
based on a referenced peak area was not possible due to the
unavailability of the (commercial) standard fluorescently
observed for the thermally eluted peptides compared to the in- ditagged Aβ peptides containing no extra FP-488. Deduction of
solution labeled ones. With this thermal elution approach, the the concentrations of fluorescently labeled peptides from
antibodies-coated magnetic beads could be used only once those of the standard peptides was not done neither because
because the antibodies could be denatured by such high temp- the precise yield of Aβ peptides fluorescent di-tagging could
erature. Due to its favorable properties in terms of sharp peak not be determined. For this reason, while the relative ratios of
shapes (Fig. 2) and high compatibility with the subsequent CE different peptides captured on micro-beads could be provided
analysis, borate buffer ( pH 10, IS 40 mM) was selected as the (see section 3.1.1), information on the absolute recovery was

Table 1 Linearity of the response, detection limits (LODs) and reproducibility for the determination of preconcentrated Aβ peptides by CE-LIF.
CE-LIF conditions as in Fig. 1

Correlation LODa LOD-Pb RSD % Tm c RSD % Ainter d RSD % Aintra e

Analyte Range (nM) coefficient (nM) (nM) (n = 4) (n = 4) (n = 3)

Aβ 1-42 40–400 0.9675 10 0.1 1.31 14.03 7.39

Aβ 1-40 40–800 0.9712 8 0.08 1.63 11.79 5.29
Aβ 1-38 40–800 0.9882 10 0.1 1.03 11.95 6.34
a
Based on peak heights corresponding to 3 times the baseline noise. b LOD-P: detectable concentrations with the calculated enrichment factor
F = 100. c Migration time, corrected to the reference peak of FP-488. RSD was measured with standard concentration of 15 nM. d RSD % Ainter: Inter-
batch RSD values for peak areas (comparison of 4 different batches). The peak areas were calculated from the mobility-based electropherograms
that were converted based on the migration times of FP-488 that was used as the referenced peak. e RSD % Aintra: RSD values for peak areas
obtained with the same batch. The batch was measured three times with CE-LIF. The peak areas were calculated from the mobility-based
electropherograms that were converted based on the migration times of FP-488 that was used as the referenced peak.

Analyst This journal is © The Royal Society of Chemistry 2015

 View Article Online

Analyst Paper

not obtainable. The smallest Aβ concentrations that were

detectable by LIF detection, in other words the detection
limits without preconcentration, where 8–10 nM, which are
almost 4 times lower than that reported in the precedent work
on CE-LIF.10 This is mainly due to the absence of exceeding
fluorescent dye in the final sample solution, which is only
made possible with on-bead labeling approach. Further
improvement in LODs could be achievable just by increasing
the amount of magnetic beads in the batchwise capture
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

process. This however was not envisaged as diminution of

beads amount would be needed for a prospective adaptation
towards a micro-fluidized bed platform. The detectable
levels of Aβ peptides by LIF after our developed immune-
capture–on-beads labeling–thermal elution procedure with an
enrichment factor of 100 were 0.08–0.1 nM which are compati-
ble with the Aβ peptides level in the CSF. In terms of sensi-
tivity, our method has not reached yet the ELISA performance
whose quantifiable level of Aβ 1-42 is 0.03 nM.15 On the other
hand, our method offers some advantages over ELISA, includ-
ing (1) concurrent separation and determination of different
structurally close Aβ peptides without recourse to simul-
taneous employment of different antibodies of high specificity
towards each Aβ peptide of interest9,10 and (2) distinguishing
of monomeric and oligomeric forms of Aβ 1-42.37
The concentrations of Aβ 1-38, Aβ 1-40 and Aβ 1-42 in CSF
samples were determined using the aforementioned sample
treatment method followed by CE-LIF analyses. Electrophero-
grams of CSF samples obtained from cognitive normal (con-
trols) or AD patients are shown in Fig. 4. Both the standards
and the CSF samples were subjected to the same enrichment –
fluorescent derivatization for appropriate comparison. The fil-
tration through 3 kDa filters was employed to remove all small
unwanted species from the sample matrix while retaining the
concerned Aβ. Much better baseline with no signal drifting
was achieved with sample pre-filtering. Among the three Aβ
peptides, Aβ 1-40 in CSF samples was present at the highest
concentrations of 6–8 nM. Aβ 1-38 in these CSF samples was
found to be at the order of 1–2 nM. The most challenging was
the determination of Aβ 1-42 due to its extremely low abun-
dance in CSF samples, as well as the presence of some
unknown peaks adjacent to that of Aβ 1-42 (see Fig. 4).
However, identification of these other compounds was not
envisaged in the scope of this work. Recourse to immuno-
capture with 12F4 was then realized for selective determination Fig. 4 CE-LIF electropherograms of CSF samples after magnetic
of Aβ 1-42, as demonstrated in Fig. 5. The employment of this immune-precipitation using the antibody 6E10 (enrichment factor of
antibody specific for the C-terminus of Aβ 1-42 led to a clear 100) and on-beads fluorescent labeling. Thermal elution conditions:
95 °C for 5 min in the presence of borate buffer ( pH 10.5, IS 40 mM). CE
visualization of Aβ 1-42 peak only. A decrease in Aβ 1-42 con-
conditions as in Fig. 1. (A) Sample treatment without pre-filtering of CSF
centration could be observed in the CSF sample from an AD samples; (B) sample treatment with pre-filtering of CSF samples using
patient compared to that of a control one. The relative concen- 3 K Dalton filters. Peak identification: (1) Aβ 1-42, (2) Aβ 1-40, (3) Aβ 1-38.
trations ratios of Aβ peptides in CSF samples obtained with CSF – AD stands for CSF samples from AD patients; CSF – C stands for
our approach were in agreement with those reported by Svobo- CSF samples from cognitive normal people used as controls; STD 1:
standard solution of Aβ 1-42 (0.2 nM), Aβ 1-40 (0.2 nM) and Aβ 1-38 (0.2
dova et al.30 who used the same antibodies and CSF samples
nM); STD 2: standard solution of Aβ 1-42 (2 nM), Aβ 1-40 (8 nM) and Aβ
provided from the same source (Ulm hospital) but not from 1-38 (8 nM). The indicated concentrations in the brackets were those of
the same patient. Pre-filtering of CSF samples would probably the standards before immune-enrichment. Both standards and CSF
increase the performance of our system. Nevertheless, our samples were subjected to the same immunocapture–label–elution
main goal was to conceive a microsystem that integrates all procedure.

This journal is © The Royal Society of Chemistry 2015 Analyst

 View Article Online

Paper Analyst

improved in this case compared to the previous microchip

design where magnetic microbeads were kept stationary inside
the microchannel.29,30 Indeed, with the previous microchip
design, fracture in the magnetic plug could occur over a
certain flowrate. In case of fracture the liquid could preferen-
tially flow through this fracture of lower hydrodynamic resist-
ance and thus reduces drastically the bed surface in contact
with the sample. With our device we obtained a homogeneous
contact of the liquid with all the beads and we could thus
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

expect a better efficiency. Compared to batch experiment, this

micro-fluidized bed configuration allows an efficient stirring
and should enhance the mass transfer between the solid and
the liquid phase. Accordingly, better capture efficiency and
labeling interaction are expected to be achievable within
shorter operation duration. The capture–label–thermal
elution procedure was then carried out in the micro-fluidized
bed and the fluorescent signal obtained at its output micro-
channel after thermal elution is shown in Fig. 6A. This fluo-
rescent fraction was subsequently collected and subjected to
CE-LIF separation for visualization of Aβ 1-42, Aβ 1-40 and Aβ
1-38 as shown in Fig. 6B. The fluorescent signal shown in
Fig. 6A was indeed the total signals of the released FP-488 and
Fig. 5 CE-LIF electropherograms of CSF samples after magnetic
labeled Aβ peptides that were well separated with CE-LIF as
immune-precipitation using the antibody 12F4 (enrichment factor of
100) and on-beads fluorescent labeling. Thermal elution conditions: displayed in Fig. 6B. The operation time was shortened from
95 °C for 5 min in the presence of borate buffer ( pH 10.5, IS 40 mM). CE more than 15 hours in the batch-based mode into only around
conditions as in Fig. 1. Sample treatment was carried out without pre- 3 hours with the fluidized bed configuration. To avoid satur-
filtering of CSF samples. CSF-AD: CSF sample collected from an AD ation of the microchannel, the amount of magnetic beads
patient; CSF-C: CSF sample from a cognitive normal person (used as
was tenfold reduced. Even under this condition, successful
control); STD: standard Aβ solution (8 nM). The indicated concentration
in the brackets was that of the standard before immune-enrichment. capture–label–elution operation in this micro-fluidized bed
Both the standard and CSF samples were subjected to the same immuno- was still achieved with an enrichment factor of 20. Higher pre-
capture–label–elution procedure. concentration gains are expected if dilution of the eluent at
the output of the microfluidized bed (at the nL range) into a
collectable volume (some µL) for subsequent CE-LIF operation
could be avoided. This would be the case when integrating the
pre- and analytical steps. As filtering in microsystems is not preconcentration step to the microchip electrophoresis oper-
trivial, this additional sample treatment step was not further ation will be realized. We demonstrated therefore that this
considered in this work. novel approach which integrated for the first time Aβ capturing
and fluorescent labeling on the same immunosupport could
3.3. Operation in a microfluidic fluidized bed: a proof of be downscaled. In the case of offline labeling followed by
concept immunocapture,29,30 manual batchwise in-solution labeling in
While the batchwise mode of the developed magnetic the absence of magnetic beads can hardly be downscaled into
immunocapture exhibited satisfactory performance for the the micro-fluidized bed platform. Operation integration and
determination of Aβ peptide mixture in CSF samples, its long automation in the magnetic-beads based microfluidic system
operation time with manual solution switching may pose for perspective point-of-care device production are therefore
some inconvenience to the operator. Our prospective objective more achievable with the developed capture–label–elution
is therefore to integrate all capture, label, elution and MCE-LIF strategy where the sample capture and labeling can be both
separation steps into an automated procedure in a micro- implemented on-chip in the same microchamber. Further-
fluidic platform. The preliminary stage towards this objective more, the undesirable preferential capture of fluorescently
relies on the implementation of the capture–label–elution pro- labeled Aβ 1-42 over the other Aβ peptides encountered in the
tocol inside a microfluidic fluidized bed whose configuration offline labeling–immunocapture technique reported by Svobo-
was reported recently.32,33 Inside its chamber, magnetic micro- dova et al.30 was not observed with our capture–label–elution
particles continuously re-circulate thanks to two counter approach. For successful perspective development of a point-
driving forces, i.e. pressure-driven hydrodynamic and magnetic of-care device that can quantitatively reach the extremely low
attraction forces. The contact of magnetic beads with Aβ pep- level of Aβ 1-42 in CSF samples, three processes have been
tides (during the capturing step) and with FP-488 (during the envisaged, including: (1) improvement of LIF detection limit
on-bead labeling process) is therefore expected to be much from 35 nM to 8 nM without the preconcentration step by the

Analyst This journal is © The Royal Society of Chemistry 2015

 View Article Online

Analyst Paper

enrichment gain. At the present stage, the presence of Aβ 1-42

in CSF samples was detected with the first two aforementioned
approaches.

4. Conclusions
This work contributed significantly to three achievements.
Firstly, an MCE/CE-LIF compatible technique of Aβ peptides
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

enrichment was successfully developed based on magneto-

immunocapture operation. Secondly, the successful fluo-
rescent labeling of Aβ peptides was for the first time done on-
beads, leading to the removal of all unwanted extra fluorescent
dyes. These pre-concentration and fluorescent labeling steps
which act as a forefront sample treatment process prior to
CE-LIF render the separation and sensitive detection of Aβ
1-42, Aβ 1-40 and Aβ 1-38 in CSF samples possible. The FP-488
that was non-specifically adsorbed onto microbeads during
the labeling process and then released into the eluent during
the elution step was profited as the internal standard for calcu-
lation of relative migration times of Aβ peptides in CE-LIF pro-
files. Finally, a proof of concept of micro-fluidized bed based
operation for enrichment and fluorescent-labeling of Aβ was
successfully demonstrated. This opens the door for production
of hand-held devices for facile and high-throughput probing of
Aβ peptides in CSF samples based on lab-on-chip electro-
kinetic separation. Perspective work on coupling the fluidized
bed with microchip electrophoresis will be soon envisaged.

Acknowledgements
This work and the post-doctoral fellowship for Dr Thanh Duc
Mai have been financially supported by the European Commu-
nity’s Seventh Framework Programme (NaDiNe FP7/2010-2015)
under the grant agreement no 246513. The post-doctoral fel-
lowship for Dr Mohamed Hiraoui was supported by DIM Ana-
lytics program (http://www.dim-analytics.fr/). We would like to
thank Dr Romain Verpillot for valuable discussion. We are
also grateful to Dr Ute Haussmann and Dr Hans Klafki (Uni-
versity of Duisburg-Essen, Germany) for their useful advices
Fig. 6 (A) Fluorescent observation at the output channel of the micro- concerning bead grafting, as well as to Prof. Dr Markus Otto
fluidized bed during the passage of the eluent after thermal elution. The at the department of Neurology, university of Ulm (Ulm,
sample is 100 µL of 60 nM standard peptides. The capture–label– Germany) for CSF sampling operation.
elution procedure can be found in section 2.3. (B) CE-LIF electrophero-
gram obtained for the eluent collected at the output channel of the
micro-fluidized bed after thermal elution. CE conditions as in Fig. 1.
References
1 J. Kang, H. G. Lemaire, A. Unterbeck, J. M. Salbaum,
removal of the exceeding fluorophores in the sample matrix C. L. Masters, K. H. Grzeschik, G. Multhaup, K. Beyreuther
after the labeling, by using microbeads (2) pre-concentration and B. Mullerhill, Nature, 1987, 325, 733–736.
of Aβ peptides with microbeads-based immunocapture and (3) 2 C. Humpel, Trends Biotechnol., 2011, 29, 26–32.
integration of all steps, i.e. preconcentration–labeling–elution– 3 E. R. Peskind, R. Riekse, J. F. Quinn, J. Kaye, C. M. Clark,
separation on a microfluidized bed coupled with a separation M. R. Farlow, C. DeCarli, C. Chabal, D. Vavrek,
microchip to eliminate the volume loss and variation at each M. A. Raskind and D. Galasko, Alzheimer Dis. Assoc. Disord.,
step and to significantly reduce the elution volume for better 2005, 19, 220–225.

This journal is © The Royal Society of Chemistry 2015 Analyst

 View Article Online

Paper Analyst

4 B. Mollenhauer, H. Esselmann, S. Roeber, W. Schulz- 22 J. Pannee, E. Portelius, M. Oppermann, A. Atkins,

Schaeffer, C. Trenkwalder, M. Bibl, P. Steinacker, M. Hornshaw, I. Zegers, P. Hojrup, L. Minthon, O. Hansson,
H. Kretzschmar, J. Wiltfang and M. Otto, J. Neural Transm., H. Zetterberg, K. Blennow and J. Gobom, J. Alzheimer’s Dis.,
2011, 118, 1261–1262. 2013, 33, 1021–1032.
5 K. Blennow and H. Hampel, Lancet Neurol., 2003, 2, 605–613. 23 M. E. Lame, E. E. Chambers and M. Blatnik, Anal.
6 P. Lewczuk, H. Esselmann, M. Otto, J. M. Maler, Biochem., 2011, 419, 133–139.
A. W. Henkel, M. K. Henkel, O. Eikenberg, C. Antz, 24 N. Ida, T. Hartmann, J. Pantel, J. Schröder, R. Zerfass,
W.-R. Krause, U. Reulbach, J. Kornhuber and J. Wiltfang, H. Förstl, R. Sandbrink, C. L. Masters and K. Beyreuther,
Neurobiol. Aging, 2004, 25, 273–281. J. Biol. Chem., 1996, 271, 22908–22914.
Published on 17 July 2015. Downloaded by UNIVERSITE PARIS SUD on 27/07/2015 14:24:49.

7 N. R. Graff-Radford, J. E. Crook, J. Lucas, et al., Arch. 25 M. Bibl, B. Mollenhauer, H. Esselmann, P. Lewczuk,

Neurol., 2007, 64, 354–362. H.-W. Klafki, K. Sparbier, A. Smirnov, L. Cepek, C. Trenkwalder,
8 V. Welge, O. Fiege, P. Lewczuk, B. Mollenhauer, E. Rüther, J. Kornhuber, M. Otto and J. Wiltfang, J. Brain, 2006,
H. Esselmann, H.-W. Klafki, S. Wolf, C. Trenkwalder, 129, 1177–1187.
M. Otto, J. Kornhuber, J. Wiltfang and M. Bibl, J. Neural 26 U. Haussmann, O. Jahn, P. Linning, C. Janssen, T. Liepold,
Transm., 2009, 116, 203–212. E. Portelius, H. Zetterberg, C. Bauer, J. Schuchhardt,
9 K. Mesbah, R. Verpillot, M. Chiari, A. Pallandre and H.-J. Knoelker, H. Klafki and J. Wiltfang, Anal. Chem., 2013,
M. Taverna, Analyst, 2014, 139, 6547–6555. 85, 8142–8149.
10 R. Verpillot, H. Esselmann, M. R. Mohamadi, H. Klafki, 27 J.-H. Kang, H. Vanderstichele, J. Q. Trojanowski and
F. Poirier, S. Lehnert, M. Otto, J. Wiltfang, J. L. Viovy and L. M. Shaw, Methods, 2012, 56, 484–493.
M. Taverna, Anal. Chem., 2011, 83, 1696–1703. 28 R. Verpillot, M. Otto, H. Klafki and M. Taverna, J. Chroma-
11 J. Wiltfang, H. Esselmann, M. Bibl, A. Smirnov, M. Otto, togr., A, 2008, 1214, 157–164.
S. Paul, B. Schmidt, H. W. Klafki, M. Maler, T. Dyrks, 29 M. R. Mohamadi, Z. Svobodova, R. Verpillot,
M. Bienert, M. Beyermann, E. Rüther and J. Kornhuber, H. Esselmann, J. Wiltfang, M. Otto, M. Taverna, Z. Bilkova
J. Neurochem., 2002, 81, 481–496. and J.-L. Viovy, Anal. Chem., 2010, 82, 7611–7617.
12 J. Wiltfang, H. Esselmann, A. Smirnov, M. Bibl, L. Cepek, 30 Z. Svobodova, M. R. Mohamadi, B. Jankovicova,
P. Steinacker, B. Mollenhauer, K. Buerger, H. Hampel, H. Esselmann, R. Verpillot, M. Otto, M. Taverna,
S. Paul, M. Neumann, M. Maler, I. Zerr, J. Kornhuber, J. Wiltfang, J.-L. Viovy and Z. Bilkova, Biomicrofluidics,
H. A. Kretzschmar, S. Poser and M. Otto, Ann. Neurol., 2012, 6, 024126.
2003, 54, 263–267. 31 J. He, M. Huang, D. Wang, Z. Zhang and G. Li, J. Pharm.
13 M. Bibl, M. Gallus, V. Welge, H. Esselmann, S. Wolf, E. Rüther Biomed. Anal., 2014, 101, 84–101.
and J. Wiltfang, J. Neural Transm., 2012, 119, 805–813. 32 S. Tabnaoui, I. Pereiro, M. Fermigier, S. Descroix, J.-L. Viovy
14 H. Esselmann, T. W. Groemer, J. Kornhuber, P. Lewczuk, and L. Malaquin, 17th International Conference on Minia-
J. M. Maler and J. Wiltfang, Patent WO2011124376A1, turized Systems for Chemistry and Life Sciences, Freiburg,
Application No. PCT/EP2011/001724, February 21, 2013. Germany, 2013, http://www.rsc.org/images/loc/2013/PDFs/
15 J.-H. Kang, M. Korecka, J. B. Toledo, J. Q. Trojanowski and Papers/467_0838.pdf.
L. M. Shaw, Clin. Chem., 2013, 59, 903–916. 33 S. Tabnaoui, PhD Thesis, Curie Institute, Université Pierre
16 S. Lista, H. Zetterberg, B. Dubois, K. Blennow and et Marie Curie - Paris VI, 2013, https://tel.archives-ouvertes.
H. Hampel, J. Neurol., 2014, 261, 1234–1243. fr/tel-00839133/document.
17 P. Gagni, L. Sola, M. Cretich and M. Chiari, Biosens. Bio- 34 K. Perez-Toralla, J. Champ, M. R. Mohamadi, O. Braun,
electron., 2013, 47, 490–495. L. Malaquin, J.-L. Viovy and S. Descroix, Lab Chip, 2013, 13,
18 N. Xia, L. Liu, M. G. Harrington, J. Wang and F. Zhou, Anal. 4409–4418.
Chem., 2010, 82, 10151–10157. 35 E. Portelius, A. J. Tran, U. Andreasson, R. Persson,
19 M. Ammar, C. Smadja, L. G. T. Phuong, M. Azzouz, G. Brinkmalm, H. Zetterberg, K. Blennow and A. Westman-
J. Vignerond, A. Etcheberry, M. Taverna and E. Dufour- Brinkmalm, J. Proteome Res., 2007, 6, 4433–4439.
Gergam, Biosens. Bioelectron., 2013, 40, 329–335. 36 E. Portelius, A. Westman-Brinkmalm, H. Zetterberg and
20 G. Grasso, Mass Spectrom. Rev., 2011, 30, 347–365. K. Blennow, J. Proteome Res., 2006, 5, 1010–1016.
21 Y. Liu, H. Qing and Y. Deng, Int. J. Mol. Sci., 2014, 15, 37 D. Brinet, J. Kaffy, F. Oukacine, S. Glumm, S. Ongeri and
7865–7882. M. Taverna, Electrophoresis, 2014, 35, 3302–3309.

Analyst This journal is © The Royal Society of Chemistry 2015