Credit: Reprinted with permission from Song et al.

In most eukaryotes, double-stranded RNAs (dsRNAs) can trigger sequence-specific gene silencing. These RNA molecules are processed into short duplex RNAs ∼21–25 nucleotides in length with 3′ overhangs of two bases. One strand of these duplexes is incorporated into a nucleoprotein complex called the RNA-induced silencing complex (RISC). The argonaute protein (Ago) recognizes the 3′ overhang of the single-stranded RNA (ssRNA) and is a crucial component of RISC. The loaded ssRNA guides the search for mRNA with complementary sequences and defines the site of cleavage (∼10 bases from the 3′ overhang), but the identifity of the RNA cleaving activity ('slicer') in RISC was not known.

A recent Science paper (Song et al. published online 29 July 2004 doi:/10.1126/science.1102514) reports the crystal structure of a full-length archaeal argonaute protein. The structure reveals that the conserved PIWI domain (purple) has a RNase H fold with an intact 'DDE' motif (ball-and-stick) important for cleavage by RNase H. The PIWI domain forms part of the crescent base, and the PAZ domain (red), which has been shown to recognize the 3′ overhang, hangs above the PIWI domain. The structure suggests that argonaute may function as the slicer and the authors model in the ssRNA as well as the mRNA.

In mammalian cells, four argonaute paralogs have been identified, but it was not clear whether each protein could affect gene silencing. Two independent studies by Meister et al. (Mol. Cell 15, 185–197; 2004) and by Liu et al. (Science, published online 29 July 2004 doi:/10.1126/science.1102513) have now characterized the properties of the human argonaute proteins. Both studies show that the human argonaute proteins seem to load ssRNAs indiscriminantly, but only the ribonucleoprotein complex containing Ago2 can cleave target mRNAs. Liu et al. also show that mice lacking Ago2 die as embryos because they have severe developmental abnormalities. These results suggest that the assembly of the catalytic Ago2 complex may be required during early mouse development. Furthermore, mouse embryonic fibroblast cells lacking Ago2 but expressing the other three argonaute proteins cannot silence gene expression in response to a small interfering RNA. These observations indicate that Ago2 has a specialized biological function distinct from the other paralogs. Finally, Liu et al. mutated two of three residues in human Ago2 that correspond to the DDE motif in the archaeal argonaute protein; these mutations abolish the RNA-cleaving activity of the human Ago2-containing complex. Taken together, these studies pinpoint Ago2 as the slicer in the RISC complex. Why do the other argonaute proteins, which also have the conserved PIWI domain, lack the RNA cleaving activity? Is each of the argonaute proteins involved in a specific pathway of dsRNA-mediated gene silencing? Future studies will have to answer these interesting questions.