Characterization of Ab-N,S-GQD@AuNP-PAni on the CSR
The AuNP-PAni nanocomposite was deposited on a finely electropolymerized polyaniline-coated CSR electrode to form a AuNP-PAni-PAni/CSR electrode. The Ab-N,S-GQDs were then bound to the AuNP-PAni to prepare Ab-N,S-GQD@AuNP-PAni-PAni/CSR as shown in Scheme 1. In the N,S-GQD@AuNP-PAni nanocomposite, the AuNPs play an anchoring role between the N,S-GQDs and the polyaniline wires via soft acid – soft base interactions between Au and S. According to the TEM analysis, the diameter of the AuNP-PAni nanowires was 50 – 70 nm, where the AuNPs dispersed evenly in the polymeric chain, and their size ranged from 6 – 14 nm (Fig. 1A). The HRTEM image of the N,S-GQD@AuNP-PAni nanocomposite shows two distinct fringe patterns (Fig. 1B) of two crystalized structures of N,S-GQDs, and the AuNPs are deciphered as shown in Fig. 1C. The characteristic fringe of 0.24 nm for AuNPs is deciphered to the adjacent position of N,S-GQDs with a fringe distance of 0.21 nm, which is the distinctive fringe of the carbon lattice (Fig. 1C) [43, 44].
The structural properties of the N,S-GQD@AuNP-PAni nanocomposite were analyzed by XRD, as shown in Fig. 1D. PAni clearly revealed specific peak at 2θ = 26.0° corresponding to the (021). AuNP peaks are observed in the nanocomposite along with the characteristic peaks at 2θ = 23.6°, 25.5°, 28.2°, 38.2°, 44.3°, 64.4°, and 78.2° corresponding to the (100), (110), (111), (200), (220), and (221) planes, respectively (Fig. 1D) [45, 46]. After the N,S-GQDs were bound, the nanocomposite showed similar peaks and intensities, indicating that the attachment of GQDs does not induce any structural lattice changes of the AuNPs [47]. The graphitic layer shows a hump at 24° in the XRD spectrum, which is completely masked by the high-intensity peaks of AuNPs. To show the binding of N,S-GQD to AuNP-PAni, FT-IR analysis was performed. Similar peaks for PAni and AuNP-PAni were observed in addition to a characteristic peak at 2570 cm–1 for the thiol group of N,S-GQDs. A strong peak at 3300–3400 cm–1 was also observed for the amino or hydroxyl group as expected (Additional file 1: Fig. S1). These peaks indicate the attachment of N,S-GQDs on the AuNP-PAni nanocomposites. The conjugation of antibody was confirmed by ELISA as the absorbance value of the Ab-N,S-GQDs significantly increased compared with bare N,S-GQDs (Fig. 1E).
The electrochemical properties of the CSR electrode surface were measured by cyclic voltammetry. Despite the functional conducting matrix, the charge storage capacity of the bare CSR is very low, and the bare CSR shows a narrow curve, which significantly increases after the polyaniline coating (Fig. 1F). Additionally, a redox peak of polyaniline appears at + 0.8/+ 0.1 V, indicating the formation of the emeraldine salt of polyaniline [48]. After the formation of the nanocomposite, the conductivity of the disposable electrode shows an enhancement of the current density, indicating successful preparation of the sensor electrode for electrochemical analysis.
Optimization of the sensing performance of Ab-N,S-GQD@AuNP-PAni
The nanocomposite layer’s thickness on the CSR matrix is an essential parameter for maintaining the disposable electrode’s reproducibility. The thickness of the base matrix of polyaniline is directly proportional to the CV cycle number in the electropolymerization step. The resistance of the polyaniline-coated CSR becomes the lowest at 15 cycles because of the emeraldine salt formation of polyaniline (green). However, at 20 cycles or more, the redox reactions showed a decrease in conductivity due to predominantly overoxidized formation of pernigraniline of polyaniline (blue) (Additional file 1: Fig S2). Furthermore, after the electrolytic polymerization, SEM analysis of the CSRs showed that the polyaniline layer’s thickness became thicker with the increasing number of cycles (Additional file 1: Fig. S3A–C). A thick polyaniline layer can lead to a reverse effect on the conductivity of CSR. The 15 cycled PAni/CSR stability was tested over 50 cycles, showing excellent stability under the optimized polyaniline layer (Additional file 1: Fig. S4). The electrochemical properties of the CSRs did not change significantly after incubation in buffer for 24 h, confirming its stability as electrodes for sensor preparation (Additional file 1: Fig. S5). The thickness of the layer was further characterized by SEM and AFM. The bare CSR with a smooth surface becomes rough with coating of polymerized polyaniline (Fig. 2A‒B). The roughness of the polyaniline layer-coated CSR becomes relatively smoother again after drop-casting Ab-N,S-GQD@AuNP-PAni to microscale order (Fig. 2C). A similar observation was noted in the corresponding AFM images, as presented in Fig. 2D‒E. The CSR’s rough surface due to the bare polyaniline becomes relatively smooth after nanoconjugates formation, following the same trend as the SEM images. The AuNP-PAni forms a nanowire structure, and the binding of N,S-GQDs onto AuNPs makes the electrode smoother after the coating of nanocomposite to cover the porous structure of polyaniline. When the AuNP-PAni was modified on CSR, electrical conductivity significantly improved compared with AuNPs-modified CSR (Additional file 1: Fig. S6). In this study, the electric resistance was used as an indicator, and it was shown that AuNPs are a suitable material for this purpose because they lead to higher conductivity by forming a complex with polyaniline nanowires.
After optimizing the PAni electropolymerization and thickness of the sensor electrode, the changes in the electrochemical properties were investigated by EIS. The conductivity and dielectric properties of the CSR surface gradually decreased after polyaniline and successive Ab-N,S-GQD-AuNP-PAni conjugation (Fig. 2F), indicating successful formation of a sensor electrode suitable for virus detection.
Furthermore, the sensing area was optimized for virus detection. Electrodes with different sensor areas from 2 mm² to 25 mm² were prepared, and WSSV was detected by following the same procedure (Additional file 1: Fig. S7). The larger the sensor area was, the more remarkable the change in the Rct value. The sensing area’s size indicates the size of the contact surface with the virus during the antigen-antibody reaction. It suggests that the larger the area is, the more virus that binds to the sensor.
On the other hand, the sensor with a large area has a low correlation coefficient (R2 value) and a high error range, particularly in the high concentration range. As the area increases, it is difficult to obtain uniformity between electrodes with simple modification by only dropping nanomaterials, resulting in a low R2 value in virus detection. The electrode with a sensing area of 10 mm², which gave the most reliable result, was used as the optimum detection electrode.
Detection of WSSV
The Nyquist impedance plots of the disposable electrode after incubation of different concentrations of the virus from 102 –109 copies/ml are shown in Fig. 3A. The EIS responses of the sensor electrodes increase with the concentration of WSSV due to the high resistance accumulation between the virus-loaded nanocomposite and CSR. When WSSV binds to the sensing electrode, a large number of nonconducting virus particles cover the conducting surface of Ab-N,S-GQD@AuNP-PAni/CSR, increasing the charge transfer resistance (Rct). The percentage change of the signal difference between the Rct values of the corresponding virus-loaded electrode and the bare electrode was adopted as the measurement signal. The calibration plot displays an excellent linear relationship between Rct and the WSSV concentration (Fig. 3B). The LOD was found as low as 48.4 copies/ml, calculated by 3σ/S (S is the slope of the linear calibration plot, and σ is the unbiased standard deviation from the lowest signal of the detection result) [49]. This value is extremely low and sensitive enough to detect the real analyte [50]. After WSSV detection, the surface of the virus-loaded electrode exhibited a significantly increased roughness, indicating the presence of WSSV on the electrode (Additional file 1: Fig. S8).
We compared our sensing performance with various WSSV detection methods in Table 1. Many studies have successfully detected DNA as the target analyte. However, it is not easy to implement on-site and rapid detection because of the need to extract DNA from the WSSV. On the other hand, antigen detection with high sensitivity has not been reported earlier, except for electrochemical methods. Our detection system is useful because it shows high sensitivity, simplicity, and adaptability for on-site detection.
Table 1. Comparison of the sensing performance of our proposed sensor with various WSSV detection methods
Materials/Method of detection
|
Target virus
(Analyte)
|
Detection range
|
LOD
|
Ref
|
Piezoelectric microcantilever sensors
|
WSSV (DNA)
|
50 to 10⁵ virions/ml
|
100 virions/ml
|
[51]
|
Lateral flow assay
|
WSSV (DNA)
|
36–1784 viral copies/ng
|
356 viral copies/ng
|
[52]
|
Surface plasmon resonance
|
WSSV (Antigen)
|
5 to 50 ng/ml
|
2.5 ng/ml
|
[53]
|
Loop-mediated isothermal amplification
|
WSSV
(DNA)
|
0.05 to 1 μg/reaction
(LAMP products)
|
2×10² copies
|
[54]
|
Electrochemical
|
WSSV
(Antigen)
|
1.37 ×10⁻³ to 1.37×10⁷ copies/μL
|
1.36×10⁻³ copies/μl
|
[22]
|
Enzyme-linked immunosorbent assay
|
WSSV
(Antigen)
|
15–240 ng/well
|
250 pg/well
|
[55]
|
Polymerase chain reaction
|
WSSV
(DNA)
|
9.0 ×10¹ – 2.0×10⁴ copies/μg
|
4 copies/sample
|
[14]
|
Impedance electrochemical detection
|
WSSV
(Antigen)
|
102 to 109 DNA copies/ml
|
48.4 DNA copies/ml
|
This work
|
Selectivity and stability of the disposable electrode
As the antibody directs the interaction between the analyte and the sensor electrode, the sensor should possess high selectivity. However, the specificity test of the sensor is still crucial for clarifying any possible cross-reactivity in real application. To confirm the specificity towards WSSV, various other viruses and some materials were tested to the sensor electrode. The sensor responses, except for WSSV (Fig. 4A), were similar to that of the bare electrode, indicating the sensor specificity for the target virus. The high selectivity of the sensors was achieved by a close coating of Ab-N,S-GQD-AuNP-PAni and effective cleaning with PBST. When many foreign substances were present in the real matrix, and the non-specific adsorption occurred on the sensor surface, the substances other than the target WSSV were removed with a highly efficient washing solution. This led to high selectivity of our proposed detection method.
The effect of interferences on sensor’s performance and recovery ratio of target analyte were also investigated [56, 57]. A fixed concentration of 10⁴ copies/ml WSSV was mixed with different matrixes and then similarly detected by the sensor electrode. The concentration of the recovered WSSV was calculated using a calibration curve (Fig. 3B) based on the obtained Rct values. The recovery ratio was compared, as shown in Table 2. When the WSSV was in PBS, L-ascorbic acid, Fe²⁺, and Cu²⁺ ions, the recovery ratio was almost 110%, while in case of Mg²⁺ and Zn²⁺, it was 90%. These results indicate that some ions affect sensing performance with a standard deviation of ±13%. There was around 4% error in the recovery ratio, indicating that this system shows reasonable performances even in real matrix samples.
The stability of the disposable electrode was tested for 8 weeks to observe its applicability for long-term usage. As depicted in Fig. 4B, the signal intensity of Rct after loading of 104 copies/ml virus remained at 86% until 35 d. However, it dropped to 73.4% after 56 d of storage due to degradation of the antibody.
To extend its application to other types of analytes, we prepared two different electrodes conjugated with different anti-HEV and anti-HA antibodies and detected their corresponding target viruses. These results demonstrate that the Nyquist impedance in both cases increases with increasing virus concentration (Additional file 1: Fig. S9A and B), and their corresponding calibration lines show excellent linearity (Additional file 1: Fig. S10A and B). The limit of detection was calculated as 34.6 DNA copies/ml for G3 HEV and 0.98 fg/ml for influenza virus A.
Table 2. Detection of WSSV in various interferences
Suspension1)
|
Average Rct value2)
|
Concentration by Rct (copies/ml)3)
|
Recovery ratio (%)4)
|
Relative error (%)5)
|
PBS
|
15,833
|
10,915.0
|
109.2
|
± 2.4
|
L-ascorbic acid (1 mM)
|
15,244
|
10,987.2
|
109.9
|
± 3.8
|
Fe²⁺ (1mM)
|
15,784
|
11,577.6
|
110.8
|
± 4.2
|
Mg²⁺ (1mM)
|
15,484
|
9,187.9
|
91.9
|
± 3.8
|
Cu²⁺ (1mM)
|
15,662
|
12,523.3
|
125.2
|
± 2.9
|
Zn²⁺ (1mM)
|
15,422
|
8,512.7
|
88.1
|
± 3.3
|
1) WSSV concentration in suspension is 10⁴ copies/ml.
2) Average Rct value of WSSV detection (n=3)
3) WSSV concentration was calculated using the calibration curve (Fig. 3B).
4) Recovery was defined as
5) Recovery error was defined as denote Rct and average Rct values (n=3), respectively.
Real virus analysis
After successful detection of WSSV in a buffer medium, real samples were collected from 10 WSSV-infected shrimp and tested with the sensor. Their DNA copy numbers were compared with the results obtained from this electrochemical detection technique. The detection results are summarized in Table 3 and Fig. 5A. According to the RT-PCR data, sample Nos. 2 and 4 do not contain any WSSV, showing 2.4 and 6.5 copies/ml according to our electrochemical method, and can be ignored. The electrochemical detection results for sample Nos. 8 and 9 significantly deviate from the RT-PCR results. However, the overall trend of the RT-PCR results for the samples shows excellent similarity to the trend of the electrochemical sensor results, confirming the reproducibility of the sensor. In the western blot analysis, the virus titer above 107 copies/ml shows VP-28 protein bands at approximately 22 kDa, but less than 107 copies/ml could not be detected (Fig. 5B). This indicates that our sensing system shows a 6–7 order of magnitude higher sensitivity than western blot. This method, which can detect WSSV from specimens in less than 20 min, is much faster than the time-consuming RT-PCR, which is currently used as a gold standard. Although the correlation coefficient between the two methods is 90%, the developed method can be used to judge WSSV infection in a short time with easy handling.
Table 3. Details of the detection results for real sample detection using the electrochemical method and RT-PCR.
Sample No
|
Rct value±SD
(n=3)
|
WSSV concentration (DNA copies/ml)
|
VP28 detection
|
Shrimp
|
by EIS*
|
by RT-PCR
|
Control
|
2680±146
|
0
|
–
|
|
|
1
|
7179±238
|
4.8×10³
|
1.2×10⁵
|
no
|
live
|
2
|
2797±72
|
2.4×10⁰
|
0
|
no
|
live
|
3
|
8988±108
|
2.0×10⁴
|
6.2×10⁵
|
no
|
dead
|
4
|
4572±143
|
6.5×10⁰
|
0
|
no
|
live
|
5
|
12101±490
|
2.6×10⁵
|
8.4×10⁷
|
no
|
dead
|
6
|
16946±406
|
1.4×10⁷
|
9.6×10⁸
|
yes
|
dead
|
7
|
26949±140
|
4.7×10¹⁰
|
2.2×10¹⁰
|
yes
|
dead
|
8
|
18946±893
|
6.9×10⁷
|
7.5×10⁹
|
yes
|
dead
|
9
|
22308±195
|
1.0×10⁹
|
3.5×10⁸
|
yes
|
dead
|
10
|
13988±406
|
1.2×10⁶
|
2.4×10⁸
|
yes
|
dead
|
* The copy number of WSSV was determined from the calibration line (Fig. 3C) using the measured Rct value.