Parts of human DNA that do not contain genes but instead turn them on and off may be just as vulnerable to cancer-causing mutations as protein-producing genes are, a new study finds.
Using computer programs to comb through the DNA of 88 cancer patients, researchers identified 98 mutations in gene-regulating parts of the genome that may be causing the patients’ breast, prostate or brain tumors, the team reports in the Oct. 4 Science.
The findings may help researchers better understand which genetic alterations lead to disease and which are harmless. “It helps to clarify a confusing question in human variation: What variants are important?” says Douglas Levine, a surgeon at Memorial Sloan-Kettering Cancer Center in New York City who was not involved with the work.
Finding one or a handful of variants that lead to disease is a daunting task because it requires sorting through more than 3 billion bases — the information-carrying chemicals that make up DNA — in each person. As a result, many scientists have narrowed their search to the 1 to 2 percent of the genome that makes proteins.
But many genetic variants implicated in common conditions such as diabetes and heart disease fall in the no-protein land between genes. Rare disease-causing mutations, such as those that spark cancer, may also fall in that vast, mysterious territory known as noncoding DNA.
“When it comes to cancer, those regions have been neglected so far,” says Jan Korbel, a geneticist at the European Molecular Biology Laboratory in Heidelberg, Germany. Cancer researchers didn’t have enough data to allow a search, says Korbel, who was not involved in the research. “This study shows how you can find these candidates.”
To narrow down which noncoding parts probably contain important variants, Ekta Khurana, a computational biologist at Yale University, and her colleagues examined DNA from 1,092 volunteers in the 1000 Genomes Project, an effort to uncover human genetic variation. The team also used information from the Encyclopedia of DNA Elements, known as ENCODE. That project maps stretches of noncoding DNA that are important for switching genes on and off. Some switches control when one or a small number of genes turn on; others govern many genes.
To find the switches most likely related to disease, Khurana and colleagues reasoned that if a stretch of noncoding DNA were particularly important for health, natural selection would have weeded out variants. So the team looked for regions that contain fewer mutations than would be expected by chance. A mutation in one of those sensitive places is about 40 times as likely to cause disease as is a mutation elsewhere in the genome, the researchers found.
Even fewer mutations showed up in a small subset of those sensitive regions, indicating that those switches may be “ultrasensitive” to change. A mutation in one of these glass-jawed pieces of DNA is about 400 times as likely to cause disease as is one in an average stretch of DNA.
The team used these data to create a computer program that identifies mutations in noncoding parts of the genome and scores how detrimental changes are likely to be. The researchers then used the program to analyze the cancer patients’ DNA. Of 98 mutations the researchers identified, 90 damage hub controllers of gene syndicates and 68 fall in sensitive noncoding areas.
One day doctors might use similar tools to find variants that contribute to their patients’ diseases. “It’s a long way to get to possible treatments from here,” Khurana says, “but it’s important to understand the biological mechanisms that are driving cancer.”