UW team clears up gene 'trees'

By SUSANNE QUICK squick@journalsentinel.com, Journal Sentinel

Monday, October 27, 2003

Most diagrams, or trees, designed to depict evolutionary relationships are like kaleidoscopes. While the image always contains the same cast of characters, the relationships between the objects change as the angle of perspective shifts.

When using DNA to sort out evolutionary relationships, the same thing happens.

Depending upon the gene you use, the tree could look like a pine or a willow -- masking or distorting the true relationships of the groups being studied, reported a team of researchers at the Howard Hughes Medical Institute at the University of Wisconsin-Madison in the latest issue of Nature.

"This paper is going to be one of those seminal studies that you see cited by scientists over and over again," said Tom Givnish, a UW botanist who was not involved in the research.

"The results are surprising and it's going to make a lot of people reconsider how much data they need" to make an accurate tree, he said.

Evolutionary trees, or phylogenies, are important to scientists not only because they depict the history of life, but they provide crucial tools in fields as disparate as pharmacology -- the science of making and testing drugs -- and conservation biology.

"You wouldn't be able to guess what model organisms" are best suited for testing drugs or studying diseases for human application without a good phylogeny, said Antonis Rokas, an author of the paper and a postdoctoral fellow at UW.

Conservation biologists would not know the biological boundaries of threatened and endangered species without phylogenies, Rokas said.

Knowing the relatives

"The overall goal is that we want to know who is related to whom," said Sean Carroll, a UW professor of genetics and senior author on the paper. "The challenge has been to decipher the true tree from those that have changed as data have been added and re-analyzed over time."

The problem with just using one or two genes is that the genes chosen might not accurately reflect a true evolutionary history, Rokas said.

For instance, if scientists were looking at three species of plants -- two that live in very cold climates, and a third from the tropics -- they might arbitrarily choose to focus on a gene that had something to do with frostbite resistance.

In this case, the tree probably would show that the two cold plants were more closely related, when in fact, they might not be.

"Natural selection can mask the real history," Rokas said.

It was this concern, among others, that led Carroll's lab to see how many genes would be needed to make an accurate phylogeny.

"We were discovering, in our own research, that it was really difficult to get an answer" about the relationships between a specific group of organisms, called choanoflagellates, and other multicellular animals, Rokas said.

"Different labs were using different genes. And as a result, they were giving us conflicting answers," he said.

Fed up with the uncertainty and confusion, three post-doctoral fellows in Carroll's lab -- Rokas, Barry Williams and Nicole King -- started talking. They decided to do something about it.

Williams, who had recently returned from a conference in which the complete genomes of eight different -- but closely related -- yeast species were presented, suggested they conduct their own analysis using this information.

A solid tree

They built a tree using 106 genes and came up with a tree that couldn't be toppled. Statistical analyses showed 100% confidence behind their tree -- and no matter what they tried to do to shake it, it wouldn't fall down.

With further analysis, they discovered they could use between eight and 20 genes and still show the same tree.

That's still a lot more than most scientists now use to build molecular phylogenies, said Givnish, the UW botanist.

"It's surprising how many genes they needed to use to obtain a high confidence," he said, considering that yeast are relatively simple when compared with organisms such as mammals and vascular plants.

Their data indicate that more genes would be needed to examine relationships in more complex organisms, he said.

"I just came out with a study using seven genes" to determine the relationships within a group of plants called monocots, a more complex organism than yeast, he said. And a colleague of his used 10.

"We really thought that would be much more than enough," he said.

The reason people have relied on so few genes to determine evolutionary relationships, Rokas said, is that the process is costly and time intensive.

The drive to sequence and publish the genomes of different species is making this task cheaper and faster, particularly if the genomes are accessible to the public.

"This (paper) would have been impossible without public access to that information," Rokas said. "It was crucial that this information be freely available."

Copyright 2003 Journal Sentinel Inc. Note: This notice does not apply to those news items already copyrighted and received through wire services or other media
Provided by ProQuest Information and Learning Company. All rights Reserved.