Clues to which diabetics are at risk for the most severe effects of the disease may come from looking for increased activity in networks of genes, rather than individual genes — like looking for a traffic jam affecting an entire town, rather than trying to pinpoint a single car as the cause of the problem.
Professor Simon Kasif (BME), graduate student Manway Liu (BME) and colleagues sifted through a trove of data from Joslin Diabetes Center and performed DNA microarray studies of their own to identify clusters of genes, related by function, that change consistently across many animal models of the disease. The research article, “Network-based Analysis of Affected Biological Processes in Type 2 Diabetes Models,” is published in PLoS Genetics.
Kasif noted that the many resources concentrated in Boston, such as Joslin, and the city’s vibrant community of scientific researchers made this study possible. The BU engineers worked jointly with Isaac Kohane of Children’s Hospital on this project, allowing the group to tackle a large, complex research question.
Type 2 diabetes mellitus has reached epidemic levels in the US, said Kasif, and it affects approximately 170 million people world wide. Insulin resistance and abnormal glucose regulation characterize the disease, which is closely linked with obesity.
“Although diabetes may appear to be a simple phenomenon — with high glucose levels indicating diabetes — it is very complex, and this is an inaccurate measure. It does not predict very well whether the person, down the road, will start experiencing the major and more traumatic consequences such as heart disease, neuropathy [nerve damage], and retinopathy [which causes blindness],” said Kasif. “The issue is, can we define a molecular signature that indicates, ‘You’ve got to watch out for this particular patient’?”
Identifying patterns of genetic activity that change consistently and markedly in diabetes might help identify subgroups of diabetes patients most at risk for the most severe effects of the disease.
The traditional biological approach of looking for one gene linked to a specific function did not yield any meaningful patterns when the researchers examined 67 animal models of diabetes.
They then zoomed out to search at the level of groups of genes that had much higher or much lower activity compared to a normal state. Using this approach, “the results basically popped out,” said Kasif.
He and Liu liken looking for the gene networks to looking for traffic jams in Boston suburbs. Initially, when they sought to find changes in specific genes, it was like a futile search for consistent changes in traffic at a specific intersection in Cambridge, for example.
But, when they stepped back to look at the entire town, regardless of what happens at a specific intersection, the researchers saw that the big-picture result was a town-wide rush-hour traffic jam.
“We take more of this network point of view to see whether Cambridge as a whole is showing a different level of activity as opposed to looking at specific junctions,” said Liu.
Across the 67 models of diabetes, the researchers saw significantly increased activity in two specific gene networks: a group of insulin-signaling genes increased its activity in 45 of the models, and a group of nuclear receptor genes increased activity in 31 of the diabetes models.
“More and more people are thinking about these network views of disease. This work emphasizes the possibility that using standard methods, something goes undetected. When you bring in this network view, suddenly the degree of unusual activity pops up,” said Kasif.
In addition to potentially helping identify diabetics most at risk for the disease’s most severe complications, Kasif’s research may open new research avenues in drug development.