Multiplexed electrochemical immunosensor for label-free detection of ...

Journal of Electroanalytical Chemistry 773 (2016) 53–62

Contents lists available at ScienceDirect

Journal of Electroanalytical Chemistry journal homepage: www.elsevier.com/locate/jelechem

Multiplexed electrochemical immunosensor for label-free detection of cardiac markers using a carbon nanofiber array chip Rakesh K. Gupta 1, Ruchi Pandya 2, Theodore Sieffert 2, M. Meyyappan, Jessica E. Koehne ⁎ Center for Nanotechnology, NASA Ames Research Center, Moffett Field, CA 94035, United States

a r t i c l e

i n f o

Article history: Received 25 September 2015 Received in revised form 11 April 2016 Accepted 15 April 2016 Available online 19 April 2016 Keywords: Biosensors Vertically aligned carbon nanofibers Cardiac proteins Nanoelectrode array Differential pulse voltammetry Electrochemical multianalyte immunosensor

a b s t r a c t We present an electrochemical multianalyte or multiplexed immunosensor for simultaneous label free detection of cardiac markers panel, comprising of C-reactive protein, cardiac troponin-I and myoglobin. The multielectrode biosensor chip contains nine identical but electrically isolated microelectrodes arranged in a 3 × 3 array configuration. Each electrode contains carbon nanofiber nanoelectrodes grown vertically using plasma enhanced chemical vapor deposition. A hydrophobic photoresist layer, lithographically etched on the chip, exposes the electrodes and helps to selectively immobilize the antibody probes for the three target cardiac biomarkers using carbodiimide chemistry. The real-time label free detection of the three cardiac markers from a mixture is demonstrated with high sensitivity and selectivity. Detection in complex protein mixtures in human blood serum does not show any false positives from non-specific protein adsorption. The results show that the present sensor can serve as a miniaturized, low cost lab-on-a-chip system for the detection of various biomarkers in healthcare, environmental monitoring and security applications. Published by Elsevier B.V.

1. Introduction Rapid and accurate diagnosis of acute myocardial infarction (AMI) can help in devising appropriate and timely surgical procedures or cardiovascular disease therapy for heart patients, thereby avoiding lifethreatening situations such as heart attack or heart failure [1–5]. An appropriate approach for cardiac risk management can be an independent monitoring of the blood concentration of standard cardiac markers before and after an AMI event and likewise determining the blood concentration levels of some of the early cardiac markers useful for diagnosis/prognosis of a heart patient [6,7]. An alternative approach is the simultaneous determination of blood concentration of biomarkers specific to AMI and systemic inflammation (SI) - a state that occurs much earlier to myocardial necrosis; this can be done using an array of electrodes or devices with each containing respective bio-recognizing elements (capture antibodies, for example) [8]. The blood concentration of standard biomarkers such as cardiac troponin T & I (cTnT and cTnI), creatine kinase (CK-MB) and myoglobin (Mb) remain elevated after the occurrence of AMI whereas the concentration levels of early markers such as cardiac reactive protein (CRP), myeloperoxidase (MPO), natriuretic peptide, both B-type natriuretic peptide (BNP) and N-terminal ProBNP, and P-selectine in blood plasma rise on the occurrence of symptoms of SI [6–9]. The simultaneous detection of certain ⁎ Corresponding author. E-mail address: [email protected] (J.E. Koehne). 1 Currently at the University of Manchester, UK. 2 These authors contributed equally.

http://dx.doi.org/10.1016/j.jelechem.2016.04.034 1572-6657/Published by Elsevier B.V.

specific cardiac biomarkers thus becomes important for the accurate detection or monitoring of cardiovascular disease, and the choice of the cardiac biomarkers for simultaneous detection can be made from the list given above. There are several examples of immunoassays developed for the accurate and sensitive detection of a single cardiac protein with limits of detection (LOD) down to ng/mL to pg/mL [10–18]. Recently, electrochemical analysis for the simultaneous detection of different targets using several parallel single-analyte immunoassays or the simultaneous amperometric determination of multianalytes using multiplexed assays has shown impressive potential [19–21]. The real time detection of analytes can be performed using either a label-free electrochemical impedimetric immunosensor [22] or a labeled amperometric immunosensor that uses labeled antibodies: the labels include florescent conjugated enzymes, metal ions, magnetic or non-magnetic nanoparticles as capture probes with enzymatic amplification features [23,24]. For example, simultaneous electrochemical multianalyte immunosensors using single label or multiple labels have recently been demonstrated [25–30]. The use of labeled antibodies for each type of antigens and florescent detection method add to the complexity of integrating such a device in hand-held point-of-care-testing (POCT) systems. Alternatively, a wide range of label-free array based immunosensors for the simultaneous detection of cardiac biomarkers has been reported. Arntz et al. described an immunosensor for early and rapid label-free real-time detection of CK-MB and Mb using an array of seven microfabricated cantilevers [31]. The relevant capture antibodies were immobilized covalently on distinct cantilevers, and

54

R.K. Gupta et al. / Journal of Electroanalytical Chemistry 773 (2016) 53–62

independent detection of dual cardiac markers was obtained through the cantilever deflection induced by the stress generated upon antigen– antibody interaction. Deflection sensitivity for Mb was reported to be below 20 μg/mL. Differential measurements between the sensors and reference cantilevers were used to reduce the effect of thermal drifting as well as cantilever instabilities generated through the reagent injection system; nevertheless, the non-specific adsorption can still generate false positive signals and the use of differential signal measurements adds to the sensor complexity. Zhang et al. reported an integrated system consisting of an array of silicon nanowires interfaced with readout circuit that can perform label-free simultaneous detection of cTnT, CK-MM and CK-MB in serum with high sensitivity (100 fg/mL) and selectivity [32]. Most of the above-mentioned biosensor platforms either involve expensive and complex instrumentation or sophisticated numerical algorithms limiting their use in research laboratory settings. In the present work, an array based immunosensor for the simultaneous detection of cardiac marker panel - comprising of an early marker CRP and two established markers, Mb and cTnI - is reported. The biosensor uses a nanoelectrode array (NEA) of vertically aligned carbon nanofibers (VACNFs), and the tips of these nanofibers are functionalized covalently with the bio-recognizing elements (cardiac capture antibodies). The target markers are electrochemically detected from the mixture solution, thus developing the sensor into a lab-on-a-chip offering advantages such as low sample volume, multiplexing capability, potential low cost per test through wafer-scale fabrication and simple electrochemical detection strategy with improved sensitivity and specificity. The sensor performance in terms of target antigen quantification, electrode cross talk and specificity is presented and also compared against devices reported previously in the literature. The physiological and disease state concentrations of these three cardiac biomarkers are well known. The CRP concentration is N3 μg/mL in human serum several hours after an AMI event [33,34]. The normal serum concentration of cTnI is below 0.4 ng/mL and any value above 2.0 ng/mL represents

high risk [35,36]. The normal myoglobin concentration is 100– 200 ng/mL, which can reach levels of 420–2000 ng/mL in serum after the AMI event [37]. The presented approach is able to meet these clinical levels while satisfying all the other metrics mentioned above. 2. Experimental 2.1. Chemicals and regents Troponin I (cTnI, 20 μg lyophilized powder) protein from human heart, monoclonal anti-Troponin I (anti-cTnI, 1 mg protein per mL (Biuret))) antibody produced in mouse, C-reactive protein (CRP, 2.1 mg protein per mL (lowry)) from human plasma, anti-human C-reactive protein (anti-CRP, 54.7 mg protein per mL (Biuret)) antibody (produced in goat), myoglobin from human heart (Mb, ≥ 95%, SDS-PAGE, 2 mg protein per mL (lowry)), anti-myoglobin (antiMb, 1:500 protein dilution) produced in rabbit, 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (EDC, ≥ 99%), glycerol (for molecular biology, ≥ 99%), N-hydroxysuccinimide sodium salt (sulfoNHS, ≥ 98% (HPLC)) were purchased from Sigma Aldrich (St. Louis, MO). Powdered skim milk as blocking agent was purchased from Saco Mix'n Drink (Middleton, WI). The stock CRP solution was stored at 4 °C whereas the anti-cTnI, anti-CRP, anti-Mb and Mb solutions were stored at − 20 °C. Highly pure de-ionized water (18.2 MΩ.cm) from super-Q Millipore system was used throughout the study. CRP-free human blood serum was purchased from Fitzgerald and used without modification. Phosphate buffered saline (PBS, 10 mM, pH 7.4) was prepared by dissolving PBS sachet (Sigma Aldrich, St. Louis, MO) in de-ionized water and was filtered using a 0.22 μm membrane filter (Millipore Durapore PVDF from Sigma Aldrich, St. Louis, MO) before every use. The 1× PBS solution contains the following 137 mM NaCl; 2.7 mM KCl; 10 mM Na2HPO4; 2 mM KH2PO4. Shipley i-Line photoresist (SPR220.7) and resist developer

Fig. 1. Fabricated biosensor chip. a) SEM image of etched 3 × 3 array device chip exposing individual electrodes (scale 800 μm), b) SEM image of an individual electrode with resist layer coatings (scale 100 μm), c) 2D AFM scan of random surface (12 μm × 12 μm) of an electrode after etching, confirming the presence of exposed CNFs (bright dots), d) AFM line profile indicating CNF height in nm (vertical axis) and e) 3D AFM micrograph revealing the VACNFs protruding above the planer oxide surface.


55

(MF-26 A) and resist remover (1165) were purchased from Rohm and Haas Electronic Materials (MA, USA). Potassium hexacyanoferrate (III) (K4[Fe(CN)6]) was prepared by mixing 5 mM K4[Fe(CN)6] in 1 M KCl (Sigma Aldrich, St. Louis MO) and used as electrolyte for electrochemical readouts. All regents used in the study were of analytic grade. 2.2. Fabrication of multiple electrodes biosensor chip A detailed description of the wafer scale fabrication of biosensor chips with VACNFs on each electrode has been reported in [38,39] and a brief account on the fabrication is given in Supplementary Information along with additional steps of coating and etching of a hydrophobic photoresist (SPR220.7) layer. Fig. 1 shows the fabricated biosensor chip. Details related to electrochemical characterization are given in Supplementary Information. A three-electrode electrochemical cell system, consisting of a high quality platinum wire as counter electrode, saturated calomel electrode (SCE, Princeton Applied Research, Oakridge, TN) as reference electrode and each of the nine electrodes as the nine distinct working electrodes, was used. A specially designed Teflon liquid cell with an o-ring can hold ~300 μL of electrolyte solution and the SCE was suspended into the electrolyte solution from the top. 2.3. Functionalization and immobilization of probe antibodies The protocols for immobilizing capture antibodies on the VACNFs for single analyte immunosensor have been previously reported [39–41]. For the multianalyte detection scheme, the stock antibody solutions (anti-CRP, anti-Mb and anti-cTnI) were diluted with 20% glycerol in PBS (10 mM, pH 7.4) for desired lower concentrations (20 μM). A 10–12 nL of specific antibody probe solution was injected onto the particular electrode (three electrodes for each type of cardiac markers in the 3 × 3 array) using a nano injection instrument (Nanoject II, Drummond Scientific Co.), which is comprised of an optical microscope, nanopipette for injecting small volumes (2.3 to 69 nL) supported by highly accurate displacement motors, injection speed control and electronic switch for solution release. The antibody immobilization scheme is presented in Supporting Information and shown in Fig. S3. 2.4. Detection and quantification of target proteins Prior to performing the simultaneous detection of multiple targets, the chip containing three cardiac antibodies (Fig. S3 (f)) was first tested for a single analyte (Mb) to evaluate the detection and cross reactivity or electrode cross talk, and the results (CV and DPV curves) obtained are shown in Fig. 2. For simultaneous detection of all three cardiac proteins, a 150 μL antigen mixture solution containing CRP (5 μg/mL), Mb (5 μg/mL) and cTnI (2 μg/mL) in PBS (10 mM, pH 7.4) was prepared in 1:1:1 volume ratio. A 15–20 μL antigen mixture solution was dropped on the chip as shown in Fig. 3 (a). The chips were incubated at room temperature for 1 h during which the specific cardiac markers in the mixture solution were allowed to react and bind covalently with the probes. That is, CRP antigens bind with anti-CRP probes (red), Mb antigens bind with anti-Mb probes (blue) and cTnI antigens bind with anticTnI probes (green) as shown in the schematic of Fig. 3(a). To remove the non-specific molecules, the chips were rinsed with deionized water for 15 min following steps similar to the immobilization of antibodies and evaluation of bare electrode kinetics. The chips, during rinsing either with buffer or deionized water, were placed in the shaker unit at 30 °C to ensure complete removal of non-specific molecules and loosely attached target antigens. The chips were allowed to dry at room temperature before recording the electrochemical readout. To evaluate the simultaneous detection of individual cardiac markers, only DPV current (base line fitted) were analyzed from the nine bare electrodes, after their subsequent modification with specific antibodies and after the antigen binding. The presence of particular target antigens, e.g., CRP on anti-CRP antibodies, Mb on

Fig. 2. Selected CV and DPV curves for bare and subsequently modified electrodes recorded in electrolyte solution containing 5 mM K4[Fe(CN)6] in 1 M KCl; a) CV: bare (blue), linker\anti-Mb (red), linker\anti-Mb\Mb[5 μg/mL] (green), b) DPV: bare (blue), linker\anti-Mb (red), linker\anti-Mb\Mb[5 μ/mL] (green). (a) and (b) present the target Mb binding on anti-Mb) and c) DPV: bare (blue), linker\anti-Mb (red), linker\anti-CRP\Mb[5 μg/mL] (green). This panel presents the nonbinding of Mb antigens on anti-CRP(cross reactivity).

anti-Mb antibodies and cTnI on anti-cTnI antibodies was detected by investigating the % change in the DPV current peaks and the results are plotted in Fig. 3(b). Additionally, to evaluate the quantification capabilities of the developed EMI, three aliquots for each cardiac biomarker (CRP (0.05 μg/mL, 0.5 μg/mL, 5 μg/mL), Mb (0.05 μg/mL, 0.5 μg/mL, 5 μg/mL) and cTnI (0.02 μg/mL, 0.2 μg/mL, 2 μg/mL)) were prepared by diluting the stock concentrations with an appropriate quantity of PBS (10 mM, pH 7.4) buffer solution. The antigen mixture solutions were then prepared by mixing various concentrations of one marker with fixed concentrations of the other two markers within the ranges specified above. DPV readouts were recorded from all nine electrodes of the array chip for each concentration of selected marker and the average % change in DPV peak currents for all cases with errors

56


Fig. 3. Simultaneous detection of three cardiac markers: a) 3-D schematic of antibody immobilization on specific electrodes of the 3 × 3 array chip using a nano-injector system and a mixture solution containing three target markers. Normalized DPV peak current with respective SDV error for: b) simultaneous detection of three cardiac markers, c) CRP (0.05, 0.5 & 5 μg/mL), Mb (5 μg/mL) and cTnI (2 μg/mL), d) CRP (5 μg/mL), Mb (0.05, 0.5 & 5 μg/mL) and cTnI (2 μg/mL) and (e) CRP (5 μg/mL), Mb (5 μg/mL) & cTnI (0.02, 0.2 & 2 μg/mL).

bars (representing the standard deviation from current data of the three electrodes for each cardiac marker) is plotted in Fig. 3(c–e). This experiment for the simultaneous detection of three cardiac markers from the mixture was repeated twice to evaluate the reproducibility of the detection results. A set of representative voltammograms from the above experiments is provided in Fig. 4. To verify the specificity and cross reactivity (electrode cross talk) of the EMI, DPV data curves collected during the CRP concentration studies were investigated for evaluating the electrode cross talk. The detection of non-specific cardiac protein binding was further investigated by performing control protein detection experiments using similar set of conditions and protocols as discussed above. Fig. 5 presents the electrode cross talk results of CRP quantification experiment. 2.5. Detection of CRP in human serum To evaluate the sensor's performance in complex biological mixtures, detection was performed after exposing the sensor to human blood serum. Briefly, anti-CRP antibody probes were attached to the VACNFs using the same approach described above. The sensor surface was blocked using 2% weight to volume skim milk in PBS by placing 50–100 μL of the skim milk blocking solution on the device surface

and incubating for 1 h at room temperature. Gentle decanting of the excess blocking solution followed by a gentle stream of nitrogen was performed. Next, 50 μL of 15% CRP-free serum and 50 μL of 15% CRP-free serum spiked with 0.1 μg/mL CRP were placed on the surface and incubated at room temperature for 60 min. DPV scans were performed of the bare VACNFs and after antibody immobilization, sensor blocking by skim milk and exposure to human serum as shown in Fig. 6. 3. Results and discussion Fig. 1 (a) shows a scanning electron microscopy (SEM) image (recorded at low beam current and energy) of a photoresist coated array chip where the resist layer is selectively etched to expose the nine electrodes. A ~50 μm wide channel in the resist layer connects each electrode to a contact pad at the far end (not shown here). The hydrophobic resist coating around each electrode helps in holding a drop (~10–12 nL) of the specific antibody solution on a selected electrode and the connecting channel (Fig. 1 (b)) that helps in holding the excess antibody solution, if any, thereby preventing the antibody solution droplet from bleeding onto the adjacent electrode. SEM images of the electrodes before the hydrophobic resist layer (SPR220.7) formation were obtained by scanning an appropriately high electron beam current at the electrode surface to


57

Fig. 4. Selected DPV curves for bare and subsequently modified electrodes recorded in electrolyte solution containing 5 mM K4[Fe(CN)6] in 1 M KCl. (a–c) Denote selective detection of CRP, cTnI and myoglobin respectively with target mixture of CRP[0.05 μg/mL] + Mb[0.05 μg/mL] + cTnI[0.02 μg/mL]. (d–f) Are the corresponding control experiments; (d) anti-CRP and target mixture of Mb[2.1 mg/mL] + cTnI[20 μg/mL], (e) anti-cTnI and target mixture of CRP[0.05 μg/mL] + Mb[0.05 μg/mL] and (f) with anti-Mb and target mixture of CRP[2.0 mg/mL] + cTnI[20 μg/mL]. In all figures, blue: bare electrode; red: after adding linker and antibody; green: detection run after admitting the mixture target containing all three antigens.

investigate the presence of exposed CNF tips e.g. bright dots seen in Fig. S1 (d) (Supplementary Information). It is, however, not possible to scan the electrode surface for CNF tips after the resist layer formation because the high beam electron current exposes the resist layer making it unstable and etched, which in turn renders it unsuitable for use. Alternatively, the CNFs tips were inspected from atomic force microscopy (AFM) scans recorded with high resolution and lower scan rates. Fig. 1 (c) shows an AFM micrograph of a random region, 2D cross-section (12 μm × 12 μm) in one of the exposed electrode, which provides a closer look at the electrode surface topography. Sparsely seen bright dots (1 μm apart) in Fig. 1 (c) are the exposed CNF tips, and Fig. 1(d) (a line section) depicts the height profile of the VACNFs embedded in the planar SiO2 oxide layer. Fig. 1(d) confirms ~70–90 nm diameter (measured at half-width points on the height plot) and ~ 14 nm (average) height of VACNFs above the SiO2 surface. Fig. 1(e) depicts the 3D schematic of the same region showing exposed CNF tips protruded above the planar oxide surface. Therefore, Fig. 1(c–e) clearly indicate the

complete removal of the resist layer from the electrode surface exposing the fiber tip, which is necessary for protein binding. 3.1. Sensor performance: label free simultaneous detection The simultaneous detection of multiple analytes is actually achieved by performing several single analyte testing operations in parallel as explained under the experimental section. The immunosensor consists of an array of nine identical but electrically isolated electrodes, in which each electrode is capable of measuring specific cardiac antigens (CRP, Mb, cTnI) with similar LOD and specificity as reported previously for single CRP immunosensor [40,41]. The reported LODs for cTnI and CRP are ~20 pg/mL (0.6 aM) and ~ 11 ng/mL (90 pM) respectively [40,41]. To keep the antibodies hydrated during the immobilization procedure, antibody solutions containing 20% glycerol in PBS and the incubation reaction between the antibodies and VACNFs was allowed for 4 h. The presence of 20% glycerol in antibody solutions, however, slows down

58


the diffusion of charged antibodies towards the –COOH groups on the fiber tips and therefore, the ~4 h incubation period has been observed to be sufficient for suitable amount of antibodies to be immobilized on the electrode surface. A ~ 10–12 nL of solution containing specific antibodies (anti-CRP, anti-Mb, anti-cTnI) was spotted on selected electrodes (Fig. S3 (e) in Supporting Information), without bleeding into other electrodes and Fig. S3 (f) shows a 3D schematic of an array chip with three cardiac antibodies (anti-CRP (20 μM), ant-Mb (20 μM) & anti-cTnI (20 μM)) immobilized on selected electrodes, which is used for label free simultaneous detection and quantification of cardiac markers panel. The electrode surface is probed by the redox probe molecule (K4[Fe(CN)6]) using CV and DPV scans in order to evaluate the immunosensor response to the binding of a target antigen from the antigen mixture solution with the corresponding antibody immobilized on a particular electrode. The probe molecules exchange electric charge, one electron per molecule, with the VACNFs when a potential is scanned between the working electrode and the counter electrode. The presence of antibody probes and target antigens bound on antibodies (i.e., formation of a dielectric layer on working electrode) impedes the exchange of charge, which reduces the bare electrode CV steady state current (Iw) or DPV peak current (Ip) providing a qualitative detection of specific antibodies and target antigens binding from the solution. The extent of decrease in “Iw” or “Ip” of an antibody-immobilized electrode is dependent on the target antigen concentration in the mixture. In addition, the successful development of a label free sensor for simultaneous detection of multiple targets should eliminate cross-talk among the different electrodes; in addition, for a specific target, the sensor response must be independent of the remaining targets present in the mixture. That is, there must be no cross-reactivity between the analytes and non-relative antibodies leading to non-specific antigen binding (false positive detection). Furthermore, the assay incubation time should be appropriate for all the analytes. The results obtained and presented below address all of these factors. Initially, to confirm specific antigen binding and prevent electrode cross talk, a 15 μL solution containing 5 μL of Mb (5 μg/mL) in PBS was prepared instead of an antigen mixture solution and spotted on the array chip with three cardiac antibodies immobilized in the order shown in Fig. S3 (f). CV and DPV scans were recorded for all nine electrodes after the washing steps, and selected scans are shown in Fig. 2. Fig. 2 (a) shows CV curves for three stages of the electrodes: bare, after immobilization of anti-Mb antibody and Mb (5 μg/mL) antigens

bound on anti-Mb. The steady state current reduces from ~174 nA for the bare electrode to ~128 nA after functionalization with the anti-Mb antibodies and further reduces to ~ 80 nA when Mb antigens bind with the anti-Mb antibodies. The decrease in steady state current is due to the fact that the diffusion of ions (K4[Fe(CN)6]) towards the working electrode is impeded first by the linker\anti-Mb bimolecular dielectric layer and later by the linker\anti-Mb\Mb dielectric layer. A similar trend can be seen in the corresponding DPV curves shown in Fig. 2 (b) and the respective DPV peak current values are approximately 54 nA, 22 nA and 10 nA for the three stages of the electrode. Nonspecific binding can lead to electrode cross talk (cross-reactivity) and result in false positive signals. The DPV scans of the electrodes immobilized with anti-CRP or anti-cTnI were analyzed for the qualitative evaluation of the electrode cross talk. Fig. 2(c) shows the DPV curves recorded for an electrode immobilized with anti-CRP antibodies. The close overlapping of linker/anti-CRP (Ip = ~22 nA) and linker/antiCRP/Mb (Ip = ~23.4 nA) curves implies the absence of Mb on anti-CRP, i.e. the absence of nonspecific binding. Similar results were obtained for electrodes immobilized with anti-cTnI as well (data not shown here). The negligible cross talk can be anticipated due to the following two key reasons: 1) binding of Mb antigens with anti-CRP or anti-cTnI is not possible as anti-Mb probes show binding affinity only for Mb antigens and 2) there is sufficient separation (~300 μm) between any two adjacent electrodes of the array in addition to the dielectric nature (SiO2) of the surface between the electrodes; this means that any nonspecific adsorption between the two electrodes does not affect the electrode current and hence, zero electrode cross talk. Fig. 2(a–c) confirm the qualitative detection of the Mb antigens by an anti-Mb immobilized electrode and negligible cross-reactivity or cross talk on electrodes with anti-CRP captures antibodies; further quantitative evidence is given below. The simultaneous detection of three cardiac proteins from a mixture solution was recorded next by using a similar array chip containing three capture antibodies. Fig. 3(a) shows the 3D schematic for label free simultaneous detection of targeted cardiac antigens from mixture solutions. Fig. 3(b) presents the normalized current data for the simultaneous detection of CRP (5 μg/mL), Mb (5 μg/mL) and cTnI (2 μg/mL) from the antigen mixture solution spotted on the array. The normalization of currents becomes necessary owing to the variation in initial currents of the bare electrodes. The error bar represents the standard deviation of the data from repetitions of the experiment. The average current plots for the simultaneous detection of CRP, Mb and cTnI closely

Fig. 5. Cross reactivity or electrodes cross talk. Normalized DPV peak currents from each set of three electrodes immobilized with three antibody probes when mixture solutions of three concentration decades of CRP concentration in constant background of Mb and cTnI were incubated.


Fig. 6. Detection of CRP in human blood serum. a) Control experiment with differential pulse voltammograms of electrodes with bare VACNF (blue), immobilized anti-CRP (red), skim milk blocking layer (purple), and CRP-free serum (green). b) Differential pulse voltammograms with electrodes of bare VACNF (blue), immobilized anti-CRP (red), skim milk blocking layer (purple), and CRP-free serum spiked with 100 ng/mL CRP (green).

follows the trend obtained for the single antigen detection plots (e.g., ΔIp(%) = ~54 from Fig. 2(b)). Studying the sensor response by varying just one antigen concentration while keeping the other two fixed further helps to evaluate the detection sensitivity and dynamic performance (Fig. 3(c–d)). These plots for three concentration decades of CRP, Mb and cTnI reveal that the normalized current increases with the increase in specific protein concentration in the mixture solution. Though the limit of detection down to few ng/mL can be achieved using the array chip, the three concentration decades chosen for the cardiac proteins panel and studied herein cover the clinically important range useful for prognosis /diagnosis of heart patients [2]. As mentioned earlier, these chips have been individually shown to provide sensitive detection of cTnI [40] down to 20 pg/mL and CRP [41] down to 11 ng/mL. Here, we simply kept the concentration of the three targets in the same range (5 μg/mL down to 20 ng/mL) purely to compare the device response signals, especially from the cross talk point of view. Fig. 4 give evidence on efficient low concentration detection of specific proteins and zero-cross talk or no interference from non-specific proteins. Fig. 4 (a–c) represents the selective DPV curves recorded simultaneously from three electrodes, selecting

59

randomly one from each column of specifically immobilized antibodies (see Fig. 3(b)). The antigen mixture solution contains three cardiac proteins in low concentrations (CRP (0.05 μg/mL), Mb (0.05 μg/mL) and cTnI (0.02 μg/mL). It is clear from Fig. 4 (a–c) that the sensor efficiently detects low concentrations of specific proteins from the mixture. A 7–10% change in DPV peak currents is observed for CRP (Fig. 4 (a)), Mb (Fig. 4 (b)) and cTnI (Fig. 4 (c) by the respectively modified electrodes. Fig. 4 (d-f) represents the selective DPV curves recorded from three electrodes, selecting randomly one from each device immobilized with different antibodies (see Fig. S4 a–c under Supplementary Information). The antigen mixture contains two non-specific control proteins of their stock concentration (CRP (2.0 μg/mL), Mb (2.1 μg/mL), cTnI (20 μg/mL) for each type of antibodies, for instance, a binary mixture of Mb (2.1 mg/mL) and cTnI (20 μg/mL) on anti-CRP immobilized electrode (Fig. S4a). DPV curves in Fig. 4 (d-f) clearly provide evidence that there is no cross talk or interference (false positive) from the detection of non-specific control protein mixture. The nanoelectrodes (which are about 1 μm apart and embedded in SiO2 matrix) containing capture antibodies do not show any binding at all with the non-specific control targets, and the adsorbed non-specific proteins (if any) on the SiO2 surface do not contribute to the redox current. Hence the DPV antibodies and target currents overlap. Fig. 3 shows the normalized DPV data collected from the similar curves as shown in Fig. 4 (a–c) for each concentration of the three proteins studied in this work. The error bars represent the standard deviation of three repetitions. Fig. S4 (Supplementary Information) presents the normalized DVP date collected from three repeats each control tests. Fig. S4 further confirms that there was no interference (or cross talk) from the non-specific proteins in the mixture solution, even when the non-specific proteins were of very high concentrations (stock concentrations). Specificity of the developed sensor was further evaluated and validated by incubating a mixture of non-specific control proteins. The specificity tests were repeatedly performed for all three different cardiac proteins as mentioned earlier. Fig. S4 presents the specificity tests, which show no binding at all for the non-specific target proteins mixture irrespective of the higher concentrations of the control proteins in mixtures. Fig. S4(a–c) depict the 3D schematics of the actual scheme employed for incubating a mixture of two mismatched cardiac proteins on an array chip immobilized with some particular cardiac antibodies, for instance, a binary mixture of Mb (2.1 mg/mL) and cTnI (20 μg/mL) on anti-CRP immobilized electrode (Fig. S4a). Fig. S4 (d) plots the normalized currents from all the three specificity tests for CRP, Mb and cTnI biomarkers. On an average, a 10–14% increase (negative change, y-axis) in normalized current after non-specific antigens mixture incubation is measured for all the three tests, which is contrary to what has been seen in the case of specific proteins binding, confirming the zero non-specific protein capturing (false positive signal) by the sensor. The negative change in normalized currents can be attributed to the removal of loosely bonded antibodies from the electrode surface due to stringent 15 min washing after non-specific protein mixture incubation and before the electrochemical readout, which is in consistent with the results shown in Fig. 2(c). Zero false positive signals as well as no cross talk confirm the high specificity of the sensor. Fig. 5 shows the normalized DPV data obtained from the CRP concentration study from Fig. 3(c) in excess cTnI and Mb. The average normalized currents obtained for Mb and cTnI from the antigen mixture solutions containing CRP (0.05 μg/mL, 0.5 μg/mL, 5 μg/mL) in Mb (5 μg/mL) and cTnI (2 μg/mL) do not show large variations, thus verifying the absence of protein cross reactivity or electrode cross talk; this feature has been absent in previous multiplexed, enzymatic labeled immunosensors [20,21,26,42]. 3.2. Detection of CRP in human blood serum To further test the specificity, the sensor was prepared for CRP detection with immobilized anti-CRP and then exposed to a complex mixture

60


of proteins in human CRP-free serum. For these studies, the VACNF sensor surface needed to be blocked with 2% skim milk to ensure no nonspecific adsorption of proteins. No further decrease in current was observed in Fig. 6(a) upon exposure to CRP-free serum after adding the blocking agent, indicating no non-specific adsorption of proteins in the human serum. When the VACNFs were exposed to CRP-free serum spiked with 0.1 μg/mL CRP, a dramatic decrease in current was observed compared to that after adding the blocking agent as seen in Fig. 6(b). This indicates that CRP binds specifically to VACNFs in complex protein mixtures such as human serum. It should be noted that the high currents (10−6 A) in Fig. 6 (a, b) are simply due to the very high density of VACNFs exposed on the electrodes after CMP. 3.3. Utility of the developed multiplexed immunosensor and comparison with the literature The present multiplex sensor enables facile label free multianalyte detection as discussed above. Here we did not use a PDMS based microfluidic channel or any other complex lithographically etched fluid transport mechanism on the chip; nevertheless, we successfully immobilized individual target specific probe proteins (anti-CEP, anti-cTnI, anti-Mb) on specific electrodes. Probes immobilization was performed by dropping 10–12 nL of probe solution through nano-injector focused on an individual electrode of 3 × 3 array. The positive tone photoresist around each electrode helps in holding the drop of 10–12 nL of probe solution without bleeding into the other electrodes, those are just 200 μm apart (see Supplementary Information). Thus, there is no use of syringe injected flow of chemical, biological and washing solutions through microfluidic channels, which is common in most sensors [27–30]. The other advantages of the proposed approach include the use of traditional carbodiimide chemistry to immobilize three different non-labeled antibodies without decorating the nanofiber electrodes with any nanoparticles. The significantly low sample volume (15–20 μL) is useful in the event of frequent monitoring of the cardiac proteins in the blood serum of a patient. The approach can be extended to detect nine different targets from other mixture solutions under physiological conditions. The multiplexed immunosensor reproduces the results of the single

analyte immunosensor, reported earlier with accuracy and precision [40,41]. Table 1 presents a comparison of the present multiplexed immunosensor with devices reported in the literature based on the number of markers detected, labeling, analysis time, detection limits, dynamic range, sample size and different detection methods involved. The present sensor is superior in terms of LOD, dynamic ranges and sample size for Mb analyte compared with nanomechanical cantilever (NMC) sensor [31], however, it is slower than NMC. The electrochemical immunosensor presented in ref. [28] is superior to our device in terms of LOD and dynamic range for both cTnI and CRP analytes, but the use of labeled nanoparticles adds to the overall cost and number of steps involved; given that our detection range is within clinical expectations for all three markers, this is not an issue. The acoustic wave immunosensors [29] present similar LOD for cTnI and Mb with faster analysis time but the lower sample size requirements and label free detection of three cardiac markers in the clinically important range make our sensor a better choice. Similar to carbon nanofibers, silicon nanowires can also provide ultrasensitive detection of cardiac proteins with LOD in the range of fg/mL as reported in ref. [32], which is superior in almost all the parameters mentioned in Table 1; however, the mass production of reproducible nanowire based immunosensor with same LOD and detection sensitivity is very difficult because of the dispersion of individual nanowires and assembling of the nanowire based two terminal devices. In contrast, our work is already on wafer scale. All other multianalyte sensors reported in ref. [27,30,43–48] are fluorescent/beads label based sensors that use optical means for the detection of cardiac proteins in the range of fg/mL to μg/mL, which is no doubt essential for the early diagnosis of AMI but their use is limited to central lab testing facilities only. The handheld point-of-care systems have a high demand in medical applications, where either electrochemical or electronics based biosensors need miniaturized electronics, low power consumption, small sample volume etc. for the direct and simultaneous monitoring of multiples analytes at high speed. The multianalyte immunosensor developed here, when integrated with an electrochemical readout circuit, can help in designing a novel miniaturized lab-on-a-chip device for

Table 1 Comparison of the present multiplexed immunosensor with other devices in the literature. Markers

Label

Analysis Limit of detection (LOD) time (ng/mL) (m)

Dynamic range (ng/mL)

@

CK, Mb

Label free

10

50,000, 20,000

50,000, 20 × 103–100 × 103 Electro-mechanical

40

[31]

cTnI, CRP

Quantum dots (CdTe, ZnSe) Gold nanoparticles

–

0.01–50, 0.5–200

Electro-chemical

30

[28]

0.05–100, 1.1–100.6, 16–1016

Mass vs frequency

160

[29]

1–10−5, 1–10−5, 1–10−5

Direct electronic detection

–

[32]

1–190, 0.4–150

Optical

100

[30]

1–120,

Optical

2

[27]

0.36–42.9 1.47–50.1 0.86–69.3 1–2 × 103, 0.1–0.5 4.8–80, 26.9–230 3.0–15, 3.0–30 1.5–125, 0.27–40 0.75–125, 0.19–50

Optical

200

[43]

Optical Optical Optical Optical Optical

30 – 500 – –

[44] [45] [46] [47] [48]

Nano-mechanical cantilever (NMC) Electrochemical Immunosensor (ECI) Acoustic wave immunosensor (AWI) Nanowires based immunosensor

CK-MB, CK-MM, cTnI

Label free

30

Capillary waveguide fluoroimmunosensor *Micromosaic immunoassays

CK-MB, cTnI

HRP and TSA-AF488 Cy3, Fluorescence

30

HRP/PS-Beads

20

0.1 × 10−3 (5 aM), 7.3 × 10−3 (307aM) 0.05 1.1, 16.0 10−6, 10−6, 10−6 0.2, 0.5 30, 100,000, 50,000 0.30, 0.51, 0.24

Fluorescence Cy5 FAMB, FEB HRP/luminol Fluorescence

– 5 270 60 15

1–2 × 103, 5 4.8, 26.9 3.0, 2.0 1.5, 0.27 0.75, 0.19

ELISA sandwich $ Multiplexed fluorescent Planar TIRF Cleaved tag immunoassay Evidence cardiac panel Triage cardiac panel

cTnI, CK-MB, Mb

CRP, Mb, cTnI hsCRP, cTnI, NTproBNP CRP, NTproBNP CK-MB, cTnI CK-MB, cTnI CK-MB, cTnI CK-MB, cTnI

12

10

Detection method

Sample Ref. size (μL)

Immunosensor/device

CK (creatine kinase), CK-MB (isoenzymes of creatine kinase-MB), CK-MM(isoenzymes of creatine kinase-MM), NTproBNP (N-terminal pro brain natriuretic peptide), @ (Dynamic range of Mb only in fixed CK),* (Dynamic range of CRP in fixed Mb and cTnI), $ (Dynamic range of NTproBNP in fixed high CRP (1–2 μg/mL)). – (Data not provided)


point-of-care diagnostic applications for cardiac health monitoring or risk stratification with less complex fluid transport and target binding process. 3.4. Other sensor metrics

[6] [7] [8]

[9]

The operational stability and repeatability of the electrodes were discussed previously in ref. [39] showing that each device could be reused 5–7 times by employing a device recovery scheme. Besides the device recovery, the device reproducibility is evident from Fig. 3 through error bars from repetitive measurements taken from different electrodes with similar concentrations of capture probes as well as target markers thereby showing the platform-to-platform variations. The storage stability of the modified electrodes (after antibodies are immobilized) in terms of the time between the antibody immobilization step and the final testing of specific or control target detection is an important issue. This requires statistical study of time vs. device performance, which will be considered in future work as we continue to develop this approach.

[11]

4. Conclusion

[15]

A label-free electrochemical multianalyte immunosensor has been developed for the simultaneous detection of three important cardiac markers. A simple method of capture probe immobilization using a few nL of antibody solution is demonstrated, which consists of formation of a hydrophobic surface layer on an individual sensor chip by spin coating and etching of a common positive photoresist using standard optical lithography. The simultaneous detection of the three markers was achieved by using an array of nine electrodes where different electrodes were selectively immobilized with three different capture probes. Simultaneous quantitative detection well within and below the clinically useful concentration range has been demonstrated with no electrode cross talk and high specificity. The multiplexing capability of the electrode array demonstrated here can be extended to other biosensing needs in clinical diagnostics, water quality monitoring and pathogen detection. Acknowledgements JK acknowledges a Presidential Early Career Award. RKG acknowledges the financial support from the J&K Council for Science and Technology, Department of Higher Education, J&K, India and University Grants Commission (UGC), New-Delhi, India. TS was a graduate student intern from the Purdue University, School of Aeronautics and Astronautics and RP was a high school student intern from Lynbrook High School, San Jose, CA.

[10]

[12]

[13]

[14]

[16] [17]

[18]

[19] [20] [21] [22]

[23]

[24]

[25] [26]

[27]

[28]

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jelechem.2016.04.034.

[29]

[30]

References [1] J. Mair, E.A. Dworzak, A. Dienstl, P. Lechleitner, B. Morass, J. Smidt, I. Wagner, C. Wettach, B. Puschendorf, Early detection of acute myocardial infarction by measurement of mass concentration of creatine kinase-MB, Am. J. Cardiol. 68 (1991) 1545–1550. [2] D.A. Morrow, E. Braunwald, Future of biomarkers in acute coronary syndromes moving toward a multimarker strategy, Circulation 108 (2003) 250–252. [3] K.T. Moe, P. Wong, Current trends in diagnostic biomarkers of acute coronary syndrome, Ann. Acad. Med. Singap. 39 (2010) 210–215. [4] G. Liu, J. Wang, J. Kim, M.R. Jan, G.E. Collins, Electrochemical coding for multiplexed immunoassays of proteins, Anal. Chem. 76 (2004) 7126–7130. [5] M.S. Sabatine, D.A. Morrow, J.A. de Lemos, C.M. Gibson, S.A. Murphy, N. Rifai, C. McCabe, E.M. Antman, C.P. Cannon, E. Braunwald, Multimarker approach to risk stratification in non-ST elevation acute coronary syndromes simultaneous

[31]

[32]

[33]

[34] [35]

61

assessment of troponin I, C-reactive protein, and B-type natriuretic peptide, Circulation 105 (2002) 1760–1763. Z. Yang, D.M. Zhou, Cardiac markers and their point-of-care testing for diagnosis of acute myocardial infarction, Clin. Biochem. 39 (2006) 771–780. D. Chan, L.L. Ng, Biomarkers in acute myocardial infarction, BMC Med. 8 (2010) 34. M. Weber, M. Rau, K. Madlener, A. Elsaesser, D. Bankovic, V. Mitrovic, C. Hamm, Diagnostic utility of new immunoassays for the cardiac markers cTnI, myoglobin and CK-MB mass, Clin. Biochem. 38 (2005) 1027–1030. F.S. Apple, A.H. Wu, J. Mair, J. Ravkilde, M. Panteghini, J. Tate, F. Pagani, R.H. Christenson, M. Mockel, O. Danne, A.S. Jaffe, Future biomarkers for detection of ischemia and risk stratification in acute coronary syndrome, Clin. Chem. 51 (2005) 810–824. B.E.F. de Ávila, V.E. Gomez, S. Campuzano, M. Pedrero, J. Salvador, M. Marco, J.M. Pingarrón, Ultrasensitive amperometric magnetoimmunosensor for human Creactive protein quantification in serum, Sensors Actuators B Chem. 188 (2013) 212–220. S. Sridevi, K.S. Vasu, S. Asokan, A.K. Sood, Sensitive detection of C-reactive protein using optical fiber Bragg gratings, Biosens. Bioelectron. (2014). W.M. Fakanya, I.E. Tothill, Detection of the inflammation biomarker C-reactive protein in serum samples: Towards an optimal biosensor formula, Biosensors 4 (2014) 340–357. Y. Zhuo, Y. Zhuo, N. Liao, Y. Chai, R. Yuan, Ultrasensitive electrochemiluminescent detection of cardiac troponin I based on a self-enhanced Ru (II) complex, Talanta (2014). S. Singal, A.K. Srivastava, A.M. Biradar, A. Mulchandani, Pt nanoparticles-chemical vapor deposited graphene composite based immunosensor for the detection of human cardiac troponin I, Sensors Actuators B Chem. 205 (2014) 363–370. V. Sharma, N.K. Puri, R.K. Singh, A.M. Biradar, A. Mulchanadani, Label-free detection of cardiac troponin-I using gold nanoparticles functionalized single-walled carbon nanotubes based chemiresistive biosensor, Appl. Phys. Lett. 103 (2013) 203703. F. Darain, P. Yager, K.L. Gan, S.C. Tjin, On-chip detection of myoglobin based on fluorescence, Biosens. Bioelectron. 24 (2009) 1744–1750. L.E. Agafonova, V.V. Shumyantseva, A.I. Archakov, Quartz crystal microbalance immunoassay for the cardiac myoglobin detection in the plasma samples, BioNanoScience 4 (2014) 46–51. D.H. Kim, S.M. Seo, H.M. Cho, S.J. Hong, D.S. Lim, S.H. Paek, Continuous immunosensing of myoglobin in human serum as potential companion diagnostics technique, Biosens. Bioelectron. 62 (2014) 234–241. P. Skladal, T.A. Kalab, A multichannel immunochemical sensor for determination of 2, 4 dichlorophenoxyacetic acid, Anal. Chim. Acta 316 (1995) 73–78. M.S. Wilson, W. Nie, Electrochemical multianalyte immunoassays using an arraybased sensor, Anal. Chem. 78 (2006) 2507–2513. M.S. Wilson, W. Nie, Multiplex measurement of seven tumor markers using an electrochemical protein chip, Anal. Chem. 78 (2006) 6476–6483. S. Pilehvar, T. Dierckx, R. Blust, T. Breugelmans, K.D. Wael, An electrochemical impedimetric aptasensing platform for sensitive and selective detection of small molecules such as chloramphenicol, Sensors 14 (2014) 12059–12069. G. Lai, F. Yan, H. Ju, Dual signal amplification of glucose oxidase-functionalized nanocomposites as a trace label for ultrasensitive simultaneous multiplexed electrochemical detection of tumor markers, Anal. Chem. 81 (2009) 9730–9736. X. Li, W. Li, Q. Yang, X. Gong, W. Guo, C. Dong, J. Liu, L. Xuan, J. Chang, Rapid and quantitative detection of prostate specific antigen with quantum dot nanobeads based immunochromatography test strip, ACS Appl. Mater. Interfaces 6 (2014) 6406–6414. J.S. Ko, H.C. Yoon, H. Yang, H.B. Pyo, K.H. Chung, S.J. Kim, Y.T.A. Kim, Polymer-based microfluidic device for immunosensing biochips, Lab Chip 3 (2003) 106–113. K. Kojima, A. Hiratsuka, H. Suzuki, K. Yano, K. Ikebukuro, I. Karube, Electrochemical protein chip with arrayed immunosensors with antibodies immobilized in a plasma polymerized film, Anal. Chem. 75 (2003) 1116–1122. M. Wolf, D. Juncker, B. Michel, P. Hunziker, E. Delamarche, Simultaneous detection of C-reactive protein and other cardiac markers in human plasma using micromosaic immunoassays and self-regulating microfluidic networks, Biosens. Bioelectron. 19 (2004) 1193–1202. F. Zhou, M. Lu, W. Wang, Z.P. Bian, J.R. Zhang, J.J. Zhu, Electrochemical immunosensor for simultaneous detection of dual cardiac markers based on a poly (dimethylsiloxane)-gold nanoparticles composite microfluidic chip: a proof of principle, Clin. Chem. 56 (2010) 1701–1707. J. Lee, Y.S. Choi, Y. Lee, H.J. Lee, J.N. Lee, S.K. Kim, K.Y. Han, E.C. Cho, J.C. Park, S.S. Lee, Sensitive and simultaneous detection of cardiac markers in human serum using surface acoustic wave immunosensor, Anal. Chem. 83 (2011) 8629–8635. A.E. Niotis, C. Mastichiadis, P.S. Petrou, I. Christofidis, S.E. Kakabakos, A.S. Kapadai, K. Misiakos, Dual-cardiac marker capillary waveguide fluoroimmunosensor based on tyramide signal amplification, Anal. Bioanal. Chem. 396 (2010) 1187–1196. Y. Arntz, J.D. Seelig, H.P. Lang, J. Zhang, P. Hunziker, J.P. Ramseyer, E. Meyer, M. Hegner, C. Gerber, Label-free protein assay based on a nanomechanical cantilever array, Nanotechnology 14 (2003) 86. G.J. Zhang, K.T.C. Chai, H.Z.H. Luo, J.M. Huang, I.G.K. Tay, A.E. Lim, M. Je, Multiplexed detection of cardiac biomarkers in serum with nanowire arrays using readout ASIC, Biosens. Bioelectron. 35 (2012) 218–223. T. Bryan, X. Luo, P.R. Bueno, J.J. Davis, An optimized electrochemical biosensor for the label-free detection of C-reactive protein in blood, Biosens. Bioelectron. 39 (2013) 94–98. I. Kushner, A.R. Sehgal, Is high-sensitivity C-reactive protein an effective screening test for cardiovascular risk? Arch. Intern. Med. 162 (2002) 867–869. E.M. Antman, M.J. Tanasijevic, B. Thompson, M. Schactman, C.H. McCabe, C.P. Cannon, G.A. Fisher, A.R. Fung, C. Tompson, D. Wybenga, E. Braunwald, Cardiac-

62

[36] [37]

[38] [39]

[40]

[41]

[42]

R.K. Gupta et al. / Journal of Electroanalytical Chemistry 773 (2016) 53–62 specific troponin I levels to predict the risk of mortality in patients with acute coronary syndromes, N. Engl. J. Med. 335 (1996) 1342–1349. P.O. Collinson, F.G. Boa, D.C. Gaze, Measurement of cardiac troponins, Ann. Clin. Biochem. 38 (2001) 423–449. T.C.M. Felismina, S. Sharma, R.A.F. Dutra, J.P.C. Noronha, A.E.G. Cass, M.G.F. Sales, Detection of cardiac biomarker proteins using a disposable based on a molecularly imprinted polymer grafted onto graphite, Microchim. Acta 182 (2015) 975–983. P.U. Arumugam, E. Yu, R. Riviere, M. Meyyappan, Vertically aligned carbon nanofiber electrode arrays for nucleic acid detection, Chem. Phys. Lett. 499 (2010) 241–246. R.K. Gupta, M. Meyyappan, J.E. Koehne, Vertically aligned carbon nanofiber nanoelectrode arrays: electrochemical etching and electrode reusability, RSC Adv. 4 (2014) 22642–22650. A. Periyakaruppan, R.P. Gandhiraman, M. Meyyappan, J.E. Koehne, Label-free detection of cardiac troponin-I using carbon nanofiber based nanoelectrode arrays, Anal. Chem. 85 (2013) 3858–3863. R.K. Gupta, A. Periyakaruppan, M. Meyyappan, J.E. Koehne, Label-free detection of Creactive protein using a carbon nanofiber based biosensor, Biosens. Bioelectron 59 (2014) 112–119. R. Polsky, J.C. Harper, D.R. Wheeler, S.M. Dirk, S.M. Brozik, Electrically addressable diazonium functionalized antibodies for multianalyte electrochemical sensor applications, Biosens. Bioelectron. 23 (2008) 757–764.

[43] J. Park, V. Sunkara, T.H. Kim, H. Hwang, Y.K. Cho, Lab-on-a-disc for fully integrated multiplex immunoassays, Anal. Chem. 84 (2012) 2133–2140. [44] J. Melin, G. Rundström, C. Peterson, J. Bakker, B.D. MacCraith, M. Read, et al., A multiplexed point-of-care assay for C-reactive protein and N-terminal pro-brain natriuretic peptide, Anal. Biochem. 409 (2011) 7–13. [45] T.E. Plowman, J.D. Durstchi, H.K. Wang, J.N. Herron, W.M. Reichert, Multiple-analyte fluoroimmunoassay using an integrated optical waveguide sensor, Anal. Chem. 71 (1999) 4344–4352. [46] M.M. Caulum, B.M. Murphy, L.M. Ramsay, C.S. Henry, Detection of cardiac biomarkers using micellar electrokinetic chromatography and a cleavable tag immunoassay, Anal. Chem. 79 (2007) 5249–5256. [47] M.M. Mion, E. Novello, S. Altinier, S. Rocco, M. Zaninotto, M. Plebani, Analytical and clinical performance of a fully automated cardiac multi-markers strategy based on protein biochip microarray technology, Clin. Biochem. 40 (2007) 1245–1251. [48] F.S. Apple, R.H. Christenson, R. Valdes, A.J. Andriak, A. Berg, Y.J. Duh, S.A. Jortani, N.A. Johnson, B. Koplen, K. Mascotti, A.H.B. Wu, Simultaneous rapid measurement of whole blood myoglobin, creatine kinase MB, and cardiac troponin I by the triage cardiac panel for detection of myocardial infarction, Clin. Chem. 45 (1999) 199–205.