Separase Regulates INCENP-Aurora B Anaphase Spindle Function Through Cdc14

See allHide authors and affiliations

Science  19 Dec 2003:
Vol. 302, Issue 5653, pp. 2120-2124
DOI: 10.1126/science.1091936


The inner centromere-like protein (INCENP) forms a complex with the evolutionarily conserved family of Aurora Bkinases. The INCENP-Aurora complex helps coordinate chromosome segregation, spindle behavior, and cytokinesis during mitosis. INCENP-Aurora associates with kinetochores in metaphase and with spindle microtubules in anaphase, yet the trigger for this abrupt transfer is unknown. Here we show that the conserved phosphatase Cdc14 regulated the yeast INCENP-Aurora complex, Sli15-Ipl1. Cdc14 dephosphorylated Sli15 and thereby directed the complex to spindles. Activation of Cdc14 by separase was sufficient for Sli15 dephosphorylation and relocalization. Cdc14 not only regulates mitotic exit but also modulates spindle midzone assembly through Sli15-Ipl1.

At the metaphase-anaphase transition, the activated separase, Esp1, promotes sister chromatid separation by cleaving the cohesin complex (1). The spindle elongates dramatically and separates the sister chromatids to opposite poles. Separase also plays an ill-defined yet essential role in regulating the stability of anaphase spindles (2, 3). This function may be mediated by separase-dependent activation of the conserved phosphatase Cdc14 (4). In yeast, Cdc14 is kept inactive by entrapment in the nucleolus (5, 6). In early anaphase, separase, as part of the FEAR pathway (Cdc fourteen early-anaphase release), promotes a transient and partial release of Cdc14 from the nucleolus. In a second phase, the mitotic exit network (MEN), a GTPase-driven signaling cascade, releases the remaining Cdc14, which then facilitates mitotic exit (7).

Cdc14 was released from the nucleolus into the nucleoplasm and cytoplasm at the onset of anaphase (5, 6). The elongating spindle therefore became exposed to Cdc14 (fig. S1), and Cdc14 associated with spindle poles (Fig. 1A). Part of the spindle pole signal in mid-late anaphase cells arose from the binding of Cdc14 to the Bub2-Bfa1 complex, which is associated with the spindle pole body (SPB) that is in the daughter cell (8). However, Cdc14 staining at both spindle poles was also observed in cells in which Cdc14 had just been released from the nucleolus. At this early stage of anaphase, Cdc14 preferentially colocalized with the kinetochore protein Nuf2 (9) but not with the SPB marker Spc42 (10) (in 87% of cells with distinctive Nuf2-Spc42 signals) (Fig. 1A). Kinetochore association of Cdc14 was also observed in cells lacking the Bub2-Bfa1 complex (fig. S1). Thus, Cdc14 may regulate events at kinetochores and anaphase spindles.

Fig. 1.

Cdc14 associates with kinetochores and dephosphorylates Sli15. (A) Colocalization of Cdc14-GFP and the kinetochore marker Nuf2-CFP, but not the SPB component Spc42-RFP (red fluorescent protein), in wild-type cells at the beginning of anaphase. Inserts are enlargements of the spindle poles marked with arrows. Two focal planes of the same cell are shown. Note, in yeast, kinetochores are clustered close to SPBs during most of the cell cycle (23). Scale bar, 5 μm. (B) Immobilized GST proteins (lanes 1 to 3) were incubated with MBP-Sli15 proteins (lane 4). Bound proteins were analyzed by immunoblotting (lanes 5 to 6). (C) Immunoprecipitated Sli15-3HA (lane 1) was incubated as indicated. (D and E) CDC14 (D) and cdc14-td (E) cells with CFP-TUB1 were synchronized (G1: t = 0) and incubated at 37°C. Sli15-9Myc, Clb2, Sic1, Pds1, and β-tubulin Tub2 were determined by immunoblotting.

Kinetochore and spindle proteins were tested for interaction with Cdc14 in a two-hybrid assay. The central, microtubule-binding region of the budding yeast inner centromerelike protein (INCENP) Sli15 showed strong interaction with Cdc14 (fig. S1). In vitro binding experiments confirmed that the microtubule-binding region of Sli15, located from amino acids 227 to 558 (Sli15227-558) (11), interacted directly with Cdc14 (Fig. 1B).

Because Sli15 is a phosphoprotein (12), its association with Cdc14 could lead to its dephosphorylation. Immunoprecipitated Sli15-3HA was efficiently dephosphorylated by incubation with purified wild-type Cdc14 protein cells but not by incubation with the inactive mutant Cdc14C283A protein (Fig. 1C). Dephosphorylation of Sli15 by Cdc14 was also observed in vivo (fig. S2).

To investigate whether the cell cycle–dependent dephosphorylation of Sli15 required Cdc14 activity, we synchronized wild-type and Cdc14-depleted (cdc14-td) cells and monitored the phosphorylation status of Sli15. In cdc14-td cells, CDC14 was fused to the degron element (13, 14). This made cells temperature sensitive because of the complete degradation of Cdc14-td (15). In wild-type and cdc14-td cells, Sli15 phosphorylation was maximal when levels of the mitotic B-type cyclin Clb2 peaked (Fig. 1, D and E). In wild-type cells, Sli15 dephosphorylation started at the beginning of anaphase as Cdc14 was released from the nucleolus (Fig. 1D). In cdc14-td cells, Sli15 phosphorylation persisted and remained high even when cells were arrested in anaphase (Fig. 1E). Dephosphorylation of Sli15 is therefore dependent on Cdc14.

We next tested whether dephosphorylation of Sli15 by Cdc14 is a prerequisite for the translocation of Sli15-Ipl1 to anaphase spindles (16). In CDC14-containing wild-type cells, Sli15 and Ipl1 associated with spindles that had just begun to elongate (Fig. 2, A and B). In contrast, the Sli15 and Ipl1 of cdc14-td cells failed to bind to anaphase spindles (Fig. 2, A and B). Thus, Cdc14 was not only essential to dephosphorylate Sli15 but also to target the Sli15-Ipl1 complex to spindles. Consistently, activation of Cdc14 in metaphase cells led to the premature association of Sli15-Ipl1 with spindle microtubules (fig. S2).

Fig. 2.

Cdc14 targets Sli15 and Ipl1 to anaphase spindles. (A and B) Synchronized CDC14 and cdc14-td cells were analyzed at 37°C for (A) Sli15-GFP or (B) Ipl1-GFP spindle association. n > 100. (C) Mutated Sli15 proteins are underphosphorylated in metaphase-arrested cells. (D) Sli156A binds to spindles independently of Cdc14 activity and targets Ipl1 to anaphase spindles. (E) Quantification of (D). Average of two experiments; n > 50. Scale bars, 5 μm.

If Sli15 dephosphorylation is sufficient to trigger microtubule binding, a mutated Sli15 that fails to become phosphorylated should bind to anaphase spindles in the absence of Cdc14. Cdc14 preferentially dephosphorylates cyclin-dependent kinase (Cdk) sites (7). We replaced serine and threonine residues of six Cdk1 consensus sites within the microtubule-binding domain of Sli15 with alanine (14). Sli15S335A and Sli156A (serines 335, 373, 427, 437, 462 and threonine 474 to alamine) of metaphase-arrested cells were considerably underphosphorylated (Fig. 2C). The functional Sli15S335A and Sli156A proteins associated prematurely with metaphase spindles of wild-type cells (fig. S3) and with anaphase spindles of cdc14-2 cells (Fig. 2, D and E). In addition, Sli156A directed Ipl1 to spindles in the absence of Cdc14 activity (Fig. 2, D and E). Dephosphorylation of Sli15 is thus sufficient to target Sli15-Ipl1 to spindles.

The FEAR functions before the MEN but results in only a partial release of Cdc14 from the nucleolus (4). If FEAR-activated Cdc14 were sufficient to dephosphorylate Sli15, this would have important implications for the timing of Sli15-Ipl1 translocation. Sli15 was directed to anaphase spindles of cdc15-1 cells (FEAR active, MEN inactive) with an efficiency similar to that seen in wild-type cells (Fig. 3A and fig. S4). This recruitment of Sli15 to cdc15-1 spindles was still Cdc14 dependent (fig. S4). When the FEAR pathway was inactivated in cdc15-1 cells (cdc15-1 Δspo12), Sli15 failed to bind to anaphase spindles (Fig. 3A and fig. S4). Moreover, the ratio of phosphorylated to nonphosphorylated Sli15 forms of cdc15-1 cells showed that FEAR-induced Cdc14 was able to dephosphorylate Sli15 (Fig. 3B). Thus, Cdc14 released by the FEAR network is sufficient to dephosphorylate and target Sli15 to anaphase spindles.

Fig. 3.

The FEAR-released Cdc14 targets Sli15-Ipl1 to spindles. (A) Sli15 spindle localization in the indicated cell types incubated for 2 hours at 37°C. (B) Phosphorylation of Sli15-3HA was analyzed in synchronized wild-type, cdc14-2, and cdc15-1 cells (G1, t = 0) at 37°C. The ratio of phosphorylated and nonphosphorylated Sli15 was determined from three independent experiments. (C) Synchronized Δcdc26 esp1-1 SLI15-GFP CFP-TUB1 cells were arrested in metaphase at 37°C, followed by induction (+galactose) or no induction (–galactose) of ESP1C1531A (t = 0). Sli15 localization with metaphase spindles was determined over time. n > 100. (D) The length of spindles with which Sli15 associates in synchronized populations incubated at 37°C. n > 50. Scale bars, 5 μm.

To confirm the finding that FEAR-controlled Cdc14 regulates Sli15, we made use of the observation that overexpression of protease-deficient separase (Gal1-ESP1C1531A) releases Cdc14 from nucleoli of cells arrested in metaphase in a manner that is independent of MEN activity (17). ESP1C1531A expression triggered binding of Ipl1 and Sli15 to metaphase spindles (Fig. 3C). Thus, separase regulates the relocalization of Sli15 to spindles via Cdc14 release.

In synchronized wild-type and cdc15-1 cells (having only FEAR active), Sli15 associated with anaphase spindles as soon as they started to elongate (Fig. 3D). In contrast, in Δspo12 cells (no FEAR) (4), only long spindles carried Sli15 (Fig. 3D). Thus, during normal cell-cycle progression, the FEAR pathway directs Sli15 to spindles in early anaphase. In the absence of FEAR function, MEN-regulated Cdc14 targets Sli15 to late anaphase spindles.

We next addressed the functional consequences of Sli15-Ipl1 deregulation. SLI156A cells did not show altered cell-cycle progression (15). However, the constitutive targeting of Sli15-Ipl1 to spindles impaired the high efficiency of chromosome segregation. The chromosome loss rate of wild-type cells was >10–4 per generation compared with 0.1 in SLI156A cells. SLI156A cells were still checkpoint proficient when tested in Cdc6-depleted cells (18), and Sli156A colocalized with the kinetochore marker Nuf2 in metaphase cells (15). Thus, Sli15-Ipl1 can be directed to spindles without disrupting essential kinetochore functions.

We then tested whether Sli15-Ipl1 stabilizes spindles. Anaphase spindles can be formed in the absence of separase activity by Tev protease cleavage of the Scc1Tev-cohesin complex. These spindles collapse after elongation (3). The lack of separase activity in this mitosis also stopped Sli15 from associating with these spindles (Fig. 4A). In contrast, Sli156A associated and stabilized anaphase spindles induced by Tev protease (Fig. 4A). Thus, the association of the Sli15-Ipl1 complex with spindles stabilizes them.

Fig. 4.

Sli15-Ipl1 stabilizes anaphase spindles and targets Slk19 to spindle midzones. (A) Strains ESM1908 (SCC1-TEV-HA3Gal1-9Myc-TEV-NLS2x10 Met3-CDC20 SLI15-GFP) and ESM1921 (SCC1-TEV-HA3Gal1-9Myc-TEV-NLS2x10 Met3-CDC20 SLI156A-GFP) were arrested in metaphase by Cdc20 depletion, and TEV protease expression was induced (t = 0). Samples were prepared for DNA staining and immunostaining for tubulin. (B and C) Cells with (B) SLK19-4GFP CFP-TUB1 or (C) ASE1-4GFP were incubated for 2 hours at 37°C. Slk19 and Ase1 localization were determined by fluorescence microscopy. Scale bars, 5 μm.

Ase1 and Slk19 are two proteins at the spindle midzone that are required for spindle stability (3, 19). Both proteins fail to go to spindle midzones when spindles elongate without separase activity (3) and may therefore depend on Sli15-Ipl1 for spindle localization. Slk19 was strongly focused within the middle of the spindle of wild-type, SLI156A, and cdc15-1 anaphase cells (∼90% of cells) (Fig. 4B). In ∼90% of cdc14-2 and sli15-3 cells, Slk19 failed to localize to the spindle midzone. Sli156A restored, in part, the defect of cdc14-2 cells (Fig. 4B). In 52% of cdc14-2 SLI156A cells, Slk19 was enriched in the middle of the spindle. Thus, Cdc14 and Sli15 are required to direct Slk19 to the spindle midzone. The rescuing effect of Sli156A suggests that Sli15-Ipl1 mediates the Cdc14 function in targeting Slk19 to the spindle midzone. In contrast, Ase1 associated with the spindle midzone of cdc14-2 and sli15-3 cells (Fig. 4C). This spindle midzone association of the stabilizer Ase1 in sli15-3 and cdc14-2 mutants partly explains why anaphase spindles of these cells do not break, although they collapse in cells that attempt anaphase in the absence of active separase (3). However, whereas anaphase spindles break in Δslk19 cells (20), they are stable in cdc14-2 cells. In both cell types, Ase1 is correctly positioned at the spindle midzone (3) (Fig. 4C). A likely explanation for this is that the spindle-associated but misplaced Slk19 of cdc14-2 cells has some residual stabilizing function.

With anaphase onset the mitotic spindle shows dramatic rearrangements, which are essential for spindle elongation, spindle stability, timely spindle disassembly at the end of anaphase, and regulation of cytokinesis (21). Here, we provide a molecular understanding of how the activation of separase at the metaphase-anaphase transition targets the chromosomal passenger proteins INCENP-Aurora to spindles. At the metaphase-anaphase transition, activated separase partially releases the phosphatase Cdc14 from the nucleolus (4, 17). Cdc14 removes inhibiting phosphates within the microtubule-binding domain of Sli15, a step that is essential and sufficient to target the Sli15-Ipl1 kinase complex to anaphase spindles. This timely regulation of Sli15-Ipl1 by Cdc14 is important to prevent aneuploidy, one of the hallmarks of cancer cells. In addition, Sli15-Ipl1 stabilizes anaphase spindles, in part by directing the Slk19 protein to the spindle midzone. The additional participation of Slk19 in the FEAR pathway indicates a complex pattern of regulation (4). A second spindle-stabilizing pathway controls Ase1. This second pathway is dependent on separase (3) but not on Cdc14 and Sli15-Ipl1. Given that the Caenorhabditis elegans ceCdc-14 also has a function in central spindle formation (22), it is likely that the spindle stabilizing role of Cdc14 phosphatases is conserved.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S4

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article