We have produced a comprehensive nucleosome map of DNA replication origins in S. cerevisiae
. Our analysis is distinct from previous genome-wide views of nucleosome position at replication origins 
in that we combined a comprehensively curated set of origins in which the ACS element was accurately mapped 
with the most comprehensive genome-wide nucleosome maps. In this manner, we detected the NDR flanked by nucleosomes that was evident in previous views (derived without critical alignment parameters 
). But more importantly, we extend this view by detecting phased arrays of positioned nucleosomes extending from either side of the origin NDR.
Considerable diversity was evident in the replication origin nucleosome maps, reinforcing the notion that the average view does not reflect the different nucleosome occupancy patterns that exist at active, functional replication origins. We found that adjacent genomic features, most notably TSS elements and gene ends, can influence the nucleosome patterns at replication origins. In particular, the presence of an adjacent TSS can result in a second NDR in addition to the NDR at the ACS. We found that TSSs are distributed asymmetrically at replication origins and that maxima in the TSS distribution correlate with early origin firing. The presence of a second NDR could improve the accessibility of the replication origin for ORC or the proteins that are recruited by ORC, or factors bound at the promoter element within the second NDR could play a direct role in recruiting replication proteins to the pre-initiation complex. In either case the activity of the replication origin would be promoted, consistent with increased likelihood that these origins will be active in early S phase. We also found that extremes of NDR width, either narrow or wide, were characteristic of late origins. For example, origins with the narrowest NDR have higher than average occupancy at the ACS. This architecture could lead to a competition between nucleosomes and ORC for binding at the ACS, resulting in a reduced efficiency of origin firing, as previously suggested 
. We conclude that functional replication origins can be built with different chromatin architectures, and that adjacent genomic features can influence the timing of replication origin firing.
Unfortunately, due to a lack of appropriate genome-wide datasets we were unable to test more sophisticated measures of origin robustness. Origin efficiency, or the likelihood that a given origin will fire in a given cell cycle, is an important parameter to test with respect to origin nucleosome architecture. This parameter is quite complex, however, encompassing not simply the intrinsic efficiency of an origin, but also the time during S phase when it fires (as later firing origins are more likely to be replicated passively from a neighboring origin), as well as the proximity of other origins, which also have unique efficiencies. As genome-wide origin efficiency datasets become available in S. cerevisiae our classification of different nucleosome patterns at replication origins will be an important tool for further investigating the impact of nucleosome structure on origin function. Accordingly, we expect the analysis presented here to represent a benchmark for future large-scale studies.
One attractive model of nucleosome positioning posits that uniformly-spaced arrays of nucleosomes, such as those seen downstream of TSSs, are the result of nucleosome packing adjacent to a barrier element 
. This uniform spacing decays further away from the barrier element, and this decay is seen as a decrease in the peak to trough height. Our data suggests that, on average, replication origins conform to this statistical positioning model. The average ACS-centered view of replication origins revealed strongly positioned +1 and −1 nucleosomes flanked by arrays of phased nucleosomes in which the uniform spacing decays as one moves away from the ACS. As is the case with the +1 nucleosome at TSSs 
, the key to understanding nucleosome positioning at replication origins likely lies in understanding the elements responsible for positioning the +1 and −1 nucleosomes that flank the ACS. Analysis of the underlying sequence at replication origins gave conflicting results. On one hand, assembly of nucleosomes in vitro
(in the complete absence of ORC) resulted in a larger NDR at the ACS than that observed in vivo
, indicating that the intrinsic sequence preference of histones does not accurately describe the positions of the +1 and −1 nucleosomes. However, this large NDR is likely the result of lower nucleosome density (approximately 50% of the in vivo
density) in the chromatin assembled in vitro 
, which might prevent the more dramatic encroachment of nucleosomes towards the ACS that is observed in vivo
. Analysis of dinucleotide patterns revealed some sequence properties that predicted both an NDR of the expected size, and the positions of the +1 and −1 nucleosomes, indicating a role for sequence in positioning these critical nucleosomes. Perhaps the most compelling evidence that DNA sequence alone does not position the +1 and −1 nucleosomes at replication origins comes from genetic perturbation of the origin recognition complex. Upon depletion of ORC we found that most origins displayed a change in the position of the nucleosomes flanking the ACS, with nucleosomes shifting inwards towards the ACS. In addition, in many cases the flanking nucleosomes became delocalized. These changes result in a shift in the phasing of adjacent nucleosomes and in delocalization of adjacent nucleosomes. Thus, when ORC binding is compromised the position of the +1 and −1 nucleosomes is altered, consistent with ACS-bound ORC serving as a barrier element component. However, the nucleosome-free region that we observe in vivo
when ORC is present is, at ~130 bp, considerably larger than both the in vitro
binding footprint of purified ORC 
and the ORC footprint seen in vivo 
, suggesting that bound ORC is not the sole barrier element. We propose that ORC, in concert with additional protein factors recruited by ORC, positions the nucleosomes that flank the NDR at origins of replication.
Together our data suggest a model of nucleosome assembly at replication origins () in which the NDR is specified by the DNA sequence of the ARS. This NDR is narrower in vivo
than in vitro
due to the presence of chromatin remodeling and modifying activities, yet wider than the ORC binding site. This sequence-specified NDR creates a chromatin environment that is permissive for ORC binding to the ACS. Binding of ORC, and perhaps recruitment of chromatin remodelers and modifiers by ORC (such as Rpd3, Sir1, Hat1, and Hat2 
) specifies the position of the +1 and −1 nucleosomes, resulting in arrays of phased nucleosomes on either side of the ACS. These positioned nucleosomes then become important for the assembly of the pre-replicative complex of replication initiation proteins 
prior to origin firing. One particularly attractive feature of this model is that it is consistent with the suspected role of chromatin structure in regulating replication origins in metazoans 
. Perhaps in the more complex replication origins of higher eukaryotes the role of DNA sequence recognition by ORC has been partially replaced by a more direct interplay between nucleosomes and the origin recognition complex. That the bromo-adjacent homology domain, which interacts with nucleosomes in some contexts 
, facilitates the binding of human ORC with chromosomes 
suggests a mechanism by which this could be achieved. It will of course be of tremendous interest to test whether nucleosome positioning at DNA replication origins is dictated by a combination of DNA sequence and ORC binding in other organisms, as appears to be the case in yeast.
Note added in proof: While this manuscript was under review a similar study was published 
. Although the studies utilized different (but largely overlapping) origin/ACS lists and methodologies (sequencing vs. microarray hybridization), they reached complementary conclusions.