Aurora controls sister kinetochore mono-orientation and homolog bi-orientation in meiosis-I

Ark1 is required for the correction of malattachment

In both budding yeast and metazoans, Aurora kinases act in the bi-orientation of chromosomes by correcting improper, syntelic attachment of chromosomes to the same spindle pole (Tanaka et al, 2002; Hauf et al, 2003; Lampson et al, 2004). When we inactivated Ark1 in an otherwise unperturbed mitosis, segregation of both sisters to one spindle pole was rare (Figure 1). This could be because in an unperturbed mitosis, initial syntelic attachment is rare or because syntelic attachment can still be corrected in ark1-mutant cells. We therefore increased the frequency of misattachment by first arresting cells in mitosis without microtubules, using a cold-sensitive tubulin mutant nda3-KM311 and then releasing to permissive temperature (Trautmann et al, 2004; Grishchuk and McIntosh, 2006; Kawashima et al, 2007). Because ark1-mutant cells do not arrest in mitosis under these conditions (Petersen and Hagan, 2003; data not shown), we could only perform the experiment with the ark1-as3 allele, which is functional when cells are grown without inhibitor and thus allows arrest by nda3-KM311. Shortly before release, we inhibited Ark1-as3 by the specific inhibitor 1NM-PP1, which inactivates the kinase within 10 min (data not shown and Figure 2A). When checking anaphase spindles 10 min after the release, both cen2-GFP marks were found close to one edge of the spindle in about 20% of cells. No such missegregation was observed in wild-type cells treated with inhibitor or in ark1-as3 cells grown without inhibitor (Figure 2A), indicating that its occurrence depends on the inactivation of Ark1. To determine whether this indeed represents initial misattachment that fails to correct, we filmed Ark1-inactivated cells being released from the nda3 arrest. Among those cells that we could image from the start of mitosis, about half had both sister centromeres located entirely at one pole throughout prometaphase and anaphase (Figure 2B and C; Supplementary Figure S2). These data indicate that Ark1 is required to correct syntelic misattachment of chromosomes (also see Supplementary Note 1).

Open in a separate window

Figure 2

Ark1 is required for the correction of malattachment. (A) The indicated strains carrying the nda3-KM311 mutation were arrested in mitosis by incubation at 19°C for 6 h, and released by transfer to 32°C. Where indicated (+inhib.) 5 μM 1NM-PP1 were added to the culture 10 min before release. Cells were fixed with methanol 10 min after upshift to 32°C. Chromosome segregation was assessed by determining cen2-GFP localization on anaphase spindles, which were labeled by mCherry-Atb2 (tubulin). (B, C) nda3-KM311 ark1-as3 cells were arrested in mitosis as in panel A and released by transferring cells to a microscope stage kept at 32°C. 1NM-PP1 (5 μM) was added 5 min before release. Only cells that could be followed through mitosis starting from very short spindle length were considered. Cells in which the spindle was defective and those in which centromere 2 did not attach to the spindle were excluded from the analysis. A kymograph of a cell that showed chromosome 2 segregating with one SPB is shown in panel C. Example kymographs for all phenotypes as well as example kymographs from ark1-as3 cells released from the nda3-KM311 arrest without any inhibitor are shown in Supplementary Figure S2A. (D) The indicated strains marked with cen2-GFP and mCherry-Atb2 (tubulin) were followed by live-cell microscopy at the restrictive temperature (34°C). A centromere was considered to have ‘switched' when the trajectory of cen2-GFP from one pole to the other could be clearly observed (arrows). The number of ‘switches' per hour in a total of 2 h prometaphase observation time is given.

To confirm in a different setting that malattachment of chromosomes cannot be corrected in ark1-mutant cells, we used a temperature-sensitive mutant of fission yeast cohesin, psc3-1T (Nonaka et al, 2002). Sister chromatids precociously detach from each other in the absence of cohesin, but remain competent to attach to microtubules. However, in the absence of cohesion between sister chromatids, kinetochore–microtubule attachment cannot be stabilized because of a lack of tension. Therefore, attachment remains unstable, which in budding yeast depends on Ipl1 (Biggins and Murray, 2001). In live-cell microscopy experiments, the instability of sister chromatid attachment in psc3-1T cells was exemplified by an occasional switching of at least one sister chromatid from one spindle pole to the other (Figure 2D, arrows). When Ark1 was additionally inactivated, such switching was abolished (Figure 2D), indicating that Ark1 is required to keep sister chromatid attachment unstable in psc3-1T cells. This result is consistent with Ark1 being needed to correct kinetochore–microtubule attachments that fail to generate tension through bi-orientation. However, the fact that sister chromatids were not very motile when Ark1 is inhibited (Figure 2D) could also indicate a more profound defect in kinetochore–microtubule dynamics. Inactivation of condensin did not seem to have the same effect as Ark1 inactivation (Supplementary Figure S2B), which again suggests that Ark1 regulates chromosome attachment independent of condensin.

Ark1 has a role in nuclear division during meiosis-I

To study the requirement of Ark1 in meiosis, we used an ‘ark1 shut-off' (ark1 s.o.) strain, where the ark1+ gene is under control of the mitosis-specific promoter of the rad21+ gene. In this strain, the level of Ark1 protein in meiosis was largely reduced, but minor amounts of Ark1 were still observed at centromeres and midspindles (Supplementary Figure S4). For some experiments, we therefore either additionally or alternatively used the ark1-as2 allele together with the specific inhibitor.

The most prominent phenotype observed after Ark1 depletion in mitosis is a failure to condense chromosomes, which leads to a defect in nuclear division (Petersen et al, 2001). We similarly found a defect in chromosome condensation during meiosis in ark1 s.o. cells (Supplementary Figure S5). Nevertheless, four distinct, albeit often unequally sized, nuclei formed after meiosis-II in ark1 s.o. cells (Supplementary Figure S5). However, in about 50% of ark1 s.o. cells, the two meiosis-II spindles formed extremely close to each other and often in what seemed to be one nucleus (Figure 3A), indicating a failure of nuclear division during meiosis-I. To assess this phenotype better, we arrested cells after meiosis-I using the mes1 mutation (Izawa et al, 2005). Under these conditions, more than 80% of ark1+ cells but only 30% of ark1 s.o. cells formed two nuclei (Figure 3B and C). Nevertheless, the mononucleated ark1 s.o. cells seemed to have undergone anaphase-I, because Rec8-GFP was largely removed from chromatin (Supplementary Figure S6). The failure in nuclear segregation after Ark1 depletion presumably reflects a failure in condensin function, since mutants in the condensin subunit cut3 also show a slightly increased number of cells with one nucleus when arrested by the mes1 mutation at a semi-permissive temperature (data not shown).

Open in a separate window

Figure 3

Ark1 is required for nuclear division during meiosis-I. (A) The position of metaphase-II spindles was evaluated in fixed cells. DNA was stained with Hoechst 33342. (B, C) The indicated strains were arrested after anaphase-I by the mes1 mutation. Cells were fixed and DNA was stained with DAPI. The number of nuclei was determined in at least 200 asci (C); exemplary cells are shown in panel B.

Ark1 promotes homolog bi-orientation in meiosis-I

To address whether Ark1 has any role in meiotic chromosome segregation, we examined the segregation of cen2-GFP in ark1 s.o. and ark1-as2 cells during anaphase-I. When Ark1 was inhibited or depleted, both homologous chromosomes 2 segregated to the same pole in about 30–40% of anaphase-I cells (Figure 4A). In another 30% of cells, at least one of the homologs was lagging on the anaphase-I spindle. These segregation problems could be caused either by a failure to bi-orient bivalents or by a failure to join homologous chromosomes through chiasmata. Since the intergenic recombination frequency between the lys3+ and ura1+ locus on chromosome 1 was similar in ark1+ and ark1 s.o. strains (data not shown), the latter is unlikely. Therefore, these data suggest that Ark1 is required for the bi-orientation of connected homologs during meiosis-I (also see Supplementary Note 2).

Open in a separate window

Figure 4

Ark1 is required for proper homolog bi-orientation in meiosis-I. (A) The indicated strains with both homologous chromosomes 2 marked by GFP were observed during anaphase-I using CFP-Atb2 (tubulin) as a marker. (B) The indicated strains, in which one of the homologous chromosomes 2 was marked by GFP, were fixed during meiosis by methanol. DNA was stained by Hoechst 33342 and the segregation of sister centromeres was determined in binucleated cells.

The role of Ark1 in bi-orientation is independent of kinetochore geometry

Paliulis and Nicklas (2000) have shown that specific features of the chromosome and not of the spindle determine the special chromosome segregation of meiosis-I. In fission yeast, genetic tricks can be used to create mitosis-like chromosomes during meiosis-I. Deletion of the gene for the meiosis-specific cohesin subunit Rec8 causes a failure in recombination and therefore chiasmata generation. Additionally, sister kinetochores in each homolog are faithfully bi-oriented presumably because geometry at the centromeric region is mitosis-like and sister chromatids remain cohered by mitotic cohesin complexes, which persist into meiosis (Yokobayashi et al, 2003). Similarly, mitosis-like chromosomes can be generated by deleting rec12+, which is required for recombination, and moa1+, the gene product of which is required for the mono-orientation of sister kinetochores in concert with Rec8 (Yokobayashi and Watanabe, 2005). In both these genetic backgrounds, Ark1 shut-off caused missegregation of sister centromeres in meiosis-I (Figure 4B), with the effect being greater in a rec12Δ moa1Δ strain than in a rec8Δ strain (see Supplementary Note 3). Thus, during meiosis-I, Ark1 can promote the equational segregation of sister chromatids (Figure 4B), or the bipolar segregation of homologs (Figure 4A), depending on the chromosome structure. We suggest that Ark1 promotes the bi-orientation of any two kinetochore-containing entities that are connected. This is in accordance with findings by Dewar et al (2004), who showed that budding yeast Ipl1 ensures the bi-orientation of two separate kinetochores on an unreplicated plasmid.

Ark1 is required for the mono-orientation of sister chromatids in meiosis-I

When we tested the segregation of homologs during anaphase-I, we found that frequently at least one of the homologs was lagging and often the cen2-GFP signal of lagging homologs split in two (Figure 4A). We therefore hypothesized that the sister kinetochores on one homolog, which normally attach to only one spindle pole, were pulled in opposite direction. To verify this assumption, we labeled only one of the homologous chromosomes 2 with GFP and determined the segregation pattern. Indeed, in about 12% of anaphase-I cells, the GFP signal split in two when Ark1 was depleted (Figure 5A), indicating that sister chromatids were pulled to opposite poles and separated precociously. These data suggested that in the absence of Ark1, sister kinetochores erroneously become attached to opposite spindle poles during meiosis-I. In accordance, we found that a visible separation of sister centromeres could already be observed during metaphase-I (Supplementary Figure S7). We therefore considered the possibility that Ark1 is required for the localization and function of Moa1, which prevents bi-orientation of sister chromatids at meiosis-I (Yokobayashi and Watanabe, 2005). However, visualization of Moa1-GFP did not reveal any difference between wild-type and ark1 s.o. cells during meiosis-I (Figure 5B). Similarly, no significant difference in Moa1 localization was observed by chromatin immunoprecipitation (ChIP; Figure 5C). Furthermore, the localization of Rec8 at the centromeric central core, which is important for mono-orientation of kinetochores (Watanabe et al, 2001; Yokobayashi et al, 2003), was intact in the ark1 s.o. cells (Supplementary Figure S8). Thus, the mono-orientation defect caused by the reduction of Ark1 is likely different from the one caused by the absence of Moa1 or Rec8, suggesting that Ark1 and Moa1/Rec8 influence sister kinetochore mono-orientation in meiosis-I through distinct mechanisms.

Open in a separate window

Figure 5

Ark1 is required for proper sister chromatid mono-orientation in meiosis-I independent of Moa1. (A) One of the homologous chromosomes 2 was marked by GFP in the indicated strains and the segregation pattern was determined during anaphase-I using CFP- or mCherry-Atb2 (tubulin) as a marker. (B) Cells from the indicated strains expressing Moa1-GFP and CFP-Atb2 (tubulin) were observed during metaphase-I. (C) Diploid strains of the indicated genotype were arrested in prophase-I by deletion of mei4+, and the amount of Moa1 at centromeres was determined by ChIP using a Moa1 antibody.

Sister centromeres separate precociously after Ark1 depletion because of merotelic attachment rather than complete bi-orientation

Cells in which moa1+ is deleted show entirely equational segregation of sister chromatids in meiosis-I if recombination is abolished by rec12Δ (see Figure 4B). In this situation lagging chromatids can be observed in anaphase-I. Their appearance is completely suppressed by deleting sgo1+, the meiosis-I-specific protector of centromeric cohesion (Yokobayashi and Watanabe, 2005). In contrast, neither deletion of rec12+ nor the additional deletion of sgo1+ leads to completely equational segregation in ark1 s.o. cells (Supplementary Figure S8C, and data not shown). When contemplating the reason for the mono-orientation defect in ark1 s.o. cells, we noticed the high frequency of lagging chromosomes or lagging sister chromatids during anaphase-I (Figures 4A and ​and5A).5A). Given that lagging chromosomes may originate from merotelic attachment and Ark1 is involved in its correction (Figure 1), we envisaged that the primary defect in ark1 s.o. cells preventing monopolar attachment could be the inability to correct merotelic attachment of a unified pair of sister kinetochores (Figure 6B). In rec12Δ moa1Δ cells, all sister centromeres eventually segregated at anaphase-I even in the presence of Sgo1, implying that microtubule-mediated pulling on bi-oriented sister kinetochores can overcome Sgo1-mediated protection (Vaur et al, 2005; Yokobayashi and Watanabe, 2005; Figure 6B). The tension on homologs that are attached in a merotelic manner is expected to be less, because some microtubules on both sister kinetochores likely attach to the same pole (Figure 6B, ‘sgo1+'). This reduced tension might not be sufficient to overcome protection by Sgo1. This hypothesis makes the key prediction that deprotection of sister chromatid cohesion by sgo1Δ would increase the equational segregation of sister centromeres in ark1 s.o. cells, different from rec12Δ moa1Δ cells (see Figure 6B). Consistent with this scenario, the ratio of cells in which sister centromeres were moving entirely to opposite poles in anaphase-I increased from ∼1% in ark1 s.o. to 17% in ark1 s.o. sgo1Δ cells (Figure 5A). Furthermore, lagging sister chromatids could be observed in ark1 s.o. cells even after sgo1+ deletion, indicating that single chromatids were attached in a merotelic manner (Figures 5A and ​and6B,6B, ‘sgo1Δ').

Open in a separate window

Figure 6

Sgo2 promotes mono-orientation similar to Ark1. (A) To assess the percentage of equational segregation (light gray bars), one of the homologous chromosomes 2 was marked by GFP in the indicated strains and the segregation pattern was judged in asci that had formed four spores. The percentage of lagging chromosomes or chromatids (dark gray bars) was determined by staining fixed cells with DAPI and the TAT-1 antibody recognizing tubulin. (B) Model for the chromosome segregation defects after Sgo2/Ark1 versus Moa1/Rec8 depletion. The centromeric region of one homolog is shown. In sgo2Δ or ark1 s.o. cells, the pair of sister kinetochores on one homolog tends to attach in a merotelic manner, that is, to both spindle poles, during metaphase-I. In anaphase, lagging homologs are common because of merotelic attachment. Eventually, both sister centromeres segregate to the same pole in most cases. If the protector Sgo1 is removed, sister centromeres segregate to opposite poles more frequently. Lagging chromatids are still observed after deletion of sgo1+, indicating merotelic attachment of single sister kinetochores. When moa1+ or rec8+ is deleted, sister kinetochores on one homolog lose their intimate connection. When rec12+ is deleted in addition to moa1+, sister chromatids segregate completely equationally in anaphase-I. Nevertheless, lagging chromosomes occur, but deleting sgo1+ abolishes their appearance.

Thus, we suggest that Ark1 promotes monopolar attachment of sister kinetochores at meiosis-I most likely by correcting merotelic attachment of paired sister kinetochores, a mechanism that is fundamentally different from that of Moa1 and Rec8, which are thought to promote the side-by-side orientation of sister kinetochores by fostering cohesion in the central core region of the centromere (Figure 6B).

Ark1 acts on chromosome attachment in a similar way during mitosis and meiosis

Since Ark1 promotes sister chromatid bi-orientation in mitosis, but sister chromatids have to mono-orient during meiosis-I, it was unclear how Ark1 would influence chromosome segregation in meiosis-I. We found that Ark1 normally promotes the bi-orientation of homologs in meiosis-I, but if the morphology of the bivalent is disrupted and mitosis-like chromosomes are created, Ark1 promotes the bi-orientation of sister chromatids (Figure 4). This indicates that the molecular mechanism of Ark1 function is the same in mitosis and meiosis, and the different outcome is determined by the structure of the bivalent.

In contrast to mitosis, where Ark1 inhibition causes only about 10–20% sister chromatid co-segregation, depletion of Ark1 in meiosis causes a more pronounced co-segregation of homologous chromosomes (30–40%). The most likely explanation is that tension-controlled correction of attachment is more important in meiosis, because the two pairs of sister kinetochores on a bivalent are not as tightly coupled as the sister kinetochores on a mitotic chromosome, which might favor their syntelic attachment (Shonn et al, 2000).

Very recently it has been demonstrated that the budding yeast Aurora kinase, Ipl1, is similarly required for homolog bi-orientation and sister chromatid bi-orientation of artificial mitosis-like chromosomes in meiosis-I (Monje-Casas et al, 2007). Because of the high conservation of Aurora functions in all eukaryotes, we expect that this will also hold true for metazoans.

Does Ark1 facilitate cleavage of Rec8?

In C. elegans, the Aurora-B kinase AIR-2 seems to be required to efficiently remove cohesin complexes containing Rec8 from chromosome arms (Kaitna et al, 2002; Rogers et al, 2002). In contrast, we find that Ark1 is not essential for Rec8 removal during meiosis-I in fission yeast (Supplementary Figure S10). In time-lapse movies of fission yeast expressing Rec8-GFP, one can clearly observe the solubilization of Rec8 at the onset of anaphase-I, presumably at the moment when it is cleaved and removed from chromatin (Supplementary Figure S10). Although this step is not as easy to discern when Ark1 is inhibited, because chromosomes are less condensed, it is clear that this solubilization also happens fairly efficiently if Ark1 is inhibited and Rec8 is subsequently degraded with kinetics similar to wild-type cells (Supplementary Figure S10). Nevertheless, it is possible that Ark1 facilitates but is not essential for Rec8 cleavage (see Supplementary Note 3).

A novel mechanism to promote mono-orientation of sister chromatids during meiosis-I

We show that fission yeast Ark1 is required for the faithful mono-orientation of sister chromatids in meiosis-I (Figure 5). In fission yeast, mono-orientation of sister kinetochores is regulated by Moa1 and Rec8, which may cooperatively promote the formation of a side-by-side structure of sister kinetochores through cohesion of the centromeric central core region (Yokobayashi and Watanabe, 2005). The depletion of Moa1 together with Rec12 or of Rec8 entirely disrupts the mono-orientation of sister chromatids at meiosis-I. The mono-orientation defect in ark1 s.o. cells is less pronounced, and Moa1- or Rec8-localization is not disrupted when Ark1 is depleted (Figure 5; Supplementary Figure S8). This indicates that Ark1 and Moa1 act in separate pathways to promote mono-orientation. Our experiments suggest that in the absence of Ark1, sister kinetochores on one homolog become attached in a merotelic manner so that they are torn apart at anaphase-I, even though they have the proper side-by-side configuration that favors mono-orientation (Figure 6B). Thus, the complicated chromosome segregation defects in meiotic cells depleted of Ark1 can be explained by the well-recognized role of Aurora in correcting malattachment of chromosomes. Syntelic and merotelic attachment of bivalents in meiosis-I provokes non-disjunction of homologs and precocious sister separation, respectively.

In budding yeast, it has been proposed that the two pairs of sister kinetochores on a bivalent only attach to one microtubule each (Winey et al, 2005) and one sister kinetochore may thus be inactivated. Consequently, merotelic attachment might not be possible, which would explain the unperturbed mono-orientation in Ipl1-depleted cells (Monje-Casas et al, 2007) despite the otherwise similar function of Ipl1 and Ark1. There is, however, controversy in the literature whether the depletion of Ipl1 causes a mono-orientation defect (Yu and Koshland, 2007). In any case, our data clearly indicate that in fission yeast, both sister kinetochores are active in meiosis-I and can attach to microtubules. As attachment of both kinetochores to microtubules at meiosis-I is observed in several organisms (Lee et al, 2000; Parra et al, 2004), the mechanism we identified here may be functional in other eukaryotes as well.

Culture conditions

Medium for mitotic cultures was YEA (YE with additional 50 mg/l adenine) or minimal medium (MM) containing 5 g/l NH₄Cl and additional nutrients if required (Alfa et al, 1993). To synchronize cells in mitosis (Figure 1A), we arrested cells in S-phase by incubation in 12 mM hydroxyurea for 4.5 h at 25 or 30°C depending on the strain. The arrest was released by washing the cells twice with fresh, warm medium before reculturing. Cells carrying the nda3-KM311 mutation (Figure 2A and B) were arrested in mitosis by incubation at 19°C for 6 h and subsequently released by shifting to 32°C. To observe cells in meiosis, cells were first grown to logarithmic phase. If the nmt41 or nmt81 promoter should be induced by thiamine depletion, cells were grown in MM containing 5 g/l NH₄Cl and, if necessary, 200 mg/l leucine and 50 mg/l adenine for about 14 h at 30°C. Cells were washed, collected and spotted on sporulation agar (SPA; Gutz et al, 1974) to which leucine or adenine had been added if necessary. After a further 7–8 h of incubation at 30°C, cells were observed directly or fixed by methanol at −80°C. To observe ark1-as2 cells in meiosis, the cells were first incubated in MM with 5 g/l NH₄Cl for 8–9 h, and then washed and incubated in MM without NH₄Cl for 4–5 h before spotting on a plate with synthetic sporulation agar (SSA; SSL with agar; Egel, 1971) containing 5 μM 1NA-PP1. Cells were observed after 7–10 h of incubation at 30°C.

Immunostaining and DNA staining

For immunostaining, cells were fixed with paraformaldehyde. To stain microtubules, we used the mouse anti-tubulin TAT1 antibody (kind gift from K Gull) at a dilution of 1:200, followed by an Alexa568-coupled anti-mouse secondary antibody (Invitrogen) at 2 μg/ml. To stain DNA, methanol-fixed cells were washed, resuspended in PEM buffer (100 mM PIPES, 5 mM EGTA, 5 mM MgCl₂, pH 6.9) and stained by 1 μg/ml Hoechst 33342 or 1 μg/ml DAPI.

Image acquisition

Images were acquired on a Zeiss AxioImager microscope (Zeiss, Jena, Germany) with MetaMorph software (Molecular Devices Corporation, Downingtown, PA). Typically, a Z-stack of about 4-μm thickness, with single planes spaced by 0.25–0.4 μm, was acquired and subsequently projected to a single image. To compare signal intensities, all images were taken with the same exposure conditions and processed similarly.

Time-lapse imaging

Live-cell recordings were performed on a DeltaVision RT system (Applied Precision, Issaquah, WA) equipped with a heating chamber. For imaging mitosis in ark1-mutants, cells were grown in liquid medium at permissive temperature, transferred to a glass-bottom culture dish (MatTek, Ashland, MA) coated with lectin and incubated on the microscope stage at the restrictive temperature (34°C) for at least 1 h, before starting image acquisition. To image cells in an nda3-KM311 release, cells were transferred from a liquid culture at 19°C to a glass-bottom culture dish coated with lectin, which was placed in the microscope chamber heated to 32°C. Image acquisition was started immediately. Images usually were acquired with the Z-sweep acquisition (OAI) feature and deconvolved using softWoRx software. Kymographs were assembled with Adobe Photoshop and Image Ready software.

We thank Iain Hagan, Jonathan Millar, Keith Gull and Roger Tsien for sharing strains, antibodies and plasmids; Eva Illgen for excellent technical assistance; Andrι Koch for generating the ark1-as3 strain; Alwin Kϕhler, Stephanie Küng and Hanna Windecker for comments on the manuscript; Dmitri Ivanov for discussions and the Watanabe laboratory, in particular Shihori Yokobayashi, Tomoya Kitajima and Takeshi Sakuno, for support. We are grateful for the financial support by the German Research Foundation (DFG), the Human Frontier Science Program (HFSP) and the Max Planck Society.