Functional gene networks reveal distinct mechanisms segregating in migraine families

Abstract

Migraine is the most common neurological disorder worldwide and it has been shown to have complex polygenic origins with a heritability of estimated 40-70%. Both common and rare genetic variants are believed to underlie the pathophysiology of the prevalent types of migraine, migraine with typical aura and migraine without aura. However, only common variants have been identified so far. Here we identify for the first time a gene module with rare mutations through a systems genetics approach integrating RNA sequencing data from brain and vascular tissues likely to be involved in migraine pathology in combination with whole genome sequencing of 117 migraine families. We found a gene module in the visual cortex, based on single nuclei RNA sequencing data, that had increased rare mutations in the migraine families and replicated this in a second independent cohort of 1930 patients. This module was mainly expressed by interneurons, pyramidal CA1, and pyramidal SS cells, and pathway analysis showed association with hormonal signalling (thyrotropin-releasing hormone receptor and oxytocin receptor signalling pathways), Alzheimer's disease pathway, serotonin receptor pathway and general heterotrimeric G-protein signalling pathways. Our results demonstrate that rare functional gene variants are strongly implicated in the pathophysiology of migraine. Furthermore, we anticipate that the results can be used to explain the critical mechanisms behind migraine and potentially improving the treatment regime for migraine patients.

Keywords: complex trait; gene-gene interaction; genetic network; migraine; transcriptomics.

Figure 1

Study design. Co-expression networks were constructed for the trigeminal ganglion, visual cortex and aorta, to identify tissue-specific modules of highly co-expressed genes. In parallel, the WGS data were analysed to identify mutations. Modules with over-representations of mutations were defined as potential ‘migraine modules’. Similarly, we defined mutations in an independent cohort compared to population controls and replicated the potential migraine modules. Replicated modules were further analysed using functional annotation. RVIS = residual variation intolerance score.

Figure 2

Overview of genes and mutations in trigeminal ganglion, visual cortex and aorta. (A and B) Number of genes and mutated genes in the trigeminal ganglia (TG), visual cortex (VC) network and aorta. (C) Module overlap based on the mutated genes in both trigeminal ganglia, visual cortex and aorta. (D) Distribution of mutations in mutated genes in trigeminal ganglion, visual cortex and aorta. (E) Gene-set enrichment analysis using brain single-cell RNA sequencing data. *P < 0.05.

Figure 3

Identification of migraine-associated modules. (A) Over-representation of mutated genes for familial migraine within modules of the trigeminal ganglia, the visual cortex and in the aorta network. *Modules with an over-representation of mutated genes in familial migraine than are assessed for replication analysis. (B) Replication in sporadic migraine (Spo. mig.) compared to 1000G. The odds ratio (OR) is displayed for each module and the error bars show the 95% confidence interval.

Figure 4

Analysis of modules on protein level. (A) Network of proteins encoded by genes in the brown module (visual cortex, VC). A large fraction of the proteins contribute to the network. (B) Three networks of proteins encoded by genes in the yellow (trigeminal ganglia, TG) module are depicted.

Figure 5

Count of mutated genes in visual cortex brown and in how many families the genes co-segregate with migraine.