Chapter 15:  Cell Communication

Extracellular Signal Molecules Bind to Specific Receptors

Yeast cells communicate with one another for mating by secreting a few kinds of small peptides. In contrast, cells in higher animals communicate by means of hundreds of kinds of signal molecules. These include proteins, small peptides, amino acids, nucleotides, steroids, retinoids, fatty acid derivatives, and even dissolved gases such as nitric oxide and carbon monoxide. Most of these signal molecules are secreted from the signaling cell into the extracellular space by exocytosis (discussed in Chapter 13). Others are released by diffusion through the plasma membrane, and some are exposed to the extracellular space while remaining tightly bound to the signaling cell's surface.

Regardless of the nature of the signal, the target cell responds by means of a specific protein called a receptor, which specifically binds the signal molecule and then initiates a response in the target cell. The extracellular signal molecules often act at very low concentrations (typically ≤ 10^-8 M), and the receptors that recognize them usually bind them with high affinity (affinity constant K_α ≥ 10⁸ liters/mole; see Figure 3-44). In most cases, these receptors are transmembrane proteins on the target cell surface. When they bind an extracellular signal molecule (a ligand), they become activated and generate a cascade of intracellular signals that alter the behavior of the cell. In other cases, the receptors are inside the target cell, and the signal molecule has to enter the cell to activate them: this requires that the signal molecules be sufficiently small and hydrophobic to diffuse across the plasma membrane (Figure 15-3).

Extracellular Signal Molecules Can Act Over Either Short or Long Distances

Many signal molecules remain bound to the surface of the signaling cell and influence only cells that contact it (Figure 15-4A). Such contact-dependent signaling is especially important during development and in immune responses. In most cases, however, signal molecules are secreted. The secreted molecules may be carried far afield to act on distant targets, or they may act as local mediators, affecting only cells in the immediate environment of the signaling cell. This latter process is called paracrine signaling (Figure 15-4B). For paracrine signals to be delivered only to their proper target cells, the secreted molecules must not be allowed to diffuse too far; for this reason they are often rapidly taken up by neighboring target cells, destroyed by extracellular enzymes, or immobilized by the extracellular matrix.

For a large, complex multicellular organism, short-range signaling is not sufficient on its own to coordinate the behavior of its cells. In these organisms, sets of specialized cells have evolved with a specific role in communication between widely separate parts of the body. The most sophisticated of these are nerve cells, or neurons, which typically extend long processes (axons) that enable them to contact target cells far away. When activated by signals from the environment or from other nerve cells, a neuron sends electrical impulses (action potentials) rapidly along its axon; when such an impulse reaches the end of the axon, it causes the nerve terminals located there to secrete a chemical signal called a neurotransmitter. These signals are secreted at specialized cell junctions called chemical synapses, which are designed to ensure that the neurotransmitter is delivered specifically to the postsynaptic target cell (Figure 15-4C). The details of this synaptic signaling process are discussed in Chapter 11.

A second type of specialized signaling cell that controls the behavior of the organism as a whole is an endocrine cell. These cells secrete their signal molecules, called hormones, into the bloodstream, which carries the signal to target cells distributed widely throughout the body (Figure 15-4D).

The mechanisms that allow endocrine cells and nerve cells to coordinate cell behavior in animals are compared in Figure 15-5. Because endocrine signaling relies on diffusion and blood flow, it is relatively slow. Synaptic signaling, by contrast, can be much faster, as well as more precise. Nerve cells can transmit information over long distances by electrical impulses that travel at rates of up to 100 meters per second; once released from a nerve terminal, a neurotransmitter has to diffuse less than 100 nm to the target cell, a process that takes less than a millisecond. Another difference between endocrine and synaptic signaling is that, whereas hormones are greatly diluted in the bloodstream and interstitial fluid and therefore must be able to act at very low concentrations (typically < 10^-8 M), neurotransmitters are diluted much less and can achieve high local concentrations. The concentration of acetylcholine in the synaptic cleft of an active neuromuscular junction, for example, is about 5 × 10^-4 M. Correspondingly, neurotransmitter receptors have a relatively low affinity for their ligand, which means that the neurotransmitter can dissociate rapidly from the receptor to terminate a response. Moreover, after its release from a nerve terminal, a neurotransmitter is quickly removed from the synaptic cleft, either by specific hydrolytic enzymes that destroy it or by specific membrane transport proteins that pump it back into either the nerve terminal or neighboring glial cells. Thus, synaptic signaling is much more precise than endocrine signaling, both in time and in space.

The speed of a response to an extracellular signal depends not only on the mechanism of signal delivery, but also on the nature of the response in the target cell. Where the response requires only changes in proteins already present in the cell, it can occur in seconds or even milliseconds. When the response involves changes in gene expression and the synthesis of new proteins, however, it usually requires hours, irrespective of the mode of signal delivery.

Autocrine Signaling Can Coordinate Decisions by Groups of Identical Cells

All of the forms of signaling discussed so far allow one cell to influence another. Often, the signaling cell and target are different cell types. Cells, however, can also send signals to other cells of the same type, as well as to themselves. In such autocrine signaling, a cell secretes signal molecules that can bind back to its own receptors. During development, for example, once a cell has been directed along a particular pathway of differentiation, it may begin to secrete autocrine signals to itself that reinforce this developmental decision.

Autocrine signaling is most effective when performed simultaneously by neighboring cells of the same type, and it is likely to be used to encourage groups of identical cells to make the same developmental decisions. Thus, autocrine signaling is thought to be one possible mechanism underlying the “community effect” that is observed in early development, during which a group of identical cells can respond to a differentiation-inducing signal but a single isolated cell of the same type cannot (Figure 15-6).

Unfortunately, cancer cells often use autocrine signaling to overcome the normal controls on cell proliferation and survival that we discuss later. By secreting signals that act back on the cell's own receptors, cancer cells can stimulate their own survival and proliferation and thereby survive and proliferate in places where normal cells of the same type could not. How this dangerous perturbation of normal cell behavior comes about is discussed in Chapter 23.

Gap Junctions Allow Signaling Information to Be Shared by Neighboring Cells

Another way to coordinate the activities of neighboring cells is through gap junctions. These are specialized cell-cell junctions that can form between closely apposed plasma membranes and directly connect the cytoplasms of the joined cells via narrow water-filled channels (see Figure 19-15). The channels allow the exchange of small intracellular signaling molecules (intracellular mediators), such as Ca²⁺ and cyclic AMP (discussed later), but not of macromolecules, such as proteins or nucleic acids. Thus, cells connected by gap junctions can communicate with each other directly, without having to surmount the barrier presented by the intervening plasma membranes (Figure 15-7).

As discussed in Chapter 19, the pattern of gap-junction connections in a tissue can be revealed either electrically, with intracellular electrodes, or visually, after the microinjection of small water-soluble dyes. Studies of this kind indicate that the cells in a developing embryo make and break gap-junction connections in specific and interesting patterns, strongly suggesting that these junctions have an important role in the signaling processes that occur between these cells. Mice and humans that are deficient in one particular gap-junction protein (connexin 43), for example, have severe defects in heart development. Like the autocrine signaling described above, gap-junction communication helps adjacent cells of a similar type to coordinate their behavior. It is still not known, however, which particular small molecules are important as carriers of signals through gap junctions, and the specific functions of gap-junction communication in animal development remain uncertain.

Each Cell Is Programmed to Respond to Specific Combinations of Extracellular Signal Molecules

A typical cell in a multicellular organism is exposed to hundreds of different signals in its environment. These signals can be soluble, bound to the extracellular matrix, or bound to the surface of a neighboring cell, and they can act in many millions of combinations. The cell must respond to this babel of signals selectively, according to its own specific character, which it has acquired through progressive cell specialization in the course of development. A cell may be programmed to respond to one combination of signals by differentiating, to another combination by multiplying, and to yet another by performing some specialized function such as contraction or secretion.

Most of the cells in a complex animal are also programmed to depend on a specific combination of signals simply to survive. When deprived of these signals (in a culture dish, for example), a cell activates a suicide program and kills itself—a process called programmed cell death, or apoptosis (Figure 15-8). Because different types of cells require different combinations of survival signals, each cell type is restricted to different environments in the body. The ability to undergo apoptosis is a fundamental property of animal cells, and it is discussed in Chapter 17.

In principle, the hundreds of signal molecules that animals make can be used to create an almost unlimited number of signaling combinations. The use of these combinations to control cell behavior enables an animal to control its cells in highly specific ways by using a limited diversity of signal molecules.

Different Cells Can Respond Differently to the Same Extracellular Signal Molecule

The specific way in which a cell reacts to its environment varies. It varies according to the set of receptor proteins the cell possesses, which determines the particular subset of signals it can respond to, and it varies according to the intracellular machinery by which the cell integrates and interprets the signals it receives (see Figure 15-1). Thus, a single signal molecule often has different effects on different target cells. The neurotransmitter acetylcholine, for example, stimulates the contraction of skeletal muscle cells, but it decreases the rate and force of contraction in heart muscle cells. This is because the acetylcholine receptor proteins on skeletal muscle cells are different from those on heart muscle cells. But receptor differences are not always the explanation for the different effects. In many cases, the same signal molecule binds to identical receptor proteins yet produces very different responses in different types of target cells, reflecting differences in the internal machinery to which the receptors are coupled (Figure 15-9).

The Concentration of a Molecule Can Be Adjusted Quickly Only If the Lifetime of the Molecule Is Short

It is natural to think of signaling systems in terms of the changes produced when a signal is delivered. But it is just as important to consider what happens when a signal is withdrawn. During development, transient signals often produce lasting effects: they can trigger a change in the cell's development that persists indefinitely, through cell memory mechanisms such as those discussed in Chapters 7 and 21. In most cases in adult tissues, however, the response fades when a signal ceases. The effect is transitory because the signal exerts its effects by altering a set of molecules that are unstable, undergoing continual turnover. Thus, once the signal is shut off, the replacement of the old molecules by new ones wipes out all traces of its action. It follows that the speed with which a cell responds to signal removal depends on the rate of destruction, or turnover, of the molecules the signal affects.

It is also true, although much less obvious, that this turnover rate also determines the promptness of the response when a signal is turned on. Consider, for example, two intracellular signaling molecules X and Y, both of which are normally maintained at a concentration of 1000 molecules per cell. Molecule Y is synthesized and degraded at a rate of 100 molecules per second, with each molecule having an average lifetime of 10 seconds. Molecule X has a turnover rate that is 10 times slower than that of Y: it is both synthesized and degraded at a rate of 10 molecules per second, so that each molecule has an average lifetime in the cell of 100 seconds. If a signal acting on the cell boosts the rates of synthesis of both X and Y tenfold without any change in the molecular lifetimes, at the end of 1 second the concentration of Y will have increased by nearly 900 molecules per cell (10 × 100 - 100), while the concentration of X will have increased by only 90 molecules per cell. In fact, after a molecule's synthesis rate has been either increased or decreased abruptly, the time required for the molecule to shift halfway from its old to its new equilibrium concentration is equal to its normal half-life—that is, equal to the time that would be required for its concentration to fall by half if all synthesis were stopped (Figure 15-10).

The same principles apply to proteins and small molecules, and to molecules in the extracellular space and inside cells. Many intracellular proteins have short half-lives, some surviving for less than 10 minutes. In most cases, these are proteins with key regulatory roles, whose concentrations are rapidly regulated in the cell by changes in their rates of synthesis. Likewise, any covalent modifications of proteins that occur as part of a rapid signaling process—most commonly, the addition of a phosphate group to an amino acid side chain—must be continuously removed at a rapid rate to make rapid signaling possible.

We shall discuss some of these molecular events in detail later for signaling pathways that operate via cell-surface receptors. But the principles apply quite generally, as the next example illustrates.

Nitric Oxide Gas Signals by Binding Directly to an Enzyme Inside the Target Cell

Although most extracellular signals are hydrophilic molecules that bind to receptors on the surface of the target cell, some signal molecules are hydrophobic enough and/or small enough to pass readily across the target-cell plasma membrane. Once inside, they directly regulate the activity of a specific intracellular protein. An important and remarkable example is the gas nitric oxide (NO), which acts as a signal molecule in both animals and plants. In mammals, one of its functions is to regulate smooth muscle contraction. Acetylcholine, for example, is released by autonomic nerves in the walls of a blood vessel, and it causes smooth muscle cells in the vessel wall to relax. The acetylcholine acts indirectly by inducing the nearby endothelial cells to make and release NO, which then signals the underlying smooth muscle cells to relax. This effect of NO on blood vessels provides an explanation for the mechanism of action of nitroglycerine, which has been used for about 100 years to treat patients with angina (pain resulting from inadequate blood flow to the heart muscle). The nitroglycerine is converted to NO, which relaxes blood vessels. This reduces the workload on the heart and, as a consequence, it reduces the oxygen requirement of the heart muscle.

Many types of nerve cells use NO gas to signal to their neighbors. The NO released by autonomic nerves in the penis, for example, causes the local blood vessel dilation that is responsible for penile erection. NO is also produced as a local mediator by activated macrophages and neutrophils to help them to kill invading microorganisms. In plants, NO is involved in the defensive responses to injury or infection.

NO gas is made by the deamination of the amino acid arginine, catalyzed by the enzyme NO synthase. Because it passes readily across membranes, dissolved NO rapidly diffuses out of the cell where it is produced and into neighboring cells. It acts only locally because it has a short half-life—about 5–10 seconds—in the extracellular space before it is converted to nitrates and nitrites by oxygen and water. In many target cells, including endothelial cells, NO binds to iron in the active site of the enzyme guanylyl cyclase, stimulating this enzyme to produce the small intracellular mediator cyclic GMP, which we discuss later (Figure 15-11). The effects of NO can occur within seconds, because the normal rate of turnover of cyclic GMP is high: a rapid degradation to GMP by a phosphodiesterase constantly balances the production of cyclic GMP from GTP by guanylyl cyclase. The drug Viagra inhibits this cyclic GMP phosphodiesterase in the penis, thereby increasing the amount of time that cyclic GMP levels remain elevated after NO production is induced by local nerve terminals. The cyclic GMP, in turn, keeps blood vessels relaxed and the penis erect.

Carbon monoxide (CO) is another gas that is used as an intercellular signal. It can act in the same way as NO, by stimulating guanylyl cyclase. These gases are not the only signal molecules that can pass directly across the target-cell plasma membrane. A group of small, hydrophobic, nongaseous hormones and local mediators also enter target cells in this way. But instead of binding to enzymes, they bind to intracellular receptor proteins that directly regulate gene transcription, as we discuss next.

Nuclear Receptors Are Ligand-activated Gene Regulatory Proteins

A number of small hydrophobic signal molecules diffuse directly across the plasma membrane of target cells and bind to intracellular receptor proteins. These signal molecules include steroid hormones, thyroid hormones, retinoids, and vitamin D. Although they differ greatly from one another in both chemical structure (Figure 15-12) and function, they all act by a similar mechanism. When these signal molecules bind to their receptor proteins, they activate the receptors, which bind to DNA to regulate the transcription of specific genes. The receptors are all structurally related, being part of the nuclear receptor superfamily. This very large superfamily also includes some receptor proteins that are activated by intracellular metabolites rather than by secreted signal molecules. Many family members have been identified by DNA sequencing only, and their ligand is not yet known; these proteins are therefore referred to as orphan nuclear receptors. The importance of such nuclear receptors in some animals is indicated by the fact that 1–2% of the genes in the nematode C. elegans code for them, although there are fewer than 50 in humans (see Figure 7-114).

Steroid hormones—which include cortisol, the steroid sex hormones, vitamin D (in vertebrates), and the moulting hormone ecdysone (in insects)—are all made from cholesterol. Cortisol is produced in the cortex of the adrenal glands and influences the metabolism of many types of cells. The steroid sex hormones are made in the testes and ovaries, and they are responsible for the secondary sex characteristics that distinguish males from females. Vitamin D is synthesized in the skin in response to sunlight; after it has been converted to its active form in the liver or kidneys, it regulates Ca²⁺ metabolism, promoting Ca²⁺ uptake in the gut and reducing its excretion in the kidneys. The thyroid hormones, which are made from the amino acid tyrosine, act to increase the metabolic rate in a wide variety of cell types, while the retinoids, such as retinoic acid, are made from vitamin A and have important roles as local mediators in vertebrate development. Although all of these signal molecules are relatively insoluble in water, they are made soluble for transport in the bloodstream and other extracellular fluids by binding to specific carrier proteins, from which they dissociate before entering a target cell (see Figure 15-3).

Beside a fundamental difference in the way they signal their target cells, most water-insoluble signal molecules differ from water-soluble ones in the length of time they persist in the bloodstream or tissue fluids. Most water-soluble hormones are removed and/or broken down within minutes of entering the blood, and local mediators and neurotransmitters are removed from the extracellular space even faster—within seconds or milliseconds. Steroid hormones, by contrast, persist in the blood for hours and thyroid hormones for days. Consequently, water-soluble signal molecules usually mediate responses of short duration, whereas water-insoluble ones tend to mediate responses that are longer lasting.

The intracellular receptors for the steroid and thyroid hormones, retinoids, and vitamin D all bind to specific DNA sequences adjacent to the genes the ligand regulates. Some receptors, such as those for cortisol, are located primarily in the cytosol and enter the nucleus after ligand binding; others, such as the thyroid and retinoid receptors, are bound to DNA in the nucleus even in the absence of ligand. In either case, the inactive receptors are bound to inhibitory protein complexes, and ligand binding alters the conformation of the receptor protein, causing the inhibitory complex to dissociate. The ligand binding also causes the receptor to bind to coactivator proteins that induce gene transcription (Figure 15-13). The transcriptional response usually takes place in successive steps: the direct activation of a small number of specific genes occurs within about 30 minutes and constitutes the primary response; the protein products of these genes in turn activate other genes to produce a delayed, secondary response; and so on. In this way, a simple hormonal trigger can cause a very complex change in the pattern of gene expression (Figure 15-14).

The responses to steroid and thyroid hormones, vitamin D, and retinoids, like responses to extracellular signals in general, are determined as much by the nature of the target cell as by the nature of the signal molecule. Many types of cells have the identical intracellular receptor, but the set of genes that the receptor regulates is different in each cell type. This is because more than one type of gene regulatory protein generally must bind to a eucaryotic gene to activate its transcription. An intracellular receptor can therefore activate a gene only if there is the right combination of other gene regulatory proteins, and many of these are cell-type specific. Thus, each of these hormones induces a characteristic set of responses in an animal for two reasons. First, only certain types of cells have receptors for it. Second, each of these cell types contains a different combination of other cell-type-specific gene regulatory proteins that collaborate with the activated receptor to influence the transcription of specific sets of genes.

The molecular details of how nuclear receptors and other gene regulatory proteins control specific gene transcription are discussed in Chapter 7.

The Three Largest Classes of Cell-Surface Receptor Proteins Are Ion-Channel-linked, G-Protein-linked, and Enzyme-linked Receptors

As mentioned previously, all water-soluble signal molecules (including neurotransmitters and all signal proteins) bind to specific receptor proteins on the surface of the target cells that they influence. These cell-surface receptor proteins act as signal transducers. They convert an extracellular ligand-binding event into intracellular signals that alter the behavior of the target cell.

Most cell-surface receptor proteins belong to one of three classes, defined by the transduction mechanism they use. Ion-channel-linked receptors, also known as transmitter-gated ion channels or ionotropic receptors, are involved in rapid synaptic signaling between electrically excitable cells (Figure 15-15A). This type of signaling is mediated by a small number of neurotransmitters that transiently open or close an ion channel formed by the protein to which they bind, briefly changing the ion permeability of the plasma membrane and thereby the excitability of the postsynaptic cell. The ion-channel-linked receptors belong to a large family of homologous, multipass transmembrane proteins. Because they are discussed in detail in Chapter 11, we shall not consider them further here.

G-protein-linked receptors act indirectly to regulate the activity of a separate plasma-membrane-bound target protein, which can be either an enzyme or an ion channel. The interaction between the receptor and this target protein is mediated by a third protein, called a trimeric GTP-binding protein (G protein) (Figure 15-15B). The activation of the target protein can change the concentration of one or more intracellular mediators (if the target protein is an enzyme), or it can change the ion permeability of the plasma membrane (if the target protein is an ion channel). The intracellular mediators affected act in turn to alter the behavior of yet other signaling proteins in the cell. All of the G-protein-linked receptors belong to a large family of homologous, seven-pass transmembrane proteins.

Enzyme-linked receptors, when activated, either function directly as enzymes or are directly associated with enzymes that they activate (Figure 15-15C). They are formed by single-pass transmembrane proteins that have their ligand-binding site outside the cell and their catalytic or enzyme-binding site inside. Enzyme-linked receptors are heterogeneous in structure compared with the other two classes. The great majority, however, are protein kinases, or are associated with protein kinases, and ligand binding to them causes the phosphorylation of specific sets of proteins in the target cell.

There are some cell-surface receptors that do not fit into any of the above classes. Some of these depend on intracellular proteolytic events to signal the cell, and we discuss them only after we explain in detail how G-protein-linked receptors and enzyme-linked receptors operate. We start with some general principles of signaling via cell-surface receptors.

Most Activated Cell-Surface Receptors Relay Signals Via Small Molecules and a Network of Intracellular Signaling Proteins

Signals received at the surface of a cell by either G-protein-linked or enzyme-linked receptors are relayed into the cell interior by a combination of small and large intracellular signaling molecules. The resulting chain of intracellular signaling events ultimately alters target proteins, and these altered target proteins are responsible for modifying the behavior of the cell (see Figure 15-1).

The small intracellular signaling molecules are called small intracellular mediators, or second messengers (the “first messengers” being the extracellular signals). They are generated in large numbers in response to receptor activation and rapidly diffuse away from their source, broadcasting the signal to other parts of the cell. Some, such as cyclic AMP and Ca ²⁺ , are water-soluble and diffuse in the cytosol, while others, such as diacylglycerol, are lipid-soluble and diffuse in the plane of the plasma membrane. In either case, they pass the signal on by binding to and altering the behavior of selected signaling proteins or target proteins.

The large intracellular signaling molecules are intracellular signaling proteins. Many of these relay the signal into the cell by either activating the next signaling protein in the chain or generating small intracellular mediators. These proteins can be classified according to their particular function, although many fall into more than one category (Figure 15-16):

• 1

  Relay proteins simply pass the message to the next signaling component in the chain.

• 2

  Messenger proteins carry the signal from one part of the cell to another, such as from the cytosol to the nucleus.

• 3

  Adaptor proteins link one signaling protein to another, without themselves conveying a signal.

• 4

  Amplifier proteins, which are usually either enzymes or ion channels, greatly increase the signal they receive, either by producing large amounts of small intracellular mediators or by activating large numbers of downstream intracellular signaling proteins. When there are multiple amplification steps in a relay chain, the chain is often referred to as a signaling cascade.

• 5

  Transducer proteins convert the signal into a different form. The enzyme that makes cyclic AMP is an example: it both converts the signal and amplifies it, thus acting as both a transducer and an amplifier.

• 6

  Bifurcation proteins spread the signal from one signaling pathway to another.

• 7

  Integrator proteins receive signals from two or more signaling pathways and integrate them before relaying a signal onward.

• 8

  Latent gene regulatory proteins are activated at the cell surface by activated receptors and then migrate to the nucleus to stimulate gene transcription.

As shown in blue in Figure 15-16, other types of intracellular proteins also have important roles in intracellular signaling. Modulator proteins modify the activity of intracellular signaling proteins and thereby regulate the strength of signaling along the pathway. Anchoring proteins maintain specific signaling proteins at a precise location in the cell by tethering them to a membrane or the cytoskeleton. Scaffold proteins are adaptor and/or anchoring proteins that bind multiple signaling proteins together in a functional complex and often hold them at a specific location.

Some Intracellular Signaling Proteins Act as Molecular Switches

Many intracellular signaling proteins behave like molecular switches: on receipt of a signal they switch from an inactive to an active state, until another process switches them off. As we discussed earlier, the switching off is just as important as the switching on. If a signaling pathway is to recover after transmitting a signal so that it can be ready to transmit another, every activated molecule in the pathway must be returned to its original inactivated state.

The molecular switches fall into two main classes that operate in different ways, although in both cases it is the gain or loss of phosphate groups that determines whether the protein is active or inactive. The largest class consists of proteins that are activated or inactivated by phosphorylation (discussed in Chapter 3). For these proteins, the switch is thrown in one direction by a protein kinase, which adds one or more phosphate groups to the signaling protein, and in the other direction by a protein phosphatase, which removes the phosphate groups from the protein (Figure 15-17A). It is estimated that one-third of the proteins in a eucaryotic cell are phosphorylated at any given time.

Many of the signaling proteins controlled by phosphorylation are themselves protein kinases, and these are often organized into phosphorylation cascades. One protein kinase, activated by phosphorylation, phosphorylates the next protein kinase in the sequence, and so on, relaying the signal onward and, in the process, amplifying it and sometimes spreading it to other signaling pathways. Two main types of protein kinases operate as intracellular signaling proteins. The great majority are serine/threonine kinases, which phosphorylate proteins on serines and (less often) threonines. Others are tyrosine kinases, which phosphorylate proteins on tyrosines. An occasional kinase can do both. Genome sequencing reveals that about 2% of our genes encode protein kinases, and it is thought that hundreds of distinct types of protein kinases are present in a typical mammalian cell.

The other main class of molecular switches involved in signaling are GTP-binding proteins (discussed in Chapter 3). These switch between an active state when GTP is bound and an inactive state when GDP is bound. Once activated, they have intrinsic GTPase activity and shut themselves off by hydrolyzing their bound GTP to GDP (Figure 15-17B). There are two major types of GTP-binding proteins—large trimeric GTP-binding proteins (also called G proteins), which relay the signals from G-protein-linked receptors (see Figure 15-15B), and small monomeric GTPases (also called monomeric GTP-binding proteins). The latter also help to relay intracellular signals, but in addition they are involved in regulating vesicular traffic and many other processes in eucaryotic cells.

As discussed earlier, complex cell behaviors, such as cell survival and cell proliferation, are generally stimulated by specific combinations of extracellular signals rather than by a single signal acting alone (see Figure 15-8). The cell therefore has to integrate the information coming from separate signals so as to make an appropriate response—to live or die, to divide or not, and so on. This integration usually depends on integrator proteins (see Figure 15-16), which are equivalent to the microprocessors in a computer: they require multiple signal inputs to produce an output that causes the desired biological effect. Two examples that show how such integrator proteins can operate are illustrated in Figure 15-18.

Intracellular Signaling Complexes Enhance the Speed, Efficiency, and Specificity of the Response

Even a single type of extracellular signal acting through a single type of G-protein-linked or enzyme-linked receptor usually activates multiple parallel signaling pathways and can thereby influence multiple aspects of cell behavior—such as shape, movement, metabolism, and gene expression. Indeed, these two main classes of cell-surface receptors often activate some of the same signaling pathways, and there is usually no obvious reason why a particular extracellular signal utilizes one class of receptors rather than the other.

The complexity of these signal-response systems, with multiple interacting relay chains of signaling proteins, is daunting. It is not clear how an individual cell manages to display specific responses to so many different extracellular signals, many of which bind to the same class of receptor and activate many of the same signaling pathways. One strategy that the cell uses to achieve specificity involves scaffold proteins (see Figure 15-16), which organize groups of interacting signaling proteins into signaling complexes (Figure 15-19A). Because the scaffold guides the interactions between the successive components in such a complex, the signal can be relayed with precision, speed, and efficiency; moreover, unwanted cross-talk between signaling pathways is avoided. In order to amplify a signal, however, and spread it to other parts of the cell, at least some of the components in most signaling pathways are likely to be freely diffusible.

In other cases, signaling complexes form only transiently, as when signaling proteins assemble around a receptor after an extracellular signal molecule has activated it. In some of these cases, the cytoplasmic tail of the activated receptor is phosphorylated during the activation process, and the phosphorylated amino acids then serve as docking sites for the assembly of other signaling proteins (Figure 15-19B). In yet other cases, receptor activation leads to the production of modified phospholipid molecules in the adjacent plasma membrane, and these lipids then recruit specific intracellular signaling proteins to this region of membrane. All such signaling complexes form only transiently and rapidly disassemble after the extracellular ligand dissociates from the receptor.

Interactions Between Intracellular Signaling Proteins Are Mediated by Modular Binding Domains

The assembly of both stable and transient signaling complexes depends on a variety of highly conserved, small binding domains that are found in many intracellular signaling proteins. Each of these compact protein modules binds to a particular structural motif in the protein (or lipid) with which the signaling protein interacts. Because of these modular domains, signaling proteins bind to one another in multiple combinations, like Lego bricks, with the proteins often forming a three-dimensional network of interactions that determines the route followed by the signaling pathway. By joining existing domains together in novel combinations, the use of such modular binding domains has presumably facilitated the rapid evolution of new signaling pathways.

Src homology 2 (SH2) domains and phosphotyrosine-binding (PTB) domains, for example, bind to phosphorylated tyrosines in a particular peptide sequence on activated receptors or intracellular signaling proteins. Src homology 3 (SH3) domains bind to a short proline-rich amino acid sequence. Pleckstrin homology (PH) domains (first described in the Pleckstrin protein in blood platelets) bind to the charged headgroups of specific phosphorylated inositol phospholipids that are produced in the plasma membrane in response to an extracellular signal; they thereby enable the protein they are part of to dock on the membrane and interact with other recruited signaling proteins. Some signaling proteins function only as adaptors to link two other proteins together in a signaling pathway, and they consist solely of two or more binding domains (Figure 15-20).

Scaffold proteins often contain multiple PDZ domains (originally found in a region of a synapse called the postsynaptic density), each of which binds to a specific motif on a receptor or signaling protein. The InaD scaffold protein in Drosophila photoreceptor cells is a striking example. It contains five PDZ domains, one of which binds a light-activated ion channel, while the others each bind to a different signaling protein involved in the response of the cell to light. If any of these PDZ domains are missing, the corresponding signaling protein fails to assemble in the complex, and the fly's vision is defective.

Some cell-surface receptors and intracellular signaling proteins are thought to cluster together transiently in specific microdomains in the lipid bilayer of the plasma membrane that are enriched in cholesterol and glycolipids. Some of the proteins are directed to these lipid rafts by covalently attached lipid molecules. Like scaffold proteins, these lipid scaffolds may promote speed and efficiency in the signaling process by serving as sites where signaling molecules can assemble and interact (see Figure 10-13).

Cells Can Respond Abruptly to a Gradually Increasing Concentration of an Extracellular Signal

Some cellular responses to extracellular signal molecules are smoothly graded in simple proportion to the concentration of the molecule. The primary responses to steroid hormones (see Figure 15-14) often follow this pattern, presumably because the nuclear hormone receptor protein binds a single molecule of hormone and each specific DNA recognition sequence in a steroid-hormone-responsive gene acts independently. As the concentration of hormone increases, the concentration of activated receptor-hormone complexes increases proportionally, as does the number of complexes bound to specific recognition sequences in the responsive genes; the response of the cell is therefore a gradual and linear one.

Many responses to extracellular signal molecules, however, begin more abruptly as the concentration of the molecule increases. Some may even occur in a nearly all-or-none manner, being undetectable below a threshold concentration of the molecule and then reaching a maximum as soon as this concentration is exceeded. What might be the molecular basis for such steep or even switchlike responses to graded signals?

One mechanism for sharpening the response is to require that more than one intracellular effector molecule or complex bind to some target macromolecule to induce a response. In some steroid-hormone-induced responses, for example, it seems that more than one activated receptor-hormone complex must bind simultaneously to specific regulatory sequences in the DNA to activate a particular gene. As a result, as the hormone concentration rises, gene activation begins more abruptly than it would if only one bound complex were sufficient for activation (Figure 15-21). A similar cooperative mechanism often operates in the signaling cascades activated by cell-surface receptors. As we discuss later, four molecules of the small intracellular mediator cyclic AMP, for example, must bind to each molecule of cyclic-AMP-dependent protein kinase to activate the kinase. Such responses become sharper as the number of cooperating molecules increases, and if the number is large enough, responses approaching the all-or-none type can be achieved (Figures 15-22 and 15-23).

Responses are also sharpened when an intracellular signaling molecule activates one enzyme and, at the same time, inhibits another enzyme that catalyzes the opposite reaction. A well-studied example of this common type of regulation is the stimulation of glycogen breakdown in skeletal muscle cells induced by the hormone adrenaline (epinephrine). Adrenaline's binding to a G-protein-linked cell-surface receptor leads to an increase in intracellular cyclic AMP concentration, which both activates an enzyme that promotes glycogen breakdown and inhibits an enzyme that promotes glycogen synthesis.

All of these mechanisms can produce responses that are very steep but, nevertheless, always smoothly graded according to the concentration of the extracellular signal molecule. Another mechanism, however, can produce true all-or-none responses, such that raising the signal above a critical threshold level trips a sudden switch in the responding cell. All-or-none threshold responses of this type generally depend on positive feedback; by this mechanism, nerve and muscle cells generate all-or-none action potentials in response to neurotransmitters (discussed in Chapter 11). The activation of ion-channel-linked acetylcholine receptors at a neuromuscular junction, for example, results in a net influx of Na⁺ that locally depolarizes the muscle plasma membrane. This causes voltage-gated Na⁺ channels to open in the same membrane region, producing a further influx of Na⁺, which further depolarizes the membrane and thereby opens more Na⁺ channels. If the initial depolarization exceeds a certain threshold value, this positive feedback has an explosive “runaway” effect, producing an action potential that propagates to involve the entire muscle membrane.

An accelerating positive feedback mechanism can also operate through signaling proteins that are enzymes rather than ion channels. Suppose, for example, that a particular intracellular signaling ligand activates an enzyme located downstream in a signaling pathway and that two or more molecules of the product of the enzymatic reaction bind back to the same enzyme to activate it further (Figure 15-24). The consequence is a very low rate of synthesis of the product in the absence of the ligand. The rate increases slowly with the concentration of ligand until, at some threshold level of ligand, enough of the product has been synthesized to activate the enzyme in a self-accelerating, runaway fashion. The concentration of the product then suddenly increases to a much higher level. Through these and a number of other mechanisms not discussed here, the cell will often translate a gradual change in the concentration of a signaling ligand into a switchlike change, creating an all-or-none response by the cell.

A Cell Can Remember The Effect of Some Signals

The effect of an extracellular signal on a target cell can, in some cases, persist well after the signal has disappeared. The enzymatic accelerating positive feedback system just described represents one type of mechanism that displays this kind of persistence. If such a system has been switched on by raising the concentration of intracellular activating ligand above threshold, it will generally remain switched on even when the extracellular signal disappears; instead of faithfully reflecting the current level of signal, the response system displays a memory. We shall encounter a specific example of this later, when we discuss a protein kinase that is activated by Ca²⁺ to phosphorylate itself and other proteins; the autophosphorylation keeps the kinase active long after Ca²⁺ levels return to normal, providing a memory trace of the initial signal.

Transient extracellular signals often induce much longer-term changes in cells during the development of a multicellular organism. Some of these changes can persist for the lifetime of the organism. They usually depend on self-activating memory mechanisms that operate further downstream in a signaling pathway, at the level of gene transcription. The signals that trigger muscle cell determination, for example, turn on a series of muscle-specific gene regulatory proteins that stimulate the transcription of their own genes, as well as genes producing many other muscle cell proteins. In this way, the decision to become a muscle cell is made permanent (see Figure 7-72B).

Cells Can Adjust Their Sensitivity to a Signal

In responding to many types of stimuli, cells and organisms are able to detect the same percentage of change in a signal over a very wide range of stimulus intensities. This requires that the target cells undergo a reversible process of adaptation, or desensitization, whereby a prolonged exposure to a stimulus decreases the cells' response to that level of exposure. In chemical signaling, adaptation enables cells to respond to changes in the concentration of a signaling ligand (rather than to the absolute concentration of the ligand) over a very wide range of ligand concentrations. The general principle is one of a negative feedback that operates with a delay. A strong response modifies the machinery for making that response, such that the machinery resets itself to an off position. Owing to the delay, however, a sudden change in the stimulus is able to make itself felt strongly for a short period before the negative feedback has time to kick in.

Desensitization to a signal molecule can occur in various ways. Ligand binding to cell-surface receptors, for example, may induce their endocytosis and temporary sequestration in endosomes. Such ligand-induced receptor endocytosis can lead to the destruction of the receptors in lysosomes, a process referred to as receptor down-regulation. In other cases, desensitization results from a rapid inactivation of the receptors—for example, as a result of a receptor phosphorylation that follows its activation, with a delay. Desensitization can also be caused by a change in a protein involved in transducing the signal or by the production of an inhibitor that blocks the transduction process (Figure 15-25).

Having discussed some of the general principles of cell signaling, we now turn to the G-protein-linked receptors. These are by far the largest class of cell-surface receptors, and they mediate the responses to the great majority of extracellular signals. This superfamily of receptor proteins not only mediates intercellular communication; it is also central to vision, smell, and taste perception.

Summary

Each cell in a multicellular animal has been programmed during development to respond to a specific set of extracellular signals produced by other cells. These signals act in various combinations to regulate the behavior of the cell. Most of the signals mediate a form of signaling in which local mediators are secreted, but then are rapidly taken up, destroyed, or immobilized, so that they act only on neighboring cells. Other signals remain bound to the outer surface of the signaling cell and mediate contact-dependent signaling. Centralized control is exerted both by endocrine signaling, in which hormones secreted by endocrine cells are carried in the blood to target cells throughout the body, and by synaptic signaling, in which neurotransmitters secreted by nerve cell axons act locally on the postsynaptic cells that the axons contact.

Cell signaling requires not only extracellular signal molecules, but also a complementary set of receptor proteins in each cell that enable it to bind and respond to the signal molecules in a characteristic way. Some small hydrophobic signal molecules, including steroid and thyroid hormones, diffuse across the plasma membrane of the target cell and activate intracellular receptor proteins that directly regulate the transcription of specific genes. The dissolved gases nitric oxide and carbon monoxide act as local mediators by diffusing across the plasma membrane of the target cell and activating an intracellular enzyme—usually guanylyl cyclase, which produces cyclic GMP in the target cell. But most extracellular signal molecules are hydrophilic and can activate receptor proteins only on the surface of the target cell; these receptors act as signal transducers, converting the extracellular binding event into intracellular signals that alter the behavior of the target cell.

There are three main families of cell-surface receptors, each of which transduces extracellular signals in a different way. Ion-channel-linked receptors are transmitter-gated ion channels that open or close briefly in response to the binding of a neurotransmitter. G-protein-linked receptors indirectly activate or inactivate plasma-membrane-bound enzymes or ion channels via trimeric GTP-binding proteins (G proteins). Enzyme-linked receptors either act directly as enzymes or are associated with enzymes; these enzymes are usually protein kinases that phosphorylate specific proteins in the target cell.

Once activated, enzyme- and G-protein-linked receptors relay a signal into the cell interior by activating chains of intracellular signaling proteins; some transduce, amplify, or spread the signal as they relay it, while others integrate signals from different signaling pathways. Many of these signaling proteins function as switches that are transiently activated by phosphorylation or GTP binding. Functional signaling complexes are often formed by means of modular binding domains in the signaling proteins; these domains allow complicated protein assemblies to function in signaling networks.

Target cells can use a variety of intracellular mechanisms to respond abruptly to a gradually increasing concentration of an extracellular signal or to convert a short-lasting signal into a long-lasting response. In addition, through adaptation, they can often reversibly adjust their sensitivity to a signal to allow the cells to respond to changes in the concentration of a particular signal molecule over a large range of concentrations.

Trimeric G Proteins Disassemble to Relay Signals from G-Protein-linked Receptors

When extracellular signaling molecules bind to serpentine receptors, the receptors undergo a conformational change that enables them to activate trimeric GTP-binding proteins (G proteins). These G proteins are attached to the cytoplasmic face of the plasma membrane, where they serve as relay molecules, functionally coupling the receptors to enzymes or ion channels in this membrane. There are various types of G proteins, each specific for a particular set of serpentine receptors and for a particular set of downstream target proteins in the plasma membrane. All have a similar structure, however, and they operate in a similar way.

G proteins are composed of three protein subunits—α, β, and γ. In the unstimulated state, the α subunit has GDP bound and the G protein is inactive (Figure 15-27). When stimulated by an activated receptor, the α subunit releases its bound GDP, allowing GTP to bind in its place. This exchange causes the trimer to dissociate into two activated components—an α subunit and a βγ complex (Figure 15-28).

The dissociation of the trimeric G protein activates its two components in different ways. GTP binding causes a conformational change that affects the surface of the α subunit that associates with the βγ complex in the trimer. This change causes the release of the βγ complex, but it also causes and the α subunit to adopt a new shape that allows it to interact with its target proteins. The βγ complex does not change its conformation, but the surface previously masked by the α subunit is now available to interact with a second set of target proteins. The targets of the dissociated components of the G protein are either enzymes or ion channels in the plasma membrane, and they relay the signal onward.

The α subunit is a GTPase, and once it hydrolyzes its bound GTP to GDP, it reassociates with a βγ complex to re-form an inactive G protein, reversing the activation process (Figure 15-29). The time during which the α subunit and βγ complex remain apart and active is usually short, and it depends on how quickly the α subunit hydrolyzes its bound GTP. An isolated α subunit is an inefficient GTPase, and, left to its own devices, the subunit would inactivate only after several minutes. Its activation is usually reversed much faster than this, however, because the GTPase activity of the α subunit is greatly enhanced by the binding of a second protein, which can be either its target protein or a specific modulator known as a regulator of G protein signaling (RGS). RGS proteins act as α-subunit-specific GTPase activating proteins (GAPs), and they are thought to have a crucial role in shutting off G-protein-mediated responses in all eucaryotes. There are about 25 RGS proteins encoded in the human genome, each of which is thought to interact with a particular set of G proteins.

The importance of the GTPase activity in shutting off the response can be easily demonstrated in a test tube. If cells are broken open and exposed to an analogue of GTP (GTPγS) in which the terminal phosphate cannot be hydrolyzed, the activated α subunits remain active for a very long time.

Some G Proteins Signal By Regulating the Production of Cyclic AMP

Cyclic AMP (cAMP) was first identified as a small intracellular mediator in the 1950s. It has since been found to act in this role in all procaryotic and animal cells that have been studied. The normal concentration of cyclic AMP inside the cell is about 10^-7 M, but an extracellular signal can cause cyclic AMP levels to change by more than twentyfold in seconds (Figure 15-30). As explained earlier (see Figure 15-10), such a rapid response requires that a rapid synthesis of the molecule be balanced by its rapid breakdown or removal. In fact, cyclic AMP is synthesized from ATP by a plasma-membrane-bound enzyme adenylyl cyclase, and it is rapidly and continuously destroyed by one or more cyclic AMP phosphodiesterases that hydrolyze cyclic AMP to adenosine 5′-monophosphate (5′-AMP) (Figure 15-31).

Many extracellular signal molecules work by increasing cyclic AMP content, and they do so by increasing the activity of adenylyl cyclase rather than decreasing the activity of phosphodiesterase. Adenylyl cyclase is a large multipass transmembrane protein with its catalytic domain on the cytosolic side of the plasma membrane. There are at least eight isoforms in mammals, most of which are regulated by both G proteins and Ca²⁺. All receptors that act via cyclic AMP are coupled to a stimulatory G protein (G_s), which activates adenylyl cyclase and thereby increases cyclic AMP concentration. Another G protein, called inhibitory G protein (G_i), inhibits adenylyl cyclase, but it mainly acts by directly regulating ion channels (as we discuss later) rather than by decreasing cyclic AMP content. Although it is usually the α subunit that regulates the cyclase, the βγ complex sometimes does so as well, either increasing or decreasing the enzyme's activity, depending on the particular βγ complex and the isoform of the cyclase.

Both G_s and G_i are targets for some medically important bacterial toxins. Cholera toxin, which is produced by the bacterium that causes cholera, is an enzyme that catalyzes the transfer of ADP ribose from intracellular NAD⁺ to the α subunit of G_s. This ADP ribosylation alters the α subunit so that it can no longer hydrolyze its bound GTP, causing it to remain in an active state that stimulates adenylyl cyclase indefinitely. The resulting prolonged elevation in cyclic AMP levels within intestinal epithelial cells causes a large efflux of Cl^- and water into the gut, thereby causing the severe diarrhea that characterizes cholera. Pertussis toxin, which is made by the bacterium that causes pertussis (whooping cough), catalyzes the ADP ribosylation of the α subunit of G_i, preventing the subunit from interacting with receptors; as a result, this α subunit retains its bound GDP and is unable to regulate its target proteins. These two toxins are widely used as tools to determine whether a cell's response to a signal is mediated by G_s or by G_i.

Some of the responses mediated by a G_s-stimulated increase in cyclic AMP concentration are listed in Table 15-1. It is clear that different cell types respond differently to an increase in cyclic AMP concentration, and that any one cell type usually responds in the same way, even if different extracellular signals induce the increase. At least four hormones activate adenylyl cyclase in fat cells, for example, and all of them stimulate the breakdown of triglyceride (the storage form of fat) to fatty acids.

Individuals who are genetically deficient in a particular G_s α subunit show decreased responses to certain hormones. As a consequence, they display metabolic abnormalities, have abnormal bone development, and are mentally retarded.

Cyclic-AMP-dependent Protein Kinase (PKA) Mediates Most of the Effects of Cyclic AMP

Although cyclic AMP can directly activate certain types of ion channels in the plasma membrane of some highly specialized cells, in most animal cells it exerts its effects mainly by activating cyclic-AMP-dependent protein kinase (PKA). This enzyme catalyzes the transfer of the terminal phosphate group from ATP to specific serines or threonines of selected target proteins, thereby regulating their activity.

PKA is found in all animal cells and is thought to account for the effects of cyclic AMP in most of these cells. The substrates for PKA differ in different cell types, which explains why the effects of cyclic AMP vary so markedly depending on the cell type.

In the inactive state, PKA consists of a complex of two catalytic subunits and two regulatory subunits. The binding of cyclic AMP to the regulatory subunits alters their conformation, causing them to dissociate from the complex. The released catalytic subunits are thereby activated to phosphorylate specific substrate protein molecules (Figure 15-32). The regulatory subunits of PKA also are important for localizing the kinase inside the cell: special PKA anchoring proteins bind both to the regulatory subunits and to a membrane or a component of the cytoskeleton, thereby tethering the enzyme complex to a particular subcellular compartment. Some of these anchoring proteins also bind other kinases and some phosphatases, creating a signaling complex.

Some responses mediated by cyclic AMP are rapid while others are slow. In skeletal muscle cells, for example, activated PKA phosphorylates enzymes involved in glycogen metabolism, which simultaneously triggers the breakdown of glycogen to glucose and inhibits glycogen synthesis, thereby increasing the amount of glucose available to the muscle cell within seconds (see also Figure 15-30). At the other extreme are responses that take hours to develop fully and involve changes in the transcription of specific genes. In cells that secrete the peptide hormone somatostatin, for example, cyclic AMP activates the gene that encodes this hormone. The regulatory region of the somatostatin gene contains a short DNA sequence, called the cyclic AMP response element (CRE), that is also found in the regulatory region of many other genes activated by cyclic AMP. A specific gene regulatory protein called CRE-binding (CREB) protein recognizes this sequence. When CREB is phosphorylated by PKA on a single serine, it recruits a transcriptional coactivator called CREB-binding protein (CBP), which stimulates the transcription of these genes (Figure 15-33). If this serine is mutated, CREB cannot recruit CBP, and it no longer stimulates gene transcription in response to a rise in cyclic AMP levels.

Protein Phosphatases Make the Effects of PKA and Other Protein Kinases Transitory

Since the effects of cyclic AMP are usually transient, cells must be able to dephosphorylate the proteins that have been phosphorylated by PKA. Indeed, the activity of any protein regulated by phosphorylation depends on the balance at any instant between the activities of the kinases that phosphorylate it and the phosphatases that are constantly dephosphorylating it. In general, the dephosphorylation of phosphorylated serines and threonines is catalyzed by four types of serine/threonine phosphoprotein phosphatases—protein phosphatases I, IIA, IIB, and IIC. Except for protein phosphatase-IIC (which is a minor phosphatase, unrelated to the others), all of these phosphatases are composed of a homologous catalytic subunit complexed with one or more of a large set of regulatory subunits; the regulatory subunits help to control the phosphatase activity and enable the enzyme to select specific targets. Protein phosphatase I is responsible for dephosphorylating many of the proteins phosphorylated by PKA. It inactivates CREB, for example, by removing its activating phosphate, thereby turning off the transcriptional response caused by a rise in cyclic AMP concentration. Protein phosphatase IIA has a broad specificity and seems to be the main phosphatase responsible for reversing many of the phosphorylations catalyzed by serine/threonine kinases. Protein phosphatase IIB, also called calcineurin, is activated by Ca²⁺ and is especially abundant in the brain.

Having discussed how trimeric G proteins link activated receptors to adenylyl cyclase, we now consider how they couple activated receptors to another crucial enzyme, phospholipase C. The activation of this enzyme leads to an increase in the concentration of Ca²⁺ in the cytosol, which helps to relay the signal onward. Ca²⁺ is even more widely used as an intracellular mediator than is cyclic AMP.

Some G Proteins Activate the Inositol Phospholipid Signaling Pathway by Activating Phospholipase C-β

Many G-protein-linked receptors exert their effects mainly via G proteins that activate the plasma-membrane-bound enzyme phospholipase C-β. Several examples of responses activated in this way are listed in Table 15-2. The phospholipase acts on an inositol phospholipid (a phosphoinositide) called phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂], which is present in small amounts in the inner half of the plasma membrane lipid bilayer (Figure 15-34). Receptors that operate through this inositol phospholipid signaling pathway mainly activate a G protein called G_q, which in turn activates phospholipase C-β, in much the same way that G_s activates adenylyl cyclase. The activated phospholipase cleaves PI(4,5)P₂ to generate two products: inositol 1,4,5-trisphosphate and diacylglycerol (Figure 15-35). At this step, the signaling pathway splits into two branches.

Inositol 1,4,5-trisphosphate (IP₃) is a small, water-soluble molecule that leaves the plasma membrane and diffuses rapidly through the cytosol. When it reaches the endoplasmic reticulum (ER), it binds to and opens IP₃-gated Ca²⁺-release channels in the ER membrane. Ca²⁺ stored in the ER is released through the open channels, quickly raising the concentration of Ca²⁺ in the cytosol (Figure 15-36). We discuss later how Ca²⁺ acts to propagate the signal. Several mechanisms operate to terminate the initial Ca²⁺ response: (1) IP₃ is rapidly dephosphorylated by specific phosphatases to form IP₂; (2) IP₃ is phosphorylated to IP₄ (which may function as another intracellular mediator); and (3) Ca²⁺ that enters the cytosol is rapidly pumped out, mainly to the exterior of the cell.

At the same time that the IP₃ produced by the hydrolysis of PI(4,5)P₂ is increasing the concentration of Ca²⁺ in the cytosol, the other cleavage product of PI(4,5)P₂—diacylglycerol—is exerting different effects. Diacylglycerol remains embedded in the membrane, where it has two potential signaling roles. First, it can be further cleaved to release arachidonic acid, which can either act as a messenger in its own right or be used in the synthesis of other small lipid messengers called eicosanoids. Eicosanoids, such as the prostaglandins, are made by most vertebrate cell types and have a wide variety of biological activities. They participate in pain and inflammatory responses, for example, and most anti-inflammatory drugs (such as aspirin, ibuprofen, and cortisone) act—in part, at least—by inhibiting their synthesis.

The second, and more important, function of diacylglycerol is to activate a crucial serine/threonine protein kinase called protein kinase C (PKC), so named because it is Ca²⁺-dependent. The initial rise in cytosolic Ca²⁺ induced by IP₃ alters the PKC so that it translocates from the cytosol to the cytoplasmic face of the plasma membrane. There it is activated by the combination of Ca²⁺, diacylglycerol, and the negatively charged membrane phospholipid phosphatidylserine (see Figure 15-36). Once activated, PKC phosphorylates target proteins that vary depending on the cell type. The principles are the same as discussed earlier for PKA, although most of the target proteins are different.

Each of the two branches of the inositol phospholipid signaling pathway can be mimicked by the addition of specific pharmacological agents to intact cells. The effects of IP₃ can be mimicked by using a Ca ²⁺ ionophore, such as A23187 or ionomycin, which allows Ca²⁺ to move into the cytosol from the extracellular fluid (discussed in Chapter 11). The effects of diacylglycerol can be mimicked by phorbol esters, plant products that bind to PKC and activate it directly. Using these reagents, it has been shown that the two branches of the pathway often collaborate in producing a full cellular response. Some cell types, such as lymphocytes, for example, can be stimulated to proliferate in culture when treated with both a Ca²⁺ ionophore and a PKC activator, but not when they are treated with either reagent alone.

Ca²⁺ Functions as a Ubiquitous Intracellular Messenger

Many extracellular signals induce an increase in cytosolic Ca²⁺ level, not just those that work via G proteins. In egg cells, for example, a sudden rise in cytosolic Ca²⁺ concentration upon fertilization by a sperm triggers a Ca²⁺ wave that is responsible for the onset of embryonic development (Figure 15-37). In muscle cells, Ca²⁺ triggers contraction, and in many secretory cells, including nerve cells, it triggers secretion. Ca²⁺ can be used as a signal in this way because its concentration in the cytosol is normally kept very low (~10^-7 M), whereas its concentration in the extracellular fluid (~10^-3 M) and in the ER lumen is high. Thus, there is a large gradient tending to drive Ca²⁺ into the cytosol across both the plasma membrane and the ER membrane. When a signal transiently opens Ca²⁺ channels in either of these membranes, Ca²⁺ rushes into the cytosol, increasing the local Ca²⁺ concentration by 10–20-fold and triggering Ca²⁺-responsive proteins in the cell.

Three main types of Ca²⁺ channels can mediate this Ca²⁺ signaling:

• 1

  Voltage-dependent Ca²⁺ channels in the plasma membrane open in response to membrane depolarization and allow, for example, Ca²⁺ to enter activated nerve terminals and trigger neurotransmitter secretion.

• 2
  IP₃-gated Ca²⁺-release channels allow Ca²⁺ to escape from the ER when the inositol phospholipid signaling pathway is activated, as just discussed (see Figure 15-36).

• 3

  Ryanodine receptors (so called because they are sensitive to the plant alkaloid ryanodine) react to a change in plasma membrane potential to release Ca²⁺ from the sarcoplasmic reticulum and thereby stimulate the contraction of muscle cells; they are also present in the ER of many nonmuscle cells, including neurons, where they can contribute to Ca²⁺ signaling.

The concentration of Ca²⁺ in the cytosol is kept low in resting cells by several mechanisms (Figure 15-38). Most notably, all eucaryotic cells have a Ca²⁺-pump in their plasma membrane that uses the energy of ATP hydrolysis to pump Ca²⁺ out of the cytosol. Cells such as muscle and nerve cells, which make extensive use of Ca²⁺ signaling, have an additional Ca²⁺ transport protein (exchanger) in their plasma membrane that couples the efflux of Ca²⁺ to the influx of Na⁺. A Ca²⁺ pump in the ER membrane also has an important role in keeping the cytosolic Ca²⁺ concentration low: this Ca²⁺-pump enables the ER to take up large amounts of Ca²⁺ from the cytosol against a steep concentration gradient, even when Ca²⁺ levels in the cytosol are low. In addition, a low-affinity, high-capacity Ca²⁺ pump in the inner mitochondrial membrane has an important role in returning the Ca²⁺ concentration to normal after a Ca²⁺ signal; it uses the electrochemical gradient generated across this membrane during the electron-transfer steps of oxidative phosphorylation to take up Ca²⁺ from the cytosol.

The Frequency of Ca²⁺ Oscillations Influences a Cell's Response

Ca²⁺-sensitive fluorescent indicators, such as aequorin or fura-2 (discussed in Chapter 9), are often used to monitor cytosolic Ca²⁺ in individual cells after the inositol phospholipid signaling pathway has been activated. When viewed in this way, the initial Ca²⁺ signal is often seen to be small and localized to one or more discrete regions of the cell. These signals have been called Ca²⁺ blips, quarks, puffs, or sparks, and they are thought to reflect the local opening of individual (or small groups of) Ca²⁺-release channels in the ER and to represent elementary Ca²⁺ signaling units. If the extracellular signal is sufficiently strong and persistent, this localized signal can propagate as a regenerative Ca²⁺ wave through the cytosol, much like an action potential in an axon (see Figure 15-37). Such a Ca²⁺ “spike” is often followed by a series of further spikes, each usually lasting seconds (Figure 15-39). These Ca²⁺ oscillations can persist for as long as receptors are activated at the cell surface. Both the waves and the oscillations are thought to depend, in part at least, on a combination of positive and negative feedback by Ca²⁺ on both the IP₃-gated Ca²⁺-release channels and the ryanodine receptors: the released Ca²⁺ initially stimulates more Ca²⁺ release, a process known as Ca²⁺-induced Ca²⁺ release. But then, as its concentration gets high enough, the Ca²⁺ inhibits further release.

The frequency of the Ca²⁺ oscillations reflects the strength of the extracellular stimulus (see Figure 15-39), and it can be translated into a frequency-dependent cell response. In some cases, the frequency-dependent response itself is also oscillatory. In hormone-secreting pituitary cells, for example, stimulation by an extracellular signal induces repeated Ca²⁺ spikes, each of which is associated with a burst of hormone secretion. The frequency-dependent response can also be nonoscillatory. In some types of cells, for instance, one frequency of Ca²⁺ spikes activates the transcription of one set of genes, while a higher frequency activates the transcription of a different set. How do cells sense the frequency of Ca²⁺ spikes and change their response accordingly? The mechanism presumably depends on Ca²⁺-sensitive proteins that change their activity as a function of Ca²⁺ spike frequency. A protein kinase that acts as a molecular memory device seems to have this remarkable property, as we discuss next.

Ca²⁺/Calmodulin-dependent Protein Kinases (CaM-Kinases) Mediate Many of the Actions of Ca²⁺ in Animal Cells

Ca ²⁺ -binding proteins serve as transducers of the cytosolic Ca²⁺ signal. The first such protein to be discovered was troponin C in skeletal muscle cells; its role in muscle contraction is discussed in Chapter 16. A closely related Ca²⁺-binding protein, known as calmodulin, is found in all eucaryotic cells, where it can constitute as much as 1% of the total protein mass. Calmodulin functions as a multipurpose intracellular Ca²⁺ receptor, mediating many Ca²⁺-regulated processes. It consists of a highly conserved, single polypeptide chain with four high-affinity Ca²⁺-binding sites (Figure 15-40A). When activated by binding Ca²⁺, it undergoes a conformational change. Because two or more Ca²⁺ ions must bind before calmodulin adopts its active conformation, the protein responds in a switchlike manner to increasing concentrations of Ca²⁺ (see Figure 15-22): a tenfold increase in Ca²⁺ concentration, for example, typically causes a fiftyfold increase in calmodulin activation.

The allosteric activation of calmodulin by Ca²⁺ is analogous to the allosteric activation of PKA by cyclic AMP, except that Ca²⁺/calmodulin has no enzymic activity itself but instead acts by binding to other proteins. In some cases, calmodulin serves as a permanent regulatory subunit of an enzyme complex, but mostly the binding of Ca²⁺ enables calmodulin to bind to various target proteins in the cell to alter their activity.

When an activated molecule of Ca²⁺/calmodulin binds to its target protein, it undergoes a marked change in conformation (Figure 15-40B). Among the targets regulated by calmodulin binding are many enzymes and membrane transport proteins. As one example, Ca²⁺/calmodulin binds to and activates the plasma membrane Ca²⁺-pump that pumps Ca²⁺ out of cells. Thus, whenever the concentration of Ca²⁺ in the cytosol rises, the pump is activated, which helps to return the cytosolic Ca²⁺ level to normal.

Many effects of Ca²⁺, however, are more indirect and are mediated by protein phosphorylations catalyzed by a family of Ca ²⁺ /calmodulin-dependent protein kinases (CaM-kinases). These kinases, just like PKA and PKC, phosphorylate serines or threonines in proteins, and, as with PKA and PKC, the response of a target cell depends on which CaM-kinase-regulated target proteins are present in the cell. The first CaM-kinases to be discovered—myosin light-chain kinase, which activates smooth muscle contraction, and phosphorylase kinase, which activates glycogen breakdown—have narrow substrate specificities. A number of CaM-kinases, however, have much broader specificities, and these seem to be responsible for mediating many of the actions of Ca²⁺ in animal cells. Some phosphorylate gene regulatory proteins, such as the CREB protein discussed earlier, and in this way activate or inhibit the transcription of specific genes.

The best-studied example of such a multifunctional CaM-kinase is CaM-kinase II, which is found in all animal cells but is especially enriched in the nervous system. It constitutes up to 2% of the total protein mass in some regions of the brain, and it is highly concentrated in synapses. CaM-kinase II has at least two remarkable properties that are related. First, it can function as a molecular memory device, switching to an active state when exposed to Ca²⁺/calmodulin and then remaining active even after the Ca²⁺ signal has decayed. This is because the kinase phosphorylates itself (a process called autophosphorylation) as well as other cell proteins when it is activated by Ca²⁺/calmodulin. In its autophosphorylated state, the enzyme remains active even in the absence of Ca²⁺, thereby prolonging the duration of the kinase activity beyond that of the initial activating Ca²⁺ signal. This activity is maintained until phosphatases overwhelm the autophosphorylating activity of the enzyme and shut it off (Figure 15-41). CaM-kinase II activation can thereby serve as a memory trace of a prior Ca²⁺ pulse, and it seems to have an important role in some types of memory and learning in the vertebrate nervous system. Mutant mice that lack the brain-specific subunit illustrated in Figure 15-41 have specific defects in their ability to remember where things are in space. A point mutation in CaM-kinase II that removes its autophosphorylation site, but otherwise leaves the kinase activity intact, produces the same learning defect, revealing that the autophosphorylation is critical in these animals.

The second remarkable property of CaM-kinase II is that it can use its memory mechanism to act as a frequency decoder of Ca²⁺ oscillations. This property is thought to be especially important at a nerve cell synapse, where changes in intracellular Ca²⁺ levels in an activated postsynaptic cell can lead to long-term changes in the subsequent effectiveness of that synapse (discussed in Chapter 11). When CaM-kinase II is immobilized on a solid surface and exposed to both a protein phosphatase and repetitive pulses of Ca²⁺/calmodulin at different frequencies that mimic those observed in stimulated cells, the enzyme's activity increases steeply as a function of pulse frequency (Figure 15-42). Moreover, the frequency response of this multisubunit enzyme depends on its exact subunit composition, so that a cell can tailor its response to Ca²⁺ oscillations to particular needs by adjusting the composition of the CaM-kinase II enzyme that it makes.

Some G Proteins Directly Regulate Ion Channels

G proteins do not act exclusively by regulating the activity of membrane-bound enzymes that alter the concentration of cyclic AMP or Ca²⁺ in the cytosol. The α subunit of one type of G protein (called G₁₂), for example, activates a protein that converts a monomeric GTPase of the Rho family (discussed in Chapter 16) into its active form, which then alters the actin cytoskeleton. In some other cases, G proteins directly activate or inactivate ion channels in the plasma membrane of the target cell, thereby altering the ion permeability—and hence the excitability of the membrane. Acetylcholine released by the vagus nerve, for example, reduces both the rate and strength of heart muscle cell contraction (see Figure 15-9A). A special class of acetylcholine receptors that activate the G_i protein discussed earlier mediates this effect. Once activated, the α subunit of G_i inhibits adenylyl cyclase (as described previously), while the βγ complex binds to K⁺ channels in the heart muscle cell plasma membrane to open them. The opening of these K⁺ channels makes it harder to depolarize the cell, which contributes to the inhibitory effect of acetylcholine on the heart. (These acetylcholine receptors, which can be activated by the fungal alkaloid muscarine, are called muscarinic acetylcholine receptors to distinguish them from the very different nicotinic acetylcholine receptors, which are ion-channel-linked receptors on skeletal muscle and nerve cells that can be activated by the binding of nicotine, as well as by acetylcholine.)

Other trimeric G proteins regulate the activity of ion channels less directly, either by stimulating channel phosphorylation (by PKA, PKC, or CaM-kinase, for example) or by causing the production or destruction of cyclic nucleotides that directly activate or inactivate ion channels. The cyclic-nucleotide-gated ion channels have a crucial role in both smell (olfaction) and vision, as we now discuss.

Smell and Vision Depend on G-Protein-linked Receptors That Regulate Cyclic-Nucleotide-gated Ion Channels

Humans can distinguish more than 10,000 distinct smells, which are detected by specialized olfactory receptor neurons in the lining of the nose. These cells recognize odors by means of specific G-protein-linked olfactory receptors, which are displayed on the surface of the modified cilia that extend from each cell (Figure 15-43). The receptors act through cyclic AMP. When stimulated by odorant binding, they activate an olfactory-specific G protein (known as G_olf), which in turn activates adenylyl cyclase. The resulting increase in cyclic AMP opens cyclic-AMP-gated cation channels, thereby allowing an influx of Na⁺, which depolarizes the olfactory receptor neuron and initiates a nerve impulse that travels along its axon to the brain.

There are about 1000 different olfactory receptors in a mouse, each encoded by a different gene and each recognizing a different set of odorants. All of these receptors belong to the G-protein-linked receptor superfamily. Each olfactory receptor neuron produces only one of these 1000 receptors, and the neuron responds to a specific set of odorants by means of the specific receptor it displays. The same receptor also has a crucial role in directing the elongating axon of each developing olfactory neuron to the specific target neurons that it will connect to in the brain. A different set of more than 100 G-protein-linked receptors acts in a similar way to mediate a mouse's responses to pheromones, chemical signals detected in a different part of the nose that are used in communication between members of the same species.

Vertebrate vision involves a similarly elaborate, highly sensitive, signal-detection process. Cyclic-nucleotide-gated ion channels are also involved, but the crucial cyclic nucleotide is cyclic GMP (Figure 15-44) rather than cyclic AMP. As with cyclic AMP, a continuous rapid synthesis (by guanylyl cyclase) and rapid degradation (by cyclic GMP phosphodiesterase) controls the concentration of cyclic GMP in cells.

In visual transduction responses, which are the fastest G-protein-mediated responses known in vertebrates, the receptor activation caused by light leads to a fall rather than a rise in the level of the cyclic nucleotide. The pathway has been especially well studied in rod photoreceptors (rods) in the vertebrate retina. Rods are responsible for noncolor vision in dim light, whereas cone photoreceptors (cones) are responsible for color vision in bright light. A rod photoreceptor is a highly specialized cell with outer and inner segments, a cell body, and a synaptic region where the rod passes a chemical signal to a retinal nerve cell; this nerve cell in turn relays the signal along the visual pathway (Figure 15-45). The phototransduction apparatus is in the outer segment, which contains a stack of discs, each formed by a closed sac of membrane in which many photosensitive rhodopsin molecules are embedded. The plasma membrane surrounding the outer segment contains cyclic-GMP-gated Na ⁺ channels. These channels are kept open in the dark by cyclic GMP that has bound to them. Paradoxically, light causes a hyperpolarization (which inhibits synaptic signaling) rather than a depolarization of the plasma membrane (which could stimulate synaptic signaling). Hyperpolarization (an increase in the membrane potential—discussed in Chapter 11) results because the activation by light of rhodopsin molecules in the disc membrane leads to a fall in cyclic GMP concentration and the closure of the special Na⁺ channels in the surrounding plasma membrane (Figure 15-46).

Rhodopsin is a seven-pass transmembrane molecule homologous to other members of the G-protein-linked receptor family, and, like its cousins, it acts through a trimeric G protein. The activating extracellular signal, however, is not a molecule but a photon of light. Each rhodopsin molecule contains a covalently attached chromophore, 11-cis retinal, which isomerizes almost instantaneously to all-trans retinal when it absorbs a single photon. The isomerization alters the shape of the retinal, forcing a conformational change in the protein (opsin). The activated rhodopsin molecule then alters the G-protein transducin (G_t), causing its α subunit to dissociate and activate cyclic GMP phosphodiesterase. The phosphodiesterase then hydrolyzes cyclic GMP, so that cyclic GMP levels in the cytosol fall. This drop in cyclic GMP concentration leads to a decrease in the amount of cyclic GMP bound to the plasma membrane Na⁺ channels, allowing more of these highly cyclic-GMP-sensitive channels to close. In this way, the signal quickly passes from the disc membrane to the plasma membrane, and a light signal is converted into an electrical one.

A number of mechanisms operate in rods to allow the cells to revert quickly to a resting, dark state in the aftermath of a flash of light—a requirement for perceiving the shortness of the flash. A rhodopsin-specific kinase (RK) phosphorylates the cytosolic tail of activated rhodopsin on multiple serines, partially inhibiting the ability of the rhodopsin to activate transducin. An inhibitory protein called arrestin then binds to the phosphorylated rhodopsin, further inhibiting rhodopsin's activity. If the gene encoding RK is inactivated by mutation in mice or humans, the light response of rods is greatly prolonged, and the rods eventually die.

At the same time as rhodopsin is being shut off, an RGS protein (see p. 854) binds to activated transducin, stimulating the transducin to hydrolyze its bound GTP to GDP, which returns transducin to its inactive state. In addition, the Na⁺ channels that close in response to light are also permeable to Ca²⁺, so that when they close, the normal influx of Ca²⁺ is inhibited, causing the Ca²⁺ concentration in the cytosol to fall. The decrease in Ca²⁺ concentration stimulates guanylyl cyclase to replenish the cyclic GMP, rapidly returning its level to where it was before the light was switched on. A specific Ca²⁺-sensitive protein mediates the activation of guanylyl cyclase in response to a fall in Ca²⁺ levels. In contrast to calmodulin, this protein is inactive when Ca²⁺ is bound to it and active when it is Ca²⁺-free. It therefore stimulates the cyclase when Ca²⁺ levels fall following a light response.

These shut-off mechanisms do more than just return the rod to its resting state after a light flash; they also help to enable the photoreceptor to adapt, stepping down the response when it is exposed to light continuously. Adaptation, as we discussed earlier, allows the receptor cell to function as a sensitive detector of changes in stimulus intensity over an enormously wide range of baseline levels of stimulation.

The various trimeric G proteins we have discussed in this chapter are summarized in Table 15-3.

Extracellular Signals Are Greatly Amplified by the Use of Small Intracellular Mediators and Enzymatic Cascades

Despite the differences in molecular details, the signaling systems that are triggered by G-protein-linked receptors share certain features and are governed by similar general principles. They depend on relay chains of intracellular signaling proteins and small intracellular mediators. In contrast to the more direct signaling pathways used by nuclear receptors discussed earlier, and by ion-channel-linked receptors discussed in Chapter 11, these relay chains provide numerous opportunities for amplifying the responses to extracellular signals. In the visual transduction cascade just described, for example, a single activated rhodopsin molecule catalyzes the activation of hundreds of molecules of transducin at a rate of about 1000 transducin molecules per second. Each activated transducin molecule activates a molecule of cyclic GMP phosphodiesterase, each of which hydrolyzes about 4000 molecules of cyclic GMP per second. This catalytic cascade lasts for about 1 second and results in the hydrolysis of more than 10⁵ cyclic GMP molecules for a single quantum of light absorbed, and the resulting drop in the concentration of cyclic GMP in turn transiently closes hundreds of Na²⁺ channels in the plasma membrane (Figure 15-47). As a result, a rod cell can respond to a single photon of light, in a way that is highly reproducible in its timing and magnitude.

Likewise, when an extracellular signal molecule binds to a receptor that indirectly activates adenylyl cyclase via G_s, each receptor protein may activate many molecules of G_s protein, each of which can activate a cyclase molecule. Each cyclase molecule, in turn, can catalyze the conversion of a large number of ATP molecules to cyclic AMP molecules. A similar amplification operates in the inositol-phospholipid pathway. A nanomolar (10^-9 M) change in the concentration of an extracellular signal can thereby induce micromolar (10^-6 M) changes in the concentration of a small intracellular mediator such as cyclic AMP or Ca²⁺. Because these mediators function as allosteric effectors to activate specific enzymes or ion channels, a single extracellular signal molecule can cause many thousands of protein molecules to be altered within the target cell.

Any such amplifying cascade of stimulatory signals requires that there be counterbalancing mechanisms at every step of the cascade to restore the system to its resting state when stimulation ceases. Cells therefore have efficient mechanisms for rapidly degrading (and resynthesizing) cyclic nucleotides and for buffering and removing cytosolic Ca²⁺, as well as for inactivating the responding enzymes and ion channels once they have been activated. This is not only essential for turning a response off, it is also important for defining the resting state from which a response begins. As we saw earlier, in general, the response to stimulation can be rapid only if the inactivating mechanisms are also rapid. Each protein in the relay chain of signals can be a separate target for regulation, including the receptor, as we discuss next.

G-Protein-linked Receptor Desensitization Depends on Receptor Phosphorylation

As discussed earlier, target cells use a variety of mechanisms to desensitize, or adapt, when they are exposed to a high concentration of stimulating ligand for a prolonged period (see Figure 15-25). We discuss here only those mechanisms that involve an alteration in G-protein-linked receptors themselves.

These receptors can desensitize in three general ways:

• 1

  They can become altered so that they can no longer interact with G proteins (receptor inactivation).

• 2

  They can be temporarily moved to the interior of the cell (internalized) so that they no longer have access to their ligand (receptor sequestration).

• 3

  They can be destroyed in lysosomes after internalization (receptor down-regulation).

In each case, the desensitization process depends on phosphorylation of the receptor, by PKA, PKC, or a member of the family of G-protein-linked receptor kinases (GRKs). (The GRKs include the rhodopsin-specific kinase involved in rod photoreceptor desensitization discussed earlier.) The GRKs phosphorylate multiple serine and threonines on a receptor, but they do so only after the receptor has been activated by ligand binding. As with rhodopsin, once a receptor has been phosphorylated in this way, it binds with high affinity to a member of the arrestin family of proteins (Figure 15-48).

The bound arrestin can contribute to the desensitization process in at least two ways. First, it inactivates the receptor by preventing it from interacting with G proteins, an example of receptor uncoupling. Second, it can serve as an adaptor protein to couple the receptor to clathrin-coated pits (discussed in Chapter 13), inducing receptor-mediated endocytosis. Endocytosis results in either the sequestration or degradation (down-regulation) of the receptor, depending on the specific receptor and cell type, the concentration of the stimulating ligand, and the duration of the ligand's presence.

Summary

G-protein-linked receptors can indirectly activate or inactivate either plasma-membrane-bound enzymes or ion channels via G proteins. When stimulated by an activated receptor, a G protein disassembles into an α subunit and a βγ complex, both of which can directly regulate the activity of target proteins in the plasma membrane. Some G-protein-linked receptors either activate or inactivate adenylyl cyclase, thereby altering the intracellular concentration of the intracellular mediator cyclic AMP. Others activate a phosphoinositide-specific phospholipase C (phospholipase C-β), which hydrolyzes phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂] to generate two small intracellular mediators. One is inositol 1,4,5-trisphosphate (IP₃), which releases Ca²⁺from the ER and thereby increases the concentration of Ca²⁺in the cytosol. The other is diacylglycerol, which remains in the plasma membrane and activates protein kinase C (PKC). A rise in cyclic AMP or Ca²⁺levels affects cells mainly by stimulating protein kinase A (PKA) and Ca²⁺/calmodulin-dependent protein kinases (CaM-kinases), respectively.

PKC, PKA, and CaM-kinases phosphorylate specific target proteins on serines or threonines and thereby alter the activity of the proteins. Each type of cell has characteristic sets of target proteins that are regulated in these ways, enabling the cell to make its own distinctive response to the small intracellular mediators. The intracellular signaling cascades activated by G-protein-linked receptors allow the responses to be greatly amplified, so that many target proteins are changed for each molecule of extracellular signaling ligand bound to its receptor.

The responses mediated by G-protein-linked receptors are rapidly turned off when the extracellular signaling ligand is removed. Thus, the G-protein α subunit is induced to inactivate itself by hydrolyzing its bound GTP to GDP, IP₃is rapidly dephosphorylated by a phosphatase (or phosphorylated by a kinase), cyclic nucleotides are hydrolyzed by phosphodiesterases, Ca²⁺is rapidly pumped out of the cytosol, and phosphorylated proteins are dephosphorylated by protein phosphatases. Activated G-protein-linked receptors themselves are phosphorylated by GRKs, thereby trigging arrestin binding, which uncouples the receptors from G proteins and promotes receptor endocytosis.

Activated Receptor Tyrosine Kinases Phosphorylate Themselves

The extracellular signal proteins that act through receptor tyrosine kinases consist of a large variety of secreted growth factors and hormones. Notable examples discussed elsewhere in this book include epidermal growth factor (EGF), platelet-derived growth factor (PDGF), fibroblast growth factors (FGFs), hepatocyte growth factor (HGF), insulin, insulinlike growth factor-1 (IGF-1), vascular endothelial growth factor (VEGF), macrophage-colony-stimulating factor (M-CSF), and all the neurotrophins, including nerve growth factor (NGF).

Many cell-surface-bound signal proteins also act through these receptors. The largest class of these membrane-bound ligands is the ephrins, which regulate the cell adhesion and repulsion responses that guide the migration of cells and axons along specific pathways during animal development (discussed in Chapter 21). The receptors for ephrins, called Eph receptors, are also the most numerous receptor tyrosine kinases. The ephrins and Eph receptors are unusual in that they can simultaneously act as both ligand and receptor: on binding to an Eph receptor, some ephrins not only activate the Eph receptor but also become activated themselves to transmit signals into the interior of the ephrin-expressing cell. In this way, an interaction between an ephrin protein on one cell and an Eph protein on another cell can lead to bidirectional reciprocal signaling that changes the behavior of both cells. Such bidirectional signaling between ephrins and Eph receptors is required, for example, to keep cells in particular parts of the developing brain from mixing with cells in neighboring parts.

Receptor tyrosine kinases can be classified into more than 16 structural subfamilies, each dedicated to its complementary family of protein ligands. Several of these families that operate in mammals are shown in Figure 15-49, and some of their ligands and functions are given in Table 15-4. In all cases, the binding of a signal protein to the ligand-binding domain on the outside of the cell activates the intracellular tyrosine kinase domain. Once activated, the kinase domain transfers a phosphate group from ATP to selected tyrosine side chains, both on the receptor proteins themselves and on intracellular signaling proteins that subsequently bind to the phosphorylated receptors.

How does the binding of an extracellular ligand activate the kinase domain on the other side of the plasma membrane? For a G-protein-linked receptor, ligand binding is thought to change the relative orientation of several of the transmembrane α helices, thereby shifting the position of the cytoplasmic loops relative to each other. It is difficult to imagine, however, how a conformational change could propagate across the lipid bilayer through a single transmembrane α helix. Instead, for the enzyme-linked receptors, two or more receptor chains come together in the membrane, forming a dimer or higher oligomer. In some cases, ligand binding induces the oligomerization. In other cases, the oligomerization occurs before ligand binding, and the ligand causes a reorientation of the receptor chains in the membrane. In either case, the rearrangement induced in cytosolic tails of the receptors initiates the intracellular signaling process. For receptor tyrosine kinases, the rearrangement enables the neighboring kinase domains of the receptor chains to cross-phosphorylate each other on multiple tyrosines, a process referred to as autophosphorylation.

To activate a receptor tyrosine kinase the ligand usually has to bind simultaneously to two adjacent receptor chains. PDGF, for example, is a dimer, which cross-links two receptors together (Figure 15-50A). Even some monomeric ligands, such as EGF, bind to two receptors simultaneously and cross-link them directly. By contrast, FGFs, which are also monomers, first form multimers by binding to heparan sulfate proteoglycans, either on the target cell surface or in the extracellular matrix. In this way, they are able to cross-link adjacent receptors (Figure 15-50B). In contact-dependent signaling, the ligands form clusters in the plasma membrane of the signaling cell and can thereby cross-link the receptors on the target cell (Figure15-50C); thus, whereas membrane-bound ephrins activate Eph receptors, soluble ephrins will do so only if they are aggregated.

Because of the requirement for receptor oligomerization, it is relatively easy to inactivate a specific receptor tyrosine kinase to determine its importance for a cell response. For this purpose, cells are transfected with DNA encoding a mutant form of the receptor that oligomerizes normally but has an inactive kinase domain. When coexpressed at a high level with normal receptors, the mutant receptor acts in a dominant-negative way, disabling the normal receptors by forming inactive dimers with them (Figure 15-51).

Autophosphorylation of the cytosolic tail of receptor tyrosine kinases contributes to the activation process in two ways. First, phosphorylation of tyrosines within the kinase domain increases the kinase activity of the enzyme. Second, phosphorylation of tyrosines outside the kinase domain creates high-affinity docking sites for the binding of a number of intracellular signaling proteins in the target cell. Each type of signaling protein binds to a different phosphorylated site on the activated receptor because it contains a specific phosphotyrosine-binding domain that recognizes surrounding features of the polypeptide chain in addition to the phosphotyrosine. Once bound to the activated kinase, the signaling protein may itself become phosphorylated on tyrosines and thereby activated; alternatively, the binding alone may be sufficient to activate the docked signaling protein. In summary, autophosphorylation serves as a switch to trigger the transient assembly of a large intracellular signaling complex, which then broadcasts signals along multiple routes to many destinations in the cell (Figure 15-52). Because different receptor tyrosine kinases bind different combinations of these signaling proteins, they activate different responses.

The receptors for insulin and IGF-1 act in a slightly different way. They are tetramers to start with (see Figure 15-49), and ligand binding is thought to induce a rearrangement of the transmembrane receptor chains, so that the two kinase domains come close together. Most of the phosphotyrosine docking sites generated by ligand binding are not on the receptor itself, but on a specialized docking protein called insulin receptor substrate-1 (IRS-1). The activated receptor first autophosphorylates its kinase domains, which then phosphorylate IRS-1 on multiple tyrosines, thereby creating many more docking sites than could be accommodated on the receptor alone. Other docking proteins are used in a similar way by some other receptor tyrosine kinases to enlarge the size of the signaling complex.

Phosphorylated Tyrosines Serve as Docking Sites For Proteins With SH2 Domains

A whole menagerie of intracellular signaling proteins can bind to the phosphotyrosines on activated receptor tyrosine kinases (or on special docking proteins such as IRS-1) to help to relay the signal onward. Some docked proteins are enzymes, such as phospholipase C-γ (PLC-γ), which functions in the same way as phospholipase C-β—activating the inositol phospholipid signaling pathway discussed earlier in connection with G-protein-linked receptors. Through this pathway, receptor tyrosine kinases can increase cytosolic Ca²⁺ levels. Much more often, these receptors depend more on relay chains of protein-protein interactions. For example, another enzyme that docks on these receptors is the cytoplasmic tyrosine kinase Src, which phosphorylates other signaling proteins on tyrosines. Yet another is phosphatidylinositol 3′-kinase (PI 3-kinase), which, as we discuss later, generates specific lipid molecules in the plasma membrane to attract other signaling proteins there.

Although the intracellular signaling proteins that bind to phosphotyrosines on activated receptor tyrosine kinases and docking proteins have varied structures and functions, they usually share highly conserved phosphotyrosine-binding domains. These can be either SH2 domains (for Src homology region, because it was first found in the Src protein) or, less commonly, PTB domains (for phosphotyrosine-binding). By recognizing specific phosphorylated tyrosines, these small domains serve as modules that enable the proteins that contain them to bind to activated receptor tyrosine kinases, as well as to many other intracellular signaling proteins that have been transiently phosphorylated on tyrosines (Figure 15-53). Many signaling proteins also contain other protein modules that allow them to interact specifically with other proteins as part of the signaling process. These include the SH3 domain (again, so named because it was first discovered in Src), which binds to proline-rich motifs in intracellular proteins (see Figure 15-20).

Not all proteins that bind to activated receptor tyrosine kinases via SH2 domains help to relay the signal onward. Some act to decrease the signaling process, providing negative feedback. One example is the c-Cbl protein, which can dock on some activated receptors and catalyze their conjugation with ubiquitin. This ubiquitylation promotes the internalization and degradation of the receptors—a process called receptor down-regulation (see Figure 15-25).

Some signaling proteins are composed almost entirely of SH2 and SH3 domains and function as adaptors to couple tyrosine-phosphorylated proteins to other proteins that do not have their own SH2 domains (see Figure 15-20). Such adaptor proteins help to couple activated receptors to the important downstream signaling protein Ras. As we discuss next, Ras acts as a transducer and bifurcation signaling protein, changing the nature of the signal and broadcasting it along multiple downstream pathways, including a major signaling pathway that can help stimulate cells to proliferate or differentiate. Mutations that activate this pathway, and thereby stimulate cell division inappropriately, are a causative factor in many types of cancer.

Ras Is Activated by a Guanine Nucleotide Exchange Factor

The Ras proteins belong to the large Ras superfamily of monomeric GTPases. The family also contains two other subfamilies: the Rho family, involved in relaying signals from cell-surface receptors to the actin cytoskeleton and elsewhere (discussed in Chapter 16), and the Rab family, involved in regulating the traffic of intracellular transport vesicles (discussed in Chapter 13). Like almost all of these monomeric GTPases, the Ras proteins contain a covalently attached lipid group that helps to anchor the protein to a membrane—in this case, to the cytoplasmic face of the plasma membrane where the protein functions. There are multiple Ras proteins, and different ones act in different cell types. Because they all seem to work in much the same way, we shall refer to them simply as Ras.

Ras helps to broadcast signals from the cell surface to other parts of the cell. It is often required, for example, when receptor tyrosine kinases signal to the nucleus to stimulate cell proliferation or differentiation by altering gene expression. If Ras function is inhibited by the microinjection of neutralizing anti-Ras antibodies or a dominant-negative mutant form of Ras, the cell proliferation or differentiation responses normally induced by the activated receptor tyrosine kinases do not occur. Conversely, if a hyperactive mutant Ras protein is introduced into some cell lines, the effect on cell proliferation or differentiation is sometimes the same as that induced by the binding of ligands to cell-surface receptors. In fact, Ras was first discovered as the hyperactive product of a mutant ras gene that promoted the development of cancer; we now know that about 30% of human tumors have a hyperactive ras mutation.

Like other GTP-binding proteins, Ras functions as a switch, cycling between two distinct conformational states—active when GTP is bound and inactive when GDP is bound (see Figure 15-17). Two classes of signaling proteins regulate Ras activity by influencing its transition between active and inactive states. Guanine nucleotide exchange factors (GEFs) promote the exchange of bound nucleotide by stimulating the dissociation of GDP and the subsequent uptake of GTP from the cytosol, thereby activating Ras. GTPase-activating proteins (GAPs) increase the rate of hydrolysis of bound GTP by Ras, thereby inactivating Ras (Figure 15-54). Hyperactive mutant forms of Ras are resistant to GAP-mediated GTPase stimulation and are locked permanently in the GTP-bound active state, which is why they promote the development of cancer.

In principle, receptor tyrosine kinases could activate Ras either by activating a GEF or by inhibiting a GAP. Even though some GAPs bind directly (via their SH2 domains) to activated receptor tyrosine kinases (see Figure 15-53), whereas GEFs bind only indirectly, it is the indirect coupling of the receptor to a GEF that is responsible for driving Ras into its active state. In fact, the loss of function of a Ras-specific GEF has a similar effect to the loss of function of that Ras. The activation of the other Ras-like proteins, including those of the Rho family, is also thought to occur through the activation of GEFs.

Genetic studies in flies and worms, and biochemical studies in mammalian cells, indicate that adaptor proteins link receptor tyrosine kinases to Ras. The Grb-2 protein in mammalian cells, for example, binds through its SH2 domain to specific phosphotyrosines on activated receptor tyrosine kinases and through its SH3 domains to proline-rich motifs on a GEF called Sos. Some activated receptor tyrosine kinases, however, do not display the specific phosphotyrosines required for Grb-2 docking; these receptors recruit another adaptor protein called Shc, which binds both to the activated receptor and to Grb-2, thereby coupling the receptor to Sos by a more indirect route. The assembly of the complex of receptor-Grb-2-Sos (or receptor-Shc-Grb-2-Sos) brings Sos into position to activate neighboring Ras molecules by stimulating it to exchange its bound GDP for GTP (Figure 15-55). The importance of Grb-2 is indicated by the finding that Grb-2-deficient mice die early in embryogenesis. Very similar sets of proteins are thought to operate in all animals to activate Ras.

This pathway from receptor tyrosine kinases is not the only means of activating Ras. Other Ras GEFs are activated independently of Sos. One that is found mainly in the brain, for example, is activated by Ca²⁺ and diacylglycerol and can couple G-protein-linked receptors to Ras activation.

Once activated, Ras in turn activates various other signaling proteins to relay the signal downstream along several pathways. One of the signaling pathways Ras activates is a serine/threonine phosphorylation cascade that is highly conserved in eucaryotic cells from yeasts to humans. As we discuss next, a crucial component in this cascade is a novel type of protein kinase called MAP-kinase.

Ras Activates a Downstream Serine/Threonine Phosphorylation Cascade That Includes a MAP-Kinase

Both the tyrosine phosphorylations and the activation of Ras triggered by activated receptor tyrosine kinases are short-lived. Tyrosine-specific protein phosphatases (discussed later) quickly reverse the phosphorylations, and GAPs induce activated Ras to inactivate itself by hydrolyzing its bound GTP to GDP. To stimulate cells to proliferate or differentiate, these short-lived signaling events must be converted into longer-lasting ones that can sustain the signal and relay it downstream to the nucleus to alter the pattern of gene expression. Activated Ras triggers this conversion by initiating a series of downstream serine/threonine phosphorylations, which are much longer-lived than tyrosine phosphorylations. Many serine/threonine kinases participate in this phosphorylation cascade, but three of them constitute the core module of the cascade. The last of the three is called a mitogen-activated protein kinase (MAP-kinase).

An unusual feature of a MAP-kinase is that its full activation requires the phosphorylation of both a threonine and a tyrosine, which are separated in the protein by a single amino acid. The protein kinase that catalyzes both of these phosphorylations is called a MAP-kinase-kinase, which in the mammalian Ras signaling pathway is called MEK. The requirement for both a tyrosine and a threonine phosphorylation ensures that the MAP-kinase is kept inactive unless specifically activated by a MAP-kinase-kinase, whose only known substrate is a MAP-kinase. MAP-kinase-kinase is itself activated by phosphorylation catalyzed by the first kinase in the three-component module, MAP-kinase-kinase-kinase, which in the mammalian Ras signaling pathway is called Raf. The Raf kinase is activated by activated Ras.

Once activated, the MAP-kinase relays the signal downstream by phosphorylating various proteins in the cell, including gene regulatory proteins and other protein kinases (Figure 15-56). It enters the nucleus, for example, and phosphorylates one or more components of a gene regulatory complex. This activates the transcription of a set of immediate early genes, so named because they turn on within minutes of the time that cells are stimulated by an extracellular signal, even if protein synthesis is experimentally blocked with drugs. Some of these genes encode other gene regulatory proteins that turn on other genes, a process that requires both protein synthesis and more time. In this way the Ras-MAP-kinase signaling pathway conveys signals from the cell surface to the nucleus and alters the pattern of gene expression in significant ways. Among the genes activated by this pathway are those required for cell proliferation, such as the genes encoding G₁ cyclins (discussed in Chapter 17).

MAP-kinases are usually activated only transiently in response to extracellular signals, and the period of time they remain active can profoundly influence the nature of the response. When EGF activates its receptors on a neural precursor cell line, for example, MAP-kinase activity peaks at 5 minutes and rapidly declines, and the cells later go on to divide. By contrast, when NGF activates its receptors on the same cells, MAP-kinase activity remains high for many hours, and the cells stop proliferating and differentiate into neurons.

MAP-kinases are inactivated by dephosphorylation, and the specific removal of phosphate from either the tyrosine or the threonine is enough to inactivate the enzyme. In some cases, stimulation by an extracellular signal induces the expression of a dual-specificity phosphatase that removes both phosphates and inactivates the kinase, providing a form of negative feedback. In other cases, stimulation causes the kinase to be switched off more rapidly by phosphatases that are already present.

Three-component MAP-kinase signaling modules operate in all animal cells, as well as in yeasts, with different ones mediating different responses in the same cell. In budding yeast, for example, one such module mediates the mating pheromone response via the βγ complex of a G protein, another the response to starvation, and yet another the response to osmotic shock. Some of these three-component MAP-kinase modules use one or more of the same kinases and yet manage to activate different effector proteins and hence different responses. How do cells avoid cross talk between the different parallel signaling pathways to ensure that each response is specific? One way is to use scaffold proteins that bind all or some of the kinases in a specific module to form a complex, as illustrated in Figure 15-57 and discussed earlier (see Figure 15-19A).

Mammalian cells also use this strategy to prevent cross talk between MAP-kinase signaling pathways. At least 5 parallel MAP-kinase modules can operate in a mammalian cell. These modules are composed of at least 12 MAP-kinases, 7 MAP-kinase-kinases, and 7 MAP-kinase-kinase-kinases. Several of these modules are activated by different kinds of cell stresses, such as UV irradiation, heat shock, osmotic stress, and stimulation by inflammatory cytokines. The three kinases in at least some of these stress-activated modules are held together by binding to a common scaffold protein, just as in yeast. The scaffold strategy provides precision, helps to create a large change in MAP-kinase activity in response to small changes in signal molecule concentration, and avoids cross-talk. However, it reduces the opportunities for amplification and spreading of the signal to different parts of the cell, which require at least some of the components to be diffusible (see Figure 15-16).

When Ras is activated by receptor tyrosine kinases, it usually activates more than just the MAP-kinase signaling pathway. It also usually helps activate PI3-kinase, which can signal cells to survive and grow.

PI 3-Kinase Produces Inositol Phospholipid Docking Sites in the Plasma Membrane

Extracellular signal proteins stimulate cells to divide, in part by activating the Ras-MAP-kinase pathway just discussed. If cells continually divided without growing, however, they would get progressively smaller and would eventually disappear. Thus, to proliferate, most cells need to be stimulated to enlarge (grow), as well as to divide. In some cases, one signal protein does both; in others one signal protein (a mitogen) mainly stimulates cell division, while another (a growth factor) mainly stimulates cell growth. One of the major intracellular signaling pathways leading to cell growth involves phosphatidylinositol 3-kinase (PI 3-kinase). This kinase principally phosphorylates inositol phospholipids rather than proteins; it can be activated by receptor tyrosine kinases, as well as by many other types of cell-surface receptors, including some that are G-protein-linked.

Phosphatidylinositol (PI) is unique among membrane lipids because it can undergo reversible phosphorylation at multiple sites to generate a variety of distinct inositol phospholipids. When activated, PI 3-kinase catalyzes the phosphorylation of inositol phospholipids at the 3 position of the inositol ring to generate lipids called PI(3,4)P₂ or PI(3,4,5)P₃ (Figure 15-58). The PI(3,4)P₂ and PI(3,4,5)P₃ then serve as docking sites for intracellular signaling proteins, bringing these proteins together into signaling complexes, which relay the signal into the cell from the cytosolic face of the plasma membrane.

It is important to distinguish this use of inositol phospholipids from their use we discussed earlier. We considered earlier how PI(4,5)P₂ is cleaved by PLC-β (in the case of G-protein-linked receptors) or PLC-γ (in the case of receptor tyrosine kinases) to generate soluble IP₃ and membrane-bound diacylglycerol. The IP₃ releases Ca²⁺ from the ER, while the diacylglycerol activates PKC (see Figures 15-58 and 15-35). By contrast, PI(3,4)P₂ and PI(3,4,5)P₃ are not cleaved by PLC. They remain in the plasma membrane until they are dephosphorylated by specific inositol phospholipid phosphatases that remove phosphate from the 3 position of the inositol ring. Mutations that inactivate one such phosphatase (called PTEN), and thereby prolong signaling by PI 3-kinase, promote the development of cancer, and they are found in many human cancers. The mutations result in prolonged cell survival, indicating that signaling through PI 3-kinase normally promotes cell survival, as well as cell growth.

There are various types of PI 3-kinases. The one that is activated by receptor tyrosine kinases consists of a catalytic and regulatory subunit. The regulatory subunit is an adaptor protein that binds to phosphotyrosines on activated receptor tyrosine kinases through its SH2 domains (see Figure 15-53). Another PI 3-kinase has a different regulatory subunit and is activated by the βγ complex of a trimeric G protein when G-protein-linked receptors are activated by their extracellular ligand. The catalytic subunit, which is similar in both cases, also has a binding site for activated Ras, which allows Ras to directly stimulate PI 3-kinases.

Intracellular signaling proteins bind to the PI(3,4)P₂ and PI(3,4,5)P₃ that are produced by activated PI 3-kinase mainly through their Pleckstrin homology (PH) domain, first identified in the platelet protein Pleckstrin. PH domains are found in about 200 human proteins, including Sos (the GEF discussed earlier that activates Ras), and some atypical PKCs that do not depend on Ca²⁺ for their activation. The importance of these domains is illustrated dramatically by certain genetic immunodeficiency diseases in both humans and mice, where the PH domain in a cytoplasmic tyrosine kinase called BTK is inactivated by mutation. Normally, when antigen receptors on B lymphocytes (B cells) activate PI 3-kinase, the resulting inositol lipid docking sites recruit both BTK and PLC-γ to the cytoplasmic face of the plasma membrane. There, the two proteins interact: BTK phosphorylates and activates PLC-γ, which then cleaves PI(4,5)P₂ to generate IP₃ and diacylglycerol to relay the signal onward (Figure 15-59). Because the mutant BTK cannot bind to the lipid docking sites produced after receptor activation, the receptors cannot signal the B cells to proliferate or survive, resulting in a severe deficiency in antibody production.

The PI 3-Kinase/Protein Kinase B Signaling Pathway Can Stimulate Cells to Survive and Grow

One way in which PI 3-kinase signals cells to survive is by indirectly activating protein kinase B (PKB) (also called Akt). This kinase contains a PH domain, which directs it to the plasma membrane when PI 3-kinase is activated there by an extracellular survival signal. After binding to PI(3,4,5)P₃ on the cytosolic face of the membrane, the PKB alters its conformation so that it can now be activated in a process that requires phosphorylation by a phosphatidylinositol-dependent protein kinase called PDK1, which is recruited to the membrane in the same way. Once activated, the PKB returns to the cytoplasm and phosphorylates a variety of target proteins. One of these, called BAD, is a protein that normally encourages cells to undergo programmed cell death, or apoptosis (mentioned earlier and discussed in detail in Chapter 17). By phosphorylating BAD, PKB inactivates it, thereby promoting cell survival (Figure 15-60). PKB also promotes cell survival by inhibiting other cell death activators, in some cases by inhibiting the transcription of the genes that encode them.

The pathways by which PI 3-kinase signals cells to grow (and increase their metabolism generally) are complex and still poorly understood. One way in which growth factors stimulate cell growth is by increasing the rate of protein synthesis through enhancing the efficiency with which ribosomes translate certain mRNAs into protein. A protein kinase called S6 kinase is part of one of the signaling pathways from PI 3-kinase to the ribosome. It phosphorylates and thereby activates the S6 subunit of ribosomes, which helps to increase the translation of a subset of mRNAs that encode ribosomal proteins and other components of the translational apparatus. The activation of S6 kinase is itself a complex process that depends on PDK1 and the phosphorylation of many sites on the protein. PDK1 may phosphorylate one of these sites in response to PI 3-kinase activation.

Figure 15-61 summarizes the five parallel intracellular signaling pathways we have discussed so far—one triggered by G-protein-linked receptors, two triggered by receptor tyrosine kinases, and two triggered by both kinds of receptors.

Tyrosine-Kinase-associated Receptors Depend on Cytoplasmic Tyrosine Kinases for Their Activity

Many cell-surface receptors depend on tyrosine phosphorylation for their activity and yet lack an obvious tyrosine kinase domain. These receptors act through cytoplasmic tyrosine kinases, which are associated with the receptors and phosphorylate various target proteins, often including the receptors themselves, when the receptors bind their ligand. The receptors thus function in much the same way as receptor tyrosine kinases, except that their kinase domain is encoded by a separate gene and is noncovalently associated with the receptor polypeptide chain. As with receptor tyrosine kinases, these receptors must oligomerize to function (Figure 15-62).

Many of these receptors depend on members of the largest family of mammalian cytoplasmic tyrosine kinases, the Src family of protein kinases (see Figure 3-68). This family includes the following members: Src, Yes, Fgr, Fyn, Lck, Lyn, Hck, and Blk. These protein kinases all contain SH2 and SH3 domains and are located on the cytoplasmic side of the plasma membrane, held there partly by their interaction with transmembrane receptor proteins and partly by covalently attached lipid chains. Different family members are associated with different receptors and phosphorylate overlapping but distinct sets of target proteins. Lyn, Fyn, and Lck, for example, are each associated with different sets of receptors in lymphocytes. In each case the kinase is activated when an extracellular ligand binds to the appropriate receptor protein. Src itself, as well as several other family members, can also bind to activated receptor tyrosine kinases; in these cases, the receptor and cytoplasmic kinases mutually stimulate each other's catalytic activity, thereby strengthening and prolonging the signal.

Another type of cytoplasmic tyrosine kinase associates with integrins, the main family of receptors that cells use to bind to the extracellular matrix (discussed in Chapter 19). The binding of matrix components to integrins can activate intracellular signaling pathways that influence the behavior of the cell. When integrins cluster at sites of matrix contact, they help trigger the assembly of cell-matrix junctions called focal adhesions. Among the many proteins recruited into these junctions is the cytoplasmic tyrosine kinase called focal adhesion kinase (FAK), which binds to the cytosolic tail of one of the integrin subunits with the assistance of other cytoskeletal protein. The clustered FAK molecules cross-phosphorylate each other, creating phosphotyrosine docking sites where the Src kinase can bind. Src and FAK now phosphorylate each other and other proteins that assemble in the junction, including many of the signaling proteins used by receptor tyrosine kinases. In this way, the two kinases signal to the cell that it has adhered to a suitable substratum, where the cell can now survive, grow, divide, migrate, and so on. Mice deficient in FAK die early in development, and their cells do not migrate normally in a culture dish.

Cytokine receptors are the subfamily of enzyme-linked receptors that we discuss next. They constitute the largest and most diverse class of receptors that rely on cytoplasmic kinases to relay signals into the cell. They include receptors for many kinds of local mediators (collectively called cytokines), as well as receptors for some hormones, such as growth hormone (see Figure 15-62) and prolactin. As we discuss next, these receptors are stably associated with a class of cytoplasmic tyrosine kinases called Jaks, which activate latent gene regulatory proteins called STATs. The STAT proteins are normally inactive, being located at the cell surface; cytokine or hormone binding causes them to migrate to the nucleus and activate gene transcription.

Cytokine Receptors Activate the Jak-STAT Signaling Pathway, Providing a Fast Track to the Nucleus

Many intracellular signaling pathways lead from cell-surface receptors to the nucleus, where they alter gene transcription. The Jak-STAT signaling pathway, however, provides one of the most direct routes. It was initially discovered in studies on the effects of interferons, which are cytokines secreted by cells (especially white blood cells) in response to viral infection. Interferons bind to receptors on noninfected neighboring cells and induce the cells to produce proteins that increase their resistance to viral infection. When activated, interferon receptors activate a novel class of cytoplasmic tyrosine kinases called Janus kinases (Jaks) (after the two-faced Roman god). The Jaks then phosphorylate and activate a set of latent gene regulatory proteins called STATs (signal transducers and activators of transcription), which move into the nucleus and stimulate the transcription of specific genes. More than 30 cytokines and hormones activate the Jak-STAT pathway by binding to cytokine receptors, some of which are listed in Table 15-5.

All STATs also have an SH2 domain that enables them to dock onto specific phosphotyrosines on some activated receptor tyrosine kinase receptors. These receptors can directly activate the bound STAT, independently of Jaks. In fact, the nematode C. elegans uses STATs for signaling but does not make any Jaks or cytokine receptors, suggesting that STATs evolved before Jaks and cytokine receptors.

Cytokine receptors are composed of two or more polypeptide chains. Some cytokine receptor chains are specific to a particular cytokine receptor, while others are shared among several such receptors. All cytokine receptors, however, are associated with one or more Jaks. There are four known Jaks—Jak1, Jak2, Jak3, and Tyk2—and each is associated with particular cytokine receptors. The receptors for α-interferon, for example, are associated with Jak1 and Tyk2, whereas the receptors for γ-interferon are associated with Jak1 and Jak2 (see Table 15-5). As expected, mice that lack Jak1 do not respond to either of these interferons. The receptor for the hormone erythropoietin, which stimulates erythrocyte precursor cells to survive, proliferate, and differentiate, is associated with only Jak2. In Jak2-deficient mice, erythrocyte development fails, and the mice die early in development.

Cytokine binding either induces the receptor chains to oligomerize or reorients the chains in a preformed oligomer. In either case, the binding brings the associated Jaks close enough together for them to cross-phosphorylate each other, thereby increasing the activity of their tyrosine kinase domains. The Jaks then phosphorylate tyrosines on the cytokine receptors, creating phosphotyrosine docking sites for STATs and other signaling proteins.

There are seven known STATs, each with an SH2 domain that performs two functions. First, it mediates the binding of the STAT protein to a phosphotyrosine docking site on an activated cytokine receptor (or receptor tyrosine kinase); once bound, the Jaks phosphorylate the STAT on tyrosines, causing it to dissociate from the receptor. Second, the SH2 domain on the released STAT now mediates its binding to a phosphotyrosine on another STAT molecule, forming either a STAT homodimer or heterodimer. The STAT dimer then moves into the nucleus, where, in combination with other gene regulatory proteins, it binds to a specific DNA response element in various genes and stimulates their transcription (Figure 15-63). In response to the hormone prolactin, for example, which stimulates breast cells to produce milk, activated STAT5 stimulates the transcription of genes that encode milk proteins.

Cytokine receptors activate the appropriate STAT proteins because the SH2 domain of these STATs recognizes only the specific phosphotyrosine docking sites on these receptors. Activated receptors for α-interferon, for example, recruit both STAT1 and STAT2, whereas activated receptors for γ-interferon recruit only STAT1. If the SH2 domain of the α-interferon receptor is replaced with the SH2 domain of the γ-interferon receptor, the activated hybrid receptor recruits both STAT1 and STAT2, just like the α-interferon receptor itself.

The responses mediated by STATs are often regulated by negative feedback. In addition to activating genes that encode proteins mediating the cytokine-induced response, the STAT dimers may also activate genes that encode inhibitory proteins. In some cases, the inhibitor binds to both the activated cytokine receptors and STAT proteins, which blocks further STAT activation and helps to shut off the response; in other cases, the inhibitor achieves the same result by blocking Jak function.

Such negative feedback mechanisms, however, are not enough on their own to turn off the response. The activated Jaks and STATs also have to be inactivated by dephosphorylation of their phosphotyrosines. As in all signaling pathways that use tyrosine phosphorylation, the dephosphorylation is performed by protein tyrosine phosphatases, which are as important in the signaling process as the protein tyrosine kinases that add the phosphates.

Some Protein Tyrosine Phosphatases May Act as Cell-Surface Receptors

As discussed earlier, only a small number of serine/threonine phosphatase catalytic subunits are responsible for removing phosphate groups from phosphorylated serines and threonines on proteins. By contrast, there are about 30 protein tyrosine phosphatases (PTPs) encoded in the human genome. Like tyrosine kinases, they occur in both cytoplasmic and transmembrane forms, none of which are structurally related to serine/threonine protein phosphatases. Individual protein tyrosine phosphatases display exquisite specificity for their substrates, removing phosphate groups from only selected phosphotyrosines on a subset of tyrosine-phosphorylated proteins. Together, these phosphatases ensure that tyrosine phosphorylations are short-lived and that the level of tyrosine phosphorylation in resting cells is very low. They do not, however, simply continuously reverse the effects of protein tyrosine kinases; they are regulated to act only at the appropriate time in a signaling response or in the cell-division cycle (discussed in Chapter 17).

Two cytoplasmic tyrosine phosphatases in vertebrates have SH2 domains and are therefore called SHP-1 and SHP-2 (Figure 15-64). SHP-1 helps to terminate some cytokine responses in blood cells by dephosphorylating activated Jaks: mutant erythropoietin receptors that cannot recruit SHP-1, for example, activate Jak2 for much longer than normal. Moreover, SHP-1-deficient mice have abnormalities in almost all blood cell lineages, emphasizing the importance of SHP-1 in blood cell development. Both SHP-1 and SHP-2 also help terminate responses mediated by some receptor tyrosine kinases.

There are a large number of transmembrane protein tyrosine phosphatases, but the functions of most of them are unknown. At least some are thought to function as receptors; as this has not been directly demonstrated, however, they are referred to as receptorlike tyrosine phosphatases. They all have a single transmembrane segment and usually possess two tyrosine phosphatase domains on the cytosolic side of the plasma membrane. An important example is the CD45 protein (see Figure 15-64), which is found on the surface of all white blood cells and has an essential role in the activation of both T and B lymphocytes by foreign antigens. The ligand that is presumed to bind to the extracellular domain of the CD45 protein has not been identified. However, the role of CD45 in signal transduction has been studied by using recombinant DNA techniques to construct a hybrid protein with an extracellular EGF-binding domain and intracellular CD45 tyrosine phosphatase domains. The surprising result is that EGF binding seems to inactivate the phosphatase activity of the hybrid protein rather than activating it.

This finding raises the possibility that some receptor tyrosine kinases and receptor tyrosine phosphatases may collaborate when they bind their respective cell-surface-bound ligands—with the kinases adding more phosphates and the phosphatase removing fewer—to maximally stimulate the tyrosine phosphorylation of selected intracellular signaling proteins. The significance of ligand-induced inhibition of CD45 phosphatase is still uncertain, however, and it seems unlikely to be the whole story; CD45 requires its phosphatase activity to function in lymphocyte activation.

Some receptorlike tyrosine phosphatases display features of cell-adhesion proteins and can even mediate homophilic cell-cell binding in cell adhesion assays (see Figure 19-26). In the developing nervous system, for example, they may have an important role in guiding the growing tips of developing nerve cell axons to their targets. In Drosophila, the genes encoding several receptorlike tyrosine phosphatases are expressed exclusively in the nervous system, and when some of them are inactivated by mutation, the axons of certain developing neurons fail to find their way to their normal targets. In some cases at least, the phosphatase activity of the protein is required to counteract the action of a cytoplasmic tyrosine kinase for normal axon guidance.

Transmembrane tyrosine phosphatases can also serve as signaling ligands that activate receptors on a neighboring cell. An example is the protein tyrosine phosphatase ζ/β (see Figure 15-64), which is expressed on the surface of certain glial cells in the mammalian brain. It binds to a receptor protein (called contactin) on developing nerve cells, stimulating the cells to extend long processes. It is possible that the phosphatase also conveys a signal to the glial cell in this interaction, but such bidirectional signaling has not been directly demonstrated for transmembrane tyrosine phosphatases.

Having discussed the crucial role of tyrosine phosphorylation and dephosphorylation in the intracellular signaling pathways activated by many enzyme-linked receptors, we now turn to a class of enzyme-linked receptors that rely entirely on serine/threonine phosphorylation. These transmembrane serine/ threonine kinases activate an even more direct signaling pathway to the nucleus than does the Jak-STAT pathway discussed earlier. They directly phosphorylate latent gene regulatory proteins called Smads, which then migrate into the nucleus to activate gene transcription.

Signal Proteins of the TGF-β Superfamily Act Through Receptor Serine/Threonine Kinases and Smads

The transforming growth factor-β (TGF-β) superfamily consists of a large number of structurally related, secreted, dimeric proteins. They act either as hormones or, more commonly, as local mediators to regulate a wide range of biological functions in all animals. During development, they regulate pattern formation and influence various cell behaviors, including proliferation, differentiation, extracellular matrix production, and cell death. In adults, they are involved in tissue repair and in immune regulation, as well as in many other processes. The superfamily includes the TGF-βs themselves, the activins, and the bone morphogenetic proteins (BMPs). The BMPs constitute the largest family.

All of these proteins act through enzyme-linked receptors that are single-pass transmembrane proteins with a serine/threonine kinase domain on the cytosolic side of the plasma membrane. There are two classes of these receptor serine/threonine kinases—type I and type II—which are structurally similar. Each member of the TGF-β superfamily binds to a characteristic combination of type-I and type-II receptors, both of which are required for signaling. Typically, the ligand first binds to and activates a type-II receptor homodimer, which recruits, phosphorylates, and activates a type-I receptor homodimer, forming an active tetrameric receptor complex.

Once activated, the receptor complex uses a strategy for rapidly relaying the signal to the nucleus that is very similar to the Jak-STAT strategy used by cytokine receptors. The route to the nucleus, however, is even more direct. The type-I receptor directly binds and phosphorylates a latent gene regulatory protein of the Smad family (named after the first two identified, Sma in C. elegans and Mad in Drosophila). Activated TGF-β receptors and activin receptors phosphorylate Smad2 or Smad3, while activated BMP receptors phosphorylate Smad1, Smad5, or Smad8. Once one of these Smads has been phosphorylated, it dissociates from the receptor and binds to Smad4, which can form a complex with any of the above five receptor-activated Smads. The Smad complex then moves into the nucleus, where it associates with other gene regulatory proteins, binds to specific sites in DNA, and activates a particular set of target genes (Figure 15-65).

Some TGF-β family members serve as graded morphogens during development, inducing different responses in a developing cell depending on their concentration (discussed in Chapter 21). The different responses can be reproduced by experimentally altering the amount of active Smad complexes in the nucleus, suggesting that the level of these complexes may provide a direct readout of the level of receptor activation. If the DNA-binding sites in different target genes have different affinities for the complexes, then the particular genes activated would reflect the cell's position in the concentration gradient of the morphogen.

As with the Jak-STAT pathway, the Smad pathway is also often regulated by feedback inhibition. Among the target genes activated by Smad complexes are those that encode inhibitory Smads, including Smad6 and Smad7. These Smads act as decoys. They bind to activated type-I receptors and prevent other Smads from binding there. This blocks the formation of active Smad complexes and shuts off the response to the TGF-β family ligand. Other types of extracellular ligands can also stimulate the production of inhibitory Smads to antagonize signaling by a TGF-β ligand; γ-interferon, for example, activates the Jak-STAT pathway, and the resulting activated STAT dimers induce the production of Smad7, which inhibits signaling by TGF-β.

In addition to these intracellular inhibitors, a number of secreted extracellular inhibitory proteins can also neutralize signaling mediated by TGF-β family members. They directly bind to the signal molecules and prevent them from activating their receptors on target cells. Noggin and chordin, for example, inhibit BMPs, and follistatin inhibits activins. Noggin and chordin help to induce the development of the vertebrate nervous system by preventing BMPs from inhibiting this development (discussed in Chapter 21). The TGF-β family members, as well as some of their inhibitors, are usually secreted as inactive precursors that are subsequently activated by proteolytic cleavage.

We turn now to enzyme-linked receptors that are neither kinases nor associated with kinases. We saw earlier that nitric oxide is widely used as a signaling molecule, diffusing through the plasma membrane of a target cell and stimulating a cytoplasmic guanylyl cyclase to produce the intracellular mediator cyclic GMP. The receptors we now consider are transmembrane proteins with guanylyl cyclase activity.

Receptor Guanylyl Cyclases Generate Cyclic GMP Directly

Receptor guanylyl cyclases are single-pass transmembrane proteins with an extracellular binding site for a signal molecule and an intracellular guanylyl cyclase catalytic domain. The binding of the signal molecule activates the cyclase domain to produce cyclic GMP, which in turn binds to and activates a cyclic GMP-dependent protein kinase (PKG), which phosphorylates specific proteins on serine or threonine. Thus, receptor guanylyl cyclases use cyclic GMP as an intracellular mediator in the same way that some G-protein-linked receptors use cyclic AMP, except that the linkage between ligand binding and cyclase activity is a direct one.

Among the signal molecules that use receptor guanylyl cyclase receptors are the natriuretic peptides (NPs), a family of structurally related secreted signal peptides that regulate salt and water balance and dilate blood vessels. There are several types of NPs, including atrial natriuretic peptide (ANP) and brain natriuretic peptide (BNP). Muscle cells in the atrium of the heart secrete ANP when blood pressure rises. The ANP stimulates the kidneys to secrete Na⁺ and water and induces the smooth muscle cells in blood vessels walls to relax. Both of these effects tend to lower the blood pressure. When gene targeting is used to inactivate the ANP receptor guanylyl cyclase in mice, the mice have chronically elevated blood pressure, resulting in progressive heart enlargement.

An increasing number of receptor guanylyl cyclases are being discovered, but in most cases they are orphan receptors, where the ligand that normally activates them is unknown. The genome of the nematode C. elegans, for example, encodes 26 of these receptors. Most of those that have been studied are expressed in specific subsets of sensory neurons, suggesting that they may be involved in detecting particular molecules in the worm's environment. Some of the orphan receptors in mammals are found in sensory neurons in the part of the nose involved in detecting pheromones.

All the signaling pathways activated by G-protein-linked and enzyme-linked receptors we have discussed so far depend on serine/threonine-specific protein kinases, tyrosine-specific protein kinases, or both. These kinases are all structurally related, as reviewed in Figure 15-66. Some enzyme-linked receptors, however, depend on an entirely unrelated type of protein kinase, as we now discuss.

Bacterial Chemotaxis Depends on a Two-Component Signaling Pathway Activated by Histidine-Kinase-associated Receptors

As pointed out earlier, many of the mechanisms involved in chemical signaling between cells in multicellular animals are thought to have evolved from mechanisms used by unicellular organisms to respond to chemical changes in their environment. In fact, some of the same intracellular mediators, such as cyclic nucleotides and Ca²⁺, are used by both types of organisms. Among the best-studied reactions of unicellular organisms to extracellular signals are their chemotactic responses, in which cell movement is oriented toward or away from a source of some chemical in the environment. We conclude this section on enzyme-linked receptors with a brief account of bacterial chemotaxis, which depends on a two-component signaling pathway, involving histidine-kinase-associated receptors. The same type of signaling pathway is used by yeasts and plants, although apparently not by animals.

Motile bacteria will swim toward higher concentrations of nutrients (attractants), such as sugars, amino acids, and small peptides, and away from higher concentrations of various noxious chemicals (repellents). They swim by means of flagella, each of which is attached by a short, flexible hook at its base to a small protein disc embedded in the bacterial membrane. This disc is part of a tiny motor that uses the energy stored in the transmembrane H⁺ gradient to rotate rapidly and turn the helical flagellum (Figure 15-67). Because the flagella on the bacterial surface have an intrinsic “handedness,” different directions of rotation have different effects on movement. Counterclockwise rotation allows all the flagella to draw together into a coherent bundle, so that the bacterium swims uniformly in one direction. Clockwise rotation causes them to fly apart, so that the bacterium tumbles chaotically without moving forward (Figure 15-68). In the absence of any environmental stimulus, the direction of rotation of the disc reverses every few seconds, producing a characteristic pattern of movement in which smooth swimming in a straight line is interrupted by abrupt, random changes in direction caused by tumbling.

The normal swimming behavior of bacteria is modified by chemotactic attractants or repellents, which bind to specific receptor proteins and affect the frequency of tumbling by increasing or decreasing the time that elapses between successive changes in direction of flagellar rotation. When bacteria are swimming in a favorable direction (toward a higher concentration of an attractant or away from a higher concentration of a repellent), they tumble less frequently than when they are swimming in an unfavorable direction (or when no gradient is present). Since the periods of smooth swimming are longer when a bacterium is traveling in a favorable direction, it will gradually progress in that direction—toward an attractant or away from a repellent.

These responses are mediated by histidine-kinase-associated chemotaxis receptors, which typically are dimeric transmembrane proteins that bind specific attractants and repellents on the outside of the plasma membrane. The cytoplasmic tails of the receptors are stably associated with an adaptor protein CheW and a histidine kinase CheA, which help to couple the receptors to the flagellar motor. Repellent binding activates the receptors, whereas attractant binding inactivates them; a single receptor can bind either type of molecule, with opposite consequences. The binding of a repellent to the receptor activates CheA, which phosphorylates itself on a histidine and almost immediately transfers the phosphate to an aspartic acid on a messenger protein CheY. The phosphorylated CheY dissociates from the receptor, diffuses through the cytosol, binds to the flagellar motor, and causes the motor to rotate clockwise, so that the bacterium tumbles. CheY has intrinsic phosphatase activity and dephosphorylates itself in a process that is greatly accelerated by the CheZ protein (Figure 15-69).

The response to an increase in the concentration of an attractant or repellent is only transient, even if the higher level of ligand is maintained, as the bacteria desensitize, or adapt, to the increased stimulus. Whereas the initial effect on tumbling occurs in less than a second, adaptation takes minutes. The adaptation is a crucial part of the response, as it enables the bacteria to respond to changes in concentration of ligand rather than to steady-state levels. It is mediated by the covalent methylation (catalyzed by a methyl transferase) and demethylation (catalyzed by a methylase) of the chemotaxis receptors, which change their responsiveness to ligand binding when methylated.

All of the genes and proteins involved in this highly adaptive behavior have now been identified. It therefore seems likely that bacterial chemotaxis will be the first signaling system to be completely understood in molecular terms. Even in this relatively simple signaling network, computer-based simulations are required to comprehend how the system works as an integrated network. Cell signaling pathways will provide an especially rich area of investigation for a new generation of computational biologists, as their network properties will not be understandable without powerful computational tools.

There are some cell-surface receptor proteins that do not fit into the three major classes we have discussed thus far—ion-channel-linked, G-protein-linked, and enzyme-linked. In the next section, we consider cell-surface receptors that activate signaling pathways that depend on proteolysis. These pathways have especially important roles in animal development.

Summary

There are five known classes of enzyme-linked receptors: (1) receptor tyrosine kinases, (2) tyrosine-kinase-associated receptors, (3) receptor serine/threonine kinases, (4) transmembrane guanylyl cyclases, and (5) histidine-kinase-associated receptors. In addition, some transmembrane tyrosine phosphatases, which remove phosphate from phosphotyrosine side chains of specific proteins, are thought to function as receptors, although for the most part their ligands are unknown. The first two classes of receptors are by far the most numerous.

Ligand binding to receptor tyrosine kinases induces the receptors to cross-phosphorylate their cytoplasmic domains on multiple tyrosines. The autophosphorylation activates the kinases, as well as producing a set of phosphotyrosines that then serve as docking sites for a set of intracellular signaling proteins, which bind via their SH2 (or PTB) domains. Some of the docked proteins serve as adaptors to couple the receptors to the small GTPase Ras, which, in turn, activates a cascade of serine/threonine phosphorylations that converge on a MAP-kinase, which relays the signal to the nucleus by phosphorylating gene regulatory proteins there. Ras can also activate another protein that docks on activated receptor tyrosine kinases—PI 3-kinase—which generates specific inositol phospholipids that serve as docking sites in the plasma membrane for signaling proteins with PH domains, including protein kinase B (PKB).

Tyrosine-kinase-associated receptors depend on various cytoplasmic tyrosine kinases for their action. These kinases include members of the Src family, which associate with many kinds of receptors, and the focal adhesion kinase (FAK), which associates with integrins at focal adhesions. The cytoplasmic tyrosine kinases then phosphorylate a variety of signaling proteins to relay the signal onward. The largest family of receptors in this class is the cytokine receptors family. When stimulated by ligand binding, these receptors activate Jak cytoplasmic tyrosine kinases, which phosphorylate STATs. The STATs then dimerize, migrate to the nucleus, and activate the transcription of specific genes. Receptor serine/threonine kinases, which are activated by signaling proteins of the TGF-β superfamily, act similarly: they directly phosphorylate and activate Smads, which then oligomerize with another Smad, migrate to the nucleus, and activate gene transcription.

Bacterial chemotaxis is mediated by histidine-kinase-associated chemotaxis receptors. When activated by the binding of a repellent, the receptors stimulate their associated protein kinase to phosphorylate itself on histidine and then transfer that phosphate to a messenger protein, which relays the signal to the flagellar motor to alter the bacterium's swimming behavior. Attractants have the opposite effect on this kinase and therefore on swimming.

The Receptor Protein Notch Is Activated by Cleavage

Signaling through the Notch receptor protein may be the most widely used signaling pathway in animal development. As discussed in Chapter 21, it has a general role in controlling cell fate choices during development, mainly by amplifying and consolidating molecular differences between adjacent cells. Although Notch signaling is involved in the development of most tissues, it is best known for its role in nerve cell production in Drosophila. The nerve cells usually arise as isolated single cells within an epithelial sheet of precursor cells. During the process, each future nerve cell or committed nerve-cell precursor signals to its immediate neighbors not to develop in the same way at the same time, a process known as lateral inhibition. In a fly embryo, for example, the inhibited cells around the future nerve-cell precursors develop into epidermal cells. Lateral inhibition depends on a contact-dependent signaling mechanism that is mediated by a signal protein called Delta, displayed on the surface of the future neural cell. By binding to Notch on a neighboring cell, Delta signals to the neighbor not to become neural (Figure 15-70). When this signaling process is defective in flies, the neighbors of neural cells also develop as neural cells, producing a huge excess of neurons at the expense of epidermal cells, which is lethal. Signaling between adjacent cells via Notch and Delta (or the Deltalike ligand Serrate) regulates cell fate choices in a wide variety of tissues and animals, helping to create fine-grained patterns of distinct cell types. The Notch-mediated signal can have other effects beside lateral inhibition; in some tissues, for example, it works in the opposite way, causing neighboring cells to behave similarly.

Both Notch and Delta are single-pass transmembrane proteins, and both require proteolytic processing to function. Although it is still unclear why Delta has to be cleaved, the cleavage of Notch is central to how Notch activation alters gene expression in the nucleus. When activated by the binding of Delta on another cell, an intracellular protease cleaves off the cytoplasmic tail of Notch, and the released tail moves into the nucleus to activate the transcription of a set of Notch-response genes. The Notch tail acts by binding to a gene regulatory protein called CSL (so named because it is called CBF1 in mammals, Suppressor of Hairless in flies, and Lag-1 in worms); this converts CSL from a transcriptional repressor into a transcriptional activator. The products of the main genes directly activated by Notch signaling are themselves gene regulatory proteins, but with an inhibitory action: they block the expression of genes required for neural differentiation (in the nervous system), and of various other genes in other tissues.

The Notch receptor undergoes three proteolytic cleavages, but only the last two depend on Delta. As part of its normal biosynthesis, a protease called furin acts in the Golgi apparatus to cleave the newly synthesized Notch protein in its future extracellular domain. This cleavage converts Notch into a heterodimer, which is then transported to the cell surface as the mature receptor. The binding of Delta to Notch induces a second cleavage in the extracellular domain, mediated by a different protease. A final cleavage quickly follows, cutting free the cytoplasmic tail of the activated receptor (Figure 15-71).

The cleavage of the Notch tail occurs very close to the plasma membrane, just within the transmembrane segment. In this respect it resembles the cleavage of another, more sinister transmembrane protein—the β-amyloid precursor protein (APP), which is expressed in neurons and is implicated in Alzheimer's disease. APP is cleaved within its transmembrane segment, releasing one peptide fragment into the extracellular space of the brain and another into the cytosol of the neuron. In Alzheimer's disease, the extracellular fragments accumulate in excessive amounts and aggregate into filaments that form amyloid plaques, which are believed to injure nerve cells and contribute to their loss. The most frequent genetic cause of early-onset Alzheimer's disease is a mutation in the presenilin-1 (PS-1) gene, which encodes an 8-pass transmembrane protein that participates in the cleavage of APP. The mutations in PS-1 cause cleavage of APP into amyloid-plaque-forming fragments at an increased rate. Genetic evidence in C. elegans, Drosophila, and mice indicates that the PS-1 protein is a required component of the Notch signaling pathway, helping to perform the final cleavage that activates Notch. Indeed, Notch signaling and cleavage are greatly impaired in PS-1-deficient cells.

Remarkably, Notch signaling is regulated by glycosylation. The Fringe family of glycosyltransferases adds extra sugars to the O-linked oligosaccharide (discussed in Chapter 13) on Notch, which alters the specificity of Notch for its ligands. This has provided the first example of the modulation of ligand-receptor signaling by differential receptor glycosylation.

Wnt Proteins Bind to Frizzled Receptors and Inhibit the Degradation of β-Catenin

Wnt proteins are secreted signal molecules that act as local mediators to control many aspects of development in all animals that have been studied. They were discovered independently in flies and in mice: in Drosophila, the wingless (wg) gene originally came to light because of its role in wing development, while in mice, the Int-1 gene was found because it promoted the development of breast tumors when activated by the integration of a virus next to it. The cell-surface receptors for the Wnts belong to the Frizzled family of seven-pass transmembrane proteins. They resemble G-protein-linked receptors in structure, and some of them can signal through G proteins and the inositol phospholipid pathway discussed earlier. They mainly signal, however, through G-protein-independent pathways, which require a cytoplasmic signaling protein called Dishevelled.

The best characterized of the Dishevelled-dependent pathways acts by regulating the proteolysis of a multifunctional protein called β-catenin (or Armadillo in flies), which functions both in cell-cell adhesion and as a latent gene regulatory protein. Wnts activate this pathway by binding to both a Frizzled protein and a co-receptor protein. The co-receptor protein is related to the low density lipoprotein (LDL) receptor protein (discussed in Chapter 13) and is therefore called LDL-receptor-related protein (LRP). It is uncertain how Frizzled and LRP activate Dishevelled, which relays the signal onward.

In the absence of Wnt signaling, most of a cell's β-catenin is located at cell-cell adherens junctions, where it is associated with cadherins, which are transmembrane adhesion proteins. As discussed in Chapter 19, the β-catenin in these junctions helps link the cadherins to the actin cytoskeleton. Any β-catenin not associated with cadherins is rapidly degraded in the cytoplasm. This degradation depends on a large degradation complex, which recruits β-catenin and contains at least three other proteins (Figure 15-72A):

• 1

  A serine/threonine kinase called glycogen synthase kinase-3β (GSK-3β) phosphorylates β-catenin, thereby marking the protein for ubiquitylation and rapid degradation in proteasomes.

• 2

  The tumor-suppressor protein adenomatous polyposis coli (APC) is so named because the gene encoding it is often mutated in a type of benign tumor (adenoma) of the colon. The tumor projects into the lumen as a polyp, which can eventually become malignant. APC helps promote the degradation of β-catenin by increasing the affinity of the degradation complex for β-catenin, as required for effective phosphorylation of β-catenin by GSK-3β.

• 3

  A scaffold protein called axin holds the protein complex together.

The binding of a Wnt protein to Frizzled and LRP leads to the inhibition of β-catenin phosphorylation and degradation. The mechanism is not understood in detail, but it requires Dishevelled and several other signaling proteins that bind to Dishevelled, including the serine/threonine kinase called casein kinase 1. As a result, unphosphorylated β-catenin accumulates in the cytoplasm and nucleus (Figure 15-72B).

In the nucleus, the target genes for Wnt signaling are normally kept silent by an inhibitory complex of gene regulatory proteins, which includes proteins of the LEF-1/TCF family bound to the corepressor protein Groucho (see Figure 15-72A). The increase in undegraded β-catenin caused by Wnt signaling allows β-catenin to enter the nucleus and bind to LEF-1/TCF, displacing Groucho. The β-catenin now functions as a coactivator, inducing the transcription of the Wnt target genes (see Figure 15-72B).

Among the genes activated by β-catenin is c-myc, which encodes a protein (c-Myc) that is a powerful stimulator of cell growth and proliferation (discussed in Chapter 17). Mutations of the APC gene occur in 80% of human colon cancers. These mutations inhibit the protein's ability to bind β-catenin, so that β-catenin accumulates in the nucleus and stimulates the transcription of c-myc and other Wnt target genes, even in the absence of Wnt signaling. The resulting uncontrolled cell proliferation promotes the development of cancer.

Hedgehog Proteins Act Through a Receptor Complex of Patched and Smoothened, Which Oppose Each Other

Like Wnt proteins, the Hedgehog proteins are a family of secreted signal molecules that act as local mediators in many developmental processes in both invertebrates and vertebrates. Abnormalities in the Hedgehog pathway during development can be lethal and in adult cells can also lead to cancer. The Hedgehog proteins were discovered in Drosophila, where a mutation in the only gene encoding such a protein produces a larva with spiky processes (denticles) resembling a hedgehog. At least three genes encode Hedgehog proteins in vertebrates—sonic, desert, and indian hedgehog. The active form of all Hedgehog proteins is unusual in that it is covalently coupled to cholesterol, which helps to restrict its diffusion following secretion. The cholesterol is added during a remarkable processing step, in which the protein cleaves itself. The proteins are also modified by the addition of a fatty acid chain, which, for unknown reasons, can be required for their signaling activity.

Two transmembrane proteins, Patched and Smoothened, mediate the responses to all Hedgehog proteins. Patched is predicted to cross the plasma membrane 12 times, and it is the receptor that binds the Hedgehog protein. In the absence of a Hedgehog signal, Patched inhibits the activity of Smoothened, which is a 7-pass transmembrane protein with a structure similar to a Frizzled protein. This inhibition is relieved when a Hedgehog protein binds to Patched, allowing Smoothened to relay the signal into the cell. Most of what we know about the downstream signaling pathway activated by Smoothened comes from genetic studies in flies, and it is the fly pathway that we summarize here.

In some respects the Hedgehog signaling pathway in Drosophila operates similarly to the Wnt pathway. In the absence of a Hedgehog signal, a gene regulatory protein called Cubitus interruptus (Ci) is proteolytically cleaved in proteasomes. Instead of being completely degraded, however, it is processed to form a smaller protein that accumulates in the nucleus, where it acts as a transcriptional repressor, helping to keep some Hedgehog-responsive genes silent. The proteolytic processing of the Ci protein depends on a large multiprotein complex. The complex contains a serine/threonine kinase (called Fused) of unknown function, an anchoring protein (called Costal) that binds the complex to microtubules (keeping Ci out of the nucleus), and an adaptor protein (called Suppressor of Fused) (Figure 15-73A). When Hedgehog binds to Patched to activate the signaling pathway, Ci processing is suppressed, and the unprocessed Ci protein is released from its complex and enters the nucleus, where it activates the transcription of Hedgehog target genes (Figure 15-73B).

Among the genes activated by Ci is the gene that encodes the Wnt protein Wingless, which helps pattern tissues in the fly embryo (discussed in Chapter 21). Another target gene is patched itself; the resulting increase in Patched protein on the cell surface inhibits further Hedgehog signaling—a form of negative feedback.

Many gaps in the Hedgehog signaling pathway still remain to be filled in. It is not known, for example, how Patched inhibits Smoothened, how Smoothened activates the pathway, how the proteolysis of Ci is regulated (although it is known that Ci phosphorylation by PKA is required for the processing), or how the release of the complex from microtubules and unprocessed Ci from the complex is controlled.

Even less is known about the Hedgehog pathway in vertebrate cells. In addition to there being at least three types of vertebrate Hedgehog proteins, there are two forms of Patched and three Ci-like proteins (Gli1, Gli2, and Gli3). Unlike in flies, Hedgehog signaling stimulates the transcription of the Gli genes, and it is unclear whether all of the Gli proteins undergo proteolytic processing, although there is evidence that Gli3 does. Inactivating mutations in one of the human patched genes, which leads to excessive Hedgehog signaling, occur frequently in the most common form of skin cancer (basal cell carcinoma), suggesting that Patched normally helps to keep skin cell proliferation in check.

Multiple Stressful and Proinflammatory Stimuli Act Through an NF-κB-Dependent Signaling Pathway

The NF-κB proteins are latent gene regulatory proteins that lie at the heart of most inflammatory responses. These responses occur as a reaction to infection or injury and help protect the animal and its cells from these stresses. When excessive or inappropriate, however, inflammatory responses can also damage tissue and cause severe pain, as happens in joints in rheumatoid arthritis, for example. NF-κB proteins also have an important role in intercellular signaling during normal vertebrate development, although the extracellular signals that activate NF-κB in these circumstances are unknown. In Drosophila, however, genetic studies have identified both the extracellular and the intracellular proteins that activate the NF-κB family member Dorsal, which has a crucial role in specifying the dorsal-ventral axis of the developing fly embryo (discussed in Chapter 21). The same intracellular signaling pathway is also involved in defending the fly from infection, just as in vertebrates.

Two vertebrate cytokines are especially important in inducing inflammatory responses—tumor necrosis factor α (TNF-α) and interleukin-1 (IL-1). Both are made by cells of the innate immune system, such as macrophages, in response to infection or tissue injury. These proinflammatory cytokines bind to cell-surface receptors and activate NF-κB, which is normally sequestered in an inactive form in the cytoplasm of almost all of our cells. Once activated, NF-κB turns on the transcription of more than 60 known genes that participate in inflammatory responses. Although TNF-α receptors and IL-1 receptors are structurally unrelated, they operate in much the same way.

There are five NF-κB proteins in mammals (RelA, RelB, c-Rel, NF-κB1, and NF-κB2), and they form a variety of homodimers and heterodimers, each of which activates its own characteristic set of genes. Inhibitory proteins called IκB bind tightly to the dimers and hold them in an inactive state within large protein complexes in the cytoplasm. Signals such as TNF-α or IL-1 activate the dimers by triggering a signaling pathway that leads to the phosphorylation, ubiquitylation, and consequent degradation of IκB. The degradation of IκB exposes a nuclear localization signal on the NF-κB proteins, which now move into the nucleus and stimulate the transcription of specific genes. The phosphorylation of IκB is performed by a specific serine/threonine kinase called IκB kinase (IKK).

The mechanism by which the binding of a proinflammatory cytokine to its cell-surface receptors activates IκB kinase is illustrated for the TNF-α receptor in Figure 15-74. Ligand binding causes the cytosolic tails of the clustered receptors to recruit various adaptor proteins and cytoplasmic serine/threonine kinases. One of the recruited kinases is thought to be an IκB kinase kinase (IKKK) that directly phosphorylates and activates the IκB kinase (IKK).

Not all of the signaling proteins recruited to the cytosolic tail of the TNF-α receptor contribute to NF-κB activation, however. Some can trigger a MAP-kinase cascade, while others can activate a proteolytic cascade that leads to apoptosis (discussed in Chapter 17).

Thus far, we have discussed cell signaling mainly in animals, with a few diversions into yeasts and bacteria. But intercellular signaling is just as important for plants as it is for animals, although the mechanisms and molecules used are mainly different, as we discuss next.

Summary

Some signaling pathways that are especially important in animal development depend on proteolysis for at least part of their action. Notch receptors are activated by cleavage when Delta (or a related ligand) on another cell binds to them; the cleaved cytosolic tail of Notch migrates into the nucleus, where it stimulates gene transcription. In the Wnt signaling pathway, by contrast, the proteolysis of the latent gene regulatory protein β-catenin is inhibited when secreted Wnt proteins bind to their receptors; as a result, β-catenin accumulates in the nucleus and activates the transcription of Wnt target genes.

Hedgehog signaling in flies works much like Wnt signaling: in the absence of a signal, a bifunctional, cytoplasmic gene regulatory protein Ci is proteolytically cleaved to form a transcriptional repressor that keeps Hedgehog target genes silenced. The binding of Hedgehog to its receptor inhibits the proteolytic processing of Ci; as a result, the larger form of Ci accumulates in the nucleus and activates the transcription of Hedgehog-responsive genes. Signaling through the latent gene regulatory protein NF-κ B also depends on proteolysis. NF-κ B is normally held in an inactive state by the inhibitory protein Iκ B within a multiprotein complex in the cytoplasm. A variety of extracellular stimuli, including proinflammatory cytokines, trigger a phosphorylation cascade that ultimately phosphorylates Iκ B, marking it for degradation; this enables the freed NF-κ B to enter the nucleus and activate the transcription of its target genes.

Multicellularity and Cell Communication Evolved Independently in Plants and Animals

Although plants and animals are both eucaryotes, they have had separate evolutionary histories for more than a billion years. Their last common ancestor was a unicellular eucaryote that had mitochondria but no chloroplasts. The plant lineage acquired chloroplasts after plants and animals diverged. The earliest fossils of multicellular animals and plants date from almost 600 million years ago. Thus, it seems that plants and animals evolved multicellularity independently, each starting from a different unicellular eucaryote, sometime between 1.6 and 0.6 billion years ago (Figure 15-75).

If multicellularity evolved independently in plants and animals, the molecules and mechanisms used for cell communication will have evolved separately and would be expected to be different. Some degree of resemblance is expected, however, as both plant and animal genes diverged from the set of genes contained by the unicellular eucaryote that was the last common ancestor of plants and animals. Nitric oxide and Ca²⁺ are widely used for signaling in both plants and animals. However, because the genome of Arabidopsis thaliana, a widely studied small flowering plant, has been completely sequenced, we know that there are no homologs of Wnt, Hedgehog, Notch, Jak/STAT, TGF-β, Ras, or the nuclear receptor family in this organism. Similarly, cyclic AMP has not been definitively implicated in intracellular signaling in plants, although cyclic GMP has.

Much of what is known about the molecular mechanisms involved in signaling in plants has come from genetic studies on Arabidopsis. Although the specific molecules used in cell communication in plants often differ from those used in animals, the general strategies are frequently very similar. Enzyme-linked cell-surface receptors, for example, are used in both lineages, as we now discuss.

Receptor Serine/Threonine Kinases Function as Cell-Surface Receptors in Plants

Like animals, plants make extensive use of cell-surface receptors. Whereas most cell-surface receptors in animals are G-protein-linked, most found so far in plants are enzyme-linked. Moreover, whereas the largest class of enzyme-linked receptors in animals is receptor tyrosine kinases, this type of receptor is extremely rare in plants, even though they contain many cytoplasmic tyrosine kinases, and tyrosine phosphorylation and dephosphorylation have important roles in plant cell signaling. Instead, plants seem to rely on a great diversity of transmembrane receptor serine/threonine kinases, which are distinct from this type of receptor used by animal cells. Like the animal receptors, however, they have a typical serine/threonine kinase cytoplasmic domain and an extracellular ligand-binding domain. The most abundant types identified so far have a tandem array of extracellular leucine-rich repeats (Figure 15-76) and are therefore called leucine-rich repeat (LRR) proteins.

There are about 80 LRR receptor kinases encoded by the Arabidopsis genome. One of the best-studied examples is CLAVATA 1 (CLV1), which was originally identified in genetic studies. Mutations that inactivate the protein cause the production of flowers with extra floral organs and a progressive enlargement of both the shoot and floral meristems, which are groups of self-renewing stem cells producing the cells that give rise to stems, leaves, and flowers (discussed in Chapter 21). The extracellular signal molecule for the receptor is thought to be a small protein called CLV3, which is secreted by neighboring cells. The binding of CLV3 to its receptor, CLV1, suppresses meristem growth, either by inhibiting cell division there or, more probably, by stimulating cell differentiation (Figure 15-77A). The intracellular signaling pathway from CLV1 to the cell response is largely unknown, but it includes a serine/threonine protein phosphatase that inhibits CLV1 signaling; also involved is a small GTP-binding protein of the Rho class and a nuclear gene regulatory protein that is distantly related to homeodomain proteins. Mutations that inactivate this gene regulatory protein have the opposite effect of mutations that inactivate CLV1: cell division is greatly decreased in the shoot meristem, and the plant produces flowers with too few organs. Thus, the intracellular signaling pathway activated by CLV1 is thought to normally stimulate cell differentiation by inhibiting the gene regulatory protein that normally inhibits cell differentiation (Figure 15-77B).

A different LRR receptor kinase called BRI1 acts as a cell-surface steroid hormone receptor in Arabidopsis. Plants synthesize a class of steroids called brassinosteroids, because they were originally identified in the mustard family Brassicaceae, which includes Arabidopsis. During development, these plant growth regulators stimulate cell expansion and help mediate responses to darkness. Mutant plants that are deficient in the BRI1 receptor kinase are insensitive to brassinosteroids. Normally, Arabidopsis plants grown in darkness are white and gangly as a result of brassinosteroid signaling; in the absence of brassinosteroid signaling, they become green, as though they were growing in light, and the mature plant is severely dwarfed. As for the other known LRR receptor kinases in plants, the nature of the signal transduction pathway that leads from the receptor to the response remains a mystery.

The LRR receptor kinases are only one of many classes of transmembrane receptor serine/threonine kinases in plants. There are at least six additional families, each with its own characteristic set of extracellular domains. The lectin receptor kinases, for example, have extracellular domains that bind carbohydrate signal molecules. The Arabidopsis genome encodes over 300 receptor serine/threonine kinases, which makes this family of receptors the largest one known in plants. Many of these are involved in defense responses against pathogens.

Ethylene Activates a Two-Component Signaling Pathway

Various growth regulators (also called plant hormones) help to coordinate plant development. They include ethylene, auxin, cytokinins, gibberellins, and abscisic acid. Growth regulators are all small molecules made by most plant cells. They diffuse readily through cell walls and they can act locally or be transported to influence cells further away. Each growth regulator can have multiple effects. The specific effect depends on which other growth regulators are acting, on environmental conditions, on the nutritional state of the plant, and on the responsiveness of the target cell.

Ethylene is an important example. This small gas molecule can influence plant development in various ways, including the promotion of fruit ripening, leaf abscission, and plant senescence. It also functions as a stress signal in response to wounding, infection, flooding, and so on. When the shoot of a germinating seedling, for example, encounters an obstacle, such as a piece of gravel underground in the soil, the seedling responds to the encounter in three ways. First, it thickens its stem, which can then exert more force on the obstacle. Second, it shields the tip of the shoot by increasing the curvature of a specialized hook structure. Third, it reduces the shoot's tendency to grow away from the direction of gravity to avoid the obstacle. This “triple response” is controlled by ethylene (Figure 15-78C, D).

Plants have a number of ethylene receptors, which are all structurally related. They are dimeric transmembrane proteins that are thought to function as histidine kinases. Ethylene receptors have an extracellular domain, which contains a copper atom that binds ethylene, and an intracellular histidine-kinase-like domain. In a manner similar to the two-component signaling pathway involved in bacterial chemotaxis discussed earlier, the kinase domain, when active, phosphorylates itself on histidine and then is believed to transfer the phosphate to an aspartic acid in another domain of the receptor. In bacterial chemotaxis, the binding of an attractant inactivates the receptor. Similarly, the binding of ethylene inactivates ethylene receptors, inhibiting the kinase domain and the downstream signaling pathway emanating from it. In its unbound, active state, the receptor activates the first component of a MAP-kinase signaling module (see Figure 15-56). The activation of this MAP-kinase cascade leads to the inactivation of gene regulatory proteins in the nucleus that are responsible for stimulating the transcription of ethylene-responsive genes. The binding of ethylene to the receptors inactivates this signaling pathway, thereby turning these genes on (Figure 15-78A, B).

Two-component signaling systems operate in bacteria and fungi, as well as in plants, but apparently not in animals. Why animals should have given up this way of signaling remains a mystery.

Plant development is greatly influenced by environmental conditions. Unlike animals, plants cannot move on when conditions become unfavorable; they have to adapt, or they die. The most important environmental influence is light, which is their energy source and has a major role throughout their entire life cycle—from germination, through seedling development, to flowering and senescence. Plants have evolved a large set of light-sensitive proteins to monitor the quantity, quality, direction, and duration of light, as we now discuss.

Phytochromes Detect Red Light, and Cryptochromes Detect Blue Light

Both plants and animals use a variety of light-responsive proteins to sense light of different wavelengths. In plants, these are usually referred to as photoreceptors. However, because the term photoreceptor is also used for light-sensitive cells in the animal retina (see p. 867), we shall use the term photoprotein instead. All photoproteins sense light by means of a covalently attached light-absorbing chromophore, which changes its shape in response to light and then induces a change in the protein's conformation. Animals use some of the same photoprotein families used by plants. The most extensively studied animal photoproteins are the rhodopsins, which are membrane-bound, G-protein-linked proteins that regulate ion channels in the light-sensitive cells of the retina, as discussed earlier.

The best-known plant photoproteins are the phytochromes, which are present in all plants and in some algae. These are dimeric, cytoplasmic serine/threonine kinases that respond differentially and reversibly to red and far-red light: whereas red light usually activates the kinase activity of the phytochrome, far-red light inactivates it. When activated by red light, the phytochrome is thought to phosphorylate itself and then to phosphorylate one or more other proteins in the cell. In some light responses, the activated phytochrome migrates into the nucleus, where it interacts with gene regulatory proteins to alter gene transcription (Figure 15-79). In other cases, the activated phytochrome activates a gene regulatory protein in the cytoplasm, which then migrates into the nucleus to regulate gene transcription. In still other cases, the photoprotein triggers signaling pathways in the cytosol that alter the cell's behavior without involving the nucleus.

Although the phytochromes possess serine/threonine kinase activity, parts of their structure resemble the histidine kinases involved in bacterial chemotaxis. This finding suggests that the plant phytochromes originally descended from bacterial histidine kinases and only later in evolution altered their substrate specificity from histidine to serine and threonine.

Plants sense blue light using two types of photoproteins, phototropin and cryptochromes. Phototropin is associated with the plasma membrane and is partly responsible for phototropism, the tendency of plants to grow toward light. Phototropism occurs by directional cell elongation, which is stimulated by the growth regulator auxin, but the links between phototropin and auxin are unknown.

Cryptochromes are flavoproteins that are sensitive to blue light. They are structurally related to blue-light-sensitive enzymes called photolyases, which are involved in the repair of ultraviolet-induced DNA damage in all organisms, except most mammals. Unlike phytochromes, cryptochromes are also found in animals, where they have an important role in circadian clocks that operate in most cells and cycle with a 24-hour rhythm (discussed in Chapter 7). The cryptochromes do not have a DNA repair activity, but they are thought to have evolved from the photolyases.

In this chapter, we have discussed how extracellular signals can influence cell behavior. One crucial intracellular target of these signals is the cytoskeleton, which determines cell shape and is responsible for cell movements, as we discuss in the next chapter.

Summary

Plants and animals are thought to have evolved multicellularity and cell communication mechanisms independently, each starting from a different unicellular eucaryote, which in turn evolved from a common unicellular eucaryotic ancestor. Not surprisingly, therefore, the mechanisms of signaling between cells in animals and plants have both similarities and differences. Whereas animals rely mainly on G-protein-linked surface receptors, for example, plants rely mainly on enzyme-linked receptors of the receptor serine/threonine type, especially ones with extracellular leucine-rich repeats. A number of growth regulators, including ethylene, help coordinate plant development. Ethylene acts through receptor histidine kinases in a two-component signaling pathway that resembles the pathway used in bacterial chemotaxis. Light has an important role in regulating plant development. These light responses are mediated by a variety of light-sensitive photoproteins, including phytochromes, which are responsive to red light, and cryptochromes and phototropin, which are sensitive to blue light.