RNA silencing allows cells to block invading viruses or mobile DNAs. An RNA-cleaving enzyme involved in the first step of silencing has now been identified.
Several years ago, I heard a famous geneticist describe the difference between talks given by geneticists and those by biochemists. To him, a geneticist's talk was one with ideas but no data; a biochemist's talk was just the opposite. I doubt that such an outrageous suggestion would be made nowadays, because so many ideas and data involve a combination of genetics and biochemistry. One such example is provided on page 363 of this issue1, where Bernstein and colleagues describe how they have identified — in the fruitfly Drosophila melanogaster — an enzyme that they call 'Dicer' because it chops RNA into small pieces of uniform size.
Bernstein et al. were investigating a process known as RNA interference or RNA silencing. This phenomenon provides organisms with a defence against mobile DNA elements, which cause mutations when they insert themselves within, or close to, a gene2. In plants, RNA silencing may also offer protection against viruses3. The existence of RNA silencing, at least in animals, has been inferred from experiments in which double-stranded (ds)RNA was introduced into cells4. Most RNA in a cell is single-stranded, so the presence of dsRNA might signal to the cell that it is being invaded by mobile DNA or viruses.
RNA is cut up into smaller chunks during at least two steps of RNA silencing. First, dsRNA is processed into many short pieces, each of about 22 nucleotides5. These fragments are then incorporated into a multi-subunit complex, which carries out the second RNA-degradation reaction. Here, the target RNA is single-stranded messenger RNA6. Apparently, the role of the 22-nucleotide RNAs generated in the first step is to interact, by base pairing, with mRNA in which the sequence of nucleotides is the same as in the dsRNA. The short RNA molecules thus guide the RNA-degrading enzyme (the RNase) part of the multi-subunit complex, ensuring that it degrades only those mRNA species that are related to the dsRNA.
An advantage of this two-stage process is that each molecule of dsRNA primes several RNase molecules with a guide RNA. Consequently the cell can mount a large response to only a few dsRNA molecules. This feature is ideal for defence against viruses or mobile DNAs that produce both single- and double-stranded RNA. The virus would be arrested early in an infection and the mobile DNA would be frozen before it had damaged the genome.
Bernstein and colleagues1 now show that different RNases are required for the two RNA-degrading steps. They give the name Dicer to the enzyme needed for the first step. The authors started their search for this enzyme by assuming that it would be similar to the known RNases that are specific for dsRNA. These enzymes are mostly members of the RNase III family, and have characteristic motifs in their protein sequence. Bernstein et al. identified candidate enzymes by scanning the Drosophila genome for genes encoding proteins with these RNase III signatures.
The authors then used three tests to confirm that one of these genes, dicer, encodes the RNA-processing enzyme needed for the first step of RNA silencing. First, they showed that dicer encodes a protein that processes dsRNA into 22-nucleotide fragments in vitro. Second, antibodies raised against this protein recognized an enzyme in Drosophila extracts that process dsRNA into 22-nucleotide fragments. Third, when Bernstein et al. introduced dicer dsRNA into Drosophila cells, they found that the cells showed a diminished ability to carry out RNA silencing of a gene encoding green fluorescent protein. In effect, the authors were using RNA interference to suppress a gene required for RNA interference. It would not be possible to use RNA interference to suppress dicer completely if this gene is indeed required for silencing. But, as observed by Bernstein et al., it was possible to get partial suppression.
One feature of the amino-acid sequence of Dicer — in addition to the RNase III motifs — is a sequence known as the PAZ motif7. This sequence is also present in the Drosophila ARGONAUTE protein and related proteins in the thale cress Arabidopsis thaliana8, the nematode worm Caenorhabditis elegans9 and the fungus Neurospora crassa10. These proteins do not have the RNase III motifs, but they are nevertheless required for RNA silencing. A clue to the role of the PAZ motif comes from Bernstein et al.'s unpublished data. Apparently, the multi-subunit RNA-processing complex needed for the second RNA-silencing step in Drosophila contains a relative of ARGONAUTE; perhaps Dicer and this protein interact through their PAZ domains (Fig. 1). The interaction would explain how the 22-nucleotide RNAs produced by Dicer could be incorporated into the multi-subunit complex that degrades single-stranded RNA.
No mutants of dicer have yet been identified. But mutation of the CARPEL FACTORY gene — a relative of dicer in Arabidopsis — leads to plants that grow abnormally and have defective flowers11. Strikingly, this mutation and others that affect both RNA silencing and growth8,12 all affect either animal germline cells (from which reproductive cells form) or their plant equivalent, meristems. A defensive RNA-silencing system would be especially important in these cells, because a virus infecting such a cell would be transmitted from one generation to the next. Similarly, mutations induced in these cells by mobile DNA would result in heritable genetic instability.
But why should loss of a defensive RNA-silencing system result in defective growth of these mutants? One possibility is that RNA silencing is also involved in regulating genes required in the germ line for normal growth and development. Alternatively, the growth defects might be a side effect of the loss of the defence mechanism in germline cells. A combination of biochemical and genetic approaches, laced with ingenuity, will be needed to dig further into the mechanism and role of RNA silencing in defence and growth.