**3. Conclusions**

Results showed that potentially probiotic *L. fermentum* 111, *L. paracasei* 108, and *L. plantarum* 49 isolated from fruit processing by-products are capable of binding AFM1 in vitro when assayed as either viable or non-viable cells. The recovery of AFM1 from bacterial cell complexes varied with the examined isolate and contact time. Non-viable cells had a higher capability for retaining AFM1 after 1 or 24 h of incubation. These results indicate that *Lactobacillus* isolates recovered from fruit with performance compatible to use as probiotics could have a satisfactory aflatoxin binding capacity, which could be exploited as a biological tool for the detoxification of foods and feeds, particularly, for the removal and restoration of AFM1 to safe levels. Further studies are needed to investigate the mechanisms involved in removal of AFM1 by these isolates and possible factors affecting the stability of formed complexes, including when exposed to conditions mimicking the human gastrointestinal tract.

#### **4. Materials and Methods**

#### *4.1. Chemicals, Bacterial Isolates, and Inoculum Preparation*

The AFM1 standard was obtained from Sigma Aldrich (St. Louis, MO, USA). Highperformance liquid chromatography (HPLC) grade solvents were obtained from Merck (Darmstadt, Germany).

The isolates *Lactobacillus plantarum* 49, *L. fermentum* 111, and *L. paracasei* 108 were examined separately for the removal of AFM1. These isolates were recovered from fruit processing by-products, identified with a partial 16S rRNA gene sequence analysis and characterized as potential candidates for use as probiotics [17]. Stocks were stored at −20 ◦C in de Man, Rogosa, and Sharpe (MRS) broth (HiMedia, Mumbai, India) with glycerol (20 mL/100 mL; Sigma-Aldrich, St. Louis, MO, USA). Working cultures were maintained aerobically on MRS agar (HiMedia, Mumbai, India) at 4 ◦C and transferred to a new media monthly. Prior to use in assays, each isolate was cultivated anaerobically (Anaerobic System Anaerogen, Oxoid, Hampshire, UK) in MRS broth at 37 ◦C for 20–24 h (to reach the stationary growth phase), harvested by centrifugation (4500× *g*, 15 min, 4 ◦C), washed twice, and resuspended in phosphate buffer solution (PBS; 50 mM K2HPO4/KH2PO4; pH 6.9) to obtain cell suspensions with an optical density reading at 660 nm (OD660) of 0.5. This suspension had viable counts of approximately 1.1 × 10<sup>9</sup> CFU/mL for each isolate when plated in MRS agar.

#### *4.2. Evaluation of AFM1 Removal and Recovery of AFM1 from Bacterial Cells*

The capability of examined *Lactobacillus* isolates to remove AFM1 in PBS was assessed with viable and non-viable bacterial cell suspensions. To obtain non-viable bacterial cells, *Lactobacillus* cell suspensions were inactivated by boiling at 100 ◦C for 1 h. No visible colonies were found when heat-treated cell suspensions (named heat-killed cells) were

plated onto MRS agar and followed by anaerobic incubation (using Anaerobic System Anaerogen, Oxoid, Hampshire, UK) for 48 h. For testing the AFM1 removal capability, 1 mL of test isolate suspension (pure culture of viable and heat-killed cells) was mixed with 1.5 mL of PBS, previously spiked with 0.15 μg/mL AFM1, and incubated aerobically at 37 ◦C [28]. After 1 and 24 h of incubation, the mixture was centrifuged (1500× *g*, 15 min, 4 ◦C) and the AFM1 content in the supernatant was determined by HPLC, as detailed in Section 4.3.

Cell pellets collected from each monitored incubation period (contact time) were evaluated for the recovery of AFM1 from cell complexes. Obtained pellets were washed with 1.5 mL of fresh PBS, the cells were re-pelleted (1500× *g*, 15 min, 4 ◦C), and supernatant was collected for the quantification of released AFM1 [18]. For each isolate, a positive control consisting of free cells suspended in PBS with 0.15 μg/mL AFM1, and a negative control, consisting of bacterial cells (viable or heat-killed), suspended in PBS were used.

#### *4.3. Quantification of AFM1*

The quantification of AFM1 in supernatants was done with high-performance liquid chromatography (HPLC) using a Shimadzu (Prominense, Tokyo, Japan) HPLC system, equipped with an auto sampler SIL 20A HT (Prominense, Shimadzu, Tokyo, Japan), fluorescence detector RF-20A (Prominense, Shimadzu, Tokyo, Japan), an LC-20AT pump (Prominense, Shimadzu, Tokyo, Japan), oven CTO-20A (Prominense, Shimadzu, <sup>T</sup>óquio, Japão), a CBM-20A controller (Prominense, Shimadzu, Tokyo, Japan), a CLC-ODS (M) reverse phase column (4.6 × 150 mm; Shim-Pack, Prominense, Shimadzu, Tokyo, Japan) and pre-column G-ODS-4 (1.0 × 4.0 mm; Shim-Pack, Prominense, Shimadzu, Tokyo, Japan).

Chromatographic conditions were the same as those described in a previous study [7]. Excitation and emission wavelengths were 366 and 428 nm, and the injection volume was 20 μL. The mobile phase was water:methanol:acetonitrile (6:2:2) and the flow rate was 1 mL/min. The calibration curve was constructed using six concentrations of AFM1 standard diluted in acetonitrile (20–60 ng/mL), performed in triplicate. From this analysis, the equation *y* = 2E+07*x* + 873,267 (*r*2 > 0.99) was obtained. The limit of detection (LOD) and limit of quantification (LOQ) were estimated based on Resolution n◦ 899 of the Brazilian Agency for Health Surveillance [29]. The LOD and LOQ of AFM1 were 0.20 and 0.67 ng/mL, respectively.

The percentage of AFM1 removed by each isolate was determined with the Equation (1) [22,27,30]:

100 × [1 − (peak area of chromatographic peak of sample)/area of positive control chromatographic peak)]. (1)

### *4.4. Statistical Analysis*

Assays were done in triplicate in three independent experiments (repetitions). A Kolmogorov–Smirnov normality test was run to assess whether obtained results had normal distribution. Results (average data ± standard deviation) were submitted to a one-way analysis of variance (ANOVA), followed by Tukey's test, considering a *p* value of ≤ 0.05 for significance. Statistical analyses were done with IBM SPSS Statistics 20 (Armonk, NY, USA).

**Author Contributions:** Conceptualization, P.O.d.C., H.I.F.M., E.L.d.S.; methodology, P.O.d.C., H.I.F.M., E.L.d.S., C.J.d.M., Y.M.N., J.F.T.; validation, P.O.d.C., H.I.F.M., C.J.d.M., Y.M.N., J.F.T.; investigation, P.O.d.C., C.J.d.M., Y.M.N.; writing—original draft preparation, P.O.d.C., H.I.F.M., E.L.d.S.; writing— review and editing; supervision, H.I.F.M.; funding acquisition, H.I.F.M., E.L.d.S., J.F.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by CAPES (Brazil), finance code 001.

**Acknowledgments:** Authors thank the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, Brazil) for partial funding of this research.

#### **Conflicts of Interest:** The authors declare no conflict of interest.
