DNA proofreading is the correction of newly incorporated errors during DNA replication, usually by a 3′→5′ exonuclease associated with a DNA polymerase. When the polymerase adds an incorrect nucleotide to a growing DNA strand, the mismatched end can be transferred to an editing site, where the incorrect nucleotide is removed before synthesis resumes. Proofreading is a major contributor to the accuracy of genome duplication, acting alongside nucleotide selection and DNA mismatch repair to prevent replication errors from becoming permanent mutations.
Molecular mechanism
DNA polymerases copy a template strand by extending a primer: an existing strand with a free 3′ hydroxyl group. During each addition, that hydroxyl group reacts with an incoming deoxyribonucleoside triphosphate, forming a new phosphodiester bond and releasing pyrophosphate. The new strand therefore grows in the 5′→3′ direction.
Correct nucleotide selection depends on complementary base pairing and on the geometry and chemical interactions of the polymerase active site. These mechanisms are highly selective but not infallible. An incorrect nucleotide can occasionally be incorporated, producing a mismatch between the template and the newly synthesized strand.
Proofreading typically proceeds through four stages:
- Slowing of synthesis. A mismatched primer terminus is often extended much less efficiently than a correctly paired terminus. This pause creates an opportunity for editing.
- Transfer to the editing site. Several nucleotides at the primer end become unpaired, allowing the 3′ end to enter an exonuclease active site distinct from the polymerase active site.
- Excision. The editing enzyme removes the terminal nucleotide through hydrolysis of its phosphodiester bond. If necessary, more than one nucleotide can be removed.
- Resumption of synthesis. The corrected primer terminus returns to the polymerase active site, where extension continues.
The editing reaction is performed by a 3′→5′ exonuclease, so called because it removes nucleotides from a strand end in the direction opposite to DNA synthesis. Its active site commonly uses divalent metal ions to catalyze bond cleavage.
Proofreading does not require an enzyme to recognize an error in an absolute sense. Correctly paired termini can also enter an editing site, and mismatched termini sometimes escape correction. High fidelity emerges from the different rates at which correct and incorrect termini undergo extension, transfer, and excision.
Organization of proofreading enzymes
Proofreading activity may reside in the same polypeptide as the polymerase or in a separate component of the replication machinery.
Bacteria
In Escherichia coli, the main chromosomal replicative enzyme, DNA polymerase III, contains a polymerizing α subunit and a separate ε subunit that supplies 3′→5′ exonuclease activity. Their association couples synthesis to proofreading within a multiprotein complex.
DNA polymerase I, which participates in processing replication intermediates and in DNA repair, contains both polymerase and proofreading activities in the same polypeptide. It also has a distinct 5′→3′ exonuclease activity involved in removing DNA ahead of a synthesis site. This latter activity is not the proofreading reaction.
These arrangements illustrate that proofreading is a functional relationship between synthesis and editing, rather than a requirement that both active sites belong to one protein.
Eukaryotes and organelles
In eukaryotes, DNA polymerases ε and δ perform much of nuclear genome replication and contain intrinsic proofreading exonuclease domains. Under the predominant model of an undisturbed replication fork, polymerase ε carries out most leading-strand synthesis, whereas polymerase δ carries out most lagging-strand synthesis. Their contributions can change during replication initiation, stress, or repair.
DNA polymerase α, which works with primase to initiate DNA synthesis, lacks intrinsic proofreading. Some errors made by polymerase α can be removed by polymerase δ during subsequent processing and extension. Correction by an enzyme other than the one that introduced the error is often called extrinsic proofreading.
DNA polymerase γ, responsible for DNA replication in the mitochondrion, also possesses 3′→5′ proofreading activity.
Archaea and viruses
Many replicative polymerases in archaea also combine DNA synthesis with proofreading, although their subunit organization differs among polymerase families. Some viruses, including many DNA viruses, encode proofreading-capable replication enzymes.
Not all DNA polymerases proofread. Several specialized polymerases involved in damage tolerance lack 3′→5′ exonuclease activity. Their ability to synthesize across damaged or otherwise difficult templates can come at the cost of lower copying accuracy.
Contribution to replication fidelity
Replication accuracy is produced by several partially overlapping safeguards:
| Safeguard | Main action | Typical timing |
|---|---|---|
| Nucleotide selection | Favors incorporation of the correct nucleotide | Before and during bond formation |
| Proofreading | Removes an incorrectly incorporated nucleotide from the growing strand end | During DNA synthesis |
| Mismatch repair | Removes mismatches that escaped initial selection and proofreading | After incorporation, usually in newly replicated DNA |
For many well-studied replicative polymerases, proofreading improves fidelity by roughly one to several hundredfold, and sometimes more. The effect depends on the enzyme, mismatch, DNA sequence, nucleotide concentrations, and experimental conditions. No single enhancement factor applies to all polymerases or all types of error.
A useful schematic relationship is
where is the initial probability of misincorporation, and each escape fraction represents the proportion of relevant errors that survive a subsequent safeguard. This is a conceptual approximation: the mechanisms are not fully independent, and different mismatches are processed with different efficiencies.
The final accuracy of cellular replication should therefore not be attributed to proofreading alone. Measured mutation rates also reflect mismatch repair, DNA damage, repair pathways, sequence context, and the biological consequences of mutations.
Distinction from mismatch repair and other editing processes
Proofreading and mismatch repair both correct copying errors, but their substrates and mechanisms differ.
Proofreading primarily acts at, or very near, the growing 3′ end of a primer. Its opportunity to correct an error is closely linked to the competition between further extension and exonucleolytic removal.
Mismatch repair recognizes mismatches or small insertion–deletion loops within double-stranded DNA. It identifies the newly synthesized strand through organism-specific signals, removes a segment containing the error, and replaces that segment by DNA synthesis. It can therefore correct errors that are no longer accessible at the primer terminus.
Proofreading is also distinct from removal of an RNA primer, excision of a chemically damaged base, and cleavage of DNA ahead of a polymerase by a 5′→3′ nuclease. Similar terminology is sometimes used for editing during transcription or for quality control during protein synthesis, but those processes use different molecular machinery.
Biological consequences and limitations
An uncorrected mismatch is not necessarily an immediately fixed mutation. Subsequent repair may restore the original sequence. If the mismatch persists into another replication cycle, however, copying of the two alternative bases can produce daughter DNA molecules with different sequences, establishing a mutation in one lineage.
Loss of proofreading commonly produces a mutator phenotype, an increased rate of spontaneous mutation. Its consequences depend on the polymerase affected, the remaining repair systems, and the organism. Proofreading defects can increase base substitutions and some small insertion or deletion errors.
Certain pathogenic variants in the exonuclease domains of human POLE and POLD1, which encode catalytic subunits of polymerases ε and δ, are associated with increased mutation burdens and inherited predisposition to particular cancers. The effects are variant-specific; not every change in a proofreading domain disrupts proofreading or has the same biological consequences.
Proofreading has several inherent limits:
- Mismatch dependence: some incorrect base pairs resemble normal pairs closely enough to evade efficient editing.
- Competition with extension: an incorrect terminus may be extended before it is excised, reducing its accessibility to the proofreading site.
- Sequence context: neighboring bases and repetitive sequences affect misincorporation, strand slippage, and editing.
- Template damage: removal of a newly added nucleotide does not itself repair damage in the template strand.
- Incomplete coverage: errors made by polymerases without proofreading are not necessarily accessible to another proofreading enzyme.
The balance between synthesis and editing also affects replication efficiency. Excessive exonuclease activity can remove correctly incorporated nucleotides and impede productive synthesis; insufficient editing allows more errors to persist.
Discovery and experimental study
The mechanistic basis of proofreading emerged from biochemical studies of bacterial DNA polymerases during the 1960s and early 1970s. Work on DNA polymerase I established the connection between its 3′→5′ exonuclease activity and preferential removal of mismatched nucleotides. Subsequent genetic and biochemical studies showed that disrupting proofreading increases replication errors.
Proofreading is studied using purified enzymes, defined DNA substrates, and cellular mutation assays. Important approaches include:
- Comparing otherwise similar polymerases with active or disabled exonuclease sites.
- Measuring extension and excision of matched and mismatched primer termini.
- Determining structures of polymerase–DNA complexes by X-ray crystallography or cryo-electron microscopy.
- Using DNA sequencing to identify mutation frequencies and patterns after changes to proofreading activity.
Interpretation requires distinguishing errors generated during DNA synthesis from mutations that remain after cellular repair and selection. An exonuclease-deficient enzyme can also differ in DNA binding or synthesis behavior, so a mutation-rate change does not always measure proofreading efficiency alone.
Use in biotechnology
Proofreading polymerases are widely used in high-fidelity polymerase chain reaction (PCR), particularly when amplified DNA will be cloned, expressed, assembled, or sequenced. Many such enzymes derive from archaeal polymerases, including Pfu polymerase from Pyrococcus furiosus, or from engineered derivatives.
Conventional Taq DNA polymerase lacks 3′→5′ proofreading activity. Under comparable suitable conditions, proofreading PCR enzymes generally introduce fewer copying errors, although actual fidelity also depends on buffer composition, nucleotide balance, template quality, cycle number, and sequence context.
Proofreading activity can also influence experimental compatibility. Some proofreading enzymes can degrade primer ends, and many yield predominantly blunt-ended PCR products rather than the 3′ adenine overhangs commonly produced by Taq. Engineered enzymes and polymerase blends can modify these properties, so proofreading status alone does not determine every feature of a PCR product.