In this study, Z. mobilis ZM4 (ATCC 31821) was used as the parental strain and cultured at 30°C with shaking at 100 rpm in RM medium (50 g/L glucose, 10 g/L yeast extract, 2 g/L KH₂PO₄, and 1.5% agar for solid). E. coli DH5α was used for plasmid construction, and all E. coli strains were cultured in Luria–Bertani (LB) medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, and 1.5% agar for solid) at 37°C, 250 rpm. When required, 100 or 300 μg/mL kanamycin was added to E. coli or Z. mobilis, respectively.

The nucleotide sequence of the TtdAgo gene (WP_055429304.1; Thermococcus thioreducens) was retrieved from the NCBI database. Online software Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) was used to compare some of the currently studied Ago proteins for multiple sequence comparisons. To further analyze the affinities of TtdAgo with other studied Ago proteins, phylogenetic tree analysis was performed by using the maximum likelihood method using MEGA X software based on the results of multiple sequence alignment.

The nucleotide sequences of the TtdAgo and TtdAgo_DM genes (D538A and D608A) were codon-optimized for expression in E. coli. The TtdAgo and TtdAgo_DM genes were synthesized (GeneCreate Biotechnology, Wuhan, China) and cloned into a pET23a expression vector in frame with the C-terminal ×6 His tag. The TtdAgo and TtdAgo_DM proteins were expressed in E. coli Rosetta (DE3) (Novagen, Darmstadt, Germany). Cultures were grown at 37°C in LB medium containing 50 μg/mL ampicillin induced by adding isopropyl-β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM until OD₆₀₀ reached 0.8. During the expression, cells were incubated at 18°C for 20 h with continuous shaking. The cells were collected by centrifugation and stored at −80°C for further protein purification.

The cell pellets were resuspended in Buffer A (20 mM Tris-HCl pH 7.5, 500 mM NaCl, and 10 mM imidazole) supplemented with an EDTA-free protease inhibitor cocktail tablet (Roche, Shanghai, China) and disrupted by sonication (Scientz-IID: 350 W, 2 s on/4 s off for 30 min). The lysates were clarified by centrifugation at 21,000 g for 20 min, and the supernatant was loaded onto Ni-NTA agarose resin at 4°C for 1–2 h with rotation and subsequently extensively washed with Buffer A containing 50 mM imidazole. The bound protein was eluted with Buffer A containing 300 mM imidazole. The eluted protein was concentrated against Buffer B (20 mM HEPES pH 7.5, 500 mM NaCl, and 1 mM dithiothreitol (DTT)) by ultrafiltration using an Amicon 50K filter unit (Millipore, United States). Next, the protein was diluted in 20 mM HEPES pH 7.5 to lower the final salt concentration to 125 mM NaCl. The diluted protein was applied to a heparin column (GE Healthcare, Boston, United States of America) and equilibrated with Buffer C (20 mM HEPES pH 7.5, 125 mM NaCl, and 1 mM DTT), then washed with at least 10 column volumes of the same buffer and eluted with a linear NaCl gradient (0.125–1 M). Fractions containing TtdAgo were concentrated by ultrafiltration using an Amicon 50K filter unit and purified on a Superdex 200 16/600 column (GE Healthcare, Boston, United States). The protein was eluted with Buffer B (20 mM HEPES pH 7.5, 500 mM NaCl, and 1 mM DTT). Purified TtdAgo was concentrated using an Amicon 50K filter unit and diluted in Buffer B to a final concentration of 8 μM, aliquoted, and flash-frozen in liquid nitrogen. The purified protein was stored at −80°C.

Cleavage assays were performed using synthetic guides and targets. Most reactions were performed with pAgo, guide, and target at the molar ratio of 4:4:1. A measure of 800 nM TtdAgo was mixed with 400 nM gDNA or gRNA and incubated for 10 min at 37°C using a PCR thermocycler (T100, Bio-Rad, CA, United States) for guide loading in buffer RB (10 mM HEPES pH 7.5, 100 mM NaCl, and 5% glycerol) with 5 mM MnCl₂. Then added 200 nM of nucleic acid target. The reactions were performed in PCR tubes at 75°C for 20 min and stopped after indicated time intervals by mixing the samples with the ×2 RNA loading dye (95% formamide, 18 mM EDTA, 0.025% SDS, and 0.025% bromophenol blue) and heating it for 5 min at 95°C. The cleavage products were resolved by 20% denaturing PAGE, stained with SYBR Gold (Invitrogen, CA, United States), visualized using Gel DocTM XR+ (Bio-Rad, CA, United States), and analyzed using ImageJ software.

In two half-reactions, 800 nM TtdAgo was preloaded with either 1,000 nM forward or reverse DNA guide in a reaction buffer containing 5 mM HEPES-NaOH pH 7.5, 100 mM or 250 mM NaCl, 5 mM or 10 mM MgCl₂, and 2.5% glycerol. The half-reactions were incubated for 30 min at 37°C. Next, both half-reactions were mixed, and 200 ng target plasmid was added. Then, the mixture was incubated for 20 min at 75 °C. A ×6 DNA loading dye (NEB, MA, United States) was added to the plasmid sample prior to resolving it on a 1% agarose gel stained with ethidium bromide.

The shuttle plasmid pTZ22b was used for TtdAgo expression in Z. mobilis ZM4. For the plasmid construction, primers synthesized by Tsingke (Beijing, China) were used for the polymerase chain reaction (PCR) using DNA polymerase (Takara, Japan) to obtain DNA fragments. All plasmids were assembled by the Gibson Assembly method (Gibson et al., 2009). After purification, gene and vector fragments were ligated through the T5 exonuclease (NEB, WA, United States)-dependent DNA assembly method as described previously (Tang et al., 2022). After transformation in E. coli, the correct colonies were selected by colony PCR and confirmed by Sanger sequencing (Tsingke Biotechnology, Beijing, China).

The recombinant plasmid was transformed into Z. mobilis competent cells via electroporation (0.1-cm electrode gap, 1600 V, 200 Ω, and 25 μF) using a Gene Pulser® (Bio-Rad, CA, United States of America). The correct colonies were selected by colony PCR. Recombinant cells were placed on RM agar plates with 300 μg/mL kanamycin supplementation, and then the plates were stored at −80°C.

To prepare the seed culture, Z. mobilis strains from frozen glycerol stocks were revived in 50-mL flasks containing 40 mL RM medium. After culturing overnight without shaking to the mid-exponential phase, the seed culture was harvested and inoculated in 50-mL shake flasks containing 40 mL RM medium with an initial OD₆₀₀ nm value of 0.1. Cell growth in terms of the absorbance value at 600 nm (OD₆₀₀) was measured using a spectrophotometer (UV-1800, AOE, China) at different time points.

The multiple sequence alignment of TtdAgo with other Agos (Supplementary Figure S1A) suggests that although TtdAgo is phylogenetically closely related to the PfAgo from Pyrococcus furiosus (Figure 1A) and contains the canonical catalytic tetrad residues (D538, E576, D608, and H725) in the PIWI domain (Supplementary Figure S1A) that is essential for the nuclease activity (Hegge et al., 2019), TtdAgo (WP_055429304.1) is distantly related to most other characterized eAgos and pAgos with a sequence identity < 20% (Figure 1A; Supplementary Figure S1A). The TtdAgo gene was codon-optimized for expression in E. coli, and the catalytically inactive variant of TtdAgo (TtdAgo_DM) was obtained with substitutions of two out of four catalytic tetrad residues (D538A/D608A, Supplementary Figure S1A).

To investigate its biochemical properties and in vivo function, TtdAgo and TtdAgo_DM were expressed successfully in E. coli using a T7-based pET expression system. The protein was purified using Ni-NTA affinity, heparin column affinity, and size-exclusion chromatography (see Supplementary Figure S1B and Materials and methods for details). We first studied the nucleic acid specificity of TtdAgo with in vitro cleavage assay using synthetic oligonucleotides. TtdAgo was preloaded with 18-nt DNA or RNA guides containing a 5′-phosphate (5′-P) or 5′-hydroxyl (5′-OH) group at 37°C for 10 min followed by the addition of complementary 45-nt long ssDNA or RNA targets (Figure 1B). After incubation for 30 min at 75°C in reaction buffer containing 5 mM Mn²⁺, the cleavage products were resolved on 20% denaturing gel and visualized by SYBR Gold staining.

As most pAgos studied strongly prefer to cleave DNA targets (Nakanishi et al., 2012; Lisitskaya et al., 2018; Jolly et al., 2020), TtdAgo can use both 5′-P-gDNA and 5′-OH-gDNA to cleave DNA targets, resulting in the appearance of the 34-nt-long 5′-fragment of the DNA target, and no RNA target cleavage was observed (Figures 1C, D). However, for the guide RNAs, no TtdAgo-mediated cleavage was observed for either DNA or RNA targets, and no cleavage products were observed in the absence of TtdAgo protein or guides (Figures 1C, D). TtdAgo cleavage required the intact catalytic tetrad in the PIWI domain, and point mutations in the tetrad eliminated the activity of TtdAgo (Figures 1C, D).

In summary, similar to previously characterized pAgos from thermophilic prokaryotes such as TtAgo, PfAgo, MjAgo, and FpAgo (Swarts et al., 2014a; Swarts et al., 2015; Willkomm et al., 2017; Zander et al., 2017; Guo et al., 2021), TtdAgo prefers to use DNA guides to cleave DNA targets. This result is similar to a simultaneous study of TtrAgo (TtdAgo) reported earlier by Fang et al. (2022). However, in contrast to most eAgos and pAgos including hAgo2, IbAgo, KmAgo, and MbpAgo (Rivas et al., 2005; Liu Y. et al., 2021; Kropocheva et al., 2021; Li W. et al., 2022), whose target cleavage resulted in a shift of the cleavage site mediated by 5′-OH guides compared to that using 5′-P-gDNA as shown by Fang et al. (2022), we did not observe the cleavage position shift for TtdAgo with 5′-P-gDNA compared to that with 5′-OH-gDNA. This result may be due to the different targets and reaction conditions such as the concentration of divalent ion and the temperature both being different. In addition, although Ago proteins do not need the PAM (protospacer adjacent motif) sequence, there were still differences in the efficiency between different targets (Hunt et al., 2021).

To further determine the prerequisites for TtdAgo-mediated target cleavage, the influence of temperatures and divalent cations on cleavage activity was tested. To explore the active temperature range of TtdAgo, we tested the effect of temperature on DNA cleavage activity mediated by 5′-P-gDNA or 5′-OH-gDNA at temperatures ranging from 30°C to 95°C. The results of the temperature-dependent DNA cleavage activity revealed that TtdAgo bound to 5′-P-gDNA or 5′-OH-gDNA showed activity in the range of 55–95°C, and the best temperature was 75°C (Figure 2A; Supplementary Figure S2A). In summary, the activity of TtdAgo mediated by 5′-P-gDNA or 5′-OH-gDNA exhibited no significant difference between 30°C and 37°C, for example, for 5′-P-gDNA, the cleavage percentage is 0.037 and 0.042, respectively. However, enhanced cleavage activity was displayed from 45°C to 75°C, which then decreased at higher temperatures above 75°C (Figure 2A; Supplementary Figure S2A).

Previous studies also demonstrated that divalent metal ions are crucial for Ago protein activities (Song et al., 2004). Thus, we tested the cleavage activity between 5′-P-DNA and 5′-OH-DNA guides with the DNA target under different divalent metal ions. In the presence of different divalent metal ions (Mg²⁺, Ca²⁺, Mn²⁺, Fe²⁺, Co²⁺, Ni²⁺, Cu²⁺, or Zn²⁺), TtdAgo was active only when Mg²⁺ or Mn²⁺ were used as cations with Mn²⁺ giving a higher activity than Mg²⁺ (Figure 2B; Supplementary Figure S2B). Titration of Mn²⁺ ions showed that TtdAgo was active in the concentration range of 0.02–10 mM and had an increased cleavage activity at Mn²⁺ concentrations ≥ 1.0 mM (Figure 2C). The cleavage efficiency when using 5′-P-gDNA was higher than that when using 5′-OH-gDNA when Mn²⁺ was used, while TtdAgo was active at Mg²⁺ concentrations ≥ 2.5 mM, and the activity exhibited no significant difference between 2.5 and 10 mM no matter whether the guide is 5′-P-gDNA or 5′-OH-gDNA (Figure 2D). Thus, TtdAgo-mediated cleavage was more efficient in the presence of Mn²⁺.

Our study thus demonstrated that TtdAgo cleaved DNA at temperatures ranging from 30°C to 95 °C and had good DNA cleavage activity at 70–80°C, which is consistent with the previous study (Fang et al., 2022). Furthermore, Fang et al. (2022) also found that TtdAgo was active at Mn²⁺ concentrations ≥ 0.1 mM, which is consistent with our results. However, we did not observe effective cutting under Co²⁺, which may be due to the 5 mM Co²⁺ used in this study being much higher than 0.5 mM Co²⁺ used in the previous study. The presence of excess Co²⁺ is likely to inhibit the activity of TtdAgo. In contrast, Fang et al. (2022) did not observe the cleavage product under the presence of Mg²⁺, which may be due to the 0.5 mM Mg²⁺ they used being lower than that required for TtdAgo to be actively functional as we demonstrated that TtdAgo functioned at Mg²⁺ concentrations ≥ 2.5 mM (Figure 2D). These studies thus demonstrated that it is crucial to obtain accurate concentrations of divalent cations for the functionality of pAgos.

Previous studies indicated that the guide length may affect cleavage efficiency (Kropocheva et al., 2021). Thus, we investigated the cleavage activity of TtdAgo using 5′-P-gDNA and 5′-OH-gDNA of different lengths ranging from 11 to 40 nt at 75°C for 30 min with 5 mM Mn²⁺. The cleavage products were observed using 14–40-nt 5′-P-gDNA, and 17–19-nt 5′-P-gDNA was optimal (Figures 3A, B). In the case of 5′-OH-gDNA, TtdAgo was most active using 16–18-nt gDNA with a lower efficiency observed for longer or shorter gDNA (Figures 3A, B). Although we did not observe the cleavage position shift as seen in other Ago proteins reported previously such as KmAgo and MbpAgo (Liu Y. et al., 2021; Li W. et al., 2022), the cleavage occurred only between the 10th and 11th guide positions for TtdAgo. In conclusion, according to the relative highest cleavage efficiency of TtdAgo, the most appropriate guide length for both 5′-P-gDNA and 5′-OH-gDNA is 18 nt (Figure 3B). Thus, most experiments were performed using guides with a length of 18 nt subsequently.

Previous studies also suggested that the presence or absence of 5′-P may affect cleavage efficiency (Cao et al., 2019; Kuzmenko et al., 2019). To investigate the catalytic properties of TtdAgo mediated by 5′-P-gDNA and 5′-OH-gDNA, we performed a cleavage kinetics assay at 75°C with 5 mM Mn²⁺. The reaction process of DNA cleavage guided with 5′-P-gDNA was obviously faster than that with 5′-OH-gDNA (Figure 3C). At the beginning of the cleavage reaction, there was no significant difference between these two reactions, but the cleavage percentage mediated by 5′-P-gDNA was higher than that mediated by 5′-OH-gDNA with the extension of the reaction time. For example, when the reaction time was extended to 10 min or 15 min, the cleavage percentages were 0.48 or 0.67 when mediated by 5′-P-gDNA and 0.33 or 0.40 when mediated by 5′-OH-gDNA, respectively. Therefore, TtdAgo prefers to use 5′-P-gDNA as a guide strand.

To evaluate the influence of guide concentration on cleavage activity, the DNA cleavage was monitored under different ratios of TtdAgo, guide, and target using 5′-P-gDNA or 5′-OH-gDNA under the condition of 5 mM Mn²⁺ at 75°C for 30 min. In this experiment, TtdAgo cleavage efficiency was decreased correspondingly when DNA guide supplementation was gradually decreased (Figure 3D and Supplementary Figure S3A). Therefore, most experiments were displayed with TtdAgo, guide, and target at the molar ratio of 4:4:1 (800 nM TtdAgo preloaded with 800 nM guide plus 200 nM target).

Similar to the majority of pAgos, TtdAgo utilizes 5′-phosphorylated DNA guides preferentially. In addition, thermophilic pAgos characterized previously indicate that complementary base pairing of approximately 15 nt between the guide and target is required to form a stable double helix structure at high temperatures. For TtdAgo, a minimum of 16 nt gDNA is required. Furthermore, TtdAgo is most active with a length of 17–19 nt for 5′-P-gDNA and 16–19 nt for 5′-OH-gDNA, with a lower efficiency observed with longer or shorter guides, which is consistent with the previous study (Fang et al., 2022).

The 5′-end-nucleotide of the guide strand is bound to the MID pocket of Ago proteins, which has certain preference for it (Nakanishi et al., 2012). To determine whether TtdAgo has a bias for the first nucleotide of the guide, cleavage assays were performed using four variants of DNA guides with different 5′-end-nucleotides but otherwise identical sequences. Slightly reduced cleavage rates were observed when TtdAgo loaded with 5′-P-gDNA containing a 5′-G and TtdAgo loaded with 5′-P-gDNA containing 5′-C, 5′-A, or 5′-T cleaved the target comparably (Figure 4A). However, when TtdAgo was loaded with 5′-OH-gDNA with different 5′-end-nucleotides, there were no obvious changes observed for the cleavage efficiency (Figure 4B). In conclusion, in contrast to some thermophilic Agos which have a bias for the first nucleotide of the guides such as TpsAgo and FpAgo (Guo et al., 2021; Sun et al., 2022), TtdAgo has no obvious preference for the 5′-end-nucleotide of the DNA guides (Figures 4A, B; Supplementary Figure S3B).

Previous studies also showed that mismatches between the guide and target may have large interference on the cleavage efficiency and precision of Ago proteins (Dayeh et al., 2018; Liu Y. et al., 2021; Li W. et al., 2022), and even a single mismatch in the seed region (guide positions g2–g8) of the guide can greatly reduce target recognition and cleavage (Miyoshi et al., 2016; Liu et al., 2018). To explore the mismatch tolerance of TtdAgo, we designed a set of DNA guides containing a single-nucleotide or dinucleotide mismatches at certain positions (Supplementary Table S1) to investigate the DNA cleavage activity with TtdAgo (Figure 5, Supplementary Figure S4). When a single-nucleotide mismatch was introduced, such as at positions 4, 7–9, or 14–18, mismatches at most positions affected the cleavage efficiency but did not lead to a dramatic decrease in cleavage efficiency, except for mismatches at positions 4 and 8 with cleavage efficiency reduced by 50%. In summary, TtdAgo has a high tolerance for mismatches between the guide and target strands.

It has been demonstrated that thermophilic Agos such as TpsAgo and FpAgo had biases for the first nucleotide of the guides (Cao et al., 2019; Guo et al., 2021). However, TtdAgo has no obvious preference for the 5′-end-nucleotide of a guide, which is similar to other pAgos such as KmAgo and MbpAgo (Liu Y. et al., 2021; Li W. et al., 2022). Previous studies revealed the significance of complementarity between the guide and target for Ago cleavage. TtdAgo has a high tolerance for mismatches between the guide and target strands. Therefore, TtdAgo can potentially be used to clear DNA virus because it is difficult for the virus to escape by mutating single bases.

Previous studies showed that pAgos can not only use gDNA to cleave ssDNA specifically but can also cleave plasmid DNA targets in vitro under the guidance of a complementary pair of DNA guides to generate double-stranded (dsDNA) breaks, or in a guide-independent manner (Swarts et al., 2014a; Swarts et al., 2015; Zander et al., 2017; Liu Y. et al., 2021; Guo et al., 2021). We first evaluated dsDNA cleavage in the absence of gDNA. When TtdAgo was incubated with plasmid pUC19 (Figure 6A) at 75°C in the reaction buffer with 5 mM Mn²⁺ for 20 min, there were no cleavage products observed and target plasmids were basically completely degraded (Supplementary Figure S5). Considering the facts that TtdAgo has more than 50% sequence similarity with PfAgo and the presence of Mn²⁺ results in the degradation of plasmid in the plasmid cleavage assay of PfAgo (Swarts et al., 2015), we replaced the divalent metal ions in the reaction buffer and performed the assay at 75°C in the reaction buffer containing 10 mM Mg²⁺ for 20 min. Our result exhibited that the plasmid generated a small amount of open-circular plasmid even in the absence of TtdAgo in buffer with 100 mM NaCl (Figure 6B).

Surprisingly, when TtdAgo was added, a linearized plasmid was detected even in the absence of guides, but that was not observed when TtdAgo_DM was used. These findings suggested that this guide-independent non-specific plasmid relaxation is caused by TtdAgo, which is consistent with previous plasmid cleavage experiments using pAgos from other thermophilic organisms such as TtAgo, PfAgo, MjAgo, and TpsAgo (Swarts et al., 2014a; Swarts et al., 2015; Zander et al., 2017; Sun et al., 2022). Fang et al. (2022) predicted the structure of TtdAgo using the AlphaFold web tool and proposed that the cavity of TtdAgo could freely bind to dsDNA when the guide is not loaded, and then dsDNA could be cleaved by the conserved DEDH catalytic center. Furthermore, when the assay was performed in a buffer with 250 mM NaCl, the guide-free TtdAgo-mediated cleavage was not observed. These results indicated that this guide-independent non-specific degradation of plasmid DNA decreased with increasing NaCl concentration.

We then investigated whether TtdAgo can utilize gDNA to cleave plasmid DNA. A pair of 18-nt 5′-P-gDNAs (Supplementary Table S1) named forward and reverse guides (“FW”’ and “RV” guides, respectively) were designed corresponding to the same target region of the pUC19 plasmid. In addition, two non-complementary guides (“NC” guides) were also designed with a random sequence with no overlap with pUC19 (Figure 6C). TtdAgo and gDNAs were incubated with pUC19 in buffers containing 100 or 250 mM NaCl, and both TtdAgo complexes with FW and RV guides were mixed and incubated with pUC19. Compared to the reaction with no guides or NC guides, the linearized plasmid was only detected in the assay containing FW or RV guides (Figure 6D). Moreover, the cleavage efficiency was higher in the reaction containing both FW and RV guides than that containing either FW or RV guide only. In summary, TtdAgo can target the dsDNA plasmid with gDNA resulting in a dsDNA break.

Like the bacterial TtAgo and archaeal PfAgo (Swarts et al., 2015; Jolly et al., 2020), TtdAgo can utilize a single gDNA or a pair of gDNAs to generate dsDNA breaks, though they do not belong to the same branch in the phylogenetic tree. These findings indicate that there is a high conservation of pAgos in function, and pAgos such as TtdAgo might be used for the development of pAgo-based genome-editing tools.

To explore the potential of applying TtdAgo to develop pAgo-based genome-editing tools in Z. mobilis, the pTZ22b-TtdAgo plasmid expressing TtdAgo was constructed and then introduced into Z. mobilis to generate the recombinant strain ZM4-TtdAgo. The impact of TtdAgo expression on Z. mobilis was then evaluated, and the result indicated that the introduction of TtdAgo reduced the cell growth of ZM4-TtdAgo dramatically. As shown in Figure 7, compared to the wild-type with an OD₆₀₀ value of 5.32 ± 0.12 and growth rate of 0.39, ZM4-TtdAgo had a final OD₆₀₀ value of 1.06 ± 0.07 and growth rate of 0.25. These results indicated that TtdAgo is toxic to Z. mobilis, which may be due to the nuclease activity of TtdAgo as reported for the programmable nucleases Cas9 (Pruett-Miller et al., 2009; Morgens et al., 2016) and the capability of TtdAgo to freely bind and cleave dsDNA even when the guide is not loaded (Fang et al., 2022). Therefore, further study is needed to alleviate the cell toxicity of TtdAgo for pAgo-based genome-editing tool development in Z. mobilis.