By developing new statistical methods to analyze incomplete DNA sequences from thirteen strategically selected plant species, the researchers uncovered a previously hidden "paleopolyploidy" event, an ancient whole-genome duplication that preceded the appearance of the ancestral flowering plant.

Claude dePamphilis, associate professor of biology at Penn State, is the principal investigator of the Floral Genome Project and the senior author of the paper. "We found a concentration of duplicated genes that suggests a whole-genome duplication event in the earliest flowering plants," he says. "A polyploidy event early in the history of flowering plants could explain their sudden evolution." The results appear in a June issue of Genome Research.

One unexpected observation from the study is the relatively slow accumulation of mutations in primitive flowering plants like the yellow water lily (Nuphar). "We can view these basal angiosperms like the Hubble Space Telescope, which helps us get a deeper look into the early history of the universe--these plants allow us to take a deeper look into genomic history."

Darwin noticed that flowering plants appear suddenly in the fossil record and then radiate quite rapidly. Technically known as angiosperms, "flowering plants exhibit a number of evolutionary innovations that appeared rapidly, including novel structures like carpels and primitive petals and sepals, the sine qua non of flowering plants," dePamphilis says. Angiosperms also boast plenty of unique biochemistry. "They're a rich source of medicinal compounds. Even the kind of wood they make is special."

The new results support the idea that "whole-genome duplications are rare in vertebrates, but common in plants," according to dePamphilis. Independent whole-genome duplications occurred relatively recently in soy, potato, and tobacco, and longer ago in maize. But the thinking was that human breeders might be artificially selecting for duplication events in crop species by "selecting desirable traits like rapid growth, high yield, and even large stature," says dePamphilis.

Detection of still more ancient whole-genome duplications previously has relied on direct observation of genomes that have been completely sequenced. Arabidopsis, for example, the first plant whose genome was entirely sequenced, was shown to harbor many small blocks of related genes in identical order along the chromosome--the telltale remnants of a whole-genome duplication. In Arabidopsis and other cases, however, the signs of whole-genome duplication are few and far between: most of the duplicate genes are quickly lost, leaving few obvious traces.

Whole-genome duplications have attracted attention as a possible mechanism to drive sudden bursts of evolution, like the one that so vexed Darwin over a century ago. While the vast majority of duplicate genes quickly accumulate mutations and are deleted from the genome, a few mutations will be selected for evolutionarily advantageous function. Rather than gradually collecting genetic novelty by single-gene duplications, simultaneously having a full genome's worth of raw material to elaborate new genetic function could drive sudden evolution. But because of the rapid, massive gene loss after a whole-genome duplication, these events are notoriously difficult to detect after millions of years. So dePamphilis and colleagues relied on a statistical filter to hunt for ancient duplications.

In order to show that such an event occurred early in angiosperm history, dePamphilis and his colleagues had to compare the genomes of "basal" angiosperms--those whose ancestors diverged from the rest early in the lineage--with more recently derived angiosperm species, as well as with plant species outside the angiosperms. They took the opportunity to examine the genomes of a number of plant species for which only a partial DNA sequence has been determined. The team uncovered evidence of whole-genome duplications by confining their study to duplicated genes. The challenge was to distinguish isolated duplications from whole-genome duplications after a period of time in which direct evidence for a genome-wide duplication may have disappeared.

Any duplicated gene gives rise to two "paralogs," which are each subject to random, independent mutation. Most mutations in a gene's DNA sequence cause a change in the corresponding gene product. But "synonymous" mutations in DNA do not lead to a change in gene product, so natural selection has little effect on them and they remain in the genome. Because such mutations accumulate through time, the frequency of synonymous mutations between paralogs "can be used as a proxy for the time following the duplication," explains dePamphilis. Even after obvious signs of a whole-genome duplication have been lost, statistical analysis can detect a group of paralogs with very similar frequencies of synonymous mutations, indicating that all of those paralogs arose simultaneously--the hallmark of a whole-genome duplication.

A paleopolyploidy event previously demonstrated by other investigators is associated with a burst of evolution in the economically important grass family. The new results from the Penn State paper confirm a previously-reported paleopolyploidy event in eudicots (a group that includes beans, tomatoes, sunflowers, roses, and apples) associated with their rapid divergence, and demonstrates the first evidence of a paleopolyploidy event associated with the ancient explosion of all angiosperms.

Mere guilt by association? "We can take it farther than just correlation," asserts dePamphilis. The MADS-box genes, a family of transcription factors that are required for flower development, are known to have undergone an expansion through duplication that was critical to the evolution of angiosperm flowers. A whole-genome duplication explains the sudden emergence of novel traits better than a series of single-gene duplications, explains dePamphilis. "Some of the MADS-box genes and many other genes important in plant development were produced by paleopolyploidy."

In addition to de Pamphilis, coauthors at Penn State include the lead author Liyiung Cui and Kerr Wall, who developed the software for the statistical analysis along with Bruce Lindsay, Distinguished Professor and head of the Department of Statistics. Additional coauthors who contributed to the paper are at the University of Oslo, the Benaroya Research Institute in Seattle, Cornell University, and the University of Florida in Gainesville. The research was sponsored by the National Science Foundation.

Related Links
Floral Genome Project at Penn State University