CLASPs prevent irreversible multipolarity by ensuring spindle-pole resistance to traction forces during chromosome alignment

Article

Reference

CLASPs prevent irreversible multipolarity by ensuring spindle-pole resistance to traction forces during chromosome alignment

LOGARINHO, Elsa, et al.

Abstract

Loss of spindle-pole integrity during mitosis leads to multipolarity independent of centrosome amplification. Multipolar-spindle conformation favours incorrect kinetochore-microtubule attachments, compromising faithful chromosome segregation and daughter-cell viability.

Spindle-pole organization influences and is influenced by kinetochore activity, but the molecular nature behind this critical force balance is unknown. CLASPs are microtubule-, kinetochore- and centrosome-associated proteins whose functional perturbation leads to three main spindle abnormalities: monopolarity, short spindles and multipolarity. The first two reflect a role at the kinetochore-microtubule interface through interaction with specific kinetochore partners, but how CLASPs prevent spindle multipolarity remains unclear. Here we found that human CLASPs ensure spindle-pole integrity after bipolarization in response to CENP-E- and Kid-mediated forces from misaligned chromosomes. This function is independent of end-on kinetochore-microtubule attachments and involves the recruitment of ninein to residual pericentriolar satellites. Distinctively, [...]

LOGARINHO, Elsa, et al . CLASPs prevent irreversible multipolarity by ensuring spindle-pole resistance to traction forces during chromosome alignment. Nature Cell Biology , 2012, vol. 14, no. 3, p. 295-303

DOI : 10.1038/ncb2423 PMID : 22307330

Available at:

http://archive-ouverte.unige.ch/unige:28850

Disclaimer: layout of this document may differ from the published version.

L E T T E R S

CLASPs prevent irreversible multipolarity by ensuring spindle-pole resistance to traction forces during

chromosome alignment

Elsa Logarinho

, Stefano Maffini

, Marin Barisic

, Andrea Marques

, Alberto Toso

, Patrick Meraldi

and Helder Maiato

Loss of spindle-pole integrity during mitosis leads to multipolarity independent of centrosome amplification^1–4. Multipolar-spindle conformation favours incorrect

kinetochore–microtubule attachments, compromising faithful chromosome segregation and daughter-cell viability^5,6. Spindle-pole organization influences and is influenced by kinetochore activity^7,8, but the molecular nature behind this critical force balance is unknown. CLASPs are microtubule-, kinetochore- and centrosome-associated proteins whose functional perturbation leads to three main spindle abnormalities: monopolarity, short spindles and multipolarity^9–13. The first two reflect a role at the

kinetochore–microtubule interface through interaction with specific kinetochore partners^10,11,14, but how CLASPs prevent spindle multipolarity remains unclear. Here we found that human CLASPs ensure spindle-pole integrity after

bipolarization in response to CENP-E- and Kid-mediated forces from misaligned chromosomes. This function is independent of end-on kinetochore–microtubule attachments and involves the recruitment of ninein to residual pericentriolar satellites.

Distinctively, multipolarity arising through this mechanism often persists through anaphase. We propose that CLASPs and ninein confer spindle-pole resistance to traction forces exerted during chromosome congression, thereby preventing

irreversible spindle multipolarity and aneuploidy.

Multipolar spindles are a hallmark of tumour cells and may arise owing to supernumerary centrosomes resulting from centrosome overduplication or cytokinesis failure. However, multipolar mitosis is usually incompatible with cell viability and normally assumes a transient nature due to the coalescence of extra centrosomes into two functional spindle poles^5,6. An alternative but less understood

1Instituto de Biologia Molecular e Celular, Universidade do Porto, Rua do Campo Alegre 823, 4150-180 Porto, Portugal.²Institute of Biochemistry, ETH Zurich, 8093 Zurich, Switzerland.³Department of Experimental Biology, Faculdade de Medicina, Universidade do Porto, 4200-319 Porto, Portugal.⁴Present addresses: Max Planck Institute of Molecular Physiology, 44202 Dortmund, Germany (S.M.); Cell Biology, Faculty of Science, Utrecht University, 3584-CH Utrecht, The Netherlands (A.M.);

Institute of Oncology Research, CH-6500 Bellinzona, Switzerland (A.T.).

5Correspondence should be addressed to H.M. (e-mail: [email protected])

Received 30 August 2011; accepted 15 December 2011; published online 5 February 2012; DOI: 10.1038/ncb2423

mechanism is related to multipolar-spindle formation independent of centrosome amplification, for example due to premature centriole disengagement or loss of spindle-pole integrity^1–4,15. To investigate the mechanism by which CLASPs (cytoplasmic linker-associated proteins) prevent spindle multipolarity in human cells, we used HeLa cells stably expressing the centriole marker centrin–GFP and immunodetection ofγ-tubulin to determine the number of centrioles in each individual pole on CLASP1/2 depletion by RNAi. As positive controls, we used cells treated with either 2µM cytochalasin D, an inhibitor of actin polymerization and cytokinesis; astrin (also known asSPAG5, sperm- associated antigen 5) short interfering RNA (siRNA), which leads to premature centriole disengagement¹⁵; orTOGp(also known asCKAP5, cytoskeleton-associated protein 5) siRNA, which perturbs spindle-pole integrity^2,3 (Fig. 1a–c and Supplementary Fig. S1a–d). All of these treatments caused a significant increase in the percentage of mitotic cells with multipolar spindles (Fig. 1a). Simultaneous depletion of CLASP1 and CLASP2 (Supplementary Fig. S1b) resulted in 17.3±4.1% of mitotic cells with multipolar spindles, whereas individual depletion of CLASP1 or CLASP2 caused, respectively, 6.3±0.9% and 5.1±0.2% of multipolar mitosis (Fig. 1a). These frequencies are significantly higher than in control cells (∼1%; Fig. 1a), indicating that the two human CLASPs cooperate to prevent spindle multipolarity. Interestingly, 29.9±9.5% of poles in CLASP1/2-depleted multipolar cells were acentriolar and 20.7±8.5% had a single centriole, a phenotypic distribution that was distinct from that of cytochalasin D treatment or astrin RNAi, but resembled that ofTOGpRNAi (Fig. 1b,c). However, CLASP1/2 depletion did not affect TOGp localization at spindle poles (Supplementary Fig. S2), indicating that the observed multipolarity is independent of TOGp and is primarily caused by loss of spindle-pole integrity and, to a lesser extent, due to centriole disengagement.

We have previously identified astrin as part of a CLASP1 complex involved in the regulation of kinetochore–microtubule dynamics^11,14.

L E T T E R S

Centrin–GFP DAPI

RNAi

RNAi

1

2

1

1

2 3

4

1 2

3 1

2

3

4

1

2

3

1

1

1

1 2

2

2

2 3

3

3

3 4

4 Luc RNAi

b

2 Cytochalasin D

Untreate d

CLASP1/

2 RNAi CLASP1

RNAi CLASP2

RNAi TOGp

RNAi Astrin RNAi

Percentage of multipolar cells Percentage of poles with n centrioles

α-tubulin γ-tubulin

0 5 50 45 40 35 30 25 20 15 10

Cytoch D Astrin TOGp CLASP1/2 0

100 80 60 40 20

0 1 2 3 4

AstrinCLASP1/2ControlCytochalasin DTOGp

a

c

Centrin–GFP α-tubulin γ-tubulin

Figure 1CLASP1/2 depletion causes the formation of multipolar spindles.

HeLa cells stably expressing the centriole marker centrin–GFP were either treated with 2µM cytochalasin D or transfected with the indicated specific siRNA, and then stained for α-tubulin, γ-tubulin and DNA.

(a) Percentage of mitotic cells with multipolar spindles (mean±s.d.; n=1,000 cells). Luciferase (Luc) RNAi was used as a negative control.

(b) Percentage of poles withn centrioles found in cells with multipolar spindles generated by treatment with cytochalasin D (Cytoch D) or

siRNAs against astrin, TOGp and CLASPs (mean±s.d.; n=30 cells).

(c) Examples of cells with multipolar spindles most typically found for each treatment. Insets are magnifications of the centrin–GFP signal at the indicated poles. Note that the possibility of centrosome overduplication was discarded because the frequency of CLASP1/2-depleted mononucleated cells entering mitosis with more than two centrosomes was less than 1% (our unpublished observations). Quantitative data are means of three independent experiments. Scale bars, 5µm.

To determine whether the eventual centriole disengagement observed in CLASP1/2-depleted multipolar spindles arises from premature activation of separase as seen on astrin depletion¹⁵, we quantified loss of sister-chromatid cohesion in chromosome spreads from cells transfected with control,SGOL1(which encodes a centromeric protein involved in sister-chromatid cohesion, Sgo1)¹⁶, astrin orCLASP1/2

siRNAs. We found that∼70% of CLASP1/2-depleted cells showed normal sister-chromatid cohesion, corresponding to 5- and 15-fold the respective percentages in astrin- or Sgo1-depleted cells (Supplementary Fig. S3a,b). Moreover, separase depletion, which rescues spindle bipolarity on astrin RNAi (ref. 15), had no effect in the observed multipolarity after CLASP1/2 depletion (Supplementary Fig. S3c).

L E T T E R S

0:00 0:00 NEB

NEB 1

2

2

1

2

1

2

a d

Time (min)

Control CLASP1/

2 RNAi BPCLASP1/

2 RNAi MP

CLASP1/

2 RNAi + Mps1-IN-1CLASP1/

2 RNAi MP Control +

Mps1-IN-1 b

c

e

f RNAi

1

2

0:17

Pole fragmentation

Pole fragmentation

1 0:16

1

2

3 1

2 1 3

2

1

1

2 1

2

2 3

3 2

1

2 1

2 1

2 1

2 3 3

2 1

2 1

2 1

3

2 1

3

Mock Nuf2 CLASP1/

2 CLASP1/

2 + Nuf2 RNAi 2

1

2 1

1

2 2 1

ControlCLASP1/2 RNAiCLASP1/2 RNAi

1:06

0:46

3:06

1:06

0:00 0:16 0:36 0:42

NEB Anaphase onset

-tubulinα Centrin–GFP + Hec1DAPI

Mock CLASP1/2 Nuf2 CLASP1/2 + Nuf2

Percentage of multipolar cells

1,000

100

10

1

0 10 20 30 40 50 1

NEB–ANA NEB–MP

Figure 2CLASPs ensure spindle-pole integrity independently of forces relying on end-on kinetochore–microtubule attachments. (a) HeLa cells stably co-expressing EGFP–centrin-2 andα-tubulin–mRFP were either mock transfected (control) or RNAi transfected (CLASP1/2RNAi) for 72 h.

(b,c) Following CLASP1/2 depletion, spindles shorten and multipolarity arises as a consequence of pole fragmentation without (b) or with centriole disengagement (c). For each cell, magnifications of the centrioles in the indicated poles are shown after contrast inversion of the EGFP–centrin-2 signal (insets). Arrowheads indicate newly formed poles. Time, h:min.

(d) Quantitative analysis of the time between NEB and anaphase onset (NEB–ANA) in control (n=19 cells) andCLASP1/2-RNAi cells (BP, bipolar; MP, multipolar;n=64 cells), untreated (n=9 cells) or treated

(n=12 cells) with Mps1 inhibitor (Mps1-IN-1). Time from NEB to multipolarity (NEB–MP) inCLASP1/2-RNAi cells was also measured. The mean is shown as horizontal bars and variability within the population is provided in the main text as s.d. (e) Percentages of mitotic cells with multipolar spindles in the indicated RNAi depletions (mean±s.d.; three independent experiments). (f) Centrin–GFP HeLa cells were transfected with control,CLASP1/2,NUF2 andCLASP 1/2+NUF 2 siRNAs, and immunostained forα-tubulin and Hec1. Hec1 detection was used to quantify NUF2 depletion because the stability of Hec1, a subunit of the Ndc80 complex, depends on NUF2. Absence of end-on attachments induced byNUF2 RNAi exacerbated spindle multipolarity onCLASP1/2 RNAi. Scale bars, 5µm.

Finally, the recruitment of astrin to spindle poles was unaffected in CLASP1/2-depleted cells (Supplementary Fig. S2). These data indicate that loss of spindle pole integrity and centriole disengagement observed in CLASP1/2-depleted cells is independent of astrin and premature separase activation.

To directly correlate centriole and mitotic spindle behaviour in CLASP1/2-depleted cells, we carried out live imaging of HeLa cells stably co-expressing EGFP–centrin-2 andα-tubulin–mRFP. Control cells assembled a bipolar spindle and progressed from nuclear envelope breakdown (NEB) into anaphase in 33.1±11.3 min (Fig. 2a,d and Supplementary Movie S1). CLASP1/2-depleted cells established short bipolar spindles, with 50% entering anaphase in 96.7±75.8 min and the other 50% becoming multipolar after a variable delay

(NEB to multipolarity: 80.9±55.7 min; Fig. 2b–d). A significant fraction of CLASP1/2-depleted multipolar cells (8/34) ended up satisfying the mitotic checkpoint and underwent a multipolar anaphase during our recordings (NEB to anaphase: 333.1±113.8 min; Fig. 2d;

Supplementary Movie S2). Inspection of spindle poles in CLASP1/2- depleted cells as they become multipolar revealed that in ∼70%

of the cases the newly formed poles were acentriolar and derived from fragmentation of pre-existing poles (Fig. 2b and Supplementary Movie S2). In the other 30%, there was centriole disengagement leading to the formation of two mono-centriolar poles (Fig. 2c and Supplementary Movie S3). Importantly, all multipolar spindles in CLASP1/2-depleted cells arose after the establishment of a bipolar spindle and their maintenance required pushing forces mediated by

L E T T E R S

the kinesin-5 Eg5 (Supplementary Fig. S4 and Movies S4 and S5). The observation that 50% of CLASP1/2-depleted cells with short spindles ultimately become multipolar indicates that the relatively moderate levels of multipolarity in fixed cells is an underestimation of the real penetrance of this phenotype.

It was recently reported that extensive metaphase delay causes

‘cohesion fatigue’, leading to precocious sister-chromatid separation, followed by centriole disengagement and spindle multipolarity in a separase-independent manner^17,18. To rule out the possibility that multipolarity in CLASP1/2-depleted cells was mainly due to a metaphase delay, we treated control cells with the proteasome inhibitor MG132 for 3 h, which blocks anaphase onset without preventing mitotic-checkpoint satisfaction. No increase in the normal number of cells with multipolar spindles was observed (our unpublished observations). Although we cannot exclude that cumulative events due to CLASP1/2 depletion result in some cohesion fatigue (see Supplementary Fig. S3a,b), the initial spindle-pole fragmentation events inCLASP1/2RNAi occur at least 1.5 h before the first manifestations of centriole disengagement and consequent multipolarity normally associated with cohesion fatigue^17,18. Although not sufficient, a delay in mitotic-checkpoint satisfaction might be necessary to cause loss of spindle-pole integrity in CLASP1/2-depleted cells. To investigate this, we filmed CLASP1/2-depleted cells in the presence of an Mps1 inhibitor necessary for mitotic-checkpoint activity¹⁹. Under these conditions, all control andCLASP1/2-RNAi cells entered anaphase with bipolar spindles after 11.7±2.4 min and 14.6±4.7 min, respectively (Fig. 2d and Supplementary Fig.

S5a), indicating that spindle multipolarity observed on CLASP1/2 depletion is associated with conditions that prevent mitotic-checkpoint satisfaction (presumably due to unattached kinetochores).

We next investigated whether loss of spindle-pole integrity on CLASP1/2 depletion was related to a role in kinetochore–microtubule dynamics¹¹. First, we perturbed end-on kinetochore–microtubule attachments by co-depleting CLASPs with NUF2 (ref. 8). Surprisingly,

∼20% of NUF2-depleted cells assembled multipolar spindles (Fig. 2e,f and Supplementary Fig. S1e), which was probably due to cohesion fatigue^17,18, because extensive sister-chromatid separation was observed in some, but not all, chromosomes (Supplementary Fig. S3a,b) and the finding that even laterally attached kinetochores can be significantly stretched²⁰. Importantly, co-depletion of CLASPs and NUF2 synergistically exacerbated the spindle-multipolarity phenotype (Fig. 2e,f and Supplementary Fig. S1e), indicating that the role of CLASPs in the maintenance of spindle-pole integrity is independent of end-on kinetochore–microtubule attachments^11,14and that the multipolarity observed on CLASP1/2 and NUF2 individual depletions arises through independent mechanisms. Finally, as CLASP1/2 depletion increases kinetochore-microtubule stability¹¹, we investigated whether this could explain the observed multipolarity. To do so, we hyperstabilized kinetochore microtubules with the aurora B inhibitor ZM447439 (ref. 21), which in control cells recovering from Eg5 inhibition always led to permanent bipolar spindles (8/8), but did not prevent multipolarization of CLASP1/2-depleted bipolar spindles (7/8; Supplementary Fig. S5b). Thus, spindle multipolarity associated with CLASP1/2 depletion is not related to their impact on kinetochore–microtubule dynamics and is independent of end-on kinetochore–microtubule attachments.

To directly investigate how a delay in mitotic-checkpoint satisfaction leads to loss of spindle-pole integrity on CLASP1/2 depletion, we switched to HeLa cells stably co-expressing H2B-histone–GFP and α-tubulin–mRFP to simultaneously monitor chromosome and spindle behaviour. Whereas control cells rapidly completed chromosome congression and entered anaphase shortly thereafter (Fig. 3a), CLASP1/2-depleted cells showed few chromosomes that were unable to congress and/or stabilize their position at the metaphase plate and were significantly delayed in mitosis (Fig. 3b and Supplementary Movie S6). Inspection of recorded cells assisted by four-dimensional reconstructions revealed that in 8/10 cells, loss of spindle-pole integrity on CLASP1/2 depletion was preceded by the presence of misaligned chromosomes along the vector of spindle-pole fragmentation (Fig. 3c and Supplementary Movie S7). Immunofluorescence microscopy analysis further revealed that misaligned chromosomes on CLASP1/2 depletion showed at least one MAD2-positive kinetochore (Fig. 3d).

Chromosome congression in human cells relies on the combined action of chromokinesins, such as Kid (also known as kinesin family member 22, KIF22; a kinesin-10), which generate polar ejection forces that push chromosome arms away from spindle poles^20,22,23, and the plus-end-directed motor CENP-E (centromere-associated protein E; a kinesin-7) at unattached kinetochores, which slides misaligned chromosomes along pre-existing spindle microtubules^24,25. Both activities eventually contribute to microtubule capture from the opposing pole and consequently to chromosome bi-orientation.

CENP-E exists in a complex with CLASPs during mitosis and is required for their proper recruitment to kinetochores¹¹. However, CENP-E is still normally targeted to kinetochores in the absence of CLASPs (ref. 11). To investigate whether loss of spindle-pole integrity on CLASP1/2 depletion was due to CENP-E- and/or Kid-mediated forces at kinetochores/chromosome arms, we co-depleted CLASPs with CENP-E or Kid by RNAi. We observed that virtually all CLASP1/2- and CENP-E-co-depleted cells showed bipolar spindles, representing a 90% reduction in spindle multipolarity (Fig. 4a,b and Supplementary Fig. S1f). CLASP1/2 and Kid co-depletion also significantly reverted spindle multipolarity, albeit only in ∼50%

of the cases (Fig. 4b). These results indicate that, in coordination with Kid-mediated forces on chromosome arms, CENP-E-mediated forces at kinetochores are responsible for the observed spindle multipolarity on CLASP1/2 depletion. This is in line with recent data showing that Kid favours lateral kinetochore–microtubule attachments before chromosome bi-orientation²⁰. To ascertain the nature of the specific interplay between CLASP1/2 and CENP-E in spindle architecture, we co-depleted CENP-E with astrin or TOGp.

Surprising, CENP-E depletion significantly rescued multipolarity of astrin-depleted, but not TOGp-depleted, cells (Fig. 4c). Thus, CLASPs are not the only proteins required for spindle-pole resistance to CENP-E-mediated forces. However, spindle multipolarity was aggravated on CLASP1/2 co-depletion with astrin or TOGp, indicating that these proteins affect spindle-pole integrity through cumulative mechanisms (Fig. 4c).

To distinguish between the role of CENP-E-mediated forces on congressing chromosomes and other possible motor-dependent activities of CENP-E (for example, promotion of microtubule plus-end elongation associated with spindle microtubule flux²⁶) as the cause of spindle multipolarity in CLASP1/2-depleted cells, we used a reversible

L E T T E R S

a

b

0:20 1:30 3:20

*

* 0:00

1:17:30 1:20:00 1:22:30 1:25:00

1:27:30

x y

z

x

x

x x

y y y

y

z

z

z z

x y z

∗ ∗ ∗ ∗

∗ ∗ ∗ ∗

∗ ∗

1:27:30 x y

z

∗ ∗

1:32:30

x y

z

∗

∗

1:30:00 x y

z

∗ ∗

0:00 0:10 4:16 4:28

*

*

4:48

*

*

*

*

0:00 0:15 0:24 0:40 0:48

3:50 c

d _{CLASP1/2 RNAi}

DAPI MAD2

*

∗

ControlCLASP1/2 RNAiCLASP1/2 RNAiFour-dimensional reconstruction

α-tub

Figure 3Loss of spindle-pole integrity in the absence of CLASPs is preceded by the presence of misaligned chromosomes with unattached kinetochores.

Live imaging of control andCLASP1/2-RNAi-depleted HeLa cells stably co-expressing H2B-histone–GFP/α-tubulin–mRFP. (a) Control cells rapidly aligned all chromosomes at the metaphase plate (17.7±2.5 min) and progressed normally into anaphase (NEB-to-anaphase: 33.7±1.8 min;

n=3 cells). (b,c) CLASP1/2-depleted cells typically exhibited a few misaligned chromosomes (arrowheads) for a variable amount of time until

pole fragmentation occurred (asterisks). (NEB-to-anaphase inCLASP1/2 RNAi: 310.4±214.4 min;n=13 cells.) Values indicate mean±s.d.. Three- dimensional-reconstruction image panels of the critical time frames where the presence of misaligned chromosomes preceded the pole fragmentation event are outlined in red. Time, h:min and h:min:s. (d)CLASP1/2-RNAi cells were immunostained forα-tubulin (α-tub), mitotic-checkpoint protein MAD2 and DNA (DAPI). Misaligned chromosomes inCLASP1/2-RNAi cells have at least one MAD2-positive kinetochore. Scale bars, 5µm.

allosteric inhibitor of CENP-E, GSK-923295, which allows temporal control of CENP-E motor activity²⁷. First, we used GSK-923295 in a series of inhibition–washout experiments in living cells stably co-expressing H2B-histone–GFP andα-tubulin–mRFP. All control and more than 80% of CLASP1/2-depleted cells treated with 20 nM GSK-923295 remained bipolar with few misaligned chromosomes for more than 2 h from NEB (Fig. 4d). However, whereas all control cells resumed congression and entered anaphase with bipolar spindles soon after drug washout, 50% of CLASP1/2-depleted cells became multipolar within 1 h (Fig. 4d and Supplementary Movies S8 and S9). Interestingly, addition–washout of GSK-923295 to CLASP1/2-depleted cells that had already lost spindle-pole integrity did not revert multipolarity (5/5), highlighting the irreversible nature of the process (Fig. 4b,d and Supplementary Movie S10). Next, we used photo-conversion of

mEos–α-tubulin to measure spindle microtubule flux in cells treated with GSK-923295, and found no difference relative to control cells (Supplementary Fig. S5c,d). This directly demonstrates that CENP-E ATPase/motor activity is not required for flux. Together, these data indicate that CENP-E-mediated traction forces during chromosome congression are responsible for the irreversible loss of spindle-pole integrity and multipolarity in CLASP1/2-depleted cells.

Spindle-pole integrity is ensured by motor proteins of the kinesin-14 and dynein families²⁸ that work in concert with stabilizing proteins, such as TOGp and Tpx2 (refs 2,3,29), to focus microtubule minus ends at spindle poles. However, the spindle- pole localization of all these proteins was unaffected in CLASP1/2- depleted cells (Supplementary Fig. S2), indicative of an alternative mechanism. We have previously identified the core centrosomal

L E T T E R S

0:02 1:00 2:10 2:28 3:00

d

∗

∗

∗

∗ ∗ ∗ ∗

∗

∗

∗

∗

∗

Multipolar NEB

NEB

∗

∗

∗

∗ ∗

∗

∗

∗ ∗

∗

∗

∗ ∗

∗

∗

∗ ∗

∗

∗

∗

b

c

Mock Astrin

CLASP1/

2

CLASP + CENP-E RNAi/drug:

CENP -E

Astrin + CENP- E

TOGp + CENP-E Kid TOGp

CLASP + Kid Percentage of multipolar cellsPercentage of multipolar cells

Mock CLASP1/

2

Astrin TOGp

CLA SP1/2

+ TO Gp CLASP1/2 + CE IN DAPI

Centrin–GFP

CENP-E CLASP1

RNAi

a

-tubulin +

Centrin–GFP +α-tub CENP-E CLASP1

ControlCLASP1/2CENPECLASP1/2 + CENPEMock RNAiCLASP1/2 RNAiCLASP1/2 RNAi

0:00 1:00 2:08 2:45 3:10

0:00 1:00 1:50 2:15 2:40

+ CENP-E inhibitor Washout

0 5 10 15 20 25 30 35

10 20 30 40 50 60

RNAi:

0 α

CLASP1/2 + astrin

Figure 4CENP-E-mediated traction forces on misaligned chromosomes are responsible for the irreversible loss of spindle-pole integrity in CLASP1/2-depleted cells. (a) HeLa cells stably expressing centrin–GFP were transfected with the indicated siRNA and stained for α-tubulin (α-tub), CENP-E and CLASP1 72 h later. Most CLASP1/2+CENP-E-depleted cells were bipolar. Note that CENP-E localizes normally at kinetochores of CLASP1/2-depleted cells.

(b) Quantification of spindle multipolarity and the effect of CENP-E depletion/inhibition in different experimental conditions. CE IN, CENP-E inhibitor. (c) Quantification of spindle multipolarity on co-depletion

of CLASPs with astrin or TOGp. Quantitative data inb andcare mean±s.d.of three independent experiments. (d) Live imaging of control (n=6) and CLASP1/2-depleted HeLa cells (n=32) stably co-expressing H2B-histone–GFP/α-tubulin–mRFP following addition of CENP-E inhibitor. Two hours after NEB, the drug was washed out and cells were imaged for another 1 h. Most CLASP1/2-depleted cells became multipolar only after CENP-E inhibitor washout. Addition of the drug did not rescue bipolarity in cells that were already multipolar at the time of drug addition. The asterisks indicate the position and number of spindle poles. Scale bars, 5µm. Time, h:min.

proteins CPAP and ninein as CLASP1-interacting proteins during mitosis¹¹. CPAP and ninein have well-established roles in centriole duplication and microtubule anchorage/nucleation at the centrosome during interphase, respectively^30,31, but their functional relationship with CLASPs at mitotic centrosomes was unclear. We found CPAP centrosomal localization unaffected in CLASP1/2-depleted cells (Supplementary Fig. S2), indicating that CLASPs may work downstream of CPAP to ensure spindle-pole integrity. On the other hand, we observed a strong correlation between CLASP1/2 and ninein levels at mitotic centrosomes, with multipolar spindles showing the

lowest accumulation of ninein on CLASP1/2RNAi (Fig. 5a,b and Supplementary Fig. S6a,b). In agreement, even a partial depletion of ninein by RNAi (Supplementary Fig. S1g) caused an increase in the percentage of cells with multipolar spindles, which showed a high number of poles with zero or one centriole (17.4±6.7%

and 29.8±12.2%, respectively; Fig. 5c,d). Moreover, we found that CLASP1/2 localization at centrosomes is unaffected by ninein depletion, indicating that ninein may act downstream of CLASPs (Supplementary Fig. S6c). The observation that CLASP1/2 and ninein co-depletion did not exacerbate spindle multipolarity relative to

L E T T E R S

c

DAPI Centrin–GFP

1 2

3

4 5

6

7 8

Ninein RNAi

CLASP1/2 RNAi

Ninein signal intensity

b

Control multipolar

e

d Control bipolar

1

2

3

4

6 5

Percentage of multipolar cells RNAi:

a

f

g

n + 1 DAPIα-tubulin Centrin–GFP Ninein

CLASP1/2 signal intensity

α-tubulin

1 2 3 4 5 6 7 8

Percentage of poles with n centrioles

CLASPs Ninein

CENP-E Kid Eg5 CLASPs Ninein

n – 1 30

100

30 100

5 poles 4 poles 3 poles

2 poles 1020

30 40 50 60 70

0

0 1 2 3 4

0 5 10 15 20

25 DAPI Ninein PCM-1 Ninein + PCM-1

Ninein CLASP1/2 RNAi:

Mock Ninein

CLA SP1/

2

Ninein + CLASP1/2Ninein + CEN P-E

Figure 5CLASPs ensure spindle-pole integrity through a functional relationship with ninein. (a) Centrin–GFP HeLa cells were mock or CLASP1/2-RNAi transfected and immunostained forα-tubulin, ninein and DNA. Insets are magnifications of the ninein signal at the indicated poles. In control cells, ninein localizes at mitotic centrosomes (inset 1).

InCLASP1/2-RNAi cells with multipolar spindles, ninein is extensively delocalized from the centrosomes (insets 4 and 5). Note that ninein localizes normally in the few control cells that sporadically form multipolar spindles (inset 2), as well as in cells inefficiently depleted by CLASP1/2RNAi (inset 6).CLASP1/2RNAi does not interfere with ninein localization in interphase centrosomes (inset 3). (b) Graph representing the correlation between CLASP1/2 signal intensity at the poles, ninein signal intensity and number of poles per cell. Multipolar spindles are mostly associated with low levels of both proteins. (c) Ninein RNAi increases the frequency of multipolar spindles. Insets are magnifications of the centrioles in the indicated poles. (d) Similarly toCLASP1/2RNAi, ninein RNAi results in multipolar spindles with a high percentage of poles with 0 and 1 centrioles. (e) The multipolarity phenotype is not

exacerbated following double depletion of CLASPs and ninein, and multipolarity is reduced to control levels in ninein+CENPE RNAi.

Quantitative data indandeare means±s.d.of three independent experiments. (f) Ninein co-localizes with the pericentriolar satellite marker PCM-1 at mitotic spindle poles. Cells were double stained with mouse anti-ninein and rabbit anti-PCM-1 antibodies. Insets show a higher magnification of the spindle-pole region, and those in the last panel at the right highlight co-localization (white mask). Scale bars, 5µm. (g) Proposed model for the role of CLASPs at spindle poles in response to traction forces during chromosome alignment to the metaphase plate. CLASPs provide a scaffolding platform at spindle poles by recruiting ninein to residual pericentriolar satellites. Perturbation of CLASP1/2 function causes bipolar spindles to lose spindle-pole integrity in response to CENP-E- and Kid-mediated kinetochore traction forces during congression of mono-oriented chromosomes, leading to multipolar spindle formation independent of centrosome amplification. This spindle conformation is not transient and cells normally enter anaphase, thus leading to aneuploidy.

L E T T E R S

CLASP1/2 depletion and that multipolarity in ninein RNAi was extensively reverted by co-depletion with CENP-E (Fig. 5e) further supports this conclusion. Finally, we mapped ninein localization on mitotic centrosomes and found that it extensively co-localizes (Pearson correlation=0.89) with the pericentriolar satellite marker PCM-1 (Fig. 5f; ref. 32). Overall, these data support that CLASPs ensure spindle-pole integrity by recruiting ninein to residual pericentriolar satellites during mitosis.

Molecular perturbations that lead to loss of spindle-pole integrity are normally associated with misaligned chromosomes and a significant mitotic delay2,3,7,15,29,33–36. The findings reported in this work assign to multiple forces required for normal chromosome alignment an active role as the cause for the loss of spindle-pole integrity on certain pole perturbations (Fig. 5g). Interestingly, a similar force balance between CENP-E and spindle poles has recently been reported on disruption of the Ndc80 complex³⁷, but it remained unclear how this perturbation sensitizes spindle poles. Our data indicate that, contrary to loss of CLASP1/2 function at spindle poles, this is probably due to a prolonged mitotic delay associated with cohesion fatigue. Furthermore, our work implicates CLASPs in the recruitment of ninein to residual pericentriolar satellites during mitosis. Pericentriolar satellites are electron-dense granules at the centrosome periphery involved in the recruitment of centrosomal proteins, including ninein, during interphase^32,38, but their roles in mitosis remained mysterious. Recently, other pericentriolar satellite components were shown to be required for spindle-pole integrity^34,35. CEP90 prevents the fragmentation of pericentriolar material in response to ‘forces during prometaphase’³⁴ and CEP72 interacts with the Plk1 target Kizuna, which ensures spindle-pole integrity in part due to Kid-mediated forces³⁶. Thus, a role for pericentriolar satellites during mitosis is emerging as a structural mechanism of spindle-pole resistance to CENP-E- and Kid-mediated forces exerted during chromosome alignment.

Finally, we demonstrate that mammalian CLASPs contribute to mitotic fidelity not only by regulating kinetochore–microtubule attachments^9,12,14, but also by preventing irreversible spindle multi- polarity through distinct molecular partnerships at kinetochores and centrosomes. For instance, incomplete CLASP1/2 depletion by RNAi, which does not remove a stable centrosome-associated CLASP1/2 pool^11,12, or injection of anti-CLASP1 antibodies⁹, leads mostly to short bipolar or monopolar spindles, which might be explained by preferential loss of CLASP1/2 function at kinetochores. Importantly, the irreversible nature of spindle multipolarity arising through the mechanism reported here is in marked contrast with the transient multipolar configuration due to supernumerary centrosomes^5,6, un- covering an alternative route to aneuploidy. On the other hand, as multipolar anaphase often results in daughter-cell lethality in several tumour-derived cell lines⁵, CLASPs may represent potential chemotherapeutic targets in human cancers.

METHODS

Methods and any associated references are available in the online version of the paper at http://www.nature.com/naturecellbiology Note: Supplementary Information is available on the Nature Cell Biology website

ACKNOWLEDGEMENTS

The authors would like to thank J. Macedo and P. Sampaio for technical help, A. Pereira for expertise in statistical analysis and all colleagues that

provided invaluable reagents. E.L. is supported by Programa Ciência funded by Programa Operacional Potencial Humano (POPH)/QREN, as well as grant PTDC/SAU-OBD/100261/2008 from Fundação para a Ciência e a Tecnologia of Portugal (COMPETE-FEDER). S.M. held a fellowship from the Fundação para a Ciência e a Tecnologia (FCT) of Portugal (SFRH/BPD/26780/2006). P.M. holds an SNF-Professorship and a EURYI award. Work in the laboratory of H.M. is financially supported by grants PTDC/SAU-GMG/099704/2008 and PTDC/SAU- ONC/112917/2009 from Fundação para a Ciência e a Tecnologia of Portugal (COMPETE-FEDER), the Human Frontier Research Program and the seventh framework programme grant PRECISE from the European Research Council.

AUTHOR CONTRIBUTIONS

E.L. designed, carried out and analysed most experiments. S.M. and A.M. carried out initial phenotypic characterization ofCLASP1/2RNAi. M.B. carried out flux measurements. A.T. and P.M. provided the EGFP–centrin-2/α-tubulin–mRFP stable HeLa cell line. H.M. designed experiments, analysed the data, coordinated the work and wrote the manuscript with contributions from E.L. and S.M.

COMPETING FINANCIAL INTERESTS

The authors declare no competing financial interests.

Published online at http://www.nature.com/naturecellbiology

Reprints and permissions information is available online at http://www.nature.com/

reprints

1. Brenner, S., Branch, A., Meredith, S. & Berns, M. W. The absence of centrioles from spindle poles of rat kangaroo (PtK2) cells undergoing meiotic-like reduction division in vitro.J. Cell Biol.72,368–379 (1977).

2. Gergely, F., Draviam, V. M. & Raff, J. W. The ch-TOG/XMAP215 protein is essential for spindle pole organization in human somatic cells. Genes Dev. 17, 336–341 (2003).

3. Cassimeris, L. & Morabito, J. TOGp, the human homolog of XMAP215/Dis1, is required for centrosome integrity, spindle pole organization, and bipolar spindle assembly.Mol. Biol. Cell15,1580–1590 (2004).

4. Keryer, G., Ris, H. & Borisy, G. G. Centriole distribution during tripolar mitosis in Chinese hamster ovary cells.J. Cell Biol.98,2222–2229 (1984).

5. Ganem, N. J., Godinho, S. A. & Pellman, D. A mechanism linking extra centrosomes to chromosomal instability.Nature460,278–282 (2009).

6. Silkworth, W. T., Nardi, I. K., Scholl, L. M. & Cimini, D. Multipolar spindle pole coalescence is a major source of kinetochore mis-attachment and chromosome mis-segregation in cancer cells.PLoS ONE4,e6564 (2009).

7. Gordon, M. B., Howard, L. & Compton, D. A. Chromosome movement in mitosis requires microtubule anchorage at spindle poles. J. Cell Biol. 152, 425–434 (2001).

8. Manning, A. L. & Compton, D. A. Mechanisms of spindle-pole organization are influenced by kinetochore activity in mammalian cells. Curr. Biol. 17, 260–265 (2007).

9. Maiato, H.et al. Human CLASP1 is an outer kinetochore component that regulates spindle microtubule dynamics.Cell113,891–904 (2003).

10. Maiato, H., Khodjakov, A. & Rieder, C. L. Drosophila CLASP is required for the incorporation of microtubule subunits into fluxing kinetochore fibres.Nat. Cell Biol.

7,42–47 (2005).

11. Maffini, S. et al. Motor-independent targeting of CLASPs to kinetochores by CENP-E promotes microtubule turnover and poleward flux. Curr. Biol. 19, 1566–1572 (2009).

12. Pereira, A. L. et al. Mammalian CLASP1 and CLASP2 cooperate to ensure mitotic fidelity by regulating spindle and kinetochore function.Mol. Biol. Cell17, 4526–4542 (2006).

13. Mimori-Kiyosue, Y.et al. Mammalian CLASPs are required for mitotic spindle organization and kinetochore alignment.Genes Cells11,845–857 (2006).

14. Manning, A. L.et al. CLASP1, astrin and Kif2b form a molecular switch that regulates kinetochore–microtubule dynamics to promote mitotic progression and fidelity.EMBO J.29,3531–3543 (2010).

15. Thein, K. H., Kleylein-Sohn, J., Nigg, E. A. & Gruneberg, U. Astrin is required for the maintenance of sister chromatid cohesion and centrosome integrity.J. Cell Biol.

178,345–354 (2007).

16. McGuinness, B. E., Hirota, T., Kudo, N. R., Peters, J. M. & Nasmyth, K. Shugoshin prevents dissociation of cohesin from centromeres during mitosis in vertebrate cells.

PLoS Biol.3,e86 (2005).

17. Daum, J. R.et al. Cohesion fatigue induces chromatid separation in cells delayed at metaphase.Curr. Biol.21,1018–1024 (2011).

18. Stevens, D., Gassmann, R., Oegema, K. & Desai, A. Uncoordinated loss of chromatid cohesion is a common outcome of extended metaphase arrest. PLoS ONE 6, e22969 (2011).

19. Kwiatkowski, N.et al. Small-molecule kinase inhibitors provide insight into Mps1 cell cycle function.Nat. Chem. Biol.6,359–368 (2010).

20. Magidson, V.et al. The spatial arrangement of chromosomes during prometaphase facilitates spindle assembly.Cell146,555–567 (2011).

L E T T E R S

21. Cimini, D., Wan, X., Hirel, C. B. & Salmon, E. D. Aurora kinase promotes turnover of kinetochore microtubules to reduce chromosome segregation errors.Curr. Biol.16, 1711–1718 (2006).

22. Levesque, A. A. & Compton, D. A. The chromokinesin Kid is necessary for chromosome arm orientation and oscillation, but not congression, on mitotic spindles.J. Cell Biol.154,1135–1146 (2001).

23. Levesque, A. A., Howard, L., Gordon, M. B. & Compton, D. A. A functional relationship between NuMA and kid is involved in both spindle organization and chromosome alignment in vertebrate cells.Mol. Biol. Cell14,3541–3552 (2003).

24. Kapoor, T. M.et al. Chromosomes can congress to the metaphase plate before biorientation.Science311,388–391 (2006).

25. Cai, S., O’Connell, C. B., Khodjakov, A. & Walczak, C. E. Chromosome congression in the absence of kinetochore fibres.Nat. Cell Biol.11,832–838 (2009).

26. Sardar, H. S., Luczak, V. G., Lopez, M. M., Lister, B. C. & Gilbert, S. P. Mitotic kinesin CENP-E promotes microtubule plus-end elongation. Curr. Biol. 20, 1648–1653 (2010).

27. Wood, K. W. et al. Antitumor activity of an allosteric inhibitor of centromere- associated protein-E.Proc. Natl Acad. Sci. USA107,5839–5844 (2010).

28. Compton, D. A. Focusing on spindle poles.J. Cell Sci.111,1477–1481 (1998).

29. Garrett, S., Auer, K., Compton, D. A. & Kapoor, T. M. hTPX2 is required for normal spindle morphology and centrosome integrity during vertebrate cell division.Curr.

Biol.12,2055–2059 (2002).

30. Bornens, M. Centrosome composition and microtubule anchoring mechanisms.Curr.

Opin. Cell Biol.14,25–34 (2002).

31. Bettencourt-Dias, M. & Glover, D. M. Centrosome biogenesis and function:

Centrosomics brings new understanding.Nat. Rev.8,451–463 (2007).

32. Dammermann, A. & Merdes, A. Assembly of centrosomal proteins and microtubule organization depends on PCM-1.J. Cell Biol.159,255–266 (2002).

33. Gruber, J., Harborth, J., Schnabel, J., Weber, K. & Hatzfeld, M. The mitotic-spindle- associated protein astrin is essential for progression through mitosis.J. Cell Sci.115, 4053–4059 (2002).

34. Kim, K. & Rhee, K. The pericentriolar satellite protein CEP90 is crucial for integrity of the mitotic spindle pole.J. Cell Sci.124,338–347 (2011).

35. Oshimori, N., Li, X., Ohsugi, M. & Yamamoto, T. Cep72 regulates the localization of key centrosomal proteins and proper bipolar spindle formation.EMBO J.28, 2066–2076 (2009).

36. Oshimori, N., Ohsugi, M. & Yamamoto, T. The Plk1 target Kizuna stabilizes mitotic centrosomes to ensure spindle bipolarity. Nat. Cell Biol. 8, 1095–1101 (2006).

37. Mattiuzzo, M.et al. Abnormal kinetochore-generated pulling forces from expressing a N-terminally modified Hec1.PLoS ONE6,e16307 (2011).

38. Kubo, A., Sasaki, H., Yuba-Kubo, A., Tsukita, S. & Shiina, N. Centriolar satellites: Molecular characterization, ATP-dependent movement toward centrioles and possible involvement in ciliogenesis.J. Cell Biol.147,969–980 (1999).

M E T H O D S

METHODS

Cell culture, RNAi and western blot analysis.HeLa cell lines stably expressing centrin–GFP (provided by A. Khodjakov, Wadsworth Center, Albany, USA), EGFP–centrin-2/α-tubulin–mRFP and H2B-histone–GFP/α-tubulin–mRFP were grown in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Invitrogen) and selective antibiotics (G418 and puromycin) at 37^◦C in a humidified atmosphere with 5% CO2. For RNAi, we used previously validated siRNA oligonucleotides against human CLASP1 and CLASP2 (ref. 13), astrin³³, TOGp (ref. 3), Sgo1 (ref. 16), NUF2 (ref. 8), ninein³², CENP-E (ref. 11), KID (ref. 39) and separase¹⁵. Cells were transfected 1 h after plating with Lipofectamine RNAiMAX–siRNA complexes according to the manufacturer’s instructions (Invitrogen). Phenotypes were analysed and quantified 36 h or 72 h later depending on RNAi depletion efficiency as monitored by western blotting using the following antibodies: rat anti-CLASP1 1:100 and rat anti-CLASP2 1:100 (ref. 11), mouse anti-astrin 1:1,000 (provided by M. S. Chang, Mackay Memorial Hospital, Taiwan), rabbit anti-TOGp 1:2,000 (provided by L. Cassimeris, Lehigh University, PA, USA; ref. 3), mouse anti-Hec1 1:1,000 (clone 9G3, Abcam), rabbit anti-ninein 1:2,000 (provided by Y-R. Hong, Kaohsjung Medical University, Taiwan), sheep anti-CENP-E 1:500 (provided by W. C. Earnshaw, Wellcome Trust Centre for Cell Biology, Edinburgh, UK) and mouse anti-α-tubulin 1:3,000 (clone B 5-1-2, Sigma). HRP-conjugated secondary antibodies (Amersham) were visualized using the ECL system (Pierce). A GS-800 calibrated densitometer (BioRad) was used for quantitative analysis of protein levels on RNAi.

Drug treatments.HeLa centrin–GFP cell cultures were incubated for 24 h in media containing 2µM cytochalasin D (Sigma-Aldrich) to induce cytokinesis failure. To inhibit the proteasome and induce a metaphase arrest, cells were treated with 5µM MG132 (Calbiochem). For Eg5 inhibition, 5µM STLC (Sigma-Aldrich) was added to the media and cells were either immediately filmed or fixed after 1 h. STLC washout was carried out by rinsing cells four times with fresh medium. Cells were then filmed for∼2 h (Supplementary Fig. S4 and Movies S4 and S5), or alternatively filmed for∼4 h following the addition of medium containing 2µM ZM447439 (AstraZeneca)+5µM MG132 (Calbiochem; Supplementary Fig. S5b). For mitotic- checkpoint inhibition, cells were treated before mitotic entry with 5µM Mps1-IN-1 (ref. 19). For CENP-E inhibition, 20 nM GSK-923295 (MedChemexpress) was added to the media and cells were either immediately filmed or fixed after 1 h.

GSK-923295 washout was carried out 2 h later by rinsing cells four times with fresh medium.

Immunofluorescence microscopy.HeLa centrin–GFP cells were processed for immunofluorescence microscopy as described previously⁹. Primary antibodies used were: rat anti-CLASP1 1:20 (ref. 11), mouse or rat anti-α-tubulin respectively 1:2,000 and 1:100 (Sigma-Aldrich), rabbit anti-γ-tubulin 1:2,000 (Sigma-Aldrich), rabbit anti-MAD2 1:400 (Bethyl), rabbit anti-CENP-E 1:400 (Santa Cruz), mouse anti-astrin 1:500, rabbit anti-ninein 1:1,000 (provided by Y-R. Hong, Kaohsjung Medical University, Taiwan), rabbit and mouse anti-ninein respectively 1:1,000 and 1:1 (provided by E. Nigg, Biozentrum, University of Basel, Switzerland), rabbit anti-PCM-1 1:400 (provided by A. Merdes, CNRS, Toulouse, France)³², mouse anti-Hec1 1:1,000 (9G3, Abcam), rabbit anti-TOGp 1:200 (ref. 3), rabbit anti-NuMA 1:1,000 (ref. 8), rabbit anti-HSET 1:1,000 (ref. 8), rabbit anti-TPX2 1:1,000 (ref. 8), rabbit anti-Kid 1:500 (ref. 23; all provided by D. A. Compton, Dartmouth Medical School, Hanover, USA) and mouse anti-CPAP 1:300 (provided by T. K. Tang, Institute of Biomedical Sciences, Taipei, Taiwan)²⁸. Secondary antibodies used were Alexa Fluor 488, 568 and 647 diluted 1:1,500 (Invitrogen) and DNA was counterstained with DAPI (1µg ml⁻¹). For chromosome spreads, HeLa cells were treated as described previously¹⁷with 300 ng ml⁻¹ nocodazole for 1 h, and then swollen in a 1:1 mixture of medium and double-distilled H₂O plus nocodazole for 20 min at 37^◦C. Mitotic cells were collected by shake off, suspended and washed by centrifugation in 3:1 methanol/acetone, and finally dropped from a height

of at least 45 cm onto cleaned glass coverslips. For phenotypic analysis, 150–300 cells were quantified in each of three independent experiments, unless otherwise indicated.

Image acquisition and time-lapse microscopy. Image acquisition for immunoflu- orescence microscopy analysis was carried out in a Zeiss AxioImager Z1 equipped with a Axiocam MR. Images were subsequently blind deconvolved with Autoquant X (Media Cybernetics). Adobe Photoshop CS4 (Adobe Systems) was used for image processing. CLASP1/2 and ninein levels at the centrosomes were measured using MATLAB (The Mathworks) by quantification of the pixel grey levels of the focusedzplane within a region of interest. The background signal was measured within a neighbouring region and was subtracted from the measured fluorescence intensity inside the region of interest. Co-localization analysis of ninein and PCM-1 immunostainings (respectively mouse and rabbit antibodies) was carried out using the co-localization plugin package in ImageJ (http://rsbweb.nih.gov/). For time- lapse microscopy, cells were plated in 35 mm glass-bottom dishes (14 mm, No 1.5 coverglass; MatTek Corporation), transfected and/or treated with chemical reagents and imaged in a heated chamber (37^◦C and 5% CO₂). Four-dimensional data sets were collected with a Revolution spinning-disc confocal system (Andor) equipped with an electron-multiplying CCD (charge-coupled device) iXon^EM+camera and a Yokogawa CSU-22 unit based on an Olympus IX81 inverted microscope. Two laser lines at 488 and 561 nm were used for the excitation of GFP and mRFP and the system was driven by IQ software (Andor).zstacks (0.8–1.0µm) covering the entire volume of the mitotic apparatus were collected every 1.5–2.5 min according to the experiments with a PLANAPO 60×NA 1.40 objective. All images represent maximum-intensity projections of allzplanes. ImageJ was used to process all movies.

Images for four-dimensional-reconstruction analysis were deconvolved using the classical maximum-likelihood algorithm of Huygens Professional 4.0 (Scientific Volume Imaging) and then analysed with the open-source software 3D Viewer of Fiji (www.fiji.sc).

Photoconversion and measurement of microtubule flux.For photoconversion assays, U2OS cells stably expressing mEos–tubulin (gift from S. Geley, Innsbruck Medical University, Innsbruck, Austria) were cultivated on glass-bottomed dishes (MatTek). Imaging was carried out using a PLANAPO 100×NA 1.40 DIC objective on a Nikon TE2000U inverted microscope equipped with a Yokogawa CSU-X1 spinning-disc confocal head containing two laser lines (488 nm and 561 nm) and a Mosaic (Andor) photoactivation system also containing two laser lines (405 nm and 488 nm). Photoconversion was carried out in late prometaphase/metaphase cells, identified by green fluorescent tubulin signals, by using two line-like regions of interest, placed perpendicular to the main spindle axis on both sides of the metaphase plate. Activation was carried out by one 500 ms pulse from a 405 nm laser. Photoconverted red signals as well as green fluorescence signals were then followed over using 561 nm and 488 nm lasers and an iXon^EM+electron-multiplying CCD camera every 3 s for 5 min. Microtubule flux (µm min⁻¹) was quantified by tracking photoconverted mEos–tubulin over time using a Matlab algorithm for kymograph generation and analysis⁴⁰. Generated kymographs were collapsed and the fluorescence intensity curve was created. Fluorescence intensity peaks of each time frame were defined by manually tracking regions of the highest fluorescence intensities followed by automatized curve fitting. Distances between measured peaks were then exported into an Excel table and microtubule flux values were calculated both from the distances between two activated regions and from the distances between each region and the mitotic pole of its associated half-spindle.

39. Tokai-Nishizumi, N., Ohsugi, M., Suzuki, E. & Yamamoto, T. The chromokinesin Kid is required for maintenance of proper metaphase spindle size.Mol. Biol. Cell16, 5455–5463 (2005).

40. Pereira, A. J. & Maiato, H. Improved kymography tools and its applications to mitosis.

Methods51,214–219 (2010).