lable at ScienceDirect

Biomaterials 32 (2011) 5478e5488

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomateria ls

The use of glass substrates with bi-functional silanes for designingmicropatterned cell-secreted cytokine immunoassays

Jeong Hyun Seo a, Li-Jung Chen b, Stanislav V. Verkhoturov b, Emile A. Schweikert b, Alexander Revzin a,*

aDepartment of Biomedical Engineering, University of California, Davis, CA 95616, USAbDepartment of Chemistry, Texas A&M University, College Station, TX 77843, USA

a r t i c l e i n f o

Article history:Received 18 February 2011Accepted 6 April 2011Available online 7 May 2011

Keywords:Cell functionCytokine release and immune cellsCytokine detectionImmunoassayCell micropatterningHydrogel microwells

* Corresponding author. Tel.: þ1 530 752 2383.E-mail address: arevzin@ucdavis.edu (A. Revzin).

0142-9612/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.biomaterials.2011.04.026

a b s t r a c t

It is often desirable to sequester cells in specific locations on the surface and to integrate sensingelements next to the cells. In the present study, surfaces were fabricated so as to position cytokinesensing domains inside non-fouling poly(ethylene glycol) (PEG) hydrogel microwells. Our aim was toincrease sensitivity of micropatterned cytokine immunoassays through covalent attachment of bio-recognition molecules. To achieve this, glass substrates were functionalized with a binary mixture ofacrylate- and thiol-terminated methoxysilanes. During subsequent hydrogel photopatterning steps,acrylate moieties served to anchor hydrogel microwells to glass substrates. Importantly, glass attachmentsites within the microwells contained thiol groups that could be activated with a hetero-bifunctionalcross-linker for covalent immobilization of proteins. After incubation with fluorescently-labeled avidin,microwells fabricated on a mixed acryl/thiol silane layer emitted w 6 times more fluorescence comparedto microwells fabricated on an acryl silane alone. This result highlighted the advantages of covalentattachment of avidin inside the microwells. To create cytokine immunoassays, micropatterned surfaceswere incubated with biotinylated IFN-g or TNF-a antibodies (Abs). Micropatterned immunoassaysprepared in this manner were sensitive down to 1 ng/ml or 60 pM IFN-g. To further prove utility of thisbiointerface design, macrophages were seeded into 30 mm diameter microwells fabricated on either bi-functional (acryl/thiol) or mono-functional silane layers. Both types of microwells were coated withavidin and biotin-anti-TNF-a prior to cell seeding. Short mitogenic activation followed by immunos-taining for TNF-a revealed that microwells created on bi-functional silane layer had 3 times higher signaldue to macrophage-secreted TNF-a compared to microwells fabricated on mono-functional silane. Therational design of cytokine-sensing surfaces described here, will be leveraged in the future for rapiddetection of multiple cytokines secreted by individual immune cells.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Cytokines are proteins secreted by mammalian cells in theprocess of endocrine communication. These proteins may bereleased in response to injury, causing inflammation or cell death,[1,2], or on the contrary, protecting against tissue injury [3].Cytokine production in leukocytes is an important means ofmonitoring immune competency or disease progression in patients.For example, detection of inflammatory cytokines such as IFN-g isused for diagnosing latent form of tuberculosis [4]. In an attempt toget more detailed and nuanced understanding of the roles playedby leukocyte subsets in the immune response, immunologists are

All rights reserved.

becoming increasingly interested in connecting cytokine secretionprofiles to specific leukocyte subsets and to specific single cells[5e9]. This goal is complicated by the fact that certain cytokineslike TNF-a and IFN-g may be secreted by multiple leukocytesubsets. Therefore, standard immunoassays of blood serum areinsufficient to discern which leukocyte subsets secreted cytokinesin question.

The main immunology tools used for cytokine profiling are flowcytometry and enzyme-linked immunospot (ELISpot) assays[10e12]. These technologies are robust and have been adapted bythe immunology community; however, there is a need to developnew tools that are less expensive and more suited for detectingcytokine release from live cells. Microfabrication and micro-patterning approaches are particularly well-suited for tackling thechallenge of connecting specific cells with specific secreted signals[13,14]. Several groups have been developingmicrodevices [15e18],

J.H. Seo et al. / Biomaterials 32 (2011) 5478e5488 5479

micropatterned surfaces [19e21] and microarrays [22e25] forleukocyte analysis. Previously, our group described the use ofmicroarrays for capturing groups of T-cells on anti-CD4 or anti-CD8T-cells spots (150 cells per spot) and detecting secreted cytokineson adjacent anti-cytokine antibody spots [25]. More recently, wedemonstrated patterning hydrogel microwells on top of a mixedlayer of anti-cell/anti-cytokine Abs to enable capturing individualCD4 T-cells from human blood and detecting IFN-g secreted bysingle cells [21]. However, in these past studies, Ab molecules wereimmobilized by physical adsorption on the glass attachment sitesinside the microwells. We therefore hypothesized that orienting Abmolecules inside the microwell will improve sensitivity of micro-patterned cytokine-sensing surfaces.

Surfaces functionalized with NH2, SH, COOH, NHS ester orepoxide end groups are commonly used for covalent immobiliza-tion of proteins. Avidin-biotin and protein A-Ab interactionsprovide additional routes for oriented attachment of Ab molecules[26]. Because we are interested in using micropatterned surfacesfor capturing single cells and detecting cytokines or other proteinsreleased by the cells, our desired surface needed to have a periodicpattern of non-fouling regions and cell/cytokine adhesive domains.Micropatterning strategies described in the literature includemicrocontact printing, photoresist lithography and direct write/spot approaches [27e30]. Another strategy, used extensively in ourprevious studies, is poly(ethylene glycol) (PEG) photolithographywhereby PEG prepolymer is photopatterned to create hydrogelmicrowells that are used to confine attachment of cells or proteinsto certain regions of the glass substrate surface [31,32]. This isa simple micropatterning strategy and can be used to populatelarge surface area with microwells for cell analysis [20,32].

A

1. Modify surface in a mixture of acrylate-and thiol-terminated silanes.

2. Spin coat PEG-DA and expose to UVlight through photomask.

3. Immerse in water to develop microwellarrays.

4. Close-up view of a mixed silanelayerinside a hydrogel microwell.

1

w

2

s

glass slide

Fig. 1. (A) A process flow diagram for micorpatterning hydrogel microwells on glass. In addalso contains thiol groups for covalent linking of proteins. (B) Strategy for immobilizing biomcross-linker, and then incubated with avidin and biotin-antibody. Throughout this paper thnoassays created by physical adsorption of avidin on mono-functional acrylated silanes foll

Attachment of hydrogel microstructures to glass or other oxidecontaining surfaces is typically promoted using an acrylated silanecoupling agent [20,32]

Moving beyond simple mono-functional silane layers, wewanted to design a coupling layer that would not only anchorhydrogel microstructures but could also be used for orientedattachment of cytokine-specific Abmolecules. Several strategies forcreating multi-functional silane layers have been proposed. Forexample, Stenger and Dulcey developed an approach of laserdesorption/patterning of silane molecules and backfilling withanother silane type to create periodic bi-functional silane surfaces[33]. We favored a simpler approach of defining composition offunctional groups on the surface by random co-assembly of twodifferent silanes [34e36]. This method was recently employed byLee et al. to modify glass substrates with a silane layer comprised ofally- and amine-terminated silanes for covalent attachment of bothhydrogel structures and avidin molecules [37]. In this paper, co-assembly of acrylate- and thiol-terminated silanes was used tocreate a bi-functional layer suitable for anchoring hydrogel micro-structures and for oriented attachment of antibodies. Thesemicropatterned surfaces were employed in construction ofimmunoassays for detection of exogenous and endogenous (cell-secreted) cytokines.

2. Materials and methods

2.1. Materials

Glass slides (75 � 25 mm2) were obtained from VWR (West Chester, PA). (3-acryloxypropyl) trimethoxysilane (MW 234) and 3-mercaptopropyl trimethox-ysilane (MW 196) were purchased from Gelest, Inc. (Morisville, PA).

B

. Functionalization of the mixed silaneith maleimide-PEG-NHS and neutravidin.

. Attachment of biotinylated Abspecific for IFN-γ or TNF-α.

glass slide

ition to acrylate moieties used to couple hydrogel microstructures to glass, silane layerolecules inside microwells. Mixed silane layer was activated using a hetero-bifunctionale sensitivity of immunoassays constructed on mixed silanes was compared to immu-owed by immobilization of biotin-antibody.

J.H. Seo et al. / Biomaterials 32 (2011) 5478e54885480

Poly(ethylene glycol)diacrylate (PEG-DA, MW575), 2-hydroxy-2-methyl-propiophenone (photoinitiator), bovine serum albumin (BSA), dimethylsulfoxide(DMSO), and anhydrous toluene (99.9%) were purchased from SigmaeAldrich(Saint Louis, MO). MAL-PEG2eNHS ester was purchased from Quanta Biodesign,Ltd. (Powell, OH). Alexa Fluor-546 conjugated streptavidin, Alexa Fluor-488conjugated neutravidin, and regular neutravidin were purchased from Invi-trogen (Carlsbad, CA). Recombinant human IFN-g and biotinylated goat anti-human IFN-g Ab were purchased from R&D systems (Minneapolis, MN).Poly(dimethylsiloxane) (PDMS) along with a curing agent were purchased fromDow Corning (Midland, MI).

Fig. 2. ToF-SIMS analysis of silane modification and avidin immobilization on glass substrateeach projectile creates a crater 5e10 nm in diameter and sends ions from this volume intimportant information about spatial co-existence of chemical species. (B) Comparison of nefunctional thiol/acrylated silane (bottom). This analysis reveals the presence of both acryl anmass spectra collected from glass substrates after attachment of linker (up) and streptavidinwell as to amino acids present in streptavidin.

2.2. Silane modification of glass substrates

Prior to silanization, glass slides were cleaned by immersion in piranha solutionconsisting of 3 parts of 95% (v/v) sulfuric acid and 1 part of 35% (w/v) hydrogenperoxide for 10 min. Subsequently, glass slides were thoroughly rinsed withdeionized (DI) water, dried under nitrogen and stored in class 10000 cleanroom.Immediately prior to silanization, glass substrates were treated in an oxygen plasmachamber (YES-R3, San Jose, CA) at 300 W for 5 min. Then the substrates wereimmersed for 12 h in a binary mixture of (3-acryloxypropyl) trimethoxysilane and 3-mercaptopropyl trimethoxysilane diluted to 0.1% v/v in anhydrous toluene. 1:1 M

s. (A) Schematic representation of surface bombardment with C60 projectiles. Impact ofo the SIMS instrument. Ions emitted from each impact are accounted for and providegative ion mass spectra of glass substrates modified with acrylated silane (up) vs. a bi-d thiol functional groups in the mixed silane assembled on glass. (C-D) The negative ion(bottom). These spectra demonstrate presence of masses assigned to peptide bonds as

Fig. 2. (continued).

J.H. Seo et al. / Biomaterials 32 (2011) 5478e5488 5481

ratio of the two silanes was used. Silanization was conducted in a glove bag filledwith nitrogen to minimize atmospheric moisture. After incubation, slides wererinsed with fresh toluene, dried under nitrogen and baked at 100 �C for 1 h. Thesilane-modified glass slides were stored in a desiccator before further use.

2.3. Characterizing bi-functional silane layer by TOF-SIMS and AFM

Secondary ion mass spectrometry (SIMS) equipped with time of flight (ToF)spectrometer was used to examine surfaces at different steps in the modification

procedure. A custom-built ToF-SIMS using primary C60þ of 26 keV total impactenergy generates negative secondary ion emissions from the topmost layers ofa negatively biased target. A feature technique is to run the ToF-SIMS in the event-byevent bombardment/detection mode [38]. It has been previously determined thatthe secondary ions are emitted from a hemispheric volume of w5e10 nm indiameter under a single projectile impact [39]. Each single impact was detected andrecorded as an individual event by a micro-channel plate (MCP) detector assembledwith eight metal anodes. An accumulation of several million events compriseda conventional secondary ion mass spectrum. The event-by-event bombardment/

J.H. Seo et al. / Biomaterials 32 (2011) 5478e54885482

detection approach allows us to unfold the co-emitted multiple secondary ions andextract a coincidental ion mass spectrum of co-located molecules within a nano-metric impact/emission volume [40]

Immobilization of avidin inside the microwells was also characterized topo-graphically using AFM (MFP-3D, Asylum Research Corp.) operated in tapping modeat a scan rate of 0.5 Hz.

2.4. Fabrication of hydrogel microwells

Photolithographic patterning of PEG hydrogel was conducted as previouslydescribed [32]. Briefly, a prepolymer solution containing PEG-DA and photoinitiatorwas spin-coated onto silanized substrates using Spintech S-100 (Redding, CA)operated at 800 rpm for 4 s. This prepolymer solution was then exposed to UV light(60 mWcm�2) through a chrome/sodalime photomask for 0.5 s using UV OmniCureseries 1000 light source (EXPO, Mississauga, Ontario, Canada). PEG-DA surfaceexposed to UV became cross-linked while unexposed PEG regions were easily dis-solved in DI water. This process resulted in the formation of 30 mm diametermicrowells with PEG hydrogel walls and glass bottom.

2.5. Attachment of Ab molecules on micropatterned surfaces

Micropatterned surfaces fabricated on acrylate/thiol silane layer were incubatedfor 2 h in 500 mMMAL-PEG2eNHS dissolved 1:1 mixture of DMSO and 1 � PBS (seeFig.1B). To analyze immobilization of avidin, microwells created onmono-functional(acrylated silane only) and bi-functional silane layers were incubated withstreptavidin-Alexa 546 and imaged using Agilent microarray scanner. GenePix Pro6.0 software (Molecular Devices, Downingtown, PA) was used for quantification offluorescent intensity. For cytokine capture and detection experiments, microwellarrays were incubated in 1 mg/mL of neutravidin for 1 h, followed by incubationwith 5 mg/mL of biotinylated IFN-g or TNF-a Ab. These surfaces were then challengedwith recombinant IFN-g of varying concentrations (from 1 ng/ml to 10 mg/ml). Thepresence of cytokine molecules on micropatterned surfaces was determined byattaching biotinylated anti-cytokine Abs followed by streptavidin-Alexa 546 (in1 � PBS with 1% BSA). The fluorescence in the microwells was imaged usinga confocal microscope (Zeiss LSM 5 Pa, Carl Zeiss, Inc., Thornwood, NY) and wasanalyzed with GenePix Pro 6.0 data analysis software (Molecular Devices,

Fig. 3. AFM analysis of neutravidin deposition on glass substrates modified with mono-acrylated silane before (A) and after (B) physical adsorption of neutravidin. (C-D) Surface topof neutravidin. Note presence of particles similar in size (6 nm) to neutravidin on glass sub

Downingtown, PA) to construct calibration curves of cytokine concentration vs.fluorescence intensity.

2.6. Detection of TNF-a release from micropatterned immune cells

Mouse macrophage cells (J774A) were cultured at 37 �C with 5% CO2 in phenolred-free Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetalbovine serum (FBS). These cells were grown in suspension culture in 50 mL biore-actor tubes (Techno Plastic Products) on a rolling apparatus (Stovall). The cells werepassaged two times a week by centrifuging and re-suspending in fresh culturemedia.

Glass pieces (w 1 � 1 in) with microwell arrays were outfitted with PDMSmicrofluidic channels. Design and fabrication of the microfluidic devices have beenprovided in detail in our previous publications [21,41]. Prior to cell seeding, 1 mLcell suspension was concentrated by centrifugation and was re-suspended inDMEM at w15 million cells/mL concentration. Cell suspension was infused into themicrofluidic channels and incubated for 30 min with microwells that were func-tionalized with TNF-a Abs as described in the previous section. Macrophages areknown to attach to Fc components of Abs and became adherent inside 30 mmdiameter microwells. Non-adherent cells were washed away. To induce cytokinerelease, macrophages were mitogenically stimulated for 3 h by 100 mg/mL PMA inDMEM. During cytokine release flow was stopped to minimize convection. Afterremoval of PMA solution, cellular micropatterns were incubated with biotin-anti-TNF-a for 30 min followed by avidin-Alexa546 (red color). Subsequently, cells werefixed with 4% PFA for 15 min and stained with DAPI for 5 min to visualize cellnuclei. Between each step, the sample was washed with 1�PBS for 5 min to removethe previous reagent. All steps described above were performed inside a micro-fluidic device.

3. Results and discussion

The goal of this study was to develop hydrogel microwell arraysfor sensitive cytokine detection. The novelty of this paper lies increating a bi-functional thiol/acrylate silane layer on glass, with

and bi-functional silanes. (A-B) Surface topography of glass substrates modified withography of thiol/acrylate silane layer before (C) and after (D) after covalent attachmentstrates modified with mixed silane layer.

Fig. 4. Fluorescence microscopy analysis of streptavidin adsorption in microwells. Individual microwells were 100 mm in diameter. (A) Physical adsorption of streptavidin-Alexa 546in microwells containing acrylate groups in the attachment sites. Note very weak fluorescence signal observed in the microwells that is comparable to non-specific fluorescence ofthe hydrogel walls. (B) Streptavidin-Alexa 546 covalent immobilized in microwells modified with thiol/acrylate silane layer. (C) Quantitation of fluorescence intensity showing w6times higher fluorescence in microwells created on a mixed silane layer.

J.H. Seo et al. / Biomaterials 32 (2011) 5478e5488 5483

acrylate groups promoting hydrogel attachment and thiol groupsused for orientedbindingof anti-cytokineAbs inside themicrowells.This micropatterning strategy allowed to significantly enhancesensitivity of cytokine detection inside the microwells.

3.1. Characterization of surface properties using ToF-SIMS and AFM

Fig. 1 details our surface modification strategy involving silani-zation of glass substrates and micropatterning of hydrogel PEGmicrowells. As demonstrated in this schema, the goal was to createa mixed silane layer containing acrylate moieties for hydrogelanchoring and thiol groups for attachment of avidin/biotin-Abconstructs. Throughout this study, we will be comparing thesensitivity of cytokine immunoassays created on mixed silane

Fig. 5. Characterization of IFN-g immunoassay in 30 mm microwells created on a mixed siarrays were challenged with 100 ng/mL (A) and 10 ng/mL (B) of recombinant IFN-g, thefluorescence intensity inside the microwells vs. IFN-g concentration. The limit of IFN-g det

layers to surfaces modified with mono-functional acrylated silanewhere avidin was physically adsorbed onto the surface.

We first wanted to verify that incubation of glass substrateswithsolution containing acrylate- and thiol-terminated trimethoxy-silanes resulted in deposition of both types of molecules. ToF-SIMSprovides an excellent means for characterizing chemical composi-tion of surfaces and is increasingly being used for analysis of bio-interfaces and micropatterned surfaces [40,42,43]. Cartoon inFig. 2A shows the principle of ToF-SIMS surface analysis employedhere to verify co-existence of acrylate and thiol groups on glasssubstrates. In brief, our surfaces were analyzed by 26 keV C60þ ToF-SIMS running in the event-by-event bombardment-detectionmode. C60 projectiles have previously been shown by us to produceupon impact hemispherical craters 5e10 nm in depth [39]. Usingevent-by-event bombardment-detection mode, we could then

lane layer and functionalized with neutravidin and biotin anti-IFN-g. (A-B) Microwelln stained with antieIFNeg-biotin and streptavidin-Alexa 546. (C) Calibration plot ofection inside the microwells was 1 ng/ml (60 pM).

J.H. Seo et al. / Biomaterials 32 (2011) 5478e54885484

analyze ions co-emitted from the same impact and could determineco-existence of acrylate and thiol groups. Fig. 2B shows bothfunctional groups being present on the glass surface after silani-zation. Comparison of glass slides coated with acrylated silane(upper panel Fig. 2B) to glass slides modified with bi-functionalsilane layer (lower panel Fig. 2B) reveals the presence of SHderivative peaks at m/z 32 (S-), 33(SH-), 64 (S2-/SO2

-) and 80 (SO3-)

only on the mixed silane surface. In contrast, peaks at m/z 119(SiCH2þSiO2OH-) and 179 (SiCH2þ (SiO2)2OH-) originating fromacrylated silane molecules were found on glass substrates modifiedwith both acrylated and mixed silanes.

The event-by-event technique allows to compute the fractionalsurface coverage of silanes [40,44]. Briefly, two co-emittedsecondary ions from an impacted/emitted nanovolume were

Fig. 6. Detection of multiple cytokines inside hydrogel microwells. PEG hydrogel microweincubated with a 1:1 mixture of antieIFNeg and anti-TNF-a Abs. (A) Brightfield image of hydmL IFN-g and TNF-a, subsequently microwells were incubated in a mixture of antieIFNmicrowells demonstrating that the two cytokines could be detected simultaneously. (E) ChResponses to 500 ng/mL (30 nM) IFN-g were compared for microwells modified with aninterpretation of the references to colour in this figure legend, the reader is referred to the

selected to calculate individual immobilized coverage of thiol silaneor acryl silane on the bi-functional surface. A fractional surfacecoverage of w 69% was obtained for the thiol silane (co-emittedions SH- and SO3

-) and w67% for the acryl silane (co-emitted ions:SiCH2þSiO2OH- and SiCH2þ (SiO2)2OH-). This result demonstratesthat the equimolar mixture of these two silanes in solution istransferred onto the surface of glass after silanization. Overall, ToF-SIMS analysis points to the assembly of both acrylate- and thiol-functionalized silanes from solution containing a mixture of thetwo silanes.

Beyond analysis of silane composition, SIMS was also used toverify the presence of heterobifunctional linker and avidin. Fig. 2Cand D show smaller and larger masses observed during SIMSanalysis of mixed silane surfaces incubated in MAL-PEG2eNHS

lls were fabricated on a mixed silane layer, functionalized with neutravidin and thenrogel microwells. (B-D) These microwells were simultaneously challenged with 500 ng/eg-PE and anti-TNF-FITC. Images show both red and green fluorescence inside thearacterizing sensitivity of cytokine immunoassay as a function of Ab immobilization.tieIFNeg as well as microwells containing antieIFNeg/anti-TNF-a combination. (Forweb version of this article.)

J.H. Seo et al. / Biomaterials 32 (2011) 5478e5488 5485

ester linker and avidin. Analysis of surfaces treated with linker (topspectra in Fig. 2C,D) revealed presence of acrylate and thiol groupsat m/z 32, 33, 119 and 179 as well as the linker at m/z 256(C11H16N2O5

-) and 270 (C12H18N2O5-). Surfaces incubated with

avidin, showed presence of peaks associated acrylate and thiolgroups as well as the linker and avidin (bottom spectra). Negativeions at m/z 91(Phe-) and 107 (Tyr-) were assigned to amino acidresidue peaks originating from avidin. In addition, substratescoated with avidin had lower intensities of glass-related masses,suggesting that protein coatingwasmasking the underlying surfaceduring mass spectra collection (Fig. 2C). Given that the depth ofprojectile penetration during SIMS analysis is estimated to be5e10 nm [39] the lack of glass peaks suggests dense avidin layerassembled on the substrate.

AFM was used to characterize surface topography of glasssubstrates modified with mono- and bi-functional silanes beforeand after deposition of avidin. In these experiments, physicaladsorption of avidin on mono-functional silane surfaces was vs.covalent immobilization on a bi-functional (thiol/acrylate) silanelayer. The latter surface was first activated with maleimide-PEG-NHS and then incubated with avidin. Fig. 3(AeB) shows AFMscans before and after avidin deposition onto glass substratesmodified with acrylated silane. As can be seen from these images,no appreciable changes were observed in surface topographybefore and after protein deposition. In contrast, AFM analysis on bi-functional silanes before and after avidin incubation revealsa significant difference in surface topography with appearance ofparticles w6 nm in size. These particles correlate well to thereported size of avidin [45]. Therefore, AFM study corroborates ToF-SIMS analysis pointing to the presence of avidin on the surface andsuggests that more avidin was retained on surfaces by covalentattachment.

3.2. Fluorescence characterization of avidin adsorption insidehydrogel microwells

Commercial micro-titer plates are constantly evolving toincrease the number of wells and thus improve throughput of theanalyses being performed. It may be argued that arrays ofmicrowells with dimensions similar to that of individual cellsrepresent are best suited for high-throughput cell analysis.Therefore, there is considerable interest in microfabricating arrays

Fig. 7. Schematic describing cytokine release studies. Step 1: Macrophages were captured ininduce cytokine production. Step 2: Binding of secreted TNF-a inside the microwells was detespecific cytokine-producing macrophages.

of wells for cell capture and analysis [46,47]. The approach ourgroup has employed is to photolithographically pattern PEGhydrogel microwells on glass so that each well consists of non-fouling PEG walls and glass attachment pads (see Fig. 1B)[20,32]. In this method, hydrogel attachment to glass is promotedby an acrylated silane coupling agent. However, it is almost alwaysthe case that microwells need to be further modified either withadhesive ligands to promote cell attachment or with Abs forimmunosensing applications. In such a scenario, biomoleculessuch as avidin or ECM proteins are physically adsorbed fromsolution onto acrylated glass attachment pads of microwells.Protein attachment does not occur on non-fouling PEG walls of themicrowells. While physical adsorption is sufficient for certainapplications (e.g. cell culture), it may not suffice for immuno-sensing where the density and orientation of Abs are key deter-minants of sensor response. Therefore, our present study wasfocused on modifying glass substrates with a mixture of twodifferent silane molecules to both promote gel anchoring and toenable covalent attachment of avidin as well as oriented bindingof Abs inside the microwells.

It should be noted, that there are other approaches for cova-lent attachment of proteins inside hydrogel microwells. Forexample, Koh and co-workers described a method wherebyhydrogel structures are fabricated on amine-functionalizednanoporous substrates [48]. Hydrogel attachment on thesesubstrates occurs based on intercalation of liquid prepolymerinto the pores of the substrate prior to photo-crosslinking anddoes not require a silane coupling layer. While such approach isquite promising, it necessitates the use of specific substrates (e.g.nanoporous alumina), whereas the method described in thepresent study is compatible with standard glass or plasticsurfaces used in tissue culture.

To analyze protein adsorption, hydrogel microwells werefabricated on glass substrates modified with either mono-functional or bi-functional silane layers and were then incubatedwith fluorescently-labeled streptavidin. Fig. 4(A,B) compares fluo-rescence in the microwells after incubationwith streptavidin-Alexa546. As seen from the images in Fig. 4A,B, much higher fluorescencewas observed in microwells modified with a bi-functional silanelayer compared to mono-functional silane. This suggests thata larger number of avidin molecules were retained inside micro-wells by covalent binding then by physical adsorption. Quantitative

side microwells pre-coated with anti-TNF-a Abs. Cells were mitogenically activated tormined by sandwich immunoassay. Fluorescence cytokine signal was co-localized with

J.H. Seo et al. / Biomaterials 32 (2011) 5478e54885486

analysis signal intensity (Fig. 4C) shows 6 times higher fluorescencesignal emanating frommicrowells fabricated on mixed silane layer.The data in Fig. 4 are significant as they demonstrate that covalentbinding to bi-functional silanes improves loading and retention ofprotein (streptavidin) within the microwells. It should also benoted that limited non-specific binding of avidin was observed onthe hydrogel sidewalls.

3.3. Performing cytokine immunoassays in hydrogel microwells

Building on the promising results presented in Fig. 4, wewanted to determine whether higher loading of streptavidin onmixed silane surfaces would translate into more sensitive cyto-kine immunoassays. We chose to detect IFN-g, since production ofthis cytokine is commonly used to determine presence of antigen

Fig. 8. Detection of TNF-a release from individual macrophages. PEG hydrogel microwells (Imaging of macrophages in hydrogel microwells. Brightfield images of cells captured in miorescent staining for TNF-a production in activated macrophages (B and D). Cells were mstreptavidin-Alexa 546. Cell nuclei were stained with Dapi (blue color). (E) Characterization oMicrowells with covalently immobilized avidin-biotin-Ab construct had 3 fold higher fluoreintensity was the same for cells captured in both types of microwells. The signal represents ain this figure legend, the reader is referred to the web version of this article.)

specific CD4 and CD8 T-cells in infectious diseases such as HIVand tuberculosis [5e9]. To test cytokine sensitivity, arrays ofhydrogel microwells containing anti-IFN-g Abs were exposed toIFN-g concentrations varying from 0.5 ng/ml to 10 mg/ml. Thepresence of cytokine molecules inside the microwells was deter-mined using biotinylated secondary Abs and streptavidin-Alexa546. Fig. 5(A, B) shows representative images of microwellarrays exposed to varying concentrations of IFN-g. As seen froma calibration plot presented in Fig. 5C fluorescence intensitychanged as a function of cytokine concentration with linear rangeextending up to 6 nM (100 ng/ml). The lower limit of IFN-gdetection for these microwells was 60 pM (1 ng/ml). This resultrepresents a w 10 fold improvement over the limit of detectionreported in our previous papers employing physical adsorption ofAbs [21,25].

30 mm diameter) were pre-coated with anti-TNF-a Abs and incubated with cells. (A-D)crowells constructed on mixed silane (A) and mono-functional silane (C). Immunoflu-itogenically stimulated for 3 h and then stained with anti-TNF-a-biotin followed byf TNF-a fluorescence in microwells created on bi-functional or mono-functional silanes.scence signal compared to sensing microwells created on monofunctional silanes. Dapin average fluorescence in 20 microwells. (For interpretation of the references to colour

J.H. Seo et al. / Biomaterials 32 (2011) 5478e5488 5487

Detection of multiple cytokines from the same cell population isimportant in immune cell analysis and disease [6,7]. As a steptowards multi-functional cell analysis, we wanted to co-depositanti-IFN-g and TNF-a Abs in the same microwell arrays and todemonstrate multi-cytokine detection in microwells. In theseexperiments, microwells were fabricated on glass substrates con-taining amixed silane andweremodifiedwith neutravidin followedbiotinylated anti-IFN-g and anti-TNF-a. Fig. 6(AeD) presents a seriesof images from microwells challenged with TNF-a and IFN-g dis-solved in1�PBSat500ng/mL (30nM)concentration.As canbe seenfrom these images, microwells respond to both TNF-a (red signal)and IFN-g (green signal) suggesting that two different cytokinetypes may be detected in the microwell arrays. Fig. 6E comparessensitivity of IFN-g response in microwells coated with only anti-IFN-g tomicrowellsmodifiedwith amixtureof anti-IFN-g and -TNF-a. This comparison points to decrease of IFN-g signal in microwellscontaining a mixture of anti-IFN-g/-TNF-a Abs. Such a result isexpected, given that the surface density of anti-IFN-g Abs likely 2times lower in the case when it is co-deposited with anti-TNF-a.Despite some signal loss, the sensitivity of the IFN-g immunoassayremains quite high. We envision immobilizing multiple Ab types inthe microwells in the future to enable detection of multiple cyto-kines secreted by immune cells.

3.4. Detecting TNF-a release from single macrophages in hydrogelmicrowells

Cytokine release from immune cells was investigated as the finalstep in characterization of micropatterned cytokine-sensingsurfaces. Macrophages are an important leukocyte subset respon-sible for immune surveillance in tissue and organs. These cellsrobustly secrete inflammatory cytokines such as TNF-a in order toeliminate infections. Cytokine release in macrophages may beinduced by stimulation with endotoxins (e.g. LPS) or mitogens. Inour experiments (described schematically in Fig. 7), mousemacrophages were captured in microwells created on either bi-functional or mono-functional silanes. In both cases, microwellswere modified with anti-TNF prior to cell seeding. Macrophagesbecame attached inside themicrowellse this is expected given thatmacrophages are known to adhere to Fc domains of Abs [49].Individual microwell dimensions of 30 mmdiameter were chosen tocorrespond to the size of single cells. As can be seen from bright-field images in Fig. 8, each microwell contained one or twomacrophages. Cells captured in the microwells were stimulatedwith PMA (mitogen) for 3 h to induce cytokine production.Experiments were done in Petri dishes without mixing tominimizeconvection and to ensure that secreted TNF-a is captured in thevicinity of the cell source. Presence of secreted cytokines wasdetermined by incubation of cellular micropatterns with bio-tinlyated-anti-TNF-a followed by streptavidin-Alexa 546. Fig. 8B,Dshow immunofluorescent staining images of macrophagescaptured on micropatterned glass substrates. As can be seen fromthese images, detection of cell-secreted TNF-a was much moresensitive in microwells containing covalently immobilized avidin-biotin-anti-TNF-a construct. It should be noted that TNF-a is notonly released but is also expressed on cell surfaces. This explainscell-associated fluorescence in both types of microwells tested.Quantifying fluorescence intensity revealed that the signal inmicrowells with engineered avidin-Ab attachment sites wasw3 times higher than in microwells containing physicallyadsorbed avidin-Ab molecules (see Fig. 8E) while the intensity ofDapi e fluorescent dye that stains cell nucleus, was comparable inboth types of microwells. The results presented in Fig. 8 underscorethe utility of designing attachment of recognition molecules insidemicrowells.

4. Conclusions

The objective of this study was to design micropatternedsurfaces for sensitive cytokine detection. PEG hydrogel photoli-thography was employed to create hydrogel microwells with non-fouling PEG walls and glass bottom. The question explored in thispaper was how to engineer the glass attachment pads inside themicrowells so as to increase sensitivity of cytokine immunoassays.Our solution was to move away from physical adsorption andtowards covalent/oriented attachment of the sensing elements. Forthis purpose, glass substrates were functionalized with a binarymixture of acrylate- and thiol-terminated silanes, where acrylategroups served to anchor hydrogel microwells while the thiol groupswere used to covalent link avidin in the glass attachment sitesinside the microwells. A series of experiments confirmed thatconsiderably higher density of avidin molecules was assembled inacrylate/thiol-modified microwells. This higher avidin concentra-tion likely translated into higher Ab loading and led to a 10 foldincrease in sensitivity of IFN-g detection in microwells created onmixed silane-modified glass. Importantly, we also demonstratedw3 times higher signal of cell-secreted TNF-a in microwells engi-neered for covalent avidin-Ab attachment. These more sensitivemicrowells will be used in the future for detection of multiplecytokines released by the immune cells. Improved sensitivity willalso allow more rapid detection of cytokine release from singlecells. The strategy of employing bi-functional silanes in conjunctionwith hydrogel micropatterning on glass has broad utility for cova-lent protein attachment and will have applications in cell/tissueengineering and biosensing.

Acknowledgement

The authors would like to thank Profs. Marcu and Louie for useof fluorescence microscopy equipment. The authors also thank Drs.Jun Yan and Yinghua Sun for helpful discussions. This work wassupported by NSF grants: EFRI 0937997 awarded to AR and CHE0750377 awarded to EAS.

References

[1] Tan TT, Coussens LM. Humoral immunity, inflammation and cancer. Curr OpinImmunol 2007;19:209e16.

[2] Notley CA, Ehrenstein MR. The yin and yang of regulatory T cells andinflammation in RA. Nat Rev Rheumatol 2010;6:572e7.

[3] O’Garra A, Barrat FJ, Castro AG, Vicari A, Hawrylowicz C. Strategies for use ofIL-10 or its antagonists in human disease. Immun Rev 2008;223:114e31.

[4] Diel R, Loddenkemper R, Meywald-Walter K, Niemann S, Nienhaus A.Predictive value of a whole blood IFN-g assay for the development of activetuberculosis disease after recent infection with Mycobacterium tuberculosis.Am J Respir Crit Care Med 2008;177:1164e70.

[5] Boom H. The role of T-cell subsets in Mycobacterium tuberculosis infection.Infect Agents Dis 1996;5:73e81.

[6] Caccamo N, Guggino G, Joosten SA, Gelsomino G, Matarese A, Salerno A, et al.Multifunctional CD4 T cells correlate with active Mycobacterium tuberculosisinfection. Eur J Immun 2010;40:2211e20.

[7] Casey R, Blumenkrantz D, Millington K, Montamat-Sicotte D, Kon OM,Wickremasinghe M, et al. Enumeration of functional T-Cell subsets byfluorescence-immunospot defines signatures of pathogen burden in tuber-culosis. Plos One; 2010:5.

[8] Rosenberg ES, Billingsley JM, Caliendo AM, Boswell SL, Sax PE, Kalams SA, et al.Vigorous HIV-1-specific CD4(þ) T cell responses associated with control ofviremia. Science 1997;278:1447e50.

[9] Zimmerli SC, Harari A, Cellerai C, Vallelian F, Bart PA, Pantaleo G. HIV-1-specific IFN-gamma/IL-2-secreting CD8 T cells support CD4-independentproliferation of HIV-1-specific CD8 T cells. Proc Nat Acad Sci 2005;102:7239e44.

[10] Brando B, Barnett D, Janossy G, Mandy F, Autran B, Rothe G, et al. Cyto-fluorometric methods for assessing absolute numbers of cell subsets in blood.Cytometry 2000;42:327e46.

[11] Cox JH, Ferrari G, Janetzki S. Measurement of cytokine release at the single celllevel using ELISPOT assay. Methods 2006;38:274e82.

J.H. Seo et al. / Biomaterials 32 (2011) 5478e54885488

[12] Karlsson AC, Martin JN, Younger SR, Bredt BM, Epling L, Ronquillo R, et al.Comparison of the ELISPOT and cytokine flow cytometry assays for theenumeration of antigen-specific T cells. J Immun Methods 2003;283:141e53.

[13] Folch A, Toner M. Microengineering of cellular interactions. Annu Rev BiomedEng 2000;2:227.

[14] Toner M, Irimia D. Blood-on-a-Chip. Annu Rev Biomed Eng 2005;7:77e103.[15] Cheng XH, Irimia D, Dixon M, Sekine K, Demirci U, Zamir L, et al.

A microfluidic device for practical label-free CD4þT cell counting of HIV-infected subjects. Lab Chip 2007;7:170e8.

[16] Han Q, Bradshaw EM, Nilsson B, Hafler DA, Love JC. Multidimensional analysisof the frequencies and rates of cytokine secretion from single cells by quan-titative microengraving. Lab Chip 2010;10:1391e400.

[17] Love JC, Ronan JL, Grotenbreg GM, van der Veen AG, Ploegh HL.A microengraving method for rapid selection of single cells producingantigen-specific antibodies. Nat Biotech 2006;24:703e7.

[18] Story CM, Papa E, Hu CCA, Ronan JL, Herlihy K, Ploegh HL, et al. Profilingantibody responses by multiparametric analysis of primary B cells. Proc NatAcad Sci 2008;105:17902e7.

[19] Kim H, Cohen RE, Hammond PT, Irvine DJ. Live lymphocyte arrays for bio-sensing. Adv Mat 2006;16:1313e23.

[20] Revzin A, Sekine K, Sin A, Tompkins RG, Toner M. Development of a micro-fabricated cytometry platform for characterization and sorting of individualleukocytes. Lab Chip 2005;5:30e7.

[21] Zhu H, Stybayeva GS, Silangcruz J, Yan J, Ramanculov E, Dandekar S, et al.Detecting cytokine release from single human T-cells. Anal Chem 2009;81:8150e6.

[22] Soen Y, Chen DS, Kraft DL, Davis MM, Brown PO. Detection and character-ization of cellular immune responses using peptide-MHC microarrays. PLOSBiol 2003;1:429e38.

[23] Bailey RC, Kwong GA, Radu CG, Witte ON, Heath JR. DNA-encoded antibodylibraries: a unified platform for multiplexed cell sorting and detection ofgenes and proteins. J Am Chem Soc 2007;129:1959e67.

[24] Zhu H, Macal M, George MD, Dandekar S, Revzin A. A miniature cytometryplatform for capture and characterization of T-lymphocytes from humanblood. Anal Chim Acta 2008;608:186e96.

[25] Zhu H, Stybayeva GS, Macal M, George MD, Dandekar S, Revzin AA. Micro-device for multiplexed detection of T-cell secreted cytokines. Lab Chip 2008;8:2197e205.

[26] Hermanson GT. Bioconjugate techniques, Vol. 1. Academic Press; 1996.[27] Blawas AS, Reichert WM. Protein patterning. Biomaterials 1998;19:595e609.[28] Kane RS, Takayma S, Ostuni E, Ingber DE, Whitesides GM. Patterning of

proteins and cells using soft lithography. Biomaterials 1999;20:2363e76.[29] Folch A, Toner M. Microengineering of cellular interactions. Ann Rev Biomed

Eng 2000;2. 227-þ.[30] Whitesides GM, Ostuni E, Takayama S, Jiang XY, Ingber DE. Soft lithography in

biology and biochemistry. Ann Rev Biomed Eng 2001;3:335e73.[31] Revzin A, Russell RJ, Yadavalli VK, Koh W-G, Deister C, Hile DD, et al. Fabri-

cation of poly(ethylene glycol) hydrogel microstructures using photolithog-raphy. Langmuir 2001;17:5440e7.

[32] Revzin A, Tompkins RG, Toner M. Surface engineering with poly(ethyleneglycol) photolithography to create high-density cell arrays on glass. Langmuir2003;19:9855e62.

[33] Dulcey CS, Georger JH, Krauthamer V, Stenger DA, Fare TL, Calvert JM. DeepUV photochemistry of chemisorbed monolayers-patterned coplanar molec-ular assemblies. Science 1991;252:551e4.

[34] Sagiv J. Organized monolayers by adsorption .1. Formation and structure ofoleophobicmixedmonolayers on solid-surfaces. J AmChemSoc1980;102:92e8.

[35] Wasserman SR, Tao YT, Whitesides GM. Structure and reactivity of alkylsi-loxane monolayers formed by reaction of alkyltrichlorosilanes on siliconsubstrates. Langmuir 1989;5:1074e87.

[36] Wayment JR, Harris JM. Controlling binding site densities on glass surfaces.Anal Chem 2006;78:7841e9.

[37] Lee KB, Jung YH, Lee ZW, Kim S, Choi IS. Biospecific anchoring and spatiallyconfined germination of bacterial spores in non-biofouling microwells.Biomaterials 2007;28:5594e600.

[38] Park MA, Gibson KA, Quinones L, Schweikert EA. Coincidence counting inTime-of-Flight Mass-Spectrometry - a test for chemical microhomogeneity.Science 1990;248:988e90.

[39] Li Z, VerkhoturovSV, Locklear JE, Schweikert EA. Secondary ionmass spectrometrywith C-60(þ) and Au-400(4þ) projectiles: depth and nature of secondary ionemission frommultilayer assemblies. Int J Mass Spec 2008;269:112e7.

[40] Chen L-J, Shah SS, Verkhoturov SV, Revzin A, Schweikert EA. Characterizationand quantification of biological micropatterns using cluster SIMS. Surf Inter-face Anal; 2010. doi:10.1002/sia3399. published online.

[41] Stybayeva G, Mudanyali O, Seo S, Silangcruz J, Macal M, Ramanculov E, et al.Lensfree holographic imaging of antibody microarrays for high-throughputdetection of leukocyte numbers and function. Anal Chem 2010;82:3736e44.

[42] Takahashi H, Emoto K, Dubey M, Castner DG, Grainger DW. Imaging surfaceimmobilization chemistry: Correlation with cell patterning on non-adhesivehydrogel thin films. Adv Funct Mater 2008;18:2079e88.

[43] Liu F, Dubey M, Takahashi H, Castner DG, Grainger DW. Immobilized antibodyorientation analysis using secondary ion mass spectrometry and fluorescenceimaging of affinity-generated patterns. Anal Chem 2010;82:2947e58.

[44] Chen L-J, Shah SS, Silangcruz J, Eller MJ, Verkhoturov SV, Revzin A, et al.Characterization and quantification of nanoparticle-antibody conjugates oncells using C60 ToF SIMS in the event-by-event bombardment/detection dode.Int J Mass Spect. in press:DOI 10.1016/j.ijms.2011.01.001.

[45] Kuzuya A, Numajiri K, Kimura M, Komiyama M. Single-molecule accommo-dation of streptavidin in nanometer-scale wells formed in DNA nano-structures. Nucl Acid Symp Ser 2008;52:681e2.

[46] Rettig JR, Folch A. Large-scale single-cell trapping and imaging using micro-well arrays. Anal Chem 2005;77:5628e34.

[47] Charnley M, Textor M, Khademhosseini A, Lutolf MP. Integration column:microwell arrays for mammalian cell culture. Integ Biol 2009;1:625e34.

[48] Lee HJ, Kim DN, Park S, Lee Y, Koh W- G. Micropatterning of nanoporousalumina membrane with poly(ethylene glycol) hydrogel to create cellularmicropatterns on nanotopographic substrates. Acta Biomater 2011;7:1281e9.

[49] Unkeless JC, Eisen HN. Binding of monomeric immunoglobulins to Fc recep-tors of mouse macrophages. J Exp Med 1975;142:1520e33.