Archaeal DNA Polymerase D but Not DNA Polymerase B Is Required for Genome Replication in Thermococcus kodakarensis

T. kodakarensis strains were grown in artificial seawater (ASW) with 5 g/liter each of yeast extract and tryptone (YT) and 2 g/liter sulfur (S⁰) at 85°C, with the growth of cultures measured by an increase in the optical density at 600 nm (OD₆₀₀), as previously described (33). Standard molecular biology techniques were used to construct plasmids pTR1, pTR2, pTR3, pTR4, pTS742, and pTS744, which were maintained and amplified in Escherichia coli. The sequences of the oligonucleotides employed are available upon request. T. kodakarensis strains TR1, TR2, TR3, TR4, TS742, and TS744 were constructed by transformation of T. kodakarensis KW128 (35) with transformants selected by growth in the absence of tryptophan, as previously described (36). Construction of T. kodakarensis TS742 introduced an additional DdeI cleavage site; DdeI digestion of the appropriate amplicon from T. kodakarensis TS742 genomic DNA confirmed the presence of this site, and sequencing confirmed the presence of the nonsense mutation introduced at codon 3 of TK0001. Approximately 1.7 kbp was deleted from the 5′ terminus of TK0001 in T. kodakarensis TS744, and this was confirmed by the sizes of the amplicons generated using primers that flanked the locus and by failure to generate amplicons using primers that would have hybridized within the deleted region. The inability to generate amplicons from within TK0001 also confirmed that this coding region was not present elsewhere in the T. kodakarensis TS744 genome. Additional confirmation of the deletion in TK0001 in T. kodakarensis TS744 was provided by Southern blotting of PsiI-digested genomic DNA. The PsiI restriction fragments that hybridized to a digoxigenin (DIG)-labeled amplicon probe, generated by PCR from within TK0001, were identified by using anti-DIG antibodies coupled to alkaline phosphatase, as previously described (37).

Cells were concentrated from exponentially growing cultures of T. kodakarensis KW128, TS742, and TS744 by centrifugation; resuspended in a solution containing 20 mM Tris-HCl (pH 7.5), 10% (vol/vol) glycerol, and 100 mM NaCl; and lysed by repeated freezing and thawing. After centrifugation, the resulting clarified lysates were quantified via Bradford assays (38), and the polypeptides present in 50-μg protein aliquots were separated by electrophoresis through denaturing 10% polyacrylamide gels and electroblotted onto a nitrocellulose membrane (Protran BA83; Whatman Inc.). Western blots were generated by using a combination of guinea pig polyclonal antibodies (Cocalico Biologicals Inc.) raised against recombinant Pol B and the small subunit of Pol D and rabbit anti-guinea pig antibodies coupled to horseradish peroxidase (Sigma-Aldrich). The blots were developed by using enhanced chemiluminescence (ECL; GE Healthcare).

Aliquots (5 μl) of serial dilutions of cells sampled from growing cultures of T. kodakarensis KW128, TR1, TS742, and TS744 were plated, under anaerobic conditions, on ASW-S⁰ medium that contained all 20 amino acids (33) with or without 0.1% methylmethane sulfonate (MMS) or 100 μg mitomycin C/ml. The colonies present on each plate after 60 h of incubation at 85°C were counted. Similarly, aliquots of dilutions of T. kodakarensis KW128, TR1, TS742, and TS744 cultures were spotted onto ASW-S⁰ plates containing all 20 amino acids and were then exposed to UV irradiation (model UVH-46550; International Biotechnologies Inc., New Haven, CT) for 1 min before the plates were incubated anaerobically at 85°C for 60 h to allow colony formation.

Pol B is encoded by TK0001 (14, 34), a >5-kbp open reading frame that also encodes two inteins (Fig. 2B). By using standard molecular biology techniques, chromosomal regions flanking and containing TK0001 were PCR amplified from T. kodakarensis genomic DNA, cloned, sequenced, and replicated within plasmids in E. coli. Subcloning and site-specific mutagenesis generated plasmids pTR1 (Fig. 2A), pTR2, pTR3, pTR4, pTS742, and pTS744 (Fig. 2B), which were used as donor DNAs to transform T. kodakarensis KW128 (Table 1) with transformants selected by the expression of trpE (TK0254). Sequencing of genomic DNA amplified from the transformants studied further in detail, designated T. kodakarensis TR1, TR2, TR3, TR4, TS742, and TS744 (Table 1), confirmed that they differed only by the mutation(s) introduced into TK0001 and that there were no changes to the promoter or ribosome binding site of TK0001. In T. kodakarensis TR2 and TR3, the mutations in TK0001 result in D215A and V93Q substitutions, which inactivate 3′→5′ exonuclease proofreading (18–20) and deaminated base recognition activities of the encoded Pol B, respectively (21, 22). Both mutations were introduced into TK0001 in T. kodakarensis strain TR4, resulting in a Pol B variant that lacked both accuracy-conferring features (Fig. 2B and Table 1). Initially, two strains were constructed with mutations designed to eliminate Pol B synthesis. In T. kodakarensis TS742, the TK0001 reading frame was terminated by a nonsense mutation at codon 3, and in T. kodakarensis TS744, ∼1.7 kbp was deleted from the 5′ terminus of TK0001, removing the first extein, the extein-1–intein-1 boundary, and part of the first intein (Fig. 2B). Surprisingly, none of the mutations introduced into TK0001 had any detectable negative effects on growth or viability (see below). Given this result, and to confirm that there was no essential function encoded anywhere within the TK0001 sequence, T. kodakarensis TS745 was constructed with TK0001 completely deleted (see below; see also the supplemental material). The presence of each mutation and all genome structures was unequivocally verified by multiple diagnostic PCRs, restriction enzyme digestions, sequencing of amplicons, and Southern blotting (Fig. 3B to E; see also Fig. S2 in the supplemental material).

In contrast to the ready construction of T. kodakarensis strains with mutations in TK0001, despite using the same procedures and selections with many different plasmid donor DNAs, we were unable to generate a T. kodakarensis strain with an inactivating mutation in TK1902 and/or TK1903, encoding DP1 and DP2, respectively.

Cultures of T. kodakarensis TS742, TS744, and TS745 grew at the same rate and reached the same final cell densities as cultures of T. kodakarensis KW128 and TR1 (Fig. 4; see also Fig. S3 in the supplemental material). These strains also had indistinguishable plating efficiencies and formed colonies with the same sizes on several different Gelrite-solidified media. Western blotting confirmed that Pol B was present in lysates of T. kodakarensis KW128 cells taken from cultures during exponential growth, but identical processing and probing failed to detect any full-length or truncated Pol B in lysates of T. kodakarensis TS742 and TS744 (Fig. 4). In contrast, Western blotting using antiserum raised against DP1 confirmed the presence of Pol D in lysates of all three strains (Fig. 4). In the absence of Pol B, there was no detectable increase in the abundance of Pol D in T. kodakarensis TS742 and TS744 (Fig. 4).

T. kodakarensis incorporates exogenous purines, and this scavenging activity results in sensitivity to the analogue 6-methylpurine (6MP) (33, 39). Sensitivity is conferred by the activity of a hypoxanthine/guanine phosphoribosyltransferase encoded by TK0664. Virtually all mutants isolated as spontaneously resistant to 6MP (6MP^r) have mutations in TK0664, and this was used as an assay to measure the frequency and determine the patterns of spontaneous mutations that occurred in T. kodakarensis strains KW128, TR1, TR2, TR3, TR4, and TS744. Dilutions of cultures were plated onto media with or without 100 μM 6MP, and comparisons of the numbers of colonies formed, in at least three separate experiments, revealed that mutations conferring 6MP^r occurred spontaneously at statistically the same frequency (1 per 9.4 × 10⁷ ± 7 × 10⁷ plated cells) in the wild-type, Pol B-defective, and Pol B-deleted strains. To determine if these mutations were qualitatively different, the TK0664 locus was PCR amplified from >100 6MP^r mutants from each strain and sequenced. Given that only inactivating mutations were likely to result in 6MP^r, as expected, the majority of mutations in all strains were deletions or insertions, and many mutations changed residues predicted to be components of the enzyme's active site (see Fig. S1 in the supplemental material). In total, 163 different transitions and transversions were identified, most resulting in translation-terminating nonsense mutations, and transitions outnumbered transversions 2-fold in all strains (see Table S1 in the supplemental material). Consistent with a hyperthermophilic life-style increasing cytosine deamination (40, 41), C-to-T transitions occurred three times more often than T-to-C transitions, although the overall percentage of C-to-T transitions remained relatively constant, ranging from 22% to 28% in all strains. There were no striking differences in the profiles of spontaneous mutations that conferred 6MP^r in the strains that contained versus those that lacked Pol B, although deletions and insertions of >3 bp occurred more often in the absence of Pol B (see Table S1 in the supplemental material).

Direct comparisons of the sensitivity of T. kodakarensis KW128, TS742, and TS744 to methyl methanesulfonate (MMS), mitomycin C, and UV light irradiation were made. All three treatments were toxic, causing substantial cell death, but whereas there were no differences in the sensitivities of these strains to MMS and mitomycin C, T. kodakarensis TS742 and TS744 were more sensitive to UV irradiation than T. kodakarensis KW128 (Fig. 5).

A review of the original submission recommended that we construct a strain with the entire TK0001 sequence deleted to categorically establish that no TK0001-encoded function remained. As described in the supplemental material, T. kodakarensis TS745 (ΔTK0001) was therefore constructed (see Fig. S2A in the supplemental material), and the complete and precise deletion of TK0001 was confirmed by multiple diagnostic PCR amplifications and Southern blotting (see Fig. S2B to S2F in the supplemental material). T. kodakarensis TS745 grows at the same rate and to the same final cell density as the parental strain T. kodakarensis KW128 in liquid media (see Fig. S3 in the supplemental material) and has the same plating efficiency on solidified media. As documented for T. kodakarensis TS742 and TS744 (Fig. 5), T. kodakarensis TS745 also has increased sensitivity to UV irradiation.

In the absence of Pol B, Pol D must be the polymerase responsible for synthesizing both the leading and lagging DNA strands of the T. kodakarensis genome. All of the biochemical properties established for Pol D in vitro are consistent with a replication function (13–15, 25–30), and Pol D exists in vivo in a stable complex(es) that also contains many other components predicted for the T. kodakarensis replisome, including MCM1, the GINS subcomplex, DNA ligase, Cdc6, PCNA1, PCNA2, and GAN (31, 32). In contrast, when similarly isolated from cell lysates, Pol B was not present in a stable complex with these replisome components. The location of the genes (TK1902 and TK1903) that encode Pol D, directly adjacent to the gene (TK1901) that encodes Cdc6 and next to the presumed origin of chromosomal replication in T. kodakarensis, also hints strongly at a replication role (4, 34).

Pol B is abundant in T. kodakarensis and is encoded in all archaeal genomes (4, 6, 8, 13); however, given that its absence does not compromise the viability or growth of T. kodakarensis, the simplest argument is that Pol B does not play a critical role in genome replication in this archaeon. It is, however, important to note that DNA polymerase ϵ (DNAP-ϵ) has been shown conclusively to replicate the leading strand of chromosomal DNA in eukaryotes (2, 42, 43), and despite this, deletions of the N-terminal DNA-polymerizing catalytic domain of DNAP-ϵ are not lethal in yeasts (44–46). These deletion mutations do result in substantial growth defects, whereas point mutations in the same gene, which inactivate only the catalytic activity of DNAP-ϵ, are lethal (47). The explanation appears to be that catalytically active DNAP-ϵ is essential when full-length DNAP-ϵ is assembled into the replisome, but replisomes that support viability are formed with DNAP-δ when full-length DNAP-ϵ is not available for incorporation (2, 42, 47). Given this precedent, it could be argued that Pol B does normally catalyze genome replication in T. kodakarensis but that this function can be fully replaced by Pol D, as unlike the growth defects of yeast strains that lack the N terminus of DNAP-ϵ, T. kodakarensis strains lacking Pol B grow normally (Fig. 4; see also Fig. S3 in the supplemental material). Arguing against Pol B normally catalyzing genome replication in T. kodakarensis, the introduction of mutations that resulted in Pol B variants that lack 3′→5′ exonuclease proofreading and uracil-sensing mechanisms had no detectable effects on the frequency or profile of spontaneous mutation (see Table S1 in the supplemental material).

If Pol B is not primarily involved in replication, likely alternatives would be a role(s) in DNA repair and/or recombination, and there are precedents for the participation of Pol B family members in repair. For example, bacterial Pol II is involved in the restart of collapsed replication forks (48), and after a repair polymerase inserts a normal base opposite a damaged template base, eukaryotic Pol ζ extends the resulting mismatch (49). The increased UV sensitivity of T. kodakarensis strains lacking Pol B argues for a role in resolving DNA damage, but it is then difficult to explain why there were no similar increases in sensitivity to MMS and mitomycin C. It is nevertheless well established that the Thermococcales are unusually resistant to ionizing radiation (50–52), and this could reflect the constitutive and abundant presence of Pol B facilitating the reassembly of a radiation-fragmented genome.

The existence of analogous but nonorthologous replicative DNA replication proteins, including many DNA polymerases of the B family, in different biological lineages has led to the hypothesis that multiple replicative systems were established in the last universal common ancestor (LUCA) (Fig. 1) (8). One or more systems and replicative polymerases were then subsequently retained and/or lost in the lineages that now constitute the three domains of life. An alternative hypothesis posits that the different DNA replicative systems evolved successively over time, with transfer and replacement of the existing system in some lineages, likely facilitated by viral infections given that many viral genomes encode DNA polymerases (7). Providing experimental support for these hypotheses, establishing an evolutionary sequence, and determining the consequences of a nonorthologous DNA replication system replacement are clearly major research challenges. Currently, the literature arguments focus on the dichotomy of Pol C in Bacteria versus Pol B in Archaea, with the presence of Pol B in the eukaryotes being a consequence and consistent with Archaea and eukaryotes more recently sharing a common ancestor (11, 12). The discovery that Pol D is likely the replicative enzyme in T. kodakarensis, and, by extrapolation, perhaps in many archaeal lineages, adds a further complication. Most archaeal research to date has investigated members of only two lineages, the Euryarchaea, which have Pol D, and the Crenarchaea, which do not have Pol D. However, now with the rapid increases in genome and metagenome sequencing, it has become clear that there are many other early-branching archaeal lineages and that Pol D is also present in the Korarchaeota, Aigarchaeota, Nanoarchaeota, and Thaumarchaeota (Fig. 1) (13). Based on parsimony, possibly both Pol B and Pol D were present in the ancestral archaeal lineage and Pol D was lost in a branch that led to the Crenarchaea and eukaryotic replication systems. Alternatively, Pol D could have evolved in the archaeal lineage after the divergence of this putative crenarchaeal-eukaryotic lineage. The distribution of histones would, in contrast, argue against this lineage and hypothesis, as the Pol D-containing Archaea, in common with all eukaryotes, have histones, whereas the Crenarchaea do not (37, 53). There are clearly exceptions, but as a current generalization, we now propose that genomic replication can also be viewed meaningfully as conforming to the biological domain trichotomy (11, 12): Archaea employ Pol D, Bacteria employ Pol C, and eukaryotes employ Pol B.