Distinct roles for Argonaute proteins in small RNA-directed RNA cleavage pathways

Abstract

In mammalian cells, both microRNAs (miRNAs) and small interfering RNAs (siRNAs) are thought to be loaded into the same RNA-induced silencing complex (RISC), where they guide mRNA degradation or translation silencing depending on the complementarity of the target. In Drosophila, Argonaute2 (AGO2) was identified as part of the RISC complex. Here we show that AGO2 is an essential component for siRNA-directed RNA interference (RNAi) response and is required for the unwinding of siRNA duplex and in consequence assembly of siRNA into RISC in Drosophila embryos. However, Drosophila embryos lacking AGO2, which are siRNA-directed RNAi-defective, are still capable of miRNA-directed target RNA cleavage. In contrast, Argonaute1 (AGO1), another Argonaute protein in fly, which is dispensable for siRNA-directed target RNA cleavage, is required for mature miRNA production that impacts on miRNA-directed RNA cleavage. The association of AGO1 with Dicer-1 and pre-miRNA also suggests that AGO1 is involved in miRNA biogenesis. Our findings show that distinct Argonaute proteins act at different steps of the small RNA silencing mechanism and suggest that there are inherent differences between siRNA-initiated RISCs and miRNA-initiated RISCs in Drosophila.

Figure 1.

Characterization of the AGO2 deletion mutant AGO2⁴¹⁴. (A) Structure of the AGO2 locus. Exons are indicated with open boxes; the closed portion is the protein-coding region. The translation initiation start site (Met) is located in the first exon. The position of EP(3)3417 is represented as triangle with an arrow pointing in the direction of GAL4-induced transcription. The deleted genomic region of the AGO2⁴¹⁴ chromosome is shown as a dotted line. In this mutant, imprecise excision of the EP element generated a 2.3-kb deletion of genomic DNA, which included exons 1 and 2 of the AGO2 gene. Sequence information from the Berkeley Drosophila Genome Project reveals that the AGO2 locus is positioned in a small region on the cytological location 71E1, on the left arm of the third chromosome. (B) Western blotting on protein extracts from adult ovaries. Western blotting was carried out using anti-AGO2 (4D2), anti-dFMR1 (5A11), and antiribosomal protein P0 antibodies. [WT(yw)] Yellow-white wild type; (AGO2⁴¹⁴) AGO2⁴¹⁴ homozygote; (AGO2⁴¹⁴;WTrescue) AGO2⁴¹⁴ homozygote with two copies of wild-type rescue. The genomic rescue construct contained a 10.1-kb XhoI-FspI fragment of the AGO2 locus in A in which no other known genes are included.

Figure 2.

RNA interference of ftz activity in embryos. Representative embryonic phenotypes of wild-type and AGO2⁴¹⁴ mutant embryos injected with ftz dsRNA are shown. (A) A wild-type embryo injected with water. (B) A wild-type embryo injected with ftz dsRNA has a phenotype similar to that of ftz. (C) An AGO2⁴¹⁴ mutant embryo injected with ftz dsRNA has a wild-type phenotype. (D) An AGO2⁴¹⁴;WTrescue embryo has a phenotype similar to that of ftz.

Figure 3.

AGO2 is required for interference after the formation of the siRNA duplex. (A) Wild-type embryo injected with synthetic ftz siRNA has a phenotype similar to ftz. (B) AGO2⁴¹⁴ embryo injected with synthetic ftz siRNA has a wild-type phenotype. (C) AGO2 is not required for the production of the siRNA duplex. Uniformly ³²P-labeled dsRNA corresponding to the ftz gene, incubated in extracts from yw and AGO2⁴¹⁴ embryos. RNA products of the reaction were analyzed on a polyacrylamide gel. Incubation of lysates with labeled dsRNA generates RNA fragments ∼22 nucleotides (nt) long. [WT(yw)] Yellow-white wild type; (AGO2⁴¹⁴) AGO2⁴¹⁴ homozygote. (D) Analysis of siRNA unwinding in AGO2⁴¹⁴ embryo lysate. The siRNA duplex was incubated in the extracts from wild-type (yw) and AGO2⁴¹⁴ embryos. RNA products of the reaction were analyzed on a native acrylamide gel. The migration patterns of the siRNA duplex, the starting material, and the single-stranded siRNA, the expected product after unwinding, are shown on the left. (Lower panel) The amounts of input siRNA duplex and unwound single-stranded (ss) siRNA at 0 and 3 h incubation are shown. The average of five independent experiments is shown here. (E) Native gel analysis of labeled siRNA-protein complexes from wild-type and AGO2⁴¹⁴ embryo lysates. B and A complexes are observed in both lanes from wild-type and AGO2⁴¹⁴ reactions. These two complexes correspond to intermediate precursors to RISC (Tomari et al. 2004). The complex corresponding to RISC (Tomari et al. 2004) is not detected in the lane from an AGO2⁴¹⁴ reaction.

Figure 4.

AGO2 is not required for miRNA-initiated RNA cleavage. (A) Schematic drawing of the target RNA and its pairing with ftz siRNA and miRNA (miR-2b or let-7). (Magenta) miR-2b, or let-7 and their complementary sequences (below); (green) ftz siRNA and its complementary sequence. An asterisk indicates the 5′-³²P radiolabel. (B) miRNA-directed target RNA cleavage is catalyzed by embryo lysates with and without AGO2. In vitro RNAi assays were carried out with 14-16-h embryo lysates of yw and AGO2⁴¹⁴. Target RNA contains the ftz siRNA and the miR-2b target sequences as shown in A. (C) In vitro processing of pre-let-7 RNAs internally labeled with α-³²P-GTP. Mature let-7-generating activities were compared between lysates from 0- to 2-h yw and AGO2⁴¹⁴ embryos. Both synthetic 21-nt let-7 RNA (not shown) and in vitro processed let-7 migrate at ∼25 nt relative to the markers. The discrepancy between molecular mass and mobility in polyacrylamide gel electrophoresis is frequently observed for small RNAs (Hutvagner et al. 2001; Lee et al. 2003). (D) In vitro RNAi assays with lysates prepared from 0- to 2-h yw and AGO2⁴¹⁴ embryos. The RNA target contains let-7 and ftz siRNA target sequences as shown in A.

Figure 5.

AGO1 is required for efficient miRNA-initiated RNA cleavage. (A) When AGO2 is suppressed by introducing specific dsRNA in S2 cells expressing EGFP, the ability of the cells to silence EGFP by RNAi is severely reduced. In contrast, when AGO1 is suppressed, the EGFP silencing effect is unaffected, indicating that AGO1 is not essential for the RNAi pathway in S2 cells. (B) Lysates were prepared from 14- to 16-h yw, AGO2⁴¹⁴ mutant, and AGO1^k08121 mutant embryos. Western blots were performed using the antibodies against AGO2 (upper), AGO1 (middle), and ribosomal protein P0 (lower) as a loading control. (C) In vitro RNAi assays with lysates prepared from 14- to 16-h yw, AGO2⁴¹⁴, and AGO1^k08121 embryos. The RNA target was the same as that used in Figure 4B. The amount of specifically cleaved bands was quantified and normalized to that of bands in wild-type embryos. (D) AGO1 is not required for the production of the siRNA duplex. Uniformly ³²P-labeled dsRNA corresponding to the ftz gene, incubated in extracts from yw and AGO1^k08121 embryos. RNA products of the reaction were analyzed on a polyacrylamide gel. Incubation of lysates with labeled dsRNA generates RNA fragments ∼22 nt long. [WT(yw)] Yellow-white wild type; (AGO1^k08121) AGO1^k08121 embryos. (E) RISC formation is not impaired in AGO1^k08121 embryo lysates.

Figure 6.

AGO1 is involved in miRNA biogenesis. (A) AGO1 is required for stable, mature miRNA production. Expression of AGO1, AGO2, or EGFP (control) was suppressed by RNAi in S2 cells. Western blots (lower panels) show that the Argonaute proteins are indeed down-regulated. RNAs were obtained from each cell and subjected to Northern blots to analyze the amounts of miR-ban using a specific probe. (B) The miRNA level in the wild-type, AGO1, and AGO2 mutant embryos. Northern blot analysis on the AGO1 mutant (AGO1^k08121) shows a marked reduction of mature miR-ban. (C) Dicer-1 and AGO1 are related to miRNA processing and stabilization, respectively. Expression of either Dicer-1 (Dcr-1), Dicer-2 (Dcr-2), AGO1, AGO2, or EGFP (control) was reduced by RNAi in S2 cells for indicated days, then total RNAs were obtained from each cell and subjected to Northern blots to visualize mature and pre-miR-ban. (D) AGO1 associates with Dicer-1. TEV extract prepared from S2 cells expressing AGO1-TAP, AGO2-TAP, or dFMR1-TAP, or the parental S2 cells (control), were subjected to Western blots using antibodies against Dicer-1, AGO1, dFMR1, and AGO2. Protein bands with asterisks indicate AGO1, AGO2, and dFMR1 with a CBP tag, a converted form of a TAP tag after TEV cleavage. (E) AGO1 associates with both pre-miRNA and mature miRNA. Northern blots show that all AGO1-TAP, AGO2-TAP, and dFMR1-TAP complexes contain mature miR-ban, but the precursor is associated only with the AGO1-TAP complex.