Replication error rates for G.dGTP, T.dGTP, and A.dGTP mispairs and evidence for differential proofreading by leading and lagging strand DNA replication complexes in human cells.

We have determined the fidelity of DNA replication by human cell extracts in reactions containing excess dGTP. Replication errors were scored using two M13 DNA substrates having the replication origin on opposite sides of the lacZ alpha-complementation gene. The data suggest that the average rates for replication errors resulting from G(template), T.dGTP, and A.dGTP mispairs are 25 x 10(-6), 12 x 10(-6), and 3 x 10(-6), respectively. The data also suggest that error rates for both the (+) and (-) strands differ by less than 2-fold when they are replicated either as the leading or lagging strand. This is in contrast to the 33- and 8-fold differences observed earlier for G.dTTP and C.dTTP mispairs on the (+) strand when replicated by the leading or lagging strand complex (Roberts, J. D., Izuta, S., Thomas, D. C., and Kunkel, T. A. (1994) J. Biol. Chem. 269, 1711-1717). Thus, the relative fidelity of the leading and lagging strand replication proteins varies with the mispair and sequence considered. Misincorporation of dGTP preferentially occurs at template positions where dGTP is the next correct nucleotide to be incorporated. This "next nucleotide" effect is characteristic of reduced exonucleolytic proofreading and suggests that these replication errors are normally proofread efficiently. Fidelity measurements performed in the absence or presence of dGMP, an inhibitor of proofreading exonuclease activity, suggest that the leading strand replication complex proofreads some mispairs more efficiently than does the lagging strand replication complex.