Report

The Centromeric Protein Sgo1 Is Required to Sense Lack of Tension on Mitotic Chromosomes

See allHide authors and affiliations

Science  07 Jan 2005:
Vol. 307, Issue 5706, pp. 130-133
DOI: 10.1126/science.1101366

Abstract

Chromosome alignment on the mitotic spindle is monitored by the spindle checkpoint. We identify Sgo1, a protein involved in meiotic chromosome cohesion, as a spindle checkpoint component. Budding yeast cells with mutations in SGO1 respond normally to microtubule depolymerization but not to lack of tension at the kinetochore, and they have difficulty attaching sister chromatids to opposite poles of the spindle. Sgo1 is thus required for sensing tension between sister chromatids during mitosis, and its degradation when they separate may prevent cell cycle arrest and chromosome loss in anaphase, a time when sister chromatids are no longer under tension.

Errors in chromosome segregation lead to disease and death. To prevent such errors, the protein complex cohesin (1) holds replicated chromosomes together, and this linkage is not broken until every pair of sister chromatids is bi-oriented on the mitotic spindle, with the two sisters attached to microtubules that emanate from opposite poles of the spindle (2). Bi-orientation generates tension on the chromosomes because the links between the sister chromatids resist the pulling forces of the spindle. Chromosome orientation is monitored by the spindle checkpoint, which detects unattached kinetochores (3) or the lack of tension between sister chromatids (4, 5). Either lesion inhibits the signal that induces chromosome segregation. Several components of the spindle checkpoint (Mad2, Mad3, and Bub3) form a complex that prevents entry into anaphase by binding to Cdc20, an essential activator of the anaphase-promoting complex (APC) (6). The APC triggers the destruction of securin (Pds1), the protein that inhibits separase (Esp1), which is the protease that triggers sister separation by cleaving cohesin (1) (Fig. 1A). When the spindle checkpoint inhibits the APC, Pds1 is stable, separase is inhibited, and the sisters remain unseparated.

Fig. 1.

Isolation and characterization of the spindle checkpoint component Sgo1. (A) LMCs separate in part because of their inability to withstand spindle forces. As a result, their linkage to microtubules is no longer under tension, and this activates the spindle checkpoint (stars) (30). In a CDC28-VF strain with LMCs, the APC is kept inactive and the cells arrest in metaphase. Abolishing the spindle checkpoint by, for example, deleting mad2 (mad2Δ) alleviates this arrest, as does the expression of dominant Cdc20 that cannot be inhibited by the spindle checkpoint complex. After ethyl methanesulfonate mutagenesis of CDC28-VF cells with a tetracycline-repressible dominant Cdc20, spindle checkpoint mutants were recovered on the basis of their ability to tolerate LMCs in the absence of dominant Cdc20. Two alleles of sgo1 were isolated, sgo1-100 and sgo1-700, and their growth (fourfold serial dilutions) in the presence or absence of dominant Cdc20 is shown. All strains shown are in the CDC28-VF background. (B to D) A strain with epitope-tagged Pds1 (Pds1-18Myc) and Sgo1 (Sgo1-13Myc) was released from a G1 (α-factor, αF) block; at the indicated times, samples were taken for Western blot analysis against the Myc epitopes. (B) Sgo1 and Pds1 expression during the normal cell cycle in rich medium with glucose (YPD) at 30°C, and (C) after release into benomyl (30 μg/ml) and nocodazole (30 μg/ml) at 23°C; (D) Sgo1 and Pds1 expression after release from G1 to galactose at 30°C in the presence or absence of nondegradable, destruction box–deleted Pds1 (Pds1-db Δ) under control of the GAL1 promoter. In both cases the APC is activated, as evidenced by the destruction of the endogenous Pds1-18Myc. However, in the presence of Pds1-dbΔ the strains remained arrested as large budded cells, whereas in its absence the cells rebudded after about 120 min (22).

Little is known about how cells sense the absence of tension between a pair of sister chromatids and send this information to the spindle checkpoint. Ipl1, the budding yeast member of the Aurora family of mitotic protein kinases (7), is involved in this process (8) and is also needed to detach microtubules from kinetochores that are not under tension (911). The only role of Ipl1 may be to induce chromosomes to detach from microtubules, thus creating naked kinetochores, which inhibit the APC by recruiting other checkpoint components. Alternatively, the lack of tension at the kinetochore may signal directly to the checkpoint without requiring microtubule detachment (8, 12, 13).

To identify components of the tension-sensing machinery, we looked for mutants that ignored chromosomes that were not under tension. The screen used a budding yeast (Saccharomyces cerevisiae) strain harboring linear minichromosomes (LMCs) that segregate poorly and activate the spindle checkpoint (14). In wild-type budding yeast, lack of tension delays APC activation, but cells eventually enter anaphase (5, 8). This delay becomes a lethal arrest in cells that contain CDC28-VF, a mutation in Cdc28 (the budding yeast homolog of the cyclin dependent kinase Cdk1) that reduces APC activity (15). This arrest is dependent on the spindle checkpoint and can be overcome by expressing CDC20-127, a dominant Cdc20 allele that is refractory to Mad2 inhibition and therefore overrides the checkpoint (16) (Fig. 1A). We mutagenized the CDC28-VF strain carrying the LMCs and selected mutants that grow in the absence of the checkpoint-resistant Cdc20 and therefore bypass the minichromosome-induced mitotic arrest (Fig. 1A). Two of the mutants lie in SGO1, the fungal homolog of Drosophila MEI-S332, a gene whose product protects centromeric cohesion during the first meiotic division (1720). Each allele carries a mutation in the most conserved region of the protein (Thr379 → Ile in sgo1-100 and Pro390 → His in sgo1-700); the sgo1-700 allele also has a second substitution (Asp519 → Asn) in a region that is strongly conserved among fungi.

Sgo1 is cell cycle regulated. Its expression increased at the G1-S transition and decreased during mitosis in a pattern similar to that of Pds1 (Fig. 1B). Like Pds1, Sgo1 was stabilized in the presence of benomyl and nocodazole, drugs that destabilize microtubules (Fig. 1C). Sgo1 could be a substrate of either the APC or separase. Because Sgo1 was still degraded in cells that expressed a nondegradable version of Pds1 that inhibits separase but has no effect on the APC (21) (Fig. 1D), Sgo1 is likely a direct target of the APC. Four potential APC recognition motifs (D-boxes) in Sgo1 further support this possibility.

Our selection was designed to identify components that signal and/or sense tension at the kinetochores. To examine the response to chromosomes that lack tension, we placed sgo1 mutants in cells that also express Mcd1 (also known as Scc1), a component of cohesin (1), under the glucose-repressible GAL1 promoter. In the presence of glucose, Mcd1 is not expressed and the sister chromatids are not linked. The positioning of the sisterless chromosomes on the spindle indicates that they attach to the spindle (5), although we cannot prove that every kinetochore is attached to a microtubule at all times. In the absence of Mcd1, reduced tension at the kinetochores activates the spindle checkpoint and delays APC activation (5, 8), as indicated by stabilization of the APC target Pds1 (Fig. 2A). The sgo1 mutants, however, failed to delay APC activation, hence they are unable to sense or respond to lack of tension.

Fig. 2.

Sgo1 is involved in tension sensing and chromosome segregation. (A) The indicated GAL1-MCD1 strains, all with epitope-tagged Pds1 (Pds1-18Myc), were released from α-factor in rich medium with glucose (Mcd1 not expressed) or galactose (Mcd1 expressed) at 30°C. At the indicated times, samples were taken for Western blot analysis against the Myc epitopes. (B) The indicated strains underwent fourfold serial dilution and were spotted on plates with or without benomyl (10 μg/ml). (C) The viability (defined as the ability to give rise to colonies on rich medium) of the indicated strains was measured at the indicated times after they were released from G1 (an α-factor block) into medium containing benomyl (10 μg/ml) and nocodazole (15 μg/ml) at 23°C. Data are from three experiments; error bars represent SDs. (D) The indicated strains all contained epitope-tagged Pds1 (Pds1-18Myc) and were released from α-factor in media containing benomyl (30 μg/ml) and nocodazole (30 μg/ml) at 23°C. Samples were taken at the indicated times for Western blot analysis against the Myc epitopes.

We also examined cells lacking Sgo1 (sgo1Δ). In the W303 strain background, sgo1Δ cells showed poor viability and acquired suppressor mutations (22). Reduced viability was largely overcome by slowing down DNA replication, an approach previously used to reduce the sensitivity of spindle checkpoint mutants to benomyl (23). The exact timing of Pds1 degradation was somewhat variable in sgo1Δ cells, but within an experiment the timing of destruction was similar regardless of whether Mcd1 was expressed (Fig. 2A); this result implies that the sgo1Δ cells cannot sense the lack of tension. The sgo1 point mutants also failed to delay APC activation in cells lacking Cdc6 expression (22). Cdc6 is required for DNA replication, and in its absence chromosomes enter mitosis without the sisters needed to generate tension at the kinetochore (5, 8).

The sgo1 mutants shared properties with other spindle checkpoint mutants such as mad2Δ. The sgo1 mutants showed chromosome loss rates (24) approximately equal to those of mad2Δ cells, five times the rate for wild-type cells, and one-third the rates for bub1Δ and bub3Δ strains (table S1) (25). Like mad2Δ, the growth of the sgo1 mutants was sensitive to benomyl (Fig. 2B). To examine their response to microtubule depolymerization, we arrested cells in G1 by exposure to mating pheromone (α-factor) and then released them into medium containing benomyl and nocodazole. The viability of sgo1 mutants dropped markedly after microtubule depolymerization (Fig. 2C) even though the sgo1 mutants, like wild-type cells, arrested with high levels of Pds1 for at least 4 hours (Fig. 2D). In contrast, mad2Δ cells failed to detect the damaged spindle and degraded Pds1 about 2 hours after their release from G1 arrest (Fig. 2D). Therefore, the sgo1 mutants, unlike mad2Δ, are still able to arrest in mitosis in response to unattached kinetochores. Although we cannot exclude the possibility that weak defects in the spindle checkpoint allow cells to respond to unattached kinetochores, but not to those that are not under tension, the observation that sgo1 mutants that have different effects on viability are all inactivate to tension sensing without altering the ability to detect unattached kinetochores argues against this possibility.

The behavior of sgo1 mutants seems paradoxical; they die after microtubule depolymerization even though they arrest. Because Sgo1 protects centromeric cohesion during meiosis (1719), one explanation is that the sgo1 mutants have mitotic cohesion defects. We introduced the sgo1 alleles into strains that have chromosome IV labeled with 256 tandem repeats of the Lac operator inserted at the TRP1 locus (12 kb from the centromere). This array is seen as a fluorescent dot when cells express a green fluorescent protein–Lac repressor fusion protein (GFP-LacI) (26). As controls, we used wild-type, mad2Δ, and GAL1-MCD1 strains in the same background. The mad2Δ cells lack all known aspects of the spindle checkpoint, and GAL1-MCD1 cells lack cohesin when grown in glucose. The strains were released from G1 into medium that contained glucose, benomyl, and nocodazole. In wild-type cells, the absence of kinetochore-microtubule attachment activates the spindle checkpoint, stabilizing cohesion. As a result, sister chromatids stay together and the GFP-LacI dots corresponding to the two sisters of chromosome IV cannot be resolved, appearing as a single dot under the fluorescent microscope. Roughly half the cells lacking Mad2 or cohesin had two green dots, showing that they had prematurely separated their sisters (Fig. 3A). The sgo1 mutants behaved much more like wild-type cells and predominantly arrested as large-budded cells with a single visible dot.

Fig. 3.

The sgo1 alleles do not have a strong cohesion defect in mitosis but fail to bi-orient their chromatids after spindle reformation. (A) The indicated strains, all with GFP dots labeling the TRP1 locus on chromosome IV, were released from α-factor block in glucose, benomyl (30 μg/ml), and nocodazole (30 μg/ml) at 23°C for 3 hours. Percentages of cells with two dots at the time of release from α-factor arrest (white bars) and at the end of 3 hours (black bars) are shown. At least 100 cells were counted at each time point of each experiment; results from three independent experiments were combined. Error bars show SDs. (B) Wild-type, mad2Δ, and sgo1-100 strains with (black bars) or without (white bars) GAL1-MPS1 were released from an α-factor block into galactose (Gal), benomyl (30 μg/ml), and nocodazole (Noc) (30 μg/ml) for 5 hours at 23°C. The samples were then treated with galactose alone for 1.5 hours (experimental) or with glucose (Glc), benomyl, and nocodazole for 0.5 hours (control). Both sets were then released into glucose and α-factor at 23°C for 2 hours. Percentages of viable cells and cells with two green fluorescent dots were determined. Data from four independent experiments are shown; error bars show SDs.

These results show that sgo1 mutants do not have a major mitotic cohesion defect. Their rapid death in the absence of microtubules could be explained in two ways. One is that the sgo1 mutants have a minor cohesion defect that is small enough to differentiate them from cells lacking cohesin, yet ensures that at least one chromosome missegregates in most benomyl- and nocodazole-treated cells. The other is that cohesion is normal, but sgo1 mutants make errors in chromosome alignment as the spindle reforms after benomyl and nocodazole have been removed. We distinguished these hypotheses by delaying sister chromatid segregation until well after benomyl and nocodazole had been removed, a treatment that has no effect on chromosomes that have already separated but allows more time for linked pairs of sisters to align correctly on the spindle.

We compared sgo1-100, wild-type, and mad2Δ cells, all containing the GFP-marked chromosome IV. Two versions of each strain were made, differing only in the presence of an extra copy of the MPS1 gene driven by the GAL1 promoter. Mps1 is part of the spindle checkpoint, and its overexpression causes a spindle checkpoint and APC-dependent metaphase arrest (27). The APC and Mps1 mutually oppose each other because the APC induces the destruction of Mps1 (28). Hence, the metaphase arrest induced by overexpressing Mps1 can be rapidly reversed by adding glucose. All the strains were released from G1 into medium that contained galactose, benomyl, and nocodazole. Regardless of whether they overexpressed Mps1, wild-type and sgo1-100 cells arrested in metaphase, whereas mad2Δ cells did not and separated their sisters prematurely (22). The cells were then transferred to medium that contained galactose but lacked benomyl and nocodazole for 1.5 hours to allow the spindle to reassemble well before chromosome segregation (Fig. 3B). In this experiment, overexpression of Mps1 increased the viability of sgo1-100 cells but had no effect on mad2Δ cells whose sisters had already separated. To confirm that rescue depended on delaying anaphase until the spindle had reformed, we transferred sgo1-100 cells that had been treated with galactose, benomyl, and nocodazole into medium containing glucose, benomyl, and nocodazole for 0.5 hours, allowing Mps1 levels to fall before washing out the microtubule-destabilizing drugs. These cells died, indicating that Mps1 overexpression was only effective when it continued after the benomyl and nocodazole had been removed (Fig. 3B). We conclude that although sgo1 mutants arrest normally when they lack a spindle, they die because they make but cannot detect errors in chromosome alignment as the spindle reforms.

We monitored chromosome segregation as cells recovered from microtubule depolymerization. Cells that had overexpressed Mps1 were released from their final mitotic block into α-factor, which prevents their escape from G1, and numbers of GFP dots (a proxy for the number of copies of chromosome IV) per cell were counted. Bi-orientation of a chromosome leads to each daughter cell inheriting one sister chromatid and thus having a single GFP dot. Failure to bi-orient the sister chromatids leaves one daughter with two GFP dots and its sister with none. As expected from the rescue of viability, allowing Mps1 overexpression as the spindle reformed substantially reduced the fraction of sgo1-100 cells that had two dots in the following G1 (Fig. 3B). These results argue that sgo1-100 cells cannot properly align their sister chromatids on the spindle but do successfully hold them together.

We detected two new defects in sgo1 mutants: an inability to arrest the cell cycle in cells whose chromosomes were not under tension, and a defect in chromosome segregation that could be rescued by delaying the onset of anaphase. Vertebrate homologs of Sgo1 contain a strong microtubule-binding domain (29), and Sgo1 is found at the kinetochore (18). Both observations argue that Sgo1 is directly involved in interactions between chromosomes and microtubules, and raise the possibility that it is the tension sensor. For example, Sgo1 could be located in a region of the kinetochore that could only be reached by microtubules that were not under tension (Fig. 4). As long as it is bound to a microtubule, Sgo1 would send a signal that destabilizes the attachment of the kinetochore to its microtubule and might also send a signal directly to the spindle checkpoint. Sister centromeres separate from each other as anaphase begins. Once chromosomes have left their sisters, arresting the cell cycle would be pointless and the destabilization of microtubule-kinetochore attachment could have disastrous consequences for chromosome segregation. The degradation of Sgo1 as anaphase begins and the resulting ablation of the tension-signaling pathway may explain why sister chromatids remain stably attached to the spindle once they reach its poles.

Fig. 4.

A model of Sgo1 function in sensing and/or signaling tension and promoting bi-orientation of mitotic chromosomes (blue). Spindle microtubules (light gray) regulate Sgo1 activity, which switches between an active (red) and inactive (green) state. In the absence of tension, Sgo1 is in contact with spindle microtubules and sends signals to promote bi-orientation and delay progression into anaphase. The dashed arrow represents uncertainty about whether the lack of tension can inhibit APC activity without generating unattached kinetochores. A separate mechanism (dark gray) may monitor attachment of kinetochores to spindle microtubules.

Supporting Online Material

www.sciencemag.org/cgi/content/full/307/5706/130/DC1

Materials and Methods

Tables S1 and S2

References and Notes

View Abstract

Navigate This Article