Possible function of the second RecJ-like protein in stalled replication fork repair by interacting with Hef

RecJ was originally identified in Escherichia coli and plays an important role in the DNA repair and recombination pathways. Thermococcus kodakarensis, a hyperthermophilic archaeon, has two RecJ-like nucleases. These proteins are designated as GAN (GINS-associated nuclease) and HAN (Hef-associated nuclease), based on the protein they interact with. GAN is probably a counterpart of Cdc45 in the eukaryotic CMG replicative helicase complex. HAN is considered mainly to function with Hef for restoration of the stalled replication fork. In this study, we characterized HAN to clarify its functions in Thermococcus cells. HAN showed single-strand specific 3′ to 5′ exonuclease activity, which was stimulated in the presence of Hef. A gene disruption analysis revealed that HAN was non-essential for viability, but the ΔganΔhan double mutant did not grow under optimal conditions at 85 °C. This deficiency was not fully recovered by introducing the mutant han gene, encoding the nuclease-deficient HAN protein, back into the genome. These results suggest that the unstable replicative helicase complex without GAN performs ineffective fork progression, and thus the stalled fork repair system including HAN becomes more important. The nuclease activity of HAN is required for the function of this protein in T. kodakarensis.


Results
HAN constitutes an independent cluster in the phylogenetic tree of RecJ family proteins. RecJ family nucleases have been actively studied, as the predicted archaeal counterparts of Cdc45 23,25,27,28,[37][38][39] . To understand their structure-function relationships, published RecJ-like proteins were picked up with their properties (referred above) and were schematically added to the phylogenetic tree, which were calculated by Makarova et al. 19 (Fig. 1a). Bacterial RecJ has been regarded as a 5′ to 3′ exonuclease. However, no experimental data are available for the nuclease activity of eukaryotic Cdc45. The GAN ortholog is well conserved in many archaeal genomes (arCOG00427) 19 . However, it is interesting that a different group of proteins, which are not associated with GINS, is also included in this cluster. HAN, the second RecJ-like protein in T. kodakarensis, has typical DHH motifs, which are essential for the nuclease function in bacterial RecJs [40][41][42] . The characteristic feature is that HAN lacks motif V and the CID (CMG-Interaction domain)-like domain, which are unique to the GAN-family 24,28,39 (Supplementary Fig. S1). This C-terminal region containing the DHH motifs is fused to three nucleic acid (NA)-binding domains, a DnaJ-like zinc finger (ZF) domain and two distinct OB-fold domains, at its N-terminal region. Based on these structural features, we expected that HAN possesses a unique exonuclease activity. Interestingly, HAN is conserved only in the euryarchaeal phylum, except for Thermoplasmatales, in Archaea ( Supplementary Fig. S2).
Biochemical characterization of the nuclease activity of HAN. We first investigated the nuclease activity of the highly purified HAN (Fig. 1b). As shown in Fig. 1c, the 5′-32 P-labeled ssDNA substrate was shortened during the reaction. The substrate DNA became shorter than 10 nt, but was not degraded to single nucleotides (lanes 1-4, "normal" DNA). When HAN was reacted with the substrate DNA with four successive phosphorothioate modifications, the degradation was inhibited at the modification sites to produce 11-14 nt DNAs (lanes 6-9, "S" DNA). These results indicate that HAN cleaved ssDNA in the 3′ to 5′ direction. We also tested the nuclease activity of HAN using 3′-FITC-labeled DNA with or without phosphorothioated modification at 3′ terminus ( Supplementary Fig. S3a), and found a product shorter than 1 nt from only DNA without phosphorothioated modification (lane 4). The DNA with phosphorothioated modification was not cleaved. The reaction rate of HAN nuclease was high at 70 °C ( Supplementary Fig. S3b), although it was slow down to show the products as a ladder pattern at 50 °C ( Supplementary Fig. S3c, lanes 3 and 4). To investigate whether the HAN reaction is processive, excess amounts (10-and 100-times) of non-labeled substrate DNAs were used to protect HAN, which fell down from the substrate DNA, from reloading. These experiments also showed only the very short product (Fig. S3d), and therefore, HAN should be a highly processive exonuclease. HAN cleaved 3′-overhanged dsDNA, but not blunt ended dsDNA and 5′-overhanged dsDNA ( Supplementary Fig. S3c, lanes [6][7][8][9][10][11][12]. The cleavage of the 3′-overhanged DNA stopped at about 10 nt from the double-stranded region, supporting the conclusion that HAN specifically cleaves ssDNA in the 3′ to 5′ direction. HAN did not degrade circular ssDNA and dsDNA ( Supplementary Fig. S3e), indicating that HAN lacks endonuclease activity. In addition, HAN cleaved RNA with the same or slightly better efficiency ( Supplementary Fig. S3f). HAN cleaved 5′-FITC-labeled RNA and generated <14 nt products (Fig. 1d, lanes 2 and 3), but not RNA phosphorothioated at the same region as the DNA substrates ( Fig. 1d lanes 7 and 8). GAN did not cleave RNA from either the 5′ or 3′ end (lanes 4 and 11), as previously observed 23 . As a control, the DNase activity of the GAN protein used for this experiment was shown in Fig. S3g. The nuclease activity of HAN was clearly dependent on manganese, and magnesium could also be used with much less efficiency. Calcium and zinc were not utilized by HAN ( Supplementary Fig. S4). The metal-binding residue in motif I was predicted from the bacterial RecJ and GAN, and the alanine substitution mutant, HAN-D366A, was prepared (Fig. 1b) and used for the nuclease assay. This mutation almost completely inactivated HAN (Fig. 1c, lanes 5 and 10; Fig. 1d, lane 5). The loss of the exonuclease activity in HAN-D366A is consistent with the results of equivalent mutations in bacterial RecJ and archaeal RecJ-like proteins 25,38,40,43 . HAN (WT and D366A mutant) (5 nM each) was incubated with 5′-32 P-labeled 5 nM ssDNA (dA30) at 70 °C for 1, 2, and 5 min. The 30-nt ssDNAs with or without four successive phosphorothioate modifications from the 10th to 14th bases were used for the substrates and indicated as "normal" (dA30, lanes 1-5) and "S" (dA30ssss14, lanes 6-10), respectively. Lane 11 shows the size marker DNAs (14, 10, and 1 nt, respectively). (d) HAN exhibited 3′ to 5′ exonuclease activity for RNA. GAN and HAN (WT and D366A) (5 nM each) were incubated with 5′-or 3′-FITC-labeled 10 nM RNA at 80 °C for 0.5, 3, and 5 min. HAN cleaved RNA processively (lanes 2 and 3). Substrates in these assays were 5′-FITC-labeled "normal" RNA (FITCrA30, lanes 1-5), 5′-FITClabeled phosphorothioated ("S") RNA (FITCrA30ssss14, lanes 6-8), and 3′-FITC-labeled RNA (rA30FITC, lanes 10-11). Products were subjected to 8 M urea-15% PAGE. Lane 9 shows 5′ FITC-labeled RNA (20 nt and 14nt) and DNA (1 nt) size markers.

Specific interactions of HAN.
We previously reported the physical interactions of GAN-GINS and HAN-Hef, detected by SPR-analyses 24,25,29 . We tested whether GAN and HAN could substitute for each other by SPR analyses, using the highly purified GAN, GINS, HAN, and Hef (IDR) proteins. GINS bound the GAN-immobilized sensorchip, but Hef-IDR did not (Fig. 2a). Hef-IDR bound the HAN-immobilized sensorchip, but GINS did not (Fig. 2b). These results clearly showed that the interactions of GAN-GINS and HAN-Hef are specific, and there is no complementarity between the two RecJ-like proteins.

Co-localization of HAN and Hef in T. kodakarensis. The intracellular amounts of HAN and Hef
in exponentially-and stationary-phase T. kodakarensis cells were measured by western blot analyses, using anti-HAN and anti-Hef antisera. The amounts of the HAN and Hef proteins were calculated to be 10,000-15,000 molecules/cell (as the monomer) and 50-200 molecules/cell (as the dimer), respectively, based on a comparison of the band intensities from the cell extract and serial dilutions of each purified protein (Fig. 3a). The intracellular concentrations of HAN and Hef should be about 5 µM and 0.1 µM, respectively, in a T. kodakarensis cell, calculated from a spherical cell with a 1 μm diameter. HAN is over 10-fold more abundant than Hef and GAN 25 . To investigate whether the HAN and Hef proteins could form a complex in T. kodakarensis cells, an immunoprecipitation experiment was performed. The immunocomplexes were captured with either anti-HAN or anti-Hef antibodies from the whole cell extracts. The immunocomplexes were subjected to SDS-PAGE, followed by western blot analyses using each antibody. As shown in Fig. 3b, HAN and Hef co-precipitated with each of the antibodies, suggesting that HAN and Hef exist in the same complex in T. kodakarensis cells. However, the stoichiometric ratio of the two proteins is clearly different as described above, and therefore, many free HAN molecules should exist to share some other functions in the cells.
Effects of HAN and Hef on stalled fork DNA. Synthetic fork-structured DNAs were used for the nuclease assays of Hef and HAN. To detect the precise cleavage site, the template strand for leading synthesis and the nascent leading strand were labeled with FITC at the 3′-terminus and Cy5 at the 5′-terminus, respectively ( Supplementary Fig. S5). The splayed arm DNA (sub1) was not cleaved by Hef, as reported previously. The fork DNA without lagging strand synthesis (sub2) was slightly cleaved at the nascent leading strand on the template. However, the fork DNA with no gap at the nascent lagging strand (sub3) was clearly cleaved at the template for  The immunocomplexes were captured individually with each antiserum from the whole cell extract (as shown at the top), and were subjected to SDS-10% PAGE followed by western blot analyses using each antiserum (shown on the right side). The whole cell extracts without immunoprecipitation (Input) and those precipitated after a treatment with preimmune serum (PI) were also loaded as positive and negative controls, respectively. The relative loading amounts are indicated at the bottom of the panels. leading strand synthesis. These results confirmed that the fork-structured DNA without a lagging strand gap is a preferable substrate for Hef, and the single-stranded nascent DNA strand, produced by the fork regression, is not cleaved. Furthermore, we also confirmed that Hef cleaves the template strand for leading strand synthesis with the same efficiency in either the presence or absence of the ssDNA region corresponding to the nascent leading strand produced by fork regression (Supplementary Fig. S6).
Based on the interaction with Hef, we assumed that HAN could work at stalled fork DNA, and possibly cleaved the exposed-ssDNA produced by regression of the nascent leading strand. We tested the effects of HAN and Hef on the regressed fork DNA containing a single-stranded nascent leading strand, as shown in Fig. 4a (corresponding to sub3 in Supplementary Fig. S5). We incubated 100 nM HAN and 5 nM Hef (as the dimer) with this fluorescently-labeled regressed fork DNA with 100 mM Mg 2+ and 0.1 mM Mn 2+ , according to their intracellular concentrations 25 . As shown in Fig. 4b, the FITC-labeled products corresponding to the Hef reaction were decreased in the presence of HAN, as compared to those by Hef alone. In contrast, the Cy5-labeled products by HAN were clearly increased in the presence of Hef, as compared with those in the absence of Hef. These inhibitions and stimulations were observed repeatedly, and the quantitative results are shown in Fig. 4c and d. When we analyzed the sites cleaved by Hef, the nicking site in the template strand for leading synthesis was the same in the presence and absence of HAN. The hydrolysis from the 3′-terminus of the strand corresponding to the nascent leading strand by HAN was stopped at 11-13 nt from the junction ( Supplementary Fig. S7), which is longer than that of the products from the simple ssDNA ( Fig. 1c and Supplementary Fig. S3), probably because the fork-structure sterically hindered the grasping of the single-stranded DNA region.

HAN is not essential for viability, but the disruption caused a feeble phenotype in T. kodakarensis.
To analyze the functions of HAN in T. kodakarensis cells, we tried to disrupt the han gene on the genome, and isolated the han deletion mutant strain. We previously reported the Δgan mutant, and in this study, the gan and han (ΔganΔhan) deletion mutant was also isolated successfully. PCR amplification of the gan and han regions from each mutant strain produced DNA products with the predicted size, corresponding to the lengths of the 5′ and 3′ flanking regions of the targeted loci without the GAN-or HAN-coding regions (Fig. 5a-c). Therefore, both GAN and HAN are not essential for T. kodakarensis viability, at least under the normal growth conditions employed for selection (growth in ASW-YT-S 0 medium at 85 °C). No effect of the gan gene deletion on neighboring gene expressions was carefully confirmed in our previous report, because other ORFs exist in the immediate vicinities of the gan gene 25 . There are some intergenic regions both up and downstreams of the han gene. (Fig. S8a), and we predicted no effect of the deletion of just the coding region of HAN on the vicinities in this study. We measured the growth properties of the mutants in ASW-YT-Pr medium at 85 °C, and detected a slight growth defect of Δhan as compared with the parental strain, KUW1, and the growth-plateau of Δhan seemed to be lower than those of KUW1 and Δgan. We also found that the ΔganΔhan mutant cells did not grow in ASW-YT-Pr medium at 85 °C (Fig. 5b). The effect of the cell morphology on turbidity of the cell culture was checked by microscopic observations of the growing cells with different conditions. The cells were always observed as a typical cocci after growing at 93 °C or irradiated by 100 and 500 J/m 2 of UV, and were not different from the cells grown in the normal cultivation condition. The growth defects described above were reproducibly observed more than three times, with three independently isolated strains. The Δgan, Δhan, and ΔganΔhan mutants, as well as the parental KUW1 cells, grew normally in ASW-YT-Pr medium at 70 °C ( Supplementary Fig. S8). We tested whether the growth defects of Δhan and ΔganΔhan are derived from the loss of the HAN nuclease activity, by inserting the nuclease-deficient han gene (han-D364A/D366A) into the Δhan and Δgan strains, Δhan::han-D364A/ D366A and Δgan::han-D364A/D366A. The growth of the Δhan::han-D364A/D366A cells did not recover to that of KUW1, and the Δgan::han-D364A/D366A cells showed additional growth defects as compared with those of Δgan (Fig. 5b). These observations indicate that the nuclease activity of HAN is important for its functions in the T. kodakarensis cells. However, Δgan::han-D364A/D366A did not show the severe temperature-sensitive phenotype at 85 °C, in contrast to the ΔganΔhan cells, suggesting that HAN functions not only as a nuclease but also as an important component for survival at high temperature.
Damage sensitivity of the Δhan mutant. Since our previous analyses showed that the Δhef mutant was highly sensitive to MMC, Hef may perform a critical function in stalled fork repair, especially those caused by ICLs 32 . We examined whether the Δhan cells displayed increased sensitivity to DNA damaging agents, as HAN may function with Hef in a DNA repair pathway. We monitored how the deletion of the han gene impacted the cell viability after DNA-damaging treatments, by a drop dilution assay (Fig. 6a). Under normal conditions without any DNA-damaging treatment, the Δhan mutant cells exhibited growth comparable to the parental KUW1 and the Δhef mutant. To investigate the effects of damage induction on the growth of the Δhan and Δhef mutants, their viabilities were compared at a higher temperature (93 °C), in the presence of MMC, and after UV irradiation. The growth defect of the Δhef cells under these damage-inducing conditions was remarkable. The Δhef cells could not form colonies upon cultivation at 93 °C for 24 h, as observed for the Δgan cells previously 25 . The sensitivity of the Δhan cells to the high temperature increased slightly as compared with the parental KUW1 cells (Fig. 6a), and the viability was quantified by counting the survival rates of the cells (Fig. 6b). The Δhef cells showed high sensitivity for MMC, as previously reported in T. kodakarensis and H. volcanii 32,44 . The Δhan cells also showed some sensitivity to MMC, but less as compared to the Δhef cells (Fig. 6a,b). We further examined the UV-sensitivity of the Δhan mutant. The Δhef cells showed high sensitivity to 5 J/m 2 of UV irradiation, consistent with our previous observation. However, the Δhan cells were obviously more resistant to this dose of UV irradiation, although they were slightly more sensitive than the parental KUW1 cells (Fig. 6a,b). This sensitivity of the Δhan cells was in the limits of error at higher doses (8, 11 J/m 2 ) of UV irradiation, and the mean sensitivities of the parental strain and the Δhan cells were about the same (Fig. 6c).

Discussion
We have demonstrated that T. kodakarensis HAN, the second RecJ-family protein, actually possesses 3′ to 5′ exonuclease activity for ssDNA and RNA, in contrast to bacterial RecJ, a 5′ to 3′ exonuclease. The GAN protein, identified as the first RecJ-like protein in T. kodakarensis, has 5′ to 3′ exonuclease activity that is stimulated by GINS, a component of the replicative helicase complex. We reported recently that the GAN nuclease lacks activity in vitro in the presence of 100 mM Mg 2+ and 0.1 mM Mn 2+ , which are the intracellular concentrations in T. kodakarensis cells. In addition, the growth defect of the Δgan mutant cells was complemented by the introduction of the nuclease-deficient mutant gan gene. From these results, we proposed that GAN works as a component of the replicative helicase and fulfills a structural role, similar to the case of eukaryotic Cdc45 in the CMG complex, even though GAN clearly has 5′ to 3′ exonuclease activity in vitro [23][24][25] . In contrast to GAN, in this study we showed that HAN possesses 3′ to 5′ exonuclease activity under conditions with 100 mM Mg 2+ and 0.1 mM Mn 2+ . Furthermore, the nuclease-deficient mutant han gene did not fully recover the temperature-sensitive phenotype of the ΔganΔhan mutant cells. Therefore, the 3′ to 5′ exonuclease activity is actually important for HAN to function in T. kodakarensis cells.
HAN specifically interacts with Hef, which is probably involved in stalled replication fork repair, and therefore, the functional relationships between the two proteins have been predicted. We showed here that the 3′ to 5′ exonuclease activity of HAN was stimulated and the endonuclease activity of Hef was inhibited in the presence of both proteins in vitro. Based on these results, we propose a model in which Hef and HAN cooperatively work at the stalled replication fork (Fig. 7). The basic properties of Hef revealed that the endonuclease, which cleaves the junction of fork-structured DNA at the strand corresponding to the template for leading synthesis, does not work for a fork-structure containing a single-stranded gap on the lagging side (which happens when the lagging strand synthesis stops ahead and only leading strand synthesis proceeds). In this case, the helicase activity of Hef can unwind the nascent leading strand to promote the fork regression. The helicase domain and the nuclease domain are clearly separated and connected by an IDR in the Hef protein (schematically drawn in Fig. 7). Due to steric hindrance, these two domains cannot simultaneously access the junction site of the fork DNA. We predict that the HAN-bound Hef accesses the fork junction predominantly with the helicase domain. Then, the nascent leading strand is dissociated from the template to become a single strand with an exposed 3′-terminus, which is a preferable substrate for HAN. When the fork structured DNA is converted to the gap-less form in the lagging strand, the domain exchange from helicase to nuclease occurs. In the case of the stalled fork with a single-strand gap in the leading strand, the nuclease domain binds to the junction and cleaves the template DNA, as we observed previously. Notably, HAN may serve to determine which domain (helicase or nuclease) in Hef preferentially binds the stalled fork junction.
The effects of the han deletion mutation on the growth of T. kodakarensis cells were not serious under the normal growth conditions. Some 3′ to 5′ exonuclease activities derived from other proteins, such as DNA polymerase B and D 11 , and Exonuclease I 45 , may function as substitutes in the absence of HAN. However, the GAN and HAN double mutant (ΔganΔhan) exhibited a temperature-sensitive phenotype at 85 °C. We previously reported that the Δgan mutant showed a temperature-sensitive phenotype at 93 °C, but was normal at 85 °C 25 . We assume that the replicative helicase complex without GAN becomes unstable for sufficient progression at higher temperatures, and the replication fork tends to stop, as compared with the wild type. In this situation, the effect of the HAN deficiency becomes higher in the ΔganΔhan mutant. We performed a targeted insertion of the nuclease-defective han gene (han-D364A/D366A) into the Δgan strain. However, the han-D364A/D366A gene did not fully complement the temperature-sensitivity of the ΔganΔhan mutant, in contrast to the case of GAN, in which the growth defects of the Δgan strain were fully recovered by the introduction of the nuclease-deficient gan gene (gan-D34A/ D36A) into the genome 25 . These results indicate that HAN works as a nuclease in T. kodakarensis cells, in contrast to GAN, which probably lacks nuclease activity in the cells. The partial complementation of the growth defect of the ΔganΔhan mutant by the nuclease-deficient HAN may be due to the fact that HAN maintains the function to select the helicase domain in Hef to bind the stalled fork DNA, even though its nuclease active site is disrupted. There are 50-300 times more HAN molecules than Hef in T. kodakarensis cells, and it is possible that HAN plays another role independently from Hef. HAN works on both RNA and DNA, and it may also be involved in RNA metabolism. Screening for additional interacting partners of HAN in T. kodakarensis cells will be useful to clarify this issue.
Hef was originally identified in P. furiosus, which belongs to the euryarchaeal phylum in Archaea. The phylogenetic relationships of the RecJ family proteins revealed that the HAN family is separated from other groups, including archaeal GAN-like proteins, eukaryotic Cdc45s, and bacterial RecJs (Fig. 1a). Interestingly, the HAN family is conserved only in Euryarchaeota that have the Hef protein 29 (Supplementary Fig. S2). These distributions support the cooperative functions of Hef and HAN in a euryarchaeal-specific repair pathway. Given their specific habitat, the euryarchaeal organisms may have developed a unique repair pathway with HAN and Hef for maintaining their genetic information. The homologous recombination process could follow the Hef-mediated fork disruption. Polyploidy may be a common trait of euryarchaeal cells 46 , and the copy number of chromosomal DNA in T. kodakarensis is 7-19 copies per cell 47 . The multiple copies of DNA would be advantageous for efficient homologous recombinational repair in euryarchaeal cells.
The domain organizations are clearly different between the GAN and HAN proteins. This difference is probably related to the specific properties of each protein. For example, the cleavage activity of GAN works only for ssDNA, starting from the 5′-end and proceeding to the 3′-end of the DNA strand. In contrast, HAN works on both ssDNA and RNA, and the cleavage starts from the 3′-end and stops at about <10 nt from the 5′-end. HAN probably needs to grasp about 10 nucleotides of the DNA strand to express its nuclease activity. The DnaJ-like ZF domain and two OB-fold domains (OB1 and OB2), found in the N-terminal region, are unique features of the HAN protein, although some of them lack ZF (Supplementary Fig. S2). These domains probably work for the DNA binding and confer the specificity. The structures and functions of the GAN family are more complicated. The domain organizations of the proteins in this family are unified, as shown in Supplementary Fig. S1. However, some are not associated with GINS (therefore, they are actually not GAN). RecJ1 from T. acidophilum has 5′ to 3′ exonuclease activity, but it is not associated with GINS, whereas RecJ2 has 3′ to 5′ exonuclease activity that is clearly stimulated by GINS. RecJ1 from Methanocaldococcus jannaschii has 5′ to 3′ exonuclease activity, and the recJ1 gene can complement the UV sensitivity of E. coli recJ mutant cells 37 . It is not known whether M. jannaschii RecJ1 interacts with GINS. RecJ2 from this organism has 3′ to 5′ exonuclease activity, but it does not interact with GINS. The RecJ-like protein from P. furiosus (PfRecJ) reportedly showed other different properties. PfRecJ interacts with GINS, and its 5′ to 3′ exonuclease is stimulated by GINS. However, it also has 3′ to 5′ exonuclease activity for RNA, which is not stimulated by GINS. The authors of this report proposed that PfRecJ functions for the proof-reading of the primers generated by primase mistakes 38 . Considering these results, it is difficult at present to propose a unified vision of the properties of the GAN family proteins in Archaea.
Studies of RecJ family proteins are becoming more popular. We have presented the properties of the euryarchaeal-specific HAN protein. It is interesting that two RecJ family proteins, GAN and HAN, may function in progression and repair, respectively, for the replication fork in T. kodakarensis. Further studies are necessary to understand the functions of the variously evolved RecJ family proteins.

Methods
Site-specific mutagenesis. To construct the pET-21a(+) (Novagen)-based expression plasmid for HAN-D366A and the plasmid to replace the native han gene with han-D364A/D366A, PCR-mediated mutagenesis was performed with a QuikChange TM site-directed mutagenesis kit (Agilent Technologies). The templates were pET-HAN [the gene encoding TkoHAN inserted into the NdeI-NotI sites of the expression plasmid pET-21a(+)] 29 and pUD3-HAN (described below). The designed mutations were confirmed by nucleotide sequencing. The primers used for each mutagenesis are listed in Supplementary Table S1.

Recombinant protein purification. The recombinant HAN (TK_RS00760) and Hef (TK_RS05025)
(full-length and His-tagged IDR proteins) were prepared according to the procedures described in our previous reports 29,32 . HAN-D366A was prepared in the same manner as the wild-type protein.
Nuclease assay. Surface plasmon resonance (SPR) analysis. The Biacore J system (GE Healthcare) was used to examine the physical interactions of HAN-GINS, HAN-Hef (IDR), GAN-GINS, and GAN-Hef (IDR). Either HAN or GAN was fixed on a CM5 sensor Chip (GE Healthcare), according to the manufacturer's recommendations. The solutions containing purified 2 µM GINS (as the tetramer) and 1 µM Hef-IDR proteins, in running buffer (10 mM HEPES-NaOH, pH 7.4, 150 mM NaCl, 3 mM EDTA, and 0.2% Tween 20), were applied for 120 s to the protein-immobilized chips, at a flow rate of 30 μl/min at 25 °C. The bound analytes were removed by regeneration buffer (10 mM HEPES-NaOH, pH 7.5, 1 M NaCl, and 0.2% Tween 20) at the end of each cycle.
Immunoprecipitation assay. T. kodakarensis KOD1 cells were cultured in 2 liters of MA-YT-Pr medium at 85 °C, and harvested at the late exponential phase by centrifugation for 10 min at 5,000 × g. The cells (9.1 × 10 11 cells) were resuspended in 45.5 ml of 50 mM Bis-Tris buffer, pH 7.0, to a final concentration of 2 × 10 10 cells/ml, and were disrupted by sonication. The cell extracts were obtained by centrifugation for 10 min at 23,000 × g. Polyclonal anti-HAN and anti-Hef antisera were raised independently, by immunizing rabbits with the purified recombinant proteins as antigens. A 20 µl portion of rProtein A Sepharose FF (GE Healthcare) was washed three times with phosphate buffered saline-Tween 20 (PBS-T: 10 mM sodium phosphate, pH 7.5, 150 mM NaCl, 0.1% Tween 20), mixed with 500 µl PBS-T containing 100 µl of each antiserum, and incubated on a rotating wheel at room temperature for 1 h. Each mixture was washed three times with 500 µl of 0.2 M triethanolamine, pH 8.0. Each antibody was cross-linked to the rProtein A Sepharose with 10 mM dimethyl sulfide (Thermo Fisher Scientific), according to the manufacturer's protocol. After equilibration of the antibody-conjugated rProtein A Sepharose with 50 mM Bis-Tris, pH 7.0, a 7.5 ml aliquot of the cell extract was added. The mixture was incubated on a rotating wheel at room temperature for 1 h. The precipitates were washed three times with 50 mM Bis-Tris, pH 7.0, and the immunoprecipitated proteins were eluted with 75 µl of gel-loading solution, containing 50 mM Tris-HCl, pH 6.8, 10% glycerol, 5% β-mercaptoethanol, 0.2% bromophenol blue, and 2% SDS. The eluted proteins were separated by SDS-PAGE and analyzed by western blotting.
Western blot analysis. The proteins on the gel were electroblotted onto a polyvinylidene difluoride (PVDF) membrane (Bio-Rad) using a Trans-Blot Turbo Transfer System (Bio-Rad), and reacted with the anti-HAN and anti-Hef antisera, prepared with each recombinant protein. Anti-Rabbit IgG HRP (Rabbit TrueBlot, Rockland Immunochemicals, Inc.) was used as the secondary antibody. The proteins were visualized by an enhanced chemiluminescence system (Millipore), and images were obtained and quantified with an LAS-3000 image analyzer (Fujifilm).
Construction of the T. kodakarensis Δhan, ΔganΔhan, Δhan::han-D364A/D366A, and Δgan::han-D364A/D366A strains. The gan gene, along with 1,000 bp of its 5′-and 3′-flanking regions, was amplified from T. kodakarensis genomic DNA, using the primer sets fhan-F/R (Supplementary Table S4). The PCR fragment was inserted into pUD3 52 after digestion with BamHI and EcoRI, and the resultant plasmid was designated as pUD3-HAN. An inverse PCR to remove the gan-encoding region was performed, using the dhan-F/R primer set, and the amplified fragment was self-ligated (pUD3-ΔHAN). The sequences of the 5′-and 3′-flanking regions were confirmed. The transformation was performed as described previously 52 . The ΔhanΔgan strains were obtained by transforming the Δhan cells with pUD3-ΔGAN 25 . The genotypes of the isolated transformants were analyzed by PCR, using han-F/R and fhan-F/R, and the gan gene locus of the candidate strains was sequenced to confirm the gene disruption. The plasmid bearing the gene encoding HAN-D364A/ D366A (pUD3-HAN-D364A/D366A) was prepared by site-specific mutagenesis. The Δhan and Δgan strains were transformed with pUD3-HAN-D364A/D366A, and the strains harboring the nuclease-deficient han gene were isolated in the same manner, and designated as Δhan::han-D364A/D366A and Δgan::han-D364A/D366A, respectively.
Growth measurements. The growth characteristics of the T. kodakarensis strains KUW1, Δgan, Δhan, ΔganΔhan, Δhan::han-D364A/D366A, and Δgan::han-D364A/D366A were measured, as follows. Each strain was precultured in ASW-YT-Pr medium at 85 °C (except ΔganΔhan) or 70 °C (ΔganΔhan) for 6-12 h, until the culture attained an OD 660 of 0.2-0.3. After preculturing, the cells were inoculated into 15 ml of ASW-YT-Pr to produce an OD 660 of 0.01, and cultured at 85 and 70 °C in glass test tubes. Cell densities were measured at appropriate intervals at 660 nm (OD 660 ), with a Novaspec II spectrophotometer (Pharmacia Biotech). Each strain was analyzed in duplicate, and the measurements were performed three times independently.
Drop dilution assay. The T. kodakarensis KUW1, Δhan, and Δhef strains were cultivated anaerobically at 85 °C for 12 h, in ASW-YT-Pr medium. The cultures (5 × 10 8 cells/ml) were serially diluted ten-fold with 0.8 × ASW. The diluted cultures were spotted on ASW-YT plates, and some of them were irradiated with an acrylic-filtered UV lamp (Kenis) (wavelength 254 nm, 5, 8, and 11 J/m 2 , calculated by an ATV-3W UV counter, ATTO). For exposure to MMC, the cells in 1 ml of culture were collected by centrifugation and resuspended in 100 µl of 0.8 × ASW, and mitomycin C (MMC) (100 µg/ml) was added. The cells were incubated at room temperature for 200 min, and then serial dilutions were spotted on ASW-YT plates. The plates were incubated anaerobically at 85 °C or 93 °C for 24 h, and the living cells were counted and visualized by staining with Coomassie Brilliant Blue (CBB), after transfer to a PVDF membrane, as described previously 53 . Experiments were performed in duplicate each time, and repeated three times independently. Data availability. No datasets were generated or analyzed during the current study.