Cell wall component and mycotoxin moieties involved in binding of fumonisin B1 and B2 by lactic acid bacteria

Aims: The ability of lactic acid bacteria (LAB) to bind fumonisins B1 and B2 (FB1, FB2) in fermented foods and feeds and in the gastrointestinal tract could contribute to decrease their bioavailability and toxic effects on farm animals and humans. The aim of this work was to identify the bacterial cell wall component(s) and the functional group(s) of FB involved in the LAB–FB interaction. Methods and Results: The effect of physicochemical, enzymatic and genetic treatments of bacteria and the removal/inactivation of the functional groups of FB on toxin binding were evaluated. Treatments affecting the bacterial wall polysaccharides, lipids and proteins increased binding, while those degrading peptidoglycan (PG) partially decreased it. In addition, purified PG from Gram-positive bacteria bound FB in a manner analogue to that of intact LAB. For FB, tricarballylic acid (TCA) chains play a significant role in binding as hydrolysed FB had less affinity for LAB. Conclusions: Peptidoglycan and TCA are important components of LAB and FB, respectively, involved in the binding interaction. Significance and Impact of the Study: Lactic acid bacteria binding efficiency seems related to the peptide moiety structure of the PG. This information can be used to select probiotics with increased FB binding efficiency. Introduction Fumonisins, a structurally related mycotoxin group produced by Fusarium verticillioides and Fusarium proliferatum, are common contaminants of corn and corn-based products worldwide (Shephard et al. 1996). There are several identified fumonisins, but fumonisin B1 (FB1) and B2 (FB2) are the most important and constitute up to 70% of the fumonisins found in naturally contaminated foods and feeds. FB1 is the diester of propane-1,2,3-tricarboxylic acid (tricarballylic acid, TCA) and 2-amino-12,16dimethyl-3,5,10,14,15-pentahydroxyeicosane, in which the C14 and C15 hydroxyl groups are esterified with the terminal carboxyl group of TCA. FB2 is the C10 deoxy analogue of FB1, in which the corresponding stereogenic units on the icosane backbone have the same configurations (Fig. 1). Figure 1 Absolute configuration of fumonisin B1 (FB1) and B2 (FB2) Fumonisins B1 and B2 are phytotoxic to corn (Lamprecht et al. 1994), cytotoxic to various mammalian cell lines (Abbas et al. 1993) and FB1 is a carcinogen in rat liver and kidney (IARC 2002). The occurrence of these analogues in home-grown corn has been associated with an increased risk of esophageal cancer in humans (Shephard et al. 2000). FB1 is considered possible carcinogens to human and classified as class 2B (IARC 2002). These mycotoxins are the causal agent of two well described diseases in domestic animals: equine leukoencephalomacia (Riley et al. 1997) and porcine pulmonary edema syndrome (Harrison et al. 1990). In addition, they have also been associated with nephrotoxic, hepatotoxic and immunosupressing effects in various animal species (Morgavi and Riley 2007). The mechanism of action appears to involve mainly disruption of sphingolipid biosynthesis by the inhibition of the enzyme sphingosine N-acetyltransferase (ceramide synthase) (reviewed by Voss et al. 2007). FB are more toxic than their hydrolysed or N-acetylated derivatives (Gelderblom et al. 1993). The free amino group appears to play a specific role in the biological activity of fumonisins. Binding of FB by lactic acid bacteria (LAB) from fermented foods and feeds, and by LAB present in the gastrointestinal tract (GIT) could contribute to decrease the toxin bioavailability. This property could also decrease the exposure of intestinal mucosa to FB. Gut tissues exposed to FB have a diminished immune response and an altered barrier function against colonization by pathogenic Escherichia coli (Bouhet et al. 2004). Viable and nonviable LAB are able to bind FB in a pH, genus, bacterial density and analogue (FB2 > FB1) dependent manner in vitro (Niderkorn et al. 2006). FB binding is rapid and particularly effective in acidic conditions, forming a stable complex in the range of pH present in the GIT. This activity is probably present in a variety of fermented foods and feeds (Mokoena et al. 2005; Niderkorn et al. 2007) and might also operate in the stomach. Binding of other major mycotoxins: aflatoxin B1 (Haskard et al. 2000), zearalenone (El-Nezami et al. 2002a) and certain trichothecenes (El-Nezami et al. 2002b) by some probiotic LAB has also been shown in vitro. In the absence of a simple detoxification method for foods and feeds contaminated by FB, the use of selected strains of LAB appears as a promising approach to reduce their toxicological effects. However, an understanding of the binding mechanism is required to allow the optimization and safe dietary application of this technology. The aim of this work was to identify the component of the bacterial cell wall and the chemical structure of FB involved in the mechanism of binding. Materials and methods Bacteria and bacteria-derived materials Strains Lactobacillus paraplantarum CNRZ 1885 (CNRS, FRE2326 Strasbourg, France) and Streptococcus thermophilus RAR1 (LAB collection of the Research Unit for Food Process Engineering and Microbiology, INRA, Thivernal-Grignon, France) were used in most experiments. Streptococcus thermophilus CNRZ 1066 and its non-capsular, non-exopolysaccharide (EPS) producing mutant Strep. thermophilus JIM 8752 (delta epsE) were obtained from the Microbial Genetics Unit, INRA, Jouy-enJosas, France. Lactococcus lactis subsp. cremoris MG1363 and mutants, in which the synthesis of certain cell wall components and adhesion properties are affected, were from the LAB and Opportunistic Pathogens Laboratory, INRA, Jouy-en-Josas, France. Bacterial strains were grown at optimal temperature (30 or 37°C) in De Man, Rogosa, Sharpe broth for lactobacilli and M17 broth (Oxoïd Ltd., Basingstoke, UK), supplemented with 0·5% of glucose for lactococci or 10% of lactose for streptococci. Commercial purified peptidoglycans (PG) from Gram-positive bacteria Micrococcus luteus and Bacillus subtilis were purchased from Sigma, Steinheim, Germany. Determination of the bacterial cell wall component involved in binding To identify the binding site, bacteria were subjected to different physicochemical and enzymatic treatments. Bacteria (Lact. paraplantarum CNRZ 1885 and Strep. thermophilus RAR1) were prepared in advance and stored at −18°C until use. Optimization tests showed that freezing did not negatively affected the binding ability of these strains (shown in results). For experiments, bacteria were thawed at room temperature, washed twice with 0·01 mol l phosphate-buffered saline (PBS), pH 7·4 and treated by one of the following methods: water (25 or 100°C, 15 min), hydrochloric acid (1 mol l HCl, 100°C, 15 min), sodium dodecyl sulphate (SDS, 2% w/v, 100°C, 15 min) or trichloracetic acid (10% w/v, 100°C, 15 min). After treatment, suspensions were centrifuged (3000 g, 10 min, 5°C). For enzymatic treatments, washed bacteria were resuspended in 1 ml lysozyme (Sigma; 45 000 U ml in phosphate buffer, pH 6), mutanolysin (Sigma; 5000 U ml in phosphate buffer, pH 6), pronase E (Sigma; 1 mg ml in 0·01 mol l PBS, pH 7·4), lipase (Sigma; 1 mg ml in 0·01 mol l PBS, pH 7·4) or trypsin (Sigma; 1 mg ml in Tris-HCl buffer, pH 8, 10 mmol l CaCl2). Suspensions were incubated at 37°C for 2 h with shaking (240 rev min) and centrifuged (12 000 g, 10 min, 5°C). All bacterial pellets from both the physicochemical and enzymatic treatments were washed three times with 4 ml of PBS and used for the binding assay. Non-treated controls were added at each experimental run. All experiments were performed in triplicate. Determination of the functional group of fumonisins involved in binding To identify which functional group of FB can interact with bacteria, different chemical reactions were applied at different sites of FB derivatives. FB1 and FB2, purchased from Sigma and Promec (Tygerberg, South Africa), respectively, were dissolved in an exact volume of acetonitrile–water in a 1 : 1 (v/v) ratio to achieve the desired concentration of stock solutions. Hydrolysed FB1 (HFB1) and FB2 (HFB2) were obtained according to Pagliuca et al. (2005). Total hydrolysis of pure FB1 and FB2 was checked by HPLC. The chromatograms showed absence of FB peak and presence of a single peak with retention times corresponding to the expected HFB product (Pagliuca et al. 2005). An optimized procedure was also used to determine the effect of the amine group in binding. Free amine of both FB was hidden by reaction with ortho-phthalaldehyde (OPA). This option was chosen because the fumonisins of the group A in which the free amine is naturally absent are not commercially available. In vitro binding assay Treated and non-treated bacteria (10 or 10 CFU ml for certain experiments, see footnotes of tables) were tested as previously described (Niderkorn et al. 2007). Briefly, bacterial material was suspended in 1 ml of corn infusion adjusted to pH 4 with lactic acid and containing FB1 and FB2 (5 μg ml each) or their derivative compounds. The corn infusion was prepared by steeping dry whole-plant corn in water and filtering as described by Niderkorn et al. (2007). For each experiment, positive controls containing no bacterial material and a negative control containing no toxin were included. Assays and controls were incubated at 25°C for 1 h and centrifuged (3000 g, 10 min, 5°C). Supernatants and bacterial pellets were analysed for FB by reversed-phase HPLC to determine free and bound fractions respectively. Because of the instability of the FB-OPA derivative (Williams et al. 2004), assays with the free amine hidden were performed following an exact timing: At t = 0, a pure FB solution (800 μg ml) and reagent with (or without) OPA were mixed (1 : 1 v/v). At t = 2 min, 50 μl of this mixture was mixed to 950 μl of acidified corn infusion containing bacteria (10 CFU ml), then incubated for 9·25 min at 25°C. At t = 12 min, tubes were centrifuged (4500 g, 3 min, 4°C). At t = 20 min, supernatants containing free FB were derivatized with OPA. All samples were injected at t = 22 min. In these conditions, preliminary assays have shown that the complex FB-OPA remains sufficiently stable to carry out measurements. For this experiment, pellets were not analysed. Fumonisins analysis Supernatants from all samples and pellets extracts were fourfold diluted in acetonitrile-water (1 : 1 v/v), then 40 μl were added to 60 μl 0·1 mol l borate buffer at pH 10 and 100 μl of OPA reagent were added. The preparation was mixed and allowed to react for 2 min before injection of 20 μl into the HPLC system. For FB extraction, 1 ml acetonitrile–water (1 : 1 v/v) was added to the bacterial pellets and this mixture was vigorously vortexed, placed in an ultrasonic bath for 6 min, then centrifuged (4500 g, 3 min, 5°C). Analysis of FB and their hydrolysed derivatives were done at room temperature by HPLC, using fluorimetric detection. The HPLC system consisted of a GOLD 126 solvent module (Beckman Coulter, Fullerton, CA, USA), an automatic sampler (Spectra-Physics, San Jose, CA, USA) equipped with a 100μl loop and a fluorescence detector FL3000 (Spectra-System, San Jose, CA, USA). Separation of FB1, FB2, HFB1 and HFB2 was performed on a C18 reversed-phase column (Prontosil, 150 × 4·6 mm, 3 μm, Bishoff Chromatography) with a gradient elution using acetonitrile (A) and water–methanol (1 : 1 v/v) acidified at pH 3·35 with pure acetic acid (B). The gradient was started at 10% of solvent A, which increased to 60% in 6 min, then maintained at 60% for 7 min, before returned to the initial condition in 1 min. The flow rate was 1 ml min and detection was set at 336 nm excitation and 440 nm emission. The retention times of FB1, FB2, HFB1 and HFB2 were 9·9, 12·2, 10·2, 13·4 min respectively. The percentage of free (or bound) mycotoxin was calculated as 100× [Peak area of mycotoxin in the supernatant (or pellet extract)/Peak area of mycotoxin in the positive control]. Statistical analysis Data was subjected to the analysis of variance (ANOVA). A significant difference between means of controls and assays (P < 0·05) was determined by Dunett's test using the STATISTICAL ANALYSIS SYSTEM (SAS) software package, ver. 8 (SAS Institute Inc., Cary, NC, USA). Results Bacterial cell wall components affecting binding None of the physicochemical treatments applied to bacteria decreased binding of FB1 or FB2. On the contrary, freezing/thawing and thermal treatments of bacteria increased the bound fractions of FB1 and FB2 in both tested strains (P < 0·05) (Table 1). Among the chemical treatments, trichloracetic acid caused a large increase in bound FB proportion (P < 0·05). HCl also produced the same effect although it was only significant on Streptococcus cells (P < 0·05). For the enzymatic treatments, lysozyme and mutanolysin were the only treatments which caused a partial, but significant decrease of this activity (P < 0·05) (Table 1). In contrast, lipase, trypsin and pronase E, an unspecific protease from Streptomyces griseus, had no effect on binding (P > 0·05) (data not shown). Table 1 Effect of freezing, chemical and enzymatic treatments of bacteria on binding of fumonisin B1 and B2 by Streptococcus thermophilus RAR1 and Lactobacillus paraplantarum CNRZ 1885*