Ordering of myosin II filaments driven by mechanical forces: experiments and theory

1. Introduction

Myosins comprise a superfamily of molecular motors that interact with actin filaments and use the energy of ATP hydrolysis to move along these filaments thereby producing active mechanical forces [1,2]. The part of the myosin heavy chain molecule that interacts with actin filaments, known as the myosin head, is conserved among all members of the myosin superfamily, consisting of more than 30 classes [2]. Other parts of the myosin molecule, the neck and the tail, can differ significantly in different myosin classes. Compared to other myosin classes, the myosin II sub-family has the unique feature that its extended tails can interact with each other to form functional, bipolar filaments. Interaction of the bipolar myosin II filaments with arrays of oppositely oriented actin filaments pulls these arrays towards each other producing contractions typical for both muscle and non-muscle cells [3,4].

This review is devoted to the organization of myosin II filaments in muscle and in particular non-muscle cells. We discuss experimental data that elucidate the organization of myosin II filaments into superstructures comprising of tens to hundreds of filaments. These superstructures or arrays not only generate cell contractility but also determine the global organization of actin filaments at the level of the entire cell. Actin and myosin II filaments, together with numerous accessory proteins, form characteristic actomyosin structures such as sarcomeres in cross-striated muscle cells and different kinds of actin bundles and networks in non-muscle cells. These various types of organization of the actomyosin cytoskeleton determine cell contractility, adhesion, locomotion and morphogenesis. We further propose and discuss a unifying theory that explains the formation of the myosin superstructures comprising numerous, laterally aligned myosin filaments. This theory considers myosin II filaments as force dipoles embedded in an elastic medium and predicts registered organization of these filaments by attractive forces between these dipoles.

2. Myosin II filaments and their arrangement in cross-striated muscle

Each myosin II molecule is a hexamer consisting of two heavy chains and two pairs of different light chains. Each heavy chain consists of an N-terminal head (motor) domain and a long α-helical rod/tail domain connected by a neck (lever arm) domain (figure 1a). A two-headed myosin II molecule is assembled by formation of a coiled coil between the α-helical tail domains of two heavy chains forming a myosin II rod. In a majority of myosin II heavy chain types, the short sequence at the C-terminus is non-helical and forms a so called C-terminal tailpiece [5–7]. Two different light chains associate with the neck regions of each heavy chain (figure 1a) [3,4]. These myosin II hexamers assemble into different types of superstructures in different cell types. The most complex and highly ordered myosin II organization is found in cross-striated (skeletal and cardiac) muscles [4,8] (figure 1b).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

(a) A cartoon depicting the generic organization of myosin II molecules. Myosin II is a hexameric protein composed of two heavy chains and two pairs of light chains: the essential light chains (ELC) and the regulatory light chains (RLC). Each myosin II heavy chain is composed of an N-terminal motor domain containing ATP- and actin-binding sites, a neck domain (lever arm) that binds ELC and RLC, α-helical rod domain and C-terminal tail. The two myosin II heavy chains dimerize via interactions between the rod domains, together forming an α-helical coiled coil. (b) Modes of formation of myosin II dimers. Two myosin molecules can interact in either antiparallel (left) or parallel (right) fashion. The interaction is mediated by the rods and tails of individual myosin molecules and depends on the charge distribution along the rods. The shifts between individual myosin II molecules in the dimers can have only particular discrete values. (c) Three general types of myosin II filaments. Left: Myosin II filaments of striated muscles consist of several hundreds of myosin II molecules and accessory proteins (see text). The myosin heads are symmetrically located on the sides of bipolar filaments leaving a bare zone in the middle. The length of human striated muscle filaments is about 1600 nm. Middle: Myosin II filaments in non-muscle cells have a uniform length of 300 nm and consist of approximately 30 myosin hexameric molecules (for A and B heavy chain isoforms) or approximately 15 molecules (for C isoform). Right: Smooth muscle myosin II molecules usually form long side-polar (face-polar) filaments lacking the bare zone. (Online version in colour.)

Muscle tissue is highly contractile and produces a wide range of active forces in order to move the skeleton of all bilaterian animals. The basic contractile unit of cross-striated body muscles is known as the sarcomere [4,8,9]. Each stereotyped sarcomere is bordered by two Z-discs, which anchor the barbed (plus) ends of two arrays of highly organized polar actin (thin) filaments, facing in opposite directions (figure 2). This organization ensures that the linear actin (thin) filaments face with their pointed (minus) ends towards the centre of the sarcomere, where the bipolar myosin (thick) filaments are located. Each Z-disc contains many accessory proteins: most importantly, actin filament crosslinker α-actinin, α-actinin binding ZASP family proteins (ZASP stands for ‘Z-band alternatively spliced PDZ motif-containing protein’), actin filament capping cap-Z protein [8,14], as well as formins [15–17]. These proteins crosslink, cap, or promote polymerization of actin filaments. The Z-disc also anchors the N-terminus of the largest human protein named titin, which extends with its C-terminus all the way to the M-line in the middle of the sarcomere (see below) and thus stably links the actin with the myosin filaments (figure 2) [18]. In its fully extended state, titin is more than 1.5 µm long, which fuelled the idea of it being a molecular ruler that determines the stereotyped length of sarcomeres [19]. The sarcomeres in relaxed human skeletal muscles are between 3.0 and 3.4 µm long, depending on the muscle sub-type [20,21]. In mammalian cardiomyocytes, the sarcomeres are shorter (1.6–2.3 µm in length) [10,22,23]. In mature muscles, titin molecules span half of the sarcomere, such that 6 titin molecules connect each myosin filament with Z-discs at each side [24].

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Striated muscle morphology. (a) Top: A cartoon depicting skeletal muscle fibres attached to tendons at both ends. Note the straightness of the muscle fibres indicating tension in the muscle-tendon system. Middle: Zoom-in showing the linear organization of striated myofibrils. Note the lateral alignment of the myofibrils resulting in muscle fibre cross-striation. Bottom: Further zoom-in depicting the organization of an individual sarcomere in striated muscle. Note the stereotyped length of sarcomeres (3.0–3.4 µm), which is generally orders of magnitude smaller than the total muscle fibre length. In cardiomyocytes, the sarcomere length is slightly smaller than in skeletal muscles (1.6–2.3 µm) [10]. This scheme was modified from Lemke & Schnorrer [11]. (b,c) Immuno-stainings of myofibrils in a cultured rat cardiomyocyte (b, MHC in red, myomesin in yellow) and in a Drosophila leg muscle (c, actin in red, Z-disc associated protein kettin [12] in green) displaying highly registered myofibrils resulting in cross-striated muscle patterns. (d) Not all Drosophila muscle types are cross-striated. Although the individual myofibrils of the Drosophila flight muscles (actin in red, kettin in green) are highly regular, they are not registered with their neighbouring myofibrils. See Spletter & Schnorrer [13] for a brief review of different muscle types. Scale bars correspond to 10 µm. Part b is courtesy of Dr. Yfat Yahalom-Ronen (Weizmann Institute of Science), c and d were acquired by Christiane Barz (Max Planck Institute of Biochemistry).

Thick filaments of striated muscles are very large, elongated molecular complexes of about 1600 nm in length and 30 nm in diameter (in vertebrates), containing about 300 muscle myosin II isoform hexamers [25] together with a number of accessory proteins [10]. The muscle myosin II hexamers are organized within a bipolar filament, with the heads present at both ends and a bare zone in the middle part of the filament (figure 1c). The exact structure of a thick filament is somewhat different for different species, but the general principles proposed more than 40 years ago [26] are still valid for all types of striated muscle thick filaments [4]. Recent cryo-electron tomography studies revealed the details of thick filament organization in insect flight muscle at almost atomic resolution [27]. The filaments are formed by interactions of the rod domains of the individual myosin heavy chain molecules. The central bare zone is built of anti-parallel myosin rods, being the base for a perfectly bipolar filament. The laterally protruding myosin motor heads are organized in a helical lattice [4,28]. It was established that only particular shifts between myosin rods interacting in antiparallel or parallel fashion are allowed (figure 1b) [26], which can be explained by an uneven distribution of positive and negative charges along the length of the rod [29,30]. These permitted configurations are thought to determine the organization of the entire filament [4,26].

In mature vertebrate sarcomeres, a prominent myosin accessory protein is myosin binding protein C (MyBP-C or C-protein) [31]. The C-terminal domains of MyBP-C run along the myosin filament surface, while the N-terminus extends toward neighbouring actin filaments [32] and is proposed to modulate the actomyosin interactions during muscle contractions [33]. In insects and other invertebrates, the thick filaments have a core made of the protein paramyosin, which is homologous to the rod part of the myosin II molecule [27]. Different versions of paramyosin are present in functionally different insect muscle types [34,35]. Another feature of myofibrils in the specialized stretch-activated insect indirect flight muscles is that individual fibrils do not align with the neighbours (the organization type is called ‘fibrillar’ muscle, figure 2c,d) [35]. However, the molecular mechanism or particular proteins responsible for the lack of alignment of the individual myofibrils in insect flight muscles remains to be identified.

At a larger scale, all myosin filaments in each sarcomere are organized into an ordered array aligned at a structure known as the M-line that connects the centre of the bare zones of the individual filaments (figure 2a) [8,9,36]. The M-line consists of several protein components among which the myomesin family elastic proteins, including M-protein, play a major role [36,37]. Another large sarcomeric protein that is found at the Z-disc and the M-line of mature vertebrate muscles is obscurin. It has significant similarities to the titin protein family [38]. However, in insects, obscurin is largely present at the M-line, where it is required for the symmetrical assembly and alignment of the thick filaments during flight muscle development [39].

At the extreme end are the very long and well-aligned thick filaments found in the catch muscle of certain molluscs (e.g. in anterior byssus retractor muscle (ABRM) muscle of Mytilus), in which the thick filaments are up to 25 µm long and stacked up to 75 µm in diameter [40]. This particular organization is believed to be caused by particularly high paramyosin concentrations present in the core of the thick filament [41]. These very long thick filaments in mollusc catch muscle contain a long region of acto-myosin overlap and thus enable a very high force production, three to four times greater than other muscles [42].

Actin filaments in the sarcomere are also associated with several important accessory proteins. One of them is nebulin, another giant protein, which in vertebrate muscle spans almost the entire length of each actin filament. Nebulin regulates actin dynamics and possibly determines thin filament length [43]. In addition, each actin filament is associated with two strands of tropomyosin molecules assembled head-to-tail along two helices of the actin filament [4,44]. The regulatory protein troponin is periodically located along actin/tropomyosin filament and regulates the position of the tropomyosin strands on actin filament in a Ca⁺⁺-dependent manner. This ensures that an interaction of the myosin II motor heads with the actin filament, and thus productive force generation, is possible only when the Ca⁺⁺ concentration in the cytoplasm exceeds a threshold value [44]. This type of contractility regulation is specific for striated muscles but not for smooth muscles or non-muscle cells which contain several isoforms of tropomyosin but do not contain troponin.

Most insect body muscles and probably all vertebrate skeletal muscles as well as the heart are cross-striated fibres. This means that in addition to the precise organization within each sarcomere and the periodic organization of the sarcomeres into chains called myofibrils, even the myofibrils are eventually organized in registry, such that Z-bands of the neighbouring myofibrils are minimally shifted relative to each other (figure 2). In mature cardiomyocytes, the Z-bands of the neighbouring myofibrils are connected by intermediate filaments consisting of the protein desmin [8,45]. In addition, the Z-bands of the peripheral myofibrils are linked to the transmembrane integrin molecules and glycoproteins of the dystroglycan complex that together form structures called ‘costameres’, the regions where the cross-striated muscle fibres adhere laterally to the extracellular matrix (ECM) [45,46]. At least as important is the solid anchorage of the two terminal Z-disc of each myofibril at the long muscle ends to the ECM, a process that is also integrin mediated [47–49]. Among the integrin linker proteins that mediate the association of integrin with myofibrils at costameres or terminal Z-discs are talin, kindlin, vinculin, filamin and other molecules found also in cell matrix adhesions of other types [46,50,51].

How the ordered organization of myosin molecules and the entire structure of sarcomeres and myofibrils emerge during muscle differentiation is still poorly understood. The sarcomeres do not exist in isolation. They are connected at their symmetrical Z-discs forming myofibrils (figure 2). Each myofibril spans the entire muscle cell, mechanically connecting the two muscle-tendon attachment sites in the case of human skeletal or insect body muscles. In cardiac muscle, the myofibrils of the cardiomyocytes are connected with intercalated discs (fascia adherens), the junctions between the neighbouring cardiomyocytes [52]. In Drosophila, it was shown that mechanical tension is essential for myofibrillogenesis in vivo [49]. During muscle development myotubes connect both ends to tendons and build up mechanical tension. In turn, tension triggers the simultaneous self-organization of actin, myosin and titin complexes into immature myofibrils, which span from one muscle-tendon attachment site to the other [11,49]. These immature myofibrils then become contractile and their spontaneous twitchings are required for the lateral alignment of neighbouring myofibrils into the highly registered lateral organization of Z-discs and M-lines in cross-striated muscle [53]. These conclusions from developing insect muscles are also supported by data gained in the developing zebrafish body muscles, which showed that developmental contractions are required to form regular cross-striated sarcomeres [54].

Apart from actin and muscle myosin filaments, titin is essential to assemble sarcomeres and myofibrils [55–58]. As titin stably connects the Z-discs with the thick filaments and contains an endogenous mechanical spring domain, it is the major source of passive muscle elasticity in mature muscle fibres [43,59]. The extension of titin's spring occurs at the low pN range and is fully reversible [60,61]. Hence, it is very likely that forces across the titin molecule, which will eventually connect thin and thick filaments, play an important role in the myofibril assembly process [11].

Thus, it is becoming increasingly clear that forces generated by the myosin filaments play an important role in the myofibril self-organization process. A theory explaining one aspect of this process, the registered organization of myofibrils, will be discussed below.

3. Registry of myofibrils in striated muscle in culture and in vivo

The organization of myofibrils into registry occurs by lateral ordering of neighbouring myofibrils. This order is manifested by the alignment of the location of their Z-discs and M-lines (figure 2) [53,62].

Experimental observations of cross-striation in embryonic muscle cells cultured on synthetic, deformable substrates [63–66] suggest that the registry of neighbouring myofibrils is a mechanically regulated process. The extent of striation is maximal on optimally rigid substrates, which are neither too soft nor too rigid [63,64]. The observation of strong striations indicates the local registry of myofibrils. By systematically quantifying the registry while varying the rigidity of the underlying gel substrate, it was demonstrated that both striation and the contractile strains produced by beating cardiac muscle cells show a maximal value in a certain range of substrate stiffness [64,65]. Interestingly, the intermediate stiffness found for maximal striation and beating strength of about 5 kPa corresponds to the native stiffness of embryonic heart tissue. The requirement of proper mechanical forces during muscle development is consistent with aforementioned results that the twitchings of immature myofibrils are required for their lateral alignment [53].

This dependence of muscle structure and function on the stiffness of the environment and the forces produced by the muscle can be understood in terms of elastic interactions between fibrils by deformations of the underlying substrate induced by actomyosin contractile forces. The theoretical treatment of the properties and interactions of active force dipoles [67] provides a coarse-grained model for a cell's cytoskeletal contractility. Such ideas have been applied to quantitatively measure the mechanical response of single cells [63,68,69] or cytoskeletal elements in various situations [70]. Examples include the orientation of fibroblasts under an external stretch [71] or the substrate rigidity-dependent mutual orientation of stress fibres of mesenchymal stem cells [72]. In the case of muscle, each myofibril is modelled as a periodic array of contractile, equal and opposite forces of each of the myosin II heads; these are termed force dipoles (figure 3a,b) by analogy with electrical dipoles that are equal and opposite charges separated by a finite distance. (We stress, however, that the force dipoles are only analogous to electrical dipoles and not identical with them; even the mathematical properties of the two are different since charge is a scalar (direction-independent) quantity, while force is a vector (direction-dependent) quantity; below we denote vectors by boldface and their components by an index.) The force dipoles deform the elastic substrate via the coupling of the contractile acto-myosin to the crosslinkers (Z-discs); these forces are then exerted on the substrate at the costamere adhesion sites of the Z-discs [62].

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Illustration of striated muscle myofibrils interacting through deformations of soft substrate. (a) View of whole muscle cell cultured on an elastic substrate. Three myofibrils that span the cell are shown, with dark squares indicating the position of the myosin filament. Z-discs are not shown. Each end of a myofibril is anchored to a grey rectangle symbolizing a muscle-tendon attachment site in the case of human skeletal or insect body muscles, or an intercalated disc connecting neighbouring cardiomyocytes in the case of heart muscle. (b) Each myofibril, a series of sarcomeres, is mechanically coupled to the elastic substrate at the Z-discs through the transmembrane integrin-containing adhesive complexes known as costameres. (c) Schematic of deformation pattern transduced in the substrate by a single fibril. The light and dark regions represent regions where the substrate is expanded or contracted. The deformations decay with transverse distance away from the myofibril. (d) A pair of neighbouring fibrils represented as arrays of force dipoles. They are out of registry by a distance Δ and experience a substrate-mediated elastic force that tends to drive them towards mutual registry [62].

We now sketch the basic elements of the force dipole theory of deformation-induced interactions among distinct and well-separated acto-myosin units using a model in which the intervening elastic medium (or substrate) is represented as a linear elastic medium (i.e. where the stress is proportional to the strain). The displacement at a point r located on the surface (z = 0) of a semi-infinite linearly elastic medium (or substrate) caused by a force acting in the direction j = x,y at another location (chosen to be the origin) on the surface of this semi-infinite medium is given by Landau et al. [73], 3.1where u_i(r) is the displacement in the ith direction at point r = (x,y) on the surface of the medium caused by the jth component of the force F at the origin; the relevant elastic constants are the Young's modulus, E, and Poisson ratio, ν, of the elastic substrate. For nearly incompressible gels such as polyacrylamide, this can be taken to be at short times when water flow is negligible. The Young's modulus, a measure of substrate stiffness, can typically be varied from kPa to 100s of kPa [64]. If instead of a point force as in equation (3.1), there is a pair of equal and opposite forces (e.g. from the two heads of a bipolar myosin filament bound to anti-parallel actin filaments) that are separated by a small distance (corresponding to the contractile actomyosin force dipole denoted by P₁) along the x-direction (see figure 5b,c), the displacement is related to the derivative of the expression in the right-hand side of equation (3.1) with respect to x [74]. The resulting medium strain, which is a spatial derivative of the displacement u_i(r) in equation (3.1), is . Only the strain, which is a measure of deformation (i.e. shape or volume change) of the elastic medium—but not the displacement—can give rise to physical forces.

Considering about 100 myosin filament heads in a sarcomere [25], each producing a contractile force of about 1 pN [75], a reasonable estimate for the contractile force per sarcomere is about 100 pN [62]. In this case, each force dipole is the pair of forces acting on the elastic substrate through the costameres at the Z-discs as shown in figure 3b. The spatial extent of such a dipole therefore corresponds to the size of a sarcomere. The transduction of these sarcomeric forces into the medium or substrate by adhesions coupling the two, induces a spatially, periodically patterned strain field within the substrate with alternating regions of compression and expansion shown respectively in figure 3c as dark and light regions. Given the strains induced by one myofibril, a sarcomere in the neighbouring myofibril located in an expanded region of the medium can then actively compress this region, restoring the medium closer to its undeformed state and thereby lowering the overall deformation energy of the elastic medium. This medium-based bias in the positioning of a neighbouring contractile unit (figure 3d), templates a registered configuration of neighbouring fibrils.

By including the molecular noise inherent in cells, such a theoretical approach can map the substrate rigidity dependence of registry onto that of the measured beating strains generated by cardiomyocytes [76]. The good agreement of theory with experiments relates the correlated beating of heart cells to the structural registry of the myofibrils, which in turn is regulated by their elastic environment. Finally, we comment on the possibility of registry driven by mechanical interactions in muscle tissue in vivo. Both the cell culture experiments [63] described in this section and in vivo studies of insect muscle [49] show that mechanical factors such as stiffness and tension are important in the registry or cross-striation of myofibrils. The proposed theoretical model is based on elastic deformations that arise from actively contractile myofibrils that are under tension. Further, immature muscle fibrils (premyofibrils) are found to be less ordered in both cases. This is consistent with a gradual progression towards order suggested by a force-driven mechanism.

4. Is there higher-order organization of myosin II filaments in smooth muscle cells?

Vertebrate smooth muscle cells, as evident from their nomenclature, have no apparent striations and therefore do not exhibit the highly ordered sarcomeric organization typical for striated muscle [4,77]. The homologues of Z-discs in smooth muscle are the dense bodies, which in analogy to Z-discs keep together the barbed ends of the actin filaments and are enriched in α-actinin [78]. As in the Z-discs of vertebrate striated muscle, the dense bodies are linked by desmin intermediate filaments, which are very abundant in smooth muscle [78,79]. Structures similar to dense bodies, known as dense plaques, are associated with the plasma membrane and mediate the adhesion of smooth muscle cells to the ECM. These structures are homologous to focal adhesions in non-muscle cells and contain characteristic proteins such as vinculin and integrin [80,81]. There is no homologue of the M-band in smooth muscle cells and the smooth muscle myosin filaments do not form ordered arrays, but rather interdigitate with the actin filaments out of registry. The organization and dynamics of the filaments formed by smooth muscle myosin molecules containing a smooth muscle-specific heavy chain isoform is generally different from that in striated muscle. In the majority of vertebrate smooth muscles, myosin filaments are not bipolar but contain myosin heads all along their length (so-called face-polar or side-polar filaments; figure 1c) [82,83]. Such filaments or ribbons can be polymerized from purified smooth muscle myosin in vitro [83]. It is even claimed that the length of myosin II filaments in smooth muscle cells is not constant but varies within a broad range [84] even though there is no complete agreement on this subject [85,86].

Smooth muscle cells do not express troponin and, therefore, the Ca⁺⁺-dependent regulation of the actin and myosin filament interactions is significantly different from that of striated muscle. Both ATPase activity of the smooth muscle myosin II and the ability to assemble into filaments are regulated by phosphorylation of the regulatory light chains (RLCs) [87–91]. A similar type of myosin II regulation occurs in non-muscle cells [3]. In cardiac striated muscle, RLC phosphorylation is constitutive and seems to be needed for the optimal contractility function [92].

Even though the degree of order in the organization of smooth muscle actin and myosin filaments is apparently less than in striated muscle, the degree of actomyosin order in the smooth muscles could be underestimated. In fact, some structures in smooth muscle could exhibit rather regular organization. For example, vinculin-containing dense plaques are quite regularly arranged in cultured smooth muscle cells [80]. Further regularity in the organization of the actomyosin cytoskeleton in smooth muscle cells will hopefully be revealed in future studies using super-resolution light microscopy.

5. Ordered arrays of myosin II filaments in non-muscle cells

The first indication that non-muscle cells can also assemble myosin filaments was obtained long ago using immuno-electron microscopy [93]. Higher quality images of such filaments were later provided by the technique of platinum replicas of permeabilized cells, from which the actin filaments were removed by incubation with the actin depolymerizing protein gelsolin [94,95]. Transmission electron microscopy of such replicas revealed that in cultured fibroblasts, numerous bipolar non-muscle myosin filaments of uniform length (approx. 300 nm) were present. These often appeared as groups of parallel filaments termed ribbons or stacks [94]. Later, such stacks were also observed in other cell types using similar techniques [95,96]. Biochemically, non-muscle isoforms of vertebrates are represented by three types of heavy chain, MHC-A (Myh9), -B (Myh10) and -C (Myh14), which interact with the same essential and regulatory light chains [3,97]. Each of the non-muscle myosin II types can assemble into filaments in vitro [98]. It was documented that myosin IIA and IIB molecules can co-assemble into a single filament [99–101], even though these isoforms usually segregate from each other resulting in anterior and posterior accumulation of myosin IIA and IIB, respectively, in polarized cells [101–103]. Interestingly, all isoforms of non-muscle myosin II can co-assemble with myosin 18A, which by itself can form antiparallel dimers, but not large filaments, and lacks the ATPase activity [104]. Thus, myosin 18A may regulate the assembly of other non-muscle myosin II filaments in vivo or mediate their interactions with some associated proteins [104].

Experiments with selective knockdown of myosin IIA or IIB revealed that myosin IIA tends to contribute more to overall cell contractility, while myosin IIB tends to be more involved in the regulation of cell front-rear polarity, guidance of cell migration, and matrix remodelling [3,105–109]. Partition between myosin IIA and IIB functions may also depend on cell type.

Regular fluorescence microscopy has insufficient resolution for the visualization of the 300 nm small myosin II filaments in non-muscle cells. However, even in the early immunofluorescence studies of myosin II localization, the obvious periodic distribution of myosin II entities (striations) was observed [110,111]. The recent introduction of super-resolution imaging, especially structured illumination microscopy (SIM) [112,113], has permitted the better visualization of myosin II filaments and super-structures formed by such filaments, as well as their dynamics in living cells [114–117].

The myosin filaments and filament stacks in the cells are mainly associated with bundles of actin filaments. The main myosin II-containing structures are the organized, parallel actin bundles or stress fibres typical of polarized fibroblast-like cells (figure 4). Related types of structures comprising myosin filament stacks are the so-called arcs or circumferential actin bundles that emerge at the periphery of spreading cells [114]. A recent study of the distribution of myosin IIA filaments revealed that in both stress fibres and arcs, myosin filaments are oriented parallel to the actin filaments and co-localize with the regions enriched by pointed (minus) ends of actin filaments associated with the protein tropomodulin (figure 4) [117]. This means that myosin filaments connect arrays of actin filaments with opposite polarity. Consistent with previous studies [93,94,111], along the length of the stress fibres or arcs, the myosin filaments are distributed in a periodic fashion, alternating with regions enriched in the actin filament crosslinking proteins α-actinin-1 and -4 [117]. The α-actinin enriched regions are also enriched with the barbed (plus) ends of actin filaments, which can incorporate monomeric (G) actin [117]. The sites with the most prominent G-actin incorporation are focal adhesions, molecular complexes connecting stress fibre ends with transmembrane integrin molecules [118].

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Myosin IIA filaments and filament stacks in REF52 rat embryo fibroblasts. (a) Complementary distribution of myosin II filaments and α-actinin-enriched domains in REF52 cells co-expressing α-actinin-1-mCherry (blue) and RLC-GFP (yellow). The RLC images of the boxed regions shown at high magnification in the inset on the right display a putative myosin filament and a myosin filament stack, as indicated in the schematic. The images were taken using a Nikon 3D-SIM microscope. Scale bar, 5 µm. (b) Schematic depicting the organization of actin and myosin II filaments in the quasi-sarcomeric units of stress fibres or transverse arcs. A single ‘sarcomere’ is shown. Barbed ends of actin filaments (visualized by G-actin incorporation) are located in the α-actinin-enriched zones, while pointed ends (decorated by tropmodulin 3) overlap with myosin filament stacks.

Periodic organization of myosin II and α-actinin in stress fibres and circumferential arcs is somewhat similar to the organization of periodic sarcomeres in myofibrils (see above). However, there are several important differences between the structure of myofibrils compared with that of non-muscle stress fibres or arcs. First of all, the myosin II filaments of non-muscle cells are about fivefold shorter than thick filaments in striated muscles (300 nm versus 1600 nm). Further, the widths of the α-actinin-rich zones in stress fibres and centrally located arcs, unlike those in Z-discs, are not uniform, but show a broad length distribution (in a range between 300 and 1000 nm). In peripheral arcs, such zones are even wider than in central arcs and in stress fibres [117]. Moreover, while sarcomeric contraction causes a shortening of the distance between Z-discs, the contraction of stress fibres or arcs seems to involve a decrease in the lengths of the α-actinin-rich zones themselves [117].

Homologues of the M-line protein obscurin which hold together myosin filaments in myofibrils [39] were recently found in several types of non-muscle tissues and organs including brain, skin, kidney, liver, spleen, and lung [119] as well as in cultured epithelial cells [120]. However, direct evidence that obscurin or obscurin-like proteins mediate the link between myosin II filaments in non-muscle cells is missing. Moreover, even though there are data in the literature indicating the presence of titin-related molecules in non-muscle cells [121], and some authors attribute the high stress fibre elasticity to the presence of such molecules [122,123], the existence of the giant elastic titin-like filaments connecting the actin and myosin II filaments in non-muscle cells has not been confirmed. The elastic response of the stress fibres may rather be related to the function of the mechanoresponsive α-actinin-binding protein zyxin [124,125].

The deep structural differences between myosin filament organization in myofibrils as compared to the stress fibres and arcs of non-muscle cells are in line with the very different rates of filament turnover. In experiments with fluorescence recovery after photo-bleaching (FRAP), the recovery of the fluorescence of labelled non-muscle myosin light or heavy chains was observed in less than a minute suggesting rapid turnover of the myosin filaments [117], while the characteristic FRAP time for myosin filaments in muscle cells exceeded an hour [126].

Interestingly, besides stress fibres and circumferential arcs, registered stacks of myosin II filaments were also found in other domains of the actin cytoskeleton. SIM microscopy revealed such structures in the contractile ring of dividing human cells during cytokinesis [115]. Periodic distribution of myosin II clusters in the contractile ring was also noticed in other studies [127]. The architecture of the actomyosin contractile ring is not fully understood even though several models have been suggested [128–130]. The periodic distribution of myosin stacks in the contractile ring highlight an interesting similarity with circumferential actin bundles in interphase cells. Another type of actomyosin ring containing periodically distributed myosin II filament stacks is represented by adhesion belts of epithelial cells in the highly ordered epithelial layer of the organ of Corti [131]. Here, unlike other types of myosin filament structures in non-muscle cells, the myosin filaments are formed by myosin IIC isoform molecules. A peculiar feature of these adhesion belts is that they are organized in registry so that arrays of myosin filaments in one cell are located exactly opposite to the symmetrical array in a neighbouring adherent cell [131]. An ordered organization of myosin IIA filament clusters was also found in the cadherin-mediated junctions between human intestinal Caco-2 epithelial cells [132].

How do these highly organized myosin II structures form and what could be their functions in non-muscle cells? To answer the first question, the processes of formation and assembly of myosin filaments should be studied. Myosin II in non-muscle cells is regulated mainly by phosphorylation of RLCs. In particular, myosin II molecules assemble into filaments only if the RLCs are phosphorylated [3,133]. This phosphorylation is mediated by several enzymes, notably myosin light chain kinase (MLCK) and Rho kinase (ROCK). ROCK, in addition, phosphorylates and inactivates the myosin light chain phosphatase (MLCP), which normally antagonizes with RLC phosphorylation [3]. Thus, ROCK and its upstream activator, small G-protein Rho, act as master regulators of myosin II filament formation in non-muscle cells. Inhibition of ROCK by a specific drug Y27632 results in the total disassembly of myosin II filaments in the central part of the cell [117]. There could, however, be spatial differences in the regulation of RLC phosphorylation. At the cell periphery, formation of the new myosin filaments is less sensitive to Y27632 treatment [116], probably because in this region it depends more on MLCK than on ROCK [116,134].

The assembly of myosin II molecules into filaments depends also on several other regulatory events including phosphorylation of the myosin II heavy chain [135–137] and action of a myosin chaperone UNC-45a [138]. Signalling networks regulating these processes are not completely known. In general, the conditions under which assembly of myosin II molecules into filaments can occur are satisfied at the cell periphery. This probably includes the activity of small G-proteins of the Rho family and their downstream protein kinase targets, which are required for proper phosphorylation of the subunits of myosin molecules. The assembly of new myosin filaments was found at focal adhesions, where the active G-protein Rac1 together with protein kinase C (PKC) stimulate the phosphorylation of myosin heavy chain at serine 1916 [137]. The assembly of new myosin II filaments was also observed in other peripheral regions of the cell such as at the lamellipodium–lamellum interface [115–117,139]. It might be that formation of new myosin II filaments in vivo is facilitated by their interactions with newly formed actin bundles which appear as a result of debranching of the Arp2/3 network in lamellipodia. This process may lead to formation of the transverse arcs at the rear border of lamellipodia [140–142].

As soon as a myosin filament is formed, it appears to be associated with a thin bundle of actin filaments [116,117] and begins to move towards the cell centre. In the course of such motion, the filament is incorporated into pre-existing superstructures (stacks) formed by other filaments. This process appears to underlie the formation and maintenance of the observed myosin stacks. It was recently shown that in addition to this process (known as concatenation), myosin filaments can undergo splitting or partitioning, such that a single filament is transformed into two separated ‘daughter’ filaments [115,116]. Such partitioning of myosin filaments was shown to depend on actin filament polymerization as well as the activity of MLCK [116]. The exact mechanism of the myosin filament partitioning is yet to be established conclusively. Obviously, understanding this phenomenon would shed light on the mechanism of myosin filament self-organization in non-muscle cells.

While expansion of filaments based on their partitioning could be one of the mechanisms by which myosin filaments form stacks, Hu et al. demonstrated several events when the movement of existing myosin filaments into registry with another and that of two existing myosin stacks into registry with each other occurred [117]. The formation of myosin filament stacks is sometimes preceded by a movement of myosin filaments for a distance of about 1 µm. Such long-distance movement is consistent with an emergent force between the filaments acting over a long range. In vitro experiments with purified myosin IIB have demonstrated formation of complexes consisting of a small number of neighbouring filaments [98] but not the large micron-sized stacks where registry occurs over more than 10 filaments observed in cells [117].

To systematically investigate the molecular requirements for myosin filament stack formation, Hu et al. disrupted the myosin filaments by cell treatment with ROCK inhibitor Y27632 and observed the recovery of the filaments and the filament stacks under various conditions. It was shown that not only the mechanochemical activity of myosin, but also the dynamics and organization of the actin filaments, are important for stack formation. Inhibitors of formin-driven actin polymerization or depolymerization did not interfere with the recovery of myosin II filaments but blocked their assembly into stacks. The actin-associated proteins involved in the process of stack formation include the formin Fmnl3, known to be a potent activator of actin filament elongation [143]. Similarly, formin mDia1 was shown to be required for the ordered organization of myosin IIA filaments associated with cell–cell junctions [132]. Cofilin1, a known actin filament severing and depolymerizing factor [144], was also shown to be involved in myosin IIA filament stack formation [117]. Finally, the formation of these stacks depended on the actin cross linking protein α-actinin-4 and to a lesser degree on α-actinin-1 [117].

To explain the process of actin-dependent myosin filament stack formation in non-muscle cells and, in particular, the emergence of a force that acts between myosin filaments with a range much larger than a typical molecular scale, we adapted the theory developed for the explanation of registered organization of sarcomeres in striated muscle.

6. Interactions between myosin II filaments through the intervening actin network: a theoretical model

In this section, we summarize a theory that shows how myosin filaments can interact via their mutual deformations of the intervening elastic medium. As discussed above for cells cultured on soft deformable substrates, parallel organization of stress fibres [72], as well as registered organization of myofibrils in cardiomyocytes [76], can be explained by the effective interaction of cellular contractile elements through the elastic substrates. We now adapt these ideas to the cytoskeleton of cells cultured on rigid substrates where interactions between myosin filaments via deformations of the relatively disorganized, intervening cytoskeletal network that surrounds and connects them [117,145,146] may register the myosin filaments into stacks.

The theory is motivated by the following observations [117]: (i) Registered stacks develop dynamically and can involve micron scale displacements of the myosin II filaments over a timescale of minutes. (ii) Myosin filaments move in association with actin fibres and the registry requires actin dynamics; the stacks disassemble when actin polymerization is inhibited. (iii) Registry is lost when contractility is blocked by contractility inhibitors such as blebbistatin. The sensitivity to actomyosin dynamics cannot be attributed to only molecular-level, local turnover of actin and myosin filaments, since these kinetics that take place over tens of seconds, as determined by FRAP experiments, are much faster than the time-scales of minutes required for registry to be established [117].

These considerations lead us to focus theoretically on the role of force transmission due to actomyosin contractility instead of direct, molecular associations of myosin filaments. Even in the absence of a soft deformable substrate, elastic interactions between neighbouring actomyosin elements on organized stress fibres may be mediated by the actomyosin-induced deformations of the intervening, disordered network of actin and possibly other cytoskeletal components (henceforth termed intervening network or ‘IVN’). The existence of such a network can be inferred from previous light and electron microscopy studies. In particular, an actin intervening network can be seen in the recent STORM images [145] as well as in earlier electron microscopy studies [146]. We hypothesize that myosin filaments are interacting with each other through such a network (figure 5a–c). An actomyosin unit on one stress fibre deforms the IVN. This deformation interacts mechanically with a contractile actomyosin unit on a neighbouring stress fibre and results in forces (transmitted by the IVN) that cause the myosin filaments on the two neighbouring stress fibres to be ‘pushed’ into registry with each other to form the stacks across stress fibres that are not in direct molecular contact with each other [117].

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

(a) Cartoon showing myosin filaments associated with neighbouring stress fibres with an intervening network of disordered actin filaments (IVN) between the neighbouring fibres. A similar picture applies to transverse arcs. As the sketch shows, the same actin filament in the IVN may bind two neighbouring stress fibres and thus physically connect and mechanically couple them. (b) Schematic geometry of interaction of two parallel force dipoles (P₁ and P₂) separated by a transverse distance d. Each dipole corresponds to the pair of contractile forces applied by the bipolar myosin filament at the points of adhesive contact with the IVN. The two dipoles representing the pair of myosin filaments are initially shifted with respect to each other (out of registry) by a distance Δ along the x-direction. The dipole P₁ deforms the elastic medium (IVN) which thus exerts a force on the second dipole P₂. The component of this force in the x-direction, F_x, tends to move P₂ into registry with P₁. (c) The same geometry as in (b) except now we consider a line of dipoles uniformly separated by a distance a as a model for an entire fibre. The deformation of the elastic medium by the line of dipoles causes a spatially periodic force on a single dipole, P₂, a transverse distance d apart and shifted from registry by Δ as in (b). (d) Plot showing elastic medium-mediated force (for a Poisson's ratio ν = 0.5, incompressible medium) on an actomyosin unit (force dipole) by another dipole as shown in (b). We plot the force, F_x on a dipole transmitted through the elastic medium (the intervening IVN), as a function of the distance of registry mismatch Δ/d. A positive displacement of P₁ by a distance Δ away from registry, results in a force that acts on P₁ in the negative x-direction, demonstrating a tendency towards registry (Δ = 0) of the two dipoles. (e) Plot showing elastic medium-mediated force F_x (for a ν = 0.5) on an actomyosin unit (force dipole P₂) by a line of dipoles (as a model of a stress fibre) as shown in (c). This shows that the elastic medium-mediated forces promote registry of P₂ with the nearest dipole on the line of dipoles. (f) Plot of the transverse force F_y between two force dipoles oriented in and offset along the x-direction by a distance Δ, as shown (b), as a function of their transverse separation d. The negative values of F_y for a range of transverse separations implies that the two dipoles with parallel organization are acted upon by attractive elastic forces mediated by the medium. (g) Plot of the transverse force F_y on a force dipole by a line of dipoles (as a model of a stress fibre) as shown in (c). This shows that the elastic medium-mediated forces promote attraction of a single actomyosin unit towards an already formed actin bundle. All forces are rescaled here by suitable scales for force dipole, separation distance and elastic modulus, and can be taken to be in arbitrary units.

ATP hydrolysis-induced activity causes actin-bound myosin II motors to produce pairwise contractile forces on actin filaments of reverse polarity [75]. Such a ‘pinching’ force pattern can be modelled as a force dipole [67,70] of magnitude P which comprises a pair of equal and oppositely directed myosin forces (figure 5b,c), each of magnitude F, separated by a distance a, so that P = F a. Two such contractile actomyosin units located on neighbouring bundles can interact mechanically through their mutual deformations of the intervening elastic medium with which they are in contact. The force of one such unit results in an elastic stress and displacement of the IVN near a neighbouring actomyosin unit located on an adjacent actin bundle, which then can respond by translating or orienting in response to the deformation of the IVN caused by the first unit. The actin filament IVN between neighbouring actin bundles (figure 5) is a crosslinked gel and can be modelled as a continuous, deformable elastic medium at sufficiently short timescales of seconds [147] appropriate to the local force transmission in our case. The continuum description is valid for a dense IVN (i.e. as long as the pore size of the disordered network is small compared to the size of a force dipole), here an actomyosin unit comprising the myosin filament and associated actin filaments.

We now consider the elastic deformation of the IVN in the presence of two such force dipoles (figure 5b), P₁ at the origin of coordinates and P₂ at r = (Δ, d, 0), representing two neighbouring myosin filaments on two adjacent actin bundles (oriented in the x-direction) that are separated by a transverse distance (in the y-direction), d. The corresponding force in the x-direction exerted on the dipole P₂ by the elastic IVN due to its deformation by forces exerted by the first dipole, P₁, is proportional to P₂ (and inversely proportional to the rigidity of the IVN, E) as well as to the derivative of the local stress field (which is calculated from the displacement in equation (3.1)) induced by P₁ [74], 6.1

The force between a pair of dipoles given by equation (6.1) is plotted in figure 5d as a function of the distance Δ by which they are misregistered in their mutual positions in the x-direction. This shows that if Δ > 0, that is P₂ is shifted away from P₁ in one direction, then the force from the medium on P₂ (see equation (6.1)), due to the deformations induced by P₁, acts in the opposite direction, i.e. it pushes P₂ in the negative x-direction, towards Δ = 0. This shows that the deformation-induced mechanical interactions between dipoles in this simple geometry tend to register them. The interaction of a dipole with a neighbouring periodic series of dipoles arranged in parallel (representing myosin filaments along a stress fibre or a transverse arc) results in a periodic force profile which moves it into registry with the neighbouring array (figure 5e). (The preferred positions for myosin filaments are those with zero force from the medium and are seen to correspond to spatial registry, Δ = 0.) The details behind this theory have been derived in connection with the registry of sarcomeres [62,70]. The deformation-induced force in the longitudinal direction (along the stress fibre) may explain the sliding of two different stacks of myosin filaments into mutual registry as observed by Hu et al. [117].

We can also calculate the force on P₂ in the direction transverse to the parallel bundles (stress fibres), i.e. y-direction, due to the elastic deformation of the IVN by P₁. The resulting expression for the transverse force is 6.2This is plotted as a function of the transverse separation between the two dipoles when they are close to being registered in figure 5f. The negative sign of the calculated transverse force implies an attraction between the two dipoles which tends to reduce their transverse separation. The result is easily generalized to that of a dipole in the neighbourhood of a line of dipoles (figure 5g). Thus, given enough mobility, parallel myosin filaments associated with actin filaments would tend to minimize their transverse distance of separation up to steric and/or electrostatic repulsion effects not included here. This perhaps explains the experimentally observed formation of myosin filament bridges across stress fibres as well as the incorporation of a single myosin filament into a bundle of filaments [117].

The physical origin of these forces is the tendency of the passive, intervening elastic medium to minimize its deformations in response to the active, ATP-dependent contractile forces generated by the actomyosin dipoles. Placing a contractile dipole in a region where the medium is already expanded by the other dipoles reduces the overall medium deformation. The contractile forces exerted by the two dipoles deform the elastic IVN and are everywhere locally balanced by the elastic restoring forces of the IVN that resist mechanical deformations. However, this local equilibrium does not necessarily imply global mechanical equilibrium (minimal elastic energy) of the system of dipoles plus medium. In addition, the contacts of the actomyosin force dipoles with the IVN are subject to dynamically changing, stochastic forces due to the thermal and molecular fluctuations within the actin IVN as well as deterministic forces that arise from the tendency of the elastic medium to resist deformation. The stochastic forces allow for local rearrangements of the dipoles while the deterministic forces eventually—at times significantly longer than those related to molecular turnover—result in translations of the dipoles in a manner that reduces the global deformation of the medium (here, the IVN). This occurs when the dipoles in neighbouring bundles are in registry, as we showed above. The recently demonstrated load-dependent changes in binding lifetimes of myosin to actin [148] may in principle affect the extent of myosin stacking.

A reason why the myosin stacking is lost when actin polymerization/depolymerization is suppressed might be related to the necessity for translocation of myosin filaments along the actin bundles/networks. In order for these actomyosin elements to move into registry and thus reach mechanical equilibrium with their neighbouring bundles, actin filaments in a bundle must depolymerize at one end of the myosin filament and polymerize at the other; unless that happens, the myosin filaments are sterically hindered from moving and registry cannot occur. Since actin polymerization/depolymerization is also stochastic, this suggests that the time scales required for net motion in the direction required for stacking may be quite long, which is consistent with the observed stacking times of about tens of minutes [117] for myosins that must move about 1 µm into registry. Specifically, actin depolymerization proceeds at about 1 nm s⁻¹ [75], so that for this to occur all along the acto-myosin unit of length about 1 µm will require thousands of seconds (tens of minutes).

In contrast to the deformation-induced forces described above, processes such as the expansion of myosin filaments by splitting or the preferential assembly of multiple filaments in close proximity to each other [115,116] might indeed explain the initial formation of filament stacks. However, as already discussed above, it is not at all clear that such an initial configuration can be maintained despite a variety of disordering forces in the cytoskeleton—random molecular-scale and motor-driven noise as well as fluctuations of the actin through polymerization/depolymerization processes. The interactions between myosin filaments through IVN provides a plausible mechanism for the observed long-range movements of myosin filaments to assemble stacks far from their point of initial formation [117]; it is also robust to the presence of noise, which does, however, control the probability that the myosin position relative to those in a neighbouring stress fiber is controlled by the force. Local expansion events [115,116] could be acting in concert with a long-ranged mechanism such as those we describe here. Further experiments will be necessary to unravel the effects of these different mechanisms. Genetic perturbations that vary the degree of crosslinking and density of the intervening network could be a test of the effect of mechanical interactions. Another possible evidence would be the dependence of myosin filament stacking on substrate stiffness in cells cultured on compliant substrates.

7. Concluding remarks

Here we reviewed some aspects about the emergence of highly ordered organization of myosin II filaments in striated muscle and non-muscle cells. Cross-striated muscles show the highest-level of order—crystal-like organization of the myosin, actin and titin filaments. Recent studies showed that in non-muscle cells, myosin II filaments also form relatively well-ordered structures, such as myosin filament stacks associated with actin bundles in interphase cells, the contractile ring in dividing cells, or adhesion belts in cells that form cadherin mediated cell–cell junctions. The mechanisms involved in the establishment and maintenance of such organization are diverse and include interactions of myosin II molecules with each other, with the actin filaments, and with a variety of accessory proteins that regulate actin and myosin filament assembly and crosslinking. In this review, we focused on a basic process that may be a common denominator of these self-organized structures: long-range mechanical forces between actively contractile units (force dipoles) embedded in an elastic medium.

Previous studies demonstrated that the formation of striations corresponding to registered sarcomeric structures in newly formed myotubes [63] and embryonic cardiomyocytes [64] is better established on substrates of an optimal, intermediate rigidity rather than on those that are too soft or too rigid. This is corroborated by in vivo experiments demonstrating that mechanical forces are required throughout the muscle fibre in order to assemble the regular myofibrils during early muscle development [49]. The observation of an optimal substrate stiffness was successfully explained by a theory that assumed sarcomeric contractile elements (force dipoles) effectively interact with each other via the substrate to which they are connected via the costamere adhesions.

Non-muscle cells also form ordered actomyosin bundle arrays in a matrix rigidity-dependent manner [72,149,150]. However, well-ordered myosin stacks are formed even in cells attached to very stiff substrates (glass coverslips), which cannot mediate elastic interactions between contractile elements. Therefore, here we have extended the theory considering individual non-muscle myosin II filaments (instead of sarcomeres) as force dipoles and the disordered, intracellular intervening actin network (instead of the substrate) as the force-transmitting, elastic medium. This theory qualitatively explains the formation of registered organization of the myosin II filaments in non-muscle cells.

Contemporary models of myofibrillogenesis suggest that myofibrils are formed from precursors, the periodic actomyosin structures that comprise non-muscle myosin II filament arrays [151,152]. Formation of such precursor structures may proceed, similarly to formation of myosin II stacks in non-muscle cells, via interactions of myosin filaments with the intervening actin network. Thus, self-organization of myosin II filaments driven by effective, attractive forces between them (due to their mutual deformations of the intervening actin network) may be a general mechanism that drives the organization of actomyosin systems.

This mechanism can operate in concert with other mechanisms based on direct or indirect molecular interactions between myosin filaments. While the accessory molecules responsible for the formation of thick filaments and their arrays in cross-striated muscle are known, our knowledge of molecules with similar functions in non-muscle cells is limited. Factors regulating assembly and disassembly of non-muscle myosin filaments and, in particular, the duplication or partitioning of such filaments [115,116] obviously require further studies.

Finally, the processes of formation of myosin stacks could underlie the global self-organization of the entire actomyosin cytoskeleton in cells. While the formation of the actin filament arrays such as stress fibres or transverse arcs obviously depends on actin polymerization and crosslinking, it is well known that inhibition of the myosin filament assembly and/or activity results in disintegration of these arrays. We hypothesize that mechanical forces between myosin filaments play an important role in the organization and maintenance of these structures. An interesting question is how the ordered organization of actomyosin bundles affects the overall cell contractility. In cardiac muscle, more registered myofibrils tend to beat in concert [76] generating larger beating forces, but further study is required on the relationship between stacking of myosin filaments and contractility in non-muscle cells. The functions of ordered organization of myosin II filaments in the contractile ring [115] and adhesion belt [131,132] are also not clear and provide an interesting avenue for future research. All in all, the discovery of a global ordered organization of myosin II filaments in non-muscle cells opens a new page that challenges our experimental and theoretical understanding of the non-muscle actomyosin cytoskeleton.

Data Accessibility

This article has no additional data.