In all organisms studied to date, UV irradiation causes inducible responses. In some bacteria the inducible SOS response involves about 40 genes that are up-regulated dependent on the pleiotropic regulator, LexA 1. In eukaryotes a variety of genes are up- and down-regulated in response to UV-damage including several DNA repair genes, although no eukaryotic equivalent of the bacterial SOS response has been identified 2. Among archaea, most studies have heretofore been limited to comparative genomic approaches where inducible mechanisms have not been discernable 34. The hyperthermophilic archaea have been found to carry homologs of several eukaryotic nucleotide excision repair (NER) genes, including rad2 (FEN1/XPG), rad3 (XPD), eif4A (rad1/XPF), and rad25 (XPB) 56. Remarkably, extremely halophilic archaea, such as the model organism Halobacterium sp. strain NRC-1, and some non-thermophilic methanogenic archaea, were found to harbor homologs of both eukaryotic NER genes and bacterial NER genes, uvrA, uvrB, uvrC and uvrD 678. The classic SOS response system regulated by LexA is lacking in archaea, however, and the relationship between the bacterial and eukaryotic repair systems in archaea is not currently known 9.

Halophilic archaea, such as Halobacterium spp., are excellent experimental systems for studies of DNA repair because they are amongst the few archaeal microorganisms to encounter high levels of sunlight in their natural environment. They occupy an extreme environmental niche, where exposure to intense solar radiation leads to evaporation and concentration of NaCl to near- or even super-saturation. These microorganisms, including the sequenced wild-type model, Halobacterium sp. NRC-1, are highly resistant to the damaging effects of UV radiation in sunlight, principally due to extremely efficient photoreactivation of DNA damage 1011. In the presence of visible light they can survive UV doses many times higher than they would ever be exposed to naturally. Cell survival approaches 100 % after doses up to 100 J/m² while 1 hour exposure to sunlight inflicts damage equivalent to, at most, only a few J/m² [12; SM, unpublished]. UV tolerance of Halobacterium sp. NRC-1 is compared to other key organisms in Figure 1. Halobacterium is significantly more UV-tolerant, even without photoreactivating light, than Escherichia coli or Saccharomyces cerevisiae, though not as resistant as the extremely radiation-resistant Deinococcus radiodurans. Halophilic archaea also have excision repair mechanisms that can operate in the absence of photoreactivating light 1213. In addition, like bacteria and eukaryotes, they are likely to possess mechanisms that enable them to tolerate the presence of some unrepaired UV lesions, including lesion bypass by DNA polymerases that can circumvent photoproducts 14) and recombination to facilitate recovery of stalled replication forks 1516.

We have taken a multifaceted approach to the study of UV responses in Halobacterium sp. NRC-1. Previously, the critical role for phr2 in light repair was demonstrated through a combination of genetics and biochemistry 11. Here, we use DNA microarrays to show the importance of homologous recombination genes in the response of cells to UV damage. Interestingly, our results are distinct from those obtained in a previous study on Halobacterium sp. NRC-1 using significantly higher doses of UV irradiation 17.

In order to understand gene expression responses to UV at the whole genome level, we studied the model archaeon, Halobacterium sp. strain NRC-1, employing DNA microarrays. We used three different doses of UV-C, 30 J/m², 50 J/m², and 70 J/m², with post-irradiation incubation times in the dark of 1 hour and 3 hours (Materials and Methods). At these UV doses, survival of cells is close to 100 % following DNA damage either in the light or in the dark (Figure 1) 1118. Cells were irradiated in growth medium with post-irradiation incubation in the same medium so as to introduce minimal additional stresses. To give an indication of repair rates at the doses used, we determined the relative occurrences of cyclobutane pyrimidine dimers (CPDs) and 6-4 photoproducts over time after irradiation with an intermediate dose of 50 J/m² (Figure 2). The majority of photoproducts were repaired during the 3-hour period, though some cyclobutane dimers still remained at 3 hours.

Custom DNA microarrays were fabricated using inkjet technology (Agilent Technologies, Palo Alto, CA) with in-house oligonucleotide design performed with the program OligoPicker 19. The arrays contained 8,455 60-mer nucleotide features representing 2474 open reading frames (ORFs). The high specificity of 60-mer oligonucoletide arrays have been demonstrated previously 20. Up to three probes were designed per ORF with a mean T_m of 81°C and a T_m range of 3°C. The microarray slides harbor both gene-probes (~8,000 features per array) as well as ~400 negative and positive control spots to test hybridization conditions and allow for error modeling. These microarrays were thoroughly tested for linearity of response and statistical significance in a related study of anaerobic respiration in Halobacterium sp. NRC-1 21. Signal intensities with a dynamic range in excess of three orders of magnitude were found allowing simultaneous analysis of low- and high intensity features.

For microarray analysis, two doses of UV-C, 30 J/m² and 70 J/m² were used. Microarray data quality was considered high. After removal of outliers, features replicated within a single array showed low differences in absolute processed signal intensities (7 % on average); and spot-to-spot variation for replicate experiments was 9 %. Of the 2474 open reading frames (ORFs) of Halobacterium sp. NRC-1 represented, 100 were significantly up-regulated and 150 were significantly down-regulated (1.5-fold above or below, respectively, p-value < 0.05) in at least one experimental setting. In the current study, we focused on genes involved in homologous recombination and DNA metabolism, which are the most significantly induced (Table 1). Expression levels for all genes represented on the array for the 30 J/m² dose are also provided (see Additional file 1). Data obtained from UV irradiation with 70 J/m² were essentially the same (not shown). At either dose, the pattern of inducible transcripts was very different from what would be seen in E. coli, not surprisingly, since there is no LexA homolog in Halobacterium sp. NRC-1. Moreover, there was no evidence of a classic coordinated SOS response and neither the prokaryotic DNA repair genes, including uvrA, uvrB, uvrC and uvrD nor any of the eukaryotic repair gene homologs were up-regulated.

Strikingly, however, the radA1 gene showed the most inducible transcript (~7 fold) at 1 hr and 3 hrs after both UV doses (Table 1; Figure 3). RadA1 is the Halobacterium sp. NRC-1 homolog of RecA/Rad51, which catalyses strand invasion and exchange during homologous recombination. RadA1 is more similar to Rad51 in eukaryotes than to bacterial RecA. Deletion of radA has been shown to cause severe UV sensitivity in the related halophililc archaeon, Haloferax volcanii 22. The expression change of radA1 was considered statistically highly significant based on the following: The radA1 gene is represented by 3 different probes per array with one of those probes being duplicated, combining to a total of 24 data points for that gene in the present report. In addition, cDNA was prepared from three independently treated cultures per condition and array. The average standard deviation of the fold changes for all radA1-probes within an array was 0.85, while the array-to-array difference for identical probes in replicated experiments was 18 %. The average p-value of log ratios was < 10^-22. Results from previous transcriptome profiling after environmental perturbations indicate that a 7-fold expression change is high in comparison to any gene in Halobacterium sp. NRC-1 [21; JM and SD, unpublished data]. In those experiments (5 different conditions not involving UV irradiation) the radA1 gene was not differentially expressed demonstrating that the induction presented here was caused by UV irradiation.

The radA2 (radB) gene, a second homolog of recA in archaea, encoding a protein with unknown role in homologous recombination, was not up-regulated, in agreement with observations in other archaea 23. Accompanying radA1 induction, several other genes involved in homologous recombination were significantly induced after UV irradiation. The dbp gene, encoding a eukaryote-like DNA binding protein of the superfamily I DNA and RNA helicases, was upregulated 1.95 (p-value 0.04) at 1 hr post UV-irradiation with 30 J/m². The arj1 gene, encoding a recJ-like exonuclease was up-regulated 1.5 fold (p-value 0.008) at 3 hrs (30 J/m²). Additionally, an apparent operon encoding RPA ssDNA binding protein complex, rfa3, (VNG2160, RPA41 homolog, Zn-finger containing), rfa8 (VNG2162, RPA32 homolog), and an uncharacterized linked ORF (VNG2163) was induced 1.53 ± 0.02 fold (p-values < 0.001) at both time points after irradiation with 70 J/m². Similar fold changes were measured after irradiation with 30 J/m², however, p-values were around 0.1. In eukaryotes, RPA-ssDNA complexes are formed during almost all DNA-damage repair pathways. For example, RPA-proteins are recruited to Rad51 foci, protein complexes that accumulate at sites of DNA damage and stalled replication forks 2425. None of the other four RPA homologs of Halobacterium sp. NRC-1 (VNG0133, 1253, 1255, 6403) was significantly differentially expressed. As for radA1, neither of the above genes were found to be differentially expressed in other environmental perturbation experiments [21; JM and SD, unpublished data].

Our microarray data, together with these findings, strongly suggest a key role for homologous recombination in survival of UV damage in this class of archaea. By analogy with other organisms, there are at least two ways that recombination might contribute to survival of UV damage in Halobacterium sp. NRC-1. (i) Recombinational rescue of stalled replication forks may take place 1526. Rescue of stalled forks is likely to be error-free and this would fit with the low mutation rate in Halobacterium spp. [127; SM, unpublished data]. (ii) Recombinational repair may occur by using duplicate copies of the genome, a mechanism that has been suggested to facilitate post-irradiation survival of D. radiodurans 28. Halobacterium sp. NRC-1 is thought to contain multiple copies of its genome (J. Soppa pers. com.), which may greatly enhance recombinational repair.

In addition to genes involved in homologous recombination, several other genes were also highly induced. Most interestingly, the gene encoding a cobalamin-dependent class II RNR (nrdJ, formerly nrdB2) was strongly up-regulated (Figure 3, Table 1). A small open reading frame (VNG1642) located immediately downstream to nrdJ encoding a small uncharacterized zinc-finger containing protein (COG1645) was also strongly induced. RNR is the rate-limiting enzyme in the de novo synthesis of deoxyribonucleotide triphosphates which are utilized in both DNA synthesis and DNA repair; the up-regulation of the nrd genes may reflect the need for increased deoxynucleotide concentrations to allow rapid and accurate excision repair. In yeast, DNA damage elicits an increase in dNTP levels; and increased dNTP pools improve cell survival post DNA damage 29. In many other organisms, RNR genes are up-regulated by UV and were among the very first UV-inducible genes identified in an early study of UV-inducible promoters in budding yeast 30. RNR is regulated during the cell cycle in eukaryotes and the UV response depends on binding of transcription factor E2F to sites in the promoter 31. It is noteworthy that, in E. coli, nrd genes are UV-inducible even in the absence of lexA 1.

In addition to nrdJ, arcBCA, required for fermentation of arginine 32 were also particularly strongly induced, though only at the early time, and with some variation in magnitude of induction between repeat experiments. Whether this induction reflects a demand for rapid supply of ATP during periods of DNA-damage repair is not known. Another possibility is that the expression of these genes is exquisitely responsive to small stresses and we view these results with some caution at this stage.

A previous transcriptome analysis of UV-irradiated Halobacterium sp. strain NRC-1 cultures 17 showed, as we do, the lack of an SOS-like response and the lack of up-regulation of any of the NER genes by UV. However, these authors did not report strong induction of recombination genes or of RNR genes. The UV dose used in the latter study was substantially higher (200 J/m²). This dose, which induces, approximately one photoproduct per 600 bp of DNA, causes about one hundred times as much DNA damage as induced by natural sunlight and resulted in compromised cell survival. It may be that the high UV dose caused interference with gene expression. This is in contrast to approximately one photoproduct per 4 kb at 30 J/m² and one per 1.7 kb at 70 J/m² 333435, doses tolerated by Halobacterium sp. NRC-1 and used in the current study.

There have been questions raised as to the significance of gene expression responses to UV and other DNA damaging agents. In the yeast, S. cerevisiae, an analysis of deletion mutants lacking UV-inducible genes suggested that most do not contribute to survival of UV irradiation 36. In addition, most yeast genes identified as having a role in surviving UV damage, by isolation of UV-sensitive mutants, including most NER genes, are not UV-inducible. In human cells too, most repair genes are not UV-inducible 37. Thus, transcriptome analysis must be interpreted with caution and inferences about the extent of gene involvement should be supported by physiological and functional studies. In the present report, the absolute contribution of the different DNA-damage repair systems has not been investigated. However, the high level of up-regulation of radA1 and induction of other recombination genes while comparable few other gene expression changes occur, clearly suggests a significant role of homologous recombination in DNA-damage repair in Halobacterium sp. NRC-1.

Our transcriptome profiling work, together with our studies of the physiological response of a model archaeon, has shown that genes involved in homologous recombination are induced by UV irradiation in relatively low doses. Our results are consistent with homologous recombination playing an important role in the cellular response to UV damage in Halobacterium sp. NRC-1, either to permit rescue of stalled replication forks or to facilitate recombinational repair. In either case, we find that induction of recombination genes is prominent in the response of Halobacterium sp. NRC-1 to UV irradiation, which is particularly significant as it has been shown recently that recombination is important in facilitating genetic exchange in wild populations of halophilic archaea 38. Our results suggest that homologous recombination is stimulated by sunlight in this model archaeon.

The authors declare that they have no competing interests.

SM designed the experiments, analyzed the data, and drafted the manuscript. JM assisted with experimental design, conducted the DNA microarray hybridization, and carried out bioinformatics and data analysis. IB assisted with experimental design and conducted the UV irradiation, RNA preparation, and cDNA labelling. BB assisted with bioinformatics analysis. WN assisted with UV irradiation, RNA preparation, and cDNA labelling. SD designed the experiments, assisted with data interpretation, and prepared the manuscript.