An archaeal Cas3 protein facilitates rapid recovery from DNA damage

Abstract CRISPR-Cas systems provide heritable acquired immunity against viruses to archaea and bacteria. Cas3 is a CRISPR-associated protein that is common to all Type I systems, possesses both nuclease and helicase activities, and is responsible for degradation of invading DNA. Involvement of Cas3 in DNA repair had been suggested in the past, but then set aside when the role of CRISPR-Cas as an adaptive immune system was realized. Here we show that in the model archaeon Haloferax volcanii a cas3 deletion mutant exhibits increased resistance to DNA damaging agents compared with the wild-type strain, but its ability to recover quickly from such damage is reduced. Analysis of cas3 point mutants revealed that the helicase domain of the protein is responsible for the DNA damage sensitivity phenotype. Epistasis analysis indicated that cas3 operates with mre11 and rad50 in restraining the homologous recombination pathway of DNA repair. Mutants deleted for Cas3 or deficient in its helicase activity showed higher rates of homologous recombination, as measured in pop-in assays using non-replicating plasmids. These results demonstrate that Cas proteins act in DNA repair, in addition to their role in defense against selfish elements and are an integral part of the cellular response to DNA damage.


Introduction
Haloferax volcanii is a halophilic archaeon and an important archaeal model species, with many genetic tools and different "omics"-approaches available. Work on DNA damage repair in H. volcanii has shown that DNA double-strand breaks can be repaired either by homologous recombination (HR) or by a process known as microhomology-mediated end joining (MMEJ) (Marshall and Santangelo 2020;Pérez-Arnaiz et al. 2020). MMEJ relies on the resection of exposed DNA ends by exonucleases, which expose short stretches of single-stranded DNA that are homologous to one another and enable the two strands to anneal (White and Allers 2018). Unlike MMEJ, which inevitably results in deletion of a DNA fragment due to the resection and reannealing process, repair by HR generally keeps the genetic information intact and is usually error-free. Nonetheless, in H. volcanii, HR is restrained by the presence of two proteins, Mre11 and Rad50, that act to prevent HR and thereby allowing MMEJ to act as the primary double-strand break repair pathway (Delmas et al. 2009). These two proteins, which are highly conserved in archaea and eukaryotes, are also required for DNA compaction following DNA damage (Delmas, Duggin, and Allers 2013). When the genes for Mre11 and Rad50 are deleted, HR becomes the dominant mode of repair in H. volcanii, which results in slower recovery from DNA damage but also improved cell survival rates (Delmas et al. 2009). Thus, there appears to be a tradeoff in H. volcanii between repair speed (using rapid but inaccurate MMEJ) and survival (using slow but accurate HR).
CRISPR-Cas systems provide acquired and heritable immunity to archaea and bacteria against selfish mobile elements, especially viruses, by degrading the nucleic acids of invading elements that match the DNA-based immune memory stored in CRISPR arrays. Cas3 is a CRISPR-associated protein that is common to all Type I systems. Cas3 possesses both nuclease and helicase activities (Makarova et al. 2002(Makarova et al. , 2006Sinkunas et al. 2011;Gong et al. 2014) and after being recruited by a multiprotein complex bound to invader-targeting crRNA, is responsible for degradation of invading DNA (Maier, Dyall-Smith, and Marchfelder 2015;Hille et al. 2018). Involvement of Cas3 in DNA repair had been suggested in the past (Makarova et al. 2002), but this proposal was set aside when the canonical role of CRISPR-Cas as an adaptive immune system was realized. Previous functionality observation of the CRISPR-Cas system in DNA repair in Escherichia coli (Babu et al. 2011), encouraged us to reexamine the earlier ideas of Cas3 and its specific involvement in DNA repair. Here we use a genetic approach to unravel the interaction of Cas3 with the DNA repair machinery of H. volcanii, and how it affects HR in this species.

A cas3 H. volcanii mutant shows higher resistance to DNA damaging agents
To test whether cas genes play a role in DNA repair in H. volcanii, we constructed two H. volcanii mutants. The first was deleted for all cas genes, including both CRISPR loci from pHV4 ( cas genes) and the second had a deletion of just the cas3 gene (cas3-KO) (Supplementary Fig. S1). We examined the sensitivity of both mutants to DNA damage, utilizing ultraviolet (UV-C) exposure as a DNA damaging agent (Delmas et al. 2009). Notably, the cas3-KO mutant showed significantly higher resistance to UV damage compared to both the wild-type (WT) strain and the cas genes mutant, as observed by colony formation at higher dilutions (Supplementary Fig. S2). Since the cas3-KO mutant included the deletion of the entire cas3 gene, which resulted in deletion of a potential cas4 promoter, we constructed new separate cas3 and cas4 mutants ( Supplementary Fig. S1), to determine which of these two cas genes is responsible for the UV-resistance phenotype. Those new mutants were then exposed to DNA damaging agents, both UV-C and the DNA-alkylating mutagen methyl methanesulphonate (MMS) (Delmas et al. 2009). In both cases, the resistance of the cas3 mutant to the DNA damaging agents was significantly higher than the WT strain and the cas4 mutant, as observed by colony formation of the cas3 mutant in the higher dilution series (Fig. 1). To confirm that DNA damage resistance was related to the deletion of the cas3 gene, we conducted a complementation assay, transforming an expression plasmid (pTA927) with the cas3 gene under the tryptophan-inducible promoter p.tnaA to the cas3 strain. As expected, the MMS sensitivity of cas3 mutant was restored to the WT levels by the induced expression of the cas3 gene from that expression plasmid ( Fig. 1 C). These results suggested a possible involvement of Cas3 in DNA damage response in H. volcanii. Moreover, the contrast between the resistance of the cas3 mutant and the sensitivity of the cas genes mutant to DNA damaging agents, suggests that different cas genes might affect the susceptibility to DNA damage differently and in opposing directions in H. volcanii. Indeed, Cas1 contributes to DNA repair in E. coli (Babu et al. 2011), has been recently shown to contribute to DNA repair in H. volcanii, and its deletion alongside the other cas genes could increase sensitivity to DNA damage (Wörtz et al. 2022).

Cas3 promotes rapid recovery from DNA damage
In the absence of foreign invading elements, the maintenance of CRISPR-Cas systems is assumed to be a burden on cells expressing them, due to the energetic cost of constitutive expression of the interference machinery, including the cas3 gene that is always expressed in H. volcanii (Brendel et al. 2014). However, the cas3 mutants show no difference in growth rate compared to the WT strain under optimal conditions (Supplementary Fig. S3 A), a result which was also corroborated by direct head-to-head competition in liquid medium ( Supplementary Fig. S3 B). This suggests that cas3 does not have a significant impact on the cell's fitness in the absence of infection by viruses or other exogenous stress.
Nonetheless, when we exposed the H. volcanii cultures to MMS, the cas3 mutant showed significant delay in growth compared with the WT strain ( Fig. 2 A). A similar result was observed in DNA damage recovery assay (Fig. 2 B), where after the first hour of MMS treatment, the mutagen was washed off and removed, and the cultures were left to recover under the same conditions for different time periods.
MMS predominantly creates methylated DNA lesions such as 7-methylguanine (7meG) and 3-methlyladenine (3meA), causing base mispairing and replication blocks (Beranek 1990). DNA damage caused by alkylating agents is primarily repaired by the base excision repair (BER) pathway or by DNA alkyltransferases (Lindahl and Wood 1999), but the sensitivity of cells to MMS also increases significantly when other DNA repair pathways are com-promised (Lundin et al. 2005;Delmas et al. 2009). For better understanding of the recovery process and the repair of the MMS lesions, we used RAre DAmage and Repair sequencing (RADAR-seq) (Zatopek et al. 2019) on DNA extractions from WT strain and cas3 mutant to quantitate 7meG and 3meA levels immediately after MMS treatment and post recovery. This method uses a DNA glycosylase in combination with Endonuclease IV to generate nicks at damaged DNA sites and then a DNA polymerase and a ligase that convert those nicks to modified bases that are detectable by PacBio sequencing (see Materials and methods section). Both the WT and the cas3 mutant appear to possess a relatively similar number of lesions, with an average of 850-880 (respectively) lesions per genome after 1 h of MMS treatment (Fig. 3 A). On the other hand, the results after the recovery time showed a higher repair efficiency in the WT strain with about 30% of repair compared to 5% of repair in the cas3 mutant (Fig. 3 B). Importantly, no particular hotspots of lesions could be observed in either mutant or WT, not even near the locus that partially matches one of the H. volcanii spacers (Fischer et al. 2012), indicating that there is no underlying site-specific CRISPR targeting involved (Supplementary Fig. S5).
Taken together, these results show that while the amount of damage caused to both strains is comparable, after recovery there is an advantage to the WT strain, containing the functional Cas3 protein, which repairs lesions faster.

Cas3 helicase activity is responsible for DNA damage sensitivity
Since the H. volcanii Cas3 protein is predicted to have both helicase and nuclease domains, as seen in well-studied bacterial homologs (Makarova et al. 2002(Makarova et al. , 2006Haft et al. 2005;van der Oost et al. 2009), our next goal was to distinguish between these two activities to clarify which domain is responsible for sensitivity to DNA damage. For that end, we used two expression vectors (pTA927) that carried mutated H. volcanii cas3 genes under the p.tnaA promoter. One of the cloned genes was defective in its nuclease domain (encoding a nuclease-dead Cas3), with the HD63-64AA double mutation that has been shown to abolish the Cas3 nuclease activity Howard et al. 2011;Mulepati and Bailey 2011;Sinkunas et al. 2011;). The other plasmid encoded a Cas3 protein with a defect in its helicase domain (helicase-dead Cas3, D444A), that has been shown to be important for Cas3 helicase activity (Sinkunas et al. 2011;). These two plasmids were transformed separately into the cas3 mutant background and tested using the DNA damage assay based on MMS treatment. Since the plasmid-encoded cas3 mutant genes were under the tryptophan induced p.tnaA promoter, we tested growth without tryptophan addition and with 2 mM tryptophan added.
Under the noninducing conditions, all the four mutants ( cas3, cas3:: cas3, cas3:: cas3 H.D, and cas3:: cas3 N.D) showed increased cell survival in comparison to the WT strain, following the MMS treatment. (Fig. 4). This indicates that none of the exogenous cas3 genes were induced (or that their expression was minor due to a promoter leakage), as expected, and the resistance phenotype they all shared was due to their joint cas3 background. In contrast, in the presence of tryptophan, only the cas3 and the cas3:: cas3 H.D (helicase dead) mutants were still more resistant to MMS, while the cas3:: cas3 and cas3:: cas3 N.D (nuclease dead) strains were more sensitive to the treatment (Fig. 4). The noncomplemented cas3 strain benefited from the addition of tryptophan to the medium in terms of its survival phenotype. These results    helicase-dead (UG618), and cas3:: cas3 nuclease-dead (UG645) were grown in Hv-YPC (containing 2 mM tryptophan for promoter activation) or in Hv-CA (noninduced medium) and exposed to DNA damage by incubation with 0.7 mg/ml MMS and plated on Hv-YPC. Surviving colonies were enumerated after 5-10 days of incubation and fraction of surviving cells for each mutant was normalized to the WT strain. In each case, the mean and SE of nine experiments are shown. * * * = P-value <.001, Mann-Whitney test, compared to WT.
suggest that the Cas3 helicase activity mediates the sensitivity to DNA damaging agents in H. volcanii, while Cas3 nuclease activity has a minor role, if any, in the response to DNA damage.

Cas3 participates in the Mre11-Rad50 DNA repair pathway
Double strand breaks (DSBs) are considered to be one of the most destructive forms of DNA damage. Such a break, in both strands of the DNA, may be subjected to illegitimate recombination or ligation if left unrepaired, and these processes can lead to chromosomal rearrangements. Both HR and MMEJ were previously shown to be DSB repair pathways in H. volcanii. It is well known that HR is highly accurate and error-free with a higher energetic cost, while MMEJ is a faster, but error-prone process (Blackwood et al. 2013;Marshall and Santangelo 2020;Pérez-Arnaiz et al. 2020). Correct regulation of both error-prone and HR pathways is required for optimal DNA repair, and it appears that in H. volcanii the Mre11-Rad50 complex has a significant impact in committing to a specific pathway, by restraining the HR pathway and thereby allowing MMEJ to act as the primary pathway of DSB repair (Delmas et al. 2009). Interestingly, it was shown that mre11-rad50 mutants were Figure 5. cas3 shows an epistatic (nonadditive) effect in the background of mre11-rad50. WT (H115), mre11 (H203), mre11 cas3 (UG669), rad50 (H202), rad50 cas3 (UG670), mre11 rad50 (H204), and mre11 rad50 cas3 (UG671) cells were exposed to DNA damage by 0.7 mg/ml MMS treatment and plated on Hv-YPC. Surviving colonies were counted after 5-10 days of incubation and the fraction of survival of each mutant was normalized to the WT strain. In each case, the mean and SE of nine experiments are shown. * * * = P-value <.001, Mann-Whitney test, compared to WT. much more resistant to DNA damaging agents, but their recovery from the damage was slower (Delmas et al. 2009), similar to what we observed with the cas3 mutant.
Those observations encouraged us to examine the resistance to MMS of a mutant that has all three genes deleted: cas3, mre11, and rad50. Such an experiment might result in 2 different outcomes: the resistance effect might be enhanced and indicate that cas3 and mre11-rad50 operate in two different pathways (synergistic effect), or alternatively, the resistance effect might not change indicating that all three genes act in the same pathway (neutral effect, a form of classic epistasis). By conducting the DNA damage assay with MMS in all deletion combinations, the phenotype was highly similar, with around 10-fold higher resistance in the different mutation combinations compared to the WT strain (Fig. 5). These results suggest that cas3 most likely participates with mre11-rad50, by restraining HR and thereby allowing MMEJ to act as primary DSB repair pathway.

Cas3 decreases the efficiency of HR
To obtain more direct results about the role of cas3 in HR pathway, we decided to execute a "pop-in" recombination assay (Dattani et al. 2022), which enabled us to measure the integration efficiency of a "suicide" plasmid (that does not contain a H. volcanii origin of replication), into the cas3 mutant genome, which could be achieved only via HR of the plasmid with the chromosome.
When transforming the cas3 mutant with the integrative plasmid, higher pop-in (recombination) efficiency was obtained compared to WT (Fig. 6 A), suggesting an increased HR efficiency in the cas3 mutant (see above). To validate that these differences are HR-based and not due to transformation efficacy, we performed a transformation efficiency assay with the replicative plasmid pTA927 (containing a H. volcanii origin of replication, which does not require recombination for successful transformation). The control assay resulted in a minor and insignificant differences in transformation efficiency between WT and cas3 strains ( Supplementary Fig. S4). Normalized recombination efficiency values were obtained by dividing the pop-in transformation efficiency values (using the integrative plasmid) by the transformation efficiency values (obtained for the same strains using the replicating plasmid). This calculation showed the same trend, namely that the cas3 mutant has higher HR rates than the WT by one order of magnitude (Fig. 6 B). Furthermore, when these transformation assays were performed in cas3 nuclease-dead and cas3 helicase-dead backgrounds, they revealed that deficiency in Cas3 helicase domain was not only responsible for greater cellular survival following MMS treatment (Fig. 4), but also for the increased rate of HR (Fig. 6B). A strain deleted in all cas genes showed an intermediate phenotype, closer to that of the WT ( Supplementary  Fig. S6).

Discussion
CRISPR-Cas systems play important roles in prokaryotic immunity. The three stages of the CRISPR-Cas immunity process (adaptation, maturation, and interference) have been well studied and characterized, and the function of the cas genes has been elucidated in several types and subtypes, including I-B systems, such as that of the haloarchaeon H. volcanii (Maier et al. 2015). Recent work has shown that at least one protein, namely Cas1, also participates in DNA repair (Babu et al. 2011;Wörtz et al. 2022). However, the role of other cas genes in DNA repair is still unknown. Here we provide evidence for involvement of Cas3 in the process of DNA repair in archaea, potentially in restraining the HR pathway.
By conducting survival assays after UV exposure and MMS treatment, we have observed that a H. volcanii cas3 mutant shows greater cell survival than the WT (Fig. 1, Supplementary Fig. S2). These results suggested a possible role for cas3 gene in DNA damage response in H. volcanii. Surprisingly, although the H. volcanii cas3 mutant was found to be more resistant to DNA damaging agents than the WT, its ability to quickly resume growth was impaired (Fig. 2). In addition, the RADAR-seq analysis revealed that its ability to quickly repair MMS lesions was also reduced. (Fig. 3). These observations suggest that WT and cas3 cells utilize two different DNA repair strategies. While cas3 uses a slower but more accurate repair strategy, this will benefit the organism only in noncompetitive situations, the WT uses a repair strategy with a decisive advantage in natural competition scenarios, because cells will be faster to resume growth and replication. Delmas et al., (2009) previously showed that H. volcanii mre11-rad50 mutant cells are more resistant to DNA damaging agents but are also slower to recover. Further, they demonstrated that the DNA repair strategy used in mre11-rad50 mutants cells involves unrestrained HR, while in WT cells, DSBs were mainly repaired by MMEJ. By conducting DNA damage assays using MMS in all deletion combinations (mre11, rad50, and cas3), we observed that the resistance phenotype did not show additivity. All combinations of deletions showed an about 10-fold higher resistance on average, compared with the WT strain (Fig. 5). Furthermore, similar to deletion of the mre11-rad50 genes, deletion of cas3 also leads to increased efficiency in HR (Fig. 6 B). These results suggest that Cas3 participates with Mre11-Rad50 in restraining HR, thereby allowing the faster MMEJ to act as primary DSB repair pathway. It is believed that restraining HR may hold a significant advantage for a highly polyploid organism, such as H. volcanii, leading to a to ensure rapid repair of DSBs at the expense of accuracy by using MMEJ in the first step, and later using HR to resolve any mutations and thereby maintain genome fidelity in the second step (Delmas et al. 2009).
Analysis of cas3 mutations revealed that the helicase activity of Cas3 is responsible for both DNA damage sensitivity (Fig. 4) and decreased HR efficiency (Fig. 6 B). These results provide a starting point for a mechanistic understanding that needs to be further explored. Our (nonexclusive) hypothesis, which best fits with the results, is that Cas3 helicase activity functions in restraining HR, possibly by removal of the D-loops and/or recombinase proteins to prevent HR. Consequently, this activity of Cas3 promotes the alternative MMEJ mechanism to act as the primary DSB repair pathway.
CRISPR-Cas systems require DNA nucleases and helicases to exert protection from selfish elements, and therefore are prone to interactions with the DNA repair machinery enzymes that perform these functions. Bernheim et al. (2019) suggested that the repertoire of DNA DSB repair systems an organism possesses determines to a great extent, which CRISPR-Cas systems it can acquire, if any, and showed that different subtypes are associated (either negatively or positively) with different repair components. It has been observed, e.g. that type II-A CRISPR-Cas systems in bacteria are very rarely found in genomes that also have the machinery for nonhomologous end joining that requires the Ku protein, probably because the Cas protein Csn2 interferes with this mechanism of DNA repair (Bernheim et al. 2017).
AddAB, which is involved in strand resection and in some ways is functionally analogous to Mre11-Rad50, is negatively associated with the I-B subtype in Firmicutes, but positively associated with this subtype in Proteobacteria (Bernheim et al. 2019). This implies that interactions between archaeal I-B systems and DNA repair are not just possible but rather probable. Indeed, recent work in H. volcanii has shown that the Cas1 integrase, whose primarily role is to integrate new spacers into the CRISPR array, is important for cellular survival of DNA damage caused by oxidative stress and that it can substitute for the DNA repair enzyme Flap endonuclease 1 (Fen1) in cleaving branched intermediate structures (flaps) (Wörtz et al. 2022). Given that such flap structures are also formed during MMEJ and our observations that Cas3 is involved in the MMEJ process, it is likely that both Cas1 and Cas3 can independently be found at repair sites. This could help direct spacer acquisition to elements already cut by other defensive nucleases such as restriction enzymes, but also cause acquisition of self-targeting "spacers" (Stachler et al. 2017(Stachler et al. , 2020). An additional advantage may be that by restraining HR, Cas3 will reduce the chances of foreign DNA elements, such as plasmids [which the H. volcanii system successfully targets and destroys (Fischer et al. 2012)], recombining with the host chromosome and integrated into it via shared insertion sequences or other repetitive sequences present in both the DNA of both the invader and the host.
Our results imply a certain degree of coevolution between CRISPR-Cas systems and the core DNA repair machinery, which is surprising given that CRISPR-Cas systems in halophilic archaea tend to be plasmid-encoded and have been frequently transferred by horizontal gene transfer, and that some Haloferax strains, such as H. gibonsii LR2-5 do not encode CRISPR-Cas systems (Mizuno et al. 2019;Tittes et al. 2021). It will be interesting to compare the DNA damage response of such naturally CRISPR-deficient strains to their CRISPR-positive relatives, which will help address the question of under which conditions, if any is the net contribution of CRISPR-Cas to DNA repair positive rather than negative and whether it can select in favor of CRISPR-Cas retention even under low viral pressures. However, since having an intact CRISPR-Cas reduces the frequency of HR (Fig. 6), it can reduce the rates of horizontal gene transfer in this and other species of halophilic archaea (Naor et al. 2012), and thus have undesirable evolutionary consequences.

Strains, plasmids, and oligonucleotides
Strains are shown in Table S1, plasmids in Table S2, and oligonucleotides in Table S3.

Culture conditions
H. volcanii strains were routinely grown aerobically at 45 • C in either Hv-YPC (rich medium) or in Hv-Ca / Hv-Ca + (minimal medium) as described in (Allers et al. 2004). Hv-YPC containing (per liter) 144 g of NaCl, 21 g of MgSO 4 ·7H 2 O, 18 g of MgCl 2 ·6H 2 O, 4.2 g of KCl, and 12 mM Tris HCl (pH 7.5). For solid media, agar (Difco) was added at a concentration of 15 g per liter. 0.5% (w/v) yeast extract (Difco), 0.1% (w/v) peptone (Oxoid), and 0.1% (w/v) casamino acids (Difco) were added, and the medium was autoclaved. After cooling, CaCl 2 was added to a final concentration of 3 mM. Casamino acids medium (Hv-Ca) was made in a similar manner, except that yeast extract and peptone were excluded, casamino acids were added to a final concentration of 0.5% (w/v), 0.8 mg of thiamine, and 0.1 mg of biotin were added per liter. Enhanced Ca (Hv-Ca + ) contained the same concentration of salts as Hv-Ca, except that Tris HCl (pH 7.5) was added to a concentration of 42 mM. After autoclaving and cooling, 4.25 ml of a sodium DLlactate solution (60%, w/v), 3.83 g of disodium succinic acid · 6H2O, 0.25 ml of glycerol, 5 ml of a 1 M NH 4 Cl solution, 6 ml of a 0.5 M CaCl 2 solution, 2 ml of 0.5 M potassium phosphate buffer (pH 7.5), 1 ml of trace elements solution (Mevarech and Werczberger 1985), 0.8 mg of thiamine, and 0.1 mg of biotin were added per liter. When required, thymidine was added to a concentration of 40 μg/ml, and uracil was added at a concentration of 50 μg/ml. For p.tnaA promoter activation, tryptophan was added to a final concentration of 2 mM (Large et al. 2007;Allers 2010;Allers et al. 2010). For pop-out selection medium, 5-FOA was added to a concentration of 50 μg/ml.
Escherichia coli strains DH12S and ns2626 were grown aerobically at 37 • C in LB medium, containing 200 μg/ml of ampicillin when required. The latter strain was used to prepare unmethylated plasmid DNA for efficient transformation of H. volcanii (Holmes, Nuttall, and Dyall-Smith 1991).

Plasmid cloning
pUG334 was generated as follows. Upstream and downstream constructs to cas3 gene were amplified from H. volcanii genomic DNA using IS225/6 (up) and IS227/8 (down) primers. The primers contained additional restriction sites for NotI (5 -up), BamHI (3down), and additional 20 reverse complement bases (3 -up, 5down). Constructs were fused by overlapping polymerase chain reaction (PCR), then the product was digested with restriction enzyme and ligated to pTA131 (digested with NotI and BamHI).
pUG427 was generated as follows. Upstream construct to HVO_CRISPR_2 locus and downstream construct to HVO_CRISPR_3 locus were amplified from H. volcanii genomic DNA using IS307/9 (up) and IS308/10 (down) primers. The primers contained additional restriction sites for BamHI (5 -up), HindIII (3 -down) and additional 19 reverse complement bases (3 -up, 5 -down). Constructs were fused by overlapping PCR, then the product was digested with restriction enzyme and ligated to pTA131 (digested with BamHI and HindIII).
pUG523 was generated as follows. Upstream and downstream constructs of cas3 gene were amplified from H. volcanii genomic DNA using IS596/7 (up) and IS598/9 (down) primers. The primers contained additional 20 homologous bases to pTA131 cloning site (5 -up, 3 -down) and additional 15 reverse complement bases to each other (3 -up, 5 -down). pTA131 was amplified and linearized using IS600/1 primers. The purified products were cloned together through gibson assembly. pUG536 was generated as follows. Upstream and downstream constructs of cas4 gene were amplified from H. volcanii genomic DNA using IS590/1 (up) and IS592/3 (down) primers. The primers contained additional 20 homologous bases to pTA131 cloning site (5 -up, 3 -down) and additional 15 reverse complement bases to each other (3 -up, 5 -down). pTA131 was amplified and linearized using IS594/5 primers. The purified products were cloned together through gibson assembly. pUG636 was generated by GeneScript. Synthetic fragment of the cas3 H. volcanii gene contained the HD63-64AA double mutation flanked by NdeI -ATG -3x flag from the 5 terminus and TGA -NotI from the 3 terminus cloned into pTA927 using restriction enzymes and ligation.
pUG734 was generated as follows. Cas3 helicase dead gene was amplified from pTA927-cas3-D444A-Flag-N using GM23/4 primers. The primers contained additional restriction sites for BamHI (5 ) and EcoRI (3 ). Product was digested with restriction enzyme and ligated to pTA131 (digested with BamHI and EcoRI).
pUG753 was generated as follows. Upstream construct of cas3 gene was amplified from H. volcanii genomic DNA using GM29/30 (up) primers and downstream construct of cas3 nuclease dead was amplified from pUG636 using GM31/2 (down) primers. The primers contained additional restriction sites for HindII (5 -up) and BamHI (3 -down). Constructs were fused by overlapping PCR, then the product was digested with restriction enzyme and ligated to pTA131 (digested with HindIII and BamHI).

Construction of H. volcanii strains
Strain construction was performed according to the pop-in/popout protocol described in (Bitan-Banin, Ortenberg, and Mevarech 2003;Allers et al. 2004). In this method, the upstream and downstream flanking regions of the sequence to be exchanged are amplified by PCR and cloned together into the "suicide plasmid" pTA131, that cannot replicate autonomously and carries the pyrE2 selectable genetic marker. The plasmids are then transformed into H. volcanii pyrE2 mutants, and the transformants, in which the plasmids have been integrated into the chromosome, are selected for on plates that lack uracil ("pop-in"). Upon counterselection on plates containing uracil and 5-FOA, the only cells that survive are those in which the integrated plasmids have been excised by spontaneous intrachromosomal homologues recombination ("pop-out"), either restoring the WT gene or resulting in allele exchange. The "pop-out" strains were screened using pairs of primers located upstream and downstream to the desired deletion site and the sequence validated by Sanger sequencing. cas genes, cas3-KO, cas3, cas4, cas3 helicase dead, and cas3 nuclease dead were constructed using the pop-in pop-out methodology as described above. Parental strains and plasmids used for the constructions are listed in Table S1 and S2.

Transformation
Transformation of H. volcanii was carried out using the polyethylene glycol 600 method as described in (Cline et al. 1989;Allers et al. 2004). A volume of 1.5 ml liquid culture were grown in Hv-YPC to OD 600nm of 1.5, and then centrifuged at 3 600 × g for 5 min. The supernatant was discarded, and the cells were resuspended in 200 μl spheroplasting solution (1 M NaCl, 27 mM KCl, 50 mM Tris HCl PH 8.5, and 15% sucrose) and incubated at room temperature for 5 min. A volume of 20 μl of 0.5 M EDTA were added and cells were incubated at room temperature for 10 min. A volume of 10 μl of purified plasmid DNA were mixed with 15 μl spheroplasting solution and 5 μl of 0.5 M EDTA, and added to the cells, followed by incubation of 5 min at room temperature. Subsequently, 250 μl of PEG solution (60% PEG 600 in spheroplasting solution) was added, and cells were incubated for 30 more minutes at room temperature. Following the incubation, 1 ml of regeneration solution (3.4 M NaCl, 175 mM MgSO 4 , 34 mM KCl, 5 mM CaCl 2 , 50 mM Tris HCl pH 7.5, and 15% sucrose) was added and cells were centrifuged at 6 000 rpm for 7 min. The supernatant was discarded, and cells were resuspended in Hv-YPC medium supplemented with 15% sucrose and left to incubate without shaking overnight at 37 • C. The cultures were then transferred to a 37 • C shaker and left for an incubation of three more hours, then washed and plated on selective media.
Transformation of E. coli was carried out by using a standard electroporation protocol (Sambrook, J. and Russell 2001).

DNA damage assays
For UV radiation assays, cultures were grown in Hv-YPC broth to a mid-log phase (OD 600nm of 0.7), serially diluted (10-fold) in 18% saltwater and 20 μl aliquots spotted on Hv-YPC plates. After drying, plates were exposed to UV-C radiation (50-250 J m −2 ) in a Hoefer TM UVC 500 ultraviolet crosslinker. Control plates were not exposed to UV radiation. Plates were then covered by aluminum foil to prevent the penetration of visible light.
For chemical mutagenesis assays, mid-log phase cultures were divided into 2 ml aliquots and methyl methanesulphonate (MMS) was added (0.1-0.7 mg/ml final concentration). Cultures were returned to 45 • C for 1 h, serially diluted (10-fold) and 20 μl aliquots were spotted on Hv-YPC plates.
In both UV and MMS exposure experiments, survivor colonies were counted after 5-10 days of incubation. Survival rate was calculated by dividing the number of colonies grown on the irradiated plates by the number of colonies grown on the control plates.
For the post-MMS recovery experiments, the MMS was removed after the first hour by centrifugation followed by resuspension in an equal volume of fresh Hv-YPC medium (X2), and the cultures were left to recover in 45 • C incubation for different time points.

Growth curves
To compare the growth of the H. volcanii strains and mutants, each sample was grown over-night in appropriate media at 45 • C to the mid-log phase and then diluted to a fresh medium to OD 600nm of 0.1. The growth curves were carried out in 96-well plates at 45 • C with continuous shaking, using the Biotek ELX808IU-PC microplate reader. Optical density was measured every 30 min at a wavelength of 595 nm.

DNA extraction
Total genomic DNA extraction of H. volcanii was done by DNA spooling protocol described in (Allers et al. 2004). Saturated culture grown in Hv-YPC broth, with or without MMS, was centrifuged at 6 000 rpm for 5 min and resuspended in 200 μl of ST buffer (1 M NaCl, 20 mM Tris HCl pH 7.5). Then 200 μl of lysis solution (100 mM EDTA pH 8.0, 0.2% SDS) was added to lyse the cells. The aqueous phase was overlaid with 1 ml 70% ethanol and the DNA was spooled onto a glass capillary until liquid was homogeneous and clear. The DNA washed three times by transferring the spooled DNA to a microcentrifuge tube with 1 ml fresh 70% ethanol, and then allowed to dry. DNA was then solubilized in 100 μl TE (10 mM Tris HCl pH 7.5, 1 mM EDTA pH 8.0) containing 0.1 mg of RNase A per ml.
Plasmid extraction from E. coli was done by Sigma-Aldrich's GenElute™ Plasmid Miniprep Kit.

DNA damage quantification
Damaged DNA (by MMS) and undamaged DNA extractions were sequenced by RAre DAmage and Repair (RADAR) sequencing method (Zatopek et al. 2019). Briefly, PacBio libraries are created from isolated genomic DNA by shearing into 2 kb fragments and ligation of PacBio SMRTbell adapters. Nick translation is performed on PacBio libraries, in which hAAG (human alkyladenine DNA glycosylase), in combination with Endonuclease IV, nick the DNA backbone at 3 mA and 7 mG (caused by the MMS). Bst FL DNA polymerase, a dNTP pool containing dTTP, dGTP, d6mATP, and d4mCTP, Taq DNA ligase and NAD+, were used to replace the nick with a patch of modified bases. Nick translated libraries are sequenced using PacBio SMRT sequencing on Sequel instrument, followed by downstream analysis to determine location and frequency of patches per million bases sequenced. The amount of the total sequenced patches is equal to the amount of the DNA lesions present in the sequenced sample.

Measuring transformation / recombination efficiencies
Transformation efficiencies were calculated for H. volcanii strains after transforming them with pTA927, a replicating plasmid or with pTA131+metX-volcanii, "suicide" integrative plasmid (both carrying the pyrE2 marker for selection). Transformation efficiency was calculated as the number of successfully transformed CFUs with the replicating plasmid (pTA927) grown on the selective plates (Hv-Ca) divided by the number of CFU grown on nonselective plates (Hv-YPC). Pop-in transformation could be achieved only by a recombination event between the plasmid metX gene and the chromosomal metX, leading to integration of the plasmid into the H. volcanii chromosome. Pop-in transformation efficiency was calculated as the number of successfully transformed CFUs with the integrative plasmid (pTA131+metX-volcanii) grown on the selective plates (Hv-Ca) divided by the number of CFU grown on nonselective plates (Hv-YPC). Recombination efficiency values were obtained by dividing the transformation efficiency obtained with the integrative plasmid (pTA131-metX-volcanii) divided by the transformation efficiency with a replicating plasmid (pTA927).