DNA Polymerase

Why DNA polymerase matters

• It is the cell's only general-purpose DNA copier. Every dividing cell, every viral replication cycle, every PCR reaction depends on a polymerase. The human genome's 6.4 billion bp duplicates in ~8 hours of S-phase, requiring roughly 10¹⁴ phosphodiester bonds formed across thousands of forks running in parallel.
• Its fidelity sets the mutation rate of life. Replicative polymerases plus mismatch repair give a per-base-pair mutation rate of ~10^-9-10^-10. Drift, selection, and human cancer all begin in the rare polymerase mistakes that escape proofreading and MMR.
• The two-metal-ion mechanism unifies enzymology. Thomas Steitz's 1993 crystal structures of Klenow fragment, HIV reverse transcriptase, and T7 RNA polymerase all converged on the same active-site geometry. The mechanism extends to RNA polymerases, group I/II introns, RNase H, and the catalytic core of the spliceosome.
• Pol III speed is one of biology's most studied enzyme rates. E. coli Pol III holoenzyme catalyzes ~1000 nt/s with k_cat ≈ 1000 s^-1 per active site, near the diffusion limit for dNTP binding. The reaction is fast enough that every base pair takes ~1 ms.
• It is the target of major antiviral and anticancer drugs. AZT, tenofovir, acyclovir, and remdesivir are all chain-terminating nucleoside analogs that block viral polymerases. Cisplatin produces lesions that stall replicative polymerases, killing dividing cancer cells. Cytarabine (Ara-C) is incorporated by Pol α and stalls leukemia cells.
• Taq polymerase enabled PCR. Kary Mullis's 1985 PCR concept needed a thermostable polymerase. Taq (Thermus aquaticus, isolated from Yellowstone hot springs by Thomas Brock) survives 95 °C denaturation cycles; one billion-fold amplifications became routine, and biotechnology was born.
• Translesion polymerases let cells survive damage. Y-family enzymes (Pol η, ι, κ, Rev1) tolerate UV photoproducts, abasic sites, and bulky adducts at the cost of higher error rates (10^-2 to 10^-4). Loss of Pol η causes xeroderma pigmentosum variant — UV-driven skin cancer at very young ages.

Common misconceptions

• Polymerase reads DNA both directions. No. Synthesis is strictly 5' to 3' on the new strand, which means reading the template 3' to 5'. Every polymerase ever characterized obeys this — there is no known reverse-direction DNA polymerase.
• Pol I replicates the genome. Pol I, the first polymerase isolated (Arthur Kornberg, 1956), does primer removal and gap filling, not bulk replication. Pol III holoenzyme is the actual replicative enzyme in E. coli — discovered by Thomas Kornberg (Arthur's son) in 1970.
• Eukaryotic polymerases are slower because they are weaker. They are slower per fork (~50 vs 1000 nt/s) but the eukaryotic genome is 1000x larger, so cells run thousands of forks in parallel. Total replication throughput per cell is comparable when normalized to genome size.
• Proofreading and mismatch repair are the same thing. Proofreading is a 3' to 5' exonuclease activity within the polymerase that catches errors during synthesis (within 1-2 bases). Mismatch repair (MMR) is a separate post-replicative system using MutS/MutL homologs that scans the daughter strand minutes to hours later and patches mismatches.
• All polymerases need Mg²⁺. The two-metal-ion mechanism prefers Mg²⁺ in vivo, but Mn²⁺ works (with reduced fidelity) and some translesion polymerases use either. Crystal structures often use higher-Z metals like Ca²⁺ or Zn²⁺ as inactive substitutes to trap pre-catalytic complexes.
• Polymerase is a single protein. Pol III holoenzyme is a 17-subunit complex (~900 kDa) including the catalytic α, the ε proofreader, the θ subunit, plus the β clamp and γ clamp loader. Eukaryotic Pol δ and Pol ε are 4-subunit assemblies, and the full replisome adds another dozen factors.

How a single nucleotide is added

The cycle begins with the polymerase bound to a primer-template, with the 3' end of the primer in the catalytic site. A dNTP diffuses in from solution and base-pairs (or fails to base-pair) with the next template nucleotide. If correct, the polymerase fingers domain rotates inward by 30-40° — the open-to-closed transition described by Lorena Beese and Tom Steitz from Klenow fragment crystal structures. This conformational change brings the catalytic aspartates and the dNTP triphosphate into alignment with the primer's 3'-OH and the two Mg²⁺ ions. The 3'-OH attacks the α-phosphate, displacing pyrophosphate (PPi) and forming the new phosphodiester bond. The fingers domain reopens, the polymerase translocates one nucleotide along the template, and the cycle repeats. Each addition consumes ~1 ms in Pol III at saturating substrate.

If a mismatch slips past the geometric checkpoint, the mispaired 3' end fits poorly into the active site, kinks the primer-template, and increases the off-rate. The primer-template is shuttled ~30 angstroms to the 3' to 5' exonuclease site, where the mispaired terminal base is hydrolyzed off (one of the few hydrolytic, not synthetic, activities in the polymerase). The polymerase then attempts the addition again. Proofreading roughly doubles the per-nucleotide free-energy cost and lowers error rates ~100-1000-fold below the geometric baseline. Post-replicative mismatch repair, run by MutS-MutL homologs scanning for mismatches by hemi-methylated GATC strand-discrimination signals (bacteria) or ribonucleotide marks (eukaryotes), removes another 99% of remaining errors. The combined system gives genomic fidelity that is essentially unmatched in any other engineered or evolved copying machinery.

Polymerase family comparison

Polymerase	Organism / role	Speed	Fidelity (error rate)	Proofreading?	Distinctive feature
Pol I	E. coli; primer removal + repair	~20 nt/s	10-6	Yes (3'-5')	Has 5'-3' exo for nick translation
Pol III	E. coli; replicative	~1000 nt/s	10-7	Yes (ε subunit)	17-subunit holoenzyme, dual replisome
Pol α-primase	Eukaryotic; primer synthesis	~20 nt/s	10-4	No	Only enzyme that synthesizes RNA primers
Pol δ	Eukaryotic; lagging strand + repair	~50 nt/s	10-7	Yes	Loads on PCNA, strand-displacement for FEN1
Pol ε	Eukaryotic; leading strand	~50 nt/s	10-7	Yes	Identified as leading-strand specific by Kunkel 2007
Pol γ	Mitochondrial replication	~70 nt/s	10-6	Yes	Sole mtDNA polymerase; mutations cause MELAS, Alpers
Pol β	Base-excision repair gap filling	~10 nt/s	10-3-10-4	No	Smallest replicative polymerase, 39 kDa
Pol η	Translesion across UV dimers	~5 nt/s	10-2-10-3	No	Loss causes xeroderma pigmentosum variant
Taq polymerase	Thermus aquaticus; PCR	~50 nt/s at 72 °C	10-5	No	Stable at 95 °C; foundation of PCR
Phi29 polymerase	Bacteriophage phi29; isothermal amp	~50 nt/s	10-6	Yes	~70 kb processivity, used for whole-genome amplification

Famous experiments

• Arthur Kornberg, 1956 (J Biol Chem). Purified DNA polymerase from E. coli (now called Pol I) and demonstrated template-directed DNA synthesis in vitro. Won the Nobel Prize in 1959. Severo Ochoa shared it for RNA polymerase work.
• Thomas Kornberg & Malcolm Gefter, 1970-1972. Showed Pol I null mutants of E. coli (polA1) still grow, identified Pol II and then Pol III as the actual replicative enzyme. Resolved the embarrassing mystery of why losing Kornberg's polymerase didn't kill the cell.
• Steitz lab, 1985-1995. Solved the crystal structures of Klenow fragment and HIV reverse transcriptase, defined the right-hand polymerase fold (palm, fingers, thumb), and proposed the two-metal-ion mechanism that turned out to apply across all nucleic acid polymerases.
• Kary Mullis, 1985. Conceived PCR while driving on Highway 128 in California. Replacing the heat-labile Klenow fragment with thermostable Taq polymerase (Saiki et al., 1988) made the technique practical. Mullis won the 1993 Nobel.
• Tom Kunkel, 2007 (Nature). Used catalytic-domain mutations that increase ribonucleotide misincorporation in Pol ε and Pol δ separately to map which polymerase replicates which strand in S. cerevisiae. Concluded Pol ε is the dedicated leading-strand polymerase, Pol δ the lagging.