Chapter 16:  The Cytoskeleton

Introduction

A eucaryotic cell contains a billion or so protein molecules, which constitute about 60% of its dry mass. There are thought to be about 10,000 different types of protein in an individual vertebrate cell, and most of them are highly organized spatially. This organization is present at multiple levels. In all cells proteins are arranged into functional complexes, most consisting of perhaps 5 to 10 proteins but others as large or larger than ribosomes. A further level of organization involves the confinement of functionally related proteins within the same membrane or aqueous compartment of a membrane-bounded organelle, such as the nucleus, mitochondria, or Golgi apparatus. An even higher level of organization is created and maintained by the cytoskeleton. It enables the living cell, like a city, to have many specialized services concentrated in different areas but extensively interconnected by paths of communication. In this section we review some of the basic strategies that enable the cytoskeleton to control the spatial location of protein complexes and organelles, as well as to provide communication paths between them.

The Cytoplasm of a Eucaryotic Cell Is Spatially Organized by Actin Filaments, Microtubules, and Intermediate Filaments ¹

How can a eucaryotic cell, with a diameter of 10 mm or more, be spatially organized by cytoskeletal protein molecules that are typically 2000 times smaller in linear dimensions? The answer lies in polymerization.For each of the three major types of cytoskeletal protein, thousands of identical protein molecules assemble into linear filaments that can be long enough, if necessary, to stretch from one side of the cell to the other. Such filaments connect protein complexes and organelles in different regions of the cell and serve as tracks for transport between them. In addition, they provide mechanical support, which is especially important for animal cells, since they do not have rigid external walls. The cytoskeleton forms an internal framework for the large volume of cytoplasm, supporting it like a framework of girders supporting a building.

It is easy to see how filaments arose in evolution: any protein with an appropriately oriented pair of complementary self-binding sites on its surface can form a long helical filament (see p. 124). Each of the three principal types of protein filaments that make up the cytoskeleton is a helical polymer that has a different arrangement in the cell and a distinct function ( Figure16-2). By themselves, however, the three types of filaments could provide neither shape nor strength to the cell. Their functions depend on a large retinue of accessory proteins that link the filaments to one another and to other cell components. Accessory proteins are also essential for the controlled assembly of the protein filaments in particular locations, and they provide the motors that either move organelles along the filaments or move the filaments themselves.

Dynamic Microtubules Emanate from the Centrosome ²

Microtubules are polar structures: one end (the plus end) is capable of rapid growth, while the other end (the minus end) tends to lose subunits if not stabilized. In most cells, the minus ends of microtubules are stabilized by embedding them in a structure called the centrosome, and the rapidly growing ends are then free to add tubulin molecules ( Figure 16-3). The centrosome generally lies next to the nucleus, near the center of the cell.

At any one time, several hundred microtubules are growing outward from a centrosome, with some extending for many microns, so that their plus end is at the edge of the cell. Each of these microtubules is a highly dynamic structure that can shorten as well as lengthen: after growing outward for many minutes by adding subunits, its plus end may undergo a sudden transition that causes it to lose subunits, so that the microtubule shrinks rapidly inward and may disappear. The microtubule network that emanates starlike from the centrosome is constantly sending out new microtubules to replace the old ones that have depolymerized ( Figure 16-4).

The Microtubule Network Can Find the Center of the Cell ³

What determines how the cytoplasmic array of microtubules is normally positioned in a cell? Important clues have been provided by experiments on cultured pigment cells isolated from fish scales: large flat cells containing many pigment granules. The granules, which can be dark brown, yellow, red, or iridescent, depending on the species of fish, are attached to microtubules and can either aggregate in the center of the cell or disperse throughout the cytoplasm. The movement of the pigment granules occurs along the microtubules and can be controlled by the fish to change its skin color. In a cultured pigment cell, the movement can be conveniently controlled by applying hormones or other reagents that change the concentration of cyclic AMP in the cytosol: raising cyclic AMP causes the granules to disperse, whereas lowering it causes the granules to aggregate. The pigment granules therefore provide a useful marker for the arrangement of microtubules in the cell ( Figure 16-5).

If one part of a fish pigment cell is cut off with a needle, the cell fragment can survive for long periods even though it lacks a nucleus. The same operation, performed when the pigment granules are dispersed, causes some granules to be trapped in the cell fragment. If the pigment granules in the fragment are induced to aggregate by hormonal treatment immediately after the surgery, they move toward the site of the cut. But if they are induced to aggregate 4 hours after the surgery, they do not move to the cut site but instead move to the exact center of the cell fragment. Further investigation shows that this change results from a major rearrangement of the microtubules within the fragment, so that their minus ends are now at the center of the fragment, just as they were at the center of the intact cell. In effect, the isolated cell fragment has become a minicell with respect to its microtubule organization, the microtubules having reorganized around a new microtubule organizing center ( Figure 16-6).

This simple experiment suggests that the cytoplasmic array of microtubules emanating from the centrosome can act as a surveying device that is able to find the center of the cell. This is a useful starting point if the array is to be able to organize the cell interior. But it is only a starting point; as we see later in this introductory section, a cell can position the array by specifically moving its centrosome to a location displaced from the cell center.

Motor Proteins Use the Microtubule Network as a Scaffold to Position Membrane-bounded Organelles ⁴

As we have just seen in the case of fish pigment cells, cytoskeletal filaments serve not only as structural supports but also as lines of transport. If a living vertebrate cell is observed in a light microscope, its cytoplasm is seen to be in continual motion. Over the course of minutes, mitochondria and smaller membrane-bounded organelles change their positions by periodic saltatory movements, which are much more sustained and directional than the continual small Brownian movements caused by random thermal motions. These and other intracellular movements in eucaryotic cells are generated by motor proteins, which bind to either an actin filament or a microtubule and use the energy derived from repeated cycles of ATP hydrolysis to move steadily along it (see p. 208). Dozens of different motor proteins have now been identified. They differ in the type of filament they bind to, the direction in which they move along the filament, and the "cargo" they carry.

The first motor protein to be discovered was myosin, a protein that moves along actin filaments and is especially abundant in skeletal muscle, where it forms a major part of the contractile apparatus. Other types of myosins were subsequently found in nonmuscle cells. All myosins have similar motor domains (the part of the protein that generates movement), but they differ markedly in the domains that are responsible for attaching the myosin molecule to other components of the cell.

The motor proteins that move along microtubules are distinct from the myosins and belong to one of two families: the kinesins, which generally move toward the plus end of a microtubule (away from the centrosome), and the dyneins, which move toward the minus end (toward the centrosome). As with the myosins, each type of microtubule-dependent motor protein carries a distinct cargo with it as it moves ( Figure 16-7).

Microtubule-dependent motor proteins play an important part in positioning membrane-bounded organelles within a eucaryotic cell. The membrane tubules of the endoplasmic reticulum (ER), for example, align with microtubules and extend almost to the edge of the cell, whereas the Golgi apparatus is located near the centrosome. When cells are treated with a drug that depolymerizes microtubules, both of these organelles change their location: the ER collapses to the center of the cell, while the Golgi apparatus fragments into small vesicles that disperse throughout the cytoplasm. When the drug is removed, the organelles return to their original positions, dragged by motor proteins moving along the re-formed microtubules. Thus the normal position of each of these organelles is thought to be determined by a receptor protein on the cytosolic surface of its membrane that binds a specific microtubule-dependent motor - a kinesin for the ER and a dynein for the Golgi apparatus ( Figure16-8).

The Actin Cortex Can Generate and Maintain Cell Polarity ⁵

In general, microtubules in the cytoplasm function as individuals, whereas actin filaments work in networks or bundles. Actin filaments lying just beneath the plasma membrane, for example, are cross-linked into a network by various actin-binding proteins to form the cell cortex. As we discuss later, the network is highly dynamic and functions with various myosins to control cell-surface movements. The location and orientation of the cortical actin filaments are controlled by nucleation sites in the plasma membrane, and different regions of the membrane direct the formation of distinct actin-filament-based structures.

Localized extracellular signals that impinge on a portion of the cell surface can induce a local restructuring of the actin cortex beneath the corresponding part of the plasma membrane. In a reciprocal way the organization of the actin cortex can have a major influence on the behavior of the overlying plasma membrane. Mechanisms based on cortical actin filaments, for example, can push the plasma membrane outward to form long, thin microspikes or sheetlike lamellipodia, or they can pull the plasma membrane inward to divide the cell in two ( Figure 16-9).

In extreme cases the actin cortex can integrate movements of an animal cell over its entire surface and maintain cell polarity independently of the microtubule array. This is illustrated by experiments on a nonpigmented type of cell isolated from fish scales. These epidermal cells, known as keratocytes, migrate unusually rapidly in culture, traveling at speeds of 30 µm/minute or more. Immunostaining with antibodies indicates that intermediate filaments and microtubules are present only in the trailing region around the cell nucleus, whereas the flattened leading edge of the cells is rich in actin filaments ( Figure16-10). Furthermore, cells that are treated with a drug that depolymerizes microtubules migrate just as rapidly as untreated cells, whereas the migration is immediately halted by agents that interfere with actin filaments. Evidently, actin filaments (acting with other proteins) are able to move a keratocyte over a surface and also maintain this cell's distinctive shape and polarity; the details of the mechanism involved, however, are unclear.

Actin Filaments and Microtubules Usually Act Together to Polarize the Cell ⁶

In a living cell the three major types of cytoskeletal filaments are connected to one another and their functions are coordinated. The distribution of intermediate filaments in an epithelial cell in culture, for example, is radically altered if the microtubules are depolymerized by drug treatment: the intermediate filaments, which are normally arrayed throughout the cytoplasm, pull back to a region close to the nucleus. There are also many situations in which microtubules and actin filaments act in a coordinated way to polarize the whole cell. We discuss just one example: the killing of specific target cells by cytotoxic T lymphocytes.

Cytotoxic T cells kill other cells that carry foreign antigens on their surface. This is an important part of a vertebrate's immune response to infection, as discussed in Chapter 23. When receptors on the surface of the T cell recognize antigen on the surface of a target cell, the receptors signal to the underlying cortex of the T cell, altering the cytoskeleton in several ways. First, proteins associated with actin filaments in the T cell reorganize under the zone of contact between the two cells. The centrosome then reorients, moving with its microtubules to the zone of T-cell-target contact ( Figure 16-11A). The microtubules, in turn, position the Golgi apparatus right under the contact zone, focusing the killing machinery - which is associated with secretion from the Golgi - on the target cell.

In this example, as in many others, a cell becomes polarized in the following general way. First, the plasma membrane senses some difference on one side of the cell that generates a transmembrane signal. The actin cortex is then reorganized in a local area beneath the affected membrane, which in turn moves the centrosome to that part of the cell, presumably by pulling on its microtubules. The centrosome in turn positions the internal membrane systems in a polarized way. The net result is a cell with a strong directional focus ( Figure 16-11B).

The Functions of the Cytoskeleton Are Difficult to Study

Although the main subunits of the three classes of cytoskeletal polymers, as well as many of the hundreds of accessory proteins that associate with them, have been isolated and their amino acid sequences determined, it has been frustratingly difficult to establish how these proteins function in the cell. Besides the complexity that stems from the large number of proteins involved, two general features make the cytoskeleton especially difficult to understand. First, the function of the cytoskeleton depends on complex assemblies of proteins, which bind in cooperative groups to the cytoskeletal filaments. It is relatively straightforward to examine the effect on a filament of a single accessory protein but very much more difficult to analyze the effects of a mixture of many different proteins. This problem is not unique to the cytoskeleton, but it is especially acute here. Secondly, the functions of the cytoskeleton are much more difficult to analyze than the functions of many other large protein complexes. The processes of RNA and DNA synthesis, for example, which involve the formation of new polymers held together by covalent bonds, can be readily analyzed in vitro, in part because the products of the in vitro reactions can easily be measured and compared with the corresponding products made in a cell. The cytoskeleton, in contrast, exerts forces and generates movements without any major chemical change. This makes it especially difficult to assay the function of a cytoskeletal system that has been reconstituted in vitro from purified components.

Summary

The cytoplasm of eucaryotic cells is spatially organized by a network of protein filaments known as the cytoskeleton. This network contains three principal types of filaments: microtubules, actin filaments, and intermediate filaments. Microtubules are stiff structures that usually have one end anchored in the centrosome and the other free in the cytoplasm. In many cells microtubules are highly dynamic structures that alternately grow and shrink by the addition and loss of tubulin subunits. Motor proteins move in one direction or the other along microtubules, carrying specific membrane-bounded organelles to desired locations in the cell. Actin filaments are also dynamic structures, but they normally exist in bundles or networks rather than as single filaments. A layer called the cortex is formed just beneath the plasma membrane from actin filaments and a variety of actin-binding proteins. This actin-rich layer controls the shape and surface movements of most animal cells. Intermediate filaments are relatively tough, ropelike structures that provide mechanical stability to cells and tissues. The three types of filaments are connected to one another, and their functions are coordinated.

Introduction

Intermediate filaments are tough and durable protein fibers found in the cytoplasm of most, but not all, animal cells. They are called "intermediate" because in electron micrographs their apparent diameter (8-10 nm) is between that of the thin actin filaments and the thick myosin filaments of muscle cells, where they were first described (they are also intermediate in diameter between actin filaments and microtubules). In most animal cells an extensive network of intermediate filaments surrounds the nucleus and extends out to the cell periphery, where they interact with the plasma membrane ( Figure16-12). In addition, a tightly woven basketwork of intermediate filaments - the nuclear lamina - underlies the nuclear envelope.

Intermediate filaments are particularly prominent in the cytoplasm of cells that are subject to mechanical stress. They are present in large numbers, for example, in epithelia, where they are linked from cell to cell at specialized junctions, along the length of nerve cell axons, and in all kinds of muscle cells. When cells are treated with concentrated salt solutions and nonionic detergents, the intermediate filaments remain behind while most of the rest of the cytoskeleton is lost. In fact, the term "cytoskeleton" was originally coined to describe this unusually stable and insoluble fiber system.

Intermediate Filaments Are Polymers of Fibrous Proteins ⁸

Unlike actin and tubulin, which are globular proteins, the many types of intermediate filament protein monomers are all highly elongated fibrous molecules that have an amino-terminal head, a carboxyl-terminal tail, and a central rod domain( Figure16-13). The central rod domain consists of an extended α-helical region containing long tandem repeats of a distinctive amino acid sequence motif called the heptad repeat.As discussed in Chapter 3, this seven amino acid motif promotes the formation of coiled-coil dimers between two parallel α helices (see Figure 3-48). Long stretches of heptad repeats are also found in many other elongated cytoskeletal proteins with coiled-coil dimeric structures, including tropomyosin and the tail of myosin, which we discuss later.

In the next stage of assembly, two of the coiled-coil dimers associate in an antiparallel manner to form a tetrameric subunit ( Figure16-14). Soluble tetramers are found in small amounts in cells, suggesting that they are the fundamental subunit from which intermediate filaments assemble. The antiparallel arrangement of dimers implies that the tetramer, and hence the intermediate filament that it forms, is a nonpolarized structure - that is, it is the same at both ends and symmetrical along its length. This distinguishes intermediate filaments from microtubules and actin filaments, which are polarized and whose functions depend on this polarity. The final stages of intermediate filament assembly are less well characterized, but it seems that tetramers add to an elongating intermediate filament in a simple binding reaction in which they align along the axis of the filament and pack together in a helical pattern (see Figure16-14).

The central rod domain, which is structurally similar in all intermediate filament proteins, mediates the lateral interactions that form the assembled filament. The globular head and tail domains, by contrast, can vary greatly in both size and amino acid sequence without affecting the basic axial structure of the filament; they often project from the surface of the filament and mediate its interactions with other components. This structural design means that intermediate filaments can be made from proteins of a surprisingly wide range of sizes (from about 40,000 to about 200,000 daltons).

In most cells, almost all intermediate filament protein molecules are in the fully polymerized state, with very little free tetramer. Nonetheless, a cell can regulate the assembly of its intermediate filaments and determine their number, length, and position. One mechanism of control involves the phosphorylation of specific serine residues in the amino-terminal head domain of intermediate filament proteins. In the most dramatic example, phosphorylation of the protein subunits that form the nuclear lamina causes them to disassemble completely at mitosis; when mitosis finishes, the specific serines are dephosphorylated and the nuclear lamina re-forms (see Figure12-18). Cytoplasmic intermediate filaments can also undergo a radical reorganization during mitosis, as well as in response to some extracellular signals. Although these changes are usually accompanied by an increase in subunit phosphorylation, other factors may also help mediate them.

Epithelial Cells Contain a Highly Diverse Family of Keratin Filaments ⁹

The cytoplasmic intermediate filaments in vertebrate cells can be grouped into three classes: (1) keratin filaments, (2) vimentin and vimentin-related filaments, and (3) neurofilaments, each formed by polymerization of their corresponding subunit proteins ( Table 16-1). By far the most diverse family of these subunits is the keratins (also called cytokeratins), which form keratin filaments, primarily in epithelial cells. There are over 20 distinct keratins in human epithelia. At least 8 more keratins, called hard keratins, are specific to hair and nails. (The keratins of epithelial cells, hair, and nails are sometimes referred to as α-keratins to distinguish them from the evolutionarily distinct β-keratins found in bird feathers, which have an entirely different structure and are not discussed in this chapter.)

Based on their amino acid sequence, the keratins can be subdivided into two types: the type I (acidic) keratins and the type II (neutral/basic) keratins. In reassembly experiments it is found that heterodimers of type I and type II keratins can form intermediate filaments but homodimers cannot, which explains why keratin filaments are always heteropolymers formed from equal numbers of type I and type II keratin polypeptides.

A single epithelial cell can make a variety of keratins, all of which copolymerize into a single keratin filament system. The simplest epithelia, such as those found in early embryos and in some adult tissues such as the liver, contain only a single type I and a single type II keratin. Epithelia in other locations, such as the tongue, bladder, and sweat glands, contain six or more keratins - the particular blend depending on the cell's location in the organ. The diversity is most pronounced in skin, where distinct sets of keratins are expressed by the cells in the different layers of the epidermis (see Figure22-19). There are also keratins characteristic of actively proliferating epithelial cells. This heterogeneity of keratins is clinically useful: in the diagnosis of epithelial cancers ( carcinomas), the particular set of keratins expressed can be used to determine the epithelial tissue in which the tumor originated and thus help to decide the type of treatment that is likely to be most effective.

Many Nonepithelial Cells Contain Their Own Distinctive Cytoplasmic Intermediate Filaments ¹⁰

Unlike keratins, vimentin and the vimentin-related proteins can form intermediate filaments that are polymers of a single protein species. Vimentinitself is the most widely distributed of the cytoplasmic intermediate filament proteins, being present in many cells of mesodermal origin, including fibroblasts, endothelial cells, and white blood cells; in addition, many cells express it transiently during development. Desminis found mainly in muscle cells: it is distributed throughout the cytoplasm of smooth muscle cells, and it links together adjacent myofibrils (ordered bundles of filamentous actin and myosin, discussed later) in skeletal and heart muscle cells. Glial fibrillary acidic proteinforms glial filaments in astrocytes in the central nervous system and in some Schwann cells in peripheral nerves ( Figure 16-15). All of these proteins co-polymerize readily with one another, and co-polymers of vimentin and a vimentin-related protein are found in a number of adult cell types. By contrast, none of these proteins co-polymerize with keratins: when keratins and vimentin-related proteins are expressed in the same cell, they form separate filament systems.

Nerve cells contain a variety of unique intermediate filaments, which are expressed in different regions of the nervous system or at specific stages of development. By far the most abundant are the neurofilaments, which extend along the length of an axon and form its primary cytoskeletal component, especially in mature nerve cells. In mammals, three neurofilament proteinshave long been recognized: termed NF-L, NF-M, and NF-H, for low, middle, and high molecular weight, respectively, all three are usually found in each neurofilament. NF-M and NF-H have especially long carboxyl-terminal tails, which are thought to project from the neurofilament axis and contribute to the regular side-to-side spacing of neuro-filaments in an axon ( Figure16-16).

If a cell in culture is stained with an antibody to a cytoplasmic intermediate filament protein, a delicate network of threadlike filaments is usually seen surrounding the nucleus and extending through the cytoplasm to the plasma membrane (see Figure 16-12). In epithelial cells, keratin filaments are attached to specialized cell junctions - both to desmosomes, which bond neighboring cells together, and to hemidesmosomes, which anchor cells to the underlying basal lamina (discussed in Chapter 19). Because the keratin filaments in each cell are connected via desmosomes to those of its neighbors, they form a continuous network throughout the entire epithelium ( Figure16-17). Similarly, desmin filaments are often anchored to specialized cell junctions in muscle cells.

The Nuclear Lamina Is Constructed from a Special Class of Intermediate Filament Proteins - the Lamins ¹¹

The nuclear lamina is a meshwork of intermediate filaments that lines the inside surface of the inner nuclear membrane in eucaryotic cells ( Figure16-18). It is typically 10-20 nm thick and is interrupted in the region of nuclear pores to provide a passageway for macromolecules entering and leaving the nucleus. In mammalian cells the nuclear lamina is composed of lamins, which are homologous to other intermediate filament proteins but differ from them in at least four ways: (1) Their central rod domain is somewhat longer (see Figure16-13). (2) They contain a nuclear transport signal that directs them from the cytosol, where they are made, into the nucleus. (3) They assemble into a two-dimensional, sheetlike lattice, which is thought to require their association with other proteins. (4) The meshwork they form is unusually dynamic and rapidly disassembles at the start of mitosis and reassembles at the end of mitosis; as already mentioned, the disassembly and reassembly are mediated by the phosphorylation and dephosphorylation of several serine residues on the lamins.

Unlike microtubules and actin filaments, which are a defining characteristic of eucaryotic cells, cytoplasmic intermediate filaments have been described only in multicellular animals, and even in these organisms they are not required in every cell type. The specialized glial cells that make myelin in the vertebrate central nervous system, for example, do not contain intermediate filaments. Moreover, intermediate filaments can be disrupted in muscle cells, fibroblasts, and epithelial cells in culture without detectable effects on cell behavior.

It seems likely that the first type of intermediate filament protein to appear in evolution was a nuclear lamin and that the various kinds of cytoplasmic intermediate filaments are later adaptations of this primitive form. The intermediate filament proteins in invertebrates, for example, more closely resemble lamins than vertebrate cytoplasmic intermediate filament proteins.

Intermediate Filaments Provide Mechanical Stability to Animal Cells ¹²

There is increasing evidence that a major function of cytoplasmic intermediate filaments is to resist mechanical stress. In the human genetic disease epidermolysis bullosa simplex, mutations in keratin genes that are normally expressed in the basal cell layer of the epidermis disrupt the keratin filament network in these cells, making them very sensitive to mechanical injury: a gentle squeeze can cause the mutant basal cells to rupture, and the skin in affected individuals is blistered. A similar condition can be produced in transgenic mice that express mutant keratins of this type ( Figure16-19). In both humans and mice the epidermis can be so weakened that individuals carrying the mutation can die from mechanical trauma. Cytoplasmic intermediate filaments are thought to strengthen nonepithelial cells in a similar way.

The structure of intermediate filaments is ideally suited for such a mechanical function. Because the fibrous subunits associate side by side in overlapping arrays, the filaments can withstand very much larger stretching forces than microtubules or actin filaments ( Figure 16-20). In the skin, keratin filaments in the outermost layers of the epidermis become covalently cross-linked to one another and to associated proteins, and as the cells die, the cross-linked keratins persist as a major part of the protective outer layer of the animal. Specialized epithelial cells at particular locations in the skin provide regional variation by generating surface appendages rich in keratin, such as hairs and nails.

But if intermediate filaments function simply to provide tensile strength to cells and tissues, why are there so many different types? And what is the function of the head and tail domains of the proteins, which show such large variations in sequence? Detailed answers to these questions cannot be given at present, but it is clear that the way that intermediate filaments are linked to other cellular components varies greatly among cell types. The desmin filaments that tie the edges of the myofibrils together in skeletal muscle cells are likely to have binding sites for specific myofibril-associated proteins. Neurofilaments in axons are linked side by side by their carboxyl-terminal tail domains to provide a continuous rope of filaments that can be a meter or more in length. Some keratins are specialized to form the tough, protective outer layer of the skin, while others specifically strengthen epithelia undergoing shape changes during morphogenesis. These different functional requirements must be accommodated by the variable regions of the different intermediate filament proteins, which project from the surface of the intermediate filaments and determine their ability to associate with one another and with other components in the cell. In a sense, therefore, the variable regions of intermediate filament proteins serve functions similar to those of the accessory proteins of actin filaments and microtubules. The difference is that the variable regions are an integral part of the intermediate filament subunit, rather than being a separate protein.

Summary

Intermediate filaments are strong, ropelike polymers of fibrous polypeptides that resist stretch and play a structural or tension-bearing role in the cell. A variety of tissue-specific forms are known that differ in the type of polypeptide they contain: these include the keratin filaments of epithelial cells, the neurofilaments of nerve cells, the glial filaments of astrocytes and Schwann cells, the desmin filaments of muscle cells, and the vimentin filaments of fibroblasts and many other cell types. Nuclear lamins, which form the fibrous lamina that underlies the nuclear envelope, are a separate family of intermediate filament proteins.

The monomers of the different types of intermediate filaments differ in amino acid sequence and have very different molecular weights. But they all contain a homologous central rod domain that forms an extended coiled-coil structure when the protein dimerizes. Two coiled-coil dimers associate with each other to form a symmetrical tetramer, which in turn assembles in large overlapping arrays to form the nonpolarized intermediate filament. The rod domains of the subunits form the structural core of the intermediate filament, whereas the domains at either end can project outward. One function of the variable terminal domains may be to allow each type of filament to associate with specific other components in the cell, so as to position the filaments appropriately for a particular cell type.

Introduction

Microtubules, as we have seen, are long, stiff polymers that extend throughout the cytoplasm and govern the location of membrane-bounded organelles and other cell components. In this section we discuss the assembly of these remarkable structures from tubulin molecules and explain how their polymerization and depolymerization are controlled by the nucleotide GTP. We then examine some ways in which selected microtubules are stabilized in the cell by their association with specific accessory proteins. Finally, we discuss the importance of microtubule-dependent motors that transport membrane vesicles and various protein complexes along microtubules.

Microtubules Are Hollow Tubes Formed from Tubulin ¹⁴

Microtubules are formed from molecules of tubulin, each of which is a heterodimer consisting of two closely related and tightly linked globular polypeptides called α-tubulin and β-tubulin. Although tubulin is present in virtually all eucaryotic cells, the most abundant source for biochemical studies is the vertebrate brain. Extraction procedures yield 10 to 20% of the total soluble protein in brain as tubulin, reflecting the unusually high density of microtubules in the elongated processes of nerve cells.

Tubulin molecules themselves are diverse. In mammals there are at least six forms of α-tubulin and a similar number of forms of β-tubulin, each encoded by a different gene. The different forms of tubulin are very similar, and they will generally co-polymerize into mixed microtubules in the test tube, although they can have distinct locations in the cell and perform subtly different functions. The microtubules in six specialized touch-sensitive neurons in the nematode Caenorhabditis elegans, for example, contain a specific form of β-tubulin, and mutations in the gene for this protein result in the specific loss of touch-sensitivity with no apparent defect in other cell functions.

A microtubule can be regarded as a cylindrical structure in which the tubulin heterodimers are packed around a central core, which appears empty in electron micrographs. More accurately, perhaps, one can view the structure as being built from 13 linear protofilaments, each composed of alternating α- and β-tubulin subunits and bundled in parallel to form a cylinder ( Figure 16-21). Since the 13 protofilaments are aligned in parallel with the same polarity, the microtubule itself is a polar structure, and it is possible to distinguish a plus (fast-growing) and a minus (slow-growing) end.

Microtubules Are Highly Labile Structures That Are Sensitive to Specific Antimitotic Drugs ¹⁵

Many of the microtubule arrays in cells are labile and depend on this lability for their function. One of the most striking examples is the mitotic spindle, which forms after the cytoplasmic microtubules disassemble at the onset of mitosis. The mitotic spindle is the target of a variety of specific antimitotic drugs that act by interfering with the exchange of tubulin subunits between the microtubules and the free tubulin pool. One of these is colchicine( Figure 16-22), an alkaloid extracted from the meadow saffron that has been used medicinally in the treatment of gout since ancient Egyptian times. Each molecule of colchicine binds tightly to one tubulin molecule and prevents its polymerization, but it cannot bind to tubulin once the tubulin has polymerized into a microtubule. The exposure of a dividing cell to colchicine, or to the closely related drug colcemid, causes the rapid disappearance of the mitotic spindle, indicating that a chemical equilibrium is maintained through continual exchange of subunits between the spindle microtubules and the pool of free tubulin. Because the temporary disruption of spindle microtubules preferentially kills many abnormally dividing cells, antimitotic drugs, such as vinblastine and vincristine (whose effects are similar to those of colcemid), are widely used in the treatment of cancer.

The drug taxol ( Figure16-22), extracted from the bark of yew trees, has the opposite effect. It binds tightly to microtubules and stabilizes them, and when added to cells, it causes much of the free tubulin to assemble into microtubules. The stabilization of microtubules by taxol arrests dividing cells in mitosis, indicating that microtubules must be able not only to polymerize but also to depolymerize during mitosis. Taxol is also widely used as an anticancer drug.

Elongation of a Microtubule Is Rapid, Whereas the Nucleation of a New Microtubule Is Slow ¹⁶

Microtubule polymerization and depolymerization are complex and interesting processes with important biological roles. Most of what we know about the dynamic behavior of microtubules has come from studying the polymerization of purified tubulin molecules in vitro. Pure tubulin will polymerize into microtubules at 37°C in a test tube as long as Mg²⁺ and GTP are present. If the polymerization is followed either by light-scattering measurements or by microscopy, it shows an initial lag phase, after which microtubules form rapidly until a plateau level of polymerization is reached. The lag phase occurs because it is much easier to add subunits to an existing microtubule, a process called elongation, than to start a new microtubule de novo, a process called nucleation.

During the rapid polymerization phase, the high concentration of free tubulin causes microtubules to polymerize faster than they depolymerize (see below). When the plateau of polymerization is reached, however, not all of the tubulin will have polymerized because subunits are dissociating (depolymerizing) from the ends of microtubules as well as adding to them. The rate of polymerization drops with time because this rate is proportional to the concentration of free tubulin; the final concentration of free tubulin at the plateau, where the polymerization and depolymerization rates are exactly balanced, is called the critical concentration ( Figure 16-23).

We saw at the beginning of the chapter that the microtubules in a cell usually grow from a specific nucleating site (in most cases, the centrosome); because of a kinetic barrier to nucleation in solution, tubulin polymerization occurs only at this site. As in the test tube, not all the tubulin in the cell becomes polymerized. A typical fibroblast cell contains approximately 20 micromolar tubulin (2mg/ml), of which 50% is in microtubules and 50% is free.

The Two Ends of a Microtubule Are Different and Grow at Different Rates ¹⁷

The structural polarity of a microtubule, which reflects the regular orientation of its tubulin subunits, makes the two ends of the polymer different in ways that have a profound effect on its rate of growth. If purified tubulin molecules are allowed to polymerize for a short time at the ends of fragments of stable microtubules and the mixture is then examined in the electron microscope, one end can be seen to elongate at three times the rate of the other ( Figure16-24). The fast-growing end is thereby defined as the plus end and the other as the minus end.

It is possible to detect the polarity of microtubules in cross-section by adding free tubulin molecules to existing microtubules: under special conditions the tubulin monomers, instead of adding to the ends of the microtubules, add to the sides, forming curved protofilament sheets. In cross-section the sheets resemble hooks and, depending on the orientation of the microtubule, will appear to point either clockwise or counterclockwise ( Figure16-25). In this way it has been shown that the plus ends of the microtubules in a cell extend away from microtubule-nucleating sites such as the centrosome, the poles of a mitotic spindle, or the basal body of a cilium ( Figure 16-26).

Centrosomes Are the Primary Site of Nucleation of Microtubules in Animal Cells ¹⁸

The microtubules in the cytoplasm of an interphase cell in culture can be visualized by staining the cell with fluorescent anti-tubulin antibodies after the cells have been fixed. The microtubules are seen in greatest density around the nucleus and radiate out into the cell periphery in fine lacelike threads ( Figure16-27). The origin of the microtubules is seen most clearly if they are first depolymerized with colcemid and then allowed to repolymerize after the drug is washed out. The new microtubules grow out from the centrosome to form a small starlike structure called an aster and then elongate toward the cell periphery until the original microtubule distribution is reestablished ( Figure16-28). If the microtubules in cultured cells are decorated with tubulin hooks to determine their polarity, they are all seen to have their plus ends facing away from the centrosome, indicating that this organizing center has the capacity to nucleate microtubule polymerization with a specific polarity.

The centrosome is the major microtubule-organizing center in almost all animal cells. In interphase it is typically located to one side of the nucleus, close to the outer surface of the nuclear envelope. Embedded in the centrosome is a pair of cylindrical structures arranged at right angles to each other in an L-shaped configuration. These are centrioles, and we discuss their structure later. The centrosome duplicates and splits into two equal parts during interphase, each half containing a duplicated centriole pair. These two daughter centrosomes move to opposite sides of the nucleus when mitosis begins, and they form the two poles of the mitotic spindle (see Figure 18-5).

Surrounding each centriole pair, in both interphase and metaphase, is a region of the cytoplasm that stains darkly when viewed by electron microscopy and appears in the best micrographs to be made of a network of small fibers ( Figure 16-29A). This is the pericentriolar material, or centrosome matrix, and it is the part of the centrosome that nucleates microtubule polymerization. The protein composition of the centrosome matrix is only partly known, as is the mechanism by which it nucleates microtubules. However, it contains a number of centrosome-specific proteins, including a special minor form of tubulin, called γ-tubulin ( Figure 16-29B), which may interact with the normal α/β tubulin dimer to help nucleate microtubules.

Not all microtubule-organizing centers contain centrioles. In mitotic cells of higher plants, for example, the microtubules terminate in poorly defined regions of electron density that are completely devoid of centrioles. Similarly, centrioles are not present in the meiotic spindle of mouse oocytes, although they appear later in the developing embryo. In fungi and diatoms the microtubule-organizing center is a plaque called the spindle pole body, which is embedded in the nuclear envelope. Despite these morphological differences ( Figure16-30), all of the organizing centers contain a matrix that nucleates microtubule polymerization, and they usually contain gamma-tubulin and other centrosome-specific proteins. Thus the molecular mechanism of microtubule nucleation is likely to be highly conserved.

Microtubules Depolymerize and Repolymerize Continually in Animal Cells ¹⁹

In a cell such as a cultured fibroblast the entire microtubule array is turning over rapidly. The half-life of an individual microtubule is about 10 minutes, while the average lifetime of a tubulin molecule, between its synthesis and proteolytic degradation, is more than 20 hours. Thus each tubulin molecule will participate in the formation and dismantling of many microtubules in its lifetime, a process that can be investigated by direct observation of living cells. One way is to inject tubulin that has been covalently linked to a fluorescent dye and then follow the behavior of microtubules that incorporate the tagged tubulin using fluorescence microscopy. Alternatively, in certain very flat cells one can visualize microtubules directly, without labeling them, using video-enhanced differential-interference-contrast microscopy (see Figure4-12). When microtubules in a cell are watched over time by either method, a remarkable phenomenon is observed. Individual microtubules grow toward the cell periphery at a constant rate for some period and then suddenly shrink rapidly back toward the centrosome. They may shrink partially and then recommence growing, or they may disappear completely, to be replaced by a different microtubule ( Figure16-31). These fluctuations in length occur over many micrometers and involve the polymerization and then depolymerization of tens of thousands of tubulin subunits. Transitions between prolonged periods of polymerization and depolymerization are also seen when pure microtubules are studied in a test tube ( Figure16-32). This behavior, called dynamic instability, plays a major role in positioning microtubules in the cell, as we discuss below.

GTP Hydrolysis Can Explain the Dynamic Instability of Individual Microtubules ²⁰

The dynamic instability of microtubules requires an input of energy to shift the chemical balance between polymerization and depolymerization - energy that comes from the hydrolysis of GTP. GTP binds to the β-tubulin subunit of the heterodimeric tubulin molecule, and when a tubulin molecule adds to the end of a microtubule, this GTP molecule is hydrolyzed to GDP. (The α-tubulin subunit also carries GTP, but this cannot be exchanged for free GTP and is not hydrolyzed, so we can consider it a fixed part of the tubulin protein structure.)

The role of GTP hydrolysis in microtubule polymerization has been examined using analogues of GTP that cannot be hydrolyzed. Tubulin molecules containing such nonhydrolyzable GTP analogues form microtubules normally, indicating that, while the binding of this nucleotide is required for microtubule polymerization, its hydrolysis is not. These microtubules, however, are abnormally stable and do not depolymerize like normal microtubules when the tubulin concentration in the surrounding fluid is lowered or when they are treated with colchicine. Thus the normal role of GTP hydrolysis is apparently to allow microtubules to depolymerize by weakening the bonds between tubulin subunits in the microtubule.

Dynamic instability is thought to be a consequence of the delayed hydrolysis of GTP after tubulin assembly. When a microtubule grows rapidly, tubulin molecules add to a polymer end faster than the GTP they carry can be hydrolyzed. This results in the presence of a GTP capon the end of the microtubule, and because tubulin molecules carrying GTP bind to one another with higher affinity than tubulin molecules carrying GDP, the GTP cap will encourage a growing microtubule to continue growing. Conversely, once a microtubule has lost its GTP cap - for example, if the instantaneous rate of polymerization slows down - it will start to shrink and then tend to go on shrinking.

A model for the structural changes that accompany dynamic instability is shown schematically in Figure 16-33. Some general principles that apply to the polymerization of both actin filaments and microtubules are discussed in Panel 16-1, pages 824-825.

Cells can modify the dynamic instability of their microtubules for specific purposes. In each M phase of the cell cycle, for example, the rapidity with which microtubules form and break down is greatly increased, so that the chromosomes can readily capture growing microtubules and a mitotic spindle can rapidly assemble (discussed in Chapter 18). Conversely, when a cell differentiates and takes on a defined morphology, the dynamic instability of its microtubules is often suppressed by proteins that bind to the microtubules and stabilize them against depolymerization. The ability to stabilize microtubules in a particular configuration provides an important mechanism by which a cell can organize its cytoplasm.

The Dynamic Instability of Microtubules Provides an Organizing Principle for Cell Morphogenesis ²¹

Cytoplasmic microtubules in animal cells tend to radiate out in all directions from the centrosome, where their minus ends are anchored. Most animal cells are polarized, however, and the assembly and disassembly of tubulin molecules are spatially controlled so that microtubules extending toward specific regions of the cell predominate. It is not known for certain how this is achieved, but it seems likely that the mechanisms depend on the dynamic instability of microtubules.

We have seen that individual microtubules in vitro tend to exist in one of two states - steady growth or rapid, "catastrophic" disassembly - and that microtubules in a cell can also exist in these two states. The inherent instability of microtubules helps to explain how they can become organized in specific directions in a cell - toward the leading edge of a crawling cell, for example. The array of microtubules radiating from the centrosome is continually changing as new microtubules grow and replace others that have depolymerized. A microtubule that grows from a centrosome can be stabilized if its plus end is somehow stabilized, or capped, so as to prevent its depolymerization. If capped by a structure in a particular region of the cell, it will establish a relatively stable link between that structure and the centrosome. Microtubules originating in the centrosome can thus be selectively stabilized by events elsewhere in the cell. Cell polarity is thought to be determined in this way by unknown structures or factors localized in particular regions of the cell cortex that "capture" the plus ends of microtubules ( Figure 16-34).

In many cells the initial stabilization of microtubules at their plus ends is consolidated to produce a more permanent polarization of the cell, as we now discuss.

Microtubules Undergo a Slow "Maturation" Revealed by Posttranslational Modifications of Their Tubulin ²²

Tubulin subunits can be covalently modified after they polymerize. Two such modifications are especially interesting in that they provide a form of molecular clock, which can be used to tell how long it has been since a given microtubule polymerized. These modifications are the acetylation of α-tubulin on a particular lysine and the removal of the tyrosine residue from the carboxyl terminus of α-tubulin. Acetylation and detyrosination are both relatively slow enzymatic reactions that occur only on microtubules and not on free tubulin molecules; moreover, they are rapidly reversed as soon as a tubulin molecule depolymerizes. Thus the longer the time that has elapsed since a particular microtubule polymerized, the higher will be the fraction of its subunits that are acetylated and detyrosinated. Complete modification takes several hours, so that in fibroblasts, where microtubules turn over rapidly, relatively few of them are modified. In nerve axons, by contrast, the majority of microtubules are stable and most are modified.

Acetylation and detyrosination can be detected by specific antibodies, and they provide a useful indication of the stability of microtubules in cells in which it is difficult to study microtubule dynamics directly. The role of these modifications is unknown, but it is thought that they provide sites for the binding of specific microtubule-associated proteins that further stabilize mature microtubules.

Microtubule-associated Proteins (MAPs) Bind to Microtubules and Modify Their Properties ²³

Whereas the posttranslational modification of tubulin marks certain microtubules as "mature" and may promote their stability, the most far-reaching and versatile modifications of microtubules are those conferred by the binding of other proteins. These microtubule-associated proteins, or MAPs, serve both to stabilize microtubules against disassembly and to mediate their interaction with other cell components. As one might expect from the diverse functions of microtubules, there are many kinds of MAPs; some are widely distributed in most cells, whereas others are found only in specific cell types.

Two major classes of MAPs can be isolated from brain in association with microtubules: HMW proteins (high-molecular-weight proteins), which have molecular weights of 200,000 to 300,000 or more and include MAP-1 and MAP-2; and tau proteins, which have molecular weights of 55,000 to 62,000. Proteins in both classes have two domains, only one of which binds to microtubules; the other is thought to help link the microtubule to other cell components ( Figure 16-35). Because the microtubule-binding domain binds to several unpolymerized tubulin molecules simultaneously, these MAPs speed up the nucleation step of tubulin polymerization in vitro. More important, they inhibit the dissociation of tubulin from the microtubule ends and thus stabilize the microtubules once they have formed. Staining with antibodies to MAP-2 and tau shows that both proteins bind along the entire length of cytoplasmic microtubules.

Many other MAPs have been isolated. Some act as structural components and provide permanent links to other cell components, including other parts of the cytoskeleton. Others are microtubule motors, which use the energy of ATP hydrolysis to move along microtubules, as we discuss below.

MAPs Help Create Functionally Differentiated Cytoplasm ²⁴

Many cell types specifically stabilize microtubules in specialized regions of cytoplasm. An especially well-studied example is provided by nerve cells, which extend two kinds of processes axons and dendrites. Axons, which are uniform in diameter and can be many centimeters long, are responsible for propagating electrical signals away from the cell body, whereas dendrites, which taper away from the cell body and rarely exceed 500 µm in length, are responsible for receiving electrical information from other neurons and relaying it to the cell body. Most nerve cells form several dendrites but only a single axon (see Figure 11-20).

Axons and dendrites are both packed with microtubules, although with different arrangements. In axons microtubules are very long and are all oriented with their plus ends away from the cell body. In dendrites the microtubules are shorter and their polarity is mixed: some have their plus ends pointing away from the cell body, while others have their plus ends pointing toward the cell body. When the distribution of MAPs in cultured neurons is studied with specific antibodies, certain forms of the tau protein are found to be present only in axons; MAP-2, on the other hand, is present in both dendrites and the cell body but completely excluded from axons ( Figure 16-36). Axons and dendrites are different in many other ways as well: mRNAs, ribosomes, and some kinds of ion channels, for example, are present in dendrites and the cell body but are excluded from axons, while certain cell-adhesion molecules and the Na⁺ channels involved in the generation of action potentials are selectively localized to axons. Thus both the cytoplasm and the plasma membrane of a nerve cell are divided into axonal and dendritic compartments. These compartments within a single cell differ from membrane-bounded compartments such as the endoplasmic reticulum or mitochondria, since they are not separated from each other by a membrane; instead, the difference seems to be one of structural organization and the types of proteins present.

The generation of axons and dendrites during the differentiation of nerve cells is discussed in Chapter 21. Although it is unclear how the cytoplasm and plasma membrane of a nerve cell become compartmentalized, MAPs may be essential for this process. When the production of tau protein is inhibited in cultured neurons by treatment with specific antisense oligonucleotides, the formation of axons is suppressed, whereas the formation of dendrites is unaffected. Conversely, when nonneuronal cells are genetically manipulated so that they express tau protein (which is normally expressed only in nerve cells), they form long axonlike processes, which contain bundles of microtubules arranged with their plus ends pointing away from the cell body, just as in nerve cells.

Because different components of the cell move along microtubules in different directions, one can postulate that an initial difference in microtubule polarity is created by a different distribution of MAPs, which will in turn lead to further differences between dendrites and axons. Secretory vesicles, for example, move toward the plus end of microtubules and therefore will be carried down the axon to the nerve terminals where they function; conversely, if ribosomes and mRNAs move toward the minus end of microtubules, they could be excluded from axons.

Kinesin and Dynein Direct Organelle Movement Along Microtubules ²⁵

Important advances in cell biology have often followed the introduction of a new experimental technique, and it was the improved ability to see small faint objects by video-enhanced light microscopy that led to the discovery of the microtubule motors responsible for organelle transport. Once it became possible to visualize single microtubules in an unfixed specimen, investigators could follow the movement of organelles and other particles along these microtubules in vitro. Alternatively, they could observe and measure the gliding movement of individual microtubules over glass surfaces coated with cell extracts.

Such in vitro motility assays were used to identify and isolate two classes of microtubule-dependent motor proteins - the kinesins and the cytoplasmic dyneins. Cytoplasmic dyneins are involved in organelle transport and mitosis and are closely related to ciliary dynein,the motor protein in cilia and flagella (discussed later). Kinesins are more diverse than the dyneins, and different family members are involved in organelle transport, in mitosis, in meiosis, and in the transport of synaptic vesicles along axons. Both the cytoplasmic dyneins and the kinesins are composed of two heavy chains plus several light chains. Each heavy chain contains a conserved, globular, ATP-binding head and a tail composed of a string of rodlike domains. The two head domains are ATPase motors that bind to microtubules, while the tails generally bind to specific cell components and thereby specify the type of cargo that the protein transports ( Figure16-37).

The Rate and Direction of Movement Along a Microtubule Are Specified by the Head Domain of Motor Proteins ²⁶

Most known motor proteins move in only one direction along microtubules - either toward the plus end or toward the minus end. This directionality can be analyzed in vitro by allowing polystyrene beads coated with the motor protein to move along microtubules that have been polymerized on centrosomes. Because the microtubules in such arrays have their plus ends outermost, the direction of movement can be readily determined with a light microscope. Whereas polystyrene beads coated with crude extracts of cytoplasm move in both directions, beads coated with kinesin isolated from axons move only outward toward the plus end of the microtubules. Beads coated with cytoplasmic dyneins, by contrast, move toward the minus ends of the microtubules, which are embedded in the centrosome.

Studies of intact nerve axons have confirmed the results obtained in in vitro experiments: organelle movement away from the cell body is driven mainly by kinesin, whereas organelle movement back from the nerve terminal toward the cell body is driven by cytoplasmic dynein ( Figure16-38). Since all proteins are made in the nerve cell body, cytoplasmic dynein must be carried first in a nonfunctional state to the nerve terminal before it can begin to work to transport organelles back to the cell body.

Surprisingly, not all kinesins move organelles toward the plus end of microtubules. A Drosophila kinesin called Ncd, for example, which is required for normal meiosis, differs from axonal kinesin in both the direction and the rate at which it moves along microtubules: whereas axonal kinesin walks toward the plus end at approximately 2 µm/second, the Ncd protein walks toward the minus end at about 0.1 µm/second.

The mechanism by which these motor proteins convert the energy of ATP hydrolysis into vectorial movement is not known. Finding out how two closely related head domains can move in opposite directions along a microtubule will require detailed structural studies and is likely to illuminate the energy transduction process itself.

Summary

Microtubules are stiff polymers of tubulin molecules. They assemble by addition of GTP-containing tubulin molecules to the free end of the microtubule, with one end (the plus end) growing faster than the other. Hydrolysis of the bound GTP takes place after assembly and weakens the bonds that hold the microtubule together. Slowly growing microtubules are especially unstable and liable to catastrophic disassembly, but they can be stabilized in cells by association with other structures that cap their two ends. Microtubule-organizing centers such as centrosomes protect the minus ends of microtubules and continually nucleate the formation of new microtubules, which grow out in random directions. Any microtubule that happens to encounter a structure that stabilizes its free plus end will be selectively retained, while other microtubules will depolymerize. It is thought that this selective process largely determines the position of the microtubule arrays in a cell.

The tubulin subunits in microtubules that have been selectively stabilized are modified by acetylation and detyrosination. These alterations are thought to label the microtubule as "mature" and provide sites for the binding of specific microtubule-associated proteins (MAPs), which further stabilize the microtubule against disassembly. Microtubule motor proteins constitute an important class of MAPs that use the energy of ATP hydrolysis to move unidirectionally along a microtubule, carrying specific cargo. In general, dyneins move cargo toward the minus ends of microtubules, while most kinesins move cargo toward the plus ends. Such motor proteins are largely responsible for the spatial organization and directed movements of organelles in the cytoplasm.

Introduction

Ciliary beating is an extensively studied form of cellular movement. Cilia are tiny hairlike appendages about 0.25 µm in diameter with a bundle of microtubules at their core; they extend from the surface of many kinds of cells and are found in most animal species, many protozoa, and some lower plants. The primary function of cilia is to move fluid over the surface of the cell or to propel single cells through a fluid. Protozoa, for example, use cilia both to collect food particles and for locomotion. On the epithelial cells lining the human respiratory tract, huge numbers of cilia (10⁹/cm² or more) sweep layers of mucus, together with trapped particles of dust and dead cells, up toward the mouth, where they are swallowed and eliminated. Cilia also help to sweep eggs along the oviduct, and a related structure, the flagellum, propels sperm.

Cilia Move by the Bending of an Axoneme - a Complex Bundle of Microtubules ²⁷

Fields of cilia bend in coordinated unidirectional waves ( Figure 16-39). Each cilium moves with a whiplike motion: a forward active stroke, in which the cilium is fully extended and beating against the surrounding liquid, is followed by a recovery phase, in which the cilium returns to its original position with an unrolling movement that minimizes viscous drag ( Figure16-40A). The cycles of adjacent cilia are almost but not quite in synchrony, creating the wavelike patterns that can be seen in fields of beating cilia under the microscope.

The simple flagella of sperm and of many protozoa are much like cilia in their internal structure, but they are usually very much longer. Instead of making whiplike movements, they propagate quasi-sinusoidal waves ( Figure16-40B). Nevertheless, the molecular basis for their movement is the same as that in cilia. It should be noted that the flagella of bacteria (described in Chapter 15) are completely different from the cilia and flagella of eucaryotic cells.

The movement of a cilium or a flagellum is produced by the bending of its core, which is called the axoneme. The axoneme is composed entirely of microtubules and their associated proteins. The microtubules are modified and arranged in a pattern whose curious and distinctive appearance was one of the most striking revelations of early electron microscopy: nine special doublet microtubules are arranged in a ring around a pair of single microtubules ( Figure 16-41). This "9 + 2" array is characteristic of almost all forms of cilia and eucaryotic flagella - from those of protozoa to those found in humans. The microtubules extend continuously for the length of the axoneme, which is usually about 10 µ long but may be as long as 200 µ in some cells.

While each member of the pair of single microtubules (the central pair) is a complete microtubule, each of the outer doublets is composed of one complete and one partial microtubule fused together so that they share a common tubule wall. In transverse sections each complete microtubule appears to be formed from a ring of 13 subunits, while the incomplete tubule of the outer doublet is formed from only 11.

Dynein Drives the Movements of Cilia and Flagella ²⁸

The microtubules of an axoneme are associated with numerous proteins, which project at regular positions along the length of the microtubules. Some serve as cross-links that hold the bundle of microtubules together. Others generate the force that drives the bending motion, while still others form a mechanically activated relay system that controls the motion to produce the desired waveform. The most important of these accessory proteins is ciliary dynein, whose heads interact with adjacent microtubules to generate a sliding force between the microtubules. Because of the multiple links that hold adjacent microtubule doublets together, what would be a sliding movement between free microtubules ( Figure 16-42) is converted to a bending motion in the cilium ( Figure 16-43).

Like cytoplasmic dynein, ciliary dynein has a motor domain, which hydrolyzes ATP to move along a microtubule toward its minus end, and a tail region that carries a cargo, which in this case is an adjacent microtubule. Ciliary dynein is considerably larger than cytoplasmic dynein, both in the size of its heavy chains and in the number and complexity of its polypeptide chains. In flagella of the unicellular green algae Chlamydomonas, for example, the dynein is composed of either 2 or 3 heavy chains (there are multiple forms of dynein in the flagellum) and 10 or more smaller polypeptides ( Figure16-44). Note that the tail of ciliary dynein binds only to the A tubule and not to the B tubule, which has a slightly different structure. The resulting asymmetry in the arrangement of the dynein molecules is required to prevent a fruitless tug-of-war between neighboring microtubules, which presumably explains why each of the nine outer microtubules is an A-B doublet.

Flagella and Cilia Grow from Basal Bodies That Are Closely Related to Centrioles ²⁹

If the two flagella of the green alga Chlamydomonas are sheared from the cell, they rapidly re-form by elongating from structures called basal bodies. The basal bodies have the same structure as the centrioles that are found embedded in the center of animal centrosomes. Indeed, in some organisms, basal bodies and centrioles seem to be functionally interconvertible: during each mitosis in Chlamydomonas, for example, the flagella are resorbed and the basal bodies move into the cell interior and become embedded in the spindle poles.

Centrioles and basal bodies are cylindrical structures about 0.2 µ wide and 0.4 µ long. Nine groups of three microtubules, fused into triplets, form the wall of the centriole, each triplet being tilted inward like the blades of a turbine ( Figure 16-45). Adjacent triplets are linked at intervals along their length, while faint protein spokes can often be seen in electron micrographs to radiate out to each triplet from a central core, forming a pattern like a cartwheel (see Figure 16-45A).

During the formation or regeneration of a cilium, each doublet microtubule of the axoneme grows from two of the microtubules in the triplet microtubules of the basal body so that the ninefold symmetry of the basal body microtubules is preserved in the ciliary axoneme. Autoradiographic evidence suggests that the addition of tubulin and other proteins of the axoneme takes place at the distal tip of the structure, at the plus end of the microtubules. How the central pair of single microtubules forms in the axoneme is not known; there is no central pair in basal bodies or centrioles.

It is not known how the length of flagella and cilia is determined. The length is constant for a given species of cell, and it is not limited by the availability of components or the kinetics of elongation. If one of the two flagella is removed in Chlamydomonas, for example, the remaining flagellum begins to shrink while the lost flagellum simultaneously regenerates. Once the shrinking flagellum and the regrowing flagellum reach the same length, they then both grow out together to reach their final characteristic length. This experiment suggests that flagellar length is constantly monitored in some way ( Figure16-46).

Centrioles Usually Arise by the Duplication of Preexisting Centrioles ³⁰

The otherwise continuous increase in cell mass throughout the animal cell cycle is punctuated by two discrete duplication events: the replication of DNA and the doubling of the centrosome, which usually has a centriole pair at its center. The two centrioles of the pair are positioned at right angles to each other ( Figure16-47). In cultured fibroblasts centriole doubling begins at around the time that DNA synthesis begins: first the two members of a pair separate, and then a daughter centriole is formed perpendicular to each original centriole (see Figure 18-4). An immature centriole contains a ninefold symmetric array of single microtubules; each microtubule then presumably acts as a template for the assembly of the triplet microtubule of mature centrioles.

The two centrioles of a pair are not identical: the daughter centriole not only has a distinct orientation but differs also in detailed morphology and function. In many vertebrate cells, for example, one of the two centrioles is distinguished by its ability to nucleate a so-called primary cilium - an isolated nonmotile cilium that has no known function.

Parent/daughter differences also exist in basal bodies and can lead to asymmetries in the cytoskeleton. In ciliated protozoa, basal body replication is coordinated with cell division and the stereospecificity of the duplication process is thought to be important for maintaining the orientation of cilia on the cell surface. This was clearly demonstrated in a classic experiment performed in the 1960s on Paramecium, a large protozoan whose surface is covered with rows of motile cilia. Normally, all of the rows are aligned with the same polarity through the coordinated replication of basal bodies, which consistently produce daughter basal bodies with the same orientation relative to the cell surface. The array of cilia growing from these basal bodies enables the cell to swim with great efficiency. By grafting experiments, however, it is possible to disturb this pattern and produce some inverted rows of cilia that beat in the direction opposite to that of their neighbors ( Figure 16-48). Once established, such altered patterns are passed on from parent to daughter Parameciumfor more than 100 generations. This form of heredity has nothing to do with DNA: the modified cells inherit a particular pattern of ciliary rows through the stereospecific replication of their basal bodies.

Summary

The axoneme of a cilium and a eucaryotic flagellum contains a cylindrical bundle of nine outer doublet microtubules. Dynein side arms extend between adjacent microtubule doublets and hydrolyze ATP to generate a sliding force between the doublets. Accessory proteins bundle the ring of microtubule doublets together and convert the sliding force into the bending movement that underlies ciliary beating. The complex structure of the ciliary axoneme forms by the self-assembly of its component proteins and is nucleated by a centriole (basal body), which serves as a template for the distinct 9 + 2 pattern of microtubules that forms the core axoneme. The centriole duplicates in a highly controlled process in which a daughter centriole is nucleated from the side of a mother centriole and grows at right angles to it. Oriented replication of basal bodies underlies the heritable pattern of beating cilia on the surface of ciliated protozoa.

Introduction

All eucaryotic species contain actin. This cytoskeletal protein is the most abundant protein in many eucaryotic cells, often constituting 5% or more of the total cell protein. Vertebrate skeletal muscle cells are the usual source of actin for experiments done in vitro, as about 20% of their mass is actin. If dry powdered muscle is treated with a very dilute salt solution, the actin filaments dissociate into their actin subunits. Each actin molecule is a single polypeptide 375 amino acids long that has a molecule of ATP tightly associated with it.

Actin filaments can form both stable and labile structures in cells. Stable actin filaments form the core of microvilli and are a crucial component of the contractile apparatus of muscle cells. Many cell movements, however, depend on labile structures constructed from actin filaments. In this section we focus on the question of how the cell controls the assembly of dynamic actin filaments from pools of soluble actin subunits in the cytosol.

Actin Filaments Are Thin and Flexible ³²

Actin filaments appear in electron micrographs as threads about 8 nm wide. They consist of a tight helix of uniformly oriented actin molecules (also known as globular actin, or G actin) ( Figure16-49). Like a microtubule, an actin filament is a polar structure, with two structurally different ends - a relatively inert and slow-growing minus end and a faster-growing plus end. Because of the oriented "arrowhead" appearance of the complex formed between actin filaments and the motor protein myosin, which we describe later, the minus end is also referred to as the "pointed end" and the plus end as the "barbed end." The three-dimensional structure of the actin molecule has been solved by x-ray diffraction analysis, and this information has been used to deduce the structure of an actin filament at the level of individual amino acids ( Figure16-50).

Some lower eucaryotes, such as yeasts, have only one actin gene, encoding a single protein. All higher eucaryotes, however, have several isoforms encoded by a family of actin genes. At least six types of actin are present in mammalian tissues; these fall into three classes, depending on their isoelectric point. Alpha actins are found in various types of muscle, whereas β and γ actins are the principal constituents of nonmuscle cells. Although there are subtle differences in the properties of different forms of actin, the amino acid sequences have been highly conserved in evolution, and all assemble into filaments that are essentially identical in most tests performed in vitro.

The total length of all of the actin filaments in a cell is at least 30 times greater than the total length of the microtubules, reflecting a fundamental difference in the way these two cytoskeletal polymers are organized and function in cells. Actin filaments are thinner and more flexible, and usually much shorter, than microtubules. We shall see that actin filaments rarely occur in isolation in the cell but rather in cross-linked aggregates and bundles, which are much stronger than the individual filaments.

Actin and Tubulin Polymerize by Similar Mechanisms ³³

Polymerization of pure actin in vitro requires ATP as well as both monovalent and divalent cations, which are usually K⁺ and Mg²⁺. The reaction is often studied either by observing the change in the light emission from a fluorescent probe that has been covalently attached to the actin or by monitoring the large increase in viscosity caused by the polymerization. When K⁺ and Mg²⁺are added to monomeric actin in the presence of ATP, there is initially a lag phase, as new filaments are nucleated, and then a rapid polymerization phase, as the short filaments elongate. The lag in polymerization with pure actin is due to the same kinetic barrier to nucleation that we discussed for tubulin polymerization (see Figure 16-23). For actin the rate of nucleation is proportional to the cube of the actin concentration, suggesting that the nucleating structure for the spontaneous polymerization of pure actin is a trimer of actin molecules. By contrast, the rate at which each filament elongates is proportional, as for microtubules, to the concentration of the free subunit, indicating that the filament elongates by the addition of one actin molecule at a time.

The polymerization rate is different at the two ends of the actin filament, and this difference is greater than for microtubules: the plus (or barbed) end of actin filaments polymerizes at up to 10 times the rate of the minus (or pointed) end. The critical concentration for actin polymerization - that is, the free actin monomer concentration at which the proportion of actin in polymer stops increasing - is around 0.2 micromolar (about 8 µg/ml). This concentration is very much lower than the concentration of unpolymerized actin in a cell, and the cell has evolved special mechanisms to prevent most of its monomeric actin from assembling into filaments, as we discuss later.

Shortly after polymerization, the terminal phosphate of the ATP bound to the actin molecule is hydrolyzed, leaving the resulting ADP trapped in the polymer. The hydrolysis of ATP during actin polymerization is analogous to the GTP hydrolysis that accompanies microtubule assembly, but in the case of actin we can understand the conformational changes involved because the three-dimensional structure of actin is known. The actin molecule is clam-shaped and binds ATP in the crevice between its two halves; like a clam shell, it can open and close. When actin polymerizes, the shell is clamped shut by interactions between amino acids on both lips of the shell and the back side of the next subunit in the polymer. It is thought that ATP hydrolysis is triggered by the closing of the clam shell as each actin molecule is incorporated into the filament, leaving ADP trapped inside ( Figure 16-51).

ATP Hydrolysis Is Required for the Dynamic Behavior of Actin Filaments ³⁴

The role of ATP hydrolysis in actin polymerization is similar to the role of GTP hydrolysis in tubulin polymerization, as explained in Panel16-1 (pp. 824-825). In neither case is hydrolysis required to form the filament; instead, it serves to weaken the bonds in the polymer and thereby promote depolymerization. There are, however, important differences in the behavior of the bound nucleotide in the subunits of these two polymers. An especially interesting difference is that ATP-ADP exchange (the replacement of bound ADP by ATP) is relatively slow for free actin (half-time of minutes), while GTP-GDP exchange is very rapid for free tubulin (half-time of seconds); thus, when actin molecules are released by disassembly of a filament, there is a relatively long delay before they can be re-used in filament assembly. In principle, this property of actin allows the cell to maintain a high cytosolic concentration of unpolymerized actin molecules in the form of ADP actin; furthermore, the ADP-actin monomer in a cell can be stabilized by binding to another protein, and this could provide a way to regulate actin polymerization.

The effect of ATP hydrolysis on actin is subtle, and there are still many questions about its precise consequences for the cell. Actin filaments, unlike microtubules, do not seem to show drastic dynamic instability in vitro. Instead, they can engage in an interesting dynamic behavior called treadmilling, which occurs when actin molecules are added continually to the plus end of the filament and are lost continually from the minus end, with no net change in filament length (see Panel 16-1, pp. 824-825). Treadmilling, like dynamic instability, is a nonequilibrium behavior that requires an input of energy, which is provided by the ATP hydrolysis that accompanies polymerization. This phenomenon is thought to contribute to the rapid exchange of the subunits of actin filaments that takes place in cells.

It is remarkable that actin and tubulin have both evolved nucleoside tri-phosphate hydrolysis for the same basic reason - to enable them, having polymerized, to depolymerize readily. Actin and tubulin are completely unrelated in amino acid sequence: actin is distantly related in structure to the glycolytic enzyme hexokinase, whereas tubulin is distantly related to a large family of GTPases that includes the heterotrimeric G proteins and monomeric GTPases such as Ras. (Both types of structures are discussed in detail in Chapter 5.) The convergent evolution of the capacity for nucleotide hydrolysis in actin and tubulin demonstrates just how important it is to microtubule and actin filament function: the dynamic assembly and disassembly of these cytoskeletal polymers that hydrolysis makes possible lies at the heart of cytoplasmic organization.

The Functions of Actin Filaments Are Inhibited by Both Polymer-stabilizing and Polymer-destabilizing Drugs ³⁵

Drugs that stabilize or destabilize actin filaments provide important tools to investigate their dynamic behavior in cells. The cytochalasins are fungal products that prevent actin from polymerizing by binding to the plus end of actin filaments. The phalloidins are toxins isolated from the Amanita mushroom that bind tightly all along the side of actin filaments and stabilize them against depolymerization. (One remedy for Amanita mushroom poisoning is to eat a large quantity of raw meat: the high concentration of actin filaments in the muscle tissue binds the phalloidin and thereby reduces its toxicity.) Both of these drugs cause dramatic changes in the actin cytoskeleton. We saw earlier for microtubules that both polymer-destabilizing drugs such as colchicine and polymer-stabilizing drugs such as taxol are toxic to cells, and the same is true for drugs affecting the stability of actin filaments, indicating that the function of actin filaments also depends on a dynamic equilibrium between the filaments and actin monomer.

Cytochalasin has found its greatest use in studying cell locomotion. In particular, the leading edge of a moving cell contains actin filaments that are continually polymerizing and are therefore very sensitive to cytochalasin. In most moving cells cytochalasin causes the leading edge rapidly to retract. If the plasma membrane of the leading edge is very firmly attached to the substratum, however, cytochalasin causes the actin filaments to retract but leaves the membrane behind, stuck to the substratum ( Figure16-52).

Phalloidin is widely used, as a fluorescent derivative, to stain actin filaments in fixed cells, and it also has a profound effect on living cells. When it is microinjected into a living fibroblast, for example, it drives all of the actin monomer into filaments at random positions in the cytoplasm, causing a drastic blebbing and contraction that often destroys the cell.

The Actin Molecule Binds to Small Proteins That Help to Control Its Polymerization ³⁶

In a fibroblast cell approximately 50% of the actin is in filaments and 50% is in monomer. The monomer concentration is typically 50-200 micromolar (2-8 mg/ml) in a variety of cell types; this is surprisingly high, given the low critical concentration of pure actin (less than 1 micromolar), and it reflects the presence of special proteins that bind to the actin molecule and inhibit its addition to the ends of actin filaments. The most abundant of these actin-monomer-binding proteins in many cells is thymosin, an unusually small protein with a molecular weight of about 5000. In the cells in which it has been most carefully studied (blood platelets and neutrophils), it is present in concentrations that are sufficient to sequester all of the monomeric actin. It is not clear how this protein inhibits actin polymerization: it could sterically block polymerization by covering a site where one monomer binds to another, or it could trap ADP on actin by inhibiting ADP-ATP exchange, thereby making the actin molecule unlikely to polymerize ( Figure 16-53).

Another actin-monomer-binding protein is profilin, which is present in all cells and is thought to play a part in controlling actin polymerization in response to extracellular stimuli. Profilin, which in many cells is largely associated with the plasma membrane, accelerates the exchange of ATP for ADP when bound to actin monomers and is thought to play a part in promoting the regulated polymerization of actin during cell movement, although this is still controversial. A mutant yeast cell that is deficient in profilin has a deficit of actin filaments, which supports a role for this molecule in stimulating the polymerization of actin.

In addition to thymosin and profilin, cells contain other abundant proteins that are able to bind actin monomers, and some of these, such as actin-depolymerizing factor (ADF), inhibit the assembly of actin into filaments. Evidently cells have a variety of mechanisms, the details of which are not yet understood, by which they hold stocks of actin monomer in reserve in order to assemble actin filaments only when and where they are needed.

Many Cells Extend Dynamic Actin-containing Microspikes and Lamellipodia from Their Leading Edge ³⁷

Dynamic surface extensions containing actin filaments are a common feature of animal cells, especially when the cells are moving or changing shape. The large, free-living cells of Amoeba proteus,for example, produce pseudopodia - stubby distensions of the actin cortex - with which they walk over surfaces. Many cells in vertebrate tissues are also capable of independent migration over surfaces, especially when put into tissue culture. The leading edge of a crawling fibroblast regularly extends a thin, sheetlike process known as a lamellipodium, which contains a dense meshwork of actin filaments. Many cells also extend thin, stiff protrusions called microspikes, which are about 0.1 µ wide and 5 to 10 µ long and contain a loose bundle of about 20 actin filaments oriented with their plus ends pointing outward (see Figure 16-9). The growing tip (growth cone) of a developing nerve cell axon extends even longer microspikes, called filopodia, which can be up to 50 µ long.

A lamellipodium can be viewed as a two-dimensional version of a micro-spike; indeed, short microspikes often project from the edges of a lamellipodium. When carefully fixed and stained for examination in an electron microscope, the actin filaments in the lamellipodium of a moving cell appear to be more organized than they are in other regions of the cell cortex. Many of the filaments project outward in an orderly array, with their plus ends inserted into the leading edge of the plasma membrane ( Figure16-54). The lamellipodium behaves as a structural unit; if it fails to adhere to the substratum, it is usually swept rapidly backward over the top of the cell as a "ruffle" ( Figure16-55).

Both lamellipodia and microspikes are motile structures that can form and retract with great speed. As we discuss next, it is thought that microspikes and lamellipodia are generated by local actin polymerization at the plasma membrane and that such actin polymerization can rapidly push out the plasma membrane without tearing it.

The Leading Edge of Motile Cells Nucleates Actin Polymerization ³⁸

When the behavior of actin filaments at the leading edge is studied by labeling a small patch of actin and following its movement, it is seen that actin is continually moving back toward the cell body at a speed of about 1 µm/minute, suggesting that actin is continuously polymerizing near the tip of the leading edge and continuously depolymerizing at more internal sites ( Figure16-56). This highly dynamic behavior of actin filaments at the leading edge is thought to be crucial for such processes as directed cell locomotion and chemotaxis. It gives the impression that the leading edge is propelling itself forward by pushing actin filaments backward.

The leading edge of a cell seems to organize actin filaments much as a centrosome organizes microtubules but with one crucial difference: it not only nucleates the growth of new filaments, but also seems to be the site at which monomers are added subsequently to enable the filaments to elongate. This role can be demonstrated by gently lysing a fibroblast and then adding rhodamine-tagged actin monomers, which are seen to polymerize preferentially at the tip of the leading edge ( Figure 16-57). Moreover, if the actin filaments in a cell are decorated to reveal their polarity, the fast-growing plus end of each actin filament is found to be attached to the membrane at the leading edge.

There are many unanswered questions about the mechanism by which the leading edge nucleates actin filament polymerization. Does the leading edge hold on to the plus end of a filament that it nucleates, for example, or does it nucleate a new filament and then quickly release it? Because of the continuous backward movement of actin (see Figure 16-56), any model that postulates that the leading edge holds onto actin filament ends would require that the filaments in the lamellipodium undergo continuous treadmilling by insertion of actin monomers at the site where the filaments are held by the membrane. In an alternative model, individual actin filaments are released and move away from the membrane (presumably as a cross-linked meshwork) soon after they form ( Figure16-58).

The rapid assembly of actin filaments at the leading edge of a moving cell requires that actin monomers be released from the actin-monomer-binding proteins that normally restrain their polymerization into filaments. We discuss below how signals in the cell's environment may regulate the release of actin monomers for polymerization at the tip of the leading edge.

Some Pathogenic Bacteria Use Actin to Move Within and Between Cells ³⁹

Listeria monocytogenes, a bacterium that causes a severe form of food poisoning, has provided unexpected insights into the mechanism by which the local polymerization of actin is controlled in cells. This pathogenic bacterium enters cells by being phagocytosed; it then escapes into the host cell cytosol by secreting enzymes that break down the membrane of the phagosome. Once in the cytosol, the bacteria not only grow and divide, but they also spread to adjoining cells by mobilizing the actin-based motility system of the host cell. By nucleating actin filaments at one region of its surface, an individual bacterium moves through the cytosol at rates of 10 µm/minute or more, laying down a tail of actin filaments behind it. When it collides with the plasma membrane of the host cell, it keeps moving outward, inducing the formation of a long, thin microspike with a bacterium at its tip. This projection is often engulfed by a neighboring cell, allowing the bacterium to enter its cytoplasm without exposure to the extracellular environment, thereby avoiding recognition by antibodies produced by the host ( Figure 16-59).

This form of movement suggests that the bacterium may be using actin to propel itself forward in the same way that the plasma membrane of a eucaryotic cell uses actin to propel itself forward during the formation of a normal microspike or lamellipodium.

If the actin filaments in the tail behind a Listeria bacterium migrating in the cytosol are marked with a fluorescent tag and observed by fluorescence microscopy, they are found to be stationary. The filaments form at the rear of the bacterium and are left behind like a rocket trail as the bacterium advances, depolymerizing again within a minute or so as they encounter depolymerizing factors in the cytosol. Assembly is induced by a specific protein on the surface of the bacterium that acts indirectly by sequestering host-cell proteins, including profilin. Since bacterium-induced movement can be reproduced in a concentrated cell-free extract, details of the mechanism should emerge from biochemical studies. These details should help us to understand how actin nucleation and polymerization occur in the microspikes and lamellipodia of a normal, uninfected cell and how these processes power the forward movement of the cell.

Polymerization of Actin in the Cell Cortex Is Controlled by Cell-Surface Receptors ⁴⁰

The production of movement is of little use unless it is properly directed according to the environment. As discussed earlier, the dynamic cortical meshwork of actin filaments rearranges rapidly in response to signals from outside the cell that impinge on the plasma membrane. The actin cytoskeleton can therefore be considered to be an integral part of the cell's signal-transduction systems, discussed in Chapter 15: when certain growth factors are added to the medium bathing quiescent cells in culture, for example, they immediately cause actin-containing lamellipodia to form and move over the cell surface.

The response of the actin cortex to external signals conveying spatial information can be highly localized. We considered one example earlier when we discussed the polarization of a cytotoxic T cell that is induced by contact with the target cell it subsequently kills (see Figure16-11). A signal-induced polarization of the actin cortex also occurs in animal cells that are capable of chemotaxis, which is defined as movement in a direction controlled by a gradient of a diffusible chemical sensed by the cell. One well-studied example is the chemotactic movement of certain white blood cells ( neutrophils) toward a source of bacterial infection. Neutrophils have receptor proteins on their surface that enable them to detect the very low concentrations of the N-formylated peptides derived from bacterial proteins (only procaryotes begin protein synthesis with N-formyl methionine). The neutrophils can be guided to their targets by a difference of only 1% in the concentration of these diffusible peptides on one side of the cell versus the other.

Another example of chemotaxis is provided by the cellular slime mold Dictyostelium discoideum. These eucaryotes live on the forest floor as independent motile cells called amoebae, which feed on bacteria and yeast and, under optimal conditions, divide every few hours. When their food supply is exhausted, the amoebae stop dividing and gather together to form tiny (1-2 mm), multicellular, wormlike structures, which crawl about as glistening slugs and leave trails of slime behind them ( Figure 16-60). As the slug migrates, the cells begin to differentiate, initiating a process that ends with the production of a tiny plantlike structure consisting of a stalk and a fruiting body some 30 hours after the beginning of aggregation ( Figure 16-61). The fruiting body contains large numbers of spores, which can survive for long periods of time even in extremely hostile environments. Only when conditions are favorable do the spores germinate to produce the free-living amoebae that start the cycle again.

The Dictyostelium amoebae aggregate by chemotaxis, migrating toward a source of cyclic AMP, which is secreted by the starved amoebae. Like neutrophils, the amoebae reorient their leading edge in order to migrate up a shallow chemoattractant gradient. And when they are exposed to a local source of cyclic AMP leaking from a micropipette, they extend actin-containing processes directly toward the pipette ( Figure 16-62). This experiment shows that eucaryotic chemotaxis involves detecting a spatial gradient of attractant concentration directly, in contrast to bacterial chemotaxis, which uses a time-dependent variation in concentration to detect gradients, as discussed in Chapter 15.

The cytoskeletal reaction of Dictyostelium amoebae to cyclic AMP can be examined by making lysates of these cells very shortly after bulk stimulation with cyclic AMP in solution. As shown in Figure16-63, a dramatic burst of actin polymerization occurs 5-10 seconds after adding cyclic AMP, which corresponds to the time required for flattening of the cells on the substratum. Between 20 and 40 seconds after the pulse of cyclic AMP, actin depolymerizes and the cells round up. Then there is a more prolonged burst of actin polymerization as actin-binding proteins are recruited into the cytoskeleton from soluble pools; during this latter period the cells that respond to cyclic AMP begin to extend lamellipodia and other actin-rich processes.

Heterotrimeric G Proteins and Small GTPases Relay Signals from the Cell Surface to the Actin Cortex ⁴¹

How does cyclic AMP binding to its receptor in Dictyostelium amoebae trigger massive actin polymerization? The receptor is known to activate a heterotrimeric G protein. The cytoplasm contains a reservoir of actin monomers, which, as we saw earlier, are stabilized by actin-monomer-binding proteins. Stimulation of actin polymerization requires that these actin molecules be made available in a form that can polymerize and also that nucleation sites for actin filaments be provided to overcome the kinetic barrier to nucleation. The actin-monomer-binding protein profilin binds tightly to the inositol phospholipids in the plasma membrane that generate intracellular signals in response to extracellular ligands (see Figure 15-30). According to one hypothesis, activation of this signaling pathway (which occurs via a heterotrimeric G protein) could release profilin from the plasma membrane into the cytosol. Profilin can catalyze ATP-ADP exchange on actin in vitro, and so when it is released from the plasma membrane, it may rapidly convert inactive ADP actin to active ATP actin to induce the local formation of actin filaments.

G proteins have also been implicated in the signaling processes that activate the actin cortex during the chemotactic response of neutrophils and the activation of blood platelets. There is evidence that two Ras-related small GTPases known as Rho and Rac act downstream; these proteins have been shown to have distinct effects on the actin cytoskeleton in fibroblasts. Microinjection of Rac protein into cultured cells causes a dramatic increase in the formation of lamellipodia within 5 minutes. Moreover, a dominant-negative mutant form of Rac inhibits the formation of lamellipodia normally induced by various growth factors, indicating that this response to growth factors depends on Rac. Microinjection of Rho protein leads to the appearance of large bundles of actin filaments known as stress fibers and to the enhancement of focal contacts, where the cell is attached to the substratum externally and stress fibers are anchored internally (as we discuss later). Rho is also thought to be needed to assemble the contractile ring during cell division. Thus Rac and Rho not only control the polymerization of actin into filaments but also govern the organization of these filaments into specific types of structures.

Mechanisms of Cell Polarization Can Be Analyzed in Yeast Cells ⁴²

Further clues to how cells may orient the activities of their cytoskeleton have come from the behavior of yeast cells. The ease of genetic analysis in yeasts has made them an important source of fundamental information about biological mechanisms that are common to all eucaryotic cells. In particular, studies on the interactions between yeast cells during mating have begun to identify mechanisms by which eucaryotic cells become structurally polarized. In the budding yeast Saccharomyces cerevisiae, cells of two mating types, a and α, secrete hormones, known as a-factor and α-factor, respectively. These hormones act by binding to cell-surface receptors that belong to the large G-protein-linked receptors discussed in Chapter 15. One consequence of the binding of α-factor to its receptors on an a-cell is to cause the cell to become polarized so that it adopts a shape known as a "shmoo" ( Figure16-64). If an α-factor gradient is present, the shmoo tip is directed toward the highest concentration of this signaling molecule.

During this polarization response the yeast cell undergoes cytoskeletal reorganizations that parallel those of an animal cell that is becoming polarized. Actin filaments congregate at the pointed shmoo tip, where they are thought to direct the local secretion of cell-wall components - possibly by directing the transport vesicles carrying these components to the shmoo tip. At the same time, the microtubule organizing center (in this case the spindle pole body, see Figure 17-24) moves to the side of the nucleus that is closest to the shmoo tip, and microtubules extend from it toward the tip. By screening for mutant cells that fail to form a shmoo during mating, many of the genes involved in yeast-cell polarization are being identified. It is likely that some of the proteins that these genes encode will also be involved in polarizing an animal cell.

Summary

Actin is a highly conserved cytoskeletal protein that is present at high concentrations in nearly all eucaryotic cells. Purified actin exists as a monomer in low ionic strength solutions and spontaneously assembles into actin filaments on addition of salt provided ATP is present. As with tubulin, the polymerization of actin is a dynamic process that is regulated by the hydrolysis of a tightly bound nucleotide (ATP in this case). In cells, approximately half of the actin is kept in a monomeric form through its binding to small proteins such as thymosin. In the cortex of animal cells, actin molecules continually polymerize and depolymerize to generate cell-surface protrusions such as lamellipodia and microspikes. Polymerization can be regulated by extracellular signals binding to cell-surface receptors that act through heterotrimeric G proteins and the small GTPases Rac and Rho.

Introduction

Actin is involved in a remarkably wide range of structures, from stiff and relatively permanent extensions of the cell surface to the dynamic three-dimensional networks at the leading edge of a migrating cell. Very different structures based on actin coexist in every living cell. In every case the fundamental structure of the actin filament is the same. It is the length of these filaments, their stability, and the number and geometry of their attachments (both to one another and to other components of the cell) that varies in different cytoskeletal assemblies. These properties in turn depend on a large retinue of actin-binding proteins, which bind to actin filaments and modulate their properties and functions.

In this section we describe some of the most important actin-binding proteins and the structures they form. Many of these are found at the perimeter of the cell in the actin-rich layer just beneath the plasma membrane called the cell cortex. This layer gives an animal cell mechanical strength and enables it to perform a variety of surface movements, such as phagocytosis, cytokinesis (cell division), and cell locomotion.

A Simple Membrane-attached Cytoskeleton Provides Mechanical Support to the Plasma Membrane of Erythrocytes ⁴⁴

As noted in Chapter 10, the proteins spectrin and ankyrin were first discovered as prominent components of the membrane-associated cytoskeleton of mammalian red blood cells (erythrocytes). These unusual cells have lost their nucleus and internal membranes, and so the plasma membrane is the only membrane. It is supported by a two-dimensional network of spectrin tetramers that are connected at their ends by very short actin filaments. The spectrin is linked to the cytoplasmic tail of an abundant transmembrane carrier protein (band 3) by means of ankyrin bridges (see Figure 10-26). Close relatives of spectrin (also called fodrin) and of ankyrin are found in the cortex of many vertebrate cells. Thus the detailed arrangement of proteins in the erythrocyte cortex provides a simplified model for the actin-based cytoskeletal network that supports the plasma membrane in all other animal cells.

The actin filaments in the erythrocyte cortex are very short, acting only as cross-linking elements between spectrin tetramers. Those in a more typical cell cortex, by contrast, are much longer and thus project into the cytoplasm, where they form the basis of a three-dimensional actin filament network. It is uncertain whether ankyrinlike molecules anchor these more typical cortical arrays to the plasma membrane, although in some epithelial cells the transmembrane Na⁺/K⁺ ATPase (discussed in Chapter 11) is thought to link the plasma membrane to the cortical actin filament network through such molecules.

The cortical actin filament network generally determines the shape and mechanical properties of the plasma membrane. Many types of membrane attachments are needed for actin filaments to perform their various functions in the cortex; coupling to transmembrane proteins through ankyrin is only one. More dynamic attachments also exist, but the proteins that mediate them are just beginning to be characterized.

Cross-linking Proteins with Different Properties Organize Particular Actin Assemblies ⁴³

The cortical actin filaments in animal cells are organized into three general types of arrays ( Figure 16-65). In parallel bundles, as found in microspikes and filopodia, the filaments are oriented with the same polarity and are often closely spaced (10-20 nm apart). In contractile bundles, as found in stress fibers and in the contractile ring that divides cells in two during mitosis, filaments are arranged with opposite polarities; they are more loosely spaced (30-60 nm apart) and contain the motor protein myosin-II (discussed later). In the gel-like networks of the cell cortex the filaments are arranged in a relatively loose, open array with many orthogonal interconnections. How are these different arrangements of the same actin filament generated and maintained within a single cell? While we do not know the complete answer, actin filament cross-linking proteins are clearly of central importance.

Actin filament cross-linking proteins can be divided into two classes - bundling proteins and gel-forming proteins - according to their effect on pure actin filaments in vitro. Bundling proteins cross-link actin filaments into a parallel array and are important for forming both the tight parallel arrays and the looser contractile bundles of actin filaments described above. Gel-forming proteins, by contrast, cross-link actin filaments at crosswise intersections, creating loose gels.

Fimbrin and α-actinin are widely distributed bundling proteins. Fimbrin is enriched in the parallel filament bundles at the leading edge of cells, particularly in microspikes and filopodia, and it is thought to be responsible for the tight association of actin filaments in these arrays. The second actin-bundling protein, α-actinin, is concentrated in stress fibers, where it is thought to be partly responsible for the relatively loose cross-linking of actin filaments in these contractile bundles; it also helps to form the anchorage for the ends of stress fibers where they terminate on the plasma membrane at focal contacts. As explained later, myosin is the motor protein in stress fibers and other contractile arrays that is responsible for their contractility. It seems likely that the very close packing of actin filaments caused by fimbrin excludes myosin, whereas the looser packing caused by α-actinin allows myosin molecules to enter; likewise, the very different spacing causes each of the two bundling proteins to exclude the other ( Figure 16-66).

Filamin is a widely distributed gel-forming protein. Although it is not present in stress fibers or the leading edge, it is enriched elsewhere in the cortex. Filamin is a homodimer that promotes the formation of a loose and highly viscous network by clamping together two actin filaments that cross each other ( Figure16-67). It is an abundant protein in many animal cells, reflecting the prevalence of the loose-network type of actin organization.

Actin-binding Proteins with Different Properties Are Built Up from Similar Modules ⁴⁵

Fimbrin, α-actinin, filamin, and spectrin each contain two actin-filament-binding domains, which is not surprising given that each needs to cross-link two filaments. Unexpectedly, however, in all of these proteins the actin-binding domains have a similar structure. The length and flexibility of the spacer sequences that separate the two actin-binding sites differ in the four proteins, and these differences determine the different properties of the four cross-linkers. Evidently, these proteins have diverged from a common ancestral actin-binding protein by adding different spacer sequences ( Figure 16-68).

Gelsolin Fragments Actin Filaments in Response to Ca²⁺ Activation ⁴⁶

Extracts prepared from many types of animal cells form a gel in the presence of ATP when they are warmed to 37°C. Although this gelation depends on both actin filaments and a cross-linking protein such as filamin, the gels exhibit more complex behavior than simple mixtures of actin filaments and filamin. If the Ca²⁺ concentration is raised above 10^-7M, for example, the semisolid actin gel begins to liquefy - a process known as solation - and regions of the solating gel show vigorous local streaming when examined under a microscope. Clearly, there must be components besides actin and filamin in the extracts to account for this behavior. These components are likely to be involved in the cytoplasmic streaming observed in some large cells, where vigorous flowing movements are required to maintain an even distribution of metabolites and other cytoplasmic components. These movements seem to be associated with sudden local changes in the cytoplasm from a solid gel-like consistency to a more fluid state.

A number of proteins have been isolated from cell extracts that, when added to a gel formed from purified actin filaments and filamin, cause it to change to a more fluid state in the presence of Ca²⁺. The best characterized of these is gelsolin, which, when activated by the binding of Ca²⁺, severs an actin filament and forms a cap on the newly exposed plus end of the filament, thus breaking up the cross-linked network of actin filaments. Similar proteins are found in the cortex of many types of vertebrate cells; these severing proteins are activated by concentrations of Ca²⁺ (about 10^-6 M) that occur only transiently in the cytosol.

One of the postulated functions of severing proteins is to help loosen or liquefy the cell cortex locally to allow membrane fusion events. When a phagocytic white blood cell engulfs a microorganism, for example, the resulting phagosome is initially coated on its cytoplasmic side with a thick network of actin filaments originating from the cortex. In order for this phagosome to fuse with lysosomes, these actin filaments must be depolymerized to allow intimate contact between the phagosome and lysosome membranes. This removal of actin can be prevented by artificially reducing the Ca²⁺ ion concentration, and it is thought that removal may depend on a local rise in Ca²⁺ through the action of gelsolin (or a similar protein). Gelsolin is also thought to be required for a cell to crawl along a substratum, although its exact role in this process is not clear.

While a mixture of purified actin filaments, filamin, and gelsolin is capable of undergoing Ca²⁺-dependent gel-to-sol transitions, it will not contract or show the streaming movements displayed by the cruder actin-rich gels obtained from cells. These activities require another type of actin-binding protein - the motor protein myosin. If myosin is selectively removed from the crude actin-rich gels, contractions and streaming no longer occur, suggesting that an interaction between actin and myosin generates the force for cytoplasmic streaming.

Multiple Types of Myosin Are Found in Eucaryotic Cells ⁴⁷

Time-lapse cinematography reveals the cortex of cells to be continually moving. In the previous section we emphasized the importance of actin filament polymerization and depolymerization in these movements, but, as with microtubules, motor proteins are also important. All of the actin filament motor proteins identified to date belong to the myosin family. Myosins were originally isolated on the basis of their ability to hydrolyze ATP to ADP and P_i when stimulated by binding to actin filaments, and this remains a useful biochemical criterion for their identification. It is also possible to observe the motor activity of myosins directly by adsorbing them onto a glass coverslip: when fluorescent actin filaments are added together with ATP, the filaments can be observed with a fluorescence microscope to glide over the myosin-coated glass surface. Novel myosins have also been identified by DNA sequencing even before being characterized biochemically or functionally.

Myosin, along with actin, was first discovered in skeletal muscle, and much of what we know about the interaction of these two proteins was learned there. Muscle myosin belongs to the myosin-II subfamily of myosins, all of which have two heads and a long, rodlike tail: each head has both ATPase and motor activity. A myosin-II protein is composed of two identical heavy chains, each of which is complexed to a pair of light chains. The amino-terminal portion of the heavy chain forms the motor-domain head, while the carboxyl-terminal half of the heavy chain forms an extended α helix. Two heavy chains associate by twisting their α-helical tail domains together into a coiled-coil to form a stable dimer that has two heads and a single rodlike tail ( Figure16-69).

A major role of the rodlike tail of myosin-II is to allow the molecules to polymerize into bipolar filaments. This polymerization is crucial for the function of myosin-II, which is to move groups of oppositely oriented actin filaments past each other, as seen most clearly in muscle contraction. Myosin-II is relatively abundant in the cell cortex; in fibroblasts, for example, there is roughly one myosin-II molecule per 100 actin molecules. Myosin-II filaments in the contractile ring are responsible for driving membrane furrowing during cell division, as discussed in Chapter 18, and they are thought to generate tension in stress fibers as well as much of the cortical tension that keeps the cell surface taut. Their role in muscle contraction is described at the end of the chapter.

In addition to myosin-II, which is generally the most abundant myosin in the cell, nonmuscle cells contain various smaller myosins, the best characterized of which is called myosin-I ( Figure 16-70). Myosin-I is thought to be more like the original, more primitive myosin from which myosin-II evolved. A single cell can contain multiple smaller myosins, each encoded by a different gene and performing a distinct function; the cellular slime mold Dictyostelium, for example, has at least nine. The common feature of all myosins is a conserved motor domain (motor head); the other domains vary from myosin to myosin and determine the specific role of the molecule in the cell. Thus myosin tails may have a membrane-binding site and/or a site that binds to a second actin filament independently of the head domain. Depending on its tail, a myosin molecule can move a vesicle along an actin filament, attach an actin filament to the plasma membrane, or cause two actin filaments to align closely and then slide past each other ( Figure 16-71).

All known myosins hydrolyze ATP to move along actin filaments from the minus end toward the plus end. Given the importance of oppositely directed motor proteins that move along microtubules (see Figure16-37), it would not be surprising to discover an additional class of motor proteins that move toward the minus end of an actin filament.

There Are Transient Musclelike Assemblies in Nonmuscle Cells ⁴⁸

In higher eucaryotic cells, organized contractile bundles of actin filaments and myosin-II filaments often form transiently to perform a specific function and then disassemble. Most notably, cell division in animal cells is made possible by a beltlike bundle of actin filaments and myosin-II filaments known as the contractile ring. This ring appears beneath the plasma membrane during the M phase of the cell-division cycle; forces generated by it pull inward on the plasma membrane and thereby constrict the middle of the cell, leading to the eventual separation of the two daughter cells by a process known as cytokinesis ( Figure16-72). The contractile ring must be assembled from actin, myosin, and other proteins at the start of cell division, a process that can be monitored by staining dividing cells with fluorescent anti-myosin antibodies. In sea urchin eggs that are about to divide, for example, myosin-II molecules are at first distributed evenly beneath the plasma membrane and then move to the equatorial region as the contractile ring forms. Once cell division is complete, the myosin-II molecules disperse. It is not known how this process is controlled, but it seems likely that Ca²⁺ is involved, as Ca²⁺-dependent phosphorylation of myosin-II both increases its interaction with actin and promotes its assembly into short bipolar filaments ( Figure 16-73).

Stress fibers, which are prominent components of the cytoskeleton of fibroblast cells in culture (see Figure 16-72), are a second example of a temporary contractile bundle of actin filaments and myosin-II. Although smaller and less highly organized, they resemble the tiny myofibrils in muscle (discussed later) in their structure and function. At one end they insert into the plasma membrane at special sites called focal contacts, where the external face of the cell is closely attached to the extracellular matrix ( Figure16-74); at the other end they insert into a second focal contact or into a meshwork of intermediate filaments that surrounds the cell nucleus. Stress fibers form in response to tension generated across a cell and are disassembled at mitosis when the cell rounds up and loses its attachments to the substratum. They also disappear rapidly if tension is released by suddenly detaching one end of the stress fiber from the focal contact by means of a laser beam. Stress fibers within fibroblasts in tissues are thought to allow the cells to exert tension on the matrix of collagen surrounding them - an essential process in both wound healing and morphogenesis (see Figure19-48). In epithelia, actin filament bundles spanning the cytoplasm from one cell-cell junction to another can appear and disappear in a similar way; such filament bundles, linked end to end via the cell-cell junctions, can form cables that transmit and generate tension along lines of particular stress in the multicellular sheet.

Not all contractile assemblies of actin filaments and myosin in nonmuscle cells are transitory. Those associated with the intercellular anchoring junctions called adhesion belts, for example, are often more lasting. Adhesion belts (discussed in Chapter 19) are found near the apical surface of epithelial cells (see Figure 16-72). Among other functions, they are thought to play an important part in the folding of epithelial cell sheets during embryogenesis.

The mechanism of contraction of all of these cytoskeletal bundles is based on the ATP-driven sliding of interdigitated actin and myosin filaments, and it is thought to require a particular type of ordered assembly, which will be explained later when we discuss muscle.

Focal Contacts Allow Actin Filaments to Pull Against the Substratum ⁴⁹

To pull on the extracellular matrix or on another cell, a stress fiber must be strongly anchored in the plasma membrane at the appropriate site. Attachments between actin filaments inside the cell and extracellular matrix on the outside of the cell are mediated by transmembrane linker glycoproteins in the plasma membrane. Those formed by cultured fibroblasts with the extracellular matrix are the best characterized. When fibroblasts grow on a culture dish, most of their cell surface is separated from the substratum by a gap of more than 50 nm; but at focal contacts ( adhesion plaques), this gap is reduced to 10 to 15 nm. Here the plasma membrane is attached to components of the extracellular matrix that have become adsorbed to the culture dish. Staining with anti-actin antibodies clearly shows these regions to be the sites where the ends of stress fibers attach to the plasma membrane (see Figure 16-74).

The main transmembrane linker proteins of focal contacts are members of the integrin family, whose external domain binds to an extracellular matrix component while the cytoplasmic domain is linked to actin filaments in stress fibers. The linkage is indirect and is mediated by multiple attachment proteins ( Figure 16-75). The cytoplasmic domain of the integrin binds to the protein talin, which in turn binds to vinculin, a protein found also in other actin-containing cell junctions, such as adherens junctions (discussed in Chapter 19). Vinculin associates with α-actinin and is thereby linked to an actin filament. Although the exact topology of protein interactions in the focal contact has not been established, a possible arrangement is shown in Figure 16-75B.

Besides their role as anchors for the cell, focal contacts can also relay signals from the extracellular matrix to the cytoskeleton. Several protein kinases, including the tyrosine kinase encoded by the src gene, are localized to focal contacts, and there are indications that their activity changes with the type of substratum on which the cell rests. These kinases can phosphorylate various target proteins, including components of the cytoskeleton, and hence regulate the survival, growth, morphology, movement, and differentiation of cells in response to the extracellular matrix in their environment.

Microvilli Illustrate How Bundles of Cross-linked Actin Filaments Can Stabilize Local Extensions of the Plasma Membrane ⁵⁰

Microvilli are fingerlike extensions found on the surface of many animal cells. They are especially abundant on those epithelial cells that require a very large surface area to function efficiently. A single absorptive epithelial cell in the human small intestine, for example, has several thousand microvilli on its apical surface. Each is about 0.08 µm wide and 1 µm long, making the cell's absorptive surface area 20 times greater than it would be without them. The plasma membrane that covers these microvilli is highly specialized, bearing a thick extracellular coat of polysaccharide and digestive enzymes. The cytoskeleton of the microvillus has been studied in detail - a task that is made easier by its highly ordered structure, compared with the less specialized regions of cell cortex.

At the core of each intestinal microvillus is a rigid bundle of 20 to 30 parallel actin filaments that extend from the tip of the microvillus down into the cell cortex. The actin filaments in the bundle are all oriented with their plus ends pointing away from the cell body and are held together at regular intervals by actin-bundling proteins. Although fimbrin, the bundling protein in microspikes and filopodia, helps to bundle actin filaments into microvilli, the most important bundling protein is villin, which is found only in microvilli ( Figure 16-76). Villin, like fimbrin, cross-links actin filaments into tight parallel bundles, but it has a different actin-binding sequence. When villin is introduced into cultured fibroblasts, which do not normally contain villin and have only a few small microvilli, the existing microvilli become greatly elongated and stabilized, and new ones may also be induced, suggesting that villin is mainly responsible for the formation of the long microvilli in epithelial cells.

At the base of the microvillus the actin filament bundle is anchored into a specialized region of cortex at the apical end of the intestinal epithelial cell. This cortex, known as the terminal web, contains a dense network of spectrin molecules overlying a layer of intermediate filaments ( Figure16-77). The spectrin is thought to provide rigidity and stability to the cortex in this region, and the anchoring of the actin filaments to the terminal web is thought to stiffen the microvilli, keeping their actin bundles projecting outward at a right angle to the apical cell surface.

The actin filament bundle is attached to the overlying plasma membrane of the microvillus by lateral bridges that can be seen in electron micrographs. The bridges are composed of a form of myosin-I that has several molecules of calmodulin (discussed in Chapter 15) bound to its tail region. The myosin is oriented with this tail region embedded in the membrane and its active ATP-binding head contacting the actin filaments. It is a mystery why a motor protein is used to link actin filaments to the membrane in microvilli. If the myosin-I in microvilli is motile, it should move toward the plus end of the actin filaments at the microvillus tip. This has lead to speculation that the myosin-I helps to pull the membrane up over the microvillus core, forming vesicles at its tip that are then released into the lumen of the intestine, where the digestive enzymes they carry continue their action.

The Behavior of the Cell Cortex Depends on a Balance of Cooperative and Competitive Interactions Among a Large Set of Actin-binding Proteins

The preceding examples show that the same actin filament can interact with different sets of actin-binding proteins at different locations in the cortex and that the actin-binding proteins can be segregated to different parts of the cell. What prevents the various sets of actin-binding proteins from mixing in the cytoplasm? It seems likely that both cooperative and competitive interactions among these proteins are important. One class of actin-binding proteins not yet discussed, for example, binds along the length of the actin filaments. The most widespread members of this class are the tropomyosins, which are rigid rod-shaped proteins named for similarities in their x-ray diffraction pattern to myosin-II. Like the tail of myosin-II, tropomyosin is a dimer of two identical α-helical chains that wind around each other in a coiled-coil. By binding along the length of an actin filament, the tropomyosin stabilizes and stiffens the filament. It also inhibits the binding of filamin to actin filaments, which probably explains why tropomyosin and filamin tend to be differentially distributed in cells. By contrast, tropomyosin binding to an actin filament increases the binding of myosin-II to the filament - an example of a cooperative interaction ( Figure16-78).

We can now begin to see how stress fibers and cortical networks of actin filaments can coexist in a common cytoplasm. At one site in a cell - perhaps nucleated at a forming focal adhesion under the influence of activated Rho protein - tropomyosin, myosin-II, and α-actinin associate with actin filaments and exclude filamin; the contractile activity of myosin-II then promotes further organizational changes to produce a stress fiber. At another site in the cell tropomyosin-deficient actin filaments bind filamin, producing a loose network that provides few sites where α-actinin can bind to two filaments at once, so that it is excluded; bending of the filaments in the loose meshwork may also discourage tropomyosin-binding, since this molecule prefers a straight filament. While this picture is partly speculative, it illustrates the basic pathway by which a combination of cooperative and competitive interactions can give rise to spatially differentiated actin filament arrays in a common cytoplasm. It is not known how the postulated local differences that initiate the formation of these assemblages are established; nor is it known how many distinct types of actin filament arrays can coexist in the same cell - there are certainly more than the two we have just mentioned.

Some of the actin-binding proteins discussed in this section are summarized in Figure 16-79.

The Migration of Animal Cells Can Be Divided into Three Distinct Actin-dependent Subprocesses ⁵¹

The crawling movements of animal cells are among the most difficult to explain at the molecular level. Different parts of the cell change at the same time, and there is not a single, easily identifiable locomotory organelle (analogous to a flagellum, for example). Although actin forms the basis of animal cell migration, it undergoes many different transformations as the cell moves forward, assembling into lamellipodia and microspikes, associating with focal contacts, forming stress fibers, and so on. A complete account would have to give a molecular explanation for these transformations, explain how they are coordinated in time and space, and also account for important biophysical parameters such as the development of tension in the cortex and the formation of strong adhesions between the cell and its substratum.

In broad terms, three distinct processes can be identified in the crawling movements of animal cells: protrusion, in which lamellipodia and microspikes (or filopodia) are extended from the front of the cell; attachment, where the actin cytoskeleton makes a connection with the substratum; and traction, where the body of the cell moves forward.

Protrusion is a function of the leading edge of the cell. Actin-rich lamellipodia and microspikes (or filopodia) extend forward over the substratum, a process that is accompanied by actin polymerization, as described previously. It seems likely that the protrusion is driven by actin polymerization at the leading edge (see Figure 16-58), although this is still debated. Myosin-I motors attached to the plasma membrane could also drive the cell forward by actively walking along actin filaments. Yet another possibility, which has been suggested to apply in particular to the locomotion of giant amoebae, is that protrusions are squeezed out of the front of the cell by hydrostatic pressure generated by the contraction of the cortex elsewhere in the cell.

The attachment of cortical actin filaments to the substratum was discussed earlier when we described focal contacts, although these are specialized attachment structures present in fibroblasts in culture and associated with the ends of stress fibers. Rapidly motile cells - such as Dictyostelium amoebae and white blood cells - make more diffuse contacts with the substratum. It is thought, however, that similar principles apply to these contacts: transmembrane receptors for extracellular matrix proteins link the plasma membrane to the substratum, and actin filaments in the cytoplasm interact with the cytoplasmic domains of these receptors through actin-binding proteins. The details of these important interactions are uncertain, but it is clear that the cell contacts with the substratum must be continually made and broken as the cell moves forward.

Traction is perhaps the most mysterious part of cell locomotion. In many cases it is thought that the force for cell locomotion is generated near the front of the cell and that the nucleus and bulk cytoplasm are dragged forward passively. The force generation can be viewed in different ways. The leading part of the cell might actively contract like a muscle fiber and thus pull on the back of the cell. In another view polymerization of actin filaments at the front of the cell extends the actin cortex forward, and the rear of the cell is then carried forward by the contractile force of the resulting cortical tension ( Figure16-80).

The Mechanism of Cell Locomotion Can Be Dissected Genetically ⁵²

One of the most powerful ways to analyze the mechanism of a complex cellular process is to examine the effect of mutations that result in the deletion, overexpression, or modification of specific proteins. In the case of eucaryotic cell locomotion, the amoeboid cells of the slime mold Dictyostelium are particularly suitable for genetic analysis. These cells have a shape and a manner of moving that closely resemble those of the cells of higher organisms. But because they are haploid, they are readily manipulated by reverse genetic methods. Thus it has been possible to delete a number of actin-binding proteins from these cells and examine the consequences for cell locomotion.

The role of myosin-II, for example, has been tested genetically by two methods. In one, a defective form of the myosin-II gene is substituted for the existing gene by homologous recombination (see Figure7-47), leading to a mutant strain with no myosin-II. The second strategy is to use anti-sense RNA (see Figure7-43) to inactivate myosin-II mRNA, which has essentially the same effect. In a normal crawling Dictyostelium myosin-II is concentrated near the rear of the cell and myosin-I is concentrated in the leading edge ( Figure16-81). In the mutants the myosin-I is unchanged but myosin-II is gone.

Remarkably, Dictyosteliumcells without myosin-II can still move over the substratum and respond chemotactically to a source of cyclic AMP, although both processes are somewhat impaired. Thus myosin-II is not absolutely essential for cell locomotion. Although protrusive activity at the leading edge of such mutant cells is quite normal, movement of the cell body forward is somewhat impaired, suggesting that myosin-II plays a role in generating traction. Nevertheless, myosin-I and/or actin polymerization must be able to drive the cell forward at a reasonable rate without the help of myosin-II.

Not surprisingly, the mutant cells are unable to form a contractile ring following mitosis and therefore develop into multinucleated giant cells. These cells eventually divide by using cell locomotion to tear themselves in two. It is interesting to speculate that such locomotion-dependent cytokinesis may represent a primitive cell division mechanism and that myosin-II might have evolved from myosin-I through natural selection for a more efficient cytokinetic apparatus.

Summary

The varied forms and functions of actin in eucaryotic cells depend on a versatile repertoire of actin-binding proteins that cross-link actin filaments into loose gels, bind them into stiff bundles, attach them to the plasma membrane, or forcibly move them relative to one another. Tropomyosin, for example, binds along the length of actin filaments, making them more rigid and altering their affinity for other proteins. Filamin cross-links actin filaments into a loose gel. Fimbrin and a-actinin form bundles of parallel actin filaments. Gelsolin mediates Ca²⁺-dependent fragmentation of actin filaments, thereby causing a rapid solation of actin gels. Various forms of myosin use the energy of ATP hydrolysis to move along actin filaments, either carrying membrane-bounded organelles from one location in the cell to another or moving adjacent actin filaments against each other. Sets of actin-binding proteins are thought to act cooperatively in generating the movements of the cell surface, including cytokinesis, phagocytosis, and cell locomotion. These movements are difficult to analyze because of the many components involved, but genetic approaches, in which genes encoding specific actin-binding proteins are mutated, can show the function of individual proteins in each process.

Introduction

Many of the proteins that associate with actin filaments in eucaryotic cells were first discovered in muscle. Muscle contraction is the most familiar and the best understood of all the kinds of movement of which animals are capable. In vertebrates, for example, running, walking, swimming, and flying all depend on the ability of skeletal muscle to contract rapidly on its scaffolding of bone, while involuntary movements such as heart pumping and gut peristalsis depend on the contraction of cardiac and smooth muscle, respectively.

Although muscle is the best-understood example of actin-based motility, it was a relatively late development in evolution, and it is highly specialized compared with more typical animal cells. In particular, the actin- and myosin-based contractile units of muscle cells, the myofibrils, are not labile like the actin- and myosin-based structures of nonmuscle cells.

Myofibrils Are Composed of Repeating Assemblies of Thick and Thin Filaments ⁵⁴

The long thin muscle fibers of skeletal muscle are huge single cells formed during development by the fusion of many separate cells (discussed in Chapter 22). The nuclei of the contributing cells are retained in this large cell and lie just beneath the plasma membrane. But the bulk of the cytoplasm (about two-thirds of its dry mass) is made up of myofibrils, which are the contractile elements of the muscle cell. They are cylindrical structures 1 to 2 µm in diameter and are often as long as the muscle cell itself ( Figure16-82).

Each myofibril consists of a chain of tiny contractile units, or sarcomeres, each about 2.2 µm long, which give the vertebrate myofibril its striated appearance. At high magnification a series of broad light and dark bands can be seen in each sarcomere; a dense line in the center of each light band separates one sarcomere from the next and is known as the Z line, or Z disc ( Figure 16-83).

Each sarcomere comprises a miniature, precisely arranged assembly of parallel and partly overlapping filaments. Thin filaments composed of actin with associated proteins are attached to the Z discs at either end of the sarcomere. They extend in toward the middle of the sarcomere, where they overlap with thick filaments, which are polymers of specific muscle isoforms of myosin-II ( Figure 16-83C). When this region of overlap is examined in cross-section by electron microscopy, the myosin filaments are seen to be arranged in a regular hexagonal lattice, with the actin filaments placed regularly between them ( Figure16-84).

Contraction Occurs as the Myosin and Actin Filaments Slide Past Each Other

Sarcomere shortening is caused by the myosin filaments sliding past the actin filaments with no change in the length of either type of filament ( Figure16-85). This sliding filament model,first proposed in 1954, was crucial to understanding the contractile mechanism.

The ultrastructural basis for the force-generating interaction is visible at very high magnification in electron micrographs. The myosin filaments are seen to possess numerous tiny side arms, or cross-bridges, that extend about 13 nm to make contact with adjacent actin filaments ( Figure 16-86). These cross-bridges are myosin-II heads, and when a muscle contracts, the myosin and actin filaments are pulled past each other by the cross-bridges acting cyclically, like banks of tiny oars.

As stated previously, the globular head, or motor domain, of the myosin-II molecule both binds to actin filaments and hydrolyzes ATP. Isolated myosin-II heads, which can be prepared by papain digestion, retain both the ATPase activity and the actin-filament-binding properties of the intact myosin-II molecule and therefore can be used to analyze the interaction between actin and myosin.

Each actin molecule in an actin filament is capable of binding one myosin-II head to form a complex that reveals the structural polarity of the actin filament. With negative staining, such complexes can be seen in the electron microscope to have a regular and distinctive form: each myosin head forms a lateral projection, and the superimposed image of many such projections gives the appearance of arrowheads along the actin filament. The pointed end created by these myosin arrows corresponds to the slow-growing minus end of the actin filament described earlier (see p. 821). The other, barbed end corresponds to the fast-growing plus end ( Figure16-87).

As shown in Figure16-88, myosin heads face in opposite directions on either side of the bare central region of a myosin-II filament. Since the heads must interact with actin filaments in the region of overlap, the actin filaments on either side of the sarcomere should be of opposite polarity. This has been demonstrated by using myosin-II heads to decorate the actin filaments attached to isolated Z discs. All of the myosin arrowheads are found to point away from the Z disc. Therefore, the plus end of each actin filament is embedded in the Z disc, while the minus end points toward the myosin filaments ( Figure16-89).

A Myosin Head "Walks" Toward the Plus End of an Actin Filament ⁵⁵

Muscle contraction is driven by the interaction between myosin-II heads and adjacent actin filaments, during which the myosin head hydrolyzes ATP. The ATP hydrolysis and subsequent dissociation of the tightly bound products (ADP and P_i) produce an ordered series of allosteric changes in the conformation of myosin. As a result, part of the energy released is coupled to the production of movement. A major advance in understanding these concerted changes in protein structure, and hence in understanding how ATP hydrolysis is coupled to directed movement of the myosin molecule, came with the determination of the three-dimensional structure of the myosin head by x-ray diffraction analysis ( Figure16-90). In conjunction with a wealth of other data, this structure suggests that unidi-rectional movement is generated by the sequence of events illustrated in Figure 16-91.

Because each turn of the cycle illustrated in Figure 16-91 results in the hydrolysis and release of one ATP molecule, the series of conformational changes just described is driven by a large favorable change in free energy, making it unidirectional. Each individual myosin head, therefore, "walks" in a single direction along an adjacent actin filament, always moving toward the filament's plus end (see Figure 16-89). As it undergoes its cyclical change in conformation, the myosin head pulls against the actin filament, causing this filament to slide against the myosin filament. Once an individual myosin head has detached from the actin filament, it is carried along by the action of other myosin heads in the same myosin filament, so that a snapshot of an entire myosin filament in a contracting muscle would show some of the myosin heads attached to actin filaments and others unattached. (A certain amount of springlike elasticity in the myosin molecule is essential to allow this to happen.) Each myosin filament has about 300 myosin heads (294 in frog muscle), and each head cycles about 5 times per second in the course of a rapid contraction - sliding the myosin and actin filaments past one another at rates of up to 15 µm/second.

Muscle Contraction Is Initiated by a Sudden Rise in Cytosolic Ca²⁺⁵⁶

The force-generating molecular interaction just described takes place only when a signal passes to the skeletal muscle from its motor nerve. The signal from the nerve triggers an action potential in the muscle cell plasma membrane, and this electrical excitation spreads rapidly into a series of membranous folds, the transverse tubules, or T tubules, that extend inward from the plasma membrane around each myofibril. The signal is then relayed across a small gap to the sarcoplasmic reticulum, an adjacent sheath of anastomosing flattened vesicles that surrounds each myofibril like a net stocking ( Figure16-92A).

In the junctional region, large Ca²⁺ release channels extend like pillars from the sarcoplasmic reticulum membrane to make contact with the T-tubule membrane on the other side ( Figure 16-92C). When voltage-sensitive proteins in the T-tubule membrane are activated by the incoming action potential, they trigger some of the Ca²⁺ release channels to open, probably by direct mechanical coupling. Ca²⁺ ions then escape from the sarcoplasmic reticulum (where they are stored in high concentration) into the cleft of the junction, causing more of the Ca²⁺ release channels to open, thereby amplifying the response. Ca²⁺ ions flooding into the cytosol then initiate the contraction of each myofibril.

Because the signal from the muscle-cell plasma membrane is passed within milliseconds (via the T tubules and sarcoplasmic reticulum) to every sarcomere in the cell, all of the myofibrils in the cell contract at the same time. The increase in Ca²⁺ concentration in the cytosol is transient because the Ca²⁺ is rapidly pumped back into the sarcoplasmic reticulum by an abundant Ca²⁺-ATPase in its membrane (discussed in Chapter 11). Typically, the cytosolic Ca²⁺ concentration is restored to resting levels within 30 milliseconds, causing the myofibrils to relax.

Troponin and Tropomyosin Mediate the Ca²⁺ Regulation of Skeletal Muscle Contraction ⁵⁷

The Ca²⁺ dependence of vertebrate skeletal muscle contraction, and hence its dependence on motor commands transmitted via nerves, is due entirely to a set of specialized accessory proteins closely associated with actin filaments. If myosin is mixed with pure actin filaments in a test tube, the ATPase activity of myosin is stimulated whether or not Ca²⁺ is present; in a normal myofibril, on the other hand, where the actin filaments are associated with accessory proteins, the stimulation of myosin ATPase activity depends on Ca²⁺.

One of these accessory proteins is a muscle form of tropomyosin, the rod-shaped molecule introduced earlier that binds in the groove of the actin helix (see Figure 16-78). The other major accessory protein involved in Ca²⁺ regulation in vertebrate skeletal muscle is troponin, a complex of three polypeptides - troponins T, I, and C (named for their Tropomyosin-binding, Inhibitory, and Calcium-binding activities). The troponin complex has an elongated shape, with subunits C and I forming a globular head region and T forming a long tail. The tail of troponin T binds to tropomyosin and is thought to be responsible for positioning the complex on the thin filament ( Figure16-93A). Troponin I binds to actin, and when it is added to troponin T and tropomyosin, the complex inhibits the interaction of actin and myosin, even in the presence of Ca²⁺.

The further addition of troponin C completes the troponin complex and makes its effects sensitive to Ca²⁺. Troponin C binds up to four molecules of Ca²⁺, and with Ca²⁺ bound, it relieves the inhibition of myosin binding to actin produced by the other two troponin components. Troponin C is closely related to calmodulin, which mediates Ca²⁺-signaled responses in all cells, including the activation of smooth muscle myosin. Troponin C may therefore be regarded as a specialized form of calmodulin that has evolved permanent binding sites for troponin I and troponin T, thereby ensuring that the myofibril responds extremely rapidly to an increase in Ca²⁺ concentration.

There is only one molecule of the troponin complex for every seven actin monomers in an actin filament (see Figure16-93A). Structural studies suggest that in a resting muscle the binding of troponin I to actin moves the tropomyosin molecules to a position on the actin filaments that in an actively contracting muscle is occupied by the myosin heads and thus inhibits the interaction of actin and myosin. When the level of Ca²⁺ is raised, troponin C causes the troponin I to release its hold on actin, thereby allowing the tropomyosin molecules to shift their position slightly so that the myosin heads can bind to the actin filament ( Figure 16-93B).

Other Accessory Proteins Maintain the Architecture of the Myofibril and Provide It with Elasticity ⁵⁸

The remarkable speed and power of muscle contraction depend on the filaments of actin and myosin in each myofibril being held at the optimal distance from one another and in correct alignment. More than a dozen structural proteins contribute to the precise architecture of the myofibril: the order in which they assemble, and the controls over this process, are important topics of contemporary research.

Actin filaments are anchored by their plus ends to the Z disc, where they are held in a square lattice arrangement by other proteins. One of the most important structural proteins in this region is α-actinin, the actin cross-linking protein discussed earlier that is abundant in most animal cells and is concentrated in the Z-disc region of the myofibril ( Figure 16-94). Myosin filaments are also held in a regular lattice - in this case a hexagonal one - through associated proteins that bind midway along the bipolar thick filaments.

Skeletal muscle cells contain two extraordinarily large proteins, called titin and nebulin, which form a network of fibers associated with the actin and myosin filaments. Titin, which has a molecular weight of 3 x 10⁶, is the largest polypeptide yet described. Stringlike titin molecules extend from the thick filaments to the Z disc; they are thought to act like springs to keep the myosin thick filaments centered in the sarcomere ( Figure 16-95). By contrast, nebulin, which is also large, is closely associated with the actin thin filaments and consists almost entirely of a repeating, 35-amino-acid actin-binding motif. The number of these motifs, and hence the total length of the nebulin molecule, is that needed to extend from one end of the actin filament to the other. Nebulin therefore could act as a "molecular ruler" to regulate the assembly of actin and the length of the actin filaments during muscle development (see Figure 3-52).

The myofibrils are bound to one another side by side by a system of desmin intermediate filaments, and the entire array is then anchored to the plasma membrane of the muscle cell by various proteins, including a flexible, elongated actin-binding protein called dystrophin. This protein, which is either absent or defective in patients with muscular dystrophy, has a close structural resemblance to spectrin, and it may link specific muscle membrane proteins to actin filaments in the myofibril.

The Same Contractile Machinery, in Modified Form, Is Found in Heart Muscle and Smooth Muscle ⁵⁹

Thus far we have described only one of the three major types of muscle present in vertebrates - skeletal muscle. The others are heart (cardiac) muscle, which contracts about 3 billion times in the course of an average human life-span, and smooth muscle, which produces the slower and longer-lasting contractions characteristic of organs such as the intestines. All three types of muscle cells, together with another class of contractile cells known as myoepithelial cells (see Figure22-36E), contract by an actin and myosin-II sliding filament mechanism.

Like skeletal muscle, heart muscle is striated, reflecting a very similar organization of actin filaments and myosin filaments. It is also triggered to contract by a similar mechanism: an action potential triggers the sarcoplasmic reticulum to release Ca²⁺, which activates contraction by means of a troponin-tropomyosin complex. Heart muscle cells, however, are not syncytial but are cells with a single nucleus. They are joined end to end by special structures called inter-calated discs ( Figure 16-96). The intercalated discs serve at least three functions. (1) They attach one cell to the next by means of desmosomes (discussed in Chapter 19). (2) They connect the actin filaments of the myofibrils of adjacent cells (performing a function analogous to that of the Z discs inside the cells). (3) They contain gap junctions, which allow an action potential to spread rapidly from one cell to the next, synchronizing the contractions of the heart muscle cells.

The most "primitive" muscle, in the sense of being most like nonmuscle cells, has no striations and is therefore called smooth muscle. It forms the contractile portion of the stomach, intestine, and uterus, the walls of arteries, and many other structures in which slow and sustained contractions are needed. It is composed of sheets of highly elongated spindle-shaped cells, each with a single nucleus. The cells contain both myosin-II and actin filaments, but these are not arranged in the strictly ordered pattern found in skeletal and cardiac muscle and do not form distinct myofibrils. Instead, the filaments form a more loosely arranged contractile apparatus, which is roughly aligned with the long axis of the cell - but is attached obliquely to the plasma membrane at disclike junctions connecting adjacent cells together.

Although the contractile apparatus in smooth muscle does not contract as rapidly as the myofibrils in a striated muscle cell, it has the advantage of permitting a much greater degree of shortening and therefore can produce large movements even though it lacks the leverage provided by attachments to bones. The organization of the actin filaments and myosin that makes this possible is poorly understood; one model is presented in Figure16-97.

The Activation of Myosin in Many Cells Depends on Myosin Light-Chain Phosphorylation ⁶⁰

The highly specialized contractile mechanisms that we have described in muscle cells evolved from the simpler force-generating mechanisms found in all eucaryotic cells. Not surprisingly, the myosin-II in nonmuscle cells most closely resembles the myosin-II in smooth muscle cells, the least specialized type of muscle. Contraction in smooth muscle cells is triggered by a rise in cytosolic Ca²⁺, but unlike the mechanism in skeletal and heart muscle, contraction is initiated mainly by phosphorylation of one of the two myosin-II light chains, which in turn controls the interaction of myosin with actin. A similar mechanism regulates nonmuscle myosin-II activity.

The two light chains on each head of the myosin-II molecule (see Figure5-23) are different, and one of them is phosphorylated during nonmuscle and smooth muscle contraction. When this light chain is phosphorylated, the myosin head can interact with an actin filament and thereby cause contraction; when it is dephosphorylated, the myosin head tends to dissociate from actin and becomes inactive. In both smooth muscle and nonmuscle cells the phosphorylation is catalyzed by the enzyme myosin light-chain kinase, whose action requires the binding of a Ca²⁺/calmodulin complex. As a result, contraction is controlled by the level of cytosolic Ca²⁺, as in cardiac and skeletal muscle ( Figure16-98).

Light-chain phosphorylation can also influence the state of aggregation of myosin-II molecules in the cell, as already mentioned in connection with motility in nonmuscle cells (see Figure 16-73). The phosphorylation of myosin-II occurs relatively slowly, so that even though it has assembled into a contractile bundle with actin, maximum contraction often requires nearly a second (compared with the few milliseconds required for a striated muscle cell). But rapid activation of contraction is not important in smooth muscle or nonmuscle cells: myosin-IIs in such cells hydrolyze ATP about 10 times more slowly than skeletal muscle myosin, producing a slow cross-bridge cycle and a slow contraction.

Summary

Muscle contraction is produced by the sliding of actin filaments against myosin filaments. The head regions of myosin molecules, which project from myosin filaments, engage in an ATP-driven cycle in which they attach to adjacent actin filaments, undergo a conformational change that pulls the myosin filament against the actin filament, and then detach. This cycle is facilitated by special accessory proteins in muscle that hold the actin and myosin filaments in parallel overlapping arrays with the correct orientation and spacing for sliding to occur. Two other accessory proteins - troponin and tropomyosin - allow the contraction of skeletal and cardiac muscle to be regulated by Ca²⁺.

In smooth muscle cells, and in most nonmuscle cells, actin and myosin produce contraction in fundamentally the same way as in skeletal and cardiac muscle. The contractile units are smaller, however, and less highly ordered in such cells; both their activity and their state of assembly are controlled by Ca²⁺-regulated phosphorylation of a myosin light chain.