Anal. Chem. 2010, 82, 49–52
Letters to Analytical Chemistry
Monolithic Silicon Chip for Immunofluorescence
Detection on Single Magnetic Beads
Emile P. Dupont,*,† Estelle Labonne,‡,§ Caroline Vandevyver,| Ulrike Lehmann,†,⊥
Edoardo Charbon,‡,X and Martin A. M. Gijs†
Laboratory of Microsystems, School of Computer and Communication Sciences, and Research Commission
EPFL-SNF, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland
While fluorescence detection is widely used for bioassays
owing to its high sensitivity, a complete fluorescent microscopy
setup, comprised of a light source, optical filters, a microscope
body, and a camera, still is bulky equipment, compromising
its use in a point-of-care environment. Here we propose an
integrated monolithic silicon chip for integrated magnetic
manipulation and optical detection of fluorescently labeled
magnetic beads. Our approach permits microscopeless measurement of the fluorescence of a single microparticle. We
demonstrate the viability of this approach by the detection of
cancer biomarker 5D10 monoclonal antibodies (mAbs) in a
noncompetitive sandwich immunoassay performed on the
surface of magnetic beads, in a phosphate buffered salinebovine serum albumin (PBS-BSA) solution, with a detection
limit of 1 ng mL-1.
A microtiter plate-based enzyme-linked immunosorbent assay (ELISA)
exploits the sensitivity and specificity of the interaction between a target
antigen (Ag), its specific capture antibody (Ab), and detection Ab to
form a traceable sandwich immunocomplex. Such a test typically takes
a few hours and is based on either colorimetric or fluorescent detection
principles. Miniaturized microchannel-based bioanalysis systems offer
the advantage of strong analysis time reduction, while consuming only
minute quantities of samples and reagents (typically a few microliters).1-4
Moreover, the use in such systems of micro- or nanoparticles (“beads”)
that can be functionalized using a wide range of available surface
chemistries prevents the need for functionalization of the micro* To whom correspondence should be addressed. E. P. Dupont, Laboratory
of Microsystems, Ecole Polytechnique Fédérale de Lausanne (EPFL), STILMIS2-BM 3135-Station 17, CH-1015 Lausanne (Switzerland). Fax: (+41)216935950.
E-mail: emile.dupont@epfl.ch.
†
Laboratory of Microsystems.
‡
School of Computer and Communication Sciences.
§
Current address: CEA-LETI, 38054 Grenoble, France.
|
Research Commission EPFL-SNF.
⊥
Current address: Microsens SA, 2002 Neuchâtel, Switzerland.
X
Current address: Circuits and Systems Group, Faculty of EEMCS, Delft
University of Technology, Delft, The Netherlands.
(1) Auroux, P. A.; Iossifidis, D.; Reyes, D. R.; Manz, A. Anal. Chem. 2002, 74,
2637–2652.
(2) Li, D. Encyclopedia of Microfluidics and Nanofluidics; Springer: New York,
2008.
(3) Vilkner, T.; Janasek, D.; Manz, A. Anal. Chem. 2004, 76, 3373–3385.
(4) West, J.; Becker, M.; Tombrink, S.; Manz, A. Anal. Chem. 2008, 80, 4403–
4419.
10.1021/ac902241j 2010 American Chemical Society
Published on Web 11/13/2009
channels.5-9 Magnetic beads, in addition, can be easily magnetically
actuated and recovered, explaining why they are increasingly employed
as a solid support in bioanalysis.10-14 Several quantitative magnetic beadbased immunoassays using optical or fluorescent detection have been
presented in a miniaturized format.15-17 In all of these approaches, the
use of a high-end (fluorescent) microscope detection platform was key
for achieving a good system performance and high detection sensitivity.
Few examples of successful integration of optical detection and microfluidics can be found in the literature. On-chip optical imaging was
demonstrated by Heng et al.18 through the use of CCD sensors
underneath a microfluidic channel. Fluorescence detection was achieved
on-chip by Balslev et al.,19 integrating a laser source, waveguides, and
photodiodes in a microsystem. However, complete integration of
optoelectronics and microfluidics in the same disposable device may
not be the most cost-effective option.20
We report here a complementary metal-oxide-semiconductor
(CMOS) chip for detecting in a microfluidic channel fluorescent
immunocomplexes located on the surface of magnetic beads. Our
approach is not based on a fluorescence microscope detection
platform but on the measurement of photon-induced electrical current
(5) Haes, A. J.; Terray, A.; Collins, G. E. Anal. Chem. 2006, 78, 8412–8420.
(6) Hong, J. W.; Studer, V.; Hang, G.; Anderson, W. F.; Quake, S. R. Nat.
Biotechnol. 2004, 22, 435–439.
(7) Kurita, R.; Yokota, Y.; Sato, Y.; Mizutani, F.; Niwa, O. Anal. Chem. 2006,
78, 5525–5531.
(8) Lim, C. T.; Zhang, Y. Biosens. Bioelectron. 2007, 22, 1197–1204.
(9) Verpoorte, E. Lab Chip 2003, 3, 60N–68N.
(10) Doyle, P. S.; Bibette, J.; Bancaud, A.; Viovy, J. L. Science 2002, 295, 2237–
2237.
(11) Dreyfus, R.; Baudry, J.; Roper, M. L.; Fermigier, M.; Stone, H. A.; Bibette,
J. Nature 2005, 437, 862–865.
(12) Gijs, M. A. M. Microfluid. Nanofluid. 2004, 1, 22–40.
(13) Lehmann, U.; Hadjidj, S.; Parashar, V. K.; Vandevyver, C.; Rida, A.; Gijs,
M. A. M. Sens. Actuators, B: Chem. 2006, 117, 457–463.
(14) Pamme, N. Lab Chip 2006, 6, 24–38.
(15) Choi, J. W.; Oh, K. W.; Thomas, J. H.; Heineman, W. R.; Halsall, H. B.;
Nevin, J. H.; Helmicki, A. J.; Henderson, H. T.; Ahn, C. H. Lab Chip 2002,
2, 27–30.
(16) Kim, K. S.; Park, J. K. Lab Chip 2005, 5, 657–664.
(17) Lacharme, F.; Vandevyver, C.; Gijs, M. A. M. Anal. Chem. 2008, 80, 2905–
2910.
(18) Heng, X.; Erickson, D.; Baugh, L. R.; Yaqoob, Z.; Sternberg, P. W.; Psaltis,
D.; Yang, C. H. Lab Chip 2006, 6, 1274–1276.
(19) Balslev, S.; Jorgensen, A. M.; Bilenberg, B.; Mogensen, K. B.; Snakenborg,
D.; Geschke, O.; Kutter, J. P.; Kristensen, A. Lab Chip 2006, 6, 213–217.
(20) Myers, F. B.; Lee, L. P. Lab Chip 2008, 8, 2015–2031.
Analytical Chemistry, Vol. 82, No. 1, January 1, 2010
49
Figure 1. (a) Micrograph of the CMOS chip, showing the top layer of
octagonal microcoils. Insert: zoom on one of the SPAD detectors in the
center of a microcoil. (b) Cross-section view of the chip with the
superposed PDMS microfluidic cartridge with an indication of the dimensions. (c) Schematic view and scheme of the SPAD structure with its
electrical control circuit. (d) Applying current in a microcoil causes an
attractive magnetic force to the nearby magnetic beads, whereby a single
magnetic bead is positioned over the SPAD active area.
pulses in single photon avalanche diodes (SPADs)21 integrated in
the center of magnetic actuation microcoils on a silicon chip (see
Figure 1a,b) (see Supporting Information, CMOS chip design and
fabrication). We profit from the advantageous signal-to-noise ratio
and integration properties of a SPAD, which is a p-n junction reverse
biased above the breakdown voltage Vbd by an excess bias voltage
Ve ) Vdd - Vbd. The electrical configuration, schematically shown
in Figure 1c, causes the optical gain to become virtually infinite,
thus making single photon detection possible. When a photon is
absorbed in the multiplication region, a set of avalanche current
pulses is triggered, the integration of which forms the SPAD output
signal.
Fluorescent immunocomplexes are formed on streptavidin-coated
magnetic beads by first incubating the latter off-chip with biotinylated
polyclonal rabbit antimouse immunoglobulin (IgG) for providing the
capture Ab (see Supporting Information, Immunoassay protocol). As
target Ag, we use either a commercial mouse IgG mAb in varying
concentrations (0.25-5000 ng/mL) or a 5D10 murine mAb solution
obtained from a hybridoma cell culture medium (see Supporting
Information, 5D10 mAb production). A polyclonal goat Cy3conjugated antimouse IgG is used as the fluorescent detection Ab.
After injection of the bead-immunocomplex solution in a low-cost,
disposable polydimethylsiloxane (PDMS) microfluidic channel (see
Supporting Information, Microfluidic cartridge fabrication), the beads
are transported using time-dependent current actuation of the
microcoil array, allowing precise positioning of single beads over a
SPAD. To achieve this, the chip and microfluidic channel are first
put in a constant magnetic field generated by a 4 cm diameter
external copper coil. Before the microcoil’s current is switched on,
the beads are at arbitrary positions in the microchannel. When
applying a current through a microcoil, we can obtain an attractive
or repulsive force on the beads, depending if the microcoil-induced
magnetic field is in the same or opposite sense of the field generated
(21) Niclass, C.; Rochas, A.; Besse, P. A.; Charbon, E. IEEE J. Solid-State Circuits
2005, 9, 1847–1854.
50
Analytical Chemistry, Vol. 82, No. 1, January 1, 2010
Figure 2. Normalized photocurrents detected by translation in the
PDMS microchannel of a magnetic bead over the SPAD active area.
I0 is the SPAD output when no bead is above the detector. Signals in
part a are for a nonfluorescent bare bead (Ibare) and a fluorescent
bead (Ifluo). The fluorescent beads are prepared by a first incubation
with the capture Ab, then with a 5 ng/mL mouse IgG target Ag
solution, and finally with Cy3-conjugated antimouse IgG fluorescent
detection Ab. The normalized differential signal in part b represents
the extra-counted photons originating from the presence of fluorescent
molecules on the bead surface.
by the external coil, respectively. Figure 1d shows magnetic beads
attracted to the center of the coil. Using the combined actuation of
adjacent coils, we are able to isolate a single bead and position it
over a SPAD; interference due to the presence of other beads is
thereby avoided. We then measure the fluorescent signal due to a
single stationary bead on top of the SPAD.
Figure 2a shows examples of the normalized photocurrent
detected by the SPAD when examining the translation in the
microchannel of a nonfluorescent bare bead and a bead functionalized
by applying first 5 ng/mL mouse IgG as target Ag, followed by
exposure to the fluorescent detection Ab solution. When the
magnetic bead is not fluorescently labeled, its presence above the
SPAD is detected by its shadow, as in a microeclipse effect, resulting
in a ∼9% reduced photocurrent.22 In the case of the fluorescent bead,
the SPAD registers an additional photon count, caused by the
fluorescent aura around the bead (see further), and the decrease in
photon count during transition is lower. Figure 2b is the differential
curve obtained using the two curves of Figure 2a and represents
the extra-counted photons originating from the presence of fluorescent detection Ab.
Figure 3 is a graph of the normalized time-averaged fluorescent
response of the SPAD detection system to different target Ag
(22) Lehmann, U.; Sergio, M.; Pietrocola, S.; Dupont, E.; Niclass, C.; Gijs,
M. A. M.; Charbon, E. Sens. Actuators, B: Chem. 2008, 132, 411–417.
Figure 3. Average normalized differential signal due to a single magnetic bead positioned over the SPAD active area for different target Ag
concentrations. I0 is the SPAD output when no bead is above the detector. Ifluo is the time-averaged signal when a fluorescent magnetic bead is
positioned over the SPAD active area, while Ibare is the time-averaged signal corresponding to the presence of a bare bead. The 9 and × represent
the data using either mouse IgG mAb or 5D10 mAb as target Ag, respectively. The data points and error bars correspond to the average and variance
of 4-6 measurements performed on nominally identical single beads, respectively. The dashed curve is a guide to the eye, while the full curve is a
logarithmic fit (correlation coefficient: R ) 0.976). The inset details the fluorescence at low concentrations on a linear concentration scale.
concentrations. The data points correspond to the fluorescent
intensity of single magnetic beads positioned over a SPAD. We
combine here the results obtained using as target Ag both the mouse
IgG mAb (9) and the 5D10 mAb (×). The graph shows that our
system has a sensitive range of around 3 decades of concentration,
which is a typical value for fluorescent detection using synthetic dyes
like Cy3.23 When the analyte concentration is higher than 10 ng/
mL, we observe saturation of the bead surface with target Ag. The
inset of Figure 3 shows the data points for the lowest concentrations
on a linear concentration scale. The first point corresponds to the
SPAD output obtained when using 0 ng/mL target Ag but after
performing the complete immunoassay protocol. The point coincides
with the fluorescent response of a bare magnetic bead, indicating a
negligible aspecific adsorption of the fluorescent detection Abs on
the bead surface. We estimate that the detection limit of our system
is around 1 ng/mL, the fluorescence response at lower concentrations being influenced by statistical bead-to-bead variations. This limit
is comparable to the one achieved by performing a similar sandwich
immunoassay on beads using a normal fluorescent microscope
setup.17
We also could correlate the SPAD detection signal to the
presence of a fluorescent layer on the surface on a magnetic bead,
observed as a fluorescent “aura” using an inverted microscope
setup (Zeiss Axio Observer.A1 with Hal 100 light-source). In these
experiments, the Cy3 dye is excited at λ1 ) 550 nm. Parts a and
c of Figure 4 show the fluorescence intensity profiles and
corresponding images of magnetic beads obtained with a 32×
objective, (a) without and (c) with application of a fluorescent
emission filter (λ2 ) 605 nm) placed between the particle and
the microscope detection objective. Curves i represent the
intensity profile for a bare bead, while curves ii correspond to
the intensity profile of a fluorescently labeled bead. Curves iii
of parts b and d of Figure 4 represent each the intensity
difference between the curves i and ii of parts a and c of Figure
(23) Rick, W. Luminescence 2003, 18, 25–30.
4, respectively. The curves iii confirm essentially that a photocurrent increase can be associated with the presence of fluorescent
molecules on the bead surface. The bead images inserted at the
right in Figure 4a,c confirm that these extra photons can be
perceived by the microscope as originating from a fluorescent aura
around the magnetic bead indeed.
Moreover, we simulated in two dimensions the experimental
diffraction data using the finite element method (FEM) in
COMSOL Multiphysics. The FEM simulation model applies
Kirchoff’s approximation to the problem, in which the scalar
complex amplitude of the wave field is determined by solving the
Helmholtz equation (eq 1) (where k0 ) 2π/λi is the wavenumber
determined by the wavelength λi of the coherent light source)
with appropriate boundary conditions.
∆u(x, y) + k02u(x, y) ) 0
(1)
Solving the Helmholtz equation by numerical methods and using
the Kirchoff’s approximation are widely used approaches to model
electromagnetic waves propagation in dielectric media, when
considering monochromatic light.24,25 In our first model (Figure
4e), we simulate a bare bead (without fluorescent labels) observed
by the microscope. In the second simulation (Figure 4g), we
model a fluorescent bead, when observed through the λ2 filter.
Further details are reported in the Supporting Information,
Finite Element Method. Parts f and h of Figure 4 show the
agreement between the simulations and the experimental observations, without and with the use of the emission filter, respectively.
The squares correspond to the experimental curve i in Figure 4a
and curve ii in Figure 4c, respectively, and the full lines are
simulated curves, as obtained along the dashed line segments in
(24) Gross, H.; Model, R.; Bär, M.; Wurm, M.; Bodermann, B.; Rathsfeld, A.
Measurement 2006, 39, 782–794.
(25) Testorf, M. E.; Fiddy, M. A. Opt. Commun. 2000, 176, 365–372.
Analytical Chemistry, Vol. 82, No. 1, January 1, 2010
51
Figure 4. (a-d) Fluorescence intensity profiles and images of magnetic beads measured using an inverted microscope setup (32× objective) and (e-h)
two-dimensional finite element simulations of light diffraction. Experimental data are taken (a, b) without and (c, d) with the use of a fluorescent emission filter
(λ2 ) 605 nm) placed between the bead and the microscope detection objective; the role of the filter is to remove the fluorescent excitation signal at λ1 ) 550
nm. In parts a and c, curves i and ii are the intensity profiles corresponding to a bare and fluorescent bead, respectively. The left and right insert show the
microscopic images of a bare bead and a fluorescent bead (for a target Ag concentration of 1 mg/mL), respectively. Curve iii in parts b and d is obtained by
subtracting curve ii from curve i for the data of parts a and c, respectively. Parts e and g show the FEM simulations corresponding to the experimental data
of parts a and c, respectively. In parts f and h, the 9 correspond to the experimental curve i of part a and ii of part c, respectively, while the full lines are the
simulated curves along the dashed line segment in parts e and g, respectively.
parts e and g of Figure 4, respectively. The good agreement
between experiments and simulations demonstrates that our FEM
model correctly represents the experimental situation.
In conclusion, we reported an integrated monolithic CMOS
chip for sensitive fluorescent detection of immunocomplexes
(down to 1 ng/mL of target Ag concentration) on the surface of
single magnetic beads. The latter are transported in a PDMS
microchannel by current-actuation of microcoils, allowing precise
positioning of single beads over a SPAD. Finally, we elucidated
the auralike origin of the additional photons detected by the SPAD
when fluorescent molecules are present on the magnetic bead
surface. We anticipate that our integrated magnetic bead manipulation and detection chip can be combined with a variety of
52
Analytical Chemistry, Vol. 82, No. 1, January 1, 2010
available magnetic bead surface chemistries, offering high flexibility and great potential to our approach for a large number of
sensitive bioassays and will be of interest especially for miniaturized portable applications that involve sensitive fluorescent detection.
SUPPORTING INFORMATION AVAILABLE
Additional information as noted in text. This material is
available free of charge via the Internet at http://pubs.acs.org.
Received for review October 5, 2009. Accepted October
30, 2009.
AC902241J