Amino acid sequences of RuvC proteins from 35 species were obtained from the NCBI website.¹ A phylogenetic tree was constructed using the complete amino acid sequences of 35 proteins from SynRuvC and other species. Protein amino acid sequences were aligned using the ClustalX1.81 program and phylogenetic trees were constructed by neighbour-joining (NJ) method using MEGA6 software.

All oligonucleotides utilized in this work were bought from Sangon (Shanghai, China), and the sequences are given in Supplementary Table 1. For all DNA substrates, one strand of DNA is fluorescently labeled at the 5′ end with 6-carboxyfluorescein (FAM). DNA annealing was performed by mixing a 1:2 ratio of labeled and unlabeled oligonucleotides in the annealing buffer (20 mM Tris-HCl [pH 8.0], 50 mM NaCl), boiled in a water bath, and cooled to room temperature overnight (Qin et al., 2022).

The WT and other mutant variants of SynRuvC were expressed and purified in a similar way. In brief, the gene encoding SynRuvC (KEGG ID: sll0896) was amplified by PCR (polymerase chain reaction) and cloned into the expression vector pET15b with an N-terminal His-SUMO (small ubiquitin-like modifier) tag. The constructed recombinant vector was transformed into BL21(DE3) competent cells and grown on LB plate with 100 mg/L ampicillin. Transformed BL21(DE3) clones were cultured to an optical density of 0.4 to 0.6 at 600 nm in an LB medium containing 100 mg/L ampicillin at 37°C. IPTG (isopropyl β-D-1-thiogalactopyranoside) at a final concentration of 0.5 mM was used to stimulate protein expression at 18 °C for 18 h.

After harvesting, cells were resuspended in lysis buffer (20 mM Tris-HCl [pH 8.0], 0.5 mM NaCl, 10% [vol/vol] glycerol, and 5 mM imidazole), disrupted by sonication, and centrifuged at 6200 g at 4°C for 60 min. Additionally, the supernatants were injected into a Ni-NTA, desalting, and ion exchange (HiTrap Q HP column, GE Healthcare) columns utilizing AKTA pure (GE Healthcare). SDS-PAGE was used to confirm each fraction before being concentrated, aliquoted, and kept at −80°C.

DNA binding assay was performed according to a previously reported approach (Sun et al., 2022), with some modifications. 50 nM 5′-FAM labeled substrates were mixed with various concentrations of SynRuvC in a 10 μL reaction buffer containing 20 mM HEPES (pH 7.5), 50 mM NaCl, 1 mM DTT, 10% (vol/vol) DMSO, and 1% (vol/vol) glycerol. Binding reactions were incubated at 37°C for 60 min, and the products were resolved by 5% native-PAGE (native polyacrylamide gels) in 1× Tris-Acetate-EDTA buffer. Gels were scanned by a fluorescent imaging system (Tanon 5200Multi, China).

50 nM 5′-FAM labeled substrates were mixed with various concentrations of SynRuvC in a 10 μL reaction buffer containing 20 mM Tris-HCl (pH 7.5), 50 mM NaCl, 1 mM DTT, 10% (vol/vol) DMSO, 10 mM Mn²⁺, and 1% (vol/vol) glycerol. The reactions were incubated at 37°C for 60 min before being stopped with a solution containing proteinase K, SDS, and EDTA (Qin et al., 2022). All products were resolved by 10% native-PAGE in 1× Tris-Acetate-EDTA buffer. Gels were scanned by a fluorescent imaging system (Tanon 5200Multi, China). For the metal ion dependence experiment, MnCl₂ was replaced with various concentrations of MgCl₂, ZnCl₂, CaCl₂, CuCl₂, CoCl₂, and NiCl₂. For the temperature effects experiment, reactions were carried out at the specified temperatures for 60 min. For the heat stability experiment, SynRuvC alone was pre-incubated at various temperature for 15 min in the reaction buffer and further incubated with DNA substrates at 37°C for 60 min. For the pH dependence experiment, Tris-HCl pH 7.5 was replaced by different pH of Tris-HCl. For the effect of salt concentrations experiment, 50 mM NaCl was replaced with various concentrations of KCl and NaCl. For the time-course analysis, reactions were carried out at 37 °C for different min.

For the determination of cleavage sites, each DNA substrates (50 nM) with a uniquely 5′-strand FAM-labeled arm were incubated with SynRuvC (2 μM) at 37°C for 60 min in reaction buffer, as mentioned above. The reaction was stopped by a solution containing proteinase K, SDS, and EDTA, and the products were resolved by 15 or 18% denaturing PAGE (containing 8 M urea) in 0.5× TBE (Tris-borate-EDTA) buffer. Gels were scanned by a fluorescent imaging system (Tanon 5200Multi, China). GA ladders of each labeled oligonucleotide produced by the Maxam-Gilbert technique were put alongside to serve as markers.

5′-strand FAM-labeled 4Jhs (50 nM) was incubated with or without SynRuvC (2 μM) at 37°C for 60 min in reaction buffer (200 μL) and then stopped by phenol extraction (Komori et al., 2000). These products were precipitated using ethanol and suspended in 1× T4 DNA ligation buffer. The products of the previous step were equally divided and incubated at 16°C for 8 h with or without T4 DNA ligase. The reaction products were resolved by 15% denaturing PAGE in 0.5× TBE buffer. Gels were scanned by a fluorescent imaging system (Tanon 5200Multi, China). Moreover, we used 59 nt and 70 nt of 5′-strand FAM-labeled DNA oligonucleotides as Markers.

The tertiary structure of SynRuvC, predicted by AlphaFold2, was downloaded from Uniprot (Q55506). The secondary structure of 5′-overhang was modeled by BIOVIA Discovery Studio Visualizer², and modified using PyMOL. Docking simulations of SynRuvC and 5′-overhang was performed with GRAMM³ and the optimal binding mode was chosen based on the lowest docking energy (Majumder et al., 2012). PyMOL⁴ was used to display the three-dimensional figures and interactions of SynRuvC–5′-overhang (Laskowski and Swindells, 2011).

The cyanobacterium Synechocystis sp. PCC6803 was obtained from the Freshwater Algae Culture Collection of the Institute of Hydrobiology, Chinese Academy of Sciences. Synechocystis sp. PCC6803 strains were grown at 30°C in liquid BG-11 medium at pH 7.5 with 20 mM HEPES-NaOH on a rotary shaker (Chen et al., 2014). For growth on plates, 1.5% (weight/vol) agar were added to BG11. Cells were cultured at a light intensity of 40 μmol photons m^–2 s^–1 using cool-white fluorescent lamps. Mutant and overexpressed strains were maintained on solid BG11 agar plates supplemented with 50 μg/mL kanamycin. The optical density of the cells at 730 nm was measured to monitor growth.

A conventional genetic approach was used to construct mutant and overexpression strains of Synechocystis sp. PCC6803 (Cameron and Pakrasi, 2010; Uchiyama et al., 2020). The km^r gene replaced the coding sequences of synruvC (sll0896) through homologous recombination. Firstly, using appropriate primers, the upstream and downstream regions of the sll0896 gene were amplified by PCR (Supplementary Table 1). Next, upstream and downstream PCR products were cloned into the pMD19T plasmid on either side of the km^r gene to generate pMD19T-sll0896-delete. As shown in Supplementary Figure 10, the pMD19T-sll0896-overexpress plasmid was produced in the same way as the pMD19T-sll0896-delete plasmid. We used the strong bidirectional promoter biPpsbA2 to overexpress synruvC gene expression (Zheng et al., 2023). Finally, wild type Synechocystis sp. PCC6803 was transformed using pMD19T-sll0896-delete and pMD19T-sll0896-overexpress to create mutant and overexpress SynRuvC strains, respectively. All mutant and overexpressed RuvC strains of Synechocystis sp. PCC6803 were identified by PCR, as shown in Supplementary Figure 10. Similarly, WT of Synechocystis sp. PCC6803 strains were also inserted the km^r gene. Supplementary Table 1 lists all of the primers utilized in the study.

The Synechocystis sp. PCC6803 strains were cultivated at 30°C in liquid BG-11 medium supplemented with 50 μg/mL kanamycin until reaching the mid-exponential growth phase. Subsequently, bacterial cells were harvested through centrifugation. Total RNA was isolated utilizing RNAprep Pure Cell/Bacteria Kit (Tiangen Biotech, Beijing, China) and treated with DNase I (Sigma-Aldrich). RNA purity and concentration were assessed using both gel electrophoresis and a spectrophotometer (NanoDrop, Thermo Scientific). And then, RNAs were reverse transcribed into cDNA using a Reverse Transcription kit (TransGen Biotech, Beijing, China) with random primers. Quantitative real-time PCR was performed with SYBR FAST qPCR Kit (Kapa Biosystems, USA) on a LightCycler^® 96 System (Roche), and the data are presented as the mRNA accumulation index (2-^ΔΔCt). Data were normalized to WT strains (set as 1). Rnpb was used as the housekeeping gene (Zhang et al., 2009). All primers are listed in Supplementary Table 1.

The strains of Synechocystis sp. PCC6803 (WT, sll0896 knockdown mutant cells, and sll0896 overexpressed cells) were cultivated to mid-logarithmic phase in BG11 (Sigma C#3061) medium supplemented with 50 μg/mL of kanamycin. Subsequently, the cultures were standardized to an OD₇₃₀ = 1. Ten-fold serial dilutions were spotted onto fresh BG11 plates containing 50 μg/mL kanamycin and supplemented with or without 2 or 4 mM MMS (methyl methane sulfonate) (Beam et al., 2002; Nowosielska et al., 2006; Blanco et al., 2010). These plates were incubated at a light intensity of 40 μmol photons m^–2 s^–1 for 10 days at 30°C. Similarly, cyanobacterial growth curves were measured. All strains were cultured in liquid BG-11 medium containing 50 μg/mL kanamycin and supplemented with 2 mM MMS or 0.5 mM HU (Beam et al., 2002; Blanco et al., 2010), and the optical density was measured at 730 nm every day for 10 days. MMS and HU were purchased from Solarbio (Solarbio Life Sciences, Beijing, China). The experiments were performed in duplicate, at least three times.

To test the tolerance of Synechocystis sp. PCC6803 strains to H₂O₂ stress, cells were subjected to increasing concentrations of H₂O₂ for 7 days (Sein-Echaluce et al., 2015). Cultures with an approximate OD₇₃₀ of 1.5 were washed once with fresh BG-11 medium, and then 200 μL of the culture was dispensed into each well of a 96-well plate. H₂O₂ was supplemented to the wells at final concentrations of 0, 1, 2, 5, and 10 mM. The plate was incubated for 7 days in dark conditions at 30°C.

Experimental data analyzed for significance were performed by using GraphPad Prism 6 (GraphPad Software, San Diego, California USA). Statistical analyses for the rest of the assays were performed using paired two-tailed Student’s t-test. Error bars represent ± SEM. *p < 0.05; **p < 0.01; ****p < 0.0001; n.s., not significant.

A gene (sll0896) in the Synechocystis sp. PCC6803 genome is predicted to encode a RuvC homolog, and so was designated as synruvC. To compare SynRuvC with RuvCs from other bacterial taxa, the amino acid sequence of SynRuvC was aligned with those of RuvCs from E. coli, P. aeruginosa, Mycobacterium leprae, Thermotoga maritima, Thermus thermophilus, and D. radiodurans using the ESPript online server. As the details of conserved amino acid residues and secondary structure displayed in Supplementary Figure 1A, SynRuvC had >30% sequence identity with the RuvCs of other bacteria. Next, a phylogenetic tree of the RuvCs of a variety of species was generated using MEGA 6. Despite their high amino-acid sequence similarities, SynRuvC was genetically distant from those of other bacteria (Supplementary Figures 1A, B). The tertiary structures of SynRuvC dimer (predicted using Swiss-model), EcRuvC (PDB ID: 1HJR), and DrRuvC (PDB ID: 7W8D) were similar, suggesting functional conservation (Supplementary Figure 1C). Altogether, these results suggest that SynRuvC probably has conserved functions, but different properties compared to other RuvCs.

Recombinant SynRuvC was expressed in BL21(DE3) cells with a His-SUMO-tag fused to the N-terminus. The His-SUMO-tag was removed after elution of His-SUMO-SynRuvC from the Ni-NTA column, and after desalting purified SynRuvC was obtained by ion-exchange chromatography (HiTrap Q HP column) (Supplementary Figure 1D).

HJ resolvase recognizes three- and four-way junctions in a structure-specific manner (Qin et al., 2022). The binding of SynRuvC to 3J (three-way junction with a non-homologous core), 3Jh (three-way junction with a 14 bp homologous core), 4J (four-way junction with a non-homologous core), and 4Jh (four-way junction with a 13 bp homologous core) (Supplementary Figure 2) was determined by electrophoresis mobility shift assay. SynRuvC slightly retarded the mobility of three- and four-way junctions and did not form stable complexes. SynRuvC bound to HJs with a variety of affinities (Figure 1A). However, SynRuvC specifically cleaved 4Jh, but not the other HJ substrates (Figure 1B). 4Jh has a 13 bp homologous core, allowing it to migrate within a certain range, whereas 4J is fixed and cannot migrate (Supplementary Figure 2). SynRuvC cleaved 4Jh but not 4J (Figure 1B), comparable to other RuvC homologs (Qin et al., 2022). However, there was no difference in the binding of these two types of four-way junctions by SynRuvC. Therefore, the homologous core of 4Jh affects the digesting activity, but not the binding activity, of SynRuvC.

RuvC cleaves four-way junctions in a sequence-dependent manner, such as DrRuvC (5′-(G/C)TC↓(G/C)-3′), EcRuvC (5′-(A/T)TT↓(G/C)-3′), and PaRuvC (TTC) (Shah et al., 1994; Hu et al., 2020; Sun et al., 2022). To investigate whether SynRuvC cleaves in a sequence-dependent manner, the four oligonucleotide strand cleavage sites of 4Jh were analyzed. Surprisingly, 4Jh was cleaved at different sites on each oligonucleotide strand (Figures 2A, B). Additionally, some cleavage sites were not located in the homologous mobile region of 4Jh, similar to Hjc of Pyrococcus furiosus (Komori et al., 1999). The cleavage of 4Jh by SynRuvC exhibit no apparent sequence specificity. However, it is noticeable that the cleavage sites predominantly occur in the 5′-TG↓(G/A)-3′ sequence (Figure 2C).

To confirm that the nicked duplex DNA produced by SynRuvC can be ligated in a repair process similar to Hjc, we conducted an analysis using 4Jhs (4Jh with a labeled short strand) to examine the ligation products. After junction cleavage, the cleavage products of 4Jhs were treated with T4 DNA ligase, which converted them into a 70-nucleotides (nt) strand (Figures 2D, E). This suggests that nicked dsDNA is repaired by ligation. This indicated that SynRuvC cleaves 4Jhs symmetrically at related sites on both strands, leaving 5′-phosphate and 3′-hydroxyl termini.

Nicked HJs are involved in HR by acting as precursors for fully ligated HJ. Moreover, they are crucial intermediates in the generation of crossovers (Osman et al., 2003; Machin, 2020). To assess whether SynRuvC resolves these structures, we designed 12 possible nicked HJs (4Jhn) based on the 4Jh structure, where n is the position of the nick (nicked at position J2, J5, J6, or J7) and independently labeled the intact strands. The set of 12 substrates was incubated with SynRuvC as described in the Materials and Methods.

When the strand opposing the nick was labeled, no obvious cleavage by SynRuvC was observed in 4Jhn 7-2*, 4Jhn 6-5*, 4Jhn 2-7*, and 4Jhn 5-6* (Figure 3A). These four substrates were resolved by SynRuvC as determined by native PAGE (Figure 3B). These results suggested that these 4Jhn substrates may be cleaved on the other two strands, but not the labeled strands opposite the nick. As expected, when 4Jhn was labeled on the 5′- or 3′-strand with respect to the nick, cleavage by SynRuvC was detected (Figures 3C, E, G). Consistently, we detected nick or gap duplexes by native PAGE (Figure 3D). The cleavage sites within the homologous core predominantly had the pattern 5′-G↓(G/A)-3′ (Figure 3G).

Notably, the primary product of cleavage by SynRuvC of 4Jhn 7-5*, 4Jhn 6-7*, 4Jhn 2-6*, and 4Jhn 5-2* substrates was a short dsDNA (Figure 3F), indicating cleavage on the 5′-strand relative to the nick. Interestingly, the cleavage sites on the 4Jhn 7-5*, 4Jhn 6-7*, 4Jhn 2-6*, and 4Jhn 5-2* substrates were not perfectly symmetrical to the pre-existing nick in position 3′, resulting in products with overhangs and flaps (Figure 3G). The excision of one arm from a nicked HJ to form a three-arm structure has been reported for Yen1, AtGEN1, and AtSEND1 (Bauknecht and Kobbe, 2014), and is referred to as replication fork intermediate (Ref-I) cleavage activity (Supplementary Figure 3). Together, these results show that SynRuvC cleaves nicked 4Jh primarily by Ref-I and RF-like cleavage activities.

Divalent metal ions are required as cofactors by HJ resolvases (Komori et al., 2000; Hu et al., 2020; Qin et al., 2022). Therefore, we used 4Jh as a substrate to investigate the metal-ion dependence of SynRuvC. We added MnCl₂, MgCl₂, ZnCl₂, CaCl₂, CuCl₂, CoCl₂, and NiCl₂, to the reaction buffer and evaluated their effects on cleavage activity. Like DrRuvC, the activity of SynRuvC was significantly increased by MnCl₂. Although MgCl₂ could in part substitute for MnCl₂, its stimulatory effect was considerably weaker. No activity was detected in the presence of ZnCl₂, CaCl₂, CuCl₂, CoCl₂, and NiCl₂ (Supplementary Figure 4A). Mn²⁺ increased the cleavage activity of SynRuvC on 4Jh in a dose-dependent manner (Supplementary Figure 4B). At ≤10 mM Mg²⁺, the cleavage by SynRuvC of 4Jh increased in a dose-dependent manner (Supplementary Figure 4C). However, a high concentration of Mg²⁺ inhibited the cleavage activity of SynRuvC, consistent with a report on EcRuvC (Bennett et al., 1993).

To determine the optimum temperature, cleavage reactions were conducted at 4°C to 70°C. The activity of SynRuvC was enhanced at 37°C to 60°C but markedly reduced at >65°C (Supplementary Figure 4D). To further investigate its temperature stability, SynRuvC was pre-incubated at 30°C to 60°C for 15 min and incubated with 4Jh at 37°C for 1 h. SynRuvC activity was lost after incubation for 15 min at >55°C (Supplementary Figure 4E). Interestingly, SynRuvC showed considerable HJ cleavage activity at high temperatures such as 60°C but the pretreated protein was not active at 60°C. This might because that the protein quickly became inactivated within the 15-min timeframe (Supplementary Figures 4D, E).

Similar to PaRuvC (Hu et al., 2020), SynRuvC preferred alkaline conditions for 4Jh cleavage, with an optimum at pH 8.5–9.0 (Supplementary Figure 4F). SynRuvC was sensitive to the concentration of salt in the reaction mixture, with minor variations between K⁺ and Na⁺ (Supplementary Figure 4G). SynRuvC cleaved 4Jh in as little as 10 min, and the generation of cleavage products increased in a time-dependent manner for 60 min (Supplementary Figure 4H). Taken together, the optimum reaction conditions for the cleavage of 4Jh by SynRuvC is in a reaction buffer (20 mM Tris-HCl [pH 8.5], 50 mM NaCl, 1 mM DTT, 10% DMSO, 10 mM Mn²⁺, and 1% glycerol) at 50°C. To eliminate the possibility that SynRuvC failed to cleave 3J, 3Jh, and 4J because of unsuitable reaction conditions, these three substrates were evaluated under the optimal conditions. The results reaffirmed that SynRuvC remained incapable of cleaving any of these substrates (Supplementary Figure 5).

The HJ resolvase, OsGEN-L, has 5′-flap endonuclease activity (Yang et al., 2012). To determine whether SynRuvC has novel activities other than conventional HJ recognition and cleavage, we evaluated its binding to (Supplementary Figure 6) and digestion of Figures 4A, B, ssDNA, 5′-overhang, 3′-overhang, duplex, nicked duplex, gapped duplex, bubble, splayed arms, 5′-flap, 3′-flap, and nicked 4Jh. The 5′-ends of the DNA structures were labeled with 6-FAM (Supplementary Table 1). SynRuvC cleaved 5′-overhang, splayed arms, 3′-flap, 5′-flap, and RF, in addition to classic HJs (Figure 4A). This indicated that SynRuvC exhibits robust FEN activity. Native PAGE showed that 5′-overhang was the optimum substrate for SynRuvC (Figure 4B). Also, SynRuvC primarily cleaved DNA substrates with 5′-arms. For instance, SynRuvC exhibited higher cleavage activities on DNA substrates with 5′- compared to 3′-overhang, as well as 5′- than 3′-flap (Figures 4A, B). Similar to DrRuvC, SynRuvC showed slightly stronger cleavage activity of pre-nicked 4Jh substrate compared to intact 4Jh (Figure 4B).

Next, we investigated whether the differences in cleavage activities were linked to binding affinities. According to a DNA binding assay, although all the DNA substrates were bound, SynRuvC preferred to bind DNA substrates with 5′-single strands than double strands (Supplementary Figure 6). Therefore, SynRuvC has a structural preference for DNA substrates. For high-resolution mapping of cleavage sites on branched DNA structures (5′-overhang, splayed arms, and 5′-flap), the reaction products were separated by 15% denaturing PAGE (Figure 5A). SynRuvC preferred to cleave the 5′-arms with a cleavage site on the 3′-strand of the duplex near the branch (Figure 5B) and showed a preference for cleavage sites with a 5′-G↓G/A-3′ pattern. These results showed that SynRuvC has FEN-like activity and cleaves a variety of non-HJ DNA substrates similarly to HJs.

To further explore the binding and cleavage of SynRuvC to non-HJ DNA substrate, we used DNA substrates with 5′-overhangs of a variety of lengths. Surprisingly, the 5′-overhang cleavage activities varied with increasing overhang length, as determined by native PAGE (Figure 6A). To investigate the underlying mechanism, we evaluated the cleavage sites by 15% denaturing PAGE. The cleavage efficiency of 5′-CG↓GC-3′ decreased with decreasing length of 5′-overhang (30–40 bp) (Figure 6B and Supplementary Figure 7). A 5′-overhang of length 33 nt (opposite to the 37th on the complementary strand) generated a new cleavage site (5′-TC↓AA-3′) (Figures 6B, C and Supplementary Figure 7). These findings suggested that in addition to structural recognition, SynRuvC has a preferred cleavage sequence.

To evaluate the mechanism of SynRuvC binding and cleavage, we performed a docking simulation using the 3D structure of SynRuvC predicted by AlphaFold2 from Uniprot data and 5′-overhang (35 nt) predicted by BIOVIA Discovery Studio Visualizer. The optimum docking conformation, as determined by GRAMM, indicated that SynRuvC binds to the junction of the 5′-overhang (35 nt) at the single-double strand (Figure 6D). Deoxyribonucleotides dA-35, dA-36, dC-38, dG-39, and dG-40 of the 5′-overhang formed hydrogen bonds with the corresponding amino acid residues of SynRuvC (Figure 6D). The binding sites of Glu-42, Arg-52, Lys-117, and Lys-127 of SynRuvC with the 5′-overhang coincide with those reported for the interaction of RuvCs with HJs (Gorecka et al., 2013; Sun et al., 2022). These results indicated that after SynRuvC binds the single-double strand regions of 5′-overhangs of different lengths, it selects recognition and cleavage sites.

Subsequently, we designed a 3′-overhang and gapped duplex with a 33 nt overhang or 12 nt gap, distinct from the previous substrates, to investigate the cleavage activity of SynRuvC. The reaction conditions were as in Figure 4, and the products were resolved by 10% native or 15% denaturing PAGE (Supplementary Figure 8). SynRuvC cleaved both 3′-overhang-38 nt and gap-12 nt duplexes. In addition to its structural specificity, SynRuvC may also exhibit sequence specificity, necessitating the presence of an appropriate cleavage site in the binding region.

To assess the function of the C-terminus in DNA binding and cleavage, we compared the binding and cleavage activities of the truncated protein SynRuvC^1–144 and SynRuvC^FL in vitro. As the control, the highly conserved 75^th amino acid residue of SynRuvC (Supplementary Figure 1A), glutamic acid, was substituted with alanine. The mutant proteins were purified identically to SynRuvC^FL (Supplementary Figure 1D). DNA binding and cleavage reactions were carried out on 5′-FAM labeled 4Jh. The SynRuvC^E75A mutation in SynRuvC abolished its 4Jh binding and cleavage activities (Supplementary Figures 9A, B). Although deletion of the C-terminus slightly altered the migration of the protein-DNA complexes (Supplementary Figure 9A), it abolished the DNA cleavage activity of SynRuvC (Supplementary Figure 9B). A plausible explanation is that the amino acid D147 in α-helix 5 mediates substrate binding (Figure 6D). Hence, deletion of the C-terminus, which contains α-helix 5, resulted in the loss of SynRuvC^1–144 cleavage activity. Therefore, deletion of the C-terminus hinders substrate binding, leading to loss of the cleavage activity of SynRuvC.

HR prevents methylation-induced toxicity in E. coli (Nowosielska et al., 2006). RadA, RecG, RecBCD, and RuvC, which function in HR, are involved in mitigating methylation-induced toxicity (Beam et al., 2002; Nowosielska et al., 2006). To gain insight into the physiological functions of SynRuvC, we employed conventional gene knockout and knock-in techniques for Synechocystis sp. PCC6803 (Cameron and Pakrasi, 2010; Uchiyama et al., 2020). We thus generated a strain overexpressing synruvC, designated synruvC^OE (Supplementary Figures 10A, B). However, after several rounds of screening, we generated only synruvC knockdown strains (Figure 7A and Supplementary Figures 10C–E).

Photoautotrophic growth of synruvC^KD in BG11 was slower than the WT and synruvC^OE strains (Figure 7B). We analyzed growth of the WT, synruvC^KD, and synruvC^OE strains in BG11 medium supplemented with methyl methane sulfonate (MMS) or hydroxyurea (HU). Although growth of the WT and synruvC^OE strains was inhibited by MMS or HU, the synruvC^KD strain exhibited poorer proliferation than WT and synruvC^OE strains (Figures 7B, C). Similarly, the WT and synruvC^OE strains had a notable growth advantage in BG11 supplemented with or without MMS (Figures 7D, E). These results implicate SynRuvC in DNA repair in Synechocystis sp. PCC6803.

DNA repair proteins reverse the damage caused by the reactive oxygen species (ROS) (Stohl and Seifert, 2006) generated during photosynthesis. To investigate whether the expression level of SynRuvC affects tolerance to oxidative stress caused by exogenous hydrogen peroxide (H₂O₂) in Synechocystis sp. PCC6803 strains, 200 μL suspensions of the WT, synruvC^KD, and synruvC^OE strains were dispensed into wells of a 96-well plate containing H₂O₂. The synruvC^KD strain was more sensitive to oxidative stress than the WT and synruvC^OE strains (Figure 7F). Collectively, these results show that SynRuvC enhances tolerance to MMS, HU, and ROS.