Slicer function of Drosophila Argonautes and its involvement in RISC formation

Abstract

Argonaute proteins play important yet distinct roles in RNA silencing. Human Argonaute2 (hAgo2) was shown to be responsible for target RNA cleavage ("Slicer") activity in RNA interference (RNAi), whereas other Argonaute subfamily members do not exhibit the Slicer activity in humans. In Drosophila, AGO2 was shown to possess the Slicer activity. Here we show that AGO1, another member of the Drosophila Argonaute subfamily, immunopurified from Schneider2 (S2) cells associates with microRNA (miRNA) and cleaves target RNA completely complementary to the miRNA. Slicer activity is reconstituted with recombinant full-length AGO1. Thus, in Drosophila, unlike in humans, both AGO1 and AGO2 have Slicer functions. Further, reconstitution of Slicer activity with recombinant PIWI domains of AGO1 and AGO2 demonstrates that other regions in the Argonautes are not strictly necessary for small interfering RNA (siRNA)-binding and cleavage activities. It has been shown that in circumstances with AGO2-lacking, the siRNA duplex is not unwound and consequently an RNA-induced silencing complex (RISC) is not formed. We show that upon addition of an siRNA duplex in S2 lysate, the passenger strand is cleaved in an AGO2-dependent manner, and nuclease-resistant modification of the passenger strand impairs RISC formation. These findings give rise to a new model in which AGO2 is directly involved in RISC formation as "Slicer" of the passenger strand of the siRNA duplex.

Figure 1.

AGO2 immunopurified from siRNA-programmed S2 cells is associated with guide siRNA and exhibits target RNA cleavage activity. (A) Silver-staining of the immunoprecipitate using a specific monoclonal antibody against AGO2. The lane indicated as -lysate shows the proteins that originated from the antibody. The lane indicated as +lysate contains a band corresponding to AGO2 (shown with an asterisk). Other protein bands were not observed in the +lysate lane compared with the -lysate lane, indicating that the AGO2 immunoprecipitated was nearly uniformly purified. (h.c.) Heavy chains of the antibody; (l.c.) light chains of the antibody. (B) Northern blot shows that luc guide siRNA is specifically associated with the AGO2 immunopurified in A. The lane indicated as n.i.-IP shows that luc guide siRNA is not in the immunoprecipitate with nonimmune IgG used as a negative control. (C) In vitro target RNA cleavage assay using luc target RNA. Immunopurified AGO2, as shown in A, shows ability for cleaving the target RNA (luc180; 5′-radiolabeled with ³²P), whereas the control (n.i.-IP) does not.

Figure 2.

AGO1 immunopurified from naïve S2 cells is associated with endogenous miRNA and cleaves target RNA harboring a sequence completely complementary to the miRNA. (A) Silver-staining of the immunoprecipitate with anti-AGO1 monoclonal antibody. The lane indicated as anti-AGO1 shows the protein components in the immunoprecipitate with anti-AGO1 antibody. The n.i. lane contains the protein components in the immunoprecipitate with negative control nonimmune IgG. A doublet corresponding to AGO1 is shown with an asterisk. The same protein band is not observed in the n.i. lane. (h.c.) Heavy chains of the antibody; (l.c.) light chains of the antibody. (B, top) Northern blot shows that luc guide siRNA is not associated with AGO1 immunopurified from luc siRNA-programmed S2 lysate. (Bottom) miR-ban, a miRNA known to be expressed in S2 cells, is associated with immunopurified AGO1. (Bottom) Under the same conditions, the AGO2 immunoprecipitated was not associated with miR-ban. (C) In vitro target RNA cleavage assay using miR-ban target RNA (bantam38), which was radiolabeled with ³²P at the 5′-end. Immunopurified AGO1 (AGO1-IP) (A) and AGO2 (AGO2-IP) (Fig. 1A) was incubated with bantam38, and then the reaction RNAs were prepared and analyzed on a gel. AGO1-IP shows the ability to cleave bantam38, whereas the control (n.i.) and AGO2-IP showed no such activity for cleaving bantam38. (D) The in vitro target RNA cleavage assay shown in C was repeated. The supernatant and the bead fractions were separated after reaction and RNAs were prepared and analyzed. The cleaved product is observed in the supernatant fraction, meaning that the product was released in buffer after cleavage. The same was observed even under conditions without ATP (data not shown); indicating that the releasing of cleaved target RNA from AGO1 occurs in an ATP-independent manner.

Figure 3.

Recombinant Drosophila full-length AGO1 and the PIWI domains of AGO1 and AGO2 cleave target RNA, depending on the small RNA sequences with which they are associated. (A) luc target RNA was efficiently cleaved by GST-AGO1 associated with luc guide siRNA. Purified GST-AGO1 (Supplementary Fig. S3) was first incubated with luc guide siRNA and then luc target RNA (luc180) radiolabeled at the 5′-end was added. Addition to the reaction of EDTA at 10 mM abolished the activity. Without siRNA addition, the target RNA cleavage is not observed. GST itself does not show the activity. (B) A similar experiment to A was carried out using miRNA let-7 (miR-let-7) and the miR-let-7 target RNA (Okamura et al. 2004). Target RNA is cleaved in an AGO1-miR-let-7-dependent manner. (C) Reconstitution of Slicer activity with the PIWI domain of AGO2. All of the recombinants used were first incubated with luc guide siRNA and then luc target RNA (luc180) radiolabeled at the 5′-end was added and incubated. GST-AGO2-PIWI-I and GST-AGO2-PIWI-G cover the PIWI domain of AGO2, but GST-AGO2-PIWI-D contains only a portion of the PIWI domain as indicated below. (D) The PIWI domain of AGO1 preassociated with guide siRNA also shows activity to cleave target RNA.

Figure 4.

Cleavage of the passenger strand of siRNA duplex by AGO2. (A) AGO2 immunopurified from siRNA-programmed S2 lysate under high-salt conditions was incubated with luc passenger siRNA radiolabeled at the 5′-end. The passenger strand is cleaved by AGO2 preloaded with luc guide siRNA. luc guide and passenger strand sequences are shown on the right. The cleavage site on the passenger is indicated with an arrow. (B) Detection of the cleaved passenger strand of siRNA duplex programmed in S2 lysates. Two sets of luc siRNA duplex (the nucleotide sequences are the same, but only passenger or guide strand of the duplex was radiolabeled at the 5′-end) were used. The passenger strand, but not guide siRNA being cleaved (9 nt) is detected. (C) In AGO2-lacking embryo lysate (AGO2⁴¹⁴ lysate) (Okamura et al. 2004), the cleaved passenger is not detected, as opposed to the wild-type embryo lysate, which indicates that cleavage of the passenger strand of the siRNA duplex occurs in an AGO2-dependent manner. The luc siRNA duplex (the passenger was radiolabeled at the 5′-end) was used.

Figure 5.

Nuclease-resistant passenger of the siRNA duplex blocks RISC formation, but ATP depletion does not. (A) The reaction shown in Figure 4B was repeated. Either EDTA (to 10 mM) or 2′-OMe-modified passenger was added when the reactions were initiated. Total incubation time was 15 min. Addition of EDTA blocks the cleavage of the passenger; in contrast, 2′-O-methyl-modified passenger does not. (B) The siRNA duplex, of which the passenger strand was modified with the 2′-O-methyl group at the ninth nucleotide (OMe-9) and radiolabeled with ³²P at the 5′-end, was programmed in S2 lysate. In si-duplex (OMe-9), the cleaved passenger is not observed, unlike with nonmodified si-duplex. (C, left) Native agarose gel electrophoresis shows that even after 30 min incubation, RISC formation is strongly impaired with si-duplex (OMe-9) (the passenger was modified with the 2′-O-methyl group at the ninth nucleotide, and the guide was labeled with ³²P at the 5′-end). (Right) The target RNA cleavage activity of S2 lysate was also markedly impaired when the si-duplex (OMe-9) was preprogrammed to the lysate. These results suggest that cleavage of the passenger strand of the siRNA duplex by AGO2 is required for efficient formation of active RISC. A target RNA cleavage assay was performed as described previously (Okamura et al. 2004). luc180 target RNA was used as in Figure 1C. (D) Even in ATP-depleted lysate (indicated as -ATP), RISC is formed as well as that with ATP (+ATP). To create -ATP conditions, S2 lysate was treated with hexokinase in the presence of glucose as previously reported (Nykanen et al. 2001). RISC formation was performed with gel-purified luc siRNA duplex (the guide strand was radiolabeled with ³²P at the 5′-end and the passenger was phosphorylated with cold ATP). ATP, GTP, creatine phosphate, and creatine kinase were omitted for the -ATP reaction. (E) RISC formed in ATP-depleted lysate is active. ATP-depleted S2 lysate was prepared as in D. Target RNA cleavage assay was performed with luc180 target RNA.

Figure 6.

(A) Recombinant GST-AGO1 was incubated with luc siRNA duplex. A 9-nt RNA product is observed in the GST-AGO1 lane, but not in the GST-only lane. (B) Reconstitution of Slicer activity with recombinant GST-AGO1, the siRNA duplex, and its target RNA in vitro. GST-AGO1 preprogrammed with the siRNA duplex shows Slicer activity, although it is not as robust compared with GST-AGO1 preassociated with guide siRNA.