Label-Free Cell Separation Using a Tunable MagnetophoreticRepulsion ForceFengshan Shen,† Hyundoo Hwang,† Young Ki Hahn,† and Je-Kyun Park*,†,‡

†Department of Bio and Brain Engineering, and ‡KAIST Institute for the NanoCentury, Korea Advanced Institute of Science andTechnology, 291 Daehak-ro, Yuseong-gu, Daejeon 305-701, Republic of Korea

*S Supporting Information

ABSTRACT: This paper describes a new label-free cell separation methodusing a magnetic repulsion force resulting from the magnetic susceptibilitydifference between cells and a paramagnetic buffer solution in amicrochannel. The difference in the magnetic forces acting on different-sized cells is enhanced by adjusting the magnetic susceptibility of thesurrounding medium, which depends on the concentration of para-magnetic salts, such as biocompatible gadolinium diethylenetriaminepentaacetic acid (Gd−DTPA), dissolved therein. As a proof-of-conceptdemonstration, Gd−DTPA solutions at concentrations of 0−80 mM wereapplied to separate U937 cells from red blood cells (RBCs) and todistinguish two different-sized polystyrene (PS) beads (8 and 10 μm indiameter). By increasing the Gd−DTPA concentration from 0 to 40 mM,the separation resolution of PS beads was increased from 0.08 to 0.91.Additionally, we successfully achieved label-free separation of U937 cells from RBCs with >90% purity and 1 × 105 cells/hthroughput using a 40 mM Gd−DTPA solution.

Efficient separation of target cells from a heterogeneouspopulation is an essential step in cell biology, immunology,

and stem cell research, as well as in clinical applications such ascancer diagnostics or transplantation.1−3 Recently, fluores-cence-activated cell sorting4,5 and magnetically activated cellseparation6,7 have become the most widely used tools for theisolation of specific target cells because of their high-throughputseparation performance. However, these methods demand acomplicated experimental setup and a relatively large sampleand reagent volume. Such requirements increase the total costof the system and reduce its availability and ease of use. Toovercome previous limitations, microfluidics-based cell separa-tion technologies have been developed. Microfluidic technol-ogies, which involve precise manipulation of a small volume offluid constrained to microchannels, offer numerous advantagessuch as reduced sample and reagent volume, rapid analysis, andimproved separation performance by high spatial resolution.8

Label-free separation techniques have received muchattention because they can greatly benefit sample preparationand postanalysis procedures. Microfluidics-based label-free cellseparation can be broadly categorized into passive and activemethods according to whether external power is applied to thesystem. For passive separation, there are several methods thatrely on the geometry of channels and hydrodynamic forces: forexample, pillar and weir structures,9,10 pinched-flow fractiona-tion,11 and deterministic lateral displacement.12 Although thesemethods are dependent on geometrical microstructures, theysuffer from clogging and require complicated, precise channeldesign. Recently, inertial forces induced by microchannels havebeen applied.13,14 This novel passive method allows high-

throughput separation; however, the use of high-velocity fluidinduces fluid shear stress, which has the potential to influencecell fate and function.15 The active separation techniquesexploit various external forces such as dielectrophoretic,16

acoustic,17 optical,18 and magnetic forces.19,20 Magnetophoresis,a migration phenomenon of particles under the influence ofmagnetic fields, is a simple manipulation technique that isunaffected by changes in solution pH, ionic concentration,surface charge, and temperature. Other active methods inflictdamage on cells by electrical Joule heating or mechanical stressfrom ultrasound. Furthermore, they often require complexelectrode designs and expensive experimental equipment, forexample, sophisticated laser setups.Due to their advantages, a number of methods using

magnetophoresis to separate biological cells have been recentlyreported. Biological cells, being a nonmagnetic material, aremagnetized in the direction opposite to the applied magneticfield and are thus repulsed. The repulsive force on cells is soweak that it is not even possible to observe it. Therefore, cellsare labeled with magnetic particles and separated by anattractive force. However, there are other approaches thatincrease the repulsive force on nonmagnetic cells so as to allowmanipulation without magnetic labeling. One example is theuse of a high-field electromagnet21,22 to enhance a magneticfield gradient. The extremely high magnetic field gradient

Received: June 14, 2011Accepted: February 29, 2012Published: February 29, 2012

Article

pubs.acs.org/ac

© 2012 American Chemical Society 3075 dx.doi.org/10.1021/ac201505j | Anal. Chem. 2012, 84, 3075−3081

increases the repulsive force to such a degree that it is evenpossible to levitate a diamagnetic, live frog against gravity.However, this method requires expensive and relativelyunavailable superconducting magnets. Alternatively, the use offerrofluids23−25 and paramagnetic ions26 enhances the repulsiveforce by increasing medium magnetic susceptibility. Kose et al.utilized ferrofluids, which are colloidal mixtures of magneticnanoparticles, to manipulate and sort cells.27 Zhu et al.demonstrated label-free separation of different-sized micro-particles surrounded by ferrofluids.28 However, the opaqueproperty of ferrofluids limits their application in cells, asobservation is impossible unless a fluorescent dye is used.Solutions of paramagnetic salts such as manganese(II) andgadolinium(III) are more beneficial than ferrofluids due to theirtransparency. Such paramagnetic solutions have been widelyused for measurement of magnetic susceptibility in particles orcells,29,30 characterization of the magnetic force of particles,31

and trapping particles and cells in a paramagnetic solution usingtwo closely facing magnets.32,33 Additionally, Rodriguez-Villarreal et al. reported focusing particles and cells with arepulsive force using paramagnetic salts in a capillary.34 Mooreet al. demonstrated improved separation of attractive-mode andrepulsive-mode particles by addition of paramagnetic salts tothe quadrupole magnetic flow sorter.35 Despite theseapplications, a concern regarding biocompatibility remainsbecause their much smaller magnetic susceptibility, comparedwith that of ferrofluids, results in the need for a highconcentration of paramagnetic salts to induce the requiredmagnetic movement of cells.To avoid the use of high paramagnetic salt concentrations

and simply apply magnetic fields to the poly(dimethylsiloxane)(PDMS) microchannels only with permanent magnets, micro-fabricated ferromagnetic structures have been embedded in thesystem.36−38 This allows the concentration of magnetic fluxlines from external permanent magnets, thereby providing astrong magnetic field gradient.Here, we report development of a label-free microfluidic cell

separation based on the difference in magnetic repulsive forcesbetween two different-sized cells in a paramagnetic saltsolution. To understand the effect of paramagnetic salts ondifferent-sized cells, we investigated the magnetic migration ofmonodisperse polystyrene (PS) beads and U937 cells byvarying the gadolinium diethylenetriamine pentaacetic acid(Gd−DTPA) concentration. Improvement of cell separationresolution was also investigated by changing the concentrationof paramagnetic salts.Theory. The magnetic force acting on a particle, including a

cell, is given by eq 1:39

=χ − χ ∇·

μ

VF

B B( ) ( )mag

p m p

0 (1)

where Vp is the volume of the particle, χp is the volumetricmagnetic susceptibility of particle, χm is the volumetricmagnetic susceptibility of the surrounding medium, B is themagnetic flux density, ∇·B is the strength of the magnetic fluxdensity gradient, and μ0 is the vacuum permeability. From eq 1,the parameters that influence magnetic force strength are theparticle volume, the magnetic flux density and its gradient, andthe magnetic susceptibility of both the particle and the medium.Traditionally, the magnetic susceptibility of particles has been

dramatically enhanced by labeling with magnetic materials toincrease the magnetic force strength. Alternatively, the

magnetic field gradient may be amplified using powerfulsuperconducting magnets. In this study, instead of magnetictags for cell labeling, we enhanced the magnetic susceptibility ofthe medium to increase the magnetic force acting on the cells.There are two possible ways to change the magneticsusceptibility of the medium. One is to make the mediummore diamagnetic than the cells using diamagnetic salts.However, there are few chemicals that are more diamagneticthan water. Though bismuth(III) subsalicylate is morediamagnetic than water, it dissolves under strong acidconditions, thereby making it incompatible with biologicalsamples. Therefore, the magnetic susceptibility of the mediumwas tuned by dissolving biocompatible paramagnetic salts suchas Gd−DTPA in the medium. The relationship betweenmedium magnetic susceptibility and paramagnetic salt concen-tration is given by eq 2:40

χ = − ×−−C(0.35 9.04) 10m Gd DTPA

6(2)

where CGd−DTPA is the concentration of Gd−DTPA (mM).Thus, the medium magnetic susceptibility may be linearlycorrelated with paramagnetic solution concentration.Scheme 1 illustrates magnetophoretic cell separation using a

magnetic repulsion force in a paramagnetic solution. The

particle flow aligns along the wall near a nickel (Ni)microstructure, where the maximum magnetic field gradient iscreated. When a magnetic field perpendicular to the directionof flow is applied by a permanent magnet placed next to the Nimicrostructure, cells in a paramagnetic solution are deflectedlaterally due to the magnetic repulsion force. Contrary toconventional magnetophoresis using a magnetic attractionforce, we exploited magnetic repulsion of the particle fromthe magnet. Because the medium is a paramagnetic solution,the net magnetic susceptibility of diamagnetic particles andmedium becomes less than zero, and thus the particles are

Scheme 1. Schematic Illustration of Magnetophoretic CellSeparation Using a Magnetic Repulsion Force in aParamagnetic Solutiona

aWhen a magnetic field perpendicular to the direction of flow isapplied by a permanent magnet placed next to the Ni microstructure,cells in a paramagnetic solution are deflected laterally due to themagnetic repulsion force. The separation efficiency is greatly enhancedby changing medium magnetic susceptibility from χm1 to χm2 via use ofparamagnetic salt solutions.

Analytical Chemistry Article

dx.doi.org/10.1021/ac201505j | Anal. Chem. 2012, 84, 3075−30813076

repelled from the magnet. Although particle volume is also aseparation parameter, it is very difficult to distinguish twodifferent-sized particles by only their size difference. However,if the magnetic susceptibility of the medium increases, the tinydifference in the forces acting on two different-sized particlesmay increase according to eq 1. This theoretical backgroundsuggests that we could achieve more efficient size-based label-free particle or cell separation by altering medium magneticsusceptibility via use of paramagnetic salt solutions.

■ EXPERIMENTAL SECTIONDevice Design and Fabrication. A microfluidic device for

magnetophoretic separation was fabricated by PDMS (Sylgard184, Dow Corning, MI) molding processes. A mixture ofPDMS prepolymer and curing agent was poured on thephotoresist mold and cured for 1 h at 70 °C on the hot plate.The cured PDMS channel was aligned and bonded to thenickel-embedded PDMS membrane under a microscope. Asshown in Figure 1a, the width and length of the separation

region in the microchannel were 300 μm and 12 mm,respectively, whereas those of the expansion region were 1and 4 mm, respectively. The overall height of the channel wasdesigned to be 25 μm. Although there have been manyattempts to use complicated Ni microstructures to enhancemagnetic field and its gradient, here we applied one of thesimplest and routinely used Ni microstructures (Figure S1 inthe Supporting Information), which we have reported onpreviously.37,38

Experimental Setup. Figure 1b shows the experimentalsetup for magnetophoretic separation. Sample and focusingflows were injected into the microfluidic device using a syringepump (Pump 11 Pico Plus; Harvard Apparatus, Inc., MA). Toapply a magnetic field, a permanent magnet (NdFeB35, w × l ×h = 25 × 50 × 10 mm3; Magtopia, Korea) was placed 1 cmfrom the Ni microstructure. Particle images were taken with aCCD camera (DS-2MBWc; Nikon Co., Japan) attached to aninverted optical microscope (TS100; Nikon Co.). The captured

images were analyzed using image analysis software (ImageJ;http://rsb.info.nih.gov/ij/).

Sample Preparation. Plain PS beads with diameters of 8and 10 μm (Sigma-Aldrich Co., St. Louis, MO) were used forlabel-free magnetophoretic separation. Gd−DTPA (Sigma-Aldrich Co.) was dissolved in phosphate-buffered saline(PBS) and the mixture was adjusted to pH 7.2. The beadswere prepared in 0.2% Tween 20 Gd−DTPA buffer atconcentrations of 0, 20, 40, 60, and 80 mM for both 8 and10 μm beads. Human histolytic lymphoma monocytes (U937)were cultured in RPMI 1640 medium (Gibco, Rockville, MD)supplemented with 20% fetal bovine serum and 1% penicillin/streptomycin. The cells were grown at 37 °C under ahumidified atmosphere of 5% CO2. To prepare deoxyhemo-globin red blood cells (RBCs), RBCs were treated with 3 mMisotonic sodium hydrosulfite (Fisher Scientific, Pittsburgh, PA)solution for 30 min at room temperature. DeoxyhemoglobinRBCs were diluted with PBS buffer at a ratio of 1:10. Theconcentration of U937 cells was adjusted to 1 × 108 cells/mL.RBC and U937 cell samples were prepared in Gd−DTPAbuffer at concentrations of 0, 9.5, 20, 40, and 80 mM. Forseparation, a sample mixture was prepared by mixing U937 cellsand RBCs at a ratio of 1:400; the final cell concentration was 8× 107 cells/mL. To distinguish U937 cells from RBCs, U937cells were stained with a cell membrane-permeable fluorescentdye (CellTracker; Invitrogen, Carlsbad, CA) by incubation at37 °C for 30 min.

Computer Simulation. Numerical analysis of magnetic fluxdensity at the cross section of the microfluidic device wassolved by the finite-element method magnetic (FEMM)program (Aladdin Enterprises, Menlo Park, CA) in a two-dimensional analysis. Parameters of the permanent magnet andboundary conditions used for solving the problem wereprovided by the FEMM program, and pure Ni permeabilitywas as described previously.41 Particle lateral deflection by themagnetic repulsive force was calculated using the MATLAB 7.0program (The MathWorks, MA). Some parameters, such as theinitial lateral position and the magnetic field exposure time,were obtained experimentally (see the Supporting Informa-tion).

Data Acquisition. Samples were injected into the micro-fluidic device at a flow rate of 0.32 μL/min. All particles or cellsinjected were focused within 100 μm from the magnet side inthe separation region. As the particles passed through the longseparation region, they experienced a magnetic repulsive force.The lateral deflection distance caused by the magnetic forcewas measured at the 1 mm wide expansion region from themagnet side of the microchannel. The separation resolution(Rs) between two particles was defined as Rs = 0.5(Δx/(σ1 +σ2)), where Δx is the distance between the average position oftwo particles and σ is the standard deviation of each particleposition.

■ RESULTS AND DISCUSSIONSimulation Analysis. To simulate the magnetic lateral

deflection under different Gd−DTPA concentrations, adistribution of the magnetic flux density across the separationregion in a microfluidic device should first be estimated. Due tothe low magnetic force induced by a permanent magnet alone,we introduced a ferromagnetic Ni microstructure to our system.The distribution of a magnetic field gradient enhanced by theNi microstructure was simulated using FEMM in a two-dimensional analysis. The numerical analysis of the magnetic

Figure 1. Configuration of the experimental setup for magneto-phoretic separation: (a) dimensions of the designed microfluidicdevice; (b) schematic diagram of the experimental setup.

Analytical Chemistry Article

dx.doi.org/10.1021/ac201505j | Anal. Chem. 2012, 84, 3075−30813077

flux density at the cross section of the separation channel region(300 μm width) is shown in Figure 2a. There was no noticeable

change along the lateral separation region with a permanentmagnet alone, whereas a significant enhancement effect ofmagnetic flux density was observed with the Ni microstructure.The magnetic field strength exponentially decreased as thedistance from the Ni microstructure increased. On the basis ofthese numerical magnetic field analysis and eq 1, we calculatedthe magnetic lateral deflection position of particles at differentGd−DTPA concentrations.Here the magnetic field exposure time and the initial

focusing positions of the particles could be obtainedexperimentally: field exposure time = 16.8 s; initial y-axislateral position of PS beads = 13.5 μm in the separation regionand 45 μm in the expansion region; initial y-axis lateral positionof cells = 18 μm in the separation region and 60 μm in theexpansion region. The magnetic susceptibility of PS beads wasassumed to be −8.21 × 10−6.33 As shown in Figure 2b, therewas no lateral shift, and both PS beads come out to the sameposition of initial point (45 μm) at 0 mM Gd−DTPA.However, the difference between the lateral positions of thedifferent-sized particles increased with Gd−DTPA concen-tration, as we expected theoretically. Nevertheless, the lateraldisplacement difference between the 8 and 10 μm particles didnot increase linearly with Gd−DTPA concentration butplateaued in the relatively high concentration range. Thisnonlinear relationship might be due to the nonuniform

magnetic flux density, which exponentially decreased as thedistance between the Ni microstructure and the particleincreased.To calculate the cell lateral displacement against different

Gd−DTPA concentrations, we assumed that U937 cells have asimilar magnetic susceptibility to white blood cells. From theliterature,19 the magnetic susceptibility of white blood cells wasfound to be −9.02 × 10−6. The hydrodynamic diameters ofU937 cells and RBCs were 10 and 6 μm, respectively. As shownin Figure 2b, U937 cells showed similar behavior to 10 μm PSbeads overall, although their magnetic susceptibility was slightlydifferent. This indicates that the size effect may be moreimportant than magnetic susceptibility due to the inherentlylow magnetic susceptibility of particles. At 0 mM Gd−DTPA,U937 cells were not influenced by the applied magnetic fieldand remained in their initial focusing position (60 μm),whereas RBCs were slightly shifted toward the magnet. ThisRBC movement might be due to the larger magneticsusceptibility of RBCs than of PBS buffer, which may resultin attraction toward the higher magnetic field. The lateralposition difference between cells was more evident than that ofPS beads because the size difference between cells (4 μm) wasgreater than that between PS beads (2 μm).

Displacement of Polystyrene Beads. As a proof-of-concept demonstration, we selected 8 and 10 μm diameterplain PS beads as a substitute for biological cells. Figure 3

shows the lateral deflection of the particles with different Gd−DTPA concentrations. The experimental results showed goodagreement with those of the simulation. As the concentration ofGd−DTPA increased, both the lateral displacement ofindividual beads and the distance between two PS beadsgradually increased. To quantitatively evaluate the effect of saltconcentration, the resolution of separation of two different-sized beads at different Gd−DTPA concentrations wascalculated according to the mean lateral displacement and thestandard deviation of each particle position. The separationresolution for 8 and 10 μm PS beads with 0, 20, 40, 60, and 80mM Gd−DTPA were 0.08, 0.62, 0.91, 0.65, and 0.47,respectively. The separation resolution for PS beads increasedat 0−40 mM Gd−DTPA but decreased at concentrationshigher than 40 mM Gd−DTPA due to changes in variousparameters that determine magnetophoretic mobility. Aspreviously described, the increased medium magnetic suscept-ibility induced by addition of paramagnetic salts could increase

Figure 2. (a) Numerical analysis of the magnetic field at the crosssection (A−A′) of the separation region in the microfluidic deviceshown in Figure 1. (b) Theoretical estimation of the magneticdisplacement of polystyrene (PS) beads and two cell types in theexpansion channel at different Gd−DTPA concentrations.

Figure 3. Measured lateral positions and separation resolutions (Rs) oftwo different-sized polystyrene (PS) beads in the expansion regionaccording to Gd−DTPA concentration.

Analytical Chemistry Article

dx.doi.org/10.1021/ac201505j | Anal. Chem. 2012, 84, 3075−30813078

both the lateral displacement of two particles and the gapbetween two beads. Accordingly, at concentrations lower than40 mM Gd−DTPA, the lateral deflection and the distancebetween two different-sized PS beads also increased, leading toan increased separation resolution. In contrast, at higher Gd−DTPA concentrations, the resolution was decreased regardlessof the increased medium magnetic susceptibility, resulting indecreased separation resolution. These phenomena showexcellent agreement with our expectations based on thesimulation analysis. As shown in Figure 2a, the magnetic fluxdensity exponentially decreased as the distance from the Nimicrostructure increased. In the case of high Gd−DTPAconcentration, the beads were repelled beyond an effectivemagnetic field region (<100 μm) where the magnetic force wassufficiently strong. These results indicate that more efficientseparation of different-sized particles might be possible if othermicrofluidic devices, which are more complicated but form auniform magnetic field over the entire area of the microchannel,utilized this concept. However, our device provides one of thesimplest magnetophoretic microfluidic device formats and aresolution sufficient for separation of target cells.According to these results, the optimum paramagnetic salt

concentration for separation in our magnetophoretic device was40 mM Gd−DTPA. The measured distributions of two types ofPS bead at 0 and 40 mM Gd−DTPA both with and withoutapplication of magnetic fields are shown in Figure 4. With 0mM Gd−DTPA, there was no significant difference in thedistribution of lateral positions of PS beads regardless of theapplication of a magnetic field. This was likely because of thenegligibly small difference in the magnetic susceptibility of themedium and beads (−1.02 × 10−6), which induces littlemagnetic force difference. In contrast, at 40 mM Gd−DTPA,both PS beads were laterally deflected under a magnetic field,and the degree of their displacement differed, facilitatingdiscrimination.Separation of U937 Cells from Red Blood Cells. To

determine the optimum concentration for separation of U937

cells and RBCs, we measured the lateral deflection of both cellsaccording to Gd−DTPA concentration. The measured valuesof the cells’ lateral positions are shown in Figure 5. The

experimental results displayed good agreement with those ofthe simulation. However, with 0 mM Gd−DTPA, no lateralshift of either cell type occurred due to the negligible magneticforce, whereas the simulation results estimated RBC movementtoward the higher magnetic field gradient as a result of the weakparamagnetic property of RBCs. This result implies that ourmagnetophoretic cell separation was achieved under relativelyweak magnetic fields compared with previous RBC separationstudies that involved enhancement of the weak paramagneticproperty under a magnetic field by adding a Ni microstructure,which is more complex.19,42 As the Gd−DTPA concentrationincreased, lateral deflection of both cells occurred, and thedistance between the two cell types increased rapidly.To quantify separation, the resolution of each Gd−DTPA

concentration for U937 cells and RBCs was considered. The

Figure 4. Measured distribution of polystyrene (PS) particles of different sizes in (a) 0 and (b) 40 mM Gd−DTPA solutions under a magnetic field.Control data (without a magnetic field) are shown in the top panel.

Figure 5. Measured lateral positions and separation resolution (Rs) ofU937 cells and RBCs in the expansion region against Gd−DTPAconcentration.

Analytical Chemistry Article

dx.doi.org/10.1021/ac201505j | Anal. Chem. 2012, 84, 3075−30813079

separation resolution of 9.5, 20, 40, and 80 mM Gd−DTPAconcentration was 0.28, 0.51, 0.71, and 0.51, respectively. Theseparation resolution of two cells is lower than that of PS beads,regardless of the bigger size difference. This might be due to thelarge deviation in the mean position of U937 cells because oftheir heterogeneous size. On the other hand, the osmoticpressure effect of high Gd−DTPA concentrations on biologicalcells might also cause some limitations in cell separation. As amagnetic resonance imaging contrast agent approved by theFood and Drug Administration,43 Gd−DTPA is itself benign tobiological cells. However, higher Gd−DTPA concentrationsbrought about stresses to the cells resulting from increasedosmotic pressure. The larger standard deviation at 80 mM thanat 20 and 40 mM Gd−DTPA might be due to these osmoticpressure effects.32 To minimize this, an osmotic pressure similarto mammalian cells, 40 mM Gd−DTPA, with the pH adjustedto 7.2, was selected as the most appropriate concentration forthe separation of U937 cells and RBCs.Figure 6a shows the measured position distribution of two

cells at 0 and 40 mM Gd−DTPA under a magnetic field. With

0 mM Gd−DTPA, there was no significant difference in theposition distribution profile after application of a magnetic field.In contrast, at 40 mM Gd−DTPA, both U937 cells and RBCsshifted laterally, and U937 cells moved more rapidly thanRBCs. Most RBCs were displaced less than 250 μm, whereasmost U937 cells were displaced more than 250 μm. On thebasis of these results, U937 cells were separated from themixture with the 250 μm line as a separation boundary using 40mM Gd−DTPA. A microscopic image of separation of U937cells from RBCs is shown in Figure 6b. Thus, we successfullyachieved label-free separation of U937 cells from RBCs with>90% purity and 1 × 105 cells/h throughput at a volumetricflow rate of 0.32 μL/min. The separation purity and throughputcould be improved using a ferromagnetic microstructure with ahigher magnetic flux density gradient, using stronger magnets(such as electromagnets), or using alternative paramagnetic andbiocompatible salts.

Figure 6. Cell distribution at (a) 0 and (b) 40 mM under a magnetic field. Control data (without a magnetic field) are shown in the top panel. (c)Microscopic image of U937 cells separated from RBCs at 40 mM Gd−DTPA.

Analytical Chemistry Article

dx.doi.org/10.1021/ac201505j | Anal. Chem. 2012, 84, 3075−30813080

■ CONCLUSIONSIn conclusion, we achieved label-free magnetophoreticseparation of cells based on size differences using paramagneticsalts. To investigate this method, Gd−DTPA paramagneticsolutions at concentrations of 0−80 mM were utilized toseparate U937 cells and RBCs and 8 and 10 μm diameter PSbeads. Separation performance was evaluated based onseparation resolution, and the optimal Gd−DTPA concen-tration for the most efficient separation of different-sizedparticles was calculated. At 40 mM Gd−DTPA concentration,U937 cells could be most efficiently separated from RBCs. Thisnovel approach represents a useful tool for label-free magneto-phoretic separation of microparticles, including biological cells.Furthermore, this concept could also be applied to the isolationof white blood cell subpopulations or to blood cleansing withimproved resolution.

■ ASSOCIATED CONTENT*S Supporting InformationDetails of the magnetophoretic device setup (Figure S1) andsimulation study. This material is available free of charge via theInternet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: jekyun@kaist.ac.kr. Phone: +82-42-350-4315. Fax:+82-42-350-4310.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis research was supported by a National Leading ResearchLaboratory Program Grant (No. 2011-0018607), a Nano/BioScience and Technology Program Grant (No. 2011-0002188),and a Converging Research Center Program Grant (ProjectNo. 2011K000864) through the National Research Foundationof Korea funded by the Ministry of Education, Science andTechnology (MEST).

■ REFERENCES(1) Battistini, L.; Piccio, L. R. B.; Bach, S.; Galgani, S.; Gasperini, C.;Ottoboni, L.; Ciabini, D.; Caramia, M. D.; Bernardi, G.; Laudanna, C.;Scarpini, E.; McEver, R. P.; Butcher, E. C.; Borsellino, G.; Constantin,G. Blood 2003, 101, 4775−4782.(2) Thomas, F. D. S. R.; Clift, R. A.; Fefer, A.; Johnson, F. L.;Neiman, P. E.; Lerner, K. G.; Glucksberg, H.; Buckner, C. D. N. Engl. J.Med. 1975, 292, 895−902.(3) Muller, M.; Fleischmann, B. K.; Selbert, S.; Ji, G. J.; Endl, E.;Middeler, G.; Muller, O. J.; Schlenke, P.; Frese, S.; Wobus, A. M.;Hescheler, J.; Katus, H. A.; Franz, W. M. FASEB J. 2000, 14, 2540−2548.(4) Shapiro, H. M. Practical Flow Cytometry, 4th ed.; Wiley-Liss:Hoboken, NJ, 2003.(5) Mattanovich, D.; Borth, N. Microb. Cell Fact. 2006, 5, 12.(6) Miltenyi, S.; Muller, W.; Weichel, W.; Radbruch, A. Cytometry1990, 11, 231−238.(7) Safarik, I.; Safarikova, M. J. Chromatogr., B 1999, 722, 33−53.(8) Whitesides, G. M. Nature 2006, 442, 368−373.(9) Brody, J. P.; Osbom, T. D.; Forster, F. K.; Yager, P. Sens.Actuators, A 1996, 54, 704−708.(10) Crowley, T. A.; Pizziconi, V. Lab Chip 2005, 5, 922−929.(11) Yamada, M.; Nakashima, M.; Seki, M. Anal. Chem. 2004, 76,5465−5471.

(12) Inglis, D. W.; Herman, N.; Vesey, G. Biomicrofluidics 2010, 4,024109.(13) Di Carlo, D.; Irimia, D.; Tompkins, R. G.; Toner, M. Proc. Natl.Acad. Sci. U.S.A. 2007, 104, 18892−18897.(14) Lee, W. C.; Bhagat, A. A. S.; Huang, S.; Van Vliet, K. J.; Han, J.;Lim, C. T. Lab Chip 2011, 11, 1359−1367.(15) Kumar, A.; Bhardwaj, A. Biomed. Mater. 2008, 3, 034008.(16) Chang, S.; Cho, Y.-H. Lab Chip 2008, 8, 1930−1936.(17) Thevoz, P. F.; Adams, J. D.; Shea, H.; Bruus, H.; Soh, H. T.Anal. Chem. 2010, 82, 3094−3098.(18) Wang, M. M.; Tu, E.; Raymond, D. E.; Yang, J. M.; Zhang, H.;Hagen, N.; Dees, B.; Mercer, E. M.; Forster, A. H.; Kariv, I.;Marchand, P. J.; Butler, W. F. Nat. Biotechnol. 2004, 23, 83−87.(19) Jung, J.; Han, K.-H. Appl. Phys. Lett. 2008, 93, 223902.(20) Smistrup, K.; Hansen, O.; Bruus, H.; Hansen, M. J. Magn. Magn.Mater. 2005, 293, 597−604.(21) Beaugnon, E.; Tournier, R. Nature 1991, 349, 470.(22) Berry, M. V.; Geim, A. K. Eur. J. Phys. 1997, 18, 307−313.(23) Rosensweig, R. E. Ferrohydrodynamics; Carmbridge UniversityPress: Cambridge, MA, 1985.(24) Helgesen, G.; Pieranski, P.; Skjeltorp, A. Phys. Rev. A 1990, 42,7271−7280.(25) Hwang, J. Y.; Takayasu, M.; Friedlaender, F. J.; Kullerud, G. J.Appl. Phys. 1984, 55, 2592−2594.(26) Winkleman, A.; Perez-Castillejos, R.; Gudiksen, K. L.; Phillips, S.T.; Prentiss, M.; Whitesides, G. M. Anal. Chem. 2007, 79, 6542−6550.(27) Kose, A. R.; Fischer, B.; Mao, L.; Koser, H. Proc. Natl. Acad. Sci.U.S.A. 2009, 106, 21478−21483.(28) Zhu, T.; Marrero, F.; Mao, L. Microfluid. Nanofluid. 2010, 9,1003−1009.(29) Gill, S. J.; Malone, C. P.; Downing, M. Rev. Sci. Instrum. 1960,31, 1299−1303.(30) Russell, A. P.; Evans, C. H.; Westcott, V. C. Anal. Biochem. 1987,164, 181−189.(31) Zimmels, Y.; Yaniv, I. IEEE Trans. Magn. 1976, 12, 359−368.(32) Winkleman, A.; Gudiksen, K. L.; Ryan, D.; Whitesides, G. M.;Greenfield, D.; Prentiss, M. Appl. Phys. Lett. 2004, 85, 2411−2413.(33) Watarai, H.; Namba, M. J. Chromatogr. 2002, 961, 3−8.(34) Rodriguez-Villarreal, A. I.; Tarn, M. D.; Madden, L. A.; Lutz, J.B.; Greenman, J.; Samitier, J.; Pamme, N. Lab Chip 2011, 11, 1240−1248.(35) Moore, L. R.; Milliron, S.; Williams, P. S.; Chalmers, J. J.;Margel, S.; Zborowski, M. Anal. Chem. 2004, 76, 3899−3907.(36) Xia, N.; Hunt, T. P.; Mayers, B. T.; Alsberg, E.; Whitesides, G.M.; Westervelt, R. M.; Ingber, D. E. Biomed. Microdevices 2006, 8,299−308.(37) Hahn, Y. K.; Park, J.-K. Lab Chip 2011, 11, 2045−2048.(38) Hahn, Y. K.; Jin, Z.; Kang, J. H.; Oh, E.; Han, M. K.; Kim, H. S.;Jang, J. T.; Lee, J. H.; Cheon, J.; Kim, S. H.; Park, H. S.; Park, J. K.Anal. Chem. 2007, 79, 2214−2220.(39) Pamme, N.; Manz, A. Anal. Chem. 2004, 76, 7250−7256.(40) Zhang, H.; Moore, L. R.; Zborowski, M.; Williams, P. S.; Margel,S.; Chalmers, J. J. Analyst 2005, 130, 514−527.(41) Suzuki, T.; Baba, H.; Matsumoto, E. Int. J. Appl. Eletromagn.Mech. 2001, 13, 307−310.(42) Adams, J. D.; Kim, U.; Soh, H. T. Proc. Natl. Acad. Sci. U.S.A.2008, 105, 18165−18170.(43) Weinmann, H. J.; Laniado, M.; Mutzel, W. Physiol. Chem. Phys.Med. NMR 1984, 16, 167−172.

Analytical Chemistry Article

dx.doi.org/10.1021/ac201505j | Anal. Chem. 2012, 84, 3075−30813081