Target RNA-directed trimming and tailing of small silencing RNAs

Abstract

In Drosophila, microRNAs (miRNAs) typically guide Argonaute1 to repress messenger RNA (mRNA), whereas small interfering RNAs (siRNAs) guide Argonaute2 to destroy viral and transposon RNA. Unlike siRNAs, miRNAs rarely form extensive numbers of base pairs to the mRNAs they regulate. We find that extensive complementarity between a target RNA and an Argonaute1-bound miRNA triggers miRNA tailing and 3'-to-5' trimming. In flies, Argonaute2-bound small RNAs--but not those bound to Argonaute1--bear a 2'-O-methyl group at their 3' ends. This modification blocks target-directed small RNA remodeling: In flies lacking Hen1, the enzyme that adds the 2'-O-methyl group, Argonaute2-associated siRNAs are tailed and trimmed. Target complementarity also affects small RNA stability in human cells. These results provide an explanation for the partial complementarity between animal miRNAs and their targets.

Fig. 1

Target-directed destabilization of miRNAs. (A) Northern analysis of total RNA from S2 cells stably expressing egfp mRNA bearing in its 3′ UTR two target sites for miR-1 [(miR-1)₂] or miR-34 [(miR-34)₂] or four target sites for let-7 [(let-7)₄]. Time of ecdysone (right panels) or control treatment (left panel) is indicated. Relative levels of the miRNAs are indicated below the lanes. (B) Representative Northern analysis of total RNA from a clonal S2 cell line expressing egfp mRNA bearing in its 3′ UTR two target sites for bantam [(bantam)₂] and a clonal control cell line expressing sole egfp mRNA. (C) Mean ± standard deviation for three biologically independent replicates of the experiment in (B). Levels of each miRNA were first normalized to the 2S rRNA loading control.

Fig. 2

Target-dependent destabilization of miR-277. (A) Change in miR-277 or bantam abundance in total RNA from a clonal S2 cell line expressing egfp mRNA bearing in its 3´ UTR two target sites for miR-277 [(miR-277)₂], relative to a clonal control cell line expressing sole egfp mRNA. Prior to comparison, miRNA levels were normalized to the 2S rRNA loading control. (B) Change in 2′-O-methylated (estimated by resistance to oxidation and β-elimination) and Ago1-associated miR-277 (determined by immunoprecipitation) (see SOM text and fig. S6 for details). In all experiments at least two technical replicates of three biologically independent measurements were used to determine mean ± standard deviation.

Fig. 3

Terminal 2′-O-methyl modification and Argonaute protein identity control small RNA stability. (A) The in vitro miRNA stability assay used in (B) and (D). (B) In the presence of a fully complementary target RNA, but not an unrelated control RNA, let-7 was tailed and trimmed. (C) Endogenous bantam miRNA was tailed and trimmed when Drosophila embryo extract was incubated overnight with a fully complementary target RNA, but not a control target. bantam and 2S rRNA were detected by Northern blotting. (D) 2′-O-methyl modification of an Ago1-loaded miRNA blocked tailing and trimming. (E) The in vitro siRNA stability assay used in (F). (F) A fully complementary target RNA directed tailing and trimming of an Ago2-loaded siRNA in the absence of Hen1. C, control target RNA; P, perfectly complementary target RNA; o/n, overnight.

Fig. 4

Target sequence requirements for miRNA tailing and trimming. (A) Target RNAs assayed for their ability to direct tailing and trimming in vitro. Mismatches are indicated with red, lower case letters. The miRNA seed sequence is highlighted in gray. (B) Seed pairing plus extensive central and 3′ pairing between the miRNA and the target was required for efficient tailing and trimming. The miRNA duplex was assembled into Ago1 by incubation in Drosophila embryo lysate, then the target RNA was added and the reaction incubated overnight.

Fig. 5

Small RNA tailing and trimming in vivo. (A) Length distribution of miRNAs from heterozygous (black) or homozygous (red) hen1^f00810 fly heads. For each annotated pre-miRNA, reads were normalized to sequencing depth, then normalized miRNA reads for each distinct pre-miRNA were weighted equally to eliminate the influence of differences in transcriptional rates. (B) Length distribution of endo-siRNA sequence reads perfectly matching the fly genome (top panels) or matching only within a 5′ prefix (bottom panels) for heterozygous (black) or homozygous (red) hen1^f00810 heads. The three classes of endo-siRNAs were analyzed separately: siRNAs derived from natural antisense transcripts (cis-NATs), from transposons, or from hairpin loci (esiRNAs). The most frequent non-genome matching nucleotide additions are indicated in the gray boxes as the percentage of all non-templated additions for specific prefix lengths. Reads are in parts per million (ppm). (C) The sequence of the esi-2.1 duplex and its cellular mRNA target, mus308. Northern analysis was used to detect esi-2.1 in total RNA from whole Oregon R (wild type), hen1^f00810 or ago2⁴¹⁴ mutant flies, and the absolute signal intensities (I, log₁₀ scale) determined for each lane. Tailing and trimming products are marked with red arrowheads.

Fig. 6

Target-dependent small RNA remodeling and destabilization in cultured human HeLa cells. (A) Northern analysis of total RNA from HeLa cells transfected with increasing amounts of 21 nt target RNA analogs (antagomirs) fully complementary to miR-16 or miR-21. Arrowheads mark tailed or trimmed miRNAs. (B) Total miRNA length distribution in mock transfected HeLa cells and in cells transfected with 1 nM antagomir targeting miR-16 (top) or 10 nM antagomir targeting miR-21 (bottom). (C) and (D) Small RNA profile from HeLa cells mock transfected (black) or transfected (red) with 1 nM antagomir targeting miR-16 (C) or 10 nM antagomir targeting miR-21 (D). Left panels: reads perfectly matching the miR-16-1 and miR-16-2 loci (C) or the miR-21 locus (D). Right panels: reads with prefixes matching miR-16-1 and miR-16-2 (C) or miR-21 (D). Non-genome matching nucleotides added to the 3′ end of the prefixes that most increased in response to antagomir transfection are shown (grey box). Reads are in ppm. (D) Endogenous miR-16 and miR-21 were destabilized when HeLa cytoplasmic extract was incubated with a fully complementary target mRNA, but not a control target. miR-16 and miR-21 were detected by Northern blotting.