POLG (DNA polymerase gamma)

The sole mitochondrial DNA replicative polymerase in animals. POLG is the catalytic subunit of a heterotrimer responsible for copying and repairing the mitochondrial genome. Because errors in mtDNA replication accumulate over a lifetime and are non-correctable by nuclear repair machinery, POLG fidelity is a central mechanistic node linking mitochondrial-dysfunction to genomic-instability in aging. The POLG D257A “mutator mouse” — engineered to lack proofreading — is the foundational experimental model demonstrating that accumulated mtDNA mutations are sufficient to drive a multi-system premature aging phenotype.

Identity

UniProt: P54098 (DPOG1_HUMAN)
NCBI Gene: 5428
HGNC symbol: POLG
Ensembl: ENSG00000140521
Mouse ortholog: Polg (one-to-one; D257A mutator corresponds to D257A in mouse)
GenAge ID: 72 (GenAge-human curated entry)
Length: 1,239 amino acids (canonical human isoform)
Synonyms: POLG1, POLGA, MDP1

Subunit architecture

POLG functions as a heterotrimer: one catalytic subunit (POLG, this page) plus a homodimer of the accessory subunit POLG2 (UniProt Q9UHN1). The accessory dimer tethers the holoenzyme to the mitochondrial nucleoid via interactions with tfam and stimulates processivity and DNA binding affinity of the catalytic subunit. Neither subunit is active alone at physiologically relevant rates.

Subunit	Gene	UniProt	Role
Catalytic	POLG	P54098	Polymerase + 3’→5’ proofreading exonuclease
Accessory (x2)	POLG2	Q9UHN1	Processivity factor; DNA binding

Additional nucleoid-associated interactors: twnk (helicase), ssbp1 (single-strand binding protein), LIG3 (ligation). Together these form the minimal mtDNA replisome.

Biochemistry and mtDNA replication

POLG is the only polymerase responsible for replicating mtDNA in mammalian cells — there is no mtDNA-specific backup replicase. Each human cell contains 100–10,000 copies of mtDNA; all are copied by this enzyme.

Key activities:

5’→3’ polymerase (EC 2.7.7.7) — extends the nascent DNA strand using dNTPs
3’→5’ proofreading exonuclease — removes misincorporated nucleotides immediately after insertion (encoded by the Exo I/II/III motifs in the N-terminal domain)
5’-dRP lyase — participates in base-excision repair of oxidative lesions in mtDNA

The proofreading activity confers a total error rate of approximately 10^-6 to 10^-7 per base pair per replication — considerably higher than nuclear DNA polymerases (delta/epsilon reach ~10^-8 with mismatch repair), but sufficient for a genome without introns that must replicate continually in oxidizing conditions. Importantly, mtDNA has no mismatch repair system, so errors escaping proofreading are permanent unsourced — the 10^-6 figure is widely cited in reviews; primary biochemical measurement citation needed.

The POLG mutator mouse — foundational mtDNA aging model

Trifunovic 2004: proofreading-deficient mice age prematurely

Trifunovic et al. introduced the D257A knock-in — a point mutation in the exonuclease active site (Asp→Ala at position 257) that abolishes proofreading without affecting polymerase activity ¹. Homozygous knock-in mice (Polg^{D257A/D257A}) accumulate mtDNA point mutations and deletions at 3–5× higher rates than wild-type littermates and develop a multi-system premature aging syndrome detectable by 9 months of age:

Kyphosis (spinal curvature)
Alopecia and graying
Sarcopenia (severe muscle wasting by ~12 months)
Anemia
Reduced subcutaneous fat
Cardiomyopathy
Osteoporosis
Infertility (in both sexes)
Median lifespan: ~14 months (vs ~26 months WT) — approximately halved

Heterozygous (Polg^{D257A/+}) mice appeared normal in lifespan and gross phenotype, consistent with a threshold effect where the remaining proofreading capacity of one wild-type allele is sufficient.

Dimension	Status	Notes
Pathway conserved in humans?	yes	Mammalian mtDNA replication mechanism conserved; same enzyme
Phenotype conserved in humans?	partial	Human POLG disease (see below) overlaps but is not identical to mouse mutator phenotype; human aging shows mtDNA deletion accumulation in post-mitotic tissues
Replicated in humans?	no	Cannot genetically engineer human mutator; correlative evidence from POLG variant carriers and tissue aging data only

needs-human-replication — direct causal evidence that POLG-driven mtDNA mutation accumulation is a rate-limiting driver of human aging is not established. The mouse model is highly influential but the pace of mtDNA mutagenesis in WT mice is orders of magnitude higher than in WT humans.

Kujoth 2005: apoptosis is the proximate tissue-loss mechanism

In an independent parallel study using the same D257A knock-in line, Kujoth et al. reported that the premature aging tissues of mutator mice showed elevated apoptosis rather than increased oxidative stress markers ². This distinguished the mechanism of tissue decline from the then-dominant reactive-oxygen-species (ROS) model: mtDNA mutations appeared to drive tissue loss via apoptotic cell death rather than oxidative damage per se. The observation implies that even sub-lethal mtDNA mutation burdens can trigger cell-intrinsic apoptosis programs in post-mitotic tissues, depleting stem and differentiated cell pools over time.

contradictory-evidence — Subsequent work has debated whether mtDNA mutation rates in WT aged mice are high enough to cause comparable cell death, or whether the mutator mouse is a pathological rather than physiological model of aging. See mitochondrial-dysfunction for the broader ROS vs. mtDNA-mutation debate.

Disease: POLG mutation spectrum

Over 150 pathogenic POLG variants are known in humans, making it one of the most common nuclear gene causes of mitochondrial disease. Disease manifestations cluster into overlapping syndromes depending on mutation type (dominant vs recessive) and affected functional domain:

Syndrome	Inheritance	Key Features
Progressive external ophthalmoplegia (PEO)	AD or AR	Ptosis, ophthalmoparesis, proximal myopathy; mtDNA deletions in muscle
Alpers-Huttenlocher syndrome	AR	Childhood-onset hepatocerebral degeneration; mtDNA depletion
Sensory ataxic neuropathy, dysarthria, ophthalmoparesis (SANDO)	AR	Sensory neuropathy, ataxia, ophthalmoplegia
MNGIE-like	AR	Mitochondrial neurogastrointestinal encephalomyopathy phenocopy
Leigh syndrome	AR (severe)	Subacute necrotizing encephalopathy; infancy onset

Van Goethem et al. first linked dominant POLG mutations to PEO with mtDNA deletions in 2001 ³. The dominant mutations cluster around the polymerase active site; recessive mutations more often affect the exonuclease domain or interdomain linker.

Pathomechanism: loss of POLG fidelity or processivity → stalling/error-prone replication → mtDNA deletions or depletion → respiratory chain deficiency → tissue failure, with highest impact in high-energy-demand post-mitotic tissues (muscle, neurons).

Pharmacology

POLG as an off-target of antiviral nucleoside analogs

POLG is the primary toxicity mechanism for several nucleoside analog antivirals used in HIV and hepatitis B treatment:

Zalcitabine (ddC), stavudine (d4T), didanosine (ddI) — older NRTIs with high POLG affinity; associated with iatrogenic mitochondrial myopathy and neuropathy from mtDNA depletion unsourced — general toxicity mechanism is established; specific affinity data citation needed
Telbivudine, tenofovir — hepatitis B antivirals; telbivudine implicated in mtDNA depletion in hepatocytes at therapeutic doses

These represent POLG inhibition as an adverse effect, not therapeutic intention. No approved drug intentionally targets POLG.

POLG as an aging therapeutic target: current assessment

POLG is not currently an actionable drug target for aging. Reasoning:

POLG-driven mtDNA mutation accumulation in WT aging is slow and probably rate-limited by metabolic demand, not enzyme error rate
Increasing POLG fidelity pharmacologically has no validated small-molecule approach
The druggable node for mtDNA stress is more likely downstream (clearing damaged mitochondria via mitophagy; preventing apoptotic depletion via senolytic/anti-apoptotic approaches)

No ClinicalTrials.gov entries targeting POLG for aging or longevity as of 2026-05-05. unsourced — confirm with ClinicalTrials.gov search.

Cross-references

mitochondrial-dysfunction — verified hallmark page (this protein’s relevance is central)
mtdna-heteroplasmy — downstream consequence of POLG error accumulation
tfam — mitochondrial transcription factor A; co-localizes in nucleoid with POLG; R21 page
polg2 — accessory subunit; stub page needed
twnk — Twinkle helicase; replication fork partner
mus-musculus — D257A mutator mouse model; verified model-organism page
mitophagy — mtDNA quality control by organelle autophagy
sarcopenia — major phenotypic output of mutator mouse; verified phenotype page

Limitations and gaps

needs-human-replication — The causal sufficiency of POLG-driven mtDNA mutations for aging (as demonstrated in the mutator mouse) has no direct human parallel. The mutator accumulates mutations ~2 orders of magnitude faster than observed in aged human tissues.
needs-replication — Kujoth 2005 apoptosis mechanism is foundational but derived from one lab’s analysis; independent confirmation of the apoptosis-vs-ROS distinction would strengthen the claim.
no-mechanism — Why aging tissues show tissue-specific mtDNA deletion patterns (e.g., brain substantia nigra) rather than uniform accumulation is not mechanistically resolved. POLG is implicated but strand-break-based deletion models are also proposed.
needs-canonical-id — druggability-tier is null; no Open Targets Platform entry confirmed for POLG as an aging target. Recommend Open Targets lookup on next lint pass.
unsourced — POLG error rate (~10^-6) and NRTI POLG affinity comparative data: widely cited in reviews but primary biochemical measurement citations not confirmed here.

Footnotes

trifunovic-2004-polg-mutator-mouse · n not specified in DOI lookup (typical mouse study: ~10–30/group) · in-vivo · model: Polg D257A knock-in mouse (C57BL/6 background) · doi:10.1038/nature02517 · local PDF: ↩
kujoth-2005-polg-apoptosis-aging · in-vivo · model: same Polg D257A knock-in line as Trifunovic 2004 · doi:10.1126/science.1112125 · not downloaded (closed access); verify against publisher PDF ↩
vangoethem-2001-polg-peo · observational (human genetics) · n=4 families (original report) · model: human PEO kindreds · doi:10.1038/90034 · local PDF: ↩

🧬 Aging Wiki

Explorer

polg