Mismatch repair in human nuclear extracts. Time courses and ATP requirements for kinetically distinguishable steps leading to tightly controlled 5' to 3' and aphidicolin-sensitive 3' to 5' mispair-provoked excision.

Mismatch repair (MMR) systems enhance genomic stability by correcting DNA replication errors. The events in mammalian MMR pathways remain poorly understood. Using HeLa cell nuclear extracts, we analyzed correction of mispairs in circular DNA substrates with single defined nicks and measured excision in the absence of exogenous dNTPs by annealing specific oligonucleotide probes. In reactions initiated by concomitant temperature shift and addition of ATP or Mg(2+) to otherwise complete mixtures on ice, ATP-initiated excision and final error correction lagged behind Mg(2+)-initiated reactions, suggesting a very early requirement for ATP but not its hydrolysis. Subsequent stable commitment (resistance to added excess competitor substrate) began within 30 s, required hydrolyzable ATP, and plateaued after 60-70 s. This may reflect formation of hydrolysis-dependent translocating and/or pre-excision complexes. Excision along shorter nick-mispair paths began 15 s later than commitment. Both 3' to 5' and 5' to 3' excision gaps appeared at rates of approximately 0.0055 of final yields per second, respectively, 30 or 2.5 times the nonspecific excision rates. The lag between 3' to 5' excision gaps at two different positions yielded an excision progress rate of 5.2 nucleotides/s. In both substrates, corrected products appeared at fractional rates of 0.0027 of final yield per second. Aphidicolin, known to inhibit both the DNA synthesis and 3' to 5' exonuclease activities of polymerases delta and epsilon, reduced appearance of 3' to 5' excision tracts roughly 4-fold at 90 microm but had no effect on 5' to 3' excision.