Reduced Mad2 expression increases chromosome loss
Mutations that inactivate the spindle checkpoint cause increased chromosome loss. We measured the chromosome loss frequency of MAD2/MAD2
Δ, and mad2Δ/mad2
Δ diploids using a colony color assay (Hieter et al., 1985
Spencer et al., 1990
). The strains carry a nonessential chromosome fragment carrying an ochre
-suppressing tRNA gene (SUP11
). The host strains carry the ade2-101 ochre
mutation; loss of the chromosome fragment leads to accumulation of a red pigment. Cells that lose the chromosome fragment during their initial division produce a sectored colony with one red and one white half.
We examined the effect of two variables on the rate of chromosome loss: the level of Mad2 and the presence or absence of a low dose (5 μg/ml) of benomyl, an inhibitor of microtubule polymerization ( and
). In the absence of benomyl, the rate of chromosome loss was significantly increased in both the MAD2/mad2Δ and mad2Δ/mad2Δ diploids relative to the MAD2/MAD2 control (p < 0.0001, χ2 test;
). The addition of benomyl led to a statistically significant increase in the chromosome loss rate of MAD2/MAD2 (p = 0.001) and mad2Δ/mad2Δ (p < 0.0001) cells. However, in MAD2/mad2Δ cells, adding benomyl did not significantly increase the chromosome loss rate (p = 0.2). Because microtubule depolymerization creates kinetochores that are not bound to microtubules, this result suggests that MAD2/mad2Δ cells respond normally to unattached kinetochores.
FIGURE 1: Chromosome loss rates in MAD2/MAD2, MAD2/mad2Δ, and mad2Δ/mad2Δ diploids. Diploid cells contained a genetically marked chromosome fragment whose loss rates was measured using a colony color-sectoring assay. Measurements were made (more ...)
Comparing chromosome loss rates by the χ2 test. p values are shown.
MAD2/mad2Δ heterozygotes respond to microtubule depolymerization
To directly determine whether MAD2/mad2
Δ diploids respond to microtubule depolymerization, we compared the ability of benomyl to arrest MAD2/MAD2
Δ cells in mitosis. We arrested cells in S phase, released them into different concentrations of benomyl, and measured their passage through mitosis by detecting the bud they formed as they entered the subsequent cell cycle. To ensure that any delay we measured occurred at the metaphase–anaphase transition rather than at the transition between anaphase and G1 (mitotic exit), we performed these experiments in diploid bub2Δ/bub2
Δ mutant cells, which lack the spindle positioning checkpoint (Li, 1999
Bardin et al., 2000
Bloecher et al., 2000
Pereira et al., 2000
). MAD2/MAD2 bub2Δ/bub2
Δ diploids rebud after release from S-phase arrest, with the fraction of rebudded cells peaking at 135 min (). Cells heterozygous for MAD2
consistently progress through the cell cycle even faster, with the fraction of rebudded cells peaking at 120 min. This faster cell cycle progression is reminiscent of the observed acceleration of anaphase in haploid mad2
Δ cells compared with their MAD2
counterparts (Li, 1999
Shonn et al., 2000
). A low benomyl level (5 μg/ml) makes cells rebud more slowly, and the kinetics of rebudding are identical for MAD2/MAD2
Δ cells. A higher benomyl level (40 μg/ml) severely delays rebudding in both cell types. These results show that MAD2/mad2
Δ cells respond normally to kinetochores that are not attached to microtubules.
FIGURE 2: MAD2/mad2Δ cells have a mitotic delay in response to microtubule depolymerization. (A) Exponentially dividing MAD2/MAD2 bub2Δ/bub2Δ cells and MAD2/mad2Δ bub2Δ/bub2Δ cells were arrested in S phase with hydroxyurea. (more ...)
We also performed a microcolony assay, which monitors the ability of cells to divide under conditions that prevent spindle formation. MAD2/MAD2, MAD2/mad2Δ, and mad2Δ/mad2Δ cells (all bub2Δ/bub2Δ) were placed on rich media containing 60 μg/ml benomyl, which completely depolymerizes microtubules (). For all three strains, the average number of cells per microcolony increased in the first 3 h because cells that had completed mitosis but not yet budded were able to bud, replicate their DNA, and arrest at the next mitosis. After 3 h, mad2Δ/mad2Δ cells kept dividing, but MAD2/MAD2 and MAD2/mad2Δ cells did not, confirming that MAD2/mad2Δ cells respond normally to kinetochores that are not attached to microtubules.
MAD2/mad2Δ heterozygotes fail to respond to short linear chromosomes
The behavior of MAD2/mad2Δ cells puzzled us: they respond normally to microtubule depolymerization, but they lose chromosomes frequently. One explanation is that they respond to chromosomes that are not attached to microtubules but cannot respond to chromosomes that are attached to microtubules but are not under tension.
To test this possibility, we examined the response of MAD2/mad2
Δ cells to chromosomes that are not under tension. Short linear chromosomes activate the spindle checkpoint because their sister chromatids separate prematurely; once this happens, they lose tension at their kinetochores (Wells and Murray, 1996
). In the presence of the CDC28-VF
mutation, which slows the exit from mitosis (Rudner et al., 2000
), this mitotic delay is strong enough to block cell proliferation. To allow these strains to proliferate, we expressed CDC20-127
, a dominant allele of CDC20
that the spindle checkpoint cannot inhibit (Hwang et al., 1998
). In strains where this allele is expressed from a tetracycline-regulated promoter (Ptet
), the response to the spindle checkpoint can be manipulated (). We used diploid strains that contain short linear chromosomes and are homozygous for Ptet-CDC20-127
. When these strains are grown in the absence of doxycycline the checkpoint is turned off: CDC20-127
is expressed, the APC is resistant to signals from the checkpoint, and cells proliferate normally. Adding doxycycline represses CDC20-127
and allows the short linear chromosomes to activate the checkpoint by inhibiting the endogenous Cdc20 and arresting cells in mitosis, thus preventing cell proliferation.
FIGURE 3: MAD2/mad2Δ cells do not arrest in response to short linear chromosomes. (A) A diagram of the effects of short linear chromosomes on the spindle checkpoint. CDC20-127 is a dominant allele of CDC20 that the spindle checkpoint cannot inhibit; in (more ...)
We compared the ability of short linear chromosomes to arrest MAD2/MAD2 and MAD2/mad2Δ cells. Strains were spotted on plates in the absence (checkpoint off) or presence of 10 μg/ml of doxycycline (checkpoint on). MAD2/MAD2 cells arrested on doxycycline plates, but MAD2/mad2Δ cells proliferated (). Among the MAD genes, the effect of heterozygosity is specific for MAD2: MAD1/mad1Δ and MAD3/mad3Δ strains did not proliferate on doxycycline plates.
We monitored the effect of short linear chromosomes on the passage through a single cell cycle. We arrested cells in S phase, released them from the arrest, and followed their progress through mitosis and into the next cell cycle, as detected by rebudding. In the absence of doxycycline, the rebudding of MAD2/MAD2 cells peaked at 120 min, but the presence of doxycycline strongly delayed rebudding (). In contrast, the rebudding of MAD2/mad2Δ was only slightly reduced by the presence of doxycycline, showing that halving the dose of MAD2 greatly reduces the ability of kinetochores that are not under tension to activate the spindle checkpoint.
The loss of sister chromatid cohesion does not arrest MAD2/mad2Δ heterozygotes
We used a second method to remove tension from the linkage between kinetochores and microtubules. Abolishing the linkage between chromosomes reduces tension at their kinetochores (Li and Nicklas, 1995
), and this manipulation activates the spindle checkpoint in meiotic (Shonn et al., 2000
) and mitotic (Biggins and Murray, 2001
Stern and Murray, 2001
King et al., 2007
) yeast cells. To remove tension from all chromosomes, we regulated the expression of the cohesin subunit Mcd1 (also known as Scc1) by fusing its gene to the galactose-inducible GAL1
). We measured cell cycle progression by following the disappearance of epitope-tagged securin (Pds1–18xMyc), which marks the activation of the APC and normally leads to sister chromatid separation. When Mcd1 was expressed, the sister chromatids of MAD2/MAD2
cells were held together by cohesin until anaphase, and the cells degraded Pds1 promptly (). When it was not, there was no cohesion, and a delay in the onset of Pds1 destruction was observed. In contrast, both MAD2/mad2
Δ and mad2Δ/mad2
Δ showed little or no response to a loss of tension on mitotic chromosomes: both cells initiated Pds1 destruction at the same time as cells that had normal cohesin. Our experiments show that the MAD2
heterozygotes have different responses to the two signals that activate the spindle checkpoint: MAD2/mad2
Δ cells respond normally to an antimicrotubule drug, but they are defective in their ability to respond to chromosomes that are not under tension.
FIGURE 4: MAD2/mad2Δ heterozygotes do not arrest in response to a loss of tension on all mitotic chromosomes. We measured cell cycle progression by Western blotting after release from a G1 arrest in the presence or absence of the MCD1, a component of sister (more ...)
Reducing Mad1 expression allows MAD2/mad2Δ cells to respond to short linear chromosomes
Because the relative levels of Mad1 and Mad2 affect the spindle checkpoint (Chung and Chen, 2002
Sironi et al., 2002
), we hypothesized that decreasing Mad1 expression would allow MAD2/mad2
Δ cells to respond to chromosomes that were not under tension. We confirmed that Mad1 and Mad2 levels in diploid cells reflected gene dose: MAD1/mad1
Δ and MAD2/mad2
Δ heterozygotes express half as much of the corresponding proteins as wild-type cells (). We tested whether reducing the level of Mad1 affected the response of MAD2/mad2
Δ diploids to two perturbations that affect tension at kinetochores: short linear chromosomes and the absence of cohesin. As a control, we generated a MAD2
Δ double heterozygote.
FIGURE 5: MAD1/mad1Δ MAD2/mad2Δ cells arrest in response to loss of tension and loss of attachment. (A) Western blots to detect the relative levels of Mad1-TAP (top) or Mad2-TAP (bottom) epitope-tagged strains that are homozygous diploids, heterozygous (more ...)
The dose of Mad1 affected the response to short linear chromosomes. Two strains that had reduced levels of Mad2 and normal levels of Mad1 (a MAD2/mad2Δ single heterozygote and a MAD2/mad2Δ MAD3/mad3Δ double heterozygote) failed to respond to the short linear chromosomes. In contrast, the response of the MAD1/mad1Δ MAD2/mad2Δ double heterozygote resembled that of wild-type MAD2/MAD2 cells. This conclusion held for both long-term proliferation on plates () and passage through mitosis in a single cell cycle ().
The dose of Mad1 did not affect the response to loss of cohesin. We measured the response of the MAD1/mad1Δ MAD2/mad2Δ double heterozygote to the loss of tension produced by repressing Mcd1. Like wild-type cells, MAD1/mad1Δ cells arrested in mitosis in response to the loss of cohesin (). However, MAD1/mad1Δ MAD2/mad2Δ double heterozygotes showed no cell cycle arrest. We discuss the difference between the responses to short linear chromosomes and cohesin depletion in what follows.
MAD1 heterozygosity suppresses MAD2/mad2Δ haploinsufficiency for chromosome loss
Reducing Mad2 expression increases chromosome loss (). To determine whether a corresponding reduction in Mad1 expression can suppress this phenotype, we measured chromosome loss in MAD1/mad1Δ MAD2/mad2Δ cells (). The rate of chromosome loss in MAD1/mad1Δ MAD2/mad2Δ cells is not significantly different from the loss rate in wild-type MAD1/MAD1 MAD2/MAD2 cells (p = 0.13, χ2 test,
), but is significantly lower than that of MAD2/mad2Δ cells (p < 0.0001, χ2 test).