4302 Research Article

IntroductionCytokinesis is the final step of mitosis. In most animal cells, acontractile ring, which comprises mainly actin and myosin, isformed at the cleavage plane and divides a parent cell bodyinto two daughter cells. The contraction of the contractile ringis mediated by an interaction between actin and myosin in amechanism analogous to the contraction of muscle. Insuspension culture, Dictyostelium cells become multinucleatedand finally lyse when a single copy of the gene encoding theheavy chain of myosin II is deleted by homologousrecombination (De Lozanne and Spudich, 1987) or silenced byantisense RNA (Knecht and Loomis, 1987). However, thesemyosin-null cells can still divide on a substratum.

When multinucleated myosin-II-null cells in suspensionculture are placed on the substratum, they adhere tightly to thesubstratum and divide without a preceding ordinary nucleardivision. This phenomenon is called traction-mediatedcytofission, and the traction force against the substratumgenerated by the two daughter cells is considered to mediatethis type of cytokinesis (Spudich, 1989; Fukui et al., 1990).Interestingly, when myosin-II-null cells are cultured on thesubstratum, they can divide in a cell-cycle-dependent mannerby means of an almost morphologically normal cytokinesis(Gerisch and Weber, 2000). This cell-cycle-dependentcytokinesis is distinguishable from traction-mediatedcytofission and is referred to as cytokinesis B (Neujahr et al.,1997; Zang et al., 1997). Uyeda and colleagues (Uyeda et al.,2000) renamed these three modes of cytokinesis as follows: themyosin-II-dependent (contractile-ring-dependent) cytokinesis

is cytokinesis A, cell-cycle-dependent and myosin-II-independent cytokinesis is cytokinesis B, and cell-cycle-independent and myosin-II-independent cytokinesis iscytokinesis C. The molecular mechanisms of cytokinesis B andC have not been clarified to date. Probably, these types ofcytokinesis are mediated by the traction force generated by thetwo daughter cells. Recent research has shown similarsubstratum-dependent and contractile-ring-independentcytokinesis in mammalian cells (Kanada et al., 2005).

Several actin-containing structures have been described inDictyostelium cells. Filopods, long thin extensions containingactin-bundles; microvilli, short thin extensions containingactin-bundles; and pseudopods or lamellipodia, largeextensions containing dense actin-meshworks that are mainlyobserved at the anterior half of migrating cells and the polarregions of dividing cells (Yumura et al., 1984), which arecommon to many animal cells. Dictyostelium cells have nolarge actin bundles inside the cell, such as stress fibers. Actinfilaments are localized mainly at the cortex to form an actin-rich cortex. During phagocytosis, the cell forms a phagocyticcup surrounding the solid food, which is formed by an actinmesh. Crown-like structures (De Hostos et al., 1991) areobserved mainly in cells drinking fluid medium and sometimesbehave as pseudopods. Actin filaments also form clusters atfocal adhesions for cell migration (Uchida and Yumura, 2004).

In the present study, to investigate the role of actin insubstratum-dependent cytokinesis, the dynamics of actin inmyosin-II-null cells expressing GFP-actin were observed bytotal internal reflection fluorescence (TIRF) microscopy. We

Cell division of various animal cells depends on theirattachment to a substratum. Dictyostelium cells deficient intype II myosin, analogous to myosin in muscle, can divideon a substratum without the contractile ring. To investigatethe mechanism of this substratum-dependent cytokinesis,the dynamics of actin in the ventral cortex were observedby confocal and total internal reflection fluorescencemicroscopy. Specifically during mitosis, we found novelactin-containing structures (mitosis-specific dynamic actinstructures, MiDASes) underneath the nuclei andcentrosomes. When the nucleus divided, the MiDAS alsosplit in two and followed the movement of the daughternuclei. At that time, the distal ends of astral microtubulesreached mainly the MiDAS regions of the ventral cortex.An inhibitor of microtubules induced disappearance ofMiDASes, leading to aborted cytokinesis, suggesting that

astral microtubules are required for the formation andmaintenance of MiDASes. Fluorescence recovery afterphotobleaching experiments revealed that the MiDAS washighly dynamic and comprised small actin-containing dot-like structures. Interference reflection microscopy andassays blowing away the cell bodies by jet streamingshowed that MiDASes were major attachment sites ofdividing cells. Thus, the MiDASes are strong candidates forscaffolds for substratum-dependent cytokinesis, serving totransmit mechanical force to the substratum.

Supplementary material available online athttp://jcs.biologists.org/cgi/content/full/120/24/4302/DC1

Key words: Actin, Cell-substratum adhesion, Cytokinesis, GFP,Myosin

Summary

A novel mitosis-specific dynamic actin structure inDictyostelium cellsGo Itoh and Shigehiko Yumura*Department of Biology, Faculty of Science, Yamaguchi University, Yamaguchi 753-8512, Japan*Author for correspondence (e-mail: yumura@yamaguchi-u.ac.jp)

Accepted 9 October 2007Journal of Cell Science 120, 4302-4309 Published by The Company of Biologists 2007doi:10.1242/jcs.015875

Jour

nal o

f Cel

l Sci

ence

4303Mitosis-specific dynamic actin structure

found novel dynamic actin structures, which appearedspecifically during mitosis. We referred to these novelstructures as mitosis-specific dynamic actin structures(MiDASes) and investigated their dynamics and functions. Ourresults demonstrate that MiDASes are formed beneath thecentrosomes presumably as a result of some signals transmittedalong astral microtubules and play an important role incytokinesis as scaffolds for transmitting the traction force tothe substratum.

ResultsA novel mitosis-specific dynamic actin structureDictyostelium cells deficient in myosin II can divide on asubstratum, although they cannot divide in suspension culture.In order to investigate the mechanism of this substratum-dependent cytokinesis, we observed the dynamics of actin in

the ventral cortex of dividing myosin-II-null cells expressingGFP-actin on the substratum by TIRF microscopy, whichilluminates only ~100 nm above the coverslip. Consistent withobservations by conventional fluorescence microscopy(Yumura et al., 1984), actin localized to polar pseudopods ofdividing cells (arrowheads in Fig. 1A). TIRF microscopyrevealed other major structures containing actin (circles in Fig.1A), which were not observed in interphase cells (Fig. 1B) butwere present only in the mitotic cells. These structures arereferred to as MiDASes hereafter. A MiDAS emerged as anirregularly shaped and large flat sheet in the middle of theventral cell cortex when the spindle began to elongate. TheMiDAS split in two in accord with the nuclear division andmoved to each daughter cell. The MiDASes dynamicallychanged their shape and size from 2-5 �m in diameter (see alsosupplementary material Movie 1). The trajectories of two

MiDASes during cell division are shown in Fig.1C. Fig. 1D shows the time-course of the area ofMiDASes, showing that they disappeared shortlyafter the completion of cytokinesis. When thecells were treated with latrunculin A, whichdepolymerizes actin filaments, MiDASesdisappeared within ~40 seconds (data notshown). Furthermore, MiDASes stained withtetramethyl rhodamine-phalloidin, which bindsspecifically to actin filaments, as shown below.Therefore, MiDASes comprise actin filaments.MiDASes were also observed in dividing wild-type cells, although their appearance was verytransient (circles in Fig. 1E). We conclude that theMiDAS is a novel mitosis-specific actin-containing structure.

Fig. 1. The dynamics of actin in myosin-null (HS1)cells during cell division and interphase. HS1 cellsexpressing GFP-actin were observed during celldivision (A) and interphase (B) by TIRF microscopy.(A) During cell division, actin localized mainly atboth polar pseudopods (arrowheads) and at the centersof each daughter cell (circles). The latter structures arereferred to as MiDASes in the present study. (B) Ininterphase, actin localized mainly in pseudopods(arrowheads) at the leading edge of a migrating cell,but any large actin-containing structures such asMiDASes could not be observed. (C) Trajectories ofcentroids of two MiDASes observed until thecompletion of cytokinesis. The arrowhead indicatesthe original position, and the arrows indicate the finalpositions of each MiDAS. (D) Time-course of the areaof MiDASes in both daughter halves. The arrowshows the timing of complete scission of the cell (410seconds), showing that they disappeared shortly afterthe completion of cytokinesis. A movie(supplementary material Movie 1) is available,showing that MiDASes move into each daughter half.(E) Typical MiDASes (circles) in wild-type cells; theyonly appeared transiently. Bars, 10 �m. Note that,interestingly, the fluorescence of MiDASes alternatedon each side in ~20% of the cells examined (10/44;see supplementary material Fig. S1).

Jour

nal o

f Cel

l Sci

ence

4304

Astral microtubules are required for the formation andmaintenance of MiDASesFig. 2A-F shows a series of optical sections by confocalmicroscopy of fixed multinucleated cells expressing GFP-histone1, which shows the position of nuclei, and stained withtetramethyl rhodamine-phalloidin. Multinucleated cells wereprepared in suspension culture as myosin-null cells cannotdivide in suspension. Fig. 2H-M shows multinucleated cellsexpressing GFP–�-tubulin stained with tetramethylrhodamine-phalloidin. MiDASes were located underneath thenuclei and the centrosomes. The latter two organellesassociated with each other. The number of MiDASes was equalto that of nuclei and centrosomes. These observations suggestthe possibility that nuclei and/or centrosomes might induce theformation of MiDASes. Centrosomes especially are plausibleas inducers because astral microtubules emerging from themcan directly reach the ventral cortex to induce formation ofMiDASes.

To assess this possibility, MiDASes and microtubules wereobserved simultaneously in live cells expressing mCherry-actinand GFP–�-tubulin by confocal microscopy (Fig. 3A-C).Consistent with the observation of fixed cells, MiDASes werelocated underneath the centrosomes and, interestingly, relocatedfollowing the movement of centrosomes (Fig. 3B,C). As TIRFmicroscopy can limit illumination to ~100 nm above thecoverslip, which corresponds to the thickness of the ventral

Journal of Cell Science 120 (24)

cortex, including the actin cortex and MiDASes, it is possible toobserve the correlation between the distal ends of eachmicrotubule and the MiDASes. Fig. 3D shows typical images ofthe relationship between the distal ends of microtubules andMiDASes that were captured by an exposure of 100 milliseconds.Fig. 3E shows a sum of traces of microtubules (green) andoutlines of MiDAS (red) for 15 seconds, indicating that the distalends of microtubules reached mainly into the MiDAS regions.These observations strongly suggest that astral microtubulesemerging from the centrosomes determine the position ofMiDASes, probably by sending signals to the ventral cortex.

To examine further the possible role of microtubules in theformation of MiDASes, myosin-null cells were exposed tothiabendazole, which specifically disrupts microtubules in amanner similar to that of nocodazole (Kitanishi et al., 1984).Shortly after the addition of thiabendazole, the astralmicrotubules shortened (Fig. 4A) and MiDASes graduallyregressed in size and finally disappeared (Fig. 4B). Whenthiabendazole was removed by washing with buffer, MiDASesreappeared underneath the centrosomes (arrows in Fig. 4B).These results strongly suggest that the astral microtubules carrysome signals to the cortex that are involved in the formationand maintenance of MiDASes.

Fig. 2. Distribution of MiDASes, nuclei and centrosomes inmultinucleated HS1 cells. Shown is a series of vertical opticalsections of a representative multinucleated HS1 cell during celldivision. To visualize F-actin and nuclei (A-G), HS1 cells expressingGFP-histone1 (green), which represents the position of nuclei, werefixed and stained with tetramethyl rhodamine-phalloidin (red).(H-N) The distribution of F-actin (red) and GFP–�-tubulin (green).The optical sections are 1.5 �m in thickness and separated from eachother by 0.6 �m (A-F,H-M). Panels G and N show 3D-reconstructedimages from panels A-F and H-M, respectively. The positions ofMiDASes (arrows) correspond to those of nuclei and centrosomes.Bars, 10 �m.

Fig. 3. Centrosomes and astral microtubules determine the positionof MiDASes. (A) MiDASes and centrosomes were observedsimultaneously in live HS1 cells expressing mCherry-actin andGFP–�-tubulin. (B) Sequential images in the box of panel A. Thecentroid of the MiDAS (closed circle) relocated, following themovement of the centrosome (open circles). (C) Trajectories of acentrosome (green line) and a MiDAS (red line) observed for 99.9seconds. Arrowheads indicate the original positions and black arrowsindicate the final positions of the centrosome and the MiDAS,respectively. (D) TIRF microscopy of a typical dividingmultinucleated cell possessing four MiDASes, showing the distalends of microtubules (green) in the MiDAS regions (arrows). Theseimages were captured by an exposure of 100 mseconds. (E) Sum oftraces of microtubules (green) and outlines of MiDASes (red)acquired for 15 seconds, indicating that the distal ends ofmicrotubules reaching to the ventral cell cortex are limited mainly tothe MiDAS regions. Bars, 2 �m.

Jour

nal o

f Cel

l Sci

ence

4305Mitosis-specific dynamic actin structure

MiDASes are required for substratum-dependentcytokinesisMyosin-null cells fail to complete cytokinesis on a substratumwith a frequency of about 10% (Neujahr et al., 1997). In suchcases, one half of the dividing cell was resorbed by the otherhalf, probably owing to an imbalance in the motile activitiesbetween the two halves. We frequently observed abortivecytokinesis when MiDASes disappeared by accident. When oneof the MiDASes disappeared (arrows in Fig. 5A), this half of thecell was finally resorbed by the other. This phenomenon wasobserved in 21 out of 24 cells examined. The other three cellsalso showed abortive cytokinesis, leaving only small fragments.Quantitative analysis shows that the area of the MiDASdecreased coincidentally with that of the resorbed half (Fig. 5B).

To assess the role of MiDASes in the completion ofcytokinesis, MiDASes were artificially disrupted by treatmentwith the microtubule inhibitor thiabendazole. All of the cellsin the early stage of cytokinesis (n=26) that lost MiDASes afterthe treatment with thiabendazole failed to complete cytokinesis(Fig. 5C). These observations indicate that the MiDAS isrequired for cytokinesis of myosin-null cells or substratum-dependent cytokinesis.

MiDASes function as scaffolds for substratum-dependent cytokinesisSince MiDASes exist in the ventral cell cortex and arenecessary for substratum-dependent cytokinesis, they mighttransmit the traction force generated by the cell to thesubstratum. If this is the case, the cell membrane underneath

MiDASes must be attached to the substratum. To assess thispossibility, myosin-null cells were observed by interferencereflection microscopy (IRM), which is generally used to detectthe adhesion sites of cells. Fig. 6A shows dual images of GFP-actin and IRM, which shows the existence of a darker tone atthe MiDAS regions in comparison with the tone at otherventral membrane areas, suggesting that MiDAS regions arecloser to the substratum than the other regions of the ventralmembrane.

To assess further whether the dividing cells attach to thesubstratum at the MiDAS regions, cell bodies were blownaway by a jet of buffer from a pipette under confocalmicroscopy. Fig. 6B,C shows MiDASes before and after

Fig. 4. Astral microtubules are required for the formation andmaintenance of MiDASes. HS1 cells expressing GFP–�-tubulin (A)or GFP-actin (B) were exposed to thiabendazole, an inhibitor ofmicrotubules, under confocal microscopy. Shortly after the additionof thiabendazole, astral microtubules shortened (A), and MiDASes(arrowheads in B) gradually regressed in size and finally disappeared(0-65 seconds in panel B). When thiabendazole was removed bywashing with buffer 65 seconds after addition of thiabendazole,MiDASes gradually reappeared underneath the centrosomes (105-160 seconds). The arrows indicate the restored MiDASes (160seconds). Bars, 5 �m.

Fig. 5. Loss of MiDASes results in the failure of cytokinesis.Dividing HS1 cells expressing GFP-actin were observed by confocalmicroscopy. When one of the MiDASes (arrows) disappeared byaccident, the other half of the cell resorbed this half, and eventuallythe cell failed to complete cytokinesis (A). The arrows indicate theposition of the MiDAS in the resorbed half of the cell. (B) Time-course of area of left (filled squares) and right (filled triangles)halves, which are divided by a line in panel A. Note that the area ofthe resorbing half (open arrowhead) rapidly decreased after thedisappearance of the MiDAS (closed arrowhead). Also, when thedisappearance of MiDASes was artificially induced by treatmentwith thiabendazole, all of the cells (n=26 cells) failed to divide (C).Bars, 10 �m.

Jour

nal o

f Cel

l Sci

ence

4306

blowing away the cell bodies. In every case, only MiDASregions remained attached to the substratum, although otherparts of the cells were detached by the blowing, indicating thatdividing cells were strongly attached to the substratum mainlyat the MiDAS regions. These observations strongly suggestthat MiDASes are scaffolds for substratum-dependentcytokinesis and platforms for the cell to transmit mechanicalforce to the substratum.

Dynamics of actin in MiDASesThe size and shape of MiDASes were not constant but changeddynamically (see supplementary material Movie 1). Toinvestigate the dynamics of actin in MiDASes, a part of theMiDAS region was photobleached by scanning laserillumination. The fluorescence of the bleached regionsrecovered immediately (Fig. 7A,B, half time of the recovery,

Journal of Cell Science 120 (24)

2.15±0.89 seconds, mean ± s.d., n=17), which indicates thatactin in the MiDASes undergoes continuous rapid turnover.Fig. 7C,D shows high magnifications of the fluorescencerecovery processes. After photobleaching, fluorescenceappeared as individual small dots and finally became anaggregate of these dots.

Next, we examined how MiDASes first emerged in the earlyanaphase and how they disappeared after the completion ofcytokinesis. Nascent MiDASes comprised multiple smallactin-containing dots. Importantly, these small actin dots didnot change their position; they appeared and then disappearedat the same position (arrows in Fig. 8A). The average durationof the appearance was 18.17±3.13 seconds (n=400 dots, 10cells). The dots gradually increased in number and they fusedwith each other to become a large MiDAS. Likewise, after thefinal stage of cytokinesis, small actin-containing dots inMiDASes gradually disappeared, and the size of the MiDASgradually decreased (Fig. 8B). These small actin dots seemedto be identical to actin foci, which are considered to be focaladhesions for migration of Dictyostelium cells (Yumura andKitanishi-Yumura, 1990; Uchida and Yumura, 2004). Actinfoci also appeared and disappeared with a duration similar tothat of actin dots in MiDASes. However, as actin foci still existeven after the disappearance of MiDASes induced by thedisruption of microtubules, the formation of actin foci does notseem to be regulated by microtubules (Fig. 5C).

These observations suggest that the MiDASes comprisesmall actin-containing dots, which turn over rapidly, and theMiDASes can change their location as a whole through thisdynamic turnover.

DiscussionIn Dictyostelium cells, myosin-II-null cells can still divide ona substratum, although they will die in suspension culture. Ithas been considered that the substratum-dependent cell-division progresses through the traction force generated as thesplitting daughter halves migrate in opposite directions. Weinvestigated the role of actin in this substratum-dependent andmyosin-II-independent cytokinesis. Careful observation ofmyosin-II-null cells expressing GFP-actin by TIRFmicroscopy allowed us to find novel actin structures(MiDASes), which appear specifically during mitosis.MiDASes emerged underneath the centrosomes, split in twoand moved, following the centrosomes. The distal ends ofastral microtubules emerging from the centrosomes reachedmainly the MiDAS regions. When MiDASes disappeared,accidentally or induced artificially after application ofmicrotubule inhibitors, the cells showed abortive cytokinesis.In addition, the cell membrane at the MiDAS regions wastightly attached to the substratum, suggesting that the tractionforce is transmitted through these regions. Therefore,MiDASes can play an important role as scaffolds for thesubstratum-dependent cytokinesis.

Because astral microtubules were required for the formationof MiDASes, there might be some signals along the astralmicrotubules that induce the formation and maintenance ofMiDASes. In vertebrate animal cells, CLIP 170, EB1 and APC(collectively termed +Tips), which associate to the plus-endsof microtubules, regulate the actin cytoskeleton at the leadingedges (Fukata et al., 2002; Kawasaki et al., 2000; Wen et al.,2004). EB1 has been identified also in Dictyostelium cells and

Fig. 6. Dividing cells attach to the substratum at MiDAS regions. Arepresentative dividing multinucleated HS1 cell simultaneouslyobserved by fluorescence microscopy and interference reflectionmicroscopy (IRM). Left panels show fluorescence images of GFP-actin, and right panels show IRM images. (A) IRM images representa darker tone (arrowheads) at MiDAS regions compared with thatpresent at other ventral membrane regions. Therefore, the cellmembrane at MiDAS regions is closer to the substratum than theother ventral cell membrane regions. (B,C) Cell bodies blown awayby a jet of buffer from a pipette under confocal microscopy. Leftpanels show a half of dividing cells before blowing, and right panelsshow the same areas after blowing away the cell bodies, as revealedby fluorescence microscopy for GFP actin and phase-contrastmicroscopy. The figures are representative results from 20 cells.Note that only MiDAS regions remained attached to the substratum,indicating that dividing cells strongly attach to the substratum mainlyat the MiDAS regions. Bar, 5 �m.

Jour

nal o

f Cel

l Sci

ence

4307Mitosis-specific dynamic actin structure

colocalizes with actin (Rehberg and Graf, 2002), but whetherDdEB1 induces the expansion of pseudopods is still unknown.In the case of 3T3 fibroblasts, the ruffling is induced by Rac1,a small GTPase that is activated by polymerizing microtubules(Waterman-Storer et al., 1999). As the plus-ends of astralmicrotubules are concentrated at the polar regions of a dividingcell, they might activate ruffling there (Neujahr et al., 1998).In our observations, MiDASes sometimes fused with actin ofpseudopods (data not shown), suggesting that similar signalsby the plus-ends of astral microtubules might induce actinpolymerization both in pseudopods at the leading edges and inMiDASes in the ventral cell cortex.

As shown in Fig. 8A, small actin-containing dots increasedin number and formed a nascent MiDAS in early anaphase.Conversely, when MiDASes disappeared after the completion

of cytokinesis, the actin dots decreased in number (Fig. 8B).These observations suggest that MiDASes comprise actin-containing dots. Interestingly, the individual actin dots didnot change their positions: they appeared and disappeared atthe same positions. These actin dots were indistinguishablefrom actin foci – the duration of their appearance and theirsizes were identical. In a previous study, we showed that actinfoci are fixed in position relative to the substratum and are acandidate for the focal adhesions of Dictyostelium during cellmigration (Uchida and Yumura, 2004). It would seem thatMiDASes changed their position so as to coordinate with theposition of the centrosomes (Figs 1 and 3). The change in theposition of MiDASes can be explained by this rapid turnoverof actin dots; their appearance in the front leading edge of theMiDAS and their disappearance in the rear result in a gradual

Fig. 7. Actin rapidly turnsover in MiDASes. Thefluorescence of a part of aMiDAS region (within blacksquares) was bleached in HS1cells expressing GFP-actin(A). The fluorescence of thebleached region recoveredshortly after photobleaching(B). The half-time of recoverywas 2.15±0.89 seconds (mean± s.d., n=17). (C,D) A high-magnification view of thefluorescence recoveryexperiments. (D) Imagesinside the square in panel C.After photobleaching,fluorescence appeared asindividual small fluorescentdots (arrowheads) and finallyformed an aggregate. Bar,2 �m.

Fig. 8. MiDASes comprise dots of actin. HS1 cellsexpressing GFP-actin were observed by TIRF microscopy.In the early stage of mitosis, a nascent MiDAS(arrowhead) was formed by aggregation of many smallactin dots (A). These small actin dots did not change theirpositions; they appeared and then disappeared at the sameposition (panel A, arrows). By contrast, during the finalstage of cytokinesis, these actin dots in MiDASesgradually disappeared, and the size of the MiDASdecreased gradually (B). Bar, 2 �m.

Jour

nal o

f Cel

l Sci

ence

4308

change in the location of the MiDAS as a whole. A recentTIRF microscopy study revealed mobile actin clusters, thesizes of which were similar to actin foci, during thereorganization of actin networks after they weredepolymerized by latrunculin A and after drug removal(Gerisch et al., 2004). The actin foci and the actin dots in theMiDASes seem to be different to actin clusters because theformer are stationary with respect to the substratum.

Do wild-type Dictyostelium cells perform cytokinesis by dualmechanisms, both myosin-II-dependent and -independent?MiDASes appear transiently, even in wild-type cells (Fig. 1E).When wild-type cells divide on a substratum, most of themround up, suggesting that most of the contact sites are detachedfrom the substratum. However, some of the population of cellsdivides when adhering strongly to the substratum, maintainingtheir flat morphology. In this case, the cellular shape duringmitosis is similar to that of myosin-II-null cells (Neujahr et al.,1997; Nagasaki et al., 2001). Probably, wild-type cells dividethrough both myosin-II-dependent and -independentmechanisms. It is also plausible that the appearance ofMiDASes might be a result of adaptation in myosin-null cells,which is transitory and less important in wild-type cells becausethey can divide in the absence of a substratum. Myosin-II-independent cell division has also been observed in mammaliancells. Epitheloid kidney cells injected with antibodies againstmyosin to diminish the levels of myosin II in the equatorialregion can still divide (Zurek et al., 1990). Normal rat kidneycells can divide in the presence of blebbistatin, a potent inhibitorof the nonmuscle myosin II ATPase, in a manner similar to thatof Dictyostelium myosin-null cells (Kanada et al., 2005).Normal kidney cells and 3T3 fibroblasts that weremicroinjected with C3 ribosyltransferase, an inhibitor of Rho,could not accumulate myosin at the cleavage furrow but stilldivided (O’Connell et al., 1999). This cleavage is consideredcell-substratum dependent because HeLa cells, which detachfrom the substratum during cell division, cannot divide wheninjected with C3 enzyme. The same authors described how,when fluorescent beads were attached to the surface of C3-injected normal rat kidney cells, they moved towards thechromosomes, and the resultant cluster of beads followed themovement of chromosomes and split in two as if they weretethered to the chromosomes. At that time, clusters of corticalF-actin were localized under the chromosomes, which isreminiscent of MiDASes (O’Connell et al., 1999). Morerecently, Guha and colleagues (Guha et al., 2005) showed thatlarge actin structures similar to MiDASes appeared in dividingnormal rat kidney cells after local treatment with the myosin IIinhibitor blebbistatin. It should be noted that MiDASes can beobserved even in myosin-null Dictyostelium cells expressingmotorless myosin II (G.I and S.Y., unpublished). Therefore, theloss of myosin ATPase activities seems to induce or enhancethe emergence of MiDASes.

Myosin-II-independent and adhesion-dependent celldivision is probably an ancient mechanism that emerged beforethe evolution of myosin and is still available in various cells(Uyeda and Nagasaki, 2004). Therefore, further investigationof this mechanism is required for a complete understanding ofthe mechanism of cytokinesis. In future studies, it will beimportant to clarify the putative signals transmitted along theastral microtubules that regulate the formation andmaintenance of MiDASes.

Journal of Cell Science 120 (24)

Materials and MethodsCell culture and transformationDictyostelium discoideum cells were cultured in a plastic dish at 22°C in HL5medium (1.3% bacteriological peptone, 0.75% yeast extract, 85.5 mM D-glucose,3.5 mM Na2HPO4 12H2O, 3.5 mM KH2PO4, pH 6.5). Plasmid DNA constructsincluding GFP-actin, mCherry-actin, GFP–�-tubulin and GFP-histone1 weretransformed into cells by electroporation (Yumura et al., 1995). Briefly, cells thatgrew to confluence in plastic dishes were washed twice with 15 mM Na/Kphosphate buffer, and they were centrifuged at 350 g for 90 seconds at 4°C. Aftercentrifugation, an equal volume of the phosphate buffer was added to the pellet,which was then suspended by mild tapping. 30 �l of the cell suspension and 3 �lof plasmid DNA solution were mixed and charged at 380 V, which was repeatedtwice. After electroporation, the solution in the cuvette was suspended in 10 ml ofHL5 medium including 40 �g/ml of streptomycin in a plastic Petri dish. After 12hours, 10 �g/ml of G418 (Sigma) and/or blasticidin (Kaken) was added forselection.

Treatment with thiabendazoleHS1 cells null for the myosin II heavy chain and expressing GFP-actin or GFP–�-tubulin that grew in HL5 medium were washed with BSS (10 mM NaCl2, 10 mMKCl, 3 mM CaCl2, 2 mM MES, pH 6.3). After incubation for 3 hours, cells wereplaced in a chamber that was made by attaching a silicon sheet with a hole (10 mmin diameter) to a 24�50 mm coverslip. An equal amount of thiabendazole(2�10–4 M in BSS) was added to the chamber. HS1 cells took 4-7 minutes to divide.Microtubules disappeared within ~30 seconds, as shown in Fig. 4A. Next,thiabendazole was removed by replacing it with BSS. The cells were observed byconfocal microscopy (LSM510, Carl Zeiss).

Fixation and stainingHS1 cells expressing GFP-histone1 or GFP–�-tubulin that grew in HL5 mediumwere suspended and then placed on an 18�18 mm coverslip. After 20 minutes,fixation was performed with ethanol containing 1% (w/v) formaldehyde at –17°Cfor 5 minutes. After washing three times with PBS (135 mM NaCl2, 2.7 mM KCl,1.5 mM KH2PO4, 80 mM NaH2PO4, pH 7.3) at intervals of 5 minutes, cells wereincubated with tetramethyl rhodamine (TRITC)-phalloidin for 30 minutes and thenwashed three times with PBS at intervals of 5 minutes. After mounting the samplesin mounting medium [10% polyvinyl alcohol in PBS containing 50% (v/v)glycerol], the cells were observed by confocal microscopy.

Confocal microscopyFluorescence images of HS1 cells expressing GFP-actin or GFP–�-tubulin wereobtained with the confocal microscope system, a LSM 510 microscope (Carl Zeiss)equipped with a 100� Plan Neofluar objective (NA 1.3). For excitation of GFP andTRITC, an argon laser (488 nm line) and HeNe laser (543 nm line) were used,respectively. Time-lapse microscopy of HS1 cells expressing GFP-actin waspreformed at an optical section of 1.5 �m, with focusing on the ventral surface ofthe cell. To acquire sequential fluorescence images from bottom to top, confocalsections were scanned up at distances of 0.3 �m in the Z-axis. IRM images andfluorescence images were acquired simultaneously, as described previously (Uchidaand Yumura, 2004). Photobleaching was performed as described previously(Yumura, 2001). Measurement of fluorescence intensity was performed by Scionsoftware (Scion Corporation). To perform an assay blowing away the cell bodies,the cell bodies were blown away by a jet of the buffer from a pipette while underconfocal microscopy.

TIRF microscopyA TIRF microscopy system (based on I X70, Olympus) was constructed asdescribed previously (Tokunaga et al., 1997). The fluorescence of the cellsexpressing GFP-actin was observed with the TIRF microscope equipped with aPlanApo oil objective (60�, NA 1.45, Olympus). The wild-type cells were slightlycompressed by an agar sheet to observe the ventral surface, as described previously(Yumura et al., 1984). Illumination was performed by an argon laser (40 mW, theNational Laser) and a HeNe laser (2 mW, Research Electro Optics). Thefluorescence images were captured by a digital cooled CCD camera (C4742-95-12ER, Hamamatsu) and digitized with IP lab software for MS Windows (Scanalytics).

We thank T. Q. P. Uyeda and A. Nagasaki for providing constructsof GFP–�-tubulin, GFP-histone1 and mCherry-actin. We also thankT. Kitanishi-Yumura for helpful discussions. A part of this work wassupported by Grants-in-aid for Scientific Research in Priority Areasfrom the Japan MEXT.

ReferencesDe Hostos, E. L., Bradtke, B., Lottspeich, F., Guggenheim, R. and Gerisch, G.

(1991). Coronin, an actin binding protein of Dictyostelium discoideum localized tocell surface projections, has sequence similarities to G protein � subunits. EMBO J.10, 4097-4104.

Jour

nal o

f Cel

l Sci

ence

4309Mitosis-specific dynamic actin structure

De Lozanne, A. and Spudich, J. A. (1987). Disruption of the Dictyostelium myosinheavy chain gene by homologous recombination. Science 236, 1086-1091.

Fukata, M., Watanabe, T., Noritake, J., Nakagawa, M., Yamaga, M., Kuroda, S.,Matsuura, Y., Iwamatsu, A., Perez, F. and Kaibuchi, K. (2002). Rac1 and Cdc42capture microtubules through IQGAP1 and CLIP-170. Cell 109, 873-885.

Fukui, Y., De Lozanne, A. and Spudich, J. A. (1990). Structure and function of thecytoskeleton of a Dictyostelium myosin-defective mutant. J. Cell Biol. 110, 367-378.

Gerisch, G. and Weber, I. (2000). Cytokinesis without myosin II. Curr. Opin. Cell Biol.12, 126-132.

Gerisch, G., Bretschneider, T., Muller-Taubenberger, A., Simmeth, E., Ecke, M.,Diez, S. and Anderson, K. (2004). Mobile actin clusters and traveling waves in cellsrecovering from actin depolymerization. Biophys. J. 87, 3493-3503.

Guha, M., Zhou, M. and Wang, Y. L. (2005). Cortical actin turnover during cytokinesisrequires myosin II. Curr. Biol. 15, 732-736.

Kanada, M., Nagasaki, A. and Uyeda, T. Q. P. (2005). Adhesion-dependent andcontractile-ring-independent equatorial furrowing during cytokinesis in mammaliancells. Mol. Biol. Cell 16, 3865-3872.

Kawasaki, Y., Senda, T., Ishidate, T., Koyama, R., Morishita, T., Iwayama, Y.,Higuchi, O. and Akiyama, T. (2000). Asef, a link between the tumor suppressor APCand G-protein signaling. Science 289, 1194-1197.

Kitanishi, T., Shibaoka, H. and Fukui, Y. (1984). Disruption of microtubules andretardation of development of Dictyostelium with ethyl N-phenylcarbamate andthiabendazole. Protoplasma 120, 185-196.

Knecht, D. A. and Loomis, W. F. (1987). Antisense RNA inactivation of myosin heavychain gene expression in Dictyostelium discoideum. Science 236, 1081-1086.

Nagasaki, A., Hibi, M., Asano, Y. and Uyeda, T. Q. P. (2001). Genetic approaches todissect the mechanisms of two distinct pathways of cell cycle-coupled cytokinesis inDictyostelium. Cell Struct. Funct. 26, 585-591.

Neujahr, R., Heizer, C. and Gerisch, G. (1997). Myosin II-independent processes inmitotic cells of Dictyostelium discoideum: redistribution of the nuclei, re-arrangementof the actin system and formation of the cleavage furrow. J. Cell Sci. 110, 123-137.

Neujahr, R., Albrecht, R., Kohler, J., Matzner, M., Schwartz, J. M., Westphal, M.and Gerisch, G. (1998). Microtubule-mediated centrosome motility and thepositioning of cleavage furrows in multinucleate myosin II-null cells. J. Cell Sci. 111,1227-1240.

O’Connell, C. B., Wheatley, S. P., Ahmed, S. and Wang, Y. L. (1999). The small GTP-binding protein Rho regulates cortical activities in cultured cells during division. J.Cell Biol. 144, 305-313.

Rehberg, M. and Graf, R. (2002). Dictyostelium EB1 is a genuine centrosomalcomponent required for proper spindle formation. Mol. Biol. Cell 13, 2301-2310.

Spudich, J. A. (1989). In pursuit of myosin function. Cell Regul. 1, 1-11.Tokunaga, M., Kitamura, K., Saito, K., Hikikoshi-Iwane, A. and Yanagida, T. (1997).

Single molecule imaging of fluorophores and enzymatic reactions achieved byobjective-type total internal reflection fluorescence microscopy. Biochem. Biophys.Res. Commun. 235, 47-53.

Uchida, K. S. K. and Yumura, S. (2004). Dynamics of novel feet of Dictyostelium cellsduring migration. J. Cell Sci. 117, 1443-1455.

Uyeda, T. Q. P. and Nagasaki, A. (2004). Variations on a theme: the many modes ofcytokinesis. Curr. Opin. Cell Biol. 16, 55-60.

Uyeda, T. Q. P., Kitayama, C. and Yumura, S. (2000). Myosin II-independentcytokinesis in Dictyostelium: its mechanism and implications. Cell Struct. Funct. 25,1-10.

Waterman-Storer, C. M., Worthylake, R. A., Liu, B. P., Burridge, K. and Salmon,E. D. (1999). Microtubule growth activates Rac1 to promote lamellipodial protrusionin fibroblasts. Nat. Cell Biol. 1, 45-50.

Wen, Y., Eng, C. H., Schmoranzer, J., Cabrera-Poch, N., Morris, E. J. S., Chen, M.,Wallar, B. J., Alberts, A. S. and Gundersen, G. G. (2004). EB1 and APC bind tomDia to stabilize microtubules downstream of Rho and promote cell migration. Nat.Cell Biol. 6, 820-830.

Yumura, S. (2001). Myosin II dynamics and cortical flow during contractile ringformation in Dictyostelium cells. J. Cell Biol. 154, 137-145.

Yumura, S. and Kitanishi-Yumura, T. (1990). Fluorescence-mediated visualization ofactin and myosin filaments in the contractile membrane-cytoskeleton complex ofDictyostelium discoideum. Cell Struct. Funct. 15, 355-364.

Yumura, S., Mori, H. and Fukui, Y. (1984). Localization of actin and myosin for thestudy of ameboid movement in Dictyostelium using improved immunofluorescence. J.Cell Biol. 99, 894-899.

Yumura, S., Matsuzaki, R. and Kitanishi-Yumura, T. (1995). Introduction ofmacromolecules into living Dictyostelium cells by electroporation. Cell Struct. Funct.20, 185-190.

Zang, J. H., Cavet, G., Sabry, J. H., Wagner, P., Moores, S. L. and Spudich, J. A.(1997). On the role of myosin-II in cytokinesis: division of Dictyostelium cells underadhesive and nonadhesive conditions. Mol. Biol. Cell 8, 2617-2629.

Zurek, B., Sanger, J. M., Sanger, J. W. and Jockusch, B. M. (1990). Differential effectsof myosin-antibody complexes on contractile rings and circumferential belts inepitheloid cells. J. Cell Sci. 97, 297-306.

Jour

nal o

f Cel

l Sci

ence