INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

When DNA is damaged or chromosomes are incompletely replicated, cells become checkpoint arrested. These checkpoints avoid replication of damaged template DNA and prevent aberrant segregation of damaged or partly replicated chromosomes. In budding yeast, proteolysis of anaphase inhibitors is regulated by these checkpoint systems. Progression from metaphase to anaphase is inhibited by Pds1p in Saccharomyces cerevisiae (6, 7, 29, 30). Before anaphase, Pds1p binds to Esp1p, inhibiting its anaphase-promoting activity (3). During an unperturbed cell cycle, Pds1p becomes polyubiquitinated at the metaphase-to-anaphase transition by multienzyme anaphase-promoting complex (APC)-cyclosome complexes. The modified forms are then recognized and degraded by 26S proteasomes (7). Once released from Pds1p, Esp1p activity induces the onset of anaphase.

pds1 mutants fail to execute checkpoint control in response to DNA damage, spindle poisons, or replication inhibition (4, 29, 30). Pds1p is required for replication checkpoint control only late in S phase, not in the context of an early S-phase replication block enforced by hydroxyurea (HU) (4, 29, 30). In the presence of 0.1 M HU, replication proceeds more slowly. Under these conditions, cells perform other aspects of cell cycle progression, budding, and spindle assembly as rapidly as in the absence of HU; cell cycle arrest in G₂ is then necessary to delay anaphase while replication is completed. Growth in the presence of a non-replication-arresting concentration of HU therefore challenges the ability of cells to accurately couple S phase with mitosis. Although pds1 mutants can prevent mitotic progression when replication is blocked, defective S-phase checkpoint control can be observed when replication is partially inhibited. Such experiments revealed a novel checkpoint pathway that is essential late in S phase. Components of this pathway, other than Pds1p and Mec1p (4; D. J. Clarke, M. Segal, S. Jensen, and S. I. Reed, submitted for publication), have not been identified. Moreover, the essential function of Pds1p at high temperature is not known. To gain insight into this function, we screened for dosage suppressors of pds1-128 and have characterized two, encoded by RAD23 and DDI1. Ubiquitin-associated (UBA) domains located at the C termini of both Rad23p and Ddi1p are required for suppression of pds1-128. Interestingly, these domains are found in several proteins involved in ubiquitin-dependent proteolysis, though no function has been assigned to UBA domains. Further experiments implicate these UBA domains in checkpoint signaling: strains with mutant versions of the proteins lacking the UBA domains are S-phase checkpoint defective.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Yeast strain construction and growth conditions. All strains are derivatives of BF264-15DU MATa ade1 his2 leu2-3,112 trp1-1^a ura3 Dns (20). Cultures were grown at 30°C, unless otherwise stated, in yeast extract-peptone (YEP) medium containing 2% dextrose, raffinose, or galactose. Strains were constructed according to standard procedures (22) except for gene disruptions (27). pds1 cells were temperature sensitive at 28°C on YEP-dextrose plates in the BF264-15DU genetic background. Rad23p and Ddi1p were epitope tagged at the C-terminal extremities of their genomic coding regions with either six Myc or six His epitopes. The C-terminal region of each coding sequence was amplified by PCR using oligonucleotides that included NotI or SalI sites and then cloned into tagging vectors in frame with the epitope. Such constructs were linearized by restriction enzyme digestion within the RAD23 or DDI1 open reading frame (ORF) and then integrated at the endogenous RAD23 and DDI1 loci. GAL1:RAD23 and GAL1:DDI1 strains were constructed by amplifying the RAD23 and DDI1 coding sequences, cloning into a yeast integration vector behind a GAL1 promoter and then integrating at the LEU2 locus. Epitope tags could then be integrated behind the GAL1:RAD23 and GAL1:DDI1 sequences. rad23^uba2 and ddi1^uba mutant alleles, expressed endogenously or under GAL1 promoter control, were constructed by integrating epitope tags directly 3' of the UBA domain sequences, thus truncating the corresponding coding sequence. Temperature-sensitive esp1 strains were generated by PCR-based mutagenesis (13, 18).

Yeast strain genotypes. Genotypes used are shown in Table 1. Oligonucleotides used are shown in Table 2.

Isolation of pds1-128 dosage suppressors. pds1-128 suppressors were isolated by transformation with an S. cerevisiae genomic library contained in a vector which is maintained at high copy number in yeast cells (YEp24 library [2]). Transformants were replicated onto YEP-galactose plates, and suppressors were selected at 37°C. About 160,000 transformants were analyzed to identify 110 colonies rescued in a plasmid-dependent manner; 80 of these colonies appeared wild type (WT) at 37°C on YEP-galactose medium. Of the 13 clones examined further, all contained PDS1; 30 of the 110 colonies grew more poorly than WT at 37°C on YEP-galactose medium. Plasmids isolated from each of these clones contained either ESP1, RAD23, or DDI1.

Cell biology protocols. Samples were prepared for FACScan analysis as described elsewhere (14). Spindles were visualized by fluorescence microscopy by using strains expressing a TUB1-GFP construct (23). Cells were gamma irradiated in liquid medium (3.5 Gy per min). For G₁ arrest, strains were grown overnight to saturation and then diluted to an optical density of 0.15 to 0.2; then -factor (200 ng/ml) was added; 90 to 100% arrest was complete within 2 to 2.5 h at 26 to 30°C. For experiments in which spindle morphology was analyzed, it was necessary to grow cells overnight in synthetic medium to allow good visualization of the spindles. When induction of genes from the GAL1 promoter was required, overnight cultures were grown in YEP-raffinose. Synchrony was carried out in rich medium as described in the text.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Dosage suppressors of pds1-128. A genetic strategy was used to find proteins that functionally interact with Pds1p. We screened for genes whose increased dosage suppressed the temperature sensitivity of pds1-128. From a genomic S. cerevisiae library (maintained at high copy number in a yeast episomal vector), three dosage suppressors were identified: (i) ESP1 (13 independent isolates, representing six plasmid types), (ii) RAD23 (4 isolates, one plasmid type), and (iii) DDI1 (8 isolates, two plasmid types). Since a genetic interaction between ESP1 and pds1 was previously reported (7), two novel suppressors of pds1-128, RAD23 and DDI1, were identified (Fig. 1a).

View larger version (47K): [in this window] [in a new window]   FIG. 1.   Dosage suppressors of pds1-128. (a) Suppression of pds1-128 but not pds1 temperature sensitivity by episomal plasmids maintained at high copy number in yeast cells carrying YEp24(URA3), YEp-PDS1, YEp-RAD23, YEp-DDI1, and YEp-ESP1. Patches were grown at 25°C on synthetic medium containing dextrose but without uracil to prevent loss of the YEp24-based plasmids. These patches were replicated on to YEP-dextrose plates and grown at 25 to 37°C for 2 days. (b) Genetic interactions between GAL1:RAD23 or GAL1:DDI1 and temperature-sensitive esp1 strains. Strains carrying esp1 alleles (esp1-N5 and esp1-B7 [13]) with or without RAD23 and DDI1 alleles under control of the inducible GAL1 promoter were grown at 30°C on YEP-dextrose plates, then replicated to YEP-galactose plates, and grown for 2 days at 30 to 35°C. (c) Suppression of pds1-128 mec1-1 temperature sensitivity by episomal plasmids maintained at high copy number in yeast cells carrying YEp-PDS1, YEp-RAD23, YEp-DDI1, YEp24, and YCp50, grown as for panel a.

Specificity of pds1-128 dosage suppression. Additional criteria were used to assess the specificity of the genetic interactions with pds1-128: first, whether the suppressors could rescue a pds1 null mutant (pds1); and second, whether genetic interactions with esp1 (encoding a protein that physically interacts with Pds1p) could be detected (Fig. 1a and b).

Dosage suppression of the pds1-128 S-phase checkpoint defect. The genetic interactions of RAD23 and DDI1 with pds1-128 were examined in the context of documented pds1 checkpoint defects. We tested whether induction of GAL1:RAD23 or GAL1:DDI1 was able to suppress the sensitivity of pds1-128 to nocodazole, gamma irradiation, or HU to examine whether Rad23p and Ddi1p could function in the spindle assembly, DNA damage, or S-phase checkpoint pathways. There was no suppression of the gamma irradiation sensitivity of pds1-128 (data not shown). In contrast, a high-copy-number plasmid containing the pds1-128 allele (YEp-pds1-128) did rescue the gamma irradiation sensitivity. Thus, increasing the amount of the mutant Pds1-128p corrected the DNA damage checkpoint defect of these cells. Similarly, YEp-pds1-128 rescued the temperature sensitivity of pds1-128 cells. By Western blot analysis, pds1-128 cells have substantially less Pds1 than do WT cells (13). Therefore, in the context of the DNA damage checkpoint, RAD23 or DDI1 overexpression presumably cannot increase the amount of Pds1-128p. Overexpression of RAD23 or DDI1 was also unable to rescue the nocodazole sensitivity of pds1-128 cells (data not shown). In this case, however, YEp-pds1-128 was similarly unable to rescue lethality caused by nocodazole treatment. Therefore, it was not possible to infer whether or not RAD23 and DDI1 overexpression affects Pds1-128p levels in the context of the spindle assembly checkpoint.

View larger version (57K): [in this window] [in a new window]   FIG. 2.   Suppression of pds1-128 S-phase checkpoint defects. (a) Suppression of pds1-128 HU sensitivity by RAD23 and DDI1 overexpression. Strains were grown to midlog phase, and then serial dilutions were spotted on to YEP-dextrose or YEP-galactose plates (to repress or induce GAL1:RAD23 and GAL1:DDI1) containing HU and grown for 2 to 3 days. (b) Overexpression of RAD23 increases Pds1-128p levels in HU-treated cells. Cells containing a GAL1:RAD23 gene and a hemagglutinin (HA) epitope-tagged pds1-128 allele were grown to midlog phase (minus HU) or accumulated in S-phase with 100 mM HU for 3 h (plus HU) in dextrose () or galactose (+) medium to repress or induce GAL1:RAD23. Cellular protein extracts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Pds1-128-p-HA was detected after Western blotting. Cdc28p levels served as a loading control. A light exposure and a darker exposure of the same blot are shown.

Role of Rad23p and Ddi1p UBA domains in dosage suppression of pds1-128. One striking similarity between the C termini of Rad23p and Ddi1p is that both contain a UBA domain (Rad23p has a second, internally located UBA domain [Fig. 3a]). These are small domains (each about 10% of the full-length protein) of unknown function. Six UBA domain-containing proteins have been identified in budding yeast.

View larger version (49K): [in this window] [in a new window]   FIG. 3.   Suppression of pds1-128 temperature and HU sensitivity depends on Rad23p and Ddi1p UBA domains. (a) Structural similarity between RAD23 and DDI1. Box diagrams show relative sizes of each ORF and the size and position of UBA, XPC, and UBL domains. C-terminal UBA domains (right) share 38% identity. The position of Ddi1p L426 is indicated in the alignment below. A hypothetical fission yeast protein (gb|Z69728 and gi|1204230) has an overall 32% identity to budding yeast Ddi1p. Probable Ddi1p homologues giving high identity scores are present in Homo sapiens (gb|AA406136 and gi|2064117; dbest database), Caenorhabditis elegans (gb|U50068 and gi|1208859) and Leishmania major (gb|AC002305 and gi|2429118). (b and c) Temperature sensitivity (b) and HU sensitivity (c) of pds1-128 strains overexpressing RAD23 and DDI1 or truncated versions of these genes expressing proteins lacking UBA domains or the RAD23 UBL. Similar results were obtained when the WT or truncated proteins were tested as (Rad23p-MYC6 and Ddi1p-His6) fusions. Strains were handled as described in the legends to Fig. 1b and 2a. (d) Lack of genetic interaction between GAL1:rad23uba2 and temperature-sensitive esp1. Strains were handled as described in the legend to Fig. 1b.

GAL1:rad23^uba2

GAL1:ddi1^uba

GAL1:ddi1^uba

rad23 ddi1 null mutants are sensitive to HU and defective in cell cycle progression in the presence of HU. The genetic and functional interactions of RAD23 and DDI1 with pds1-128 suggested that Rad23p and Ddi1p may function in the S-phase checkpoint pathway and that the putative checkpoint function of Rad23p and Ddi1p depends on conserved UBA domains. It could therefore be expected that loss-of-function alleles of RAD23 or DDI1 would result in loss of S-phase checkpoint control. The common UBA domains of Rad23p and Ddi1p suggested that the two proteins may have some functional redundancy. We therefore constructed RAD23 and DDI1 deletion strains and tested their HU sensitivity. (Fig. 4a). rad23 ddi1 cells were as sensitive to HU as are pds1-128 cells, though rad23 and ddi1 single mutants were either much less sensitive (rad23) or not sensitive at all (ddi1). rad23 ddi1 cells were not sensitive to gamma irradiation (Fig. 4b) or nocodazole (not shown).

View larger version (44K): [in this window] [in a new window]   FIG. 4.   rad23 ddi1 cells are HU sensitive but not radiation sensitive. (a) Sensitivity of pds1 and rad23 ddi1 mutants to HU. Strains were grown to midlog phase, then serial dilutions were spotted on to YEP-dextrose or YEP-dextrose plates containing HU, and growth continued for 2 to 3 days at 30°C. (b) Sensitivity of pds1, rad23, ddi1, and rad9 mutants to gamma irradiation. Strains were grown to midlog phase at 30°C in YEP-dextrose medium, irradiated, and then plated on YEP dextrose. Colonies were counted after 2 to 4 days.

rad23 ddi1

rad23 ddi1

rad23 ddi1

rad23 ddi1

rad23 ddi1

View larger version (17K): [in this window] [in a new window]   FIG. 5.   S-phase checkpoint defect of rad23 ddi1 mutants (a) Rebudding assay on HU plates. WT, pds1-128, and rad23 ddi1 cells were arrested in G1 at 30°C (-factor), washed, and plated on YEP-dextrose plates containing 100 mM HU. Each strain budded with similar timing (not shown). Rebudding (the appearance of microcolonies containing three or more cell bodies) was scored at various time intervals (see also Fig. 6). (b) Rebudding assay following gamma irradiation. The procedure was as described for Fig. 5a. The ability of cells to rebud on YEP-dextrose plates following gamma irradiation in G1 was scored. (c) Timing of budding, G2 spindle formation, and spindle elongation in WT and mutant cells grown in liquid YEP-dextrose medium containing 100 mM HU following release from G1 (-factor).

rad23 ddi1

rad23 ddi1

rad23 ddi1

rad23 ddi1

rad23 ddi1

rad23 ddi1

Role of Rad23p and Ddi1p UBA domains in S-phase checkpoint control. The potential dual roles of Rad23p in nucleotide excision repair (NER) and checkpoint control might result in a somewhat ambiguous checkpoint-defective phenotype in the rad23 ddi1 mutant. Although these cells appeared to initiate anaphase before S phase was complete, they also appeared to be delayed for progression through the latter parts of mitosis. To resolve this issue, we attempted to create strains with mutant versions of Rad23p and Ddi1p that would be defective for checkpoint control but not compromised in NER or in the ability to progress through mitosis. We constructed strains in which endogenous RAD23 and DDI1 were replaced with mutant versions lacking the C-terminal UBA domains (rad23^uba2 and ddi1^uba), based on the genetic evidence that these mutant versions cannot suppress pds1-128. The UBA domain deletion alleles were epitope tagged to confirm that expression levels of the resulting products were similar to those of the WT proteins (not shown). We reasoned that since these domains are at least partly required for suppression of pds1-128 HU sensitivity, it may be possible to detect a checkpoint defect in strains with UBA domain deletions. Either a rad23^uba2 or a ddi1^uba mutation showed only a marginal sensitivity to HU (not shown). In contrast, strains with both mutated alleles were HU sensitive (Fig. 6a), though not as sensitive as pds1-128 strains. This suggested that rad23^uba2 ddi1^uba cells might be partially defective in S-phase checkpoint regulation. rad23^uba2 ddi1^uba cells were not sensitive to nocodazole, gamma irradiation, or UV irradiation (not shown), correlating with the failure of high dosage RAD23 and DDI1 to rescue the sensitivity of pds1-128 to these agents and confirming that the rad23^uba2 ddi1^uba strain is NER competent.

View larger version (35K): [in this window] [in a new window]   FIG. 6.   S-phase checkpoint defect of rad23uba2 ddi1uba mutant cells. (a) Sensitivity of rad23uba2 ddi1uba mutant cells to HU. Strains were grown to midlog phase, then serial dilutions were spotted onto YEP-dextrose or YEP-dextrose plates containing HU, and growth continued for 2 to 3 days at 30°C. (b) Rebudding assay on HU plates or following gamma irradiation (see legend to Fig. 5 for details). Microcolonies were scored based on the number of cell bodies per microcolony. (c) Timing of short spindle formation, S-phase progression, and spindle elongation in WT and mutant cells grown in liquid YEP-dextrose medium containing 100 mM HU following release from G1 (-factor). S-phase progression was monitored by FACScan analysis; histogram plots show DNA content throughout the experiment (in grey), and overlaid solid lines indicate positions of 1C and 2C DNA peaks taken from the asynchronous cultures prior to G1 arrest. The percentage of the total cells that had elongated spindles above each histogram. (d) Timing of loss of sister chromatid cohesion in WT and rad23uba2 ddi1uba mutant cells grown in liquid YEP-dextrose medium containing 100 mM HU following release from G1 (-factor). Both strains budded and replicated DNA with similar timing (FACScan analysis not shown), but loss of cohesion occurred prematurely in the mutant cells. Cohesion was monitored at the TRP1 locus using the tetR-GFP/tetO system as previously described (4).

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23 ddi1

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Genetic interactions of pds1-128 with RAD23 and DDI1. The key role of Pds1p in M-phase cell cycle control has been well documented, but less is known about factors regulating Pds1p in response to checkpoint signals. We have used a genetic screen to identify proteins that interact with Pds1p. High dosage of RAD23 or DDI1 suppressed the temperature sensitivity of pds1-128 and rescued the HU sensitivity of pds1-128 at the permissive temperature. Neither the temperature sensitivity nor the HU sensitivity was rescued by RAD23 or DDI1 overexpression in pds1 strains; there was no bypass of the cellular requirement for Pds1p. In agreement, high-dosage RAD23 and DDI1 enhanced the temperature sensitivity of esp1 mutants. This would be expected of proteins that positively regulate Pds1p, since Esp1p separin activity, required for progression through mitosis, is negatively regulated by Pds1p. These interactions implicate Rad23p and Ddi1p in promoting Pds1p-dependent functions. Moreover, RAD23 and DDI1 specifically corrected the S-phase checkpoint defect of pds1-128 cells. Neither the gamma irradiation sensitivity nor the nocodazole sensitivity of pds1-128 was rescued by overexpressing RAD23 and DDI1. Thus, these data implicate Rad23p and Ddi1p in subset of the Pds1p-dependent checkpoint pathways.

UBA domains of Rad23p and Ddi1p are required for S-phase checkpoint control. Rad23p is a highly conserved protein with a NER function. The role of Rad23 in NER depends on binding to the NER complex via its XPC-binding domain and on an interaction with the proteasome cap via its UBL domain (9-11, 16, 17, 24, 28). Almost nothing is known about the function of Ddi1p. It has an upstream activation domain identical to that of RAD23 and is induced by DNA damage and HU, but unlike rad23 mutants, ddi1 strains are not sensitive to UV light (data not shown) and are therefore not required for NER.

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

rad23^uba2 ddi1^uba

Regulation of Pds1p by Rad23p and Ddi1p? DNA damage, spindle perturbations, and replication inhibition increase Pds1p stability, which in turn inhibits the onset of anaphase. We have shown that RAD23 overexpression increases the level of Pds1-128p in a manner that is enhanced by S-phase checkpoint activation. The proposal that Rad23p and Ddi1p regulate Pds1p stability is an attractive one given the dependence of rescue on UBA domains. The function of UBA domains is not known, though they are present in different classes of enzyme involved in ubiquitin-dependent proteolysis (12), an intriguing coincidence given the dependence of Pds1p proteolysis on the ubiquitin system. Rad23p also has a UBL domain at its N terminus; this domain was recently shown to be required for a physical interaction with 26S proteasomes, the particles that degrade ubiquitinated Pds1p to initiate the onset of anaphase. However, deletion of this domain did not abrogate the ability of Rad23p to rescue pds1-128 when overproduced, and other data implicate the UBL domain in the NER function of Rad23p (21, 28). An alternative model for suppression of pds1-128 by RAD23 and DDI1 was recently suggested by the finding that both Rad23p and Ddi1p could bind to ubiquitin and that this interaction was dependent on the UBA domains (Bertolaet and Reed, unpublished). These data make it tempting to propose an intriguing mechanism for Rad23p- and Ddi1p-dependent checkpoint regulation. Rad23p and Ddi1p could regulate Pds1p stability by binding to ubiquitinated Pds1p or to the ubiquitin ligase complex that ubiquitinates Pds1p via the UBA domains (Fig. 7). Indeed, a recent report described a novel activity of Rad23p as an inhibitor of ubiquitin chain elongation (19). Rad23p was shown to bind to a mono- or a diubiquitinated substrate and prevent extension of the ubiquitin chain. A similar interaction with ubiquitinated Pds1p, or the ubiquitin ligase complex, stimulated by ongoing DNA replication, may provide one mechanism through which the S-phase checkpoint is enforced. How UBA domains participate in such control systems is currently being addressed.

View larger version (37K): [in this window] [in a new window]   FIG. 7.   Model for control of Pds1p stability. Pds1p can become polyubiquitinated by APCCdc20, a multienzyme E3 ubiquitin ligase. The ubiquitin chain enables a 26S proteasome to recognize and degrade Pds1p. These events must occur prior to the onset of anaphase, since Pds1p inhibits the activity of separin Esp1p, needed for anaphase spindle elongation and loss of sister chromatid cohesion. To block anaphase onset, checkpoint controls prevent Pds1p degradation. The spindle assembly checkpoint directly inhibits APCCdc20, and the DNA damage checkpoint is thought to promote Pds1p stability by a Chk1p-dependent phosphorylation of Pds1p. We propose a novel mechanism for anaphase control, in which Rad23p and Ddi1p are able to recognize ubiquitinated Pds1p, or the APC (not shown), and thereby inhibit the ubiquitin chain elongation that is required for targeting Pds1p to the 26S proteasome.