Bifunctional M13 Phage as Enzyme Container for the Reinforced Colorimetric–Photothermal Dual-Modal Sensing of Ochratoxin A

“Point of care” (POC) methods without expensive instruments and special technicians are greatly needed for high-throughput analysis of mycotoxins. In comparison, the most widely used screening method of the conventional enzyme-linked immunosorbent assay (ELISA) confronts low sensitivity and harmful competing antigens. Herein, we develop a plasmonic-photothermal ELISA that allows precise readout by color-temperature dual-modal signals based on enzymatic reaction-induced AuNP aggregation for highly sensitive detection of ochratoxin A (OTA). The bifunctional M13 phage carrying OTA that mimics the mimotope on the end of p3 proteins and abundant biotin molecules on the major p8 proteins is adopted as an eco-friendly competing antigen and enzyme container for amplifying the signal intensity. Under optimal conditions, both colorimetric and photothermal signals enable good dynamic linearity for quantitative OTA detection with the limits of detection at 12.1 and 8.6 pg mL−1, respectively. Additionally, the proposed ELISA was adapted to visual determination with a cutoff limit of 78 pg mL−1 according to a vivid color change from deep blue to red. The recoveries of OTA-spiked corn samples indicate the high accuracy and robustness of the proposed method. In conclusion, our proposed strategy provides a promising method for eco-friendly and sensitive POC screening of OTA. Moreover, it can be easily applied to other analytes by changing the involved specific mimotope sequence.


Introduction
Mycotoxin contamination in agricultural and food products is one of the major food safety concerns worldwide [1,2]. Penicilliumand Aspergillus-derived ochratoxin A (OTA) [3] is one of the most toxic mycotoxins that widely contaminate various crops (corn, wheat, beans, and cocoa), resulting in a serious threat to human health and consequential economic losses worldwide [4]. Many countries and international organizations have set strict limit standards for OTA residues [5,6], which are 0.5-10 µg kg −1 in different food commodities according to European Union regulations [7]. Regarding the universality and severe consequence of mycotoxin pollution, cheap "point of care" (POC) methods without expensive instruments and special technicians play a crucial role in the high-throughput analysis of mycotoxins [8]. Enzyme-linked immunosorbent assay (ELISA) is the most widely used POC method because of its merits of rapidity, simplicity, cost-effectiveness, and excellent specificity [9]. However, the conventional competitive ELISA for mycotoxin detection suffers from two major shortcomings [10,11]. First, the low-intensity and single-color colorimetric signal of enzymatic reaction-derived products leads to limited sensitivity and poor adaptability of on-site detection based on naked-eye interpretation in resource-constrained areas [12,13]. Second, using toxic target mycotoxin and organic reagents for preparing enzyme-or carrier protein-analyte conjugates as competing antigens leads to serious secondary pollution and a nonnegligible occupational hazard [14]. Hence, highly sensitive and eco-friendly immunoassays for mycotoxin screening must be developed urgently.
In recent decades, various colloidal nanoparticles have been used in developing biosensors to enhance or supersede current analytical techniques, making a great impact in research and practice applications [15,16]. In particular, plasmonic nanoparticles are interesting because their unique localized surface plasmon resonance (LSPR) enables them to produce intense responses to incident light [17], which can be linked to the presence of a target analyte to yield extremely sensitive detection [18,19]. Gold nanoparticles (AuNPs) are the most widely used plasmonic nanoparticles due to their merits, including easy preparation, excellent colloidal stability, and high accessibility for functionalization [20,21]. Highly sensitive AuNP-based sensing methods adopting various output signals, such as colorimetric signals [12], surface-enhanced Raman scattering [22], fluorescence [23], and photothermal signals [24] have been developed. AuNP-based colorimetric and photothermal sensing methods are particularly well suited for POC applications because of their simple signal readout via the naked eye or a thermometer [25]. Both colorimetric and photothermal signals generated from AuNPs are far more sensitive than those from conventional ELISAs because of the strong LSPR absorbance and photothermal effect of AuNPs. In recent years, plasmonic and photothermal ELISAs, which integrate LSPR modulation of AuNPs with conventional ELISA platforms, have attracted considerable attention because of their enhanced sensitivity [26]. The distinct color changes of plasmonic ELISAs are convenient for naked-eye readout without an excitation; the photothermal signal is more suitable than the colorimetric signal for muddy or colored samples [24]. Therefore, the combination of colorimetric and photothermal signals provides enhanced feasibility in POC applications and enhanced accuracy and reliability because of the mutual verification of the two signals [27]. Given that the two signals synchronously follow the LSPR modulation, dual modal plasmonic-photothermal ELISAs (ppELISAs) have been developed using AuNPs as signal transducers for the detection of disease biomarkers [28], nucleic acids [24], microorganisms [29], and mycotoxins [30]. A slight change in the compositions, shapes, sizes, or aggregation states of AuNPs may give rise to a remarkable LSPR variation. Among these modulation strategies, the AuNP aggregation-induced redshift of the LSPR absorbance always results in remarkably contrasting color changes and a distinct photothermal effect with a certain excitation. Previously, we reported a tyramine-mediated AuNP aggregation system based on the horseradish peroxidase (HRP)-catalyzed polymerization of phenolic hydroxyl in tyramine [31]. On this basis, a direct competitive ppELISA (dc-ppELISA) with enhanced sensitivity, accuracy, and reliability over conventional ELISAs can be promoted for mycotoxin detection.
The use of competing antigens of the enzyme-analyte conjugates in conventional direct competitive ELISAs for mycotoxin detection limits the sensitivity because one target molecule can only competitively inhibit the binding of one enzyme molecule. This approach also poses severe secondary pollution problems because of the consumption of toxic target mycotoxin and organic reagents. Therefore, novel competing antigens should be explored, particularly those that are eco-friendly and loaded with enhanced amounts of enzymes. Therefore, the M13 bacteriophage (M13 phage) carrying a mimic antigen epitope and multiple enzymes was proposed as a promising surrogate for competing antigens. The filamentous M13 is noninfectious to humans and composed of 2700 copies of the major p8 proteins and 3-5 copies of minor p3, p6, p7, and p9 proteins at the two ends [32]. It can be easily modified to mimic various target analytes, such as small molecules or proteins, by integrating mimotopes at the N-terminal of p3 proteins through a gene modificationfusion expression process [33]. The abundant copies of p8 proteins can be extensively functionalized as a container for high-density loading of signal transducers or regulators (e.g., fluorescent dyes and enzymes), resulting in remarkably enhanced signal intensities of biosensors [34]. Previously, we adopted an OTA-mimicking M13 phage (M13 OTA ) integrated with an OTA mimotope at the p3 proteins as an enzyme container to improve the sensitivity of a fluorescent ELISA for OTA detection [35].
Herein, we developed an eco-friendly dc-ppELISA for the highly sensitive detection of OTA in corn, and a colorimetric-photothermal dual model gave the developed method more applicability in POC screening. A biotinylated M13 phage (bio-M13 OTA ) was used as a competing antigen and container for glucose oxidase (GOx) enzymes, and citrate-AuNPs were applied as dual-modal signal transducers. The aggregation of citrate-capped AuNPs was induced by the HRP-catalyzed polymerization of tyramine, which was electrostatically adsorbed onto AuNPs in the presence of H 2 O 2 produced from GOx-catalyzed oxidation of glucose. The analytical performance of the proposed dc-ppELISA, including the limit of detection (LOD), 50% competitive inhibition concentration (IC 50 ), accuracy, and reliability, was evaluated and compared with that of conventional HRP-based ELISAs.

Principle of The Proposed dc-ppELISA Method
In this study, we developed a novel dc-ppELISA for highly sensitive detection of OTA using a bio-M13 OTA phage as an eco-friendly competing antigen and enzyme container. Colorimetric-photothermal responses from enzymatic reaction-induced AuNP aggregation were also used as dual output signals (Scheme 1). In detail, the p8 proteins of M13 OTA are functionalized with abundant biotin molecules to enable the high-capacity loading of GOx enzymes via the biotin-streptavidin system. The bio-M13 OTA phage is captured by anti-OTA antibodies immobilized on the microplate in the absence of OTA. Then, large amounts of GOx are captured, generating a large amount of H 2 O 2 via catalyzed oxidation of glucose. Afterward, H 2 O 2 is catalyzed by HRP to produce hydroxyl radicals that can stimulate cross-linking among the phenolic hydroxyl moieties of tyramine. The tyramine is absorbed on the AuNP surface via the electrostatic interaction between the amino group of tyramine and the negative charge on citrate-capped AuNPs, thereby bringing AuNPs together. In OTA-positive samples, the binding of M13 OTA and AuNP aggregation is inhibited via a competitive binding process, resulting in a minor degree of AuNP aggregation.
Toxins 2022, 14, x FOR PEER REVIEW 4 of 13 Scheme 1. The principle of dc-ppELISA using a bifunctional M13 phage as a competing antigen for colorimetric-photothermal response arising from enzymatic reaction-induced AuNP aggregation.

Optimization of a Signal Transduction System based on AuNP Aggregation
We first demonstrated the colorimetric-photothermal dual signal response to AuNP aggregation induced by the HRP-H2O2-tyramine system. Seven tests (a, b, c, d, e, f, and g) were conducted by adding one, two, or three components of the HRP-H2O2-tyramine system into a solution of 13 nm AuNPs. As shown in Figure 1A, only the addition of all HRP, H2O2, and Scheme 1. The principle of dc-ppELISA using a bifunctional M13 phage as a competing antigen for colorimetric-photothermal response arising from enzymatic reaction-induced AuNP aggregation.

Optimization of a Signal Transduction System Based on AuNP Aggregation
We first demonstrated the colorimetric-photothermal dual signal response to AuNP aggregation induced by the HRP-H 2 O 2 -tyramine system. Seven tests (a, b, c, d, e, f, and g) were conducted by adding one, two, or three components of the HRP-H 2 O 2 -tyramine system into a solution of 13 nm AuNPs. As shown in Figure 1A, only the addition of all HRP, H 2 O 2 , and tyramine (test b) results in a color change of the AuNP solution from red to blue, a redshift of LSPR from 520 to 630 nm, and a remarkable increase in LSPR absorbance in the range of 600-900 nm. Furthermore, the temperature of the reaction solution increases considerably after 5 min excitation with an 808 nm laser ( Figure 1B). Conversely, the absence of any HRP-H 2 O 2 -tyramine system components provides no obvious signal change in the AuNP solution. The resultant AuNP solutions of test b and test g were further characterized. Figure 1C,D and Figure S1 show that in the presence of HRP and tyramine, the dispersive AuNPs in the solution form bulk agglomerations accompanied by a dramatically increased hydrodynamic particle size from 13.9 to 108.7 nm when H 2 O 2 is added. All these results demonstrate the feasibility of dual-modal signal output arising from the AuNP aggregation induced by the HRP-H 2 O 2 -tyramine system. Moreover, H 2 O 2 can be used as an effective signal regulator in the presence of HRP and tyramine. The substrate solution of HRP-tyramine-AuNPs should be optimized to provide signal response toward low-concentration H2O2 and achieve a high sensing performance using the signal transduction system. In theory, a low tyramine concentration is conducive to achieving high sensitivity, whereas a high tyramine dosage can broaden the dynamic linearity of ppELISA. Nevertheless, tyramine overdosage may cause the self-aggregation of AuNPs due to charge neutralization. Therefore, the effect of tyramine dosage on AuNP aggregation was evaluated by adding different tyramine concentrations to the substrate solution containing 1.7 μg mL −1 of HRP and 5.3 nM of AuNPs. The ratio of the OD values at 630 and 520 nm (OD630nm/OD520nm) was chosen as the quantitative index for the AuNP aggregation. As shown in Figure S2A, the color of the AuNP solution is red when the tyramine concentration in the substrate solution is less than 33.3 µ g mL −1 . Conversely, when the tyramine concentration increases from 33.3 µ g mL −1 to 267 µ g mL −1 , the color of The substrate solution of HRP-tyramine-AuNPs should be optimized to provide signal response toward low-concentration H 2 O 2 and achieve a high sensing performance using the signal transduction system. In theory, a low tyramine concentration is conducive to achieving high sensitivity, whereas a high tyramine dosage can broaden the dynamic linearity of ppELISA. Nevertheless, tyramine overdosage may cause the self-aggregation of AuNPs due to charge neutralization. Therefore, the effect of tyramine dosage on AuNP aggregation was evaluated by adding different tyramine concentrations to the substrate solution containing 1.7 µg mL −1 of HRP and 5.3 nM of AuNPs. The ratio of the OD values at 630 and 520 nm (OD 630nm /OD 520nm ) was chosen as the quantitative index for the AuNP aggregation. As shown in Figure S2A, the color of the AuNP solution is red when the tyramine concentration in the substrate solution is less than 33.3 µg mL −1 . Conversely, when the tyramine concentration increases from 33.3 µg mL −1 to 267 µg mL −1 , the color of the AuNP solution obviously changes from red to blue, and the OD 630nm /OD 520nm value increases sharply from 0.2 to 1.1. These results show that an excess of tyramine can induce AuNP self-aggregation in the absence of H 2 O 2 . Moreover, 33.3 µg mL −1 of tyramine dosage is optimal in the substrate solution. Sufficient HRP is also vital in the signal transduction system. However, excess HRP may be absorbed on the AuNP surface, thereby deteriorating the AuNP aggregation triggered by the HRP-H 2 O 2 -tyramine system. The HRP dosage in the substrate solution was optimized by incubating different HRP concentrations in the substrate solution containing 33.3 µg mL −1 tyramine, 5.3 nM AuNPs, and 4 µM H 2 O 2 . The results in Figure S2B show that OD 630nm /OD 520nm increases remarkably when the HRP final concentration increases from 0 to 1.67 µg mL −1 , and the color of the AuNP solution varies from red to deep blue. Nevertheless, the OD 630nm /OD 520nm value decreases gradually when the HRP concentration further increases from 1.7 µg mL −1 to 26.7 µg mL −1 . Thus, 1.7 µg mL −1 of HPR in the substrate solution was used for all the succeeding studies.
Additionally, the sensitivity of the signal transduction system to H 2 O 2 is a key factor in determining the H 2 O 2 -dependent immunoassay sensitivity. Thus, the responses of colorimetric and photothermal signal outputs to H 2 O 2 were assessed in the optimal AuNP, tyramine, and HPR concentrations. As shown in Figure 2A, OD 630nm /OD 520nm increases with the increase of H 2 O 2 concentration from 0.4 to 3.2 µM. Then, it reaches a plateau with further increasing H 2 O 2 . A naked-eye cutoff limit of 1.6 µM for H 2 O 2 is achieved, which is addressed by a distinct color change from red to purple at this point.  Additionally, the sensitivity of the signal transduction system to H2O2 is a key factor in determining the H2O2-dependent immunoassay sensitivity. Thus, the responses of colorimetric and photothermal signal outputs to H2O2 were assessed in the optimal AuNP, tyramine, and HPR concentrations. As shown in Figure 2A, OD630nm/OD520nm increases with the increase of H2O2 concentration from 0.4 to 3.2 μM. Then, it reaches a plateau with further increasing H2O2. A naked-eye cutoff limit of 1.6 μM for H2O2 is achieved, which is addressed by a distinct color change from red to purple at this point. The corresponding regression equation in the linear detection range of 0.8-2.4 μM can be expressed by y = 0.7x − 0.4 (R 2 = 0.9905), and the LOD for H2O2 detection is calculated as 0.8 μM ( Figure  2A). After 5 min exposure under 808 nm excitation, the redshift of LSPR and the increase in the absorbance at 808 nm arising from AuNP aggregation led to a distinct upregulated photothermal response ( Figure 2B). The increase in temperature induced by the photothermal effect increases from 2.3 to 9.2 °C with the H2O2 increase from 0.4 to 4 μM. The corresponding regression equation in the linear detection range of 0.4-2.4 μM can be expressed by y = 2.8x + 1.3 (R 2 = 0.9912), and the LOD for H2O2 detection is calculated as 0.4 μM. The above results imply that compared with the plasmonic signal, the photothermal signal can provide enhanced sensitivity for the immunoassay.

Development of The dc-ppELISA Method
Thereafter, we developed a dc-ppELISA using a bifunctional phage (biotin-M13OTA) as a competing antigen, where the OTA-mimicking peptides were expressed at the end of p3 proteins. Moreover, numerous biotin molecules were modified on the major p8 proteins. According to our previous report, the as-prepared bio-M13OTA exhibits a high load-

Development of The dc-ppELISA Method
Thereafter, we developed a dc-ppELISA using a bifunctional phage (biotin-M13 OTA ) as a competing antigen, where the OTA-mimicking peptides were expressed at the end of p3 proteins. Moreover, numerous biotin molecules were modified on the major p8 proteins. According to our previous report, the as-prepared bio-M13 OTA exhibits a high loading capacity of 269 streptavidin molecules [35]. This finding indicates that biotin-M13 OTA can be used as the enzyme container for carrying a mass of biotinylated GOx (biotin-GOx). Unpurified anti-OTA ascitic fluids were used to coat the plate wells directly to maintain the maximum bio-activity of anti-OTA antibodies, and the photothermal mode have been used to proof the feasibility ( Figure S3). Moreover, the concentrations of anti-OTA ascites and bifunctional phages were optimized via a checkerboard titration method. The competitive inhibition rate based on the colorimetric signal was used to evaluate the optimal parameters. As shown in Table S1, the proposed dc-ppELISA exhibits the maximum competitive inhibition rate at 82.4% for 1 ng mL −1 OTA when the concentrations of anti-OTA ascites and bio-M13 OTA are 0.8 µg mL −1 and 2 × 10 9 pfu mL −1 , respectively. Subsequently, the effects of pH, ionic strength, and methanol content of the sample solution were optimized to achieve the best detection performance of the dc-ppELISA. As shown in Figure 3A, the competitive inhibition rate of 1.0 ng mL −1 OTA in the PB buffer increases as pH increases from 5.0 to 7.5. It then decreases with a higher pH value. Methanol is a typically organic phase in the extraction solution for the hydrophobic mycotoxins. The high methanol concentration may interfere with immunoreaction efficiency. The results in Figure 3B show that the inhibition rate of PB (pH = 7.5) is unchanged when methanol content is less than 5%. However, it obviously decreases when the methanol content exceeds 5%. Hence, the highest methanol content in the sample solution should be less than 5%. The results in Figure 3C show that 10 mM NaCl in the sample solution (pH = 7.5, 5% methanol) provides the highest competitive inhibition rate. Therefore, the corn extract sample using 80% methanol solution (v/v in water) should be diluted with 10 mM PB buffer containing 10 mM NaCl (pH = 7.5) to methanol, final concentration of 5%.  Figure 3B show that the inhibition rate of PB (pH = 7.5) is unchanged when methanol content is less than 5%. However, it obviously decreases when the methanol content exceeds 5%. Hence, the highest methanol content in the sample solution should be less than 5%. The results in Figure 3C show that 10 mM NaCl in the sample solution (pH = 7.5, 5% methanol) provides the highest competitive inhibition rate. Therefore, the corn extract sample using 80% methanol solution (v/v in water) should be diluted with 10 mM PB buffer containing 10 mM NaCl (pH = 7.5) to methanol, final concentration of 5%.

Analytical Performance of The Proposed dc-ppELISA Method
Under optimal conditions, competitive inhibition curves based on the dual readout were obtained by detecting standard OTA samples and plotting the resultant competitive inhibition rates against OTA concentration. Figure 4A shows a sudden color change from blue to purple at an OTA concentration of 78 pg mL −1 , which corresponds to 6.24 μg kg −1 , regarded as the naked-eye cutoff limit. The corresponding regression equation in the linear range of colorimetric sensing from 9.8 pg mL −1 to 312 pg mL −1 can be expressed as y = 19.1 ln(x)-37.7 (R 2 = 0.9943). The half-inhibition concentration (IC50) and 10% competitive inhibition concentration (IC10, equal to LOD) were calculated to be 98.4 pg mL −1 and 12.1 pg mL −1 , which correspond to 7.9 μg kg −1 and 1 μg kg −1 , respectively. Figure 4B shows the competitive inhibition curve based on the photothermal readout. The corresponding regression equation in a linear range from 9.8 pg mL −1 to 312 pg mL −1 can be expressed as y = 20.1 ln(x)-33.1 (R 2 = 0.9917). The IC50 and IC10 based on the photothermal signal are calculated to be 63.1 pg mL −1 and 8.6 pg mL −1 , which correspond to 5.0 μg kg −1 and 0.7 μg kg −1 , respectively. The LOD values of the dc-ppELISA based on the colorimetric and photothermal signals are 12.3-and 17.5-fold lower than those of the conventional HRP-based

Analytical Performance of The Proposed dc-ppELISA Method
Under optimal conditions, competitive inhibition curves based on the dual readout were obtained by detecting standard OTA samples and plotting the resultant competitive inhibition rates against OTA concentration. Figure 4A shows a sudden color change from blue to purple at an OTA concentration of 78 pg mL −1 , which corresponds to 6.24 µg kg −1 , regarded as the naked-eye cutoff limit. The corresponding regression equation in the linear range of colorimetric sensing from 9.8 pg mL −1 to 312 pg mL −1 can be expressed as y = 19.1 ln(x)-37.7 (R 2 = 0.9943). The half-inhibition concentration (IC 50 ) and 10% competitive inhibition concentration (IC 10 , equal to LOD) were calculated to be 98.4 pg mL −1 and 12.1 pg mL −1 , which correspond to 7.9 µg kg −1 and 1 µg kg −1 , respectively. Figure 4B shows the competitive inhibition curve based on the photothermal readout. The corre-sponding regression equation in a linear range from 9.8 pg mL −1 to 312 pg mL −1 can be expressed as y = 20.1 ln(x)-33.1 (R 2 = 0.9917). The IC 50 and IC 10 based on the photothermal signal are calculated to be 63.1 pg mL −1 and 8.6 pg mL −1 , which correspond to 5.0 µg kg −1 and 0.7 µg kg −1 , respectively. The LOD values of the dc-ppELISA based on the colorimetric and photothermal signals are 12.3-and 17.5-fold lower than those of the conventional HRP-based ELISA (0.15 ng mL −1 , Figure S4). This finding demonstrates the high potential of our proposed method for the sensitive detection of OTA in resource-limited regions. Later, the specificity of the proposed assay was determined by detecting several other common mycotoxins, including AFB1, DON, ZEN, FB1, and CIT. The results in Figure S5 show the negligible signal response of the proposed dc-ppELISA to AFB1, DON, ZEN, FB1, and CIT with a high concentration at 1000 ng mL −1 , indicating a cross-reactivity lower than 1‰. These results indicate the excellent selectivity of our proposed method for OTA detection.
The accuracy and precision of the dc-ppELISA method were determined by recovery evaluation from the OTA-fortified corn samples (2-100 μg kg −1 ). As shown in Table 1, the average recoveries (n = 3) based on the colorimetric readout in intra-assays and inter-assays range from 89.7% to 116.3% and from 95.4% to 119.4%, respectively; the average recoveries based on the temperature readout in intra-assays and inter-assays range from 95.4% to 113.6% and from 91.5% to 104.7%, respectively. These results indicate that our method can accurately detect OTA in real corn samples. The relative standard derivation for all the assays is less than 16%, revealing the good precision of our proposed method. Subsequently, 10 OTA-contaminated real corn samples were blindly analyzed through ultraperformance liquid chromatography-fluorescence detection (UPLC-FLD) method, and the results verified the practicability of our proposed method. Table 2 shows that all the detection results of our proposed method are basically consistent with those of the UPLC-FLD method, indicating that the proposed dc-ppELISA method is reliable for OTA screening in real corn samples.  Later, the specificity of the proposed assay was determined by detecting several other common mycotoxins, including AFB 1 , DON, ZEN, FB 1 , and CIT. The results in Figure S5 show the negligible signal response of the proposed dc-ppELISA to AFB 1 , DON, ZEN, FB 1 , and CIT with a high concentration at 1000 ng mL −1 , indicating a cross-reactivity lower than 1‰. These results indicate the excellent selectivity of our proposed method for OTA detection.
The accuracy and precision of the dc-ppELISA method were determined by recovery evaluation from the OTA-fortified corn samples (2-100 µg kg −1 ). As shown in Table 1, the average recoveries (n = 3) based on the colorimetric readout in intra-assays and interassays range from 89.7% to 116.3% and from 95.4% to 119.4%, respectively; the average recoveries based on the temperature readout in intra-assays and inter-assays range from 95.4% to 113.6% and from 91.5% to 104.7%, respectively. These results indicate that our method can accurately detect OTA in real corn samples. The relative standard derivation for all the assays is less than 16%, revealing the good precision of our proposed method. Subsequently, 10 OTA-contaminated real corn samples were blindly analyzed through ultraperformance liquid chromatography-fluorescence detection (UPLC-FLD) method, and the results verified the practicability of our proposed method. Table 2 shows that all the detection results of our proposed method are basically consistent with those of the UPLC-FLD method, indicating that the proposed dc-ppELISA method is reliable for OTA screening in real corn samples.

Conclusions
In conclusion, a plasmonic and photothermal immunoassay for OTA screening was developed using a bio-M13 OTA phage as a green competing antigen. A dual readout signal transduction system based on the HRP-tyramine-H 2 O 2 system gave the developed ELISA method more applicability in POC screening by replacing testing instruments. This novel method can achieve highly sensitive OTA detection based on a vivid color change that can be addressed by the naked eye, the redshift of LSPR spectra, and a temperature rise that can be read by a thermometer. The LODs based on colorimetric and photothermal signals are 12.3-and 17.5-fold lower than those of conventional HRP-based ELISAs. The results also demonstrate that our proposed method exhibits excellent selectivity, good accuracy, acceptable precision, and high reliability for sensing OTA in real corn samples. Although the detection time is increased compared with conventional ELISAs, we believe that the increase in time cost is acceptable compared with the improvement in sensing performance. We think that future work can be carried out from the efficient and rapid combination of enzymes and bacteriophages (such as click chemistry) to reduce the time cost. In summary, we proposed a promising POC screening strategy for the highly sensitive detection of OTA, and this strategy can be applied for other mycotoxin and contaminant screening by panning different mimotopes or ligands to the target analytes from phage display libraries.

Materials and Instruments
The following reagents were used: ochratoxin A (OTA), aflatoxin B 1 (AFB 1 ), fumonisin Transmission electron microscopy (TEM) images were obtained with a JEOL JEM-2100 electron microscope (Tokyo, Japan). Ultraviolet-visible (UV-Vis) absorption spectra were recorded using a double-beam UV-vis spectrophotometer (Cintra 10e; GBC, VI, Melbourne, Australia). Average hydrodynamic diameter was determined via dynamic light scattering (DLS) particle size analyzer (Zeta Sizer Nano ZS90, Malvern Instruments Ltd., Worcestershire, UK). The surface plasmon resonance (SPR) signal intensity of the AuNPs was determined using a Multiskan GO multimode reader (Thermo Fisher, Vantaa, Finland). Temperature of the AuNP solution under the excitation of an 808 nm laser was measured by an infrared camera, and all measurements were performed at room temperature.

Synthesis of Citrate-capped AuNPs
Citrate-capped AuNPs with an average diameter of 13 nm were synthesized via a one-step reduction method. Briefly, 250 mL of 1 mM HAuCl 4 solution was heated to a boil under vigorous stirring, and 25 mL of 38.8 mM citrate solution was quickly added followed by 15 min of reaction under a stirring and boiling condition. During this period, the solution color changed immediately from pale yellow to blue and then gradually to dark red. Subsequently, the reaction solution was placed in an ice bath for 15 min, and the resultant AuNP solution was filtered with a 220 nm microporous membrane and stored at 4 • C for later use. The concentration of AuNPs was calculated to be 15.9 nM according to Beer's Law.

Propagation of M13 OTA Bacteriophage
Escherichia coli ER2738 cells were cultured at 37 • C overnight in tetracycline-containing (10 µg mL −1 ) Luria-Bertani (LB) medium. Then, 200 µL of the above LB with enriched Escherichia coli cells and 1 µL of M13 OTA bacteriophage solution was inoculated into 20 mL of LB solution and cultured at 37 • C for 4~5 h under constant shaking at a speed of 250 rpm. Subsequently, the cell debris was removed via 10 min centrifugation at 5000 rpm, and then the proliferated M13 OTA in the supernatant was precipitated at 4 • C overnight after adding a PEG-NaCl solution (2.5 M NaCl, 20% PEG-8000). Thereafter, the precipitates of M13 OTA were collected via centrifugation (13,500 rpm, 10 min) and resuspended in 1 mL of phosphate-buffered (PB) solution. The cell debris was removed by spinning for 10 min at 5000 rpm. Then, 200 µL of PEG-NaCl solution was added to the supernatant for 1 h incubation on ice. The M13 OTA phage was obtained via centrifugation (13,500 rpm, 10 min) and resuspended with 200 µL of PB buffer. The concentration of amplified M13 OTA phage bacteriophages was determined via a plate count method.

Preparation of Bio-M13 OTA and Bio-GOx
Bio-M13 OTA and biotinylated GOx (bio-GOx) were obtained via the coupling between the amine groups on p8 proteins of the phage or GOx and the active ester group of NHS-biotin. During the preparation of bio-M13 OTA at the different molar ratios, different volumes of NHS-biotin (1 mg mL −1 , dissolved in DMF) were added to 1 mL of M13 OTA solution (2 × 10 9 pfu mL −1 , pH = 8.

dc-ppELISA Procedure for OTA
The procedure for the dc-ppELISA using bio-M13 OTA as a competing antigen is as follows. First, 100 µL of protein G (25 µg mL −1 , 0.01 M PBS pH 8.6) is added into each well of a 96-well microplate and incubated overnight at 4 • C. After washing three times with PBST (PBS containing 0.05% Tween 20) and once with PBS, 100 µL of anti-OTA ascitic fluids (0.75 µg mL −1 in PBS, pH = 8.6) is added to each well and incubated for 2 h at 37 • C. After removing excess anti-OTA ascitic fluids, 300 µL of blocking buffer (1% BSA in 0.01M PBS 7.4) is added for another 2 h incubation at 37 • C. After a repeated washing process, 50 µL of the sample solution and 50 µL of bio-M13 OTA solution (2 × 10 9 pfu mL −1 ) are added to each well. After incubating for 60 min at 37 • C, the microplate is washed again. Then, 100 µL of streptavidin solution is added and incubated for 30 min at 37 • C. After washing away the unbound streptavidin, 100 µL of bio-GOx is added and incubated for another 30 min at 37 • C. After unreacted bio-GOx is wiped off, 100 µL of D (+)-glucose (1 mg mL −1 in PBS, pH = 7.4) is added for 1 h incubation at 37 • C. Finally, 150 µL of substrate solution containing HRP (50 µL, 5 µg mL −1 ), tyramine (50 µL, 100 µg mL −1 ), and AuNPs (50 µL, 15.9 nM) is added into each well. After 5 min incubation, the microplate is photographed to record the color development and the optical density (OD) of each well at 520 and 630 nm are detected by multimode microplate reader. Furthermore, an infrared camera is used to record the temperature of each well after irradiation with an 808 nm laser for another 5 min. The inhibition rates were calculated according to the following formula: inhibition rate (%) = (1 − B/B 0 ) * 100%, where B and B 0 represent OD 630nm /OD 520nm values (colorimetric signal) or temperature changes (∆T, photothermal signal) obtained from detecting OTA-positive and -negative samples.

Sample Preparation
Corn samples were purchased from grain procurement agencies (Shandong, China), and verified to be OTA-free via high-performance liquid chromatography (HPLC). The HPLC method was according to the Chinese national standard GB 5009.96-2016, which is listed below: (a) chromatographic column: C18 column, column length 150 mm, inner diameter 4.6 mm, particle size 5 µm; (b) mobile phase: acetonitrile water glacial acetic acid (96 + 102 + 2); (c) flow rate: 1.0 mL/min; (d) column temperature: 35 • C; (e) injection volume: 50 µL; (f) detection wavelength: excitation wavelength 333 nm, emission wavelength 460 nm. All samples were thoroughly ground and mixed before use, and the corn samples for assay validation were prepared following a previously reported method and Chinese national standard GB 5009.96-2016. In brief, several portions of finely ground corn samples (5.0 g each portion) were spiked with OTA at different levels (2-100 µg kg −1 ). Each portion of OTA-spiked corn sample was extracted with 25 mL methanol-H 2 O solution (80% methanol: 20% H 2 O) under vigorous shaking for 20 min. After centrifugation at 10,000 rpm for 5 min, the supernatant was then collected and stored at 4 • C for further use.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/toxins15010005/s1, Figure S1: Particle size of AuNP aggregation induced by HRP-H 2 O 2tyramine system and dispersion of AuNPs in the absence of H 2 O 2 ; Figure S2: Optimization of the substrate solution of HRP-tyramine-AuNP signal transduction; Figure S3: Feasibility of adopting the biotinylated M13 OTA phage as competing antigen in the proposed dc-ppELISA for OTA detection; Figure S4: Cross-reactivity of the proposed dc-ppELISA toward other common mycotoxins; Figure S5: Calibration curve of conventional HRP-based ELISA; Table S1: Optimization of the working conditions of coating antibody and biotinylated M13 OTA phage using checkerboard method.