Temporary increase in plasma membrane tensioncoordinates the activation of exocytosis andcontraction during cell spreadingNils C. Gauthiera,b, Marc Antoine Fardina,c, Pere Roca-Cusachsa,d, and Michael P. Sheetza,b,1

aDepartment of Biological Sciences, Fairchild Building, Columbia University, New York, NY 10027; bMechanobiology Institute, National University ofSingapore, Singapore T-Lab, Singapore 117411; cLaboratoire Matiere et Systemes Complexes, Centre National de la Recherche Scientifique, Unité Mixtede Recherche 7057, Universite Paris Diderot, 10 rue Alice Domont et Leonie Duquet, 75205 Paris Cedex 13, France; and dInstitute for Bioengineering ofCatalonia and University of Barcelona, c/ Baldiri Reixac 10-12, 08028 Barcelona, Spain

Edited by Alexander Mogilner, University of California, Davis, CA, and accepted by the Editorial Board June 30, 2011 (received for review April 12, 2011)

Cell migration and spreading involve the coordination of mem-brane trafficking, actomyosin contraction, and modifications toplasmamembrane tension and area. The biochemical or biophysicalbasis for this coordination is however unknown. In this study, weshow that during cell spreading, lamellipodia protrusion flattensplasmamembrane folds and blebs and, once the plasmamembranearea is depleted, there is a temporary increase inmembrane tensionby over twofold that is followed by activation of exocytosis andmyosin contraction. Further, an artificial increase in plasma mem-brane tension stopped lamellipodia protrusion and activated anexocytotic burst. Subsequent decrease in tension restored spread-ing with activation of contraction. Conversely, blebbistatin inhibi-tion of actomyosin contraction resulted in an even greater increasein plasmamembrane tension and exocytosis activation. This spatio-temporal synchronization indicates that membrane tension is thesignal that coordinates membrane trafficking, actomyosin contrac-tion, and plasma membrane area change. We suggest that cellsuse plasma membrane tension as a global physical parameter tocontrol cell motility.

In several animal cell types (1–4), and even in plants (5), mod-ifications in plasma membrane (PM) tension induced by osmo-

larity change can control PMarea. Regulation of PMarea is linkedto membrane trafficking, with endocytosis and exocytosis, respec-tively, decreasing or increasing the PM area. Moreover, mem-brane trafficking appears to be amajor factor in cell migration (6).

Membrane tension also controls the dynamics of importantfunctions (3, 4), such as, endocytosis, membrane edge extension,and membrane resealing (7, 8). Recently, modeling of cell migra-tion has confirmed that membrane tension provides the feedbackcontrol that fits best with the observed motile behavior (9, 10).

During fibroblast spreading, there is a dramatic increase in PMarea of 50% through the activation of exocytosis of a recyclingcompartment (11). Because spreading also correlates with adecrease in apparent PM tension, the cause of this exocytosisis unclear (11). To understand this phenomenon, it is particularlyuseful to study isotropic fibroblast spreading, which is a rapid typeof spreading with abrupt transitions from one motile phase toanother (12–14). The early phase of spreading (P1) involves therapid extension of lamellipodia circumferentially with only weakcontractions and a very homogenous edge velocity (13, 15). Whenthe spreading cells reach about half of their final spread area,there is a transition to a contractile phase of spreading (P2) inwhich local cycles of protrusion–contraction activate the forma-tion of early adhesions (13–15).

Experimentally, this transition can be easily and primarilydefined by decreases in edge velocity, associated with the pro-gressive appearance of a more heterogeneous edge behavior(i.e., protruding and contracting regions alternate along the cellcircumference) (13–15). Gradual reduction in the edge velocitycauses a progressive decrease in spread area growth rate untilthe cell reaches its final spread area. The robust nature of this

transition enables us to link changes in exocytosis with changesin the motility phase.

Results and DiscussionDuring the initial phase of spreading, P1, we observed a verylow level of exocytosis using total internal reflection fluorescencemicroscopy (TIR-FM) of cells labeled with a fluorescent proteinlocalized in the Golgi apparatus (Fig. 1 A–E and Movie S1) orloaded before spreading with the membrane dye FM1-43 (11)(Fig. 1 F–H andMovie S2). However, at the beginning of the con-tractile phase, P2, there was a burst in exocytosis of both labels(Fig. 1 D, E, and H, and Movies S1 and S2). The onset of thecontractile phase was characterized by a clear decrease in theedge velocity (change in color on the velocity map in Fig. 1 Cand G), and more edge heterogeneity with local cycles of protru-sion–retraction and some ruffling (Fig. 1F, and brightfield record-ing in Movie S2). These results indicated that increases inexocytosis through constitutive secretion (Golgi) or the recyclingpathway (FM1-43) were remarkably linked to the slowing of edgeextension and activation of contraction at the start of P2. Thesefindings raised two questions: What was the source of membranethat enabled round cells to spread during P1, and what mechan-ism activated both exocytosis and contraction simultaneously?

To determine how the PMwas distributed during P1, we labeledthe PMwith FM1-43 just before spreading and maintained the dyein the medium during spreading. Thus, when unlabeled vesiclesfrom any internal compartment fused with the PM, they were im-mediately labeled with FM1-43 and the overall cell fluorescenceincreased (11). Throughout P1, cell fluorescence was constant, in-dicating that no exocytosis occurred (Fig. 2 A–C and Movie S3).However, there was a sharp increase in total fluorescence at thebeginning of P2, indicating that exocytosis was activated when cellsstarted to contract (Fig. 2C and Fig. S1). These results were inagreement with Golgi exocytosis and dye release studies (Fig. 1),which raised the following question: What was the source of themembrane that enabled P1 spreading without membrane additionthrough exocytosis? Careful observation of early P1 revealed thatcells generally (85%, n ¼ 31 cells from 12 independent experi-ments) had one or more very bright fluorescent regions (Fig. 2A,red dashed circle, and other cells analyzed in Figs. S1 and S2, andMovies S3 and S4). The fluorescence in the bright regions de-

Author contributions: N.C.G. and M.P.S. designed research; N.C.G. performed research;M.A.F. contributed new reagents/analytic tools; N.C.G. and P.R.-C. analyzed data; andN.C.G. and M.P.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. A.M. is a guest editor invited by the EditorialBoard.

See Commentary on page 14379.1To whom correspondence should be addressed. E-mail: ms2001@columbia.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1105845108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1105845108 PNAS ∣ August 30, 2011 ∣ vol. 108 ∣ no. 35 ∣ 14467–14472

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

9, 2

020

creased over time until the cell reached P2 (Fig. 2D and Figs. S1and S2). Because there was no exocytosis, there appeared to be apreexisting PM “buffer” that unfolded during P1 to enable lamel-lipodial protrusion and cell flattening. When the fluorescence inthe bright regions and the rest of cell was quantified over time, aloss in fluorescence of the PM buffer was equaled by the gain influorescence of the rest of the cell, confirming that the membranearea of the buffer provided the membrane needed for lamellipo-dial protrusion (Fig. 2E and Fig. S2). In other cases, we observed

bleb retraction before the P1–P2 transition (Movie S2) that mayhave provided additional membrane. To clearly identify whichmembrane compartment formed the buffer, we used time-lapse3D confocal microscopy of FM1-43 tagged plasma membraneto determine the distribution of membrane in a spreading cell overtime (Fig. 2F and Movie S5). Strikingly, we confirmed that duringP1, blebs and membrane folds dissipated. When all available mem-brane folds appeared depleted, lamellipodia extension sloweddown, in association with the appearance of membrane ruffles,

C

B

A

D E

F G H

Fig. 1. Exocytosis and contraction activations aresynchronized during spreading. (A) One exampleof a typical fibroblast cell analyzed for its exocytic be-havior (TIR-FM) during spreading. (see alsoMovie S1)(Scale bar: 10 μm.) (B) Example of single Golgi vesicleexocytosis analyzed by TIR-FM. (C) Motility map ofthe cell. The cell perimeter is plotted as a functionof time and the color range shows the local protru-sive or retractile behavior of the edge. The mapclearly shows the transition between the fast non-contractile spreading P1 (first 2–3 min) and the con-tractile spreading P2 (dashed line). (D) Spread areaand exocytic events per minute plotted as a functionof time. (E) Same results but the numbers of exocyticevents perminute are plotted as a function of spreadarea to outline the sharp transition between P1 andP2. The number of exocytic events was fewer thanone per minute before phase transition but in-creased to 13� 3.2 events per minute after 10 minof spreading (n ¼ 4 cells). (F) Brightfield images ofthe cell analyzed in G and H (see also Movie S2).The arrow shows the loss of edge cohesionwhen cellenters P2. (Scale bar: 5 μm.) (G) Motility map of thecell. (H) Analysis of the fluorescence intensity of thecells in the TIR-FM field during spreading (FM1-43 se-cretion by exocytosis) plotted as a function of spreadarea. This experimentwas reproduced two times andthe three cells presenting the same type of motilitypresented the same exocytic behavior.

A

B C

D

F

EFig. 2. Plasma membrane area depletion precedesexocytosis and contraction activation. (A) Time se-quence of a representative cell during spreading ana-lyzed with DIC (Upper) and FM1-43 epifluorescence(Lower). The dashed box outlines themembrane buf-fer region depicted inD. The dashed circle representsthe membrane buffer region analyzed in E. Theimages are extracted from Movie S3. (Scale bar:10 μm.) (B) Motility map of the cell depicted in A.The dashed line represents the transition betweenP1 and P2. (C) FM1-43 fluorescence intensity repre-senting membrane exocytosis at each time point(10 s) is plotted versus the spread area to outlinethe sharp transition between P1 and P2. All the cellswith a clear P1–P2 transition also had a clear andsharp exocytosis activation (n ¼ 31 cells from 12 inde-pendent experiments). (D) Time sequence (10 s) ofthe dashed box d showing the membrane buffer de-pletion during P1. (E) Fluorescence intensity analysisof the different parts of the cell depicted in A show-ing that the apparent constant membrane area ob-served during P1 (whole cell fluorescence, black) isdue to the unfolding of a membrane buffer (bufferregion, red) that provides excess membrane area tothe rest of the cell (whole cell–buffer, blue). Note thatthe fluorescence intensity lost in the buffer region(−0.52 a:u:) is similar to the gain in the rest of thecell (þ0.54 a:u:). (F) Time-lapse 3D reconstructionof the PM during spreading using FM1-43 (see alsoMovie S5). Arrowhead points to the membrane foldsand white arrows to the blebs. Folds and blebs disap-pear before the cell enters P2 phase, where contrac-tion signs (ruffling) appear at the edge (blue arrows).

14468 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1105845108 Gauthier et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

9, 2

020

signs of the beginning of the contractile phase of spreading (14,16). Thus, we found that folds and blebs are the source of PM areathat enables rapid lamellipodia protrusion in P1. When spreadingcells arrived at the limit of the PM area, there was activation ofexocytosis to provide extra membrane area for further spreading(11) and myosin contraction to promote early adhesion site forma-tion (14, 16). Interestingly, the membrane reservoir in P1 appearsto be different than caveolea, which are small plasma membraneinvaginations at the surface of the cells, recently described to dis-assemble under membrane mechanical stress (17). During spread-ing, two types of buffer may participate, folds and blebs during P1,and, maybe, caveolin disassembly during P2. Indeed we previouslyobserved that a glycosylphosphatidylinositol (GPI)-anchoredprotein recycling pathway caused the increase in membrane areaduring spreading, and caveolae have been shown to contain GPI-anchored proteins. However, it seems unlikely that the total mem-brane area increases during spreading (about 50% increase in areabetween nonspread and fully spread cells) is primarily providedby caveolea. Indeed caveolea are very small structures, about100 nm in diameter, and not normally present in the large numbersneeded to account for the increase in area.

What signaled to the cell that all PM area was depleted? Onepossibility was that cells were activated by a mechanical signal, suchas a transient increase in apparent PM tension due to the suddendepletion of the membrane folds. Changes in plasma membranetension were monitored through membrane tethers that werepulled either by a glass pipette or an optically trapped bead. Bothapproaches led to similar results. With the pipette system, wefrequently observed a narrowing of the tether at the beginningof P2, sometimes associated with pearling (Fig. 3A, Fig. S3, andMovies S6, S7, and S8). As previously described (14, 16), the be-ginning of the contractile phase was clearly identified in this experi-ment by the appearance of waves, or periodic contractions, at theleading edge (14, 16), moving backward toward the cell center(Fig. 3C andMovie S6). The pearling was followed by tether break-age and immediate retraction into the cell edge (Fig. 3B, Fig. S4,and Movie S8). It is well known that increased membrane tensionwill lead to the thinning and, ultimately, to the breakage of a tether(18–20), but the pipette system does not allow direct force mea-surement. To directly estimate the membrane tension throughthe measurement of tether force, we used optical tweezers. Tetherforce was constant during P1 (Fig. 3D and E, step I and Movie S9)with an average value of 13 pN (�3 pN SD, n ¼ 3).When the edgestarted to slow down and the first sign of contraction appeared(waves or ruffling), marking the beginning of P2, we observed aprogressive increase in tether force (Fig. 3E, step II). The tetherforce reached a peak 65% higher than the initial force (22 pN� 1SD, n ¼ 3). Because membrane tension is proportional to thesquare of the tether force (3), there was about a 2.5-fold increasein membrane tension. After the peak (Fig. 3E, step II), tether forcedropped when the edge slightly retracted due to contraction(Fig. 3E, step III), and stabilized (Fig. 3E, step IV). This stabiliza-tion correlated with the edge starting to spread again, (Fig. 3E, stepIV), a situation similar to step I, but at a lower level of membranetension. Rapidly, and before releasing the trap, a new slowdown ofthe edge was observed (Fig. 3E, step V) associated with a new in-crease in membrane tension. Both methods showed an increase inmembrane tension at the P1–P2 transition as the edge was slowingdown. Further spreading depended upon the subsequent reductionin tension. Interestingly, the membrane tension seems to be modu-lated cyclically. We already reported that membrane tension isglobally decreasing during spreading (14, 16). Combined with theseobservations, we propose that the decrease in membrane tension isprobably occurring through repetitive cycles of stabilization-increase-decrease (steps I-II-III), with each new stabilization oc-curring at lower membrane tension. The edge of the cell, at thetransition, is also behaving cyclically with periodic contractions

A

C

D

E

B

Fig. 3. The PM tension increases as the cell starts to contract. (A) A re-presentative cell visualized by DIC microscopy. The tether is the thin tubeof membrane between the pipette and the edge of the cell. The dashedline represents the portion of the edge analyzed by the kymograph in C.The image is extracted from Movie S6. (Scale bar: 10 μm.) (B) Timesequence showing tether breakage at the transition between P1 andP2. (C) Kymograph of the edge outline in A showing the periodic contrac-tion which characterizes the beginning of P2. The total time represents612 s. (D) Cell morphology in P1 (Left) and P2 (Right) during laser tweezersexperiment. The dashed line represents the portion of the edge analyzedby the kymograph in E. Dark arrowhead point to the tether, dark arrow tothe trap bead. White arrowhead points to the wave of material movingbackward observable in DIC and also observable in E at the back of theleading edge, sign of the contraction. The white arrows point to two ruf-fles, also a clear sign of contraction activation. (E) Composite figure oftether force (graph) and kymograph of the cell edge (blue line) at the P1to P2 transition. The spike in force at the beginning represents the forceneeded to extract the tether from the edge. After stabilization (step I), thetether force increased (step II), decreased (step III) to stabilize again at alower value than previously (step I compared to step IV), and increasedagain when the edge started to slow down a second time (step V). Atthis point, the bead was released from the trap to verify that the tetherwas still there.

Gauthier et al. PNAS ∣ August 30, 2011 ∣ vol. 108 ∣ no. 35 ∣ 14469

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

9, 2

020

(14, 16) (Fig. 3 C and E). If we estimate the time required for thetension to increase (step II, 21 s� 3.6 SD, n ¼ 3), this duration isremarkably compatible with the duration of one cycle of periodiccontraction that was previously reported to range between 18 and30 s (with an average of 24 s), with the edge protrusion accountingfor about three-fourths of the cycle duration (comparable to step IIin Fig. 3E) and retraction occurring during the remaining one-fourth of the time (14, 16) (step III in Fig. 3E). Thus, it is reason-able to think that each cycle of periodic contraction is correlatedwith cycles of membrane tension variations during the transitionbetween P1 and P2.

An alternative hypothesis was that the change in membranetension was the result (and not the cause) of the increased con-tractility. To test this hypothesis, we added the Myosin II inhibi-tor, blebbistatin. As previously described (14, 16), Myosin IIinhibition altered P2 by decreasing the magnitude of the periodiccontractions and overall cell contractility. However, we observedthat neither the activation of exocytosis nor the increase in PMtension was affected by Myosin II inhibition (Fig. S4). Inhibitedcells spread as well as control cells in P1. When the edge velocitydecreased, exocytosis simultaneously increased (Fig. S4 A and B).P2 was also characterized in this case by lateral propagation ofprotrusive waves without the clear cycle of protrusion–contrac-tion (Fig. S4A compared with Fig. 2B). Periodic contractions wereless defined (Fig. S4E), but tether breakage still occurred at thebeginning of P2 (Fig. S4D andMovies S10 and S11). Remarkably,blebbistatin inhibited the pulling of new tethers with the pipettelater in P2 (Movie S12), indicating that under myosin inhibition,membrane tension was kept high well into P2. With the opticaltrapped beads, we observed that tether force increased dramati-cally at the P1–P2 transition in blebbistatin treated cells(Movie S13). Indeed, we were unable to accurately measure thetether force because, in over half of the cases, beads were pulledout of the trap at the transition (n ¼ 12, force of the trap was70 pN). Moreover, tether force in P1 was also surprisingly highin blebbistatin treated cells (30 pN� 10 SD, n ¼ 12). Thus, themembrane tension at the transition increased over 10-fold in cir-cular cells after myosin inhibition compared to control cells, anddespite an already very high tension in P1, membrane exocytosiswas normally activated at the beginning of P2.

We reported previously that myosin-II-inhibited cells werespreading to a larger area (21). A careful analysis of those resultsshowed that myosin inhibition does not affect the beginning ofspreading (P1). At the end of P1, at around 1;000 μm2 of spreadarea, myosin-II-inhibited cells started to spread significantly lar-ger and faster than control cells (more than 30% after 15 min ofspreading). However, during that phase, even for myosin-inhib-ited cells, spreading speed was progressively reduced comparedto P1. We also reported previously that at a similar thresholdvalue of 1;000 μm2 of spread area, spreading cells started to in-crease the plasma membrane area by exocytosis (11). In view ofthe current results, we propose now that, in the control case, thePM tension is alleviated in P2 by both exocytosis, increasing mem-brane area, and myosin contraction pulling actin filaments awayfrom the edge. However, when myosin II is inhibited and actinfilaments are no longer pulled away from the edge, there is anincreased lamellipodia actin polymerization pressure against theplasma membrane, and, in consequence, an overspreading and anabnormal increase in membrane tension.

If a rise in membrane tension was the activation event forexocytosis and contraction, then increasing tension by an alterna-tive method such as hypotonic shock might activate the P1 to P2transition. Hypotonic shock caused a dramatic, but reversibleincrease in PM tension (1–4). After exposure of cells to a 50%decrease in tonicity, we observed that cells immediately stoppedlamellipodia extension (Fig. 4 A and B and Movie S14). Incontrast, control cells treated with an isotonic shock continuedto spread (Fig. S5). In addition, hypotonically treated cells

immediately activated exocytosis, strongly suggesting that the in-crease in membrane tension activated exocytosis (Fig. 4C). Afterisotonic media was restored, P2 began immediately and celllamellipodia spread with clear cycles of protrusion–contraction,indicating myosin activation (Fig. 4A). Thus, an early increase inmembrane tension by hypotonic shock caused a premature P1 toP2 transition, indicating that a rise in membrane tension wasnecessary and sufficient to induce the change in motility mode.

A

B

C

Fig. 4. Hypotonically induced increase and subsequent decrease in mem-brane tension induces P1–P2 transition with activation of exocytosis andcontraction. (A) Kymograph analysis of a DIC movie for a representative cellin P1 that is exposed to hypotonic medium for30 s (ringer 0.5×, lamellipodiaprotrusion stops) and subsequently restored to isotonic ringer (spreadingresumes with a contractile lamellipodia presenting cycles of protrusion–contraction). The cell analyzed is presented in the bottom left corner ofMovie S14 with three other cells with similar behavior. (B) Plot of the cell areaversus time showing the strong effect of the temporary increase in plasmamembrane tension on the overall spread area for the cell depicted in A.When PM tension increases (arrowhead) the cell stops spreading; spreadingis resumed during the isotonic recovery phase (arrow). (C) The increase inmembrane tension activates the exocytosis (lower graph, blue arrow).FM1-43 fluorescence is plotted as a function of time for a representative con-trol cell (Upper) and for a cell exposed to hypotonic media (Lower). The PMarea addition induced by hypotonicity was 37� 7% of the original PM area in15 s (SEM, n ¼ 6 cells).

14470 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1105845108 Gauthier et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

9, 2

020

If actin polymerization in the lamellipodia was limited by thePM area because of an increase in membrane tension, actin poly-merization activity should behave accordingly during spreading.In the initial P1, where there was excess membrane area and lowmyosin contraction, no clear reward flow of actin was observableby TIR-FM in control- or blebbistation-treated cells (Fig. 5 A andB, and Movies S15 and S16), similar to previous reports withGFP-alpha-actinin (14, 16). When control cells started P2, thelamellipodia actin buckled and retracted cyclically from the edge(Fig. 5A, arrows) as previously described during periodic contrac-tions (14, 16). Myosin II was primarily responsible for the shortedge retraction occurring during each cycle of periodic contrac-tion, and was also thought to be the main cause of the buckling(16). When myosin-inhibited cells entered P2, they spread to alarger area and had greatly decreased periodic contractions(Fig. 5B and Movie S16) as previously observed (14, 16, 21),but there was clear actin buckling at the leading edge (Fig. 5Band Movie S16). If increased membrane tension was causing ahigher local density of polymerized actin in the lamellipodia,leading to buckling, then alternative methods of increasing ten-sion such as hypotonic shock should have a similar effect. Asnoted previously (Fig. 4), an artificial increase in membrane ten-sion inhibited lamellipodia extension but we also observed an in-crease in actin density at the cell edge (Fig. 5C and Movie S17).When tension was restored, lamellipodia actin polymerization re-sumed (Fig. 5C). Thus, membrane tension can directly exert afeedback control on the actin polymerization, the driving forcefor lamellipodia extension.

Membrane tension provides a physical means of coordinatingthe activation of the multiple cellular activities that define P2 andwe suggest that a transient increase in membrane tension at theP1 to P2 transition provides the signal for the transition.

It is possible to estimate if the membrane tension changescould cause the changes in extension rates. The relationshipbetween membrane tension and the tether force has been de-scribed as T ¼ F2∕ð8π2BÞ, where F is the tether force and B andT are the effective bending stiffness and effective membrane ten-sion, respectively (18, 19, 22). The adjective “effective” takes intoaccount the addition of a cytoskeletal adhesion component (3).At the P1–P2 transition, cells reduce their overall lamellipodialprotrusion rate by one-half from approximately 70 to 35 nm∕s(13, 15). Previous studies showed that the amount of pressureP applied from atomic force microscopy tips needed to reducelamellipodial protrusions by half on a length scale of approxi-mately 1 μm was of the order of 1 nN∕μm2 (23). If we considerthat the membrane is mainly curved along the lamellipodiumheight, with characteristic radius of curvature r of approximately100 nm, we can calculate the increase in membrane tensionT caused by this applied pressure using Laplace’s law as T ¼Pr ¼ 10−4 N∕m. Therefore, if the transition from P1 to P2 isindeed caused by the stalling of actin polymerization due to thedepletion of membrane reservoirs, it should be accompanied byan increase in membrane tension of this magnitude. Given thatB is around 3 × 10−19 Nm (3) and that the measured tether forceduring P1 is 13 pN, an increase in tension of 10−4 N∕m gives atether force increase of 37 pN, in the same order of magnitudeas the observed change in experiments. These estimates, coupledwith our observations, strongly suggest that actin polymerizationforces can cause the increase in tension observed.

In myosin-II-inhibited cells, the tension in P1 is higher than incontrol cells. However, cells can still spread efficiently and re-spond to the increase in membrane tension. One explanation forthis phenomenon is that the polymerization of the lamellipodialactin in P1 is a process that can adjust the number of polymeriz-ing filaments to counteract the force exerted by the membranetension. Once a threshold is reached to push the membrane for-ward, lamellipodia extension occurs at a constant pace during P1.For any cells, myosin-inhibited or not, the critical point of entry

into P1 is the activation of the actin polymerization process. Onceactivated, the cell is at a steady state during P1, as long as plasmamembrane area from folds is available. At the transition into P2,the increase in membrane load against actin polymerization sim-ply destabilizes the preexisting steady state in the system and ac-tivates the transition to a new phase of behavior, P2, as observed.

C

P1 P2 P1 “P2”

P1 P2 P1 “P2”Control BlebbistatinA B

60 s

5µm

60 s

5µm

Ringer0.5X

Ringer1X

Ringer1X

Ringer0.5X

Ringer1X

Ringer1X

60 s5µ

m

Fig. 5. Actin dynamics correlated with spreading behavior. (A) Characteristickymograph analysis of GFP-actin in TIR-FM (upper picture, see alsoMovie S16).Arrows point to the periodic contractions easily identified by the actin buck-ling at the beginning of P2. Total time is 350 s. Lower pictures present the actinmorphology at the leading edge in P1 and P2. Arrow points again to the buck-ling. (B) Same analysis for a characteristic blebbistatin treated cell (see alsoMovie S16). (C) Same analysis as inA and B but for a cell with an hypotonicallyinduced increase and subsequent decrease in membrane tension (see alsoMovie S17). Arrow points to the actin polymerization pushing forward theleading edge after isotonicity restoration. Lower pictures present the typicalactin morphology at the leading edge during the experiment.

Gauthier et al. PNAS ∣ August 30, 2011 ∣ vol. 108 ∣ no. 35 ∣ 14471

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

9, 2

020

One surprising aspect of our results is that membrane tension inmyosin-inhibited cells is higher in P1 than control cells. However,membrane exocytosis stills occurs normally at the transition intoP2, when the tension spikes. How is the tension then linked to exo-cytosis activation in that case? Because the major component ofthe membrane tension is the adhesion between the membrane andthe cytoskeleton, we suggest that the absence of contractile forceis linked to an increase in membrane-cytoskeleton adhesion. Analternative view is that there is an alteration in membrane traffick-ing, because exocytosis during spreading involves two pathways: arecycling pathway containing GPI-anchored proteins, responsiblefor the increase in membrane area, and a constitutive secretionfrom the Golgi apparatus (11). The GPI recycling pathway couldbe very sensitive to membrane tension increase and even be di-rectly exocytosed because of membrane tension variations (24).Golgi secretion, however, could be activated through other signals.The global actin remodeling occurring at the transition and obser-vable in TIR-FM, even under myosin inhibition, could be involved(Movies S15 and S16). Indeed, actin dynamics seems to be inte-grally linked to Golgi dynamics (25). Moreover, in myosin-inhib-ited cells, entry into P2 seems to occur at a larger spread area thanin control cells. We currently understand relatively little about theproteins involved in membrane-cytoskeleton adhesion and howtheir activation is linked to myosin activity and membrane traffic.

ConclusionIn this study, we find that membrane tension correlates with theactivation of exocytosis and myosin contraction. We think thatour results, in this respect, bring the physical properties of themem-brane to the heart of understanding coordination in cell motility.Motile functions rely on local biochemical reactions between mo-lecules such as actin filament assembly, but those functions need tobe coordinated at the whole cell level in complex behaviors like cellmigration. Physical forces provide a robust and simple way to coor-dinate biochemical processes to produce integrated cell functions.

Materials and MethodsCell Culture Transfection and Reagents. Immortalized mouse embryonic fibro-blasts were cultured and transfected with YFP-Golgi (Clontech), as previouslydescribed (11).

Tether Pulling with Micropipette. Borosilicate glass pipettes (o.d. 1.0 mm; i.d.,0.58 mm; 10-cm length; Sutter Instrument) were pulled to obtain an outsidediameter below 1 μm. After acid wash (nitric acid 20%), the pipette tip wascoated with concanavalin A (1 mg∕mL, in PBS) for 24 h. Pipettes were

mounted on a micromanipulator and observed with differential interferencecontrast (DIC) microscopy. The tethers were pulled by moving the stage.

Tether Pulling with Optical Trap. As described previously (11), the cells wereviewed by a video-enhanced DIC microscope (Carl Zeiss MicroImaging)equippedwith laser optical tweezers. Carboxylated, 1-μm-diameter latex beads(Polysciences)were coatedwith concanavalin A by use of carbodiimide linkage.Beads were attached to the cell membrane and pulled away from the cellsurface to form a membrane tether. The tether force was calculated fromthe displacement of the bead from the center of the laser tweezers duringtether formationandthecalibrationof the trap.Membranetensionwas relatedto the square of the tether force as deduced from the lipid bilayer systems.

Spreading Assay. Cells were detached with trypsin/EDTA. Trypsin was then in-activated with soybean trypsin inhibitor (1 mg∕mL in DMEM). The cells werethen suspended in Ringer solution [150 mM NaCl, 5 mM KCl, 1 mM CaCl2,1 mM MgCl2, 20 mM Hepes, and 2 g∕L glucose (pH 7.4)] and incubatedfor at least 30 min before plating on coated surfaces. Ringer 0.5× contains1 mM CaCl2 and MgCl2. Cover glasses were acid-washed, treated with hex-amethyldisilazane, and coated with fibronectin (10 μg∕mL, 2 h in PBS at37 °C, Roche Diagnostics GmbH).

FM1-43 Before Spreading and FM1-43 During Spreading Assays. These assayswere used exactly as we previously described (11).

Confocal Microscopy. Cells were analyzed using a PerkinElmer Spinning Diskcoupled to an Olympus IX81 microscope using a 100× oil objective (N.A. 1.40).The cells where analyzed in the same condition as previously reported for theFM1-43 during spreading assay.

Image Analysis and Statistics. The free-access ImageJ software was used. Themotility map was drawn from a manual mask extracted from the DIC with anImageJ plug-in developed by Marc Antoine Fardin and inspired from whatwas previously developed in our lab (13). The plug-in is freely available uponrequest. The fluorescence level was quantified as follows: An area containingone cell was selected from the 16 bit epifluorescence or 8 bit TIR-FM (Fig. 1)recording. The intensity of an identical area in the same field but withoutcells was subtracted from that value to obtain only the fluorescence intensityof the cell. The same protocol was followed for each time point.

ACKNOWLEDGMENTS. We thank all the members of the Sheetz lab for theirsupport and help, particularly Nicolas Biais, SimonMoore, Olivier Rossier, andAdam Walpert. We also thank Pascal Hersen, Steven Wolf, and Baixue Zhengfor fruitful discussions (Mechanobiology Institute, National University ofSingapore, Singapore). P.R.-C. was supported by a Marie Curie InternationalOutgoing Fellowship within the Seventh European Community FrameworkProgramme. This work was supported by National Institutes of Health GrantGM064561 (to M.P.S.) and the Mechanobiology Institute, National Universityof Singapore, Singapore.

1. Apodaca G (2002)Modulation of membrane traffic by mechanical stimuli.Am J PhysiolRenal Physiol 282:F179–190.

2. Morris CE, Homann U (2001) Cell surface area regulation and membrane tension.J Membr Biol 179:79–102.

3. Sheetz MP (2001) Cell control by membrane-cytoskeleton adhesion. Nat Rev Mol CellBiol 2:392–396.

4. Sheetz MP, Sable JE, Dobereiner HG (2006) Continuous membrane-cytoskeletonadhesion requires continuous accommodation to lipid and cytoskeleton dynamics.Annu Rev Biophys Biomol Struct 35:417–434.

5. Kell A, Glaser RW (1993) On the mechanical and dynamic properties of plant cellmembranes: Their role in growth, direct gene transfer and protoplast fusion. J TheorBiol 160:41–62.

6. Jones MC, Caswell PT, Norman JC (2006) Endocytic recycling pathways: Emergingregulators of cell migration. Curr Opin Cell Biol 18:549–557.

7. Raucher D, Sheetz MP (2000) Cell spreading and lamellipodial extension rate is regu-lated by membrane tension. J Cell Biol 148:127–136.

8. Togo T, Krasieva TB, Steinhardt RA (2000) A decrease in membrane tension precedessuccessful cell-membrane repair. Mol Biol Cell 11:4339–4346.

9. Neves SR, et al. (2008) Cell shape and negative links in regulatory motifs togethercontrol spatial information flow in signaling networks. Cell 133:666–680.

10. Keren K, et al. (2008) Mechanism of shape determination in motile cells. Nature453:475–480.

11. Gauthier NC, Rossier OM, Mathur A, Hone JC, Sheetz MP (2009) Plasma membranearea increases with spread area by exocytosis of a GPI anchored protein compartment.Mol Biol Cell 20:3261–3272.

12. Dobereiner HG, Dubin-Thaler BJ, Giannone G, Sheetz MP (2005) Force sensing andgeneration in cell phases: Analyses of complex functions. J Appl Physiol 98:1542–1546.

13. Dubin-Thaler BJ, Giannone G, Dobereiner HG, Sheetz MP (2004) Nanometer analysisof cell spreading on matrix-coated surfaces reveals two distinct cell states and STEPs.Biophys J 86:1794–1806.

14. Giannone G, et al. (2004) Periodic lamellipodial contractions correlate with rearwardactin waves. Cell 116:431–443.

15. Dubin-Thaler BJ, et al. (2008) Quantification of cell edge velocities and traction forcesreveals distinct motility modules during cell spreading. PLoS ONE 3:e3735.

16. Giannone G, et al. (2007) Lamellipodial actin mechanically links myosin activity withadhesion-site formation. Cell 128:561–575.

17. Sinha B, et al. (2011) Cells respond to mechanical stress by rapid disassembly of caveo-lae. Cell 144:402–413.

18. Hochmuth FM, Shao JY, Dai J, Sheetz MP (1996) Deformation and flow of membraneinto tethers extracted from neuronal growth cones. Biophys J 70:358–369.

19. Hochmuth RM, Evans EA (1982) Extensional flow of erythrocyte membrane from cellbody to elastic tether. I. Analysis. Biophys J 39:71–81.

20. Hochmuth RM, Wiles HC, Evans EA, McCown JT (1982) Extensional flow of erythrocytemembrane from cell body to elastic tether. II. Experiment. Biophys J 39:83–89.

21. Cai Y, et al. Cytoskeletal coherence requires myosin-IIA contractility. J Cell Sci123:413–423.

22. Hochmuth RM (1982) Solid and liquid behavior of red cell membrane. Annu RevBiophys Bioeng 11:43–55.

23. Prass M, Jacobson K, Mogilner A, Radmacher M (2006) Direct measurement of thelamellipodial protrusive force in a migrating cell. J Cell Biol 174:767–772.

24. StaykovaM, Holmes DP, Read C, Stone HA (2011) Mechanics of surface area regulationin cells examined with confined lipid membranes. Proc Natl Acad Sci USA108:9084–9088.

25. Zilberman Y, et al. (2011) Involvement of the Rho-mDia1 pathway in the regulation ofGolgi complex architecture and dynamics. Mol Biol Cell, 10.1091/mbc.E11-01-0007.

14472 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1105845108 Gauthier et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

9, 2

020