S. pneumoniae serotype 2 strain, D39 (WT), and its isogenic mutants were used in this work. Routinely, pneumococci were grown anaerobically in closed, completely filled 20 ml test-tubes, in a water bath without shaking, at 37°C in Todd–Hewitt (TH) broth with 0.5% (w/v) yeast extract (THY) to an approximate OD₆₂₀ of 0.25. Where specified, bacteria were grown in chemically defined medium (CDM) (Yesilkaya et al., 2009), with the exception that BCAA were omitted from the amino acids mixture, and added separately. D39 strains mutated in tpxD (ΔtpxD) and brnQ (ΔbrnQ), were grown in medium supplemented with 100 μg ml⁻¹ spectinomycin. The 6 ΔtpxD complemented strains 1–6 (ΔtpxDcomp1-6; Figure 2), were grown in medium supplemented with 100 μg ml⁻¹ spectinomycin and 250 μg ml⁻¹ kanamycin. The ΔcodY mutant (kindly donated by Dr. Calum Johnston, Laboratoire de Microbiologie et Genetique Moleculaires, France) was grown in medium supplemented with 20 μg ml⁻¹ trimethoprim.

The construction of ΔtpxD and its genetically complemented strain containing the entire tpxD-coding sequence (SPD1464) plus 83 bp upstream, was described previously (Hajaj et al., 2012). The brnQ gene disruption was achieved by allelic replacement mutagenesis as described previously (Song et al., 2005). Briefly, the upstream and downstream flanking regions of brnQ were amplified using the primer combinations LFSPD0546F/LFSPD0546R, for the upstream region, and RFSPD0546F/RFSPD0546R, for the downstream region, respectively (Table S1). Subsequently, using LFSPD0546F and RFSPD0546R primers, the flanking fragments were fused to the spectinomycin resistance marker, which had been amplified with specF and specR primers from pDL278 (Yesilkaya et al., 2000). The resulting fused fragments were purified and transformed into S. pneumoniae D39 as described before using competence stimulating peptide (Alloing et al., 1996). The transformants were selected on blood agar base (BAB) containing 5% (v/v) sheep blood supplemented with spectinomycin. The mutation was confirmed by PCR using different primer combinations, and DNA sequencing.

To construct the 6 genetically cis-complemented strains (ΔtpxDcomp1-6), the chromosomal region containing the entire tpxD-coding sequence preceded by a variable length of the proximal upstream sequence harboring different parts of tpxD putative regulatory elements, was amplified using the appropriate primers (Table S1), which were designed to include NcoI or BamHI sites. The amplicons were purified (Qiagen), digested and cloned into similarly digested pCEP (Guiral et al., 2006). An aliquot of ligation mixture was transformed into E. coli BL21 (DE3), and kanamycin resistant transformants were analyzed for the presence of recombinant plasmid using primers, malF and pCEPR, that anneal to either side of the cloning site. Then the purified recombinant plasmid was transformed into ΔtpxD as described previously (Alloing et al., 1996). The successful construction of modified strains was then confirmed by PCR using malF and pCEPR primers, and DNA sequencing.

The replacement of cys⁵⁸ to serine in TpxD was done by splicing overlap extension (Song et al., 2005). For this, the region surrounding cys⁵⁸ was amplified using malF and TPX-S1, and pCEPR and TPX-S2 primers (Table S1). The amplicons were then fused together using malF and pCEPR primers. The fused products were digested with BamHI and NcoI, and ligated into the same sites of pCEP. The replacement of nucleotide sequence was confirmed by sequencing using malF and pCEPR primers, and the recombinant plasmid was transformed into ΔtpxD as before. The resulting strain was designated as TpxDC58S.

Bacteria were grown anaerobically at 37°C in THY medium to OD₆₂₀ = 0.25 and then challenged with 100, 250 or 1000 μM H₂O₂ for 40 min while maintaining anaerobic conditions. Control cultures were incubated with Double-distilled water instead of H₂O₂. Viability of WT and ΔtpxD bacteria, challenged with 1 mM H₂O₂, was confirmed by colony forming unit counts, as previously described (Hajaj et al., 2012). Experiments were repeated at least twice, each with two biological replicates. RNA was prepared by using MasterPure^TM RNA purification kit (Epicentre®, USA). RNA was treated with DNase I according to the above kit protocol. cDNA was synthesized with Verso^TM cDNA kit (ABgene, UK). Real-time PCR reactions contained Absolute^TM blue QPCR SYBR mix ROX (ABgene, UK), and the expression of individual genes was determined using gene specific primers (Table S1). The transcription level of target genes was normalized to gyrA. The results were analyzed by the comparative C_T method (Livak and Schmittgen, 2001).

D39 and its isogenic ΔtpxD were grown anaerobically at 37°C in THY medium to OD₆₂₀ = 0.25 and then challenged with 1 mM H₂O₂ for 40 min while maintaining anaerobic conditions. Control cultures were incubated with Double-distilled water instead of H₂O₂. All other procedures regarding DNA microarray experiments (RNA isolation, RNA quality test, cDNA synthesis and labeling) were performed as described (Shafeeq et al., 2011). Microarray slide images were scanned using GenPix Pro 6.1 (MSD analytical technologies). Processing and normalization (LOWESS spotpin-based) of slides was done with the in-house developed MicroPrep software. DNA microarray data were obtained from three independent biological replicates hybridized to glass slides with a dye swap. Expression ratios were calculated from the measurements of at least seven spots. Differential expression tests were performed on expression ratios with a local copy of the Cyber-T implementation of a variant of the t-test. False discovery rates were calculated as described (van Hijum et al., 2005). A gene with p-value of < 0.005 and a fold change cut-off of 1.8 was considered differential expressed. We have used the Cyber-T server to analyze our Microarray data. Therefore, the p-values mentioned in this study (Supplementary Material) should be considered as Bayes corrected p-values.

Microarray data have been submitted to Gene Expression Omnibus (GEO) database under the accession number GSE65157.

The coding sequence of tpxD was amplified by PCR with tpx-for and tpx-rev primers (Table S1) from S. pneumoniae D39 and cloned into pRSETc vector (Invitrogen) between XhoI and KpnI sites. E. coli BL21 cells harboring the constructed plasmid were grown in Luria-Bertani broth supplemented with ampicillin (100 μg/ml) for 24 h. Cells were harvested by centrifugation and stored at −70 °C. The pellet was suspended in lysis buffer [50 mM Tris pH 8, 100 mM phenylmethylsulfonyl fluoride (PMSF)], disintegrated by sonication, and centrifuged at 4000 × g for 1 h. Proteins in the supernatant were loaded onto a Ni-NTA column (Adar biotec, Israel), and incubated for 1 h at 4°C. The column was then washed with 10 mM imidazole, and the recombinant protein eluted from the column with 100 mM imidazole. The eluted protein was run on SDS-PAGE and a band which was of the predicted molecular weight of TpxD was visualized. Polyclonal antibodies against TpxD were custom made by Sigma Aldrich, Israel, using the purified recombinant TpxD.

Pneumococci were lysed and subjected to SDS-PAGE. Separated proteins were electroblotted onto 0.45 μm nitrocellulose membrane (Bio-Rad), and probed with the polyclonal antibodies against TpxD. Antigen complexes were detected using Peroxidase Affinity Pure Goat anti rabbit IgG (Jackson ImmunoResearch) and visualized with SuperSignal West Pico Chemiluminescent substrate (Pierce). Densitometry was performed using ImageJ software (Schneider et al., 2012).

The sequence of tpxD upstream region in D39 was extracted from the National Center for Biotechnology Information (NCBI) database using the following entries: tpxD sequence gene ID: 4441865. Streptococcus pneumoniae D39 strain sequence ID: NC_008533.1. Promoter elements were predicted using the Softberry BPROM algorithm of bacterial promoters (http://www.softberry.com/berry.phtml?topic=bprom&group=programs&subgroup=gfindb). Known CodY-regulated genes in S. pneumoniae were retrieved from the RegPrecise database (Novichkov et al., 2010), and their CodY binding site aligned with the upstream sequence of tpxD, using ClustalW in BioEdit 7.0.9 (Hall, 1999). CodY binding motif was visualized using WebLogo 3.2 (Lewis et al., 2000).

The coding sequence of codY was amplified by PCR from Genomic DNA from D39 strain with CodYF and CodYR primers (Table S1). The amplicons were cloned into plasmid pLEIC using In-fusion Cloning kit (Clontech), which allows ligation independent cloning method based on the annealing of complementary ends of a cloning insert and linearized cloning vector. The recombinant construct was transformed into Fusion-Blue competent E. coli, and transformants were plated on LA containing 100 μg/ml ampicillin. A transformant was selected for sequencing to rule out any mutation. The construct DNA then was transformed into E. coli BL21(DE3) (Novagen, United Kingdom) for recombinant protein expression. The strain with the recombinant plasmid was grown in Luria-Bertani broth containing ampicillin (100 μg/ml) for 16 h at 25°C, and the expression was induced with 0.5 mM Isopropyl β-D-1-thiogalactopyranoside. The pellet was collected by centrifugation, resuspended in lysis buffer (50 mM Tris pH 8, 100 mM PMSF), and subjected to sonication. The protein purification used immobilized metal affinity chromatography (IMAC) resin and non-denaturing conditions as instructed by the manufacturer (Clontech). The column was washed with 20 mM imidazole and the recombinant protein was eluted using 400 mM imidazole. The identity of the protein was verified as previously described (Kazakov et al., 2007) by matrix-assisted laser desorption ionization–time-of-flight (MALDI-TOF) mass spectrometric analysis of tryptic digests of the products by the Protein nucleic acid chemistry laboratory at the University of Leicester.

Fluorescently labeled DNA probe representing the putative promoter region of tpxD was generated using Spd1464E1FAM and Spd1464E2 primers (Table S1), and S. pneumoniae D39 genomic DNA. After amplification, the probe DNA was gel purified. The binding reaction was set up by mixing a constant amount of DNA promoter probe (10 nM), and increasing amounts of purified and dialyzed His-tagged recombinant CodY (0–1 μM) in 5X binding buffer (20 mM Tris-HCl pH 7.5, 30 mM KCl, 1 mM dithiothreitol (DTT), 1 mM ethylenediaminetetraacetic acid (EDTA) pH 8.0 and 10% glycerol). When required, 2 mM H₂O₂ and BCAA (isoleucine 1.62 mM, leucine 3.48 mM, and valine 2.77 mM) were also added into the binding buffer 20 min after addition of the DNA probe. The binding reaction was incubated at room temperature for 30 min in a total volume of 20 μl. Then, the reaction mixture was subjected to non-denaturing PAGE (8%) for 40 min at 200 V. DNA-protein complexes were visualized using a Typhoon Trio+ scanner (GE Healthcare Life Sciences) with a 526 nm short-pass wavelength filter. The image was analyzed using LI-COR image analysis software.

D39 were grown under anaerobic conditions to OD₆₂₀ = 0.25 and then challenged with 1 mM H₂O₂ for 40 min while maintaining anaerobic conditions. Alkylation was performed as previously described (Hajaj et al., 2012). Briefly, in vivo reduced protein-thiols were blocked by resuspending phosphate buffered saline washed bacteria in urea buffer (0.1 M Tris pH 8.2, 1 mM EDTA, 8 M urea) containing 32.5 mM iodoacetic acid (IAA). Samples were then treated with DTT (3.5 mM) and incubated with iodoacetamide (IAM) (10 mM). Following alkylation, samples were resolved on 18% (w/v) SDS-PAGE, stained with Coomassie blue, and a band corresponding to ~29.5 kDa cut from the gel, reduced with 3 mM DTT (60°C for 30 min), and modified with 10 mM IAM in 100 mM ammonium bicarbonate (in the dark, room temperature for 30 min). The modified protein was then digested with modified trypsin (Promega) at a 1:10 enzyme-to-substrate ratio, overnight at 37°C. The resulting peptides were resolved by reverse-phase chromatography on 0.075 × 200-mm fused silica capillaries (J&W) packed with Reprosil reversed phase material (Dr Maisch GmbH, Germany). Mass spectrometry was performed by a Q-Exactive plus mass spectrometer (Thermo) in a positive mode using repetitively full MS scan followed by High energy Collision Dissociation (HCD) of the 10 most dominant ion selected from the first MS scan. The mass spectrometry data was analyzed using Proteome Discoverer 1.4 software Using Sequest (Thermo) algorithm searching against the Streptococcus pneumoniae D39 proteome from Uniprot and specific database. Semi quantitation was done by calculating the peak area of each peptide based its extracted ion currents (XICs). The area of the protein is the average of the three most intense peptides from each protein.

Results were filtered with 1% false discovery rate.

The significance of differences was determined by the unpaired t-test. p < 0.05 was considered significant.

For research involving biohazards, correct standard procedures have been carried out.

In a previous study (Hajaj et al., 2012), we showed that TpxD is an efficient H₂O₂ scavenger, and that its expression is up-regulated under aerobic compared to anaerobic growth conditions. In order to isolate the effect of H₂O₂ from other ROS formed under aerobic environment, D39 and ΔtpxD were challenged with sub-lethal H₂O₂ concentrations under anaerobic conditions. Similar viability, confirmed by colony forming unit counts, was measured in WT and the mutant cells challenged with 1 mM H₂O₂. Microarray profiling of D39 challenged with 1 mM H₂O₂ revealed that 217 genes were differentially expressed by a factor of 1.8 or more compared to unchallenged bacteria: 146 were down-regulated and 71 were up-regulated (Table S2). When the ΔtpxD mutant was challenged with 1 mM H₂O₂, global transcriptional response was not observed (Table S3). Classification of the genes affected by H₂O₂ in the WT into functional categories revealed that the most prominent differentially expressed genes included those associated with amino acid transport and metabolism (Table S4).

It was shown in eukaryotes that thiol peroxidases serve as transcriptional regulators of global gene expression in response to H₂O₂, in addition to enzymatic removal of H₂O₂ (Crooks et al., 2004). The finding that ΔtpxD response to H₂O₂ was significantly lower than that of the WT strain could originate from bacterial inability to respond to 1 mM H₂O₂ in the absence of TpxD scavenging activity. Alternatively, the low response could be due to TpxD role as a global transcriptional regulator in S. pneumoniae. Hence, we constructed a mutant in which the catalytic activity was disrupted by replacing the peroxidatic cys⁵⁸ by serine (TpxDC58S). We selected 7 genes, whose expression was affected by 1 mM H₂O₂ in the WT but not in ΔtpxD mutant and checked their transcription levels in TpxDC58S. As shown in Figure 1, the expression levels of these genes were unaffected by 1 mM H₂O₂ in TpxDC58S, unlike the WT strain. However, when lower H₂O₂ concentrations (100 and 250 μM) were applied, TpxDC58S mutant could respond in a way similar to that of the WT strain (Figure 1). These data suggest that TpxD scavenging activity is crucial for pneumococcal global response to increased H₂O₂ levels. The fact that a transcriptional response to 100 and 250 μM H₂O₂ was induced in cells harboring inactive TpxD (TpxDC58S), rules out the possibility that TpxD serves as a transcriptional regulator.

We have shown that TpxD expression and synthesis were significantly increased following a challenge with H₂O₂ (Hajaj et al., 2012). To discover the specific DNA-elements responsible for TpxD synthesis and regulation by H₂O₂, in silico tools were applied. BPROM algorithm predicted 5 core elements upstream of tpxD-coding sequence (Figure 2A). Based on these predictions, we constructed 6 genetically modified strains, in which ΔtpxD was cis-complemented with an intact copy of tpxD-coding sequence, preceded by a variable length of tpxD proximal upstream region: ΔtpxDcomp1 harboring only tpxD-coding sequence; ΔtpxDcomp2 harboring the sequence starting from the predicted Shine-Dalgarno element and up to and including tpxD-coding sequence; ΔtpxDcomp3 harboring the sequence starting from the predicted transcription start site and up to and including tpxD-coding sequence; ΔtpxDcomp4 and 5 harboring the sequence starting from the predicted −10 and −35 consensus regions of the promoter, respectively, and up to and including tpxD-coding sequence; ΔtpxDcomp6 harboring tpxD-coding sequence plus a region of 83 bp upstream of the gene (Figure 2A). Lysates of ΔtpxDcomp1-6 strains, grown anaerobically, were subjected to Western blotting using polyclonal antibodies against TpxD. No TpxD-band was seen in ΔtpxDcomp1-4 strains (Figure 2B). In ΔtpxDcomp5, basal TpxD protein level was detected, indicating that the Shine-Dalgarno, transcription start site, −10 and −35 consensus regions are essential for TpxD synthesis. However, ΔtpxDcomp5 did not show any up-regulation in response to 1 mM H₂O₂compared to the WT strain (Figure 2B). Only ΔtpxDcomp6 showed a significant increase in TpxD synthesis following a challenge with H₂O₂as measured by densitometry: 1.70 ± 0.13-fold, p < 0.05 (Figure 2B). These results point out that the regulatory elements essential for tpxD up-regulation by H₂O₂ are located within the immediate 83 bp upstream of tpxD-coding sequence.

The next step was to identify a candidate TF responsible for tpxD up-regulation under H₂O₂ stress. To this end, we clustered the genes, shown in the microarray to be affected by H₂O₂, according to their specific regulator, by using RegTransBase database (Fomenko et al., 2011). Clustering was done solely to highlight a potential TF, being aware of the fact that H₂O₂ invokes a general stress response, involving many transcription factors and genes, making it very hard to distinguish direct from indirect effects of H₂O₂. This analysis yielded 14 TF-clusters covering 60/217 of the genes found by the microarray to be significantly affected by 1 mM H₂O₂ (Table 1). It should be noted that 16 of the 60 genes were classified in 2 TF-clusters (based on RegTransBase database) and that the regulon grouping of some genes is controversial.

We then used ClustalW to check whether tpxD upstream sequence contains a putative binding site for one of the TFs reported in Table 1. This analysis identified a CodY-binding site of 15 bp, between −26 and −40, upstream of tpxD-transcription start site (AATCATTGGAAAATT) which is highly homologous with CodY-binding motifs of pneumococcal genes known to be under CodY regulation (Figure 3). This putative binding site shares 73% homology with CodY-consensus sequence in Lactococcus lactis (den Hengst et al., 2005; Guedon et al., 2005).

We then expressed and purified the pneumococcal CodY protein with the intent to test its in vitro binding to the motif upstream of tpxD-coding sequence. EMSA results showed indeed that CodY interacts with the DNA probe containing this region in a concentration dependent manner (Figure 4). CodY binding to tpxD upstream region was specific, first, as we demonstrated previously that CodY does not bind to the psaR promoter, which is known not to be regulated by CodY (Hendriksen et al., 2008). Second, we deleted the putative CodY binding sequence in the upstream region of tpxD, and found that the binding was abolished (Figure 4C). The effect of H₂O₂ on CodY binding to the motif upstream of tpxD-coding sequence was checked by the incubation of CodY with 2 mM H₂O₂ prior to the addition of target DNA. This experiment resulted in increased CodY affinity to its target: 400 nM were sufficient for CodY binding to DNA in the presence of H₂O₂ whereas 600 nM were needed without H₂O₂ (Figure 4A). CodY-DNA binding activity was shown to be increased by BCAA (Brinsmade et al., 2010). When the reaction mixture was supplemented with BCAA, DNA binding was observed already at 200 nM. In this case addition of H₂O₂ had no impact (Figure 4B). These data indicate that the 15 bp sequence serves as CodY binding site. Furthermore, H₂O₂ increases CodY binding to its DNA target under conditions where BCAA are not present.

Our attempts to inactivate codY were unsuccessful, probably due to its essentiality. Hence we used a codY-deletion mutant harboring suppressing mutations inactivating the fatC and amiC genes (Caymaris et al., 2010). To elucidate the influence of CodY on TpxD expression, bacteria were grown in CDM without BCAA. Under anaerobic conditions, tpxD expression was significantly reduced in ΔcodY compared to its wild type, 2.43 ± 0.25-fold reduction, indicating that CodY is an activator of tpxD, even without BCAA. A 4.00 ± 0.12-fold reduction in tpxD expression was measured when ΔcodY cells were challenged with 1 mM H₂O₂. Hence we conclude that CodY is required for tpxD up-regulation under high H₂O₂ levels.

To elucidate the role of BCAA in CodY regulation of TpxD expression and synthesis, D39 were grown in the absence of BCAA (Figure 5). This experiment revealed that BCAA had no effect on the basal level of TpxD under anaerobic growth conditions. In addition, TpxD up-regulation by H₂O₂ was less pronounced in bacteria grown without BCAA. These data indicate that BCAA are required for CodY-mediated, TpxD up-regulation by H₂O₂. We then checked whether BCAA itself affect tpxD expression in a mechanism independent of CodY. To this end, tpxD expression was assessed in ΔcodY grown with or without BCAA and challenged with H₂O₂. This comparison revealed that the presence of BCAA was accompanied by a 2.27 ± 0.20-fold reduction in tpxD expression in ΔcodY, suggesting that in the absence of CodY, BCAA interfere with tpxD expression in as yet unresolved mechanism.

BCAA intracellular level is determined by their biosynthesis through the ilv operon (Brinsmade et al., 2010) and by their transport through specific transporters (Belitsky, 2015). The microarray data showed that the BCAA- biosynthetic enzymatic machinery (ilv) was down-regulated following a challenge with H₂O₂ (Table S2). Concomitantly, the expression of brnQ, a BCAA transport system II carrier protein, showed a remarkable increase, 7.6-fold, following a challenge with 1 mM H₂O₂ (Figure 1), probably to compensate for the decrease in BCAA biosynthesis. Noteworthy is the fact that a ΔbrnQ mutant could not overcome a challenge with 0.5–1.0 mM H₂O₂ (data not shown), indicating the importance of BCAA for oxidative stress resistance. ClustalW analysis revealed a putative CodY binding site in the upstream region of brnQ (Figure 3). Taken together the fact that CodY functions as a transcriptional repressor of the ilv operon (Hendriksen et al., 2008) on the one hand, and our the finding of a putative CodY binding site upstream of brnQ coding sequence on the other, implies that CodY is involved in maintaining homeostatic levels of BCAA in the cell.

Microarray data showed that codY expression was increased by 1.8-fold in the presence of 1 mM H₂O₂ (Table S2). Validation by RT-PCR and Western blotting showed no significant change in CodY expression (Figure S1). This led us to conclude that CodY regulation by H₂O₂ is carried out by a different mechanism. Hence we checked the effect of H₂O₂ on the redox state of CodY-cysteines by a thiol double trapping method followed by mass spectrometry. Our data show that under anaerobic conditions, the 2 CodY-cysteines were always in the reduced state, as the pick area of the oxidized, IAM-alkylated cysteines, was zero. Following a challenge with H₂O₂, the pick area of the reduced, IAA-labeled cysteines, was 13E7 whereas that of the oxidized, IAM labeled cysteines, was 3.75E7. These data indicate that H₂O₂ triggers the oxidation of the 2 cysteines in about 1/4 of CodY molecules, which might influence CodY activity.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.