Olivier Hyrien1, Guillaume Guilbaud2 and Torsten Krude3 1Département de Biologie, École Normale Supérieure, Université Paris Science and Letters, Centre National de la Recherche Scientifique, Institut National de la Santé et de la Recherche Médicale, Institut de Biologie de l'Ecole Normale Superieure, 75005 Paris, France; 2Division of Protein and Nucleic Acid Chemistry, MRC Laboratory of Molecular Biology, Cambridge CB2 0QH, United Kingdom; 3Department of Zoology, University of Cambridge, Cambridge CB2 3EJ, United Kingdom Corresponding authors: tk218cam.ac.uk, olivier.hyrienbio.ens.psl.eu, guilbaudmrc-lmb.cam.ac.uk Abstract

Mammalian DNA replication origins have been historically difficult to identify and their determinants are still unresolved. Here, we first review methods developed over the last decades to map replication initiation sites either directly via initiation intermediates or indirectly via determining replication fork directionality profiles. We also discuss the factors that may specify these sites as replication initiation sites. Second, we address the controversy that has emerged from these results over whether origins are narrowly defined and localized to specific sites or are more dispersed and organized into broad zones. Ample evidence in favor of both scenarios currently creates an impression of unresolved confusion in the field. We attempt to formulate a synthesis of both models and to reconcile discrepant findings. It is evident that not only one approach is sufficient in isolation but that the combination of several is instrumental toward understanding initiation sites in mammalian genomes. We argue that an aggregation of several individual and often inefficient initiation sites into larger initiation zones and the existence of efficient unidirectional initiation sites and fork stalling at the borders of initiation zones can reconcile the different observations.