Of seeds and supplements: structural insights into extended microRNA–target pairing

Abstract

Metazoan micro(mi) RNAs guide Argonaute proteins to their targets via perfect pairing to the seed region, located near the 5′ end of the miRNA. In this issue of The EMBO Journal, Sheu‐Gruttadauria et al report the crystal structure of human Argonaute 2 in complex with both a miRNA and target RNA and show that miRNA 3′ supplementary nucleotides can increase target affinity and may contribute more to miRNA‐mediated silencing than is currently appreciated.

MicroRNAs (miRNAs) are a class of short (~21 nt) RNAs that post‐transcriptionally repress gene expression by hybridizing to targeted mRNAs and repressing protein synthesis. However, instead of hybridizing to cognate mRNAs as naked molecules, miRNAs are loaded into Argonaute (Ago) proteins, which serve as the core of the miRNA‐induced silencing complex (miRISC). Ago proteins are comprised of four domains (amino [N]/PAZ and MID/PIWI) in a bilobal organization (Fig 1A) (Song et al, 2004). Connected by linker regions (L1 and L2), the two lobes form a central cleft and positions both the guide (miRNA) and the target RNA (Wang et al, 2008). When loaded into Ago proteins, the miRNA sequence can be broken down into subdomains: the 5′ anchor (g1), the seed (g2‐g8), the center (g9‐g12), the 3′ supplementary (g13‐g16), and the tail sequence (the 3′‐most nucleotides) (Fig 1B) (Wee et al, 2012; Schirle et al, 2014; Salomon et al, 2016). The seed region is the evolutionarily most conserved portion of a miRNA, and seed‐target RNA base pairing is essential for dictating target specificity (Bartel, 2009). Supplementary base pairing between a miRNA and a target has also been predicted to enhance target recognition (Grimson et al, 2007). However, in contrast to complementary seed regions in targets, supplementary target nucleotides are generally poorly conserved (Friedman et al, 2009). Thus, while the significance of seed‐target base pairing to miRNA‐mediated repression is well established, the relevance of supplementary pairing between miRNAs and targets has remained a matter of debate.

Figure 1. microRNA–target site interactions in human Ago2.

(A) Schematic diagram of human Ago2 protein. Domains and loops are marked accordingly. (B) Schematic of miRNA–target interactions in the Ago2 crystal structure (C). Vertical lines denote base pairing between a miRNA (red) and a target sequence (blue). Central miRNA nucleotides are compacted, bringing the seed and supplementary duplexes into proximity in Ago2. Figures were adapted from Sheu‐Gruttadauria et al (2019).

In this issue of The EMBO Journal, Sheu‐Gruttadauria et al (2019) describe the crystal structure of a human Argonaute 2 (Ago2)–miR122–target ternary complex structure (Fig 1C). Importantly, while other structural studies of human Argonaute have used targets that only base pair to a miRNA seed region, this is the first structure containing both seed and supplementary base pairing between a miRNA and a target. Initially, Sheu‐Gruttadauria and colleagues had set out to determine the structural basis for target RNA cleavage by Ago2. To this end, they utilized a catalytically inactive mutant Ago2 to avoid cleavage and a target that was fully complementary from g2‐g16. Unexpectedly, despite the availability of central base pairing, the complex crystallized in a conformation lacking it. Instead, the miRNA forms two discontinuous duplexes with the target RNA: one with the seed region (g2‐g8) and the other with the supplementary region (g13‐g16) in a separate chamber in the central cleft that can accommodate up to 5 contiguous base pairs. While theoretically the central base pairs should be able to form, the narrowing of the central cleft and the position of several loops in Ago2 prevent this from happening. Notwithstanding the discontinued nature of the two duplexes, the crystal structure also indicates that the seed and supplementary duplexes are in close proximity to each other as a result of compaction of central miRNA nucleotides g9‐g12. The authors also compare the crystal structures of seed‐only versus seed plus supplementary‐paired Ago2, which reveals that Ago2 undergoes a number of conformational shifts upon seed pairing allowing for supplementary pairing to proceed.

Sheu‐Gruttadauria and colleagues provide evidence that targets capable of supplementary pairing can have over 20‐fold higher affinities for an Ago2‐loaded miRNA as compared to targets that can only pair to a miRNA seed region, with GC‐rich supplementary base pairing outperforming weaker AU‐rich pairing. The authors also tested target affinity by varying the length of the nucleotides that bridge the seed and supplementary pairing nucleotides. They observed that while a 4 nucleotide bridge in the target provides optimal affinity, the seed and supplementary duplexes can be bridged by a target loop anywhere from 1 to 15 nucleotides. The authors also tested whether optimal affinity translates into enhanced silencing of miRNA‐targeted mRNAs. To this end, they utilized a reporter RNA with a single miRNA–target site that was transfected into HEK293 cells. The authors observed that a target containing a 4 nucleotide bridge between a seed and a GC‐rich supplementary sequence was repressed 2‐fold better than a reporter with a target maintaining only seed pairing. Even targets containing up to a 15 nucleotide bridge were repressed better than a reporter containing a seed‐only target.

miRNAs can exhibit differences from their annotated mature sequences. These miRNA isoforms (isomiRs) can include additional 3′ terminal nucleotides (3′ isomiRs) (Morin et al, 2008). Sheu‐Gruttadauria and colleagues also observed that 3′ isomiRs generate stronger supplementary pairing as compared to a canonical miRNA, potentially as a result of reduced tension in the miRNA tail region, which could make the supplementary chamber more available. This may explain, as the authors point out, why other groups have previously found supplementary base pairing on 21‐mer miRNAs to have a more modest effect on miRNA–target affinities (Wee et al, 2012; Salomon et al, 2015).

In summary, Sheu‐Gruttadauria and colleagues demonstrate how supplementary base pairing occurs in an Ago2–miRNA–target complex, and that supplementary miRNA–target pairing can indeed enhance both target affinity and miRNA‐mediated repression. As the authors mention, supplementary sites of even two GC base pairs can significantly increase target affinity, which may make some supplementary sites difficult to predict. Nevertheless, this paper provides us with an updated set of rules detailing how miRNAs engage their targets in order to post‐transcriptionally regulate gene expression.

The EMBO Journal (2019) 38: e102477