Genetic information that determines hair color or whether an individual might develop a particular cancer is passed from one generation to the next through DNA. Genes encoded in the DNA contain information, but a process called methylation is one factor that often controls how that information is expressed.
A group of University of Washington researchers has devised a method that combines DNA sampling and mathematical modeling to find out how accurately methylation patterns are copied during DNA replication. That could pave the way for understanding the role methylation plays in normal gene expression and how it factors in the development of human disease.
In methylation, a methyl group (made up of a carbon atom and three hydrogen atoms) is attached to a specific gene sequence in one part of DNA. The density of methyl saturation determines how the gene is expressed. The densest saturation turns the gene off so that it is not expressed at all, and less-dense saturation allows the gene to be expressed at different levels.
The result can be obvious, for instance, in a calico cat and its multicolored coat, said Diane Genereux, a visiting UW biology graduate student and lead author of a paper describing the new measuring technique in the April 19 edition of the Proceedings of the National Academy of Sciences.
"In a calico cat, different genes that express coat color are on and off in different parts of the cat's coat, so you get patches of different-colored fur," Genereux said.
Methylation typically passes from genes in a DNA strand to the same genes in a daughter strand created during DNA replication. The new technique allows researchers to examine how faithfully this "maintenance" methylation is carried out across generations and how consistently it occurs on the same gene sequence, said Brooks Miner, a UW research scientist in biology and a co-author of the paper.
But DNA molecules also can undergo what is called "de novo," or new, methylation in which a methyl group shows up on a DNA strand at a place where it did not appear before. That could change how that particular gene sequence is expressed.
Understanding methylation rates is important because the rate of genetic mutation is very low, a tiny fraction of 1 percent, while the rate of methylation changes that alter or suppress gene expression is substantially higher.
In the past, researchers could look at only one strand of DNA at a time and so could not conduct a side-by-side comparison of where methylation was occurring. An earlier paper from the same research group, led by Charles Laird, a UW biology professor, introduced a molecular method to look at both DNA strands together and observe methylation differences between them.
"When we look at both DNA strands, we know one strand has to be the parent and one has to be the daughter, but we don't know which is which," Miner said. "The new mathematical model allows us to infer the rates of both maintenance and de novo methylation without directly identifying parent and daughter DNA strands."
Ultimately, such knowledge could lead to better understanding of, perhaps even treatment for, some cancers or genetic conditions such as one called fragile X syndrome, the most common cause of genetic mental impairments, from slight learning disabilities to severe cognitive disorders. Fragile X is caused by abnormal methylation of a gene called FMR1, Miner said, and other conditions have similar causes.
"Methylation is a normal biological process that, in the case of fragile X, is happening at the wrong place at the wrong time," he said. "It's a basic process, but it's not fully understood."
The new method lets the researchers see how consistently methylation occurs in different places on a DNA strand. Applying mathematical models to a DNA sequence allows them to measure methylation rates for different areas of the genome, Genereux said.
"As with any inference, we know we're not going to get the precise rates," she said. "To get an exact answer, we'd have to look at all the cells in an individual. Our method provides a way to get useful approximations from a small DNA sample."