A Conversation with Nicholas Proudfoot.

Dr. Proudfoot: The concept of the gene is that genes make proteins, so the critical genes in mammalian genomes would be protein-coding genes. Obviously, it was appreciated that there were lots of other transcription units that made structural RNAs like ribosomal RNA or tRNA, and these are actually made by different RNA polymerases. But the RNA polymerase II that makes the proteincoding genes also does a lot of other transcription that doesn’t seem to be directly related to producing a messenger RNA and then a protein. When mammalian proteincoding geneswere first picked apart, it was clear that a large fraction of the gene is actually noncoding; it’s intronic. You get extraordinary situations where 90%–95% of the transcription unit is actually intronic and is removed by splicing and then largely degraded. Then, once genomic or transcriptomic analysis became possible and you could really get a more complete and higher-resolution profile of transcription across the genome, there was the realization that there are a bunch of transcripts that were entirely noncoding. Initially, the simplest interpretation on the discovery of these noncoding RNAs was that if they exist, they must be there for a purpose. The problem is that, in general, these noncoding RNAs are very unstable and they are rapidly degraded, so the cell doesn’t usually take a lot of care over producing very much of them.