Supported by Crohn's in Childhood Research Appeal (CICRA) charity.
Aberrant response to commensal Bacteroides thetaiotaomicron in Crohn's disease: An ex vivo human organ culture study†
Article first published online: 25 OCT 2010
Copyright © 2010 Crohn's & Colitis Foundation of America, Inc.
Inflammatory Bowel Diseases
Volume 17, Issue 5, pages 1201–1208, May 2011
How to Cite
Edwards, L.A., Lucas, M., Edwards, E.A., Torrente, F., Heuschkel, R.B., Klein, N.J., Murch, S.H., Bajaj-Elliott, M. and Phillips, A.D. (2011), Aberrant response to commensal Bacteroides thetaiotaomicron in Crohn's disease: An ex vivo human organ culture study. Inflamm Bowel Dis, 17: 1201–1208. doi: 10.1002/ibd.21501
- Issue published online: 11 APR 2011
- Article first published online: 25 OCT 2010
- Manuscript Accepted: 23 AUG 2010
- Manuscript Received: 16 AUG 2010
- microbial epithelial cell interactions;
- innate immune system in IBD;
- mucosal immunology;
- infectious agents in IBD;
Human ex vivo evidence indicating that an inappropriate immune response(s) to nonpathogenic bacteria contributes to disease pathogenesis in pediatric Crohn's disease (CD) is limited. The aim of the present study was to compare and contrast the early innate immune response of pediatric “healthy” versus CD mucosa to pathogenic, probiotic, and commensal bacteria.
“Healthy control” and CD pediatric mucosal biopsies (terminal ileum and transverse colon) were cocultured for 8 hours with E. coli O42, Lactobacillus GG (LGG), Bacteroidesthetaiotaomicron (B. theta), or stimulated with interleukin (IL)-1β (positive control). Matched nonstimulated biopsies served as experimental controls. IL-8 was the immune marker of choice. IL-8 mRNA and protein levels were quantified by quantitative polymerase chain reaction and sandwich enzyme-linked immunosorbent assay, respectively.
IL-8 secretion was observed when control, ileal biopsies were exposed to pathogenic O42 and probiotic LGG, with no response noted to commensal B. theta. In comparison, Crohn's ileal biopsies showed impaired ability to induce IL-8 in response to O42 and LGG. Control colonic tissue showed a limited response to O42 or B. theta and LGG significantly reduced IL-8 secretion. Unlike control tissue, however, Crohn's ileal and colonic tissue did respond to B. theta, with more enhanced expression in the colon.
We provide the first ex vivo data to support the notion that aberrant mucosal recognition of commensal bacteria may contribute to pediatric CD. While IL-8 responses to O42 and LGG varied with disease status and anatomical location, B. theta consistently induced significant IL-8 both in ileal and colonic CD tissue, which was not seen in control, healthy tissue. (Inflamm Bowel Dis 2010;)
It is well established that luminal microbiota play an important role in the pathogenesis of Crohn's disease (CD).1 Commensal enteric bacteria have been found to exacerbate the adaptive immune system in CD.2 However, it is thought that this is a secondary response, as no single species has been consistently linked to all cases of CD.3 Moreover, T cells cloned from CD patients (i.e., T cells that should be monospecific) respond to multiple and divergent commensal species.4 One leading hypothesis is that commensal bacteria trigger an inappropriate innate mucosal immunity in susceptible individuals leading to the chronic adaptive inflammatory reaction(s) seen in CD. Paradoxically, innate immune responses to nonpathogenic bacteria are purported to underlie the therapeutic benefits of probiotics in CD.5 To date, studies of probiotics in CD have been contradictory6; the Cochrane systematic review has concluded that their use cannot be recommended until a more rigorous scientific understanding of the predicted immune and bacterial effects is achieved.7, 8 The aim of this study was to investigate how healthy control versus CD mucosa respond to commensal, probiotic, and pathogenic microbes.
Interleukin (IL)-8 was chosen as an “indicator” of mucosal immune activation ex vivo. The role of IL-8 in recruiting neutrophils, T lymphocytes, and natural killer cells to the site of infection is well established9–11; thus, IL-8 is capable of amplifying innate immune responses. In addition, a significant role for IL-8 in the etiology of mucosal ulceration in inflammatory bowel disease (IBD) has been postulated.12 Previous studies have evaluated adult patients with longstanding or steroid-modulated disease and have reported conflicting results regarding the role of IL-8 in IBD pathogenesis. To date, only one study has evaluated the ex vivo levels of IL-8 in pediatric IBD,13 and no studies to date have reported the ex vivo response of pediatric CD tissue to bacterial stimulus. The factors that initiate early mucosal lesions are likely to be identified by studying these responses in children. In addition, understanding cytokine production in pediatric IBD patients may provide insight into disease pathogenesis.13
MATERIALS AND METHODS
Bacterial Strains and Cultures
The human commensal strain Bacteroidesthetaiotaomicron VPI-5482 (ATCC 29148, B. theta) was kindly provided by Prof. Fredrick Backhed.14 The probiotic strain Lactobacillirhamnosus GG (LGG) was kindly provided by Dr. David R Mack.15 The pathogenic Escherichia coli strain, enteroaggregative E. coli (EAEC) O42, Serotype: 044:H18 was kindly provided by Prof. James P Nataro. EAEC O42 expresses plasmid encoded cytotoxin, flagella, and fimbriae.16
B. theta was cultured on blood agar plates (Brain Heart Infusion [BHI] base supplemented with 5% v/v defibrinated horse blood [Oxoid Basingstoke, UK]) at 37°C in an 85% N2/10% O2/5% CO2 atmosphere. A single colony was cultured overnight in 3 mL of thioglycollate medium (supplemented with vitamin K solution [1 mg/L] and histidine-hematin [42 mg/L] all Sigma-Aldrich, Poole, UK) at 37°C without shaking, producing a culture broth containing bacteria in the stationary phase of growth. All other bacterial strains were first cultured on BHI agar plates, followed by overnight growth (without shaking) in 3 mL BHI broth at 37°C in 95% O2, 5% CO2 conditions.
Recombinant cytokine interleukin-1β (IL-1β) was reconstituted according to the manufacturer's instructions (Sigma-Aldrich, Poole, UK). Ten ng/mL of IL-1β was routinely used.
Mucosal endoscopic biopsies were taken from the terminal ileum (TI) and transverse colon (TC) from patients being investigated for abdominal pain, chronic diarrhea, failure to thrive, bloody diarrhea, polyps, or CD. Adjacent specimens were formalin-fixed for histological confirmation and were examined by an experienced pathologist. Biopsies were obtained from two groups, a CD group and a control group, where the final diagnosis was not CD. All diagnoses of CD were made on the basis of clinical and histological findings according to the Porto Criteria (IBD Working Group of ESPGHAN).17 Tissue from the non-CD group, referred to as “Control” was biopsied from regions that showed no disease macroscopically and was reported as histologically normal. Tissue from the CD group referred to as “Crohn's” also showed no or mild disease macroscopically during endoscopy, but was found histologically to range from normal architecture to chronically inflamed. All procedures were performed using Olympus PCF or Fujinon EG/EC-41 pediatric endoscopes and took place between 8:30 and 13:00. The collective age range of all the patients was 2–16 years with a median of 12, and a ratio of 23 males to 22 females. No patients were on therapy at the time of endoscopy.
In Vitro Organ Culture (IVOC)
Immediately after removal, biopsy explants were placed in 5 mL of prewarmed (37°C) IVOC medium for coculture studies. Intestinal biopsies were cultured as previously described.18 Tissue samples were mounted on a foam support and examined using a dissecting microscope. The tissue was oriented with the mucosal surface facing upwards. Only intact nonhemorrhagic tissue (biopsies) was incubated in 12-well plates containing 0.8–1.0 ml of prewarmed (37°C) IVOC medium on a rocking platform at 37°C in 95% O2, 5% CO2 for 8 hours. The IVOC assay was carried out by inoculating 20 μL (≈2 × 107 bacteria) of an overnight or log-phase bacterial culture onto the mucosal surface of the explants. To prevent bacterial overgrowth and drop in pH, the IVOC medium was changed every 2 hours using an aseptic technique. As the bacteria could not be completely removed from the foam support, reinoculation was found to be unnecessary.
IL-8 mRNA Expression
Following infection, biopsies were subjected to RNA extraction using TRIZOL (Invitrogen, Paisley, UK) according to the manufacturer's instructions. RNA samples were subjected to DNase treatment using a TURBO DNase-free kit (Ambion, Cambridgeshire, UK), following the manufacturer's protocol for extended digestion. Approximately 1.5 μg of total RNA was reverse-transcribed using the BioScript cDNA synthesis kit (Bioline, London, UK). Quantitative polymerase chain reaction (qPCR) was performed with a Mastercycler EP realplex thermal cycler (Eppendorf, Histon, Cambridge, UK), using a QuantiTect SYBR Green PCR kit (Qiagen, Crawley, UK). The difference in IL-8 expression between infected and matched noninfected controls was determined using the ΔCt method, with correction for differing PCR efficiencies.19 The housekeeping gene ribosomal phosphoprotein P0 (RPLP0) was used to normalize IL-8 expression. Primers were synthesized by VH-BIO (Gateshead, Newcastle, UK) Primers used were IL-8: (sense GAA CTG AGA GTG ATT GAG AGT GGA and antisense primer CTC TTC AAA AAC TTC TCC ACA [product size 88 basepairs and 85% amplification efficiency]). RPLPO: (sense GCA ATG TTG CCA GTG TCT G and antisense primer GCC TTG ACC TTT TCA GCA A [product size 103 basepairs and 90% amplification efficiency]). The primer sequence IL-8 (2627) was downloaded from the public RTPrimerDB database (http://medgen.UGent.be/rtprimerdb/).20 Primer sequences for RPLP0 were obtained from Dydensborg et al.21 Primers were intron–exon spanning in order to minimize the amplification of genomic DNA and tested to obtain optimum PCR efficiency and minimal primer-dimer formation.
For the detection of IL-8 protein, an IL-8 sandwich enzyme-linked immunosorbent assay (ELISA) kit was used following the manufacturer's instructions (Quantikine, R&D, Abingdon, UK).
Statistical analyses were performed using GraphPad Prism 4 (San Diego, CA). Differences in gene or protein expression between the matched unstimulated biopsies and stimulated biopsies (denoted with an asterisk on reaching statistical significance) were evaluated using a paired Wilcoxon Signed Rank Test. A P value of <0.05 was considered statistically significant (*P ≤ 0.05 **P ≤ 0.01). Differences in mRNA or protein expression between control or CD patient groups were evaluated using a two-tailed Mann–Whitney U-test.
Ethical approval for obtaining intestinal mucosal biopsies from patients undergoing routine endoscopic procedure was granted to the Centre for Paediatric Gastroenterology, Royal Free and University College London Medical School by the local Ethical Committee of the Royal Free Hampstead NHS Trust Hospital (Ethics Reference 04/Q0501/122). The biopsies were taken under the direction of the clinician in charge after fully informed parental consent was obtained.
IL-8 mRNA Expression in Response to IL-1β and Bacterial Stimulus
To validate our ex vivo model using human intestinal tissue explants, initial experiments focused on quantifying mucosal IL-8 induction in response to a well-known potent agonist, IL-1β. Matched biopsies from the same patient were exposed to IL-1β or IVOC medium alone for 8 hours prior to RNA extraction or measurement of IL-8 secretion. The IL-8 mRNA response was variable between the TI and TC in control patients following IL-1β exposure (Fig. 1A). Tissue from both regions of CD patients, however, was hyporesponsive to IL-1β (Fig. 1A). Subsequently, IL-8 mRNA expression in response to bacterial stimuli was investigated. TI and TC biopsies from control and CD groups were inoculated with EAEC strain O42, LGG, B. theta, or medium alone for 8 hours. The IL-8 mRNA response was variable between the TI and TC in control and CD patients, with no statistically significant modulation in expression observed (Fig. 1B–D).
IL-8 Secretion in Response to IL-1β
Secretion of IL-8 protein increased on exposure to IL-1β in both the TI and TC tissue in the majority of patients (Fig. 2A), except biopsies from the TC of CD patients, a cohort that showed spontaneous high basal IL-8 production, with minimal increase on exposure to IL-1β (Fig. 2A).
IL-8 Secretion in Response to Pathogenic Bacterial Stimulus
The pathogenic E. coli strain EAEC O42 was included as a candidate for comparison with LGG and B. theta because 1) EAEC O42 is known to strongly adhere to pediatric colonic mucosa under IVOC conditions18 and 2) O42 strongly induces IL-8 secretion in vitro.22 TI and TC biopsies from control and CD groups were inoculated with overnight cultures of EAEC strain O42 or medium alone (n = 31/group). Control TI responded by substantial induction of IL-8 protein (no stimulated tissue mean 1237.0 pg/mL [SD ± 899.9] versus O42 infected tissue mean 2864.5 pg/mL [SD ± 2084.7]; P = 0.0156). In contrast, Crohn's TI and TC showed less ability to induce IL-8 above basal levels (TI no stimulated tissue mean 2827.5 pg/mL [SD ± 2050.761] versus O42 infected TI tissue mean 2900.7 pg/mL [SD ± 1716.7]; TC no stimulated tissue mean 2126.5 pg/mL [SD ± 785.3] versus O42 infected TC tissue mean 2688.3 pg/mL [SD ± 2207.09] P = 0.0415). IL-8 expression was found to be more variable in control TC (Fig. 2B). Differences observed in IL-8 secretion between the control and CD groups in either location were not found to be statistically significant.
IL-8 Secretion in Response to Nonpathogenic Bacterial Stimulus
LGG is one of the best-studied probiotic bacteria in clinical trials and has been found to have a benefit in CD.23, 24 To investigate the response of CD tissue to LGG, biopsies were exposed to overnight cultures of the probiotic (n = 30) and IL-8 protein quantified. A significant induction of IL-8 protein was observed in the TI of control patients on bacterial exposure (Fig. 2C) (no stimulated TI tissue mean 1583.7 pg/mL [SD ± 1094.6] versus LGG infected TI tissue mean 2020.7 pg/mL [SD ± 1351.7]; P = 0.05). In contrast, control colonic tissue responded to LGG by causing inhibition of the basal IL-8 protein levels (no stimulated TC tissue mean 2551.8 pg/mL [SD ± 2020.3] versus LGG infected TC tissue mean 1936.9 pg/mL [SD ± 1695.0]; P = 0.0313). No statistically significant modulation in secretion was observed in the CD group (Fig. 2C). However, there was a trend towards decreased IL-8 secretion to LGG in the TI (no stimulated TI tissue mean 3190.4 pg/mL [SD ± 1906.7] versus LGG infected TI tissue mean 1617.6 pg/mL [SD ± 820.1]; P = 0.06). A statistically significant difference in IL-8 secretion between the control and CD groups in both the TI and TC was observed (P = 0.014 and P = 0.028, respectively) (Fig. 2C).
B. theta is a prominent member of the normal human intestinal microbiota.25 To further understand B. theta-human gut crosstalk, tissue explants were cocultured with B. theta (n = 27). No statistically significant induction of IL-8 protein was observed in either the TI or the TC of control patients on exposure to the commensal B. theta (Fig. 2D) (TI none stimulated tissue mean 1853.7 pg/mL [SD ± 1851.5] versus B. theta infected TI tissue mean 1660.7 pg/mL [SD ± 1754.5]; P = 0.513; TC no stimulated tissue mean 3489.3 pg/mL [SD ± 3399.6] versus B. theta infected TC tissue mean 3827.2 pg/mL [SD ± 3958.7]; P = 0.188). A significant increase in IL-8 protein was observed in the TI and this increase was more marked in TC of CD patients (TI no stimulated tissue mean 1358.1 pg/mL [SD ± 1029.3] versus B. theta infected TI tissue mean 2710.1 pg/mL [SD ± 2073.0]; P = 0.0034; TC no stimulated tissue mean 5374.5 pg/mL [SD ± 6928.0] versus B. theta infected TC tissue mean 7686.7 pg/mL [SD ± 6040.6]; P = 0.0156). A statistically significant increase in IL-8 secretion between the control and CD groups in both the TI and TC was observed (P = 0.019 and P = 0.028, respectively) (Fig. 2D).
Despite major strides in our understanding of the etiology of CD, the complex interactions between gut microbiota and the human gastrointestinal mucosal immune system remain largely undefined. In recent years probiotics have been investigated as a potential biologic therapy for IBD, as their antiinflammatory actions may aid in restoring immune homeostasis and health. However, studies to date of probiotics in CD have been contradictory and disappointing.
The aim of this study was to characterize and compare the innate mucosal signals (IL-8 expression) generated ex vivo in response to “commensal,” “probiotic,” and “pathogenic” bacteria in order to ascertain if CD mucosa responds differentially to healthy mucosa. The major finding of our study was the differential response of healthy control versus CD tissue to B. theta. Both the TI and the TC of pediatric CD patients reacted with a significant increase in IL-8 secretion, a response not observed in the control group (Fig. 2D). This response was not observed with the proinflammatory cytokine IL-1β, and indeed CD tissue tended towards a “hyporesponsive state” to IL-1β (Fig. 2A). The TI and the TC of CD patients also reacted with a significant increase in IL-8 secretion on exposure to the pathogen O42, although this response was also observed with the control group (Fig. 2B), suggesting that the CD response to B. theta parallels that of control tissue to a pathogen. To date there are no reported ex vivo pediatric studies of the response of CD tissue to bacterial stimulus. Thus, we present here the first ex vivo data to support the current theory that commensal bacteria may be initially responsible for the inappropriate induction of the innate immune response. Why and how CD mucosa responds differentially to B. theta remains unclear. It will be of great interest to elucidate the mechanism for this aberrant response.
Induction of epithelial IL-8 production may drive recruitment of both neutrophils, blood-borne DC precursors, and T lymphocytes, which will not have been subject to conditioning and can be directly activated by the bacteria and the proinflammatory cytokine milieu.26 The cytokine milieu created can prime either a tolerogenic or an inflammatory immune response by modulating T-cell lineage differentiation.27, 28 Dependent on DC priming, naïve CD4+T cells can differentiate into either of three subsets of effector T cells (Th1, Th2, and Th17 cells) or into regulatory T cells (Tregs).29 Therefore, it is plausible that the inappropriate IL-8 response to the commensal strain B. theta observed in this study could indeed contribute to the chronic inflammatory reaction seen in CD. Indeed, another strain of commensal Bacteroides, B.vulgatus, has been reported previously to induce colitis in several animal models of IBD30, 31 as well as in humans.32, 33 By contrast, a protective role for B.vulgatus has been noted in other models of experimental colitis,34 pointing to a critical role for the host genetic background.
How the gut microbiota shapes intestinal helper T-cell response at steady-state has recently stimulated much interest given the accumulating evidence of microbial dysbiosis in IBD.35 The first bacterial product shown to modulate T-cell lineage differentiation of the mammalian immune system was the zwitterionic polymer polysaccharide A (PSA), a surface carbohydrate structure of another commensal strain of Bacteroides, B. fragilis.36, 37 Another commensal species able to modulate T-cell lineage differentiation is the segmented filamentous bacteria (SFB), which adhere to the ileal mucosa and to peyer's patches and can alone orchestrate a striking range of T-cell functions, simultaneously stimulating T cells of Th17, Th1, Th2, and regulatory cell phenotype.38 Intriguingly, lack of these bacteria in even a diverse flora prevents normal establishment of mucosal lymphocyte populations.28 It is not known in humans which are the species capable of driving this response. This may be important, as there is evidence that aberrant proinflammatory cytokine production may promote conversion of Treg cells into inflammatory effector T cells that could cause harmful inflammation.39
In the ex vivo model employed in this study the tissue can only be maintained for 8 hours, allowing study only of the immediate proinflammatory response and not the subsequent adaptive immune response. As IL-8 is capable of recruiting neutrophils, T lymphocytes, and natural killer cells,9–11 we suggest that this recruitment would modulate downstream T-cell responses. It will be of interest to further study in this system the DC IL-12/IL-23 axis. Limitations of biopsy size and number mean that we have not determined the cell type(s) secreting IL-8 in this system. This may be a direct epithelial response9 or secondary to B. theta-induced IL-17 production.40
Further studies are required to determine what bacterial products drive this innate response, which appears specific for both bacterial species and anatomical location. Indeed, normal control biopsies showed a greater ileal than colonic response to the pathogenic E. coli and probiotic strain, with minimal or even reduced IL-8 response to commensal B. theta. By contrast, biopsies from the CD patients showed less difference between ileal and colonic responses, and showed a response to the commensal B. theta similar to the pathogenic E. coli and quite distinct from the exogenous probiotic LGG. Differences between ileal and colonic responses have previously been ascribed to the increased colonic mucus barrier.28 This barrier is disrupted in CD41 and may explain why a lesser difference is seen the CD group.
The suppression of colonic IL-8 secretion by LGG in the control group might suggest good potential for use as a probiotic for ameliorating intestinal inflammation. However, this suppression was limited to the ileum in CD. Probiotics including LGG have general been less effective in CD than UC42 and adverse responses have been reported.43–46 Given the immunoregulatory potential of bacteroides species36 the question arises whether B. theta might be a potentially useful probiotic agent in CD. If the IL-8 response seen was indeed secondary to innate immune cell production of IL-17,40 then this may indeed promote beneficial maintenance of epithelial integrity and clearance of extracellular pathogens.38, 47, 48 There is evidence that the Bacteroides group is significantly decreased in CD and their use as probiotics has previously been suggested.49
We noted a degree of variability between patient samples. The CD biopsies ranged in inflammatory status, and some had relatively more longstanding disease. While these patients were not on therapy at the time of investigation, we cannot rule out any effects of their previous treatments. However, no significant differences in IL-8 production have been reported between previously and newly diagnosed pediatric patients.13 The greatest variability seen in this study was with IL-8 mRNA, which did not always correlate with protein secretion, as previously reported by others.50 Similar discordance between intestinal mRNA and protein expression has been seen with other cytokines, suggesting that it may be more physiologically relevant to analyze protein concentration rather than mRNA levels.50
In conclusion, we identified significant differences in IL-8 protein response to B. theta in tissues from CD patients compared to controls. In the absence of an animal model that accurately recapitulates the features of CD, this ex vivo model may be the closest we can currently get to the in vivo situation. Thus, we present here the first pediatric ex vivo data to support the role of aberrant recognition of commensal bacteria by the innate immune system in CD. We further hypothesize that specific strains of the commensal microbiota may well have the capacity to prime an individual for the development of CD or evoke relapses in enteric inflammatory disease. The in vitro organ culture model system used in this study will be a useful tool in further determining the immunoregulatory properties of other bacterial species. We identified differences between CD and normal controls in mucosal responses to the established probiotic LGG which may explain its lack of beneficial effect in CD. The experimental data suggests dominant immunoregulatory functions of Bacteroides species; therefore, B. theta might be a better choice for probiotic in CD.
We thank all the patients and parents for consenting to provide the tissue. Also, we thank the Registrars who obtained informed consent, especially Dr. Camilla Salvestrini. We also thank Prof. Fredrick Backhed, Prof. James P Nataro, and Dr. David R Mack for provision of the bacterial species, and the Histopathology Department at the Royal Free.
- 7Probiotics for maintenance of remission in Crohn's disease. Cochrane Database Syst Rev. 2006; CD004826., , , et al.