DICER cleavage fidelity is governed by 5′-end binding pockets

Steven

DICER cleavage fidelity is governed by 5′-end binding pockets

Article Date: 04 March 2026
Source: https://www.nature.com/articles/s41586-026-10211-5

Summary

This study combines massively parallel dicing assays and high-resolution cryo-electron microscopy to revise how DICER recognises RNA termini and selects cleavage registers. The authors identify two distinct 5′-end binding pockets in human DICER: a previously described U-favouring pocket (involving R1003 and R821) that biases cleavage to 22 nt from the 5′-end (DC22), and a newly discovered G-favouring pocket (centred on D991 and H992) that biases cleavage to 21 nt (DC21). Biochemical assays, mutant DICER proteins and cellular sequencing validate that 5′-G and 5′-A tend to promote DC21 whereas 5′-U and 5′-C favour DC22. Cryo-EM shows coordinated inward motions of the PAZ and dsRBD domains that clamp and align the duplex for precise catalysis, and reveals how sequence motifs (for example, mWCU and YCR) can cooperate with — or override — end-binding preferences by inducing RNA conformational changes. The dual-pocket model is supported across species (including Drosophila Dcr-1) and has implications for shRNA design and prediction of miRNA isomiR registers.

Key Points

  • DICER has two distinct 5′-end binding pockets: a G-favoured pocket (D991, H992) that promotes DC21, and a U-favoured pocket (R1003, R821) that promotes DC22.
  • Massively parallel dicing assays across hundreds of variants show 5′-G and 5′-A bias cleavage to DC21; 5′-U and 5′-C bias to DC22. 5′-G does not reduce accuracy as previously suggested — it often enhances DC21 precision.
  • Cryo-EM reconstructions (3.0–3.4 Å) reveal PAZ and dsRBD movements that compact and align RNA in a dicing-ready conformation; these domain motions act like ‘chopsticks’ to position the duplex.
  • Sequence motifs (mWCU, YCR), stem length and structural features cooperate with end-binding rules; strong motifs can override a 5′-end preference by distorting the RNA backbone to reposition the scissile bond.
  • Mutations of D991/H992 disrupt the G-pocket and reduce DC21 preference for 5′-G substrates; cross-species analyses and fly Dcr-1 assays indicate evolutionary conservation of the dual-pocket mechanism among metazoans.
  • Practical implication: shRNA design (often starting with a 5′-G from Pol III transcription) should account for the G-pocket bias to optimise desired cleavage registers and knockdown efficiency.

Context and relevance

DICER is central to small-RNA biogenesis and RNAi-based technologies. This work updates the mechanistic model for how DICER counts from the 5′ end and explains variability in miRNA length/isomiR registers observed across species and in cellular data. By resolving two discrete 5′-end pockets and showing how RNA motifs and domain dynamics integrate with end recognition, the paper connects sequence-level rules to atomic interactions — bridging biochemical assay results with structural detail. The findings are directly relevant to anyone designing shRNAs or interpreting isomiR datasets, and they refine predictive rules for how a pre-miRNA sequence will be processed.

Why should I read this?

Short version: if you work with miRNAs, RNAi tools or want to design better shRNAs, this paper tells you why the first nucleotide matters — and how. It’s a neat bit of structure + high-throughput biochemistry that explains odd cleavage patterns you may have seen and gives actionable guidance for design tweaks (eg, mind the first base and motifs). If you hate surprises in your knockdown results, read this — it’ll save you fiddling around blind.

Author take

Punchy and important: the discovery of a genuine 5′-G pocket overturns the simple notion that G is universally disfavoured and provides a concrete structural handle (D991/H992) that determines register. The study is methodologically robust — combining multiplexed assays, mutagenesis, cell-based sequencing and multiple cryo-EM structures — so the model is persuasive and immediately useful to the field.

Source

Source: https://www.nature.com/articles/s41586-026-10211-5

Data & resources: cryo-EM maps and PDB models are deposited (EMDB and PDB accession codes listed in the paper); massively parallel dicing data are in GEO (GSE296721).