Transcriptional responses to biologically relevant doses of UV-B radiation in the model archaeon, Halobacterium sp. NRC-1

Background Most studies of the transcriptional response to UV radiation in living cells have used UV doses that are much higher than those encountered in the natural environment, and most focus on short-wave UV (UV-C) at 254 nm, a wavelength that never reaches the Earth's surface. We have studied the transcriptional response of the sunlight-tolerant model archaeon, Halobacterium sp. NRC-1, to low doses of mid-wave UV (UV-B) to assess its response to UV radiation that is likely to be more biologically relevant. Results Halobacterium NRC-1 cells were irradiated with UV-B at doses equivalent to 30 J/m2 and 5 J/m2 of UV-C. Transcriptional profiling showed that only 11 genes were up-regulated 1.5-fold or more by both UV-B doses. The most strongly up-regulated gene was radA1 (vng2473), the archaeal homologue of RAD51/recA recombinase. The others included arj1 (vng779) (recJ-like exonuclease), top6A (vng884) and top6B (vng885) (coding for Topoisomerase VI subunits), and nrdJ (vng1644) (which encodes a subunit of ribonucleotide reductase). We have found that four of the consistently UV-B up-regulated genes, radA1 (vng2473), vng17, top6B (vng885) and vng280, share a common 11-base pair motif in their promoter region, TTTCACTTTCA. Similar sequences were found in radA promoters in other halophilic archaea, as well as in the radA promoter of Methanospirillum hungatei. We analysed the transcriptional response of a repair-deficient ΔuvrA (vng2636) ΔuvrC (vng2381) double-deletion mutant and found common themes between it and the response in repair proficient cells. Conclusion Our results show a core set of genes is consistently up-regulated after exposure to UV-B light at low, biologically relevant doses. Eleven genes were up-regulated, in wild-type cells, after two UV-B doses (comparable to UV-C doses of 30 J/m2 and 5 J/m2), and only four genes were up-regulated by all doses of UV-B and UV-C that we have used in this work and previously. These results suggest that high doses of UV-C radiation do not necessarily provide a good model for the natural response to environmental UV. We have found an 11-base pair motif upstream of the TATA box in four of the UV-B up-regulated genes and suggest that this motif is the binding site for a transcriptional regulator involved in their response to UV damage in this model archaeon.


Background
Solar radiation encompasses UV wavelengths ranging from <290 nm (UV-C, which is filtered out by ozone in the stratosphere), 290-320 nm (UV-B) and 320-400 nm (UV-A) as well as non-UV radiation, including visible (400-700 nm) and infrared (>700 nm). It has been well established that UV-B and UV-C cause mutagenic and cytotoxic damage to cells resulting from the induction of photoproducts in DNA, principally cyclobutane pyrimidine dimers (CPDs), 6-4 photoproducts (6-4 pps), and their Dewar isomer. More recently, it has been shown that CPDs are also the predominant DNA lesions caused by UV-A [1][2][3][4]. Most laboratory studies of the responses of living cells to UV have used high UV doses and mainly energy emitted from germicidal lamps at 254 nm (UV-C). However, these studies reflect neither biologically relevant doses nor wavelengths, because UV-C never reaches the Earth's surface and because the doses of UV in natural sunlight are low in comparison to the doses commonly used in the laboratory.
Most organisms have developed multiple strategies for surviving UV radiation. These can include protection from damaging wavelengths, cell cycle arrest, and activation of various pathways for repair of UV-damaged DNA. Tolerance mechanisms, such as recombination and lesion bypass, which allow cells to survive when lesions remain unrepaired in the DNA are also critical for survival [5]. In consequence of this variety of responses, even organisms exposed to high levels of sunlight in their natural environment show considerable variation in their UV sensitivities [1,6,7]. Among these are the highly radiation-resistant halophilic archaea, such as Halobacterium species NRC-1, which are exposed to intense solar radiation in their natural hypersaline environments.
The sequenced model archaeon, Halobacterium sp. NRC-1, is highly resistant to the damaging effects of UV light. One reason for this is highly efficient photoreactivation of DNA damage [8,9] but, even in the absence of photoreactivation, Halobacterium is significantly more UV-tolerant than Escherichia coli or Saccharomyces cerevisiae [10,11]. It is not yet clear why this is so. When the genome sequence of Halobacterium sp. NRC-1 became available, it seemed that a likely explanation was the existence of multiple repair systems because the genome contains homologues of both eukaryotic and bacterial nucleotide excision repair (NER) genes [12]. However a functional analysis of key repair genes has shown Halobacterium sp. NRC-1 depends primarily and possibly solely on a bacterial-type NER, involving UvrA, UvrB and UvrC proteins for repair of CPDs and 6-4pps [10]. There have been suggestions that carotenoid pigments may also play a part in protection of Halobacterium from UV radiation [13,14]. Two possible roles for carotenoids are in protecting DNA by directly absorbing UV and thus preventing formation of photoproducts, or in providing energy for excision repair. It has been shown that mutants lacking carotenoid pigments are more sensitive to UV irradiation than wild-type cells [14] and there is evidence for protection of DNA by bacterioruberin in vitro [13]. Another interesting observation is the very low occurrence of dipyrimidines in the genome of Halobacterium sp. NRC-1 which should result in fewer photoproducts [15,16]. However, a comparison of photoproducts in DNA from UV-irradiated Halobacterium and yeast cells has not shown any detectable difference in numbers or types of photoproducts induced, suggesting carotenoid protection and dipyrimidine number are not major factors [17]. Another possible contributor to the high UV tolerance may be the existence in Halobacterium cells of multiple copies of the genome, with 15 to 25 copies of the 2-Mbp chromosome per cell [18]. However, although high copy number and its accompanying genetic redundancy might be expected to increase a cell's chances of surviving DNA damage, the relationship between UV resistance and ploidy is not clear-cut at these high copy numbers [19,20]. In Deinococcus radiodurans an important factor seems to be that the recombination/repair protein, RecA, plays a critical role in UV tolerance [21], and this may also be the case in Halobacterium.
Many of the cellular responses to UV irradiation are constitutive but in all organisms studied to date there are also inducible responses. These have been investigated by a number of groups using whole genome transcriptome profiling. The best studied example of transcriptional regulation in microorganisms is the SOS response in bacteria such as E. coli, which involves LexA-dependent up-regulation of about 40 genes, including excision repair genes [22]. In addition, a number of genes, including nrdA, and nrdB (coding for ribonucleotide reductase subunits) are up-regulated independently of LexA, though mostly not more than 2-fold [22]. Depending on the eukaryote, a variety of genes are up-and down-regulated in response to UV-damage, but no eukaryotic equivalent of the bacterial SOS response has been identified [23].
There have been several studies of transcriptional responses to UV in the archaea [11,[24][25][26][27][28]. Although these studies have used different experimental regimes, there are certain common observations, including the absence of a coordinated SOS-like response. A study by Salerno et al. [24] suggested that, in Sulfolobus solfataricus, the homologues of human repair genes XPF, XPG and XPB (homologues of Saccharomyces cerevisiae RAD1, RAD2 and RAD25 respectively) were UV-inducible. However, this was not confirmed by more recent analyses [25,28] and transcriptome analysis in Halobacterium, has not shown excision repair genes to be up-regulated by UV [11,26].
Most laboratory studies of UV damage have used shortwave UV because low-pressure mercury vapour germicidal lamps, which emit at 254 nm, are readily available and they produce essentially the same type of DNA damage as UV-B, the most damaging wavelengths in sunlight at the Earth's surface. They focus on short-wave UV (UV-C) at 254 nm, which is blocked by oxygen and ozone in the stratosphere and therefore is a wavelength that never actually reaches the Earth's surface. Most studies of transcriptional responses to UV radiation have also used UV doses that are very much higher than those encountered in the natural environment. Two archaeal studies used UV-C doses of 200 J/m 2 , a more recent one used 75 J/m 2 , and in our own previous study we used 30 J/m 2 and 70 J/m 2 . High UV doses have traditionally been used for studies of repair of photoproducts because the assays for measuring DNA damage are rarely sensitive enough to allow the use of lower doses. However, for transcriptional studies, there is little justification for using doses that are many-fold higher than organisms are ever exposed to under sunlight. Table 1 shows the amount of damage produced by various doses of UV used in transcriptional studies and shows that a dose of 200 J/m 2 (administered over a period of only 1 minute) produces more DNA damage than 12 hours of sunlight [29,30]. It is well known that initiation of DNA replication and transcription are inhibited by UV in a dose-dependent manner, so we believe that high doses are likely to produce artefacts, making it important to use more biologically relevant doses. In order to approach biologically relevant radiation conditions for our transcriptional analysis, we used a broad-band UV-B lamp and low doses of UV, producing equivalent damage (in terms of CPDs) to 5 J/m 2 and 30 J/m 2 of UV-C. The lower UV-B dose used in this study, equivalent to 5 J/m 2 UV-C, induces the same amount of damage (in terms of CPDs) in 30 seconds as about 20-30 minutes of sunlight (see Table 1).

Results
We irradiated wild-type Halobacterium sp. NRC-1 with a dose of UV-B that induces the same number of CPDs per kb DNA as 5 and 30 J/m 2 UV-C (we will refer to these regimes as *5 J/m 2 and *30 J/m 2 ) and irradiated a ΔuvrA ΔuvrC double deletion mutant, which lacks the capacity for nucleotide excision repair, with a dose of *5 J/m 2 . We have compared the transcriptional response to these UV-B doses and to the response to irradiation with 30 J/m 2 UV-C, which we reported previously [26].

The transcriptional response to a UV-B dose equivalent to 30 J/m 2 UV-C
After a UV-B dose of *30 J/m 2 , 103 genes were significantly up-regulated (1.5-fold or above, p-value < 0.001) at 1 hour and/or 3 hours after irradiation. The most strongly up-regulated genes included radA1 (vng2473) (gene for RecA/Rad51 recombination protein), nrdJ (vng1644) (ribonucleotide reductase α subunit), vng1642 (a conserved hypothetical halophile ORF adjacent to nrdJ), arcA (vng6317), arcB (vng6315) and arcC (vng6316) (all of which are required for fermentation of arginine), dbp (vng2167) (coding for a eukaryote-like DNA binding protein of the superfamily I DNA and RNA helicases) and vng17 and vng261, small ORFs unique to Halobacterium sp. NRC-1 and with unknown functions.
We compared the results of this experiment to our previously published 30 J/m 2 UV-C data and found that, of the 103 genes identified as up-regulated, 29 were also up-regulated in the 30 J/m 2 UV-C arrays ( Figure 1 and Table 2).

Genes up-regulated in wild-type cells after *5 J/m 2 UV-B
At the lower UV-B dose, only 41 genes were significantly up-regulated in the wild-type strain. Of these, 11 were also up-regulated at *30 J/m 2 UV-B ( Figure 2 and Table 3). These are the genes whose transcriptional control is most likely to be significant for the response to biologically relevant UV doses.  [30] Diagram showing overlap between genes up-regulated 1.

Genes up-regulated after *5 J/m 2 UV-B in a repair-deficient mutant
In addition to analysing the transcriptional response to UV in wild-type Halobacterium sp. NRC-1 cells, we measured the response to *5 J/m 2 UV-B in a ΔuvrA ΔuvrC knockout strain which lacks the capacity for nucleotide excision repair [10] so that we could examine responses in the absence of repair (and, presumably, the persistence of DNA damage). NRC-1 cells are able to remove UV damage by excision repair relatively rapidly [17] and most photoproducts are repaired within 3 hours after irradiation. So we anticipated that, if the response was related to amount of damage in DNA, the transcriptional response to UV in a repair-deficient mutant might resemble the response to a higher dose in wild-type cells. However, we found that the response to a dose of *5 J/m 2 UV-B was very similar in both the wild-type and repair-deficient mutant. The total number of genes up-regulated was very similar, 41 and 47 respectively, and there was considerable overlap, with 13 genes up-regulated in common ( Figure 2). The fold changes were also similar to the wild-type at the same dose and lower than the fold changes seen after the higher dose, with the possible exception of arcC (see Table 3). This suggests that the nature of the transcriptional response does not simply depend on the number of DNA photoproducts present in the DNA. Table 4 shows the fold-changes for selected transcripts in the five experiments we have carried out, irradiating wildtype and mutant cells with various doses of UV-C and UV-B [31]. The table highlights the fact that some genes, including radA1 (vng2473), nrdJ (vng1644), vng1642, arj1 (vng779) and trxA2 (vng2600) were up-regulated by all or most UV-irradiation regimes. Other genes, notably hjr (vng2252), vng261, vng1800, rfa3 (vng2160), and the arcABC genes, were up-regulated only by higher doses or only by short-wave UV. Most interestingly several genesnpa (vng6361), vng17, vng6359 (which is similar to vng17 and is located directly upstream of npa), top6A (vng884) and top6B (vng885) were significantly up-regulated only by lower doses or by UV-B. Only four genes were up-regulated 1.5-fold or more in response to all of the doses of UV-C and UV-B we have used, at at least one time point. These are radA1 (vng2473), arj1 (vng779), nrdJ (vng1644) and vng1642.

Confirmation of up-regulation with quantitative real time PCR
Six genes, including radA1 (vng2473), were selected for confirmation of the up-regulation noted from microarray data using qRT-PCR ( Figure 3). The results agree well with the microarray data, for all doses and all wavelengths, and confirm that these genes are indeed up-regulated by UV in most cases. In a few cases the RT-PCR results do not agree quantitatively with the microarray data; in these instances, qRT-PCR showed somewhat greater up-regulation than was evident from the microarray data. The most dramatically up-regulated gene, radA1, is up-regulated 9.7-fold, three hours after 30 J/m 2 UV-C, 7.6-fold after an equivalent dose of UV-B and over 4-fold after the much lower UV-B dose (*5 J/m 2 ).

A motif common to the promoter regions of five UV-B upregulated genes
Since radA1 (vng2473) was consistently the most highly up-regulated gene in all our experiments, we examined its promoter region and noticed a striking sequence motif, TTTCACTTTCA, with an internal 5 bp repeat (TTTCA), located about 50 bases upstream of the start codon. A findpatterns search of the Halobacterium sp. NRC-1 genome revealed seven matches of this 11-base sequence. Four were in UV-B up-regulated genes (radA1, vng280, top6B, and vng17) and one was on a non-coding strand. Alignments of the promoters of these genes are shown in Figure 4A. A proviso is that the alignment in the figure uses the second ATG in the vng280 ORF as the translational start codon rather than the first predicted using Glimmer in the genome sequence [12]. Interestingly, a near-match (TTTTACTTTCA) to the 11-base pair motif is found 52-62 bases upstream of the start codon of npa, a putative transposase gene, which is also up-regulated after UV-B irradiation. A similar motif is found located in the upstream regions of radA genes in other halophilic archaea and, interestingly, Methanospirillum ( Figure 4B). It is not found in any of the radA2 (vng1665) promoter regions examined (not shown).

Discussion
Previous genomic transcriptional analyses in the archaea have shown large numbers of genes to be up-regulated after irradiation with high doses of UV-C and experiments by different groups have shown considerable differences in the genes identified [11,25,26]. The use of low doses of UV-B has enabled us to focus on a smaller set of genes, whose transcriptional response is more likely to be biologically and environmentally significant than the genes identified previously. These low-dose experiments have confirmed the upregulation of radA1 (vng2473) previously identified in high dose UV-C experiments and revealed the up-regulation of several genes that were not, including top6B (vng885), vng17 and npa (vng6361). We have shown that top6B and vng17, as well as vng280, all share a common motif with radA1 (vng2473) in the promoter region which seems very likely to be involved in transcriptional regulation in response to DNA damage. A nearly identical motif is also present upstream of npa. The proteins encoded by these genes may have related functions in the cell's response to UV radiation. RadA1 is likely to play a major role in resolving stalled replication forks and/or promoting repair [32][33][34] and it is likely to be required in large amounts because it coats single-stranded DNA to form nucleoprotein filaments [35], hence the greatest fold-induction observed after UV radiation. top6A (vng884) and top6B (vng885) code for DNA topoisomerase VI subunits A and B. The little-studied archaeal topoisomerase VI enzymes are members of the topoisomerase IIB family and have been shown to be important in both Sulfolobus and halophilic archaea [36]. They have ATPdependent nicking-closing activity as well as ability to generate double-strand breaks and they are able to release positive supercoils that are formed ahead of replication forks and during transcription [37,38]. It is likely that they are involved in processing stalled forks in UV-damaged DNA in Halobacterium. arj1 (vng779), which is up-regulated by all UV-B doses examined, encodes a RecJR-like protein, so, by analogy to E. coli RecJR it, too, is likely to be involved in recovery of DNA replication at stalled forks, possibly by making DNA lesions at stalled forks accessible for repair [39]. We do not know the functions of the vng17 and vng280 gene products. If these genes are indeed up-regulated because of their role in recovery of DNA replication, we speculate that the reason why they are not significantly up-regulated after high UV doses is that high doses may largely halt initiation and/or elongation of DNA replication [40,41]. Therefore, after high doses of UV irradiation, there are fewer replication forks that become blocked. However, the precise roles for these genes must await further experimentation, including genetic knockouts and perturbations.
After the higher dose of UV-B, we observed up-regulation of arcA, arcB and arcC, though only at the earlier time after irradiation (Table 5); this is similar to the response we saw after UV-C irradiation at 30 J/m 2 and 70 J/m 2 [26]. We do not see these genes up-regulated after low UV-B doses, except for slight up-regulation of arcC in the repair-deficient mutant and we do not know the significance of this response. We suggested in an earlier report that up-regulation of these genes may reflect a demand for rapid supply of ATP during periods of DNA-damage repair [26] or it may be a more general stress response.
The level of up-regulation of radA1 that we see in Halobacterium sp. NRC1 is similar to that reported for the archaeal mesophiles, Methanococcus maripaludis and Methanococcus voltae. Reich et al. [27], using Northern blot analysis of transcripts and Western blots to study RadA protein levels, found that radA transcription was up-regulated, and RadA protein levels increased, in the four archaea studied. The up-regulation was greater (about 6-fold after a UV dose of 50 J/m 2 ) in the mesophiles, Methanococcus maripaludis and Methanococcus voltae, than in the thermophiles (about 2fold), Sulfolobus solfataricus and Methanococcus jannaschii. A recent transcriptomic study using microarrays after a range of UV doses did not show significant up-regulation of radA in Sulfolobus solfataricus [28], possibly reflecting the low level of the response or, perhaps, the use of different growth conditions.
It is seems likely that the 11-bp motif, TTTCACTTTCA, that we have identified upstream of the start codon is involved in regulation of the genes that share it -radA1 (vng2473), vng17, vng 280, and top6B (vng885) -and it may be the binding site for a transcriptional regulator. It Histograms showing the fold changes in transcripts from microarray data (blue) and confirmation by qRT-PCR (maroon) of six selected genes is interesting that three of the genes that have this motif were not originally identified in our high-dose UV-C experiments but that they were all up-regulated after UV-B exposure. Neither top6B, vng280, nor vng17 is up-regulated to as high a level as radA1 ( Figure 3 and Table 4). However we are currently carrying out a detailed study of the radA1 promoter region and have found that the radA1 upstream region contains an additional putative regula-tory sequence that is not present in the other three genes (unpublished).
Interestingly, in Sulfolobus solfataricus, a crenarchaeon, SSO0777, which is a paralogue of the radA gene, is regulated in response to DNA damage, by the activator Sta1, which binds within the sequence ATTTTTTATTTTCACAT-GTAAGATGTTTATT [42]. There is no obvious homologue   Table  1) [11]. A similar observation has been made in Schizosaccharomyces pombe, where transcription of rhp51, the radA homologue, was up-regulated after low doses of UV-C but not after high doses (200 J/m 2 and above) [43] and it was suggested that extensive DNA damage and blocking of DNA replication prevented up-regulation. In Saccharomyces cerevisiae, too, high doses of UV have not been informative. Genes shown to play a role in survival of UV irradiation (with deletion mutants that were sensitive to UV) failed to correlate with genes that were transcriptionally up-regulated by a high dose of UV-C (200 J/m 2 ) [44], so studies of transcriptional response to high doses of UV-C could not identify genes involved in surviving UV irradiation.
One of the distinguishing features of the current study is that we used UV-B light, in contrast to short-wave UV-C commonly used in laboratory studies of UV damage. Whilst it is true that the photoproducts induced by UV-C, UV-B and sunlight are broadly similar and that they are all repaired by nucleotide excision repair, there are significant differences in the damage induced by different UV light sources. Perdiz et al. [1] measured the proportions of the three major types of photoproduct formed in DNA on exposure to different sources of UV light -a UV-C lamp emitting at 254 nm, a broad-band UV-B lamp and a solar simulator. They found that the proportions of cyclobutane pyrimidine dimers (CPDs) to 6-4 photoproducts (6-4 pps) to Dewars induced in DNA were 1.0:0.25:0 for the UV-C lamp, 1.0:0.12:0.014 for the broad-band UV-B lamp and 1.0:0.18:0.06 for the solar simulator [1]. These results showed that UV-B, though not identical in its effects to sunlight, is a closer model than UV-C because both sunlight and UV-B induce a significant number of Dewars as well as inducing relatively fewer 6-4pps. They also measured repair of the three types of photolesion and found that both CPDs and Dewars are repaired much more slowly than 6-4 pps [1].
We have compared the doses used in published microarray studies to the UV doses found in sunlight (Table 1).
These are, inevitably, approximations since the UV doses and wavelengths in sunlight vary with latitude, altitude, time of day and local conditions. The figures we have used are based on the maximum number of CPDs induced by sunlight during a whole day's exposure, measured by Wilhelm et al. using a DNA dosimeter, at equatorial latitudes off the coast of South America [29] and Visser et al., also using a DNA dosimeter, off the south coast of Curacao (12° 07' N) [30].
Finally, Halobacterium sp. NRC-1 has also been the subject of studies with ionizing radiation from both gamma and electron beam sources. In one study conducted by DeVeaux et al [45], two highly radiation resistant Halobacterium mutants were reported which, with a LD 50 of nearly 12 KGy, are even more resistant than Deinococcus radiodurans, previously the most radiation resistant organism known. The mutants upregulated the expression of rfa3 and two transcriptionally-linked downstream genes, which are also inducible after high UV-C exposure. The ability of Halobacterium to survive both ionising and nonionising radiation is a remarkable property of these species and suggests that more detailed investigations will provide a much better understanding of the DNA repair and replication systems operating in these model Archaea.

Culture conditions and UV-irradiation
Halobacterium sp. strain NRC-1 and the ΔuvrA ΔuvrC mutant, were grown in the dark, at 37°C, in an orbital shaker-incubator at 225 rpm, under aerobic conditions to early exponential growth phase (OD 600 0.19-0.23) in complete medium, CM [46]. 50-ml cultures were grown up in triplicate for each time point. For irradiation, cultures were transferred individually into pre-warmed plastic boxes and irradiated in the dark, in CM + medium with gentle agitation, using two unfiltered FS20 fluorescent tubes as the UV-B source. In order to compare the transcriptional profiles after UV-B irradiation with our previous studies, in which we irradiated with 30 and 70 J/m 2 UV-C, from a mercury vapour lamp emitting at 254 nm, we irradiated plasmid DNA and measured cyclobutane dimers (i.e. sites sensitive to nicking by micrococcal UVendonuclease [47]). The number of cyclobutane pyrimidine dimers induced in plasmid DNA by the UV-B lamp in 30 sec was shown to be equal to the number induced by 5 J/m 2 UV-C. An equivalent UV-B dose to 30 J/m 2 UV-C was administered by irradiating for 3 minutes. UV-B doses are referred to as 'damage-equivalent' doses. For post-UV incubation, cultures were returned to the original warmed flasks and incubation was continued at 37°C in the dark. We avoided changing the medium, so as to avoid any additional stress caused to the cells by harvesting and changing media.
tion, cDNA labelling and analysis of the data. IB assisted with experimental design and conducted the UV irradiation, RNA preparation, and cDNA labelling. DJC constructed and characterised the ΔuvrA ΔuvrC double deletion mutant. PD conducted the DNA microarray hybridization, data processing and analysis and assisted extensively with preparation of the figures, tables, and manuscript text. SD assisted with experimental design, data interpretation and finalising the manuscript. All authors read and approved the final manuscript.