Using genetic linkage analysis to identify nuclear genetic modifiers of the pathogenic mtDNA variation m.3243A>G

Abstract

Mitochondrial function is under bi-genomic control; pathogenic variants in both the nuclear and mitochondrial genomes (mtDNA) can result in clinical mitochondrial disease. The most common pathogenic, heteroplasmic mtDNA variant is m.3243A>G (NC_012920.1), which is associated with extensive clinical heterogeneity. Established risk factors (age and m.3243A>G variant level) explain only a small proportion of phenotypic variability, whereas high to moderate estimates of heritability for some m.3243A>G-related phenotypes provide evidence for the influence of unidentified nuclear factors. This work aimed to explore the genetic aetiology of this variability and identify plausible modifying factors using genetic linkage analysis. A cohort of 488 m.3243A>G carriers were recruited from: the UK Mitochondrial Disease Patient Cohort; the German Network for Mitochondrial Diseases; the Nationwide Italian Collaborative Network of Mitochondrial Diseases; and the Exeter Centre of Excellence for Diabetes Research. Analysis of nineteen recognised m.3243A>G-related phenotypes enabled the evaluation of the effect of the previously established risk factors, age and m.3243A>G variant level on the presence of disease-related phenotypes within this cohort, and also identified the influence of sex for some phenotypes; these factors explain a small proportion of observed phenotypic variance (pseudo-R range = 0.05-0.26). The performance of several genetic linkage analysis methods which facilitated the incorporation of known risk factors was evaluated in a representative sub-population of the available data. This established that genome-wide Haseman-Elston regression-based linkage analysis as implemented by Merlin-REGRESS was the most suitable approach for the data. This method was then applied to the full cohort of 208 individuals from 83 pedigrees who were informative for linkage analysis. For eight m.3243A>G-related phenotypes (cardiovascular involvement, cognition, dysphoniadysarthria, gastro-intestinal dysmotility, hearing impairment, migraine, neuropathy, and seizures), no regions of interest (LOD>1.8) were identified, indicating that the aetiology of these traits is likely to be highly complex, with no major nuclear loci influencing risk. For seven phenotypes (cerebellar ataxia, chronic progressive external ophthalmoplegia, diabetes, myopathy, psychiatric disturbance, ptosis, and stroke-like episodes) at least one region of interest was identified. For encephalopathy, seven regions of interest were identified, including two (LOD>3.3) on chromosomes 5 and 11, suggesting that a small number of nuclear genetic factors play a key role in the development of this severe neurological phenotype. The significance of these results was further assessed using simulation studies. Identified linkage regions can be explored at a higher resolution using genome-wide association studies (GWAS) and whole genome sequencing (WGS). Identifying the nuclear factors involved in the development of m.3243A>G-related disease will provide a better understanding of the underlying genetic architecture of disease, subsequently informing knowledge surrounding the mechanisms of disease, as well as enabling clinicians to provide tailored prognostic advice.