December 14, 2018 - 12:11 AMT
Genome scans could reveal roots of schizophrenia, autism

More than 2000 human brains stored in tissue banks are giving up their genetic secrets. Genome scans have already revealed hundreds of locations where DNA tends to differ between people with and without a particular psychiatric disease. But those studies don't pin down specific culprit genes or what they do in the brain. "There was kind of a missing link," says Daniel Geschwind, a neurogeneticist at the University of California (UC), Los Angeles. He and others in the 3-year-old PsychENCODE Consortium, fueled by roughly $50 million from the U.S. National Institutes of Health (NIH) in Bethesda, Maryland, have tried to bridge that gap by tracking which genes are expressed, and where.

The consortium focuses on regulatory regions, which control the expression of protein-coding genes, and which previous studies implicated as drivers of psychiatric disease risk. PsychENCODE collaborators have cataloged differences in the activity of these regulatory regions in different parts of the brain, at different stages of brain development, and in brains affected by different disorders—chiefly schizophrenia, autism, and bipolar.

The result, outlined this week in a series of papers in Science and its sister journals Science Advances and Science Translational Medicine, is the most complete picture yet of how regulatory regions influence the brain. In one of the new papers, for example, researchers describe DNA sites where a variation in a sequence changes the expression of a protein-coding gene elsewhere. Before PsychENCODE, that list consisted of fewer than 5000 locations, Geschwind says, but the consortium's work has brought the total to roughly 16,000.

"These data allow us to do things we've been wanting to do for a while," says Gerome Breen, a psychiatric geneticist at King's College London who was not in the consortium but plans to use its publicly available data set. Not all researchers are optimistic that the new data set will directly lead to new drugs for illnesses. But many expect it to reveal clues to how complex diseases develop.

The collaborators analyzed their brain samples with RNA sequencing to find out which genes were transcribed. They also did various epigenetic analyses, such as measuring how DNA's folded structure brings regulatory regions into contact with distant protein-coding regions.

The immense data set allows researchers to identify genome "modules"—groups of genes that tend to be expressed together and have common functions. Unique patterns of gene expression in a module might reveal a nuanced genetic feature of a disease. For example, previous studies have shown the expression of genes involved in neural signaling tends to be unusually low in autism, and to a lesser extent, in bipolar disorder and schizophrenia. But PsychENCODE data enabled a finer-grained analysis. They revealed modules including one containing genes that control how cells package and release their chemical messengers into synapses. That set of genes, it turns out, is especially active in schizophrenia and bipolar disorder, but not in autism. Such details might point to brain processes that could be targets for therapies.

The new data set can also reveal windows of brain development when disease-associated genes seem to have the most influence, says Geetha Senthil, the NIH program director who has coordinated and overseen PsychENCODE. Those windows, in turn, might be the times when intervention would be most valuable. Doctors can already observe, based on a patient's symptoms, when a disease seems to take hold, but, she says, "having a biological clue would be thrilling."

The project's namesake, ENCODE (Encyclopedia of DNA Elements), was a broader quest to map noncoding regions of the human genome. Its initial results, unveiled in 2012, stirred controversy. Scientists disputed the team's claim that most of the genome was functional and questioned whether the project's insights would be worth NIH's $185 million investment.

Dan Graur, an evolutionary geneticist at the University of Houston in Texas and one of the most outspoken critics of ENCODE, also finds fault with some of the initial PsychENCODE results. The project targets psychiatric disorders that are themselves poorly defined, he says. "If you take something vague and correlate it with millions of genetic and epigenetic variations, you are bound to get statistical significance that will have little biological significance."

Neurogeneticist Kevin Mitchell of Trinity College Dublin echoes some of Graur's concerns. "I'm not fully convinced that we know more today than we did yesterday," he says. He doubts that a profile of gene expression can define disorders as heterogeneous as schizophrenia or autism—or give new insights into how to treat them. "It's a huge amount of work, very well intended and very well done," he says, "but there are some limits to what you can do with genomics."

But many researchers defend the project's value. "I'm sure there are researchers out there who will look at these first papers and say, … ‘Where is our paradigm-shifting finding?’" says Alexander Nord, a neurogeneticist at UC Davis who was not in the consortium. "That's a bit of a straw man, expecting us to find that in one set of analyses." The data set will grow richer as researchers work to interpret it, he says. "It's not going to go out of style."