30 May 2008. In the latest of a flurry of recent papers examining the role of copy number variations (CNVs) in schizophrenia, researchers report a strong association between de novo CNVs and the sporadic, or nonfamilial, form of the disorder. The paper is part of another flurry as well, being the third major paper this month from the Columbia University groups led by Maria Karayiorgou and Joseph Gogos, close on the heels of their recent DISC1 and 22q11 deletion syndrome mouse model papers (Kvajo et al., 2008; Stark et al., 2008).
Two studies published earlier this year (see SRF related news story) implicated CNVs in schizophrenia, but neither was designed to directly address the relative contribution of inherited versus de novo CNVs to sporadic cases.
One of these studies (Kirov et al., 2008), which reported CNV disruptions of neurexin1 (NRXN1) and amyloid precursor-binding protein A2 (APBA2) in subjects with schizophrenia, compared a sample with equal numbers of sporadic and familial cases to control subjects. In one arm of the second study (Walsh et al., 2008), parents of 83 patients with childhood-onset schizophrenia (COS), an extremely rare and severe form of the disorder (see Q&A with Nitin Gogtay accompanying SRF related news story), were genotyped to identify de novo CNVs “insofar as possible,” but the authors conceded that, aside from two mutations that were “presumptively” de novo, 25 other CNVs they reported were “either inherited or of unknown origin.”
To focus more closely on the role of de novo CNVs (referred to as "copy number mutations" by the authors) in sporadic cases of schizophrenia, Karayiorgou, Gogos, and first author Bin Xu joined forces with researchers at the University of Pretoria in a study of 1,017 individuals from the genetically homogeneous Afrikaner population of South Africa (Xu et al., 2008). Of these individuals, 200 had been diagnosed with schizophrenia or schizoaffective disorder and 159 were unaffected control subjects. However, both biological parents of all affected and control subjects were also enrolled in the study, allowing the team to confirm that 152 affected individuals were sporadic cases, with no incidence of schizophrenia in first- or second-degree relatives, while 48 were familial cases.
A fine filter
The researchers performed whole-genome scans of the entire sample at a resolution of ~30 kb. They eliminated from consideration any CNVs that had >50 percent overlap with a CNV in any parental chromosome (including across families), thus avoiding the inclusion of common CNVs that might result from high mutation rates in unstable genomic regions and any commonly inherited CNVs that might falsely appear to be de novo without this filtering procedure.
The researchers identified 15 de novo CNVs in the 152 sporadic cases (~10 percent), but only two in the 159 controls (1.3 percent), meaning that de novo mutations were about eight times more common in sporadic cases, a highly significant association (p = 0.00078). Moreover, the group found no de novo CNVs in the 48 familial cases of schizophrenia, indicating that “…the association between de novo [CNVs] seems to be primarily confined to the sporadic cases.” In a comparison of the collective rate of de novo and inherited CNVs in sporadic cases and unaffected controls, the researchers found that sporadic cases were only 1.5 times more likely to harbor inherited CNVs (p = 0.049), and they argue that “only a small portion of [these inherited mutations] should be expected to contribute to the pathogenesis of the sporadic cases.”
As for the nature of the CNVs, the researchers identified roughly equal numbers of genomic gains and losses in the sporadic cohort, but they cite three microdeletions in the 22q11.2 locus as particularly significant. These mutations, which were found in three different subjects, represent the only recurrent structural variations in the study, and confirm earlier reports by Karayiorgou’s group and others that variations in this locus confer susceptibility to schizophrenia (Karayiorgou et al., 1995; Liu et al., 2002). Of special interest in this regard, a de novo mutation identified at 14q32.13–32.2 includes DICER1, the protein product of which regulates microRNA production. Deletions at 22q11.2 disrupt DCGR8, a key miRNA processing protein, which these researchers recently tied to schizophrenia-relevant deficits in their 22q11DS mouse model (Stark et al., 2008).
The authors also report three de novo CNVs mutating single genes, including a microdeletion affecting RAPGEF6. This gene, a GTP exchange factor that acts with the GTPases Rap1 and Rap2, has been supported as a candidate gene for schizophrenia in both linkage and association studies (Lewis et al., 2003; Chen et al., 2006). Among the other genes affected by the de novo CNVs detected in this study, Xu and colleagues report statistically significant enrichment in neurodevelopmental, small GTPase, and RNA binding/processing pathways among the schizophrenia cases.
"[De novo CNVs] can explain, at least in part, how schizophrenia persists in the population despite the low fecundity of affected individuals. Second, our findings suggest that de novo genetic lesions at many different loci can contribute to schizophrenia, and this heterogeneity may account for the difficulties in finding genetic variants with a major effect on disease risk," the researchers write.—Peter Farley.
Xu B, Roos JL, Levy S, van Rensburg EJ, Gogos JA, Karayiorgou M. Strong association of de novo copy number mutations with sporadic schizophrenia. Nat. Genet. Published online 30 May 2008; doi 10.1038/ng.162.