**1. Introduction**

According to the Food and Agriculture Organization of the United Nations and a recent study by Eskola et al. in 2019, approximately 25% of food-crops worldwide are contaminated with mycotoxins, a group of secondary metabolites produced by molds [1,2]. Particularly, the toxins of *Aspergillus, Fusarium,* and *Penicillium* species are the most detrimental ones because these so-called aflatoxins, ochratoxins, trichothecenes (especially deoxynivalenol), and zearalenone have various adverse effects on the health of humans and animals [3–8]. From all mycotoxins, aflatoxin B1 (AFB1), mainly produced by *Aspergillus* species, has the strongest adverse effects on health, as it might lead to liver cancer [9–12]. Several factors can promote the enrichment of mycotoxins within food products. Key drivers are

late harvests of crops as well as elevated humidity and temperature during storage. Within the EU, but also in the USA, Brazil, and other countries, regulatory requirements regarding the maximum tolerable levels of mycotoxins in various foods have been established. Regulatory limits in the EU for the IARC group 1 carcinogen aflatoxin B1 are set to 2–8 ng/g (ppb; ng·mL−<sup>1</sup> ) in grain, corn, nuts, and fruits, as published in (EG) Nr. 1881/2006. Especially due to those harsh effects and corresponding low regulatory limits, highly sensitive and reliable detection technologies are of primary importance.

Currently, there are basically three analytical technologies used for mycotoxin testing. On the one hand, laboratory-based liquid chromatography (LC) coupled to mass spectrometry (MS) is the most common method, as well as enzyme-linked immunosorbent assay (ELISA), are routinely used for highly sensitive mycotoxin testing. On the other hand, lateral flow assays (LFA) are used for fast but less sensitive on-field testing [13–19]. Although LC-MS/MS methods have the great advantage of high sensitivity and simultaneous detection of currently more than 500 mycotoxins within a single run, expensive equipment, highly qualified staff, and a possible complex sample cleanup inhibit the applicability for fast on-site testing [20]. In comparison to LC-based methods, ELISA techniques are cheaper and, according to Renauld and colleagues (2019), faster in assay procedure, but also require a lab with the corresponding equipment for sample preparation and analysis [21].

In the field of ELISA, the most commonly used format for aflatoxin detection is the competitive assay because of the small molecular structure of the antigen that prevents the simultaneous binding of two antibodies, which would be the prerequisite for a sandwich ELISA. The basic working principle is the coating of mycotoxin-conjugate onto a microtiter plate followed by the addition and incubation of a mycotoxin-specific antibody in the presence of a liquid extract of sample material. In this step, mycotoxins in the sample compete with coated mycotoxins for antibody-binding. Subsequently, antibodies saturated with soluble mycotoxins are removed during a washing step and cannot contribute to a signal, achieved by either direct readout or indirect readout by means of a labeled secondary antibody. In this format, a low signal corresponds to a high concentration of mycotoxin within the sample, and vice versa. Most of the currently commercially available ELISA-based assay formats have a limit of detection (LOD) ranging from 1 <sup>µ</sup>g·mL−<sup>1</sup> (equals ng·mL−<sup>1</sup> or ppb) up to 50 <sup>µ</sup>g·mL−<sup>1</sup> , for example, Ridascreen Aflatoxin B1 30/15 test kit (r-biopharm, Darmstadt, Germany) or AgraQuant Aflatoxin B1 ELISA test (Romer Labs, Butzbach, Germany). Nevertheless, due to the requirement for laboratory equipment, their usability for on-site testing is very limited. In such cases, commercially available LFA have the highest potential for fast and cost-effective on-site mycotoxin testing. However, most of these assay systems provide only qualitative results with LODs ranging from 4 ppb to 5 ppb, as achieved with RIDA®Quick Aflatoxin test kit(r-biopharm, Darmstadt, Germany) or AFB1 (Aflatoxin B1) lateral flow assay kit (Elabscience, Texas, USA), respectively. Primarily due to matrix interference effects, these assays can result in false-positive detection [22,23].

A novel, portable magnetic immunodetection approach has been described in previous studies [24–28]. The detection of human and plant pathogens, as well as various proteins by sensing and quantifying superparamagnetic particles (MP) with the help of a portable magnetic reader (Figure 1A), have been successfully demonstrated. This device can be operated using a conventional external power adapter or a portable battery, allowing an on-site readout without electrical infrastructure. For this immunomagnetic detection approach, MPs were functionalized with monoclonal antibodies directed against target molecules, retained in a sandwich-based manner within an immunofiltration column, and can be detected by means of frequency mixing magnetic detection (FMMD) technology [29].

In this technique, magnetic particles are subjected to two sinusoidal magnetic fields of different frequencies generated by two excitation coils, which are schematically shown within the measurement head in Figure 1B. Here, MPs are exposed to a low- and high-frequency magnetic field, so-called driving frequency, generated by the outermost coil, and excitation frequency, generated by the middle-positioned coil. The low frequency with 61 Hz (f2) has an amplitude of a few millitesla, resulting in alternating positive and negative magnetic saturation of superparamagnetic particles oscillating with a frequency of 2f<sup>2</sup> of 122 Hz [29]. The high-frequency magnetic excitation field

(f1) with 49 kHz probes the magnetization state of the superparamagnetic particles and yields an iron oxide dose-dependent signal when the low-frequency driving field is close to zero. Finally, the resulting mixing frequency signal of f<sup>1</sup> + 2f<sup>2</sup> can be demodulated and detected by the innermost coil, composed of two adjacent sections, so-called detection coil (upper one) and reference coil (lower one). Those sections differ only in the winding-orientation of coils. With this clock- and counterclockwise orientation yielding induced voltages of opposite sign, the directly induced excitation field can be canceled out. By placing the sample carrying the MPs in the detection head, the resulting signal is amplified, measured, and directly visualized at the touchscreen of the handheld, portable magnetic reader (Figure 1). Based on a calibration curve, the detected signal can be attributed to the amount of analyte within the sample.

**Figure 1.** (**A**) Handheld, portable frequency mixing-based readout device and (**B**) schematic cross-section of detection head composed of driving coil providing the low driving frequency (f<sup>2</sup> ), the excitation coil providing the high excitation frequency (f<sup>1</sup> ) and detection unit based on a detection coil detecting the resulting mixing frequency signal of magnetic particles (MPs; f<sup>1</sup> + 2f<sup>2</sup> ) together with the directly induced signal and the reference coil detecting only the directly induced signal. The finally resulting measuring signal does not contain the directly induced excitation due to the opposite winding direction of detection and reference coil.

Motivated by the above-described drawbacks of currently used analytical methods for sensitive on-field testing, the aim of this study was to develop a novel, highly sensitive, and portable assay based on competitive magnetic immunodetection (cMID) and FMMD. The sensitivity should be comparable to a laboratory-based ELISA. Hence, initially, a competitive ELISA (cELISA) was established, serving as a reference method. Assay parameters, such as the used coating and antibody concentrations, were optimized to reach a sufficient sensitivity for the detection of aflatoxin B1. Afterward, the cMID assay was established using the same optimization strategy in combination with further evaluation of the required amount of nanoparticles. The basic principle of cMID is the use of biotinylated antibodies, which can be enriched within the coated immunofiltration column by a competitive binding reaction depending on the amount of pre-captured mycotoxin. By flushing magnetic particles functionalized with streptavidin through the column by gravity flow, particles can bind to retained antibodies and subsequently be detected using FMMD. Figure 2 visualizes the basic cMID principle (Figure 2A) and the competitive binding reaction within the column with the corresponding measuring signal (Figure 2B).

**Figure 2.** Schematic overview of the competitive magnetic immunodetection principle. (**A**) Immunofiltration column coated with aflatoxin B1-BSA mycotoxin conjugate with bound biotinylated monoclonal antibodies targeting aflatoxin B1. Magnetic particles functionalized with streptavidin bind to antibodies and can be detected by FMMD. (**B**) After pre-incubation of biotinylated, monoclonal antibodies with serially diluted free aflatoxin B1, the sample is flushed through an aflatoxin B1-BSA coated immunofiltration column. Non-saturated antibodies bind to the coated antigen and are retained within the matrix. The higher the mycotoxin content within the sample, the more antibodies are saturated and are flushed through the column. Afterward, streptavidin-functionalized magnetic particles are applied onto the column, bind to retained antibodies and can be detected using FMMD.
