PP2A-Cdc55 phosphatase coordinates actomyosin ring contraction and septum formation during cytokinesis

Eukaryotic cells divide and separate all their components after chromosome segregation by a process called cytokinesis to complete cell division. Cytokinesis is regulated by exclusive elements of the process, and by some mitotic exit regulators. The mitotic kinases Cdc28-Clb2, Cdc5, and Dbf2-Mob1 phosphorylate cytokinetic proteins in budding yeast, but very little is known about the phosphatases regulating cytokinesis. The PP2A-Cdc55 phosphatase regulates mitosis counteracting Cdk1- and Cdc5-dependent phosphorylations. This prompted us to propose that PP2A-Cdc55 could also regulate cytokinesis by counteracting the mitotic kinases. Here, we demonstrate by in vivo and in vitro assays that PP2A-Cdc55 dephosphorylates the F-BAR protein Hof1 and the chitin synthase Chs2, two components of the Ingression Progression Complexes (IPC) involved in cytokinesis regulation. Primary septum formation and actomyosin ring contraction are impaired in absence of PP2A-Cdc55. Interestingly, the non-phosphorylable version of Chs2 rescue the asymmetric AMR contraction observed in absence of Cdc55, indicating that timely dephosphorylation of the IPC proteins by PP2A-Cdc55 is crucial for proper actomyosin ring contraction and septum formation. These findings reveal a new mechanism of cytokinesis regulation by the PP2A-Cdc55 phosphatase and extend our knowledge in the involvement of multiple phosphatases during cytokinesis.


Introduction 51
Cytokinesis is the final event of the cell cycle and mediates the physical separation 52 of mother and daughter cells. It is a highly ordered and regulated process that is 53 conserved among eukaryotes. The genome, packed into chromosomes, is 54 In conclusion, the phosphatase assays indicate that PP2A Cdc55 counteracts 279 phosphorylation in Hof1 and Chs2, reinforcing the idea that they are direct PP2A Cdc55 280 substrates. PP2A Cdc55 is known to counteract Cdk1 phosphorylation, but this is the 281 first time, to our knowledge, that a Dbf2 substrate has been shown to be 282 between cdc55Δ and wild-type cells (Fig. S3), suggesting that septin dynamics might 302 not be greatly affected by the absence of Cdc55. 303 13 304 PP2A Cdc55 is involved in bud morphology through actin polarization and cell-wall 305 synthesis [50]. Polarized growth must be temporally coordinated with the events of 306 the cell cycle. Actin cytoskeleton polarization at the site of bud emergence is 307 triggered at Start by the kinase activities of Cln1,2-Cdc28. In addition, Clb2-Cdc28 308 restrains repolarization to the mother bud neck [70]. Only when the Clb2-Cdc28 309 kinase is inactivated at the end of mitosis, the actin cytoskeleton is directed to the 310 neck to complete cytokinesis. Actin is then recruited in an Iqg1-dependent manner to 311 the division site and it is essential for the formation of the AMR [9,43]. We envisage 312 the possibility that PP2A Cdc55 affects actin cytoskeleton polarization, not only during 313 the period of apical growth, but also at the end of mitosis. For this reason, we 314 wondered whether actin polarization was defective in cdc55∆ cells. To examine this 315 possibility, actin filaments were stained with rhodamine-labelled phalloidin in cells 316 progressing through mitosis and cytokinesis. Actin signal was found to be 317 depolarized in metaphase-arrested cells both in the wild-type and cdc55Δ strains, as 318 expected (Fig. S4a). When cells completed mitosis, wild-type cells polarized the 319 actin cytoskeleton to the division site. In cdc55Δ mutants, actin was also repolarized 320 in a timely manner at the division site during cytokinesis (Fig. S4a). However, in a 321 subpopulation of cdc55Δ cells, actin polarization at the new bud was observed 322 before cytokinesis had been completed (Fig. S4b). We were able to observe actin 323 signals simultaneously at the division site during cytokinesis and at the new bud site 324 in 12.5% of cdc55Δ mutant cells, whereas there was no premature actin 325 repolarization at the new bud site in wild-type cells (Fig. S4b). The premature actin 326 localization indicates that actin re-polarization of the next cell cycle occurs before cell 327 division in a fraction of cdc55Δ cells. Therefore, PP2A Cdc55 could act to prevent actin 328 re-polarization until cytokinesis is completed. 329 Cytokinesis in budding yeast is accomplished by the concerted action of the 331 actomyosin contractile ring (AMR) and the formation of the septum. We investigated 332 how these two processes occur in the absence of Cdc55. First, to clarify whether the 333 actomyosin contractile ring is functional, we analyzed the localization of a Myo1-334 tdTomato fusion protein in cells progressing through mitosis and cytokinesis by the 335 synchronous release from the metaphase arrest by Cdc20 depletion. It has been 336 described that Myo1 localizes to the division site immediately after budding, so in 337 metaphase-arrested cells Myo1-tdTomato was localized at the bud neck in control 338 cells. Similarly, Myo1-tdTomato was detected at the bud neck at metaphase in 339 cdc55Δ strains. The signal size is reduced during anaphase, reflecting the 340 contraction of the AMR, until the signal becomes a single dot and finally disappears 341 (Fig. 3a). The dynamics of AMR contraction are similar in the two strains (6 minutes 342 from the start of contraction to the disappearance of the signal), but the "dot" signal 343 collapsed to one side of the division site in almost all the cdc55Δ cells ( Fig. 3b  344 cdc55Δ). We observed that the AMR contraction was asymmetric with respect to the 345 centripetal axis in 96% (N=30) of cdc55Δ cells (Fig. 3b). In order to check whether 346 the asymmetric Myo1 signal is not due to an adaptive mechanism of the cdc55Δ 347 deletion mutant cells, we investigated the Myo1 contraction after inducing the Cdc55 348 degradation during metaphase using an auxin-degradation system [71]. We observed 349 that after the degradation of Cdc55, Myo1 contraction was asymmetric (86% of cells) 350 as in cdc55Δ cells (Fig. 3c). The results indicate that the lack of PP2A Cdc55 activity 351 promotes the asymmetric contraction of Myo1. 352 We next examined whether Myo1 asymmetry could also be detected in the inactive 353 version of Cdc55 (cdc55-ED). We synchronized cells at the metaphase-anaphase 354 transition by Cdc20 depletion and analyzed the contraction of the Myo1-tdTomato 355 after the release. We observed that 65% of cdc55-ED mutant cells showed an 356 asymmetric Myo1-tdTomato signal upon contraction (Fig. 3c) To confirm that the asymmetric AMR contraction phenotype was not affected by the 366 tdTomato-tagging or the synchronization method, we compared fixed cdc55Δ cells 367 tagged with Myo1-GFP upon release from metaphase-arrested cells by Cdc20 368 depletion with cdc55Δ cells tagged with Myo1-tdTomato synchronizing cells with 369 alpha factor in G1. The cdc55Δ Myo1-GFP tagged cells showed an asymmetry of 370 87.5% of the population (N=23) (Fig. S5), representing a similar phenotypic 371 penetrance to that of Myo1-tdTomato in cdc55Δ cells. This asymmetry was not 372 observed in WT cells (1 asymmetric cell in the 41 cells studied). 373 Next, we performed the assay synchronizing cells in G1 by alpha factor in absence 374 of Cdc55. cdc55∆ cells enter mitosis with a delay due to compromised Cdk1 activity 375 because of inhibitory Cdc28-Y19 phosphorylation [72]. To correct for this delay, we 376 introduced the cdc28_Y19F allele, which is refractory to Cdk1 inhibition. The cdc55Δ 377 cells containing cdc28_Y19F progressed normally through mitosis upon release from 378 metaphase-arrested cells [56]. Again, we observed that Myo1-tdTomato constriction 379 was asymmetric in 80% of the cdc55Δ cdc28_Y19F cell population (Fig. S5). 380 Therefore, the asymmetric Myo1 signal in the absence of Cdc55 was observed 381 independently of the epitope and the synchronization method used. at metaphase by Cdc20 depletion and released them into mitosis and took images 388 30-45 min after the release when we found cells at cytokinesis. Calcofluor intensity 389 was then measured and quantified in wild-type and cdc55Δ cells (Fig. 3d). There 390 was a statistically significant reduction of the 45% in the intensity of the calcofluor 391 staining in the absence of Cdc55 compared with control cells. Chitin is incorporated 392 in primary septa (PS) and secondary septa (SS), so we cannot distinguish at which 393 of them this reduction occurs. For this reason, we repeated the calcofluor staining in 394 cells containing a deletion of CHS3, the chitin synthase responsible for secondary 395 septum formation [21]. A reduction of 70% in the calcofluor intensity was observed in 396 the absence of Cdc55 relative to control cells (Fig. 3d). The above results suggest 397 that primary and secondary septum formation is reduced in cells lacking PP2A Cdc55 398 activity. Our results suggest that, similar to IPCs, PP2A Cdc55 has a role coordinating 399 AMR contraction with septum formation, probably by dephosphorylating IPC 400

components. 401
To determine whether ingression of the plasma membrane could occur in cdc55Δ 403 cells, we performed time-lapse video microscopy of cells expressing Myo1-tdTomato 404 to follow AMR contraction, and the small G-protein Ras2 fused to 3 copies of GFP to 405 study plasma membrane dynamics. The visualization of the plasma membrane at 406 the site of division revealed no cytoplasm connection between mother and daughter 407 cells, confirming that cytoplasmic division had been completed in control and cdc55Δ 408 cells (Fig. S6). In conclusion, cdc55Δ cells have a defective AMR "collapsed" to one 409 side of the division site and a reduction in septum formation, but nevertheless 410 manage to complete cell division. 411

412
To confirm the defects in septum formation we investigated the cytokinetic structure 413 by transmission electron microscopy (TEM). We synchronized cells at the 414 metaphase-anaphase transition by Cdc20 depletion and captured images 40-50 415 mins after metaphase release. Cells synthesizing the primary septum were identified 416 and the structure of the septum was then examined. In wild-type cells, we observed 417 that the PS formed on both sides of the membrane invagination (Fig. 4a), as 418 expected. Once the PS was finished, the two SS was formed on both sides of the PS 419 ( Fig. 4b). Finally, the PS is degraded, and the cells physically separated (Fig. 4c). 420 Conversely, in cdc55∆ cells, the PS formed on only one side of the division site (Fig.  421 4d). From the cells performing primary septum formation, we quantified 85% (N=25) 422 with asymmetric PS in absence of Cdc55. Some cells showed aberrant, thicker 423 structures with diverse morphologies that resemble the remedial septa (Fig. 4e). The 424 remedial septa were first described in IPC mutant cells, i.e., myo1∆ and chs2∆ [23]. To determine whether PP2A Cdc55 regulates IPC localization, we used GFP-tagged 454 IPCs strains in controls and cdc55∆ cells. Consistent with the results from the Myo1 455 protein, Iqg1, Hof1, Cyk3, Inn1, and Chs2 were contracted asymmetrically at the 456 division site in cdc55Δ cells (Fig. 5a). These findings indicate that the lack of 457  Our results suggest that Cdc55 regulates Hof1 and Chs2-dependent processes 500 during cytokinesis, and that these proteins are Cdc55 substrates. To screen for new 501 PP2A Cdc55 substrates during mitosis, we previously performed a global study of the 502 PP2A Cdc55 phosphoproteome by a quantitative phosphoproteomic analysis based on 503 SILAC labeling [58]. Wild-type cells were labeled using 13 C6-lysine and 13 C6-arginine 504 (heavy), and cdc55∆ mutant cells were grown in the presence of unmodified arginine 505 and lysine (light). Cells were arrested in metaphase and protein extracts were 506 prepared. Phosphopeptides were enriched by TiSH-based enrichment. 507 Phosphopeptide analysis of the heavy/light-labelled cells was done by LC-MS/MS. 508 The screening revealed that a phosphopeptide corresponding to Chs2 protein was 509 hyperphosphorylated in cdc55∆ mutant cells. The Chs2 peptide contained one Cdk1 510 minimal S/TP site: S133 (Figure 6a). The phosphosite was detected with the highest 511 confidence (pRS site probability 99.9% and p-value < 0.00001) and was also 512 identified in a second phospho-proteomic study for cdc55∆ mutant cells [46]. This 513 result confirms that Chs2 is a PP2A Cdc55 substrate. The S133 is one of the 6 Cdk1 514 consensus sites (S/T-P) previously reported to be phosphorylated by Cdk1 [74]. In 515 addition, Chs2 S14, S60, S69 and S100 can be efficiently dephosphorylated by 516 Cdc14 while the other two sites S86 and S133 can also be dephosphorylated by 517 Cdc14 although less efficiently [39]. It is important to remember that Cdc14 is 518 prematurely release from the nucleolus and active as a phosphatase in cdc55∆ 519 mutant cells [56,57]. Therefore, despite the fact that the two phosphatases can 520 contribute to the dephosphorylation of Chs2, in the phosphoproteomic analysis the 521 identification of a hyperphosphorylated peptide in absence of Cdc55 implies that this 522 site is not mainly affected by Cdc14 (the phosphosites mainly dephosphorylated by 523 Cdc14 will be identified as hypophosphorylated since Cdc14 is prematurely active in 524 cdc55∆ cells). To advance in the mechanistic understanding of the role of PP2A Cdc55 525 22 in cytokinesis we prepared phospho-mutants for Chs2 and determined the 526 phenotypic effects of these mutations. 527 To demonstrate that the cytokinetic defects observed in cdc55∆ mutant cells were 528 due to the increase of Chs2 phosphorylation levels, we introduced the chs2-6A-YFP 529 mutant, were all the SP sites were mutated to alanine [75], in cdc55∆ mutant cells. 530 Myo1 and Chs2 signals were asymmetric in the presence of the control CHS2-YFP 531 (79% of cells; N=13) in absence of Cdc55, as expected ( Fig. 6b) Remarkably, in the 532 chs2-6A-YFP non-phosphorylable version Myo1 and Chs2 localization became 533 symmetric in 78% of cells (N=18), indicating that the non-phosphorylable chs2-6A 534 mutant rescued the asymmetric localization of Myo1 in cdc55∆ mutant cells (Fig. 6b). 535 The Chs2-YFP signal was detected earlier at the bud neck, probably due to the 536 overexpression of Chs2, but the Myo1 and Chs2 contraction time were similar in the 537 presence of Chs2 and chs2-6A (Fig. 6c). We can conclude that timely 538 dephosphorylation of Chs2 by PP2A-Cdc55 is required for proper AMR contraction. 539 540 These findings demonstrate that the lack of PP2A Cdc55 activity provokes a collapse of 541 the AMR, representative of a dysfunctional AMR, and that it also alters the residence 542 time at the division site of IPC proteins. Therefore, PP2A Cdc55 dephosphorylation of 543 the IPC subunits is essential for maintaining proper AMR contraction and septum 544 formation; both required for an efficient cytokinesis. formation, we propose that PP2A Cdc55 is also involved in this efficient mechanism of 576 coordination through IPC dephosphorylation (Fig. 7). 577

578
The role of PP2A Cdc55 during cytokinesis 579 In the absence of Cdc55, Myo1 was constricted and unanchored in a timely fashion, 580 but the constriction was displaced from the central axis, becoming asymmetric (Figs. 581 3a and S5). This phenotype is characteristic of IPC mutants [6,25,67] and denotes a 582 dysfunctional AMR. 583 Consistent with this, Iqg1, Cyk3, Hof1, Inn1, and Chs2 localization were also 584 asymmetric upon AMR constriction (Fig. 5a). The asymmetric Cyk3 and Inn1 585 localization have also been described in hof1∆ mutants [13,19]. Hof1 586 dephosphorylation by PP2A Cdc55 (Fig. 1b)  also contribute to the asymmetry seen in the IPC proteins. 593 594 PP2A Cdc55 also regulates septum formation. Chitin incorporation was decreased in 595 the cdc55∆ and cdc55∆ chs3∆ cells (Fig. 3d), indicating that Chs2 activity might be 596 impaired. The electron microscopy study of septum formation demonstrated that PS 597 is still formed in cdc55∆ cells, being predominantly asymmetric on one side of the 598 division site during its formation (Figs. 4d, j). PS was reported to be asymmetric in 599 around 50% of hof1∆ mutant cells [35]; while more than 80% of the cells showed 600 asymmetric PS in cdc55∆ cells. This difference in magnitude can be explained taking 601 into consideration that at least two IPC, Chs2 and Hof1, are PP2A Cdc55 substrates. 602 603 Later on, a remedial septum is formed that contains even cytosol fractions 604 embedded in the septa (Fig. 4). This remedial septum formation has been seen 605

Yeast strains, plasmids and cell-cycle synchronization procedures 646
All yeast strains used in this study were derivatives of W303 and are listed in Table  647 S1. Epitope tagging of endogenous genes and gene deletions were performed by 648 gene targeting using polymerase chain reaction (PCR) products [81]. Cell 649 synchronization using α-factor and metaphase arrest by Cdc20 depletion and entry 650 into synchronous anaphase by Cdc20 re-introduction were performed as previously 651 described [82]. To obtain the pGAL1-CDC55 construct, the DNA fragment containing 652 the GAL1 promoter was cut with SpeI and subcloned into the CDC55 containing 653 plasmid previously digested with NheI. For kinase assays, immunoprecipitations were performed as above and incubated 697 with α-Pk clone SV5-Pk1 (Serotec) antibody. Beads were washed with ten volumes 698 of lysis buffer and twice with the kinase reaction buffer (50 mM Tris-HCl, pH7.4, 10 699 mM MgCl2, 1 mM DTT). The kinase reaction (50 mM Tris-HCl, pH7.4, 10 mM MgCl2, 700 1 mM DTT, 5 mM β-glycerophosphate, 25 μM ATP, 10 mCi/mL 32 gamma-ATP and 2 701 ng-1 μg of protein substrate (plus 2 mM EGTA for the Cdc5 kinase assay) was 702 incubated at 30ºC for 30 minutes for 6His-Inn1, 6His-Hof1, and Streptag-Chs2, and 1 703 hour for 6Hys-Cyk3. Kinase assays were stopped by placing the tubes on ice. The 704 supernatant containing the phosphorylated substrate was separated from the 705 magnetic beads and stored at -80ºC. An aliquot of the kinase assay was mixed with 706 SDS-PAGE loading buffer, proteins were separated by electrophoresis, transferred 707 to nitrocellulose membranes and radioactivity detected in a Typhoon FLA950 (GE 708 Healthcare). Immunopurified protein was quantified by western blot and the 709 membrane was stained with Coomassie to detect the recombinant substrate.