Mitochondrial DNA (mtDNA) is the DNA contained within a mitochondrion, the organelle responsible for much of the energy conversion that sustains most eukaryotic cells. It constitutes a small but functionally indispensable part of the cellular genome, distinct from the DNA of the cell nucleus. In humans, it is ordinarily a circular molecule containing 16,569 base pairs and 37 genes. Across other organisms, mitochondrial genomes vary greatly in size, organization, and gene content, and some are linear or divided among multiple molecules.
Mitochondrial DNA encodes components required for oxidative phosphorylation, the process that couples nutrient oxidation to the production of adenosine triphosphate (ATP). Its maintenance and expression nevertheless depend overwhelmingly on proteins encoded in the nuclear genome. This division of genetic responsibility reflects the bacterial ancestry of mitochondria and requires close coordination between two genomes.
Its predominantly maternal inheritance in humans, high copy number, and distinctive patterns of variation make mtDNA important in medical genetics, evolutionary biology, archaeology, and forensic identification. Understanding it requires distinguishing the properties of individual DNA molecules from those of the many copies distributed among cells, tissues, organisms, and populations.
Discovery and evolutionary origin
The discovery of mitochondrial DNA established that hereditary information in eukaryotic cells is not confined to nuclear chromosomes. In 1963, Margit Nass and Sylvan Nass reported electron-microscopic evidence of DNA-containing structures inside chick-embryo mitochondria. The structures were sensitive to treatment with a DNA-degrading enzyme, supporting their identification as DNA. Biochemical isolation and characterization subsequently confirmed that mitochondria possess their own genetic material. Earlier work on non-Mendelian inheritance in organisms such as yeast had already indicated the existence of hereditary determinants outside the nucleus.
A major milestone came in 1981, when Frederick Sanger’s group and collaborators published the complete human mitochondrial genome sequence. Its small size and unusually compact organization made it one of the earliest completely sequenced human genetic systems. The sequence provided a framework for identifying mitochondrial genes, studying population variation, and locating disease-associated changes. The revised Cambridge Reference Sequence, published in 1999, remains a widely used coordinate reference for human mtDNA. A reference sequence is an indexing standard, not a universal ancestral sequence or a definition of a healthy genome.
The existence of mtDNA strongly supports the endosymbiotic theory of mitochondrial origin. Under this model, an ancestral host cell incorporated a bacterium that became a permanent intracellular partner. Evidence includes similarities between mitochondrial and bacterial translation systems, bacterial affinities of many mitochondrial genes, and the evolutionary relationships reconstructed from sequence data. The mitochondrial ancestor belonged to, or was closely related to, the bacterial lineage containing modern alphaproteobacteria; its exact position relative to living groups remains an active subject of research.
During evolution, most genes of the ancestral endosymbiont were lost or transferred to the host nucleus. Proteins encoded by transferred genes could subsequently be imported back into the organelle. This process of endosymbiotic gene transfer explains why modern mitochondria retain only a small genome while depending on a much larger set of nuclear genes. Human mitochondrial function involves more than a thousand proteins, whereas human mtDNA encodes only 13 proteins.
Why mitochondria retained any genes is not explained by a single universally sufficient mechanism. Proposed factors include the difficulty of importing certain highly hydrophobic membrane proteins and advantages of regulating respiratory components close to their site of action. These explanations are not mutually exclusive. Some mitochondrion-derived organelles have lost their genomes entirely, demonstrating that DNA retention depends on the organelle’s remaining functions rather than on its evolutionary origin alone.
Genome architecture and diversity
The human mitochondrial genome is a double-stranded, usually circular DNA molecule. Its two strands are conventionally called heavy and light strands because their different nucleotide compositions produce different buoyant densities. The genome contains 13 protein-coding genes, 22 transfer RNA genes, and two ribosomal RNA genes. Most genes are encoded on the heavy strand; the light strand encodes the ND6 protein and eight transfer RNAs. Gene names such as MT-ND1 and MT-CO1 identify specific mitochondrial loci.
Human mtDNA is exceptionally compact. It has very little noncoding sequence between genes, lacks introns, and includes overlapping coding regions. Its largest noncoding segment is the control region, which contains important regulatory sequences involved in replication and transcription. The term D-loop is often used informally for this region, but more precisely denotes a displacement-loop structure formed when a short third DNA strand, called 7S DNA, associates with part of the molecule. The control region and the physical D-loop are therefore related but not identical concepts.
DNA is packaged into mitochondrial nucleoids, protein–DNA assemblies located in the matrix and frequently associated with the inner mitochondrial membrane. The nuclear-encoded protein TFAM binds and bends mtDNA, contributing to its compaction and organization. Mammalian nucleoids often contain approximately one genome copy, although measurements depend on the organism and experimental method. They are not miniature nuclear chromosomes: they lack the conventional histone-based nucleosomes characteristic of nuclear chromatin.
Genome copy number varies greatly among cell types and physiological states. A nucleated human cell commonly contains hundreds to thousands of mtDNA copies, whereas a mature oocyte can contain more than 100,000. Mature mammalian erythrocytes have neither mitochondria nor mtDNA. Copy number does not translate directly into a fixed number of organelles because mitochondria continually fuse and divide, and an individual mitochondrial compartment may contain several nucleoids.
The familiar human arrangement is not universal. Many animal mitochondrial genomes are similarly compact, but plant mitochondrial genomes commonly extend to hundreds of thousands of base pairs and contain repeats, introns, and substantial noncoding DNA. Their physical organization can involve branched, linear, and recombining molecules rather than one stable circular chromosome. Fungal mitochondrial genomes also vary widely in size and intron content. Some protists possess fragmented genomes or unusual collections of DNA circles; kinetoplastid organisms, including trypanosomes, contain elaborate networks of maxicircles and minicircles. Such diversity makes “a small circular chromosome” a useful description of human mtDNA, but an inadequate definition of mitochondrial genomes generally.
Gene expression and respiratory function
Mitochondrial genes contribute directly to the machinery of cellular respiration. In humans, seven mtDNA-encoded proteins are subunits of respiratory complex I, one belongs to complex III, three belong to complex IV, and two belong to ATP synthase, also called complex V. Complex II is entirely nuclear-encoded. The mitochondrial products are largely hydrophobic components embedded in the inner membrane, where they work with many nuclear-encoded subunits.
The electron transport chain transfers electrons through a succession of redox reactions. Complexes I, III, and IV use part of the released energy to move protons across the inner mitochondrial membrane. The resulting electrochemical gradient powers ATP synthase. Mitochondrial genes therefore encode only a small fraction of the relevant proteins, but defects in their products can compromise the assembly, stability, or operation of entire respiratory complexes.
Human mitochondrial transcription is performed by a dedicated, nuclear-encoded RNA polymerase, POLRMT, assisted by transcription factors including TFAM and TFB2M. Transcription produces long RNA molecules containing sequences for multiple genes. Many transfer RNA sequences lie between protein-coding or ribosomal RNA sequences. Their recognition and removal by RNA-processing enzymes help release the neighboring mature RNAs, an arrangement known as the tRNA punctuation model. Some transcript boundaries require additional processing because they do not conform to this pattern.
Mitochondrial translation occurs on specialized ribosomes, or mitoribosomes. Mammalian mitoribosomes contain the two mtDNA-encoded ribosomal RNAs together with many nuclear-encoded proteins. Their structure differs substantially from both bacterial ribosomes and the ribosomes in the cytosol. The 22 human mitochondrial transfer RNAs support translation through modified decoding rules and flexible recognition of certain codons.
The mitochondrial genetic code also differs from the standard nuclear code. In vertebrate mitochondria, for example, UGA specifies tryptophan rather than acting as an ordinary stop codon, and AUA can specify methionine rather than isoleucine. Mitochondrial codes differ among evolutionary lineages, so vertebrate rules cannot be applied indiscriminately to fungi, plants, or protists. Several human mitochondrial transcripts acquire complete stop codons through polyadenylation, illustrating how RNA processing contributes directly to the final coding message.
Expression must be coordinated across cellular compartments. Nuclear genes supply replication enzymes, transcription factors, ribosomal proteins, respiratory subunits, assembly factors, and protein-import machinery. Mitochondrial gene products must be synthesized and assembled in appropriate proportions with these imported components. Consequently, a defect in a nuclear gene can impair mitochondrial translation or respiration even when the mtDNA sequence itself is unchanged.
Replication, repair, and genome maintenance
Mitochondrial DNA replication is not restricted to the nuclear S phase of the cell cycle. Copies can be synthesized as cells grow, change metabolic state, or replace damaged organelles. The principal replicative enzyme in human mitochondria is DNA polymerase gamma, whose catalytic subunit is encoded by POLG. It works with an accessory subunit, the TWINKLE helicase, mitochondrial single-stranded DNA-binding protein, and other maintenance factors. Its proofreading activity helps limit replication errors.
Several models describe how the two strands are copied. In the classical strand-displacement model, synthesis begins near the heavy-strand origin, and replication exposes a separate light-strand origin after progressing around part of the molecule. Other observations support replication intermediates in which RNA associates extensively with a displaced template strand, often discussed under the RITOLS model, meaning RNA incorporation throughout the lagging strand. Strand-coupled DNA synthesis has also been observed. These models describe different features or populations of intermediates; their relative importance can depend on tissue, physiological conditions, and experimental preparation.
Human mtDNA is subject to base damage, replication errors, strand breaks, and deletions. Its proximity to respiratory activity once encouraged the view that reactive oxygen species were the dominant source of mitochondrial mutations. More recent analyses show that many somatic point-mutation patterns are more consistent with replication-associated errors and spontaneous base changes. Oxidative damage remains biologically important, but damage to a DNA molecule is not equivalent to a heritable sequence mutation.
Mitochondria possess DNA repair pathways, particularly base-excision repair, which removes certain damaged bases and repairs the resulting gaps. Their repair repertoire is nevertheless different from that of the nucleus. Canonical nucleotide-excision repair, which removes bulky lesions from nuclear DNA, has not been established as a comparable general pathway in mammalian mitochondria. Severely damaged mtDNA molecules may instead be degraded, allowing remaining intact copies to sustain or restore the genome population.
Genome maintenance also involves events above the level of individual DNA molecules. Mitochondrial fusion permits exchange of many organelle contents, while fission can help segregate dysfunctional components. Mitophagy, the selective removal of mitochondria through cellular degradation pathways, contributes to organelle quality control. These processes do not guarantee elimination of every harmful mtDNA variant: the relationship between a genome’s sequence and the function of its surrounding mitochondrial compartment can be indirect.
Nuclear defects can produce mtDNA depletion, multiple deletions, or abnormal nucleotide supplies. POLG-associated disorders and TWINKLE-related disorders illustrate how failure of a nuclear-encoded maintenance protein destabilizes an otherwise separate genome. Genome integrity therefore depends on the coordination of DNA synthesis, nucleotide metabolism, repair, molecular turnover, and organelle dynamics.
Maternal inheritance, heteroplasmy, and segregation
In humans, mtDNA is ordinarily inherited from the mother. The oocyte supplies nearly all the cytoplasm of the fertilized egg, whereas sperm mitochondria generally do not contribute a lasting mitochondrial lineage to the embryo. This inheritance pattern differs from that of nuclear autosomes, for which offspring normally receive one copy from each parent. Maternal inheritance is widespread among animals, but it is not universal across eukaryotes: paternal, biparental, and specialized inheritance systems occur in other groups.
Two terms describe the composition of a mitochondrial genome population. Homoplasmy means that the copies considered at a particular position have the same sequence. Heteroplasmy means that more than one sequence variant is present. Heteroplasmy can occur within a cell, among cells in a tissue, or among tissues of one individual. Its measured level depends on the sampled material and the sensitivity of the assay; a specimen that appears homoplasmic by one method may contain low-frequency variants detectable by another.
For a particular variant, its heteroplasmy fraction can be expressed as
where the quantities represent the numbers of mtDNA molecules carrying the variant and the alternative sequence in the sample. This fraction does not by itself determine physiological effect. The identity of the variant, its distribution among cells, total mtDNA copy number, nuclear genetic background, and tissue energy demands also matter.
During cell proliferation and development, mtDNA copies are replicated and partitioned. Random sampling, together with differences in replication or survival, can alter variant proportions. This process is often called replicative segregation. Unlike nuclear chromosomes, mitochondrial genomes do not undergo the same orderly pairing and segregation at each mitotic division. A variant can consequently become abundant in one cellular lineage and remain rare in another.
A particularly important restriction and redistribution of mitochondrial genetic diversity occurs during female germline development. The mitochondrial genetic bottleneck reduces the effective number of genomes that determine transmission to the next generation. Its operation involves developmental changes in copy number, sampling of genome populations, and selective processes; it is not simply a single reduction to an invariant number of molecules. As a result, children of the same mother can inherit markedly different heteroplasmy levels.
Genetic drift and selection both influence transmission. Some damaging variants are preferentially removed during germline development, but selection is neither complete nor equally effective against all variants. Reported exceptions to strictly maternal inheritance in humans require careful interpretation because technical artifacts, sample mixtures, and nuclear copies of mitochondrial sequences can mimic unexpected transmission. Predominantly maternal inheritance remains the established framework for ordinary human mtDNA transmission.
Mitochondrial DNA and human disease
Pathogenic mtDNA variants can impair oxidative phosphorylation and cause disease affecting one organ or multiple systems. The brain, skeletal muscle, heart, retina, inner ear, and endocrine tissues are frequently involved because of their functional dependence on mitochondrial metabolism. Nevertheless, mitochondrial disease is broader than disease caused by mtDNA: many disorders arise from nuclear variants affecting respiratory proteins, mitochondrial translation, maintenance of mtDNA, or related metabolic pathways.
Clinical effects often depend on a threshold phenomenon. A tissue may retain sufficient respiratory function while the proportion of a pathogenic variant remains below a certain level, then become dysfunctional as that level rises. There is no universal threshold percentage. Different variants have different biochemical effects, and individual cells may differ substantially from the average measured in a tissue sample. Some pathogenic variants cause disease in a homoplasmic state, showing that heteroplasmy is not required for mitochondrial pathology.
Several well-characterized disorders illustrate these principles. MELAS, named for mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes, is frequently associated with the m.3243A>G variant in MT-TL1, a mitochondrial transfer RNA gene. The same variant can also be associated with other presentations, including maternally inherited diabetes and deafness. The m.8344A>G variant in MT-TK is a common cause of MERRF, or myoclonic epilepsy with ragged-red fibers. These associations are not one-to-one rules: a single variant can produce different manifestations.
Leber hereditary optic neuropathy commonly involves variants in genes encoding complex I subunits, particularly MT-ND1, MT-ND4, and MT-ND6. It primarily affects retinal ganglion cells and central vision, and it has incomplete penetrance even in families carrying a homoplasmic variant. Large-scale mtDNA deletions are associated with disorders including Pearson syndrome, Kearns–Sayre syndrome, and progressive external ophthalmoplegia. The clinical pattern depends partly on which tissues contain the deletion and on how its abundance changes over time.
Molecular investigation must account for tissue distribution. A variant present at a high level in muscle or urinary epithelial cells may be less abundant in blood, and some pathogenic variants decline in blood with age. Sequence analysis can identify point variants and small insertions or deletions; specialized analyses assess large rearrangements and copy-number depletion. Detection of a sequence difference alone does not establish pathogenicity. Interpretation also considers inheritance, population frequency, functional evidence, clinical features, and heteroplasmy.
Changes in mtDNA also occur in cancers and other common diseases, but association does not automatically establish a causal role. Tumors may accumulate mitochondrial variants during their clonal history, and some variants can alter metabolism without being the initiating cause. Distinguishing inherited pathogenic variants, acquired functional changes, and neutral passengers is central to interpreting mitochondrial findings in medicine.
Population history, species relationships, and identification
Mitochondrial DNA became an influential tool in population genetics because it is abundant, often recoverable from degraded samples, and inherited largely as a single linked unit in humans. Groups of related mitochondrial sequences are classified into haplogroups, defined by shared variants inherited from common maternal ancestors. Their geographical distributions provide evidence about population history, migration, and demographic change, although any one person’s mtDNA represents only one maternal line among many genealogical ancestors.
A landmark 1987 study by Rebecca Cann, Mark Stoneking, and Allan Wilson used mitochondrial variation to investigate modern human origins. Subsequent work with larger datasets and complete mitochondrial genomes refined the maternal genealogy and supported a recent African origin of living human mitochondrial lineages. The expression “mitochondrial Eve” refers to the most recent woman from whom all present-day human mtDNA lineages descend. It does not mean that she was the only woman alive at the time, the first anatomically modern human, or the source of all human genetic ancestry.
Mitochondrial sequence comparisons are also widely used in phylogenetics. In animals, a region of the cytochrome c oxidase I gene is a common marker for DNA barcoding, which compares standardized sequences to help identify specimens. Mitochondrial markers can reveal otherwise difficult-to-distinguish lineages, but they do not always coincide with species boundaries. Introgression following hybridization, incomplete lineage sorting, selection, and sex-biased dispersal can produce a mitochondrial history different from that inferred from nuclear loci.
The use of mtDNA as a molecular clock requires calibration and explicit assumptions. Mutation rates estimated from family transmission can differ from substitution rates inferred over long evolutionary intervals because selection and loss of rare variants affect which changes persist. Different parts of the genome evolve at different rates, and repeated mutations at the same position can obscure relationships. Reliable historical inference therefore combines appropriate evolutionary models with archaeological, fossil, geographical, or independent genetic evidence.
In forensic genetics, mtDNA is especially useful for hair shafts, old bones, teeth, and other material with little recoverable nuclear DNA. Its high copy number increases the chance of obtaining a sequence. A match can support association with a maternal lineage, but it ordinarily cannot distinguish close maternal relatives and is not equivalent to the individualizing power of a high-quality nuclear DNA profile. Statistical interpretation depends on sequence frequency, database composition, laboratory quality controls, and the possibility of heteroplasmy.
Research on ancient DNA initially relied heavily on mitochondrial sequences because of their relative abundance. Complete ancient mitochondrial genomes have helped reconstruct relationships among extinct and living populations. Nuclear genome analysis now supplies a much broader picture, but mtDNA remains valuable when preservation is poor. Contamination, postmortem chemical damage, and the distinction between maternal ancestry and total ancestry remain essential considerations.
Somatic variation, immune signaling, and experimental manipulation
Mitochondrial genomes continue to change after development. Somatic mutations can arise in individual cells and expand clonally, producing patches of tissue in which a particular variant becomes abundant. In aging skeletal muscle, for example, some fibers accumulate high levels of mtDNA deletions together with respiratory-chain defects. Related patterns occur in other tissues, although the amount and biological significance of somatic variation differ among cell types and individuals.
These observations helped motivate mitochondrial theories of aging. An influential early proposal held that respiratory activity generates reactive oxygen species, which damage mtDNA and create a self-amplifying cycle of respiratory failure. Experimental results support roles for mitochondrial dysfunction in aging but do not establish this simple cycle as a universal explanation. Mice carrying proofreading-deficient mitochondrial DNA polymerase accumulate mutations and develop several premature-aging features; interpreting such models requires distinguishing artificially elevated mutation burdens from the processes operating during ordinary aging.
Mitochondrial DNA also acts as a signal when it appears outside its normal compartment. Release into the cytosol can activate the cGAS–STING pathway, which detects misplaced DNA and induces inflammatory and interferon responses. Mitochondrial DNA can also engage TLR9, an endosomal DNA-sensing receptor. These mechanisms connect mitochondrial stress with innate immunity, but the effects depend on how DNA is released, chemically modified, transported, and cleared. Detecting circulating cell-free mtDNA is therefore not equivalent to identifying one specific disease mechanism.
Experimental analysis has been complicated by nuclear mitochondrial DNA segments, or NUMTs, which are fragments of mitochondrial origin inserted into nuclear chromosomes. These sequences can be mistakenly amplified or aligned as mtDNA, particularly in short-read sequencing or samples with low mitochondrial copy number. Modern analyses use combinations of mapping strategies, long reads, molecular enrichment, and independent validation to distinguish authentic mitochondrial variants from nuclear copies and technical errors.
Manipulating mtDNA presents challenges because delivery into mitochondria differs from delivery into the nucleus. Sequence-specific nucleases, including mitochondrially targeted TALENs, can preferentially destroy selected mtDNA molecules and thereby shift heteroplasmy toward remaining genomes. Such approaches generally change the composition of a genome population rather than repair each targeted molecule. Mitochondrial base editors developed from bacterial deaminases have enabled particular nucleotide conversions without requiring the conventional guide-RNA import used by many nuclear editing systems. Their sequence constraints, unintended changes, delivery, and long-term effects remain important research questions.
Mitochondrial replacement uses reproductive techniques to combine parental nuclear genetic material with mitochondria supplied by a donor oocyte. It is distinct from mtDNA sequence editing. Techniques such as maternal spindle transfer and pronuclear transfer aim to reduce transmission of pathogenic maternal mtDNA, but small amounts of the original mtDNA may be carried over and can change in proportion during development. Its clinical use is subject to jurisdiction-specific regulation, while research examines developmental outcomes, residual heteroplasmy, and interactions between mitochondrial and nuclear genetic backgrounds.