Mad2 affects the timing of the meiotic cell cycle
To investigate the role of Mad2 in meiosis, we analyzed sporulation in wild-type and mad2Δ
cells. We used the W303 budding yeast strain because the mitotic and meiotic phenotypes of mad2Δ
were previously characterized in this background (Hwang et al., 1998
; Shonn et al., 2000
). The process of sporulation, which includes meiosis and spore formation, can be induced through nutrient starvation of diploid budding yeast cells. Sixty-five percent of wild-type W303 cells sporulate, and of those, 96% package the four products of meiosis into four spores, forming a tetrad. Sixty-five percent of mad2Δ
cells sporulate as well. However, sporulation of mad2Δ
cells results in two major populations of spores: 1) 58% form tetrads, and 2) 34% form dyads, or asci containing two spores (). A small fraction of wild-type and mad2Δ
cells form triads (2 and 6%, respectively). We reasoned that investigating the differences between meiotic cell cycle events in wild-type cells, mad2Δ
cells that form dyads, and mad2Δ
cells that form tetrads might uncover a role of Mad2 in regulating the meiotic divisions.
Figure 1: Mad2 affects the duration of the meiotic cell cycle. (A) Wild-type and mad2Δ/mad2Δ sporulated cells were counted for the number of spores in each ascus. Nine hundred sporulated cells were counted in three biological replicates. (B, C) (more ...)
We first asked whether Mad2 is required for the proper timing of the meiotic divisions. Past studies did not detect a difference in timing of mad2Δ
cells, but these studies analyzed fixed cells and, due to asynchrony in meiotic induction, may not have detected small changes in specific phases of the meiotic cell cycle (Shonn et al., 2000
). To analyze more carefully the timing of the meiotic divisions, we used time-lapse microscopy to measure the duration of meiotic phases in individual wild-type and mad2Δ
cells. We expressed two green fluorescent protein (GFP)–tagged proteins, ZIP1-GFP and TUB1-GFP, to follow progression through meiosis. Zip1 is a component of the synaptonemal complex that assembles in zygotene and disassembles in diplotene, and it serves as a visual marker for prophase I (Sym et al., 1993
; Scherthan et al., 2007
). TUB1-GFP encodes a tagged α-tubulin, permitting observation of spindle formation and breakdown in meiosis I and meiosis II (Carminati and Stearns, 1997
; Straight et al., 1997
). Although we used the same fluorescent tag for both proteins, they are distinguishable because they are both morphologically and temporally different during the meiotic cell cycle (). Time-lapse images taken during sporulation allow us to measure the duration of each stage. We define the cell cycle stages based on disappearance of Zip1 and spindle morphology ().
The time-lapse microscopy indicates that the mad2Δ cells that form dyads undergo only one meiotic division, but the mad2Δ cells that form tetrads undergo both meiotic divisions (). We will refer to the mad2Δ cells that undergo one meiotic division and form dyads as “1 division mad2Δ cells” and the mad2Δ cells that undergo both divisions and form tetrads as “2 division mad2Δ cells.” The small fraction (6%) of mad2Δ cells that form triads also undergo two divisions; however, there is an error in spore formation, and only three of the four products of meiosis are packaged into spores. Therefore, 64% of sporulating cells undergo both divisions; in our analysis of the 2 division mad2Δ cells, we only include those that form tetrads since we are unsure about the underlying cause of the spore formation error resulting in a triad.
Observation of the 2 division mad2Δ cells shows that these cells have a faster prometaphase I/metaphase, 23 ± 8 min, compared with 41 ± 12 min in wild-type cells (). This difference is highly significant (p < 0.0001; unpaired Student's t test). This change in cell cycle timing only occurs in meiosis I; the duration of each stage in meiosis II of 2 division mad2Δ cells is similar to that of wild-type cells. The shorter duration of prometaphase I/metaphase I may cause the increase in chromosome missegregation that occurs in 2 division mad2Δ cells.
In contrast, the 1 division mad2Δ cells have an extended prometaphase I/metaphase I at 59 ± 16 min, compared with 41 ± 12 min in wild-type cells (). The difference in the duration of prometaphase I/metaphase I between wild-type and 1 division mad2Δ cells is highly statistically significant (p < 0.0001; unpaired Student's t test). The duration of anaphase I in 1 division mad2Δ cells is 69 min ± 13 min, substantially greater than the 13 ± 5 min in wild-type cells but much more similar to anaphase II in wild-type cells (76 ± 9 min). In the 1 division mad2Δ cells, the anaphase I spindle elongates and bends around the cell, which is a characteristic of anaphase II spindles but not anaphase I spindles in wild-type cells (). In summary, the 2 division mad2Δ cells have a faster prometaphase I/metaphase I, and the 1 division mad2Δ cells have a longer prometaphase I/metaphase I, but the entire meiotic cell cycle is shorter than in wild-type cells.
The 1 division mad2Δ cells undergo an aberrant meiotic division in which the homologous chromosomes pair and recombine, but sister chromatids separate inappropriately
To determine the role of Mad2 in meiosis, we further analyzed the 1 division mad2Δ
cells. We monitored the segregation of chromosome IV by placing a lactose operator (LacO) array near the centromere and expressing a GFP–lactose repressor fusion protein (GFP-LacI), targeting GFP to the chromosome (Straight et al., 1996
; Shonn et al., 2000
). When both homologous chromosomes have GFP targeted near CEN4, 97% of the mad2Δ
dyads have two GFP-marked chromosomes in each spore, suggesting that the spores are diploid (). As a further confirmation that the spores are diploid, we dissected the dyad spores and found that 78% were viable and able to sporulate.
Figure 2: The mad2Δ dyad spores are the result of an aberrant meiotic division in which homologous chromosomes pair and recombine and then sister chromatids separate. (A) mad2Δ/mad2Δ cells with both chromosome IV's marked with a LacO array (more ...)
The formation of two diploid spores after a single division in mad2Δ cells could be the result of either 1) segregating homologous chromosomes and ending the cell cycle after meiosis I or 2) separating sister chromatids inappropriately. To determine how the chromosomes separated in the 1 division mad2Δ cells, we labeled one of the two homologous chromosomes with a LacO array near the centromere of chromosome IV. If homologous chromosomes separate, one of the two spores will have two copies of the marked chromosome, and the other spore will have two copies of the unmarked chromosome. If sister chromatids separate, each spore will have one marked chromosome. When the LacO array is placed at the TRP1 locus, approximately 12 kb from CEN4, 97% of the dyads contain one marked chromosome in each spore (). We verified that chromosome III also segregates sister chromatids by placing the LacO array at the LEU2 locus, approximately 22 kb away from CEN3. Ninety-three percent of mad2Δ dyads contained one marked chromosome in each spore (). Surprisingly, the data show that sister chromatids separate inappropriately in the single meiotic division. These results suggest that in the 1 division mad2Δ cells, meiosis II events occur in the single meiotic division.
To investigate whether other events of meiosis I were perturbed in the 1 division mad2Δ cells, we examined whether the cells initiate meiosis correctly by pairing and recombining homologous chromosomes in prophase I. To monitor pairing, we examined spread meiotic nuclei of mad2Δ cells in the pachytene stage of prophase. In pachytene, homologous chromosomes have paired, synapsed, and initiated recombination. We marked both homologous chromosomes with a LacO array near CEN4 and expressed GFP-LacI. Chromosomes that are paired will have two GFP marked chromosomes in close proximity. One hundred percent of wild-type and 96% of mad2Δ cells have paired homologous chromosomes (). Using this assay, we cannot determine which cells will form dyads, but the proportion of cells with paired chromosomes is so great that it must include the mad2Δ cells that undergo one division, showing that there is not a defect in pairing in mad2Δ cells.
To determine whether crossovers occur in the 1 division mad2Δ cells, we monitored the segregation pattern of a LacO array placed at different locations along one of the homologous chromosome IVs. As shown earlier, if a LacO array is placed only 12 kb from the centromere on one of the homologous chromosomes, the sister centromeres split, and 97% of the dyads have one chromosome with the array in each spore. Only 3% of the dyads have two chromosomes with the array in one spore. We figured that if an array is placed further from the centromere, a crossover could occur between the array and the centromere, and the dyads would inherit two marked chromosomes in one spore. Indeed, with an array located approximately 100 kb from the centromere, 27% of the dyads have two marked chromosomes in one spore. With an array located 350 kb from the centromere, 36% of the dyads have two marked chromosomes in one spore (). In summary, the mad2Δ dyads undergo a single meiotic division in which homologous chromosomes pair, recombine, and separate sister chromatids inappropriately ().
In the 1 division mad2Δ cells, kinetochores are not clamped together, and cohesin does not remain protected around the centromere
Our results suggest that in the 1 division mad2Δ
cells, the paired homologous chromosomes attach sister chromatids to opposite spindle poles in metaphase I instead of homologous chromosomes (). In wild-type cells, sister chromatids do not attach to opposite spindle poles because the monopolin complex holds the sister chromatids' kinetochores together, assembling one microtubule-binding site (Toth et al., 2000
; Rabitsch et al., 2003
; Winey et al., 2005
; Petronczki et al., 2006
; Monje-Casas et al., 2007
). To determine whether monopolin localization is disrupted in 1 division mad2Δ
cells, we monitored the localization of the monopolin component Lrs4 tagged with GFP, using time-lapse microscopy (). In wild-type cells, Lrs4 normally resides in the nucleolus until the end of prophase I. Then Lrs4 leaves the nucleolus and binds to the kinetochores until the end of anaphase I (Rabitsch et al., 2003
). In 2 division mad2Δ
cells, Lrs4-GFP behaves similarly to wild-type cells (unpublished data). In 1 division mad2Δ
cells, Lrs4-GFP leaves the nucleolus but does not bind to kinetochores ().
Figure 3: In the mad2Δ cells that undergo one meiotic division, the monopolin complex does not bind sister kinetochores. (A) Time lapse images of meiosis in wild-type and 1 division mad2Δ/mad2Δ cells. Both strains are expressing Lrs4-GFP (more ...)
The chromosome segregation pattern of 1 division mad2Δ
cells is quite distinct from that in monopolin mutants. In monopolin mutants, sister chromatids cannot separate in meiosis I due to the protected centromeric cohesins. However, the spindle poles will separate, and the cells will undergo meiosis II, making four mostly inviable spores (Toth et al., 2000
; Rabitsch et al., 2003
). However, in the 1 division mad2
Δ cells, sister chromatids separate in the single division, making two viable diploid spores, suggesting that centromeric cohesins are not protected.
We monitored the meiotic cohesin Rec8 to determine whether centromeric cohesin is lost prematurely in the 1 division mad2
Δ cells. In wild-type cells, previous reports showed that there is stepwise cleavage of Rec8; the Rec8 along chromosome arms is cleaved in meiosis I, and the centromeric Rec8 is cleaved in meiosis II (Klein et al., 1999
; Buonomo et al., 2000
; Kitajima et al., 2003
). To monitor cohesin cleavage during the meiotic cell cycle, we tagged Rec8 with GFP in cells also expressing mCherry-TUB1. As expected, wild-type and 2 division mad2Δ
cells show a stepwise loss of Rec8. The majority of Rec8 is cleaved concurrently with anaphase I spindle assembly, leaving a fraction of Rec8 until meiosis II (). In the 1 division mad2Δ
cells, there is still a stepwise loss of cohesin, except that the first cleavage occurs prematurely, 45 ± 12 min prior to anaphase I.
Figure 4: In the mad2Δ cells that undergo one meiotic division, protection of the meiotic cohesin Rec8 is lost prematurely. (A) Time lapse images of meiosis in wild-type and 1 division mad2Δ/mad2Δ cells. Cells are expressing Tub1-mCherry (more ...)
Our results suggest that in the 1 division mad2Δ
cells, centromeric cohesins do not remain protected in anaphase I, allowing the cells to separate sister chromatids. We used time-lapse microscopy to monitor Sgo1, one of the factors required for protection of centromeric cohesins (Kerrebrock et al., 1995
; Katis et al., 2004
; Kitajima et al., 2004
; Marston et al., 2004
). We made an Sgo1-GFP fusion protein and expressed mCherry-TUB1 in wild-type and mad2Δ
cells. As expected, Sgo1-GFP associates with chromosomes throughout meiosis I in wild-type and in 2 division mad2Δ
cells (). In contrast, in 1 division mad2Δ
cells, Sgo1 is lost 38 ± 7 min prior to anaphase I. The loss of Sgo1 most likely results in the cleavage of centromeric Rec8 (Kitajima et al., 2004
; Marston et al., 2004
). Therefore, our data show that in the 1 division mad2Δ
cells, sister chromatids separate because sister kinetochores are not clamped together by monopolin, Sgo1 is lost prematurely, and centromeric cohesins are cleaved in the single meiotic division. Clamping of sister kinetochores by monopolin is independent of cohesin and cohesin protection (Monje-Casas et al., 2007
), suggesting that the phenotype seen in mad2Δ
cells is due to pleiotropic misregulation of the meiotic cell cycle. We diagram the timing of cell cycle events with respect to the stages of meiosis (as defined by spindle morphology) in .
Schematic representing the timing of different cell cycle events with respect to the stage of the cell cycle in wild-type, 2 division mad2Δ, and 1 division mad2Δ cells.
In the absence of Mad2, the APC/C is prematurely active in prometaphase of meiosis I
We considered that Mad2 regulates the meiotic cell cycle by modulating APCCdc20
activity. During spindle checkpoint signaling, Mad2, together with other checkpoint proteins, inhibits APCCdc20
activity to delay the cell cycle in metaphase I (Musacchio and Salmon, 2007
) To determine whether APCCdc20
is prematurely active in the absence of Mad2, we investigated the timing of the degradation of the APCCdc20
substrate securin/Pds1. Using time-lapse microscopy, we monitored Pds1-GFP and mCherry-TUB1. In wild-type cells, Pds1-GFP is degraded, and the cells enter anaphase I (). We were surprised to find that in the 2 division mad2Δ
cells, Pds1 is degraded on average 13 ± 12 min prior to anaphase I spindle assembly ( and ). Because prometaphase I/metaphase I is ~18 min faster in 2 division mad2Δ
cells than in wild-type cells (), Pds1 is in fact degraded ~31 min early (). It is not clear why the cells do not enter anaphase I immediately after loss of Pds1, but there may also be a misregulation of other cell cycle events preventing cohesin cleavage. It is striking that in the 1 division mad2Δ
cells, Pds1 is degraded even more prematurely: 81 min prior to anaphase I spindle assembly ( and ).
Figure 6: The APC is prematurely active in mad2Δ cells. (A) Time-lapse images of meiosis in wild-type and mad2Δ/mad2Δ cells. Both strains are expressing Pds1-GFP and Tub1-mCherry. One hundred wild-type cells, 100 mad2Δ cells that (more ...)
Average time from securin/Pds1 degradation to anaphase I spindle formation.
If prematurely active APC is indeed the cause of premature Pds1 degradation, then decreasing APC/C activity should prevent the early Pds1 degradation. We measured the timing of the loss of Pds1-GFP in mad2Δ
cells that also have a deletion of Swm1, a nonessential component of the APC/C. Cells that lack Swm1 target substrates for ubiquitination less efficiently than wild-type cells (Hall et al., 2003
; Schwickart et al., 2004
; Oelschlaegel et al., 2005
), but Pds1 is degraded at anaphase I in swm1Δ
cells ( and ). In mad2Δ swm1Δ
cells, we do not see premature degradation of Pds1; the degradation of Pds1 occurs within 1 ± 2 min of anaphase I spindle formation ( and ). Using time-lapse microscopy, we find that no mad2Δ swm1Δ
cells undergo only one meiotic division. Therefore, down-regulating APC/C activity prevents premature degradation of Pds1 in mad2Δ
cells and rescues the single-division phenotype. We conclude that in mad2Δ
cells, the APC/C is prematurely active.
The role of Mad2 in down-regulating APC/C activity in metaphase I is distinct from its role in delaying the cell cycle if a chromosome is not attached to the spindle
Because the activity of the APC/C is inhibited during spindle checkpoint signaling when a chromosome is not attached to spindle microtubules, we wanted to determine whether another protein required for the spindle checkpoint signaling in meiosis, Mad3, is also required to decrease APC/C activity during prometaphase I and metaphase I (Shonn et al., 2000
; Musacchio and Salmon, 2007
). We monitored loss of Pds1-GFP with respect to anaphase I spindle assembly in mad3Δ
cells. Pds1-GFP is degraded within 1 ± 2 min of anaphase I spindle formation in mad3Δ
cells, similar to what occurs in wild-type cells (). We also do not see the formation of dyad spores or the single-division phenotype. Our results indicate that in the absence of Mad3, the APC/C is not prematurely active, suggesting that Mad2 functions independently of Mad3 to down-regulate APC/C activity during prometaphase I/metaphase I. This is consistent with previous studies showing that mad3Δ
cells do not have an increase in meiosis I nondisjunction (Shonn et al., 2003
Both APCCdc20 and APCAma1 are prematurely active in 1 division mad2Δ cells
The meiosis-specific cofactor of the APC/C, Ama1, also targets Pds1 for ubiquitination and subsequent degradation and, therefore, may be prematurely active in mad2Δ
cells (Cooper et al., 2000
; Oelschlaegel et al., 2005
; Penkner et al., 2005
). Ama1 is not essential for meiosis but does function in meiosis I to promote the rapid degradation of APC/C substrates (Oelschlaegel et al., 2005
). To determine whether APCAma1
is prematurely active in mad2Δ
cells, we deleted Ama1 in wild-type and mad2Δ
cells and followed the degradation of Pds1-GFP with respect to spindle assembly by live-cell imaging. In ama1Δ
cells, Pds1 is degraded within 1 ± 2 min of anaphase I spindle assembly (). In ama1Δ mad2Δ
cells, Pds1 is degraded ~11 ± 8 min prior to anaphase I spindle assembly, similar to 2 division mad2Δ
cells. We do not see any cells that undergo a single meiotic division. We conclude that premature APCAma1
activity results in the very premature Pds1 degradation and the single-division phenotype in mad2Δ
Our results suggest that Mad2 may have a role in preventing premature APCAma1
activity. However, Mad2 could directly or indirectly inhibit APCAma1
. In wild-type cells, Cdc20 is required for Pds1 degradation, suggesting that APCAma1
is only active after APCCdc20
targets substrates for degradation (Salah and Nasmyth, 2000
; Oelschlaegel et al., 2005
). We reasoned that Mad2 may not directly inhibit APCAma1
, but might instead inhibit APCCdc20
activity, and that this inhibition prevents APCAma1
activity. To determine whether Cdc20 is required for the premature activity of APCAma1
, we analyzed whether Pds1 can be degraded in the absence of Cdc20 in mad2Δ
cells. We replaced the Cdc20 promoter with the mitosis-specific Clb2 promoter to make a Cdc20 meiotic null (cdc20-mn
; Lee and Amon, 2003
) and monitored Pds1 degradation with respect to metaphase I spindle assembly. The cdc20-mn
and cdc20-mn mad2Δ
cells do not degrade Pds1 prematurely. Ninety-eight percent of the cdc20-mn
and 96% of cdc20-mn mad2Δ
cells that enter meiosis are blocked in metaphase I with Pds1-GFP present for at least 200 min (). Our results demonstrate that APCCdc20
activity is required for the activation of APCAma1
in 1 division mad2Δ
cells. We conclude that in prometaphase I, Mad2 indirectly prevents premature APCAma1
activity by inhibiting APCCdc20
Figure 7: APCCdc20 activity is required for the premature APCAma1 activity in mad2Δ cells. (A) Time-lapse images of meiosis in cdc20-mn/cdc20-mn and mad2Δ/mad2Δ cdc20-mn/cdc20-mn cells. All strains are expressing Pds1-GFP and Tub1-mCherry. (more ...)
We decided to further explore the regulation of APCAma1
. A previous study demonstrates that APCAma1
activity is inhibited during metaphase I by cyclin-dependent kinase (CDK; Oelschlaegel et al., 2005
). However, an allele of AMA1
, with all eight putative CDK phosphorylation sites mutated to alanine does not result in premature APCAma1
activity, suggesting that the inhibition of APCAma1
by CDK is not direct or that there is redundancy in the regulatory pathway. To determine whether Ama1 phosphorylation regulates APCAma1
activity in mad2Δ
cells, we analyzed mad2Δ Ama1-m8
cells expressing mCherry-TUB1 by time-lapse microscopy. Surprisingly, 87% of sporulated mad2Δ Ama1-m8
cells undergo a single meiotic division (). The percentage sporulation of mad2Δ Ama1-m8
cells was similar to that in wild-type cells. In accordance with previous observations, we did not see a phenotype of Ama1-m8
in the wild-type background (Oelschlaegel et al., 2005
). Our results suggest that there is redundancy in the regulatory network to prevent premature activity of APCAma1
in metaphase I. We propose that Ama1 is inhibited by phosphorylation and by another activity of CDK. In the absence of Mad2, premature APCCdc20
activity could lead to less CDK activity and, in some cells, the dephosphorylation of Ama1, resulting in the one-division phenotype.
A nonphosphorylatable form of Ama1 increases the percentage of mad2Δ cells that undergo 1 meiotic division.
Premature APCAma1 activity results in the premature release of the Cdc14 phosphatase
Our results demonstrate that in the absence of Mad2, securin/Pds1 is degraded prematurely. The degradation of securin results in the release of the protease separase, which cleaves sister chromatid cohesion. Separase also functions in the Cdc14 early anaphase release (FEAR) network, promoting the release of the Cdc14 phosphatase from the nucleolus (Stegmeier et al., 2002
; Buonomo et al., 2003
; Marston et al., 2003
). Once released, Cdc14 counteracts CDK activity by dephosphorylating CDK substrates. Because the 1 division mad2Δ
cells degrade securin/Pds1 prematurely, we analyzed whether Cdc14 is released prematurely. We monitored the timing of release and resequestration of Cdc14-GFP in wild-type and mad2Δ
cells using time-lapse microscopy. In wild-type cells, Cdc14-GFP is released from the nucleolus 1 ± 4 min before anaphase I, is resequestered into the nucleolus in metaphase II, and then is released again 5 ± 5 min before anaphase II (). The 2 division mad2Δ
cells have similar timing of Cdc14 release; Cdc14 is released 6 ± 6 min before anaphase I, resequestered, and released again 6 ± 5 min before anaphase II. Surprisingly, in 1 division mad2Δ
cells, Cdc14 is also released twice, but both releases occur before the first division. Cdc14 is first released 54 ± 7 min before anaphase I spindle formation, resequestered, and released again 6 ± 6 min before anaphase I spindle formation (). These results support our conclusion that meiosis II cell cycle events are uncoupled from the chromosome segregation cycle in the 1 division mad2
Δ cells ().
Figure 8: The phosphatase Cdc14 is released from the nucleolus prematurely in the 1 division mad2Δ cells, and the premature release is dependent on APCAma1 activity. (A) Time-lapse images of meiosis in wild-type, mad2Δ/mad2Δ, and mad2Δ/mad2Δ (more ...)
Our results suggest that the premature degradation of Pds1 leads to separase activation of the FEAR network and Cdc14 release. Because Ama1 is required for the premature degradation of Pds1 in mad2Δ cells that undergo one division, Ama1 should also be required for the premature release of Cdc14. Analysis of Cdc14 release in mad2Δ ama1Δ cells reveals that it occurs with timing similar to that of wild-type cells during both meiotic divisions (). These data show that the activity of AMA1 is required for the premature release of Cdc14, most likely through targeting Pds1 for ubiquitination and subsequent degradation.