Palladium Nanoparticles-Based Fluorescence Resonance Energy Transfer Aptasensor for Highly Sensitive Detection of Aflatoxin M1 in Milk

A highly sensitive aptasensor for aflatoxin M1 (AFM1) detection was constructed based on fluorescence resonance energy transfer (FRET) between 5-carboxyfluorescein (FAM) and palladium nanoparticles (PdNPs). PdNPs (33 nm) were synthesized through a seed-mediated growth method and exhibited broad and strong absorption in the whole ultraviolet-visible (UV-Vis) range. The strong coordination interaction between nitrogen functional groups of the AFM1 aptamer and PdNPs brought FAM and PdNPs in close proximity, which resulted in the fluorescence quenching of FAM to a maximum extent of 95%. The non-specific fluorescence quenching caused by PdNPs towards fluorescein was negligible. After the introduction of AFM1 into the FAM-AFM1 aptamer-PdNPs FRET system, the AFM1 aptamer preferentially combined with AFM1 accompanied by conformational change, which greatly weakened the coordination interaction between the AFM1 aptamer and PdNPs. Thus, fluorescence recovery of FAM was observed and a linear relationship between the fluorescence recovery and the concentration of AFM1 was obtained in the range of 5–150 pg/mL in aqueous buffer with the detection limit of 1.5 pg/mL. AFM1 detection was also realized in milk samples with a linear detection range from 6 pg/mL to 150 pg/mL. The highly sensitive FRET aptasensor with simple configuration shows promising prospect in detecting a variety of food contaminants.


Introduction
Aflatoxins (AFs), which are highly toxic mycotoxins produced by Aspergillus parasiticus, Aspergillus flavus and Aspergillus nomius (rarely), present in a wide range of food and feed commodities [1,2]. The exposure of aflatoxin B 1 (AFB 1 )-contaminated feed to lactating mammals will lead to the conversion of AFB 1 into aflatoxin M 1 (AFM 1 ) through hydroxylation under liver cytochrome P450 catalysis [3,4]. AFM 1 , which have intense hepatotoxic and carcinogenic effects and have been designated as group1 carcinogen by the International Agency for Research on Cancer (IARC) of the World Health Organization (WHO) in 2002 [5], will subsequently be secreted in the milk of lactating mammals. As AFM 1 is quite stable during general pasteurization and storage process and will not be destroyed until the temperature exceeds 299 • C, it will enter human body through milk hybridization between ssDNA and its complementary chains occurred, the coordination effect was greatly weakened as fewer heteroatoms were exposed to PdNPs in the double helix structure of DNA. In our design, the strong coordination effect between the 5 -FAM-labeled AFM 1 aptamer and PdNPs brought the fluorescence donor FAM close to the fluorescence acceptor PdNPs, which resulted in the occurrence of FRET, and the fluorescence quenching of FAM was observed. After AFM 1 was introduced into the FAM-AFM 1 aptamer-PdNPs FRET system, AFM 1 aptamer preferentially bound to AFM 1 accompanied with its conformational change, which largely weakened the coordination effect between the AFM 1 aptamer and PdNPs. Thus FAM was separated from PdNPs and the FRET process was blocked. Therefore, the fluorescence recovery of FAM was observed and the degree of fluorescence recovery was in a positive AFM 1 concentration-dependent manner.

Construction of the FRET Aptasensor for AFM1
The AFM1 aptasensor was constructed based on aptamer-bridged FRET between FAM and PdNPs, as shown in Scheme 1. It has been reported that there was a strong coordination effect between nitrogen functional groups of single-stranded DNA (ssDNA) and PdNPs [32,33]. However, after the hybridization between ssDNA and its complementary chains occurred, the coordination effect was greatly weakened as fewer heteroatoms were exposed to PdNPs in the double helix structure of DNA. In our design, the strong coordination effect between the 5′-FAM-labeled AFM1 aptamer and PdNPs brought the fluorescence donor FAM close to the fluorescence acceptor PdNPs, which resulted in the occurrence of FRET, and the fluorescence quenching of FAM was observed. After AFM1 was introduced into the FAM-AFM1 aptamer-PdNPs FRET system, AFM1 aptamer preferentially bound to AFM1 accompanied with its conformational change, which largely weakened the coordination effect between the AFM1 aptamer and PdNPs. Thus FAM was separated from PdNPs and the FRET process was blocked. Therefore, the fluorescence recovery of FAM was observed and the degree of fluorescence recovery was in a positive AFM1 concentration-dependent manner. Scheme 1. Schematic illustration of the biosensor for aflatoxin M1 (AFM1) detection based on aptamer-bridged fluorescence resonance energy transfer (FRET) from 5-carboxyfluorescein (FAM) to palladium nanoparticles (PdNPs).

Properties Characterization of the Energy Acceptor
As the 5′-FAM-labeled AFM1 aptamer was negatively charged, sodium citrate-modified PdNPs with negative charge were used in this biosensor to avoid the side effect caused by electrostatic attraction which would also bring FAM close to PdNPs. Firstly, 12 nm Pd seeds (Figure 1b) whose light absorption was centered in the ultraviolet region and was very weak in the visible range ( Figure 1a) were synthesized using a sodium ascorbate reduction method. Then, larger PdNPs with an average diameter of 33 nm (Figure 1d) were synthesized on the basis of the 12 nm Pd seeds according to a seed-mediated growth method. The UV-Vis absorption spectrum of the 33 nm PdNPs in Figure 1c clearly showed that it exhibited strong absorption in nearly the whole UV-Vis spectral range, which overlaps well with the emission spectrum of FAM, which was essential for FRET occurrence between FAM and PdNPs. Scheme 1. Schematic illustration of the biosensor for aflatoxin M 1 (AFM 1 ) detection based on aptamer-bridged fluorescence resonance energy transfer (FRET) from 5-carboxyfluorescein (FAM) to palladium nanoparticles (PdNPs).

Properties Characterization of the Energy Acceptor
As the 5 -FAM-labeled AFM 1 aptamer was negatively charged, sodium citrate-modified PdNPs with negative charge were used in this biosensor to avoid the side effect caused by electrostatic attraction which would also bring FAM close to PdNPs. Firstly, 12 nm Pd seeds (Figure 1b) whose light absorption was centered in the ultraviolet region and was very weak in the visible range ( Figure 1a) were synthesized using a sodium ascorbate reduction method. Then, larger PdNPs with an average diameter of 33 nm (Figure 1d) were synthesized on the basis of the 12 nm Pd seeds according to a seed-mediated growth method. The UV-Vis absorption spectrum of the 33 nm PdNPs in Figure 1c clearly showed that it exhibited strong absorption in nearly the whole UV-Vis spectral range, which overlaps well with the emission spectrum of FAM, which was essential for FRET occurrence between FAM and PdNPs.

Construction of the AFM1 Aptasensor
To investigate the energy transfer efficiency between the FAM donor and PdNPs acceptor pair, an increasing concentration of 33 nm PdNPs were added into a fixed amount of 5′-FAM-labeled AFM1 aptamer (80 nM). After incubation in HEPES buffer (20 mM, pH = 7.0) containing 5 mM KCl and 5 mM MgCl2 for a short while, a PdNPs concentration-dependent fluorescence quenching phenomenon of 5′-FAM-labeled AFM1 aptamer was observed with the maximum quenching efficiency reaching 95%, as indicated in Figure 2a. In order to investigate the non-specific fluorescence quenching caused by PdNPs towards fluorescein dye, fluorescein dye at a final concentration of 80 nM was mixed with PdNPs (0.060 mg/mL) in HEPES buffer for 1 h and then the fluorescence intensity of fluorescein was measured. From Figure 2b it clearly indicated that the non-specific fluorescence quenching caused by PdNPs towards fluorescein could be eliminated. Therefore, the effective fluorescence quenching of FAM caused by PdNPs was ascribed to the strong coordination effect between nitrogen functional groups of the AFM1 aptamer and PdNPs, which brought FAM close to PdNPs resulting the occurrence of FRET. The time dependence of fluorescence quenching efficiency indicated in Figure 2c suggested that it only took 30 min to reach the quenching equilibrium. In the following fluorescence recovery experiments, In order to ensure reaching the quenching equilibrium and obtain stable fluorescence signal, 1 h incubation time was chosen for the fluorescence quenching experiment.

Construction of the AFM 1 Aptasensor
To investigate the energy transfer efficiency between the FAM donor and PdNPs acceptor pair, an increasing concentration of 33 nm PdNPs were added into a fixed amount of 5 -FAM-labeled AFM 1 aptamer (80 nM). After incubation in HEPES buffer (20 mM, pH = 7.0) containing 5 mM KCl and 5 mM MgCl 2 for a short while, a PdNPs concentration-dependent fluorescence quenching phenomenon of 5 -FAM-labeled AFM 1 aptamer was observed with the maximum quenching efficiency reaching 95%, as indicated in Figure 2a. In order to investigate the non-specific fluorescence quenching caused by PdNPs towards fluorescein dye, fluorescein dye at a final concentration of 80 nM was mixed with PdNPs (0.060 mg/mL) in HEPES buffer for 1 h and then the fluorescence intensity of fluorescein was measured. From Figure 2b it clearly indicated that the non-specific fluorescence quenching caused by PdNPs towards fluorescein could be eliminated. Therefore, the effective fluorescence quenching of FAM caused by PdNPs was ascribed to the strong coordination effect between nitrogen functional groups of the AFM 1 aptamer and PdNPs, which brought FAM close to PdNPs resulting the occurrence of FRET. The time dependence of fluorescence quenching efficiency indicated in Figure 2c suggested that it only took 30 min to reach the quenching equilibrium. In the following fluorescence recovery experiments, In order to ensure reaching the quenching equilibrium and obtain stable fluorescence signal, 1 h incubation time was chosen for the fluorescence quenching experiment.

AFM1 Detection in Aqueous Buffer Solution
As illustrated in Scheme 1, after AFM1 was introduced into the 5′-FAM-AFM1 aptamer-PdNPs FRET system in the HEPES buffer, the AFM1 aptamer preferentially bound to AFM1 accompanied by its conformational change, which greatly weakened the coordination effect between the AFM1 aptamer and PdNPs. Therefore, FAM was separated from PdNPs and the FRET process was inhibited. Meanwhile, the fluorescence of FAM was restored and the degree of fluorescence recovery was in an AFM1 concentration-dependent manner, as indicated in Figure 3a. A linear relationship between the fluorescence recovery of FAM and the concentration of AFM1 in the range from 5 pg/mL to 150 pg/mL was obtained in the HEPES buffer, with the detection limit of 1.5 pg/mL (calculated as the concentration corresponding to three times of the standard deviation of the background signal from seven independent measurements) ( Figure 3b). Compared to the previously reported structure, switching aptamer-based FRET assay for AFM1 detection with a linear detection range from 25 ng/kg to 2000 ng/kg [25], and the time-resolved fluorescent competitive immunochromatographic assay for AFM1 detection in milk with a linear dynamic range of 0.1-2.0 ng/mL [34], the sensitivity of the present sensor is significantly improved, which shows great potential to detect lower concentration of AFM1 in milk samples. The performance improvement of this FRET largely relied on the excellent fluorescence quenching ability of PdNPs towards FAM, with almost negligible non-specific fluorescence quenching. Other interfering toxins, including AFB1, OTA, ZEN, FB1 and T-2 toxin, were added individually into the FAM-AFM1 aptamer-PdNPs FRET system in the place of AFM1 under the same experimental procedures to examine the specificity of this FRET biosensor for AFM1. It can be seen from Figure 4 that the interference toxins all cause negligible fluorescence variation of FAM compared to AFM1, which firmly indicated the excellent specificity of this

AFM 1 Detection in Aqueous Buffer Solution
As illustrated in Scheme 1, after AFM 1 was introduced into the 5 -FAM-AFM 1 aptamer-PdNPs FRET system in the HEPES buffer, the AFM 1 aptamer preferentially bound to AFM 1 accompanied by its conformational change, which greatly weakened the coordination effect between the AFM 1 aptamer and PdNPs. Therefore, FAM was separated from PdNPs and the FRET process was inhibited. Meanwhile, the fluorescence of FAM was restored and the degree of fluorescence recovery was in an AFM 1 concentration-dependent manner, as indicated in Figure 3a. A linear relationship between the fluorescence recovery of FAM and the concentration of AFM 1 in the range from 5 pg/mL to 150 pg/mL was obtained in the HEPES buffer, with the detection limit of 1.5 pg/mL (calculated as the concentration corresponding to three times of the standard deviation of the background signal from seven independent measurements) (Figure 3b). Compared to the previously reported structure, switching aptamer-based FRET assay for AFM 1 detection with a linear detection range from 25 ng/kg to 2000 ng/kg [25], and the time-resolved fluorescent competitive immunochromatographic assay for AFM 1 detection in milk with a linear dynamic range of 0.1-2.0 ng/mL [34], the sensitivity of the present sensor is significantly improved, which shows great potential to detect lower concentration of AFM 1 in milk samples. The performance improvement of this FRET largely relied on the excellent fluorescence quenching ability of PdNPs towards FAM, with almost negligible non-specific fluorescence quenching. Other interfering toxins, including AFB 1 , OTA, ZEN, FB 1 and T-2 toxin, were added individually into the FAM-AFM 1 aptamer-PdNPs FRET system in the place of AFM 1 under the same experimental procedures to examine the specificity of this FRET biosensor for AFM 1 . It can be seen from Figure 4 that the interference toxins all cause negligible fluorescence variation of FAM compared to AFM 1 , which firmly indicated the excellent specificity of this developed FRET biosensor towards AFM 1 as a result of the high binding affinity between AFM 1 aptamer and AFM 1 . developed FRET biosensor towards AFM1 as a result of the high binding affinity between AFM1 aptamer and AFM1.

AFM1 Detection in Milk Samples
In order to ensure consumption safety and human health, it is very important to monitor the concentration of AFM1 in milk. In this paper, AFM1 detection was also realized in 100-fold diluted milk sample with HEPES buffer under the same experimental procedures as that in the aqueous buffer solution. It can be seen from Figure 5a that the fluorescence of FAM was restored in a AFM1 concentration-dependent manner. And the degree of fluorescence restoration was linear related to the concentration of AFM1 in the range from 6 pg/mL to 150 pg/mL, with a detection limit of 1.8 pg/mL (calculated as the concentration corresponding to three times of the standard deviation of the background signal from seven independent measurements) (Figure 5b). The relatively narrower linear range and higher detection limit may be ascribed to the complexity of the milk sample. developed FRET biosensor towards AFM1 as a result of the high binding affinity between AFM1 aptamer and AFM1.

AFM1 Detection in Milk Samples
In order to ensure consumption safety and human health, it is very important to monitor the concentration of AFM1 in milk. In this paper, AFM1 detection was also realized in 100-fold diluted milk sample with HEPES buffer under the same experimental procedures as that in the aqueous buffer solution. It can be seen from Figure 5a that the fluorescence of FAM was restored in a AFM1 concentration-dependent manner. And the degree of fluorescence restoration was linear related to the concentration of AFM1 in the range from 6 pg/mL to 150 pg/mL, with a detection limit of 1.8 pg/mL (calculated as the concentration corresponding to three times of the standard deviation of the background signal from seven independent measurements) (Figure 5b). The relatively narrower linear range and higher detection limit may be ascribed to the complexity of the milk sample.

AFM 1 Detection in Milk Samples
In order to ensure consumption safety and human health, it is very important to monitor the concentration of AFM 1 in milk. In this paper, AFM 1 detection was also realized in 100-fold diluted milk sample with HEPES buffer under the same experimental procedures as that in the aqueous buffer solution. It can be seen from Figure 5a that the fluorescence of FAM was restored in a AFM 1 concentration-dependent manner. And the degree of fluorescence restoration was linear related to the concentration of AFM 1 in the range from 6 pg/mL to 150 pg/mL, with a detection limit of 1.8 pg/mL (calculated as the concentration corresponding to three times of the standard deviation of the background signal from seven independent measurements) (Figure 5b). The relatively narrower linear range and higher detection limit may be ascribed to the complexity of the milk sample. Standard addition experiments were conducted to examine the feasibility of this AFM 1 biosensor in practical AFM 1 -free milk samples. The satisfactory recoveries from 92% to 106.5% in Table 1 convincingly demonstrates that this FRET biosensor based on the efficient fluorescence resonance energy transfer between FAM and PdNPs has great potential in practical application.

2017, 9, 318 7 of 11
Standard addition experiments were conducted to examine the feasibility of this AFM1 biosensor in practical AFM1-free milk samples. The satisfactory recoveries from 92% to 106.5% in Table 1 convincingly demonstrates that this FRET biosensor based on the efficient fluorescence resonance energy transfer between FAM and PdNPs has great potential in practical application.

Conclusions
In summary, a highly sensitive FRET aptasensor for AFM1 detection has been constructed based on the excellent fluorescence quenching ability of PdNPs towards FAM with negligible non-specific fluorescence quenching. The application of AFM1 aptamer with high affinity and specificity towards AFM1 also contributes to the good performance of this biosensor in both aqueous buffer solution and milk samples. In consideration of its simple configuration and operation, the homogeneous FRET aptasensor can be widely used to detect a variety of food contaminants, such as other biotoxins in the future.

Conclusions
In summary, a highly sensitive FRET aptasensor for AFM 1 detection has been constructed based on the excellent fluorescence quenching ability of PdNPs towards FAM with negligible non-specific fluorescence quenching. The application of AFM 1 aptamer with high affinity and specificity towards AFM 1 also contributes to the good performance of this biosensor in both aqueous buffer solution and milk samples. In consideration of its simple configuration and operation, the homogeneous FRET aptasensor can be widely used to detect a variety of food contaminants, such as other biotoxins in the future.

Instrumentation
The size and morphology of palladium seeds and larger PdNPs were characterized by a FEI Tecnai G2 F30 transmission electron microscope with an acceleration voltage of 200 kV. The UV-vis absorption measurements were conducted on a Thermo-Spectronic Unicam UV500 spectrometer (Thermo Spectronic, Waltham, MA, USA). The fluorescence spectra were recorded on a HITACHI F-4500 fluorescence spectrometer (HITACHI, Tokyo, Japan).

Synthesis of Sodium Citrate Capped Palladium Seeds
Palladium seeds were synthesized according to a reported procedure [35]. 20% freshly prepared aqueous solutions of sodium citrate (100 µL) and 1% Na 2 PdCl 4 in water (735 µL) were both added into 47 mL ultrapure water. After the solution was heated to boiling, 0.1% hot sodium ascorbate (2.5 mL) was introduced quickly into the mixture. Boiling under reflux was continued for another 30 min. Then the solution was cooled down to room temperature naturally and filtered through a 0.22 um Millipore membrane filter.

The Synthesis of 33 nm PdNPs
The synthesis of 33 nm PdNPs was accomplished by a seed-mediated growth method reported by Lu et al. [36]. 10 mL aqueous solution of H 2 PdCl 4 with a concentration of 1 mM was placed in a 50 mL round-bottom flask. And 3 mL of the synthesized palladium seeds were added. Then an excess amount of aqueous solution of ascorbic acid (100 mM, 1.2 mL) was introduced into the above solutions under extensive stirring. The color of the solution readily changed from pale yellow to a dark brown color, which suggested the formation of larger PdNPs. The resultant solution was stirred for another 5 min at room temperature. Next, the obtained PdNPs were centrifuged and washed with ultrapure water for three times. Finally, the products were redispersed in 3 mL of ultrapure water for further use.

Quenching Measurements
The concentration of PdNPs used in the FRET system were optimized against a fixed concentration of 5 -FAM-labeled AFM 1 aptamer, that is, 80 nM. For optimization, the concentrations of PdNPs were set at 0, 0.015 mg/mL, 0.030 mg/mL, 0.045 mg/mL, 0.060 mg/mL, 0.075 mg/mL, 0.090 mg/mL. They were incubated in HEPES buffer (20 mM, pH = 7.0) containing 5 mM KCl and 5 mM MgCl 2 for 1 h and then the fluorescence intensities were recorded under excitation at 480 nm and emission at 520 nm. The time-dependent fluorescence intensities were obtained by incubating a fixed concentration of 5 -FAM-labeled AFM1 aptamer (80 nM) with PdNPs in a concentration of 0.060 mg/mL from 1 min to 60 min.

AFM 1 Detection in Aqueous Buffer Solution
In a typical FRET analysis process, various concentrations of AFM 1 (0, 5 pg/mL, 20 pg/mL, 40 pg/mL, 80 pg/mL, 100 pg/mL, 150 pg/mL, 300 pg/mL, 600 pg/mL, 900 pg/mL, 1200 pg/mL) were first mixed with the 5 -FAM-labeled AFM 1 aptamer (80 nM) in HEPES buffer, respectively, and the mixtures were all incubated at room temperature for 2 h. Afterwards PdNPs was added individually into the above mixtures with an ultimate concentration of 0.060 mg/mL, followed by incubation for another 1 h at room temperature. Finally, the fluorescence intensity of the reaction mixture was recorded under excitation at 480 nm and emission at 520 nm. To examine the specificity of the FRET aptasensor, a list of other mycotoxins including AFB 1 , OTA, ZEN, FB 1 and T-2 toxin were added into the FAM-AFM 1 aptamer-PdNPs FRET system in place of AFM 1 following the same experimental procedures.

AFM 1 Detection in Milk Samples
The milk samples (7% fat content) were purchased from the local market of Wuhan, China. They were first centrifuged at 5000 rpm for 10 min at 25 • C to remove the fat and the supernatant were collected respectively. For the determination of AFM 1 in milk samples, the supernatant was 100-fold diluted with HEPES buffer without further processing, and the same assay procedure as in the HEPES buffer solution was followed. Standard addition method was adopted to determine the concentration of AFM 1 in AFM 1 -free milk samples.