Figure 1. Adherent cells crawl along surfaces by extending protrusions (pseudopods) in the direction of motion, as shown in the central panel. These protrusions attach to the substrate, and then the cell draws its body forward. In many cases this forward movement is driven by internal contractions, followed by release of the rearmost attachments. The cycle repeats with the extension of new pseudopods. Different cell types show a variety of mechanisms. The velocity, persistence of directionality, and morphology of cells differ widely. Shown are several examples of extensively studied motile cell types. Keratocytes move rapidly while maintaining a constant half-moon shape. Amoeba proteus is a traditional model for amoeboid movement; the mechanism of protrusion formation is quite different from that in most tissue cells. Crawling fibroblasts and leucocytes differ in their degree of attachment to the substrate, and in their typical velocities. Fibroblasts move slowly while applying forces to the extracellular matrix much larger than those required for movement, while leucocytes are weakly attached to the substrate and move more rapidly. Nerve cells do not translocate their bodies, but they extend long processes by means of growth cones at the tips that behave very much like motile cells.

Figure 2. Two mechanisms for the formation of pseudopods are shown. They are probably extreme idealizations of what would normally be a mixture of the two. The mechanism most intensively studied in recent years is based on polymerization of cytoskeletal filaments, normally actin, within the lipid plasma membrane envelope. In the case of nematode sperm cells, protrusion is driven instead by polymerization of a protein known as major sperm protein (MSP). Another mechanism is based on the effect of fluid pressure created by contraction of the cell cortex, which forces the membrane to bulge in a direction where the cortex is weakened. While this classic model is most suggestive in amoebae, for which it was first proposed, it may operate to some extent in other cell types as well.

Figure 3. (a) Lamellipodial and filopodial extensions are filled with a network of actin filaments, organized into very specific structures. It is generally believed that the formation of these networks provides the driving force to extend the membrane. (b) An experimental model showing that actin polymerization can produce forces is provided by intracellular movement of bacteria such as Listeria. It has been shown that the ‘comet tail’ behind the bacterium has many features in common with the lamellopodium.

Figure 4. The production of force by polymerization raises questions about how filaments can extend when near a wall or membrane. Fluctuation-based models (Mogilner and Oster, 1996) suggest that undulations of the membrane, or of the filaments themselves, leave enough space for occasional addition of monomers. When these undulations straighten, the membrane is driven forward. Quantitative estimates indicate that, in order for this mechanism to work in the specific case of actin and cell membranes, the filaments must approach the membrane at an angle.

Figure 5. The formation of a branching actin network requires several accessory proteins. The key element is the Arp2/3 complex, shown in yellow-green, which is regulated by binding of a protein called N-WASP that in turn is under control of the small G protein Cdc42 and a phospholipid PIP2. The activated Arp2/3 complex binds to the sides or tips of existing actin filaments (chevron elements representing actin monomers) and nucleates the growth of new filaments at a typical angle of 70°. The source of new actin monomers is from a complex of actin with profilin, which limits extension to one end (the ‘plus’ end) of the polar filament. At the same time the older parts of the filaments bind cofilin, which promotes their disassembly. The extension of the filaments is also limited by binding of a capping protein to the growing ends. Each of these elements provides a means of regulating the network formation.

Figure 6. Many types of myosin act as molecular motors that apply mechanical forces to actin filaments. The myosins differ primarily in the tail region that governs their attachment to other cellular components. The head regions of most types (designated collectively by the asterisk) are similar in structure, and move directionally along the actin from the minus to the plus end. Myosin VI is an exception that moves in the opposite direction. This movement can drive various types of cargo along the actin. If anchored to a stationary object, myosins can also cause directional sliding of actin filaments. Myosin II has the unique ability to form bipolar filaments, which can move oppositely oriented actin arrays towards each other. It serves as the drive for most types of contraction in the cell, including muscle contraction.

Figure 7. A network of small G proteins of the Rho family act as a sort of biochemical computer to integrate external signals and induce restructuring of the cytoskeleton. The main players are Cdc42, Rac and Rho; they are active in GTP-bound form and inactive when bound to GDP. They work via an assortment of proteins, some of which are listed in red, which lead to the reactions listed below the arrows via a further sequence of events. External factors can activate each of these G proteins via the response of membrane receptors. In addition to their effects on the cytoskeleton, several feedback paths have been identified by which these proteins regulate each other's activity. Basically, Cdc42 activates (indicated by arrowheads) Rac, and Rac activates Rho, but under some conditions Rac and Cdc42 can inhibit Rho (indicated by round head). Rho can also inhibit Rac. The interactions among these controlling elements are a subject of active research.

Figure 8. The scheme shows a hypothesis (Elbaum et al., 1999) for the role of microtubules in development of focal adhesions and cell polarization. In (1), new actin-filled protrusions are formed ahead of the cell body. In (2), microtubules (Mt) then grow into the newly formed compartments. Microtubule growth is associated with activation of Rac (Waterman-Storer et al., 1999) and Rac stimulates the formation of lamellipodia and small focal complexes (fc). It is possible that microtubules may deliver some components (c) involved in development of these structures. The maturation and strengthening of the initial cell–substrate contacts is blocked as long as microtubules remain nearby (Kaverina et al., 1999). Microtubules are dynamic structures, however, and depolymerize in a process known as dynamic instability. Depolymerization of microtubules is associated with activation of Rho (Ren et al., 1999), which in turn promotes actomyosin contraction and tensile force applied to the focal complexes, as seen in (3). The contraction may cause a retraction of the cell edge (4a), or conversely the strengthening of the initial focal complexes into mature focal adhesions (fa) associated with actin bundles called stress fibres (4b). In either case the process can then repeat from step (1). In this way the dynamic instability of microtubules can test the strength of initial contacts around the cell edge, and select the strongest for further development. Such a positive feedback mechanism can lead to polarization of the shape and selection of a direction for locomotion.