Sir2 mitigates an intrinsic imbalance in origin licensing efficiency between early- and late-replicating euchromatin

Significance

In eukaryotes, DNA replication origins, the sites where new DNA synthesis begins during the process of cell division, must be adequately distributed across chromosomes to maintain normal cell proliferation and genome stability. This study describes a repressive chromatin-mediated mechanism that acts at the level of individual origins to attenuate the efficiency of origin function. This attenuation is essential for achieving the normal spatial distribution of origins across the chromosomes of the eukaryotic microbe Saccharomyces cerevisiae. While the importance of chromosomal origin distribution to genome stability and cellular fitness is acknowledged, this study defines a chromatin modification that establishes the normal spatial distribution of origins across a eukaryotic genome.

Abstract

A eukaryotic chromosome relies on the function of multiple spatially distributed DNA replication origins for its stable inheritance. The spatial location of an origin is determined by the chromosomal position of an MCM complex, the inactive form of the DNA replicative helicase that is assembled onto DNA in G1-phase (also known as origin licensing). While the biochemistry of origin licensing is understood, the mechanisms that promote an adequate spatial distribution of MCM complexes across chromosomes are not. We have elucidated a role for the Sir2 histone deacetylase in establishing the normal distribution of MCM complexes across Saccharomyces cerevisiae chromosomes. In the absence of Sir2, MCM complexes accumulated within both early-replicating euchromatin and telomeric heterochromatin, and replication activity within these regions was enhanced. Concomitantly, the duplication of several regions of late-replicating euchromatin were delayed. Thus, Sir2-mediated attenuation of origin licensing within both euchromatin and telomeric heterochromatin established the normal spatial distribution of origins across yeast chromosomes important for normal genome duplication.

The distribution of DNA replication origins across eukaryotic chromosomes is important for maintaining cell proliferation and genome stability through multiple cell divisions. Regions that contain a paucity of origins are linked to chromosome fragility and cancer-associated deletions (1–4). Origin function relies on two distinct cell-cycle restricted reactions (5). In G1 phase, the ORC (origin recognition complex) and Cdc6 protein bind DNA to direct the assembly of a stable catalytically inactive MCM (minichromosome maintenance) complex, also known as origin licensing. In S phase, kinases and accessory proteins convert the MCM complex into two bidirectionally oriented helicases that unwind the DNA to allow for new DNA synthesis, also known as origin activation. Thus, the chromosomal distribution of two distinct reactions, MCM complex assembly (origin licensing) and MCM complex activation (origin activation), establishes the spatiotemporal distribution of origins. While recent progress in Saccharomyces cerevisiae provides insights into the how the distribution of origin activation is regulated, little is known about how the chromosomal distribution of origin licensing is achieved (6–11).

Several attributes of S. cerevisiae help address mechanisms relevant to chromosome duplication. The sequence-specific binding of yeast ORC and the organism’s small genome have allowed mapping of the ORC and MCM binding positions and initiation sites of ∼400 individual yeast origins at near-nucleotide resolution (12–17). Such studies have also identified origin-adjacent chromatin features (e.g., nucleosome positioning, modification states, nonhistone chromatin-associated proteins) and functional properties (e.g., replication time, origin efficiency, ORC binding modes) (18–23). Thus, yeast origins can be parsed by selected criteria into large cohorts whose comparisons can illuminate issues relevant to genome replication that would be difficult to glean from classical approaches. It is now established that approximately half of yeast origins, distributed broadly over the central regions of chromosomes, act in the first portion of S phase (early and mid origins), while the remaining origins, generally located between ∼15 and 75 kb from the telomeric ends, act later (late origins). Most of the yeast genome is euchromatic, and its duplication requires both early and late origins (Fig. 1). The terminal 5 to 10 kb of yeast chromosomes exist in a heterochromatic structure. While many telomeres harbor DNA elements with the potential to act as origins, these elements are repressed by heterochromatin, making telomeres among the last regions of the genome to be duplicated. This origin distribution establishes a reproducible spatiotemporal pattern of chromosome duplication in a proliferating yeast population. While recent reports have defined specific chromatin-associated proteins enriched within the central regions of yeast chromosomes that recruit a limiting S-phase kinase required for origin activation (8, 11), it is unclear whether and how chromatin impinges on the distribution of origin licensing.

Fig. 1.

Spatiotemporal progression of yeast chromosome VI replication. Replication of yeast chromosome VI (horizontal line) in a proliferating yeast population is shown. Origins are indicated with perpendicular lines. Centromere is indicated by the square. Origins with the highest probability of activation, depicted by black or dark gray vertical lines, are used in most cell cycles (efficient) and act in early S. Less efficient origins, which are used in <50% of cell cycles, are depicted in light gray or gray-dotted lines. Late origins, enriched for inefficient origins, usually act later in S. The majority of yeast DNA euchromatic, and origins represented by black or gray bars, are considered euchromatic origins, regardless of when or how efficiently they function. Origins within telomeres are red. Telomeres are in Sir2 heterochromatin that represses origins (24). Thus telomeric origins do not function in wild-type cells but do in sir2∆ cells.

A yeast heterochromatic deacetylase, Sir2, and a nucleosome binding protein, Sir3, are components of telomeric heterochromatin. Recent work reveals that Sir2 and Sir3 act directly on nucleosomes adjacent to euchromatic origins (25), an unanticipated result, given the paradigm for Sirs in yeast heterochromatic structures that inhibit both transcription and origin function (25–27). Genetic analyses reveal that molecular features of Sir2 chromatin found at euchromatic origins are functionally relevant (25, 28, 29), e.g., a deletion of SIR2 (sir2∆) suppresses the temperature-sensitive growth and origin licensing defects caused by the cdc6-4 mutation. However, because the relevance of Sir2 in origin licensing has only been assessed in mutant yeast defective for this reaction, the physiological role of this Sir2 chromatin at origins is unclear.

We used genome-wide mapping of MCM to show that Sir2 promotes an equitable distribution of origin licensing between early- and late-euchromatic and telomeric X origins. In SIR2 cells, these three distinct origin cohorts showed similar levels of MCM binding, whereas in sir2∆ cells, telomeric X origins and early-euchromatic origins gained MCM relative to late-euchromatic origins. Telomeric X origins exist within Sir2 heterochromatin. Sir2-chromatin marks were higher at early- compared to late-euchromatic origins. Thus, Sir2 attenuation of origin licensing correlated with Sir2-chromatin levels. Replication assays revealed that the function of both early-euchromatic and telomeric X origins was enhanced in sir2∆ cells, providing evidence that Sir2 attenuation of origin licensing helped limit origin function at these loci (24). In the absence of Sir2, several regions of late-replicating euchromatin failed to complete duplication by the end of S phase by a mechanism that was independent of the replication capacity of the Sir2-controlled rDNA locus that can alter euchromatic replication (30, 31). Thus, varying degrees of Sir2-mediated attenuation of origin licensing at individual origins balances the distribution of MCM complexes across chromosomes to promote complete duplication of euchromatin by the end of S phase.

Results

Early Origins Were Enriched among Euchromatic Origins Most Responsive to SIR2.

Origin licensing requires Cdc6. Thus, MCM ChIP-Seq signals (MCM signals) are lost at chromosomal origins in cdc6-4 cells cultured at 37 °C (25, 32). However, MCM signals are restored at many origins, including euchromatic origins, in cdc6-4 sir2∆ cells (25, 32). Not all euchromatic origins in cdc6-4 sir2∆ cells are rescued to the same extent. Because Sir2-heterochromatic regions are late replicating and inhibitory to origin function, we initially predicted that the euchromatic origins most affected by SIR2, i.e., those most rescued for MCM binding in cdc6-4 sir2∆ cells compared to cdc6-4 cells, would be late origins (24, 25, 30, 33, 34). Instead we found that the most Sir2-reponsive euchromatic origins were enriched for early and depleted for late origins (Fig. 2A). The analyses in Fig. 2A used MCM signals derived from combining data from three biological replicates and smoothing the data by binning nucleotide signals (25). We reexamined these data at nucleotide resolution as depicted in Fig. 2B. MCM signals, abolished in cdc6-4 cells, were rescued at euchromatic origins in cdc6-4 sir2∆ cells, as expected (Fig. 2C, all). Early-euchromatic origins were rescued more substantially than late-euchromatic origins, consistent with the outcome in Fig. 2A. Thus, while each of the euchromatic origin cohorts parsed by their replication time (Trep values) included Sir2-responsive origins, the early origins were more likely than late origins to show Sir2 sensitivity.

Fig. 2.

Early origins were enriched among euchromatic origins most Sir2 responsive. (A) Sir2-responsive origins were parsed into quintiles. The first (low), third (mid), and fifth (high) quintiles had median Sir2-reponsive values of 0.4, 0.7, and 1.0, respectively (25). Euchromatic origins that generated a Trep value in ref. 21 were parsed into quintiles from earliest (early) to latest (late). Hypergeometric distribution was used to compare the enrichments (+) or depletions (−) within each Sir2-responsive group relative to “all” origins. The P values are indicated (+/−: P < 0.05; ++/−−: P < 0.01). (B) The MCM signal associated with a given nucleotide was the median value derived from three biological replicates. To derive the MCM distribution for the origins within a given cohort, the per-nucleotide distribution of MCM signals for all origins in that cohort was determined. The median value (dark lines) and first and third quartiles (shaded coloring) for each cohort are shown. (C) MCM signal distributions for wild type (WT) (CDC6 SIR2), cdc6-4, and cdc6-4 sir2∆ cells were determined for each of the indicated cohorts. The number at the Left corner of the box is the ratio for median MCM signals summed between nucleotides −100 through +150 in cdc6-4 sir2∆ to that in CDC6 SIR2 cells.

Sir2-Chromatin Marks Were Higher at Early- Compared to Late-Euchromatic Origins.

Molecular hallmarks of Sir2-heterochromatin, Sir2-dependent depletion of H4K16ac, and Sir3 binding to nucleosomes, are present at euchromatic origins (25). While both early- and late-euchromatic origins are flanked by nucleosomes showing Sir2-dependent depletion of H4K16ac and Sir3 binding (25), early-euchromatic origins showed greater levels of these marks compared to late-euchromatic origins (Fig. 3 and SI Appendix, Fig. S1). The difference was most pronounced at the nucleosomes flanking ORC. Thus Sir2-chromatin levels were greater at early- compared to late-euchromatic origins.

Fig. 3.

Sir2-chromatin marks were higher at early- compared to late-euchromatic origins. (A) Early- and late-euchromatic origin cohorts were examined for Sir2-chromatin marks on six origin-adjacent nucleosomes. (B) Normalized H4K16ac assessed in SIR2 cells at early- and late-euchromatic origins. (C) The change in H4K16ac observed for early- and late-euchromatic origin-adjacent nucleosomes in sir2∆ cells. (D) Sir3 MNase ChIP-Seq signals at early (purple)- and late (mint)-euchromatic origin cohorts. The median Sir3 signal at the indicated −1 nucleosome differed between early- and late-euchromatic origins with the indicated P value (Wilcoxon rank sum tests). The significance cutoff used demanded that >90% of the nucleotide positions within the signal apexes had P values <0.05. (E) The origins in the early and late cohorts were randomly assigned to two groups and the Sir3 MNase ChIP-Seq signals from ref. 32 were determined for each randomized cohort.

SIR2 Promoted Equitable Distribution of MCM Complexes between Early- and Late-Euchromatic Origins.

SIR2 has a profound effect on origin licensing at euchromatic origins in cdc6-4 cells (Fig. 2C), but its effect in wild-type (CDC6) cells was unclear (25). Therefore, MCM signals were examined at higher resolution in CDC6 cells that differed in their SIR2 genotype (Figs. 2B and 4). In SIR2 cells, late-euchromatic origins generated lower median MCM signals than the early-euchromatic origins, as reported (10). However, substantial overlap in the distributions of signals for these two cohorts was observed, suggesting that the MCM binding differences between early- and late-euchromatic origins were minimal (Fig. 4A). In contrast, in sir2∆ cells, the difference in MCM signals between these origin cohorts was clear. MCM signals at >50% of the nucleotides differed between late- and early-euchromatic origin cohorts in sir2∆ cells, whereas no nucleotides differed at a comparable P value cutoff in SIR2 cells (SI Appendix, Fig. S2). MCM signals were also assessed at telomeric X origins that are repressed by Sir2-dependent heterochromatin (24). In SIR2 cells, telomeric X origins generated MCM signals most similar to those generated by the late-euchromatic origins, whereas in sir2∆ cells, their MCM signals were similar to those of early origins and greater than those of late origins (Fig. 4B). Thus, Sir2 limited MCM binding at telomeric X and early-euchromatic origins relative to late-euchromatic origins.

Fig. 4.

SIR2 promoted more equitable distribution of MCM complexes between early- and late-euchromatic origins. (A) MCM signals were determined as in Fig. 2C for the indicated origins in congenic SIR2 CDC6 and sir2∆ CDC6 cells. (B) Analyses as in A, with data for the telomeric X origins included. (C) Magnified view of the median values between nucleotides −10 and +100 for the early- and late-euchromatic origins in A. (D) Median signals between nucleotides −10 and +100 were summed to generate an MCM area for each origin, Z scores were assigned and displayed in box-and-whiskers plots for each indicated cohort. The P values (Wilcoxon rank sum test) for the difference between the early- and late-euchromatic cohorts’ median MCM Z scores are indicated. (E) Analysis as in D after exclusion of inefficient origins (17).

To quantify these differences, MCM signals between the −10 and +100 nucleotides for each of the fragments were summed, and each origin was assigned a Z score to indicate how far the MCM signal for that origin diverged from the mean behavior of the entire collection of origin fragments (mean defined as “0”) (SI Appendix, Fig. S1). The Z scores were displayed in box-and-whiskers plots (Fig. 4D). In SIR2 cells, each of the origin cohorts generated similar MCM signals, indicating that MCM complexes were equitably distributed. In contrast, in sir2∆ cells, the median Z score of the early-origin cohort fell above the mean, while the median Z score for the late-origin cohort fell below the mean, indicating that the distributions of MCM signals for these two cohorts differed. The telomeric X origins also generated higher MCM Z scores in sir2∆ cells compared to late-euchromatic origins. Thus, in the absence of Sir2, MCM complexes were more efficiently distributed to early-euchromatic and telomeric X origins compared to late-euchromatic origins. The Sir2-controlled rDNA origin did not show enhanced MCM binding in sir2∆ cells, consistent with a published study (SI Appendix, Fig. S3) (35).

Some origins are inefficient, possibly because they are not licensed in many cells (17). To focus on origins that were licensed to some minimal extent, we repeated the analysis using only origins within each cohort that generated an origin efficiency value in a genome-scale study (17). Considering only these origins, the difference between MCM signals at the early and late origins observed in sir2∆ cells was enhanced (Fig. 4E). Thus, Sir2 limited MCM accumulating within heterochromatic telomeres and early-replicating euchromatin relative to late-replicating euchromatin.

Changes in MCM Distribution Correlated with Changes in Replication Dynamics.

S-phase Sort-Seq experiments were used to address whether alterations in MCM distribution had the potential to alter replication dynamics. The same yeast strains used for MCM ChIP-Seq were cultured at the permissive growth temperature for cdc6-4 so that the effect of this mutation on replication dynamics could also be examined. The number of sequence reads for a given region in S are normalized to the corresponding reads from G1 to generate a Sort-Seq value. Genome annotation produces chromosomal replication profiles with peaks indicating an active origin (SI Appendix, Fig. S4) (36). Wild-type and sir2∆ cells generated virtually indistinguishable profiles. Therefore, Sir2 effects on euchromatin replication dynamics were not observable at this resolution. In contrast, origin function in cdc6-4 cells, regardless of SIR2 genotype, was reduced over several regions of the genome. Thus, only a fraction of origins functioned normally in cdc6-4 cells even at permissive temperature, and SIR2 did not affect origin preference. Thus, Sir2 did not alter the origins that remained the most functional when origin licensing was compromised by cdc6-4.

To enhance quantitative comparison of the relevant origins, Sort-Seq values were assigned to 30-kb replicons parsed into three cohorts by their origins’ replication times (Treps) or Sir2 responsiveness, and presented in box-and-whiskers plots (Fig. 5 A and B). The results validated the approach: the early replicons generated greater Sort-Seq values than the late replicons, and SIR2 delayed replication of the heterochromatic telomeric X origins. In cdc6-4 cells, regardless of SIR2 genotype, the relative behavior of these replicons was maintained. However, when these replicons were parsed by Sir2 responsiveness, a cdc6-4 effect on origin usage was observed (Fig. 5B). In CDC6 cells, regardless of SIR2 genotype, the three Sir2-responsive cohorts produced similar broad box-and-whiskers plots, an expected outcome given that each of these cohorts contained a mixture of replicons duplicated at different times in S (Fig. 2A). In contrast, in cdc6-4 cells, regardless of SIR2 genotype, the different Sir2-responsive cohorts produced box-and-whiskers plots similar to those produced for the cohorts parsed by their S-phase replication times (compare cdc6-4 data in Fig. 5 A and B). Thus, in cdc6-4 cells, the origin cohort containing the most Sir2-responsive origins also showed the highest replication activity. The origin cohorts most effective at binding MCM when origin licensing was compromised by the cdc6-4 mutation were also the cohorts that showed the highest replication function in cdc6-4 cells, indicating that alterations in MCM distribution had an impact on replication dynamics.

Fig. 5.

Changes in MCM distribution coincided with changes in replication dynamics. (A) Sort-Seq values for 30-kb replicons centered on euchromatic origins parsed by their Trep values from ref. 21. Telomeric X origin replicons also shown. (B) As in A except with replicons parsed by euchromatic origins’ assigned Sir2 responsiveness (25). (C) Assigning Sort-Seq Z scores to distinct 1,001-bp origin fragments. (D) Euchromatic origins parsed into 10 distinct cohorts by their Trep values. Z scores are plotted from the earliest 10% (10th) to the latest 10% (100th) decile in SIR2 and sir2∆ cells. Telomeric X origin Z scores are also plotted. The Wilcoxon rank sum P values for the differences between the 50th and 60th and the 60th and 70th cohorts are indicated: **< 0.01 and ***< 0.001.

Sir2 Promoted the Normal Progression of Euchromatic Origin Replication in an Unperturbed S Phase.

The above analyses suggested that in CDC6 cells, Sir2 did not alter euchromatic origin function, yet Sir2 is important to completing replication of late-replicated regions by early-G2 phase (31). The dynamic range of the Sort-Seq analyses is limited (1-2n). In the spatiotemporal control of chromosome duplication, a more meaningful value is how specific genomic regions are duplicated relative to one another (37). Therefore, to examine the replication behavior of origin cohorts at higher resolution and over a greater dynamic range, we assigned Z scores to origin fragments using their Sort-Seq ratios (Fig. 5C), ranked the origins by their replication time, portioned the ranked origins into deciles, and displayed the Z scores for each decile in box-and-whiskers plots (Fig. 5D). Effects of Sir2 on euchromatic origin replication were revealed. First, origin function of 50th and 60th deciles (early/mid-S origins) differed more substantially in SIR2 cells than in sir2∆ cells. Second, the median values for the early-origin deciles (10th to 50th) in SIR2 cells were similar, whereas they continually decreased in sir2∆ cells, suggesting SIR2 imposed greater stochasticity of origin function (SI Appendix, Fig. S5). Third, in sir2∆ cells the difference in the behavior of the 60th and 70th ranked deciles became more acute, creating a gap in replication efficiency of these cohorts. Thus, SIR2 promoted the normal progression of euchromatic origin function in an unperturbed S phase.

SIR2 Was Required for Completing Duplication of Late-Replicating Euchromatin Independent of rDNA Replication Demands.

While the S Sort-Seq data differences were subtle, they revealed a Sir2 effect on euchromatic origin replication: preventing late-origin replication efficiency from falling further behind that of early origins and even trailing telomeric X origins (Fig. 5D). To challenge this outcome, we applied the Z-score approach to independent S Sort-Seq data from ref. 31 and observed similar alterations in the replication, strengthening the conclusion that normal progression of euchromatic origin function in S required SIR2 (SI Appendix, Fig. S6).

Sir2 can indirectly promote euchromatic origin activity by acting directly to form heterochromatin that inhibits the function of the rDNA origin, present in hundreds of copies within the yeast rDNA repeat array (30). Thus, in sir2∆ cells, the rDNA origins (r-ORIGIN) are more active, and because there are so many, the limiting origin activation factors are sequestered away from euchromatic origins. This sir2∆-caused rDNA sequestration of origin activation factors contributes to the incomplete duplication, or “replication gaps,” of late-replicating regions in early G2 phase (31). To assess whether sir2∆-induced MCM distribution changes might also be linked to these replication gaps, the effects of SIR2 on the replication of relevant origin cohorts in G2 cells was examined applying the Z-score approach to the data from ref. 31 (Fig. 6A). The cohorts produced the expected box-and-whiskers plots in S phase. However, in G2 phase, early- and late-euchromatic origins generated indistinguishable Sort-Seq Z scores near 0 in SIR2 cells, indicating equivalent duplication. In contrast, in sir2∆ cells, early- and late-euchromatic origins Z scores diverged, with the former generating Z scores above the population mean and the latter generating Z scores below the mean (Fig. 6A). Thus, while in SIR2 cells only the telomere X origins remained unduplicated in G2, in sir2∆ cells both late-euchromatic and telomeric X origins remained unduplicated.

Fig. 6.

SIR2 was required for completing replication of late-euchromatic regions independent of rDNA replication demands. (A) Z scores for the indicated origins from S- and early G2-phase SIR2 and sir2∆ cells derived from raw data in ref. 35 were determined as in Fig. 5C and displayed as box-and-whiskers plots for early-, mid, and late-euchromatic origins, and telomeric X origins. (B) Analyses as in A except with data from the sir2∆ r-origin* mutant. (C) Sir2 controls the relative replication probability of the three indicated origin cohorts at the levels of origin licensing efficiency (probability MCM load), and, at least in part via Sir2 control of the rDNA array, availability of origin activation factors (probability of MCM activation). While origin licensing and activation are distinct steps, loaded MCM complex is the substrate for origin activation factors. Thus alterations in origin licensing efficiencies affect the cohorts’ competitiveness for origin activation factors.

In sir2∆ cells with a cis mutation that weakens the rDNA origin, referred to here as r-origin*, rDNA-dependent sequestration of origin activation factors is reduced, allowing for partial rescue of sir2∆-induced replication gaps (31) (SI Appendix, Fig. S3). If changes in MCM distribution contributed to the sir2∆-induced incomplete duplication of late-euchromatic origins in G2 (Fig. 6A), then duplication of these elements should not be rescued by the r-origin* mutation, whereas duplication of telomeric X origins that had gained MCM in sir2∆ cells should be (Fig. 4). Therefore, the cohorts were assessed in sir2∆ r-origin* cells (Fig. 6B). In sir2∆ r-origin* cells, duplication of the telomeric X origin cohort in G2 was rescued (Fig. 6 A and B; compare early-euchromatic and telomeric X origins in G2 in sir2∆ r-ORIGIN cells to sir2∆ r-origin* cells), supporting the model wherein duplication of telomeric X origins in sir2∆ cells is limited by the availability of origin activation factors (31). In contrast, the late-euchromatic origins remained underduplicated even in sir2∆ r-origin* cells (Fig. 6 A and B, compare late-euchromatic origin cohort behavior in sir2∆ r-origin* to SIR2 r-ORIGIN cells). Thus, the r-origin* mutant substantially reduced the delayed replication of telomeric X but not late-euchromatic origins, providing evidence that that telomeric X origins’ gain of MCM complexes in sir2∆ cells was relevant to their enhanced replication efficiency over late-euchromatic origins. Notably, early-euchromatic origins also gained MCM complexes relative to late-euchromatic origins (Fig. 4), and the Sort-Seq data indicated that several of these origins generated higher-ranking Z scores in sir2∆ cells, indicative of enhanced replication efficiency. Two-dimensional origin mapping of early-euchromatic origin, ORI922 that gained MCM signals and acquired an enhanced S-phase Z score in sir2∆ cells, also revealed that its origin activity was enhanced in sir2∆ cells (SI Appendix, Fig. S7). Thus Sir2’s direct attenuation of origin licensing at telomeric X origins and early-euchromatic origins limited their origin function and prevented replication gaps from forming in late-replicating euchromatin.

Discussion

Recent evidence reveals that the yeast Sir proteins, known for their roles in forming yeast heterochromatin (also known as transcriptionally silent chromatin), also act directly within euchromatin where they can have an impact on chromosome replication (25, 27, 38). In particular, Sir2 and Sir3 act directly on nucleosomes adjacent to euchromatic DNA replication origins to inhibit origin licensing (25). However, the function of this Sir2-chromatin state has thus far only been assessed in cells where one of the core origin licensing proteins is defective (e.g., cdc6-4, orc5-1, mcm2-1 yeast) (25, 28, 29), leaving a physiological role for Sir2 in origin licensing an open issue.

Here, genomic and computational approaches revealed that in the absence of Sir2, MCM complexes shift toward early- relative to late-replicated euchromatin. Sir2’s negative effects on origin licensing depend on its catalytic deacetylation of nucleosomes (25, 28, 29). Therefore, the repressive properties of Sir2 chromatin (e.g., reduced DNA accessibility and nucleosome mobility) act to attenuate origin licensing by reducing the probability that a complete reaction will be completed. Any one or several discrete steps (e.g., ORC binding, Cdc6 binding, Cdt1-MCM association, etc.) could be affected, and to varying degrees at individual origins. Higher levels of Sir2 chromatin at Sir2-responsive origins indicate that the levels of this repressive state are likely one component that affects the probability of origin licensing, but other variables could also be relevant, enhancing the stochasticity of this attenuating mechanism (25).

In terms of cell physiology, a key point is that small reductions in the probability of origin licensing at any one origin act cumulatively over multiple origins to impose a genome-scale impact: origin density is concentrated in early-replicating euchromatin and at telomeres relative to late-replicating euchromatin. In essence, Sir2 prevents early-replicating origins from being as effective at origin licensing as they could be, thus limiting the collective origin activity that would otherwise be possible in early euchromatin (Fig. 6D). Because MCM complexes are the substrate for the limiting origin activation factors, and early-replicating euchromatin has active mechanisms for recruiting these factors, a shift in MCM complexes toward early-replicating euchromatin would be expected to lead to a concomitant shift in origin activation factors, resulting in more origin activity in early-replicating euchromatin than required for efficient duplication, and a depletion of origin activity from late-replicating euchromatin. The ultimate outcome is a reduced probability that late-replicating euchromatin will complete duplication in a timely manner (Fig. 6D). While this simple model helps explain the data, additional experiments will be required to address the relative contributions of Sir2’s distinct roles at the rDNA array and in origin licensing at non-rDNA origins to the regulation of genome duplication.

While Sir2-heterochromatin inhibition of origin function at telomeres is established, it was unknown whether this inhibition occurs at the level of origin licensing or activation (24). Thus, an important observation was that Sir2 attenuated origin licensing at telomeric X origins substantially, though clearly not completely. The increased licensing of telomeric X origins in sir2∆ cells provided a mechanism for them to compete more effectively than late-euchromatic origins for origin activation factors (Fig. 6D). In the absence of Sir2, both early-euchromatic and telomeric X origins were more competitive for MCM complexes, which should make them more competitive for origin activation factors. Notably, the rDNA origin itself did not show a relative gain in MCM signals in sir2∆ cells, in agreement with an earlier report (35). However, activation of a Sir2-suppressed transcription unit within rDNA in sir2∆ cells shifts MCM complexes to regions adjacent to the rDNA origin (35). The repeat nature of the rDNA origin and this reported transcription-dependent MCM-complex shifting limited our ability to interpret Sir2’s role in rDNA origin licensing by the approach used in this study.

There is a consensus that early origins exist within the most “open” and transcriptionally active regions of the genome (3, 8, 11, 18). Thus, it was initially perplexing that Sir2 chromatin, known for its ability to inhibit origin function, was present at higher levels and had a greater impact on origin licensing at early- compared to late-euchromatic origins. We posit that the same molecular properties that make early-euchromatic origins more accessible to origin regulatory machinery also make them more accessible to Sir proteins diffusing through the nucleoplasm. While Sir proteins are concentrated within heterochromatin domains, Sir proteins associate with these regions dynamically, releasing and rebinding (27, 39, 40). Released Sir proteins will have opportunities to sample other regions of the genome. The greater accessibility of early euchromatin might facilitate the more frequent sampling of early-euchromatic origins by “free-agent” Sir proteins (26, 41, 42).

Classic suppressor genetics uncovered Sir2 chromatin’s negative effect on origin licensing, a phenomenon that, like other negative forms of origin regulation, would be difficult to uncover without genetics (28, 43). Building on this foundation, the approaches used here to quantitatively analyze large cohorts of precisely mapped origins benefitted from the deeply annotated yeast genome and were essential to revealing the potential physiological relevance of Sir2-mediated attenuation of origin licensing. In a microbe-like yeast, such a genome-scale effect could impinge on organismal fitness by reducing cell proliferation rates and/or genome stability, as delayed replication enhances mutation frequency (44, 45). In mammalian cells, SirT1, the human ortholog of yeast Sir2, as well as Set8-mediated chromatin compaction, limit origin function and promote genome stability (46–48). Notably, Set8-modified chromatin limits MCM binding to chromatin. Thus, the use of repressive chromatin modifiers for achieving genome-scale control of origin licensing is emerging as an important regulatory node of eukaryotic chromosome duplication and stability. The tools available to budding yeast allowed us to show how such a mechanism could promote the normal distribution of origins across chromosomes. While the importance of origin distribution to genome stability has become clear, this study identifies a specific chromatin-mediated mechanism to establish the spatial distribution of MCM complexes across a eukaryotic genome (49–51).

Materials and Methods

Yeast strains were congenic derivatives of W303-1A and have been published (25).

Sequencing Data.

Raw data were assigned the following BioProject ID’s: PRJNA428768 and PRJNA601998 for reads from the MCM ChIP-Seq experiments; PRJNA601998 for reads from the S Sort-Seq experiments; and GEO accession GSE90151 for reads from the S and G2 Sort-Seq experiments (31).

Determining Per-Nucleotide MCM ChIP-Seq Signals and Distributions.

MCM ChIP-Seq data from three independent experiments were mapped to sacCer3 using Bowtie2 and default parameters. Duplicates were removed using MarkDuplicates within Picard tools. Per-nucleotide coverages were determined using Samtool’s BedCov. Coverages for each independent ChIP were normalized for sequencing depth and breadth (52). ChIP/input ratios were mapped to the nucleotides within origin-containing fragments, and internally scaled for each nucleotide by dividing each ratio by the median ChIP/input ratio measured between coordinates −600 to −400. Per-nucleotide signal distribution for an origin cohort was the per-nucleotide median ± 1 quartile of the distribution of scaled ratios for all origins within a cohort.

Determining H4K16ac Levels at Nucleosomes Adjacent to Euchromatic Origins.

H4K16ac ChIP/input ratios for each nucleotide within each nucleosome annotated in ref. 53 were summed. Six nucleosomes, three 5′ and three 3′ to the T-rich ORC binding site, were mapped to 1.2-kb fragments (Fig. 3A). Nucleosomes were also mapped to a control group of loci, 391 randomly selected distinct euchromatic 1.2-kb regions that were not annotated in the OriDB and lacked an ORCACS match. For each control fragment, the mean H4K16ac level for six nucleosomes, was used to normalize the H4K16ac levels for each nucleosome surrounding the experimental loci.

S-phase Sort-Seq experiments were performed on yeast growing at the permissive-growth temperature for cdc6-4 (23 °C) as described in ref. 36.

Data Availability.

Raw data and methods used in this study are available on public databases or provided within this manuscript.