November 6, 2013. Different psychiatric illnesses share genetic roots, and researchers are searching for the genes that increase the risk for several disorders. This year, two studies have pinpointed common variants that show some overlap in their contributions to risk for schizophrenia, bipolar disorder, depression, autism, and attention-deficit/hyperactivity disorder (ADHD) (see SRF related news story and SRF news story). The talks in a symposium at the World Congress on Psychiatric Genetics in Boston on Monday, October 21, focused on similarities and differences in the genetic factors for these disorders, particularly between schizophrenia and bipolar disorder.
Overlaps between schizophrenia and bipolar disorder
Although the contribution of copy number variants (CNVs)—the loss or gain of DNA segments often containing many genes—to schizophrenia has considerable support (see SRF schizophrenia genetics primer), a similar CNV connection to bipolar disorder has been less apparent (see SRF related news story). Elaine Green of Plymouth University, UK, presented new results comparing the frequency of rare CNVs in bipolar disorder and schizophrenia. CNVs did turn up in bipolar disorder, but they did not amount to the CNV burden found in schizophrenia. For example, large deletions (>1 Mb) were found in schizophrenia 2.3 times more often than in bipolar disorder. Removing CNVs already known to be associated with schizophrenia abolished this difference between disorders, suggesting that their shared genetic roots do not include rare, large CNVs. Green speculated that large CNVs incur more severe phenotypes, such as IQ deficits or early onset.
Among rare point mutations in voltage-gated calcium channel genes, however, the tables were turned, with bipolar disorder carrying a larger burden than schizophrenia. These preliminary results, presented by Douglas Ruderfer of Mount Sinai School of Medicine, New York, came from exome sequencing 2,500 cases of schizophrenia, 1,100 of bipolar disorder, and 2,500 controls. Given the clear signal at the calcium channel gene CACNA1C delivered by genomewide association studies (GWASs) of common variants (see SRF related news story), Ruderfer combed through the exome sequences of this and other voltage-gated calcium channel genes for rare, loss-of-function variants. Though both disorders had more of these than were found in controls, bipolar disorder had a significantly greater burden than schizophrenia did. This offers additional support for calcium channel alterations in psychiatric illness, particularly bipolar disorder.
Psychiatric disorders often come with other neurological problems, such as migraine headaches. This suggests that disease-related genetic risk factors may also be responsible for co-morbid conditions. To explore this, Verneri Antilla of the Broad Institute, Cambridge, Massachusetts, described the Brainstorm Project, which combines cases of various psychiatric illnesses (schizophrenia, bipolar, major depression, autism, ADHD, Tourette's syndrome) with cases of neurological disorders (stroke, migraine). This leads to a giant dataset of 100,000 cases and 100,000 controls—big enough to find genomewide correlations. Preliminary analyses confirmed the close relationship between schizophrenia and bipolar disorder, but also suggested a relationship between schizophrenia and stroke, as well as between bipolar disorder and migraine.
In GWASs, an exciting genomewide-significant SNP can still leave researchers guessing about the precise gene or location of the genome it implicates. This is tricky if a signal occurs in a region of extended linkage disequilibrium (LD), which means that it marks a long stretch of DNA that tends to stay together through many cycles of cell division. Two talks in the session proposed ways of resolving signals within these so-called LD blocks. Giulio Genovese of the Broad Institute turned to African-American and Latino populations collected from the Genomic Psychiatry Cohort (GPC). African-Americans and Latinos are admixed populations, meaning that their ancestors come from more than one continent, and this is reflected in their genomes, which have a different LD structure than the European populations used in most GWASs. Studying thousands of cases and controls of these admixed populations, Genovese found similar schizophrenia-related signals: Most known schizophrenia CNVs replicated in this sample with a similar frequency, and though a GWAS of these cases was underpowered, many SNPs had the same direction of association with disease as those flagged so far by the Psychiatric Genomics Consortium’s GWAS (see SRF related conference story).
Beyond this, the African-American and Latino data may also clarify things in the European-based PGC data. Genovese noted an extensive region along 3q26.33 containing many genomewide-significant single nucleotide polymorphisms (SNPs) in the PGC GWAS. Adding the African-American and Latino data to this bumped up some signals in a narrower region of 3q26.33, suggesting a more precise locale for a disease-related variant.
Instead of diversifying samples, Mark Reimers of Virginia Commonwealth University diversified the information used to sort through genomewide-significant SNPs obtained for schizophrenia. Reimers presented a method for integrating information about how conserved an SNP is across mammals and whether it is a DNAse hypersensitivity site, indicating a regulatory region (see SRF related news story). He found SNPs previously associated with schizophrenia were more likely to be conserved and lie in a DNAse hypersensitivity site. Also, in regions where several SNPs achieved similar levels of genomewide significance, only a subset was also conserved or had DNAse hypersensitivity. Taking into account this information could help home in on the disease-related signals and prioritize variants for follow-up.—Michele Solis.