The human kinesin Kif18A is a motile microtubule depolymerase essential for chromosome congression

The Human Kinesin Kif18A

Is a Motile Microtubule Depolymerase Essential for Chromosome Congression

Monika I. Mayr,1 Stefan HOmmer,1 Jenny Bormann,1 Tamara GrOner,1,2 Sarah Adio,3 Guenther Woehlke,3 and Thomas U. Mayer¹,.

1 Chemical Genetics

Independent Research Group Department of Cell Biology

Max-Planck-Institute of Biochemistry Am Klopferspitz 18

82152 Martinsried Germany

2London Research Institute Lincoln's Inn Fields Laboratories 44 Lincoln's Inn Fields

London WC2A 3PX United Kingdom 31nstitute for Cell Biology University of Munich Schillerstr. 42 80336 Munich Germany

Summary

Background: The accurate alignment of chromosomes at the spindle equator is fundamental for the equal distri- bution of the genome in mitosis and thus for the genetic integrity of eukaryotes. Although it is well established that chromosome movements are coupled to micro- tubule dynamics, the underlying mechanism is not well understood.

Results: By combining RNAi-depletion experiments with in vitro biochemical assays, we demonstrate that the human kinesin Kif18A is a motile microtubule depo- lymerase essential for chromosome congression in mammalian tissue culture cells. We show that in vitro Kif18A is a slow plus-end-directed kinesin that pos- sesses microtubule depolymerizing activity. Notably, Kif18A like its yeast ortholog Kip3p depolymerizes longer microtubules more quickly than shorter ones.

In vivo, Kif18A accumulates in mitosis where it localizes close to the plus ends of kinetochore microtubules. The depletion of Kif18A induces aberrantly long mitotic spin- dles and loss of tension across sister kinetochores, resulting in the activation of the Mad2-dependent spin- dle-assembly checkpoint. Live-cell microscopy studies revealed that in Kif18A-depleted cells, chromosomes move at reduced speed and completely fail to align at the spindle equator.

Conclusions: These studies identify Kif18A as a dual- functional kinesin and a key component of chromosome congression in mammalian cells.

'Correspondence: mayer@biochem.mpg.de

Introduction

The function of mitosis is to equally divide the dupli- cated parental genome into two newly forming daughter cells. A key component of this process is the mitotic spindle, a microtubule-based bipolar structure whose dynamic properties are vital for both chromosome align- ment and separation before and upon anaphase onset, respectively [1]. The assembly and function of the mi- totic spindle depends on a complex network regulat- ing-in a temporally and spatially controlled manner- microtubule dynamics and activities of motor proteins of the dynein and kinesin superfamilies, microtubule- based ATPases that convert chemical energy into mechanical force [2-4].

Prerequisite for the equal partitioning of replicated chromosomes is their proper bipolar attachment on the mitotic spindle; i.e., sister kinetochores of chromatid pairs are attached to microtubules emanating from op- posite spindle poles. In higher eukaryotes, the capture of kinetochores by spindle microtubules starts with nuclear-envelope breakdown resulting, commonly, in mono-oriented chromosomes with only one kinetochore attached to microtubules. Subsequently, mono- oriented chromosomes alternate between movements away from and toward the spindle pole until the unat- tached kinetochore is encountered by microtubules from the opposite spindle pole. Bivalent-oriented chro- mosomes congress toward the spindle equator where they oscillate until the sudden loss of cohesion at ana- phase onset allows the separation of sister chromatids to opposite spindle poles.

Although the basic concepts of chromosome con- gression have been recognized for a long time (see [5]

for review), the discovery of the underlying molecular mechanisms has been started only recently. Yet, given that microtubule plus ends remain attached to kineto- chores of congressing and oscillating chromosomes, it is apparent that chromosome movements have to be tightly coupled to microtubule dynamics. The identifica- tion of kinetochores as the major sites of microtubule assembly and disassembly associated with chromo- some movements [6, 7] suggested that the kineto- chore-associated motor proteins CENP-E and MCAKI XKCM1 play a key role in chromosome congression.

However, inactivation of neither CENP-E, a plus-end- directed kinesin [8], nor MCAKlXKCM1, a nonmotile microtubule depolymerase [9, 10], impaired the con- gression of bivalently oriented chromosomes [11, 12].

Members of the Kinesin-8 family of kinesins (Kif18A, H. sapiens; Klp67 A, D. me/anogaster; KipB, A. nidulans;

Kip3p, S. cerevisiae; klp5/6+, S. pombe) [13] have been classified as microtubule depolymerases based on the observation that loss of their activity results in aberrantly long spindles with hyperstable microtubules [14-23]. In the current study, we show that the human kinesin Kif18A, which is enriched at the plus ends of kinetochore microtubules, possesses both plus-end-directed

Konstanzer Online-Publikations-System (KOPS) URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-1401416

A

A'

367 453 898

~:W1!:]·.i,.r4'iii'M*,1

Figure 1. The Human Kinesin Kif18A Peaks in Mitosis

Kif18AAbN Kif18AAbC

(A) Schematic representation of Kif18A indicating the regions used for antibody production.

(6) HeLa cell extracts; in vitro-translated Kif18A^FL and un programmed wheat-germ ex- tract were probed by western blotting with Kif18A antibody (Kif18A^AbN). Anti-a.-tubulin immunoblot serves as a loading control.

B

..J 0

'4:

^0;)

g

^+:>

.9

I~

"... c

....

8

::2 ⁽¹¹

~ ~

^...I

Mr(K) <l>

:r:

9 8 - ••••

6 4 -

<l>

(f)

.c (l)

.$

e-

.5

(l)

...I

<l>

:r:

4: z

0::

e

C

8

4: z

0::

~ 0;)

..-~

~

a-Kif18A

(C) HeLa cells synchronized in S phase were released and analyzed at the indicated time points by western blotting with Kif18A (Kif18A^AbN) and cyclin 61 antibodies. Anti-a;- tubulin immunoblot serves as a loading con- trol. The total cell extract loaded was 15 I-'g.

c

o 6 8 9 q,'? ,,<:> ,,<:>'? ,," '? ,,'l- .:},,:-'o ,,<>;,'9 hr after thymidine ,...-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ -, release

I - - -,....,. -

⁺ ··la-Kif18A

~================~

I·

⁰

--~----

• • . • • • "la-CYCB1

~================~

motility and microtubule depolymerizing activity. Be- cause recently a dual functionality has been demon- strated for the yeast ortholog Kip3p [24, 25], it is sug- gested that this remarkable characteristic applies to all members of the Kinesin-8 family. Our cellular studies re- vealed that chromosome congress ion relies on the pres- ence of Kif18A, suggesting that this motile microtubule depolymerase is key for the dynamics of kinetochore microtubules driving chromosome alignment in the pre- anaphase state of the cell cycle in human cells.

Results and Discussion

Kif18A Is a Cell-Cycle-Regulated Kinesin

For exploring the function and localization of endoge- nous human Kif18A, polyclonal Kif18A antibodies were raised against an N-terminal GST -tagged fragment (res- idues 1-367) and against a peptide composed of amino acids 879-894 (Figure 1A). Western-blot analyses of Hela extracts revealed that the affinity-purified anti- body raised against the N terminus of Kif18A (Kif18AbN) detected a protein of the expected size of approximately 100 kDa (Figure 1 B). The antibody also recognized in vitro-translated (IVT) Kif18A but did not detect a pro- tein in unprogrammed wheat-germ extract (Figure 1 B).

The depletion of endogenous Kif18A with short-inter- fering RNAs (siRNAs) caused the almost complete loss of the immunoreactive signal in Hela extracts,

demonstrating that the detected protein was indeed Kif18A (Figure 1B).ldentical results were obtained with the immunopurified peptide antibody (Kif18Abc) (data not shown).

Immunoblot analyses indicated that Kif18A was more abundant in mitotic than in interphase cells (Figure 1 B).

For investigating regulation of abundance in more detail, extracts of synchronized Hela cells released from a thy- midine arrest were immunoblotted for Kif18A. Figure 1 C shows that the endogenous levels of Kif18A were low in S phase, increased approximately 9 hr after release, and decreased abruptly after approximately 10.5 hr (Fig- ure 1 C). For defining the timing of entry into and exit from mitosis, identical samples were immunoblotted for cyclin B1. As shown in Figure 1 C, endogenous levels of cyclin B1 increased and declined with similar kinetics as Kif18A. Immunoblots against (I.-tubulin confirmed that for each time point equal amounts of proteins were loaded (Figure 1 C). Taken together, these data show that Kif18A accumulates during mitosiS, suggest- ing a mitotic function for this kinesin.

Kif18A localizes Close to Plus Ends of Kinetochore Microtubules

Next, we determined Kif18A's intracellular localization by indirect immunofluorescence with affinity-purified Kif18Abc antibodies. In interphase cells, Kif18A was detected in the nucleus (Figure 2A). As a bipolar spindle

B

Figure 2. Kif18A Localizes near the Plus Ends of Kinetochore Microtubules

(A) Immunofluorescence images of HeLa cells at different cell-cycle stages stained with DAPI (blue) and antibodies directed against Kif18A (red) and anti ,,-tubulin (green). The scale bar represents 10 IIm.

(8) HeLa cells stained for Kif18A (red), the kinetochore protein Hec1 (green), and chromatin (blue). The scale bar represents 1 0 ~lm.

(C) Localization of transiently expressed GFP-tagged full-length Kif18A (GFP- Kif18A^FL, 1-898) during metaphase. HeLa cells were stained for the kinetochore-microtubule-binding protein HURP (red), and Kif18A was visualized by GFP-tag. The scale bars represent 1 0 ~lm.

formed, Kif18A localized along spindle microtubules before it accumulated in metaphase at the tips of micro- tubules pointing toward chromosomes or being con- nected to kinetochores (Figure 2A). Upon anaphase on- set, the Kif18A signal diminished and finally disappeared almost completely as cells left mitosis (Figure 2A), and this is in line with the immunoblot analyses (see Figure 1 C). No signal was detected in HeLa cells depleted of Kif18A by siRNA approaches and thus con- firmed that the observed immunofluorescence signal was specific for Kif18A (Figure 5A). Identical localization patterns were observed with the affinity-purified Kif18^AbN antibody (data not shown). To determine the lo- calization of Kif18A on the metaphase spindle in greater detail, we costained cells with the outer-kinetochore protein Hec1 [26, 27]. As shown in Figure 2B, Kif18A was enriched at the kinetochore microtubules' tips, distal to the outer-kinetochore protein Hec1. Next, we stained HeLa cells transiently expressing green- fluorescent-protein (GFP)-tagged full-length Kif18A (GFP-Kif18A^FL) for Hec1 and HURP, a mitotic spindle protein that localizes to kinetochore microtubules in the vicinity of chromosomes [28, 29]. As shown by triple-immunofluorescence images, Kif18A was en- riched at kinetochore microtubules flanked by Hec1 and HURP on the chromosome-proximal and -distal site, respectively (Figure 2C). Quantification of Kif18A- immunoreactive punctae and the number of kineto- chores per cell-with Hec1 as a marker-revealed that both numbers closely match (average Hec1 and Kif18A punctae per HeLa cell: 62.6 and 58.3 respectively [n

=

10]), further supporting the observation that under these experimental conditions, Kif18A binds exclusively to ki- netochore microtubules. In nocodazole-release experi- ments, we observed that GFP-Kif18A^FL was enriched in the proximity of centrosomes in the presence of noco- dazole before it accumulated upon nocodazole washout at the tips of regrown microtubules in the vicinity of chromosomes (Figure S1 in the Supplemental Data available online). Taken together, these data demon- strate that Kif18A is enriched near the plus ends of kinet- ochore microtubules.

Loss of Kif18A Activates the Mad2-Dependent Spindle-Assembly Checkpoint

Previously performed organism-wide kinesin-siRNA studies have shown that Kif18A seems to be required for chromosome congression in mitotic cells [23]. How- ever, the underlying mechanism leading to chromosome alignment defects in the absence of Kif18A has not been addressed in this study. To gain detailed insights into the mitotic function of Kif18A, we first depleted Kif18A in asynchronous HeLa cells by using two different siRNA oligos. Immunoblot analyses confirmed that both siRNA oligos decreased the endogenous levels of Kif18A to less than 25% compared to cells treated with a luciferase control (GL2) siRNA oligo (Figure 3A). Cell-cycle analy- ses with laser scanning microscopy (LSM) (see Supple- mental Experimental Procedures) revealed that cells de- pleted of Kif18A accumulated in a G2IM state with a 4N DNA content (Figure 3B). This indicates that Kif18A is re- quired for timed progression through mitosis. The ob- served delay in mitosis was mediated by the spindle-as- sembly checkpoint (SAC) because cells depleted of

both Kif18A and Mad2 did not show an accumulation of mitotic cells but a dramatic increase in the number of multinucleated cells (Figure 3C). In line with these re- sults, we observed that Mad2 was enriched at kineto- chores of mal oriented chromosomes in Kif18A-depleted cells (Figure 3D). Notably, in the rare cases when chro- mosomes were properly aligned at the metaphase plate in Kif18A-RNAi cells, Mad2 did not accumulate at these kinetochores (Figure 3D) and thus suggests that Kif18A-in contrast to dynein-dynactin [30]-does not play a role in checkpoint inactivation, i.e., in the deple- tion of checkpoint proteins from correctly attached ki- netochores.

Cells Depleted of Kif18A Fail to Establish a Stable Metaphase Plate

To gain further insight into the defects caused by Kif18A depletion, we performed live-cell analyses of control and Kif18A-depleted HeLa cells expressing GFP-tagged histone-H2B. Whereas control-treated cells efficiently aligned chromosomes at the metaphase plate, Kif18A- RNAi cells had severe problems in chromosome con- gression and displayed a prolonged prometaphase characterized by oscillating chromosomes that failed to form a stable metaphase plate (Figure 4A; Movies S1 and S2). Within 60 min (median: 52 min, n

=

31) after the first signs of chromosome condensation, 65% of control RNAi cells aligned all chromosomes correctly (Figure 4B). In contrast, only approximately 6% of Kif18A-depleted cells managed to do so within the same period of time (median: 170 min, n = 35) (Figure 4B and Movie S2). Next, we examined whether alterations in the velocity of chromosomes in Kif18A-depleted cells could account for the observed chromosome alignment defects. To this end, we quantified the rate of chromo- some movements in GL2- and Kif18A-RNAi cells by us- ing a yellow fluorescent-protein (yFP)-tagged form of the centromeric protein CENP-A to visualize individual chromosomes. These analyses revealed that in the pe- riod between nuclear-envelope breakdown and the end of metaphase (PM-M, prometa-, metaphase), chro- mosomes moved away from and toward the spindle equator in GL2-RNAi cells at an average rate of approx- imately 2l-tm/min (Figure 4C, left, and Movie S3), a veloc- ity consistent with previous reports [31, 32]. Upon anaphase onset, chromosomes started to move pole- wards at 2.5 I-tm/min (EA, early anaphase) before they slowed down to 1.1 I-tm/min during late anaphase (LA in the left panel of Figure 4C). In contrast, chromosomes in Kif18A-depleted cells displayed sustained but slow movements (betwe.en 0.8 and 1.3 I-tm/min and Movie S4) while trying to align properly at the metaphase spin- dle (Figure 4C, right). Because ofthe prolonged prometa- phase arrest of Kif18A-depleted cells (Figure 3B), no ana- phase movements of chromosomes could be recorded during the course of the experiments. Taken together, these data indicate that Kif18A is essential for dynamic movements of chromosomes during prometaphase and their proper alignment at the metaphase plate.

Depletion of Kif18A Induces Elongated Spindles and Decreases Tension across Sister Kinetochores The observed defects in chromosome congression upon Kif18A depletion prompted us to examine more

A B Kif18ARNAi

---

^{GL2 RNAi 31S 416} G1/S (66.4%)

«

^2S2

0::>

m 10 ~lg 5 ~g 2,5119 #1 #2

r- IM

....

2

_I

<:)

249

1~·*;;'4

1~

..0 ii!l3

::::l

... _I

<:)

dna Integral 10000000

~o

=

GL2 RNAi Kif18ARNAi

c

100 80

.!!2

60

multinucleated

(i3 ⁽⁾ • mononucleated

....

0 • mitotic

'#. 40 20

RNAi: GL2IGL2 GL2/Mad2 GL2IKif18A Mad2lKif18A

Figure 3. Loss of Kif18A Activates the Mad2-Dependent Spindle-Assembly Checkpoint

(A) HeLa cells treated with GL2 control or Kif18A-RNAi oligos were probed by western blotting with Kif18A (Kif18A^AbN) antibody. For Kif18A-RNAi cells, 10 j.lg of total protein was loaded. An anti-a-tubulin blot serves as a loading control.

(B) Cell-cycle analyses of control (GL2) or Kif18A-depleted cells.

A 8

C

⁴⁰⁰

'E

₃₅₀

...- Q) ^(/) 300

c 0

<I> 250

(/) m .c 200

0-m

c 150 m C ¹⁰⁰

:::l

<I>

~

⁵⁰

RNAi: GL2 Kif18A

c

3.0

C 2.5

:§

E 2.0

:::l.

PM-like

GL2 RNAi Kif18ARNAi

Figure 4. Kif18A Is Required for Timed Progression through Mitosis and Chromosome Congression

(A) Selected live-cell images of HeLa cells expressing GFP-tagged Histon-H2B transfected with control or Kif18A siRNA oligos. The scale bar represents 10 !!m.

(B) Quantitative analyses of the live-cell images shown in (A) documenting the time required for anaphase onset in control or Kif18A-depleted HeLa cells (n = 30). The first signs of chromosome condensation is defined at t = 0 min.

(C) Quantification of chromosome movements in GL2-treated (left panel) and Kif18A-depleted (right panel) HeLa cells stably expressing YFP-CENP-A during different cell-cycle stages (I, interphase; PM-M, prometa-, metaphase; EA, early anaphase; LA, late anaphase; C, cyto- kinesis). (Regarding GL2-RNAi cells, each bar represents average chromosome velocity at the indicated cell-cycle stages calculated by the anal- yses of a total of 116 individual CENP-A dots in five different cells; for Kif18A-RNAi cells, each bar represents average chromosome velocity in one Kif18A-RNAi cell. In total, 108 individual CENP-A dots in five different cells were analyzed.) Values represent the average ± SE.

closely the spindle morphology in these cells. As shown in Figure 5A, HeLa cells depleted of Kif18A below the detection limit frequently formed elongated "banana- shaped" spindles with wavy and bent microtubules, in- dicating that these microtubules are abnormally long.

Quantitative analyses with pericentrin as a marker for spindle poles demonstrated that the average distance between the two spindle poles in Kif18A-RNAi cells (n = 10) was approximately 18.1 I1m (±1.5) compared to 9.3 I1m (±1.1) in GL2-RNAi cells, (n

=

10, Figure 58

and Figure 84). To rule out the possibility that the in- crease in spindle length in Kif18A-RNAi cells was caused by the prolonged arrest in prometaphase, we determined spindle length in cells depleted of both Mad2 and Kif18A. Although the codepletion of Mad2 eliminated the mitotic delay caused by Kif18A depletion (Figure 3e), spindle length in Kif18-/Mad2-RNAi cells (n = 10) was still significantly increased to 16.0 I1m (±1.0) compared to GL2-/Mad2-RNAi control cells (Figure 58 and Figure 84). Thus, abnormally long spindle

(C) Quantitative cell-cycle analyses of HeLa cells 48 hr after transfection with GL2IGL2, GL2IMad2, GL2IKif18A, or Kif18A1Mad2 siRNA-oligos.

For each combination of double siRNA, 100 cells were counted.

(D) The checkpoint protein Mad2 is enriched at kinetochores of maloriented chromosomes but not at those of properly aligned chromosomes in GL2- and Kif18A-siRNA HeLa cells. Cells were transfected with GL2 control or Kif18A-RNAi oligos for 48 hr and then stained for Mad2 (red), Hec1 (blue), and a-tubulin (green). The scale bar represents 1 0 ~lm.

A B

25

E

²⁰

.E:

.r:

+' c ^Cl 15

~ (j) 10 '6 c 'Q.

r/)

5

RNAi: GL21 GL21 Mad21 Mad2 Kif18A Kif18A

c

a-tubulin

o

E

2,1

E

1,9 ,2;

~ ^1,7

•

c

I

1J!. 1,5 '6 ^{~ 0 1,3}

I

I

""

^{0 1,1}

•

.s

.~

•

-t: 0,9

•

2 _0,7 .S

0,5

RNAi: Gl2 GL2 + Noc Kif18A

Figure 5. Kif18A Depletion Affects the Integrity of the Mitotic Spindle

(A) Hela cells transfected with GL2- or Kif18A-RNAi oligos were stained with Kif18A (Kif18AAbc, red) and et-tubulin (green) antibodies and DAPI (blue). The scale bar represents 1

°

^Ilm.

(B) Quantification of pole-to-pole distance in Gl2lMad2, Gl2lKif18A, or Kif18A1Mad2-depleted mitotic cells (n

=

10 for each condition). Values represent the average ± SE.

(C) Spindle microtubules are hyperstable in Kif18A-depleted cells. Hela cells were treated for 48 hr with control (GL2-) or Kif18A-siRNA, and this was followed by a 35 min incubation on ice. Cells were stained for et-tubulin. The scale bar represents 1

°

^I'm.

(D) Depletion of Kif18A decreases tension on sister kinetochore pairs. Representative immunofluorescence images of Hela cells 48 hr after transfection with Gl2- or Kif18A-siRNA are shown. Tubulin and DNA are shown in green and blue, respectively. Kinetochore pairs on sister chro- matids were observed by CREST signal (red) at either side of the sister chromatid. The scale bar represents 15 l'ffi.

(E) Quantification of interkinetochore distances. In total, the interkinetochore distance of 113 (Gl2-RNAi cells, n = 10), 78 (Kif18A-RNAi cells, n

=

10), and 73 (GL2-RNAi cells + 5 I'M nocodazole for 6 hr, n

=

10) bioriented chromatid pairs were measured.

A

B OnM

c

Kif18AFl

1 0 0 - 9 0 - 8 0 _ 7 0 -

6 0 - 5 0 -

E

16 14

E

¹²

3 ₁₀

~ c: 8 .!Q 6

...

::'l: 4 2 0

__ +_2~.~5~m~M~ ____ +~5~m~M~ _____ +_7~.~5~m~M~AMP_PNP SOnM

OnM

s P

100nM

S P

100nM

2.SmM 100nM

S P

o

min 5 min 10 min 0 min 5 min 10 min + buffer + Kif18AFl

long GMP-CPP MT

100 nM 100 nM 100 nM Kif18AFl

AMP-PNP 5mM 100nM

S P

7.5 mM 100nM

S P

7 6

E

3 ⁵

:E 4

i3 ...

::'l: 2

o

o

low salt

100nM Kif18A^Fl

- Kif18A^Fl

- tubulin

+ buffer + Kif18AFl short GMP-CPP MT

Figure 6. Kif18A Is a Plus-End-Directed Microtubule Depolymerase

(A) Immunofluorescence images of polarity-marked microtubules moving in the presence of His-Kif18A^FL and ATP with the brightly labeled plus- end trailing. The scale bar represents 10 flm.

(B) GMP-CPP microtubules (final tubulin concentration: 0.35 flM) in BRB80 buffer supplemented with KCI (final CI- concentration: 110 mM) were incubated with buffer or two different concentrations of insect-derived full-length Kif18A protein in the presence of ATP (at a final concentration of 1 mM). Where indicated, the reactions were supplemented with various concentrations of AMP-PNP, a nonhydrolyzable ATP analog. After 25 min at RT, samples were fixed and analyzed by fluorescence microscopy. The scale bar represents 1 0 ~,m.

microtubules and elongated spindle shapes are direct consequences of Kif18A depletion, suggesting that Kif18A negatively affects length of spindle microtubules.

Consistently, in the absence of Kif18A, spindle microtu- bules were more resistant to cold-induced depolymer- ization than those in GL2-RNAi control cells (Figure 5C).

These data suggest that Kif18A is required for the dy- namic behavior of kinetochore microtubules. Changes in the dynamics of kinetochore microtubules affect tension acting on sister kinetochores [33], and this can be analyzed by measurement of the distance between sister kinetochores. As shown in Figures 5D and 5E, loss of Kif18A decreased the interkinetochore distance of bioriented chromosomes from 1.55 Ilm (±0.17) (113 kinetochore pairs from ten different GL2-RNAi cells were analyzed) to 1.31 Ilm (±0.17) (78 kinetochore pairs from ten different Kif18A-RNAi cells were analyzed). In the absence of microtubules, the distance between kinetochore pairs was reduced to 0.86 Ilm (±0.12) (73 kinetochore pairs from ten different nocodazole-treated GL2-RNAi cells were analyzed), indicating that in the ab- sence of Kif18A, kinetochores are still attached to micro- tubules albeit under reduced tension. Together, all data obtained so far indicate that Kif18A has a destabilizing effect on kinetochore microtubules, and this function seems to be vital for tension acting on sister kineto- chores (Figure 5E) and thus for proper chromosome congression during prometaphase (Figure 4A).

Kif18A Is a Plus-End-Directed Microtubule Depolymerase

The phenotype of Kif18A-depleted cells prompted us to investigate the enzymatic properties of Kif18A in vitro.

First, we tested whether Kif18A possesses motility ac- tivity by using in vitro microtubule-gliding assays. In the presence of recombinant full-length Kif18A (His- Kif18A^FL) and ATP, we observed motility offluorescently labeled microtubules at an average speed of approxi- mately 0.02 (±0.007) Ilm/s (Movie 85). To determine the directionality of Kif18A, we analyzed the movement of polarity-marked microtubules. As shown in Figure 6A and Movie 85, microtubules moved with the brightly la- beled plus-end trailing (n = 20) indicative of a plus-end- directed motor activity. Thus, these studies identify Kif18A similarly to its orthologs Klp67A and Kip3p from Drosophila and budding yeast [25, 34J, respectively, as a slow plus-end-directed kinesin. Next, it was investi- gated whether Kif18A possesses microtubule-destabi- lizing activity in vitro. For these experiments, full-length His-tagged Kif18A (His-Kif18A^FL) expressed in insect cells and purified to homogeneity was incubated with fluorescently labeled microtubule polymer that were stabilized by polymerization in the presence of the

slowly hydrolyzable GTP analog GMPCPP (guanylyl- (rx,13)-methylene-diphosphonate) [35]. In the presence of ATP and recombinant His-Kif18A^FL, rhodamine- labeled GMP-CPP-microtubules (resuspended to a 0.35 IlM tubulin-dimer concentration) de polymerized within 25 min at RT (Figure 6B). This effect was dose de- pendent on His-Kif18A^FL and specific as indicated by the fact that identical buffer conditions did not affect microtubule polymers (Figure 6B). Previously, it was shown that the nonmotile Kinesin-13 kinesins are capa- ble of de polymerizing microtubules in an ATP-depen- dent manner [10, 36, 37]. To address whether Kif18A- like the Kinesin-13 proteins-requires ATP hydrolysis for its microtubule depolymerization activity, the effect of Kif18A on microtubule polymers was studied in the presence of the nonhydrolyzable ATP analog AMP- PNP (13,y-imidoadenosine 5'-triphosphate). As shown in Figure 6B, 2.5 mM AMP-PNP did not significantly affect His-Kif18A^FL-dependent microtubule destabiliza- tion, whereas 5 mM attenuated and 7.5 mM completely blocked the depolymerization of microtubules. In these assays, ATP was present at a concentration of 1 mM because His-Kif18A^FL was purified and stored in ATP- containing buffer for ensuring protein stability. Thus, these data suggest that Kif18A's destabilizing activity is sensitive to high concentrations ofthe ATP competitor AMP-PNP. The fact that identical results were previ- ously reported for Kip3p, whose microtubule-destabiliz- ing activity was attenuated but not blocked in the pres- ence of 2 mM AMP-PNP [24], suggests that Kinesin-8 proteins share common mechanochemical properties.

To corroborate our findings, we incubated GMP-CPP microtubules (resuspended to a 0.35 IlM tubulin-dimer concentration) with His-Kif18A^FL and 1 mM ATP and then ultracentrifugated them to separate soluble tubulin dimer from pelletable microtubule polymer. When GMP- CPP microtubule polymer was incubated with 100 nM His-Kif18A^FL, almost all tubulin was recovered from the supernatant fraction, whereas the majority of tubulin was recovered from the pellet fraction in the absence of Kif18A (Figure 6C). In line with the microscopic analy- ses, AMP-PNP suppressed in a dose-dependent man- ner the destabilizing activity of 100 nM His-Kif18A^FL in the presence of 1 mM ATP (Figure 6C). Under lower- salt condition, 100 nM His-Kif18^FL did not efficiently depolymerize GMPCPP microtubules in the presence of 1 mM ATP (Figure 6D, 60mM final CI- concentration compared to 110 mM in Figure 6C). Thus, the micro- tubule depolymerization activity of Kif18A is highly sensitive to lower-salt conditions.

Recently, Kip3p was identified as a length-dependent depolymerase that de polymerizes long microtubules more efficiently than short ones [25]. As further noted

(C) Microtubule pelleting assay. Samples treated as in (A) (control buffer or 100 nM His-Kif18A^FL, 0.35 JlM tubulin as polymer, 1 mM ATP, 110 mM CI-, and the indicated concentrations of AMP-PNP) were not fixed but pelleted by ultracentrifugation. Supernatant and pellet fractions were analyzed by SDS-PAGE. Asterisks mark His-Kif18A^FL•

(D) Kif18A does not depolymerize GMP-CPP microtubules under low-salt conditions. GMP-CPP microtubules (final tubulin concentration of 0.35 I'M) in BRB80 buffer were incubated with 100 nM HiS-Kif18A^FL (final CI- concentration: 60 mM), and 1 mM ATP were incubated for 25 min at RT, pelleted by ultracentrifugation and analyzed by SDS-PAGE.

(E) Kif18A depolymerizes longer GMP-CPP microtubules more efficiently than shorter ones. Short or long fluorescently labeled GMP-CPP microtubules in the amount of 0.35 I'M were incubated with 100 nM His-Kif18A^FL and 1 mM ATP for the indicated time points at RT. Microtubule length was determined by fluorescence microscopy. For each bar, the lengths of 20 microtubules were measured. Values represent the aver- age ± SE.

in this study, longer microtubules accumulated more Kip3p at their plus ends than shorter ones, a finding that seems to be critical for the rate of microtubule depolymerization. In agreement with these results, we observed that in the presence of 1 mM ATP, 100 nM His-Kif1SA^FL depolymerized long (14.9 ± 1.44 llm) GMPCPP microtubules at a rate of 1.25 ± 0.14 llm/min, whereas short ones (5.7 ± 0.4S11m) were depolymerized at a rate of 0.21 ± O.OS llm/min. As expected, this effect was specific because in the absence of Kif1SA, long and short microtubules did not significantly change length (Figure 60). Thus, under conditions where Kif1SA cannot accumulate to sufficient amounts at the plus ends (50 nM His-Kif1SA^FL or 100 nM His-Kif1SA^FL + high concen- trations of AMP-PNP), microtubules are not efficiently depolymerized. Because AMP-PNP inhibits Kif1SA's motility (data not shown), we were not able to elucidate whether Kif1SA also requires its ATPase activity to de- polymerize microtubules. Taken together, these data demonstrate that Kif1SA is a motile microtubule depoly- merase that depolymerizes longer microtubules more quickly than shorter ones.

The ATPase Activity of Kif1SA Is Stimulated Not Only by Microtubule Polymer but Also by Tubulin Oimer The nonmotile microtubule depolymerases of the Kine- sin-13 family share the characteristic that their ATPase activity-unlike that of conventional kinesins-is stimu- lated not only by microtubule polymer but also by tubu- lin dimers [36]. In vitro assays revealed that Kif1SA's ATPase activity like that of Kip3p [24] was stimulated in a concentration-dependent manner by microtubule polymer and albeit to a lesser extent by tubulin dimer (Figure S6 in the Supplemental Data). Thus, the microtu- bule depolymerases of the Kinesin-S and the Kinesin-13 families distinguish themselves from the Kar3p-a mo- tile kinesin that utilizes its minus-end-directed motility to depolymerize microtubules from their plus ends [3S]-in their tubulin-stimulated ATPase activity.

In conclusion, we have identified the human kinesin Kif1SA as a motile plus-end-directed microtubule depo- lymerase. Thus, Kif1SA integrates both motility and microtubule depolymerization. Most recent reports demonstrated that this dual functionality also applies for Kip3p, the yeast Kinesin-S member [24, 25]. Previ- ously, it has been demonstrated that microtubule depo- Iymerization that occurs predominantly at kinetochore sites [6] can create forces sufficient for moving chromo- somes [39]. For the congression of chromosomes in prometaphase, the depolymerase activities of the non- motile Kinesin-13 proteins Kif2a and Kif2c (MCAK) seem to be dispensable because cells depleted of both kinesins have no defect in chromosome movement and alignment and mitotic progression [40]. In contrast, in Kif1SA-depleted cells, chromosomes oscillate at reduced velocity and completely fail to align at the meta- phase plate. Furthermore, a lack of Kif1SA causes hy- perstable spindle microtubules and the loss of tension across sister kinetochores; the latter results in the acti- vation of the Mad2-dependent spindle-assembly check- point. This loss of tension could also accounts for sus- tained chromosomes oscillations in Kif1S-RNAi cells because, as previously shown, tension on kinetochores controls chromosome movements [41]. Thus, in view of

our data we postulate that Kif1SA is an essential compo- nent for chromosome congression in prometaphase by regulating microtubule dynamics at the plus ends of ki- netochore microtubules. Clearly, the activity of Kif1SA has to be tightly controlled. It is tempting to speculate that the length dependency of the Kif1SA-mediated de- polymerization might play a role in the alignment of chro- mosomes at the mitotic spindle's center, where microtu- bules on both sites of kinetochore pairs are equally long.

In the future, it will be important to elucidate the molec- ular mechanism and regulation of Kinesin-S proteins to understand precisely how this family of motile depoly- merases contribute to the function of the mitotic spindle in eukaryotes.

Acknowledgments

We are exceptionally grateful to D.R. Foltz and D.W. Cleveland for providing HeLa stably expressing YFP-CENPA and Vladimir Varga for providing helpful comments on experiments. We also thank members of the Mayer lab and Nigg department for helpful com- ments on the manuscript. TUM and SH were supported by the Deut- schen Forschungsgemeinschaft (MA 1559/4-4) and the Fonds der Chemischen Industrie, respectively.

References

1. Wittmann, T., Hyman, A., and Desai, A. (2001). The spindle: A dynamic assembly of microtubules and motors. Nat. Cell BioI.

3, E28-E34.

2. Vale, R.D., and Fletterick, R.J. (1997). The design plan of kinesin motors. Annu. Rev. Cell Dev. BioI. 13, 745-777.

3. Sharp, D.J., Rogers, G.C., and Scholey, J.M. (2000). Microtubule motors in mitosis. Nature 407,41-47.

4. Kline-Smith, S.L., and Walczak, C.E. (2004). Mitotic spindle as- sembly and chromosome segregation: Refocusing on microtu- bule dynamics. Mol. Cell 15, 317-327.

5. Mitchison, T.J., and Salmon, E.D. (2001). Mitosis: A history of division. Nat. Cell Bioi. 3, E17-E21.

6. Mitchison, T., Evans, L, Schulze, E, and Kirschner, M. (1986).

Sites of microtubule assembly and disassembly in the mitotic spindle. Cell 45, 515-527.

7. Gorbsky, G.J., Sammak, P.J., and Borisy, G.G. (1988). Microtu- bule dynamics and chromosome motion visualized in living ana- phase cells. J. Cell BioI. 106,1185-1192.

8. Yen, T.J., Li, G., Schaar, B.T., Szilak, I., and Cleveland, D.W.

(1992). CENP-E is a putative kinetochore motor that accumu- lates just before mitosis. Nature 359, 536-539.

9. Walczak, C.E, Mitchison, T.J., and Desai, A. (1996). XKCM1: A Xenopus kinesin-related protein that regulates microtubule dy- namics during mitotic spindle assembly. Cell 84, 37-47.

10. Desai, A., Verma, S., Mitchison, T.J., and Walczak, C.E. (1999).

Kin I kinesins are microtubule-destabilizing enzymes. Cell 96, 69-78.

11. Maney, T., Hunter, A.W., Wagenbach, M., and Wordeman, L (1998). Mitotic centromere-associated kinesin is important for anaphase chromosome segregation. J. Cell BioI. 142, 787-801.

12. McEwen, B.F., Chan, G.K., Zubrowski, B., Savoian, M.S., Sauer, M.T., and Yen, T.J. (2001). CENP-E is essential for reliable bio- riented spindle attachment, but chromosome alignment can be achieved via redundant mechanisms in mammalian cells.

Mol. Bioi. Cell 12, 2776-2789.

13. Lawrence, C.J., Dawe, R.K., Christie, K.R., Cleveland, D.W., Dawson, S.C., Endow, S.A., Goldstein, L.S., Goodson, H.V., Hlrokawa, N., Howard, J., et al. (2004). A standardized kinesin nomenclature. J. Cell Bioi. 167, 19-22.

14. Cottingham, F.R., and Hoyt, M.A. (1997). Mitotic spindle posi- tioning in Saccharomyces cerevisiae is accomplished byantag- onistically acting microtubule motor proteins. J. Cell BioI. 138, 1041-1053.

15. DeZwaan, T.M., Ellingson, E., Pellman, D., and Roof, D.M. (1997).

Kinesin-related KIP3 of Saccharomyces cerevisiae is required for a distinct step in nuclear migration. J. Cell BioI. 138, 1023- 1040.

16. Garcia, M.A., Koonrugsa, N., and Toda, T. (2002). Two kinesin- like Kin I family proteins in fission yeast regulate the establish- ment of metaphase and the onset of anaphase A Curr. BioI.

12, 610-621.

17. Garcia, M.A., Koonrugsa, N., and Toda, T. (2002). Spindle-kinet- ochore attachment requires the combined action of Kin I-like Klp5f6 and Alp14fDisl-MAPs in fission yeast. EMBO J. 21, 6015-6024.

18. West, R.R., Malmstrom, T., and Mclntosh, J.R. (2002). Kinesins klp5(+) and klp6(+) are required for normal chromosome move- ment in mitosis. J. Cell Sei. 115,931-940.

19. West, R.R., Malmstrom, T., Troxell, C.L., and Mclntosh, J.R.

(2001). Two related kinesins, klp5+ and klp6+, foster microtubule disassembly and are required for meiosis in fission yeast. Mol.

BioI. Cell 12, 3919-3932.

20. Gandhi, R., Bonaccorsi, S., Wentworth, D., Doxsey, S., Gatti, M., and Pereira, A (2004). The Drosophila kinesin-like protein KLP67 A is essential for mitotic and male meiotic spindle assem- bly. Mol. BioI. Cell 15, 121-131.

21. Savoian, M.S., Gatt, M.K., Riparbelli, M.G., Callaini, G., and Glover, D.M. (2004). Drosophila Klp67 A is required for proper chromosome congression and segregation during meiosis I.

J. Cell Sci. 117, 3669-3677.

22. Gatt, M.K., Savoian, M.S., Riparbelli, M.G., Massarelli, C., Callaini, G., and Glover, D.M. (2005). Klp67A destabilises pre- anaphase microtubules but subsequently is required to stabilise the central spindle. J. Cell Sci. 118,2671-2682.

23. Zhu, C., Zhao, J., Bibikova, M., Leverson, J.D., Bossy-Wetzel, E., Fan, J.B., Abraham, R.T., and Jiang, W. (2005). Functional anal- ysis of human microtubule-based motor proteins, the kinesins and dyneins, in mitosisfcytokinesis using RNA interference.

Mol. BioI. Cell 16,3187-3199.

24. Gupta, M.L., Jr., Carvalho, P., Roof, D.M., and Pellman, D. (2006).

Plus end-specific depolymerase activity of Kip3, a kinesin-8 pro- tein, explains its role in positioning the yeast mitotic spindle. Nat.

Cell BioI. 8, 913-923.

25. Varga, V., Helenius, J., Tanaka, K., Hyman, A.A., Tanaka, T.U., and Howard, J. (2006). Yeast kinesin-8 depolymerizes microtu- bules in a length-dependent manner. Nat. Cell BioI. 8, 957-962.

26. Martin-Uuesma, S., Stucke, V.M., and Nigg, E.A. (2002). Role of Hecl in spindle checkpoint signaling and kinetochore recruit- ment of MadlfMad2. Science 297, 2267-2270.

27. DeLuca, J.G., Dong, Y., Hergert, P., Strauss, J., Hickey, J.M., Salmon, E.D., and McEwen, B.F. (2005). Hecl and nuf2 are core components of the kinetochore outer plate essential for organizing microtubule attachment sites. Mol. BioI. Cell 16, 519-531.

28. Sillje, H.H., Nagel, S., Korner, R., and Nigg, E.A. (2006). HURP is a Ran-importin beta-regUlated protein that stabilizes kineto- chore microtubules in the vicinity of chromosomes. Curr. BioI.

16,731-742.

29. Koffa, M.D., Casanova, C.M., Santarella, R., Kocher, T., Wilm, M., and Mattaj, I.W. (2006). HURP is part of a Ran-dependent complex involved in spindle formation. Curr. Bioi. 16, 743-754.

30. Howell, B.J., McEwen, B.F., Canman, J.C., Hoffman, D.B., Farrar, E.M., Rieder, C.L., and Salmon, E.D. (2001). Cytoplasmic dyneinfdynactin drives kinetochore protein transport to the spindle poles and has a role in mitotic spindle checkpoint inac- tivation. J. Cell BioI. 155, 1159-1172.

31. Skibbens, R.V., Skeen, V.P., and Salmon, E.D. (1993). Directional instability of kinetochore motility during chromosome

congression and se9regation in mitotic newt lung cells: A push-pull mechanism. J. Cell Bioi. 122,859-675.

32. Gordon, M.B., Howard, L., and Compton, D.A. (2001). Chromo- some movement in mitosis requires microtubule anchorage at spindle poles. J. Cell BioI. 152,425-434.

33. Waters, J.C., Chen, R.H., Murray, AW., and Salmon, E.D. (1998).

Localization of Mad2 to kinetochores depends on microtubule attachment, not tension. J. Cell BioI. 141,1181-1191.

34. Pereira, A.J., Dalby, B., Stewart, R.J., Doxsey, S.J., and Goldstein, L.S. (1997). Mitochondrial association of a plus end- directed microtubule motor expressed during mitosis in Drosophila. J. Cell BioI. 136,1081-1090. . 35. Hyman, A.A., Salser, S., Drechsel, D.N., Unwin, N., and

Mitchison, T.J. (1992). Role of GTP hydrolysis in microtubule dynamics: Information from a slowly hydrolyzable analogue, GMPCPP.Mol. BioI. Cell 3,1155-1167.

36. Hunter, AW., Caplow, M., Coy, D.L., Hancock, W.O., Diez, S., Wordeman, L., and Howard, J. (2003). The kinesin-related pro- tein MCAK is a microtubule depolymerase that forms an ATP- hydrolyzing complex at microtubule ends. Mol. Cell 11,445-457.

37. Moore, A., and Wordeman, L. (2004). The mechanism, function and regulation of depolymerizing kinesins during mitosis.

Trends Cell Bioi. 14, 537-546.

38. Sproul, L.R., Anderson, D.J., Mackey, AT., Saunders, W.S., and Gilbert, S.P. (2005). Cikl targets the minus-end kinesin depoly- merase kar3 to microtubule plus ends. Curr. BioI. 15, 1420-1427.

39. Koshland, D.E., Mitchison, T.J., and Kirschner, M.W. (1988).

Polewards chromosome movement driven by microtubule de- polymerization in vitro. Nature 331, 499-504.

40. Ganem, N.J., Upton, K., and Compton, D.A (2005). Efficient mitosis in human cells lacking poleward microtubule flux. Curr.

Bioi. 15, 1827-1832.

41. Skibbens, R.V., and Salmon, E.D. (1997). Micromanipulation of chromosomes in mitotic vertebrate tissue cells: Tension con- trols the state of kinetochore movement. Exp. Cell Res. 235, 314-324.