GST activity and membrane lipid saturation prevents mesotrione-induced cellular damage in Pantoea ananatis

Callisto®, containing the active ingredient mesotrione (2-[4-methylsulfonyl-2-nitrobenzoyl]1,3-cyclohenanedione), is a selective herbicide that controls weeds in corn crops and is a potential environmental contaminant. The objective of this work was to evaluate enzymatic and structural changes in Pantoea ananatis, a strain isolated from water, in response to exposure to this herbicide. Despite degradation of mesotrione, probably due a glutathione-S-transferase (GST) pathway in Pantoea ananatis, this herbicide induced oxidative stress by increasing hydrogen peroxide production. Thiol fragments, eventually produced after mesotrione degradation, could be involved in increased GST activity. Nevertheless, there was no peroxidation damage related to this production, as malondialdehyde (MDA) synthesis, which is due to lipid peroxidation, was highest in the controls, followed by the mesotrione- and Callisto®-treated cultures at log growth phase. Therefore, P. ananatis can tolerate and grow in the presence of the herbicide, probably due an efficient control of oxidative stress by a polymorphic catalase system. MDA rates depend on lipid saturation due to a pattern change to a higher level of saturation. These changes are likely related to the formation of GST-mesotrione conjugates and mesotrione degradation-specific metabolites and to the presence of cytotoxic adjuvants. These features may shift lipid membrane saturation, possibly providing a protective effect to bacteria through an increase in membrane impermeability. This response system in P. ananatis provides a novel model for bacterial herbicide tolerance and adaptation in the environment. Electronic supplementary material The online version of this article (doi:10.1186/s13568-016-0240-x) contains supplementary material, which is available to authorized users.


Introduction
Pesticides have been widely used to increase crop production, yet there are concerns about the adverse effects that these chemicals have on wildlife because many agrochemicals are not readily degraded by microorganisms (Copley 2009). It is estimated that ~140,000 tons of synthetic pesticides are applied annually in the USA, and approximately 300 different pesticides have been reported to be contaminants of European food products (Bjørling-Poulsen et al. 2008). The metabolism and genetic diversity of microbial communities, in both soil and plants, can be altered by pesticides as well as by the presence and expression of genes encoding enzymes that facilitate herbicide degradation (Simon and Daniel 2011;Tétard-Jones and Robert 2015). The biological degradation of pesticides appears to be one of the best strategies for potentially removing xenobiotics from the environment. Indeed, the application of herbicide biodegradation and bioremediation methodologies to pesticide-contaminated environments is one of the most promising areas of biotechnological research (Martins et al. 2007).
Oxidative stress is characterized by an increase in the production of reactive oxygen species (ROS) to a level greater than the cell's ability to defend against them (Ghelfi et al. 2011;Peters et al. 2014). Most ROS in bacteria are derived from the sequential reduction of O 2 catalyzed by enzymes in the electron transport chain associated with the plasma membrane (Lushchak 2001). Highly reactive products of aerobic metabolism, such as hydrogen peroxide (H 2 O 2 ), superoxide (O 2 • ) and hydroxyl (OH • ) radicals, can damage DNA, RNA, proteins and lipids (Gratão et al. 2005), and antioxidant systems, such as the enzymes catalase, peroxidase and glutathione reductase (Olchanheski et al. 2014;Peters et al. 2014), are invoked to combat reactive oxygen intermediates. For example, bacteria exposed to the herbicide acetochlor at 62 and 620 mM concentrations exhibited an increase in lipid peroxidation by 39 and 34 %, respectively, suggesting that microorganisms can tolerate some cytotoxic agrochemicals via induction of antioxidant stress responses (Martins et al. 2011).
Pantoea ananatis CCT 7673 is a mesotrione-degrading bacterium that we previously isolated from water, together with 359 other mesotrione-tolerant microorganisms (Pileggi et al. 2012). Bacterial growth studies showed that P. ananatis tolerates high concentrations of mesotrione and Callisto ® . high-performance liquid chromatography (HPLC) analysis demonstrated that P. ananatis CCT 7673 can degrade this herbicide within ~17 h, resulting in products that are less toxic than those produced by a Bacillus sp. strain (Crouzet et al. 2010). Despite the ability of P. ananatis to degrade mesotrione, this bacterium did not utilize the herbicide as a nutrient source for growth (Pileggi et al. 2012).
The aim of this current study was to determine the mechanism(s) by which P. ananatis resists the toxic effects of mesotrione and its commercial formulation Callisto ® . We also examined whether mesotrione creates stress responses in P. ananatis, ultimately affecting cellular metabolism through enzymatic responses.

Mesotrione-degrading strain
The bacterium used in this study, P. ananatis CCT 7673, was previously isolated at Capão da Onça School Farm-Ponta Grossa State University, Ponta Grossa-PR, Brazil, and was previously shown to be a mesotrione-degrading bacterium (Pileggi et al. 2012).

Herbicide treatment and growth curve determination
Bacterial growth (600 µL) was measured spectrophotometrically at 600 nm using 600 µL of sample collected every 2 for 24 h. When the samples reached an OD of greater than 0.6, bacterial growth was also measured by dilution-plating on MM, MMM and CMM media.

Hydrogen peroxide quantification (H 2 O 2 )
Cells (100 mg) were homogenized in liquid nitrogen containing 0.1 % tricloroacetic acid (TCA) using a mortar and pestle and centrifuged at 11,600×g for 15 min at 4 °C. A 0.2 mL aliquot of the supernatant was added to 0.2 mL 100 mM phosphate buffer, pH 7.5; 0.8 mL of 1 M KI solution was added. The reaction was mixed and incubated in the dark at 4 °C for 1 h, followed by 20 min at room temperature. The absorbance was measured at 390 nm, and the results are expressed as mmol H 2 O 2 /10 6 C.F.U (Dourado et al. 2013).

Lipid peroxidation
Lipid peroxidation was quantified by estimating the levels of thiobarbituric acid (TBA)-reactive substances, as previously described (Heath and Packer 1968). The concentration of malondialdehyde (MDA) was monitored at 535 and 600 nm using a Perkin Elmer Lambda 40 spectrophotometer, and the MDA concentration was calculated using a molar extinction coefficient of 155 mM cm −1 .

Membrane lipid evaluation
After 12, 17.5 and 19 h of incubation, bacterial samples were centrifuged for 5 min at 11,600×g at 4 °C, and the precipitate was frozen. The material was lyophilized for 15 h; a solution of chloroform, methanol and water (1: 2: 0.8) was added, and the mixture was stirred at 180 rpm for 18 h at room temperature. After this period, a mixture of chloroform, methanol and water (1: 1: 0.9) was added to a final volume of 5.8 mL (Bligh et al. 1959). The chloroform phase was collected, and the membrane lipid fraction was analyzed by Fourier Transformed Infrared Spectroscopy (FTIR), from 400 to 4000 cm −1 . Statistical analysis was performed using principal component analysis (PCA) and partial least squares (PLS), as implemented in the Pirouette v 4.0 software (Infometrix, Bothell, WA, EUA). PCA was performed using the average value of triplicates of the spectra, and PLS was performed with the value of each triplicate.

Protein extraction and quantification
Total proteins were extracted, in triplicate, from P. ananatis CCT 7673 cells after 12 and 24 h of incubation after each treatment. Antioxidant enzymes were isolated as follows, at 4 °C, unless otherwise stated. The culture was centrifuged at 10,000×g for 10 min. The pellet was ground in liquid nitrogen using a mortar and pestle and homogenized (10:1, v/w) in 100 mM potassium phosphate buffer (pH 7.5) containing 1 mM ethylenediaminetetraacetic acid (EDTA), 3 mM DL-dithiothreitol and 5 % (w/v) polyvinylpolypyrrolidone. The homogenate was centrifuged at 10,000×g for 30 min, and the supernatant was stored in separate aliquots at -80 °C prior to enzymatic activity assays. The protein concentration was determined as described in Bradford (1976), using bovine serum albumin (BSA) as the standard.

Polyacrylamide gel electrophoresis (PAGE)
Electrophoresis was carried out using 12 % polyacrylamide gels, with a 4 % stacking gel, as previously described (Monteiro et al. 2011). Sodium dodecyl sulfate (SDS) was omitted for non-denaturing gels. Electrophoresis was carried out at 4 °C at a constant current of 15 mA gel −1 for 3 h for gels stained for superoxide dismutase (SOD) activity or for 21 h for gels stained for catalase (CAT) activity. Equal amounts of protein (20 µg) were loaded per lane onto non-denaturing PAGE gels. For SDS-PAGE analysis, gels were stained as previously described (Azevedo et al. 1998).

SOD isoforms
Classification of superoxide dismutase (EC 1.15.1.1) isoforms was performed by non-denaturing PAGE (12 % gels) with 300 μg of protein from bacterial extracts, as described by Gratão et al. (2015). Prior to staining, the gel was divided vertically into three parts: the first part was maintained at 4 °C in 100 mM potassium phosphate buffer, pH 7.8; the second was immersed in 100 mL of the same buffer but containing 2 mM KCN and 1 mM EDTA; and the third was immersed in 100 mL of buffer containing 5 mM H 2 O 2 and 1.0 mM EDTA. All of the steps were performed in the dark. Isoforms were classified as Cu/Zn-SOD, Fe-SOD or Mn-SOD (Cabiscol et al. 2000).

SOD activity staining
SOD activity was assayed using non-denaturing PAGE gels, essentially as described (Gratão et al. 2015). Following electrophoresis, gels were rinsed in distilled deionized water and incubated in the dark for 30 min in 50 mM potassium phosphate buffer, pH 7.8, containing 1 mM ethylenediamine tetraacetic acid (EDTA), 0.005 mM riboflavin, 0.1 mM nitroblue tetrazolium and 0.3 % N,N,N′,N′-tetramethylethylenediamine. One unit of bovine liver SOD (Sigma, St. Louis, MO, USA) was included as a positive control for the chemical reaction. The gels were immersed in water until the achromatic bands of SOD were revealed (Gratão et al. 2012).

Catalase activity staining
Catalase (EC 1.11.1.6) activity was observed after protein separation by non-denaturing electrophoresis. Following staining, the gels were incubated in 0.003 % H 2 O 2 for 10 min and developed in a 1 % (m/v) FeCl 3 and 1 % K 3 Fe (CN 6 ) solution for 10 min. One unit of bovine liver CAT (Sigma, St. Louis, USA) was used as a control for the chemical reaction (Boaretto et al. 2014).

Glutathione-S-transferase (GST) assays
The activity of GST was assayed in a reaction mixture containing 0.1 M potassium phosphate buffer, pH 6.5, 0.1 mM glutathione sulfhydryl (GSH), and 0.04 M 1-cloro-2,4-dinitrobenzene (CDNB), as previously described (Ghelfi et al. 2011). The CDNB was mixed with buffer and GSH, and 25 µL of protein extract was added. The reaction was initiated 1 min after the mixture was prepared and incubated at 30 °C. Activity was determined by monitoring the kinetics of the formation of 1, 2-dichloro-4-nitrobenzene (DCNB) in samples for 3 min using a spectrophotometer at 610 nm. The results are expressed as µmol/min/mg protein.

Statistical analysis
Data on H 2 O 2 , cell viability, MDA quantification and GST activity were obtained, in triplicate, for each treatment and analyzed statistically using a randomized complete design. The significance of differences was assessed by analysis of variance (p < 0.05). Statistical analyses were performed with Stata 12 software using factor analysis and the ANOVA test for comparisons between strains, treatments, and growth periods, and using the Bonferroni's test for the analyses. We also used the Bartlett's Test for inequality of population variances and Mann-Whitney/Wilcoxon Two-Sample Test (Kruskal-Wallis test for two groups).

Results
Experimental pilot tests were conducted to determine the best incubation time of analysis. Two sets of time period were selected based on the growth curves ( Fig. 1) and mesotrione degradation by P. ananatis. Based on associations with mesotrione degradation, quantification of H 2 O 2 , viability, GST, MDA and lipid saturation was performed at 12, 17.5 and 19 h (the late log phase, peak of mesotrione degradation and early stationary phases, respectively). In contrast, CAT and SOD activities were analyzed only at 12 and 24 h (the late log, before the mesotrione degradation peak, and stationary phases, after mesotrione degradation, respectively) because they were not related to the mesotrione degradation peak.

Evaluation of peroxide levels
The peroxide levels found in in P. ananatis, in response to mesotrione and Callisto ® , are shown in Fig. 2.

Viability and bacterial growth
The results of viability and bacterial growth experiments in MM (control) as well as MMM and CMM are shown in Figs. 1 and 3, respectively.

SOD profile
Classification of SOD isoforms was performed by nondenaturing PAGE. The Cu/Zn-SOD isoform is indicated by band I, and Mn-SOD by bands II and III (Fig. 4).
Three SOD bands were found to be active in P. ananatis CCT 7673 exposed to the herbicides (Fig. 5), and all bands were more clearly present after 24 h of treatment.

CAT activity and gel profiles
The CAT activity in P. ananatis CCT 7673 at 12 and 24 h in mesotrione and Callisto ® treatments, and control, are shown at Fig. 6.

GST activity
GST analysis was performed at 12 h (before mesotrione degradation), 17.5 h (immediately before mesotrione degradation) and 19 h (after mesotrione degradation and in the presence of byproducts in the culture medium) (Fig. 7).

Lipid membrane changes in response to herbicide cytotoxicity
The concentration of MDA was examined in the controls, mesotrione-and Callisto ® -treated cultures to determine the amount of lipid peroxidation and herbicide cytotoxicity (Fig. 8).
The lipid saturation data was obtained by FTIR and PCA (Fig. 9a) and PLS (Fig. 9b) statistical analyses. In the regression vector graphic (Fig. 6a), the regions with variables having the highest weight are those that are far from zero and have a higher magnitude coefficient. Positive and negative values indicate directly and inversely proportional relationships between the parameters, respectively. Exploratory PCA and PLS  were studied via construction of a matrix with three classes (at 12, 17.5 and 19 h). All spectral regions were evaluated.

Evaluation of oxidative stress generated by mesotrione
Cellular metabolism in aerobic organisms results in O 2 reduction and formation of ROS, such as H 2 O 2 and O 2 • . These ROS are generated via normal aerobic metabolic processes of the electron transport chain or due to adverse environmental conditions (Dourado et al. 2015;Gratão et al. 2015), and an increase in ROS can cause oxidative stress in the cell, negatively impacting DNA, RNA and lipids (Gratão et al. 2005). To avoid such damage, organisms possess antioxidant mechanisms that are mediated by enzymes to reduce the impact of ROS, which generally occurs via transcriptional activation of the genes oxyR and soxRS. ROS-induced damage can be assessed by quantification of H 2 O 2 in cells, which is one of the main causes of oxidative stress (Gratão et al. 2012), and MDA levels can be used as a marker of oxidative stress (Cabiscol et al. 2000).
The peroxide levels found in this study indicated that mesotrione (p < 0.061, Additional file 1: Section 1.2) and Callisto ® (p < 0.015, Additional file 1: Section 1.2) induced oxidative stress in P. ananatis (Fig. 2). Moreover, the mesotrione molecule (Additional file 1) contains two groups at its benzoyl positions, 2 and 4, which produce a strong electron-withdrawing effect (Mitchell et al. 2001). These nitro-and methanesulfonyl groups, which are likely responsible for the observed oxidative stress, are also found in MNBA and AMBA, the products of mesotrione degradation by Bacillus sp (Durand et al. 2006). In fact, AMBA is more cytotoxic than the active ingredient mesotrione (Mitchell et al. 2001;Bonnet et al. 2008). In contrast, the nitro-and methanesulfonyl groups are removed or modified after 18 h of degradation by P. ananatis CCT 7673, producing oxidative stress-generating molecules other than AMBA or MNBA (Pileggi et al. 2012). Although no difference in microbial growth was observed among the treatments and control, the H 2 O 2 data indicated that mesotrione caused oxidative stress in P. ananatis.

Herbicide tolerance according to bacterial growth
The results of viability and bacterial growth experiments in the MM (control) as well as in MMM and CMM are shown in Figs. 1 and 3, respectively. Viability reached proportional levels in all experiments and controls, compatible with the growth stage. Except at 12 h with mesotrione treatment, before herbicide degradation was complete, in which low viability was found (p < 0.012, Additional file 1: Section 2.1). These analyses suggested that mesotrione decreased the capacity of cell division prior to the time of complete herbicide degradation (18 h). However, this did not diminish overall cell viability, indicating that and 4 correspond to control, lanes 2 and 5 to mesotrione treatment, and lanes 3 and 6 to Callisto ® treatment after 12 (lanes 1, 2 and 3) and 24 h (lanes 4, 5 and 6). Band I corresponds to Cu/Zn-SOD. Bands II and III correspond to Mn-SOD Fig. 6 CAT activity staining following non-denaturing PAGE analysis of P. ananatis CCT 7673 extracts. Lane S shows SOD standard, lanes 1 and 4 correspond to control, lanes 2 and 5 to mesotrione treatment, and lanes 3 and 6 to Callisto ® treatment after 12 (lanes 1, 2 and 3) and 24 h (lanes 4, 5 and 6) P. ananatis tolerates the herbicide, as previously shown by Pileggi et al. (2012). Therefore, any difference in oxidative stress response in these cultures was likely due to specific induction by the herbicide treatment. Moreover, the elevated H 2 O 2 levels indicate that the herbicides caused oxidative stress in this bacterium.

SOD profile
Three SOD bands were found to be active in P. ananatis CCT 7673 exposed to the herbicides (Figs. 4, 5), and all bands were more clearly present after 24 h of treatment. The electrophoretic pattern of SOD activity in P. ananatis CCT 7673 did not indicate a specific SOD  The Cu/Zn-SOD isoform, indicated by band I, is located within the periplasmic space, but studies of this isoform in this bacterium are scarce; bands II and III correspond to Mn-SOD (Fig. 4), a SOD-type enzyme commonly found in the bacterial cytoplasm that likely responds to increased levels of the O 2 • radical ). The lack of intense change in SOD activity suggested that the treatments did not alter bacterial metabolism because changes in metabolism would result in changes in the production of O 2 • . However, the SOD band III showed approximately the same level of activity in all treatments and at all times, whereas the other isoenzymes exhibited some increase in activity, particularly after 24 h.

CAT activity and gel profiles
The mesotrione molecule is less cytotoxic than Callisto ® , which is the commercial formulation containing adjuvants (Mitchell et al. 2001;Bonnet et al. 2008). The peroxide rates and levels of CAT activity in P. ananatis CCT 7673 at 12 h of treatment supported this hypothesis (Figs. 2, 6). There was increase in CAT activity from 12 to 24 h in control and herbicide-treated cultures. H 2 O 2 levels may increase 5-10-fold during the transition from the lag phase to the exponential growth phase in Escherichia coli (González-Flecha and Demple 1997). In this study, CAT activity in P. ananatis CCT 7673 was dependent on the growth phase (Fig. 3), and the H 2 O 2 levels increased no more than twofold, probably due the presence of five different isoforms of CAT (Fig. 6). This number of isoforms had not been reported to P. ananatis yet. In a previous study with this same bacterial strain, mesotrione was not detectable by HPLC analysis in the culture medium after 24 h (Pileggi et al. 2012), though H 2 O 2 was produced by the bacterial cells for up to 12 h in response to changes in cell growth as well as to mesotrione and Callisto ® (Fig. 2). This production of H 2 O 2 likely induced higher levels of CAT expression.

GST activity
With two carboxylic groups with reducing properties, GSH is a non-enzymatic antioxidant that can act as an electron donor and efficiently removes ROS due to its ability to transfer electrons to GST (Ghelfi et al. 2011). GST, in turn, catalyzes the modification of hydrophobic electrolytic substrates, which are normally cytotoxic (Li et al. 2009;Masip and Veeravalli 2006). Furthermore, this enzyme, which is located within the bacterial periplasm, can detoxify xenobiotics (Kullisaar et al. 2010). Pileggi et al. (2012) demonstrated that P. ananatis degrades mesotrione after 18 h of incubation. In our study, GST analysis was performed at 12 h (before mesotrione degradation), 17.5 h (immediately before mesotrione degradation) and 19 h (after mesotrione degradation and in the presence of byproducts in the culture medium), and activity was found to be greatest at the last two time points (Fig. 7), corresponding to the mesotrione degradation period in P. ananatis CCT 7673. The p value for Bartlett's Chi square (p = 0.00) suggests that the variances are not homogeneous and that the ANOVA may not be appropriate. So Mann-Whitney/Wilcoxon Two-Sample Test (Kruskal-Wallis test for two groups) was used with p < 0.0180 (Additional file 1: Chapter 3). Therefore, GST is suggested to be involved in mesotrione degradation in this strain. A relationship between GST and herbicide degradation has been observed with several herbicides, such as diuron, oxyfluorfen (Geoffroy 2002), atrazine, chloroacetanilide (Van Eerd et al. 2003) and metolachor (Stamper et al. 2003). However, such a relationship with GST was also found with 2,4-dichlorophenoxyacetic acid (2,4-D) and primisulfuron, two herbicides not known to be metabolized through glutathione conjugation (Pang et al. 2012). Enhancement of the level of free thiol groups has been related to augmented GST activity in different plants (Miteva et al. 2003(Miteva et al. , 2004. P. ananatis CCT 7673 can degrade 100 % of added mesotrione (Pileggi et al. 2012), eventually producing thiol fragments that could be involved in increased GST activity (Fig. 7).

Lipid membrane changes in response to herbicide cytotoxicity
Malondialdehyde (MDA) was examined to determine the amount of lipid peroxidation and herbicide cytotoxicity. As shown in Fig. 8, the MDA values were greater in the controls than mesotrione (p < 0.008, Additional file 1: Section 4.2) and Callisto ® -treated (p < 0.003, Additional file 1: Section 4.2) cultures mainly at 17.5 h growth phase. It is an unexpected inversion of proportion observed for H 2 O 2 results (Fig. 2). MDA is a cytotoxic aldehyde product that is released when ROS react with unsaturated fatty acids of the cell membrane, thus causing cell damage (Heath and Packer 1968). Reducing agents, such as glutathione or sulfite, can activate H + -ATPases or prevent their inhibition by pro-oxidants, and these effects may be attributable to the location of the H + -ATPase in the lipid bilayer (Maeshima 2000). Depending on the stress conditions, the composition of membrane fatty acid residues can prevent lipid peroxidation and MDA production (Ayala et al. 2014). In the presence of oxidants, the stability (mechanical strength) of biological membranes decreases, and antioxidant enzymes may decelerate or inhibit lipid peroxidation reactions, thus enhancing membrane stability. Furthermore, changes in membrane stability, such as those provided by ion channels, can stabilize membranes (Nurminsky et al. 2009).
As MDA serves as an index of peroxidation damage (Heath and Packer 1968), the lipid fraction has been further investigated by mid-infrared spectroscopy to detect changes (Costa Filho 2014). Ogliari et al. (2009) demonstrated that mesotrione promotes differential activation of the primary proton transport system for enzymatic detoxification of this herbicide in maize plants, which can be attributed to changes in lipid conformation after treatment with mesotrione. Freezing and salt stress induce the production of desaturases, which alter unsaturated fatty acids in the membranes of plants and yeasts (Rodríguez-Vargas et al. 2007). Changes in the conformation of saturated and unsaturated fatty acids of the plasma membrane could be considered a protective mechanism by bacteria in contact with herbicides (Murínová and Dercová 2014;Segura et al. 1999). Balague et al. (2001) reported a decrease in unsaturated lipids in the plasma membrane of E. coli HB101 after treatment with the herbicide 2,4-D, and these authors considered the reduction in membrane fluidity as a possible defense mechanism against cell damage. Moreover, Sánchez et al. (2005) reported that an increase in the saturation level of membrane lipids in Klebsiella planticola DSZ allowed growth in a culture medium containing the herbicide simazine. Similar results were reported by Danilo et al. (1996) for Ochrobactrum anthropi cells cultured in the presence of atrazine.
As peroxidation only occurs when lipids are in an unsaturated form, the MDA level depends on the saturation level of lipids (Ayala et al. 2014). Thus, the lower levels of MDA, at log growth phase, compared to the control indicated a change to a higher level of saturation of lipids in P. ananatis after treatments with mesotrione and Callisto ® . Thus, the results suggest changes toward lower membrane permeability, possibly conferring protection against herbicide cytotoxicity.
The hypotheses described above were tested using lipid saturation data obtained by FTIR, and PCA and PLS statistical analyses. PLS relates the variation in incubation time with changes in the FTIR spectrum to assess the functional groups formed or consumed. The regression was performed using the set of FTIR data for those variables (from 950 to 1500 cm −1 ) that showed better discrimination in PCA. The correlation coefficient obtained was 0.93. The others studied variables (Glutathione, GST, H 2 O 2 , MDA, SOD and cell viability) presented lower correlation coefficient values with the FTIR spectrum and therefore were not considered in the analysis of lipid saturation changes.
In the regression vector graphic (Fig. 9a), the regions with variables having the highest weight are those that are far from zero and have a higher magnitude coefficient. Positive and negative values indicate directly and inversely proportional relationships between the parameters, respectively. For example, the variable 1381 cm −1 (infrared band) indicates a decrease in the angular deformation of CH bond-plane cis-olefinic groups (unsaturated) with increasing incubation time, possibly increasing membrane permeability. In addition, one phospholipid region at 1085 cm −1 , a symmetric stretching of the carbonyl (C=O) of COO − groups of fatty acids at 1460 cm −1 and acetate groups of saturated alcohols at 1271 cm −1 were also identified.
Exploratory PCA and PLS were studied via construction of a matrix with three classes (at 12, 17.5 and 19 h). All spectral regions were evaluated. By applying the preprocessing average centered with multiplicative scatter correction (MSC) and first derived with 99.26 % (PCA) and 99.60 % (PLS) for 7 and 12 principal components (PC), the most informative was between 950.90631 and 1500.6188 cm −1 , respectively. PC1 and PC2 in the score chart (Fig. 9a) demonstrated the greatest discrimination among the samples, and differences were related to the unsaturated and saturated fatty acids present. The data indicated that membrane permeability in P. ananatis cells after 12 h of incubation is related to unsaturated or saturated/near unsaturated fatty acids in the control and treatments at spectral regions with a number of bands at 1165, 1246, and 1397 cm −1 . This incubation time represents the log phase of growth ( Fig. 1), suggesting that cells require permeable membranes to perform metabolism processes at maximal rates. After 17.5 h of incubation, the band at 1099 cm −1 (unsaturated fatty acid) was predominant, characterizing a state of membrane permeability, and this was more conspicuous in MMM medium (Fig. 9b) than in MM medium. According to peroxide production data, mesotrione had a cytotoxic effect on bacterial cells (Fig. 2). In contrast, no CAT-and SOD-specific responses were found at this time. In addition to changes in fatty acid saturation, GST was responsive to mesotrione treatment, with an increase in enzyme activity (Fig. 7) correlated to complete mesotrione degradation at 18 h (Pileggi et al. 2012).
A GST-mesotrione conjugate or induction of GST by free thiol may have been generated during 12-17.5 h of incubation, and such a molecule would need to interact with the cell membrane (Kullisaar et al. 2010) or cytoplasm (Vuilleumier 1997). Pang et al. (2012) demonstrated that glutathione transporters located in the plasma membranes of plants are important components in the glutathione conjugation-related detoxification system. For this reason, the data from the 17.5 h CMM treatment might indicate that the membrane was in a state of impermeability. The herbicide surfactants present in Callisto ® , 2-(8-methylnonoxy) ethanol and octan-1-ol (https://www3.syngenta.com/country/uk/ en/ProductGuide/Pages/Callisto.aspx) are designed to alter the permeability of plant cells. Moreover, 1,2-benzisothiazolin-3-one (https://www.archive.epa.gov/pesticides/reregistration/web/pdf/benzisothiazolin_red. pdf ) has an antimicrobial function, which likely helps to prevent microbial degradation of mesotrione. PCA and PLS analyses of the 17.5 h CMM treatment suggested that Callisto ® adjuvants interfered with and changed the structure of the P. ananatis cytoplasmic membrane (Mesnage et al. 2013).
At the 19 h time point, the presence of the band at 1240 cm −1 (Silverstein and Webster 1998) indicated that P. ananatis membrane impermeability increased in the MMM treatment and even more so in CMM.
In general, an increase in saturated fatty acids in P. ananatis was observed in response to mesotrione treatment, with a greater increase in Callisto ® -treated cells (Fig. 9). These data appear to be correlated to a decrease in membrane permeability, as a response to the cytotoxicity of the herbicide and adjuvants. Despite the increase in H 2 O 2 (Fig. 2), the decrease in MDA concentration in the mesotrione and Callisto ® treatments (Fig. 8) corroborated this hypothesis because the MDA measured originated from the peroxidation of unsaturated fatty acids.
The commercial formulation of the Callisto ® herbicide, and its active ingredient mesotrione at concentrations that are commonly used in the environment, cause oxidative stress in P. ananatis CCT 7673 through increased H 2 O 2 production and greater GST enzyme expression, which is likely involved in herbicide degradation. Despite this, the bacterium can tolerate and grow in the presence of the herbicides. No CAT-and SOD-specific responses were observed to explain this herbicide tolerance, and the bacterium exhibited low amounts of MDA, an index of peroxidation damage, under herbicide treatments. A pattern change to a higher level of lipid saturation was suggested by PCA and PLS analyses, possibly conferring a protective effect to bacteria cells through reduced bacterial membrane permeability. In the mesotrione treatments, this increase was likely related to the formation of a GST-mesotrione conjugated adduct. Changes in membrane fatty acid saturation in the Callisto ® treatments were of a greater degree than in the mesotrione treatments. These events were likely related to the adjuvant content, which interfered with the structure of the lipid membrane. Taken together, these features make P. ananatis an excellent model for studying herbicide tolerance in both soil and water environments.