Key for accurate chromosome partitioning to the offspring is the ability of mitotic spindle microtubules to respond to different molecular signals and remodel their dynamics accordingly. Spindle microtubules are conventionally divided into three classes: kinetochore, interpolar, and astral microtubules (kMTs, iMTs, and aMTs, respectively). Among all, aMT regulation remains elusive. Here, we show that aMT dynamics are tightly regulated. aMTs remain unstable up to metaphase and are stabilized at anaphase onset. This switch in aMT dynamics, important for proper spindle orientation, specifically requires the degradation of the mitotic cyclin Clb4 by the Anaphase Promoting Complex bound to its activator subunit Cdc20 (APC/CCdc20). These data highlight a unique role for mitotic cyclin Clb4 in controlling aMT regulating factors, of which Kip2 is a prime candidate, provide a framework to understand aMT regulation in vertebrates, and uncover mechanistic principles of how the APC/CCdc20 choreographs the timing of late mitotic events by sequentially impacting on the three classes of spindle microtubules.
Chromosome segregation requires remodeling of the mitotic spindle. The mitotic spindle is composed of microtubules, microtubule-associated proteins (MAPs), and motor proteins (Prosser and Pelletier, 2017). Based on their function, microtubules are divided into three categories: (i) kinetochore microtubules (kMTs), which connect the spindle poles to chromosomes and direct their segregation; (ii) interpolar microtubules (iMTs), which form a bundle of antiparallel microtubules and promote the distancing of the two sister chromatids via spindle elongation; and (iii) astral microtubules (aMTs), which connect the spindle poles to the cellular cortex and guide chromosomes along the polarity axis by dictating spindle positioning and orientation (Winey and Bloom, 2012). Proper spindle positioning is fundamental for the correct segregation of chromosomes, and it is crucial for many cellular processes such as stem cell maintenance, tissue homeostasis, and development (Lechler and Mapelli, 2021). Despite their fundamental role, the molecular mechanisms that regulate aMT dynamics remain elusive and largely overlooked. Here, we investigate aMT dynamics in Saccharomyces cerevisiae and ask whether and how they are coordinated with other cell cycle events.
Chromosome segregation is initiated by the activation of the anaphase promoting complex or cyclosome (APC/C) in complex with its activator subunit Cdc20 (Sudakin et al., 2001; Izawa and Pines, 2014). The APC/C is an E3-ubiquitin ligase whose specificity is dictated by the interaction with its regulatory subunits, Cdc20 and Cdh1 (Visintin et al., 1997). Activation of the APC/CCdc20 at the metaphase-to-anaphase transition initiates a three-step signaling cascade that culminates in cohesin cleavage—the point of “no return” for mitotic exit (Fig. 1 A). Cohesin is a protein complex that holds chromosomes together from the moment of their replication up to their separation (Uhlmann et al., 2000). Following cohesin cleavage, the coordination between sister chromatid separation and segregation is directed by changes of mitotic spindle microtubule dynamics. kMTs retract and pull sister chromatids toward the spindle poles while iMTs drive spindle elongation, thereby segregating sister chromatids apart from each other. Consistent with their functions (kMTs search and capture chromosomes and iMTs promote the formation of a short bipolar spindle without forcing elongation), kMTs and iMTs are unstable in metaphase (Higuchi and Uhlmann, 2005). Vice versa, kMTs and iMTs are stabilized in anaphase to preserve proper kMT-chromosome interactions and to promote spindle elongation (Higuchi and Uhlmann, 2005; Mallavarapu et al., 1999). Notably, since budding yeast exhibits a “closed” mitosis—the nuclear envelope does not break down—the mitotic spindle extends within the nucleus between two spindle pole bodies (SPBs), the centrosome equivalent (Byers and Goetsch, 1975). Microtubules extend from both the cytoplasmic (aMTs) and the nuclear (kMTs and iMTs) faces of the SPBs during the cell cycle. By removing the opposing forces to spindle pulling, cohesin cleavage indirectly affects the dynamics of nuclear microtubules.
Molecularly, the switch in microtubule dynamics from an unstable to a stable state has long been associated with cyclin dependent kinase (CDK) activity. CDK1 activity is high in metaphase and decreases in anaphase due to APC/C-mediated degradation of the cyclin-regulatory subunits and to the activation of the main yeast CDK-counteracting phosphatase Cdc14 (Fig. 1 A). The observation that several motors and MAPs are regulated by phosphorylation and de-phosphorylation events mediated by the antagonistic couple CDK1/Cdc14 gave rise to a model linking spindle microtubules stabilization to the overall change in the phosphorylation landscape in favor of de-phosphorylation (Khmelinskii et al., 2007; Avunie-Masala et al., 2011; Khmelinskii et al., 2009). This view has been recently challenged by an elegant study in budding yeast reporting that the number of phosphorylated or de-phosphorylated residues in late mitosis is similar, thus suggesting that mitotic kinases can compensate for the drop in CDK-mediated phosphorylation (Touati et al., 2018). Supporting this model is the observation that the phosphatase Cdc14 and the Polo-like kinase Cdc5 are redundant in triggering spindle elongation (Roccuzzo et al., 2015). The contribution of spatially defined mechanisms renders the dynamics of microtubule regulation more complex. An example is the phospho-regulation of single kMTs following their binding to kinetochores, which relies on the inhibition of the Aurora B kinase to promote kMT stabilization when the attachment that is limited to a single kinetochore creates tension (Akiyoshi et al., 2010; Sarangapani et al., 2013). Altogether these observations unveiled a sophisticated regulation of nuclear spindle microtubules at the metaphase-to-anaphase transition, which relies both on phosphorylation and de-phosphorylation events—depending on the residue—and takes into account the spatially defined modulation of individual microtubules.
How aMTs fit into this picture remains unknown. Most studies focused on the regulation of aMT binding to the cellular cortex. It emerged that, both in budding yeast and multi-cellular eukaryotes, CDKs negatively affect this interaction. In yeast, the cyclin Clb4 promotes the detachment of aMTs from the cellular cortex in early mitosis by phosphorylating a yet-to-define substrate (Maekawa and Schiebel, 2004). In human, CDK1 reduces aMTs binding to the membrane by phosphorylating the nuclear mitotic apparatus (NuMa; Kotak et al., 2013). NuMa localizes at the cortex and it is a component of the evolutionary conserved cortical machinery essential for spindle orientation (Kiyomitsu and Boerner, 2021). The human Polo-like kinase 1 (Plk1) phosphorylates NuMa (Sana et al., 2018) and contributes to negatively regulating its cortex localization (Kiyomitsu and Cheeseman, 2012). Instead, little is known as to whether aMT dynamics are regulated in a cell cycle dependent manner. The only evidence of a direct regulation of aMT dynamics comes from a recent study highlighting that CDK1-dependent phosphorylation of the plus-end tracking protein GTSE1 is required to destabilize aMTs in prometaphase (Singh et al., 2021).
Here we show that, similar to nuclear microtubules, aMTs dynamics switch from an unstable to a more stable status at anaphase onset. At the heart of this switch, required to maintain aMTs in close proximity of the cellular cortex and ultimately to establish proper spindle positioning, is the APC/CCdc20-dependent degradation of the mitotic cyclin Clb4 that likely impacts aMT dynamics by affecting the activity/localization of aMT regulators, of which Kip2 is a prime candidate. Besides identifying the unique function of Clb4 among all cyclins in this process, our data evidence a central role for the APC/CCdc20 in choreographing late mitotic events by sequentially instructing the three classes of spindle microtubules.
Astral microtubules are stabilized in anaphase
To gain insights into aMT regulation in mitosis, we probed aMT morphology in wild-type S. cerevisiae cells undergoing a synchronous cycle. At each time-point, cell cycle progression was assessed by monitoring spindle morphologies and aMT length and number, two bona fide indicators of aMT stability (Drechsler et al., 2015). A correlation emerged between the population approaching metaphase and a decrease in both aMT length and number (Fig. 1 B). Contrariwise, these parameters increased when the population entered anaphase (Fig. 1 B). This relationship became particularly evident when we correlated the morphology of aMTs with metaphase and anaphase spindles (Fig. 1, C and D).
Since the distinction between cells in late G2 and in early metaphase, or in late metaphase and in early anaphase, is somewhat arbitrary, we validated our findings by probing aMT dynamics in homogenous populations. More precisely, we assessed aMT morphology in cdc20 and cdc15 mutant cells, which arrest in metaphase and anaphase, respectively (Fig. 1 A and Fig. 2 A). To arrest cells in metaphase, unless otherwise specified, we used a conditional allele of CDC20 where the wild-type CDC20 gene is fused to an Auxin-inducible degron sequence (CDC20-AID, henceforth cdc20; Shetty et al., 2019). To arrest cells in anaphase, we used an ATP analog-sensitive allele of the kinase CDC15 (cdc15-as1; Bishop et al., 2001). cdc20 and cdc15 mutants were released from a G1 block into the restrictive conditions, and aMTs were analyzed at their terminal phenotype (Fig. 2 A). While the arrest uniformed the average aMT length between metaphase and anaphase cells, the number of aMTs remained significantly higher in cdc15 mutants (Fig. 2, B and C). To exclude possible mutant-specific effects, the same assays were performed arresting the cells in metaphase by different mutations (cdc20-1, cdc23-1 [Zachariae and Nasmyth, 1999], GAL-MAD2 [Rossio et al., 2010] and cdc13-1 [Hartwell et al., 1973] cells) or in anaphase (cdc5-as1 [Snead et al., 2007] and cdc14-1 [Hartwell and Smith, 1985] cells). The data obtained confirmed what we found in CDC20-AID and cdc15-as1 cells (Fig. S1 A and Fig. 2, B and C), thus indicating that changes in aMT morphology are determined by the cell cycle phase rather than by the mutant strains used.
The observation that upon protracted arrest aMT length uniformed between the two cell cycle phases prompted us to explore aMT dynamics in depth. For this purpose, we moved to live-cell imaging and measured parameters that ultimately define dynamic properties, such as catastrophe and rescues rates (Kosco et al., 2001; Estrem et al., 2016). The length of individual aMTs in cdc20 and cdc15 cells expressing a GFP-tagged Tub1 fusion was measured over time in single cells at their terminal phenotype (Fig. 2 D and Fig. S1 B). In line with previous findings, aMTs of cdc15 cells resulted less dynamic than the ones of cdc20 cells as assessed by the decrease of both catastrophe and rescue rates (Fig. 2 E and Table S1). Taken together, the observations that: (i) aMTs increase in number and length when wild-type cells move from metaphase to anaphase; (ii) aMT number is higher in mutants arrested in anaphase when compared to their metaphase counterpart; and (iii) anaphase aMTs appear less dynamic, support the conclusion that aMTs are stabilized at anaphase onset. Why aMT length is similar in the two conditions remains puzzling.
Stabilization of aMTs in anaphase relies on a specific signature
Having established that aMTs are stabilized at anaphase onset, we wished to characterize the molecular mechanism at the heart of this switch. It is known that anaphase stabilization of iMTs requires the activity of either the phosphatase Cdc14 or the Polo-like kinase Cdc5 (Roccuzzo et al., 2015). To investigate whether aMTs and iMTs share the same regulatory mechanism, we examined the aMT phenotype of cdc14 cdc5 double-mutant cells. These cells arrest in mini-anaphase (after cohesin cleavage, the hallmark of anaphase onset), with short bipolar spindles and undivided nuclei, due to defects in spindle elongation (Fig. 1 A and Fig. 3 A; Roccuzzo et al., 2015).
aMT dynamics of cdc14 cdc5 cells were probed at their terminal phenotype both by indirect immunofluorescence and live-cell imaging, and compared with the ones of cdc20 and cdc15 arrested cells, which share with the double mutant spindle morphology and mitotic phase, respectively. Consistent with their mini-anaphase arrest, aMTs of cdc14 cdc5 cells were significantly more stable than the ones of metaphase-arrested cdc20 cells: they were both longer and more numerous (Fig. 3, B and C), thus suggesting that the molecular mechanism at the core of the anaphase aMT stabilization differs from the one controlling iMTs, as it is independent of Cdc14 and/or Cdc5 activities.
The aMTs of cdc14 cdc5 arrested cells were also more stable than the ones of late anaphase arrested cdc15 mutants. aMTs of cdc14 cdc5 cells were similar in number, but longer than their cdc15 counterpart (Fig. 3, B and C). This phenotype is not caused by an inappropriate activation of checkpoint pathways since abrogation of checkpoint activities (spindle assembly, DNA damage, and spindle positioning checkpoints) in cdc14 cdc5 cells did not alter aMT morphology (Fig. S2, A–C).
To understand why cdc14 cdc5 and cdc15 cells carry aMTs of different length, we moved to live-cell imaging. It turned out that while the aMT catastrophe and rescue rates of metaphase (cdc20) and anaphase (cdc15) cells always follow the same trend, either by increasing or decreasing simultaneously, thus possibly reaching an equilibrium, the two parameters were uncoupled in the aMTs of the cdc14 cdc5 cells (Fig. 3, D and E). Here, aMTs are characterized by a low catastrophe rate associated with a high rescue rate (Fig. 3, D and E; and Table S1), a combination that favors polymerization and could provide an explanation to the different aMT length observed.
Why catastrophe and rescue rates are uncoupled in cdc14 cdc5 cells remains unclear. Different scenarios can be envisioned: (i) spindle elongation may negatively impact on aMT dynamics (e.g., cross-talk between iMTs and aMTs or spatial constraints imposed by bringing the poles in close proximity to the cellular cortex); (ii) Cdc14 and/or Cdc5 may counteract aMT stabilization; or (iii) a combination of the two. To assess if spindle elongation is impacting aMT dynamics, we allowed cells to elongate their spindles but prevented the establishment of aMT-cortex connections by inactivating the spindle positioning factor Kar9 (Beach et al., 2000) and the minus-end directed motor Dyn1 (Li et al., 1993; Falk et al., 2016) in a cdc15 background. We found that the aMTs of cdc15 kar9 dyn1 cells are significantly longer than the ones of cdc15 cells and resemble aMTs of cdc14 cdc5 cells, indicating that spindle elongation could affect aMT dynamics in an indirect manner likely by bringing these microtubules in the proximity of the cortex (Fig. S2, A, D, and E). To test if Cdc14 and Cdc5 counteract aMT stability, we probed aMT morphology in metaphase-arrested cells lacking either Cdc14 or Cdc5 and found that aMTs of cdc20, cdc20 cdc5, and cdc20 cdc14 cells were similar both in number and length (Fig. S2 F). To exclude redundancy, we looked at the aMTs of a cdc20 cdc14 cdc5 triple mutant and observed that the simultaneous inactivation of Cdc14 and Cdc5 didn’t alter metaphase aMT dynamics (Fig. S2 G). Here, we used an allele of CDC20 whose expression is under the control of the methionine-repressible promoter (pMET-CDC20) because the concomitant inactivation of Cdc14 and Cdc5 is synthetically lethal with all tested CDC20 mutant alleles (cdc20-1, cdc20-3, and cdc20-AID) and with a cdc23 temperature sensitive allele of a gene encoding for a core subunit of the APC/C (Care et al., 1999). aMTs of pMET-CDC20 cdc14, pMET-CDC20 cdc5, and pMET-CDC20 cdc14 cdc5 cells were similar in number to the ones of metaphase cells (pMET-CDC20), albeit slightly longer (Fig. S2 G), a phenotype likely due to the less stringent pMET-CDC20 metaphase arrest (Fig. S2 H) as it was not observed in cdc20 cdc5 and cdc20 cdc14 cells (compare Fig. S2, F with G). All together these data indicate anaphase onset as a critical step for aMT stabilization.
APC/CCdc20 activation is sufficient to stabilize aMTs at anaphase onset
To identify the anaphase-specific trait that elicits aMT stabilization, we dissected the signaling cascade directing cohesin cleavage into its three critical steps, namely: (1) the activation of the APC/CCdc20; (2) the activation of the separase/Esp1, mediated by the APC/CCdc20-dependent degradation of securin/Pds1; and (3) the Esp1-mediated cleavage of the Scc1 subunit of the cohesin complex and, next, probed aMTs in mutant strains impaired in the completion of sequential steps of the cascade: cdc20, esp1, scc1nc (these cells overexpress under the control of the galactose-inducible promoter an uncleavable allele of Scc1—GAL-scc1R180DR268D; Uhlmann et al., 1999; Uhlmann et al., 2000), and cdc14 cdc5 cells (Fig. 4, A and B).
To assess aMT morphology and dynamics, the mutants of interest were released from the G1 arrest into restrictive conditions. During the analysis, we noticed that, differently from cdc20 and cdc14 cdc5 cells, which maintained a stable short bipolar spindle for the entire duration of the experiment, both esp1 and scc1nc cells disassembled their mitotic spindle soon after reaching a short bipolar spindle configuration (metaphase-like morphology; Fig. 4, C and D). This observation is consistent with the ability of these cells to pull the nucleus, hence the spindle, into the bud and, as a consequence, exit from mitosis prematurely, albeit having failed to separate sister chromatids (Campbell et al., 2019). To perturb the system as little as possible, we initially analyzed aMTs of esp1 and scc1nc cells at the time point preceding spindle disassembly, with the caveat that we might have a mixed population of cells, some at the terminal phenotype and others still in metaphase (Fig. 4 D). Consistent with this possibility, the difference in aMT length and number between cdc14 cdc5 and cdc20 cells was less obvious when these parameters were measured at the time point chosen for esp1 and scc1nc cells (compare Fig. 4 E with Fig. 3 B). Nevertheless, aMTs of esp1 and scc1nc cells were slightly more numerous and longer than aMTs of cdc20 cells and resembled the ones of cdc14 cdc5 cells, suggesting that APC/CCdc20 activation is the molecular event triggering aMT stabilization (Fig. 4 E).
To confirm this observation, we engineered esp1 strains to prevent spindle collapse. Since spindle collapse is caused by the movement of the nucleus into the bud and the consequent activation of the mitotic exit network (MEN; Campbell et al., 2019), a signaling cascade that triggers mitotic exit through the phosphatase Cdc14 (Fig. 1 A), we: (i) precluded spindle movement into the bud by disrupting Dyn1-mediated pulling forces; (ii) impaired MEN activity by inhibiting Cdc15 or Cdc5 (Campbell et al., 2019); and (iii) inactivated Cdc14 (Fig. 1 A and Fig. 5, A and B). In agreement with our hypothesis, most esp1 dyn1, esp1 cdc15, esp1 cdc5, and esp1 cdc14 cells retained a short bipolar spindle for the entire duration of the experiment (Fig. S3) and exhibited, at their terminal arrest, aMTs significantly longer and more numerous than the ones of cdc20 cells, pheno-copying the ones of the cdc14 cdc5 mutants (Fig. 5, C and D). These data further support that the activation of the APC/CCdc20 is the molecular event required to dictate aMT stabilization at anaphase onset.
The APC/CCdc20 stabilizes aMTs through the degradation of the mitotic cyclin Clb4
The APC/CCdc20 complex targets proteins for proteasomal degradation (Pines, 2011); hence, to elucidate the molecular mechanism driving aMT stabilization, we searched among its substrates for the one(s) whose degradation impacts on aMT dynamics. Given that critical APC/CCdc20 metaphase substrates are securin (Pds1 in yeast; Cohen-Fix et al., 1996; Yamamoto et al., 1996) and cyclin B (Clb cyclins in yeast; Sullivan and Morgan, 2007), we asked whether their removal could prompt aMT stabilization in phases of the cell cycle when they are normally unstable.
As Pds1 degradation is sufficient to trigger spindle elongation and chromosome segregation in cdc20 mutants (Cohen-Fix et al., 1996; Yamamoto et al., 1996), to avoid interference by the elongated spindle, we analyzed the impact of Pds1 deletion on aMTs in cdc20 mutant cells also defective in spindle elongation (pMET-CDC20 cdc14 cdc5; Fig. S4 A). We found that the aMTs of pMET-CDC20 pds1 cdc14 cdc5 cells are similar in length and number to the ones of pMET-CDC20 cdc14 cdc5 cells, and significantly less and shorter than the ones of cdc14 cdc5 pds1 cells (Fig. S4 B), thus indicating that Pds1 removal is not sufficient to promote aMT stabilization and pointing toward other putative candidates.
We then assessed whether Clb cyclins could be such a candidate. Since most cdc20 mutant alleles are synthetically lethal with CLB2 deletion (Costanzo et al., 2016 and Fig. S4 C), to avoid issues of synthetic lethality we tested the consequences of deleting individual cyclin subunits in cells arrested in S-phase with hydroxyurea (HU; Koç et al., 2004; Chabes et al., 2003). Of all tested cyclins, only deletion of Clb4 significantly altered aMT length and number, pinpointing Clb4 as the critical substrate of the APC/CCdc20 affecting aMT stabilization (Fig. 6, A and B).
To validate this result, we deleted CLB4 in a cdc20 background and analyzed aMTs of cdc20 clb4 cells at the metaphase arrest. We found that lack of Clb4 led to abnormally stable metaphase aMTs. In particular aMTs of cdc20 clb4 cells were as stable as the ones of esp1 cdc15 (Fig. 6, C and D). As Clb4 was never formally ascertained as an APC/CCdc20 substrate, we next probed Clb4 levels in: (i) wild-type cells overexpressing Cdc20 (GAL-CDC20; Visintin et al., 1997) and (ii) wild-type, cdc20, cdc15, and cdc14 cdc5 cells undergoing a synchronous cell cycle. As reported for bona fide APC/CCdc20 substrates (e.g., Pds1 [Visintin et al., 1997]), we found that: (i) in S-phase arrested cells, high levels of Cdc20 led to fast Clb4 degradation, while Clb2, whose degradation is mainly dependent on the APC/CCdh1 complex (Visintin et al., 1997; Schwab et al., 1997), remained stable and eventually accumulated (Fig. 6 E); and (ii) Clb4 protein levels remained high in cdc20 cells, but dropped when cells entered anaphase in wild-type, cdc14 cdc5, and cdc15 cells (Fig. 7, A–C). Additionally, the kinetics of Clb4 degradation resembled that of Pds1 degradation and differed from that of Clb2 degradation, which has delayed degradation with respect to Clb4 in cycling cells (Fig. 7 B); and was only partially degraded in anaphase arrested cells (Fig. 7 C; Baumeret al., 2000).
To complete our analysis and to exclude the contribution of other APC/CCdc20 targets, we also tested the kinesin motor protein Kip1 (Gordon and Roof, 2001), the Haspin-like kinase Alk2 (Nespoli et al., 2006), the APC/CCdh1 inhibitor Acm1 (Enquist-newman et al., 2008), and the DNA replication-promoting kinase Dbf4 (Ferreira et al., 2000). Individual proteins were depleted in a cdc20 mutant background, and the resulting double mutants were next investigated for aMT morphology. Of all the proteins tested, only deletion of Kip1 or Acm1 exhibited aMTs with an intermediate phenotype—slightly increased in length than their cdc20 counterpart (Fig. S4, A and D).
The findings that: (i) aMTs of cdc20 clb4 cells phenocopy the ones of esp1 cdc15, (ii) this phenotype is recapitulated only by the removal of Clb4, and finally (iii) Clb4 is an APC/CCdc20 substrate, identify Clb4 as the main target of the APC/CCdc20 whose degradation is required for aMT stabilization in anaphase and highlight a unique and specialized function of Clb4 among cyclin subunits.
Multiplexed quantitative phospho-proteomics associates aMT stabilization to de-phosphorylation
To gain insights into the complex molecular mechanism underlying aMT dynamics, we interrogated the proteome and phospho-proteome of cdc20, cdc14 cdc5, and cdc15 cells representing the metaphase, mini-anaphase, and anaphase cell cycle phase, respectively. Samples were collected at the terminal phenotype and subjected to tandem mass tags-based multiplexed quantitative phospho-proteomics. Our dataset covered 4,641 proteins (Table S3) and identified 5,324 phospho-sites (Table S4). To validate our experimental setup, we looked at CDK1 substrates (Fin1 [Bouchoux and Uhlmann, 2011], Ase1 [Khmelinskii et al., 2007], and Orc6 [Bouchoux and Uhlmann, 2011]) whose de-phosphorylation status is known to change from the metaphase-to-anaphase transition and throughout anaphase with different kinetics (Fig. S5, A–C).
Having established that the chosen mutants are a good proxy for the mitotic cell cycle phases of interest (Fig. S5, A–C), we assessed global phosphorylation changes. We identified 974 phosphosites that changed their status when going from metaphase to anaphase, as assessed by comparing cdc20 with cdc15 cells. Among the 974 phosphosites, 197 were phosphorylated (69 of which reside in the minimal CDK1 consensus site) and 777 were de-phosphorylated (of which 407 are putative CDK substrates; Fig. 8 A). Since aMT stabilization occurs at anaphase but is independent of Cdc14 and Cdc5, we restricted the analysis by comparing the phospho-proteome of cdc20 cells with the combination of the datasets of cdc14 cdc5 and cdc15 cells. By these means, we identified 340 phospho-sites that changed from metaphase to anaphase, of which 81 were phosphorylated (34 of which reside in the minimal CDK1 consensus site) and 259 were de-phosphorylated (of which 160 are putative CDK substrates; Fig. 8 B). The changes of these phosphorylation sites were unlikely caused by changes in protein abundance as the respective protein amounts remained largely unchanged (Fig. S5, D and E). Taken together, our data evidence a bias towards de-phosphorylation for anaphase cells.
To identify possible regulators of aMT dynamics, we scrutinized the phosphorylation status of residues within proteins known to be involved in microtubule regulation (de Gramont et al., 2007; Yeh et al., 2000; Kosco et al., 2001; Blake-Hodek et al., 2010), namely: (i) five kinesin motor proteins (Cin8 [Hoyt et al., 1992], Kip1 [Hoyt et al., 1992], Kip2 [Roof et al., 1992], Kip3 [DeZwaan et al., 1997], and Kar3 [Meluh and Rose, 1990]), (ii) the single minus-end directed Dyn1 protein (Li et al., 1993), and (iii) five MAPs (Bim1 [Schwartz et al., 1997], Bik1 [Berlin et al., 1990], Ase1 [Pellman et al., 1995], Stu1 [Pasqualone and Huffaker, 1994], and Stu2 [Wang and Huffaker, 1997]). Since the activity of these proteins is often modulated by microtubule-cortex interaction (Yeh et al., 2000; Omer et al., 2018; Hoopen et al., 2012), we extended our analysis to proteins that mediate the connection between aMTs and the cellular cortex, such as the cortical receptor Num1 (Farkasovsky and Küntze, 1995), the actin motor protein Myo2 (Beach et al., 2000), and the actin nucleator Bni1 (Yeh et al., 2000). Interestingly, we found that Cin8, Kip2, Bim1, Bik1, Ase1, Stu1, Num1, Myo2, and Bni1 showed at least one residue with a different phosphorylation status between metaphase and anaphase arrested cells with a bias in favor of de-phosphorylation in anaphase (Fig. S5 F).
To validate our analysis and as a proof of concept we chose Kip2, a kinesin with a microtubule polymerization function (Fig. 8 C). Kip2 is inhibited by GSK3-dependent phosphorylation. This event likely requires priming at the S63 and T275 Kip2 sites by an additional kinase, of which CDK1 is one candidate (Drechsler et al., 2015; Chen et al., 2019). Notably, the S28 residue highlighted by our phospho-proteomic analysis falls into a stretch of residues within the GSK3 consensus motif (Fig. 8 C). To test if CDK-mediated phosphorylation of Kip2 contributes to keep aMTs unstable in metaphase, we, first, probed aMTs in cdc20 kip2-S63AT275A mutant cells (henceforth kip2-2A) and found that they pheno-copy the ones of cdc14 cdc5 double mutant cells (Fig. 8, D and E). Next, we confirmed that anaphase stabilization of aMTs requires Kip2 activity, as deleting KIP2 in cdc14 cdc5 double mutant cells reduced both aMT length and number (Fig. 8, D and E). To establish if Kip2 is a substrate of Clb4, we compared the phosphorylation profiles of Kip2 wild-type and 2A allelic variants in S-phase-arrested cells following Clb4 overexpression. Since we didn’t observe significant differences when protein samples were run on regular acrylamide gels, to exacerbate differences among the phospho-species, we moved to Phos-tag gels. We found that high levels of Clb4 alter the phosphorylation pattern of both Kip2 allelic variants (Fig. 8 F). Indeed, we observed the appearance of a slower migrating smear above wild-type Kip2 and the disappearance of one fast migrating band (band b) concomitantly with the enrichment of a slower migrating one (band c) in Kip2-2A. Next, we compared the phosphorylation profiles of the two Kip2 allelic variants in cdc20 metaphase-arrested cells in the presence or absence of Clb4 at physiological levels (Fig. 8 G). Deletion of CLB4 led to the appearance of a faster migrating band (band a) in wild-type Kip2. In Kip2-2A, when Clb4 was present, the phosphorylation profile of the kinesin pheno-copied the one triggered by Clb4 overexpression, vice versa when the cyclin was absent the slower migrating band (band c) collapsed with concomitant increase of the faster one (band b; Fig. 8 G). Taken together these data identify Kip2 as a bona fide Clb4 substrate and pave the way for further testing of putative substrates/sites within other candidates including Ase1 (S834), Myo2 (T1097), and Bni1 (T1918).
Dynamic aMTs and Cdc5 guide proper spindle orientation
The identification of Clb4 as central to the regulation of aMT dynamics contributes to our understanding of the mechanism by which budding yeast properly aligns its mitotic spindle along the mother-bud axis, an essential requisite for survival. Schiebel and colleagues proposed that Clb4 in complex with CDK1 facilitates spindle alignment by mediating the interactions of aMTs with the bud cortex and enables the turnover of established aMT-bud cortex attachments (Maekawa and Schiebel, 2004). Our data are now suggesting that it does so by controlling aMT stability. This model allows clear predictions: (i) stable microtubules should exhibit an increased residence time at the cortex and (ii) premature aMT stabilization should result in an increased number of wrong attachments.
To assess if the increased aMT stability observed in cdc14 cdc5 double mutant cells reflects changes in the dynamics of aMT-cortex interactions, we analyzed the behavior of aMTs at the cellular cortex in cdc20 and cdc14 cdc5 arrested cells, and measured the time that each individual aMT spent in proximity of the cellular cortex. In agreement with our hypothesis, aMTs of cdc14 cdc5 cells remained close to the cellular cortex longer than aMTs of cdc20 cells, thus suggesting that the dynamics of aMT-cortex connections are altered in double mutant cells (Fig. 9, A and B).
To assess if premature aMT stabilization favors wrong attachments, we probed aMT directionality. Proper cortical attachments foresee that aMTs originating from the bud-directed SPB enter the bud, while the others point to the cortex of the mother cell. If stabilization prevents error correction, then mutants with prematurely stable microtubules should manifest directionality problems, with aMTs originating from two different SPBs pointing to the same cellular compartment, and aMTs originating from a single SPB pointing to both mother and daughter cells (in our analysis, both categories were grouped as abnormal aMTs). Consistent with our hypothesis, we found that all mutants with stable aMTs (cdc20 clb4, esp1 cdc15, and cdc14 cdc5) showed an increased percentage of abnormal aMTs when compared to cdc20 mutants (Fig. 9, C–E).
Since proper attachment of aMTs with the bud cortex directs the spindle toward the emerging bud and orients it to the polarity axis (Shaw et al., 1997; Maddox et al., 2000; Segal et al., 2002), we asked whether anomalous aMTs affect spindle orientation. We probed spindle orientation by measuring the angle formed by the spindle and the bud-neck in cdc20, cdc14 cdc5, esp1 cdc15, and cdc20 clb4 cells. We found that, while most metaphase arrested cells correctly orient their spindle, cdc14 cdc5 cells do so in a random manner suggesting that stable aMTs may lead to spindle positioning defects. The finding that in esp1 cdc15 and cdc20 clb4 cells spindles were properly oriented indicates that aMT stability may predisposes to, but it is not sufficient to compromise spindle orientation (Fig. 9 F). Since among the tested mutants only cdc14 cdc5 manifested spindle orientation defects, we wondered if Cdc14 and/or Cdc5 inactivation could account for the phenotype. To test this hypothesis, we analyzed spindle orientation in cdc20 cdc14, cdc20 cdc5, and cdc20 cdc14 cdc5 cells, in which aMTs are unstable, and in esp1 cdc5, esp1 cdc14 cdc5, and cdc20 clb4 cdc5 cells, in which aMTs are stable. We found that all tested strains with unstable aMTs or with stable aMTs but active Cdc5 carry properly oriented spindles, instead cells in which aMTs are stable and Cdc5 is inactivated, showed orientation defects (Fig. 9 F). These data indicate that unstable aMTs and Cdc5 activity are redundantly required for proper spindle orientation.
Despite the importance of aMTs in controlling spindle stability and orientation, their regulation remains largely understudied. Here we report that in budding yeast, similarly to kMTs and iMTs, aMT dynamics are also intrinsically regulated in a cell cycle dependent manner. More precisely, they remain unstable up to metaphase and are stabilized as soon as cells enter anaphase. This finding is consistent with their role in searching for cortical anchor sites and correcting erroneous attachments in metaphase and next in stabilizing the binding of the mitotic spindle with the cortex to properly guide spindle positioning and elongation in anaphase.
Our study in budding yeast identifies two evolutionary conserved mitotic machineries, the Clb4-CDK1 and APC/CCdc20 complexes as central regulators of aMT dynamics. The presence of Clb4 is sufficient to render aMTs unstable up to metaphase, likely by introducing one or multiple inhibitory phosphorylation events in factors affecting microtubule stability, of which Kip2 is an example. Vice versa, aMT stabilization, observed at anaphase, is dictated by the APC/CCdc20-mediated degradation of the mitotic cyclin Clb4. Of note, this function of Clb4 in aMT regulation is unique as its removal is sufficient to stabilize aMTs in cell cycle phases when they are normally unstable, whereas the same does not stand true for the other mitotic cyclins. Finally, aMT stability in late anaphase is counteracted by a mechanism that likely integrates two anaphase specific requirements, namely spindle elongation and stable aMT-cortex links.
The identification of a kinase as central to aMT regulation immediately calls for the phosphatase required to reverse these Clb4-CDK1–mediated phosphorylation events. If in yeast the main mitotic CDK-counteracting phosphatase is Cdc14, our data indicate that at least one additional phosphatase is involved. A possible candidate is Glc7 (the sole PP1 catalytic subunit in yeast) in combination with its regulatory subunit Bud14. Indeed, overexpression of Bud14 increases aMT length in a way reminiscent of cells lacking Clb4 (Knaus et al., 2005). The involvement of the Glc7-Bud14 complex is also supported by its localization. Both Glc7 and Bud14 accumulate at bud cellular cortex, where they could counteract the activity of the Clb4-CDK1 complex, which mainly localizes at the plus-end of aMTs directed toward the bud (Ni and Snyder, 2001; Knaus et al., 2005; Maekawa and Schiebel, 2004). Importantly, this interplay between Clb4-CDK1 and Bud14-Glc7 at the bud cellular cortex would not only satisfy the requisite of the yet-to-identify “spatial cue” necessary for the specific stabilization of bud-directed aMTs as reported by Barral and colleagues (Lengefeld et al., 2018), but also integrates intrinsic (e.g., cell cycle regulation of cyclin levels) and extrinsic signals (e.g., interaction between bud-directed aMTs and the bud cellular cortex).
The mechanisms of regulation of aMT dynamics by CDK1 and APC/CCdc20 activities that this work identified in yeast are likely retained in vertebrates. At least three pieces of evidence support this hypothesis. First, in all vertebrates CDK1-CyclinB activity is high in metaphase and decreases at anaphase onset due to APC/CCdc20-mediated degradation of the cyclin B subunit (Sullivan and Morgan, 2007). Similarly to the Clb4-CDK1 complex in yeast, CDK1-CyclinB activity destabilizes aMTs in prometaphase by phosphorylating the EB1-dependent microtubule plus-end tracking protein GTSE1 in human cells (Singh et al., 2021). Second, in mammalian cells, Dyn1 activity is low in metaphase and high in anaphase, and, since, in budding yeast, the increase in Dyn1 activity is directed by aMT stabilization, it is possible that aMTs are stabilized in anaphase also in vertebrates (Kiyomitsu and Cheeseman, 2012; Kiyomitsu and Cheeseman, 2013; Kotak et al., 2013; Estrem et al., 2016). Third, aMTs are often reported to shrink during late anaphase in human cells, thus suggesting that a mechanism counteracting aMT stability connected to spindle elongation/cortex proximity may also exist in these cells.
This work identifies budding yeast Cdc5 as a novel regulator of spindle positioning redundant to aMT stabilization. Interestingly, Plk1, the human homologue of Cdc5, has also been intensively linked to the regulation of spindle positioning (di Pietro et al., 2016). Indeed, Plk1-dependent phosphorylation of several components of the cortex platform that bind to aMTs is required to correctly orient and position the spindle along the cleavage plane (Bergstralh et al., 2017). Investigating how Cdc5 influences these processes may further clarify the role of this kinase in multi-cellular eukaryotes and shed light into how it coordinates spindle positioning with the plethora of mitotic events under its control. Taken together, our work legitimizes yeast as an optimal model system to investigate aMT regulation and spindle positioning at molecular level. Besides the fact that each yeast cell has few aMTs (1–6 aMTs per cell; Shaw et al., 1997), easily distinguishable from other types of microtubules and singularly traceable over time (Fees et al., 2017), our mass spectrometric analysis clearly illustrates the advantages of exploiting the power of yeast genetics to clarify complex molecular processes.
An important corollary of our study in yeast is uncovering the APC/CCdc20 complex as the master “choreographer” of late mitotic events, where it precisely coordinates, in time and space, the sequence of events required for the faithful execution of mitosis by sequentially impacting on the stability of the three classes of spindle microtubule: astral, kinetochore, and interpolar. More precisely, active APC/CCdc20 targets several substrates for degradation, including Clb4, Pds1, and other mitotic cyclins (Fig. 10, Step 1). Degradation of Clb4 is the signal for aMT stabilization that drives proper spindle positioning (Fig. 10, Step 2). Instead, Pds1/Securin degradation unleashes the protease Esp1/Separase (Fig. 10, Step 2). Esp1 has a dual role. On the one hand it indirectly affects kMT dynamics by cleaving the cohesin subunit Scc1 (Uhlmann et al., 1999; Baskerville et al., 2008) that leads to chromosome separation and movement to the poles (Asbury, 2017; Fig. 10, Step 3). On the other hand, as a component of the cdc14 Early Anaphase Release (FEAR) network (Stegmeier et al., 2002), it promotes the transient activation of Cdc14 (Fig. 10, Step 3). In turn, Cdc14 acts on motors and MAPs to stabilize iMTs (Khmelinskii and Schiebel, 2008; Roccuzzo et al., 2015; Higuchi and Uhlmann, 2005), thereby prompting spindle elongation and ultimately chromosome segregation (Fig. 10, Step 4). The time delay imposed by the incremental number of molecular events impacting specific classes of spindle microtubules guarantees that proper spindle positioning precedes sister chromatid separation (cohesin cleavage) which in turn precedes their segregation (spindle elongation; Fig. 10).
Several observations suggest that this logic of mitotic progression stands true in vertebrates. While clear differences exist between budding yeast and multi-cellular eukaryotes mitosis (i.e., budding yeast cells do not break the nuclear envelope during cell division, DNA is only partially condensed, etc.), we already mentioned data supporting a regulation of aMTs mediated by CDK1 activity (Singh et al., 2021). In respect to chromosome movement toward the poles (anaphase A), evidence exists that kMT de-polymerization is coupled to cohesin cleavage (Oliveira et al., 2010). Finally, spindle elongation, hence chromosomes segregation, relies on the assembly of the central spindle (the Metazoa equivalent of the yeast spindle midzone), and this process requires, as in yeast, iMT stabilization and bundling. The latter is promoted by the centralspindlin complex comprising the Caenorhabditis elegans ZEN-4 (mammalian orthologue MKLP1) kinesin-like protein and the Rho family GAP CYK-4 (MgcRacGAP), which in turn is regulated by phosphorylation and de-phosphorylation events of the kinesin motor domain (Mishima et al., 2004). In yeast, the phospho-regulation controlling spindle midzone assembly is mediated by a proline-directed kinase-phosphatase switch, with Cdc14 being a likely candidate (Mishima et al., 2004). Interestingly, the two human homologues of Cdc14 bind to microtubules and promote both their stabilization and bundling activity during mitosis (Cho et al., 2005). This finding in addition to the knowledge that, in human cells, a CDK counteracting phosphatase(s), other than Cdc14 homologues, whose activation necessitates proteasomal activity (Skoufias et al., 2007), drives mitotic exit, makes it likely that the stepwise regulation of late mitotic events dictated by the APC/CCdc20 is an evolutionary conserved mechanism of regulation.
Materials and methods
All strains used in this study are isogenic to W303 and are listed in Table S2.
Cell cycle arrest and synchronization experiments were performed as previously described (Amon, 2002). Yeast cells were grown in yeast extract peptone (YEP) media unless differently specified in the figure legend. Carbon sources (glucose, raffinose, or galactose) were used at a final concentration of 2%. In all experiments cells were pre-arrested in G1 phase with 5 µg/ml α-factor (RP01002; Genscript) and released in a synchronous cell cycle or into the arrest of interest. To release cells from G1, cells were washed with 10 volumes of fresh medium and transferred into medium lacking the pheromone. When pertinent, drug(s) were added and restrictive conditions applied.
In detail: Temperature-sensitive alleles were inactivated by incubating the culture at the restrictive temperature of 37°C. cdc5-as1 and cdc15-as1 alleles were inhibited by adding to the media 5 µM of CMK (custom-made, Accendatech; Snead et al., 2007) or of 1NM-PP1 analogue 9 (A603003; Toronto Research Chemicals), respectively. Cells carrying an AID degron were inactivated with Auxin (I5148; Sigma-Aldrich) at a final concentration of 500 µM. Methionine (M9625; Sigma-Aldrich) was added at a final concentration of 8 mg/ml to inactivate the MET-CDC20 construct. HU (H8627; Sigma-Aldrich) was used at a final concentration of 10 mg/ml to arrest cells in S-phase.
Indirect in situ immunofluorescence
1 ml of cell culture OD600 = ∼0.4–0.6 was incubated overnight in fixing solution (3.7% formaldehyde in 0.1 M potassium phosphate buffer, pH 6.4) at 4°C. Next, cells were washed three times in 0.1 M potassium phosphate buffer and once in sorbitol buffer (1.2 M sorbitol, 0.33 M citric acid, pH 5.9). Cells were lysed with 0.1 mg/ml Zymolyase 100 T in sorbitol buffer for ∼30 min at 30°C. When digestion was complete spheroplasts were washed in sorbitol buffer and deposited on a poly-L-lysine coated multi-wells slide for 10 min. Slides were soaked in −20°C cold methanol for 3 min and in −20°C cold acetone for 10 s.
For spindle microtubule visualization spheroplasts were: (a) incubated 90 min with rat anti-tubulin alpha YOL34 monoclonal antibody (MCA78G; AbD Serotec) diluted 1:100 in PBS/BSA (1% crude BSA, 0.04 M K2HPO4, 0.01 M KH2PO4, 0.15 M NaCl, 0.1% NaN3); (b) washed five times in PBS/BSA; (c) incubated 90 min with FITC-conjugated donkey anti-rat antibody (712-095-153; Jackson ImmunoResearch Laboratories) used at 1:100 dilution; (d) washed five times in PBS/BSA; and (e) added with a DAPI mounting solution (90 ml of glycerol 100%, 10 ml of KPBS, 100 mg of p-phenylenediamine, 5 μg of DAPI). The KPBS solution is 0.04 M K2HPO4, 0.01 M KH2PO4, 0.15 M NaCl. Slides were closed with a cover slip and sealed with nail polish (Senic-Matuglia and Visintin, 2017).
Astral microtubule analysis in fixed cells
Images of stained cells were acquired at room temperature with an upright LEICA DM6 B microscope with a 100X/1.40 oil UPlanSApo ∞/0.17/DFN 25 Olympus objective and Andor Zyla.4.2P camera using Leica Application Suite X software. Optimized z-stacks were taken to cover a thickness of 6.1 µm. Image analysis was performed using the “Fiji Is Just ImageJ” (FIJI) software. aMT length was measured in three dimensions using the FIJI plug-in “simple neurite tracer.” To properly score aMT number, the presence of abnormal aMTs and the bud-neck/spindle angles, Z-series were collapsed into a maximum-intensity two-dimensional projection using the Z-project function. Note that the aMT number is likely underestimated due to the difficulty to separate two or more aMTs (bundle) close to each other. The bud-neck was defined based on the Differential Interference Contrast (DIC) image of the cell.
Astral microtubule analysis in live cells
50 µl of arrested cells carrying a GFP-tagged Tub1 fusion were collected at OD600 = 0.2–0.4 and loaded in a CellASIC ONIX plate for haploid yeast cells (Millipore). This procedure allows a constant addition of fresh medium to the culture and prevents cellular movements during image acquisition. Cells were incubated in filtered YEP media with glucose (YEPD) media at 37°C. Images were acquired every 10 s for a total of 5 min using a Nikon Eclipse Ti inverted microscope with a 100X/1.40 oil Olympus objective and an Andor Zyla sCMOS camera using the NIS software version 5.10.00. At each time point, 17 z-stack images were taken (0.4 µm apart) covering a total thickness of 6.4 µm. Image acquisition was followed by a deconvolution process automatically performed by the Huygens software. Image analysis was performed using FIJI. To reduce the complexity and the noise of the analysis, aMT length measurements were performed using the FIJI plug-in “simple neurite tracer” in maximum-intensity two-dimensional projection using the Z-project function. As described in Fees et al. (2017), different events were identified: (i) polymerization events—defined as an increase in microtubule length by at least 0.5 µm across a minimum of three time points; (ii) depolymerization events—defined as a decrease in microtubule length by at least 0.5 µm across a minimum of three time points; and (iii) pause events—defined as net changes in microtubule length <0.5 µm across a minimum of three time points. Next, catastrophe frequencies were calculated by dividing the number of polymerization/pause after polymerization-to-depolymerization transitions by the total time of all growth events. Rescue frequencies were calculated by dividing the number of depolymerization/pause after depolymerization-to-polymerization transitions by the total time of all shrinkage events. Polymerization rates were calculated by dividing the net change in length by the change in time for each polymerization event. Depolymerization rates were calculated by dividing the net change in length by the change in time for each depolymerization event. Dynamicity was calculated by dividing the sum of the absolute value of all length changes (in µm) by the total duration of the image acquisition (in seconds), and by multiplying the resulting values by 1690 (number of tubulin subunits corresponding to 1 µm of microtubule length) to obtain tubulin subunits per second (Toso et al., 1993). aMT-cortex connection was evaluated based on the DIC image of the cell.
Definition of astral microtubule stability
In the literature, the term “stability” is often used to express two slightly different concepts: (i) the tendency of aMTs to polymerize and elongate, instead of depolymerize and shorten; and (ii) the capacity of aMTs to maintain a determined length over time, without growing or shrinking. The first one can be depicted by different indicators, like aMT length and number in fixed cells, or the maximum length reached by each single aMT in live cell imaging. The second one can be measured using the parameter “dynamicity” in live cell imaging, which indicates the net exchange of tubulin subunits over time—independently of polymerization or depolymerization. In this manuscript, the term “stability” refers to the first definition.
Cells were lysed in 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM p-nitrophenyl phosphate, 50 mM dithiothreitol, 1 mM phenylmethylsulphonyl fluoride, and 2 µg/ml pepstatin with glass beads and boiled in 1× sample buffer. The primary antibodies used were: mouse 9E10 monoclonal anti-myc (CVMMS-150R-1000, used at 1:1,000 dilution; Covance) to detect Clb4-13myc and Kip2-13myc; mouse 16B12 anti-HA (901515, used at 1:1,000; Bio Legend) to detect 3HA-Clb4; rabbit anti-Clb2 (Y-180; SC-9071, used at 1:1,000 dilution; Santa Cruz Biotechnology) for Clb2; mouse anti-Pgk1 (A-6457, used at 1:5,000 dilution; Molecular Probes) for Pgk1; rabbit anti-Pds1 (used at 1:1,000 dilution) for Pds1 was a kind gift of Adam Rudner (Ottawa Institute of Systems Biology, Ottawa, Canada); rabbit anti-Kar2 (used at 1: 200,000 dilution) for Kar2 was a kind gift of Mark D. Rose (Princeton University, Princeton, NJ); goat anti-rabbit IgG (H + L)–HRP conjugate (170-6515, used at 1:5,000 dilution; Bio-Rad) and goat anti-mouse IgG (H + L)–HRP conjugate (170-6516, used at 1:10,000 dilution; Bio-Rad) were used as secondary antibodies to visualize proteins using chemiluminescence (ECL; GE Healthcare). Phos-tag gels were obtained adding Phos-tag AAL-107 (304-93521, Fujifilm Wako Chemicals Europe GmbH, used at the final concentration of 50 µm) and MnCl2 10 mM in 6% Acrylamide gels.
In Figs 1 C, 2, B and C, and 9 B, P values were determined by unpaired two-tailed Student’s t test using the GraphPad Software. In Figs. 3, B and E, 4 E, 5 D, 6, A and D, 8 E, 9, E and F, S1 A, S2, C, D, F, and G, and S4, B and D, P values were determined by One-Way Anova-Tukey’s multiple comparisons test using the GraphPad Software. Data distribution was assumed to be normal, but this was not formally tested. P value of <0.05 was considered statistically significant (∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001; ∗∗∗∗ = P < 0.0001). In graphs, averages ± SEM is normally shown.
Repeatability of the experiments
Each experiment has been repeated at least three times. The results were highly reproducible.
Liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis
Sample preparation followed a previously reported procedure (Li et al., 2019). Digested samples were labeled with Tandem Mass Tag (TMT)-11plex reagents (90406, A34807; Thermo Fisher Scientific) in the following order: 126 (cdc15-as1), 127n (cdc15-as1), 129n (cdc20-AID), 129c (cdc20-AID), 130n (cdc20-AID), 130c (cdc14-1 cdc5-as1), 131 (cdc14-1 cdc5-as1), and 131c (cdc14-1 cdc5-as1). Proteomic and phospho-proteomic data were collected on an Orbitrap Fusion and an Orbitrap Lumos mass spectrometer (Thermo Fisher Scientific), respectively. An online real-time search algorithm (Orbiter) was used in proteomic data collection as described previously (Li et al., 2020; Schweppe et al., 2020). MS raw files were initially converted to mzXML and monoisotopic peaks were re-assigned using Monocle (Rad et al., 2020). Database searching with SEQUEST included all entries from the Saccharomyces Genome Database (2014). Peptide-spectrum matches were adjusted to 1% false discovery rate (FDR; Elias and Gygi, 2007). Protein-level FDR was filtered to the target 1% FDR level. Phosphorylation site localization was determined using AScore algorithm (Beausoleil et al., 2006) and filtered at 13 (95% confidence). For TMT reporter ion quantification, a 0.003-Da window around the theoretical m/z of each reporter ion was scanned, and the most intense m/z was used. Reporter ion intensities were adjusted to correct for the isotopic impurities of the TMT reagents according to the manufacturer’s specifications. Peptides were filtered for a summed signal-to-noise of 100 across all channels. For each protein, peptide TMT values were summed to create protein quantifications. To compensate for differential protein loading within a TMT plex, the summed protein quantities were adjusted to be equal within the plex. Phosphorylation site quantifications were also normalized by correction factors generated in this process to account for protein loading variance. For each protein or phosphorylation site within a TMT plex, the signal-to-noise value was scaled to sum to 100 for subsequent analysis. Unpaired student’s t test was performed in Perseus (version 18.104.22.168) to identify the phospho-residues that were significantly different between the different mutant cells. An FDR of 0.001 (S0 at 0.05) was applied, with an additional cut-off of a log2 fold change <−0.5 or >0.5. To identify the phospho-residues likely phosphorylated by CDK, and the phospho-peptides were filtered using the minimal CDK consensus (S/TP).
Online supplemental material
Fig. S1 shows aMT length and number of different mutant strains that arrest either in metaphase or in anaphase and individual frames of two aMTs of a cdc20 mutant cell analyzed by live-cell imaging; Fig. S2 shows experiments performed to explain the aMT phenotype of cdc14 cdc5 cells; Fig. S3 shows spindle kinetics of different esp1 mutants; Fig. S4 shows experiments to identify the APC/CCdc20 substrate(s) whose degradation is required for anaphase aMT stabilization; Fig. S5 shows details of the phospho-proteomic analysis performed; Table S1 contains aMT dynamic parameters; Table S2 lists the strains used in this study; Table S3 is available as an Excel file and contains the protein quantification obtained by the proteomic analysis; Table S4 is available as an Excel file and contains phospho-site quantification obtained by the phospho-proteomic analysis.
Strains, reagents, and protocols used in the manuscript are available to the scientific community upon request. The MS proteomics data have been deposited to the ProteomeXchange Consortium with the dataset identifier PXD028828.
In memory of Angelika Amon an unmatchable mentor and dear friend.
We thank A. Amon (Massachusetts Institute of Technology, Cambridge, MA), A. Rudner (Ottawa Institute of Systems Biology, Ottawa, Canada), M.D. Rose (Princeton University, Princeton, NJ), D. Branzei (FIRC Institute of Molecular Oncology, Milan, Italy), M. Muzi-Falcone (University of Milan, Milan, Italy), O. Cohen-Fix (National Institute of Diabetes and Digestive and Kidney diseases, Bethesda, MD), and D. Liakopoulos (Centre de Recherche de Biochimie Macromoléculaire, Montpellier, France) for strains and reagents; W. Maruwge for English language editing; E. Schiebel, M. Mapelli, I. Cheeseman, and members of the Visintin laboratory for critical discussions and for critical reading of the manuscript.
This work was partially supported by an International Early Career Scientist grant from Howard Hughes Medical Institute and a grant from the Italian Ministry of Health (RF-2011-02347470) to R. Visintin, a grant from National Institutes of Health GM67945 to S.P. Gygi, and in part by a Fondazione Italiana per la Ricerca sul Cancro-Associazione Italiana per la Ricerca sul Cancro fellowship (Giorgio Boglio) to F. Zucca. F. Zucca was a PhD student within the European School of Molecular Medicine.
The authors declare no competing financial interests.
Author contributions: Conceptualization: F. Zucca and R. Visintin; Funding acquisition: R. Visintin and S.P. Gygi; Investigation: F. Zucca, J. Li, and C. Visintin; Project Administration, Supervision, and Validation: R. Visintin. Visualization: F. Zucca and R. Visintin. Writing—original draft: F. Zucca and R. Visintin. Writing—review & editing: All authors.