Review article: serotonin receptors and transporters — roles in normal and abnormal gastrointestinal motility


Dr M. D. Gershon, Department of Anatomy & Cell Biology, Columbia University, P&S, 630 West 168th Street, New York,
NY 10032, USA.


The gut is the only organ that can display reflexes and integrative neuronal activity even when isolated from the central nervous system. This activity can be triggered by luminal stimuli that are detected by nerves via epithelial intermediation. Epithelial enterochromaffin cells act as sensory transducers that activate the mucosal processes of both intrinsic and extrinsic primary afferent neurones through their release of 5-hydroxytryptamine (5-HT). Intrinsic primary afferent neurones are present in both the submucosal and myenteric plexuses. Peristaltic and secretory reflexes are initiated by submucosal intrinsic primary afferent neurones, which are stimulated by 5-HT acting at 5-HT1P receptors. 5-HT acting at 5-HT4 receptors enhances the release of transmitters from their terminals and from other terminals in prokinetic reflex pathways. Signalling to the central nervous system is predominantly 5-HT3 mediated, although serotonergic transmission within the enteric nervous system and the activation of myenteric intrinsic primary afferent neurones are also 5-HT3 mediated. The differential distribution of 5-HT receptor subtypes makes it possible to use 5-HT3 antagonists and 5-HT4 agonists to treat intestinal discomfort and motility. 5-HT3 antagonists alleviate the nausea and vomiting associated with cancer chemotherapy and the discomfort from the bowel in irritable bowel syndrome; however, because 5-HT-mediated fast neurotransmission within the enteric nervous system and the stimulation of mucosal processes of myenteric intrinsic primary afferent neurones are 5-HT3 mediated, 5-HT3 antagonists tend to be constipating and should be used only when pre-existing constipation is not a significant component of the problem to be treated. In contrast, 5-HT4 agonists, such as tegaserod, are safe and effective in the treatment of irritable bowel syndrome with constipation and chronic constipation. They do not stimulate nociceptive extrinsic nerves nor initiate peristaltic and secretory reflexes. Instead, they rely on natural stimuli to activate reflexes, which they strengthen by enhancing the release of transmitters in prokinetic pathways. Finally, when all the signalling by 5-HT is over, its action is terminated by uptake into enterocytes or neurones, which is mediated by the serotonin reuptake transporter. In inflammation, serotonergic signalling is specifically diminished in the mucosa. Transcripts encoding tryptophan hydroxylase-1 and serotonin reuptake transporter are both markedly decreased. Successive potentiation of 5-HT and/or desensitization of its receptor could account for the symptoms seen in diarrhoea-predominant and constipation-predominant irritable bowel syndrome, respectively. Symptoms associated with the down-regulation of the serotonin reuptake transporter in the human mucosa in irritable bowel syndrome are similar to the symptoms associated with the knockout of the serotonin reuptake transporter in mice. The observation that molecular defects occur in the human gut in irritable bowel syndrome strengthens the hand of those seeking to legitimize the disease. At least it is not ‘all in your head’. The bowel contributes.

The behaviour of the bowel can be controlled by the intrinsic innervation of the gut independently of input from the central nervous system

The surprising thing about the enteric nervous system (ENS) is that it surprises individuals, especially neuroscientists, who are accustomed to the central nervous system (CNS). It is almost always axiomatic amongst such individuals that the brain controls the behaviour of tissues and organs, either directly or in concert with the spinal cord. The bowel stands out as different. Alone amongst the organs of the body, the gut contains an intrinsic nervous system that can, when it is called upon to do so, control not only the behaviour of that organ, but that of the neighbouring pancreas as well.1, 2 Not that the discovery of the independence of the intrinsic innervation of the bowel is a recent occurrence. Over a century has passed since Bayliss and Starling described what they called the ‘law of the intestine’ and attributed it to the ‘local nervous mechanism’ of the gut.3–5 The ‘law of the intestine’, now called the peristaltic reflex, referred to the ability of the bowel to ‘sense’ an increase in pressure within its lumen, and to respond with a propulsive wave of oral contraction and anal relaxation. Because the gut continued to manifest this activity after they had cut off all of its extrinsic innervation, it was clear to Bayliss and Starling that the nerve cells and fibres required for the mediation of the ‘law of the intestine’ were all intrinsic elements of the bowel wall. This clarity became still more crystalline when Trendelenburg demonstrated, at the time of World War I, that the peristaltic reflex could even be evoked by increasing the intraluminal pressure in a loop of guinea pig small intestine isolated in vitro, a preparation from which the CNS and extrinsic sensory nerves could not be more absent.6 Later, in 1921, Langley paid tribute to the ENS by classifying it as a separate division of his newly defined autonomic nervous system.7 The other two divisions were classified on the basis of their connections to the CNS: thoracic and lumbar for the sympathetic, cranial and sacral for the parasympathetic. Langley thought that most of the neurones of the ENS were not directly innervated by the CNS and thus met neither the thoraco-lumbar nor cranio-sacral criteria; ergo, the ENS was yet another division. Because of its long recognized independence, it should, today, surprise no one that the ENS is unique, both in the complexity of its organization and in its resemblance to the brain.

The ens resembles the cns

The unique organizational features of the ENS are many. It resembles the CNS and differs from the remainder of the peripheral nervous system (PNS) in that internal collagen is lacking and the support of neurones is derived not from Schwann cells but from enteric glia.8–12 The ENS is also very large; there are at least as many neurones in the human small intestine (> 108) as there are in the spinal cord;8, 13 moreover, the phenotypic diversity of enteric neurones exceeds that of any other region of the PNS. In fact, all of the classes of neurotransmitter found in the CNS have also been found in the ENS.14, 15 Because of its resemblance to the brain, the ENS is often cited as a ‘simple nervous system’. Unfortunately, a simple nervous system is an oxymoron; the ENS is anything but simple. In recent years, however, the efforts of a small but productive cadre of neurogastroenterologists have begun to unravel the twisted web of microcircuitry upon which the independent mediation of the behaviour of this particular organ depends.

Neuronal monitoring of luminal conditions is a transepithelial phenomenon

To respond to changes in luminal pressure or, for that matter, to respond to changes in pH, the presence of nutrients or any other luminal stimuli, the ENS obviously requires a system of sensors to monitor conditions prevailing in the lumen of the bowel. Despite this necessity, no nerve fibres actually penetrate the intestinal epithelium or even the basement membrane of that epithelium. There are thus no intraluminal or intraepithelial nerve endings. Sensation is therefore a transepithelial phenomenon.1, 2 One mechanism, but probably not the only one, by which transepithelial sensation is accomplished in the bowel is to utilize the enteroendocrine cells as sensory transducers. The best characterized of these cells is the enterochromaffin cell16–18(Figure 1). These cells store serotonin (5-hydroxytryptamine, 5-HT) in prodigious quantities. Over 95% of the body's serotonin is located in the gut and over 90% of that store is present in the enterochromaffin cells that are scattered in the enteric epithelium from the stomach through the colon. The remainder is located in the ENS, where 5-HT serves as the neurotransmitter of a population of descending myenteric interneurones.19, 20 Enterochromaffin cells have a microvillous border that projects into the lumen of the bowel, and store their 5-HT in secretion granules at the base of the cells.21 Serotonin is secreted primarily into the lamina propria, where it gains access to nerve processes in the connective tissue.1, 2 The amount of 5-HT that enterochromaffin cells secrete constitutively is large, and that which they secrete in response to stimulation is even larger. High concentrations of 5-HT thus overflow into the portal circulation and intestinal lumen constantly, and even higher concentrations overflow postprandially.22–28 The large amounts of 5-HT secreted by enterochromaffin cells, which dwarf those secreted by neurones, are evidently necessitated by the fact that the enteric epithelium, including its enterochromaffin cells, turns over. The cells of the intestinal epithelium are thus in motion. New cells are generated in intestinal crypts and old cells slough into the lumen from the tips of villi (in the small intestine). Nerves thus cannot form morphologically recognizable junctions with enterochromaffin cells, but are separated from them by distances that are large and variable (Figures 2 and 3). Secretion thus makes up in mass of released transmitter what it lacks in subtlety of contact. An analogy with a garden hose spraying water at a flowerbed is more apt than one to a micropipette. The specificity of responses to a paracrine signal, like that provided in the mucosa by 5-HT, is thus dependent on the receptors reached by the transmitter. The ultimate efficacy of this kind of signalling system requires an efficient mechanism to inactivate and terminate the action of the transmitter.

Figure 1.

An electron micrograph of an enterochromaffin (EC) cell. Note that the cell has an abundance of electron-dense secretory granules (arrows), which contain 5-hydroxytryptamine (5-HT) and are located in the basolateral cytoplasm of the cell. These granules average 0.2–0.3 µm in diameter. The cell rests on the basement membrane (BM) that separates the mucosal epithelial layer from the underlying lamina propria. The apical boundary of this enterochromaffin cell is not visible.

Figure 2.

An electron micrograph showing the junction of the mucosal epithelium and lamina propria. The basal surface of an enterocyte is visible at the upper right. It is separated by the basal lamina from the lamina propria. A thick nerve bundle traverses the lamina propria (from upper left to lower right) under the epithelium. Note the abundance of nerve fibres within the bundle and its distance from the epithelium. A fibroblast and a plasma cell lie between the nerve and the epithelium.

Figure 3.

An electron micrograph of the mucosa and the epithelium–lamina propria junction. The basal lamina and associated reticular fibres, which comprise the basement membrane, are well shown and underlie the basal surfaces of enterocytes. A small nerve branch approaches the mucosal epithelium but does not give rise to a recognizable synapse. The large, dense, cored vesicle within a varicosity of one of the axons inside the nerve measures ∼ 150 nm and displays a morphology characteristic of peptide storage.

Enterocytes inactivate mucosally released 5-ht by serotonin reuptake transporter (sert)-mediated uptake

Responses to 5-HT are terminated by its reuptake.29, 30 Uptake into cells is required to inactivate 5-HT because, in contrast with acetylcholine, there are no extracellular enzymes that catabolize 5-HT; moreover, passive diffusion is too slow to prevent the accumulation of 5-HT in contact with its receptors. Accumulation excessively potentiates the action of 5-HT and, if not prevented (as will be discussed later in the context of the irritable bowel syndrome), causes 5-HT receptors to desensitize. In the nervous system, serotonergic neurones express a specific SERT in their plasma membranes that mediates 5-HT reuptake.31–33 There are, however, no serotonergic nerves in the enteric mucosa. For a long time, it was not clear how 5-HT could be inactivated in the mucosa of the bowel. The resolution of that mystery came with the discovery that mucosal epithelial cells in mice, rats, guinea pigs and humans express the very same molecule, SERT, which mediates 5-HT uptake into nerve fibres.34–37 These cells are well equipped to catabolize the 5-HT they take up by means of monoamine oxidase-catalysed oxidative deamination or by glucuronidation.38, 39

SERT is well known as the target of tricyclic antidepressants, serotonin selective reuptake inhibitors and cocaine.30 These compounds affect the gut;34, 35 however, the effects of SERT inhibition on 5-HT signalling in the bowel are mitigated by the expression in the bowel of backup transporters, which have a lower affinity for 5-HT than SERT, but a high capacity.36 These backup transporters include organic cation transporters, which parallel SERT in their distributions, and the dopamine transporter, which is present in the bowel in dopaminergic neurones.40 The 5-HT signalling component of mucosal sensory transduction therefore consists, in its essentials, of enterochromaffin cells, which secrete 5-HT as a first messenger, 5-HT-responsive sensory nerves in the lamina propria that confer specificity on the responses because of the 5-HT receptors they express, and enterocytes that express SERT and thus remove 5-HT in order to terminate its action.

Intrinsic primary afferent neurones of the submucosal and myenteric plexuses enable the gut to mediate reflex activity

In order for the gut to be able to respond to luminal stimuli and regulate its own behaviour in the absence of input from the CNS, it must contain intrinsic primary afferent neurones (IPANs) that do for the gut what dorsal root and cranial nerve ganglion cells do for the remainder of the body. Dorsal root and cranial nerve ganglion neurones carry afferent information to the CNS. IPANs are the analogous neurones that carry information to the interneurones and motor neurones of the ENS. IPANs have been found in both enteric plexuses.41–44 Submucosal IPANs are involved in the mediation of mucosally driven peristaltic and secretory reflexes.41, 45, 46 Myenteric IPANs are involved in other types of gastrointestinal reflexes.47, 48 IPANs in both plexuses are cholinergic neurones.41, 49 Those in the submucosal plexus co-release calcitonin gene-related peptide, together with acetylcholine.41, 50, 51 Acetylcholine is responsible for fast excitatory neurotransmission, and calcitonin gene-related peptide is a co-transmitter that mediates slow excitatory neurotransmission.41 The slow calcitonin gene-related peptide-mediated responses increase the irritability of the bowel and, when they are antagonized, the circumferential spread of excitation around the gut is inhibited. Many of the myenteric IPANs in the guinea pig bowel contain the calcium-binding protein, calbindin,52, 53 which has been used as an IPAN marker; however, calbindin may not be a general marker for myenteric IPANs because its presence appears to be species dependent.14 IPANs resemble dorsal root and cranial nerve sensory ganglion cells in their common expression of certain differentiation antigens;54 nevertheless, neither submucosal nor myenteric IPANs are identical to dorsal root and cranial nerve sensory ganglion neurones. In contrast with the latter cells, IPANs are innervated and are not just the first neurones stimulated in the afferent limb of enteric reflexes, but act as interneurones as well.55–58

5-Ht1p receptor activity initiates peristaltic and secretory reflexes

The receptor activity that activates the submucosal IPANs which, in turn, initiate peristaltic and secretory reflexes is called 5-HT1P.59–62 This receptor activity has not been cloned and has not been detected in the CNS, although it has also been reported in the skin and lymphoid organs.59 5-HT1P-mediated responses resist antagonism by conventional 5-HT antagonists and are unusual in their structure–activity relationship, in that agonists require an unsubstituted hydroxyl moiety at the 5 or 6 position of an indole ring. 5-HT1P agonists include hydroxylated indalpines and 5-HT1P antagonists include a dipeptide, N-acetyl-5-hydroxytryptophyl-5-hydroxytryptophan amide (5-HTP-DP), and renzapride. Renzapride, however, is not selective because it is also a 5-HT4 agonist and a 5-HT3 antagonist. Thus far, no drugs have been developed to target the 5-HT1P receptor in order to alter gastrointestinal motility. Peristaltic and secretory reflexes are both antagonized by 5-HTP-DP and by anti-idiotypic antibodies that recognize and block the 5-HT1P site.20 5-HTP-DP also prevents the activation of submucosal IPANs following mucosal stimulation.41 These observations are interesting in that they confirm the central role of 5-HT1P receptors in the stimulation of submucosal IPANs by enterochromaffin cell-released 5-HT; however, they also suggest that a 5-HT1P antagonist might induce paralytic ileus, which would be an unpromising response for a drug used to influence motility. Similarly, a 5-HT1P agonist might be expected to cause intractable diarrhoea by perpetually activating peristaltic reflexes. Indeed, 5-hydroxyindalpine was first tried as a 5-HT1P agonist because it was found to be the metabolite responsible for diarrhoea in human subjects treated with the antidepressant indalpine.59 Therefore, the enterochromaffin cell–submucosal IPAN junction may not be the best site to target the action of drugs that increase or decrease intestinal motility. Fortunately, the distal terminals of IPANs are also affected by 5-HT, and the receptors located there make much more attractive sites for prokinetic drug action.

5-Ht4 receptors are presynaptic, enhance transmitter release and thus strengthen neurotransmission in prokinetic pathways

5-HT has been shown to increase the release of both acetylcholine and calcitonin gene-related peptide from stimulated submucosal IPANs.41, 50, 51 This increase potentiates the effect of both of these transmitters and enhances neurotransmission. This enhancement increases the spread of excitation in the bowel following mucosal stimulation, and thus the tendency of the gut to manifest propulsive peristaltic and secretory reflexes. The response of the distal terminals of submucosal IPANs to 5-HT is mediated by 5-HT4 receptors. These receptors also promote the release of transmitters from myenteric neurones at intraganglionic synapses,63, 64 and the release of acetylcholine at the terminals of the final common motor neurones that activate smooth muscle.65 Again, the effectiveness of nerve stimulation or intestinal smooth muscle is amplified by 5-HT4 receptors. Electrophysiological recordings have documented the effects of 5-HT4 receptor stimulation.63, 64 The fast excitatory postsynaptic potentials (EPSPs) or, if patch-clamp recordings are obtained, fast excitatory postsynaptic currents (EPSCs) that are evoked by stimulating the fibre tract inputs to enteric neurones are increased markedly in amplitude by 5-HT and by 5-HT4 agonists, such as tegaserod, renzapride or cisapride.63, 64, 66 The enhanced EPSPs or EPSCs elicited by 5-HT or 5-HT4 agonists are specifically blocked by 5-HT4 antagonists, such as GR113808. The EPSPs or EPSCs are themselves cholinergic and mediated by nicotinic receptors. EPSPs or EPSCs are thus mimicked by nicotine and blocked by nicotinic antagonists, such as hexamethonium. 5-HT4 antagonists thus do not affect the EPSPs or EPSCs directly; they only block the amplification of the responses evoked by 5-HT4 receptor stimulation. No other effects are evoked by 5-HT4 agonists. They do not evoke postsynaptic responses and thus do not directly activate or inhibit enteric neurones. 5-HT evokes postsynaptic responses, but these are all mediated by other subtypes of 5-HT receptor. They include a fast depolarization or fast inward current that is 5-HT3 mediated, and a slow depolarization or slow inward current that is 5-HT1P mediated. Exposure of enteric neurones to 5-HT4 agonists has also been observed to elicit showers of EPSCs from presynaptic inputs (M. D. Gershon, personal observation, 2004).

The location of plasmalemmal 5-HT4 receptors in the bowel is restricted to the presynaptic sites at which their action has been demonstrated (internal pools of newly synthesized receptors in transport vesicles en route to the cell surface are not accessible to neurotransmitters or applied drugs).67 This location has recently been ascertained using in situ hybridization to identify the cells that express 5-HT4 receptors and selective antibodies to the receptor. The antibodies that have been useful include those that react with the 5-HT4a isoform and antibodies raised in chickens that react with all isoforms of the receptor (many isoforms are expressed in the bowel) and thus are pan-5-HT4 antibodies (M. D. Gershon, personal observation, 2004). These antibodies recognize isotopes in the ectodomain of the 5-HT4 receptor, and thus can even be used to detect 5-HT4 receptors in living cells because they can be employed without prior permeabilization. Transcripts encoding the 5-HT4 receptor have been observed by in situ hybridization in subsets of neurones in both submucosal and myenteric plexuses and in some of the smooth muscle fibres of the muscularis externa. Antibodies have immunocytochemically revealed 5-HT4 receptors in subsets of neurones in both plexuses; however, 5-HT4 immunoreactivity has mainly been found on nerve fibres in the ganglionic neuropil and on nerve fibres within the circular layer of smooth muscle. At the electron microscopic level, the receptor sites have been demonstrated to be primarily located on the plasma membranes of neurites (presumably axons), and to be concentrated at the presynaptic membranes of synapses (Figure 4). Drugs that act as 5-HT4 agonists would therefore be expected to promote transmission in prokinetic pathways by enhancing neurotransmitter release and, in contrast with a drug that targets the 5-HT1P sites, would depend on natural stimuli to evoke peristaltic and secretory reflexes. Such drugs would not be anticipated to induce perpetual or overwhelming motility by permanently activating reflexes at their origins. A 5-HT4 agonist that lacks other non-specific effects would therefore be an attractive prokinetic agent because it would, in principle, be expected not only to be effective, but also safe. Tegaserod, a partial 5-HT4 agonist, is such a drug.

Figure 4.

Immunoelectron microscopic demonstration of 5-HT4 receptor immunoreactivity in a terminal varicosity within the neuropil of a myenteric ganglion. The appearance of the tissue is different from that obtained with standard methods of fixation with glutaraldehyde and OsO4. That type of preparation denatures proteins and destroys their antigenicity. To preserve 5-HT4 receptor immunoreactivity, tissue was fixed with 3.75% acrolein and embedded in a hydrophilic resin (LR Gold). Antibodies to the 5-HT4 receptors and secondary antibodies coupled to 10 nm particles of colloid gold were applied to sections, which were then osmicated to improve contrast. Note that 5-HT4 receptor immunoreactivity (arrows) is presynaptic and associated with membranes.

The locations of 5-HT4 receptors and their actions (to enhance acetylcholine and calcitonin gene-related peptide release in prokinetic pathways) define the uses and limitations of tegaserod. If the motility of the gut is deficient, as it is when irritable bowel syndrome is associated with constipation, in chronic constipation and, possibly, in functional dyspepsia, tegaserod is likely to be extremely beneficial; thus tegaserod consistently outperforms placebo when matched against these conditions.68–74 In contrast, tegaserod cannot be of benefit if enteric nerves have degenerated or have become non-functional. Chronic intestinal pseudo-obstruction and diabetic gastroparesis are therefore challenges that the mechanism of action of tegaserod is not equipped to meet. Synaptic transmission in prokinetic pathways must be there in order to be strengthened by tegaserod. Similarly, if prokinetic pathways have degenerated, tegaserod cannot make them regenerate or find a means of circumventing missing enteric microcircuits. Moreover, strengthening transmission in prokinetic pathways is likely to be counterproductive when excessive motility is part of the problem, as it is in irritable bowel syndrome associated with diarrhoea.

5-Ht3 receptors mediate transmission of signals from the gut to the cns

The gut is not just a secreting motility machine. Motile and secretory output is obvious, but sensory reception from the bowel is also important and, when not right, enteric sensation can be devastating. Nausea, pain, bloating and urgency have a profound tendency to focus the mind. Visceral hypersensitivity is a prominent component of irritable bowel syndrome and is often the most distressing. Over 90% of vagal fibres are sensory, carrying information from the gut to the brain.75–77 These fibres, moreover, are amply supplemented by spinal nerves. Not all of the information that is sent by the bowel to the brain necessarily comes to consciousness. Vagal stimulation has been used to treat depression and epilepsy, and it can improve learning and memory.78–81 It is thus conceivable that natural stimulation of vagal fibres in the bowel can do the same things, or that their opposites occur when the vagus is inappropriately activated in an abnormal bowel. The gut can function when cut off from the CNS but, in real life, it rarely, if ever, is called upon to do so. The CNS, no less than the ENS, is informed about conditions prevailing in the lumen of the bowel. To a great extent, the mechanisms of sensation appear to be similar, except that the responding nerves for carrying signals to the CNS are the extrinsic fibres originating from neurones that reside in cranial nerve or dorsal root ganglia. There is, however, a profound difference in the receptors used by 5-HT to activate extrinsic sensory nerves and submucosal IPANs. In contrast with 5-HT1P and 5-HT4 receptors, which stimulate submucosal IPANs and enhance neurotransmission from them, respectively, 5-HT3 receptors activate extrinsic sensory nerves.82–85 As a result, 5-HT3 antagonists are very effective in curbing the untoward effects of visceral hypersensitivity. They are also effective in treating the nausea and vomiting associated with cancer chemotherapy, which evidently stems in part from the actions of 5-HT in the gut.86, 87 Ondansetron and granisetron are used to counteract the nausea of chemotherapy, and alosetron has demonstrated real efficacy in treating irritable bowel syndrome. Importantly, the centrality of the 5-HT3 receptor in the transmission of noxious signals to the CNS means that a 5-HT4 agonist, such as tegaserod, can be administered without fear of activating sensory nerves and inducing unwanted sensations from the gut.

Motility can be slowed by antagonizing 5-Ht3 receptors within the ens

The distribution of 5-HT3 receptors in the gut includes, but is not by any means limited to, extrinsic sensory nerves.88, 89 They are also expressed by myenteric neurones, where they mediate fast excitatory neurotransmission at the synapses of serotonergic interneurones.90–93 5-HT3 receptors are also responsible for mediating the activation of mucosal terminals of myenteric IPANs by 5-HT.44, 94, 95 Therefore, 5-HT3 antagonists not only inhibit the transmission of sensory signals to the CNS by acting on the receptors expressed by extrinsic sensory nerves, but also slow intestinal motility, presumably by interfering with serotonergic neurotransmission within the ENS and by blocking the initiation of reflexes, such as giant migrating contractions, that may be initiated by myenteric IPANs. Peristaltic and secretory reflexes persist in the face of 5-HT3 receptor antagonism because these reflexes are triggered by submucosal IPANs, which are activated by other 5-HT receptors.95–97 5-HT3 antagonists thus do not paralyse the bowel; nevertheless, 5-HT3 antagonists do have a tendency to be constipating,98–100 and thus must be used in irritable bowel syndrome only when diarrhoea is the predominant component, and certainly should never be administered to an individual when constipation is a significant element of the condition. In individuals who do experience constipation, moreover, it is not necessary to resort to 5-HT3 antagonists to combat visceral hypersensitivity and, in particular, the problem of bloating. There is evidence that tegaserod alleviates both of these problems.71, 101 The mechanism by which tegaserod relieves intestinal discomfort is not entirely clear, but it is possible that sensation from the gut improves when abnormalities of intestinal motility, compliance and tone are corrected by tegaserod.

Colonic motility is abnormally slow in transgenic mice that lack 5-Ht4 receptors

Considerable recent insight has been obtained from studying the bowel of transgenic mice that lack 5-HT4 receptors.67 Colonic motility is abnormally slow in these animals, confirming that the 5-HT4 receptor plays a critical role in the maintenance of normal colonic motility. Defects in the motility of the upper gastrointestinal tract, however, have not yet been observed, probably because the gut of the 5-HT4 knockout mice has thus far been investigated only with relatively crude tools and only under basal conditions. The bowel, however, may behave relatively normally when it is not stressed. Under postprandial or other conditions in which the gut is called upon to do work, the 5-HT4 receptor may be more necessary for maintaining adequate motility than when it is resting. Still, the observations to date imply that 5-HT4 receptors are critically important, even under resting conditions, for normal motility to occur in the colon. Surprisingly, the wall of the colon of 5-HT4 knockout mice appears to be quite thin. The thickness of the mucosa is the same in 5-HT4–/– and wild-type mice; the difference between knockout and normal mice is mainly in the muscularis externa. The 5-HT4 knockout animals also have fewer myenteric and submucosal neurones in the colon than do their wild-type littermates. These abnormalities appear to be more severe in older animals than at birth, suggesting that long-standing failure to receive 5-HT4 receptor stimulation may lead to atrophy of colonic neurones and smooth muscle. If this is true, it could suggest that patients with irritable bowel syndrome associated with constipation and those with chronic constipation might benefit from early treatment with tegaserod. If a lack of 5-HT4 stimulation contributes to constipation in these individuals, it might be important to prevent it from becoming a long-standing problem, lest it lead to atrophy of the neuromuscular apparatus of the colon.

Mucosal expression of sert is decreased in the bowel in inflammation and irritable bowel syndrome

A difficulty that has plagued research into irritable bowel syndrome for a long time is the lack of a physical defect characteristic of the condition. As a result, irritable bowel syndrome must be defined as a symptom complex, increasing the likelihood that patients with heterogeneous ailments are included in the category of irritable bowel syndrome.102–104 The definition of irritable bowel syndrome, moreover, has changed over the years and, although an internationally agreed upon definition (Rome II) now exists, a new panel is meeting and the definition is thus unlikely to be immutable. Clearly, as long as the definition of the problem changes, there can be no guarantee that all patients experience the same illness. Patients, no less than the scientists who want to help them, are troubled by the need to define irritable bowel syndrome by its characteristic symptoms. Irritable bowel syndrome is dismissed by much of the public as ‘a state of mind’, and thus is de-legitimized as a disease. In the absence of anatomical or chemical evidence to the contrary, it is difficult to refute the argument that irritable bowel syndrome is not a disease but ‘a state of mind’, and indeed many physicians probably believe that is just what irritable bowel syndrome is. Certainly, it is true that anxiety or depression can change bowel function. On the other hand, it is also correct that changes in bowel function can be a cause of anxiety or depression. Apart from the obviously depressing nature of intractable cramps, urgency, bloating, diarrhoea and constipation, there is also the possibility alluded to above that signals from the bowel that do not come to consciousness might alter mood. Recent investigations of SERT expression in the mucosa, however, suggest that irritable bowel syndrome is more than just ‘a state of mind’. There is a molecular defect in the bowel in irritable bowel syndrome.

As noted above, SERT is normally expressed in mucosal enterocytes and plays a critical role in the termination of the action of 5-HT in the mucosa. Inflammation of the mucosa is associated with a profound decrease in the expression of SERT. This decrease has been detected in experimental inflammation induced in the guinea pig colon,105 and has also been observed in biopsies of human colonic mucosa from patients with ulcerative colitis.37, 106, 107 Interestingly, mucosal SERT expression is decreased in both constipation-predominant and diarrhoea-predominant irritable bowel syndrome.37 The decrease in SERT expression in the gastrointestinal mucosa of human subjects with ulcerative colitis and constipation-predominant or diarrhoea-predominant irritable bowel syndrome has been quantified by measurements of mRNA encoding SERT by real-time reverse transcriptase-polymerase chain reaction, and has been demonstrated by immunocytochemistry to extend to SERT protein. Other elements of serotonergic signalling are also affected. The expression of tryptophan hydroxylase-1 is reduced in the mucosa together with SERT. The effects predicted to be exerted by a decrease in SERT expression are similar to the abnormalities of gastrointestinal function and sensation that are observed in ulcerative colitis and constipation-predominant or diarrhoea-predominant irritable bowel syndrome. An increase in the availability of free 5-HT, which would be expected to be caused by a defect in its inactivation due to decreased SERT, would be expected to prolong the contact time of 5-HT with its receptors and to recruit more cells responding to a stimulus of a given strength. The resulting potentiation of the effects at 5-HT3 receptors would be expected to cause discomfort due to stimulation of the 5-HT3 receptors on extrinsic sensory nerves. Potentiation of 5-HT1P- and 5-HT4-mediated effects would increase secretion and tend to cause diarrhoea. Continued occupancy of the receptors would lead to their desensitization, which would mimic the effects of 5-HT1P and 5-HT4 antagonists and cause constipation. Interestingly, the targeted deletion of SERT in transgenic mice leads to a state of alternating diarrhoea and constipation.36 The diarrhoea in SERT knockout mice is associated with excessively rapid colonic motility and results in increased excretion of water in the stool. The diarrhoea in SERT knockout mice is not permanent, however, but alternates with constipation. During periods of constipation, colonic motility is excessively slow, as it is in mice that lack 5-HT4 receptors. The mice spend most of their time with diarrhoea; episodes of constipation are transient. The pattern can be explained as diarrhoea due to 5-HT potentiation, transiently interrupted when something as yet unidentified intervenes in the life of the mouse and increases the release of 5-HT. Animals that lack SERT cannot handle this challenge and receptors desensitize. In fact, 5-HT3 receptors undergo a molecular change in response to SERT knockout and become less sensitive to 5-HT and more likely to desensitize.107


The original work described in this paper was supported by NIH grants NS12939 and NS15547 and a research grant from Novartis.