|Home | About | Journals | Submit | Contact Us | Français|
Cdc14-family phosphatases play a conserved role in promoting mitotic exit and cytokinesis by dephosphorylating substrates of cyclin dependent kinase (Cdk). Cdc14-family phosphatases have been best studied in yeast (for review see  ), where budding yeast Cdc14 and its fission yeast homolog Clp1 are regulated in part by their localization, with both proteins thought to be sequestered in the nucleolus in interphase. Cdc14/Clp1 are released from the nucleolus in mitosis, and in late mitosis a conserved signaling pathway termed the MEN/SIN acts through an unknown mechanism to keep Cdc14 and Clp1 respectively out of the nucleolus [3-6]. Here we show that the most downstream SIN component, the Ndr-family kinase Sid2, acts to maintain Clp1 in the cytoplasm in late mitosis by phosphorylating Clp1 directly and thereby creating binding sites for the 14-3-3 protein Rad24. Mutation of the Sid2 phosphorylation sites on Clp1 disrupts the interaction between Clp1 and Rad24, and causes premature return of Clp1 to the nucleolus during cytokinesis. Loss of Clp1 from the cytoplasm in telophase renders cells sensitive to perturbation of the actomyosin ring, but does not affect other functions of Clp1. Because all components of this pathway are conserved, this might be a broadly conserved mechanism for regulation of Cdc14-family phosphatases.
Despite considerable work on the SIN/MEN pathways in fission and budding yeast the key question of how each pathway acts to keep its respective Cdc14-family phosphatase out of the nucleolus has remained unknown. Previous studies showed that in late mitosis the SIN maintains Clp1 in the cytoplasm until cytokinesis is completed by regulating the nuclear shuttling of Clp1, perhaps through the action of the 14-3-3 protein Rad24 [7, 8]. Binding of Rad24 to Clp1 depends on the most downstream SIN pathway kinase Sid2 . 14-3-3 proteins are known to bind phosphopeptides, particularly the RXXpS motif , and RXXpS matches the predicted consensus phosphorylation site for Sid2 family kinases . Because Rad24 is restricted to the cytoplasm, we hypothesized that Sid2 phosphorylation of Clp1 might allow Rad24 to bind to and retain Clp1 in the cytoplasm. Therefore, we tested whether Sid2 could phosphorylate Clp1 directly, and whether Sid2 phosphorylation of Clp1 created binding sites for the 14-3-3 protein Rad24. We found that Sid2 kinase purified by tandem affinity purification (TAP) from yeast cells was capable of directly phosphorylating bacterially produced Clp1 (Figure 1A). Furthermore, Clp1 only bound Rad24 when it had been pre-phosphorylated by Sid2 kinase (Figure 1B).
To ascertain the significance of Clp1 phosphorylation by Sid2 in vivo, we sought to identify and mutate sites on Clp1 phosphorylated by Sid2. Phosphoamino acid analysis of in vitro phosphorylated Clp1 showed that it was phosphorylated exclusively on serine residues (Figure 1C). In vitro phosphorylated Clp1 was analyzed by two-dimensional phosphopeptide mapping, which identified 6 major tryptic peptides and a number of less abundant spots (Figure 1D). Analysis of in vitro phosphorylated Clp1 using mass spectrometry identified 5 sites of phosphorylation in Clp1 that were all within the C-terminal half (Figure 1E). Analysis of Clp1 purified from yeast cells using mass spectrometry identified the same 5 sites (Figure S1). Mutation of the 5 sites to alanine (Clp1-5A) significantly reduced the overall levels of Clp1 phosphorylation in vitro (Figure 1F, lane 3) and eliminated 5 of the 6 major tryptic phosphopeptides (Figure 1D). Through a combination of mutagenesis of additional sites followed by in vitro phosphorylation and 2 dimensional phosphopeptide analyses, we identified serine 493 as the last remaining site of significant phosphorylation. Mutation of S493 in addition to the previously identified 5 sites (Clp1-6A) eliminated the last major phosphopeptide, and caused almost complete elimination of phosphorylation of Clp1 by Sid2 in vitro (Figure 1F, lane 4, and data not shown). Mutation of any site singly, including S493, did not cause a major reduction in Clp1 phosphorylation in vitro, or binding to Rad24 in vitro (data not shown), suggesting that no single site is crucial. Bacterially expressed Clp1-6A retained wild-type in vitro phosphatase activity suggesting that the mutations did not grossly affect the structure of the protein (Figure 1G). All 6 sites of phosphorylation fit the consensus RXXS motif predicted for Sid2 family kinases . Mutation of an additional single RXXS motif at amino acid 499 (Clp1-7A) did not cause further reduction of overall level of phosphorylation (Figure 1F, lane 2), and resulted in reduced in vitro phosphatase activity of recombinant Clp1 and therefore was not pursued further (Figure 1G).
To determine the role of Clp1 phosphorylation by Sid2, Clp1-6A-GFP was integrated into the clp1+ locus such that it was expressed from the endogenous promoter, and was the only expressed copy of clp1 in the cell (supplemental methods). The level of Clp1-6A protein was similar to wild-type Clp1 (Figure 1H and data not shown). Like wild-type Clp1-GFP, Clp1-6A-GFP localized in interphase to the SPB and nucleolus, was released from the nucleolus as cells enter mitosis, localized to the kinetochores and actomyosin ring in early mitosis (Figure S2A). In anaphase cells Clp1-6A-GFP localized to the spindle, often appearing somewhat brighter than wild-type Clp-GFP (Figure 2A). In contrast, the Clp1-6A-GFP actomyosin ring signal began to diminish in anaphase compared to wild-type Clp1-GFP (Figure S2A). In telophase cells where the spindle has broken down but cells have not completed cytokinesis, wild-type Clp1 remained out of the nucleolus in the cytoplasm and faintly at the contractile ring until cytokinesis was completed. In contrast, Clp1-6A appeared to return to the nucleolus prematurely and was observed only faintly if at all in the contractile ring (Figure S2A-B). To examine the timing of Clp1-6A release more carefully, we performed time-lapse analysis of Clp1-GFP and Clp1-6A-GFP cells expressing a marker for the actomyosin ring (Rlc1-GFP) (Figure 2A, and Movies S1A-B) and quantified the nucleolar/cytoplasmic ratios of the GFP signal (Figure S2C). This analysis showed that Clp1-6A re-accumulated in the nucleolus as soon as the spindle broke down prior to actomyosin ring constriction (Figure 2A (30 min.), and Figure S2C). In fact, Clp1-6A most likely begins to accumulate in the nucleolus at anaphase onset since there are already significantly higher levels in the nucleolus at the time of spindle breakdown (Figure S2C). In contrast wild-type Clp1 did not re-accumulate in the nucleolus until 75 minutes later after the actomyosin ring had finished constriction and disappeared (Figure 2A (95 min.), and Figure S2C).
When cytokinesis is perturbed by low doses of the actin depolymerizing drug Latrunculin B, Clp1 remains cytoplasmic during the resulting cytokinesis delay (Figure 2B). In contrast, Clp1-6A returns to the nucleolus (Figure 2B, Figure S3A). This relocalization is similar to the behavior of wild-type Clp1 in SIN mutants . Interestingly, unlike clp1Δ cells, and like wild-type cells, clp1-6A cells halt further rounds of nuclear division when cytokinesis is delayed by Latrunculin B treatment and remain in a binucleate state with interphase microtubules (Figure S3B-C). Similarly, when the cytokinesis checkpoint is activated using the cps1-191 mutant defective in septum assembly and actomyosin ring constriction , clp1-6A cps1-191 mutant cells arrest like cps1-191 single mutant cells at restrictive temperature as binucleates with active SIN, interphase microtubules, and actomyosin rings consistent with the cytokinesis checkpoint being intact in clp1-6A cells (Figure S4).
We previously showed that when the SIN is constitutively activated in telophase by inactivating a component of its GTPase activating protein, Cdc16, cells undergo repeated rounds of cytokinesis and Clp1 persists in the cytoplasm once it is released from the nucleolus in the first mitosis . However, constitutive activation of the SIN in telophase is unable to keep Clp1-6A in the cytoplasm, and the mutant protein returns to the nucleolus (Figure 2C). Interestingly Clp1-6A-GFP appears to localize more strongly to the SPB than the wild-type protein in both cdc16 cells and cells arrested by the cytokinesis checkpoint (Figure 2B-C). Since the SIN is active in both situations, it suggests that the SIN may antagonize both nucleolar and SPB localization of Clp1.
We also expected that loss of Sid2 phosphorylation sites on Clp1 would disrupt binding of Rad24 to Clp1. To test this hypothesis, we examined whether bacterially produced GST-Rad24 would bind to Clp1-6A from yeast lysate. Unlike wild-type Clp1, Clp1-6A failed to bind to Rad24 (Figure 2D, Figure S5) suggesting that the cause of premature return of Clp1-6A to the nucleolus might be loss of Rad24 binding. To try to make a phosphomimetic version of Clp1 that bound Rad24 independently of Sid2, we mutated the six Sid2 phosphorylation sites on Clp1 to asparate residues, generating clp1-6D. However Clp1-6D did not bind Rad24 (data not shown) suggesting that aspartic acid residues cannot substitute in Clp1 for phosphorylated serines for 14-3-3 binding. The clp1-6D cells also displayed a general loss of function phenotype (data not shown) indicating that the asparate mutations caused defects in the structure of the protein and therefore this mutant was not analyzed further (data not shown).
The clp1-6A mutant allowed us to test the function of SIN mediated retention of Clp1 in the cytoplasm during telophase. We assayed whether the clp1-6A strain displayed any of the defects found in clp1Δ cells. Clp1 has roles in chromosome segregation, cytokinesis, the cytokinesis checkpoint, and regulation of cell size [5, 6, 12, 13]. Unlike clp1Δ, the clp1-6A mutation does not have negative interactions with mutations in genes involved in chromosome segregation such as dis1 (Figure S6A, and data not shown). Clp1 negatively regulates Cdc25 explaining both why clp1Δ cells have a reduced cell size and why overexpression of Clp1 causes a block in mitotic entry and cell elongation [5, 6, 14, 15]. Clp1-6A presumably is able to regulate Cdc25 normally since clp1-6A cells have a wild-type cell size and overexpression of Clp1-6A blocks mitotic entry like wild-type Clp1 (Figure S6B-C). As shown earlier, clp1-6A is also wild type for the cytokinesis checkpoint. It has been previously shown that the main function of Clp1 in the cytokinesis checkpoint is to promote SIN activity . Consistent with this, clp1-6A, unlike clp1Δ, did not show any negative interactions with the SIN mutants sid1-239, sid4-A1, cdc11-136, sid2-250, cdc14-118, spg1-B8, or mob1-R4 (data not shown). However, we did find that clp1-6A is sensitive to perturbations of the actomyosin ring, showing sensitivity to low doses of the actin inhibitor Latrunculin B, and negative genetic interactions with several mutations affecting actomyosin ring assembly and cytokinesis (Figure 3). In particular, clp1-6A cells had negative interactions with the actomyosin ring assembly mutants cdc15-140, mid1-18, and myo2-E1, with the double mutants showing synthetic growth defects at semi-permissive temperatures (Figure 3A). Examination of double mutant cells in liquid culture at semi-permissive temperatures showed enhanced cytokinetic defects (Figure S7). For example, after 8 hours at 30°C both myo2-E1 and clp1-6A myo2-E1 cells showed single nuclei separated by relatively complete but misformed septa. In contrast, clp1Δ myo2-E1 cells, which lack the cytokinesis checkpoint, have only occasional partial septa and are highly multinucleate (Figure S8). However, the myo2-E1 single mutant, unlike clp1-6A myo2-E1, was able to complete cytokinesis since there was a significant number of mononucleate cells and fewer tetranucleate cells (Figure S7). Overall, these results suggest that maintenance of Clp1 in the cytoplasm is important for completion of cytokinesis when the cell division apparatus is perturbed.
Previous studies showed that both the SIN and Clp1 are required for a cytokinesis checkpoint that, in response to perturbation of the cell division apparatus, halts further cell cycle progression until cytokinesis can be completed [6, 11, 12]. When the cell division apparatus is perturbed, Clp1 is required to maintain SIN activity, and the SIN is required keep Clp1 in the cytoplasm [6, 12]. Based on these results, it was proposed that the SIN and Clp1 act in a positive feedback loop whereby they promote each others activity, with the SIN promoting cytoplasmic retention of Clp1, and cytoplasmic Clp1 acting to maintain SIN activity [6, 12]. However we found that cytoplasmic retention of Clp1 is not required to maintain SIN activity and halt cell cycle progression when the actomyosin ring is damaged, (Figure S3, and S4), but it is required to complete cytokinesis (Figure 3), presumably by maintaining the cell division apparatus. This is consistent with recent results showing that inability to target Clp1 to the actomyosin ring causes similar cytokinetic defects when the ring is perturbed but not cytokinesis checkpoint defects .
Although many studies have shown that the SIN and MEN pathways regulate the conserved phosphatases Clp1 and Cdc14 respectively to keep them out of the nucleolus during late mitosis, the mechanism has been unclear. Here we show that the most downstream kinase in the SIN pathway, Sid2, phosphorylates Clp1 to promote binding of the 14-3-3 protein Rad24. Binding to Rad24 results in cytoplasmic retention of Clp1. A recent study showed that the Cds1 kinase phosphorylates Clp1 on similar residues to promote cytoplasmic retention of Clp1 in response to blocks in DNA replication , suggesting that the same mechanism could be used by multiple inputs to regulate Clp1. In addition, the Sid2 homolog in animal cells, the Lats1/2 tumor suppressor, might regulate targets using a similar strategy. Lats1/2 phosphorylates the oncogene YAP1 causing it to bind a 14-3-3 protein and be retained in the cytoplasm [18-22]. Given that mammalian cells have at least two Cdc14 homologs, it is tempting to speculate that they too may be regulated through Lats1/2 phosphorylation and 14-3-3 binding as we observe in yeast.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.