A survey of genomic studies supports association of circadian clock genes with bipolar disorder spectrum illnesses and lithium response

Abstract

Circadian rhythm abnormalities in bipolar disorder (BD) have led to a search for genetic abnormalities in circadian "clock genes" associated with BD. However, no significant clock gene findings have emerged from genome-wide association studies (GWAS). At least three factors could account for this discrepancy: complex traits are polygenic, the organization of the clock is more complex than previously recognized, and/or genetic risk for BD may be shared across multiple illnesses. To investigate these issues, we considered the clock gene network at three levels: essential "core" clock genes, upstream circadian clock modulators, and downstream clock controlled genes. Using relaxed thresholds for GWAS statistical significance, we determined the rates of clock vs. control genetic associations with BD, and four additional illnesses that share clinical features and/or genetic risk with BD (major depression, schizophrenia, attention deficit/hyperactivity). Then we compared the results to a set of lithium-responsive genes. Associations with BD-spectrum illnesses and lithium-responsiveness were both enriched among core clock genes but not among upstream clock modulators. Associations with BD-spectrum illnesses and lithium-responsiveness were also enriched among pervasively rhythmic clock-controlled genes but not among genes that were less pervasively rhythmic or non-rhythmic. Our analysis reveals previously unrecognized associations between clock genes and BD-spectrum illnesses, partly reconciling previously discordant results from past GWAS and candidate gene studies.

Figure 1. Organization of the extended clock network.

The core clock oscillator is a set of ∼18 genes that encode for transcriptional regulators (middle). These proteins are organized in complex feedback loops with positive (green) and negative (red) limbs that generate the ∼24 hr rhythms in gene expression responsible for maintaining circadian rhythms. Upstream clock modulators influence the period and/or amplitude of rhythms by altering protein stability, cellular distribution, or phosphorylation of proteins within the core clock (top). Core clock transcriptional regulators generate expression rhythms in numerous downstream clock controlled genes that are not the “gears of the clock” involved in generating rhythms, but may be important effectors or “hands of the clock” (bottom).

Figure 2. The method for clock gene and random gene analysis is summarized for three test sets.

The rate of association in SLEP was determined for sets of randomly selected genes that contained gene numbers similar to A) core clock genes and B) clock modulator genes. C) Pervasively rhythmic (PRCCGs), weakly rhythmic (WRCCGs) and non-rhythmic genes were directly compared. Of 9950 clock controlled genes, 5187 were rhythmic in 2–5 tissues. Genes of intermediate rhythmicity were not examined.

Figure 3. Overlap among clock genes, genes associated with BD spectrum illnesses by GWAS, and lithium-responsive genes.

For each set of genes, the % associated with BD spectrum illnesses by GWAS (blue), the % found to be lithium-responsive (red), and the overlap between these (purple) is shown for 18 core clock genes (A), 342 clock modulator genes (C) and 136 pervasively rhythmic clock controlled genes (CCGs) (E) compared to their respective randomly selected controls (B and D), CCGs that are less pervasively rhythmic (F), and genes with no evidence for rhythmicity (G). Gray indicates % of genes with no relationship to either BD or lithium response.