Label-Free Impedimetric Immunosensors Modulated by Protein A/Bovine Serum Albumin Layer for Ultrasensitive Detection of Salbutamol

The sensing properties of immunosensors are determined not only by the amount of immobilized antibodies but also by the number of effective antigen-binding sites of the immobilized antibody. Protein A (PA) exhibits a high degree of affinity with the Fc part of IgG antibody to feasibly produce oriented antibody immobilization. This work proposes a simple method to control the PA surface density on gold nanostructure (AuNS)-deposited screen-printed carbon electrodes (SPCEs) by mixing concentration-varied PA and bovine serum albumin (BSA), and to explore the effect of PA density on the affinity attachment of anti-salbutamol (SAL) antibodies by electrochemical impedance spectroscopy. A concentration of 100 μg/mL PA and 100 μg/mL BSA can obtain a saturated coverage on the 3-mercaptoproponic acid (MPA)/AuNS/SPCEs and exhibit a 50% PA density to adsorb the amount of anti-SAL, more than other concentration-varied PA/BSA-modified electrodes. Compared with the randomly immobilized anti-SAL/MPA/AuNS/SPCEs and the anti-SAL/PA(100 μg/mL):BSA(0 μg/mL)/MPA/AuNS/SPCE, the anti-SAL/PA(100 μg/mL): BSA(100 μg/mL)/MPA/AuNS/SPCE-based immunosensors have better sensing properties for SAL detection, with an extremely low detection limit of 0.2 fg/mL and high reproducibility (<2.5% relative standard deviation). The mixture of PA(100 μg/mL):BSA(100 μg/mL) for the modification of AuNS/SPCEs has great promise for forming an optimal protein layer for the oriented adsorption of IgG antibodies to construct ultrasensitive SAL immunosensors.


Introduction
In past few decades, immunosensors have emerged as a crucial tool for detecting various chemicals and biological compounds in applications including environmental monitoring, clinical diagnostics and food safety [1][2][3][4]. Their detection mechanism relies on the specificity and capturing ability of immobilized antibodies versus antigens. Recently, label-free techniques have been developed to directly monitor antigen-antibody interactions, such as surface plasmon resonance [5], quartz crystal

Reagents and Chemicals
Recombinant PA and porcine serum were respectively purchased from BioVision (MW:39210 Da, Cat. 6500B) and Invitrogen Corp (Cat.26250084, Bibco). Highly specific sheep monoclonal anti-SAL antibody was obtained from Randox Biosciences (MAB9343). Potassium hexacyanoferrate(III) (K 3  SPCEs with a working area of 3.2 mm 2 were obtained from Tyson Bioresearch (Taiwan). All chemicals were of reagent grade and were used without further purification. All solutions were prepared with water purified through a Milli-Q system.

AuNS Deposition
All electrochemical experiments were performed with an IM-6 impedance analyzer (Zahner Electrik GmbH, Germany) in a three-electrode system by using a Pt wire as the counter electrode and an Ag/AgCl electrode as the reference electrode. The bare SPCEs were first electrocleaned by sweeping the potential from 0 to 1.3 V for 20 cycles with a 0.1 V/s scanning rate in 0.1 M PBS (pH 7.0), and then oxidized at 2.0 V for 30 s to obtain more hydrophilic surfaces. After being rinsed with double distilled water, the electrocleaned SPCEs were dipped in the 8 mM HAuCl 4 solution (pH 2.0), containing 100 mM KCl, for the two-step AuNS deposition [26]. The first step is to nucleate Au nanoparticles on the electrocleaned SPCE surface in the cyclic potential range from 0.5 to −0.5 V with a scan rate of 50 mV/s for seven cycles. The second step was performed at the half peak potential of gold reduction (~0.62 V) for 10 min to grow nanostructures on the Au nanoparticles. The AuNS/SPCEs were then rinsed with double-distilled water to remove the free ions from the electrode surface for subsequent surface modification.
The morphology of the AuNS/SPCEs was observed by field-emission scanning electron microscopy (SEM, JEOL JSM-7401F, Tokyo, Japan) with an accelerating voltage of 3.0 kV. The roughness factor of AuNS/SPCEs was defined as the ratio of the real area of AuNS to the geometric area of SPCE. The real surface area of AuNS was determined in the 50 mM HClO 4 solution in the potential range of 0.3-1.5 V with a scan rate of 50 mV/s [26], and calculated from the oxide desorption charge density of 386 µC/cm 2 by integrating the reductive current of the gold oxide monolayer to obtain the roughness factor [26].

Immunosensor Preparation
A total of 10 µL aliquot of 10 mM MPA was dripped on the surface of AuNS/SPCEs at 30 • C for 1 h in an incubator with 30% relative humidity to form an SAM, which was used as a link for the immobilization of PA and BSA. The unbound MPA molecules were removed using double-distilled water. The MPA-modified AuNS/SPCEs were soaked overnight in 1 mL 20 mM MES solution (pH 4.6), containing 30 mM EDC and 2 mM NHS, to activate the carboxyl group of MPA SAM. Subsequently, 10 µL mixture of concentration-varied PA and BSA prepared in 10 mM PBS was placed on the EDC/NHS-activated MPA/AuNS/SPCEs at 30 • C for 40 min, and then the PA-modified electrodes were dipped in PBS containing 0.05% Tween-20, called PBST, and blank PBS to remove the unbound PA and BSA. A 10 µL 0.1 mg/mL anti-SAL antibody was placed on the PA-modified AuNS/SPCEs at 30 • C for 1 h, and then rinsed with PBST and PBS. The fabricating procedures for the immunosensors are depicted in Scheme 1. After performing the oriented attachment of anti-SAL, the immunosensors were incubated with SAL samples with different concentrations prepared in PBS for 40 min at room temperature, and then cleaned with PBS. The change in the electrochemical properties of the electrode/solution interface was estimated by cyclic voltammetry (CV) and EIS.

Electrochemical Measurements
An equimolar Fe(CN) 6 3−/4− mixture (2.5 mM) in 10 mM PBS (pH 7.4) was used as a mediator to estimate the electrochemical properties of electrodes in each modification step and quantify the immunoreaction. The CV was performed in the potential range of −0.1 V-+0.5 V at a scanning rate of 20 mV/s to evaluate the redox behavior of Fe(CN) 6 3−/4− mediator on the modified electrodes. EIS measurements were performed in a frequency range of 1-100 kHz at a +0.21 V potential, added by a 5 mV amplitude sine wave. Measurement of the impedance spectra and simulation of equivalent circuits were performed using the IM-6/THALES software package. Figure 1a shows the SEM image of the morphology of AuNS/SPCEs. The result indicates that the surface of AuNS/SPCEs has well-dispersed binary gold structures with submicrometer-scaled pyramidal structures on the micrometer-scaled particles. The mechanism of two-step AuNS formation is discussed in our previous study [26]. However, the AuNS morphology in this work is slightly different from that in our previous study [26]; this inconsistency is attributed to the effect of the SPEC carbon substrate on Au particle nucleation. The highly rough AuNS helps immobilize greater bioaffinity recognition molecules or antibodies. The roughness factor of AuNS/SPCEs was evaluated by the reductive charge of gold oxide desorption, as shown in Figure 1b. By integrating the reductive current of gold oxide in the range of 0.75-1.0 V, the real surface area of the AuNS/SPCEs was 28.89 ± 0.36 mm 2 with a corresponding roughness factor of 9.03 ± 0.11 (n = four electrodes). In contrast, the polished gold disk electrodes had a smaller roughness factor, 3.38, calculated from Figure 1b. The result suggests that the AuNS can significantly increase the surface area of SPCEs.

Effect of PA and BSA Mixture
Four kinds of PA and BSA mixtures with concentration (µg/mL) ratios of 100:0, 100:50, 100:100 and 100:200 were used to explore the effect of PA coverage on the amount of attached antibody. Figure 2 shows the impedance spectra measured at the AuNS/SPCEs followed by MPA modification, the immobilization of the concentration ratio-varied PA:BSA mixture, and the anti-SAL adsorption. The impedance spectra measured at the AuNS/SPCEs (curves (i) of Figure 2) exhibit a dominant linear region, implying that the redox reaction of the Fe(CN) 6 3−/4− mediator at the AuNS/SPCEs is a diffusion-controlled behavior, due to its fast electron-transfer rate [26]. Following MPA modification, the Nyquist plots show a linear part at lower frequencies and a semicircle part at higher frequencies, indicative of a kinetics-controlled region. Moreover, the radius of the semicircle increased with the modification steps of the PA:BSA immobilization and the anti-SAL adsorption. The results indicate that the impedance of the solution/electrode interface is increased by the modification.  The Randles equivalent circuit, consisting of four elements-the solution resistance (R s ), the Warburg impedance (Z w ), the pure capacitance of electrical double layer (C dl ) and the electron transfer resistance (R et )-is frequently used to explain the diffusive and kinetic behavior of solution/electrode interfaces. As the modified MPA molecules and immobilized PA/BSA proteins have difficulty forming a dense and pinhole-free modification layer on the rough AuNS, the constant phase element (CPE) was used to replace the C dl to elucidate the inhomogeneity of the electrode surface [9,26]. The modified Randles equivalent model is shown in the inset of Figure 3. The impedance of the CPE can be presented by Z CPE (ω) = Z 0 (jω) −α , where Z 0 is a constant, j is an imaginary number, ω is the angular frequency, and 0 < α < 1. When α is closer to 1, the CPE becomes more capacitive. The modified Randles equivalent circuit, using CPE instead of C dl , is more suitable to explain the electrochemical properties of the modified electrodes [27]. Figure 3 shows the Bode plots for the PA(100):BSA(100)/MPA/AuNS/SPCE, obtained from computer fitting using the modified Randles equivalent circuit and the experimental measurement. The fitting result shows a good consistency with the experimental measurement. The modified Randles circuit model was used to obtain the element values for the AuNS/SPCEs, followed by MPA modification, PA/BSA immobilization and anti-SAL affinity, which have a mean error of less than 0.2% and a maximum error of 2.1% for all fitting data. In Faradic impedance measurement, the R et value is more sensitive to the change in the interfacial impedance than the CPE value [9,26,28]. Therefore, the R et value is used to quantitatively evaluate the effect of different PA:BSA modifications on the amount of adsorbed anti-SAL, as listed in Table 1. The R et values, measured at the AuNS/SPCEs, were only 70−80 Ω, implying that the AuNS deposition supplies a high conductance and a large surface to significantly promote the redox current of Fe(CN) 6 3    Table 1 to quantify the magnitude of adsorbed anti-SAL on the density-varied PA-modified electrodes. The result shows that ∆R et-anti-SAL increased with the decreasing PA density, in the 83%−50% range. Similarly, Boujday et al. found that a PA layer with about 60% coverage adsorbed more IgG than a PA layer with 100% coverage, as observed by infrared spectroscopy and quartz crystal microbalance [25]. The phenomenon is attributed to the PA molecules having a lower coverage with a greater flexibility and accessibility for antibody adsorption. However, we think that, in addition to contributing to the flexibility of a low coverage PA layer, the ordered arrangement of PA molecules also plays an important role in the antibody adsorption. In principle, the R et measured on an immunosensor is affected by the thickness, structure and polarity of the biorecognition layer. It is difficult to realize the actual orientation of the adsorbed anti-SAL by using EIS measurement. The amount and orientation of adsorbed anti-SAL determines the thickness and the structure of the anti-SAL/PA:BSA biorecognition layer. The relationship between the R et value and the amount/structure parameters of anti-SAL/PA:BSA may be represented by a linear regression or an exponential regression. According to the results in Table 1, the ∆R et-anti-SAL didn't show a linear relationship with the PA density, because the ∆R et-anti-SAL obtained from the 33%PA density of PA(100):BSA(200)-immobilized electrode is much lower than that obtained from the 50%PA density of PA(100):BSA(100)-immobilized electrode. It is worth studying the effect of PA/BSA ratio on the amount of adsorbed antibodies by using different analytical techniques.
Furthermore, the PA(100):BSA(100) immobilization produces a saturated and compact protein layer, which may reduce the PA flexibility, but the 50% PA density of PA(100):BSA(100)-modified electrodes had a significantly higher ∆R et-anti-SAL (p<0.05) than other PA density-modified electrodes, evaluated by Student's t-test. The result implies that the 50% density PA molecules of the ordered and compact arrangement help absorb more antibodies than the more flexible PA layer with 83% density, because the 83% coverage PA layer had a shorter interval between different PA molecules and a more flexible structure, which caused the adsorbed antibodies to tilt and produce steric hindrance, thus reducing the amount of antibodies adsorbed. Insets of Figure 2a-c respectively show antibody adsorption schemes on the protein layers with 83%, 63% and 50% density PA. Furthermore, the electrodes with 33% density PA had the lowest ∆R et-anti-SAL , indicating that a reduced PA density correlates with lower amounts of antibodies adsorbed, as the PA is immobilized in a compact protein layer. The corresponding scheme is depicted in the inset of Figure 2d. These results suggest that both the density and flexibility of immobilized PA molecules determine the numbers and orientation of the adsorbed antibodies.

Calibration Curves
The amount of effective paratopes of adsorbed antibody determines the magnitude of antigen binding. In the study, three kinds of immunosensors based on the anti-SAL/MPA/AuNS/SPCEs, the anti-SAL/PA (100) Figure 4a shows the impedance spectra, measured at the anti-SAL/MPA/AuNS/SPCEs for concentration-varied SAL samples. All Nyquist plots obtained from the anti-SAL/MPA/AuNS/SPCEs presented an obvious linear part and a small semicircular part, indicating the low interfacial impedance required to produce the significant diffusion-controlled behavior due to the smaller amount of randomly oriented anti-SAL [26]. Moreover, the semicircle's radius increased with the SAL concentrations, implying increased interfacial impedance. After fitting the experimental spectra by using the modified Randles circuit, the relationship between the ∆R et-SAL and the SAL concentration is shown in Figure 4a    immunosensors measured for the lower concentrations was much higher than for the higher concentrations, implying the effective paratopes of oriented anti-SAL with high accessibility and immunoreacting affinity with SAL of lower concentration. The phenomenon is similar to the results of Hafaiedh's study, which reported protein G-modified immunosensors with a significantly increased sensitivity in the C-reactive protein (CRP) detection of low concentrations compared to the random physisorption of anti-CRP-modified immunosensors [24]. The calculated LOD was 0.2 fg/mL (S/N>3) which is much lower than that obtained at the anti-SAL/MPA/AuNS/SPCEs and the anti-SAL/PA(100):BSA(0)/MPA/AuNS/SPCEs. The result proves that the anti-SAL adsorbed on a protein layer of 50% density PA has the more effective paratopes to obtain a more sensitive immunoreaction and a lower LOD. Moreover, the relative standard deviation (<2.5%) of ∆R et-SAL for each SAL measurement obtained at the anti-SAL/PA(100):BSA(100)/MPA/AuNS/SPCE-based immunosensors (n = 3 electrodes) is significantly smaller than that (<23.4%) of the ∆R et-SAL measured at the anti-SAL/MPA/AuNS/SPCE-based immunosensors (n = 3). The result implies that the protein layer with a 50% PA density of constructs a more compact and stable sensing interface on the sensor surface to promote reproducibility. Moreover, the label-free anti-SAL/PA(100):BSA(100)/MPA/AuNS/SPCEs-based immunosensors had a lower LOD than other reported SAL immunosensors with sandwich and competitive immunoassay methods, as compared in Table 2 [26,[29][30][31]. This implies that the anti-SAL/PA:BSA/AuNS/SPCEs have promise to act as highly sensitive immunosensors for the detection of small molecule chemicals. The monoclonal anti-SAL used in the literature [26] and this work is a monovalent antibody, which is suitable for label-free detection. If the antibodies are required for sandwich immunoassay [29], the paratope number of antibodies should be larger than 1.

Other Sensing Properties
In real sample measurement, matrix interference is an unavoidable challenge for label-free impedimetric immunosensors. Although the matrix influence can be reduced or eliminated by diluting real samples [34,35], the analyte concentration is simultaneously diluted. Therefore, a high sensitivity and low LOD immunosensor is needed for the dilution sensing strategy. An SAL of 0.01, 0.1 and 1 ng/mL was spiked in a commercial porcine serum to estimate the practical application of the immunosensors. After diluting the SAL-containing serum by a factor of 1000 with 10 mM PBS to the 0.01, 0.1 and 1 pg/mL SAL concentrations, the ∆R et-SAL values were measured and calculated using the regression equation, as shown in Table 3. The recovery of detected 0.01−1 pg/mL SAL was in the range of 95.1-99.0%. The result shows a high feasibility for using the anti-SAL/PA(100):BSA(100)/MPA/AuNS/SPCEs to develop a commercial immunosensor for the practical measurement of SAL concentrations in real samples. Moreover, the selectivity of the developed immunosensor was investigated from the individual responses to 1 ng/mL ractopamine (RAC), clenbuterol (CLB) and SAL. The ratios of ∆R et-RAC and ∆R et-CLB to the ∆R et-SAL were, respectively, 6.25% and 3.27% (data not shown). The result implies that the immunosensors have good specificity and selectivity.

Conclusions
The mixture of PA and BSA was used to modify the EDC/NHS-activated MPA/AuNS/SPCEs for the oriented immobilization of anti-SAL. Particularly, the PA/BSA protein layer constructed by the concentrations of 100 µg/mL PA and 100 µg/mL BSA can supply greater adsorption of monoclonal anti-SAL, which can improve the sensing properties of EIS-based immunsensors. The anti-SAL/PA(100):BSA(100)/MPA/AuNS/SPCE-based immunosensors present an ultrasensitive detection for SAL, with an extremely low LOD (0.2 fg/mL) and good recovery for the practical measurement of SAL in 1000-times-diluted porcine serum samples. The PA/BSA protein layer with 50% PA density has great promise for greater adsorption of different IgG antibodies to construct highly sensitive immunosensors.