Structural basis for the recognition of guide RNA and target DNA heteroduplex by Argonaute

Argonaute proteins are key players in the gene silencing mechanisms mediated by small nucleic acids in all domains of life from bacteria to eukaryotes. However, little is known about the Argonaute protein that recognizes guide RNA/target DNA. Here, we determine the 2 Å crystal structure of Rhodobacter sphaeroides Argonaute (RsAgo) in a complex with 18-nucleotide guide RNA and its complementary target DNA. The heteroduplex maintains Watson–Crick base-pairing even in the 3′-region of the guide RNA between the N-terminal and PIWI domains, suggesting a recognition mode by RsAgo for stable interaction with the target strand. In addition, the MID/PIWI interface of RsAgo has a system that specifically recognizes the 5′ base-U of the guide RNA, and the duplex-recognition loop of the PAZ domain is important for the DNA silencing activity. Furthermore, we show that Argonaute discriminates the nucleic acid type (RNA/DNA) by recognition of the duplex structure of the seed region.

are in the order Rhodobacter sphaeroides Ago (RsAgo), Thermus thermophilus Ago (TtAgo), Pyrococcus furiosus Ago (PfAgo), Aquifex aeolicus Ago (AaAgo), human Argonaute2 (HsAgo2), human Argonaute1 (HsAgo1), Kluyveromyces polysporus (KpAgo), Neurospora crassa QDE-2 Argonaute (NcQDE-2), Drosophila melanogaster Argonaute1 (DmAgo1) and Arabidopsis thaliana Argonaute1 (AtAgo1). Residue numbers and the secondary structure of RsAgo are indicated above the alignment. Conserved residues are shown in red, and essentially invariant residues are shaded in red. Conserved binding residues in the MID domain and slicer catalytic residues are shown by green and blue circles, respectively. The amino acid residue highly conserved in eukaryotic Argonautes corresponding to K566 in HsAgo2 is shown by the orange circle. The amino acid residue that is highly conserved in RNA-guided Argonautes corresponding to R731 in RsAgo is shown by the red circle.
The RNA and DNA strands are shown in upper case red letters and lower case blue letters, respectively. The phosphorylated 5ʹ end of each guide strand is denoted by the letter "p".

Widespread recognition of the heteroduplex by the PIWI domain and the mechanism of defective slicer activity of RsAgo
The PIWI domain directly interacted with the heteroduplex through four main sites ( Supplementary Fig. 6a).  Fig. 3e). In addition, the side chain of K609 and the main chain of R610 interacted with the phosphate groups of nucleotides 15 and 16 of the guide RNA strand, respectively ( Supplementary Fig. 3e). In TtAgo, loop L3 also interacted with both the guide DNA and the target DNA/RNA strands in several conformational states 8,9,16 . At the third site, the side chains of N686 and K692 formed hydrogen bonds with nucleotides 6 (phosphate group) and 4 (2′-OH and O2) of the guide RNA strand, respectively ( Supplementary Fig. 3b). In addition, the side chain of R731 interacted with the phosphate groups of nucleotides 3 and 4 of the guide RNA strand (Fig. 3a). Similar interactions via an arginine residue equivalent to R731 have also been observed in RNA-guided eukaryotic Argonaute proteins 10-14 . This residue is not conserved in DNA-guided prokaryotic Argonaute proteins ( Supplementary Fig. 4) suggesting that the interaction via the arginine residue at the third site is a characteristic feature of RNA-guided Argonaute proteins. At the fourth site, the side chain of S638 and the main chains of H639 and D640 formed hydrogen bonds with the phosphate group of nucleotide 11′ of the target DNA strand (Supplementary Fig. 3j). In the vicinity of the fourth site, the three catalytic residues DDD/H (D478, D546,and D660 in TtAgo) were replaced by GHE (G529, H605, and E746) in RsAgo ( Supplementary Fig. 6b,c).
Additionally, access to the active site by the E569-containing loop (corresponding to loop L2 of TtAgo that contained the fourth catalytic residue E512), a conformational change to form a catalytic tetrad 12,16 was completely blocked by loop L1 (Supplementary Fig. 6b,c). These are likely the main causes of slicer-activity deficiency in RsAgo.

Interactions and additional functions of the L1 and L2 linkers
Our structural data showed that the linker regions directly interacted with the guide and the target strands ( Fig. 2c). At the L1 linker region, a long β-hairpin of the L1 linker interacted with the heteroduplex at two sites ( Supplementary Fig. 7a). At the basal site that is close to the PAZ domain, the R151 side chain and the M176 and Y178 main chains formed hydrogen bonds with the phosphate groups of nucleotides 8 and 9 of the guide RNA strand (Supplementary Fig.   3c). At the front-end site, the K157 side chain formed a hydrogen bond with the phosphate group of nucleotide 13′ of the target DNA strand (Supplementary Fig. 3j). To clarify the functional roles of the interaction between the L1 linker and the heteroduplex, we performed a plasmid DNA silencing assay using R151A and K157A variants. The analysis showed that both the mutations significantly reduced DNA silencing activity compared to wild-type ( Supplementary Fig. 7c). These results indicated that the interactions of R151 and K157 with the target strand were crucial for DNA silencing activity. On the other hand, the front-end site of the equivalent long β-hairpin is separate from the target strand in TtAgo, unlike in RsAgo ( Supplementary Fig. 7a,b). This may be due to the presence of a negatively charged amino acid residue (D154) in TtAgo at the equivalent position of K157. The interaction between K157 and the target DNA strand might be a specific feature of RNA-guided DNA-targeting Argonaute.
The L2 linker interacted with the duplex at two sites ( Fig. 2c). At the first site, the helix α8-turn-helix α9 segment interacted with the seed region of the duplex (Supplementary Fig. 3h).
On helix α8 of this segment, the K245 side chain formed hydrogen bonds with nucleotides 7 (2′-OH and O2) and 8 (O4′) of the guide RNA strand, and the T249 side chain formed a hydrogen bond with nucleotide 6′ (O3′) of the guide DNA strand ( Supplementary Fig. 3d,h). On the other hand, on helix α9 of this segment, the Y260 side chain formed a hydrogen bond with nucleotide 5′ (phosphate group) of the target DNA strand, and the R275 side chain interacted with nucleotides 7 (O3′) and 8 (phosphate group) of the guide RNA strand ( Supplementary Fig. 3c,h).
To evaluate the importance of these hydrogen bonds, we used a plasmid DNA silencing assay and found that substitution of K245 or R275 with alanine showed the reduction of the activity ( Supplementary Fig. 7c). In HsAgo2, it has been suggested that the fitting of helix α7, which corresponds to the helix α8-turn-helix α9 segment of RsAgo, into the minor groove of the seed region is crucial for the target binding (Fig. 7a,b) 15 . Based on this view, we infer that the interaction of the helix α8-turn-helix α9 segment with the heteroduplex is important for the functional expression of RsAgo. At the second site, the adenine of nucleotide 1′ of the target DNA strand penetrated a narrow pocket formed by the L2 linker (Y329, E340 and Y341) and the PIWI domain (R670, Q689 and L690), where the Y329 side chain of the L2 linker formed a hydrogen bond to the adenine N6 amine ( Supplementary Fig. 3f,g). In a plasmid DNA silencing assay, we found that mutation of Y329 to alanine reduced the activity of RsAgo (2.4-fold), suggesting the importance of this residue ( Supplementary Fig. 7c).