Despite a recent surge in interest in the neurobiology SPECIAL REPORTS AND REVIEWS

Transcription

1 GASTROENTEROLOGY 2002;122: SPECIAL REPORTS AND REVIEWS Evolving Pathophysiologic Models of Functional Gastrointestinal Disorders EMERAN A. MAYER*, and STEPHEN M. COLLINS *CURE Neuroenteric Disease Program, UCLA Division of Digestive Diseases and Brain Research Institute, Los Angeles, California; and Department of Gastroenterology, McMaster University, Hamilton, Ontario In contrast to most other disorders of the digestive system, functional disorders of the gut continue to be defined by symptom criteria rather than by biological markers. At the same time, animal models of functional gastrointestinal disorders in which to test pathophysiologic hypotheses are lacking. The aim of this report is to critically review recently proposed conceptual as well as animal models of functional gastrointestinal disorders. Converging disease models have been proposed that postulate an enhanced responsiveness of neural, immune, or neuroimmune circuits in the central nervous system or in the gut to exteroceptive (psychosocial) or interoceptive (tissue irritation, inflammation, infection) perturbations of the organism s homeostasis. The enhanced responsiveness results in dysregulation of gut motility, epithelial function (immune, permeability), and visceral hypersensitivity, which in turn produce irritable bowel syndrome symptoms. These conceptual models provide plausible mechanisms for irritable bowel syndrome symptom generation and are consistent with extensive epidemiologic and pathophysiologic data. Several animal models have recently been proposed that mimic key features of these conceptual disease models. They fall into models triggered by centrally targeted stimuli (neonatal stress, post-traumatic stress disorder) or those triggered by peripherally targeted stimuli (infection, inflammation). Depending on the timing of the trigger (neonatal vs. adult), the changes induced in the animal may be permanent or transient. Future development of existing and novel models involves the use of transgenic and knockout animals, as well as the demonstration of predictive validity in terms of responsiveness to candidate drugs. Despite a recent surge in interest in the neurobiology and drug development of functional gastrointestinal (GI) disorders, wide gaps remain in our knowledge of the etiology and pathophysiology of the most common functional GI syndromes, such as irritable bowel syndrome (IBS), functional dyspepsia, and noncardiac chest pain. However, although functional GI disorders continue to be defined by symptom criteria, 1 significant progress in the mechanistic understanding of these syndromes has recently been made 2 : well-designed epidemiologic information has clearly implicated psychosocial stressors 3 6 and acute gastroenteric infections 6,7 as central and peripheral triggers of first symptom onset or exacerbation. Perceptual hypersensitivity to physiological and experimental visceral stimuli ( visceral hypersensitivity ) has emerged as an important component for symptom generation 8 that may interact with coexisting alterations in GI motility and in intestinal secretory function. Recent data also implicate alterations in autonomic 9 and neuroendocrine responses to visceral stimuli 10 consistent with altered responsiveness of central stress circuits. The current review was aimed at integrating published clinical and human pathophysiological data 11 recently reviewed by a panel of experts 2 into conceptual models of the disease suitable for the development of relevant animal models. Rather than a representative evaluation of all existing conceptual models of IBS, the current review has selected 2 (interrelated) disease models that meet the following criteria: (1) They can provide plausible explanations for certain key pathophysiological and clinical features of IBS (face validity) outlined in Tables 1 and 2. (2) Key pathophysiological features can be reproduced in animal models (Table 2). (3) The models allow for the generation of hypotheses that can be tested in both patients and animal models. Abbreviations used in this paper: CRD, colorectal distention; CRF, corticotropin-releasing factor; EC, enterochromaffin cells; EMS, Emotional Motor System; GABA, -aminobutyric acid; GI, gastrointestinal; HPA, hypothalamus-pituitary-adrenal; IBS, irritable bowel syndrome; LC, locus coeruleus; NA, noradrenergic; NE, norepinephrine; PAG, periaqueductal grey; PFC, prefrontal cortex; PI-IBS, postinfective IBS; PTSD, post-traumatic stress disorder; SERT, serotonin reuptake transporter by the American Gastroenterological Association /02/$35.00 doi: /gast

2 June 2002 IBS ANIMAL MODELS 2033 Table 1. Existing Animal Models of IBS CNS-directed pathogenesis ( stress memory ) Early life Neonatal psychological stress 12 Adult life PTSD 13 Gut-directed pathogenesis ( pain and immune memory ) Early life Colonic inflammation/pain 14 Adult life Colitis 15 Postinfectious 16 PTSD, post-traumatic stress disorder. Heterogeneity Even though currently used symptom criteria implicitly assume a homogeneous disorder of IBS, it is likely that the non-specific symptom complex of abdominal pain and discomfort associated with alterations in bowel habits reflects a heterogeneous group of disorders with different etiologies (e.g., primary central nervous system [CNS] vs. gut-directed etiologies; genetic vulnerability), pathophysiologies (gut motor, sensory, or secretory changes), and clinical presentations (diarrhea vs. constipation vs. pain). One of the reasons for similar GI symptom expression may result from the fact that there are a limited number of perceptual (pain/discomfort) and behavioral (bowel movements) responses to gut stimulation, regardless if the stimuli are primarily targeted centrally or intraluminally, and regardless of underlying mechanisms. This point is illustrated by another group of GI disorders previously referred to as the syndrome of chronic idiopathic intestinal pseudo-obstruction. 27 Although initially, the clinical syndrome (abdominal pain and discomfort associated with radiological evidence for gut dilatation) was considered a homogeneous disorder of GI motility, it has since been shown that a variety of different syndromes with different etiologies and different underlying pathophysiologies can produce identical symptoms. 27 Further heterogeneity in clinical presentation of functional GI disorders may arise from the presence of various co-morbid disorders, such as affective disorders (reviewed in Mayer et al., 2001), 21 fibromyalgia, 23,28 or interstitial cystitis. 24 For example, the fact that such extraintestinal symptoms/syndromes are not seen in all patients suggests that they may not be a necessary element in the pathophysiology of IBS. Alternatively, there may be a non-specific core abnormality in IBS patients (vulnerability factor[s]), in the absence of Table 2. Relationship Between Findings and Mechanisms of IBS Found in Humans and Animal Experimental Data Clinical symptom Human experimental finding Putative pathophysiologic mechanisms Animal experimental data Diarrhea Passage of mucus Abdominal pain Abdominal discomfort Overlap with functional dyspepsia Comorbidity with affective disorders Extraintestinal symptoms Increased distal colonic motility and secretory response to food and stress Increased epithelial secretory response Tenderness to palpation and endoscopic manipulation, lower rectosigmoid pain thresholds to balloon distention Non-painful discomfort (bloating type symptoms); lower discomfort thresholds in GI tract to ascending series of distentions Delayed gastric emptying, altered small intestinal motility 40% comorbidity with anxiety disorders 21 Fibromyalgia 23 Interstitial cystitis 24 Fatigue, decreased energy Decreased libido 25 Modulation of distal colonic motor and secretory function by increased output from LC complex via sacral parasympathetic nucleus 17 Stress-induced visceral hyperalgesia, hypervigilance, mechanism(s) unknown Preattentive and attentive hypervigilance towards visceral afferent input (ascending NA outputs from LC complex) 20 Stress-induced delay in gastric emptying Enhanced responsiveness of central CRF/NA systems 22 Failure to activate stress-induced somatic analgesia mechanisms? Hypervigilance Overlap of afferent and efferent innervation of pelvic organs 17 Increased activity of central NA systems, altered HPA axis regulation Stress-induced down-regulation of sex hormone secretion 26 Stress-induced and CRF-induced increase in pellet output and colonic motility (central and peripheral CRF 1 -R) 18,19 Stress-induced visceral hyperalgesia Stress-induced hypervigilance Stress-induced and CRF-induced delayed gastric emptying Enhanced anxiety-like behavior LC, locus coeruleus; CRF, corticotropin-releasing factor; CRF 1 -R, corticotropin-releasing factor; NA, noradrenergic; HPA, hypothalamus-pituitaryadrenal.

3 2034 MAYER AND COLLINS GASTROENTEROLOGY Vol. 122, No. 7 which neither peripheral nor central factors may be able to trigger IBS symptoms. Evolving IBS Models Disease models (conceptual models and animal models) provide testable hypotheses essential for rational research into pathophysiological mechanisms underlying IBS and other functional GI disorders. Without such models, the field of functional bowel research will lack credibility compared with other fields of gastroenterology, and progress in the area of drug development will continue to be hampered by lack of rational targets. Disease models can be assessed based on their face validity (if the model resembles one or several clinical or pathophysiological characteristics of the clinical condition), construct validity (if the model is consistent with a particular theoretical rationale, e.g., enhanced stress responsiveness or postinfectious dysmotility), and their predictive validity (how well the model predicts treatment responses to specific drugs or nonpharmacologic interventions). 29 Even though considerable interactions between perturbations aimed at the gut (physical stressors) and the CNS (psychological stressors) are likely, we will discuss these models in terms of their pathogenesis as primary gut-directed (physical stressors), and primary CNS-directed (psychosocial stressors). Conceptual IBS Models Different conceptual models for IBS pathophysiology have been proposed to take into account the available body of pathophysiologic and epidemiologic data at a given time. Although traditional reductionistic models had focused primarily on isolated peripheral aspects of IBS, such as epithelial function ( mucous colitis ), GI motility ( spastic colon ), or a combination of both ( spastic colitis ), more recent conceptual models (visceral hypersensitivity) 8 and postinfective IBS (PI-IBS) 6,7 have taken into account both peripheral and central dimensions of the disorder. Different authors have proposed mind-body or cognitive-behavioral models of the disease and Drossman 33 first applied the concept of a biopsychosocial model to IBS, to accommodate the interacting role of genetic and biological factors with psychosocial and cognitive factors in the pathophysiology of IBS. Conceptual Model Focusing on Primary CNS-Directed Pathogenesis of IBS The following information has been selected from the literature that supports the concept of an enhanced stress responsiveness playing a role in development of IBS symptoms in a subset of patients. Valentino et al. 17 have proposed a neurobiological model of IBS that is based primarily on mechanistic studies in animals. It has as its key component an altered responsiveness of pontine nuclei resulting in abnormal brain and gut responses to peripheral or central stimuli. Increased responsiveness of central noradrenergic (NA) systems, in particular of the locus coeruleus (LC) complex (including Barrington s nucleus), has been implicated as a central mechanism in IBS by different investigators (reviewed in Mayer, 2000) 34 to explain the wide variety of extraintestinal symptoms such as overlap with anxiety disorders, 35 stress sensitization, 36 and sleep disturbances 37 (see next paragraph). Incorporating key concepts previously proposed in the psychiatric literature and applied to the area of functional GI disorders by Tache et al., 38 Valentino et al., 17 Stam et al., 36 and Lydiard et al., 35 we have developed a comprehensive model based on the concept of enhanced responsiveness of central stress circuits ( emotional motor system [EMS]), 39,40 which has been reported in detail elsewhere. 34,41 The key features and components of this model are summarized in Figure 1A. Suitability of enhanced EMS responsiveness model for development of animal model. An IBS disease model based on enhanced stress responsiveness fulfills the 2 criteria listed at the beginning of this review: (1) It provides a plausible explanation for intestinal and extraintestinal symptom complexes in IBS. (2) There are several animal models available to evaluate the validity of the concept (Table 2). As shown in Table 1, the proposed model provides plausible and testable explanations for the many reported clinical and pathophysiological aspects of IBS: (1) dysregulation of GI motility and secretion 44 ; (2) enhanced perceptual responses to visceral stimuli and decreased responses to somatic stimuli 45 ; (3) altered gut epithelial function; (4) overlap with other functional GI syndromes, such as functional dyspepsia and noncardiac chest pain 46 ; (5) comorbidity with affective disorders 21 ; (6) stress-sensitivity 3 5 ; (7) presence of common extraintestinal symptoms, such as fatigue, decrease in libido, and poor sleep. 25,47 Conceptual Model Focusing on Primary Gut-Directed Pathogenetic Mechanisms in IBS The following clinical observations in IBS patients have been selected to support the model of alterations in the gut immune system and gut neuro-immune interactions that may play a role in development of IBS symptoms in a subset of patients. Putative role of inflammation. Based on results from animal studies, low-grade inflammation or immune

4 June 2002 IBS ANIMAL MODELS 2035 Figure 1. (A) Enhanced stress responsiveness model of IBS. 41 Brain mechanisms. The central component of the model is the EMS, a system of specific networks concerned with the induction of emotions (fear, anger, sadness, etc.) and the emotion-specific activation of parallel effector systems. 39,40 These effector systems ultimately modulate target cells in the gut, producing alterations in various GI functions. 41 Medial prefrontal structures (ventromedial cortex, perigenual and infragenual cingulate cortex) receive input from lateral prefrontal cortex (PFC). Output from medial PFC modulates integrative structures of the forebrain (amygdala, hypothalamus) and hindbrain (periaqueductal grey [PAG]). Output from these regions in turn projects to pontomedullary nuclei and to the pituitary gland. The integrated output occurs in terms of autonomic, neuroendocrine, and pain modulatory responses to emotional stimuli and stressors. 42,43 PFC, prefrontal cortex; Hypoth., hypothalamus; LH, lateral hypothalamus; MH, medial hypothalamus; Ce, central nucleus of amygdala; Bvl, basal ventrolateral nucleus of amygdala; dlpag, dorsolateral PAG; lpag, lateral PAG; vlpag, ventrolateral PAG; Pit., pituitary. Modified from Bandler et al. 42 (B) Enhanced stress responsiveness model of IBS. 41 Putative bidirectional brain-gut interactions involved in modulating responsiveness of organism to CNS- and gut-directed stressors. Psychosocial stressors activate stress circuits within the EMS, and the resulting peripheral output in form of neuroendocrine (cortisol), corticotropin-releasing factor (CRF), and autonomic (norepinephrine [NE], epinephrine) responses shifts the mucosal immune system towards a Th2 response (increased mast cells, inducible nitric oxide synthase expression). 26,74 Autonomic responses can also directly or indirectly modulate gut permeability, thereby changing the access of luminal factors (antigens, bacteria) to the gut immune system. Luminal factors (physical stressors) modulate gut immune function, and immune products from the gut such as cytokines and chemokines can modulate the responsiveness of the EMS. 26 Temporal properties of the stressor and age to the animal at stress exposure are important determinants of type of neuroimmune interaction. ACTH, adreno corticotropin hormone; AVP, arginine vasopressin peptide; EPI, epinephrine; IL, interleukin; MC, mast cell; LIF, leukemia inhibitory factor; PVN, paraventricular nucleus; Th, T helper; TNF, tumor necrosis factor. Modified from Elenkov et al. 74 activation may be a basis for altered motility, and/or afferent and epithelial function of the gut in IBS, analogous to reported changes in asthma. 48 Several independent studies showed increased mast cell numbers in the muscularis externa of the colon, 49 and the ileal and colonic mucosa. 50,51 Increased cellularity of the colonic mucosa and lamina propria of IBS patients has been documented in selected 52 and in unselected IBS patients. 53 Preliminary results suggest increased inducible nitric oxide synthase expression in colonic biopsies of unselected IBS patients, 54 and it has been suggested that IBS patients may be genetically more susceptible to

5 2036 MAYER AND COLLINS GASTROENTEROLOGY Vol. 122, No. 7 inflammatory injury because of an inability to secrete sufficient quantities of anti-inflammatory cytokines. 55 PI-IBS. It has long been recognized that IBS-like symptoms may develop in patients recovering from enteric infection. 56,57 More recent studies have monitored the development of IBS symptoms in patients recovering from proven bacterial gastroenteritis 3 12 months previously. IBS-like symptoms were present in 7% 30% of patients. 7,58 60 A large cohort study identified a history of acute gastroenteritis as a major risk factor (relative risk 11.9) for the development of IBS. 61 Reported risk factors for the development of PI-IBS include female gender, the duration of the acute diarrheal illness, the absence of vomiting, and the presence of sustained psychosocial stressors around the time of infection. PI-IBS is not restricted to a particular organism and has been documented with a variety of bacterial infections (Salmonella, Campylobacter, and Escherichia coli) as well as parasitic infection. 62 However, the role of acute viral gastroenteritis in PI-IBS is unknown. Studies on function and tissue in PI-IBS patients have shown changes in intestinal motility 63,64 and epithelial function, including increased intestinal permeability and bile acid malabsorption, 65,66 and increase in the number of colonic enteroendocrine cells. 66 There is also histological evidence for alterations in mucosal immune function in PI-IBS, including an increased expression of interleukin-1 messenger RNA (mrna), 64 increased cellularity of the lamina propria and an increase in CD3 lymphocytes. 6 However, because increased cellularity of the colonic mucosa and lamina propria has been reported in unselected IBS patients as well, 53 it remains to be determined if altered gut immune function is a general characteristic of IBS patients, possibly related to an inability to efficiently down-regulate an inflammatory response. 55 The implication of stressful life events in the development of PI-IBS suggests a convergence of central and peripheral mechanisms in the expression of this syndrome. IBS in patients in remission from inflammatory bowel disease. The hypothesis that an abnormal inflammatory response plays a role in etiology and symptom expression of a chronic IBS-like syndrome is supported by observations in patients with idiopathic inflammatory bowel disease (IBD). Patients in histologically proven remission from ulcerative colitis express IBS-like symptoms with a higher than expected frequency. 67,68 Furthermore, functional studies have shown that patients in remission from ulcerative colitis and inactive Crohn s disease have changes in anorectal 69 and colonic motility. 70 However, surprisingly, IBD patients in clinical remission have not been found to demonstrate evidence for visceral hypersensitivity 71,72 as assessed by rectal or sigmoid balloon distention. These observations provide a basis for exploring a causal linkage between colitis and the development of persistent intestinal motility (and possibly afferent) abnormalities, and a putative model of postinflammatory IBS. Suitability of tissue irritation model for development of animal model. An IBS disease model based on persistent effects of transient gut irritation (infection, inflammation) fulfills 2 criteria listed above: (1) it provides a plausible explanation for persistent alterations, gut motility, and gut immune function (e.g., increased mast cells); (2) there are animal models available to test the validity of the concept. Interaction Between CNS-Directed and Gut- Directed Pathogenetic Mechanisms Every conceptual model has to take into account that events within the CNS or in the GI tract do not occur in isolation, but that both systems interact with each other both under normal conditions and particularly during perturbations of homeostasis. As shown in Figure 1B, psychosocial stressors may modulate the immune response of the gut to an infectious organism 6 as suggested by observations in PI-IBS, and gut-directed physical stressors may modulate the responsiveness of central stress circuits. 73 A series of observations support bidirectional interactions between the CNS and the immune system, 75,76 and with the enterochromaffin cells (EC) of the gut. 77,78 Mast cell degranulation in the gut occurs in response to psychological stress and can even result from Pavlovian conditioning. 80 Mast cell products released in the gut (such as proteases or histamine) have the potential of activating and/or sensitizing visceral afferent fibers. 82 In analogy to neuroimmune interactions in other parts of the immune system, 26,74 altered outputs of central stress circuits in terms of hypothalamus-pituitary-adrenal (HPA) axis and sympathetic and sympathoadrenal responses (described under comprehensive model above and summarized in Figure 1A) may have profound influences on the gut immune system, including cellularity, mast cell numbers, cytokine profiles, and Th1/Th2 balance. 26,74 Altered stress responsiveness can be associated with increased permeability of the intestine, 83 altering the access of luminal organisms and antigens to the gut immune system and thereby increasing susceptibility to inflammatory triggers in the gut lumen. 84,85 Alternatively, immune cell products including different cytokines and chemokines released in the gut may have profound influences on the responsiveness of

6 June 2002 IBS ANIMAL MODELS 2037 central stress circuits 86 including gene expression of corticotropin-releasing factor (CRF) and vasopressin. 73 Although acute peripheral inflammation may be associated with an up-regulation of central stress responsiveness, chronic inflammatory processes such as rheumatoid arthritis or ulcerative colitis may be associated with a down-regulation. 26 Immune cell products signaling to the CNS via the vagus nerve play a major role in the pathogenesis of a symptom complex including hyperalgesia, fever, anorexia, and taste aversion ( sickness syndrome ). 75,76 Animal Models of IBS Animal models can be assessed based on their face validity, i.e., if the animal model resembles the human conditions in terms of etiology, behavioral features, and underlying biology. Even though high face validity of an animal model for the entire syndrome is desirable, it is not essential. An example of an animal model with high face validity for interstitial cystitis, a condition frequently associated with IBS, develops spontaneously in the cat. 87 However, given some of the unique clinical features of IBS patients (for example, cognitive and behavioral features), it is unlikely for any animal model to reach perfect face validity. However, an animal model that adequately models specific aspects of the syndrome (target organ and/or tissue dysfunction), such as stressinduced visceral hyperalgesia, stress-induced motility changes in upper and lower GI tract, or postinfectious mucosal changes (increase in EC cells, increased intestinal permeability), would be of significant benefit, in particular if such limited models showed good predictive validity. Existing animal models can be divided into 2 broad categories based on their primary pathogenetic mechanism: those that are initiated by a CNS-directed (psychosocial) stressor and those that are initiated by a gutdirected (physical) stressor (gut inflammation, infection). Based on the bidirectional brain-gut interactions discussed earlier, the primary target may be directed at brain while secondary changes in the gut are likely to occur and vice versa. These categories can be further subdivided based on the age during which the animal has been exposed to the stressor, for example, during the perinatal and neonatal period, or during adult life. Regardless of the mode and time of initiation, it is likely that responses of both peripheral and central elements of the brain-gut axis will be altered in the fully expressed model. In addition to creating changes in the animal, an important aspect of animal models is the way responses are elicited. For example, assessment of readouts may be normal under baseline conditions, but only show abnormalities after the animal has been exposed to a particular psychological or inflammatory stressor. Also, the pattern of responses may vary depending on the nature of the stressor. For example, animals may show normal responses to interoceptive (physical) stressors (inflammation, hemorrhage), but abnormal responses to exteroceptive (psychological) stressors. Animals may show abnormal behavioral responses in one situation (open field) and abnormal autonomic responses to a different stressor (prod shock). 88 Animal Models Initiated by CNS-Directed Mechanisms ( Stress Memory Models) Several central models with relevance for IBS have been adapted from the psychiatric literature. Even though all models discussed below were developed as animal models of common affective disorders (e.g., anxiety disorders and depression), the subsequent demonstration of alterations in the brain-gut axis in some of these models focused attention on their potential usefulness as IBS models. Furthermore, in view of the high comorbidity of IBS with anxiety disorders, 21 and the reported stress-sensitivity of IBS symptom expression, 3,5,6 these models exhibit face validity for the human disease (Table 3). These models can be separated into those that are induced by environmental influences and those induced by genetic manipulations of the central stress responsiveness. Alterations in CRF System Gene Expression Induced by Environmental Manipulations: Effects of Repeated or Developmental Stressors Neonatal psychosocial stress model. The quality of the early family environment can serve as a major source of stress vulnerability in later life. Individuals who are the victims of physical, emotional, or sexual abuse are at considerably greater risk for anxiety disorders and depression in later life Other less dramatic influences in early life, such as cold and distant parent-child relationships, and divorce of the parents or death of the mother before age 10, have also been associated with such enhanced vulnerability later in life. The enhanced vulnerability is not limited to affective disorders, but extends to a greater risk of chronic illness in general, including a greater risk for development of IBS It has been suggested that this influence of early life events is mediated in part by parental influences on the development of neural systems that underlie the expression of behavioral, autonomic, and neuroendocrine responses to stress

7 2038 MAYER AND COLLINS GASTROENTEROLOGY Vol. 122, No. 7 Table 3. Animal Models of IBS PTSD 13,88 Neonatal psychological stress 12,89,90 Neonatal inflammation/pain 14 PI-IBS IBD 15,92 Key findings relevant for clinical IBS features Pathophysiology Pathogenesis Stress-induced colonic motility; visceral afferent hypersensitivity Alterations in central CRF/vasopressin, NE systems Intense stressful experience as adult Stress-induced colonic hyperalgesia, stressinduced colonic motility response; anxiogenic behavior; hypervigilance; increased intestinal permeability Alteration in central CRF, NE, and GABA A receptor systems 96,97 Aversive early life experiences Colonic hyperalgesia; altered bowel habits Persistent dorsal horn neuronal sensitization; possible alteration in dorsal column pathway Pain, inflammation in neonatal period Altered intestinal motility; colonic hyperalgesia; autonomic imbalance Altered smooth muscle contractility and inhibitory innervation; increased substance P in spinal cord; increased responsiveness of primary afferents Transient intestinal infection with nematode; Th2 immune activation; COX-2 activation in smooth muscle Altered colonic motility and hyperalgesia Persistent, altered smooth muscle contractility Transient acute inflammation; immune activation PTSD, post-traumatic stress disorder; NE, norepinephrine; GABA, -aminobutyric acid; COX, cyclooxygenase. Some of the best characterized alterations in this central adaptation to pathologic stress are an increase in CRF synthesis and secretion, 114 an increase in the activity and sensitivity of central NA systems, 96,115 and down-regulation of glucocorticoid receptors 116 suggestive of an enhanced HPA response to stress. 41,96 As a consequence of these alterations in the central stress circuitry, secondary changes are likely to occur in spinal (dorsal horn neurons) and peripheral target cells (gut immune and epithelial cells, enteric nervous system, smooth muscle cells), including permeability changes of gut epithelium 83 and blood-brain barrier 117 (see also paragraph: Interaction Between CNS-Directed and Gut-Directed Pathogenetic Mechanisms, above). Similar to the lifelong persistence of CNS changes, such peripheral changes may be of a permanent nature. Validated animal models of aversive early life events include situations in which the normal mother-infant interaction is compromised. In the rodent, one of the major maternal behaviors that determines the quality of maternal care is the characteristic licking/grooming of the pups by the mother. 97 Daily separation of the litter from their mothers for 180 minutes each day during postnatal days 4 18 will result in an alteration of maternal behavior, with significantly reduced times of the licking/grooming behavior. 96 This neonatal stress in the form of compromised mother-infant interaction results in permanent changes in the CNS, which have been documented at the level of gene expression, neurochemistry, electrophysiology, and morphology. 22,43,96,97 A key component of the model is the compromised ability to restrain the synthesis and release of CRF in response to acute psychological stressors. 43 Neurocircuits affected by the neonatal maternal separation protocol include hypothalamic and extrahypothalamic CRF systems, and ascending monoaminergic systems. 96 The increase in CRF and norepinephrine (NE) drive in maternally deprived rats is associated with a decrease in tone of the inhibitory -aminobutyric acid (GABA)/benzodiazepine system. Studies by Caldji et al. 97 have shown that these changes in the GABA A receptor are associated with altered expression of and subunits and with decreased GABA binding. These neurochemical changes are associated with enhanced fearfulness, increased HPA responsiveness to stressors, and an increased risk of developing depressionlike behaviors. Coutinho et al. 12,89,90 recently showed that maternally separated rats show evidence for alterations in stress-induced visceral and somatic pain sensitivity consistent with compromised stress-induced engagement of opioid systems. Additionally, colonic motor function in response to stress is also enhanced in these animals. Preliminary results suggest that rats exposed to neonatal stress also demonstrate evidence for increased

8 June 2002 IBS ANIMAL MODELS 2039 intestinal permeability 83 and that stress-induced visceral hyperalgesia is inhibited by central administration of a CRF 1 receptor antagonist. 12 Strengths and limitations. The demonstration of enhanced anxiety-like behaviors, hypervigilance, increased vulnerability to anhedonia under stress, alterations in the HPA axis responsiveness, stress-induced visceral hyperalgesia, and enhanced stress-induced colonic motility in adult animals give this model both face and construct validity. The model mimics the following clinical features seen in a subset of IBS patients: visceral hyperalgesia associated with normal somatic pain sensitivity, diarrhea, increased fearfulness, and stress-sensitivity. The increased prevalence of functional GI symptoms in patients with a history of early adverse life events (see above) further contributes to the similarity of animal model and the human condition. However, neither the neonatal stress model nor any of the other models showing enhanced stress responsiveness discussed in this review are specific models for IBS symptoms alone. All of them have behavioral and affective components, e.g., they are both models of GI dysfunction as well as enhanced fearfulness and stress hyperreactivity. Even though changes in several neurotransmitter systems have been characterized, little is known about the central circuits underlying stress-induced visceral hyperalgesia, in particular interactions between the CRF and opioid systems. In addition, it is currently unknown if the chronically altered stress responsiveness is associated with alterations in peripheral systems, in particular the gut-associated immune system, and visceral afferent pathways. Finally, the model may only be relevant for the subset of IBS patients with a history of adverse early life events. Post-traumatic stress disorder model. In adult man and in animals, exposure to an uncontrollable stressor can induce an alteration in the behavioral, autonomic, and neuroendocrine responses to future stressors. Depending on its intensity, either repeated or single exposure to the same stressor can produce long-lasting sensitization to a wide range of stimuli that are unrelated to the initial stressor (reviewed in Stam et al., 1997, 2000). 36,88 Diagnostic criteria based on the Diagnostic and Statistical Manual (DSM-IV) for post-traumatic stress disorder (PTSD) state that the clinical syndrome must follow experiencing or witnessing events that involve actual or threatened death or serious injury, or a threat to the physical integrity of self or others. 123 Increased prevalence of GI symptoms in PTSD has been reported in prisoners of war, Vietnam combat veterans, 124,125 and former hostages. 36 The reported increase in IBS prevalence in women with sexual abuse is highest in those who were raped at gun point or perceived life threat, meeting PTSD criteria. 118 Validated animal models use daily sessions of brief electric footshock to induce a sensitized state (reviewed in Stam et al., 2000). 88 Other interventions include social defeat by an aggressive conspecific, 126 repeated administration of psychostimulants, or repeated central injection of CRF. 127 In addition to the considerable overlap of IBS with PTSD in patients, animal models of PTSD have shown long lasting alterations in reactivity (both hypo-responsiveness and hyperresponsiveness) of the EMS, including alterations in colonic motility responses, 13 cardiovascular responses, neuroendocrine responses, and regional Fos-responsivity in the brain to stressors (reviewed in Stam et al., 2000). 88 Preliminary evidence suggests that such animals also show increased visceral afferent responses to colonic distention. 128 It is of interest, that in contrast to the autonomic and neuroendocrine sensitization observed in response to the acute stressor, behavioral responses of these animals were only seen in response to other types of stressors, such as the open field test or novel environments. Mechanisms underlying the observed alterations in stress responsiveness in both animal models and PTSD patients include enhanced responses of central CRF/vasopressin and NA responses, 129 increased spinal CRF levels, 115 altered HPA axis responses, 130 pre-attentive CNS hyperresponsiveness, 131 and altered responses of limbic brain regions, including amygdala, paraventricular nucleus, LC, and frontal cortex. 132 Strengths and limitations. Animal models of adult stress sensitization exhibit both face and construct validity. Stress-induced bowel dysfunction, visceral hyperalgesia, and development of abnormalities following a severe stressful life event (PTSD) mimic clinical features seen in a subset of IBS patients, e.g., those who develop IBS symptoms after a life-threatening event. Evidence for altered perceptual responses in the unanesthetized animal is currently lacking. The influence of vulnerability factors (Figure 2) on the expression of the adult phenotype needs to be established. Also, like all reported IBS animal models, this model may only be relevant for a subset of IBS patients. The Potential of Models With Genetically Induced Changes of the CRF System: Use of Transgenic Mice and Antisense Oligonucleotides Both behaviorally induced animal models described above share the increased responsiveness of the reciprocally related central CRF and NA systems as a

9 2040 MAYER AND COLLINS GASTROENTEROLOGY Vol. 122, No. 7 Figure 2. Putative role of psychological and physical stressors on IBS symptom development and severity. Different types of stressors play distinct pathogenetic roles in the development of chronic IBS symptoms. Vulnerability factors for the development of enhanced stress responsiveness in adult life include genetic factors, 133 and the quality of the early life environment. 97 Psychological trauma occurring early in life (in humans: first decade) can result in permanent neuroplastic changes in the EMS, 134,135 increasing the vulnerability of the individual to stressors later in life. Physical stressors in the early life period may also result in permanent changes in pain systems. Trigger factors for the first development of IBS symptoms in the predisposed individual include certain psychosocial stressors as well as physical stressors (e.g., gut infection). Perpetuating factors include, but are not limited to, symptom-related anxiety that may play a major role in chronicity of symptoms, even in the absence of other psychosocial stressors. 41 Modified from Mayer et al. 41 core pathophysiologic mechanism. In view of the well documented role of central (and possibly peripheral) CRF systems in modulation of stress-induced motility changes of the upper and lower GI tract, and the postulated role of CRF in stress-induced visceral hyperalgesia, one may speculate that not only animal models with behaviorally induced but also those with genetically induced changes in the CRF system (CRF 1 and CRF 2 receptor, CRF binding protein, CRF) (reviewed in Bakshi and Kalin, 2000) 22 might be useful models for IBS. For the same reasons, animals with genetic manipulations of the GABA A and the NA system may be useful models for increased stress responsiveness in IBS. However, key features of IBS, such as visceral hypersensitivity or stressinduced motility alterations have not been reported in any of these models. Animal Models with Primary Gut-Directed Etiologies ( Pain Memory and Immune Memory Models) The newborn gut may be exposed to a variety of factors resulting in mucosal inflammation and tissue irritation. 8 These factors include but are not limited to inflammation resulting from food allergies, viral infections, and acid reflux, all of which are common in early life. In view of underdeveloped responses of the nervous system (HPA axis, endogenous pain inhibition systems), these insults may result in permanent neuroplastic alterations manifesting as altered pain and neuroendocrine response to visceral stimuli as adults. For example, in rats, neonatal glucocorticoid treatment 136 or low dose endotoxin 137 has permanent programming effects on endocrine as well as immune functioning in adult life. Neonatal inflammation/neonatal pain. Based on the concept of neonatal sensitization, a rat model of IBS was recently published by Al-Chaer et al. 14 Daily colon irritation was produced in the neonatal period (days 8 21) either in the form of daily noxious colorectal distention (CRD) (two 60-second distentions separated by 30 minutes at rest) or in the form of daily intracolonic injection of mustard oil. Behavioral pain responses to CRD were significantly increased in the neonatally irritated rats when assessed at postnatal week 5 and persisted up to postnatal week 12, whereas colon histologies were not different from controls. Electrophysiologic recordings of individual dorsal horn neurons from segments L6-S1 showed increased responses at baseline as well as increased responses to CRD and to somatic stimuli applied to the same dermatome. Central (spinal) sensitization was further shown by increased expression of c-fos in spinal segments receiving colorectal input. 138 These findings suggest that 2 different perturbations of the intestinal homeostasis in early life (chemical and mechanical irritation) can result in permanent sensitization of dorsal horn neurons receiving convergent input. Preliminary results suggest that neurons at the thoracolumbar level of the spinal cord that normally receive little colorectal input sensitize as well, and become highly responsive to CRD. 139 The spinal sensitization can be attenuated by blockade of spinal NK 1 receptors 140 or by lesions in the dorsal column of the spinal cord rostral to the recording site. 141 These preliminary findings suggest that the dorsal column may play a role in facilitating the sensitization of dorsal horn neurons to CRD. Finally, the adult animals demonstrate evidence for altered defecation. Strengths and limitations. The model has face validity because it mimics the key features of the human condition, i.e., visceral hypersensitivity and alteration in bowel habits. It is the only model discussed here that presents with subsets of animals showing diarrhea and constipation. Even though no convincing longitudinal studies exist to show irritation of the gut in early life as a risk factor for later development of IBS, such a course of events is plausible at least in a subset of IBS (or chronic abdominal pain) patients. There is also construct validity of the model: visceral hyperalgesia secondary to long lasting or permanent sensitization of dorsal horn neurons

10 June 2002 IBS ANIMAL MODELS 2041 is a plausible mechanism that could result in enhanced pain and autonomic responses to physiologic visceral events. It remains to be determined if IBS patients exhibit somatic hyperalgesia in the same dermatome that receives afferent input from the colon: some studies have shown hypoalgesia, 142 whereas others have shown hyperalgesia. 143 Also, it is not known if IBS patients demonstrate permanently altered visceral pain perception, as would be expected from a permanent sensitization of visceral afferent pathways, or if the alteration in IBS is present only during times of symptom exacerbation. Intestinal infection in the adult. Clinical, epidemiologic, and physiologic studies have shown that acute transient infection of the gut is associated with a syndrome that, in many instances, meets diagnostic criteria for the diagnosis of IBS. 6,7,64 These patients recovering from acute enteritis exhibit changes in sensory-motor function in the colon, with evidence of increased EC cell number, increased intestinal permeability, and some evidence of immune activation. 66 Psychosocial stress and behavioral factors, and female gender are some of the risk factors associated with the development of PI-IBS. Thus, to achieve face validity by an animal model of PI-IBS, the infection must be transient, there must be evidence of neuromuscular dysfunction that persists after recovery from the infection, and, that this should be evident not only at the site of infection, but also in distal non-inflamed sites. Ideally, stress and gender should influence postinfective gut pathology. Transient infection with the nematode parasite Trichinella spiralis caused profound changes in the function of myenteric nerves, 48,144,145 smooth muscle, 146,147 and interstitial cells of Cajal. 148 Inflammation-induced changes in nerve and muscle function occur not only at inflamed sites but also at non-inflamed sites. 149,150 Inflammation results in an increase in Substance P in the gut, and in the spinal cord, and a down-regulation of NK 1 receptors in this model. 151 Furthermore, stress can alter enteric neural function in the post-inflamed gut. 152 Transient infection of the proximal small intestine with T. spiralis results in the persistence of smooth muscle hypercontractility and loss of inhibitory nerves at the site of inflammation. 91 Even though this effect on gut motility occurs in the absence of any discernible inflammatory cell infiltrate in the mucosa or myenteric plexus, the initiation of the hypercontractility state requires CD4 cells 16 and Th2 cytokines acting via Stat6 pathway. 153 The Th2-mediated initiation is required for the hypercontractility that persists postinfection in this model for days postinfection. 91 In addition, this hypercontractility appears to be actively maintained by the local production of prostaglandin E2 by the muscle cells, via cyclooxygenase 2. This is reminiscent of previous work showing exaggerated prostaglandin production by muscle in a model of colitis. 154 More recent work has also implicated transforming growth factor in maintaining this hypercontractility. 155 These findings raise the possibility that the progression to persistent dysfunction may be reversible and this is indeed the case. Treatment with a short course of steroid shortly after recovery from infection prevents the progression of the hypercontractility. 91 The in vitro changes in muscle contractility and innervation are reflected in altered intestinal motor activity recorded in vivo in mice both during and after infection; these changes are also reversible with steroid treatment postinfection. 92 Preliminary results suggest increased immunoreactive substance P in the colon and dorsal horn of the spinal cord, and an enhanced responsiveness of primary visceral afferents after balloon distention of the colon of these animals. 93 Thus, the combination of altered intestinal motility and possible altered visceral afferent nerve function attributes construct validity to the model. It is of interest to point out that the postinfective changes seen in this model are strain specific, being prominent in NIH/Swiss mice, less so in balb/c mice, and absent from B10.BR mice. 156 These observations suggest a genetic susceptibility, but the locus of susceptibility remains to be determined. As the phenomenon occurs in strong Th2 responders, it is likely to be found in immune regulatory genes. Strengths and limitations. The T. spiralis model has face validity because it mimics persistent hypercontractility of the colon after a transient gut infection. The model mimics altered bowel habits seen in a subset of patients following an enteric infection. It also has construct validity because it shows that long-term immune modulation of smooth muscle and enteric nervous system can result in persistently altered gut motility. However, the existing model of PI-IBS uses a parasitic model, whereas most clinical cases of PI-IBS follow bacterial infections with Salmonella, Campylobacter, and Shigella. Thus, models of PI-IBS based on more common pathogens need to be developed. The preliminary evidence for persistent visceral hypersensitivity needs to be corroborated. Noninfectious intestinal inflammation in the adult. Clinical observation has suggested that IBS may be more common in patients in remission from ulcerative colitis and that this is accompanied by changes in rectosigmoid physiology. Studies in animal models of colitis

11 2042 MAYER AND COLLINS GASTROENTEROLOGY Vol. 122, No. 7 have suggested that motility changes may persist and even become more prominent after recovery from experimental colitis. 157 Preliminary work in the mouse following colitis induced by intra-colonic administration of the hapten dinitrobenzene sulfonic acid supports this notion. Using a dose of 6-mg dinitrobenzene sulfonic acid in C57Bl/6 mice, Ma et al. 15 were able to show that although colonic muscle contractility was not substantially altered during the acute phase of colitis, hypercontractility emerged during the recovery phase in which lymphocytes transiently infiltrated the tissue (including the muscle layer). As the structural integrity of the tissue was restored, the magnitude of hypercontractility appeared to increase. 15 This preliminary finding requires further work, but preliminary data suggest that like the postinfective model, Th2 cytokines are critical and that colonic sensory changes also occur. Strengths and limitations. Preliminary data suggest that the model may have face and construct validity, because it shows changes of gut function outlasting a transient inflammatory event. The model mimics persistent altered bowel habits seen in a subset of patients with IBD in remission, or in patients after an acute enteric infection. However, a series of issues need to be addressed before it can be accepted as a model of IBS. For example, it is unclear if gut inflammation in adult humans or in animals is sufficient to produce long-lasting changes in motility or visceral sensory mechanisms in the absence of vulnerability factors. Furthermore, the model of IBS following active IBD needs to be extended to other animal models of IBD. Other animal models of IBS. At least 2 other models have been reported that mimic certain features of IBS. Gue et al. 158 have reported the development of stress-induced visceral hyperalgesia in normal male Wistar rats following restraint stress. Because no other intervention is required in these animals to produce enhanced stress responsiveness, this model may represent a genetic model of altered stress sensitivity. Chen et al. 159 recently reported altered bowel habits in a mouse model with targeted deletion of the serotonin reuptake transporter (SERT), and have implicated a genetic alteration in SERT as a possible model for IBS. Challenges Improvement of Face and Construct Validity To enhance the face and construct validity of existing animal models and the clinical condition, more work is required. For example, the relative contributions and the interactions between peripheral and central stressors need to be delineated. Similarly, the influence of vulnerability factors (such as those determining enhanced stress responsiveness) on the phenotype in gut-directed animal models needs to be assessed. Carefully designed studies are required to determine the presence and nature of visceral hypersensitivity, motility alterations, and autonomic dysregulation in all animal models. Ideally, assessment of these functional alterations in different models should be done with comparable experimental techniques. There is a need to capitalize on the anecdotal observations that surgical manipulation (cholecystectomy, hysterectomy) 160,161 can adversely affect the prognosis in IBS. There is sufficient evidence that manipulation alters the cellularity and physiology of the gut that these findings should be brought into an experimental paradigm more closely resembling IBS, either based on central or peripheral processes. There is a particular interest in bile acids and IBS, not only based on the relationship of cholecystectomy and IBS, but also on the demonstrated sensitivity of IBS patients to the secretory and motor effects of bile acids. Even though currently not supported by reliable scientific data, the notion that luminal nonpathogenic bacteria may play a role in IBS 162 is of considerable topical interest in light of the therapeutic potential of probiotics. The experimental controlled manipulation of gut flora is now feasible in certain centers and attempts should be made to reach an end point in gut function that resembles IBS. That is, altered sensory-motor function without overt inflammation, with or without changes in permeability or changes in behavior. On the other side, observations made in existing animal models will generate hypotheses to be tested in the human condition. For example, is there evidence for persistent central sensitization (as shown in the neonatal colon irritation model) based on hyperalgesia to both visceral and somatic stimuli in IBS? Are alterations in gut immune function present in all IBS patients or only in certain subsets, such as PI-IBS? Are the alterations in immune function observed in animal models detectable in human intestinal biopsies? Demonstration of Predictive Validity Evidence for predictive validity of existing animal models in terms of effective IBS therapies is currently lacking. 163 The demonstration of such predictive validity will be essential to accept or reject any of the conceptual and animal models, or to assign validity for a subset of IBS patients. For example, the experience with the kappa agonist fedotozine that showed high efficacy in the acute rat CRD model, but failed to show therapeutic effects on