Special Reviews

Morphogenetic Cell Movements: Diversity from Modular Mechanical Properties

See allHide authors and affiliations

Science  05 Dec 2008:
Vol. 322, Issue 5907, pp. 1502-1505
DOI: 10.1126/science.1164073


Animal tissue and organ development requires the orchestration of cell movements, including those of interconnected cell groups, termed collective cell movements. Such movements are incredibly diverse. Recent work suggests that two core cellular properties, cell-cell adhesion and contractility, can largely determine geometry, packing, sorting, and rearrangement of epithelial cell layers. Two additional force-generating properties, the ability to generate cell protrusions and cell adhesion to the extracellular matrix, contribute to active motility. These mechanical properties can be regulated independently in cells, suggesting that they can be employed in a combinatorial manner. A small number of properties used in combination could, in principle, generate a diverse array of cell shapes and arrangements and thus orchestrate the varied morphogenetic events observed during metazoan organ development.

The formation of complex organs requires the concerted development of groups of cells, including coordinated regulation of their shapes and movements. Many different types of cell movements occur during normal embryonic development, including those of interconnected cell groups, known as collective cell movements. A striking feature of such movements is their diversity: Cells move in clusters, strands, sheets, and tubes, in addition to the well-studied migrations of individual cells. The variety of such movements raises the question of whether they share common cellular and molecular mechanisms, and if so, how distinct morphologies and arrangements arise.

The dominant conceptual framework over the past 15 to 20 years for the regulation of cell motility during embryonic development and tumor cell dissemination has been the epithelial-to-mesenchymal transition (EMT) and its reverse. The idea is that polarized epithelial cells exhibit a high degree of cell-cell adhesion and are not motile, whereas mesenchymal cells exhibit little cell-cell adhesion and are often highly motile (14). Conversion of epithelial cells into mesenchymal cells is thought to result from down-regulation of E-cadherin gene expression by the transcriptional repressor Snail and its relatives and from up-regulation of cell/matrix interactions via integrins (1, 4). The mechanisms of integrin-mediated migration of individual mesenchymal cells are well-established (5), and thus the conversion of epithelial cells to mesenchymal cells is thought to render them motile.

The EMT framework works well to explain the conversion of stationary epithelial cells, such as cells of the neural tube, into individual migratory mesenchymal cells, such as neural crest cells. However, EMT does not describe the myriad other types of morphogenetic cell movements that drive embryonic development, as discussed below. In particular, this concept does not offer insight into collective cell movements because such cells by definition retain cell-cell adhesion while moving.

A great deal of work in model organisms such as flies, frogs, and fish, has resulted in the identification of many specific molecules that participate in signaling networks and cytoskeletal dynamics and thereby contribute to different types of collective cell movements (6). However, at times the molecules seem to confound rather than clarify. For example, fibroblast growth factor signaling pathways exert virtually opposite effects on different cell types: stimulating protrusion and migration in Drosophila tracheal cells but causing zebrafish cells to stop moving (7). Planar cell polarity takes a lead role in cell movements of gastrulation (8, 9) but a negligible one in others (10). So how do we organize all of this information?

In his treatise On Growth and Form, originally published in 1917, D'Arcy Thompson articulated the view that morphogenesis should be approached as a problem of mechanics, the branch of physics concerned with forces and motion (11). This highly original concept did not reach the mainstream and fell further out of favor as the notion of genetic control of morphogenesis emerged (12). However, recent studies, reviewed here revive the contention that mechanical forces are the fundamental shapers of cells and embryos. They further suggest that, if used in a combinatorial manner, a small number of modular mechanical (i.e., force-generating) properties—namely cell-cell adhesion, cell-matrix adhesion, protrusion, and contractility—could orchestrate morphogenesis. Interpreting the functions of specific molecules and signaling pathways within the framework of the elementary forces that shape cells and tissues may provide an opportunity to unite the mechanical view of morphogenesis with modern molecular genetics and to explain the otherwise daunting variety of cell shapes, arrangements, and movements.

Developmental Cell Movements Are Diverse

The idea that developmental cell movements are diverse and can be distinct from EMT is illustrated with a few examples from Drosophila development. The early fruit fly embryo consists of a single layer of epithelial cells that the process of gastrulation converts into multiple layers via numerous and complex cell motions (1315). One of these is known as germband elongation (Fig. 1A), which doubles the length the participating cell group (the germband) and reduces its width by half (16). At the tissue level, the effect is dramatic, but at the level of each individual cell, the movements are relatively subtle: Cells exchange neighbors in a directional manner, never losing their epithelial characteristics.

Fig. 1.

Morphogenetic cell movements are diverse. (A) Schematic representation of germband elongation in the early Drosophila embryo. Three rows of cells are marked with colored dots to highlight their changes in position and shape. (Right) The same cells 30 min later. Individual cells intercalate, and the group lengthens and narrows. (B) Schematic drawing of one branch of the developing Drosophila tracheal tree at three time points. Cells at the tip exhibit obvious, elaborate protrusions whereas others do not. (C) Schematic representation of Drosophila follicle cell rearrangements during egg chamber development. Follicle cell nuclei, dark blue; nurse cell nuclei, gray; basement membrane, pink. At stage 7 of development (left), follicle cells are uniformly cuboidal in morphology. At stage 9 (right), follicle cells in different regions (boxed) undergo different morphological transformations. (D) Migration of the zebrafish lateral line primordium. (Top) Schematic of a fish embryo showing the overall migration path. (Bottom) Micrograph of migrating cells and one rosette (white bracket) being left behind.

After gastrulation, cells that form the branched tubular network of the Drosophila tracheal system begin their extraordinary movements (Fig. 1B) [see also (17)]. Initially, ∼80 cells on the surface of each half-segment of the embryo invaginate (18). Specific subsets become protrusive and actively migrate. The rest do not extend obvious protrusions but instead intercalate and thus lengthen and narrow the tube. During this process, tracheal cells retain their epithelial apical/basal polarization and adhere to one another; thus, they do not undergo EMT (6).

In the adult Drosophila ovary (Fig. 1C), morphogenetic movements occur that are distinct from those described above. The functional unit of the ovary is the egg chamber. Each egg chamber is composed of ∼650 cuboidal epithelial follicle cells surrounding 16 germline cells. The germline cells include 15 nurse cells, which are support cells, and one oocyte, which develops into the egg. As oogenesis progresses, three distinct morphogenetic changes occur within the follicle cell epithelium (19). Most of the cells assume a columnar morphology (Fig. 1C), elongating their lateral cell surfaces while shrinking their apical and basal domains. Meanwhile, cells in the anterior half of the egg chamber flatten out into a squamous morphology. This transition requires the lateral membrane to shrink while the apical and basal surfaces expand. In contrast, cells at the anterior pole of the egg chamber (the border cells) cluster together, extend protrusions in between the nurse cells, detach from the rest of the epithelium, and then migrate as a cohesive cluster to the oocyte (20).

The morphogenetic movements described above do not resemble each other or EMT. Anterior follicle cells, for example, reduce cell-cell adhesion and expand their contact area with the basement membrane, alterations normally associated with EMT. Yet these cells do not leave the epithelium, migrate long distances, or change neighbors. Border cells, on the other hand, express a higher level of E-cadherin than do the squamous anterior follicle cells (21) and retain apical basal polarity (22), properties usually associated with stationary epithelial cells. Yet, they invade the neighboring group of nurse cells, leave the follicle cell epithelium completely, and migrate.

Vertebrate embryos also exhibit collective cell movements that do not resemble classical EMT. One dramatic example is migration of the zebrafish lateral line. The lateral line is a group of ∼100 cells that migrates down the length of the zebrafish embryo (Fig. 1D). A small group of leader cells provides direction to the followers, though most of the cells are actively motile (23). Periodically, at the rear of the mass, a group of cells coalesces, stops moving, and is left behind to form a sensory organ. Motile cells of the lateral line exhibit some characteristics of epithelial cells, including apical/basal polarity, and other characteristics of mesenchymal cells, such as filopodial projections (7).

Some cells, such as neural crest cells and limb muscle precursors (24), undergo complete EMT, but as illustrated by the examples provided above, EMT does not provide an adequate conceptual framework for understanding collective cell movements. How then do groups of cells, some with features of both epithelial and mesenchymal cells, organize themselves in diverse and complex ways? The answer may lie in considering variations in the mechanical properties of cells.

Core Mechanical Properties Control Cell Geometry, Sorting, and Motility

To understand collective cell movements, we need to consider cell shapes, arrangement, and motility. Two cellular properties that repeatedly emerge as critical determinants of cell shape, packing geometry, and sorting are cell-cell adhesion and cell contractility (2530). It has long been appreciated that quantitative and qualitative differences in cell-cell adhesion promote sorting of distinct cell types in vitro (31). For example, when cells expressing two different types of cell-cell adhesion molecules (such as E-cadherin and N-cadherin) are mixed together, they sort apart from one another. However, it is now becoming clear that surface tension, a property that depends on the opposing forces of both cell-cell adhesion and contractility of the cortical cytoskeleton, is likely to be a fundamental determinant of cell shape, packing, and sorting (26, 30, 32).

In a uniform, non-dividing epithelium, cells tend to form a hexagonal array that can be modeled mathematically by considering the forces of cell-cell adhesion and cortical contractility, acting in opposition (28). Whereas contractility (a constricting force exerted by myosin on the cortical F-actin cytoskeleton) acts to minimize a cell's contact surface with its neighbors, the effect of cell-cell adhesion is to increase it. An ordered hexagonal array of cells represents a ground state and an energy minimum in which these forces are balanced (28, 29). At a molecular level, cadherins contribute substantially to overall cell-cell adhesion in many embryonic cell types, and cortical F-actin and its associated myosin are major determinants of contractility. Recent atomic force measurements in zebrafish embryos provide direct evidence for this (30).

The introduction of an asymmetry in either adhesion or contractility would be expected to perturb the orderly hexagonal array of an epithelium. This is precisely the mechanism that drives directional cell intercalation during Drosophila germband elongation. Cells of the early Drosophila embryo initially form a regular array (Fig. 2A) with symmetrical localization of myosin (33, 34), F-actin, and E-cadherin (8). However, during germband elongation, myosin becomes concentrated specifically on anterior and posterior surfaces and depleted from dorsal and ventral cell surfaces (Fig. 2B). This asymmetric myosin accumulation exerts a directional contractile force that preferentially shrinks particular sides of each cell (33) (Fig. 2C) and increases the disorder of the epithelium (8). As a result, instead of a fairly uniform hexagonal array, cells of the elongating germband exhibit a variety of polygonal shapes, some of which form clusters of four to seven cells that all meet at a vertex. These transient structures consistently resolve in a directional manner so as to promote the overall lengthening and narrowing of the cell sheet. These cell rearrangements fail if myosin activity is inhibited or remains uniformly distributed (33). Thus, enhancement of contractility in one region of a cell over another can yield directed cell rearrangement within the plane of the epithelium.

Fig. 2.

Localized myosin drives cell shape changes and germband elongation. (A) At stage 6, before germband elongation, the epithelium is uniform, as are myosin (red) and bazooka (green). (B) During germband elongation, myosin II (red) accumulates specifically in anterior and posterior (vertical) membranes and is depleted from dorsal and ventral (horizontal) membranes. Bazooka localizes in a complementary pattern. (C) Membranes rich in myosin (red) shrink.

Competing forces of cell-cell adhesion and cortical contractility also determine the particular shapes of cone and pigment cells in the Drosophila eye. The eye is composed of repeating units of a defined number of cells. Within each unit, in a particular plane of cross section, two primary pigment cells surround four central cone cells and are in turn framed by secondary and tertiary pigment cells in a highly reproducible pattern. Beginning with random shapes, a computer simulation that allows the cells to reorganize so as to achieve an energy minimum recapitulates the precise cell shapes found in the wild-type eye (27). The simulations are based on a mathematical model that takes into account the balance of forces created by differential cell-cell adhesion and contractility of the cell cortex. This model reproduces both the beautiful regular wild-type pattern (Fig. 3) and altered shapes in a handful of mutant conditions (26). In contrast, models that account for only differential cell-cell adhesion, without considering membrane tension due to the contractility of the underlying cortical cytoskeleton, do not reproduce the proper patterns. In the future, we might hope that mathematical modeling and computer simulations will be powerful enough to account for the relative arrangements of cells in the cluster (in addition to cell shapes), starting from random values.

Fig. 3.

Computer simulation reproduces cell shapes in the fly eye. (Left) Micrograph of a single unit in the Drosophila eye. (Right) Result of a computer simulation that starts with random cell shapes. Cells are allowed to change shape according to competing forces of cell-cell adhesion and cortical contractility. Four cone cells (c) are surrounded by two primary pigment cells (1°) that are, in turn, framed by secondary and tertiary pigment cells.

In early zebrafish embryos, the forces of cell-cell adhesion and cortical contractility have been measured directly for three different cell types, using atomic force microscopy (30, 35). Moreover, these measured values have been used in a mathematical model to predict cell sorting behavior. Both in vitro and in vivo tests of the model confirm that proper layering of ectoderm, mesoderm, and endoderm precursor cells can be predicted on the basis of their known adhesive and contractile properties, as long as extraembryonic cells that are normally present in the embryo are included in the simulation (30).

Whereas variation in cell-cell adhesion and cortical contractility is sufficient to determine geometry and sorting of cell types in the Drosophila eye and the zebrafish embryo, active cell motility requires an additional property, that of protrusion. The ability to extend and retract protrusions is an obvious feature of migrating border cells, tracheal tip cells, and the migrating lateral line, even though all of these cells retain epithelial character. In contrast, cells in the Drosophila germband are not protrusive, nor are cells in the developing eye. Tracheal cells that are not at the tip do not extend obvious protrusions, nor do the rosette cells that have stopped migrating at the back of the zebrafish lateral line. Border cells, tracheal tip, and migrating lateral line cells migrate actively and directionally into neighboring tissues, whereas the cell types that lack protrusions do not. Therefore, although cells seem to be able to undergo local rearrangements within the plane of an epithelium in the absence of protrusions, directional migration is highly correlated with this property.

Variations in cell/matrix interactions are also apparent among different collective cell movements. In the egg chamber, for example, there is a basement membrane, produced by the follicle cells, that surrounds the whole structure (Fig. 1C). Border cells detach completely from it and discontinue production of matrix proteins such as collagen IV and laminin (36). In contrast, squamous cells increase their contact area with the basement membrane as they spread, and columnar follicle cells reduce their contact area although they remain attached.

Diversity from Combinatorial Use of Protrusion, Contractility, Cell-Cell Adhesion, and Cell-Matrix Adhesion Modules

At the molecular level, cell-cell adhesion, cell-matrix adhesion, cortical contractility, and cellular protrusions are modular properties controlled by largely different sets of molecules. Protrusion is driven by actin polymerization (37). Cell-cell adhesion is controlled by cadherins (38) and a variety of other molecules. Cell-matrix adhesion is regulated primarily by the interactions of integrins and their ligands (39), although other matrix components and receptors certainly contribute. Myosin and its regulators (such as Rho kinase) control contractility. There is evidence for cross talk between these modules (40, 41) through the actin cytoskeleton, which is probably important for coordinating these mechanical properties; however, it is also clear that each characteristic can vary independently among cell types (30).

There are numerous examples in biology where small numbers of modular components are used in a combinatorial manner to generate complexity. In this way, four nucleotides (in combinations of three), generate 64 codons, and 20 amino acids combine together in various numbers to create hundreds of thousands of (and potentially many more) proteins. Similarly, cells may use a relatively small number of modular mechanical properties in various combinations to create diverse morphologies, arrangements, and movements. Simply varying four properties qualitatively, quantitatively, and spatially can result in hundreds of combinations, providing sufficient complexity to explain the diversity of observed cell geometries, topologies, and movements.

Different follicle cell types in the Drosophila ovary illustrate the concept of combinatorial use of mechanical properties to create diverse collective cell behaviors (Fig. 4). All of the cells start out in a uniform, cuboidal, hexagonal array. Border cells increase their expression of E-cadherin so as to adhere more strongly to each other (21). They extend long protrusions and require myosin activity to retract them (42, 43), and they down-regulate expression of matrix proteins (36). In contrast, squamous cells down-regulate expression of multiple cell-cell adhesion molecules, including E-cadherin, so as to adhere less well to one another, a prerequisite for their change in shape (19). These cells concomitantly expand their contact with the basement membrane, suggestive of elevated cell-matrix adhesion. Their flat shape suggests reduced cortical tension, but this notion remains to be tested. Columnar follicle cells exhibit a distinct combination of properties: E-cadherin expression greater than that of squamous cells and less than that of border cells (19), cell-matrix contact less than that of squamous cells but greater than that of border cells, and no evidence of protrusiveness. The observed differences in expression of cadherin, cell shape, and motility are consistent with the idea that each cell type has a distinctive combination of adhesive, contractile, and protrusive mechanical properties (Fig. 3). In principle, this prediction could be tested directly, using atomic force measurements such as those obtained from cells of the early zebrafish embryo (30).

Fig. 4.

Distinct combinations of mechanical properties characterize follicle cells with different shapes and behaviors. The relative strengths of the indicated properties are deduced from experimental observation of the cells' interactions with each other, the distribution of cadherins, and genetic evidence. They have not been measured directly. Cell-cell and cell-matrix refer to adhesiveness.


The great diversity of collective cell movements observed throughout development and organogenesis may result from combinatorial use of a small number of modular mechanical properties. EMTs represent a morphogenetic program that converts cells from one particular cell type (stationary columnar epithelial cells) into another cell type (individual mesenchymal cells). This particular transition involves a loss of cell-cell adhesion, up-regulation of cell-matrix adhesion, and increased protrusiveness and contractility. However, many other combinations of properties are possible, allowing cells to achieve a great diversity of individual cell shapes, as well as collective arrangements and movements. Cells with some properties of epithelial cells and some properties of mesenchymal cells may simply possess unique combinations of mechanical properties distinct from epithelial cells, mesenchymal cells, and other tissues.

Many interesting questions remain to be reviewed and investigated further. Key questions include how the myriad molecules and signaling pathways that regulate collective cell migration affect the core mechanical properties described here. It is probably generally the case that signaling pathways governing cell polarity (apical/basal, leading/lagging, and planar polarity) function to localize mechanical forces asymmetrically within cells. By definition, an asymmetry in force will cause dynamics. The mechanisms by which cell forces are coordinated in space and time represent another fascinating topic. By considering the effects of specific molecules and signaling pathways on the magnitude and direction of mechanical forces within and between cells, it may be possible to explain how individual cell shapes are achieved as well as how groups of cells sort and move, which, taken together, should go a long way toward explaining organ and tissue morphogenesis. Were he alive today, D'Arcy Thompson would probably be thrilled at the prospect of unifying molecular and cell biology with mechanics.

References and Notes

View Abstract

Navigate This Article