A Genome-Wide Search for Ionizing-Radiation-Responsive Elements in Deinococcus radiodurans Reveals a Regulatory Role for the DNA Gyrase Subunit A Gene's 5′ Untranslated Region in the Radiation and Desiccation Response

ABSTRACT

Tight regulation of gene expression is important for the survival of Deinococcus radiodurans, a model bacterium of extreme stress resistance. Few studies have examined the use of regulatory RNAs as a possible contributing mechanism to ionizing radiation (IR) resistance, despite their proffered efficient and dynamic gene expression regulation under IR stress. This work presents a transcriptome-based approach for the identification of stress-responsive regulatory 5′ untranslated region (5′-UTR) elements in D. radiodurans R1 that can be broadly applied to other bacteria. Using this platform and an in vivo fluorescence screen, we uncovered the presence of a radiation-responsive regulatory motif in the 5′ UTR of the DNA gyrase subunit A gene. Additional screens under H₂O₂-induced oxidative stress revealed the specificity of the response of this element to IR stress. Further examination of the sequence revealed a regulatory motif of the radiation and desiccation response (RDR) in the 5′ UTR that is necessary for the recovery of D. radiodurans from high doses of IR. Furthermore, we suggest that it is the preservation of predicted RNA structure, in addition to DNA sequence consensus of the motif, that permits this important regulatory ability.

IMPORTANCE Deinococcus radiodurans is an extremely stress-resistant bacterium capable of tolerating up to 3,000 times more ionizing radiation than human cells. As an integral part of the stress response mechanism of this organism, we suspect that it maintains stringent control of gene expression. However, understanding of its regulatory pathways remains incomplete to date. Untranslated RNA elements have been demonstrated to play crucial roles in gene regulation throughout bacteria. In this work, we focus on searching for and characterizing responsive RNA elements under radiation stress and propose that multiple levels of gene regulation work simultaneously to enable this organism to efficiently recover from exposure to ionizing radiation. The model we propose serves as a generic template to investigate similar mechanisms of gene regulation under stress that have likely evolved in other bacterial species.

INTRODUCTION

Deinococcus radiodurans is a model organism for extreme stress resistance that exhibits unique tolerance to ionizing radiation (IR) (15 to 20 kGy), desiccation, and oxidative stresses (1–3). To comprehensively study global mechanisms of gene expression dynamics that could aid survival under environmental stress, a number of proteomics and transcriptomics studies have recently been conducted throughout bacteria (4–9). In Deinococcus species, these studies have highlighted that the genome of D. radiodurans largely contains the same DNA repair enzymes as other radiosensitive organisms, alluding to the complexity of the radiation resistance mechanism (10–18). Indeed, some of the uncovered stress responses in this organism (1–3, 19–22) are similar to many well-characterized bacterial stress responses that involve gene expression regulation based on repressors/activators directly regulating at the promoter level; an example of this in Escherichia coli is the SOS response (23, 24). Additionally, based on the higher concentration of manganese in D. radiodurans, it has been proposed that Mn²⁺ forms complexes with amino acids and peptides to act as an antioxidant to protect proteins from reactive oxygen species (ROS) (25–28). While there is compelling evidence that the protection of the proteome via Mn²⁺-peptide complexes is a critical component of D. radiodurans survival, there is significant interest in uncovering other contributing stress response mechanisms and global regulatory networks that might be unique to this organism.

While a unique set of IR-protective genes has not yet been determined in D. radiodurans, a comparative bioinformatics analysis between D. radiodurans and Deinococcus geothermalis, another radiation-resistant bacterium of the same family, revealed a unique gene regulatory network that aids in gene expression responses during recovery from IR and desiccation stress (29). The radiation and desiccation response (RDR) regulon is defined by a 17-bp palindromic motif (called the RDRM) predicted to be present in the promoter of ∼25 genes in the regulon; many of the predicted RDR-regulated genes are DNA repair enzymes that are crucial to IR resistance (29). The regulon is controlled by a repressor protein, DdrO (encoded by DR2574), that binds as a dimer to the RDRM and is released when site specifically cleaved by the post-IR activated protease, IrrE (also called PprI [DR0167]) (14, 22, 30–32). While further exploration of this network is needed, the RDR regulon suggests that unique regulatory networks (rather than unique IR protection genes) are important contributors to D. radiodurans IR survival.

While regulation at the gene level has been explored (1, 19, 33–37), RNA regulatory elements have not been highly explored in D. radiodurans, despite the well-documented ability of RNA to provide a dynamic response to stress. To date, noncoding RNAs (ncRNAs) have been implicated in the stress tolerance and regulation mechanisms of many bacteria (38–45). In D. radiodurans, transcriptomics studies have largely enhanced the discovery of ncRNAs that have been demonstrated to exhibit differential expression during recovery of ionizing radiation under various growth conditions (13, 17, 38). This use of regulatory RNA in stress recovery is not surprising given that RNA-based gene regulation permits fine-tuning of gene expression that cannot be accomplished with DNA regulation alone. Furthermore, regulatory RNAs pose advantageous features for efficient gene expression control given that they pose a lower metabolic load since they are not translated and can provide acute response times due to the short half-life of RNAs (43, 44).

RNA-level regulation proffers a possible mechanism that D. radiodurans might have exploited during its evolution to adapt to high levels of stress. Based on the small size of ncRNAs, there is also a decreased likelihood of radiation damage, which would permit a more rapid and efficient response to radiation (1). Recent work has uncovered several small RNAs that are significantly upregulated under IR (38). Additionally, two regulatory 5′ untranslated regions (5′ UTRs) have been experimentally validated in D. radiodurans, confirming the presence of regulatory 5′ UTRs in this organism (46, 47). 5′ UTRs of mRNAs represent a class of cis-regulatory RNAs that can directly regulate their adjacent mRNA transcript through structural changes in the RNA that can reveal/hide the ribosome binding site (RBS) or through promotion/repression of mRNA degradation (43, 48). Additionally, 5′ UTRs can contain sequence and/or structural motifs that act as binding sites for other regulatory elements, such as proteins, RNA, and small molecules (45, 49). As such, these regulatory elements can provide a direct sensing mechanism of stress through structural changes elicited by directly sensing environmental cues such as temperature (RNA thermosensors), pH, and specific metabolites (riboswitches) (44, 50, 51).

In this work, we explore the possibility that the RNA-based response at the 5′ UTR could be important for recovery mechanisms post-IR. We couple bioinformatics analysis of previously published D. radiodurans transcriptomics data (38), structural predictions, and structural conservation analysis to identify 31 5′-UTR candidates that could exhibit gene regulation activity after exposure to 1 to 10 kGy of IR. Initially, we screened the regulatory ability of these 5′ UTRs after IR-induced stress in the context of GFP expression by establishing an in vivo fluorescence reporter system containing a translational fusion of each 5′-UTR candidate to a D. radiodurans codon-optimized green fluorescent protein (GFP). Ultimately, our methods identified the 5′ UTR of the DNA gyrase subunit A gene (DR1913; gyrA) as a regulatory 5′ UTR capable of activating gene expression during recovery from 10 kGy of IR, based on the presence of a conserved RDRM (22, 29, 30, 32). Additionally, we predict that the conserved RNA structure (rather than just DNA sequence consensus) of this regulatory motif is also important to D. radiodurans recovery from IR. Our characterization of the 5′ UTR of gyrA suggests an additional RNA level of regulation in the 5′ UTR of gyrA to regulate gene expression during recovery from IR.

RESULTS

Bioinformatics analysis of RNA-sequencing data reveals novel 5′-UTR candidates that could regulate gene expression post-IR exposure in D. radiodurans.

A bioinformatics approach developed for this work (Fig. 1A to C) identified a set of 31 5′ UTRs that could contribute to tuning gene regulation in D. radiodurans R1 during recovery from exposure to ionizing radiation (1 to 15 kGy). We identified an initial set of 380 potential 5′-UTR candidates by screening a previously published RNA sequencing data set of D. radiodurans R1 for transcripts mapped to 5′ UTRs adjacent to annotated coding regions (38) (see Table S1 in the supplemental material). Putative 5′ UTRs were described as being less than 300 bp away from the start codon, without overlapping the neighboring annotated open reading frame. Additionally, we selected for 5′-UTR candidates greater than 35 bp, as this represents the length of the shortest known 5′-UTR regulatory element (preQ1 class I riboswitch) (52). Any 5′ UTRs corresponding to tRNA synthetases or ribosomal proteins were also discarded from possible candidates. This analysis also ruled out potential 3′ UTRs (of an adjacent upstream coding region), noncoding intergenic RNAs, or coding spacers on polycistronic mRNAs.

FIG 1.

Bioinformatics analysis to identify potential regulatory 5′ UTR candidates. (A) Identify potential 5′ UTRs based on read counts and proximity to annotated genes from visual inspection of RNA sequencing data. (B) Filter 5′-UTR candidate selection by removal of candidates with low transcript read counts and selection of those with differential expression of adjacent genes. (C) Verify the presence and sequence of 5′-UTR candidates determined by RT-PCR and 5′ RACE.

To narrow down the number of putative 5′-UTR candidates for further analysis, we filtered the 380 potential candidate regions on the basis of the detected transcript counts of the 5′ UTR and ranked more favorably candidates that had transcript counts comparable to their adjacent coding sequence. This initial ranking served as a preliminary analysis to remove possible overlapping candidates; 5′ rapid amplification of cDNA ends (RACE) analysis followed (see below) for all the final candidates to characterize the transcript start site. Sequences that had average transcript counts less than twice the average read per genome nucleotide (3.25) were discarded from the candidate list to remove low-expression candidates, producing 282 5′-UTR candidates.

To reduce the number of 5′-UTR candidates to experimentally determine transcription start sites by reverse transcription PCR (RT-PCR) and 5′ RACE, we cross-referenced the candidate list to all previously published differentially expressed proteins during recovery from ionizing radiation, which ranged from 1 to 15 kGy (10–14). We prioritized those differentially expressed candidates demonstrating greater than a 2-fold change after IR exposure based on the rationale that these types of differentially expressed proteins could be regulated by their associated 5′ UTRs. We also selected putative 5′ UTRs associated with well-characterized metabolic enzymes (such as thymidylate synthase [DR2630] and CTP synthase [DR1573]), as similar enzymes have been shown in the literature to be regulated by 5′ UTRs (50, 51).

This filtering produced 89 candidates (Table S2) that were experimentally tested for contiguous expression and transcription start site (TSS) location by either RT-PCR (Fig. S4) or 5′ RACE (depending on the efficiency of each method for each candidate). Using the RT-PCR, 5′ RACE, and transcriptome sequencing (RNA-seq) data, we were able to designate specific UTR sequences for 31 5′-UTR candidates, which were then selected for further study (Table S3).

As a prelude to experimentation, structural conservation was analyzed for the 31 5′-UTR candidates based on the rationale that structural conservation among organisms closely related to D. radiodurans R1 of the 31 5′-UTR candidates could indicate true (functionally conserved) regulatory ability. We conducted an RNA structural conservation analysis using LocARNA, a published algorithm that uses an alignment tool to determine sequence conservation and a covariance model to determine structure conservation (53–55). For this analysis, we utilized all sequenced members of the Deinococcus-Thermus phylum (Fig. S1). Each 5′-UTR candidate region was searched using BLAST (with the NCBI blastn tool) against this specific list of organisms to collect a set of similar sequences (E value > 10⁻⁸) that were then analyzed for structural conservation by LocARNA. While some candidates did not have any sequence similarity with the selected list of organisms, most candidates contained structurally conserved motifs (Fig. S1). Importantly, many of these structured motifs appeared quite complex and thermodynamically favorable, displaying distinct base pairing and hairpin structures. Results from this analysis indicated the possibility that some of the 5′-UTR candidates could be regulatory in gene expression based on the possibility of conserved structure-based regulation.

Our bioinformatics approach resulted in the identification of two widely conserved bacterial riboswitches that have been deposited in the RNA families (RFAM) database (56): the flavin mononucleotide (FMN) switch (DR0153) and the glmS ribozyme (DR0302). These riboswitches have been demonstrated to control regulation of their adjacent genes in response to flavin mononucleotide (FMN switch) (46) and glucosamine-6-phosphate (glmS ribozyme) (47). Importantly, both the FMN riboswitch (46) and glmS ribozyme (47) have been experimentally confirmed in D. radiodurans, validating that our method was capturing true regulatory 5′ UTRs. Although possible connections to a response to radiation have not yet been elucidated for these two riboswitches, many of the remaining 29 5′-UTR candidates selected for further study are associated with differentially expressed proteins during ionizing irradiation and oxidative stress (Table S3).

In vivo fluorescence-based screen uncovers the gyrA (DR1913) 5′ UTR as regulatory for gene expression during recovery from 10 kGy of IR.

To test the independent ability of the uncovered 5′ UTRs to regulate gene expression in D. radiodurans R1 during recovery from ionizing radiation (10 kGy), we designed an in vivo fluorescence screen in the context of the pRADgro shuttle plasmid (Fig. S2A and S3A). This vector has been shown to be successful in the overexpression of exogenous proteins and is able to replicate in both E. coli (where it confers ampicillin resistance) and D. radiodurans (where it confers chloramphenicol resistance) (57). This pRADgro reporter plasmid (Fig. S2B and S3C) contained an insertion site for a 5′-UTR element, a constitutive D. radiodurans R1 groESL promoter, and a codon-optimized enhanced GFP reporter. Initially, we established this screen in the context of the well-characterized synthetic theophylline riboswitch (ThRS) (58) (Fig. 2A; Fig. S3B). This specific regulatory region operates at the translational level by sequestering the RBS, which is released by the binding of the theophylline ligand, causing a conformational shift that exposes it for translation. A consistent significant fluorescence increase (∼4.7-fold) was observed following induction of the pThRS-GFP construct with 2 mM theophylline (relative to a dimethyl sulfoxide [DMSO] control), as indicated by the median fluorescence determined using flow cytometry and whole-cell fluorescence microscopy (Fig. 2C). We therefore reasoned that this assay could readily detect shifts in translational activity mediated by a 5′ UTR poststress during recovery from ionizing radiation. Recently, another laboratory reported the development of a similar in situ fluorescence assay to evaluate promoter activity in D. radiodurans (59). However, it is an important distinction that this already published method aims to determine promoter activity, while the one presented in this work interrogates 5′-UTR regulation.

FIG 2.

Development of an in vivo fluorescence screen in D. radiodurans. (A) Plasmid map of the pThRS-GFP construct. (B) Flow cytometry data acquisition was gated for the population determined to be D. radiodurans cells based on forward scattering (FSC) versus side scattering (SSC) as a first gate and forward scattering versus fluorescence (FL1) as a second gate. (C) Median fluorescence values from the pThRS-GFP construct expressed in D. radiodurans. Error bars represent standard deviations of results from biological triplicates. Confocal fluorescence microscopy images of D. radiodurans expressing pThRS-GFP appear above the bar graph (scale bar, 10 μm).

The 31 5′-UTR candidate sequences were used to construct a library of potential regulatory RNA regions in the context of the GFP reporter described above (Table S3; Fig. S2B and S3C). Each construct contained the 5′ UTR and 60 additional nucleotides corresponding to the nearby coding region. We included the sequence for 20 additional amino acids of the native gene, as we reasoned that nucleotides in the coding region may affect the structure of the 5′ UTR, as previously reported in other bacteria (60). This assay was used to screen all 5′-UTR candidates for their ability to regulate GFP expression in D. radiodurans R1 cells during recovery from ionizing radiation. It is important to note that our screen specifically focused on RNA regulatory responses that increase GFP expression, as we theorized it would be difficult to decouple decreased expression from possible GFP degradation postirradiation. Although a rarer regulatory response for 5′ UTRs, RNA regulatory elements have previously been demonstrated to result in an increase of gene expression after activation (61–63).

Based on the construction of the reporter, any activation of the 5′ UTR would result in a downstream increase in GFP expression, measured by an increase in fluorescence. Given that the groESL promoter (within the pRADgro plasmid) has not been previously reported to be induced by IR, we expected any changes in GFP expression to be true reflections of 5′-UTR regulation and not of any plasmid-based artifact (11, 57). However, to verify a lack of post-IR response from the groESL promoter itself (independently from the 5′ UTR under interrogation), we constructed and tested fluorescence shifts in cells harboring the pRADgro-GFP construct postirradiation. As shown in Fig. 3A, we observed a slight increase in median fluorescence fold change at 10 kGy; however, this was not significant by Student's t test in comparison to the innate basal level of autofluorescence of D. radiodurans R1 cells determined by the empty-vector (pRADgro) control. We expected that the minor contribution of the groESL promoter to fluorescence shifts during recovery from IR would be consistent across all tested candidates and considered this part of the background fluorescence of this plasmid system. Given this observation, we focused only on 5′-UTR candidates that had a significant (by Student's t test, P value of <0.05) average fold change (>2-fold) in median fluorescence during recovery from IR relative to that of the empty-vector control.

FIG 3.

In vivo fluorescence screen of 5′-UTR reporter constructs (pDR-GFP) reveals the 5′ UTR of DR1913 (UTR_gyrA) as a 10-kGy-IR-responsive RNA element. (A) Normalized median fluorescence (median fluorescence of sample treated with 10 kGy of IR/median fluorescence of 0-kGy sample) of all 31 pDR-GFP candidates referred to by gene number (see Table S3 in the supplemental material), in addition to empty plasmid (pRADgro) and constitutive reporter (pRADgro-GFP). Error bars represent standard deviations of results of at least technical duplicates from two independent experiments (n = 5). Significant shifts in median fluorescence include that of DR1913 (DNA gyrase subunit A, UTR_gyrA), where significance by Student's t test is denoted by asterisks (**, P value < 0.005). (B) Fluorescence histogram of a single replicate of pDR1913-GFP (UTR_gyrA) after recovery from 0 and 10 kGy of IR.

We conducted our in vivo fluorescence screen of the 31 5′-UTR candidates under acute IR (10 kGy). In selecting this dose, we reasoned that an acute dose higher than the survival threshold of D. radiodurans (2) would elicit stronger potential 5′-UTR activity. For these assays, D. radiodurans R1 samples expressing the pRADgro–5′-UTR–GFP (pDR-GFP) reporter were irradiated with a linear accelerator (LINAC) β-ray source. Of the 31 candidates examined by flow cytometry during IR recovery, one 5′-UTR candidate demonstrated a significant post-IR fluorescence response: the 5′ UTR of DR1913 (the DNA gyrase subunit A gene, gyrA), here referred to as UTR_gyrA. D. radiodurans R1 cells expressing pDR1913-GFP had a significant fold change in median fluorescence during recovery from 10 kGy of IR (2.7-fold, P value = 3.1 × 10⁻⁵, compared to pRADgro control), as indicated by flow cytometry and whole-cell fluorescence microscopy (Fig. 3A and B and 4A and B). No other candidates demonstrated a fold change in fluorescence greater than 2-fold or significant changes in fluorescence compared to that of the empty-vector control (pRADgro) by Student's t test; therefore, they were not examined further (Fig. 3A).

FIG 4.

UTR_gyrA responds in a specific IR dose-dependent manner. (A) D. radiodurans cells expressing pDR1913-GFP (UTR_gyrA) exhibited a dose-dependent fluorescence response with increasing IR dose (R² = 0.89), as measured by flow cytometry. Error bars represent standard deviations of results of triplicate samples. (B) D. radiodurans expressing the pDR1913-GFP reporter (UTR_gyrA) imaged using a confocal fluorescence microscope (scale bar, 5 μm). (C) Western blot analysis for GFP normalized to total protein loaded and normalized again to the 0-kGy value to present fold changes in GFP. (D) RT-qPCR data analyzed using the 2^−ΔΔCT method probing for GFP and rRNA as a reference. Sample values were normalized to those of the reference gene and to the sample values at 0 kGy to present fold changes in GFP expression post-IR. Error was propagated from differences in triplicate samples. (E) Fold change in median fluorescence for D. radiodurans expressing pDR1913-GFP (UTR_gyrA) posttreatment with 20 mM H₂O₂, as measured by flow cytometry. pDR1857-GFP (UTR_ohrP) represents a positive control for H₂O₂ stress with a significant fold change compared to the pRADGro control (**, P value < 0.005, Student's t test). Error bars represent standard deviations of results for biological triplicates.

In vivo fluorescence-based screening determines that UTR_gyrA regulates gene expression in an IR dose-dependent manner.

To assess the regulatory ability of UTR_gyrA during recovery from lower doses of ionizing radiation, we tested the 5′-UTR response during recovery from a range of IR doses (0.25 to 15 kGy). Remarkably, fluorescence measurements by flow cytometry and whole-cell fluorescence microscopy of D. radiodurans expressing the pDR1913-GFP reporter showed a unique sensitivity within this range of doses that is well within D. radiodurans R1 IR survival (Fig. 4A and B). While dose-dependent activation has not been extensively studied at the proteomics and transcriptomics level in D. radiodurans R1, the dose-dependent regulation captured by pDR1913-GFP hints at an additional mechanism to control this differential expression (13).

To verify accurate capturing of GFP levels during recovery from IR by flow cytometry, we performed Western blotting analyses. Fold changes in GFP levels seen by Western blotting correlated with the fold changes seen by flow cytometry during recovery from 10 kGy of IR (2.67-fold change relative to the 0-kGy control), confirming the increase in GFP during recovery from IR (Fig. 4C). These changes in GFP protein expression from the pDR1913-GFP construct also correlated with the native levels of GyrA upregulation by mass spectrometry (2.72-fold change relative to the unirradiated control) (Fig. 5D; Fig. S6A to C). This upregulation of GFP during IR recovery correlated with the transcript levels, as demonstrated by reverse transcription quantitative PCR (RT-qPCR) (3.08-fold increase after 10 kGy of IR compared to the 0-kGy control) (Fig. 4D). Previous studies of 5′-UTR RNA regulatory elements have confirmed regulatory effects at both the transcriptional and translational levels, suggesting the possibility that the 5′ UTR of gyrA might modulate both levels of gene expression (61, 64). Interestingly, previous proteomics and transcriptomics studies have reported a high upregulation of gyrA following irradiation that could function in the repair of the genome (previously reported 5.61-fold maximum protein upregulation at 1 h after 6 kGy of IR [11] and 3.29-fold upregulation for transcript at 0.5 h after 15 kGy of IR [17]), which suggests the importance of tight control of this gene (Table S3).

FIG 5.

UTR_gyrA mutants demonstrate changes in IR survival compared to strain R1. (A) Genomic 5′-UTR gyrA mutant construction strategy. ΔUTR_gyrA was constructed by completely removing the highlighted region (red lines with chromosome 1 coordinates indicated) of the 5′ UTR, leaving the RBS intact. The RDRM (green) was scrambled as shown by swapping the regions in green to maintain the predicted local structure (NuPACK predictions). The bold nucleotides indicate the RDRM highly conserved sequence. (B) Survival curve of R1 (WT), the ΔUTR_gyrA mutant, gyrA complementation strains (ΔUTR_gyrA-pGyrUTR and ΔUTR_gyrA-pGroUTR), and the UTRScramgyrA mutant under IR stress. Each point represents the average CFU count post-IR stress, normalized to the sham (0 kGy) count. Error bars represent the standard deviations of results for biological triplicate samples. (C) gyrA gene expression fold changes (compared to 0-kGy samples) from RT-qPCR results analyzed using the 2^−ΔΔCT method (relative to the expression changes for the 0-kGy-treated samples) from samples treated with 1 or 10 kGy IR. The promoter-expressed GyrA (pGyrUTR/pGroUTR) served as a positive control for the experiments to ensure proper detection of gyrA. Error was propagated from differences in triplicate samples. (D) R1 and ΔUTR_gyrA strains treated with 0, 1, and 10 kGy were run on a liquid chromatography tandem mass spectrometer (LC-MS/MS) to determine the changes in GyrA protein levels upon the removal of the 5′ UTR. Triplicate samples were injected into the mass spectrometer, and the peak area was determined for four GyrA peptides to quantify the total protein in the sample. Additional proteins that had reported expression level changes in the literature were also quantitated as sample control (data not shown). GyrA LC-MS/MS peptide peak areas are normalized by the total ion current for each peptide detected (data here are from the IPDIGALR peptide).

Finally, we explored whether UTR_gyrA regulatory effects were specific to IR recovery. While IR causes damaging double-stranded DNA breaks, additional damage is due to the production of reactive oxygen species (ROS) from the irradiation-based cleavage of water that can further damage cellular components (2). To test whether the observed regulatory features of the UTR_gyrA post-IR were specific to an IR response, and not part of a generic oxidative stress response, we screened the regulatory activity of UTR_gyrA under recovery from 20 mM H₂O₂. As a positive control for these experiments, we employed a regulatory region from the DR1857 5′ UTR (which encodes the organic hydroperoxide resistance protein; pDR1857-GFP, UTR_ohrP), which we predicted to be regulatory in gene expression during H₂O₂ recovery based on high sequence similarity of this region in other bacteria (65, 66). As shown in Fig. 4E, UTR_gyrA did not induce significant changes in GFP expression during recovery from 20 mM H₂O₂ stress relative to UTR_ohrP (0.99-fold [P value of 0.46] for UTR_gyrA versus 1.28-fold [P value of 0.0017] for UTR_ohrP). These results suggest that the regulatory activity of UTR_gyrA is likely a specific part of post-IR recovery rather than general oxidative stress recovery. This hypothesis was further supported by the observation that no regulatory activation of UTR_gyrA was observed even when tested in a catalase knockout strain (ΔkatA; strain KKW7003) (67), which is more sensitive to hydrogen peroxide than the R1 strain of D. radiodurans used in this study (data not shown).

Physiological characterization of UTR_gyrA reveals the importance of the regulatory region for D. radiodurans survival post-IR stress.

To evaluate the physiological relevance of the regulatory 5′ UTR within its native genomic context, we constructed a strain in which the 5′ UTR of gyrA was disrupted (ΔUTR_gyrA). In this strain, the 5′ UTR was deleted using a homology-directed markerless gene replacement strategy that replaced the endogenous 5′ UTR with a deletion variant in the context of the wild-type (WT) D. radiodurans R1 strain. A sketch of the specific construct illustrating the specific deletions/insertions is shown in Fig. 5A, and a confirmation of these genetic disruptions was performed by PCR (data not shown). The ΔUTR_gyrA mutant displayed a drastic decrease in survivability under a range of IR doses (0, 1, 5, 10, and 15 kGy) compared to WT D. radiodurans R1, and we were unable to obtain any measurable survival of ΔUTR_gyrA at doses greater than 5 kGy (Fig. 5B). Since expression of gyrA has been determined to be critical for genome maintenance (68), we examined the effect of this deletion on the native transcription of gyrA by RT-qPCR (Fig. 5C) and native protein levels by mass spectrometry (Fig. 5D; Fig. S6A to C). Both transcript and protein levels were reduced during normal growth conditions and were not upregulated during IR recovery. These results demonstrated the disruption of critical IR-responsive regulatory portions of gyrA but confirmed that the ability to express GyrA was not eliminated.

To further isolate the role of UTR_gyrA in IR recovery, we used a genetic complementation strategy. Here we expressed GyrA protein (FLAG tagged) using the groESL promoter and either the native gyrA 5′ UTR or a nonnative 5′ UTR (groESL) (pGyrUTRgyrA and pGroUTRgyrA, respectively) in the ΔUTR_gyrA mutant (ΔUTR_gyrA-pGyrUTR and ΔUTR_gyrA-pGroUTR, respectively). While the gyrA complementation strains were unable to fully restore wild-type survival (likely due to the nonnative promoter), ΔUTR_gyrA-pGroUTR increased survival to 10 kGy, and ΔUTR_gyrA-pGyrUTR extended survival to 15 kGy (Fig. 5B). Interestingly, use of the UTR_gyrA regulatory element to regulate gyrA expression led to a significant increase in survival of the ΔUTR_gyrA deletion strain relative to the nonnative 5′ UTR (groESL) (Fig. 5B). These differences in survival were compared to protein levels by Western blotting (Fig. S6D) and to transcript levels by RT-qPCR (Fig. 5C), demonstrating that the UTR_gyrA is required for significant upregulation of gyrA during post-IR recovery. Collectively, these results suggest the physiological relevance of the UTR_gyrA for regulating gene expression in D. radiodurans during recovery from IR.

Further analysis of the UTR_gyrA sequence revealed a previously reported DNA sequence-based motif that has been implicated in the radiation and desiccation response (RDR) of D. radiodurans and D. geothermalis (29) (Fig. 6A). To elucidate if the IR-dependent regulation by UTR_gyrA was based on this mechanism, we repeated the same 5′-UTR–GFP fluorescence screen in a strain lacking IrrE (DR0167; ΔirrE-pDR1913-GFP), the protease that cleaves DdrO (DR2574) during IR recovery to derepress the RDR regulon (22). This strain demonstrated no significant upregulation of GFP expression during post-IR recovery (Fig. 6B), suggesting that UTR_gyrA regulatory activity is dependent on the RDR regulatory system.

FIG 6.

Binding of DdrO to RDRM found in the 5′ UTR of gyrA. (A) Predicted structure of UTR_gyrA based on conservation with other Deinococcus-Thermus species (from LocARNA) with the RDRM highlighted in red. (B) Normalized median fluorescence of D. radiodurans R1 and ΔirrE mutant expressing pDR1913-GFP. Error bars represent standard deviations of results for biological triplicates.

Given these results, we hypothesized that the decreased survival of the ΔUTR_gyrA strain relative to the WT strain during IR recovery was due to removal of the RDRM sequence contained in the 5′ UTR. Previous research has suggested that the RDRM functions as an enhancer element to activate upregulation of downstream genes during post-IR recovery (69). This enhancement has been described as a sequence consensus-based binding of DdrO to the region, to block enhancement under native conditions. The RDRM consensus sequence contains four highly conserved regions (Fig. 5A) that have been determined to be critical for transcriptional regulation, as disruption of this consensus sequence results in decreased enhancement of gene expression during post-IR recovery (22, 59, 69). Based on the presence of the RDRM in the 5′ UTR of gyrA, we reasoned that the UTR_gyrA RNA provides the possibility of an RNA structural element that could be an additional contributor to UTR_gyrA regulation during recovery from IR. We rationally created genomic RDRM compensatory mutations (UTRScramgyrA) that disrupted the consensus sequence while maintaining predicted local structure of the RDRM (as predicted by NuPACK) (70) (Fig. 5A).

Testing the survival of the UTRScramgyrA mutant to IR revealed a survival similar to that of the WT (Fig. 5B). Interestingly, examination of gyrA transcript levels during post-IR recovery in the UTRScramgyrA strain revealed a 4.5-fold lower change in gene expression than that of the WT during recovery from 10 kGy of IR (Fig. 5C). This decrease in upregulation of gyrA during post-IR recovery is consistent with the presumed enhancer role of the RDRM consensus sequence (69). However, our previous disruption of UTR_gyrA (ΔUTR_gyrA) resulted in a significant decrease in survival under IR (Fig. 5B). Given the viability of the UTRScramgyrA mutant under high IR doses, we hypothesized that the presence of the 5′-UTR element itself, specifically its structural consensus, still enabled post-IR recovery mechanisms independent of the previously described DdrO sequence consensus-based regulation. In this model, the RDRM sequence motif is not required; rather, preserving RNA structure is important for recovery from IR. Importantly, these results suggest that the 5′ UTR of gyrA contributes to other mechanisms besides the DNA-level RDRM-mediated response known to depend on sequence consensus.

Conservation of UTR_gyrA structure suggests possible alternative regulatory mechanisms in UTR_gyrA during recovery from IR.

One possible alternative mechanism of UTR_gyrA that we initially explored was the possible direct binding of DdrO to RNA as well as DNA. However, after several attempts, we only saw weak evidence for this mechanism, as measured by in vitro gel shift analyses (data not shown). Given the dependence on IrrE/DdrO that we noted earlier, the failure to observe strong direct binding between the UTR_gyrA and DdrO in vitro suggests an alternative mechanism of UTR_gyrA regulation, where perhaps other factors are involved in vivo (to mediate an indirect relationship between DdrO and the UTR_gyrA). Indeed, a structure conservation analysis using LocARNA demonstrates that the 5′ UTR of gyrA contains some structural conservation among several Deinococcus species (Fig. 6A). This conservation of structure suggests a shared regulatory mechanism dependent on RNA structure.

DISCUSSION

As RNA-based regulation is hypothesized to be dynamic and efficient, it represents a useful mechanism for extremophiles to employ. Recent studies have demonstrated the function of regulatory RNAs in D. radiodurans for radiation resistance (38). Although there is some evidence that Deinococcus species also employ leaderless mRNA (mRNA without a 5′ UTR) for rapid turnover of important genes (71), our analysis demonstrates that the 5′ UTR can provide an important function in gene expression regulation during recovery from IR. In this work, we uncovered a 5′ UTR (UTR_gyrA; 5′ UTR of DR1913) that can upregulate GFP expression during recovery from 10 kGy of IR (Fig. 3 and 4). DNA gyrase subunit A (GyrA) is an important protein for genomic maintenance and DNA repair by negatively supercoiling DNA. As such, the function of the gyrA protein is critical for DNA replication and for genome reconstruction after severe IR exposure (68, 72). Indeed, this 5′ UTR was demonstrated to be essential to D. radiodurans survival after exposure to high doses of IR (>5 kGy) (Fig. 5B). Importantly, these results support the proposed hypothesis of an additional RNA-level regulatory mechanism of the 5′ UTR of gyrA during recovery from IR. Additionally, we have demonstrated a generic in vivo fluorescence screen to determine the regulatory ability of 5′ UTRs (including riboswitches such as FMN and glmS ribozyme) in response to a variety of environmental stresses.

Based on our results, we add to the current regulation model by proposing the possibility of gyrA regulation at both the DNA and RNA level to permit tight control of gyrA gene expression during IR recovery (Fig. 7). In this model, both the RDRM at the DNA level and an additional RNA-dependent mechanism serve to regulate gyrA expression during recovery from IR. Our data still support the current model in the literature whereby the repressor protein DdrO binds to DNA to restrict expression to a low basal level until activation of the IrrE protease and subsequent cleavage of DdrO after IR permits a rapid upregulation of gyrA for function in DNA repair mechanisms. However, it is plausible that the UTR_gyrA is also activated by IR by an IrrE-dependent mechanism (Fig. 6B) that involves additional cellular factors to aid survival. While further characterization is necessary to determine the regulatory mechanism of UTR_gyrA, the proposed mechanism of UTR_gyrA could be utilized by D. radiodurans as a fine-tuning mechanism of gene expression for recovery after irradiation.

FIG 7.

Model for UTR_gyrA regulation in which gyrA expression is controlled by the RDRM sequence consensus for DdrO regulation and a currently unknown mechanism for RNA-based regulation at the 5′ UTR. The current model for gyrA RDR-based regulation involves the radiation-induced activation of IrrE that cleaves DdrO to remove its ability to repress at the RDRM of the RDR gene. The release of the repressor DdrO then permits rapid upregulation of gyrA. In the expanded model, the 5′ UTR of gyrA serves as an additional regulatory region that is not dependent on RDRM sequence consensus to regulate gyrA expression.

It is important to note that in this work we have determined that RNA structure is an important element of UTR_gyrA regulatory function. Specifically, our RDRM sequence mutant (UTRScramgyrA) did not result in a decrease in survival to IR (Fig. 5C), despite the importance of RDRM sequence consensus for RDR function and the importance of gyrA expression for survival during IR recovery. These results are in stark contrast to other reports whereby RDRM mutants for both DR0070 (ddrB) (Fig. S7) (22) and DR906 (gyrB) (69) have resulted in a disruption of RDR-based regulation. Most importantly, our results suggest that other features of the UTR_gyrA, besides RDRM sequence consensus, are important in regulating gyrA expression during IR recovery. Interestingly, gyrA is not the only RDR-containing gene that contains its RDRM in the 5′ UTR (most RDRMs are near the TSS; however, some are located in the promoter region, the 5′ UTR, and even the coding sequence). Another RDR gene, pprA (DRA0346), also has been shown to contain the RDRM in the 5′ UTR by primer extension analysis, suggesting a common RNA regulatory mechanism of these genes (73). While the transcriptional and posttranscriptional activities of these 5′ UTRs have not been explored fully in previous research, it is possible that our expanded RDR model (Fig. 7) might apply to these other genes in the RDR regulon. However, further research is necessary to determine the range of this mechanism in D. radiodurans. Overall, our results suggest that the RDR stress response mechanism might rely on two aspects: conservation of a DNA consensus sequence (RDRM) and preservation of 5′-UTR RNA structure.

It is worth noting that some of the 31 5′-UTR candidates presented in this research are likely to respond to different stresses that this study did not examine. The specificity of UTR response is demonstrated by the lack of overlapping responsive 5′-UTR candidates between IR and H₂O₂-induced stress conditions tested in this work. This specificity further reinforces the specialization of these regulatory regions for particular environmental stresses and even levels/doses thereof. The possibility is maintained that other Deinococcus species utilize similar RNA regulatory systems for efficient gene expression control under a variety of stressors, including IR. It is also important to note that we did not observe a significant post-IR response of the other RDRM-containing gene in this study (uvrA, DR1771) (Fig. 3A); however, this could be due to the lower binding affinity of DdrO for that region (32). Overall, our research suggests that other features besides the RDRM sequence consensus are important in eliciting RDR regulatory activity during IR recovery.

MATERIALS AND METHODS

Plasmids and strains.

All plasmids and strains are included in Table 1. Wild-type Deinococcus radiodurans R1 (ATCC 13939; reference genome NC_001263) served as the parent strain utilized for all the D. radiodurans test strains and mutants, unless otherwise noted. The ΔkatA strain was kindly provided by Michael Daly (67). The 5′-UTR disruption (ΔUTR_gyrA) and RDRM scramble (UTRScramgyrA) mutants were constructed in this study by using a suicide plasmid (carrying a selective marker, a counterselectable marker, and a rare cutting endonuclease site) to introduce the intended mutation by homologous recombination (R. Saldanha and T. J. Lamkin, personal communication) The resulting merodiploid integrant had a duplication of the genomic sequence with the wild-type and mutant sequences separated by plasmid sequences. Introduction of a plasmid expressing the rare cutting endonuclease introduced a double-stranded break stimulating homologous recombination, which resulted in the eviction of integrated plasmid sequences, and restoring the merodiploid chromosome to either the wild-type configuration or fixing the intended mutation into the chromosome. Counterselection was used to eliminate the endonuclease-expressing plasmid. Genetic testing for plasmid-borne selection markers and PCR screening was used to verify the creation of the markerless mutant gene replacement and loss of plasmid markers. The ΔirrE strain was gifted by Pascale Servant (22). The plasmids used for testing differential expression of 5′-UTR–GFP fusions (pDR-GFP) were synthesized (GenScript) from the pRADgro vector (57) with each 31 5′ UTRs (see Table S3 in the supplemental material) inserted between the groESL promoter and GFP sequence. Upon sequencing of the pRADgro vector, we determined a 20- to 30-random-nucleotide insertion upstream of the promoter compared to the published plasmid (Fig. S2A and S3A); however, we determined that this derivation does not impact expression. The pRADgro-GFP construct was synthesized using restriction cloning with ApaI and XbaI cut sites to insert a synthesized D. radiodurans codon-optimized GFP after the groESL promoter. The gyrA complementation strains (pGyrUTRgyrA and pGroUTRgyrA) were constructed using Gibson cloning with the pRADgro backbone and gyrA PCR fragments from genomic DNA. The theophylline riboswitch construct (pThRS-GFP) was a gift from Nancy Kelley-Loughnane. Sequences for all plasmids are contained in Fig. S3, and those for primers are in Table S4. All D. radiodurans strains were cultured in TGY (1% tryptone, 0.5% yeast extract, and 0.1% glucose) medium (BD Difco) at 32°C.

TABLE 1.

Plasmid and strain list

Plasmid or strain	Description	Source, reference, and/or strain	Plasmid map	Sequence
Plasmids
pRADgro	Shuttle vector for E. coli and D. radiodurans using groESL promoter; Chlr (D. radiodurans) Ampr (E. coli)	57	Fig. S2A	Fig. S3A
pRADgro-GFP	Shuttle vector for E. coli and D. radiodurans with codon-optimized GFP and groESL promoter; Chlr Ampr	This study
pThRS-GFP	pRADgro with theophylline riboswitch upstream of GFP (pRADgro-ThRS-GFP); Chlr Ampr	This study; 79	Fig. 3A	Fig. S3B
pDR-GFP	pRADgro reporter plasmid with candidate 5′-UTR sequences named by gene number (Table S1) inserted upstream of GFP (pRADgro–5′UTR–GFP); Chlr Ampr	This study; GenScript	Fig. S2B	Fig. S3C
pDR1913-GFP (UTR_gyrA)	pRADgro reporter plasmid with DR1913 5′ UTR (DNA gyrase subunit A gene, UTR_gyrA) upstream of GFP; Chlr Ampr	This study; GenScript
pDR1857-GFP (UTR_ohrP)	pRADgro reporter plasmid with DR1857 5′ UTR (gene encoding organic hydroperoxide resistance protein; UTR_ohrP) upstream of GFP; Chlr Ampr	This study; GenScript
pGyrUTRgyrA	pRADgro expression plasmid with DNA gyrase subunit A gene (DR1913) FLAG tagged and expressed with native gyrA 5′ UTR (pRADgro-1913UTR-gyrase A-FLAG); Chlr Ampr	This study	Fig. S2C	Fig. S3D
pGroUTRgyrA	pRADgro expression plasmid with DNA gyrase subunit A gene (DR1913) FLAG tagged and expressed with groESL 5′ UTR (pRADgro-GroESLUTR-gyrase A-FLAG); Chlr Ampr	This study	Fig. S3E	Fig. S3E
Strains
D. radiodurans R1	Wild-type D. radiodurans	ATCC 13939
ΔkatA mutant	R1 strain with genomic deletion of catalase, katA (DR1998); Kanr	67; KKW7003
ΔUTR_gyrA mutant	R1 strain with genomic deletion of the DNA gyrase subunit A gene (DR1913) 5′ UTR	This study	Fig. 5A
UTRScramgyrA mutant	R1 strain with the DNA gyrase subunit A gene RDRM 5′ UTR genomically scrambled by compensatory mutations	This study	Fig. 5A
ΔUTR_gyrA-pGyrUTR mutant	ΔUTR_gyrA strain expressing pGyrUTRgyrA plasmid; Chlr Ampr	This study
ΔUTR_gyrA-pGroUTR mutant	ΔUTR_gyrA strain expressing pRADgro-GroELSUTR-gyrase A-FLAG plasmid; Chlr Ampr	This study
ΔirrE mutant	ΔirrEΩkan (R1 strain with knockout of irrE); Kanr	22; GY14127
ΔirrE-pDR1913-GFP mutant	ΔirrEΩkan (R1 strain with knockout of irrE) expressing pDR1913GFP; Chlr Ampr Kanr	This study

Transformation of D. radiodurans.

Transformation of D. radiodurans was performed based on a previously described procedure with minimal modifications (38, 74). D. radiodurans R1 cells grown to late log phase (optical density at 600 nm [OD₆₀₀], ∼1) were mixed with 30 mM CaCl₂ (J. T. Baker) and 10% glycerol (Sigma-Aldrich) to make them competent and stored at −80°C. Competent cells were incubated with 1 μg of plasmid DNA on ice for 30 min, followed by incubation at 30°C for 1 h. Transformed cells were then diluted 1:4 with fresh TGY medium in a new tube and grown at 32°C for 18 h. After incubation, cells were pelleted at 3,000 rpm for 5 min and resuspended in 150 μl TGY medium for plating onto TGY plates with the appropriate antibiotic. Plates were then incubated for 2 days at 32°C.

RT-PCR.

As an initial means of experimentally confirming the presence of predicted 5′ UTRs in their corresponding mRNAs, reverse transcription PCR (RT-PCR) analysis was conducted. Total RNA was extracted from wild-type D. radiodurans R1 at mid-log phase (OD₆₀₀, 0.6 to 0.8) by using vigorous bead beating and TRIzol (Invitrogen) as previously described (38). The extracted RNA was then treated with DNase I (New England BioLabs) for 1 h at 37°C and denatured at 65°C. cDNA was obtained using random hexamer oligonucleotides (N₆) (gene-specific primers were used to troubleshoot some reactions [Table S4]) and a Superscript III reverse transcription kit (Invitrogen) according to the manufacturer's protocols. The cDNA was then subjected to PCR with primers designed to yield amplicons of just the coding region (primers Y and Z, as labeled in Fig. 1C and Table S1) and the coding region plus the 5′ UTR (if present) (primers X and Z, as labeled in Fig. 1C and Table S1). These primers were designed arbitrarily to yield an amplicon of a convenient length (100 to 800 nucleotides) for visualization by agarose gel electrophoresis. A negative control (lacking the RT polymerase) was used to control for potential genomic DNA contamination. The PCR products were run on 1% agarose gels and visualized with EZ-vision dye (Amresco) on a ChemiDoc XRS+ imager (Bio-Rad).

RT-qPCR.

Reverse transcription-quantitative PCR (RT-qPCR) was performed using cDNA samples prepared as described above using random hexamer oligonucleotides (N₆). qPCR was run using gene-specific primers for 16S rRNA (as a reference gene) and the DNA gyrase subunit A gene (DR1913) or GFP (as the gene of interest) (Table S4). RT-qPCR was performed using Power SYBR green master mix (Life Technologies) in the Applied Biosystems ViiA 7 real-time PCR system (Foster City, CA) in the Core DNA facility at the Institute for Cellular and Molecular Biology, University of Texas at Austin. Fold changes in gene expression were determined using the 2^−ΔΔCT method as described in reference 75 for technical triplicate samples in at least two independent experiments.

5′ RACE.

For determining the transcriptional start sites (TSS) of 5′ UTRs, rapid amplification of cDNA ends (RACE) was conducted as previously published (38, 76). Specifically, the FirstChoice RLM-RACE kit (Ambion) was utilized per the manufacturer's protocol. In brief, total RNA was extracted from wild-type D. radiodurans R1 and 10 μg was ligated to the 5′-RACE adaptor (provided in the kit) using T4 RNA ligase at 37°C for 1 h. The RNA-5′-RACE adaptor product was then reverse transcribed, as before, using the provided Moloney murine leukemia virus (MMLV) reverse transcriptase enzyme and random decamer primers (N₁₂) at 42°C for 1 h. The 5′-UTR cDNA was then PCR amplified and sequenced to determine the TSS (Table S4).

Conservation analysis using LocARNA.

As a means of identifying functional 5′-UTR candidates, conservation of the candidate 5′ UTRs in D. radiodurans R1 was compared to that of the sequences of all members of the most closely related Deinococcus and Thermus phylum (2) with annotated genomes. Candidate 5′ UTRs were subjected to the NCBI blastn tool for identifying sequence similarity among the Deinococcus-Thermus phylum. The BLAST parameters were designed for the least-stringent matching, as we hypothesized that structural conservation could still be present with low sequence similarity (word size at 7, match/mismatch at 1/−1, and gap cost existence 0:extension 2). Any sequences identified with an E value of less than 10⁻⁸ were utilized for conservation analysis and structure prediction with LocARNA (default parameters), which utilizes ClustalW to align multiple sequences and generate consensus structures based on conserved sequences and base pairing probability (53–55).

Fluorescence-based flow cytometry.

A BD FACScalibur flow cytometer with a 488-nm argon laser and 530-nm fluorescence (FL1) logarithmic amplifier was utilized for observing differential expression of green fluorescent protein (GFP) in our 5′-UTR test strains (D. radiodurans expressing pDR-GFP) under various stressors. One hundred microliters of each technical triplicate sample from at least two independent experiments was pelleted and resuspended in 1 ml of filter-sterilized 1× phosphate-buffered saline (PBS) (Amresco) in polyethylene cytometer tubes (Falcon). As we were unable to identify any prior protocols for assaying fluorescence in D. radiodurans with a flow cytometer, some parameters had to be established. Initially, D. radiodurans cells had to be gated, or isolated, based on forward scatter (FSC) versus side scatter (SSC) to gate for the viable D. radiodurans population (Fig. 2B). In the case of a bimodal population distribution in which the first peak was consistent with the empty-vector (pRADgro) control population, the fluorescent population was gated and the median fluorescence (FL1) was derived from that population to remove bimodal bias from the uninduced cell population (Fig. 2B). CellQuest Pro (BD) software (or FlowJo, LLC, software for the bimodal populations) was used to attain the median fluorescence of each cell population in triplicate. Median fluorescence values were normalized using a ratio of the IR-treated fluorescence to the sham 0 kGy fluorescence to obtain the fold change in fluorescence that permitted comparison of multiple replicate experiments. Significance was determined based on a one-tailed unpaired Student's t test between the median fluorescence values of triplicate irradiated samples/sham-treated samples for the empty vector (pRADgro) and each individual 5′-UTR construct. Any sample with results with a P value of <0.05 was determined to be significantly different from the empty-vector control and was further examined.

Theophylline riboswitch activity assay.

Establishment of the fluorescence screen for the 5′-UTR–GFP fusion candidates was done using the well-characterized synthetic theophylline riboswitch (ThRS) (58). Biological triplicates of D. radiodurans R1 containing the pThRS-GFP plasmid (with codon-optimized GFP) (Fig. S2B and S3B) were grown to mid-log phase (OD₆₀₀, 0.6 to 0.8) and induced with 2 mM theophylline (Sigma-Aldrich) (control with DMSO). Fluorescence was measured via flow cytometry 2 h after induction as described above.

High-dose irradiations (1 to 15 kGy).

Samples (5-ml-volume cultures) of D. radiodurans R1 were harvested at mid-log phase (OD₆₀₀, 0.6 to 0.8) and double sealed in polyethylene bags (2-oz Whirl-Pak Bags; Nasco) that were subsequently placed inside larger polyethylene specimen bags (7-oz Whirl-Pak Bags; Nasco) in technical triplicates in at least two independent experiments. Samples were then refrigerated in coolers with dry ice and transported for irradiation. Acute doses of ionizing radiation (1 to 15 kGy) and sham irradiations were conducted with a 10-MeV, 18-kW linear accelerator (LINAC) β-ray source at the National Center for Electron Beam Research, Texas A&M, College Station, TX. The samples were irradiated on wet ice to keep the cell growth static. Shortly after irradiation, samples were transferred to 15-ml conical tubes and diluted 2-fold in fresh TGY medium. The samples were allowed to recover in incubators at 32°C for 2 h prior to flow cytometry (as described above) (38) (Fig. S5).

IR survival assay.

To obtain the comparable percentages of survival for each of the IR dosages tested (0, 1, 5, 10, and 15 kGy), samples of D. radiodurans R1 were grown and irradiated as described above. However, samples were not permitted to recover and were instead kept static on dry ice until plating. Biological triplicate samples were plated at various dilutions (from 10⁻³ to 10⁻⁷) on TGY agar plates and incubated at 32°C for 2 days. Colonies were then counted and normalized to the counts at 0 kGy. Additionally, spot plates were done for each dilution and dose. Survival fractions of 0.23 were comparable to those previously reported in the literature for the same corresponding dose of 10 kGy, while survival fractions of 0.01 at 15 kGy were more equivalent to the published fractions with survival of 13 to 14 kGy (2).

Low-dose irradiations (0.1 to 0.5 kGy).

Samples were prepared as before for the high-dose irradiations in double-sealed polyethylene bags. For the low-dose irradiations (0.1 to 0.5 kGy), a dual-source Cs¹³⁷ gamma irradiator was utilized. The samples were placed in the center of the irradiation chamber, and dry ice was placed along the sides of the chamber to keep samples at 4°C (ice was replenished as necessary). The samples were exposed at a rate of 47.1 Gy/h and recovered as described before in fresh medium for 2 h. Samples were then analyzed using flow cytometry as described above.

Hydrogen peroxide stress assay.

Biological triplicate samples (5-ml-volume cultures) of D. radiodurans strain R1 or ΔkatA mutant were grown to an OD₆₀₀ of 0.6 to 0.8 for treatment with 20 mM H₂O₂. The stress concentration was determined based on literature concentration ranges (46). Samples were incubated with 20 mM H₂O₂ (Fisher Chemical) for 30 min at 4°C. Following treatment, samples were recovered with fresh medium for a 2-fold dilution and incubated for 2 h at 32°C. After recovery, a 100-μl sample was resuspended in 1 ml PBS for analysis by flow cytometry as described above. Survival assays of R1 and ΔkatA strains treated with H₂O₂ were performed as described above for IR survival (data not shown).

Western blotting.

D. radiodurans protein lysate was obtained from cells grown to an OD₆₀₀ of ∼0.8 (10-ml cultures) and treated with the appropriate IR stress as described above. The cell pellet was stored at −80°C postirradiation until processing. The pellet was resuspended in 500 μl of 1×PBS and lysed using a probe sonicator (XL-2000 Microson ultrasonic liquid processor; QSonica) voltage output of approximately 10 V for 1 min for six bursts with 10 min of rest on ice between each burst to prevent overheating and denaturing of proteins. Following sonication, the sample was then centrifuged at 15,000 rpm to pellet the cellular debris and insoluble protein. Soluble protein lysate was obtained from the supernatant and quantified using a Bradford assay (utilizing Coomassie protein assay reagent by Thermo Scientific).

Western blotting was performed with minimum modifications from previously published protocols (77). Briefly, SDS-PAGE using a 12% bisacrylamide gel was loaded with 4 μg total protein of each boiled protein sample in denaturing sodium dodecyl sulfate (SDS) loading buffer (125 mM Tris-HCl [pH 6.8; Fisher Scientific], 25% glycerol, 2% SDS [Sigma-Aldrich], 0.01% bromophenol blue [Sigma-Aldrich], and 0.5% β-mercaptoethanol) (Amresco). Gels were run in duplicate to allow for one as a loading control (SYPRO-Ruby stained using the manufacturer's protocol [Invitrogen]), as typical housekeeping protein expression was not assumed to be constant under irradiation. Following SDS-PAGE, we transferred the gel to a 0.2-μm nitrocellulose membrane (Bio-Rad) by electroblotting at 17 V for 35 min using a Trans-Blot semidry transfer cell (Bio-Rad). Detection of GFP was achieved by using an anti-GFP antibody (Roche; 11814460001) at a 1:1,000 dilution and an anti-mouse–horseradish peroxidase (HRP) conjugate (Promega; W4021) at a 1:2,500 dilution. Detection of gyrA-FLAG utilized an anti-FLAG antibody (Agilent) at a 1:10,000 dilution and the anti-mouse–HRP conjugate. All antibodies were diluted in 1% nonfat milk (in Tris-buffered saline [TBS]: 20 mM Tris, 500 mM NaCl), and membranes were blocked overnight at 4°C with 5% nonfat milk (in TBS) to minimize nonspecific binding. Chemiluminescent detection using Bio-Rad's Clarity Western enhanced chemiluminescence (ECL) substrate provided images of bands of interest (GFP). Quantitation of Western blot bands was done utilizing the TotalLab Quant band volume detection protocol that was normalized to total lane intensity in the SYPRO-ruby gel.

Mass spectrometry.

Protein lysate was obtained as described above, and the concentration was measured using the Bradford assay. Equal amounts of protein were run on a 12% SDS-PAGE gel. Gel bands between 60 and 100 kDa (as determined by the ladder) were cut and in-gel trypsin digested based on previous published protocols (78). Briefly, cut gel bands were dehydrated with 100% acetonitrile and then reduced with 10 mM dithiothreitol (DTT) for 30 min at room temperature. The gel was then alkylated with 50 mM iodoacetamide in the dark at room temperature for 30 min. Gel bands were washed with 100 mM ammonium bicarbonate solution and then dehydrated again with 100% acetonitrile. The dehydrated gel was then digested with 10 ng/μl trypsin (in 50 mM ammonium bicarbonate) overnight at 37°C. Protein was then extracted from the gel using 5% formic acid and 1:2 (vol/vol) 5% formic acid–acetonitrile. Before being run on a mass spectrometer, protein samples were dried in a SpeedVac and then resuspended in 0.1% formic acid. The samples were then run through a Zip Tip (Millipore) with C₁₈ resin prepared according to the manufacturer's instructions. Following a wash with 0.5% trifluoroacetic acid (TFA), the peptides were eluted from the resin using elution buffer (67% acetonitrile [ACN], 32.8% water, 0.2% TFA). The eluted protein samples were then dried and resuspended in 7 μl of 0.1% formic acid. Samples were then injected into a Thermo Orbitrap Elite hybrid linear ion trap Fourier transform mass spectrometer (FT-MS) with Dionex 3000 nanospray ultraperformance liquid chromatography (UPLC) and run for 1 h per sample. Peak areas were calculated using the Skyline Proteomics software (MacCoss Lab Software) and normalized to total ion intensity.