Fe3O4@PDA/MIL‐101(Cr) as magnetic solid‐phase extraction sorbent for mycotoxins in licorice prior to ultrahigh‐performance liquid chromatography‐tandem mass spectrometry analysis

Abstract Magnetic solid‐phase extraction (MSPE) strategy based on the Fe3O4@PDA/MIL‐101(Cr) has been proposed to separate and purify five common mycotoxins in licorice, including aflatoxin B1, aflatoxin G1, sterigmatocystin, zearalenone, and ochratoxin A. Integrating the MSPE and solid–liquid extraction/partitioning, a modified QuEChERS was established to adapt to the complex licorice samples. The Fe3O4@PDA/MIL‐101(Cr) was successfully synthesized and characterized by Fourier transform infrared spectroscopy (FT‐IR), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and nitrogen adsorption–desorption isotherms. Sorbents with superior advantages for exclusion of matrix interference and extraction of target analytes in a short time were obtained, according to their ability of magnetic separation, high surface area (287.75 m2/g), large pore volume (0.61 cm3/g), and nanosized structure with mesopores. Prior to analysis with ultrahigh‐performance liquid chromatography‐tandem mass spectrometry (UHPLC‐MS/MS), several key parameters that would affect the sorbents’ extraction efficiency were extensively investigated. Under the optimized conditions, the practicality of the developed method for analysis of mycotoxins in licorice samples was confirmed by adequate linearity (R 2 ≥ 0.9967), high sensitivity (LODs and LOQs, respectively, in the ranges 0.01–0.09 and 0.02–0.30 μg/kg), acceptable recovery (78.53%–116.28%), satisfactory reusability, and good interbatch precision of the sorbents (RSDs in the ranges 6.70%–11.20% and 6.02%–10.35%, respectively). The results indicated that the established method was feasible and reliable for the environment‐friendly and rapid screening of mycotoxins in complex licorice samples.


| INTRODUC TI ON
Licorice is the root of Glycyrrhiza uralensis Fisch. or Glycyrrhiza glabra L., Leguminosae, which is one of the ancient and worldwide popular herbs native to southern Europe and parts of Asia. With multifunctional ingredients, licorice has beneficial applications in both fields of food and medicine (Peng et al., 2021). It has been on the list of homologous materials of medicine and food with great value, according to the National Health Commission of China. For the value as food, licorice was mainly used in confectionery sectors to produce candies due to the sweet-tasting compound (Herrera et al., 2009).
As for medicinal aspects, licorice has been regarded as the "guide drug" in many traditional Chinese medicine (TCM) prescriptions for thousands of years in China (Wang et al., 2013). Meanwhile, the history of its therapeutic applications has been documented in Europe (Fiore et al., 2005). With the exploitation of its extracts and active ingredients, licorice is extensively applied in food additives, health supplements, food flavoring agents, and cosmetics, more than sweets, herbal remedies, and pharmaceuticals (Hayashi & Sudo, 2009). However, the quality and safety of licorice has encountered the challenge of mycotoxin contamination, which is a common problem in food safety (Fung et al., 2018). Licorice is susceptible to mycotoxin contamination since licorice belongs to crops and can be easily infected by toxigenic fungi during postharvest handling, transportation, and storage processes (Atyn & Twaruek, 2020).
Mycotoxins are a heterogeneous group of toxic metabolites mainly produced by filamentous fungi species of Aspergillus, Penicillium, Alternaria, and Fusarium (Palumbo et al., 2020). They are stable and hard to remove or destroy during the general cooking process. Until now, more than 400 mycotoxins have been discovered with great structural and toxic diversity, some of which are of great concern in food safety, such as aflatoxins (AFs), ochratoxin A (OTA), and zearalenone (ZEN). As is well known, the exposure of mycotoxins to human beings and animals could induce neurotoxic, carcinogenic, nephrotoxic, immunosuppressive, and estrogenic effects (Lee & Ryu, 2016). The existence of mycotoxins in food and agricultural commodities has been recognized as an adverse threat to human health and the economy. Therefore, it is of great importance to determine and monitor the contamination levels of mycotoxins in licorice. With multiple chemical constituents, licorice has an extremely complex matrix, which increases the difficulty of mycotoxin analysis Shakeri et al., 2018). To be specific, challenges mainly come from the interfering effect on extraction and purification efficiency by starches and polysaccharides as well as the matrix effect, which is defined as suppression or enhancement of the analyte response induced by the coextracted flavonoids, organic acids, or volatile oils (Wei et al., 2018).
Currently, liquid chromatography-tandem mass spectrometry (LC-MS/MS) has been widely used as a confirmatory method for mycotoxin analysis on account of its high sensitivity and compatibility with almost the whole range of compound polarities (Dominguez et al., 2020;He et al., 2019;Li et al., 2013). However, the application has been blocked by the matrix effect, which could impact quantification results of the LC-MS/MS approach, especially for samples with complex matrices. Therefore, reliable and accurate quantification of mycotoxin is subject to efficient sample preparation (Capriotti et al., 2019;Cho et al., 2019). As a critical procedure in instrumental analysis, appropriate sample extraction and clean-up are effective in reducing the matrix interferences with enrichment of target analytes. Immunoaffinity column and QuEChERS have been widely used as sample preparation methods for mycotoxin analysis (Wei et al., 2018;Zhang et al., 2018). Nevertheless, limitations of expensive cost, cross-reactivity, low recovery, or sensitivity to matrix types have restricted their use in some cases. As a quick, cheap, and efficient method in analytical chemistry, dispersive solid-phase extraction (d-SPE) has gained substantial attention in the analysis of mycotoxins in complex matrices Ran et al., 2017;Reinholds et al., 2019;Tanveer et al., 2020). Especially with the development of novel sorbents, magnetic sorbents are considered in combination with d-SPE, which is called magnetic solid-phase extraction (MSPE). MSPE can further simplify the d-SPE process and is a powerful tool for the application of micro/nanomaterials in sample preparation (Maya et al., 2017). Recently, metal-organic frameworks (MOFs) combined with magnetic nanoparticles have emerged as valuable sorbents for d-SPE due to their ultrahigh porosity, enormous surface area, tunable pore size, and the ability to realize functionalization and magnetic separation (Gao et al., 2020). Like other advanced materials, MOFs have been widely used for the determination of heavy metals, pesticides, and some other food and environmental pollutants. However, relatively limited research has been conducted to meet the challenge of the determination of trace-level mycotoxins in complex matrices. Comparatively, mycotoxin analysis with advanced materials as d-SPE sorbents is rather a new topic Li et al., 2021;Zhao et al., 2020). Challenges might ascribe to complex sample matrices, trace-level analytes, and diverse physicochemical properties of mycotoxins. Currently, improvements in mycotoxin analysis have been presented by the application of some advanced sorbents, such as carbon nanotubes, metallic nanoparticles, and graphene-based materials. It deserves to further investigate the fabrication of advanced sorbents and their application in mycotoxin analysis.
Comprehensively considering the property of mycotoxins and the mechanism of adsorption interaction of MOFs (Hasan & Jhung, 2015;Joseph et al., 2019), this study aims to synthesize
The mass spectrometry was equipped with an electrospray ionization source (ESI). Nitrogen was used as cone, nebulization, and desolvation gas. The MS system was operated with optimal parameters as follows: the source temperature and desolvation gas temperature were, respectively, set at 150 and 250℃. The cone gas flow was maintained at 150 L/h, and desolvation gas flow was 600 L/h. Multiple reaction monitoring (MRM) in positive mode (ESI + ) was developed for qualification and quantification of target mycotoxins.
For each mycotoxin, one precursor ion and two fragment ions were monitored. The cone voltage and capillary voltage of mycotoxins were optimized separately. The optimal MS parameters for analysis of target mycotoxins were listed in Table 1.

| Synthesis and characterization of Fe 3 O 4 @ PDA/MIL-101(Cr)
Initially, Fe 3 O 4 was prepared by the solvothermal method according to the reference (Shao et al., 2012).  To characterize the prepared Fe 3 O 4 @PDA/MIL-101(Cr), Fourier transform infrared spectroscopy, scanning electron microscopy, transmission electron microscopy, and specific surface area and porosity analyzer were applied to investigate the functional groups, morphology, and porous structure, respectively.

| Sample collection and pretreatment
Licorice samples were randomly collected from local pharmacies in Chengdu, China. All samples were separately ground and stored in sealed plastic bags below 4°C for further use. The blank samples were confirmed to be free of the target mycotoxins with the analytical method proposed in this work.
Sample powder (2.0 g) was weighed accurately in a 50 ml centrifuge tube. In a typical run, the sample was ultrasonicated for 30 min and centrifuged after being macerated with 20 ml acetonitrile/water (84/16, v/v) for 5 min (Han et al., 2010;Jiang et al., 2017;Zheng et al., 2014). To improve the partition of the mycotoxins into the organic phase, the obtained extraction solution was mixed vigorously for 30 s with 4 g anhydrous magnesium sulfate, 1 g anhydrous sodium chloride, 1 g trisodium citrate dihydrate, and 0.5 g disodium hydrogen citrate sesquihydrate, followed by another centrifugation.
Afterward, 5 ml of the resulting ACN-based supernatant was diluted with ultrapure water to decrease the concentration of organic solvent to 50%. Besides, the pH value of the solution was adjusted to 5.
Twenty milligram of the synthesized Fe 3 O 4 @PDA/MIL-101(Cr) was added into the sample solution obtained as abovementioned.
After being vortexed for 30 min, the mixture was separated by magnetism. Then, the supernatant was discarded, followed by desorption procedure, which was carried out with the following processes: 1 ml acetonitrile containing 2% formic acid was added into

| Optimization of MSPE procedure
To achieve an optimal separation and purification procedure, experimental parameters that could influence the extraction and desorp-

| Effect of pH and adsorption mechanism
The pH of the sample solution would determine the existing forms of the analytes as well as the surface-binding sites of the sorbent. As a consequence, pH has a significant influence on the extraction efficiency of the target mycotoxins. In this study, the effect of pH was investigated in the range from pH 3 to pH 9. In optimization of pH, other parameters were as follows: sorbent amount of 15 mg; adsorption time of 20 min; desorption solvent acetonitrile (2% formic acid); and desorption time of 5 min. As shown in Figure 2a, the highest recovery rates were obtained at pH 5, except for ZEN. Increasing with the pH value, ZEN would lose protons and become negative.
The increased recovery of ZEN might be attributed to the electrostatic attraction between the negatively charged analyte and the unsaturated Cr (III) sites of the sorbent. Thus, the main adsorption mechanism of ZEN on the sorbent could be speculated to be electrostatic interaction more than π-π stacking. As for STER and OTA, recovery increased from pH 3 to pH 5, while obviously decreased from pH 5 to pH 7, and increased again at pH 9. Considering the pKa value (8.38, Figure S3) of STER, the analyte will be in neutral form with good hydrophobicity below pH 6. In this way, high recovery was obtained due to the hydrophobic effect. However, protons would accumulate on the analyte under highly acidic conditions. Thus, poor adsorption recovery was obtained at pH 3 owing to the electrostatic repulsion between the analyte and the sorbent. It is speculated that the main adsorption mechanism of STER might be hydrophobic effect or hydrogen bonding in acidic conditions, while electrostatic interaction in alkaline conditions. In consideration of the pKa values of OTA, its adsorption behavior in the investigated pH range was speculated to be the result of the synergistic effect of electrostatic interaction and hydrogen bonding. As for AFB 1 and AFG 1 , speculation of the adsorption mechanism was dissimilar to that of other mycotoxins due to the fact that no dissociation equilibrium exists for the analytes. Nevertheless, the structures of aflatoxins have oxygen atoms with lone electrons, which would coordinate with H + in highly acidic conditions. Then, poor recovery was similarly obtained at pH 3 for electrostatic repulsion. Theoretically, the adsorption of AFB 1 and AFG 1 was possibly driven by π-π stacking, hydrogen bonding, and hydrophobic effect. When the pH value was more than 5, the carboxyl group on the sorbent was deprotonated along with the hard formation of hydrogen bond between the analyte and the sorbent, thus the adsorption recovery was relatively decreased. In comprehensive F I G U R E

| Effect of sorbent amount
The sorbent amount was evaluated in the range 5-25 mg. As shown in Figure 2b, it revealed that the sorbent amount had a significant impact on the recovery of target mycotoxins, especially when the sorbent amount was increased from 5 to 15 mg. In optimization of the sorbent amount, other parameters were as follows: pH 5; adsorption time of 20 min; desorption solvent acetonitrile (2% formic acid); and desorption time of 5 min. According to the result, the recovery of five target mycotoxins was obviously increased, with the sorbent amount varying from 5 to 15 mg. In comparison with the experiment using 15 mg of sorbent, an experiment with 20 mg of sorbent resulted in better recovery for AFB 1 , AFG 1 , STER, and ZEN.
As for OTA, a further increase in the sorbent amount nearly had no influence on its recovery. Furthermore, it was shown that 25 mg of sorbent induced a slight increase on the recovery of ZEN, while no obvious change on the recovery of other mycotoxins. Consequently, 20 mg of sorbent was a reasonable compromise to ensure satisfactory extraction efficiency and recovery.

| Effect of adsorption time
As the adsorption process is an equilibrium-based extraction procedure, adsorption time is another significant factor to impact the extraction efficiency and recovery of the mycotoxins. In this study, vortex mixing was employed in the adsorption process to facilitate the achievement of adsorption equilibrium. In optimization of the adsorption time, other parameters were as follows: pH 5; sorbent amount of 20 mg; desorption solvent acetonitrile (2% formic acid); and desorption time of 5 min. From the result in Figure 2c, the equilibrium of target mycotoxins was almost reached at an adsorption F I G U R E 2 Effect of (a) pH, (b) sorbent amount, (c) adsorption time, and (d) desorption solvent on the recovery (%) of target mycotoxins time of 30 min. As the adsorption time further increased, the recovery of target mycotoxins remained nearly the same. Therefore, 30 min was chosen as the optimal adsorption time.

| Effect of desorption solvent and time
To guarantee satisfactory desorption efficiency of the analytes from sorbent, desorption solvent and desorption time have been, respectively, investigated in this study. In optimization of the desorption procedure, other parameters were as follows: pH 5; sorbent amount of 20 mg; and adsorption time of 30 min.
According to the result of the pre-experiment, the addition of formic acid in desorption solvent could increase the elution efficiency of target mycotoxins in this study. Therefore, four kinds of desorption solvents were specifically considered, including 2% formic acid-methanol, 2% formic acid-acetonitrile, 2% formic acid-methanol/acetonitrile (v/v = 1/1), and 2% formic acid-2% methylbenzene-methanol/acetonitrile (v/v = 1/1). The result in Figure 2d indicated that the highest recovery rates were obtained when 2% formic acid-acetonitrile was used as the desorption solvent. Experiments with desorption time of 2 and 5 min were also compared, resulting in no obvious distinction. It is implied that 2 min was enough to obtain satisfactory recovery of the mycotoxins.

| Determination of the matrix effect
As a well-known problem for electrospray ionization, the matrix effect appeared as signal suppression or signal enhancement of analyte is unavoidable due to the co-elution of matrix interferants. To characterize the matrix effect of the proposed method, matrix-matched calibration curves were established by the addition of standard so-  the signal of STER was suppressed by 39%. In other words, the purified sample matrix had no significant influence on the response of AFB 1 , AFG 1 , ZEN, and OTA, while having a relatively prominent effect on the response of STER. Considering the fact that matrixmatched standards are commonly used to compensate matrix effect, matrix-matched calibration was essential for the accurate determination of target mycotoxins in this study.

| Method validation
To validate the proposed method, parameters that reflect the efficiency and feasibility of the method were estimated by investigating the analytical characteristics in terms of linearity, limits of detection (LODs), limits of quantification (LOQs), the reusability, and interbatch precision of the sorbents. The obtained results were listed in Table 3.  To further confirm the accuracy of the proposed method, recovery of target mycotoxins spiked in licorice samples at three different concentration levels was evaluated. Each concentration level was investigated as three replicates of the spiked samples, which were extracted, purified, and analyzed by the established method.
Specifically, recovery was calculated as the ratio of the actual response to the theoretical response on the basis of the matrixmatched calibration. The obtained results were summarized in

| Comparison of the proposed method with other methods
To evaluate the separation and purification effect of the synthesized nanomaterial, adsorption of the target mycotoxins in the licorice sample was compared between the synthesized Fe 3 O 4 @ PDA and Fe 3 O 4 @PDA/MIL-101(Cr). As presented in Figure 3, It can be clearly seen from  (Xing et al., 2016) MIL-101(Cr) were shown in Figure 4. After being processed by the salts to enhance the partition of the extracted solution, one part of the resulting ACN-based supernatant was diluted with water (1:1, v/v) and filtered through a 0.22 μm syringe filter for analysis. And the other part was diluted and treated with the MSPE procedure as described above. It has been clearly shown that the interference of sample matrix has been greatly reduced with the MSPE procedure, which resulted in better chromatographic separation.
To further evaluate the applicability of the proposed method, target mycotoxins in five commercial licorice samples were analyzed.
The result of real sample analysis showed that no other mycotoxins were detected in five licorice samples, except for OTA. The contamination levels of OTA were below the LOQ value of the method, which was far lower than the MRL established by the European Union (20 µg/kg).

| CON CLUS ION
In With the ability of rapid screening of mycotoxins in licorice samples, the proposed method suggests a potential application for the determination of multimycotoxins in samples with complex matrices. Since a few mycotoxins have been considered in this study, more work is still needed in the following studies.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available on request from the authors.