Investigating the causal and consequential role of the nuclear encoded mitochondrial genome in Amyotrophic Lateral Sclerosis

Abstract

Mitochondrial dysfunction is implicated in amyotrophic lateral sclerosis (ALS), but which mitochondrial pathways are involved, and whether their involvement is a cause or a consequence of disease remains unclear. We investigated whether nuclear-encoded mitochondrial gene (NEMG) pathways are enriched for ALS genetic risk, which are dysregulated in ALS tissue, whether pathway activity defines patient subgroups that differ in survival, and whether co-expression between the nuclear and mitochondrial genomes is altered. We tested MitoCarta3.0 pathways for enrichment of ALS genetic risk using MAGMA and the largest European ALS GWAS to date (27,205 cases and 110,881 controls), using hierarchical testing across pathway tiers, and annotated the direction of gene-level effects using published TWAS and SMR results. In >1,500 bulk RNA-seq samples from post-mortem motor cortex and spinal cord (KCL Brain Bank, NYGC ALS Consortium) and iPSC-derived motor neurones (Answer ALS), we performed cell type adjusted gene set enrichment analysis (GSEA), gene set variation analysis (GSVA)-based clustering with exploratory survival analysis, and nuclear-mitochondrial co-expression analysis. Of seven broad mitochondrial categories, only the mitochondrial central dogma was enriched for ALS risk after correction (Padj = 0.044). Hierarchical testing within this pathway localised the signal to translation (Padj = 0.008) and the mitoribosome (Padj = 0.014). Genetically predicted expression pointed in opposite directions for different mitoribosomal genes (lower MRPS33 and higher MRPL28 expression with increased risk), and did not agree in direction with differential expression in ALS tissue more often than expected by chance. A nominal enrichment in ROS and glutathione metabolism was largely attributable to SOD1. In tissue, mitochondrial genes were broadly more highly expressed in ALS motor cortex in both cohorts, largely as a global shift across mitochondrial genes. OXPHOS subunits and the mitoribosome were enriched beyond this shift in ALS Consortium motor cortex (both PFDR = 0.027), as were OXPHOS subunits in iPSC-derived motor neurones (PFDR = 2.4 x 10-), whereas spinal cord showed no genome-wide enrichment. GSVA based clusters were associated with diagnosis in motor cortex, and nominally with survival in lumbar spinal cord (adjusted HR 0.69, Punadj = 0.013), where cluster structure was weakest. No individual pathway was associated with survival after correction in adjusted models. In all four post-mortem tissues, nuclear-encoded mitochondrial genes were more negatively co-expressed with mtDNA-encoded genes in ALS than in controls, most strongly in lumbar spinal cord, whereas iPSC-derived motor neurones showed the opposite pattern. In conclusion, inherited ALS risk shows a modest enrichment centred on mitochondrial translation, with gene specific rather than uniform directions of effect. The same pathway, alongside OXPHOS, is more highly expressed in end-stage ALS motor cortex, which may reflect a response to disease, residual differences in cell composition, or both. At the gene level, genetically predicted and observed expression changes did not agree in direction, so whether inherited risk and these tissue changes are linked remains unresolved.