Deinococcus species, especially Deinococcus radiodurans has been described as the most radiation resistant bacteria on earth. It can withstand up to 5 kGy of ionizing radiation (IR) with no loss of viability (Moseley and Mattingly, 1971; Battista, 1997). Exposure to such high doses of radiation generates hundreds of double strand breaks (DSBs), thousands of single strand breaks (SBDs) and greater than 1,000 sites of base damage, which D. radiodurans can repair within hours (Battista, 1997; White et al., 1999; Slade and Radman, 2011; Daly, 2012). The mechanism of this extreme resistance is not known but generally believed to be due to a combination of factors such as its densely packed genome, high levels of pigments (e.g., deinoxanthin) in the membrane, a high level of intracellular Manganese, and an extremely efficient DNA repair mechanism.

The genome of D. radiodurans encodes several DNA glycosylases with homologs in all three domains of life involved in the Base Excision repair (BER) pathway (Makarova et al., 2001; Daly, 2012). Endonuclease III (EndoIII) is a bifunctional DNA glycosylase, belonging to the Helix-hairpin-Helix (HhH)-GDP superfamily of DNA glycosylases, which removes a broad range of oxidized DNA lesions like thymine glycol (Tg) and 5-hydroxycytosine (5OHC) in the BER pathway (Dizdaroglu, 2015; Sarre et al., 2019). The most common oxidized lesions are 5OHC and 5-hydroxyuracil (5OHU) which is generated by oxidation of cytosine and Tg which is generated through either oxidation of thymine or oxidation and deamination of 5-methylcytosine. Abasic sites (AP-sites) is caused by loss of a nucleobase, and dihydrothymine (DHT) and dihydrouracil (DHU) originate from IR damage of thymine and cytosine (Dizdaroglu, 1985; Breen and Murphy, 1995). The main substrate of EndoIII enzymes is Tg. This lesion is not mutagenic, but cytotoxic and has been shown to affect cell division by blocking DNA polymerases due to the perturbation induced in the DNA structure (Ide et al., 1985; Clark and Beardsley, 1986; Wallace, 2002; Dizdaroglu, 2015).

In chromosome I, D. radiodurans possesses three genes encoding for EndoIII-like glycosylases DR2438, DR0289 and DR0928 (endoIII1, endoIII2 and endoIII3, respectively). The gene products of endoIII1 and endoIII3 (DrEndoIII1 and DrEndoIII3) were initially considered to be of archaeal origin, while endoIII2 (DrEndoIII2) was found to be closest to yeast EndoIII (Makarova et al., 2001). Homologs of these three enzymes have been found in other Deinococcus species except in D. peraridilitoris which lacks the endoIII3 (Lim et al., 2019; Sarre et al., 2019). Later, sequence alignments of the three EndoIII enzymes with homologs found in other bacterial, archaeal and eukaryotic species led to the hypothesis that these enzymes could be of bacterial origin and that DrEndoIII3 is specific to the Deinococcaceae genus (Sarre et al., 2015).

Structure determination of DrEndoIII1 and 3, and homology modelling of DrEndoIII2 revealed that they consist of two α-helical domains (A and B) divided by a positively charged DNA binding cleft, which contains the highly conserved HhH motif and the catalytic residues (aspartate and lysine). Domain A harbors a [4Fe-4S] cluster and a FeS cluster loop and domain B two DNA intercalating loops (DIL1 and DIL2) which are involved in the stabilization of DNA when the damaged base is flipped out of the DNA into the active site pocket. Even though all DrEndoIII homologs can bind undamaged and damaged DNA with various affinities, only DrEndoIII1 and DrEndoIII2 demonstrated activity towards the common EndoIII substrates (Sarre et al., 2015, 2019). DrEndoIII2 revealed catalytic properties that exceed those of E. coli EndoIII (EcEndoIII), while DrEndoIII1 acts mostly as a monofunctional DNA glycosylase with lower activity, being also able to process single stranded DNA (ssDNA) (Sarre et al., 2019). Mutational studies have been conducted to unravel the catalytic differences between these enzymes. Substitutions introduced in DIL1 and DIL2 of DrEndoIII1, revealed that these mutations led to a decreased activity due to a lower substrate affinity (Rollo et al., 2022). In the case of DrEndoIII3, none of the substitutions performed in the active site pocket led to an induction of EndoIII type activity (Rollo et al., 2022).

To unravel the role of these three enzymes in the repair of oxidative damages, transcriptome profiling and EndoIII knockout (KO) mutants have been generated and analyzed (Liu et al., 2003; Chen et al., 2007; Hua et al., 2012). However, several questions remain to be answered from these studies. Cells irradiated with a total dose of 15 kGy revealed a down regulation of endoIII1 with no alterations of the expression of endoIII2 and endoIII3 (Liu et al., 2003). On the other hand, only endoIII1 and endoIII3 were observed to be damage activated, by gene up regulation, immediately after exposure to 2 kGy of IR (Chen et al., 2007). DrEndoIII KO mutants (Hua et al., 2012) showed no significant difference between the mutants and wild type (WT) when exposed to both IR and hydrogen peroxide (H₂O₂). This has also been observed in EndoIII KO mutant in E. coli, which was shown to be as resistant as the WT (Cunningham and Weiss, 1985). Moreover, in D. radiodurans each EndoIII KO mutant showed elevated levels of spontaneous mutation which were increased in the triple KO mutant leading to the hypothesis that the three EndoIII’s are not only required to repair DNA damage but also have non-overlapping substrates (Hua et al., 2012).

To shed new light on the role of these enzymes at a cellular level in D. radiodurans, endoIII1, endoIII2, and endoIII3 single KO mutants were constructed and analyzed after being subjected to various oxidative stress related conditions. Here we report the characterization of these strains in terms of their resistance to ionizing and UV-C radiation and exposure to H₂O₂ and methyl viologen (MV) (also known as the common herbicide Paraquat). In the latter case, the effect of stress on the expression level of the endoIII genes was also analyzed. Our data suggests that the KO mutants are as resistant as the WT to IR, UV-C and H₂O₂ exposure, however upon exposure to MV they were revealed to be more resistant than the WT which could be attributed to the observed upregulation of EndoIII homologs gene expression levels. Due to an increasing amount of Deinococcus sequenced genomes, we also performed a phylogenetic analysis in order to increase our insight regarding the origin of the genes.

Based on more available genome sequenced Deinococcus strains, we started this work by performing a phylogenetic analysis of the three Endonuclease III enzymes among these strains. Based on an analysis of 17 fully genome sequenced Deinococcus species, our data suggests that DrEndoIII1 and DrEndoIII3 are archaeal type enzymes as previously suggested (Makarova et al., 2001). We could also hypothesize that DrEndoIII3 may not be a specific variant for the Deinococcus species, however this requires further analysis and validation (Figure 1). Regarding DrEndoIII2 our data suggest that it is more closely related to bacteria than to yeast.

The survival of the ∆EndoIII strains upon exposure to IR revealed no significant differences when compared to the WT (Figure 2A). This was surprising, especially in the case of ΔEndoIII2, since DrEndoIII2 is the apparent main EndoIII enzyme in D. radiodurans (Sarre et al., 2019). However, similar results from other IR experiments have been observed previously (Cunningham and Weiss, 1985; Hua et al., 2012), even in the case of multiple KO strains. In these cases, it was suggested that other repair pathways and enzymes from the BER pathway are able to repair the lesions that are introduced into the DNA under these conditions. It is likely that this is the case also for our experiment, knowing that IR causes both direct and indirect damage to DNA, leading to base modifications (e.g., 8-oxoG and Tg) and strand breaks which can be dealt with both in BER and homologous recombination (Cox and Battista, 2005; Slade and Radman, 2011; Krokan and Bjørås, 2013; Dizdaroglu, 2015; Timmins and Moe, 2016). We can also not exclude the possibility that the ROS scavenging system (e.g., catalases and SOD) and other irradiation protection mechanisms (e.g., high level of pigments in the membrane) is protecting D. radiodurans from severe damage in this case.

Exposure of our KO strains to UV-C radiation revealed no observable response and resistance differences to D. radiodurans WT (Figure 2B). The most abundant DNA lesions generated by UV-C radiation, are the mutagenic and cytotoxic cyclobutane pyrimidine dimers (CPDs) and pyrimidine–pyrimidone (6-4) photoproducts (6–4PPs) (Battista, 1997; Sinha and Häder, 2002; Slade and Radman, 2011; Timmins and Moe, 2016), which are repaired via the NER and the UV damage endonuclease (UVDE) pathways (Tanaka et al., 2005; Blasius et al., 2008; Slade and Radman, 2011; Timmins and Moe, 2016). UV radiation also generates other photoproducts and lesions normally associated with oxidative damages in DNA such as pyrimidine hydrate, Tg, dipurine adducts, ROS (mainly singlet oxygen (¹O₂), protein oxidation and 8-oxoG) (Demple and Linn, 1982; Zhang et al., 1997; Friedberg et al., 2006; Krisko and Radman, 2010; Chatterjee and Walker, 2017). Thus, we suggest that the UV-damages generated in this experiment might have been repaired in the NER and UVDE pathways and that the remaining oxidative damages might have been efficiently repaired by the BER DNA glycosylases (8-oxoG DNA glycosylase and the EndoIII’s) even in the absence of one of the DrEndoIII enzymes.

In order to analyze the response of the D. radiodurans strains to oxidative stress, they were exposed to either H₂O₂ or MV (Figure 3). Both agents are responsible for damage of nucleobases, nucleotides (oxidized pyrimidines and purines (eg. Tg and 8 oxo-G)) (Demple and Linn, 1982; Petrovská and Dušinská, 1999; Slade and Radman, 2011), and proteins containing iron–sulfur and heme groups (Munteanu et al., 2015). Superoxide radicals can be converted to H₂O₂ and consequently to Hydroxyl radicals leading to major DNA modifications (Maynard et al., 2009; Dizdaroglu, 2015; Gonzalez-Hunt et al., 2018). Both H₂O₂ and superoxide radicals are generally detoxified by different scavengers systems such as superoxide dismutase, catalases and Mn complexes (Culotta and Daly, 2013). Here, exposure of the KO and WT strains to H₂O₂ demonstrated that they possessed similar sensitivity to this oxidation damaging agent (Figure 3A). This response have also been observed previously of other D. radiodurans and E. coli EndoIII KO mutants (Cunningham and Weiss, 1985; Hua et al., 2012).

Regarding the MV stress experiments, our data clearly revealed an effect on the growth demonstrating a reduction in both growth rate and final OD on all strains tested. As previously observed, the growth of the WT was severely affected, leading to a decrease of the growth before entering a lag phase (Santos et al., 2019) (Figure 4B). The three DrEndoIII KO strains proved to be more resistant being able to grow to a higher final optical density upon addition of the damaging agent than the WT (> 17x). Out of the three KO strains, ∆EndoIII1 was most severely affected by the addition of MV, reducing the final OD₆₀₀ from 10 to almost 4 and a reduction in the growth rate from 0.2563 to 0.175 h⁻¹ (Figure 4B; Supplementary Table S1). ∆EndoIII2 showed higher resistance levels than the other mutants (Figure 4B). This was not expected since DrEndoIII2 is considered the main EndoIII enzyme responsible for repair of oxidation damaged pyrimidines in DNA (Sarre et al., 2019). Although ΔEndoIII3 unveiled to be the strain with the lowest growth rate under normal conditions it was also the strain with the smallest difference in growth rates when comparing both growth conditions (Supplementary Table S1). This result suggests that DrEndoIII3, in general, is important for oxidative stress resistance in D. radiodurans despite its lack of classic EndoIII substrate specificity (Sarre et al., 2019).

A gene expression analysis was carried out to evaluate the importance of the three DrEndoIII homologs in oxidative stress resistance (MV) at the gene level (Figure 5). The results revealed that endoIII1 is expressed at similar levels in all strains and conditions suggesting that it is important during cell growth (Figure 5A). These results do not correspond to findings in other transcriptome experiments where irradiation was used to induce oxidative stress. In one case it was shown that the expression levels of endoIII1 were down regulated upon exposure to 15 kGy of IR compared to control conditions (Liu et al., 2003). In the other case it was shown that endoIII1 was DNA-damage inducible immediately upon exposure to 2 kGy of IR (Chen et al., 2007). However, the stress induction in both cases is different and is thus not directly comparable to our experiment. Even so, it indicates that the level of expression may be different depending on the damage that is inflicted into the DNA. In our case, a generally high expression level of endoIII1 was observed in both normal and stressed condition in WT as well as KO strains, indicating a generally high importance of this protein during cell growth.

Opposite to endoIII1, both endoIII2 and endoIII3 genes are upregulated in the KO strains under oxidative stress conditions compared to the WT (Figures 5B,C). We even observed that endoIII2 is overexpressed surpassing the expression of the housekeeping genes when endoIII1 is knocked out (ΔEndoIII1) under normal conditions and that endoIII3 reaches the expression levels of the housekeeping genes. This is also in contrast to what has been reported previously. In these studies the level of expression of the endoIII2 and endoIII3 genes did not change (Liu et al., 2003), and the endoIII2 gene was not significantly induced upon IR exposure (Chen et al., 2007). Again, the stress induction in those studies is not directly comparable to our study, and we can only assume that the level of damage induced by exposure of the cells to MV is not as dramatic as for IR. However, this brings support to our hypothesis about endoIII1 being in general very important for D. radiodurans in both normal and stressed conditions, suggesting that this enzyme may play additional roles in the cells which goes beyond DNA repair. This corresponds well to our knowledge that DrEndoIII1 possesses specificity for oxidation damages in single stranded DNA in addition to its primary function as a DNA glycosylase in the presence of Tg in double stranded DNA substrates (Sarre et al., 2019). Multiple roles of DNA glycosylases have been observed previously, e.g., of human uracil DNA N-glycosylases (UNG) which is involved in both BER and replication (Krokan et al., 2002). Due to its activity on single stranded DNA, it is thus not unlikely that DrEndoIII1 is involved in DNA replication repair, however, this will have to be explored in future experiments.

Here we have provided further insights into the role/importance of the three Endonuclease III enzymes in D. radiodurans when this bacterium is exposed to oxidative stress. Despite the lack of observed in vitro activity/function for DrEndoIII3 from previous studies, our data suggests that this gene product is extremely important for the bacteriums’ oxidation damage resistance. This is noticeable in all the KO strains, with observed upregulation of endoIII3 in both ΔEndoIII2 and ΔEndoIII1 KO strains. Also, in ΔEndoIII2 the overexpression of endoIII3 coupled with endoIII1 was revealed to be the best combination to cope with oxidative stress and is a good example of how the lack of one of the endoIII genes may be compensated for by overexpression of the other two. Moreover, the observation that the expression of endoIII1 is important in all strains in both normal and stressed conditions, suggests that this enzyme may play multiple roles in D. radiodurans. The role of both DrEndoIII1 and DrEndoIII3 will therefore be further explored in future studies. Thus, we conclude that all three DrEndoIII enzymes are important in the resistance mechanisms of D. radiodurans towards oxidative stress, however, the contribution of the individual enzymes in this process is yet to be fully disclosed. One possible approach to reveal new information in this regard is to perform a comparative gene clustering analysis of each EndoIII group using available genome data of sequenced Deinococcus strains.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

FR: Investigation, Methodology, Visualization, Writing – original draft, Formal analysis. GM: Investigation, Visualization, Writing – review & editing, Formal analysis. AG: Writing – review & editing, Methodology, Supervision, Formal analysis. SI: Investigation, Writing – review & editing. PS: Methodology, Writing – review & editing, Supervision, Visualization. CR: Supervision, Writing – review & editing, Conceptualization, Funding acquisition. EM: Conceptualization, Funding acquisition, Supervision, Writing – review & editing, Project administration.

The author(s) declare financial support was received for the research, authorship, and/or publication of this article. This work was supported by FCT—Fundação para a Ciência e a Tecnologia, I.P., through MOSTMICRO-ITQB R&D Unit (UIDB/04612/2020, UIDP/04612/2020) and LS4FUTURE Associated Laboratory (LA/P/0087/2020), Centre National de la Recherche Scientifique and the University Paris-Saclay, research projects PTDC/QUI/BIQ/100007/2008, PTDC/BBBBEP/0561/2014, PTDC/BIA-BFS/31026/2017, PTDC/BIA-BQM/31317/2017, post doc fellowship SFRH/BPD/97493/2013 (EM), PhD fellowships SFRH/BD/132966/2017 and COVID/BD/152598/2022 (FR) and SFRH/BD/06723/2020 (AGG). Funding is also acknowledged for the TIMB3 and IMpaCT project, European Union’s Horizon 2020 research and innovation program, under grant agreement No 810856 and No 857203, respectively, and the Research Council of Norway, project number: 183626.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.