Our main results can be summarized as follows: 1) ORC binding to fission yeast origins is periodic in the cell cycle, rising during M and peaking at the M/G1 transition. Pre-RC formation is also periodic, rising and peaking in G1. 2) Compared to an early efficient origin (ori2004), ORC binding and pre-RC formation are delayed at a late inefficient origin (ori2060) and at a cryptic origin not used in it normal chromosomal context (ars727). 3) The extent of pre-IC formation at these three origins reflects the efficiency of their usage. 4) Decreasing ORC binding through removal of AT-hook motifs at ori2004 results in delays in pre-RC and pre-IC formation and delayed replication. 5) Delaying cells in M equalizes ORC binding and leads to changes in origin usage across the genome for many origins such that early efficient origins are utilized less efficiently, and some late inefficient origins are used more efficiently. 6) Overproduction of pre-IC factors Cdc45, Hsk1, or Dfp1 results in increased origin efficiency across the genome.
Peak ORC association with fission yeast origins at M/G1 is similar to the timing of ORC binding to origins in metazoa (DePamphilis, 2005
). Our results contrast with conclusions from earlier studies in both budding and fission yeasts, which have reported an unchanging association of ORC with chromatin (Lygerou and Nurse, 1999
) and origins (Aparicio et al., 1997
; Ogawa et al., 1999
). The differences between the present and earlier fission yeast results concerning origin binding are likely due to the increased temporal resolution and quantitative sensitivity of the experiments reported here. Using ChIP, we cannot exclude the possibility that a conformational change at the origin results in alterations in Orp1 ChIP signals. If this is the case, our results may reveal a specific event such as chromatin remodeling occurring at the origin during the M/G1 transition, prior to pre-RC formation. Our analysis also does not allow us to distinguish if all cells exhibit periodic ORC behavior at a given origin or if this occurs in a subpopulation of cells in which the origin is competent to fire.
Our work indicates that the timing of ORC and MCM binding during mitosis and G1 contributes to establishing the replication program during S. We show that pre-RCs are formed at origins regardless of their efficiencies, consistent with work from both budding and fission yeasts (Hayashi et al., 2007
; Santocanale and Diffley, 1996
), but that ORC binding and pre-RC formation are both delayed at a later firing origin. In contrast, no differences have been reported for MCM association with early and late firing origins in budding yeast (Aparicio et al., 1999
). We also demonstrate that removal of AT-hook motifs in ori2004
which reduces ORC binding delays pre-RC and pre-IC formation as well as replication timing, suggesting that differences in affinity of origins for ORC during M play a role in the timing of origin firing during the S of the subsequent cell cycle.
We propose that delays in ORC and MCM binding during M and G1 lead to a non-equal association of Cdc45 and other replication factors among origins, thus establishing the replication timing program and determining the efficiency of origin usage. This can be understood if origins compete for a limited pool of replication factors such that origins at which ORC and pre-RC assemble late have a smaller number of these factors available to them. Consistently, we observe that the level of Cdc45 bound to an origin is closely related to origin efficiency; it should be noted that our results represent a population analysis and that Cdc45 may bind only to a subpopulation of active origins. Cdc45 has been suggested to be present in limiting quantities in human cells (Pollok et al., 2007
), and the replication efficiency in Xenopus
extracts correlates with the amount of Cdc45 on chromatin (Edwards et al., 2002
). Moreover, a recent study using DNA combing has shown that Hsk1-Dfp1 in fission yeast regulates origin efficiency (Patel et al., 2008
). Pre-IC formation at late origins may be restricted by decreasing levels of available factors as cells enter S and by passive replication through a region inactivating an origin so it cannot bind such factors. Therefore, the efficiency of an origin would be determined by both the timing of origin firing and its context in relation to other origins, with proximity to an early origin inactivating a late origin due to passive replication.
Our model proposes that the timing and therefore efficiency of origin firing are established by the recruitment of ORC to origins during M followed by competition among origins for limiting replication factors. ORC binding to origins may be determined by a combination of the primary sequence of an origin as well as its chromatin context. The timing of the increase in ORC binding during M and of pre-RC formation during G1 determines origin usage during the subsequent S. Our model could explain replication timing and efficiency as follows: ORC binding to early origins reaches a maximum at the exit of M, at which time ORC association with late origins is still low. At the beginning of G1, when MCM is competent to bind to origins, early origins have sufficient ORC to assemble a pre-RC, while late origins do not. Early origins with pre-RCs can then bind Cdc45 and other replication factors, which we suggest are present in limiting amounts. At late origins, ORC binding increases and reaches a maximum only later in G1, when they can then recruit MCMs. These origins complete pre-RC assembly toward the end of G1 and will be delayed for Cdc45 binding compared to early efficient origins, and it is this delay that results in a difference in replication timing. Late origins will also have a smaller pool of available Cdc45 for pre-IC formation, which decreases the likelihood of pre-IC formation and origin usage. Support for this model comes from our experiments prolonging M, which allows time for ORC to bind fully at more origins, reducing the differences in usage of many efficient and some inefficient origins. If ORC is limiting, as has been shown in budding yeast, then the equilibration of ORC binding may reflect a dynamic on-off process for ORC at origins during the prolonged M (Rowley et al., 1995
). Alternatively, ORC may simply accumulate to a maximal level at both efficient and inefficient origins, allowing them to compete more equally for limiting replication factors. When mitosis is extended, a decrease in efficiency of many of the most frequently used origins is observed, while some inefficiently replicated regions become more active. Moreover, overproducing pre-IC factors Cdc45, Hsk1, or Dfp1 increases replication in both efficient and inefficient regions, suggesting that these proteins, among others, are limiting for replication.
The fact that we observe partial equilibration of origin efficiencies following MBC treatment may be due to the inaccessibility of an origin in its chromatin context. Certain regions of the fission yeast genome are generally repressed for origin activity, as reported by Heichinger et al. (2006)
and Hayashi et al. (2006), and interestingly the MBC-induced origins identified in this study are mostly located outside these regions (Supp. Fig. 6
). It is also possible that as some replication factors localize to nuclear foci (Meister et al., 2007
), an origin may need to be positioned near a local pool for recruitment to occur. In addition, previous work investigating the program of origin firing in other organisms has implicated chromatin structure and histone modifications as important regulators. For example, acetylation status has been shown to change the pattern of origin firing, as deletion of the Rpd3 histone deacetylase in budding yeast results in increased histone acetylation and earlier firing of several origins (Aparicio et al., 2004
; Vogelauer et al., 2002
). These results are consistent with work in Drosophila
follicle cells demonstrating that histones at active origins are hyperacetylated during the period of ORC binding and that RPD3 mutations result in redistribution of Orc2 in amplification stage cells (Aggarwal and Calvi, 2004
). The timing of origin firing in fission yeast may likewise be regulated by chromatin modifications during the cell cycle, which modulate the affinity of an origin for ORC. In our experiments, delaying cells in mitosis may lead to a subset of late origins acquiring the same modifications as early origins, thus altering the origin firing program by allowing more equivalent binding of ORC.
Our work demonstrates that origin efficiency and timing in fission yeast are regulated by events that occur during mitosis of the previous cell cycle and suggests that competition for limiting factors by origins during G1 and S determines their activity. Support for aspects of this model comes from experiments in budding yeast where the late replication of origins at telomeres is established between mitosis and G1 (Raghuraman et al., 1997
), in CHO cells where ORC binds chromatin during M/G1, and in mammalian nuclei introduced into Xenopus
extracts where replication timing is established by early G1 (Dimitrova and Gilbert, 1999
; Okuno et al., 2001
). In human cells, time-lapse imaging studies have shown that Orc1 exhibits a distinctive localization pattern during G1 resembling the pattern of DNA replication that occurs during S (S. Prasanth and B. Stillman, unpublished results). Thus, it is possible that as in fission yeast, ORC binding and pre-RC assembly during M of the previous cell cycle and the G1 prior to S phase are key determinants for origin selection and the timing of origin firing in the cells of metazoan eukaryotes.