Budding yeast complete DNA replication after chromosome segregation begins

To faithfully transmit genetic information, cells must replicate their entire genome before division. This is thought to be ensured by the temporal separation of replication and chromosome segregation. Here we show that in a substantial fraction of unperturbed yeast cells, DNA replication finishes during anaphase, late in mitosis. High cyclin-Cdk activity inhibits replication in metaphase, and the decrease in cyclin-Cdk activity during mitotic exit allows DNA replication to finish at difficult-to-replicate regions. Replication during late mitosis correlates with elevated mutation rates, including copy number variation. Thus, yeast cells temporally overlap replication and chromosome segregation during normal growth, possibly allowing cells to maximize population-level growth rate while simultaneously exploring greater genetic space. One Sentence Summary Completion of DNA replication is coupled to downregulation of Cyclin-Dependent Kinase during mitotic exit.


Abstract:
To faithfully transmit genetic information, cells must replicate their entire genome before division. This is thought to be ensured by the temporal separation of replication and chromosome segregation. Here we show that in a substantial fraction of unperturbed yeast cells, DNA replication finishes during anaphase, late in mitosis. High cyclin-Cdk activity inhibits replication in metaphase, and the decrease in cyclin-Cdk activity during mitotic exit allows DNA replication to finish at difficult-to-replicate regions. Replication during late mitosis correlates with elevated mutation rates, including copy number variation. Thus, yeast cells temporally overlap replication and chromosome segregation during normal growth, possibly allowing cells to maximize population-level growth rate while simultaneously exploring greater genetic space.

Main text:
Eukaryotic cells must complete DNA replication before chromosome segregation in order to maintain genomic stability. Complete replication is thought to be ensured by the temporal separation of DNA synthesis (S-phase) from mitosis (M-phase) (1).
The ordering of S and M phases is established by increasing levels of cyclindependent kinase (Cdk) activity during the cell cycle (2) and is enforced by checkpoints that inhibit chromosome segregation when cells are exposed to severe replication stress (3). However, some yeast mutants, and cancer cells exposed to mild DNA replication stress, perform DNA synthesis in mitosis and possibly even in the subsequent G1 (4,5), suggesting that DNA synthesis and mitosis may not be fully incompatible. Supporting this view, several lines of evidence suggest that budding yeast lack a checkpoint to detect if DNA replication has completed before entry into mitosis (6)(7)(8)(9)(10). To directly test if DNA synthesis occurs during mitosis in unstressed cells we arrested yeast in metaphase via depletion of the anaphase promoting complex activator Cdc20 and measured incorporation of the nucleotide analogue 5-ethynyl-2'-deoxyuridine (EdU) as cells were, or were not, released into a G1 arrest. Cells held in metaphase showed no nuclear EdU signal after a 60-minute pulse, whereas cells released from metaphase into G1 arrest incorporated EdU into the nucleus (Fig.1A,   S1A). EdU incorporation was higher in G1 than in metaphase cells in both DAPI-rich and DAPI-poor nuclear regions, which contained the nucleolar marker Net1 (Fig.   S1B). Freely-cycling unstressed cells showed significant nuclear EdU incorporation in late mitosis and G1, whereas mitotic cells with actively segregating nuclei did not ( Fig. S2). Further, 45% of log-phase cells entered anaphase with single-stranded DNA, detected as Replication Protein A (RPA) foci (Rfa2-GFP) (Fig. S3). These observations suggest that metaphase is refractive to nuclear DNA synthesis, but that some DNA synthesis occurs between metaphase and the following G1 in freely cycling, unstressed cells.
Next, we evaluated whether DNA synthesis during mitosis is important for chromosome segregation and cell division. We arrested cells in metaphase by depletion of Cdc20 or by treatment with nocodazole, an inhibitor of microtubule polymerization. Upon release we inhibited DNA synthesis by treatment with the ribonucleotide reductase inhibitor hydroxyurea (HU) or by inactivating DNA replication factors with temperature-sensitive (ts) mutations, including DNA polymerase delta and epsilon, and the GINS complex component PSF2 (Fig. 1B, C).
In all of these cases, disruption of DNA synthesis during mitosis delayed or inhibited nuclear and cell division (visualized with the Histone H2B (Htb2)-mCherry reporter and the membrane marker GFP-CAAX, respectively) and triggered long-lived chromatin bridges ( Fig. 1B-D, S4). Disruption of mitotic DNA synthesis also delayed cytokinesis in cells without tagged histones, although with reduced severity in response to HU (Fig. 1C, S4E). Ablation of the RAD9 checkpoint, which prevents cell cycle progression in response to DNA damage during both S phase and anaphase (11), abolished nuclear division delays in response to challenges in DNA synthesis during mitosis (Fig. S4A-B). These results suggest that mitotic DNA synthesis promotes timely chromosome segregation.
Surprisingly, cells arrested in metaphase for prolonged periods of time did not undergo DNA synthesis during the arrest (Fig. 1A). We hypothesized that high Cdk activity inhibits DNA synthesis in metaphase, and that the inhibition of Cdk during mitotic exit enables synthesis to complete. To test this, we examined cells arrested in late anaphase with separated nuclei and high levels of mitotic cyclins by inactivation of the Mitotic Exit Network (MEN) (ts mutants in TEM1, CDC15, or DBF2) (12). MEN mutants displayed a high frequency of chromatin bridges visualized with Htb2-mCherry or the DNA dye YOYO-1, indicative of incomplete chromosome segregation in at least 40% of late anaphase cells ( Fig. 2A, S5).
Moreover, time-lapse imaging of fluorescent loci in chromosome XII showed defects in the segregation of telomere-proximal regions in MEN mutant cells (Fig. S6).
Chromatin bridges were stable in MEN-arrested cells (cdc15-as1 treated with the ATP analogue 1-NA-PP1) even after inactivation of the cohesin Scc1p (Fig. S7). MEN reactivation (by washout of 1-NA-PP1) led to the resolution of chromatin bridges before cytokinesis even in the absence of Topoisomerase II activity (Fig. 2B). Thus, MEN bridges are not due to persistent cohesion or catenation between replicated sister DNA molecules. However, timely bridge resolution after MEN reactivation required DNA polymerase delta, indicating that MEN-deficient bridges require DNA synthesis for their resolution (Fig. 2B). Consistent with this, RPA foci indicative of ssDNA are present in most MEN-deficient cells during anaphase (Fig. 2C).
DNA synthesis during late mitosis may reflect mitotic repair of already replicated DNA, mitotic replication of diverse genomic regions, or mitotic replication of specific genomic regions. To distinguish between these possibilities, we used Illumina sequencing to measure DNA copy number (13) in cells arrested in (i) G1, (ii) metaphase, via depletion of Cdc20, or (iii) late anaphase/telophase, via inactivation of MEN (dbf2-2). To obtain >95% synchrony we isolated mitotically arrested cells by sucrose gradient centrifugation and used the fraction with the highest synchrony for DNA extraction (Fig. S8, Supplementary Table S1 and Methods). From the copy number ratios between mitotic and G1-arrested cells we calculated the percentage of cells in which each region of the genome is under-represented during mitosis (see Methods). We excluded long repeats (rDNA and telomeres) from our analysis, as in these cases DNA copy number cannot be distinguished from repeat copy number. Nonetheless, we found that chromosome ends were underrepresented in a high percentage of mitotic cells: on average, 60 kb at the end of each chromosome are under-represented in metaphase (Fig. 3A, S9, 10A). Over 40% of cells have lower metaphase copy number in the 1 kb closest to each telomere  Table S2). Regions under-represented in mitosis corresponded to a subset of late-replicating regions (Fig. 3D, S11). We therefore refer to these regions as "under-replicated in mitosis". Difficult-to-replicate regions such as those with G-quadruplexes, transposable elements and fragile sites (14) showed significant under-replication in metaphase (Fig. 3E). Under-replication was higher in metaphase (CDC20 depletion) than in late anaphase (dbf2 inactivation) ( Fig. 3A). Moreover, under-replication for transposable elements and fragile sites becomes negligible in late anaphase (Fig. S12), suggesting that these regions complete replication in anaphase.
To directly test if high mitotic Cdk activity inhibits DNA synthesis in difficult-toreplicate genomic regions we determined DNA copy number in Cdc20-depleted cells after inactivation of Cdk using the cdc28-as allele. Inactivation of Cdk enabled Cdc20depleted cells to finish replication of chromosome ends (Fig. 3F). Furthermore, Cdk inactivation (cdc28-4) also enabled chromatin bridge resolution independently of cytokinesis in MEN-deficient cells (Fig. 3G). Consistent with this, over-expression of the mitotic cyclin CLB2 in S phase slowed completion of DNA replication (Fig. S13).
Therefore, Cdk inactivation is required to complete replication of specific regions during mitotic exit, preventing the formation of stable chromatin bridges.
There is some correlation between replication timing and mutation rates in yeast and human cells (18,19) and a strong relationship between distance to the telomere and mutation rate in budding yeast (20). Deletion of genes that are replicated in late mitosis has no apparent fitness cost, consistent with these genes being dispensable under non-challenging conditions (Fig. 4A). Interestingly, genes in regions replicated during mitosis exhibited higher levels of intraspecies genetic diversity at the single-nucleotide level, for small insertions and deletions (<50 nt), and at the level of gene gain and loss (Fig. 4B). Further, gene ontology analysis shows that genes under-replicated in mitosis are mostly subtelomeric, and are enriched in carbon signaling and transport functions ( Fig. 4C and Supplementary Table S3).
Thus, replication during late mitosis may result in increased mutation rates of specific classes of genes.
Together, these results suggest that a substantial fraction of wild-type unstressed cells finish replication of specific chromosome regions late in mitosis, long after the initiation of chromosome segregation (Fig. 4D). We speculate that replication forks likely stall at difficult-to-replicate sequences, such as G-quadruplexes, and that their replication is inhibited by high Cdk levels in metaphase. Although high Cdk levels are compatible with bulk DNA replication in fission yeast (2), chromatin replication is inhibited in Xenopus mitotic extracts (15). Perhaps the replication of specific genomic regions must be paused during metaphase to prevent damage and chromosome pulverization (16,17). Our data indicate that regions under-replicated in metaphase have one last opportunity to complete their replication before cytokinesis when Cdk levels fall below a critical threshold during mitotic exit.
Why do yeast not complete replication prior to mitosis? Starting mitosis before completion of DNA replication may allow for an increased cell division rate which could outweigh occasional mutations, aneuploidy and gene loss. For instance, a decrease in the number of viable cells can be easily compensated for by an accelerated division rate (Fig. S14). In addition, our data raise the possibility that regions that replicate during mitosis, devoid of essential genes, serve as a genetic playground in which high rates of mutation and copy-number variation (21) allow cells to more rapidly explore genotypic space. Further, continued replication during late mitosis may explain the high frequency in animal cells of ultrafine chromatin bridges, whose defective resolution is associated with genome instability (22). raffinose) medium to mid-log phase, and galactose was added to 0.5%. To determine DNA content with Sytox-Green, we used the staining protocol from (26).

EdU incorporation in cells arrested in metaphase. EdU click and DAPI staining
were carried out as in (27)  EdU labeling with Alexa Fluor 647 (for microscopy) as described (27). Cells were imaged in a Leica TCS SP5 confocal microscope (63x objective, 3 planes spaced 0.67 µm). Nuclear EdU incorporation was measured in maximum projections as the background-subtracted mean intensity in the nucleus (defined by DAPI staining).
The background was measured in multiple small cytoplasmic areas to avoid signal from mitochondrial DNA. As not all cells are permeabilized during fixation, we discarded those cells that showed no EdU signal in mitochondria.
EdU incorporation in freely-cycling cells was carried out in the same manner as above, in cells with wild-type CDC20 (w303, RAD5 bar1 GPDpr-TK(5x) ADH1pr-hENT1), except that cells were pregrown in SCD to mid-log phase and α-Factor (20 μg/ml) was added for 30 minutes to prevent cells already in G1 from entering into S phase. Cells were then incubated in EdU (25 μM

MET3pr-CDC20
We generated 200-bp non-overlapping windows across the genome (bedtools makewindows) and computed coverage for each window (bedtools coverage). have not yet replicated a given region of the genome (under-replication), is defined as 2*(1-CNR).
The above was performed for six biological replicates for each strain and condition (see Supplementary Table S1)    Regions with high frequency of G-quadruplexes, transposable elements and fragile sites exhibit higher under-replication in metaphase (* p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001, Student T-test). (F) Genome sequencing was performed in metaphase-arrested (Cdc20-depleted) cells before and after inhibition of Cdk using Budding yeast complete DNA replication after chromosome segregation begins. The order of cell cycle phases is represented by grey (G1), blue (S1, S2) and red (M) bars.
S phase is divided into pre-mitotic S1, in which the bulk of DNA replication occurs, and S2, which overlaps with chromosome segregation and in which subtelomere, transposons and fragile sites are replicated in a fraction of cells. DNA replication is inhibited by high M-Cdk levels during metaphase.