E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ loca te /yexc r

Research Article

Regulation of multiple cell cycle events by Cdc14 homologuesin vertebrates

Liliana Krasinska1,2, Geoffroy de Bettignies, Daniel Fisher2, Ariane Abrieu,Didier Fesquet⁎, Nathalie Morin⁎

Centre de Recherches de Biochimie Macromoleculaire, Université Montpellier II, CNRS UMR5237 1919, Route de Mende,34293 Montpellier Cedex 5, France

A R T I C L E I N F O R M A T I O N

⁎ Corresponding authors. Fax: +33 467 52 15 5E-mail addresses: didier.fesquet@crbm.cnAbbreviations: MT, microtubule; GLU, dety

1 On leave from the Medical University of G2 Institute de Genetique Moleculaire de Mo

0014-4827/$ – see front matter © 2007 Elsevidoi:10.1016/j.yexcr.2006.12.022

A B S T R A C T

Article Chronology:Received 18 September 2006Revised version received6 December 2006Accepted 20 December 2006Available online 10 January 2007

Whereas early cytokinesis events have been relatively well studied, little is known about itsfinal stage, abscission. The Cdc14 phosphatase is involved in the regulation of multiple cellcycle events, and in all systems studied Cdc14misexpression leads to cytokinesis defects. Inthis work, we have cloned two CDC14 cDNA from Xenopus, including a previously unreportedCDC14B homologue.We useXenopus and human cell lines and demonstrate that localizationof Cdc14 proteins is independent of both cell-type and species specificity. Ectopicallyexpressed XCdc14A is centrosomal in interphase and localizes to themidbody in cytokinesis.By using XCdc14A misregulation, we confirm its control over different cell cycle events andunravel new functions during abscission. XCdc14A regulates the G1/S and G2/M transitions.We show that Cdc25 is an in vitro substrate for XCdc14A and might be its target at the G2/Mtransition. Upregulatedwild-type or phosphatase-deadXCdc14A arrest cells in a late stage ofcytokinesis, connected by thin cytoplasmic bridges. It does not interfere with central spindleformation, norwith the relocalization of passenger protein and centralspindlin complexes tothemidbody. We demonstrate that XCdc14A upregulation prevents targeting of exocyst andSNARE complexes to the midbody, both essential for abscission to occur.

© 2007 Elsevier Inc. All rights reserved.

Keywords:Cdc14CytokinesisAbscissionexocystSNARE

Introduction

Themolecular events that couplemitotic exit with cytokinesishavebeen best characterized in yeast. Themitotic exit network(MEN) in budding yeast [1] and septation initiation network(SIN) in fission yeast are homologous signaling pathways withhighly conserved components [reviewed in 2]. At the base ofthese two pathways, however, one key downstream compo-nent is found: the dual specificity phosphatase, Cdc14p in S.cerevisae and Flp1 in S. pombe. The activation of Cdc14p in late

9.rs.fr (D. Fesquet), nathalirosinated tubulin; WT, wdansk, Poland.ntpellier, CNRS UMR 5535

er Inc. All rights reserved

anaphase is required for mitotic exit and it promotes cyclindependent kinase inactivation through APC/Cdh1 regulationand Sic1 accumulation [1,3]. In contrast, although Flp1 canfunctionally complement the loss of Cdc14p, the SIN pathwayis not required formitotic exit per se. Instead, it induces a G2/Marrest as part of a cytokinesis checkpoint [4,5].

Although potential homologues of MEN components havebeen identified in higher eukaryotes, their functions andrelationship with potential MEN-like regulatory pathways isstill amatter of debate. Inhumans, twoCdc14 isoforms, hCDC14

e.morin@crbm.cnrs.fr (N. Morin).ild-type; PD, phosphatase-dead; RT, room temperature

1919, Route de Mende, 34293 Montpellier Cedex 5, France.

.

1226 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

A and hCDC14B, have been described [6,7]. Similarly to Cdc14p,hCDC14A can dephosphorylate Cdh1 in vitro and thus mightpromote activation of APC in late anaphase [8]. And, similarly toFlp1, hCDC14A and hCDC14B might regulate the activity ofCDK1-cyclin B through Cdc25 phosphatase at the G2/M transi-tion [9]. Deregulation of hCDC14A levels revealed its role inregulation of the centrosomecycle [6,7]. Finally, RNAidepletionsof C. elegans Cdc14, CeCdc14, although leading to contradictoryresults, suggest a role for the phosphatase in cytokinesis aswellas preventing extra cell divisions during quiescence [10,11].Interestingly, whereas Gruneberg et al. show that CeCdc14depletion leads to complete loss of central spindle and cytokin-esis defect, Saito et al. demonstrate yet different consequencesand a role for CeCdc14 in G1 to ensure cell cycle arrest. Thediscrepancies between the two studies have been ascribed tostrain and RNAi concentration differences [12].

The complexity of the observed phenotypes might wellcorrespond to involvement of Cdc14 in multiple regulatoryevents throughout the cell cycle. Some controversy oversubcellular localization of human Cdc14 homologues exists,and the studies of phenotypes due to overexpression/depletionhave not yet clearly identified Cdc14 functions. Nevertheless, acommon feature of all Cdc14 proteins studied to date is the tightregulation of their cellular localization, which may also controltheir activity. In yeast, Cdc14 is sequestered inside thenucleolusand released in a regulated manner [13–16]. In C. elegans,CeCdc14 localizes to the central spindle and the midbody [10].In humans, hCDC14B is found in nucleoli during interphase,becoming diffuse throughout the cell atmitosis entry. hCDC14Alocalizes to the centrosome in interphase and is lost from thisstructure in earlymitosis [6,7]. However, further studies showedthat both hCDC14A and hCDC14B could localize to themidbody[17,18], which was not described in previous studies.

Another feature of Cdc14 in all systems studied is itsregulation of cytokinesis. CeCdc14 colocalizes at the centralspindle with a component of the centralspindlin complex, themitotic kinesin MKLP1/ZEN-4 [19], which it activates bydephosphorylation. In parallel, coordination of chromosomesegregation and cytokinesis is assured by relocation ofchromosomal passengers (including INCENP, Aurora B, Survi-vin, and Borealin) from metaphase kinetochores to theanaphase central spindle. While Cdc14p mediated depho-sphorylation of yeast INCENP, sli5, is required for thistranslocation at the onset of anaphase [20], a similar role forCdc14 in higher eukaryotes has not been investigated. HumanCDC14A has recently been observed at the central spindle [17],a localization that would be consistent with its proposedregulation of potential substrates, MKLP1 and INCENP.

In Xenopus, Kaiser et al. [21] identified two Cdc14A para-logues but no Cdc14B homologue. Microinjection of Xenopusembryos with antibodies against XCdc14 interfered with thecell cycle, although the phenotype was not characterized.Given the still incomplete data on the involvement of Cdc14 inthe control of cell cycle events, the different localizationsascribed to human CDC14A and B, and the unidentified cellcycle phenotype in Xenopus, in this report we attempt to betterdefine the impact of vertebrate Cdc14 on cell cycle progres-sion. To do so, we use amixed Xenopus/ HeLa heterologous cellsystem, to enable study of functional conservation. Weidentify two Xenopus paralogues, XCdc14A and XCdc14B. We

demonstrate that ectopically expressed XCdc14A is centroso-mal during interphase while at the end of mitosis it isrecruited to the midbody. Ectopic XCdc14A expression inter-feres with three important stages of cell cycle progression, theG1/S and G2/M transitions, and a late stage of cytokinesis, theabscission.We provide evidence that XCdc14A can prevent theG2/M transition by dephosphorylating Cdc25. During cytokin-esis, it prevents the recruitment of SNARE/exocyst complexes,required for abscission to occur.

Materials and methods

cDNA cloning, immunization procedures, proteinpurification and antibodies

Database (http://www.ncbi.nlm.nih.gov) searches were usedto identify Expressed Sequence Tag (EST) encoding proteinhomologous to metazoan Cdc14 (GenBank accession nos.BG018336 and BJ08353). cDNAs were obtained from eitherOpen Biosystems, USA or the Xenopus laevis cDNA ressource.Sequence analysis with DNA Strider software revealed thatthey encode two different proteins related to Cdc14, which wecalled XCdc14A and XCdc14B. These sequences have beendeposited in GenBank (GenBank accession nos. AAV66581 andAAV66582, respectively).

For antibody production, PCR amplified XCdc14A (aa 1-489)and B ORF were subcloned into pET100d using a Topo cloningprocess (Invitrogen) and the 6His-tagged full-length XCdc14AandBwere expressed in E. coli andpurified over TALONaffinityresin according to manufacturer instruction (BD Biosciences).For in vitro transcription/translation, theCdc14AORFwasPCR-amplified with primers containing an XmaI site, with 6Hissequence included in the reverse primer, and subcloned intoSmaI site of pCS2 vector. XCdc14A wild type and phosphatasedead form were translated in rabbit reticulocyte lysate (TNTSp6 Quick Master Mix; Promega) in the presence of [35S]methionine, according to the manufacturer's instructions.

For GFP expression studies, full-length wild type XCdc14AcDNA was subcloned into the XmaI site of pEGFP-C1 vector(CLONTECH). A point mutation in XCdc14A (phosphatase-deadmutant C276S) was generated using QuickChange site directedmutagenesis kit according to the manufacturer's protocol(Stratagene). All constructswere confirmed byDNA sequencing.

Xenopus egg extracts and in vitro kinase andphosphatase assays

Interphase egg extracts were prepared as described [22],snap-frozen in liquid nitrogen and stored at −80 °C. For G2/Mtransition studies, spermchromatin (1000 spermheads/μl)wasadded to interphase extract (40 μl) supplemented with gfp-NLS-GST construct, and incubated at RT for 20 minutes. Then1,5 μl in vitro translated XCdc14A (WT or PD) or reticulocytelysate (RL) was added, and incubated for 70 min. Entry intomitosis was induced with recombinant GST-tagged Δ90 cyclinB. For assessment of XCdc25 phosphorylation status, samplesof extract (1 μl) were frozen in liquid nitrogen. For phosphataseassay, approximately 2 μg of bacterially produced GST-hCDC25C (clone gift from M. Morris) was phosphorylated by

1227E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

CDK1/cyclin B complex (New England BioLabs), supplementedwith 2 μCi γ 32P ATP. Reactions were incubated at RT for10 minutes and stopped by addition of 5 mM EDTA. Phos-phorylated CDC25C was then incubated with 1,5 μg ofrecombinant 6His-XCdc14A for 1 hour at RT. Reaction wasstopped by the addition of sample buffer, separated by SDS-PAGE and CDC25C was visualized by Coomassie staining andautoradiography.

Western blot analysis

Xenopus egg or cell extracts were separated on 12% SDS-PAGEand transferred to nitrocellulose filters (BioRad). After blockingin 3% skimmedmilk in TBST buffer (0,9%NaCl; 100mMTrisHClpH 7,5; 0,1% Tween 20) for 1 hour, the membranes wereincubatedwithprimary antibodies diluted inTBST for 1hour atRT or overnight at 4 °C. Antibodies used were: anti-XCdc14Adepleted of XCdc14B (through incubation for 2 hours at RT ofanti-XCdc14A antibodies with ∼30 μg of recombinant 6His-tagged XCdc14B transferred to a nitrocellulose membrane),anti-XCdc25 (gift from Thierry Lorca). After incubation withproper secondary antibodies, Western Lightning Chemilumi-nescence Reagent kit (PerkinElmer Life Sciences) was used.

Cells and immunofluorescence

XL2 cells were grown in Leibowitz media (L15medium, GIBCO)supplementedwith 10%FCS (inactivated foetal calf serum), 100U/mL penicillin and 100 μg/mL streptomycin, and incubated at25 °C in the absence of any additional CO2. For immunofluor-escence, cells were fixed in 100% cold methanol at −20 °C for5minutes, and blocked in PBS+0,05% Triton-X-100+2% bovineserum albumin (BSA). All the subsequent steps were per-formed in the same buffer. Cells were incubated for 1 hour atRT with primary antibodies, washed five times, incubated for30 minutes with secondary antibodies, washed five times,followed by 5 minute staining of DNA with DAPI. Coverslipswere mounted with Mowiol and sealed with nail varnish.

Forwestern blotting, cells grown on 35mmplateswere lysedin TLB buffer (20 mMTris, pH 7,5; 10% glycerol; 1% Triton X-100;0,137 M NaCl; 25 mM β-glycerophosphate; 2 mM EDTA; 0,5 mMDTT; 1 mM orthovanadate; 2 mM sodium pyro-phosphate;10 mg/ml leupeptin; 1 mM phenylmethylsulfonyl fluoride),supplemented with protease inhibitor cocktail (Roche AppliedScience) and 1 μMmicrocystin, then blocked in Laemmli buffer,sonicated and loaded onminigel for analysis.

HeLa cellsTransfections were carried out using FuGENE 6 (Roche AppliedScience) according to the manufacturer's instructions.

For immunofluorescence, cells were fixed either with 100%coldmethanol for 5minutes at −20 °C or with DSPmethod [23].For actin staining, cells were fixed in 3,7% paraformaldehydein PBS+2% BSA+0,05% Triton X-100 for 5 minutes at RT,permeabilized in PBS+2% BSA+0,5% Triton X-100, five washesin PBS+2% BSA+0,05% Triton X-100. Secondary antibody wasused together with 1:1000 dilution of TRITC- phalloidin (anactin F binding reagent) and incubated for 30 min.

Primary antibodies used: anti-XCdc14A; anti-γ Tubulin(Sigma, clone GTU-88); anti-β Tubulin; anti-Ninein (gift from

JB Rattner); anti-GLU Tubulin (L4 antibodies; gift fromDr D. Joband L. Lafanechere); anti-polyglutamylated tubulin GT335 (giftfrom C. Janke); anti-human Aurora B (BD TransductionLaboratories); anti-human MKLP1 (H-110) (Santa Cruz); anti-Xenopus INCENP (gift from Thierry Lorca); anti-human Sec8(BD Transduction Laboratories); anti-Endobrevin (SYSY); anti-cyclin A and anti-cyclin E (gift from V. Dulic). All secondaryantibodies were AlexaFluor conjugates and used according tomanufacturer's instructions (Molecular Probes).

All images were taken on an upright microscope (LeicaDMRA) using Metamorph 6.2.6. (Molecular Devices) software.For Sec8 and endobrevin midbody localization (Fig. 7, insets),images present two-dimensional projections of three-dimen-sional deconvolved (Huygens, Science Volume Imaging)reconstructions to ensure that all stainedmaterial was shown.

For time-lapse imaging, cells were plated on 25 mm glass-bottom plates, transfected the next day with either gfp alone,gfp-XCdc14A, or gfp-centrin (kind gift from M. Bornens) and24 hours later placed in an observation chamber, in completemedium with CO2 (5%) exchange at 37 °C. Cells were imagedevery 20 minutes for up to 20 hours, using Plapo 40x objective,NA 1.25, with Nomarski contrast and with gfp filter on aninverted microscope (DMIRE2). Images were captured on aMicromax YHS 13000 (Princeton Instruments) camera andMetamorph 6.2.6. (Molecular Devices) software.

Adobe Photoshop (Adobe Systems) was used for adjust-ment of image contrast and brightness.

pNPP assay

HEK293T cells were transfected with gfp, gfp-XCdc14A WT orPD using JetPEI according to themanufacturer's recommanda-tions (Polyplus Transfection). Transfected cells were washedin PBS 20 hours post-transfection and lysed in Lysis Buffer(20 mM TrisHCl pH8.0; 150 mM NaCl; 10% Glycerol; 1%NP40;5 mM EDTA; 0,5 mM EGTA; 1 mM DTT; 1 mM PMSF and 1xComplete protease inhibitor w/o EDTA, Roche) for 30 minutesat 4 °C. After centrifugation at 13200 rpm at 4 °C for 20minutes,the supernatant was incubated with polyclonal rabbit anti-gfpantibodies (Torrey Pines Biolabs) prebound to Dynabeadsprotein A (Dynal) according to the manufacturer's recommen-dations. After 1 hour incubation at 4 °C, the beads werewashed once in Wash Buffer 1 (50 mM TrisHCl pH 8.0; 300 mMNaCl; 0,05% NP40; 0,5 mM EDTA), twice in Wash Buffer 2(50 mM TrisHCl pH 8.0; 500 mM NaCl; 0,05% NP40; 0,5 mMEDTA), once in Wash Buffer 1 again and once in PhosphataseAssay Buffer (50 mM TrisHCl pH8.0; 1 mM DTT; 1 mM EDTA).

Phosphatase assays were performed in 250 μl of Phospha-tase Assay Buffer containing 20 mM pNPP for 1 hour at 37 °C.Absorbance at 410 nm was measured at the indicated timepoints.

After the assay, the beads were resuspended in 20 μl of 1xLaemmli buffer and the proteins were separated by SDS-PAGE,transfered to a nitrocellulose membrane and analysed byWestern-Blot using monoclonal anti-gfp antibodies (Roche).

FACS analysis

For FACS analysis, unsynchronized untransfected, gfp aloneor gfp-XCdc14A (WT or PD) expressing cells (24 hours post-

1228 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

transtection) were fixed with either 70% ethanol (for gfp-XCdc14A) or 2%PFA followed by permeabilization with 0,2%Triton X-100 (for gfp alone), incubated with propidium iodide(1 μg/ml) and RNAse (200 μg/ml) for 30 minutes, and analysedwith the fluorescence-activated cell sorting; 100 000 cells werecounted.

Results

Cloning and characterization of Xenopus Cdc14homologues

In order to clone the Xenopus Cdc14A proteins, we performedBLAST searches against the Xenopus EST database using the S.cerevisiae Cdc14 sequence as a query. We identified two ESTcDNAs (GenBank accession nos. BG018336 and BJ08353) enco-ding Cdc14 related proteins. These cDNAs (ORF1 and ORF2)encode for open reading frames of respective predictedmolecular weights of 69 kDa and 55 kDa, respectively. ORF1 isidentical to previously reported XCdc14α [21] and shares 65,1%identitywith hCDC14A. The first 416 aa of ORF2 are identical toXCdc14β but the two proteins diverge in their C-terminus. AsXCdc14B and XCdc14β share the same 5′ and 3′ untranslatedregions, this suggest that they are encoded by a single genethrough an alternative splicing event. ORF2 is closely related tohuman CDC14B2 (56,6% identity) indicating that it is probablythe hCDC14B orthologue (Suppl. data, Fig. 1). Phylogenetic treeanalysis confirms that ORF1/XCdc14α belongs to the hCDC14Asubfamily while ORF2 belongs to hCDC14B subfamily (datanot shown). For clarity, we thus renamed ORF1/XCdc14α asXCdc14A (GenBank accession no. AAV66581), and namedORF2as XCdc14B (GenBank accession no. AAV66582).

Fig. 1 – gfp-XCdc14A localizes to the centriole and the midbody.fixed and analyzed by indirect immunofluorescence 24 hours laγ-tubulin (top, red;), and with ninein, a centriolar protein (bottom(B) gfp-XCdc14A localizes to the midbody in cytokinesis; it is pre

To study the Cdc14 proteins in the Xenopus system, wegenerated specific antibodies against XCdc14A and B. How-ever, none of them allowed us to clearly identify Cdc14proteins in Xenopus eggs, tissues or cell extracts, suggesting avery low abundance of the XCdc14 proteins in these systems(data not shown).

Thereafter, we concentrated our studies on the detailedinvestigation of the role of XCdc14A in the progression of cellcycle using ectopically expressed gfp-tagged XCdc14A.

Ectopically expressed XCdc14A localizes to thecentrioles and prevents de novo microtubule nucleation

To gain insight into the function of XCdc14 we first deter-mined its cellular localization. We therefore transientlyexpressed a gfp-tagged form of XCdc14A (gfp-XCdc14A) inXenopus XL2 (epithelial) cells. We also used the heterologoussystem of HeLa cells, to see whether localization is likely to beinfluenced by the cell-type or species specificity. The sub-cellular localization of gfp-XCdc14A (see below and data notshown) was similar in the two types of cells, suggestingfunctional conservation of localization mechanisms betweenXenopus and HeLa cells. We confirmed our results using theXTC Xenopus cell line, as well as another transformed humancell line, U2OS cells (data not shown).

In interphase HeLa and XL2 (data not shown) cells, ectopicgfp-XCdc14A appears diffuse in the cytoplasm and enriched atthe centrosome, as confirmed by costaining with the pericen-triolar marker, γ-tubulin (data not shown and Fig. 1A, toppanels). Costaining with ninein, a component of the mothercentriole [24], revealed that gfp-XCdc14A specifically localizesto the centriole. Only one of the gfp-associated distinct dotscostained with ninein, indicating that gfp-XCdc14A localizes

(A) HeLa cells were transiently transfected with gfp-XCdc14A,ter. XCdc14A colocalizes with the pericentriolar component,, red). Insets, enlargements of centrosomes. Bar 5 μm.

sent at the centrosomes (arrows). β-tubulin, red. Bar 10 μm.

1229E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

to both mother and daughter centrioles (Fig. 1A, bottom). Todetermine whether the phosphatase activity of XCdc14A wasrequired for its cellular localization we constructed a phos-phatase dead (PD) mutant by replacing the conserved cysteineof the catalytic domain with serine (C276S; [25,26]). Thelocalization of the PD mutant was indistinguishable fromWT XCdc14A (data not shown) indicating that the catalyticactivity is not required for its targeting to cellular structures.

In HeLa cells, gfp-XCdc14A overexpression never inducedcentrosomal disorganization or splitting (data not shown) incontrast to a report for hCDC14A [6] (Kaiser et al., 2002).However, upregulation of XCdc14A interfered with centroso-mal function, impeding nucleation of MTs depolymerized bynocodazole treatment (Suppl. data, Fig. 2, time 10 min., upperpanel), as reported for hCDC14 [7,18]. Nocodazole treatmentof gfp-XCdc14A expressing cells showed also that itscentrosomal localization is independent of MTs (Suppl.data, Fig. 2, time 0 min., bottom panel). Thus, some but notall of hCDC14 functions might be conserved with the Xenopusproteins.

Since the localization appears to be independent of the celltype in which XCdc14A is expressed, we preferred to use theHeLa cell system in whichmuch higher transfection efficiencycan be obtained than in Xenopus cells, allowing us to studyeven low-frequency phenotypes. Such heterologous systemshave been used previously to obtain important functionalinformation for Cdc14 [9]. We suspected that a Xenopus/HeLamixed vertebrate model might also yield useful informationabout conservation of function and regulation.We nextcharacterized in more detail the phenotype induced by thegfp-XCdc14 expression in HeLa cells.

Ectopic expression of gfp-XCdc14A interferes with the cellcycle

Interestingly, we never observed mitotic cells expressing gfp-XCdc14A, suggesting that upregulation of the phosphatasemight induce a G2/M block. In contrast, many gfp-XCdc14Apositive cells (22%, n=405) displayed late telophase/cytokin-esis phenotypes compared to gfp-positive control cells (5%,n=1033), indicating that gfp-XCdc14A expression indeedinterferes with cytokinesis as would be expected if XenopusCdc14 is functionally conserved with human CDC14. In thesecells, the ectopic protein concentrated to the midbody, andwas present at the centrioles/centrosomes (Fig. 1B, arrows).Similarly, the PD mutant localized to the midbody in cytokin-esis (data not shown).

To investigate the effect of XCdc14A on cell cycle progres-sion we analysed the distribution of cells overexpressingeither WT or PD gfp-XCdc14A in different phases of the cellcycle by flow cytometry (Fig. 2A). Unsynchronised cellsexpressing WT but not PD gfp-XCdc14A showed an aberrantDNA profile, characterized by accumulation of cells in G2/Mand loss of cells in S phase (49% and 8,7% for WT gfp-XCdc14Avs 26,5% and 14,7% for the mock transfected cells, respec-tively). Cells expressing gfp only showed a similar FACS profileto control cells (data not shown), excluding a possible effect ofgfp expression per se on the progression of cells throughthe cell cycle. We also used an in vitro phosphatase assayto confirm that the gfp-tagged XCdc14A exhibits catalytic

activity. As shown in Supp. data Fig. 3, the WT but not the PDXCdc14A immunoprecipitated from transfected cells effi-ciently catalyzed the dephosphorylation of pNPP. Thus, over-expression of XCdc14A may interfere with entry into and/orexit from mitosis and possibly the G1/S transition (see below),in accordance with our observations of fixed cells. Moreover,this effect is dependent on the catalytic activity of thephosphatase.

In order to thoroughly characterize the cell cycle defectinduced by upregulated levels of XCdc14A we followed cellsexpressing WT gfp-XCdc14A by time-lapse imaging. Again, nomitotic cells expressing gfp-XCdc14A could be observed, andnone of the examined interphase cells entered into mitosis. Incontrast, in a subset of dividing cells, in which there was nodetectable gfp-XCdc14A signal, the gfp signal became detect-able as the midbody formed, upon cytokinesis, probably dueto neo-translation of accumulated mRNAs, or from recruit-ment and concentration at the midbody of very low levels ofgfp-XCdc14A.

Over 85% of control cells completed cytokinesiswithin 2,5±0,8 hours from its beginning (formation of a well definedmidbody, up to the abscission and separation of the resultingdaughter cells), while 15% died during this process (Figs. 2Band C, upper panel, and movie 1). In contrast, gfp-XCdc14Aexpressing cells captured in cytokinesis were characterized by3 different phenotypes (Fig. 2B). The majority of gfp-XCdc14Atransfected cells (almost 60%) remained connected by acytoplasmic bridge with well defined midbody for up to18 hours (medium 11,4±4,4 hours) of imaging (Suppl. data,movie 2; Fig. 2C, middle panel). These cells, connected by thinand elongated bridges, were highly dynamic, moving apartand getting closer again as if in attempts to divide. In over 30%of the cells, division ended by apoptosis of one or bothdaughter cells that was characterized by intensive blebbing(Suppl. data, movie 3; Fig. 2C, lower panel). These cells wereprobably lost during fixation procedures, which would makethe number of gfp-XCdc14A expressing cells blocked incytokinesis even higher than we observed. Finally, completionof cytokinesis was occasionally apparent (<10%), probably dueto traction-mediated “artificial” cytokinesis [24].

To analyze whether the observed defects were dependenton the catalytic activity of the phosphatase, we observed cellsexpressing PD gfp-XCdc14A by time-lapse microscopy. Simi-larly to cells overexpressing catalytically active protein, PDgfp-XCdc14A expressing cells were unable to perform abscis-sion and remained connected by cytoplasmic bridges with awell defined midbody for up to 13,1±7,2 hours (n=16 cellsanalysed). Interestingly, however, in contrast to cells expres-singWT gfp-XCdc14A, these cells apparently progressed in thecell cycle (Fig. 2D). We observed daughter cells enteringanother round of mitosis, while still interconnected. In thesecases, the thin bridges were broken upon rounding up of thecells (Fig. 2D, time 11:00). Accordingly, we could observeinterphase cells expressing PD gfp-XCdc14A entering intomitosis.

These observations indicate that upregulated XCdc14Ablocks entry into mitosis and completion of cytokinesis. Inaddition, whereas arrest of the cell cycle depends on thephosphatase activity of XCdc14A, the cytokinesis defect doesnot.

1230 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

XCdc14A inhibits the G2/M transition by targeting Cdc25phosphatase

To analyze in detail the G2/M arrest induced by XCdc14A, wetook advantage of Xenopus egg extract. In this system,interphase extracts are induced to enter mitosis by theaddition of recombinant Δ90 cyclin B [27], which we followby 1) DNA condensation status using DAPI staining, 2)

nuclear envelope integrity, using a gfp-tagged NLS sequenceconstruct, whose diffusion throughout the extract indicatesthe nuclear envelope breakdown, and 3) reduction in XCdc25electrophoretic mobility, characteristic of its activatoryphosphorylations. We show that WT but not PD XCdc14Asignificantly delays entry into mitosis (Fig. 3). 75 minutesafter Δ90 cyclin B addition, more than 95% of nuclei brokedown in the control or after XCdc14A PD addition, while less

Fig. 3 – Wild-type but not phosphatase-dead XCdc14A inhibits entry into mitosis. (A) Entry into mitosis was monitoredmicroscopically as condensation of chromatin (DNA stainedwith DAPI) and diffusion of gfp signal.Whereas control (CycB only),PD or RL containing extract entered mitosis, as indicated by the condensation of chromatin and diffusion of gfp signal (+Cyc B,50′), extract containing WT XCdc14A was delayed in the entry into mitosis, with significant number of nuclei displayingcondensed chromatin but intact nuclear membrane (CycB+WT, 50′). (B) 75 minutes after induction of mitosis entry, nuclei inmitosis were counted; the graph represents counts from seven separate experiments. (C) Analysis of XCdc25 phosphorylationstatus. Compare the shift in electrophoretic mobility of Cdc25 due to phosphorylation at 50 minutes. (D) XCdc14A reversesCDK1-dependent hCDC25C phosphorylation. GST-hCDC25C was phosphorylated in vitro by CDK1/cyclin B and subsequentlyincubatedwith orwithout 6His-XCdc14A. Reactionswere separated by SDS-PAGE and analyzed by Coomassie staining (bottom)and autoradiography (top).

1231E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

than 50% of nuclei entered mitosis in the presence of theWT phosphatase (Fig. 3B). This could reflect a directinhibition of XCdc25 by XCdc14A, as was shown in yeasts[28,29]. To determine whether XCdc14A is a functionalphosphatase for Cdc25, we tested whether recombinantXCdc14A could dephosphorylate hCDC25C phosphorylatedby CDK1/Cyclin B. As shown in Fig. 3D, XCdc14A reduced the

Fig. 2 – Ectopic expression of gfp-XCdc14A interferes with the ceanalysis of gfp-XCdc14A expressing cells. Untransfected or gfp-X(24 hours posttransfection) were stainedwith propidium iodide, agrath shows distribution of cells in different cell cycle phases as(B) HeLa cells transciently transfected with either gfp or WT gfp-Xfor up to 20 hours. Cells undergoing cytokinesis were analysed,presented as percentage of total number of cells counted (n=13 f(C) Time-lapse images of cytokinesis defect phenotypes inducedindicates dissolved cytoplasmic bridge. Middle and bottom: examgfp-XCdc14A (stalled cytoplasmic bridge and apoptosis, respectivcells in which the gfp signal appeared in late telophase, when thsignal appears weak). Left, gfp image for phase contrast; middle10 μm for control, 20 μm for gfp-XCdc14A positive cells. Time-la(D) PD gfp-XCdc14A expressing cells enter mitosis but are unableanalysis; midbodies depicted with arrows; bar 20 μm.

phosphorylation of CDC25C and reversed its mobility shift.This result indicates that Cdc25 is dephosphorylated and itsactivity likely directly regulated by XCdc14A at the entry intomitosis.

We therefore show that XCdc14A can regulate entry intomitosis via dephosphorylation of Cdc25 in vertebrates, aspreviously reported in yeasts.

ll cycle and induces cytokinesis arrest. (A) Flow cytometryCdc14A (wild type or PD) expressing asynchronous cellsnd their DNA content was determined by flow cytometry. Themedium (±STDEVE) taken from 3 independent experiments.Cdc14A were followed 24 hours later by time-lapse imaging

phenotypes observed from 3 separate experiments areor the gfp, and n=27 for gfp-XCdc14A positive cells).by WT gfp-XCdc14A. Top, control gfp-expressing cells; arrowples of cytokinesis defect phenotypes induced by WT

ely). Middle: for better comparison with the control, we chosee midbody (arrow) was being formed; therefore the gfp, time 4 h 20; bottom, time 0 h. Time in hours:minutes. Barpse movies are available as Suppl. data, movies 1–3.to undergo abscission. Phase and gfp contrast of time-lapse

1232 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

Gfp-Xcdc14 expressing cells stalled in cytokinesis also arrest inG1

To investigate whether XCdc14A upregulation could alsoinhibit the G1/S transision, HeLa cells synchronized in G1/Sby a double thymidine block, were transfected with gfp-XCdc14A after release from the block, and analysed by indirectimmunofluorescence 24 hours later. As expected, mostcontrol gfp-positive cells had progressed through the nextcell cycle and only 10±3%were in cytokinesis. In contrast, overhalf of gfp-XCdc14A expressing cells were still in cytokinesis,connected by long intercellular bridges. No significant amountof binucleated cells (the cleavage furrow did not regress) normultinucleated cells could be observed, indicating that con-nected gfp-XCdc14A expressing daughter cells might arrest inthe cell cycle, whichwas also supported by our observations oflive cells (see above). To test this hypothesis, we analysed theexpression of cell cycle regulated cyclin E and A in cellsoverexpressing gfp-XCdc14A. Fig. 4A shows late G1 control cell(control, upper panel), where cyclin E accumulates in thenucleus, displaying a characteristic punctate pattern, withmore brightly stained dots, which probably correspond toCajal bodies [30,31]. The lower panel presents a cell in S phase,with cyclin E being present almost exclusively in Cajal bodies,with markedly lower levels in the nucleoplasm. Cyclin Apositive nuclear staining (Fig. 4B, control) is considered as amarker for S phase entry [32]. All cytokinesis-arrested cellsexpressing WT gfp-XCdc14A were either negative for cyclin E(Fig. 4A, WT gfp-XCdc14A upper panel) or displayed stainingcharacteristic for mid- to late G1 phase (Fig. 4A, WT gfp-XCdc14A lower panel). None of them presented any typical S

Fig. 4 – WT but not PD gfp-XCdc14A expressing cells stalled in cexpressing HeLa cells were fixed and analysed for the expressionstaining for cyclin E in late G1 (top), and S phase, with accumulacyclin A (B, left). WT gfp-XCdc14A expressing cells are either negcorresponding to late G1 (A, bottom); none of them is positive forexpressing cells (B, PD gfp-XCdc14A). DNA in blue. Bar 5 μm for

phase positive cyclin A signal (Fig. 4B, WT gfp-XCdc14A). Incontrast, a fraction of cells expressing PD gfp-XCdc14Ashowed positive cyclin A nuclear staining indicating thatthey entered S phase (Fig. 4B, PD gfp-XCdc14A). These resultsexplain the flow cytometry profile of distribution in the cellcycle and are consistent with our observations of live cells.The lack of WT gfp-XCdc14A-expressing cells in S phase isprobably a consequence of G1 arrest induced in cells unable tocomplete cytokinesis. In contrast, phosphatase inactiveXCdc14A does not interfere with G1/S. Alternatively, the G1/Sarrest might constitute the cell response to abscission failureand the expression of the PD mutant might allow bypass thisarrest.

These results strongly suggest that cells arrested incytokinesis expressing wild type but not phosphatase inactivegfp-XCdc14A do not enter the next S phase, a phenotype thatresembles the arrest observed for centriolin depletion [33] anddisruption of the centrosome function [34,35]. In all systemsstudied to-date, a role of Cdc14 in the regulation of cytokinesiswas suggested. We thus decided to investigate in detail thenature of cytokinesis defect caused by XCdc14A.

Actin and microtubule organization at the cleavage furrow arenot affected by ectopic XCdc14A expression

Defects in cytokinesis are most often associated with furrowregression, cell relaxation and binucleation. We first analyzedthe length of intercellular bridge connecting the gfp-XCdc14Apositive daughter cells and its microtubule composition. Thelength of the cytoplasmic bridge in gfp-XCdc14A expressingcells is statistically significantly longer than in control cells

ytokinesis are blocked in G1. Control and gfp-XCdc14Aof cyclin E (A) and A (B) (red). Left, control cells. Representativetion in Cajal bodies (A, bottom). Control cell positive forative for cyclin E (early G1; A, top) or they show stainingcyclin A (B, WT gfp-XCdc14A) in contrast to PD gfp-XCdc14Athe control and 20 μm for gfp-XCdc14A.

Table 1 – Length of cytoplasmic bridge in control andgfp-XCdc14A expressing cells [μm]

Control gfp-XCdc14A

Average 11,03 18,52STD VAR 2,74 10,88Min 7,19 3,01Max 14,98 48,9n 35 47

The length of the cytoplasmic bridge connecting cells in cytokinesiswas measured in control and gfp-XCdc14A expressing cells usingthe Metamorph software, by tracking both origin points of the MTbundles. STD VAR, standard variation; Min, minimal length of thebridge; Max, maximal length of the bridge; n, number of cells incytokinesis analysed.

1233E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

(Fig. 5A and Table 1; t-test p=0,0002). This corresponds to ourobservations during live imaging, where cells blocked incytokinesis were dynamically moving apart, as if in attemptsto separate. This phenotype is reminiscent of siRNA-mediateddepletion of Rab11/FIP3 and centriolin [36–38].

Microtubule bundles in the cytoplasmic bridge result fromthe compaction of the central spindle formed in anaphase. Toanalyse them, we stained cells with antibodies against β-tubulin (Fig. 5B) and detyrosinated-tubulin (GLU), a marker ofstabilized microtubules (Fig. 5C). MT organization in earlytelophase appears identical in control and gfp-XCdc14A-expressing cells. As telophase progresses, the intercellularbridge narrows, elongates and the stabilized MT bundlesprogressively disappear (control in Fig. 5B, middle and lowerpanels, and 4C, lower panel; [24,36]). In gfp-XCdc14A expres-sing cells, however, this latter event was accompanied by aextreme lengthening and thinning of MT bundles as well ascomplete loss of GLU staining (Figs. 5B, C, lower panels for gfp-XCdc14A). In summary, the ectopic expression of XCdc14A didnot affect the general organization of microtubules in the

Fig. 5 – Microtubule and actin organization of the cytokinesis brcytoplasmic bridge connecting cells in cytokinesis was measureMetamorph software, by tracking both origin points of theMTbunanalyzed forβ-tubulin (B, red) and detyrosinated tubulin (GLU) (C,through cytokinesis. DNA in blue. (D) Actomyosin ring formationcells. Control (left) or gfp-XCdc14A (right) expressing cells were s(red). In telophase, actin is present on the actomyosin ring (contrlower bottom). For gfp-XCdc14A expressing cells lower panel showfor panels B, C and D.

cytoplasmic bridge, although it induces its lengthning. How-ever, as no gfp-XCdc14A positive cells could be observed inanaphase, we cannot rule out the possibility that XCdc14Aupregulation might cause subtle differences in telophase.

idge is not affected by gfp-XCdc14A. (A) The length of thed in control and gfp-XCdc14A expressing cells using thedles. (B andC) Control and gfp-XCdc14A expressing cellswerered). Top and bottom, representation of temporal progressionand disassembly is not disturbed in gfp-XCdc14A expressingtained with TRITC-phalloidin to visualize the actin networkol, top) that constricts in late telophase/cytokinesis (control,s an extremely elongated bridge;β-tubulin in blue. Bar 5μm

1234 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

In early telophase, the equatorial actomyosin ring isassembled midway between the spindle poles. The activationof RhoA kinase pathway allows its contraction and ingressionof the cleavage furrow, as seen in the control cells (Fig. 5D,control, upper panel). As telophase progresses into cytokin-esis, the RhoA pathway is inactivated, the actomyosin ringdisassembled and only a faint staining of cortical actinremains visible at the cytoplasmic bridge (Fig. 5D, control,lower panel). Expression of gfp-XCdc14A impedes neither theactomyosin ring formation nor its disassembly, as seen withactin staining (Fig. 5D, gfp-XCdc14A). We conclude thatXCdc14A is unlikely to be involved in regulation of the RhoApathway.

Passenger proteins and centralspindlin complexlocalization is not disrupted by XCdc14A

Given the essential role of centralspindlin and passengerproteins complexes in the formation of central spindle and theregulatory role of Cdc14 from other systems in their transloca-tion in anaphase, we tested whether the localization of MKLP1and of the two chromosomal passenger proteins, Aurora B andINCENP, might be affected by ectopic expression of gfp-XCdc14A. As our data indicate that gfp-XCdc14A begins toaccumulate in telophase, we therefore wondered whether thepresence of the up-regulated phosphatase might modify thelocalization of these regulatory complexes in the midzone/midbody region. As shown in Fig. 6, in gfp-XCdc14A expressingcells (right), the localization of Aurora B, INCENP, andMKLP1 atthe midzone and midbody is indistinguishable from thatobserved in control cells (left), at both late telophase (top) and

Fig. 6 – Ectopic expression of gfp-XCdc14A does not change theregion in latemitosis. Control (left) andWT gfp-XCdc14A expressiMKLP1 (green in the control, red in gfp-XCdc14A positive cells). Tcytokinesis, with Aurora B localized to the central spindle flankiconcentrated to the midbody. Bar 2 μm.

cytokinesis (bottom), as characterized by relocalization ofAurora B from MT bundles flanking the midbody to themidbody itself. The same was true for the PD gfp-XCdc14A(data not shown).

In summary, we show that the expression of gfp-XCdc14Aphosphatase does not disrupt the localization to specificcellular structures of key regulators of late mitotic events,including the formation of the central spindle, the cleavagefurrow, and the midbody.

The mother centriole movement towards the midbody repeatsin gfp-XCdc14A expressing cells

The movement of the mother centriole towards the midbodywas proposed to be a prerequisite for abscission to occur [24].In the gfp-XCdc14A positive cells blocked in cytokinesisfollowed by time lapse imaging, we could observe the gfpsignal associated to the midbody as well as two dots in eachdaughter cell likely to be the two centrioles identified in ourstudies of fixed cells (Fig. 1A). In cells in which the gfp signal ofthe phosphatase appeared late in telophase, we frequentlyobserved a highly dynamic movement of the centriole fromeither one or both of the daughter cells, as reported by Pielet al. [24]. However, abscission did not occur after the centriolereturned to the cell body in XCdc14A expressing cells. Interes-tingly, in over six hours of time-lapse recording (movie 4 anddata not shown), the mother centriole movement towards themidbody could repeat several times, but would never success-fully induce physical separation of the daughter cells (Fig. 7and movie 4; note time points 0 hr 40, 3 hrs, and 5 hrs, wherecentriole approaching the midbody from opposite cells can be

localization of MKLP1 and passenger proteins at the midbodyng (right) HeLa cells were stained for Aurora B (red), INCENP orop for control MKLP1 and INCENP, an earlier stage ofng the midbody; bottom, a later stage, with Aurora B

Fig. 7 – Repeated migration of mother centriole to the midbody region takes place in gfp-XCdc14A expressing cells but theabscission does not occur. HeLa cells expressing gfp-XCdc14Awere followed by time-lapse imaging.White and red arrowheadsdepict mother centrioles moving towards the midbody from both cells; Mb, midbody. Time in hours:minutes. Bar 2 μm.

1235E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

observed). To exclude the possibility that the effect on thecentriole movement towards the midbody as well as comple-tion of cytokinesis is due to up-regulation of any centriolarprotein, we performed time-lapse imaging of HeLa cells stablyexpressing gfp-tagged centrin. Our results indicate that cellsoverexpressing centrin progressed normally through cytokin-esis, accomplishing the abscission. Moreover, we could notemovement of gfp-fluorescent centriole towards the midbodyregion preceding the separation of daughter cells (Supp. Fig. 4).Our data are in agreement with published observations, thatcells ectopically expressing gfp-centrin do not show anydefects in cell cycle progression [24,39,40].

Thus XCdc14A appears to block the events that arenormally induced after mother centriole migration, sinceabscission is not observed and the movement repeats severaltimes. Ultimately, the unsuccessful attempts of migratingcentrioles are abandoned, and the gfp-XCdc14A expressingcells arrest and evolve as shown in Fig. 2B.

Gfp-XCdc14 expression interferes with exocyst and SNAREcomplexes recruitment required for abscission

Recent studies show that centriolin, a mother centriolecomponent and a homologue of MEN scaffold Nud1 protein,re-localizes to the midbody during cytokinesis to recruitprotein complexes of vesicle targeting (exocyst) and vesiclefusion (SNARE) [33,36]. Centriolin silencing prevents theselast stages of cytokinesis and daughter cells remain con-nected by long intercellular bridges, as with ectopic expres-sion of XCdc14A. We therefore investigated whether gfp-XCdc14A phosphatase may control late cytokinesis eventsrequired for abscission. As such, we analyzed by immuno-fluorescence the presence of the SNARE and exocyst compo-nents, endobrevin and Sec8, respectively, at the midbody incontrol and gfp-XCdc14A positive cells (Fig. 8). In control cells(Fig. 8A, left), the exocyst complex component sec8 isdetected in the immediate vicinity of the midbody. Early in

cytokinesis, endobrevin and aurora-B kinase colocalized oneach side of the midbody in control daughter cells (Fig. 8B, topleft), and as cytokinesis progressed, their localization becamefocused at the midbody (Fig. 8B, lower left). In contrast, in gfp-XCdc14A expressing cells (Fig. 8A, medium panels), we wereunable to visualize Sec8 staining close to the midbody. Weoccasionally did see positive staining inside the cytoplasmicbridge, but none close to or at the midbody, even when thecytoplasmic bridge became very long. Similarly, no recruit-ment of endobrevin to the midbody could be observed (Fig.8B, medium panels). As illustrated in Figs. 8A, B, (right), andwhich corresponds to our observations by time-lapse, PD gfp-XCdc14A also abolished targeting of exocyst and SNAREcomplexes to the midbody in cytokinesis.

In summary, our results show that XCdc14A, indepen-dently of its catalytic activity, interferes with the recruitmentof SNARE/exocyst complexes to the midbody in the very lateevents of cell division, required for new membrane insertionand ultimate separation of daughter cells.

Discussion

In this study we have investigated the role of Xenopus Cdc14Ain the cell cycle. Our results point to a conserved function ofthe Xenopus homologue in the regulation of G2/M transition,which is executed at least through dephosphorylation ofCdc25. Moreover, we show a novel regulatory function forXCdc14A in abscission. Upregulated XCdc14A arrests cells inlate cytokinesis, connected by cytoplasmic bridges. Ectopicexpression of XCdc14 compromises neither central spindleformation nor establishment and constriction of the acto-myosin ring. Likewise, we did not observe abnormal localiza-tion of the centralspindlin and chromosomal passengercomplex proteins. In contrast, XCdc14A impedes the correcttargeting of both SNARE and exocyst complexes components,indispensable for abscission.

Fig. 8 – Ectopic expression of XCdc14A interferes with the recruitment of SNARE and exocyst complexes to the midbody.Control and gfp-XCdc14A expressing HeLa cells were fixed and analysed for the presence of exocyst (Sec8; A) and SNARE(Endobrevin; B) complex proteins (red). For the control (left), antibodies against Xenopus INCENP or human Aurora B were usedto localize themidbody region (green); upper and lower panels represent subsequent stages of cytokinesis, as shown by AuroraB recruitment to themidbody (B). In gfp-XCdc14A expressing cells (right), no staining at themidbody is visible for either Sec8 orEndobrevin, even if occasionally the signal of these proteins might be seen along the cytoplasmic bridge. Boxed regionsenlarged with insets to show the midbody region; endob, endobrevin; gfp-XC14, gfp-XCdc14A. Bar 10 μm.

1236 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

Whereas early cytokinesis steps, involving cleavage furrowformation and constriction, are relatively well studied andunderstood, little is known about the so-called “late” cytokin-esis events, the abscission, which brings about the physiolo-gical separation of daughter cells. Recent studies imply thatvesicle trafficking and new membrane delivery and insertionare essential [41,42], thus providing evidence that the “early”and “late” cytokinesis stages are temporally segregated andmechanistically distinct. A recent report proposes centriolinas a key orchestrator of these events [36]. Centriolin requiresMKLP1 for its recruitment to the midbody. It is then itselfessential for the recruitment of the exocyst vesicle targetingproteins to themidbody ring, allowing asymmetricmovementof SNARE vesicle fusion. Intriguingly, centriolin is a mothercentriole component, whose silencing strikingly resemblesthe cytokinesis defects induced by overexpression of XCdc14A[30]. Moreover, knockdown of either exocyst or SNAREcomplex proteins mimics the abscission phenotype resultingfrom centriolin depletion. The works of Wilson et al. [38] andFielding et al. [37] add some new elements to our under-

standing of abscission. They present a model of Rab-11-dependent recycling of endosomes and membrane traffic tothe midbody. Rab are small GTPases, which function in thevesicle tethering/docking to their target compartment, leadingto membrane fusion [43]. Rab11 regulates the recruitment ofFIP3 and FIP4 effector proteins to endosomes. Another GTPase,Arf6, is involved in the localization of these vesicles to themidbody, and the final docking event involves the exocystcomplex, complementing the model of Gromley et al. [36].Unsurprisingly, depletions of Rab11 or FIP3 and FIP4 result inabrogation of abscission, similarly to centriolin depletion (andXCdc14A overexpression), which suggests that both of thesepathways might complement each other or converge. Furtherstudies would be required, however, to understand whetherand how the Rab11- and centriolin-dependent events arelinked. Whether any of the presented sequences of eventsrequires regulatory/activating phosphorylation (with Aurora Band Polo kinases present in the midbody region) and/ordephosphorylation events (apart from Cdc14, also PP1 phos-phatase has been shown to localize to the midbody; [44]),

1237E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

remains to be determined. Poor understanding of themechan-isms of delivery, targeting and fusion of transport vesicles atthe site of abscission hinders at present the identification ofcandidate substrates/partners for Cdc14.

Interestingly, all the proteins, which, when deregulated,abrogate abscission, share one common feature: theylocalize to the centrosome/centriole in interphase and earlymitosis, and appear at the midbody in cytokinesis (centrio-lin, FIP3/FIP4, and XCdc14A). In this context, it is notsurprising that centrosome ablation induces a similarcytokinesis defect [34,35]. Thus, the centrosome might notonly anchor certain regulatory proteins, but it might beinvolved more directly in the regulation of cell cycle events[45]. Mother centriole movement has been proposed to be aprerequisite for the abscission to occur [27]. Our data show,however, that it is not sufficient, and suggest the existenceof downstream control events, the proper execution of whichis abolished by overexpression of XCdc14A. Interestingly, inthis respect, centriolin shares homology with the buddingyeast Nud1p protein that anchors the MEN pathway at thespindle pole body. We speculate that the observed asym-metric movement of the mother centriole to the midbody[24] could correlate with equally asymmetric accumulationof secretory vesicles at one side of the midbody, where theabscission subsequently occurs [36]. Is it the centriolemigration to the midbody that is responsible for the deliveryof regulatory proteins or does it just convey a signal, whichtriggers further events? Similarities between the phenotypesinduced by centriolin and XCdc14A misregulation, togetherwith the homology between centriolin and the MEN/SINcomponent Nud1p/Cdc11p, and the fact that Cdc14 is a well-established MEN pathway final substrate, raise the questionof whether a pathway homologous to the yeast MEN existsin higher organisms. Another MEN homologue, the kinaseLATS1, has been found to participate in the regulation ofearly stages cytokinesis [46]. However, the phenotypeinduced by interference with early cytokinesis might havehampered investigation of any other potential role insubsequent events. Intriguingly, MEN and SIN pathwaysalso show asymmetric activation at one of the spindle polebodies [47].

We demonstrate that cells arrested in cytokinesis due toXCdc14 upregulation arrest in G1 and do not progress to Sphase. This is consistent with the results from both centriolarprotein perturbation [33] and centrosome/centriole removalstudies [34,35]. The molecular mechanism that prevents Sphase is currently unknown. It has been suggested that it isthe centrosomal targeting of cyclin E, essential for triggeringthe S phase entry independently of CDK2 activity, that is lostduring centrosome structure or function removal [48]. It istherefore likely that centrosome function is compromised byectopically expressed XCdc14A. Whether or not Cdc14 inter-feres with cyclin E centrosomal localization, remains to bedetermined.

The fact that the phosphatase dead mutant of XCdc14Aalso induced cytokinesis defect, was surprising and couldimply that the phosphatase activity is not required in theregulation of cytokinesis. However, this result could also beinterpreted differently. Firstly, both WT and PD XCdc14A maybind and thus sequester their partners/substrates, changing

their proper cellular localization and/or preventing the inter-action with other partners. Secondly, Cdc14 might have amodulatory effect on cytokinesis and any interference withthe delicate balance of phosphorylation/dephosphorylationevents might lead to similar consequences. Accordingly,protein phosphatase PTP-BL interferes with early cytokinesisevents independently of its catalytic activity [49].

Our evidence for a potential role of XCdc14A in theregulation of the G2/M transition is reminiscent of the G2arrest imposed by overexpression of Flp1p and due toinactivation of CDK1 through destabilization of Cdc25p[4,5,28]. Similarly, hCDC14A and B can bind the fission yeastCdc25p, and hCDC14A can dephosphorylate its Cdc2-depen-dent residues [9]. However, the specific activity of hCDC14against hCDC25 has not yet been addressed. Our resultsindicate that XCdc14A negatively regulates the CDK1 activityat the mitosis entry concomitant with modulation of XCdc25activity/phosphorylation status. Cdc25 activation at G2/Mtransition is accompanied by increased phosphorylation atmultiple residues [50,51]. Some of these phosphorylations areCDK1/cyclinB-dependent [50,52], which makes Cdc25 anexcellent putative substrate for Cdc14. CyclinB1/CDK1 com-plex is first phosphorylated and activated on the centrosome[53]. XCdc14A localizes to this structure in interphase and atG2/M, which would bring the substrate and the phosphatasein spatial proximity. Accordingly, we show that XCdc14Adirectly dephosphorylates hCDC25C in vitro. Although we didnot assess whether this dephosphorylation modulates theactivity and/or stability and/or localization of Cdc25, wecannot exclude this possibility given the role of Cdc14 inthe regulation of Cdc25p in fission yeast [4,5,28] together withthe proposed similar function for hCDC14 [9], and theconservation of mechanisms regulating the activity of cyclinB/CDK1.

Collectively, we demonstrate that Xenopus Cdc14A isinvolved in the regulation of abscission, and its deregulationinterferes with membrane vesicle targeting/fusion to themidbody region. More work will be required to elucidatemechanisms of function and potential links between Cdc14and other candidate MEN homologues and the existing modelof the centriolin- and Rab-11 dependent events essential forabscission. Further studies would be also required to deter-mine the nature of interaction between the XCdc14A andXCdc25 and the role of CDK1 in the regulation of XCdc14activity and/or cellular localization.

Acknowledgments

We thank Susana Prieto, Julien Espeut, Amaury Gaussen,and Violeta Morin for their assistance and working atmo-sphere. We thank Vjeko Dulic for advice and discussionregarding the cell cycle. We acknowledge MRI platform ofCRBM, including Pierre Travo, Julien Cau, Sylvain De Rossi,and Volker Baecker for assistance with microscopy. Wethank the Japanese National Bio-Resource Project (Xenopus)for providing the EST clones. This work has been supportedby grants to N. M., A. A. and D. F. from the Associationpour la Recherche sur le Cancer (grants 3147, 5479 and4600).

1238 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.yexcr.2006.12.022.

R E F E R E N C E S

[1] S.L. Jaspersen, J.F. Charles, R.L. Tinker-Kulberg, D.O. Morgan,A late mitotic regulatory network controlling cyclindestruction in Saccharomyces cerevisiae, Mol. Biol Cell. 9(1998) 2803–2817.

[2] A.J. Bardin, A. Amon, Men and sin: what's the difference? Nat.Rev. Mol. Cell Biol. 2 (2001) 815–826.

[3] R. Visintin, K. Craig, E.S. Hwang, S. Prinz, M. Tyers,A. Amon, The phosphatase Cdc14 triggers mitotic exit byreversal of Cdk-dependent phosphorylation, Mol Cell. 2(1998) 709–718.

[4] N. Cueille, E. Salimova, V. Esteban, M. Blanco, S. Moreno, A.Bueno, V. Simanis, Flp1, a fission yeast orthologue of the s.cerevisiae CDC14 gene, is not required for cyclin degradationor rum1p stabilisation at the end of mitosis, J. Cell Sci. 114(2001) 2649–2664.

[5] S. Trautmann, B.A. Wolfe, P. Jorgensen, M. Tyers, K.L. Gould,D. McCollum, Fission yeast Clp1p phosphatase regulates G2/M transition and coordination of cytokinesis with cell cycleprogression, Curr. Biol. 11 (2001) 931–940.

[6] B.K. Kaiser, Z.A. Zimmerman, H. Charbonneau, P.K. Jackson,Disruption of centrosome structure, chromosomesegregation, and cytokinesis by misexpression of humanCdc14A phosphatase, Mol. Biol. Cell. 13 (2002) 2289–2300.

[7] N. Mailand, C. Lukas, B.K. Kaiser, P.K. Jackson, J. Bartek, J.Lukas, Deregulated human Cdc14A phosphatase disruptscentrosome separation and chromosome segregation, Nat.Cell. Biol. 4 (2002) 317–322.

[8] J. Bembenek, H. Yu, Regulation of the anaphase–promotingcomplex by the dual specificity phosphatase human Cdc14a,J. Biol. Chem. 276 (2001) 48237–48242.

[9] M.D. Vazquez-Novelle, V. Esteban, A. Bueno, M.P. Sacristan,Functional homology among human and fission yeast Cdc14phosphatases, J. Biol. Chem. 280 (2005) 29144–29150.

[10] U. Gruneberg, M. Glotzer, A. Gartner, E.A. Nigg, The CeCDC-14phosphatase is required for cytokinesis in the Caenorhabditiselegans embryo, J. Cell Biol. 158 (2002) 901–914.

[11] R.M. Saito, A. Perreault, B. Peach, J.S. Satterlee, S. van denHeuvel, The CDC-14 phosphatase controls developmentalcell-cycle arrest in C. elegans, Nat. Cell Biol. 6 (2004)777–783.

[12] E.T. Kipreos, Developmental quiescence: Cdc14 moonlightingin G1, Nat. Cell Biol. 6 (2004) 693–695.

[13] R. Visintin, E.S. Hwang, A. Amon, Cfi1 prevents prematureexit from mitosis by anchoring Cdc14 phosphatase in thenucleolus, Nature 398 (1999) 818–823.

[14] F. Stegmeier, R. Visintin, A. Amon, Separase, polo kinase, thekinetochore protein Slk19, and Spo12 function in a networkthat controls Cdc14 localization during early anaphase, Cell108 (2002) 207–220.

[15] R. Azzam, S.L. Chen, W. Shou, A.S. Mah, G. Alexandru, K.Nasmyth, R.S. Annan, S.A. Carr, R.J. Deshaies,Phosphorylation by cyclin B-Cdk underlies release of mitoticexit activator Cdc14 from the nucleolus, Science 305 (2004)516–519.

[16] S. Trautmann, D. McCollum, Distinct nuclear and cytoplasmicfunctions of the S. pombe Cdc14-like phosphatase Clp1p/Flp1p and a role for nuclear shuttling in its regulation, Curr.Biol. 15 (2005) 1384–1389.

[17] U. Gruneberg, R. Neef, R. Honda, E.A. Nigg, F.A. Barr,Relocation of Aurora B from centromeres to the centralspindle at the metaphase to anaphase transition requiresMKlp2, J. Cell Biol. 166 (2004) 167–172.

[18] H.P. Cho, Y. Liu, M. Gomez, J. Dunlap, M. Tyers, Y. Wang, Thedual-specificity phosphatase CDC14B bundles and stabilizesmicrotubules, Mol Cell Biol. 25 (2005) 4541–4551.

[19] M. Mishima, V. Pavicic, U. Gruneberg, E.A. Nigg, M. Glotzer,Cell cycle regulation of central spindle assembly, Nature 430(2004) 908–913.

[20] G. Pereira, E. Schiebel, Separase regulates INCENP-Aurora Banaphase spindle function through Cdc14, Science 302 (2003)2120–2124.

[21] P.B. Bell Jr., B. Safiejko-Mroczka, Improved methods forpreserving macromolecular structures and visualizing themby fluorescence and scanning electron microscopy, ScanningMicrosc. 9 (1995) 843–857 discussion 858–60.

[22] B.K. Kaiser, M.V. Nachury, B.E. Gardner, P.K. Jackson, XenopusCdc14 alpha/beta are localized to the nucleolus andcentrosome and are required for embryonic cell division, BMCCell Biol. 5 (2004) 27.

[23] A.W. Murray, Cell cycle extracts, Methods Cell Biol. 36 (1991)581–605.

[24] M. Piel, J. Nordberg, U. Euteneuer, M. Bornens,Centrosome-dependent exit of cytokinesis in animal cells,Science 291 (2001) 1550–1553.

[25] M. Streuli, N.X. Krueger, A.Y. Tsai, H. Saito, A family ofreceptor-linked protein tyrosine phosphatases in humansand Drosophila, Proc. Natl. Acad. Sci. U. S. A. 86 (1989)8698–8702.

[26] G.S. Taylor, Y. Liu, C. Baskerville, H. Charbonneau, Theactivity of Cdc14p, an oligomeric dual specificity proteinphosphatase from Saccharomyces cerevisiae, is required forcell cycle progression, J. Biol. Chem. 272 (1997) 24054–24063.

[27] A. Abrieu, T. Lorca, J.C. Labbe, N. Morin, S. Keyse, M. Doree,MAP kinase does not inactivate, but rather prevents the cyclindegradation pathway from being turned on in Xenopus eggextracts, J. Cell Sci. 109 (Pt 1) (1996) 239–246.

[28] B.A. Wolfe, K.L. Gould, Fission yeast Clp1p phosphataseaffects G2/M transition and mitotic exit through Cdc25pinactivation, Embo J. 23 (2004) 919–929.

[29] V. Esteban, M. Blanco, N. Cueille, V. Simanis, S. Moreno, A.Bueno, A role for the Cdc14-family phosphatase Flp1p at theend of the cell cycle in controlling the rapid degradation of themitotic inducer Cdc25p in fission yeast, J Cell Sci. 117 (2004)2461–2468.

[30] S.V. Ekholm, P. Zickert, S.I. Reed, A. Zetterberg, Accumulationof cyclin E is not a prerequisite for passage through therestriction point, Mol. Cell Biol. 21 (2001) 3256–3265.

[31] J. Liu, M.D. Hebert, Y. Ye, D.J. Templeton, H. Kung, A.G. Matera,Cell cycle-dependent localization of the CDK2-cyclin Ecomplex in Cajal (coiled) bodies, J. Cell Sci. 113 (Pt 9) (2000)1543–1552.

[32] F. Erlandsson, C. Linnman, S. Ekholm, E. Bengtsson, A.Zetterberg, A detailed analysis of cyclin A accumulation at theG(1)/S border in normal and transformed cells, Exp. Cell Res.259 (2000) 86–95.

[33] A. Gromley, A. Jurczyk, J. Sillibourne, E. Halilovic, M.Mogensen, I. Groisman, M. Blomberg, S. Doxsey, A novelhuman protein of the maternal centriole is required for thefinal stages of cytokinesis and entry into S phase, J. Cell Biol.161 (2003) 535–545.

[34] E.H. Hinchcliffe, F.J. Miller, M. Cham, A. Khodjakov, G. Sluder,Requirement of a centrosomal activity for cell cycleprogression through G1 into S phase, Science 291 (2001)1547–1550.

[35] A. Khodjakov, C.L. Rieder, Centrosomes enhance the fidelityof cytokinesis in vertebrates and are required for cell cycleprogression, J. Cell Biol. 153 (2001) 237–242.

1239E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 2 2 5 – 1 2 3 9

[36] A. Gromley, C. Yeaman, J. Rosa, S. Redick, C.T. Chen, S.Mirabelle, M. Guha, J. Sillibourne, S.J. Doxsey, Centriolinanchoring of exocyst and SNARE complexes at themidbody isrequired for secretory-vesicle-mediated abscission, Cell 123(2005) 75–87.

[37] A.B. Fielding, E. Schonteich, J. Matheson, G.Wilson, X. Yu, G.R.Hickson, S. Srivastava, S.A. Baldwin, R. Prekeris, G.W. Gould,Rab11-FIP3 and FIP4 interact with Arf6 and the exocyst tocontrol membrane traffic in cytokinesis, Embo J. 24 (2005)3389–3399.

[38] G.M. Wilson, A.B. Fielding, G.C. Simon, X. Yu, P.D. Andrews,R.S. Hames, A.M. Frey, A.A. Peden, G.W. Gould, R. Prekeris,The FIP3-Rab11 protein complex regulates recyclingendosome targeting to the cleavage furrow during latecytokinesis, Mol. Biol. Cell. 16 (2005) 849–860.

[39] M. Piel, P. Meyer, A. Khodjakov, C.L. Rieder, M. Bornens, Therespective contributions of the mother and daughtercentrioles to centrosome activity and behavior in vertebratecells, J. Cell Biol. 149 (2000) 317–330.

[40] M.M. Mogensen, A. Malik, M. Piel, V. Bouckson-Castaing, M.Bornens, Microtubule minus-end anchorage at centrosomaland non-centrosomal sites: the role of ninein, J. Cell Sci. 113(Pt 17) (2000) 3013–3023.

[41] D.N. Robinson, J.A. Spudich, Towards a molecularunderstanding of cytokinesis, Trends Cell Biol. 10 (2000)228–237.

[42] S.H. Low, X. Li, M. Miura, N. Kudo, B. Quinones, T. Weimbs,Syntaxin 2 and endobrevin are required for the terminal stepof cytokinesis inmammalian cells, Dev. Cell. 4 (2003) 753–759.

[43] M. Zerial, H. McBride, Rab proteins as membrane organizers,Nat. Rev. Mol. Cell Biol. 2 (2001) 107–117.

[44] L. Trinkle-Mulcahy, P.D. Andrews, S. Wickramasinghe,J. Sleeman, A. Prescott, Y.W. Lam, C. Lyon, J.R. Swedlow, A.I.

Lamond, Time-lapse imaging reveals dynamic relocalizationof PP1gamma throughout the mammalian cell cycle, Mol.Biol. Cell. 14 (2003) 107–117.

[45] S.J. Doxsey, Molecular links between centrosome andmidbody, Mol. Cell. 20 (2005) 170–172.

[46] X. Yang, K. Yu, Y. Hao, D.M. Li, R. Stewart, K.L. Insogna, T. Xu,LATS1 tumour suppressor affects cytokinesis by inhibitingLIMK1, Nat. Cell Biol. 6 (2004) 609–617.

[47] J.N. Molk, S.C. Schuyler, J.Y. Liu, J.G. Evans, E.D. Salmon, D.Pellman, K. Bloom, The differential roles of budding yeastTem1p, Cdc15p, and Bub2p protein dynamics in mitotic exit,Mol. Biol. Cell. 15 (2004) 1519–1532.

[48] Y. Matsumoto, J.L. Maller, A centrosomal localization signal incyclin E required for Cdk2-independent S phase entry, Science306 (2004) 885–888.

[49] L. Herrmann, T. Dittmar, K.S. Erdmann, The protein tyrosinephosphatase PTP-BL associates with the midbody and isinvolved in the regulation of cytokinesis, Mol. Biol. Cell. 14(2003) 230–240.

[50] T. Izumi, J.L. Maller, Elimination of cdc2 phosphorylation sitesin the cdc25 phosphatase blocks initiation of M-phase, Mol.Biol. Cell. 4 (1993) 1337–1350.

[51] T. Izumi, D.H. Walker, J.L. Maller, Periodic changes inphosphorylation of the Xenopus cdc25 phosphatase regulateits activity, Mol. Biol. Cell. 3 (1992) 927–939.

[52] I. Hoffmann, P.R. Clarke, M.J. Marcote, E. Karsenti, G.Draetta, Phosphorylation and activation of human cdc25-Cby cdc2-cyclin B and its involvement in theself-amplification of MPF at mitosis, Embo J. 12 (1993)53–63.

[53] M. Jackman, C. Lindon, E.A. Nigg, J. Pines, Active cyclinB1-Cdk1 first appears on centrosomes in prophase, Nat. CellBiol. 5 (2003) 143–148.