Last author

Since the human genome sequence revealed that we have an unexpectedly low number of genes — around 25,000 — scientists have increasingly looked to RNA to explain the much greater number of complex biological functions that occur within us. Individual RNAs can generate and regulate the expression of many proteins, but it has been technically difficult to track RNA's biochemical fingerprint in living tissues. In 2003, Robert Darnell, a neuro-oncologist at Rockefeller University in New York City, and his colleagues tailored an in vitro technique to irreversibly attach RNA via its binding sites to proteins of interest in mouse brain tissue. On page 464, they describe how they modified this technique to create a genome-wide map of the sites where a neuronal protein binds to RNA. Darnell tells Nature more.

What led you to do this work?

We were studying a group of rare brain diseases, called paraneoplastic neurological disorders, which arise in conjunction with immune responses to common cancers that target RNA-binding proteins. We wanted to explore — mechanistically and physiologically — how RNA regulation affects both cancer and the healthy brain. To resolve that dynamic, however, we had to figure out a way to monitor RNA biochemistry in living tissue.

What “Aha!” moments drove this work?

The first occurred seven years ago when we realized that we could apply an old test-tube trick to live brains, and bind RNAs irreversibly to brain proteins. That was the basis of a technique called CLIP — crosslinking immunoprecipitation — which allowed us to sequence the bound RNAs. Next, we realized that high-throughput sequencing methods could give us the resolution needed to create a genome-wide RNA-binding map. But we also had to find a way to sort through the millions of sequences for the biologically relevant binding sites. The 'Aha!' solution rested on the assumption that biologically relevant binding should be reproducible from one mouse brain to another.

Did you find much binding that was not biologically relevant?

No, which was a big surprise. Roughly 90% of the sites where we found RNA bound to protein were consistent from animal to animal. So those sites served as big neon signs pointing to crucial points of RNA regulation. We found a large number of RNA interactions with Nova2, the neuronal RNA-binding protein we were studying. This included interactions at sites that determine how RNA can be spliced — that is, cut and pasted together in various ways — to generate different proteins, and at unexpected sites, such as in non-coding-sequence regions.