Advancedadvanced biology

DNA Polymerase Fidelity: Proofreading and Precision

Learning objectives

  • Explain the induced-fit mechanism of correct dNTP selection.
  • Describe 3'-to-5' exonuclease proofreading and when the primer shuttles.
  • Evaluate the contribution of processivity clamps (beta, PCNA) to fidelity.
  • Compare error rates across polymerase tiers and the mismatch-repair backstop.
  • Relate fidelity defects to mutator phenotypes and cancer.

Selecting the right nucleotide: induced fit

A DNA polymerase faces a brutal discrimination problem: it must pick the correct deoxynucleotide triphosphate from four lookalikes, thousands of times per second, in a crowded nucleus. The first filter is geometry. When a dNTP enters the active site, the polymerase's fingers domain closes -- but only a correct Watson-Crick base pair has the right shape to fit snugly against the template base and position the primer's 3'-OH for attack on the alpha phosphate. An incorrect dNTP distorts the active site, misaligning the chemistry and slowing incorporation dramatically; the enzyme then tends to release the wrong nucleotide rather than extend it. This induced-fit step -- conformational change gated by correct geometry -- contributes the largest share of selectivity, far more than the small difference in base-pairing energy alone. Kinetic studies show the discrimination is mostly in the forward rate of chemistry after binding: right nucleotide, fast closure and fast chemistry; wrong nucleotide, a distorted complex that stalls.

Proofreading: the 3'-to-5' exonuclease

When a wrong nucleotide does slip through, replicative polymerases get a second chance. Their 3'-to-5' exonuclease domain -- a separate active site -- excises mismatched primer termini. The mechanism is a physical shuttling: a mismatch at the primer end frays the duplex, stalling the polymerase active site, and the primer terminus swings into the exonuclease site, where the incorrect nucleotide is clipped off. Synthesis then resumes. This kinetic partitioning -- extend versus excise -- is exquisitely sensitive to the mismatch: a correct terminus extends rapidly and never visits the exonuclease site, while a mismatch pauses extension long enough for excision to win. Proofreading-deficient polymerase mutants (exonuclease-dead) show elevated mutation rates by one to two orders of magnitude, demonstrating that this single domain contributes roughly a hundredfold to fidelity. Some polymerases, like the translesion specialists that copy past DNA damage, lack proofreading entirely -- trading accuracy for the ability to keep going, a deliberate division of labor.

Staying on: processivity clamps

Fidelity is not only about choosing bases; it is about not falling off. Replicative polymerases are tethered to DNA by ring-shaped sliding clamps -- the beta clamp in bacteria, PCNA in eukaryotes and archaea -- that encircle the duplex and slide freely along it. Clamp loaders (complexes like RFC) use ATP to crack the ring open and seat it at primer-template junctions. Once loaded, the clamp raises processivity from tens of nucleotides to tens of thousands, which matters for fidelity because dissociation and reassociation at a primer terminus creates opportunities for misalignment and strand slippage, especially in repetitive sequences. The clamp also serves as a mobile platform, coordinating the polymerase with other replisome components -- and, tellingly, with mismatch repair proteins, physically linking synthesis to its quality-control backstop. Processivity, speed, and accuracy are thus engineered together, not separately.

Three tiers of error correction

Replication fidelity is a layered system with roughly multiplicative gains. Base selection by induced fit gets the error rate to about one in 10^4-10^5; proofreading excision improves it another hundredfold to about one in 10^6-10^7; post-replicative mismatch repair -- MutS homologs recognizing the mismatch, MutL homologs directing excision of the new strand -- catches most of what remains, reaching overall error rates near one in 10^9-10^10 per base per division. Each tier catches a different failure mode: selection prevents most misinsertions, proofreading removes misinsertions immediately, and mismatch repair fixes the escapees plus errors selection cannot address, like small insertion-deletion loops from polymerase slippage. The strand-discrimination problem -- how repair knows which strand is new -- is solved differently across domains: bacteria use transiently unmethylated GATC sites, eukaryotes use strand discontinuities like nicks and the replication machinery itself. Losing any tier raises the mutation rate and, in humans, cancer risk.

When fidelity fails: mutators and disease

The cost of fidelity defects is written in genomes. Proofreading-deficient mutations in the human replicative polymerases POLE and POLD1 cause ultramutated cancers -- tumors with staggering mutation burdens -- and germline variants predispose to colorectal cancer and polyposis. Defective mismatch repair causes Lynch syndrome, with its characteristic microsatellite instability as slippage errors go uncorrected. Conversely, the immune system exploits controlled infidelity: activation-induced deaminase deliberately mutates antibody genes during somatic hypermutation, and translesion polymerases deliberately misincorporate across from lesions to keep replication alive. Fidelity is therefore not maximized everywhere but optimized per context -- maximal at the replication fork, deliberately relaxed where diversity or survival demands it. Understanding these tiers also underlies practical tools: high-fidelity PCR enzymes are proofreading-proficient archaeal polymerases, and their error rates are quoted precisely because the same three-tier logic applies in the test tube.

Multiplicative quality control

Base selection (~10^-4 to 10^-5), proofreading (~100x improvement), and mismatch repair (~100-1000x) multiply to reach overall error rates near 10^-9 to 10^-10 per base -- each tier catching what the last missed.

Key points

  • Induced fit: only a correct base pair positions the 3'-OH for chemistry; wrong dNTPs stall the closed complex and dissociate.
  • The 3'-to-5' exonuclease excises mismatches after the primer frays and shuttles to the proofreading site, contributing ~100x fidelity.
  • Sliding clamps (beta, PCNA) tether polymerases for processive synthesis and coordinate mismatch repair.
  • Three tiers -- selection, proofreading, mismatch repair -- multiply to overall error rates near one in 10^9-10^10 per base.
  • POLE/POLD1 proofreading defects cause ultramutated cancers; mismatch-repair loss causes Lynch syndrome.
  • Fidelity is optimized per context: translesion polymerases and somatic hypermutation deliberately trade accuracy for survival or diversity.

Common misconceptions

✗ Myth: Polymerases never make mistakes.

✓ Truth: Base selection alone errs about once per 10^4-10^5 insertions. Near-perfect replication requires proofreading plus mismatch repair on top.

✗ Myth: Proofreading happens before the wrong base is added.

✓ Truth: Proofreading is post-insertion: the mismatched primer terminus is excised by the 3'-to-5' exonuclease after incorporation, then synthesis resumes.

✗ Myth: Higher fidelity is always better.

✓ Truth: Translesion polymerases sacrifice fidelity to replicate past damage, and the immune system deliberately mutates antibody genes -- accuracy is tuned to context.

Frequently asked questions

How does a polymerase choose the correct dNTP?

Induced fit: only a correct Watson-Crick pair fits the closed active site and aligns the 3'-OH for attack. Wrong nucleotides distort the complex, slowing chemistry so they dissociate instead.

What does the 3'-to-5' exonuclease do?

It excises a just-added mismatched nucleotide. The frayed primer end shuttles from the polymerase site to the exonuclease site, the error is clipped off, and synthesis resumes.

What is PCNA?

The eukaryotic sliding clamp -- a ring encircling DNA that tethers the polymerase for processive synthesis and serves as a platform coordinating repair proteins.

How do the three fidelity tiers combine?

Multiplicatively: selection (~10^-4 to 10^-5), proofreading (~100x better), mismatch repair (~100-1000x better) reach roughly 10^-9 to 10^-10 errors per base per division.

What happens when proofreading is defective?

Mutations in POLE/POLD1 exonuclease domains cause ultramutated cancers; germline defects predispose to colorectal tumors -- proof that one domain guards the genome.