key: cord-0032036-mfdf76xu
authors: Li, Wenqiang; Liu, Yang; He, Ruyi; Wang, Longyu; Wang, Yaping; Zeng, Wanting; Zhang, Zhiwei; Wang, Fei; Ma, Lixin
title: A programmable pAgo nuclease with RNA target preference from the psychrotolerant bacterium Mucilaginibacter paludis
date: 2022-05-07
journal: Nucleic Acids Res
DOI: 10.1093/nar/gkac315
sha: 5459987f794292228b2a7b73bcd75718e75459a1
doc_id: 32036
cord_uid: mfdf76xu

Argonaute (Ago) proteins are programmable nucleases found in eukaryotes and prokaryotes. Prokaryotic Agos (pAgos) share a high degree of structural homology with eukaryotic Agos (eAgos), and eAgos originate from pAgos. Although eAgos exclusively cleave RNA targets, most characterized pAgos cleave DNA targets. This study characterized a novel pAgo, MbpAgo, from the psychrotolerant bacterium Mucilaginibacter paludis which prefers to cleave RNA targets rather than DNA targets. Compared to previously studied Agos, MbpAgo can utilize both 5′phosphorylated(5′P) and 5′hydroxylated(5′OH) DNA guides (gDNAs) to efficiently cleave RNA targets at the canonical cleavage site if the guide is between 15 and 17 nt long. Furthermore, MbpAgo is active at a wide range of temperatures (4–65°C) and displays no obvious preference for the 5′-nucleotide of a guide. Single-nucleotide and most dinucleotide mismatches have no or little effects on cleavage efficiency, except for dinucleotide mismatches at positions 11–13 that dramatically reduce target cleavage. MbpAgo can efficiently cleave highly structured RNA targets using both 5′P and 5′OH gDNAs in the presence of Mg(2+) or Mn(2+). The biochemical characterization of MbpAgo paves the way for its use in RNA manipulations such as nucleic acid detection and clearance of RNA viruses.

Eukaryotic Argonautes (eAgos) are key components of the RNA-induced silencing complex (RISC) and participate in posttranscriptional gene regulation and antivirus defense (1) . Argonautes (Agos) are also found in many bacterial and archaeal organisms (2) . Currently, only a few prokaryotic Agos (pAgos) have been characterized, and biochemical studies showed that they could function as programmable endonucleases in vitro and protect cells from foreign genetic elements in vivo (3) (4) (5) (6) . A recent study showed that pAgos may function in DNA replication (7) , suggesting that the cellular functions of pAgos are still obscure and remain to be explored.

Genomic studies revealed that pAgos are much more diverse than eAgos (8) , and pAgos can be classified into long pAgos, short pAgos and PIWI-RE (9) . Despite these divergent cellular functions and diversity, eAgos and most long pAgos adopt a highly conserved six-domain architecture, including the N-terminal (N), linker 1 (L1), middle (MID), linker 2 (L2), P-element induced wimpy testis (PIWI), and PIWI-Ago-Zwille (PAZ) domains (5) . Structural studies of Agos showed that the 5 end and the 3 end of nucleic acid guides are anchored in the MID and PAZ domains, respectively (10, 11) . Agos with endonuclease activity have a catalytic tetrad DEDX (X is D, H, N, or K) in the PIWI domain, which is essential in binding divalent metal ions and responsible for catalysis (5, 12, 13) . After binding small nucleic acid guides, active Agos can precisely cleave the complementary targets between the 10 and 11 nucleotides of the guide nucleic acid, which is the canonical cleavage pattern (14) .

Although eAgos exclusively cleave RNA targets, most previously characterized pAgos cleave DNA targets. Initially, pAgos have been characterized from thermophilic prokaryotes, including TtAgo (Thermus thermophilus) (3), PfAgo (Pyrococcus furiosus) (4) , MpAgo (Marinitoga piezophila) (13) , and MjAgo (Methanocaldococcus jannaschii) (15, 16) , which can be programmed with DNA guides (gDNAs) or RNA guides (gRNAs) to cleave DNA targets effectively at elevated temperatures but not moderate temperatures. Recently, active pAgos from mesophilic prokaryotes including CbAgo (Clostridium butyricum) (17, 18) , LrAgo (Limnothrix rosea) (17) , SeAgo (Synechococcus elongatus) (19) , KmAgo (Kurthia massiliensis) (20, 21) , CpAgo (Clostridium perfringens) and IbAgo (Intestinibacter bartlettii) (22) were characterized to search for a pAgo that can cleave both single-stranded DNA (ss-DNA) and double-stranded DNA (dsDNA) at moderate temperatures and be used for genome editing. Some of these also have RNA target cleavage activity, including TtAgo, MpAgo, CbAgo, CpAgo and KmAgo, but they all cleave DNA targets better than RNA targets (13, 17, (20) (21) (22) (23) .

The clustered regularly interspaced short palindromic repeats-associated (CRISPR-Cas) system is the most widely used enzymatic tool for programmable nucleic acid cleavage. As another programmable nuclease with diverse binding and cleavage activities, pAgos are being exploited in genome editing applications (9) , molecular cloning (24), and nucleic acid detection (25) (26) (27) (28) . To date, catalytically active pAgos characterized have been isolated from thermophilic and mesophilic species (29) . Interestingly, pAgos from mesophilic species analyzed so far are active in a wide temperature range from 37 • C to 50 • C (17, 22) , suggesting the possibility of finding pAgos cleaving targets with a high efficiency at 37 • C also in psychrotolerant or psychrophilic species. In addition, the properties of pAgos from psychrotolerant or psychrophilic species remain unknown.

In this work, we characterized a novel pAgo, MbpAgo, from the psychrotolerant organism Mucilaginibacter paludis (30) , which is distantly related to other characterized pAgos, and contains the canonical catalytic tetrad in the PIWI domain (residues D566, E601, D635, and D768). This study demonstrated that, different from other pAgos, MbpAgo binds gDNAs to cleave RNA with high efficiency but DNA with very low efficiency at physiological temperatures. Furthermore, we demonstrated that MbpAgo could utilize both 5 phosphorylated(5 P) and 5 hydroxylated(5 OH) gDNAs to cleave unstructured and highly-structured RNA, suggesting that it can expand the toolkit for RNA manipulations.

Considering possible applications in human cells, the nucleotide sequence of the MbpAgo gene (WP 008504757.1; M. paludis) and MbpAgo double mutant (MpbAgo DM: D566A, D635A) gene were codon-optimized for expression in Homo sapiens. The codon-optimized genes were synthesized by Wuhan Genecreate Biotechnology Co., Ltd and cloned into pET28a expression vectors in frame with the N-terminal His 6 tag (Supplementary Figure S1B ; Supplementary Table S4 ). MbpAgo variants H482Y/K486 (YK), H482R/K486 (RK), H482A/K486 (AK), H482/K486A (HA) and H482A/K486A (AA) that were used for in vitro activity assays were constructed by polymerase chain reaction (PCR) mediated site-directed mutagenesis (31) ; the primers, template, and purpose are listed in Supplementary  Table S1 . All cloned constructs were verified by DNA sequencing. Escherichia coli Rosetta (DE3) (Novagen), con-taining eukaryotic tRNAs rarely used in E. coli, was used to express MbpAgo or MbpAgo variant proteins. Cultures were grown at 37 • C in Luria-Bertani (LB) medium containing 50 g/ml kanamycin and induced by adding isopropyl-␤-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM until OD 600 reached 0.8. Cells were incubated at 18 • C for 16 h with continuous shaking for expression. Centrifugally collected cells were stored in a −80 • C refrigerator for further protein purification.

The cell pellet was resuspended in Buffer A [20 mM Tris-HCl (pH 7.4), 500 mM NaCl, and 20 mM imidazole] supplemented with 1 mM phenylmethylsulfonyl fluoride and disrupted by sonication (SCIENTZ-IID: 400 W, 2 s on/4 s off for 20 min). The lysate was clarified by centrifugation and the supernatant was loaded onto Ni-NTA agarose resin for 1 h with rotation. The beads were washed with Buffer A, then with the same buffer containing 50 mM imidazole and eluted with Buffer A containing 150 mM imidazole. Fractions containing MbpAgo or variants were concentrated by ultrafiltration using an Amicon 50K filter unit (Millipore) and purified on a Superdex 200 16/600 column (GE Healthcare) equilibrated with Buffer B [20 mM Tris-HCl (pH 7.4), and 500 mM NaCl]. Fractions containing MbpAgo were concentrated with an Amicon 50K filter unit (Millipore), placed in Buffer B, aliquoted and flash-frozen in liquid nitrogen.

Cleavage assays were performed using the synthetic guides and targets (see Supplementary Table S2 for oligonucleotide sequences) under conditions described previously study (20) . 5 -FAM-labeled targets and 5 -P guides were synthesized for some experiments. 800 nM MbpAgo was mixed with 400 nM gDNA or gRNA, and incubated for 10 min at 37 • C for guide loading in buffer RB [10 mM HEPES-NaOH (pH 7.5), 100 mM NaCl, and 5% glycerol] with 5 mM MnCl 2 . Target nucleic acids were added to the final concentration of 200 nM. The cleavage reactions were performed in PCR tubes at 37 • C, stopped after the indicated time intervals by mixing the samples with equal volumes of 2 × RNA loading dye [95% formamide, 18 mM EDTA, and 0.025% sodium dodecyl sulfate (SDS), and 0.025% bromophenol blue], and heated for 5 min at 95 • C. Kinetic analyses of RNA cleavage were performed under singleor multiple-turnover conditions, and the detailed methods were described in the literature (17, 18) . Briefly, in singleturnover reactions (the [MbpAgo-gDNA] /RNA target ratio was > 1.0), the data were fitted to the following singleexponential equation:

where C is the cleavage efficiency at a given time point, C max is the maximum cleavage, and k obs is the observed rate constant. In multiple-turnover reactions (the [MbpAgo-gDNA] /RNA target ratio was < 1.0), the data were fitted to the following two-phase equation:

where B is the burst fraction, k burst is the observed cleavage rate in the burst fraction, and V is the steady-state velocity of RNA cleavage. To analyze the effect of various divalent cations, 5 mM Mn 2+ , Mg 2+ , Ni 2+ , Co 2+ , Cu 2+ , Fe 2+ , Ca 2+ , or Zn 2+ was used in the reactions, respectively. To analyze the temperature dependence of RNA cleavage, MbpAgo was loaded with gDNAs for 10 min at 37 • C; the samples were transferred to indicated temperatures in a PCR thermocycler (T100, Bio-Rad), RNA targets were added and the samples were incubated for 15 min. All reactions were carried out at 37 • C if not indicated. For reactions not containing fluorescent labels, the gels were resolved by 20% denaturing PAGE, stained with SYBR Gold (Invitrogen), and visualized with Gel Doc TM XR+ (Bio-Rad). For reactions containing FAM labels, the gels were visualized with Gel Doc TM XR+ (Bio-Rad) and analyzed by the NIH program ImageJ and Prism 8 (GraphPad).

The HIV-1 DIS 5 -untranslated region (UTR) RNA (32) was in vitro transcribed using T7 RNA polymerase (Thermo Fisher Scientific) and synthetic DNA templates carrying a T7 promoter sequence. The transcripts were treated with DNase I, gel-purified, and ethanol precipitated. The gDNAs used for cleavage (Supplementary Table S3 ) were 5 -phosphorylated using T4 PNK (New England Biolabs) except for cleavage with 5 -OH guides. 800 nM MbpAgo was mixed with 400 nM gDNA and incubated for 10 min at 37 • C for guide loading. The RNA target was added to the final concentration of 200 nM and incubated for 30 min at 37 • C for cleavage. Reactions were stopped with 2 × RNA loading dye and heated for 5 min at 95 • C. The cleavage products were resolved by 8% denaturing PAGE and stained with SYBR Gold.

To examine the loading of the guide onto MbpAgo, MbpAgo and the guide were incubated in buffer RB with 5 mM Mn 2+ or 1 mM EDTA for 10 min at 37 • C. The concentration of the internally FAM-labeled fluorescent guide was fixed as 2.5 nM, whereas the concentration of MbpAgo varied. The samples were mixed with 1 l of 10 × loading buffer [250 mM Tris-HCl (pH 7.5), 40% glycerol] and resolved by 10% native PAGE with 0.5 × Trisborate-EDTA (TBE) buffer. The gels were visualized with the ChemiDoc MP Imaging System (Bio-Rad). To analyze the loading of the target onto the MbpAgo-gDNA complex, MbpAgo DM and label-free guide (1:1) with varying concentrations were incubated for 10 min at 37 • C, and the 100 nM 5 -FAM-labeled target was added to the reaction for 30 min at 37 • C. The samples were mixed with 1 l of 10 × loading buffer and resolved by 10% native PAGE with 0.5 × TBE. For reactions containing FAMlabeled targets, the gels were visualized with the Gel Doc TM XR+ (Bio-Rad). To determine the apparent dissociation constants (Kd) for guide binding by MbpAgo and target binding by the MbpAgo-gDNA complex, the gel images obtained were analyzed with the NIH program ImageJ and Prism 8 (GraphPad). Data were fitted with the Hill equation with a Hill coefficient of 2-2.5. All nucleic acids used in this study are listed in Supplementary Table S2 .

CD spectra and melting curves were acquired with a Chirascan V100 CD spectrometer (Applied Photophysics) equipped with a thermostatted cell holder at 25 • C under a constant nitrogen flow, holding the protein solutions in quartz cuvettes of 0.1 mm optical path, scanning from 290-195 nm at 1 nm/s with a slit width of 1 nm against an air background obtained with the same settings without any cuvette. The protein concentration was 0.32 mg/ml (3.5 M) in all experiments with addition of 3.5 M gDNA or without gDNA, and the buffer contained 7.5 mM Tris-HCl (pH 7.4), and 150 mM NaCl with 5 mM Mn 2+ or 1 mM EDTA. A reference spectrum of the corresponding buffer showed only noise within ± 0.5 mdeg. Melting curves were acquired with 2.5 • C stepwise increments and 30 s intervals from 25 • C to 75 • C. Measurements were performed in duplicate. Data were normalized to the fraction of unfolded protein and analyzed using the software GraphPad Prism 8 in nonlinear regression using the Boltzmann sigmoidal function for melting temperature (Tm) determination.

Isolation of co-purification nucleic acids was carried out as described in Hegge et al.'s (18) method, with minor modifications. Briefly, to 2 mg of purified MbpAgo in Buffer B, CaCl 2 and proteinase K (Zomanbio) were added to final concentrations of 5 mM CaCl 2 and 1 mg/ml proteinase K. The sample was incubated for 50 min at 55 • C. The nucleic acids were separated from the organic fraction by adding Roti-phenol/chloroform/isoamyl alcohol (pH 7.5-8.0) at a 25:24:1 ratio. The top layer was isolated and nucleic acids were precipitated using ethanol precipitation by adding 99% ethanol at a 1:2 ratio supplied with 0.5% linear polymerized acrylamide as a co-precipitating agent. This mixture was incubated for 16 h at −20 • C and centrifuged at 13,000 rpm for 30 min. Next, the nucleic acid pellet was washed with 500 l of 70% ethanol and solved in 50 l nuclease-free water. The purified nucleic acids were treated with either 100 g/ml RNase A (Thermo Fisher Scientific), 2 units DNase I (NEB), or both for 1 h at 37 • C and resolved on 20% denaturing PAGE and stained with SYBR Gold.

MbpAgo (accession no. WP 008504757.1 in the National Center for Biotechnology Information protein database) is distantly related to other characterized eAgos and pAgos (sequence identity < 20%; Figure 1A ; Supplementary Figure S1A) and contains the canonical catalytic tetrad in the PIWI domain (residues D566, E601, D635, and D768; Supplementary Figure S1A ). To study its biochemical properties and in vivo functions, MbpAgo was expressed and purified. The codon-optimized gene encoding MbpAgo was chemically synthesized and cloned into the pET28a plasmid (Supplementary Figure S1B ). In addition to the wild-type protein, its catalytically inactive variant (MbpAgo DM) was obtained by substituting two of four catalytic tetrad residues (D566A/D635A) (Supplementary Figure S1A) . The protein was expressed in E. coli and purified using Ni-NTA-affinity and size-exclusion chromatography (see Supplementary Figure S1 and Materials and Methods for de- Figure S1 ).

The nucleic acid specificity of MbpAgo was next studied by in vitro cleavage assay using synthetic fluorescently labeled oligonucleotide targets ( Figure 1B) . MbpAgo was loaded with 16 nt gDNAs or gRNAs containing a 5 P or 5 OH group at 37 • C for 10 min followed by the addition of complementary 5 -end FAM-labeled 45-nt-long ssDNA or RNA targets ( Figure 1B , Supplementary Table S2 ). After incubation for 30 min at 37 • C, the cleavage products were resolved by 20% denaturing gel ( Figure 1C and D). Unexpectedly, although most studied pAgos strongly prefer to cleave DNA targets (5, 17, 20) , MbpAgo can use both 5 P-gDNA and 5 OH-gDNA to cleave almost all RNA targets, and no DNA target cleavage was observed in 30 min (Figure 1C and D) . In addition, only very weak DNA target cleavage was observed even after incubation for 6 h (Supplementary Figure S1E ). Interestingly, eAgos also can use gDNA to cleave RNA targets with decreased activity relative to the use of gRNA (32, 33) . However, for the gR-NAs, no MbpAgo-mediated cleavage was observed, neither of DNA nor RNA targets, even after incubation for 12 h ( Figure 1C and D; Supplementary Figure S1E ). This DNAguided RNA target preference has not been observed in other eAgos or pAgos homologs, as characterized eAgos use gRNAs to cleave RNA targets and characterized pAgos always prefer to cleave DNA targets. Cleavage of the target strand by MbpAgo occurs at a single site after the 10th nucleotide counting from the 5 -end of the 5 P-gDNA, consistent with previously characterized Ago homologs (Supplementary Figure S1F and 1G) (14) . Surprisingly, when guided with 5 OH-gDNA, target cleavage also only occurred between target position 10 -11 relative to the 5 -end of the guide (Supplementary Figure S1F and 1G), different from that catalyzed by LrAgo (17) and KmAgo (20) , as they cleave beyond the canonical cleavage site when guided with 5 OH-gDNA. No cleavage activity was detected for a catalytically dead MbpAgo variant with substitutions of two of the tetrad residues (D566A/D635A; Figure 1C ). Moreover, the nucleic acids that co-purify with MbpAgo after expression in E. coli were analyzed ( Supplementary Figure S1H) . Although no small DNAs were observed, DNAs longer than 45 nt and RNAs of undefined length were found in association with MbpAgo. Previous studies showed that no small DNAs were found in association with the catalytically dead version of TtAgo and SeAgo, demonstrating that their catalytic activity is essential for small DNA biogenesis (4, 19) . MbpAgo might not acquire short DNA as guides in E. coli for its weak DNA catalytic activity. Future studies are necessary to explore the in vivo function of MbpAgo.

In summary, in contrast to previously characterized pAgos that prefer to cleave DNA targets, MbpAgo prefers to cleave RNA targets rather than DNA targets and can potentially be exploited for RNA manipulation.

Previous studies indicated that the presence or absence of 5 P may affect cleavage efficiency and cause the cleavage site to shift relative to the canonical cleavage site (17, 22) , but MbpAgo can cleave RNA targets with no obvious difference between 5 P-gDNA and 5 OH-gDNA ( Figure 1C and Supplementary Figure S1G) . Furthermore, previous studies showed that the guide length could also cause the same effect on cleavage efficiency and precision (21) . Thus, MbpAgo cleavage efficiency was first investigated using 5 P-gDNA and 5 OH-gDNA of different lengths. MbpAgo was most active with 14-21 nt for 5 P -gDNA and 15-18 nt for 5 OH-gDNA, with a lower efficiency observed with longer or shorter guides (Figure 2A and B). Similar to KmAgo, the cleavage positions were shifted if shorter (11-13 nt) and longer (21-40 nt) 5 P-gDNAs were used (21) . Interestingly, although the cleavage position was shifted if shorter (10-14 nt) or longer (18-30 nt) 5 OH -gDNA were used, the cleavage occurred only between the 10th and 11th guide positions when using 15 to 17 nt long 5 OH-gDNA, and the same occurred with 14 to 21 nt long 5 P-gDNA ( Figure 2A) .

In this experiment, a change in the guide length and 5group (5 P or 5 OH) can affect the target cleavage efficiency and even the precision of the cleavage site, also found in some other pAgos, such as LrAgo and KmAgo (17, 20) . To further explore the catalytic properties, the kinetics of RNA cleavage by MbpAgo were analyzed with 14 and 18 nt gDNA, which had a markedly different cleavage efficiency or precision compared to 16 nt gDNA (Figure 2A ). Under single-turnover conditions, the observed rates (k obs ) of RNA cleavage with 5 P-gDNA were faster than those with 5 OH-gDNA, and this distinction was bigger in the 14 and 18 nt guides than in the 16 nt guides (Figure 2C-E; Supplementary Figure S2A -D). For 5 P-gDNA, the k obs value of the 16 nt (0.539 ± 0.042 min -1 ) was almost identical to that of the 18 nt (0.517 ± 0.057 min -1 ), but almost two times faster than that of 14 nt (0.281 ± 0.032 min -1 ). Besides, 18 nt gDNA had an adverse effect on cleavage activity under multiple-round conditions, which is discussed in the next section. However, for 5 OH-gDNA, the k obs value of the 16 nt (0.376 ± 0.043 min -1 ) was markedly faster than those of 18 nt (0.197 ± 0.030 min -1 ) and 14 nt (0.097 ± 0.011 min -1 ). Overall, the most appropriate length of the guides is 16 nt for MbpAgo, and both 16 nt 5 P-gDNA and 5 OH-gDNA direct effective and accurate cleavage of RNA targets. Therefore, most experiments were performed with 16 nt guides.

To further determine the prerequisites for MbpAgomediated target cleavage, the influence of divalent cation type, temperature, and [MbpAgo-gDNA]/RNA target ratios was tested. As divalent metal ions are crucial for Ago protein activity, we test which divalent cations MbpAgo can utilize to mediate DNA-guided RNA target cleavage. MbpAgo was active with Mn 2+ , Mg 2+ , Ni 2+ , and Co 2+ as cations, with Mn 2+ and Mg 2+ giving higher activity than the others ( Figure 3A; Supplementary Figure S3A ). Titration of Mn 2+ and Mg 2+ ions showed that, whether the guide is 5 P-gDNA or 5 OH-gDNA, MbpAgo was active at Mn 2+ concentrations ≥ 0.01 mM and showed increased cleavage activity at Mn 2+ concentrations > 1.0 mM (Supplementary Figure S3B) . However, MbpAgo was active at Mg 2+ concentrations ≥ 0.2 mM and showed increased cleavage activity at Mg 2+ concentrations > 5.0 mM (Supplementary Figure  S3C) . Thus, MbpAgo-mediated cleavage was more efficient the in presence of Mn 2+ .

Analyses of the temperature-dependent RNA cleavage activity revealed that MbpAgo bound to 5 P-gDNA displayed comparable levels of RNA cleavage activity between 25 • C and 60 • C and retained good target RNA cleavage activity at 4-20 • C ( Figure 3B ). The observed absence of differences in the efficiencies of RNA cleavage between 25 • C and 60 • C is most likely explained by the fact that almost all target is already cleaved at 25 • C, so a further increase in the cleavage rate with increasing temperature could not be detected. In the case of 5 OH-gDNA, MbpAgo cleavage activity increased from 0 to 50 • C and decreased at >50 • C ( Figure 3B; Supplementary Figure S3D ), suggesting that interactions with the 5 P may be essential in stabilizing the binary MbpAgo-guide complex at elevated temperatures (17) . Therefore, although MbpAgo originates from a psychrotolerant bacterium, it is an active DNA-guided RNA nuclease at physiological temperatures and even sufficiently stable to cleave RNA at elevated temperatures.

To investigate the substrate turnover kinetics of MbpAgo, the cleavage in a time course was monitored using variable [MbpAgo-gDNA]/RNA target ratios at 37 • C ( Figure  3C ). Two-phase kinetics with a rapid burst in the initial phase was followed by a slower linear steady state at 37 • C under multiple-turnover conditions (when the [MbpAgo-gDNA]/RNA target ratio was < 1.0), which were noticeably slower than the rate of RNA cleavage under singleturnover conditions (when the [MbpAgo-gDNA]/RNA target ratio was > 1.0). According to previously studied Ago proteins (17, 18, 21, 34) , the initial burst phase corresponds to the first round of catalysis, following the formation of the ternary pAgo/gDNA/RNA target complex. In contrast, the slow linear phase corresponds to the subsequent rounds of catalysis, after the pAgo-gDNA complex dissociates from the first substrate and rebinds another. From the burst amplitude in the reaction in which the molar ratio of the MbpAgo-gDNA complex to target is 0.25:1, the concentration of active binary MbpAgo/gDNA complexes was estimated as 40 nM (of 50 nM MbpAgo taken in the reaction), corresponding to ∼80% of active MbpAgo in the preparations. The steady-state velocity of RNA cleavage by MbpAgo under these conditions was 0.6436 nM × min −1 [95% confidence interval (95% CI): 0.4593-0.7967 nM × min −1 ], corresponding to the k obs value of 0.0161 min −1 (based on the 40 nM effective concentration of the binary complex). Then the substrate turnover kinetics experiment was performed with 18 nt gDNA at 37 • C ( Figure 3D ). From the burst amplitude in the reaction in which the molar ratio of the MbpAgo-gDNA com- plex to target is 0.25:1, the concentration of active binary MbpAgo-gDNA complexes (44 nM) was comparable with the concentration for 16 nt gDNA (40 nM). Interestingly, the steady-state velocity of MbpAgo-mediated target RNA cleavage directed by 18-nt gDNA (0.3121 nM × min −1 ; 95% CI: 0.2166 -0.4008 nM × min −1 ), corresponding to the k obs value of 0.0071 min −1 (based on the 44 nM effective concentration of the binary complex), was two times slower than the velocity measured in the same setup using a 16 nt gDNA. This suggested that the velocity of the binary complex dissociation from the target substrate is decreased in the case of longer gDNAs. In addition, almost all RNA target was cleaved within 60 min when the reaction was performed at 50 • C, indicative of a multiple-round reaction ( Figure 3E and F). In conclusion, although MbpAgo functions as a multiple-turnover enzyme, its steady-state rate may be limited by product release and exchange of target molecules, such as CbAgo and KmAgo at 37 • C (18, 21) .

Previous studies demonstrated that the 5 -nucleotide of the guide strand is bound in the MID pocket of Ago, and many studied eAgos and pAgos have a certain bias for it (5) . To determine whether MbpAgo has specificity for the 5nucleotide of gDNAs, four guide variants were tested with different 5 -nucleotides but otherwise identical sequences. Like in case of CbAgo and KmAgo, there were no obvious changes in cleavage efficiency and rate when MbpAgo was loaded with gDNAs with different 5 -nucleotides (Figure 4A and B, Supplementary Figure S4A ) (17, 21) .

Mismatches between the guide and target affect the cleavage activities of Ago proteins (17, 20, 32) . To check the mismatch tolerance of MbpAgo, the effects of mismatches between the guide and target strands on its RNA cleavage activity were analyzed. A set of gDNAs was designed, each containing a single-nucleotide or dinucleotide mismatches at a certain position (Supplementary Table S2 ), and tested them in the RNA cleavage reaction with MbpAgo ( Figure  4C; Supplementary Figure S4B ). When a single-nucleotide mismatch was introduced, mismatches at positions 8 and 9 and 11-15 affected the cleavage efficiency. However, a dramatic decrease in cleavage efficiency was not observed, and only mismatches at positions 9 and 13 reduced cleavage efficiency to ∼50%. Furthermore, although most dinucleotide mismatches affected cleavage efficiency, dinucleotide mismatches at positions 11-13 dramatically reduced target cleavage. Thus, MbpAgo has a high tolerance for mismatches between the guide and target strands.

Previous structural and bioinformatic studies showed that the vast majority of pAgos contain the MID and PIWI domains to anchor the 5 -end of a guide by a set of conserved amino acid residues, which constitute a conserved six-amino-acid motif (8) . The first two residues of this mo-tif are the most conserved, they might be important for the function of different groups of pAgos, by which pAgos can be classified into several subtypes, including YK, HK, RK, and MID-OH (8) . The multiple sequence alignment of the 5 -end guide binding pocket of the MID domain from MbpAgo with several other characterized Ago proteins showed that the first two residues of most characterized Ago proteins are YK, whereas the first two residues of MbpAgo are HK (H482 and K486; Supplementary Figure  S5A ). Furthermore, the three-dimensional (3D) model built using SWISS-MODEL revealed that the first two amino acid residues in the specific motif involved in the interaction with the guide 5 -end consistent with the multi-sequence alignment results and interactions of the guide 5 end with the MID pocket are similar overall for MbpAgo and CbAgo ( Figure 5A and Supplementary Figure S5A) . Analysis of the 5 -binding pocket in the MID domain of MbpAgo revealed H482 and K486 in the specific motif involved in interactions with the Me 2+ ion and the guide 5 P similar to Y472 and K476 in CbAgo, respectively ( Figure 5A ) (17) .

These results indicated that MbpAgo could use 5 OH-gDNAs to cleave RNA targets, with only slightly lower efficiency than 5 P-gDNAs of identical sequence ( Figure 2D ; Supplementary Figure S2D ). The K d value was further measured for guide binding by MbpAgo using an EMSA (Figure 5B; Supplementary Figure S5B and C). In the presence of Mn 2+ , the K d value of MbpAgo associated with a 5 P-gDNA (111.2 ± 3 nM) was lower than that for complexes composed of MbpAgo and 5 OH-gDNA (147.7 ± 5.8 nM), which may be the reason why MbpAgo prefers to use 5 P-gDNAs for efficient target RNA cleavage. Interestingly, the K d value for binding 5 OH-gDNA in the absence of Mn 2+ (161.5 ± 4.8 nM) was slightly higher than the K d value measured in the presence of Mn 2+ . In contrast, the K d value for binding 5 P-gDNA in the absence of Mn 2+ (205.4 ± 7.6 nM) was relatively higher than the K d value measured in the presence of Mn 2+ . Meanwhile, the thermostability of MbpAgo was also measured using circular dichroism ( Figure 5C ; Supplementary Figure S5D -F). gDNA binding remarkably improves the thermostability of MbpAgo, similar to hAgo2 and KmAgo (10, 20) . MbpAgo associated with 5 P-gDNAs has higher thermostability than 5 OH-gDNAs in the presence of Mn 2+ , which may explain why MbpAgo guided with 5 P-gDNAs can cleave RNA at higher temperatures (Figure 3B) . Moreover, the addition of Mn 2+ can significantly improve the thermostability of MbpAgo associated with 5 P-gDNAs but has no significant effect on the thermostability of MbpAgo associated with 5 OH-gDNAs. Both the gDNA-binding kinetics and thermal denaturation analyses suggested that divalent metal ions are important for MbpAgo to bind the 5 P group of the guide. The K d value for target binding was also measured by the MbpAgo-gDNA complex using an EMSA ( Figure 5D ; Supplementary Figure S5G and H) . The K d value of MbpAgo-gDNA complex binding RNA targets is 207.22 ± 22 nM, lower than the K d value of MbpAgo-gDNA complex binding DNA targets (322.1 ± 35.5 nM). The higher affinity to RNA targets might be one of the reasons why MbpAgo prefers RNA tar- gets. The real K d values for guide binding by MbpAgo and target binding by the MbpAgo-gDNA complex are probably lower, given that the effective concentration of MbpAgo is < 100%, owing to the detection limits of measurements.

To determine the role of HK in the specific motif, HK was substituted with YK, RK, AK, HA or AA and the resulting MbpAgo variants were tested in cleavage assays. When guided with 5 P-gDNA, the cleavage efficiency and k obs of the variants were comparable to the wild-type (WT) MbpAgo ( Figure 6A WT MbpAgo had a slightly higher binding affinity for 5 P-gDNA than other variants, except for the YK variant (Figure 6C; Supplementary Figure S7A ). However, when guided with 5 OH-gDNA, although the cleavage efficiency and k obs of RK, AK, HA and AA variants were comparable to the WT, the k obs value of the YK variant increased compared with that of the WT (Figure 6D and E; Supplementary Figure S6A-6F ). In addition, for the YK variant, the cleavage efficiencies with 5 P-gDNA and 5 OH-gDNA as well as the k obs were similar, which is not the case for all other variants. The binding affinity of the YK variant to 5 OH-gDNA was ∼25% higher than that of the WT for the same 5 OH-gDNA ( Figure 6F; Supplementary Figure S7B ), which might be one of the reasons why the YK variant has a higher cleavage activity with 5 OH-gDNA. Although previously studied pAgos with YK or RK in the motif have good activity for DNA targets, the DNA target cleavage efficiency of these MbpAgo variants did not change significantly compared to WT (Supplementary Figure S6G ). This suggested that the preference of MbpAgo for RNA targets may not be related to the motif. Moreover, the cleavage sites of WT, YK and RK variants guided with 14 or 18 nt 5 P-gDNAs occurred only between target position 10 -11 relative to the guide 5end. Still, the cleavage sites of HA, AK and AA variants guided with 14 or 18 nt 5 P-gDNAs were shifted ( Figure  6G ). Overall, these results suggested that the HK in the mo- tif indeed was involved in interactions with the 5 -end of a guide.

Considering that MbpAgo guided with both 5 P-gDNAs and 5 OH-gDNAs could cleave RNA with high efficiency, this study next examined whether the MbpAgo-guide complex can cleave target sequences in a highly-structured RNA that contains a diverse set of conformational features, such as varying lengths of helices, bulges, hairpin loops, and single-stranded regions. Dayeh et al. predicted the secondary structure of HIV-1 DIS 5-UTR with SHAPE and designed a set of gDNAs to span the HIV-1 DIS 5-UTR sequence in 23-nt increments (Supplementary Figure  S8A) (32) . As 45-nt unstructured RNA targets are cleaved best when gDNAs are 16 nt long (Figure 2A Figure S8A ). Cleavage products were detected with most 5 P-gDNAs and 5 OH-gDNAs, except gDNA-1 located within the highly-stable stems of the transactivation response (TAR; nt 1-57) element when reactions were performed with 5 mM Mn 2+ (Figure 7B and C) . Then, we tested whether MbpAgo can also cleave highlystructured RNAs under Mg 2+ , because Mg 2+ is the recommended divalent metal ion in the reaction system of some RNA manipulation tool enzymes (such as reverse tran- scriptase RNA polymerase, and RNA ligase). The results showed that MbpAgo can cleave the highly-structured RNA at the expected position with most gDNA except gDNA-1 with 5 mM Mg 2+ , albeit a little less efficiently (Supplementary Figure S8B and C). This suggested the possibility of the combined use of MbpAgo and these RNA manipulation tool enzymes. Besides, in vitro cleavage of a double-stranded RNA target could not be detected (Supplementary Figure  S8D , lanes 5 and 6). Thus, similar to KpAgo (Kluyveromyces polysporus) and KmAgo (21, 32) , the MbpAgo-gDNA complex can cleave the target sequences in highly-structured RNAs, and the cleavage efficiency is modulated by the secondary RNA structure.

This study characterized a novel pAgo from the psychrotolerant bacterium M. paludis. Compared to known Agos, MbpAgo has an unusual preference for cleaving RNA targets with high efficiency at moderate temperatures but has very weak activity in cleaving DNA targets. Previously, most characterized mesophilic pAgo proteins strongly prefer DNA targets, and have only very weak or undetectable RNA target cleavage activity, including CbAgo, LrAgo, SeAgo, CpAgo and IbAgo (17) (18) (19) 22) . KmAgo can efficiently and precisely cleave RNA targets, but KmAgo also prefers to cleave DNA targets (20) . Therefore, MbpAgo is the first example of a pAgo protein that prefers to cleave RNA targets at moderate temperatures. Furthermore, MbpAgo can target RNA under a wide range of reac-tion conditions. The efficiency and accuracy of cleavage are modulated by temperature, divalent ions, and the phosphorylation and length of gDNAs and their complementarity to the RNA targets.

MbpAgo can utilize both 16 nt 5 OH guides and 5 P guides for efficient RNA target cleavage. MbpAgo is most active with 14-21 nt lengths for 5 P-gDNA and 15-18 nt lengths for 5 OH-gDNA, with a lower cleavage efficiency observed with shorter or longer guides, which is similar to that of other studied pAgos. Although some mesophilic pAgos, including CbAgo and KmAgo, were reported to can use 5 P guides for RNA target cleavage, almost no 5 OH guide-mediated RNA target cleavage activity was detected (17, 20) . Moreover, target cleavage by most eAgos and pAgos, including hAgo2, CpAgo, IbAgo, LrAgo and KmAgo, with 5 OH guides resulted in a shift of the cleavage site compared to cleavage using 5 P-gDNAs (17, 20, 22, 35) . For MbpAgo, although the cleavage site is shifted if shorter or longer guides are used, the cleavage position is no longer shifted with 15 to 17 nt long 5 OH guides. Thus, except for the 5 P that can help to determine the correct register of the guide-target duplex relative to the active site of pAgo, the length of the guide can also modulate the cleavage efficiency and position.

MbpAgo utilizes Mn 2+ and Mg 2+ as cations, and MbpAgo-mediated cleavage is more efficient in the presence of Mn 2+ . MbpAgo can cleave RNA target at a wide range of temperatures and still has good RNA cleavage activity at 4-20 • C, probably because MbpAgo is from a psychrotolerant bacterium. This suggested that pAgos with good activity at mesophilic temperatures could also be found from psychrotolerant and psychrophilic bacteria. Furthermore, MbpAgo cannot use 5 OH guides to cut targets at 60 • C. This indicated that MbpAgo loaded with 5 P-gDNAs may be more stable than that with 5 OH-gDNAs, as demonstrated by measuring the thermostability of MbpAgo using circular dichroism. Several Agos have strong sequence preferences for the 5 -nucleotide of a guide, including KpAgo and MjAgo (32, 36) . However, MbpAgo has no obvious preference for the 5 -nucleotide of a guide, which is similar to several other pAgos, including KmAgo and CbAgo (17, 20) . Previous studies on eAgos and pAgos demonstrated the importance of complementarity between the guide and the target for efficient repression. MbpAgo has a high tolerance for mismatches between the guide and target strands. However, only activity comparisons were performed at a single time point. The real effects of mismatches at other positions may be higher in different conditions (e.g. shorter incubation times). Given the weak effects of mismatches on MbpAgo dependent cleavage, MbpAgo can potentially be used to detect or clear the RNA virus, because the virus will not be able to escape easily by mutating single bases.

Except for several pAgos, including MpAgo, which exclusively bind 5 OH guides (13) , most eAgos and pAgos were shown to use 5 P guides for RNA cleavage, and multiple interactions between the 5 P group and the MID domain are observed in the structure of some Ago-guide complexes (5, 18) . A bioinformatic study revealed several subtypes of the MID domain, with substitutions of key residues involved in interactions with the 5 -end of guide molecule (8) . The MID domain of most Agos, including CbAgo, RsAgo, and KmAgo, contains the YK subtype, whereas MbpAgo contains the HK subtype (Supplementary Figure S5A) . As far as we know, MbpAgo might be the first studied pAgo with the HK subtype in the motif. Although substituting HK with YK, RK, AK, HA or AA has no obvious effect on 16 nt 5 P-gDNA mediated RNA cleavage activity, a single alanine mutation in the HK motif affects the 5 P-gDNA binding affinity and the 14 and 18 nt 5 P-gDNA mediated cleavage precision. Thus, H482 and K486 in the motif contribute to anchoring the 5 phosphate of 5 P-gDNA to ensure the cleavage precision. Furthermore, substituting HK with YK can increase the 5 OH-gDNA mediated RNA cleavage activity, and substituting HK with RK, AK, HA or AA has little effect on 16-nt 5 OH-gDNA mediated RNA cleavage activity. In addition, the affinity of MbpAgo for 5 P-gDNA is 30% higher than the affinity of MbpAgo for 5 -OH-gDNA. The actual differences may be much higher because of the limitations of the method. Interactions with other parts of the guide may be important for stabilizing the 5 OH guide in MbpAgo to ensure that MbpAgo can use 5 OH-gDNA for effective cleavage. Furthermore, the affinity of the MbpAgo-gDNA complex for RNA targets is higher than that for DNA targets, which may be one of the reasons why MbpAgo prefers to cleave RNA targets. Thus, structural studies are necessary to determine the structural basis of the apparent preference for RNA targets.

Finally, we have demonstrated that MbpAgo can efficiently cleave highly-structured RNA targets using both 5 P-gDNAs and 5 OH-gDNAs in the presence of Mg 2+ or Mn 2+ . Previously, eAgo from the budding yeast K. polysporus and pAgo from the mesophilic bacterium K. massiliensis were used for highly-structured RNA cleavage (20, 21, 32) . Similar to these findings, the cleavage efficiency of MbpAgo is modulated by the secondary structure at 37 • C. However, using MbpAgo to cleave RNA is more convenient and cost-effective, as the synthesis of 5 OH-gDNAs is easier and more inexpensive than 5 P-gDNAs. Thus, MbpAgo can potentially be applied in RNA-centric in vivo and in vitro methods, such as RNA targeting, antiviral and nucleic acid detection (37, 38) . In conclusion, MbpAgo is a unique programmable nuclease with a strong preference for RNA targets and can potentially be used in RNA manipulations.

Supplementary Data are available at NAR Online. 

The evolutionary journey of Argonaute proteins

Prokaryotic homologs of Argonaute proteins are predicted to function as key components of a novel system of defense against mobile genetic elements

DNA-guided DNA interference by a prokaryotic Argonaute

Argonaute of the archaeon Pyrococcus furiosus is a DNA-guided nuclease that targets cognate DNA

RNA interference and beyond: structure and functions of prokaryotic Argonaute proteins

DNA targeting and interference by a bacterial Argonaute nuclease

Thermus thermophilus Argonaute functions in the completion of DNA replication

The expanded universe of prokaryotic Argonaute proteins

Prokaryotic Argonaute proteins: novel genome-editing tools?

The structure of human argonaute-2 in complex with miR-20a

Structure-based cleavage mechanism of Thermus thermophilus Argonaute DNA guide strand-mediated DNA target cleavage

Structure of yeast Argonaute with guide RNA

A bacterial Argonaute with noncanonical guide RNA specificity

Argonaute proteins: structural features, functions and emerging roles

Guide-independent DNA cleavage by archaeal Argonaute from Methanocaldococcus jannaschii

Structural and mechanistic insights into an archaeal DNA-guided Argonaute protein

Programmable DNA cleavage by ago nucleases from mesophilic bacteria Clostridium butyricum and Limnothrix rosea

DNA-guided DNA cleavage at moderate temperatures by Clostridium butyricum Argonaute

Genome-wide DNA sampling by ago nuclease from the cyanobacterium Synechococcus elongatus

A programmable omnipotent argonaute nuclease from mesophilic bacteria Kurthia massiliensis

A programmable pAgo nuclease with universal guide and target specificity from the mesophilic bacterium Kurthia massiliensis

Argonaute proteins from human gastrointestinal bacteria catalyze DNA-guided cleavage of single-and double-stranded DNA at 37 • C

Structure of an argonaute silencing complex with a seed-containing guide DNA and target RNA duplex

Highly specific enrichment of rare nucleic acid fractions using Thermus thermophilus argonaute with applications in cancer diagnostics

Pyrococcus furiosus Argonaute-mediated nucleic acid detection

PfAgo-based detection of SARS-CoV-2

Pyrococcus furiosus Argonaute coupled with modified ligase chain reaction for detection of SARS-CoV-2 and HPV

Argonaute proteins: structures and their endonuclease activity

Mucilaginibacter paludis gen. nov., sp. nov. and Mucilaginibacter gracilis sp. nov., pectin-, xylan-and laminarin-degrading members of the family Sphingobacteriaceae from acidic Sphagnum peat bog

PCR-mediated site-directed mutagenesis

Argonaute-based programmable RNase as a tool for cleavage of highly-structured RNA

Binding and cleavage specificities of human Argonaute2

Kinetic analysis of the RNAi enzyme complex

Purified Argonaute2 and an siRNA form recombinant human RISC

Mechanistic insights into archaeal and human argonaute substrate binding and cleavage properties

RNA targeting with CRISPR-Cas13

Development of CRISPR as an antiviral strategy to combat SARS-CoV-2 and influenza