|The Flys Sensational Sequence|
|Columbia University and Celera researchers
use bioinformatics and genomics to
uncover taste receptors in Drosophila
Edward R. Winstead
March 26, 2001
A year after storing the DNA sequences on his computers hard drive, Peter J. Clyne took a closer look at two unidentified fruit fly genes. He had found the sequences among data at the Berkeley Drosophila Genome Project Web site months earlier and set them aside. Now, the Yale University researcher thought he had a pair of the flys taste receptor genessomething no one had ever foundand he was right.
Clyne's success came from using algorithms to search for odorant receptor proteins by their structural features rather than their DNA sequence. The search yielded the two taste receptors, which he used to find others. Having mined several dozen candidates from the fly database in May 1999, he cancelled a vacation to India and teamed up with a Yale postdoc, Coral Warr, to run experiments asking where in the fly the genes were expressed. Eighteen of 19 genes turned up in taste organs. The Yale group, led by John Carlson, published the new gene family in the March 10, 2000 issue of Science.
Now, one year later, Richard Axel, of the Columbia University College of Physicians and Surgeons, in New York, and colleagues confirm the Carlson lab results and expand the gene family. The number of fly taste receptors is at least 56, Axel's group reports in the March 9, 2001 issue of Cell. The researchers systematically screened a variety of tissues for the expression of hundreds of potential taste receptor genes. They pinpointed the expression of taste receptor genes in chemosensory cells of gustatory tissue and found that some genes were also turned on in olfactory tissue.
The gene candidates had been identified by Anibal I. Cravchik, of the Medical Affairs Department at Celera Genomics. He mined the genes from the complete fly genome sequence, which was determined in a collaboration between Celera and the Berkeley Drosophila group. Cravchik's algorithms identified 310 gene candidates based on the predicted structural properties of taste receptor proteins. He then emailed the sequences to Kristin Scott at Columbia University.
Scott used the data to create 'probes' and screen fly tissues for the expression of individual taste receptors. It became apparent to her that these genes are expressed at very low levels, a finding also reported by the Yale group. As the project progressed, discoveries in the flies led to new criteria for additional gene searches and more refined algorithms, a process that eventually yielded 56 genes now known as gustatory receptors (GRs), including 23 new genes.
"None of this would have been possible without the sequencing of the Drosophila genome," says Scott. "Having the taste receptors means we can begin to characterize the neurocircuitry involved in the representation of different chemosensory information in the brain."
Outside the fly's mouth, its legs, wings and female genitalia are decorated with receptor proteins that relay sensory information to the brain. Males use the information to select mates, and females use it to identify safe environments for laying eggs. The anatomical locations of taste receptors were mapped using experiments such as dipping the legs of flies in sugar, which triggers instinctive feeding behavior.
When many of the world's Drosophila experts met at Celera for an initial annotation of the fly genome in November 1999, no taste receptors had been identified among the 13,000 genes in the fly sequence. The Science and Cell papers show why it took bioinformatics and genomics to identify the taste receptors.
"Taste receptors don't look like anything else and are outliers in the genome, so finding them requires individual attention from a specialist," says Leslie B. Vosshall, who represented the Axel laboratory at the annotation. "The commonality among genes is subtle and the algorithms are tricky to run. Anibal was instrumental in finding the range of gustatory receptors that were hiding in the genome of the fly."
Like the Yale team, Cravchik created novel algorithms to search for structural characteristics of proteins rather than DNA sequences. One of the structural traits of chemosensory genes in other species is a series of amino acids that snake back and forth across the cell membrane, resulting in seven distinct domains. "We didn't know what the genes looked like, but we knew they had the 7-transmembrane domain, and we used this secondary structure to search the data," says Cravchik.
The researchers established criteria for their search: Candidate taste receptors had to have no known function, be expressed in taste organs, and have the predicted structural characteristic of a 7-transmembrane domain, which could be determined from a statistical analysis of the sequence. It was also thought that some taste receptors genes might reside together in clusters.
Of the 13,000 Drosophila genes in the sequence, 5,600 had no known function. Cravchik used the algorithm to screen this set and came up with the 310 sequences. Of this group, Scott detected the expression of about eight genes, some of which were expressed in olfactory tissue.
Cravchik revised the algorithm to take into account the possibility that this gene family included features of odorant receptors and then screened the entire genome. This was followed by BLAST searches and more tissue studies, resulting in the 56 taste receptors. More are likely to be found, the researchers say, because the genes are so diverseothers may be lurking undetected in the genome.
Relatively little of the Drosophila genome had been sequenced when Peter Clyne began searching for odorant receptor genes, the project that unearthed the taste receptors. But he thought the odds for finding odorant receptor candidates in even seven percent of the genome were good. Many in the field assumed that chemosensory gene families in the fly would prove to be large.
"The genes were definitely thereanimals are smelling," says Clyne. It was clear, however, that searching the genome for odorant receptors based on their predicted DNA sequence had not yet worked.
"This was like doing library searches by the words in a book compared to doing a search that asks 'What is the meaning of the book?'" says Junhyong Kim, a theoretical biologist at Yale interested in applying bioinformatics to 'real-world' questions in biology.
Kim teamed up with Clyne, Warr and Carlson, who were the experts on olfaction in the fly, to search for the odorant receptors. After the structural characteristics of the predicted proteins had been identified, such as the 7-transmembrane domain, Kim developed numerical measures to describe the structures, constructing what he calls 'feature space.'
"Feature space for people might be composed of body weight and height," explains Kim. "These are numerical measures of some features of the object and several such measures comprise feature space." Kim described the strategy in detail last year in Bioinformatics.
"To be honest, the discovery of the odorant receptor family involved a lot of luck," says Kim. "It just happened that among the first 40 candidates we pulled out two were in the final group. Even when the algorithm works well there are a lot of false positives."
Carlson's group published the family of candidate odorant receptors in Neuron in February 1999. "The two taste receptors sat on the shelf for a year because we became engrossed in the odorant receptors," says Clyne.
Once Clyne had the DNA sequences of the candidate taste receptors, showing that the genes were expressed in taste tissue proved difficult, and he resorted to a technique called RTPCR. The technique is better at finding very low levels of gene expression, but is not as persuasive as direct tissue staining. Experiments with a genetic mutation in Drosophila suggested that in fact the genes were expressed in taste organs.
"I really enjoyed the Cell paper and think it was the logical next step in the research," says Clyne, now at the University of California, San Francisco.
Many in the field view the studies as complementary. "The strength of Kristin Scott's group was their careful characterization of the gene family and filling in of the details," says Leslie Vosshall, now at The Rockefeller University in New York. "The strength of Peter Clyne's group was pulling out of the air the genes everyone had missed."
The Yale group, she adds, published the family without overwhelming evidence that the genes really were fly taste receptors. "And in the end they were absolutely right."
See related GNN article
. . .