DNA Polymerase and Mismatch Proofreading
Kornberg's Replication Fork, Exonuclease Editing, and Okazaki Fragment Stitching
Arthur Kornberg purified the first DNA-synthesizing enzyme, DNA Polymerase I, from Escherichia coli extracts in 1956. Skeptical biochemists doubted that a blind enzyme could copy genetic sequences accurately. Kornberg silenced the doubt by showing that varying the adenine-to-guanine ratio of the input DNA template altered the newly synthesized product strand in exact matching proportions.
Thomas Steitz solved the crystal structure of the DNA Polymerase I Klenow fragment in 1985, revealing a right-handed architecture shared across nature. The DNA duplex rests across a $\beta$-sheet palm holding two catalytic $\text{Mg}^{2+}$ ions while a thumb domain grips the duplex. Alpha-helical fingers snap shut by 40 degrees over each incoming deoxynucleoside triphosphate.
Base-pairing hydrogen bonds between adenine-thymine and guanine-cytosine pairs provide a free-energy preference of only 8 to $12\text{ kJ/mol}$ over mismatches in water. That small gap alone would yield one error every 100 bases. Polymerases sharpen selection to one error in 100,000 by clamping the closed fingers into a steric pocket demanding an exact $10.8\text{-\AA}$ pair width.
When a rare tautomer or wobble mismatch slips past the insertion pocket, its skewed $3'\text{-OH}$ terminus misaligns from the catalytic magnesium ions and stalls forward addition a hundredfold. That pause lets the frayed DNA end melt across $30\text{ \AA}$ into a separate $3'\to 5'$ exonuclease proofreading site. Clipping off the bad base drops the error rate to $10^{-7}$.
After the replication fork moves on, the MutS-MutL-MutH mismatch repair complex scans the newly made duplex for residual kinks and excises the faulty patch. In E. coli, MutH tells the new strand from the old parent strand because the parent carries methyl tags on GATC sequences. That final sweep pushes genomic accuracy below one error in $10^9$ bases.
Left by itself, a core DNA polymerase falls off its template strand after adding only 10 to 20 nucleotides. Mike O'Donnell solved the crystal structure of the $\beta$ sliding clamp in 1992, revealing a closed ring that encircles the DNA duplex. Tethered to that sliding washer, E. coli Polymerase III copies 50,000 bases at 1,000 nucleotides per second.
Because every known DNA polymerase adds nucleotides exclusively to a $3'\text{-OH}$ group in the $5'\to 3'$ direction, the antiparallel lagging strand cannot be copied continuously. Reiji and Tsuneko Okazaki proved the mechanism in 1968 using two-second tritiated-thymidine pulses in E. coli. The lagging strand is stitched together backward from 1,000-nucleotide Okazaki fragments.
Unlike RNA polymerases, no DNA polymerase can start a brand-new chain from bare nucleotides; it requires a pre-existing $3'\text{-OH}$ hook. On the lagging strand, DnaG primase synthesizes a 10-nucleotide RNA primer every second. DNA Polymerase I later chews away those RNA primers with a $5'\to 3'$ nuclease domain before DNA ligase seals the nicked backbone.
To keep leading- and lagging-strand synthesis moving forward together at identical speeds, Bruce Alberts proposed the trombone model in 1980. The lagging-strand template loops back 180 degrees through a coupled twin-polymerase replisome. As the fork advances, the looped DNA feeds out like a trombone slide during each Okazaki fragment and resets every second.
As the hexameric DnaB helicase unwinds the E. coli double helix at 1,000 base pairs per second, it twists the parental DNA ahead of the fork by 100 turns every second. Discovered by Martin Gellert in 1976, DNA gyrase relieves that torsional lockup. It slices both strands, passes another duplex segment through the break, and reseals the cut.
Removing the final RNA primer at the tip of a linear eukaryotic chromosome leaves an uncopied $3'$ gap that would shorten the chromosome at every division. Elizabeth Blackburn and Carol Greider discovered telomerase in Tetrahymena in 1985. Carrying its own internal RNA template, this reverse transcriptase adds repetitive TTAGGG buffer caps to chromosome ends.
Eukaryotic DNA polymerases copy chromatin-wrapped DNA at just 30 to 50 nucleotides per second—twenty times slower than bacterial Polymerase III. A single origin would need six weeks to copy a human chromosome. To duplicate all six billion base pairs inside an eight-hour S-phase window, human cells fire 30,000 to 50,000 replication origins in parallel.