SEARCH

SEARCH BY CITATION

Keywords:

  • bacterial translocation;
  • endotoxin;
  • NOD2;
  • SNPs;
  • NF-κB;
  • Crohn's disease

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Background:

Recent insights into the pathogenesis of Crohn's disease (CD) point to an important role of the mucosal barrier and intestinal microflora that may induce a chronic inflammation after crossing the intestinal barrier. The first detected susceptibility gene for CD, NOD2, is a pattern recognition receptor (PRR) for the recognition of the bacterial cell wall component muramyldipeptide (MDP). Binding of MDP to NOD2 is followed by activation of proinflammatory pathways mainly regulated by nuclear factor kappa B (NF-κB). In this study we investigated whether impaired recognition of MDP via NOD2 variants is associated with increased bacterial translocation across the epithelial barrier and whether this is followed by increased or decreased NF-κB activation.

Methods:

NOD2 variants were analyzed in 36 CD patients and 30 controls. Endotoxin was stained by immunohistochemistry in 30 intestinal biopsies from patients carrying NOD2 variants (NOD2-mut) or being NOD2 wildtype (WT). Junctional proteins were visualized by immunofluorescence and quantified by Western blotting. NF-κB activation was analyzed by immunohistochemistry in specimens from NOD2-WT and NOD2-mut CD and control patients.

Results:

We demonstrated the increased presence of endotoxin in the mucosal lamina propria of CD patients carrying NOD2 variants. This was associated with an altered composition of epithelial cell–cell contacts. Patients carrying NOD2 variants displayed increased NF-κB activation in the mucosa.

Conclusions:

This study for the first time demonstrates that translocation of luminal bacteria and/or bacterial products into the intestinal mucosa is increased in patients carrying NOD2 variants, leading to higher activation of proinflammatory signaling cascades. (Inflamm Bowel Dis 2010)

The human mucosa faces an enormous density of bacteria, antigens, and pathogens in the gut lumen. The epithelial cell layer has the task of selectively blocking the passage of toxins and pathogens into the mucosa, while ensuring resorption of food and fluid. Defects in the tightly regulated barrier function of the mucosa and the epithelium can lead to the influx of antigens into the intestinal wall. Recognition and clearance of these pathogens are critical to the survival of the organism. Pattern recognition receptors (PRRs), which include members of the intracellular NOD protein family, recognize molecular patterns of various pathogens (PAMPs).1, 2 One member, NOD2, has been linked to susceptibility to Crohn's disease (CD), an inflammatory bowel disease (IBD).3, 4

The ligand of NOD2 is the bacterial wall component muramyldipeptide (MDP). After MDP binding to a leucine-rich repeat (LRR) domain, NOD2-mediated signaling normally results in activation of the transcription factor nuclear factor kappa B (NF-κB) via a kinase called RICK. Activation and nuclear translocation of the NF-κB signaling pathway has different consequences. On the one hand, NF-κB induces numerous proinflammatory genes that play a role in the regulation of innate and adaptive immune responses,5 e.g., the controlled expression of immune modulators such as IL-2, IL-6, IL-8, GM-CSF, or chemokines and their receptors, enzymes (iNOS, MMP-9), or cell-surface adhesion molecules (E-selectins).6 On the other hand, expression of antiapoptotic genes like FLIPL, TRAF1, or members of the Bcl-2 family is induced via NF-κB signaling.7 For the 3 frequent NOD2 variants, so-called single nucleotide polymorphisms (SNPs) 8, 12, and 13, there exist conflicting data regarding their functionality. The “loss of function” theory is supported by the observation that a defect of NOD2 variants to activate defense mechanisms results in an increased bacterial translocation, at least in animal models.8–11 On the other hand, mice with SNP13 mutation showed an excessive IL-1β production upon bacterial challenge, rather supporting the “gain of function” scenario.12 Most of the data on NOD2 function have been derived in cell lines overexpressing the wildtype (WT) protein and its variants or in different mouse models. However, it is obvious that patients have to be studied to understand the impact of NOD2 variants in human disease.

To date, the etiology of CD is not yet clear. However, recent years have brought new and important insights. Besides genetic and environmental factors, the luminal flora seems to be involved in the pathogenesis of chronic intestinal inflammation.13 In CD patients the mucosal barrier may become leaky, leading to uncontrolled uptake of antigens and proinflammatory molecules, including luminal bacteria and bacterial products from the gut lumen.13 An increased permeability in patients (and their relatives) with CD was shown 20 years ago.14, 15

Cell–cell contacts between intestinal epithelial cells for obvious reasons play an important role in the barrier function of the epithelium. The junctional complex is composed of different adhesion proteins, forming apical tight junctions and subjacent adherens junctions. Three prominent tight junction proteins or protein families are occludin, the claudins, and the Zonula occludens (ZO) proteins. Occludin was shown to be a central mediator of tight junction formation as occludin-deficient embryonic stem cells were unable to form functional tight junctions.16 In addition, overexpression of mutated occludin variants changed barrier capacities.17–19 The claudin family is composed of 24 transmembrane proteins. The most prominent members claudin-1, -2, and -4 regulate the “tightness” and specific ion selectivity of the epithelium.20–22 ZO-1 acts as a linker between the actin cytoskeleton and other tight junction-associated proteins and regulates paracellular permeability.23–28 The adherens junction protein family prominently includes cadherins and catenins that fulfil anchoring functions.29, 30

Our hypothesis was that in CD patients luminal bacteria might reach the lamina propria via modified cell–cell contacts followed by a dysregulated immune response. In NOD2 variant patients a lack of detection of invaded bacteria and bacterial material could be followed by impaired clearance of those bacterial products and subsequent activation of proinflammatory pathways. Therefore, the aim of the study was to establish a link between NOD2-mediated signaling and bacterial translocation in CD patients.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Patients

Surgical specimens were taken from healthy areas of the colonic mucosa (ascending or sigmoid colon) of patients undergoing surgery for colorectal carcinoma (>10 cm distance from the tumor) or from the inflamed colonic mucosa of CD patients. Histological evaluation was performed by an experienced IBD pathologist. IBD patients were treated with 5-aminosalicylic acid (5-ASA) and/or steroids. No patients treated with immunosuppressants were included in this study. The therapies had no influence on the results. The study was approved by the University of Regensburg Ethics Committee. Thirty samples from colon without inflammation and 36 CD samples with a moderate/severe inflammation were used.

Immunohistochemistry

Demasking of paraffin-embedded sections was performed as previously described.31 For the detection of endotoxin, specimens were incubated with primary mouse anti-gram-negative endotoxin antibody (IgG2a, monoclonal, clone B40/23; Acris, Hiddenhausen, Germany). Mouse IgG2a (Sigma, Taufkirchen, Germany) was used as isotype control at identical concentrations. For the detection of NF-κB(p65)-positive cells slides were incubated with primary mouse anti-NF-κB, p65 subunit (monoclonal, Millipore, Schwalbach, Germany) and mouse IgG3 (Acris) as isotype control, respectively. Biotin-conjugated goat antimouse secondary antibody (IgG(H+I); Jackson ImmunoResearch, Hamburg, Germany) and consecutively the ABC-Elite-standard-system (Vector Laboratories, Burlingame, CA) were applied. After washing, the tissue was incubated with NovaRED (AEC, Vector Laboratories) for red immunostaining. Double-labeling immunohistochemistry was performed as previously described.31 Immunohistochemically stained sections were captured (4 high-power fields each) with a microscope at the indicated magnifications (Zeiss Axiovert, Goettingen, Germany). For semiquantitative analysis NF-κB positively stained cells were counted in 3 high-power fields by an independent investigator blinded to the NOD2 genotype.

Immunofluorescence

Immunofluorescent stainings were performed for claudin-1, -2, -4, occludin, ZO-1, β-catenin, and E-cadherin in surgical specimens. After blocking of endogenous peroxidases31 for nonspecific binding slides were incubated in 1% bovine serum albumin / phosphate-buffered saline (BSA)/(PBS) for 30 minutes. After washing in PBS primary antibodies (rabbit antihuman claudin antibodies: Acris; rabbit anti-occludin antibody: Santa Cruz Biotechnology, Santa Cruz, CA; mouse anti-ZO-1 antibody: BD, Heidelberg, Germany; mouse anti-β-catenin antibody: Biomol, Hamburg, Germany; rabbit anti-E-cadherin antibody: Cell Signaling, Danvers, MA) were applied corresponding to the advised dilutions of the manufacturers' protocol. Isotype stainings assured specific staining results. The slides were then incubated with goat antirabbit Alexa546 or goat antimouse Alexa546 secondary antibody, respectively (Molecular Probes/Invitrogen, Karlsruhe, Germany). Nuclei were 4′,6-diamidino-2-phenylindole (DAPI)-stained with a mounting medium (Vectashield, Vector Laboratories). Immunofluorescently stained sections were captured (4 high-power fields each) with a microscope at the indicated magnifications using fluorescent light (Zeiss Axiovert).

Fluorescence In Situ Hybridization (FISH)

Paraffin-embedded intestinal tissue sections (5 μm) were fixed in 4% paraformaldehyde and rehydrated in a graded ethanol series. Slides were incubated for 30 minutes in a preheated prehybridization solution (0.9 M NaCl, 20 mM Tris-HCl, 0.01% SDS) at 46°C. This was followed by a 3-hour incubation with the probe-containing hybridization solution (0.9 M NaCl, 20 mM Tris-HCl, 0.05% SDS, 5 ng/mL probe) at constant 46°C in a humid chamber. Probe sequences were as follows: Universal Bacteria Antisense 16S rRNA (EUB338) Cy3 – 5′-GCT GCC TCC CGT AGG AGT-3′. Subsequently, slides were washed for 15 minutes in preheated washing buffer (0.9 M NaCl, 20 mM Tris-HCl, 0.01% SDS). After washing in H2O slides were air-dried and counterstained with DAPI. Microscopic evaluation of stained bacteria was performed by fluorescent microscopy.

Isolation of Intestinal Epithelial Cells and Generation of Whole Protein Lysates

Isolation of intestinal epithelial cells was performed as described previously.32, 33 For the generation of whole protein lysates surgical specimens were dissected into small pieces, kept in RIPA buffer (Tris, NaCl, deoxycholic acid, Triton-X-100, SDS, complete proteinase inhibitor mixture; Roche, Mannheim, Germany) and homogenized. Subsequently, lysates were incubated on ice for 30 minutes followed by centrifugation. Supernatants were collected and the protein concentration was determined.

Western Blot

Western blot analysis was performed according to standard protocols using the following primary antibodies: mouse anti-β-catenin (Biomol, Hamburg, Germany, 5 μg/mL), rabbit anti-claudin-1 (Zymed, South San Francisco, CA; 2 μg/mL), rabbit anti-claudin-2 (Zymed, 2 μg/mL), mouse anti-claudin-4 (Zymed, clone 3E2C1, 2 μg/mL), rabbit anti-E-cadherin (Cell Signaling/NEB, Frankfurt, Germany, 1:1000), mouse anti-NF-κBp65 (Chemicon, Hampshire, UK, 10 μg/mL), mouse anti-occludin (BD, San Diego, CA; clone 19, 1:250) and mouse anti-ZO-1 (BD, clone 1, 1:250). Nitrocellulose membranes were incubated with primary antibodies for 1 hour at room temperature under slight shaking. Specific staining was controlled with corresponding isotypes. Equal loading was ensured by staining with mouse anti-β-actin antibody (Chemicon, Temecula, CA; clone C4, 1:3000). After washing the according horseradish peroxidase (HRP)-conjugated secondary antibody (goat antirabbit IgG-HRP, 1:8000; goat antimouse IgG-HRP, 1:3000) was added and the membrane was incubated for another hour under gently shaking. Proteins were detected using the ECL-Plus-western blotting detection system (Amersham Life Science, Braunschweig, Germany).

Statistical Analysis

Data are expressed as means ± standard deviation (SD). Statistical analyses were performed using the SigmaPlot 8.0 Student's t-test. Differences were considered significant at P < 0.05.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Strong Colonization of the Intestinal Mucus Layer and Translocation of Luminal Bacterial Compounds in CD Patients

We first investigated the bacterial colonization of the intestinal mucus and the epithelium in CD patients and controls using FISH technology. The mucus layer lining the intestinal wall showed a strong colonization of adherent bacteria in CD patients (Fig. 1B). In the mucus of controls bacterial colonization was almost absent (Fig. 1A).

thumbnail image

Figure 1. Bacterial colonization of the intestinal mucus layer. Detection of bacteria in the colonic mucus layer of a control patient (A) and a patient with CD (B) using FISH technology. Bacteria were visualized with the universal antisense EUB338 probe. Nuclei were counterstained with DAPI. The mucus of the inflamed intestinal tissue of the CD-patient showed a strong bacterial colonization (B), in the control patient no bacteria where detected (A). Representative for 3 patients each. Magnification 630×.

Download figure to PowerPoint

However, in contrast to other reports, we were not able to detect translocated bacteria with FISH technology in the lamina propria neither in CD patients nor in controls. In order to obtain information about a potential translocation of luminal bacteria or bacterial products into the tissue we immunohistochemically localized mucosal endotoxin deposits. We found the level of endotoxin accumulation to be correlated with the NOD2 genotype of the patients. Semiquantitative analysis was performed visually by evaluating the whole section. In the intestinal tissue of controls neither in NOD2 WT individuals (Fig. 2A) nor in individuals with NOD2 variants (Fig. 2B,C) was a positive endotoxin staining obtained. By contrast, CD patients showed a clear accumulation of endotoxin in the intestinal tissue. The distribution of positively stained cells corresponded to the translocation route—strongly positive-stained areas were predominantly located on the mucosa border as primary gate for invading bacteria (Fig. 2D,F) as well as in areas of tissue destruction (Fig. 2E) and faded toward the submucosa. Regarding the NOD2 genotype, clear differences were obvious in the group of CD patients: patients with the WT gene only revealed a moderate accumulation of endotoxin (Fig. 2D), patients with heterozygous SNP8 in NOD2 showed a strong accumulation of endotoxin reaching from basolateral to far below the crypt basis (Fig. 2E). CD patients with heterozygous SNP12 had similar endotoxin levels preferably located in the mucosal border areas as patients without NOD2 mutation (Fig. 2F). In patients with heterozygous SNP13 a comparably strong accumulation of endotoxin, which was distributed from the epithelium to areas close to the crypts, was observed (Fig. 2G). Unspecific staining was excluded by isotype controls (Fig. 2H).

thumbnail image

Figure 2. Immunohistochemical detection of endotoxin in intestinal tissue. Immunohistochemical staining of bacterial endotoxin in paraffin-embedded human intestinal tissue of control patients and CD patients. (A) Control patient without polymorphism, (B) control patient SNP8, (C) control patient SNP13, (D) CD patient without polymorphism, (E) CD patient SNP8, (F) CD patient SNP12, (G) CD patient SNP13, (H) isotype control. All indicated NOD2 polymorphisms were heterozygous. Representative for 12 control patients and 18 CD patients. Magnification 200×.

Download figure to PowerPoint

Identification of Cell–cell Contacts as Route for Luminal Bacteria to Overcome the Epithelium

Both the strong colonization of the mucus layer and accumulation of endotoxin in the mucosa in CD patients suggested that the primary access could occur by overcoming the epithelial barrier. To clarify whether cell–cell contacts in the intestinal epithelium of CD patients are modified we investigated the presence of essential tight junction and adherens junction proteins.

The functionality of intercellular contacts is maintained by a complex network of numerous integral membrane proteins involved in sealing (claudin-1, claudin-4), pore-formation (claudin-2), adhesion (ZO-1), connection (β-catenin, E-cadherin), or hitherto unknown (occludin) functions. In this regard a morphological (light or electron microscopical) intact structure of cell–cell contacts that we observed both in CD patients and controls (data not shown) does not necessarily ensure a functional epithelial barrier. To determine potential differences in the expression of these proteins in epithelial cells from CD patients immunofluorescent stainings and Western blot analyses were performed, indicating significant differences between CD patients and controls. In inflamed CD-associated tissue increased levels of claudin-1 and -2, occludin, and ZO-1 could be detected by immunofluorescence staining (Fig. 3A(a); CD) as compared to control patients (Fig. 3A(a); NI). Staining of claudin-1 and the pore forming claudin-2 was increased in crypts of CD patients. ZO-1 and occludin accumulated in the entire inflamed mucosa (Fig. 3A(a)). Regarding claudin-4 expression, no difference was observed between CD patients and controls (Fig. 3A(a)). Immunostaining revealed clear differences in the expression of intestinal adherens junction proteins β-catenin and E-cadherin between biopsy samples from CD patients as compared to noninflamed controls. Staining for proteins was reduced in the intestinal tissue of CD patients versus control patients (Fig. 3A(b)). Unspecific staining was excluded by isotype controls (Fig. 3C; exemplarily shown for 1 mouse (left) and 1 rabbit (right) isotype antibody). To confirm the differences of relevant cell–cell contact proteins found in immunofluorescent staining, Western blot analyses were performed using lysates from isolated intestinal epithelial cells (IEC) or whole biopsies (whole protein). In samples derived from inflamed areas of CD patients both in IEC lysates and whole tissue lysates, claudin-1, -2, occludin, and ZO-1 expression was increased as compared to noninflamed controls (NI). Claudin-4 expression was unchanged (Fig. 3B). Expression of the adherens junction proteins β-catenin and E-cadherin was clearly reduced in IEC-lysates as well as whole tissue lysates derived from CD patients (Fig. 3B). Equal loading was confirmed by Western blots for β-actin (Fig. 3B, lower lanes each). Thus, the Western blot results completely supported the immunostaining findings.

thumbnail image

Figure 3. Alterations of epithelial cell–cell contact proteins in CD. (A) Immunofluorescent localization of the (a) tight-junction proteins claudin-1, -2, -4, occludin, and ZO-1 and (b) adherens junction proteins β-catenin and E-cadherin in intestinal tissue of control patients and patients with CD. (c) Representative isotype controls for primary mouse antibodies (left) and rabbit antibodies (right). Magnification 200×. Paraffin-embedded sections. This figure is representative for 12 patients each. (B) Protein chemical detection of cell–cell-contact proteins in lysates of intestinal tissue (whole protein) and isolated IECs of control patients and CD patients. The upper band shows the indicated protein; the upper band represents the loading control β-actin. In inflammation-associated samples (CD) both in cell lysates and IEC the TJ proteins claudin-1, -2, occludin and ZO-1 were heightened detectable versus control samples. Claudin-4 did not show differences between CD and controls. The AJ proteins β-catenin and E-cadherin were diminished in lysates of CD patients. Using Western blot technology ZO-1 was not detectable in IEC. This figure is representative for 3 patients each.

Download figure to PowerPoint

NOD2-dependent Activation of NF-κB in the Intestine of CD Patients

Recognition of the bacterial cell wall component MDP by NOD2 LRR domains induces the activation of NF-κB signaling and subsequently the induction of proinflammatory genes modulating innate and adaptive immune responses. To investigate the influence of NOD2 polymorphisms on the activation of NF-κB we examined its induction in human intestinal tissue with respect to NOD2 genotypes. All NOD2 variant patients in the investigated group were heterozygous carriers of SNP8, SNP12, or SNP13 variants; no compound heterozygous or homozygous carrier was found in the study population. In controls no significant differences in the amount of activated NF-κB were observed between carriers of the NOD2 WT allele (2.2 ± 4.9 NF-κB(p65)-positive cells per 3 high-power fields [HPF]; Fig. 4A; NI WT) and patients with the respective SNPs (3.6 ± 6.3 NF-κB(p65)-positive cells per 3 HPF; Fig. 4A; NI MUT). In the intestinal tissue of NOD2-WT CD patients the number of NF-κB(p65)-positive cells (4.9 ± 5.5 per 3 HPF) was comparable to the control groups (Fig. 4A; CD WT). CD patients with NOD2 variants showed a significant increase of NF-κB activation that was reflected by 40.1 ± 42.6 NF-κB(p65) positively stained cells per 3 HPF as compared to NOD2-WT CD patients (Fig. 4A; CD MUT, P < 0.05).

thumbnail image

Figure 4. Activation of NF-κB in CD. Immunohistochemical detection and semiquantitative capture of NF-κB(p65)-positive cells in intestinal tissue of control patients and CD patients with NOD2 wildtype (WT) and NOD2 mutated (MUT) genotype, respectively. (A) NF-κB(p65)-positive cells per 3 fields of sight were counted. (B) Representative microscopic images: (a) control patient with NOD2 WT, (b) control with NOD2 SNP8, (c) CD patient with NOD2 WT, (d) CD patient with NOD2 SNP8, (e) CD patient with NOD2 SNP12, (f) CD patient with NOD2 SNP13, (g) double staining for NF-κB(p65) (red) and intestinal macrophages (CD68, blue). (h) Single IMAC, (i) isotype control—single staining with substituted primary antibody, (k) isotype control—double staining with substituted primary antibody. *P < 0.05; **P < 0.01. magnification 200× (a–g,i,k) or 1000× (h). This figure is representative of ≥3 patients per group.

Download figure to PowerPoint

Detailing of CD patients on the basis of their SNP variant demonstrated that in the intestine of patients with SNP8 (81.4 ± 38.3 positively stained cells/3 HPF) and SNP12 (34.0 ± 10.8 positively stained cells/3 HPF) a significantly higher activation of NF-κB occurred compared to CD patients without polymorphism (Fig. 4A, P < 0.05). In the intestinal tissue of CD patients with NOD2 SNP13 NF-κB activation was comparable to nonmutated CD patients (2.0 ± 3.5 positively stained cells/3 HPF; Fig. 4A). In Figure 4B representative micrographs of immunohistochemically localized NF-κB(p65) and the corresponding semiquantitative analyses are shown. Within the group of control patients independently from the NOD2 genotype no NF-κB(p65) positively stained cells were detected (Fig. 4B(b)). Double-staining immunohistochemistry revealed colocalization of NF-κB(p65)-staining (red) with the macrophage marker CD68 (blue deposit) (Fig. 4B(g)). Unspecific staining results were excluded by isotype controls (Fig. 4B(i,k)).

High Production of IL-8 upon Expression of the NOD2 SNP13 Variant

Looking at conflicting data regarding the effect of these mutations on the functionality of the protein9–12 we additionally investigated how a modified NOD2 signaling pathway may influence the activation of the downstream mediator NF-κB in vitro. NOD2 negative HEK293T cells were transiently transfected with 2 expression vectors under which 1 expressed the NOD2 WT gene and 1 the NOD2 SNP13 variant. The “empty” vector backbone served as a control. To mimic bacterial translocation cells were specifically stimulated with the NOD2 ligand MDP. NF-κB activation was indirectly measured by quantification of IL-8 secretion. Cells expressing the control plasmid did not show any changes to basal IL-8 secretion following MDP stimulation (Fig. 5; pcDNA3.1). In cells that were transfected with the NOD2 WT vector exclusive overexpression of the transgene resulted in a significantly elevated IL-8 secretion (Fig. 5; NOD2, *P < 0.05). After stimulation with MDP a 1.7-fold induction of IL-8 production versus unstimulated cells was obtained at a dose of 100 ng/ml MDP (§P < 0.05). Higher concentrations did not further increase the secretion of IL-8.

thumbnail image

Figure 5. Synthesis of IL-8 in MDP stimulated HEK293T-cells. HEK293T-cells were stably transfected with pcDNA3.1 (control), pcDNA3.1_NOD2/CARD15 (NOD2/CARD15 WT gene) and pcDNA3.1_SNP13 (NOD2/CARD15 mutation SNP13) and stimulated with increasing concentrations of MDP. The overexpression of both gene variants resulted in an increased IL-8 production. The expression of the WT gene led to a significant induction of IL-8 secretion through stimulation with 100 ng/mL MDP. In cells expressing the mutated variant no induction of IL-8 secretion was obtained. Significance: basal secretion referred to *pcDNA3.1 unstimulated (P < 0.05) and §pcDNA3.1_NOD2/CARD15 unstimulated (P < 0.005); Student's t-test. This figure is representative of 3 independent experiments.

Download figure to PowerPoint

Furthermore, we demonstrated for the first time that in cells producing the truncated SNP13 variant solely overexpression of the mutated transgene led to increased IL-8 production (Fig. 5; *P < 0.05). In cells expressing mutated NOD2_SNP13 IL-8 secretion could not be triggered by stimulation with MDP. Therefore, discrimination between “loss” or “gain” of function seems to be not as easy as suggested in many publications. On the one hand, the lost ability of cells that were transfected with the SNP13 variant to respond to MDP stimulation supports the “loss of function” scenario, whereas the high basal IL-8 secretion rather encourages the “gain of function” theory that is also supported by the patient data.

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

To avoid the invasion of bacteria or specific pathogens from the intestinal lumen into the body the healthy mucosa possesses local protection mechanisms like mucus production, synthesis of antimicrobial peptides, intraluminal IgA-secretion, or gut motility.34, 35 In CD these protective factors are impaired. The intestinal barrier becomes “leaky,” resulting in an invasion of luminal bacteria.

Therefore, in this study we asked 1) whether the transepithelial passage of bacteria is associated with the NOD2 genotype of the patients, 2) whether modified cell–cell contacts may be the route for translocation, and 3) what consequence bacterial stimulation after mucosal translocation has on activation of NOD2-associated proinflammatory signaling pathways.

In the last years several studies focused on the fecal microflora in CD patients and changes in the biodiversity were found.36–38 Investigations regarding the adherens or potential translocation of luminal bacteria are yet limited due to their unspecificity and a lack of quantitative methods.39, 40 Similar to other groups,41, 42 we found that the intestinal mucus layer of CD patients is highly colonized with bacteria, whereas in the mucus layer of normal individuals almost no adherent bacteria were detectable. The abnormal penetration of bacteria into the mucus during CD may be facilitated by a biochemical modification of the mucin layer, a lack of defensins,43 as well as bacterial enzymes that are able to degrade mucin oligosaccharides.44 This is followed by a direct contact between the epithelium and the aggressive microflora, which is normally absent.37 Using FISH technology we effectively detected mucus-associated bacteria. However, in contrast to other reports we could not identify intraepithelial/mucosal bacteria. To obtain more information about potentially translocated bacteria, endotoxin found in the outer membrane of various gram-negative bacteria was localized by immunohistochemical staining. We demonstrate here that accumulation of endotoxin in the intestinal tissue of CD patients reflects the NOD2 genotype: Patients with heterozygous SNP8 or SNP13 variants revealed much stronger endotoxin staining than patients with SNP12 or the WT variant. NOD2 as an MDP sensor protein mediates activation of NF-κB. In this context various in vitro and in vivo experiments proved that NOD2 mutations lead to a loss of MDP recognition.9–11 The NOD2 genotype and the amount of bacterial translocation are likely not directly linked. However, an intracellular accumulation of bacterial products may subsequently lead secondarily to an increased inflammatory reaction. In vitro studies indicated an induction of NOD2 protein synthesis in response to tumor necrosis factor (TNF) stimulation of epithelial cells.45 Similarly, in this study we show that HEK293T-cells are sensitized to bacterial stimuli solely by overexpressing NOD2, as indicated by an increased secretion of IL-8.

Epithelial paracellular permeability as well as barrier function are maintained by tight and adherens junctions.46 In IBD an increase in permeability of the tight junctions mediated by proinflammatory cytokines has been observed. We show here that cell–cell contacts in the intestine of CD patients are morphologically unchanged, but their protein composition is altered. The expression of the tight junction proteins claudin-1, -2, occludin, and ZO-1 was increased in inflamed tissue of CD patients versus controls, claudin-4 was unchanged, and the expression of the adhesion proteins E-cadherin and β-catenin was clearly reduced in the tissue of CD patients.

Claudin-1 and claudin-2 have sealing functions in the epithelial barrier. A change in their expression levels affects the intestinal barrier. An increased expression of claudin-1 was also found in colorectal tumors.47 An induction of claudin-2, as also shown here in CD-associated tissue, was demonstrated to result in the formation of leaky pores20, 48 and an increased permeability. Furthermore, the increased detection of these 2 proteins and their ability to recruit and activate pro-matrix-metallo-proteinase (MMP)-2 or -947, 49, 50 may result in increased proteolysis of extracellular matrix proteins causing tissue remodeling and degradation. Claudin-4 mainly has regulatory functions for transcellular Na+ permeability and no sealing functions.22 Numerous studies showed that occludin plays a major role within the tight junctions. Overexpression of mutated occludin variants resulted in a reduced barrier function of the tight junction.17–19 With respect to this, the finding that occludin was increased in the intestinal tissue of CD patients is a further indicator for the reduced tightness of cell–cell contacts in CD. ZO-1 interacts with the transcription factor ZONAB for regulation of epithelial differentiation and morphogenesis.51 An increased amount of ZO-1 in the tissue may influence the communication between the nucleus and the tight junctions, leading to changes in epithelial differentiation and cell division. In addition, the reduced expression of adherens junction-associated E-cadherin and β-catenin may support the paraepithelial passage of potential noxes.

In summary, these data clearly demonstrate changes in the composition of junctional complex proteins in inflamed CD epithelium that may result in a dysregulated barrier function and loss of cell–cell contacts smooth the way for luminal bacteria to pass the epithelial barrier.

Furthermore, we show that translocated bacterial components are able to activate NF-κB in a NOD2 genotype-specific manner. The effects of NOD2 mutations on intestinal inflammation in CD are still controversial. There exist both “loss” and “gain” of function theories.9–12 In this study we demonstrated in vitro that overexpression of NOD2 in cells expressing the WT gene and remarkably also in cells producing the SNP13 variant activates the NF-κB signaling pathway (IL-8 secretion) without any bacterial stimulus. MDP stimulation increased IL-8 secretion levels in cells that were transfected with the WT NOD2 gene, whereas in cells that produce the truncated SNP13 variant this inducibility was lost. These data correlate with observations from Hugot et al,3 who achieved the same data using lipopolysaccharide (LPS). Our results point out parallels to the regulation of extracellular Toll-like receptors (TLRs) in IEC. Similarly to TLR4,52 NOD2 is only weakly expressed in unstimulated IEC that can be induced by proinflammatory cytokines.53 Due to the demonstrated induction of NOD2 (WT) by bacterial stimulation this protein may additionally fulfil regulatory functions in the gut.

In the intestinal tissue of CD patients an increased activation of NF-κB was found that was further increased in patients with NOD2 variants. NF-κB not only plays a role in inflammatory reactions but also in the regulation of epithelial integrity and intestinal homeostasis.7 In vivo studies showed that an IEC-specific inhibition of NF-κB through NEMO deficiency causes a severe chronic intestinal inflammation that results from epithelial cell apoptosis, reduced expression of antimicrobial peptides, and bacterial translocation.7 In this regard the weakly detected NF-κB activation in the controls reflects a necessity of the intestine to sustain homeostasis. Mainly, a primary defect in innate immunity is presumed that induces proliferation of bacteria and secondarily a NOD2-independent inflammatory reaction via adaptive effector T cells in the host tissue.9 Despite many controversial studies and hypotheses, NOD2 seems to play a key role in the immune defense of the gut against bacterial infections. Mutations in NOD2 advance a predisposition for CD through defect regulation of immune reactions to commensal and/or pathogenic bacteria.

This study for the first time shows that the extent of bacterial translocation occurring in CD is associated with the NOD2 genotype of the patients. Therefore, in patients with NOD2 polymorphism(s) a higher mucosal bacterial translocation additionally triggers intestinal inflammation.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES