5 October 2012. A new round of sequencing in schizophrenia offers up a new batch of genetic mutations in sporadic cases of the disorder. Published online October 3 in Nature Genetics, the study undertakes the largest hunt yet for “de novo” mutations in the protein-coding exomes of 231 cases of schizophrenia and 34 controls. The researchers, led by Maria Karayiorgou and Joseph Gogos at Columbia University in New York, report more function-changing mutations in schizophrenia than in controls, and these tended to land in genes associated with prenatal brain development. They also identified four genes hit by mutations in two different people (LAMA2, DPYD, TTRAP, and VPS39). The researchers estimate that 46 percent of their mutations are true risk variants, and they highlight genes new to the field.
The study marks the latest installment in a series of exome sequencing studies of schizophrenia published in the past year, and the Karayiorgou and Gogos group’s second look at de novo mutations specifically. Arising anew in sperm or egg cells, de novo mutations haven’t been weeded out by natural selection. If they give rise to severe biological effects, they could make it easier to localize the genes disrupted in a disorder, compared to subtly acting, common variants (see SRF genetics overview). So far, exome sequencing has turned up a scattering of de novo variants in people with schizophrenia, which outnumber those in controls and localize to a diverse group of genes (see SRF related news story and SRF news story).
But the difficulty lies in interpreting these rare variants once they are detected: Are they pathogenic, or harmless bystanders? This has been made tricky by the discovery of similar rare variants in healthy people: For example, the de novo mutation rate in the exome in people with autism is similar to that of their unaffected siblings (see SRF related news story). In fact, we are all effectively knockouts of some sort or another (MacArthur et al., 2012). This conundrum dogs any kind of rare event, de novo or not: Another recent exome sequencing study did not find that moderately rare variants found in schizophrenia cases occurred more frequently than in controls (see SRF related news story).
To deal with this, researchers are trying various ways to validate their hard-won mutations as true risk variants. This includes statistical arguments about their enrichment in cases versus controls; predicted effects on the encoded protein; their membership in a particularly tantalizing set of genes, such as synaptic ones; and replication in other cohorts. The new study tries these and more, notably expanding its original Afrikaner sample (see SRF related news story) and including a cohort from the United States.
By the numbers
First author Bin Xu and colleagues sequenced 795 exomes in mother-father-child trios to find de novo mutations—present in the child, but not in either parent. They focused on single-nucleotide variants (SNVs) in which one base is replaced by another, and indels, small deletions and/or insertions of bases within the sequence. They did this in 146 Afrikaner trios (53 of these trios were sequenced in their study last year), 85 trios from the United States, and 34 unaffected control trios (22 of which were also in their previous exome study).
In the Afrikaner sample, the researchers identified 93 SNVs and nine indels. The bulk of the SNVs consisted of non-synonymous mutations predicted to change an amino acid in the resulting protein; others occurred in splice sites, which could also change amino acid composition. The indels predicted either amino acid deletion or premature protein truncation. These function-altering mutations (i.e., non-synonymous, splice site, and indels) outnumbered the synonymous ones by 7.6 to 1. The U.S. sample turned up 53 SNVs and four indels, with function-altering mutations dominating the synonymous ones by 4.3 to 1. This theme continued in the combined sample, which had a functional-to-synonymous ratio of 6. Compared to a ratio of 2.6 in controls, they found the enrichment in schizophrenia greater than expected by chance.
For these different cohorts, including the controls, the researchers measured similar mutation rates in the range of 1.28-1.7 x 10-8 mutations per base per generation, which is similar to rates reported in other de novo studies. They also found a greater number of de novo mutations in those born to older fathers, which jibes with a recent study reporting a higher chance of de novo events with increasing paternal age (see SRF related news story).
Finding rare mutations in the same gene in different people with schizophrenia, but not in controls, could strengthen the case for their pathogenicity. The researchers found four instances of this, with two hits found in four different genes: LAMA2, which encodes a component of the extracellular matrix; DPYD, an enzyme that breaks down pyrimidine; TTRAP, or transformation/transcription domain-associated protein; and VPS39, which encodes a protein related to vacuoles inside cells. None of these genes harbored de novo mutations in the controls, nor in the unaffected siblings in the autism exome studies, and problems with these genes have been previously associated with other neurodevelopmental disorders. The researchers also note that DPYD is near MIR137, a microRNA that is one of the top hits in the largest schizophrenia genomewide association study (GWAS; see SRF related news story), and they suggest that the GWAS signal might stem from DPYD.
To extend the search for “repeat offender” genes implicated multiple times in different people, the researchers turned to their previous data on de novo copy number variations (CNVs), the loss or gain of a chunk of DNA (see SRF related news story). Though CNVs are usually large enough to disrupt multiple genes, any overlap between these and the single genes fingered by the de novo variants found here might flag true risk variants. They report five genes hit by both, including DGCR2, a key gene in the 22q11.2 deletion associated with schizophrenia.
Guilt by functional association?
To gather more evidence implicating these de novo mutations in schizophrenia, the researchers examined different aspects of gene function. The set of genes marked by de novo mutation in this study was not enriched for genes belonging to synaptic categories, and pathway analysis found it was not particularly enriched for genes that work together. The researchers did find, however, a tendency for these genes to be strongly expressed prenatally. For example, the ratio of functional to synonymous variant was 9.50 in the subset of genes that showed the strongest expression in prenatal hippocampus and dorsal lateral prefrontal cortex, regions associated with schizophrenia. In contrast, this ratio was 5.44 in genes expressed more strongly postnatally, and 4.75 in genes with constant expression. Among de novo variants found in controls, this ratio did not vary. This suggests that mutations affecting genes involved in early stages of brain development are relevant to schizophrenia, write the authors. Consistent with this, the people with schizophrenia who carried mutations to these prenatally biased genes were more likely to exhibit multiple abnormal behaviors as children, and had worse functional outcomes than those who did not.
Whether this insight really incriminates these variants in schizophrenia remains to be seen, but it illustrates how creative ways of looking at gene function may help pin blame. Larger sample sizes will also be necessary, and may help replicate some of these findings and narrow in on true risk variants.—Michele Solis.
Xu B, Ionita-Laza I, Roos JL, Boone B, Woodrick S, Sun Y, Levy S, Gogos JA, Karayiorgou M. De novo gene mutations highlight patterns of genetic and neural complexity in schizophrenia. Nat Genet. 2012 Oct 3.