Analysis of bacterial systems has demonstrated that strand-specific DNA mismatch correction protects the genome against mutation by removing biosynthetic errors from newly replicated DNA and by aborting recombination events between related, but non-allelic DNA sequences. We have recently reconstituted E. coli methyl-directed mismatch repair in a pure system comprised of 8 proteins, and have also demonstrated the existence of a similar pathway in nuclear extracts derived from human cells. Like the bacterial pathway, the system we have identified in human cells is able to recognize and to process in a strand-specific manner the different classes of base-base mispairs in a reaction that requires a replicative (aphidicolin sensitive) DNA polymerase. The primary goal of this proposal is to establish the molecular nature of strand-specific mismatch repair as it occurs in human cells. To this end, the reaction occurring in HeLa nuclear extracts will be characterized with respect to mismatch specificity, size and location of excision repair tracts, involvement of alpha and delta DNA polymerase, and possible involvement of the mammalian homologue of MutS, the protein that mediates mismatch recognition in bacterial systems. A major aim of this study is the isolation and characterization of the components required for the human reaction, with the hope that, as in the case of E. coli, we will be able to reconstruct the reaction in a defined system. Since it is not possible to monitor the course of a mismatch repair event occurring within a living cell, the nuclear extract studies outlined above will provide the criteria for evaluating successful reconstitution of the reaction in a purified system. As an initial attempt to address the relationship between mismatch repair proficiency and genetic stability in mammalian systems, we will also participate in a collaborative study that will compare, with respect to their proficiency in mismatch correction, a "wild type" CHO cell line and a mutant derivative that contains a deletion spanning the gene encoding the mammalian MutS homologue. Should a defect in mismatch repair be associated with the mammalian MutS deletion, the spontaneous mutabilities of wild type and mutant lines will be assessed.