Additive effects of metal excess and superoxide, a highly toxic mixture in bacteria.

Heavy metal contamination is a serious environmental problem. Understanding the toxicity mechanisms may allow to lower concentration of metals in the metal-based antimicrobial treatments of crops, and reduce metal content in soil and groundwater. Here, we investigate the interplay between metal efflux systems and the superoxide dismutase (SOD) in the purple bacterium Rubrivivax gelatinosus and other bacteria through analysis of the impact of metal accumulation. Exposure of the Cd2+ -efflux mutant ΔcadA to Cd2+ caused an increase in the amount and activity of the cytosolic Fe-Sod SodB, thereby suggesting a role of SodB in the protection against Cd2+ . In support of this conclusion, inactivation of sodB gene in the ΔcadA cells alleviated detoxification of superoxide and enhanced Cd2+ toxicity. Similar findings were described in the Cu+ -efflux mutant with Cu+ . Induction of the Mn-Sod or Fe-Sod in response to metals in other bacteria, including Escherichia coli, Pseudomonas aeruginosa, Pseudomonas putida, Vibrio cholera and Bacillus subtilis, was also shown. Both excess Cd2+ or Cu+ and superoxide can damage [4Fe-4S] clusters. The additive effect of metal and superoxide on the [4Fe-4S] could therefore explain the hypersensitive phenotype in mutants lacking SOD and the efflux ATPase. These findings underscore that ROS defence system becomes decisive for bacterial survival under metal excess.

Introduction

Heavy metal pollution, mainly caused by anthropogenic activities, may have adverse effects on the ecosystems and living organisms unless sustainable management practices are implemented. Cu2+ and Cd2+ pollution originating from activities including excessive use of fertilizers, antimicrobials and agrochemicals is an important environmental concern (Ballabio et al., 2018).

https://ec.europa.eu/environment/integration/research/newsalert/pdf/agricultural_management_practices_influence_copper_concentrations_european_topsoils_518_na2_en.pdf

Metal pollution contributes also to the selection and spread of antibiotic resistance factors. In fact, cross‐resistance in bacteria can occur through transfer of genes encoding efflux pumps that can expel antibiotics in addition to metals (Baker‐Austin et al., 2006; Rensing et al., 2018; Asante and Osei Sekyere, 2019). Metal pollution affects groundwater and soils properties by altering the microbial diversity and evenness (Nunes et al., 2016). Limiting the concentration of heavy metals in agrochemicals or in metal‐based antimicrobial treatments can lead to a sustainable use of metals in agriculture and farming (Turner, 2017) and may limit the metal‐related co‐selection for metal and antibiotic resistance in bacteria.

In contaminated soils, exposed organisms have to deal with excess metal. In bacteria, most metal ions presumably reach the bacterial cytoplasm through porins and specific or non‐specific transporters in the inner membrane. The lack of rigorous control or specific uptake system for some metals is often compensated for by the presence of effective efflux systems. This allows cells to tolerate the excess of metal(s) in their immediate environment by expelling the surplus of metal ions (Arguello et al., 2007; von Rozycki and Nies, 2009; Capdevila et al., 2017; Chandrangsu et al., 2017). The P1B‐type ATPases family of heavy metal transporters is the most frequently present in bacteria. They are efficient efflux pumps that extrude excess of toxic metal ions such as copper (Cu+), zinc (Zn2+), cadmium (Cd2+) or silver (Ag+) from the cytoplasm to the periplasm in which metal is handled by other detoxifying proteins (Arguello et al., 2007). The effectiveness of these metal efflux systems prevents the molecular and physiological damages that metal can cause. Therefore, mutants defective in metal homoeostasis machineries, specifically in the efflux systems, shed light on the toxicity mechanisms and the molecular events following metal accumulation within the cell. Lines of evidence showed that while excess of metal can cause damage to [4Fe‐4S] clusters and mismetallation of metalloproteins under strict anaerobic condition (Macomber and Imlay, 2009), contribution of reactive oxygen species (ROS) should be taken into account under aerobic condition.

Superoxide is produced in presence of oxygen by some aerobic respiratory complexes and other non‐respiratory flavoproteins, like the lipoamide or xanthine dehydrogenases. It can directly damage exposed catalytic [4Fe‐4S] clusters in dehydratases or mononuclear Fe2+ enzymes, or indirectly damage other cell components by contributing to the production of hydroxyl radicals (Imlay, 2013; Imlay, 2019). To deal with superoxide in the cytoplasm, bacteria possess either one or two superoxide dismutases (Fe‐Sod and Mn‐Sod) whose expression is differently regulated by redox‐cycling compounds and by iron (Imlay, 2013; Imlay, 2019). In enteric bacteria such as Escherichia (E.) coli, sodA gene is regulated by the SoxRS system. The transcriptional factor SoxR functions as a sensor of oxidative stress in response to changes in the redox state of its [2Fe‐2S] cluster (Imlay, 2013; Kobayashi, 2017; Imlay, 2019). SoxR activates the transcriptional factor SoxS that in turn induces the transcription of many genes including sodA. Nonetheless, the majority of SoxR regulons in bacteria, including Vibrio and Pseudomonas, lack the soxS gene (Dietrich et al., 2008). Expression of Sod in these bacteria does not involve SoxR and relies mainly on iron homoeostasis and regulators. In E. coli and other bacteria, when iron is available, the Fe2+‐Fur repressor inhibits Mn‐Sod and small RNA RyhB production and activates the Fe‐Sod synthesis. In contrast, under iron limitation, Fur is inactivated, which induces the synthesis of the Mn‐Sod and the transcription of the small RyhB RNA. RyhB interaction with Fe‐Sod mRNA triggers degradation of the mRNA and thus promotes decrease of Fe‐Sod in the cells (Masse and Gottesman, 2002; Imlay, 2013; Imlay, 2019).

Metal stress promotes increase in the level of oxidative stress‐associated defence enzymes such as superoxide dismutases (SOD), catalases and peroxidases, which suggest elevated ROS in metal‐stressed cells (Krumschnabel et al., 2005; Macomber et al., 2007; Helbig et al., 2008; Ray et al., 2013; Xu et al., 2019). The involvement of these enzymes in metal stress response was further supported by the metal‐sensitive phenotype of E. coli mutant of which two genes encoding SOD were concomitantly inactivated (Geslin et al., 2001).

Macomber and co‐workers reported that while copper did not induce significant oxidative DNA damage in copper‐overloaded E. coli cells, they exhibited increased catalase (KatG and KatE) and superoxide dismutase activities. This suggested an induction of SoxRS regulons and the presence of elevated amounts of hydrogen peroxide and superoxide in these cells (Macomber et al., 2007). Metals such as Fe2+ and Cu+ are redox‐active ions which, in presence of oxygen, can induce the formation of reactive oxygen species via Fenton and Haber–Weiss‐type reactions (Gunther et al., 1995). These metals are also toxic under anaerobic condition, as Cu+ was shown to directly poison bacteria by damaging the exposed [4Fe‐4S] clusters of dehydratases in E. coli (Macomber and Imlay, 2009) and those of porphyrin biosynthesis enzymes in the photosynthetic bacterium Rubrivivax (R.) gelatinosus and the human pathogen Neisseria gonorrhea (Azzouzi et al., 2013; Djoko and McEwan, 2013). Toxicity of non‐redox‐active metals such as Cd2+ or Zn2+ was also associated with destruction of exposed [4Fe‐4S] clusters, and release of iron, which in turn might produce hydroxyl radicals via the Fenton reaction (18).

In the present study, we investigated the role of superoxide dismutase and superoxide stress in heavy metal toxicity. We found that Cd2+ or Cu+ stress induced the expression and activity of the cytosolic Fe‐Sod or Mn‐Sod in many bacteria including R. gelatinosus, E. coli, Vibrio (V.) cholerae, Pseudomonas (P.) aeruginosa, P. putida and Bacillus (B.) subtilis, when the metal efflux systems are missing.

Although SOD induction by excess metal was previously reported in many transcriptomic data sets, the importance and consequences of SOD inactivation under excess metal in metal efflux deficient mutants were never reported. Here, the Cd2+ and Cu+ hypersensitive phenotypes of mutants lacking SOD along with the Cd2+ and Cu+ ‐ATPases, respectively, were shown to emphasize the importance of SOD and ROS detoxification in metal toxicity. Altogether, the data demonstrated the involvement of SOD in the protection of heavy metal‐exposed cells and shed light on the synergy of excess metal and superoxide in heavy metal toxicity.

Results

Induction of a 22 kDa protein in cadR − and ΔcadA deficient strains

To deal with toxic Cd2+ exposure, bacteria usually induce the expression of gene encoding the metal efflux ATPase, CadA (also known as ZntA in some species, as it expels Zn2+ as well). The major transcriptional regulator CadR interacts with Cd2+, thereby allowing induction of the expression of the CadA ATPase. While these genes are widely distributed in bacterial species, purple photosynthetic bacteria can encounter Cd2+ in their environment and are therefore suitable models to assess the effect of Cd2+ on their growth. Requirement for CadR in CadA expression and metal tolerance was assessed in the purple photosynthetic bacterium R. gelatinosus wild‐type and cadR − mutant cells. To this end, total cell extracts from overnight culture of wild‐type and cadR − mutant grown under photosynthetic condition with different concentrations of CdCl2 were analysed by Western blot. Note that CadA and the periplasmic copper protein CopI can be readily detected by conventional HisProbe due to the presence of histidine‐rich motifs within these proteins. In the wild‐type, Cd2+ induced a steady increase in the amount of CadA and CopI proteins (Fig. 1A). In the cadR − mutant, CadA was not induced below 600 µM of CdCl2. Unexpectedly, however, apparent induction of CadA was observed with excess CdCl2 (800 and 1000 µM). Furthermore, a 22 kDa protein band revealed by the HisProbe was clearly induced up to 600 but not with 800 and 1000 µM of CdCl2 (Fig. 1A). Interestingly, the corresponding band was only slightly expressed in the wild‐type cells, even for those challenged with excess Cd2+. Concomitantly to CadA induction, the 22 kDa protein amount decreased upon elevated Cd2+ stress (800 and 1000 µM) in the cadR − mutant (Fig. 1A). Taken together, these data confirmed that Cd2+ induced the expression of the Cd2+‐efflux ATPase CadA and the production of a 22 kDa protein in CadR‐deficient cells. Moreover, these data suggested a direct link between the expressions of these two proteins. To test this possibility, CadA‐deficient mutant ΔcadA of R. gelatinosus was subjected to Cd2+ stress. In contrast to cadR −, growth of the ΔcadA mutant was inhibited beyond 200 µM CdCl2 (see below); therefore, the analysis included only low CdCl2 concentrations. As expected, Cd2+ excess (100 or 200 µM) caused an increase in the amount of CadA and CopI proteins, but the amount of the 22 kDa protein remained unchanged in the wild‐type strain (Fig. 1B). In contrast, both CopI and the 22 kDa protein were induced in the Cd2+‐stressed ΔcadA mutant cells (Fig. 1B). Therefore, in the absence of the CadA ATPase, Cd2+ induced the expression of the 22 kDa protein which could be involved in Cd2+ response and tolerance. Similar results showing the induction of the 22 kDa protein in cadR − and ΔcadA strains were obtained under respiratory (aerobic) condition (Fig. S1). Alongside this study, biochemical experiments in our group suggested that the 22 kDa protein may correspond to the Fe‐Sod superoxide dismutase, SodB (unpublished data).

Fig. 1

Induction of His‐rich 22 kDa protein in Cd2+ efflux pump mutants in R. gelatinosus. Induction of CadA, CopI and the 22 kDa proteins in R. gelatinosus wild‐type (WT) strain, cadR − (A) and ΔcadA mutants (B).Total protein extracts from the same amount of cells (0.1 OD680nm) grown overnight in photosynthetic condition with indicated concentration of CdCl2 were separated on 15% SDS‐PAGE and analysed by Western blot using the HRP‐HisProbe. Growth of cadR − with 800 and 1000 µM of CdCl2 started after 3 days (very likely suppressors). CadA, CopI and the 22 kDa protein are indicated.

Induction of SodB in response to excess CdcadA mutant

Various SOD genes encode the enzymes responsible for superoxide dismutation and contribute to the protection of bacteria against ROS (Beyer et al., 1991; Imlay, 2003). These SODs vary by their metal cofactors but also by their subcellular localization. Analysis of the genome of R. gelatinosus highlighted the presence of two SOD encoding genes: sodC and sodB, which, respectively, encode the 15 kDa periplasmic Cu‐Zn SodC and a cytosolic 21.5 kDa Fe‐Sod. SodB displays a five‐histidine‐rich motif in its amino terminus sequence, suggesting that the HisProbe could allow SodB detection. To demonstrate that SodB corresponded to the 22 kDa band, we disrupted the sodB gene by insertion of a kanamycin resistance cassette within the coding region and checked for the presence of the 22 kDa protein. Western blot analysis confirmed that the band at 22 kDa was indeed absent in the inactivated strain ΔsodB (Fig. 2A). Next, we tested SOD activity in R. gelatinosus ΔsodB mutant. It is established that SOD mutants are more sensitive to the redox‐cycling agent menadione in the presence of oxygen. The susceptibility of ΔsodB to menadione was tested in comparison with the wild‐type by disc diffusion assays under photosynthetic and respiratory conditions. The ΔsodB mutant was found to be more sensitive to menadione than the wild‐type under both conditions. Growth inhibition zone for ΔsodB mutant reached 26 and 29 mm, whereas the inhibition diameter only spanned 16 and 18 mm under photosynthetic and aerobic conditions, respectively, for the wild‐type strain (Fig. 2B). Complementation of the ΔsodB mutant with the sodB gene cloned on a replicative plasmid restored a wild‐type phenotype on menadione, further confirming the role and activity of SodB in R. gelatinosus (Fig. S2). In‐gel SOD activity assay showed only one active band in the wild‐type soluble fraction, which was absent in the ΔsodB strain (Fig. 2C). These results strongly indicated that the 22 kDa protein corresponds to the SodB protein which has bona fide SOD activity. Note that, among Rubrivivax strains, two other strains (R. gelatinosus IL144 and R. benzoalyticus) have their sodB gene fused with the upstream gene, xseA, which encodes the 55‐kDa exonuclease VII large subunit (Fig. S3). While RT‐PCR experiment showed that these two genes were cotranscribed (Fig. S3), we did not detect any corresponding band at 70 kDa with the HisProbe in all tested conditions (Fig. 1 and Fig. S1). The presence of an internal promoter, a post‐transcriptional or post‐translational regulatory mechanism may lead to the expression of the intact SodB protein of 22 kDa. Nonetheless, the ΔxseA mutant was not sensitive to ROS compared to ΔsodB (data not shown). Furthermore, complementation of ΔsodB strain with pBxseAsodB or pBsodB restored a wild‐type phenotype (Fig. S2), confirming that XseA did not play a role in the SOD activity.

Fig. 2

R. gelatinosus SodB expression and activity. (A) Western blot analysis of wild‐type (WT) and ΔsodB mutant cells. Cells were grown by photosynthesis and total protein extracts from the same amount of cells (0.1 OD680nm) were analysed by Western blot. (B) Involvement of SodB in oxidative stress under photosynthetic (PS) or respiratory (RES) condition. Representative growth inhibition (left) and mean ± stdev from 3 independent experiments of the diameter of inhibition zone (right) are presented. (C) SOD in‐gel activity assay of indicated strains. Expression and activity profile of SodB in the WT, ΔcadA strain after various CdCl2 treatment of cells grown under PS (D) and RES (E) conditions.

To further examine the impact of increasing concentration of Cd2+ on the SodB expression and activity in the ΔcadA mutant under different oxygen tensions, cells were grown under photosynthetic or respiratory condition, and their soluble fractions analysed. Western blot analysis and in‐gel activity assay showed that the amount and activity of SodB increased in ΔcadA cells with increasing cadmium concentration under both growth conditions (Fig. 2D and E). However, the increase in expression and activity was more substantial under respiratory condition (Fig. 2E).

Loss of SodB did not influence Cd2+ tolerance in the wild‐type background, but remarkably, SodB was required in cells unable to detoxify Cd2+

Changes in the amount and activity of oxidative stress‐associated defence enzymes including SODs can be used as evidence for an increased oxidative stress in the cell. Yet, direct involvement of SodB in Cd2+ tolerance has not been demonstrated. To address this, growth inhibition in presence of increasing concentration of Cd2+ was monitored for the wild‐type, ΔsodB and ΔcadA strains under respiratory and photosynthetic conditions. ΔsodB and wild‐type strains exhibited comparable growth inhibition profiles under both conditions (Fig. 3A and B). In contrast, growth of the ΔcadA mutant was compromised by low concentration of Cd2+ (Fig. 3A and B).

Fig. 3

Involvement of SodB in Cd2+ stress in R. gelatinosus. Dose–response curves of the wild‐type (WT), ΔsodB, ΔcadA and ΔcadAsodB strains. Cells were grown for 24 h in the presence of different concentrations of CdCl2 either aerobically (A) or by photosynthesis (B). Results represent the mean of 3 independent experiments. Menadione disc diffusion assay of cells in RES condition with or without 25 µM CdCl2 (C). Results represent mean ± stdev from 3 independent experiments.

The Cd2+‐efflux ATPase CadA extrudes Cd2+ from the cytoplasm in R. gelatinosus. Thus, inactivation of cadA led to a decreased Cd2+ resistance (Steunou et al., 2020a). Notably, growth inhibition experiments showed that the ΔcadA mutant was more sensitive to cadmium under the respiratory condition than the photosynthetic condition (Fig. 3A and B), which correlated well with the stronger induction of SodB under respiratory condition than photosynthetic condition (Fig. 2D and E). Taken together, these data suggested that SodB would be involved in Cd2+ tolerance. Wild‐type R. gelatinosus was fairly resistant to low concentrations of CdCl2 in both respiratory and photosynthesis conditions (Fig 3A and B). Consistent to no apparent induction of SodB in the wild‐type (Fig. 1), ΔsodB mutant exhibited comparable growth inhibition profiles under both conditions (Fig. 3A and B). In contrast, growth of ΔcadAsodB double mutant was dramatically affected by cadmium in both respiratory and photosynthetic conditions in comparison with the wild‐type, ΔcadA and ΔsodB strains. Hypersensitivity of ΔcadAsodB strain to the concomitant presence of Cd2+ and superoxide was further investigated by disc diffusion assay (Fig. 3C and Fig. S4). To this end, cells were spread on agar plates with or without 25 µM CdCl2. Menadione soaked filter discs were used to assess growth inhibition. Under respiratory condition, growth inhibition zones were comparable, regardless of the supplement of 25 µM CdCl2 for the wild‐type, ΔsodB and ΔcadA strains. In contrast, the ΔcadAsodB strain exhibited a very strong phenotype in presence of both CdCl2 and menadione, as growth was completely inhibited on plate (Fig. 3C and Fig. S4). In presence of menadione alone, however, the diameter of bacterium‐free zone for ΔcadAsodB and ΔsodB was comparable. Altogether, these data showed that the high toxicity of Cd2+ in the absence of the CadA efflux pump was exacerbated by the absence of the superoxide detoxification system.

SodB was also crucial for survival under copper stress only in the Cu+‐efflux ATPase‐deficient mutant copA −

Similar to cadmium, copper is extruded from the bacterial cytoplasm mainly by the dedicated Cu+‐efflux ATPase CopA (Rensing et al., 2000; Arguello et al., 2007). In R. gelatinosus and E. coli, it was reported that accumulation of Cu+ in copA − mutant led to [4Fe‐4S] degradation (Macomber and Imlay, 2009; Azzouzi et al., 2013). To test whether Cu+ could also lead to the induction of SodB in copA − mutant, protein expression profiles and SOD in‐gel activity of R. gelatinosus wild‐type and copA − cells grown in presence of excess CuSO4 were compared to those of the ΔcadA strain grown in the presence of CdCl2. In the wild‐type background, cells grown with excess CuSO4 showed increased level of CopI, while amount and activity of SodB remained unchanged (Fig. 4A and B). In contrast, induction of CopI in the copA − cells was associated with an increased amount and activity of the superoxide dismutase SodB, with similar degree to ΔcadA cells exposed to CdCl2 (Fig. 4A and B). Then, it is conceivable that SodB would also be required for tolerance to excess Cu+. To address this, growth inhibition of the double mutant ΔsodBcopA − strain in presence of increasing concentration of CuSO4 was monitored under photosynthetic and respiratory conditions and compared with ΔsodB or copA − single mutants (Fig. 4C and D). Disruption of the sodB gene in the wild‐type background did not affect copper tolerance since ΔsodB mutant growth was comparable to that of the wild‐type. However, disruption of sodB gene in the copA − mutant background led to a strong decrease in Cu+ tolerance (Fig. 4C and D). Disc diffusion assay was also performed in presence of CuSO4 and menadione, and the results indicated the hypersusceptibility of the ΔsodBcopA − strain to Cu2+ and superoxide on solid medium (Fig. S5). These results confirmed that SOD was also required for bacterial survival under copper stress when the Cu+‐efflux system was defective.

Fig. 4

Involvement of SodB in copper stress in R. gelatinosus. Western blot analysis of wild‐type (WT), ΔcadA, copA − and ΔsodB mutants grown in malate medium supplemented or not (‐) with 100 µM CdCl2 (Cd2+) or CuSO4 (Cu2+) (A). Cells were grown by photosynthesis, and total protein extracts from the same amount of cells (0.1 OD680nm) were analysed by Western blot using the HRP‐HisProbe. SOD activity assay on non‐denaturing PAGE. Soluble fractions from the ΔcadA, copA, and ΔsodB mutants were separated on 15% PAGE (B). Dose–response curves of the WT, ΔsodB, copA − and ΔsodBcopA − strains. Cells were grown in presence of increasing concentration of CuSO4 by respiration (C) or under photosynthesis condition (D) for 24 h. Results represent mean ± stdev from 3 independent experiments.

Altogether, association between the induction of SodB by metal excess and the hypersensitivity to the cognate metal in the SOD‐efflux pump double mutant appeared to be common in R. gelatinosus. To understand whether this is a general phenomenon in bacteria, we tested induction of SOD by excess metal in different bacterial species.

Fe‐Sod activity and amount were also increased in V. cholerae under Cd2+ stress

Vibrio cholerae is an aquatic bacterium but can infect human intestine to cause diarrhoeal diseases. V. cholerae encodes genes for two cytoplasmic SODs, the Mn‐Sod (vc2696 or sodA) and the Fe‐Sod (vc2045 or sodB) on the genome. However, in‐gel SOD activity assay indicated that only SodB was active in our condition (Fig. 5A). V. cholerae also possesses Cd2+/Zn2+‐ and Cu+‐efflux ATPases, encoded by zntA (vc1033, hereafter referred to as cadA for clarity) and copA (vc2215) respectively. To assess the effect of increasing concentration of Cd2+ on SodB activity in V. cholerae, wild‐type and the cadA − mutants were grown in the presence of excess CdCl2 and the soluble fractions were analysed on non‐denaturing PAGE for SOD activity. In the wild‐type fractions, SodB activity was comparable between CdCl2 exposed and non‐exposed cells (Fig. 5B). In contrast, the cadA − mutant exposed to 5 µM CdCl2 showed significant increase in SodB activity in comparison with non‐exposed cells and the wild‐type. Western blot analysis confirmed the increase in SodB amount only in the cadA −‐deficient mutant when exposed to CdCl2 (Fig. 5B).

Fig. 5

SOD activity was induced in the ATPase mutants of V. cholerae and E. coli. Western blot and SOD in‐gel activity assay of soluble fractions from wild‐type (WT) and SodB − strains of V. cholerae (A). SodB activity and expression in V. cholerae were induced in the ATPase‐deficient mutants cadA − and copA − in response to 5 µM CdCl2 (Cd2+) (B) or 200 µM CuSO4 (Cu2+) (C) as shown by the in‐gel activity assay and on the Western blots. Effect of excess CuSO4 on growth of the V. cholerae ΔsodBcopA deletion strain. WT, the single mutants ΔcopA, sodB − and the ΔsodBcopA mutant were grown overnight in LB (D) or LB supplemented with 200 µM of CuSO4 (E), growth was monitored at 600 nm with a Tecan Infinite. Western blot and SOD in‐gel activity assay of soluble fractions from WT, SodA − and SodB − strains of E. coli (F). SodB activity was reduced, while SodA increased in the ATPase mutants zntA − and copA − in response to 20 µM CdCl2 (Cd2+) (G) or 100 µM CuSO4 (Cu2+) (H) as shown by the in‐gel activity assays and on the Western blots.

Similarly to the effect of cadmium in cadA −, both the activity and amount of SodB were also induced in the copA − V. cholerae cells when challenged with 200 µM CuSO4 (Fig. 5C). Note that induction of CopI (vc0260) was appreciated in copA − mutant and not in cadA − V. cholerae cells (Fig. 5B and C). We then assessed the requirement of SodB for excess Cu+ tolerance in the absence of CopA. We introduced ΔcopA mutation in sodB − strain and showed that the ΔsodBcopA double mutant has comparable growth rate in LB medium with the wild‐type and the single mutants (Fig. 5D). When cells were challenged by 200 µM CuSO4, sodB mutation did not affect copper tolerance, while ΔcopA mutant showed slight growth defect (Fig. 5E). Importantly, the ΔsodBcopA double mutant showed more sensitivity to copper than the ΔcopA mutant (Fig. 5E). Altogether, even though the phenotype was not as strong as in R. gelatinosus, we concluded that SodB was also important for metal tolerance in V. cholerae given its high induction under excess metal and the sensitive phenotype of the double mutant.

Mn‐Sod and Fe‐Sod activity and amount, respectively, increased and decreased in response to excess Cd2+ or Cu+ in E. coli

In contrast to R. gelatinosus and V. cholerae, both the Fe‐Sod (SodB) and the Mn‐Sod (SodA) are active in E. coli (Carlioz and Touati, 1986) (Fig. 5F). Note that unlike R. gelatinosus and V. cholerae, E. coli SodB contains only four histidines in its amino terminus and is not detectable on Western blot with the HisProbe. In contrast, SodA could be detected thanks to a five‐histidine motif (Fig. 5F). Geslin et al reported that the absence of SodB and SodA in E. coli resulted in an increased sensitivity to Cd2+, suggesting an interplay between SOD and excess cadmium (Geslin et al., 2001). To address how E. coli handles the expression and activity of the two SODs in response to excess Cd2+ or Cu+, impact of Cd2+ on SOD activity and amount in E. coli, wild‐type and cadA − (zntA −) mutants were examined. In‐gel SOD activity assay of soluble fractions showed no difference in SodB and SodA activities in the wild‐type samples in response to 20 µM Cd2+ (Fig. 5G). On the contrary, in the zntA − mutant, exposure to 20 µM Cd2+ resulted in a decreased SodB activity concomitant to a slight increase in SodA activity. Western blot analyses did not show any significant increase in the amount of SodA in the wild‐type samples, but showed a slight induction of SodA expression in the zntA − strain when exposed to excess CdCl2.

Similar to the cadmium effect, the addition of 100 µM CuSO4 to the growth medium of copA − strain significantly decreased the SodB and induced the SodA activity (Fig. 5H). This is in agreement with previous report in E. coli by Macomber and co‐workers (Macomber et al., 2007). Together, these findings clearly showed that both Cd2+ (redox non‐active) and Cu+ (redox‐active) stress in E. coli resulted in a decreased Fe‐Sod activity and increased Mn‐Sod activity. This suggested that Cd2+ or Cu+ may directly or indirectly activate SoxRS or trigger an iron limitation situation in E. coli.

Excess cadmium or copper led to the induction of SOD in P. aeruginosa, P. putida and B. subtilis

Previous transcriptional studies in P. aeruginosa, P. putida and in Gram + bacteria also reported slight increase of the SOD transcripts under many metal excess stresses, suggesting that SOD might protect cells under such conditions (Teitzel et al., 2006; Cheng et al., 2009; Yang et al., 2012; Tarrant et al., 2019).

P. aeruginosa genome encodes two different SODs, Mn‐Sod (SodA) and Fe‐Sod (SodB). SodB is the most abundant SOD in iron‐replete condition and stationary growth phase, while SodA is only expressed under iron limitation (Martins et al., 2018). We therefore examined whether excess Cd2+(20 µM) or Cu+ (200 µM) stress in the ATPase efflux mutants lacking CadA (ZntA) and CopA, respectively, affected activity and expression of SODs. Under both stresses, in the wild‐type soluble fractions, SodB activity and expression were not affected by excess metal (Fig. 6A). In contrast, SodB activity and expression were increased in the copA − and zntA − mutants exposed to CuSO4 and CdCl2 respectively. These data showed that as in R. gelatinosus and V. cholerae, excess copper or cadmium induced the expression of SodB in P. aeruginosa.

Fig. 6

SOD activity was induced in the ATPase‐deficient mutants of P. aeruginosa, P. putida and B. subtilis. SodB activity and expression in P. aeruginosa were induced in the ATPase mutants zntA − and copA − in response to 20 µM CdCl2 or 200 µM CuSO4 as shown by the in‐gel activity assay and on the Western blot (A). SOD (SodAB) activity and expression in P. putida were induced in the ATPase mutant copA − in response to 50 µM CuSO4 (B). SodA activity and expression in B. subtilis were induced in the ATPase mutant copA − under 200 µM CuSO4 stress (C).

In P. putida, in‐gel SOD activity assay showed only one active band, which corresponded to a SodA‐SodB heterodimer (Fig. 6B and (Heim et al., 2003)). To check the effect of metal stress on this Sod heterodimer, wild‐type and copA deletion mutants (Adaikkalam and Swarup, 2002) were subjected to 50 µM CuSO4 copper stress and SOD activity and amount were assessed. For the wild‐type, copper had no effect on SOD. In contrast, the copA − mutant showed increased SOD activity and amount under copper excess only (Fig. 6B). Both SodA and SodB are detectable by the HisProbe (both contain 5 histidine), and we were not able to discriminate between the two SODs due to a similar molecular weight. Nevertheless, excess copper was also linked to an increase in the superoxide dismutase activity in P. putida.

Bacillus subtilis wild‐type and ΔcopA mutants were also grown in presence of CuSO4, and their SOD activity and amount probed as above. Only one detectable band corresponding to the Mn‐Sod was present in B. subtilis as previously reported (Henriques et al., 1998; Inaoka et al., 1998). The activity and amount of SodA were significantly induced in the ΔcopA mutant challenged with 200 µM CuSO4 in contrast to the wild‐type (Fig. 6C).

Taken together, these data are in favour of a rather general mechanism involving the induction of SOD under the excess metal stress in bacteria. Importantly, the induction was triggered by both active and non‐redox‐active cations. The putative underlying molecular mechanisms that may govern such an induction are discussed below.

Discussion

Metal ions such as Fe2+, Cu+ or Co2+ are able to undergo different redox states and can induce in presence of oxygen, the production of reactive oxygen species via the Fenton and Haber–Weiss reactions (Elzanowska et al., 1995; Gunther et al., 1995). Even though Cd2+, Zn2+ or Hg2+ are not redox active, they can also induce the production of ROS (Xu and Imlay, 2012; Norambuena et al., 2019). Given their affinity for sulphur atoms, metal ions are also able to directly interact and inhibit the activity of many key enzymes whose assembly and/or activity depends on thiol/cysteine groups, even under strict anaerobic condition (Macomber and Imlay, 2009; Macomber and Hausinger, 2011; Barwinska‐Sendra and Waldron, 2017). Therefore, accumulation of metal in the cell is a real threat that can ultimately lead to cell death if the detoxification systems fail to remove the excess. In this study, we showed that the cell protection, particularly against Cd2+ and Cu+, relied primarily on the metal export system and then secondarily on the ROS detoxification enzymes. Indeed, the interplay between the two systems became manifest when both detoxification systems were inactivated. In R. gelatinosus, mutants devoid of both the efflux ATPase and the unique cytosolic SOD enzyme became much more sensitive to Cd2+ or Cu+ than single mutants. This highlights (i) the critical role of the metal efflux ATPases in the protection of bacteria against metal homoeostasis dysregulation and (ii) the importance of superoxide detoxification system when metal homoeostasis was dysregulated. In our proposed toxicity model (Fig. 7), CadA or CopA protects the cell and prevents metal‐related damage in the wild‐type strain and SodB is constitutively expressed to a basal level. In the absence of CadA or CopA, accumulation of Cd2+ or Cu+ in the cytoplasm would damage exposed [4Fe‐4S] clusters and limit growth of the bacterium. Degradation of [4Fe‐4S] clusters should increase temporarily the iron pool (released iron and/or iron uptake) that can ultimately generate oxidative stress if superoxide detoxification is not efficient. Like metals, reactive oxygen species such as superoxide damage [4Fe‐4S] clusters and increase intracellular released iron (Keyer and Imlay, 1996; Imlay, 2019). Therefore, in cells defective for both metal and superoxide detoxification systems, the simultaneous accumulation of metal and superoxide was not endurable by the cell and inevitably led to metal hypersensitive cells and growth inhibition. Cells facing both excess Cd2+ and Cu+ and superoxide accumulation would have significant decrease in their [4Fe‐4S] clusters content. Moreover, Cu+ and Cd2+ appeared to directly bind and inhibit components of the Fe‐S biogenesis ISC machinery in bacteria (Chillappagari et al., 2010; Tan et al., 2017; Roy et al., 2018), thus exacerbating the toxicity of metals and superoxide. Under such harsh conditions, stressed cells would not be able to repair or provide sufficient active [Fe‐S] clusters to sustain growth (Fig. 7). In addition, the presence of elevated level of superoxide with released iron from damaged clusters or imported by the uptake systems would participate in Fenton chemistry and catalyse hydroxyl radical formation in these cells, thereby contributing to cell growth inhibition. A similar scenario was reported to explain the toxicity of copper in macrophages, in which bacterial killing is caused by copper overloading and hydrogen peroxide production (White et al., 2009). Our findings, with both redox‐active Cu+ and non‐active Cd2+, support this scenario and suggest that metal overloading and superoxide synergistically poison cells because they both destabilize exposed [4Fe‐4S] clusters and likely target the same [4Fe‐4S] enzymes.

Fig. 7

Interplay between the metal efflux and ROS detoxifying system. Wild‐type (WT) cells induce the expression of the ATPase CadA to prevent Cd2+ accumulation and related damages and SodB to scavenge superoxide. In the ΔcadA deficient mutant, accumulation of Cd2+ in the cytosol damage exposed [4Fe‐4S] clusters, cells perceive the situation as an ‘iron‐starvation’ condition and respond by increasing Fe2+ import presumably to rebuilt [4Fe‐4S]. This led to the induction of the superoxide dismutase SodB in response to iron status. In cells defective in both metal and ROS detoxification systems, the concomitant accumulation of Cd2+ and ROS leads to metal hypersensitive cells and growth inhibition, because both superoxide and Cd2+ target enzymes with exposed [4Fe‐4S] clusters. This imbalance and loss of [4Fe‐4S] clusters is no longer tolerable by the bacterium.

One remarkable finding in this study is the induction of SOD expression and activity in the efflux ATPase mutants from different bacterial species when exposed to excess metal. This raised the question of the molecular mechanisms that underpinned SOD induction by excess Cd2+ or Cu+. It is well established in E. coli and other bacteria that Mn‐Sod and Fe‐Sod expressions are partly regulated by the iron status within the cell. This regulation involves the Fur repressor and the sRNA RyhB that downregulate non‐essential iron‐containing proteins when iron is limiting (Masse and Gottesman, 2002; Imlay, 2019). Under such condition, the Fe‐Sod expression is repressed, whereas the Mn‐Sod is upregulated to convert superoxide into H2O2 and protect cells from oxidative stress. Excess Cu2+ or Cd2+ in bacteria and eukaryotes was shown to interfere with iron homoeostasis. Many studies suggest that excess Cd2+, Ni2+, Co2+, Zn2+ or Cu+ may generate an Fe‐starvation signal and lead to the induction of iron uptake systems (Stadler and Schweyen, 2002; Yoshihara et al., 2006; Houot et al., 2007; Chillappagari et al., 2010; Xu et al., 2019). This could be a direct effect of [4Fe‐4S] cluster degradation. Indeed, Keyer and Imlay elegantly showed that ‘released iron’ originating from [4Fe‐4S] degradation was rapidly sequestrated, and iron uptake was then required to supply iron to the Fe‐S machinery in response to peroxynitrite stress (Keyer and Imlay, 1997). In E. coli, we observed a decrease of the Fe‐Sod and increase of the Mn‐Sod, which suggest that under excess Cu+ or Cd2+ stress, cells encounter a state of iron limitation. Furthermore, in E. coli, SoxR/SoxS directly induced the expression of the Mn‐Sod in response to superoxide or redox‐recycling compounds (Imlay, 2019). Therefore, if superoxide is produced under metal stress, then it could also explain the induction of Mn‐Sod under Cu+ or Cd2+ stress by SoxRS. Oxidation of the 2Fe‐2S cluster of SoxR by superoxide or redox‐active compounds activates SoxR (Imlay, 2013; Kobayashi, 2017; Imlay, 2019). Cu+ is a potent oxidant, Cd2+is not. We observed the same SodA induction in cells challenged with excess Cu+ or Cd2+, suggesting that if Cu+ and Cd2+ induction of SodA involves SoxRS, SoxR would be indirectly activated by these cations.

Unlike E. coli, soxRS genes are missing in R. gelatinosus genome. V. cholerae encodes SoxR homologue but not SoxS; however, the only available superoxide dismutase to protect the cell from the deleterious effects of superoxide is the iron‐dependent SodB. Therefore, in these bacteria, there is no other choice but to express SodB in spite of the putative iron limitation status. To corroborate this hypothesis, we have shown that like iron starvation, Cu+ and Cd2+ excess induced iron uptake system as attested by the accumulation of the iron‐binding protein FbpA in the periplasm of R. gelatinosus (Steunou et al., 2020b). The induction of Sod in the ATPase mutants is therefore indirect and is probably the consequence of iron homoeostasis dysregulation by excess Cd2+ or Cu+.

Metals such as Cd2+ or Cu+ can displace or replace Fe2+ in the ferric uptake regulator Fur and may therefore affect Fur‐DNA binding (Adrait et al., 1999; Mills & Marletta, 2005; Vitale et al., 2009). Nevertheless, if Cd2+ or Cu+ metallates Fur in vivo and allows binding to the Fur box, we may then expect a repression of the iron uptake systems and induction of the Fe‐Sod. Our results showed, however, an induction of Fe‐Sod and the FbpABC iron uptake system (Steunou et al., 2020b), which rather suggest that Fur is unloaded in the presence of Cd2+ or Cu+.

In Vibrio tasmaniensis, a pathogenic strain of oysters, the Mn‐Sod is active. In this vibrio, it was shown that the Cu+‐efflux system (including CopA) and the ROS detoxification system (including SodA) were highly induced intracellularly (more than 50 fold) and that they are required for vibrio intracellular survival in phagocytes and cytotoxicity to haemocytes (Vanhove et al., 2016). As for V. cholerae, for Vibrio shiloi (a coral pathogen), the Fe‐Sod was required for survival of the bacterium in the coral ectodermal cells (Vanhove et al., 2016).

In P. aeruginosa, transcriptional profiles of exposed cells to Cu2+ ions (Teitzel et al., 2006), Cd2+ or Zn2+ nanoparticles also showed an increase in the relative amount of the Mn‐Sod superoxide dismutase SodA transcript (Manara et al., 2012; Yang et al., 2012). The increase of SodA in these strains could be also related to iron homoeostasis, as several genes induced under iron‐limiting condition were upregulated in copper‐stressed cells. In P. putida, quantitative proteomic studies showed upregulation of SOD in Ni2+, Co2+ or Cd2+‐treated cells, suggesting an increased production of superoxide radicals due to the presence of metal (Cheng et al., 2009; Manara et al., 2012; Ray et al., 2013). These studies also pointed out for a role of iron starvation in the induction of SOD since they also report the increased level of iron import proteins under metal stress. In Gram + bacteria, the superoxide dismutase SodM was also significantly induced by CuSO4 in Staphylococcus aureus (Tarrant et al., 2019). In B. subtilis, transcriptional profiles of exposed cells to Cu2+ ions also demonstrated that copper stress induced genes required for iron uptake and a slight increase in the SodA transcript (Chillappagari et al., 2010). In the cyanobacterium Synechocystis sp. PCC 6803, high copper treatment also induced the expression of the iron superoxide dismutase sodB (Giner‐Lamia et al., 2014). Very recently, mercury (Hg2+) was shown to induce ROS scavenging genes transcripts including the Mn‐Sod in the thermophilic aerobe Thermus thermophiles. T. thermophiles Δsod mutant exhibited an increased sensitivity to Hg2+ (Norambuena et al., 2019).

Collectively, these results showed that SOD was induced under metal excess stress in bacteria. The mechanisms of induction might be both direct and indirect and very likely involved iron homoeostasis dysregulation and SoxRS system when present as in E. coli. The data also showed that the SOD scavenging activity was vital when the metal efflux systems were defective. Increase in the SOD amount and activity would protect metal‐exposed cells by decreasing superoxide concentration and hence would save some exposed [4Fe‐4S] clusters from degradation (Imlay, 2019).

Extensive use of antibiotics in health care and agriculture has led to an increase in antibiotic resistance (Asante and Osei Sekyere, 2019). Copper and zinc exhibit antimicrobial properties and are used as alternatives to antibiotics in farming and agriculture. Nevertheless, the use of high concentrations of these cations led to environment contamination and to co‐selection of antibiotic resistance genes (Baker‐Austin et al., 2006; Purves et al., 2018; Rensing et al., 2018; Bischofberger et al., 2020). To avoid such impacts, we need to significantly reduce the concentration of metal in metal‐based antimicrobial treatments. Our findings showed that superoxide detoxification system became determinant for bacterial survival under metal stress. Moreover, induction of SOD by excess metal appears to be a general phenomenon in bacteria; therefore, targeting the ROS defence system together with the metal efflux systems may allow lowering the concentration of metals in the metal‐based antimicrobial treatments in agriculture and farming.

Experimental procedures

Bacterial strains and construction of mutants

Bacterial strains and plasmids used in this study are listed in Table S1. Standard methods were performed according to Sambrook et al. (1989) unless indicated otherwise. To inactivate sodB, a 1.7 kb fragment was amplified using the primers RG345_sodBF and RG346_sodBR (Table S2) and cloned into the PCR cloning vector pGEM‐T to give pGsodB. sodB gene was inactivated by deletion of a 50 bp fragment and the insertion of the 1.2 kb Km cassette at the MscI and StuI sites within the sodB coding sequence. The resulting recombinant plasmid was designated pGsodB::Km. A 2.4 kb fragment obtained by PCR using RG369_xseAF and RG370_sodBR containing the entire xseAsodB gene was cloned into pGEM‐T to give pGxseAsodB. xseAsodB was subcloned in pBBR1MCS‐3 at the SmaI‐SacI sites. The resulting plasmid was designated pBxseAsodB. From this plasmid, xseA was deleted by digestion with SacII and SmaI giving the pBsodB plasmid. Those plasmids were used for the complementation of ∆sodB. Transformation of R. gelatinosus cells was performed by electroporation (Ouchane et al., 1996). The plasmid pGsodB::Km was used to transform the wild‐type, the copA − and ΔcadA strains. Transformants were selected on malate plates supplemented with the appropriate antibiotics under respiratory condition. Following transformant selection, template genomic DNA was prepared from the ampicillin‐sensitive transformants and confirmation of the antibiotic resistance marker’s presence at the desired locus was performed by PCR.

To construct ∆copA in V. cholerae, ~ 600 bp fragments upstream and downstream to the copA were amplified using primers oYo848 and oYo849, and oYo850 and oYo851 respectively (Table S2). Resulting fragments were cloned into SmaI site of pCVD442 vector (Donnenberg and Kaper, 1991) using Gibson Assembly (Gibson et al., 2009), resulting in pEYY345. ΔcopA mutation was then introduced in C6706 (wild‐type) and C6706 sodB::Tn strains, kindly provided by J. Mekalanos (Cameron et al., 2008), by allelic exchange (Donnenberg and Kaper, 1991).

Single‐gene deletion mutant library of E. coli (KEIO collection (Baba et al., 2006)) was obtained from the National BioResource Project, National Institute of Genetics, Japan. P. aeruginosa mutants were obtained from the Comprehensive P. aeruginosa Transposon Mutant Library at the University of Washington Genome Center (Jacobs et al., 2003). P. putida strains were kindly provided by Sanjay Swarup (Adaikkalam and Swarup, 2002), and B. subtilis strains were gift from Peter Graumann (Chillappagari et al., 2010).

Cell preparation, growth and growth inhibition

Rubrivivax gelatinosus was grown at 30°C, in the dark aerobically (high oxygenation: 250 ml flasks containing 20 ml medium, referred to as respiratory condition) or under light microaerobically (in filled tubes with residual oxygen in the medium, termed as photosynthetic condition) in malate growth medium. E. coli, V. cholerae, B. subtilis and P. aeruginosa were grown overnight at 37°C in LB medium. P. putida was grown at 30°C in LB medium. Antibiotics were added at following concentrations when appropriate: kanamycin, ampicillin, tetracycline, spectinomycin and trimethoprim (50 µg ml−1), streptomycin (100 µg ml−1).

Growth curves in respiratory condition were monitored at OD680nm with measurements taken every 15 min for 24 h using a Tecan Infinite M200 luminometer (Tecan, Mannerdorf, Switzerland). For growth inhibition under photosynthetic condition, strains were grown in filled tubes and OD680nm was measured after 24 h using the Tecan luminometer.

For disc diffusion assay, 200 µl (1 OD680nm) of overnight grown cells was mixed with 7 ml semi‐solid agar and uniformly spread onto solidified agar plates. Sterile discs (6 mm) were soaked with 10 μl of 10 mM menadione and placed in the middle of the plate. Plates were incubated overnight at 30°C under either photosynthetic or respiratory condition. The disc diffusion method was performed in triplicate, and the mean of the measured inhibition zones was calculated.

SOD in‐gel activity assay on non‐denaturing gel electrophoresis

About 20 µg of soluble proteins was separated on a 10% non‐denaturing polyacrylamide gel and stained for SOD activity as described in Weydert and Cullen (2010), with minor modifications. Incubation with TEMED (0.85%) and Riboflavin‐5‐Phosphate (56 µM) was performed for 15 min at light and room temperature (RT), followed by the addition of nitroblue tetrazolium (2 mg ml−1) and a 15 min incubation in the dark at RT. Gel was washed twice in ddH2O and left in ddH2O at RT on a light box until SOD‐positive staining appeared.

Western blot and HisProbe‐HRP detection

Equal amount of soluble proteins (20 µg) or disrupted cells (1 OD680nm) when indicated was separated on SDS‐PAGE and transferred onto a Hybond ECL Polyvinylidene difluoride membrane (GE Healthcare). Membrane was then probed with the HisProbe‐HRP (horseradish peroxidase, from Pierce) according to the manufacturer’s instruction. The HisProbe‐HRP allows detection of proteins (including Sod) exhibiting at least five His residues in their N‐ter domain. Those with only four His residues (E. coli SodB) are not detected with this probe. Positive bands were detected using a chemiluminescent HRP substrate according to the method of Haan and Behrmann (Haan and Behrmann, 2007). Image capture was performed with a ChemiDoc camera system (Bio‐Rad).

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this manuscript.

Author contributions

ASS and SO conceived of the study. ASS, MB, MLB, RT, AD, YY and SO acquired and analysed data. ASS, MB, AD, AKL and SO interpreted the data. ASS and SO wrote the manuscript.

Table S1 . Bacterial strains and plasmids used in this work.

Table S2 . DNA primers used in this work.

Fig. S1 . A 22 kDa protein was induced in the Cd2+‐ATPase deficient mutants under respiration. Induction of SodB in R. gelatinosus wild‐type (WT), cadR ‐ and ΔcadA strains with the increase of CdCl2 concentration. Cells were grown under respiration. Total protein extracts from the same amount of cells (OD680nm = 1) were separated on 15% SDS‐PAGE and analyzed by western blot using the HisProbe.

Fig. S2 . XseA is not required for menadione tolerance. A. Menadione disc‐diffusion assay (as described in Experimental procedures) results showing the complementation by pBxseAsodB and pBsodB. pBTc: empty plasmid. B. sodB constructs used in this work. pGsodB::Km was used to generate the ΔsodB mutant. pB derivatives (pBxseAsodB and pBsodB) were used for complementations of ΔsodB strain.

Fig. S3 . Genetic organization of xseA sodB orfs. A. xseA‐sodB gene fusion. xseA coding sequence is in blue and sodB in red. No stop codon was found between the two orfs. Putative promoters and transcripts and primers used in RT‐PCR are indicated. B. Protein sequence of the fusion. The five His detected by the HisProbe in SodB are shown in black and boldstyle. C. Analysis of sodB expression by RT‐PCR. Total RNA from wild‐type (WT) and ΔsodB strains grown under respiration was extracted according to Steunou et al. (2004). 2 µg of total RNA was reverse transcribed with the superscript II (Invitrogen) using the specific sodB primer RG348. 2 µl of the RT reaction was used for the PCR. We used either the primers RG433‐RG434 specific to xseA or the RG347–RG348 primers specific to sodB. Both PCRs were positive showing the presence of the xseAsodB transcript (arrow). Genomic DNA and 16SrRNA were used for controls. This experiment suggested the presence of xseAsodB transcript but does not exclude the presence of an internal promoter upstream of sodB. Putative promoters are shown in gray.

Fig. S4 . Effect of Cd2 and menadione stress in ΔsodB, and ΔcadAsodB strains. A. Menadione disc‐diffusion assay. Plates were incubated at 30°C overnight either under photosynthetic (PS) or respiratory (RES) conditions. B. Western blot analysis of wild‐type (WT), ΔsodB, ΔcadA, and ΔcadAsodB strains grown by photosynthesis with (50 µM) or without (−) cadmium, showing the induction of SodB. C. Effect of Cd2+ and menadione stress in ΔsodB, and ΔcadAsodB strains. Menadionedisc‐diffusion assay in plates supplemented or not with 25 mM CdCl2. Plates were incubated at 30°C over night under aerobic respiration conditions. The simultaneous presence of Cd2+ and menadione is deleterious for the ΔcadAsodB hyper‐sensitive strain.

Fig. S5 . Effect of CuSO4 and menadione stress in the wild‐type (WT), ΔsodB, copA −, and ΔsodBcopA − strains. Menadione disc‐diffusion assay in plates supplemented or not with 25 mM CuSO4 (as described in Experimental procedures). Plates were incubated at 30°C overnight under aerobic respiration conditions. The simultaneous presence of Cu2+ and menadione is deleterious for the ΔsodBcopA − hyper‐sensitive strain.

Click here for additional data file.