Strains and Growth Conditions

The strains of Halobacterium used were derivatives of strain NRC-1 (Fig. 1). Deinococcus radiodurans R1 (ATCC 13939) was a gift from John Battista at Louisiana State University. Halobacterium cultures were grown in CM⁺ medium (24) at 42°C with shaking, without supplemental illumination. D. radiodurans was grown in TGY (0.5% tryptone, 0.3% yeast extract, 0.1% glucose) at 30°C with aeration (25). For solid medium, 2% agar was added. Halobacterium plates were incubated at 42°C for 7 days, then removed and allowed to continued to grow at room temperature until no new colonies appeared. D. radiodurans plates were incubated at 30°C for 3–5 days. Liquid cultures for irradiation were inoculated from purified isolates on solid medium. Stock cultures were maintained in glycerol at −80°C. For short-term use, purified cultures were maintained on stock plates at 4°C. Growth rates were monitored by measuring the optical density at 600 nm (OD₆₀₀) of cultures in liquid medium over time. Samples of growing cultures were diluted and plated at various times to ensure that vastly different cell sizes, the presence of nonviable cells, or differences in gas vesicle production did not skew the OD₆₀₀ readings. For determination of survival, irradiated liquid cultures were immediately diluted and spread on plates. Survival after irradiation was determined by comparing surviving cells to an unirradiated control.

Irradiation and Dosimetry

The 20 MeV pulsed electron LINAC delivered 18–20 MeV electrons at a peak current of 80 mA, with a peak dose rate of 2.5 × 10⁵ Gy/s and an average dose rate of 30 Gy/s. The pulse width was 2 ms, and the repetition rate was 60 Hz. Samples were irradiated at room temperature in 0.2-ml thin-walled PCR tubes held 2.5 m from the beam port. At this distance, the uniformity of the beam was within 10% of the peak dose over a circular area of 10 cm diameter, which encompassed the sample holder size. Samples were located on isocontours of dose with sample dose variations less than 1%. Beam location was determined before and after irradiations with PIN diodes, and the sample doses delivered were measured using a GEX Corporation thin-film dosimetry system and GEX B3 radiochromic film (26).

Irradiation–Growth Cycles

A stationary-phase culture (4 × 10⁸ cells/ml) of Halobacterium NRC-1 in 0.1-ml aliquots was exposed to electrons from a 20 MeV LINAC at doses of 16.4 ± 1.6 and 19.5 ± 2 kGy. These aliquots were diluted into 2 ml of CM⁺ medium and incubated with shaking for 2–5 days at 42°C until growth was visible. Glycerol was added to 15%, and 1 ml of each suspension was frozen at −80°C. Prior to the next irradiation, the frozen stocks were used to inoculate 2 ml of CM⁺ medium and grown for 2 days at 42°C. Aliquots of these cultures were irradiated at 23 ± 2.3 kGy. Each was then diluted into 2 ml broth and incubated until growth was visible. Once again, 0.1-ml aliquots of these cultures were irradiated at 20.6 ± 2.1 kGy. Fifty microliters of each irradiated sample was diluted into 2 ml CM⁺ medium and grown for 7 days, until growth was visible. These cultures were irradiated in 0.1-ml aliquots at 21 ± 2.1 kGy. Fifty microliters of each aliquot was diluted into 1 ml CM⁺ medium until growth was visible, approximately 7 days. These two cultures were streaked on solid medium and individual isolates were purified four times. LH5 was derived from the cultures originally irradiated at 16.4 kGy and LH7a from the culture irradiated at 19.5 kGy. This procedure is summarized in Fig. 1.

Cell Irradiation and Survival Measurement

Each strain was grown in liquid culture to the appropriate density and culture state (exponential or stationary phase). In an individual experiment, a culture was divided into aliquots and subjected to a predetermined set of doses, including no radiation. The aliquots were diluted in growth medium immediately after irradiation and plated in duplicate or triplicate. In a given experiment, the multiple platings of the unirradiated (0 Gy) aliquot were counted and averaged to give the initial cell density in colony-forming units per milliliter (cfu/ml). This value represented 100% survival and was used as the basis of comparison for all irradiated aliquots of that culture. The duplicate or triplicate platings of each irradiated aliquot were averaged to determine cfu/ml at that dose. The surviving fraction for that dose was determined by dividing the cfu/ml by the cfu/ ml representing 100% survival (0 Gy, or initial cell density) of that culture. For each strain, multiple experiments (independent cultures) were performed on at least two different days, and the surviving fractions were combined (Figs. 2 and 3).

Survival Curve Fitting

For each strain, surviving fractions from multiple irradiations were combined and plotted in OriginPro 7 (OriginLab Corporation, Northampton, MA). The resulting data were fitted to a sigmoidal Boltzmann function,

Microarray Analysis

Halobacterium sp. strain NRC-1 and derivatives LH5 and LH7a were grown under aerobic conditions to early exponential growth phase (OD₆₀₀ of 0.15 to 0.3). Cells were harvested by chilling the incubation vessels in an ethanol-dry ice bath for 1 min followed by centrifugation of the culture (8,000g, 5 min, 2°C) in chilled beakers. Total RNA was isolated immediately after harvest at 25°C using an Agilent Total RNA isolation kit (Agilent Technologies, Palo Alto, CA), and DNA was hydrolyzed using amplification grade DNase (Sigma-Aldrich, St. Louis, MO). Lysis of cells was carried out directly in lysis buffer. Two treatments with DNase (Invitrogen, Carlsbad, CA) were carried out to remove contaminating DNA from the RNA preparations. To minimize biological noise, RNA preparations from three 50-ml cultures of each strain grown under identical conditions were pooled to equal parts for cDNA synthesis. cDNA was prepared from 7 μg total RNA with Super Script III reverse transcriptase (Invitrogen) and Cy3- or Cy5-dCTP (Amersham Biosciences, Piscataway, NJ), and purified after alkaline hydrolysis of RNA on Qiagen mini-elute columns (Qiagen, Valencia, CA). The labeled cDNA targets were mixed with hybridization buffer and control targets and hybridized to microarray slides assembled into a hybridization chamber (Agilent) for 17 h at 60°C in the dark. After hybridization, the slides were washed as described in the Agilent protocol and scanned for the Cy3 and Cy5 fluorescent signals with an Agilent DNA-microarray scanner (Model no. G2565BA). Transcriptome profiling of cells was carried out in duplicate. Relative mRNA levels were determined by parallel two-color hybridization to oligonucleotide (60-mer) microarrays representing 2474 open reading frames (ORFs) representing 97% of unique Halobacterium sp. NRC-1 ORFs according to Müller and DasSarma (10). Previous results showed that differences in the relative intensity of the channels could be adjusted for by intensity-dependent LOWESS normalization (10, 27). Image processing and statistical analysis were carried out using Feature Extraction Software Version 7.1 (Agilent). Log ratios for each feature were calculated and the significance of the log ratio was assessed by calculating the most conservative log ratio error and significance value (P value) using a standard error propagation algorithm (Agilent) and a universal error model (Rosetta Biosoftware, Seattle, WA).

Reverse-Transcriptase PCR (RT-PCR)

Total RNA was isolated as described above. DNA was removed from the RNA by three treatments with RNase-free DNase I (Qiagen). cDNA was synthesized from 0.2 to 0.8 μg RNA and 2 pmol of specific primer with SuperScript II RNase H⁻ Reverse Transcriptase (Invitrogen) as described by the supplier. The PCR amplification mixtures contained 6 μl cDNA, 200 nM of each primer, 200 μM dNTP, 1 U Taq polymerase in PCR buffer with 1.5 mM MgCl₂. For amplification of parts of rfa3-rfa8 and rfa8-ral, and to map the transcriptional start site of rfa3, the following primer pairs were used (positions in the NRC-1 chromosome of primers in parentheses): −29F (1594051–1594069)/−29R (1594539–1594556); +3F (1594082–1594103)/+3R (1594603–1594622); rfa3–rfa8F (1594829–1594848)/rfa3–rfa8R (1595225–1595244); rfa8–ralF (1595510–1595534)/rfa8–ralR (1595754–1595777). PCR parameters were as follows: 3 min at 92°C, 30 cycles of 0.45 min at 94°C, 1 min at 50°C for the primer pair −29F/−29R, 1 min at 58°C for +3F/+3R, and 1 min at 60°C for the other two pairs, 1 min at 72°C, followed by 7 min at 72°C.

Isolation of Extremely Radiation-Resistant Mutants of Halobacterium sp. NRC-1

Using a series of irradiation–growth enhancements based on a procedure used by Davies and Sinskey on Salmonella typhimurium (22), we enriched a culture of Halobacterium sp. NRC-1 for individuals with increased tolerance to high doses of electron-beam radiation. The two isolates chosen for further study, LH5 and LH7a, were derived from independently irradiated and derived cultures (Fig. 1). These strains were also phenotypically distinct. Strain LH5 had lost the gas vesicle production capability of the parent NRC-1, due to an IS element insertion into the gvpJ gene (data not shown) rendering colonies translucent, whereas strain LH7a retained the gas vesicle-positive phenotype (opaque colonies) of the parent strain, NRC-1.

Characterization of Radiation Resistance in Halobacterium sp. NRC-1 and Radiation-Resistant Mutants LH5 and LH7a

To determine the extent of the radiation resistance acquired by the two mutants, exponential-phase cultures (OD₆₀₀ < 0.2) of these strains and the parent strain NRC-1 were irradiated with 18–20 MeV electrons from a standard medical LINAC over a range of doses. For comparison, we also measured the survival of D. radiodurans R1, one of the most radiation-resistant organisms known (28). Exponentially growing Halobacterium wild-type parent strain, NRC-1, had an LD₅₀ of 5.4 kGy (±0.2). The LD₅₀ of D. radiodurans, irradiated under the same conditions, was 7.9 kGy (±0.2). In contrast, the LD₅₀ for the Halobacterium isolate LH5 was 11.9 kGy (±0.4) and that of LH7a was 12.1 kGy (±0.3). Moreover, survivors of LH5 and LH7a were reproducibly detected at doses as high as 25 kGy, where there were few or no survivors of D. radiodurans. The graph showing surviving fractions and fitting of the data for the exponentially growing cultures is presented in Fig. 2.

Since microorganisms generally display greater resistance in stationary phase (29–31), each strain was also tested in stationary phase (Fig. 3). We confirmed the observation (20) that Halobacterium sp. NRC-1, unlike most other microbes, shows greater resistance when actively growing and tested the response of LH5 and LH7a at this growth density. As expected, the LD₅₀ of NRC-1 in the stationary phase was significantly less (2.9 kGy ± 0.2) than the value for the exponential-phase cultures (5.4 kGy). Both radiation-resistant isolates also showed decreased resistance when in the stationary phase. The LD₅₀ of LH5 decreased to 6.8 kGy (±0.3) and that of LH7a decreased to 8.7 kGy (±0.2).

One simple explanation for the increased radiation resistance of strains LH5 and LH7a would be reduced growth rate. Reduction in growth rate has been shown to increase survival of D. radiodurans after irradiation (29, 30). Given the radiation doses received by these Halobacterium cells, it is highly likely that each individual contains numerous mutations, some of which may affect the growth rate. To rule out the latter possibility, we verified that the doubling times of LH5 and LH7a were the same as that of the parent strain NRC-1 by measuring the OD₆₀₀ of cultures grown under standard conditions (data not shown).

Transcriptional Analysis of Strains LH5 and LH7a

In principle, the molecular basis for increased radiation resistance of strains LH5 and LH7a may be attributed to one or more mutations in the genome. These mutations may result in increased activity of proteins involved in protection against and/or repair of radiation-induced damage. Alternatively, regulatory elements that govern expression of genes involved in protection or repair may be affected. To test for the latter scenario on a genome-wide scale, we used previously described in situ synthesized oligonucleotide arrays (10) to compare the transcriptomes of the mutant strains LH5 and LH7a to that of the wild-type strain NRC-1 under our standard conditions (see the Materials and Methods). In strain LH5, 47 and 72 genes out of 2474 ORFs (97% genome coverage) displayed more than a 1.5-fold (P < 0.01) increase or reduction, respectively, in transcript level compared to the wild type (Fig. 4; Table 2 in Supplementary Information). Most interestingly, genes involved in homologous recombination, e.g., radA1 (archaeal homolog of rad51/recA) and rfa3, rfa8, ral (single-stranded DNA-binding protein complex), and protection against reactive oxygen species, e.g., sod1 and sod2 (Mn-dependent superoxide dismutases), were induced in LH5. As expected from the gas vesicle-deficient phenotype of LH5, the gvp (gas vesicle protein) genes (7) had much lower expression levels in that strain. In LH7a, 11 and 36 genes had transcript levels that were significantly increased or decreased, respectively (Fig. 4; Table 2 in Supplementary Information). Among those, rfa3, rfa8 and ral were the only genes that had induced transcript levels in both LH5 and LH7a (2.1- to 3.7-fold and 3.4- to 5.5-fold, respectively; Table 2 in Supplementary Information).

Transcription and Promoter Analysis of the rfa3 Operon

To gain further insight into the transcriptional organization and regulation of the rfa3, rfa8 and ral gene region, we used RT-PCR analysis first to determine whether rfa3 and rfa8 and rfa8 and ral are transcriptionally linked and likely form an operon. Primer pairs flanking the two intergenic regions produced amplification products (Fig. 5A), confirming that transcription proceeds from rfa3 through rfa8 to ral, resulting in a polycistronic transcript. To determine the transcriptional start site, two primer pairs were designed, with one set (−29F and R) having one end 29 bp upstream of the predicted transcription start point and the other within the coding region of the gene, while the second set (+3F and R) had one end 3 bp within the predicted transcription start point and the other within the coding region of the gene (see Fig. 5B for the upstream primer positions). The results of RT-PCR showed that the second set (+3F and R) produced a PCR product, whereas the first set did not (−29F and R) (Fig. 5A). These results are consistent with the location of the transcription start point at ∼40 bp upstream of the ATG start codon. Although this codon corresponds to the second ATG in the predicted rfa3 ORF from genome annotation (13), subsequent alignment of rfa3 orthologs (data not shown) in other archaea showed it to be consistent with the translation start. A classical TATA-box motif (TTTAAA) was found centered about 25 bp upstream of the putative transcriptional start site mapped (Fig. 5B).