Better risk assessments through molecular biology

As director of the U.S. Environmental Protection Agency’s NexGen Program, toxicologist Ila Cote leads a collaboration that brings together data, methods, skills and brains from diverse fields to better understand how chemicals interact with living things and the environment. In doing so, scientists hope to answer questions about potential risks from chemical exposure more quickly and cheaply. Cote, who recently hosted a conference about NexGen in Washington, D.C., discussed the effort with Science News chemistry writer Rachel Ehrenberg.

One could take a sample of spit … to understand what people had been exposed to and how that changed their risk for some specific diseases. Robert E. Walline

What is risk assessment?
Risk assessment is a process of evaluating information to determine how likely you think some event is; in the case of the EPA, the likelihood of public health or environmental damage. One example is air pollution regulations: We … would take all the information that’s available on the health effects and environmental effects of air pollution, organize it, synthesize it and interpret it and then provide that information to decision makers at EPA.

What does molecular biology bring to risk assessment?
It really got a big leap forward with the Human Genome Project and the invention of robots that can do lab work. Now you can generate data both faster and cheaper, and it’s a different kind of data. Molecular biology is really the study of the machinery of cells and how they function, particularly in regard to these important molecules in the cell like DNA, genes and proteins. So with those new ways of looking at the function and the machinery of the cell, we’re gaining new insights.

What are some of these insights?
Understanding how chemicals cause disease, how individuals might differ from one to the other in terms of their response to the same exposure. To go back to [air pollution]: We know that about 30 percent of the population is much more sensitive to ozone. For a long time we didn’t know why. Now we know that part of that is some specific genes that make you more susceptible. That’s probably not the whole story, but that’s one piece. There are differences in metabolism — big differences in the population in how you metabolize things. So any drug where metabolism is involved in an important way, we know that individual responses can differ. We all know people who smoke. And not all of those people will get lung cancer or a smoking-related disease, and it’s been a perplexing question for many years as to why that is. These new methods are providing some insight.

What methods are most promising?
It’s very hard to figure out how to use data in the abstract, so we’re developing some case studies in collaboration with other agencies. We’re taking chemicals that we know a lot about — we know what the public health risks are for these chemicals as well as we know it for anything … — and we want to take the matching molecular biology data and see if we can reverse-engineer to the right answer…. For ozone, we’re looking at lung injury and how that happens through an inflammatory process. So we know, at a chemical level, most of the steps that lead to inflammation. And what we want to do is look at the genes that turn on and off that are associated with inflammation. By looking at the patterns of those genes can you predict the health effects and the magnitude of the response for that individual?

One of the prototypes we’re doing is with benzene, in collaboration with a University of California, Berkeley lab that’s done epidemiology studies of benzene-exposed workers in China. So they can measure how the genes turn on in those people and accurately predict how much benzene they’ve been exposed to and how likely they are to get cancer. So potentially in the future, if we were concerned about a school or a community, one could take a sample of spit or a cheek scraping … and be able to understand what people had been exposed to and how that changed their risk for some specific disease. The technology is just now coming online, and we don’t have enough examples to be able to reliably do it now, but I don’t think it’s that far off in the future.

What are the challenges to implementing a molecular biology approach to toxicology?

In some ways this new data is not very intuitively obvious. In the good old days, you exposed a bunch of rats and half died and half didn’t, and you could say, “That doesn’t look good.” Now what you get is these little bright lights of readouts of a bunch of genes that get up-regulated or down-regulated. And you show that to somebody and they go “What does that mean?” … It requires people who understand biology and computers to understand all these data.