Spo13 prevents premature cohesin cleavage during meiosis

Background: Meiosis produces gametes through two successive nuclear divisions, meiosis I and meiosis II. In contrast to mitosis and meiosis II, where sister chromatids are segregated, during meiosis I, homologous chromosomes are segregated. This requires the monopolar attachment of sister kinetochores and the loss of cohesion from chromosome arms, but not centromeres, during meiosis I. The establishment of both sister kinetochore mono-orientation and cohesion protection rely on the budding yeast meiosis I-specific Spo13 protein, the functional homolog of fission yeast Moa1 and mouse MEIKIN. Methods: Here we investigate the effects of loss of SPO13 on cohesion during meiosis I using a live-cell imaging approach. Results: Unlike wild type, cells lacking SPO13 fail to maintain the meiosis-specific cohesin subunit, Rec8, at centromeres and segregate sister chromatids to opposite poles during anaphase I. We show that the cohesin-destabilizing factor, Wpl1, is not primarily responsible for the loss of cohesion during meiosis I. Instead, premature loss of centromeric cohesin during anaphase I in spo13 Δ cells relies on separase-dependent cohesin cleavage. Further, cohesin loss in spo13 Δ anaphase I cells is blocked by forcibly tethering the regulatory subunit of protein phosphatase 2A, Rts1, to Rec8. Conclusions: Our findings indicate that separase-dependent cleavage of phosphorylated Rec8 causes premature cohesin loss in spo13 Δ cells.


Introduction
Sexual reproduction relies on a cell division programme called meiosis. In humans, this is highly error-prone and may give rise to infertility, miscarriage or chromosomal abnormalities such as Down syndrome (reviewed by Hassold & Hunt, 2001). Meiosis consists of two consecutive divisions, where homologous chromosome segregation in meiosis I is followed by mitosis-like sister chromatid segregation in meiosis II. Homologue segregation requires a number of adaptations to the chromosome segregation machinery (Marston & Amon, 2004), including recombination of homologues, mono-orientation of sister kinetochores and the protection of pericentromeric cohesin in meiosis I.
Cohesin is a multi-subunit protein complex made up of the core subunits Smc1, Smc3 and the kleisin α-Scc1 (Losada et al., 1998;Michaelis et al., 1997) as well as the accessory subunits Scc3 (Tóth et al., 1999) and Pds5 (Hartman et al., 2000;Panizza et al., 2000). In mitosis, cohesin resists the spindle forces that pull sister chromatids towards opposite poles, likely by topologically linking sister chromatids (Gruber et al., 2003;Haering et al., 2002). Upon successful bi-orientation, securin (Pds1 in yeast) is ubiquitinated and destroyed by the proteasome, freeing separase (Esp1), which proteolytically cleaves Scc1 and thereby allows chromosome segregation. While previous research has identified key mechanisms governing cohesin protection, a number of additional proteins have been implicated in this process, but their roles remain unclear. Amongst them is the meiosis I-specific Spo13 (Wang et al., 1987). Cells without SPO13 only undergo a single meiotic division and show a variety of meiotic defects, including failure to mono-orient sister kinetochores in meiosis I and inability to protect cohesin (Katis et al., 2004b;Klapholz & Esposito, 1980;Lee et al., 2004;Shonn et al., 2002). Spo13 is thought to have functional orthologs in both fission yeast (Moa1) and mouse (MEIKIN) (Kim et al., 2015). The unifying feature of these proteins is their interaction with Polo kinases, whose kinetochore recruitment by Spo13, Moa1 and MEIKIN has been proposed to enable mono-orientation and cohesin protec- The exact role of Spo13 in cohesin protection is currently unclear. Interestingly, SPO13 overexpression blocks cohesin

Amendments from Version 1
Our revised manuscript consists of the changes in response to the reviewers and we additionally added a further figure (Figure 3) which presents data adding further support to the conclusion that Spo13 prevents loss of all cohesion during anaphase I, and responds to point 2 raised by reviewer 1 (Hochwagen). This figure includes three new experiments in which we analysed spo13Δ mad2Δ cells which undergo two meiotic divisions. Similar to one-division spo13Δ meiosis, in two-division spo13Δ meiosis, we observed reduced Rec8 at centromeres in anaphase I and frequent segregation of sister chromatids to opposite poles. The result is that spo13Δ mad2Δ mutants exhibit profound aneuploidy. We believe that this interesting data confirms the central conclusion of our manuscript which is that Spo13 is important for the retention of cohesion during meiosis I. Here, we take a live cell imaging approach to re-evaluate the importance of Spo13 for cohesin protection. We show that both cohesin and sister chromatid cohesion are lost upon anaphase I onset in spo13Δ cells. Furthermore, we confirm that cohesin removal results from separase-mediated cleavage rather than removal by the prophase pathway. We also provide evidence that PP2A is capable of preventing cohesin cleavage in spo13Δ cells.

Results
Pericentromeric cohesin is prematurely lost in spo13Δ cells Previous analyses of fixed cells found that centromeric Rec8 is undetectable or greatly diminished in Further evidence that Spo13 is important for protection of centromeric cohesion came from the analysis of cells lacking the monopolin subunit, Mam1, which biorient, rather than monoorient sister kinetochores, yet fail to segregate sister chromatids due to the persistence of centromeric cohesion. Importantly inactivation of SPO13 allowed mam1Δ cells to segregate sister chromatids during anaphase I (Katis et al., 2004b;Lee et al., 2004). Together, these findings provide evidence that centromeric cohesion is impaired in spo13Δ cells. However, it has been argued that residual centromeric cohesin persists after securin destruction in spo13Δ cells and prevents timely spindle elongation (Katis et al., 2004b). To clarify the importance of Spo13 in centromeric cohesion, we used live cell imaging of cells progressing through meiosis. We scored the percentage of cells where cohesin (Rec8-GFP) was retained at the pericentromere in anaphase I, as indicated by co-localization with the kinetochore protein Mtw1 ( Figure 1A, B). To ensure that observed effects in spo13Δ cells were not a consequence of mono-orientation loss, which partially impacts cohesion (Nerusheva et al., 2014), we simultaneously imaged mam1Δ cells for comparison. Quantification of pericentromeric Rec8 ( Figure 1C) showed that, strikingly, deletion of SPO13 leads to complete loss of cohesin in anaphase I. This is not due to impaired cohesin loading in early meiosis, since prophase I-arrested spo13Δ cells have similar levels of Rec8 on centromeres compared to wild type ( Figure 1D). We conclude that Spo13 is required for the retention of pericentromeric cohesin in anaphase I.

spo13Δ cells prematurely segregate sister chromatids
To assess sister chromatid cohesion in spo13Δ cells, we labelled one copy of chromosome V near the centromere with an array of tet operators (tetO), expressed GFP-tagged TetR repressor (Michaelis et al., 1997) and imaged CEN5-GFP foci in live meiotic cells. Upon anaphase I entry (as judged by degradation of yeast securin Pds1 (Salah & Nasmyth, 2000)), three different phenotypes may be observed, depending on whether cells successfully mono-orient sister kinetochores and protect pericentromeric cohesin (Figure 2A). In wild-type cells, a single GFP focus segregates to one of the spindle poles (as marked by the spindle pole body component Spc42-tdTomato). Alternatively, in case of defective mono-orientation, split GFP foci stay in close proximity (<2 µm) because sister chromatids are cohered by pericentromeric cohesin. Lastly, in cells lacking both mono-orientation and sister chromatid cohesion, GFP foci split over a greater distance (>2 µm). Note that, using this assay, pericentromeric cohesion loss during anaphase I can only be readily observed where it is accompanied by sister kinetochore bi-orientation. We subsequently scored the number of cells falling into either of these categories for each of the mutants analysed. This revealed that sister centromeres separate over large (>2 µm) distances in the half of spo13Δ anaphase I cells that bi-orient sister kinetochores ( Figure 2B), consistent with all cohesion being lost. A small fraction of centromeres in spo13Δ mam1Δ cells, which bi-orient almost exclusively, stay in close proximity in the 30-minute time frame measured ( Figure 2B), indicating that these cells at least temporarily retain sister chromatid cohesion. However, the loss of cohesion in all spo13Δ cells with bi-oriented kinetochores, the near-complete absence of Rec8, and the fact that deletion of SPO13 permits efficient sister chromatid segregation in most mam1Δ cells ( Figure  Restoring the second nuclear division in spo13Δ cells does not prevent chromosome missegregation We reasoned that the chromosome missegregation events seen in spo13Δ mutants might be related to the absence of the second nuclear division in these cells. Thus, restoring two Images for ΔCEN5=0µm, ΔCEN5=0-2µm and ΔCEN5>2µm were taken from wild-type, mam1Δ and spo13Δ cells, respectively. (B) Frequency of CEN5 distance categories is shown for the indicated genotypes after live-cell imaging. Wild-type (AM15190), spo13Δ (AM15118), mam1Δ (AM15119) and spo13Δ mam1Δ (AM15120) cells carrying SPC42-tdTomato, PDS1-tdTomato and heterozygous TetR-GFP foci at CEN5, were sporulated for 2.5 h before imaging on a microfluidics plate. divisions to spo13Δ cells by deletion of MAD2 (Shonn et al., 2002) would be expected to allow accurate chromosome segregation in the absence of Spo13. Our analysis of pericentromeric Rec8-GFP in anaphase I showed that, while pericentromeric cohesin in anaphase I is retained in wild-type and mad2Δ strains, it is lost to a similar degree in spo13Δ and spo13Δ mad2Δ mutants ( Figure 3A-C). Intriguingly, mad2Δ cells were frequently unable to separate kinetochores in anaphase I, despite successful cleavage of arm cohesin ( Figure 3A). While the reasons for this phenotype are unclear, we speculate that unattached kinetochores might persist into anaphase I when MAD2 is deleted.
To analyse chromosome segregation in more detail, we followed cells carrying chromosomes labelled with Htb1-mCherry and heterozygous CEN5-GFP foci through meiosis ( Figure 3D). To assess global chromosome segregation, we assayed the chromosomal content of spores by measuring the area occupied by Htb1-mCherry after meiosis II and calculated the ratio of the largest and smallest chromosomal mass in each cell. In wild-type cells, this ratio is close to 1 in most cells ( Figure 3E) since all four nuclei are expected to be of similar size. In contrast, spo13Δ mad2Δ cells show a large variation in the chromosomal content of nuclei destined for spores ( Figure 3E), indicating gross chromosome missegregation. We additionally investigated the segregation of heterozygous CEN5-GFP foci in these cells ( Figure 3F, G). Similar to spo13Δ single mutants, a large proportion of spo13Δ mad2Δ double mutant cells split sister chromatids upon the first nuclear division ( Figure 3F). Furthermore, 20% of spo13Δ mad2Δ cells display CEN5-GFP dot(s) in only one out of four spores after meiosis II ( Figure 3G). This is largely caused by the absence of Spo13, since mad2Δ single mutants display a more modest missegregation phenotype ( Figure 3G). Therefore, spo13Δ mad2Δ cells fail to accurately segregate chromosomes during both the first and second nuclear divisions. We conclude that the lack of Spo13 causes loss of centromeric cohesion during meiosis I and severe chromosome missegregation even when the second nuclear division is restored.
Sister chromatid cohesion is restored by preventing cohesin cleavage A cleavage-independent, Rad61/Wpl1-dependent, cohesin removal pathway, similar to that which occurs in mammalian mitosis, operates during prophase I of budding yeast meiosis We considered the possibility that cells lacking Spo13 lose cohesion, not due to its cleavage, but as a result of ectopic Rad61 activity. However, deletion of RAD61 did not restore cohesion to spo13Δ cells ( Figure 4A), indicating that a failure to counteract cleavage-independent cohesin removal is not solely responsible for the cohesion defect of cells lacking Spo13.
Next, we assessed whether cohesin cleavage is required for cohesion loss during anaphase I in spo13Δ cells. First, we inactivated Esp1 (separase), using the temperature-sensitive esp1-2 mutant (Buonomo et al., 2000) and followed Rec8-GFP by live cell imaging ( Figure 4B-D). As expected, cohesin remained on chromosomes even after anaphase I onset in both in esp1-2 and esp1-2 spo13Δ cells and, consequently, sister chromatid segregation was largely prevented ( Figure 4E).
Additionally, we prevented cohesin cleavage by mutating the separase cleavage site in Rec8 (Rec8-N) (Buonomo et al., 2000). We followed GFP-tagged versions of this Rec8 variant through meiosis in wild-and spo13Δ cells ( Figure 5A). Similar to esp1-2 mutants, rec8-N prevents cleavage of cohesin along the length of the chromosome in spo13Δ cells ( Figure 5B) and pericentromeric cohesin intensity is greatly increased ( Figure 5C). Furthermore, we find that Rec8-N prevented the segregation of sister chromatids in spo13Δ mutants ( Figure 5D). We conclude that cohesin cleavage is required for sister chromatid segregation in spo13Δ cells.

PP2A is functional in the absence of Spo13
Rec8 cleavage during wild-type meiosis relies on its prior phosphorylation (Brar et al., 2006; Katis et al., 2010) which is reversed in the pericentromere by PP2A. We considered the possibility that PP2A function may be impaired in spo13Δ cells, rendering it unable to dephosphorylate, and therefore protect, cohesin. We assessed whether tethering PP2A directly to cohesin could prevent Rec8 cleavage in the absence of Spo13. We fused GFPbinding protein (GBP), a nanobody specifically recognising GFP (Rothbauer et al., 2006), to the PP2A regulatory subunit Rts1 to irreversibly tether PP2A to GFP-tagged Rec8. This was sufficient to prevent cohesin removal, both in pCLB2-SGO1 and spo13Δ cells ( Figure 6A-C). To further confirm the full functionality of Rts1 in spo13Δ cells, we utilised a separase biosensor (Yaakov et al., 2012) where a cleavable Rec8 moiety is fused to GFP and LacI, with the latter allowing targeting of the biosensor to a lacO array on chromosome arms ( Figure 7A). In wild-type and spo13Δ cells, this biosensor appears as a single GFP focus in meiosis I until separase is  Figure 1C. ***p<0.001, n.s. = not significant (Welch two-sample t-test). For spo13Δ mad2Δ mutants, the analysis in (B) and (C) was performed exclusively for cells that performed two divisions (as judged by the presence of four Mtw1-tdTomato foci after meiosis II). (D-G) Severe chromosome missegregation occurs in spo13Δ mad2Δ cells. (D) Representative images of wild-type (AM24848), spo13Δ (AM24849), mad2Δ (AM25221) and spo13Δ mad2Δ (AM25222) cells carrying heterozygous TetR-GFP foci at CEN5 and HTB1-mCherry. Green arrows indicate CEN5-GFP segregation outcomes after meiosis I, cyan arrows indicate CEN5-GFP segregation outcomes after meiosis II. (E) Spores of spo13Δ mad2Δ vary greatly in the amount of nuclear DNA, as estimated by Htb1-mCherry area, thus indicating gross chromosome missegregation. The area occupied by Htb1-mCherry was measured in cells with four (wild type (n=45), mad2Δ (n=31) and spo13Δ mad2Δ (n=33)), or two (spo13Δ (n=50)) nuclear masses after meiosis II and variation in chromosomal area estimated by obtaining the ratio of the largest and smallest nuclear mass for each cell. **p<0.01, n.s. = not significant (Welch two-sample t-test). (F-G) CEN5 missegregation in spo13Δ mad2Δ cells. Segregation of heterozygous CEN5-GFP foci was scored in 50 cells after the first (F) and second (G) nuclear division in the indicated strains. For spo13Δ mad2Δ mutants, the analysis in (F) and (G) was performed exclusively for cells that performed two divisions (as judged by the presence of four distinct Htb1-mCherry signals after meiosis II). Note that while a large proportion of spo13Δ mad2Δ cells end up with CEN5-GFP foci in two separate spores after meiosis II (similar to wild type), many of these cells have already segregated sister chromosomes in meiosis I (purple stripes), rather than meiosis II (gray).

Conclusions
The successful protection of pericentromeric cohesin is a key modification to the meiotic chromosome segregation machinery   et al., 2019). Future work should focus on elucidating how Spo13 elicits its effects on kinase function, and how this might be linked to its functions in both sister kinetochore monoorientation and meiotic cell cycle control.

Yeast strains and plasmids
All strains are SK1-derivatives and are listed in Table 1. Plasmids generated in this study are listed in Table 2

Chromatin immunoprecipitation
ChIP-qPCR was performed as previously described (Vincenten  et al., 2015), using mouse anti-Ha (12CA5, Roche). All parameters and equipment are identical to those described previously, including qPCR mixes and thermocycling conditions. Primers for qPCR analysis are listed in Table 3.
Live cell imaging Live cell imaging was performed on a DeltaVision Elite system (Applied Precision) connected to an inverted Olympus IX-71 microscope with a 100x UPlanSApo NA 1.4 oil lens. Images were taken using a Photometrics Cascade II EMCCD camera. The Deltavision system was controlled using SoftWoRx software, version 5.5. Live-cell imaging for Figure 3 was performed on a Zeiss Axio Observer Z1 (Zeiss UK, Cambridge) equipped with a Hamamatsu Flash 4 sCMOS camera, Prior motorised stage and Zen 2.3 acquisition software.
Cells were imaged at 30˚C (unless stated) on an ONIX microfluidic perfusion platform by CellASIC. Cells were pre-grown in culture flasks for ~3 h before transfer to microfluidics plates.
Imaging began about 30 min later with images being acquired every 15 min for 12-15 h. Seven z-stacks were acquired with 0.85µm spacing. Image panels were assembled using Image-Pro Premier 3D, version 9.1 (Media Cybernetics). Images were analysed using ImageJ 1.48v (National Institutes of Health). Final image assembly was carried out using Adobe Photoshop CS5.1 and Adobe Illustrator CS5.1. Rec8-GFP intensities were measured using the DV_DotCounter custom plugin for ImageJ (Kelly, 2019a). The plugin applied a Z projection to each colour channel and allowed the user to select a cell of interest. Kinetochores in the red channel were identified by Yen Auto Threshold (Yen et al., 1995) and their XY central coordinates, mean intensity and area recorded. The coordinates were then used to measure mean intensity in the corresponding location in the green channel, equivalent to pericentromeric Rec8-GFP. In experiments where pericentromeric cohesin was likely to be found in between kinetochores (which is thought to occur in cells that bi-orient in meiosis I but retain cohesin), the XY coordinates in the red channel were used to generate a line profile between the 2 kinetochores in both colour channels over   Table 2. Plasmids generated in this study.
AMp1368 YIplac128-rec8-N-GFP LEU2 integration plasmid carrying rec8-N-GFP. Table 3. qPCR primers used in this study. For distances from centromeres, "-" indicates the location is upstream of the centromere, whereas "+" indicates the location is downstream of the centromere. exactly the same pixels. The two brightest peaks in the line profile of the green channel were calculated to give the maximum intensity value for each. Rec8-GFP intensity was measured in this manner for Figure 4C and Figure 5C. The plugin used was the custom YeastLineProfiler for ImageJ (Kelly, 2019b). Chromosomal area in Figure 3E was measured using a custom ImageJ plugin (Kelly, 2019c) that identifies the regions of bright fluorescence in the red channel using Yen Auto Threshold and subsequently measures the area of these regions of interest.

Chr. Location
An earlier version of this article can be found on bioRxiv ( by separase, therefore the pool of cohesin bound to centromeric regions must be protected from phosphorylation during the first meiotic division to prevent premature loss of cohesion. Galander et al. use in vivo imaging to investigate cohesin protection in budding yeast, focusing on Spo13, which role in this process remains poorly understood.
The authors show convincingly that spo13 mutants display premature loss of Rec8 and sister chromatid cohesion during the first meiotic division, that this premature loss of cohesion requires separase but not Wapl, and that expression of a separase-resistant Rec8 rescues cohesin loss in spo13 mutants. These results demonstrate an important role for Spo13 in preventing separase-dependent Rec8 removal during meiosis I.
Specific comments: Figure 4 shows that expression of Rec8-N (separase resistant) prevents loss of Rec8 from pericentromeric regions in spo13 mutants, but despite this, sister centromeres still show substantial separation in ~50% of the cells. How can sister centromeres achieve this level of separation despite extensive Rec8 binding?
The introduction doesn't mention the Challa et al 2019 paper describing the role of Wapl in promoting Rec8 removal before anaphase I. Since distinguishing the contribution of the Wapl and separase pathways to the cohesion defects of spo13 mutants is a key aspect of the manuscript, mentioning the cohesion in meiosis I. Previous work has shown that Spo13 affects the disposition of cohesin at centromeres but the molecular basis for the modulation of centromeric cohesion by Spo13 has remained unclear. Recent work has shown that there are two pathways for cohesin removal in meiosis in budding yeast. Here the authors test which pathway is impacted by Spo13. The experiments have moved the field forward by using live cell imaging methods to address this question. The results demonstrate that Spo13, at least in part, protects Rec8 at the centromeres from cleavage by separase.

Introduction
Para. 3, line 7. Might be good to adjust the sentence saying Rec8 cleavage is dependent on Cdc5. The next sentence indicates the lack of clarity on this point.

Results and Discussion
Page 4, second column, 4 lines from bottom: "withbi" -N mutants. This shows centromeres are more able to separate in spo13 mutants even without Rec8 cleavage. There are multiple possible explanations for these results. Is it because sister centromeres are more easily bi-oriented in spo13 mutants? Alternatively, could it be that Spo13 also promotes sister centromere cohesion also protects pericentromeric cohesion through a pathway that doesn't involve cleavage? The manuscript would benefit from brief comments from the authors on the implications of these observations.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results?  Thanks for pointing out this omission, we have added the suggested text and references.

Results and Discussion
Page 4, second column, 4 lines from bottom: "withbi" This has been corrected Fig. 3 E/Fig. 4 D -spo13 delete allows meiosis I sister centromere separation in esp1 and REC8-N mutants. This shows centromeres are more able to separate in spo13 mutants even without Rec8 cleavage. There are multiple possible explanations for these results. Is it because sister centromeres are more easily bi-oriented in spo13 mutants? Alternatively, could it be that Spo13 also promotes sister centromere cohesion also protects pericentromeric cohesion through a pathway that doesn't involve cleavage? The manuscript would benefit from brief comments from the authors on the implications of these observations. We believe that this is due to the fact that sister centromeres are bioriented in mutants. We spo13 have added a few sentences explaining these observations and the previous evidence that shows that defects in monoorientation result in sister centromere splitting even without loss of cohesion.

3.
However, our main conclusion from these experiments is that PP2A-Rts1 is functional in the absence of Spo13 and we believe are data supports this interpretation.
We also respectfully disagree with the conclusion in Mehta et al. based on the available data. Their conclusion that Spo13 acts independently of Rts1 was derived from data presented in Figure 3 of their paper where they showed that foci segregate to opposite poles during anaphase I CEN5-GFP in more than cells. However, these genotypes are not complete: the methods spo13Δ rts1Δ spo13 section and strain table reveal that all cells (but none of the other mutations) also carry spo13Δ delete to allow the cells to go through two meiotic divisions. Therefore, this experiment is mad2Δ not properly controlled, especially because cells themselves are compromised in mad2 chromosome segregation (Figure 3 in our current manuscript).
In conclusion, while Rts1-independent Rec8 removal by Spo13 remains a possibility, there is currently no evidence for it.
The typo on Page 4 has been corrected.
No competing interests were disclosed. Competing Interests: