CHAPTER 4: Signal Transduction in the Brain

• Signal transduction refers to the processes by which signals between cells carried by neurotransmitters, hormones, trophic factors, and cytokines are converted into biochemical signals within cells.

• Most neurotransmitter receptors can be divided into two classes by their signal transduction mechanism—one class involving activation of an ion channel that is intrinsic to the receptor and the other involving activation of G proteins.

• Signal transduction can alter neuronal function on vastly different time scales ranging from very rapid (millisecond) changes in membrane potential produced by ligand-gated channels to changes over seconds produced by intracellular second messengers and protein kinases.

• Many critical drugs that act on the nervous system are agonists or antagonists at G protein–coupled receptors.

• Although second messengers such as cyclic nucleotides and Ca²⁺ may directly gate ion channels, their major role in intracellular signaling systems is to regulate protein serine–threonine kinases that phosphorylate other proteins.

• The brain contains numerous other types of protein serine–threonine kinases that also play important roles in the regulation of cell function.

• Neurotrophins, such as nerve growth factor and brain-derived neurotrophic factor, interact with a family of receptors termed receptor tyrosine kinases (Trks).

• Certain cytokines act on receptors that activate Janus kinases, which in turn activate a family of transcription factors called signal transducers and activators of transcription (STATs).

• Protein phosphatases, also categorized into serine–threonine or tyrosine subfamilies, reverse the actions of protein kinases and serve critical functions in cell regulation.

• Many intracellular signaling pathways ultimately regulate gene expression.

• Transcription is stimulated when an activator protein displaces nucleosomes, the major component of chromatin, permitting a complex of proteins, called general transcription factors, to bind DNA at a core promoter and recruit RNA polymerase.

• DNA-binding sites for regulatory proteins are called regulatory elements, and the proteins that bind them are called transcription factors.

• Each gene has a unique pattern of cellular expression and response to physiologic signals based on the combinatorial interaction of regulatory elements found within its regulatory regions.

• Eukaryotic cells increase the diversity of proteins that can be produced from a single gene by alternatively splicing the exons within the primary transcript.

• Mature (spliced) messenger RNAs are transported from the nucleus into the cytoplasm, where they are translated to proteins on organelles called ribosomes.

• During and after translation, proteins are processed by cleavage into smaller proteins and by a variety of covalent modifications such as glycosylation.

• Numerous families of transcription factors, with diverse regulatory and functional properties, control gene expression in the CNS under normal and pathological conditions.

Signal transduction refers to the processes by which intercellular signals such as neurotransmitters, neurotrophic factors, circulating hormones, and cytokines produce intracellular biochemical alterations that in turn modify neuronal functioning including the regulation of gene expression. Intercellular signals and their receptors, generally located in the neuronal plasma membrane, represent only a small amount of all information processing in the brain. An understanding of the complex biochemical mechanisms that operate inside neurons and ultimately mediate the actions of all intercellular signals is required to appreciate not only the ways in which the brain responds to drugs and other stimuli but also the ways in which it continuously adapts to a host of environmental changes.

Four general patterns of signal transduction occur in the brain 4–1. One pattern 4–1A, discussed briefly in Chapter 3, involves the binding of neurotransmitter to a multimeric plasma membrane receptor complex that contains a ligand-gated ion channel. Protein–protein interactions tether such ion channels, or receptor ionophores, at proper subcellular locations and often to other signaling proteins. This mechanism of signal transduction is used primarily by the amino acid neurotransmitters at their ionotropic receptors (Chapter 5) as well as by acetylcholine at nicotinic receptors (Chapter 6), serotonin at 5HT₃ receptors (Chapter 6), adenosine triphosphate (ATP) at P2X purinergic receptors (Chapter 8), and heat and many other environmental signals at transient receptor potential (TRP) channels (Chapters 2 and 11). The structural features of ligand-gated ion channels are discussed in detail in the chapters devoted to these transmitters.

A second pattern, also briefly described in Chapter 3, is characterized by the binding of neurotransmitter to a plasma membrane receptor that couples with a guanine nucleotide–binding protein or G protein 4–1B. Most neurotransmitters bind to this superfamily of G protein–coupled receptors, as do several cytokines. All such receptors have a seven-transmembrane domain structure, whose N terminus faces the extracellular space and whose C terminus faces the cytoplasm; other structural features of these receptors are discussed in detail in Chapter 6, wherein the β-adrenergic receptor is presented as a prototype. The binding of ligands to these receptors initiates a range of biologic effects on target neurons, which include direct G protein regulation of certain ion channels (see 4–1B, left) and the triggering of complex cascades of intracellular messengers. Activation of intracellular messenger pathways leads to the generation of second messengers and the regulation of protein phosphorylation, and ultimately to diverse physiologic responses to extracellular stimuli 4–2, including regulation of ion channels (see 4–1B, right). As will become evident, protein phosphorylation is the major molecular currency of intracellular signal transduction pathways. Protein phosphorylation describes a process by which protein kinases add phosphate groups to specific target proteins, and protein phosphatases (PPPs) remove these phosphate groups. Phosphate groups, because of their large size and negative charge, alter the conformation and overall charge of a protein and hence its function.

A third pattern of signal transduction involves the direct activation of a class of protein kinases, called protein tyrosine kinases, which phosphorylate proteins on tyrosine residues 4–1C. This type of signaling is used by most types of neurotrophic factors and cytokines. In some cases the neurotrophic factor receptor and the protein kinase reside in a single protein; in other cases the receptor must recruit a cytoplasmic kinase to effect its signaling (see 4–2). Activation of the protein tyrosine kinase triggers cascades of further protein phosphorylation that ultimately lead to the many effects of neurotrophic factors on brain function, including the regulation of ion channels.

A fourth pattern characterizes transduction by all known steroid hormones and certain other lipophilic extracellular signals that cross the plasma membrane and activate receptors in the neuronal cytoplasm 4–1D. On binding their ligand, these cytoplasmic receptors translocate to the nucleus, where they bind DNA and function as transcription factors. Thus, these receptors can be considered ligand-activated transcription factors.

Much of the specificity and rapidity of cell signaling comes from the spatial apposition of signaling proteins in microdomains of the cell’s plasma membrane termed lipid rafts. These specializations, which are rich in cholesterol and sphingolipids, organize sets of signaling proteins into discrete compartments, which greatly facilitates complex signaling cascades.

G Proteins

G proteins perform a central function in the process of transmembrane signaling in the nervous system. These proteins were named because of their ability to bind the guanine nucleotides, guanosine triphosphate (GTP), and guanosine diphosphate (GDP). G proteins couple receptors to specific intracellular effector systems (see 4–1). Three major types of G protein are involved in the transduction of signals produced by neurotransmitter binding: G_s, G_i/o, and G_q 4–1. Other subtypes serve more specialized functions in discrete cell types. Each type of G protein is a heterotrimer composed of single α, β, and γ subunits. Individual α_s, α_i, α_q, etc, subunits are primarily responsible for the unique functions of the G proteins that contain them 4–1. Additionally, several types of β and γ subunits complex with one another and differentially associate with subtypes of α subunits to confer still greater degrees of specificity on intracellular signaling proteins. The functional cycle of G proteins is shown schematically in 4–3. All eukaryotic cells also contain a distinct class of monomeric guanine nucleotide–binding proteins that subserve numerous critical roles in the regulation of cell function; these are termed small-molecular-weight G proteins and are described in 4–1.

Several bacterial toxins produce their pathogenic effects by regulating the activity of specific G proteins through a process termed ADP-ribosylation (see 4–1). This process involves the addition of an ADP-ribose group from nicotinamide adenine dinucleotide (NAD) to an amino acid residue in the G protein. Cholera toxin ADP-ribosylates and irreversibly activates G_αs by inhibiting its GTPase activity, whereas pertussis toxin ADP-ribosylates and inactivates G_αi and G_αo by stabilizing their association with βγ subunits. Diphtheria toxin ADP-ribosylates and inactivates a small G protein, known as eukaryotic elongation factor 2, which regulates ribosomal function (see 4–1). G proteins also are the targets of certain insect venoms. The best characterized is mastoparan, a 10 amino acid–long peptide contained in wasp venom. The peptide stimulates the GTPase activity of G_αi subunits, thereby shortening the duration of the GTP-bound activated subunits.

G protein function can be regulated by several modulatory proteins. A neurotransmitter receptor may be seen as one type of modulatory protein, specifically a guanine nucleotide exchange factor (GEF), because it triggers the release of GDP from the α subunit. Regulators of G protein signaling (RGS proteins) are GTPase-activating proteins (GAPs) that activate the GTPase activity intrinsic to the α subunits of G proteins; thus, these proteins inhibit G protein function by shortening the duration of the signals from both the activated GTP-α subunit and the free βγ subunits. All G protein α subunits, except for G_αs, are known to be regulated in this manner. More than 20 subtypes of RGS proteins, each with a characteristic expression pattern in the brain, have been identified. Some RGS proteins also exert scaffolding functions by virtue of binding other signaling proteins that together form signaling complexes for neurotransmitter receptors. Phosducin is another modulatory protein that binds to βγ subunits and thereby competes with these subunits for binding to α subunits and possibly to effector proteins.

G Protein Regulation of Ion Channels

Many types of neurotransmitter receptors regulate ion channels by means of G proteins. The process by which this regulation occurs is best established for neurotransmitter receptors that couple to G_αi or G_αo. Via G protein coupling, these receptors activate inwardly rectifying K⁺ channels (GIRKs) or inhibit voltage-gated Ca²⁺ channels, depending on the cell type involved. This regulation occurs primarily through the βγ subunits of G proteins, which directly open, or gate, GIRKs and limit the opening of Ca²⁺ channels in response to membrane depolarization.

Second Messengers

G proteins also transduce the activation of neurotransmitter receptors by neurotransmitter binding to altered intracellular levels of second messengers in target neurons. Prominent second messengers in the brain include cyclic adenosine monophosphate (cAMP), cyclic guanosine monophosphate (cGMP), Ca²⁺, nitric oxide (NO), and the major metabolites of both phosphatidylinositol—inositol triphosphate (IP₃) and diacylglycerol (DAG)—and arachidonic acid, such as prostaglandins. Altered levels of second messengers mediate the effects of receptor activation on many types of ion channels and on numerous physiologic responses.

Cyclic nucleotides

cAMP and cGMP are classified as cyclic nucleotides because they are synthesized from ATP and GTP, respectively, through the formation of a cyclic phosphodiester ring 4–4. Neurotransmitter receptors coupled via G_s stimulate cAMP levels via activation of adenylyl cyclase, the enzyme responsible for the synthesis of cAMP 4–5. In contrast, receptors coupled via G_i inhibit adenylyl cyclase. Nine forms of adenylyl cyclase (types I–IX) have been identified, each of which exhibits a distinctive pattern of expression in brain and peripheral tissues 4–6. All but type IX are activated by the natural plant molecule, forskolin.

Neurotransmitters regulate cGMP levels by means of two mechanisms. A very small number of plasma membrane receptors, such as atrial natriuretic peptide receptors, contain guanylyl cyclase that is activated when a ligand binds to the receptor. In most cases, however, guanylyl cyclase is a cytosolic enzyme that is activated by NO (see 4–5). NO, in turn, is generated via nitric oxide synthase (NOS), which is activated by Ca²⁺ in conjunction with the Ca²⁺-binding protein calmodulin. Accordingly, neurotransmitters that increase intracellular Ca²⁺ levels typically increase cGMP levels by generating NO. Organic nitrates that are used as vasodilators in the treatment of ischemic heart disease work by generating free NO in vascular smooth muscle, which increases cellular cGMP levels and in turn leads to muscle relaxation. NO also regulates cell function through the S-nitrosylation of other proteins (Chapter 8). Because NO is a soluble gas, it can diffuse out of the neuron in which it is synthesized and stimulate guanylyl cyclase in neighboring cells, thereby increasing cGMP levels. Thus, NO functions as both an intercellular and an intracellular messenger in the brain. The potential utility of NOS inhibitors in the treatment of stroke is described in Chapter 20. Linaclotide is a small peptide agonist of guanylyl cyclase 2C, which is approved for the treatment of irritable bowel syndrome.

cAMP and cGMP are enzymatically degraded by phosphodiesterases (PDEs), which are expressed in numerous forms in brain and other tissues. As outlined in 4–2, these enzymes differ in their relative selectivity for cAMP and cGMP and in their regulation by other cellular signals, such as by the cyclic nucleotides themselves or by Ca²⁺/calmodulin. At high concentrations, caffeine and related methylxanthines inhibit PDE activity, which may contribute to some of the pharmacologic effects of these drugs, particularly at higher doses.

Nonselective PDE inhibitors, which affect many forms of the enzyme, are associated with pervasive side effects, but progress has been made in the development of selective PDE inhibitors. Sildenafil (Viagra) and related drugs, specific inhibitors of PDE5, are prescribed for the treatment of erectile dysfunction. PDE5 is concentrated in vascular smooth muscle and especially, but not exclusively, in the vasculature of the penis. Considerable effort also has been made toward developing PDE inhibitors that are selective for isoforms of the enzyme expressed predominantly in the brain. Rolipram, for instance, inhibits all isoforms of PDE4; this drug initially showed promise as an antidepressant, but its clinical utility was limited by side effects such as nausea. However, because PDE4 comprises many subtypes (see 4–2), an inhibitor of one subtype may lead to the development of an effective antidepressant without rolipram’s side effects. Another PDE4 inhibitor, ibudilast, which also acts on other PDE subtypes, is marketed as an anti-inflammatory agent in Japan. It is being considered for several CNS indications. There is interest as well in generating inhibitors of PDE10A, which is highly enriched in striatum. Such inhibitors are being evaluated for the treatment of schizophrenia (Chapter 17).

Calcium

The regulation of intracellular Ca²⁺ levels by neurotransmitter receptors occurs via two major mechanisms 4–7. Neurotransmitter receptor activation can alter the flux of extracellular Ca²⁺ into neurons in several ways: (1) Ca²⁺ passes directly through certain subtypes of ligand-gated channels, such as activated N-methyl-d-aspartate (NMDA) glutamate receptors and nicotinic cholinergic receptors (Chapters 5 and 6); (2) specific voltage-gated Ca²⁺ channels are inhibited by G_i, as previously outlined; (3) the depolarization of a neuron, which causes the activation of voltage-gated Ca²⁺ channels, leads to large increases in intracellular Ca²⁺ levels (Chapter 2); or (4) the activation of other second messenger systems may alter Ca²⁺ channel properties; for example, cAMP and neurotransmitters that act through cAMP can modulate voltage-gated Ca²⁺ channels by regulating their phosphorylation.

The second major mechanism by which neurotransmitters can increase intracellular levels of free Ca²⁺ is through regulation of the phosphatidylinositol system and subsequent actions on intracellular Ca²⁺ stores (see 4–7). Such regulation is possible because many types of neurotransmitter receptors are coupled through G proteins to an enzyme termed phospholipase C (PLC). Receptor activation of PLC is mediated most often by subtypes of G_q, whose βγ subunits are primarily responsible for this regulation.

Phospholipase C and the phosphatidylinositol pathway

PLC catalyzes the breakdown of phosphatidylinositol, which results in the generation of two lipid molecules that function as second messengers in the brain: IP₃ and DAG (4–8 and 4–9). The most important forms of PLC in the brain are designated β and γ 4–10. The β form is predominantly responsible for mediating the effects of neurotransmitters that are coupled to G_q, and the γ form is predominantly responsible for mediating the effects of neurotrophic factors on this enzyme.

PLC action on phosphatidylinositol results in the generation of free inositol-1-phosphate that, through a series of lipid phosphorylation events, is converted into active IP₃ or inositol-1,4,5-triphosphate 4–9. IP₃ is subsequently recycled by means of successive dephosphorylation into inositol, which cells use to replenish phosphatidylinositol in their membranes. Lithium, which remains an important drug in the treatment of mania and other mood disorders, inhibits several of these inositol phosphatases, although it is not known whether this action is related to the drug’s clinical effects (Chapter 17). IP₃ acts on an IP₃ receptor, which functions to release Ca²⁺ from intracellular organelles such as the endoplasmic reticulum. Thapsigargin inhibits this effect. Some organelles also contain a related ryanodine receptor, which triggers the release of Ca²⁺ from intracellular stores in response to a rise in Ca²⁺ itself. This process is inhibited by rapamycin, although this drug is best known for its inhibition of mammalian target of rapamycin (mTOR) (see below). DAG, the other metabolite of phosphatidylinositol, functions as a second messenger in cells by activating protein kinase C (PKC), as will be discussed below.

Arachidonic acid metabolites

The prostaglandin and leukotriene families of intracellular messengers play important roles in the regulation of signal transduction in the brain and elsewhere. Prostaglandins and leukotrienes are generated by a complex biochemical cascade, depicted in 4–11, initiated by phospholipase A₂, which cleaves membrane phospholipids to yield free arachidonic acid. Next, arachidonic acid is cleaved by cyclooxygenase to yield, after numerous enzymatic steps, several types of prostaglandins and other cyclic endoperoxides, such as prostacyclins and thromboxanes, or it is cleaved by lipoxygenase to yield the leukotrienes 4–11.

The varied biologic activities of these endoperoxides and leukotrienes serve to regulate adenylyl cyclase, guanylyl cyclase, ion channels, protein kinases, and other cellular proteins, either directly or by serving as the endogenous ligands for G protein–coupled receptors.

Receptors for the leukotrienes are termed BLT or LT receptors, and those for prostaglandins and thromboxanes are termed prostanoid receptors. Several prostanoid receptors are known, and each is named for the arachidonic acid metabolite with the highest affinity for that receptor. EP receptors, for example, are activated preferentially by prostaglandin E₂ (PGE₂). All LT receptors are coupled by G_q and presumably act by stimulating the PLC pathway. In contrast, the numerous subtypes of prostanoid receptors are coupled by G_q, G_s, or G_i.

Aspirin, acetaminophen, and all nonsteroidal anti-inflammatory drugs (NSAIDs) such as ibuprofen exert their antipyretic, analgesic, and anti-inflammatory effects by inhibiting cyclooxygenase, which has both constitutive (COX₁) and inducible (COX₂) forms. COX₁ is expressed in many locations, including the upper gastrointestinal tract, where nonselective COX inhibitors block the protective functions of prostaglandins. Such interference increases the risk of peptic ulceration, a common side effect of these drugs. Selective COX₂ inhibitors, such as rofecoxib, that are analgesic and anti-inflammatory but with less disturbance of the gastrointestinal tract were in wide use until recent reports of cardiovascular toxicity (Chapter 11).

Ligand-directed signaling

Classic studies in neuropharmacology assumed that different agonists at a single receptor would, of course, trigger the receptor to signal through its one and only effector pathway, a β-adrenergic receptor through G_s and cAMP, a 5HT_2A receptor through G_q and PLC, and so on. Only the agonist’s potency and efficacy would vary, not its qualitative functional effect. In fact, we now know that many receptors can signal through multiple pathways (eg, a β-adrenergic receptor can signal not only through G_s and cAMP but also through arrestin pathways; see 4–3) and that receptors can display qualitatively different signaling effects depending on the agonist. Such ligand-directed or biased signaling, discussed further in Chapters 1 and 6, has greatly complicated efforts to understand drug action, but has also raised the very real possibility of creating novel agonists to a receptor that target one signaling pathway versus another and thus dialing in wanted efficacy while avoiding unwanted side effects.

Most of the effects of intracellular second messengers are produced through their regulation of protein phosphorylation. Protein phosphorylation may inactivate a neurotransmitter receptor, cause an ion channel to open more or less readily, or enable the synthesis of a neurotransmitter much more rapidly. Indeed, virtually every class of neuronal and glial protein is regulated through phosphorylation 4–3, and consequently many prominent forms of CNS plasticity are mediated by such phosphorylation events. Although proteins are covalently modified in many ways—for example, by ADP-ribosylation, acylation, and glycosylation, among many others—none of these mechanisms is as widespread or as readily subject to regulation by synaptic and hormonal stimuli as phosphorylation.

Second Messenger–Dependent Protein Phosphorylation Cascades

Protein kinases

Protein kinases can be classified according to their sequence homologies 4–4; however, they are presented here based on shared regulatory and functional characteristics. Among the best characterized protein kinases in the brain are those activated by the second messengers cAMP, cGMP, Ca²⁺, and DAG. All of these protein kinases, named for the second messengers that activate them, phosphorylate substrate proteins on serine or threonine residues, and thus are referred to as protein serine–threonine kinases.

The brain contains one major type of cAMP-dependent protein kinase (termed protein kinase A [PKA]) and of cGMP-dependent protein kinase (termed protein kinase G [PKG]). The structure and function of these enzymes are shown in 4–12. Virtually all cellular actions of cAMP in eukaryotic cells, except for those related to cyclic nucleotide–gated channels (CNG and HCN channels; Chapter 2), are mediated by the activation of PKA (see 4–5). PKA’s integral role in these cellular processes stems from its ubiquitous expression and its ability to phosphorylate a wide array of substrate proteins, including many types of ion channels, receptors, cytoskeletal proteins, and nuclear transcription factors. It is believed that many of the cellular actions of cGMP are mediated by the activation of PKG (see 4–5), however; less is known about the function of PKG, in part because it is not as widely expressed as PKA. Like cAMP, cGMP gates certain ion channels and directly regulates several forms of PDE (see 4–2).

There are two major classes of Ca²⁺-dependent protein kinases (see 4–7). One class is activated by Ca²⁺ in conjunction with calmodulin and is referred to as Ca²⁺/calmodulin-dependent protein kinase or CaM-kinase. The other is activated by Ca²⁺ in conjunction with DAG and other lipids and is referred to as PKC. The phorbol esters, which activate all known PKC isoforms, have been a valuable tool in PKC research. The brain contains several forms of each of these enzymes, which exhibit very different regulatory properties and are expressed in distinct neuronal cell types throughout the nervous system. Examples of this diversity are shown in 4–5 and 4–6, while some of the unique structural properties of PKC enzymes are depicted in 4–12. In addition to regulating CaM-kinases and PKC, Ca²⁺ is known to bind to and regulate a large number of proteins in the cell.

Certain Ca²⁺-dependent protein kinases engage in complex modes of regulation. CaM-kinase II, for example, can autophosphorylate itself and in turn create a form of the enzyme that remains active even after Ca²⁺ levels have returned to normal 4–12. Such a sustained period of kinase activation may represent a molecular mechanism that contributes to learning and memory (Chapter 14). CaM-kinases I and IV can be activated by another kinase known as CaM-kinase kinase, which in turn is activated by Ca²⁺/calmodulin and inhibited by PKA phosphorylation. The N-terminal truncated form of PKCζ (also referred to as PKMζ) is a constitutively active PKC enzyme, which is induced largely in brain and lacks the autoregulatory domain found in other PKC isoforms 4–12. Induction of PKMζ has been implicated in the regulation of long-term memory.

Protein kinase–anchoring proteins

Protein kinases are localized to distinct subcellular regions in neurons and glia via specific anchoring proteins. Such proteins sequester each type of kinase to the region of a cell, such as the postsynaptic specialization or cell nucleus, that requires its function, and to bring the kinases in close approximation to some of their protein substrates. Examples include the A kinase–anchoring proteins (AKAPs) for PKA and receptors for activated C kinase (RACKs) for PKC.

Protein kinase inhibitors

There has been intense interest in the development of protein kinase inhibitors (PKIs) for use as research tools and as potential therapeutic agents. The most specific of these are peptide inhibitors, which are directed at the active site of a particular enzyme. The prototype, a naturally occurring 17-kDa protein known as PKI, is a highly specific inhibitor of PKA that is expressed in many tissues. Small molecule PKIs have been explored in the treatment of numerous diseases. Their development is furthest along for several cancers, and remains more hypothetical for neurodegenerative disorders, such as Alzheimer disease (Chapter 18). Examples of small molecular PKIs used experimentally in neuroscience research are listed in 4–7.

Protein phosphatases

The brain contains a relatively small number of PPPs that undo the actions of the second messenger–dependent protein kinases. These PPPs, termed protein serine–threonine phosphatases, differ in their regional distribution in the brain and in their regulatory properties 4–8.

Neurotransmitters can regulate PPPs, and in turn influence protein phosphorylation, by means of three primary mechanisms. PPP3 (formerly phosphatase 2B), also referred to as calcineurin, is activated directly when it is bound to Ca²⁺/calmodulin. Thus, neurotransmitters that alter cellular Ca²⁺ levels can influence the phosphorylation of cellular proteins by influencing calcineurin activity. Calcineurin activity is also regulated by immunophilins. This class of proteins, including cyclophilin and FK506-binding protein (FKBP), was originally discovered in lymphocytes and is the target of several important immunosuppressive agents, such as cyclosporin and FK506. Immunophilins are also expressed in the brain, although less is known about their physiologic regulation of neuronal function. However, when they are bound to FK506 and related drugs, the immunophilins inhibit calcineurin activity.

Another mechanism of PPP regulation is indirect and involves a class of proteins referred to as PPP inhibitors, examples of which are listed in 4–8. Among these, the best known are phosphatase inhibitors 1 and 2 and dopamine- and cAMP-regulated phosphoprotein of 32 kDa (DARPP-32). DARPP-32 is especially concentrated in neurons of the striatum, which receive dense dopaminergic innervation. These proteins are highly potent inhibitors of protein phosphatase 1 (PPP1), the predominant protein serine–threonine phosphatase in mammalian cells. Phosphorylation of most of these inhibitor proteins by PKA or by other protein kinases greatly enhances their inhibitory activity. In neurons that contain these inhibitors, neurotransmitters that alter cellular cAMP levels can influence the phosphorylation of cellular proteins by altering PPP1 activity as well as by activating PKA. This process is illustrated in 4–13.

The third mechanism of regulation is through regulatory subunits, which can be tissue-specific and control the subcellular localization of phosphatases as well as their substrate specificity. For example, spinophilin functions as a PPP1-anchoring protein that selectively tethers the enzyme to dendritic spines. In contrast, the PPP1 nuclear targeting subunit (PNUTS) targets the same phosphatase to the cell nucleus. PPP2 (formerly protein phosphatase 2A) is composed of a catalytic and two regulatory subunits, numerous subtypes of which are known. Mutations in the regulatory subunit B cause spinocerebellar ataxia 12, one of several neurodegenerative diseases caused by an expansion of CAG repeats within the untranslated region (UTR) of the gene (Chapter 18).

Studies of the physiologic functions of phosphatases have been facilitated by the identification of natural small molecules that are relatively specific phosphatase inhibitors. Among these, the best known is okadaic acid, a toxin that can accumulate in shellfish and inhibits PPP1 and PPP2.

As mentioned previously, the intracellular signal transduction pathways for most neurotrophic factors (Chapter 8) involve the activation of protein tyrosine kinases (see 4–4), which occurs by different mechanisms depending on the neurotrophic factor involved. In turn, protein tyrosine phosphatases undo those phosphorylation reactions (see 4–8). These pathways illustrate the extraordinary complexity of intracellular regulation, and keeping track of them can be daunting, even for scientists who study them. Although neurotrophic factors and their signaling pathways typically have been studied because of their central role in neural development and differentiation, research has confirmed that they are critical for mediating responses to a wide array of external stimuli throughout the adult life of an organism.

Given the intricate role of protein tyrosine kinases and phosphatases in the control of cell growth, it is not surprising that there is considerable effort at generating drugs that target these enzymes for the treatment of cancer and inflammatory diseases. Several marketed tyrosine kinase inhibitors are monoclonal antibodies directed against receptor tyrosine kinases, although an increasing number of small molecule inhibitors are being introduced (see 4–7). Likewise, work is under way to develop small molecules that target protein tyrosine phosphatases.

Neurotrophin Signaling Pathways

Neurotrophic factor signal transduction cascades are best established for the neurotrophins, as summarized in 4–14 (Chapter 8). Neurotrophins bind to a plasma membrane receptor, termed Trk, and in turn trigger the dimerization, phosphorylation, and intrinsic protein tyrosine kinase activity of Trk (see 4–4). Once activated, tyrosine kinase signals via three main intracellular pathways—the Ras-Raf-MEK-ERK, IRS-PI3-kinase-AKT, and PLCγ pathways 4–14—which are distinguished by several features. First, the Ras-Raf-MEK-ERK and IRS-PI3-kinase-AKT pathways involve cascades of multiple serine–threonine protein kinases that phosphorylate and activate successive kinases. An example of these cascades is shown in 4–15. This leads to an extraordinary level of amplification in the cell. Second, all three pathways culminate in the phosphorylation of numerous cell proteins, which mediate the diverse effects of neurotrophins on neuronal function. Third, each of these pathways utilizes highly specific modes of protein–protein interactions, termed src homology (SH2) domains. Elaboration of SH2 domains in neurotrophin signaling led to a series of major discoveries of diverse types of protein–protein interaction domains (eg, PDZ, leucine zipper domains) that are integral to cell signaling 4–2.

A pathway that is functionally parallel to the Ras-Raf-MEK-ERK pathway in neurotrophin signaling is activated by certain cytokines and noxious stimuli 4–15. Tumor necrosis factor-α or UV light, for example, leads to the activation of the small G protein Rac. In response, Rac activates a cascade of protein kinases that culminates in the phosphorylation and activation of Jun N-terminal kinase (JNK). JNK phosphorylates the transcription factors Jun and activating transcription factor-2 (ATF2) and thereby enables them to regulate gene expression. Because this pathway is involved in responses to cellular stress, JNK is often referred to as a stress-activated protein kinase, or SAP-kinase. Another SAP-kinase is a 38-kDa protein known as p38.

While these various signaling pathways are best understood for mediating the functional effects of neurotrophins, it is important to emphasize they are also highly regulated by the cell’s G protein and second messenger pathways. G protein βγ subunits, PKA, and PKC, for example, are known to influence the Ras-Raf-MEK-ERK pathway, which highlights the enormous cross-talk that occurs among a cell’s complex web of intracellular signaling networks.

JAK–STAT Pathway

Another prominent family of cytokines and hormones employs a very different mechanism of signal transduction. This family includes ciliary neurotrophic factor (CNTF) and interleukin-6 (IL-6), among many others (Chapter 8). Each interacts with a specific plasma membrane α receptor named for the ligand (eg, CNTF α receptor, IL-6 α receptor). The α subunits of these receptors subsequently form heterotrimeric complexes with other proteins (Chapter 8). After these complexes are formed, the tripartite receptor associates with a protein tyrosine kinase called Janus kinase (JAK), which normally exists in the cytoplasm in its inactive form. The association of JAK with the receptor complex leads to JAK’s activation and to the phosphorylation and activation of a family of transcription factors called signal transducers and activators of transcription (STATs). The activation of STATs in turn leads to many of the long-term effects of the cytokine on cell function.

GDNF Signaling Pathway

A similar pattern characterizes signaling by glial cell line–derived neurotrophic factor (GDNF) and the related neurotrophic factors known as neurturin and persephin. These factors are distant relatives of the transforming growth factor-β family and are of particular interest because of their potent neurotrophic effects on dopaminergic and other neurons. GDNF, neurturin, and persephin each bind to a unique receptor that dimerizes and interacts with the tyrosine kinase Ret (Chapter 8), which then mediates the diverse actions of these factors on cell function.

Cytoplasmic Protein Tyrosine Kinases

The first protein tyrosine kinase discovered was Src, which we now know is a prototype of a large family of such kinases that exist in the cell cytoplasm and in general are not physically part of a plasma membrane receptor complex (see 4–4). Src family members that are expressed at high levels in neurons include Fyn, Yes, Lck, and Lyn, in addition to Src itself. By phosphorylating many types of substrates, such as ion channels, glutamate receptor subunits, synaptic vesicle proteins, and transcription factors, Src family kinases have been implicated in the regulation of numerous neuronal processes, including synaptic transmission, neurotransmitter release, cell survival, and gene expression.

Protein Tyrosine Phosphatases

Interestingly, far more types of protein tyrosine phosphatases occur in the brain than do types of protein serine–threonine phosphatases (see 4–8). Some protein tyrosine phosphatases are selective for ERK and related MAP-kinases. Other enzymes are membrane associated and contain extracellular domains that may represent recognition sites for as yet unknown intercellular signals. Still other forms of protein tyrosine phosphatases are soluble and are present throughout the neuronal cytoplasm. Certain membrane- and cytosol-associated forms that exhibit highly restricted distributions in the nervous system may subserve specialized functions. For example, the striatal-enriched phosphatases (STEPs) are expressed at high levels in striatal medium spiny neurons and contribute to the unique features of these cells. Although inhibitors of protein tyrosine phosphatases might be expected to promote the survival of neurons, this remains an unproven hypothesis. Vanadate, which inhibits most if not all protein tyrosine phosphatases, has been used as an experimental tool to study these enzymes in neural phenomena. Since mutations in numerous forms of protein tyrosine phosphatases are linked to several cancers, there is considerable interest in generating drugs to counter the effects of the mutations.

Other Protein Phosphorylation Cascades

Second messenger–dependent protein kinases and the protein kinases that function in neurotrophic factor signaling cascades represent a relatively small fraction of the ~500 protein kinase genes that have been identified in mammalian cells 4–4. Research is now revealing how most of these other kinases affect neural functioning. G protein receptor kinases (GRKs), for example, phosphorylate G protein–coupled receptors and mediate receptor desensitization 4–3. Many protein kinases were first characterized according to their role in basic cellular functions, such as cell division, protein synthesis, or intermediary metabolism; however, recent evidence indicates that most of these enzymes operate in particular signal transduction cascades in the brain and contribute to synaptic transmission and neuronal adaptation. Key examples are described here.

mTOR is a protein serine–threonine kinase, first identified as the main target of rapamycin, a natural antifungal agent synthesized by certain bacteria. mTOR exists in two complexes, termed TORC1 and TORC2, distinguished by unique subunits. TORC1 is downstream of AKT (see the section “Neurotrophin Signaling Pathways” and 4–14) and plays a central role in the regulation of protein synthesis 4–16. In more recent years it has been shown to contribute to synaptic and other forms of neuronal plasticity. Glycogen synthase kinase-3β (GSK3β), which is also upstream of TORC1, has been implicated in several neuropsychiatric disorders: for example, its hyperphosphorylation of the microtubule-associated protein tau might contribute to Alzheimer disease and other dementias (Chapter 18), while its inhibition by lithium might contribute to the drug’s antimanic activity (Chapter 17). As well, mutations in the genes that encode the tuberous sclerosis complex cause tuberous sclerosis, characterized by nonmalignant tumors in the brain and other organs, while mutations in several other genes that encode proteins in this signaling pathway cause rare forms of autism spectrum disorders. Much less is known about the function of the TORC2 complex in brain and peripheral tissues.

Cyclin-dependent kinases (CDKs) were discovered for their role in controlling the cell cycle. However, one isoform, CDK5, is most enriched in brain where it is controlled not by cyclin but by another regulatory subunit, termed p35. CDK5 is a key signaling molecule in neurons throughout the brain, where it has been shown to modulate synaptic transmission and contribute to activity-dependent plasticity. One of its many substrates is DARPP-32; phosphorylation of DARPP-32 by CDK5 converts DARPP-32 into a potent inhibitor of PKA. Like GSK3β, it too contributes to the hyperphosphorylation of tau (Chapter 18).

Casein kinase 1 (CK1) refers to a family of enzymes that are central regulators of diverse cellular functions. They control WNT signaling pathways, important in cellular differentiation and adult neuronal function (Chapter 17). They phosphorylate several proteins that comprise the molecular “clock” and thereby influence circadian rhythms (Chapter 13). They also phosphorylate numerous proteins involved in gene transcription and protein translation.

All living cells depend on the regulation of gene expression by extracellular signals for their development, homeostasis, and adaptation to the environment. Indeed, many signal transduction pathways function primarily to modify transcription factors that alter the expression of specific genes. Thus, neurotransmitters, growth factors, and drugs change patterns of gene expression in cells and in turn affect many aspects of nervous system functioning, including the formation of long-term memories. Many drugs that require prolonged administration, such as antidepressants and antipsychotics, trigger changes in gene expression that are thought to be therapeutic adaptations to the initial action of the drug.

The mechanisms underlying the control of gene expression are becoming increasingly well understood. These are subject to various forms of dynamic regulation in the cell, including structural changes in chromatin, transcription of DNA into RNA, splicing of RNA into messenger RNA (mRNA), editing and other covalent modifications of mRNA, translation of mRNA into protein, and posttranslational modification of a protein into its mature, functional form. Although we have learned the molecular details of some of these regulatory processes, such information has not yet been exploited for the medical treatment of CNS disorders. This section provides a detailed overview of the regulation of gene expression in cells, underscoring its profound impact on neural and behavioral plasticity. Central to this section is the belief that the therapeutic effects of drugs that influence the CNS often depend on alterations in gene expression, as well as the expectation that novel therapeutic agents for neuropsychiatric disorders will one day target some of the mechanisms that govern the transcription and translation of genes into mature proteins.

Regulation of Gene Expression by Chromatin

In eukaryotic cells, DNA is contained within a discrete organelle called the nucleus, which is the site of DNA replication and transcription. Transcriptionally quiescent regions of DNA are tightly packed into a coil, whereas regions characterized by active transcription may be more than 1000-fold more extended. Chromosomes are extremely long molecules of DNA, which are wrapped around histone proteins to form nucleosomes, the major subunits of chromatin. Chromatin plays a critical role in transcriptional regulation, because it can repress gene expression by inhibiting the ability of transcription factors to access DNA. In fact, chromatin ensures that genes are inactive unless their expression is required 4–17. To activate gene expression, cells must attenuate nucleosome-mediated repression of an appropriate subset of genes by means of activator proteins that modify chromatin structure. This activation process, which involves transcription factors, histones, and many other regulatory proteins, remodels chromatin and opens up regions of DNA to permit their transcription into RNA.

Activation or repression of chromatin is mediated in part via the covalent modification of histone proteins, including their acetylation, methylation, and phosphorylation, among many other chemical changes (eg, ubiquitylation, glycosylation, ADP-ribosylation) 4–18. Histone acetylation is associated with an opening of chromatin and activation of gene expression. Histone methylation is associated with either gene activation or repression depending on the amino acid residue undergoing methylation. DNA itself can undergo methylation (typically on cytosines), which usually mediates repression of gene expression, although 5-hydroxymethylation of cytosines might promote gene expression. The enzymatic machinery mediating these changes in chromatin structure is becoming increasingly well understood. For example, histone acetyltransferases (HATs) catalyze histone acetylation, and histone methyltransferases catalyze histone methylation, with these reactions reversed by histone deacetylases (HDACs) and histone demethylases, respectively. Likewise, DNA methyltransferases (DNMTs) catalyze 5-methylcytosine, which is converted to 5-hydroxymethylcytosine and then to fully demethylated cytosine by ten eleven translocation (TET) proteins, which likely work in conjunction with several other DNA repair enzymes. Regulation of histone and DNA modifications has been implicated in recent years in a wide range of neuroscientific phenomena, including learning and memory and the actions of several classes of psychotropic drugs. Indeed, there is a great deal of interest in the potential therapeutic utility of HDAC inhibitors, which have been shown to promote learning and memory and exert antidepressant-like and neuroprotective effects in animal models. Most available HDAC inhibitors, for example, sodium butyrate, trichostatin A, and valproate, are highly nonselective and affect many other types of proteins. Even more selective HDAC inhibitors—for example, SAHA (vorinostat) and MS-275—can affect several HDACs, but work is under way to synthesize inhibitors specific for a given subtype. HATs represent another putative drug target, although less explored to date than HDACs. Curcumin, a major ingredient in the curry spice, turmeric, is an inhibitor of HAT activity, among many other actions, and is purported to have beneficial anticancer and anti-inflammatory effects in open trials.

Research has demonstrated that genetic defects in several chromatin remodeling proteins cause several neurologic and psychiatric syndromes 4–9. An example is Rett syndrome, an autism spectrum disorder, caused by mutations in a gene whose protein product interacts with methylated DNA 4–4. Recent discoveries have directly implicated DNA methylation in responses to pharmacologic agents. Temozolomide is used to treat glioblastoma; it acts by damaging DNA. The enzyme encoded by the O-6-methylguanine-DNA methyltransferase (MGMT) gene can repair this damage. In some tumors, this gene is itself methylated that suppresses its expression and renders the tumors more sensitive to temozolomide.

Regulated Steps of Transcription

Genes contain a coding region that specifies the sequence of the encoded mRNA and its protein along with much longer regulatory regions that determine when, where, and to what extent that gene is expressed. These regulatory regions have been termed promoter or enhancer regions. The best characterized regulatory regions of genes are located immediately upstream from (or 5´ to) their coding regions. However, DNA sequences located 3´ of the coding region, and even within the coding region, have been shown to contain regulatory information. Only a small percentage of chromosomal DNA in the human genome (approximately 4%) is responsible for encoding the roughly 20,000 genes that encode RNA strands. mRNA serves as an intermediate between DNA and protein. Other, noncoding, RNAs have direct cellular functions. These include ribosomal RNA (rRNA), transfer RNA (tRNA), and small nuclear RNA (snRNA), along with a large number of noncoding RNAs that serve a range of regulatory activities. The spacing of genes on chromosomes is far from uniform: some chromosomal regions, and indeed whole chromosomes, are gene rich or gene poor. Intergenic regions alternately consist of unique sequences and long stretches of tandemly repeated sequences referred to as satellite DNA. This extragenic DNA likely performs structural and regulatory roles that remain the subject of active research.

Transcription occurs in regions of open chromatin, where a complex of proteins, called general transcription factors, binds DNA at a core promoter and recruits RNA polymerase (see 4–17). The construction of this protein complex at the transcription start site and the first steps of RNA synthesis are referred to as transcription initiation. For many genes, transcription initiation occurs at a so-called TATA box (a sequence rich in T and A nucleotides). In the elongation step, RNA polymerase transcribes RNA across the length of a gene. The final step in the transcription of RNA is termination.

Posttranscriptional Regulation

The resulting primary RNA transcript is subjected to a posttranscriptional process called splicing, which removes internal sequences (introns) that will not be translated into protein in that particular cell 4–19. snRNAs play an important role in this splicing. Alternative splicing provides a mechanism by which a single gene can give rise to several mRNAs and proteins. This is because different exons (sequences of genes that are retained in mRNAs) can be used to construct the mature mRNA depending on the cell type and stage of development involved. In fact, some genes are heavily spliced, with numerous mRNAs formed through alternative splicing. One of the most extraordinary examples is neurexin: three neurexin genes are thought to give rise to hundreds of neurexin proteins through alternative splicing, which help control the formation of specific contacts in the developing nervous system. Thus, the true molecular diversity of mammalian cells is significantly underrepresented by the approximately 20,000 genes contained in the genome.

Subsequently the mature mRNA is exported from the nucleus into the cytoplasm. All mRNA molecules contain UTRs at their 5´ and 3´ ends that affect mRNA stability and translatability. For example, a poly(A) tail is coupled to mRNA by means of an endonucleolytic cleavage step that produces the 3´ end of a mature mRNA molecule. While the stability and translatability of an mRNA molecule are regulated by its polyadenylation, the underlying mechanisms remain incompletely understood. Eukaryotic mRNAs are also modified at their 5´ ends with the addition of a modified guanosine cap. This cap increases the stability of the mRNA and promotes efficient protein synthesis by binding protein factors that are required to initiate the assembly of a translation-competent ribosome.

The translation of mRNA into protein is governed by the genetic code. The sequence of nucleotides in a molecule of mRNA is read on ribosomes in serial order and in groups of three. Each triplet of nucleotides, called a codon, specifies a single amino acid. The translation of mRNA into protein is mediated by tRNAs, which recognize the three bases in a codon and carry the corresponding amino acid. Translation initiation always occurs at the codon AUG (specifying methionine), and translation is terminated when the ribosome reaches a stop codon in the mRNA.

Finally, the newly synthesized protein may undergo a variety of chemical modifications and cleavages. Such cleavages can yield multiple protein or polypeptide products from a single mRNA. One of the best characterized examples is the proopiomelanocortin gene, whose single mRNA gives rise to several hormones, including β-endorphin, α-melanocyte-stimulating hormone (α-MSH), and adrenocorticotropin (ACTH), among other biologically active products (Chapter 7). A protein must also be targeted to the appropriate cellular compartment, which is determined by the protein’s amino acid sequence, by its covalent modifications (eg, glycosylation, phosphorylation), and by the presence of other proteins in the cell that serve scaffolding or anchoring functions. Details regarding posttranslational alterations and the associated sorting of proteins to different cellular locations are addressed in subsequent chapters in this book.

The stability of a protein, like that of an mRNA molecule, is highly regulated. Some proteins, such as those in the Fos family of transcription factors (see below), are highly unstable because they contain specific amino acid sequences, termed degrons, that serve as substrates for proteolytic enzymes, or proteases. As well, the stability of certain proteins depends on their association with other proteins. Many cellular proteins undergo proteolysis by means of the ubiquitin–proteasome pathway, whereby specific enzymes add ubiquitin groups to the targeted protein. This addition tags the protein for proteolysis by proteasomes, which are large multiprotein complexes. Abnormal functioning of proteasomes is implicated in neurodegenerative disorders, such as Parkinson disease (Chapter 18). Specific inhibitors of proteasomes, such as the synthetic peptide MG132 and the nonpeptide natural product lactacystin, are used experimentally to study the role of proteasomes in cell regulation.

Noncoding RNAs

Recent research has uncovered several important roles for noncoding RNAs. Best studied are small RNAs, termed microRNAs (miRNAs), which are transported to the cytoplasm where they serve a regulatory function by binding to specific mRNAs and inhibiting their translation or reducing their stability. The molecular details by which this occurs are now well known 4–20. Many hundred miRNAs have been identified in mammalian cells to date, and, interestingly, many show dynamic regulation in the CNS in response to a host of stimuli. Accordingly, miRNAs represent a newly discovered mechanism of gene regulation, the role of which in neurologic and psychiatric disorders is now being explored. Suppression of an mRNA by miRNA is termed RNA interference (or RNAi), and such regulation is exploited experimentally, where miRNAs (or the related small interfering RNAs—siRNAs) are used to knock down a target mRNA of interest in cultured neurons or glia or even in the brain in vivo in order to understand the biologic activity of the encoded protein. There is interest in the possible utility of miRNAs, perhaps delivered via viral vectors, in the treatment of a wide range of human disorders. In addition to small noncoding RNAs, there is increasing recognition of the actions of large noncoding RNAs in regulating gene transcription and other cellular processes.

All of the steps of gene expression, from RNA transcription to protein translation to posttranslational modifications of proteins, offer critical control points for dynamic regulation. However, the best understood is regulation of gene transcription, which is the focus of the remainder of this chapter. The precise control of transcription by extracellular signals such as neurotransmitters, growth factors, and cytokines permits the regulation of processes such as cell proliferation and differentiation and assists cells in adapting to their environments. Each cell in an organism contains a complete copy of that organism’s genome. However, selective expression of this common genome is required for the formation of distinct cell types during development, including the formation of thousands of types of neurons in the brain. In the fully differentiated adult brain, a subset of an organism’s genes remains accessible for expression and the rate of their expression is under constant control in response to the external environment. Indeed, such regulation of gene transcription is now thought to be a key mechanism for neural and behavioral plasticity.

Transcription Factors: Key Regulators of Gene Expression

The rate of transcription of a given gene is determined first by the state of its nearby chromatin. Many genes are inaccessible for transcription in neurons due to their presence in silenced chromatin. Conversely, many neural genes are not expressed in nonneural cells due to the action of the transcriptional repressor, RE1-silencing transcription factor (REST; also called neuron-restrictive silencer factor [NRSF]). Other genes available for transcription exist in varying states of activation (see 4–18), some being actively transcribed, others not transcribed but in a permissive or poised state, that is, able to be turned on in response to the right stimulus. The activation of such permissive genes involves several steps that are only briefly summarized here. The key step is the binding of a class of protein, named transcription factors, to short DNA sequences, called response elements, present along the regulatory regions of genes, that is, promoters or enhancers. Transcription factor binding provides part of the thermodynamic energy to spread nucleosomes apart. The factors also recruit to the gene enzymes that modify histones, such as HATs, which further loosens the nucleosome complex. More than 100 proteins may bind to the vicinity of an activated gene. This basal transcription apparatus contains RNA polymerase (more specifically, RNA polymerase II for virtually all genes that encode mRNAs) and a host of proteins that move nucleosomes along a strand of DNA, allowing for its transcription into RNA.

Transcription factors typically contain physically distinct functional domains 4–21. These domains contribute to the categorization of transcription factors into various families. Many transcription factors are active only when they form dimers or higher-order complexes. Multimerization domains are diverse and include so-called leucine zippers (see 4–2). Whether transcription factor dimers are homodimers or heterodimers, both partners commonly contribute to the DNA-binding domain and to the activation domain.

Interestingly, dimerization sometimes can be a mechanism for the negative control of transcription, as illustrated by the cAMP response element binding protein (CREB) family of transcription factors. CREB normally binds to its cAMP response elements (CREs) as homodimers. However, another gene, cAMP response element modulator (CREM), encodes a truncated protein inducible cAMP repressor (ICER), which can dimerize with CREB. However, because ICER lacks an activation domain, the resulting CREB–ICER dimers cannot activate transcription. ICER is thus an endogenous inhibitor (or dominant negative antagonist) of CREB-mediated transcription.

Although transcription factors may directly contact several proteins in the basal transcription complex, they sometimes interact with this apparatus through the mediation of coactivator or adapter proteins (see 4–21). In either scenario, transcription factors that bind at a distance from the core promoter can interact with the basal transcription apparatus because the DNA forms loops that bring distant regulatory regions (eg, enhancers) in contact with each other.

Many activator proteins become involved in the assembly of the mature transcription apparatus only after modification, such as phosphorylation, has occurred in response to extracellular signals. This, too, is illustrated by CREB. CREB can activate transcription only when it is phosphorylated on a particular serine residue (ser133) because phosphorylation of this residue permits CREB to interact with an adapter protein known as CREB-binding protein (CBP), which in turn contacts and activates the basal transcription apparatus in part via its HAT activity. Interestingly, mutations in CBP cause Rubinstein–Taybi syndrome, an autosomal dominant disorder characterized by mental retardation, bone and palatal abnormalities, and cardiac dysfunction 4–9.

Transcription Factors: Targets of Signaling Pathways

Two major mechanisms of transcriptional regulation by extracellular signals are illustrated in 4–22. One of these mechanisms involves transcription factors that are present at significant levels under basal conditions and are rapidly stimulated by signaling cascades to activate or repress the transcription of responsive target genes. By means of the other major mechanism, transcription factors that are expressed at very low levels under basal conditions are induced by a physiologic signal that enables them to regulate the expression of a series of genes.

A critical step in the extracellular regulation of gene expression is the transduction of signals from the cell membrane to the nucleus, which can be accomplished by several mechanisms. Some transcription factors translocate to the nucleus in response to their activation. These include steroid hormone receptors, whose translocation is triggered by the binding of their ligand, and nuclear factor-kappaB (NF-κB), a transcription factor retained in the cytoplasm by a binding protein (IκB) that masks its nuclear localization signal. Signal-regulated phosphorylation of IκB by I kappa kinase (IκK) leads to the dissociation of NF-κB, which in turn is permitted to enter the nucleus to bind DNA; IκB subsequently undergoes proteolysis in the cytoplasm 4–23. Other transcription factors must be phosphorylated or dephosphorylated directly before they can bind to DNA; for example, the phosphorylation of STATs by the protein tyrosine kinase JAK in the cytoplasm permits STAT multimerization, which in turn permits nuclear translocation and the construction of an effective DNA-binding site in the multimer.

Some transcription factors are already bound to their cognate regulatory elements in the nucleus under basal conditions and are converted into transcriptional activators by phosphorylation. CREB, for example, is bound to CREs before cell stimulation 4–24. The critical nuclear translocation step for CREB involves the activation of protein kinases such as PKA, which, after entering the nucleus, phosphorylates CREB. Alternatively, CREB activation can involve the nuclear translocation of second messengers, such as Ca²⁺ bound to calmodulin. After entering the nucleus, these second messengers activate CaM-kinase IV that in turn phosphorylates CREB 4–24. As previously mentioned, phosphorylation converts CREB into a transcriptional activator by permitting it to recruit CBP into the transcription complex.

The remainder of this chapter focuses on several transcription factor families that have received a great deal of attention as mediators of neural and behavioral plasticity in the adult.

CREB Family of Transcription Factors

CREB regulates transcription by binding to CREs present in a subset of genes. As their name suggests, CREs enable cAMP to activate (and in some cases repress) genes and have more recently been shown to confer responsiveness of genes to Ca²⁺ and to the MAP-kinase pathway as well. CREs have been identified in many genes expressed in the nervous system, including those that encode neuropeptides, neurotransmitter synthetic enzymes, signaling proteins, and other transcription factors.

The consensus CRE sequence, TGACGTCA, illustrates the palindromic nature of many transcription factor binding sites: the DNA sequence of the site’s two complementary DNA strands, which run in opposite directions, are identical 4–25. Many regulatory elements are perfect or approximate palindromes because many transcription factors bind to DNA as dimers, whereby each member is responsible for recognizing half of the binding site. While the idealized CRE site is a palindrome, several (nonpalindromic) variations in the sequence exist in CREB-regulated genes.

Regulation of CREB by cAMP, Ca²⁺, and growth factors

cAMP, Ca²⁺, and growth factors activate CREB by causing its phosphorylation at ser133. cAMP activates PKA, whereas Ca²⁺ activates CaM-kinase IV, both of which phosphorylate ser133 4–24. CREB is also phosphorylated on ser133 by a growth factor–activated kinase, known as ribosomal S6 kinase (specifically RSK-90), which is phosphorylated and activated by MAP-kinases. The activation of CREB thereby illustrates a striking convergence of several signaling pathways on the phosphorylation of a single amino acid residue in the protein. As mentioned earlier, such phosphorylation of CREB activates it by enabling its interaction with CBP. CREB contains sites other than ser133 that can be phosphorylated, which may assist in fine-tuning the regulation of CREB-mediated transcription.

CREB mediation of neural plasticity

The activation of a single transcription factor by convergent signaling pathways is particularly important in the nervous system because it may represent a mechanism for long-term neural adaptations. A case in point is learning and memory, which depend on the temporally coordinated arrival of two different signals that subsequently must be integrated in target neurons and their circuits. CREB has been shown to be important for learning and memory in several species. Genetic knockout of CREB in Aplysia, Drosophila, and rodents impairs the formation of new memory, while overexpression of wild-type or constitutively active forms of CREB causes the opposite effects. Likewise, CREB is required for cellular models of learning and memory, such as long-term potentiation. The role of CREB in these processes is discussed in greater detail in Chapter 14.

CREB-like proteins

CREB, like many transcription factors, is a member of a large family of related proteins. Other members of this family may compensate for CREB when it is inactivated, and also provide independent forms of positive and negative gene regulation. Among the proteins that are closely related to CREB are activating transcription factors (ATFs) and CREMs, which are generated by distinct genes. Several alternative splice forms of CREB, ATFs, and CREMs have been identified. All of these proteins bind CREs as dimers, and many can heterodimerize with CREB itself. The similarities between ATF1 and CREB are especially striking because both are activated by cAMP and Ca²⁺ signaling pathways. Many ATF proteins and CREM isoforms also appear to activate transcription; however, some CREMs, such as ICER, act to repress it, as stated earlier.

The dimerization domain used by CREB proteins, and several other families of transcription factors, is called a leucine zipper, mentioned above. This domain is composed of an α helix in which every seventh residue is a leucine; the leucines line up along one face of the α helix two turns apart. The aligned leucines of the two dimerization partners interact hydrophobically to stabilize the dimer. The many proteins, in addition to CREB, that utilize these mechanisms of dimerization belong to a superfamily known as basic-leucine zipper (bZIP) proteins.

AP-1 Family of Transcription Factors

Activator protein-1 (AP-1) transcription factors are indispensable in the regulation of neural gene expression by extracellular signals. AP-1 proteins bind DNA, as heterodimers or homodimers, to the DNA sequence TGACTCA, which is called the AP-1 sequence. This consensus sequence is a heptamer that forms a palindrome flanking a central C or G base (see 4–25). Although it differs from the CRE sequence by only a single base, AP-1 sites strongly prefer AP-1 proteins as opposed to CREB, which requires an intact CGTCA sequence. As a result, this single base difference between CRE and AP-1 sites significantly influences which transcription factors, and hence which intracellular signaling pathways, can regulate a particular gene. Many genes expressed in the nervous system contain AP-1 sites in their regulatory regions. Among these are genes that encode neuropeptides, neurotransmitter receptors, neurotransmitter synthetic enzymes, and cytoskeletal proteins.

AP-1 transcription factors bind to DNA as dimers that comprise two families of proteins, Fos and Jun, both of which are bZIP proteins. The known members of the Fos family are c-Fos, Fos-related antigen-1 (FRA1), FRA2, and FosB and its alternative spliced variant ΔFosB. The known members of the Jun family are c-Jun, JunB, and JunD. Most AP-1 complexes are formed by one Fos family member and one Jun family member. Unlike Fos proteins, certain Jun proteins also can form homodimers that bind to AP-1 sites, albeit with lower affinity than Fos–Jun heterodimers. Transcriptional regulation is further complicated by the fact that some AP-1 proteins can heterodimerize, by means of the leucine zipper, with members of the CREB–ATF family. AP-1 proteins also can form higher-order complexes with apparently unrelated families of transcription factors. They can, for example, complex with and inhibit the transcriptional activity of steroid hormone receptors.

Among Fos and Jun proteins, only JunD is expressed constitutively at high levels in many cell types. The other AP-1 proteins tend to be expressed at low or even undetectable levels under basal conditions, but with stimulation may be induced to high levels of expression. Thus, unlike regulation by constitutively expressed transcription factors such as CREB, regulation by most Fos/Jun heterodimers requires new transcription and translation of the transcription factors themselves. Most AP-1 proteins have short half-lives, which means that they exert highly transient effects of some stimulus on gene expression. An exception is ΔFosB (a truncated product of FosB) that, unlike all other Fos family proteins, is highly stable and thereby can mediate relatively persistent changes in gene expression 4–26. Accordingly, ΔFosB has been implicated in long-lasting neural and behavioral plasticity, in particular, drug addiction (Chapter 16).

Activation of cellular genes by AP-1

The genes that encode most AP-1 transcription factors are termed immediate early genes (IEGs). IEGs, a prototype of which is the c-Fos gene, are activated rapidly (within minutes) and transiently and do not require new protein synthesis. Late response genes, in contrast, are induced or repressed more slowly (over hours) and depend on the synthesis of new proteins. The term IEG initially applied to viral genes in eukaryotic cells that are activated immediately after infection by commandeering host cell transcription factors. Viral IEGs generally encode transcription factors that activate the late expression of viral genes.

The application of IEG terminology to nonviral genes has created some confusion. Many cellular genes are induced independently of protein synthesis but with a time course whose duration is between that of classic IEGs and late response genes. Moreover, many cellular genes that are regulated as IEGs encode proteins that are not transcription factors; for example, any gene induced by CREB could potentially show temporal features of induction of an IEG. Despite these complications, the concept of IEG-encoded transcription factors has assisted our understanding of gene regulation in the nervous system. In addition, several IEGs (in particular, c-Fos) have been used as cellular markers of neural activation because of their rapid induction from low basal levels in response to neuronal depolarization and various second messenger and growth factor pathways 4–27.

Activation by multiple signaling pathways

The best characterized cellular IEG is c-Fos. Because this gene contains binding sites for CREB 4–28, it is not surprising that it can be activated rapidly by neurotransmitters or drugs that stimulate the cAMP or Ca²⁺ pathways. The c-Fos gene also can be induced by the Ras/MEK/MAP-kinase pathway, discussed earlier in this chapter. This activation may occur in part via phosphorylation and activation of CREB by RSK. However, growth factor induction of c-Fos also occurs via a CREB-independent mechanism: subtypes of ERK translocate into the nucleus where they phosphorylate the transcription factor Elk-1 (also called the ternary complex factor [TCF]). Elk-1 subsequently complexes with the serum response factor (SRF) to bind to and activate the serum response element (SRE) in the c-Fos gene (see 4–28). SREs are also present in many other growth factor–inducible genes.

Still another mechanism of c-Fos induction involves cytokine-activated signaling pathways that trigger the activation of STATs. As previously mentioned, STATs are activated in response to their phosphorylation by JAKs, which in turn are activated by a variety of cytokines and other signals, as stated earlier. After STATs are activated, they form multimeric complexes, translocate to the nucleus, and bind to their specific DNA response elements, called STAT sites. The c-Fos gene (see 4–28) contains a STAT site (also called SIF-inducible element [SIE]), which mediates the induction of the gene by cytokines. STAT sites are found in many genes expressed in the nervous system.

Regulation by phosphorylation

Several AP-1 proteins are regulated at the posttranslational level by phosphorylation. The best studied example is c-Jun, which is phosphorylated in response to the activation of a MAP-kinase signaling pathway, called SAP-kinases or JNKs, by some form of cellular stress (see 4–15). JNKs phosphorylate c-Jun in its transcriptional activation domain, and thereby increase its ability to activate transcription. This process has been implicated in the modulation of synaptic transmission and may, under unique circumstances, trigger the activation of apoptosis (programmed cell death) pathways (Chapter 18).

Steroid Hormone Receptor Superfamily

Steroid hormones (eg, glucocorticoids, estrogen, testosterone, and mineralocorticoids) and related signals (retinoids, thyroid hormones, and vitamin D₃) are small, lipid-soluble ligands that diffuse readily across cell membranes. Unlike the receptors for neurotransmitters and peptide hormones, which are located in the cell membrane, the receptors for these ligands are localized in the cytoplasm. In response to ligand binding, steroid hormone receptors translocate to the nucleus, where they regulate the expression of certain genes by binding to specific hormone response elements (HREs) in their regulatory regions. Thus, these receptors are sometimes referred to as the nuclear receptor superfamily. Steroid, thyroid, and retinoid hormones and their nuclear receptors are critical for nervous system function, and are extremely important in CNS pharmacology (Chapters 10 and 15).

The mechanisms by which the glucocorticoid receptor (GR) operates are representative of those utilized by most receptors in this superfamily. Under basal conditions, the GR is retained in the cytoplasm by a large multiprotein complex of chaperone proteins, including the heat shock protein Hsp90 and the immunophilin Hsp56. After it is bound by glucocorticoid, the GR dissociates from its chaperones, translocates to the nucleus, and subsequently binds to glucocorticoid response elements (GREs). GREs typically are 15 bases in length and consist of 2 palindromic half-sites. GRs and related nuclear receptors have a modular structure similar to that of other transcription factors 4–29. The DNA-binding domain of the GR is characterized by multiple cysteines organized around a central zinc ion, an arrangement often referred to as a zinc finger. Many other transcription factors possess this DNA-binding domain.

GREs can confer either positive or negative regulation on the genes to which they are linked (see 4–29). A well-characterized negative GRE is located in the proopiomelanocortin gene mentioned above (Chapters 7 and 10); the GRE permits glucocorticoids to repress the gene, which encodes ACTH, and thus acts as an important feedback mechanism by inhibiting further glucocorticoid synthesis.

GRs are responsible for many important physiologic actions that do not appear to be mediated by DNA binding. Rather, they can interfere with transcriptional activity mediated by other transcription factors—particularly those mediated by AP-1 and NF-κB proteins (see 4–29). Glucocorticoids can also interfere with NF-κB activity by inducing expression of IκB, the protein that holds NF-κB in the cytoplasm.

Additional members of the nuclear hormone receptor superfamily have been identified in recent years. Some respond to cholesterol or lipids and are important in the regulation of intermediary metabolism 4–5. The endogenous ligands for many others are not yet known. Increasing evidence supports the involvement of many such nuclear receptors in neural functioning. Nerve growth factor inducible factor B (NGFI-B) and Nurr1, for example, are regulated in the brain by exposure to certain psychotropic drugs, including cocaine and antipsychotic agents.

Other Transcription Factors

The CREB, AP-1, NF-κB, STAT, and steroid hormone receptor families represent just a small portion of the transcription factors that are expressed in neurons and glia. However, many other transcription factors are crucial in neural signaling. CAATT enhancing binding protein (C/EBP) and its family members mediate some of the effects of the cAMP pathway on gene expression. Egr family members (eg, Egr-1/Zif268/NGFI-A) are zinc finger transcription factors that, like c-Fos, are induced rapidly and transiently in the brain by stimuli whose temporal features resemble those of IEGs. The induction of EGR family proteins has been correlated with induction of hippocampal long-term potentiation (Chapter 14); however, the target genes of these proteins remain poorly characterized. Circadian genes, the prototype of which is clock, are transcription factors that control diurnal variations in cell function through the regulation of gene expression (Chapter 13).

The functional mechanisms addressed in this chapter are far more intricate than their descriptions suggest. Regulatory regions of genes are usually far longer than the coding regions of genes, and regulatory information is contained not only in the 5´ promoter regions of genes but also in transcribed sequences and in 3´ untranscribed regions. Moreover, within the 5´ regions, this chapter focuses on a relatively small number of response elements. Most genes contain many regulatory sites, and these sites do not function in isolation but influence one another. Consequently, the temporal and spatial synthesis of multiple signaling pathways affects the expression of most genes. Unraveling this complexity is a daunting task, particularly in vivo, but is likely to yield important clues for understanding neural and behavioral plasticity. One goal of future research, outlined in 4–5, is to take advantage of the enormous complexity and specificity of these mechanisms to generate agents that act more rapidly and more effectively in the treatment of neuropsychiatric disorders.

The large majority of medications used to treat neurologic and psychiatric disorders target neurotransmitter receptors or proteins that influence levels of neurotransmitters, such as transporters and synthetic enzymes, and there has been a bias against targeting intracellular proteins based in part on the belief that the latter lack the specificity required for a safe and effective drug target. However, as is evident from this chapter, molecular biologic research has revealed a degree of heterogeneity in intracellular messenger pathways—from G proteins to transcription factors—not suspected by classic biochemical, pharmacologic, or physiologic studies. Moreover, individual subtypes of intracellular signaling proteins possess unique regulatory properties and exhibit varying levels of expression in different neuronal and glial cell types. Such heterogeneity has raised the specter of developing new medications that target intracellular messenger proteins. This point can be illustrated by consideration of RGS proteins. With more than 20 RGS subtypes known, many of which show highly restricted patterns of expression within the brain, it is conceivable that drugs aimed at antagonizing one particular subtype could exert a potent, clinically useful, and safe effect on the signaling of G protein–coupled receptors within brain regions of interest. Similar considerations operate for a host of brain proteins, even transcription factors 4–5, and this vastly increases the range of proteins that should be evaluated as future drug targets.