Both systems illustrate a solution to a puzzling problem in chromatin biology. How can such a wide variety of sequences be recognized and directed toward a single fate—either silencing in heterochromatin or processing in Tetrahymena? RNAi may provide the answer.
The story begins with transposons. These DNA invaders are characterized by little other than their overactive transcription from either end, leading to complementary transcripts that can anneal to each other. The resulting double-stranded RNA (dsRNA) “is the one thing a cell can grab onto,” says Gorovsky. The RNAi machinery chops up dsRNAs, and the resulting small RNAs target any further transcripts for destruction.
Now, the CSHL and Whitehead groups have found that a similar process may be occurring at the repeats flanking the fission yeast centromere. The Whitehead group found small RNAs complementary to these repeats, with the CSHL group characterizing forward and reverse transcripts and centromere-localized RNA-dependent RNA polymerase that are probably the source of the small RNAs. The CSHL group also found that forward transcription, centromeric silencing, and histone modifications at the centromere are RNAi dependent. Thus, chromodomain proteins may bind the dsRNAs and methylate histone H3, leading to recruitment of chromatin silencing proteins such as Swi6.
The proposal in Tetrahymena is similar, but here the small RNAs are used to mark the chromatin for elimination. Widespread transcripts from one nucleus are matched against the DNA of another, and anything left over is characterized as a recent invader that should be purged. Gorovsky's group found small RNAs and movement of RNAi machinery between nuclei that fits such a model, and a group led by David Allis (University of Virginia, Charlottesville, VA) made the connection between histone H3 modification and DNA elimination.
Suspicions about a DNA connection are not new—both transposons and dsRNA constructed from untranscribed regions can cause chromatin changes. But the new work provides a direct connection. It does not, however, indicate whether RNAi causes silencing of noncentromeric heterochromatin. Heterochromatin is packed with repeats that could yield dsRNA transcripts, but worms and plants lacking RNAi components, although reported as embyronic lethal, have not been examined for heterochromatin defects.
A more imponderable question involves evolution. Did a system for shutting down parasitic transposons get coopted to build centromeres, mark Tetrahymena sequences for deletion, and stabilize repetitive, rearrangement-prone areas of the mammalian genome? Or did evolution run that sequence in reverse? Either way, RNAi has been a key determinant of genome dynamics. ▪