Inkjet-Print Micromagnet Array on Glass Slides for Immunomagnetic

Enrichment of Circulating Tumor Cells

PENG CHEN,1 YU-YEN HUANG,3 GAURI BHAVE,1 KAZUNORI HOSHINO,2 and XIAOJING ZHANG1,3

1Department of Biomedical Engineering, University of Texas at Austin, Austin, TX 78712, USA; 2Department of BiomedicalEngineering, University of Connecticut, Storrs, CT 06269, USA; and 3Thayer School of Engineering, Dartmouth College,

Hanover, NH 03755, USA

(Received 30 April 2015; accepted 11 August 2015)

Associate Editor Jennifer West oversaw the review of this article.

Abstract—We report an inkjet-printed microscale magneticstructure that can be integrated on regular glass slides for theimmunomagnetic screening of rare circulating tumor cells(CTCs). CTCs detach from the primary tumor site, circulatewith the bloodstream, and initiate the cancer metastasisprocess. Therefore, a liquid biopsy in the form of capturingand analyzing CTCs may provide key information for cancerprognosis and diagnosis. Inkjet printing technology providesa non-contact, layer-by-layer and mask-less approach todeposit defined magnetic patterns on an arbitrary substrate.Such thin film patterns, when placed in an external magneticfield, significantly enhance the attractive force in the near-field close to the CTCs to facilitate the separation. Wedemonstrated the efficacy of the inkjet-print micromagnetarray integrated immunomagnetic assay in separatingCOLO205 (human colorectal cancer cell line) from wholeblood samples. The micromagnets increased the captureefficiency by 26% compared with using plain glass slide asthe substrate.

Keywords—Inkjet printing, Micromagnets, Circulating

tumor cells, Microfluidic, Immunomagnetic cell separation.

INTRODUCTION

Circulating tumor cells (CTCs) are the cells thathave shed into the vasculature from a primary tumorsite, and circulate in the bloodstream. Detecting andanalyzing CTC have been an active research area be-cause of their clinical values in assisting cancer diag-nosis and prognosis.27 To overcome the challengesassociated with the rareness of the CTC, a variety oftechnologies have been developed to separate the CTC

from interfering background hematocyte cells.3,19 Oneeffective approach is the immunomagnetic assay,which works by labeling the cancer cells with magnetictags through cancer-specific antibodies, and usingpermanent magnets to drive the labeled cancer cells forseparation.8

Traditional immunomagnetic assay uses permanentmagnets (in the scale of centimeter or millimeter) as themagnetic flux source. They are limited in the lowmagnetic field gradient, which might cause cell losses.Limitation also comes from the low density of effectivemagnetic traps, which might lead to cell aggregation.The aggregation may impact the structural integrityand fluorescent signals of the CTC, and interfere withcell imaging and identifying. To advance theimmunomagnetic assay, especially for rare cell detec-tion applications, integrations of microfluidic chip andmicroscale magnetic flux source have been pursued.1

A NdFeB hard magnetic film is deposited on siliconsubstrate using thermal patterning technique, and isused to trap cells functionalized with super paramag-netic nanoparticles.30 A shrink induced magnetic trapis formed on a shape memory polymer and integratedinto a microfluidic device to extract DNA samples forqPCR studies.15 Imprinting technology is used topattern magnetic microparticles on polymer substrateusing a magnetic grid as the template. The imprintedstructures can be used to trap magnetically-labeledfibroblast cells.4 Other than static micromagnet, micro-electromagnet can also be used to clean pathogensfrom human blood.29 We have recently developed apatterned thin-film micromagnet array using pho-tolithography and deposition technique, and demon-strated its impact in enhancing the capture anddistribution of the CTCs.2 Despite the differences ingeometry and fabrication methodology, these micro-

Address correspondence to Xiaojing Zhang, Department of

Biomedical Engineering, University of Texas at Austin, Austin,

TX 78712, USA. Electronic mail: john.zhang@dartmouth.edu

Annals of Biomedical Engineering (� 2015)

DOI: 10.1007/s10439-015-1427-z

� 2015 Biomedical Engineering Society

magnets work with the same principle—the micro-magnets generate localized strong magnetic field afterbeing magnetized by the external magnetic field, andincrease the magnetic force applied on the targets tofacilitate the magnetic separation process.

Inkjet printing technology, which is widely used forhard-copy documents production and high definitionphotographic printing. It is emerging as an alternativepatterning technology to traditional photolithographyand enables fabrication of two-dimensional or three-dimensional structures with sub-micrometer resolu-tion. It has been applied in various fields such aselectrochemical sensor,11 biological nanodevices,25

nanophotonic devices,18,23 and organic electronics.28 Ithas also been used as a versatile patterning techniquewith cells, polymers,12 and magnetic materials.26 Theprinting process includes the ejection of a fixed quan-tity of ink in a chamber, which is sensitive to externalvoltage, from a nozzle through a sudden, quasi-adia-batic reduction of the chamber volume via piezoelec-tric action. This sudden reduction sets up a shockwavein the ink, which causes a liquid to eject from thenozzle. The inks have to meet strict physicochemicalproperties (viscosity, surface tension, adhesion to asubstrate etc.) to achieve optimal performance andreliability of the inkjet printing process.5 This tech-nology is appealing in it being a noncontact, additivepatterning and mask-less approach. Versatile thin filmscan be directly written on the substrate, and the designcan be easily modified from batch to batch.20,24

Here we report a micromagnet array fabricatedusing inkjet-printing technology and integrate it withimmunomagnetic microchip to optimize the detectionof the rare CTCs. We first present the fabricationprinciple and detailed characterization results of theinkjet micromagnet structure. Then, we introduce theexperimental results of using the inkjet micromagnetintegrated system to separate CTCs from whole bloodsamples. We choose COLO205 cells (a type of humancolorectal cancer cell line) as the separation target.13

Figure 1a shows the principle of the inkjet micro-magnet array integrated immunomagnetic CTCscreening system. The whole blood sample, whichcontains large number of normal blood cells as well asspiked cancer cells, is mixed with magnetic nanopar-ticle suspension for labeling. The nanoparticles arefunctionalized with antibodies, such as anti-epithelialcell adhesion molecule (anti-EpCAM), and bind withCTC specifically. When the blood sample flowsthrough the microchannel, the cancer cells are mag-netically separated due to the magnetic momentumprovided by the attached magnetic nanoparticles.Meanwhile, normal blood cells, such as red blood cells(RBC) and white blood cells (WBC) flow out of themicrochannel unaffected. Figure 1b is a three-dimen-

sional illustration of the assembled device structure.Two magnetic flux sources exist in the system—exter-nal magnetic field generated by the permanent magnetsoutside the microchannel, and internal magnetic fieldgenerated by the inkjet-printed micromagnets. Spacersare introduced to the system to adjust the externalmagnetic field. The combination of the two magneticflux sources enhances the attractive interactionsbetween the cancer cells and the microchannel. To takeadvantage of the gravitational force, the microchannelis kept in the inverted orientation during the separationprocess.7

Compared with other microfluidic based CTCdetection technologies,3,9,16 the micromagnet inte-grated immunomagnetic assay retains the high sensi-tivity of magnetic separation and high specificity ofimmuno-recognition. As a patterning technique, theinkjet printing technique is a versatile approach forrapid prototyping, and can be easily integrated withglass slides. It is a more cost-effective technology thatallows the deposition with high accuracy, flexibilityand great reproducibility.6,21 In addition, the micro-magnet design can be easily modified toward differenttargets, such as different cancer types or differentcancer stages. It provides a promising platform to pushtoward clinical and translational applications.17

FIGURE 1. Inkjet-printed micromagnet integrated immuno-magnetic microchip for the detection of rare circulating tumorcells. (a) Principle of the immunomagnetic CTC separationand the micromagnet integration. (b) 3-dimensional illustra-tion of the assembled microchip device.

CHEN et al.

MATERIALS AND METHODS

Device Fabrication

Figure 2a illustrates the workflow of the micro-magnet fabrication using inkjet printing technology.Magnetic nanoparticle suspension (3 lL, fluidMAG-ARA from chemicellTM with hydrodynamic diameterof 100 nm) is mixed with 100 lL ink medium to makethe printing ink. Commercial inkjet material printer(Fuji DMP-2800 Dimatrix Materials Printer, FUJI-FILM) is used to print the ink (jetting temperature is40 �C) following a defined pattern on a standard glassslide (pre-heated to 60 �C). After printing, the glassslide is moved to a hot plate (pre-heated to 100 �C) toevaporate the printing ink liquid. The magneticnanoparticles remain on the glass slide and self-assemble the micromagnet structures. Then plasma-enhanced chemical vapor deposition (PECVD) is usedto coat the entire structure with SiO2 (10 nm thick) toprotect the patterns against the blood flows in themicrochannel. The microchannel is made by a stan-

dard molding technique using PDMS (Sylgard184,Dow Corning, Midland, MI, 10:1 pre-polymer to curingagent). UV-patterned SU-8 photoresist (MicroChem,Newton, MA) on a silicon wafer is used as the master.After removal from the master template, the PDMSchip is manually cut into a proper size and bondedwith the glass slide that has micromagnets on thesurface to form the microfluidic chamber for CTCscreening.

We measure the viscosity of the printing ink med-ium in advance to meet the requirements of the inkjetprinter. Ink medium is the mixture of glycerol and DIwater. We prepare ink samples with different volumeratios (1:0.8, 1:0.75, 1:0.7, 1:0.64, 1:0.6) between glyc-erol (Glycerol, ‡99% from Sigma-Aldrich) and DIwater. When the concentration of glycerol is higher,the viscosity becomes larger. The viscosities of thesamples are measured using vibrational viscometer(SV-10 vibration viscometer, A&D Company, Japan) atthe jetting temperatures (40 �C). At this temperature,the viscosity of the medium with 1:0.75 mix ratio is 8[cP], and it’s in the working range of the printer. Weselect this medium as the printing ink.

The cartridge voltage is set to be 40 V. The piezo-jetting of the cartridge is controlled by a customizedwaveform, which contains four segments as shown inFig. 4b. Each segment has three important properties:level, duration and slew rate. The level value is a per-centage of the cartridge voltage. The level values insegment one and two have the most impact on thejetting process. Changing duration of segment one andslew rate and/or duration of segment two changes dropformation largely. The applied voltage relates directlyto the volume of the pumping chamber. Faster changesin voltage change the volume faster, bigger changes involtage cause bigger volume changes. The slew ratedetermines the rate of the volume change. The waveform we use is shown in Table 1.

Cell Screening Experiment

The cultured COLO205 cell suspension is first mixedand incubated with trypsin (0.05% Trypsin–EDTA(1X), Phenol Red, Life Technology) for 5 min to breakthe cell clusters and ensure the cells flow through themicrochannel individually. Observe the cells under themicroscope until over 90% of the cells are individuallydispersed, otherwise increase the incubation time a fewmore minutes, and check for dissociation every 30 s.Then same amount of cell culture medium (RPMI 1640with 5% fetal bovine serum) is added to the suspensionto neutralize the trypsin. A cell suspension (10–20 lL)containing approximately ~150 cells is spiked into2.5 mL aliquot of blood samples acquired from heal-thy donors. Then, magnetic nanoparticles (Fer-

FIGURE 2. Fabrication of the micromagnet array. (a) Fabri-cation process of the micromagnet array using inkjet printingtechnology. (b) Printing waveform used to control the inkejection from the inkjet printer.

Inkjet-Print Micromagnet Array on Glass Slides

rofluidTM, Janssen Diagnostic, J&J), which are pre-functionalized with cancer specific antibodies anti-Ep-CAM, are added to the blood samples to label theCOLO205 cells. In our previous studies, we did TEMwith the particles, and found the average diameter ofthis particle is around 100 nm. We did simulation toestimate the number of particles per cell was 2250–4500. The magnetic momentum and labeling effec-tiveness of the particles have been experimentallydemonstrated.8 PBS is used to fill the microchannelbefore introducing the blood sample to eject air bub-bles at the flow rate of 5 mL/h. The blood sample isthen driven through the microchannel at a flow rate of2.5 mL/h using a syringe pump. After the screening,PBS is introduced to wash the remaining blood and toremove unwanted cells. After flushing, 1 mL of ice-cold acetone is introduced at the flow rate of 2.5 mL/hto the channel to fix cancer cells on the substrate. Thesample slide is then disassembled, dried and stored infridge (4 �C) before staining.

Fluorescent Staining and Imaging

After the screening steps, the samples slides arefluorescently stained with anti-cytokeratin (CK, proteinin epithelial tissue, mouse anti-cytokeratin, pan-FITC,Sigma-Aldrich, St Louis, MO), anti-CD45 (found onleukocytes, AlexaFluor 568, Invitrogen, Carlsbad, CA,bound to mouse anti-human clone 9.1 made in Dr.Jonathan Uhr’s lab at the University of Texas South-western Medical Center), and DAPI (stain DNA in cellnucleus, Vectashield Mountain Medium with DPA,Vector Laboratories, Inc, Burlingame, CA). CTCs ex-hibit CK +/CD45-/DAPI + signals, while back-ground white blood cells show CK-/CD45 +/DAPI + signals. The slides are first rinsed with PBST(PBS + 0.1% Tween 20, Sigma-Aldrich, St. Louis,MO) for 5 min, and then excessive solution is removedusing Kimwipe. 300 lL blocking buffer (Boca ScienceInc., Boca Raton, FL) is added to each slide followedby a 60 min incubation at 37 �C. The slides are thenstained with anti-cytokeratin and anti-CD45 in stain-ing solution (1X PBS, 0.1% Tween-20, and 1% BSA).After another 45 min incubation at 37 �C, the slidesare rinsed with PBST thoroughly. Next, the slides arestained with DAPI. The samples are stored in 4 �Cfridge for 30 min before being observed with the in-verted microscope (IX81, Olympus, Center Valley, PA).

Capture Rate Definition

The capture rate is defined as follows: whenpreparing the cancer cell suspension for the spikedsample, the same amount of cancer cell suspension isdropped on two glass slides as control samples. Thecontrol samples do not go through the microchannel

(please note that control slides do not contain bloodcells). After the suspension gets dried on the controlslide, they are stained and observed as smears. Thecapture rate is calculated by dividing the number of thecancer cells found from the spiked samples by theaverage number of cancer cells found on the controlslides. Since the number of cancer cells in each aliquotcannot be precisely known, chances are that more cellsare spiked into the blood samples than the controlslides, which could result in a nominal over 100%capture rate. One can normalize the data to 100%, butwe choose to present the original data for comparison.

Fluorescent Magnetic Beads Screening andSpectrofluorometer Measurement

Threemicrolitre ferromagnetic fluorescentmicrobeads(2 lm in diameter, pre-conjugated with green fluorescentdyes 470/490 nm, FCM-2052-2, Spherotech, IL) suspen-sion is added to 1 mL PBS tomake the screening sample.Fluorescent emission and excitation scans are performedon a FluoroMax-4 spectrofluorometer from Horiba Sci-entific. The screeningmicrobead sample is placed in a 100lL quartz cuvette (16.100F-Q-10/Z15, Starna Cells) forfluorometer measurement. The excitation wavelength isset to 470 nm and the emission wavelength is scannedfrom 400 to 600 nm using 5 nm slit size, 1 nm incrementstep, 1000 detector gain, and 0.5 s integration time. Thesample is then flowed through the micromagnet-inte-grated immunomagnetic microchip at the flow rate of5 mL/h, and gets collected at the outlet. The emissionsignal of the processedmicrobead suspension is examinedwith the spectrofluorometer again at 400–600 nm. Thefluorescent signals are then subtracted by the backgroundsignal from the PBS solution. The depletion rate is cal-culated by comparing the relative fluorescent intensity atthe emission wavelength of the microbead sample beforeand after the screening.

Statistical Analysis

Data are reported as mean ± standard deviation ofthe mean as noted. Assuming groups have a normaldistribution and homogenous variances, the groupmeans are compared by an independent two sample Ttest. Differences are considered significant at the 95%confidence level (p< 0.05).

TABLE 1. Key parameter and waveform used to control theink ejection during the micromagnet fabrication process.

Segment Level (%) Slew rate Duration (ls)

1 0 0.36 5.376

2 100 1.35 8.384

3 67 0.6 4.928

4 40 0.8 1.344

CHEN et al.

Blood Specimen Collection

Blood samples are drawn from multiple healthydonors after obtaining informed consent under anIRB-approved protocol. Written informed consentswere obtained from all participants. This study wasapproved by the Institutional Biosafety Committee(IBC) and the Advisory Committee on HumanResearch at the University of Texas at Austin. Allexperiments were performed in accordance with thedeclaration of Helsinki. All specimens are collectedand stored in CellSave tubes (Janssen Diagnostic, J&J).

Permanent Magnet Configurations

As an important part of the microfluidic basedimmunomagnetic assay, the permanent magnet arrayserves for two purposes. It provides the externalmagnetic field for the long-range magnetic attractionto separate the CTCs from blood. Meanwhile, itmagnetizes the micromagnets inside the microchannel,which generate short-range magnetic field to retain theCTCs. Strong permanent magnets can enhance theseparation efficiency in both ways. However, it mightcause problematic aggregation of free nanoparticlesand cells in sharp areas such as the corners and edgesof the permanent magnets. This aggregation may laterinterfere with the cell identification by quenching thefluorescent signals or damaging the cell structural in-tegrity. We design a pixel permanent magnet array tointroduce a smooth magnetic field increment from theinlet to the outlet of the microchannel. Figure 3ashows the schematic of the permanent magnet array,where thirty-six pieces of N48 Neodymium magnets (1/8 inch cube, NB001-N48, Applied Magnets, Plano, TX)assemble a 6 9 6 array with alternating polarizations.Spacers are inserted between the magnets and themicrochannel to reduce the magnetic field at the frontpart of the array. The spacers are very weakly mag-netized 1 mm thick elastic sheets (AFG-13346 MagnetSheets, ProMAG Products, Marietta, OH). In ourprevious study, we measured the magnetic field in-duced by the spacer, and they are small enough to beneglected. Thus the spacers can be treated as simpledistance offset.10 It is noteworthy that the spacers onlycover the front part of the channel so that the strongmagnetic field in the later part of the channel canprevent potential CTCs loss.

We first examine the magnetic field generated by thepixel permanent magnet array using simulation tools,as is shown in Fig. 3b. The plot shows the magneticfield intensity at the bottom of the microchannel, i.e.,1 mm on top of the permanent magnet array. Themaximum value is ~0.03 T, and it’s located at the laterpart of the array, where there is no spacer. The pixel

design creates a smooth magnetic field increment fromthe front to the later part of the micorchannel.

In addition, we calculate the magnetic potentialenergy of the permanent magnet array. A potentialwell is the region surrounding a local minimum ofpotential energy. Due to entropy, an object tends to

FIGURE 3. Theoretical analyses of the magnetic field gen-erated by the pixel permanent magnets array. (a) Schematic ofthe permanent magnet array and the spacers. (b) Magneticfield at the bottom of the microchannel. (c) Magnetic potentialenergy distribution at the bottom of the microchannel.

Inkjet-Print Micromagnet Array on Glass Slides

stay in the potential well. It can be used to explain themagnetic cell separation process. Consider an object ofvolume V within a fixed magnetic field M, it possessesa magneto-static self-energy that is given by:

Ws ¼l02

Z

V

M �HMdv ð1Þ

where HM is the field in the object due to M. If amagnetized object is subject to an applied field Ha, itacquires a potential energy:

Wa ¼ �l0

Z

V

M �Hadv ð2Þ

This potential energy can be viewed as the workrequired to move the object from an environment withzero field to a region permeated by Ha. Now weconsider the model of a cell labeled with magneticnanoparticles in an external magnetic field. Due to thesmall size of a cell, the variation of the magnetic fieldapplied on the cell can be neglected. The potentialenergy can now be simplified to:

Wa ¼ �l0VMHa cos hð Þ ð3Þ

Here h is the angle between M and Ha. The mag-netization M is dependent on the applied magneticfield Ha and the volume magnetic susceptibility of thematerials, give as:

M ¼ vmHa ð4Þ

Substitute Eq. (4) into (3), the potential energy nowcan be expressed as:

Wa ¼ �l0Vvm �H2a ð5Þ

Given that l0, V, vm are all constants for the sametype of cell and nanoparticle, the potential energy istotally determined by the intensity of the magnetic fieldHa generated by the permanent magnets. The effectivemagnetic susceptibility of the cell can be calculatedusing:

vc ¼ N � R3P

R3c

� vp ð6Þ

here Rc is the hydrodynamic radius of the cell, Rp is thehydrodynamic radius of the particle, and vp is the

magnetic susceptibility of the labeling nanoparticle.Substitute Eq. (6) into (5), we get the final expressionof the magnetic potential energy of the cell as:

Wa ¼ � 4

3l0pNR3

Pvp �H2a ð7Þ

Here the magnetic permeability is l0 ¼ 4p�10�7H m�1, and the radius of the nanoparticle isRp ¼ 50 nm, and the volume susceptibility of the

nanoparticle is vp ¼ 5ðSIÞ. The number of nanoparti-

cles per cell is set to be N = 2500, which is estimatedbased on our device model and verified with experi-mental results.8

Due to the large variance in the magnetic potentialwell energy, we take a logarithmic calculate of theenergy to show the distribution, and the result is shownin Fig. 3c. The pixel permanent magnet array displaysscattered potential wells all over the microchannel dueto the alternating polarizations. In the meantime, thearray shows enhanced potential well toward the end ofthe array, which guarantees that any target cells thatmight escape from the front part can be captured be-fore flowing out of the channel.

RESULTS

Microchannel and Micromagnet Array Design

The geometry and key dimensions of the microflu-idic channel are shown in Fig. 4 (blue solid line). Theheight of the microchannel is 500 lm. Themicrochannel is designed to be a flat chamber to in-crease the effective contact area and decrease the dis-tance between the channel and the permanent magnets.The position of the permanent magnet array is shownas orange dash lines in Fig. 4. The permanent magnetsare placed with alternating polarizations to maximizethe magnetic field gradient. The layout of the inkjetmicromagnet array is illustrated by the black dots inFig. 4. The resolution, i.e., the dimension of a single

FIGURE 4. Design of the microchannel and the micromagnetarray. Blue solid line—geometry and dimensions of themicrochannel. Orange dash line—position and layout of thepermanent magnet array. Black dot array—layout and distri-bution of the inkjet micromagnets. Inset zoom-in view of themicromagnet array and the array periodicity.

CHEN et al.

micromagnet element, is determined by the volume ofa single inkjet droplet and the concentration ofnanoparticles in the ink. The periodicity of themicromagnet array (shown in the inset of Fig. 4) d isdecided by the inkjet printer cartridge, and it rangesbetween 0 and 250 lm. The dimensions of a micro-magnet determine the strength and effective range ofindividual micromagnet. The spatial periodicity plays asignificant role in adjusting the distributions of thecaptured CTCs to facilitate cell staining and imaging.According to our previous theoretical investigation ofthe micromagnet approach, we choose the dimensionof the inkjet micromagnet to be 50 lm, and micro-magnet array periodicity to be 150 lm.2

Inkjet-Printed Micromagnets Fabrication

Figure 5a shows the SEM (scanning electronmicroscope) images of the inkjet-printed micromag-nets, and the inset shows a zoom-in view of a singleprinted pattern. The magnetic nanoparticles assemblemultiple ‘‘ring-shaped’’ structures as an array on thesubstrate. Most of the nanoparticles aggregate aroundthe edges of the ring, and a small amount ofnanoparticles are scattered in the middle of the ring.The ring structure is formed because of the thermo-dynamics in the evaporation stage. The evaporationrate is higher at the edge than the center, which createsoutward convective flow and brings the nanoparticlesto the edges. Such high concentration of magnetic

nanoparticles on the edge creates a distributed localmaximum of magnetic gradient, which enhances thecapture of labeled cancer cells.

To calibrate the fabricated structure, we take imagesof the printed pattern using bright-field microscopeand perform measurement with ImageJ. The outerdiameter is determined to be 48.5 ± 3.5 lm, the innerdiameter 25.9 ± 3.1 lm, and the width of the edge is11.2 ± 1.3 lm. The dimension of the inkjet-printedmicromagnet is on the same order as a cancer cell,which indicates that only a few cells can interact fullywith each micromagnet. It helps distribute the cellsover the micromagnet array on the substrate. Theperiodicity of the printed micromagnet array is mea-sured to be 150 lm. We also characterize the printedpattern with AFM (atomic force microscope) to obtainthe morphological information. The result is shown inFig. 5b. We take a cross-section of Fig. 5b, as indi-cated by the red dash line, and the profile is shown inFig. 5c. The peak thickness at the edge of the pattern isabout 300 nm. AFM measurement also confirms thering-shaped geometry we observe in the SEM images.

Magnetic Fluorescent Microbeads ScreeningExperiments

To test the efficacy of the inkjet-printed patterns andthe immunomagnetic system, we first run experimentswith magnetic fluorescent beads. When the microbeadsolution flows through the microchannel, the mi-

FIGURE 5. Inkjet-printed micromagnets. (a) SEM of the inkjet- printed micromagets, and zoom-in view of a single micromangetelement. (b) AFM profile of a single micromagnet element. (c) Cross-section profile of a single micromagnet element showing themaximum height at the edge of the micromagnet is 300 nm.

Inkjet-Print Micromagnet Array on Glass Slides

crobead gets separated. Figure 6a shows the fluores-cent image of part of the slides after the screeningexperiments. The printed patterns work as microscalemagnetic flux sources and attract the magnetic mi-crobeads. For the experiments with the fluorescencebeads, the periodicity of 75 um is used to better visu-alize the effect of the printed pattern. We use theperiodicity of 150 lm for all the other experimentswith cancer cells. We plot the relative fluorescentintensity of the image as shown in Fig. 6b. The fluo-rescent signal profile exhibits regular repetition, andthe distance between two adjacent peaks is about75 lm, which is the value we used for the slides. Itvalidates the functionality of the printed micromagnetpatterns in generating surface magnetic field. In addi-tion, we examine and compare the fluorescent signalintensity of the sample microbead suspension beforeand after the screening experiments with spectrofluo-rometer. The depletion rate is 100%, indicating thatthe system effectively removes all the magnetic beads.

Cancer Cell Screening Experiments

We perform blood screening experiments withCOLO205 cells as the separation targets using the in-

kjet-printed micromagnet integrated immunomagneticassay. Blood samples are acquired from healthy do-nors. Before the screening, blood plasma is replacedusing a dilution buffer to keep the physiologicalproperties of the screening samples consistent. Fer-rofluid nanoparticles are used to label the cells andprovide the magnet momentum. During the screeningexperiment, flow rate is 2.5 mL/h. After the screening,we stain the slides with anti-CK/DAPI/CD45 for cellidentifying, counting and locating. We then observethe slides using bright-field and fluorescent microscoperespectively.

After the separation step, a flushing step is per-formed to remove the excessive blood from themicrofluidic channel. However, we find a small numberof background cells (RBCs, WBCs) left on the slides,due to non-specific binding between the cells and thesubstrate. Therefore, immunofluorescent staining withthree fluorescent dyes (anti-CK, anti-CD45, DAPI) isan essential step to identify the cells as discussed inseveral previous studies.8,10,14,22 Bright-field, fluores-cent, and overlay images of sample cells found on thesubstrate are shown in Fig. 7. The arrows in Fig. 7point to the micromagnets. Sample COLO205 cells arefound directly attached to the printed micromagnets,or indirectly trapped by the ferrofluid nanoparticlesaggregated around the printed patterns, as shown inthe overlay images in Fig. 7. The direct and indirectattachment between the COLO205 cells and the inkjetmicromagnets confirm the interactions between thecells and the micromagnet patterns.

DISCUSSION

Now we study the effect of the inkjet-printedmicromagnets in capturing CTCs. Same microfluidicchannel is used with two types of substrate, namelypatterned slide and plain slide. The patterned slides arestandard glass slides with the inkjet-printed patterns.Plain slides are standard glass slides exposed to thesame external magnetic field but without the patterns.On plain slides, the average capture rate is69.1 ± 12.6% (n = 4), while on the inkjet-printedslides, the average capture rate is 95.6 ± 6% (n = 4).The micromagnet patterns increase the capture rate by26% (p< 0.05) compared with plain slides. Themicromagnet-integrated system presented here is anoptimization based on our previous generation of theimmunomagnetic CTC detection microchip. In theprevious design, plain glass slide without any micro-magnet pattern and large permanent magnets wereused as the substrate. It yielded an average capture rateof 79.5 ± 24% in the screening experiments withCOLO205.8 The integration of the micromagnets and

FIGURE 6. Magnetic screening of fluorescent magnetic mi-crobeads. (a) Fluorescent image of substrate showing regulardistribution of the fluorescent magnetic beads. (b) Relativefluorescent intensity profile of the substrate. The peak-to-peak distance is 75 lm, equals to the periodicity of themicromagnet array.

CHEN et al.

pixel permanent magnet increases the capture rate by16% compared with the previous system.

To study the effect of the patterns, we divide thecaptured cells into two groups based on whether theyare captured by the printed patterns or not. The resultsare shown in Fig. 8a (blue dots: cell attached to amicromagnet, orange dots: cells not attached to amicromagnet). We find 78.5% of the captured cells arefound to be captured by the micromagnets, only 21.5%of the cells are not attached to any micromagnets.With this large portion of cells captured by the inkjetmicromagnets, it confirms the contributions of theinkjet micromagnets in enhancing the surface magneticfield and increasing the CTC capture efficiency.

Figures 8a and 8b show the locations and distribu-tion histograms of the captured COLO205 cells on twosample micromagnet and plain slides respectively. Theposition of the permanent magnet array is also shownin Fig. 8. The gray dots in Fig. 8a indicate the positionof the micromagnet array. The pixel permanent mag-net array and the spacers reduce the magnetic field atthe front of the channel, thus allowing the target cellsto be captured all over the substrate. With the previousdesign of our immunomagnetic microchip, more than95% of the cancer cells were captured at the front edgeof the permanent magnet array.2 While with the cur-rent design, this number was lowered down to 61%.

The reduced aggregation facilitates the subsequentstudies including staining and imaging.

On plain slides (Fig. 8b), more cancer cells arecaptured close to the outlet of the microchannel. Thisis because the spacers reduce the magnetic field at thefront part of the channel, and the magnetic force isthe strongest in the area close to the outlet. While onpatterned slides, more cancer cells are captured closeto the inlet of the microchannel (Fig. 8a). It is be-cause the micromagnets in the front part of themicrochannel are magnetized and increase theattractive forces applied on the cancer cells. It can beconsidered as a strong indicator of the functionalityof the micromagnets.

CONCLUSION

In this paper, we present the design and fabricationof a micromagnet structure using inkjet printing tech-nology. The outer diameter of the fabricated structureis 48.5 ± 3.5 lm, the inner diameter is 25.9 ± 3.1 lm,the width of the edge is 11.2 ± 1.3 lm, and the peakthickness is 300 nm. The micromagnets are designed togenerate sufficient magnetic force to retain the targetrare CTCs. The periodicity of the micromagnet array is150 lm, which distributes the CTCs across the sub-

FIGURE 7. COLO 205 screening experiment results using the inkjet-printed micromagnet integrated immunomagnetic assay. Thearrows point to the locations of the micromagnets. Bright-field and fluorescent panel show the interactions between the COLO205cells and the micromagnet elements.

Inkjet-Print Micromagnet Array on Glass Slides

strate to facilitate cell imaging and identification. Wethen integrate the inkjet micromagnets into theimmunomagnetic assay and perform blood screeningexperiments to separate COLO205 cells. The inkjetmicromagnets significantly increase the capture rate by26% compared with using plain glass slide as thesubstrate.

ACKNOWLEDGMENT

We thank our collaborator Professor Konstantin V.Sokolov at the University of Texas MD AndersonCancer Center and Dr. Zhigang Li at the Geisel Schoolof Medicine at Dartmouth for the supports in vali-dating the presented results. We thank Ms. Nancy

Lane and Drs. Michael Huebschman, Jonathan W.Uhr, and Eugene P. Frenkel at the University of TexasSouthwestern Medical Center for their invaluablesuggestions on the experiment design. We appreciatethe help from Professor Tim Yeh and Dr. Judy Ob-liosca at the University of Texas at Austin with thespectrofluorometer measurement. We are grateful forthe financial support from the National Institute ofHealth (NIH) National Cancer Institute (NCI) CancerDiagnosis Program under Grant 1R01CA139070.

REFERENCES

1Chen, P., Y.-Y. Huang, K. Hoshino, and X. Zhang.Multiscale immunomagnetic enrichment of circulatingtumor cells: from tubes to microchips. Lab Chip 14:446–458, 2014.2Chen, P., Y.-Y. Huang, K. Hoshino, and J. X. J. Zhang.Microscale magnetic field modulation for enhanced cap-ture and distribution of rare circulating tumor cells. Sci.Rep. 5:8745, 2015.3Chen, Y., P. Li, P.-H. Huang, Y. Xie, J. D. Mai, L. Wang,N.-T. Nguyen, and T. J. Huang. Rare cell isolation andanalysis in microfluidics. Lab Chip 14:626–645, 2014.4Dempsey, N. M., D. Le Roy, H. Marelli-Mathevon, G.Shaw, A. Dias, R. B. G. Kramer, L. Viet Cuong, M.Kustov, L. F. Zanini, C. Villard, K. Hasselbach, C.Tomba, and F. Dumas-Bouchiat. Micro-magneticimprinting of high field gradient magnetic flux sources.Appl. Phys. Lett. 104:262401, 2014.5Fuller, S. B., E. J. Wilhelm, and J. M. Jacobson. Ink-jetprinted nanoparticle microelectromechanical systems. J.Microelectromech. Syst. 11:54–60, 2002.6Gonzalez-Macia, L., A. Morrin, M. R. Smyth, and A. J.Killard. Advanced printing and deposition methodologiesfor the fabrication of biosensors and biodevices. Analyst135:845–867, 2010.7Hoshino, K., P. Chen, Y.-Y. Huang, and X. Zhang.Computational analysis of microfluidic immunomagneticrare cell separation from a particulate blood flow. Anal.Chem. 84:4292–4299, 2012.8Hoshino, K., Y.-Y. Huang, N. Lane, M. Huebschman, J.W. Uhr, E. P. Frenkel, and X. Zhang. Microchip-basedimmunomagnetic detection of circulating tumor cells. LabChip 11:3449–3457, 2011.9Hou, H. W., M. E. Warkiani, B. L. Khoo, Z. R. Li, R. A.Soo, D. S. W. Tan, W. T. Lim, J. Han, A. A. S. Bhagat,and C. T. Lim. Isolation and retrieval of circulating tumorcells using centrifugal forces. Sci. Rep. 3:1259, 2013.

10Huang, Y. Y., K. Hoshino, P. Chen, C. H. Wu, N. Lane,M. Huebschman, H. Liu, K. Sokolov, J. W. Uhr, and E. P.Frenkel. Immunomagnetic nanoscreening of circulatingtumor cells with a motion controlled microfluidic system.Biomed. Microdevices 15(4):673–681, 2013.

11Jensen, G. C., C. E. Krause, G. A. Sotzing, and J. F.Rusling. Inkjet-printed gold nanoparticle electrochemicalarrays on plastic. Application to immunodetection of acancer biomarker protein. Phys. Chem. Chem. Phys.13:4888–4894, 2011.

12Kim, J. D., J. S. Choi, B. S. Kim, Y. C. Choi, and Y. W.Cho. Piezoelectric inkjet printing of polymers: stem cell

FIGURE 8. Locations and distributions of the capturedCOLO 205 cells on (a) inkjet-print micromagnet integratedslides and (b) plain slide respectively. In (a), blue dots repre-sent the cells that are attached to the micromagnets, while theorange dots are the cells that are found not attached to anymicromagnet.

CHEN et al.

patterning on polymer substrates. Polymer 51:2147–2154,2010.

13Mitchell, M. J., E. Wayne, K. Rana, C. B. Schaffer, and M.R. King. TRAIL-coated leukocytes that kill cancer cells inthe circulation. Proc. Natl. Acad. Sci. 111:930–935, 2014.

14Nagrath, S., L. V. Sequist, S. Maheswaran, D. W. Bell, D.Irimia, L. Ulkus, M. R. Smith, E. L. Kwak, S. Digumarthy,and A. Muzikansky. Isolation of rare circulating tumourcells in cancer patients by microchip technology. Nature450:1235–1239, 2007.

15Nawarathna, D., N. Norouzi, J. McLane, H. Sharma, N.Sharac, T. Grant, A. Chen, S. Strayer, R. Ragan, and M.Khine. Shrink-induced sorting using integrated nanoscalemagnetic traps. Appl. Phys. Lett. 102:063504, 2013.

16Ozkumur, E., A. M. Shah, J. C. Ciciliano, B. L. Em-mink, D. T. Miyamoto, E. Brachtel, M. Yu, P.-I. Chen,B. Morgan, and J. Trautwein. Inertial focusing fortumor antigen–dependent and–independent sorting ofrare circulating tumor cells. Sci. Trans. Med. 5:179ra47,2013.

17Riethdorf, S., H. Fritsche, V. Muller, T. Rau, C. Schindl-beck, B. Rack, W. Janni, C. Coith, K. Beck, and F. Ja-nicke. Detection of circulating tumor cells in peripheralblood of patients with metastatic breast cancer: a validationstudy of the Cell Search system. Clin. Cancer Res. 13:920–928, 2007.

18Seekamp, J., S. Zankovych, A. Helfer, P. Maury, C. S.Torres, G. Boettger, C. Liguda, M. Eich, B. Heidari, and L.Montelius. Nanoimprinted passive optical devices. Nan-otechnology 13:581, 2002.

19Shields, I. V., C. Wyatt, C. D. Reyes, and G. P. Lopez.Microfluidic cell sorting: a review of the advances in theseparation of cells from debulking to rare cell isolation. LabChip 15(5):1230–1249, 2015.

20Singh, M., H. M. Haverinen, P. Dhagat, and G. E. Jab-bour. Inkjet printing—process and its applications. Adv.Mater. 22:673–685, 2010.

21Singh, M., H. M. Haverinen, P. Dhagat, and G. E. Jab-bour. Inkjet printing-process and its applications. Adv.Mater. 22:673, 2010.

22Stott, S. L., C.-H. Hsu, D. I. Tsukrov, M. Yu, D. T.Miyamoto, B. A. Waltman, S. M. Rothenberg, A. M.Shah, M. E. Smas, and G. K. Korir. Isolation of circu-lating tumor cells using a microvortex-generating her-ringbone-chip. Proc. Natl. Acad. Sci. 107:18392–18397,2010.

23Tekin, E., P. J. Smith, S. Hoeppener, A. M. van den Berg,A. S. Susha, A. L. Rogach, J. Feldmann, and U. S. Schu-bert. Inkjet printing of luminescent CdTe nanocrystal–polymer composites. Adv. Funct. Mater. 17:23–28, 2007.

24Tekin, E., P. J. Smith, and U. S. Schubert. Inkjet printingas a deposition and patterning tool for polymers andinorganic particles. Soft Matter 4:703–713, 2008.

25Truskett, V. N., and M. P. Watts. Trends in imprintlithography for biological applications. Trends Biotechnol.24:312–317, 2006.

26Voit, W., W. Zapka, L. Belova, and K. Rao. Application ofinkjet technology for the deposition of magnetic nanopar-ticles to form micron-scale structures. IEE Proc. Sci. Meas.Technol. 150:252–256, 2003.

27Williams, S. C. P. Circulating tumor cells. Proc. Natl. Acad.Sci. 110:4861, 2013.

28Wu, W., G.-Y. Jung, D. Olynick, J. Straznicky, Z. Li, X.Li, D. Ohlberg, Y. Chen, S.-Y. Wang, and J. Liddle. One-kilobit cross-bar molecular memory circuits at 30-nm half-pitch fabricated by nanoimprint lithography. Appl. Phys. A80:1173–1178, 2005.

29Yung, C. W., J. Fiering, A. J. Mueller, and D. E. Ingber.Micromagnetic-microfluidic blood cleansing device. LabChip 9:1171–1177, 2009.

30Zanini, L. F., N. M. Dempsey, D. Givord, G. Reyne, andF. Dumas-Bouchiat. Autonomous micro-magnet basedsystems for highly efficient magnetic separation. Appl.Phys. Lett. 99:232504, 2011.

Inkjet-Print Micromagnet Array on Glass Slides