CHAPTER 8: Atypical Neurotransmitters

• The common designation of a group of neurotransmitters as purines is a misnomer; what are called purinergic signaling molecules are the nucleoside and nucleotide derivatives of purine and perhaps pyrimidine bases.

• The two principal purinergic signaling molecules are adenosine and adenosine triphosphate (ATP). ATP is stored in small synaptic vesicles and released in a Ca²⁺-dependent fashion, whereas adenosine is released from nonvesicular cytoplasmic stores, likely via bidirectional nucleoside transporters.

• Purine receptors form a relatively large and diverse group and have been categorized as P1 and P2 receptors.

• P1 receptors, also called adenosine receptors, bind adenosine and its analogs and are G protein–coupled. Stimulant drugs of the methylxanthine family, including caffeine, are antagonists of adenosine receptors. P2 receptors consist of both ligand-gated ion channels termed P2X receptors and G protein–coupled receptors termed P2Y receptors. P2X receptors play an important role in pain processing.

• Cannabinoids, the principal active ingredients of marijuana, primarily act in the brain on the CB₁ receptor, a G protein–coupled receptor found on presynaptic terminals in the central nervous system (CNS).

• Anandamide and 2-arachidonylglycerol are endogenous cannabinoids (endocannabinoids), which are released from postsynaptic cells and activate presynaptic CB₁ receptors.

• Nitric oxide (NO) is generated from arginine by NO synthase, which is stimulated by activation of postsynaptic NMDA receptors and increases in cellular Ca²⁺ levels. It diffuses out of cells and activates soluble guanylyl cyclase leading to the production of cGMP in adjacent cells and nerve terminals.

• Carbon monoxide (CO) is produced by the breakdown of heme by heme oxygenase-2 and also may function as an atypical, diffusible messenger. Hydrogen sulfide (H₂S) is another “gas transmitter”: it is generated from cyst(e)ine by the enzyme cystathionine β-synthase.

• Neurotrophic factors are polypeptides or small proteins that support the growth, differentiation, and survival of neurons. They produce their effects by activation of tyrosine kinases.

• The neurotrophins, which comprise nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophin-3 (NT-3) and neurotrophin-4 (NT-4), act by binding to a family of tropomyosin receptor kinase (Trk) receptors, TrkA, TrkB, and TrkC, with intrinsic tyrosine kinase activity.

• Numerous other growth factors, such as glial cell line–derived neurotrophic factor (GDNF), vascular endothelial growth factor (VEGF), and neuregulin, are important in regulating the nervous system.

• Several cytokine-like factors, including ciliary neurotrophic factor (CNTF) and interleukin-6 (IL-6), are characterized by binding to receptors that activate a family of protein tyrosine kinases called Janus kinases (JAKs), which in turn activate transcription factors called signal transducers and activators of transcription (STATs).

• Many additional cytokines, best understood for their role in the immune system and inflammatory responses, also are important in the regulation of CNS function. Prominent examples include interleukin-1 (IL-1), tumor necrosis factor-α (TNF-α), and transforming growth factor-β (TGF-β).

• Chemokines are small proteins involved in immune responses; in the brain, chemokines are expressed predominately by microglia.

Atypical neurotransmitters include a host of intercellular signaling molecules that have unusual properties or more recent dates of discovery compared with better known small-molecule and neuropeptide neurotransmitters that were reviewed in Chapters 5 to 7. The atypical neurotransmitters include the purinergic neurotransmitters adenosine and adenosine triphosphate (ATP), endogenous cannabinoids (endocannabinoids), the gases nitric oxide (NO) and carbon monoxide (CO), and families of neurotrophic factors and cytokines. Each of these neurotransmitters is found in specific subsets of neurons (and in some cases glia), has specific biochemic machinery for its synthesis, and exerts its biologic effects via activation of specific receptors or enzymes. Because of their discrete anatomic distributions and, in many cases, modulatory functions, these atypical neurotransmitters, their receptors, and the proteins involved in their production and degradation are potentially significant targets for drug development. Not discussed in this chapter are transient receptor potential (TRP) channels, which are covered in Chapters 2 and 11. Many of these channels respond to a host of endogenous molecules (eg, lipids, nucleotides, inorganic ions) or environmental signals (eg, temperature, mechanical stress) and thus function analogously to other signaling mechanisms presented in this chapter.

The purinergic neurotransmitters, adenosine and ATP, were long considered improbable mediators of neurotransmission because of their roles in intermediary metabolism and as building blocks of RNA and DNA. However, we now know that they are concentrated at certain synapses, they are released in response to synaptic stimulation, and they activate specific receptors to produce significant responses in target neurons. It is also established that purinergic receptors mediate the actions of several pharmacologic agents, most notably caffeine and related stimulants.

Biochemistry

The practice of referring to these signaling molecules as purines is to some degree a misnomer. Purines are ring-structured basic compounds that include adenine and guanine. These molecules do not function as neurotransmitters, nor do the pyrimidines, which include uracil, thymidine, and cytosine. So-called purinergic signaling molecules are in fact the nucleoside and nucleotide derivatives of purine bases 8–1A. The two principal neurotransmitters in this family are adenosine and ATP. Related to these are the adenine dinucleotides, which consist of two adenosine molecules covalently linked by a chain of two to six phosphate groups. Adenine dinucleotides are represented by the abbreviation ApnA, wherein n equals the number of phosphates between the two adenosines 8–1B. ApnA molecules are released by neurons and thus may be considered part of the purinergic signaling family. There is also some evidence that nucleoside and nucleotide derivatives of pyrimidine bases may serve as neurotransmitters, although this possibility remains poorly characterized.

Storage and Release

Despite their similarities, adenosine and ATP have distinct properties. ATP and ApnA are stored in small synaptic vesicles and are released in response to action potentials in a Ca²⁺-dependent process similar to the release process for classic neurotransmitters (Chapter 3). Indeed, ATP and classic neurotransmitters often can be detected in the same synapses and even in the same synaptic vesicles indicating that they can be coreleased. Release of ATP from sympathetic nerve terminals as well as vascular endothelial cells and tumor cells may play a particularly important role in various pain states. In contrast, adenosine is released from nonvesicular cytoplasmic stores and most likely reaches the extracellular space by one of two means. First, any of several bidirectional nucleoside transporters can secrete adenosine into the extracellular space, including the synapse. Second, ATP is very rapidly metabolized into adenosine once it is released from a cell. A membrane-bound ectodiphosphohydrolase converts ATP into ADP and AMP, and subsequently a membrane-associated or soluble ecto-5´-nucleotidase converts AMP into adenosine. Because the conversion of ATP to adenosine takes place in less than a second, the synaptic release of ATP should be regarded as an important source of extracellular adenosine.

The schematic representation of the purinergic synapse in 8–2 shows the release of ATP leading to the production of other purinergic compounds and to the subsequent activation of several types of purinergic receptors. It should be noted that ApnA is hydrolyzed much more slowly than ATP and can reside in the synaptic cleft for longer periods of time.

Transporters

Nucleoside transporters are membrane-bound proteins that shuttle purine and pyrimidine nucleosides into and out of many cells, including neurons. The transporters are purine-selective or pyrimidine-selective. Some act to concentrate the nucleosides in a cell in a Na⁺-dependent fashion. Others transport nucleosides down their concentration gradients. Pharmacologic and genetic studies indicate that there are at least seven nucleoside transporters, which belong to two gene families. The so-called equilibrative nucleoside transporter gene family (Slc29A1–A4 or ENT1–4), which consists of four subtypes (Slc29A1 or ENT1–4), mediates both efflux and influx of nucleosides; the concentrative nucleoside transporter family, which has three members (Slc28A1–A3 or CNT1–3), mediates Na⁺-dependent influx of nucleosides. Although much remains to be learned about the structure and function of these proteins, they have been exploited in the development of powerful therapeutic agents. A number of cancer chemotherapeutic drugs, such as gemcitabine, and several potent antiviral compounds, such as zidovudine (AZT), which are used in the treatment of AIDS, are nucleoside analogs that enter target cells by means of nucleoside transporters. Given the powerful effects that adenosine exerts on brain function, it is possible that drugs that regulate nucleoside transporters might eventually prove useful in the treatment of neuropsychiatric disorders.

Receptors

Purine receptors form a relatively large and diverse group of proteins that comprise two main subgroups, referred to as P1 and P2 receptors. Characteristics of both of these receptor families are discussed here and are summarized in 8–1. P1 receptors, also known as adenosine receptors (A₁ and A₂), bind adenosine and its analogs and are coupled to G proteins. Four subtypes have been identified (A₁, A_2A, A_2B, A₃), and each displays the canonical seven-transmembrane domains found in other members of the G protein–coupled receptor superfamily. The A₁ receptor subtype is the most widely expressed in the brain and spinal cord and has the highest affinity for adenosine. Its activation has been implicated in the putative anxiolytic, anticonvulsant, analgesic, and sedative properties of adenosine. Conversely, antagonism of A₁ receptors by methylxanthines such as caffeine and similar drugs results in stimulatory effects, including increased alertness at low doses and anxiety and irritability at much higher doses (8–1; see also Chapter 13).

The two subtypes of the A₂ receptor, A_2A and A_2B, have a somewhat lower affinity for adenosine than do A₁ receptors. The A_2B receptor is ubiquitous in the human body yet is expressed only at very low levels in the brain and spinal cord. In contrast, the A_2A receptor is highly concentrated in the dorsal striatum, nucleus accumbens, and olfactory tubercle—three brain regions that receive rich dopaminergic innervation (Chapters 6, 16–18). Indeed, the actions of adenosine and dopamine are interrelated in these regions. In the striatum, A_2A receptor agonists inhibit dopamine D₂ receptor–mediated behaviors, and A_2A antagonists mimic D₂ receptor agonists. The latter action likely contributes to the locomotor-activating properties of methylxanthines. The inverse relationship between adenosine and dopamine has led to speculation that selective A_2A antagonists might be useful in the treatment of Parkinson disease, as discussed in the next section of this chapter.

The A₃ receptor is expressed at low levels in the brain, and its function is not yet known. It is distinguished from the other subtypes in that it has a much lower affinity for adenosine. Whereas A₁ and A₂ receptors bind adenosine with nanomolar affinity, micromolar concentrations are necessary to activate A₃ receptors.

The P2 receptor family is intriguing because it comprises both G protein–coupled receptors (the P2Y family) and ligand-gated ion channels (the P2X family). Eight human P2Y receptor subtypes have been cloned and characterized. These receptors display unique pharmacologic profiles, and bind purine and pyrimidine nucleotide diphosphates and triphosphates, as well as ApnA molecules, with varying affinities 8–1. The P2Y₁ receptor, for example, binds ATP and ADP but not UTP or UDP. In contrast, the P2Y₂ receptor is activated by ATP and UTP with equal affinity. The regional distribution and functional significance of P2Y receptors in the central nervous system (CNS) are still being characterized.

Seven P2X receptor subtypes are known. Each is an ATP- or ApnA-gated cation channel composed of multiple subunits. Although the exact stoichiometry of P2X receptors is unknown, functional homomeric receptors have been constituted in vitro, and there is evidence for heteromeric P2X_2/3 receptors. Because these receptors have multiple subunits, each of the cloned proteins in 8–1 should be regarded as a subunit rather than as a complete receptor. P2X receptor activation leads to rapid Na⁺, K⁺, and Ca²⁺ flux and subsequent membrane depolarization. These fast-acting channels have been detected in the peripheral nervous system, neuromuscular junction, spinal cord, and many brain regions. When on postsynaptic cells, their activation leads to membrane depolarization. They are also found on many presynaptic nerve terminals throughout the CNS, and their activation routinely leads to enhancement of transmitter release. Via their actions in both the periphery and the spinal cord, they appear to play important roles in mediating a variety of different forms of pain and thus are important targets for the development of novel analgesic medications (Chapter 11).

Purine Functions

The cloning of the P1 and P2 receptor families opened up numerous avenues for investigating purine function in the nervous system. Receptor-selective agonists and antagonists are being developed, and transgenic animals that underexpress or overexpress individual receptor subtypes have been generated. Although many putative roles for nucleosides and nucleotides have been hypothesized, only a handful have been clearly established in the CNS.

Adenosine has both anxiolytic and hypnotic properties, and the administration of adenosine receptor antagonists confirms these observations. As described in 8–1, caffeine and other methylxanthine compounds, by blocking endogenous adenosine action, increase alertness and improve cognitive performance. The cognition-enhancing properties of these drugs are particularly interesting, partly because of the wide use of caffeine, and partly because adenosine receptor antagonists may prove useful as symptomatic treatments for the cognitive deficits associated with Alzheimer disease and other forms of dementia (Chapter 18).

Adenosine also is being investigated for its neuroprotective effects; it is hoped that these properties might be exploited in the treatment of stroke. Stroke is characterized, in part, by an ischemia-induced increase in glutamate levels followed by an overstimulation of glutamate receptors and massive influx of Ca²⁺ into neurons, which unleashes a cascade of cytotoxic events (Chapter 20). Thus, an ideal neuroprotective agent would inhibit glutamate release presynaptically and prevent postsynaptic membrane depolarization and subsequent Ca²⁺ influx. Adenosine can accomplish both feats. Adenosine activation of presynaptic A₁ receptors inhibits the release of glutamate and other neurotransmitters at many synapses throughout the brain. Activation of postsynaptic A₁ receptors opens K⁺ channels, and thereby hyperpolarizes neurons and counters the excitatory effects of glutamate. The net result is believed to be decreased Ca²⁺ influx and decreased neuronal death. Experiments in animal models support this hypothesis. The administration of selective A₁ agonists just before or during an ischemic event in animals significantly reduces neuronal loss. Conversely, the administration of A₁ antagonists augments ischemic brain damage. It is important to note that ischemia dramatically increases adenosine levels in the brain that will likely influence the net effects of drugs that target adenosine receptors.

Although activation of A₁ receptors may provide some measure of neuroprotection, the actions of adenosine at other receptor subtypes may exert the opposite effect. Activation of A₂ receptors, for example, has had deleterious effects on animals in models of stroke. Conversely, A₂-selective antagonists can produce neuroprotective effects. Acute activation of A₃ receptors with selective agonists also profoundly increases brain damage and mortality in animal models. In contrast, chronic administration of these adenosine agonists improves survival. Adenosine’s actions on vascular tone and platelet function likely contribute to these various observations. The interplay that occurs among adenosine receptor subtypes during ischemia clearly requires further investigation.

Adenosine systems also may prove to be promising targets for drugs designed to treat Parkinson disease. As discussed in Chapter 18, this disease is caused by a loss of dopaminergic innervation to the striatum, which leads to rigidity, tremor, and slowness of movement. The striatum, as previously mentioned, expresses high levels of adenosine A_2A receptors. A_2A receptor agonists produce biochemical and behavioral effects (decreased motor activity) that mimic those associated with Parkinson disease because they decrease dopamine neurotransmission in the striatum. Accordingly, blocking A_2A receptors might have the opposite effect; that is, dopamine function in the striatum and motor activity might be expected to increase.

Several orally active compounds that selectively block A_2A receptors have been described. When used in animal models of Parkinson disease, these drugs significantly reverse motor deficits while inducing little, if any, dyskinesia, a side effect associated with L-dopa therapy (Chapter 18). Moreover, when coadministered with L-dopa, A_2A antagonists diminish L-dopa-induced dyskinesia. Currently, A_2A receptor antagonists are in development for the treatment of Parkinson disease.

Adenosine also has been reported to exhibit anticonvulsant, anxiolytic, and antidepressant activity, as well as both analgesic and pain-enhancing properties. Although these purported functions require further investigation, they point to the therapeutic potential of selective adenosine receptor agonists and antagonists.

Over the last decade, a role for ATP and P2X receptors in pain has become apparent. Application of ATP to human skin elicits acute pain. Several different subtypes of P2X receptors are found in sensory neurons in dorsal root ganglia and other sensory ganglia with P2X₃ and P2X_2/3 receptors generally having the highest levels of expression. These receptors are also found at nociceptive nerve terminals in the periphery where they can respond to the release of ATP from various sources including tumors and vascular endothelium 8–3.

Molecular and pharmacologic studies of P2X receptors, in particular P2X₃ and P2X_2/3 receptors, suggest that while they may not be critically involved in acute pain sensation, changes in their levels and functions importantly contribute to the development and maintenance of various forms of neuropathic and inflammatory pain. As these pain states are common, and among the most resistant to treatment, P2X receptors appear to be important targets for potential novel analgesic medications (Chapter 11).

In 1964, identification of the psychoactive ingredient in marijuana as Δ⁹-tetrahydrocannabinol (THC) led to the synthesis of related compounds that displayed saturable binding and pharmacology consistent with the existence of a so-called cannabinoid receptor. Some of these agents 8–4 were found to cause a GTP-dependent inhibition of adenylyl cyclase, which suggested that the receptor for cannabinoids is a member of the large superfamily of G protein–coupled receptors. These observations led to the eventual cloning of two cannabinoid receptors known as the CB₁ and CB₂ receptors. CB₁ receptors are found at very high concentrations throughout the brain including the cerebellum, hippocampus, basal ganglia, cortex, brainstem, thalamus, and hypothalamus. This broad anatomic distribution of the CB₁ receptor helps to explain the diverse effects elicited by marijuana 8–2. The CB₂ receptor has low overall homology with CB₁ and is found primarily in immune cells with some limited expression in the brain.

CB₁ and CB₂ receptors couple to effectors by means of G_i/o proteins. Like other receptors that couple with these G proteins, CB₁ receptors inhibit adenylyl cyclase and, depending on the cell type, inhibit voltage-gated Ca²⁺ channels or stimulate inwardly rectifying K⁺ channels. As they are primarily found on presynaptic terminals of both excitatory and inhibitory neurons, activation of CB₁ receptors inhibits transmitter release and thereby synaptic transmission in a wide range of brain regions.

The identification of cannabinoid receptors led to the speculation that an endogenous cannabinoid-like substance might exist; such speculation was partly based on the discovery of endogenous opioid peptides soon after the detection of opioid receptors (Chapter 7). Because cannabinoids are highly lipophilic, the search for an endogenous substance focused on membrane extracts from the brain. This led to the identification of two substances, anandamide and 2-arachidonylglycerol (2-AG), which function as endogenous cannabinoids or endocannabinoids: they are released in response to neural activity and activate CB₁ receptors. As shown in 8–5A, both anandamide and 2-AG are thought to be synthesized from membrane-derived lipid precursors. Abundant evidence suggests that their release from postsynaptic neurons is coupled to their synthesis. Specifically, in a variety of cell types, strong depolarization leads to Ca²⁺ influx via voltage-dependent Ca²⁺ channels and this in turn causes activation of the enzymes responsible for generating anandamide or 2-AG 8–6. In certain cell types, the generation of these endocannabinoids can also be stimulated by activation of postsynaptic G protein–coupled receptors, specifically G_q-linked metabotropic glutamate receptors or muscarinic receptors 8–6.

There is extensive evidence that anandamide and 2-AG can function as true retrograde messengers in that they can escape from postsynaptic cells and bind to adjacent presynaptic CB₁ receptors to inhibit transmitter release 8–6. Depending on the cell types being examined, this endocannabinoid-mediated depression of synaptic transmission can last for seconds, minutes, or even hours. This action of endocannabinoids has been shown to greatly influence neural circuit function. Endocannabinoids are not stored in vesicles like classic neurotransmitters, and it remains unclear whether they passively diffuse through the postsynaptic plasma membrane or are transported by a specific transporter protein. The physiologic actions of anandamide and 2-AG are terminated by their transport into cells followed by intracellular hydrolysis. There is strong evidence for a specific transporter process, likely involving facilitated diffusion, although the key transporter protein has not yet been identified. Nevertheless, selective inhibitors of this transport activity have been developed 8–5B. Two specific intracellular enzymes, fatty acid amide hydrolase (FAAH) and monoacylglycerol lipase (MGL), break down anandamide and 2-AG, respectively 8–5. Inhibitors of these enzymes increase levels of the endocannabinoids 8–5B, which likely accounts for the broad behavioral effects of this class of compounds.

There has been great interest in the possible therapeutic potential of drugs that target individual components of the endocannabinoid signaling pathways. Given the strong stimulatory effects on appetite of CB₁ agonists such as marijuana, it may not be surprising that one of the first clinically available cannabinoid drugs is the CB₁ inverse agonist rimonabant, which was made available in Europe for the treatment of obesity. It is thought that the key site of action of rimonabant that leads to weight loss is in areas of the hypothalamus that are involved in the control of food intake and energy expenditure (Chapter 10). Unfortunately, the clinical use of CB₁ inverse agonists or antagonists is limited by their negative side effects, in particular, the induction of depressive symptoms and suicidality, which led to their withdrawal for use in Europe and prevented their approval in the United States.

A CB₁ agonist, dronabinol, is currently available for the treatment of the nausea and vomiting associated with cancer chemotherapy and for stimulating appetite in people with advanced HIV/AIDS. Cannabinoids are also useful in treating certain forms of glaucoma. Because activation of CB₁ receptors has significant analgesic effects in models of neuropathic and inflammatory pain, likely because of actions on peripheral nerve endings, there are efforts to develop CB₁ agonists that do not cross the blood–brain barrier and therefore will not have CNS effects. The range of potentially useful effects of cannabinoids drives the medical marijuana movement. However, since most of marijuana’s effects in the CNS are mediated via a single receptor—CB₁—it is not possible to develop agonists selective for a given desired effect without causing its psychoactive effects and associated addiction liability (Chapter 16).

Compounds that inhibit the degradative enzymes for anandamide and 2-AG (FAAH and MGL, 8–5) also may have therapeutic potential. In animal models, these compounds affect behavior in a manner suggesting that they may have analgesic and anxiolytic as well as antidepressant activities. Furthermore, when combined with dopamine receptor agonists, they improve motor performance in Parkinson disease models. These drugs may exert effects different from CB1 receptor agonists because, unlike the agonists, which indiscriminately activate all CB1 receptors in the brain, inhibition of degradative enzymes should enhance only the specific actions of endogenously released cannabinoids.

Nitric Oxide

NO was first identified as a cellular messenger when it was established to be the much sought endothelial-derived relaxing factor that is released from vascular endothelial cells and causes vasodilation. Soon thereafter it was found that NO is released from neurons by stimulation of nitric oxide synthase (NOS) in response to NMDA receptor activation. Three different isoforms of NOS exist, each the product of a distinct gene. Neuronal NOS (nNOS) is expressed exclusively in neurons, endothelial NOS (eNOS) originally was identified in endothelial cells, and inducible NOS (iNOS) originally was identified in certain immune system cells. Some neurons express eNOS and iNOS in addition to nNOS. NO is synthesized from arginine by nNOS, which is activated on binding to Ca²⁺–calmodulin complexes. nNOS is ideally positioned to respond to the rise in intracellular Ca²⁺ elicited by activation of NMDA receptors since it directly binds to PSD-95, a key synaptic scaffold protein that also binds to NMDA receptors 8–7. Because NO is a soluble gas, it is not stored in vesicles but rather freely diffuses out of the neurons in which it is generated. Its major action is to stimulate cytosolic guanylyl cyclase, which catalyzes the formation of cGMP (Chapter 4) in neighboring cells. Thus, NO can function as a retrograde messenger to activate guanylyl cyclase in adjacent presynaptic terminals, an action that at some synapses can influence neurotransmitter release. Another important action of NO is selective and reversible S-nitrosylation of cysteine residues in a wide variety of proteins including intracellular signaling enzymes, pumps, G proteins, and ion channels. Like phosphorylation, S-nitrosylation is a posttranslational modification that can either increase or decrease the target protein’s activity. NO also can inhibit cellular enzymes containing iron–sulfur complexes, such as mitochondrial NADH–ubiquinone oxidoreductase and NADH–succinate oxidoreductase, which are crucial to metabolism. Finally, NO can react with oxygen and other free radicals to yield reactive NO species such as peroxynitrite. This is particularly important in mediating the neurotoxic effects of excessive NO production, which likely contributes to the neural damage caused by strokes (Chapter 20).

Carbon Monoxide

CO, best known as a highly toxic gas when breathed in large amounts, also may function as a neurotransmitter, although its roles in the brain are still being elucidated. It is generated in neurons by heme oxygenase (HO), an enzyme that degrades heme to form biliverdin, iron, and CO. There are three isoforms of HO (HO1–3) with HO2 being selectively enriched in neurons. Like NO, CO stimulates soluble guanylyl cyclase, a finding that is consistent with the observation that the expression of HO2 and guanylyl cyclase overlap in many brain regions. The regulation of HO2 activity in neurons is not as well understood as that of NOS. One potentially important mechanism for increasing HO2 activity involves activation of protein kinase C by G_q-linked metabotropic glutamate receptors or phorbol esters, which in turn leads to the direct phosphorylation of HO2 by casein kinase 2 (CK2) 8–8. Like NOS, there is also evidence that HO2 can be directly activated by Ca²⁺–calmodulin.

Evidence for a role for CO as a neurotransmitter is strongest for the enteric nervous system and olfactory receptor neurons. HO2 as well as nNOS are enriched in neurons of the myenteric plexus, and genetic deletion or pharmacologic inhibition of HO2 significantly reduces noradrenergic neurotransmission in the gut. Similarly, HO inhibition can reduce the generation of cGMP in olfactory neurons that normally generate sufficient CO to activate guanylyl cyclase. CO production also has been suggested to play a role in the regulation of circadian rhythms by influencing the DNA binding activity of key circadian transcription factors.

Hydrogen sulfide (H₂S) is yet another gas that has been implicated in serving intercellular signaling functions. It is synthesized in the brain from cyst(e)ine by the enzyme cystathionine β-synthase. H₂S is thought to act primarily by S-sulfhydrating cysteine residues in target proteins.

The belief that extracellular signals can promote the growth and differentiation of nerve cells is almost a century old, yet the molecular diversity of neurotrophic factors and of their intracellular signaling cascades did not become apparent until the past two decades. Knowledge of neurotrophic factor signaling has dramatically enhanced our understanding of the ways in which the nervous system evolves during development and adapts throughout the adult life of an organism. Such knowledge also has provided insight into mechanisms responsible for neuronal survival, whose failure may underlie neurodegenerative disorders such as Alzheimer disease, Parkinson disease, Huntington disease, and amyotropic lateral sclerosis (ALS) (Chapter 18). Moreover, neurotrophic factors and their signaling proteins represent a large number of potential targets for the pharmacotherapeutic treatment of these and other neuropsychiatric disorders.

A discussion of neurotrophic factors requires the definition of several terms. Neurotrophic factors themselves are peptide growth factors for nerve cells: they influence the cell cycle, growth, differentiation, and survival of neurons. Due to overlap among growth factors for neurons and glia, we also apply the term neurotrophic factor to any molecule that produces trophic effects on the nervous system by influencing glia. To add to the confusion, the term neurotrophin applies to only one family of neurotrophic factors. A broader term is cytokine, which is borrowed from the field of immunology and refers to molecules released by one cell to modulate the activity of other cells. The term cytokine therefore applies to all growth factors, including neurotrophic factors. The term neurotrophic factor is restricted to proteins and excludes the many nonpeptide molecules—for example, steroid hormones, retinoic acid, and many small-molecule neurotransmitters—that also can influence the growth and integrity of the nervous system.

Although neurotrophic factors originally were distinguished from neurotransmitters based on their role in nervous system development, as opposed to synaptic transmission in the adult, we now know that there is considerable overlap in the actions of these molecules. Like neurotransmitters, many neurotrophic factors are synthesized by neurons and alter the functioning of other neurons; under some circumstances, they may even be released as a result of neuronal activity. Moreover, some neurotrophic factors, such as brain-derived neurotrophic factor (BDNF), may produce rapid changes in target neurons that are indistinguishable from those elicited by conventional neurotransmitters. Likewise, many small-molecule neurotransmitters not only elicit the rapid changes associated with synaptic transmission but also affect the growth, differentiation, and survival of neurons during development and in adulthood.

Certain key features of neurotrophic factors distinguish them from peptide neurotransmitters. Notably, neurotrophic factors tend to be larger proteins, whereas peptide neurotransmitters typically are very small peptides (Chapter 7). In addition, most neurotrophic factors produce their biologic effects through the regulation of protein tyrosine kinases, whereas most peptide neurotransmitters signal through G protein–coupled receptors and classic second messenger cascades.

Functional Characteristics of Neurotrophic Factors

Neurotrophic factors are synthesized by transcription and translation in the cell bodies of particular neurons and glia. Some are stored in these cells, perhaps in large dense core vesicles (Chapter 3), and are transported either to nerve terminals or to dendrites. Many factors, such as interleukin-1 (IL-1), BDNF, and glial cell line–derived neurotrophic factor (GDNF), are encoded by immediate early genes (Chapter 4); consequently, activity-dependent transcription may be the major determinant of release of these factors. The release of many neurotrophic factors can be triggered by depolarization, and there is evidence that such activity-dependent release of a neurotrophic factor requires more extended periods of neuronal activation compared with the release of small-molecule transmitters. The major mechanism responsible for the termination of neurotrophic factor signals is proteolytic degradation. However, some factors, such as BDNF, are sequestered by a functionally inactive, truncated receptor that limits their diffusion and thereby perhaps the duration of their action.

According to the neurotrophic hypothesis, neurotrophic factors are indispensable in the establishment of suitable nerve–target connections during development. While the sites of action of many neurotrophic factors are unclear, the case of nerve growth factor (NGF) in the peripheral nervous system is well understood. The first growth factor to be identified, NGF was discovered in the 1950s when certain sympathetic and sensory neurons were shown to require this protein for their survival 8–9. NGF was subsequently demonstrated to be synthesized by the target organs of these nerve cells and required to maintain the sympathetic and sensory neurons that innervate the organs. Because the supply of NGF is limited, competition for the factor exists among growing nerve fibers. Growing neurons that do not receive the NGF signal from their target do not survive. Likewise, only neurons that successfully respond to NGF survive and make appropriate connections with their targets. These findings demonstrate that NGF is required for proper nerve–target connections during development 8–10.

Although this pattern of NGF synthesis and action predominates in the peripheral nervous system, different patterns emerge in the brain and spinal cord (see 8–10). In the CNS, a target neuron may supply neurotrophic factor for an innervating neuron, but may also synthesize many other neurotrophic factors and receptors. Thus, factors in the brain and spinal cord may serve additional autoregulatory, or autocrine, functions. Furthermore, certain neurotrophic factors can be transported in an anterograde fashion to axon terminals where, after release, they act on the cell bodies or nerve terminals of other nerve cells. As well, neurotrophic factor–receptor complexes, formed on the plasma membrane of nerve terminals, can be retrogradely transported back to cell bodies, where they exert some of their biologic effects.

Glia further complicate our understanding of the production of neurotrophic factors. Some factors are synthesized by both neurons and glia and act on receptors expressed by both cell types. Such patterns of synthesis and activity result in highly complex forms of intercellular communication among neurons and glia that investigators have yet to disentangle.

Families of Neurotrophic Factors

The classification of neurotrophic factors is complicated by the history of their discovery: the names of many factors were based on the actions with which they were originally associated. Interleukins were given their name because they were identified as proteins that mediate communication among white blood cells, even though they are produced by glia and perhaps certain neurons as well. GDNF was named for its original source, even though it is made by many types of neurons and other cells in the body. Fibroblast growth factor (FGF) was regarded as a growth factor for fibroblasts, even though it is also produced by glia. Ciliary neurotrophic factor (CNTF) was named for its contribution to the growth and maintenance of ciliary ganglion neurons in the eye, even though it is generated by glia and neurons and is important for the survival of many types of neurons, notably motor neurons.

Currently, neurotrophic factors can be categorized based on their homologies and on the shared signal transduction mechanisms through which they produce their biologic effects 8–2. This chapter focuses primarily on the neurotrophins, which are among the best characterized in the nervous system. The chapter also provides brief discussions of GDNF, CNTF, some neurotrophic factors better known for their role as immune-response cytokines or chemokines, as well as vascular endothelial growth factor (VEGF), neuregulin, and nonacronymic VGF (which is not an abbreviation).

Neurotrophins

The neurotrophin family comprises NGF and subsequently identified factors that employ similar signaling mechanisms, including BDNF, neurotrophin-3 (NT-3), and neurotrophin-4 (NT-4; also known as neurotrophin-4/5). All neurotrophins are small proteins. BDNF, for example, has a molecular mass of approximately 14 kDa. Another common feature of neurotrophins is that they produce their physiologic effects by means of the tropomyosin receptor kinase (Trk) receptor family (also known as the tyrosine receptor kinase family).

Neurotrophins produce effects on a wide range of neurons. As previously mentioned in connection with NGF, their role in ensuring the survival of neurons in the peripheral nervous system is well established. NGF, for example, is present in target fields of small sympathetic and sensory neurons that have nociceptive and temperature-sensing functions, and the NGF receptor is expressed in these neurons. BDNF is produced in skeletal muscle (innervated by motor neurons). BDNF, NT-4, and NT-3 each supports the survival of a subset of peripheral sensory neurons.

Neurotrophins similarly affect the survival and maintenance of some neurons in the CNS, although their precise roles are less clear. In different circumstances their primary function might be target-derived support of afferent neurons, the support of efferent neurons, or the generation and maintenance of differentiated neurons. For example, NGF promotes the survival of cholinergic neurons in the septal nuclei of the basal forebrain, and BDNF, NT-3, and NT-4 promote the survival of some cortical motor and hippocampal neurons, and of noradrenergic, dopaminergic, and serotoninergic neurons located in the brainstem. Because cholinergic neurons are affected early in Alzheimer disease (Chapter 18), NGF has been proposed as a potential treatment; the use of other neurotrophins has been proposed for the treatment of Parkinson disease. Unfortunately, clinical trials have not shown significant efficacy as shown later in this chapter. To complicate matters further, recent work has shown that precursors of mature neurotrophins (eg, pro-NGF or pro-BDNF) are themselves released from neurons and antagonize the actions of mature neurotrophins. These proneurotrophins may, therefore, also be important for regulating cell death during development and after injury.

We now know that neurotrophins continue to play a role in the adult CNS, where they have been shown to support the survival and plasticity of fully differentiated neurons and in some instances to contribute to synaptic transmission. BDNF, in particular, is important for some forms of long-term potentiation (LTP) as well as certain types of learning and memory. BDNF also has been implicated in regulation of adult hippocampal neurogenesis, epilepsies, and animal models of several psychiatric disorders, in particular, depression, as will be discussed in later chapters of this book. NGF, which acts by activating the TrkA receptor (next paragraph), has been implicated in neuropathic pain, as well (Chapter 11). This has prompted the evaluation of anti-NGF antibodies or small-molecule antagonists of TrkA as potential treatments of chronic pain syndromes.

Trk receptors

All neurotrophins bind to a class of highly homologous receptor tyrosine kinases known as Trk receptors, of which three types are known: TrkA, TrkB, and TrkC. These transmembrane receptors are glycoproteins whose molecular masses range from 140 to 145 kDa. Each type of Trk receptor binds specific neurotrophins: TrkA is the receptor for NGF, TrkB the receptor for BDNF and NT-4, and TrkC the receptor for NT-3. However, some overlap in the specificity of these receptors has been noted 8–11.

The characteristic structural domains of the Trk receptor are represented in 8–12. The binding of a neurotrophin to its Trk receptor causes activation of the receptor’s catalytic domain. Neurotrophins bind as dimers, and in turn cause the dimerization of Trk molecules, which results in the autophosphorylation of Trk on several key cytoplasmic tyrosine residues. Such autophosphorylation initiates intracellular signaling cascades that lead to many of the biologic effects of receptor activation (Chapter 4). Intracellular signaling occurs by way of phosphorylated tyrosine residues that form a recognition sequence for the SH2 domains that are present on several types of cellular proteins. SH2 domains on Shc and Grb2, for example, eventually link Trk receptors to the activation of the small G protein Ras, which in turn triggers the activation of the microtubule-associated protein (MAP)-kinase cascade. Genetic abnormalities in Ras-related proteins cause neurofibromatosis in humans, which involves excessive growth and occasionally cancerous degeneration of Schwann cells (Chapter 4).

Other biologic effects of Trk receptor activation result from the phosphorylation of different signaling proteins on tyrosine residues. The most important of these are phospholipase Cγ (PLCγ), which triggers the phosphatidylinositol cascade, and insulin receptor substrate (IRS), which leads to activation of the phosphatidylinositol-3-kinase cascade. Mutations in the ATM gene, which encodes one subtype of phosphatidylinositol-3-kinase, have been shown to cause ataxia telangiectasia, a disease characterized by progressive degeneration and atrophy of several brain regions, particularly that of the cerebellum. The various intracellular signaling cascades activated by the neurotrophins and their Trk receptors are discussed in greater detail in Chapter 4.

Trk receptors display multiple splice variants. The best understood are the TrkB and TrkC isoforms that contain normal extracellular ligand-binding domains but lack the catalytic tyrosine kinase domain. As previously discussed, truncated Trk receptors limit the scope and duration of neurotrophin activity, either by heterodimerizing with full-length Trk receptors and preventing their phosphorylation or by sequestering extracellular neurotrophins and preventing their activation of full-length Trk receptors.

p75 receptor

The first neurotrophin receptor to be cloned was not a Trk receptor, but p75, a 75-kDa protein that is currently described as the low-affinity neurotrophin receptor. The p75 receptor binds to all neurotrophins with roughly equal affinity. It also appears to modulate Trk signaling, perhaps by allowing Trk receptors to respond to lower concentrations of NGF and other neurotrophins—a highly desirable trait in a receptor that competes for limited quantities of growth factor. In addition, p75 may make TrkA and TrkB more selective for their respective primary ligands, NGF and BDNF. While the mechanism by which p75 exerts these effects is not well understood, it is hypothesized that p75, Trk, and a neurotrophin form a stable trimolecular complex. Importantly, however, Trk receptors are fully capable of mediating functional responses to neurotrophins in the absence of p75.

Despite evidence that p75 can enhance neurotrophin and Trk function and promote cell survival under some conditions, p75 has been implicated in cell death pathways, which are described in greater detail in Chapter 18. p75 has a so-called “death domain,” analogous to similar domains in the tumor necrosis factor-α (TNF-α) receptor (see below), through which it activates programmed cell death or apoptosis (Chapter 18). Recent discoveries suggest that the opposing actions of p75 may depend on the presence of the coreceptor sortilin, a member of the Vps10p-receptor family. The p75–sortilin complex has high affinity for the proneurotrophins (eg, pro-BDNF and pro-NGF), which have strong proapoptotic activity. Studies to determine the functional implications of proneurotrophin action as well as that of p75 and sortilin are ongoing. Interestingly, genetic variations in certain sortilin family genes have been implicated in risk for Alzheimer disease (Chapter 18).

Neurotrophins and synaptic plasticity

Studies of developing visual cortex illustrate that neurotrophins play a role in some forms of synaptic plasticity. As axons from the lateral geniculate nucleus (LGN) of the thalamus grow into the primary visual cortex and form synapses, they segregate into eye-specific, alternating patches known as ocular dominance columns. Monocular deprivation experiments have revealed that LGN neurons that receive inputs from the deprived eye exhibit weakened synaptic connections. NT-4 can rescue these neurons from such plastic changes. It is possible that LGN neurons might compete for TrkB ligands and that this competition might be crucial for the formation of ocular dominance columns. In support of this hypothesis, both infusion of excess BDNF or NT-4 and infusion of a neurotrophin antagonist into the visual cortex are sufficient to block formation of ocular dominance columns.

Evidence suggests that neurotrophins also are involved in regulating synaptic plasticity in the adult brain. Neuronal activity dramatically and rapidly regulates the expression of certain neurotrophins and their Trk receptors in adult neurons. The induction of BDNF and TrkB, for example, has been observed in neurons of the hippocampus in response to trains of synaptic stimuli (including those associated with the formation of LTP; Chapter 5). The rapidity of BDNF induction is consistent with that of other immediate early genes, such as c-Fos (Chapter 4), and induction of BDNF is mediated by the activation of preexisting transcription factors such as CREB.

Increasing evidence suggests that neurotrophins can modulate synaptic transmission and regulate the formation and strengthening of synapses. Such actions may be mediated by cross-talk between neurotrophin and neurotransmitter signaling pathways. For instance, NT-3, via its Trk receptor, rapidly enhances synaptic transmission at the neuromuscular junction by increasing the probability of acetylcholine release from the presynaptic neuron. BDNF and NT-3 are reported to increase the size of excitatory postsynaptic potentials at the Schaffer collateral–CA1 synapse in the hippocampus. Accordingly, BDNF knockout mice exhibit reduced basal synaptic transmission at this synapse, and hippocampal slices from these mice are reportedly deficient in LTP. These data clearly implicate the involvement of BDNF in hippocampal function in the adult, although further research is needed to better understand the underlying mechanisms involved.

GDNF Family

GDNF, a glycosylated protein of approximately 18 kDa, was first isolated from a glial cell line that supports the survival of dopaminergic neurons from the midbrain. Because Parkinson disease is caused by the degeneration of dopaminergic neurons in the midbrain (Chapter 18), GDNF has received considerable attention as a potential therapeutic agent for this disease, although obstacles remain as will be discussed toward the end of this chapter.

More recently, GDNF has been proven to support the survival of many other neuronal cells, including those of the myenteric plexus within the gut. However, the most profound abnormality observed in GDNF knockout mice is maldevelopment of the kidney, which causes death shortly after birth. This finding demonstrates that GDNF plays a critical role outside of the nervous system.

Like the neurotrophins, GDNF produces its biologic effects through the activation of a protein tyrosine kinase, but such activation is achieved indirectly through an intervening receptor protein 8–13. This receptor, termed GFRα1, binds to and activates Ret, a transmembrane protein tyrosine kinase of about 150 kDa, which then mediates GDNF action. Loss-of-function mutations in Ret are associated with Hirschsprung disease in humans, a disorder characterized by abnormal gut motility, consistent with Hirschsprung-like abnormalities in GDNF knockout mice. In contrast, gain-of-function mutations in Ret are associated with neural crest malignancies in humans, such as multiple endocrine neoplasias and medullary thyroid carcinoma.

Other members of the GDNF family of neurotrophic factors include neurturin, artemin, and persephin, which also signal by means of Ret. Like GDNF, each binds to a specific receptor-α (Rα) subunit, which subsequently converges on Ret. Also like GDNF, neurturin can support the survival of dopaminergic neurons in the midbrain, which raises its potential utility for Parkinson disease.

CNTF

As previously mentioned, CNTF initially was studied as a survival factor for chick ciliary ganglion neurons where it upregulates choline acetyltransferase (Chapter 6). More recently, CNTF, a protein of 24 kDa in size, has been proven to regulate the survival or differentiation of many other neuronal cells, including motor neurons, hippocampal neurons, and midbrain dopaminergic neurons. Its effects on motor neurons are perhaps the most dramatic: CNTF prevents their degeneration after axotomy and improves some motor defects in murine models of motor neuron disease. Consequently, CNTF has been tested as a therapeutic agent in the treatment of ALS, which is characterized by the degeneration of motor neurons. However, clinical trials were abandoned due to toxic side effects. Attempts to overcome these side effects have been unsuccessful (see below), but there is still interest in CNTF in the treatment of Huntington disease as well as in recovery from spinal cord injury.

CNTF, along with several cytokines (see next section), exerts its effects through a characteristic glycoprotein of 130-kDa (gp130) signaling pathway 8–14. Each ligand binds a unique Rα subunit, which is tethered to the plasma membrane by glycophosphatidylinositol (GPI). On ligand binding, the receptor complexes with gp130 and in some cases with other proteins. The formation of this receptor complex triggers activation of Janus kinase (JAK) and of related protein tyrosine kinases, such as Tyk, which subsequently mediate biologic effects such as activation of the signal transducer and activator of transcription (STAT) family of transcription factors (Chapter 4).

Surprisingly, CNTF knockout mice develop normally and exhibit only mild motor neuron defects during adulthood. Also surprising is a related finding in humans: approximately 2.5% of the Japanese population are homozygous for inactivating mutations of CNTF and thus are, in a sense, human CNTF “knockouts”; these individuals develop without obvious deficits. In contrast, mice lacking the CNTF receptor CNTFRα lose almost all of their motor neurons and die within 24 hours of birth. The striking discrepancies between CNTF and CNTFRα knockout mice suggest the existence of additional endogenous ligands for CNTFRα that have yet to be discovered.

CNTF and its receptor complex are expressed by neurons and glial cells in the CNS. However, much remains to be learned about the details of how CNTF contributes to cell–cell communication and its functional consequences in health and disease.

Immune-Response Cytokines and the CNS

As alluded to previously, the best studied effects of many cytokines are their actions on the immune system. Only recently has it emerged that many of these factors, particularly IL-1 and interleukin-6 (IL-6), TNF-α, and transforming growth factor-β (TGF-β), also mediate CNS responses to immunologic challenges. In addition, increasing evidence supports their involvement in the regulation of neural function and plasticity far more broadly.

Several cytokine signaling pathways are known. Many cytokines exert their actions at least partly via the activation of the JAK–STAT pathway. In some instances, such as for IL-6, leukemia inhibitory factor, and oncostatin M, such activation is achieved via the gp130 signaling pathway mentioned above for CNTF 8–14. However, many other cytokines (eg, several interferons) activate JAK–STAT via other receptor mechanisms. Interesting, several peripheral hormones—prolactin, growth hormone, leptin, and erythropoietin, among others—also signal through JAK–STAT (Chapter 10).

IL-1 activates Toll-like receptors, which act via nuclear factor κB (NF-κB) and MAP-kinase pathways, among others (Chapter 4). Toll-like receptors, numerous subtypes of which are known, are crucial for mammalian immunity and the ability to fight infections (Chapter 12). TGF-β activates the TGF-β receptor, which is a serine–threonine protein kinase that phosphorylates a family of transcription factors known as SMADs. Bone morphogenetic proteins (BMPs), classified as part of the TGF-β family, also signal through serine–threonine protein kinase receptors and SMADs. Finally, TNF-α signals through a complex receptor mechanism, involving the TNF-α receptor (which binds TNF-α) and its subsequent interaction with any of several effector proteins that determine the specific functional response elicited, which includes activation of NF-κB, MAP-kinases, and caspases (proteolytic enzymes involved in cell death; Chapter 18). One key adaptor is tumor necrosis factor receptor type 1–associated death domain protein (TRADD). Antibodies directed against several of these cytokines or their receptors are being used increasingly to treat a range of inflammatory and autoimmune diseases, including those that affect the nervous system (Chapter 12).

Cytokines implicated in immune function are critical for systemic homeostasis. Serious illness or stressors elicit a defense response that aids in the body’s recovery. This response involves processes that are mediated in part by the CNS, such as fever, reduced appetite, cardiovascular changes, sleep disturbances, and malaise. A more detailed discussion of these processes is provided in Chapters 10 and 11; the section that follows is devoted to the general schemes by which cytokines affect brain function, whether they are generated in the CNS or periphery.

The impact that immune-response cytokines can have on the brain is exemplified by the fever—which is mediated by the brain—that occurs in animals and humans in response to the peripheral injection of IL-1. Although the blood–brain barrier is believed to limit cytokine entry to most of the brain, some cytokines enter at sites where the blood–brain barrier is incomplete, such as certain periventricular areas. Cytokines also may act by inducing lipophilic signals, such as prostaglandins, within endothelial cells, which can diffuse into the brain parenchyma.

Interestingly, the brain itself can synthesize many of these immune-response cytokines. Such cytokines are believed to be synthesized primarily in glia—particularly in microglia, the CNS equivalent of macrophages—and in astrocytes. Expression by certain types of neurons now appears likely as well. Likewise, while receptors for these cytokines are expressed primarily in glia, increasingly such expression is being demonstrated in neurons. After almost any dramatic perturbation, such as brain infection or injury, activated microglia and astrocytes produce cytokines, including IL-1, IL-6, TNF-α, and TGF-β. These factors may further activate glial cells and in turn stimulate gliosis—the generation of new astrocytes and microglia. The activated glia presumably help the brain recover by restoring normal homeostasis.

However, excessive levels of these cytokines can contribute to neural injury. This phenomenon is demonstrated dramatically in mutant mice that overexpress specific factors such as IL-1 or IL-6; such mice exhibit significant neurodegeneration in the vicinities of cytokine overexpression. Moreover, elevated levels of immune-response cytokines have been associated with several neurodegenerative disorders, including Alzheimer disease. Elevated cytokine levels are also observed in individuals with multiple sclerosis, an autoimmune disease that results in the degeneration of myelin sheaths around axons. Treatment of multiple sclerosis now centers on the systemic delivery of antibodies that neutralize certain immune-response cytokines or of certain cytokines themselves that suppress immune responses (eg, interferon-β) (Chapter 12). In addition, there is increasing evidence for a role of immune-response cytokines in psychiatric disorders, in particular, depression (Chapter 15).

In some cases, immune-response cytokines produce effects similar to those produced by CNTF and related proteins. IL-1, like IL-6, enhances survival of several types of central neurons. TGF-β, like CNTF, is important for neural crest cell differentiation during development. On the other hand, TNF-α released from macrophages contributes to the waves of programmed cell death of motor neurons required for proper development. Like other neurotrophic factors, these cytokines may also directly affect neurons in the adult brain. Receptors for IL-1, TNF-α, and TGF-β are most densely concentrated in the hippocampus and hypothalamus, where some of their corresponding factors have been proven to influence synaptic plasticity. For example, IL-1 is reported to attenuate the generation of hippocampal LTP while TNF-α may directly influence synaptic properties in developing neural circuits. Immune-response cytokines also may influence the rate of neurogenesis and the survival of newly formed neurons in the dentate gyrus of the adult hippocampus. More research will be required to better characterize the direct effects of these cytokines on neurons, and to explore their possible role in the regulation of nervous system function under normal and pathologic conditions.

Chemokines are a family of small (8–10 kDa) proteins originally characterized for their involvement in immune responses. Recent studies have revealed that many chemokines and their receptors are expressed in the brain, predominantly by microglia but also in astrocytes and certain neurons. Chemokines are categorized based on the spacing between two required cysteine residues in their primary structure, with either contiguous (CC) residues, one intervening residue (CXC), or several intervening residues (CX₃C). Receptors for these chemokines—all of which belong to the G protein–coupled receptor family—are likewise named according to these categories (see 8–2).

Chemokines may mediate part of the brain’s inflammatory response, and they may be involved in several human disorders, including multiple sclerosis, brain tumors, and stroke. Chemokines also may play a role in regulating the migration of neurons during development and in attracting growing nerve fibers toward target neurons, for example, in axonal guidance. For instance, mice that lack the chemokine receptor CXC₄R exhibit abnormal migration of cerebellar granule cells during late stages of development. Investigators have speculated that chemokines may assist in regulating neuron–neuron interactions during adulthood. According to this hypothesis, chemokines may contribute to the formation or retraction of synapses in the context of long-term plasticity.

VEGF

This factor comprises several related proteins of ~20 kDa that have long been known for their effects in growth and permeability of blood vessels. Only recently has it been recognized for its neurotrophic properties. VEGF exerts its effects through three receptor tyrosine kinases, VEGFR1 (also known as Flt-1), VEGFR2 (or Flk-1), and VEGFR3 (or Flt-4). VEGF and its receptors are expressed in neurons and glia, and this expression is highly upregulated by hypoxia.

The neurotrophic properties of VEGF were first identified when mutations in the VEGF promoter of mice resulted in ALS-like symptoms. Subsequently, VEGF was found to rescue hypoxia-induced motor neuron death both in vivo and in vitro. Recently, a polymorphism in the VEGF promoter sequence was identified in a subset of ALS patients. It is thought that low VEGF levels may underlie motor neuron degeneration in at least one group of patients, but exploitation of these findings to treat ALS patients has proven difficult. VEGF may also be important for response to stroke and other forms of neural injury. Antibody or small-molecule inhibitors of VEGF are currently used in the treatment of macular degeneration and several cancers, where inhibition of vascular growth is therapeutic.

Neuregulin

This group of proteins exert potent trophic effects on the nervous system through activation of the Erb family of receptor tyrosine kinases (Chapter 4). Neuregulins and their receptors are considered part of the epidermal growth factor (EGF) signaling family. Certain forms of neuregulins are transmembrane proteins that are localized to the presynaptic nerve terminal plasma membrane, where they are cleaved into smaller fragments, which activate Erb receptors on the postsynaptic membrane to modulate glutamatergic signaling. Mutations in the genes that encode neuregulins and their receptors are implicated in schizophrenia (Chapter 17).

VGF

First identified as a factor that promotes the growth of cultured pheochromocytoma cells, the name VGF is nonacronymic, as noted above. The VGF gene encodes a larger precursor protein that is processed by proteolysis into a large number of biologically active peptides with diverse functional activities. VGF expression is highest in brain and certain endocrine tissues (ie, pituitary gland, adrenal gland, and pancreas), where its expression is induced by other growth factors such as NGF and BDNF. Clinical interest in VGF comes from a growing number of animal studies. As just two examples, VGF administration promotes the survival of pancreatic β-cells and ameliorates the symptoms of diabetes, and it promotes the generation of new neurons in the hippocampal dentate gyrus and exerts antidepressant-like effects, in rodent models. Much further work is needed to evaluate these and other potential therapeutic applications of VGF.

Although this chapter highlights only a few of the many known families of neurotrophic factors, it is clear that certain neurotrophic factors would be ideal candidates for use in the treatment of neurodegenerative disease, given their ability to promote neuronal survival. Yet clinical trials of such agents have been disappointing. Such setbacks can be attributed in part to inadequate animal models of many human neurodegenerative diseases. Moreover, because neurotrophic factors are relatively large proteins, their delivery to the CNS is particularly challenging. This is compounded by the fact that many neurotrophic factors have multiple roles in different parts of the brain and body. The delivery of quantities large enough to ensure bioavailability can therefore result in serious side effects. Neurotrophins have the additional complication of activating p75 in addition to Trk receptors, often with antagonistic effects. Consequently, small-molecule ligands of Trk or other receptors might be useful, but have been particularly challenging to generate. Other possible avenues for therapies—with their own challenges—involve direct delivery of neurotrophic factors, or vectors that express them, locally in the brain (see below).

Clinical experience with neurotrophic factors is reviewed only briefly here. As previously mentioned in this chapter, BDNF, CNTF, insulin-like growth factor-1 (IGF-1), and VEGF have been proven to support motor neuron survival in vitro and in vivo. In addition, treatment with BDNF, CNTF, or a combination of these factors has been shown to slow or arrest the progression of motor neuron degeneration in the wobbler mouse. VEGF has been found to have similar effects in the superoxide dismutase-1 (SOD1) knockout mouse, another ALS model (Chapter 18).

Encouraged by these results, investigators have used BDNF, CNTF, and IGF-1 in clinical trials involving patients with ALS. While BDNF has been well tolerated, it has produced only modest clinical effects, a result also obtained with IGF-1. Clinical trials involving CNTF have been even more problematic. Although treatment with CNTF slowed deterioration of muscle strength, it produced severe toxicity in most patients—especially in those receiving higher doses. Yet such high doses are 30 times lower than the effective dose (corrected for body weight) in wobbler mice. In addition, patients treated with this neurotrophic factor can develop neutralizing antibodies to CNTF that limit its bioavailability.

The therapeutic use of neurotrophic factors in the treatment of other diseases has been equally disappointing. Because NGF supports the survival of many sensory neurons, investigators have attempted to treat sensory neuropathy with this factor. However, as in studies of motor neuron therapies, humans have not been able to tolerate the highest doses of NGF, although such doses are 10,000 times lower than effective doses used in animal models.

Because BDNF, GDNF, and other factors have enhanced the survival of dopaminergic neurons in animal models, the use of neurotrophic factors in the treatment of Parkinson disease is a top research priority (Chapter 18). Although experiments with GDNF in a nonhuman primate model of Parkinson disease were promising, clinical trials in humans have demonstrated only modest benefits, and only when GDNF, or a viral vector expressing GDNF, is infused directly into the striatum. Trials with BDNF have been even less successful.

Likewise, because cholinergic afferent systems, whose function appears to be supported by neurotrophic factors such as NGF and BDNF, may be important for memory function, investigators are exploring the use of NGF and BDNF as treatments for Alzheimer disease (Chapter 18). Indeed, after treatment with NGF, aged rats improve in their ability to perform memory tasks. A recent small clinical trial in which fibroblasts expressing NGF were implanted into the brains of Alzheimer disease patients yielded moderate reductions in cognitive decline, but larger-scale trials are required.

A major challenge in all of these clinical trials is the difficulty in delivering neurotrophic factors directly to targeted neurons of the CNS. Intracranial injection of these factors in slow-release preparations and injection of cultured cells or viral vectors that overexpress these factors are currently under evaluation. These systems would have the obvious benefit of being able to target neurotrophic factors directly to the affected areas of the brain, thereby avoiding side effects caused by actions on other areas of the brain and periphery. Ideally, it would be useful to develop viral vectors that allow for modification in the quantity of neurotrophic factor produced, but this is not yet feasible.

The challenges associated with administering neurotrophic factors have inspired hopes of developing small molecules that might promote or antagonize neurotrophic factor function. Only recently have such reagents begun to be developed, and it remains to be seen whether they will have clinical utility. The large size of neurotrophic factors, and presumably of their binding sites on their receptors, by comparison to small-molecule neurotransmitters, makes this a challenging prospect in drug development. One recent success is the use of EGF receptor antagonists, such as gefitinib or erlotinib, for the treatment of several types of cancers, including glioblastomas (malignancies of astrocytes) (Chapter 1). More creative approaches to such research may be useful as well. For example, for BDNF, rather than development of a traditional TrkB agonist, small molecules that promote the dimerization and hence activation of TrkB might be developed. Such strategies are currently under way. Finally, as noted earlier, antibodies directed against a neurotrophic factor or its receptor are being used increasingly to treat a range of neurologic disorders, with interest in applying such approaches to psychiatric disorders as well (Chapter 15).