New research has led to the discovery of an important way that genomic variation can affect metabolic health and response to one class of medication for type 2 diabetes; these findings could lead to development of personalized treatments for the disease. Our genes contain the encoded recipes for making thousands of different proteins. When genetic differences alter these recipes in ways that inactivate or change the activities of the proteins they encode, there can be significant consequences for human health. But a surprisingly large portion of the genome is made up of stretches of DNA that do not code for proteins, and it has been something of a mystery as to why many genetic differences that are known to influence risk for various diseases have been found in these non-coding regions. One part of the explanation likely stems from the fact that proteins are not produced at the same rate throughout the body: their respective genes are “expressed” (activated so that the protein they encode can be produced) to differing extents in different tissues, at various stages of development and disease, and in response to nutritional and environmental cues. This pattern of gene utilization is enforced by expression-regulating proteins that bind to DNA sequences typically located near to (but not usually within) the genes they regulate. Thus, changes in non-coding DNA could affect health by altering the binding sites of these regulatory proteins, thereby influencing the expression of nearby genes.
In the new study, researchers investigated binding sites (in both mice and humans) for one such regulatory protein, called PPARγ, which plays a key role in controlling metabolic gene expression. They began by comparing male mice from two genetically different strains, one of which was more susceptible to type 2 diabetes and other metabolic diseases than the other. PPARγ can bind approximately 35,000 sites in the mouse genome, and the investigators found that in male mice genetic differences between the two strains affected PPARγ’s ability to bind about 2,000 of these sites. In some cases, variations in the DNA sequence within PPARγ binding sites (or within the binding sites of other regulatory proteins PPARγ sometimes works with) had large effects on PPARγ binding. The researchers were able to document that, in general, stronger PPARγ binding correlated with stronger expression of nearby genes, indicating that sequence variation at these sites had a signifi ant impact on gene expression in the two strains of mice. These differences in gene expression were accentuated when the mice were treated with rosiglitazone, a medication for type 2 diabetes that works by activating PPARγ. What makes this particular finding so significant is that the response to rosiglitazone is quite variable: it dramatically lowers blood glucose (sugar) in most (but not all) people with type 2 diabetes, and the drug is linked to significant side effects in some who take it. These findings suggest that the positive and negative variations in rosiglitazone response may be linked to differences in PPARγ binding sites near certain key genes.
To test whether people have significant, physiologically relevant differences in the binding sites of PPARγ and the proteins it works with, the investigators compared samples from five obese female volunteers (one Hispanic, one African American, and three White). As in mice, numerous differences were found that affect how well PPARγ binds various regulatory sites. The researchers looked at expression of genes near these sites using data from a previous study of samples from a group of 1,381 Finnish men with diabetes or at risk for the disease. They found that stronger binding of PPARγ or its partners correlated well with higher expression of nearby genes, suggesting that natural differences in PPARγ binding sites correlate with markedly different gene expression patterns in people as well as mice. In fact, the analysis uncovered a previously undescribed PPARγ binding site difference that appears to have a significant impact on levels of good cholesterol and other metabolic factors and thus, presumably, on human health. Further analyses of variation in human PPARγ binding sites may lead to identification of genetic variations associated with beneficial and/or harmful responses to rosiglitazone, and may one day facilitate personalized treatment for type 2 diabetes by identifying those most likely to benefit from and least likely to be harmed by the drug. Indeed, these findings chart a course toward better understanding the effects of any drug that acts on proteins that regulate gene expression.