RecA is essential for extreme radiation tolerance in Deinococcus radiodurans. Interestingly, Sahara bacterium Deinococcus deserti has three recA genes (recAC, recAP1, recAP3) that code for two different RecA proteins (RecAC, RecAP). Moreover, and in contrast to other sequenced Deinococcus species, D. deserti possesses homologues of translesion synthesis (TLS) DNA polymerases, including ImuY and DnaE2. Together with a lexA homologue, imuY and dnaE2 form a gene cluster similar to a widespread RecA/LexA-controlled mutagenesis cassette. After having developed genetic tools, we have constructed mutant strains to characterize these recA and TLS polymerase genes in D. deserti. Both RecAC and RecAP are functional and allow D. deserti to survive, and thus repair massive DNA damage, after exposure to high doses of radiation. D. deserti is mutable by UV, which requires ImuY, DnaE2 and RecAC, but not RecAP. RecAC, but not RecAP, facilitates induced expression of imuY and dnaE2 following UV exposure. We propose that the extra recAP1 and recAP3 genes may provide higher levels of RecA protein for efficient error-free repair of DNA damage, without further increasing error-prone lesion bypass by ImuY and DnaE2, whereas limited TLS may contribute to adaptation to harsh conditions by generating genetic variability.
To preserve genome integrity, cells have developed strategies to protect and repair their genome. The bacterium Deinococcus radiodurans and other members of the Deinococcaceae are famous for their extraordinary tolerance towards high doses of gamma and UV radiation or long periods of desiccation. This extreme tolerance is linked to the ability of these bacteria to accurately repair massive DNA damage, including hundreds of double-strand DNA breaks and thousands of other lesions (Cox and Battista, 2005; Blasius et al., 2008). It has been proposed that DNA repair in Deinococcus could be facilitated by a high intracellular Mn/Fe ratio that prevents oxidation of proteins (Daly et al., 2007) and by a highly condensed nucleoid that may restrict diffusion of DNA fragments generated by irradiation (Zimmerman and Battista, 2005).
Recently, the genome of Deinococcus deserti, a bacterium isolated from surface sand of the Sahara desert (de Groot et al., 2005), was sequenced and analysed (de Groot et al., 2009). Its genome consists of a 2.8 Mb chromosome and three large plasmids of 324, 314 and 396 kb. Besides many common genes, D. deserti has several additional DNA repair genes compared with D. radiodurans and D. geothermalis. Interestingly, among these are genes encoding three putative TLS polymerases and a potential photoproduct lyase. Moreover, and highly uncommon, D. deserti has three different recA genes, one located on the chromosome and the other two on different plasmids. The plasmid recA genes code for identical proteins that have 81% identity with RecA encoded by the chromosome. RecA is nearly ubiquitous (Rocha et al., 2005), and most known bacterial species, including D. radiodurans and D. geothermalis, possess only one recA. Two recA were found in Bacillus megaterium (Nahrstedt et al., 2005), Myxococcus xanthus (Norioka et al., 1995), and seven in Acaryochloris marina (Swingley et al., 2008). A recA was also found on a 65 kb conjugative lactococcal plasmid (Garvey et al., 1997) and on a 53 kb plasmid from an environmental Vibrio strain (Hazen et al., 2007). RecA from E. coli has been studied most extensively. It is a multifunctional protein with a key role in DNA repair and genetic recombination (Cox, 2007a,b). Through its coprotease activity, it also has a regulatory role in induction of the SOS response. Following DNA damage, activated RecA nucleoprotein filaments are formed that facilitate the autocleavage of repressor LexA, resulting in induction of many DNA repair genes including recA itself and TLS polymerase genes (Sutton et al., 2000; Erill et al., 2007; Butala et al., 2009). The RecA coprotease activity also facilitates cleavage of phage repressor proteins and of UmuD to UmuD′, a component of TLS polymerase V (Pol V). In D. radiodurans, RecA stimulates cleavage of its two LexA homologues, but contrary to E. coli, recA induction is not LexA-dependent (Narumi et al., 2001; Bonacossa de Almeida et al., 2002; Sheng et al., 2004; Satoh et al., 2006). Instead, recA induction is dependent on a novel regulatory protein IrrE, also called PprI (Earl et al., 2002a; Hua et al., 2003).
In this study, we characterized the three recA genes, as well as the genes encoding the potential TLS polymerases and photoproduct lyase in D. deserti. This work included the development of tools allowing the construction of D. deserti gene deletion mutants. Survival after exposure to radiation, UV-induced mutagenesis and regulation of gene expression were investigated in wild-type and mutant strains. Analysis of various recA mutant strains showed that both RecA proteins are functional for DNA repair. Induced expression of each recA was dependent on irrE. UV-induced mutagenesis was observed and shown to depend on two TLS polymerases, DnaE2 and a Y-family DNA polymerase, ImuY. Their radiation-induced expression was dependent on the chromosome-encoded RecA, but not on IrrE and the plasmid-encoded RecA. To our knowledge, this is the first time that different functions for different RecA proteins within a bacterial species are demonstrated.
Three recA, three TLS polymerases and a putative photoproduct lyase in D. deserti
In contrast to D. radiodurans and D. geothermalis, D. deserti has genes encoding putative TLS DNA polymerases (Deide_1p00180, Deide_1p01880 and Deide_1p01900) and a putative photoproduct lyase (Deide_3p02150), located on plasmid P1 and plasmid P3 respectively (Fig. 1). It has also three different recA genes (Deide_19450, Deide_1p01260 and Deide_3p00210, present on the chromosome, plasmid P1 and plasmid P3 respectively). Except for Deide_1p00180, transcription of all these genes was induced after exposure of D. deserti to UV, suggesting a role in the DNA damage response (de Groot et al., 2009).
Deide_1p00180 (PolB) shares up to 48% identity with DNA polymerases of the B-family (DNA polymerase II). Deide_1p01880 encodes a distantly related homologue of the DinB subfamily of Y-family DNA polymerases. After blast analysis with Deide_1p01880, several proteins annotated as DNA polymerase IV were found, but with less than 30% identity. As experiments in this work demonstrated that Deide_1p01880 is involved in induced mutagenesis (see below), it will be named imuY. Deide_1p01900 (DnaE2) is a potentially error-prone polymerase homologous to the alpha subunit DnaE (Deide_21950) of DNA polymerase III. The genes imuY and dnaE2 are located close to each other, separated by a small hypothetical orphan gene of 243 bp (Fig. 1). The gene immediately upstream of imuY encodes a LexA homologue (Deide_1p01870), with the highest similarity to D. radiodurans LexA (DR_A0344, 50% identity). RT-PCR analysis indicated that lexA, imuY and dnaE2 are part of a single operon (Fig. S1). Similar so-called mutagenesis cassettes, in different configurations, have recently been described in various groups of bacteria, but not previously in members of the Deinococcus–Thermus phylum, and shown to be under RecA/LexA control (Erill et al., 2006; 2007). In most of these cassettes, a gene called imuA is present upstream of the Y polymerase gene (imuB), but an imuA homologue was not found in the genome of D. deserti. Furthermore, similarity between D. deserti ImuY and ImuB proteins is very low, e.g. only 15% identity with ImuB of Caulobacter crescentus and Pseudomonas putida.
Using the Conserved Domain Database (Marchler-Bauer et al., 2007), analysis of Deide_3p02150 revealed a COG1533 domain, defined as SplB DNA repair photolyase. However, Deide_3p02150 has only limited homology with spore photoproduct lyases such as Bacillus subtilis SplB (26% identity). It should be noted that SplB is not a photolyase as it does not use light for activity but a radical mechanism. blast analysis with Deide_3p02150 revealed higher levels of identity (up to 64%) with proteins annotated as radical SAM domain proteins (SAM, S-adenosylmethionine). D. deserti and several of the species that contain homologues of Deide_3p02150 do not form spores. Like Deide_1p00180 (polB), Deide_3p02150 is not adjacent to other known DNA repair genes (Fig. 1).
The genetic environments of the three recA genes in D. deserti are different (Fig. 1). As in D. radiodurans and D. geothermalis, the chromosomal recA of D. deserti is the last gene in an operon that also contains cinA and ligT, encoding a DNA damage/competence-inducible CinA-like protein and a 2′−5′ RNA ligase respectively. The two recA genes on the plasmids share 97% identity at the DNA level and code for 100% identical proteins (RecAP), highly homologous to the chromosome-encoded RecAC (81% identity). Both for RecAC (about 90% identity) and for RecAP (about 80% identity), the next best hits in blast analyses were the RecA proteins from D. radiodurans and D. geothermalis, suggesting that RecAP is of deinococcal origin. Figure 2A permits a direct comparison of RecAC and RecAP. A multiple alignment of these proteins with RecA from other Deinococcaceae and E. coli is shown in Fig. 2B. The major differences between RecAC and RecAP are present at the N- and C-termini, and between residues 240 and 255 (RecAC numbering). Compared with RecAP and RecA from E. coli, a conserved extension of 11–13 residues is present at the N-terminus of RecAC and RecA from D. radiodurans and D. geothermalis. This region is flexible in the crystal structure of D. radiodurans RecA (Rajan and Bell, 2004). The C-terminal 15–25 residues are variable in sequence in RecA proteins and are disordered in published crystal structures. E. coli RecA activity is autoregulated by its C-terminus; deletion of the last 17 residues enhances most RecA activities (Cox, 2007a,b). For the region of E. coli RecA that corresponds to residues 240–255 of D. deserti RecAC, analysis of point mutations and structural data have established that residues in this region are implicated in coprotease substrate binding and cleavage (Dutreix et al., 1989; Mustard and Little, 2000; McGrew and Knight, 2003; VanLoock et al., 2003).
Development of genetic tools for D. deserti
To investigate the role in D. deserti of the genes described in the previous section, it was necessary to obtain genetic tools for this bacterium. Unlike D. radiodurans, D. deserti appeared not to be naturally transformable under the conditions tested. Therefore, other methods were sought to genetically modify D. deserti. The aim was to obtain a transformation method, a plasmid that can replicate in D. deserti and a system to inactivate/delete genes. We tested different transformation techniques (electroporation, transformation using a CaCl2 method and conjugation) while varying the parameters in each method, as well as different DNA molecules (various available and newly constructed plasmids, purified from various strains, PCR fragments). Details and results of these assays are described in Supporting information. Although transformants were obtained with each method, a classic CaCl2 transformation method appeared to be most appropriate for D. deserti. Moreover, the D. radiodurans–E. coli shuttle vector pI3 could be introduced in D. deserti as a replicating plasmid. About 100–1000 transformants per microgram of pI3 plasmid were obtained by this technique when pI3 was extracted from E. coli, against 10000 transformants per microgram when pI3 was extracted from D. deserti. Gene deletion mutants of D. deserti could be obtained by allelic replacement via double homologous recombination after transformation with plasmid constructs derived from E. coli vector pUC19, which does not replicate in D. deserti. In these plasmids, fragments corresponding to DNA upstream and downstream of the gene of interest are cloned in the correct orientation, respectively, upstream and downstream of a kanamycin resistance cassette. Expression of the kanamycin resistance gene in this cassette is driven by the constitutive promoter of the tuf gene (Deide_18970, encoding elongation factor EF-Tu). Only 10–20 kanamycin-resistant transformants were usually obtained, which was sufficient for the work described here, i.e. the construction and characterization of D. deserti gene deletion mutants. Double homologous recombination at the correct locus and complete absence of the gene of interest was systematically confirmed by diagnostic PCR. After D. radiodurans, D. deserti is the second Deinococcus species for which a system for gene disruption has been developed.
Each single recA mutant is as UV and gamma radiation-resistant as the wild type
To analyse the role of each recA, three mutant strains were constructed in which one of the recA genes was deleted. The chromosomal recA mutant, ΔrecAC, showed growth characteristics as the wild type (data not shown). For several other bacteria, various growth defects upon recA inactivation have been described, such as reduced growth rate and plating efficiency for D. radiodurans (Bonacossa de Almeida et al., 2002) and filamentous growth for Thermus thermophilus (Castan et al., 2003) and B. subtilis (Sciochetti et al., 2001). Moreover, whereas deletion of recA in D. radiodurans results in extreme sensitivity to radiation (Bonacossa de Almeida et al., 2002), the D. desertiΔrecAC strain is as resistant to UV and gamma irradiation as the wild type (Fig. 3). These results strongly suggest that the supplementary recA genes on the plasmids code for a functional RecAP protein, allowing normal growth and radiation resistance of the ΔrecAC strain. The two other single mutants, ΔrecAP1 and ΔrecAP3, also presented a normal growth and radiation resistance (data not shown).
In addition to the kanamycin resistance cassette described above, a second antibiotic resistance cassette was needed to construct D. deserti double mutants. For this, it appeared to be possible to use the DrPtuf::cat cassette, containing the chloramphenicol resistance gene fused to the promoter of the D. radiodurans tuf gene (Earl et al., 2002b). The ΔrecAP1ΔrecAP3 double mutant showed growth (data not shown) and radiation resistance characteristics as the wild-type strain (Fig. 3), indicating that RecAC is also functional. Two other double mutants, in which either recAP1 or recAP3 remains intact, were also constructed. Note that the ΔrecAC strain RD40 was used as the parental strain for the construction of the ΔrecACΔrecAP3 double mutant, showing that RecAC is not specifically required for gene replacement by homologous recombination. The ΔrecACΔrecAP3 strain grew normally and was radiation resistant (data not shown). However, the ΔrecACΔrecAP1 mutant showed a severely affected growth rate and formed long filaments (results not shown). Therefore, in this double mutant the remaining recAP3 did not permit normal growth. As recAP1 and recAP3 code for identical proteins, these results indicate that recAP3 expression is weak compared with that of recAP1, at least under the experimental conditions used. This is in agreement with our previously reported RT-PCR data that indicated that recAP3 transcription levels are lower than that of recAP1 and recAC (de Groot et al., 2009). The promoter activities of recAP1 and recAP3 may be different: the first 72 bp upstream of their start codons is nearly identical (only one difference), but the sequences further upstream are different. Radiation resistance was strongly reduced in the ΔrecACΔrecAP1 strain (Fig. 3). However, survivors were still observed at the highest doses tested (0.001% at 15 kGy γ rays), in contrast to D. radioduransΔrecA mutants that are extremely sensitive to radiation (< 0.001% at 2 kGy) (Bonacossa de Almeida et al., 2002; Tanaka et al., 2004), suggesting that weak expression of recAP3 allows repair of some DNA damage. A triple recA mutant may provide evidence for the contribution of recAP3 to DNA repair and survival. However, as a third antibiotic resistance cassette functional in D. deserti is currently unavailable, this triple recA mutant, if viable, could not be constructed.
In D. radiodurans, radiation tolerance and DNA damage-induced expression of recA requires the presence of irrE (Earl et al., 2002a; Hua et al., 2003), encoding a Deinococcus-specific regulator protein (de Groot et al., 2009; Vujicic-Zagar et al., 2009). A D. desertiΔirrE mutant, which grows normally under standard conditions, is extremely sensitive to radiation (Fig. 3). The ΔirrE strain is less tolerant to radiation than the ΔrecACΔrecAP1 double mutant, probably because IrrE regulates multiple DNA repair and protection pathways (Hua et al., 2003; Lu et al., 2009). Moreover, the ΔrecACΔrecAP1 strain still contains intact recAP3.
To get further insight in the expression levels of the three different recA genes, Western blotting experiments were performed with cell extracts of the different mutant strains that were irradiated or not with UV. Due to the small difference in migration, both RecAC and RecAP could be detected, confirming that D. deserti produces two different RecA proteins (Fig. 4). Only RecAC was visible in un-irradiated cell extracts, whereas both RecA proteins were clearly induced after irradiation of the wild-type strain. Analysis of the various mutants showed that the large majority of RecAP in the cells was encoded by recAP1, in agreement with the observed weak transcription of recAP3. Nevertheless, the immunoblot results indicate that low levels of RecAP3 were induced, which probably contribute to DNA repair and the observed residual radiation tolerance of the ΔrecACΔrecAP1 double mutant. The results further showed that irrE is required for induction of both RecAC and RecAP (Fig. 4).
TLS DNA polymerases ImuY and DnaE2 are involved in UV-induced mutagenesis in D. deserti
To investigate the role of the putative TLS DNA polymerase and photoproduct lyase genes in D. deserti, four mutant strains with deletions of the corresponding genes were constructed, ΔpolB, ΔimuY, ΔdnaE2 and ΔDeide_3p02150. A double mutant, Δ(imuY-dnaE2), in which the DNA fragment encompassing both imuY and dnaE2 was deleted, was also constructed. The mutants grew normally and showed the same survival rates as the wild-type strain after exposure to UV or gamma irradiation (data not shown).
Translesion synthesis polymerases often generate mutations due to their low fidelity. To determine whether D. deserti's TLS polymerases are involved in error-prone lesion bypass, rifampicin assays were performed after UV irradiation. This test detects point mutations in the rpoB gene that confer a rifampicin-resistant phenotype (RifR). First, we analysed whether induced mutagenesis could be demonstrated in the wild-type strain. This was indeed the case, since an approximately 60-fold increase in the number of RifR mutants was observed after exposure to UV (Fig. 5A). For the ΔpolB strain, the amount of RifR mutants was similar as for the wild type (Fig. 5A). However, an about 10-fold decrease in the number of UV-induced RifR mutants was observed with the ΔimuY and ΔdnaE2 strains (Fig. 5A). The double mutant Δ(imuY-dnaE2) showed the same decrease of induction of RifR mutants as the ΔimuY and ΔdnaE2 single mutants, suggesting that ImuY and DnaE2 act in the same pathway. Complementation of the ΔimuY mutant by plasmid pRD52, carrying lexA-imuY, indicated that the decreased mutagenesis in this mutant was not due to a potential negative effect on dnaE2 expression (Fig. 5C). RT-PCR analysis indicated that dnaE2 expression in the ΔimuY mutant occurred by read-through transcription from the kanamycin resistance cassette (results not shown). If the putative photoproduct lyase Deide_3p02150 is involved in removal of DNA lesions, its inactivation might result in increased mutagenesis. However, the rifampicin assay revealed a similar amount of RifR mutants for the ΔDeide_3p02150 strain as for the wild type (data not shown). In conclusion, the results indicate that ImuY and DnaE2 are functional and involved in error-prone lesion bypass in D. deserti, whereas our experiments did not reveal a role for polB and Deide_3p02150.
Chromosome-encoded RecA, but not plasmid-encoded RecA, is required for induced-mutagenesis
In E. coli and other bacteria, RecA is involved in induced mutagenesis. Following DNA damage, RecA stimulates autocleavage of repressor protein LexA, resulting in induced expression of genes encoding TLS polymerases and other DNA repair proteins. E. coli RecA also facilitates cleavage of UmuD to UmuD′, a component of TLS Pol V. It also participates directly in lesion bypass by interacting with Pol V (Fuchs and Fujii, 2007; Schlacher and Goodman, 2007). In D. deserti, a lexA homologue is present immediately upstream of imuY and dnaE2, like in similar, RecA/LexA-regulated mutagenesis cassettes identified in several Proteobacteria (Abella et al., 2004; Erill et al., 2006).
To see if one or several recA genes are involved in UV-induced mutagenesis in D. deserti, the rifampicin assay was performed with the various recA mutant strains. A strong decrease in number of RifR mutants was observed in the mutants lacking recAC (Fig. 5B). This reduction was comparable with that found in the ΔimuY and ΔdnaE2 mutant strains. In contrast, high amounts of RifR mutants were still obtained with the ΔrecAP1, ΔrecAP3 and ΔrecAP1ΔrecAP3 strains. Introduction of plasmid pRD76 carrying recAC in the ΔrecAC mutant restored induced mutagenesis (Fig. 5C). Therefore, chromosome-encoded RecAC, but not plasmid-encoded RecAP, is required for UV-induced mutagenesis.
RecAC is required for induction of imuY and dnaE2 expression
Our results show that recAC, imuY and dnaE2 are required for UV-induced mutagenesis in D. deserti. To investigate if recAC is involved in transcriptional regulation of imuY and dnaE2, RT-PCR experiments were performed. Whereas the expression of imuY and dnaE2 is induced upon UV irradiation in the wild-type strain, induction of these genes was not visible in the ΔrecAC strain (Fig. 6). In contrast, UV-induced expression of imuY and dnaE2 was still present in the ΔrecAP1ΔrecAP3 double mutant (Fig. 6). Therefore, the chromosome-encoded RecAC protein, but not plasmid-encoded RecAP, is required for induction of the TLS polymerase genes imuY and dnaE2 that are present on plasmid P1. These results are in agreement with those obtained in the rifampicin assays.
Transcription of the various genes characterized in the present study was also analysed in the D. desertiΔirrE mutant (Fig. 6). Induction of each recA gene was dependent on irrE (note that 35 PCR cycles were used for amplification of the recAP3 fragment, and 30 for recAP1 and recAC). However, UV-induced transcription of imuY and dnaE2 was still observed in the ΔirrE strain, indicating that un-induced levels of RecAC are sufficient to facilitate induction of imuY and dnaE2 following UV damage. When UV-induced mutagenesis was analysed in the ΔirrE strain, a 6- to 10-fold increase in the number of RifR mutants was observed in comparison with the wild type (data not shown), most likely because error-free repair pathways are no longer induced in the absence of IrrE.
Expression of the polymerase II (polB) and the putative photoproduct lyase (Deide_P302150) genes was also analysed in the various strains (Fig. 6). As in the wild type, exposure to UV did not result in induced transcription of polB in the ΔirrE and ΔrecA mutant strains (if anything, polB expression is rather slightly diminished after UV exposure). For the putative photoproduct lyase, the observed induction was dependent on irrE.
Both RecAC and RecAP promote LexA cleavage in vitro
As described for similar mutagenesis cassettes in other bacteria, e.g. P. putida (Abella et al., 2004), expression of lexA-imuY-dnaE2 in D. deserti is likely repressed by its own encoded LexA, and derepressed by RecA-mediated cleavage of LexA. Since RecAC, but not RecAP, is required for induced expression of imuY and dnaE2, only RecAC might be able to stimulate cleavage of LexA. This was tested in vitro, using RecAC, RecAP and LexA proteins that were purified after expression in E. coli. Unlike RecAC (and LexA), purified RecAP started to precipitate rapidly after purification, and this could not be prevented by immediate dialysis against another buffer. Experiments were therefore performed immediately after purification of the proteins. Purified LexA, RecAC and RecAP migrated on SDS polyacrylamide gel with their expected molecular masses (Fig. 7). Similar to E. coli and D. radiodurans LexA (Narumi et al., 2001), D. deserti LexA undergoes autodigestion when subjected to alkaline conditions leading to two breakdown products (Fig. S2). When testing RecA-mediated digestion at pH 7.4, LexA was cleaved by incubation with either RecAC or RecAP to produce two breakdown products of the same size as those observed in the autodigestion (Fig. 7). The observed cleavage was less efficient with RecAP than with RecAC. However, we cannot exclude that the concentration of available RecAP in the reaction mixture was decreased due to protein precipitation (see above). Both for RecAC and for RecAP, this cleavage required the presence of the oligonucleotide, ATPγS and MgCl2 (Fig. 7 and Fig. S2).
Bacterial species belonging to the Deinococcaceae are famous for their extreme tolerance to ionizing and UV radiation and other stresses that provoke massive DNA damage. Genome sequences are available for D. radiodurans (White et al., 1999), D. geothermalis (Makarova et al., 2007) and D. deserti (de Groot et al., 2009), species that were isolated from irradiated canned meat, a hot spring and Sahara surface sand respectively. A common radiation/desiccation response motif (RDRM) in these species is present upstream of the same DNA repair genes, such as recA, pprA, ddrA, uvrA, uvrB (Makarova et al., 2007; de Groot et al., 2009). A conserved radiation response regulon is further supported by the observation that D. deserti irrE is able to restore radiation resistance in a D. radioduransΔirrE mutant (Vujicic-Zagar et al., 2009). D. deserti possesses several DNA repair genes that are absent from the two other sequenced Deinococcus species (de Groot et al., 2009), including a putative photoproduct lyase (Deide_3p02150) and three TLS polymerase genes (polB, imuY, dnaE2). Moreover, D. deserti has three recA genes (recAC, recAP1, recAP3), which code for two different RecA proteins (RecAC, RecAP), whereas D. radiodurans and D. geothermalis have only one recA. The aims of this work were (i) the development of genetic tools for D. deserti, (ii) to investigate the functional expression of the three recA, the three TLS polymerases and the putative photoproduct lyase, and (iii) to see if the presence of the extra recA genes is correlated with the other supplementary genes. After construction and characterization of various single and double mutant strains, we conclude that both RecAC and RecAP are functional proteins that allow repair of massive DNA damage after exposure of D. deserti to high doses of gamma and UV radiation. ImuY and DnaE2 are involved in UV-induced point mutagenesis. Not the supplementary RecAP but RecAC facilitates induction of imuY and dnaE2 expression. IrrE is involved in induction of each recA and of the putative photoproduct lyase gene, but it is not required for imuY and dnaE2 induction.
Of the genes characterized in this work, the RDRM was found upstream of each recA (de Groot et al., 2009) (Fig. 1). Together with data on D. radiodurans, their IrrE-dependent radiation-induced expression suggests a link between IrrE and this motif. Indeed, IrrE-dependent induction of several other genes with the RDRM in their upstream region, such as ddrB and ddrC, was also observed in D. deserti (S. Fochesato and A. de Groot, unpubl. data). IrrE possibly regulates the putative photoproduct lyase gene in a different manner, because the RDRM was not found upstream of Deide_3p02150. It should be noted that direct binding of IrrE to the RDRM has not been demonstrated (Ohba et al., 2005; Vujicic-Zagar et al., 2009).
The absence of TLS polymerase genes in the genome of D. radiodurans (and D. geothermalis) is in agreement with previous studies that reported that D. radiodurans and other Deinococcus species are immutable by UV and that they repair UV-induced DNA damage accurately (Sweet and Moseley, 1974; Tempest and Moseley, 1982; Tanaka et al., 2005). Moreover, it was thought that error-prone TLS might be imprudent in Deinococcus (Sale, 2007). However, our results with D. deserti demonstrate that an error-prone pathway for the repair of such damage is not necessarily absent from all Deinococcaceae. As most generated mutations will be deleterious, expression and/or activity of ImuY and DnaE2 should be tightly regulated to allow the majority of the many lesions generated by intense radiation to be repaired by error-free mechanisms. Various bacterial proteins are involved in accurate repair of UV-induced lesions (Goosen and Moolenaar, 2008), and D. deserti expresses several of these, including UV-damage endonuclease (UVDE/UvsE) and UvrABC proteins (de Groot et al., 2009).
Deletion of polB or the photoproduct lyase-like gene (Deide_3p02150) did not result in decrease or increase of induced mutagenesis. For polB, this might be because PolB activity is error-free, PolB is expressed under conditions different from those applied here, or PolB generates frameshift mutations (Becherel and Fuchs, 2001), which cannot be detected in the rifampicin assay that detects point mutations. Deide_3p02150 has only low similarity with known spore photoproduct lyases, and there is no evidence that Deide_3p02150 is indeed a protein involved in DNA repair. However, its UV-induced and IrrE-dependent expression suggests a role in the DNA damage response. Deide_3p02150 might be involved in repair of specific lesion(s) formed during concomitant exposure to UV and desiccation. Although deletion of the TLS polymerase genes and Deide_3p02150 did not affect tolerance to irradiation under the experimental conditions, i.e. in the presence of water and nutrients, their role in survival in harsh environmental conditions such as the hot and arid desert may be more important. In addition, mutations generated by TLS will increase genetic variability, which may lead to better adaptation of D. deserti to its hostile habitat (Nohmi, 2006).
In this article we report, for the first time, the presence of two different RecA proteins with partly different functions in a single bacterial species. Only RecAC facilitates UV-induced mutagenesis, most likely by promoting cleavage of LexA encoded by the lexA-imuY-dnaE2 cassette. Moreover, as observed in the ΔirrE mutant, only basal levels of RecAC are sufficient for induction of imuY and dnaE2. In contrast, induced levels of RecAP, although functional for repair of massive DNA damage in the ΔrecAC mutant, do not allow induction of imuY and dnaE2, possibly resulting from its inability to promote (sufficient) LexA cleavage in vivo. Indeed, in vitro experiments showed that LexA cleavage is less efficient with RecAP than with RecAC. Interestingly, one of the regions that are different in RecAC and RecAP, i.e. residues 240–255 (RecAC numbering), corresponds to a region in E. coli RecA that has been implicated in coprotease substrate binding and cleavage (Dutreix et al., 1989; Mustard and Little, 2000; McGrew and Knight, 2003; VanLoock et al., 2003). Therefore, the difference in this region of the RecA proteins might be an explanation for the inability of RecAP to induce TLS. Nevertheless, coprotease activity of RecAP is not abolished, and TLS induction may be controlled by another mechanism in addition to LexA binding and cleavage efficiency. This is supported by observations that not all RecA-DNA filaments induce the SOS response in E. coli. It has been suggested that initially formed RecA nucleoprotein filaments are competent for recombination, but that SOS induction requires a change in property of the filament or formation of extended filaments (Gruenig et al., 2008; Long et al., 2008). In D. deserti, only RecAC filaments might adopt this special conformation, whereas both RecAC and RecAP filaments are competent for recombination. Like for E. coli RecA, not all RecAC filaments may adopt the conformation necessary for LexA cleavage. Previous studies have shown a competition between secondary DNA and LexA for binding to a RecA nucleofilament (Harmon et al., 1996; Rehrauer et al., 1996). Proteins that modulate RecA function (Cox, 2007b) may play a role in determining which process is undertaken, recombination or LexA cleavage (Gruenig et al., 2008; Long et al., 2008).
Several interesting questions remain. In addition to TLS induction, are there other functional differences between RecAC and RecAP? They likely have largely overlapping functions, but RecAC and RecAP may be optimized for different activities in for example homologous recombination, repair of stalled replication forks or RecA-mediated excision repair (Ishimori et al., 1996; Namsaraev et al., 1998; Bichara et al., 2007; Cox, 2007a,b). Furthermore, RecAC and RecAP may have yet unidentified different coprotease targets. Do RecAC and RecAP form mixed nucleoprotein filaments, or are they targeted to different DNA substrates? What is the advantage to possess two different RecA proteins (three recA genes)? Under the experimental conditions, either RecAC or RecAP is sufficient for extreme radiation resistance. However, as described for the TLS polymerases, the presence of multiple recA genes may contribute to survival under nutrient-poor, dry and UV-exposed conditions in the desert. In such an environment, growth will be slow and frequently arrested while DNA damage will accumulate due to desiccation and UV. The extra recAP genes may allow synthesis of higher levels of RecA for a rapid error-free repair of the accumulated DNA damage, without further inducing error-prone lesion bypass.
The RecA/LexA-regulated SOS system, a response mechanism to address DNA damage in bacteria, has been studied most extensively in E. coli where it regulates about 40 genes. Studies in other bacteria have revealed substantial heterogeneity in LexA regulon contents. In D. radiodurans, induction of recA and other repair genes is LexA-independent and requires IrrE. The presence in D. deserti of two different RecA and both IrrE- and RecA-dependent regulation of DNA repair genes illustrates further diversity of DNA damage response regulons among different bacterial species.
Bacterial strains, plasmids and growth conditions
The strains and plasmids used in this study are listed in Table 1. D. deserti was grown at 30°C in 10-fold diluted tryptic soy broth (TSB/10) supplemented with trace elements (Vujicic-Zagar et al., 2009). Antibiotics were used at the following concentrations for D. deserti: streptomycin, 10 μg ml−1; kanamycin, 10 μg ml−1; chloramphenicol, 3 μg ml−1; rifampicin, 10 μg ml−1. E. coli was grown in Luria–Bertani medium (LB), when necessary supplemented with 50 μg ml−1 kanamycin or 100 μg ml−1 ampicillin.
Table 1. List of strains and plasmids used in this study.
An exponential culture of D. deserti was centrifuged (5000 g) and cells were re-suspended in 0.1 M CaCl2 (half of the initial culture volume) and placed overnight at 4°C. Cells were centrifuged again and re-suspended in 1/15 of initial culture volume of 0.1 M CaCl2 with 20% glycerol. The cells were stored at −80°C until use. The mixture of cells and DNA was placed on ice for at least 30 min, followed by a heat shock at 42°C for 2 min, and then again on ice for 1–2 min. Five hundred microlitres of TSB/10 plus trace elements was added and cells were incubated for 6 h at 30°C before plating.
Deinococcus deserti chromosomal DNA was extracted from 10 ml of stationary-phase cells that were harvested by centrifugation (10 min, 5000 g) and re-suspended in 500 μl of 50 mM Tris-HCl, pH 8, 50 mM EDTA, 0.15 M NaCl. Thirty microlitres of lysozyme (10 mg ml−1) was added and the mixture was incubated for 1 h at 37°C. Then 100 μl of 10% SDS, 15 μl of proteinase K (20 mg ml−1) and 15 μl of pronase E (20 mg ml−1) were added and the mixture was incubated for 1 h at 37°C. One volume of phenol–chloroform–isoamyl alcohol (25:24:1) was added and the solution was vigorously mixed and centrifuged at 13200 r.p.m. for 5 min in a table centrifuge (Eppendorf). After a second phenol–chloroform–isoamyl alcohol extraction, the DNA was precipitated by addition of 1 vol. of isopropanol and centrifugation at 13200 r.p.m. for 30 min. DNA was washed with 70% ethanol, and centrifuged again at 13200 r.p.m. DNA was re-suspended in 100–200 μl of water. DNA concentration was measured at OD260 (Eppendorf Biophotometer). The QIAprep spin miniprep Kit (Qiagen) was used for plasmid extraction from E. coli.
PCR and plasmid construction
For PCR amplification of DNA fragments for cloning purposes, Pfx polymerase (Invitrogen) was used. For diagnostic PCR, HotGoldstar (Eurogentec) was used following the instructions of the manufacturer but with addition of DMSO (5% final concentration) in the PCR mix. Primers are listed in Tables S1–S4.
For cloning, the blunt-end PCR products were purified with the QIAquick gel extraction kit (Qiagen) and first inserted into the pCR®4Blunt-TOPO® vector (Invitrogen). DNA fragments were recloned in other plasmids using appropriate restriction enzymes. Derivatives of plasmids pRD0 and pRD36, which do not replicate in D. deserti, were used to construct D. deserti gene deletion mutants by double homologous recombination. Plasmid pRD0 contains a BamHI–PstI fragment with the kanamycin resistance gene translationally fused to the 370 bp promoter region of D. deserti gene tuf (Deide_18970; encoding elongation factor EF-Tu), and was constructed as described in Supporting information. Plasmid pRD36 contains a BamHI–PstI fragment with a fusion of the D. radiodurans tuf promoter to the chloramphenicol resistance gene. This DrPtuf::cat cassette was amplified from pGTC101 with primers DRtuf1Bam and CatPst (Table S1). Fragments of 1–1.5 kb corresponding to DNA upstream and downstream of genes to be deleted were amplified from D. deserti chromosomal DNA (with primers listed in Table S2), and cloned in the correct orientation, respectively, upstream and downstream of the cassettes in pRD0 (Ptuf::kan) or pRD36 (DrPtuf::cat). The resulting plasmids were used to transform D. deserti. Diagnostic PCR to confirm double homologous recombination at the correct locus and complete absence of the gene of interest was performed using primers listed in Table S3.
For pRD52, a 2.1 kb XbaI–HindIII DNA fragment carrying lexA-imuY (Deide_1p01870-Deide_1p01880) and 185 bp upstream of lexA with its probable promoter was cloned in pI3. Plasmid pRD76 contains a 385 bp XbaI–EcoRI fragment, corresponding to the probable promoter of the cinA-ligT-recA operon and 85 bp of the beginning of cinA, fused at the EcoRI site to a 1.2 kb EcoRI–HindIII DNA fragment carrying the last 96 bp of ligT and entire recAC (Deide_19450). The pET-TEV derivatives pET-TEV/LexA, pET-TEV/RecAC and pET-TEV/RecAP contain NdeI–XhoI fragments with the genes lexA (Deide_1p01870), recAC (Deide_19450) and recAP1 (Deide_1p01260) respectively. The cloned fragments in the pI3 and pET-TEV derivatives were amplified with the primers listed in Table S1. Cloned DNA fragments were verified by nucleotide sequencing (Genome Express, France).
RNA isolation and RT-PCR
For RNA isolation from D. deserti, cells were treated with RNAprotect Bacteria Reagent (Qiagen) and RNA was isolated using the RNeasy Mini Kit (Qiagen). RNA samples were treated twice with DNase. RT-PCR was performed in two steps. First, cDNA was synthesized in 20 μl of reactions using 1 μg of RNA and the Transcriptor First Strand cDNA Synthesis Kit (Roche). In the second step, DNA fragments of 150–250 bp were amplified in 25 μl reactions using 1 μl of cDNA from the first step, Taq polymerase (Sigma) and gene-specific primers (Table S4). These amplifications were carried out by incubating reactions at 94°C for 10 min prior to 30 cycles of 1 min at 94°C, 30 s at 56°C and 30 s at 72°C, followed by a final step at 72°C for 2 min, with modifications for tuf (25 cycles), Deide_3p02150 (32 cycles), recAP3 and dnaE2 (both 35 cycles). Controls for DNA contamination were performed with reactions lacking reverse transcriptase.
Gamma and UV irradiation and induced mutagenesis
For gamma irradiation, cells were grown overnight to exponential phase (OD600 0.1–0.3), concentrated 10-fold in growth medium and exposed on ice to gamma rays for the desired dose (39.6 Gy min−1, 60Co source, CEA-Cadarache, France). Then, 20 μl of serial dilutions were spotted on solidified medium to determine survival.
To determine survival after UV irradiation, cells were grown to exponential phase (OD600 0.2–0.4). The cultures were serially diluted until 10−5 and 4 ml of dilutions in Petri dishes were exposed to UV-C (254 nm) for the desired dose, and then 100 μl were spread on plates. Colonies were counted after 3–4 days of incubation. For analysis of gene or protein expression after UV irradiation, two samples from the same culture in exponential phase (OD600 0.2–0.4) were taken. One sample was exposed to UV-C (4 ml portions in Petri dishes) and the other was not irradiated. Then both were further incubated, and samples for RNA isolation or total-cell extract preparation were taken at the desired time point. For mutagenesis experiments, the same cultures as those for testing survival to UV were used. Four millilitres of non-diluted cultures were UV-irradiated for the desired dose and then incubated for 20 h at 30°C with shaking to permit fixation of mutations. In case of strains containing plasmids (complementation), irradiated cultures were first centrifuged and cells were re-suspended in fresh medium with chloramphenicol prior to incubation for 20 h. Then, 1 ml of culture was spread on plates supplemented with rifampicin. Serial dilutions of these cultures were also spread on plates without antibiotic. The number of rifampicin-resistant mutants was observed after 3–4 days at 30°C. The frequency of mutation per 109 colony-forming units (cfu) was calculated by (number of rifampicin-resistant cfu ml−1/total number of cfu ml−1) × 109.
Preparation of cell extracts for immunoblotting
Cells from 10 ml of culture (OD600 0.3–0.6) were collected by centrifugation, washed with 1 ml of 10 mM Tris-HCl pH 8.0, and stored at −80°C until use. Cells were re-suspended in 0.5 ml of 50 mM Tris-HCl pH 8.0, 5 mM EDTA, supplemented with protease inhibitor cocktail as recommended by the manufacturer (Sigma, P8849), and disrupted by sonication on ice (six times for 10 s with pauses of 1 min; microtip, Vibra Cell 72408, Bioblock Scientific). After centrifugation for 5 min at 13200 r.p.m. at 4°C (Eppendorf table centrifuge) to remove cell debris, the supernatant was kept as total-cell extract. Protein concentrations were measured using CooAssay Protein Dosage Reagent UPF86420 (Uptima/Interchim). After addition of SDS sample buffer, extracts were heated for 10 min at 95°C.
Protein separation and immunoblot analysis
Fractions of cell extracts were subjected to electrophoresis through a 10% Bis-Tris NuPAGE gel (Invitrogen). The proteins were transferred to a nitrocellulose membrane (Whatman, 10401 196) using a semi-dry transfer system (Bio-Rad) and then treated with a 1:1000 dilution of polyclonal antibodies raised against RecA protein from E. coli (generous gifts of S. Sommer and S. Marsin). The goat anti-rabbit peroxidase (Sigma, A6154) was used as a second antibody. The antigen–antibody complexes were visualized with G:BOX (SynGene) using the SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific).
Expression and purification of LexA and RecA proteins
In plasmids pET-TEV/LexA, pET-TEV/RecAC and pET-TEV/RecAP, the lexA, recAC and recAP genes are under the control of the T7 promoter. Each protein produced contains an N-terminal His tag of 21 residues (MGSSHHHHHHSSGENLYFQGH) including a TEV protease cleavage site (underlined). For expression of LexA, E. coli BL21(DE3) cells (Invitrogen) freshly transformed with pET-TEV/LexA were grown at 37°C overnight to saturation in 10 ml of LB medium containing 50 μg ml−1 kanamycin. This pre-culture was then diluted in 1 l of LB medium, 50 μg ml−1 kanamycin (in a 3 l flask) and grown at 37°C, 160 r.p.m. At OD600 of 0.6–0.7, isopropyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 0.1 mM, and the cells were grown at 17°C for 17 h. RecAC and RecAP were expressed similarly, except that E. coli BL21 (AI) cells (Invitrogen) were used, and not only IPTG but also l-arabinose (0.2% final concentration) was added to induce gene expression. The induced cells were harvested by centrifugation and cell pellets were frozen at −20°C.
For purification, cell pellets containing the recombinant proteins were re-suspended in 30 ml of sodium phosphate buffer (20 mM) pH 8, 500 mM NaCl, supplemented with 25 μl of protease inhibitor cocktail (Sigma, P8849) and 1 mg of DNase (Sigma, DN25). Cells were disrupted by French press (two cycles) and the soluble extracts were then recovered after centrifugation at 10 000 g for 10 min at 4°C and ultracentrifugation at 150 000 g for 45 min at 4°C. The supernatants were injected onto HisTrapTM HP columns (1 ml) (GE Healthcare), previously equilibrated in sodium phosphate buffer (20 mM) pH 8, 500 mM NaCl, containing 20 mM imidazole. A step gradient of imidazole (20, 40, 100 and 500 mM) was used for elution (20 ml was used for each fraction except for the 500 mM fraction where 4 ml was used for RecAC and RecAP and 2.5 ml for LexA). Eluted fractions were analysed by SDS-PAGE for the presence of each protein and verification of the protein purity. Protein concentration was determined by the CooAssay Protein Dosage Reagent UPF86420 (Uptima/Interchim). The mass of each purified protein was verified on a micrOTOF-Q mass spectrometer (Bruker, Wissembourg, France).
RecA-mediated LexA cleavage reaction
The protocol was adapted from Satoh et al. (2006). The reaction mixture contained 20 mM Tris-HCl pH 7.4, 10 mM MgCl2, 10 μM oligonucleotide (5′-GCGTGTGTGGTGGTGTGC-3′) (Giese et al., 2008), 1 mM ATPγS, 8 μM RecA (RecAC or RecAP) and 7 μM LexA. To activate RecAC and RecAP by binding with ATPγS and oligonucleotide, the mixture without LexA was first pre-incubated at 37°C for 20 min. The reaction was then initiated by addition of LexA, and the reaction mixture was incubated at 37°C for 2 h. After quenching by addition of NuPAGE® LDS sample buffer (Invitrogen) and heating at 95°C for 10 min, the reaction products were subjected to a 10% Bis-Tris NuPAGE gel (Invitrogen). LexA and breakdown products were visualized by staining with ImperialTM protein stain (Pierce). The LexA autodigestion reaction in alkaline conditions was performed by incubating 7 μM LexA in 50 mM Tris-HCl pH 10.5 at 37°C for 20 h.
We are grateful to S. Sommer and S. Marsin for generous gifts of anti-RecA antiserum, and to P. Servant for plasmids. We thank S. Cuiné and C. Brutesco for help with Western blotting and detection, J. Vicente for access to gamma irradiation facilities, G. Brandelet for help with protein purification and D. Lemaire for mass spectrometry. This work was supported by the Commissariat à l'Energie Atomique and the Agence Nationale de la Recherche (ANR-07-BLAN-0106-02).