Aptamer-Based Fluorescence Quenching Approach for Detection of Aflatoxin M1 in Milk

Aflatoxin M1 (AFM1), one of the most toxic mycotoxins, is a feed and food contaminant of global concern. In this study, we developed a fast and simple method for detection of AFM1 based on a structure-switching signaling aptamer. This aptasensor is based on the change in fluorescence signal due to formation of an AFM1/aptamer complex. To generate the aptasensor, the specific aptamer was modified with FAM (carboxyfluorescein), and their complementary DNAs (cDNA) were modified with a carboxytetramethylrhodamine (TAMRA) quenching group. In the absence of AFM1, the aptamers were hybridized with cDNA, resulting in quenching of the aptamer fluorescence due to the proximity of the aptamer’s fluorophore to the quenching group on the cDNA. On the other hand, in the presence of AFM1, a structural switch in the aptamer was induced by formation of an AFM1/aptamer complex. Changes in the structure of the aptamer led to the release of the cDNA, causing the generation of a fluorescence signal. Thus, AFM1 concentrations could be quantitatively monitored based on the changes in fluorescences. Under optimized conditions, this assay exhibited a linear response to AFM1 in the range of 1–100 ng/mL and a limit of detection of 0.5 ng/mL was calculated. This proposed aptasensor was applied to milk samples spiked with a dilution series of AFM1, yielding satisfactory recoveries from 93.4 to 101.3%. These results demonstrated that this detection technique could be useful for high-throughput and quantitative determination of mycotoxin levels in milk and dairy products.


INTRODUCTION
Mycotoxins are highly toxic secondary metabolites and produced by several fungal genera, including Aspergillus, Fusarium, and Penicillium (Li et al., 2012;Liu et al., 2015). To date, more than 400 mycotoxins have been identified (Selvaraj et al., 2015). Among them, aflatoxins (AFs), deoxynivalenol (DON), ochratoxin (OTA), fumonisin B 1 (FB 1 ), zearalenone (ZEN), and T-2 toxin (a trichothecene mycotoxin) are common mycotoxins (Yang et al., 2017). As the most toxic AF, aflatoxin B 1 (AFB 1 ) is classified as a group 1 carcinogen by the International Agency for Research on Cancer (IARC) (Iqbal et al., 2015). In addition, aflatoxin M 1 (AFM 1 ) is a hydroxylated metabolite of AFB 1 , and approximately 2-6% of AFB 1 in contaminated animal feed is converted to AFM 1 in milk (Abera et al., 2019). AFM 1 mainly causes chronic hepatitis and liver tumors in humans, which was classified as group 2 and changed to group 1 carcinogen by IARC (Karczmarczyk et al., 2016). Due to the serious harm caused to human and animal health by mycotoxins, many countries and international organizations (e.g., the European Union and the Joint FAO/ WHO Expert Committee on Food Additives) have established relevant standards for limiting AFM 1 contamination in foods to extremely low levels Song et al., 2014). In Europe, the maximum residue level of AFM 1 in milk is 0.05 μg/kg (European Commission, 2006). In China and the United States, the maximum tolerated level for AFM 1 in milk is 0.5 μg/kg (Busman et al., 2015;Mao et al., 2015). Once raw milk is contaminated with AFM 1 , conventional heat treatment (pasteurization or ultra-high temperature) for milk and dairy products is not easy to degrade AFM 1 (Beitollahi et al., 2020).
Despite strict regulation and control of AFM 1 levels in milk, AFM 1 pollution still occurs frequently Zheng et al., 2017). Consequently, methods for detecting AFM 1 in milk are attracting increasing public attention. Many methods for detecting AFM 1 have been developed, including highperformance liquid chromatography (HPLC) (Mao et al., 2015;Pietri et al., 2016), enzyme-linked immunosorbent assay (ELISA) (Li et al., 2009;Kav et al., 2011), immunochromatography (Busman et al., 2015), thin-layer chromatography (TLC) (Sassahara et al., 2005), and liquid chromatography coupled with mass spectroscopy (LC-MS) (Vaclavik et al., 2010). However, instrument-based analytical methods suffer from several drawbacks, including the requirements of technical expertize and complicated, time-consuming sample pretreatment, making these approaches unsuitable for field operation (Eivazzadeh et al., 2016). Moreover, antibody-based immunoassay systems require high-quality antibodies, which are often too expensive, unstable, or susceptible to degradation and denaturation for field applications (Istamboulia et al., 2016). In the past few years, several groups of researchers have pursued AFM 1 detection technology on immunoassays combined with fluorescence , electrochemistry (Parker and Tothill, 2009) or chemiluminescence (Vdovenko et al., 2014). Biosensor-based aptamers have attracted increasing attention due to their numerous advantages, including high sensitivity and specificity, easy synthesis and modification, labeling, reusability, and portability (Rhouati et al., 2016). Malhotra et al. used laminated magnetic beads to select this AFM 1 aptamer by exponential enrichment (SELEX technique). The aptamer has a unique structure that encompasses two overlapping stem loops without conserved motif or G-quadruplex, which has lowest dissociation constant value (K d 35 nM). The researchers confirmed that the aptamer has good affinity against AFM 1 . In addition, the specific aptamer has previously been applied for AFM 1 sensing for our research team (Malhotra et al., 2014). Guo X et al. developed a biosensor based on this aptamer to detect AFM 1 and combined the advantages of strong recognition ability of the aptamer to AFM 1 and excellent amplification efficiency of the RT-qPCR technique to improve sensitivity (Guo et al., 2016). Recently, Kordasht H et al. used cyclic voltammetry and differential pulse voltammetry techniques to a bioassay based on this aptamer for the determination of AFM 1 (Kordasht et al., 2019).
In particular, structure-switching fluorescence-quenching signal adaptation platforms are considered promising approaches for detection of various biomolecules due to their high sensitivity and specificity, and the fluorescence-quenching approach has been used for the detection of AFB 1 in our previous research . Sharma et al. developed a structure switching aptamer assay for AFM 1 detection. However, the selectivity of the aptamer for AFM 1 was not clear since only irrelevant OTA and AFB 1 were chosen as interferences to study the cross-reaction (Sharma et al., 2016). Cross-activity tests between other toxins (especially the structural analogs AFB 1 , AFB 2 , AFG 1 , and AFG 2 ) should be carried out to answer the question whether the aptasensor is suitable for quantifying the AFM 1 concentration in real samples. In addition, the coexistence of multiple mycotoxins in foods is a common and complex phenomenon. Therefore, the effect of mix-mycotoxins on the specificity of the aptasensor should be investigated.
Thus, we developed the biosensor for the detection of AFM 1 based on the specific aptamer, as well as the study of optimal aptamer complementary DNA and structure conversion signal. In the specificity experiment part, we detected the cross-reaction of AFB 1 , AFB 2 , AFG 1 , AFG 2 , OTA, ZEN, FB 1 , and mixmycotoxins for the aptasensor performance. In addition, this proposed aptasensor was confirmed to be simple, fast, convenient, economical and effective. The method represents a new practical application of aptamers for the detection of AFM 1 in milk and other dairy products.

Materials and Reagents
AFM 1 , AFB 1 , aflatoxin G 1 (AFG 1 ), aflatoxin G 2 (AFG 2 ), OTA, ZEN, and FB 1 were purchased from Qingdao Pribolab Pte. Ltd. (Qingdao, China). The Tris-HCl buffer (pH 7.0) used in this study contained 10 mM 2-amino-2-(hydroxymethyl)-1,3propanediol (Tris), 120 mM sodium chloride (NaCl), 5 mM potassium chloride (KCl), and 20 mM anhydrous calcium chloride (CaCl 2 ). All chemical reagents were obtained from Shanghai Chemical Reagent Company (Shanghai, China). Phosphate-buffered saline (PBS) was obtained from Thermo Fisher Scientific (Waltham, MA, USA). Liquid whole milk samples were purchased from a local supermarket (Mengniu Dairy, Beijing, China). Distilled water was purified using a Milli-Q purification system. Measurement of fluorescence was performed in Tris-HCl buffer. AFM 1 -aptamer labeled with fluorescein (FAM) and its complementary sequences labeled with carboxytetramethylrhodamine (TAMRA) were synthesized by Wuhan GeneCreate Biological Engineering (Wuhan, China) and purified by HPLC. Their sequences were as follows: AFM Aptamer-Based Fluorescence-Quenching Assay for AFM 1 Detection First, the AFM 1 aptamer and cDNA were dissolved and diluted in Tris-HCl buffer, and 500 μL AFM 1 aptamers (10 nM) were mixed with 500 μL cDNA. To optimization, the ratios of AFM 1 aptamer: cDNA were set at 1:1, 1:2, 1:3, 1:4, or 1:5 (unless otherwise stated, experiments used a molar ratio of 1:2). Then, the mixture of AFM 1 aptamer and cDNA was heated at 88°C for 5 min and allowed to stand at room temperature for at least 30 min. Next, 500 μL of different concentrations of AFM 1 standard solution was added to the mixture while vortexing. The final reaction volume was 1.5 mL. Similarly, 500 μL Tris-HCl buffer was added to the mixture of AFM 1 aptamer and cDNA while vortexing without addition of AFM 1 and then the fluorescence intensity of fluorescence quenching was measured. Fluorescence intensities at excitation/emission wavelengths of 495/520 nm were recorded on a F-7000 fluorescence spectrophotometer (Hitachi, Japan). To avoid the influence of aflatoxin background fluorescence signal, we performed background correction by measuring the fluorescence of a control containing AFM 1 and buffer without FAM-modified AFM 1 aptamer or cDNA .

Specificity Analysis
To evaluate the ability of this assay to detect AFM 1 with high specificity, seven mycotoxins structurally related to AFM 1 (AFB 1 , AFB 2 , AFG 1 , AFG 2 , OTA, ZEN, and FB 1 ) were selected for testing. All mycotoxins were used at the same concentration (40 ng/mL). All other detection conditions were identical to those used in the AFM 1 procedure, allowing comparison of fluorescence intensity for all toxins texted.

Real Sample Analysis
The assay was validated for quantitative detection of AFM 1 in liquid milk samples that were spiked with known concentrations of AFM 1 . When the concentration of AFM 1 in milk samples was in the range of 1-100 ng/mL, the following pretreatment methods were used. Samples were spiked with AFM 1 at a concentration of 0.5, 5, 10, or 20 ng/mL. Each sample was accurately weighed (0.5 mL) and adding into 10 mL centrifuge tubes, and then 2.5 mL of 70% methanol in water was added to extract AFM 1 from the sample. The entire mixture was vortexed for 5 min on a Vortex-Genie (Scientific Industries, Bohemia, NY, United States), and then centrifuged at 10,000 × rpm for 10 min. The supernatant was collected, filtered through a 0.22-μm filter, and concentrated to 0.5 mL using a N1-Automatic Nitrogen Concentrator (Preekem Scientific, Shanghai, China). Finally, each of the residues was redissolved in 2 mL of aqueous methanol solution (5%, v/v) and stored at 4°C until needed. Five replicates were measured for each sample to assess the accuracy of the assay. When the concentration of AFM 1 in milk samples was in the range of 0-1.0 ng/mL, samples were pretreated to increase sample quantity. The samples accurately weighed (25 mL), spiked with AFM 1 at a concentration of 0.5 ng/mL, heated to approximately 35-37°C, and then centrifuged at 4,000 rpm for 10 min. A 25-mL volume of milk sample was passed through the AFLAPREP ® M column (Biopharm, Germany) by gravity flow to facilitate toxin binding. The column was washed with 20 mL of PBS at a flow rate of approximately 5 mL per min. Toxin was eluted from the column with 1.25 mL of methanol:acetonitrile (40:60 v/v) at a flow rate of one drop per second; following elution, 1.25 mL of water was run through the column and collected in the same vial to yield a total volume of 2.5 mL, which was concentrated to 1 mL in a N1-Automatic Nitrogen Concentrator. After concentration, 1.5 mL Tris-HCl buffer was added. Subsequently, the fluorescence signal was determined. Five replicates of each sample were measured to evaluate the accuracy of the method.

Design of the Aptamer-Based Fluorescence Quenching Platform
A schematic diagram of the fluorimetric design for AFM 1 detection is shown in Figure 1A. Theoretically, the sensor functionality was based on changes in the conformation of the aptamer after formation of the AFM 1 /aptamer complex and the resultant change in fluorescence signal. In our sensor system, AFM 1 aptamers were modified with FAM, and 7-mer cDNAs with the TAMRA quenching group. In the absence of AFM 1 , the aptamers were hybridized with cDNA, resulting in quenching of their fluorescence due to the proximity of TAMRA-cDNA (Sharma et al., 2016). Upon the AFM 1 addition, a structural switch was induced in the AFM 1 aptamer, leading to formation of an AFM 1 /aptamer complex (Seok et al., 2015). Changes in the structure of the aptamer caused the release of cDNA from the aptamer, and ultimately to the recovery of aptamer fluorescence. Therefore, the concentration of AFM 1 and the change in fluorescence intensity after the reaction were positively correlated, and the fluorescence change could be used to quantify the level of AFM 1 . To confirm that AFM 1 could dissociate the aptamer-cDNA complex and restore the aptamer fluorescence, we added 150 ng/ mL AFM 1 to 10 nM aptamer and 20 nM cDNA in Tris-HCl buffer. Figure 1B shows that fluorescence intensity increased at least 11-fold after the addition of AFM 1 . Moreover, these findings confirm that the covalently labeled fluorophore

Optimization of the cDNA Sequence
Next, we optimized the number of cDNA bases to stabilize the aptamer/cDNA duplex and minimize background fluorescence in the absence of AFM 1 . Accordingly, to optimize the sensor, we designed a series of complementary single-stranded DNA sequences. In this experiment, 10 nM AFM 1 aptamer was added to 20 nM of cDNAs 1-7, incubated for 5 min at 88°C, and then cooled to room temperature for 30 min to prepare add 150 ng/mL AFM 1 . In Figure 2A, black columns represent fluorescence background after hybridization of AFM 1 aptamers with cDNA1-cDNA7 in the absence of AFM 1 , and the blue columns represent fluorescence intensity after addition of 150 ng/mL AFM 1 .
The results revealed that cDNA5-7 could effectively quench AFM 1 aptamer fluorescence and decrease background signal in the absence of AFM 1 . Due to the short lengths of cDNAs 1-4, hybridization with AFM1 aptamer was unstable, resulting in a higher background signal. When cDNA5 was hybridized with the aptamer, a significant increase in fluorescence (F/F 0 ) was observed after addition of AFM 1 , whereas the longer cDNA sequences (cDNA6 and cDNA7) effectively quenched fluorescence ( Figure 2B). This may be because of cDNA5, which contains more nucleotides than the shorter sequences, and could promote the formation of cDNA-aptamer duplexes while limiting AFM 1induced structural transitions of the aptamer. Therefore, we considered cDNA5 to be the optimal single-stranded cDNA sequence, and used it in subsequent experiments.

Optimization of cDNA5 Concentration
To further improve the performance of the sensor, we optimized the cDNA5 concentration. In this experiment, 10 nM aptamer was mixed with cDNA5 at a molar ratio of 1:1, 1:2, 1:3, 1:4, or 1:5. Subsequently, 150 ng/mL AFM 1 was added, and the change in fluorescence signal was monitored. As shown in Figure 2C, the maximum fluorescence enhancement (F/F 0 ) was highest at a cDNA5 concentration of 20 nM. Thus, even at higher

Detection methods Transduction principle LOD References
Instrumental methods HPLC 6 × 10 −3 ng/mL concentrations of cDNA5, the formation of stable cDNA5/ aptamer duplexes generated low background fluorescence. When the concentration was higher than 20 nM (twice the AFM 1 aptamer concentration), it affected the optimal AFM 1 interaction, thereby limiting the increase in fluorescence. Accordingly, for subsequent sensing experiments, we determined the optimum concentration of cDNA5 to be 20 nM.

Stability of the Reaction
To generate the desired fluorescence signal response, we analyzed the reaction time of AFM 1 with the aptamer. As shown in Figure 2D, the aptamer and AFM 1 combined rapidly: the concentration of AFM 1 was detected at 1 min, and the fluorescence value remained stable for up to 30 min. Therefore, the fluorescence sensor had acceptable stability. We determined that a reaction time of 15 min was optimal for this assay.

Quantitative Determination of AFM 1
For all experimental conditions, the excitation and emission wavelengths were 495 and 520 nm, respectively. The sensor detected different concentrations of AFM 1 (0,1,5,10,20,40,60,80,100,150,200,300, and 400 ng/mL) with corresponding fluorescence signal intensities. As the concentrations of AFM 1 in the reaction system were increased, the released amounts of cDNA were also increased, leading to quenching groups located away from the fluorophore and an increase in fluorescence value. As shown in Figure 3A, the fluorescence intensity increased with the increased AFM 1 concentration, and fluorescence intensity was maximal at an AFM 1 concentration of 150 ng/mL. The concentration of AFM 1 was linear in the range of 1-100 ng/mL ( Figure 3B), and the linear regression equation was F 1.665C + 7.652 (R 2 0.9963), in which F denotes the fluorescence intensity and C denotes the AFM 1 concentration. The limit of detection (LOD) was 0.5 ng/mL with a signal-to-noize ratio of 3 (S/N 3). Although the proposed detection method gave lower or comparable detection limits compared to other instruments and rapid screening methods (Table 1), this sensor provides a novel, simple, inexpensive and portable idea for detecting AFM 1 .

Specificity of the Assay
Specificity is an important measure of an aptamer sensor. To evaluate the specificity of our aptasensors, we selected seven mycotoxins with structural similarities to AFM 1 (AFB 1 , AFB 2 , AFG 1 , AFG 2 , OTA, ZEN, and FB 1 ) as controls. Each of these mycotoxins was examined at the same concentration (40 ng/mL) under the same experimental conditions used for AFM 1 . As shown in Figure 4, neither the individual mycotoxins nor a mixture of all compounds other than AFM 1 (MIX1) significantly altered fluorescence relative to the no-mycotoxin control (p > 0.05). A mixture of all toxins including AFM 1 (MIX2) had a slightly weaker effect than AFM 1 individually, but the difference was not significant. These results confirmed that this fluorescence biosensor did not react mycotoxins other than AFM 1 , indicating that the method was highly specific for detection of AFM 1 .

Milk Sample Analysis
To further verify practicality and feasibility, we applied our method to the detection of AFM 1 in two different brands of spiked milk samples. Four different concentrations of AFM 1 (0.5, 5, 10, and 20 ng/mL) were spiked into milk samples, and sample preparation was performed as described in Real Sample Analysis section, and five replicates were tested for each concentration. As shown in Table 2, the recoveries of the spiked milk samples were in the range 93.4-101.3% (n 5), respectively, indicating that the FIGURE 4 | Fluorescence intensity in the absence (control) and presence of 40 ng/mL mycotoxins: AFM 1 , AFB 1 , AFB 2 , AFG 1 , AFG 2 , OTA, ZEN, and FB 1 , MIX1 (AFB 1 , AFB 2 , AFG 1 , AFG 2 , OTA, ZEN, and FB 1 ), and MIX2 (AFM 1 , AFB 1 , AFB 2 , AFG 1 , AFG 2 , OTA, ZEN, and FB 1 ). Excitation/ emission λ ex /λ em 495/520 nm. Reaction conditions: 10 nM AFM1 aptamer and 20 nm cDNA in Tris-HCl buffer (10 mM, pH 7.0). aptasensor could be used as a quantitative method for AFM 1 detection in real milk samples.

CONCLUSION
In summary, we successfully developed an aptamer-based fluorescence quenching assay for direct detection of AFM 1 and confirmed the feasibility of this method for use on milk samples. Under optimal conditions, the aptasensor exhibited linear increases in fluorescence intensity in response to increasing concentrations of AFM 1 from 1 to 100 ng/mL, with a LOD of 0.5 ng/mL. This assay was highly specific for AFM 1 and did not detect other toxins with similar structures, including AFB 1 , AFB 2 , AFG 1 , AFG 2 , OTA, ZEA, and FB 1 . Recoveries of AFM 1 from spiked milk samples were in the range of 93.4-101.3% (n 5). This assay offers a novel, simple, rapid, specific, convenient, and cost-effective approach for detection of AFM 1 , and provides a means of quantitatively determining mycotoxins in dairy products. In future studies, the method of target recycling of signal amplification to improve the sensitivity of aptamer-based fluorescence biosensors will be conducted. At the same time, we clearly understand that the interference from the milk matrix to optical measurement is typically very challenging. Therefore, strict sample pre-treatment was required for eliminating the effects of the background due to milk on the aptasensor detection, which is also the challenge and future development trend of nucleic acid aptamer detection of mycotoxins.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENTS
All participants were informed about the aim of the study and gave written consent.