For scientists studying genes and the genome, nucleosomes present an interesting puzzle. What relationship these protein 'beads' strung onto DNA strands have to genetic sequences has been a long-standing question.

This is important because nucleosomes, which pack DNA into a cell's nucleus, block the binding of other proteins. So the presence of a nucleosome at a specific location means that transcription cannot take place there. But genomicists haven't been able to work out how nucleosome location is decided or what makes these proteins bind to specific sites. “There's been a debate in the literature about what determines the nucleosomes' positions,” says Eran Segal, a computational biologist at the Weizmann Institute of Science in Rehovot, Israel. “Is there some specificity in the DNA?”

To answer this question, Segal's team joined up with a computational biology group at Northwestern University in Evanston, Illinois, led by Jonathan Widom. They looked at yeast, because it is a small, well-known organism, and used a DNA-sequencing approach to compile a collection of genomic locations that are bound by nucleosomes in vivo. Then they planned to look for sequence commonalities among all nucleosome locations.

“What we tried to do — which did not work — was take the sequences, build models for them, and plot a score to show how well the nucleosomes should fit on the basis of binding sites,” Segal says. “But when we tried to do that, we basically didn't get a signal.”

So Segal and his colleagues considered a number of different approaches. “We ended up using a framework from statistical mechanics,” Segal says. Their approach was akin to solving a jigsaw puzzle, using a few basic rules. For instance, nucleosomes are about 150 base pairs long and cannot overlap, so a section of the DNA sequence could be excluded from the search each time the researchers located one nucleosome. They used this information, combined with statistical rules, to determine the probability of a nucleosome residing at any particular base pair. Just like in a jigsaw puzzle, the more nucleosomes they placed the easier it became to determine the positions of others.

This information was then used to establish a computational model for placing nucleosomes, with each location grounded in solid statistical rules. They used this model to predict the location of nucleosomes along the full length of the yeast geneome. Then, they carried out experiments to see how well their predictions worked, using mutations in short stretches of DNA to see how nucleosome binding was affected.

Both the computational and experimental approaches, described on page 772, were necessary. “Widom's lab carried out the biological experiments, whereas my lab carried out the model construction,” says Segal. “I don't think either of us would have succeeded alone.”

There is still much work to be done. The researchers' model explains only about 50% of nucleosome organization. And a better understanding of how nucleosomes and DNA interact is still needed. But this work has established that DNA sequence does have a role in determining where nucleosomes occur, effectively dictating how and where DNA can wrap itself around protein packages.