Structural insights into C-terminus-mediated RNA target cleavage by a mesophilic prokaryotic argonaute

Abstract

Prokaryotic Argonaute proteins (pAgos) are programmable nucleases that always utilize DNA guides to cleave DNA targets. Recent studies show that some pAgos preferentially utilize DNA guides to cleave RNA targets rather than DNA targets. VbAgo, derived from a Verrucomicrobia bacterium, is a nuclease capable of specifically cleaving single-stranded RNA and highly structured RNA substrates at 37 °C, making it an ideal candidate for developing RNA manipulation toolkits. An in-depth investigation of its mechanism contributes to understanding the functional characteristics of gDtR-type Ago proteins. Here, we present cryo-electron microscopy structures of VbAgo, the VbAgo-guide DNA binary complex, multiple wild-type VbAgo-guide DNA-target RNA ternary complexes, and the catalytically inactive mutant (VbAgo-DM) guide DNA-target RNA ternary complex, with resolutions ranging from 2.5 to 3.2 Å. By integrating these cryo-EM structures with biochemical data, we elucidate the entire catalytic process of VbAgo, revealing its unique C-terminal regulatory mechanism. Specifically, in its apo state, VbAgo’s C-terminus occupies the nucleic acid binding channel, partially impeding its catalytic activity while enhancing its stability. The binding of guide DNA displaces the C-terminus, and subsequent binding of target RNA, along with conformational changes in the N-terminal and PAZ domains, facilitates VbAgo dimerization. Following this, the C-terminus transitions from a loop to a helix, enabling maturation of the catalytic center and inducing movements in the MID-PIWI’ interactions at the dimer interfaces, ultimately leading to dimer dissociation. Concurrently, cleavage of the target RNA and subsequent product release occur, after which VbAgo reverts to its binary state to initiate the next cleavage cycle. Moreover, we demonstrate that VbAgo exhibits guide DNA mediated RNA knockdown activity in mammalian cells. In summary, our study provides a comprehensive understanding of the molecular mechanisms governing self-inhibition, guide binding, target recognition, and product release in VbAgo. These findings offer valuable insights into the diverse mechanisms of pAgos, broadening their functional scope and enhancing the biotechnological potential of pAgo proteins.

Prokaryotic Argonaute proteins are programmable nucleases. Here, the authors capturing structures of VbAgo at functional stages to show how its C terminus acts as a regulator and demonstrate its ability to cleave RNA in mammalian cells.

Introduction

Argonaute (Ago) proteins are a class of nucleic acid-guided programmable nucleases that play essential roles in DNA/RNA interference pathways^1,2. In eukaryotes, Ago proteins (eAgos) act as core components of the RNA-induced silencing complex (RISC), contributing to gene regulation, defense against RNA viruses, and the maintenance of genome stability³. Genomic analysis reveals that prokaryotic Ago proteins (pAgos) exhibit greater diversity compared to their eukaryotic counterparts and can be classified into three subtypes: long-A pAgos, long-B pAgos and short pAgos⁴. Unlike eAgos, which exclusively target RNAs⁵, pAgos demonstrate broader substrate-binding and cleavage capabilities, including four distinct modes: DNA-guided DNA cleavage (gD-tD), DNA-guided RNA cleavage (gD-tR), RNA-guided DNA cleavage (gR-tD), and RNA-guided RNA cleavage (gR-tR)⁶.

Early studies primarily focused on thermophilic long-A pAgos, such as TtAgo from Thermus thermophilus, PfAgo from Pyrococcus furiosus, and MjAgo from Methanocaldococcus jannaschii. These enzymes exhibit optimal activity within the temperature range of 55–99.9 °C and predominantly target DNA substrates^7–9. The catalytic mechanisms of these enzymes have been well elucidated^10,11. Nonetheless, their reliance on high temperatures and strict substrate specificity limits their applicability in in vivo genome editing and diagnostic technologies¹². Besides, short pAgos require auxiliary proteins encoded within the same operon to form functional complexes for bacterial immunity^4,13,14. However, their activation triggers collateral cleavage activity, and detection relies on signal transduction processes. Furthermore, the inherent complexity of these integrated detection systems poses significant challenges for tool development^15,16. Consequently, research efforts have intensified in the search for mesophilic candidates capable of cleaving nucleic acid substrates at moderate temperatures^17–22. Notably, although some exhibit dual-substrate (DNA/RNA) cleavage activity, the majority primarily target DNA, exhibiting cleavage patterns similar to those of thermophilic pAgos. Thus, the mechanisms underlying RNA-targeting capabilities remain relatively poorly understood.

Recent studies have identified two gD-tR type pAgos (the PliAgo and PnyAgo groups) with unique RNA-targeting preferences. However, their practical applications are constrained by several challenges, including difficulties in protein purification, cleavage sites variability, and residual DNA cleavage activity. Structural predictions have revealed an extended C-terminus in both groups. Biochemical data demonstrate that amino acid modifications in this region do not affect guide binding, whereas truncation mutations in certain Ago proteins markedly reduce target RNA cleavage efficiency^22,23. These findings suggest another catalytic mechanism within these pAgos. Nevertheless, the molecular details remain elusive due to the lack of high-resolution structural data on RNA-targeting pAgos.

In this study, we present a comprehensive mechanistic analysis of VbAgo, a mesophilic pAgo derived from a Verrucomicrobia bacterium, which exhibits RNA-specific cleavage activity at physiological temperatures. Previous studies have shown that VbAgo possesses conserved cleavage sites, insensitivity to guide 5’-end modifications, and robust activity in cleaving highly structured RNA targets at mesophilic temperatures, making it an ideal candidate for RNA-editing tool development²⁴. Utilizing cryo-electron microscopy (cryo-EM), we determine a series of high-resolution structures of VbAgo in multiple states. Combined with extensive biochemical data, we reveal that VbAgo possesses direct RNA-targeting catalytic activity in vivo and elucidated the molecular mechanisms regulating self-inhibition, guide DNA loading, RNA target recognition, and cleavage product release. These findings illustrate the diverse operational principles of pAgos, thereby expanding their functional versatility and enhancing the biotechnological applicability of pAgo proteins.

Results

The acidic C-terminus mediates auto-inhibition in the apo state of VbAgo

Multiple sequence alignment analysis reveals that VbAgo consists of 782 amino acids, a length comparable to other gD-tR type pAgos, and exhibits the highest sequence similarity (21.33%) with MbpAgo (Supplementary Fig. 1a). Within the PIWI domain, VbAgo contains the conserved DEDD catalytic tetrad motif (D550, E584, D617, and D756), supporting its capacity to cleave nucleic acid substrates (Fig. 1a). We first determined the 3.2 Å cryo-EM structure of full-length VbAgo in its apo state (VbAgo_apo) (Supplementary Fig. 2 and Supplementary Table 2). Like most Ago proteins, VbAgo_apo adopts a typical two-lobed architecture. One lobe comprises the conserved MID domain (residues 397–537) and PIWI domain (residues 538–782), while the other consists of the N-terminal domain (residues 1–134) and PAZ domain (residues 188–294). Two linker regions, L1 (residues 135–187) and L2 (residues 295–396), are positioned front and back, respectively, arranged perpendicularly to interweave with the four main domains, thereby enhancing structural connectivity (Fig. 1a, b).

Fig. 1. The C-terminus occupies the nucleic acid-binding channel in the apo state of VbAgo.

a Schematic diagram of the domain organization of VbAgo. Domains and their catalytic residues (DEDD motif) are colored and labeled. C-ter, C-terminus. b On the left, the cryo-EM map of VbAgo_apo is presented, with protein domains colored as in (a). On the right, a close-up view of the acidic C-terminus is provided, with coloring based on electrostatic potential: red indicates strongly electronegative regions, while blue represents more electropositive areas. c Cleavage activity of VbAgo C-terminus truncation mutants (de[truncation]) is shown with wild-type (WT) in gray. Data represent the mean ± SD from n = 3 replicates per group. Pairwise comparisons between groups were performed using Student’s t-tests, with wild-type protein or strains serving as controls. Mutants exhibiting significantly lower activity are shown in cyan, while those with higher activity are shown in purple. Statistical significance was defined as p < 0.05, source data are provided in the Source data file. d Thermal stability analysis of wild-type VbAgo and its C-terminal truncation mutants with higher cleavage activity in (c) was conducted from 37 to 70 °C. The assay used a 5′-FAM-labeled 45 nt RNA as the substrate and produced 34 nt 5′-FAM-labeled products upon cleavage.

Interestingly, VbAgo features an extended C-terminal segment (residues 736–782) embedded within its PIWI domain, forming a disordered loop that interacts with all four structural domains. In all previously structurally characterized long pAgo and eAgo structures, the C-terminus is located at the interface between the MID and PIWI domains. While most adopt short, ordered helices, only KmAgo displays a comparably flexible loop²⁵ (Supplementary Fig. 8a). In contrast, VbAgo’s acidic C-terminus is sequestered within a positively charged channel, suggesting a potential regulatory role through nucleic acid mimicry (Fig. 1b). It interacts with various domains via the formation of salt bridges (D743-R520, E745-R381, R760-E298, E761-K488, E778-R231), hydrogen bonds (H736-Y380, H736-Y377, T751-H615, S755-D550, H758-N521, E761-Q497, E761-Y484, D770-T296, E775-H656), stacking interactions (W739-P534, W754-R573), and a putative disulfide bond between C780 and C281 (Supplementary Fig. 8b). Such auto-inhibition has previously been observed exclusively in the short pAgo SPARTA^26–29. However, SPARTA employs an ordered α-helix, whereas VbAgo, a representative long pAgo, utilizes an extended, disordered loop.

To further investigate the function of the C-terminus, we designed several C-terminal truncation mutants (de[truncation]) and systematically assessed their thermal stability and cleavage activity relative to the wild type. Surprisingly, truncations within residues 759–762 resulted in diverse cleavage patterns, including complete loss, reduction, or enhancement (Fig. 1c). Given that VbAgo functions as a multiple-turnover enzyme²⁴, we further quantified the observed rate constant (k_obs) under both single- and multiple-turnover conditions. The linear time course of product formation, without an initial burst phase, indicates that the catalytic rate remains constant across successive turnover cycles, suggesting that product release is not rate-limiting. Consistent with the cleavage activity pattern, kinetic analysis revealed that de[764–782] exhibits a significantly accelerated observed rate constant (k_obs) compared to wild-type VbAgo, whereas de[761–782] shows a markedly reduced k_obs at 37 °C (Supplementary Fig. 9). The cleavage activity turning points at positions 759 and 761 require additional structural information for analysis, which we will discuss in the following sections.

In thermal stability analysis, the de[764–782] mutant exhibited the largest increase in cleavage activity but experienced a 5 °C decrease in thermal stability compared to the wild type (Fig. 1d). This inverse correlation suggests that the C-terminal region stabilizes key intramolecular interactions—such as salt bridges, stacking contacts, and hydrogen bonds—thereby enhancing global conformational rigidity and thermal stability. However, this rigidity restricts also restricts essential conformational dynamics at the catalytic center, limiting catalytic efficiency. Removal of the C-terminus alleviates the auto-inhibitory conformation, partially disrupting the stabilizing network, increasing conformational flexibility, and accelerating RNA cleavage kinetics at 37 °C. The observed thermodynamic destabilization results directly from the enthalpic penalty associated with loss of favorable intramolecular interactions upon truncation³⁰. Collectively, our data indicate that the C-terminus not only plays a critical role in modulating the cleavage activity of VbAgo but also contributes to its thermal stability.

The C-terminus flips upward to facilitate guide binding

To investigate the conformational changes of the C-terminus within cleavage cycle, we next resolved the structure of the binary complex formed by VbAgo with guide DNA, termed VbAgo_gDNA, at an overall resolution of 3.0 Å (Fig. 2a, Supplementary Fig. 3, and Supplementary Table 2). Compared to the apo state, where we were able to build a full-length atomic model, we could only trace the structure up to valine at position 763 in the binary state (Fig. 2b). Notably, L759 faces a relatively hydrophobic region comprising residues P400, L402, L470, L494, A495, A529, A530, and L532, while R760 is positioned at the interface between the MID and PIWI domains, oriented outward without interacting with any amino acids or nucleic acids (Fig. 3f, left panel).

Fig. 2. Binary structure of VbAgo in complex with guide DNA.

a Cryo-EM density map (left) and ribbon diagram (right) of VbAgo_gDNA with domains colored as VbAgo_apo. The density of guide DNA is indicated by a blue color. b On the left, structural comparison between VbAgo_apo (gray) and VbAgo_gDNA binary complex (colored). The movements of the MID domain upon gDNA binding are indicated by black arrows. On the right, a magnified view of the C-terminus comparison in VbAgo from the same perspective, with gDNA shown as a blue stick representation. The steric clash is colored in red. The C-terminus of apo is shown in orange, the binary C-terminus is displayed in cyan, and the curved black arrow illustrates the conformational flipping of the C-terminus. c Structure of the guide DNA in VbAgo_gDNA is shown with cryo-EM density as a mesh, with four kink positions marked by red Roman numerals on a yellow background. d Structure of the guide DNA in PliAgo_gDNA (PDB: 7R8H), MjAgo_gDNA (PDB: 5G5T), TtAgo_gDNA (PDB: 3DLH) and KmAgo_gDNA (PDB: 8XHV) binary complexes. e Close-up views of the four kink positions correspond to (c). The nucleic acid sequence of the guide DNA is presented in the upper panel, with the kink positions highlighted against a yellow background. Detailed interactions between the guide strand and VbAgo are illustrated in the lower panel. f Cleavage activity of VbAgo and its mutants with alanine substitutions at residues responsible for gDNA kinks. Data represent the mean ± SD from n = 3 replicates per group. Pairwise comparisons between groups were performed using Student’s t-tests, with wild-type protein or strains serving as controls. Wild-type (WT) and those with no significant effect on cleavage activity are shown in gray, cyan indicates mutants with significantly lower cleavage activity than WT, and purple indicates mutants with significantly higher cleavage activity than WT. Statistical significance was defined as p < 0.05, source data are provided in the Source data file.

Fig. 3. Ternary structures of VbAgo_gDtR captured in distinct states and conformational dynamics of the C-terminal.

a Cryo-EM density maps reveal three distinct conformational states of the ternary complex. (Left) Dimer-VbAgo_gDtR, with the protein domains colored wheat (promoter A) and teal (promoter B). (Middle) Monomer-VbAgo_gDtR, depicted in orchid. (Right) Released-VbAgo_gDtR, shown in purple. At the top, a schematic illustrates the base pairing between guide DNA (blue) and target RNA (dark red), with black scissors indicating the cleavage site. b Overview of structural changes during the catalytic cycle (color coding as in (a)), explicitly illustrating their temporal sequence with arrows. The C-terminus is highlighted in cyan. c Structural superposition reveals conformational transitions of the N-terminal domain between binary and monomeric ternary complexes, with black curved arrows indicating the direction and angle of rearrangements. d C-terminus conformations in binary and monomeric ternary complexes are shown, with the region (residues 750–762) containing the last catalytic residue D756 highlighted in black. e Cleavage activity of VbAgo and its double mutant (L552G/I557G). Data represent the mean ± SD from n = 3 replicates per group. Pairwise comparisons between groups were performed using Student’s t-tests, with wild-type protein or strains serving as controls. Statistical significance was defined as p < 0.05. Wild-type (WT) and those with no significant effect on cleavage activity are shown in gray, cyan indicates mutants with significantly lower cleavage activity than WT. Source data are provided in the Source data file. f The interaction network associated with C-terminal conformational changes in the binary and monomeric ternary complexes corresponds to the three conformational states depicted in (d).

Upon superposition with the apo form, the C-terminus undergoes significant conformational changes to accommodate the entry of gDNA. The initial segment (residues 736–752) of the corresponding loop region experiences a lateral shift, creating sufficient space to anchor the 5′-end of gDNA (Fig. 2b). Concurrently, several helical segments within the MID domain, namely residues R412-G431, R500-D510, and G516-A530, shift upward by 8–15 Å. Furthermore, starting from residue 753, the C-terminus undergoes a 125.9° upward rotation. These collective structural rearrangements effectively mitigate potential steric clashes between VbAgo and gDNA (Fig. 2b). Therefore, this complete exposure of the nucleic acid-binding channel enables sequential base-by-base downward propagation of gDNA following 5′-end engagement.

VbAgo employs a Mg²⁺-dependent, C-terminus-independent mechanism for 5′-phosphate DNA guide recognition

The entire gDNA strand in VbAgo_gDNA displays a well-defined density map enabling accurate modeling of each nucleotide (Fig. 2c). Similar to most Ago-guide binary structures^25,31–34, the 5′- and 3′-ends of the gDNA in VbAgo_gDNA are anchored within the binding pockets of the MID and PAZ domains, respectively. Currently, the recognition mechanism of the 5′-phosphate (5′-P) of the guide by the MID domain exhibits significant variation. Eukaryotic Argonaute proteins, such as hAgo2 (PDB: 4OLA) and KpAgo (PDB: 4F1N), neutralize the charge between the 5′-terminal phosphate of the guide RNA and the carboxyl group at the C-terminus of the PIWI domain through a conserved lysine residue^31,35. In contrast, most prokaryotic Argonaute proteins and PIWI-Clade Argonaute proteins that prefer 5′-P guides, such as Drosophila Piwi (PDB: 6KR6) and Bombyx mori Siwi (PDB: 5GUH), rely on divalent metal ions to achieve charge neutralization^36,37. Although this mechanism, which depends on a rigid C-terminus, is crucial for stabilizing the 5′-terminal of the guide, it also restricts protein modification or functional fusion in this region, limiting its utility in engineering applications. Conversely, the C-terminus of the gDtR class prokaryotic Argonaute protein is more flexible. For example, the 5′-terminal stability of PliAgo does not depend on the C-terminus, and PliAgo can function without Mg²⁺ ²², whereas VbAgo still requires Mg²⁺ assistance (Supplementary Fig. 11).

Specifically, in the 5’-end binding pocket, the first nucleotide, T1, undergoes a 180° flip, with its phosphate group forming stable hydrogen bonds with the side chains of Y484, K488, and Q497, while its base stacks against the side chains of Y484 and R500. Additionally, a Mg²⁺ ion is centrally positioned within the pocket, coordinating with the phosphate groups of T1, G2, and A3. Moreover, a hydrogen bond forms between the oxygen atom of the phosphate backbone linking T1 and G2 and the hydroxyl group of T503. The base of G2 engages in stacking with R518, further stabilizing the 5′-end of the gDNA (Supplementary Fig. 10).

In the 3′-end binding pocket, the base of A17 forms a π-π stacking interaction with the side chain of W284, while the base of T18 stacks with the side chain of F245. The deoxyribose moieties of these nucleotides also engage in π-π stacking interactions with the side chains of P285 and Y257. Additionally, the phosphate group between T16 and A17 establishes polar interactions with the side chain of R261, whereas the phosphate group between A17 and T18 forms multiple hydrogen bonds with the side chains of N223, Y262, R221, and Y257. Collectively, these interactions stabilize the 3’-end of the gDNA (Supplementary Fig. 10).

Furthermore, the guide in VbAgo_gDNA bears remarkable resemblance to that observed in the PliAgo complex (Fig. 2d). Within VbAgo, four base twists occur in the gDNA, disrupting continuous π-π stacking interactions at specific base positions: A7-G8, A10-G11, T13-T14, and A17-T18. Each of these twists is accompanied by at least one set of interactions, including T297-A7, Q310-A7, R619-G11, W50-T13, R122-T14, W140-T14, W284-A17, F245-T18, and Y257-T18 (Fig. 2e). Alanine mutagenesis at these interaction sites resulted in modest reductions in activity, except for the R619A mutant, which exhibited significantly diminished cleavage activity compared to the wild type, likely due to impaired stability of the gDNA-tRNA duplex in the ternary cleavage state (Supplementary Fig. 10). Conversely, the W140A mutant showed enhanced cleavage activity, possibly attributable to improved target RNA pairing efficiency (Fig. 2f).

Target RNA binding induces the dimerization of VbAgo

By varying the reaction times with targets (7, 10, and 40 min), we successfully captured ternary structures corresponding to multiple states, including target binding and product release, with overall resolutions ranging from 2.7 to 3.0 Å (Supplementary Figs. 4–6 and Supplementary Table 2). Within this series of ternary structures, both monomeric and dimeric conformations were observed (Fig. 3a). To test whether target RNA binding drives VbAgo dimerization, we performed a quantitative titration experiment in which increasing concentrations of target RNA were incrementally added to pre-assembled VbAgo-gDNA binary complexes maintained at a fixed, sub-stoichiometric protein concentration. Native-PAGE analysis confirmed that VbAgo remained exclusively monomeric in the absence of target RNA, whereas dimerization occurred only after reaching a threshold target RNA concentration (Supplementary Fig. 12a). Concurrently, the dimer-to-monomer ratio progressively decreased over longer reaction times, consistent with the kinetic progression of the catalytic cycle as presented in the data processing workflows (Supplementary Figs. 4–6).

At an incubation time of 7 min, the corresponding dataset revealed that a subset of VbAgo-gDtR particles adopted a dimeric conformation (Supplementary Fig. 4). A certain proportion of dimers was still present at 10 min (Supplementary Fig. 5). However, due to severe preferred orientation, we were unable to resolve the three-dimensional structure of dimers. In contrast, the structures from the 40-min dataset predominantly exhibited monomeric forms (Supplementary Fig. 6). These conformations are designated as dimer-VbAgo_gDtR and monomer-VbAgo_gDtR, respectively (Fig. 3a). Combined with our observation that VbAgo_gDNA exists exclusively in the monomeric form (Supplementary Fig. 3), these structural and biochemical data collectively indicate that target RNA binding is necessary and sufficient to induce dimerization of this mesophilic RNA-targeting long pAgo, consistent with recent findings in thermophilic DNA-targeting long pAgos (PfAgo and TtdAgo). During peer review, a parallel study on a mesophilic DNA-targeting long pAgo (CpAgo from Clostridium perfringens) observed similar target DNA-induced dimerization, supporting nucleic acid-triggered oligomerization as a conserved regulatory feature across diverse Ago family members^38,39.

Structural analysis of VbAgo ternary complexes during the dimer-to-monomer transition

In the structures obtained from the 7-min and 10-min datasets, the RNA targets remained intact in both monomeric and dimeric forms. Further analysis revealed that the nucleic acid duplex in all conformations comprises 15 base pairs, ranging from G2-C2′ to T16-A16′, with the RNA target remaining uncleaved. Using the monomeric structure from the 7-min dataset (monomer-VbAgo_gDtR) as a reference, we conducted pairwise structural alignments (RMSD analysis) with the left/right protomers from the 7-min dimeric complex and the monomeric complex from 10-min dataset (Supplementary Figs. 12b, c and 13). The low RMSD values achieved confirm the structural identity of these pre-catalytic states (Supplementary Fig. 13).

Within the active site, a pair of magnesium ions (Mg²⁺ A and Mg²⁺ B) form a conserved interaction network with four catalytic residues (Supplementary Fig. 14). The side chain of D550 consistently interacts with both Mg²⁺ A and Mg²⁺ B; D756 interacts exclusively with Mg²⁺ A; and D617 interacts exclusively with Mg²⁺ B. Notably, in the conformation corresponding to the 7-min dataset, Mg²⁺ A is coordinated by two water molecules, whereas in the 10-min dataset conformation (Supplementary Fig. 14a–c), only one water molecule is observed near Mg²⁺ A (Supplementary Fig. 14d). Additionally, Mg²⁺ B is consistently surrounded by several scattered water molecules, and the side chain of the glutamate finger residue E584 is oriented outward from the pocket across all conformations (Supplementary Fig. 14a–d). This observation is consistent with the dynamic movement of the glutamate finger during cleavage.

Interestingly, within the 40-min dataset, two distinct monomeric classes were identified through classification (Supplementary Fig. 6 and Supplementary Table 2). One of these classes was assigned to a cleavage-inactive state, based on the density map showing only a single magnesium ion (Mg²⁺ A) within the catalytic active site, while the position typically occupied by the second magnesium ion (Mg²⁺ B) was instead occupied by a water molecule (Supplementary Fig. 14e). This finding deviates from the canonical two-metal-ion cleavage mechanism observed in long pAgos¹¹. Additionally, the density map of its C-terminus allowed us to model only up to residue L759, whereas in monomeric VbAgo_gDtR the model extends to V763 at the end of the helix. In contrast, in dimeric VbAgo_gDtR, a loop region extends beyond this point up to S767. Apart from this, the remaining regions of this cleavage-inactive state resemble those of monomer-VbAgo_gDtR (Supplementary Fig. 12e). We speculate that this class represents an intermediate state within the cleavage cycle and will not be further discussed here. The other class exhibited a unique state of the target RNA, distinct from all previously described states, characterized by cleavage and accompanied by partial product release—designated as “released-VbAgo_gDtR” for clarity (Fig. 3a). The target RNA strand is cleaved between positions 10′ and 11′, with the double-helical structure maintained only on the seed region side. The guide DNA strand becomes disordered beyond position 14. Moreover, the absence of the density corresponding to the target RNA segment from C11′ to T19′ results in a downward bending of the 3′-end of the guide DNA strand by a certain angle (Supplementary Fig. 15b, d). Taken together with all the aforementioned ternary structures, this suggests that target cleavage occurs after dimerization and most likely takes place following dimer dissociation. However, further investigation is required to elucidate the precise function of the observed dimerization and the specific mechanism involved in transition from dimers back to monomers during cleavage as the reaction progresses.

The C-terminal loop-helix-loop transition regulates the positioning of the catalytic residue D756, toggling it between engaged and disengaged states within the catalytic pocket

To gain insight into the molecular mechanism that drives RNA cleavage by VbAgo, we performed structural alignments comparing the catalytically active monomer-VbAgo_gDtR complex with both its precursor binary VbAgo_gDNA state and the post-cleavage released-VbAgo_gDtR state (Fig. 3b). These comparisons uncovered a remarkable structural transformation in the C-terminus (residues 750–763). In the monomer-VbAgo_gDtR, this region transitions from a disordered loop—as observed in both binary and post-cleavage states—to a well-defined α-helix (Fig. 3b). Notably, within this dynamic helical segment, D756—the final residue of the catalytic tetrad—undergoes a significant reorientation, positioning its side chain into the catalytic pocket (Fig. 3d). Meanwhile, residues 764–782 exhibit absent or weak electron density in both the VbAgo binary and ternary complexes (Fig. 2b and Supplementary Fig. 12e), indicating that this region remains flexible upon gDNA binding and target RNA engagement and likely does not participate in these processes. Deleting this region alleviates auto-inhibition and enhances conformational flexibility, thereby increasing cleavage activity. Analysis of the C-terminus (residues 736–767), which shows electron density in the VbAgo ternary complexes, reveals that E761 is located at an α-helix of the C-terminus (Supplementary Fig. 12e). Truncation at E761 disrupts a key hydrogen bond between E761 and R760, compromising the stability of the helical structure and impairing the loop-to-helix transition, thereby destabilizing the catalytic center.

Structurally, L759 within the formed helix is sandwiched between two hydrophobic residues, L552 and I557, further stabilizing the helical conformation (Fig. 3f, middle panel). To validate the importance of helix stability, cleavage assays were conducted using a double mutant (L552G/I557G) engineered to disrupt stability at the helix terminus. This mutant demonstrated a markedly reduced cleavage activity relative to the wild-type protein (Fig. 3e). Additionally, truncation at L759 abolishes a critical intrahelical hydrogen bond between L759 and D756, which is essential for maintaining the local architecture within the same helical turn. This disruption alters the spatial orientation of D756, impeding its proper involvement in catalysis and ultimately resulting in a complete loss of cleavage activity (Figs. 1c and 3f). Collectively, these structural and biochemical analyses indicate that the C-terminus can be divided into two distinct functional regions: one (residues 762–782) just responsible for auto-inhibition and another (residues 736–761) indispensable for catalysis. This finding also accounts for previous C-terminal truncation data, which align with observed cleavage activity turning points at positions 759 and 761 (Fig. 1c).

Following RNA cleavage, the dissolution of the helix coincides with D756 disengaging from the catalytic center, and the C-terminus reverting to its original disordered conformation (Fig. 3d, f). Based on these findings, we hypothesize that the C-terminus facilitates the insertion of the catalytic residue D756 into the catalytic pocket by undergoing local conformational changes from a loop to a helix. This transition promotes maturation of the catalytic center and modulates the target RNA cleavage process. At the same time, we conducted a comprehensive examination of the changes taking place in various domains and nucleic acid segments during target binding and product release. Through pairwise overlay analysis comparing adjacent states, we noted that the MID domain exhibited subtle differences (Fig. 3b and Supplementary Fig. 15c). As the target RNA binds, facilitated by base complementarity pairing, the 3′-end of the gDNA progressively dissociates from the PAZ pocket. This dissociation ultimately leads to the formation of 15 base pairs, ranging from G2-C2’ to T16-A16′, which are arranged diagonally across the N-terminal domain. However, during the cleavage of the target RNA and subsequent product release, only the seed region of the gDNA maintains base pairing. The unpaired 3′ complementary region of the gDNA shifts closer to the PAZ domain once again, with its terminal position aligning with the gDNA in the binary conformation (Supplementary Fig. 15b, d).

Interestingly, unlike other Ago cleavage processes where the PAZ domain typically undergoes a pronounced flip^25,38, VbAgo’s PAZ domain exhibits only a lateral shift of approximately 6 Å from target RNA binding to cleavage completion (Fig. 3b and Supplementary Fig. 15a). Instead, the N-terminal domain undergoes a substantial flip followed by subsequent recovery (Fig. 3b, c). Furthermore, both the PAZ and N-terminal domains progressively revert to their binary conformation following cleavage, suggesting the potential initiation of a next cleavage cycle from this state. Previous biochemical experiments have also confirmed that VbAgo functions as a multiple-turnover enzyme due to its efficient product release²⁴. In summary, this mechanism, which involves C-terminus regulation and utilizes the binary conformation as the initiation point for successive cleavage cycles, represents an another catalytic mode within the Ago protein family, thereby advancing our comprehension of the molecular mechanism underlying RNA endonuclease activity.

The dimeric form of VbAgo serves as a checkpoint against off-target effect, dissociating upon a C-terminal conformational shift from loop to helix

To investigate whether VbAgo dimerization is related to its cleavage activity, we resolved the cryo-EM structure of the catalytically inactive double mutant (D550A and D617A), VbAgo-DM, in complex with guide DNA and target RNA at a resolution of 2.5 Å, designated as dimer-VbAgo-DM_gDtR (Supplementary Fig. 7 and Supplementary Table 2). Notably, nearly all particles exhibited a dimeric form. Therefore, we aligned dimer-VbAgo-DM_gDtR with dimer-VbAgo_gDtR for comparative analysis. The overall conformations of both structures are highly similar, with their left and right halves adopting identical states (Fig. 4a, b, d). Their dimerization interfaces are also highly similar (Supplementary Fig. 16g, h, j, k). Minor differences exist in the catalytic loop at the PIWI-PIWI’ dimer interfaces, specifically, the three consecutive arginine residues on the glutamate finger adopt distinct conformations. In the wild-type, R578 and R580 face the opposing monomer and interact with C8′ of the target strand and G11 of the guide strand, respectively (Supplementary Fig. 16e), whereas in the DM mutant, R579 instead faces the opposing monomer and interacts simultaneously with U10′ of the target strand and E584 (Supplementary Fig. 16d). Mutating these three arginine residues individually or collectively to alanine slightly reduces cleavage activity (Supplementary Fig. 18d, e).

Fig. 4. Comparison of dimeric structures between the wild-type VbAgo ternary complex and the double mutant VbAgo-DM ternary complex.

a The cryo-EM map (left) and atomic model (right) of the VbAgo-DM_gDtR complex are presented. The dimer comprises protein domains colored medium purple (promoter A) and thistle (promoter B), with guide DNA depicted in blue and target RNA in red, respectively. b The atomic model of the dimer-VbAgo_gDtR complex follows the color scheme established in (Fig. 3a). c Structural superimposition of the binary complex VbAgo_gDNA and the ternary complex dimer-VbAgo-DM_gDtR, highlighting the C-terminus in the magnified inset (black box). d Structural superimposition of the ternary complexes dimer-VbAgo-DM_gDtR and dimer-VbAgo-DM_gDtR, highlighting the C-terminus in the magnified inset (black box). e Schematic diagram illustrates the spatial proximity between Cy3 and BHQ2 fluorophores during the dimerization process (left). Kinetic analysis of dimer formation for wild-type VbAgo (cyan), de[759-782] mutant (purple), and DM mutant (pink) at 37 °C was performed using FRET (n = 3). Data are presented as mean ± SD, with VbAgo-gDNA (sky blue) serving as a negative control (right). Kinetic traces were fitted to a biphasic equation (see “Methods”). The decay phase rate constants (K₁) were: wild-type, 0.01030 ± 0.001019 s⁻¹; de[759-782], 0.00174 ± 0.000344 s⁻¹; and DM, 0.00243 ± 0.000147 s⁻¹. The corresponding rise phase constants (K₂) were 0.0004985 ± 4.842 × 10⁻⁵ s⁻¹, 6.760 × 10⁻⁵ ± 4.055 × 10⁻⁵ s⁻¹, and 1.007 × 10⁻⁴ ± 1.564 × 10⁻⁵ s⁻¹, respectively.

To further elucidate the functional implications of VbAgo dimerization, we introduced mutations at the amino acid residues within the N-PAZ′ and MID-PIWI′ interaction interfaces, designated as Group 1 (D35G/R276G/F274G/P33A/R90G/243G) and Group 2 (R490A/L486A/H482A/T597A/D593A/V589A). Native gel electrophoresis confirmed that these mutants exist only as monomers, and in vitro RNA cleavage assays revealed that monomeric VbAgo retains catalytic activity comparable to the wild-type but produces a shifted pattern of RNA products (Supplementary Fig. 17a, e). These results suggest that dimeric VbAgo likely represents a pre-cleavage surveillance conformation that ensures base-pairing fidelity prior to entering the catalytically active state. To identify residues responsible for checking guide-target duplex stability, we compared dimeric and monomeric VbAgo, revealing additional molecular interactions with the duplex in the dimeric conformation. Specifically, Q114 forms a hydrogen bond with the 2′-hydroxyl oxygen of the A14 base in the target RNA, while R278 forms a hydrogen bond with the phosphate backbone oxygen at the A10 base of the gDNA (Supplementary Fig. 17c). RNA cleavage assays showed that alanine substitutions disrupting these interactions led to cleavage of the RNA target in different registers (Supplementary Fig. 17d), supporting our hypothesis that dimeric VbAgo functions as a checkpoint intermediate within the catalytic cycle to prevent off-target cleavage event.

Actually, the primary difference between wild-type and double mutant dimeric ternary complexes lies in the conformation of the C-terminus: in dimer-VbAgo-DM_gDtR, the C-terminus of VbAgo appears as disordered loops, similar to those observed in the binary and released-VbAgo_gDtR states, whereas in dimer-VbAgo_gDtR, it locally forms helical conformation, consistent with that observed in monomer-VbAgo_gDtR. This also induces subtle multi-directional displacements (0.9 Å−2.1 Å) of the wild-type MID and PIWI domains relative to the DM variant (Fig. 4c, d), increasing the distance at the MID-PIWI interaction interface. Consequently, the reduced interdomain binding affinity facilitates dissociation into monomeric states. Our fluorescence resonance energy transfer (FRET) experiments also revealed that the DM mutant of VbAgo is unable to dissociate after dimerization. The mutations D550A and D617A within the catalytic pocket disrupt the stabilizing interactions with D756, preventing spontaneous formation of the associated helix, which in turn leads to the inability of the dimer to dissociate. These observations indicate that the formation of the C-terminal helix not only facilitates the maturation of the catalytic pocket but also promotes spontaneous dissociation of the dimer. Furthermore, the FRET results of the inactive C-terminal truncation mutant VbAgo-de[759-782] behaved similarly to those of VbAgo-DM (Fig. 4e), suggesting that truncation at residue 759 prevents helix formation, thereby inhibiting dimer dissociation. These findings highlight the central role of dimerization and the C-terminal conformational transition during the VbAgo catalytic cycle, providing a unified explanation for substrate recognition and dimer dissociation.

Working model of VbAgo

Based on structural and biochemical findings, we propose a working model for VbAgo (Fig. 5). In its apo state, VbAgo assumes an auto-inhibited conformation, where the acidic C-terminus blocks the nucleic acid binding channel. Then, the C-terminus flips upward to facilitate the binding of guide DNA. Once VbAgo recognizes and pairs with the target RNA, it dimerizes, and the C-terminus transitions from a loop to a helix configuration. This conformational change allows D756 to enter the active site and execute its catalytic role. Our findings from cryo-electron microscopy and FRET experiments reveal that the conformational change in the C-terminus triggers the maturation of the active site, leading to dimer dissociation. Following this, VbAgo reverts to its monomeric form and cleaves the target RNA. Upon cleavage and product release, VbAgo returns to its binary state to initiate the next cleavage cycle. Notably, throughout the entire cleavage process, VbAgo undergoes a “monomer-dimer-monomer” transformation, accompanied by a “loop-helix-loop” conformational change in its C-terminus.

Fig. 5. Working model of VbAgo.

The upper panel illustrates the sequential stages of VbAgo auto-inhibition, activation, target cleavage, and product release. Color coding follows Fig. 1a: guide DNA is shown in blue, target RNA in dark red, and the four catalytic residues are represented as sticks, with D756 specifically highlighted in yellow. The lower panel presents enlarged views of the catalytic center, illustrating structural transitions in the C-terminus across different functional states. The region spanning residues 750-762 undergoes conformational rearrangement and is indicated in black.

Discussion

The efficient cleavage capability of pAgos, unrestricted by protospacer flanking sequence (PFS) limitations, holds promise as an effective RNA editing tool^24,40. In this study, we resolved the cryo-EM structures of VbAgo in multiple states. We discovered that, unlike other long Agos, VbAgo possesses an additional sequence at its C-terminus within the PIWI domain that binds within the nucleic acid channel in its apo state, exerting an auto-inhibitory role. Subsequently, upon the successive binding of guide and target RNA, this segment undergoes a series of conformational changes, participating in the maturation of the catalytic center and dimer dissociation. Moreover, for the first time, we report target RNA binding induced dimerization in a mesophilic pAgo. This dimerization is essential for VbAgo to prevent off-target RNA cleavage. Compared to the tightly bound dimers found in thermophilic PfAgo, the interactions within VbAgo dimers are more transient (Supplementary Fig. 16), spontaneously dissociating after the C-terminus transitions from a loop to a helix. These phenomena suggest that VbAgo employs a catalytic cycle distinct from currently known pAgos, significantly expanding our understanding of the diversity of pAgos’ catalytic molecular mechanisms.

Indeed, the auto-inhibition mechanism exhibited by the acidic C-terminus is conserved in SPARTA systems. This design ensures tight regulation over SPARTA activation, thereby preventing mis-activation by partially complementary nucleic acids or excessive immune response triggered by low-copy plasmid invasion^26–29. However, our findings suggest that the C-terminus of VbAgo exhibits robust dual functionality, comprising two distinct functional regions: one (residues 736–761) essential for catalysis, and another (residues 764–782) responsible solely for auto-inhibition. In the apo state, this whole region (residues 736–782)—which includes the catalytic tetrad residue D756—adopts an auto-inhibitory conformation that sterically blocks nonspecific nucleic acid interactions. During the transition to the substrate-cleavage state, the segment harboring D756 undergoes a conformational change from a loop to a helix, stabilizing the active center. Activity is impaired when the stability of this helix is compromised. This architecture is entirely distinct from that of SPARTA systems, whose C-terminal regions utilize an ordered α-helix functioning solely as auto-inhibitory elements^26–29.

Recent structural studies of thermophilic Ago proteins have reported the dimerization of long Ago, induced by target DNA in PfAgo and TtdAgo³⁸. Upon comparing the dimerized VbAgo ternary complex with its PfAgo counterpart, we observed distinct differences in their dimerization interface. Specifically, the dimerization interface of VbAgo is relatively smaller and involves fewer interactions than that of PfAgo (Supplementary Fig. 16). This observation offers a possible explanation for why conformational changes in the C-terminus of VbAgo might facilitate dimer dissociation. Furthermore, our study identified that VbAgo requires dimerization as a critical checkpoint within its catalytic cycle to ensure the stability of the duplex, thereby preventing cleavage of the RNA target at alternative sites. By resolving cryo-electron microscopy structures corresponding to distinct states of the VbAgo catalytic process, we elucidated the molecular mechanisms underlying both dimer formation and dissociation, as well as clarified the functional significance of dimerization. Compared to previous structural and mechanistic studies of dimerized pAgos, our findings not only enhance the understanding of dimerization modes but also expand the mechanistic knowledge of pAgos, providing valuable guidance for future optimization in applied contexts.

Both single-stranded DNA and RNA can function as guides for recognizing DNA or RNA targets, resulting in four possible types of programmable enzymes⁴¹. Unlike previously characterized prokaryotic Agos such as PfAgo, TtAgo, and KmAgo, which preferentially target DNA and have undergone detailed mechanistic studies^10,25,33,38, the molecular mechanism of the gDtR class, which was characterized last, has remained poorly understood²². However, based on the requirements of an ideal RNA-programmable enzyme, including (i) effective RNA cleavage without off-target effects, (ii) convenient and inexpensive RNP assembly, (iii) no target site sequence limitations, and (iv) easy programmability to target both structured and unstructured RNA sequences without modifying the catalytic machinery⁴², the gDtR class of pAgo is undoubtedly a promising option⁴³. To our knowledge, currently characterized Agos capable of cleaving both structured and unstructured RNA sequences include MbpAgo from Mucilaginibacter paludis, KpAgo from Kluyveromyces polysporus, and VbAgo^21,37. Among them, the first two show some limitations, including the residual DNA-cleaving ability of MbpAgo and the complex purification process and ineffective removal of endogenous nucleic acids for the eAgo, KpAgo^21,24,35. Given that VbAgo emerged as a more promising candidate.

It is worth noting that we established a robust, orthogonal dual-luciferase reporter system in HEK293T cells—using Renilla luciferase as an internal control and firefly luciferase as the target—alongside an enhanced green fluorescent protein (eGFP) reporter system to quantitatively assess the knockdown efficiency of VbAgo on endogenous-like transcripts in mammalian cells. Pre-assembled VbAgo–gDNA complexes targeting either the firefly luciferase (Fluc) or eGFP coding sequences were delivered via a cell-penetrating peptide, resulting in dose-dependent and statistically significant reductions in both reporter outputs. VbAgo-WT exhibited potent knockdown activity across both reporters, whereas the catalytically inactive mutant VbAgo-DM showed no detectable activity in vivo (Supplementary Fig. 19), confirming that cleavage is strictly dependent on nuclease function. These findings provide the direct functional evidence that pAgo family proteins achieve targeted RNA degradation through their intrinsic DNA-guided RNA cleavage activity, thereby establishing VbAgo as a viable platform for programmable endogenous RNA editing in mammalian cells. Although the C-terminal truncation mutant de[764–782] failed to exhibit knockdown activity, the live-cell, high-throughput screening platform established in this study provides a foundation for further structure-guided engineering to improve VbAgo’s in vivo stability, target-binding fidelity, and catalytic efficiency. This rational engineering framework positions VbAgo as a promising candidate for next-generation in vivo RNA editing therapeutics.

Methods

Protein expression and purification

The wild-type VbAgo gene and its catalytically inactive double mutant variant (VbAgo-DM, D550A, D617A) were cloned into pET23a expression vectors containing a C-terminal 6×His tag and are stored in our laboratory. Site-directed point mutants for in vitro activity assays were generated using PCR-mediated mutagenesis⁴⁴. The resulting recombinant plasmids were transformed into Escherichia coli BL21(DE3) competent cells (Novagen). Bacterial cultures were grown in LB medium containing 100 μg/mL ampicillin at 37 °C with shaking until reaching an optical density at 600 nm (OD₆₀₀) of 0.8–1.0. Protein expression was induced by adding 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), followed by incubation at 18 °C for 16 h. Cells were harvested by centrifugation at 6000 rpm for 10 min at 4 °C, and the supernatant was discarded. The collected cell pellets were immediately flash-frozen in liquid nitrogen and stored at −80 °C for subsequent purification.

Initially, cell pellets were resuspended in Buffer A (50 mM Tris-HCl, pH 7.5, 500 mM NaCl, 5% glycerol, 10 mM imidazole) and lysed using a JN-02C cell disrupter (JNBIO, Inc.). The lysate was clarified by centrifugation at 14,000 rpm for 30 min at 4 °C. Subsequently, the supernatant was incubated with Ni-NTA resin (Qiagen) at 4 °C for 1 h with gentle rotation. The resin was extensively washed with Buffer A, and the target protein was eluted stepwise using Buffer A supplemented with 50 mM and 100 mM imidazole. Eluted fractions were analyzed by SDS-PAGE, and those containing the target protein were desalted using a G-25 desalting column (Cytiva) and buffer-exchanged into Buffer B (50 mM HEPES, pH 7.5, 100 mM NaCl, 5% glycerol). Further purification was performed on a Hi-Trap Heparin HP column (Cytiva), equilibrated with Buffer B, and eluted using a linear NaCl gradient from 0.1 to 1.0 M. Peak fractions were concentrated using an Amicon 50 K filter unit (Millipore) and subjected to size-exclusion chromatography on a Superdex 200 Increase 10/300 column (Cytiva) pre-equilibrated with Buffer C (20 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM DTT). Finally, the purified protein was concentrated to 20 mg/mL and stored at −80 °C.

Target RNA cleavage assay

Cleavage assays were performed using synthetic guide DNAs and 5′-FAM-labeled target RNAs. The sequences of the oligonucleotides are listed in Supplementary Table 1. Reactions were carried out in a 10 μL mixture containing 800 nM VbAgo protein or its mutants, 400 nM 18-nt guide DNA, and reaction buffer (20 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM MgCl₂). The mixture was incubated at 37 °C for 10 min, after which 200 nM 5′-FAM-labeled target RNA was added. The reaction proceeded for an additional 30 min and was subsequently terminated by adding 2×RNA loading buffer (95% formamide, 18 mM EDTA, 0.025% SDS, 0.025% bromophenol blue). Samples were denatured at 95 °C for 5 min and resolved on a 20% denaturing polyacrylamide gel. Gel images were acquired using the GelDoc Go Gel Imaging System (Bio-Rad), and data analysis was performed with Image J and Prism 8.0 (GraphPad).

RNA cleavage kinetics were analyzed with slight modifications to the procedures described above. For single-turnover reactions, the pAgo protein and guide DNA were combined at a 1:1 molar ratio to form the pAgo-gDNA binary complex, which was then mixed with target RNA at a 2:1 ratio to initiate catalysis (final concentrations: 400 nM pAgo, 400 nM gDNA, 200 nM target RNA). For multiple-turnover reactions, the pAgo protein and guide DNA were mixed at a 1:2 molar ratio to form the binary complex. Target RNA was then added in a four-fold molar excess relative to the gDNA to start the reaction (final concentrations: 80 nM pAgo, 160 nM gDNA, 320 nM target RNA). Single-turnover cleavage kinetics data were fitted to the equation: Y = Y_max × [1 – exp(–k_obs × X)], where X is time, Y_max is the maximum cleavage, Y is the cleavage efficiency at a given time point, and k_obs is the observed rate constant. Multiple-turnover kinetics data were fitted to the linear equation Y = B + K_obs × t, where t is time, Y is the cleavage efficiency at time t, and B is the cleavage efficiency in the absence of pAgo.

Thermal stability assay

Thermal stability was assessed by pre-incubating 800 nM VbAgo protein or its mutants at various temperatures for 30 min. Subsequently, samples were mixed with 400 nM phosphorylated 18-nt guide DNA and incubated at 37 °C for 10 min to promote binary complex formation. Then, 200 nM target RNA was added, and the reaction proceeded at 37 °C for 1 h. The reaction was terminated by adding 2× RNA loading buffer, and the products were resolved on a 20% denaturing polyacrylamide gel. Gel visualization and data analysis were performed as previously described for the cleavage assay.

Native polyacrylamide gel electrophoresis (Native-PAGE)

The reaction volume was 10 µL. Purified VbAgo, diluted to 0.2 mg/mL, was combined with an equimolar amount of guide DNA in Buffer N (20 mM HEPES-NaOH, pH 7.9, 50 mM NaCl, 5 mM MgCl₂) and incubated at 50 °C for 10 min to assemble the VbAgo-gDNA binary complex. To form the VbAgo-gDtR ternary complex, increasing concentrations of target RNA (1, 2, 3, and 5 µM) were added to the 0.2 mg/mL binary complex, and the samples were incubated at 37 °C for 7 min. The samples were then separated on a 4% native PAGE gel, and protein bands were visualized by Coomassie Brilliant Blue staining. The mutant protein was analyzed under identical conditions, except that the ternary complex was formed by adding target RNA at a 1.5-fold molar excess relative to the guide DNA.

Fluorescence resonance energy transfer (FRET) assay

The 5′-phosphate (5′-P)-labeled and 3′-Cy3/BHQ2-labeled 18-nt guide DNAs (gDNAs) were synthesized by Sangon. The 3′-Cy3-labeled and 3′-BHQ2-labeled gDNAs were separately pre-incubated with VbAgo at 50 °C for 15 min to form Cy3-labeled gDNA/VbAgo (C1) and BHQ2-labeled gDNA/VbAgo (C2) complexes. Subsequently, C1, C2, and 19-nt target RNA were rapidly mixed and transferred into a 200 μL fluorescence cuvette. VbAgo, guide DNAs, and target RNAs were mixed at a molar ratio of 1:1.1:1.1 (100 nM VbAgo, 110 nM guide, 110 nM target) in reaction buffer (20 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM MgCl₂). Fluorescence signals of Cy3 (donor) and BHQ2 (quencher) were recorded at 1 s intervals using a fluorescence spectrophotometer (F4600, HITACHI) for 1 h at 37 °C. The fluorescence signal from the C1 component was monitored simultaneously. The excitation and emission wavelengths for Cy3 were 550 nm and 570 nm, respectively. The Cy3 fluorescence intensity was plotted as a function of time. The relative fluorescence intensity was calculated using the equation F = F_x/C_x × 1, where F_x is the fluorescence intensity at a given time and C_x is the fluorescence intensity for the C1-only component at that same time. Data were analyzed using GraphPad Prism 8.0. The resulting data were fitted to a fall-and-rise equation: F = [−C/(K₁ − K₂)] × [exp(−K₂ × t) − exp(-K₁ × t)] + F_max, where t is time, F is the relative fluorescence intensity at a given time, C is the initial rate, and F_max is the maximum relative fluorescence intensity. Here, K₁ represents the observed rate constant for the falling phase (dimer formation), while K₂ is the observed rate constant for the rising phase (dimer dissociation).

Mammalian cell culture and transfection for knockdown analysis with VbAgo

All experiments in mammalian cells were conducted using the HEK293T cell line, which stably expressed either a dual-luciferase or an enhanced green fluorescent protein (eGFP) reporter system. The cells were cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. Transfection was performed via cell-penetrating peptide-mediated delivery. For each well, 200 nM of pre-incubated VbAgo-gDNA binary complexes (VbAgo:gDNA = 1:1) and 10 µM of LAH5 cell-penetrating peptide (Nanjing Jie Tai Biotechnology Co., Ltd) were transfected (gDNA sequences are listed in Table S1). Prior to transfection, cells were seeded in a 48-well plate at 250 µL per well. When cell confluence reached 60%–70%, the medium was replaced with Opti-MEM, and the cells were maintained in this medium for over 1 h. Cells were harvested 24 h post-transfection and washed once with phosphate-buffered saline (PBS). For the dual-luciferase reporter system, cells were lysed on ice, and luciferase activity was measured using a Dual-Luciferase Reporter Assay Kit (Yeasen, Cat# 11402ES60). For the eGFP reporter system, fluorescence intensity was analyzed by flow cytometry using a CytoFLEX instrument (Beckman Coulter, Inc.), with 10,000 cells examined per sample. All experiments were performed with three biological replicates.

Cryo-EM sample preparation and data acquisition

For the binary complex, 5′-phosphorylated guide DNA was added to purified VbAgo protein at a 1:1 molar ratio and incubated at 37 °C for 30 min. A 4 µL aliquot of the mixture (1.0 mg/mL) was applied to glow-discharged 300 mesh holey gold grids (Au, R1.2/1.3, Quantifoil). The grids were blotted for 4–5 seconds under 100% humidity at 4 °C and vitrified in liquid ethane using an FEI Vitrobot Mark IV (Thermo Fisher Scientific). For ternary complex, the protein was initially incubated with the guide at 37 °C for 10 min, after which the target was added, and the reaction was allowed to proceed at the same temperature for varying durations (7, 10, 40 min). The molar ratio of protein, guide, and target in the reactions was maintained at 1:1:1. All subsequent procedures for ternary complex grids preparation were identical to those for the binary complex. The final magnesium ion concentration was maintained at 5 mM in all reactions.

Cryo-EM grids were loaded onto a 300 kV Titan Krios microscope (Thermo Fisher Scientific) equipped with a Gatan K3 direct electron detector and a BioQuantum energy filter with a slit width of 20 eV. Micrographs were automatically collected using EPU software following standard protocols. Data were recorded in counting mode at a magnification of ×105,000, corresponding to a pixel size of 0.851 Å/pixel. The defocus range was set between −1.0 and −1.5 μm. Each micrograph consisted of 40 frames captured over a total exposure time of 2.5 s, with an accumulated electron dose of approximately 54 e⁻/Ų.

Image processing

For all datasets, image stacks were summed and corrected for drift and beam-induced motion using MotionCor2⁴⁵. Subsequent data processing was performed using cryoSPARCv4.2.1⁴⁶. Using the VbAgo_apo dataset as a representative workflow, we processed a total of 4009 micrographs. The initial steps included motion correction and contrast transfer function (CTF) estimation. Automated picking with blob picker on 2000 micrographs yielded 3,821,806 particles. These particles underwent two rounds of ab-initio reconstruction, heterogeneous refinement, and 2D classification. From the most distinct 2D classes, we selected 180,989 particles for two additional rounds of reconstruction and refinement. The refined particles were split into five classes for 3D classification. The best class was subjected to non-uniform refinement, resulting in a 3.8 Å density map, which was then used to create 2D templates.

Utilizing these templates, the template picker identified 11,019,420 particles from the entire dataset. After the initial reconstruction and refinement, 3,482,833 particles were retained. To further refine these particles, five iterative cycles of ab -initio reconstruction and heterogeneous refinement were conducted to remove bad particles and contaminants, resulting in 460,522 particles available for 3D classification. The best class contained 84,950 high-quality particles, achieving a final electron density map at a resolution of 3.2 Å.

Parallel processing pipelines generated final datasets for VbAgo_gDNA, dimer-VbAgo_gDtR, monomer-VbAgo_gDtR, (10-min) monomer-VbAgo_gDtR, (40-min) monomer-VbAgo_gDtR, released-VbAgo_gDtR, and dimer-VbAgo-DM_gDtR, comprising 186,978, 125,387, 143,760, 132,500, 114,582, 112,179, and 1,016,135 particles, respectively. These datasets yielded maps with resolutions of 3.0, 2.7, 2.8, 2.9, 2.9, 3.0, and 2.5 Å, respectively. Comprehensive image processing parameters are cataloged in Table S2, with complete workflows presented in Supplementary Figs. 2–7.

Model building and refinement

The initial template for VbAgo_apo was generated using AlphaFold2, followed by rigid-body fitting into the cryo-EM density map using Chimera^47,48. Subsequent manual adjustments and model rebuilding were performed in COOT⁴⁹. The model was further refined through multiple rounds of real-space refinement in PHENIX⁵⁰. For the remaining seven structures, guide DNA and target RNA were built de novo in COOT. The well-refined VbAgo_apo model served as the starting template for these structures, with iterative manual adjustments in COOT and real-space refinements in PHENIX. Final model statistics were validated using MolProbity⁵¹ and are summarized in Table S2. Structural figures were prepared using UCSF Chimera.

Statistics and reproducibility

Pearson’s correlation coefficients and p-values were calculated using GraphPad Prism 8.0 software. P-values for multi-group comparisons (≥3 groups) were determined by one-way ANOVA followed by Dunnett’s multiple comparisons test, while pairwise analyses between two groups employed two-tailed independent Student’s t-tests, using wild-type proteins or strains as controls. All statistical experiments were repeated with at least three independent preparations. Error bars represent the standard error of the mean (n = 3; *p < 0.05, **p < 0.01, ***p < 0.001). Source data are provided in the Source Data file. Each statistical analysis was performed on data obtained from a minimum of three independent biological replicates. For representative electron microscopy images, more than five randomly selected micrographs were acquired per condition from each independent replicate.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

S.W., L.M., and Yang L. supervised the project and designed the experiments. L.M. and Yang L. provided the original VbAgo plasmid. T.C., K.M., and K.C. performed the expression and purification of VbAgo wild-type and mutants proteins. T.C., Q.D., and S.L. carried out biochemical assays. T.C. also prepared cryo-EM grids. J.W. and H.Y. collected the cryo-EM data. S.W., J.W., and Yu L. conducted the data processing. T.C. and X.T. built the atomic models. S.W. and X.T. analyzed the data and wrote the manuscript. All authors participated in the discussion and revised the manuscript.

Peer review

Peer review information

Nature Communications thanks Tian-Min Fu and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The cryo-EM maps have been deposited into the Electron Microscopy Data Bank with the accession numbers: EMD-64797 for VbAgoapo, EMD-64798 for VbAgogDNA, EMD-64832 for dimer-VbAgogDtR, EMD-64859 [https://www.ebi.ac.uk/pdbe/entry/emdb/] for monomer-VbAgogDtR, EMD-64871 for (10 min) monomer-VbAgogDtR, EMD-64873 for (40 min) monomer-VbAgogDtR, EMD-64826 for released-VbAgogDtR, EMD-64822 for dimer-VbAgo-DMgDtR). The coordinates have been deposited into the Protein Data Bank (accession numbers: 9V65 for VbAgoapo, 9V66 for VbAgogDNA, 9V8A for dimer-VbAgogDtR, 9V93 for monomer-VbAgogDtR, 9V9G for (10 min) monomer-VbAgogDtR, 9V9I for (40 min) monomer-VbAgogDtR, 9V7Z for released-VbAgogDtR, 9V7S for dimer-VbAgo-DMgDtR). Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-71422-y.