Interrelationships between replication infidelity and genome composition

An enduring issue in evolutionary and cancer biology is how replication infidelity influences genome composition and vice versa. Here we examine this issue by sequencing the genomes of diploid budding yeast strains that are either mismatch repair (MMR) proficient or deficient and encode wild type or mutator variants of the three major nuclear DNA replicases. Analysis of over 43,000 mutations that accumulated in the absence of selective pressure demonstrates that the nuclear DNA replication machinery generates less than one mismatch per genome and in combination with MMR, achieves a genome-wide per base error rate of 1.7 x 10-10. Absent both MMR and purifying selection, replication error patterns strongly depend on replication origin proximity, replication fork direction, and the local DNA sequence. Preferred sequences were observed for base substitutions and deletions. Error rates also vary with replication time, in linker versus nucleosome-bound DNA, in 5'- and 3'-untranslated regions, in coding regions and in intergenic DNA. This genome-wide view shows that replication fidelity is amazingly high but heterogeneous, in patterns that suggest the underlying mechanisms by which replication modulates genome stability and composition and vice versa.