Three-Dimensional Magnetic Focusing of Superparamagnetic Beads

Jan 7, 2011 - Chad I. Rogers , Jayson V. Pagaduan , Gregory P. Nordin , and Adam T. Woolley. Analytical ... Anas Alazzam , Bobby Mathew , Saud Khashan...
33 downloads 0 Views 3MB Size
Download PDF
ARTICLE pubs.acs.org/ac

Three-Dimensional Magnetic Focusing of Superparamagnetic Beads for On-Chip Agglutination Assays R. Afshar,* Y. Moser, T. Lehnert, and M. A. M. Gijs Laboratory of Microsystems, Ecole Polytechnique Federale de Lausanne, STI-LMIS2 BM, 3135-Station 17, CH-1015 Lausanne, Switzerland ABSTRACT: We present a new method for three-dimensional (3D) magnetic focusing of magnetic microparticles in a microfluidic system. On-chip magnetic particle manipulation in the microchannel is achieved by a high-gradient magnetic field generated by means of a micromachined field concentrator. The system allows retention of functionalized beads in a dense plug while flowing through buffer or analyte. Slowly reducing the magnetic retention force in the presence of a flow results in controlled release of the particles into a fine streamline with regular longitudinal interparticle spacing. Alignment at half-height of the channel is readily obtained through the symmetry of the magnetic field. A single lateral sheath flow is required to provide full 3D focusing of the microparticles in the middle of the microchannel with a maximum deviation of (5 μm from the center position. With the use of this system, a new approach for performing an immunoagglutination assay on-chip has been implemented. Three-dimensional focusing allowed reliable counting of singlets and agglutinated doublets. We demonstrate the potential of the agglutination assay in a microfluidic format using a streptavidin/biotinylated bovine serum albumin (bBSA) model system. A detection limit of about 400 pg/mL (6 pM) is achieved.

M

agnetic micro- and nanoparticles (beads) are versatile mobile carriers in microfluidic systems, in particular for bioanalytical and diagnostic applications. The possibility to combine magnetic and fluidic forces for bead manipulation, extraction of biomolecules from complex sample matrixes, or for size separation in small liquid volumes opens the way to innovative concepts with previously unmatched performance.1 For instance, immunoassays based on a number of different magneto-microfluidic protocols have been proposed earlier.2 A particular advantage of using magnetic beads is that they can be magnetically retained in the flow channel when passing a sample solution.2a,3 Thereby, they can specifically capture a target molecule from the analyte flow. In general, minimizing the number of beads in the system favors purification or concentration of the target molecules, which is beneficial for sensitive analyte detection.4 It may be even particularly attractive to implement an individual bead analysis approach on-chip for the accurate counting or discrimination of particles with different sizes. Such on-chip cytometric approaches require three-dimensional (3D) focusing of the particle suspension into a single focused stream for reliable detection. Particles should pass one-by-one, in line and with a constant speed through the illumination spot of a laser or the sensitive zone of an optical or impedimetric detector that is directly interfacing with the microchannel.5 The requirement of a very narrow focused particle stream is particularly important for detecting small cells or micrometer-size beads. A novel feature of our work is that we present a microfluidic magnetic actuation system that provides in a straightforward way the 3D focusing of magnetic beads. Furthermore, a magnetic r 2011 American Chemical Society

immunoagglutination assay was performed on-chip, taking advantage of the particle alignment in the 3D focused stream. Immunoagglutination assays represent a simple and cost-efficient alternative in the field of medical diagnostics.6 In such assays, the analyte or antigen (Ag) binds to colloidal particles grafted with an antibody (Ab) by forming a sandwich-like structure (Ab-Ag-Ab). It has been demonstrated that the detection limit can be reduced down to 1 pM in the bulk format by introducing immunoagglutination of superparamagnetic nanoparticles for a model immunoassay.7 This significant result is based on a strong increase of the colliding frequency between the magnetic carriers in the presence of a magnetic field leading to the formation of magnetic chains. A first approach for a chip-based agglutination test with magnetic beads was presented by Degre et al.8 The high potential of on-chip agglutination assays could be demonstrated by Moser et al. using an integrated magnetic actuation system.3a Three-dimensional hydrodynamic focusing in a microfluidic device may be realized in a variety of ways, as was summarized in a recent review.9 Both horizontal and vertical laminar sheath flow focusing concepts have been proposed, generally requiring a multilayer chip design with several fluidic inlets and eventually with different microchannel heights.10 Chung et al. presented a microfluidic device using a sample injection nozzle centered in a microfluidic channel realized using complex 3D lithography.11 Received: October 26, 2010 Accepted: December 15, 2010 Published: January 7, 2011 1022

dx.doi.org/10.1021/ac102813x | Anal. Chem. 2011, 83, 1022–1029

Analytical Chemistry Watkins et al. developed an electrical microcytometer, in which T-cells were forced by a sheath flow to the bottom of a microchannel where they passed just over electrical detection electrodes.5b A high-throughput microfluidic flow cytometer was presented by Simonnet and Groisman.12 This device was made of a single cast of poly(dimethlysiloxane) (PDMS) on an SU8 master microstructure. Adjusting the hydrostatic pressure applied to four fluidic inlets, together with the variation of the microchannel’s geometrical parameters, resulted in a focused 10 μm wide stream of particles or cells. This approach was developed further to allow studying of biochemical kinetics based on a highly confined 3D sample flow with a lateral dimension below 0.5 μm.13 Mao et al. took advantage of Dean flow in a curved channel for focusing 7-8 μm polystyrene particles in the vertical direction, thus eliminating a complex multilayer chip design. In this work, lateral focusing prior to detection was achieved by conventional sheath flow focusing,14 whereby the focused bead stream had a diameter of less than 15 μm.5a Dean flow-assisted sheath flow focusing of human red blood cells was demonstrated by Lee et al. in a microchannel composed of an array of contraction and expansion regions.15 In a 3D flow focusing method using grooved channels, two lateral sheath flows first forced a dye sample in the middle of the channel and then pairs of chevrons cut into the top and bottom walls of the channel directed the sheath fluid from the sides to the middle over and under the sample stream.16 Chevron-based 3D sheath flow focusing has also been demonstrated to focus 5.6 μm diameter polystyrene beads into a microflow cytometer.17 In another design, a combination of sheath flows and a microweir positioned right beneath the optical detection system was used to perform focusing of 5 and 10 μm polystyrene beads.18 Sheathless particle focusers rely on a force to manipulate the suspended particles laterally with respect to the main stream to their equilibrium positions. This force can be of external origin or can be internally induced by the channel topology. Sheathless inertial focusing of 7 μm polymer beads has been achieved in straight microfluidic channels comprising regular alternating geometrical structures.19 Also devices exploiting lift forces in a serpentine channel for continuous focusing, self-ordering, and separation of particles and red blood cells have been demonstrated.20 (Di)electrophoretic forces or acoustophoresis are other interesting options for focusing particles, cells, or single molecules.21 Kim and Yoo performed axisymmetric flow focusing in a circular microcapillary of 5 μm diameter polystyrene beads that were lagging behind the flow of a suspending fluid (a solution of water-glycerol with the same density as the particle) due to application of a negative electrophoresis force; the same principle has also been used to focus red blood cells.21c,22 The concept of “hydrodynamic filtration” has been applied to focus 5 μm fluorescent beads into a stream along the centerline of a planar microchannel.23 In this method, multiple 10 μm wide side channels first split the carrier fluid flows from the main channel, while leaving the beads in the main stream. Then, these split flows were reinjected into the main channel from both sides simultaneously focusing the beads to the center. Using the narrow microfluidic side channels required sophisticated microfabrication techniques and sufficiently clean flow solutions during operation to prevent clogging. In this paper, we propose for the first time a magneto-microfluidic system comprising a magnetic microtip, which functions as a field concentrator, for magnetic force-based 3D focusing of magnetic beads in a microchannel. Focusing in the vertical direction is readily achieved through the magnetic field configuration. A

ARTICLE

Figure 1. (a) Magnetic bead actuation system comprising an electromagnet, a magnetic yoke, and a pair of soft-magnetic poles that form an interface with the microfluidic chip. (b) Zoom on the microchip featuring a straight microfluidic channel (inlet 1), a secondary channel (inlet 2), and the asymmetric arrangement of the two magnetic tips. (c) Enlarged view of the chip, showing the release of the magnetic beads and the deviation of a 3D focused stream of individual beads toward the center of the main channel using a second flow.

single lateral sheath flow is necessary to subsequently position and align individual beads in the center of the flow. Furthermore, the system permits bead retention and controlled release. It is designed to handle a very small amount of beads (∼500-1000), a low number being beneficial for optimum concentration of analyte molecules on the bead surface in general. We demonstrate here the high potential of this platform by performing an on-chip magnetic bead-based immunoagglutination assay with accurate detection. Magnetic actuation allows efficient analyte capture on the surface of the functionalized superparamagnetic particles. The amount of agglutinated particles and thereby the analyte concentration is determined by observation of the 3D focused stream of released particles. A biotin-streptavidin model assay was used to test the system.

’ MATERIALS AND METHODS Microfluidic Setup. The magneto-microfluidic setup comprising the chip and the magnetic circuit is schematically shown in Figure 1a, with a zoom on the microfluidic chip in Figure 1b. The chip has a principal straight microfluidic channel (10 mm long, 200 μm wide, 100 μm high) and a secondary microchannel (200 μm wide, 100 μm high). The fluidic junction is located 0.5 mm downstream with respect to the magnetic tip location (Figure 1b). Figure 1c is an enlarged schematic view of the region of interest, illustrating the release of a magnetically retained bead plug into a 3D focused stream. A single syringe pump is connected simultaneously to both fluidic inlets of the chip for fluidic control. As both channel portions before the fluidic junction have equal hydraulic resistance, the flow rates in both sections are equal. Consequently, the secondary flow deviates the released magnetic bead stream exactly toward the middle of the main channel. 1023

dx.doi.org/10.1021/ac102813x |Anal. Chem. 2011, 83, 1022–1029

Analytical Chemistry

Figure 2. (a) 2D FEM simulation of the magnetic field lines in the channel close to the adjacent tip. The resulting magnetic force field (eq 1) exerted on a magnetic particle is represented by arrows of different sizes. Significant field concentration, thus attractive force, is observed only on one channel side wall due to the asymmetric positioning of the two tips. (b) 3D FEM simulation of the magnetic field flux density in the channel. Images represent calculations of the magnetic flux density in the x-z plane at different distances y from the adjacent tip (positioned at y = 0 μm). The first image (at y = 40 μm) corresponds to the field distribution at the channel side wall. The two other images correspond to the middle and the opposite side of the channel.

The microfluidic chips are cast in a single piece of PDMS (Sylgard 184, Dow Corning Inc. Midland, MI) using a standard SU8/Si mold. Sealing of the channels is achieved by mechanically clamping the PDMS part onto a glass slide by means of a PMMA chip holder with suitable fluidic connections. Magnetic Setup and Field Simulation. We use magnetophoretic force to attract the magnetic beads toward the tip. The magnetic force FBmag on the beads in an induction field B is calculated by 1 ΔχVm rB2 ð1Þ B F mag ¼ 2μ0 where Δχ is the volume magnetic susceptibility of the particle relative to the suspending medium, μ0 is the vacuum permeability, and Vm is the magnetically active volume of the bead.1a,24 This expression holds for magnetically unsaturated particles, which is a good approximation for the present experiments. In the present device, one of the soft-magnetic tips is placed adjacent to the channel wall (wall thickness is 40 μm, Figure 1c). This tip concentrates the external magnetic field generated by the coil in a small region of the microchannel. The other tip is positioned at a distance of 500 μm and is used for flux closure. This arrangement leads to a strong magnetic gradient only on one side of the microchannel. The magnetic microtips are cut by laser from a 100 μm thick magnetic foil (Vacoflux 50, Vacuumschmelze, Germany). The width of the narrow tip end is 50 μm. The PDMS part of the chip comprises recesses to receive the pair of magnetic poles in a reliable manner. Figure 2a shows a two-dimensional (2D) finite element method (FEM) simulation of the magnetic field lines in the channel plane (x-y plane) and the magnetic force field close to the tip that is adjacent to the microchannel. The resulting force field is asymmetric with respect to the channel middle axis. Beads

ARTICLE

will therefore be attracted toward this tip and form a plug on the channel side wall. Due to the rectangular cross section of the tip end, the force field distribution has a similar symmetry in the channel cross section (y-z plane) as in the channel plane (x-y plane). Figure 2b shows the result of a 3D FEM simulation of the magnetic flux density in the x-z plane. The strong flux gradient in the channel is illustrated by three plots taken at different distances in the y-direction from the tip adjacent to the channel. The flux density obviously is highest at the channel wall near the tip (i.e., at a distance of 40 μm from the tip) and becomes very low close to the opposite side wall of the channel (i.e., at a distance of 120 μm from the tip). The external coil is connected to a bipolar operational power supply (Kepco, BOP 50-2M), which is addressed by a function generator (Tektronix, AFG3021B). A 50 Hz sinusoidal current is generated in the coil in all experiments. Damping of the current amplitude allows controlled demagnetization of the magnetic tips for release of the beads. Due to the low coercivity of the superparamagnetic beads, their kinetic behavior depends only on the maximum value of the field intensity. This means that a static plug forms on the channel wall, even if an ac field is applied.25 For a current of 80 mA in the external coil (1400 Cu windings), the effective field strength at the channel wall was calculated to be ∼50 mT (see Figure 2b). Protocol for Bead Capture and Release. Two types of superparamagnetic beads with a diameter of 1.05 μm (Δχ = 1.38, 1.8 g/cm3, Dynabeads MyOne) and 2.83 μm (Δχ = 0.84, 1.6 g/cm3, Dynabeads M-270), respectively, are used in the experiments. Bead solutions are diluted and suspended in 15 mM phosphate buffer saline (PBS) solution with 0.5% surfactant (Tween 20) to improve the colloidal stability. The experimental procedure starts by sucking 2 μL of bead solution with a concentration of 5  108 beads/mL from the outlet into the main channel at a flow rate of 20 nL/s (1 mm/s). A syringe pump is used for this purpose (neMESYS, Cetoni, Germany). The flow is stopped and the tip is magnetized, once the particle suspension uniformly fills the channel portion close to the tip. The amount of captured particles forming a dense plug is proportional to the coil current and can be estimated from the depleted area in the channel close to the tip.26 The surplus of beads is flushed away by pumping buffer solution to the outlet. The system is now ready for performing an agglutination assay or for bead release with 3D focusing. For the 3D focusing experiments, the 50 Hz sinusoidal coil current is exponentially attenuated to zero (with a damping constant 0.018 ms-1). Controlled release of the bead plug is achieved by applying simultaneously a constant and equal flow rate through inlets 1 and 2 of the microchip (typically 6 nL/s, mean velocity 0.3 mm/s). Particle trajectories are observed on an inverted optical microscope (Zeiss Axiovert S100) equipped with a fast monochrome digital camera (PixeLINK PL-B741), using video recording software (StreamPix) and image processing for particle counting (ImageJ, open source). Reconstructed fluorescent images in the vertical x-z plane are obtained by recording the autofluorescent images of 2.8 μm beads with a Leica SP5 inverted confocal scanning microscope. For that, a series of images is collected in the x-y plane, spanning over a range of 100 μm in the z-direction with z-sectioning steps of 3 μm. All images are analyzed using LAS AF 9000 software to reconstruct the image in the x-z plane. The confocal microscope is equipped with a 488 nm argon laser for fluorescent excitation. Due to the lower fluorescence, the 1 μm beads could not be clearly visualized with this system. 1024

dx.doi.org/10.1021/ac102813x |Anal. Chem. 2011, 83, 1022–1029

Analytical Chemistry

Figure 3. Principle of the microfluidic agglutination immunoassay. (a) Formation of a dense magnetic bead plug at the channel side wall. (b) Release and 3D focusing of the beads in a solution flow by controlled attenuation of the field. A detection window is defined (dashed rectangle) for automated counting of the particles in the focused stream. (c) Inverting the flow direction allows capturing again the whole amount of beads and the formation of a dense plug. Steps a-c are performed in buffer solution and serve as in situ reference experiment. (d) The buffer solution is replaced by the analyte solution. Dissociation of the confined bead plug allows capturing the analyte (no flow is applied). Dense plug formation and dissociation may be repeated several times to increase the amount of antigens on the beads. Subsequently, beads are released for detection (step b).

Protocol for On-Chip Immunoagglutination. We take advantage of the strong affinity of streptavidin for biotin to develop a model assay for agglutination on-chip. In our system, the analyte (or target Ag) is biotinylated bovine serum albumin (bBSA) (Sigma-Aldrich, Buchs, Switzerland) that may link to streptavidincoated beads, thus forming doublets or larger aggregates. The analyte test solution is 15 mM PBS containing variable bBSA concentrations in the range of 10 pg/mL to 1 μg/mL (0.15 pM to 0.15 μM). A surfactant (0.5%, Tween 20) was added to reduce nonspecific agglutination. The principle of the on-chip agglutination assay is illustrated in Figure 3. First, a colloid of functionalized streptavidin-coated magnetic beads is injected into the microchannel and magnetically immobilized in the form of a dense plug on one of the side walls (Figure 3a). In order to determine precisely the amount of specifically agglutinated particles (i.e., linked by bBSA), it is of critical importance to record accurately an in situ reference signal prior to each measurement. For that, the beads are simply released in the buffer solution and the focused stream is analyzed (Figure 3b). Beads are counted by automated image recording in an observation window in order to determine the amount of single beads and aggregates. The center of this window is located 350 μm downstream the fluidic junction on the chip (window size 230 μm  600 μm). The bead stream passing the window is recorded, and the video sequence is analyzed. Subsequently, the flow direction is reversed and the beads are captured again in a plug at the initial location (Figure 3c). The majority of the initial amount of beads (more than 95%) is recovered at the end of this procedure. The proper agglutination protocol starts by introducing the analyte solution in the microchannel. Once the channel is filled,

ARTICLE

Figure 4. (a) Top view of a magnetic plug immobilized on the side wall of the microchannel under the effect of an applied magnetic field (about 400 beads of 1 μm diameter are retained). (b) Beads are progressively released from the plug into the flow by controlled attenuation of the sinusoidal coil current. (c) Fluorescent image of the plug and the released particle stream in the x-z plane. Vertical focusing of the beads at half-height of the channel is observed (bead diameter 2.8 μm). (d) Photograph of a focused bead stream at the channel junction (top view, bead diameter 1 μm). (e) Cross section of the focused 2.8 μm bead stream in the main channel (dashed rectangle) taken 50 μm after the fluidic junction. The diameter of the bead stream diameter is about 10 μm. The stream is positioned nearly in the center of the channel (bead velocity of 0.6 mm/s). (f) Reducing the bead velocity to 0.1 mm/s increases the impact of sedimentation, and the bead stream is close to the bottom of the channel. Images c, e, and f are reconstructed from a series of fluorescent confocal images taken in the x-y plane (vertical sectioning 3 μm).

the flow is stopped and the field is removed (Figure 3d). The highly localized bead plug dissociates in the bBSA solution allowing capturing the analyte molecules on the beads. After this step, the magnetic field is applied and the beads form again a dense plug. This cycle may be repeated several times, replacing the bBSA solution after each cycle. The impact of repeated cycling and a variable residence time of the beads in the dissociated state are studied in the following examples. Subsequently, the agglutinated colloid is released and focused for detection.

’ RESULTS AND DISCUSSION Magnetic 3D Focusing of the Beads. Experimental bead capture and plug formation near the magnetic tip on the channel wall, as well as subsequent release, is demonstrated in the images shown in Figure 4, parts a and b. In the present case, experimental parameters have been chosen to form a plug that contains 400 ( 20 beads with a diameter of 1 μm (coil actuation current of 80 mA) (Figure 4a). The possibility of controlling the amount of beads (dosing) has been described previously.26 Basically, a magnetic field is applied to the tip, after filling the channel with a homogeneous magnetic beads solution. The area in which the beads are depleted close to the tip provides a direct estimate of 1025

dx.doi.org/10.1021/ac102813x |Anal. Chem. 2011, 83, 1022–1029

Analytical Chemistry the number of beads that are in the plug. The superparamagnetic beads decrease their magnetic moment during exponential attenuation of the magnetic field. The hydrodynamic shear force progressively overrules the decreasing magnetic retention force and tears off the beads from the plug (Figure 4b). The released particles thereby are confined in a narrow stream close to the channel wall. Figure 4c shows a fluorescent image of a plug of 2.8 μm beads at the vertical channel wall adjacent to the magnetic tip (x-z plane). The total extension of the plug in the z-direction is less than 50 μm. With the present chip design, prefocusing in the z-direction is readily achieved thanks to the magnetic field configuration. Experimental observation reveals that beads are first attracted toward the middle of the microtip area, where magnetophoretic forces are highest. The plug forms around this center position, thus is confined in a volume that is much smaller than the volume defined by the lateral and vertical dimensions of the tip, depending on the amount of retained beads. The extension of the released particle stream in the x-z plane shrinks further down to about 10 μm at a distance of 300 μm from the plug location (flow rate 6 nL/s). This feature is at the origin of the magnetic focusing of the beads along the z-axis. Shrinking of the bead bead stream diameter after release is most likely related to the laminar flow streams forming around the dense plug close to the channel wall. At the fluidic junction on the chip, the released magnetic bead stream is deviated toward the center of the main channel (see Figure 4d). Here, the flow rate and the maximum flow velocity are twice as high as in the channel sections upstream. Furthermore, beads are pushed from a region with very low flow velocity near the channel wall toward the center of the parabolic flow profile. Particles in the released stream are therefore strongly accelerated. This simple fluidic arrangement allows stretching and one-by-one alignment of the released beads in the flow direction. Figure 4d shows the experimental observation using 1 μm beads. For the flow parameters chosen in Figure 4d, the particle speed after the junction is 0.6 mm/s (12 nL/s) and beads align in the middle of the flow channel with a maximum deviation of (5 μm in the lateral direction (y-direction) from the center position. The linear bead density in the stream (x-direction) is about 20 beads/mm in this case. The rate of release can be tuned by varying the field damping constant and the flow rate. Three-dimensional focusing of a released stream of the 2.8 μm beads is demonstrated in Figure 4e. The image shows a cross section of the bead stream in the microchannel in the y-z plane taken 50 μm after the fluidic junction. The flow linear velocity at the center of the main channel is 0.6 mm/s. The focused bead stream diameter is about 10 μm, i.e., 10% of the channel height and 5% of the channel width. Considering the parabolic flow profile, the maximum variation of the bead velocity in the focused stream can be estimated to about 1%. At the location of the channel considered in Figure 4e, the beads appear not to be exactly focused at half-height, but the center of the bead stream is deviated by 10 μm from the middle of the channel toward the bottom. Figure 4f shows an image of the bead stream taken at the same location using a lower flow rate (bead velocity of 0.1 mm/s (2 nL/s) in the main channel). In Figure 4f, the beam center is deviated by 45 μm and nearly reaches the bottom of the channel. The total travel time after release being about 4 times longer than for the situation of Figure 4e, we attribute this deviation to a stronger impact of sedimentation of the beads. A quantitative force analysis confirms this assumption. The deviation of 1 μm beads from the center position at the

ARTICLE

Figure 5. (a) Photograph shows a stream of released and focused particles. The insets show a zoom where single beads and doublets can be distinguished (bead diameter 1 μm). (b) Curve showing the impact of the number of standard cycles on nonspecific agglutination in buffer solution, as monitored by the reduction of the number of singlets. One standard cycle consists of forming a dense plug for 1 min prior to plug dissociation during 1 min (no flow applied). (c) Curve demonstrating the increase of specific agglutination with the number of cycles. One cycle consists of forming a dense plug for 1 min prior to plug dissociation during either 1 or 4 min (no flow applied). Analyte capture was performed by dissociating the plug in a solution of 10 ng/mL bBSA. Two different times (1 and 4 min), during which the beads are kept in the dissociated state, have been tested.

same location was estimated to be negligible at a flow speed of 0.6 mm/s and only 4 μm at a flow speed of 0.1 mm/s. Analyte Capture in the Microchannel. Analyte capture on functionalized beads is achieved by releasing the plug in the sample solution without applying a flow (see Figure 3d). The dissociated plug increases the bead/analyte encounter probability and thereby the capture efficiency.3a Subsequent formation of aggregates is achieved under reconfinement of the beads in a dense plug. The analyte concentration is related to the number of aggregates formed. As an example, Figure 5a shows the alignment of single 1026

dx.doi.org/10.1021/ac102813x |Anal. Chem. 2011, 83, 1022–1029

Analytical Chemistry beads and doublets in a released and focused stream. For quantitative measurements, the number of aggregates/doublets and single beads in the stream may be counted by using standard image processing. In our case, the resulting decrease of the number of singlets was determined. The area of each particle is recorded (number of pixels), and below a certain threshold, the particle is counted as singlet. In order to obtain the best signal possible, i.e., high specific agglutination for a given analyte concentration, while minimizing the amount of nonspecific aggregates, the parameters for analyte capture have been optimized. Two parameters have been investigated: (i) the number of cycle repetitions and (ii) the residence time of the beads in the dissociated state of the plug. One standard cycle consists of dissociating the plug during 1 min followed by the reformation of a dense plug for 1 min. First, nonspecific agglutination was investigated. Figure 5b demonstrates the impact of repeated cycling in buffer solution. The initial amount of singlets before cycling  Sref,N=0 was 92.5%. The graph shows a decrease of the amount of singlets with the number of cycles N, i.e., an increase of nonspecific agglutination. Nonspecific interaction and sticking of the beads to each other occurs during the formation of the dense plug. The cycles have been repeated up to 10 times, resulting in a decrease of single beads to 80%. For the proper agglutination assay, we use three cycles and the corresponding reference signal  Sref,N=3, defining a correction factor F = (Sref,N=3/Sref,N=0) = 1.02, which corresponds to a 2% reduction of the initial number of singlets due to nonspecific agglutination. This value is acceptable compared to the significant increase of specific agglutination after three cycles (see below). In a second step, the analyte capture was evaluated for different configurations. Figure 5c compares the normalized signal for an increasing number of cycles and for two different dissociation times. The experiments have been carried out with a bBSA concentration of 10 ng/mL. The bBSA solution has been replaced after each cycle. An on-chip reference test (one cycle taking buffer solution without analyte) was carried out before each measurement in order to determine the amount of nonspecific agglutination (see Figure 3). We consider the normalized signal 1 - SbBSA,N/Sref,N, defining the percentage of the single beads in the released stream after a number of cycles N, based on the signal Sref,N and SbBSA,N, defined by the percentage of the number of singlets in the reference and the analyte measurement, respectively. Increasing the number of cycles from one to three resulted in an about 7-fold increase of the detection signal. Squares in Figure 5c correspond to a dissociation time of 1 min and circles to a time of 4 min. The incubation time, i.e., the formation of the dense plug, was 1 min in both cases. Only a very small change was found between cycle 3 and 4. We conclude that the optimum density of bead surface-bound antigens is obtained after three cycles. Furthermore, we may deduce from the curve in Figure 5c that the plug dissociation time seems to have no significant impact on the capture efficiency. This result indicates that the analyte capture is not limited by diffusion under the present conditions. Dose-Response Curve of the Immunoagglutination Assay. An on-chip immunoagglutination assay according to the above-described protocol was performed. Agglutination occurs when a bBSA molecule captured on the surface of a bead binds to another bead, resulting in a stream of beads that contains a mixture of aggregates of different sizes as well as single beads. In the limit of low concentrations, mainly doublets will form.7a We note that, without 3D focusing, it is extremely difficult to distinguish in a stream singlets and doublets of very small beads (typical diameter

ARTICLE

Figure 6. Dose-response curve for a magnetic agglutination assay using a streptavidin-bBSA model system. A full on-chip assay (squares), where analyte capture was achieved by magnetic actuation on-chip, and a partially off-chip assay (circles) are compared. A detection limit of 400 pg/mL was obtained for the on-chip assay.

1 μm) with a normal microscope setup. As described above, we take here clear advantage of the 3D focused bead stream for accurate counting of the particles. All particles are well-aligned and in the same focal plane; thus, reliable detection of single beads and doublets can be made by using simple image processing software. In order to demonstrate the feasibility of a sensitive agglutination assay, we established a dose-response curve for the bBSA-streptavidin model assay. Figure 6 compares the dose-response curves for the on-chip assay (squares) and for an off-chip experiment (circles). An on-chip reference test (one cycle, buffer solution without analyte) is carried out before each measurement in order to determine the amount of nonspecific agglutination (see Figure 3). As explained before, we consider the normalized signal 1 - SbBSA,N=3/(Sref,N=0F), defining the percentage of the single beads in the released stream for the reference and the analyte measurement, respectively. The curve recorded in Figure 5b in principle would allow determining a nonspecific correction factor F as a function of cycle number N. In our studies, three standard cycles have always been performed, resulting in a reduction of 2% of the amount of singlets, i.e., F = 1.02. This method results in a reliable readout of the specific agglutination, even at very low bBSA concentrations. The reference test is very accurate as it is performed in situ, i.e., the same beads are used for the reference test and the assay. The dose-response curve was recorded for bBSA concentrations in the range of 0.01 ng/mL (0.15 pM) to 1 μg/mL (15 nM). We can safely determine a detection limit of about 400 pg/mL (≈6 pM). The signal for this concentration is more than 3 times the standard deviation above zero. Saturation occurs for concentrations above 10 ng/mL (150 pM), where most of the beads are agglomerated. As the total number of retained beads can be controlled during the formation of the plug,26 the nominal number of beads for each measurement was the same for reproducibility and reliability of the assay. The microfluidic agglutination protocol might seem complex due to the repetitive trapping and release of the particles to perform the assay. However, doing this permits us to explicitly take advantage of the control offered by the microfluidic format. Indeed, the possibility to manipulate essentially the same ensemble of beads under quasi-continuous observation allows us to determine the reference signal with the utmost accuracy. This is of 1027

dx.doi.org/10.1021/ac102813x |Anal. Chem. 2011, 83, 1022–1029

Analytical Chemistry particular interest for the lowest concentrations of bBSA (smaller than 1 ng/mL), where the specific agglutination signal (the number of doublets) is small. In principle, it would not have been necessary to consider such dynamic reference signal for the higher concentrations, but we have always performed the same protocol. Also, the drawback of this apparently complex protocol is very relative, as the whole manipulation is done under computer control. For the off-chip assay, the beads are first mixed off-chip with solutions containing different analyte concentrations (incubation time of 6 min). Subsequently, the agglutinated colloid is introduced in the microfluidic chip and immobilized. The plug is released (no cycling) for counting in the 3D focused stream. It can be seen in Figure 6, that the standard deviation for the offchip experiment is much larger than for the on-chip experiment, as no accurate in situ reference signal can be recorded in this case. Also controlling the total number of beads is more complicated. As a consequence, the detection limit can be estimated to about 10 ng/mL, which is a factor of 25 higher than for the full on-chip assay.

’ CONCLUSION We presented a new magneto-microfluidic method for 3D focusing of magnetic beads in the flow direction, using a very simple and robust experimental protocol. Only a single sheath flow is required to align the particles one-by-one in the center of the channel with a maximum deviation of (5 μm. Focusing in the vertical direction is achieved by taking advantage of the symmetry of the strongly focused magnetic field generated by a soft-magnetic microtip integrated on-chip. We showed that this type of device is of particular interest for fully integrated on-chip immunoagglutination assays, which are based on determining the ratio of single beads and aggregates. Capture of the analyte was achieved by repeated concentration and dissociation of a plug of functionalized beads. Subsequent use of the 3D focusing effect allowed accurate counting of single or agglutinated particles. A detection limit of about 400 pg/mL (6 pM) was found using a bBSA-streptavidin model assay. This is competitive with stateof-the-art microfluidics-based assays,1a several orders of magnitude lower than typical values obtained for standard (nonmagnetic) bulk assays, and comparable to a previously reported magnetic assay in the bulk format.7a Thereby, our work demonstrated the high potential of 3D focusing of a magnetic bead stream for the automatic detection of low-concentration target analytes. Indeed, positioning the beads exactly in focus of the optical observation plane using our extremely simple magnetic method, which avoids the application of complex multiple fluid focusing streams, allowed establishing an easy counting protocol, performing the in situ reference test, and the precise dosing of a low number of beads. The latter aspect could be of special interest in future experiments, where very dilute target analytes can be preconcentrated on a minimum number of beads for subsequent detection and analysis. Moreover, in future, the present system could be further exploited when combined with on-chip particle detectors5c and may be developed into a fully integrated versatile platform for performing different types of magnetic bead-based bioassays requiring several manipulation steps on-chip.

’ AUTHOR INFORMATION Corresponding Author

*Phone: (þ41)216934834. Fax: þ41 (0) 216935950. E-mail: rana.afshar@epfl.ch.

ARTICLE

’ ACKNOWLEDGMENT The authors thank the BioImaging and Optics Platform (BIOP) of EPFL (Lausanne), in particular Thierry Laroche for assistance in confocal imaging. Microfabrication of the SU8/Si chips has been carried out in the EPFL Center for MicroNano Technology. The present work was supported by the European Commission funded project DETECTHIV (Grant No. 037118, Sensitive nanoparticle assay for the detection of HIV). ’ REFERENCES (1) (a) Gijs, M. A. M.; Lacharme, F.; Lehmann, U. Chem. Rev. 2010, 110 (3), 1518–1563. (b) Gijs, M. A. M. Microfluid. Nanofluid. 2004, 1 (1), 22–40. (c) Pamme, N. Lab Chip 2006, 6 (1), 24–38. (d) Verpoorte, E. Lab Chip 2003, 3 (4), 60N–68N. (2) (a) Lacharme, F.; Vandevyver, C.; Gijs, M. A. M. Anal. Chem. 2008, 80 (8), 2905–2910. (b) Herrmann, M.; Roy, E.; Veres, T.; Tabrizian, M. Lab Chip 2007, 7 (11), 1546–1552. (c) Do, J.; Ahn, C. H. Lab Chip 2008, 8 (4), 542–549. (d) Do, J.; Choi, J.-W.; Ahn, C. H. IEEE Trans. Magn. 2004, 40 (4), 3009–3011. (3) (a) Moser, Y.; Lehnert, T.; Gijs, M. A. M. Lab Chip 2009, 9 (22), 3261–3267. (b) Nel, A. L.; Minc, N.; Smadja, C.; Slovakova, M.; Bilkova, Z.; Peyrin, J.-M.; Viovy, J.-L.; Taverna, M. Lab Chip 2008, 8 (2), 294– 301. (c) Bronzeau, S.; Pamme, N. Anal. Chim. Acta 2008, 609 (1), 105–112. (4) Sivagnanam, V.; Song, B.; Vandevyver, C.; Gijs, M. A. M. Anal. Chem. 2009, 81 (15), 6509–6515. (5) (a) Mao, X.; Lin, S. C. S.; Dong, C.; Huang, T. J. Lab Chip 2009, 9 (11), 1583–1589. (b) Watkins, N.; Venkatesan, B. M.; Toner, M.; Rodriguez, W.; Bashir, R. Lab Chip 2009, 9 (22), 3177–3184. (c) Dupont, E. P.; Labonne, E.; Vandevyver, C.; Lehmann, U.; Charbon, E.; Gijs, M. A. M. Anal. Chem. 2010, 82 (1), 49–52. (6) Singer, J. M.; Plotz, C. M. Am. J. Med. 1956, 21 (6), 888–892. (7) (a) Baudry, J.; Rouzeau, C.; Goubault, C.; Robic, C.; CohenTannoudji, L.; Koenig, A.; Bertrand, E.; Bibette, J. Proc. Natl. Acad. Sci. 2006, 103 (44), 16076–16078. (b) Baudry, J. J. Phys.: Condens. Matter 2004, 16 (15), R469. (8) Degre, G.; Brunet, E.; Dodge, A.; Tabeling, P. Lab Chip 2005, 5 (6), 691–694. (9) Xuan, X.; Zhu, J.; Church, C. Microfluid. Nanofluid. 2010, 9 (1), 1–16. (10) (a) Sundararajan, N.; Pio, M. S.; Lee, L. P.; Berlin, A. A. J. Microelectromech. Syst. 2004, 13 (4), 559–567. (b) Chang, C. C.; Huang, Z. X.; Yang, R. J. J. Micromech. Microeng. 2007, 17 (8), 1479–1486. (c) Tsai, C. H.; Hou, H. H.; Fu, L. M. Microfluid. Nanofluid. 2008, 5 (6), 827–836. (d) Scott, R.; Sethu, P.; Harnett, C. K. Rev. Sci. Instrum. 2008, 79 (4), 046104. (11) Chung, S.; Park, S. J.; Kim, J. K.; Chung, C.; Han, D. C.; Chang, J. K. Microsyst. Technol. 2003, 9 (8), 525–533. (12) Simonnet, C.; Groisman, A. Anal. Chem. 2006, 78 (16), 5653– 5663. (13) Gambin, Y.; Simonnet, C.; VanDelinder, V.; Deniz, A.; Groisman, A. Lab Chip 2010, 598–609. (14) Mao, X.; Waldeisen, J. R.; Huang, T. J. Lab Chip 2007, 7 (10), 1260–1262. (15) Lee, M. G.; Choi, S.; Park, J.-K. Lab Chip 2009, 9 (21), 3155– 3160. (16) Howell, P. B., Jr.; Golden, J. P.; Hilliard, L. R.; Erickson, J. S.; Mott, D. R.; Ligler, F. S. Lab Chip 2008, 8 (7), 1097–1103. (17) (a) Kim, Y. W.; Yoo, J. Y. Lab Chip 2009, 9 (8), 1043–1045. (b) Golden, J. P.; Kim, J. S.; Erickson, J. S.; Hilliard, L. R.; Howell, P. B.; Anderson, G. P.; Nasir, M.; Ligler, F. S. Lab Chip 2009, 9 (13), 1942– 1950. (18) Fu, L. M.; Tsai, C. H.; Lin, C. H. Electrophoresis 2008, 29 (9), 1874–1880. (19) (a) Choi, S.; Song, S.; Choi, C.; Park, J. K. Small 2008, 4 (5), 634–641. (b) Sato, H.; Sasamoto, Y.; Yagyu, D.; Sekiguchi, T.; Shoji, S. J. Micromech. Microeng. 2007, 17 (11), 2211–2216. 1028

dx.doi.org/10.1021/ac102813x |Anal. Chem. 2011, 83, 1022–1029

Analytical Chemistry

ARTICLE

(20) Di Carlo, D.; Edd, J. F.; Irimia, D.; Tompkins, R. G.; Toner, M. Anal. Chem. 2008, 80 (6), 2204–2211. (21) (a) Wang, T. H.; Peng, Y.; Zhang, C.; Pak, K. W.; Ho, C. M. J. Am. Chem. Soc. 2005, 127 (15), 5354–5359. (b) Shi, J.; Mao, X.; Ahmed, D.; Colletti, A.; Huang, T. J. Lab Chip 2008, 8 (2), 221–223. (c) Kim, Y. W.; Yoo, J. Y. Lab Chip 2009, 9 (8), 1043–1045. (22) Kim, Y. W.; Yoo, J. Y. Biosens. Bioelectron. 2009, 24 (12), 3677– 3682. (23) Aoki, R.; Yamada, M.; Yasuda, M.; Seki, M. Microfluid. Nanofluid. 2009, 6 (4), 571–576. (24) (a) Zborowski, M.; Sun, L.; Moore, L. R.; Stephen Williams, P.; Chalmers, J. J. J. Magn. Magn. Mater. 1999, 194 (1-3), 224–230. (b) Shevkoplyas, S. S.; Siegel, A. C.; Westervelt, R. M.; Prentiss, M. G.; Whitesides, G. M. Lab Chip 2007, 7 (10), 1294–1302. (25) Moser, Y.; Lehnert, T.; Gijs, M. A. M. Appl. Phys. Lett. 2009, 94 (2), 022505. (26) Afshar, R.; Moser, Y.; Lehnert, T.; Gijs, M. A. M. Sens. Actuators, B [Online, early access]. DOI: 10.1016/j.snb.2009.08.044. Published Online: Aug 20, 2009. http://www.sciencedirect.com/science/article/ B6THH-4X3W457-7/2/d22fe8649e241c3f67d00fc0a396f3c2.

1029

dx.doi.org/10.1021/ac102813x |Anal. Chem. 2011, 83, 1022–1029