The CF was acquired from Gleba (La Plata, Buenos Aires, Argentina) as a commercial formulation of 48% w/v and stored at 16°C before use. DeMan-Rogosa-Sharpe (MRS) broth and agar were purchased from Britania (Buenos Aires, Argentina). The kit for the identification of API CH-50 strains (REF 50300, API systems, BioMerieux^®, France) was stored at 4°C. The chemicals and organic solvents used in the different tests were of analytical grade and were acquired from Sigma-Aldrich (Merck - USA). The water used was deionized by membrane filtration (OSMOION 5, APEMA water purifier equipment). Fresh white cabbage (Brassica oleracea) was purchased in the local market in Mar del Plata, Argentina.

Two LAB strains, L. mesenteroides and L. paramesenteroides, were isolated from white cabbage (Brassica oleracea), and two reference strains of LAB, L. fermentum (CRL 220) and P. pentosaceus (CRL 922), provided by Reference Center for Lactobacilli (CERELA, Tucumán, Argentina), were used in this study.

Raw materials were washed with tap water and sliced. Ten grams of vegetable were suspended in 90 mL Butterfield’s phosphate sterile buffered dilution water (pH 7.2 ± 0.1). The standard pour plate technique, using MRS agar, was employed to isolate LAB strains, and plates were incubated for 48 h at 35°C under conditions of anaerobiosis [21]. The pure colonies were subjected to morphological and preliminary tests (catalase and oxidase test, nitrate reduction, indole production, gelatin and casein metabolization). Afterwards, LAB strains were re-isolated with MRS agar and broth, and finally identified as L. mesenteroides and L. paramesenteroides, using the kit API CH-50.

The reference strains and isolates from cabbage were grown overnight in 9 mL of MRS broth. Incubation was carried out at 35°C for 19 h. The cultures were centrifuged at 6,000 g for 10 min, and suspended in sterile saline solution (0.85% w/v). Then, the inoculum was adjusted to a McFarland standard of 2, which corresponds to a concentration of 6 × 10⁸ CFU mL^–1. One milliliter of this solution was re-suspended in 9.0 mL of 0.9% w/v sterilized saline solution, thus obtaining a 6 × 10⁷ CFU mL^–1 inoculum.

CF sterile solution was added to 20 mL MRS broth for a final concentration of 1.2 mg L^–1, and then vortexed homogenized for 30 s. The bacterial suspension was inoculated into the flask containing MRS medium supplemented with CF to achieve a concentration of 10⁵ CFU mL^–1. The reaction mixture was vortexed for 2 min to ensure a homogeneous distribution of LAB strain cells and CF, and then incubated for 24 h at 35°C. Aliquots of 1 mL were taken at 0, 6, 12, and 24 h in order to quantify the residual CF. At the same time, LAB counts were performed in order to evaluate the population over time. The standard pour plate technique using MRS agar was carried out, and the plates were incubated at 35°C for 48 h. A control containing MRS and CF was carried out without bacterial suspension.

Extraction and purification were performed according to Pegoraro et al. [22] (2016) with some modifications. One milliliter from the OPP degradation assay was centrifuged at 10,000 g for 5 min, and the supernatant (300 μL) was cleaned using a glass mini-column filled with 1 g sodium sulfate (Na₂SO₄) and 0.6 g of activated silica. n-Hexane (270504-1L, Sigma-Aldrich, PRO - ACS analysis) was used as elution solvent (5 mL final volume), and a 1 mL prewash was performed. The clean extract was evaporated to dryness using N₂ stream and finally, reconstituted with 1.0 mL of n-hexane for final analysis.

The CF was quantified by gas chromatography-mass spectrometry (GC-MS; Shimadzu CGMSQP2010 ULTRA-AOC20I) according to Zhang et al. [17] (2016) with light modifications. The column was an HP-5MS capillary column (15 m × 250 μm, 0.25 μm). The flow rate of the carrier gas (99.99% helium gas) was maintained at 35.2 cm s^–1. The injection mode was split less. The injector temperature was set at 300°C. The interface and the ionization source were maintained at 280 and 230°C, respectively. Ionization was carried out by electron impact at 70 eV. Purified extracts (2.0 μL) were analyzed by a temperature ramp programmed as follows: held at 80°C for 0.5 min, ramped up to 230°C at 25°C/min, and then up to 300°C at a rate of 35°C/min and kept for 5 min, with a total run time of 13.5 min. Using a CF standard solution on the GC-MS system allowed the selection of three ion mass-to-charge ratio (m/z) in the selected ion monitoring (SIM) mode. The ion m/z = 314 was used for quantification and, m/z = 199 and m/z = 197 for identification. The instrument detection limit (IDL) was 5 μg/L, defined as a signal to noise ratio of 3:1 of the amount of pesticide, and was determined by calculating the signal to noise ratio for the CF in the lowest calibration standard. The method detection limit (MDL) was determined in accordance with the USEPA. To achieve this, we prepared seven solutions, which were spiked with CF at 15 μg/L. The resulting data were used to calculate the standard deviation of the measured concentrations. The MDL was then calculated by multiplying this standard deviation by the appropriate one-sided Student’s t-value for six degrees of freedom and a 99% confidence level (t = 3.14), resulting in 7.2 μg/L.

The products of CF degradation were also detected. GC-MS scan mode was employed to qualitatively distinguish these products, 3,5,6 trichloro-2-pyridinol (TCP) and diethyl thiophosphate (DETP), and they were identified by comparing the MS spectrum with data reported at the National Institute of Standards and Technology (NIST) Library. The GC-MS conditions used were according to the method of Ahir et al. [23] (2020) with slight modifications. The injection mode was split less. The oven temperature was held initially at 130°C for 1 min, programmed from 130 to 290°C at 15°C/min, and held at 290°C for 4 min. Run time was 15.67 min. Subsequently, the main m/z peaks were corroborated by SIM mode.

Cells were counted for growth evaluation at each time by the spread-plate method in MRS agar. Serial dilutions were spread on culture dishes and then incubated at 35°C for 48 h. Results were expressed as log₁₀ CFU mL⁻¹.

A solution of MRS broth, inoculated with 5% v/v of LAB strains overnight culture and 1.2 mg L^–1 of CF, was incubated for 20 h at 35°C. Five milliliters of carbonate/bicarbonate buffer (10.0 mmol L^–1, pH 10.5) was added and stirred for 1 min. The reaction mixture was treated by sonication 10 times (10 s with 15 s intervals) to break the bacterial cell wall and release the enzyme, then the suspension was centrifuged for 15 min at 10,000 g according to Sasikala et al. [24] (2012). The supernatant (crude enzyme solution) was used for the phosphatase activity assay.

Phosphatase activity was evaluated using the colorless reagent p-nitrophenyl phosphate (p-NPP) that produces p-nitrophenol (p-NP) as a yellow product [25]. The methodology was based on Zhang et al. [18] (2014) with minor modifications. The reaction mixture consisted of 1.0 mL of 5.0 mmol L^–1 p-NPP [prepared in carbonate buffer with 1.0 mmol L^–1 MgCl₂·6H₂O, Mg (II) acting as a cofactor], 1.0 mL of the carbonate buffer, 1.0 mL of the crude enzyme solution and distilled water, with a fixed final volume of 5.0 mL. After a reaction time of 60 min at 35°C, 3.0 mL of buffer termination (0.1 mol L^–1 NaOH with 5.0 mmol L^–1 EDTA) was added to stop the reaction. The generated p-NP was quantified by UV-VIS spectroscopy (Agilent 8435) at 400 nm. One unit (U) of phosphatase activity was defined as the amount of enzyme liberating 1 μmol of p-NP per minute at 35°C [18]. The reaction control was the assay without incubation.

The phosphatase activity calculation was made from Equation 1 [26]. The activity coefficient of p-NP at 400 nm was calculated from a calibration curve with a p-NP standard (Sigma-Aldrich Lot.: 13621 PD-166) in carbonate buffer 10.0 mmol L^–1 pH 10.5.

(1) U   m L - 1 =   A p - N P   -   A c o n t r o l E   L   V t o t a l   10 6   1 t i m e

where A_p-NP = absorbance of the incubation reaction mixture; A_control = absorbance of the reaction mixture without incubation; E = molar extinction coefficient of p-NP; L = optical path of the cell; V_total = total reaction volume; 10⁶ = correction factor; and time = incubation time.

The interaction between the CF and a phosphatase from L. mesenteroides was analyzed through molecular docking investigation. The study aimed to identify the most probable binding sites, determining their position, binding energy, and inhibition constant. A structural prediction of L. mesenteroides hydrolase was used to search for potential CF binding sites.

The trials were performed in triplicate and subjected to statistical tests to evaluate variability. Kinetic parameters were studied by linear regression. Data analysis was performed using Statistica 8.0 software. Statistical significance between groups was assessed by comparing the difference in their mean values relative to the pooled standard error using independent samples t-test. The resulting test statistic was used to derive the corresponding P-value, which determined whether the observed differences were statistically significant (P < 0.05). All analyses were performed using standard parametric procedures unless otherwise indicated.

Several pesticide residue testing methods use dichloromethane in a liquid-liquid extraction [17]. However, chlorinated hydrocarbons should be replaced with less harmful alternatives for ecological and toxicological reasons [18, 28]. In this work, a solid-liquid partition method was used, employing a homemade mini-column assembled with a Pasteur pipette. Unlike liquid-liquid extraction, only a few milliliters of n-hexane were required as the eluent. The stationary phase consisted of activated silica, which acted as a purifying agent for organic and biological residues [28], and sodium sulfate acted as a desiccant. For quantification, 2 μL of the purified sample was injected into the GC-MS. The concentration of CF at each time point was calculated from a standard calibration curve.

Figure 1 shows the degradation rate of CF by four LAB strains. All strains exhibited the ability to degrade CF, achieving degradation rates ranging from 37 ± 2% to 97 ± 5% after 24 h of incubation. L. mesenteroides reached 80% degradation after just 12 h, while P. pentosaceus, L. paramesenteroides, and L. fermentum reached 34%, 27%, and 23%, respectively.

The CF degradation products were consistent with previous reports, in which TCP and DETP were described as the major metabolites formed during CF biodegradation [29–31]. Based on a comparison of GC-MS spectral data with literature reports on CF hydrolysis, the peaks at m/z 199, 197, 171, 169, 134, and 107 confirmed the formation of TCP, detected at 7.2 min (Figure 2). DETP, the second major metabolite, was detected at 9.2 min through peaks at m/z 95, 141, and 78. Regarding their toxicity, TCP has been found in samples of different origins (milk, urine, and blood) [32]. Unfortunately, TCP, like other products formed through the transformation of pesticides, has no established biological exposure limits, making it difficult to interpret the results in terms of human health risk [33, 34].

Degradation kinetics followed a first-order model in all strains (Figure 3, Table 1). The degradation rate constants for L. paramesenteroides, L. fermentum, and P. pentosaceus were comparable to those reported for other LAB strains [18]. Zhang et al. [18] (2014) found similar values for L. plantarum and S. thermophilus in milk, which showed degradation constants of 0.0186 and 0.0197 h^–1 and degradation rates of 64 and 70%, respectively. In contrast, L. mesenteroides exhibited the highest rate constant and degradation efficiency, reaching 75% degradation after only 6 h of incubation.

As shown in Figures 1 and 3, after 24 h of incubation, the strains had a positive impact on the degradation of CF, with percentages of 97 ± 5%, 47 ± 3%, 42 ± 3%, and 37 ± 2% for L. mesenteroides, L. paramesenteroides, P. pentosaceus, and L. fermentum, respectively. These results support previous findings on the bioremediation potential of LAB strains commonly used in food processing [15, 18, 35].

The bacterial population growth during CF degradation is shown in Figure 4. All four strains exhibited tolerance to CF. L. mesenteroides, L. paramesenteroides, and P. pentosaceus showed significant population increases during the first 6 h, with counts rising by 3 log cycles. After 12 h, increases of 4–6 log cycles were observed. However, a gradual decline was noted during the final 12 h, suggesting that CF imposes stress on bacterial cells [36].

L. fermentum displayed a longer lag phase, with little change in population during the first 6 h (Figure 4). After that, its population increased continuously, consistent with delayed adaptation and enzyme synthesis. This behavior also matched its lower CF degradation rate. Notably, no significant correlation was found between bacterial growth and CF degradation in any strain.

Tolerance to CF may be linked to enzymatic activity. LAB strains are known to hydrolyze OPPs using enzymes such as phosphatases [1, 15, 18]. GC-MS analysis supported this, showing a tentative identification of TCP and DETP as hydrolysis products. Table 2 summarizes phosphatase activity and degradation constants for each strain. L. mesenteroides showed significantly higher degradation and phosphatase activity (P < 0.05) compared to the other strains.

Phosphatase activity values were 0.2130 ± 0.0053, 0.2600 ± 0.0097, and 0.2640 ± 0.0015 U mL^–1 for L. paramesenteroides, L. fermentum, and P. pentosaceus, which showed CF degradation ranging between 37 and 47%. These results were similar to those of Zhang et al. [18] (2014), who reported phosphatase activities of 0.266–0.357 U mL^–1 in LAB. Similarly, Zhou and Zhao [35] (2015) reported 0.081–0.175 U mL^–1 in yogurt cultures. In contrast, L. mesenteroides reached 0.889 ± 0.018 U mL^–1, suggesting greater metabolic versatility and stress resistance.

Docking analysis was performed using aligned protein structures and an optimized CF ligand. Although the 4ZRS protein had a greater number of residues, the structures overlapped satisfactorily. Binding interactions in pocket 1 involved van der Waals forces and hydrogen bonds with Arg59 and Lys66, while in pocket 3, they involved Lys41 and Ala42. These results differ from Haque et al. [27] (2020), who identified Ser128, Asp256, and His285 as key residues, likely due to methodological differences in pocket identification.

Table 3 summarizes the pocket parameters, binding energies, and inhibition constants. Pockets 1 and 3 showed similar free energies (ΔG ≈ –5.77 to –5.79 kcal mol^–1), but pocket 1 contained a canonical hydrolase motif (Gly127-Glu128-Ser129-Ser130-Gly131), suggesting that catalysis likely occurs there. Hydrogen bonding also contributes to ligand stability [37, 38]. Figure 5 shows the loop (in orange) containing the Gly-Glu-Ser-Ser-Gly motif, where the dotted line represents the hydrogen-bond distance between the hydrogen of Ser129 and the oxygen of the thiophosphate group (2.99 Å). Pocket 3 may represent an allosteric site.

Although L. brevis showed stronger binding energy in previous reports (ΔG = –7.7 kcal mol^–1) [37], the inhibition constant for L. mesenteroides was higher (> 50 μM), indicating greater resistance to CF and better degradation performance. This aligns with the 97% CF degradation observed in L. mesenteroides, compared to 76.5% for L. brevis [39].

This study confirmed that the ability of LAB strains to remove OPPs is positively correlated with phosphatase production. Indeed, TCP and DETP were tentatively identified by GC-MS as the main degradation metabolites. Significant removal of CF, measured in terms of degradation kinetic parameters, was observed in the four tested LAB strains. Among them, L. mesenteroides isolated from vegetables efficiently degraded CF and reached values above 90%, indicating a metabolism better suited to OPP stress conditions. Notably, the binding energy between CF and phosphatase was exergonic, indicating that the interaction occurs spontaneously. These findings imply that enzyme catalysis plays a crucial role in the degradation of CF by L. mesenteroides. This study is one of the key supporting studies needed to assess the remediation efficiency of bacteria commonly found in food. In this regard, L. mesenteroides could be applied post-harvest to remove OPP residues.

Finally, this study provides experimental and computational evidence supporting the role of LAB phosphatases in CF degradation. The combination of microbial screening, enzymatic assays, and in silico analyses offers a comprehensive approach to understanding pesticide biodegradation by food-grade bacteria.