Main

Cytoskeletal systems based on actomyosin and microtubules cooperate in various cellular functions, yet they also have opposing effects. Disrupting microtubule function can enhance cellular contractility, increase cell tension1 and induce formation of stress fibres and focal adhesions. Steady-state cultured cells maintain equilibrium between these systems, which shifts during the cell cycle and in response to various cues. During mitosis, cells downregulate stress fibres and reorganize microtubules to form the mitotic spindle, whereas formation of the cleavage furrow and cytokinesis require repositioning of myosin II by microtubular signals2. Another interruption of steady state occurs during wound healing, where wound-edge cells initiate rapid migration3. Cells in culture normally maintain high contractility and exert forces on rigid substrates using myosin II to contract actin fibres and induce maturation of adhesion complexes to focal adhesions. Myosin II is required for F-actin anterograde flow in the cell body and retrograde flow in the lamella4, but it is absent from the lamellipodium5,6. F-actin, myosin II, microtubules and focal adhesions interact dynamically to mediate efficient cell migration7,8,9. Microinjection of antibodies against brush-border myosin II could suppress stress fibres and enhance lamellae and locomotion of chicken fibroblasts10.

Mammalian cells have three isoforms of non-muscle myosin II, termed IIA, IIB and IIC, encoded by three different genes. Despite considerable homology, these isoforms exhibit differences in enzymatic properties11, subcellular localization and tissue expression patterns. Genetic ablations of the different isoforms demonstrate that they are not functionally redundant. Myosin IIA-knockout mice die at embryonic day (E) 6.5–7 and fail to organize normal germ layers12, whereas myosin IIB-null mice die between E14.5 and birth due to brain and heart defects13. Evolutionary divergence of function between these myosin isoforms would permit distinct roles at specific developmental stages14 while retaining some capacity for partial compensation. Myosin IIB is involved in stabilizing normal cell polarity15, whereas myosin IIA is associated with Rho kinase-dependent functions (including organization of stress fibres and focal adhesions). Depletion of myosin IIA from human cancer cells enhanced rates of wound closure, but suppressed net single-cell motility16. Despite the wealth of knowledge about myosin II in general, the unique functions of each isoform are still under investigation. We have used models of myosin IIA deficiency to characterize its specific roles in contractility, maintenance of cytoskeletal equilibrium and the resulting regulation of cell migration.

Results

Spatial regulation of myosin IIA and microtubules during a change of state

Wound healing is a perturbation of steady-state conditions where fibroblasts are induced to migrate into a wound gap. A traditional model of scratch-wounding monolayers of human foreskin fibroblasts (HFFs) was used to examine for changes in myosin IIA and microtubules. In confluent cultures, and shortly after in vitro wounding when cells are non-migratory, myosin IIA and microtubules overlap extensively in distribution — both were observed close to membrane edges. However, 8 h later in migrating cells, microtubules protruded into the leading lamellae, whereas myosin IIA accumulated farther behind (Fig. 1). Genetic and RNA interference (RNAi) approaches were used to examine this spatial switching of microtubule and actomyosin localization, and the roles of myosin IIA.

Figure 1: The steady-state balance between myosin IIA and microtubules is altered after in vitro wounding.
figure 1

Confluent monolayer of non-migrating HFFs before (control) and immediately after inducing a scratch wound (0 h) showing overlapping distribution of microtubules labelled by α-tubulin (green) and myosin IIA (red). As cells start to migrate into the wound gap (grey arrows) after 8 h, microtubules expand into the active leading edge (white arrow), whereas myosin IIA distribution is limited to areas behind the leading lamella. The scale bar represents 20 μm.

Full size image

Myosin IIA mediates cell tension and contractility but suppresses cell migration

The non-muscle myosin heavy chain IIA gene, Myh9, was genetically ablated in RW4 mouse embryonic stem (ES) cells to generate myosin IIA-knockout cells, which also express myosin IIB but not myosin IIC12. As we previously described12, myosin IIA-knockout ES cells failed to cluster and instead formed flat, dispersing colonies with defects in cell–cell adhesion. Individual myosin IIA-deficient cells were flatter, and 20–50% of the cells displayed extremely elongated trailing tails with delayed retraction (Fig. 2a), even though cell motility was rapid (see below). F-actin staining revealed a dramatic decrease in actin stress-fibre formation in these myosin IIA-null cells and a loss of focal adhesions, as indicated by vinculin staining (Fig. 2b). These effects were mimicked by 60–65% knockdown of myosin IIA in HFFs using small interfering RNA (siRNA; Fig. 3a and see Supplementary Information, Fig. S1). Knockdown fibroblasts often showed elongated tails, consistent with a dominant role for myosin IIA in cellular contractility, as absence of the IIB isoform did not result in this phenotype (Fig. 2a). Actomyosin-dependent cellular contractility is important for wound contraction by fibroblasts17 and can be quantified by in vitro fibrin gel contraction assays. As shown in Fig. 3c, reducing myosin IIA levels by siRNA markedly reduced the capacity of HFFs plated on top of polymerized fibrin to contract the fibrin, consistent with a key role for myosin IIA in mediating cell contractility.

Figure 2: Genetic myosin IIA deficiency impairs formation of actin stress fibres and focal adhesions.
figure 2

Actin stress fibres are stained with Alexa Fluor 488–phalloidin (green). (a) Wild-type (WT), myosin IIA-null (IIA−/−) and myosin IIB-null (IIB−/−) mouse ES cells. A myosin IIA-null cell exhibiting elongated trailing tails (white arrowheads) is shown in the lower right panel. (b) Vinculin (stained red, arrowheads) or actin (green) + vinculin staining of focal adhesions in wild-type and myosin IIA-null ES cells. The scale bars represent 20 μm.

Full size image
Figure 3: Myosin IIA depletion impairs contractility but increases migration rates.
figure 3

(a) Primary HFFs treated with myosin IIA siRNA or control siRNA; fluorescence confocal image of F-actin stained by Alexa Fluor 488–phalloidin. (b) Inhibition of myosin II ATPase in wild-type ES cells by 25 μΜ blebbistatin mimics the morphological effect of myosin IIA ablation; F-actin stained by Alexa Fluor 488–phalloidin. (c) Depletion of myosin IIA by siRNA (50 nM) inhibits fibroblast contraction of fibrin gels stained with Ponceau S. The bar graph quantifies diameter of the corresponding contracted gels (n = 20, P = 0.0018). (d, e) Migration velocities increase on gene ablation. Wild-type versus myosin IIA-null ES cells (d; n = 19, P <0.0001), or myosin IIA siRNA knockdown in HFF (e; n = 35, P = 0.0002). (f) Increased velocity of migration is induced by 25 μΜ blebbistatin in MCF10A breast epithelial cells (n = 37 and 31; P <0.0001), HFF (n = 37 and 25; P = 0.0031), wild-type ES cells (WT ES; n = 28 and 26; P <0.0001) and primary embryonic mouse fibroblasts (MEF; n = 28 and 30; P <0.0001). Con, control; bleb, blebbistatin. All error bars represent s.e.m. The scale bars represent 20 μm in a and b.

Full size image

Rather than disrupting cell migration, myosin IIA ablation in ES cells resulted in a striking 2–3-fold increased migration velocity according to time-lapse microscopy (Fig. 3d). These cells displayed unusually large lamellipodia and lamellae with extensive ruffling activity, unlike wild-type or myosin IIB-null cells (Fig. 2a, b). siRNA against the non-muscle myosin heavy chaing IIA gene MYH9 (hereafter termed myosin IIA siRNA) also increased lamellipodial ruffling in HFFs, concomitant with loss of stress fibres and focal adhesions (Fig. 3a and data not shown), and doubled the migration velocity (Fig. 3e). None of these effects were observed in myosin IIB-null ES cells or myosin IIB siRNA-transfected cells, including no increases in cell migration (see below and Supplementary Information, Fig. S2).

We compared the effects of these specific myosin IIA ablations with more general pharmacological inhibitors of myosins. Treatment with ML-7, a broad-spectrum inhibitor of myosin light-chain kinase, resulted in cell flattening with no effect on ruffling. In contrast, treating ES cells or HFFs with blebbistatin, a specific inhibitor of myosin II ATPase activity2, induced membrane ruffling and decreased contractility similar to myosin IIA gene ablation and siRNA knockdown (Fig. 3b). Blebbistatin treatment induced substantial increases in migration rates of HFFs, mouse ES cells, human MCF10A breast epithelial cells and primary mouse embryonic fibroblasts (Fig. 3f). Despite intrinsic differences in initial migration rates between cell types, they all exhibited substantial stimulation of cell migration after blebbistatin treatment. Interestingly, even though migration velocity increased in myosin IIA-deficient cells, overall directional persistence decreased (see Supplementary Information, Fig. S3).

As we, and others, have reported, Rac1 can switch cells between directional and random modes of migration18,19. The enhanced membrane ruffling, spreading and random migration of myosin IIA-deficient cells suggested the involvement of Rac1.

Roles of Rac and Rac–GEF in myosin IIA-deficient cells

Small GTPases have major roles in regulating cell tension through the cytoskeleton and in cell migration. Membrane ruffling and spreading are often associated with increased Rac activity, whereas contractility is associated with Rho. Both require specific guanine exchange factors (GEFs) for activation. Pulldown assays for GTP–Rac revealed that myosin IIA-null ES cells have elevated levels of active Rac (Fig. 4b). Wild-type ES cells treated with blebbistatin also showed increased Rac activation (data not shown). Immunofluorescence staining of Rac1 in myosin IIA-null ES cells showed enhanced localization at the lamellar edges (Fig. 4a). Rac1 activation was accompanied by colocalization with its specific GEF, Tiam1, at the leading edge (Fig. 4a), as confirmed by line scans of immunofluorescence staining intensity (see Supplementary Information, Fig. S4).

Figure 4: Rac activation in myosin IIA-deficient cells.
figure 4

(a) Confocal images of fluorescence immunostaining for Rac1 (green) and Tiam1 (red) show that Rac and Tiam1 colocalize at the leading membrane edge in myosin IIA-null ES cells, but not in wild-type ES cells. (b) Rac activation levels are increased in myosin IIA-null cells (solid bars) compared with wild-type cells (hatched bars) without altered activation of RhoA or Cdc42, and with no change in total Rac. Active Rac, Rho and Cdc42 levels were determined by pulldown assays with PAK–PBD (p21-activated kinase binding domain)-coupled beads or Rhotekin-coupled beads, and compared with their total protein levels in total cell lysates using densitometric quantification by Scion Image software. (c) Myosin II deficiency-induced increase in migration rate is blocked by inhibiting Rac. Migration velocities of wild-type ES cells treated with the Rac inhibitor NSC23766 (NSC), blebbistatin (bleb) or blebbistatin + NSC23766 were assessed by video time-lapse microscopy analysis of migration (n = 19, 20, 17 and 20 cells, respectively; P <0.001). (d) Rac inhibition abolishes membrane ruffling in myosin IIA-deficient cells. F-actin staining by Alexa Fluor 488–phalloidin (green) of non-treated (con) HFFs, blebbistatin-treated (bleb) and blebbistatin treatment following Rac inhibition by NSC23766 (bleb + NSC). The error bars represent s.e.m. The scale bars represent 20 μm in a and d.

Full size image

The contribution of Rac activation to the induction of cell migration and ruffling was examined in myosin-deficient cells by two approaches to reduce Rac activity in HFFs: Rac1 knockdown by siRNA; and inhibition of Rac activation by the Rac-specific inhibitor, NSC23766. The first approach reduces both total Rac expression and Rac activity levels, whereas the second approach blocks Rac interaction with GEFs, but not interactions with other effector molecules such as PAK1. Rac1 siRNA transfection followed by blebbistatin treatment abolished blebbistatin-induced ruffling, and the cells became elongated. Inhibition of Rac activation by NSC23766 produced a similar reversal of the blebbistatin phenotype (Fig. 4d). Moreover, siRNA knockdown of the Rac GEF Tiam1 also reversed this morphological phenotype; knockdown also resulted in reduced acetylated tubulin, suggesting a role in microtubule stabilization (Fig. 6d, e). Reducing Rac activation also blocked the stimulation of cell migration after loss of myosin IIA function. Rac inhibition in wild-type ES cells by NSC23766 suppressed migration velocity from a 2.8-fold increase to 0.6-fold (Fig. 4c). Control cells treated with NSC23766 alone showed no significant effects on migration, indicating that partial reduction in Rac levels did not per se alter migration velocity.

Myosin IIA regulates microtubule distribution and dynamics

The balance of actomyosin and microtubule activities is thought to regulate direction of migration20 and contractility1,21. We first examined whether attenuation of actomyosin function affected microtubule distribution or function. Immunostaining for α-tubulin revealed increased total staining intensity in myosin IIA-null cells compared with wild-type cells, with enhanced tubulin localization in the extended membrane lamellae (Fig. 5a) and a twofold increase in the total numbers of microtubules near the membrane edge of lamellae (134 ± 16 in wild-type versus 251 ± 10 in myosin IIA-null ES cells; mean ± s.e.m., n = 12 cells each). In myosin IIA-null ES cells, the majority of the microtubules stained positively for acetylated tubulin (Fig. 5a), a posttranslational modification associated with microtubule stabilization. Blebbistatin suppression of myosin II in wild-type ES cells showed similar effects as early as 20 min after treatment (Fig. 5b), representing altered localization with no change in α-tubulin protein levels (see Supplementary Information, Fig. S1). We tested whether increasing Rac activation could mimic these effects, but no changes in distribution, staining intensity and/or numbers, or bending of microtubules were observed in cells transfected with either wild-type or constitutively activated Rac (data not shown).

Figure 5: Microtubule expansion in myosin IIA-deficient cells.
figure 5

(a) Immunostaining of α-tubulin (red) and acetylated tubulin (acetyl, red) in wild-type and myosin IIA-null (IIA−/−) ES cells. Actin was stained with Alexa Fluor 488–phalloidin (green). (b) Increased number of microtubules and expansion into membrane ruffles 20 min after addition of blebbistatin to wild-type ES cells; F-actin (Alexa Fluor 488–phalloidin, green) and acetylated tubulin (red). (c) Microtubule disruption by 100 ng ml−1 vinblastine (vin) or 660 nM nocodazole (nocod) reduces migration rates in myosin IIA-null cells (P <0.0001, n = 20, 29 and 21 cells, respectively) but not significantly in wild-type cells (P = 0.11, n = 21 and 19). (d) COS-7 cells transfected with mCherry–tubulin in combination with EGFP–myosin IIA or EGFP–myosin IIB. Microtubule dynamics were observed using time-lapse fluorescence confocal microscopy over a period of 6 min. Arrowheads indicate fixed points of reference near selected microtubules. Dashed lines in the top panels delineate the cell edge. (e) Dwell time of individual microtubules located in a zone of cytoplasm within 10 μm of the membrane edge of lamellae was determined by analysing data from time-lapse movies (n = 38, 35 and 29). (f) Reduced velocity of COS-7 cell migration after expression of EGFP–myosin IIA but not EGFP–myosin IIB (n = 38, 35 and 29 cells, respectively). The error bars represent s.e.m. The scale bars represent 20 μm in a and b, and 5 μm in d.

Full size image

To directly examine the link between myosin IIA and microtubule dynamics, time-lapse fluorescence confocal microscopy was used to analyse COS-7 cells, which are normally devoid of myosin IIA. Cells were cotransfected with mCherry–tubulin and GFP–myosin IIA or IIB. Cells transfected with mCherry–tubulin in the presence or absence of myosin IIA recapitulated the distinctive localization patterns described above (Fig. 5d and data not shown). Time-lapse imaging revealed striking changes in the dynamics of microtubules (see Supplementary Information, Movies 1 and 2). GFP–myosin IIA-expressing cells showed rapid, frequent changes in individual microtubule length, localization and stability, accompanied by vigorous retrograde movement of myosin IIA clusters. In contrast, untransfected or GFP–myosin IIB transfected COS-7 cells exhibited stable microtubules that showed much less change in overall location or length over the filming period (6 min). Myosin IIB was present at substantially lower levels in lamellae, displayed considerably slower retrograde movement than myosin IIA (0.60 ± 0.12 μm min−1 versus 1.99 ± 0.26 μm min−1, mean ± s.e.m., n = 34 and 28 observations, respectively; P <0.0001), and associated more with actin stress fibres (see Supplementary Information, Movie 2 and data not shown). The average dwell time of individual mCherry-labelled microtubules within a 10 μm zone from the cell edge, as determined by time-lapse confocal microscopy, dropped 4.3-fold in cells expressing GFP–myosin IIA, with no change observed in GFP–myosin IIB transfectants, indicating a substantial increase in microtubule dynamics in the presence of myosin IIA (Fig. 5e). GFP–myosin IIA- but not GFP–myosin IIB-transfected COS-7 cells showed reduced migration rates, consistent with our other data establishing myosin IIA regulation of cell migration (Fig. 5f). The striking ability of microtubules to undergo retrograde translocation in the presence of myosin IIA compared with myosin IIB was confirmed in HFFs transfectants that normally express both isoforms (see Supplementary Information, Movies 3 and 4). These cells showed normal microtubule dynamics, but GFP–myosin IIA and GFP–myosin IIB demonstrated different patterns of retrograde flow.

To determine whether microtubule polymerization was required for the observed stimulation of cell migration, myosin IIA-null ES cells were treated with the microtubule-disrupting agents nocodazole or vinblastine. Both drugs reduced migration velocities of myosin IIA-null cells by 48% to 71%, respectively, with minimal effects on migration of wild-type cells (17% decrease; Fig. 5c). However, consistent with previous reports, microtubule disruption reduced migration polarity, resulting in a loss of persistence from a D/T (the distance between starting and end points of migration path divided by the total distance traversed by the cell) value of 0.14 ± 0.02 to 0.06 ± 0.01. In summary, a functional polymerized microtubule system is essential for the increase in overall migration velocity in myosin IIA-deficient cells.

Myosin IIB can restore contractility when microtubules are suppressed

Microtubule-disrupting agents, such as nocodazole, stimulate fibroblast contractility1. Myosin IIA-null cells with severe impairment in forming actin stress fibres and focal adhesions were able to generate stress fibres and resume formation of focal adhesions 2 h after treatment with nocodazole (Fig. 6a), with increasing effect by 16 h (Fig. 6b). This result suggests that a component capable of promptly generating a contractile phenotype was still present. Immunolocalization of myosin IIB in nocodazole-treated cells showed enhanced colocalization of myosin IIB with the newly-formed stress fibres (Fig. 6a). Blebbistatin, which inhibits both myosin isoforms, blocked this nocodazole-induced rescue of stress fibres (Fig. 6a). Moreover, myosin IIA-null ES cells unable to contract fibrin gels regained the ability to do so after treatment with nocodazole (Fig. 6c). Knocking down myosin IIA by siRNA did not result in any compensatory increase in the levels of myosin IIB (see Supplementary Information, Fig. S1), indicating that redistribution of myosin IIB, but not any increase in total protein, is involved in the restoration of contractility. Taken together, these results show that myosin IIB can potentially compensate for the loss of myosin IIA in generating contractility, but only when microtubules are depolymerized, demonstrating overall regulation by microtubules.

Figure 6: Disruption of microtubules restores contractility of myosin IIA-deficient cells.
figure 6

(a) Myosin IIB staining (red) and vinculin (blue) in myosin IIA-null cells after nocodazole treatment for 2 h show colocalization with the newly formed stress fibres (actin, green). Both stress fibre formation and localization of myosin IIB are abolished if the cells are pretreated with blebbistatin. (b) Nocodazole-treated myosin IIA-null cells at 16 h regain actin stress fibres and focal adhesions as indicated by staining for vinculin (blue), F-actin (phalloidin, green), and α-tubulin (red). (c) Contraction of fibrin gels by wild-type versus myosin IIA-null cells plated on top of polymerized fibrin with and without 660 nM nocodazole quantified by image analysis of the upper contracted layer (n = 30 for each condition). (d) Myosin IIA-null ES cells were transfected with 25 nM siRNA directed against Tiam1 and were stained after 48 h for Tiam1 (red), Rac1 (blue) and α-tubulin (green). These siRNA-treated cells showed reversal of the ruffling phenotype. (e) Tiam1 siRNA-knockdown myosin IIA-null ES cells stained for acetylated tubulin (red), actin (green) and Tiam1 (blue) show reduction in acetylated tubulin. The scale bars represent 20 μm.

Full size image

Myosin IIB cannot compensate for myosin IIA in restraining migration

Because myosin IIB could compensate for myosin IIA deficiency in mediating contractility, we examined whether it could similarly suppress migration rates. Migration rates of a fibroblast cell line isolated from myosin IIB knockout mice at E13 were compared with wild-type fibroblasts. Myosin IIB-null ES cells and myosin IIB siRNA-knockdown human fibroblasts were also compared with their wild-type counterparts. In all cases, myosin IIB-deficient cells did not migrate at higher velocities (see Supplementary Information, Fig. S2 and data not shown). In addition, no statistically significant changes in migration rates of myosin IIA-null cells were detected when the remaining myosin IIB isoform was inhibited by blebbistatin, although a further reduction in stress fibres was observed (see Supplementary Information, Fig. S2) and even more pronounced defects in tail retraction. These results suggest that myosin IIB can support contractility under certain conditions, but unlike myosin IIA, does not seem to act as a negative regulator of cell migration.

Role of microtubule motor molecules

To determine whether the microtubule-dependent stimulation of migration in myosin IIA-deficient cells involves microtubule motor function, we examined the involvement of dynein and kinesin in this process by chemical inhibition. The general dynein inhibitor erythro-9-3-(2-hydroxynonyl) adenine (EHNA) had no effect on the blebbistatin-induced stimulation of fibroblast migration (data not shown). In contrast, inhibitors of the kinesin Eg5 affected migration. Treatment of human fibroblasts with monastrol, a cell-permeable Eg5 inhibitor, did not inhibit cell migration in control cells, yet it abolished the blebbistatin-induced increase in cell migration (Fig. 7a). Monastrol also inhibited formation of blebbistatin-induced membrane ruffles (Fig. 7d). Similarly, transfection of human fibroblasts with Eg5-specific siRNA suppressed the blebbistatin-stimulated migration (Fig. 7b). To further confirm the unexpected involvement of Eg5 kinesin, several additional Eg5-specific inhibitors were tested: Eg5 inhibitor II (NSC83265, S-trityl-l-cysteine), Eg5 inhibitor III (dimethylenastron)22 and trans-HR22C16 (ref. 23). All Eg5 kinesin inhibitors reduced the velocity of migration of cells lacking myosin IIA, whereas no such effect was observed in wild-type cells (Fig. 7c). Consequently, Eg5 is selectively involved in the microtubule-dependent acceleration of cell migration in the absence of myosin IIA.

Figure 7: Inhibition of Eg5 kinesin abolishes the increase in migration and membrane ruffling of myosin II-deficient cells.
figure 7

(a, b) HFFs treated with monastrol (a) or transfected with Eg5 siRNA (b) fail to show increased migration in response to blebbistatin treatment, in contrast with non-treated or mock-transfected cells (control versus 25 nM Eg5 siRNA, P >0.05 and control versus 100 nM Eg5 siRNA, P >0.05, n = 33, 26 and 31; control blebbistatin versus blebbistatin plus 25 nM Eg5 siRNA, P <0.0001 or versus blebbistatin plus 100 nM siRNA, P <0.0001, n = 26, 32 and 29). (c) The Eg5 inhibitors monastrol (mon, 50 μM; P <0.001), Eg5 inhibitor II (EgII, 2 μM; P <0.001), Eg5 inhibitor III (EgIII, 5 μM; P <0.0001) and trans HR22C16 (HR, 5 μM; P <0.05) reduce migration in myosin IIA-null ES cells, but not in wild-type ES cells (P=0.13, non-significant; n = 23, 26, 23 and 23 cells, respectively). (d) HFFs treated with monastrol are not morphologically different from their non-treated counterparts, but blebbistatin-induced membrane ruffling and microtubule expansion were blocked by monastrol treatment. (e) Schematic representation of myosin IIA regulatory crosstalk linking actomyosin and microtubule cytoskeletal systems. Myosin IIA plays a central role in cellular contractility and promoting microtubule dynamics dependent on kinesin Eg5. Ablation or inhibition of myosin IIA results in loss of stress fibres, focal adhesions and ability to contract fibrin gels, as well as an increase in microtubule stabilization with more acetylation, translocation of Tiam1 and Rac1 to the plasma membrane and activation of Rac, and stimulation of membrane ruffling and random motility. Myosin IIB can compensate for myosin IIA in contractile processes, but only after microtubule disruption. Other studies (red line) suggest that organization of the actin cytoskeleton can also promote cell contractility, and that enhanced adhesion to substrates through focal adhesions can reduce cell motility. Dashed boxes indicate cellular functions affected by myosin IIA; green line from myosin IIB to stress fibres and focal adhesions indicates dependence on microtubule status. The error bars represent s.e.m. in a–c. The scale bars represent 20 μm.

Full size image

Discussion

Cellular actomyosin and microtubule systems maintain a dynamic balance essential for cell contractility, polarization and migration. Initiation of migration in resting cells shifts this balance to restrict myosin IIA distribution and allows microtubule expansion into the leading lamella. We used cells genetically or functionally deficient in myosin IIA to decipher its specific roles in regulating the equilibrium between these cytoskeletal systems and in contractility, microtubule dynamics and cell migration. Myosin IIA-deficient cells displayed not only pronounced contractile defects, but also dramatic increases in non-directional migration, accompanied by exaggerated non-polarized membrane ruffling associated with stabilization of microtubules in lamellae. This phenotype was mimicked by blebbistatin, which specifically inhibits myosin II, but not by the MLCK inhibitor ML-7, which reduces cell contractility and tension. Our findings indicate that myosin IIA continually restrains random cell migration under normal conditions in multiple cell types by crosstalk with the microtubule system.

We demonstrated that Rac activation is necessary, but not by itself sufficient, for the exaggerated ruffling and increased migration of myosin IIA-deficient cells. The Rac GEF Tiam1 is translocated to the membrane of myosin IIA-null cells, colocalizing with Rac at membrane edges. Suppression of Tiam1 by siRNA knockdown reverted the myosin IIA-null morphological phenotype, consistent with its known roles in Rac-dependent ruffling24. Our observation that the absence of myosin IIA can signal back to activate Rac is also consistent with known effects of cellular tension on Rac activity25.

We documented an unusual expansion of microtubules into the overly active lamellae of myosin IIA-deficient cells, with increased numbers and levels of acetylated tubulin. Direct tracking of individual microtubules by time-lapse fluorescence microscopy revealed that myosin IIA promotes microtubule dynamics, and deficiency of this isoform results in microtubule stabilization near the cell edge. Myosin II inhibition by blebbistatin mimicked this regulation of microtubule assembly. Conversely, depolymerization of microtubules reversed the effects on lamellar expansion, ruffling, Rac–Tiam1 localization and migration in myosin IIA-null cells. Microtubule disruption also permitted colocalization of myosin IIB with new stress fibres and restored the ability of cells to contract fibrin gels, demonstrating: the capability of myosin IIB to compensate for IIA in formation of stress fibres; and the restricting effect of microtubules on myosin IIB in mediating contractility. Interestingly, it was previously reported that myosin IIB-null fibroblasts are impaired in their ability to contract three-dimensional collagen gels26. Previous reports have shown coupling between anterograde and/or retrograde flow of microtubules and actin fibres27 and myosin II functions in actin-bundle severing and regulation of retrograde flow28. Our data indicate that myosin IIA undergoes vigorous retrograde movement, whereas myosin IIB does not.

We propose that myosin IIA has a critical function in coupling the actomyosin and microtubule cytoskeletal systems (Fig. 7e). In the absence of myosin IIA, the increased stability of microtubules explains their accumulation in lamellae. Stabilized microtubules could drive the excessive ruffling either by exerting force against the membrane29 or particularly by enhancing actin polymerization at the leading edge through activation of Rac by its GEF Tiam1 — which is known to associate with microtubules30 and has been shown by us to be essential for the ruffling phenotype. This global Rac-interdependent regulation of microtubule stabilization in lamellae is distinct from the phenomenon of local rapid, non-GTPase modulation of microtubule assembly by local cell tensile stress31.

The unexpected involvement of the Eg5 kinesin, also known as mitosis-associated kinesin32, suggests yet another level of regulation. Multiple Eg5 inhibitors and Eg5 RNAi inhibited the increased migration of myosin IIA-deficient cells. Eg5, unlike cargo kinesins, reportedly moves microtubules against each other and may actively contribute to pushing microtubules forward. These data suggest a new role for this kinesin, distinct from its role in mitosis, and consistent with a previously reported role in regulating microtubules in neuronal dendrites and axons33. Involvement of a kinesin in crosstalk between the microtubule and actin cytoskeletons is plausible, as kinesin-1 has been shown to regulate focal adhesion size34.

In summary, we propose that myosin IIA mediates cell contractility and provides coupling between microtubules and the actomyosin system, which under normal conditions promotes microtubule dynamics and negatively regulates membrane ruffling and cell migration. Suppression or loss of myosin IIA function leads to microtubule stabilization and expansion into lamellae, extensive membrane activity and increased migration velocity in a variety of cell types. This novel function underscores the complexity of myosin–microtubule regulatory interactions in cellular contractility and migration.

Methods

Cell lines.

RW4 ES cells and the associated myosin IIA-knockout, described elsewhere12, were cultured in DMEM with 10% ES-approved FBS supplemented with leukaemia inhibitory factor (LIF; Chemicon, Temecula, CA), HEPES, MEM amino acids, 100 U ml−1 penicillin, 100 μg ml−1 streptomycin, and 1:1000 β-mercaptoethanol (GIBCO/Invitrogen, Carlsbad, CA). Primary HFFs were a gift from Susan Yamada (NIDCR, NIH, Bethesda, MD) and were used at passages 5–18. Human fibroblasts, primary embryonic mouse fibroblasts derived from 12-day-old embryos35, and COS-7 cells from American Type Culture Collection (ATCC, Manassas, VA) were cultured at 37 °C in 10% CO2 in DMEM containing 10% FBS (Hyclone, Logan, UT), 100 U ml−1 penicillin and 100 μg ml−1 streptomycin. The MCF10A human breast epithelial cell line from ATCC was cultured in DMEM–Ham's F12 medium supplemented with 20 ng ml−1 EGF, 10 μg ml−1 insulin, 500 ng ml−1 hydrocortisone, and 5% equine serum. For time-lapse fluorescence imaging, a phenol red-free version of DMEM was used.

Plasmids.

An expression plasmid containing α-tubulin with mCherry fluorescence tag was kindly provided by J. Hammer and J. Martina (NHLBI, Bethesda, MD) and Roger Tsien (UC San Diego, San Diego, CA), and the Rac1 cDNA (wild-type Rac) plasmid was from J. Silvio Gutkind (NIDCR, Bethesda, MD). The constitutively activated Rac1 construct pRK VSV RacQ61L (RacQ61L) with a VSV epitope tag36 and human NMHC-IIA and NMHC-IIB fused to EGFP26 were described previously.

Antibodies and reagents.

Sources were: mouse anti-α-tubulin and mouse anti-acetylated tubulin (Sigma, St Louis, MO), rabbit anti-NMHC-IIA C-terminal peptide and rabbit anti-NMHC-IIB C-terminal peptide37, mouse anti Rac (Upstate Cell Signaling Solutions, Bellerica, MA), rabbit anti-Tiam1 (Santa Cruz Biotechnology, Santa Cruz, CA), mouse anti-vinculin (BD Biosciences Pharmingen, San Jose, CA), Alexa Fluor 488 phalloidin (Molecular Probes/Invitrogen, Carlsbad, CA), blebbistatin (−/−; Toronto Chemicals, North York, Canada; or Calbiochem/EMD, San Diego, CA), monastrol (Sigma), Eg5 inhibitor II, Eg5 inhibitor III, trans-HR22C16, nocodazole (Calbiochem/EMD) and vinblastine (Sigma). NSC23766 was designed, synthesized and provided by the Drug Synthesis and Chemistry Branch (National Cancer Institute, Bethesda, MD) and is commercially available (Calbiochem/EMD).

Transfection, pulldown assays and immunoblotting.

HFFs were transfected with 25–100 nM siRNA using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) and OptiMEM medium (GIBCO/Invitrogen) without antibiotics according to the manufacturer's recommendations. The control siRNA was Dharmacon siGLO RISC-free siRNA. Pulldown assays for Rho, Rac and Cdc42 were performed as recommended by the manufacturer (Cytoskeleton, Denver, CO), and the released proteins were resolved on 8–16% gradient gels (Novex/Invitrogen, Carlsbad, CA). Blots were probed with the indicated antibodies, followed by the appropriate secondary horseradish peroxidase-conjugated antibodies (Amersham/GE Healthcare, Piscataway, NJ; and Sigma). Immunoblots were visualized using the ECL system and Hyperfilm X-ray film (Amersham Biosciences) and intensity was quantified by Scion Image software (Scion Corp., Frederick, MD). Non-muscle myosin IIA, myosin IIB and Rac1 siRNAs were obtained as SMARTpool preparations from Dharmacon. Eg5 kinesin siRNA was obtained from Santa Cruz Biotechnology. Plasmids encoding mCherry–α-tubulin with or without EGFP-tagged myosin IIA or IIB were transfected into COS-7 cells or HFFs by electroporation using a Bio-Rad Gene Pulsar TM at 170V, 960 μFd with external capacitance, with a time constant of 17–22 in 0.4 cm gap cuvettes. Cells were used for experiments 24–72 h after transfection.

Laser-scanning confocal microscopy.

Cells for immunofluorescence microscopy analysis were plated on 20 mm glass coverslips (Carolina Biological Supply Company, Burlington, NC) and cultured overnight. Cells were fixed with 4% paraformaldehyde in PBS containing 5% sucrose for 15 min and permeabilized with 0.5% Triton X-100 in PBS for 4 min. Immunostaining was performed with the indicated antibodies at room temperature for 1 h, followed by secondary Cy2, rhodamine X or Cy5-conjugated antibodies and/or Alexa Fluor 488–phalloidin. Stained samples were mounted in Gel/Mount (Biomeda, Foster City, CA). Immunofluorescence microscopy images were obtained using an LSM 510 triple laser confocal microscope equipped with A-Plan-Apochromat 63× 1.4 or plan-Neofluar 40× 1.3 objectives (Zeiss, Thornwood, NY). A 488 nm argon (∼10% power), a 543 nm HeNe1 (∼ 95% power) and a 633 nm HeNe2 (∼80% power) were used to excite 488 Alexa fluor and Cy2, rhodamine X and Cy5, respectively. The pinholes for each laser line were aligned for optimal confocality. For details regarding time-lapse fluorescence microscopy, image filtering and quantification, please see the Supplementary Information, Methods.

Time-lapse migration assay.

For time-lapse microscopy, cells were plated on tissue culture dishes (BD Falcon, San Jose, CA). After culturing overnight, the medium was changed, inhibitors added as indicated, and cell movements monitored using inverted microscopes (Axiovert 25, Zeiss) with 37 °C environmental chambers using A-Plan objectives 5× 0.12 NA or 10× 0.25 NA. Images were collected with CCD video cameras (model XC-ST50; Sony, San Diego, CA) at 10 min intervals and saved as image stacks using MetaMorph software (Universal Imaging Corporation /Molecular Devices, Sunnyvale, CA). Velocity was determined by tracking the positions of cell nuclei using the Track Point function of MetaMorph. Statistical analyses performed by Instat software (GraphPad Software, San Diego, CA) used one-way ANOVA tests for multiple samples and unpaired t-test, two-tailed, to compare two samples. Normality was confirmed by Kolmogorov and Smirnov's method. Error bars represent standard error of means s.e.m. unless otherwise indicated.

Inhibitor assays.

Blebbistatin was used at 25 μM unless indicated otherwise. Vinblastine was used at 100 ng ml−1 and nocodazole at 660 nM. When used to block blebbistatin effects, they were added 20 min before blebbistatin. Eg5 kinesin inhibitors were used as follows: monastrol, 50 μM; Eg5 inhibitor II, 2 μM; Eg5 inhibitor III, 5 μM; Trans HR22C16, 5 μM. The Rac inhibitor NSC23766 was used at 200 μΜ.

Fibrin gel contraction.

Gels were prepared in 24-well plates with 2 mg ml−1 bovine fibrinogen (Sigma) and 10 IU of thrombin in regular culture medium, and were left to polymerize at 37 °C for 30 min. Fibroblasts (2 × 104) or ES (1 × 105) cells were plated on top of the gels and cultured overnight at 37 °C in a 10% CO2 incubator. Blebbistatin or nocodazole were added after the cells had spread. The contracted fibrin gels were stained with 10% Ponceau S (Sigma), and images were obtained with a Stemi SV11 binocular microscope (Zeiss) equipped with a video camera at 0.6× magnification. The top contracted area was quantified using MetaMorph software (Molecular Devices/Universal Imaging).

Note: Supplementary Information is available on the Nature Cell Biology website.