The Serotonin Signaling System: From Basic Understanding To Drug Development for Functional GI Disorders

Transcription

1 GASTROENTEROLOGY 2007;132: REVIEWS IN BASIC AND CLINICAL GASTROENTEROLOGY The Serotonin Signaling System: From Basic Understanding To Drug Development for Functional GI Disorders MICHAEL D. GERSHON* and JAN TACK *Department of Pathology & Cell Biology, Columbia University, College of Physicians and Surgeons, New York, New York; and Section of Gastroenterology, Catholic University of Leuven, Gastro-enterologie, O& N, Leuven, Belgium Serotonin is an important gastrointestinal signaling molecule. It is a paracrine messenger utilized by enterochromaffin (EC) cells, which function as sensory transducers. Serotonin activates intrinsic and extrinsic primary afferent neurons to, respectively, initiate peristaltic and secretory reflexes and to transmit information to the central nervous system. Serotonin is also a neurotransmitter utilized by a system of long descending myenteric interneurons. Serotonin is synthesized through the actions of 2 different tryptophan hydroxylases, TpH1 and TpH2, which are found, respectively, in EC cells and neurons. Serotonin is inactivated by the serotonin reuptake transporter (SERT)-mediated uptake into enterocytes or neurons. The presence of many serotonin receptor subtypes enables selective drugs to be designed to therapeutically modulate gastrointestinal motility, secretion, and sensation. Current examples include tegaserod, a 5-HT 4 partial agonist, which has been approved for treatment of irritable bowel syndrome (IBS) with constipation in women and for chronic constipation in men and women. The 5-HT 3 antagonists, granisetron and ondansetron, are useful in combating the nausea associated with cancer chemotherapy, and alosetron is employed in the treatment of IBS with diarrhea. Serotonergic signaling abnormalities have also been putatively implicated in the pathogenesis of functional bowel diseases. Other compounds, for which efficacy has not been rigorously established, but which may have value, include tricyclic antidepressants and serotonin selective reuptake inhibitors to combat IBS, and 5-HT 1 agonists, which enhance gastric accommodation, to treat functional dyspepsia. The initial success encountered with serotonergic agents holds promise for newer and more potent insights and therapies of brain-gut disorders. Digestive diseases are a prevalent and expensive problem. According to the National Digestive Diseases Information Clearinghouse, they yearly affect million people in the United States, cause 13% of all hospitalizations, provoke approximately 50 million physician office visits, and cost about $107 billion (direct indirect costs). Disorders that involve the enteric innervation affect motility, and while not often lethal, they cause considerable morbidity. Pathogenesis may be infectious, inflammatory, neurological or, as in the case of functional bowel diseases, unknown. These conditions, which include irritable bowel syndrome (IBS), are highly prevalent. In the United States, they affect million people, initiate million physician visits per year, and are involved in approximately 20% 40% of all visits to gastroenterologists. Because causative biochemical or anatomical abnormalities of the bowel have not been identified, functional bowel diseases are commonly assumed to be psychogenic; nevertheless, it is conceivable that abnormalities within the gut, in at least a subset of patients, also contribute to their causation. This possibility is supported by the complexity of the intrinsic innervation of the bowel, the enteric nervous system (ENS), and that of the interactions of central and peripheral mechanisms in the control of gastrointestinal motility. The ENS is a center of integrative neuronal activity that is able to regulate the behavior of the gut, even in the absence of input from the CNS, and is engaged in a two-way dialogue with the CNS. 1 The ENS and the CNS thus influence each other. The ENS, moreover, is extremely large, and with regard to its ultrastructural organization and neuronal diversity, more like brain than peripheral nerve. It follows that an adequate understanding of signaling within the bowel and from the bowel to the brain can contribute to the development of improved means of treating and, ultimately preventing functional bowel disease and other Abbreviations used in this paper: EC, enterochromaffin; IPANs, intrinsic primary afferent neurons; SSRIs, selective serotonin reuptake inhibitors; TpH, tryptophan hydroxylase by the AGA Institute /07/$32.00 doi: /j.gastro

2 398 GERSHON AND TACK GASTROENTEROLOGY Vol. 132, No. 1 disorders of gastrointestinal motility. Because serotonin plays critical roles in enteric neurotransmission, the initiation and propagation of intrinsic enteric reflexes, and in gut-to-brain signaling, 1 it is important to understand serotonin s contributions to normal and abnormal gastrointestinal function. The physiological role of serotonin has still not been completely elucidated, partly because of the presence of multiple serotonin receptor subtypes in the gut wall and partly because suitable ligands for in vivo studies are lacking. Still, some serotonin receptor agonists and antagonists are now in use for the treatment of functional gastrointestinal disorders and others are under investigation or in development. This review summarizes our current understanding of serotonergic signaling in the gastrointestinal tract and its implications for the pathogenesis and treatment of functional gastrointestinal disorders. Part 1. Basic Mechanisms and Translational Considerations Serotonin Is an Important Signaling Molecule Used by Multiple Neural Systems in Brain and Gut It is common in medical jargon to talk about the serotonin system, as if such a singular entity actually exists. Serotonin, however, is used for signaling in many diverse systems both in the brain and in the periphery. In the brain, in which serotonin is best known, most serotonergic neurons are located in the nuclei of the median raphe. 2 These nuclei, however, are functionally diverse; thus, the descending medullary neurons of the raphe obscurus, magnus, and pallidus are quite different from their ascending pontine and mesencephalic counterparts in the raphe pontis, centralis, and dorsalis. The descending projections of raphe nuclei influence spinal and brainstem mechanisms and are very much involved in the central modulation of pain and in pathologic pain syndromes, 3 often, unfortunately, from the bowel. By contrast, the ascending raphe projections are directed toward areas of higher integrative function and are implicated in the regulation of mood, sleep, sex, appetite, and just about everything that makes life worth living. 4,5 Despite the importance and diversity of central serotonergic mechanisms, the brain is not the ne plus ultra of serotonin. The brain actually contains very little serotonin in relative terms. 6 Most of the body s serotonin ( 95%) resides in the gut. Within the bowel, serotonin is synthesized by the enterochromaffin (EC) subtype of enteroendocrine cell 7 9 and by serotonergic neurons of the myenteric plexus EC cells contain one form of the rate-limiting enzyme in serotonin s biosynthesis, tryptophan hydroxylase-1 (TpH-1), 14,15 whereas enteric and central serotonergic neurons contain another, TpH-2, which is a different gene product. 15 EC cells produce and secrete far more serotonin than either central or peripheral serotonergic neurons, such that the serotonin secreted by EC cells overflows to reach the gastrointestinal (GI) lumen 16,17 and blood Overflowing serotonin from EC cells, which is taken up and concentrated in platelets, is virtually the sole source of blood serotonin; platelets lack TpH and thus cannot produce serotonin on their own. The primary targets of the serotonin that is secreted by EC cells are the mucosal projections of primary afferent neurons. These include extrinsic nerves, which transmit sensations of nausea and discomfort to the central nervous system (CNS), and the mucosal projections of intrinsic primary afferent neurons (IPANs); submucosal IPANs initiate peristaltic and secretory reflexes, while myenteric IPANs initiate giant migrating contractions. 33,34 Serotonin secreted by myenteric neurons mediates fast and slow excitatory neurotransmission and is involved in the regulation of GI motility. 13 Clearly, the variety in serotonin-secreting cells may not be kaleidoscopic, but it is impressive and counterproductive to discussions of a unitary serotonin system. Serotonergic cells include ascending and descending raphe neurons, enteric neurons, EC cells, and platelets (plus mast cells in rats and mice 6 ); moreover, the various serotoninsecreting cells have little or nothing to do with one another. They play many different roles in normal physiology and contribute to at least an equal number of pathophysiologies of disease states. Serotonin Is a Paracrine Messenger and a Neurotransmitter in the Gut The central serotonergic neurons provide inputs to a multiplicity of other neurons throughout much of the CNS and do not function as a unit. Central serotonergic neurons, moreover, are separated from peripheral serotonergic neurons, platelets, and EC cells by the blood-brain barrier, which is impermeable to serotonin. The projections of enteric serotonergic neurons are not thoroughly understood, but they are believed to be descending interneurons that innervate follower cells in both plexuses. 13,35 37 They do not, however, project to smooth muscle or other effectors. The serotonin secreted by EC cells acts locally in a paracrine fashion, 29,38,39 although the nerves it activates are mostly situated relatively far from them. 40 The EC cell-to-nerve junctions thus bear no resemblance to neuromuscular junctions or to any other synapses of the central or peripheral nervous systems (Figure 1). EC cells, like other cells of the GI epithelium, are in motion. They turn over and are replaced by stem cells in the necks of gastric glands and the crypts of Lieberkühn in the small and large intestines As a result of the movement and transience of EC cells, traditional synapses cannot be found on them (nerves are not good at innervating moving targets), 40 and EC cells do not focus the transmitter they release as precisely as do neurons. In fact, the hypersecretion of serotonin by EC cells may have evolved as a compensa-

3 January 2007 SEROTONERGIC SIGNALING AND GI DISORDERS 399 Figure 1. An electron micrograph of an EC cell from the mouse ileum. Note the clustering of dark serotonin-containing storage granules in the basolateral cytoplasm. Serotonin is released (5-HT, arrow) into the loose areolar connective tissue of the lamina propria where it can gain access to nerve fibers (nerve). After it has acted on 5-HT 3 (extrinsic nerves; myenteric IPANs) or 5-HT 1P (submucosal IPANs) receptors expressed by these nerves, serotonin is inactivated by SERT-mediated uptake (dotted arrow) into enterocytes where it is catabolized. The marker 500 nm. A Specific Transporter, Serotonin Reuptake Transporter, Inactivates Serotonin The continual secretion of serotonin in mammoth concentrations has necessitated the evolution of effective mechanisms for its inactivation. Receptors for which serotonin is a ligand, like all receptors, desensitize; therefore, serotonin cannot be allowed to linger on them. Responses of the bowel, moreover, while slow, are not so slow that they can depend on unaided diffusion to regulate the timing of their initiation. Enzyme-mediated catabolism of signaling molecules, such as that of acetylcholine (ACh) by acetylcholinesterase, is a well-known mechanism of transmitter inactivation. Serotonin is catabolized by monoamine oxidase 47,48 and, in the gut, by transferases and other enzymes ; however, because all these enzymes are intracellular molecules, they do not contribute to the termination of serotonin-mediated intercellular signaling events. There is no extracellular enzyme, analogous to acetylcholinesterase, that rapidly catabolizes serotonin. Because serotonin is a base, it is positively charged at a physiologic ph. As a result, serotonin penetrates poorly through the lipid bilayers of plasma membranes. Serotonin inactivation is therefore accomplished mainly by transporter-mediated uptake into either the cells that secrete it or into neighboring cells, which in the mucosa are enterocytes. 52,53 Knockout of SERT: Animal models of IBS? The serotonin reuptake transporter (SERT; 5-HTT) is the primary molecule responsible for inactivating serotonin in the CNS and gut, 21,59 61 but, when SERT is inhibited or knocked out, organic cation transporters and the dopamine transporter provide backups of lower affinity but high capacity. 21 Because SERT is the target of antidepressants and cocaine, the existence of the backup transporters is highly significant. These molecules compensate for the loss of SERT function; however, the degree of compensation is not sufficient to maintain normal function. Mice that lack SERT exhibit increased colonic motility and increased water in stools and display an alternating pattern of diarrhea and constipation. 21 Transcription of SERT, furthermore, is decreased as a consequence of experimental inflammation of the bowel 62 and is also decreased in patients with inflammatory bowel disease (IBD) or irritable bowel syndrome (IBS). 63 In those conditions, the behavior of the gut is not dissimilar to that tion for the lack of close connections at EC cell-toneuron junctions. The local concentration of serotonin in the mucosa is thus quite high, 39 which suggests that the functions of mucosal serotonin are unlikely to be limited to stimulating nerves. Serotonin, for example, has been postulated to affect crypt epithelial secretion and proliferation Figure 2. Illustration showing actions of serotonin in the bowel wall. EC cells in the mucosal epithelium secrete serotonin (5-HT) in response to pressure or other appropriate stimuli (arrows). Serotonin can stimulate extrinsic (not illustrated) or intrinsic primary afferent neurons (IPANs). IPANs are located in submucosal and myenteric plexuses. Submucosal IPANs are activated by 5-HT 1P receptors, while myenteric IPANs are activated by 5-HT 3 receptors. Both submucosal and myenteric IPANs are cholinergic, but submucosal IPANs also release CGRP. Release of ACh and CGRP from IPANs is amplified by 5-HT 4 receptors, which are presynaptic. Myenteric IPANs (Dogiel type II neurons) are thought to project multiples processes and terminate on numerous other cells. Following the action of serotonin, it is inactivated by SERT-mediated uptake into enterocytes. Submucosal IPANs have been linked to mucosally driven peristaltic and secretory reflexes, while myenteric IPANs have been associated with giant migrating contractions.

4 400 GERSHON AND TACK GASTROENTEROLOGY Vol. 132, No. 1 of the bowel of mice that lack SERT. For this reason, SERT knockout mice may be considered to be a possible animal model of IBS. Of course, it is not possible to interview mice to ascertain whether the cyclical alternations in their bowel habits are associated with pain and/or discomfort. Techniques that utilize contractions of the abdominal musculature as a surrogate marker for pain are useful in the analysis of visceral hypersensitivity, 64,65 which might be expected in mice that lack SERT, but not very good to evaluate a problem stemming from the knockout of a gene, and have not yet been applied to this model. The reduction in SERT transcription induced by inflammation has led to the suggestion that a common defect related to mucosal SERT expression accounts for the common symptoms seen in patients with IBD and IBS. The observations also support the idea that one or more defects in enteric serotonergic signaling contribute to the pathophysiology of IBS (see discussion in Part 2). A number of animal models have been developed, which attempt to duplicate conditions prevailing in IBS. None of these models, however, have been validated as precise mimics of the human condition. Most utilize mucosal inflammation, which can readily be induced and leads to a visceral hypersensitivity that remains present and detectable in the animals as a legacy long after the inflammation has passed. Inflammation of the colon of neonatal rats, for example, causes persistent visceral hypersensitivity that remains in adult animals. 66,67 A link of this syndrome to serotonergic signaling is tenuous at best, although the persistent visceral hypersensitivity of the adult bowel can be blunted by administration of a 5-HT 4 agonist (tegaserod). 68 Mucosal inflammation, moreover, produces long-lasting changes in the properties of intrinsic neurons of the ENS as well as visceral hypersensitivity Colitis induced by the administration of TNBS leads to dysmotility and enhanced excitability of IPANs (AH neurons), which display increased numbers of fast excitatory inputs. 69,70 The latter is a reflection of greatly enhanced synaptic facilitation, which occurs in the submucosal plexus following inflammation along with the appearance of serotonergic fast excitatory postsynaptic potentials on S neurons, which are mediated by 5-HT 3 receptors. 71 These events are not observed in tissue that has not been subjected to inflammation. Serotonin, furthermore, by stimulating G protein-coupled receptors, such as 5-HT 2, can potentiate nociceptive responses mediated by TRPV1 channels. 22 Sensitization of this kind may play a role in the development of visceral hypersensitivity, 74 although there is no good evidence in humans that a 5-HT 4 agonist abolishes visceral hypersensitivity. Normal GI Physiology Is Serotonin Dependent The hypothesis that serotonin participates in the pathophysiology of IBS requires that the roles played by serotonin in normal bowel physiology be identified. Identification of these roles, however, has been surprisingly elusive. It is not that serotonin lacks activity when applied to GI preparations in vitro or in vivo. To paraphrase H. L. Menken, no pharmacologist ever went broke throwing serotonin at the gut. The problem is that serotonin does so many things to the bowel that it is difficult to determine which of these actions are physiologically important. The multiplicity of responses to applied serotonin can be attributed to the expression within the gut wall of a wide variety of subtypes of 5-HT receptors. 13 Serotonin Initiates Peristaltic and Secretory Reflexes and Also Activates Extrinsic Sensory Nerves The first coherent idea about the role of serotonin in GI motility can be attributed to Edith Bülbring et al. Bülbring is responsible for the idea that EC cells of the mucosal epithelium secrete serotonin into the wall of the gut to stimulate intrinsic sensory (primary afferent) neurons (IPANs), which initiate the peristaltic reflex. 38,75 77 As noted above, Bülbring s ideas have stood up quite well despite the passage of time. It is now believed that pressure and/or other relevant stimuli, such as acid, lead EC cells to secrete serotonin (Figure 2). 38,39,78 85 This serotonin can activate intrinsic or extrinsic primary afferent nerves in the mucosa. 1 Extrinsic nerves are activated primarily by 5-HT 3 receptors and convey discomfort of one kind or another to the CNS ,86,87 Cancer chemotherapeutic agents also release serotonin from EC cells, which activates extrinsic primary afferent nerves to induce nausea. 88 Because these nerves are excited by 5-HT 3 receptors, 5-HT 3 antagonists, such as ondansetron (Zofran), granisetron (Kytril), or alosetron (Lotronex), alleviate the serotonin-evoked discomfort that extrinsic sensory nerves convey to the CNS. 1,88 These drugs are useful in cancer chemotherapy because they do not interfere with the serotonin s ability to activate submucosal IPANs. 1,13 Other subtypes of serotonin receptor, including 5-HT 1P and 5-HT 7, stimulate submucosal IPANs, which then initiate peristaltic 29,89,90 and secretory 30,31,91 93 reflexes. Submucosal IPANs are cholinergic neurons that appear to corelease calcitonin gene-related peptide. 29 ACh and calcitonin gene-related peptide are responsible, respectively, for fast and slow excitatory neurotransmission. 1 Although primary transmission in reflex initiation is cholinergic, 94 responses do not spread efficiently to distant regions when receptors to the cotransmitter, calcitonin gene-related peptide, are blocked. 29 In fact, amplification of the strength of synaptic signaling is important in this circuit and can be accomplished by stimulation of 5-HT 4 receptors, which are presynaptic and are located on cholinergic nerve terminals. 95 Stimulation of 5-HT 4 receptors enhances the release of ACh and thus has the effect of increasing the

5 January 2007 SEROTONERGIC SIGNALING AND GI DISORDERS 401 Figure 3. An electron micrograph showing a portion of the myenteric plexus (MyP), a nearby smooth muscle cell and an interstitial cell of Cajal (ICC). A cholinergic synapse is illustrated. Note the cluster of small (50 nm diameter) clear synaptic vesicles around a presynaptic density (active zone). The postsynaptic element is an extremely attenuated neurite. The myenteric plexus contains millions of synapses, most of which are on fine processes, such as the one illustrated here, suggesting that neuronal activity is capable of a degree of modulation that is difficult to study by current techniques, which record mainly activity in nerve cell bodies. Most synapses, like this one, are cholingergic. amplitude of fast excitatory postsynaptic currents and thus the strength of synaptic transmission HT 4 receptors are also located on the terminals of the final common motor nerves that utilize ACh to excite smooth muscle. 99,100 5-HT 4 agonists thus are prokinetic because they strengthen neurotransmission in responsible neural circuits. The ability of 5-HT 4 agonists to enhance propulsive movements of the bowel relies on natural stimuli to activate reflex activity. It is of interest that 5-HT 4 receptors have been postulated to be the receptors that EC cells utilize at the EC cell-to-ipan junction to initiate peristaltic reflexes This is largely because the application of 5-HT 4 agonists to the mucosa in vitro evokes peristaltic reflexes 101,102,104 and the desensitization of the mucosa to 5-HT 4 agonists blocks these reflexes. 103 What these observations demonstrate, however, is not that 5-HT 4 receptors themselves trigger peristaltic reflexes but that 5-HT 4 receptor stimulation is a critical component of them. The application of agonists to the mucosa of the gut does not assure that such compounds will not diffuse beyond the surface to which they are applied. The time course of pharmacologic studies of the bowel is such that any drug that is at all absorbed will reach the submucosal plexus (and in vitro the epithelial barrier is compromised, thereby assuring that drugs penetrate into the gut wall). Once there, 5-HT 4 agonists will amplify the strength of cholinergic neurotransmission, 95,97 which is what they do, and also enhance the release of calcitonin gene-related peptide to promote the spread of stimuli around and through the gut wall. 29 As a result, even subliminal stimuli, or perhaps none at all, will suffice to reach the threshold for the activation of reflexes; thus, the application of a 5-HT 4 agonist to the mucosal surface of the bowel can be so facilitatory in effect that it provokes the occurrence of reflexes without actually starting them. IPANs are probably constitutively active in intact preparations of bowel. 5-HT 4 receptors, moreover, are not located on nerves in the mucosa and thus are not present at the site at which serotonin that is released from EC cells meets the mucosal terminals of IPANs HT 4 Table 1. Mechanisms and Effects of 5-HT Receptor Ligands in Humans Class Compounds Effects Clinical applications (potential) 5-HT 3 receptor antagonists Ondansetron Slower small bowel transit Diarrhea-predominant IBS granisetron Decreased intestinal secretion alosetron Decreased colonic tone cilansetron Inhibition of colonic response to feeding Slower colonic transit 5-HT 3 receptor agonists MKC-733 Slower emptying of liquids Not established Faster small bowel transit Stimulation of interdigestive phase 3 (Constipation?) 5-HT 4 receptor agonists Tegaserod Faster gastric emptying Constipation-predominant IBS prucalopride Enhanced gastric accommodation renzapride a Faster small bowel and colonic transit Chronic constipation cisapride a Enhanced intestinal secretion (Functional dyspepsia?) (Decreased visceral sensitivity?) (Gastroparesis?) 5-HT 1 receptor agonists Sumatriptan b Enhanced gastric accommodation Not established buspirone c Slower gastric emptying R Stimulation of interdigestive phase 3 a Also 5-HT 3 antagonistic properties. b 5-HT 1p agonist, may be identical to 5-HT 7. c 5-HT 1A agonist.

6 402 GERSHON AND TACK GASTROENTEROLOGY Vol. 132, No. 1 agonists, furthermore, do not increase the activity of submucosal IPANs when applied to the mucosa 29 as they clearly would do if the 5-HT 4 receptor were to act at the EC cell-to-ipan junction. Finally, the activity of the 5-HT 4 receptor is presynaptic, and it does not directly stimulate enteric neurons. 95,97 To say that 5-HT 4 receptors are critical for the peristaltic reflex but not its initiators is not to denigrate the importance of these receptors in enteric physiology. It is common to focus on EC cells as the primary serotonergic entity in the gut because they are the largest source of serotonin; nevertheless, it is likely that serotonergic neurons are also physiologically significant, even though, like those of the brain, their store of serotonin pales by comparison with that of EC cells. Quantity, of course, is not the only measure of physiologic significance. Enteric serotonergic neurons represent only a small proportion of the total number of neurons in the enteric nervous system (ENS), 36,106 but they innervate many other neurons. 1 Most enteric neurons are cholinergic, 107 as are most of the classical synapses of the ENS (Figure 3). These are often found on processes of neurons a great distance from the cell bodies, which are utilized to obtain recordings in most studies of the effects of compounds on enteric neurotransmission. The abundance of cholinergic synapses within the ENS means that serotonergic modulation of ACh release (for example by stimulating 5-HT 4 receptors) is likely to have profound effects on the output of the ENS. Enteric serotonergic neurons have been linked both to fast excitatory neurotransmission, via 5-HT 3 receptors, and to slow excitatory neurotransmission, via 5-HT 1P1 and perhaps 5-HT 7 receptors, 111 and they appear to be important as interneurons in the mediation of a variety of reflex pathways. 1 In fact, depletion of serotonin or destruction of serotonergic neurons greatly impairs the motility of the bowel. 112 Determination of exactly which enteric reflex pathways utilize serotonergic neurons would be very helpful in intelligent drug design because it would enable motility to be subtly modulated with selective compounds that act on specific subtypes or even splice variants of receptors. Enteric motile and secretory behaviors are more likely to be beneficially affected by tweaking the system than by dramatically turning large numbers of neurons on or off simultaneously (Table 1). Future Directions Future research on the role(s) played by serotonin in gastrointestinal physiology is likely to go well beyond the known role serotonin plays in the initiation of peristaltic and secretory reflexes. Investigations will have to concentrate on the poorly characterized effects of serotonin on neuron-neuronal synapses (see previously) and the more subtle effects mediated by the serotonergic interneurons of the myenteric plexus. The Figure 4. Typical profiles of 5-hydroxytryptamine (5-HT) concentrations before and after a standardized meal in individual patients with constipation ( )- and diarrhea ( )-predominant IBS and healthy controls (Œ). From reference Atkinson et al. 125 actions of interneurons, serotonergic or otherwise, are not apparent in the typical pharmacological preparations of guinea pig ileum used to investigate drugs in vitro. Typically, these preparations monitor only the most gross of responses of final common myenteric motor neurons by recording the contractions of longitudinal muscle or, less commonly, circular muscle, to applications of drugs or electrical field stimulation. When motor neurons are activated directly by these stimuli, the sophisticated synaptic networks that lie upstream are short circuited and not monitored. The upstream synaptic networks, however, determine the type of motility the bowel actually exhibits in vivo, which is not to twitch or restrict itself to peristaltic reflexes. 113 What are needed are methods that will permit the activities of individual neurons to be related to the behaviors of the bowel that they control. These techniques will allow the functions of interesting interneurons to be explored, which, in turn will likely pay off handsomely, not only in enhanced comprehension of the functions of serotonin and the ENS, but also in therapeutic dividends. Other arenas in which serotonin is a player, but which have been much less well studied, are those of development and plasticity. Serotonin is a growth factor. It promotes enteric neuronal development by stimulating 5-HT 2B receptors 114 and enhances neuronal survival and plasticity by stimulating 5-HT 4 receptors (personal observation). Serotonin, which is present in the earliest neurons to develop in the ENS and is then closely followed during development by EC cells, is in a position to sculpt the ENS. The activity of early developing neurons and EC cells can affect the numbers and types of neurons that follow. As a result, early experience of the environment can alter, with unforeseen consequences, the nature of the ENS. At the other end of life, neurons are lost from the gut as part of the ageing process By affecting neuronal survival, serotonin can enhance the mainte-

7 January 2007 SEROTONERGIC SIGNALING AND GI DISORDERS 403 nance and resistance of the ENS to perturbations of life and age. These effects of serotonin and its multitude of receptors are significant for future drug development because they may provide a window of opportunity to intervene pharmacologically to ward off at least some of the ravages of time and experience. Part 2. Human Studies and Pharmacotherapy Presence of Serotonin, Serotonin Receptors, and the Serotonin Reuptake Transporter in the Human GI Tract The presence of serotonin, serotonin receptors and the serotonin reuptake transporter in the human gastrointestinal tract has not been studied in detail, but the available data are in line with the observations in animal studies (see Part 1). TpH-1 and serotonin are present in EC cells, and serotonin is present in enteric neurons in the human gastrointestinal tract Serotonin immunoreactivity can be found in neuronal cell bodies in both the myenteric and submucosal plexuses, of the small intestine and of the colon. No reports, however, are available on the expression of serotonin in the enteric nervous system of the human esophagus and the stomach. SERT is expressed at the basolateral surface of mucosal epithelial cells in the human rectum. 133 No data are available on its expression in deeper layers or in other parts of the human gastrointestinal tract. Very few studies have examined the expression of serotonin receptors in the human gastrointestinal tract. Radioautographic studies suggest that 5-HT 3 receptors exist in the human colon and rectum, where they are mainly expressed on myenteric neurons. 126,127 Transcripts encoding 5-HT 4 receptors have been detected in all segments of the human gastrointestinal tract. 128,129 Radioautographic studies have confirmed the expression of the 5-HT 4 receptors in the muscle layers and in the myenteric plexus of stomach, colon, and rectum in humans. 127,130,131 Transcripts encoding the 5-HT 7 receptor have also been detected in the human gastrointestinal tract. 132 Physiologic Role of Serotonin in the Human Bowel The role of serotonin in the control of GI function is incompletely understood. This is mainly due to the presence of many serotonin receptors in the GI tract and the lack of suitable and selective antagonists for use in humans. Studies of the effects of exogenous administration of serotonin have been hampered by their potential for causing harm, which can occur even when low doses of serotonin are given. Information on the physiologic role of serotonin in humans has thus mainly been obtained from studies of the release of endogenous serotonin, from inhibition of SERT, from the use of selective serotonin receptor antagonists, and from observations of enteric behavior under pathologic conditions. As was noted earlier, most of the serotonin in human peripheral blood originates from the GI tract. 134,135 Platelets do not produce serotonin but avidly take up serotonin from the circulation; therefore, the serotonin concentration in platelet-depleted plasma is thought to reflect serotonin release from the GI tract. Because platelets are fragile, however, and the concentration of serotonin in platelets far exceeds whatever might be free in plasma, there is always a danger that what is measured in plateletdepleted plasma reflects leakage from platelets during preparation. Acute administration of selective serotonin reuptake inhibitors (SSRIs) prolongs the availability of physiologically released serotonin, thereby enhancing effects of serotonin released from the GI tract but also from the CNS. 136 Tricyclic antidepressants inhibit SERT, but they also exert a variety of other pharmacologic actions, as, for example, inhibition of the norepinephrine and dopamine transporters. Although the effects of tricyclic antidepressants on GI tract motility and sensitivity have been studied intensively, the extent to which observed effects are actually attributable to SERT inhibition is unclear. The effects of SSRIs are almost entirely due to SERT inhibition, but these agents have not been studied that extensively. An acute decrease in serotonin synthesis can be achieved by acutely depleting L-tryptophan, the amino acid precursor of serotonin (see above). Between 4 and 7 hours after oral administration of an amino acid mixture devoid of L-tryptophan, a pronounced and selective lowering of serotonin biosynthesis is achieved, both centrally and peripherally. 137 The clinical picture of carcinoid syndrome, a disorder characterized by excessive release of serotonin from hepatic metastases of carcinoid tumors, indicates an overall stimulatory effect of serotonin on GI secretomotor function. Release of Serotonin Several studies have measured serotonin in deproteinated, platelet-depleted plasma as an estimate of serotonin release from the gastrointestinal tract Ingestion of a meal is followed by a rise in serotonin levels in platelet-depleted plasma which persists for several hours. In the individual subjects, the rise in serotonin is not gradual, but rather exhibits a pattern of repeated sharp rises and falls 141 (Figure 4). However, the relationship between serotonin release in the gastrointestinal tract and levels in platelet poor plasma is confounded by a number of variables. These include expression and activity of SERT, platelet fragility (see above), the reuptake of serotonin within the gut, and the transport of serotonin in the liver, lungs, and platelets themselves. Role in Perception No consistent effects of SSRIs on sensation throughout the gastrointestinal tract have been found. In

8 404 GERSHON AND TACK GASTROENTEROLOGY Vol. 132, No. 1 the esophagus, acute administration of the SSRI, citalopram, significantly lowered the perception of mechanical and chemical stimulation in healthy subjects with esophageal hypersensitivity. 142 This action of citalopram occurred without alteration of basal esophageal motility. This observation supports the idea that serotonin is involved in the modulation of esophageal sensitivity but does not identify whether a central or a peripheral mechanism of action is the cause. In contrast to their effects on the esophagus, acute or short-term administration of SSRIs does not appear to affect sensitivity to luminal distention in the proximal stomach, the antrum, the colon, or the rectum One study reported that acute L-tryptophan depletion was associated with increased sensitivity to rectal distention. 147 Studies with selective 5-HT 3 receptor antagonists failed to demonstrate a significant effect on sensitivity to gastric, colonic, or rectal distention The 5-HT 3 antagonist, ondansetron, did inhibit symptoms induced by duodenal lipids or acid, but the physiological implications of these duodenal intubation and perfusion studies are unclear. 154,155 One investigation on healthy subjects, evaluating a 5-HT 4 antagonist, failed to demonstrate an effect on colonic sensation. 156 Role in Motility and Secretion Control Administration of SSRIs seems to exert more consistent effects on GI motility than those on sensation described above. Short-term pretreatment with the SSRI paroxetine has been shown to enhance gastric accommodation to a meal in healthy volunteers, in the absence of any effect on fasting gastric compliance. 144 This observation suggests that serotonin release is involved in the control of the accommodation reflex. Paroxetine was not observed to affect gastric emptying. 157 In the small bowel, short-term pretreatment with paroxetine increased the occurrence of phase 3 of the migrating motor complex and shortened orocaecal transit time. 157,158 In a colonic barostat/manometry study, acute intravenous administration of citalopram was shown to increase colonic phasic contractility, frequency of occurrence of high-amplitude propagated contractions, and colonic compliance while at the same time, citalopram suppressed the colonic tonic response to a meal. 145 In addition to sensation and motility, evidence suggests that serotonin is also involved in the control of intestinal secretion and absorption. The 5-HT 3 receptor antagonist alosetron inhibits basal secretion in the healthy jejunum. 159 In addition, there is evidence that 5-HT 4 receptor agonists stimulate chloride and water secretion in the human jejunum. 160 Patients with carcinoid syndrome often suffer from diarrhea, which has both a secretory and a motor component. The secretory component of carcinoid diarrhea is attributable to excessive serotonergic stimulation of submucosal secretomotor neurons; the motor component includes faster small bowel and colon transit and an exaggerated tonic response of the colon to ingestion of a meal. 161 These observations suggest that serotonin is involved in the gastric accommodation reflex, in small bowel transit, and in the colonic response to feeding. 5-HT 4 receptor antagonism failed to alter esophageal contractility, gastric emptying, and small bowel or colonic transit. 156,162 5-HT 3 receptor antagonists do not significantly alter esophageal contractility, 163,164 rate of gastric emptying, 165,166 or gastric accommodation HT 3 antagonists, however, do inhibit the gastric component of phase 3 of the migrating motor complex, and they slow small intestinal transit. 168,169 The colonic tonic and phasic response to feeding is inhibited by the 5-HT 3 antagonist ondansetron. 169 The same pathway is activated by antral balloon distention or by intraduodenal lipid infusion, and this increase in colonic tone is also inhibited by the 5-HT 3 antagonist granisetron. 170,171 The 5-HT 3 antagonist, alosetron, decreased tone in the descending colon, and this was associated with increased volume but not pressure thresholds to colonic balloon distention. 151 Pathophysiologic Aspects of Serotonin in the Human GI Tract Altered GI serotonin content has been reported in a number of inflammatory conditions. In untreated celiac disease, increased numbers of duodenal EC cells and significantly higher postprandial plasma serotonin levels have been found. 172 A significant correlation was observed between peak postprandial serotonin levels and postprandial dyspepsia scores, suggestive of a role for serotonin in symptom pathogenesis. In ulcerative colitis, EC cell counts in rectal biopsy specimens are decreased, and the expression of mucosal serotonin, messenger RNA (mrna) encoding TpH-1 and SERT, as well as SERT immunoreactivity were all significantly reduced. 133 Increased serotonin immunoreactivity has been reported in the myenteric plexus of the ileum in patients with Crohn s disease. 123 Several studies have investigated abnormalities of serotonin signalling in IBS. Decreased mrna encoding mucosal serotonin, TpH-1, and SERT, as well as SERT immunoreactivity, have been reported in both diarrheapredominant and constipation-predominant IBS. 133 A recent abstract, however, reported a study, albeit with significant differences in methods, which failed to reproduce the findings of altered SERT expression in IBS. 173 Decreased postprandial serotonin plasma levels have been reported to occur in constipation-predominant IBS, 140,141 and increased plasma levels have been reported to occur in diarrhea-predominant IBS. 138,139,141 Increased numbers of rectal EC cells and an increase in the peak of postprandial serotonin release have been reported to occur in patients with postinfectious IBS. 140,174 Further work needs to be carried out on the role of serotonin in IBS, taking into account the limitations associated with

9 January 2007 SEROTONERGIC SIGNALING AND GI DISORDERS 405 the use of serotonin in platelet-poor plasma as a marker for serotonin release in the GI tract. Altered release of serotonin, if it occurs in the bowel, moreover, could either be a cause of altered motility or a consequence of it. Changes in the turnover or biosynthesis of serotonin because of alterations in TpH-1 activity may also modify levels of serotonin in platelet-poor plasma and make a contribution to IBS. In slow transit constipation, increased numbers of serotonin-containing myenteric neurons in the ascending colon have been reported. 124 Using a rabbit anti-idiotypic antibody, lowered serotonin receptor immunoreactivity was found in the muscularis mucosae and circular muscle layer of patients with slow transit constipation. 175 The same group found a positive correlation between serotonin receptor(s) immunoreactivity levels in the circular muscle and the transit time observed in patients with colonic inertia. 175 Functional studies in slow transit constipation confirmed that the colonic response to antral distention or duodenal lipid infusion is impaired and that 5-HT 3 -dependent and -independent components are dysfunctional. 176 Serotonin is a key player in the causation of chemotherapy-induced nausea and vomiting. Several cancer chemotherapy agents induce the release of serotonin from EC cells in the duodenum. Released serotonin induces nausea and vomiting through stimulation of 5-HT 3 receptors on vagal afferent nerves that convey information to the medullary vomiting system including the area postrema. 177 Serotonergic Modulation of GI Function: Therapeutic Implications As was noted above, ligands of serotonin receptors have been found to exert multiple effects on the human GI tract and the brain gut axis, and some can be employed as therpeutic agents. These ligands do so, in part, because serotonin plays a variety of roles in the control of GI function but also because a multiplicity of serotonin receptor subtypes are expressed within the ENS, on smooth muscle cells, enterocytes, extrinsic afferent nerve fibers, and the central nervous system. Targeting serotonin receptors to alter motility has a number of potential advantages. Although most enteric neurons are cholinergic, 107 directly targeting cholinergic neurotransmission is not very attractive as a therapeutic target. Cholinergic antagonists would virtually stop enteric neurotransmission and induce paralytic ileus at effective doses, while cholinergic agonists would induce massive spasm. 94 Anticholinergics are used therapeutically, but only at low and probably ineffective doses, because of their multiple side effects. Due to the activity of endogenous serotonin in GI function, SSRIs exert powerful effects on the gut. Ligands of serotonin receptors that exert at least potential effects on GI sensorimotor function in humans are those belonging to the 5-HT 1, 5HT 2, 5-HT 3, 5-HT 4, and 5-HT 7 subtypes of serotonin receptors SSRIs Although mechanistic studies provide a rationale for their evaluation, no data are available on therapeutic effects of SSRIs in noncardiac chest pain or in functional dyspepsia. Two small uncontrolled studies, and 1 study in which the comparator was psychotherapy of 6-weeks duration, suggest a therapeutic potential for SSRIs in the treatment of IBS Four small placebocontrolled studies have been conducted, of which 3 reported symptomatic benefit of the SSRI over placebo In 1 of these studies, the therapeutic effect was unrelated to effects on anxiety, depression, and colonic sensorimotor function, and, in contradistinction to its actions in healthy controls, the intravenous administration of citalopram failed to exert an acute effect on colonic sensorimotor function in IBS patients, at least as measured by use of a colonic barostat HT 3 Receptors 5-HT 3 receptor antagonists, such as ondansetron and granisetron, were originally introduced for the treatment of chemotherapy-induced nausea and vomiting. 5-HT 3 receptor antagonists, however, also slow small bowel transit, inhibit small bowel secretion, and decrease colonic compliance. 153,159,169 Because studies in animals have suggested that 5-HT 3 receptors are excitatory mediators in visceral sensory pathways (see above), 5-HT 3 antagonists might be expected to reduce the level of perception of visceral distention. 188 Similar to findings in rodent models, 5-HT 3 antagonists have not been found to exert consistent effects on sensitivity to balloon distention in humans Ondansetron has been reported to reduce nausea and sensitization to gastric distention during duodenal lipid infusion in healthy subjects, 154 and alosetron has been observed to reduce the colonic hypersensitivity that is induced by duodenal lipids in women with diarrhea-predominant IBS. 189 In addition, there is evidence that 5-HT 3 antagonists modulate visceral perception at the level of the CNS. In a placebo-controlled study, alosetron decreased activity and activation in response to rectal balloon distention in structures of the emotional motor system, and this was associated with a decrease in GI symptoms. 190,191 Three placebo-controlled studies with ondansetron have reported improvements in stool pattern and also in abdominal pain in patients with IBS The 5-HT 3 antagonist alosetron was first approved for the treatment of diarrhea-predominant IBS in women but was later withdrawn by the manufacturer because of severe constipation and ischemic colitis occurring in some patients using the drug. 200 That withdrawal, however, caused an uproar from disappointed users and led alosetron to be reintroduced in the United States under an

10 406 GERSHON AND TACK GASTROENTEROLOGY Vol. 132, No. 1 extremely cautious safety monitoring program and is now available but only when conditions set by that program are met. Other 5-HT 3 antagonists, in particular cilansetron, have been investigated or are also under development for diarrhea-predominant IBS. 201 In a placebo-controlled, dose-finding study in functional dyspepsia, alosetron showed therapeutic potential for relieving the symptoms of early satiety and postprandial fullness. 202 The effects of 5-HT 3 receptor activation have not been thoroughly studied in humans. In 1 study, a 5-HT 3 receptor agonist was shown to delay liquid emptying, to enhance the rate of small bowel transit, and to stimulate interdigestive antroduodenal phase 3 activity of migrating motor complexes. 203 No therapeutic applications for 5-HT 3 receptor agonists have been reported, and, in view of the role of the 5-HT 3 receptor in signalling nociceptive information, such as nausea from the bowel to the CNS, none are anticipated. 5-HT 4 Receptors 5-HT 4 receptor agonists stimulate motility and secretion through enhanced release of ACh from excitatory motor neurons and interneurons Although it has been suggested that serotonin released by mucosal stimulation initiates the peristaltic reflex in animals by activating 5-HT 4 receptors on IPANs, 208 these receptors are, as discussed earlier, presynaptic. The fact that 5-HT 4 agonists strengthen but do not directly activate reflexes probably enhances the safety of these compounds in therapy. They rely on natural stimuli to start reflexes and do not cause spasm or perpetual uncontrolled propulsion. Diarrhea can occur as a result of therapy, but it is manageable and readily overcome by discontinuing treatment or lowering the dose. In addition to their actions within the ENS, 5-HT 4 receptor agonists directly relax GI smooth muscle; however, it is unlikely that this effect contributes to the motility effects exerted by this receptor in vivo. 209 Several 5-HT 4 agonists have been developed for the treatment of hypomotility disorders of the human GI tract. Tegaserod is a partial agonist and prucalopride a full agonist at 5-HT 4 receptors. Both cisapride and renzapride are 5-HT 4 receptor agonists, but neither is selective. These drugs are also antagonists at 5-HT 3 receptors and may exert additional actions unrelated to 5-HT receptors (see above). 5-HT 4 receptor agonists have been reported to exert modest but inconsistent effects on esophageal motility and on gastroesophageal reflux in patients with gastroesophageal reflux disease (GERD) HT 4 agonists do not compare well to the gold standard of therapy of GERD with a proton pump inhibitor; a place for 5-HT 4 agonists in the management of straightforward GERD has therefore not been established. 5-HT 4 agonists stimulate the rate of gastric emptying Small scale studies suggest a therapeutic potential for 5-HT 4 agonists in gastroparesis, but no large scale controlled trials of these agents in gastroparesis have yet been published Impaired gastric accommodation has been proposed as an important pathophysiologic mechanism in functional dyspepsia. 222 The 5-HT 4 agonists cisapride and tegaserod were shown to enhance gastric accommodation to a meal, 222,225,226 suggesting a therapeutic potential in functional dyspepsia with impaired accommodation. In a phase II study, tegaserod tended to improve dyspeptic symptoms in female functional dyspepsia patients with normal gastric emptying, 227 which may, at least in part, be related to tegaserod s beneficial effect on gastric accommodation. Recently, a preliminary report of 2 phase III studies of tegaserod in the treatment of functional dyspepsia suggested efficacy in motility-like dyspepsia. 228 Whether 5-HT 4 receptor agonists can affect visceral hypersensitivity by acting at 5-HT 4 receptors is still controversial. The 5-HT 4 receptor agonists, cisapride and tegaserod, did not alter sensitivity either to gastric distention 223,224 or to colonic distention in healthy subjects. 229 Assessment of these classical end points of visceral mechanosensitivity testing would argue against a visceral analgesic effect of these 5-HT 4 agonists in humans. On the other hand, analgesic properties of drugs can also be assessed by studying the RIII reflex, a polysynaptic reflex that is elicited by electrical stimulation of a cutaneous sensory nerve and recorded from a flexor muscle on the ipsilateral limb. Graded visceral distension inhibits the RIII reflex response, and this is considered a marker for visceral pain. In healthy subjects, pretreatment with tegaserod prevented the inhibition of the RIII reflex evoked by slow rectal distension up to the pain threshold, suggestive of a visceral analgesic effect. 230 Finally, the results of a recent clinical study suggested an effect of tegaserod on esophageal pain thresholds in patients with functional heartburn. 231 In all mechanistic studies in humans, the lack of data with selective antagonists prevents conclusions on the involvement of 5-HT 4 receptors. Clearly, the ability of 5-HT 4 agonists to strengthen cholinergic transmission within the ENS could, by restoring abnormal bowel function toward normal, reduce the perception of discomfort from the gut, no matter whether or not the drugs directly affect visceral nociceptive neurons. In keeping with the idea that it strengthens colonic peristaltic activity, tegaserod has been shown to promote small intestinal transit time and to enhance proximal colonic emptying in patients with constipation-predominant IBS. 232 Tegaserod also stimulates intestinal secretion and promotes evacuation of jejunally perfused gas mixtures in healthy humans who tended to retain gas. 160,233 Several studies have established that treatment with tegaserod for up to 12 weeks is able to improve constipation and to provide relief of pain/discomfort and bloating in constipation-predominant IBS More recently, it has been demonstrated that the effects of