The conserved fission yeast protein Rad26ATRIP preserves genomic stability by occupying central positions within DNA-structure checkpoint pathways. It is also required for proper cellular morphology, chromosome stability and following treatment with microtubule poisons. Here, we report that mutation of a putative nuclear export sequence in Rad26ATRIP disrupted its cytoplasmic localization in untreated cells and conferred abnormal cellular morphology, minichromosome instability and sensitivity to microtubule poisons without affecting DNA-structure checkpoint signaling. This mutation also disrupted a delay to spindle-pole-body separation that occurred following microtubule damage in G2. Together, these results demonstrate that Rad26ATRIP participates in two genetically defined checkpoint pathways – one that responds to genomic damage and the other to microtubule damage. This response to microtubule damage delays spindle-pole-body separation and, in doing so, might preserve both cellular morphology and chromosome stability.
Introduction
Prokaryotes and eukaryotes delay cell-cycle progression in response to DNA damage (Hartwell and Weinert, 1989; Opperman et al., 1999). In eukaryotes, DNA-structure checkpoints are required for these cell-cycle delays, which coordinate DNA repair with division (Weinert and Hartwell, 1988; Carr, 1995). As failure to repair DNA damage can lead to genetic instability and cancer (Hartwell, 1992; Weinert and Lydall, 1993; Weinert and Hartwell, 1988), these checkpoints might form barriers between precursor lesions and tumorigenesis (Bartkova et al., 2005).
Elegant yeast screens led to the discovery of mutants defining DNA-structure checkpoint proteins (Weinert and Hartwell, 1988; al-Khodairy and Carr, 1992; Rowley et al., 1992; Enoch et al., 1992; Walworth et al., 1993; al-Khodairy et al., 1994; Weinert and Hartwell, 1988; Weinert et al., 1994). These proteins participate in pathways composed of sensors and transducers that detect and transmit the presence of unreplicated and damaged DNA to the cell-cycle machinery (for reviews, see Kastan and Bartek, 2004; Rouse and Jackson, 2002b; Melo and Toczyski, 2002). The checkpoint sensor complexes detect abnormal DNA structures and/or processed lesions using parallel pathways. Following detection, physical assembly of these complexes somehow activates a PIKK (phosphoinositide 3-kinase related kinase) family member (Bonilla et al., 2008). PIKKs are evolutionarily related to phosphoinositide 3-kinases; members of both families share a highly conserved C-terminal kinase domain of ~300 residues that distinguishes them from all other eukaryotic kinases (Keith and Schreiber, 1995). Two PIKKs, called ataxia telangiectasia mutated (ATM) and ataxia telangiectasia and Rad3-related (ATR), are central to DNA checkpoint pathways in humans (for a review, see Abraham, 2001). Once activated by the sensor complexes, they phosphorylate downstream transducing kinases, including Chk1 and Chk2, which negatively regulate mitotic cyclin-dependent kinase.
It is becoming clear that these DNA-structure checkpoint pathways also influence spindle dynamics and cellular morphology. Crosstalk between the DNA-structure and spindle-assembly checkpoints was first observed in budding yeast, in which the checkpoints jointly mediate arrest during replication stress (Garber and Rine, 2002). Coordination between these pathways has now been observed in many eukaryotes and acts to sustain the anaphase delay following either genomic (Clemenson and Marsolier-Kergoat, 2006; Collura et al., 2005; Krishnan et al., 2004; Mikhailov et al., 2002) or microtubule stress (Zachos et al., 2007). These delays, which probably help to ensure proper chromosome segregation, might also be relevant during unperturbed division cycles that require extra time due to stochastic errors in kinetochore-microtubule attachment (for a review, see Zachos and Gillespie, 2007).
The altered morphology of ataxia-telangiectasia fibroblasts that carry primary lesions in ATM originally suggested that DNA checkpoint activity somehow influences morphology (McKinnon and Burgoyne, 1985). Studies with fission yeast, budding yeast and filamentous fungi confirm that this influence exists across evolution (Baschal et al., 2006; Malavazi et al., 2006; Enserink et al., 2006). It remains to be determined whether checkpoint proteins directly influence morphology by genetically defined pathways or whether morphological alterations arise as indirect consequences of the genetic instability that occurs following loss of DNA-structure checkpoints.
Fission yeast Rad3ATR is a PIKK that is central to DNA-structure checkpoint signaling (Bentley et al., 1996). Rad3ATR physically associates with Rad26ATRIP, a regulatory subunit required for normal levels of Rad3 kinase activity (Edwards et al., 1999; Wolkow and Enoch, 2002). This Rad3-Rad26 checkpoint complex is conserved throughout evolution and exists in humans (ATR-ATRIP), budding yeast (MEC1-LCD1DDC2/PIE1), Xenopus (xATR-xATRIP) and possibly filamentous fungi (UvsB-UvsD) (Cortez et al., 2001; Rouse and Jackson, 2002a; Paciotti et al., 2000; Wakayama et al., 2001; De Souza et al., 1999).
Fission yeast Rad3 and Rad26 are also required when cells are challenged with the microtubule poisons methyl benzimidazole carbamate (carbendazim; MBC) and thiabendazole (TBZ) (Staron and Allard, 1964; Wolkow and Enoch, 2003; Baschal et al., 2006). This sensitivity is not caused by loss of DNA checkpoint activity, because checkpoint-deficient hus1Δ and rad17Δ cells exhibited wild-type growth on plates containing these drugs (Wolkow and Enoch, 2003). Furthermore, cytoplasmic Rad26-GFP accumulated during treatment with microtubule poisons, but not during treatment with genotoxins, suggesting that Rad26 responds specifically to microtubule damage (Baschal et al., 2006).
Here, we report the consequences of mutating a conserved putative nuclear export signal of Rad26ATRIP. We observed that this mutation disrupts the localization of cytoplasmic Rad26ATRIP in untreated cells and confers sensitivity to microtubule toxins but not genotoxins. These mutant cells also exhibit aberrant morphology, minor chromosome instability and a failure to delay spindle-pole-body separation during MBC treatment in G2. Rad26 performs these functions independently of the spindle-assembly checkpoint protein Mad2. These results suggest that Rad26ATRIP participates in a checkpoint response to microtubule damage in G2 that delays entry into mitosis and, in doing so, might preserve both cellular morphology and chromosome stability.
Results
Rad26 and Mad2 respond to microtubule damage using different pathways
A benzimidazole-resistant α-tubulin allele (Yamamoto, 1980; Umesono et al., 1983) rescued the growth of rad26Δ cells on MBC (Fig. 1A), but not on the DNA-damaging agent phleomycin (Fig. 1A). We conclude that rad26+ is required when microtubules are damaged.
The mad2-dependent spindle-assembly checkpoint delays exit from mitosis when microtubules are damaged (Li and Murray, 1991). Given that crosstalk between DNA-structure and spindle-assembly checkpoints exists, we hypothesized that rad26+ might function to delay exit from mitosis during microtubule damage. Here, we investigated this hypothesis using a nda3-311 cold-sensitive β-tubulin allele that blocks spindle formation and has been used to assay spindle-assembly checkpoint proficiency (Kanbe et al., 1990; He et al., 1997; Sczaniecka et al., 2008). When nda3-311 cells are downshifted to 20°C, strains with an intact spindle-assembly checkpoint delay mitosis and accumulate condensed chromatin, whereas strains deficient in the spindle-assembly checkpoint progress through mitosis and into the next cell cycle. We found that 76% of rad26+ and 77% of rad26Δ cells arrested with highly condensed nuclei following the downshift, compared to 22% of mad2Δ cells and 21% of rad26Δmad2Δ double-mutant cells (Fig. 1B). Therefore, rad26+ does not participate in the mad2-dependent pathway that blocks exit from mitosis when microtubules are damaged in an nda3-311 background.
We also tested whether rad26+ is required to delay DNA replication during MBC treatment. For these experiments, flow cytometry was used to quantitate DNA in cells treated with 8 μg/ml MBC during a four-hour time course (Fig. 1C). The data show that rad26+ and rad26Δ strains respond to MBC treatment by delaying further rounds of DNA replication. The same was not true for mad2Δ cells, which continued to replicate DNA despite the presence of MBC. Again, we observe that rad26+ is not required in a mad2-dependent pathway.
We note that, whereas rad26+ cells began to synthesize DNA and recover from MBC in the third hour of treatment, rad26Δ cells displayed a sharp arrest (Fig. 1C). This suggests that rad26+ might be required to recover from microtubule stress; experiments to test this are underway.
If Rad26 and Mad2 perform different roles during microtubule damage, then the rad26Δmad2Δ double mutant should exhibit additive defects. Using plate assays, we observed that the mad2Δ allele conferred greater MBC sensitivity than the rad26Δ allele, whereas the rad26Δmad2Δ double mutant displayed an additive phenotype (Fig. 1D). We conclude that Rad26 and Mad2 participate in different pathways that, when disrupted, cause greater sensitivity to MBC than disruption of either pathway alone.
Rad26 has a putative nuclear export sequence
Our previous model suggested that cytoplasmic Rad26 participates in a response to microtubule damage (Baschal et al., 2006). Using the Minimotif Miner tool (Balla et al., 2006), we identified a cluster of hydrophobic residues near the C terminus of Rad26 that resembles a nuclear export sequence (NES) (Kutay and Guttinger, 2005). The ClustalW2 multiple sequence alignment tool (Larkin et al., 2007) revealed that the C-terminal location of this putative NES is conserved among homologous proteins in fission yeast, budding yeast, filamentous fungi and humans (Fig. 2A) (al-Khodairy et al., 1994; Wakayama et al., 2001; Paciotti et al., 2000; De Souza et al., 1999; Cortez et al., 2001). In a yeast two-hybrid screen for proteins with nuclear export activity, it was discovered that this region of Lcd1ATRIP (referred to as YDR499W in the report) physically interacts with Crm1 and, when mutated (L670A or L673A), causes nuclear accumulation of GFP-Lcd1ATRIP (Jensen et al., 2000). Together, these results suggest that this C-terminal motif is a bona fide NES in fission yeast Rad26ATRIP.
The rad26:4A allele confers sensitivity to MBC
To test whether this conserved motif of Rad26 is required during microtubule damage, we changed four of the hydrophobic residues to alanines (Fig. 2A). The allele (rad26:4A) was then integrated into a rad26Δ strain, where its expression was driven by the endogenous rad26+ genomic promoter. Plating assays were used to characterize the fitness of this strain during exposure to three different toxins (Fig. 2B). Phleomycin was used to induce DNA breaks (Sleigh and Grigg, 1977), hydroxyurea (HU) was used to stall replication forks (Zhao et al., 1998) and MBC was used to damage microtubules (Jacobs et al., 1988). The control dilutions in Fig. 2B show that the growth of rad26Δ cells was inhibited by all three toxins. Growth of the rad26:4A strain on HU and phleomycin was actually enhanced in comparison with the rad26+ strain. However, rad26:4A growth was inhibited by MBC. Therefore, the rad26:4A allele specifically compromises growth on medium containing MBC.
Cytoplasmic localization of Rad26:4A-YFP is compromised in untreated cycling cells
If this conserved C-terminal motif functions as an NES, then the cytoplasmic localization of Rad26:4A should be affected. This was tested by tagging Rad26 and Rad26:4A with YFP, a tag that did not affect the growth characteristics of rad26:4A cells (Fig. 2B) or rad26+ cells (data not shown). Whereas control experiments verified that Rad26-YFP localized to both the cytoplasm and nuclei of untreated cycling cells, we observed little Rad26:4A-YFP in the cytoplasm and an elevated signal in the nuclei (Fig. 3A). Fluorescence intensity measurements verified that the cytoplasmic Rad26:4A-YFP signal was weaker than the cytoplasmic Rad26-YFP signal, and that the nuclear Rad26:4A-YFP signal was stronger than the nuclear Rad26-YFP signal (Fig. 3B). Therefore, this conserved hydrophobic C-terminal motif is required for cytoplasmic localization of Rad26 in untreated cycling cells.
Rad26 accumulates in the cytoplasm following MBC treatment (Baschal et al., 2006). Here, we observed that both Rad26-YFP and Rad26:4A-YFP accumulate in the cytoplasm during MBC treatment (Fig. 3A). Therefore, this C-terminal motif is not required for cytoplasmic accumulation of Rad26 during MBC treatment, a response that might be directed by other putative NESs that appear in the Rad26 primary sequence (data not shown). That this response occurs in rad26:4A cells only demonstrates that, in and of itself, cytoplasmic accumulation of Rad26 is not sufficient for a proper response to microtubule damage.
The DNA-damage checkpoint of rad26:4A cells is intact
Although the plate assays in Fig. 2B suggest that Rad26:4A functions normally in DNA checkpoint pathways, we used two more assays to test the possibility that it does not. First, liquid cultures treated with phleomycin were used to test whether rad26:4A cells exhibit a phenotypically normal checkpoint response. Control experiments showed that the DNA-damage checkpoint functions normally in rad26+ and rad26:4A cells, which continued to elongate while restraining nuclear division and septation during phleomycin treatment (Fig. 4A). The rad26Δ cells displayed the classic checkpoint-defective phenotype (Enoch and Nurse, 1990) and progressed through mitosis and septation in the presence of phleomycin. Therefore, rad26:4A cells exhibit a phenotypically normal checkpoint response.
Rad3-dependent phosphorylation of Rad26 occurs during DNA-damage checkpoint activation and is used as a biochemical marker of checkpoint function (Edwards et al., 1999). To verify that the rad26:4A cells have a functional DNA-damage checkpoint, we tested whether Rad26:4A was phosphorylated during phleomycin treatment. Western blots showed that both Rad26 and Rad26:4A were expressed at very similar levels and phosphorylated during phleomycin treatment (Fig. 4B). We conclude that the DNA-damage checkpoint functions properly in the rad26:4A background.
The rad3-gfp allele also genetically separates the microtubule-damage response from the DNA-structure checkpoint response
In the course of these studies, we observed that the growth of rad3-gfp cells was compromised on phleomycin and HU plates, but not on MBC plates, whereas the growth of rad3Δ and rad26Δ cells was compromised on all three types of medium (Fig. 5). The rad3-gfp cells also failed to arrest mitosis in the presence of phleomycin (Fig. 4A), confirming that they have lost the DNA-structure checkpoint. The rad3-gfp allele also co-segregated with sensitivity to growth on plates containing HU (data not shown). Therefore, the rad3-gfp allele separates the role of Rad3 during microtubule damage from its roles during genomic damage. This allele is used within the next series of experiments to help categorize phenotypes associated with the rad26:4A allele and concomitant loss of the Rad26-dependent response to microtubule damage.
The rad26:4A allele influences morphology and chromosome stability
We reported previously that rad26Δ and rad3Δ cells display morphological defects and spontaneous chromosome instability (Baschal et al., 2006). Here, we used rad26:4A cells (sensitive to microtubule damage but not genomic damage) and rad3-gfp cells (sensitive to genomic damage but not microtubule damage) to determine whether morphology and/or chromosome stability are influenced by this response to microtubule damage.
Fission yeast are rod-shaped, cylindrical cells that elongate from each end (Mitchison and Nurse, 1985). We imaged untreated cycling cells with differential interference contrast (DIC) microscopy and observed that rad26+ cells were 2.53-fold longer than they were wide (L/W ratio=2.53; Fig. 6A). Similar to our previous report (Baschal et al., 2006), rad26Δ cells were more spherical, with a L/W ratio of 2.03. The rad26:4A, rad3Δ and mad2Δ cells shared this spherical appearance and very similar L/W ratios of 1.99, 2.05 and 2.01, respectively. By contrast, rad3-gfp cells retained the normal cylindrical rod shape and the wild-type L/W ratio of 2.47. We conclude that morphology is influenced by microtubule-damage responses and not by the DNA-damage response.
Fission yeast chromosome-loss assays can be performed using ade− strains that carry a genomic ade6-210 allele (Javerzat et al., 1996). These strains are then converted to ade+ strains by means of intragenic complementation with the ade6-216 allele of a centromere-containing minichromosome. Estimates of chromosome loss are calculated by determining the number of cells that lose the minichromosome and become ade− after 40 hours of growth in complete liquid media containing adenine (Baschal et al., 2006). We observed that 0.19% of rad26+ and 3.21% of rad26Δ cells spontaneously lost the minichromosome during the growth period (Fig. 6B), showing that rad26+ prevents spontaneous minichromosome loss. The rad26:4A cells also lost the minichromosome at an elevated level (0.62%) that was significantly greater than that of rad26+ cells but significantly less than rad26Δ cells (χ2P<0.05). We conclude that the rad26:4A allele defines a function that contributes slightly, but significantly, to chromosome maintenance.
Next, we used the rad3-gfp allele (only DNA checkpoint defective) to test whether DNA checkpoint function contributes to chromosome stability (Fig. 6B). Our results show that the minichromosome instability of rad3-gfp cells (2.83%) was significantly greater than that of rad26:4A cells (0.62%), but less than that of rad3Δ cells (3.12%). Minichromosome loss in the rad3-gfp rad26:4A double mutant (3.19%) was not significantly different from that of rad26Δ (3.21%) or rad3Δ (3.12%). Together, these data demonstrate that Rad26 and Rad3 influence spontaneous chromosome instability by at least two different mechanisms: a minor mechanism that is absent in rad26:4A cells and a more significant mechanism that is absent in rad3-gfp cells.
Lastly, we used this chromosome-loss assay to test whether Mad2 participates in either of these two chromosome-loss mechanisms (Fig. 6B). First, we observed that mad2Δ and rad26:4A produced an additive effect in which the mad2Δrad26:4A double mutant (1.67%) lost a chromosome amount almost identical to the sum of the single mutants (mad2Δ=0.96%; rad26:4A=0.62%). This is consistent with our conclusion that Rad26 and Mad2 operate in different pathways. Second, striking phenotypes appeared when the mad2Δ allele was crossed into rad26Δ and rad3-gfp backgrounds, as ≥12% of the mad2Δrad26Δ and mad2Δrad3-gfp double mutants experienced minichromosome loss. This suggests that the spindle-assembly checkpoint and the DNA checkpoint pathways cooperate synergistically to preserve faithful chromosome transmission.
Rad26 delays mitotic entry during MBC treatment
MBC is known to inhibit late interphase events of the fission yeast cell cycle (Walker, 1982). Recently, it was reported that 75 μg/ml MBC causes a G2 delay in fission yeast (Balestra and Jimenezk, 2008). We tested whether rad26+ was required for this MBC-dependent delay using a temperature-sensitive cdc25.22 allele that reversibly blocks cells in G2 and the spindle-pole-body marker Cut12-EGFP to indicate mitotic entry (Fantes, 1979; Bridge et al., 1998; Craven et al., 1998). A representative G2 cell with one Cut12-EGFP signal and a representative mitotic cell with two Cut12-EGFP foci are shown in Fig. 7.
When released to 20°C in the absence of MBC, the timing of spindle-pole-body separation was similar in rad26+, mad2Δ and rad3-gfp cells, but slightly accelerated in both rad26Δ and rad26:4A cells. When released to 20°C in the presence of 16 μg/ml MBC, rad26+, mad2Δ and rad3-gfp cells delayed spindle-pole-body separation for more than 180 minutes. However, rad26Δ and rad26:4A cells progressed slowly into mitosis, reaching ~50% with two Cut12 foci by 180 minutes. Therefore, the rad26:4A allele compromises this delay, which occurs independently of the DNA-structure and spindle-assembly checkpoints.
Discussion
We identified a conserved, hydrophobic C-terminal motif in Rad26ATRIP that was shown previously to function as an NES for Lcd1ATRIP in undamaged cycling cells (Jensen et al., 2000). Here, we report that this motif is also required for proper cytoplasmic localization of Rad26ATRIP in undamaged cycling cells (Fig. 3). Loss of cytoplasmic Rad26ATRIP by the rad26:4A allele accompanied phenotypes that all implicate a compromised response to microtubule damage: sensitivity to microtubule toxins; morphological abnormalities; spontaneous minichromosome segregation errors; and failure to properly delay mitosis when microtubules are damaged. Meanwhile, the DNA-damage response of these rad26:4A cells was preserved. In fact, these cells displayed greater resistance to phleomycin than rad26+ cells (Fig. 2B), possibly due to higher concentrations of nuclear Rad26ATRIP (Fig. 3). An explanation for these data is that nuclear Rad26ATRIP is required when DNA is damaged and cytoplasmic Rad26ATRIP is required when microtubules are damaged.
Results with budding yeast, Drosophila melanogaster and human cells have previously established that DNA-checkpoint proteins influence cytoplasmic responses. However, these responses depend on genomic damage. In budding yeast, DNA damage initiates a nuclearly delimited cascade of checkpoint events that delay mitosis (Demeter et al., 2000) and a cytoplasmic cascade that influences nuclear movements (Dotiwala et al., 2007). Experiments with Drosophila Chk2 (Takada et al., 2003) and human Chk1 (Kramer et al., 2004; Loffler et al., 2007) have established that cytoplasmically localized metazoan checkpoint proteins associate with centromeres, where they delay mitotic events during genotoxic stress. Here, we found that fission yeast Rad26ATRIP possibly controls a cytoplasmic response to microtubule damage that is independent of genomic damage.
TBZ, a benzimadazole derivative that targets microtubules, is also known to inhibit late interphase events of the fission yeast cell cycle (Staron and Allard, 1964; Walker, 1982). Using 20 μg/ml TBZ, we also observed that TBZ caused fission yeast to delay spindle-pole-body separation (our unpublished data) and septation, and that both delays were independent of rad26+ (Baschal et al., 2006). This might occur because TBZ, and not MBC, depolymerizes cortical actin and disrupts cell-polarity markers, in addition to its affects on microtubules (Sawin and Snaith, 2004). The TBZ-dependent delay to mitosis might therefore result from the activation of multiple checkpoint pathways that might or might not include the Rad26-dependent pathway discussed here.
On the mechanism of this response
A mechanism in fission yeast that delays mitotic entry in response to MBC-dependent microtubule damage during G2 was reported recently (Balestra and Jimenez, 2008). The authors found that growth of wee1Δ and cdc2-1w cells (insensitive to Wee1 inhibition; Enoch and Nurse, 1990) was MBC sensitive, whereas growth of cdc25Δ and cdc2-3w cells (insensitive to Cdc25 activation) (Enoch and Nurse, 1990) was not. G2-synchronized wee1 cells also failed to delay mitosis following the addition of 75 μg/ml MBC, whereas cdc25 cells delayed mitosis normally. Moreover, they observed that Wee1, degradation of which triggers mitosis (McGowan and Russell, 1995; Aligue et al., 1997; Muñoz et al., 1999; Watanabe et al., 2005), was stabilized in MBC-treated cells. These results show that a checkpoint mechanism responds to MBC-dependent G2 microtubule damage by stabilizing Wee1 and delaying mitotic entry.
Many questions about this Rad26-dependent response remain. For example, does Rad26ATRIP function to target and stabilize the Wee1 kinase when interphase microtubules are damaged? Does Rad26 react to physical perturbations of the microtubule cytoskeleton or a secondary consequence of microtubule damage? Do the morphological abnormalities and chromosome-segregation errors of rad26:4A cells arise because they enter mitosis with microtubule damage?
Morphology and chromosome segregation
Morphological abnormalities of rad26:4A cells might develop as these cells progress into mitosis despite the presence of interphase microtubule damage. In other words, abnormal morphology might be an indirect consequence of the checkpoint defect. However, it is possible that Rad26 directly influences morphology. For example, human ATM has been shown to specifically affect RhoA activity during the DNA-damage response and physically interact with CKIP-1, a regulator of the actin cytoskeleton (Canton et al., 2005; Frisan et al., 2003; Zhang et al., 2006).
Rad26 preserves chromosome stability by at least two different mechanisms. The more minor mechanism is defined by the rad26:4A allele and might result when cells enter mitosis despite the presence of damaged microtubules. The major mechanism is defined by DNA-structure checkpoint defects that probably compromise the repair of spontaneous double-strand breaks to encourage chromosome loss (reviewed by Elledge, 1996; Paulovich et al., 1997; Lengauer et al., 1998). The existence of these two mechanisms in fission yeast predicts that multiple mechanisms of chromosome loss are also present in vertebrate DNA-checkpoint-defective cells.
Conclusion
Fission yeast use the Rad26-dependent response when microtubules are damaged in G2 and the Mad2-dependent response when microtubules are damaged in mitosis. These pathways cooperate to influence the fidelity of chromosome segregation, which suffers in an additive fashion following loss of both pathways (Fig. 6). Crosstalk between the pathways was not readily apparent, because Rad26 was not required for the Mad2-dependent response and vice versa (Figs 1 and 7). The questions of what this Rad26-dependent checkpoint pathway responds to and how it delays mitosis remain.
Materials and Methods
Strains, growth conditions and chemical stock solutions
The strains used in this study were grown under standard conditions (Moreno et al., 1991), unless noted otherwise (Table 1). Chemical reagents and stock solutions are as follows: MBC (Sigma, St Louis, MO, USA) was stored as an 8 mg/ml DMSO solution; phleomycin (Research Products International, Mt Prospect, IL, USA) as a 5 mg/ml DMSO solution; and HU (Sigma) as a 200 mM H2O solution.
Physiological methods
To perform plate assays, cultures grown to an optical density (OD) of 0.3 in YE5S liquid medium were serially diluted by a factor of 5. From each dilution, 5 μl aliquots were manually spotted onto plates using a Pipetman. Spot assays were repeated twice with similar results.
Minichromosome stability assays were performed using cells cultured in YE5S liquid medium for 40 hours to OD 0.5. Cultures were then diluted and plated in YE5S medium for 2 days at 30°C. Colonies were then replica plated to EMM minimal media lacking adenine (EMM–adenine) for 2 days at 30°C. Dark red colonies unable to grow well on these EMM–adenine plates had lost the minichromosome. Three trials were performed and 200 cells were scored per trial.
Spindle-pole-body separation was monitored in different strains containing cdc25.22 and cut12-egfp. Cells of each strain were cultured in YE5S liquid medium to OD 0.3 at 30°C, then shifted to 37°C for 3 hours 5 minutes before 16 μg/ml MBC was added. Cultures were maintained in the presence of MBC at 37°C for another 25 minutes before downshifting to ~20°C and releasing cells from the cdc25.22 block. The percentage of cells containing two Cut12-EGFP foci or a septum was determined every 20 minutes following this downshift. Three trials were performed and 200 cells were scored at every 20 minute time point.
Flow cytometry and nda3-km311 experiments were used to investigate spindle-assembly checkpoint integrity. Flow cytometry experiments were performed using cells cultured in YE5S liquid medium to OD 0.3. Cultures were then treated with MBC (8 μg/ml) and 1 ml aliquots were collected every hour for four hours and processed according to the procedure found on the Forsburg laboratory web site (www-rcf.usc.edu/~forsburg/yeast-flow-protocol.html). Briefly, cells were fixed in 70% ethanol and washed with 50 mM sodium citrate before treating with RNase and staining with Sytox Green (Molecular Probes, Eugene, OR, USA). Flow analysis was performed using a Coulter Elite Epics flow cytometer. Strains carrying the cold-sensitive nda3-km311 allele were cultured at 30°C in YE5S liquid medium to OD 0.3 before downshifting to 20°C for 5 hours. The cells were then fixed in cold methanol and stained with DAPI (Sigma) to observe chromatin and calcofluor (Sigma) to stain septa. Images were captured using a Leica DM5000 equipped with a Leica DFC350FX R2 digital camera and Leica FW4000 software. The percentage of cells containing condensed chromatin was measured using data acquired from two independent trials in which the number of cells containing condensed chromatin was determined for 200 cells.
Microscopy, L/W ratios and fluorescence intensity measurements
For DIC microscopy, cells were imaged directly from cultures grown in liquid YE5S to OD 0.3. To visualize EGFP and YFP fusion proteins, 1 ml aliquots from cultures grown in liquid YE5S to OD 0.3 were centrifuged and resuspended in cold methanol for one minute, washed twice in 100 μl SlowFade Component C (SlowFade Antifade Kit, Molecular Probes) and air dried on coverglass (Fisher). Once dried, 4.5 μl SlowFade Component A was dropped on the coverglass, which was then placed onto a slide. Achieving yeast monolayers that adhered tightly to the coverslips was crucial to observing YFP signals. To help ensure that such layers formed, the coverglass was soaked in acetone for one day, scrubbed with dishwashing soap, wiped with 70% ethanol (Sigma) and air dried prior to use. When compared with live cells expressing YFP fusions, methanol fixation did not affect the localization of YFP signals, but greatly facilitated formation of adherent monolayers (data not shown). Images were acquired using a Leica DM5000 equipped with a Leica DFC350FX R2 digital camera and Leica FW4000 software. All YFP fluorescence images were acquired with 10 second exposure times, and the contrast and brightness parameters of these images were corrected identically.
To calculate length/width (L/W) ratios, Leica software was used to acquire DIC images and measure the length and width of 100 cells. The reported L/W ratios represent the average L/W ratio calculated from two independent experiments.
Fluorescence intensity measurements were calculated using ImageJ (rsbweb.nih.gov/ij/). The data in Fig. 3B represent the relative YFP intensities derived by averaging fluorescence measurements from 50 cells of each strain.
Construction of rad26:4A, rad26-yfp, rad26:4A-yfp and rad3-gfp strains
An ~3 kb genomic PstI-BamHI fragment containing the rad26+ locus cloned into pBSK+ (pTW910) vector was mutagenized using the QuikChange site-directed mutagenesis kit (Stratagene). Briefly, four hydrophobic residues within the C-terminal motif were changed to alanines (Fig. 2A) using the mutagenic primer 5′ CCCTCAAAATGAATGCGTAGAGATTTTAGTATCTGCTGCTCGGGCTGCGTACATTTTATCTTCCGAAGATTTATCATC 3′. The mutagenized PstI-BamHI fragment containing the rad26:4A allele was then cloned into pJK148, a leu+ integration vector (Keeney and Boeke, 1994). The resulting vector (pTW919) was linearized at the leu+ marker using EcoNI and transformed into the rad26Δ leu1-32 fission yeast strain TE257. Of 20 leu+ transformants selected, 100% displayed wild-type sensitivities to the genotoxic agents HU and phleomycin, and rad26Δ sensitivity to MBC (data not shown). We arbitrarily chose one of these strains to use in the experiments presented here (TW1275), backcrosses with which showed that the phenotypes described above (sensitivity to MBC, but not HU or phleomycin) co-segregated with leu+ (data not shown).
Rad26 and Rad26:4A were C-terminally tagged with YFP using the technique of Bahler et al. (Bahler et al., 1998). Briefly, two primers (pRad26 forward: 5′ TATTTTCTCACTACAGAATTGTTGGAAGTTTGCGTCTCTCCCGAAGAGCTGGAGCAGTTGTACACTAATTTTCGGATCCCCGGGTTATTAA 3′; pRad26 reverse: 5′ GATGTGGGTGCGGGACGGGAAAGAACAACACTGAAGAAACAAGTATCATTATTTCATTTGAAAAATTAGGGAAATGAATTCGAGCTCGTTTAAAC 3′) were used to amplify a YFP-kanMX6 module (gift of Dave Kovar, University of Chicago, IL, USA) using the high-fidelity polymerase Accuzyme (Bioline, Randolph, MA, USA). The resulting PCR fragment directed integration of yfp to the 3′ end of rad26+ following yeast transformation, and both rad26+ and rad26:4A were tagged with yfp in this manner. Similarly, Rad3 was C-terminally tagged with GFP using primers (pRad3 forward: 5′ CAAGAATTGATCAAATCTGCTGTCAACCCAAAAAACCTGGTAGAAATGTACATTGGTTGGGCTGCTTATTTCCGGATCCCCGGGTTAATTAA 3′; pRad3 reverse: 5′ AATTCTTCATCGGATTAATAAATAAAATATCTTCGATTCAAATCATAAGTTTAATAATGGGTAGCTTGTTCATTGGAATTCGAGCTCGTTTAAAC 3′) to amplify the GFP (S65T)-kanMX6 module of pFA6a-GFP (Bahler et al., 1998).
Acknowledgements
We thank Dan McCollum, Dave Kovar, Hiroshi Murakami, Ian Hagan and the Japanese Yeast Genetic Resource Center National BioResource Project (YGRC-NBRP) for strains and plasmids. We also thank Zach Krych and members of the Sclafani, McIntosh and Forsburg laboratories for helpful suggestions and technical assistance. This work was supported by UCHSC and an American Cancer Society institutional research grant (57-001-47), a National Science Foundation Major Research Initiative equipment grant (4540122) and a UCCS CRCW award.