Differential activity of nonmuscle myosin IIA and IIB isoforms generates a dynamic actomyosin network in a concentration-dependent manner

TL;DR

This study shows that nonmuscle myosin IIA (NMIIA) generates more dynamic actin networks in cells compared to IIB, influencing cell-scale actin remodeling.

Contribution

The study identifies NMIIA as the key isoform driving dynamic actin network behavior, distinct from NMIIB.

Findings

NMIIA generates higher network tension and faster rupture compared to NMIIB in simulations.

Live-cell imaging shows NMIIA causes more actin severing and peripheral actin arc formation.

Increased NMIIA expression enhances actin network dynamics, as shown by multiple experimental techniques.

Abstract

Cell-scale actin remodeling requires rapid actin depolymerization beyond that generated by cofilin and gelsolin. Previous reports had indicated that the activity of myosin restricted the length of actin bundles. However, it was unknown whether the ubiquitous nonmuscle myosin II isoforms (NMIIA and IIB) could generate cell-scale actin dynamics. Using linear actomyosin network simulation, we observed higher network tension and faster network rupture with NMIIA than NMIIB. Live-cell imaging of the actin network in COS7 cells also showed a similar result with numerous network severing events recorded in the presence of NMIIA while NMIIB produced fewer bundle severing events. Moreover, NMIIA was required for the formation of peripheral actin arcs and long actin fibers that were absent in cells-expressing NMIIB. We also observed the peripheral localization of cofilin in the presence of NMIIA supporting the live-cell imaging data of increased actin severing by NMIIA. Finally, using fluorescence recovery after photobleaching, optical trap–based cortical force measurements and live-cell imaging of actin network, we showed that the dynamics of the actin network increased with the increasing expression of NMIIA but not NMIIB. Thus, we established NMIIA as the predominant generator of cell-scale actin dynamics.

Figures

Full text

Cellular plasticity is maintained via the actomyosin network that regulates physiological processes including cell migration and cytokinesis. This requires dynamic actomyosin structures such as those in lamellipodia, filopodia, or cytokinetic ring. Actin fibers create branched networks at the periphery that continuously remodel themselves during cell migration (1). Actin-severing protein such as ADF/cofilin and gelsolin act on actin filaments at local levels to debranch and depolymerize the network (2). However, cellular scale changes require large fluxes of actin monomers; beyond those produced by ADF/cofilin or gelsolin. Nonmuscle myosin II (NMII) is an essential part of the actomyosin network. NMII is a hexamer with two heavy chains, and a pair each of regulatory light chains and essential light chains. The heavy chain contains the motor (binds to actin), neck, and alpha-helical coiled-coil domains that ends with a nonhelical tailpiece (3). There are three isoforms of the heavy chain giving rise to three NMII (NMIIA, NMIIB, and NMIIC) that are structurally similar but differ in their biochemical and biomechanical properties. NMIIA and NMIIB expression is almost universal in all cell types and forms the predominant NMII in actomyosin networks. Compared to NMIIB, NMIIA is a fast co-operative motor with low duty ratio (shorter duration of motor binding to the actin bundles) (3). NMIIB has a strong affinity for ADP (4) and shows long actin-attachment lifetime under resisting load when compared to NMIIA (5). Therefore, the biomechanical properties of the actomyosin network depends upon the type of NMII isoform expressed and the ratio of NMIIA and NMIIB in a mixed filament network. Such a mixed filament scenario was shown previously (6, 7).

Previously, it was reported that in fast moving fish keratocytes, NMII assembled at the rear of the cell body and that its activity was required for the disassembly of actin bundles (8). A similar finding was made in the neuronal growth cone where actin bundles in the filopodium were restricted by a zone of NMII (9). Inhibition of NMII activity using blebbistatin ablated the NMII zone leading to longer actin fibers. In line with these experimental results, modeling of actomyosin network in vitro and in silico showed a tensile force dependent buckling and shortening of actin filaments (10). Although these studies indicate the effects of myosin and NMII on the dynamics of actin network, NMII isoform–specific regulation of actin dynamics is yet to be unraveled. This is important as cellular scale changes in actin network dynamics would be governed by individual and combinatorial properties of NMIIA and NMIIB. A scenario of NMII isoform–specific regulation of actomyosin network has been postulated in silico (11) and in vivo (12). Furthermore, previous reports analyzing the effects of NMII isoforms using isoform-specific knockouts did not indicate the effects of individual isoforms in a mixed isoform scenario (13, 14, 15). Using a combination of linear actomyosin network simulation and live-cell imaging, we identified NMIIA as the predominant generator of network rupture and dynamics. Although NMIIA showed a positive correlation with cell stiffness, it maintained a dynamic cortical actin. Contrary to this, NMIIB expression positively correlated with cell stiffness without maintaining a dynamic cortical actin. Furthermore, cofilin localized to the peripheral actin network in the presence of NMIIA. This correlated with the increased actin severing and dynamics observed in these cells. We attributed the increased actin network dynamics in the presence of NMIIA to its motor-dependent actin bundle severing activity. Our study thus highlights the specific role of NMIIA-dependent actin bundle severing inducing cellular scale actin network remodeling.

Results

Linear actomyosin network simulation reveals NMIIA-dependent fast network severing

We simulated a linear two-dimensional actomyosin network using the Cytosim package while considering the catch-slip model as published previously (16). NMII isoform–specific linear actomyosin contractile network was produced by crosslinking the NMII isoforms with the actin filaments. We analyzed the network for the time of network rupture and the maximum tension of the network before rupture. Our simulations showed that NMIIA generated greater network tension (Fig. 1, A and B), leading to faster network rupture (Fig. 1C and Table 1) than NMIIB. We also simulated a condition with a linear network of crosslinked 360 NMIIB motor ensemble and 360 actin filaments spiked with increasing NMIIA concentrations. Increasing concentrations of NMIIB were used as a control (Fig. S1, A and C). Increasing concentration of NMIIA was found to increase the network tension (Fig. 1, D and F). However, such an increase in network tension was not observed with increasing numbers of NMIIB motors in the control simulation (Fig. S1, A and C). Next, we generated a linear network of crosslinked 360 NMIIA motor ensemble and 360 actin filaments and spiked it with increasing concentrations of NMIIB (Fig. 1, E and G). Increasing concentration of NMIIA was used as simulation control (Fig. S1, B and D). A network of NMIIA spiked with increasing concentration of NMIIB did not show a similar scaling of NMIIB concentration with network tension (Fig. 1, E and G). In fact, the NMIIA crosslinked actomyosin network already showed a high network tension that did not increase appreciably with the addition of NMIIB. The time of network rupture increased in networks with increasing concentrations of NMIIB (Fig. S1E and Table 1). This was unlike the network with increasing NMIIA where network rupture time showed a decreasing trend (Table 1). Thus, NMIIA was found to generate greater network tension contributing to faster network rupture. Moreover, the linear actomyosin network simulation hinted at the possibility of NMIIA motor activity-driven network rupture. To check such a possibility, we utilized COS7 cells that do not express the NMIIA isoform but express the NMIIB isoform. We overexpressed fluorescent lifeact (a fluorescent actin-binding probe) alone or with NMIIA in these cells and imaged the actin network at 10 s intervals. In control COS7 cells, we observed short actin bundles with sparse actin bundle severing events (Fig. 2, A and B and Fig. S3, Video S1). In contrast, upon NMIIA expression, we observed many actin bundle severing events averaging 34 severing events per cell (Fig. 2, A and B, Video S2). Moreover, overexpression of NMIIB in COS7 cells could not produce a similar increase in actin bundle severing and showed sparse severing events similar to COS7 (Fig. 2, A and B, Video S3). We also checked the length and width of actin bundles using FilamentSensor tool (17) in all these conditions to check the effects of NMIIA and NMIIB overexpression. The data clearly showed that overexpression of NMIIA decreased the actin bundle length (average length = 6.03 ± 1.65 μm) when compared to control (6.98 ± 1.69 μm) or when NMIIB was overexpressed (6.90 ± 1.97 μm) (Fig. 2C). Simultaneously the average width of actin bundles increased with NMIIB overexpression (control 0.16 ± 0.04 μm, NMIIA 0.117 ± 0.047 μm, NMIIB 0.167 ± 0.458 μm) (Fig. S2B). To further confirm the effect of NMII isoforms on actin dynamics, we overexpressed either NMIIA or NMIIB and EGFP-actin in COS7 cells expressing opto-RhoA-mCherry, a photoactivable RhoA construct (18). Upon blue light exposure opto-RhoA is recruited to the membrane leading to activation of NMII isoforms. We performed fluorescence recovery after photobleaching (FRAP) to measure actin dynamics preactivation and postactivation of opto-RhoA (Fig. S2C). We observed decreased half time of recovery of EGFP-actin in case of NMIIA overexpression (Fig. 2D). However, NMIIB overexpression showed increased half time of recovery post opto-RhoA activation (Fig. 2E). This indicated that NMIIA but not NMIIB increased the actin dynamics. Together, linear actomyosin network simulations and live-cell imaging of the actin network and dynamics suggested NMIIA isoform–specific remodelling of the actin network.Figure 1NMIIA-generated forces break the linear actomyosin network faster than NMIIB. The simulations include linear periodic contractile networks composed of white actin filaments and blue crosslinkers two types of motors are simulated: NMIIA (A) and NMIIB (B). NMIIA and NMIIB motor ensembles shown in red, crosslinkers are shown in blue, actin is shown in white. The simulation time steps are 0.5 s apart and the total time of simulation is 400 s. C, the time to network rupture was observed in five different simulations, and the average rupture time was plotted for networks containing only NMIIA and NMIIB. Data represent mean ± SEM, n = 5. p values ∗∗∗p < 0.001. Statistical significance was assessed using unpaired two-tailed Student’s t tests with Welch’s correction. D, a network was simulated with 360 NMIIB molecules, with NMIIA filaments in increments of 60 until reaching 360 molecules. Time of simulation is indicated. NMIIA motor ensembles are shown in red, whereas NMIIB motor ensembles are shown in yellow when both motors are added. E, in another scenario, NMIIB was added incrementally by 60 molecules on top of a network already containing 360 NMIIA molecules. Time of simulation is indicated. NMIIA motor ensembles are shown in red, whereas NMIIB motor ensembles are shown in yellow when both motors are added. F, the maximum tension generated in NMIIB crosslinked actin network with incremental addition of NMIIA was plotted. Data represent mean ± SEM, n = 5. G, the maximum tension generated in NMIIA crosslinked actin network with incremental addition of NMIIB was plotted. Data represent mean ± SEM, n = 5, p values ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. Statistical significance was assessed using Welch’s one-way ANOVA with Games–Howell post hoc tests. NMII, nonmuscle myosin II.Table 1. Linear simulation of actomyosin network was analyzed for the time of network ruptureNMIIB crosslinked actin networked spikes with increasing numbers of NMIIANetwork rupture time (sec) indicated as mean ± SEM of 5 simulationsNMIIA crosslinked actin networked spikes with increasing numbers of NMIIBNetwork rupture time (sec) indicated as mean ± SEM of 5 simulationsNMIIB55.12 ± 3.44NMIIA16.6 ± 0.95NMIIB + NMIIA6049.375 ± 6.91NMIIA + NMIIB6017.2 ± 1.50NMIIB + NMIIA12049.7 ± 2.31NMIIA + NMIIB12027.1 ± 2.67NMIIB + NMIIA18036.6 ± 3.50NMIIA + NMIIB18030.5 ± 2.87NMIIB + NMIIA24043.8 ± 6.43NMIIA + NMIIB24035.2 ± 4.88NMIIB + NMIIA30039.9 ± 2.36NMIIA + NMIIB30031.375 ± 1.89NMIIB + NMIIA36041.7 ± 5.06NMIIA + NMIIB36043.25 ± 1.34Actomyosin network was produced by crosslinking either NMIIA or NMIIB with the actin. To mimic the mixed isoform scenario, increasing numbers (indicated in the table) of the other isoform was added. The first two columns show the network rupture time when there are 360 NMIIB filaments with NMIIA filaments increasing in increment of 60. The last two columns show the network rupture time when there are 360 NMIIA filaments with NMIIB filaments increasing in increment of 60.NMII, nonmuscle myosin II.Figure 2Overexpression of NMIIA in COS7 cells increases breakage of actin bundles. A, COS7 cells were transfected with either Lifeact-GFP (green), which binds to filamentous actin, or NMII-mCherry along with Lifeact-GFP (green). Imaging was performed at 10 s intervals, with cells maintained at 37 °C and 5% CO2. Representative images at 0, 40 and 80 s of imaging and corresponding kymographs are shown for COS7, NMIIA overexpression, and NMIIB overexpression. White arrows highlight the bundle breakage. The portion of the actin bundles for which the kymographs have been generated is shown in red. The scale bar represents 5 μm. B, quantification of actin bundle severing events. Breakage events were manually quantified across 120 frames for each cell, and the average number of events was plotted (n = 3–9 cells per condition; ∗∗∗p < 0.001). Statistical significance was assessed using unpaired two-tailed Student’s t tests with Welch’s correction. C, actin bundle length values extracted using the filament sensor were quantified and bundle length distribution >5 μm was plotted as indicated. D and E, the half-time of eGFP fluorescence recovery was measured in cells cotransfected with opto-RhoA–mCherry, eGFP-actin, and the indicated nonmuscle myosin II (NMII) isoform. Fluorescence recovery after photobleaching (FRAP) was performed on the indicated regions before and after 3 min of UV exposure to activate opto-RhoA. Recovery kinetics were analyzed to assess changes in actin and NMII dynamics upon optogenetic stimulation.

NMIIA remodels the actin network in cells

Next, we analyzed the actin network in COS7 cells and COS7-overexpressing NMIIA or NMIIB. COS7 cells showed the presence of shorter actin bundles and crosslinked peripheral actin network (Fig. 3, A and B, Fig. S3). On the contrary, NMIIA-expressing COS7 cells showed long actin stress fibres and peripheral actin arc structures. Such peripheral arcs were also observed in the live-cell images (Fig. S3). Moreover, peripheral arcs and long actin bundles were not observed with overexpression of NMIIB (Fig. 3, A and B). This indicated that NMIIA specifically altered the actin network. To further verify whether NMIIA could induce a differential localization of actin remodeling proteins such as cofilin, we performed immunofluorescence staining for cofilin, an actin-binding protein that aids in actin filament severing. In COS7 cells, cofilin localized to the actin network away from the peripheral lamella toward the nuclear proximal cytoskeleton network (Fig. 4, A and B). Interestingly, upon NMIIA expression, cofilin was redistributed to the peripheral actin network, indicating remodeling of the actin network in the presence of NMIIA. This agreed with the previous reports where NMIIA-bound actin network sequestered cofilin, orchestrating actin disassembly (19). NMII isoforms have polymerization and motor properties. We asked which of the two properties of NMIIA might be responsible for increased actin dynamics. To check the contribution of polymerization, we utilized the EGFP-NMIIA-deltaIQ2 mutant (20) which has high polymerization activity and to check the contribution of motor activity, we used blebbistatin which is a known inhibitor of ATP-dependent motor activity (21). COS7 cells expressing NMIIA, when treated with blebbistatin, retained the peripheral actin arcs (Fig. 3, A and B), although the number of arcs per cell was less compared to NMIIA control cells. However, peripheral arcs and long actin bundles were sparsely present in COS7 cells expressing the highly polymerizing mutant NMIIA-deltaIQ2 (Fig. 3, A and B). Moreover, COS7-expressing NMIIA-deltaIQ2 showed a diminished and diffused cofilin localization (Fig. 4, A and B). This indicated that increasing the polymerization activity of NMIIA was insufficient to remodel the actin network. At the same time, decreased motor activity under blebbistatin treatment drastically reduced the number of peripheral actin arcs (Fig. 3B), indicating an effect of the motor activity on the remodeling of the actomyosin network.Figure 3NMIIA overexpression changes actin structures in COS7 cells. A, COS7 cells were transfected with either NMIIA, NMIIB, or NMIIA deltaIQ2. Immunofluorescence image showing actin (magenta) and the corresponding NMII isoform (red) was taken under a confocal microscope. The white arrow shows the location of actin arcs. The scale bar represents 10 μm. B, the number of arcs was manually counted from each image, and the mean value was plotted (n = 41–133 cells per condition). p values ∗∗p < 0.01, ∗∗∗p < 0.001. Statistical significance was assessed using Welch’s one-way ANOVA with Games–Howell post hoc tests. NMII, nonmuscle myosin II.Figure 4NMIIA alters cofilin distribution in COS7 cells. A, COS7 cells were transfected with either NMIIA, NMIIB, or NMIIA deltaIQ2. Representative immunofluorescence image showing actin (magenta) and NMII isoform (red). Cofilin is shown here in green. The scale bar represents 10 μm. B, cells were divided into three zones: peripheral, middle zone, and perinuclear, and the number of cofilin particles was counted in each zone, and their averages were plotted (n = 7–9 cells per condition). p values ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. Statistical significance was assessed using Welch’s one-way ANOVA with Games–Howell post hoc tests. NMII, nonmuscle myosin II.

Optical trap–based cortical stiffness measurements revealed NMIIA-specific dynamic cytoskeletal remodeling

The actomyosin network also governs the mechanical properties of a cell. We, therefore, probed the cortical stiffness of the cells through optical tweezers based force spectroscopy experiments to investigate the signature of NMIIA-dependent remodeling of the peripheral actin network (22). Although indirect, these measurements provide confirmatory indication of the underlying actin dynamics. We generated a gradient of expression of mCherry-tagged NMIIA in COS7 cells by flow sorting cells on the intensity of mCherry for the cortical stiffness measurements (Fig. 5A). Thus, COS7 cells were flow sorted into varied NMIIA expressions, starting from no NMIIA expression to COS7 with low, medium, and high expression of NMIIA (Fig. S4C). Similarly, we transfected COS7 (endogenous NMIIB expression—low) with fluorescent-tagged NMIIB. COS7 cells were flow-sorted into medium and high overexpression of NMIIB (Fig. S4D). Thereby, we successfully established a gradient of expression of NMIIA and NMIIB in COS7 cells (Fig. 5A). For the force spectroscopy measurements, polystyrene beads were attached to the cell boundary of coverslip attached COS7 cells. This constituted the optical handles. A polystyrene bead was optically trapped and pulled at a constant speed; consequently, a tether was formed consisting of cell membrane and underlying cortex. While the force required to pull the tether was measured from the displacement of the trapped bead from the center of the optical trap, the total movement of the stage provided the elongation of the tether. The average slope of the variation of force with the elongation determined the elastic constant, measuring the cortical stiffness.Figure 5NMIIA increases cortical stiffness while maintaining cortex dynamics. A, schematic of the experiment exhibiting sorting of cells with different levels of myosin expression tagged with mCherry. On the right is a schematic explaining the force spectroscopy measurement protocol. B and C, typical force-elongation plots exhibiting a few measurements with overexpression of NMIIA and NMIIB, respectively. The average variation of these few measurements is shown in black, and the red lines are linear fits to the average variations. D, COS7 cells (control) and overexpressing NMIIA were analyzed by the force spectroscopy protocol. Cells with medium and high NMIIA expression exhibited two force constant values (prior to and after yielding), indicating that the cells expressing NMIIA have two elastic regimes, suggestive of a dynamic cortical network that withstands longer tether elongations. In the case of NMIIA medium and high overexpression, the force constant measured after yielding transition is indicated in gray. Number of experiments performed in each condition: COS7—seven cells, 17 measurements; NMIIA (low)—six cells, 14 measurements; NMIIA (medium)—six cell, 14 measurements; and NMIIA (high)—nine cells, 21 measurements. E, COS7 cells expressing different concentrations of NMIIB were analyzed using the force spectroscopy protocol. An elastic constant from only the initial elastic regime is shown for NMIIB-overexpressed cells, as in most cases, the tethers broke from the cell at a longer elongation (corresponding to higher force) before reaching the yield point. Data represent mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. The number of experiments performed in each condition: COS7—seven cells, 17 measurements; NMIIB (medium)—six cells, 11 measurements; and NMIIB (high)—five cells, 14 measurements. Statistical significance was assessed using Welch’s one-way ANOVA with Games–Howell post hoc tests. NMII, nonmuscle myosin II.

We observed a gradual increase in the cortical stiffness of cells with increasing expression of NMIIA (Fig. 5D). However, the cells exhibited a yielding transition, characterized by an abrupt decrease in the elastic constant at longer elongation, that is, an increase in deformability under higher applied force at higher concentrations of NMIIA (Fig. 5, B and D). Considering the average behavior (before and after yielding), the cortical stiffness showed a moderate increase with the expression of NMIIA. Thus, increased expression of NMIIA modestly increased the cortical stiffness while maintaining a highly dynamic actin network. On the contrary, we observed a substantial increase in the elastic constant of cells with increasing NMIIB expression (Fig. 5E). Remarkably, in most of the cases with NMIIB, either the polystyrene bead detached from the cells, or the tethers broke at a much lower force, that is, at shorter elongation. This was indicative of the stiffness of the underlying cortex that resisted tether pulling.

A similar observation was made in linear actomyosin simulations: varying NMIIB levels while keeping NMIIA constant increased the network rupture time (Table 1 and Fig. S1E). This indicated that NMIIB generated a stable actomyosin network contributing to increased cortical stiffness. On the other hand, NMIIA generated a dynamic cortical cytoskeletal network that resisted tether breakage.

NMIIA and NMIIB produce contrasting effects on actin dynamics in cells

Next, we asked if overexpression of either of the two NMII isoforms modulates the actin dynamics in these cells. To estimate the dynamic actin content, we expressed EGFP-actin. By bleaching a cortical actin region of the cells for EGFP-actin followed by analyzing its fluorescence recovery (FRAP), we obtained two parameters that indicated the dynamics of filamentous actin–the half-life of recovery (t_1/2_) and the percent mobile fraction.

We transfected COS7 cells with mCherry-tagged NMIIA along with EGFP-actin. We flow-sorted the COS7 cells overexpressing mCherry-tagged NMIIA based on mCherry fluorescence into low, medium, and high NMIIA-expressing cells (Fig. S4, A and C). We ensure that the COS7 cells with low, medium, and high NMIIA expression had a low but equal expression of EGFP-Actin (Fig. S4A). Using FRAP, we checked the dynamics of EGFP-actin under these gradients of expression of NMIIA (Fig. 6, A and B). Half-time (t_1/2_) of recovery (Fig. 6C) and percent mobile fraction (Fig. 6D) of EGFP-actin was calculated for each condition. With increasing NMIIA expression, we observed a decrease in the t_1/2_ of EGFP-actin, indicating increased actin dynamics (Fig. 6C). We then flow-sorted COS7 cells based on the levels of overexpression of NMIIB into medium- and high-expressing cells (Fig. S4, B and D). We thus established a gradient of expression of NMIIB in COS7 cells. Similar to NMIIA overexpression, increasing expression of NMIIB showed a significant decrease in the half-time recovery of EGFP-actin. However, unlike NMIIA, a significant decrease in the percent mobile fraction of EGFP-actin was observed in NMIIB medium and NMIIB high conditions (Fig. 6, F and G). Together these results indicated that NMIIB decreased the levels of dynamic actin, whereas NMIIA increased the dynamics of the actin network.Figure 6Dynamics of actin with increasing expression of NMIIA. COS7 cells–expressing EGFP-actin and mCherry-NMIIA were sorted based on NMIIA fluorescence levels. Photobleaching of EGFP-actin followed by an analysis of its recovery kinetics was performed. A, a representative montage of COS7 cells–expressing EGFP-actin and mCherry-NMIIA shows the prebleach and postbleach recovery of EGFP-actin fluorescence (green). Bleaching was performed for a 20 × 20 μm area toward the periphery of the cells. The scale bar represents 10 μm. B, a representative fluorescence recovery curve of EGFP-actin for each indicated condition. C, half-life of recovery (t_1/2_) and (D) percent mobile fraction of EGFP-actin was estimated for cells with low, medium, and high NMIIA expression levels. (n = 9–13 cells per condition). ∗p < 0.05. Statistical significance was assessed using Welch’s one-way ANOVA with Games–Howell post hoc tests. COS7 cells–expressing EGFP-actin and mCherry-NMIIB were sorted based on NMIIB fluorescence levels. Photobleaching of EGFP-actin was followed by an analysis of its recovery kinetics. E, a representative fluorescence recovery curve of EGFP-actin for each indicated condition. F, half-life of recovery (t_1/2_) and (G) percent mobile fraction of EGFP-actin was estimated for NMIIB gradient conditions (endogenous NMIIB, medium NMIIB, and high NMIIB). (n = 7–13 cells per condition). ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. Statistical significance was assessed unpaired two-tailed Student’s t tests with Welch’s correction. NMII, nonmuscle myosin II.

Increasing concentration of NMIIA increased the actin severing events in COS7 cells

The previous results hinted at the possibility of an NMIIA gradient-dependent actin remodeling. Since our simulation and live-cell imaging data indicated NMIIA-dependent actin severing (Figs. 1 and 2), we verified whether a gradient of NMIIA expression could produce a progressive increase in actin severing events. We performed live cell imaging of COS7 cells alone or with overexpression of NMIIA at 10 s intervals. We sorted the cells based on NMIIA expression and performed live-cell imaging with low, medium, and high expression of NMIIA. In NMIIA low conditions, isolated actin bundle severing events were observed along with shorter actin bundles (Fig. 7, A and B, Fig. S3, Video S4). On average, 20 severing events per cell were recorded. With NMIIA medium, distinct actin arcs were visible near the cell periphery, along with severing events averaging 37 severing events per cell (Fig. 7, A and B, Fig. S3, Video S5). With high NMIIA expression, many severing events were observed, averaging 43 severing events per cell (Fig. 7, A and B, Fig. S3, Video S6). Remarkably, high NMIIA expression also showed the presence of shorter bent actin bundles along with prominent actin arcs. With increasing expression of NMIIA, actin bundles showed nonsignificant decrease in average length (average length: low = 6.97 ± 2.09 μm, medium = 6.91 ± 2.17 μm, high = 6.68 ± 1.65 μm) (Fig. 7C), indicating NMIIA expression-dependent alteration of actin network. As a control, we treated COS7 cells–expressing NMIIA with blebbistatin. With the inhibition of motor activity, we did not observe any actin severing event (Fig. 7A, Video S7). Instead, as expected, we observed thick and shorter actin bundles and blebbing of the membrane. Taken together, our experimental data was in concurrence with our simulation data and revealed increased NMIIA concentration-dependent actin bundle severing. These results indicated that NMIIA motor activity generated tensile forces sufficient to create a dynamic actin network in a concentration-dependent manner.Figure 7The gradient of NMIIA enhances actin breakages in COS7 cells. COS7 cells were transfected with either Lifeact-GFP alone or cotransfected with NMIIA-mCherry. Following transfection, cells were flow-sorted into populations with low, medium, and high NMIIA expression levels. These sorted cells were then plated on bottom-coverslip dishes and adhered to the surface overnight. After a 24-h incubation period, time-lapse imaging was performed at 10-s intervals, with the cells maintained under physiological conditions at 37 °C and 5% CO_2_. A, representative images at 0, 40 and 80 s of imaging and corresponding kymographs are shown for NMIIA low, medium, and high expressing cells. White arrows highlight the actin bundle breakage events. The portion of the actin bundles for which the kymographs have been generated is shown in red. The scale bar represents 5 μm. The last panel shows NMIIA-overexpressing cells after 30 min of Blebbistatin treatment. B, quantification of actin bundle severing events. Breakage events were manually observed across 120 frames in all the cells, and their average was plotted. (n = 3 cells per condition), ∗p < 0.05. C, actin bundle length values extracted using the FilamentSensor were quantified and length distribution >5 μm was plotted as indicated. Statistical significance was assessed unpaired two-tailed Student’s t tests with Welch’s correction. NMII, nonmuscle myosin II.

Discussion

Various actin remodeling proteins have been implicated in the regulation of actin network dynamics, such as the actin severing proteins cofilin and gelsolin, the actin branching Arp2/3 complex, and actin polymerization promoter profilin. Because the actin remodeling proteins are spatially regulated, they control the network dynamics at distinct subcellular regions. However, large-scale remodeling of the actin cytoskeleton is required to support physiological processes such as neuronal growth cones, dendritic branching, and generation of proplatelets from a mature megakaryocyte. Recently, we reported a NMIIA-dependent actin remodeling and corresponding increase in the traffic and signaling of the thrombopoietin receptor (23). These require actin remodeling at cellular scale that must be dependent upon proteins with generic association with the actin network. Myosin proteins were found to be effective in generating such remodeled actin network in fish keratocytes or neuronal growth cones (8, 9). However, it was still unknown if NMII isoforms which are abundant in all cell types contribute toward cell-scale actin remodeling. Previous network simulations had indicated that myosin activity caused a collapse and disruption of the actomyosin network (10, 24). We performed linear actomyosin network simulation and found that actomyosin network tension and time of rupture was dependent upon the isoform of NMII. NMIIA being a fast cooperative motor generated greater network tension leading to faster network rupture. NMIIB crosslinked actomyosin network showed an increase in maximum network tension with progressive increase in NMIIA motor units. On the contrary, actomyosin network established with NMIIA crosslinked to actin showed a high network tension that did not progressively increase upon further increase in NMIIB motors. To further probe the actin network dynamics with the overexpression of NMIIA, we utilized EGFP-actin probe for a FRAP-based actin dynamics assay. We observed a direct correlation between the expression of NMIIA and the increased dynamics of the actin network as revealed through decreased t_1/2_ of recovery of actin without any change in its percentage mobile fraction. On the contrary, increased expression of NMIIB decreased the dynamics of the actin cytoskeleton by decreasing the percentage of mobile fraction of actin. Moreover optogenetic activation of RhoA also showed a similar trend of reduced t_1/2_ of EGFP actin in the presence of NMIIA. In fact, it was observed that the NMIIB was a better actin crosslinker protein than NMIIA. This was indicated through the observed correlation between increased levels of NMIIB expression, increased cortical stiffness, decreased percent mobile fraction of actin and increased average width of actin bundles. Although increased NMIIA expression did show an increase in the cortical stiffness, force measurements showed a yielding transition at higher force of extension that was not observed for NMIIB. This indicated a highly dynamic actomyosin cortex produced in the presence of NMIIA but not in the presence of NMIIB. These results were in agreement to previous observation where NMIIB decorated stable actin networks observed at the rear of a migrating cell while NMIIA localized at the dynamic leading edge (25). Additionally, our results concur with previous observation that fast network remodeling similar to the one required during directional cell migration required NMIIA (26). Incidentally, NMIIA expression in COS7 cells produced actin arcs that were not observed with NMIIB. This result agreed with previous reports (15, 26, 27). Surprisingly, actin arc structures were observed when NMIIA was expressed above a threshold level, indicating a NMIIA concentration dependent actin remodeling.

It was, therefore, possible that NMIIA motor activity–driven actin network rupture contributed to the observed increase in actin dynamics. We performed live-cell imaging of the actin network in COS7 cells with different levels of expression of NMIIA. In the absence of NMIIA or with low NMIIA expression, short actin bundles were observed. With further increase in NMIIA, long transverse fibers and actin arcs were seen along with increased bundle severing. Interestingly, with high NMIIA expression, both long fibers/arcs as well as short bundles were observed possibly due to increased network rupture. Furthermore, actin bundle length showed a decreasing trend with increased expression of NMIIA. Blebbistatin, an inhibitor of ATP-dependent NMII motor activity confirmed that the observed increase in actin severing with increased NMIIA expression was dependent upon its motor activity. Our data thus show that NMIIA motor activity-dependent actin severing results in a cellular scale increase in dynamic actin.

Experimental procedures

Cell culture and transfection

COS7 cells were grown in Dulbecco's modified Eagle's medium with 10% FBS and 1% penicillin-streptomycin in a humidified CO2 incubator at 37 °C. The plasmids with the desired gene were transfected in the cells using Lipofectamine 2000 Transfection Reagent (Invitrogen, Thermo Fisher Scientific) following the manufacturer’s protocol. For blebbistatin treatment, cells were treated for 30 min at a concentration of 10 μM. The plasmids that are used for this study are mCherry-MyosinIIA-C-18, pDEST/LIfeAct-mCherry-N1, pCMV-EGFP-NMHC-IIAdeltaIQ2, pTRE-GFP-NMHC II-B, mCherry-MyosinIIB-N-18, EGFP actin 7, and opto-RhoA-mCherry_pcDNA3.1.

Immunofluorescence and confocal microscopy

Cells were seeded onto coated coverslips and then grown up to the confluency of 40 to 50%. Cells were transfected with the plasmids having the desired construct of NMII isoform along with lifeact. Media were removed, and cells were washed with KHM buffer (25 mM Hepes pH 7.2, 125 mM potassium acetate, and 2.5 mM magnesium acetate). To remove soluble proteins, cells were transiently permeabilized with 40 μg/ml digitonin solution in KHM buffer for 2 min. Cells were washed once with KHM buffer and fixed in 4% formaldehyde for 20 min. Following fixation, cells were washed several times with PBS + 10% fetal bovine serum (FBS) and then blocked and permeabilized with PBS + 10% FBS + 0.2% Triton X-100 for 30 min. After washing with PBS, cells were incubated with the primary antibody (1:200 in PBS + 1% FBS) overnight at 4 ^o^C. The next day, cells were incubated with a secondary antibody (1:1000 in PBS + 1% FBS) for 1 h. For fluorescent phalloidin staining, cells were stained with 66 nM Alexa 647–tagged phalloidin (Thermo Fisher Scientific, cat.no. A22286) for 30 min, washed with PBS, and then mounted on slides. We used another protocol for cofilin staining. Cells were washed with KHM buffer, permeabilized with 80 μg/ml digitonin for 2 min, fixed with 4% formaldehyde in PBS for 20 min, and washed with PBS + 10% FBS. Permeabilization and blocking were performed in PBS + 10% FBS+100 μM digitonin for 15 min. Primary antibody staining was done overnight at 4 °C in PBS + 1% FBS, followed by secondary antibody incubation for 2 h at room temperature in PBS + 1% FBS. Phalloidin staining (1:400) was carried out for 30 min at room temperature, followed by PBS wash and imaging. Confocal imaging was performed with 63X oil-objective using Leica TCS SP8 Confocal Laser Scanning Microscope (CRF, IIT Delhi) equipped with lasers of wavelengths of 405 nm, 488 nm, 552 nm, and 635 nm.

Following antibodies have been used for immunofluorescence study: CFL-1 (Abclonal, cat no. A1704), mCherry (Abclonal, cat no. AE002), GFP (Abclonal, Cat no. AE012), Alexa Fluor 488 rabbit (Thermo Fisher Scientific, Cat no. A11070), Alexa Fluor 488 mouse (Thermo Fisher Scientific, Cat no. A11017), Alexa Fluor 568 rabbit (Thermo Fisher Scientific. Cat no. A21069), Alexa Fluor 568 mouse (Thermo Fisher Scientific, Cat no. A11019).

Live-cell imaging and FRAP

Live-cell imaging of actin network was performed in a Zeiss Elyra SIM microscope (SATHI, IIT Delhi). Ten to twenty thousand cells were seeded in 35 mm bottom coverslip plates and then transfected with the desired NMII isoform plasmid along with Lifeact-GFP. Live-cell imaging was performed for 20 min at 37 °C and 5% CO_2_. Images were acquired at 10 s intervals for 20 min using a 63X oil-immersion objective. The acquired images were processed by Zen Blue software (https://www.zeiss.com/microscopy/en/products/software/zeiss-zen.html) to generate the super-resolution images of the actin network. We observed the time lapse images for breakage using a kymograph. To discriminate between fluctuations during imaging vis-a-vis actual bundle severing event, we observed for the breakage (gaps in the kymograph) that persisted until the end of imaging. For breakages appearing at the very end of the time-lapse imaging, they were counted only if we could follow them for at least 10 subsequent frames. For quantification, we manually counted these events from the kymographs. For FRAP, cells were seeded onto coated coverslips in 35 mm plates. Cells were grown up to the confluence of 50 to 60% and then transfected with the desired NMII construct and fluorescent EGFP-actin.

FRAP was performed using the Leica TCS SP8 Confocal Laser Scanning Microscope under a 63X oil objective at confocal conditions. All recordings were performed on a selected region of interest (ROI) of 18 to 20 μm^2^ at the cell cortex. For bleaching, 30 to 50% laser intensity was used. Ten prebleach frames were acquired, and 60 to 180 postbleach frames were recorded at 1-s intervals. Fluorescence intensity at ROIs was exported for further analysis on Origin. Data were plotted and fitted with ExpAssoc2 double exponential equation.

$$[eqn]$$ $$[eqn]$$

where TD1 = first time offset: x value at which first exponential begins, TD2 = second time offset: x value at which second exponential begins, Yb = baseline: y value at which exponential begins, A1 = first amplitude: change in response for first exponential, A2 = second amplitude: change in response for second exponential, τ_1_ = first time constant, τ_2_ = second time constant.

From the fitted graph, the t_1/2_ and percent mobile fraction was calculated.

Percentage recovery = ((maximum fluorescence intensity after recovery – intensity at t_0_ sec)/(prebleach intensity – intensity at t_0_))∗100.

t_1/2_ was calculated according to the following formula from the fitted graph.

$$[eqn]$$

where y(t_1/2_): The fluorescence intensity at the half-time of recovery.

y_max_: the maximum fluorescence intensity after full recovery.

y_0_: the baseline fluorescence intensity immediately after bleaching.

t_1/2_: the time at which the intensity reaches halfway between y_0_ and y_max_.

At least seven of such plots were fitted per condition and the mean ± SEM for the t_1/2_ and percent mobile fraction was obtained.

FRAP with photoactivatable RhoA was performed using the same protocol described above, with the exception that two distinct ROIs were analyzed within the same cell. FRAP was first performed on one ROI prior to UV exposure. Cells were then exposed to UV light for 3 min to activate photoactivatable RhoA, after which FRAP was performed on a second ROI located on the opposite side of the same cell.

Fluorescence-activated cell sorting

COS7 cells were grown for 24 h to reach the confluency of 50 to 60%. Cells were transfected with the desired plasmid. Cells were kept at 37 °C and 5% CO_2_ in a humidified incubator for at least 24 h for the proteins to express. Cells were trypsinized, washed three times with PBS, and then resuspended in PBS for sorting. Cells were sorted using BD fluorescence-activated cell sorting AREA III flow sorter, based on the expression of GFP and mCherry fluorophores. Sorted cells were collected in Dulbecco's modified Eagle's medium complete media with 20% FBS. Cells were pellet down and washed two times with PBS and then seeded in 35 mm bottom coverslip plates. Cells were grown until firmly attached to the coverslip before imaging.

Actomyosin network simulation

We simulated a system of connected actin and myosin fibers using the Cytosim package, details of which may be found in (28). The physics of the system is described by a Langevin equation that considers the Brownian motion of the fibers, bending elasticity, fiber–fiber interactions and external force fields. Details of the equations and the constraints used may be found in (28). We assessed the NMII isoform–specific motor activity on actin network using the Cytosim DC package which is standalone version of Cytosim package developed by (16). NMII motor was defined as rods containing 15 motor entities at both the ends. Motors followed catch-slip unbinding dynamics. Periodic linear contractile system simulations were performed as described by (16). A 9.42 mm network with 360 actin filaments, 360 motor ensembles, and 1000 cross-linkers was used. Simulations ran for 400 s during which most networks ruptured, forming filament-free gaps and recoiling into clusters. We exported the network tension data from the simulations using Cytosim's report function (Fiber: Tension), which captures the forces generated by the filaments in each simulation frame. This allowed us to track the maximum tension produced in the network over time. In addition to the quantitative tension data, we analyzed the simulation videos to visually observe the behavior of the network during rupture events. These rupture events were characterized by the formation of filament-free gaps, followed by the recoil of the network into dense clusters. The parameters for NMIIA and NMIIB used in the simulation are summarized in Table 2.Table 2NMII isoform–specific parameters used in the linear simulation of actomyosin networkParametersNMIIAReferencesNMIIBReferencesbinding_range0.06 μm(16)0.06 μm(16)binding_rate0.2 s^−1^(31)0.2 s^−1^(4)unbinding_rate1.71 s^−1^(31)0.35 s^−1^(4)alpha_catch0.92(11, 32)0.92(11, 32)x_catch0.0025(11, 32)0.0025(11, 32)alpha_slip0.08(11, 32)0.08(11, 32)x_slip0.0004(11, 32)0.0004(11, 32)unloaded_speed0.3 μm s^−1^(33)0.1 μm s^−1^(34)stall_force3.4 pN(35)2.2 pN(36)NMII, nonmuscle myosin II.

Elastic constant measurements

The elastic constants of the cells are determined by pulling a tether from the cells and measuring the force required using optical tweezers (22, 29). The cells are grown on the culture dish and thus are inherently attached to the substrate. A trace amount of polystyrene microspheres with diameter 1.98 μm (microparticles GmbH) are added to the culture dish containing the cells and kept in an incubator overnight. The microbeads get spontaneously attached to the cell membrane at nonspecific sites. A chosen beads among those attached at cortical region of the cell is then trapped optically using optical tweezers, which is setup around an inverted microscope by tightly focusing a 1064 nm laser beam through a 60 × oil-immersion objective (Nikon) (29). As the bead is held in the optical trap, a piezo-controlled sample stage is translated linearly at a constant speed (v = 1 μm/s) along the x-axis until the bead escapes from the trap due to the tension from the tether, or the bead gets detached from the tether, or the tether breaks. Since the tether is not visible under bright-field microscopy, its formation is verified from indirect observations, such as, requirement of stronger force as the membrane-attached-bead is pulled farther away from the cell, and the recoil of the bead due to the elastic force from the tether when the optical trap is switched off (30). While the stage is translated, the tether is elongated by a length l as the attached bead is held by the optical trap, and the applied force (F) displaces the bead from the center of the trap by an amount Δx (Fig. 5A). This displacement, Δx, measures the force as $[eqn]$ , where k is the stiffness of the optical trap. The instantaneous elongation is measured from the stage movement as l = (v t - Δx), where t is the time duration over which the stage has moved. The measurements are repeated a few times on the same cell by pulling different microbeads attached to the same cell, and on multiple cells with similar NMIIA/NMIIB expressions, to enhance the statistics. The average slope of the linear variation of force (F) with elongation (l) provides the elastic constant of the cell. In some cases, this slope, and hence the elastic constant, changes at longer elongation, upon yielding.

Actin bundle length and width analysis

Actin bundle length and width distributions were quantified from structured illumination microscopy images using FilamentSensor (v2.0) (17), a Java-based image analysis software for filament detection and quantification. Raw structured illumination microscopy images were imported in OME-TIFF format and analyzed using the Filament analysis module. All images were processed using the default parameter settings provided by the software, without manual modification of segmentation or feature extraction thresholds. The analysis workflow included automated preprocessing, filament segmentation, and skeletonization to identify individual filament structures. Filament length was calculated along the extracted filament skeletons, whereas filament width was estimated from intensity profiles perpendicular to the filament axis, as implemented in FilamentSensor. Detected filaments were visually inspected to confirm appropriate segmentation. Quantitative measurements were exported as comma-separated value (.csv) files and used for subsequent statistical analysis.

Analysis of cofilin distribution

To measure the number of cofilin dots, the entire cell area was segmented into 5 μm sections from the cell periphery using the Edit > Enlarge function in Fiji (ImageJ, https://imagej.net/software/fiji/downloads). For quantitative analysis, images were segmented into concentric regions of 0 to 5 μm, 5 to 10 μm, and the remaining perinuclear region. Signals outside the region of interest were excluded by generating a binary mask using the createMask function. Cofilin dots were detected through thresholding, applied via image > adjust > auto local threshold > Bernsen. Finally, particle analysis was performed using the analyze particles function, with a size cut-off range of 0.05 μm to infinity.

Statistical analysis

All data are presented as mean value ± SEM. For comparisons involving more than two independent groups, Welch’s one-way ANOVA was used due to unequal sample sizes and unequal variances, followed by Games–Howell post hoc tests for multiple comparisons. Actin bundle length and width were analyzed using a two-tailed Mann–Whitney U test. For two-group comparisons, unpaired two-tailed Student’s t tests assuming unequal variances (Welch’s correction) were applied and p values less than or equal to 0.05 were considered significant (∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001).

Data availability

The data generated in this article are depicted in figures and supplementary figures and videos.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.