The 5′ Nuclease Domain of DNA Polymerase I Mediates a Novel DNA Transfer Pathway during Proofreading

Abstract

The high fidelity of many DNA polymerases is due, in part, to their ability to remove erroneously incorporated nucleotides at the terminus of the growing primer strand. This “proofreading” activity is catalyzed by the 3′-5′ exonuclease (exo) domain in E. coli DNA polymerase I (Pol I). Structural studies have shown that this domain is spatially separated from the 5′-3′ polymerase domain (pol), which is responsible for incorporation of dNTPs by template-directed polymerization. Pol I also contains another spatially distinct domain, the 5′ nuclease (nuc) domain, which is connected to the enzyme core by a flexible amino acid linker. This domain is involved in cleavage of downstream flaps formed during Okazaki fragment maturation and base excision repair. Using a single-molecule FRET system, we resolved three distinct conformations of Pol I-DNA complexes. Experiments with Pol I mutants established that these states correspond to DNA bound at the pol, exo, or nuc domains. Notably, the DNA substrate can transfer among all three domains without dissociation from the enzyme. We determined rate constants for site switching of model DNA substrates containing a correctly paired primer terminus, a terminal mismatch or an internal mismatch. The presence of mismatches accelerated transfer of DNA from the pol domain to the exo domain, as expected during proofreading. Surprisingly, the mismatches also accelerated transfer of DNA along the pol to nuc to exo domain pathway. These observations reveal the existence of a second intramolecular proofreading pathway, in which mismatched DNA first transfers from the pol domain to the nuc domain and then to the exo domain. Our results suggest an unexpected role for the 5′ nuclease domain in the proofreading activity of Pol I.