Semi-continuous, real-time monitoring of protein biomarker using a recyclable surface plasmon resonance sensor

Author’s Accepted Manuscript Semi-continuous, Real-time Monitoring of Protein Biomarker using a Recyclable Surface Plasmon Resonance Sensor Dong-Hyung Kim, Il-Hoon Cho, Ji-Na Park, SungHo Paek, Hyun-Mo Cho, Se-Hwan Paek www.elsevier.com/locate/bios

PII: DOI: Reference:

S0956-5663(16)30792-8 http://dx.doi.org/10.1016/j.bios.2016.08.035 BIOS9032

To appear in: Biosensors and Bioelectronic Received date: 15 June 2016 Revised date: 4 August 2016 Accepted date: 12 August 2016 Cite this article as: Dong-Hyung Kim, Il-Hoon Cho, Ji-Na Park, Sung-Ho Paek, Hyun-Mo Cho and Se-Hwan Paek, Semi-continuous, Real-time Monitoring of Protein Biomarker using a Recyclable Surface Plasmon Resonance Sensor, Biosensors and Bioelectronic, http://dx.doi.org/10.1016/j.bios.2016.08.035 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Semi-continuous, Real-time Monitoring of Protein Biomarker using a Recyclable Surface Plasmon Resonance Sensor Dong-Hyung Kima,1, Il-Hoon Chob,1, Ji-Na Parka, Sung-Ho Paeka, Hyun-Mo Choc, Se-Hwan Paeka,d* a

Department of Bio-Microsystem Technology, Korea University, 145Anam-ro, SeongbukGu, Seoul 02841, Korea

b

Department of Biomedical Laboratory Science, College of Health Science, Eulji University, Seongnam13135, Korea;

c

Korea Research Institute of Standards and Science, P.O. Box 102, Yuseong, Taejon 34113, Korea

d *

Department of Biotechnology, Korea University, 2511 Sejong-ro, Sejong30019, Korea Address for correspondence: Se-Hwan Paek, Professor. 540 R&D Center (Green Campus).

Korea University. 145 Anam-ro, Seongbuk-gu. Seoul 02841. Republic of Korea. Tel: 82-23290-3438. FAX: 82-2-927-2797. E-mail: [email protected]

Abstract Although label-free immunosensors based on, for example, surface plasmon resonance (SPR) provide advantages of real-time monitoring of the analyte concentration, its application to routine clinical analysis in a semi-continuous manner is problematic because of the high cost of the sensor chip. The sensor chip is in most cases regenerated by employing an acidic pH. However, this causes gradual deterioration of the activity of the capture antibody immobilized on the sensor surface. To use sensor chips repeatedly, we investigated a novel surface modification method that enables regeneration of the sensor surface under mild conditions. We introduced a monoclonal antibody (anti-CBP Ab) that detects the conformational change in calcium binding protein (CBP) upon Ca2+ binding (>1 mM). To construct a regenerable SPR-based immunosensor, anti-CBP Ab was first immobilized on the sensor surface, and CBP conjugated to the capture antibody (specific for creatine kinase-MB 1

These authors equally contributed to this work. 1

isoform (CK-MB); CBP-CAb) then bound in the presence of Ca2+. A serum sample was mixed with the detection antibody to CK-MB, which generated an SPR signal proportional to the analyte concentration. After each analysis, the sensor surface was regenerated using medium (pH 7) without Ca2+, and then adding fresh CBP-CAb in the presence of Ca2+ for the subsequent analysis. Analysis of multiple samples using the same sensor was reproducible at a rate >98.7%. The dose-response curve was linear for 1.75 to 500.75 ng/mL CK-MB, with an acceptable coefficient of variation of <8.8%. The performance of the immunosensor showed a strong correlation with that of the Pathfast reference system (R2 >96%), and exhibited analytical stability for 1 month. To our knowledge, this is the first report of a renewal of a sensor surface with fresh antibody after each analysis, providing high consistency in the assay during a long-term use (e.g., a month at least).

Keywords Label-free sensor regeneration, Conformation change-sensitive antibody, Semi-continuous immunosensor, Acute myocardial infarction markers, Companion diagnostics

1. Introduction

Real-time monitoring of biomarkers in body fluids (e.g., interstitial fluid, blood, and urine) will enable realization of companion diagnostics to improve treatment (Best et al. 2015). Several currently available devices measure physical signals, such as cardiac impulse and pulsation in real-time (Patel et al. 2012). Continuous biochemical monitoring is applied typically to glucose for management of diabetes (Wang 2008). The device is miniaturized into a wearable format to continuously monitor glucose for several days without interruption. However, application of such technology to other biomarkers, such as hormones and proteins, is in its infancy. These biomarkers are relatively large and structurally complex, and can at present be assayed only using immunological methods (Koivunen and Krogsrud 2006). Comparing to enzyme assay, the method may demand sophisticated technologies that circumvent current problems, for examples, labeling-based signal generation, the sample matrix effect, slow binding kinetics, and trace signal detection (Kim et al. 2014). For immunoassay like in continuous glucose monitoring, a label-free sensor can be used by investigating a technique providing fresh capture antibody every each analysis as shown in 2

this study (refer to below for details). Acute myocardial infarction (AMI) must be diagnosed at the incipient stage (e.g., <2−4 h after onset) and treated appropriately. To this end, although disposable sensors for biomarkers (e.g., creatinine kinase-MB isoform; CK-MB) are used, frequent sample drawings may not allow timely detection (Al-Hadi and Fox 2009; Gibler et al. 1995). We reported previously an alternative method of diagnosis involving monitoring of the serum myoglobin concentration using a label-free sensor (e.g., surface plasmon resonance (SPR)) in a real-time, continuous manner (Hsieh et al. 2004). The monitoringon SPR sensor was able to be demonstrated by controlling sample matrix effect and using a marker-specific antibody revealing rapid reaction kinetics. The method performed together with ECG has an increased diagnostic accuracy during the early period (<2-4 h after onset) of 88.2%, compared to 50−60% for ECG alone (Kim et al. 2010). Addition of a specific marker such as CK-MB can further enhance the analytical accuracy to 80.0% at <0−2 h after AMI onset, 92% at <2−4 h, and 100% at <4−6 h (Al-Hadi and Fox 2009; Jagannadharao et al. 2010; Kontos et al. 1997; Lindahl et al. 1995). However, CK-MB is present in trace amounts (clinical range: 1–500 ng/mL), which hampers continuous measurement using a rapidly reactive antibody due to its low affinity (Schiel and Hage 2009).

Although a label-free immunosensor based on, for example, SPR, can facilitate real-time monitoring of analyte concentration (Kim et al. 2014), its application to routine clinical analysis using high-affinity antibodies may be problematic. High-affinity antibodies usually have a very low dissociation rate, and so an additional regeneration step would be required after each analysis to repeatedly measure the analyte using the same sensor. Sensor chips are regenerated usually employing a medium of acidic pH to dissociate the antigen from the capture antibody (Goode et al. 2015). However, this leads to gradual deterioration of antibody activity. The dissociated antigen can remain near the surface and may reassociate with the antibody, causing incomplete regeneration (Chiu et al. 1986). There are other regeneration methods enabling repetitive assays on the same SPR sensor, they could fail to show both of firm binding of the capture antibody and mildness in the exchange with a new one at the same time. In general, repeated analyses using the same sensor, particularly employing a high-affinity antibody, result in low reproducibility.

3

We propose in this study a novel technique for real-time monitoring of CK-MB in a semicontinuous manner through repetitive immunoassays using the same sensor. This could be facilitated by exchanging the used capture antibody for a fresh antibody for analysis of the next sample. As we previously produced an immunological binder sensitive to the conformational change of a linker molecule that can be tethered to the capture antibody (Paek et al. 2014), the hypothesis was approached by immobilizing it on the SPR sensor surface. The linker was calcium-binding protein (CBP), which became antigenic due to the conformational change upon binding of an inducer (i.e., Ca2+). If the capture antibody is subsequently immobilized, retaining high affinity, by tethering it to the linker on the labelfree sensor surface, the target analyte can be analyzed with high sensitivity. The tethered linker and other substances may be subsequently removed by eliminating the inducer from the medium. For the next analysis, the linker conjugated to the fresh capture antibody can be added in the presence of the inducer to prepare a new immunosorbent. We attempted to demonstrate here analyses of CK-MB in multiple samples using a single SPR sensor.

2. Materials and Methods 2.1. Materials A monoclonal antibody (clone 3-6F) specific to Ca2+-bound configuration of CBP (anti-CBP antibody) was produced in our laboratory (Paek et al. 2014). CBP designated mutated glucose-galactose binding protein (GGBP) was produced in Escherichia coli and purified by our research group (Paek et al. 2014). Calcium chloride dihydrate was obtained from Daejung (Siheung, Korea). Human CK-MB (from human heart tissue) was purchased from Biospacific (Emeryville, CA, USA), and monoclonal antibodies (clone 7501 and 7502 specific to CKMB) were supplied by MedixBiochemica (Kauniainen, Finland). Human serum (from human male AB plasma), sodium dodecyl sulfate (SDS), casein (from bovine milk), and bovine serum albumin (BSA) were obtained from Sigma-Aldrich (St. Louis, MO). Succinimidyl 4(N-maleimidomethyl)

cyclohexane-1-carboxylate

(SMCC),

N-succinimidyl

3-(2-

pyridyldithio)-propionate (SPDP), EZ-Link™ succinimidyl-6-(biotinamido)-6-hexanamido hexanoate (NHS-LC-LC-Biotin), dithiothreitol (DTT), and streptavidin (SA) were supplied

4

by Thermo Fisher Scientific (Rockford, IL, USA). The SPR sensor chip (BIAcore CM5), amine coupling kit containing 100 mM N-hydroxysuccinimide (NHS), 400 mM N-ethyl-N’(dimethylaminopropyl) carbodiimide (EDC), 1 M ethanolamine hydrochloride, 40% glycerol, surfactant polysorbate 20 (P20), and 10 mM HEPES containing 150 mM sodium chloride, 3 mM EDTA, and 0.005% surfactant P20 (HBS-EP) were obtained from GE Healthcare (Uppsala, Sweden). The Octet Red streptavidin (Octet SA) sensor tip was purchased from ForteBio (San Francisco, CA, USA). Other reagents used in this study were of analytical grade.

2.2. Generation of a regenerable SPR sensor surface Preparation of the capture binder. To prepare the capture binder, CBP was first coupled to streptavidin (CBP-SA) and used for immobilization of a biotinylated antibody specific to CK-MB (b-CAb). CBP (200 µg) in 10mM phosphate, pH 7.4 (PB) was activated with SMCC in 20 M excess for 1 h. Streptavidin (1 mg) in PB was reacted with SPDP in 5 M excess for 1 h, and then chemically reduced to produce sulfhydryl groups by treating with 10 mM DTT and 5 mM EDTA at 37C for 1 h. The activated components were separately fractionated on a Sephadex G-15 gel filtration column (10 mL volume) by removing excess reagents. Next, the two activated proteins, maleimide-CBP and thiolated-SA, were mixed at a molar ratio of 1:5 and reacted at room temperature (RT) for 2 h. For biotinylation of the capture antibody, anti-CK-MB (1 mg) in 10 mM phosphate, pH 7.4, containing 140 mM NaCl (PBS) was reacted with NHS-LCLC-biotin (20 molar excess) dissolved in dimethyl sulfoxide at RT for 2 h. Excess reagents were removed in a Sephadex G-15 gel filtration column (10 mL volume).

Optimization of sandwich complex formation. For preliminary tests, the biotinylated antibody (clone 7501; 1 µg/mL), diluted in buffer (10 mM Tris, pH 7.4, and 140 mM NaCl containing 0.5% casein and 0.1% Tween; casein-Tris), was bound to Octet SA sensor tip surfaces at 30ºC, followed by incubation for 20 min. The residual surfaces were blocked with casein-Tris for 15 min. To detect the analyte directly on the sensor, the sensor tip was dipped into a sample containing CK-MB (0.01–10 µg/mL) in casein-Tris for 15 min. For sandwich immunoassay, the detection antibody (clone 7502; 1 µg/mL) was pre-reacted with CK-MB for 5 min, added to the sensor, and incubated for 15 5

min. The same protocol was also applied to the control sensor, on which non-specific protein (BSA, 1 µg/mL) was immobilized. The specific binding signal was measured by subtracting the control signal from the test signal using the analysis software (Data Analysis 7.0) provided by the manufacturer. This procedure was repeated, with the exception of use of clone 7502 as the capture antibody and 7501 as the detection antibody.

Preparation of regenerable SPR sensor surface. A monoclonal antibody (clone 3-6F) specific to CBP was immobilized on the SPR CM5 sensor chip of the Biacore 3000 SPR system (GE Healthcare) to prepare the immunosorbent. The chip was first activated using NHS (0.4 M) and EDC (0.1 M), according to the manufacturer’s instructions, and installed within the SPR sensor system. The anti-CBP antibody (50 µg/mL) in 10 mM sodium acetate (pH 4.5) was injected into the chip at a rate of 5 µL/min through flow channel #2 of the system for 10 min. Residual reactive groups were inactivated by sequentially injecting 1 M ethanolamine hydrochloride solution, pH 8.5, for 10 min. The chip surface was cleaned with 10 mM HEPES containing 150 mM NaCl, 3 mM EDTA, and 0.005% surfactant P20 (HBS-EP). The control site (channel #1) was prepared using the same procedure, with the exception of immobilization of a non-specific protein (BSA, 50 µg/mL). The prepared sensor chip was stored in casein-Tris when not in use.

2.3. Construction of an SPR immunosensor for semi-continuous monitoring Basic analytical procedure. The prepared chip was mounted within the Biacore 3000 SPR sensor system and operated according to the protocol provided by the manufacturer (GE Healthcare). The chip was initially equilibrated at 25ºC with a flow of test medium into channel #2 via #1 at a flow rate of 5 µL/min (‘preparation’ mode). To immobilize the capture binder (clone 7502 biotinylated; b-CAb) on the sensor surface, CBP-SA (1 µg/mL) and b-CAb (5 µg/mL) separately diluted in casein-Tris containing Ca2+ (>1 mM) were sequentially injected into the sensor chip at a rate of 5 µL/min. A human serum sample containing CK-MB (75 µL) was mixed with the detection antibody (clone 7501) and subsequently combined with an equal volume of pretreatment medium prepared by adding 0.2% SDS, 0.2% P20, and 5 mM CaCl2 to casein-Tris (casein-SDS-P20-Ca2+). The pre-treated sample was supplied to the sensor for 15 min, and the same medium lacking analyte was subsequently added for 15 min (‘sample injection’ 6

mode). The sensor was finally regenerated by injecting casein-Tris containing 0.3% SDS (casein-SDS; 50 µL) after sample analysis. The specific SPR signal was measured by subtracting the non-specific background signal measured at the control site (channel #1) from that at the test site (channel #2). Baseline drift was corrected using a linear equation passing through the two initial and final points, and a corrected signal vs. time plot was generated by compensating for the slope (Kim et al. 2014). The kinetic reaction curve was plotted using the BIAevaluation 2.0 software provided by the manufacturer. To analyze the next sample on the same chip, the capture binder was re-applied to the sensor under the condition containing calcium ion. An identical analytical procedure was used in subsequent experiments unless stated otherwise.

Optimization of analytical conditions. After initial equilibration of the SPR immunosensor chip, CBP-SA and b-CAb diluted in casein-Tris wereseparately diluted in medium containing 0–20 mM CaCl2, and sequentially injected into the sensor system for optimization. The unbound reagents were washed away by applying casein-Tris containing the corresponding calcium ion concentration, and the amount of capture binder was quantitatively analyzed to determine immobilization yield according to calcium concentration. The sensor was subsequently regenerated by applying 50 µL of casein-SDS as described above. Next, the capture binder was diluted to different concentrations (1, i.e., 5 µg/mL, and 5) in casein-Tris containing 5 mM CaCl2 (casein-TrisCa2+), and then used for comparison following the identical protocol as described.

2.4. Assessment of semi-continuous sensing performance Dose response. Human serum was used as sample medium and its CK-MB concentration (0.75 ng/mL) was determined using a commercial device, Pathfast (Mitsubishi Chemical, Tokyo, Japan). Each serum sample spiked with CK-MB (0.25–500 ng/mL) was pre-treated with an equal volume of casein-SDS-P20-Ca2+. After immobilization of the capture binder under optimal conditions, the sample was analyzed by injecting it into the immunosensor for 15 min, followed by washing with serum mixed with casein-SDS-P20-Ca2+ at the same volume ratio for 15 min. The sensor response was assessed by determining the average SPR signal at a quasi-steady state between 10 and 13 min during the washing step. 7

Multiple samples were analyzed using the same procedure in a semi-continuous manner and the signal was plotted against CK-MB concentration. The standard curve was linearized via log-logit transformation (Cho et al. 2009), and a regression line was drawn using the leastsquares method. Logit was calculated as log{B/(Bs -B)}, where B and Bs were the signal values for the standard sample and at saturation, respectively. Although the saturated SPR signal may be measured by employing a sample containing excess CK-MB, such trial was restricted by the analyte concentration available in stock. Therefore, the value (integrated intensity ~ 10,000 RU) was estimated by trial-and-error (Seo et al. 2016) until maximum correlation coefficient (R2) was obtained in drawing a log-logit plot for linearization.

Correlation with a reference system. The SPR-based semi-continuous immunosensor was analyzed using blinded samples, and the results were compared with those of the Pathfast reference system. Serum containing basal CK-MB (0.75 ng/mL) was used as the sample matrix. A total of 24 blinded samples containing 1–100 ng/mL CK-MB were prepared. The samples were separately pre-treated withanequal volume of casein-SDS-P20-Ca2+, and analyzed using the immunosensor in a semi-continuous manner. Each signal was converted to the analyte concentration using the linearized standard curve. The same samples were also analyzed using the reference system according to the procedure recommended by the manufacturer. The two sets of quantitative data were plotted on each axis, and the degree of correlation was determined via linear regression analysis.

Reproducibility test. The same immunosensor was used repeatedly to assay CK-MB in different samples at seven pre-designated dates (i.e., 1, 3, 5, 7, 10, 20 and 30 days) for 1 month after initial use. The samples were prepared by spiking human serum with 20, 40, 100 and 160 ng/mL CK-MB. Each sample was pre-treated with an equal volume of casein-SDS-P20-Ca2+ and analyzed sequentially as described above. After first use of the immunosensor, the experiments were repeated 1, 3, 5, 7, 10, 20 and 30 days later. The sensor was stored in casein-Tris at 4ºC when not in use. The signals were averaged and the coefficient of variation (CV) was determined to evaluate the reproducibility of the immunosensor.

8

3. Results and Discussion 3.1. Smart Sensor Surface Renewal Technique The capture antibody switching technique was used to regenerate the intact sensor surface and maintain the performance of the immunoassay. We produced a monoclonal antibody (anti-CBP Ab) sensitive to the conformational change of CBP upon Ca2+ binding (>1 mM) (Paek et al. 2014). The presence of ionic calcium results in association of the two components ('switched-on'; Figure 1, A). The antigen-antibody complex is dissociated upon depletion of the switching ion from the medium ('switched-off'). As complex formation is reversible depending on the Ca2+ concentration, the ion acts as 'molecular switch' of the binding reaction.

We investigated a novel surface modification method enabling regeneration of the sensor surface under mild conditions (Figure 1, B). A regenerable SPR-based immunosensor was constructed by first immobilizing an anti-CBP Ab on the sensor surface (step I), followed by binding of CBP conjugated to the capture antibody (specific to CK-MB; CBP-CAb) in the presence of Ca2+ (step II). A serum sample mixed with an antibody to CK-MB was then added to generate an SPR signal proportional to the analyte concentration (step III). After analysis, the sensor surface was regenerated by addition of medium (pH 7) lacking Ca2+ (step IV). The immunosorbent for the next analysis was then prepared following identical steps. As the regeneration conditions were mild, the new assay was able to be reproduced at high rate (see below for details). Thus the sensor surface was renewed with fresh antibody after each analysis. Thus, semi-continuous monitoring can be performed using the same sensor in a cyclic manner. We have measured the binding affinity of the anti-CBP antibody, i.e., 1.11 x 10-9 M (kon = 1.31 x 105 M-1s-1 and koff = 1.44 x 10-4 s-1) (Park et al. 2016), immobilized on the sensor surface in the presence of calcium ion. Such high affinity and slow dissociation kinetics were suitable to stably hold CBP-CAb during immunoassay. The washing effect on the complex formation between CBP and anti-CBP antibody was minimal as far as ionic calcium was present in the medium (Park et al. 2016). However, upon calcium depletion, the CBP conjugate was almost completely dissociated from the anti-CBP antibody on the sensor in 15 9

min and, therefore, the residual binding was negligible.

The time interval between sample measurements varies according to the biomarker in question. For CK-MB, as the level in blood changes only slowly, measurements with an interval of hours are appropriate (Gibler et al. 1995). Thus, repeated measurements can be carried out using semi-continuous real-time monitoring.

The utility of the surface renewal technique was demonstrated by constructing an analytical system and comparing its performance with that of a reference system. The analytical conditions for a biomarker of AMI, CK-MB, were then optimized. Performance was characterized by analyzing multiple samples on a single sensor in a semi-continuous manner. Analytical utility was assessed by correlating the results with those of a reference system, Pathfast (Mitsubishi Chemical; Tokyo, Japan).

3.2. Optimization of Analytical Conditions for CK-MB We first determined the optimal arrangement of antibodies to form the sandwich complex with CK-MB in a solid-phase immunoassay. Two antibodies specific to CK-MB, clones 7501 and 7502, were used. Clone 7501 was biotinylated and immobilized on the surface of a streptavidin-coated label-free Octet sensor (ForteBio; San Francisco, CA, USA). This was used as the capture antibody to analyze standard CK-MB samples in a defined medium (e.g., 0.5% Casein-Tris containing 0.1% Tween). The procedure was repeated using samples combined with clone 7502. Although the dose responses of the sensor were proportional to the analyte concentration, the detection sensitivities were low (>0.1 µg/mL; Figure 2, A). Next, clone 7502, which was also biotinylated, was used as the capture antibody for analyses of standard samples containing CK-MB only (2, B). The sensor in this configuration showed about 10-fold higher detection limit of the marker than that of the sensor of the former arrangement using clone 7501 as the capture antibody. Use of clone 7501 as the detection antibody further enhanced the performance. This was likely due to increased accessibility to the capture antibody using clone 7502.

The distinct results shown in Figures 2A and 2B were reproducible when the same experiments were repeated using the two monoclonal antibodies specific to CK-MB. The lack 10

of difference regardless of the use of detection antibody, shown in Figure 2A, could be caused by steric hindrance. This implements that the initial binding of one antibody to antigen masks the antigenic site of the same antigen molecule for the subsequent binding of the other antibody. The identical effect may be often met in large molecular interactions that sandwich complex formation in the solid phase can be largely variable according to arrangement of the two antibodies in spatial configuration (Gao et al. 2014; Zhang and Meyerhoff 2006). Such phenomenon can be significant when the antigen-antibody bindings occur in the solid phase due to, so called, solid matrix effect restricting the molecular arrangement in stacking manner (Gao et al. 2014).

Immobilization of the capture antibody on the regenerable SPR sensor (step II in Figure 1, B) was then optimized in terms of maximum surface binding and minimal residue after regeneration. Under the calcium switch-on condition, CBP-streptavidin conjugate (CBP-SA; step II-1 of Figure 3, A) and biotinylated capture antibody (b-CAb; step II-2) were sequentially bound via the biotin-streptavidin linkage. After each binding, excess reagent was washed out and immobilized capture antibody was quantified by detecting the SPR signal. The capture antibody level was controlled by varying two primary parameters, i.e., Ca2+ and capture antibody concentrations. The signal increased in proportion with the Ca2+ concentration from 2 to 20 mM (Figure 3, A). However, since the baseline level after regeneration was 8.2-fold increased at high Ca2+ concentrations compared to the optimal concentration, 5 mM Ca2+ was used. Under the high calcium concentrations, the capture antibody molecules may be crowded on the CBP layer of the sensor, enabling to form a rigid layer by molecular interaction. This caused incomplete dissociation of the antibody from the surface and subsequently the increased baseline. Under the optimal condition, the baseline was recovered >98.7% after regeneration. The capture antibody concentration was then increased 5 times, showing about 3.6-fold enhancement of the binding without deteriorating the baseline recovery (3, B). It is notable that the SPR sensor typically showed a drift of 0.48 RU/min, < 1.0 RU/min as guideline for maintaining the sensor stability without additional cleaning (Tanious et al. 2008; Wang et al. 2015). As the baseline drift was stable, it was then corrected by compensating for the slope (Kim et al. 2014). The optimal conditions were used in subsequent experiments.

3.3. Performance Characterization of Semi-continuous Biosensing for CK-MB 11

The sensor surface renewal technique shown in Figure 3, A was applied to analysis of CKMB in human serum in a semi-continuous fashion using the same SPR sensor (Figure 4, upper panel). The SPR sensor with an anti-CBP antibody immobilized on the surface was initially equilibrated with human normal serum (step I). CBP-SA (II-1) and b-CAb (clone 7502; II-2) were sequentially bound to the immunosensor with the fresh capture antibody and then washed with the same basal medium to establish a baseline (step II). Standard sample (75 L) combined with the detection antibody (clone 7501) was added, incubated for 15 min and then washed with pretreated normal serum for 15 min (step III). After regeneration with buffer lacking ionic calcium (step IV), steps I to IV were repeated for subsequent samples to evaluate the responses of the sensor to various doses of CK-MB (Figure 4, A). From the timeresponse curve at each dose, the sensor signal was measured by averaging the SPR signals obtained at quasi-steady state between 10 and 13 min during the washing step. The signal for each dose was plotted against the CK-MB concentration, yielding a sigmoidal dose-response curve (4, B). The limits of detection (2.71 ng/mL) and quantification (3.32 ng/mL) were calculated as the concentrations corresponding to the signals obtained by multiplying the standard deviation of the blank by three and five, respectively. The blank was measured to be 0.41±0.06 RU as background level when Sigma human serum was used as medium. The curve was then linearized via log-logit transformation and then regressed to obtain a linear equation with a correlation coefficient (R2) >0.99. We used a four-parameter log-logistic curve to determine the straight line that best fits a set of data points (Seo et al. 2016). The dynamic range of the linear region was 1.75 to 500.75 ng/mL CK-MB and the CV for triplicate measurements was <8.8%. The linear region closely covered the clinical range of cut-off value of about 1 ng/mL to upper detection of about 200 ng/mL (Cho et al. 2014; Panteghini and Pagani 1988).

The novel immunosensor was characterized by measuring blinded samples and comparing the results with those of the reference system, Pathfast (Figure 5). Samples (n=24) were prepared by spiking human serum with CK-MB (2–150 ng/mL) followed by analysis using both systems simultaneously. The signal produced from the test immunosensor was converted to the analyte concentration using the standard curve shown in Figure 4, B. The dose values for each sample using the two systems were plotted on an arithmetic scale. The CK-MB levels measured by Pathfast were about 5 times higher than those determined by the SPR 12

sensor in semi-continuous manner. Such difference can occur according to variable sources such as standards used for calibration, antibodies recognizing distinct isotopes of the antigen, analytical methods, and detection technologies employed to construct each system (Tate and Ward 2004). This effect may also be magnified if the target concentration range of the system was lowered (Cho et al. 2014; Vanderstichele et al. 2012). Therefore, correlation test is essential to compare the performance of newly developed system with that of the reference system. The data were linearly regressed to assess the correlation between the two systems, showing R2 >0.96. This indicated that the semi-continuous immunosensor study can be stepped forward for tests assessing the diagnostic accuracy with clinical samples.

The long-term reproducibility of the semi-continuous immunosensor for CK-MB for repeated measurements of standard samples was evaluated (Figure 6). The immunosensor was used to analyze 20–160 ng/mL CK-MB standards according to the protocol in Figure 4 (6, A). The analyses were repeated up to 30 days after the first use. The average of repeated measurements (n = 7) was plotted against the analyte dose, which yielded a dose-response curve with a CV of <8.0% (6, B). Therefore, the SPR-based immunosensor facilitates semicontinuous monitoring of biomarkers present at trace concentrations without the need to exchange the sensor chip under the analytical conditions employed in this study.

In contrast, when the conventional regeneration technique (i.e., use of acidic pH) was used, an immunosorbent became deteriorated during the repetitive uses for a long-term period. The stability indeed was tested with a SPR immunosensor for CK-MB prepared on CM5 sensor chip, showing gradual decrease of the sensor signal after temporarily being stable during the first few days (Figure S1 in Supplementary Information). Other examples regarding unstable bioactivity during repetitive uses can also be found in immunoaffinity chromatography (Moser and Hage 2010) and immunoassay based on streptavidin layer used for the capture antibody immobilization (Zhang et al. 2002). Furthermore, the same information was also presented by the manufacturer of the SPR sensor (Biacore Sensor Surface Handbook, 2008).

4. Conclusions A technique for real-time monitoring of low-abundance disease biomarkers was developed by

13

introducing an ‘ionic calcium switch’, which controls engagement and disengagement of a high-affinity capture antibody on the sensor surface. Thus, the switching method was used to supply fresh reagent for analysis of multiple samples on a single SPR sensor chip. The immunosorbent was regenerated almost completely under mild conditions, enabling highly reproducible long-term monitoring of an analyte in a semi-continuous manner. This immunoanalytical technique will facilitate detection of biomarkers in sample such as blood making feasible early diagnosis. Furthermore, simultaneous measurement of more than one biomarker of a disease, e.g., AMI, will enhance the sensitivity of the method (Kim et al. 2014). These suggest that semi-continuous immunosensing can be developed for companion diagnostics.

Acknowledgements This work was supported by the Research and Development Convergence Program of the Korea Research Council of Fundamental Science and Technology (KRCF; Project in the year 2013), and by the Converging Research Center Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning, Korea (Project number 2013K000249).

References Al-Hadi, H.A., Fox, K.A., 2009. SQUMJ. 9(3), 231-246. Best, M.G., Sol, N., Zijl, S., Reijneveld, J.C., Wesseling, P., Wurdinger, T., 2015. Acta Neuropathol. 129(6), 849-865. Biacore Sensor Surface Handbook., 2008. BR-1005-71 Edition AB. Chiu, A.Y., Matthew, W.D., Patterson, P.H., 1986. JCB. 103(4), 1383-1398. Cho, I.-H., Paek, E.-H., Kim, Y.-K., Kim, J.-H., Paek, S.-H., 2009. ANAL. CHIM. ACTA. 632(2), 247-255. Cho, J.H., Kim, M.H., Mok, R.S., Jeon, J.W., Lim, G.S., Chai, C.Y., Paek, S.H., 2014. Journal of chromatography B. 967, 139-146. Gao, Z., Hou, L., Xu, M., Tang, D., 2014. Scientific reports 4, 3966. Gibler, W.B., Runyon, J.P., Levy, R.C., Sayre, M.R., Kacich, R., Hattemer, C.R., Hamilton, C., Gerlach, J.W., Walsh, R.A., 1995. Ann. Emerg. Med. 25(1), 1-8. 14

Goode, J.A., Rushworth, J.V., Millner, P.A., 2015. Langmuir. 31(23), 6267-6276. Hsieh, H.V., Pfeiffer, Z.A., Amiss, T.J., Sherman, D.B., Pitner, J.B., 2004. Biosens. Bioelectron. 19(7), 653-660. Jagannadharao, P.R., Jarari, A.M., Hai, A., Rawal, A.K., Kolla, S.D., Sreekumar, S., Khurana, L., Sidhanathi, N.R., 2010. IJMBS. 2(5), 190-197. Kim, D.H., Seo, S.M., Cho, H.M., Hong, S.J., Lim, D.S., Paek, S.H., 2014. Biosens. Bioelectron. 62, 234-241. Kim, Y., Kim, H., Kim, S.Y., Lee, H.K., Kwon, H.J., Kim, Y.G., Lee, J., Kim, H.M., So, B.H., 2010. Am. J. Clin. Pathol. 134(1), 157-162. Koivunen, M.E., Krogsrud, R.L., 2006. Lab. Medicine. 37(8), 490-497. Kontos, M.C., Anderson, F.P., Hanbury, C.M., Roberts, C.S., Miller, W.G., Jesse, R.L., 1997. Am. J. Emerg. Med. 15(1), 14-19. Lindahl, B., Venge, P., Wallentin, L., 1995. Coron. Artery Dis. 6(4), 321-328. Moser, A.C., Hage, D.S., 2010. Bioanalysis 2(4), 769-790. Paek, S.H., Park, J.N., Kim, D.H., Kim, H.S., Ha, U.H., Seo, S.K., Paek, S.H., 2014. Analyst. 139(15), 3781-3789. Panteghini, M., Pagani, F., 1988. Z. klin. Chem. u. klin. Biochem. 26(5), 277-280. Park, J.N., Paek, S.H., Kim, D.H., Seo, S.M., Lim, G.S., Kang, J.H., Paek, S.P., Cho, I.H., Paek, S.H., 2016. Biosens. Bioelectron. 85, 611-617. Patel, S., Park, H., Bonato, P., Chan, L., Rodgers, M., 2012. J. Neuroeng. Rehabil. 9, 21. Schiel, J.E., Hage, D.S., 2009. J. Sep. Sci. 32(10), 1507-1522. Seo, S.-M., Kim, S.-W., Park, J.-N., Cho, J.-H., Kim, H.-S., Paek, S.-H., 2016. Biosens. Bioelectron. 83, 19-26. Tanious, F.A., Nguyen, B., Wilson, W.D., 2008. Methods Cell Biol. 84, 53-77. Tate, J., Ward, G., 2004. Clin. Biochem. Rev. 25(2), 105-120. Vanderstichele, H., Stoops, E., Vanmechelen, E., Jeromin, A., 2012. Alzheimer's research & therapy 4(5), 39. Wang, J., 2008. Chem. Rev. 108(2), 814-825. Wang, S., Poon, G.M., Wilson, W.D., 2015. Meth. Mol. Biol. 1334, 313-332. Zhang, F., Fath, M., Marks, R., Linhardt, R.J., 2002. Anal. Biochem. 304(2), 271-273. Zhang, H., Meyerhoff, M.E., 2006. Anal. Chem. 78(2), 609-616.

15

Figure 1. The technical basis of renewing the sensor surface under mild conditions to enable repeated measurements of different samples using the same sensor. To renew the sensor surface, we introduced an immunological binder sensitive to the conformational change of calcium binding protein (CBP) upon Ca2+ binding (A). To apply a switching reaction to semicontinuous biosensing, the antibody was first coated on the surface of the SPR sensor (step I, B). CBP conjugated to the capture antibody specific for CK-MB as the target analyte was then bound to the surface in the presence of ionic calcium (‘switched-on’ state; step II). A serum sample containing the analyte mixed with the detection antibody was added for analysis via formation of a sandwich complex on the sensor (step III). The surface was then regenerated in the absence of the inducer (‘switched-off’ state; step IV). Steps I to IV were repeated to analyze multiple samples using the same sensor.

Figure 2. Determination of the optimal arrangement of antibodies to form a sandwich complex with CK-MB in a solid-phase immunoassay. Two antibodies specific to CK-MB, clones 7501 and 7502, were used to analyze the marker. When clone 7501 was immobilized on the label-free sensor (e.g., Octet), the configuration showed low detection capabilities >0.1 g/mL (A) irrespective of whether the standard sample contained CK-MB only or also clone 7502 as the detection antibody. Next, when clone 7502 was used as the capture antibody, this configuration showed enhanced performances under the respective conditions comparing to the former arrangement (B).

Figure 3. Determination of the optimal conditions for immobilization of the capture antibody on the SPR sensor surface. The capture antibody was immobilized under the calcium switchon condition (step II in Figure 1, A), which was proceeded by sequentially binding a CBPstreptavidin conjugate (CBP-SA; A, step II-1) and biotinylated capture antibody (b-CAb; step II-2) via a biotin-streptavidin linkage. The signal increased in proportion with the increase in Ca2+ concentration from 2 to 20 mM (A). The capture antibody concentration was then increased fivefold, which resulted in 3.6-fold enhancement of binding without deterioration of the recovery rate (B).

16

Figure 4. Dose responses of the semi-continuous immunosensor to CK-MB under optimal conditions. The sensor surface renewal technique shown in Figure 3 was applied to analysis of CK-MB in human serum in a semi-continuous fashion using the same sensor (the upper panel). Steps I to IV were repeated for the various standard samples to obtain the sensor responses (A). The signal measured at each dose was plotted against the CK-MB concentration (B) and the curve was linearized via log-logit transformation (inset).

Figure 5. Correlation of the performance of the semi-continuous immunosensor with the reference system. The novel immunosensor was characterized by measuring blinded samples and comparing the analytical results with those of the reference system, Pathfast. Samples (n = 24) were prepared by spiking CK-MB (2–150 ng/mL) in human serum and then analyzed using both systems simultaneously. The dose values for each sample were plotted using an arithmetic scale. The data were subjected to linear regression to examine the correlation between the two systems (R2 >0.96).

Figure 6. Long-term reproducibility of repeated measurements of CK-MB samples on the same immunosensor. The surface-renewal-based immunosensor assay was performed according to the protocol in Figure 4 to analyze 20–160 ng/mL CK-MB standards (A). The analyses were repeated for up to 30 days after first use of the sensor. The average of repeated measurements (n = 7) was plotted against the analyte dose, which yielded adose-response curve with a CV of <8.0% (B).

17

Figure 1 (A) Ionic calcium (switch)-dependent antigen-antibody binding Conformation change of CBP upon Ca2+ binding

Antigen-antibody binding

‘Switched-on’

A) With Ca2+ Calcium binding protein (CBP) Ca2+

‘Switched-off’

Association Anti-CBP antibody

Reversible reaction depending on calcium concentration

B) Without Ca2+ Dissociation

(B) Semi-continuous biosensing via surface renewals ‘Switched-on’ state

I. Establishment of baseline

‘Switched-off’ state

II. Immobilization of fresh capture antibody

Capture binder in medium with calcium

CBP-capture antibody conjugate

III. Detection of target analyte

IV. Regeneration of sensor surface Ca2+

Detection antibody

Serum sample

T arget molecule

Ca2+

Anti-CBP antibody

Regeneration Medium without calcium Cyclic measurement by repeating steps I to IV

Figure 2 B) Sensor with clone 7502 as capture antibody (Octet)

A) Sensor with clone 7501 as capture antibody (Octet) 2.0

2.0 CK-MB only

Binding signal (nm)

Binding signal (nm)

CK-MB only CK-MB-clone 7502 binding complex

1.5

1.0

0.5

CK-MB-clone 7501 binding complex

1.5

1.0

0.5 0

0 0

0.01

0.1

1

10

0

100

0.01

0.1

1

10

100

Concentration of CK-MB (µg/mL)

Concentration of CK-MB (µg/mL)

18

Figure 3 A) Effect of ionic calcium concentration on the immobilization of capture antibody

CBP-SA

Step II-2

Washing

b-CAb

Washing

Step IV. Medium free of Ca2+

SPR signal (RU)

300 20 mM Ca2+

200 10 mM

Baseline 100 established

Baseline after regeneration

5 mM 2 mM

0

Baseline after regeneration

300

100

Capture antibody immobilized

200

80 60 40

100

20

0 mM

0

0 -100 1000

0

3000

2000

5000

4000

Recovery rate (%)

Step II-1

Calcium switch-off state

SPR signal (RU)

Calcium switch-on state

B) Effect of capture antibody concentration

1X

6000

5X

1X

5X

Concentration of capture antibody

Time (s)

Figure 4 Calcium switch-on state Step I. Establishment of baseline

Step II. Immobilization of capture antibody Step II-1

Step II-2 Baseline

Calcium switch-off state Step III. Detection of target analyte

Step IV. Regeneration of sensor surface

SPR signal (RU)

300 b-CAb 25 µg/mL

200

Washing

Neutral pH medium without calcium

Serum sample

CBP-SA 5 µg/mL

100 0

Washing

Baseline with normal serum

-100

Baseline with normal serum

-200

0

1000

2000

3000

4000

6000

5000

7000

8000

Time (s)

A) Time-response curve at each dose of CK-MB

50

SPR signal (RU)

50

SPR signal (RU)

60

CK-MB, ng/mL 500.75

250.75

40

100.75

30

10.75

20

1.75

10

Logit

60

B) Dose-response curve for CK-MB

40 30

y = 0.8696ln(x) - 12.941 R² = 0.9922

-7 -8 -9 -10 -11 -12 -13 1

20

10

100

1000

CK-MB (ng/mL)

10

0

1 -10

0 0

500

1000

1500

2000

1

2500

10

100

1000

Concentration of CK-MB (ng/mL)

Time (s)

19

Semi-continuous immunosensor (CK-MB, ng/mL)

Figure 5

100

y = 0.1943x + 3.6555 R² = 0.9693

80 60 40

20 0 0

100

200

300

400

500

PathFast (CK-MB, ng/mL)

Figure 6

A) Repeated use of the same immunosensor for analyzing CK-MB during a month

B) Reproducibility of the repeated sensor measurements

35

25

20 ng/mL CK-MB 40 100 160

SPR signal (RU)

SPR signal (RU)

30

25

Sample #1 Sample #2 Sample #3 Sample #4

20 15

10

20 15 10 5

5 0

0 1

3

5

7

10

20

Time (days)

30

20

40

100

CK-MB (ng/mL)

20

160

Highlights: • We investigated a novel surface modification to mildly regenerate SPR sensor chip.

• This was attained by a new binder changing the used capture antibody with fresh one.

• Analysis of multiple samples using the same sensor was reproducible at a rate >98.7%.

• The dose-response curve was linear for CK-MB, with an acceptable CV <8.8%.

• CK-MB was semi-continuously monitored on the same sensor up to 30 days.

21