RISC-y Business

Recent studies are revealing more about how the protein complex called RISC--RNA-induced silencing complex--works. RISC is at the heart of the gene-silencing technique called RNA interference, which has skyrocketed as a tool for functional genomics and as an approach to potential therapeutics.

In RNAi, one of the strands from double-stranded short-interfering RNAs (siRNAs) 19 to 21 nucleotides long enters RISC and guides the complex to a complementary strand of a target messenger RNA; the two strands form an siRNA-mRNA duplex. The complex then cleaves the mRNA, preventing its expression. How the complex chooses the correct strand of the siRNA is not yet fully understood; scientists believe that it involves recognition of the 3 and 5 termini of the siRNA strand--also called the guide strand--by RISC.

Now, two papers from independent research groups, one led by David Barford at the Institute for Cancer Research, London, and the other led by Dinshaw J. Patel of Memorial Sloan-Kettering Cancer Center, New York City, suggest how RISC recognizes the 5 phosphate of the siRNA (Nature 2005, 434, 663 and 666).

A specific domain in a protein called Argonaute, found in many species in different forms, appears to be key. Humans, for example, are known to produce four Argonautes, designated Ago1-Ago4. Argonautes contain four domains, including the PIWI domain, which is thought to contain the catalytic site that cuts the mRNA.

Last year, Leemor Joshua-Tor, Gregory J. Hannon, and coworkers at Cold Spring Harbor Laboratory in Long Island, N.Y., found that the PIWI domain in a full-length Argonaute from the archaebacterium Pyrococcus furiosus contains a fold similar to that in RNAse H enzymes, which normally cleave the RNA in RNA-DNA hybrids (Science 2004, 305, 1434). Also last year, Barford's group found that the Piwi protein from another archaebacterium, Archaeoglobus fulgidus, also contains an RNAse H fold (EMBO J. 2004, 23, 4727). The A. fulgidus Piwi protein contains a domain similar to the PIWI domain of Ago2, but is only half the size of the more complicated protein.

"Though there was an appreciation that siRNA 5-end recognition was critical for RISC function, no definitive information was available on which domain of Argonaute is involved in the recognition," Patel says. That information is important, he adds, "because the cutting of the message strand is counted from the 5 end" of the guide strand. Earlier biochemical studies have shown that the mRNA is cleaved between the bases corresponding to nucleotides 10 and 11 on the guide strand.

The new work from both the Barford and Patel groups centers on crystal structures of a complex of an RNA strand with the Piwi protein from A. fulgidus. Piwi "turned out to be an interesting module because it had all the elements necessary to recognize the 5 phosphate," Patel says.

Both studies showed that the 5 phosphate dips into a binding pocket in a cleft that Barford had spotted in the earlier structural studies of Piwi alone. That cleft is present in Argonaute proteins across all species. Within the cleft, the 5 phosphate directly contacts a bound metal ion, two lysine residues, a tyrosine, and a glutamine. "Those four residues are absolutely invariant throughout all Argonaute sequences," Barford says. "That's a very tight binding site to bind the phosphate."

IN ADDITION, in the cleft, the 5 nucleotide on the guide strand is not paired with the corresponding nucleotide on the target strand. Instead, it stacks with the conserved tyrosine. "The entire nucleotide at the 5 position of the guide strand is really buried within this cavity," causing the siRNA-mRNA duplex to unwind, Barford says.

In addition to obtaining the structure, Patel teamed up with Thomas Tuschl at Rockefeller University, New York City, to see what happens if those key amino acids seen in the Piwi structure are mutated in human Ago2. Such mutations result in reduced mRNA cleavage, they found.

"The challenge for the field right now is not just to solve the structure of an Argonaute but to solve it with an siRNA in it," Patel says. Because all the successful structural work so far has involved bacterial systems, he points out, "the other challenge is to solve the structure of human Argonaute, either in the free state or bound to siRNA."

One of the hurdles in getting the crystal structure of human Argonaute has been obtaining a sufficient quantity of the protein. Joshua-Tor and Hannon may have a way around that problem. They can now use the bacterium Escherichia coli to produce recombinant human Argonaute, something that hadn't been done previously (Nat. Struct. Mol. Biol. 2005, 12, 340).

Using recombinant Ago2, Joshua-Tor and Hannon discovered the surprising role of the 5 phosphate in the cleavage reaction. "People call the 5 phosphate a licensing factor to get into the RNAi pathway," Joshua-Tor says. But the phosphate isn't absolutely necessary for cleavage activity, they found. Instead, it is important for the fidelity of cleavage.

Joshua-Tor and Hannon determined that the phosphate holds the target strand in the right position. "If you don't have the phosphate, you can make the guide strand slip," Joshua-Tor says, causing the cleavage to occur one phosphate over from the place it normally happens.

By combining just the recombinant Ago2 with an siRNA, Joshua-Tor and Hannon were able to cleave a target mRNA--even though there's more to the full RISC complex than just Ago2. "E. coli does not have an RNAi pathway," Joshua-Tor points out. "We settled any remaining doubts that Ago2 was indeed the activity that cuts the target mRNA."

Another experiment described in the same paper may explain why Ago2 is the only one of the four human Argonaute proteins that cleaves RNA. RNAse H enzymes usually have an active site that contains two aspartates and a glutamate. Yet, when Joshua-Tor and Hannon soaked crystals of the P. furiosus Argonaute with manganese, they found that manganese binds the two aspartates and an unexpected histidine. This unexpected catalytic motif could help explain why Ago1 can't cleave RNA: because it has arginine instead of the histidine. Ago3's lack of catalytic activity is still a mystery, because it has the unusual motif yet is unable to cleave RNA.

Joshua-Tor and Hannon are also trying to figure out where the 5 phosphate goes. To do that, they soaked the Argonaute from P. furiosus in a solution containing tungstate, which they thought might preferentially bind to the same site that binds the 5 phosphate. The tungstate binding they found suggests a binding site that is different from that predicted in Barford's and Patel's papers. Joshua-Tor and Hannon predict that 5 phosphate binds at the end of a long groove in a region called the PIWI box.

In most ways, the minimal RISC--Ago2 with its associated siRNA--behaves just like the full RISC complex. The only way it differs is in turnover. In the full complex, turnover is known to depend on the presence of adenosine triphosphate. "We knew that Ago2 didn't have an ATP-binding site, so we weren't surprised that there was no ATP dependence," Joshua-Tor says. She was surprised that the kinetics of the cleavage reaction with just Ago2 were so similar to those of the cleavage reaction with the full RISC.

There are still hurdles to understanding RNAi fully. "The challenge is out there: Try to solve a full-length human Argonaute structure, preferably with an siRNA bound; look at the contacts; and tell us which contacts are important and which ones are less important," Patel says. "We could then design our guide siRNAs to be most effective" at targeting particular mRNAs.