The irritable bowel syndrome (IBS) is a common disorder that may affect as many as 9%-15% of the population in Western countries [1–3]. IBS is characterized by abdominal pain and disturbed bowel function in the absence of a detectable organic disease, which may explain the symptoms [4]. The presence of disturbed gut function in IBS may indicate an underlying pathology in control systems or effectors of the gut. We have previously reported lymphocytic infiltration and neuron damage in myenteric ganglia when full-thickness biopsies from the jejunum in 10 patients with severe IBS were investigated [5], and there are many reports that have highlighted signs of an activated immune system as a putative pathogenetic mechanism in IBS (for review see de Giorgio [6]). However, the aetiology of observed immune activation remains unsolved.

Studies have repeatedly indicated that IBS can arise after an acute gastroenteritis. The underlying mechanism of post-infectious IBS has not been established but ongoing inflammation appears to play a role, with an increase in enteroendocrine cells, lymphocytes, mast cells, and proinflammatory cytokines (for review, see Spiller[7]). However, in a recent study of ours the actual agent causing gastroenteritis was not a predictor of risk for IBS [8]. Consequently, a host factor, such as a pre-existing chronic infection with a different microbe than the agent causing gastroenteritis, might explain the development of IBS. We presumed that a candidate agent should be compatible with an asymptomatic carrier-ship, have a preference for female gender, and have the ability to become persistent and to live in bowel epithelium. There are some observations to support the idea that a persistent infection with Chlamydia trachomatis might constitute such a host factor. Trachoma-related blindness is 2-4 times more likely to affect females compared to males [9]. It is known that IBS occurs in 35%-50% of females with chronic pelvic pain syndrome, which is believed to often be caused by chronic infection with C. trachomatis [10–12]. A previous attempt to link C. trachomatis to IBS using serum IgG antibodies failed [13], but IgG antibody patterns may be insufficient to rule out persistence of Chlamydia due a dominating cellular immune response to infection [14, 15]. Since we had previously found inflammation in mucosa and enteric ganglia of the jejunum in patients with IBS we decided to reanalyze archived biopsy material to find out if Chlamydia antigens are present in the small bowel in this group of patients.

Immunofluorescence (IF) staining showed that 53/60 patients were positive for Chlamydia LPS in full thickness jejunum biopsies. Positive staining occurred in a few cells both within the epithelium and in lamina propria (Figure 1). No Chlamydia LPS-positive cells were found in the submucosa and muscularis propria or the enteric nervous system. Since we had full thickness biopsies from only 10 controls, we investigated if the finding of LPS-positive cells could be reproduced in biopsies from small bowel mucosa. We investigated mucosal biopsies from 24 of our patients with IBS and 32 controls. Staining of slides from these biopsies yielded similar rates of positive stainings: 21/24 (88%) patients were positive for Chlamydia LPS again in a few cells within the epithelium and in l. propria. In 19 patients biopsies had been taken with a time interval of more than 1 year. Chlamydia LPS was present in biopsies with a median time difference of 5.2 (range 1-11) years. Overall, 58/65 (89%) patients were classified as Chlamydia LPS positive. Six women and one man with IBS were negative in all biopsies.

In 90% of positive biopsies from patients Chlamydia LPS was localised to mucosal cells at the level of the crypts. Double staining with antibodies to chromogranin-A showed that Chlamydia LPS was present in enteroendocrine cells within the epithelium (Figure 1). Chlamydia LPS was found in l. propria cells in 69% of the biopsies from patients. Double staining with antibodies to CD117, CD11c and CD68 revealed that in this location Chlamydia LPS was present in macrophages. Seventy-nine percent of Chlamydia LPS positive biopsies were also positive to C. trachomatis MOMP (Figure 1). Staining for C. pneumoniae was negative in all patients.

Only 6/42 (14%) controls (5 women) were positive for Chlamydia LPS (6 in l. propria macrophages and 2 in enteroendocrine cells) and 2/6 LPS-positive controls were positive to C. trachomatis MOMP. No biopsies from controls were positive for C. pneumoniae. Slides from 10 patients who were positive for Chlamydia LPS in immunofluorescence and 10 negative controls were studied using immunohistochemistry with polyclonal antibodies to Chlamydia LPS and streptavidin-biotin complex. Chlamydia LPS was found in enteroendocrine cells and/or macrophages in all cases whereas all controls remained negative.

The prevalence of Chlamydia LPS was much higher in patients with IBS (58/65) than it was among controls (6/42). The odds ratio, corrected for differences in age and gender distributions, for mucosal Chlamydia LPS being indicative for presence of IBS was 43.1 (95% CI: 13.2-140.7). Biopsies were analysed by two independent investigators unaware of clinical data. The agreement between the two investigators regarding individual biopsies was 94%, whereas their agreement regarding patient classification was 100%.

We found no correlation between presence of Chlamydia antigens and the type of neuropathy in patients with full thickness biopsies. Although the majority of patients had neuropathy with inflammation, Chlamydia antigens were common also among those who had neuropathic changes without inflammation (LEG: 19/20 = 95%; LG 23/26 = 88%; DN 9/12 = 75%). We found no difference between IBS subgroups with regard to positivity for Chlamydia antigens and 19/21 with D-IBS, 22/22 with C-IBS, and 17/22 with A-IBS were LPS positive.

New biopsies were taken from the duodenum mucosa in 4 LPS-positive patients (Table 1). We confirmed the presence of Chlamydia LPS antigen in patients with Western blotting but the new biopsies were negative for C. trachomatis MOMP both in immunofluorescense and Western blot (Figure 2). Electron microscopy revealed small oval structures resembling intermediate bodies of Chlamydia with characteristic condensed nucleoids of nucleic acid (Figure 3). Mitochondria were observed surrounding the inclusions but did not appear to be in close association with the inclusions. We were unable to confirm the presence of Chlamydia DNA in the same biopsies using 23S ribosomal DNA as target. Results were negative for Chlamydia DNA also when we analyzed micro-dissected Chlamydia LPS positive cells.

We report the novel finding of chlamydial antigens in enteroendocrine cells and macrophages of the small bowel mucosa in patients with severe IBS. LPS and MOMP antigens were detected in mucosa and l. propria of the small bowel in a large majority of patients with IBS, but rarely so in healthy controls. The odds ratio for mucosal Chlamydia LPS being indicative for presence of IBS is much higher than any previously described pathogenetic marker in IBS [21]. These findings raise several questions. The first is whether or not observed immunofluorescence findings represent presence of bacterial antigens. The specificity of the LPS antibody is important in this respect. There is a risk that antibodies may give rise to unspecific binding but such binding should not differ between patients and controls. We therefore think that unspecific binding is an unlikely explanation for our findings. It has previously been shown that this monoclonal anti-LPS antibody does not bind to environmental Chlamydiae [22, 23]. The genus-specific LPS epitope is not shared by other known gram-negative bacteria and monoclonal antibodies do not bind LPS of those organisms [24]. Thus, positive staining of LPS can be considered as a marker of past or present Chlamydia infection. We used different techniques to visualize Chlamydia LPS. In addition to immunofluorescence with a FITC-conjugated monoclonal antibody, we used a polyclonal antibody to Chlamydia LPS and the standard streptavidin-biotin technique for light microscopy. The latter technique also showed presence of antigen in the same cell types. We used Western blot on new biopsies from a limited number of patients and confirmed that Chlamydia LPS was present in the tissue samples from these patients. These findings further support the hypothesis that observed antigens have a bacterial origin.

The second question concerns the identity of the species involved. We used a genus-specific antibody to Chlamydia LPS and species-specific antibodies to C. pneumoniae and C. trachomatis MOMP. Since we found positive staining for C. trachomatis MOMP in 79% of Chlamydia LPS biopsies and none of them was positive for C. pneumoniae we hypothesized that the origin of these antigens is a past or present infection with C. trachomatis.

The third question is if observed antigens represent an ongoing infection or not. We investigated archived biopsy material from the small bowel and this puts certain limits on our ability to ascertain the underlying cause for observed findings as well as their specificity. To confirm and strengthen our immunohistochemical findings we used antibodies to different antigens and also different methods of visualisation. In a limited number of cases we had the opportunity to take new biopsies for more advanced analyses using molecular biology techniques, proteomics and electron microscopy. Whereas electron microscopy showed presence of Chlamydia-like organisms in the cytoplasm of enteroendocrine cells, similar to those described in persistent C. trachomatis infection [25], we were unable to detect chlamydial DNA in a small number of patients using standard extraction and amplification protocols. It is possible that the standard methods we used for nucleic acid retrieval were inadequate for the detection of a persistent infection [26]. Another explanation for the contradictory results could be that chlamydial antigens were remainders of a past, but no longer present infection [27]. The latter seems unlikely in light of the long-term presence of C. trachomatis antigens observed in several of our patients. Long-term presence of antigens is more likely to be attributable to replicating Chlamydiae residing in the diseased tissue [28]. At present, however, we cannot determine if patients with severe IBS have an ongoing intestinal infection with C. trachomatis or not. Neither can we explain the presence of inflammation in myenteric ganglia or neuropathy by the finding of Chlamydia antigens in enteroendocrine cells. A possible link between the presence of Chlamydia antigens in enteroendocrine cells and myenteric inflammation could be autoimmune mechanisms related to heat shock protein 60 [29], but at this time such a mechanism remains purely speculative.

We found Chlamydia antigen also in macrophages. It is known that Chlamydiae become spontaneously persistent following the infection of monocytes and the monocyte is the common host cell for Chlamydiae during persistent infection [30–32]. Chlamydiae may participate in the maintenance of local immunological response and inflammation via infected monocytes/macrophages and also use them to spread infection to other organs [33].

Some of our Chlamydia LPS positive patients did not show the presence of MOMP antigen. Chlamydiae down-regulate the major outer membrane protein (MOMP) expression in the persistent state [34]. Attenuated synthesis of MOMP, usually considered the immuno-dominant surface epitope of the organism, in combination with its almost exclusively intracellular location during persistence may help to provide some relative invisibility from immune surveillance [34].

The presence of C. trachomatis antigens in small bowel mucosa has not been demonstrated in IBS patients before. However, it is known that Chlamydiae can persist in the intestinal mucosa of certain animals for a long time without infection-induced inflammation [35]. The specificity of C. trachomatis antigens for IBS compared to other gastrointestinal diseases is unknown. We did not have access to biopsies from the upper small intestine of patients with for example Crohn's disease or ulcerative colitis. At the same time, it is difficult to rule out the co-existence of IBS with such diseases and this might blur the results of such a comparison. We therefore decided to compare our patients with severe IBS to healthy controls in which IBS and other motility disorders could be excluded.

The finding of C. trachomatis antigens in enteroendocrine cells makes it tempting to suggest a novel pathogenetic mechanism in IBS. Enteroendocrine cells play a pivotal role in the control of gut motility and secretion and increased numbers of enteroendocrine cells have been detected in patients that developed IBS after an acute gastroenteritis [36]. It is unknown if infection alters the function of enteroendocrine cells in man but animal experiments using different models of enteric infection have shown pronounced changes in both numbers of enteroendocrine cells and their function [37, 38]. Serotonin-producing enteroendocrine cells may present an ideal location for Chlamydia due to their abundance of tryptophan. Tryptophan is required for normal development in Chlamydia species and tryptophan metabolism has been implicated in Chlamydia persistence and tissue tropism [39]. We hypothesize that Chlamydia actively may enter EEC searching for tryptofan. Infection may impair EEC serotonin and hormone secretion and cause motility disturbances but further studies on enteroendocrine cell lines or cultured EEC are required for elucidating such a mechanism.

We found chlamydial antigens in enteroendocrine cells and macrophages of the small bowel mucosa in 89% of patients with IBS but in only 14% of controls. Even though we were unable to prove the presence of viable Chlamydia, lack of positive PCR being the major obstacle, the evidence for an intracellular organism or at least a protein structure with an antigen in common with Chlamydia in patients with IBS is strong. Our results suggest that the presence of C. trachomatis antigens in enteroendocrine cells may be involved in the pathogenesis in IBS.