Esp1 is phosphorylated by Cdk1 and dephosphorylated by PP2A^Cdc55

In vivo metabolic labeling with ³²P-orthophosphate followed by immunoprecipitation of a tagged Esp1 revealed that Esp1 is phosphorylated in vivo in mitotically arrested cells (Fig 1A). This phosphorylation depends on Cdk1 activity, because after induction of the yeast Wee1 kinase (GAL-SWE1), a treatment that inhibits Cdk1 but maintains a mitotic arrest [45], Esp1 phosphorylation is reduced.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Cdk1^Clb2 phosphorylates Esp1.

(A) Esp1 is phosphorylated in vivo. ESP1, ESP1-myc13 and pGAL-SWE1 ESP1-13myc cells were grown in YEP + raffinose, arrested in mitosis with nocodazole and switched to YEP + galactose media to induce expression of Swe1. After 1 h, cells were washed in medium lacking phosphate, and grown for 30 min in the presence of [³²P]orthophosphate. Esp1-13myc was immunoprecipitated with 9E10 antibody, run on a polyacrylamide gel and exposed to a phosphorimager screen or immunoblotted. (B) Esp1 contains six minimal Cdk1 consensus sites (S/TP): S13 and T16 (termed N-terminal), T1014, S1027 and T1034 (termed central) and S1280 (termed C-terminal). (C) Mutating Esp1 central residues prevents Esp1 phosphorylation in vivo. Left panels, ESP1, esp1-2A, esp1-3A, esp1-1A, esp1-3A+2A, esp1-2A+1A, esp1-3A+1A and esp1-2A+3A+1A cells were grown in YEP + dextrose, arrested in mitosis with nocodazole and labeled with [³²P]orthophosphate as described in (A). Right panels, wild-type, cdc55Δ and rts1Δ cells were grown in YEP + dextrose, arrested in G1 with α-factor and released into the cell cycle in the presence of nocodazole. After 90 min cells were labeled with [³²P]orthophosphate as described in (A). Esp1 was immunoprecipitated with anti-Esp1 antibody, run on a polyacrylamide gel and exposed to a phosphorimager screen or immunoblotted. (D) Wild-type and esp1-3A cells were grown to log phase, arrested in G1 with α-factor, and released into the cell cycle (t = 0). α-factor was re-added at t = 60 min to arrest cells in the following G1. Samples were taken for immunoblotting at the indicated timepoints, and run on a polyacrylamide gel containing Phos-tag reagent (top panel), or a standard polyacrylamide gel (bottom panels) and immunoblotted with the indicated antibodies. Note that running and transferring of Phos-tag polyacrylamide gels is inconsistent and cell cycle-dependent changes in protein abundance observed in these panels may not accurately reflect changes in protein abundance. The Esp1 panels from the standard polyacrylamide gel more accurately reflect cell cycle changes in Esp1 abundance. (E) Cdk1^Clb2 phosphorylates the central region of Esp1 in vitro. Esp1 was immunoprecipitated from the strains in (C) growing asynchronously, incubated with γ-[³²P]ATP and purified Cdk1^Clb2-CBP, washed, run on a polyacrylamide gel, and exposed to a phosphorimager screen or immunoblotted with anti-Esp1 antibody. (F) PP2A^Cdc55 dephosphorylates Esp1 in vitro. Esp1 was immunoprecipitated from wild-type cells and phosphorylated with purified Cdk1^Clb2-CBP and γ-[³²P]ATP while immobilized on IgG-coupled magnetic beads. The beads were washed and incubated for the indicated times at room temperature with no addition (yellow lines), TAP-purified PP2A^Cdc55 (blue lines), or PP2A^Cdc55 and okadaic acid (OA) (red lines). The three reactions share a t = 0 sample that was taken before the additions. The dephosphorylation of Esp1 was quantified on a phosphorimager and the extent of dephosphorylation relative to t = 0 (average ± SEM) was graphed. The experiment shown is representative of one of three repeats.

https://doi.org/10.1371/journal.pgen.1007029.g001

Esp1 has six minimal Cdk1 phosphorylation sites (S/TP) distributed into three groups; two sites near the N-terminus (termed ‘N-terminal’), three sites near the N-terminal end of the protease domain (termed ‘central’) and a single site close to the C-terminus (termed ‘C-terminal’) (Fig 1B). To determine if these sites are phosphorylated by Cdk1, we mutated each group individually, and in combination, to unphosphorylatable alanine residues (S/T to A). These Esp1 mutants are all expressed at endogenous levels (S1A Fig), interact with Pds1 normally (S1B Fig) and support viability in otherwise wild-type cells. We also created a set of Esp1 mutants that substitute two phospho-mimetic aspartic acid residues at each potential phosphorylation site (SP/TP to DD). These Esp1 mutants are also expressed at endogenous levels and support full viability in otherwise wild-type cells (S1C Fig).

In vivo metabolic labeling using ³²P-orthophosphate demonstrates that Esp1 phosphorylation is lost only in cells lacking the three central Cdk1 sites (Fig 1C), suggesting that Cdk1 phosphorylates Esp1 in the central region. Esp1 with the central phosphorylation sites mutated to alanine migrates more quickly than wild-type Esp1 or other alanine-substituted Esp1 mutants (S1A Fig), while mutating these central sites to aspartate slightly retards Esp1 mobility (S1C Fig).

To confirm this phosphorylation, lysates from cells synchronized in G1 and released into the cell cycle were examined on Phos-Tag polyacrylamide gels that retard the mobility of phosphorylated proteins [47,48]. Wild-type Esp1 is phosphorylated on at least two residues following release from G1, and this phosphorylation peaks at the same time as the mitotic B-type cyclin, Clb2, when both Pds1 and Mcd1 levels are falling during anaphase (Fig 1D). Esp1 phosphorylation is detected 30 minutes after release from G1, before Clb2 appearance, suggesting that other mitotic cyclins, like Clb5, may also phosphorylate Esp1 in vivo. No mobility shift on the Phos-Tag gel is detected in the esp1-3A mutant, in which the three central Cdk1 sites are mutated, and the protein resolves as a single unphosphorylated form, demonstrating that the mobility shift of phosphorylated Esp1 depends on these sites.

To determine if Cdk1 can phosphorylate these residues directly, we incubated immunoprecipitated wild-type and mutant Esp1 proteins with purified Cdk1^Clb2 and Cdk1^Clb5 complexes in an in vitro phosphorylation reaction (Fig 1E and S1D Fig). Both Cdk1^Clb complexes phosphorylate Esp1 in vitro and mutation of the three central sites prevents Cdk1^Clb2 phosphorylation. No phosphorylation of Esp1 is observed when Cdk1^Clb2 or Cdk1^Clb5 is omitted from these reactions, demonstrating this phosphorylation is not due to a co-precipitated kinase (S1D and S1E Fig). Taken together, these data show that Cdk1 phosphorylates Esp1 in vivo and in vitro on sites in the central region of the protein.

Purified PP2A^Cdc55 dephosphorylates several Cdk1 substrates in budding yeast [45,49], and is also able to dephosphorylate immunoprecipitated Esp1 that has been phosphorylated in vitro by purified Cdk1^Clb2 (Fig 1F). Mitotic cdc55Δ cells also display increased in vivo phosphorylation of Esp1 relative to wild-type cells, and relative to cells deleted for the second PP2A regulatory subunit in yeast, RTS1, the homologue of the vertebrate B56 subunit of PP2A (Fig 1C).

Depletion of Pds1 in ESP1-3D cells causes premature spindle elongation

esp1-3A and ESP1-3D cells exhibit no apparent differences in cell cycle progression, the timing of anaphase onset, or the behaviour of the mitotic spindle as compared to wild-type cells (S1F–S1I Fig), suggesting that if Esp1 phosphorylation regulates anaphase it may act redundantly with Pds1 function.

We therefore constructed strains expressing an auxin-inducible degron tagged PDS1 (PDS1-AID) and the rice, Oryza sativa, F-box protein Tir1 (OsTir1) [50]. Treating these cells with the plant hormone auxin (indole-3-acetic acid) causes rapid degradation of Pds1 to 10–20% of its normal levels (S2A Fig). This reduction inhibits the growth of PDS1-AID cells on plates containing auxin (Fig 2A). We observe that the esp1-3A mutant, which prevents Esp1 phosphorylation, partially suppresses the growth defect caused by Pds1 depletion. In contrast, the ESP1-3D allele, which mimics Esp1 phosphorylation, exacerbates this defect (Fig 2A). These results suggest a model in which Pds1 and phospho-regulation of Esp1 work redundantly to regulate the essential function of Esp1, with destruction of Pds1 and phosphorylation of Esp1 both acting to activate Esp1, and preventing phosphorylation, in the esp1-3A allele, inhibiting Esp1 activity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Pds1 depletion causes immediate spindle elongation in a phospho-mimetic Esp1 mutant.

(A) Pds1 depletion causes synthetic sickness in ESP1-3D cells. Eight-fold serial dilutions of the indicated strains were spotted onto the indicated plates and grown at 25°C. (B) The ESP1-3D allele is semi-dominant in diploids. Ten-fold serial dilutions of the indicated diploids were grown in YEP + dextrose media, spotted onto the indicated plates and grown at 25°C. (C) The ESP1-3D allele is semi-dominant when complemented by a plasmid borne ESP1 (ESP1-CEN-HIS3). Ten-fold serial dilutions of the indicated strains were grown in media lacking histidine, spotted onto the indicated plates and grown at 25°C. (D) Mitotic spindle morphology of individual ESP1-3D cells depleted of Pds1. PDS1-AID SPC42-eGFP and ESP1-3D PDS1-AID SPC42-eGFP cells were grown at 25°C to log phase and arrested in G1 with α-factor. 30 min before α-factor release +/- auxin was added. Cells were released at t = 0 and at t = 25 min cells were plated onto YPD live microscopy pads +/- auxin and Spc42-eGFP was imaged every minute. Each strain was imaged at least two times in each condition. Cells that undergo normal metaphase spindle formation are shown in black. Cells that undergo immediate spindle elongation upon spindle formation are shown in green. Cells that exhibit failed or no anaphase spindle elongation are shown in red. See S3 Fig for cell traces of experiments done in the absence of auxin and Fig 3 for tabulation of all imaging data. The scoring metric is described in the text and in the material and methods.

https://doi.org/10.1371/journal.pgen.1007029.g002

These interactions are partially dominant when tested in diploids that express ESP1-3D or esp1-3A and one wild-type copy of ESP1 (Fig 2B), or in haploid cells that contain wild-type ESP1 on a low-copy centromeric plasmid (Fig 2C). Simply increasing the copy number of wild-type ESP1 in PDS1-AID cells causes a greater growth defect on auxin, which suggests that the dominant phenotypes of ESP1-3D and esp1-3A are caused by an increase or decrease of Esp1 activity, respectively (Fig 2C).

We observe similar negative genetic interactions between pds1Δ and ESP1-3D cells (S2B and S2C Fig), but were surprised that esp1-3A is synthetically sick when combined with pds1Δ, the opposite phenotype to that observed in esp1-3A PDS1-AID cells (Fig 2A and S2C Fig). We observe additional differences between pds1Δ and PDS1-AID cells (presented below), and hypothesize that they are caused by the requirement of Pds1 for effective nuclear import and activation of Esp1 [20]. Unlike pds1Δ cells that produce no Pds1 protein, PDS1-AID cells treated with auxin retain some Pds1-AID protein (S2A Fig), suggesting that this residual protein may be sufficient to fulfill Pds1’s positive function on Esp1, but insufficient to fully inhibit Esp1 activity after mitotic entry.

To understand the lethality of ESP1-3D PDS1-AID cells we analyzed mitotic progression using live cell imaging of an endogenously tagged spindle pole body (SPB) protein (SPC42-eGFP) and directly measured the length of SPB separation as a correlate for spindle elongation ([45]; Fig 2D). The formation of a short mitotic spindle (1–2 μm in length), caused by the rapid separation of SPBs, marks entry into mitosis. Wild-type cells spend an average of 22.4 minutes with a short mitotic spindle before SPBs and sister centromeres rapidly separate at the onset of anaphase (S2D Fig). In the presence of auxin, both PDS1-AID and ESP1-3D PDS1-AID cells display striking defects in spindle elongation, with many cells undergoing continual elongation of the spindle as soon as SPB separation occurs (Fig 2D).

To categorize the behaviour of these cells we quantified spindle dynamics: cells whose spindles do not elongate to 2 and 2.5 μm within the first 10 and 15 minutes, respectively, are classified as “normal metaphase spindle formation” (or “normal”; black traces), and those that elongate beyond 2 and 2.5 μm within these time-intervals are classified as “immediate spindle elongation” (or “immediate”; green traces) (Fig 3). We used these two rules to identify cells that are difficult to score, and among all cells treated with auxin (Fig 3), 15% of the cells produce conflicting scores (i.e., immediate/normal or normal/immediate in the 10/15 minute intervals). These cells were manually curated (see Materials and Methods for details). Of the wild-type, pds1Δ, and PDS1-AID cells not treated with auxin, 97% score as “normal” (Fig 3, S2D and S3 Figs), showing this metric differentiates between auxin-treated and -untreated cells. In a small number of cells the spindle fails to elongate to 6 μm in the 60 minutes of observation, and these are scored as “failed anaphase” (red traces).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Spindle characteristics during mitosis.

Cells with “normal metaphase spindle formation” do not elongate their spindle more than 2 μm in the first 10 minutes after SPB separation and more than 2.5 μm in the first 15 minutes after SPB separation. Cells whose spindles elongate more than 2 and 2.5 μm in these time-intervals display “immediate spindle elongation.” Cells with “failed anaphase” don’t elongate their spindles to 6 μm in the 60 minutes after SPB separation. A small number of cells (15% for cells treated with auxin, and 11% for untreated cells) produce conflicting scores using these two rules (i.e., immediate/normal or normal/immediate in the 10/15 minute intervals), and we categorized these cells manually (see Materials and Methods for details).

https://doi.org/10.1371/journal.pgen.1007029.g003

Of the PDS1-AID cells treated with auxin, 40% undergo immediate spindle elongation, and this number increases to 70% of ESP1-3D PDS1-AID cells (Fig 2D). The rate of spindle elongation varies among these “immediate” cells, but it is faster than the rate of spindle elongation during metaphase in wild-type cells (which lengthens from ~ 1 to 2 μm during 20 minutes), and slower than the rapid anaphase elongation that occurs at the metaphase-to-anaphase transition in untreated cells (compare “immediate” cells in Fig 2D to “normal” cells in S3 Fig). A subset of “immediate” cells also display uncharacteristic shortening of the spindle after a period of continual spindle lengthening. These behaviours are not observed in untreated ESP1-3D PDS1-AID or PDS1-AID cells, which undergo spindle formation, maintain a short metaphase spindle and then abruptly initiate anaphase spindle elongation in a manner indistinguishable from wild-type cells (S2D and S3 Figs). Together these results suggest that the combination of mimicking phosphorylation on Esp1 and reducing Pds1 allows precocious anaphase spindle elongation which may explain their synthetic effects on viability.

PDS1-AID cells grown in the presence of auxin have strikingly different behaviour than pds1Δ cells which undergo relatively normal spindle formation and then delay in initiating anaphase spindle elongation (S2D Fig). This difference, like the differences in the viability of pds1Δ esp1-3A and PDS1-AID esp1-3A (Fig 2A and S2C Fig), may be caused by residual Pds1-AID protein in PDS1-AID cells (S2A Fig).

Depletion of Pds1 in cdc55Δ cells causes premature anaphase onset

Past work has shown that cells lacking both Pds1 and Cdc55 are inviable and undergo premature Mcd1 cleavage and anaphase onset [46]. Changes in Esp1 phosphorylation and activity could explain this phenotype. cdc55Δ PDS1-AID cells are inviable when grown in the presence of auxin (Fig 4A) and 88% of the cells undergo immediate spindle elongation (Figs 3 and 4B), a phenotype more severe than observed in ESP1-3D PDS1-AID cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Pds1 depletion in cdc55Δ cells induce premature anaphase onset and chromosome mis-segregation.

(A) cdc55Δ PDS1-AID cells are inviable on media containing auxin. Eight-fold serial dilutions of the indicated strains were spotted onto the indicated plates and grown at 25°C. (B) Mitotic spindle morphology of individual cdc55Δ cells depleted of Pds1. cdc55Δ PDS1-AID SPC42-eGFP and swe1Δ cdc55Δ PDS1-AID SPC42-eGFP cells were imaged as in Fig 2D. See S3 Fig for cell traces of experiments done in the absence of auxin, and Fig 3 for tabulation of all imaging data. (C) Pds1 depletion in cdc55Δ cells cause premature sister chromatid separation. PDS1-AID, cdc55Δ PDS1-AID and ESP1-3D PDS1-AID cells containing ura3::240lacO pCUP1-eGFP-lacI SPC42-mCherry were imaged as in Fig 2D. Data from individual experiments were combined for subsequent analysis (PDS1-AID–auxin [n = 17], PDS1-AID + auxin [n = 26], cdc55Δ PDS1-AID + auxin [n = 14], and ESP1-3D PDS1-AID [n = 19]). The length of time between spindle formation and sister chromatid separation was determined for each imaged cell. Cells in which sister chromatids segregated to the same pole during anaphase were characterized as failed and analyzed separately. If sister chromatid separation preceded spindle formation, this length of time was defined as 0. All measured times are displayed with the population average (± SEM). The time between spindle formation and sister separation in cdc55Δ cells is significantly different from auxin-treated and untreated PDS1-AID cells (Student’s t-tests; p < 0.05), and in auxin-treated PDS1-AID cells is significantly different from untreated PDS1-AID cells (Student’s t-test; p < 0.05). (D) An example of one cdc55Δ PDS1-AID cell imaged in (C). Inter-spindle pole body (spindle length; red squares) and inter-sister chromatid (green diamonds) distance is graphed, and in this cell, sister chromatid separation preceded spindle formation, and both sisters segregated to the same pole. (E) Pds1 depletion in cdc55Δ cells cause failed sister chromatid segregation. Segregation of sister chromatids was monitored in the cells imaged in (C). Cells in which sister chromatids segregated to the same pole during anaphase were characterized as failed. Cells in which sister chromatids segregated to the separate poles, regardless of timing were characterized as accurate. PDS1-AID–auxin (n = 17), PDS1-AID + auxin (n = 36), cdc55Δ PDS1-AID + auxin (n = 27).

https://doi.org/10.1371/journal.pgen.1007029.g004

PP2A^Cdc55 has been shown to inhibit Swe1, the budding yeast Wee1 kinase, and activate Mih1, the budding yeast Cdc25 phosphatase [49,51,52]. Swe1 phosphorylates and Mih1 dephosphorylates a conserved tyrosine on Cdk1 (Y19 on yeast Cdk1) that when phosphorylated inhibits Cdk1 activity before and during mitosis. cdc55Δ mutants have increased inhibitory tyrosine 19 phosphorylation on Cdk1 [52], so we tested if the immediate spindle elongation observed in cdc55Δ mutants depends on SWE1 and Cdk1 inhibition. cdc55Δ swe1Δ PDS1-AID cells are inviable on media containing auxin (Fig 4A), and 85% of the analyzed cells undergo immediate spindle elongation (Fig 4B). A similar percentage of cells undergo immediate spindle elongation in the absence or presence of SWE1 showing the immediate spindle elongation in cdc55Δ PDS1-AID cells is not caused by increased inhibitory tyrosine phosphorylation on Cdk1, but through a different mechanism. However, spindle elongation in cdc55Δ swe1Δ PDS1-AID cells is less variable, with few dramatic spindle shortening events, suggesting that sustained high levels of inhibitory phosphorylation on Cdk1 in cdc55Δ cells affects spindle dynamics.

Immediate spindle elongation may be caused by premature separation of sister chromatids and a similar phenotype has been observed in fixed cdc55Δ cells with reduced Pds1, and in mcd1 and mcm21 mutants [46,53,54]. Additionally, spo11Δ diploids, which lack meiotic recombination between homologous chromosomes, initiate anaphase I of meiosis prematurely [55]. We observed sister chromatid cohesion and SPB movement directly using a lacO array on the arm of chromosome V at the URA3 locus in cells expressing lacI-GFP and Spc42-mCherry (Fig 4C and 4D). In cdc55Δ PDS1-AID cells treated with auxin, the lacO array separated, on average, 6.71 minutes after SPB separation compared to 13.85 minutes in auxin treated PDS1-AID control cells. In some cdc55Δ PDS1-AID cells, the lacO arrays separate prior to SPB separation, an event not observed in controls (Fig 4C and 4D). Consistent with our observation that many PDS1-AID cells treated with auxin display immediate spindle elongation, these cells also advance the timing of chromosome V separation compared to untreated cells (13.85 vs. 17.65 minutes). Although ESP1-3D PDS1-AID cells display more severe spindle elongation defects than control PDS1-AID cells (Fig 2D), the timing of chromosome V separation is not significantly different in these two strains (15.89 vs. 13.85 minutes; Fig 4C).

The advanced timing of sister separation in cdc55Δ PDS1-AID cells may cause severe chromosome segregation defects. When we follow the fate of the lacO arrays during anaphase, cdc55Δ PDS1-AID cells treated with auxin segregate the lacO arrays randomly to the two daughter cells (Fig 4E) and this defect may explain the potent lethality observed in these cells after auxin treatment. Although control PDS1-AID cells also show high rates of chromosome mis-segregation, their defect is less severe. Importantly, in PDS-AID and cdc55Δ PDS1-AID cells both chromatids remain attached to a SPB (though often the same one), suggesting that neither the immediate spindle elongation, nor the chromosome segregation defects, are caused by a failure in kinetochore attachment to microtubules.

Blocking phosphorylation of Esp1 and Mcd1 does not suppress cdc55Δ defects

Because PP2A^Cdc55 can dephosphorylate Esp1 in vitro (Fig 1F) and deletion of CDC55 increases Esp1 phosphorylation in vivo (Fig 1C), we tested whether blocking Esp1 phosphorylation, in the esp1-3A mutant, suppresses the lethality of cdc55Δ PDS1-AID. Although esp1-3A partially suppresses the growth defect of PDS1–AID (Fig 2A), we see no suppression in esp1-3A cdc55Δ PDS1-AID or esp1-3A cdc55Δ swe1Δ PDS1-AID cells (Fig 5A). Additionally, esp1-3A has no impact on the immediate spindle elongation we observe in cdc55Δ swe1Δ PDS1-AID cells treated with auxin (S3 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. mcd1-10A and esp1-3A do not suppress cdc55Δ phenotypes.

(A) esp1-3A, mcd1-10A and esp1-3A mcd1-10A do not rescue cdc55Δ cells depleted of Pds1. Ten-fold serial dilutions of the indicated strains were spotted onto the indicated plates and grown at 25°C. (B) Cdc14 is released at a spindle length of approximately 2 μm in swe1Δ cdc55Δ cells regardless of Pds1 depletion. swe1Δ cdc55Δ PDS1-AID CDC14-eGFP SPC42-mCherry cells were grown at 25˚C to log phase and arrested in G1 with α-factor. 30 min before α-factor release +/- auxin was added. Cells were released at t = 0 and at t = 90 min samples were fixed for microscopy. The distance between spindle pole bodies was measured in each cell. Each cell was categorized as pre-mitotic (a single Spc42-mCherry focus), short spindle (Spc42-mCherry foci separated by < 2μm) or long spindle (Spc42-mCherry foci separated by > 2μm). In each cell Cdc14 was characterized as nucleolar or released qualitatively. The proportion of cells displaying nucleolar and released Cdc14 at each cell cycle stage relative to spindle length was calculated. swe1Δ cdc55Δ PDS1-AID–auxin (n = 88), swe1Δ cdc55Δ PDS1-AID + auxin (n = 102). All raw images are shown in S4A Fig.

https://doi.org/10.1371/journal.pgen.1007029.g005

In budding yeast, Cdc5 (the budding yeast Polo kinase) phosphorylation of Mcd1 promotes its cleavage, and this phosphorylation is essential in the absence of Pds1 [30,56]. PP2A^Cdc55 has also been shown to dephosphorylate Mcd1 in vitro, and deletion of CDC55 increases phosphorylation of Mcd1 in vivo [57]. We therefore tested whether phosphorylation of both Esp1 and Mcd1 work redundantly to promote sister chromatid separation. Mutation of ten phosphorylation sites in Mcd1, in the mcd1-10A mutant [56] (note that the mcd1-10A mutation has been named scc1-10A in previous reports, but for clarity we use the standard name (http://www.yeastgenome.org/locus/S000002161/overview)), suppresses growth defects of PDS1-AID cells treated with auxin, but neither mcd1-10A, nor the double mutant esp1-3A mcd1-10A, suppress the lethality of cdc55Δ PDS1-AID cells (Fig 5A).

Cdk1-dependent phosphorylation of the APC, which targets Pds1 for degradation, is also regulated by PP2A^Cdc55 in vivo and in vitro, and mutation of twelve Cdk1 sites on three APC subunits can partially suppress the SAC defect of cdc55Δ cells [45]. Combining these APC mutations with esp1-3A and mcd1-10A does not increase this suppression (manuscript in preparation).

PP2A^Cdc55 and Esp1 also function in the Cdc Fourteen Early Anaphase Release (FEAR) pathway [58,59]. Though not essential, this pathway promotes release of the Cdc14 phosphatase from the nucleolus early in anaphase to activate the essential mitotic exit network (MEN) in late anaphase. Cdc14 release from the nucleolus in early anaphase has been proposed to be an important trigger of anaphase onset [60–62]. We therefore examined if the lethality and premature spindle elongation of cdc55Δ swe1Δ PDS1-AID cells correlate with earlier Cdc14 release from the nucleolus. To carefully monitor any changes in FEAR activation we correlated release of Cdc14-GFP to spindle length (S4A Fig) and found that early degradation of Pds1-AID does not cause Cdc14 release at shorter spindle lengths (Fig 5B). This result suggests that premature FEAR activation is not responsible for the lethality, and the premature spindle elongation and sister chromatid separation in cdc55Δ PDS1-AID cells. Supporting this data, deletion of SPO12, a component of the FEAR pathway, delays Cdc14 release [58], but does not suppress the growth defects of cdc55Δ PDS1-AID or ESP1-3D PDS1-AID cells when grown on auxin (S4B and S4C Fig).

In conclusion, we find no evidence that the premature spindle elongation and sister chromatid separation in cdc55Δ PDS1-AID cells are caused by increased phosphorylation on Esp1, Mcd1 and the APC, or by premature activation of the FEAR network.

Depletion of Pds1 in slk19Δ cells causes premature anaphase onset

While examining whether disruption of the FEAR pathway suppresses the lethality of cdc55Δ PDS1-AID cells (S4B and S4C Fig), we tested mutants in SLK19, a spindle-associated protein that is also a component of the FEAR pathway and a substrate of Esp1 [23,58]. Unlike spo12Δ cells, which are insensitive to Pds1 depletion, slk19Δ cells are as sensitive to Pds1 depletion as cdc55Δ cells (Fig 6A and 6B), and when combined with cdc55Δ PDS1-AID or ESP1-3D PDS1-AID do not suppress their growth defects. High throughput synthetic lethal screening has also identified synthetic interactions between pds1Δ and slk19Δ [63,64].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Pds1 depletion in slk19Δ cells induce premature anaphase onset and chromosome mis-segregation.

(A) slk19Δ PDS1-AID cells are inviable in media containing auxin, and slk19Δ does not interact genetically with ESP1-3D. Eight-fold serial dilutions of the indicated strains were spotted onto the indicated plates and grown at 25°C. (B) slk19Δ does not interact genetically with cdc55Δ. Eight-fold serial dilutions of the indicated strains were spotted onto the indicated plates and grown at 25°C. (C) Mitotic spindle morphology of individual slk19Δ cells depleted of Pds1. slk19Δ PDS1-AID SPC42-eGFP cells were imaged as in Fig 2D. See S3 Fig for cell traces of experiments done in the absence of auxin, and Fig 3 for tabulation of all imaging data. (D) cdc55Δ and slk19Δ cells mis-segregate chromosomes. The indicated strains containing pCUP1-eGFP-lacI ura3::240lacO SPC42-mCherry were grown at 25°C to log phase and arrested in G1 with α-factor. 30 min before α-factor release auxin was added. Cells were released at t = 0 and at t = 120 min samples were fixed for microscopy. Telophase cells were identified using the Spc42-mCherry signal. The presence of GFP-lacI foci at either both spindle poles or at a single spindle pole was scored (n = 100 for each strain).

https://doi.org/10.1371/journal.pgen.1007029.g006

Because slk19Δ cells are sensitive to reduced Pds1 protein, we imaged slk19Δ PDS1-AID cells in mitosis. Similar to cdc55Δ and ESP1-3D cells, depletion of Pds1 in slk19Δ cells cause 95% of cells to undergo immediate spindle elongation (Figs 3 and 6C).

slk19Δ PDS1-AID cells also have a severe defect in the segregation of a lacO array integrated on chromosome V (Fig 6D). In this experiment, performed on fixed cells, we score whether sister lacO arrays segregate to opposite poles or to the same pole 120 minutes after release from a G1 arrest, when most cells have completed anaphase. Segregation of chromosome V is nearly random in slk19Δ PDS1-AID cells (Fig 6D), while cdc55Δ PDS1-AID cells mis-segregate chromosome V in 25% of divisions, a defect less severe than our measurements of chromosome V mis-segregation in live cells (Fig 4E). PDS1-AID and ESP1-3D PDS1-AID cells have less severe defects, mis-segregating chromosome V in 10% of divisions.

Premature spindle elongation is accompanied by advanced Mcd1 proteolysis and Slk19 cleavage

To determine if premature spindle elongation in ESP1-3D PDS1-AID, cdc55Δ swe1Δ PDS1-AID and slk19Δ PDS1-AID cells is caused by premature activation of Esp1, we monitored cleavage of the Esp1 substrates Mcd1 and Slk19 by western blot following release from a G1 arrest. In the absence of auxin, control, ESP1-3D, cdc55Δ swe1Δ and slk19Δ cells show similar kinetics of Pds1 and Mcd1 proteolysis and Slk19 cleavage (Fig 7). The behaviour of these cells differs very little from wild-type cells without PDS1-AID. In the presence of auxin, little Pds1 accumulates as cells transition through the cell cycle, and Mcd1 proteolysis occurs 10 to 20 minutes earlier in control PDS1-AID cells (beginning at 90 minutes following G1 release compared to 110 minutes). In cdc55Δ swe1Δ PDS1-AID and slk19Δ PDS1-AID cells, Mcd1 proteolysis occurs an additional 10 to 20 minutes earlier (80 and 70 minutes after G1 release, respectively) (Fig 7A and 7B). Although Mcd1 proteolysis may initiate slightly earlier in ESP1-3D PDS1-AID cells, the kinetics of Mcd1 disappearance is very similar to PDS1-AID cells. In addition to changes in Mcd1 proteolysis, very little full length Slk19 accumulates prior to mitosis and its cleavage is advanced relative to cells grown in the absence of auxin (Fig 7A). This difference is very similar in all cells examined, and the defect occurs early in the cell cycle, suggesting that Pds1 inhibition of Esp1 normally allows full-length Slk19 to accumulate.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Pds1 depletion in cdc55Δ swe1Δ and slk19Δ cells induce premature Slk19 cleavage and Mcd1 proteolysis.

(A) The indicated strains were grown to log phase, arrested in G1 with α-factor, and 30 min before α-factor release +/- auxin was added to the cultures. Cells were released from the arrest (t = 0) at 25°C into media containing +/- auxin. α-factor was added at t = 80 min to arrest cells in the following G1. Samples were harvested for immunoblotting at the indicated timepoints, run on a polyacrylamide gel, and immunoblotted with the indicated antibodies. Wild-type cells were grown in parallel, but not treated with auxin and are included as a comparison to untreated PDS1-AID cells. Pds1-AID migrates adjacent to a background band (indicated by an *). (B) PDS1-AID and slk19Δ PDS1-AID cells were grown as in part (A). Samples were harvested for immunoblotting at the indicated timepoints, run on a polyacrylamide gel, and immunoblotted with the indicated antibodies. Pds1-AID migrates adjacent to a background band (indicated by an *).

https://doi.org/10.1371/journal.pgen.1007029.g007

The mitotic cyclin, Clb2, accumulates similarly in both the absence and presence of auxin in all mutants (Fig 7), indicating that premature Esp1 activation is not due to premature mitotic entry. The destruction of Clb2 is blocked in ESP1-3D PDS1-AID cells treated with auxin (Fig 7A) and we hypothesize that defects in anaphase onset may activate the SAC, causing a delay in APC activation. cdc55Δ swe1Δ PDS1-AID cells only partially stabilize Clb2, but cdc55Δ mutants are defective in the SAC [52,65]. slk19Δ mutants activate the SAC [66] and irrespective of auxin addition we see stabilization of Clb2, Clb5 and Pds1-AID (only in the absence of auxin) (Fig 7B).

ESP1-3D, cdc55Δ and slk19Δ mutants impair the pericentric organization of Cohesin

Cohesin is bound along the length of paired sister chromatids, but is concentrated within the pericentromere where it forms a barrel structure [8,11]. Cleavage of Cohesin triggers sister chromatid separation, and the local cleavage of Cohesin within the pericentromere is essential for the separation of kinetochores at anaphase onset [14,54]. Because the premature proteolysis of total Mcd1 is subtle in ESP1-3D, cdc55Δ swe1Δ and slk19Δ cells (Fig 7), we wondered if these mutants might have a specific defect in the cleavage of pericentric Cohesin. We imaged pericentric Cohesin by tagging the Smc3 subunit of Cohesin with GFP. When control and mutant PDS1-AID cells are released from a G1 arrest in the absence of auxin, the Smc3-GFP barrel forms normally, persists during metaphase and rapidly disappears at anaphase onset (Fig 8). Prior to anaphase, the pericentric Cohesin barrel fluorescence is ~1.5-fold over the non-barrel nuclear fluorescence, which represents the binding of Cohesin along chromosomes arms. When cells are released in the presence of auxin, Smc3-GFP barrels form in control PDS1-AID cells, though with significantly decreased intensity compared to wild-type cells [11]. Strikingly, Cohesin barrels do not form in ESP1-3D, cdc55Δ, cdc55Δ swe1Δ and slk19Δ cells depleted of Pds1 and Smc3-GFP localization between the SPBs is reduced to background levels.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. The pericentric Cohesin barrel does not form when Pds1 is depleted from cdc55Δ, cdc55Δ swe1Δ, ESP1-3D and slk19Δ cells.

The indicated strains containing SMC3-GFP SPC29-RFP were imaged as in Fig 2D. Maximal Cohesin fluorescence between SPBs was measured in cells with spindles shorter than 2 μm (metaphase, left), and in cells with a single SPB (pre-mitotic, right). The ratio of barrel/nuclear fluorescence is plotted (average ± SD). A value of 1 is the background nuclear fluorescence. There are significant differences between all untreated and auxin-treated strains (p < 0.05; Student’s t-tests).

https://doi.org/10.1371/journal.pgen.1007029.g008

When untreated or auxin-treated cells are imaged prior to mitosis, when SPBs have not yet separated, Smc3-GFP localization is similar in all strains, forming a focus adjacent to the paired SPBs (Fig 8). This pre-mitotic localization is consistent with our observation that total Mcd1 accumulates normally early in the cell cycle in both untreated and auxin-treated cells (Fig 7).

ESP1-3D and slk19Δ mutants bypass a LatA-dependent block to Mcd1 proteolysis

We began investigating Cdk1-dependent Esp1 phosphorylation as a possible mechanism to explain the metaphase arrest of pds1Δ cells grown in latrunculin A (LatA), a treatment that activates a Swe1-dependent checkpoint characterized by low Cdk1 activity. In contrast, sister chromatids separate in pds1Δ cells grown in nocodazole, a treatment that activates the SAC and maintains high Cdk1 activity. Using PDS1-AID cells we confirmed that Pds1 is not required for the maintenance of Mcd1 in LatA-arrested cells, but it is required for the maintenance of Mcd1 in nocodazole-arrested cells (Fig 9). Strikingly, Mcd1 is destabilized in LatA arrested ESP1-3D PDS1-AID and slk19Δ PDS1-AID cells soon after auxin addition, suggesting these mutants bypass the Pds1-independent block to sister chromatid separation (Fig 9). During this bypass, Swe1 remains stabilized and Y19 phosphorylation on Cdk1 is unchanged, showing that this bypass occurs downstream of Cdk1 inhibition. In contrast to the behaviour of Mcd1, Slk19 is cleaved during the first 15 minutes after auxin addition in both nocodazole and LatA arrested cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. ESP1-3D and slk19Δ bypass a Pds1-independent arrest.

The indicated strains were grown to log phase, arrested in G1 with α-factor, and released from the arrest into media containing nocodazole or latrunculin A at 25°C. After 2 h (t = 0), +/- auxin was added to the cultures. Samples were harvested for immunoblotting at the indicated timepoints, run on a polyacrylamide gel, and immunoblotted with the indicated antibodies. cdc55Δ was not analyzed in this experiment because it is defective in both the spindle assembly and morphogenesis checkpoints, and does not arrest in nocodazole or LatA. Pds1-AID migrates adjacent to a background band (indicated by an *).

https://doi.org/10.1371/journal.pgen.1007029.g009

PDS1-AID vs. pds1Δ

In this study we made use of cells in which Pds1 is fused to an auxin-inducible degron (PDS1-AID) in order to induce the rapid degradation of Pds1 [50]. Like pds1Δ cells grown at permissive temperature, PDS1-AID cells grow poorly in the presence of auxin [29] (Fig 2A). We were surprised, however, that these cells advance proteolysis of Mcd1 and often trigger immediate anaphase onset (Figs 2D and 7), phenotypes not seen in pds1Δ cells [16,30]. pds1Δ cells are thought to delay Mcd1 cleavage both via Cdc5/PP2A phosphoregulation, and also because of a defect in Esp1 nuclear localization [20,30,57,67]. Although some Esp1 must enter the nucleus in pds1Δ cells, it is insufficient to trigger premature anaphase, and in fact causes delays in anaphase onset (S2D and S2E Fig) [30]. Although most Pds1-AID is degraded in our experiments, we observe some Pds1-AID accumulation prior to mitosis (Fig 7 and S2A Fig), and speculate that this Pds1 is responsible for the increased Esp1 activity and advancement of anaphase onset in these cells.

Anaphase onset at mitotic entry

Past work has shown that depletion of Pds1 in cells deleted for CDC55 led to the initiation of anaphase soon after mitotic entry [46]. We have extended this study using live cell imaging and confirmed that more than half of ESP1-3D, cdc55Δ and slk19Δ cells depleted for Pds1 initiate anaphase soon after mitotic entry (Figs 2D, 3, 4B and 6C). In many of these cells spindle elongation is initiated immediately after mitotic entry, and sister chromatids separate and segregate to a pole of the spindle, indicating that anaphase onset occurs prematurely.

cdc55Δ and slk19Δ cells display nearly random segregation of chromosome V which would occur if spindle elongation began prior to the formation of bipolar attachments to the mitotic spindle, and may indicate that these mutants also have bi-orientation defects caused by premature Cohesin loss from the pericentromere. Similar bi-orientation defects can be observed in mcm21Δ mutants, which are defective in pericentric Cohesin loading, and load similar amounts of Cohesin within the pericentromeres as on chromosome arms [68]. Unlike mutants in kinetochore components that prevent kinetochore attachment to the spindle and prematurely elongate their spindles [69], we see no evidence for attachment defects, as sister chromatids separate and then segregate (or mis-segregate) to one pole or the other.

We hypothesize that premature anaphase onset in these mutants is caused by the absence of the pericentric Cohesin barrel in early mitosis (Fig 8). The Cohesin barrel, and the pericentric chromatin contained within it, has been proposed to be an integral component of the mitotic spindle [11,14,70,71] that functions to orient sister kinetochores towards opposite poles and to resist the pulling forces of the spindle. Although immediate spindle elongation is uninterrupted in many of these cells, the rate of elongation is slower than during normal anaphase onset, and a subset of cells contract their spindles (Figs 2D, 4B and 6C and S3 Fig). These differences from normal anaphase onset may be caused by persistent sister chromatid linkages outside the pericentromere that slow (and in some cases reverse) spindle elongation.

Separase phosphorylation and dephosphorylation

We have shown that budding yeast Esp1/Separase, as in vertebrates, is phosphorylated in vitro by Cdk1, and its in vivo phosphorylation depends on Cdk1 activity and on three central phosphorylation sites (Fig 1). Two of these three Cdk1 sites (S1027 and T1034) are conserved in related yeasts, and a recent crystal structure of the Esp1/Pds1 complex reveals that these sites lie in a region that forms part of the substrate binding domain of Esp1 [72], raising the possibility that phosphorylation of these sites could affect substrate binding. We think it is unlikely Esp1 phosphorylation stimulates its catalytic activity because the activity of immunopurified Esp1 doesn’t vary during the cell cycle, and Cdk1 phosphorylation of Esp1 doesn’t increase Mcd1 cleavage in vitro ([30] and Frank Uhlmann, personal communication).

Several lines of evidence suggest that phosphorylation stimulates Esp1 activity in vivo: 1) Blocking phosphorylation in esp1-3A and mcd1-10A cells suppress the growth defects of PDS1-AID cells (Figs 2A and 5A), 2) the ESP1-3D mutant acts semi-dominantly to cause synthetic growth defects in combination with PDS1-AID, and exacerbates the immediate spindle elongation phenotype of PDS1–AID (Fig 2B–2D), 3) increasing the dosage of esp1-3A, ESP1 and ESP1-3D increases the growth defects of PDS1-AID cells, and form an allelic series in the strength of this effect (Fig 2C), 4) ESP1-3D cells lacking Pds1 prevent the assembly of the pericentric Cohesin barrel (Fig 8), and 5) ESP1-3D cells share phenotypes with cells lacking PP2A^Cdc55, which can dephosphorylate the same residues that Cdk1 phosphorylates in vitro (Figs 4 and 1F) and regulates Esp1 phosphorylation in vivo (Fig 1C).

PP2A^Cdc55 has also been shown to dephosphorylate Mcd1 in vitro, and deletion of CDC55 increases phosphorylation on Mcd1 in vivo [57]. However, we were surprised that the esp1-3A and mcd1-10A mutants have no effect on the lethality and spindle morphology of cdc55Δ PDS1-AID cells (Fig 5A). These results suggest either that PP2A^Cdc55 dephosphorylates and inhibits other targets that promote anaphase onset, or that the physical interaction of PP2A^Cdc55 to a known substrate plays a more important role than its dephosphorylation. We favor the latter model because previous work has shown that PP2A^Cdc55 can stably bind Esp1, and this interaction is reduced after anaphase onset [59].

SLK19 inhibits anaphase onset

We have identified SLK19 as an inhibitor of anaphase onset in vivo. A function for Slk19 in protecting pericentric Cohesin may provide a mechanism for previous observations that slk19Δ cells have defects in kinetochore clustering and change the elasticity of pericentromeric chromatin [24,25].

We propose a model in which Slk19 acts redundantly with Pds1 to inhibit Esp1 (Fig 10). Phosphorylation of Esp1 relieves inhibition by Slk19, while dephosphorylation and binding of PP2A^Cdc55 enhances Slk19 inhibition. Past data has shown stable physical interactions of Esp1 to both PP2A^Cdc55 and Slk19. Consistent with this model, Slk19 binding to Esp1 occurs throughout most of the cell cycle, and like the interaction between Cdc55 and Esp1, is reduced after anaphase onset [23,59,73]. Our finding that ESP1-3D and slk19Δ cells bypass the Pds1-independent arrest caused by LatA (Fig 9) provides additional evidence that phosphoregulation of Esp1, and Slk19, function redundantly with Pds1 to regulate Mcd1 proteolysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Proposed model for redundant inhibition of Esp1 by Pds1 and Slk19.

Our data support a model in which both Pds1 and Slk19 (19) inhibit Esp1. Anaphase onset is triggered by both APC^Cdc20-mediated proteolysis of Pds1 and Cdk1-dependent activation of Esp1. Slk19 and PP2A^Cdc55 inhibit Esp1, and Cdk1 phosphorylation of Esp1 overcomes this inhibition. We speculate that Esp1 phosphorylation triggers the release of Slk19 from the active site of Esp1. Although PP2A^Cdc55 can dephosphorylate Esp1 in vitro and cdc55Δ cells increase Esp1 phosphorylation in vivo, the lack of suppression of cdc55Δ by esp1-3A suggest that PP2A^Cdc55 binding to Separase may be more important for Esp1 inhibition. Slk19 can inhibit Separase either as an uncleaved or cleaved (as depicted) protein.

https://doi.org/10.1371/journal.pgen.1007029.g010

This model, however, does not provide an explanation for why ESP1-3D has milder phenotypes than slk19Δ and cdc55Δ. We speculate that either the aspartate residues do not fully mimic Esp1 phosphorylation, or an additional regulator of this process is also regulated by Cdk1 phosphorylation. Slk19 itself is phosphorylated by Cdk1 on several sites that are adjacent to the Esp1 cleavage site [74].

Regulation of pericentric Cohesin

ESP1-3D, cdc55Δ and slk19Δ cells all share a defect in the formation of a pericentric Cohesin barrel in mitosis (Fig 8) and this defect is more penetrant in all three mutants than is the advancement in bulk Mcd1 proteolysis (Fig 7). This observation suggests that this regulatory network is involved in the control of pericentric cohesion. Modeling of the Cohesin barrel suggests it may create an outward pushing force along the spindle axis at anaphase onset and may therefore assist in the initial movement of chromosomes towards the spindle poles [70,71]. In this model, ordered loss of Cohesin from the barrel may be needed to transform this outward force into movement of sister chromatids toward the spindle poles. Both Esp1 and Slk19 localize to the kinetochore and at anaphase onset move together to the central spindle [21,22,75], providing a mechanism to localize Separase along the axis of the Cohesin barrel. We speculate that this re-localization is accompanied by relief of Esp1 inhibition by Slk19 and PP2A^Cdc55, leading to cleavage of pericentric Mcd1 from within the Cohesin barrel. The localization of Esp1 and Slk19 along the spindle axis is interdependent [75], so like Pds1, Slk19 may play a positive function delivering Esp1 to its site of action, but also maintain inhibition of Esp1 until Cdk1 activity reaches maximal levels at the metaphase-to-anaphase transition [16,20,67,76].

Although we see premature proteolysis of Mcd1 and loss of pericentric Cohesin when Pds1 is depleted in cdc55Δ and slk19Δ cells, Mcd1 still accumulates normally and localizes properly in pre-mitotic cells (Figs 7 and 8). Cdc5/PP2A^Cdc55 regulation of Mcd1 phosphorylation [30,57] is likely the mechanism for Mcd1 protection early in the cell cycle because Cdc5 activity only rises in mitosis [77–79]. An alternative mechanism may involve a poorly understood function of the mitotic cyclins Clb5 and Clb6, which function redundantly with Pds1 during S-phase to prevent loss of pericentric cohesion [80,81].

Conservation of Separase regulation

In vertebrates, Cdk1 also phosphorylates the central region of Separase, but this phosphorylation is inhibitory, and works in parallel with Securin inhibition [34,82]. Although this is opposite to the regulation we have identified in this work, the mechanism by which Cdk1 inhibits Separase is poorly understood. Separase inhibition may be caused both by the stable binding of Cdk1/Cyclin B1 to Separase, as well as phosphorylation-dependent Separase aggregation [35,83,84]. Although phosphorylation itself triggers aggregation, recent work has shown that stable binding of Cdk1/Cyclin B1 to Separase prevents aggregation [85], indicating that Cdk1 also promotes Separase activity in vertebrates.

PP2A associated with its B56 regulatory subunit (homologous to Rts1 in yeast) interacts directly with human Separase and this interaction has also been shown to both promote and inhibit Separase function [36,37]. Similar to our model for PP2A^Cdc55 function (Fig 10), stable binding of PP2A^B56, rather than dephosphorylation, regulates Separase function.

Slk19 homologues have not been clearly identified outside of budding yeasts. A few reports have speculated that mitosin/CENP-F or fission yeast alp7 may share homology with the C-terminal coiled-coil domains of Slk19 [26,86–88]. Although these regions may mediate Slk19 interaction with microtubules and depletion of CENP-F causes cohesion defects at the kinetochore [89], our proposed mechanism suggests that proteins with homology to the N-terminus of Slk19, which contains the Esp1 cleavage site, may be more relevant to understanding whether vertebrate Separase is regulated by a similar mechanism. If this function were conserved it could be mediated by an unidentified Slk19 homologue, a different Separase substrate, or perhaps even Separase itself, which has three internal autocleavage sites [90]. A recent cryo-EM structure of the C. elegans Separase/Securin complex suggest that the autocleavage sites are accessible to the catalytic site of Separase [91] and persistent binding between the protease domain and these sites, as seen in Slk19 binding to Esp1, would be an effective mechanism to inhibit Separase activity. Mutation of the Separase autocleavage sites in vivo causes poorly understood delays in G2 [92], but when this mutant is expressed in wild-type cells it causes premature loss of centromeric cohesion and separation of sister chromatids [37], suggesting these sites may indeed regulate the initiation of anaphase.

Ethics statement

This study was performed in strict accordance with standards for animal care and use outlined in the Canadian Council on Animal Care Standards. The University of Ottawa is a registered research facility under the Province of Ontario's Animals for Research Act. The protocol was approved by the University of Ottawa Animal Care Committee (Permit Number: BMI-113). All surgery was performed under sodium pentobarbital anesthesia and every effort was made to minimize suffering.

Strain and plasmid construction

S1 Table lists the strains used in this work and S2 Table lists the strains used in each figure. All strains are derivatives of the W303 strain background (W303-1a; see S1 Table for complete genotypes). All deletions and replacements were confirmed by immunoblotting, phenotype or PCR. Strains were constructed by genetic cross and transformation. The sequences of all primers used in this study are available upon request. Phusion polymerase (NEB) was used for all PCR reactions. The bacterial strains TG1 and DH5α were used for amplification of DNA, and Rosetta (Novagen) was used for protein purification.

SPC42-mCherry-NAT^R was constructed in the following manner. The mCherry coding sequence (BBa_K165004) was obtained in the vector BBa_J63009 (iGEM). mCherry was amplified by PCR as a PacI/AscI fragment and cloned into pKT127[93], resulting in pAR733. mCherry along with the KAN^R marker was amplified by PCR off pAR733 and integrated at SPC42. SPC42-mCherry-KAN^R was switched to SPC42-NAT^R using pAG25 [94]. SPC42-eGFP-KAN^R and SPC42-eGFP-Sphis5⁺ were constructed by amplifying eGFP and either the KAN^R marker or the Sphis5⁺ marker from pKT127 and pKT128, respectively [93], and integrating the resulting PCR product at SPC42. SPC42-eGFP-NAT^R and SPC42-eGFP-HYG^R were constructed by switching KAN^R for NAT^R or HYG^R using pAG25 and pAG32, respectively [94]. CDC14-eGFP- Sphis5⁺ was created by amplifying GFP and the Sphis5⁺ marker from pKT128 and integrating the resulting PCR product at CDC14. SMC3-GFP-URA3 was made using the plasmid pLF639 (A. Strunnikov, National Institutes of Health, Bethesda, MD) cut with Hpa1. SPC29 was tagged with RFP by PCR amplifying a SPC29-RFP-HYG^R fragment from yeast strain KBY4999, or a SPC29-RFP-NAT^R fragment from ADR9045. his3::pCUP1-GFP12-lacI-12::HIS3 was made by integrating pSB116 [95]. To create the ura3::240lacO-URA3 allele the lacO array was cut out of pLAU43 (a gift from D.Z. Rudner, Harvard Medical School, Boston, MA) [96] with XbaI/BamHI and cloned into pRS406 [97] to make pAR615. pAR615 was cut with StuI and transformed into yeast to integrate the lacO array at URA3.

cdc55Δ::HIS3 was created using pJM6 [52]. pds1Δ::HYG^R, cdc55Δ::HYG^R and esp1Δ:: HYG^R were constructed by amplifying HYG^R off pAG32 and deleting PDS1, CDC55 or ESP1, respectively. slk19Δ::NAT^R and spo12Δ::NAT^R were constructed by amplifying NAT^R off pAG25 and deleting SLK19 and SPO12, respectively. BAR1 was deleted using pJGsst1 (J. Thorner, University of California, Berkeley, CA). MIH1 was deleted using pIP33 (P. Sorger, Harvard Medical School, Boston, MA). swe1Δ::TRP1 strains were made by crossing JM449 (J. Minshull, Atum, Newark, CA) to the appropriate strains.

The 2μ-pGAL-CLB2-TAP-URA3 (pAR546) and 2μ-pGAL-CLB5-TAP-URA3 (pAR547) plasmids were created as follows. The CLB2 or CLB5 ORF was amplified and the resultant PCR, designed to have overlapping homology, was co-transformed into yeast along with pRS-AB1234 (C. Carroll and D.O. Morgan, UC San Francisco, San Francisco, CA) cut with BamH1 and HindIII.

CEN-PDS1-URA3 was constructed by PCR amplifying PDS1 with upstream and downstream regions and cloning the PCR fragment into pRS316 digested with EcoR1/BamH1 to create pAR1060. PDS1-AID-KAN^R was constructed by amplifying AID and KAN^R from pAID1 [50] and integrating the resulting PCR product at PDS1. PDS-AID-NAT^R was made by switching KAN^R to NAT^R using pAG25. leu2::pGPD1-OsTIR1-LEU2 was constructed by digesting pTIR4 [50] with PmeI to integrate it at LEU2 (plasmids were gifts of T. Eng and D. Koshland, UC, Berkeley, Berkeley, CA).

The mcd1::pGAL-MCD1-18myc-URA3, trp1::pMCD1-MCD1-3HA-TRP1, trp1::pMCD1-mcd1-10A-HA3-TRP1 and leu2::pMCD1-mcd1-10A-3HA-LEU2 alleles were derived from strains Y1287, Y1288 and Y1296 (gifts of N. Hornig and F. Uhlmann, The Francis Crick Institute, London, UK) [56]. ESP1-13myc-KAN^R was created by amplifying 13myc-KAN^R from pFA6a-13Myc-kanMX6 [98] and integrating the resulting PCR product at ESP1. The ESP1-18myc-TRP1 allele was derived from K7024 (a gift of F. Uhlmann, The Francis Crick Institute, London, UK).

Construction of ESP1 mutants

A 5 kb region containing the ESP1 ORF was amplified and cloned into pRS316 between XhoI and NotI resulting in pAR745 (CEN-ESP1-URA3). The XhoI/NotI fragment from pAR745 was cloned into pRS315 and pRS313 resulting in pAR797 (CEN-ESP1-LEU2) and pAR800 (CEN-ESP1-HIS3), respectively.

The esp1-2A-NAT^R, esp1-3A-NAT^R, esp1-1A-NAT^R, esp1-2A+3A-NAT^R, esp1-2A+1A-NAT^R, esp1-2A+3A+1A-NAT^R, esp1-2D-NAT^R, ESP1-3D-NAT^R, esp1-1D-NAT^R, ESP1-2D+3D-NAT^R, esp1-2D+1D-NAT^R and ESP1-2D+3D+1D-NAT^R alleles were constructed in the following manner: six DNA sequences were synthesized (DNA2.0) corresponding to each of the N-terminal (containing SP13 and TP16), central (containing TP1014, SP1027 and TP1034) and C-terminal (containing SP1280) sites and surrounding residues with each site mutated to either alanine or tandem aspartic acid residues. Restriction sites were engineered within or adjacent to the mutated codons for later identification. Each mutated region was amplified and cut with restriction sites found in the ESP1 gene (AvrII/MscI for esp1-2A and esp1-2D; SpeI/NheI for esp1-3A and ESP1-3D; and SalI/AatII for esp1-1A and esp1-1D). These fragments were then cloned into pAR797 to make pAR873 (CEN-esp1-2A-LEU2), pAR875 (CEN-esp1-3A-LEU2), pAR871 (CEN-esp1-1A-LEU2), pAR872 (CEN-esp1-2D-LEU2), pAR874 (CEN-ESP1-3D-LEU2) and pAR870 (CEN-esp1-1D-LEU2). pAR875 was subsequently used to make pAR902 (CEN-esp1-2A+3A-LEU2); pAR873 to make pAR924 (CEN-esp1-2A+1A-LEU2); pAR871 to make pAR886 (CEN-esp1-3A+1A-LEU2); pAR872 to make pAR903 (CEN-ESP1-2D+3D-LEU2); pAR870 to make pAR915 (CEN-esp1-2D+1D-LEU2); pAR874 to make pAR901 (CEN-ESP1-3D+1D-LEU2); pAR886 to make pAR1005 (CEN-esp1-2A+3A+1A-LEU2); and pAR915 to make pAR927 (CEN-ESP1-2D+3D+1D-LEU2).

To add a marker to pAR797, pAR873, pAR875, pAR871, pAR872, pAR874, pAR870, pAR902, pAR924, pAR886, pAR903, pAR915, pAR901, pAR1005 and pAR927, each plasmid was cut with SnaBI and co-transformed into yeast with a PCR product containing the NAT^R cassette amplified from pAG25 and ends overlapping the cut backbone plasmid. Plasmids were rescued and confirmed by restriction digest. The resulting plasmids were pAR906 (CEN-ESP1-NAT^R-LEU2), pAR936 (CEN-esp1-2A-NAT^R-LEU2), pAR905 (CEN-esp1-3A-NAT^R-LEU2), pAR904 (CEN-esp1-1A-NAT^R-LEU2), pAR921 (CEN-esp1-2D-NAT^R-LEU2), pAR922 (CEN-ESP1-3D-NAT^R-LEU2), pAR920 (CEN-esp1-1D-NAT^R-LEU2), pAR990 (CEN-esp1-2A+3A-NAT^R-LEU2), pAR992 (CEN-esp1-2A+1A-NAT^R-LEU2), pAR930 (CEN-esp1-3A+1A-NAT^R-LEU2), pAR991 (CEN-ESP1-2D+3D-NAT^R-LEU2), pAR932 (CEN-esp1-2D+1D-NAT^R-LEU2), pAR931 (CEN-ESP1-3D+1D-NAT^R-LEU2), pAR1006 (CEN-esp1-2A+3A+1A-NAT^R-LEU2), and pAR994 (CEN-ESP1-2D+3D+1D-NAT^R-LEU2).

ESP1 mutants along with the NAT^R cassette were amplified from these plasmids and transformed into wild-type yeast. Presence of mutated phosphorylation sites was verified by amplifying the mutated region and digesting the amplified product with the appropriate restriction enzyme. ESP1-3D-HYG^R and esp1-3A-HYG^R were constructed by switching NAT^R for HYG^R using pAG32. esp1-3A-ADE2 and ESP1-3D-ADE2 were constructed by amplifying ADE2 from pRS412 [97] and using it to replace NAT^R and HYG^R respectively.

3FLAG tagged ESP1 mutants were constructed in the following manner. The ESP1 ORF was amplified off pAR745 and cloned into pBS-KS (Stratagene) between SalI and NotI to make pAR868. A BglII site was inserted downstream of the ESP1 stop codon in pAR868 using site-directed mutagenesis to make pAR877. ESP1-BglII was amplified from pAR877 and co-transformed into yeast along with pAR797 cut with SnaBI and NcoI, and rescued to create pAR888. pAR888 was then cut with BglII and co-transformed into yeast with 3FLAG-KAN^R amplified off pDAM278 [99] and rescued to create pAR911. 3FLAG-KAN^R was then cut out of pAR911 with BsgI and NotI and cloned into pAR797, pAR873, pAR875, pAR871, pAR902, pAR924, pAR886 and pAR1005 to create pAR911 (CEN-ESP1-3FLAG-KAN^R-LEU2), pAR965 (CEN-esp1-2A-3FLAG-KAN^R-LEU2), pAR968 (CEN-esp1-3A-3FLAG-KAN^R-LEU2), pAR973 (CEN-esp1-1A-3FLAG-KAN^R-LEU2), pAR971 (CEN-esp1-2A+3A-3FLAG-KAN^R-LEU2), pAR964 (CEN-esp1-2A+1A-3FLAG-KAN^R-LEU2), pAR966 (CEN-esp1-3A+1A-3FLAG-KAN^R-LEU2) and pAR975 (CEN-esp1-2A+3A+1A-3FLAG-KAN^R-LEU2). These plasmids were then transformed into the appropriate yeast strain.

Physiology

Unless noted in the figure legend, cells were grown in yeast extract peptone media + 2% dextrose (YEPD) at 25°C or 30°C. Cells cycle arrests were performed with 10 μg/mL nocodazole (Sigma-Aldrich) or 100 ng/mL α-factor (Biosynthesis) for 3 hours. Auxin (indole-3-acetic acid, Sigma-Aldrich) was used at 500 μM in liquid and solid media. The morphogenesis checkpoint was activated using 2.5–5 μM LatA (Sigma-Aldrich or Tocris Biosciences). LatA efficacy varied between batches and suppliers so the amount needed to induce a fully Swe1-dependent checkpoint arrest was determined empirically.

Plate-based viability assays were performed using a multi-pronged serial dilution fork (DAN-KAR). Liquid culture viability assays were performed by diluting cultures 1000X and/or 10000X into YPD and sonicated to disrupt cell adhesion. Viability was calculated relative to viability at t = 0. Dilutions were adjusted to ensure that > 100 colonies grew at each timepoint.

Fixed microscopy

For fixed cell microscopy, ~ 2.0 x 10⁶ cells were harvested and fixed with 4% paraformaldehyde in PBS pH 7.5 for 15 minutes. Cells were washed with 100 mM KPO₄/1.2 M sorbitol pH 7.5, sonicated to break cell adhesions and resuspended in KPO₄/1.2 M sorbitol. Samples were imaged using a Nikon TI microscope (Nikon) with a Nikon Plan Apo 60X 1.4 NA objective and FITC and/or TRITC filter sets (FITC (41001); TRITC (41002c), Chroma) at room temperature. Images were obtained using a Photometrics CoolsnapHQ2 camera (Photometrics) and NIS-Elements software (Nikon). Unless otherwise noted a minimum of 200 cells were visually scored per data point. Spindles were measured in three dimensions using a stack of 17 fluorescence images spaced every 0.5 μm, covering the entire height of the cell. All measurements were made using NIS-Elements software (Nikon).

Live microscopy

Imaging pads were made by adding 25% Gelatin (w/v) to SC or YEPD media at 55–60°C, pipetting 50 μL between two microscope slides and allowing it to cool. 1–2 μL of cultures were pipetted onto live imaging pads, covered by a coverslip and sealed with 1:1:1 vaseline:lanolin:petroleum jelly (VALAP). Strains were imaged at 25°C for 2 hours using brightfield and FITC and/or TRITC filter sets (FITC (41001); TRITC (41002c), Chroma) on a Nikon TI microscope (Nikon) with a Nikon Plan Apo 60X 1.40 NA objective and a Photometrics CoolsnapHQ2 camera (Photometrics). 17 Z-slices, spaced every 0.5 μm were imaged at each timepoint. Fluorescence excitation was attenuated using neutral density filters and 100–200 ms exposure times were used for GFP, mCherry and brightfield acquisition. Measurements were made using NIS-Elements software (Nikon). Look up tables were manually adjusted linearly. Example images were prepared using ImageJ software (National Institutes of Health).

Spindle behaviour was classified using three rules. Cells with “normal metaphase spindle formation” did not elongate their spindle more than 2 μm in the first 10 minutes after SPB separation and more than 2.5 μm in the first 15 minutes after SPB separation. Cells whose spindles elongate more than 2 and 2.5 μm in these time-intervals display “immediate spindle elongation.” A small number of cells (15% for cells treated with auxin, and 11% for untreated cells) produce conflicting scores using these two rules (i.e., immediate/normal or normal/immediate in the 10/15 minute intervals), and we categorized these cells manually. For immediate/normal cells: If the spindle elongated to a length greater than 2 μm in a single time point and there was a clear inflection point that defined anaphase onset, these cells were classified as “normal.” If there was no clear anaphase onset inflection point and there was spindle shortening between 10 and 15 minutes, these cells were classified as “immediate.” For normal/immediate cells: If rapid anaphase spindle elongation began between 10–15 minutes with a clear inflection point, these cells were classified as “normal.” If there was no clear inflection point and the spindle elongated at a slow continuous rate, these cells were classified as “immediate”.

Cells with “failed anaphase” do not elongate their spindles to 6 μm in the 60 minutes after SPB separation.

Smc3-GFP fluorescence as cells progress from metaphase to anaphase was quantified according to the method described in Hoffman et al. and Yeh et al. [11,100]. Live-cell images were obtained from cells immobilized on 25% gelatin/media slabs. Five plane Z sections at 200 nm steps through the cell were acquired at 1 min intervals. The microscope used for wide-field imaging was a Nikon Eclipse TE2000E stand (Nikon) with 100 PlanApo NA 1.4 objective with a Hamamatsu Orca ER camera (Hamamatsu). Images were acquired at room temperature with MetaMorph imaging software (Molecular Devices). In brief, a computer-generated 5 x 5 and 6 x 6 pixel regions were centered over the region of interest, and the total integrated fluorescence counts were obtained for each region. Inner- and outer-region data were transferred into Microsoft Excel (Microsoft) with the use of the MetaMorph Dynamic Data Exchange function. The measured value for the 5 x 5 pixel region includes both cohesin fluorescence and local background fluorescence. The background component was obtained by subtraction of the integrated value of the 5 x 5 pixel region from the larger 6 x 6 pixel region. This result was scaled in proportion to the smaller area of the 5 x 5 pixel region and then subtracted from the integrated value of the 5 x 5 pixel region to yield a value for cohesin fluorescence.

Western blots and immunoprecipitation

These methods have been described previously [45]. Immunoprecipitations of wild-type and mutant Esp1 and Esp1-FLAG were performed in APC lysis buffer (50 mM Hepes-KOH pH 7.8, 700 mM NaCl, 150 mM NaF, 150 mM Na-β-glycerophosphate pH 8.3, 1 mM EDTA, 1 mM EGTA, 5% glycerol, 0.25% NP-40, 1 mM DTT, 1 mM PMSF, 1 mM Na₃VO₄, 1 mM benzamidine, and leupeptin, bestatin, pepstatin A and chymostatin all at 1 mM). 1–2μl of α-Esp1 and α-FLAG-M2 (F1804, Sigma-Aldrich) were used in each immunoprecipitate.

The following antibodies were used for Western blots and immunoprecipitations: The use of 9E10 ascites (BabCO), α-Swe1, α-Clb2, α-Pds1, α-Clb5, α-Cdk1, α-P-Cdc2-Y15 (#9111, Cell Signaling Technology) antibodies have been described previously [45,101]. Rabbit polyclonal α-Esp1, α-Mcd1, α-G-6-PDH antibodies (A9521, Sigma-Aldrich) were used at 1:1000, and α-Slk19 at 1:2000, in TBS-T with 4% Fat Free Milk Powder, 5% glycerol, 0.02% NaN₃. An autoclaved solution of 5% milk was used to make the 4% milk dilution buffer to increase the longevity of the antibody solution. Membranes were pre-blocked with TBS-T with 4% Fat Free Milk Powder before incubation with all primary antibodies.

α-Esp1 antibodies were generated as follows: coding sequences for the truncated protein Esp1_230–414 was amplified using PCR and cloned into pHIS-parallel2 [102] as a BamHI/SalI fragment to create pAR882. Denatured His₆-Esp1_230–414 protein was purified on Ni-NTA columns, dialyzed and ~0.5 mg of precipitated protein was injected into rabbits every 4 weeks for 8 to 16 weeks (uOttawa animal facility). Rabbit serum was harvested and the α-Esp1 antibodies purified on Affigel-15 (Bio-rad) columns coupled to purified His₆-Esp1_230–414 that had been solubilized in 0.3% SDS.

α-Mcd1 antibodies were generated as follows: coding sequence for the truncated protein Mcd1_201–301 was amplified by PCR and cloned into pGEX6P-1 (Promega) as a BamHI/EcoRI fragment to create pAR742. GST-Mcd1_201–301 was purified and 1 mg of the fusion protein was injected into rabbits every 4 weeks for 8 to 16 weeks (uOttawa animal facility). Rabbit serum was harvested, and the α-Mcd1 antibodies purified on an Affigel-10 (Bio-rad) column coupled to purified malE-Mcd1_201–301. malE-Mcd1_201–301 was expressed from the plasmid pAR1117 which contains Mcd1_201–301 cloned as a BamH1/Sal1 fragment into pMAL-c2 (NEB).

α-Slk19 antibodies were generated as follow: coding sequence for the truncated protein Slk19_700–817 was amplified by PCR and cloned into pGEX6P-1 (Promega) as a BamHI/EcoRI fragment to create pAR1230. GST-Slk19_700–817 was purified and 1 mg of the fusion protein was injected into rabbits every 4 weeks for 8 to 16 weeks (uOttawa animal facility). Rabbit serum was harvested, α-GST antibodies were removed on an Affigel-10 (Bio-rad) column coupled to GST, and the α-Slk19 antibodies were purified on an Affigel-10 (Bio-rad) column coupled to purified GST-Slk19_700–817.

HRP-conjugated α-rabbit and α-mouse secondary antibodies (Bio-rad) were used at a 1:5000 dilution in TBS-T + 4% Fat Free Milk Powder for 30 min to 1 hr., washed with TBS-T and incubated in Western Lightning Plus-ECL (PerkinElmer). Signal detection was done on HyBlot CL (Harvard Scientific) autoradiography film.

PhosTag polyacrylamide gels

3.0 x 10⁷ cells were harvested for immunoblotting and cell pellets were washed twice with 50 mM HEPES pH 8.0. Cell extracts were prepared by bead beating frozen cell pellets in a Mini-Beadbeater (BioSpec Products) in 1X urea sample buffer (2% SDS, 65 mM Tris-Cl pH 6.8, 10% glycerol, 4 M Urea, 0.02% bromophenol blue, 5% betamercaptoethanol, and 1mM PMSF) and an excess of acid washed glass beads (BioSpec Products) for 60 sec. Samples were run on 10% acrylamide gels with 100 μM Phos-tag reagent (Wako), 200 μM MnCl₂ and 0.1% SDS. Gels were run for 5 h at 200 V and 25 mA. Following electrophoresis, gels were washed 2 x 10 min in transfer buffer with 1 mM EDTA, 1 x 10 min in transfer buffer and transferred to nitrocellulose using standard wet-transfer protocol at 60 V and 500 mA for 2 hr. at 4°C.

In vivo labeling of Esp1

10 x 10⁷ cells were harvested and labeled in 2 mL phosphate-free medium containing 0.5–1 mCi ³²PO₄ (PerkinElmer) as described previously [45]. Uptake of label was monitored by scintillation counting (TriCarb 2910TR; PerkinElmer) of the cells and media, and exceeds 98%. Esp1-myc13 or Esp1 was immunoprecipitated using 9E10 or α-Esp1 antibodies, respectively.

Kinase and phosphatase assays

Cdk1/Clb2-CBP and Cdk1/Clb5-TAP complexes were purified from cells containing pAR546 or pAR547, respectively (2μ-pGAL-CLB2-TAP and 2μ-pGAL-CLB5-TAP, described above). Clb2-TAP was overexpressed by growth in galactose. Cdc55-CBP complexes were purified from asynchronously growing CDC55-TAP (ADR5465) cells. Protein complexes were purified as described previously [45,103].

To phosphorylate Esp1, Esp1 was immunoprecipitated with α-Esp1 antibody and treated with purified Cdk1/Clb2-CBP complexes. Kinase reactions were performed with 1 μCi γ-[³²P]ATP as previously described [45]. Dephosphorylation of Esp1 was measured by incubating in vitro phosphorylated Esp1, still bound to beads, with TAP purified PP2A^Cdc55 complexes as previously described [45]. Okadaic acid (LC laboratories) was used at 2 nM. Phosphatase assays were quantified using a Typhoon Trio Phosphorimager and ImageQuant software (GE).