Address for Correspondence Robin Spiller, Professor of Gastroenterology, Wolfson Digestive Diseases Centre, Nottingham, UK. Tel: +44 115 823 1032; fax: +44 115 942 2232; e-mail: email@example.com
Abstract Serotonin (5-hydroxytryptamine, 5-HT) is present in abundance within the gut, most stored in enterochromaffin cell granules. It is released by a range of stimuli, most potently by mucosal stroking. Released 5-HT stimulates local enteric nervous reflexes to initiate secretion and propulsive motility. It also acts on vagal afferents altering motility and in large amounts induces nausea. Rapid reuptake by a specific transporter (serotonin transporter, SERT) limits its diffusion and actions. Abnormally increased 5-HT is found in a range of gastrointestinal disorders including chemotherapy-induced nausea and vomiting, carcinoid syndrome, coeliac disease, inflammatory bowel disease and irritable bowel syndrome (IBS) with diarrhoea (IBS-D), especially that developing following enteric infection. Impaired SERT has been described in IBS-D and might account for some of the increase in mucosal 5-HT availability. 5-HT3 receptor antagonists inhibit chemotherapy-induced nausea and diarrhoea associated with both carcinoid syndrome and IBS. While IBS-D is associated with increased 5-HT postprandially, IBS with constipation (IBS-C) is associated with impaired 5-HT response and responds to 5-HT4 agonists such as Prucalopride and 5-HT4 partial agonists such as Tegaserod.
There are a range of disorders in which serotonin (5-hydroyxtryptamine, 5-HT) metabolism is known to be abnormal (Fig. 1). Before addressing the role of serotonin in functional gut disorders, it may be valuable initially to examine some organic diseases in which the mechanisms of abnormal 5-HT metabolism are more clearly understood. It was the recognition that cisplatinum caused nausea and vomiting via the release of 5-HT1 which stimulated the development in the 1990s of the 5-HT3 receptor antagonists. Experience with these new drugs demonstrated their profoundly constipating and minor anxiolytic effects which lead to their exploitation in the treatment of diarrhoeal disorders. The early studies examined the effect of 5-HT3 receptor antagonists in the carcinoid syndrome2 and healthy volunteers.3 This lead to an increase in our understanding of the role of 5-HT in normal gastrointestinal (GI) physiology which will be considered first.
Role of 5-HT in normal physiology
The majority of the body's stored serotonin lies within the gut enterochromaffin (EC) cells but we should not ignore the important role 5-HT plays in signalling between interneurones in the enteric nervous system mediating motor and secretory responses.
Release of serotonin
The most reliable laboratory stimulus for release of mucosal 5-HT is stroking or brushing the mucosa.4 Cultured enteroendocrine cells are directly sensitive to deformation,5 responses which seem to depend in part on the local release of adenosine which acts on stimulatory A2 receptors and inhibitory A3 receptors. Hypertonic glucose solutions can also induce the release of 5-HT; however, similar effects are seen with both hypertonic mannitol and saline suggesting this is a non-specific effect which appears to require neural mediation as tetrodotoxin inhibits 5-HT release by 66%.6 Duodenal acid also stimulates 5-HT release, again dependent in part on neural input, being blocked by propranolol and atropine.7 Release in intact animals can also occur indirectly via neural pathways as is seen in animals with a Thiry Vella loop in whom a meal stimulates an immediate increase in 5-HT in the loop, which is not exposed to any nutrients.8 5-Hydroxytryptamine released in the mucosa is rapidly taken up by an active transport process (serotonin transporter, SERT) either by enterocytes or vascular endothelial cells during its passage through the liver and lung during which time around 40% is removed.7 Once within the cell 5-HT is metabolized by monoamine oxidases to 5-hydroxyindole acetic acid (5-HIAA) and by UDP-glucuronosyl-transferase to serotonin 5-O-glucuronide which, in CaCo2 cells at least, accounts for 35% of 5-HT metabolism.9
Action of 5-HT released into the mucosa
Exogenous 5-HT added to the mucosa stimulates mucosal secretions both directly acting on epithelial 5-HT2 and 5-HT4 receptors and indirectly acting by 5-HT1P/4 receptors on primary afferent neurones and 5-HT3 receptors on interneurones and secretor motor neurones (Fig. 2).10 Neural activation of submucosal and myenteric neurones by puffs of nitrogen appear mediated via 5-HT release, as they are blocked by 5-HT1P receptor antagonist N-acetyl-5-hydroxytryptophyl 5-hydroxytryptophan amide (5-HT-DP).11 Intestine secretion in the rodent is mediated by 5-HT4 receptors12 in both the ileum and the colon,13 and studies in stripped human jejunal mucosa show secretion was not blocked by tetrodotoxin nor hexamethonium but was by the 5-HT4 receptor antagonist ICS 205-930, suggesting a direct effect via 5-HT4 receptors.14 In addition to stimulating secretions, 5-HT also stimulates the peristaltic reflex by activating ascending contractions and descending inhibition. These reflexes, which show important species differences, are mediated by 5-HT4 receptors in man15 and a combination of 5-HT3 and 5-HT4 receptors in guinea pigs. The 5-HT1P antagonist, 5-HT-DP, inhibited activation of primary afferent neurones as assessed by CGRP release. These studies suggested that 5-HT activates the primary afferent neurones via 5-HT1P receptors and that 5-HT3 and 5-HT4 receptors are involved later in the reflex.
Effects of serotonin agonists in intact human beings
Most of the above studies have been carried out on isolated preparations which may not give an accurate reflection of effects in an intact organism. Numerous studies in animals have demonstrated that 5-HT infusions stimulate gastric antral, duodenal and ileal contractions16 and increase the frequency of the phase III of the migrating motor complex (MMC),17 an effect which is inhibited by both the 5-HT3 receptor antagonist Ondansetron and the 5-HT4 receptor antagonist, GR113808. However, there are much fewer studies in man owing to the associated side effects including nausea, vomiting and diarrhoea, though these can be minimized using low doses. Lordal et al.18 infused 5-HT at 5 and 15 nmol min kg−1 which increased plasma 5-HT from a basal of 2 nmol L−1 to 10 and 30 nmol L−1, respectively. This caused marked stimulation of small bowel motor activity and an increase in MMC frequency which occurred without inducing nausea in the majority of subjects given the low dose. A 5-HT3 agonist, MKC-733 given by mouth (2 and 4 mg), similarly increased the number of MMCs in the antrum and duodenum, delayed gastric emptying but accelerated small bowel transit. This was associated with subsequent passage of loose stools later in the day in the majority of subjects taking the highest (4 mg) dose. Subsequent clinical trials have shown a laxative effect accelerating colonic transit in patients with constipation even at much lower doses, 0.2 and 0.5 mg b.i.d.19
Effect of elevated plasma 5-HT in carcinoid syndrome
Carcinoids are slow-growing neoplasms most commonly originating in the midgut. Symptoms vary but initially include obstructive symptoms and diarrhoea. The well known ‘carcinoid syndrome’ of cutaneous flushing, bronchospasm and diarrhoea are usually only seen once the tumour has metastasized to the liver allowing tumour products, which include 5-HT, substance P and other neuroendocrine products, to escape hepatic metabolism. Plasma 5-HT levels are often markedly elevated and can be further increased by i.v. pentagastrin. Stimulating release in this way causes flushing and diarrhoea which can be inhibited by the 5-HT2 receptor antagonist ketanserin.20 Pentagastrin appears to act indirectly by releasing adrenaline which acts on β-adrenoreceptors on EC cells, an effect which can be blocked by adrenalectomy or by the β-adrenoreceptor antagonist, propranolol.21 Somatostatin can inhibit 5-HT release induced by pentagastrin22 and is widely used preoperatively to avoid a carcinoid crisis due to release of tumour products. Somatostatin has become an important part of the treatment of carcinoid syndrome.23 It reduces urine 5-HIAA around 25%24 suggesting a partial control of 5-HT secretion. Other benefits may include inhibition of tumour growth and secretion of other tumour products. Physiological studies in patients demonstrate marked acceleration of small bowel and colonic transit together with an increase in colonic tone, particularly exaggerated postprandially.25 This exaggerated response is mediated in part by serotonin, as it can be reduced by about 50% by i.v. Ondansetron 0.15 mg kg−2.
Cisplatinum-induced GI disturbances
Cisplatinum, above all other components of chemotherapy regimens, appear responsible for inducing profound nausea and vomiting largely due to 5-HT release. Microdialysis techniques show doubling of plasma 5-HIAA in the hours immediately following Cisplatinum injection, in a manner which parallels the onset of vomiting. There was an associated marked increase in urine excretion of 5-HIAA over the next 6 h. 5-HT3 antagonists Ondansetron and Granesetron, both effective in inhibiting the nausea fail to alter either 5-HIAA blood levels or urinary excretion26 indicating they do not substantially alter cisplatinum-induced 5-HT release. The emetic reflex appears mediated via the vagus rather than via systemic action, as vagotomy totally abolishes the vomiting induced by Cisplatinum in ferrets.27 This fits with nerve recoding studies which show that the vagus has abundant 5-HT3 receptors and responds to local 5-HT.28,29
The introduction of screening endomysial and tissue transglutaminase antibody tests means that coeliac disease is now being increasingly recognized, found with a frequency of around 3% in patients presenting with otherwise typical irritable bowel syndrome (IBS). The characteristic features include nausea, bloating and diarrhoea leading to a reduction in food intake. This is the main cause of weight loss rather than malabsorption, which is often relatively minor. Early studies showed increased concentrations of 5-HT in the duodenal mucosa,30 increased plasma 5-HT levels31 and increased urine excretion of the 5-HT metabolite, 5-HIAA.32 A recent study in which all these assessments were performed simultaneously confirmed these findings and demonstrated a marked increase in postprandial 5-HT which correlated with dyspepsia.33 Duodenal biopsies confirmed crypt hyperplasia and villous atrophy in the coeliac patients which was associated with a more than twofold increase in enteroendocrine cells per millimetre of mucosa. Peak postprandial plasma levels 99 (59–139) nmol L−1 approached the lower limit of those seen in carcinoid syndrome25 and correlated with dyspepsia.
Duodenal biopsies confirmed a doubling of 5-HT mucosal concentrations but interestingly 5-HIAA levels were depressed giving a significantly reduced 5-HIAA/5-HT ratio. Within a mucosal biopsy, 5-HT comes from the EC cells while 5-HIAA will come from 5-HT which has been taken up into enterocytes and metabolized by intracellular monamino-oxidases (MAO). Reduction in 5-HIAA could reflect either impairment of MAOs or impairment of serotonin uptake. 5-HT being a polar molecule requires an energy-dependant sodium-linked active transporter to allow its uptake into the enterocyte. Preliminary data34 indicates a reduction in mRNA for SERT, out of proportion to the reduction in mRNA of another intracellular protein cytokeratin 20, a marker of enterocytes. Serotonin transporter in enterocytes appears similar to that in platelets and previous studies have indicated that platelets from patients with coeliac disease express fewer H3-impramine-binding sites, which is used as a measure of SERT. The impairment of SERT both in the enterocyte and in platelets might explain the higher plasma levels.
Coeliac disease is an allergy to the wheat protein gliadin in which mucosal injury is mediated by activated T cells. There is an associated acceleration of cell turnover and increased enterocyte apoptosis. Enterochromaffin cells are differentiated from stem cells during their proliferation but may have a longer half-life than enterocytes. There is also an increase in CCK-containing enteroendocrine cells, though in this case, CCK release is impaired.35 This T-cell dependence of enteroendocrine cell hyperplasia is a common theme and noted in numerous animal models also characterized by EC hyperplasia including infection with Trichinella spiralis and the rat tape worm Hymenolepis diminuita. In T. spiralis, T-cell receptor knock out mice fail to show EC hyperplasia in response to infection,36,37 which can also be inhibited by steroid therapy in wild-type animals.37 This T-cell dependence is likely to be true in humans and CD3+ T-lymphocyte counts correlate strongly with enteroendocrine cell counts in patients after enteritis caused by Campylobacter jejuni.38,39
Acute Campylobacter enteritis
A study of serial rectal biopsies in patients with enteritis due to C. jejuni showed an increase in enteroendocrine cell numbers which peaked at 2 weeks postinfection but had still not returned to baseline by 3 months. Interestingly, the ratio of 5-HT/Peptide YY (PYY)-containing cells reversed, with the normal predominance of PYY, being replaced with a predominance of 5-HT-containing EC cells within the rectum.38 The same study showed that patients with a history of IBS developing after infection also had abnormally elevated EC cell count [see postinfective IBS (PI-IBS) below]. These findings were confirmed in a much larger study of over 750 patients with Campylobacter enteritis 105 of whom developed PI-IBS. When 28 who developed PI-IBS were compared with 28 age- and sex-matched individuals who were infected but did not develop PI-IBS the PI-IBS patients showed a 20% increase in EC cells.39 Furthermore, in a multivariate analysis, each standard deviation increase in EC cells was associated with a relative risk of developing PI-IBS of 3.8 [1.3–9.5; 95% confidence intervals (CI)].
Inflammation in general appears to increase mucosal 5-HT content so the change associated with Campylobacter infection is unlikely to be specific to any one bacterial species and may well underlie all cases of PI-IBS.
Chronic inflammatory bowel disease
Acute inflammation in trinitrobenzene sulphonic acid (TNBS) colitis4 and TNBS ileitis40 is associated with increased EC cell numbers, increased 5-HT release and evidence of impaired SERT, changes which all lead to increased 5-HT availability. Numerous studies in man have also shown increases in enteroendocrine cell numbers in inflammatory bowel disease. In colectomy specimens for ulcerative colitis and Crohn's there was a near doubling in serotonin-containing enteroendocrine cell numbers though interestingly, some peptides, such as PYY, showed a small decline.41 This confirmed earlier studies of increased EC cell numbers in ulcerative colitis.42 Similarly, in Crohn's ileitis EC cells were also doubled.43 EC cell numbers were also increased nearly fourfold in surgical specimens resected for radiation enteritis.44 Ulcerated and damaged mucosa might also be expected to show impairment of metabolism and in a biochemical study of both Crohn's and ulcerative colitis. Magro et al.45 showed depressed levels of noradrenaline and 5-HT in patients with active Crohn's disease. There was further evidence of impaired synthesis of dopamine with depressed levels associated with elevated levels of its precursor l-DOPA. Similarly, when studying the precursors and metabolites of serotonin, the ratio of 5-HT to its precursor 5-hydroxytryptophan was depressed in active Crohn's disease and in ulcerative colitis implying impairment of synthesis. Paradoxically, there was an increase in 5-HIAA/5-HT ratio, implying that uptake and metabolism of the released 5-HT was unimpaired.45 Regrettably, no measure of 5-HT release was made, so the precise impact on 5-HT availability is uncertain. Furthermore, there have been no studies of 5-HT3 antagonists in inflammatory bowel disease so the significance of these changes in serotonin metabolism is uncertain.
Irritable bowel syndrome
Increased numbers of EC cells were reported in IBS in the 1970s42 but it was not until Bearcroft et al.46 showed an increase in postprandial plasma 5-HT levels that interest in this area was rekindled. Soon after, several large clinical trials showed beneficial effects of 5-HT3 antagonists in diarrhoea-predominant IBS,47 supporting the concept that excess 5-HT might be contributing to diarrhoea in IBS. Postprandial levels of 5-HT were noted to be significantly elevated after a spaghetti meal in 39 IBS patients with diarrhoea. When patients were divided into those who were symptomatic following ingestion of the meal, 31 who were symptomatic had significantly greater increase in postprandial plasma 5-HT compared to those without symptoms and controls.48
Increased EC cells had been noted in individuals with Campylobacter enteritis38 and in PI-IBS.39 Subsequent studies in 15 individuals with PI-IBS using the same test meal as used by Houghton et al., showed a significant increase in area under the curve of 5-HT in female PI-IBS patients.49 Furthermore, in the same study, significant impairment of release was noted in 15 constipated IBS patients who also had prolonged colonic transit time.
Alterations in mucosal 5-HT metabolism in IBS
Mucosal 5-HT turnover was assessed by measuring the 5-HIAA/5-HT ratio and also calculating 5-HT content per EC cells. Constipated IBS patients showed similar numbers of EC cells to healthy controls but elevated 5-HT content such that the 5-HT content per EC cell was nearly double normal. The 5-HIAA/5-HT ratio was significantly depressed at 0.14 compared with 1.2 for health controls in keeping with the impaired release. In contrast, PI-IBS patients showed increased EC cell numbers, normal 5-HT content, but also a significant reduction in 5-HIAA/5-HT ratio to 0.21. This implies an impairment of 5-HT reuptake and metabolism. These findings were supported in a larger independent study of 29 IBS patients with constipation and 55 IBS patients with diarrhoea compared with 35 healthy controls. Irritable bowel syndrome patients with diarrhoea had significantly elevated postprandial concentrations while the constipated patients had reduced values.50 Constipated IBS patients had increased platelet 5-HT compared with IBS-D patients, again supporting the idea that SERT might be defective in IBS-D.50 The higher levels of platelet 5-HT in constipated IBS patients was also observed in Dunlop's study when compared with PI-IBS.49
Interpretation of these results is complicated by the fact that minor activation of platelets could lead to substantial increases in 5-HT. However, markers of platelet activation such as thromboglobulin were negative in these studies and plasma 5-HIAA which is not stored in platelets and hence not subject to this error showed very similar effects in the Atkinson study.50
These findings of increased mucosal 5-HT in constipated IBS are supported by previous small studies in eight constipation-predominant IBS patients who were shown to have higher levels than two diarrhoea-predominant IBS and seven normal controls, a pattern which was found throughout the colon.51
5-HT modulating agents for treatment of IBS
5-Hydroxytryptamine excess may be responsible for many features in IBS-D which would explain the success of 5-HT3 antagonists. It may also be true that in constipation, impaired release of serotonin contributes to the symptoms which may be alleviated in some patients by 5-HT4 agonists such as Prucalopride52 and Tegaserod.53 The precise site of action of these 5-HT-modulating agents remains uncertain as autoreceptors on the EC cells may themselves modulate endogenous 5-HT release. MKC-733, for example, has a very short half-life in plasma and is relatively poorly absorbed. Its prolonged duration of action may therefore be due to action of unabsorbed drug on 5-HT3 receptors on the mucosal EC cells as it passes down the gut. Similarly, Tegaserod, also has a bioavailability of only 10% implying that most of the drug is available to act locally on EC cells. Modulating the availability of 5-HT induces many changes with receptor desensitization which may reduce its efficacy with chronic usage. Nevertheless, both agonists and antagonists have proven clinical efficacy, though in both cases, the large number needed to treat suggests that improved ways are needed to target and characterize patients who are likely to respond. Simple biomarkers of responsiveness to these treatments remain to be discovered.
Conflict of interest statement
RS has received funding from Novartis Pharmaceuticals for research on serotonin metabolism in FGIDs.