Intercellular mechanotransduction during multicellular morphodynamics

Jin-Hong Kim, Lawrence J. Dooling, Anand R. Asthagiri

Abstract

Multicellular structures are held together by cell adhesions. Forces that act upon these adhesions play an integral role in dynamically re-shaping multicellular structures during development and disease. Here, we describe different modes by which mechanical forces are transduced in a multicellular context: (i) indirect mechanosensing through compliant substratum, (ii) cytoskeletal ‘tug-of-war’ between cell–matrix and cell–cell adhesions, (iii) cortical contractility contributing to line tension, (iv) stresses associated with cell proliferation, and (v) forces mediating collective migration. These modes of mechanotransduction are recurring motifs as they play a key role in shaping multicellular structures in a wide range of biological contexts. Tissue morphodynamics may ultimately be understood as different spatio-temporal combinations of a select few multicellular transformations, which in turn are driven by these mechanotransduction motifs that operate at the bicellular to multicellular length scale.

1. Introduction

The remarkable dynamism of multicellular structures is in full display during development and continues in adult tissues, such as the intestinal epithelium (van der Flier & Clevers 2009) and the mammary gland (Nelson & Bissell 2006). Meanwhile, disruptions in multicellular morphology, and consequently tissue function, play a major role in diseases, such as cancer (Debnath & Brugge 2005). Elucidating the forces that form and re-shape multicellular structures is integral to our understanding of development and disease, and has clear implications for biomedical applications, such as tissue engineering.

Transformations in multicellular structure are achieved through mechanical forces that act upon cell adhesions. Cells adhere to their neighbours and to the surrounding extracellular matrix (ECM). Significant advances have been made in our understanding of the molecular composition of adhesions (Zamir & Geiger 2001; Zaidel-Bar et al. 2007) and their mechanosensitivity (Geiger et al. 2009). Acting as mechanosensors and as an interface for force transmission, cell adhesions play a pivotal role in regulating single-cell behaviours, such as rolling (Chang et al. 2000), spreading and migration (Beningo et al. 2001), survival and proliferation (Chen et al. 1997), and differentiation (Engler et al. 2004b, 2006).

Multicellular morphodynamics, however, is not the simple consequence of cell autonomous responses to local forces. Local forces are transmitted over longer length scales and propagate their effects at a mesoscopic level. In this review, we discuss different modes by which mechanical forces are transduced in a multicellular context, ranging from bicellular interactions to larger tissue-scale structures (figure 1). Here, we use the term ‘mechnotransduction’ broadly to include both this transmission and re-distribution of mechanical forces and the interconversion of mechanical forces and biochemical signals.

Figure 1.

Modes of force transmission in multicellular systems. In a multicellular context, several intercellular mechanotransduction motifs can be identified: (a) indirect mechanosensing through compliant substratum (black arrows); (b) cytoskeletal ‘tug-of-war’ between cell–matrix and cell–cell adhesions (blue arrows); (c) cortical contractility contributing to line tension (red arrows); (d) compressive stresses (green arrows) acting on the planes represented by the dashed lines and resulting from the proliferation of neighbouring cells; and (e) forces mediating collective migration (purple arrows) including traction forces, such as those depicted at the leading edge, and tension that is propagated through cell–cell contacts.

2. Indirect cell–cell mechanosensing through a compliant ECM

An emerging mechanism for cell–cell communication involves exerting and sensing traction forces on the ECM. When a cell contracts, it pulls on its surroundings through integrin-mediated adhesions. This allows the cell to sense the mechanical response of its environment and react appropriately (Discher et al. 2005; Vogel & Sheetz 2006). As a result, the physical properties of the matrix, in particular its compliance, have a significant effect on cell behaviours such as spreading (Engler et al. 2004a; Yeung et al. 2005), migration (Pelham & Wang 1997; Lo et al. 2000; Peyton & Putnam 2005), proliferation (Wang et al. 2000) and differentiation (Engler et al. 2004b, 2006). In in vitro studies of contractility, substrates of varying compliance are commonly prepared using synthetic polymers, such as polyacrylamide, by varying the extent of cross-linking while keeping the adhesive ligand composition constant (Pelham & Wang 1997). Fluorescent beads can be embedded within these substrates, and their displacements are measured to produce a map of the traction forces (Dembo & Wang 1999).

It is becoming increasingly apparent that contractile forces generated against the ECM not only influence the behaviour of individual cells but also play a role in governing how cells interact with each other. As a cell contracts on a compliant substrate, it produces stress and strain that can be sensed by its neighbours, thus providing a mechanical pathway for cell–cell communication even in the absence of direct contact. Reinhart-King et al. (2008) have demonstrated this concept by investigating how substrate compliance influences the contact and migratory behaviours of pairs of bovine aortic endothelial cells. Using traction force microscopy, they show that the distance over which a cell significantly deforms its substrate decreases with increasing substrate stiffness, and they postulate that this distance represents the maximum range of ECM-mediated cell–cell mechanosensing. On soft surfaces, two cells that collide remain in contact throughout the duration of the experiment, most likely because the soft substrate prevents them from generating enough traction force to break the cadherin bonds formed at the cell–cell junction. Conversely, cellular collisions on stiff surfaces are very elastic and cells remain in contact for short durations before migrating away from one another. On substrates of intermediate compliance, a pair of cells repeatedly forms a contact and breaks it. As a result, they exhibit a lower dispersion than isolated cells and fail to migrate beyond the measured distance of significant substrate deformation. This behaviour suggests that, even after contact is broken, the cells still communicate mechanically through the matrix and that the substrate compliance influences cell–cell interactions.

Cell–cell communication mediated by the ECM has also been observed between human mesenchymal stem cells (hMSCs) on fibrin (Winer et al. 2009). In this case, the communication is believed to involve the strain-stiffening property of nonlinear elastic matrices. The strain produced by cell contraction stiffens the substrate by several orders of magnitude, thereby changing the microenvironment of nearby cells. This results in an alignment and elongation of hMSCs cultured on such substrates.

The two previous examples demonstrate how contractile forces generated on the ECM may be responsible for influencing the interactions between cells cultured in vitro on compliant substrates. Similar behaviour may be observed at the tissue level as well. Epithelial and endothelial cells are often separated from underlying stromal cells by a basement membrane consisting of proteins, such as laminin and collagen. The presence of stromal cells significantly alters the mechanical properties of the ECM through contractility and matrix remodelling. Elson and colleagues have shown that fibroblasts compress and stiffen collagen gels in vitro (Wakatsuki et al. 2000), and that the mechanical properties of the tissue vary with fibroblast concentration (Marquez et al. 2006). These effects can be sensed by the basal surface of the epithelium and endothelium, and may play an important role in tissue homeostasis, development and tumour progression (Grinnell 2003; Lopez et al. 2008).

3. Direct cell–cell interactions and their mechanical interplay with cell–matrix adhesions

While cells are capable of communicating indirectly with each other through the ECM, as cells get close enough to interact directly using cell–cell adhesion receptors, such as cadherins, various short-range modes of cross talk unfold between cell–cell and cell–matrix adhesions. The differential adhesion paradigm considers the antagonistic interplay between cell–cell and cell–matrix interactions at the level of the cell surface. Steinberg and colleagues observed this antagonism during the transition between aggregation and spread phenotypes of multicellular clusters (Ryan et al. 2001). Cells with minimal cadherin expression level exhibited low cohesivity and a spread phenotype even on substrata that are only moderately adhesive. However, increasing cell–cell cohesivity by raising cadherin expression reverts this spread phenotype and promotes aggregation. Their results demonstrate that tissue spreading is the outcome of a competition between cell–cell cohesivity and cell–substratum adhesivity (figure 2a).

Figure 2.

Cross talk between cell–matrix and cell–cell adhesions. (a) The differential adhesion paradigm states that aggregation is preferred when cell–cell cohesivity outweighs cell–matrix adhesivity, while a spread phenotype is promoted when cell–matrix adhesivity dominates. (b) Cell-generated contractile forces mediate a ‘tug-of-war’ between cell–cell and cell–matrix adhesions. (c) A multicellular implication of (b) is depicted in two-dimensional cell culture on elastomeric micropillars. Peripheral cells exhibit high traction forces on the substrate resulting in the bending of the pillars, while cells in the interior of the cluster dissipate cell-generated contractile forces against their neighbours.

In addition to this antagonism at the level of the cell surface, cell–cell and cell–matrix adhesions are also coupled mechanically through their joint affiliation with the cytoskeleton. At the molecular level, actin cables associate with adherens junctions at cell–cell contacts and provide a physical mechanism for cell-generated contractile forces to act upon cell–cell adhesions. This non-muscle myosin-mediated tension at sites of cell–cell adhesion is necessary for the formation and maturation of cell–cell contacts, which are destabilized upon loss of myosin-generated contractility (Conti et al. 2004; de Rooij et al. 2005). However, excessive contractile forces can compromise cell–cell adhesions (de Rooij et al. 2005). Precisely how much contractile force is imposed upon cell–cell adhesions will depend on the level of cell–matrix adhesions, which are also linked to the actin cytoskeleton (figure 2b). In situations where cell–matrix adhesions are enhanced, as observed upon hepatocyte growth factor treatment and on stiff substrates, they are able to withstand contractile forces better, while cell–cell adhesions are compromised, thereby promoting cell scatter (de Rooij et al. 2005). Consistent with these observations, cells are able to form multicellular aggregates better and undergo tissue-like compaction on a compliant substratum more than on stiff substrates (Guo et al. 2006). Furthermore, using mammary epithelial cells cultured in three-dimensional matrix, Weaver and colleagues showed that increasing matrix stiffness elevates Rho kinase (ROCK)-generated contractility and focal adhesion formation among mammary epithelial cells, in turn weakening adherens junctions and disrupting organized acinar structures (Paszek et al. 2005). In this manner, cell-generated contractile forces mediate a ‘tug-of-war’ between cell–cell adhesions and cell–matrix adhesions that has implications for multicellular organization in both two- and three-dimensional contexts. It is important to note, however, that these forces at cell adhesions may also induce changes in gene expression that contribute to cell scatter. For example, enhanced adhesion-mediated signalling on stiff surfaces may lead to gene-expression patterns facilitating the loss of cell–cell contacts and cell scatter as observed in epithelial–mesenchymal transition.

The cross talk between cell–cell and cell–matrix interactions can also promote spatial gradients in mechanical stresses within multicellular structures. Cells at the periphery of a cluster extend their free edge into the surrounding ECM and exert greater traction forces through their adhesions to the matrix than cells in the interior of the cluster. In contrast, interior cells are surrounded by neighbours, and the contractile forces generated within these cells are imposed upon their neighbours through cell–cell contacts. By measuring the deflection of vertical elastomeric micropillars, Chen and colleagues directly quantified the gradient in traction forces in multicellular clusters and correlated this gradient to spatial patterns in proliferation (figure 2c) (Nelson et al. 2005). The introduction of cytoplasmic-deletion mutant of VE-cadherin, which is defective in linking cadherin to the actin cytoskeleton, ablated the spatial gradient in traction forces and the pattern in cell cycle activity across cell clusters.

In addition to the distribution of traction forces within multicellular aggregates, the level of soluble growth factors, such as epidermal growth factor (EGF), also plays an important role in shaping spatial patterns in proliferation. We have recently demonstrated that, in epithelial clusters, cadherin-dependent contact inhibition is enforced only below a critical threshold level of EGF (Kim et al. 2009). Thus, only when the growth-promoting activity of EGF dips below a threshold is cell–cell contact able to effectively inhibit the proliferation of cells in the interior of a cluster, leading to a spatial pattern in proliferation. When EGF concentration is raised above this threshold, epithelial cells exhibit contact-independent, uniform proliferation. Intriguingly, this threshold amount of EGF is tunable: augmenting cell–cell interactions increases the EGF threshold at which the system transitions from contact-inhibited to contact-independent proliferation. Thus, it is evident that a cross talk between hormonal/growth factor pathways and the physical distribution of traction forces is involved in regulating patterns in cell proliferation in epithelial clusters.

The maturation of cell–cell contacts in epithelial sheets can be accompanied by the recruitment of the focal adhesion protein vinculin from sites of cell–matrix adhesions to cell–cell junctions. This change in vinculin localization leads to the reorganization of stress fibres associated with focal adhesions at the cell–substratum interface into cortical bundles that run parallel with cell–cell contacts (Maddugoda et al. 2007). During epidermal stratification, cortical actin bundles further polarize into the apical plane and form a continuous cytoskeletal network spanning the entire epithelial sheet. Coordinated tension developed through these apical actin cables enables cells to slide under neighbouring cells by transiently disrupting their cell–substratum interactions (Vaezi et al. 2002).

4. Cortical contractility and line tension along cell–cell contacts

Cell-generated contractile forces along cortical actin structures in the apical region of epithelial cells also contribute to line tension along cell–cell interfaces. This line tension plays a significant role in cellular rearrangements during processes such as intercalation in response to external and internal forces, in shaping and sizing cells in growing epithelial sheets, and in maintaining multicellular compartments (figure 3).

Figure 3.

Cortical contractility and interfacial line tension as a recurring motif in multicellular morphodynamics. Cell-generated contractile forces acting along the cortical actin structures are counterbalanced by cell–cell adhesion, contributing to the development of line tension at the interface between cells. (a) This line tension presents an energy barrier to cellular rearrangements under externally applied forces, giving rise to irreversible deformation of multicellular aggregates. (b) Internal asymmetry in local cortical tension drives intercalation involving the collapse of dorsal–ventral (DV) junctions (indicted by the red line) followed by the addition of anterior–posterior contacts (indicated by blue lines). (c) Collective cell intercalation propels tissue-wide change in morphology such as tissue elongation. (d) In a growing epithelium, cell division predominantly results in daughter cells sharing an edge and cleaving at a side rather than a vertex, posing ‘geometric rules’ governing distribution of cell shapes. (e) Anisotropic line tension developed at the boundary of two cell populations (indicated by the red line) is responsible for the maintenance of multicellular compartmentation such as DV demarcation of wing imaginal discs.

4.1. Line tension as an energy barrier for plastic deformation

At a macroscopic scale, line tension is involved in the plasticity of multicellular aggregates (i.e. irreversible shape change of the aggregates) exposed to external compressive load (Marmottant et al. 2009). Line tension provides an energy barrier for cellular rearrangements within the aggregates. Cell aggregates under high compressive stress overcome this barrier and undergo not only elastic cell shape change, but also cellular rearrangements involving shuffling of cells (intercalation). These cellular rearrangements persist even after the imposed external stress is removed, rendering a plastic deformation (figure 3a). In contrast, in a low-stress regime where line tension is not overcome, aggregates exhibit only cell shape changes through spontaneous membrane fluctuations, and, when the external force is removed, the original aggregate shape is recovered.

4.2. Line tension in intercalation

Asymmetric line tension provides the driving force for intercalation during germ band elongation in Drosophila embryos (Rauzi et al. 2008). In this process, the epithelial tissue elongates along the anterior–posterior (AP) axis through the intercalation of cells along the dorsal–ventral (DV) axis. This process involves the shrinkage of DV contacts (v-junctions) followed by the establishment of new cell–cell contacts parallel to the AP axis (t-junctions) (figure 3b). Myosin II is preferentially localized at v-junctions (Lecuit & Lenne 2007), and this localization corresponds to greater tension at v-junctions than along t-junctions as quantified by local laser ablation and the consequent recoil speed of the cell–cell interface (Rauzi et al. 2008). This asymmetry in local cortical tension drives the tissue-wide change in morphology (figure 3c). Furthermore, this tension increases as the v-junction collapses, suggesting that cortical elasticity is also a critical factor.

4.3. Line tension in shaping cells in growing epithelium

During the intercalation process described above, the number of epithelial cells remains fixed and the predominant activity involves the relative shuffling in the position of cells within the epithelium. In other situations, the epithelium undergoes a significant change in cell number while the relative position of cells does not change markedly. An important feature of such growing epithelial sheets is the distribution of polygonal cell shapes. While most cells are hexagonal, there are also significant numbers of cells with a shape that ranges from quadrilateral to octagon. Gibson et al. (2006) show that simple ‘geometric rules’ of epithelial cell divisions are sufficient to predict the distribution of polygonal shapes in the developing epithelial wing primordium of Drosophila melanogaster. These rules were based on observations such as the following: the vast majority of epithelial divisions (94%) result in daughter cells that share an edge, and cell divisions tend to cleave a side rather than a vertex (figure 3d). A Markov model based on these and other geometric rules predicts that a growing epithelial sheet reaches a distribution of polygonal shapes consistent with that observed in developing wings. In fact, the predicted distribution of cell shapes matches that observed in epithelial tissues from vertebrate, arthropod and cnidarian organisms, suggesting that a common set of geometric division rules governs the shapes of epithelial cells in growing tissues throughout the metazoa.

It should be noted, however, that the distribution of polygonal shapes is not the only feature of interest in a growing epithelium. For example, the average size of a cell increases as cell number increases, and cells occasionally desorb or delaminate from the epithelium. Furthermore, the geometric constraints of cell divisions likely arise from mechanical forces and biophysical properties, such as membrane elasticity, contractility and cell–cell adhesion. This raises the question of how these forces and biophysical properties shape cells within a growing epithelium. Farhadifar et al. (2007) examined this issue by developing a model in which the positions of vertices in a growing epithelium are determined by the minimization of energy associated with the contractility of the cortical actin–myosin network, line tension along apical junctions and cell surface elasticity. Their model predictions of frequency of cell delamination, cell area and polygonal shapes matched those with the developing Drosophila wing disc only for specific ranges of parameter values. These results suggest that the biophysical properties of epithelial cells are wired to give rise to the observed cell shapes in growing epithelial tissues. It would be interesting to determine whether these parameter values are also necessary to give rise to the geometric rules of cell divisions used in the Gibson model.

4.4. Line tension in compartmentation

Anisotropic line tension is involved not only in local re-shuffling of neighbouring cells during intercalation (figure 3b), but also in maintaining long-range barriers between two cell populations (figure 3e). This role of partitioning cell populations was first suggested in the context of DV demarcation of wing imaginal discs (Major & Irvine 2005, 2006). More recently, the magnitude of anisotropic line tensions has been directly measured and computationally modelled in AP demarcation of wing imaginal discs (Landsberg et al. 2009; Vincent & Irons 2009), and eliminating this line tension has been shown to compromise the re-establishment of AP compartmentation following cell divisions at this interface during Drosophila embryonic development (Monier et al. 2010). Consistent with the idea that anisotropic line tension may be a recurring motif for maintaining cell compartments, the above studies span AP and DV compartmentation in Drosophila wing discs and AP compartmentation in Drosophila embryonic development.

4.5. Contractility and cell–cell adhesion: opposing contributions to line tension?

In the above models of line tension along cell–cell junctions, contractility opposes cell–cell adhesion (figure 3). However, there is some evidence that contractility can influence the endocytosis of cell adhesion receptors (Sahai & Marshall 2002) while planar cell polarity proteins regulate the exocytosis and recycling of cell-adhesion proteins (Classen et al. 2005). Furthermore, in the case of cell adhesion to the ECM, contractility is essential for forming and maintaining focal adhesions; in a similar manner, contractile forces are involved in promoting the maturation of cell–cell adhesions (Yamada & Nelson 2007). Thus, it remains an open question to what extent cortical contractility and cell–cell adhesion ought to be viewed as independent opposing contributions to line tension.

5. Forces associated with cell behaviours

The loss, accrual and movement of cells owing to apoptosis, proliferation and migration, respectively, can generate local forces on direct neighbours and even propagate to affect tissue morphology at a mesoscopic scale.

5.1. Forces associated with apoptosis

The extrusion of apoptotic cells from an epithelial sheet has been observed in the context of various developmental processes and is essential to maintain the integrity of the epithelium and its barrier function. Rosenblatt et al. (2001) showed that an apoptotic cell within an epithelial layer rapidly develops an actomyosin ring around its periphery and signals to its neighbouring cells to induce actin cable formation at the interface between the apoptotic cell and neighbouring live cells (figure 4a). Rho-mediated contraction of these actin cables pulls neighbouring cells towards the apoptotic cell and extrudes the apoptotic cell out of its parental epithelia, rapidly sealing the opening that could have been left by the removal of the dead cell. In fact, selective blocking of Rho activity in neighbouring cells aborted the extrusion of the apoptotic cells completely, disrupting the integrity of epithelia. Thus, apoptotic force involves not only an autonomous contractile force in a cell undergoing death, but also collective force developed among live cells surrounding the apoptotic cell.

Figure 4.

Forces associated with apoptosis and their implications for Drosophila dorsal closure. (a) An actomyosin ring forms in both the cell undergoing apoptosis (grey) and its live neighbours. Contractile forces along this actomyosin ring generate a force (red arrows) that pulls neighbouring cells into the space occupied by the extruded cell and prevents gaps in the epithelium. (b) This force associated with apoptosis contributes to dorsal closure in the Drosophila embryo. The contractile forces generated by apoptotic cells (grey) in the AS contribute to tension (blue arrows) along the leading edge of the lateral epithelia tissues.

Such forces involved in the extrusion of apoptotic cells also propagate through cell–cell interactions to affect the long-range morphology of tissues. An example involves dorsal closure of the Drosophila embryo. During this process, an elliptical opening in the dorsal epidermis is occupied by the amnioserosa (AS) and is covered by two dorsally migrating epithelial leading edges with the two flanks advancing along the dorsal midline (figure 4b). A precise coordination of forces, including the contractility of the AS, contributes to sheet migration and dorsal closure (Kiehart et al. 2000). The apoptosis of AS cells contributes significantly to the contractility of this tissue and thus the rate of dorsal closure (Toyama et al. 2008). By quantitatively comparing the recoiling velocity of the leading edge of lateral epidermis upon laser ablation in wild-type and apoptotic mutants, it was estimated that apoptosis of aminoserosa cells accounts for approximately one-third to one-half of the net force developed at the leading edge of lateral epidermis. The contractile forces involved in extruding apoptotic cells may be transmitted by cell–cell contacts to the lateral epidermis, contributing the force needed for dorsal migration of lateral epithelia and fusion.

5.2. Mechanical stresses imposed by proliferation

In a growing tissue in which cellular rearrangements are restricted in the time scale of cell division, mechanical stresses imposed by an increase in cell mass (i.e. proliferation) are not fully released and thus rapidly accumulate in a local environment. One of the phenotypic features of rapidly growing tissue is that cell spreading against underlying substrate decreases with increasing cell density. Restricted cell spreading further correlates with a decrease in stress fibre formation, which in turn destabilizes focal adhesions. Consistent with these changes, when plated on varying sizes of adhesive patterns consisting of micropillars, cells grown on smaller islands exhibited significantly reduced cytoskeletal tension and contraction force (Tan et al. 2003). In addition, accumulation of mechanical stresses accompanying the aforementioned events has been correlated with cell cycle arrests in high-density culture (Liu et al. 2006).

Notably, coupled with other microenvironmental factors, force induced by proliferation plays a central role in patterning multicellular behaviours in the context of developmental processes (figure 5). Patterning and growth regulation of Drosophila wing imaginal discs involves the gradients of morphogens including Decapentaplegic (Dpp). However, while it is clear how reduced morphogen concentration far from the source would halt cell proliferation at the edge of a developing tissue, how cell proliferation and tissue growth stop near the morphogen source remained unclear. Shraiman (2005) theoretically showed that, at the region of high morphogen concentration, mechanical stresses rapidly accumulate as a result of the high rate of proliferation. This accumulated mechanical stress in turn inhibits morphogen-induced proliferation. Thus, once cell proliferation ceases at the edge of a developing tissue owing to low morphogen concentration, continued proliferation near the morphogen source would escalate the local mechanical stress and stop the growth of tissue as a whole. Thus, mechanical stresses would serve as a local negative regulator of growth, thereby affecting growth patterns and organ size (Hufnagel et al. 2007).

Figure 5.

Role of proliferation-induced mechanical stresses in growth patterning and organ size determination. During development of Drosophila wing imaginal disc, spatial gradients in morphogens regulating cell growth are established by their localized secretion and transport from source cells (red cells). The graph depicts the steady-state gradient in morphogen concentration as a function of distance from the morphogen source. Below the graphs, grey-scale colour gradients in the cells indicate the level of mechanical stress owing to crowding. At early stages of the development (top cell array), cells proliferate uniformly, expanding the epithelium without local accumulation of mechanical stresses. Later, cell growth ceases at the edge of the epithelium owing to low morphogen level (bottom cell array), and mechanical stresses accumulate near the morphogen source as a result of imbalanced growth. It is postulated that this mechanical stress may desensitize cells to the locally high levels of growth factor and lead to mechanics-induced cell cycle arrest (Shraiman 2005).

5.3. Forces driving collective migration

The mechanics of migration in single cells have been widely studied, revealing the importance of protrusive forces that drive the extension of the leading edge of the cell and contractile forces that detach the trailing edge and pull the cell body forward (Lauffenburger & Horwitz 1996). However, less is known about the mechanics of sheets and strands of cells moving together, a process known as collective migration. Given that these cells not only adhere to the surrounding matrix but also remain in contact with each other through cell–cell adhesion proteins, such as cadherins, one would expect the interplay between mechanical forces involved in cell–cell and cell–matrix adhesion to play a major role in the behaviour of such systems. Understanding how collective migration forces are generated and transmitted between cells has important implications in disease and physiology.

Collective migration is a key phenomenon in tissue morphogenesis and is widely observed in developing organisms (Friedl & Gilmour 2009; Rørth 2009). Wound healing is a classic example of collective motility, and in vitro assays of this process have provided a powerful model system to study the movement of two-dimensional cell sheets. Other examples include border cell motility during Drosophila ovary development (Rørth 2002; Montell 2003) and branching morphogenesis in mammary epithelia (Ewald et al. 2008).

Two important questions arise concerning how the forces that lead to wound closure are generated. The first is whether wound healing is driven predominantly by proliferation within the monolayer that pushes it forward or whether cell migration propels the healing process by pulling the sheet into the wound. The emerging consensus appears to be that cell migration at the healing front is the key driver with proliferation helping to maintain the monolayer (Gov 2007). Several studies have suggested a leader–follower model, wherein the leading cells at the wounded edge migrate and pull along the trailing cells. For example, in both IAR-2 and MDCK epithelial sheets exposed to model wounds, leaders temporarily lose their epithelial character and develop lamellipodia and focal adhesions that protrude into the wound (Omelchenko et al. 2003; Poujade et al. 2007). Leader–follower behaviour is also observed in a wounded endothelial monolayer in the presence of fibroblast growth factor (Vitorino & Meyer 2008).

Rho-dependent cytoskeleton reorganization appears to play a significant role in the leader–follower model of wound healing. Omelchenko et al. (2003) note that leader cells disassemble their cortical actin cables upon wounding and reorient filaments perpendicular to the advancing front. Follower cells maintain their cortical actin cables but exhibit radial, rather than tangential, cell–cell contacts with leaders, indicating that tension is generated by the leader cells. Poujade et al. (2007) observe similar behaviour in leader cells and also note the development of a supracellular actin belt in follower cells that may transmit force as in the purse-string wound closure mechanism.

If we accept the notion that wound healing is driven primarily by cell migration rather than proliferation, a second question that arises is where the traction forces necessary for migration are generated. In the simplest leader–follower model, traction forces would be generated by the first row of cells (i.e. the leaders), so that followers need only to release their attachments and be pulled forward. Recent findings, however, suggest that this is not the case and that, instead, the traction forces involved in propelling wound healing may be generated by cells much further into the monolayer (Trepat et al. 2009). Thus, in growing MDCK sheets, significant traction forces are observed far away from the leading edge. Furthermore, a force balance shows that the tensional stress is propagated into and accumulates within the sheet, suggesting long-range transmission of forces from the leading edge into the interior of a growing epithelial sheet (Ladoux 2009). This view may be supported by the observation of cryptic lamellipodia protruding from submarginal cells in the direction of the wound as well as the ability for these cells to compensate for a loss of motility in the first row of the advancing edge (Fenteany et al. 2000; Farooqui & Fenteany 2005).

The observations from these studies indicate that collective sheet migration and wound healing may occur by different modes depending on the tissue environment. In one extreme, cells within a monolayer may behave nearly autonomously and generate their own motile forces (Bindschadler & McGrath 2007). While leader cells may be present in such cases, they act primarily to guide or polarize their followers to move in the direction of the wound. In the other extreme, leader cells may exert enough force to physically drag follower cells behind them (Friedl et al. 2004). Most observations appear to suggest a mode in which both behaviours are important. As a result, the migration of each cell arises from its own traction forces as well as the forces exerted by its neighbours. The relative strength of these forces could depend on a number of factors, including monolayer size and density, the strength of cell–cell adhesions, the matrix over which the sheet is migrating and the presence of soluble factors. Such behaviour would be consistent with the recent hypothesis that collective morphogenic movements are controlled in vivo by modular mechanical properties (Montell 2008).

6. Concluding remarks

The modes of force propagation described in this review are recurring motifs as they contribute to morphodynamics across several distinct multicellular contexts. An intriguing possibility is that these and other force transmission modalities may enable a well-defined set of multicellular transformations. Indeed, seemingly diverse morphological patterns observed in vivo may be an outcome of different coupling and executions of these common motifs. For example, diverse epithelial morphogenetic phenotypes observed during dorsal closure and germ band extension in the Drosophila embryo and during convergence of the zebrafish trunk neural ectoderm are simply quantitative combinations of cellular deformation and intercalation (Blanchard et al. 2009). The rapidly growing interest in dynamical imaging of development in several model organisms should add to these findings and provide a more complete description of possible multicellular transformations. Tissue morphodynamics may ultimately be understood as different spatio-temporal combinations of a select few multicellular transformations, which in turn are driven by a small group of mechanotransduction motifs that operate at the bicellular to multicellular length scale.

Acknowledgements

We thank the Research Centre of Excellence in Mechanobiology at the National University of Singapore for hosting A.R.A. during the development of some of the ideas presented in this review; the members of the Asthagiri group for helpful discussions; and two anonymous reviewers for their suggestions.

This work was supported by the NCI Physical Sciences of Oncology Center at USC (U54CA143907) and the National Institutes of Health grant R01-CA138899.

Footnotes

  • One contribution to a Theme Supplement ‘Mechanobiology’.

    • Received February 8, 2010.
    • Accepted March 12, 2010.

References

View Abstract