Biosensors and Bioelectronics 64 (2015) 43–50

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

http://d0956-56

n CorrE-m1 Cu

journal homepage: www.elsevier.com/locate/bios

Detecting multiple cell-secreted cytokines from the sameaptamer-functionalized electrode

Ying Liu, Ying Liu 1, Zimple Matharu, Ali Rahimian, Alexander Revzin n

Department of Biomedical Engineering, University of California, Davis, Davis, CA 95616, USA

a r t i c l e i n f o

Article history:Received 27 May 2014Received in revised form30 July 2014Accepted 15 August 2014Available online 22 August 2014

Keywords:AptasensorElectrochemical detectionCytokineCell secretionMicrofluidics

x.doi.org/10.1016/j.bios.2014.08.03463/& 2014 Elsevier B.V. All rights reserved.

esponding author.ail address: arevzin@ucdavis.edu (A. Revzin).rrent address: ReLIA Diagnostic Systems Inc.,

a b s t r a c t

Inflammatory cytokines are secreted by immune cells in response to infection or injury. Quantification ofmultiple cytokines in parallel may help with disease diagnosis by illuminating inflammatory pathwaysrelated to disease onset and progression. This paper describes development of an electrochemicalaptasensor for simultaneous detection of two important inflammatory cytokines, interferon gamma (IFN-γ) and tumor necrosis factor alpha (TNF-α). To enable multiplexing, IFN-γ and TNF-α aptamers werelabeled with anthraquinone (AQ) and methylene blue (MB) redox reporters respectively. Randomimmobilization of two aptamer on gold exhibited redox peaks at �0.37V (AQ) and �0.15V (MB) vs.Ag/AgCl reference. When challenged with either IFN-γ or TNF-α, redox signal of the appropriate reporterchanged in concentration dependent manner. To demonstrate one possible application of this sensingapproach, electrodes were integrated into microfluidic devices and used to dynamically monitor cytokinerelease from immune cells. Two cell types, primary human CD4 T-cells and U937 monocytic cells, wereused to compare differences in cytokine secretions upon stimulation. These cells were infused into themicrofluidic devices and stimulated to commence cytokine production. Release of IFN-γ and TNF-α wasmonitored concurrently from the same small group of cells over the course of 2 h. The strategy ofencoding specific aptamer types with unique redox reporters allows sensitive and specific detection ofmultiple protein biomarkers from the same electrode.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

Cytokines are signaling proteins secreted by immune cells inorder to regulate a multitude of immune responses, includingproliferation, migration or activation of cells. Outside of keepinginfections at bay, production of inflammatory cytokines is closelytied to cancer (Moss and Blaser, 2005), atherosclerosis (Sack,2002), rheumatoid arthritis (Sarzi-Puttini et al., 2005), Alzheimer'sdisease, Parkinson's disease and multiple sclerosis (Lee et al.,2010). Interferon gamma (IFN-γ) and tumor necrosis factor alpha(TNF-α) are two common pro-inflammatory cytokines producedby immune cells upon stimulation with antigen, and they havebeen found to play an important role in multiple pathologicalconditions. For example, in HIV infected individuals, robust pro-duction of IFN-γ by CD4 T-cells has been associated with non-progression of the disease. In another example, detection of IFN-γrelease by T-cells is supplanting a century old skin test fordetection of tuberculosis (TB). In fact, in an effort to identify

Burlingame, CA 94040, USA.

blood-based correlates/markers of TB it was noted that T-cellproduction of only IFN-γ was associated with latent form of thisdisease whereas T-cell secretion of both IFN-γ and TNF-α corre-lated to active form of TB. Therefore, multiplexed detection ofcytokines from immune cells is highly significant.

Cytokines are typically detected using antibody-based immu-noassays. These immunoassays are sensitive specific and robust,however, they are time-consuming; requiring multiple washingand handling steps to achieve the readout. Also, the need formultiple washing and labeling steps makes antibody-based assayssuboptimal for dynamic monitoring of cellular secretions.

Aptamers are emerging as an effective alternative to antibodies,offering the advantages of high thermal/chemical stability, regen-erability and ease of modification. One of the most attractivefeatures of aptasensors is the simplicity with which an oligonu-cleotide can be designed into a beacons that produce signaldirectly upon binding of target analyte, without the need forlabeling and washing steps. Simplicity of getting the readoutmakes aptamer-based biosensors particularly suitable for pointof care testing. The fact that the signal comes on without the needfor additional labeling or washing steps, positions aptasensors wellfor dynamic monitoring of cellular activity.

Y. Liu et al. / Biosensors and Bioelectronics 64 (2015) 43–5044

Previously, our lab has developed hairpin structure aptamer forIFN-γ detection (Liu et al., 2013, 2010). The hairpin was thiolated,conjugated with a redox reporter methylene blue (MB), andimmobilized on a gold electrode. Binding of cytokine caused aconformational change in the aptamer, resulting in a decrease ofthe redox current. More recently, we have demonstrated anaptasensor for TNF-α detection (Kwa et al., 2014) and fabricatedelectrodes for simultaneous detection of IFN-γ and TNF-α from thesame small group of cells (Kwa et al., 2014; Liu et al., 2011). In thisrecent study, IFN-γ and TNF-α aptamers were labeled with sameredox reporter (MB) and were assembled on specific electrodesthrough a series of electrode protection and de-protection steps.We could then monitor electrochemical signals from specificelectrodes to discern changes in IFN-γ or TNF-α concentrations.However, we were dissatisfied with the complexity of functiona-lization protocol employed in this previous study and sought todevelop a more facile multiplexing strategy. Instead of using thesame redox reporter and encoding cytokine specificity based onelectrode location, we wanted to encode different aptamer typeswith redox reporters and then use the same electrode to decodechanges in cytokine concentrations (Scheme 1). There have beenreports describing the use of nanostructured electrodes for sensingbased on electrochemical redox spectra that arose due to multipleredox/electroactive moieties diffusing from solution and reactingon the electrode (Karimi-Maleh et al., 2013, 2014). Plaxco labrecently demonstrated the concept of redox encoding usingoligonucleotide sequences as target analytes (Kang et al., 2012).However, to the best of our knowledge, this strategy has not beenapplied to aptamer-based biosensors and protein detection. Toprove the concept, thiolated aptamers against IFN-γ and TNF-αwere labeled with anthraquinone (AQ) and methylene blue (MB)respectively. Electrodes functionalized with a mixture of the twoaptamers showed redox peaks at �0.37 V (AQ) and �0.15 V (MB)vs. Ag/AgCl reference. Importantly, correct redox peak shifteddownwards upon introduction of either IFN-γ or TNF-α(Scheme 1). No change in redox activity of the electrodes was

Scheme 1. Layout of the microfluidic sensing platform. The electrode array is encaseelectrodes are modified with different aptamers-redox reporter constructs. Either T-captasensors. Upon cytokine binding, aptamer changes conformation resulting in dedynamically monitored on the same electrode using square wave voltammetry (SWV) (

observed after challenges with nonspecific proteins. As the nextstep in this study, micropatterned electrodes were integrated withmicrofluidic devices and used for immune cell analysis. PrimaryT-cells and monocyte cell line were analyzed for production ofIFN-γ and TNF-α to demonstrate that secretion pattern was celltype specific. The production of both cytokines from a small groupof immune cells was monitored concurrently (from the samemicroelectrode) over the course of 2 h. In the future, we envisionincreasing the number of redox reporters and aptamers assembledon the electrode surface to enhance multiplexing power of thisstrategy.

2. Materials and methods

2.1. Materials

10� phosphate-buffered saline (PBS) without calcium andmagnesium, sodium bicarbonate (NaHCO3), anhydrous toluene(99.9%), dimethylformamide (DMF), acetone, bovine serum albu-min (BSA), 6-mercapto-1-hexanol (MCH), Tris-(2-carboxyethyl)phosphine hydrochloride (TCEP), poly(ethylene glycol) diacrylate(PEG-DA, MW 700), and 2-hydroxy-2-methyl-propiophenone(photoinitiator), T cell and monocyte activation reagents: phorbol12-myristate-13-acetate (PMA) and ionomycin were purchasedfrom Sigma-Aldrich (St. Louis, MO). Chromium etchant (CR-4S)and gold etchant (Au-5) were purchased from Cyantek Corporation(Fremont, CA). Positive photoresist (S1813) and developer solution(MIF-318) were purchased from Shipley (Marlborough, MA). 3-Ac-ryloxypropyltrichlorosilane was purchased from Gelest, Inc. (Mor-risville, PA). Monoclonal purified mouse anti-human CD4 Abs andanti-human CD14 Abs were purchased from Gelest, Inc. (Fullerton,CA). Human recombinant Interferon gamma (IFN-γ), Tumor ne-crosis factor alpha (TNF-α), Interleukin-12 (IL-12), Interleukin-6(IL-6), Interleukin-10 (IL-10) were purchased from R&D systems(Minneapolis, MN). Cell culture medium RPMI 1640 with

d in PEG hydrogel layer and integrated into PDMS microfluidic channels (A). Auells or monocytes are infused into microfluidic channels and are bound next tocreased electron transfer efficiency. Two cytokine types released by T-cells areB).

Y. Liu et al. / Biosensors and Bioelectronics 64 (2015) 43–50 45

L-glutamine was purchased from Life Technologies (Carlsbad, CA).Methylene blue-carboxylic acid succinimidyl ester (MB-NHS) waspurchased from Biosearch Technologies Inc. (Novato, CA). Anthra-quinone-2-carboxylic acid NHS ester (AQ-NHS) was purchasedfrom Emp Biotech GmbH. Berlin, Germany. All chemicals wereused without further purification.

The IFN-γ-binding aptamer and TNF-α-binding aptamer weresynthesized by IDT Technologies, San Diego, CA. The 34-mer IFN-γ-binding aptamer sequence was as follows: 5′-NH2-C6-GGG GTTGGT TGT GTT GGG TGT TGT GTC CAA CCC C-C3-SH-3′. The 28-merTNF-α-binding aptamer sequence was as follows: 5′-NH2-C6-rGnrGnrAnrGnrUnrAnrUnrCnrUnrGnrAnrCnrAnrAnrUnrUnrCnrGnrGnrAnrGnrCnrUnrCn -rC-C3-SH-3′. Phosphorothioates (or S-oli-gos, marked with n) were used to stabilize RNA against RNasedegradation. Both TNF-α and IFN-γ aptamers were modified at the3-terminus with a C3-disulfide [HO(CH2)6–S–S–(CH2)6–] linkerand at the 5-terminus with an amine group for redox probe(MB) conjugation. The aptamer were dissolved in 1� PBS buffer(pH 7.4). Thiolated aptamer was reacted with 10 mM TCEP for 1 hat room temperature to reduce the disulfide bond of the probe.

2.2. Conjugation of redox moieties to aptamers

The redox reporters MB-NHS and AQ-NHS were conjugated tothe 5′-terminus of amino modified TNF-α aptamer and IFN-γaptamer through succinimide ester coupling following previouslyreported protocols (Xiao et al., 2007). The resulting conjugatedsample was filtered with a centrifugal filter (Millipore, AmiconUltra 3K 0.5 mL) and stored at �20 °C before use.

2.3. Fabrication of micropatterned gold electrode array, PEGpatterning, and electrode modification

The electrode array layout was designed in AutoCAD andconverted into plastic transparencies by CAD/Art Services (Bandon,OR). Glass slides (75 mm�25 mm) were sequentially sputter-coated with 15 nm Cr adhesion layer and 100 nm Au layer byLGA Thin films (Santa Clara, CA). Standard photolithography andmetal etching techniques were used to micropattern the goldelectrodes as described previously (Yan et al., 2010). The Au/Cr

Scheme 2. Fabrication of microelectrode array and PEG hydrogel patterning around eelectrode (A) of aptamers onto the electrode and antibodies on neighboring glass regio

layers were etched to create eight micropatterned circular shapedelectrodes (300 mm in diameter) that were individually connectedto 2 mm�2 mm contact pad via 15 mm wide leads.

The glass substrates with photoresist protected gold microelec-trode were modified with 0.05% 3-Acryloxypropyltrichlorosilane (An-hydrous Toluene) to ensure covalent anchoring of hydrogel structuresonto the glass substrates in the following (Revzin et al., 2003). Afteracryl-silane modification, substrates were sonicated in acetone for2 min to remove photoresist and then placed in an oven for 3 h at100 °C to fully crosslink the silane layer. To create non-fouling PEGhydrogel micropatterns and define the sites for cell attachment, aprepolymer solution containing 20% PEG-DA (MW 700) and 1% (v/v)photoinitiator was loaded onto the silane modified glass and coveredwith a cover glass (24 mm�30mm�0.13 mm) to create a uniformlayer. A photomask was aligned and the substrate was irradiatedthrough the photomask for 1.5 s by a 365 nm UV light source (80mW/cm2, OmniCure Series 1000, Lumen Dynamics Group, Mississau-ga, Ontario, Canada). The regions exposed to UV light underwentradical polymerization and became cross-linked, while the unexposedregions were dissolved in DI water after 5 min of development. ThisPEG patterning strategy generates 800 mm diameter glass/Au regionswithin the non-fouling hydrogel layer, allowing the targeted attach-ment of cells next to electrodes (Scheme 2A).

A solution (1 mL) containing 5 mM MB-TNF-α aptamer and12 mM AQ-IFN-γ aptamer was pipetted on each electrode andincubated in a humidity chamber overnight in the dark. Theelectrodes were rinsed with copious amounts of DI water andthen immersed in an aqueous solution of 3 mM MCH for 1 h todisplace non-specifically adsorbed aptamer molecules and topassivate the electrode surface. After the MCH solution was gentlyrinsed away, a 0.5 mL solution containing either anti-CD4 or anti-CD14 Ab (0.1 mg/mL in 1� PBS) was pipetted onto each cellcapture region and incubated for 1h, allowing the antibodies to bephysisorbed onto the exposed silane areas only (Scheme 2B).

2.4. Integration of aptasensor with microfluidics and capture ofimmune cells

Microfluidic channels were fabricated out of poly (dimethylsiloxane) (PDMS) using standard soft lithography approaches. The

lectrodes to avoid nonspecific adsorption and create cell seeding regions next tons (B).

Y. Liu et al. / Biosensors and Bioelectronics 64 (2015) 43–5046

microfluidic device contained two flow chambers with width–length–height dimensions of 3 mm�10 mm�0.1 mm as well as aweb of auxiliary channels for applying negative pressure. ThePDMS device was secured on the glass substrates by applying avacuum to a spider web feature surrounding the device asdescribed previously (He et al., 2008).

For experiments with primary T-cells, blood was collected fromhealthy adult donors through venipuncture under sterile condi-tions with informed consent and approval from UC Davis IRB. RBCswere lysed using an ammonia chloride-based erythrocyte lysissolution (89.9 g NH4Cl, 10.0 g KHCO3, and 370.0 mg tetrasodiumEDTA in 10 L of deionized water) as described previously (Zhuet al., 2008). Human monocytic U937 cell line, CRL-1593.2/U-937(American Type culture collection, ATCC) were cultured in suspen-sion in 75 cm2 tissue culture flasks and incubated in RPMI 1640supplemented with 10% FBS and 100 U/mL penicillin/streptomycinand L-glutamine. Cells were seeded at 1�106/mL, passaged threetimes per week, and cultured under a 5% CO2/95% air humidifiedatmosphere at 37 °C. For T-cell and monocyte experiments, micro-fluidic devices were functionalized with anti-CD4 and anti-CD14respectively. Microfluidic devices were infused with cell suspen-sion at the rate of 10 mL/min. Upon filling the chamber with cells,the flow rate was decreased to 3 mL/min for 20 min to allow for cellcapture. The flow rate was raised to 20 mL/min subsequently for5 min in order to wash away non-specific cells.

2.5. Electrochemical detection of cytokine IFN-γ and TNF-α

Electrochemical measurements were carried out with a CHI842B potentiostat (CH Instruments, Austin, TX) operating inside ahomemade Faraday cage. A three electrode electrochemical cellwas constructed around a microfluidic device, with a flow-throughAg/AgCl (3 m KCl) reference electrode in the outlet, a Pt wirecounter electrode in the inlet and a set of four miniature workingelectrodes inside each channel. Detailed description of the micro-fluidic device is provided elsewhere (Matharu et al., 2013).Measurements were performed using square wave voltammetry(SWV) with a voltage range of �0.5 V to 0 V and 40 mV amplitude,60 Hz frequency.

After assembling a mixture of AQ-IFN-γ and MB-TNF-α apta-mers, the electrode was allowed to equilibrate in RPMI1640medium for 30 min, as determined by stable faradaic current.Human recombinant IFN-γ and TNF-αwas diluted with RPMI 1640at concentration ranging from 10–220 ng/mL and then infusedinto the microfluidic device for 30 min incubation to allow thecytokine-aptamer binding. Subsequently, microfluidic channelswere washed with RPMI 1640 for 5 min to rinse away nonspecificcytokines and then characterized using SWV. Binding of analytecaused the redox peak to shift down. The change in redox activitywas reported as signal suppression – a ratio of (initial peak redoxcurrent � final peak redox current)/initial peak redox current.

To demonstrate specificity, electrodes containing the twoaptamer types were challenged with 100 ng/mL IFN-γ and TNF-α, as well as nonspecific cytokines including IL-12, IL-6, IL-10 andBSA.

For cell secretion experiments, microfluidic devices with inte-grated electrodes were placed into a custom-made heating stageto maintain temperature of 37 °C. After capturing T-cells or U937monocytes, microfluidic devices were infused with PMA(50 ng/mL) and ionomycin (2mM) dissolved in phenol-red-freeRPMI 1640 media without serum to stimulate cells. A surgicalclamp was secured around the outlet tubing to eliminate con-vective mixing. SWV measurements commenced immediatelyupon mitogenic stimulation and were made every 20 min overthe course of 2 h.

3. Results and discussion

The goal of this paper was to develop a facile approach formultiplexed detection of cytokines using aptasensors. In thestrategy outlined in Scheme 1A, aptamers against IFN-γ andTNF-α were tagged with unique redox reporters and then as-sembled on the same electrode. Appearance of the target cytokineat the electrode surface caused specific redox peak to shift down-ward in concentration-dependent manner.

3.1. Microelectrode array fabrication and microfluidic deviceassembly

The arrays of Au electrodes were fabricated on glass usingphotoresist lithography, metal etching, and photopolymerizationof PEG hydrogel. A layout of individual electrodes within the arrayis shown in Fig. 1A. Eight circular electrodes (300 μm in diameter)were arranged in two rows to fit four electrodes into one micro-fluidic channel. The glass substrates containing array of goldelectrodes were then modified with acrylate-terminated silane tocreate anchoring sites for subsequent attachment of PEG hydrogellayer. PEG hydrogel was micropatterned to create circular regions(800 μm in diameter) around Au electrodes (Fig. 1B). The rationalefor this step was to use non-fouling properties of PEG in order toprecisely control placement of cells around the biosensor (Yanet al., 2010, Yan et al., 2009). As illustrated by Scheme 2B, thissurface modification scheme allowed to create a complex biointer-face comprised of non-fouling PEG glass, glass functionalized withcell-capture antibodies (anti-CD4 or anti-CD-14, depending on thecell type used) and aptamer-functionalized Au electrode. Fig. 1Cdemonstrates that T-cells preferentially attached next to sensingelectrode on glass regions containing anti-CD4.

3.2. Proving the principle of simultaneous detection of IFN-γ andTNF-α from the same electrode

Signal transduction in the reagent-less electrochemical apta-sensors is presumed to occur via structure switching of the nucleicacid construct (Lai et al., 2007; Xiao et al., 2005). We havepreviously designed IFN-γ beacon to contain a hairpin with stemregion partially obstructing aptamer nucleotides (Liu et al., 2010).Our goal in this study was to demonstrate that, when labeled withunique redox reporters and assembled on the same electrode, IFN-γ and TNF-α aptamers respond individually to their cognateligands. SWV curve was used to monitor changes in the intensityof the reduction current of the redox reporter (Scheme 1A). Thisdecrease in signal is often reported as “signal suppression”: theratio of SWV peak current loss at a given cytokine concentration toSWV peak current in the absence of analyte (Lai et al., 2007; Xiaoet al., 2007).

To prove this concept we used AQ (potential at �0.37 V vs. Ag/AgCl) for IFN-γ aptamer and MB (potential at �0.15 V vs. Ag/AgCl)for TNF-α aptamer. The NHS-functionalized redox tags werecovalently conjugated to amine groups at the 5′-terminus ofaptamer molecules. Given that cytokine binding is likely asso-ciated with changes in the conformational state of the aptamersteric hindrance and surface density of aptamer molecules arelikely to play an important role in biosensor performance (Ricciet al., 2007; White et al., 2008). Therefore, we wanted tocharacterize how surface density and ratio of IFN-γ and TNF-αaptamers affected sensor responses. To optimize the ratio of AQ-IFN-γ aptamer and MB-TNF-α aptamer, we compared the perfor-mance of four different concentrations/ratios (Supporting Fig. 1).Cathodic redox peak for AQ went up from 1.15 mA to 7.15 mA whenMB-TNF-α aptamer/AQ-IFN-γ aptamer ratio was decreased from5 to 0.417 while redox peak for MB-TNF-α remained unchanged.

Fig. 1. Micropatterned sensing surfaces. (A) Low magnification view of the arraywith working electrodes, leads and contact pads. A set of four electrodes isintegrated into one microfluidic channel (B). Higher magnification view of PEGgel regions defining cell attachment sites on glass. (C) Capture of cells on antibody-modified glass region next to aptasensing electrode.

Y. Liu et al. / Biosensors and Bioelectronics 64 (2015) 43–50 47

The MB peak current was not affected by increasing AQ-IFN-γaptamer, demonstrating that the electrode surface was not satu-rated by MB-TNF-α aptamer and there was space for co-immobi-lization of multiple aptamers. Based on these experiments, theoptimal mixture was to contain 12 mM AQ-IFN-γ aptamer and5 mM MB-TNF-α aptamer. We found it important to have higherconcentration of AQ-labeled aptamer because the upward baselineshift at more negative potentials masked AQ redox peak at lowerconcentrations. Both aptamer surface coverages (Г) are calculatedby the equation Г¼Q/nFA. The total charge passed the surface (Q)is calculated by integrating the redox peaks. The AQ-IFN-γ apta-mer surface density was estimated to be (2.0070.14)�1013

molecules/cm2, and the MB-TNF-α aptamer surface density wasestimated to be (2.1870.06)�1013 molecules/cm2. This surface

density is in general agreement with other reports employinghairpin aptamers (Ricci et al., 2007; White et al., 2008), and belowthe threshold of surface saturation density (6.5770.55)�1013

molecules/cm2 as previously reported in our group (Liu et al.,2010).

A series of experiments was designed to demonstrate responseof dual cytokine aptasensors. In the first set of experiments,sensing electrodes were challenged either with IFN-γ or TNF-αindividually. Fig. 2A and B shows typical redox activity of theelectrode containing two aptamer populations with AQ (�0.37 V)and MB (�0.15 V) redox reporters. As seen from Fig. 2A, challen-ging sensing electrode with IFN-γ of varying concentrationsresulted in decrease of AQ redox current but had no effect onMB redox peak. Reversing the experiment and infusing differentconcentrations of TNF-α caused MB redox peak to decrease whileposition of AQ peak remained largely unchanged. The SWV resultswere converted into calibration curve by plotting redox peakcurrent vs. concentration of cytokines, and peak current wasreported in terms of signal suppression percentage as discussedabove (Fig. 2C and D). As shown in Fig. 2C, AQ cathodic peak (IFN-γaptamer) changed in response to varying concentrations of IFN-γprotein (9–130 ng/mL), with signal suppression reaching satura-tion at 45%. Importantly, MB peak (TNF-α aptamer) remainedunchanged during challenges with IFN-γ protein. When challen-ging dual cytokine aptasensor with recombinant TNF-α, weobserved a linear relationship from 9 to 88ng/mL for MB cathodicpeak with signal suppression saturating at 41% (Fig. 2D). The signalsaturated at lower percentage for dual aptasensor due to thecompetition between two aptamers. The limits of detection werecalculated to be 6.35 ng/mL for IFN-γ and 5.46 ng/mL for TNF-αbased three times of signal to noise ratio. These values were higherthan the detection limits of 1 and 2 ng/mL for IFN-γ and TNF-αrespectively reported by us previously for aptasensors detectingindividual cytokines (Liu et al., 2010, 2011). This result is notunexpected and may be attributed to decrease in surface densityof individual aptamer types. The stability of the dual aptasensorwas also checked by periodic electrochemical measurements overthe course of 2.5 h in RPMI 1640 medium under physiologicalconditions (Supporting Fig. 2). The signal remained unchangedover this time period suggesting that aptasensors are stable for theduration of the experiment.

3.3. Characterizing specificity of dual cytokine aptasensor

Specificity is a key requirement for any biosensor. To demon-strate that dual cytokine aptasensors did not respond to non-specific proteins, aptasensors were challenged with inflammatorycytokines that may be produced by T-cells or monocytes -IL-6, IL-10, IL-12, as well as BSA. Target analytes, TNF-α and IFN-γ, wereused as a positive control. Fig. 3 demonstrated MB and AQ cathodicpeaks in response to different cytokines. Signal suppression valuesfor both AQ (IFN-γ aptamer) and MB (TNF-α aptamer) redox peakswere less than 5% after exposure to 100ng/mL of non-specificproteins. Challenging dual aptasensor with 100ng/mL TNF-αresulted in 40.8% signal suppression of MB peak, �44-fold higherthan response from AQ peak. Signal suppression of AQ peak was35.76% due to challenge with 100ng/mL IFN-γ, �39-fold comparedwith MB peak response. The dual aptasensor demonstrated highspecificity to target cytokines and minimal cross-reactivity be-tween the two populations of aptamers assembled on theelectrode.

Fig. 2. Dual aptamer electrode responding to varying concentrations of IFN-γ and TNF-α. (A) Challenging dual cytokine aptasensor with IFN-γ concentrations ranging from 10to 220 ng/mL, causes AQ redox peak to shift downward while MB peak does not change. This suggests that only IFN-γ aptamers conjugated to AQ become engaged withcytokine molecules. (B) Challenging dual cytokine sensor with TNF-α concentrations ranging from 10–220 ng/mL leads to decrease in MB but not AQ redox current. (C, D)Calibration curves constructed by plotting sensor responses to various concentrations of cytokines. The data points and error bars represent average and standard deviationsof measurements from four different sensing electrodes.

Fig. 3. Specificity and cross-reactivity of dual cytokine aptasensor. Electrodes werechallenged with TNF-α, IFN-γ, IL-12, IL-6, IL-10 and BSA at 100 ng/mL. Error bars arestandard deviation from the average of four different aptasensors.

Y. Liu et al. / Biosensors and Bioelectronics 64 (2015) 43–5048

3.4. Simultaneous detection of IFN-γ and TNF-α release fromimmune cells

To validate application of dual cytokine aptasensor for dynamicmonitoring of cell secretions, miniature electrodes were integratedwith microfluidic devices and tested with primary human T-cellsand human monocyte cell line U937. Activated T-cells are knownfor robust production of IFN-γ as well as TNF-α, whereas mono-cytes produce significant amounts of TNF-α but little IFN-γ. Bytesting these two immune cell types, we were able to demonstratethat response of the sensing device accurately reflects cytokineprofiles of the cells. When carrying out T-cell experiments,peripheral blood mononuclear cells (PBMCs) were cultured intissue culture flasks for two days. Monocytes are known to adhereto tissue culture plastic over this time period while T-cells andB-cells are known to remain anchorage independent. Thus, float-ing cells collected from the culture flask were expected to beT- and B-lymphocytes with minor contamination from monocytes.These cells were then infused into microfluidic channels contain-ing electrodes and anti-CD4-modified glass regions. While we did

Fig. 4. Continuous monitoring of IFN-γ and TNF-α release from captured cells using dual cytokine aptasensor. Cytokine are produced from activated T-cells (A) and U937monocytes (B). After mitogenic stimulation, measurements are made every 20 min for 2.5 h. Changes in redox peak are converted into signal suppression. The error bars arestandard deviation from the average number of four different aptasensors.

Y. Liu et al. / Biosensors and Bioelectronics 64 (2015) 43–50 49

not carry out cell purity characterization in this paper, ourprevious results demonstrated that captured cells stain positivefor CD3 and CD4 surface antigen, identifying the cells as CD4T-cells (Liu et al., 2011). The dimensions of individual attachmentsites allows �800 cells to be captured. The electrochemicalexperiment commenced immediately after stimulation of immunecells with mitogen, PMA/Ionomycin. Fig. 4 summarized cytokinesecretion data from T-cells and U937 monocytes. The redox peaksof AQ and MB were measured every 20 min for 2 h and the signalsuppression values were calculated for different time points. Weobserved saturation of the signal suppression values of �50% forAQ redox peak (IFN-γ) and �40% for MB peak (TNF-α). Theelectrochemical signals measured with dual aptamer electrodeswere comparable to cell secretion signals recorded by us pre-viously using single aptamer electrodes (Kwa et al., 2014; Liu et al.,2011). Similar saturation behavior was observed in the calibrationexperiments of Fig. 2, suggesting that signal leveled off due tosaturation of binding sites and not because of decrease in cellsecretory activity. As expected, we observed T-cells producingsignificant amounts of IFN-γ and TNF-α, whereas monocytessecreted TNF-α but little IFN-γ (Fig. 4C and D).

4. Conclusion

Development of multi-analyte aptasensors for rapid and sensi-tive detection of cell-secreted cytokines has significant implica-tions for disease diagnostics and basic science. Here we describe amultiplexing concept based on labeling of specific aptamers withunique redox reporters. This strategy allows assembling redox-encoded aptamers as a random mixture on the surface of the sameelectrode and then detecting one vs. another target analyte bymonitoring redox peak shifting. This concept was proven usingIFN-γ and TNF-α aptamers functionalized with AQ and MB redoxreporters respectively. To demonstrate dynamic monitoring ofcytokine release from a small group of cells, micropatternedaptasensing electrodes were packaged within nonfouling PEG gellayer and integrated into a microfluidic device. T-cells or mono-cytes were captured and mitogenically stimulated inside micro-fluidic channels. Cytokine detection was monitored continuouslyover the course of 2 h. Significant findings of this paper included:1) simplicity of dual aptasensor construction, 2) high specificity tothe target analyte, 3) sensitivity and limit of detection of dualcytokine aptasensor was similar to that of single analyte aptasen-sor. In the future, multiplexing capability of this approach may be

Y. Liu et al. / Biosensors and Bioelectronics 64 (2015) 43–5050

enhanced by incorporating other redox reporters to create elec-trochemical spectra with multiple redox peaks where upward ordownward changes in peak position may be used to discern thepresence and the concentration of multiple target analytes.

Acknowledgments

This work was supported by National Science Foundation. Wealso thank Dr. Yandong Gao for his assistance with microfluidicsexperiments.

Appendix A. Supplementary information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.bios.2014.08.034.

References

He, Z., Stybayeva, G., Macal, M., Ramanculov, E., George, M.D., Dandekar, S., Revzin,A., 2008. Lab Chip 8, 2197–2205.

Kang, D., White, R.J., Xia, F., Zuo, X.L., Vallee-Belisle, A., Plaxco, K.W., 2012. NPG AsiaMater 4, e1–e6.

Karimi-Maleh, H., Biparva, P., Hatami, M., 2013. Biosens. Bioelectron. 48, 270–275.Karimi-Maleh, H., Tahernejad-Javazmi, F., Ensafi, A.A., Moradi, R., Mallakpour, S.,

Beitollahi, H., 2014. Biosens. Bioelectron. 60, 1–7.Kwa, T., Zhou, Q., Gao, Y., Rahimian, A., Kwon, L., Liu, Y., Revzin, A., 2014. Lab Chip

14, 1695–1704.Lai, R.Y., Plaxco, K.W., Heeger, A.J., 2007. Anal. Chem. 79, 229–233.Lee, Y.J., Han, S.B., Nam, S.Y., Oh, K.W., Hong, J.T., 2010. Arch. Pharm. Res. 33, 1539–

1556.Liu, Y., Zhou, Q., Revzin, A., 2013. Analyst 138, 4321–4326.Liu, Y., Tuleouva, N., Ramanculov, E., Revzin, A., 2010. Anal. Chem. 82, 8131–8136.Liu, Y., Yan, J., Howland, M.C., Kwa, T., Revzin, A., 2011. Anal. Chem. 83, 8286–8292.Matharu, Z., Enomoto, J., Revzin, A., 2013. Anal. Chem. 85, 932–939.Moss, S.F., Blaser, M.J., 2005. Nat. Clin. Pract. Oncol. 2, 90–97.Revzin, A., Tompkins, R.G., Toner, M., 2003. Langmuir 19, 9855–9862.Ricci, F., Lai, R.Y., Heeger, A.J., Plaxco, K.W., Sumner, J.J., 2007. Langmuir 23, 6827–

6834.Sack, M.N., 2002. Pharmacol. Ther. 94, 123–135.Sarzi-Puttini, P., Atzeni, F., Shoenfeld, Y., Ferraccioli, G., 2005. Autoimmun. Rev. 4,

153–161.White, R.J., Phares, N., Lubin, A.A., Xiao, Y., Plaxco, K.W., 2008. Langmuir 24, 10513–

10518.Xiao, Y., Lai, R.Y., Plaxco, K.W., 2007. Nat. Protoc. 2, 2875–2880.Xiao, Y., Lubin, A.A., Heeger, A.J., Plaxco, K.W., 2005. Angew. Chem. Int. Ed. 44, 2–5.Yan, J., Pedrosa, V.A., Simonian, A.L., Revzin, A., 2010. ACS Appl. Mater. Interfaces 2,

748–755.Yan, J., Sun, Y.H., Zhu, H., Marcu, L., Revzin, A., 2009. Biosens. Bioelectron. 24, 2604–

2610.Zhu, H., Macal, M., Jones, C.N., George, M.D., Dandekar, S., Revzin, A., 2008. Anal.

Chim. Acta 608, 186–196.