mtDNA Heteroplasmy

The coexistence of wild-type and mutant mitochondrial DNA molecules within the same cell or tissue. Because each mammalian cell carries hundreds to thousands of mtDNA copies across dozens of mitochondria, a mutation can be present at any fraction from near-zero to near-100 %. In young tissues most cells are effectively homoplasmic (one dominant genotype); with age, stochastic replication errors and clonal expansion produce a mosaic of cells with varying mutant burdens — a defining contributor to the mitochondrial-dysfunction hallmark.

This page covers the somatic heteroplasmy process as it relates to aging. Germline heteroplasmy (transmission of pathogenic mtDNA variants from mother to offspring, as in LHON, MERRF, and MELAS) is addressed briefly under the aging/disease distinction section below.

Mechanism — how heteroplasmy arises and expands

Step 1 — replication error introduces a variant

Mitochondrial DNA is replicated continuously, independently of the cell cycle, by the replication machinery anchored at polg (mtDNA polymerase gamma). Each replication event carries a low but non-zero probability of introducing a point mutation or deletion. The 4,977 bp “common deletion” removes a large segment between two direct repeat sequences and is the best-characterised age-associated deletion ¹.

Step 2 — random genetic drift

Within a cell, individual mtDNA molecules compete for replication. In the absence of strong selection, the variant load in any one cell undergoes a random walk — the “relaxed replication” model. Over many cell divisions (or mtDNA turnover cycles in post-mitotic cells), purely stochastic processes can drive a variant toward fixation or extinction in that cell. This drift is slow in proliferating cells with large effective population sizes and faster in post-mitotic cells with few mitochondria per nucleoid.

Step 3 — clonal expansion in tissues

In tissues that are organised as clonal units — most clearly demonstrated in colonic crypts — a single crypt-base stem cell can propagate its mtDNA genotype to all daughter cells of that crypt. Greaves et al. (2014) showed that clonally expanded mtDNA point mutations in human colorectal epithelium increase dramatically between ages 40 and 80, and that these expansions begin as early as mid-life ². By late life, a significant fraction of crypts become entirely filled with a single expanded mutant, detectable by COX (complex IV) deficiency histochemistry.

Tissue	Vulnerability	Mechanism
Colonic crypts	High	Clonal unit structure amplifies single-stem-cell drift
Skeletal muscle	High	Post-mitotic fibers; focal COX-deficient segments accumulate with age
Substantia nigra neurons	High	Long-lived post-mitotic; very low mtDNA copy repair capacity
Cardiac myocytes	Moderate–high	Post-mitotic; long-lived; heteroplasmy detectable but cardiac consequences unclear needs-replication
Liver	Lower	Regenerative capacity partially resets heteroplasmy

The post-mitotic muscle and neuron vulnerability reflects loss of the dilution mechanism: dividing cells can “dilute out” a low-frequency mutant across daughters, whereas post-mitotic cells accumulate variants indefinitely.

Heteroplasmy threshold effect

Below a tissue-specific threshold of mutant mtDNA load, residual wild-type molecules are sufficient to maintain normal oxidative phosphorylation (OXPHOS). Above the threshold, OXPHOS capacity collapses sharply. Reported thresholds for most pathogenic variants:

Complex I subunit mutations (ND genes): ~60–70% mutant load required for biochemical defect
tRNA mutations (MT-TL1, MT-TA): ~80–90% mutant load before OXPHOS failure
Common deletion: ~60% deletion load for complex I/IV deficiency in most tissues

These thresholds are variant- and tissue-specific and were characterised primarily in mitochondrial disease patients with high germline heteroplasmy, not in the low-heteroplasmy aging context. At the single-cell level, individual aging neurons and muscle fiber segments can exceed threshold even when tissue-level heteroplasmy is low — the mosaic structure of clonal expansion underlies this discrepancy. needs-human-replication

The threshold also means that tissue-level assays substantially underdetect functionally significant heteroplasmy: a sample with 5% average mutant load may contain isolated cells or clones at 80%+.

Aging vs. germline mtDNA disease — a critical distinction

Feature	Aging heteroplasmy	Germline mtDNA disease
Distribution	Tissue-mosaic; stochastic clonal expansion	Uniform across most tissues (maternal transmission)
Variant frequency	Typically low at tissue level; high locally	Often >50% at birth; fixed in offspring
Variants	Point mutations + deletions, tissue-specific	Usually single pathogenic point mutation
Age of onset	Functional impact in post-reproductive life	Childhood–mid-adulthood for threshold-crossing variants
Heritability	Not heritable (somatic only)	Maternally inherited; complex bottleneck genetics

Germline mtDNA diseases (LHON, MERRF, MELAS, Pearson, Kearns–Sayre) result from high heteroplasmy established in the oocyte and propagated through the mtDNA bottleneck during oogenesis. This is mechanistically distinct from the stochastic somatic accumulation of aging.

Key evidence in model organisms

The POLG “mutator mouse” — expressing a proofreading-deficient form of polg — provides the clearest experimental evidence that somatic mtDNA mutations drive aging phenotypes. Trifunovic et al. (2004) showed that homozygous mutator mice accumulate ~3–5× more mtDNA mutations than wild-type and develop a progeroid syndrome (kyphosis, hair loss, anemia, cardiomyopathy) with reduced lifespan ³. Vermulst et al. (2008) used the same model to demonstrate that it is specifically clonally expanded deletions, not point mutations at steady state, that are the dominant driver of the aging phenotype ⁴.

Dimension	Status
Pathway conserved in humans?	yes
Phenotype conserved in humans?	partial — humans accumulate mtDNA mutations with age but without frank progeroid syndrome
Replicated in humans?	in-progress — human clonal expansion data from colonic crypts confirmed; causal aging role not directly established

The mutator mouse debate: whether the accelerated mutation rate in mutator mice is quantitatively comparable to normal human aging remains contested. Mutator mice accumulate ~5–10× more mutations than aged wild-type mice, which may overstate the causal role of mtDNA mutagenesis in normal human aging. contradictory-evidence

UPR^mt — cellular response to heteroplasmy stress

The mitochondrial unfolded protein response (uprmt) is activated when the stoichiometry of mitochondrial components is perturbed — a condition caused by high heteroplasmy levels that disrupt coordinated expression of nuclear- and mitochondrially-encoded OXPHOS subunits. UPR^mt upregulates mitochondrial chaperones (HSP60, HSP10, mtHSP70) and proteases (CLPP, LONP1) to maintain proteostasis. In C. elegans, UPR^mt activation extends lifespan under conditions of mild mitochondrial stress. Whether UPR^mt activation in aged human tissues with heteroplasmy is beneficial or a failing compensatory response is unresolved. no-mechanism

Therapeutic angles

mtDNA editing — selective mutant elimination

Mitochondrially targeted nucleases can be engineered to recognise and cleave mutant (but not wild-type) mtDNA sequences, shifting heteroplasmy toward wild-type.

mitoZFN (mitochondrially targeted zinc-finger nucleases): Gammage et al. (2018) demonstrated that systemically administered AAV-delivered mtZFN reduced mutant mtDNA load and restored normal OXPHOS in cardiac tissue of a mouse model carrying a pathogenic heteroplasmic mtDNA point mutation ⁵. Effect was durable and without detectable off-target editing. Currently preclinical only. needs-human-replication

mitoTALEN: A companion 2018 Nature Medicine paper (doi:10.1038/s41591-018-0166-8) used mitoTALEN delivered in the same AAV system to restore tRNA^Ala levels in a separate mouse model. Both approaches share the same limitation: delivery to non-dividing cells (especially CNS neurons) remains a major barrier. long-term-unknown

mitoARCUS: Engineered ARC nucleases (derived from I-SceI homing endonuclease scaffold) have been described for mtDNA editing in preclinical settings, with improved specificity relative to early ZFN designs. No DOI confirmed for primary publication; unsourced.

Germline prevention — maternal spindle transfer and pronuclear transfer

For prevention of germline transmission of pathogenic mtDNA variants, maternal spindle transfer and pronuclear transfer (collectively “mitochondrial replacement therapy” or “three-parent IVF”) allow the nuclear genome of a prospective mother with high heteroplasmy to be transferred into an enucleated donor oocyte with healthy mitochondria. This approach is licensed for clinical use in the UK (HFEA regulatory approval, 2016) and has resulted in live births. It addresses germline disease prevention, not somatic aging heteroplasmy.

Experimental pharmacological approaches

Cyclosporine A analogs and other agents that modulate mitochondrial membrane dynamics have been proposed to alter the selective pressure on mutant vs. wild-type mtDNA within cells. Evidence is limited to in vitro models. dose-response-unclear needs-human-replication

Rapamycin (mTOR inhibition) and caloric restriction may indirectly reduce the mutant mtDNA burden by upregulating mitophagy, which selectively eliminates depolarized mitochondria enriched for high-heteroplasmy genotypes — though this link between mitophagy and heteroplasmy reduction is indirect and not specifically established. no-mechanism

Relationship to free-radical-theory-of-aging

The free-radical (oxidative stress) theory of aging, proposed by Harman, suggested that ROS generated by OXPHOS would damage mtDNA over time, accumulating mutations that further impair OXPHOS in a vicious cycle. The heteroplasmy data partially supports this: mtDNA is indeed mutation-prone and mutations accumulate with age. However, the evidence now suggests that clonal expansion of early mutations (not continuous ROS-driven de novo mutagenesis) dominates the age-related mutant load, and the mutator mouse model argues against ROS as the primary mutagen. See free-radical-theory-of-aging for the broader theory and its contested status.

Limitations and gaps

Quantitative heteroplasmy measurements at single-cell resolution in most human tissues are technically demanding; most published aging data is from bulk tissue (underestimates mosaic structure) or from disease cohorts (not representative of normal aging). needs-human-replication
Whether tissue-average heteroplasmy levels in normal aging ever reach the threshold for OXPHOS dysfunction is contested; the mosaic model suggests isolated cells do cross threshold, but this is unproven as a generalised aging mechanism. contradictory-evidence
The causal directionality — do mtDNA mutations cause aging phenotypes in normal humans, or are they a correlate of other aging processes? — is not resolved outside the mutator mouse model. no-mechanism
The therapeutic mtDNA editing window (can mutant load be reduced without off-target effects in aged, post-mitotic tissues?) is unknown; all editing data is from young mice with engineered mutations. long-term-unknown
mtDNA editing delivery to CNS neurons remains unsolved. needs-canonical-id (mitoARCUS primary publication DOI needed)

Footnotes

doi:10.1093/nar/18.23.6927 · Cortopassi GA & Arnheim N · Nucleic Acids Research 1990 · in-vitro (human tissue) · detection of the 4,977 bp common deletion in aging human brain and muscle; showed age-dependent increase · model: human post-mortem tissues ↩
doi:10.1371/journal.pgen.1004620 · Greaves LC et al. · PLoS Genetics 2014 · observational · n=~60 human subjects across age decades · clonally expanded mtDNA point mutations increase dramatically with age in human colorectal epithelium; drift begins mid-life · model: human colonic crypt biopsies ↩
doi:10.1038/nature02517 · Trifunovic A et al. · Nature 2004 · in-vivo · POLG proofreading-deficient mice develop progeroid phenotype with ~3–5× elevated mtDNA mutagenesis and reduced lifespan; no increase in ROS · model: Mus musculus (POLG knock-in) ↩
doi:10.1038/ng.95 · Vermulst M et al. · Nature Genetics 2008 · in-vivo · clonally expanded mtDNA deletions (not point mutations at steady state) are the primary driver of premature aging phenotypes in POLG mutator mice · model: Mus musculus POLG mutator homozygotes ↩
doi:10.1038/s41591-018-0165-9 · Gammage PA et al. · Nature Medicine 2018 · in-vivo · AAV-delivered mtZFN selectively reduces mutant mtDNA load and restores OXPHOS in cardiac tissue; preclinical proof-of-concept · model: Mus musculus (m.5024C>T heteroplasmic knock-in) ↩

🧬 Aging Wiki

Explorer

mtdna-heteroplasmy