Article Text

Original article
Sensory neuro-immune interactions differ between Irritable Bowel Syndrome subtypes
  1. Patrick A Hughes1,2,
  2. Andrea M Harrington1,
  3. Joel Castro1,
  4. Tobias Liebregts1,3,
  5. Birgit Adam1,3,
  6. Dallas J Grasby1,
  7. Nicole J Isaacs1,
  8. Lochana Maldeniya1,
  9. Chris M Martin1,
  10. Jenny Persson1,
  11. Jane M Andrews1,
  12. Gerald Holtmann1,
  13. L Ashley Blackshaw1,4,
  14. Stuart M Brierley1,5
  1. 1Nerve-Gut Research Laboratory, Discipline of Medicine, Faculty of Health Sciences, The University of Adelaide and Department of Gastroenterology and Hepatology, Royal Adelaide Hospital, Adelaide, South Australia, Australia 5000
  2. 2Department of Leukocyte, Biology, Women's and Children's Health Research Institute, Women's and Children's Hospital, Adelaide, South Australia, Australia
  3. 3Department of Bone Marrow Transplantation, University Hospital Essen, West German Cancer Centre, Germany
  4. 4Neurogastroenterology Group, Blizard Institute, Barts and the London School of Medicine & Dentistry, Queen Mary University of London, London, UK
  5. 5Discipline of Physiology, Faculty of Health Sciences, The University of Adelaide, Adelaide, South Australia, Australia
  1. Correspondence to Dr Patrick A Hughes, Nerve-Gut Research Laboratory, Hanson Institute, Adelaide, South Australia 5000, Australia;{at}


Objective The gut is a major site of contact between immune and sensory systems and evidence suggests that patients with irritable bowel syndrome (IBS) have immune dysfunction. Here we show how this dysfunction differs between major IBS subgroups and how immunocytes communicate with sensory nerves.

Design Peripheral blood mononuclear cell supernatants from 20 diarrhoea predominant IBS (D-IBS) patients, 15 constipation predominant IBS (C-IBS) patients and 36 healthy subjects were applied to mouse colonic sensory nerves and effects on mechanosensitivity assessed. Cytokine/chemokine concentration in the supernatants was assessed by proteomic analysis and correlated with abdominal symptoms, and expression of cytokine receptors evaluated in colonic dorsal root ganglia neurons. We then determined the effects of specific cytokines on colonic afferents.

Results D-IBS supernatants caused mechanical hypersensitivity of mouse colonic afferent endings, which was reduced by infliximab. C-IBS supernatants did not, but occasionally elevated basal discharge. Supernatants of healthy subjects inhibited afferent mechanosensitivity via an opioidergic mechanism. Several cytokines were elevated in IBS supernatants, and levels correlated with pain frequency and intensity in patients. Visceral afferents expressed receptors for four cytokines: IL-1β, IL-6, IL-10 and TNF-α. TNF-α most effectively caused mechanical hypersensitivity which was blocked by a transient receptor potential channel TRPA1 antagonist. IL-1β elevated basal firing, and this was lost after tetrodotoxin blockade of sodium channels.

Conclusions Distinct patterns of immune dysfunction and interaction with sensory pathways occur in different patient groups and through different intracellular pathways. Our results indicate IBS patient subgroups would benefit from selective targeting of the immune system.

  • Irritable bowel syndrome
  • pain
  • cytokine
  • sensory neurons
  • neurogastroenterology
  • gastro-oesophageal reflux disease
  • Helicobacter Pylori—epidemiology
  • functional dyspepsia
  • functional bowel disorder
  • visceral hypersensitivity
  • nervegut interactions
  • visceral nociception
  • lower oesophageal sphincter
  • real time PCR
  • 2,4,6-trinitrobenzene sulphonic acid
View Full Text

Statistics from

Significance of this study

What is already known about this subject?

  • Irritable bowel syndrome (IBS) is a chronic debilitating functional disorder of the gastrointestinal tract with an unknown cause.

  • Increasing evidence indicates IBS patients have altered immune profiles.

  • Immune profiles differ between IBS patient subgroups.

What are the new findings?

  • Peripheral blood mononuclear cell (PBMC) supernatants from healthy subjects, constipation predominant and diarrhoea predominant irritable bowel syndrome (IBS) patients differed from each other in terms of immune profile and effects on colonic afferent nerves.

  • PBMC supernatants from healthy subjects inhibited colonic afferent mechanosensitivity in an opioid-mediated manner.

  • The proinflammatory cytokines TNF-α, IL-1β and IL-6 are increased in PBMC supernatants from diarrhoea predominant IBS patients, their receptors are expressed on colonic afferent endings, they sensitise colonic afferents to mechanical stimuli and they correlate with symptoms of pain.

How might it impact on clinical practice in the foreseeable future?

  • We have shown there are major differences in the nature of interactions between the nervous and immune systems in constipation predominant and diarrhoea predominant irritable bowel syndrome patients. Therapies directed at restoring the specific alterations in the immune systems of these patient groups are likely to be of benefit.


Chronic abdominal pain has no effective treatment, but results in more general practitioner visits than headache, chest pain or high blood pressure and has a major impact on quality of life.1 Irritable Bowel Syndrome (IBS) is the most common manifestation of abdominal pain, which is diagnosed as a recurrent sense of abdominal discomfort or pain accompanied by altered bowel movements.2–4 There are no physiological markers, and so all approaches to the condition are based on bowel habit, with patients categorised as diarrhoea predominant (D-IBS), constipation predominant (C-IBS) or alternating between these states. While these classifications provide options for the management of bowel habit, it is important to note that pain crosses all subgroups, is the most debilitating symptom and the most intractable to treatment.1 ,5

Currently the best insight into the aetiology of IBS is the increased risk of development following gastrointestinal infection, implicating the immune system in the initiation of this disease.6 Termed postinfectious IBS (PI-IBS), the clinical characteristics of these patients correlate robustly with D-IBS.6 ,7 Thorough investigation of biopsies from PI-IBS patients consistently reveal a chronic but low grade immune activation, with increased numbers of resident immune cells including mast cells and T cells.7–10 These findings stimulated investigation of the immune status of non-PI-IBS patients, and numerous reports now indicate alterations in resident immune cell population and function occur in the IBS population as a whole.11–15 More recent investigations have extended to investigate the circulating peripheral blood mononuclear cell (PBMC) population. PBMCs from IBS patients have increased expression of the gut homing integrin β7, indicating increased homing of circulating lymphocytes to the gastrointestinal tract.16 ,17 However, the immune phenotype of these cells is controversial, with reports of shifts in the Th1/Th2 axis towards either Th1 predominance, suggestive of a type 1 inflammatory response, or Th2 predominance, suggestive of a type 2 allergic response, and several reports failing to observe any change.16 ,18–25 There is clearly a need for studies of accurately categorised cohorts of IBS patients and a systematic approach to characterising their cytokine status.

Pain and discomfort from the intestine are signalled by C-fibres of the lumbar splanchnic and sacral pelvic nerves, the soma of which are contained within the dorsal root ganglia (DRG). The splanchnic pathway predominantly consists of afferents with high thresholds to mechanical stimuli, including serosal afferents which only respond to high intensity circular stretch and focal compression, suggesting a specialised nociceptive pathway.26 ,27 The pelvic pathway constitutes a mix of both high threshold serosal afferents and afferents with a low/medium mechanical threshold, with mucosal afferents responding only to fine mucosal stroking, muscular afferents responding only to low intensity circular stretch and muscular/mucosal afferents responding to both.26 ,27 This diversity of afferent subtypes signals the full range of painful and physiological stimuli including distension, urge and contractile events.

Colonic afferent function is actively modulated by inflammation and increased peripheral levels of cytokines correlate with symptoms of altered motility in D-IBS patients, such as diarrhoea and urgency.21 ,27 ,28 Several studies have shown that supernatants of colonic mucosal biopsies from IBS patients contain elevated concentrations of mast cell mediators including histamine and serine proteases, typically associated with allergic type responses.12 ,29–31 Mast cell derived mediators released from these biopsies sensitise enteric and gastrointestinal extrinsic sensory afferents to mechanical stimuli, indicating that inflammatory mediators associated with allergic type responses are potentially important in IBS.12 ,29–31 We have previously demonstrated in a preliminary communication that immune mediators from D-IBS patients sensitise colonic afferents to mechanical stimuli.32 Here we aimed to extend these findings by characterising the circulating immune profile in subclasses of IBS patients and how altered immune function specifically interacts with sensory pathways to give rise to symptoms of pain. We also characterised the major molecular mechanisms underlying visceral sensory neuro-immune interactions thereby identifying potential targets for the treatment of this elusive disease.

Materials and methods

For comprehensive details please also see supplementary material.

All experiments were approved by the Human Ethics Committee's of the Royal Adelaide Hospital and University of Adelaide, and the Animal Ethics Committees of the Institute of Veterinary Science and University of Adelaide.

Human subjects

In all, 35 IBS patients (21 women, 14 men; median age 46.3 years) were recruited consecutively from the Department of Gastroenterology, Hepatology and General Medicine at the Royal Adelaide Hospital and from 36 age matched healthy subjects (HS) (23 women, 13 men; median age 38.9 years) were recruited by advertisement. Written informed consent was obtained prior to inclusion. All patients had chronic or relapsing symptoms of IBS consistent with ROME II criteria present for at least 3 years.33 Patients were categorised according to bowel habits (Supplementary methods) as either C-IBS (15 patients: nine women, six men; median age 38.9 years) or D-IBS (20 patients: 12 women, eight men; mean age 46.3 years). Five of the D-IBS patients were considered PI-IBS with confirmed previous infection that had cleared by at least 6 months prior to inclusion in our study (Supplementary methods). Abdominal symptoms were assessed using a valid self-report Bowel Disease Questionnaire before study participation.21 The symptoms of pain intensity and pain frequency were considered key symptoms and were preselected before data analysis.

PBMC isolation and culture

PBMCs were isolated from fresh blood by density gradient centrifugation, resuspended to 1×106 cells/ml in complete medium and cultured overnight (Supplementary methods).20 ,21 Supernatants were collected and pooled according to D-IBS, C-IBS or HS status (Supplementary methods), aliquoted and stored at −80°C until use.

Cytokine measurement

Cytokine concentrations were determined using ELISA plates against tumour necrosis factor (TNF)-α, IL-1β, IL-6 (eBioscience, San Diego, California, USA) or a human 25-plex multiplex kit according to manufacturer's instructions (Luminex, Invitrogen, Mulgrave, Victoria, Australia) run on a Bio-Plex-200 Array system (Bio-Rad, Gladesville, New South Wales, Australia) (Supplementary methods).20 ,21

Functional studies

Colonic single unit extracellular electrophysiological recordings from the pelvic or splanchnic nerve were performed essentially as previously described using 10–16-week male C57/BL6 mice (Supplementary methods).26 ,27 ,34 ,35 We have previously categorised these afferents as C-fibres based on conduction velocity.34 ,35 Following classification of afferent subtype, the receptive field was subsequently incubated in 100 μl of supernatant or agonist, which was applied via a small metal ring surrounding the receptive field, for 5 min and mechanosensitivity retested. Some supernatants were incubated with infliximab (1.5 mM; Remicade, Schering Plough, North Ryde, New South Wales, Australia) for 2 h at room temperature prior to application. Some cytokines and supernatants were added in the presence of opioid or ion channel antagonists that were first added for 10 min alone and mechanosensitivity reassessed prior to addition of cytokine/supernatant in the presence of antagonist. Direct responses were included if the magnitude of response was greater than twice basal firing.


All reagents were purchased from Sigma-Aldrich (Castle Hill, New South Wales, Australia) unless otherwise stated: TNF-α (3.9 nM), IL-1β (0.6 nM), IL-6 (4.1 nM), IL-10 (4.8 nM), (D-Ala2, N-MePhe4, Gly-ol) -enkephalin (DAMGO, TOCRIS, Bristol, UK) (1 μM), Naloxone (10 μM), D-Phe-Cys-Tyr-D-Trp-Orn-Thr-Pen-Thr-NH2 (CTOP, TOCRIS) (10 μM), capsazepine (5 μM), HC-030031 (Hydra Biosciences, Cambridge, Massachusetts, USA) (10 μM) or tetrodotoxin (TTX) (Alomone, Jerusalem, Israel) (1 μM).


Immunoreactivity for β-endorphin (Thermo Scientific, Waltham, Massachusetts, USA), PGP9.5 (UltraClone, Isle of Wight, UK) and CD3 (Abcam, Cambridge, UK) was assessed in 15 μm frozen cross-sections of formalin fixed mucosal biopsies from HS or IBS patients that underwent antigen retrieval (DAKO, Glostrup, Denmark) (Supplementary methods). μ-Opioid receptor (MOR) immunoreactivity (Chemicon, Millipore, Billerica, Massachusetts, USA) was assessed in 12 μm cryosections of DRG and distal colon of male C57/BL6 mice essentially as previously described.34 ,36

Quantitative RT-PCR

mRNA was isolated from DRG or laser captured microdissected (PALM, Carl Zeiss, North Ryde, New South Wales, Australia) retrogradely labelled (Alexa Fluor 555, Invitrogen) colonic DRG neurons essentially as previously described (Supplementary methods).36 ,37 Quantitative real time PCR (RT-PCR) was performed for the following targets: β-tubulin, TNFR1, TNFR2, IL-1R1, IL-6R, CD130, IL-10R, CCR5 and IFNAR1 (Supplementary table 1). Confirmation of amplified products was resolved by gel electrophoresis. The comparative cycle threshold method was used to quantify the abundance of targets relative to the housekeeper β-tubulin. Experiments were performed at least in triplicate and repeated twice.

Statistical analysis

In all cases, results are expressed as Mean ± SEM, N=number of subjects and n=number of observations. The relationship between the cytokine levels and the intensity of symptoms was assessed by Spearman rank correlations. χ2 Tests determined the significance of differences between proportions of afferents directly responding to PBMC supernatants or cytokines. One-way analysis of variance with Tukey's post hoc test determined the significance of changes in magnitude of direct effects between supernatants or cytokines, the differences between cytokine concentrations in HS, C-IBS and D-IBS supernatants, and numbers of β-endorphin containing lamina propria cells in colonic biopsies. Paired and unpaired Student t tests or one-way analysis of variance with repeated measures determined the significance of changes in mechanosensitivity. In all cases, *p<0.05, **p<0.01 and ***p<0001.


Effects of immune products from IBS subclasses on afferent function

We initially investigated the effects of PBMC supernatants from HS, C-IBS patients and D-IBS patients on colonic afferent function by performing single unit extracellular electrophysiological recordings of responses from intact pelvic or splanchnic afferent fibres. Using this method of recording, we were able to discriminate afferent classes based on their threshold to mechanical stimuli as either low/medium-(incorporating mucosal, muscular and muscular/mucosal) or high threshold (serosal afferents), and distinguish between direct effects of supernatants on afferents, whereby action potentials are elicited in the presence of supernatant alone, and any effects on their mechanosensory function.

Incubation of colonic afferents with PBMC supernatants significantly influenced the sensitivity of afferents to mechanical stimuli. We have previously described that D-IBS patient supernatants sensitise pelvic afferents to mechanical stimulation,32 and now show this is also the case for splanchnic serosal afferents. Furthermore, we show this sensitisation is dose-dependent, while supernatants from C-IBS patients had no effect on mechanosensitivity of any afferent class (figure 1A, Supplementary figure 1A–E). In contrast, we have previously observed that supernatants from HS inhibited muscular/mucosal afferent mechanosensitivity.32

Figure 1

Effects of peripheral blood mononuclear cell (PBMC) supernatants from irritable bowel syndrome (IBS) patients on colonic afferent function. (A) PBMC supernatants from diarrhoea predominant IBS (D-IBS) patients (N=5, n=10) caused hypersensitive responses in splanchnic serosal afferents, with healthy subjects (HS) (N=7, n=10) and constipation predominant IBS (C-IBS) (N=10, n=16) PBMC supernatants having no effect on mechanosensitivity. (B) High threshold afferents responded directly to C-IBS supernatants but not to supernatants from either D-IBS or HS, which only rarely caused direct firing in low/medium threshold afferents. A representative example of direct firing caused by application of C-IBS supernatants is shown with arrows indicating addition and removal of supernatant. Results expressed as percentage of afferents responding and total number of action potentials per 5 min incubation. All data expressed as Mean±SEM. *p<0.05.

Approximately 25% of high threshold fibres directly responded during incubation with supernatants from C-IBS patients, but they were never observed in high threshold fibres following incubation with either HS or D-IBS supernatants (figure 1B). These direct responses were reproducible when tested with a second application of supernatant after a 5 min washout (61±2 spikes/5 min at first dose vs 60±4 during repeat (mean ± SEM. n=2, N=2)) (Supplementary figure 1F).

Immune cells from HS secrete β-endorphin

We further investigated the mechanism of inhibition by HS supernatants, and found that it was blocked by prior incubation of colonic afferents with naloxone, a non-selective opioid antagonist (data not shown), indicating an underlying opioidergic mechanism. CTOP, a MOR selective antagonist, abolished the inhibitory effects induced by HS supernatants (figure 2A), and they were mimicked with the MOR selective agonist DAMGO (Supplementary figure 2A). In addition to these effects of exogenously applied HS supernatants, we localised β-endorphin immunoreactivity to a subpopulation of CD3+ve immune cells in the mucosa of the human colon (figure 2B). These cells were in close apposition to PGP9.5 labelled nerve fibres (figure 2C). In addition, we localised MOR immunoreactivity in mouse DRG neurons and colon wall (Supplementary figure 2B). The number of β-endorphin positive colonic mucosal lamina propria cells was reduced in C-IBS patients compared with HS, but was not changed in D-IBS patients (figure 2D). Our results demonstrate that healthy human immune cells actively secrete β-endorphin which acts to dampen viscero-sensory mechanosensation. This inhibitory effect from PBMC supernatants is lost in C-IBS patients, and switches to sensitisation in D-IBS patients. Candidates for the sensitising effects are investigated below.

Figure 2

Peripheral blood mononuclear cells (PBMCs) from healthy subjects (HS) inhibit mechanosensitivity via β-endorphin. (A) The inhibitory effect of PBMC supernatants from HS (N=5, n=10) on the stretch and stroke components of muscular/mucosal afferents is blocked by CTOP (10 μM) (N=4, n=6). (B) Immunofluorescence staining of human colonic tissue sections stained with antibodies against CD3 (arrows) and β-endorphin (arrows) and (C) antibodies against PGP9.5 (arrow) and β-endorphin (arrow). Merging of images shows β-endorphin expression in a subpopulation of CD3+ve cells (arrows) and β-endorphin containing cells are in close apposition to nerve fibres (arrow). Scale bar in both panels is 25 μm. (D) Mucosal biopsies from constipation predominant irritable bowel syndrome patients contained less β-endorphin containing lamina propria cells than HS or diarrhoea predominant irritable bowel syndrome patients. Results expressed as Mean ± SEM. *p<0.05. Access the article online to view this figure in colour.

Immune profiles of IBS patient subclasses differ

We next determined, using multiplex array technology, if changes in immune cell profile could explain the alterations in colonic function described above using multiplex array technology. We found the immune profiles of D-IBS and C-IBS PBMC supernatants differed significantly from each other and from HS (figure 3A, Supplementary figure 3). Compared with HS, D-IBS patients had substantially elevated concentrations of IL-1β, IL-10, TNF-α, IL-6, the soluble antibody for IL-2 (IL-2RA) and the chemokines CCL3 and CCL4, while IFN-α concentrations were raised, but not significantly (p=0.067) (Supplementary figure 3). The cytokine profile of C-IBS patients more closely resembled that of HS with the exception of IL-1β which was increased above HS levels but not to the same extent as observed in D-IBS patients, and decreased concentrations of CCL5 and interestingly IL-2RA, the opposite of that in D-IBS (Supplementary figure 3). D-IBS supernatants differed significantly from C-IBS supernatants, with substantial increases in the cytokines IL-1β, IL-10, TNF-α, IL-6, CCL3, CCL4, CCL5, IL-2RA and IL-12 (figure 3A). Importantly, the high levels of TNF-α correlated significantly with self reported symptoms of pain frequency (figure 3B) and pain intensity (figure 3C) in D-IBS patients, but not in C-IBS patients. Similar correlations were observed for IL-1β (Supplementary figure 4A,B) and IL-6 (Supplementary figure 4C,D). No difference was observed between pain frequency and intensity scores in D-IBS and C-IBS patients.

Figure 3

Diarrhoea predominant irritable bowel syndrome (D-IBS) and constipation predominant IBS (C-IBS) peripheral blood mononuclear cell supernatants differ substantially from healthy subjects (HS) and each other, and correlate with symptoms of IBS. (A) The immune profile of D-IBS patients is significantly altered compared with C-IBS and HS. All data expressed as Mean ± SEM. *p<0.05, **p<0.01, ***p<0.001. (B, C) Spearman rank correlation of association of TNF-α levels and self-reported frequency of pain (B) and pain intensity (C). The dotted line represents the 95% CI. (B) TNF-α levels correlate with frequency of pain in D-IBS patients (r=0.65, p=0.012), but not in C-IBS patients (r=0.34, p=0.367). (C) TNF-α levels correlate with pain intensity in D-IBS patients (r=0.62, p=0.024) but not in C-IBS patients (r=−0.09, p=0.825). Access the article online to view this figure in colour.

Functional effects of cytokines on colonic afferents

Little is currently known of the effects of individual cytokines on afferent function, particularly that of the colon, and while the multiplex results provided several potential targets we first needed to know whether colonic afferents express receptors for these cytokines. We found L6-S1 DRG expressed mRNA for TNFR1 and TNFR2 (receptors for TNF-α), IL-1R1 (IL-1β), IL-6R (IL-6), CD130 (IL-6 coreceptor), IL-10R (IL-10), IFNAR-1 (IFN-α) and the chemokine receptor CCR5 (receptor for both CCL3 and CCL4) (figure 4A). However, DRG contains a resident immune cell population which could potentially express these receptors, while only a small proportion (approximately 5%) of DRG neurons innervates the colon.36 ,37 Therefore, it was important to determine expression specifically in colon-innervating DRG neurons. The profile of this population differed from whole DRG, with mRNA for IFNAR-1, CCR5 and TNFR2 absent from colonic neurons, while TNFR1, IL-1R1, IL-6R, IL-10R and CD130 were each present (figure 4A). Quantitative RT-PCR analysis of colonic DRG neurons indicated mRNA expression of the IL-6R cosignalling molecule CD130 is much more abundant than the other cytokine receptors, followed by TNFR1 and IL-1R1 which are both in greater abundance than IL-10R and IL-6R (figure 4A).

Figure 4

Expression and functional effects of cytokines on colonic afferents. (A) Gel electrophoresis of quantitative real time (RT)-PCR products indicate colonic dorsal root ganglia neurons express mRNA for TNFR1, IL-1R1, IL-10R, IL-6R and CD130, but do not express IFNAR-1, CCR5 or TNFR2, each of which is expressed in whole dorsal root ganglia. Quantitative RT-PCR analysis indicated colonic neurons expressed CD130 mRNA in the greatest abundance followed by TNFR1 and IL-1R1, each of which are expressed in greater abundance than either IL-10R or IL-6R (N=4). (B) More afferents responded directly to IL-1β (0.6 nM) compared with IL-6 (4.1 nM) or IL-10 (4.8 nM) while none responded directly to TNF-α (3.9 nM). A representative example of direct firing caused by IL-1β is shown with arrows indicating addition and removal of IL-1β. Results expressed as percentage of afferents responding to supernatant and of total number of action potentials observed during 5 min supernatant incubation. (C) TNF-α (N=8, n=10), IL-1β (N=10, n=18) and IL-6 (N=5, n=6) individually caused mechanical hypersensitivity of pelvic serosal afferents, whereas IL-10 (N=7, n=8) had no effect. Overall, TNF-α caused the greatest degree of mechanical hypersensitivity. All data expressed as Mean ± SEM. *p<0.05, **p<0.01, ***p<0.001.

Combining the findings of altered cytokine expression in human PBMCs with the specific expression of cytokine receptors on colonic nerves, we identified four targets from D-IBS that may contribute to generation of pain signals: TNF-α, IL-1β, IL-6 and IL-10, and so we tested the effects of each of these individually on colonic afferent function. We used similar concentrations of cytokines, at doses relatively higher than that observed in supernatants in order to assess the relative potency of each cytokine, and also to relate results to those previously published using similar doses.38–44 Reported below are responses for high-mechanical threshold colonic afferents, as these afferents sense noxious stimuli in the colon wall. Smaller numbers of experiments were also performed in low/medium threshold afferents which displayed similar results (data not shown). Afferents responded with direct excitation to IL-1β, IL-6 and IL-10, with IL-1β affecting the largest population (figure 4B). The magnitude of response was similar between afferents. By contrast, direct responses were not observed during TNF-α incubation. Similar to the effects of supernatants, individual cytokines were capable of influencing mechanosensitivity. TNF-α caused the greatest increase in mechanosensitivity, while more modest, but significant, increases were also observed following incubation with either IL-1β or IL-6. By contrast, IL-10 did not influence mechanical sensitivity (figure 4C). In addition to its effect on pelvic afferents, TNF-α also caused mechanical hypersensitivity of high threshold splanchnic colonic afferents, while IL-1β, IL-6 and IL-10 did not (Supplementary figure 5).

Since TNF-α had the most pronounced effects on mechanosensitivity, with the greatest magnitude and effects on both pelvic and splanchnic afferent pathways, removing it from the PBMC supernatants of D-IBS patients should alter its ability to influence mechanical hypersensitivity. Prior incubation of D-IBS supernatants with infliximab, a monoclonal antibody against TNF-α, reduced the free concentrations of TNF-α to negligible levels (Supplementary figure 6). Addition of the D-IBS/infliximab cocktail to colonic afferents resulted in a significant decrease in the ability of the D-IBS supernatants to cause mechanical hypersensitivity (figure 5). Interestingly, a residual sensitising effect remained which, while modest, was significantly greater than baseline measurements, indicating potential effects of other endogenous mediators such as IL-6 and/or IL-1β on mechanical sensitisation (figure 5).

Figure 5

TNF-α is the primary mediator of colonic afferent mechanical hypersensitivity caused by immune products from diarrhoea predominant irritable bowel syndrome (D-IBS) patients. Incubation of D-IBS peripheral blood mononuclear cell supernatants with infliximab (1.5 mM) substantially reduced their ability to induce colonic afferent mechanical sensitisation (N=4, n=6) compared with D-IBS supernatants alone (N=5, n=6); however, a residual hypersensitive effect remains. All data expressed as Mean ± SEM. *p<0.05, **p<0.01.

Mechanisms of action of cytokines

Current evidence suggests TNF-α sensitises sensory ganglion neurons via interactions between TNF receptors and the pain sensing transient receptor potential V1 (TRPV1) ion channel, sodium channels and/or potassium channels.38 ,40 ,45 We confirmed the findings of others that TNFR2 is absent from neurons, indicating in colonic neurons TNF-α acts via TNFR1 (figure 4A).38 ,46 The TRPV1 inhibitor capsazepine has previously been shown to inhibit pelvic colonic sensory afferents; however, the dose required was high (500 μM) potentially suggesting off-target effects.47 We found selective antagonism of TRPV1 with 5 μM capsazepine did not affect baseline mechanosensitivity of high threshold colonic afferents or the sensitising effects of TNF-α (figure 6A). TRP Ankyrin 1 (TRPA1) is a distinct member of the TRP family, and contributes towards gastrointestinal afferent mechanosensation.34 ,48 We have previously shown selective antagonism of TRPA1 with 10 μM HC-030031 significantly inhibits the mechanosensitivity of colorectal afferents,48 and now extend this to show it also blocks the sensitising effects of TNF-α (figure 6B) and D-IBS supernatant (Supplementary figure 7), indicating in colonic afferent endings TNF-α signals via interactions between TNFR1 and TRPA1 but not TRPV1.

Figure 6

Differential mechanisms of action of cytokines on colonic afferent function. (A) Blocking TRPV1 with capsazepine (czp) (5 μM) had no effect on colonic afferent baseline mechanosensitivity or TNF-α induced hypersensitivity (N=4, n=8). In contrast (B), blocking TRPA1 with HC-030031 (10 μM) substantially reduced colonic afferent mechanosensitivity and abolished the sensitising effect of TNF-α (3.9 nM) (N=4, n=7). Traces of mechanical responses are shown in the presence of (A) capsazepine or capsazepine and TNF-α or (B) HC-030031 or HC-030031 and TNF-α. Arrows indicate mechanical stimuli on/off. (C) IL-1β induced mechanical sensitisation of colonic afferents is independent of direct activation as the magnitude of sensitisation is similar in afferents directly responding to IL-1β (0.6 nM) (responders; N=4, n=6) as those that did not (non-responders; N=6, n=12). (D) IL-1β caused direct firing of colonic afferents via a NaV1.7 mechanism and not TRPA1. Blockade of NaV1.7 with tetrodotoxin (TTX) (1 μM) reduced the magnitude of direct responses by colonic afferents to IL-1β, while HC-030031 had no effect. A representative trace of direct response to IL-1β and loss of this response in the presence of TTX is shown. Arrows indicate drug on/off. All data expressed as Mean ± SEM. *p<0.05, ***p<0.001.

The ability of cytokines and PBMC supernatants to cause either direct firing of action potentials, mechanical hypersensitivity or both suggests these effects are mediated by independent mechanisms. We were able to test this hypothesis as IL-1β caused both direct firing and mechanical hypersensitivity. IL-1β caused mechanical hypersensitivity of a similar magnitude irrespective of whether or not the afferent responded in a direct manner (figure 6C). We used TTX to block sodium channels with a concentration (1 μM) that predominantly affects the TTX-sensitive sodium channel NaV1.7 as inhibition of TTX-resistant NaV channels requires greatly increased doses of TTX. This significantly decreased the magnitude of the direct effects of IL-1β, whereas the response was unchanged after inhibition of TRPA1 (HC-030031) (figure 6D). Incubation with 1 μM TTX did not affect afferent baseline mechanosensitivity or the ability of TNF-α to mechanically sensitise colonic afferents (Supplementary figure 8). This indicates the reduction in action potential firing (direct effect) was not due to inhibition of action potential propagation at this concentration of TTX implying a selective coupling of IL-1R1 to TTX-sensitive NaV channels.


This study demonstrates the immune phenotype of IBS patients differs markedly between subgroups. We showed that immune cell products potently influence the behaviour of nociceptive and non-nociceptive afferents innervating the colon. In this and a previous publication32 we demonstrate three major mechanisms by which immune supernatants influence C-fibre colonic afferent function: potentiation of mechanosensory function in D-IBS and inhibition of mechanosensory function in HS as shown previously,32 plus direct excitation in C-IBS. Here we determined that these effects are reproducible, which individual cytokines are likely to underlie them and the mechanism by which their receptors are coupled to afferent excitability.

This study shows that PBMC supernatants from D-IBS patients had substantial increases in concentrations of a number of cytokines and chemokines including IL-1β, TNF-α, IL-6, IL-10, CCL3, CCL4, CCL5, IL-12 and IL-2RA relative to C-IBS and HS. Critically, we found that TNF-α, IL-1β and IL-6 individually sensitised colonic afferents to mechanical stimuli, via activation of their respective receptors expressed on colonic afferents, while their increased levels correlated directly with self reported symptoms of pain in D-IBS patients. We showed that the predominant effects of PBMC supernatants from D-IBS patients are due to TNF-α, because removing it substantially reduced the ability of supernatants to sensitise afferents to mechanical stimuli. The residual mechanical sensitisation may be due to IL-1β or IL-6 acting independently or in synergy, or potentially other inflammatory mediators not investigated here. TNF-α has previously been shown to sensitise afferent endings via an interaction with TRPV1 in somatosensory afferents; however, it is suggested that the functional outcome of this coupling is sensitisation to heat and not to mechanical stimuli.40 This interaction is yet to be investigated in the gastrointestinal tract. We show in colonic afferents that inhibition of TRPV1 with capsazepine at a dose of 5 μM had no effect on TNF-α induced mechanical hypersensitivity, and instead TNF-α sensitises pelvic serosal colonic afferents via an interaction between TNFR1 and TRPA1.

Nearly 25% of high threshold afferents responded directly to C-IBS supernatants, firing action potentials during incubation. This effect was not observed with supernatants from HS or D-IBS patients. Direct responses of this nature signal towards the CNS in the absence of mechanical stimuli, and as such are likely to result in acute mechanical-independent pain responses. The low proportion of afferents responding suggests a distinct subpopulation are sensitised. Afferents responded directly most often to IL-1β, and our results suggest these effects occur independently of mechanical sensitisation as the sensitising effects of IL-1β do not differ between afferents that directly respond and those that do not. We discovered a role for TTX-sensitive ion channels, most likely NaV1.7, in the direct effects of IL-1β, as blockade of these channels inhibited the magnitude of the direct responses elicited. However, 1 μM TTX did not affect colonic afferent baseline mechanosensitivity or the ability of TNF-α to sensitise afferents to mechanical stimuli. Furthermore, inhibition of TRPA1 had no effect on the magnitude of direct responses to IL-1β. It is unlikely the direct effects observed in C-IBS are mediated entirely by IL-1β, as its concentration is higher in D-IBS supernatants where direct effects are only rarely observed, raising the possibility that they are due to inflammatory mediators other than those investigated in the current study. Alternatively, they may occur due to other mechanisms, such as the selective loss of β-endorphin induced inhibition via MOR in C-IBS (discussed below), and therefore a removal of downstream inhibition by G-protein coupled receptors.49 Loss of this inhibitory effect may result in afferents becoming more likely to respond directly to inflammatory mediators. Importantly, our results show different ion channels are involved in either mechanical sensitisation or direct excitation of colonic afferents and that different cytokine receptor–ion channel coupling can occur in the same ending.

We showed PBMC supernatants from HS inhibited colonic afferent mechanosensitivity via an opioidergic effect mediated by β-endorphin. Animal studies have previously shown immune cells contain the machinery required for synthesis and release of opioids including β-endorphin, and immune cell secreted β-endorphin modulates behavioural responses to gastrointestinal stimuli.50 ,51 We found a correlate of this phenomenon in humans, with β-endorphin containing immune cells in close proximity to nerve fibres in the healthy human colon. Loss of this inhibitory effect may contribute towards the sense of discomfort/pain observed in C-IBS patients where we observed decreased numbers of β-endorphin positive cells in the lamina propria. Interestingly, such changes in β-endorphin immunoreactivity were not observed in biopsies from D-IBS patients, suggesting the inhibitory effects are still present in supernatants from these patients. However, supernatants from D-IBS patients sensitised colonic afferents to mechanical stimuli, indicating the increase in excitatory mediators secreted by these patients over-ride any β-endorphin-mediated inhibition of mechanosensitivity.

It has previously been suggested IBS patients have a low grade immune activation with a type I autoimmune phenotype, with increased proinflammatory, or Th1, cytokines, and/or decreased anti-inflammatory, or Th2 cytokines.16 ,18–25 Our results confirm previous findings that the PBMCs of D-IBS patients have increased concentrations of both Th1 and Th2 cytokines.20 ,21 The notion of immune activation in IBS is controversial, and clinical trials of probiotic or anti-inflammatory therapies are equivocal. IBS patients treated with the probiotic Bifidobacterium infantis 35624 show improvement of symptoms, but also a normalisation of cytokine ratios.22 However, the enthusiasm generated by these results is tempered by the lack of benefit shown by treatment of Campylobacter infected PI-IBS patients with the steroid Prednisolone, or in unselected IBS patients with the mast cell stabiliser Ketotifen, despite much promise from preclinical studies.11 ,12 ,29–31 ,52 ,53 Interestingly, the studies investigating mast cell function indicate that while mediator levels are increased in IBS patients, there is no difference between D-IBS and C-IBS. This may suggest mast cell derived mediators are equally important in signalling pain in both patient subtypes but are not involved in mediating motility related events. However, the Prednisolone, Ketotifen or probiotic studies outlined above were not powered appropriately to discriminate differences between IBS patients based on bowel habit. Our study differs in that we have shown distinct differences are apparent in the circulating immune profile of D-IBS and C-IBS patients, and their resultant effects on colonic sensory afferent function. This again highlights the confounding nature of grouping all IBS patients together when investigating circulating cytokine levels. One limitation of the current approach was the use of pooled supernatants, and while onerous it would be of interest to correlate the effects on colonic afferent nerves with PBMC supernatants from individual patients. Overall, these findings with mast cells and PBMCs suggest that combinations of factors may underlie pain signalling in IBS and apparent differences between IBS subgroups.

In conclusion, immune activation in IBS has been controversial, with numerous contrasting reports. We demonstrate clear differences in the immune phenotype and resultant modulation of colonic sensory function between D-IBS and C-IBS patients and our results show it is critical to stratify IBS patients according to bowel habit. Furthermore, our results indicate immune derived mediators such as TNF-α are increased specifically in D-IBS patients where they correlate with symptoms of pain intensity and pain frequency and sensitise colonic afferents to mechanical stimuli.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Correction notice This article has been corrected since it was published Online First. The author name Ashley L Blackshaw has been amended to read L Ashley Blackshaw. The author affiliations have also been amended.

  • Funding This work was supported by NHMRC Australia project grant # 626960, NHMRC Principle Research Fellowship (LAB) and NHMRC Australian Biomedical Fellowships (PAH and SMB).

  • Competing interests None.

  • Ethics approval Ethics approval was provided by the Royal Adelaide Hospital and University of Adelaide.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement This manuscript contains all data from this study.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles