Spatial confinement of active microtubule networksinduces large-scale rotational cytoplasmic flowKazuya Suzukia,b, Makito Miyazakia,b,1, Jun Takagia,2, Takeshi Itabashia,b, and Shin’ichi Ishiwataa

aDepartment of Physics, Faculty of Science and Engineering, Waseda University, Tokyo 169-8555, Japan; and bWaseda Bioscience Research Institute inSingapore, Singapore 138667, Singapore

Edited by Raymond E. Goldstein, University of Cambridge, Cambridge, United Kingdom, and accepted by Editorial Board Member Herbert Levine February 3,2017 (received for review September 28, 2016)

Collective behaviors of motile units through hydrodynamic inter-actions induce directed fluid flow on a larger length scale thanindividual units. In cells, active cytoskeletal systems composed ofpolar filaments and molecular motors drive fluid flow, a processknown as cytoplasmic streaming. The motor-driven elongation ofmicrotubule bundles generates turbulent-like flow in purifiedsystems; however, it remains unclear whether and how microtu-bule bundles induce large-scale directed flow like the cytoplasmicstreaming observed in cells. Here, we adopted Xenopus egg ex-tracts as a model system of the cytoplasm and found that micro-tubule bundle elongation induces directed flow for which thelength scale and timescale depend on the existence of geometricalconstraints. At the lower activity of dynein, kinesins bundle andslide microtubules, organizing extensile microtubule bundles. Inbulk extracts, the extensile bundles connected with each otherand formed a random network, and vortex flows with a lengthscale comparable to the bundle length continually emerged andpersisted for 1 min at multiple places. When the extracts wereencapsulated in droplets, the extensile bundles pushed the dropletboundary. This pushing force initiated symmetry breaking of therandomly oriented bundle network, leading to bundles aligninginto a rotating vortex structure. This vortex induced rotationalcytoplasmic flows on the length scale and timescale that were10- to 100-fold longer than the vortex flows emerging in bulkextracts. Our results suggest that microtubule systems use notonly hydrodynamic interactions but also mechanical interactionsto induce large-scale temporally stable cytoplasmic flow.

active matter | cytoskeleton | self-organization | symmetry breaking |directed flow

Many biological systems exhibit directed fluid flow on vari-ous length scales from the order of 10 m, such as a school

of fish (1), down to the order of 10 μm, such as a bacterial sus-pension (2–5). Swimming bacteria generate dipole fluid flowaround themselves (6). At high concentrations of bacteria, thedipole fluid flow induces orientation of neighboring bacteria,leading to large-scale spiral-pattern organization generating thevortex flow (2–5).Inside cells, active cytoskeletal networks mainly composed

of polar filaments and molecular motors generate fluid flow inthe cytoplasm, a process called cytoplasmic streaming (7–14).Although still little is known about the biological roles of cyto-plasmic streaming, streaming is thought to be necessary for long-distance transport of micrometer-sized components, such asorganelles (9). The active cytoskeletal system is classified intotwo categories, that is, actin and microtubule systems, accord-ing to filament type. In both systems, the polar filaments (actinfilaments or microtubules) are polymerized and depolymerizedusing chemical energy obtained from ATP or GTP hydrolysis,and the molecular motors (myosin for actin filaments, kinesinand dynein for microtubules) convert chemical energy of ATPhydrolysis into mechanical work, resulting in the sliding ofoverlapping filaments. Thus, actin and microtubule systems inthe cytoplasm are highly dynamic.

Actin-based cytoplasmic streaming has been extensively studied,and the following mechanism has been generally accepted:organelle-bound myosins moving on actin filaments producehydrodynamic flows, which align neighboring actin filaments toinduce large-scale streaming (10–12, 14). Similar to actin-basedstreaming, organelle-bound kinesins moving on microtubules arethought to generate large-scale streaming (9). Meanwhile, recentin vitro experiments have demonstrated that purified microtu-bules are organized in extensile bundles by artificially clusteredkinesins, and the network of these bundles generates fluid flows(15). However, it is still unclear whether and how microtubulebundle elongation can drive cytoplasmic flow in cells. Asrevealed in bacteria-driven vortex flow (2–5), the hydrodynamicinteraction between microtubule bundles is probably essential.Bundle elongation generates hydrodynamic flows around itself,which might be able to align neighboring filaments to inducelarge-scale flow. In addition to these hydrodynamic interactions,the physical boundaries of the cells potentially contribute toinitiating the organization of large-scale cytoplasmic flow. This isbecause the persistence length of microtubules (1‒2 mm: ref. 16)is nearly equal to or longer than the cell size (10‒1,000 μm). Infact, a recent theoretical study of actin-based streaming indicatedthat, in a cylindrical cell, actin filaments located on the sidepreferentially reorient from circumferential to longitudinal sothat the filaments minimize their physical curvature, therebydirecting cytoplasmic streaming (12). This work allows us tospeculate that the mechanical interactions between microtubulesand the physical boundary more strongly affect streaming, because

Significance

At the microscopic scale, collective behaviors of motile unitscan induce directed fluid flow on a larger length scale thanindividual units based on their hydrodynamic interactions.Here, we found that the motor-driven extensile behaviors ofmicrotubule bundles in the cytoplasm induce rotational flow ina cell-sized confined space on length scale and timescale thatwere 10- to 100-fold longer than the vortex flows emerging inthe bulk space. These scale differences were derived frommechanical force generation by microtubule bundle elongationnear the physical boundary and the transmission of this forceover the microtubule network. These findings suggest that themicrotubule cytoskeleton utilizes not only hydrodynamic in-teractions but also mechanical interactions to induce large-scale cytoplasmic flow.

Author contributions: K.S., M.M., J.T., T.I., and S.I. designed research; K.S. performedresearch; T.I. contributed new reagents/analytic tools; K.S. analyzed data; and K.S.,M.M., J.T., T.I., and S.I. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. R.E.G. is a Guest Editor invited by the EditorialBoard.1To whom correspondence should be addressed. Email: makito.miyazaki@aoni.waseda.jp.2Present address: Quantitative Mechanobiology Laboratory, National Institute of Genet-ics, Shizuoka 411-8540, Japan.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1616001114/-/DCSupplemental.

2922–2927 | PNAS | March 14, 2017 | vol. 114 | no. 11 www.pnas.org/cgi/doi/10.1073/pnas.1616001114

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

16, 2

020

the flexural rigidity of microtubules is much higher than that ofactin filaments (17, 18).Here, we adopted metaphase Xenopus egg extracts as a model

system of the cytoplasm (19) and examined whether and how themicrotubule system in the cytoplasm induced directed fluid flow.We found that the confinement of cytoplasmic microtubules indroplets induces rotational flow on length scale and timescalethat were 10- to 100-fold longer than the vortex flows emergingin bulk extracts. Quantitative analysis and comparison with thebacteria-driven flows (2–5) suggested that, in addition to hy-drodynamic interactions, mechanical interactions of microtu-bules play important roles in the self-organization of large-scalecytoplasmic flow.

ResultsVortex Flows Emerged in Bulk Cytoplasmic Extracts at the LowerActivity of Dynein. We first observed self-organization ofmicrotubule networks in bulk metaphase extracts. To promotethe assembly of dense microtubule networks, extracts stored onice were mixed with fluorescently labeled Taxol, a microtubule-stabilizing agent (100 nM, unless otherwise stated, at which thedepolymerization rate is decreased, but the frequency ofmicrotubule catastrophe is not affected: ref. 20). To eliminatethe effects of actin dynamics (21, 22), 2 μM cytochalasin D, anactin-depolymerizing agent, was added to the extracts. The extractswere then perfused into a large flow chamber (20 × 20 mm2 with aheight of ∼100 μm), sealed to prevent unexpected fluid flow, andwarmed to 20 °C to initiate microtubule polymerization. As thepolymerization proceeded, multiple microtubule asters assembledover the entire region and connected with each other, organizing alattice network (Fig. 1A). These results are consistent with aprevious report (23).Because dynein participates in aster formation (23), we as-

sumed that dynein played a dominant role in this lattice for-mation. As expected, adding a dynein inhibitor (24, 25)prevented lattice formation, and the network became disordered(Fig. 1B and Fig. S1). In the presence of a dynein inhibitor, cy-toplasmic granules appeared to flow. Particle image velocimetry(PIV) measurements of the granules clarified that turbulent-likecomplex flow emerged in the entire region of bulk extracts (Fig.1C). The complex flow continued for at least several hours, andthe steady-state flow velocity was 0.36 ± 0.26 μm·min−1 (mean ±SD; Fig. 1D). During this period, local vortex flows were con-tinually generated at multiple places (Fig. 1C, blue boxes) anddisappeared within a minute after generation. Consistent withthis, the time correlation of the flow evaluated over the entireregion of the complex flow was decayed within 1 min. To ex-amine the spatial correlation of the complex flow, we calculatedthe equal-time velocity–velocity correlation function C(r, t), de-fined as follows:

Cðr, tÞ=�~vð~x+~r, tÞ ·~vð~x, tÞj~vð~x+~r, tÞjj~vð~x, tÞj

�x,θ,

where h  ix,θ indicates the average over space coordinates~x= ðx, yÞand all angles θ of~r. The Cðr, tÞ shows the orientation correlationof the flow between two points with the distance r at time t. TheCðrÞ exponentially decayed and reached zero at ∼50 μm (definedas the correlation length; Fig. 1E), which was comparable to thesize of the vortex flow (Fig. 1C, blue boxes). The correlationlength was also comparable to the length of single microtubulebundles (34.4 ± 10.1 μm, mean ± SD; n = 63 bundles; Fig. S2),rather than the length of single microtubules (14.5 ± 6.4 μm,mean ± SD; n = 122 microtubules). To investigate how micro-tubule bundles induce complex flows, we observed bundle behav-iors by confocal microscopy and found that the bundleselongated (Movie S1). These results suggested that the complexflows emerged through hydrodynamic interactions between ex-tensile microtubule bundles based on a mechanism similar to

that of bacteria-driven vortex flows (2–5). The complex flow drivenby microtubule bundle elongation was also observed in vitro (15),but the correlation length was fourfold longer than that in ourwork. This fact suggests that the correlation length depends onthe bundle density, because the microtubule bundle density in theprevious work was much higher than that in our work.

Encapsulation of the Extracts in a Droplet Induces RotationalCytoplasmic Flow at Length Scale and Timescale That Are Largerthan the Vortex Flows in Bulk Extracts. We next investigated theeffects of spatial confinement by encapsulating the extracts invarious-sized droplets (Fig. 2A). Due to the high mass density ofthe extracts, the droplets were sedimented on the coverslip anddeformed in the z axis. [The ratio of the droplet height to itsdiameter was 0.465 ± 0.050 (mean ± SD), almost independentlyof the droplet size (n = 25 droplets).] After the encapsulation,the droplets were warmed to initiate microtubule polymerization.We found that, in the presence of dynein inhibitor, rotationalcytoplasmic flow emerged inside the droplets, visualized bypassivated tracer particles (Fig. 2B, Fig. S3A, and Movie S2)and the organelles (Fig. 2C and Movie S3). The flow resemblesmicrotubule-based rotational cytoplasmic streaming observed inembryos, oocytes, and eggs of several species (7–9). Probably due

30 min + p150-CC130 min

Flow velocity (µm min-1)

Cou

nts

400

0

200

0 21 3

0.36 ± 0.26 µm min-1

0 50 100 150Distance r (µm)

0

0.5

1.0

C(r)

161 min

D E

C

A B

Fig. 1. Vortex flows continually emerge and disappear at multiple places inbulk extracts. (A and B) Epifluorescence images of microtubule networksself-organized in bulk Xenopus egg extracts without (A) and with (B) thedynein inhibitor p150-CC1. (Scale bars, 100 μm.) (C) Flow field of the bulkextracts measured by PIV (grid size, 9.5 μm). The vectors (Left) or flowstreamlines (Right) were merged with the bright-field image. The blue boxesshow places where typical vortex flows emerged. Not all vectors are shownfor clarity. (Scale bar, 100 μm.) (D) Flow velocity distribution shown in C.(E) The normalized spatial correlation function C(r, t) as a function of lateralseparation r. Black, orange, green, and cyan lines indicate the spatial correlationfunctions at 155, 161, 167, and 171 min, respectively. In all microscopic images,0 min indicates the timing of elevating the temperature from 0 to 20 °C.

Suzuki et al. PNAS | March 14, 2017 | vol. 114 | no. 11 | 2923

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

16, 2

020

to the droplet deformation in the z axis, rotational flow along thez axis was rarely observed (only a few percentage of all droplets;Movie S4), suggesting that rotational dynamics could bediscussed in the context of a 2D plane. We did not include theresults of the droplets showing the rotational flow along the zaxis. Droplets containing the extracts without Taxol or withnocodazole (a microtubule-depolymerizing agent) did not showrotational flow (Fig. S3 B and C), indicating that a sufficient densityof microtubules was required to induce flow. Compared with the

vortex flows observed in bulk extracts, which persisted for 1 min,the rotational flow in the droplets was sustained for over 10- to100-fold longer. The angular velocity of the flow hωix graduallyincreased after the increase in temperature and then remainedconstant for several tens of minutes (Fig. 2D). Collectively, thespatial confinement induces temporally stable microtubule-drivingrotational flow.Rotational cytoplasmic flow was observed in droplets with a

diameter of 100‒700 μm (Fig. 2E), that is, the length scale of therotational flow in droplets was 10-fold larger than the vortexflows in bulk extracts. The mean angular velocities hωix,t in singledroplets exhibited a linear correlation with R−1 (R, radius of thedroplet; Fig. 2E), implying that the flow velocity near theboundary was nearly constant among various-sized droplets. Infact, PIV analysis showed that, although there were regionalvariations in the flow velocities in droplets (Fig. 2 F and G), theflow velocity near the boundary was almost independent of thedroplet size (Fig. 2H). In droplets smaller than 100 μm in di-ameter, the microtubule bundles formed a cortex-like structure(Fig. S4A), similar to the structures organized by purified mi-crotubules (15, 26) and actin filaments (27). The intersections ofthe fitting lines and the horizontal line in Fig. 2E predicted thatthe maximum diameter of a droplet showing rotational flow was∼800 μm, which is nearly equal to the maximum diameter of thatof droplets showing unidirectional flow. In droplets of over700 μm in diameter, multidirectional flow was observed (Fig. S4Band Movie S5). Thus, the confinement induced rotational cyto-plasmic flow not only at the larger timescale but also at the largerlength scale than the vortex flows emerging in bulk extracts.

Microtubule Bundles Are Arranged in a Rotating Vortex. To clarifythe mechanism for rotational flow, we observed microtubulenetworks by confocal microscopy. When the extracts encapsulatedin droplets were warmed to initiate microtubule polymerization,microtubule bundles were assembled in random orientations,forming a random bundle network (Fig. 3A, 11 min). The neigh-boring bundles located near the droplet boundary were then ori-ented in the same direction, and the entire network began torotate (Fig. 3A, 15 min). Eventually, most of the bundles ori-ented in the rotational direction, resulting in the formation of avortex structure (Fig. 3A, 27 min; Movie S6). This structure re-sembles the spiral-arrayed microtubules observed in embryos,oocytes, and eggs of several species (7–9, 28). However, unlikethe microtubule array observed in animal cells, most of thebundles in droplets seemed not to be anchored to the lipidmembrane (29) and the bundles did not exhibit wave-likemotion (9, 29). Meanwhile, purified microtubules and artificialkinesin clusters also formed a vortex structure in confined space,but this structure did not rotate (30). The difference suggests therole of nonmotor cross-linkers [e.g., microtubule-associated proteins(MAPs)] on the vortex rotation, because nonmotor cross-linkersbridge between microtubules and between microtubule bundles,modifying the mechanical properties of the microtubule bundlenetwork. Although time-lapse observations (Fig. 2C and MovieS3) were not sufficient to conclude whether the organelles weretransported along microtubules by kinesins, the organelles couldenhance the fluid flow even when just bound on the rotatingbundles.The microtubule bundles in the rotating vortex array elon-

gated and buckled or bent after they reached the dropletboundary (Fig. 3B). During these processes, the end of thebundle contacting the boundary did not slip along the boundary.Although there was no other experimental evidence, this factimplies the existence of friction between the end of a bundle andthe boundary. Moreover, as mentioned above, the flow velocitynear the boundary was nearly constant among droplets of varioussizes (Fig. 2H). These results suggest that the driving force forthe rotational flow was produced on the boundary. Taken to-gether, these data led us to speculate that one end of the bundlepushed the boundary, and this pushing force then rotated theentire bundle network to drive the rotational flow.

D

B

26 min 26-55 min

A

x

y

Top

x

z

Cross-section

Coverslip

Microtubules

Mineral oil Lipids

Egg extracts

C

21 min 21-80 min

(deg

min

-1)

0

6

12

-6

-12

Time (min)0 30 60 90 120 150 180 210

ωx

2.0 2.5

500 nM100 nM50 nM

0.5 1.0 1.50

2

4

6

8

0

[Taxol]

(deg

min

-1)

-1 (×10-2 µm-1)R

ωx,t

Flow

vel

ocity

(µm

min

-1)

0

2.5

5.0

0 0.5 1.0

G

E

0 100 200 300d / R R (µm)

F

H73 µm63 µm

74 µm89 µm

163 µm123 µm

169 µm

234 µm322 µm

174 µm

R:

Flow

vel

ocity

nea

r th

e bo

unda

ry (µ

m m

in-1

)

0

2.5

5.0

Fig. 2. Encapsulation of the extracts induces rotational cytoplasmic flow.(A) Schematic illustration and confocal images of the extracts encapsulatedin droplets. (B and C) Time projection of PEG-coated 0.25-μm beads (B) andorganelles (C) in droplets. Arrows represent the rotational direction. 0 minindicates the timing of elevating temperature. For the experiment in C, [OG-Taxol] = 200 nM. (D) Time course of the angular velocity of the beads hωix inthree droplets. The lines and their surrounding areas represent means ± SDs,respectively. (E) Relationship between the droplet radius R and hωix,t with-out distinction of [p150-CC1] = 400, 800, or 2,000 nM. hωix,t indicates theabsolute value of the average angular velocity of beads in the overall regionof a droplet while rotational flow continues at a constant velocity. Weconfirmed that hωix,t and the frequency of rotational flow did not differbetween the three concentrations of p150-CC1. (F) Flow field for a dropletshowing rotational flow measured by PIV. Not all vectors are shown forclarity. (G) Flow velocity profiles of various sized droplets calculated from PIVmeasurements (d, distance from the droplet center). (H) The relationshipbetween R and the flow velocity near the boundary, defined as the flowvelocity averaged over the region of d=R> 0.8. In all microscopic images,dashed lines indicate the droplet boundaries. (Scale bars, 30 μm.)

2924 | www.pnas.org/cgi/doi/10.1073/pnas.1616001114 Suzuki et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

16, 2

020

Microtubule–Microtubule Sliding Propelled by Molecular MotorsDrives Rotational Flow. To test our hypothesis, we examined thedynamics of individual bundles in detail by compressing thedroplets with two coverslips (Fig. 4). In compressed droplets,rotational flow did not occur, but the bundle dynamics wereclearly observed because the motion was restricted in a 2D plane.The bundles elongated and then buckled (or bent) when theyreached the droplet boundary (Fig. 4 A and B, and Movie S7).The elongation rate was 3.58 ± 1.61 μm·min−1 (mean ± SD; n =19 bundles; Fig. 4C), almost equal to the velocity of rotationalflow near the boundary (3.19 ± 0.63 μm·min−1, mean ± SD; n =10 droplets; Fig. 2H). Although hydrodynamic interactions in thedroplets compressed by two coverslips (height, 5‒10 μm) may bedifferent from those in droplets showing rotational flow(height, >50 μm), there was good agreement between the elon-gation rate and the flow velocity near the boundary, stronglysuggesting that microtubule bundle elongation rotated the bun-dle network, generating rotational cytoplasmic flow.Two possibilities exist for microtubule bundle elongation:

(i) microtubule polymerization and (ii) microtubule–microtubulesliding propelled by molecular motors, presumably kinesins (15,26). First, we investigated whether the microtubule polymeriza-tion reaction was essential to drive rotational flow by varyingTaxol concentrations from 50 to 500 nM. Because Taxol reducesthe rates of both growing and shortening of microtubules (20),we expected that an increase in the Taxol concentration maydecrease the angular velocity of the flow. However, the angularvelocities were not decreased by increasing Taxol concentrations(Fig. 2E), even though the microtubule-growing rates at 100 and500 nM Taxol are about 50% slower than that at 50 nM Taxolin vitro (20). Meanwhile, ATP depletion with apyrase (an ATP-hydrolyzing enzyme) or inhibition of ATP hydrolysis of kinesinswith AMP-PNP (a nonhydrolyzable analog of ATP) blocked therotational flow, even though microtubule bundles were assem-bled under both conditions (Fig. S5). Taken together, these re-sults show that the driving force for rotational cytoplasmic flow isderived from microtubule–microtubule sliding propelled bykinesins, rather than from microtubule polymerization.

Model for the Organization of a Vortex Structure Inducing RotationalCytoplasmic Flow. Based on these results, we propose a model forrotational cytoplasmic flow in a confined space under suppres-sion of dynein activity (Fig. 5A). As the microtubule polymeri-zation proceeds, microtubules are bundled in parallel orantiparallel arrangements by kinesins and MAPs (phase 1).These bundles are randomly oriented and connected by cross-linking proteins (kinesins and MAPs), forming a randommicrotubule bundle network (phase 2). Each microtubulebundle is elongated by kinesins. Once the extensile bundlesreach the droplet boundary, they start to generate the momentof force against the microtubule bundle network (phase 3).Because both the magnitude and direction of the moment offorce generated by each microtubule bundle are varied, andbecause the bundle density is low (∼0.0002 bundles·μm−3 at500 nM Taxol; estimated from Fig. 3A, 27 min), the balancebetween clockwise and anticlockwise rotational forces can beeasily disrupted. This spontaneous symmetry breaking inducesrotation of the entire microtubule bundle network, which initiatesrotational cytoplasmic flow (phase 4). The flow promotes thearrangement of microtubule bundles into a vortex pattern, andthe bundle alignment accelerates the flow. This positive-feedbackloop maintains the flow unidirectionally over several tens ofminutes (phase 5). In this context, the rotational direction is notdetermined in one direction, and this is consistent with our results(clockwise, 51%; anticlockwise, 49%; n = 47 droplets). This is incontrast to the observation that contractile actin bundles organizethe vortex structure with a certain direction of chirality in a circular

A

θ

5 min

27 min

11 min

θ

Center

15 min

0

10

20

30

0

10

20

30

-90 0 900

10

20

30

10

20

30

0

B

Cou

nts

(degree)

15 min

17 min

19 min

21 min

B

Fig. 3. Spatial confinement arranges microtubules in the vortex array. (A,Left) Time-lapse series of confocal images of microtubule bundles showingthe vortex formation process. Every image is the maximum projection imagealong the z axis (range, the equatorial plane of the droplet ± 20 μm). (Right)Each histogram shows angles of microtubule bundles (θ) measured in the leftimage. θ is defined as the angle of the bundles to the line connecting thedroplet center with the bundle midpoint, as shown in the image at 11 min(in this image, θ = −63°). (Scale bar, 30 μm.) (B) Time-lapse images of anelongating bundle near the boundary shown in the yellow box in A. Arrows,ends of bundles. (Scale bar, 20 μm.) In all microscopic images, dashed lines

indicate droplet boundaries, and 0 min indicates the timing of elevating thetemperature. [OG-Taxol] = 500 nM.

Suzuki et al. PNAS | March 14, 2017 | vol. 114 | no. 11 | 2925

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

16, 2

020

cell (31). In addition, the actin vortex did not show rotationalflow. However, these results suggest that vortex formationcould be one of the general features among extensile andcontractile systems.The flow direction occasionally changed (Figs. 2D and 5B, and

Movie S8). The mechanism of this directional change can bespeculated based on the observation that the assembly and dis-assembly of microtubule bundles continually occurred (Fig. 4B).The bundle density was so low that a few newly assembledbundles reaching the boundary may easily overcome the forcegenerated by preexisting bundles to change the rotational di-rection. In fact, with increasing Taxol concentration, the fre-quency of the directional change decreased, that is, therotational flow became more stable (Fig. 5C). Because Taxolstabilizes microtubules (i.e., the frequency of complete disas-sembly of microtubules triggered by catastrophe decreases withincreasing Taxol concentration; ref. 20), these results are con-sistent with our model.

DiscussionWe have demonstrated that, under suppression of dynein activ-ity, cytoplasmic microtubules were organized in extensile bun-dles, and these bundles induced directed cytoplasmic flow, thelength scale and timescale of which depended on the existence ofspatial confinement.Confinement of extracts in droplets induced submillimeter

length scale rotational flow, which was 10-fold larger than thevortex flows emerging in bulk extracts. The length scale differ-ence could be caused by mechanical interactions between thebundle and the droplet boundary. First, the microtubule bundlenetwork in the cytoplasm could be elastic enough to transmitmechanical force over the longer distance than the hydrody-namic interaction because microtubules in the cytoplasm are soshort (14.5 ± 6.4 μm, mean ± SD; n = 122 microtubules) thatthey can act as rigid rods (16) and are connected with nonmotorcross-linkers, such as MAPs, which enhance the elasticity of theoverall network. Accordingly, microtubule bundle elongationcan generate a pushing force against the overall microtubulebundle network once the elongating bundles reach the dropletboundary. Although hydrodynamic interactions of microtubulebundles may also contribute to the spontaneous alignment of bun-dles into the vortex pattern because extensile microtubule bundlescan be regarded as pusher-type motile units, the alignmentcannot be explained solely by the hydrodynamic interactions. Toorient the bundle with a length of 21.9 μm (the mean length ofthe bundles in droplets) by the hydrodynamic interactions fromthe angle θ = 0° to 60° (θ = 60° is the typical bundle angle in thevortex; Fig. 3A) before the bundle disassembly (bundle lifetime,20.9 ± 6.74 min; mean ± SD; Fig. 4D), the flow velocity gradientper micrometer should be larger than 0.05 min−1, which wascalculated under the assumption that the bundle is advectedat the same velocity of fluid flow. Therefore, the flow velocity inthe bulk extracts (0.36 ± 0.24 μm·min−1) is too slow to arrangethe bundles into the vortex pattern because the maximum flowvelocity gradient generated by the bulk flow was estimated asonly 0.033 min−1 (0.36 μm·min−1 × 2/21.9 μm), and it would takemore than 31.9 min to orient the bundle from θ = 0° to 60°.Thus, the mechanical interactions of microtubule bundles couldbe essential for the arrangement of the bundles into the largelength scale vortex pattern.Not only the length scale, but also the timescale of the rota-

tional flow is 10- to 100-fold longer than those of the vortex flowsemerging in bulk extracts. This timescale difference could beexplained by the positive-feedback loop (Fig. 5A); the fluid flowpromotes the alignment of the bundles, and the bundle align-ment accelerates the fluid flow. Moreover, once the microtubulebundles are arranged in the vortex array, the vortex structurewould be more easily maintained than the random network be-cause of the presence of mechanical interactions between mi-crotubule bundles. The vortex structure is primarily composed of

Bun

dle

lifet

ime

(min

)

180 min 182 min 190 min 194 min 196 minA

B C D50

0

25

0

10

5

Elon

gatio

n ra

te (µ

m m

in-1)

0

50

100

Bun

dle

leng

th (µ

m)

170 200Time (min)

230140

= 19n = 9n

Fig. 4. Microtubule bundles exhibit extensile behavior. (A) Time-lapse seriesof confocal images of microtubule bundles in a compressed droplet. Everyimage is the maximum projection image along the z axis. To examine thebehaviors of individual microtubule bundles in the steady state, we observedthem after 2–4 h from the droplet formation, which corresponded to thetime period when most droplets showed rotational flow. Arrows, both endsof the bundle. (Scale bar, 30 μm.) 0 min, timing of elevating the tempera-ture. (B) Typical time courses of the microtubule bundle length in thecompressed droplet. Asterisks and double daggers indicate the timing ofbundle disassembly and fusion to other bundles, respectively. (C and D)Distributions of the bundle elongation rate (C) and the bundle lifetime (D)(period between the assembly time and disassembly time) in compresseddroplets. Solid lines indicate the mean rate and lifetime.

Elongation

Network formation Rotational force generation by pushing

Bundle alignment near the boundary

Vortex formationand rotational flow

10 min 40 min60 180Time (min)

0

5

10

-5

-100 120

(deg

min

-1)

0

50

100

50 100150

200>250

% o

f dro

plet

s

Taxol: 50 nM 100 nM 500 nMNo rotational flow Unstable Stable

0

MT polymerizationand bundle formation

Phase 1 Phase 2 Phase 3 Phase 4 Phase 5

Droplet diameter (µm)

Symmetry breaking

ωx

0 50 100150

200>250

0 50 100150

200>250

A

B C

Fig. 5. Directional stability and mechanism of ro-tational cytoplasmic flow. (A) Schematic illustrationshowing the mechanism of the rotational flow.(B) Time projection of the beads (Left) and hωix (Right)with 50 nM Taxol. The lines and their surroundingareas represent means ± SDs, respectively. (Scale bar,30 μm.) (C) Distributions of stable rotational flow(blue), unstable rotational flow (the flow changingits direction oppositely at least once per 30 min; red),and no rotational flow (hωix,t < 0.1 deg·min−1; gray;Movie S9) at various droplet diameters. Each bar in-cludes 10–44 droplets.

2926 | www.pnas.org/cgi/doi/10.1073/pnas.1616001114 Suzuki et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

16, 2

020

parallel-arrayed bundles, and many cross-linkers can bridge theoverlapping regions simultaneously, stabilizing the vortex array.In contrast, in the case of randomly oriented arrays, cross-linkerscan bridge only the crossing points of bundles, and thus, therandom bundle network is easily remodeled by fluid flow and themechanical stress generated by bundle elongation. Accordingly,we concluded that not only the length scale, but also the time-scale difference is caused by the mechanical interaction.Both the rotational flow emerged in droplets and the vortex

flows emerged in bulk space resemble the flow patterns of bac-terial suspensions with and without spatial confinement (4, 5).However, the effects of the spatial confinement are essentiallydifferent. Although confinement increases the temporal stabilityof the bacteria vortex similar to our results, it does not change thelength scale of the vortex, that is, the vortex size of bacteria indroplets is comparable to the vortex size in bulk space (4, 5). Thedifference may be attributed to a lack of specific connectionsbetween bacteria, presumed by the bacterial motions during thecirculation in a droplet. Whereas the bacteria located near theboundary did not move, the others were smoothly swimming evenwhile contacting with those immotile bacteria (4). Compared withbacteria, the vortex-arrayed microtubule bundles rotated morelike an elastic structure. In summary, although further quantitativestudies with a combination of numerical simulations should beperformed to reveal the respective roles of hydrodynamic andmechanical interactions, our results suggest that microtubulenetworks use, in addition to hydrodynamic interaction, mechanical

interaction derived from microtubule cross-linkers (motor andnonmotor proteins) and the physical boundary to induce cyto-plasmic flow over larger length scale and timescale.

Materials and MethodsXenopus egg extracts were prepared as described previously (19). Allchemical reagents, proteins, and polyethylene glycol (PEG)-coated beads (32,33) were mixed with the extracts on ice before droplet formation. OregonGreen-labeled Taxol (100 nM, unless otherwise stated) and 2 μM cytochalasinD (for all experiments) were added to the extracts. To suppress dynein ac-tivity, p150-CC1 was added (800 nM, unless otherwise stated). The lipid–oilmixture was prepared as previously reported (34–36). The egg extracts wereadded to this mixture, and the droplets were then immediately formed bytapping on ice or at 16 °C. The extract-in-oil droplets were perfused into thechamber assembled with siliconized coverslips, which were spaced withdouble-faced tape pieces, and the chamber was sealed with Valap. All im-ages were acquired at 20 ± 1 °C and analyzed with LabVIEW and ImageJ.

ACKNOWLEDGMENTS. We gratefully acknowledge Y. Arai for providing aplug-in for ImageJ used for bead tracking. We thank M. Chiba, M. Tanabe,K. Matsuura, A. Hattori, M. Odaka, T. Kikuchi, N. Takahashi, Y. Nakata, andM. Iwamura for experimental assistance; H. Terazono for technical advice; andK. Yasuda for helpful comments. This work was supported by the ResearchFellowship for Young Scientists [DC1 (to K.S. and J.T.)], Grants for ExcellentGraduate Schools and Waseda University Grant for Special Research Projects(to K.S.), Grants-in-Aid for Young Scientists (B) and Scientific Research onInnovative Areas (to M.M.), and Grants-in-Aid for Specially PromotedResearch and Scientific Research (S) (to S.I.) from the Ministry of Education,Culture, Sports, Science and Technology of Japan.

1. Katz Y, Tunstrøm K, Ioannou CC, Huepe C, Couzin ID (2011) Inferring the structure anddynamics of interactions in schooling fish. Proc Natl Acad Sci USA 108(46):18720–18725.

2. Dombrowski C, Cisneros L, Chatkaew S, Goldstein RE, Kessler JO (2004) Self-concentrationand large-scale coherence in bacterial dynamics. Phys Rev Lett 93(9):098103.

3. Sokolov A, Aranson IS (2012) Physical properties of collective motion in suspensions ofbacteria. Phys Rev Lett 109(24):248109.

4. Wioland H, Woodhouse FG, Dunkel J, Kessler JO, Goldstein RE (2013) Confinementstabilizes a bacterial suspension into a spiral vortex. Phys Rev Lett 110(26):268102.

5. Lushi E, Wioland H, Goldstein RE (2014) Fluid flows created by swimming bacteria driveself-organization in confined suspensions. Proc Natl Acad Sci USA 111(27):9733–9738.

6. Drescher K, Dunkel J, Cisneros LH, Ganguly S, Goldstein RE (2011) Fluid dynamics andnoise in bacterial cell-cell and cell-surface scattering. Proc Natl Acad Sci USA 108(27):10940–10945.

7. Schroeder TE, Battaglia DE (1985) “Spiral asters” and cytoplasmic rotation in sea ur-chin eggs: Induction in Strongylocentrotus purpuratus eggs by elevated temperature.J Cell Biol 100(4):1056–1062.

8. Theurkauf WE (1994) Premature microtubule-dependent cytoplasmic streaming incappuccino and spire mutant oocytes. Science 265(5181):2093–2096.

9. Serbus LR, Cha BJ, Theurkauf WE, Saxton WM (2005) Dynein and the actin cytoskeletoncontrol kinesin-driven cytoplasmic streaming in Drosophila oocytes. Development132(16):3743–3752.

10. Ueda H, et al. (2010) Myosin-dependent endoplasmic reticulum motility and F-actinorganization in plant cells. Proc Natl Acad Sci USA 107(15):6894–6899.

11. Woodhouse FG, Goldstein RE (2012) Spontaneous circulation of confined active sus-pensions. Phys Rev Lett 109(16):168105.

12. Woodhouse FG, Goldstein RE (2013) Cytoplasmic streaming in plant cells emerges nat-urally by microfilament self-organization. Proc Natl Acad Sci USA 110(35):14132–14137.

13. Yi K, Rubinstein B, Li R (2013) Symmetry breaking and polarity establishment duringmouse oocyte maturation. Philos Trans R Soc Lond B Biol Sci 368(1629):20130002.

14. Goldstein RE, van de Meent JW (2015) A physical perspective on cytoplasmic streaming.Interface Focus 5(4):20150030.

15. Sanchez T, Chen DTN, DeCamp SJ, HeymannM, Dogic Z (2012) Spontaneous motion inhierarchically assembled active matter. Nature 491(7424):431–434.

16. Keller PJ, Pampaloni F, Lattanzi G, Stelzer EHK (2008) Three-dimensional microtubulebehavior in Xenopus egg extracts reveals four dynamic states and state-dependentelastic properties. Biophys J 95(3):1474–1486.

17. Gittes F, Mickey B, Nettleton J, Howard J (1993) Flexural rigidity of microtubules andactin filaments measured from thermal fluctuations in shape. J Cell Biol 120(4):923–934.

18. Isambert H, et al. (1995) Flexibility of actin filaments derived from thermal fluctua-tions. Effect of bound nucleotide, phalloidin, and muscle regulatory proteins. J BiolChem 270(19):11437–11444.

19. Desai A, Murray A, Mitchison TJ, Walczak CE (1999) The use of Xenopus egg extractsto study mitotic spindle assembly and function in vitro. Methods Cell Biol 61:385–412.

20. Derry WB, Wilson L, Jordan MA (1995) Substoichiometric binding of Taxol suppressesmicrotubule dynamics. Biochemistry 34(7):2203–2211.

21. Field CM, et al. (2011) Actin behavior in bulk cytoplasm is cell cycle regulated in earlyvertebrate embryos. J Cell Sci 124(Pt 12):2086–2095.

22. Pinot M, et al. (2012) Confinement induces actin flow in a meiotic cytoplasm. ProcNatl Acad Sci USA 109(29):11705–11710.

23. Verde F, Berrez JM, Antony C, Karsenti E (1991) Taxol-induced microtubule asters inmitotic extracts of Xenopus eggs: Requirement for phosphorylated factors and cy-toplasmic dynein. J Cell Biol 112(6):1177–1187.

24. Quintyne NJ, et al. (1999) Dynactin is required for microtubule anchoring at centro-somes. J Cell Biol 147(2):321–334.

25. Gaetz J, Kapoor TM (2004) Dynein/dynactin regulate metaphase spindle length bytargeting depolymerizing activities to spindle poles. J Cell Biol 166(4):465–471.

26. Keber FC, et al. (2014) Topology and dynamics of active nematic vesicles. Science345(6201):1135–1139.

27. Claessens MMAE, Tharmann R, Kroy K, Bausch AR (2006) Microstructure and visco-elasticity of confined semiflexible polymer networks. Nat Phys 2(3):186–189.

28. Harris P, Osborn M, Weber K (1980) A spiral array of microtubules in the fertilized seaurchin egg cortex examined by indirect immunofluorescence and electron micros-copy. Exp Cell Res 126(1):227–236.

29. Monteith CE, et al. (2016) A mechanism for cytoplasmic streaming: Kinesin-drivenalignment of microtubules and fast fluid flows. Biophys J 110(9):2053–2065.

30. Nédélec FJ, Surrey T, Maggs AC, Leibler S (1997) Self-organization of microtubulesand motors. Nature 389(6648):305–308.

31. Tee YH, et al. (2015) Cellular chirality arising from the self-organization of the actincytoskeleton. Nat Cell Biol 17(4):445–457.

32. Uchida K, et al. (2007) Creation of a mixed poly(ethylene glycol) tethered-chain surfacefor preventing the nonspecific adsorption of proteins and peptides. Biointerphases2(4):126–130.

33. Hermanson GT (2008) Bioconjugate Techniques (Elsevier, Amsterdam), 2nd Ed, pp215–223.

34. Miyazaki M, Chiba M, Eguchi H, Ohki T, Ishiwata S (2015) Cell-sized spherical con-finement induces the spontaneous formation of contractile actomyosin rings in vitro.Nat Cell Biol 17(4):480–489.

35. Miyazaki M, Chiba M, Ishiwata S (2015) Preparation of cell-sized water-in-oil droplets forin vitro reconstitution of biological processes in cellular compartments. Protoc Exch.Available at http://www.nature.com/protocolexchange/protocols/3815#/reagents. AccessedMarch 24, 2015.

36. Chiba M, Miyazaki M, Ishiwata S (2014) Quantitative analysis of the lamellarity of giantliposomes prepared by the inverted emulsion method. Biophys J 107(2):346–354.

37. King SJ, Brown CL, Maier KC, Quintyne NJ, Schroer TA (2003) Analysis of the dynein-dynactin interaction in vitro and in vivo. Mol Biol Cell 14(12):5089–5097.

38. Thielicke W, Stamhuis EJ (2014) PIVlab—towards user-friendly, affordable and accu-rate digital particle image velocimetry in MATLAB. J Open Res Softw 2(1):e30.

39. Mitchison TJ, Nguyen P, Coughlin M, Groen AC (2013) Self-organization of stabilizedmicrotubules by both spindle and midzone mechanisms in Xenopus egg cytosol. MolBiol Cell 24(10):1559–1573.

40. Newmeyer DD, Lucocq JM, Bürglin TR, De Robertis EM (1986) Assembly in vitro ofnuclei active in nuclear protein transport: ATP is required for nucleoplasmin accu-mulation. EMBO J 5(3):501–510.

41. Lockhart A, Cross RA (1996) Kinetics and motility of the Eg5 microtubule motor.Biochemistry 35(7):2365–2373.

42. Crevel IMTC, Lockhart A, Cross RA (1996) Weak and strong states of kinesin and ncd.J Mol Biol 257(1):66–76.

43. Foster PJ, Fürthauer S, Shelley MJ, Needleman DJ (2015) Active contraction ofmicrotubule networks. eLife 4:e10837.

Suzuki et al. PNAS | March 14, 2017 | vol. 114 | no. 11 | 2927

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

16, 2

020