**1. Introduction**

Mycotoxin contamination is one of the most important problems in food and feed safety. According to previous studies, 25–50% of crops harvested worldwide are contaminated with different types of mycotoxins [1]. *Fusarium* species are the most widespread pathogens in cereals, and *Fusarium* toxins are the most reported mycotoxins in raw agricultural commodities [2]. Therefore, mycotoxins produced by *Fusarium* moulds significantly affect feed quality and safety and also represent a prominent issue in feed quality control after the most hazardous contaminants aflatoxins of *Aspergillus* origin. Accordingly, as most alerts in official food and feed monitoring mostly refer to aflatoxin contamination [3], most monitoring activities and analytical method development efforts are geared towards aflatoxins. Nonetheless, growing attention is paid to *Fusarium* mycotoxins as well, partly due to their spread caused by climate change and partly due to their well-known toxicological significance. Among *Fusarium* mycotoxins, deoxynivalenol (DON) and zearalenone (ZON), as well as their metabolites 3- and 15-acetyl-DON, α-, and β-zearalenol, are of special importance as they are formed under field conditions prior to harvest, being highly stable during storage and difficult to degrade by thermal processing [4–6]. Especially wheat, barley, oats, rye, corn, and triticale are vulnerable to *Fusarium* infection, and compared to other cereals, they are also frequently contaminated mostly with DON and ZON [7].

**Citation:** Majer-Baranyi, K.; Adányi, N.; Székács, A. Biosensors for Deoxynivalenol and Zearalenone Determination in Feed Quality Control. *Toxins* **2021**, *13*, 499. https:// doi.org/10.3390/toxins13070499

Received: 1 July 2021 Accepted: 15 July 2021 Published: 17 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Low-level contamination of *Fusarium* toxins is very frequent. DON and ZON are typically found in more than 50% and about 80%, respectively, of food samples tested in studies conducted between 2010 and 2015 in the EU [8]. DON, also known as vomitoxin, is of primary concern due to its genotoxicity, but it can also cause slow growth, lowered milk production in cattle, feed refusal, reduced egg production in laying hens, intestinal haemorrhage, and suppression of immune responses. ZON is problematic due to its hormonal effects causing changes in the reproductive system and reduced fertility. The use of toxin-contaminated feeds in livestock farming can cause a variety of adverse health effects in farm animals and a corresponding high degree of economic loss. Furthermore, contaminated feed can pose a health risk to humans indirectly, while mycotoxin carry-over is possible to milk, meat, and eggs; therefore, systematic control of mycotoxin content in feeds is of great importance. Although ZON, DON, and their metabolites are not of major concern due to their occurrence in milk, their presence has been reported in several studies. In an Italian study, 185 cow's milk-based infant formula products were investigated for ZON and its metabolites. ZON, α-, and β-zearalenol were detected in 9%, 26%, and 28.6% of the samples, respectively, with a maximum level of the latter metabolite of 73.2 ng/mL [9]. A technical survey from New Zealand reported that 0.06–0.08% of ZON residues mainly in form of α- and β-zearalenol can be secreted into milk, while DON residues occur in milk mainly in form of its diepoxy derivative exerting lower toxicity than the parent mycotoxin [10]. The EU has established maximum permitted levels and guidance levels of certain mycotoxins in feed, which should be routinely monitored. The guidance levels for ZON is 100–500 µg/kg in complementary and complete feeding-stuffs and 2–3 mg/kg for feed material, and for DON, it is 900 µg/kg in complementary and complete feeding-stuffs, 8 mg/kg in cereals and cereal products, and 12 mg/kg in maize by-products [11]. Commonly used techniques, such as high-performance liquid chromatography (HPLC) hyphenated with different detectors [12–14], liquid chromatography coupled with mass spectrometry (LC-MS) [15], liquid chromatography-tandem mass spectrometry (LC-MS/MS) [16–18], and gas chromatography-tandem mass spectrometry (GC-MS/MS) [19,20], for mycotoxin determination in food and feed have been powerful tools, as they provide proper sensitivity and accuracy in quantitative determination, but they are time-consuming, laborious, expensive, and require advanced instrumentation and trained staff [21]. In contrast, technically simple thin-layer chromatography (TLC) is also an excellent tool for rapid routine testing [22,23]; however, its sensitivity is unsatisfactory because of the even stricter EU limits. It is therefore essential to develop analytical methods that can detect the target analytes with sufficient sensitivity and accuracy and at the same time are inexpensive, fast, rely on simple measurement techniques, and allow on-site applications. The development and use of biosensors in food and feed analysis may efficiently address this challenge. This paper aims to provide an overview of recent advances and current trends in biosensor development for ZON and DON determination.

#### **2. The Use of Sensorics for Determination of DON and ZON**

Biosensors can be defined as a device incorporating an active biological sensing element (an enzyme, a tissue, living cells, antibodies, molecularly imprinted polymers (MIP), aptamers, DNA/RNA) connected to a transducer that converts the observed physical or chemical changes into a measurable signal. Biosensors can be classified according to the applied recognition elements (enzyme sensors, immunosensors, aptasensors, etc.) and also according to the signal transduction method: optical, electrochemical, piezoelectric, and thermometric; however, the latter application is not common in food and feed analysis. For mycotoxin determination, immunosensors are the most commonly applied analytical tools among biosensors, but beside that, MIP-based sensors and aptasensors (as artificial recognition element-based sensors) are also emerging techniques. Immunosensors employ antibodies, antibody fragments, antigens, or antigen conjugates as biomolecular recognition elements, and the specific antigen-antibody binding event is detected and converted to a measurable signal by the transducer. The basic working principle of the immunosensor set-

up is that the specific binding of the antibody or antigen immobilised on the transducer to the antigen or antibody in the sample produces an analytical signal that varies dynamically with the concentration of the analyte of interest. The formation of the immunocomplex can be determined either by label-free methods by directly measuring the physical changes induced by the binding event or by label-based modes using detection specific labels. For mycotoxin determination, both labelled and label-free immunosensors incorporated with various types of transducers are extensively researched and developed.

#### *2.1. Optical Immunosensors*

Nowadays, beside electrochemical immunosensors, the use of optical transducers has taken the lead in immunosensor development for mycotoxin determination because optical detection allows the construction of sensitive, simple, inexpensive, and portable analytical devices for on-site monitoring and also enables direct, real-time detection of various analytes. Optical biosensing can be divided into two general modes: label-free and label-based mode. Label-free biosensors do not require the use of any label to monitor the biorecognition event, while label-based protocols use specific labels like fluorescent dyes, enzymes, or nanoparticles, and the optical signal is generated by colorimetric, fluorescent, or luminescent methods [24,25]. Although these label-based methods are very sensitive and widely used, the performance of the sensor depends on the efficiency of the labelling step. Thus, the use of label-free biosensors may be preferable to the use of label-based ones, as they offer simple, rapid measuring procedures and enable real-time monitoring of the binding reaction. Of course, label-free optical biosensors also have disadvantages, especially in the determination of small molecules such as mycotoxins, as the sensor response often depends on the size of the analyte, and these analytes are mostly not chromogenic or fluorescent [26].

#### 2.1.1. Label-Free Optical Immunosensors

Surface plasmon resonance (SPR) technique has gained great attention in biosensor development lately. The technique was introduced in the early 1990s and since then become a powerful analytical tool in the risk assessment of contaminants in food and feed [27]. The SPR phenomenon occurs at the gold surface of the sensor chip when an incident polarised laser light beam strikes the surface at a particular angle through a prism (Figure 1A). It generates electron charge density waves called plasmons, which cause intensity reduction of the reflected light at this angle [28]. In the SPR immunosensor, immunogens (antibody or antigen) are immobilised on the gold layer of the chip mounted on a glass support. The binding of the analyte to the sensor surface causes a local change in refractive index, and corresponding shifts of the coupling angle are monitored in real time. SPR-based biosensors have received considerable attention in the past decades as they allow fast, reliable, and label-free detection of analytes [29]. In addition, they are suitable for realtime monitoring of the interaction kinetics; moreover, the biosensor chips are reusable. Another advantage of the SPR technique is that several measurements can be performed in parallel on a single sensor using multi-channel measurement. As several mycotoxins may be present simultaneously in feed or food samples, multiplex analysis is particularly relevant. Despite the fact that the SPR technique in biosensor research is being studied very extensively [30–33], only a few sensor development efforts suitable for ZON or DON determination have been investigated in recent years.

**Figure 1.** Operating principles of label-free optical immunosensors. (**A**) Surface plasmon resonance (SPR); (**B**) optical waveguide lightmode spectroscopy (OWLS); (**C**) planar waveguide; (**D**) white light reflectance spectroscopy. **Figure 1.** Operating principles of label-free optical immunosensors. (**A**) Surface plasmon resonance (SPR); (**B**) optical waveguide lightmode spectroscopy (OWLS); (**C**) planar waveguide; (**D**) white light reflectance spectroscopy.

Recently, Wei et al. [34] reported an SPR-based biosensor for the simultaneous determination of aflatoxin B1 (AFB1), ochratoxin A (OTA), ZON, and DON in corn and wheat. The limit of detection (LOD) for AFB1, OTA, ZON, and DON were identified as 0.59 ng/mL, 1.27 ng/mL, 7.07 ng/mL, and 3.26 ng/mL, respectively. Average recoveries were between 85% and 115%. Joshi et al. [35] developed two types of SPR-based biosensors for the detection of mycotoxins in barley. First, a double 3-plex assay was developed for the detection of DON, ZON, and T-2 toxin on the first chip and for OTA, fumonisin B1 (FB1), and AFB1 on the second chip using SPR. After determining the optimal conditions, the assay was transferred to a 6-plex format (six different mycotoxins determined on a single chip) in a portable nanostructured imaging surface plasmon resonance (iSPR) instrument, and the two assays were compared. The advances of iSPR technique over conventional SPR are the visualisation of the entire sensor surface in real time to monitor hundreds of molecular interactions simultaneously, and also multiplex detection is available. Results showed that DON, T-2, ZON, and FB1 could be detected at sufficient levels in barley samples according to the EC guidelines, but for OTA and AFB1, sensitivities should be improved when SPR was used for determination. The portable 6-plex iSPR was less sensitive but still allowed detection of DON, T-2, ZON, and FB1 at relevant levels. The sensitivities (IC50 values) obtained by iSPR biosensor in an assay buffer for T-2, FB1, and ZON were 10 ng/mL, 8 ng/mL, and 25 ng/mL, respectively. Recently, Wei et al. [34] reported an SPR-based biosensor for the simultaneous determination of aflatoxin B1 (AFB1), ochratoxin A (OTA), ZON, and DON in corn and wheat. The limit of detection (LOD) for AFB1, OTA, ZON, and DON were identified as 0.59 ng/mL, 1.27 ng/mL, 7.07 ng/mL, and 3.26 ng/mL, respectively. Average recoveries were between 85% and 115%. Joshi et al. [35] developed two types of SPR-based biosensors for the detection of mycotoxins in barley. First, a double 3-plex assay was developed for the detection of DON, ZON, and T-2 toxin on the first chip and for OTA, fumonisin B1 (FB1), and AFB1 on the second chip using SPR. After determining the optimal conditions, the assay was transferred to a 6-plex format (six different mycotoxins determined on a single chip) in a portable nanostructured imaging surface plasmon resonance (iSPR) instrument, and the two assays were compared. The advances of iSPR technique over conventional SPR are the visualisation of the entire sensor surface in real time to monitor hundreds of molecularinteractions simultaneously, and also multiplex detection is available. Results showed that DON, T-2, ZON, and FB1 could be detected at sufficient levels in barley samples according to the EC guidelines, but for OTA and AFB1, sensitivities should be improved when SPR was used for determination. The portable 6-plex iSPR was less sensitive but still allowed detection of DON, T-2, ZON, and FB1 at relevant levels. The sensitivities (IC<sup>50</sup> values) obtained by iSPR biosensor in an assay buffer for T-2, FB1, and ZON were 10 ng/mL, 8 ng/mL, and 25 ng/mL, respectively.

A rapid and sensitive iSPR assay was developed for *Fusarium* toxins by Hossain and Maragos [36] using secondary antibody with gold nanoparticles (AuNPs) as an amplification tag to determine DON, ZON, and T-2 toxin in wheat. LODs were 15 µg/kg for DON, 24 µg/kg for ZON, and 12 µg/kg for T-2 toxin. Sensor chips could be reused for over 46 cycles without significant signal loss, and it took 17.5 min to measure a sample, including the regeneration steps. The same research group developed an iSPR-based immunosensor for T-2 and T-2 toxin 3-glucoside (T2-G), so-called "masked" mycotoxin, determination in wheat, which is a niche in the field of research [37]. In their experiment on a carboxyl functionalised sensor surface, T-2-protein conjugate was immobilised using 1-ethyl-3-(3 dimethylaminopropyl) carbodiimide with N-hydroxysuccinimide (EDC-NHS) method. A A rapid and sensitive iSPR assay was developed for *Fusarium* toxins by Hossain and Maragos [36] using secondary antibody with gold nanoparticles (AuNPs) as an amplification tag to determine DON, ZON, and T-2 toxin in wheat. LODs were 15 µg/kg for DON, 24 µg/kg for ZON, and 12 µg/kg for T-2 toxin. Sensor chips could be reused for over 46 cycles without significant signal loss, and it took 17.5 min to measure a sample, including the regeneration steps. The same research group developed an iSPR-based immunosensor for T-2 and T-2 toxin 3-glucoside (T2-G), so-called "masked" mycotoxin, determination in wheat, which is a niche in the field of research [37]. In their experiment on a carboxyl functionalised sensor surface, T-2-protein conjugate was immobilised using 1-ethyl-3-(3 dimethylaminopropyl) carbodiimide with N-hydroxysuccinimide (EDC-NHS) method. A

competitive immunoassay format was applied to detect the mycotoxins, and a secondary antibody labelled with AuNPs was used for signal amplification. The LOD was 48 µg/kg of T-2 and 36 µg/kg of T-2-G; the recoveries ranged between 86–90%. Hu et al. [38] could achieve LODs for AFB1, OTA, and ZON as low as 8, 30, and 15 pg/mL, respectively, with their iSPR immunosensor using AuNPs for signal amplification.

Another emerging technique in the field of optical immunosensor development is the optical waveguide lightmode spectroscopy (OWLS) technique that enables monitoring molecular interactions on the sensor surface in a label-free manner in real-time (Figure 1B). The basic principle of the OWLS method is that linearly polarised He-Ne laser light is coupled by a diffraction grating into the waveguide layer. The incoupling is a resonance phenomenon that occurs at a defined angle of incidence that depends on the refractive index of the medium covering the surface of the waveguide. In the waveguide layer, light is guided by total internal reflection to the edges, where it is detected by photodiodes. By varying the angle of incidence of the light, the mode spectrum can be obtained from which effective refractive indices are calculated for both the electric and magnetic modes. The sensor consists of a glass substrate with a lower refractive index and a thin (160–220 nm) waveguide layer with a higher refractive index mounted on the top in which a fine optical grating (2400–3600 line/mm) is formed for in- or outcoupling of the light [39].

For DON measurement, Majer-Baranyi et al. [40] presented a label-free OWLS-based immunosensor. In their research, the sensor was modified by 3-aminopropyltriethoxysilane (APTS), and a DON-ovalbumin conjugate was immobilised via glutaraldehyde (GA). With the optimised sensor, DON content of spiked wheat flour samples was investigated using a competitive assay method where DON was quantitatively detectable in the 0.005–50 mg/kg concentration range, and it took 10 min to measure a sample, offering fast and sensitive determination of DON. Székács et al. [41] developed a competitive OWLS-based immunosensor for ZON determination in maize samples. In the competitive assay method, a ZON-bovine serum albumin (BSA) conjugate was immobilised on the sensor surface using three different surface modification methods. According to their results, the epoxy-modified sensors provided lower binding efficacy and reproducibility; when using amino-silanised sensor chips for immobilisation either by GA (APTS/GA) or succinic anhydride (SA) and EDC-NHS (APTS/SA/EDC-NHS) the detection range of ZON were the same in both cases, but for further application, the APTS/SA/EDC-NHS sensor was chosen due to the better reproducibility and longer shelf-life. The LOD of ZON was 0.002 pg/mL, and the dynamic measuring range was between 0.01 and 1 pg/mL.

Recently, another waveguide-based immunosensor for ZON detection was published also using a planar waveguide (PW) for the sensor set-up [42] (Figure 1C). The working principle of the sensor is as follows: circularly polarised laser light is incoupled into the planar waveguide, which propagates through by multiple internal reflections, and the outcoming light is collected by a charge-coupled device (CCD) array photodetector. The sensing principle is based on the different behaviour of the s- and p-components of polarised light. Changes in the refractive index of the covering media cause phase shifts between p- and s-polarisations of light, which are converted to a multiperiodic signal by a polariser and detected by a CCD photodetector. For ZON determination, polyclonal ZON-specific antibodies were immobilised on the functionalised surface, and the binding of ZON was detected in a direct manner. The LOD of the method was 0.01 ng/mL, and the dynamic working range was between 0.01–1000 ng/mL.

Another emerging label-free optical sensor technique is white light reflectance spectroscopy (WLRS), where a broadband light from a light source is emitted and guided vertically to the surface by a reflection probe consisting of six fibres distributed on the periphery of the circle-shaped probe, while the reflected light from the sample is collected by the optical fibre positioned in the centre of the probe and directed to the spectrometer (Figure 1D). The sensor consists of two layers: a Si substrate and, on top of this, a thicker silicon dioxide layer where the biomolecules can be immobilised. The emitted white light is reflected from the sensor consisting of layers with different refractive indexes, resulting

in an interference spectrum that is recorded by the spectrometer. Due to biomolecular interactions on the surface, the spectra shift to higher wavelengths [43]. A fast WLRS-based immunosensor for DON determination in wheat and maize samples was reported by Anastasiadis et al. [44], where DON-ovalbumin conjugate was immobilised on the aminosilanised sensor surface. A competitive immunoassay was performed where DON presented in the sample and DON immobilised on the sensor surface were competed for the anti-DON monoclonal antibody binding sites. The primary immunoreaction was followed by a signal enhancement step using an anti-mouse IgG secondary antibody. With the optimised sensor, wheat and maize samples were investigated. In the spiked grain samples, the LOD of DON was 62.5 µg/kg in both cases, while the linear response range was broadened up to 12.5 mg/kg. The measurement was completed within 17 min, including regeneration step, and a single chip could be reused 20 times.

The statistical parameters of the measurements, the cross reactivity, and the matrix analysed of optical immunosensors for DON and ZON detection are summarized in Table 1.

**Table 1.** Statistics of measuring parameters, cross reactivity, and the matrix analysed of optical immunosensors for DON and ZON detection.



**Table 1.** *Cont.*

Deoxynivalenol (DON), 15-acetyl-deoxynivalenol (15-AcDON), 3-acetyl-deoxynivalenol (3-AcDON), Deoxynivalenol 3-glucoside (3DON-Glc), Zearalenone (ZON), α-zearalenol (α-ZEL), β-zearalenol (β-ZEL) α-zearalanol (Zeranol), Ochratoxin A (OTA), Ochratoxin B (OTB), Aflatoxin B1 (AFB1), Aflatoxin B2 (AFB2), Aflatoxin M1 (AFM1), HT-2-glucoside (HT-2Glc), Fumonisin B1 (FB1), Fumonisin B2 (FB2), T-2 glucoside (T2-G), signal is not significant (n.s.), no data (n.d.), surface plasmon resonance (SPR), Imaging surface plasmon resonance (iSPR), optical waveguide lightmode spectroscopy (OWLS), planar waveguide (PW), white light reflectance spectroscopy (WLRS), near-infrared fluorescence-based lateral flow immunosensor (NIR-based LFIA).
