Report

Anaphase Onset Before Complete DNA Replication with Intact Checkpoint Responses

See allHide authors and affiliations

Science  09 Mar 2007:
Vol. 315, Issue 5817, pp. 1411-1415
DOI: 10.1126/science.1134025

Abstract

Cellular checkpoints prevent mitosis in the presence of stalled replication forks. Whether checkpoints also ensure the completion of DNA replication before mitosis is unknown. Here, we show that in yeast smc5-smc6 mutants, which are related to cohesin and condensin, replication is delayed, most significantly at natural replication-impeding loci like the ribosomal DNA gene cluster. In the absence of Smc5-Smc6, chromosome nondisjunction occurs as a consequence of mitotic entry with unfinished replication despite intact checkpoint responses. Eliminating processes that obstruct replication fork progression restores the temporal uncoupling between replication and segregation in smc5-smc6 mutants. We propose that the completion of replication is not under the surveillance of known checkpoints.

Eukaryotes have acquired cellular mechanisms that prevent or delay progression through the cell cycle when DNA is damaged (1). These mechanisms are referred to as checkpoints. Completion of DNA replication before mitotic entry is thought to be subjected to regulation by a checkpoint (1), because premature entry into mitosis would be detrimental to the integrity of the genome. Such a replication-completion checkpoint should prevent mitosis by sensing the persistence of unreplicated DNA or ongoing fork progression in an otherwise normal Sphase.

The arguments supporting the existence of a replication-completion checkpoint derive from observations demonstrating that budding yeast cells activate a reversible checkpoint when cells are treated with the drug hydroxyurea (HU) (2). However, the checkpoint response to HU is caused by the accumulation of single-stranded DNA on replication forks rather than unreplicated DNA (35). Indeed, indirect evidence from several studies suggests that yeast cells might lack a replication-completion checkpoint (68).

The Smc5-Smc6 complex is related to cohesin and condensin and functions in DNA repair (9). Cells expressing the smc5-smc6 mutant alleles showed S phase– and anaphase-entry times similar to those of wild-type (WT) cells (Fig. 1A and fig. S1A), and the central check-point kinase Rad53 was activated only after the first mitosis under nonpermissive conditions (10). The ribosomal DNA (rDNA) array in the middle of chromosome XII is a major binding site for the Smc5-Smc6 complex (10). Segregation analysis of chromosome XII with the use of fluorescence detection of DNA-based tags, which were inserted at different positions along the chromosome, revealed that tags located between the centromere and the rDNA segregated equally to daughter nuclei in WT and mutant cells, whereas tags between the rDNA and the telomere missegregated to one pole in smc5-smc6 mutants (fig. S1B) (10). Thus, unresolved linkages between sister chromatids seem to cause chromosome nondisjunction in these mutants. Deletion of the recombination gene RAD52 partially suppresses the smc6-9 growth defect (10). We observed a modest reduction in nondisjunction of chromosome XII when we deleted different recombination genes (fig. S2, A and B). Therefore, the majority of nondisjunction events in smc6-9 cells are not recombination structures.

Fig. 1.

Passage through S phase, but not through anaphase, in the absence of Smc5-Smc6 produces chromosome nondisjunction. (A) Cell cycle dynamics of spindle elongation in WT and smc6-9 cells, showing that smc6-9 mutants are not delayed in cell cycle progression. The image depicts a representative micrograph of spindles during the time course. (B) Segregation of a chromosome tag (tetO:487) inserted in the telomere flank of rDNA under the indicated conditions in WT, smc6-9, and nse5-1 cells. (C) Representative micrographs of cells in (B). Red, nuclei; green, tetO:487.

Segregation analysis with the use of different growth regimes (fig. S3) showed that Smc5-Smc6 function is required during S phase (Fig. 1, B and C). To explore possible defects in replication forks, we compared actively replicating rDNA from WT and smc6-9 mutant cells by two-dimensional (2D) gel electrophoresis (see supporting online material). We detected an increase in replication fork barrier (RFB) arrest, recombination, and termination structures in smc6-9 mutants during S phase (Fig. 2, A and B). The accumulation of structures in smc6-9 was not caused by an increase in origin firing because WT levels of bubble intermediates were observed in the mutant (fig. S4). Unexpectedly, we found that, unlike WT cells, smc6-9 mutants exhibited an increase in Y-arc replication structures when cells were arrested in metaphase (Fig. 2C) (10). Thus, smc6-9 cells are still replicating rDNA during metaphase, demonstrating that there is a delay in the replication of this region. To further confirm this possibility, we monitored the replication of individual rDNA-containing chromosomes by their extension on silanized glass surfaces (11). Cells were induced to incorporate bromodeoxyuridine (BrdU) in their DNA during the previous replication. In addition, we blocked cells in metaphase to ensure that they did not enter another cell cycle. We digested chromosomes with restriction enzymes that cut throughout the genome but not within the 2-megabase rDNA region; thus, long DNA fibers that were extended on the silanized glass surfaces represent ribosomal gene arrays (12). Replicated regions incorporated BrdU and were identified after immunodetection, whereas BrdU gaps in the fibers revealed unreplicated regions. We detected a twofold increase in unreplicated gaps in smc6-9 cells relative to WT cells (Fig. 2D). Moreover, the average length of the rDNA fibers in smc6-9 cells was significantly shorter than that in WT cells (Fig. 2D), suggesting that smc6-9 fibers break during the protocol. Thus, this procedure underestimates the amount of unreplicated rDNA gaps in these mutant cells. In addition, gaps smaller than a few kilobases are not detectable by this technique (12). We conclude that replication of the rDNA region is delayed, and not completed by metaphase, in smc6-9 cells.

Fig. 2.

Replication of rDNA regions is not finished in metaphase in smc6-9 mutants. (A) Replication intermediates of a rDNA region in WT and smc6-9 cells digested with Bgl II and separated by 2D gel electrophoresis. Schematic representations of yeast rDNA and restriction enzyme sites are shown. ARS, autonomously replicating sequences. Numbers in parentheses represent different structures on the 2D gels quantified. (B) Quantification of rDNA replication structures in (A). (C) WT and smc6-9 cultures arrested in metaphase and analyzed by 2D gel electrophoresis as in (A). Quantification of corresponding rDNA structures is shown. (D) DNA combing analysis of completion of DNA replication at the ribosomal genes array. Two representative rDNA fibers, either completely replicated or presenting unreplicated gaps, are shown. Gap lengths are indicated in kb. Red, propidium iodide stain of total DNA; green, antibodies to BrdU. Error bars in (B) and (C) indicate SD from three independent experiments.

The fact that smc6-9 mutants are not delayed in mitotic entry (fig. S1A) suggests that they might execute anaphase before they finish replication, at least for chromosome XII. We evaluated whether replication is completed in smc6-9 mutants before segregation using pulsed-field gel electrophoresis (PFGE). Incompletely replicated chromosomes do not resolve by PFGE and remain in the wells. In WT cells, unreplicated (Fig. 3A; time: 0 min) and fully replicated (Fig. 3A; time: >80 min) chromosomes entered the gel; however, all chromosomes failed to enter the gel while replicating (Fig. 3A; time: 20 to 60 min). In smc6-9 cells, the amount of chromosome XII that failed to enter the gel after replication (and even segregation) was much greater than that in WT cells (Fig. 3A), and the deletion of RAD52 had little effect (Fig. 3A).

Fig. 3.

smc6-9 cells enter anaphase before completing replication of chromosome XII, despite the presence of intact checkpoint responses. (A) Synchronized WT and smc6-9 cultures analyzed by PFGE to visualize yeast chromosomes and their replication. PFGs were transferred and hybridized with a probe to various chromosomes to quantify entry of individual chromosomes. Quantification of chromosome XII entry into the gel is shown for WT, smc6-9, and smc6-9 rad52D cells. The single asterisk indicates the axis label: time (in min) after G1 release at 37°C. In (B), Rad53p is similarly phosphorylated in WT and smc6-9 cells after HU treatment. A DSB at MAT (C) or rDNA (D) induces Rad53 phosphorylation in WT and smc6-9 cells. Arrows in (B) to (D) indicate Rad53p species. Phosphorylated Rad53p is indicated (**).

To analyze whether smc6-9 cells are deficient in the activation of known checkpoint pathways, we tested whether smc6-9 mutants activate Rad53 in response to replication stress. Similar levels of Rad53 phosphorylation were detected in WT and smc6-9 cells exposed to the ribonucleotide reductase inhibitor HU (Fig. 3B) (13). Both WT and smc6-9 cultures arrested as budded cells with one nucleus in response to HU, indicating that both strains halt the cell cycle when the S phase checkpoint is activated. We obtained similar results when cells were exposed to methyl methanesulfonate. In addition, smc6-9 mutants are also competent in the activation of the DNA-damage checkpoint in response to a single double-strand break (DSB) at the MAT locus induced by expression of the HO endonuclease (HO) (Fig. 3C). After 4 hours of HO induction, both asynchronous WT and smc6-9 cells arrested in G2-M with a single focus of the checkpoint protein Ddc2 (fig. S5A), and Rad53 activated (Fig. 3C). Check-point activation also occurs when DSBs occur inside the nucleolus. In a strain carrying the I–Sce-I recognition site in the middle of the ribosomal gene array and expressing the I–Sce-I endonuclease under the galactose promoter, we detected Rad53 activation in WT and smc6-9 cells (Fig. 3D). We conclude that breaks in the rDNA are also under normal checkpoint surveillance. Finally, smc6-9 mutants are competent in the spindle check-point, because their sensitivity to the spindle-depolymerizing drug benomyl was comparable to that of WT cells (fig. S5B). We conclude that mitotic entry before the completion of rDNA replication in smc6-9 mutants is not caused by failures in the cellular checkpoint machineries.

Our data indicate that replication of the rDNA region is severely delayed in the absence of Smc5-Smc6 function (Figs. 2 and 3A). We tested whether eliminating processes that obstruct replication forks in the rDNA would reduce the delay and suppress the chromosome XII nondisjunction phenotype. The rDNA is composed of tandem repeats that are unidirectionally replicated because of the presence of a polar RFB, mediated by the Fob1 protein, which is next to the 35S transcription termination site (14, 15). In WT cells, replication of rDNA has to deal with the presence of RFBs and the clustering of active origins (12), which generates a situation where large fragments are replicated by a single rightward-moving fork. In the absence of Fob1, leftward-moving forks are not blocked at the RFB site; thus, replication is accelerated by the presence of two active forks (instead of one). Deletion of FOB1 in smc6-9 cells reduced chromosome XII nondisjunction from 63 to 34% (Fig. 4A and fig. S6A). In addition to RFBs, the high transcription rates in rDNA genes are also likely to pose a challenge for active replication across the locus. Inactivation of ribosomal RNA (rRNA) gene transcription can be achieved by the simultaneous deletion of the yeast RNA polymerase (Pol) I subunit A135 and the introduction of a multicopy plasmid containing the 35S rRNA coding region fused to an RNA Pol II promoter (16). The inactivation of chromosomal rRNA gene transcription reduced chromosome XII nondisjunction from 63 to 19% in smc6-9 (Fig. 4B). The simultaneous inactivation of transcription and replication barriers allows segregation of chromosome XII in smc6-9 mutants with virtually WT efficiency (Fig. 4B and fig. S6B). Thus, the delay in rDNA replication observed in smc6-9 mutants is a direct consequence of the inability to promote and ensure stable fork progression through the rDNA.

Fig. 4.

Eliminating replication impediments restores segregation in smc6-9 cells. (A) Segregation of tetO:487 tag in smc6-9 and smc6-9 fob1D strains. (B) Segregation of tetO:487 tag in smc6-9, smc6-9 rpa135D, and smc6-9 fob1D rpa135D strains. (C) Diagrammatic representation of the segregation of yeast chromosome XII in WT and smc6-9 mutants. (D) Delayed replication forks in yeast cells do not trigger a replication-completion checkpoint. Error bars in (A) and (B) indicate SD from three independent experiments.

A cellular checkpoint can be defined as a mechanism that halts the cell cycle in WT cells in response to a certain condition (e.g., a cellular insult such as DNA damage), and a gene is classified as part of a checkpoint if mutants of this gene do not arrest the cell cycle as WT cells do. We have shown that smc6-9 mutant cells behave exactly like WT cells with respect to activation and maintenance of all known cellular checkpoints, namely the S phase checkpoint, DNA-damage checkpoint, and spindle checkpoint. Therefore, the Smc5-Smc6 complex is not part of any of these mechanisms. We find that, despite the fact that all known checkpoints are intact in smc6-9 mutants, the replication delay in these cells does not hold up mitotic entry before completing replication (Fig. 4C). The function of the Smc5-Smc6 complex is important for coping with the replication program, particularly on challenging templates such as rDNA. We do not know the exact reason why smc5-smc6 mutants are retarded in replication. One possibility is that Smc5-Smc6 function might be required for accurate fork restart after pausing. We propose that neither ongoing forks nor unreplicated segments of rDNA triggers a classical checkpoint response (Fig. 4D). Here, we have used the rDNA as a model for replication completion. The rDNA region might be considered as a “special genomic locus.” However, we have shown that it is under normal checkpoint surveillance. Because rDNA represents 8 to 12% of the yeast genome, mitotic entry before completion of rDNA replication is detrimental to the integrity of the whole genome. We also note that failure to replicate certain genomic regions, such as centromeres, could potentially induce a cell cycle arrest through activation of the spindle checkpoint or the Rad9-dependent mid-anaphase checkpoint. In mammalian cells, fragile sites have been shown to replicate late and to be sensitive to replication delays (1719), which suggests that forks progressing through those regions in late G2 do not signal a mitotic delay and that the chromosomal breaks and gaps observed in metaphase cells are due to unreplicated DNA. The smc6-9 mutant cells show an increase in chromosomal rearrangements at a non-rDNA locus (20), which, as in mammalian fragile sites, could be attributed to the segregation of partially replicated chromosomes. Studies with the origin recognition complex have shown that, during S phase, there is a fork threshold for checkpoint activation (21). Similarly, in unperturbed cell cycles, ongoing forks below a threshold level might fail to activate a checkpoint during G2-M. Therefore, we propose that both the timely separation between replication and segregation and the reservoir of unused origins are the crucial factors ensuring replication before mitosis.

Supporting Online Material

www.sciencemag.org/cgi/content/full/315/5817/1411/DC1

Materials and Methods

SOM Text

Figs. S1 to S6

References

References and Notes

View Abstract

Navigate This Article