Defects in Saccharomyces cerevisiae protein phosphatase type I activate the spindle/kinetochore checkpoint

Results

The Glc7p point of execution is in G₂/M

Yeast cells containing the glc7-129 allele arrest at the nonpermissive temperature (12°C) as large-budded cells with 2N DNA content (Baker et al. 1997). However, this terminal phenotype may not reflect the cell cycle stage at which Glc7p activity is required for progression through the cell cycle. To identify the execution point of Glc7p, cells containing a single chromosomal copy of glc7-129(YAB128) were grown at the permissive temperature (30°C) to log phase and then placed on an agarose pad containing synthetic complete (SC) medium for microscopic observation. After images were collected, the microscope slide was incubated for 20 hr at the nonpermissive temperature (12°C) before images of the same cells were collected again. Representative images of cells before and after the shift are displayed in Figure . Cells were classified morphologically as unbudded, small-budded, medium-budded, large-budded with a short spindle, and large-budded with an anaphase spindle (for definition of each class, see Materials and Methods). An integratedNUF2–GFP fusion (Kahana et al. 1995) in the strain allowed us to locate the spindle pole bodies (SPBs). As shown in Figure , all cells that were in a cell cycle stage prior to anaphase at the time of the shift arrested as large-budded cells with a short spindle. However, many cells in anaphase at the time of the shift proceeded into the next cell cycle at the nonpermissive temperature. These results indicate that Glc7p function is required at or before anaphase to promote cell cycle progression.

The results of reciprocal shifts confirmed the point of execution of Glc7p. glc7-129 cells (YAB128) were arrested in S phase with hydroxyurea (HU) or in G₂/M with nocodazole at the permissive temperature (30°C). HU and nocodazole were then washed out, and individual cells were observed microscopically to confirm cell cycle arrest before shifting to the nonpermissive temperature. After ∼20 hr at the nonpermissive temperature (12°C), the terminal arrest phenotype of individual cells was quantified. All HU-treated cells (100%) arrested in the first cell cycle as large-budded cells with a short spindle (n = 19). Of nocodazole-treated cells, 42% of the cells completed mitosis, divided, and arrested in the next cell cycle (n = 24). The remaining 58% arrested as large-budded cells with a short spindle. In contrast, wild-type cells released from HU or nocodazole arrest were found to undergo at least one cell division after 20 hr at 12°C (data not shown). Two conclusions can be drawn from these results: (1)glc7-129 cells arrest in the first cell cycle after shift to the nonpermissive temperature; and (2) the execution point of glc7- 129 lies between S phase and anaphase, most likely very near anaphase.

glc7-129 cells delay in G₂/M with a short spindle

To characterize the cell cycle defect(s) of the glc7-129mutant in more detail, time-lapse microscopy was performed at the semipermissive temperature (22°C) on glc7-129 cells expressing Nuf2p–GFP. Cells were monitored by differential interference contrast (DIC) and fluorescence microscopy at 2-min intervals to monitor cell morphology and spindle dynamics, respectively. A representative montage of images from a wild-type cell and several glc7-129 cells is presented in FigureA. From our analysis of cell morphology and spindle length, as judged by the distance between the two Nuf2p–GFP spots, we formed three conclusions about the cell cycle delay inglc7-129 cells. First, glc7-129 cells retain short spindles for very long periods of time. The glc7-129 cell marked with the small black arrow in Figure A contained a clearly defined short spindle at the 16-min time point but did not proceed through anaphase until the 160-min time point. Other glc7-129cells in the montage maintained the short spindle for the duration of the experiment. In contrast, a wild-type cell, marked with a large arrow in Figure A, proceeded from a single, unseparated SPB into anaphase B within the 16-min interval of the montage. The distance between SPBs is presented graphically in Figure B for three wild-type and three glc7-129 cells. The GLC7 cells retained a short spindle (defined as 2 discernible SPBs no more than 2.5 μm apart) for 7 ± 2 min (n = 5) before proceeding into anaphase, whereas the glc7-129 cells retained a short spindle for 86 ± 34 min (n = 5). The average timeglc7-129 cells retained short spindles is likely an underestimate because many cells had a short spindle for the entire duration of the experiments. We note that the average time that cells of our wild-type maintained short spindles is shorter than that reported previously (Yeh et al. 1995; Straight et al. 1997). Because Nuf2p–GFP accumulates on the nuclear side of the spindle (Osborne et al. 1994), our estimate of spindle length may be an underestimate of the true length. Our analysis of two-dimensional instead of three-dimensional projections will also cause underestimates of spindle length. We also must propose that the short spindle in theglc7-129 cells is longer than wild type because we can observe a short spindle in these cells. Further analysis using a probe such as GFP–tubulin will resolve differences in short spindle length. These factors make it difficult to accurately determine the length of time wild-type cells have a short spindle, but the conclusion remains the same, that glc7-129 cells retain a short spindle for excessively long times.

The second conclusion we can make from our analysis is that the mitotic delay is not the result of some general defect in microtubule dynamics. As in wild-type cells, the short spindle in glc7-129 cells oscillates dynamically in both orientation and position between the 2-min time points of the time-lapse experiment. For example, the spindle in the glc7-129 cell marked with the white arrow (Fig.A) moved from the mother cell into the bud and back into the mother cell during the course of this experiment. Once anaphase begins, the rate of spindle elongation (from ∼2 to ∼6 μm) inglc7-129 cells is more rapid (0.61 ± 0.12 μm/min, n = 5) than in wild-type cells (0.18 ± 0.04 μm/min, n = 5).

The third conclusion from our time-lapse experiments is that cytokinesis is delayed in glc7-129 cells. The time between mid-anaphase (spindle length ∼6 μm) and cytokinesis was found to be 49 ± 12 min in wild-type cells (n = 5), similar to results previously published (Yeh et al. 1995). In contrast,glc7-129 cells took much longer to complete cytokinesis. Oneglc7-129 cell underwent cytokinesis 70 min after mid-anaphase. Four other cells had not undergone cytokinesis at the completion of the time-lapse data acquisition (>54, 92, 116, and 124 min after mid-anaphase). Interestingly, a defect or delay in cytokinesis is caused by another allele of GLC7 (Stark 1996).

The glc7-129-mediated G₂/M arrest requires the spindle/kinetochore checkpoint

To test the possibility that glc7-129-mediated G₂/M arrest is due to activation of a cell cycle checkpoint, the glc7-129 strain was crossed to strains containing null alleles of mad1 and bub3, andrad9 (required for the DNA-damage checkpoint) to recover the double mutant strains mad1 glc7-129 (YAB123), bub3 glc7-129 (YAB92), and rad9 glc7-129 (YAB82). Analysis of tetrads from these crosses revealed that spore clones containingglc7-129 and either mad1 or bub3 grew more slowly than spore clones containing glc7-129 alone. In contrast, rad9 glc7-129 double mutants showed no synthetic growth defect (data not shown). The mad1 glc7-129 or bub3 glc7-129 strains exhibit DNA content (Fig. A) and morphology (Fig.B), typical of cycling cells. At the nonpermissive temperature, the glc7-129 strain had extremely elevated levels of H1 kinase activity (Fig. C), indicative of cells arrested in mitosis (Surana et al. 1993). The mean kinase activity in theglc7-129 strain was ∼13× higher than wild-type kinase activity (Fig. D). In contrast, the mad1 glc7-129 strain had wild-type levels of kinase activity, consistent with the flow cytometry data. Cells were treated with nocodazole, a microtubule-depolymerizing drug known to activate the spindle/kinetochore checkpoint. The elevated levels of H1 kinase activity observed in nocodazole-treated cells was reproducibly less than that observed forglc7-129 cells (about four times higher than wild type).

Time-lapse microscopy was performed on mad1 glc7-129containing NUF2–GFP (YAB152). As shown in FigureA, the mad1 mutation completely eliminates the anaphase delay of glc7-129 at 22°C. mad1 glc7-129 maintains a short spindle for 6 ± 1 min similar to wild-type cells (n = 5). The rate of spindle elongation in the mad1 glc7-129 strain, as in the glc7-129 strain, is actually faster than that of wild-type cells (0.33 ± 0.20 μm/min, n = 5). However, the time from mid-anaphase to cytokinesis appears longer in the double mutant compared to wild type. Two cells underwent cytokinesis 48 and 54 min after mid-anaphase. Three other cells had not undergone cytokinesis at the completion of the time-lapse data acquisition (>106, 130, and 166 min after mid-anaphase).

glc7-129 cells require a functional spindle/kinetochore checkpoint for accurate chromosome segregation and viability

To test the possibility that the slower growth rate of mad1 glc7-129 strains is due to increased rates of chromosome loss and reduced viability, we assayed the percentages of viable cells inglc7-129 and mad1 glc7-129 strains at the permissive temperature and after 12 and 24 hr at the nonpermissive temperature. As shown in Table , only 79% of the mad1 glc7-129 and 68% of the bub3 glc7-129 cells were viable at the permissive temperature. In contrast, the glc7-129strain maintained high viability even after 24 hr at the nonpermissive temperature (85%). After 24 hr at the nonpermissive temperature, the viability of the mad1 glc7-129 and bub3 glc7-129strains was greatly reduced (17% and 22%, respectively). These strains also exhibited a rate of chromosome loss 2–3 orders of magnitude higher than in wild-type cells (Table ). In contrast,glc7-129 and mad1 strains displayed wild-type rates of chromosome loss at 30°C and rates approximately 10-fold above wild-type at 22°C. Together, these data indicate thatglc7-129 cells that lack the spindle/kinetochore checkpoint exhibit a high rate of cell death due to chromosome instability.

Discussion

We show here that cells containing the recessive cold-sensitive allele glc7-129 delay in mitosis because of activation of the spindle/kinetochore checkpoint. Sassoon et al. (1999)have likewise found that the temperature-sensitive alleleglc7-10 activates the checkpoint. The fact that loss of PP1 function in other organisms also causes mitotic arrest suggests that the role of PP1 and, thus, the consequences of its deficiency, might be highly conserved. Precedence for the conservation of PP1 function is found in glycogen metabolism where PP1 holoenzymes, composed of the PP1 catalytic subunit and a glycogen-targeting subunit, is thought to dephosphorylate glycogen synthase in both yeast (Stark 1996) and mammals (Hubbard and Cohen 1993).

The specificity of PP1 is thought to be regulated by subunits that target PP1 to its substrate and/or regulate its activity (Hubbard and Cohen 1993). Studies in yeast indicate that the physiological defects of a number of glc7 mutants are caused by the failure of the variant Glc7 proteins to associate with a specific targeting subunit rather than to defects in catalytic activity (Stuart et al. 1994; Tu and Carlson 1995). The allele described in this work, glc7-129, contains two point mutations altering residues that lie on the outer surface of the protein (D137A, E138A) (Baker et al. 1997). glc7-129 mutants have defects in glycogen biosynthesis, sporulation, as well as in mitosis but have a normal glucose response (Baker et al. 1997). Consistent with its glycogen biosynthesis defect, Glc7-129p does fail to bind Gac1p (D.L. Frederick and K. Tatchell, in prep.). It is, therefore, possible that the defect responsible for the mitotic arrest of glc7-129 is caused by a specific defect in its interaction with a mitosis-specific regulatory subunit rather than a general defect in catalytic activity. However, the nature of the Glc7p holoenzyme required for the mitotic role is unknown to us as yet.

Assuming that the primary event monitored by the spindle/kinetochore checkpoint is the attachment of kinetochores to the spindle, we envision two potential roles for PP1 in this process. First, PP1 could dephosphorylate components of the mitotic spindle and the failure to do so could cause changes in spindle dynamics that activate the checkpoint. Supporting this idea is the observation that introduction of PP1-specific phosphatase inhibitors into mammalian cells causes defects in spindle dynamics (Tournebize et al. 1997). However, our results do not support a change in spindle dynamics as a cause of the mitotic defect in glc7-129 cells. We detected no obvious defects in microtubule morphology inglc7-129 cells. Also, unlike mutants known to affect microtubule dynamics, we have observed that glc7-129 cells are not hypersensitive to the microtubule-depolymerizing drug benomyl. Finally, the dynamically oscillating short spindle and apparently normal anaphase in glc7-129 cells argue against a general role in spindle dynamics.

The second possibility is that Glc7p must dephosphorylate kinetochore proteins to allow spindle attachment. The phosphorylation state of kinetochore proteins may affect their binding to CEN DNA and/or their competence to bind microtubules. Two essential components of the S. cerevisiae kinetochore, Cbf2p/Ndc10p and Ctf13p, have been shown to be phosphorylated in vitro or as recombinant proteins in insect cells, respectively (Jiang et al. 1995; Kaplan et al. 1997). Also, treatment of yeast kinetochore proteins with phosphatase in vitro prevents binding to centromere sequences (Lechner and Carbon 1991). Therefore, the phosphorylation state of kinetochore complexes may affectCEN DNA binding. Using the transcription readthrough assay (Perier and Carbon 1992), an in vivo assay that presumably measures kinetochore protein complex formation and/or structure onCEN DNA, we did not observe significant differences in kinetochore/centromere complex formation and/or structure in wild-type and glc7-129strains (data not shown). This corroborates the results of Sassoon et al. (1999), who found that centromere binding activity is normal in extracts prepared from glc7-10 strains. Thus, defects in Glc7p do not appear to alter kinetochore complex formation and structure.

If glc7 mutants that activate the spindle/kinetochore checkpoint have normal kinetochore/CEN DNA complex formation and structure but lose chromosomes at a high rate in the absence of the checkpoint, then the most likely defect is the binding of kinetochores to microtubules. This scenario is consistent with the point-of-execution experiments placing Glc7p function at G₂/M just prior to anaphase and is directly supported by the results of Sassoon et al. (1999) that extracts prepared from glc7-10 cells are deficient in microtubule binding and that addition of recombinant Cbf2p/Ndc10p can restore this activity. It remains to be seen if Cbf2p/Ndc10p is a substrate of Glc7p in vivo.

Materials and methods

Yeast strains, media, and general methods

The yeast strains, with the exception of KT81 (MAT a hom3) and KT82 (matα hom3 ilv1), are congenic to strain KT1112 (MATα ura-52 leu2 his3) (Stuart et al. 1994). To integrate the glc7-129 allele into the yeast genome at the GLC7 locus, aBglII–KpnI fragment from pNC160–glc7-129(Baker et al. 1997) was inserted into pRS306 (Sikorski and Hieter 1989) at the BamHI and KpnI sites, which results in an amino-terminal truncation of Glc7-129p. The resulting plasmid was linearized within glc7-129 with SalI and transformed into diploid strain KT1112/1113. Ura⁺transformants were sporulated, and MAT a andMATα Ura⁺ haploids were recovered. The Ura⁺ haploids arrested in mitosis at 12°C and failed to accumulate glycogen, two traits observed previously from the plasmid-borne glc7-129mutant. The URA3 locus in these strains showed tight genetic linkage to the GLC7 locus. The cold-sensitive Ura⁺ haploids were then grown on 5-fluoro-orotic acid (5-FOA) media to select for cells that had lost the URA3 gene through recombination. A resulting Ura⁻, cold-sensitive 5-FOA revertant was designated strain YAB27. The NUF2–GFP gene fusion on plasmid pJK67 (Kahana et al. 1995) was linearized with BstE1 and integrated into the chromosomal NUF2 locus of strains KT1112, YAB27, and YAB123. Loss of viability was determined as described (Wang and Burke 1995). Inviability was scored as the inability of cells to undergo more than three divisions (i.e., ≤eight cells) within 24 hr after plating cells at the permissive temperature. At least 300 cells were scored for each strain.

Point-of-execution experiments

Yeast strain YAB128 was grown to log phase at 30°C, and cells were placed on a slide with a thin 2% agarose pad in SC medium. DIC and fluorescent images were captured from several fields of cells. Individual cells were classified as unbudded, small bud (bud size <0.3× mother cell diameter), medium bud (bud size 0.3×–0.6× mother cell diameter), large bud (bud size >0.6× mother cell diameter) with short spindle (<2 μm), and large bud (bud size >0.6× mother cell diameter) with anaphase spindle (>2 μm). The slide was then incubated at 12°C for ∼20 hr and images of the same fields were captured again.

Logarithmically growing YAB128 cells at 30°C were arrested in S phase or G₂/M by incubating for 3 hr in YPD containing 0.1m HU or 15 μg/ml nocodazole, respectively, HU or nocodazole was then washed out, images of cells were captured immediately, and cells were then shifted to 12°C for ∼20 hr. Cells were scored for their terminal phenotype as above. As a control in the preceding experiments, wild-type GLC7 cells, upon release from arrest, were found to undergo at least one cell division after 20 hr at 12°C.

Analysis of chromosome III mis-segregation

MAT a/α diploids can become mating competent through either the loss of one copy of chromosome III, resulting in hemizygosity, or mitotic recombination, resulting in homozygosity for the MAT locus. To quantitate the mating efficiency of MAT a/α diploids, single colonies from each strain were inoculated into 5 ml of liquid YPD and grown to early log phase (∼1 × 10⁷cells/ml) at 30°C or 22°C. Cells (5 × 10⁶) from each strain were then mixed with 2 × 10⁷ logarithmically growing haploid cells KT81 (MAT a) and KT82 (MATα), filtered onto a 25-mm × 0.45 μm filter, and incubated on a YPD plate at 30°C for 3.5 hr. The cells were then washed off the filters and plated onto minimal media to select for triploids. After 2 days, colonies were counted and used to calculate the rate of chromosome III mis-segregation according to the following equation: No. of colonies from MAT a mating + no. of colonies fromMATα mating/2 divided by 5 × 10⁶.

Microscopic analysis

Images were collected through an Olympus UPlanFl 100×/1.3 NA objective with a Photometrics CH250/A cooled CCD camera (KAF1400 Kodak chip) using Scanalytics IPLab Spectrum (Fairfax, VA) software. To collect time-lapse images of cells containing NUF2–GFP, cells were grown to mid-log phase in YPD, harvested by centrifugation, and placed on microscope slides on a pad of 2% agarose in SC medium as described (Waddle et al. 1996). Images of living cells were collected using a DAGE VE1000 SIT camera and DSP2000 signal processor. At each time interval, a two-dimensional projection of eight fluorescent images in different Z-axis planes was acquired using a Chroma (Brattleboro, VT) 41001 filter set. A separate DIC image was captured from a central Z-axis plane. A custom script written for IPLab Spectrum software was used to acquire both fluorescence and DIC images. For time-lapse microscopy of wild-type and glc7-129 strains shown in Figure, the wild-type cells were distinguished from the glc7-129cells by briefly staining the wild-type cells with the chitin-specific stain Calcofluor at 1 μg/ml and then imaging the field of cells with UV light using a DAPI filter cube to distinguish wild type from glc7-129. Neither the Calcofluor staining or the brief UV light treatment altered growth rates.

Protein kinase assay

Whole-cell extracts were made as described by Stuart et al. (1994) and protein concentrations determined by the Bradford assay (Bradford 1976) using BSA as a standard. The protein kinase assays were performed essentially as described by Hisamoto et al. (1994) except that 100 μg of total protein from each of the strains was incubated with 5 μg of suc1–Sepharose (Upstate New York Biotechnology) that was diluted to a volume of 100 μl in lysis buffer (50 mm Tris-HCl at pH 7.5, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 1 mm sodium PP_i) containing protease inhibitors (0.5 mm PMSF and 1 μg each of chymostatin, leupeptin, antipain, and pepstatin A/1 ml).

Flow cytometry

Cells were processed for flow cytometry as described (Baker et al. 1997). Quantitation of morphology of the cells used for flow cytometry was performed using DIC microscopy. The position of the nucleus was inferred from the location of DNA stained with propidium iodide. For each strain, n > 300.