2424 Research Article

IntroductionCyclin-dependent kinases (CDKs) are involved in a variety ofimportant regulatory pathways in eukaryotic cells, includingcell-cycle control, apoptosis, neuronal physiology,differentiation and transcription (Harper and Adams, 2001;Trembley et al., 2004; Chen et al., 2006). The human (CDC2L1and CDC2L2) and mouse (cdc21) genes both produce twodifferent protein products. The 110-kDa protein isoform ofcyclin-dependent kinase 11 (CDK11p110), the major proteinkinase isoform, is expressed throughout the cell cycle and isinvolved in transcriptional regulation and RNA processing(Trembley et al., 2002; Trembley et al., 2003; Hu et al., 2003).By contrast, the 58-kDa protein isoform of cyclin-dependentkinase 11 (CDK11p58) is specifically expressed in the G2/Mphase of the cell cycle (Cornelis et al., 2000; Tinton et al.,2005; Wilker et al., 2007). This protein is generated bytranslation from an internal ribosome entry site located in thecoding region of the CDK11 full-length mRNA (Cornelis etal., 2000). Previous studies have shown that minimaloverexpression of CDK11p58 in Chinese hamster ovary (CHO)cells results in aneuploidy, increased numbers of cells thatmaintain postmitotic bridges or midbodies, and apoptosis(Lahti et al., 1995; Bunnell et al., 1990). A significant role forat least one CDK11 protein isoform in mitotic control in vivowas also identified through gene disruption. Deletion of thesingle CDK11 (cdc2l) gene in mice leads to embryonic deathat E3.5 (Li et al., 2004). CDK11-deficient blastocysts exhibitboth proliferative defects and mitotic arrest. Moreover, theinvolvement of CDK11p58 in centrosome maturation and

bipolar spindle morphogenesis was recently reported (Petrettiet al., 2006). Here we used a series of hypomorphic smallinterfering RNA (siRNA) expression constructs todownregulate CDK11 expression to varying extents. Thesestudies revealed additional roles for CDK11 in mitoticprogression and chromosome cohesion.

Sister chromatid cohesion is mediated by the cohesincomplex. Establishment of chromosome cohesion occursduring DNA synthesis in S phase, and the association of pairedsister chromatids must be maintained until anaphase to ensureaccurate chromosome segregation. Chromosome arm cohesinis removed during prophase and prometaphase through anonproteolytic process that requires Plk1 and aurora B kinaseactivity. Plk1 can phosphorylate the cohesin subunits Scc1 andSA2 in vitro, and the phosphorylation by Plk1 triggers thedissociation of cohesin from chromosome arms (Sumara et al.,2002; Losada et al., 2002; Hauf et al., 2005). By contrast,centromeric cohesin is cleaved by separase specifically at theonset of anaphase (Gimenez-Abian et al., 2004; Hauf et al.,2001). Centromeric chromosome cohesion is protected fromproteolysis before anaphase by Sgo1 (Kitajima et al., 2005;McGuinness et al., 2005). The centromeric localization ofSgo1 is regulated by Bub1 kinase and PP2A phosphatase(Kitajima et al., 2005; Tang et al., 2006). The cleavage ofcentromeric cohesin and initiation of sister chromatidseparation is controlled by the spindle checkpoint. The spindlecheckpoint monitors both the interactions betweenkinetochores and spindle microtubules and the tension acrosspaired kinetochores, and it delays cohesin cleavage and

Cyclin-dependent kinase 11 (CDK11) mRNA produces a110-kDa protein (CDK11p110) throughout the cell cycle anda 58-kDa protein (CDK11p58) that is specifically translatedfrom an internal ribosome entry site sequence duringG2/M. CDK11p110 is involved in transcription and RNAprocessing, and CDK11p58 is involved in centrosomematuration and spindle morphogenesis. Deletion of theCDK11 gene in mice leads to embryonic lethality at E3.5,and CDK11-deficient blastocysts exhibit both proliferativedefects and mitotic arrest. Here we used hypomorphicsmall interfering RNAs (siRNAs) to demonstrate that, inaddition to playing a role in spindle formation andstructure, CDK11p58 is also required for sister chromatidcohesion and the completion of mitosis. Moderate depletionof CDK11 causes misaligned and lagging chromosomes butdoes not prevent mitotic progression. Further diminutionof CDK11 caused defective chromosome congression,

premature sister chromatid separation, permanent mitoticarrest and cell death. These cells exhibited altered Sgo1localization and premature dissociation of cohesioncomplexes. This severe phenotype was not corrected bycodepletion of CDK11 and either Plk1 or Sgo1, but it wasrescued by CDK11p58. These findings are consistent withthe mitotic arrest we observed in CDK11-deficient mouseembryos and establish that CDK11p58 is required for themaintenance of chromosome cohesion and the completionof mitosis.

Supplementary material available online athttp://jcs.biologists.org/cgi/content/full/120/14/2424/DC1

Key words: CDK11p58, Cell cycle, Mitosis, Cyclin-dependent kinase,Sister chromatid cohesion, Cohesin, Mouse

Summary

CDK11p58 is required for the maintenance of sisterchromatid cohesionDongli Hu, Marcus Valentine, Vincent J. Kidd and Jill M. Lahti*Department of Genetics and Tumor Cell Biology, St Jude Children’s Research Hospital, 332 N. Lauderdale, Memphis, TN 38105, USA*Author for correspondence (e-mail: Jill.Lahti@stjude.org)

Accepted 15 May 2007Journal of Cell Science 120, 2424-2434 Published by The Company of Biologists 2007doi:10.1242/jcs.007963

Jour

nal o

f Cel

l Sci

ence

2425CDK11 protects sister chromosome cohesion

anaphase onset until all chromosomes are properly attached tothe spindle and aligned at the metaphase plate (Lew and Burke,2003; Musacchio and Hardwick, 2002). Spindle checkpointfunction involves protein phosphorylation and proteolysis(Cleveland et al., 2003). Indeed, several of the checkpointproteins including Msp1, Bub1 and BubR1 are protein kinases.In addition to the spindle checkpoint kinases, several otherknown kinases such as CDK1, aurora kinases and Plk1 playimportant regulatory roles in mitosis (Peters, 2002; Tang et al.,2004a; Acquaviva et al., 2004; van Vugt et al., 2004; Jackmanet al., 2003; Lampson et al., 2004; Sumara et al., 2004; Barr etal., 2004; van Vugt and Medema, 2005). These mitotic kinasesalso generate signals preventing the proteolysis of anaphaseinhibitors and mitotic cyclins, thereby blocking the onset ofanaphase (Peters, 2002). Here we show that the CDK11p58

kinase also plays a crucial role in mitotic progression and isrequired for the maintenance of sister chromatid cohesion andfor the completion of mitosis in human cells.

ResultsCDK11 depletion results in spindle and centrosomeabnormalitiesMultiple siRNAs targeting different regions of CDK11 wereused to specifically downregulate CDK11 gene expression(Fig. 1A and see supplementary material Fig. S1). Bothdouble-stranded oligonucleotides and pSUPER(Brummelkamp et al., 2002) expression constructs were usedto allow us to vary the extent of CDK11 depletion. In general,the pSUPER expression constructs were more effective thanthe siRNAs in mediating CDK11 depletion. Direct comparisonof siRNA expression construct pSUPER-CDK11-A1361 andoligonucleotide A1361, both of which target the samesequence used previously (Petretti et al., 2006), revealed thatthe expression construct silenced CDK11 much moreeffectively. Most of the studies described here were performedusing HeLa cells and either the C-terminal pSUPER-CDK11-

A2184 construct, which most efficiently downregulated bothCDK11p110 and CDK11p58, or the N-terminal pSUPER-CDK11-A461 construct, because both of these constructsresulted in the most complete elimination of CDK11. However,the results were verified using constructs targeting differentregions of CDK11 (A111, A188, A461, A910 and A1361) anddifferent cell lines (293T, U2OS and SAOS2) (supplementarymaterial Fig. S1 and data not shown). The specificity of theseconstructs was confirmed by analysis of several other proteins,including tubulin and actin. Experiments using SBI SystemBiosciences’s inteferon response detection kit (# SI300A-1),which monitors induction of five genes (OAS1, OAS2, MX1,ISGF3�, IFITM1) whose expression are induced by activationof the interferon response pathway, were also performed toensure that the phenotype was not because of induction ofinterferon-mediated responses to dsRNA (data not shown).

Depletion of CDK11 caused aberrations in spindlepositioning and structure (data not shown), similar to thoseobserved previously by Petretti et al. (Petretti et al., 2006).Abnormalities in centrosome numbers and positions were alsopresent. The extent of these abnormalities and the number ofcells exhibiting these aberrant phenotypes correlated with theextent of CDK11 depletion. Expression of green fluorescentprotein (GFP)-tagged RNAi-resistant CDK11p58 proteinrescued the centrosome number abnormality completely andcorrected the centrosome positioning defects in >80% of thecells (data not shown). These findings are in agreement withthose of previous studies (Petretti et al., 2006) and support arole for CDK11p58 in centrosome migration and spindleassembly or maintenance.

CDK11 depletion causes G2/M cell-cycle block andsubsequent cell deathIn addition to the spindle and centrosome abnormalities, wealso noticed an accumulation of rounded cells and an increasein cells with an apoptotic appearance after treatment with the

Fig. 1. CDK11 RNAi causes G2/M cell-cycleaccumulation, impaired proliferation and cell death.(A) Western blot analysis showing downregulation ofthe CDK11p110 and CDK11p58 proteins. HeLa cellswere transfected with CDK11 RNAi-N(pSUPER/A461), CDK11 RNAi-C (pSUPER/A2184)or the control plasmid (pSUPER). After 72 hours, celllysates were prepared and subjected to electrophoresisand immunoblot analysis. Immunoblots were probedwith anti-CDK11 P1C antibody (Trembley et al.,2002). (B) CDK11 RNAi impaired cell proliferation.Phase-contrast photographs of HeLa cells 72 hoursafter transfection with CDK11 or control RNAiplasmids. Bar, 100 �m. (C) DNA content analysis 72hours after transfection showing increased G2/M andsub-G1 fractions resulting from CDK11 RNAi.(D) Retarded cell growth caused by CDK11 RNAi.Equal numbers of cells were transfected with pSUPER(control) or pSUPER-CDK11–RNAi(pSUPER/A2184) plasmids and plated into dishes forculturing. Cells were counted 72 hours aftertransfection. The bars represent the mean fromtriplicate experiments with standard deviation.

Jour

nal o

f Cel

l Sci

ence

2426

siRNAs that most effectively targeted CDK11. The number ofcells exhibiting these morphologies correlated with the severityof CDK11 depletion. For example, cells treated with siRNAA1361 had a near-normal appearance, whereas treatment withexpression construct A2184 caused the appearance ofnumerous rounded and dying cells (Fig. 1B). DNA contentanalysis verified an increase in the cell population in G2/Mphase of the cell cycle and in the population of sub-G1 CDK11RNAi cells in the cultures with the severe phenotype (Fig. 1C).This sub-G1 population was stained by annexin V, indicatingthat these cells were apoptotic (data not shown). The mitoticarrest and subsequent cell death resulting from CDK11depletion were also shown by time-lapse microscopy of HeLaand HeLa H2B-GFP cells 48 hours after transfection (seesupplementary material Fig. S2 and Movies 1 and 2).

G2/M accumulation is because of mitotic arrest of cellscontaining prematurely separated sister chromatidsTo clarify the composition of the enhanced G2/M populationresulting from CDK11 depletion (Fig. 1), chromosomeanalysis was performed. No obvious differences inchromosome condensation were visible upon staining withDAPI or antibodies that detect phosphorylation of serine 10 ofhistone H3, a marker that correlates with chromosomecondensation (Wei et al., 1999) (data not shown). Interestingly,however, depletion of CDK11 with all of the constructsresulted in abnormalities in chromosome segregation. Theextent of these abnormalities correlated with the level ofCDK11 depletion. Moderate CDK11 depletion with the A1361oligonucleotide and expression constructs did not preventmitosis, although the cells exhibited an increase in the presenceof misaligned and lagging chromosomes (Fig. 2A,B). Bycontrast, more complete depletion dramatically increased thepopulation of cells with condensed chromosomes and slightlyseparated sister chromatids whereas mitotic cells with a typicalanaphase or telophase appearance were absent (Fig. 2C,D). Inaddition, very few cells were observed with all chromosomesaligned on the metaphase plate. Rather, most cells exhibitedincomplete chromosome congression. Expression of thepSUPER-A2184 (CDK11 RNAi-C) resulted in the strongestphenotype, which correlated directly with the level of CDK11depletion (Fig. 1A). Of the total mitotic cells treated with theA2184 construct, 62% exhibited premature sister chromatidseparation after CDK11 depletion, although most of theseparated sister chromatids remained paired (Fig. 2C,D). Thisobservation was confirmed by measuring the inter-kinetochoredistances of 50-100 paired sister chromatids stained withCREST antibody at different points during G2/M. Theinterkinetochore distance between the sister chromatids of thecontrol and CDK11 RNAi cells was identical in G2 (control0.46±0.06 �m and CDK11 RNAi cells 0.43±0.08 �m) and inprophase (control 0.61±0.1 �m and CDK11 RNAi cells0.61±0.04 �m). However, the distance between thekinetochores was significantly larger at later stages in mitosis.The distance between the sister chromatids of thepostprophase-arrested CDK11 RNAi cells was 1.05±0.24 �m,whereas the average distance of the control cells at acomparable stage in mitosis was only 0.76±0.15 �m. In fact,the distance between the sister chromatids in CDK11 RNAicells was larger than that of metaphase control cells (1.05±0.24�m versus 0.89±0.14 �m). Importantly, overexpression of the

Journal of Cell Science 120 (14)

RNAi-resistant form of CDK11p58-GFP, but not the RNAi-sensitive form, reduced the number of cells with prematurelyseparated sister chromatids by ~75% (Fig. 3).

Because the appearance of separated sister chromatids couldresult from defects in the establishment of sister chromatidcohesion during DNA replication, we performed severaladditional experiments to examine DNA replication andchromosome cohesion, including BrdU incorporation,measurements of G2-phase interchromatid distances and time-lapse photography of double-thymidine synchronized controland CDK11 RNAi cells. These studies indicated that theabsence of CDK11 did not affect DNA replication, the distancebetween sister chromatid in G2 cells or cell cycle progressionthrough S and G2 phase (data not shown). Rather, the CDK11-depleted cells differed from control cells only after enteringG2/M phase (see supplementary material Fig. S2 and Movies1 and 2). These data suggest that the premature separation ofsister chromatids is not because of gross defects in theestablishment or maintenance of cohesion during S/G2, but ismost likely a result of the premature loss of the cohesincomplexes holding the two sister chromatids together duringmitosis.

Depletion of CDK11 results in prometaphase-metaphase cell cycle arrestTo probe the nature of the mitotic arrest, we analyzed theresponse of CDK11-depleted cells to the microtubule poisonsnocodazole and taxol. These agents trigger the spindlecheckpoint, resulting in a prometaphase-metaphase mitoticarrest. As expected, the percentage of prometaphase-metaphase cells increased in control cultures during a six-hourexposure to these agents. By contrast, the number ofprometaphase-metaphase CDK11 RNAi cells did not increasein response to either treatment, although there was a slightincrease in the population of cells with prematurely separatedsister chromatids in response to treatment with themicrotubule-stabilizing drug taxol (Fig. 2D), suggesting thatthe spindle checkpoint can delay sister chromatid separation inthe presence of tension or that sister chromatid tension per seplays a role in the retention of cohesion complexes.

As an additional means to determine the nature of themitotic arrest in the CDK11-depleted cells, we performedimmunofluorescence staining with antibodies to cyclins A andB1. Both of these cyclins are required for mitotic progression,although cyclin A is degraded during the prometaphase-metaphase transition, whereas cyclin B1 is degraded during themetaphase-anaphase transition (Lew and Kornbluth, 1996;Peters, 2002). Immunofluorescence staining of cyclins A andB1 in CDK11 RNAi cells revealed that most of the mitotic cellslacked detectable cyclin A but retained cyclin B1 (Fig. 4A).These findings support the conclusion that CDK11 depletionresults in premature sister chromatid separation and mitoticarrest at the prometaphase-metaphase transition.

Mitotic progression is monitored by the spindle checkpointto ensure that sister chromatid separation does not occur untilall of the chromosomes have achieved bipolar spindleattachment and have aligned at the metaphase plate (Meraldiet al., 2004; Nicklas et al., 1995). BubR1, a key component ofthe mitotic spindle checkpoint machinery, normally remains onthe kinetochores until the onset of anaphase when sisterchromatids start to segregate (Chen, 2002; Lampson and

Jour

nal o

f Cel

l Sci

ence

2427CDK11 protects sister chromosome cohesion

Kapor, 2005; Tang et al., 2001). Immunofluorescence stainingwith BubR1 revealed the presence of high levels of BubR1 onthe kinetochore in all mitotic CDK11 RNAi cells, despite the

presence of separated sister chromatids (Fig. 4B). These datasuggest the spindle checkpoint is activated in CDK11 RNAicells and that activation of the spindle checkpoint is at least

Fig. 2. CDK11 depletion causespremature separation of condensedsister chromatids. (A,B) Partialdepletion of CDK11 causeschromosome misalignment andlagging chromosomes but does notblock mitotic progression.(A) Immunofluorescence analysis ofHeLa cells transfected with either aCDK11 siRNA A1361 oligonucleotideor control siRNA oligonucleotide andcultured for 48 hours beforeimmunofluorescence staining withanti-�-tubulin (red) and anti-CREST(green) antibodies and DAPI (blue).Bar, 5 �m. (B) Determination of thefrequency of chromosome alignmentand segregation. More than 100mitotic cells from each treatmentgroup (si-A1361, n=138; control,n=198) were analyzed for thepresence of misaligned and laggingchromosomes. In this representativeexperiment, 29% of the CDK11A1361 siRNA cells had misalignedand/or lagging chromosomes whereasonly 1.5% of the control cellsexhibited this phenotype. (C) Giemsa-

stained mitotic chromosome spreads for cells treated witheither control siRNA expression vector or CDK11 siRNAexpression construct 2184 analyzed 48 hours aftertransfection. Bar, 5 �m. (D) CDK11 depletion results inpremature sister chromatid separation. Data from arepresentative experiment in which cells were transfected witheither the control or CDK11 siRNA expression vector and 100mitotic cells were scored for chromosome appearance.Chromosome appearance was categorized as I: partiallycondensed with closed arms; II: condensed and separated armswith kinetochore linked; III: both condensed arms andkinetochores separated, still paired; IV: condensed andseparated sister chromatids no longer paired. (Noco:nocodazole.) The mitotic index was determined from counting~1000 cells from each Giemsa-stained RNAi sample.

Jour

nal o

f Cel

l Sci

ence

2428

partially responsible for the mitotic arrest observed in thesecells. However, in CDK11 RNAi cells, the spindle checkpointis not able to prevent the premature separation of the twochromatids.

CDK11 is required for maintenance of chromosomecohesion during mitosisThe premature separation of the sister chromatids suggestedthat diminished CDK11 results in defects in maintenance ofcohesion and/or premature removal of cohesin complexes.

To investigate the effects of CDK11 on sister chromatidcohesion in more detail, we transfected HeLa cells expressingthe Myc-tagged human Scc1 cohesin component with CDK11RNAi constructs and monitored Scc1 by immunofluorescence.Our data indicated that the cohesin complexes were present onthe chromosomes and kinetochores of CDK11-depleted cells

Journal of Cell Science 120 (14)

at the early stage of mitosis; however, Scc1 was prematurelyremoved from post-prophase mitotic cells (Fig. 5). In fact, 64%of the CDK11-depleted mitotic cells were Myc-Scc1 negative,whereas only 25% of the control RNAi cells were Myc-Scc1negative (Fig. 5) (McGuinness et al., 2005). The percentage ofCDK11 RNAi cells exhibiting a premature loss of cohesion didnot change in response to nocodazole treatment, suggestingthat CDK11 plays a role in protecting and/or maintainingcohesion complexes prior to the metaphase.

Recent data indicate that centromeric cohesion is guarded bythe cohesin-protecting protein Sgo1. Sgo1, in association withBub1, protects cohesin complexes from separase cleavagebefore the onset of anaphase (Tang et al., 2004b; McGuinnesset al., 2005; Watanabe and Kitajima, 2005; Kitajima et al.,2005). Therefore, we performed experiments to determinewhether Sgo1 and Bub1 were present at the centromeres inCDK11 RNAi cells. Immunofluorescence examination of Sgo1localization in CDK11-depleted cells showed that Sgo1proteins were present at the centromeres, although the positionof the protein was altered relative to CREST (Fig. 6A). In thecontrol cells, the Sgo1 signal was flanked by CREST signals,whereas the Sgo1 signal appeared to be colocalized with theCREST signal on kinetochores in CDK11 RNAi cells. Inaddition, Bub1 was prematurely removed from kinetochores inpost-prophase CDK11 siRNA-treated cells, although theprophase Bub1 staining pattern of these cells was similar tothat of the controls (Fig. 6B).

Because a loss of centromeric cohesion was also observedin Sgo1 RNAi-treated cells and Sgo1 localization was aberrantin CDK11 RNAi cells, we performed dual depletionexperiments using constructs targeting both CDK11 and Sgo1to determine whether the loss of both proteins would result insimilar centromeric cohesion defects. Simultaneous depletionof both Sgo1 and CDK11 resulted in separated sisterchromatids, which were similar in appearance and frequencyto those produced by the depletion of Sgo1 alone (Fig. 7). Wealso treated control and CDK11 RNAi cells with RNAiexpression constructs targeting the polo-like kinase Plk1. Plk1regulates multiple steps in mitosis, including spindle formationand chromosome condensation. Importantly for theseexperiments, this kinase is also required for the removal ofchromosome arm cohesin complexes (McGuinness et al., 2005;van Vugt and Medema, 2005; Kitajima et al., 2005; Salic etal., 2004). As reported previously, 88% of the mitoticchromosomes in Plk1 RNAi-treated cells exhibited closed arms(Fig. 7). Interestingly, codepletion of Plk1 and CDK11 causeda loss of both arm and centromeric cohesion in 32% of themitotic cells. This phenotype is similar to, although less severethan, that produced by codepletion of Sgo1 and Plk1, whichcaused a loss of chromosome cohesion in 71% of the cells.Recently, data has been published demonstrating thatcentromeric Plk1 staining is reduced in CDK11 RNAi cells(Petretti et al., 2006). Therefore, it is somewhat surprising thatCDK11-Plk1 codepletion has milder effects on prematuresister chromatid separation than Sgo1-Plk1 depletion.However, it is possible that the reduction in Plk1 kinetochorebinding is not sufficient to compromise cohesin complexremoval or that other kinases, such as aurora B, cancompensate for reduced chromosomal Plk1 in CDK11 RNAicells. Nonetheless, these findings suggest that CDK11, likeSgo1, plays roles in protecting or maintaining chromosome

Fig. 3. CDK11p58 protein partially rescues the premature sisterchromatid separation phenotype caused by CDK11 depletion. HeLacells stably expressing GFP-tagged CDK11p58 protein from theCDK11p58-GFP cell line or the RNAi-resistant form of theCDK11p58m-GFP cell line were transfected with CDK11 RNAi orcontrol RNAi plasmids and analyzed 48 hours later byimmunoblotting for expression of CDK11 proteins with P1Cantibody (A) and for mitotic chromosome appearance and mitoticindex (B). Giemsa-stained cells were analyzed for mitoticchromosome appearance (100 cells) and mitotic index (>1000 cells).Representative data from one experiment are shown. Chromosomeappearance was categorized as in Fig. 2.

Jour

nal o

f Cel

l Sci

ence

2429CDK11 protects sister chromosome cohesion

cohesion both along the arms and at the kinetochore regionsduring mitosis.

DiscussionThe use of hypomorphic CDK11 RNAi constructs in thesestudies has revealed a gradient of mitotic abnormalities.Moderate depletion of CDK11 led to spindle and centrosomedefects and misaligned and lagging chromosomes, but did notsignificantly impede mitotic progression. This abnormalprogression through mitosis, which has many of the hallmarksof a mitotic catastrophe (Castedo et al., 2004; Blagosklonny,2007), frequently results in aneuploidy and/or multinucleatedcells. By contrast, severe depletion of CDK11 resulted in moredramatic mitotic abnormalities. Under these conditions, mitoticchromosomes underwent normal chromatin condensation andappeared to progress normally through prophase andprometaphase; however, the condensed chromosomes failed tocongress properly at the metaphase plate, and the cellsunderwent a permanent mitotic arrest followed by thepremature loss of sister chromatid cohesion and cell death.Based on the observations that dying cells were stained byannexin V, we believe that the loss of CDK11 leads to mitoticfailure followed by apoptosis. However, because we have notexamined other markers of apoptosis, such as activation ofeffector caspases, cytochrome C release or DNA ladderformation, it remains possible that these severely depletedCDK11 cells also die from mitotic catastrophe.

The presence of prematurely separated sister chromatidsafter downregulation of CDK11p58 suggests that the kinaseplays a role in maintenance or protection of chromosomecohesion. To test this possibility, we examined the effects ofsimultaneous depletion of CDK11 and two proteins withknown roles in chromosome cohesion, Plk1 and Sgo1.Codepletion of CDK11 and Sgo1 increased the extent of thesister chromatid separation beyond that caused by the depletionof CDK11 alone. Depletion of Plk1 alone resulted in mitoticcells with hypercondensed chromosomes with closed arms.When the loss of Plk1 was combined with the loss of CDK11,the chromosomes remained hypercondensed, but thechromatids were separated. These findings support a role forCDK11 in the maintenance of chromosome cohesion but notin chromosome condensation. A similar phenotype has alsobeen observed upon codepletion of Sgo1 and Plk1(McGuinness et al., 2005). The similarities in the codepletionphenotype suggest that CDK11, like Sgo1, is required formaintenance of both centromere and arm cohesion and thatCDK11 and Sgo1 function in the same or redundant pathways.This theory is supported by the similar appearance ofchromosomes from Sgo1 and CDK11 RNAi cells and by thealterations in the localization of Sgo1 in CDK11-depleted cells.This hypothesis is also supported by other similarities in thephenotypes of CDK11 and Sgo1 RNAi cells. RNAi-mediateddepletion of Sgo1 causes premature centromeric sisterchromatid separation and a permanent mitotic arrest

Fig. 4. CDK11 depletion results in mitoticarrest. (A) Most CDK11-depleted mitotic cellsretain cyclin B despite sister chromatidseparation. CDK11 RNAi or control RNAi cellswere stained for cyclin A (green), cyclin B (red)and DNA (DAPI) 48 hours after transfection. Aportion of these cells were treated withnocodazole for six hours before harvesting.Cyclin A and cyclin B content was analyzed byimmunofluorescence in 130-250 mitotic cellsfrom each group in a representative experiment.(B) CDK11 depletion results in mitotic arrestwith BubR1 accumulation to the kinetochores.HeLa cells were analyzed by indirectimmunofluorescence 48 hours after transfectionwith CDK11 RNAi or control plasmids usingantibodies to BubR1 (red) and CREST (green).DNA is stained with DAPI (blue). Bars, 5 �m.

Jour

nal o

f Cel

l Sci

ence

2430

comparable to that observed in CDK11 RNAi cells (Tang etal., 2004b; Salic et al., 2004; McGuinness et al., 2005; Kitajimaet al., 2005). In addition, both CDK11- and Sgo1-depleted cellsretain cyclin B expression (McGuinness et al., 2005). Sisterchromatid separation in Sgo1 RNAi cells does not appear to bemediated by separase but rather occurs through anothermechanism, possibly involving phosphorylation of cohesinsubunits (Tang et al., 2004b; McGuinness et al., 2005). Ouranalysis of CDK11-depleted cells suggests that removal ofcohesin complexes from CDK11 and Sgo1 RNAi cells mayoccur through similar mechanisms, because CDK11 RNAicells have high levels of cyclin B and a premature loss of Scc1without a significant increase in Scc1 cleavage (data notshown). Thus, it is possible that CDK11 may functioncoordinately with Sgo1. Alternatively, CDK11 may be requiredfor proper localization, binding or function of Sgo1 and Bub1or PP2A; because these two other proteins are required forproper Sgo1 localization and function and for centromeric

Journal of Cell Science 120 (14)

cohesion (Tang et al., 2004b; Tang et al., 2006). Theobservation that centromeric Bub1 is not properly retained atthe centromere in post-prophase CDK11 RNAi cells isconsistent with this possibility. The localization of PP2A hasbeen reported to be Bub1-dependent and proper localization ofSgo1 is also dependent on PP2A (Tang et al., 2006), thus thediminished binding of Bub1 in the CDK11 cells might explainthe abnormal localization of Sgo1. A model depicting the roleof CDK11 in chromosome cohesion is shown in Fig. 8. In thismodel, the effects of CDK11 on chromosome cohesion couldresult from direct interactions between CDK11 and Sgo1,Bub1 or PP2A. Alternatively, we propose that theseinteractions are transient and/or CDK11 phosphorylates one ormore of these proteins prior to their localization at thekinetochore. We currently favor the second propositionbecause we have not been able to detect either interactionbetween CDK11 and the Sgo1 complex or kinetochore-associated CDK11. Although this may be because of technicalproblems with antibody recognition or CDK11 accessibility,these data indicate that further studies are needed to addressthe exact role of CDK11 in the maintenance of sister chromatidcohesion.

Although the spindle checkpoint in CDK11-depleted cellsappears to be intact as these cells arrest in mitosis, it is not clearwhether the loss of sister chromatid cohesion occurs before orafter mitotic arrest. Cohesion defects because of prematuredissociation of cohesion complexes caused by the loss ofCDK11 could lead to checkpoint activation and cell cyclearrest. This model is consistent with recent data from studieswith Xenopus demonstrating that antibody-mediated depletionof cohesin causes defects in chromosome congression, spindleabnormalities, changes in kinetochore structure, and aberrantkinetochore attachment and chromosome movement that

trigger the spindle checkpoint (Kenneyand Heald, 2006). Alternatively,dissociation of sister chromatidcohesion could occur after mitoticprogression was halted by the spindlecheckpoint. In this scenario, inductionof the spindle checkpoint could resultfrom defects in spindle-kinetochoreattachment, faulty spindle structure,and/or improper function of the spindlecheckpoint because of aberrantphosphorylation resulting from theloss of CDK11. The absence ofCDK11 would compromise the abilityof Sgo1 to protect the arrested cellsfrom a loss of centromeric cohesionand premature sister chromatidseparation. Importantly, the inability tooverride the spindle checkpoint andproceed through mitosis in eithermodel could result from a combinationof spindle checkpoint activity,aberrations in the structure of themitotic spindle, defects in kinetochoreattachment, and/or a requirement forCDK11 kinase activity for thecompletion of anaphase.

The data from this study

Fig. 5. CDK11 depletion leads to premature removal of centromeric cohesin complexes.HeLa cells expressing an inducible Myc-tagged human Scc1 were transfected with CDK11RNAi or control RNAi plasmids for 48 hours with or without 6 hours of nocodazoletreatment before harvesting. The cells were analyzed by immunofluorescence staining usingantibodies against the Myc tag to detect Myc-human Scc1 (red) and the kinetochores CRESTantigen (green). (A) Result of the analysis of 200-300 mitotic cells examined for the presenceof a Myc-Scc1 signal on centromeric kinetochores in a representative experiment.(B) Representative mitotic cells. Bar, 5 �m.

Jour

nal o

f Cel

l Sci

ence

2431CDK11 protects sister chromosome cohesion

demonstrates that CDK11p58

is required for mitoticprogression. The severedepletion data imply thatcomplete loss of CDK11 invivo would prevent mitoticprogression and would belethal to cells and to survival ofan organism. The observationthat CDK11-null embryos dieimmediately after depletion ofmaternal CDK11 proteinstores (Li et al., 2004) isconsistent with this prediction.Importantly, CDK11-nullembryos have an increasedpercentage of mitotic andapoptotic cells, suggesting thatloss of CDK11 results in amitotic arrest followed byapoptotic death similar to thatobserved in severely depletedCDK11 RNAi cells. Bycontrast, HeLa cells treatedwith hypomorphic siRNAconstructs, which caused onlypartial loss of CDK11 protein,completed mitosis despite thepresence of spindle defects andlagging chromosomes. Thesedata suggest that loss of asingle CDK11 allele because ofdeletion of chromosome 1p36,which is frequently observed intumors such as neuroblastoma,melanoma, breast cancer andseveral types of leukemia andlymphoma (Brodeur et al.,1977; Poetsch et al., 2003;Mori et al., 2003; Melendez etal., 2003; Bieche et al., 1993),could increase the number ofaneuploid cells, therebycontributing to tumorigenesis.This conclusion is supportedby recent studies examiningthe responses of CDK11haploinsufficient mice tocarcinogens known to causeskin cancer (Chandramouli etal., 2007). Mice with a singlefunctional CDK11 alleleexhibited an increase in tumorsize and number in thisskin cancer model system(Chandramouli et al., 2007).Importantly, however, the rateof tumor formation was notincreased in this model systemnor have we observed anyincrease in the frequency of

Fig. 6. CDK11 depletion leads to altered Sgo1 localization on the kinetochores. (A) HeLa cells wereanalyzed by indirect immunofluorescence 48 hours after transfection with CDK11 RNAi or controlplasmids using antibodies to Sgo1 (red) and CREST (green). DNA was stained with DAPI (blue).(B) Bub1 dissociates prematurely from the kinetochores upon depletion of CDK11. HeLa cells treatedwith CDK11 RNAi and control RNAi were stained for Bub1 (red) and DAPI (blue) 48 hours aftertransfection. Bars, 5 �m.

Jour

nal o

f Cel

l Sci

ence

2432

spontaneous tumors in our CDK11 heterozygous mice,suggesting that other cooperating genetic events are requiredfor tumor formation and progression.

The plethora of mitotic effects observed in CDK11 RNAicells probably results from alteration of phosphorylation ofCDK11 substrates. Recently, WAPL (Kueng et al., 2006;Gandhi et al., 2006), Haspin (Dai et al., 2006) and PICH(Baumann et al., 2007) have been found to play roles inchromosome cohesion. It will be of interest to determinewhether any of these proteins are phosphorylated directly byCDK11p58. However, it is very likely that many of the mitoticabnormalities we have identified in this study are indirect andarise from defective kinase cascades. The observation that

Journal of Cell Science 120 (14)

CDK11 RNAi expression decreased Plk1localization at the centromere (Petretti et al.,2006) strongly supports this conclusion.Indeed, multiple kinase cascades involvingkinases that regulate CDK11 or are regulatedby CDK11, such as BubR1, Mps1, Plk1,aurora, CDK1, MAPK and others, are likelyto exist. In addition, the CDK11p58 amino acidsequence has strong homology to CDK1 andpossesses all of the sites and motifs thatregulate the mitotic activity of CDK1(Trembley et al., 2004), suggesting the mitotickinase activity of CDK11 might be regulatedby posttranslational modifications. It remainsto be determined whether modification ofthese sites regulates the activity of CDK11.Nonetheless, the collection of mitoticabnormalities resulting from CDK11depletion clearly establishes a role for CDK11as a mitotic kinase.

Materials and MethodsCells and antibodiesHeLa, 293T, U2OS and SAOS2 cells were maintained inDulbecco’s modified Eagle’s medium (DMEM)supplemented with 10% fetal calf serum, 2% glutamine and0.1% penicillin-streptomycin at 37°C with 5% CO2.Transfections were performed using PolyFect (Qiagen),Lipofectamine 2000 (Invitrogen) or Dharmafect1(Dharmacon) according to the manufacturer’s instructions. Inthe case of the pSUPER constructs, transfections were doneby electroporation. The following antibodies were used: P1Cantibody to CDK11 (Trembley et al., 2002), cyclin A(NeoMarkers), cyclin B1 (Santa Cruz and Abcam), �-tubulin(Sigma), �-tubulin (Cell Skeleton), anti-Myc monoclonalantibody (mAb) (Upstate-Millipore), Bub1 (ChemiconInternational), Sgo1 (kindly provided by H. Yu, Universityof Texas Southwestern Medical Center, Dallas) and actin(Santa Cruz). Secondary antibodies included anti-mouse IgG-Texas Red (Amersham Biosciences), anti-rabbit IgG-FITC(Amersham Biosciences), anti-mouse IgG-Cy3 (JacksonImmunoResearch), anti-rabbit IgG-Cy3 (JacksonImmunoResearch) and anti-human IgG-FITC (Pierce).

For studies using microtubule toxin treatment to examinethe spindle checkpoint, nocodazole (0.1 �g/ml) or taxol (33nM) were added to the culture media, and the cells wereharvested at the indicated times for further analysis.

CDK11 RNAiThe pSUPER siRNA expression vector (Brummelkamp et al.,2002), kindly provided by Reuven Agami (NetherlandsCancer Institute, Amsterdam), was used for construction ofCDK11 RNAi constructs containing 19-nucleotide sequencesdesigned to target various regions of CDK11, as previouslydescribed (Brummelkamp et al., 2002). The 19-nucleotide

CDK11-targeting sequences were A461: 5�-GGGAGATG GCAAGGGAGCA-3�(CDK11 RNAi-N); A2184: 5�-GAGCGAGCAGCAGCGT GTG-3� (CDK11 RNAi-C); A910: 5�-GGGAGCACCAGTGAAGAAT-3�; A111: 5�-TTCTGATGACC -GGGATTCC-3�; A188: 5�-GGAACTCCCCGTATAGAAG-3�; and A1361: 5�-AGCGGCTGAAGATGGAGAA-3�. For electroporation delivery of pSUPER RNAiconstructs, cells were grown to ~85% confluence, harvested, and suspended at adensity of approximately 107 cells/ml. Cells (400 �l of the suspension) were thenadded to a 4-mm electroporation cuvette (BioRad), along with 25 �g of plasmidDNA. Electroporation was performed using an Electro Cell Manipulator (BTXECM 630, BTX Instruments) at 220 V and 960 �F capacitance. Afterelectroporation, cells were transferred to plates with complete DMEM forcultivation. Apoptosis and cell-cycle analysis of the CDK11 RNAi-treated cellswere examined by annexin V staining and flow cytometry, respectively.Immunobloting with cell lysates prepared from CDK11 RNAi-treated cells andcontrol cells was performed with the indicated antibodies used as describedthroughout this study. The pSUPER Plk1 and Sgo1 siRNA expression constructswere constructed similarly using the following oligonucleotides: Sgo1,

Fig. 7. Analysis of mitotic chromosomes in cells treated with Plk1, Sgo1 and CDK11RNAi expression constructs. (A,B) Quantitation of the chromosome phenotypes fromPlk1, Sgo1 and CDK11 RNAi cells in a representative experiment. HeLa cells weretransfected with CDK11, Plk1, Sgo1, Plk1-CDK11, Sgo1-CDK11, Sgo1-Plk1 or controlRNAi expression constructs. Mitotic spreads were prepared and Giesma stained 48 hourslater. More than 100 mitotic cells were scored for chromosome appearance. (C) Theappearance of characteristic mitotic chromosomes. Bar, 3 �m.

Jour

nal o

f Cel

l Sci

ence

2433CDK11 protects sister chromosome cohesion

5�-CAGUAGAACCUGCUCAGAA-3�; and Plk1, 5�-CGAGCUGCUUAAUG AC -GAG-3�) (McGuinness et al., 2005). The A1361 CDK11 siRNA was synthesizedfrom Dharmacon and transfected into HeLa cells using 4 �l of Dharmafect1 and 50nM siRNA oligonucleotides in each well of a six-well dish according to themanufacturer’s instructions (Dharmacon).

CDK11p58-GFP cell linesThe plasmids used for generation of stable HeLa tranfectants were constructed bycloning the CDK11p58 cDNA or a CDK11p58 cDNA with a silent mutation in theCDK11 RNAi targeting region (the mutated CDK11 siRNA targeting sequence:5�-AAGTGAACAACAGCGTGTG-3�) into pEGFP-N3 vector, whose CMVpromoter was replaced with an SV40 promoter taken from a pSVL plasmid by PCRto yield a C-terminal p58-GFP fusion protein. The resulting plasmids weretransfected into HeLa cells, and single cells were cloned under neomycin selectionand GFP sorting.

Immunofluorescence analysisHeLa cells, cultured on coverslips for 48 hours after being transfected with RNAiconstructs with or without microtubule toxin treatment, were fixed for 10 minuteswith 4% formaldehyde and permeabilized for 3 minutes with 0.3% Triton X-100/phosphate-buffered saline (PBS) and then blocked with 10% fetal bovine serum(FBS) for 30 minutes followed by incubation with various primary antibodies for 1hour at 37°C. The coverslips were then washed and incubated with the fluorchrome-conjugated secondary antibodies for 1 hour at 37°C, washed with PBS and mountedwith Vectashield mounting medium containing 0.05 �g/ml DAPI. Samples wereanalyzed either on a Nikon C1si Eclipse TE2000-E using a 60� objective (NA 1.45)with images acquired using EZ C1 software or on a Nikon Eclipse E-80i using a100� objective (NA 1.35) with images acquired with a DS-SM-L digital camerausing the Nikon DS-L1 software. TIFF image files were further manipulated usingAdobe Photoshop 7.0.

FISH analysisFISH analysis was performed as described previously (Gururajan et al., 1998) witha BAC clone containing the human histone cluster located at 6p21 (RPCI-11-300H3) and a plasmid containing human chromosome 1 heterochromatin repeats(pUC 1.77). Specific hybridization signals were detected with anti-digoxigeninFITC (Roche Biochemicals) followed by counter-staining with DAPI. The distancebetween sister chromatids of 25 G2 cells from each RNAi-treated group wasdetermined by measuring the distance between paired FISH signals from the samecell for multiple sets of signals; 25 independent G2 cell measurements were madefor each G2-synchronized RNAi group from photomicrographs.

Mitotic chromosome analysis and mitotic index analysisCells transfected with RNAi constructs were harvested with trypsin and subjectedto a hypotonic treatment for chromosome mitotic spread and Giemsa staining. Foreach sample, a mitotic index was determined by dividing the number of mitotic cellsby the total number of cells examined in a sample. More than 1000 cells werecounted in each experiment and the experiments were performed multiple timeswith similar data. Chromosome appearance and occurrence frequency weredetermined by analyzing more than 100 mitotic cells for each RNAi condition in asingle experiment.

This paper is dedicated to the memory and scientific lifeachievements of Vincent J. Kidd, who died 7 May 2004. He will beremembered as a dedicated scientist, an exceptional colleague, anenthusiastic mentor and a dear friend by all of us. The authors thankJose Grenet, Simon Moshiach and Kejin Zhu for excellent technicalsupport, and the members of the Lahti and Kidd laboratory, especiallyJ. Trembley and A. Inoue, and K. Kitagawa, R. Kitagawa and J.Partridge for support, encouragement and helpful discussions. Wethank R. Agami for the pSUPER expression vector; J. Peters forproviding Myc-human Scc1-expressing HeLa cells; K. Sullivan forHeLa H2B GFP cells; K. Kitagawa, G. J. Gorbsky and H. Yu forantibodies; and R. Giet for helpful discussion and sharing data beforepublication, and D. Galloway for editorial assistance. We alsoacknowledge the Cytogenetics and Microinjection facilities and theHartwell Center for assistance and reagents. Support for these studieswas provided by NIH grants GM44088 and P30 CA21765 and theAmerican Lebanese Syrian Associated Charities (ALSAC).

ReferencesAcquaviva, C., Herzog, F., Kraft, C. and Pines, J. (2004). The anaphase promoting

complex/cyclosome is recruited to centromeres by the spindle assembly checkpoint.Nat. Cell Biol. 6, 892-898.

Barr, F. A., Sillje, H. H. and Nigg, E. A. (2004). Polo-like kinases and the orchestrationof cell division. Nat. Rev. Mol. Cell Biol. 5, 429-440.

Baumann, C., Hofmann, K. and Nigg, E. A. (2007). PICH, a centromere-associatedSNF2 family ATPase, is regulated by Plk1 and required for the spindle checkpoint.Cell 128, 101-114.

Bieche, I., Champeme, M.-H., Matifas, F., Cropp, C. S., Callahan, R. and Lidereau,

Fig. 8. Roles of CDK11p58

in regulation of sisterchromatid cohesion duringmitosis. CDK11p58 protectssister-chromatid cohesion atboth the arms and at thecentromere. The dashedcircles denote thehypothetical protectiveaction of CDK11p58 onchromosome cohesionrather than localization ofCDK11p58 to thechromosome. SACrepresents the spindlecheckpoint. Prematuredissociation of Bub1 andcohesion complexes occursin CDK11-depleted cells.*Plk1 binding tokinetochore is reduced inCDK11 RNAi cells (Petrettiet al., 2006), which mayalso influence chromosomecohesion.

Jour

nal o

f Cel

l Sci

ence

R. (1993). Two distinct regions involved in 1p deletion in human primary breast cancer.Cancer Res. 53, 1990-1994.

Blagosklonny, M. V. (2007). Mitotic arrest and cell fate: why and how mitotic inhibitionof transcription drives mutually exclusive events. Cell Cycle 6, 70-74.

Brodeur, G. M., Sekhon, G. and Goldstein, M. N. (1977). Chromosomal aberrations inhuman neuroblastomas. Cancer 40, 2256-2263.

Brummelkamp, T. R., Bernards, R. and Agami, R. (2002). A system for stableexpression of short interfering RNAs in mammalian cells. Science 296, 550-553.

Bunnell, B. A., Heath, L. S., Adams, D. E., Lahti, J. M. and Kidd, V. J. (1990).Increased expression of a 58-kDa protein kinase leads to changes in the CHO cell cycle.Proc. Natl. Acad. Sci. USA 87, 7467-7471.

Castedo, M., Perfettini, J. L., Roumier, T., Andreau, K., Medema, R. and Kroemer,G. (2004). Cell death by mitotic catastrophe: a molecular definition. Oncogene 23,2825-2837.

Chandramouli, A., Shi, J., Feng, Y., Holubec, H., Shanas, R., Bhattacharyya, A. K.,Zheng, W. and Nelson, M. A. (2007). Haploinsufficiency of the cdc2l gene contributesto skin cancer development in mice. Carcinogenesis doi: 10.1093/carcin/bgm066.

Chen, H. H., Wang, Y. C. and Fann, M. J. (2006). Identification and characterizationof the CDK12/cyclin L1 complex involved in alternative splicing regulation. Mol. Cell.Biol. 26, 2736-2745.

Chen, R. H. (2002). BubR1 is essential for kinetochore localization of other spindlecheckpoint proteins and its phosphorylation requires Mad1. J. Cell Biol. 158, 487-496.

Cleveland, D. W., Mao, Y. and Sullivan, K. F. (2003). Centromeres and kinetochores:from epigenetics to mitotic checkpoint signaling. Cell 112, 407-421.

Cornelis, S., Bruynooghe, Y., Denecker, G., Van Huffel, S., Tinton, S. and Beyaert,R. (2000). Identification and characterization of a novel cell cycle-regulated internalribosome entry site. Mol. Cell 5, 597-605.

Dai, J., Sullivan, B. A. and Higgins, J. M. (2006). Regulation of mitotic chromosomecohesion by Haspin and Aurora B. Dev. Cell 11, 741-750.

Gandhi, R., Gillespie, P. J. and Hirano, T. (2006). Human Wapl is a cohesin-bindingprotein that promotes sister-chromatid resolution in mitotic prophase. Curr. Biol. 16,2406-2417.

Gimenez-Abian, J. F., Sumara, I., Hirota, T., Hauf, S., Gerlich, D., de la Torre, C.,Ellenberg, J. and Peters, J. M. (2004). Regulation of sister chromatid cohesionbetween chromosome arms. Curr. Biol. 14, 1187-1193.

Gururajan, R., Lahti, J. M., Grenet, J., Easton, J., Gruber, I., Ambros, P. F. andKidd, V. J. (1998). Duplication of a genomic region containing the Cdc2L1-2 andMMP21-22 genes on human chromosome 1p36.3 and their linkage to D1Z2. GenomeRes. 8, 929-939.

Harper, J. W. and Adams, P. D. (2001). Cyclin-dependent kinases. Chem. Rev. 101,2511-2526.

Hauf, S., Waizenegger, I. C. and Peters, J. M. (2001). Cohesin cleavage by separaserequired for anaphase and cytokinesis in human cells. Science 293, 1320-1323.

Hauf, S., Roitinger, E., Koch, B., Dittrich, C. M., Mechtler, K. and Peters, J. M.(2005). Dissociation of cohesin from chromosome arms and loss of arm cohesionduring early mitosis depends on phosphorylation of SA2. PLoS Biol. 3, e69.

Hu, D., Mayeda, A., Trembley, J. H., Lahti, J. M. and Kidd, V. J. (2003). CDK11complexes promote pre-mRNA splicing. J. Biol. Chem. 278, 8623-8629.

Kenney, R. D. and Heald, R. (2006). Essential roles for cohesin in kinetochore andspindle function in Xenopus egg extracts. J. Cell Sci. 119, 5057-5066.

Kitajima, T. S., Hauf, S., Ohsugi, M., Yamamoto, T. and Watanabe, Y. (2005). HumanBub1 defines the persistent cohesion site along the mitotic chromosome by affectingShugoshin localization. Curr. Biol. 15, 353-359.

Kueng, S., Hegemann, B., Peters, B. H., Lipp, J. J., Schleiffer, A., Mechtler, K. andPeters, J. M. (2006). Wapl controls the dynamic association of cohesin with chromatin.Cell 127, 955-967.

Jackman, M., Lindon, C., Nigg, E. A. and Pines, J. (2003). Active cyclin B1–Cdk1first appears on centrosomes in prophase. Nat. Cell Biol. 5, 143-148.

Lahti, J. M., Xiang, J., Heath, L. S., Campana, D. and Kidd, V. J. (1995). PITSLREprotein kinase activity is associated with apoptosis. Mol. Cell. Biol. 15, 1-11.

Lampson, M. A. and Kapoor, T. M. (2005). The human mitotic checkpoint proteinBubR1 regulates chromosome-spindle attachments. Nat. Cell Biol. 7, 93-98.

Lampson, M. A., Renduchitala, K., Khodjakov, A. and Kapoor, T. M. (2004).Correcting improper chromosome-spindle attachments during cell division. Nat. CellBiol. 6, 232-237.

Lew, D. J. and Kornbluth, S. (1996). Regulatory roles of cyclin dependent kinasephosphorylation in cell cycle control. Curr. Opin. Cell Biol. 8, 795-804.

Lew, D. J. and Burke, D. J. (2003). The spindle assembly and spindle positioncheckpoints. Annu. Rev. Genet. 37, 251-282.

Li, T., Inoue, A., Lahti, J. M. and Kidd, V. J. (2004). Failure to proliferate and mitoticarrest of CDK11p110/p58-null mutant mice at the blastocyst stage of embryonic celldevelopment. Mol. Cell. Biol. 24, 3188-3197.

Losada, A., Hirano, M. and Hirano, T. (2002). Cohesin release is required for sisterchromatid resolution, but not for condensin-mediated compaction, at the onset ofmitosis. Genes Dev. 16, 3004-3016.

McGuinness, B. E., Hirota, T., Kudo, N. R., Peters, J. M. and Nasmyth, K. (2005).Shugoshin prevents dissociation of cohesin from centromeres during mitosis invertebrate cells. PLoS Biol. 3, e86.

Melendez, B., Cuadros, M., Robledo, M., Rivas, C., Fernandez-Piqueras, J.,Martinez-Delgado, B. and Benitez, J. (2003). Coincidental LOH regions in mouseand humans: evidence for novel tumor suppressor loci at 9q22-q34 in non-Hodgkin’slymphomas. Leuk. Res. 27, 627-633.

Meraldi, P., Draviam, V. M. and Sorger, P. K. (2004). Timing and checkpoints in theregulation of mitotic progression. Dev. Cell 7, 45-60.

Mori, N., Morosetti, R., Mizoguchi, H. and Koeffler, H. P. (2003). Progression ofmyelodysplastic syndrome: allelic loss on chromosomal arm 1p. Br. J. Haematol. 122,226-230.

Musacchio, A. and Hardwick, K. G. (2002). The spindle checkpoint: structural insightsinto dynamic signalling. Nat. Rev. Mol. Cell Biol. 3, 731-741.

Nicklas, R. B., Ward, S. C. and Gorbsky, G. J. (1995). Kinetochore chemistry issensitive to tension and may link mitotic forces to a cell cycle checkpoint. J. Cell Biol.130, 929-939.

Peters, J. M. (2002). The anaphase-promoting complex: proteolysis in mitosis andbeyond. Mol. Cell 9, 931-943.

Petretti, C., Savoian, M., Montembault, E., Glover, D. M., Prigent, C. and Giet, R.(2006). The PITSLRE/CDK11p58 protein kinase promotes centrosome maturation andbipolar spindle formation. EMBO Rep. 7, 418-424.

Poetsch, M., Dittberner, T. and Woenckhaus, C. (2003). Microsatellite analysis at1p36.3 in malignant melanoma of the skin: fine mapping in search of a possible tumoursuppressor gene region. Melanoma Res. 13, 29-33.

Salic, A., Waters, J. C. and Mitchison, T. J. (2004). Vertebrate Shugoshin links sistercentromere cohesion and kinetochore microtubule stability in mitosis. Cell 118, 567-578.

Sumara, I., Vorlaufer, E., Stukenberg, P. T., Kelm, O., Redemann, N., Nigg, E. A.and Peters, J. M. (2002). The dissociation of cohesin from chromosomes in prophaseis regulated by Polo-like kinase. Mol. Cell. 9, 515-525.

Sumara, I., Gimenez-Abian, J. F., Gerlich, D., Hirota, T., Kraft, C., de la Torre, C.,Ellenberg, J. and Peters, J. M. (2004). Roles of polo-like kinase 1 in the assemblyof functional mitotic spindles. Curr. Biol. 14, 1712-1722.

Tang, Z., Bharadwaj, R., Li, B. and Yu, H. (2001). Mad2-Independent inhibition ofAPCCdc20 by the mitotic checkpoint protein BubR1. Dev. Cell 1, 227-237.

Tang, Z., Shu, H., Oncel, D., Chen, S. and Yu, H. (2004a). Phosphorylation of Cdc20by Bub1 provides a catalytic mechanism for APC/C inhibition by the spindlecheckpoint. Mol. Cell 16, 387-397.

Tang, Z., Sun, Y., Harley, S. E., Zou, H. and Yu, H. (2004b). Human Bub1 protectscentromeric sister-chromatid cohesion through Shugoshin during mitosis. Proc. Natl.Acad. Sci. USA 101, 18012-18017.

Tang, Z., Shu, H., Qi, W., Mahmood, N. A., Mumby, M. C. and Yu, H. (2006). PP2Ais required for centromeric localization of Sgo1 and proper chromosome segregation.Dev. Cell 10, 575-585.

Tinton, S. A., Schepens, B., Bruynooghe, Y., Beyaert, R. and Cornelis, S. (2005).Regulation of the cell-cycle-dependent internal ribosome entry site of the PITSLREprotein kinase: roles of Unr (upstream of N-ras) protein and phosphorylated translationinitiation factor eIF-2alpha. Biochem. J. 385, 155-163.

Trembley, J. H., Hu, D., Hsu, L. C., Yeung, C. Y., Slaughter, C., Lahti, J. M. andKidd, V. J. (2002). PITSLRE p110 protein kinases associate with transcriptioncomplexes and affect their activity. J. Biol. Chem. 277, 2589-2596.

Trembley, J. H., Hu, D., Slaughter, C. A., Lahti, J. M. and Kidd, V. J. (2003). Caseinkinase 2 interacts with cyclin-dependent kinase 11 (CDK11) in vivo and phosphorylatesboth the RNA polymerase II carboxyl-terminal domain and CDK11 in vitro. J. Biol.Chem. 278, 2265-2270.

Trembley, J. H., Loyer, P., Hu, D., Li, T., Grenet, J., Lahti, J. M. and Kidd, V. J.(2004). Cyclin dependent kinase 11 in RNA transcription and splicing. Prog. NucleicAcid Res. Mol. Biol. 77, 263-288.

van Vugt, M. A. and Medema, R. H. (2005). Getting in and out of mitosis with Polo-like kinase-1. Oncogene 24, 2844-2859.

van Vugt, M. A., van de Weerdt, B. C., Vader, G., Janssen, H., Calafat, J.,Klompmaker, R., Wolthuis, R. M. and Medema, R. H. (2004). Polo-like kinase-1 isrequired for bipolar spindle formation but is dispensable for anaphase promotingcomplex/Cdc20 activation and initiation of cytokinesis. J. Biol. Chem. 279, 36841-36854.

Watanabe, Y. and Kitajima, T. S. (2005). Shugoshin protects cohesin complexes atcentromeres. Philos. Trans. R. Soc. Lond. B Biol. Sci. 360, 515-521.

Wei, Y., Yu, L., Bowen, J., Gorovsky, M. A. and Allis, C. D. (1999). Phosphorylationof histone H3 is required for proper chromosome condensation and segregation. Cell97, 99-109.

Wilker, E. W., van Vugt, M. A., Artim, S. A., Huang, P. H., Petersen, C. P.,Reinhardt, H. C., Feng, Y., Sharp, P. A., Sonenberg, N., White, F. M. et al. (2007).14-3-3sigma controls mitotic translation to facilitate cytokinesis. Nature 446, 329-332.

Journal of Cell Science 120 (14)2434

Jour

nal o

f Cel

l Sci

ence