Bacterial strains and plasmids used in this work are listed in Table S1. Streptomyces strains were cultured in rich liquid media TSBY (30 g/L tryptic soy broth, 5 g/L yeast, 340 g/L sucrose) or minimal M9 media (6.8 g/L Na₂HPO₄, 3.0 g/L KH₂PO₄, 0.5 g/L NaCl, 1.0 g/L NH₄Cl) supplemented with 5 g/L w/v glucose and 0.12 g/L MgSO₄. Antibiotics were added to the medium as needed (50 μg/ml for apramycin).

To generate the reporter vector PdndB-xylE, which contains xylE under the control of the dndB promoter, a DNA fragment that included the xylE coding region, ribosomal binding site (RBS), and upstream region containing stop codons in all three reading frames was amplified from pIJ4083 (Tunca et al., 2007) by PCR with primers 3700F and 3700R (Table S2). The PCR product was digested using XbaI, and ligated into the XbaI-digested pSET152, generating the xylE reporter vector, pJTU3700. Next, a 760-bp fragment containing the intergenic region between dndA and dndB (from positions −387 to 373 with respect to the dndB transcriptional start site, determined in this study) was amplified from chromosomal DNA of S. lividans 1326 by PCR with primer PdndBF and PdndBR (Table S2). The PCR product was digested using BamHI, and ligated with BamHI-digested plasmid pJTU3700, producing the PdndB-xylE reporter plasmid pJTU3707. The reporter pJTU3707 was then introduced into S. lividans 1326 via conjugation using E. coli ET12567 (pUZ8002). The catechol-2,3-dioxygenase activity assay was performed as described without modification (Ingram et al., 1989).

To generate plasmid pJTU3707-10, the -10 region of the dndB promoter was removed from pJTU3707 using PCR. Briefly, after PCR amplification using pJTU3707 as the template, 1 μl of DpnI was added to the PCR reaction mixture and incubated at 37°C for 6 h to remove the template plasmid. The PCR product was transformed into E. coli DH5α, resulting in plasmid pJTU3707-10.

Disruption of sigR was performed by gene replacement via homologous recombination. For this purpose, two DNA fragments containing partial sigR sequence and its flanking region were amplified from S. lividans 1326 genomic DNA by PCR using KOD plus DNA polymerase (TOYOBO, Japan). A 1582 bp PCR product (primer 5505LF and 5505LR), and a 1552 bp PCR (primer 5505RF and 5505RR) were digested with XbaI and BglII, and EcoRI and BglII respectively. The plasmid pBluescript II was digested using XbaI and EcoRI. Then the PCR DNA fragments were inserted into pBluescript II, generating pJTU3730. The cassette containing apramycin-resistance gene and OriT was obtained from pIJ773 by PCR amplification using primer 773F and 773R (Table S2). The BglII-digested cassette from pIJ773 was inserted into BglII-digested pJTU3730, generating pJTU3731, which pJTU3731 was then introduced into S. lividans 1326 via conjugation mediated by E. coli ET12567 (pUZ8002) (Kieser et al., 2000). The resulting exconjugants were cultured on MS plates. Spores were harvested and spread again on MS plates containing apramycin. After culture at 30°C for 3 days, apramycin resistance (apr^r) colonies were selected and screened for sigR disruption mutants by PCR (using primer 5505TF and 5505TR).

To test survival following exposure to oxidative agents, strains of S. lividans 1326 and HXY6 containing the plasmid pIJ702 were used (Zhu et al., 2005). Fresh spores were pre-germinated according to the standard protocol without modification (Kieser et al., 2000). Briefly, stock spores were pelleted by centrifugation and re-suspended in TES buffer (0.05 M, pH8.0). The spores were heat shocked at 50°C for 10 min, cooled using cold tap water, and then an equal volume of double-strength Germination Medium (1% Difco yeast extract, 1% Difco casaminoacids, and 0.01 M CaCl₂) was added. The spores were then incubated at 37°C with shaking at 300 rpm for 2–3 h. Approximately 1 × 10⁷ germinated spores (calibrated by spectrophotometry using OD₄₅₀) were washed twice using distilled water, and then diluted 100-fold prior to peroxide treatment. For the H₂O₂ treatment, H₂O₂ was added to a final concentration of 20 mM, and spores were incubated at 30°C for 2 h. Samples were taken at 30 min intervals, and diluted 100-fold using distilled water. PAA was added to final concentrations of 200, 230, 260, and 290 μM, and spores were incubated at 30°C for 30 min. Cumene hydroperoxide (CHP) was added to final concentrations of 10, 12.5, 15, 17.5, and 20 mM, and spores were incubated at 30°C for 1 h. Diamide was added to final concentrations of 15, 30, and 45 mM, and spores were incubated at 30°C for 1 h. Controls consisted of samples not exposed to oxidants. Aliquots were plated on MMT media, incubated for 4 days at 30°C, and the number of viable cells was determined by colony counts.

Spores of S. lividans 1326 and HXY6 were inoculated into TSBY media. Following growth for 3 days at 30°C with shaking at 220 rpm, mycelia were collected by centrifugation. To measure in vivo PAA-scavenging, mycelia were washed twice using M9 media and re-suspended in the same media to an OD₄₅₀ of 0.2. After growing for 3 h, cultures were challenged with 500 μM PAA, incubated at 30°C with shaking for 3 h, and then samples were withdrawn at 30 min intervals. The concentration of residual PAA was measured in 96-well plates using Gen5 (Biotek) with the Hydrogen Peroxide Quantitative Assay Kit (Sangon Biotech), which is based on the ferric-xylenol orange assay and which can detect peroxides in the micromolar concentration range (Gay and Gebicki, 2002).

For the in vitro PAA-scavenging assay, mycelia were washed twice and re-suspended in buffer containing 50 mM potassium phosphate (pH 7.6) (P buffer). Cells were disrupted by sonication and then centrifuged at 15000 rpm for 20 min at 4°C to remove cell debris. Protein concentration was measured using the Bradford Protein Assay, and the protein concentrations of different strains were adjusted to the same level by dilution with P buffer. The total protein extracts were treated with 5 mM PAA for 2.5 h. Samples were then taken at 5 min intervals, and the PAA concentration was measured as described above.

S. lividans 1326 was cultured in TSBY medium to an OD₄₅₀ reading of 0.4, and then treated with 1 mM paraquat, 20 mM H₂O₂, 200 μM PAA, 3 mM CHP, or 3 mM diamide. Samples were collected by centrifugation at 0, 25, 50, or 75 min post addition of oxidants to the culture. For the analysis of peroxide scavenging genes, samples were collected at 15 min intervals. RNA was then isolated and RNA concentration was measured using the Nanodrop 2000 Spectrophotometer (Thermo Fisher Scientific Inc.). Genomic DNA was removed using RNase-free DNase I, and the absence of DNA contamination was further confirmed using PCR. Reverse transcription was performed using the cDNA Synthesis Kit (Thermo Fisher Scientific Inc.), and the resulting cDNA was used as the template for PCR amplification. The PCR product for dndB was 365 bp, and rrnA encoding 16s rRNA was used as the internal control. Forward and reverse primers are listed in Table S2. PCR products were examined by electrophoresis on a 1.2% agarose gel and visualized by staining with ethidium bromide.

For quantitative real-time PCR, reactions were performed on an ABI7500 Fast Real Time System (Applied Biosystems). A standard reaction mixture (25 μl) contained 1 or 150 ng cDNA (1 ng for reference gene, 150 ng for target gene), 12.5 μl SYBR Green qPCR Master Mix (Thermo Fisher Scientific Inc.), and 50 nM of the forward and reverse primers. The PCR product for dndB was 161 bp, and transcript amounts were normalized to the amount of 16s rRNA. The conditions for PCR amplification were 95°C for 10 min, followed by 40 cycles of 95°C for 15 s, 60°C for 1 min. A dissociation curve ranging from 55°C to 90°C in 0.5°C increments, with a dwell time of 30 s, was performed to assess the specificity of the reactions.

The 5′ RACE System Kit (ThermoFisher) was used to determine the transcriptional start sites of the dndA and dndB genes, according to protocols provided by the manufacturer. RNA was isolated from S. lividans 1326 grown in TSBY medium. First-strand cDNA was synthesized using SuperScript reverse transcriptase and a gene-specific primer, GSP1 (A-GSP1 for dndA, B-GSP1 for dndB). RNA was removed by treatment with RNase Mix, and the cDNA was purified on a S.N.A.P Column (ThermoFisher). A poly(dC) tail was added to the 3′-end of the cDNA using terminal deoxynucleotidyl transferase and dCTP. The first PCR amplification was performed using rTaq polymerase (TAKARA BIO INC.), a nested, gene-specific primer, GSP2 (A-GSP2 for dndA, B-GSP2 for dndB), which anneals to a site within the cDNA, and AAP (Abridged Anchor Primer). The second PCR amplification used AUAP (Abridged Universal Amplification Primer) with another nested, gene-specific primer, GSP3 (A-GSP3 for dndA, B-GSP3 for dndB), the sequence of which is located upstream of GSP2 and within the cDNA product. The product of the second PCR was ligated into pMD18-T vector for sequencing.

Spores of the wild type S. lividans 1326 and the HXY6 were grown in TSBY media to an OD₄₅₀ reading of 0.3. The cultures were then collected by centrifugation and total RNA was isolated using RNAprotect Bacteria Reagent and RNeasy Mini kits (Qiagen), followed by further purification using a MICROBExpress kit (Ambion) to enrich the mRNA. The resulting mRNA was fragmented using an RNA fragmentation kit (Ambion), producing fragments of sizes ranging from 200 bp to 250 bp. Double-stranded cDNA was generated using the SuperScript Double-Stranded cDNA Synthesis Kit (Invitrogen) and purified using a QiaQuick PCR extraction kit. An RNA-seq library was constructed using a Paired End Sample Prep kit (Illumina), and the sequencing was performed using the HiSeq 2000 sequencer (Illumina). Raw reads were filtered to remove dirty reads as recommended by Illumina, and then the resulting clean reads were analyzed with TopHat (version 2.0.5) software (Langmead et al., 2009), using the annotated S. lividans 1326 genome as reference. We employed R packages edgeR (version3.3; Robinson et al., 2010) for differential gene expression analysis in this study, using multiple hypothesis test for the p-value correction (q-value), and taking in to account the false discovery rate (FDR). Fold-change was calculated according to the FPKM (Fragments Per Kilobase of exon model per Million mapped reads) value (screening criteria q-value 2 ≤ 0.05, and Fold-change ≥ 2). Gene Ontology (GO) enrichment analysis of differentially expressed genes was analyzed using the GOseq R package, with gene length bias corrected (Young et al., 2010). GO terms with q-value < 0.05 were considered significantly enriched by differentially expressed genes.

To investigate the potential consequences for host cells, germinated spores of S. lividans 1326 wild type (WT) strain and its mutant daughter strain HXY6, from which the dnd gene cluster was removed, were treated with H₂O₂, PAA, or CHP. PAA and CHP were tested as representatives of organic hydroperoxides. After treatment, the spores were spread onto agar plates, and colonies were counted to calculate the survival rates (Figure 1). WT had a two-fold higher survival rate than HXY6 following H₂O₂ treatment, and a 10-fold higher survival rate than HXY6 following treatment with 20 mM CHP or 290 μM PAA (using mycelia, similar result was obtained for PAA treatment, Figure S1). These observations suggest that phosphorothioation of the DNA confers an anti-oxidation survival advantage to the host.

The global change in gene expression analysis in response to phosphorothioation modification was conducted to investigate whether DNA phosphorothioation up-regulates genes that express peroxide scavengers and that contribute to the anti-oxidative properties associated with the modification. The mRNA levels of each gene were compared between WT and the dnd⁻ mutant HXY6. A total of 206 genes had significantly different expression, as measured by a 2-fold (log2- ratio. 1, p-value = 0.05) or greater difference between the two strains, with 146 genes up-regulated (Table S3) and 60 genes down-regulated (Table S4) in the mutant (Figure 2). Gene Ontology enrichment analysis shows that the differentially expressed genes scattered to seventeen pathways (or molecular functions), and none of them has p-value lower than 0.05 and at the same time contains more than 5 genes (Figure 3). Although a bunch of transcriptional regulators were seen in the differential expressed genes (Tables S3, S4), homology analysis of these genes indicated that these genes are scattered to multiple different pathways. Gene Ontology enrichment analysis also indicated that none of these regulator genes has p-value lower than 0.05 and concentrate in pathway containing more than 5 genes. These observations suggested that the global RNA transcriptions of the two strains are essentially similar. Particularly, differences of anti-oxidative genes (including catalase, alkyl hydroperoxide reductase CD, and organic hydroperoxide resistance genes) between the two strains were not seen. The resistance to hydrogen peroxide conferred by DNA phosphorothioation might not involve up-regulation of these anti-oxidation genes in advance.

Quantitative RT-PCR analysis was conducted to characterize the transcriptional response of peroxide scavenging genes following the loss of DNA phosphorothioation. There are five catalase (SLI_RS01475, SLI_RS02375, SLI_RS02905, SLI_RS31365, and SLI_RS37155), two alkylhydroperoxidase reductases (SLI_RS25315 and SLI_RS25320), and three organic hydroperoxide resistance genes (SLI_RS15855, SLI_RS12925, and SLI_RS32700) in the genome. Transcriptional changes of these genes were traced in 45 min after peroxide treatments (Figure 4, and Figure S2 for the removal of hydrogen peroxide).

Figure 4A shows that the five catalase, and two alkylhydroperoxidase reductase genes were activated by 20 mM hydrogen peroxide, reaching peaks in 15 min after the treatment. Two organic hydroperoxide resistance genes (SLI_RS12925 and SLI_RS32700) did not response to the hydrogen peroxide treatment. Notably, Sli_RS15855, the homolog of ohr in S. coelicolor, was up-regulated hundreds of folds in 15 min. The eight responsive genes were more activated in the dnd⁻ mutant than the wild type strain. Similar responses were also observed for the PAA treatment (Figure 4B, and Figure S2 for the removal of PAA).

The observation that catalase and organic hydroperoxide resistance genes (thiol-dependent oxidase; Cussiol et al., 2003, 2010), were more induced in the dnd⁻ strain by H₂O₂and PAA treatments, reaching levels higher than the wild type, suggested that the DNA phosphorothioation system compensated for the increased catalase and peroxidase activity present in the mutant.

PAA, which is an organic hydroperoxide, was used to test whether the WT strain can scavenge the peroxide directly. Results showed that PAA was scavenged more rapidly by WT than by the dnd⁻ mutant, both in vivo (500 μM) and in vitro (5 mM) immediately after peroxide exposure (Figure 5), supporting the suggestion that the modification system constitutes a new type of organic hydroperoxide resistance system in S. lividans.

Determination of the transcriptional response of the dnd promoter to oxidants might provide additional evidence for its physiological function as an antioxidation system.

5′ RACE was conducted first to determine the transcriptional start site of the dnd gene cluster. Gene specific primers were designed to amplify fragments of cDNA of dndA and dndB. After the addition of poly(dC) tails, PCR was performed, and the second round of nested PCR generated clear amplification products (Figure 6A). The fragments were then cloned and sequenced, obtaining the transcriptional start sites of dndA and dndB. The -10 regions of dndA and dndB promoters are quite similar, containing the consensus sequence TA(t/g)CGT. Two kinds of repeat sequences, the R repeat (R1-R2-R3) and the r repeat (r1-r2) were also found in the -35 region (Figure 6B).

The location of the dnd promoter was further characterized using the xylE reporter gene. A DNA fragment containing the region from −387 to 373 bp with respect to the dnd transcriptional start site was inserted in front of the xylE gene (Figure 6C), and to measure xylE expression levels, total protein was extracted and catechol 2,3-dioxygenase activity was assayed. Two strains were compared, the WT and a mutant from which the -10 region of the dndB promoter was removed from upstream of xylE. As shown in Figure 6D, with the deletion of the -10 region from the dndB promoter, the catechol 2,3-dioxygenase activity dropped sharply compared to the wild type control.

RNA was extracted and RT-PCR was performed to monitor the transcription of the dnd gene cluster. As shown in Figures 7A,B, the dnd gene cluster is transcribed early during growth, corresponding to log phase growth of the strain. The early expression of the dnd gene cluster was further confirmed using the xylE reporter construct (Figures 6B,C,7B).

At the mid-log phase of growth, the wild type strain was treated using 20 mM H₂O₂, 1 mM paraquat, 3 mM CHP, 3 mM PAA, and 3 mM diamide, the concentrations for which were chosen based on published studies (Lee et al., 1993; Kallifidas et al., 2010). These are model compounds that can induce the expression of superoxide dismutase, catalase, and organic hydroperoxide resistance genes. Diamide is special for its capability that can induce the expression of both superoxide dismutase, and catalase genes via thiol-based switches (Privalle et al., 1993; Zheng et al., 1998; Antelmann and Helmann, 2011). RNA was then isolated and quantitative RT-PCR was conducted to determine whether the dnd transcription is subjected to oxidant regulation. Transcription from the dnd gene cluster decreased slightly after 1 mM paraquat treatment (Figure 7C), but with H₂O₂, CHP, and PAA treatment, there was no obvious change in transcription within the same time period of 75 min (lower concentrations from 100 μM to 1 mM were also tried, with similar results observed). However, with diamide treatment, transcription increased about 2.5 fold in 50 min.

Since diamide induced the expression of dnd genes, we tested the survival of the strains under diamide stress. Figure 8 shows that there arenÕt significant differences between WT and the dnd⁻ mutant.

In S. coelicolor, a sigmaR-RsrA regulation system composing of a sigma factor and a redox sensor anti-sigma factor (Paget, 1998; Kang et al., 1999), responds to diamide treatment. We disrupted sigR gene, and then tested whether dnd transcription is subjected to SigR regulation. Figures 9A,B shows that there is no obvious dnd gene expression difference between WT and the sigR⁻ mutant after diamide treatment. Therefore, the response to diamide is not mediated by SigR.

DndB is the repressor of dnd gene cluster. The possibility of DndBÕs response to diamide was explored. Figure 9C shows that in dndB in-frame deletion mutant, the response to diamide stimulation is lost. DndB might therefore participate in the diamide induction process.

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.