Single‐Molecule Optical Replication Mapping (ORM) Suggests Human Replication Timing is Regulated by Stochastic Initiation

Abstract

DNA replication timing is regulated by the timing of initiation across the genome. However, there is no consensus as to how initiation timing is regulated. Deterministic models contend that different initiation sites are programed to initiate at different, well‐defined times. Stochastic models posit that different initiation sites have different initiation probabilities but can fire at any time during S. Sites with a high probability of initiation are more likely to fire early and thus will, on average, initiate early in S. Ensemble replication timing assays that report only average initiation time cannot distinguish between these two types of models.To test if replication initiates deterministically or stochastically, we developed Optical Replication Mapping (ORM) a high‐throughput, genome‐wide, single‐molecule replication mapping approach that combines in vivo replication labeling with the Bionano genomic mapping technology. We have mapped replication in over 20 human cell datasets, representing over 2000‐fold coverage of the human genome, allowing us to map sites of replication initiation that are used in fewer than 0.1% of S phases.Our analyses support three major conclusions. 1) Replication initiates at many infrequently‐used initiation sites spread across initiation zones of 30 to 100 kb. We find no evidence for discrete, high‐frequency initiation sites in the human genome. These result suggest that "replication origin", in the sense of a defined genetic element, is not a useful concept in metazoans. 2) Initiation zones correlate well with regions of DNase I accessible DNA, such as promoters. G4 sequences do not appear to be predictive of sites of initiation, beyond their enrichment in DNase I accessible chromatin. These results suggest that a major driver of initiation probability is DNA accessibility. 3) We observe a low frequency of early initiation across late‐replicating regions, and these initiation events are enriched in late initiation zones. Observed initiation rates across all regions of the genome are consistent with a stochastic mechanism of replication timing and inconsistent with a deterministic replication timing program. We propose that the dominant parameter that regulates replication timing is the initiation probability across initiation zones, with a high degree of heterogeneity of initiation sites within these zones, and no spatial or temporal coordination between individual initiation events.