Microarray-to-Microarray Transfer of Reagents by Snapping of TwoChips for Cross-Reactivity-Free Multiplex ImmunoassaysHuiyan Li,†,‡ Sebastien Bergeron,†,‡ and David Juncker†,‡,§,*†Biomedical Engineering Department, ‡McGill University and Genome Quebec Innovation Centre, §Department of Neurology andNeurosurgery, McGill University, Montreal, QC, H3A 1A4, Canada

ABSTRACT: Whereas microarray and microfluidic technolo-gies have progressed on many fronts, servicing microchips withminute amounts of reagents still constitutes an importantchallenge for many applications. Recently, chip-to-chip reagenttransfer methods were introduced that simplify the delivery ofreagents but required manual, visual alignment, custom-builtmicrowells, and only showed the reaction of a single samplewith multiple chemicals. Here, we present the snap chip, whichuses common glass slides for transfer, back-side alignment forachieving precise alignment in spite of mirroring, and a snap-apparatus for facile transfer of arrays of chemicals at once bysnapping the two slides together. We recently established that cross-reactivity was a significant problem in multiplex assays boththeoretically and experimentally and found that it can be eliminated by avoiding mixing, but which necessitates delivering eachdetection antibody to a single spot with the cognate capture antibody.18 Using the snap chip, multiplexed sandwichimmunoassays without mixing were performed: a slide with multiple arrays of 10 different capture antibodies was incubated witha sample, and then all detection antibodies transferred at once by snapping, each to the single cognate spot. All binding curveswere established and limits of detection in the pg/mL range were obtained. Snap chips were stored up to 3 months prior tousage. The snap chip, by dissociating microarray production, which requires expensive equipment, from assay execution, whichcan be achieved using a hand-held alignment apparatus, will allow for multiplex reactions to be performed using a user-friendlykit. This new liquid handling format can be easily adapted to other applications that require transfer of minute amounts ofdifferent reagents in parallel.

Microarray technologies have been developed in the pastyears. Fabrication of microarrays depends on the

transfer of minute amounts of reagents. An important challengeis to transfer and pattern chemicals stored in macroscopiccontainers as microarrays on slides. The macro-to-microchallenge was first addressed using pin spotters to transferminute amount of liquids from microtiter plates to chips byrepeatedly printing the pins onto multiple chips.1 The uploadand transfer are controlled by capillary effects that need to beprecisely engineered.2,3 Inkjet spotters for biological applica-tions with front-loading of reagents through the nozzles havesubsequently been developed and used for microarraying.4−6

The number of nozzles is typically much lower than that of pinspotters, however the programmability and rapid dispensing ofdroplets compensates for the limited parallelism. A highlyparallel system named the top spot comprises a spotting headthat is filled using capillary forces and that dispenses reagentsby compressing the air above the nozzles.7 This system issimpler than inkjet spotters but lacks individual addressing ofthe nozzles and requires larger volumes of reagents to load thehead. However, all of these systems rely on robotics, are quitecomplex, expensive to acquire and operate, and there is noprovision for removing a substrate from the spotter and placingit back subsequently for repeatedly addressing the same spot.One option to circumvent the need for expensive spotters is

to prespot reagents into miniaturized storage plates that may

then be used at a different time and place and thus dissociateproduction of the array from the execution of a multiplexreaction. Hunter pioneered this concept using preloadedmicrowell arrays.8 Whereas no peer review papers wereapparently published to our knowledge, this technology wasdeveloped into a commercial product called the OpenArraythat is used for DNA analysis.9 This technology is based oncustom chips made of 300 μm thick steel sheets with 3072holes, each 300 μm in diameter. The sheets are hydrophobic,but the inside wall of the holes is hydrophilic. Samples aretypically applied to a block of wells using pipet tips.More recently, several groups proposed novel approaches to

transfer minute amounts of reagents en bloc. Ismagilov andcolleagues developed an elegant approach called the slipchip.Nanoliter droplets of reagents are first trapped in channels andrecesses on one chip that serve as reaction chambers. A sampleis loaded in a chip comprising microchannel that runs parallelto the recesses. Next, the channels and the recesses are overlaidby sliding the two microstructured chips.10 To date, slipchipshave been used to deliver a single sample to an array of reagentssuch as 48 crystallization wells or to different chambers for

Received: February 1, 2012Accepted: April 26, 2012

Article

pubs.acs.org/ac

© XXXX American Chemical Society A dx.doi.org/10.1021/ac3003177 | Anal. Chem. XXXX, XXX, XXX−XXX

sandwich immunoassays,10,11 which are examples of reactingone chemical with a number of N reagents, (a 1-to-N transfer).Chip-to-chip transfer methods have also been performed by

clamping two chips together. The first demonstration wascarried out using sol−gel droplets immobilized on a glass slideand loaded with drugs or metabolites. The chemicals were thentransferred to cell monolayers by diffusion ensuring that onlycells close to each droplet were exposed to significantconcentration of chemicals.12 Subsequently, similar approacheswere extended to alginate gel droplets and to cells encapsulatedin collagen.13,14 More recently, Khademhosseini and colleaguesadopted an approach to transfer drugs from ∼200 μm wideposts made of either PDMS15 or a hydrogel16 that were coatedor loaded respectively with a drug library by inkjet spotting. Anarray of 400 μm wide microwells containing cells (from thesame cell line) was perfused with a drug library by clamping thechips and letting the drugs diffuse into the buffer contained ineach well. The alignment between the two slides was performedmanually with the aid of a stereomicroscope, which iscumbersome while at the same time a small number of wellswere typically misaligned, all of which limits the versatility ofthis approach. In summary, for the chip transfer methodsdescribed above, manual alignment based on visible structureson the chip was used, and the transfer followed an N-to-1 or a1-to-N arrangement with a number of N different reagentsbeing reacted or mixed with one other reagent.Multiplexed sandwich immunoassays are a chemical reaction

where N analytes are sandwiched between (bound by) Ncapture and N detection antibodies.17 In conventional multi-plexed sandwich assays in either array or bead formats, thedetection antibodies are applied as a mixture over the wholearray, and the pairwise interaction between nonmatchedreagents each constitutes a liability for cross-reactivity, whichin practice can lead to significant levels of cross-reactivity for anonly 14-plex assay.18 Cross-reactivity can be minimized bycareful optimization and selection of reagents for some classesof analytes, but notwithstanding best efforts, cross-reactivitycannot be suppressed entirely in practice. Moreover, thevulnerability to cross-reactivity scales as 4N(N − 1) irrespectiveof optimization efforts18 and cross-reactivity events can occurduring a test owing to the uniqueness of each sample. Thesecharacteristics limit the sensitivity and reliability of multiplexedsandwich assay. Recently, we proposed the antibody colocaliza-tion microarray (ACM)18 that overcomes these limitations byavoiding mixing and using two spotting rounds, one to arraythe capture antibodies and a second one to deliver eachdetection antibody to a single capture antibody spot. The assayconfiguration of the ACM reproduces the one of the enzymelinked inmunosorbent assays (ELISA) − the gold standard forimmunoassays − where only a single pair of antibodies is usedin each microwell and the dual binding of capture and detectionantibodies discriminates against nonspecific binding and cross-reactivity. The execution of an ACM entails the following keysteps: Aligned spotting of capture antibodies, removing theslide from the spotter, whole-slide blocking and incubating withsamples, washing and rinsing as needed, placing it back on thedeck for aligned spotting of the detection antibody, and finallywhole-slide incubation of labels and secondary antibodies. TheACM thus depends on the transfer of N different detectionantibodies to N different capture antibody spots, representingan N-to-N transfer. This was achieved using a custom builtmicroarrayer with alignment mechanisms for precise overlay ofthe capture and detection antibody spots.18 The ACM protocol

thus entails spotting during the assay execution, which iscumbersome and requires the use of a complex and expensivemicroarrayer.Here, we present the snap chip for (i) N-to-N transfer of a

microarray of reagents stored as semispherical liquid dropletson a transfer slide to a target microarray on an assay slide bysnapping two microscope slides together. (ii) Visualization-freetransfer thanks to a protocol with back-side alignment and ahand-held snap apparatus used to mechanically align the twoslides without the need of a microscope enabling simple andreliable transfer of reagents. (iii) Dissociation of the assayexecution from the time-consuming and costly microarrayfabrication by establishing long-term storage of both assay andtransfer slides that can be spotted ahead of time and stored.Using the snap chip, we performed multiplexed sandwichimmunoassays with colocalization of capture and detectionantibodies with detection limits in the pg/mL. Ten targets wereassayed simultaneously in both buffer and serum.

■ EXPERIMENTAL SECTIONMaterials. Rabbit antigoat IgG (H+L) labeled with the

fluorescent dye Alexa Fluor 488 and goat antimouse IgG (H+L) labeled with Alexa Fluor 647 were purchased fromInvitrogen. Matched antibody pairs (capture and biotinylateddetection antibodies) and antigens used in this study includehuman epidermal grow factor receptor 2 (HER 2), Endoglin(ENG), Leptin (LEP), fibroblast growth factor (FGF),osteopontin (OPN), tumor necrosis factor receptor-II (TNFRII), granulocyte macrophage colony-stimulating factor (GM-CSF), chemokine (C−C motif) ligand 2 (CCL 2), chemokine(C−C motif) ligand 3 (CCL 3), and interleukin-1 beta (IL 1β)are from R&D Systems. Streptavidin-conjugated Cy 5 waspurchased from Rockland. Phosphate buffered saline (PBS)tablets were purchased from Fisher Scientific. Tween-20 andglycerol were obtained from Sigma-Aldrich. Bovine serumalbumin (BSA) was purchased from Jackson ImmunoResearchLaboratories, Inc. Normal human female serum from a singledonor was purchased from Golden West Biologicals. BSA-freeStabilGuard Choice Microarray Stabilizer was obtained fromSurModics, Inc. Nitrocellulose coated slides were obtainedfrom Grace Bio-Laboratories, Inc. Aminosilane coated slideswere purchased from Schott North America.

Preparation of Microarrays on Slides. Capture antibodysolutions containing 400 μg/mL antibodies and 10% glycerol inPBS were spotted on a nitrocellulose slide (assay slide) at arelative humidity of 60%, 1.2 nl were delivered to each spot.Detection antibody solutions containing 20 μg/mL antibodies,20% glycerol, and 1% BSA were spotted on an aminosilaneslides (transfer slide) at a relative humidity of 80% to preventevaporation; 8 nl were delivered to each spot. Spotting wasperformed using an inkjet spotter (Nanoplotter 2.0, GeSiM).The center-to-center spacing between spots was 800 μm for the1024-spot array (Figure 4), and 1 mm for the immunoassays.After spotting, the assay slide was incubated for 1 h with

spotted capture antibodies at room temperature with a relativehumidity of 60%. A slide module gasket with 16 compartments(Grace Bio-Laboratories, Inc.) was then clamped on the slidedividing it into 16 wells for immunoassays. After incubation theassay slide was rinsed twice with PBS containing 0.1% Tween-20 (PBST) for 5 min on the shaker at 450 rpm and once withPBS, also for 5 min at 450 rpm.

Alignment and Snapping of Slides. A back-sidealignment approach was used to achieve mirror symmetric

Analytical Chemistry Article

dx.doi.org/10.1021/ac3003177 | Anal. Chem. XXXX, XXX, XXX−XXXB

alignment between the assay and the transfer slides. Both slideswere placed on 500 μm thick rubber sheets (McMaster-Carr) ina recess on two vacuum chucks. The vacuum chucks featuredalignment rods and holes, and were used to hold the slides inplace during snapping. Two 25 μm thick Kapton sheets(McMaster-Carr) were inserted between the slides at bothextremities. To snap the slides, the two vacuum chucks werepushed together by hand, and clamped tightly using two C-Type clamps (Canadian Tire) applied at the extremities. After 1min, the clamps were removed, the vacuum chucks wereseparated, and the slides were removed from the snapapparatus. Additional explanations are given in the Resultsand Discussion section.10-plex Sandwich Immunoassays with Snap Chips.

Ten proteins including HER 2, ENG, LEP, FGF, OPN, TNFRII, GM-CSF, CCL 2, CCL 3, and IL 1β were measured inspiked buffer or spiked, diluted serum (10% in PBS buffer).The assay slide onto which capture antibodies for these tenproteins were spotted was blocked for 1 h on the shaker at 320rpm with StabilGuard. Five-folds, seven-points dilution series often proteins in PBS or 10% human serum diluted in PBS wereapplied to the slide and incubated for 1 h at 320 rpm (startingconcentrations: 200 ng/mL for HER 2, ENG, LEP, FGF, OPNand 50 ng/mL for TNF RII, GM-CSF, CCL 2, CCL 3, IL 1β).Blank samples of PBS and 10% serum in PBS containing nospiked proteins were also incubated on the assay slide at thesame time. The slide was rinsed twice with PBST and once withPBS on the shaker at 450 rpm for 5 min, the slide modulegasket was removed, and the slide was dried under compressednitrogen gas. Next, the assay and the transfer slides wereclamped on the snap apparatus, snapped together for 1 min,separated, and the assay slide was incubated in a Petri dishsaturated with humidity for 1 h. Then a slide module gasket wasclamped on the assay slide, and the slide was rinsed three timeswith PBST and once with PBS on the shaker at 450 rpm for 5min, and incubated with 2.5 μg/mL of streptavidin conjugatedCy 5 for 20 min on the shaker at 320 rpm. The slide was thenrinsed twice with PBST, once with PBS and once withdeionized water on the shaker at 450 rpm for 5 min, and driedwith a stream of nitrogen gas.Storage of Snap Chip Assay and Transfer Slides. We

spotted both assay and transfer slides, stored them for 3months, performed the immunoassays and compared themwith freshly spotted slides. The assay slide was blocked withStabilGuard after incubation with capture antibodies and bothassay and transfer slides were immediately stored in an airtightbag with desiccant and placed in a −20 °C freezer. Sealed bagswere removed from the freezer and kept for 30 min at roomtemperature prior to opening and removing the slides toprevent condensation on the surface. The transfer slides werethen incubated in a Petri dish saturated with humidity for 30min to rehydrate the glycerol droplets before the antibodytransfer process.Scanning and Analysis. A microarray laser scanner (LS

ReloadedTM, Tecan) was used to scan slides. For the large scalearrays, a 488 and 633 nm lasers were used to image captureantibody spots and the transferred proteins, respectively. Forsandwich immunoassays, only the 633 nm laser was used. Thefluorescence intensity was computed by subtracting thebackground signal in the vicinity of each spot. All experimentswere performed in triplicate, and the data was analyzed usingArray-Pro Analyzer (MediaCybernetics). Y-intercept of the fitcurves incremented by two or three times the standard

deviation of the independent assays was calculated as thelimit of detection (LOD) using GraphPad Prism (GraphPadSof tware).

■ RESULTS AND DISCUSSIONMicroarray Fabrication. The procedure for the micro-

array-to-microarray transfer of detection antibodies to slideswith immobilized capture antibody spots is shown in Figure 1.

Corresponding detection and capture antibodies are colocalizedthus following the format of an ACM. A slide with a 12 μmthick nitrocellulose coating was used as assay slide because theporous nitrocellulose provides a high protein binding capacity,and absorbs the solution from the transfer slide thus promotingrapid binding of detection antibodies. For the transfer slides, wefirst tested uncoated glass slides, but because of the low contactangle between glass and the antibody solutions, the resultingdroplets were too thin to permit a reliable transfer to the assayslide. Slides with an aminosilane coating (contact angle withwater ∼65°)19 were chosen because they yielded a compromisebetween having a slightly enlarged diameter for relaxedalignment between the two microarrays and a droplet thicknessof tens of micrometers to ensure fluidic contact to the assayslide and reliable transfer. 1.2 nl of solution was delivered toeach capture spot yielding a 300 μm spot on the nitrocelluloseslide, whereas 8 nl of detection antibody solution was appliedon the aminosilane-coated transfer slide and produced a dropletthat was 700 μm in diameter on the nitrocellulose slide upontransfer. Following snapping and separation, all droplets weretransferred entirely and reliably to the assay slide.

Figure 1. Process flow for microarray-to-microarray transfer ofreagents for multiplexed sandwich immunoassays. (a) Captureantibody spots are formed by spotting 1.2 nl onto a nitrocelluloseassay slide, and transfer droplets by spotting 8 nl of biotinylateddetection antibodies on a transfer aminosilane slide with precisealignment. (b) Incubation of the nitrocellulose assay slide with thesample solution. (c) Microarray-to-microarray transfer of detectionantibodies to assay slide with snap apparatus. (d) Incubation of theassay slide with streptavidin-Cy 5. (e) After drying, the fluorescencewas acquired with a microarray scanner.

Analytical Chemistry Article

dx.doi.org/10.1021/ac3003177 | Anal. Chem. XXXX, XXX, XXX−XXXC

Mirror Alignment of Slides. Our goal was to establish amicroarray-to-microarray transfer without visual adjustmentusing purely mechanical alignment. This was motivated by thefact that the spots on the nitrocellulose slide are invisible whendried and that it is in any case cumbersome to do visualalignment under a microscope. When using mechanicalalignment, one however needs to consider the mirror symmetrybetween spotting and transfer (part c of Figure 1). Duringspotting, the bottom right corner of each slide was pushedagainst a ridge on the slide deck of the microarrayer. However,during the transfer the two slides face one another and whenseen from the top, the bottom-right corner of the transfer slidebecomes the bottom-left corner. In conventional microarrays,there is no need for accurate positioning of the spots and all ofthe spot positions can be defined relatively to the position ofthe first spot. However, for the microarray-to-microarraytransfer protocol proposed here this would lead to misalign-ment between the two arrays because of the mirroring.Mechanical alignment following mirroring is further compli-cated by the fact that the size of typical glass slides can vary20

and that it would thus not be possible to align the two slides byaligning relative to a single corner corresponding to twoopposing edges on the transfer and assay slide, respectively (i.e.,when looking at the assay slide through the transfer slide, thebottom right corner in fact corresponds to the bottom-leftcorner of the mirrored transfer slide).We considered two options for achieving exact overlay

during the transfer. The first was to spot at exact coordinates ina mirror pattern on both slides and to align each slide relative tothe bottom-left edge on each half of the snap apparatus. Thesecond option was to first spot an alignment mark on the back-side of the transfer side (at the same coordinate than therightmost spot of the top row of the assay slide) while aligningboth slides relative to the bottom right corner, flip the transferslide, align it again relative to the bottom right corner (as seenfrom the top), use the image recognition system of the inkjetspotter to extract the coordinate of the alignment mark, and useit as the coordinate of the first spot of the array, as seen in partsa−c of Figure 2. In this manner, both arrays will be alignedwhen they are aligned in the transfer apparatus (Figure 3) andthe alignment accuracy is independent of the size of the slides.Given the availability of an image recognition system, and of

the difficulty in correcting for size variation in the first option,we used the second option for our experiments.

Snapping of Slides. The assay and the transfer slides wereplaced in a custom-built snap apparatus comprising twoprecision milled vacuum chucks, a steel plate, and four steelrods, as shown in parts a and b of Figure 3. The vacuum chuckscontain a recess for inserting and aligning the slides and holdthem in place prior to snapping them together. To keep theprecise mirror symmetric pattern alignment between the twoslides, the assay and the transfer slides were pushed against thebottom right and the bottom left corners in the recess of theirrespective vacuum chuck. The four steel rods were fixed to oneof the vacuum chucks and served to guide the other chuck thatcomprised four alignment holes. The steel plate was used

Figure 2. Schematics of the protocol for mirror alignment. (a) Capture antibodies were spotted on an assay slide. The framing square at the bottom-right indicates the corner used for mechanical alignment. (b) The transfer slide was flipped, and aligned to the bottom right, and a dye droplet wasspotted at the same coordinate than the rightmost spot of the top row on the assay slide. (c) The transfer slide was flipped back, and placed back onthe deck. Using an image recognition system, the coordinate of the dye spot was determined, and detection antibodies spotted relative to the dyespot. (d) Schematic representation of the alignment of the slides on the two vacuum chucks relative to the bottom right and left, respectively.

Figure 3. Schematics and images of the snap apparatus. (a) Schematicof the snap apparatus and liquid bridging between the two slidesduring transfer. A rubber backing and spacers ensure a constant gapacross the slide. (b) Photograph of the snap apparatus made of twovacuum chucks each containing a recess for mechanical slidealignment, steel rods, and a precision-machined support plate(width: 12.5 cm, length: 15 cm) used for supporting the slides duringsnapping.

Analytical Chemistry Article

dx.doi.org/10.1021/ac3003177 | Anal. Chem. XXXX, XXX, XXX−XXXD

during snapping to support the two chucks while they werebeing manually clamped together. Small pieces cut out of a 25μm thick Kapton sheet were placed between the two slides atthe edges to keep a constant separation over the entire area andthus avoid excessive squeezing of the droplets during snapping.500 μm thick rubber cushions were inserted between the slidesand the vacuum chucks to accommodate small imperfectionand distribute the pressure evenly across the two slides.Following snapping of the two slides, a liquid bridge formedatop each of the spots connected the slides across the gap, andupon separation, the droplets and reagents were transferred tothe assay slide, as seen in part a of Figure 3.Accuracy of Microarray-to-Microarray Transfer. We

characterized the alignment accuracy using an assay slide with16 nitrocellulose pads each spotted with 16 droplets. IgGslabeled with green and red fluorescent dyes were spotted onboth the assay and transfer slide, respectively, snapped together,and the nitrocellulose slide was scanned immediately withoutany washing. The average center-to-center distance between thespotted and transferred droplet was 147 μm, whereas thelargest distance was 216 μm. The misalignment increased fromthe left to the right side of the slide, and in fact doubledfollowing mirrored transfer, indicating that there was an angularmisalignment between the slides and the motorized inkjet stage.The different droplet sizes of capture and detection antibodyspots however relaxed the alignment constraints and ensuredcomplete overlap in spite of some misalignment.

Microarray-to-Microarray Transfer of Antibodies. Wefirst evaluated the use of the snap chip for carrying out simpleone-step immunoassays. An array of 256 fluorescently labeledgoat IgGs (the analyte) were transferred to an assay slidepatterned with an array of 1024 fluorescently labeled rabbitantigoat IgGs (the capture antibody), incubated and washed, asseen in Figure 4. 20% glycerol was added to the detection bufferto prevent drying of the antibodies being transferred while theassay slide was dried under a stream of nitrogen prior to thetransfer to promote the absorption of the droplets in thenitrocellulose and minimize lateral spreading. Visual inspectionreveals a selective and homogeneous transfer of proteins acrossthe entire slide, as seen in part a of Figure 4. The fluorescenceintensity profiles of the two proteins shows excellent overlap, asseen in part b of Figure 4.

10-plex Sandwich Immunoassays in Buffer andSerum. To evaluate the use of microarray-to-microarraytransfer for multiplexed sandwich immunoassays, we selected10 proteins, including one breast cancer biomarker (HER 2), 4cancer related proteins (ENG, LEP, FGF, OPN), and 5cytokines (TNF RII, GM-CSF, CCL 2, CCL 3, IL 1β). HER2 isa plasma membrane-bound receptor tyrosine kinase, and it hasbeen used for the typing of breast cancer by measuring itsamount in biopsy samples by immunohistochemistry.21 WhenHER2 is overexpressed breast cancer patients can be treatedwith trastuzumab, a monoclonal antibody against the HER2receptor that could inhibit cell proliferation.22 ENG is a cellmembrane glycoprotein that is overexpressed in tumor bloodvessels but not in most normal tissue and therefore hasprognostic significance.23 Leptin is a protein hormone andoverexpressed in obese individuals and also in breast cancercells.24 FGF is important in tumor angiogenesis and has beeninvestigated for cancer therapeutics.25 OPN is correlated withmalignant transformation and its elevated level was found to beassociated with shorter survival in breast cancer patients.26 TNFRII is one of the TNF α receptors and has been reported to beincreased in the blood of breast cancer patients.27 GM-CSF is acytokine that may activate immune cells and has beeninvestigated in breast cancer treatment.28 CCL2 and CCL3are chemokine ligands that have been found to be increased intumor.29,30 IL 1β is a cytokine that has been reported tointeract with estrogen receptors in breast cancer cells and maymodulate hormonal activity in human breast tumors.31 Theexperiment flow is shown in parts b−e of Figure 1. Thespotting solution containing detection antibody was supple-mented with 1% BSA to block the aminosilane-coated slidesurface and prevent the adsorption of the detection antibodies,which helped increase the transfer efficiency.We established binding curves for all proteins spiked in PBS,

as seen in parts b and c of Figure 5. A four-parameter logisticequation was used for curve fitting and the LOD obtained werein the pg/mL range for all of the 10 proteins, as seen in Table 1.The LODs were calculated as background signal incrementedby 2 SD for the comparison with ELISAs from R&D Systems,32

and 3 SD corresponding to the scientific convention.33 Mostproteins achieved comparable LODs with commercial ELISAs,for CCL 3 the LOD obtained using snap chip exceeds that ofELISAs, and for FGF and OPN further optimization is needed.The LOD values were compared with the physiological range ofthese proteins in the serum of healthy persons as found in thescientific literature, and our assays were found to exceed thislimit for all of the 10 targets.34−42 Moreover, by selectingantibodies from other suppliers, and by thorough optimization

Figure 4. Transfer of a microarray of proteins to a microarray ofantibodies from one slide to another by snapping. An array of antigoatIgGs labeled with Alexa 488 (green) with a center-to-center spacing of800 μm was addressed with an array of Alexa 647 (red) labeled goatIgGs with a center-to-center spacing of 1600 μm. Intermediate spotswere loaded with a solution of PBS. (a) Fluorescence image of theassay slide after snapping and transfer. The transfer was successful overthe entire slide and the insert reveals a good overlap between the twofluorescence spots indicated by yellow color. Green borderscorrespond to the edge of each of the 16 nitrocellulose pads. Thered speckles on the top two nitrocellulose pads are contaminations butdo not affect the array results. (b) Fluorescence intensity profile of thegreen and red protein spots in the row marked by the arrow in theinsert.

Analytical Chemistry Article

dx.doi.org/10.1021/ac3003177 | Anal. Chem. XXXX, XXX, XXX−XXXE

using design of experiment approaches such as the Taguchimethod, significant improvement of the LOD of an order ofmagnitude or more can be achieved.43

To test the applicability of the snap chip for immunoassayswith complex biological samples, we performed a multiplexed

immunoassay for the same 10 proteins spiked in 10% serum, asseen in parts d and e of Figure 5. Serum was diluted tominimize matrix interference and background signals, as it iscommonly done in ELISA assays.33,44 For most proteins, thebackground signals at zero concentration of spiked proteins are

Figure 5. Fluorescent image and binding curves for sandwich immunoassays for 10 proteins in buffer solutions and 10% serum. (a) Fluorescentmicrograph of a representative slide with 16 replicate arrays incubated with PBS and 10% serum samples, and a close-up of a single array identified bythe dashed lines. Scale bar of close-up: 1 mm. (b) Binding curves for HER 2, ENG, LEP, FGF, and OPN in PBS. (c) Binding curves for TNF RII,GM-CSF, CCL 2, CCL 3, and IL 1β in PBS; the affinity of the antibodies for these five proteins was higher and the assay range was adjustedaccordingly. (d) Binding curves for HER 2, ENG, LEP, FGF, and OPN and (e) for TNF RII, GM-CSF, CCL 2, CCL 3, and IL 1β in 10% serum.The error bars are standard deviations between independent assays. The LOD of each curve calculated as background signal incremented by 2 SDare indicated using arrows in (b) and (c). The LOD of each curve in spiked 10% serum is not shown because the presence of endogenous proteins inserum that may lead to inaccurate results.

Table 1. LOD Values Obtained from 10-plex Immunoassays in PBS; the Units Are pg/mL

protein LOD (3 SD) LOD (2 SD)LOD from

R&D Systems32 (2 SD)average concentration inserum of healthy controlsa refs

HER 2 155 81 n/a ≤15 000 Kong et al.38

ENG 138 74 30 150 000 Takahashi et al.41

LEP 52 28 8 26 430 ± 19 400 Aliustaoglu et al.34

FGF 85 51 3OPN 263 171 24 123 000 Bramwell et al.35

TNF RII 36 21 2 3180 ± 600 Rutkowski et al.39

GM-CSF 6 3 3 900 ± 90 Scholl et al.40

CCL 2 15 10 5 173 Kim et al.37

CCL 3 3 2 10 88.3 Kim et al.37

IL 1β 14 8 1 40 Yurkovetsky et al.42

aStandard deviation indicated when available.

Analytical Chemistry Article

dx.doi.org/10.1021/ac3003177 | Anal. Chem. XXXX, XXX, XXX−XXXF

higher than those in PBS, presumably due to the presence ofendogenous proteins given that the typical physiologicalconcentrations are well beyond the LOD, as well as becauseof matrix effects.45 Whereas further investigations are neededbeyond the scope of this manuscript, the binding curvesindicate that it will be possible to quantify proteins in bloodsamples using the snap chip. These results indicate that highsensitivity can be achieved using snap chips that already rivalsthe one obtained with ELISA for some assays.Storage of Snap Chips. To dissociate the production of

the slides − which requires advanced equipment such as a pinor an inkjet spotter − from the execution of the assay, whichcan be done at low cost without need for peripheral equipment,slides must be stored prior to usage. Here, using TNF RII as amodel protein, we developed a protocol for storing transfer andassay slides in a freezer at −20 °C. After spotting, slides wereblocked, dried, and stored at −20 °C and thawed prior to usage.We noticed that when the prespotted slides were thawed in theopen, water condensation would form on the surface, and as aresult the droplets were disrupted. To prevent condensation,we sealed the slide together with a desiccant in a plastic bag andstored it in a freezer. Prior to an experiment, the bag was takenout of the freezer and equilibrated to room temperature beforebeing opened. Using this protocol, no loss of signal wasobserved between freshly prepared slides and 1 month storage,and a factor 4 of increased LOD for 3 months storage, as seenin Figure 6. After 1 month storage, there was a higher

background signal (Y-intercept of the fit curve), but also astronger signal, and for 3 months storage a higher LOD may bedue to loss of activity of the antibodies. The LODs for all threeconditions was still well below the average physiologicalconcentration in healthy patients for this protein. These resultsconfirm the possibility for storing the snap chips and set thestage for future optimizations to establish longer term storagewhile also developing protocols for storage at 4 °C, or maybeeven at room temperature.

■ CONCLUSIONSIn this work we developed a snap chip for the collective transferof reagents from microarray-to-microarray in an N-to-Nconfiguration with a density of 130 spots/cm2. The alignmentmethod enables end users to perform visualization-free transferby simply snapping two slides using a hand-held snap apparatus

without the need of a microscope. The snap chip was used formultiplexed immunoassays, and LODs in the pg/mL rangewere obtained. The LODs obtained with the snap chip weresimilar to the ones obtained with commercial ELISA. Byprespotting on the slides and storage, snap chip overcomes theneed for spotting during the experiments and the need forspotting equipment for the end users.The snap chip transfer may be expanded to collectively

transferring chemicals to a micro-structured microfluidic chipand help address the so-called “world-to-chip” interface.46−48

Next steps in the development of the snap chip will be toimprove the alignment accuracy between transfer and assaychip to increase spot density, and optimize the conditions forthe intended application, such as immunoassays while tacklingimportant practical issues such as slide storage. Finally, it will bepossible to build portable and convenient-to-use snap chip forthe use by unskilled end users in research and eventually inclinical settings as point-of-care diagnostics.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: david.juncker@mcgill.ca.

NotesThe authors declare the following competing financialinterest(s): McGill has filed a patent application on someaspects of the work reported here with H.L. and D.J. asinventors.

■ ACKNOWLEDGMENTSWe thank Saule Tourekhanova for her help, Veronique Lafortefor critical reading of the manuscript, and Rob Sladek and HaigDjambazian for use of the inkjet spotter. We thank theCanadian Institutes for Health Research (CIHR), the NaturalScience and Engineering Research Council of Canada(NSERC), and the Canada Foundation for Innovation (CFI)for financial support. H.L. acknowledges a scholarship from theNSERC-CREATE Integrated Sensor Systems program. D.J.acknowledges support from a Canada Research Chair.

■ REFERENCES(1) Schena, M.; Shalon, D.; Davis, R. W.; Brown, P. O. Science. 1995,270, 467−470.(2) George, R. A.; Woolley, J. P.; Spellman, P. T. Genome Res. 2001,11, 1780−3.(3) Safavieh, R.; Roca, M. P.; Qasaimeh, M. A.; Mirzaei, M.; Juncker,D. J. Micromech. Microeng. 2010, 20, 055001.(4) Li, H.; Leulmi, R. F.; Juncker, D. Lab Chip. 2011, 11, 528−534.(5) Rianasari, I.; Walder, L.; Burchardt, M.; Zawisza, I.; Wittstock, G.Langmuir 2008, 24, 9110−9117.(6) Pla-Roca, M.; Leulmi, R. F.; Djambazian, H.; Sundararajan, S.;Juncker, D. Anal. Chem. 2010, 82, 3848−3855.(7) Steinert, C.; Kalkandjiev, K.; Zengerle, R.; Koltay, P. Biomed.Microdevices 2009, 11, 755−761.(8) Hunter, I. W. Method for performing microassays. U.S. Patent6,743,633, June 1, 2004.(9) Morrison, T.; Hurley, J.; Garcia, J.; Yoder, K.; Katz, A.; Roberts,D.; Cho, J.; Kanigan, T.; Ilyin, S. E.; Horowitz, D.; Dixon, J. M.;Brenan, C. J. H. Nucleic Acids Res. 2006, 34, e123.(10) Du, W.; Li, L.; Nichols, K. P.; Ismagilov, R. F. Lab Chip. 2009, 9,2286−2292.(11) Liu, W.; Chen, D.; Du, W.; Nichols, K. P.; Ismagilov, R. F. Anal.Chem. 2010, 82, 3276−3282.(12) Lee, M.-Y.; Park, C. B.; Dordick, J. S.; Clark, D. S. Proc. Natl.Acad. Sci. U.S.A. 2005, 102, 983−987.

Figure 6. Binding curves of TNF RII assays in PBS obtained withfreshly spotted slides and with slides stored at −20 °C for 1 and 3months, respectively. The error bars are standard deviations betweenduplicate spots. The LOD values obtained for slides that were fresh,and stored for 1 month and 3 months were 4, 3, and 18 pg/mL,respectively. The LOD of each curve was calculated as backgroundintensity incremented by 2 SD and is indicated using an arrow.

Analytical Chemistry Article

dx.doi.org/10.1021/ac3003177 | Anal. Chem. XXXX, XXX, XXX−XXXG

(13) Fernandes, T. G.; Kwon, S.-J.; Bale, S. S.; Lee, M.-Y.; Diogo, M.M.; Clark, D. S.; Cabral, J. M. S.; Dordick, J. S. Biotechnol. Bioeng.2010, 106, 106−118.(14) Lee, M.-Y.; Kumar, R. A.; Sukumaran, S. M.; Hogg, M. G.;Clark, D. S.; Dordick, J. S. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 59−63.(15) Wu, J.; Wheeldon, I.; Guo, Y.; Lu, T.; Du, Y.; Wang, B.; He, J.;Hu, Y.; Khademhosseini, A. Biomaterials 2011, 32, 841−848.(16) Kwon, C. H.; Wheeldon, I.; Kachouie, N. N.; Lee, S. H.; Bae,H.; Sant, S.; Fukuda, J.; Kang, J. W.; Khademhosseini, A. Anal. Chem.2011, 83, 4118−4125.(17) Nielsen, U. B.; Geierstanger, B. H. J. Immunol. Methods. 2004,290, 107−120.(18) Pla-Roca, M.; Leulmi, R. F.; Tourekhanova, S.; Bergeron, S.;Laforte, V.; Moreau, E.; Gosline, S. J. C.; Bertos, N.; Hallett, M.; Park,M.; Juncker, D. Mol. Cell. Proteomics 2012, 11.(19) Briard, R.; Heitz, C.; Barthel, E. J. Non-Cryst. Solids. 2005, 351,323−330.(20) Product Specification, SCHOTT. http://www.schott.com/nexterion/english/download/spec_glass_b_cl_cl_eu.pdf (accessedApril 22, 2009).(21) Gaedcke, J.; Traub, F.; Milde, S.; Wilkens, L.; Stan, A.; Ostertag,H.; Christgen, M.; von Wasielewski, R.; Kreipe, H. H. Mod. Pathol.2007, 20, 864−870.(22) Harvey, J. M.; Clark, G. M.; Osborne, C. K.; Allred, D. C. J. ClinOncol. 1999, 17, 1474−1481.(23) Beketic-Oreskovic, L.; Ozretic, P.; Rabbani, Z.; Jackson, I.;Sarcevic, B.; Levanat, S.; Maric, P.; Babic, I.; Vujaskovic, Z. PathologyOncol. Res. 2011, 17, 593−603.(24) Zhou, W.; Guo, S.; Gonzalez-Perez, R. R. Br. J. Cancer. 2011,104, 128−137.(25) Knights, V.; Cook, S. J. Pharmacology & amp; Therapeutics 2010,125, 105−117.(26) Bramwell, V.; Doig, G.; Tuck, A.; Vandenberg, T.; Tomiak, A.;Perera, F.; Tonkin, K.; O’Malley, F.; Wilson, S.; Chambers, A. BreastCancer Res. Treat. 2001, 69, 257.(27) Fuksiewicz, M.; Kowalska, M.; Kotowicz, B.; Rubach, M.;Chechlinska, M.; Pienkowski, T.; Kaminska, J. Clin. Chem. Lab. Med.2010, 48, 1481−1486.(28) Cheng, Y. C.; Valero, V.; Davis, M. L.; Green, M. C.; Gonzalez-Angulo, A. M.; Theriault, R. L.; Murray, J. L.; Hortobagyi, G. N.;Ueno, N. T. Br. J. Cancer. 2010, 103, 1331−1334.(29) Ghoneim, H. M.; Maher, S.; Abdel-Aty, A.; Saad, A.; Kazem, A.;Demian, S. R. The Egyptian Journal of Immunology 2009, 16, 37−48.(30) Zucchetto, A.; Benedetti, D.; Tripodo, C.; Bomben, R.; Dal, Bo,M.; Marconi, D.; Bossi, F.; Lorenzon, D.; Degan, M.; Rossi, F. M.;Rossi, D.; Bulian, P.; Franco, V.; Del Poeta, G.; Deaglio, S.; Gaidano,G.; Tedesco, F.; Malavasi, F.; Gattei, V. Cancer Res. 2009, 69, 4001−4009.(31) Speirs, V.; Kerin, M. J.; Newton, C. J.; Walton, D. S.; Green, A.R.; Desai, S. B.; Atkin, S. L. Int. J. Oncol. 1999, 15, 1251−1254.(32) ELISA Reference Guide & Catalog, R&D Systems. http://www.rndsystems.com/resources/images/6836.pdf(33) Gonzalez, R. M.; Seurynck-Servoss, S. L.; Crowley, S. A.; Brown,M.; Omenn, G. S.; Hayes, D. F.; Zangar, R. C. J. Proteome Res. 2008, 7,2406−2414.(34) Aliustaoglu, M.; Bilici, A.; Gumus, M.; Colak, A.; Baloglu, G.;Irmak, R.; Seker, M.; Ustaalioglu, B.; Salman, T.; Sonmez, B.; Salepci,T.; Yaylaci, M. Med. Oncol. 2010, 27, 388−391.(35) Bramwell, V. H. C.; Doig, G. S.; Tuck, A. B.; Wilson, S. M.;Tonkin, K. S.; Tomiak, A.; Perera, F.; Vandenberg, T. A.; Chambers,A. F. Clin. Cancer. Res. 2006, 12, 3337−3343.(36) Dehqanzada, Z. A.; Storrer, C. E.; Hueman, M. T.; Foley, R. J.;Harris, K. A.; Jama, Y. H.; Shriver, C. D.; Ponniah, S.; Peoples, G. E.Oncol. Rep. 2007, 17, 687−94.(37) Kim, B.; Lee, J.; Park, P.; Shin, Y.; Lee, W.; Lee, K.; Ye, S.;Hyun, H.; Kang, K.; Yeo, D.; Kim, Y.; Ohn, S.; Noh, D.; Kim, C. BreastCancer Res. 2009, 11, R22.

(38) Kong, S.-Y.; Kang, J. H.; Kwon, Y.; Kang, H.-S.; Chung, K.-W.;Kang, S. H.; Lee, D. H.; Ro, J.; Lee, E. S. J. Clin. Pathol. 2006, 59, 373−736.(39) Rutkowski, P.; Kaminska, J.; Kowalska, M.; Ruka, W.; Steffen, J.Int. J. Cancer. 2002, 100, 463−471.(40) Scholl, S. M.; Lidereau, R.; de la Rochefordiere, A.; Cohen-SolalLe-Nir, C.; Mosseri, V.; Nogues, C.; Pouillart, P.; Stanley, E. R. BreastCancer Res. Treat. 1996, 39, 275−283.(41) Takahashi, N.; Kawanishi-Tabata, R.; Haba, A.; Tabata, M.;Haruta, Y.; Tsai, H.; Seon, B. K. Clin. Cancer. Res. 2001, 7, 524−532.(42) Yurkovetsky, Z. R.; Kirkwood, J. M.; Edington, H. D.;Marrangoni, A. M.; Velikokhatnaya, L.; Winans, M. T.; Gorelik, E.;Lokshin, A. E. Clin. Cancer. Res. 2007, 13, 2422−2428.(43) Luo, W.; Pla-Roca, M.; Juncker, D. Anal. Chem. 2011, 83, 5767−5774.(44) Lexmond, W.; der Mee, J. v.; Ruiter, F.; Platzer, B.; Stary, G.;Yen, E. H.; Dehlink, E.; Nurko, S.; Fiebiger, E. J. Immunol. Methods.2011, 373, 192−199.(45) Pfleger, C.; Schloot, N.; Veld, F. t. J. Immunol. Methods. 2008,329, 214−218.(46) Delamarche, E.; Juncker, D.; Schmid, H. Adv. Mater. 2005, 17,2911−2933.(47) Yang, H.; Luk, V. N.; Abelgawad, M.; Barbulovic-Nad, I.;Wheeler, A. R. Anal. Chem. 2008, 81, 1061−1067.(48) Cooksey, G. A.; Plant, A. L.; Atencia, J. Lab Chip. 2009, 9,1298−1300.

Analytical Chemistry Article

dx.doi.org/10.1021/ac3003177 | Anal. Chem. XXXX, XXX, XXX−XXXH