Development of an Immunofluorescence Assay Module for Determination of the Mycotoxin Zearalenone in Water

Project Aquafluosense is designed to develop prototypes for a fluorescence-based instrumentation setup for in situ measurements of several characteristic parameters of water quality. In the scope of the project an enzyme-linked fluorescent immunoassay (ELFIA) method has been developed for the detection of several environmental xenobiotics, including mycotoxin zearalenone (ZON). ZON, produced by several plant pathogenic Fusarium species, has recently been identified as an emerging pollutant in surface water, presenting a hazard to aquatic ecosystems. Due to its physico-chemical properties, detection of ZON at low concentrations in surface water is a challenging task. The 96-well microplate-based fluorescence instrument is capable of detecting ZON in the concentration range of 0.09–400 ng/mL. The sensitivity and accuracy of the analytical method has been demonstrated by a comparative assessment with detection by high-performance liquid chromatography and by total internal reflection ellipsometry. The limit of detection of the method, 0.09 ng/mL, falls in the low range compared to the other reported immunoassays, but the main advantage of this ELFIA method is its efficacy in combined in situ applications for determination of various important water quality parameters detectable by induced fluorimerty—e.g., total organic carbon content, algal density or the level of other organic micropollutants detectable by immunofluorimetry. In addition, the immunofluorescence module can readily be expanded to other target analytes if proper antibodies are available for detection.


Introduction
Natural mycotoxin contamination has been identified as an emerging problem in agriculture. These toxic secondary metabolites produced by some fungal species are often found in food and feed (especially in grains) and cause high risk for food-and feed-borne intoxication in both humans and livestock [1]. The wide range of their negative effects includes, e.g., genotoxic, cytotoxic, mutagenic, and teratogenic effects [2]. Among all the toxic filamentous fungi species, Aspergillus, Fusarium, and Penicillium are important genera, producing regularly detected and widely studied toxins including aflatoxins, ochratoxin A, deoxynivalenol, T-2 toxin, fumonisin, and zearalenone (ZON) [3]. Mycotoxins found in human urine, indicate the possibility of chronic exposure [4]. In addition, a northward migration of toxicogenic plant pathogenic fungi has been reported assumedly triggered by climate change [5][6][7]. ZON is a frequently occurring mycotoxin, produced by species of the Fusarium genus, including F. graminearum, F. culmorum, F. semitectum, F. cerealis, and F. equiseti [8]. Its most known impact on human health is endocrine disruption: ZON and derivatives trigger estrogen-like effects in mammals causing alteration in hormone-mediated processes, e.g., the production of follicle-stimulating hormone (FSH) and luteinizing hormone (LH), and reductions in the number of Leydig and granulosa cells [9]. The potential health and economic impacts of ZON necessitate its routine monitoring in food and commodities, which have led to the development and validation of analytical methods in recent decades.

Mycotoxins as Pollutants in Surface Waters
In addition to food-and feed-borne intoxication, humans can also be affected through exposures via surface water contamination. Various phytopathogenic fungi, including Fusarium species, have been demonstrated to be capable of continuing to produce their secondary metabolites in water [10,11], and this process has been indicated to be a potential route of human exposure to mycotoxins [12]. Numerous investigations have reported the presence and input pathways of the toxin in surface or groundwater [13][14][15][16][17]. Mycotoxins may occur in surface waters by direct fungal contamination, by leaching from infested soil as water runoff from agriculture, by washing out from contaminated agricultural commodities such as cereals, oil, forage, feed etc. [18,19], or by mycotoxin biosynthesis in water by fungi [20,21]. In turn, mycotoxins have been considered as emerging surface water contaminants of diffuse and point source occurrence [22], through which fungal contaminants are considered emerging evidence-based threats for drinking water quality safety regulations [23]. Thus, water contamination by aflatoxins B2 and G2 were detected in water in Southern England at levels of 0.1-1.7 ng/mL [24]; aflatoxins B1 and B2, fumonisin B3, and ochratoxin A were detected at concentrations up to 0.035 ng/mL in the Tagus river Portugal [25]; phytoestrogens and mycotoxins were monitored in agricultural stream basins in the United States in Iowa, with occasional occurrence of deoxynivalenol above 0.1 ng/mL level [26]; aflatoxins B2, B1, and G1, as well as ochratoxin A were detected at levels of no toxicological risk up to 0.0007 ng/mL in bottled water in Portugal [27]; fumonisins were detected at up to 0.048 ng/mL levels in aqueous environmental samples in Poland [28]. Along with other mycotoxins, ZON and its metabolites also appear to be water contaminants. ZON was in found in surface waters, groundwater, and wastewater in Poland at levels up to 0.081 ng/mL, originating from cereal crops [14,15], and ZON, along with the metabolites zearalanone, α-zearalenol, β-zearalenol, α-zearalanol, and β-zearalanol, was detected in surface waters in Brazil at levels up to 4.12 ng/mL [29]. In total, 32 of 159 surface water samples collected in central Illinois have been positively tested, and 10 of them were above limit of quantification (LOQ) with concentrations between 0.002 and 0.006 ng/mL [30]. ZON also has been found in nine samples collected from eight Portuguese rivers and creeks ranging between 0.006 and 0.083 ng/mL [17], and appeared in the drainage water of a F. graminearum infected field in Switzerland with higher concentrations in the summer vegetation periods in a two-year field experiment [31]. Lower but detectable (0.002-0.005 ng/mL) concentrations were found in the Tiber river in Italy [32].
Moreover, the appearance of mycotoxins in the aquatic environment can adversely influence entire ecosystems. Thus, Fusarium mycotoxins, including ZON, can exert hormonal (estrogenic), hepatotoxic or genotoxic effects on fish [33,34]. Nonetheless, maximal residue levels (MRLs) to ensure compliance with the tolerable daily intake for humans (µg/kg body weight) by the European Union legislation [35] have been set only for food and feed, e.g., MRLs from 20 µg/kg in processed maize-based foods for infants and young children up to 200 µg/kg in unprocessed maize [36]. There is also a commission recommendation for ZON (and other mycotoxins) in products intended as animal feed as recommended by two scientific opinions of the European Food Safety Authority [37,38], but no declared maximum level for drinking water or surface waters have been established yet.
Project Aquafluosense (NVKP_16-1-2016-0049) [69] aims to develop a new water analysis system for natural and artificial waters, allowing complex, systematic and in situ fluorescence-based assessment and monitoring of water quality. The modular instrument family developed for main parameters (chlorophyll-a content, chemical and biochemical oxygen demands, total organic carbon, polycyclic aromatic hydrocarbon, and certain agricultural pollutant contents) can be individually configured for target tasks at each monitoring point. Within the project, we aimed to develop an enzyme-linked fluorescent immunoassay (ELFIA) module for monitoring and quantification of ZON.

Determination of Zearalenone by Autofluorescence
Detection capability of ZON by its own induced fluorescence was assessed in direct fluorescence (autofluorescence) measurements in water. A fluorescence intensity spectral map and a calibration curve are presented on Figure 1. Fluorescence is generated by the optical excitation of electrons, which emit fluorescent light when they return to their ground state from their excited state. As a loss of vibrational energy inevitably occurs during this process, the emission spectrum is shifted to longer wavelengths than the excitation wavelength (Stokes shift). Figure 1 depicts the relevant wavelength pairs of excitation and emission of fluorescence spectra, as well as the dependence of intensity of the emitted light (fluorescence) on the concentration of ZON. Excitation mapping was carried out by scanning emission intensities as a function of excitation intensities between 250 and 830 nm wavelengths depicting emission intensity in a color scale from blue to red (Figure 1a). On the basis of the fluorescence spectral map, the optimized peak for ZON measurement by autofluorescence was obtained with excitation at 280 nm wavelength and emission detection at 520 nm wavelengths. The dependence of the emitted light at these parameters on the concentration of ZON in the aqueous sample was also tested and was found to follow a sigmoidal (logistic) regression ( Figure 1b). Based on the sigmoid curve for autofluorescence, an LOD value of 11.5 µg/mL was determined. 830 nm wavelengths depicting emission intensity in a color scale from blue to red ( Figure  1a). On the basis of the fluorescence spectral map, the optimized peak for ZON measurement by autofluorescence was obtained with excitation at 280 nm wavelength and emission detection at 520 nm wavelengths. The dependence of the emitted light at these parameters on the concentration of ZON in the aqueous sample was also tested and was found to follow a sigmoidal (logistic) regression ( Figure 1b). Based on the sigmoid curve for autofluorescence, an LOD value of 11.5 µg/mL was determined.

Titration and Inhibition of the Antiserum
Efficacy of the immunization was monitored by titration of the two rabbit antisera against the coating antigen, ZON conjugated to bovine serum albumin (BSA) (ZON-BSA) between 1:50 and 1:12,200 dilution factors. Microplates were coated with the BSA conjugate at concentrations of 1 to 5 µg/mL in coating buffer. Serum titers, defined as the serum dilution that binds 50% of the antigen under the given conditions, were determined for sera obtained from rabbits (rabbit 1 and rabbit 2). Only slight differences occurred between the efficacy of the two antisera: titer values were 1:828 and 1:448 for antisera from rabbit 1 and rabbit 2, respectively ( Figure 2a). In the subsequent immunofluorescence assay experiments, antiserum from rabbit 1, showing somewhat higher affinity to the antigen, was applied. Accuracy and reproducibility of the measurements are better the nearer they are to the titer value, and thus antiserum was applied at a dilution factor of 1:1000 in the ELFIA tests.

Titration and Inhibition of the Antiserum
Efficacy of the immunization was monitored by titration of the two rabbit antisera against the coating antigen, ZON conjugated to bovine serum albumin (BSA) (ZON-BSA) between 1:50 and 1:12,200 dilution factors. Microplates were coated with the BSA conjugate at concentrations of 1 to 5 µg/mL in coating buffer. Serum titers, defined as the serum dilution that binds 50% of the antigen under the given conditions, were determined for sera obtained from rabbits (rabbit 1 and rabbit 2). Only slight differences occurred between the efficacy of the two antisera: titer values were 1:828 and 1:448 for antisera from rabbit 1 and rabbit 2, respectively ( Figure 2a). In the subsequent immunofluorescence assay experiments, antiserum from rabbit 1, showing somewhat higher affinity to the antigen, was applied. Accuracy and reproducibility of the measurements are better the nearer they are to the titer value, and thus antiserum was applied at a dilution factor of 1:1000 in the ELFIA tests.
To avoid the risk of saturatization or weak signal detection in assays, optimizations of the coating antigen concentration and serum dilution factor were performed by checkboard titration. The coating antigen, ZON-6 -carboxymethyloxime-BSA conjugate, was applied at concentrations in the range of 0.3125-2.5 µg/mL against the antiserum from rabbit 1 at dilutions in the range of 1:3375-1:1000 dilution factor. All combinations were investigated uninhibited and under inhibition by 3.2 ng/mL of ZON, as well. The coating antigen concentration and the antiserum dilution factor consistently influenced the analytical parameters ( Figure 2b). The analytical signal (relative fluorescence unit, RFU) increased with increasing concentrations of the coating antigen and decreased with increasing dilution of the serum. The addition of ZON at a concentration of 3.2 ng/mL resulted in an average 40.0% ± 0.1% inhibition of the assay signal. To avoid the risk of saturatization or weak signal detection in assays, optimizations of the coating antigen concentration and serum dilution factor were performed by checkboard titration. The coating antigen, ZON-6′-carboxymethyloxime-BSA conjugate, was applied at concentrations in the range of 0.3125-2.5 µg/mL against the antiserum from rabbit 1 at dilutions in the range of 1:3375-1:1000 dilution factor. All combinations were investigated uninhibited and under inhibition by 3.2 ng/mL of ZON, as well. The coating antigen concentration and the antiserum dilution factor consistently influenced the analytical parameters (Figure 2b). The analytical signal (relative fluorescence unit, RFU) increased with increasing concentrations of the coating antigen and decreased with increasing dilution of the serum. The addition of ZON at a concentration of 3.2 ng/mL resulted in an average 40.0% ± 0.1% inhibition of the assay signal.

Immunoassay
Indirect competitive ELFIAs were performed to established ZON calibration curves and to determine the LOD. The detection range was investigated in a concentration series of 0.004 pg/mL-2 µg/mL ZON in assay buffer. Matrix effects were determined by comparing calibration curves obtained in assay buffer and in surface water samples. No significant differences were determined among curves (p > 0.05), thus it has been concluded that determination of ZON in surface water can be performed without modification in sample preparation. Calibration curves and LODs were determined using both absorbance and fluorescence signals ( Figure 3). For comparability of the two detection modes, assay signals are represented as relative values (signals ratios to maximal signal levels). ZON at a concentration of 2000 ng/mL and above reached its full inhibition potential on the surface binding of the antibodies. This occurs because at this concentration the avidity of the primary antibody is saturated by ZON molecules in the solution, and therefore, further increases in ZON concentration cannot push the immunocomplexation equilibrium any further-ZON has reached its full capacity to block binding of the antibody to the coating antigen ZON-BSA conjugate on the surface of the immunoplate. The average relative analytical signal corresponding to the maximal assay signal produced by the uninhibited serum was set to the upper plateau level of the sigmoid standard curve. The

Immunoassay
Indirect competitive ELFIAs were performed to established ZON calibration curves and to determine the LOD. The detection range was investigated in a concentration series of 0.004 pg/mL-2 µg/mL ZON in assay buffer. Matrix effects were determined by comparing calibration curves obtained in assay buffer and in surface water samples. No significant differences were determined among curves (p > 0.05), thus it has been concluded that determination of ZON in surface water can be performed without modification in sample preparation. Calibration curves and LODs were determined using both absorbance and fluorescence signals ( Figure 3). For comparability of the two detection modes, assay signals are represented as relative values (signals ratios to maximal signal levels). ZON at a concentration of 2000 ng/mL and above reached its full inhibition potential on the surface binding of the antibodies. This occurs because at this concentration the avidity of the primary antibody is saturated by ZON molecules in the solution, and therefore, further increases in ZON concentration cannot push the immunocomplexation equilibrium any further-ZON has reached its full capacity to block binding of the antibody to the coating antigen ZON-BSA conjugate on the surface of the immunoplate. The average relative analytical signal corresponding to the maximal assay signal produced by the uninhibited serum was set to the upper plateau level of the sigmoid standard curve. The average relative analytical signal corresponding to full inhibition of the serum was considered as the lower plateau level of the sigmoid standard curve. Analytical parameters of calibration curves are summarized in Table 1.
LOD values were calculated for the two analytical detection modes of resorufin as a chromophore product. Thus, LOD = 0.25 and 0.09 ng/mL were determined for visual absorbance and fluorescence detections, respectively. Detection by fluorescence provided a wider and steeper dynamic range, thus ELFIA proved to be a 2.8-fold more sensitive method for ZON than the corresponding ELISA. It has to be noted that absorbance detection of resorufin by the application of QuantaRed Enhanced Chemifluorescent HRP Substrate Kit with HRP enzyme reaction (Thermo Fisher Scientific Inc., Waltham, MA, USA) provided 3.4fold lower LOD than that of o-phenylenediamine dihydrochloride (OPD) as chromophore in a similar colorimetric ELISA (LOD for OPD = 0.85 ng/mL). average relative analytical signal corresponding to full inhibition of the serum was considered as the lower plateau level of the sigmoid standard curve. Analytical parameters of calibration curves are summarized in Table 1.  Table 1. Statistical parameters of the logistic mathematical fitting using the Rodbard equation [70] in the immunoassay format using assay signals by resorufin as a chromophore product with detection of absorbance and fluorescence. LOD values were calculated for the two analytical detection modes of resorufin as a chromophore product. Thus, LOD = 0.25 and 0.09 ng/mL were determined for visual absorbance and fluorescence detections, respectively. Detection by fluorescence provided a wider and steeper dynamic range, thus ELFIA proved to be a 2.8-fold more sensitive method for ZON than the corresponding ELISA. It has to be noted that absorbance detection of resorufin by the application of QuantaRed Enhanced Chemifluorescent HRP Substrate Kit with HRP enzyme reaction (Thermo Fisher Scientific Inc., Waltham, MA, USA)  Table 1. Statistical parameters of the logistic mathematical fitting using the Rodbard equation [70] in the immunoassay format using assay signals by resorufin as a chromophore product with detection of absorbance and fluorescence.

Effects of Light Source Intensity
Calibration curves were determined in the induced fluorescence method at different light source intensities. LED power can be digitally set between 0.001 µW and 4.63 mW in 256 nonequidistant steps at 532 nm wavelength, and it was tuned to provide assay signals between 20 and 30 RFU as the suggested least detectable value and 4095 RFU as the highest readable signal by the instrument. Correspondingly, the LED power values investigated ranged between 1 .1 and 314 µW (1.1, 100.2, 169.4, 256 and 314 µW). As a background, 2000 ng/mL ZON solution was applied that triggered total inhibition of the antiserum. of the calibration curves were determined with the optimized immunoassay system (see Section 4.4.3). In the data evaluation process, RFUs were corrected with the background. Among analytical parameters, IC 50 values determined from the calibration curves in a concentration range of 0.0256-2000 ng/mL ZON were compared, and were found to be 2.52 ± 0.24, 3.04 ± 0.31 and 3.68 ± 0.29 ng/mL ZON for LED power values of 100.2, 169.4, 256 µW, respectively, while the LOD values did not appear to be significantly affected by the intensity of the light source.

Cross-Reactivity of the Antisera with Zearalenone Derivatives
Inhibition of the antiserum by metabolites and structural analogues of ZON was also determined under optimized assay conditions. ZON is metabolized mostly through hydrolysis mainly to βand α-zearalenol in yeast and ovine species, respectively [71,72]. Thus, IC 50 values by ZON metabolites and their reduced derivatives (α-and β zearalenol, αand β-zearalanol, zearalanone) were determined in the immunoassay by absorbance and fluorescence and relative cross-reactivities (considering inhibition by ZON as 100%) are listed in Table 2. The results indicate that the antibodies appear to be most sensitive to the presence of the unsaturation in the resorcyclic lactone ring and exhibit lower affinity to hydroxy metabolites. Stereoconfiguration of the hydroxyl group also occurs to influence recognition.

Analytical Detection Capability Compared to Other Immunoanalytical Methods
The analytical performance of the above ELFIA method was compared to that of other immunoanalytical methods (immunoassays, immunosensors) reported in the scientific literature. Detection capabilities of the immunoanalytical methods are listed in Table 3. The immunoanalytical methods reported are mostly developed to be used for crop commodities (maize, wheat, barley, rice), and LODs are specified in the method descriptions as detectable ZON concentrations in the commodity (e.g., µg/kg). For this comparison, Table 3 enlists LODs in the final diluted extract according to the method specifications published. In these assays, organic solvent extracts in aqueous acetonitrile or methanol (MeOH) were used for ZON determination, but the solvent was diluted to 0.1% or below during dilution with the assay buffer to reach the detection range. LODs ranged between 10 ng/mL by an IgYbased ELISA [57] down to 0.002 pg/mL by an an optical waveguide ligthmode spectroscopy immunosensor [67]. As seen from the analytical sensitivity data, the current ELFIA method is located in the middle range of the reported procedures regarding the LOD values. Analytical detection performance can be possibly improved by using more specific antibodies, but the main advantage of the ELFIA method lies not primarily in its sensitivity, but in its utility in combined in situ application in the determination of other water quality parameters detectable by induced fluorimetry-e.g., total organic carbon content, algal density or other organic microcontaminants (herbicide glyphosate or pharmaceutical carbamazepin). In addition, the immunofluorescence module can be easily extended to other target analytes if proper antibodies are available for their detection.

High-Performance Liquid Chromatography (HPLC)
Concentrations of ZON were also determined by HPLC instrumental analysis on the basis of peak areas in the chromatograms at the corresponding retention time (6.12 min) with excellent linear calibration characteristics in three parallel measurements. Peak areas determined at 236 nm for ZON concentrations between 10 and 2000 ng/mL were applied in linear regression, where regression coefficient of the concentration dependence was 0.999 in all measurements. Calibration curves with standard solutions were also investigated in MeOH:water = 7:3 and MeOH:phosphate buffer saline (PBS) = 1:1, where slope coefficients of linear regression were 26.90 ± 0.06 and 25.92 ± 0.06 for MeOH:water and MeOH:PBS, respectively. Peak purity was assessed by ratios of signal intensities (peak areas) recorded at 236 and 274 nm. These values for standard solutions were 2.15. Relative SDs established for different concentration levels for three parallel injections ranged between 0.65% and 1.76%. The LOD of the method was determined to be 10 ng/mL for ZON. A chromatogram and linear calibration (average of peak areas from three parallel measurements and their SDs) are presented in Figure 4.

Total Internal Reflection Ellipsometry
The immunoassay setup has also been applied in a system based on detection by total internal reflection ellipsometry (TIRE), where two parameters Ψ and ∆ are related, respectively, to the amplitude and phase shift of p-and s-components of the polarized light detected [73]. Since variations in the refractive index and the thickness of the adsorbed layers cause 10 times higher values of ∆ than of Ψ, ∆(λ) spectra were used in the TIRE method as a sensor response. A typical series of ∆(λ) spectra for ZON competitive immunoassays and the respective calibration curve of the assay signal in TIRE (δd corresponding to the shift in the adsorbed layer thickness vs. the concentration of ZON) obtained by sigmoidal fitting are depicted in Figure 5. The response is similar to that shown in Figure 2, where the highest concentration of ZON yields the lowest response, which is typical for competitive immunoassays. On the basis of the standard calibration curve, a LOD of 0.01 ng/mL for ZON was determined.

Total Internal Reflection Ellipsometry
The immunoassay setup has also been applied in a system based on detection by total internal reflection ellipsometry (TIRE), where two parameters Ψ and Δ are related, respectively, to the amplitude and phase shift of p-and s-components of the polarized light detected [73]. Since variations in the refractive index and the thickness of the adsorbed layers cause 10 times higher values of Δ than of Ψ, Δ(λ) spectra were used in the TIRE method as a sensor response. A typical series of Δ(λ) spectra for ZON competitive immunoassays and the respective calibration curve of the assay signal in TIRE (δd corresponding to the shift in the adsorbed layer thickness vs. the concentration of ZON) obtained by sigmoidal fitting are depicted in Figure 5. The response is similar to that shown in Figure 2, where the highest concentration of ZON yields the lowest response, which is typical for competitive immunoassays. On the basis of the standard calibration curve, a LOD of 0.01 ng/mL for ZON was determined.

Total Internal Reflection Ellipsometry
The immunoassay setup has also been applied in a system based on detection by total internal reflection ellipsometry (TIRE), where two parameters Ψ and Δ are related, respectively, to the amplitude and phase shift of p-and s-components of the polarized light detected [73]. Since variations in the refractive index and the thickness of the adsorbed layers cause 10 times higher values of Δ than of Ψ, Δ(λ) spectra were used in the TIRE method as a sensor response. A typical series of Δ(λ) spectra for ZON competitive immunoassays and the respective calibration curve of the assay signal in TIRE (δd corresponding to the shift in the adsorbed layer thickness vs. the concentration of ZON) obtained by sigmoidal fitting are depicted in Figure 5. The response is similar to that shown in Figure 2, where the highest concentration of ZON yields the lowest response, which is typical for competitive immunoassays. On the basis of the standard calibration curve, a LOD of 0.01 ng/mL for ZON was determined.

Conclusions
Within project Aquafluosense, successful development resulted in a modular instrumentation setup for fluorescence-based determination of several characteristic parameters of water quality. The application of fluorescence, as an analytical signal in an enzymelinked immunoassay format, results in a method of improved sensitivity with a lower LOD value than in the colorimetric ELISA (0.09 and 0.25 ng/mL, respectively). Moreover, resorufin-based determination proved lower LOD than application of OPD in colorimetric assay (LOD OPD = 0.85 ng/mL). This benefit of this is that it allows determination of lower pollutant concentrations in surface water, which contributes to more effective monitoring. The detection results were validated by HPLC instrumental analysis and by a TIRE immunosensor method. Although the sensor technology provided orders of magnitude lower LOD than the immunofluorescence method developed, the great advantage of the latter is that it makes in situ determination possible, and the 96-well microplate format used in the immunofluorescence determination prototype allows an assay capacity of 25 samples in parallel in triplicates (with standard curves of seven calibration points). An in situ detection module with a dynamic detection range between 0.4 and 400 ng/mL was developed for ZON. The immunofluorescence method and the instrument prototype constitute a part of the Aquafluosense modular instrument family for determination of characteristic water parameters and contaminants.

Materials and Reagents
Organic chemicals and solvents, mycotoxin ZON and its derivatives, goat antirabbit immunoglobulin conjugated to horseradish peroxidase (HRP) as secondary antibody, and salts for buffers were purchased from Sigma-Aldrich Inc. (St. Louise, MO, USA). The purity of standard solutions was ≥98%. Immunoassays were carried out in high-capacity 96-well microplates (Nunc, Roskilde, Denmark) for colorimetric assay and in low profile 96-well microplates with white wells for increased fluorescence (Bio-Rad Laboratories, Hercules, CA, USA). QuantaRed Enhanced Chemifluorescent HRP Substrate Kit was used as the last step in immunoassays. Surface water samples were obtained from river Danube at Budapest and lakes Velencei at Agárd and Balaton at Tihany (sampling site GPS latitude and longitude and coordinates-Danube: 47

Instrumentation
RFUs were determined by the prototype of a novel instrumentation developed in project Aquafluosense and realized in a modular setup ( Figure 6). The instrument was developed partly (motor, optics, sample holder) at Optimal Optik Ltd. (Budapest, Hungary) and partly (detector electronics) at the Budapest University of Technology and Economics (Budapest, Hungary), and was designed to fit the 96-well ELISA microplate format using a self-designed, 3D-printed holder for ELFIA ( Figure 6). The samples were illuminated in a dual head configuration with a high-power LED (Cree XPEBGR-L1-0000-00F01 with 520 to 535 nm minimum to maximum dominant wavelength range) in each head. The emitted fluorescence is measured in a dichroic beam path with silicon photodiodes (PIN-25D, OSI Optoelectronics) having large active area (d = 27.9 mm). The necessary high-spectral blocking and contrast was achieved by a combination of dichroic (Semrock FF562-Di03, edge: 562 nm) and bandpass optical filters on both the excitation (Semrock FF01-531/40-25, peak: 531 nm, width: 40 nm) and emission (Semrock FF01-593/40-25, peak: 593 nm, width: 40 nm) paths. The photodetector signal was coupled to a 2-stage amplifier unit (1st stage: OPA129 electrometer preamplifier, Texas Instruments; 2nd stage: AD620 instrumentation amplifier, Analog Devices) and then fed to a 12-bit analog-to-digital converter (Analog Devices AD7864-2 with 0 to 5 V unipolar input range) yielding 4095 resolvable RFUs. Gain and offset of the 2nd stage and the LED optical power were controlled by 256-stage (8-bit) digital potentiometers. As the LED power adjustment was nonequidistant, the optical power-control number curve was calibrated by a FieldMaxII-TO (Coherent) power meter with an OP-2 VIS sensor head set to a nominal wavelength of 532 nm. The instrument was equipped with stepping motors to move the detector heads over the 96-well microplates which provides fast and effective determination of individual RFUs in each microplate well. The instrument development is currently in the experimental phase; further decision about possible commercialization will be made by the project consortium. distant, the optical power-control number curve was calibrated by a FieldMaxII-TO (Coherent) power meter with an OP-2 VIS sensor head set to a nominal wavelength of 532 nm. The instrument was equipped with stepping motors to move the detector heads over the 96-well microplates which provides fast and effective determination of individual RFUs in each microplate well. The instrument development is currently in the experimental phase; further decision about possible commercialization will be made by the project consortium.

Determination of Zearalenone by Autofluorescence
Fluorescence spectra of ZON were recorded on a SpectraMax iD3 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA, USA) by scanning the excitation wavelengths between 250 and 830 nm and emission wavelengths between 270 and 850 nm, with step sizes of 10 nm, where the emission wavelength must be a minimum of 20 nm greater than excitation. A spectrum map was established by RFUs measured for a solution of ZON in PBS at a concentration of 2000 ng/mL with a corresponding RFU for PBS at each point as a background. An optimized peak on the basis of the fluorescence spectral map obtained was applied to establish a calibration curve (in the concentration range of 0.6-2000 ng/mL) and a LOD of ZON based on autofluorescence.

Determination of Zearalenone by Autofluorescence
Fluorescence spectra of ZON were recorded on a SpectraMax iD3 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA, USA) by scanning the excitation wavelengths between 250 and 830 nm and emission wavelengths between 270 and 850 nm, with step sizes of 10 nm, where the emission wavelength must be a minimum of 20 nm greater than excitation. A spectrum map was established by RFUs measured for a solution of ZON in PBS at a concentration of 2000 ng/mL with a corresponding RFU for PBS at each point as a background. An optimized peak on the basis of the fluorescence spectral map obtained was applied to establish a calibration curve (in the concentration range of 0.6-2000 ng/mL) and a LOD of ZON based on autofluorescence.

Hapten Synthesis and Conjugation
The corresponding hapten, ZON-6 -carboxymethyloxime was converted from ZON by the method of Thouvenot and Morfin [61]. The reaction mixture containing 300 mg of ZON dissolved in dry pyridine and 600 mg of carboxymethoxylamine was stirred overnight at room temperature (22 ± 0.5 • C), then evaporated and the residue was taken up in 50 mL of slightly alkaline (pH = 8) water. The aqueous phase was extracted with ethyl acetate (4 × 100 mL). The organic phase was dried over sodium sulfate, then separated and evaporated to afford 156 mg of the product. The process of conversion of ZON to the corresponding hapten was followed by thin layer chromatography using hexane-ethyl acetate (1:2) as an eluent. BSA and conalbumin (CONA) were applied as carrier proteins. The hapten was conjugated to these proteins through amide bonds [55], using the active ester method for conjugation. Thus, 125 mg of the hapten ZON-6 -carboxymethyloxime was dissolved in 6.2 mL of dry tetrahydrofurane (THF), then 24 mg of N-hydroxy-succinimide and 73 mg of N,N -dicyclohexylcarbodiimide were added. The mixture was stirred for 2 h at room temperature and the precipitation formed (dicyclohexylurea) was filtered off. In the mixture of 15.5 mL of water and 0.9 mL of THF, 150 mg of the proteins were dissolved in two separate batches. To these solutions, 3.1 mL of the above THF solution of the hapten active ester was added dropwise. The mixture was stirred for 24 h at 4 • C, and then the products (hapten-protein conjugate) were dialyzed against water at 4 • C for 1 week. Conjugation was monitored by UV spectroscopy; conjugates were lyophilized and stored at −20 • C until the analytical measurements.

Serum Preparation
Two 3-month old female New Zealand white rabbits were immunized intradermally with the CONA-hapten conjugate (immunogen). Rabbit immunisation was performed under the supervision of the Ethics Committee of Research on Animals (Food Science Research Institute, Institute of Food Sciences, Hungarian University of Agriculture and Life Sciences, Budapest, Hungary) and under the authorisation and inspection by the Government Office for Pest County in Hungary (Official permit for animal testing # PE/EA/45-6/2020, last date of approval: 21 February 2020). Initial immunization was carried out with 0.1 mg immunogen dissolved in PBS and emulsified in Freund's complete adjuvant (1:1 volume fraction). Then, injection of 0.15 mg of the immunogen in PBS and Freund's incomplete adjuvant (1:1 volume fraction) was given (the first three booster injections at 3-week intervals, with the subsequent one at a 1-month interval). One week after each immunization, rabbits were bled and after coagulation of the blood (4 • C overnight) the sera were centrifuged (2400 g, 15 min) and purified by gel chromatography on PD-10 desalting columns.

Immunoassay
Plates were coated with 1 µg/mL ZON-6 -carboxymethyloxime-BSA conjugate (ZON-BSA) in carbonate buffer (15 mM Na 2 CO 3 , 35 mM NaHCO 3 , pH = 9.6) overnight (~8 h) at 4 • C. The unbound conjugate was washed 4 times with PBS with 2% Tween20 (137 mM NaCl, 2.7 KCl, 10 mM Na 2 HPO 4 × 2H 2 O, pH = 7.4). Blocking was carried out with 150 µL/well 1% gelatin in PBS, for 1.5 h of incubation at 37 • C. After washing, competition was initiated by adding 50 µL/well of both ZON analytical standard and purified rabbit anti-ZON serum (in PBS buffer with 5% Tween20, dilution 1:1000). An analytical ZON standard stock solution (1 mg/mL ZON in MeOH) was used in serial dilution (0.004 pg/mL-2 µg/mL). Thus, MeOH content in the actual ELFIA measurements was 0.2% or below. After 1 h of incubation at 37 • C and four washes, 100 µL/well goat antirabbit IgG-HRP (horseradish peroxidase, dilution 1:7500) conjugate was added as secondary antibody and incubated for 1 h at 37 • C. Unbound secondary antibodies were washed out 4 times with PBS and 100 µL/well working solution of QuantaRed Enhanced Chemifluorescent HRP Substrate Kit were added (content of the working solution: QuantaRed ADHP concentrate, QuantaRed Enhancer Solution, and QuantaRed stable Peroxide in 1:50:50 (v/v) proportion). The kit contains 10-acetyl-3,7-dihydroxyphenoxazine (ADHP), a nonfluorescent compound that is dehydrogenated (oxidized) by HRP to resorufin, a highly fluorescent reaction product, which can also be measured in a colorimetric plate reader. After 5 min of incubation, the enzymatic activity was stopped by 10 µL of QuantaRed Stop Solution. Both absorbance and fluorescence were measured at 576 and at 593 nm wavelengths, respectively. After the colorimetric assay, the liquid phase was transferred with an 8-channel pipette into a white, low profile 96-well microplate where fluorescence was determined. Absorbances were read by SpectraMax iD3 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA, USA) at 576 nm wavelength, while relative fluorescence signals were determined by the prototype fluorimeter developed in project Aquafluosense [69,74] (see Section 4.2). Standard curves were obtained in different surface water samples as well, for evaluation of matrix effects in ZON determination from natural water bodies. Statistical analysis of standard curves was performed by comparison of IC 50 values by one-way analysis of variance followed by post hoc Tukey test at a significance level of 0.05.
The immunoassay developed in this study can be performed in situ by the instrument developed and built in a laboratory motor vehicle. Coating the assay microplates by the ZON-BSA conjugate can be carried out under laboratory conditions (method in Section 4.4.3) prior the measurement, and in situ determination can be performed on precoated microplates in the mobile equipment. Thus, the total immunoassay performance time is 2.5 h. Since the immunofluorescence instrument is equipped with stepping motor units, fluorescence determination on an entire plate requires approximately 2 min. A possibility of further shortening the procedure is potentially through performing both the coating and blocking steps in the laboratory a day before the in situ determination. In this case, the coating step can be applied at room temperature for 1 h and the blocking step with 1% gelatin in PBS for 1.5 h. Coated and blocked microplates can be stored at 4 • C until measurement; however, it is necessary to prevent evaporation in the wells by covering the microplate with parafilm.

High-Performance Liquid Chromatography (HPLC)
A calibration curve and the LOD for ZON were determined also by HPLC measurements on a Younglin YL9100 HPLC system equipped with a YL9150 autosampler (YL Instruments, Anyang, Korea). Compounds were separated on a column (150 × 4.6 mm i.d., 5 µm) at 30 • C containing C18 stationary phase. The PerfectSil 100 ODS-3 column was manufactured by MZ-Analysentechnik GmbH (Mainz, Germany). UV detector signals were recorded at λ 1 = 236 nm and λ 2 = 274 nm. Eluent flow rate was 1.0 mL/min with isocratic elution for 10 min (30:70 = A:B eluents, A = 90% water:10% MeOH, B = MeOH). The retention time of ZON under the current conditions was 6.12 min. LOD, defined as an analyte concentration corresponding to a signal level of signal/noise ratio of 3, was determined with standard solutions. For establishing an analytical standard curve, a stock solution of ZON at a concentration of 1.0 g/mL was prepared in MeOH. Calibration curves were obtained with 7 standard solutions between 10 and 2000 ng/mL in MeOH:water = 7:3) and MeOH:PBS (1:1).

Total Internal Reflection Ellipsometry (TIRE)
Total internal reflection ellipsometry (TIRE) experiments were carried out on an M-2000 automatic spectroscopic ellipsometer (J.A. Woollam Co., Lincoln, NE, USA) operating in the 370-1000 nm range using glass-based sensor chips fabricated in the laboratory by vacuum evaporation. Sensor surfaces were prepared by thermal evaporation of layers of chromium (Cr)-3 nm thick and gold (Au)-25 nm on standard microscopic glass slides (BK-7). The Cr layer improves the adhesion of gold to the glass surface. The Au surface was modified with mercaptoethyl sodium sulfonate to enhance the negative surface charge. The ellipsometer was equipped with a 68 • trapezoidal prism which allowed coupling the light beam at total internal reflection conditions to the gold film on the glass slide. The 0.2 mL reaction cell with the inlet and outlet tubes was attached underneath to the gold surface, allowing the injection of the required chemicals to perform binding reactions. The ellipsometry spectral scans were performed in a standard Trisma/HCl buffer solution (pH = 7.5) after completing each adsorption (binding) stage. For a competitive immunoassay, a ZON-6 -carboxymethyloxime-BSA conjugate (ZON-BSA) was electrostatically immobilized on the Au surface via a polyallylamine hydrochloride layer. In order to block all the remaining binding sites, an additional adsorption of BSA was carried out. Then a mixture of ZON-specific antiserum and solutions of free ZON (at a concentration range of 0.01 ng/mL-10 µg/mL) were injected into the cell with the intermediate rinsing with buffer. The mixtures were preincubated for 5 min before injection. A series of ∆ spectra were recorded after binding of ZON to the antibodies immobilized on the chip surface.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy reasons.