A fluoro-microbead guiding chip for simple and quantifiable immunoassay of cardiac troponin I (cTnI)

Biosensors and Bioelectronics 26 (2011) 3818–3824

Contents lists available at ScienceDirect

Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios

A fluoro-microbead guiding chip for simple and quantifiable immunoassay of cardiac troponin I (cTnI) Seung Yeon Song a , Yong Duk Han a , Kangil Kim b , Sang Sik Yang b , Hyun C. Yoon a,∗ a b

Department of Molecular Science & Technology, Ajou University, San 5, Youngtong Ku, Suwon 443749, Republic of Korea Department of Electrical & Computer Engineering, Ajou University, Suwon 443749, Republic of Korea

a r t i c l e

i n f o

Article history: Received 13 October 2010 Received in revised form 8 January 2011 Accepted 23 February 2011 Available online 1 March 2011 Keywords: Fluoro-microbead guiding chip Cardiac troponin I Fluorescence signal Sandwich immunoassay

a b s t r a c t We have developed a fluoro-microbead guiding chip (FMGC) to perform an optical immunoassay of cardiac troponin I (cTnI). The plasma marker protein cTnI is the currently preferred marker to use for a definitive diagnosis and prognosis of myocardial infarction. The FMGC has four immunoreaction regions on a silicon oxide substrate, with five gold patterns imprinted on each region for multiple simultaneous assays. The FMGC assay clearly distinguished immunospecific binding from nonspecific binding by comparing optical signals from inside and outside of the patterns. To detect cTnI, a sandwich immunoassay was performed using antibody-tagged fluoro-microbeads. The cTnI-specific capture antibody was conjugated to the FMGC surface by reaction with 3-3 -dithiobis-propionic acid N-hydroxysuccinimide ester to create a self-assembling antigen-sensing monolayer (DTSP SAM) on the chip. A sample containing cTnI was applied to the antigen-sensing monolayer and allowed to react. To generate a binding signal, a cTnI detection antibody-linked fluoro-microbead preparation was added. The cTnI concentration in a sample was determined by counting the number of biospecifically bound fluoro-microbeads on the corresponding five patterns on the FMGC. The optical signal showed a linear correlation with cTnI concentrations in plasma samples containing from 3.4 pM to 3.4 nM (0.1–100 ng/ml) cTnI. The sensitivity of cTnI detection could be increased by reducing the non-specific binding of the beads to the antigen-sensing surfaces of the chip. Optical detection and quantification of binding by fluorescence microscopy gave results that correlated well with results from a commercial ELISA for cTnI in human plasma. Based on these findings, we propose that the FMGC-based immunoassay system may be adapted to detect and quantify a variety of clinically important targets in human samples. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Clinically important biomarkers present in blood, urine, and saliva at very low concentrations must be measured accurately to monitor states of health and disease (Giljohann and Mirkin, 2009; Zhang et al., 2007). The enzyme-linked immunosorbent assay (ELISA) provides adequate specificity and sensitivity for most markers (Mayilo et al., 2009). However, targets present at ng/ml concentrations may require a change in methodology. The limit of detection in the conventional ELISA is lies about between 0.3 ng/ml and 1 ng/ml. To increase ELISA sensitivity, a high enzyme-to-antibody ratio may be used to amplify the signal. For example, gold nanoparticles (AuNPs) (Ambrosi et al., 2010), polymer microspheres (Ke et al., 2010), or magnetic microbeads (Liu et al., 2010) may be used as scaffolds to stabilize the enzyme-antibody conjugate. Rissin et al. (2010) developed a single-molecule ELISA to detect prostatespecific antigen in human serum at picomolar concentrations.

∗ Corresponding author. Tel.: +82 31 219 2512; fax: +82 31 219 2394. E-mail address: [email protected] (H.C. Yoon). 0956-5663/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2011.02.036

In this study, we developed a fluoro-microbead guiding chip (FMGC)-based sandwich immunoassay for convenient and accurate biomarker quantification. The FMGC supported four immunoassay regions, each patterned with microchannels. Each immunosensing region contained five gold functional surfaces to perform five identical tests simultaneously and assess reproducibility (Fig. 1A). The FMGC presents several advantages for use in an immunoassay system. First, the number of beads bound in each immunosensing region can be counted directly under a fluorescence microscope. This is both convenient and accurate. Second, the fluoro-microbeads attached on one immunosensing region can be viewed all at once, which allows the analyst to distinguish the specific immuno-binding (beads bound to the patterns) from nonspecific binding (beads bound to the region outside of the patterns). Based on the fluoro-microbead counts inside and outside of the patterns and a simple numerical analysis, a biomarker may be simply and accurately quantified. Cardiovascular diseases claim 17.1 million lives each year world-wide (http://www.who.int). Myocardial infarction (MI), one of the most severe adverse cardiac events, may cause irreversible tissue injury or necrosis in the myocardium. Although the MI

S.Y. Song et al. / Biosensors and Bioelectronics 26 (2011) 3818–3824

3819

Fig. 1. (A) Design of the fluoro-microbead guiding chip (FMGC). (B) Photograph of the FMGC. (C) Schematic diagram of the sandwich immunoassay using antigen/antibody binding (System 1) and avidin/biotin affinity binding (System 2) on the FMGC. The avidin/biotin couple was used to enhance the signal.

diagnosis is based primarily on electrocardiograpy, only 57% of patients with acute myocardial infarction (AMI) exhibit electrocardiographic changes (McDonell et al., 2009; Suprun et al., 2010). A rapid and sensitive method to confirm AMI using specific cardiac markers is therefore desirable. The cardiac forms of troponin T (cTnT), troponin I (cTnI), creatine kinase-MB (CK-MB) and myoglobin provide valuable diagnostic markers for AMI and myocardial injury (Melanson and Tanasijevic, 2005; Piras and Reho, 2005; Wu et al., 2004). Among these markers, cTnI is highly specific to cardiac injury, showing little or no changes in patients with a skeletal muscle disease or trauma; creatine kinase (CK) and the CK-MB isoenzyme are less specific (Apple et al., 2004; Casals et al., 2007). No marker for AMI has yet shown greater specificity than cardiac troponin I (cTnI) (Panteghini et al., 2004). Blood levels of cTnI level in healthy humans are normally lower than 0.1 ng/ml. After the onset of AMI, however, the cTnI rises rapidly, within 3–6 h (Orbulescu et al., 2010; Wei et al., 2003), to levels ranging from 1 to 50 ng/ml, with wide variation. The cut-off cTnI concentration for an AMI diagnosis may be as low as 1 ng/ml, which corresponds to a desired detection limit of 0.1 ng/ml. This places cTnI among the biomarkers present in the lowest range of measurable concentrations (Ko et al., 2007). Following an MI, the cTnI remains elevated for 4–10 days, while most other cardiac markers decline more rapidly (Wu et al., 1999). cTnI may therefore be used to monitor patient status and predict outcome, as well as to diagnose AMI. In previous studies, several immunoassay techniques may be used to monitor cTnI (Bruls et al., 2009; Cho et al., 2009; Kiely et al., 2007; Mair et al., 1996; Todd et al., 2007). In the present study, we devised an accurate and uncomplicated sandwich immunoassay mounted on an FMGC. The detection component was prepared by conjugation of cTnI antibody to fluoromicrobeads. The sensing surface was prepared by conjugation of cTnI capture antibody to DTSP-functionalized patterns on the chip. In the sandwich immunoassay, sample cTnI was allowed to bind

to the chip surface. The fluoro-microbead conjugates were then reacted with the cTnI immobilized. The bound fluoro-microbead conjugates were counted directly using a conventional fluorescence microscope. The cTnI concentration was easily determined by counting the beads immobilized on immunosensing regions of the FMGC. Further, the optical signal from the region outside a pattern (the non-specific binding, or NSB signal) was used to control for the specificity of the immunosensing reaction. It was possible to accurately quantify cTnI through statistical analysis of data from a single test. Here we demonstrate that this method is suitable for cTnI biosensing using human blood samples. 2. Experimental 2.1. Chemicals and apparatus 3-3 -Dithiobis-propionic acid N-hydroxysuccinimide ester (DTSP), dimethyl sulfoxide (DMSO), 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (EDC), avidin, Triton X-100, and sodium bicarbonate were purchased from Sigma–Aldrich, USA; cystamine, poly(amidoamine) generation 4 dendrimer (Dend), 4formylphenylboronic acid (BA), ethanolamine and glutaraldehyde (GA), from Sigma–Aldrich; N-hydroxysulfosuccinimide sodium salt (sulfo-NHS), from Fluka; monoclonal antibodies to human cardiac troponin I (19C7 clone and 625 clone), from BioDesign; human cardiac troponin I (cTnI), skeletal isoform of troponin I (sTnI), cardiac troponin T (cTnT) and cardiac troponin C (cTnC), from Fitzgerald Industries International, USA; FluoSpheres® carboxylate-modified microspheres (excitation wavelength 540 nm and emission wavelength 560 nm, 2% solids, Ø = 0.2 ␮m), from Invitrogen; polyethylene glycol (PEG, MW = 3.4 kDa), from Polyscience Inc.; polydimethylsiloxane (PDMS) and its curing agent, from Dow Corning, USA; and a conventional cTnI ELISA kit, from AbFrontier, Korea.

3820

S.Y. Song et al. / Biosensors and Bioelectronics 26 (2011) 3818–3824

A phosphate-buffered saline solution containing 0.1 M phosphate and 0.15 M NaCl (PBS, pH 7.2), and a 50 mM 2-(Nmorpholino)ethanesulfonic acid (MES, pH 6.0) solution were prepared in doubly distilled and deionized water with a specific resistance greater than 18 M cm. The PBST1 (50 mM PBS, 0.9% NaCl, 0.1% BSA, 0.1% Triton X-100, pH 7.4) and PBST2 (50 mM PBS, 0.9% NaCl, 0.1% Triton X-100, pH 7.4) were used to resuspend and wash the fluoromicrospheres. All other materials and solvents used were of the highest quality available and purchased from regular sources. Fluorescence images were taken with the CCD camera connected to the fluorescence microscope (Leica DM 4000B) and the number of fluoro-microbeads was quantified using NIH image J software. The zeta-potential analyzer (Model ELS-Z1, Otsuka Electronics, Japan) was used to confirm the biochemical modification of the fluoro-microbeads. 2.2. Antibody conjugation to fluoro-microbeads For the troponin immunoassay, cTnI detection antibodies (19C7 clone) were coupled with carboxylated fluoro-microbeads using water soluble 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC). The fluoro-microbead suspension was diluted with MES buffer to 0.5% solid contents. To activate carboxyl groups on the fluoro-microbeads, 25 ␮l of 5 mM EDC and 25 ␮l of 8 mM sulfoNHS were added to diluted microbead solution and gently stirred for 15 min at room temperature. The cTnI detection antibody was then added to a final concentration of 9 ␮g/ml and reacted for 2 h with stirring. To minimize nonspecific electrostatic and hydrophobic binding, the microbead surface was blocked by mixing with 25 ␮l of 1 M l-lysine in 0.008% PEG for 2 h at room temperature (Supplementary information, Fig. S1). The fluoro-microbead conjugates were resuspended with PBST1 and diluted to 0.005%. The result of the blocking process was verified by z-potential measurements. 2.3. Fabrication of the fluoro-microbead guiding chip (FMGC) The FMGC consists of a channel layer and a sensing layer. Fig. S2 (in Supplementary information) describes the fabrication process of the device. The channel layer was formed through sequential casting and curing of the PDMS elastomer. The mold for the PDMS layer was made using a negative photoresist (SU8 2100, Microchem Corporation, MA, USA). To create the PDMS channel layer, PDMS prepolymer (a mixture of base polymer and curing agent, 10:1) was poured onto the patterned SU-8 masters, which are created on a silicon wafer using a lithography process with the negative photoresist. The dimensions of the channels were 7.5 mm × 2 mm × 0.12 mm and the inner volume was 8 ␮l. After curing at 60 ◦ C for 2 h in a vacuum oven, the PDMS replicas were peeled off from the masters and the holes were punched out. To fabricate the sensing layer, the silicon substrate was thermally oxidized and the silicon oxide was patterned using photolithographic technology. After oxide patterning, Ti (50 nm) and Au (200 nm) layers were deposited onto the patterned substrate using a sputter system and the lift-off method. Five identical rectangular patterns (800 ␮m × 180 ␮m) were made below one microchannel site for immunoassay of the target molecule at a single concentration. The five patterns correspond to five simultaneous replicate assays, which provide a check for reproducibility. Fig. 1B shows a photograph of the fabricated FMGC with a set of four identical immunosensing regions. 2.4. Fabrication of sensor surface Fig. 1C illustrates the interface for the cTnI sandwich immunoassay. Here we describe preparation of the sensor surface for the

sandwich immunoassay. Prior to the bottom-up layer formation process, the FMGCs were cleaned under oxygen plasma (5 min, 150 W, O2 50 sccm) for immediate use. Based on our previous work (Song and Yoon, 2009), an amine-reactive self-assembling monolayer was formed by dipping the surfaces into a 5 mM DTSP solution in DMSO for 2 h. The chip surfaces were rinsed with DMSO, ethanol, and DDW in succession. To immobilize the capture antibody, microchannels were created by covering the PDMS grid on the sensing layer of the FMGC. The capture anti-cTnI antibody (625 clone, 8 ␮l of 0.1 mg/ml in PBS) was loaded into a microchannel using a micropipet and reacted for 1 h. After washing, the remaining unreacted groups were blocked by injecting 10 mM ethanolamine in bicarbonate buffer (pH 9.4) for 10 min. Nonspecific binding was thereby greatly reduced. 2.5. Optical immunosensing with the sandwich immunoassay and fluoro-microbead counting To apply the FMGC as a diagnostic device, a sandwich immunoassay was conducted using cTnI antibody-linked fluoromicrobeads as the detection component. Eight-microliter aliquots of cTnI (3.4 pM to 3.4 nM in PBS) were loaded into microchannels of the FMGC and allowed to react with the immobilized capture antibody for 30 min. The unreacted cTnI sample was washed out of the channel by injecting PBS buffer, and a 0.005% solution of cTnI antibody-linked fluoro-microbeads was loaded into channels (Fig. 1C: System 1). After 10 min to allow the binding sandwich to form, the chips were washed with PBST1 and PBST2, and bound fluoro-microbeads were observed and counted using a fluorescence microscope. Fluorescence images were acquired using a green filter matched to the fluoro-microbeads used (absorption maximum at 540 nm and emission maximum of 560 nm). As an alternative method to amplify the antigen–antibody binding signal, we employed the avidin–biotin affinity interaction. After immobilization of the capture antibody and binding of the target antigen cTnI, biotin-conjugated cTnI detection antibody was loaded into the chip and reacted for 30 min. After washing the microchannels with PBS, 0.005% avidin-conjugated fluoro-microbeads were injected (Fig. 1C: System 2). Then the amplified signal from the avidin–biotin reaction was detected under a fluorescence microscope. To count the biospecifically bound fluoro-microbeads on the patterns, one sensing region containing five patterns was photographed using a 100× objective lens. The fluorescence signal was recorded using the NIH Image J software to count the number of microbeads. The specificity of this system was assessed by comparing the numbers of fluoro-microbeads bound inside and outside of the patterns. After microbeads specifically immobilized on the five patterns (800 ␮m × 180 ␮m) were counted, nonspecific binding (NSB) was determined by counting beads in an equivalent area (800 ␮m × 180 ␮m) outside of that region. The data were recorded as the specific binding signal less the NSB signal for each cTnI concentration analyzed. For each cTnI immunoassay performed with an FMGC, mean values and standard deviations for the multiple assays were generated. 2.6. Application to human plasma To simulate clinical application, human plasma samples were spiked with cTnI at 3.4 pM to 3.4 nM and analyzed using the sandwich immunoassay. Human plasma was prepared from whole blood by centrifugation at 3000 rpm, 4 ◦ C for 10 min and the supernatant was collected. After spiking, the plasma samples were diluted 10-fold and assayed in the FMGCs. Results obtained with the sandwich immunoassay were compared with results using conventional ELISA.

S.Y. Song et al. / Biosensors and Bioelectronics 26 (2011) 3818–3824

3821

Fig. 2. Photographs of microbead conjugates binding on different biosensing surfaces. (A) Dend-BA-; (B) Cys-GA-; and (C) DTSP-modified chip surfaces. An optimal sensing surface for the sandwich immunoassay was selected by comparing the binding signals from each surface using 1 ng/ml cTnI with the signal for 0 ng/ml as the control.

3. Results and discussion 3.1. Preparation of fluoro-microbead conjugate For this FMGC-based sandwich immunoassay, we selected a fluoro-microbead as the detection component, because it was sufficiently large (200 nm) for direct visual counting through a fluorescence microscope at 100× magnification. The cTnI concentration in the analyte could thus be simply quantified by counting the biospecifically bound fluoro-microbeads. To detect cTnI in low picomolar (∼ng/ml) concentrations requires a very low background signal. To minimize nonspecific electrostatic and hydrophobic binding of bead conjugates, we blocked the remaining carboxylic groups of the fluoromicrobeads with l-lysine, as confirmed by zeta potential analysis (Supplementary information, Table S1). Following lysine treatment, the z-potential of the bead conjugates was −17.1 mV. This value was higher than the z-potential of the unblocked conjugates (−21.8 mV), showing that the blocked microbead conjugates were less negative than the unblocked conjugates. The size of the fluoromicrobead conjugates also increased, from 325.8 nm to 350.5 nm, after the blocking reaction with l-lysine. This change in the zpotential and size confirmed the blocking of unreacted carboxylate groups. The generally hydrophobic character of polystyrene beads may further contribute to nonspecific binding, notably through attraction to an unmodified hydrophobic gold surface. To minimize this NSB, the microbead conjugates were treated with 0.008% PEG. The hydrated, neutral and flexible chains of this polymer reduce hydrophobic interactions, and PEG is thus widely used to prevent biofouling (Kim et al., 2006; Lee et al., 2006; Yang et al., 1999). PEG bound to the fluoro-microbead conjugates through hydrogen bonding between the carboxylic acid group of the fluoro-microbeads and oxygen from the hydroxyl side group of PEG (Lan et al., 2007). The z-potential of PEG-coated bead conjugates (−16.7 mV) was less negative than the z-potential of the microbead conjugates blocked with l-lysine only (−17.1 mV), due to the charge shielding effect (Supplementary information, Table S1) (Abe et al., 2010). The size of the PEG-coated bead conjugates was 360.5 nm, larger than the microbead conjugates blocked with l-lysine only. These results confirmed that the fluoro-microbeads were successfully blocked with l-lysine and coated with PEG. Fig. S3 (in Supplementary infor-

mation) shows the adsorbed fluoro-microbead conjugates on the glass surface before (a) and after (b) coating with PEG. The numbers of nonspecifically bound beads on the glass surface decreased significantly after PEG treatment, supporting the assumption that PEG would reduce nonspecific hydrophobic binding to the patterned area of the FMGC. Combined treatment of the fluoro-microbeads with l-lysine and PEG effectively increased the biospecificity of the cTnI antigen–antibody reaction on the immunosensing zone of the FMGC. 3.2. Selection of reactive surface for the cTnI immunoassay To create an optimal biosensing interface for this FMGCbased sandwich immunoassay, we compared three agents for surface modification: (i) Dend-BA, (ii) Cys-GA, and (iii) DTSP. The DTSP-modified surface was formed by dipping the gold surface of the FMGC into 5 mM 3-3 -dithiobis-propionic acid Nhydroxysuccinimide ester (DTSP) solution in DMSO. To produce the Dend-BA surface, a DTSP-modified chip was further treated with a boronic-acid functionalized dendrimer (Dend-BA) in methanol. The Cys-GA surface was formed by treating the FMGC surface with 5 mM cystamine (Cys) in DMSO and then with 0.025% glutaraldehyde (GA) in DDW. After completion of all the steps, each chip surface was washed with PBS and a sandwich immunoassay was performed using 0 ng/ml and 1 ng/ml cTnI. An optimal sensing platform was selected through comparison of immunoassay results using FMGCs with each of the three different surface modifications. Based on cTnI concentrations in human blood, the appropriate cut-off value for cTnI in an AMI diagnosis is 1 ng/ml. This requires a biosensing interface with the lowest possible background detection level. Fig. 2 shows the fluorescence microscopic images of Dend-BA, Cys-GA, and DTSP chip surfaces after they were exposed to antibody-tagged fluoro-microbeads. In the analysis of Dend-BA (Fig. 2A) and Cys-GA (Fig. 2B) surfaces, we could not distinguish between the fluorescence signals from 0 ng/ml and 1 ng/ml cTnI because of the large numbers of beads bound to these surfaces. On the DTSP chip surface (Fig. 2C), however, we could easily discriminate between the fluorescence signals of 0 ng/ml and 1 ng/ml cTnI. The Dend-BA and Cys-GA surfaces had significantly more microbead adsorption than the DTSP surface because polyamidoamine dendrimer and cystamine display positively charged amine groups while the fluorescent beads

3822

S.Y. Song et al. / Biosensors and Bioelectronics 26 (2011) 3818–3824

Fig. 3. Fluorescence images from FMGCs after sandwich immunoassay of cTnI at varying concentrations using the antigen/antibody binding system ((A), System 1) and avidin/biotin affinity binding ((B), System 2). The quantification of fluorescence intensity vs. cTnI concentration is also shown (inserted graphs).

display negatively charged carboxylate groups. By direct observation, we concluded that the microbead conjugates bound to the Dend-BA and Cys-GA chip surfaces by electrostatic interaction. However, the DTSP chip surface did not have any fluoro-microbead conjugates bound to 0 ng/ml cTnI because of its slightly negative Nhydroxysuccinimide (NHS) groups. Therefore, the DTSP-modified surface was found to be appropriate for the cTnI sandwich assay using fluoro-microbead conjugates because of the apparently small electrostatic adsorption and sufficient reaction yield to show a significant signal for a cTnI concentration of 1 ng/ml. 3.3. Fluorescence-based optical analysis of sandwich immunoassays Each FMGC supported four immunosensing regions, and five gold patterns were imprinted on each region. As shown in Fig. 3, the same concentration of target cTnI was applied to all of the five patterns in one region for one immunoassay. Four different cTnI concentrations could be assayed simultaneously on one FMGC using the four respective microchannels on the chip. In addition, Fig. 3A shows the images of FMGC with immobilized fluoromicrobead-antibody conjugates for various concentrations of cTnI (System 1). The binding of microbead conjugates onto a pattern on the FMGC surface was observed under a fluorescence microscope. The fluorescence of the immobilized beads corresponded clearly to the patterns printed on FMGC. By counting, we found that the number of fluoro-microbeads increased with the concentration of cTnI applied to the assay. However, the microbead conjugates binding outside of a pattern on the FMGC (i.e., NSB) were very few in number. To quantitatively

compare immunospecific and non-specific binding, the calibration curve was calculated from the number of fluoro-microbeads immobilized on a pattern (black solid line) and outside of a pattern (red solid line) using the NIH Image J software (Fig. 3A). The curve shows a cTnI detection range from 3.4 pM to 3.4 nM (0.1–100 ng/ml; black solid line). However, the number of beads outside a pattern was close to zero (red solid line), indicating that NSB was minimal in this chip surface. This further confirmed the effectiveness of the l-lysine blocking and PEG coating process in reducing NSB. Target detection in this concentration range satisfied the cut-off level for an accurate diagnosis and prognosis of AMI using the FMGC-based sandwich immunoassay system. As a plausible alternative to the FMGC-based system, we tested an avidin–biotin affinity immunoassay system (System 2). The signal from this assay was also recorded by a fluorescence microscope, and distinctive patterns of fluoro-microbeads immobilized on the FMGC were visually observed (Fig. 3B). As in System 1, the brightness of the patterns increased with the concentration of target antigen (cTnI) applied (see calibration curve, Fig. 3B). The fluorescence intensity in a pattern was directly proportional to the cTnI concentration from 3.4 pM to 3.4 nM, and binding to the pattern on the chip showed high contrast to binding in the region outside. From these observations, System 2 appears to induce a higher density of fluoro-microbead binding than System 1. This might be attributed to the higher association constant of the avidin–biotin couple as compared to that of the antigen/antibody pair. Typical affinities for IgG antibodies are 105 –109 M−1 . The avidin/biotin binding constant (Ka = 1 × 1015 M−1 ) is significantly higher than the antigen/antibody binding constant (Yoon et al., 2001). However, in both systems, the calibration curve shows

S.Y. Song et al. / Biosensors and Bioelectronics 26 (2011) 3818–3824

cTnI cTnT cTnC sTnI

50

The number of beads

The number of beads

80

60

40

20

3823

2 folds dilution 5 folds dilution 10 folds dilution

40

30

20

10

0

0 10

100

1000

Concentration of Troponins (pM) Fig. 4. Cross-reactivity with nonspecific antigens (cTnT, cTnC and sTnI) in the FMGCbased immunoassay. The fluorescence signal from each antigen was measured by counting the numbers of fluoro-microbeads bound on the FMGC.

a signal increase at cTnI concentrations from 3.4 pM to 3.4 nM (0.1–100 ng/ml). For the antigen–antibody binding method, the limit of detection (LOD), calculated as three times the standard deviation of the background signal, was 3.4 pM (0.1 ng/ml). The LOD for the avidin–biotin binding method was 1.36 pM (0.04 ng/ml), which was significantly lower than the LOD of System 1. From the above results, it can be concluded that this sandwich immunoassay (using either System 1 or 2) satisfied the clinical cut-off requirement for troponin I (1 ng/ml) in the diagnosis and prognosis of AMI. Further signal amplification was obtained using the avidin–biotin reaction (System 2). Both systems showed coefficients of variation (COV) lower than 9%, which supports the reproducibility and precision of the sandwich assay. From these observations, we suggest that an FMGC-based sandwich immunoassay can provide an uncomplicated and accurate determination of cTnI using a conventional fluorescence microscope in a typical laboratory setting.

10

100

Concentration of cTnI (pM) Fig. 5. Detection of cTnI in diluted plasma samples using a sandwich immunoassay on the FMGC. Calibration curves for fluorescence as a function of cTnI concentration in plasma diluted 2-, 5- and 10-fold. Each data point represents the average and standard deviation for three replicate assays.

3.5. Analysis of cTnI in human plasma

3.4. Cross-reactivity test with related troponins

To test the clinical utility of the present cTnI immunoassay in AMI diagnosis, we measured cTnI concentrations in real human samples. This required us to first test the plasma dilution effect on immunosensor signaling. Fresh human plasma samples were spiked with cTnI at concentrations ranging from 3.4 pM to 3.4 nM, then diluted 2-, 5-, and 10-fold for testing in the System 1. Fig. 5 shows the calibration curve of the fluorescence signal as a function of the cTnI concentration in each human plasma sample dilution. In the 10-fold dilutions, the numbers of immobilized beads increased linearly with the concentration of cTnI, and were disproportionately higher than for the 2-fold and 5-fold dilutions. In the 2-fold and 5-fold dilution sets, the fluorescence detection was inadequate. The LOD at 10-fold plasma dilution was 6.04 pM (0.18 ng/ml), which met the detection requirement for AMI, but was significantly lower than at 2-fold and 5-fold dilutions. We assume that the sensitivity of the sandwich immunoassay increases with the dilution factor because dilution suppresses non-specific binding by real plasma components. Despite the presence of these various components,

The cross-reactivity of immunoassay components reveals the specificity and constitutes a critical parameter of an immunosensing system. To evaluate the specificity of this system, we measured the cross-reactivity of the cTnI antiserum with related antigens, including sTnI, cTnT and cTnC. Significant cross-reaction between the cTnI-specific antibodies with these other antigens would compromise the accuracy of cTnI determination. To test cross-reactivity in this system, we prepared cTnT, cTnC, and sTnI samples at concentrations from 3.4 pM to 3.4 nM and applied them to cTnI capture antibody immobilized at a gold pattern on a FMGC chip. After the reaction, 0.005% cTnI detection antibody-linked fluoro-microbeads were applied according to the method for System 1. The cTnI control test showed linear concentration dependence from 3.4 pM to 3.4 nM (0.1–100 ng/ml) (Fig. 4). However, the three non-reactive troponin proteins gave bead counts of less than 5% of the binding value for the correctly paired cTnI–cTnI antibody reaction. Fluorescence signals from non-reactive troponins also failed to show concentration dependence, again supporting the NSB resistance of the developed surface (Fig. 4). These findings indicate an acceptably low level of cross-reactivity within the FMGC-based cTnI immunoassay system developed in this study, which fulfills an important requirement as a clinically useful method.

Fig. 6. Correlation between cTnI levels measured in spiked real plasma samples using the developed sandwich immunoassay (System 1 and System 2) and measurements using a conventional ELISA. The means of independent triplicate assays are shown and error bars indicate standard deviations. Inset: a FMGC fluorescence image for 340 pM cTnI sample.

3824

S.Y. Song et al. / Biosensors and Bioelectronics 26 (2011) 3818–3824

our assay system gives accurate results with a 10-fold dilution of plasma samples. We then compared immunoassay results for cTnI in spiked human plasma samples using the FMGC (Systems 1 and 2) and an ELISA kit (Fig. 6). The sensitivity and the specificity of the FMGC were evaluated using the commercial ELISA as a reference standard. Results using System 1 and using the clinical ELISA for the cTnI spiked plasma samples showed a good correlation. From the curve in Fig. 6, The LOD of System 1 determined by fluorescence microscopy as described above was 4.6 pM. The LOD in the ELISA was 12.9 pM under the same reaction conditions. This reveals that sandwich immunoassay developed on the FMGC achieved a 3-fold improvement in sensitivity as compared to the conventional ELISA. Use of the avidin–biotin reaction for signal enhancement (System 2) achieved an even lower LOD (1.4 pM), which may reflect the very high avidin–biotin association constant. These results overall demonstrate significantly enhanced performance by the FMGCbased sandwich immunoassay as compared to conventional ELISA, and support its use in clinical diagnosis. 4. Conclusions In this study, we used a fluoro-microbead guiding chip to develop a new bioimmunosensing device to measure cTnI in human plasma. This FMGC-based sandwich immunoassay system achieved high accuracy in part by treatment of chip surfaces to minimize non-specific binding and through quantitative analysis of fluorescence signals from the immunosensing platform. The FMGC-based system also showed greater sensitivity (a lower LOD) than the conventional ELISA method. A prominent advantage of this system is the ease of signal quantification obtained through visual counting of fluoro-microbeads immobilized on the FMGC. Each immunoassay region was integrated with a microfluidic channel, and multiple regions were constructed on each FMGC to allow for multiple simultaneous assays and quantification of a single protein over a broad concentration range. Our results revealed the potential utility of FMGC in designing a multiplex biosensing clinical diagnostic tool. We continue to develop the design and fabrication of the microfluidics components of this system to minimize procedural steps such as the dilution of real samples. Acknowledgements This research was supported by the Korean Health Technology R&D Project, Ministry for Health, Welfare & Family Affairs (A091120), the National Research Foundation grant (2010-0021666), and the Priority Research Centers Program (20100028294) through the National Research Foundation of Korea.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.bios.2011.02.036. References Abe, M., Akbarzaderaleh, P., Hamachi, M., Yoshimoto, N., Yamamoto, S., 2010. Biotechnol. J. 5 (5), 477–483. Ambrosi, A., Airò, F., Merkoc¸i, A., 2010. Anal. Chem. 82 (3), 1151–1156. Apple, F.S., Murakami, M.A.M., Christenson, R.H., Campbell, J.L., Miller, C.J., Hock, K.G., Scott, M.G., 2004. Clin. Chim. Acta 345 (1–2), 123–127. Bruls, D.M., Evers, T.H., Kahlman, J.A.H., van Lankvelt, P.J.W., Ovsyanko, M., Pelssers, E.G.M., Schleipen, J.J.H.B., de Theije, F.K., Verschuren, C.A., van der Wijk, T., van Zon, J.B.A., Dittmer, W.U., Immink, A.H.J., Nieuwenhuis, J.H., Prins, M.W.J., 2009. Lab Chip 9 (24), 3504–3510. Casals, G., Filella, X., Bedini, J.L., 2007. Clin. Biochem. 40 (18), 1406–1413. Cho, I.-H., Paek, E.-H., Kim, Y.-K., Kim, J.-H., Paek, S.-H., 2009. Anal. Chim. Acta 632 (2), 247–255. Giljohann, D.A., Mirkin, C.A., 2009. Nature 462 (26), 461–464. Ke, R., Yang, W., Xia, X., Xu, Y., Li, Q., 2010. Anal. Biochem. 406 (1), 8–13. Kiely, J., Hawkins, P., Wraith, P., Luxton, R., 2007. IET Sci. Meas. Technol. 1 (5), 270–275. Kim, P., Jeong, H.E., Khademhosseini, A., Suh, K.Y., 2006. Lab Chip 6 (11), 1432–1437. Ko, S.H., Kim, B.J., Jo, S.-S., Oh, S.Y., Park, J.-K., 2007. Biosens. Bioelectron. 23 (1), 51–59. Lan, Z., Wu, J., Lin, J., Huang, M., Yin, S., Sato, T., 2007. Electrochim. Acta 52 (24), 6673–6678. Lee, L.M., Heimark, R.L., Baygents, J.C., Zohar, Y., 2006. Nanotechnology 17 (4), S29–S33. Liu, Z., Zhang, L., Yang, H., Zhu, Y., Jin, W., Song, Q., Yang, X., 2010. Anal. Biochem. 404 (2), 127–134. Mair, J., Genser, N., Morandell, D., Maier, J., Mair, P., Lechleitner, P., Calzolari, C., Larue, C., Ambach, E., Dienstl, F., Pau, B., Puschendorf, B., 1996. Clin. Chim. Acta 245 (1), 19–38. Mayilo, S., Kloster, M.A., Wunderlich, M., Lutich, A., Klar, T.A., Nichtl, A., Kürzinger, K., Stefani, F.D., Feldmann, J., 2009. Nano Lett. 9 (12), 4558–4563. McDonell, B., Hearty, S., Leonard, P., O’Kennedy, R., 2009. Clin. Biochem. 42 (7–8), 549–561. Melanson, S.F., Tanasijevic, M.J., 2005. Cardiovsc. Pathol. 14 (3), 156–161. Orbulescu, J., Micic, M., Ensor, M., Trajkovic, S., Daunert, S., Leblanc, R.M., 2010. Langmuir 26 (5), 3268–3274. Panteghini, M., Pagani, F., Yeo, K-T.J., Apple, F.S., Christenson, R.H., Dati, F., Mair, J., Ravkilde, J., Wu, A.H.B., 2004. Clin. Chem. 50 (2), 327–332. Piras, L., Reho, S., 2005. Sens. Actuators B 111–112, 450–454. Rissin, D.M., Kan, C.W., Campbell, T.G., Howes, S.C., Fournier, D.R., Song, L., Piech, T., Patel, P.P., Chang, L., Rivnak, A.J., Ferrell, E.P., Randall, J.D., Provuncher, G.K., Walt, D.R., Duffy, D.C., 2010. Nat. Biotechnol. 28 (6), 595–600. Song, S.Y., Yoon, H.C., 2009. Sens. Actuators B 140 (1), 233–239. Suprun, E., Bulko, T., Lisitsa, A., Gnedenko, O., Ivanov, A., Shumyantseva, V., Archakov, A., 2010. Biosens. Bioelectron. 25 (7), 1694–1698. Todd, J., Freese, B., Lu, A., Held, D., Morey, J., Livingston, R., Goix, P., 2007. Clin. Chem. 53 (11), 1990–1995. Wei, J., Mu, Y., Song, D., Fang, X., Liu, X., Bu, L., Zhang, H., Zhang, G., Ding, J., Wang, W., Jin, Q., Luo, G., 2003. Anal. Biochem. 321 (2), 209–216. Wu, A.H.B., Apple, F.S., Gibler, W.B., Jesse, R.L., Warshaw, M.A.M., Valdes Jr., R., 1999. Clin. Chem. 45 (7), 1104–1121. Wu, A.H.B., Smith, A., Christenson, R.H., Murakami, M.M., Apple, F.S., 2004. Clin. Chim. Acta 346 (2), 211–219. Yang, Z., Galloway, J.A., Yu, H., 1999. Langmuir 15 (24), 8405–8411. Yoon, H.C., Hong, M.-Y., Kim, H.-S., 2001. Langmuir 17 (4), 1234–1239. Zhang, H., Zhao, Q., Li, X.-F., Le, X.C., 2007. Analyst 132 (8), 724–737.