Email Icon Facebook icon Twitter Icon GooglePlus Icon Contact

User Top Menu

Schizophrenia Genetics 2015—Part 1, Renaissance

6 Aug 2015

In SRF's five-part 2015 schizophrenia genetics update, reporter Michele Solis surveys leaders in the field about milestones, challenges, and current research.

See Part 2, From Discovery to Understanding; Part 3, Rare Allure; Part 4, Rethinking Diagnoses; and Part 5, Plan of Action.

Download a PDF of the entire series.

Editor's note: In Schizophrenia Research Forum's five-part 2015 schizophrenia genetics update, reporter Michele Solis surveys leaders in the field about milestones, challenges, and current research. It is an interesting exercise to compare it to Patricia McCaffrey's original 2010 series. A few things stand out: The first series began with a look back at linkage studies that had identified regions of the genome that likely harbored risk genes, setting the stage for the subsequent boom of candidate gene studies. The SchizophreniaGene database was considered by many the state of the art: the meta-analyses of more than 1,500 candidate gene studies had resulted in the publication of a Nature Genetics article (Allen et al., 2008).

In 2015, linkage is ancient history, and candidate genes are now in the bust portion of their gene-mining cycle, supplanted mainly by genomewide association studies (GWAS). Two developments that were heralded in the 2010 series—rare variants, especially copy number variants, and whole-genome sequencing—have yet to exert a large impact on biological research. Indeed, the "special" rare genetic abnormality of mental illness research—the DISC1 disruption—has lost its ingenue status, though it remains a consistent performer in biological studies. Finally, a new actor has entered the stage: Patient-derived induced pluripotent stem cells, which were not mentioned even once in 2010, are seen as the best hope to make sense of risk variants for schizophrenia.

With all that said, the final articles of each series do not make significantly different points. Pending the identification of the source of the more than 100 GWAS signals, we still await clear targets for biologists and drug designers, we still wonder whether endophenotypes or clinical subgrouping will provide better clues, and we still hope there are some low-hanging fruit in pharmacogenomics. We wish the geneticists the best of luck in their daunting task and look forward to surveying a new landscape in 2020.—Hakon Heimer.

August 7, 2015. With a steady stream of papers in high-profile journals, front page headlines, and proliferating collaborations, schizophrenia genetics has arrived. It even has its own T-shirt. And a lucrative sign of the field's heightened status came last summer with the $650 million pledge to the Broad Institute by philanthropist Ted Stanley for research on psychiatric disorders. This definitive vote of confidence came at a time when drug companies had all but given up on psychiatric research.

Once a scientific backwater beset by false leads, the field now claims over 100 regions of the genome firmly associated with schizophrenia, a severe mental illness that has stymied researchers for over a century. The explosive transformation has been wrought by international collaboration and ever more economical tools to explore the genome in an unbiased way.

"Honestly, schizophrenia genetics is a completely different field now," said Patrick Sullivan of the University of North Carolina at Chapel Hill. "It has shifted to a highly critical, highly empirical, enormously data-driven enterprise, which has been pretty neat to see." Sullivan is a leader of the Psychiatric Genomics Consortium (PGC), which convinced researchers around the world to share their data to get genomewide association studies (GWAS) off the ground. The most recent GWAS, published in 2014, zoomed in on 108 regions of the genome linked to the disorder.

"I think there's been enormous progress in schizophrenia genetics; that's the bottom line," said Daniel Weinberger of the Lieber Institute for Brain Development in Baltimore, Maryland. Weinberger has been on the hunt for genes in schizophrenia for decades.

"Genes related to risk have been identified with a level of statistical evidence that is above the fray of controversy—and this kind of evidence is attracting a new generation of scientists to understand psychiatric illness," he said.

Despite the good news, researchers may still find themselves bewildered. In very short order, they've gone from having no firm leads to an embarrassment of riches, casting suspicion on hundreds of genes. Turning these insights into an understanding of schizophrenia's biology or potential therapeutic targets will be a long haul.

"Realistically, until we start to understand and interpret the clues that genetics is giving us, we're not actually accomplishing that much," said Mark Daly of the Broad Institute, Cambridge, Massachusetts, who chairs the PGC's analysis group.

The lack of biological understanding looms large. "I think it would be fair to say that we haven't made any quantum leaps in understanding causal pathways and mechanisms," says David Porteous of the University of Edinburgh, Scotland. "But perhaps we now have a better feel for just how big the task is that lies ahead."

This five-part series, an update to SRF's first schizophrenia genetics series published in 2010, covers the developments in the past five years and surveys researchers on the best ways forward. Read on to find answers to these questions:

  • How will the disease-influencing genes in the 100-plus loci be found?
  • Do we need more GWAS?
  • What is the status of the old candidate genes from the pre-GWAS era?
  • Are there any new copy number variants (CNVs) on the horizon?
  • When will definitive rare variants finally turn up in sequencing studies?
  • Will paying attention to schizophrenia's different features, including endophenotypes, help find genes?
  • How do genetic signals that confer risk for multiple disorders eventually contribute to a specific disease?

Growing GWAS

The complicated genetic story mirrors the complexities of the disorder itself. Appearing in late adolescence or early adulthood, schizophrenia is a mix of frightening symptoms consisting of psychosis—hearing voices, having fixed delusions usually of a paranoid nature, and experiencing disordered thinking—as well as negative symptoms, which comprise the paralyzing lack of motivation and emotional expression that cut a person off from the rest of the world. Cognitive deficits, such as impaired attention and memory, also plague most people with schizophrenia.

People differ in their symptom profiles. One person may have little psychosis but all the motivation-crushing negative symptoms, whereas another may have a sampling of all. To complicate matters, the psychotic symptoms may wax and wane. Outcomes differ, too: some return to full lives with work and satisfying relationships, usually with the help of antipsychotic drugs, whereas others struggle with addiction, end up homeless, or even commit suicide.

One might worry that this hodgepodge would resist genetic analysis. But epidemiological studies established that schizophrenia runs in families, with an estimated heritability of 65-80 percent, meaning that a majority of the liability for the disease in the population can be attributed to genetic factors.

But how best to identify the genes involved depended on people's hunches about where risk might lie. Human genomes teem with different kinds of variation. Some are single base changes commonly found in 5 percent or more of the population, akin to a misspelled word in the book that is our genome. Others are rare, consisting of single base changes, or insertions or deletions of a few bases; in the protein-coding part of the genome book, these might scramble a word beyond recognition. Another kind of rare variant, called a copy number variant (CNV), deletes or duplicates long stretches of DNA, similar to dropping or repeating a sentence or paragraph.

Common variants, because they are widespread in the population, confer only subtle effects on risk, nudging a person toward disease. Rare variants consist of genetic glitches that natural selection has not yet weeded out, and so they can potently escalate risk.

To detect common variants, researchers use microarrays to screen millions of single nucleotide polymorphisms (SNPs) commonly found in humans. These not only tally the sort of allele a person has at each SNP location, but can also detect rare CNVs encompassing these locations. Finding the rare variants, however, requires more expensive and time-consuming sequencing, which reads out DNA letter by letter in order.

Though most geneticists agree that both common and rare variants have a hand in schizophrenia, their relative importance is still debated. In one way, the discussion boils down to the general versus the particular: should we seek genetic clues that, though weak, are generally applicable to all people with schizophrenia or go after the powerful ones that apply to a rare few, betting these would lead to key biological processes relevant to many with the disorder?

"The common variants are where the money is," said Patricio O'Donnell, head of psychiatry at Pfizer in Cambridge, Massachusetts, which is also a member of the PGC. "We want to find a target that eventually might have an impact in a relatively significant proportion of people with schizophrenia."

By contrast, rare variants might produce more personalized insights.

"Ultimately, treatment is based on the individual person who is sitting in front of you," said Jack McClellan of the University of Washington in Seattle. McClellan and colleague Mary-Claire King, who have been searching for rare variants in schizophrenia, have proposed that most people with schizophrenia have a unique or "private" genetic cause.

Five years ago, the field was divided about the best course to follow. At the time, the returns from the common variant approach, which uses GWAS to see if certain SNPs are overrepresented in disease compared to controls, fell short of what many had hoped for (see SRF Genetics Series 2010). Worried that lumping heterogeneous cases together diluted genetic signals, some felt the common variant approach should be abandoned. Proponents, on the other hand, argued that larger samples were needed.

The PGC pressed on, increasing its sample size to 150,000 in the latest GWAS—five times the number in a combined analysis of the first studies in 2009 (see SRF related news report). While some samples came from people whose symptoms had been comprehensively documented, others had only a schizophrenia diagnosis. Still others didn't even have that, coming anonymously from clozapine clinics, which regularly take blood from people on the antipsychotic to check for deadly side effects.

This approach couldn't be more different from what came before it. In the past, a single researcher might have canvassed the countryside to carefully interview and assemble people with an artisanal level of detail. Compared to that, GWAS can seem like a trip to a big-box store.

"We have to face up to the fact that the PGC has largely been focused on large numbers. With the money, I could study more people more superficially or fewer people deeply," said Kenneth Kendler of Virginia Commonwealth University in Richmond, who has participated in both styles of inquiry. "The field still doesn't know which is better."

Parts list

So far, the big-box model has prevailed. The PGC's latest schizophrenia GWAS found 128 SNPs that occurred more frequently in people with schizophrenia than in controls, with a high enough level of statistical significance to sidestep many of the concerns about false positives that had dogged the field before.

The findings also clearly endorse the PGC's data-sharing experiment and its lumping approach.

"I must say that the team science approach has been really successful," said Thomas Lehner, director of the Office of Genomic Research Coordination at the National Institute of Mental Health (NIMH) in Bethesda, Maryland. "It's delightful to see how these competitive groups can actually put aside the competitiveness and work together on problems. I think that's the future."

"I'm encouraged that we can find real associations with risk of schizophrenia, which suggests that the category is a meaningful category for genetic investigation," said David Goldstein of Columbia University in New York City, who has been a critic of GWAS.

But, he adds, "We don't yet have any genetics that provide enough biological insight to provide pointers to new treatment opportunities. I think that's what we'd like to see."

Of the hundreds of genes that have fallen under suspicion thanks to the landmark GWAS, some will turn out to be innocent bystanders, while others will take a place in the parts list for schizophrenia risk.

Most of the loci are new but implicate processes already suspected of malfunctioning in schizophrenia, including glutamatergic signaling, dopamine signaling, calcium channels, and immune signaling (see SRF related news report).

Drug companies are taking note. "In my career it's the one paper that has truly resonated within the industry," said Nick Brandon, head of discovery at AstraZeneca's neuroscience program in Cambridge, Massachusetts, who has been working on psychiatric disease in industry for 14 years. "This dataset has clearly reinvigorated industry's interest in going after schizophrenia again."

Though the exact length of the parts list remains unclear, the fact that it comprises many entries fits with the idea, first suggested by Irv Gottesman and James Shields in the 1960s, that schizophrenia is a "polygenic" disorder. This means one person's schizophrenia stems from disruptions to multiple genes.

"I am very pleased with these signs of progress after waiting all my years in the field," said Gottesman of the University of Minnesota in Minneapolis. "But we have yet to make the bridge between those SNPs and the biology of the nervous system."

The challenge is that each SNP has a tiny—some say uniformative—effect, increasing risk ever so slightly. This means that none of the SNPs by themselves explain very much about an individual's liability to develop schizophrenia.

But the effects may not be too small to matter. To wit: one of the SNPs points to DRD2, the gene encoding the dopamine 2 receptor, which is the main target of antipsychotic drugs.

Combined, these small effects could pack a punch. A polygenic risk score (PRS) adds up all of the schizophrenia-associated SNPs carried by people to give a sense of their overall risk (ISC, 2009). In the new GWAS, the highest scores add up to an effect size about 10 times that of a single SNP, and more people with schizophrenia score high compared to controls. The score does not take into account specific genes, however, so one person with a high score could have a different combination of risk SNPs than another with the same score. This suggests that different people may carry different constellations of these risk factors.

Still, the 128 SNPs combined strain to explain schizophrenia's liability, accounting for only 3.4 percent (65-80 percent would explain all heritability). Analyses of patterns of all SNPs in schizophrenia and controls—regardless of their association status—suggest that common variants will eventually explain 30-50 percent of schizophrenia's heritability (see SRF related news report; SRF news report).

The unaccounted-for heritability leaves room for risk factors not probed by GWAS. Sullivan suggests a pie metaphor, with slices of as yet unknown size due to common variation, CNVs, protein-coding variation, environment, and interactions between genes and environment.

The PGC is interested in all of these slices. "Our view is that the genome is going to tell us what the answer is, and we simply don't care if it's a common variant or a rare variant," Sullivan said. "If it's important, we want to find it."

Wait, there's more

The PGC has a much larger GWAS already in the works, called PGC3, with 60,000 cases. Some of the PGC3 samples will be genotyped with the Psych Chip, a low-cost array specifically designed to detect common and some rare variants under suspicion for a variety of psychiatric disorders, including schizophrenia. Given the rate of genomewide-significant hits per sample so far, hundreds of more genomewide-significant hits are expected.

"There's nothing magic about it," Sullivan said. "In the end we have a carefully considered, mature, and relatively inexpensive way to actually learn more."

One expectation of the PGC is that getting as complete a catalog as possible will help delineate the operative biological processes impacted by these genes. But others worry that GWAS can become a numbers game, unmoored from biology.

"With each turn of the GWAS wheel, things that were previously highly significant hits don't necessarily survive into the next round," Porteous said, referring to five SNPs from previous GWAS that were not among the PGC's latest hits. Though this may reflect weeding out of false positives, SNPs with such small effect sizes may just shuffle in and out of statistical significance as sample sizes grow.

Others praise the PGC results to date but don't see the point of much more GWAS. "The published GWAS are a landmark. What now?" said Francis McMahon of the NIMH and an SRF advisor. "I'd be more excited to see someone making sense of one of these genes than I would about another paper with another hundred genes."

"We will keep looking for more genes, because it's like going to the moon—because you can go to the moon, you have to go the moon," Weinberger said. "Frankly, we have plenty of genes. The question now becomes, What do we do with them?"

The difference between finding a genomewide-significant SNP and finding a bona fide risk gene is a chasm that will take different approaches to cross. But, as will be discussed in the second installment of this story, "From discovery to understanding," bridging this gap is an important step for turning GWAS findings into actionable biological insights about schizophrenia.—Michele Solis.

See Part 2, From Discovery to Understanding.