January 22, 2014. Rare variants in the exome, the protein-coding part of the genome, suggest disruptions to synaptic communication in schizophrenia. Published in two studies in Nature on January 22, these results come from the largest yet exome sequencing efforts, conducted by an international collaboration of researchers.
The first, led by Michael O’Donovan and Michael Owen of Cardiff University, United Kingdom, scanned the exomes of 617 people with schizophrenia and both their parents to find the spontaneously occurring “de novo” mutations. The second study, led by Shaun Purcell of The Broad Institute of Harvard and MIT in Cambridge, Massachusetts, and Pamela Sklar at Mount Sinai School of Medicine in New York City, took a case-control approach, looking for rare variants found more frequently in 2,536 people with schizophrenia than in 2,543 controls. Though the studies fell short of definitively pegging individual genes, they both found evidence for disruptions in sets of genes encoding the synaptic machinery that conveys glutamate signals from one neuron to another.
“We found out that the rarest, most severe point mutations—the most likely to damage a protein—contribute to schizophrenia risk,” Pamela Sklar told SRF. “But, as a group, they contribute more modestly than many thought.”
The two studies provide a sobering curative for anyone who still thought that scanning the exome in schizophrenia might quickly narrow down a few rare variants of large effect. Though the studies detected many rare variants, these were scattered across the exome, rather than piling up on certain genes. Not only does this bolster the idea that many genes contribute to schizophrenia risk, but it also emphasizes the need to do more sequencing to identify them and their effects.
This may require exome sequencing in at least 10,000 people, said Anna Need of Imperial College London, who was not involved in the study. “If you're looking for rare variants that are one in 1,000 or one in 10,000, you're not going to see many of those in a cohort of 2,500,” she added.
Rare variants have turned up in earlier forays into exome sequencing in smaller schizophrenia cohorts. One of these has been from Need and colleagues, which compared rare variants in cases and controls, and also did not support a role for a few rare variants of large effect (see SRF related news report). Previous studies of de novo mutations revealed a diverse group that rarely hit the same gene twice (see SRF related news report; SRF news report; news report; news report). This ambiguous situation, combined with the high degree of exome variation in each human being, leaves researchers straining to link rare variants to the disorder.
Many geneticists resort to gene set analysis, which asks whether variants found in schizophrenia predominately hit certain kinds of genes more than do variants found in controls. For example, one de novo study found that rare variants in the exome were enriched in genes highly expressed prenatally, when brain development is underway (see SRF related news report). The new studies both implicate sets of genes encoding parts of post-synaptic complexes that transduce glutamate signals important for information flow in the brain.
“It is reassuring that they are starting to identify these gene sets,” Need said. “But it’s not surprising that they’re not seeing gene-specific results yet because of their sample sizes.”
Sklar remains optimistic. “I'm actually really encouraged by these results,” she said. “This is genetics making big strides in building a strong foundation for understanding what’s going on. If the reality is more complicated than we expected, then understanding that is important.”
De novo notes
First author Menachem Fromer and colleagues sequenced the exomes of 617 people with schizophrenia, and each of their parents, from a Bulgarian cohort used in previous studies (see SRF related news report). This turned up 637 de novo variants within protein-coding or splice site sequences, giving a mutation rate of 1.61 x 10-8 per base per generation, which was not different from that expected in the general population. Similar to a previous study, this rate was associated with father’s age at the child's birth (see SRF related news report).
The burden of protein-altering variants—either non-synonymous ones that altered a protein’s amino acid sequence, or the loss-of-function ones that resulted in a premature stop codon, changed a splice site, or induced a frameshift—was the same in schizophrenia as in controls from other studies. When analyzed by gene, however, the researchers found what might be nascent pile-ups: Eighteen different genes were hit twice by these mutations, which was more than expected by chance (p = 0.03). Of these, two loss-of-function mutations hit TAF13, which encodes a subunit of the TFIID transcription initiation complex, and this attained genomewide significance.
The protein-altering variants found in schizophrenia also favored certain groups of genes, with significant enrichment in genes encoding parts of the activity-regulated cytoskeleton-associated scaffold protein (ARC) complex and the N-methyl-D-aspartate receptor (NMDAR) complex; for example, loss-of-function variants hit ARC complex genes 17 times more frequently in schizophrenia than those in controls. Disruptions to these complexes could disturb the process by which neural activity alters synaptic strength. Consistent with this, an enrichment in non-synonymous mutation was also found in targets of Fragile X mental retardation protein (FMRP), which also regulate synaptic plasticity.
The researchers also reported enrichment of their loss-of-function variants in genes hit by de novo mutation in autism (see SRF related news report) and intellectual disability, consistent with genetic overlaps suggested by copy number variations (Malhotra et al., 2012). People with mild cognitive impairments seemed to drive this overlap, however, which suggests that any overlaps between these disorders could reflect a shared cognitive component.
In the second study, first author Shaun Purcell and colleagues sequenced the exomes of 2,536 people with schizophrenia and 2,543 controls from a Swedish sample previously studied for a genomewide association study (see SRF related news report). This raked in 635,944 coding or splice-site variants, but none of the rare ones occurred more frequently in schizophrenia than in controls. Considering different variants hitting the same gene as repeat events also did not find any schizophrenia-specific associations.
To narrow the search space (and limit corrections for multiple comparisons), the researchers focused on variants occurring in 2,546 genes, pre-selected for their suspected involvement in schizophrenia by other methods. Loss-of-function mutations in this gene set occurred more frequently in schizophrenia than in controls; for example, 1,547 of the rare variants with a frequency of <0.1 percent occurred in cases and 1,383 in controls (p = 0.0001). The mean effect of these variants, however, was not large: While 46 percent of people with schizophrenia carried at least one loss-of-function variant in the schizophrenia-related gene set, 41 percent of controls did so, giving an odds ratio of 1.12. Because this number combines the effects of multiple variants across multiple genes, it is unclear whether this reflects the watering down of a few rare variants with sizeable effects on risk, or many variants, each with a small effect.
Looking for enrichment in biologically related gene sets, the researchers found it in the ARC complex, the post-synaptic density protein-95 (PSD-95) complex, and calcium channel genes. These results reinforce the de novo findings and point to the synapse as an important locus of schizophrenia risk. Apart from the FMRP gene set, the researchers did not find much enrichment in 2,507 genes linked to autism or intellectual disability, suggesting that any genetic overlap between these disorders is limited.—Michele Solis.
Fromer et al. De novo mutations in schizophrenia implicate synaptic networks. Nature (2014) doi:10.1038/nature12929. Published online 22 January 2014.
Purcell et al. A polygenic burden of rare disruptive mutations in schizophrenia. Nature (2014) doi:10.1038/nature12975. Published online 22 January 2014. Paper