Disruption of origin chromatin structure by helicase activation in the absence of DNA replication.

Prior to initiation of DNA replication, the eukaryotic helicase, Mcm2-7, must be activated to unwind DNA at replication start sites in early S phase. To study helicase activation within origin chromatin, we constructed a conditional mutant of the polymerase α subunit Cdc17 (or Pol1) to prevent priming and block replication. Recovery of these cells at permissive conditions resulted in the generation of unreplicated gaps at origins, likely due to helicase activation prior to replication initiation. We used micrococcal nuclease (MNase)-based chromatin occupancy profiling under restrictive conditions to study chromatin dynamics associated with helicase activation. Helicase activation in the absence of DNA replication resulted in the disruption and disorganization of chromatin, which extends up to 1 kb from early, efficient replication origins. The CMG holohelicase complex also moves the same distance out from the origin, producing single-stranded DNA that activates the intra-S-phase checkpoint. Loss of the checkpoint did not regulate the progression and stalling of the CMG complex but rather resulted in the disruption of chromatin at both early and late origins. Finally, we found that the local sequence context regulates helicase progression in the absence of DNA replication, suggesting that the helicase is intrinsically less processive when uncoupled from replication.

The stable DNA double helix must be melted and unwound into two single strands prior to the initiation of DNA replication. In eukaryotic organisms, the Mcm2-7 complex serves as the replicative helicase responsible for unwinding the duplex DNA at the origin and for continued separation of strands ahead of the replisome at the replication fork. Mcm2-7 is loaded at replication start sites during G1 phase of the cell cycle and is made fully active by the association of Cdc45 and the GINS complex to form the CMG (Cdc45–Mcm2-7–GINS) holohelicase in early S phase (Bell and Labib 2016). While genetic and biochemical studies have begun to elucidate the factors that are necessary and sufficient for helicase activation, we have little understanding of how the active helicase initially melts the duplex DNA in the context of the local chromatin environment or of the mechanisms that regulate CMG progression through chromatin.

Prior to helicase activation, replication origins must be selected and licensed. The conserved origin recognition complex (ORC) selects potential start sites (Bell and Stillman 1992). During G1 phase of the cell cycle, at least two copies of Mcm2-7 are recruited to each origin and loaded by ORC, Cdt1, and Cdc6 (Tanaka and Diffley 2002; Randell et al. 2006; Remus et al. 2009). These Mcm2-7 complexes encircle the DNA helix and interact to form double hexamers (Remus et al. 2009). In early S phase, DDK and S-CDK phosphorylation recruit various factors to the helicase including Cdc45, polymerase ε, and the GINS complex; together, these form the preinitiation complex (Tanaka et al. 2007; Zegerman and Diffley 2007; Heller et al. 2011). At this point, the Mcm2-7 double hexamer separates into two independent CMG complexes (Miyazawa-Onami et al. 2017; Douglas et al. 2018). Mcm10 association is necessary to activate the CMG holohelicase to initially unwind DNA (Kanke et al. 2012; Van Deursen et al. 2012; Watase et al. 2012). Now active, the CMGs must unwind the DNA double helix and pass each other on single-stranded DNA (ssDNA) in order for Mcm2-7 to travel with the correct orientation on the leading strand (Georgescu et al. 2017; Douglas et al. 2018). Following origin unwinding, replication begins on both the leading and lagging strands with the synthesis of primers by polymerase α, which are extended by polymerases δ and ε (Bell and Labib 2016).

The local chromatin environment at replication origins is important for origin selection and helicase loading (Gutiérrez and MacAlpine 2016). In budding yeast, ORC binds to a T-rich sequence (ACS) (Broach et al. 1983; Bell and Stillman 1992) that is necessary but not sufficient for origin function (Breier et al. 2004). In addition to the primary sequence, the local chromatin structure is also thought to contribute to origin activity (Thoma et al. 1984; Berbenetz et al. 2010; Eaton et al. 2010). In higher eukaryotes, open chromatin organization serves as the major determinant of origin identity (MacAlpine et al. 2010; Miotto et al. 2016; Long et al. 2020). Experiments at the yeast origin ARS1 initially established that specific nucleosome positioning around ARS1 is necessary for origin function (Simpson 1990). ORC binding establishes a precise positioning for the nucleosomes flanking origins across the budding yeast genome (Berbenetz et al. 2010; Eaton et al. 2010). During G1 phase, a Cdc6-dependent shift in the flanking nucleosome occurs around the origin, suggesting that Mcm2-7 loading expands the nucleosome-free region (NFR) at replication start sites (Belsky et al. 2015). Although the minimal set of replication and chromatin factors necessary for origin activation have been identified on chromatinized templates in vitro (Azmi et al. 2017; Devbhandari et al. 2017; Kurat et al. 2017), the chromatin dynamics in early S phase that accompany preinitiation complex formation and helicase activation remain unaddressed.

Prior to the start of replication elongation, helicase activation and subsequent DNA unwinding in early S phase serve as a model for the uncoupling of helicase and polymerase activity. In eukaryotes, helicase–polymerase uncoupling was initially observed in Xenopus extracts when aphidicolin treatment to inhibit polymerases α and δ caused extensive negative supercoiling, indicating that DNA could be unwound by the helicase even in the absence of replication (Walter and Newport 2000; Pacek and Walter 2004). Subsequently, uncoupling of the helicase and leading strand polymerization have been observed in multiple contexts that disrupt replication, including polymerase inhibition, DNA damage, and dNTP depletion (Sogo et al. 2002; Katou et al. 2003; Byun et al. 2005; Nedelcheva et al. 2005; Tanaka et al. 2009; Gan et al. 2017; Ercilla et al. 2020). Helicase–polymerase uncoupling functions as a common consequence and signal to the intra-S-phase checkpoint across these mechanistically distinct perturbations to genome replication (Byun et al. 2005). Whether the uncoupled helicase also stalls through a common mechanism remains unclear. Rad53 appears to have a regulatory role in limiting DNA unwinding following uncoupling of the replisome in response to DNA damage or hydroxyurea (Lopes et al. 2006; Gan et al. 2017; Devbhandari and Remus 2020). Recent in vitro work in the context of UV lesions has suggested that the helicase slows and eventually stalls following uncoupling simply due to the absence of leading strand replication behind the helicase (Taylor and Yeeles 2019). In E. coli, this slowing of the helicase has also been observed when uncoupled from replication and has been termed a “dead man's switch” (Graham et al. 2017). The intrinsic or extrinsic mechanisms that ultimately halt CMG progression through chromatin following uncoupling remain unresolved in the context of in vivo chromosomal DNA replication.

In studies of helicase activation and helicase–polymerase uncoupling, the following open questions remain: (1) How is local chromatin impacted by helicase activation? Is this similar at all origins, or does it vary based on the characteristics of individual origins? (2) What factors influence helicase movement and processivity when uncoupled from replication in vivo? (3) How does the cell respond to helicase–polymerase uncoupling outside of the contexts of DNA damage and replication stress? Can cells repair any unreplicated gaps left by this uncoupling? Using a conditional genetic system to target the function of polymerase α along with nucleotide-resolution genomic assays, we have captured helicase activation within chromatin at origins genome-wide, which has also allowed us to study the consequences of helicase–polymerase uncoupling in vivo in the absence of external damaging agents. Given the conservation of the CMG complex and its activity from the budding yeast to mammals (Aves et al. 2012), our results likely have implications for helicase activation and helicase–polymerase uncoupling within chromatin context in all eukaryotes.

Results

Capturing the active CMG complex at start sites of DNA replication

The formation of the CMG complex, helicase activation, DNA unwinding, and initiation of DNA replication by polymerase α priming are tightly coupled events that must occur in the context of chromatin. To better understand the chromatin dynamics associated with the transition to an active CMG complex prior to initiation of DNA replication, we sought to block the priming of leading and lagging strand DNA replication by DNA polymerase α. In the absence of Polα, the CMG complex still assembles at origins and can unwind DNA (Kanke et al. 2012; Van Deursen et al. 2012; Watase et al. 2012; Miyazawa-Onami et al. 2017; Douglas et al. 2018).

To prevent priming by Polα and thus inhibit the initiation of DNA replication, we used temperature-sensitive alleles of the catalytic domain of Polα in combination with the anchor-away system to abrogate Polα function (Haruki et al. 2008). Upon rapamycin addition, the anchor-away system rapidly depletes a protein of interest from the nucleus using components of the mTOR pathway. Using a CRISPR-Cas9 approach to mutate CDC17, we constructed cells containing the previously characterized allele cdc17-1 (Hartwell et al. 1973; Lucchini et al. 1990) as well as a new mutation (G791D). We initially characterized these alleles across restrictive conditions (temperature and/or rapamycin addition) (Fig. 1A). Cells with temperature-sensitive mutations and an anchor-away tag, cdc17-ts-FRB, exhibit the greatest sensitivity to the restrictive conditions (37°C + rapamycin).

Figure 1.

cdc17-ts-FRB cells arrest prior to replication under restrictive conditions and recover from prolonged arrest with a delay in G2/M phase. (A) Spot assay across permissive and restrictive conditions for WT and conditional cdc17 alleles. (B) Schematic of experimental designs for arrest and recovery experiments. (C) DNA content measured by flow cytometry in cdc17-ts-FRB cells under permissive (24°C) and restrictive conditions (37°C + rapamycin) during G1 phase and up to 120 min after release from G1. (D) Copy number across chromosome XV during G1 or 60 min after release from G1, under permissive (24°C) or restrictive (37°C + rapamycin) conditions. Open circles along the x-axis denote the positions of early origins. (E) DNA content measured by flow cytometry for cdc17-ts-FRB cells under permissive (24°C) and restrictive conditions (37°C). Cells were synchronized in G1 phase and either released for 1 h at 37°C and then returned to 24°C for the remaining time, or released at 24°C. (F) Aggregate copy number at early origins. Following the time course shown in B, cells were exposed to 37°C for 1 h or left at 24°C following G1 phase. Samples were either collected at 3 h with nocodazole (Nz) added upon release from G1, or following an overnight (o/n) recovery without nocodazole.

To ensure that DNA replication was completely inhibited by our approaches to remove Polα function, we assessed the progression of cdc17-ts-FRB cells through the cell cycle using flow cytometry. Cells were synchronized in G1 phase using α-factor, and monitored for 2 h following release from G1 for entry into S phase (Fig. 1B,C; Supplemental Fig. S1A). Twenty minutes after release from G1 synchronization, cells still have G1 DNA content in both permissive and restrictive conditions, indicating that they are recovering from the α-factor arrest. At permissive conditions (24°C), cells begin to enter S phase by 40 min. At 60 min after release from α-factor, most cells are in G2 phase, and by 2 h they return to an asynchronous cycling pattern. Under restrictive conditions (37°C + rapamycin), cells remain arrested, with DNA content equivalent to G1 phase, even after 2 h suggesting that no or minimal chromosomal replication occurs under restrictive conditions. To confirm this observation, we examined copy number genome-wide to determine whether any limited genome duplication was occurring (Fig. 1D). Plotting copy number across chromosome XV, we did not observe a copy number increase at early origins 60 min after cells were released from G1 arrest at 37°C with rapamycin. In contrast, at 24°C, copy number increases ∼50% at early origins 60 min after release from G1 phase. These results are representative of the pattern across other chromosomes. Taken in combination with our flow cytometry data, these experiments suggest that no or extremely minimal replication is occurring, making cdc17-ts-FRB cells a useful in vivo model of helicase activation prior to the onset of replication.

In order to assess whether restoration of Polα would allow recovery of replication following a prolonged arrest at 37°C, we synchronized cells as shown in Figure 1B and released them into 24°C or 37°C. Rapamycin was not used in the recovery experiments due to the difficulty in reversing the sequestration of the anchor-away-tagged protein in the cytosol. After 1 h, we returned cells to permissive conditions (24°C) and followed their progression through the cell cycle (Fig. 1E; Supplemental Fig. S1B). Control cells that were never exposed to the restrictive conditions progress through S phase and G2/M phase following release from α-factor. In contrast, cells that experienced the restrictive temperature exhibit a prolonged S phase and an arrest at G2/M phase; however, by 7 h, the cells eventually return to an asynchronous cell cycle distribution. Thus, loss of Polα function causes a tight but reversible arrest prior to replication. The subsequent G2-phase arrest experienced by cells recovering Polα function suggests activation of the G2/M checkpoint to allow for completion of DNA replication and/or repair of DNA damage.

In the absence of DNA replication priming, the Mcm2-7 helicase may become active and start to unwind origin DNA, which has the potential to result in unreplicated gaps at replication origins. In order to assess whether there are any unreplicated gaps or disruptions to the replication program that might cause activation of the G2/M checkpoint following recovery, we released cells from G1 phase into restrictive conditions (37°C) for 1 h and then shifted cells into 24°C for 2 h, or released cells from G1 phase into 24°C for the same amount of time as a control. Nocodazole was added upon release from G1 to both sets of cells to catch cells in G2/M phase and prevent additional rounds of replication from impacting origin copy number (Supplemental Fig. S1C). We then collected genomic DNA to examine copy number at origins in these cells 3 h after release from G1 phase (Fig. 1F). If there is helicase activation and unwinding in the absence of DNA replication, then, when permissive conditions are restored, we would expect copy number to be depleted at the origin in cells exposed to the restrictive condition. Alternatively, if there is no helicase activation or unwinding, then we would expect uniform copy number across origin sequences. We found an ∼35% decrease in copy number at replication origins (20% and 50% in each independent biological replicate) (Supplemental Fig. S1D), suggesting that delay in S-phase progression and G2/M arrest is due to the presence of unreplicated gaps. In cells released from G1 phase into permissive conditions and blocked in nocodazole, there was no decrease in copy number at early origins. As an additional control, we assessed copy number in cells exposed to restrictive conditions or control cells after allowing the cells to grow overnight (∼19 h following G1 phase) and potentially repair any gaps present at origins. Copy number was uniform in both populations of cells following the overnight recovery, supporting that the decrease in copy number at early origins at 3 h is a repairable event and specific to cells exposed to restrictive conditions.

Overall, we conclude that in cdc17-ts-FRB cells, there is likely no replication at restrictive conditions and that an unreplicated gap is left at the origin when cells recover from the absence of Polα, necessitating a delay in G2/M phase to repair these gaps. If the helicase can unwind DNA prior to the onset of replication, it could progress away from origins and leave behind the observed gap once replication begins. We thus sought to investigate the state of replication start sites under restrictive conditions in cdc17-ts-FRB cells, beginning with chromatin structure.

Origin chromatin is disrupted in the absence of Cdc17 function

Helicase activation has been modeled in vitro and at single origins in vivo (Miyazawa-Onami et al. 2017; Douglas et al. 2018). However, it has never been assessed in high resolution at origins across the genome. Additionally, despite the importance of local chromatin in origin function, the chromatin dynamics that accompany helicase activation have never been examined. We used a factor-agnostic micrococcal nuclease (MNase) chromatin profiling assay to study chromatin dynamics during helicase activation (Henikoff et al. 2011). To collect chromatin samples for this assay, we used the experimental design shown in Figure 1B, harvesting cells in G1 and at 20, 40, and 60 min after release from G1, both at 24°C and 37°C with rapamycin. Based on the FACS analysis (Fig. 1C), cells are recovered from G1 arrest and in S phase by 40 min following release from α-factor. The resulting chromatin samples were digested with MNase, and DNA fragments between 20 and 250 bp in length were recovered. By plotting these fragment lengths as a function of chromosomal position, we can infer the positions of both nucleosomes and small DNA-bound proteins, such as replication and transcription factors. Nucleosomes typically protect DNA fragments ∼140–180 bp in length, while small binding factors or subnucleosomal structures protect fragments <100 bp. For example, during S phase (both at 24°C and 37°C + rapamycin), we observed new small factors, likely transcriptional regulators, binding upstream of the gene RNR1, which encodes a ribonucleotide reductase subunit (Supplemental Fig. S2A). Additionally, chromatin structure is disrupted throughout the gene body, indicated by the lack of clearly phased nucleosomes and the appearance of broadly dispersed smaller fragments consistent with elevated transcription of RNR1 in S phase (Elledge and Davis 1990) and the disassembly and displacement of nucleosomes.

Based on the extensive DNA unwinding observed in vitro in studies of helicase–polymerase uncoupling and the transcriptional chromatin dynamics our assay can capture, we predicted that helicase activation and DNA unwinding would specifically disorganize chromatin structure around origins. We first examined chromatin structure around early and late replication origins in cells released into S phase under permissive conditions (24°C), and noted that by 40 and 60 min there was a general increase in read coverage at early origins attributable to the onset of replication, but there were no specific chromatin structure changes (Supplemental Fig. S2B). We next surveyed the genome for disorganized chromatin structure at origins under the restrictive conditions following release from α-factor and identified specific origins that exhibited a time-dependent change in chromatin organization as they entered S phase. For example, the early-activating origin ARS922 on chromosome IX exhibits precisely positioned nucleosomes flanking the origin and an ORC-dependent small DNA-binding factor footprint (Belsky et al. 2015) at the ACS in G1 and 20 min following release from α-factor (Fig. 2A). However, by 40 min (when cells should be entering S phase), we saw a disorganization in the chromatin flanking the origin that persists for at least 1 h following release. Specifically, we observed an increase in fragments smaller than a nucleosome that are broadly distributed and lack any detectable organization or phasing, suggesting that there is disruption of nucleosomes and an increase in subnucleosomal-sized fragments and/or increased accessibility of nucleosomal DNA. To quantify this increase in smaller fragments, we first plotted the fragment size distribution for each time point and calculated the change in <100-bp fragments ±1 kb from the origin. We found an increase in the number of <100-bp fragments in the chromatin after 40 min compared with G1 phase (e.g., +200% in Fig. 2A, ±1 kb from the origin), while nucleosome-sized fragments occur in a similar number (−14% in Fig. 2A).

Figure 2.

Chromatin disorganization occurs at select replication origins under restrictive conditions in cdc17-ts-FRB cells. (A) Chromatin profiles for ARS922 in cdc17-ts-FRB cells at 37°C + rapamycin, in G1 phase and 20, 40, and 60 min following release from G1. Fragment midpoints are plotted by fragment length and genomic position relative to the origin. Shading of each point is determined by a 2D kernel density estimate. Gene bodies are shown in gray across the top. The 2D Shannon entropy estimate in bits (log2) for the 1-kb window around the origin is inset on each panel. The distribution of fragment sizes ±1 kb from the origin, excluding the ORC footprint, is plotted at the right of each chromatin profile, with the ratio of 20- to 100-bp fragment counts relative to G1 printed on the plot. (B) Chromatin profiles for ARS405.5 in cdc17-ts-FRB cells at 37°C + rapamycin, in G1 phase and up to 60 min after release from G1.

Although the number of nucleosome-sized fragments does not change significantly, nucleosome-sized fragments are detected in the linker regions between G1 nucleosome positions and even in the nucleosome-free region at the origin center. These changes may indicate loss of nucleosome positioning and an accumulation of partially unwrapped nucleosomes and/or tetrasomes (Rhee et al. 2014; Ramachandran et al. 2017). However, other origins, such as ARS405.5, a late origin on chromosome IV, have a static chromatin organization across all time points (Fig. 2B). Additionally, control loci, such as binding sites for the transcription factor Abf1, also remain unchanged (Supplemental Fig. S2C). We conclude that there is an origin-specific chromatin disorganization occurring in the absence of replication upon release of cells from G1 phase and that this chromatin change is limited to a subset of replication start sites.

We hypothesized that if the observed chromatin disorganization resulted from the presence of active CMG complexes at origins, it should be limited to early, efficient origins. In the absence of replication, the intra-S-phase checkpoint will likely be activated by DNA unwinding at early origins, leading to suppression of late origins through the phosphorylation of Sld3 and Dbf4 (Zegerman and Diffley 2010). To agnostically identify disorganized origins, we developed an entropy score based on information theory to holistically capture chromatin organization (Tran et al. 2021). In organized chromatin, fragment midpoints resulting from our MNase assay tend to be well-organized at small factor (e.g., ORC) binding sites or in well-phased arrays of nucleosome dyads, while midpoint fragments recovered from disorganized chromatin are less structured and more randomly distributed. We estimated the 2D Shannon entropy around origins based on the distribution of fragment midpoints surrounding each origin (as shown in Fig. 2) and used the difference in entropy between G1 and 40-min chromatin as a discriminating score (see the Supplemental Material). A subset of 239 replication start sites that bind ORC during G1 and G2 phases was used in this analysis (Belsky et al. 2015). We assigned each origin an entropy score and plotted the distribution of these scores (Fig. 3A). Origins were classified as early activating if they activated in the presence of hydroxyurea and all others were designated as late origins (Belsky et al. 2015). The bimodal distribution of origin entropy scores demonstrates that early-firing origins have a greater entropy score, and thus more chromatin disorganization, than origins that activate later in S phase.

Figure 3.

Early, efficient origins have disorganized chromatin extending up to 1 kb away from the origin under restrictive conditions. (A) Distribution of entropy scores across early and late origins. The score represents the entropy difference between chromatin in G1 phase and chromatin 40 min after release from G1 under restrictive conditions. (B) Heat map of 239 replication origins in cdc17-ts-FRB cells at 37°C + rapamycin, ordered by decreasing entropy score. The midpoints of <100-bp fragments resulting from MNase digestion were counted around each origin and are plotted in 100-bp bins. The activation time and efficiency for each origin is annotated at the left. Origin timing indicates a binary early or late designation based on whether or not an origin activates in HU. Time labels reflect minutes following release from G1 phase.

We next ordered origins based on this score and visualized the increased entropy by only plotting those fragments with a length of <100 bp that were overrepresented in the disorganized chromatin (Fig. 3B). We chose to plot <100-bp fragments as a visual proxy for the broader chromatin disorganization. Timing and efficiency annotations for each origin further confirm that early-activating, efficient origins have disorganized chromatin relative to G1 phase (McGuffee et al. 2013; Belsky et al. 2015). In contrast, late, inefficient origins retain their G1 chromatin structure. The increase in chromatin disorganization as reflected by the entropy score extends up to 1 kb on either side of the origins. As observed at ARS922 (Fig. 2A), chromatin disorganization does not occur until 40 min after release from G1 arrest, and it persists at 60 min. The increased entropy observed at early, efficient origins only occurs under restrictive conditions at origins; it is not present at origins under permissive conditions (Supplemental Fig. S3A) or at Abf1 binding sites (Supplemental Fig. S3B). These results support that the chromatin disorganization we observed is dependent on the formation of the preinitiation complex, since the phenomenon is limited to early, efficient origins.

CMG movement and ssDNA accumulation define the extent of chromatin disorganization around origins

The observation that chromatin disorganization is limited to early, efficient origins in a Polα conditional mutant suggests that helicase activation and DNA unwinding might be responsible for these chromatin dynamics. If this is the case, Mcm2-7 would be detected outside the origin center where it is loaded in G1 phase, likely 1 kb from the origin center. To test this hypothesis, we performed Mcm2-7 ChIP-seq across the same time course in cdc17-ts-FRB cells and examined Mcm2-7 ChIP enrichment at sequences surrounding origins of replication. To visualize the data, we ordered origins by their Mcm2-7 ChIP signal at 40 min and annotated each origin with corresponding entropy scores from Figure 3. In G1 phase, Mcm2-7 is located at the origin center (ACS) across all origins (Fig. 4A). Beginning 40 min after release from G1 synchronization, Mcm2-7 is enriched ∼1 kb upstream of and downstream from early, efficient origins with high entropy scores, indicating this is the same set of origins that experience chromatin disorganization (Fig. 4B). This pattern of Mcm2-7 enrichment only occurs under restrictive conditions (Supplemental Fig. S4A,B). The bidirectional enrichment of Mcm2-7 suggests that the double hexamer has separated, which occurs when the CMG complexes are fully formed (Douglas et al. 2018). To confirm that the complete CMG complex, rather than Mcm2-7 alone, is moving away from the origin, we performed ChIP-seq for FLAG-tagged Pol2, a subunit of polymerase ε (Fig. 4C). Polε is recruited to origins during preinitiation complex formation following S-CDK phosphorylation along with the GINS complex and travels with the active CMG (Heller et al. 2011; Sengupta et al. 2013; Yu et al. 2014). We observed Polε enrichment in the same pattern as the Mcm2-7 complex flanking the origin 1 kb upstream and downstream, suggesting the intact CMG complex is moving bidirectionally away from these origins in the absence of Polα. These results support our hypothesis that the active holohelicase complex is able to unwind DNA and perturb chromatin structure prior to the onset of replication.

Figure 4.

CMG movement and ssDNA accumulation define the extent of chromatin disorganization. (A,B) Mcm2-7 ChIP-seq enrichment normalized to input across origins during G1 phase (A) and 40 min after G1 (B) at 37°C + rapamycin in cdc17-ts-FRB cells. All panels are ordered by Mcm2-7 ChIP signal in B. Enrichment plotted in 1-bp bins. Timing, efficiency, and entropy score for each origin from Figure 3 are annotated at the left. (C) Pol2 (Polε subunit) ChIP-seq enrichment normalized to input at origins 40 min after G1 phase. (D) Rfa1 (RPA subunit) ChIP-seq enrichment normalized to input at origins 40 min after G1.

During G1 the Mcm2-7 double hexamer can translocate along double-stranded DNA (dsDNA) when pushed by an RNA polymerase (Gros et al. 2015). Thus, it is possible that the CMG complex could be moved in a similar way after double hexamer separation without unwinding DNA, although this has not been previously observed in the context of the CMG complex. To determine whether ssDNA was produced by CMG movement, we performed ChIP-seq using a tagged allele of RFA1, a subunit of the ssDNA binding complex RPA. Plotting Rfa1 enrichment at 40 min alongside the Mcm2-7 and Polε ChIP data, we observed that Rfa1 is enriched on DNA in a similar pattern flanking the same set of origins (Fig. 4D). The enrichment of Rfa1 at sequences flanking early origins is specific to the restrictive conditions and does not occur under permissive conditions (Supplemental Fig. S4C). Our data suggest active CMG complexes are unwinding DNA at a subset of origins, and this ssDNA persisted long enough for us to observe RPA binding. This unwinding appears to be limited to a 1-kb window around origins, suggesting that there are mechanisms that halt CMG movement when uncoupled from replication.

The intra-S-phase checkpoint does not regulate CMG progression in the absence of replication

When tracts of ssDNA accumulate following replication stress, RPA binds to this ssDNA. At a certain threshold, this ssDNA-RPA signal activates the DNA replication arm of the intra-S-phase checkpoint (Pardo et al. 2017). The DNA replication checkpoint is mediated through Mrc1 and leads to the phosphorylation and activation of the effector kinase Rad53. Rad53 suppresses late origin firing, along with other actions that halt further replication, and has recently been shown to regulate the replisome during helicase–polymerase uncoupling in vitro (Tercero and Diffley 2001; Devbhandari and Remus 2020).

The persistent RPA accumulation we observed in cdc17-ts-FRB cells, which occurs at early and not late origins, suggests activation of the intra-S-phase checkpoint. We hypothesized that Mcm2-7 movement might be regulated by the activated intra-S-phase checkpoint in the absence of replication. In the absence of the checkpoint, we predicted that Mcm2-7 might continue to progress >1 kb from origins and that Mcm2-7 movement would occur at both late and early origins. To test these predictions, we constructed cdc17-ts-FRB MRC1-FRB cells to examine helicase movement in the absence of the intra-S-phase checkpoint; similar to cdc17-ts-FRB cells, these cells arrest with G1 DNA content under restrictive conditions as measured by flow cytometry and DNA copy number analysis (Supplemental Fig. S5A,B). To assess checkpoint status in cdc17-ts-FRB cells alone and in combination with MRC1-FRB, we performed immunoblots for Rad53 in both strains (Fig. 5A). In cdc17-ts-FRB cells, Rad53 is phosphorylated beginning 40 min after release from G1 arrest, corresponding to the time we observed chromatin disorganization, CMG movement, and RPA binding around origins. Thus, the amount of ssDNA observed in Figure 4 is sufficient to activate the intra-S-phase checkpoint. Rad53 phosphorylation does not occur at any time point in cdc17-ts-FRB MRC1-FRB cells under restrictive conditions, indicating that intra-S-phase checkpoint activation is blocked by depletion of Mrc1 from the nucleus. We performed our MNase chromatin profiling assay and Mcm2-7 ChIP-seq in cdc17-ts-FRB MRC1-FRB cells to assess Mcm2-7 movement and chromatin disorganization in the absence of the checkpoint. To compare Mcm2-7 ChIP enrichment across origins, we ordered origins based on the Mcm2-7 ChIP signal in cdc17-ts-FRB cells. As expected, Mcm2-7 occupancy and movement in the absence of the intra-S-phase checkpoint is distributed across a larger set of origins compared with cells with the checkpoint intact (Fig. 5B). However, Mcm2-7 movement is still limited to 1 kb on either side of the origins (Supplemental Fig. S5C).

Figure 5.

The intra-S-phase checkpoint is activated in the absence of Cdc17 function but does not influence the extent of Mcm2-7 movement. (A) Western blot for Rad53 and GAPDH/Tdh1 in cdc17-ts-FRB and cdc17-ts-FRB MRC1-FRB cells, at 37°C + rapamycin following release from G1 synchronization. The top Rad53 band indicates phosphorylated Rad53. (B) Heat map of Mcm2-7 ChIP-seq enrichment normalized to input in 1-bp bins at origins in cdc17-ts-FRB cells at 37°C + rapamycin, with and without MRC1-FRB. Origins are ordered by decreasing Mcm2-7 enrichment level at 40 min in cdc17-ts-FRB cells. The difference panel was calculated by subtracting the Mcm2-7 signal in cdc17-ts-FRB MRC1-FRB from the Mcm2-7 signal in cdc17-ts-FRB. (C) Chromatin profiles for late origin ARS603 and early origin ARS1114 in cdc17-ts-FRB MRC1-FRB cells, in G1 phase and 40 min after release from G1 at 37°C + rapamycin. The entropy value for ±1 kb around each origin is shown in the top left of each panel. The distribution of fragment sizes ±1 kb is plotted at the right of each chromatin profile, as in Figure 2, with the ratio of 20- to 100-bp fragment counts relative to G1 printed on the plot. Gene bodies are depicted in gray above each plot. Points are shaded as in Figure 2.

When we analyzed origin chromatin in our MNase data using the entropy score, we found that chromatin disorganization, likely due to the presence of an active helicase, now occurs at both early and late replicating origins. (Supplemental Fig. S5D). For example, ARS603 is a late origin that gained disorganized chromatin in cdc17-ts-FRB MRC1-FRB cells (Fig. 5C). Early origins, such as ARS1114, still exhibit chromatin disorganization. Thus, the intra-S-phase checkpoint is involved in halting the activation of late replication origins in the absence of replication, similar to its role under replication stress conditions. Although the helicase is active at most origins, the same limits on Mcm2-7 unwinding and movement persist, suggesting that other factors must influence helicase movement and DNA unwinding in the absence of replication.

CMG movement stalls in a sequence-dependent manner

Since the intra-S-phase checkpoint is activated by Mcm2-7-dependent DNA unwinding in the absence of replication but does not appear to limit the movement of Mcm2-7, we speculated that local chromatin features or sequence elements surrounding origins may hinder helicase progression. We first examined the local chromatin environment surrounding origins. If nucleosomes act as barrier elements, Mcm2-7 peaks should accumulate at these features. We selected highly enriched Mcm2-7 peaks on either side of the origins at 40 min after release from G1 phase. However, when we looked at nucleosome and Mcm2-7 occupancy, we did not observe specific enrichment of Mcm2-7 at nucleosome dyads or in linker regions (Supplemental Fig. S6A). Thus, nucleosome positioning does not appear to be the ultimate reason Mcm2-7 stalls, though it may still have some inhibitory effect on DNA unwinding.

We next examined the sequence content surrounding Mcm2-7 peaks. Replication origins in S. cerevisiae share an AT-rich DNA unwinding element (Umek and Kowalski 1988), and therefore GC content increases with distance from the origin until it reaches the genome-wide average. If sequence context inhibits origin unwinding, Mcm2-7 peaks would likely stall outside the easily unwound and less thermally stable AT-rich regions (Umek and Kowalski 1988; Natale et al. 1992; Ak and Benham 2005). We first plotted aggregate GC content and Mcm2-7 ChIP enrichment surrounding origins in cdc17-ts-FRB cells (Fig. 6A). GC content at replication origins decreases from the genome-wide average of 37% to 20%. Mcm2-7 is enriched just beyond the edges of this AT-rich origin region where the GC content returns to the genome-wide average. We also observed that the Mcm2-7 enrichment pattern closely aligns with the return to a normal copy number following recovery from 37°C, first shown in Figure 1F. Together, these results suggest that replication begins where Mcm2-7 stalls, leaving behind an unreplicated gap.

Figure 6.

The active CMG complex unwinds the AT-rich origin region and stalls as it moves through sequences with higher GC content. (A) Aggregate GC content and Mcm2-7 ChIP enrichment at 37°C + rapamycin at 239 origins in cdc17-ts-FRB cells. Aggregate copy number from Figure 1F, 3 h after G1 in nocodazole and exposed to 37°C, is shown for comparison. GC content calculated in 30-bp bins. Mcm2-7 ChIP enrichment shown in 1-bp bins. (B, left) Heat map of Mcm2-7 ChIP-seq enrichment at origins 40 min after G1 under restrictive conditions in cdc17-ts-FRB MRC1-FRB cells. Origins are ordered by decreasing width of the AT-rich region at origins. (Right) AT content around origins. The 3-mers containing only A/T are shaded purple; all other 3-mers are white. (C) GC content and distance from origin for each Mcm2-7 ChIP-seq peak in cdc17-ts-FRB cells during G1 phase and 40 min after release from G1, 37°C + rapamycin. GC content is calculated from a 100-bp window around each Mcm2-7 peak. (D) Proposed model for helicase activation without replication and recovery following restoration of Polα.

To further examine Mcm2-7 positioning relative to the origin sequence, we ordered origins based on the length of their AT-rich regions and plotted Mcm2-7 ChIP signal in cdc17-ts-FRB MRC1-FRB cells (Fig. 6B). Mcm2-7 is not entirely excluded from the AT-rich region, and there is not a sharp boundary at the transition to elevated GC content. Together, our data indicate that active CMG complexes, when uncoupled from DNA replication, gradually stall when encountering increased GC content, as opposed to stalling at the immediate transition from AT-rich regions to elevated GC content. To more precisely compare Mcm2-7 movement and GC content, we plotted each Mcm2-7 peak by GC content and distance from the origin (Fig. 6C). G1-phase Mcm2-7 peaks tend to be close to the origin, with a median GC content of 34%. Conversely, 40 min after release from G1 phase under restrictive conditions, Mcm2-7 peaks have a significantly higher median GC content of 41% (P-value < 2.2 × 10−16, t-test) and are located farther from the origin center than the G1 peaks. Overall, we conclude that the active CMG moves away from the origin to regions of higher GC content after release from G1 arrest in the absence of replication.

Discussion

Replication start site selection, licensing, and activation occur within the local chromatin environment, and we have previously addressed how ORC binding and Mcm2-7 loading impact origin chromatin (Eaton et al. 2010; Belsky et al. 2015). Mechanistic studies of helicase activation have established that the two CMG complexes move apart at the origin prior to replication initiation (Douglas et al. 2018). The active helicase is also able to unwind kilobases of DNA when uncoupled from replication elongation (Walter and Newport 2000; Pacek and Walter 2004). Taken together, this evidence suggests that once active, the CMG has the potential to extensively unwind DNA and disrupt chromatin at the origin even in the absence of nascent strand synthesis. We have used multiple genomic approaches in a polymerase α conditional mutant to study both helicase activation and helicase–polymerase uncoupling within origin chromatin.

Activation of CMGs in the absence of replication results in chromatin disorganization at early, efficient origins characterized by a loss of nucleosome positioning and an increase in entropy. The increased entropy observed at replication origins in the absence of polymerase α function is analogous to the chromatin changes observed at highly transcribed genes upon induction, such as RNR1. The elevated entropy within gene bodies is driven by the continual unwinding by the RNA polymerase that disrupts nucleosome structure (Tran et al. 2021). In contrast, the increased chromatin disorganization and entropy at replication origins is likely driven by CMG-dependent DNA unwinding, which would only unwind a region once in a given cell. Given the analogous chromatin dynamics at transcribed genes, it is likely that the observed dynamics are caused by changes in a common element, the nucleosome. At both sets of genomic loci, the small fragments likely reflect partially unwrapped nucleosomes or small structures such as tetrasomes, as has been postulated previously for fragment sizes <150 bp (Rhee et al. 2014; Ramachandran et al. 2017). Partially unwrapped nucleosomes would be more vulnerable to MNase digestion closer to the nucleosome dyad, resulting in shorter protected fragments, while subnucleosomal structures would simply protect fewer base pairs of DNA.

Despite the increase in entropy, well-positioned nucleosomes can still be discerned around the origin, suggesting a mixed population of chromatin organizational states. This is in contrast to the chromatin changes induced at highly transcribed genes in which well-phased nucleosome positions are barely discernible (Supplemental Fig. S2A). In some cells, the helicase may have unwound less duplex DNA, leaving some nucleosomes well positioned. In others, the helicase may not have been activated by preinitiation complex factors, given the stochastic nature of the replication program. It is also energetically favorable that unwound, complementary single strands would eventually reanneal, allowing nucleosomes to be reset behind the helicase (Fig. 6D). All of these possible states of unwinding and resetting will be captured by our chromatin profiling assay.

Chromatin disorganization, Mcm2-7 enrichment, and RPA enrichment at replication origins all support that the active CMG can uncouple from DNA replication and unwind duplex DNA up to 1 kb from origins. Our prior work demonstrated that Mcm2-7 is loaded in the nucleosome-free region adjacent to either the upstream or downstream flanking nucleosome during G1 phase (Belsky et al. 2015). Interestingly, this asymmetric loading of the helicase near either the upstream or downstream nucleosome in G1 does not impact initial unwinding, which is symmetric with respect to the origin center. This suggests that the origin recognition complex does not pose a significant obstacle to the CMG, which can disrupt up to six nucleosomes on either side of the nucleosome-free region at the origin. Previous estimates of CMG movement in the absence of replication in vivo in budding yeast range widely from 100 bp up to 3 kb, though 400–500 bp is the most common maximum estimate (Walter and Newport 2000; Sogo et al. 2002; Katou et al. 2003; Nedelcheva et al. 2005; Zellweger et al. 2015; Gan et al. 2017). This range of distances may be due to the resolutions of the methods used as well as different initial causes for helicase–polymerase uncoupling, which include dNTP depletion in checkpoint-deficient cells, multiple types of DNA lesions, and polymerase inhibition with aphidicolin. Our results demonstrate in high resolution at multiple origins that the CMG complex can unwind up to 1 kb of DNA when uncoupled from replication in the absence of damaging agents and with an active checkpoint.

Mechanistic experiments in vitro support the model that, after activation, CMG complexes establish replication forks traveling with the Mcm2-7 N-terminal domain first (Georgescu et al. 2017; Douglas et al. 2018). However, the Mcm2-7 double hexamer forms in G1 phase through the association of each Mcm2-7 complexes’ N-terminal domain, and thus this directionality requires the two CMG complexes to translocate past each other on ssDNA prior to initiation of replication. A consequence of this head-to-head model is more extensive DNA unwinding prior to replication initiation than if Mcm2-7 traveled with the opposite orientation at forks. Thus, limited helicase uncoupling at the origin may be a necessary feature for replication initiation.

Helicase–polymerase uncoupling in the absence of replication leaves behind unreplicated regions at origins when replication finally begins. If the stalled CMG complexes are disassembled, these origins may be passively replicated by CMGs initiating from other start sites. However, given that the location at which the Mcm2-7 complex stalls coincides with the return to normal copy number following recovery, we hypothesize that in cdc17-ts-FRB cells replication restarts outside the origin region (model in Fig. 6D). Cells with limited levels of polymerase α experience increased genomic instability (Song et al. 2014), potentially due to ssDNA tracts generated by uncoupling of the leading and lagging strands. Furthermore, mutations in the human homolog of CDC17/POL1, POLA1, are associated with X-linked developmental disorders, potentially caused by impaired polymerase α function and alterations to replication dynamics leading to reduced cell proliferation (Van Esch et al. 2019). The unreplicated gaps we observed suggest that helicase–polymerase uncoupling at replication origins may contribute to heightened genomic instability and altered replication dynamics, making the repair of these gaps a target for future study. Prior to recovery, RPA bound to ssDNA unwound by the stalled CMG might be replaced by Rad51, mediated by Rad52, which could participate in ssDNA reannealing (Sung 1997; Shinohara et al. 1998). When polymerase α function is restored, the CMG would resume unwinding DNA from where it stalled, and replication of the majority of the genome would begin. The resulting structure, two replicated double helices connected by an unreplicated section of reannealed parental strands at the origin, does not resemble any structure known to occur during homologous recombination or DNA repair pathways. There must exist a mechanism to repair this structure, given that recovering cells pause in G2/M phase and eventually re-enter the cell cycle. These gaps must first be recognized by a repair or checkpoint factor, and we expect that a helicase will be recruited to unwind the unreplicated strands. There are several potential helicases that could serve this role, including Sgs1 and Srs2 (Crickard and Greene 2019). Once unwound, polymerase α could establish primers extended by Polδ, and the nascent strands could ultimately be ligated together using mechanisms of Okazaki fragment maturation (Bell and Labib 2016).

Together, our flow cytometry and copy number experiments support that cdc17-ts-FRB cells do not initiate replication under restrictive conditions, but we still detected activation of the intra-S-phase checkpoint through Rad53 phosphorylation. The lack of Rad53 phosphorylation in cdc17-ts-FRB MRC1-FRB suggests that intra-S-phase checkpoint activation occurs through the DNA replication arm of the checkpoint in this context, as opposed to the DNA repair arm mediated through Rad9 (Pardo et al. 2017). Canonically, the presence of ssDNA recruits the sensor kinase Mec1 (ATR in mammals), but it is not sufficient for Mec1 activation (Zou and Elledge 2003). The 9-1-1 complex, which recognizes ssDNA-dsDNA junctions, is one of the additional factors that contributes to activation of Mec1 (Majka and Burgers 2003; Majka et al. 2006). At a minimum, a single primer is thought to be required for checkpoint activation, with additional primers amplifying activation (Byun et al. 2005; MacDougall et al. 2007; Van et al. 2010). However, if priming is absent in our system, there may be more flexibility in requirements for checkpoint activation than previously thought. Given that studies directly addressing the requirement of priming in checkpoint activation were done in Xenopus, this flexibility in checkpoint activation may be specific to budding yeast. In S. cerevisiae, additional factors such as Polε play a role in the checkpoint, and the 9-1-1 complex may not be fully necessary for checkpoint function (Crabbé et al. 2010; García-Rodríguez et al. 2015).

Helicase–polymerase uncoupling has been described in various checkpoint mutants when exposed to DNA damage or hydroxyurea, and the effector kinase of the intra-S-phase checkpoint, Rad53, has been proposed as a key regulator of CMG progression in the absence of replication (Gan et al. 2017; Devbhandari and Remus 2020). In our experiments, Rad53 is phosphorylated in the absence of polymerase α function in cdc17-ts-FRB cells, which confirms studies in other eukaryotes showing that the intra-S-phase checkpoint is activated in the presence of aphidicolin or polymerase α inhibition (Byun et al. 2005; Ercilla et al. 2020). We found that the intra-S-phase checkpoint does not ultimately regulate movement of the CMG complex when polymerase α function is perturbed. However, the context in which helicase–polymerase uncoupling occurs may modulate the role of Rad53 in regulation. For example, eSPAN experiments in rad53-1 cells demonstrate that leading strand uncoupling in hydroxyurea can be almost entirely rescued by deletion of SML1, which inhibits ribonucleotide reductase RNR (Gan et al. 2017). Additionally, the same study found that Polε is more sensitive than Polδ to dNTP levels. These results suggest that the role of Rad53 in regulation of dNTP pools is specific to uncoupling in cells exposed to hydroxyurea. In vitro reconstitution experiments have shown that an activated Rad53 can reduce the amount of DNA unwound by the CMG in the presence of aphidicolin, both at stalled forks and in the absence of replication (Devbhandari and Remus 2020). Given these results, Rad53 may be recruited to origins by the activation of the checkpoint in our experiments as well; however, Rad53 phosphorylation is not required for the helicase uncoupling we observed at DNA replication origins. Together, these results suggest that there may be other extrinsic mechanisms that can stall the helicase.

Budding yeast origins, like some prokaryotic origins, have been shown to contain a broad AT-rich DNA unwinding element in addition to the shorter T-rich motif (Umek and Kowalski 1988; Kowalski and Eddy 1989). This DNA unwinding element is energetically easier to unwind into single strands and likely serves to facilitate replication initiation (Natale et al. 1992; Ak and Benham 2005). As sequence content returns to the genome-wide average outside the origin region and becomes more difficult to denature, Mcm2-7 enrichment increases in cdc17-ts-FRB under restrictive conditions, indicating that the CMG is less processive and frequently stalls at sequences with elevated GC content. Although this pattern does not necessarily imply causation, previous studies support the model that replicative helicases are less processive following uncoupling with replication. Studies of the E. coli helicase and polymerases have established that, when there are stochastic pauses in replication or if the helicase and polymerase are physically separated, the rate of helicase progression slows to prevent excessive uncoupling (Kim et al. 1996; Yeeles and Marians 2013; Graham et al. 2017). This phenomenon has been named a biological “dead man's switch” (Graham et al. 2017). It has recently been proposed that a similar mechanism exists in the eukaryotic replisome (Taylor and Yeeles 2019). Similarly, single molecule studies of CMG unwinding in the absence of replication have produced a model of the CMG as a “lazy Brownian ratchet” that does not continuously unwind DNA, instead sometimes entering paused states (Burnham et al. 2019); this model fits well with a CMG that stalls at GC sequences. When Mcm2-7 encounters a DNA lesion on the leading strand, the replisome slows down or stalls regardless of the type of lesion or composition of the replisome, suggesting that there is an intrinsic mechanism that stalls the helicase (Sparks et al. 2019; Taylor and Yeeles 2019). Given that the CMG can bypass leading strand DNA lesions that block Polε, this slowing mechanism is likely a response to helicase–polymerase uncoupling (Sparks et al. 2019). We have shown that, without priming, the CMG stalls at more stable sequences, supporting the “dead man's switch” model. Furthermore, Polε remains with the CMG, and thus strand polymerization, and not the physical presence of a polymerase, must modulate helicase activity. In higher eukaryotes, an AT-rich unwinding element has not been found at origins, so we predict that Mcm2-7 will stall closer to origins in the absence of replication initiation in these systems. Our results suggest that in addition to impacting replisome progression, the mechanisms that regulate helicase–polymerase uncoupling contribute to replication initiation. Failure to regulate helicase uncoupling at the origin via a “dead man's switch” may trigger replication catastrophe resulting from the generation of excess ssDNA and the sequestration of RPA (Toledo et al. 2013; Ercilla et al. 2020); thus, given the conservation of the helicase complex (Liu et al. 2009), the intrinsic mechanisms regulating helicase uncoupling from the replisome are likely conserved across eukaryotes and critical for preserving genome stability.

Materials and methods

Yeast genetic methods

All yeast strains are in the W303 background and derived from HHY168 (Haruki et al. 2008). Genotypes of all strains used are shown in Table 1. HHY168 and plasmids containing the FRB tag were obtained from Euroscarf. The LiAc/ssDNA/PEG method was used for all transformations, with the following adjustments (Gietz and Woods 2002). For all transformations involving temperature-sensitive strains or mutations, cells were grown at 24°C prior to transformation and an overnight incubation at 24°C was used in place of the typical 42°C heat shock. To amplify the FRB construct, primers for CDC17 and MRC1 were designed to match the 3′ end of each gene and the plasmids pFA6a-FRB-KanMX6 or pFA6a-FRB-HisMX6 (Longtine et al. 1998). These PCR products were then transformed into yeast. Temperature-sensitive mutations were added to CDC17 using CRISPR-Cas9 gene editing (Anand et al. 2017). First, guide RNA sequences were added to the Cas9-expressing plasmid bRA89 (a gift from J. Haber at Brandeis University). To construct the repair template, a fragment of CDC17 was cloned into pCR 2.1 TOPO (Thermo Fisher 450641). Temperature-sensitive mutations were added to this fragment of CDC17 using site-directed mutagenesis (Agilent 200521). Both bRA89 with gRNA and the repair template (excised from the TOPO vector) were transformed into CDC17-FRB. RFA1 and POL2 were tagged with 3xFLAG using primers to the 3′ end of each gene and to pFA6a-6xGLY-3xFLAG-HIS3MX6 (Funakoshi and Hochstrasser 2009).

Table 1.

Yeast strains used in this study

Cell synchronization and early S-phase arrest

All experiments were done in YPD media at 24°C unless indicated otherwise. For each experiment, cells were grown to saturation overnight, and this culture was used to inoculate a larger culture the next morning. Cells were doubled (based on OD600 readings) and then arrested for 2 h using α-factor (Genway) at a final concentration of 50 ng/mL. In the final 30 min of the α-factor arrest, cultures were split in two and moved into different conditions: 24°C and 37°C alone or with 1 mg/mL rapamycin (RPI). At the end of the α-factor arrest, cells were washed twice with sterile water, resuspended in YPD media, and returned to their previous conditions (24°C or 37°C ± rapamycin) for at least 1 h. To block cells in G2/M, nocodazole (Sigma) was added to cells following release from α-factor arrest at a concentration of 15 μg/mL for 3 h.

Spot assay

Cells were grown to saturation overnight in liquid culture and then all cultures were adjusted to the same OD600. Cells were serially diluted in 10-fold increments, mixed thoroughly, and spotted onto either YPD or YPD + 1 mg/mL rapamycin plates. Plates were incubated 2 d at either 24°C or 37°C and then imaged.

Flow cytometry

Experiments to analyze DNA content were performed as previously described (Gutiérrez et al. 2019).

Chromatin preparation

Twenty-five-milliliter samples were collected from cultures grown to an OD600 of 0.8–1, cross-linked, and quenched as previously described (Belsky et al. 2015). Cells were pelleted, flash-frozen, and stored at −80°C. MNase digestions were performed on thawed pellets again according to Belsky et al. (2015). One microgram of MNase DNA was used in the Illumina library preparations.

Chromatin immunoprecipitation

ChIP experiments were performed as in Belsky et al. (2015), with the following adjustments. Fifty-milliliter samples were collected from cultures grown to an OD600 of 0.8–1. For Mcm2-7 immunoprecipitation, 25 µL of AS1.1 antibody was used (Chen et al. 2007). For Rfa1-3xFLAG and Pol2-3xFLAG, 1–2 µL of mouse anti-FLAG (Sigma F3165) antibody was used. The entire ChIP pellet was used in the Illumina library preparations.

Western blotting

One-milliliter samples were collected from cultures grown to an OD600 of 0.8–1, and prepared for analysis using TCA precipitation. Samples were run on a 10% SDS-PAGE gel and transferred to a nitrocellulose membrane with a wet transfer system. Membranes were blocked with 5% milk/1% Tween/TBS for 1 h, then incubated overnight with either rabbit polyclonal anti-Rad53 (Abcam ab104232) or rabbit polyclonal anti-GAPDH/Tdh1 (Origene AP21309AF-N; a gift from D. Lew, Duke University) in the blocking solution. Membranes were washed and incubated in goat anti-rabbit IgG HRP (Thermo Fisher 31460) in blocking solution for 40 min at room temperature. Membranes were washed, incubated with ECL solution (Pierce 34095), and imaged using film.

Genomic DNA

Cells were pelleted, flash-frozen, and stored at −80°C. Genomic DNA was isolated from thawed cells as in Hoffman (2001) with the following modifications: a wash in 70% ethanol following each ethanol precipitation and a 1-h incubation at −20°C during the final precipitation. Four micrograms to 5 µg of DNA was used in the Illumina library preparations.

Sequencing library preparation

MNase sequencing libraries were prepared as previously described (Henikoff et al. 2011; Belsky et al. 2015), and ChIP and genomic DNA libraries were prepared according to Illumina TruSeq protocols. The only adjustment made to both protocols was the use of the NEBNext multiplex oligos for Illumina kit (E7335S) in the adaptor ligation and PCR steps and the use of AMPure XP beads (Beckman) for purification of some libraries.

Data analysis

Sequencing data were aligned to sacCer3 using BWA version 0.7.12 (Li and Durbin 2009). Data analysis was performed in R version 4.0.5 (https://www.r-project.org). Scripts are available at https://gitlab.oit.duke.edu/rah60/hoffman-2021-paper. See the Supplemental Material for details on data analysis.

Data access

All sequencing data from this study are available in the NCBI Sequence Read Archive (SRA) under accession number PRJNA704995.