Bacterial pathogens encounter a variety of adverse physiological conditions during infection, including metal starvation, metal overload and oxidative stress. Here, we demonstrate that group A Streptococcus (GAS) utilises Mn(II) import via MtsABC during conditions of hydrogen peroxide stress to optimally metallate the superoxide dismutase, SodA, with Mn. MtsABC expression is controlled by the DtxR family metalloregulator MtsR, which also regulates the expression of Fe uptake systems in GAS. Our results indicate that the SodA in GAS requires Mn for full activity and has lower activity when it contains Fe. As a consequence, under conditions of hydrogen peroxide stress where Fe is elevated, we observed that the PerR-regulated Fe(II) efflux system PmtA was required to reduce intracellular Fe, thus protecting SodA from becoming mismetallated. Our findings demonstrate the co-ordinate action of MtsR-regulated Mn(II) import by MtsABC and PerR-regulated Fe(II) efflux by PmtA to ensure appropriate Mn(II) metallation of SodA for optimal superoxide dismutase function.
Trace transition metals are important bacterial nutrients, but can be toxic in excess. The concentration of available metal ions is tightly regulated in the host , particularly for Fe and Zn, which are sequestered within host tissues as pathogen control strategies . In addition to metal starvation, bacteria are also subject to killing via metal overload within innate immune cells. Examples include Cu- and Zn- dependent killing of Salmonella [3,4] and Mycobacterium tuberculosis within macrophages [5,6], and Zn-dependent killing of Streptococcus pyogenes within neutrophils . Bacteria also experience oxidative and nitrosative stress arising from reactive oxygen and nitrogen species (ROS and RNS) generated by innate immune cells, and bacterial pathogens have evolved defence systems that protect against these challenges . Importantly, there is close interplay between metal ion homeostasis and oxidative stress response. Metals such as Cu and Fe can cause toxicity by potentiating oxidative stress , while Mn is generally considered to have a protective role against ROS [10,11].
S. pyogenes (group A Streptococcus, GAS) is an obligate human pathogen that causes a wide spectrum of diseases, ranging from mild infections of the pharynx and skin through to severe, life-threatening diseases such as necrotising fasciitis and streptococcal toxic shock syndrome . GAS strains express a variety of virulence factors to subvert phagocytosis , abrogate immune responses, enhance inflammation , promote local tissue invasion , acquire metal ions , and defend against oxidative stress [17,18]. GAS uses superoxide dismutase (SOD) to detoxify the superoxide anion and, although it lacks catalase, it does possess peroxidases for the removal of hydrogen peroxide (H2O2) [18,19]. GAS encodes a single superoxide dismutase, SodA, which is thought to use Mn as a cofactor [20,21], although it should be noted that cambialistic SODs capable of using either Mn or Fe have been described in some Gram-positive bacteria, including Streptococcus pneumoniae and Staphylococcus aureus [22,23].
Given the differing pro- and anti-oxidant effects of Fe and Mn, it follows that GAS must tightly co-ordinate Fe and Mn homeostasis to meet the intracellular demand for these metal ions and maintain biochemical functions, while at the same time avoid the potential for oxidative stress. In many Gram-positive bacteria, including GAS, the peroxide response regulator PerR regulates the cellular response to oxidative stress and also contributes to Fe and Mn homeostasis [17,24–26]. The Fe-loaded form of PerR is highly sensitive to oxidation by H2O2, which in turn enables derepression of genes involved in the oxidative stress response . In GAS, these genes include pmtA, which encodes a P1B-4-type ATPase that effluxes ferrous iron and likely reduces potentiating the effects of excess Fe on H2O2 and superoxide stress [28,29]. In contrast, Mn-loaded PerR is H2O2-insensitive, and thus, it continues to repress the PerR regulon .
Mn import in GAS is mediated by the ABC-type metal transport system MtsABC, and this system is essential for protection against the superoxide generator, paraquat . The DtxR family regulator MtsR functions as a repressor that senses cellular Mn levels in GAS and regulates the expression of MtsABC [21,31]. GAS lacks a Fur homologue and instead uses its Mn sensor, MtsR, to also control Fe homeostasis via the expression of the haem acquisition accessory proteins Shr and Shp, the haem importer SiaABC and the ferric ferrichrome importer SiuABDG [32–36]. Recent findings in Streptococcus cristatus (formerly S. oligofermentans) have shown that MntR, an MtsR homologue, can also act as a H2O2 sensor and activate Mn import in response to H2O2 stress .
In this work, we examine the relationship between Mn import (via MtsABC and regulated by MtsR), Fe efflux (via PmtA and regulated by PerR), and defence against oxidative stress (via SodA) in GAS. We show that uptake of Mn occurs upon exposure to H2O2 and subsequently leads to activation of SodA. This influx of Mn requires a functional MtsABC, but does not rely upon up-regulation of mtsABC by MtsR. We further demonstrate that SodA from GAS is more active in the Mn-metallated form when compared with the Fe-metallated form and that optimal SodA activity also depends on a functional PmtA. These co-ordinated systems ensure optimal Mn metallation of GAS SodA for oxidative stress defence.
Bacterial strains and growth conditions
The invasive GAS strain M1T1 clinical isolate 5448 was used in this study . 5448 wild-type (WT) and isogenic mutant strains were routinely cultured at 37°C in Todd-Hewitt broth (Difco) supplemented with 1% (w/v) yeast extract (Merck) (THY). For growth on solid medium, strains were cultured on horse blood agar (5% (w/v)) or THY agar (1.5% (w/v)). Escherichia coli strains were grown in lysogeny broth (LB) . E. coli Top10 was used for the maintenance of pJRS233-, pLZ12-, and pBAD-derived plasmids (Supplementary Table S1). For selection in GAS, the antibiotics kanamycin (300 μg/ml), spectinomycin (100 μg/ml), and erythromycin (2 μg/ml) were used. Kanamycin (50 μg/ml), spectinomycin (100 μg/ml), and erythromycin (500 μg/ml) were used to select for plasmids in E. coli.
DNA manipulations and mutant construction
All GAS mutant strains, except 5448ΔsodA (see below) (Supplementary Table S1), were constructed as previously described . To construct the 5448ΔmtsR mutant, the 5′ and 3′ flanking regions of mtsR were amplified using 5448 WT genomic DNA as templates and primer pairs mtsR-KO-1 (incorporating an XhoI site at the 5′-end) and mtsR-KO-2 (with homology to aad9 (specR) at the 5′-end), and mtsR-KO-3 (with homology to aad9 at the 5′-end) and mtsR-KO-4 (incorporating a BamHI site at the 5′-end), respectively (Supplementary Table S2). The aad9 cassette was amplified from pUCSpec using primers spec-F and spec-R. The resulting fragments were fused by three-way PCR to form the mtsR-KO construct, which was ligated with the temperature-sensitive shuttle vector pJRS233. The pJRS233-mtsR-KO plasmid was electroporated into electrocompetent GAS as previously described . A double-crossover event was selected for by serial passaging at 37°C and 30°C, resulting in 5448ΔmtsR that contains specR (marker) and has lost eryR (shuttle vector). To generate the 5448ΔmtsR::mtsR complemented strain, the WT mtsR allele was amplified using primers mtsR-KO-1 and mtsR-KO-4, and ligated into pJRS233, yielding pJRS233-mtsR, which was electroporated into 5448ΔmtsR. The plasmid was integrated via double crossover at 37°C for replacement of aad9 with mtsR at the original locus, to yield 5448ΔmtsR::mtsR.
The construction of 5448ΔmtsABC was achieved by amplification of the 5′- and 3′-flanking regions of mtsABC using 5448 WT template DNA and primer pairs mtsABC-KO-1 (incorporating an XhoI site at the 5′-end) and mtsABC-KO-2 (with homology to aphA-3 (kmR) at the 5′-end), and primer set mtsABC-KO-3 (with homology to aphA-3 at the 5′-end) and mtsABC-KO-4 (incorporating a PstI site at the 5′-end), respectively (Supplementary Table S2). The aphA-3 gene was amplified from pUC4Ωkm2 using primers km-F and km-R. The resulting fragments were fused by three-way PCR to form the mtsABC-KO construct, which was ligated into the temperature-sensitive shuttle vector pJRS233. The pJRS233-mtsABC-KO plasmid was electroporated into electrocompetent GAS, and a double-crossover event was selected for by serial passaging as described above, resulting in 5448ΔmtsABC. To generate the 5448ΔmtsABC::mtsABC complemented mutant, the WT mtsABC allele was amplified using primers mtsABC-KO-1 and mtsABC-KO-4, and ligated into vector pJRS233, yielding pJRS233-mtsABC, which was electroporated into 5448ΔmtsABC. The plasmid was integrated via double crossover at 37°C for replacement of kmR with mtsABC at the original locus to yield 5448ΔmtsABC::mtsABC.
The construction of 5448ΔsodA was achieved using the temperature-sensitive E. coli-GAS shuttle vector pLZ12ts (Barnett et al., manuscript in preparation). The genomic regions flanking sodA from 5448 WT amplified using primer pairs sodA-1 and sodA-2 (5′), and sodA-3 and sodA-4 (3′). The kanamycin-resistance cassette was amplified from pUC4ΩKm2 using primers km-sodA-F and km-sodA-R, which also shared homology to the sodA gene (Supplementary Table S2). pLZ12ts was amplified using primers pLZ12-F and pLZ12-R. The four PCR products and linearised vector were assembled using the NEBuilder system (New England Biolabs) according to manufacturer's directions, and the resultant plasmid was sub-cloned E. coli Top10. The plasmid was electroporated into electrocompetent GAS and double-crossover mutants were selected for as described above, resulting in 5448ΔsodA that contains the kmR (marker) and has lost specR (shuttle vector).
The construction of a plasmid for the expression of GAS SodA was achieved through site-directed mutagenesis of pBAD-myc-His-A to remove the myc epitope and insert a TEV protease cut site upstream of the C-terminal 6xHis tag, using primer pairs pBAD-site-F and pBAD-site-R. The resulting plasmid, pBAD-TEV-His, was digested with NcoI and ApaI, and ligated to a sodA PCR fragment, which was prepared as follows. The sodA gene was amplified from 5448 WT using primers sodA-prot-F (incorporating a NcoI cut site at the 5′-end) and sodA-prot-R (incorporating an ApaI cut site at the 5′-end). Then, resultant PCR product was then digested with NcoI and ApaI and ligated into pBAD-TEV-His, and transformed into E. coli Top10. All strains were confirmed by DNA sequencing (Australian Equine Genetics Research Centre, The University of Queensland).
Analyses of intracellular metal content
For analysis of intracellular Mn accumulation, GAS strains were grown in THY to OD600 0.6–0.8 and challenged for 1 h with either 8 mM Fe(II) or 1 mM Mn(II). Sterile deionised water was used as a control. For metal accumulation following H2O2 exposure, strains were grown in THY to OD600 0.6–0.8. Following the collection of the T0 aliquot, the strains were challenged for 30 min with either sterile water, 1 mM, or 4 mM H2O2. Alternatively, for experiments examining metal accumulation profiles due to regulator deletion, strains were grown to OD600 0.6–0.8 in THY alone. For experiments to establish metal accumulation profile, strains were grown in THY alone or in the presence of 2,2′-bipyridyl (20 µM), Fe(II) (20 µM, 200 µM, or 2 mM), and/or Mn(II) (20 µM).
Cells were spun at 3200×g at 4°C for 15 min and washed three times in 1× PBS containing 250 mM EDTA, followed by three washes in 1× PBS, and a sample was taken to determine total protein content by the BCA assay (Thermo Scientific). Samples were digested in 70% nitric acid at 70°C for 48 h, diluted to 2% nitric acid with HPLC-grade H2O, and analysed for Fe and Mn by inductively coupled plasma mass spectrometry (ICP-MS) at the School of Earth and Environmental Sciences at the University of Queensland. Metal content was normalised to total protein content obtained from the BCA assay.
The metal content of rSodA proteins was assessed by ICP-MS. Briefly, protein concentrations of rSodA-Fe and rSodA-Mn isoforms were obtained by the BCA assay and solutions of each protein were adjusted to a final concentration of 10, 20, or 40 nM in HPLC-grade H2O with 2% HNO3. Total parts per billion values for metal were converted to nM in sample to establish molar equivalents of bound metal.
Viability of cells during hydrogen peroxide challenge
GAS strains were grown in THY to OD600 = 0.6–0.8. Upon the collection of a T0 sample, cultures were challenged with water, 1 mM H2O2, or 4 mM H2O2, and incubated at 37°C for 30 min. Following incubation, cultures were vortexed for 30 s, serially diluted, and plated onto THY agar plates to assess viable colony-forming units (CFUs).
Bacterial growth was assessed in the presence of varying amounts of Fe(II) (FeSO4·7H2O), Mn(II) (MnSO4·4H2O), Zn(II) (ZnSO4·7H2O), streptonigrin (Sigma), and methyl viologen (paraquat; Sigma). All salts were analytical grade (Sigma). Metal solutions were prepared in deionised distilled water and filter-sterilised. Streptonigrin was prepared in 100% ethanol and paraquat in distilled deionised water. Overnight cultures of GAS grown in THY were diluted in fresh THY to OD600 = 0.05 and supplemented with varying concentrations of each compound. Cultures were grown in flat-bottom 96-well plates (Greiner; final volume of 200 µl/well) without shaking at 37°C and optical densities at 595 nm were measured every 30 min using a FLUOstar Optima plate reader (BMG Labtech). Doubling time was calculated using the reciprocal of the gradient of natural log-transformed OD595 values during the exponential growth phase.
RNA extraction and qPCR analysis
Overnight GAS cultures were diluted to OD600 = 0.025 in THY broth with or without 2 mM Fe(II) or 0.5 mM Mn(II). At OD600 = 0.6–0.8, 5 ml of culture was harvested by centrifugation (8000×g, 10 min). Bacterial pellets were immediately resuspended in TRIzol (Invitrogen), transferred to lysing matrix B tubes (MP Biomedicals), and mechanically lysed using the FastPrep 120 instrument (Thermo Scientific). Chloroform (0.2 volumes) was added and the mixture was centrifuged (12 000×g, 30 min, 4°C). The aqueous phase was removed, supplemented with 70% ethanol, and loaded onto RNeasy columns (Qiagen) for purification following manufacturer's recommendations. RNA was eluted in ultrapure water and contaminating genomic DNA was removed using RNase-free DNase (TURBO, Ambion). RNA integrity was examined by gel electrophoresis and quantified by spectrophotometric analysis on the NanoDrop instrument (Thermo Scientific).
RNA (500 ng) was converted to cDNA using random hexamers and Superscript VILO (Invitrogen) following manufacturer's directions. qPCR was performed using SYBR Green mastermix (Applied Biosystems) with 100 nM primer mixes (Supplementary Table S2) and 2 ng of cDNA per reaction using an Applied Biosystems Quantistudio 6 instrument. Data were analysed using LinRegPCR  to obtain Cq values with gyrA as the reference gene, and data were plotted as ΔΔCq (Log2FC)  using 5448 WT in THY alone as the control, or WT at OD600 = 0.3 (early-exponential phase), OD600 = 0.6 (mid-exponential phase), or OD600 = 0.9 (late-exponential) for experimentation with 5448ΔperR.
Expression of rSodA
E. coli Top10 containing pBAD-Sod-TEV-His was cultured in 500 ml LB containing 100 μg/ml ampicillin supplemented with either 250 μM Fe(II) or Mn(II) at 37°C with vigorous agitation. At OD600 = 0.5, l-arabinose was added to a final concentration of 0.2% and cultures were shifted to 28°C for a further 4 h. Cells were harvested by centrifugation (5000×g, 4°C, 15 min), resuspended in buffer A [50 mM Tris (pH 8.5), 150 mM NaCl, and 5% glycerol] containing protease inhibitors (Roche complete EDTA-free), and lysed by sonication. Lysates were centrifuged (50 000×g, 4°C, 1 h) and rSodA was purified from the soluble fraction by fast protein liquid chromatography using a HisTrap column (GE Healthcare). rSodA was eluted using a 25–500 mM imidazole gradient in buffer A, and imidazole was removed by dialysis against buffer A using a 10 000 Da molecular mass cut-off membrane. The 6xHis tag from rSodA was removed by digestion with TEV  at 4°C overnight, followed by passing the reaction through a HisTrap column and collecting the flowthrough. The identity of tag-free rSodA (23.48 kDa) was confirmed by electrospray ionisation mass spectrometry. Protein concentration was estimated using absorbance at 280 nm with molar extinction coefficient of 40 910 M−1 cm−1 (calculated using ExPASy ) as well as the BCA assay (Thermo Scientific).
SOD activity assays
To determine the role of metal ions in SOD activity, overnight cultures were inoculated to OD600 = 0.025 in THY with or without Fe(II) or Mn(II). Strains were cultured to OD600 0.6–0.8, harvested by centrifugation (3200×g, 15 min, 4°C), washed with HEPES buffer (50 mM, pH 7.4), resuspended in 1 ml of HEPES buffer (50 mM, pH 7.4), and lysed mechanically on the FastPrep 120 instrument. For H2O2 challenge experiments, strains were grown to OD600 = 0.6 and exposed to water (control) or H2O2 (1 or 4 mM) for 30 min. Cells were harvested by centrifugation (3200×g, 15 min, 4°C), washed with HEPES buffer (50 mM, pH 7.4), resuspended in 1 ml of HEPES buffer (50 mM, pH 7.4), and lysed mechanically on the FastPrep 120 instrument.
GAS lysates were centrifuged (20 000×g, 2 min) and soluble fractions were used in SOD assay (Sigma–Aldrich) following the manufacturers' instructions. Protein concentrations in the lysates were determined using the BCA assay (Thermo Scientific). Lysates were assayed for SOD activity using a commercial SOD assay kit (Sigma–Aldrich) following the manufacturers’ instructions.
Purified rSodA-Fe or rSodA-Mn was prepared in HEPES buffer (50 mM, pH 7.4) and assessed for SOD activity using a commercial SOD assay kit (Sigma–Aldrich) following the manufacturers' instructions.
SOD gel activity assays
Whole cell lysates (40 μg) were resolved on native PAGE gels. Gels were incubated with 0.1% nitroblue tetrazolium (NBT) in deionised water for 1 h with gentle shaking, washed three times with water, and finally incubated with (−)-riboflavin (28 μM) in phosphate buffer (100 mM, pH 7) containing TEMED (28 mM) for 15 min in the dark with gentle shaking. Gels were washed three more times with water and subsequently exposed to light to promote colour development.
All data were analysed using GraphPad Prism 7. Analyses of ICP-MS and gene expression data were performed using either one-way ANOVA or two-way ANOVA (for comparison of strains under mixed conditions) as indicated. Two-way ANOVA was used to compare to either the WT control of that treatment or to assess differences within strains as a result of treatment, compared with H2O as control. For one-way or two-way ANOVA, Tukey's or Dunnett's post-hoc tests were used, respectively. Growth curve data represent mean ± standard deviation of at least four independent biological replicates. SOD assays show mean ± standard deviation of at least four independent biological replicates.
Mn is imported in response to hydrogen peroxide stress
We recently demonstrated that PmtA is a PerR-regulated Fe export pump . Based on these results, we hypothesised that intracellular Fe levels would fall in response to H2O2 shock as a consequence of PmtA activity. To determine changes in the intracellular metal levels of GAS during oxidative stress, we cultured 5448 WT in THY medium to the mid-exponential phase and subsequently exposed this culture to 0, 1, or 4 mM H2O2 for 30 min. Surprisingly, treatment with high concentrations of H2O2 (4 mM) led to a 1.5-fold increase in intracellular Fe levels, rather than the predicted decrease (Figure 1A), and this increase occurred independently of changes in bacterial viability, as determined by CFU counts (Supplementary Figure S1).
Metal accumulation in GAS in response to oxidative stress.
To assess the role of PmtA in the H2O2-induced increase in intracellular Fe, we next compared the levels of intracellular Fe in 5448 WT, 5448ΔpmtA, and 5448ΔpmtA::pmtA complemented strains following H2O2 challenge. Consistent with our recent observations , the 5448ΔpmtA strain accumulated 2.1-fold higher intracellular Fe levels when compared with the isogenic parent and complemented strains (Figure 1B). Similar to the 5448 WT and 5448ΔpmtA::pmtA complemented mutant, the amounts of intracellular Fe in 5448ΔpmtA also increased following exposure to 4 mM H2O2 (Figure 1B). While these results were consistent with our recent description of PmtA as an Fe efflux pump , they did not identify the mechanism underpinning the observed increase in intracellular Fe in response to oxidative stress (Figure 1A).
In addition to an increase in intracellular Fe levels, we detected a 3.7-fold increase in intracellular Mn levels in the 5448 WT strain following exposure to as low as 1 mM H2O2 (Figure 1C). This increase also occurred in the 5448ΔpmtA strain, but failed to reach levels observed in the 5448 WT and 5448ΔpmtA::pmtA complemented strains under any condition (Figure 1D). In fact, basal Mn levels in 5448ΔpmtA before exposure to H2O2 were reproducibly lower than those in the WT or complemented mutant strains (Figure 1D), an opposite phenotype to that observed for intracellular Fe levels (Figure 1B). Together, these findings suggest that there might be co-ordinated shifts of intracellular Fe and Mn levels in GAS, particularly during conditions of oxidative stress.
Mn uptake during hydrogen peroxide stress requires MtsABC
To determine whether the observed increase in intracellular Mn (Figure 1C) required the Mn importer MtsABC, we constructed the 5448ΔmtsABC deletion and 5448ΔmtsABC::mtsABC complemented mutant strains. As reported previously for M1 GAS , growth of 5448ΔmtsABC in THY was impaired, as evidenced by a prolonged doubling time (1.6 ± 0.2 h for this mutant vs. 1.2 ± 0.1 h for 5448 WT) and a lower stationary phase optical density compared with the wild-type and complemented mutant strains (Figure 2A). This growth defect was abolished by the addition of 10 µM Mn(II) (Figure 2B) but not Fe(II) or Zn(II) (Supplementary Figure S2). Consistent with the role of MtsABC in Mn uptake, we found that 5448ΔmtsABC was Mn-deficient when compared with the isogenic parent or the complemented mutant strains (Figure 2C). This deficiency was again alleviated by supplementing the culture medium with 20 µM Mn(II) (Figure 2C), indicating that alternative low-affinity or non-specific Mn import pathways may be present in GAS.
Effect of mutation of mtsABC on intracellular metal accumulation and growth.
In contrast with Mn, levels of intracellular Fe in 5448ΔmtsABC were identical with those in the wild-type or the 5448ΔmtsABC::mtsABC complemented strains, regardless of whether they were cultured in THY alone, in the presence of the Fe chelator 2,2′-bipyridyl (20 µM), or in the presence of Fe(II) (20 µM, 200 µM, or 2 mM) (Figure 2D), indicating that MtsABC did not facilitate Fe import under these experimental conditions.
We next examined the effects of oxidative stress on intracellular Mn levels in the 5448ΔmtsABC mutant. Unlike the 5448 WT or the 5448ΔmtsABC::mtsABC complemented strains, intracellular Mn levels in 5448ΔmtsABC remained low (∼8-fold less Mn compared with 5448 WT and 5448ΔmtsABC::mtsABC) under all conditions. Exposure to 1 mM H2O2 did result in a 2-fold increase in intracellular Mn levels in the 5448ΔmtsABC mutant (Figure 2E), but this effect was not statistically significant. These results established that the influx of Mn during conditions of oxidative stress was dependent on MtsABC. By contrast with Mn, intracellular Fe levels in 5448ΔmtsABC were not significantly different from that of the isogenic parent or complemented mutant strains (Figure 2F), again consistent with the proposal that MtsABC does not facilitate the import of Fe under these conditions.
Hydrogen peroxide stress does not induce the expression of mtsABC
The transcriptional response to oxidative stress in GAS is controlled by the H2O2-sensing transcriptional regulator PerR. However, a recent study in S. cristatus showed that the MtsR homologue MntR is also a H2O2 sensor . Hence, we hypothesised that the observed increase in intracellular Mn (Figure 1C) occurred following transcriptional up-regulation of MtsABC, either by MtsR or PerR, or both. To test this proposal, we examined the levels of mtsA mRNA in the 5448ΔmtsR and 5448ΔperR mutant and complemented strains, as well as the 5448ΔmtsABC mutant and its complemented strains in response to H2O2, using the same experimental conditions as in Figure 1.
Analysis of expression of mtsA in response to Fe, Mn, and H2O2.
As shown in Figure 3A, deletion of mtsR led to the constitutive up-regulation of mtsA regardless of the experimental condition, consistent with MtsR acting as a transcriptional repressor of mtsABC. As anticipated, other known MtsR-regulated genes such as shp, shr, and siaA were also constitutively up-regulated (Supplementary Figure S3A–C). In comparison, expression of pmtA, which is not regulated by MtsR, did not change in an MtsR-dependent manner (Supplementary Figure S3D). We further confirmed that transcription of mtsA was Mn-responsive and showed that the addition of Mn(II) led to down-regulation of mtsA expression in the 5448 WT strain (Figure 3A), consistent with repression by Mn–MtsR. In contrast, expression of mtsA remained unchanged in the 5448ΔperR mutant (Figure 3B). Under the same conditions, deletion of perR led to the constitutive up-regulation of known PerR-regulated genes pmtA, ahpC, and dpr (Supplementary Figure S4A–C). Together, these results indicated that mtsA is not part of the PerR regulon in GAS. Consistent with this finding, there is no Per box upstream of the mtsABC operon in GAS .
Consistent with the idea that PerR is not involved in the regulation of mtsABC, we detected a 2-fold decrease in mtsA expression, albeit not statistically significant, in the 5448 WT and 5448ΔmtsABC::mtsABC complemented strains following challenge with 1 mM H2O2 and a statistically significant decrease (∼5-fold) upon treatment with 4 mM H2O2 (Figure 3C). Transcription of mtsA in the 5448ΔpmtA mutant mirrored this trend (Supplementary Figure S5A). In comparison, the PerR-regulated genes pmtA, dpr, ahpF, and sodA were all up-regulated by H2O2 (Supplementary Figure S5C–J), the opposite trend from mtsA. This down-regulation of mtsA expression occurred as early as 5 min post-exposure to H2O2 (Figure 3D), whereas PerR-regulated genes dpr, sodA, and ahpF were all up-regulated within the same time period (Supplementary Figure S6A–C). Therefore, although the increase in intracellular Mn during oxidative stress required a functional MtsABC transporter (Figure 2E), it did not appear to involve transcriptional up-regulation of mtsABC by MtsR.
Exposure to hydrogen peroxide stimulates SOD activity and this process requires MtsABC
Having shown that exposure to H2O2 led to MtsABC-dependent uptake of Mn (Figure 2E) and increased transcription of sodA (Supplementary Figure S5I,J), we next determined whether the intracellular accumulation of Mn led to an increase in SodA enzyme activity. Following a short (30 min) exposure to 1 mM H2O2, we detected a ∼3-fold increase in SodA activity in the 5448 WT strain (Figure 4A and Supplementary Figure S7A). This result was consistent with the expected role of SodA activity in protecting the cell during conditions of oxidative stress . As a control, we confirmed that there was negligible superoxide dismutase activity in the 5448ΔsodA strain, regardless of H2O2 exposure (Figure 4A and Supplementary Figure S7A).
Superoxide dismutase activity and sensitivity analysis of 5448ΔmtsABC.
To determine whether the H2O2-mediated increase in SodA activity was dependent on the import of Mn via MtsABC, we repeated the above experiment using the 5448ΔmtsABC mutant strain (Figure 2C). We noted that basal activity of SodA in 5448ΔmtsABC, even before exposure to H2O2, was reproducibly lower when compared with the 5448 WT and 5448ΔmtsABC::mtsABC complemented strains (Figure 4A and Supplementary Figure S7A), consistent with previously reported findings . SodA activity was restored to wild-type levels by the addition of exogenous Mn(II) (40 µM) (Figure 4B and Supplementary Figure S7B), mirroring the finding that Mn deficiency in the 5448ΔmtsABC mutant was corrected by the addition of Mn salts (Figure 2C). Importantly, Mn supplementation did not affect background activity in the 5448ΔsodA mutant (Supplementary Figure S7B), indicating that any potential contribution from non-SodA Mn–protein or Mn–sugar complexes in superoxide quenching was negligible [47,48].
Following treatment with 1 mM H2O2, amounts of sodA mRNA in 5448ΔmtsABC increased to levels that were comparable with those in the isogenic parent and complemented strains (Supplementary Figure S5J). However, unlike 5448 WT and 5448ΔmtsABC::mtsABC, the increased expression of sodA in 5448ΔmtsABC was only accompanied by a moderate, 1.5-fold increase in SodA activity (Figure 4A). As shown in Figure 2E, this mutant accumulated lower intracellular Mn during conditions of oxidative stress. Thus, it was likely that some of the SodA enzymes in this mutant was produced in the apo- or incorrectly metallated form . Consistent with this view, growth curve analysis indicated that while all strains displayed identical growth in THY alone (Figure 4C), the 5448ΔmtsABC mutant was unable to grow in the presence of the superoxide generator paraquat (1 mM) (Figure 4D). Nevertheless, growth of this mutant was restored to wild-type levels upon supplementation of the growth medium with 10 µM Mn(II) (Figure 4E). As a control, 5448ΔsodA was unable to grow in the presence of 1 mM paraquat, regardless of Mn supplementation. Together, these results indicated that MtsABC is required for optimal metallation of SodA with Mn, but supplemental Mn may be imported via alternative transporters and become available to SodA. This corroborates previous findings in GAS  and Streptococcus gordonii .
SodA is less active in the Fe-metallated form
We were surprised to find that SodA in the 5448 WT and 5448ΔmtsABC::mtsABC complemented strains failed to activate upon exposure to 4 mM of H2O2 when compared with 1 mM H2O2 (Figure 4A), despite similar increases in sodA expression under these conditions (Supplementary Figure S5J). As shown in Figures 1D and 2E, Mn was imported under both conditions. However, the relative intracellular levels of Fe were higher upon exposure to 4 mM H2O2 compared with 1 mM H2O2 (Figure 2E,F). We therefore hypothesised that the increased relative abundance of intracellular Fe might outcompete Mn from SodA, leading to the formation of Fe–SodA, and that the Fe–SodA form was less active than Mn–SodA.
To investigate the metal dependence of SodA, we purified recombinant GAS SodA from E. coli. As we were unable to generate an apo-form of the enzyme for subsequent reconstitution with only Mn or Fe, we prepared Mn- or Fe-metallated forms of SodA from E. coli cultured in the presence of excess Fe(II) or Mn(II) salts (250 µM). ICP-MS analyses indicated that SodA isolated from Mn-supplemented E. coli cultures contained 1.2 ± 0.2 (SD) molar equivalents of Mn. This Mn–SodA form was relatively pure, and it contained only 0.06 ± 0.08 (SD) molar equivalents of contaminating Fe (Figure 5A). Conversely, SodA isolated from Fe-supplemented E. coli cultures contained 0.8 ± 0.1 (SD) molar equivalents of Fe and 0.2 ± 0.0 (SD) molar equivalents of Mn (Figure 5A). We were unable to remove the Mn–SodA contaminant from this mixture and so the enzyme was used as isolated. Subsequent measurements of SodA activity of both the Mn- and the Fe- forms demonstrated that the Mn–SodA displayed approximately four times as much activity as the Fe–SodA (Figure 5B). Some of the activity in the Fe–SodA sample may be attributed to Mn–SodA contamination, but nonetheless our results support our proposal that Fe–SodA was less active than Mn–SodA. Hence, we propose that GAS SodA may be a cambialistic enzyme, similar to SodM from S. aureus , though the GAS SodA appeared to display less activity with Fe as a cofactor.
Activity of recombinant SodA in Fe or Mn isoforms.
PmtA protects SodA from mis-metallation by Fe
We have recently shown that the 5448ΔpmtA mutant was sensitive to the superoxide generator paraquat . As shown in Figure 1B,D, 5448ΔpmtA displayed a higher intracellular Fe/Mn ratio when compared with the WT or complemented strains. We therefore hypothesised that the excess intracellular Fe may promote the Fenton reaction in the presence of paraquat, while Mn might quench this reactive oxygen species . As described recently , growth curve analysis showed that in THY alone, all strains grew equally well (Figure 6A), but the addition of 2 mM paraquat resulted in diminished growth of the 5448ΔpmtA mutant (Figure 6B), and we were able to rescue growth by adding excess Mn(II) into the growth medium (Figure 6C).
Superoxide sensitivity analysis and superoxide dismutase activity of 5448ΔpmtA.
Our experiments with recombinant SodA indicated that excess intracellular Fe may lead to mis-metallation of SodA, leading to generation of the less active Fe–SodA enzyme. To test this proposal, we first examined the effects of excess exogenous Fe(II) on SodA activity in the 5448 WT strains. As anticipated, we found that Fe treatment led to a complete loss of SodA activity to background levels (i.e. levels in 5448ΔsodA mutant strain) (Figure 6D and Supplementary Figure S7C). This reduction in SodA activity may arise from down-regulation of sodA by Fe(II), but there was only a small decrease (1.5-fold) in sodA gene expression under the same conditions (Supplementary Figure S9). As a control, we also measured SodA activity in 5448 WT cultures in the presence of excess exogenous Mn(II). As anticipated, we observed an increase in SodA activity in 5448 WT but not in 5448ΔsodA mutant (Figure 6D and Supplementary Figure S7C). Thus, our results indicated that the loss of SodA activity likely occurred from the incorporation of Fe into the enzyme.
In the 5448ΔpmtA mutant strain, which was Fe-rich and Mn-deficient, we detected a lower basal SodA activity when compared with the 5448 WT and 5448ΔpmtA::pmtA complemented strains (Figure 6D,E and Supplementary Figure S7C,D). Although exposure of 5448ΔpmtA to 1 mM H2O2 up-regulated transcription of sodA to levels that were comparable to those in the isogenic parent or complemented mutant strains (Supplementary Figure S5I), SodA activity in 5448ΔpmtA reached only half of that in 5448 WT and 5448ΔpmtA::pmtA (Figure 6E and Supplementary Figure S7D). These trends were mirrored by SOD in-gel activity assays (Supplementary Figure S8). Under these conditions of oxidative stress, we found that 5448ΔpmtA accumulated more Fe (Figure 1B) and less Mn (Figure 1D) than did the 5448 WT parent strain. Thus, our findings were again consistent with the suggestion that SodA may be partially metallated with Fe and that the Fe–SodA form was less active than Mn–SodA.
MtsR is critical for Mn and Fe homeostasis during oxidative stress
Although it is established that MtsR differentially controls the expression of Mn uptake genes (mtsABC) and Fe uptake genes (shp, shr, and siaA) (Figure 3A and Supplementary Figure S3A–C), its role in regulating the balance of Mn and Fe inside cells is not fully understood. Given the constitutive up-regulation of mtsA in a 5448ΔmtsR mutant, we anticipated that this strain would accumulate Mn. Contrary to this hypothesis, 5448ΔmtsR was Mn-deficient compared with 5448 WT and 5448ΔmtsR::mtsR (Figure 7A). Instead of Mn, 5448ΔmtsR appeared to accumulate Fe (Figure 7A), and consistent with this observation, 5448ΔmtsR was unable to grow in the presence of the Fe-dependent antibiotic streptonigrin (Figure 7B). These results can be explained by the constitutive up-regulation of Fe uptake genes shp, shr, and siaA in 5448ΔmtsR.
Fe and Mn content of perR and mtsR mutants and mtsA regulation in 5448ΔpmtA.
The Fe-rich and Mn-deficient phenotype of 5448ΔmtsR was reminiscent of 5448ΔpmtA (Figure 1B,D), but the underlying mechanism was likely different. The MtsR-regulated Fe uptake gene shp was not up-regulated in 5448ΔpmtA when compared with the isogenic parent or complemented strains (Supplementary Figure S5K). Hence, the increase in intracellular Fe in 5448ΔpmtA was not a consequence of Fe influx via MtsR-dependent Fe uptake systems, but rather was attributed to the loss of PmtA and the inability to remove excess Fe. Similarly, basal expression levels of mtsA in 5448ΔpmtA were largely unchanged when compared with those in the WT or complemented mutant strains (Figure 7C). Thus, the observed Mn deficiency was not related to a loss of MtsABC expression. Nevertheless, the addition of 2 mM Fe(II) to 5448ΔpmtA did result in a 28-fold down-regulation of mtsA (Figure 7C). We hypothesise that the excess intracellular Fe in 5448ΔpmtA may lead to the formation of Fe–MtsR and constitutive repression of mtsA. To date, the precise binding affinities of MtsR for Mn and Fe and their respective affinities (and/or allosteric free energies) to the mtsA, sia, shp, and shr promoters are unknown and warrant future investigation.
In the case of 5448ΔperR, we found that this mutant contained less Fe but more Mn relative to 5448 WT and 5448ΔperR::perR (Figure 7D). The Fe-deficient phenotype was consistent with constitutive up-regulation of the Fe efflux pump pmtA (Supplementary Figure S4A) [26,28]. This mutant was reproducibly less sensitive to streptonigrin when compared with the parent or the complemented mutant strains (Figure 7B). The Mn-rich phenotype was also reported previously for the ΔperR mutant of S. cristatus , but it could not be explained by our current understanding of PerR regulation, particularly considering the lack of a Per box upstream of mtsABC  and our findings that mtsA was not constitutively activated in 5448ΔperR (Figure 3B).
It is now established that during host–pathogen interactions, systemic and niche-specific Mn levels in the host are kept low by Mn-sequestering host effectors such as calprotectin [51,52]. Conversely, high-affinity uptake systems for Mn are known to be important for the virulence of many pathogenic bacteria, including PsaABC in S. pneumoniae [53,54] and MtsABC in GAS . Nonetheless, we have previously shown that excess Mn can be toxic to GAS and that this bacterium requires the cation diffusion facilitator, MntE, to efflux excess Mn . Our work herein describes the complementary roles of the Mn importer MtsABC and the Fe efflux pump PmtA in GAS Fe/Mn metal homeostasis and their contribution to oxidative stress defence, particularly in ensuring optimal metallation of SodA (Figure 8A). Our data also suggest an interplay or overlapping roles between MtsR and PerR in Fe and Mn homeostasis in GAS, but several outstanding questions remain.
Model of the interplay between Mn and Fe homeostasis and Sod metallation in 5448 WT and 5448ΔpmtA.
One surprising result from the present study is the observation that, while the increase in intracellular Mn in response to H2O2 required a functional MtsABC, the levels of mtsABC expression decreased (instead of increased) under oxidative stress conditions (Figure 3C). A potential rationalisation of this observation is that MtsABC activity increases in response to H2O2 stress. Our data demonstrated that GAS has a relatively high intracellular Fe/Mn ratio when grown in THY. This ratio mirrored the relative metal composition of THY, which contained 15.6 ± 0.7 µM Fe and 0.30 ± 0.01 µM Mn. We propose that extracellular Fe (in the Fe(II) form) in THY may compete for the Mn(II)-binding site in MtsA (the solute-binding protein) and subsequently inhibit Mn(II) import during normal conditions (Figure 8A). Consistent with this idea, intracellular Mn levels did not increase in the 5448ΔmtsR mutant, despite constitutive up-regulation of mtsABC. However, the addition of H2O2 into the culture medium (to enforce conditions of oxidative stress) may lead to oxidation of extracellular Fe(II) to Fe(III) and subsequent dissociation of Fe(III) from MtsA, allowing Mn import to occur. There was also a hint that excess intracellular Fe may block Mn uptake via MtsABC, with 5448ΔpmtA showing decreased intracellular Mn levels when compared with 5448 WT and 5448ΔpmtA::pmtA despite comparable levels of mtsA expression (Figures 7C and 8B). Whether the excess cytoplasmic Fe blocks Mn import by binding to the cytoplasmic domains of MtsABC and/or locking this transporter in a closed or non-functional state is unknown.
The Mn-rich phenotype of the 5448ΔperR mutant also requires further investigation. Previous microarray data reported that the mts operon was down-regulated in a ΔperR mutant of the M1 strain AP1 . In contrast, deletion of perR in the M3 strain 003Sm  or the M14 strain HSC5  did not appear to affect mts transcription. Furthermore, in M1T1 MGAS5005 perR mutant, mtsA was found to be up-regulated , but this experiment used metal-depleted THY medium, which may have provided different growth conditions compared with that used in the present study. Nonetheless, the pattern of regulation of mtsA in response to Fe and Mn that we observed is consistent with other published findings . In the 5448ΔperR mutant, the constitutive activation of pmtA and subsequent Fe efflux lead to low intracellular Fe. As a consequence, MtsR may sense these low Fe levels and activate MtsR-regulated genes including mtsABC. The import of Fe by Sia and Siu  may result in cycling of Fe but at the same time permit Mn to accumulate inside the cytoplasm due to activation of MtsABC .
An additional key question that arises relates to the mechanism of differential regulation of Mn uptake genes (mtsABC) and Fe uptake genes (shp, shr, and siaA) by MtsR (Figure 3A and Supplementary Figure S3). Previous work with MtsR from MGAS5005 indicated that the Mn and Fe forms of MtsR display distinct affinities for DNA. Mn–MtsR was shown to bind with high affinity to the mtsABC promoter, while Fe–MtsR exhibited a lower affinity to the same sequence. In addition, Mn–MtsR bound with high affinity to the sia and siu promoters, but exhibited low-affinity binding to the siu promoter in the presence of Fe. . In this work, we showed that, unlike mtsA, expression of shp appeared to increase (instead of decrease) in response to 4 mM H2O2. Even though MtsR from GAS and MntR from S. cristatus share high amino acid sequence identity (98% cover, 55% identity, 2 × 10−87E-value), including the H2O2-sensing cysteine residues Cys11 and Cys123, H2O2 did not induce the expression of mtsA in GAS as shown for S. cristatus . GAS MtsR lacks the Cys156 residue present in S. cristatus MntR. Cys156 has been shown to play a role in disulfide bond formation in response to H2O2 stress , and so its absence in GAS MtsR may explain the lack of H2O2-dependent gene expression of mtsA at low levels of H2O2. Thus, GAS appears to use both MtsR and PerR to control Fe homeostasis. Given the differing requirements of GAS for Fe and Mn, and the opposing effects of these two metal ions on SodA activity and the response to oxidative stress, the mechanisms for how Fe and Mn homeostasis are organised or co-ordinated represent an interesting topic for future studies.
A.G.T., K.Y.D., C.Y.O., M.J.W., and A.G.M. conceived the project and designed experiments. A.G.T. performed experimentation. T.C.B. provided reagents and critiqued experimental design. A.G.T., K.Y.D., C.Y.O., M.J.W., and A.G.M. wrote the manuscript and all authors edited the manuscript.
A.G.T. was supported by a Research Training Program Scholarship (Department of Education and Training, Commonwealth of Australia). K.Y.D. is supported by a Royal Society Research Grant [RSG/R1/180044]. C.Y.O. is a recipient of a Garnett Passe & Rodney Williams Memorial Foundation Research Fellowship. T.C.B. is supported by a Career Development Fellowship from the National Health and Medical Research Council (NHMRC)-funded ‘Improving Health Outcomes in the Tropical North: A multidisciplinary collaboration’ (Hot North; APP1131932). M.J.W. is supported by NHMRC grants APP1102621 and APP1071659. A.G.M. is supported by NHMRC Project grant 1084460.
The Authors declare that there are no competing interests associated with the manuscript.
These authors contributed equally to this work.