From heteroplasmy shift to direct base correction: mitochondrial genome editing for disease modeling and therapy

Mitochondria are essential organelles that support cellular energy metabolism and contribute broadly to tissue homeostasis, stress responses, and disease pathophysiology. As the mitochondrial genome encodes core components of oxidative phosphorylation, genetic alterations in mitochondrial DNA (mtDNA) can profoundly affect cellular and tissue function. Nevertheless, compared with the nuclear genome, mtDNA remains difficult to interrogate and manipulate because of longstanding technical barriers to mitochondrial genome engineering. Consequently, many fundamental questions surrounding mtDNA biology remain unresolved, including the mechanisms governing heteroplasmy, mutationspecific disease phenotypes, tissue-dependent pathogenic thresholds, and long-term mtDNA population dynamics. These limitations are particularly evident in mitochondrial genetic diseases, which can cause severe neurological, muscular, cardiac, and visual phenotypes but for which only a limited number of genetically defined disease models are available. Recent advances in mitochondrial genome editing are beginning to overcome these barriers. In this review, we organize the field according to therapeutic strategy, from selective depletion of mutant genomes to direct base correction and emerging precision-editing architectures. Mitochondria-targeted zinc-finger nucleases, mitoTALENs, and mitoARCUS can selectively eliminate mutant mtDNA and shift heteroplasmy toward wild-type genomes, whereas DddAderived cytosine base editors, TALE-linked deaminases, strandselective mitochondrial base editors, and their engineered derivatives enable direct C·G-to-T·A or A·T-to-G·C conversion without programmed double-strand breaks. We evaluate these platforms in relation to mutation class, heteroplasmy level, sequence context, bystander and off-target editing, target tissue, delivery format, durability, and functional rescue. Beyond therapeutic correction, mitochondrial genome editors are opening new opportunities for disease modeling by enabling defined pathogenic variants to be introduced into cells, organoids, and animal models. Validated applications include human cell and organoid systems and genetically defined mouse rat models. Extension to larger animals and nonhuman primates remains a future direction for the disease-modeling framework discussed here. Clinical translation, however, remains constrained by cargo size, co-delivery requirements, unintended nuclear exposure, long-term mtDNA dynamics, and the need for efficient and tissue-specific delivery to clinically relevant organs and cell types. Finally, we propose a framework linking genotype, heteroplasmy, and tissue context to editor selection, with the goal of moving the field from measurements of editing efficiency toward mechanistic understanding and therapeutic relevance.