Volume 4, Issue 27 April 2007
Self-Tuning Genes
Rachel Brem studies yeast genes in hopes of finding ways to easily locate and analyze potentially harmful gene variations in humans. Image credit: Courtesy Rachel Brem
When scientists first began studying the human genome, they thought its 3 billion base pairs of DNA might harbor as many as 100,000 genes. Less than ten years later, in 2004, they revised that number sharply downward to between 20,000 and 25,000—just a third more than the lowly roundworm. How can a creature as complex as Homo sapiens have so few instructions? One answer is that we make the most of the DNA we have by carefully regulating how those genes are used.
For example, in some cases, proteins are responsible for regulating their own production. This can happen directly, as when a protein stops its gene from being read; or indirectly, as when a protein interacts with other cell factors to amplify its gene's expression. Researchers don't yet know how many genes self-regulate. If the mechanism is widespread, it could be useful in tracking down mutations responsible for cardiovascular disease, diabetes, schizophrenia, and other disorders controlled by multiple genes.
Rachel Brem, QB3 faculty affiliate aand professor of genetics and development who joined the UC Berkeley faculty just last spring, studies such regulatory networks in the context of entire genomes. She hopes to find characteristics that will make it easier to identify self-regulatory genes, and home in on the mutations most likely to cause disease.
Brem uses yeast as her model organism. This member of the fungus family offers many advantages in the lab, Brem says. "We can address questions like how the genome is organized, and how genes work together, in a lot more depth because so much is already known about yeast and the experiments are easy to do."
Brem's first order of business, after setting up her lab, will be to assay each of the roughly 6,200 genes in yeast to see if their expression is controlled by a feedback mechanism. She'll insert a jellyfish gene known as green fluorescent protein (GFP) into each stretch of yeast DNA that codes for a protein. Then she'll grow these yeast on Petri dishes. In the dark, these modified colonies will glow like miniature stars. She'll compare their glow against colonies where she's deleted the yeast portion of the marked gene. If the deleted strain gets brighter or dimmer, Brem can conclude that the gene is involved in a regulatory feedback network.
Yeast colonies genetically engineered to produce a protein that glows green. Rachel Brem is analyzing how brightly these colonies grow in order to find genes that regulate their own expression. Gene feedback networks are implicated in complex diseases such as diabetes and heart disease. Image credit: Courtesy Edward Marcotte's laboratory, CSSB, University of Texas at Austin
"We're looking for patterns," Brem says. "Are there certain functions or classes of genes associated with regulatory feedback? How frequent are they in the genome and why?" Because the number of genes Brem must process are so large, she'll be fishing for commonalities among regulatory genes using computer algorithms of her own devising.
Researchers such as UC Berkeley's Adam Arkin have found that regulatory feedback is associated with chance fluctuations in mRNA or protein levels—a phenomenon called expression noise. "Even though they're all genetically identical, and grown under the same conditions, yeast clones don't express certain proteins at exactly the same level," Brem says. "Some genes are noisier than others. That makes people think the cell is actively tuning the distribution around an expression level set by the regulatory network." Noise may ensure that a few individuals can handle abrupt changes in their environment. In other words, if a colony is suddenly assaulted by toxic chemicals or high heat, a few individuals will already have expression levels suited to those conditions.
As for Brem's ultimate goal? "We'd like to have a computer program smart enough to make a prediction about which would be disease-linked without doing any other experiments," Brem says. "But we haven't seen enough of those mutations so far, so we don't know how to recognize them yet." With her perseverance and prowess at computational analysis, that lack of data should be replaced with new insight into the organizing principles of our DNA.