The genetic basis of seasonal affective disorder

<p>Family and twin studies have shown a heritable component to seasonal affective disorder (SAD). While a few studies have examined individual genetic variants in SAD, many methodological issues exist in the current literature. First, most studies combined major depression (MDD) and bipolar (BD) cases in the genetic analysis of SAD. This makes it difficult to differentiate the effect from MDD and BD. Second, most studies adopted a candidate gene approach and used fairly small sample sizes. This does not allow for testing across a wide variety of genes, and it yields less robust P-values. Third, healthy controls have been used, but not case comparisons, which makes it difficult to differentiate the effects of seasonality from that of the primary illness (MDD and BD). To overcome these issues, seasonal MDD and BD cases were separated into two different studies in this thesis; sample sizes for both studies are the largest in the current SAD molecular genetics literature; GWAS was used to test for potential risk loci in a hypothesis-free fashion; case comparisons were incorporated to exclude potential genetic contributions related generally to the primary diseases themselves (MDD and BD).</p> <p>For MDD, we performed a GWAS with 562 seasonal MDD cases and 1,225 comparison cases with non-seasonal MDD. Subjects were drawn from two iterations of the Genetics of Recurrent Early Onset Depression (GenRED) study. Seasonal cases were those whose depressive episodes typically started in fall or winter. A mega-analysis of the two GWAS datasets was done using SNPTEST. We found that two single nucleotide polymorphisms (SNPs), rs149882931 and rs77073398, on chromosome 16p12.1 were associated with seasonal depression, at a genome-wide significant level (OR= 1.66, P= 3.59 x 10-8 and OR=1.62, 4.76 x 10-8, respectively). Since SAD is more prevalent in females, a female-specific analysis was carried out. The two variants were more significant in this analysis: P=2.18x10-9 (OR=1.89) and P=2.79x10-9 (OR=1.82), respectively, and a significant sex-by-SNP interaction was observed. These SNPs are located in a conserved intergenic region between the genes HS3ST4 and C16orf82. The protein product of HS3ST4 modifies the side chains of heparan sulfate proteoglycans. We therefore tested the hypothesis that the heparan sulfate biosynthesis pathway would be enriched in nominally significant SNPs using the SNP ratio test, and found evidence for such enrichment (P=0.008, SNP ratio test, P=0.027, SKAT).</p> <p>For BD, the GWAS analysis of 818 seasonal BD cases and 1,515 healthy controls showed that BD-S is most strongly associated with two SNPs within the ZBTB20 genes. BD subjects were drawn from NIMH Bipolar Genetics Study (BIGS), and seasonal cases were defined as those with depressive episodes starting in fall or winter. An association study was carried out with SNPTEST, and we found two single nucleotide polymorphisms (SNPs) in the intronic region of ZBTB20 gene to be associated with BD-S (rs7646282, OR=2.34, P= 7.23 x 10-8 and rs139459337, OR=2.37, 8.05 x 10-8). A similar case-only study was carried out with 818 BD-S cases and 1239 cases without seasonal depressive symptoms (non-BDS), though no SNP was found to be significantly associated in this analysis. rs7646282 is the strongest SNP in cis-association with ZBTB20 gene expression, and ZBTB20 has been shown to affect the neural development of the hippocampus, a brain region implicated in the pathophysiology of BD.</p> <p>Finally, we sought to determine whether there is a role for circadian rhythm genes in BD susceptibility. In this study, we used a discovery set of 189 exome-sequenced BD patients and 105 healthy controls to look for circadian genes associated with BD. We found the DRD2 gene to be the circadian gene most strongly associated with BD. Among the rare damaging variants in the DRD2 gene, the S311C variant was the predominant SNP. To test whether this variant segregates in family members with BD, we genotyped the family members of probands from the discovery sample. This data was used for a linkage and family-based association study. Even though the linkage analysis was only very weakly positive, the family-based association study showed significant segregation of the variant in family members with BD (P< 0.05). To follow up on this finding, we further genotyped 2,185 unrelated BD cases and 1,982 healthy controls. We found no support for the S311C variant in this replication dataset. Sub-phenotype study of psychotic features and mood-incongruence also did not show significant association. Meta-analysis with 2,994 BD cases and 3,661 controls, however, revealed no association between the S311C variant and BD.</p>