Journal of Cell Science partnership with Dryad

Journal of Cell Science makes data accessibility easy with Dryad

Nonmuscle myosin II moves in new directions
Mary Anne Conti, Robert S. Adelstein

This article has a correction. Please see:


For many years, analyses of the role of the actomyosin cytoskeleton in many basic cellular processes have centered on actin. Increasingly, however, a number of investigators are examining proteins that are proximal to actin; in particular, nonmuscle myosin II (NMII). Recent experiments have increased our understanding of the role of NMII in three related cellular activities: generation of cell polarity, cell migration and cell-cell adhesion. Progress has been particularly promising thanks to the use of new microscopic, genetic and biochemical techniques. In mammalian systems, generation of transgenic mice and the introduction of specific siRNAs have been useful in deciphering the role of the three different isoforms of NMII: NMIIA, NMIIB and NMIIC. Studies in Drosophila and Aplysia, which are informative model systems for investigating the function of NMII, have also shed light on NMII. Recent work examines the contractile and structural roles that NMII plays at cell-cell boundaries, and both its contractile and actin-crosslinking roles in cell migration. In addition, NMII might also function as a scaffold molecule, anchoring signaling molecules, such as kinases and Rho GTPase guanine nucleotide exchange factors.


The myosin superfamily of actin-based molecular motors consists of at least 25 different classes. The myosin II subfamily, which includes skeletal, cardiac and smooth muscle myosins, as well as nonmuscle myosin II (NMII), has the most members. In vertebrates, there are over 15 different myosin II isoforms, each of which contains a different myosin II heavy chain (MHC). This MHC diversity is generated by multiple genes as well as by alternative splicing of the pre-mRNA that encodes them.

All myosin II molecules are hexamers composed of MHC dimers and two pairs of myosin light chains (MLCs, see Fig. 1A), and can bind reversibly to actin filaments, hydrolyze ATP in a process that is activated by actin and thereby convert chemical energy into mechanical force and movement. Although this contractile activity is most evident in differentiated muscle tissues, such as the beating heart, it is also seen in nonmuscle cells in diverse cellular processes such as cell division, cell migration, and cell-cell and cell-matrix adhesion. There is, however, a fundamental difference in regulation within the vertebrate myosin II subclass. Whereas the primary regulation of skeletal and cardiac muscle myosin II is through a set of actin-associated proteins – troponin and tropomyosin – the regulation of vertebrate nonmuscle and smooth muscle myosin II is through phosphorylation of the 20 kDa MLC (Adelstein and Conti, 1975; Amano et al., 1996). This latter type of regulation permits nonmuscle and developing muscle cells to respond to numerous signals originating both outside and inside the cell.

Whereas the N-terminal globular head of the molecule contains the MgATPase catalytic domain, the α-helical coiled-coil C-terminal domain is involved in filament formation (Fig. 1A,B). Thus, myosin can be viewed as a bipartite molecule that has two distinct but related functions. One is manifested by its ability to translocate actin filaments and resides in its globular motor domain and lever arm. This ability to act as a motor permits a number of different roles. For example, myosin plays a role in cell migration both by moving the body of the cell forward and by retracting the rear of the cell, and it can redistribute inside cells, thus playing a role in generating cell polarity (see below). The second function of NMII is structural and resides in the ability of the rod portion of the molecule to form filaments, which allows several myosin heads to maintain tension for long periods of time, similarly to smooth muscle myosin (Fig. 1B). Both functions require binding to actin. The recent availability of mutant forms of NMII with compromised motor activity has permitted separation of these functions in both cultured cells and mice.

In mammals, three different genes (Myh9, Myh10 and Myh14) encode three different nonmuscle MHCs (NMHCs), commonly referred to as NMHC IIA, NMHC IIB and NMHC IIC (reviewed in Conti et al., 2008; Krendel and Mooseker, 2005; Sellers, 2000; Bresnick, 1999). The MLCs, which are very tightly but noncovalently bound to the MHCs (230 kDa), play a role in stabilizing (MLC17) and regulating (MLC20) myosin structure and activity. MLC20 is a substrate for a number of kinases, including Ca2+-calmodulin-dependent MLC kinase (MLCK), Rho kinase (see Fig. 1C) and, most-recently reported, AMP-activated protein kinase (AMP kinase; see below). Each of these kinases phosphorylates MLC20 primarily on Ser19 (and to a lesser extent on Thr18), which increases actin-activated MgATPase activity, filament formation and contractile activity in vitro and in vivo. Dephosphorylation leads to decreased contractile activity and is catalyzed by a single class of myosin phosphatase (MYPT) that can also be regulated by a number of kinases, including Rho kinase (reviewed in Ito et al., 2004). Phosphorylation of MYPT by Rho kinase is inhibitory and thereby enhances NMII activity by increasing MLC20 phosphorylation (see Fig. 1C) (reviewed in Zhao and Manser, 2005; Matsumura, 2005; Burridge and Wennerberg, 2004; Somlyo and Somlyo, 2003).

The functions attributed to NMII appear to increase almost daily. This is partly due to an increase in the number of tools available to probe NMII activities, such as siRNAs, relatively specific chemical inhibitors such as blebbistatin (Straight et al., 2003) and transgenic animals that have mutations and deletions. Here, we focus on some of the recent developments in our understanding of the various roles played by NMII. Our emphasis is on animal cells, including those containing only a single isoform of NMII (Drosophila), as well as those from mice and humans, which contain three different isoforms. Of course, an article of this size, owing to space limitations, needs to be selective, and we realize that we have had to omit a substantial amount of relevant and outstanding work.

Fig. 1.

(A) Diagram of a myosin molecule showing the globular head region, the α-helical coiled-coil rod and the short non-helical tail (NHT). The subfragment-1 (S-1), rod and heavy meromyosin (HMM) proteolytic domains are also indicated. (B) An example of a bipolar filament, which is formed by interaction among the rod domains. (C) Regulation of myosin activity by phosphorylation of MLC20 of NMII by MLCK and Rho kinase. Whereas MLCK can only phosphorylate MLC20, Rho kinase can phosphorylate MLC20 and also a subunit of myosin phosphatase (MYPT1). Phosphorylation by Rho kinase activates myosin and inactivates MYPT. Both result in an increase in phosphorylated MLC20 and activation of myosin.

Cell polarity

Cell polarity comes in almost as many flavors as quarks: up, down, front, back and side-to-side. The apical-basal cell axis is established at right angles to the plane of the tissue. It is defined by localization of protein networks such as adherens junctions and tight junctions at the apical lateral sides of the cell (Fig. 2A). Its maintenance is necessary for the control of cell division, migration and adhesion to effect cell functions such as absorption at the apical surface in the kidney and intestines, intercellular permeability, as well as oriented movement of tissue layers during embryonic development. Front-to-back polarity is established in cells responding to a signaling gradient; the front of the cell becomes oriented towards the source of the attractive signal (Fig. 2B). Actin-generated cell protrusions begin at this edge and are suppressed at the sides and back of the cell. Nuclei, microtubule-organizing centers and the Golgi align along the new front-to-back axis, and proteins localize to particular regions to effect detachment of the back of the cell. In planar cell polarity, protein localization and cell morphology become biased in response to directional signals from neighboring cells in tissues and organs. Ordered alignment of stereocilia in the mammalian inner ear, orientation of hair, feathers or scales in the skin, elongation of kidney tubules by oriented cell division and closure of the neural tube provide examples of planar cell polarity (reviewed by Zallen, 2007; Klein and Mlodzik, 2005).

Recent work implicates NMII in the control of cell polarity. Lee et al. (Lee et al., 2007) provide evidence that MLC20 is a substrate for AMP-kinase, an enzyme primarily thought of as a metabolic regulator because it is activated in response to low levels of ATP and energy deprivation. The authors show that absence of the kinase in Drosophila epithelial cells leads to marked abnormalities in cell polarity and mitosis. These defects are attributed to a failure to phosphorylate NMII, which results in its inactivation. The defects can be rescued by introducing a phosphomimetic form of NMII, in which Glu residues are substituted in place of phosphorylated Ser19 and Thr18 residues in MLC20, into Drosophila embryos and into a mammalian epithelial cell line. Activation of AMP-kinase is sufficient to cause dramatic changes in cell shape, inducing apical-basal polarization and brush border formation in the human colon cell line LS174T. This work is important because it implicates NMII as a downstream effector of an energy-sensing and cell-polarity-generating pathway and directly demonstrates a role for NMII in establishing cell polarity. Somewhat puzzling is the apparent lack of compensation in flies in which AMP kinase has been ablated by any of the other enzymes known to phosphorylate NMII MLC20.

Phillips et al. (Phillips et al., 2005) have explored the role of NMII in planar cell polarity in a study of noncanonical Wnt signaling during vertebrate cardiac development. These authors are interested in understanding the cause of a common cardiac congenital abnormality, double outlet right ventricle (DORV), in which the pulmonary artery and the aorta both arise from the right ventricle. A DORV is usually associated with a ventricular septal defect, which shunts oxygenated blood to the right ventricle, thereby preventing lethality. Phillips et al. show that a defect in noncanonical Wnt signaling is responsible for defects in myocardialization of the pulmonary outflow tract, implicating Rho kinase and its substrates, MLC20 and the NMII phosphatase (MYPT; see Fig. 1C). Myocardialization is the process during which myocardial cells become polarized and extend lamellipodia and filipodia to initiate migration into the flanking mesenchyme cardiac cushion. A DORV arises in mice that have mutations in the Vangl2 (Van Gogh-like 2) gene, a component of the noncanonical Wnt/planar cell polarity (PCP) pathway. Vangl2 encodes a protein that is part of a multiprotein complex present at the cell membrane that activates RhoA, thus linking signals from Wnt, which are required for cardiac development, to Rho kinase. A striking finding in these mice is the failure of the mutant myocardial cells to become polarized and to invade the cardiac mesenchyme cushion, which prevents normal development of the aortic outflow tract. The authors attribute this to a failure to activate NMII by phosphorylation of MLC20 due to the inability of mutant Vangl2 to activate Rho kinase. The lack of NMII phosphorylation interferes with the initial steps required for cell polarization and subsequent migration of the myocardial cells.

As Philips et al. point out, their data are also consistent with previous findings that ablation of NMHC IIB in mice results in DORV (Tullio et al., 1997). Analysis of NMIIB-ablated mice on embryonic day (E)11.5 reveals a defect in myocardialization due to a failure of cardiac myocytes to migrate into the cardiac cushion (X. Ma and R.S.A., unpublished). Unlike the wild-type cells, the NMHC-IIB-null cardiac cells do not adopt an asymmetric morphology characterized by protrusion at a leading edge. The lack of polarization, which is required for subsequent cell migration, could explain why NMIIB-ablated and -mutated mice fail to establish the proper alignment of the aorta with the left ventricle.

Convergent extension and gastrulation

A role for NMII in cell polarity – as well as cell adhesion – can also be seen during two other important processes in metazoan development: convergent extension and gastrulation. Although much of this work has been carried out in Drosophila, it has important applications to mammalian development. Convergent extension takes place during early development as cell layers narrow and elongate and the embryonic body plan is reorganized. The elongation is mediated by intercalation of cells along one axis, during which they break the linkage to adjacent cells and form linkages to new neighboring cells (reviewed in Lecuit, 2005; Lecuit and Lenne, 2007). NMII (the product of a single gene in Drosophila) is enriched at disassembling cell-cell junctions, where it is concentrated along the anteroposterior axis (Zallen and Wieschaus, 2004), and becomes localized to temporary junctions, where it appears to drive the formation of new junctions by preventing reversion to the original cell orientation (Bertet et al., 2004). The Rho kinase inhibitor Y27632, which inhibits phosphorylation of MLC20 and thereby activation of NMII, also prevents recruitment of NMII to the cell-cell junctions. NMII contractile activity, therefore, appears to be required. Although localization of E-cadherin cell-cell adhesion molecules is not affected, Drosophila NMII mutants are unable to intercalate cells or elongate and the embryos appear to be frozen at the stage of junction remodeling. Such experiments suggest the importance of NMII-mediated contractile activity at cell-cell boundaries in the oriented disassembly of cell junctions as well as a role for NMII in the assembly of new junctions, which ultimately alters the configuration of the tissue to narrow and extend the body axis.

Phosphorylation of MLC20 and activation of NMII by Rho kinase signaling have an important role in gastrulation during Drosophila embryogenesis. Gastrulation is the stage of embryonic development during which the mesoderm and definitive endoderm are formed and the embryo becomes three-layered. It is accompanied by precisely orchestrated cell movements, which include invagination of cells derived from the ectoderm to give rise to the mesoderm. During Drosophila embryogenesis, an apically localized network of actomyosin is required for apical cell constriction during ventral furrow formation, an early step in gastrulation. Studies from a number of laboratories (Nikolaidou and Barrett, 2004; Dawes-Hoang et al., 2005; Fox and Peifer, 2007; Kolsch et al., 2007) have elucidated a pathway initiating gastrulation that starts at the transcription factor twist and involves activation of the Gα subunit of an apically located G-protein-coupled receptor. This leads to the activation of a guanine nucleotide exchange factor, RhoGEF2, which catalyzes the exchange of GDP for GTP on Rho GTPases, thereby activating them (see Fig. 1C). This results in activation of Rho kinase with subsequent phosphorylation of MLC20 and localization and contraction of the NMII-actin network and the apical cell surface. In mutants of armadillo (the Drosophila analog of β-catenin), in which apical cell junctions are disrupted, NMII is mislocalized, forming a tight, contracted ball at the center of the cell, and the constriction and cell shape changes that are necessary for gastrulation do not occur (Dawes-Hoang et al., 2005). This work is interesting because it links NMII to the cell adhesion protein β-catenin and shows the importance of placing NMII in the right location (the apical cell border) at the right time (Fig. 2A).

Cell migration

A common theme in the experiments discussed above is activation of NMII by phosphorylation of MLC20. A number of studies in mammalian cells have been directed at identifying the relevant kinases catalyzing this phosphorylation. Investigation of the front-to-back polarity that plays a crucial role in neutrophil migration implicates Ca2+ signaling and MLCK in uropod retraction, a late, vital step in cell migration (Eddy et al., 2000) (Fig. 1C). The locations of MLCK and Rho kinase in fibroblasts have been studied by Katoh et al. (Katoh et al., 2001) and Totsukawa et al. (Totsukawa et al., 2004). Both find evidence for spatial regulation of MLC20 phosphorylation. In addition, Totsukawa et al. (Totsukawa et al., 2004) show that MLCK is active at the cell periphery, where it controls membrane ruffling, and Rho kinase is active at the cell center, at which it regulates focal adhesion formation. Inhibition of Rho kinase prevents formation of focal adhesions at the center of the cell (but not at the periphery) and results in faster, more persistent cell migration. Sandquist et al. (Sandquist et al., 2006) provide evidence that NMIIA and NMIIB have different functions during cell migration and spreading in tumor cell lines, and that Rho kinase preferentially regulates phosphorylation of MLC20 associated with NMIIA in these cells. Their data raise the intriguing possibility that differential regulation of the different NMII isoforms might reside in the different kinases, or perhaps, as suggested by others, in the small GTP-binding proteins that regulate some of the kinases.

Fig. 2.

Cell polarization due to rearrangement of the actomyosin cytoskeleton. (A) An example of apical-basal cell polarity, such as that shown by epithelial cell layers in the intestines and skin. Myosin is concentrated at the apical side of the cell, where it might participate in assembly and disassembly of the apical junction complex (green rectangles), which tightly links cells, or in the process of apical constriction, which alters the shape of the cell and its tissue layer. (B) Diagram of cell migration and the establishment of front-to-back polarity. Extension of the lamellipodium at the front (right) of the migrating cell is due to addition of actin monomers to the branching actin network. The migrating cell attaches to and detaches from the surface via integrin clusters. Focal complexes form at the cell front and mature into focal adhesions owing to NMII contractility. Activation of NMII contributes to detachment of the rear (left) of the moving cell. Arrows indicate the direction of movement.

Cell migration can be viewed as a series of steps, involving cell protrusion (which is primarily actin-mediated but also involves myosin activity), front-to-back cell polarization, the attachment of the front of the cell to the extracellular matrix, assembly (and ultimately disassembly) of adhesion plaques and, finally, retraction of the tail of the cell (Fig. 2B). Although we have arbitrarily divided this Commentary into sections on cell polarity, migration and cell adhesion, these are related phenomena, as emphasized in a recent article by Vincente-Manzanares et al. (Vincente-Manzanares et al., 2007). These authors used isoform-specific siRNAs and subsequent transfections to introduce both wild-type and mutant GFP-NMHC-II constructs into fibroblasts to study protrusion, adhesion and migration. They find that an NMIIA (NMHC Asn93Lys) mutant that has markedly decreased ATPase activity can substitute for endogenous NMIIA during maturation of adhesion sites at the leading edge of migrating fibroblasts, indicating the importance of myosin as a crosslinker for actin filaments (Fig. 2B). An NMIIB mutant (NMHC Arg709Cys) can also act as an actin crosslinker and substitute for endogenous NMIIB to restore central adhesions and at least partially restore front-to-back polarity. NMII, in some cases acting at a distance, thus appears to integrate these component processes to drive cell migration.

Other studies of the role of NMII in mammalian cell migration (reviewed in Webb et al., 2005) have also focused on the role of the individual isoforms NMIIA, NMIIB and NMIIC. Betapudi et al. (Betapudi et al., 2006) used NMII-specific siRNA in a breast tumor cell line to show that depletion of NMIIA impairs cell migration but enhances lamellar spreading, whereas depletion of NMIIB impairs the initial rates of lamellar spreading as well as migration. Cai et al. (Cai et al., 2007) have used mouse embryonic fibroblasts and retrovirus-mediated expression of short hairpin RNAs to generate cell lines lacking NMIIA and NMIIB. They find that NMIIA is the principal force-generating myosin in these cells and that NMIIA and NMIIB together account for nearly all of the force-generating capacity in these cells.

Using a combination of genetically ablated cells and siRNA-treated cells, Even-Ram et al. (Even-Ram et al., 2007) provide evidence that NMIIA plays a crucial role coupling the actomyosin and microtubule cytoskeletal systems. NMIIA, under normal conditions, promotes microtubule dynamics and acts as a brake for membrane ruffling and cell migration. The authors show an unusual expansion of microtubules into the overly active lamellae of NMIIA-ablated cells, in which the microtubules become stabilized at the edge of the cell. This work is consistent with previous studies showing that fibroblasts isolated from NMIIB-null mice exhibit multiple, unstable protrusions and migrate with a lack of persistence (Lo et al., 2004). Studies of superior cervical ganglion nerve growth cone motility using cells from NMIIB-ablated mice yield similar observations (Bridgman et al., 2001).

Investigation of the invasion of 3D matrices by ameboid tumor cells has also shown the importance of phosphorylation of NMII MLC20 in migration. Sahai and Marshall (Sahai and Marshall, 2003) have implicated Rho signaling in this process. The subsequent phosphorylation of MLC20 by Rho kinase is instrumental in the invasion of Matrigel by cells that have an ameboid or rounded morphology (Wyckoff et al., 2006). In contrast to migration of elongated cells, migration of these cells does not require proteolysis of the extracellular matrix. The cells can instead generate sufficient force to deform collagen fibers and migrate through the extracellular matrix owing to the contractile activity generated by phosphorylated NMII on actin filaments. Importantly, in addition to working with cell lines, the investigators also imaged live tumors and show that the organization of GFP-labeled myosin mimics that in the cell lines in that it is present immediately behind the invading margin of the cell. Moreover, inhibition of Rho kinase, but not matrix metalloproteinases, reduces rounded tumor cell motility in vivo and points to the possible importance of targeting both Rho kinase and matrix metalloproteinases in stemming cancer cell migration.

Medeiros et al. (Medeiros et al., 2006) have used fluorescent speckle microscopy to assess actin dynamics in cultures of migrating Aplysia (sea slug) bag cell neurons. The authors define a zone of fast retrograde actin flow in the peripheral domain of the migrating nerve growth cone, which is driven by polymerization of actin at the front of the cell and depolymerization at the rear of the peripheral domain (actin treadmilling). An adjacent zone of slower retrograde actin flow in the transition domain, however, is dependent on NMII contractile activity. They find that inhibition of NMII by blebbistatin decreases rearward movement of actin filaments from the leading edge by approximately one-half, the remaining movement being almost fully accounted for by plus-end actin assembly at the leading edge of the growth cone. Retrograde actin flow is, thus, a steady state that depends on both NMII contractility and actin-network treadmilling. The authors provide evidence that NMII contractility in the transition zone, which defines the central-to-peripheral-domain boundary behind the leading edge of the nerve growth cone, potentiates the severing and recycling of polarized actin filaments near their minus ends. They speculate that myosin contractile force on actin bundles causes bending or kinking of filaments, which might exert shear forces leading to bundle severing. This work is important because it shows that cell migration is affected by a number of different aspects of cytoskeletal activity including myosin contractility, actin polymerization-depolymerization and actin severing.

Gupton and Waterman-Storer (Gupton and Waterman-Storer, 2006), and Giannone et al. (Giannone et al., 2007) contribute important signaling and biomechanical information to our understanding of the role of NMII in cell migration. Gupton and Waterman-Storer used quantitative fluorescent speckle microscopy and other microscopy techniques to show that a dynamic interdependence of actin filament organization, NMII contractility and binding of focal adhesions to the extracellular matrix contributes to optimal cell migration at intermediate extracellular matrix concentrations. NMII contractile activity and focal adhesion strength affect actin filament organization and bundling, which, in turn, alters focal adhesion turnover and cell migration speed. Giannone et al. (Giannone et al., 2007) extend these studies by analyzing the molecular mechanisms coordinating actin polymerization, myosin-force generation and focal adhesion formation (see Fig. 2B). They demonstrate that mechanical signaling is as crucial as biochemical signaling during cell migration, identifying a labile actin network, lamellipodial actin, that periodically separates from the cell edge. NMII at the rear of the lamellipodium pulls on this actin network, resulting in increased tension at the front of the cell, buckling of the cell front, retraction of the edge and formation of an adhesion site near the front of the cell. In their model, the actin network then separates from the cell tip and cell protrusion starts again. Thus, the lamellipodial actin operates like a cyclic mechanical bridge that coordinates NMII force generation with the initiation of adhesion sites and cycles of edge protrusion and retraction.

Cell-cell adhesion

The apical junction complex that is responsible for cell-cell adhesion and the barrier function of diverse tissue types plays key structural roles in cell sorting and organogenesis and, additionally, functions as part of intracellular signaling pathways (reviewed in Goodwin and Yap, 2004; Erez et al., 2005). The main proteins of the adherens junctions are the transmembrane cadherins and the cytosolic catenins; tight junctions contain transmembrane occludins and claudins and the cytosolic ZO-1 protein. Recent work indicates that NMII plays an important part in cell adhesion by controlling localization of these proteins (Fig. 3).

Genetic ablation of NMIIA from embryonic stem cells and mouse embryos results in a loss of cell-cell adhesion, which correlates with a decrease in E-cadherin and β-catenin localization at cell-cell contacts (Conti et al., 2004). Because there is no decrease in the overall content of cell adhesion proteins in these cells, they appear to have become diffusely distributed in the cytoplasm. siRNA treatment of wild-type embryonic stem cells, in order to lower NMIIA content, yields similar results. The studies suggest that tension on actin provided by NMII is necessary to maintain the proper localization of these cell adhesion proteins. Note that the earlier models in which there is a direct connection between transmembrane cadherin and the actomyosin networks have been modified in light of experiments showing that α-catenin does not simply connect a stable E-cadherin and β-catenin complex to actin (Yamada et al., 2005; Drees et al., 2005). It is possible that the connection from cadherin to the actomyosin network is achieved by locally high concentrations of α-catenin dimers (which promote actin polymerization), by linkages through the PDZ domain of β-catenin to other proteins that bind to actin or by other adhesion molecules, such as nectin (reviewed in Weis and Nelson, 2006) (Fig. 3).

Shewan et al. (Shewan et al., 2005) find interdependence in the localization of E-cadherin and NMII in the MCF7 tumor cell line. Moreover, they show that the recruitment of NMII to cadherin-based contacts is regulated by phosphorylation of Ser19/Thr18 on MLC20 by Rho-kinase. Yamada and Nelson (Yamada and Nelson, 2007) expand on this idea, showing that there are two steps in cell-cell adhesion. The first is mediated by activation of Rac1 and its downstream target Arp2/3, which regulates actin filament branching in the lamellipodia. The second step involves activation of RhoA and contraction of actomyosin. The initial step in cell-cell adhesion appears to be stochastic, involving contact between exploratory lamellipodia emanating from opposing cells and resulting in the local accumulation of E-cadherin by a still unknown process. The zone of E-cadherin accumulation spreads laterally as more contacts are formed by the Rac1-induced lamellipodia. During the second stage, the actomyosin contractile forces pull the edges of the contacting cell membranes to expand the contact to the entire width of the cells. Together, these activities drive initiation, expansion and completion of cell-cell adhesion.

Fig. 3.

Diagram showing a structural role for nonmuscle myosin II in maintaining cell adhesion. The extracellular domains of cadherin molecules interact with those of the neighboring cell. Bipolar filaments of myosin bind to actin, which, in turn, interacts with the catenins, which are shown interacting with the cytoplasmic portion of cadherins (green rectangles). The nature of the interaction between the catenin and associated proteins and actin is unclear at present (see text). As outlined above, NMII plays a role in both maintaining and disassembling the adhesion junction complex. (Right) A similar type of role can be visualized for NMII with regard to focal adhesions through which the cell interacts with the extracellular matrix (ECM). Among the various proteins binding to the integrins are α-actinin, talin, focal adhesion kinase, vinculin and paxillin. (Bottom) NMII also serves as a scaffold for kinases and small GTP-binding molecules (as well as molecules that have yet to be identified), which regulate its activity and its interaction with actin and other molecules such as microtubules. The diagram omits numerous actin-bundling and regulatory proteins.

In mice, ablation or mutations in the gene encoding NMHC IIB result in defects in the heart and brain during development. Recently, one of the brain defects, hydrocephalus, has been linked to a loss of cell-cell adhesion in the cells lining the spinal canal (Ma et al., 2007). Hydrocephalus results from an excess of cerebral spinal fluid in the ventricular system of the brain and, if untreated, can lead to irreversible brain damage. The cells lining the spinal canal contain only the NMIIB isoform, which co-localizes with N-cadherin and β-catenin (Fig. 4a,c). Closer inspection of these cells shows the presence of a polarized mesh-like adhesion structure at the apical border of wild-type cells (Fig. 4c), which collapses and becomes discontinuous in the NMIIB-ablated (Fig. 4d) and hypomorphic mutated (Fig. 4b) mice. The hypomorphic mutated mice have decreased amounts of NMIIB and a point mutation (Arg709Cys) in the myosin motor domain. The collapse of the mesh-like structure permits the underlying neuroepithelial cells to invade and obstruct the spinal canal. This blockage appears to be the primary cause of the hydrocephalus.

The defects in cell adhesion can be rescued by the introduction of two different NMII proteins by homologous recombination. Despite significant differences in the kinetic properties of NMIIA and NMIIB (Kovacs et al., 2003; Rosenfeld et al., 2003), expression of NMHC IIA in place of NMHC IIB restores the mesh-like structure in the cells lining the spinal canal, thereby preventing hydrocephalus (Bao et al., 2007). The hydrocephalus can also be rescued in vivo by elevated expression of the Arg709Cys NMIIB mutant (Ma et al., 2007). Because this myosin has a decreased ATPase activity and cannot propel actin filaments in gliding assays, this finding is somewhat surprising. One explanation is that NMIIB has a structural role rather than a motor role in the adhesion complex of the cells bordering the spinal canal. Thus, myosin could exert tension on actin filaments for prolonged periods while slowly hydrolyzing ATP. Interestingly, neither NMIIA nor the mutant NMIIB can rescue abnormal migration of the pontine or facial neurons found in NMIIB-ablated mice, which indicates a motor rather than a structural function for NMIIB in the migration of these cells.

The above experiments suggest that NMII plays two different roles in cell adhesion: it makes use of its structural properties and its ability to bind to actin to exert tension on cortical actin filaments in order to maintain cell-cell connections, and it uses its enzymatic properties and ability to translocate actin to help disassemble the adhesion complex. Additionally, there might be a scaffolding role for the rod region of the NMII molecule (Fig. 3). Because this region of the MHC is known to have sites for phosphorylation (e.g. Dulyaninova et al., 2007; Even-Faitelson and Ravid, 2006), these sites might regulate interaction with a number of proteins. In this case, the rod would act as an anchor for effector molecules – for example, kinases – or signaling molecules such as guanine nucleotide exchange factors, including MyoGEF, which has been shown to play a role in cytokinesis (Wu et al., 2006).

Concluding remarks

Cell polarization, cell adhesion and cell migration are related and interdependent processes, and understanding how one influences the other is proving to be a productive area for research. Still, important problems remain and need to be resolved. Many experiments are performed on cell lines that have undergone multiple passages. These cells often acquire significantly altered cytoskeletal properties compared with primary cultured cells and cells in vivo. Some of these changes might be reflected in the quantity and relative levels of NMII isoforms and investigators should consider this. Quantification of the various isoforms of NMII in mammalian cells has become important, particularly when one is analyzing whether an isoform plays a unique role that cannot be played by a second isoform of NMII. It is important to distinguish whether a phenotype in a mammalian cell reflects a low amount of a specific NMII isoform or of NMII in general.

A favorite caveat that is raised by most cell biologists, the reconciliation of differences between cells studied in 2D/3D culture and cell behavior under true in vivo conditions (Yamaguchi et al., 2005; Even-Ram and Yamada, 2005), is a major issue that also needs to be resolved. Recent work with mutant NMII isoforms using transfected cells and transgenic mice has permitted experiments that separate the structural and enzymatic properties of NMII. These experiments will need to be extended to include more sophisticated point mutants of the various parts of the molecule known to be involved in these two distinct functions. Finally, the manner by which NMII integrates its variety of functions with other members of the myosin superfamily is of particular interest (see Arden et al., 2007; Maddugoda et al., 2007). Despite these caveats, the future appears both challenging and intriguing as NMII moves in new directions.

Fig. 4.

Immunofluorescence confocal images of mouse spinal cords. (a) Cross-section at E12.5 of a wild-type (B+/B+) spinal cord stained for NMIIB (green) and N-cadherin (red). (b) Magnified view of a hypomorphic Arg709Cys NMIIB-mutant (BCN/BCN) spinal cord at E12.5 stained as in panel a. Note that the neuroepithelial cells have invaded and obstructed the spinal canal in the mutant mouse. (c) Apical border of the cells lining the spinal canal in a B+/B+ mouse at E11.5, showing a mesh-like adhesion structure stained for NMIIB (green) and β-catenin (red). (d) Spinal canal of an NMIIB-ablated (B/B) mouse at E11.5, showing collapse of the apical mesh-like structure, with neuroepithelial cells protruding into the canal, stained for NMIIB (green, which is not seen in the B/B mouse) and N-cadherin (red). These panels are reproduced with permission (Ma et al., 2007).


The authors thank Sachiyo Kawamoto, Xuefei Ma and James R. Sellers (NHLBI) and Kenneth M. Yamada (NIDCR) for their important and useful comments, and Catherine S. Magruder for her expert editorial assistance.

  • Accepted November 1, 2007.


View Abstract