Basal DNA repair machinery is subject to positive selection in ionizing-radiation-resistant bacteria

Background Ionizing-radiation-resistant bacteria (IRRB) show a surprising capacity for adaptation to ionizing radiation and desiccation. Positive Darwinian selection is expected to play an important role in this trait, but no data are currently available regarding the role of positive adaptive selection in resistance to ionizing-radiation and tolerance of desiccation. We analyzed the four known genome sequences of IRRB (Deinococcus geothermalis, Deinococcus radiodurans, Kineococcus radiotolerans, and Rubrobacter xylanophilus) to determine the role of positive Darwinian selection in the evolution of resistance to ionizing radiation and tolerance of desiccation. Results We used the programs MultiParanoid and DnaSP to deduce the sets of orthologs that potentially evolved due to positive Darwinian selection in IRRB. We find that positive selection targets 689 ortholog sets of IRRB. Among these, 58 ortholog sets are absent in ionizing-radiation-sensitive bacteria (IRSB: Escherichia coli and Thermus thermophilus). The most striking finding is that all basal DNA repair genes in IRRB, unlike many of their orthologs in IRSB, are subject to positive selection. Conclusion Our results provide the first in silico prediction of positively selected genes with potential roles in the molecular basis of resistance to γ-radiation and tolerance of desiccation in IRRB. Identification of these genes provides a basis for future experimental work aimed at understanding the metabolic networks in which they participate.


Background
In this paper, we consider ionizing-radiation-resistant bacteria (IRRB) as "non-spore-forming bacteria that can protect their cytosolic proteins from oxidation and tolerate many DNA double-strand breaks (DSBs) after exposure to high, acute ionizing radiation (dose greater than 1 kilogray (kGy) for 90% reduction (D 10 ) in Colony Forming Units (CFUs)), and can resist prolonged desiccation." Over the past five decades, Deinococcus radiodurans (D 10 ≈ 15 kGy) has been a model for understanding many of the basic principles that govern resistance to ionizing radiation and tolerance of desiccation (for review see [1]). Recent findings indicated that the mutual nature of D. radiodurans's γ-radiation resistance and desiccation tolerance resides in cytosolic Mn-dependent mechanisms that protect proteins from oxidative modifications that intro-duce carbonyl groups [2,3]. Like most IRRB, D. radiodurans accumulates about 300 times more Mn(II) than ionizing-radiation-sensitive bacteria (IRSB) [4] such as Escherichia coli (D 10 ≈ 0.7 kGy) and Thermus thermophilus (D 10 ≈ 0.8 kGy) [5,6]. Due to its high concentration of intracellular Mn(II) ions and the consequent protection of proteins, D. radiodurans can survive 10 kGy of ionizing radiation, a dose that causes approximately 100 DNA DSBs per genome. This species can also survive 8 days in a desiccator with no obvious DNA DSBs [2][3][4]7,8]. Early on, analysis of D. radiodurans's transcriptional response revealed that the cellular responses to ionizing radiation and desiccation exhibit substantial overlap [9]. In this context, 41 ionizing-radiation-sensitive strains of D. radiodurans were also shown to be sensitive to desiccation [7]. Furthermore, the mutational inactivation of the genes DR_1172 -encodes a homolog of LEA76, a group 3 LEA (late embryogenesis abundant) protein -and DR_B0118 -encodes a protein that is expressed during dehydration by the resurrection plant, Craterostigma plantagineumgreatly sensitized D. radiodurans to desiccation, but not to ionizing radiation [10]. The former result based on a collection of ionizing-radiation-sensitive strains of D. radiodurans is consistent with evidence that dried bacterial cells exhibit a substantial number of DNA DSBs, single-strand breaks, and DNA crosslinks [11], DNA damage that is also observed following exposure to ionizing radiation [12]. Indeed, ionizing radiation and desiccation introduce similar types of DNA damage -DNA DSBs -in D. radiodurans [7]. The latter result based on the mutational inactivation of the genes DR_1172 and DR_B0118 considerably strengthens the hypothesis that the D. radiodurans's desiccation tolerance could be a consequence of this organism's adaptation to ionizing radiation (radiation adaptation hypothesis), particularly that the origin of ionizing-radiation resistance in bacteria can be explained as an adaptation to environmental radiation [13]. For example, Deinococcus sp. have been isolated from the subsurfaces of hydrothermal vents at depths of 64.8-128.9 m below the sea floor (mbsf) [14]. D'Hondt and colleagues [15] have recently surveyed environments representative of a broad range of subsurface conditions that can be found in marine sediments. Among the most striking features of deeply buried sediments (20-100 mbsf) are Mnrich sites with high natural γ-radiation levels. Ionizingradiation levels in such deposits might have been much higher on ancient Earth (Daly MJ, personal communication, 2006). Concerning the close relationship between bacterial ionizing-radiation resistance and desiccation tolerance [2,3,7,9], many of the environments from which IRRB have been isolated can be considered to be dry, and it has been shown that many of these strains are also desiccation tolerant [16]. Moreover, it was demonstrated that IRRB are present at higher numbers in an arid soil than in a nonarid soil, and that IRRB in the arid soil are recovered after exposure to higher doses than the doses which allow recovery of bacteria from the nonarid soil [16]. In this context, many IRRB (i.e. Deinococcus deserti, Deinococcus sonorensis, etc.) have been isolated from the desert [16,17], indicating they have the capacity to adapt to harsh environments.
Three principal mechanisms may have contributed to the adaptability of IRRB: positive Darwinian selection, lateral gene transfer, and gene regulation [18]. Previous studies have discussed the importance of lateral gene transfer and gene regulation in Deinococcus evolution [5,6]. However, no previous research has addressed the role of positive Darwinian selection in resistance to ionizing-radiation and tolerance of desiccation in IRRB. In this paper, we assess the role of positive Darwinian selection by analysis of the genomes of IRRB. Among the many IRRB, completely sequenced genomes are only available for D. geothermalis (D 10 ≈ 15 kGy), D. radiodurans, Kineococcus radiotolerans (D 10 ≈ 2 kGy), and Rubrobacter xylanophilus (D 10 ≈ 5.5 kGy) [1,6,19,20]. When bacterial species of the genera Deinococcus (D. geothermalis and D. radiodurans) and Kineococcus (K. radiotolerans) were tested in vitro for desiccation tolerance, all were found to be tolerant to desiccation [4,21]. We are not aware of any studies that examined R. xylanophilus for desiccation tolerance, but since (1) many IRRB isolated from the desert -they are one way or another tolerant to desiccation -cluster within the monophyletic Rubrobacteria subgroup which is specialized for particular niches present in arid soils [22,23], (2) bacterial species of the genus Rubrobacter (R. radiotolerans, R. taiwanensis, and R. xylanophilus) belong to IRRB [24,25], and (3) there has to our knowledge never been a report of desiccation-sensitive IRRB, we included R. xylanophilus in our study.
Positive Darwinian selection leads to the fixation of advantageous mutations and is the fundamental process behind adaptive changes in genes (reference [26] and citations therein). Many statistical methods have been developed for detecting adaptive molecular evolution [27-29], such as the "conservative" PAML program [30,31]. In this report, we used a freely available software package with a user-computer interface (DnaSP) [32] to identify genes under positive selection in IRRB. This program calculates the ratios of nonsynonymous to synonymous mutation rates (Ka/Ks) in protein coding genes. The Ka/Ks ratio measures the strength of selection, with values > 1 indicative of positive selection [29].

Results and discussion
At the time we began this work, the GOLD database [20] documented four completely sequenced IRRB genomes: D. geothermalis DSM 11300 [6], D. radiodurans R 1 [19] . We identified 689 ortholog sets among IRRB as potentially subject to positive Darwinian selection (Additional file 1, genes in bold). Supporting our conclusion that these genes have been under selective pressure from niches with high natural γ-radiation or desiccation levels [13], we found that many of these positively selected genes are involved in resistance to γ-radiation resistance and tolerance of desiccation [6,9]. To find out if there were any biais due to phylogenetic relationships, patterns of positive selection in Deinococcus and Thermus species were compared, because phylogenetically the closest relative of Deinococcus is the genus Thermus [5]. As patterns of positive selection in Deinococcus and Thermus species were different [see Additional file 1], we concluded that DnaSP scores are related to resistance to ionizing-radiation and tolerance of desiccation phenotypes and not to the phylogenetic relationships of the species. But it should be noted that our bioinformatical method used to detect positive selection is based on sequence alignments, and therefore we cannot exclude that in some cases we have included false positives because of improper alignments.
We defined two important families of positively selected genes in IRRB: i) Group-I genes, which are under positive selection in IRRB but under neutral or purifying selection in IRSB (125 genes, highlighted in light gray in Additional file 1), and ii) Group-II genes, which are under positive selection in IRRB but absent in all IRSB (58 genes, highlighted in dark gray in Additional file 1). These two groups of positively selected genes in IRRB may provide insight into the molecular adaptations to ionizing radiation and desiccation. For instance, Group-II includes DNA repairrelated proteins (DR_0192, MutT/NUDIX family protein; DR_2074, putative 3-methyladenine DNA glycosylase), oxidative damage-related proteins (DR_2242, thiol-specific antioxidant protein; DR_2538, cytochrome P450) and water stress-associated proteins (DR_0463, maltooligosyltrehalose synthase; DR_0464, putative maltooligosyltrehalose trehalohydrolase). These conclusions are in concordance with DNA-centric hypotheses (for example, [33]), protein-centric hypotheses [2,3], and the water replacement with trehalose hypothesis [34] as protection mechanisms against ionizing radiation and desiccation. Our results support the idea that resistance to ionizing radiation and tolerance of desiccation are two complex phenotypes, and suggest that protection and repair mechanisms are complementary in IRRB.
The D. radiodurans genome has most of the DNA repair proteins found in E. coli [19]. Broad-based bioinformatic and experimental studies have concluded that D. radiodurans uses a relatively conventional set of DNA repair and protection mechanisms, but that these mechanisms are much more efficient than in IRSB [2,3,9,19]. We found that some of the accessory DNA repair genes in Deinococcus that are extremely important for resistance phenotypes (e.g., pprI (DR_0167, Dgeo_0395) and pprA (DR_A0346, Dgeo_2628) [35,36]) have no orthologs in Kineococcus and Rubrobacter. Similarly, the five transcripts of D. radiodurans (ddrA, DR_0423; ddrB, DR_0070; ddrC, DR_0003; ddrD, DR_0326; pprA, DR_A0346) that are most highly induced following ionizing radiation and recovery from desiccation [9] were not present in Kineococcus and Rubrobacter.
It seems likely that the shared ability of IRRB to survive the damaging effects of ionizing radiation and desiccation is the result of basal DNA repair pathways and that basal DNA repair genes have been acted upon by positive selection. Table 1 shows that, unlike many of their orthologs in IRSB [37], all DNA replication, repair, and recombination genes in IRRB were subject to positive selection. Three major DNA repair genes showed remarkably strong evidence for positive selection: DR_1707 (Ka/Ks = 3.15), DR_0906 (Ka/Ks = 3.4) and DR_1913 (Ka/Ks = 3.48).
Zahradka et al. [33] reported the remarkable efficiency of DNA repair enzymes in D. radiodurans during recovery from ionizing radiation and proposed a model named 'Extended Synthesis-Dependent Strand Annealing' (ESDSA). This model proposes that DNA polymerase I (PolA, DR_1707) accounts for the high fidelity of RecAdependent DNA DSB fragment assembly. recA (DR_2340), a central gene to genomic restoration, is upregulated in D. radiodurans cultures recovering from ionizing radiation and desiccation [9], and Table 1 shows that recA is positively selected. In particular, PolA of D. radiodurans supports very efficient DNA replication at the earliest stages of recovery and is present at higher levels than during normal DNA replication [33]. It is unclear whether polA paralogs participate in ESDSA. However, the presence of polA paralogs that are subject to positive selection (unlike orthologs in IRSB) suggests that these paralogs may be involved in genomic networks of resistance to γradiation and tolerance of desiccation (see Table 1). A positively selected polA paralog set might also explain the results of Gutman et al. [38], who showed that ionizingradiation-sensitive D. radiodurans polA mutants are fully complemented by expression of the polA gene from the relatively ionizing-radiation sensitive E. coli. A more com-plete examination of these paralogous proteins will be necessary to prove this hypothesis.
Genes encoding the two subunits of the DNA gyrase, DR_1913 (gyrA) and DR_0906 (gyrB), are also induced by ionizing radiation and desiccation [9], and this presumably explains their strong positive selection values. This implies that DNA supercoiling is important for DNA repair following the deleterious effects of ionizing radiation and desiccation. ). This is likely the result of continuous interaction between IRRB and natural niches, which have high levels of γ-radiation or high probability of desiccation [13][14][15][16][17].
The present work has identified the first suite of genes that are under positive Darwinian selection in IRRB. Our results can serve as a useful background to guide future physiological and biochemical experiments examining resistance to ionizing radiation and tolerance of desiccation. As more IRRB genome sequences become available, particularly in more distantly related species, we expect that our methods will provide a sensitive and controlled approach for detection of genes that have been subject to positive Darwinian selection for resistance to ionizing radiation and tolerance of desiccation.

Methods
To identify genes under positive selection in IRRB, the DnaSP program requires an aligned set of orthologous sequences [32]. Thus, we initiated our analyses by performing comparisons between all fully sequenced IRRB genomes (Deinococcus geothermalis DSM 11300 [6], Deinococcus radiodurans R 1 [19],