Huge DNA Databases Reveal the Recent Evolution of Humans

The real clue is what scientists did not find.

When we talk about human evolution, we usually talk about how we evolved into humans: how we lost body hair, gained brain mass, started to walk on two feet—in short, things that happened millions of years ago.

But evolution did not stop when the first modern humans emerged. A new study of two massive genetic databases—one in the United Kingdom and one in California—suggests genetic mutations that shorten lifespans have been weeded out since, and are possibly still in the process of being weeded out today.

The idea for the study, says Molly Przeworski, a geneticist at Columbia, came when the health-care company Kaiser Permanente decided to make available a massive genetic database of certain patients in California. Over lunch at the time, she and the study’s coauthor, Joseph Pickrell, did a quick back-of-the-envelope calculation: With a 100,000-person database, they could see evolution acting on a mutation found in more than 10 percent of the population.

So they dived headlong into the data from Kaiser Permanente (57,696 people) and a second database (117,648 people) compiled by the U.K. Biobank, which collects tissue samples and medical data. Two genes immediately popped up: APOE, a mutation in which is associated with late-onset Alzheimer’s disease, and CHRNA3, a gene for a nicotine receptor where a variant is associated with more smoking. In both these cases, they found genetic variants associated with dying younger. What was truly intriguing, though, is that the team only found two such genes.

“That was the thing I was most interested in,” says Gil McVean, a statistical geneticist at Oxford, who was not involved in the research. “Why are there so few mutations that have a big effect on longevity?” The most obvious effects from APOE and CHRNA3 emerge late in life, after most people have already had children. In that case, you would expect there to be no selection pressure on mutations affecting longevity. And without selection pressure, a lot of such mutations would have accumulated by chance somewhere in the 6 billion letters of human DNA. But the fact there are only two such variants suggest others have indeed been actively weeded out.

The method from Przeworski and her colleagues is indirect but clever: They’re using genetic signatures (or lack thereof) from people in the present to infer evolution that happened in the past.

But why would mutations that affect longevity get selected out? Scientists can only speculate. Perhaps these genes not only have an obvious effect at old age but also subtler harmful effects in youth. Or, perhaps, it has to do with inclusive fitness, the idea that kin who share genes have an interest in helping each other. Mutations that kill off grandparents before they can help rear their grandchildren might confer a disadvantage in this way.

But then another question: Why do the bad APOE and CHRNA3 mutations still exist? A compelling explanation is that the environment has changed. Humans are no longer living in the savanna. Perhaps the APOE mutation only causes Alzheimer’s in the context of our contemporary diet and environment. CHRNA3 and its effect on smoking are an illustrative example: Smoking is a relatively new phenomenon. “The cost would not necessarily be present until very recently,” says Przeworski, so evolution has not yet had time to weed it out.

Przeworski and her colleagues also looked at 42 more traits—such as body-mass index, timing of puberty, high cholesterol—that are determined by not one but many genetic variants. They wanted to know how these traits are related to lifespan. And they found, for instance, that genetic variants that lead to later onset of puberty and age of having first child are associated with longevity. In other words, less fertility means more longevity. That suggests any trait may have trade-offs, and this big genetic database is one way to study these trade-offs. “You can look at the entire spectrum of consequence of variants. That is a pretty big shift in how we can do evolutionary genetics,” says McVean.

Emmanuel Milot, a population geneticist at the Université du Québec à Trois-Rivières who has studied recent human evolution on an isolated Canadian island, says the new approach is a welcome addition to the toolbox. By studying church records, Milot has found evidence of selection for younger age of first birth over 140 years, but he couldn’t pinpoint the genetic variants being passed down. “When you put these two different types of approach together, then you can understand more,” he says.

As big new genetic data sets continue to grow, it’s becoming increasingly easy to study subtle effects of evolution. Przeworski says they are now analyzing the expanded U.K. Biobank, which has gone from 150,000 to 500,000 participants. With more people, they can look for the effects of even rarer mutations than the ones in this study. Mutations that appear in 10 percent of the population are still relatively common. Przeworski and Pickrell have already done another calculation, this time more rigorous than their initial back-of-the-envelope calculation: With 500,000 samples, the team could study mutations that appear in only 2 percent of population. And that will provide even more detail about recent human evolution.