Development of a microfluidic confocal fluorescence detection system for the hyphenation of nano-LC to on-line biochemical assays.

Abstract

One way to profile complex mixtures for receptor affinity is to couple liquid chromatography (LC) on-line to biochemical detection (BCD). A drawback of this hyphenated screening approach is the relatively high consumption of sample, receptor protein and (fluorescently labeled) tracer ligand. Here, we worked toward minimization of sample and reagent consumption, by coupling nano-LC on-line to a light-emitting diode (LED) based capillary confocal fluorescence detection system capable of on-line BCD with low-flow rates. In this fluorescence detection system, a capillary with an extended light path (bubble cell) was used as a detection cell in order to enhance sensitivity. The technology was applied to a fluorescent enhancement bioassay for the acetylcholine binding protein, a structural analog of the extracellular ligand-binding domain of neuronal nicotinic acetylcholine receptors. In the miniaturized setup, the sensitive and low void volume LED-induced confocal fluorescence detection system operated in flow injection analysis mode allowing the measurement of IC(50) values, which were comparable with those measured by a conventional plate reader bioassay. The current setup uses 50 nL as injection volume with a carrier flow rate of 400 nL/min. Finally, coupling of the detection system to gradient reversed-phase nano-LC allowed analysis of mixtures in order to identify the bioactive compounds present by injecting 10 nL of each mixture.

Full text

Page 1

ORIGINAL PAPER
Development of a microfluidic confocal fluorescence
detection system for the hyphenation of nano-LC to on-line
biochemical assays
Ferry Heus & Martin Giera & Gerdien E. de Kloe & Dick van Iperen & Joost Buijs &
Tariq T. Nahar & August B. Smit & Henk Lingeman & Iwan J. P. de Esch &
Wilfried M. A. Niessen & Hubertus Irth & Jeroen Kool
Received: 8 July 2010 /Revised: 7 September 2010 /Accepted: 8 September 2010 /Published online: 25 September 2010
# The Author(s) 2010. This article is published with open access at Springerlink.com
Abstract One way to profile complex mixtures for receptor
affinity is to couple liquid chromatography (LC) on-line to
biochemical detection (BCD). A drawback ofthishyphenated
screening approach is the relatively high consumption of
sample, receptor protein and (fluorescently labeled) tracer
ligand. Here, we worked toward minimization of sample and
reagent consumption, by coupling nano-LC on-line to a light-
emitting diode (LED) based capillary confocal fluorescence
detection system capable of on-line BCD with low-flowrates.
In this fluorescence detection system, a capillary with an
extended light path (bubble cell) was used as a detection cell
in order to enhance sensitivity. The technology was applied to
a fluorescent enhancement bioassay for the acetylcholine
binding protein, a structuralanalogof the extracellular ligand-
binding domain of neuronal nicotinic acetylcholine receptors.
In the miniaturized setup, the sensitive and low void volume
LED-induced confocal fluorescence detection system operat-
ed in flow injection analysis mode allowing the measurement
of IC50values, which were comparable with those measured
by a conventional plate reader bioassay. The current setup
uses 50 nL as injection volume with a carrier flow rate of
400 nL/min. Finally, coupling of the detection system to
gradient reversed-phase nano-LC allowed analysis of mix-
tures in order to identify the bioactive compounds present by
injecting 10 nL of each mixture.
Keywords Miniaturization.Acetylcholine binding
protein.LED-based confocal fluorescence detection.
Microfluidics.Bioassay
Abbreviations
4-APC4-(2-(Trimethylammonio)ethoxy)benzenaminium
bromide
Acetylcholine binding protein
Biochemical detection
Confocal fluorescence detector
(E)-3-(3-(4-Diethylamino-2-hydroxybenzylidene)-
3,4,5,6-tetrahydropyridin-2-yl)pyridine
Dimethyldichlorosilane
Formic acid
Fluorescence detector
AChBP
BCD
CFD
DAHBA
DMDCS
FA
FD
Electronic supplementary material The online version of this article
(doi:10.1007/s00216-010-4210-x) contains supplementary material,
which is available to authorized users.
F. Heus:M. Giera:J. Buijs:H. Lingeman:W. M. A. Niessen:
H. Irth:J. Kool (*)
BioMolecular Analysis, Department of Chemistry and
Pharmaceutical Sciences, Faculty of Sciences,
VU University Amsterdam,
De Boelelaan 1083,
1081 HV Amsterdam, The Netherlands
e-mail: j.kool@few.vu.nl
G. E. de Kloe:I. J. P. de Esch
Medicinal Chemistry, Department of Chemistry and
Pharmaceutical Sciences, Faculty of Sciences,
VU University Amsterdam,
De Boelelaan 1083,
1081 HV Amsterdam, The Netherlands
D. van Iperen
Mechanical and Electronic Workshops, Faculty of Sciences,
VU University Amsterdam,
De Boelelaan 1083,
1081 HV Amsterdam, The Netherlands
T. T. Nahar:A. B. Smit
Molecular and Cellular Neurobiology, Center for Neurogenomics
and Cognitive Research, Neuroscience Campus,
VU University Amsterdam,
De Boelelaan 1085,
1081 HV Amsterdam, The Netherlands
Anal Bioanal Chem (2010) 398:3023–3032
DOI 10.1007/s00216-010-4210-x

Page 2

FIA
Kd
LED
LC
MeOH
MS
PMT
RBA
Flow injection analysis
Dissociation constant
Light-emitting diode
Liquid chromatography
Methanol
Mass spectrometer
Photomultiplier tube
Radioligand-binding assay
Introduction
In search of bioactive compounds, many exogenous natural
products were demonstrated to have valuable starting
points for medicinal chemistry. For this reason, there is a
continuous effort to capture the available biodiversity, but
for some natural products, the available extracts are very
scarce, and consequently, only limited amounts are avail-
able for analysis. Liquid chromatography (LC) is the
separation workhorse in the process of screening for
bioactive compounds in complex mixtures. After separa-
tion, fractionation, and evaporation, the bioactivity of each
fraction is analyzed. This traditional dereplication process
is usually associated with many bottlenecks as it is
laborious and time consuming, and often multiple iterative
cycles are needed to identify and obtain the pure bioactive
compounds. This requires large amounts of starting
material.
By coupling LC to on-line biochemical detection
(BCD) and mass spectrometry (MS), bioactive mixtures
can be screened more rapidly and comprehensively as
bioactive compounds are simultaneously identified and
unknowns may be structurally elucidated [1–6]. When
applied for the screening of receptor ligands, the LC
effluent is continuously mixed with the assay compounds
consisting of target receptor and a fluorescent tracer ligand
that has an enhanced fluorescence yield in the receptor’s
binding pocket [7, 8]. A fluorescence detector is then used
for the BCD readout. Even though less amount of sample
is needed for successful on-line BCD than for traditional
dereplication processes, in many cases, still substantial
amounts are required. Nano-LC offers a solution to this
problem from the separation’s perspective, but needs the
on-line fluorescence BCD part also to be miniaturized.
As microfluidic technologies utilize highly miniatur-
ized and complex chip-based systems, there is a general
demand for miniaturized detection methods that are
sensitive and selective, in particular also for fluorescence
detection [9]. An example of this is laser-induced
fluorescence detection [10], but costs for adequate lasers
are high and excitation wavelengths are either fixed or
only have a limited tunable wavelength range [11, 12].
LED-induced fluorescence has been utilized in specific
cases for, e.g., cell based analysis [13] and has the
advantage of easily interchangeable LEDs for a wider
range of wavelengths. Other techniques rely on non-flow-
based or non-LC flow-based detection in confocal fluo-
rescence mode [14]. Often, microfluidic chip-based setups
are used for both incubation/chemical reaction steps and
for fluorescence readout [15–17]. A disadvantage in these
setups can be the restriction in freely changing the
incubation unit and detection readout, which can be
overcome by separating the incubation chamber and the
detection cell. As a non-chip-based detection cell, a
stripped fused silica capillary is a good option [18],
although the flow path in narrow capillaries is very
narrow, which limits the sensitivity. Even when operated
in confocal mode, the sensitivity is compromised unless
the confocal optics are of high quality and the positioning
of the confocal point is very accurate, which means that
expensive setups are necessary. Implementation of a
bubble cell capillary is able to overcome this limitation.
This study presents the development of a microfluidic
LED-based confocal fluorescence detection (CFD) system,
with a bubble cell capillary as detection cell, capable of
monitoring fluorescence bioassays in a continuous low-
flow format. In this way, reduction of reagent and sample
consumption in an on-line biochemical analysis format can
be achieved by decreasing the overall flow rates for the
fluorescence-based on-line BCD. An advantage of the new
microfluidic CFD with respect to other low-flow rate
fluorescence detection methods is the adaptation for on-
line BCD at very low-flow rates in combination with nano-
LC. The chromatographic part of the system utilizes nano-
LC separations at sub-μL/min flow rates, which was
coupled on-line to the microfluidic CFD. The detection
system consisted of a LED, excitation and emission filters,
a confocal lens, a photomultiplier tube, and a bubble cell
capillary acting as a detector cell (Fig. 1). To demonstrate
the applicability of the system developed, an on-line BCD
methodology with a fluorescence enhancement bioassay of
acetylcholine binding protein (AChBP) as receptor in
conjunction with the in-house made fluorescent tracer
ligand (E)-3-(3-(4-diethylamino-2-hydroxybenzylidene)-
3,4,5,6-tetrahydropyridin-2-yl)pyridine (DAHBA) [8] was
applied. As a biochemical incubation chamber for the
miniaturized AChBP bioassay, a microfluidic chip was
employed. This chip was similar to a previously developed
chip, which was used for a miniaturized enzymatic on-line
BCD utilizing mass spectrometry as readout [19]. Since
combining flows in the nanoliter per minute range often
leads to laminar flows, special considerations of mixing
procedures are required [20, 21]. Therefore, we used a
microfluidic chip for efficient mixing of the bioassay flows
[19]. The chip provides both efficient mixing of the nano-
3024F. Heus et al.

Page 3

LC effluent and the bioassay carrier flow and does so in a
convenient and robust manner directing both flows to the
CFD. When compared with traditional on-line bioaffinity
systems [8, 22], the current setup not only allowed injection
volumes of approximately a thousand times smaller but also
decreased the overall flow rate of the bioassay ~30 times.
This enabled the use of more convenient 2.5 mL syringes
for continuous-flow bioassay delivery instead of 100 mL
superloops operated by LC pumps.
AChBP is considered an analog for the ligand-binding
domain of neuronal nicotinic acetylcholine receptors, which
are potential pharmaceutical targets for Alzheimer’s dis-
ease, Parkinson’s disease, pain relief, anxiety, epilepsy, and
several cognitive deficits [23–27]. These facts underline the
importance of this target protein for drug discovery
purposes.
Experimental
Chemical and biological reagents
AChBP(from snail species Lymnaea stagnalis) was expressed
from Baculovirus using the pFastbac I vector in Sf9 insect
cells and purified from the medium as described by Celie et
al. [23]. The tracer ligand DAHBA, (E)-3-(3-(4-diethylamino-
2-hydroxybenzylidene)-3,4,5,6-tetrahydropyridin-2-yl)pyri-
dine, was synthesized in-house [8]. KH2PO4, Na2HPO4,
NaCl, trizma base, hydrochloric acid, formic acid (FA),
rhodamine-110, 5% dimethyldichlorosilane (DMDCS) in
toluene, and HPLC grade acetonitrile (99.9%) were purchased
from Sigma-Aldrich, Zwijndrecht, The Netherlands. ELISA
blocking reagent was purchased from Roche, Mannheim,
Germany. LC-MS grade methanol (MeOH; 99.95%) was
purchased from Biosolve, Valkenswaard, The Netherlands.
HPLC grade water was produced using a Milli-Q purification
system, Millipore, Amsterdam, The Netherlands. All ligands
used for evaluating and validating the system were
synthesized by the VU University Medicinal Chemistry
department. The synthetic routes to compounds 1 and 7
were recently published by Kool and de Kloe et al. [8].
The synthetic route of compound 5 was recently published
by Eggink et al. [28], referred to as 4-APC. The binding of
compounds 2, 3, 4, 6, 8, and 9 to AChBP was recently
studied by De Kloe and Retra et al. [29]. The structures of
all compounds including the tracer ligand DAHBA can be
found in Fig. 2.
Instrumentation
The complete system consisted of a nano-LC unit including
a UV detector, a microfluidic chip as biochemical reactor,
and a microfluidic LED-based CFD (Fig. 1).
Nano-LC and solvent delivery system
The Ultimate nano-LC system with a Famos autosampler
and UV detector was from LC Packings, Amsterdam, The
Netherlands. The system was operated at 400 nL/min.
Mobile phase solvent A consisted of water/MeOH 95:5 and
0.1% FA; solvent B of water/MeOH 5:95; and 0.1% FA.
Sample volumes ranging from 10 to 50 nL were injected
into the analytical capillary column (nano-LC; 150 mm×
100 μm i.d.) packed in-house with Waters Xbridge C18
particles (5 μm, 135 Å pore diameter; Waters, Milford,
MA, USA). For mixture analysis, gradient elution was
applied running 15 min isocratic at 0% solvent B, then
rising to 75% solvent B in 45 min. The column exit was
coupled via a low void volume Z-shaped capillary (3 nL
UV-detector cell) to the microfluidic chip and the CFD.
Microfluidic chip
The microfluidic chip and the 4515 chip holder were
produced by Micronit Microfluidics, Enschede, The Nether-
lands, as described earlier [19] and shown schematically in
Fig. 1 with all the interconnecting capillaries. Microchip
ferrules were purchased from Upchurch Scientific, Oak
Harbor, USA. The syringe pump was from Harvard
Apparatus, Holliston, USA. Syringes were purchased from
Hamilton AG, Bonaduz, Switzerland.
A schematic representation of the microfluidic chip is
shown in Fig. 3. The 45×15×2.2 mm chip consisted of
LED
PMT
Bandpass filter
(465 nm)
Lenses
Bandpass filter
(520 nm)
Dichroic mirror
(504 nm)
20× quartz
objective
Bubble cell
capillary (125 µ m)
Gradient
nano-LC pump
UV
detector
Microfluidic chip
Syringe driven by
a syringe pump
Nano column
Auto
injector
Fig. 1 Schematic view of the complete on-line microfluidic CFD
system. The bioassay carrier solution was delivered by a syringe pump
(5 μL/min) and mixed with the nano-LC effluent (400 nL/min) in the
4-μL reaction chamber of the microfluidic chip. A detailed photo-
graphic picture of the microfluidic chip can be found in de Boer et al.
[19]. The outlet of the chip was hyphenated to the bubble cell capillary
where fluorescence was detected by the PMTof the microfluidic CFD.
A detailed description of the system is given in the “Experimental”
section
Microfluidic confocal fluorescence detection system3025

Page 4

two borofloat glass 33/D263 layers with a 457 mm long×
125 μm wide and 70 μm deep open tubular microreactor
of 4 μL, comprising the total volume of the chip’s channel.
The reactor had two inlets and one outlet. One inlet was
used to connect the nano-LC carrier flow; the other inlet
was used to infuse the AChBP and tracer ligand DAHBA
at a flow rate of 5 μL/min by a syringe pump. The chip
outlet was connected to the microfluidic CFD.
All fused silica capillaries used for connections to and
from the chip were purchased from BGB Analytik AG,
Schloßböckelheim, Germany. A 300 mm×20 μm internal
diameter (i.d.) capillary connected the nano-LC system to
the chip. The capillary connecting the syringe pump to the
chip was deactivated by the supplier with 1,3-diphenyl-
1,1,3,3-tetramethyldisilazane and was 1,500 mm×50 μm
i.d. The outlet capillary to the bubble cell of the CFD was
1,000 mm×50 μm i.d. The microchip, the 20 μm i.d.
capillary, and the bubble cell capillary were deactivated with
5% DMDCS in toluene as follows: The chip and the tubing
were flushed with acetonitrile (5 min), 5% DMDCS (5 min),
acetonitrile (5 min), and MeOH (5 min) at a flow rate of
100 μL/min from a syringe pump.
Microfluidic LED-based confocal fluorescence detector
The microfluidic CFD is shown in detail in Fig. 1. The
flow cell of the CFD consisted of a 150 μm i.d. extended
light path “bubble cell” with 50 μm i.d. connecting
capillaries (no. G1600-64232) and was purchased from
Agilent Technologies, Amstelveen, The Netherlands. The
LED (465 nm, 45 mW, 350 mA; no. ELJ-465-211) was
from Roithner Lasertechnik, Vienna, Austria. The quartz
microscope objective (20×0.65, working distance
2.0 mm; no. 03-0202) was from Partec Optics, Münster,
Germany. The photomultiplier tube, incorporated in a
photosensor module (185- to 650 nm spectral response;
no. H9656-03) was from Hamamatsu, Shuzuoka, Japan.
Focusing lenses were from Bernhard Halle Nachfl.
GmbH, Berlin, Germany. The dichroic mirror (506 nm;
no. FF506-Di02-25) and bandpass filters (465 nm and
520 nm; no. FF01-465/30-25 and FF01-520/35-25) were
from Semrock, Rochester, USA.
In the CFD, the bubble cell serves as the actual flow-
through detector cell. Light emanating from the high-
intensity LED passed through the 465 nm single bandpass
filter, after which it was collimated by a lens and reflected
under 90° by the dichroic mirror. The filtered light was
subsequently focused into the center of the 150 μm bubble
cell by the 20× quartz microscope objective. Emitted light
then passed the same dichroic mirror, a focusing lens, and
the 520 nm single bandpass filter and was subsequently
detected by the photomultiplier tube. The LASER (473 nM,
10 mW, DPSS RLTMBL) used in the initial setup was also
from Roithner Lasertechnik.
+
2
3
4
1
+
AChBP + DAHBA
from syringe
Ligand eluting
from nano-LC
Ligand eluting from nano-LC displacing
DAHBA from AChBP binding pocket
Fluorescence detectio n
(by CFD) of AChBP +
DAHBA complex
Displacement of DAHBA
results in a decreased
fluorescence output
and
Fig. 3 The microfluidic chip, which comprises the bioaffinity
analysis part of the system. AChBP and DAHBA are infused to the
chip at 5 μL/min (1) together with the nano-LC effluent containing
potential ligands (2). In the 4 μL open tubular microreactor of the
chip, DAHBA is displaced by eluting ligands from the analytical
column (3) resulting in a decrease in fluorescence (4) as detected by
the microfluidic CFD
N
N
HN
Compound 1
O
H
N
N
N
Br
Compound 2
Compound 3
NH3+
O
N+
Br-
Br-
Compound 5 Compound 4
Compound 6
N
N
N
N
N
OH
Compound 7Compound 8Compound 9
DAHBA
O
H
N
N
N
N
H
N
H
N
O
N
O
H
N
N
N
O
N
N
N
OH
Br
N
N
N
O
O
O
O
HH
Fig. 2 Structures of the com-
pounds/ligands used for valida-
tion and evaluation of the on-line
microfluidic CFD bioaffinity
system
3026 F. Heus et al.

Page 5

Hardware optimization and chemical evaluation
Signal-to-noise determinations of the microfluidic CFD
at different settings
At each photomultiplier tube (PMT) control voltage (8.0 to
10.0 V at 0.5 V increments tested) and four different LED
input currents (from 200 to 350 mA at 50 mA increments),
baselines were recorded at a frequency of 10 Hz for 1 min,
while infusing a water/MeOH 50:50 solution to the micro-
fluidic CFD with the syringe pump operated at 5 μL/min.
The noise of the CFD system at each specific condition was
calculated as two times the standard deviation of the 600
data points measured. In order to obtain a fluorescence
signal, the above-described measurements were repeated
with infusion of rhodamine-110 at a concentration of
500 pM. The measured 600 data points at each specific
condition were averaged. Thus, the signal-to-noise ratio for
500 pM rhodamine-110 at five different PMT control
voltages and four LED input currents were calculated.
Linearity analysis of the microfluidic CFD
To evaluate linearity at the 20 conditions tested as described
for the signal-to-noise calculation, rhodamine-110 solutions
were infused with concentrations of 0, 0.5, 1.0, 2.5, 5.0,
10.0, and 25 nM in water/MeOH 50:50. The resulting
rhodamine-110 baselines were averaged for data analysis.
Determination of IC50values with the microfluidic CFD
operated as on-line BCD
Nano-column based flow injection analysis (FIA) in a
concentration–response manner allowed the determination
of IC50values of test compounds. These analyses were run
in isocratic LC mode with 0% to 50% B in the mobile
phase, depending on the lipophilicity of each investigated
compound. Each compound eluted from the nano-column
without retention.
AChBP biochemical bioassay
Fresh solutions of L. stagnalis Acetylcholine binding
protein (AChBP) and the fluorescent tracer ligand DAHBA
were made every day by dissolving 1 μg/mL AChBP and
40 nM DAHBA in the bioassay solution containing: 1 mM
KH2PO4, 3 mM Na2HPO4, 0.16 mM NaCl, 20 mM trizma
base/HCl at pH 7.5 and 400 μg/mL ELISA blocking
reagent. The bioassay solution was housed in a 2.5 mL
syringe. During the experiments, the AChBP/DAHBA
bioassay solution in the syringe was kept at 4 °C. The on-
line bioassay itself was performed at 22 °C. The syringe
pump was operated at a flow rate of 5 μL/min. The
resulting bioassay flow was mixed with the eluent from the
nano-LC. Continuous in-flow incubation was allowed in
the microfluidic chip and then followed by fluorescence
detection with the microfluidic CFD.
Results and discussion
Evaluation of the on-line microfluidic confocal
fluorescence detection system
Hardware setup
The microfluidic CFD setup initially contained a laser as
light source. Although the light beam from the laser tested
(from Roithner Lasertechnik, 473 nM, 10 mW, DPSS
RLTMBL) was more focused, and therefore more powerful
than the LED used in the rest of this study, it also had
serious drawbacks. Besides the higher costs, lasers as a
light source impose restrictions as they only allow one
wavelength (or a limited range) per laser to be used.
Furthermore, baseline stability and reproducibility of the
tested laser were not adequate for our purposes. Due to
these drawbacks, the LED clearly surpasses the tested laser
as LEDs are cheap and available in a large variety of
wavelengths. More importantly, they were found to produce
a more stable and reproducible fluorescence detection
system for our purposes. Therefore, a LED was incorpo-
rated into the system together with an additional collimat-
ing lens to focus the light of the LED. In order to evaluate
and optimize sensitivity and performance of the detection,
capillaries with 20, 50, 100, and 150 μm i.d. as well as two
types of bubble cell capillaries were evaluated as detection
cell by installing the capillaries into the CFD, followed by
infusion of a rhodamine-110 solution. In these experiments,
an excitation wavelength of 465 nm was applied. A
detection window was created in the tested capillaries
(except the bubble cell capillaries) by locally removing the
polyimide coating. The 100 and 150 μm i.d. capillaries
yielded higher fluorescence signals than the 20 or 50 μm i.d.
capillaries, which rendered the fluorescence system rather
insensitive due to the very limited light path. An evident
drawback of the two larger i.d. capillaries as detector cells
was the increased dead volume, which imposed problems
upon implementation into the miniaturized flow system.
Besides sensitivity issues, the 20 μm capillary was also very
susceptible to clogging; therefore, capillaries with an i.d.
smaller than 50 μm i.d. were not evaluated further. Next, two
bubble cell capillaries [30] were evaluated. Both 50 and
75 μm i.d. capillaries (with a 150 and 200 μm bubble cell,
respectively) were tested. Although the 75 μm i.d. capillary
showed a slightly higher signal-to-noise ratio than the
50 μm i.d. capillary, the 50 μm i.d. bubble cell capillary
Microfluidic confocal fluorescence detection system 3027

Page 6

was finally incorporated in the system. The 50 μm i.d.
capillary part did not significantly increase the system dead
volume (and thus band broadening), while the advantage of
the bubble cell to provide an extended light path yielded a
fluorescence readout with a 5.6 times increased sensitivity
compared with the 50 μm i.d. capillary without a bubble
cell. This was in good agreement with the six times’
increase in sensitivity as given by the supplier.
Electronic settings (LED intensity and PMT control
voltage)
The final microfluidic CFD hardware setup was subse-
quently evaluated by measuring the fluorescence of a blank
and a 500 pM rhodamine-110 solution at 20 different
combinations of LED and PMT settings. The results
obtained are shown in Fig. S1 (Electronic supplementary
material). The optimal signal-to-noise ratio was found at a
power input to the LED of 350 mA and the control voltage
of the PMT set at a value of 10 V, which were both the
maximal settings. The signal-to-noise ratio of 500 pM
rhodamine-110 was found to be 23 at these settings. Next,
the linearity was evaluated by infusing different concen-
trations (0.5 to 25 nM) of rhodamine-110 at the optimal
settings. From these results (Fig. S2: Electronic supple-
mentary material), it was concluded that the system showed
a high linearity at the optimal settings within the concen-
tration range tested. These results also showed a dynamic
range from 0.5 to 25 nM rhodamine-110 with an R2of
0.9998. Concentrations higher than 25 nM showed similar
linearity, but could only be measured by decreasing the
voltage to the PMT.
Biological evaluation and validation of the microfluidic
confocal fluorescence detection system operated in on-line
biochemical detection mode
Initial biochemical evaluation
The miniaturized on-line bioaffinity analysis part of the
system is shown in Fig. 3. DAHBA and AChBP are
continuously infused into the chip where the bioassay
carrier solution is mixed continuously with the nano-LC
eluent. Injected compounds elute from the nano-LC column
to the chip where they are allowed to incubate with the
miniaturized bioassay. In the incubation chamber of the
chip, ligands displace DAHBA from the AChBP binding
pocket. As this displacement results in a decreased
fluorescence yield from DAHBA, it is detected by the
microfluidic CFD as a negative peak in the on-line bioassay
trace. The concentration (of 1 μg/mL) AChBP was chosen
to allow a good assay window while maintaining optimal
bioassay sensitivity. The concentration of tracer DAHBA
used (40 nM) was slightly higher than its dissociation
constant (Kd) [8]. This is commonly done for ligand-
binding assays. The optimal flow rate of the bioassay
carrier solution was determined by varying the flow rate
with incremental steps of 1 μL/min, starting at 1 μL/min. A
stable baseline was observed at a minimal flow rate of
5 μL/min. At this flow rate, the syringe pump generated no
pulses, nor did the 75% MeOH and 0.1% FA in the nano-
LC effluent have any adverse effects on the bioassay.
Higher flow rates of the bioassay carrier solution had a
negative effect on the bioassay sensitivity because of
dilution effects. During the initial biochemical evaluation
and after test injections of AChBP ligands, a decline in
assay sensitivity was observed. This was due to the
adsorption of AChBP to the wall of the chip and bubble
cell capillary, causing clogs, tailing effects, and an
increased baseline. Therefore, a DMDCS deactivation
procedure was applied to the chip, connecting capillaries
and the bubble cell. With deactivated capillaries, no
significant (increases in) band broadening were observed
during a time period of 3 months. Also, a more stable
baseline was observed and no clogging occurred anymore.
Biochemical validation
Nicotine, a typical ligand for AChBP, was used to evaluate
intraday and interday repeatability of the complete system.
Nicotine was injected at increasing concentrations, with a
nano-LC carrier flow containing 0% eluent B, which
allowed nicotine to elute from the column non-retained. A
typical analysis result is shown in Fig. 4. The figure shows
the displacement profile of 50 nL of nicotine injected in
triplicate at 12 different concentrations in a serial dilution
manner with a factor of two. Injecting different concen-
trations of nicotine showed negative peaks of which the
negative peak heights represented the percentage of
DAHBA displaced in the bioassay. Over 99% tracer
DAHBA displacement by nicotine occurred at a concentra-
tion of 10 μM of nicotine in the bioassay. At full DAHBA
displacement, injecting higher nicotine concentrations
evidently did not yield higher negative signals, but only
broader signals due to tailing effects. The figure also
indicates the excellent repeatability for each concentration
injected.
The intraday repeatability was determined by measuring
multiple IC50curves during one measuring day using the
same batch of bioassay constituents. Because ligands
injected are diluted in the microfluidic BCD system, a
dilution factor has to be used to correlate DAHBA
displacements with the actual ligand concentrations in the
on-line BCD part. Therefore, a dilution factor was
calculated in order to determine IC50 values. This was
done according to a mathematical analytical formula as
3028 F. Heus et al.

Page 7

presented previously [22, 31] (see also the Electronic
supplementary material for the formula used and further
explanation) and resulted in a dilution factor of 243, which
was used further. Table 1 shows the analytical results.
These results demonstrate that the system operated in a very
repeatable manner (RSD=1.5%). Furthermore, the average
IC50for nicotine as acquired from the IC50curve shown in
Fig. 5 (6.41), correlated well with results obtained by
traditional biochemical assay formats (6.40; from Kool et
al. [8]). The bioassay window was determined as the largest
negative peak obtainable and was measured by injecting a
very high concentration of nicotine. The bioassay window
did not change markedly over a full measuring day. As
seen in Fig. 4, representing a typical measuring result, a
small baseline drift (of only 8% in this case) was observed
during 7 h.
The interday reproducibility was tested in essentially the
same way as the intraday repeatability, but in this case, one
IC50curve was measured per day with every day a fresh
solution of bioassay constituents for 3 days. These results
are depicted in Table 1 and Fig. S3 (Electronic supplemen-
tary material) and clearly show the good reproducibility of
the system (RSD=2.3%). This was also the case for the
bioassay window that did not show marked differences over
the days; an average assay window (maximum obtainable
S/N ratio) of 47.2±1.7 was found. As the average IC50
found (IC50 of 6.38±0.02) for nicotine (see Table 1)
correlated well with results obtained by a traditional
radioligand-binding assay (RBA; IC50of 6.40±0.01) [8]
and low standard deviations were obtained, it was conclud-
ed that the methodology was not only accurate but also
precise.
The results from both the repeatability as well as the
reproducibility experiments allowed us to further validate
the system by analyzing several ligands for a pharmaco-
logical comparison with traditional methodologies.
Pharmacological validation
To evaluate the biological sensitivity and validate the
system pharmacologically, IC50values of different ligands
were measured the same way as done for the intraday
repeatability and then compared with results from a
traditional RBA. For this, five ligands including nicotine
were analyzed. The structures of the ligands used can be
found in Fig. 2. These quinuclidine-containing ligands were
chosen because their binding to AChBP has been properly
characterized [29]. After analysis of the ligands with both
FIA on-line microfluidic BCD and with a traditional RBA,
IC50curves were constructed (see Fig. 5 for the on-line
microfluidic BCD FIA measurements). Subsequently, a
correlation plot was constructed by plotting the CFD pKi
values to the pKivalues of a traditional RBA. The R2of this
plot then rates the applicability of the system to assess
ligand binding. The model equation derived from Fig. 6
was y=0.696x+2.1455 with an R2=0.9651 (Fig. 6). The
correlation between results from our microfluidic system
100 200 300 400
100
150
200
IC50 = 392 nM
time (min)
fluorescence intensity [465-520]
Fig. 4 Triplicate injections (50 nL) of increasing concentrations
(100 nM to 10 mM, in-vial concentration) of the ligand nicotine in
FIA mode. The observed negative peak heights were plotted against
the concentrations inside the flow system to obtain an IC50curve; see
Fig. 5 for the IC50curve of nicotine
Table 1 Pharmacological validation by evaluation of the intraday repeatability and interday reproducibility
Parameter1 (pIC50)R2
2 (pIC50)R2
3 (pIC50)R2
Intradaya
Interdayb
6.41±0.02
6.07±0.01
0.9973
0.9989
6.55±0.02
6.35±0.01
0.9932
0.9988
n.d.
6.21±0.02
n.d.
0.9973
The numbers 1, 2, and 3 signify the respective measurements
aThe intraday repeatability was investigated by determining the IC50value of nicotine two times in triplicate within 24 h
bFor the interday repeatability, the IC50value of nicotine was determined three times in triplicate over the course of 3 days
Microfluidic confocal fluorescence detection system3029

Page 8

and traditional RBA demonstrated the excellent applicabil-
ity of the microfluidic BCD system for IC50determination
of ligands. The system also proved to be very sensitive; the
LOD of compound 1 (pKiof 8.71±0.03) for example was
412 pM. In addition, compound 5 with a choline moiety
that we used as derivatization reagent in previous work [28]
was also investigated with the complete system. This was
done because it was expected that this compound also had
affinity toward the AChBP as the choline moiety represents
an important pharmacophoric part for AChBP binding
pocket. Indeed, compound 5 showed a low affinity (pKi
of 5.45±0.01) toward AChBP.
Proof-of-principle for mixture analysis of the microfluidic
confocal fluorescence system
To show the potential of the microfluidic CFD for
bioaffinity analysis of mixtures toward the AChBP, a
mixture of eight ligands was separated by nano-LC
followed by biochemical analysis with the miniaturized
BCD format. A typical result obtained is shown in Fig. 7.
For analysis of the mixture shown, only 10 nL of mixture
was injected. The upper trace shows an UV readout at
254 nm of the eight ligands injected (100 pmol injected).
The lower trace shows the simultaneously obtained bio-
affinity trace. This trace clearly shows displacement of the
DAHBA tracer ligand from the AChBP binding pocket
upon eluting ligands, yielding negative peaks indicative of
the total bioaffinity per ligand separated. The miniaturized
system shows a very stable baseline, while bioaffinity peaks
are obtained with very low band broadening effects. The
system’s stability and low band broadening in the UV trace
as well as in the bioassay enable a clear distinction between
co-eluting compounds as can be seen for compounds 3, 7
and 8. The UV trace in Fig. 7 shows an excellent separation
for these closely related analogs. These results also indicate
that the miniaturized system could be used to establish
6789
6
7
8
9
1
2
4
nicotine
3
RBA [pKi]
on-line bioaffinity analysis [pKi]
Fig. 6 Correlation plot of five typical ligands analyzed with both the
traditional RBA and the on-line microfluidic BCD. An R2of 0.9651
was obtained
-10-9-8
log [M] ligand
-7-6-5-4
0
50
100
nicotine
4
5
2
3
1
relative fluorscence [%]
Fig. 5 IC50curves of six AChBP ligands. All ligands were injected in
triplicate with an injection volume of 50 nL
204060
time (min)
46
3 + 7
82
9
4
6
3 + 7
82
91
1
fluorescence intensity
[465-520]
absorbance 220 nm
Fig. 7 On-line microfluidic BCD and UV analysis of a mixture
containing eight AChBP ligands (10 nL injected). The upper trace
shows the UV signal of the ligands. The lower trace shows the
fluorescence bioaffinity signal, as obtained by the microfluidic CFD.
Structures of the ligands are shown in Fig. 2
3030 F. Heus et al.

Page 9

structure–activity relationships, for this implies the evalu-
ation of different substituents on one scaffold.
Both the IC50measurements of pure compounds as well as
the application of bioactive mixture analysis demonstrate that
this miniaturized system is comparable in terms of sensitivity
and stability with “non-miniaturized” on-line BCD systems
where much higher flow rates and injection volumes are used
[7, 8, 31]. Here, the additional advantage of our new
miniaturized BCD technology is the use of very low sample
volumes in a fluorescence-based detection setup. This renders
the setup suitable for analysis of very scarce and/or precious
samples, which is a topic of our current research efforts.
Conclusion
A microfluidic CFD was developed and successfully
coupled to a miniaturized continuous-flow bioassay incor-
porating a nano-LC system as the separation technology.
The implementation of a LED and a DMDCS deactivation
procedure made the system stable, rugged, and very
versatile for, e.g., on-line biochemical screening utilizing
fluorescent tracer ligands with different fluorescence prop-
erties. Additional sensitivity was obtained by the imple-
mentation of a bubble cell capillary as detection cell. The
presented system not only showed a very good stability and
robustness but also decreased sample consumption a
thousand fold as compared with conventional on-line
systems. By miniaturizing both the overall flow rates of
the system from the micro- to the nano-scale and the
dimensions used in the flow system, a miniaturized
continuous-flow system was obtained. Here, the enormous
reduction in consumption of sample and biochemical
reagents provides for a larger range in possible drug targets
for on-line bioactivity screening, which are for example
difficult to produce or obtain in larger quantities. Further-
more, the system allows chemical and simultaneous
biochemical screening of scarce sample types only obtain-
able in very low volumes, such as precious natural extracts,
animal venoms, cell, and tissue lysates. By miniaturizing,
no stability and sensitivity were sacrificed when compared
with conventional on-line systems.
Acknowledgements
grant no. 700.56.043 from the Dutch Scientific Society NWO. Martin
Giera was supported by the German Academic Exchange Service
(DAAD). Gerdien E. de Kloe and Tariq T. Nahar were financially
supported within the framework of Top Institute Pharma no. D2-103.
Ansgar Kretschmer is thanked for suggesting 4-APC as a possible
AChBP ligand.
This work was financially supported by ECHO
Open Access
Creative Commons Attribution Noncommercial License which per-
mits any noncommercial use, distribution, and reproduction in any
medium, provided the original author(s) and source are credited.
This article is distributed under the terms of the
References
1. Johnson BM, Nikolic D, van Breemen RB (2002) Applications of
pulsed ultrafiltration–mass spectrometry. Mass Spectrom Rev 21
(2):76–86
2. Sun Y, Gu C, Liu X, Liang W, Yao P, Bolton JL, van Breemen RB
(2005) Ultrafiltration tandem mass spectrometry of estrogens for
characterization of structure and affinity for human estrogen
receptors. J Am Soc Mass Spectrom 16(2):271–279
3. Niessen KV, Hofner G, Wanner KT (2005) Competitive ms
binding assays for dopamine d2 receptors employing spiperone as
a native marker. Chembiochem 6(10):1769–1775
4. Annis DA, Nazef N, Chuang CC, Scott MP, Nash HM (2004) A
general technique to rank protein–ligand binding affinities and
determine allosteric versus direct binding site competition in
compound mixtures. J Am Chem Soc 126(47):15495–15503
5. Kool J, Eggink M, van Rossum H, van Liempd SM, van Elswijk
DA, Irth H, Commandeur JN, Meerman JH, Vermeulen NP (2007)
Online biochemical detection of glutathione-s-transferase p1-
specific inhibitors in complex mixtures. J Biomol Screen 12
(3):396–405
6. Kool J, Van Liempd SM, Harmsen S, Schenk T, Irth H,
Commandeur JN, Vermeulen NP (2007) An on-line post-column
detection system for the detection of reactive-oxygen-species-
producing compounds and antioxidants in mixtures. Anal Bioanal
Chem 388(4):871–879
7. Oosterkamp AJ, Villaverde Herraiz MT, Irth H, Tjaden UR, van
der Greef J (1996) Reversed-phase liquid chromatography
coupled on-line to receptor affinity detection based on the human
estrogen receptor. Anal Chem 68(7):1201–1206
8. Kool J, de Kloe GE, Bruyneel B, de Vlieger JS, Retra K,
Wijtmans M, van Elk R, Smit AB, Leurs R, Lingeman H, de Esch
IJ, Irth H (2010) Online fluorescence enhancement assay for the
acetylcholine binding protein with parallel mass spectrometric
identification. J Med Chem 53(12):4720–4730
9. Gotz S, Karst U (2007) Recent developments in optical detection
methods for microchip separations. Anal Bioanal Chem 387
(1):183–192
10. Laing TD, Marenco AJ, Moore DM, Moore GJ, Mah DC, Lee
WE (2006) Capillary electrophoresis laser-induced fluorescence
for screening combinatorial peptide libraries in assays of
botulinum neurotoxin a. J Chromatogr B Analyt Technol Biomed
Life Sci 843(2):240–246
11. Kolivoska V, Weiss VU, Kremser L, Gas B, Blaas D, Kenndler E
(2007) Electrophoresis on a microfluidic chip for analysis of
fluorescence-labeled human rhinovirus. Electrophoresis 28
(24):4734–4740
12. Yu L, Shen Z, Mo J, Dong X, Qin J, Lin B (2007) Microfluidic
chip-based cell electrophoresis with multipoint laser-induced
fluorescence detection system. Electrophoresis 28(24):4741–4747
13. Tatosian DA, Shuler ML, Kim D (2005) Portable in situ
fluorescence cytometry of microscale cell-based assays. Opt Lett
30(13):1689–1691
14. Sung JH, Kam C, Shuler ML (2010) A microfluidic device for a
pharmacokinetic–pharmacodynamic (pk–pd) model on a chip.
Lab Chip 10(4):446–455
15. Hadd AG, Raymond DE, Halliwell JW, Jacobson SC, Ramsey JM
(1997) Microchip device for performing enzyme assays. Anal
Chem 69(17):3407–3412
16. Kong J, Jiang L, Su X, Qin J, Du Y, Lin B (2009) Integrated
microfluidic immunoassay for the rapid determination of clenbu-
terol. Lab Chip 9(11):1541–1547
17. Fonseca A, Raimundo IM Jr, Rohwedder JJ, Lima RS, Araujo MC
(2010) A microfluidic device with integrated fluorimetric detection
for flow injection analysis. Anal Bioanal Chem 396(2):715–723
Microfluidic confocal fluorescence detection system3031

Page 10

18. Du WB, Fang Q, Fang ZL (2006) Microfluidic sequential
injection analysis in a short capillary. Anal Chem 78(18):6404–
6410
19. de Boer AR, Bruyneel B, Krabbe JG, Lingeman H, Niessen WM,
Irth H (2005) A microfluidic-based enzymatic assay for bioactiv-
ity screening combined with capillary liquid chromatography and
mass spectrometry. Lab Chip 5(11):1286–1292
20. Auroux PA, Iossifidis D, Reyes DR, Manz A (2002) Micro total
analysis systems. 2. Analytical standard operations and applica-
tions. Anal Chem 74(12):2637–2652
21. Burke BJ, Regnier FE (2003) Stopped-flow enzyme assays on
a chip using a microfabricated mixer. Anal Chem 75(8):1786–
1791
22. Falck D, de Vlieger JS, Niessen WM, Kool J, Honing M, Giera
M, Irth H (2010) Development of an online p38alpha mitogen-
activated protein kinase binding assay and integration of LC-HR-
MS. Anal Bioanal Chem. doi:10.1007/s00216-010-4087-8
23. Celie PH, Kasheverov IE, Mordvintsev DY, Hogg RC, van Nierop
P, van Elk R, van Rossum-Fikkert SE, Zhmak MN, Bertrand D,
Tsetlin V, Sixma TK, Smit AB (2005) Crystal structure of
nicotinic acetylcholine receptor homolog achbp in complex with
an alpha-conotoxin pnia variant. Nat Struct Mol Biol 12(7):582–
588
24. Dutertre S, Ulens C, Buttner R, Fish A, van Elk R, Kendel Y,
Hopping G, Alewood PF, Schroeder C, Nicke A, Smit AB, Sixma
TK, Lewis RJ (2007) Achbp-targeted alpha-conotoxin correlates
distinct binding orientations with nachr subtype selectivity.
EMBO J 26(16):3858–3867
25. Hogg RC, Bertrand D (2004) Nicotinic acetylcholine receptors as
drug targets. Curr Drug Targets CNS Neurol Disord 3(2):123–130
26. Hogg RC, Raggenbass M, Bertrand D (2003) Nicotinic acetyl-
choline receptors: from structure to brain function. Rev Physiol
Biochem Pharmacol 147:1–46
27. Smit AB, Syed NI, Schaap D, van Minnen J, Klumperman J, Kits
KS, Lodder H, van der Schors RC, van Elk R, Sorgedrager B,
Brejc K, Sixma TK, Geraerts WP (2001) A glia-derived
acetylcholine-binding protein that modulates synaptic transmis-
sion. Nature 411(6835):261–268
28. Eggink M, Wijtmans M, Ekkebus R, Lingeman H, de Esch IJ, Kool
J, Niessen WM, Irth H (2008) Development of a selective esi-ms
derivatization reagent: synthesis and optimization for the analysis of
aldehydes in biological mixtures. Anal Chem 80(23):9042–9051
29. de Kloe GE, Retra R, Geitmann M, Källblad P, Nahar T, van Elk
R, Smit AB, van Muijlwijk-Koezen JE, Leurs R, Irth H,
Danielson UH, de Esch IJP (2010) Surface plasmon resonance
biosensor based fragment screening using acetylcholine binding
protein identifies ligand efficiency hot spots (le hot spots) by
deconstruction of nicotinic acetylcholine receptor α7 ligands. J
Med Chem. doi:10.1021/jm100834y
30. Hempel G (2000) Strategies to improve the sensitivity in capillary
electrophoresis for the analysis of drugs in biological fluids.
Electrophoresis 21(4):691–698
31. Schebb NH, Heus F, Saenger T, Karst U, Irth H, Kool J (2008)
Development of a countergradient parking system for gradient
liquid chromatography with online biochemical detection of serine
protease inhibitors. Anal Chem 80(17):6764–6772
3032 F. Heus et al.