Identification of Nuclear Genetic Loci Linked to Clinical Features of the m.3243A>G Mitochondrial DNA Variant

Mitochondrial DNA (mtDNA) disorders exhibit striking clinical variability that is poorly explained by known factors such as variant heteroplasmy, age, or sex. Nuclear genetic modifiers likely play a significant role in this heterogeneity. We aimed to characterize the nature of nuclear genetic involvement for 2 common syndromic presentations of the common pathogenic mtDNA variant, m.3243A>G: mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes (MELAS) and maternally inherited diabetes and deafness (MIDD).

We assembled a multicenter cohort of clinically ascertained carriers of m.3243A>G (total n = 488), identifying 198 individuals across 76 pedigrees suitable for genetic linkage analysis. We investigated 4 clinical features characteristic of MELAS and MIDD: diabetes, hearing impairment, stroke-like episodes, and encephalopathy. Haseman-Elston regression-based genetic linkage analysis was performed to identify regions of the nuclear genome cosegregating with these features. The effects of m.3243A>G heteroplasmy, age, and sex were accounted for using logistic regression; empirical significance thresholds were determined through feature-specific gene-dropping simulations. Association analyses were performed in 247 individuals using single-variant (SAIGE) and gene-based approaches (SAIGE-GENE+ and MAGMA) to refine candidate loci within a significant linkage region.

We identified significant genetic linkage to encephalopathy (chromosome 7q22; LOD = 3.72), and regions suggestive of genetic linkage on chromosomes 1, 5, 6, 11, and 13, for encephalopathy and stroke-like episodes. No linkage was identified for diabetes or hearing impairment. Association analysis within the chromosome 7 region identified variant rs62500792 (intergenic between SDHAF3 and TAC1) with the lowest p value (3.7 × 10^-5), yet no variants reached the proportional significance threshold (5.3 × 10^-6). Gene-based analyses highlighted PLOD3 (p = 3.9 × 10^-3) and IMMP2L (p = 6.4 × 10^-3) as candidates, as each showed the strongest gene-level signals within the linkage region across complementary burden-testing methods, although neither reached corrected significance thresholds.

The nuclear genetic architecture modifying m.3243A>G differs across clinical features. Severe neurologic features (encephalopathy and stroke-like episodes) may be influenced by a small number of nuclear genes with relatively large effect sizes, whereas the nuclear contribution to diabetes and hearing impairment appears more polygenic. This study highlights the value of large, well-characterized patient cohorts in identifying modifier loci and advancing knowledge of the mechanisms underlying phenotypic variability in mtDNA disease.