The genome of D. radiodurans R1 was screened for genes encoding MazEF homologs by the BlastP program¹ using established E. coli MazEF (EcMazEF) as query sequences. Two pairs of putative MazEF-encoding genes were found in the D. radiodurans genome: dr0416-dr0417 and dr0661-dr0662, which are consistent with the predicted sequences in the Toxin-Antitoxin Database (TADB) (Shao et al., 2011). DR0417 and DR0662 share 46 and 29% identities, respectively, to E. coli MazF, indicating that dr0417 and dr0662 might encode the toxin proteins. The dr0416 and dr0661, which are upstream of dr0417 and dr0662, respectively, might encode the antitoxin. DR0417 contains several conserved amino acid residues, such as Gly22, Arg29, Pro30, and Pro59 (Supplementary Figure S1), which are required for substrate binding and toxin activity (Simanshu et al., 2013), while DR0662 showed less conservation in the substrate binding sites, e.g., Ala21 and Asp59 (Supplementary Figure S1).

To determine whether two pairs of MazEF homologs are functional TA systems, each MazEF homology module was overexpressed in E. coli, and the inhibition of cell growth was monitored. Western blot assays with anti-DR0417 and anti-DR0662 antibodies confirmed that the proteins were expressed successfully in E. coli (Supplementary Figure S2). E. coli with overexpression of DR0417 exhibited substantial cell growth defect compared with the control cells containing only empty vectors (Figure 1A). In contrast, cells with co-overexpression of DR0416 and DR0417 or those with over-expression of DR0416 alone showed similar growth profiles to that of the control, indicating that the DR0417-induced growth defect could be neutralized by DR0416 (Figure 1A). However, over-expression of DR0662 and/or DR0061 in E. coli did not result in obvious changes of cell growth (Figure 1B). These results indicated that the dr0416-dr0417 locus may function as MazEF-type TA system and we therefore refer to them as MazEF-dr for MazEF in D. radiodurans.

To evaluate the biochemical properties of MazEF-dr, the in vitro activity of MazF-dr was assessed. Total cellular RNAs were extracted from D. radiodurans cells. Following incubation with purified MazF-dr, a gradual accumulation of cleavage products appeared, and they increased with increasing protein concentration (Figure 2A). Especially 16S rRNA and 23S rRNA of D. radiodurans were cleaved completely into smaller fragments by MazF-dr at high concentration (100 nM, lane 8 in Figure 2A), while few cleavage products of 5S rRNA and tRNAs were detected. When an exogenous RNA (MS2 RNA, RefSeq accession no. NC_001417) was used as a substrate of MazF-dr, a specific cleavage profile was also observed (Figure 2B). Thus, MazF-dr seems to be an endoribonuclease that cleaves RNAs. However, pre-incubation of MazF-dr with MazE-dr completely inhibited the RNase activity of MazF-dr (lanes 9–11 in Figure 2A), indicating that the activity of MazF-dr could be neutralized by MazE-dr. These results suggest that MazF-dr acts as the toxin and MazE-dr is the antitoxin that might abrogate the toxic effect of MazF-dr.

A primer extension assay was used to analyze the RNA cleavage site of MazF-dr. 16S rRNA was used as the substrate to determine the cleavage site of MazF-dr. As shown in Figures 3A–D, two cleavage patterns were identified, both of which consisted of an ACA consensus sequence (Figure 3E). In the first pattern (Figures 3A,D), cleavage occurred at the 5′ end of an adjacent A in the tetrad sequence (^↓A^↓ACA). For the other cleavage pattern (Figures 3B,C), cleavage occurred at the 5′ end of the A (^↓ACA). Notably, the cleavage activity of MazF-dr at the A^↓ACA site was higher than that at the ^↓AACA site, as determined by the product intensity, suggesting that the ^↓ACA is the major cleavage site.

To further validate the consensus cleavage sequence and the effects of the 5′ terminal nucleotide adjacent to the ACA sequence of substrates on the activity of MazF-dr, we synthesized a series of 5′ fluorescein amidite (FAM)-labeled RNA derivatives (Supplementary Table S2) and measured the cleavage activity of MazF-dr on these RNAs. As shown in Figure 4A, MazF no longer cleaved the RNAs that lacked the ACA consensus sequence. These results indicated that MazF-dr specifically cleaved RNAs containing a consensus 5′ACA3′ sequence in a ribosome-independent fashion. In addition, substitutions of the nucleotide before the ACA sequence with A, G, U, or C did not prevent cleavage function of MazF-dr, but it affected the cleavage activity of MazF-dr (Figure 4B). MazF-dr had the highest cleavage activity on the U-substituted sequence, showing a decreasing cleavage activity in the order of U>A>G>C, suggesting that the surrounding nucleotide composition of the substrates has the impact on the activity of the toxin. Unlike D. radiodurans MazF, EcMazF did not have a cleavage preference for the nucleotide before the ACA consensus sequence (Figure 4B).

We evaluated the roles of the mazEF-dr module in D. radiodurans in response to extreme stresses. A double gene knockout mutant of mazEF-dr, designated ΔMazEF-dr, was constructed (Supplementary Figure S3). The growth curve of the ΔMazEF-dr did not differ significantly from that of the wild-type strain under stress-free condition (data not shown), indicating that MazEF-dr is not essential for normal bacterial growth. However, the viability of ΔMazEF-dr differed greatly from that of the wild type under different doses of DNA damage stresses, including MMC, H₂O₂, and γ-radiation (Figures 5A–C). Under low doses of DNA damaging agents (<5 μg/ml MMC, <40 mM H₂O₂, <6 kGy γ-ray), the dose-response curve of the wild type and ΔMazEF-dr showed a similar “shoulder,” representing little decrease in the viability; however, a “steep slope” representing a substantial decrease of the viable fraction of the wild-type strain was observed at higher doses of the DNA damage agents (10–15 μg/mL MMC, 40–80 mM H₂O₂, 6 kGy γ-ray), compared with ΔMazEF-dr cells (Figures 5A–C). Mutants of a DNA repair related gene (pprI) and an antioxidant related gene (crtB), which was involved in the carotenoid biosynthesis, were used for the comparison purpose (Figure 5A). The pprI mutant was more sensitive to MMC stress than those of the wild type and ΔMazEF-dr, while the crtB mutant was as resistant as the wild type.

The cell viability data are consistent with the cell phenotypes obtained by a confocal microscopy assessment using the Live/Dead Kit (Figure 6A). The living cells were stained by SYTO 9 (green) and the dead cells were stained by propidium iodide (PI) (red). Treatment with a sub-lethal dose of MMC (15 μg/ml) led to the death of a substantial number of the wild-type cells, while most of the mutant cells remained alive after the treatment. A transmission electron microscopy (TEM) analysis demonstrated that morphological changes occurred in 24% (calculated from three independent views) of the wild-type cells, compared with the mutant cells under sub-lethal dose of the MMC stress (Figure 6B), including the loss of membrane integrity, membrane blebbing and cytoplasm condensation, which are the characteristics of PCD (Dewachter et al., 2015). The mutation of mazEF-dr relieved these cell death phenotypes (Figures 6A,B), suggesting that the MazEF-dr mediated cell death of a fraction of the population in response to sub-lethal dose of DNA damage stress.

The bacterial strains and plasmids used in this study are listed in Supplementary Table S1. The E. coli strains DH5α and BL21 (λDE3) pLysS were used for cloning and protein expression studies, respectively. E. coli BW25113Δ6 (kindly provided by Prof. Nancy A. Woychik) was used for cell growth inhibition experiments. E. coli strains were cultured in Luria-Bertani (LB) broth or LB agar at 37°C unless otherwise noted. The media were supplemented with chloramphenicol (Cm; 15 μg/ml) or kanamycin (Kan; 40 μg/ml) if necessary. D. radiodurans and its derivatives were cultured in TGY (0.5% Bacto tryptone, 0.1% glucose and 0.3 Bacto yeast extract) broth or TGY agar at 30°C. Streptomycin (Str, 10 μg/ml) was added to the cultures of D. radiodurans mutant strains.

Mutant was constructed by double crossover recombination of a streptomycin resistance cassette into the genome as described previously (Tian et al., 2007). Briefly, upstream and downstream fragments of the target gene flanked by BamHI and HindIII, respectively, were PCR-amplified with the corresponding primers listed in Supplementary Table S2. After digestion, these fragments were ligated to a streptomycin resistance cassette and transformed into the competent cells using the CaCl₂ method. Mutants were screened on TGY plates containing streptomycin, and then all the recombinants were confirmed by PCR analysis and sequencing.

The toxin and antitoxin homology genes were cloned into pBAD33 and pET28a, the resulting plasmids were transformed or co-transformed into E. coli BW25113Δ6 cells, and their expression was induced by arabinose and isopropyl-β-D-thiogalactoside (IPTG), respectively. Cultures were grown to an optical density at 600 nm (OD₆₀₀₎ of 0.6, then 0.2% arabinose and/or 0.3 mM IPTG were added to induce the expression of the toxins and/or antitoxins. E. coli cells transformed with pBAD33-dr0417 and pET28a-dr0416, or pBAD33-dr0662 and pET28a-dr0661 were cultured overnight in LB broth. Then, the cultures were diluted into fresh LB broth and grown to OD₆₀₀ of 0.55. Subsequently, each culture was divided into four equal parts. One of the cultures served as control, to which was added equal ddH₂O, while 0.2% arabinose, 0.3 mM IPTG or both 0.2% arabinose and 0.3 mM IPTG were added to the other three cultures to induce the expression of toxin and/or antitoxin. These cultures were incubated at 37°C and the OD₆₀₀ was measured hourly to evaluate the cell growth.

The viability of D. radiodurans exposed to DNA damage stresses was determined as described previously (Tian et al., 2007). Cells grown to an OD₆₀₀ of 0.6 were harvested and suspended in sterile phosphate-buffered saline (PBS). For the MMC (Sigma Co., United States) treatment, the cells were treated with 5–20 μg/ml MMC, and samples were taken after 40 min and plated on TGY agar plates after being diluted appropriately. For the H₂O₂ treatment, the cells were treated with different concentrations of H₂O₂ for 30 min. After incubation, the reaction was stopped by catalase (20 U) for 15 min. Then, the cells were diluted and plated on TGY agar plates. For the γ-radiation treatment, cell suspensions were irradiated at room temperature for 2 h with ⁶⁰Co γ-ray (point source, Zhejiang Academy of Agricultural Sciences, Zhejiang, China) at several different doses (from 3 to 12 kGy), which were adjusted by changing the distance of samples from the γ-ray source (Wang et al., 2008). After radiation, the cells were diluted appropriately with PBS and plated on TGY agar plates. All plates were incubated at 30°C for 3 days to count the viable colonies. All the tests were repeated three times independently. The viable fractions were expressed as the logarithm of the ratio of the number of colonies from the treated samples to those from the untreated control.

DR0416 and DR0417 were expressed from a pET28a+ vector containing a N-terminal 6×His tag in BL21 (λDE3) pLysS cells. When cells grown in LB media reached an OD₆₀₀ of 0.6–0.8, protein expression was induced with 0.3 mM IPTG at 30°C for 5 h. Then, cells were pelleted by centrifugation and re-suspended in lysis buffer A [20 mM pH 8.0 Tris-HCl, 500 mM NaCl, 5% (v/v) glycerol] containing PMSF (phenylmethanesulfonyl fluoride, Roche Biochemicals, Switzerland) and lysed by sonication. Cell lysates were centrifuged at 15,000 × g for 30 min to remove debris, and the supernatant was loaded onto a Ni-NTA column (GE Healthcare, United States), the target proteins were eluted with elution buffer B (20 mM Tris–HCl pH 8.0, 500 mM NaCl, and 500 mM imidazole). The collected protein fractions were dialyzed against buffer C [20 mM Tris-HCl pH 8.0, 100 mM NaCl, 5% (v/v) glycerol] and then loaded onto HiTrap Q ion exchange column (GE Healthcare) and eluted by gradient elution. Finally, proteins were further purified on a Superdex 75 column (GE Healthcare) with buffer C. Fractions containing the target proteins were pooled, concentrated, flash-frozen in liquid nitrogen, and stored in -80°C.

Total RNA of D. radiodurans was isolated using the TRIzol method according to the manufacturer’s protocol (Ambion, United States). A culture of D. radiodurans was grown in TGY medium until the OD₆₀₀ reached ∼0.6. Then, the cells were pelleted and resuspended in 1 ml of TRIzol solution. After vortexing, the samples were incubated for 15 min at room temperature.

Purified DR0417 was incubated with D. radiodurans total RNA or MS2 phage RNA (RefSeq accession no. NC_001417) at 37°C for 15 min. The reaction mixture consisted of DR0417, RNA, 0.1 mM ethylenediaminetetraacetic acid (EDTA) and 0.5 μl of RNase inhibitor (Promega, United States) in 20 mM Tris-HCl (pH 8.0), with or without DR0416. After denaturation in urea, the mixtures were loaded onto a 4.5% polyacrylamide gel containing 8 M urea, electrophoresed, and then stained with ethidium bromide (EB).

Deinococcus radiodurans 16S rRNA was transcribed in vitro using the T7 promoter method (Hook-Barnard et al., 2007). Briefly, 16S rDNA fragments used for the transcription reaction were amplified with the primers 16S-TR and 16S-TS that contain the T7 promoter sequence (Supplementary Table S2). After purified, in vitro transcription reaction was carried out at 37°C for 1.5 h with 50 U T7 RNA polymerase (New England Biolabs, United Kingdom). Then DNase I (New England Biolabs) was used to digest the 16S rDNA substrate. The obtained 16S rRNA products were extracted using phenol-chloroform and precipitated with ethanol.

For a primer extension analysis of the cleavage sites of MazF-dr, the 16S RNA was digested with MazF-dr at 37°C for 10 min. The reaction mixture contained 1 μg of 16S RNA substrate, 0.05 μg of MazF-dr and 0.5 μl of RNase inhibitor in 20 mM Tris-HCl buffer (pH 8.0). Primer extension was carried out in 10 μl of the reaction mixture containing 5′-FAM labeled primers (16S-F1, 16S-F2, and 16S-F3 as shown in Supplementary Table S2) at 42°C for 1 h, and then it was stopped by adding 2 μl of quench loading buffer (95% formamide, 20 mM EDTA, 0.05% bromophenol blue, and 0.05% xylene cyanol EF). The samples were incubated at 70°C for 5 min prior to electrophoresis on a 6% sequencing gel. The four sequence markers were produced using PCR with 16S DNA as template and the same primers that were used in the primer extension assay. The PCR mixture contained 1 μg of 16S DNA substrate, 0.05 mM ddNTPs, 0.5 mM dNTP, therminator DNA polymerase (New England Biolabs) and 1 mM FAM labeled primer. A sequence logo plot of multiple sequence alignment was generated by the Weblogo online program to analyze the conserved RNA cleavage sites of MazF-dr² (Doerks et al., 2002).

Oligoribonucleotides labeled with 5′-FAM were commercially synthesized by Takara Bio (Japan). The endoribonuclease activity of MazF was assayed in a 10 μl reaction mixture containing MazF, FAM-labeled RNA substrates, 0.2 μl of RNase inhibitor (Promega, United States), 20 mM Tris-HCl (pH 8.0), and 0.1 mM EDTA. The reaction mixture was incubated at 37°C for 15 min, and stopped by adding 2 μl of RNA loading buffer (0.25% bromophenol blue, 0.25% xylene cyanol EF, and 15% Ficoll). Then, the reaction mixture was loaded onto a 20% urea PAGE and scanned by the Typhon system (GE Healthcare, United States).

The cell death was assessed by using confocal microscopy. The cells were stained using the Live/Dead Kit containing PI and SYTO 9 (Invitrogen). Cells grown to an OD₆₀₀ of 0.8 were washed twice with 0.85% NaCl. Cells treated with a sub-lethal dose of MMC (15 μg/ml, 40 min), as well as untreated cells, were collected by centrifugation. Then the cells were stained using 1 μl of a 1:1 mixture of PI and Syto 9 from the kit and were incubated for 15 min at room temperature keeping in the dark. Following the wash by 0.85% NaCl twice, 100 μl of 4% formalin was applied for the fixation. The cells were washed twice with PBS buffer and re-suspended in 20 μl of 50% glycerol. A confocal microscopy assay was performed using a Leica DM4000 fluorescence microscope. Living cells were stained by Syto 9 (green) and death cells were stained by PI (red).

For the TEM analysis, D. radiodurans cells treated with MMC (15 μg/ml, 40 min), as well as untreated cells, were collected by centrifugation. Samples were processed as described previously (Rothfuss et al., 2006). The cells were fixed with 2.5% glutaraldehyde in phosphate buffer (pH 7.0) overnight and then embedded in 5% agar. Thin sections of the samples were stained with uranyl acetate for 15 min and observed by TEM (JEM-1230, JEOL, Japan).