Unlocking the clock

Ravi Allada (left) and Kevin Keegan (right) getting into the rhythm of their analysis. Credit: Courtesy of Ravi Allada" class="fr-fic fr-dii">Ravi Allada (left) and Kevin Keegan (right) getting into the rhythm of their analysis. Credit: Courtesy of Ravi AlladaIn October 2004, just as the clocks were changing in Illinois, the focus of Kevin Keegan's graduate work on circadian clocks was changing, too. Keegan, a PhD student at Northwestern University in Evanston, Ill., was developing computer code to analyze preliminary data coming out of a microarray study to find Drosophila circadian rhythm genes. "On a lark," he decided to test his code on some previous circadian gene-hunting studies.Together with his supervisor, Ravi Allada, Keegan ran the data from those five reports through his pilot algorithm, and out popped a complete surprise: many new genes that the reports had missed and had never been implicated before in daily timing. "I totally didn't expect to find a number of novel genes," he says. Quickly, the original "side project" of sifting through the existing data to find the best set of circadian genes became a top priority.The signature of a circadian gene is that its expression levels oscillate once each day. Some genes peak in the evening, others in the daytime, but all such genes cycle on a 24-hour period. At the heart of the fly's circadian rhythm lies a core set of "pacemaker" genes, which control the fly's sleep-wake cycle and its daily rhythms in temperature and hormones, but many other interacting genes remain to be found. "The gears of the clock are well understood, but we don't understand what turns the hands," says Allada.The problem with the original five studies, published in 2001-2002, was that each one identified a different list of clock-related genes - a total of 458 unique genes. Though they all used essentially the same fly stocks, the same lab protocols for isolating RNA, and the same Affymetrix GeneChip technologies, the average overlap between any two lists was only 23%, suggesting that something was amiss. Seven genes were found in common among all five studies, and these included most, but not all, of the known pacemaker genes. But what about the hundreds of other genes?Keegan and Allada's technique added a key element: an ANOVA test that screened the list of candidate genes and tossed out those whose expression levels did not significantly peak and trough over 24 hours. This drastically reduced the number of genes. And when they combined all five datasets, they uncovered a suite of new genes whose activity levels cycled on a daily basis. "It shows how important algorithm choice is," says Keegan."I believe their analysis is much better than ours was," says Michael Rosbash, a molecular geneticist at Brandeis University in Waltham, Mass., and the author of one of the original circadian rhythm studies. "They thought about how to do this analysis in a better way, and by applying some much more sophisticated algorithms, they did."What Keegan and Allada came up with in the end was a list of 214 genes, more than half of which were not found in any of the five original studies (PLoS Comp Biol, 3:e208, 2007). The types of genes they found also ran the gambit of functions, from protein kinases to ion channels. "It looks like a Noah's Ark of gene classes," says Allada. "There's a little bit of everything in here."To verify their results, they used quantitative real-time PCR to verify three randomly chosen genes, and the experiment confirmed that the two genes found using the most stringent ANOVA analysis were indeed involved in circadian rhythm. Molecular work showed that a gene with a particularly promising rhythmic pattern inhibited the activity of one of the core pacemaker genes called clock. Because it had a conserved amino acid sequence known as an Orange domain, Keegan and Allada decided to call the gene clockwork orange, referring to the well-known Anthony Burgess novel and Stanley Kubrick film.Keegan and other members of Allada's lab are now going through the 200-odd genes they found to systematically disrupt their functions and see if they affect the fly's periodic behavior. "My guess is we're going to see some interesting phenotypes," he says.

Unlocking the clock

Interested in reading more?

Become a Member of