Article Text

PDF

Original article
Lymphotoxin β receptor signalling executes Helicobacter pylori-driven gastric inflammation in a T4SS-dependent manner
  1. Raquel Mejías-Luque1,2,
  2. Jessica Zöller3,
  3. Florian Anderl1,
  4. Elena Loew-Gil1,
  5. Michael Vieth4,
  6. Thure Adler5,
  7. Daniela B Engler6,
  8. Sabine Urban6,
  9. Jeffrey L Browning7,
  10. Anne Müller6,
  11. Markus Gerhard1,2,
  12. Mathias Heikenwalder3,8
  1. 1 Institut für Medizinische Mikrobiologie, Immunologie und Hygiene, Technische Universität München, Munich, Germany
  2. 2 German Centre for Infection Research (DZIF), partner site Munich, Munich, Germany
  3. 3 Institut für Virologie, Technische Universität München, Helmholtz Zentrum München, Neuherberg, Germany
  4. 4 Institut für Pathologie, Klinikum Bayreuth, Bayreuth, Germany
  5. 5 Immunology Screen, German Mouse Clinic, Helmholtz Zentrum München, Neuherberg, Germany
  6. 6 Institute of Molecular Cancer Research, University of Zürich, Zürich, Switzerland
  7. 7 Department of Immunobiology, Biogen Idec, Cambridge, Massachusetts, USA
  8. 8 Division of Chronic Inflammation and Cancer, German Cancer Research Center (DKFZ), Heidelberg, Germany
  1. Correspondence to Professor Mathias Heikenwälder, Schneckenburgerstrasse 8, Munich 81675, Germany; heikenwaelder{at}helmholtz-muenchen.de, mheikenwaelder{at}dkfz.de Division of Chronic Inflammation and Cancer (German Cancer Research Center) DKFZ, or Professor Markus Gerhard, Trogerstrasse 30, Munich, 81675, Germany; markus.gerhard{at}tum.de, Institut für Medizinische Mikrobiologie, Immunologie und Hygiene, Technische Universität München, Germany

Abstract

Objective Lymphotoxin β receptor (LTβR) signalling has been implicated in inflammation-associated tumour development in different tissues. We have analysed the role of LTβR and alternative NF-κB signalling in Helicobacter pylori-mediated gastric inflammation and pathology.

Design We analysed several ligands and receptors of the alternative NF-κB pathway, RelB, p52 nuclear translocation and target genes in tissue samples of H. pylori-infected patients with different degrees of gastritis or early gastric tumours by in situ hybridisation, immunohistochemistry, Western blot and real-time PCR analyses. Molecular mechanisms involved in LTβR activation by H. pylori were assessed in vitro using human gastric cancer cell lines and distinct H. pylori isolates. The effects of blocking or agonistically activating LTβR on gastric pathology during challenge with a human pathogenic H. pylori strain were studied in a mouse model.

Results Among the tested candidates, LT was significantly increased and activated alternative NF-κB signalling was observed in the gastric mucosa of H. pylori-infected patients. H. pyloriinduced LTβR–ligand expression in a type IV secretion system-dependent but CagA-independent manner, resulting in activation of the alternative NF-κB pathway, which was further enhanced by blocking canonical NF-κB during infection. Blocking LTβR signalling in vivo suppressed H. pylori-driven gastritis, whereas LTβR activation in gastric epithelial cells of infected mice induced a broadened pro-inflammatory chemokine milieu, resulting in exacerbated pathology.

Conclusions LTβR-triggered activation of alternative NF-κB signalling in gastric epithelial cells executes H. pylori-induced chronic gastritis, representing a novel target to restrict gastric inflammation and pathology elicited by H. pylori, while exclusively targeting canonical NF-κB may aggravate pathology by enhancing the alternative pathway.

  • HELICOBACTER PYLORI - PATHOGENESIS
  • HELICOBACTER PYLORI - GASTRITIS
  • EPITHELIAL CELLS
View Full Text

Statistics from Altmetric.com

Significance of this study

What is already known on this subject?

  • Activation of alternative NF-κB signalling through Lymphotoxin β receptor (LTβR) has been implicated in inflammation-driven tumour development and progression in different tissues.

  • Helicobacter pylori infection leads to early activation of the canonical NF-κB signalling pathway.

  • Polymorphisms in the LTA gene were associated with increased risk for gastric cancer development.

  • Increased levels of chemokines regulated by alternative NF-κB signalling (eg, CXCL13) have been observed in H. pylori-infected subjects.

What are the new findings?

  • H. pylori infection triggers LTβR signalling and alternative NF-κB activation in gastric epithelial cells, contributing to gastric inflammation.

  • Blocking LTβR signalling during H. pylori infection reduces gastric inflammation, while activation of LTβR leads to a more severe pathology.

  • Activated alternative NF-κB exacerbates gastric inflammation even in the absence of H. pylori.

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

  • Our results suggest that blocking H. pylori-induced LTβR signalling might represent an interesting therapeutic approach after antibiotic treatment to control the gastric inflammation elicited by the bacterium. This strategy might be also considered in combination with inhibition of canonical NF-κB, since blocking the canonical pathway alone might have deleterious effects by enhancing the alternative NF-κB pathway. This combinatorial or sequential approach may be able to prevent the well-described progression of gastric pathology even after eradication of H. pylori and should be evaluated in clinical trials in high-risk populations.

Introduction

Chronic gastritis induced by colonisation of the gastric mucosa by Helicobacter pylori can progress to atrophic gastritis, intestinal metaplasia, dysplasia and ultimately to gastric carcinoma.1 The pro-inflammatory environment triggered by the bacterial infection initiates rapid activation of distinct signalling cascades, particularly NF-κB, thereby setting the molecular basis for malignant transformation.2

The severity of the inflammatory response to the bacterium and the risk for gastric cancer (GC) development upon infection have been directly linked to the presence of certain bacterial virulence factors, mainly the cag pathogenicity island (cagPAI), which encodes different components of a type IV secretion system (T4SS) necessary for translocation of Cytotoxin-associated gene A into the host cells.3 Another component of this T4SS, the CagL protein, interacts with and activates the integrin α5β1 on gastric epithelial cells, which initiates the delivery of CagA into the target cells.4 Inside the epithelial cell, CagA interacts with host cell molecules such as the phosphatase SHP-2 and triggers a number of downstream signalling pathways with oncogenic activity. At the same time, inflammatory signalling is induced as a consequence of this interaction. While CagA delivery mainly results in the induction of interleukin (IL)-8, the activation of canonical NF-κB signalling is mainly T4SS dependent,5 highlighting the important role of this pathway in H. pylori-induced pathology.

Activation of canonical NF-κB signalling in gastric epithelial cells and infiltrating immune cells has been well documented during H. pylori infection;2 ,6–8 however, only few data about the functional role of alternative NF-κB pathway in the context of H. pylori-driven pathology are available.

The alternative NF-κB pathway is activated via a distinct set of receptors of the tumour necrosis factor superfamily (TNFSF),9 including BAFFR, Fn14, CD40, herpes virus entry mediator and the lymphotoxin β receptor (LTβR). The latter is triggered by either binding of LTα1β2 heterotrimers expressed by activated T, B, natural killer (NK) cells and lymphoid tissue inducer cells, or LIGHT (homologous to lymphotoxin exhibits inducible expression and competes with HSV glycoprotein D for binding to herpes virus entry mediator, a receptor expressed on T lymphocytes), a TNFSF member expressed on T lymphocytes.9–11 Under pathological conditions, LTs and LIGHT can also be expressed by epithelial cells.12 ,13 LTβR is mainly expressed on epithelial and stromal cells, thereby enabling the communication with lymphocytes. Receptor activation leads to the formation of the RelB/p52 NF-κB complex, which mechanistically relies on inducible processing of the precursor p100 to p52 after activation of NF-κB inducing kinase (NIK). The precursor p100 functions as an IκB-like molecule that keeps RelB located in the cytoplasm.14 Following ligand binding, proteosomal processing of p100 to p52 results in the formation of RelB/p52 heterodimers capable of translocating into the nucleus,15 ,16 inducing the expression of several target genes, including alternative NF-κB-specific CXCL13 and other NF-κB target genes (CXCL10, CCL2 or CCL20) that could be linked to active alternative NF-κB signalling.12 ,17 Interestingly, some of these chemokines have been reported to be upregulated in stomachs of H. pylori-infected subjects.18–21 However, the upstream events leading to the secretion of these chemokines during H. pylori infection are still mostly uncharacterised.

The LT system (LT/LIGHT and LTβR) is crucial for formation and maintenance of tertiary lymphoid organs (TLOs), thereby contributing to cancer development in several tissues.22–25 However, the role of LTβR signalling in H. pylori-driven gastritis and progression to GC remained unknown. Here, we provide evidence that LTβR signalling is essential in executing H. pylori-induced chronic inflammation, a precursor of gastric carcinogenesis.

Material and methods

Cell culture and H. pylori strains

AGS, Kato III and SNU-1 (American Type Culture Collection (ATCC)), AZ521 (Japanese Collection of Research Biosources), St3051 and St2957, NUGC-4 and MKN45 were grown in Dulbecco's Modified Eagle Medium (DMEM)/10% fetal calf serum (FCS) (Gibco).26 Cells were maintained at 37°C in a CO2 atmosphere and routinely tested for Mycoplasma contamination.

H. pylori strains G27,27 SS1, PMSS128 and P1229 were grown on Wilkins–Chalgren blood agar plates and maintained under microaerophilic conditions. G27 isogenic strains G27ΔCagA, G27ΔCagE, G27ΔCagF and G27ΔCagI, G27ΔBabA, G27ΔSabA, G27ΔVacA, G27ΔgGT, G27ΔUreA/B, PMSS1ΔCagE 30 and P12slt (kindly provided by Ivo Boneca) were grown in plates containing either 20 μg/mL of kanamycin or 15 µg/mL of chloramphenicol under the same conditions. H. pylori clinical isolates were obtained from the Städtisches Klinikum München. Two of the seven clinical isolates were CagA positive (see online supplementary figure S4B).

Supplementary data

Histology and immunohistochemistry

Human gastric tissue samples were obtained from the paraffin-embedded tissue bank of the Institut für Pathologie, Klinikum Bayreuth Germany, after approval of the local ethics committee. Patients did not receive antibiotics, nonsteroidal anti-inflammatory drugs (NSAID) or proton pump inhibitor (PPI) drugs before biopsy. Paraffin sections were stained with H&E or various primary and secondary antibodies (see online supplementary table S2). Briefly, sections were incubated in Bond Primary antibody diluent (Leica) and staining was performed on a BOND-MAX immunohistochemistry robot (Leica Biosystems) using BOND polymer refine detection solution for diaminobenzidine (DAB). Image acquisition was performed with a Leica SCN400 slide scanner. The number of cells positively stained for different markers was determined using SlidePathTissueIA image analysis software (Leica) on whole gastric mucosa sections and normalised to tissue area. Pictures for representation were taken by scanning whole tissue sections using a Leica SCN400 slide scanner.

Animal experiments

Female C57BL/6 mice (6–8 weeks) were obtained from Harlan Laboratories (Rossdorf, Germany). All animal studies were conducted in compliance with European guidelines for the care and use of laboratory animals (licenses 63/2008 and 24/2013 to AM) and were approved by the Zurich Cantonal Veterinary Office.

The systemic antagonist mLTβR-mIgG1 (LTβR–Ig) and a control murine monoclonal IgG1 antibody, MOPC-21, were prepared at Biogen Idec and stored at −80°C until used. One week prior to infection, mice were injected intraperitoneally with 100 μg of either substance on day 7 and day 2. On day 0, 2 and 5, mice were infected orogastrically with one dose of 109 colony-forming units (cfu) of the H. pylori strain PMSS1 and further injected once per week with 100 μg of either LTβR–Ig or MOPC-21 for 1 month. H. pylori infection was confirmed by counting cfu from stomach homogenates and efficacy of LTβR–Ig treatment was confirmed by loss of splenic follicular dendritic cell (FDC) networks visualised by anti-FDC-M1 staining. Agonistic activation of the alternative NF-κB signalling pathway was performed by treating mice twice a week with 50 µg ACH6 antibody (kindly provided by Dr Jeffrey Browning/Biogen Idec).

Statistics

Normally distributed data were analysed by one-way analysis of variance (ANOVA) with Dunnett's or Bonferroni's correction for multiple comparisons or Student's t test where indicated, while Mann–Whitney U test or ANOVA Kruskal–Wallis with Dunn's comparison test was used to compare not normally distributed data. Statistical significance was defined when p<0.05.

Results

LTβR ligand expression and activation of alternative NF-κB signalling in H. pylori-induced gastritis and early gastric tumours

We first assessed mRNA expression of several ligands activating the alternative NF-κB signalling (eg, BAFF, CD40L, LTα, LTβ and TWEAK). We observed significant induction of LTβ mRNA in H. pylori-induced gastritis (see online supplementary figure S1A). Thus, we investigated the identity of putative LTβ-responsive cells by in situ hybridisation for LTβR in human gastric tissue samples presenting different degrees of gastric inflammation or early gastric tumours associated with H. pylori infection (see online supplementary table S1). We observed that gastric epithelial cells display LTβR mRNA expression that was not affected by the severity of gastric disease (figure 1A and see online supplementary figure S1B). In the same stomach samples increased levels of LTβ were corroborated on protein level, predominantly expressed by inflammatory cells (T and B cells) infiltrating the gastric mucosa (figure 1B, C and see online supplementary figure S1C). Moreover, we detected LTβ protein expression in LTβR+ epithelial cells in H. pylori-induced gastritis as well as in gastric tumours (figure 1C and see online supplementary figure S1D), suggesting cell autonomous and non-autonomous LTβR signalling. Interestingly, LTβ expression was higher in intestinal-type compared with diffuse-type GCs (see online supplementary figure S1E).

Figure 1

Lymphotoxin β receptor (LTβR) signalling is activated in Helicobacter pylori-induced gastritis and gastric tumours. (A) LTβR mRNA expression detected in human normal gastric mucosa, in mild, moderate and severe gastritis caused by H. pylori infection, as well as in early gastric adenocarcinomas detected by in situ hybridisation. (B) Representative images of LTβ and RelB expression in human gastric tissue samples detected by immunohistochemistry. Arrowheads indicate RelB nuclear expression in epithelial cells. Arrows denote expression in immune cells. (C) Quantification of LTβ+ cells (median and minimum to maximum values) in human normal gastric mucosa (n=20), mild (n=26), moderate (n=24) and severe gastritis (n=17), as well as in early gastric adenocarcinoma (n=23) tissue samples. (D) Representative image of RelB expression in gastric tumours detected by immunohistochemistry. Arrows indicate RelB+ nuclei. (E) Representative images of p52 expression in human gastric samples. Arrowheads indicate p52 nuclear expression in epithelial cells. Arrows denote expression in immune cells. (F) Whiskers (median and minimum to maximum) box plots show relative mRNA levels of CXCL13, CCL17, CXCL10, CCL20, CCL2 and A20 detected in human normal gastric mucosa (n=11) and H. pylori-induced gastritis (n=11). *p≤0.05, **p≤0.01, ***p≤0.001. (analysis of variance Kruskal–Wallis, Dunn's multiple comparison test.)

Alternative NF-κB signalling activation was assessed by RelB and p52 nuclear translocation in epithelial cells, which was higher in gastritis and in gastric tumour samples, when compared with controls (figure 1B, D, E and see online supplementary figure S1F).

Furthermore, we analysed expression of LTβR downstream target genes in H. pylori-induced moderate gastritis samples (figure 1F). Enhanced expression of CXCL13 and several other chemokines related to LTβR signalling (CCL17, CCL20, CXCL10 and CCL2) was found in patients with H. pylori-associated gastritis compared with controls (figure 1F). Furthermore, increased expression of A20, an endogenous inhibitor of canonical NF-κB, corroborated concomitant canonical NF-κB signalling activation in H. pylori infection.

H. pylori activates LTβR signalling and alternative NF-κB in human GC cells

To investigate whether H. pylori activates LTβR signalling in gastric epithelial cells and to define the molecular mechanisms involved, we initially characterised a panel of human GC cell lines. Except AZ521, all cell lines analysed expressed LTβR (see online supplementary figure S2A). Interestingly, constitutive p100 processing was detected in some cell lines including AZ521 and NUGC-4 (see online supplementary figure S2B). Constitutive activation of alternative NF-κB signalling was not LTβR-ligand dependent, since under basal conditions none of the cell lines studied expressed LTα, LTβ or LIGHT (see online supplementary figure S2C). In parallel, we also analysed canonical NF-κB signalling. All GC cell lines analysed expressed TNFR1 at different levels, while canonical NF-κB activation—examined by pRelA—was observed in most of the cell lines independently of basal expression of TNFα (see online supplementary figure S2C,D). Based on these results, we selected the cell lines MKN45 and St3051, which express LTβR at different levels but lack constitutive activation of the alternative NF-κB pathway, for further experiments.

We first infected these GC cell lines with the H. pylori strain G27 at different multiplicity of infection (MOI) for 12 h. H. pylori stimulation at low MOI (2–10) resulted in dose-dependent induction of p100 expression and cleavage to p52 (figure 2A and see online supplementary figure S2E). At higher infectious doses (MOI 20–50) we found a reduction in p100 processing, which we observed to be a consequence of cell death occurring at high MOI (data not shown). Thus, we selected MOI 10 for further investigations. As expected, H. pylori infection also resulted in a dose-dependent phosphorylation of p65 (see online supplementary figure S2F), demonstrating activation of canonical NF-κB.

Figure 2

Helicobacter pylori activates alternative NF-κB signalling. (A) MKN45 and St3051 cells were infected with H. pylori G27 strain at the indicated MOI and p100 processing to p52 was analysed by western blot. The agonist of lymphotoxin β receptor (LTβR) signalling BS-1 (0.5 µg/mL) was used as positive and β-actin as loading control. One representative blot and band quantification showing the ratio between p100 and p52 levels from three independent experiments is shown. OD, optical density (arbitrary units). (B) MKN45 GC cells were infected with H. pylori G27 strain at MOI 10 for different time points and processing of p100 to p52 was analysed by western blot. The agonist BS-1 (0.5 µg/mL for 12 h) was used as positive control. β-actin was used as loading control. One representative blot is shown. (C) p100 processing to p52 detected by western blot in MKN45 cells infected with different clinical isolates (I–VII) at MOI 10 for 12 h. One representative of three independent experiments is shown. (D) mRNA expression levels (mean±SD) of LTα and LTβ upon H. pylori G27 strain infection at MOI 10 for 3 (n=3), 6 (n=4) and 12 h (n=4) in MKN45 and St3051 cells (ANOVA, Dunnett's multiple comparison test). (E) CXCL13, CXCL10, CCL20 and A20 mRNA expression upon H. pylori G27 strain infection at MOI 10 for 12 h. The agonist BS-1 (0.5 µg/mL) was used as positive control. Ct values were normalised to Glyceraldehyde-3-phosphate dehydrogenase (GADPH). Results (mean±SD) from three independent experiments are presented as fold induction. *p≤0.05, ***p≤0.001. (Control and H. pylori-infected cells were compared using Student's t test.)

We next examined the kinetics of alternative NF-κB signalling activation in a time course study. p100 expression was upregulated 6 h post-infection, while its processing was apparent earliest after 8 h (figure 2B and see online supplementary figure S2G), reaching a plateau of p52 at 10 h, which was maintained even at 24 h post-infection. Using different clinical H. pylori isolates we confirmed p100 processing (figure 2C).

We further investigated whether p100 processing to p52 correlated with LTβR-ligand expression. We detected LT mRNA upregulation already 3 h post-infection (figure 2D), which was approximately 4–5 h before first signs of p100 processing (figure 2B). Notably, expression of LTβ was induced at a much higher level when compared with LTα in both cell lines, while LTβR expression did not change (see online supplementary figure S2H).

Next, we analysed the expression of different chemokines regulated by LTβR signalling at 12 h post-infection. Similar to results obtained from patients, significant induction of CXCL13, CXCL10 and CCL20 in response to H. pylori infection was found in GC cells (figure 2E and see online supplementary figure S2I). A20 levels were also significantly enhanced (figure 2E and see online supplementary figure S2I).

H. pylori-induced LTβR ligand upregulation activates alternative NF-κB in a secondary loop post-infection

H. pylori can interact with host receptors, for example, MUC5AC31 or TNFR1,32 which prompted us to investigate whether H. pylori directly interacted with the LTβR and triggered downstream signalling. However, no interaction between LTβR and H. pylori was detected via immunoprecipitation (see online supplementary figure S3A).

As H. pylori-induced LT expression precedes p100 processing (shown in figure 2B, D), we hypothesised that LTs might activate LTβR signalling in a secondary loop post-infection. Thus, we scavenged LTα1β2 and LIGHT using an LTβR–Ig fusion protein. LTβR–Ig treatment during H. pylori infection did not affect LTα and LTβ mRNA levels (see online supplementary figure S3B and data not shown). However, LTβR–Ig treatment suppressed p100 processing, whereas the isotype control MOPC-21 did not (figure 3A). This suggests that LTβR ligands induce activation of LTβR signalling in response to H. pylori infection.

Figure 3

Activation of lymphotoxin β receptor (LTβR) signalling is induced by secreted LT and LIGHT (homologous to lymphotoxin exhibits inducible expression and competes with HSV glycoprotein D for binding to herpes virus entry mediator, a receptor expressed on T lymphocytes) in response to Helicobacter pylori infection. (A) MKN45 cells were incubated with LTβR–Ig or the isotype control MOPC21 for 1 h before H. pylori infection at MOI 10. p100 processing to p52 was evaluated by western blot after 12 h infection. β-actin served as loading control. One representative blot and quantification of three independent experiments (mean±SD) are shown. (B) A20, LTα and LTβ mRNA levels were analysed by real-time PCR in MKN45 cells after 12 h infection with the H. pylori strain G27 (MOI 10). Cells were incubated with the inhibitor of human IκB kinase 2 (IKK2) TPCA-1 1 h prior to infection. Ct values were normalised to GADPH. Results expressed as mean±SD from three independent experiments are shown. (C) MKN45 cells were incubated with TPCA-1 for 1 h before infection with the H. pylori strain G27 (MOI 10). Cell lysates were obtained after 12 h infection and levels of p100/p52 were assessed by western blot. β-actin was used as loading control. One representative blot and quantification (mean±SD) of three independent experiments are shown. (D) LIGHT mRNA expression levels after infection of MKN45 cells with the H. pylori strain G27 at MOI 10 for 12 h. Cells were incubated with TPCA-1 for 1 h before infection. Results (mean±SD) from three independent experiments are shown. (E) CXCL13 and CCL22 mRNA levels in MKN45 cells infected with H. pylori G27 (MOI 10, 12 h). Where indicated, cells were incubated with TPCA-1 1 h before infection. Ct values were normalised to GADPH. Results (mean±SD) from three independent experiments are shown. (F) Whiskers (median and minimum to maximum) box plot showing relative mRNA levels of LIGHT in gastric H. pylori-induced and H. pylori-negative gastritis (n=11). (G) Representative images of LIGHT expression in human gastric tissue samples detected by immunohistochemistry. Arrows indicate positive cells. *p≤0.05, **p≤0.01, ***p≤0.001. ((A–E) analysis of variance (ANOVA), Bonferroni's multiple comparison test; (F) ANOVA Kruskal–Wallis, Dunn's multiple comparison test.)

LT expression can be regulated by canonical NF-κB.33 Therefore, we analysed whether this pathway also regulates LT expression in GC cells. Upon TNFα stimulation, LTβ expression was significantly induced, while only minor effects were detected on LTα mRNA levels. As a control, A20 was induced (see online supplementary figure S3C). We confirmed regulation of LTα and LTβ by canonical NF-κB upon H. pylori infection by using an inhibitor of human IκB kinase 2 (IKK2) [5-(p-Fluorophenyl)-2-ureido]thiophene-3-carboxamide (see online supplementary figure S3D). Reduced levels of A20 were detected after TPCA-1 treatment (figure 3B), while the expression of LTα was slightly affected. Remarkably, LTβ levels were significantly reduced (figure 3B), confirming that LTβ expression is regulated by canonical NF-κB in gastric cells. When assessing p100 processing we observed lower levels of p100 in the presence of TPCA-1 (figure 3C and see online supplementary figure S3E). Remarkably, cleavage of p100 to p52 was still found to a similar extent when compared with untreated controls, suggesting a canonical NF-κB-independent LTβR ligand to drive p100 processing in the absence of LTs.

Besides LTs, LIGHT was reported to activate LTβR. Therefore, we studied LIGHT expression in gastric cells post-H. pylori challenge. While no or minor upregulation was detected after activation of canonical NF-κB using TNFα (see online supplementary figure S3C), we observed induction of LIGHT expression upon H. pylori infection (figure 3D), which was even further enhanced when blocking canonical NF-κB by TPCA-1. Likewise, expression of the LTβR signalling-specific target gene CXCL13 was increased when blocking canonical NF-κB (figure 3E), while expression of the canonical pathway target CCL22 was completely abrogated (figure 3E). This suggests that canonical NF-κB negatively regulates LTβR activation through repressing LIGHT in H. pylori infection. Notably, expression of other TNFSF ligands such as BAFF or CD40L was not detected, neither under basal conditions nor after H. pylori infection or TPCA-1 treatment in GC cells (data not shown). In line, the expression of TWEAK was not upregulated upon bacterial infection or inhibition of canonical NF-κB (see online supplementary figure S3F), supporting the fact that LIGHT may be responsible for inducing p100 processing after blocking canonical NF-κB pathway. In parallel, H. pylori-triggered upregulation of LIGHT mRNA expression was confirmed in human gastric biopsies with moderate gastritis (figure 3F). LIGHT protein expression was mainly increased in epithelial cells of human gastric tissue samples presenting different degrees of gastric inflammation and in early gastric tumours associated with H. pylori infection (figure 3G).

H. pylori cagPAI but not CagA is required for LTβR-mediated activation of alternative NF-κB

To determine whether the expression of LTs and subsequent activation of LTβR signalling was influenced by H. pylori virulence factors, we infected epithelial cells with different mutant strains. A functional T4SS was necessary to induce LT expression, since cells infected with the mutant strain G27ΔCagE expressed lower levels of LT compared with cells infected with the wild-type strain (figure 4A). Notably, LIGHT expression did not depend on a functional T4SS, confirming canonical NF-κB-independent regulation of LIGHT (see online supplementary figure S4A).

Figure 4

Activation of alternative NF-κB is independent of CagA. (A) mRNA expression levels of lymphotoxin (LT)α and LTβ in MKN45 and St3051 cells after infection with wild-type Helicobacter pylori G27 strain and the isogenic mutant strains G27ΔCagA and G27ΔCagE. Ct values were normalised to GADPH. Results (mean±SD) from four independent experiments are shown (analysis of variance, Bonferroni's multiple comparison test). (B) MKN45 cells were infected with the H. pylori wild-type strains G27, PMSS1 and SS1 and the G27 isogenic mutant strains G27ΔCagA, G27ΔCagE, G27ΔBabA, G27ΔSabA, G27ΔVacA, G27ΔgGT and G27ΔUreA/B at MOI 10 for 12 h. p100 processing to p52 was detected by western blot. β-actin was used as loading control. One representative blot and quantification of three independent experiments is shown. *p≤0.05, **p≤0.01, ***p≤0.001. (G27 wt and G27ΔCagE-infected cells were compared using Student's t test.)

When analysing p100 processing, we observed that lack of CagA left p100 levels and p100 cleavage unaltered, while bacteria devoid of functional cagPAI (SS1, G27ΔCagE, G27ΔCagF, G27ΔCagI) did not induce p100 expression and processing (figure 4B, see online supplementary figure S4B, C). In line, clinical isolates lacking CagA induced p100 processing (figure 2E and see online supplementary figure S4D). Thus, CagA is not necessary for LT-mediated activation of alternative NF-κB signalling in gastric cells.

Another described H. pylori effector molecule translocated into target host cells through the T4SS is peptidoglycan.34 To determine whether peptidoglycan could be involved in LT-mediated activation of alternative NF-κB, we infected cells with a strain deficient in lytic transglycosylase activity (P12slt), which is involved in bacterial muropeptide release. Processing of p100 to p52 was reduced in cells infected with the slt-deficient strain (see online supplementary figure S4E), suggesting that peptidoglycan may be involved in activation of alternative NF-κB.

A functional T4SS is necessary to induce ligand expression and alternative NF-κB activation in vivo

In order to validate our results in vivo, C57BL/6 mice were infected with the pathogenic H. pylori strain PMSS128 ,30 and analysed for LTα, LTβ and LIGHT mRNA expression. One month post-infection, mice presented significantly increased gastric levels of all LTβR ligands (figure 5A), confirming a correlation between H. pylori infection and LT/LIGHT upregulation in vivo. Since our in vitro data suggested the involvement of a functional T4SS in alternative NF-κB activation, we also infected mice with PMSS1 strain deficient for CagE. mRNA levels of LTα, LTβ and LIGHT were lower in mice infected with H. pylori PMSS1ΔCagE when compared with mice infected with the wild-type strain (figure 5B), even though they were colonised at higher levels (figure 5C). In addition, mice infected with CagE-deficient bacteria showed less gastric inflammation (figure 5D), as previously reported.30 Specifically, lower infiltration of T and B cells was detected in the stomach of mice infected with PMSS1ΔCagE when compared with mice infected with wild-type bacteria (figure 5E). Infection of mice with wild-type PMSS1 induced activation of alternative NF-κB in gastric epithelial cells, as detected by nuclear translocation of RelB (figure 5E). Notably, lack of CagE correlated with reduced nuclear RelB translocation, confirming that activation of alternative NF-κB signalling by H. pylori requires a functional T4SS. Moreover, the expression of canonical as well as alternative NF-κB signalling target genes such as A20, murine IL-8 homologue KC, CXCL13, CXCL10, CCL2 and CCL20 was reduced in the gastric mucosa of mice infected with bacteria deficient for CagE (figure 5F).

Figure 5

Helicobacter pylori activates alternative NF-κB in vivo in a type IV secretion system (T4SS)-dependent manner. (A) Lymphotoxin (LT)α, LTβ and LIGHT (homologous to lymphotoxin exhibits inducible expression and competes with HSV glycoprotein D for binding to herpes virus entry mediator, a receptor expressed on T lymphocytes) mRNA levels in mice infected for 1 month with the H. pylori strain PMSS1. Ct values were normalised to GADPH. (B–F) Mice were infected for 1 month with the H. pylori strain PMSS1 or an isogenic strain lacking CagE. (B) LTα, LTβ and LIGHT mRNA levels in infected mice. Ct values were normalised to GADPH. (C) Colony-forming units (cfu) per mg of stomach. (D) Inflammatory score. Activity and chronicity of infection was evaluated in antrum and corpus of murine stomach samples according to the updated Sydney system. (E) Murine stomach samples were stained for CD3, B220 and RelB by immunohistochemistry. Representative pictures are shown. CD3+ and B220+ cells and RelB+ nuclei were calculated per area of tissue (mm2). (F) Relative mRNA expression levels of A20, KC, CXCL13, CXCL10, CCL20 and CCL2 in the stomach of H. pylori-infected mice. Ct values were normalised to GADPH. *p≤0.05, **p≤0.01, ***p≤0.001. ((A), (B) and (F) (Mann–Whitney U test); (E) analysis of variance Kruskal–Wallis, Dunn's multiple comparison test.)

Blocking LTβR signalling in vivo reduces H. pylori-induced gastric inflammation

We next assessed whether LTβR signalling is correlatively or causally linked with gastric inflammation and pathology triggered by H. pylori in vivo. Thus, we treated H. pylori-infected mice with LTβR–Ig (see online supplementary figure S5A) to study the effects of blocking LTβR signalling on bacterial colonisation and gastric inflammation induced by H. pylori. Efficacy of LTβR–Ig treatment was confirmed by staining LTβR-dependent FDCs in spleen25 (see online supplementary figure S5B). Interestingly, LTβR–Ig-treated mice presented a higher gastric bacterial load when compared with MOPC-21-injected mice (figure 6A), suggesting that LTβR signalling is important to control bacterial burden. Higher H. pylori colonisation was accompanied by less gastric inflammation (figure 6B), indicating that LTβR signalling is an important driver of gastric pathology induced by the bacterium. Remarkably, LTβR–Ig-injected mice displayed less CD3+ and CD4+ cells infiltrating the stomach (figure 6C and see online supplementary figure S5C). In addition, less B220+ cells were detected upon LTβR–Ig treatment in the gastric mucosa of infected mice (figure 6C), demonstrating a reduced inflammatory response after blocking LTβR signalling. Moreover, mice treated with LTβR–Ig presented lower nuclear translocation of RelB in gastric epithelial cells (figure 6C), confirming treatment efficacy.

Figure 6

Blocking of lymphotoxin β receptor (LTβR) signalling during Helicobacter pylori infection reduces gastric inflammation. (A–D) C57BL/6 mice infected with the H. pylori strain PMSS1 were injected weekly with 100 µg of LTβR–Ig or the isotype control MOPC-21 (Control–Ig). Mice were sacrificed after 1-month infection. Dots represent data of individual mice pooled from two independent experiments. (A) Colony-forming units (cfu) per mg of stomach. (B) Inflammatory score. Activity and chronicity of infection was evaluated in antrum and corpus of murine stomach samples according to the updated Sydney system. (C) Murine stomach samples were stained for CD3, B220 and RelB by immunohistochemistry. Representative pictures are shown. CD3+ and B220+ cells and RelB+ nuclei were calculated per area of tissue (mm2). (D) Relative mRNA expression levels of KC, CXCL10, CCL20 and CCL2 in the stomach of control or H. pylori-infected mice treated with LTβR–Ig or MOPC-21. Ct values were normalised to GADPH. *p≤0.05, **p≤0.01, ***p≤0.001. (analysis of variance Kruskal–Wallis, Dunn's multiple comparison test.)

We further assessed whether blocking LTβR signalling upon H. pylori infection altered the expression levels of several chemokines. We observed decreased levels of KC, CXCL10, CCL20 and CCL2 in the stomach of mice injected with LTβR–Ig (figure 6D), correlating with the lower inflammation found in the gastric mucosa.

Activation of LTβR signalling enhances gastric pathology induced by H. pylori

We next determined the effect of enhanced LTβR signalling during H. pylori infection. We injected H. pylori-infected mice with ACH6, a murine agonist of LTβR35 (figure 7 and see online supplementary figure S5A). ACH6-triggered activation of LTβR led to a decrease in gastric bacterial load (figure 7A), accompanied by higher gastric inflammation (figure 7B). We observed slightly lower levels of CD3+ cells in H. pylori-infected animals treated with the agonist (figure 7C), while no major differences were detected in CD4+ or F480+ macrophages (see online supplementary figure S6A). Interestingly, animals treated with the agonist presented more B cells than infected control mice (see online supplementary figure S6B). In line, the expression of CXCL13 was enhanced in mice injected with ACH6 (figure 7D). We also observed increased expression of the LTβR target genes LIGHT, ICAM—involved in leucocyte recruitment—and CCL2 (figure 7D), while KC and CCL20 levels remained unchanged (see online supplementary figure S6C).

Figure 7

Agonistic activation of lymphotoxin β receptor (LTβR) signalling during Helicobacter pylori infection increases gastric inflammation. (A–D) C57BL/6 mice infected with the H. pylori strain PMSS1 were injected weekly with 50 µg of the agonist of LTβR signalling ACH6. Mice were sacrificed after 1-month infection. Horizontal bars indicate medians. (A) Colony-forming units (cfu) per mg of stomach. (B) Gastric inflammatory score evaluated according to the updated Sydney system. (C) Representative pictures of murine stomach samples stained for CD3, B220 and RelB. CD3+ and B220+ cells and RelB+ nuclei were calculated per area of tissue (mm2). (D) Relative mRNA expression levels of CXCL13, ICAM, CCL2 and LIGHT (homologous to lymphotoxin exhibits inducible expression and competes with HSV glycoprotein D for binding to herpes virus entry mediator, a receptor expressed on T lymphocytes) in stomach samples. Ct values were normalised to GADPH. *p≤0.05, **p≤0.01, ***p≤0.001. ((A) (Mann–Whitney U test); (B–D) analysis of variance Kruskal–Wallis, Dunn's multiple comparison test.)

Discussion

Activation of LTβR triggers and maintains TLOs during chronic infection.36 Notably, lymphoid neogenesis and TLO formation were reported as a histological hallmark of chronic H. pylori infection,37–39 suggesting a possible role of the LTβR signalling pathway in H. pylori-driven pathology.

The involvement of NIK in H. pylori-induced NF-κB activation was reported in gastric epithelial cells,40 ,41 and p100/p52 processing upon H. pylori infection was reported in B lymphocytes.42 However, in the latter study, activation of alternative NF-κB reflected by p100 processing was not found in the GC cell line AGS. In addition, p100/p52 staining of human gastric samples from H. pylori-infected subjects showed only cytoplasmic staining that was weaker compared with infiltrating lymphocytes, leading to the conclusion that the alternative NF-κB pathway was not activated in epithelial cells upon H. pylori infection. In our study, we have detected p100 processing in different human gastric cell lines. Importantly, activation of the alternative NF-κB pathway was corroborated in epithelial and immune cells in gastritis and gastric tumour samples from H. pylori-infected patients. Likewise, mice infected with H. pylori showed increased nuclear translocation of RelB in gastric epithelial cells, demonstrating that H. pylori activates alternative NF-κB in vitro and in vivo.

Canonical and alternative NF-κB signalling are closely related and crosstalk at different levels, for example, expression of p100 and LTβ, have been reported to be regulated by canonical NF-κB.33 ,43–45 In line, we detected enhanced expression of p100 and LTs upon H. pylori infection, which was reduced when activation of canonical NF-κB was blocked. However, p100 processing was still observed, indicating that p100 overexpression is insufficient to induce its processing.46 It rather necessitates a specific activation by LTs through LTβR upon infection. Further, p100 processing upon canonical NF-κB blockade implied another canonical NF-κB-independent, LTβR ligand to trigger this pathway, which we found to be LIGHT. In contrast, treatment with LTβR–Ig, which scavenges all LTβR ligands, fully disables alternative NF-κB signalling. Therefore, our results demonstrate a tight crosstalk between canonical and alternative pathway during H. pylori infection, since activation of canonical NF-κB regulates the expression of ligands important for activation of the alternative pathway.

We observed that the T4SS played a crucial role in controlling the expression of LTs and subsequent activation of LTβR signalling in vitro and in vivo, whereas the presence of CagA was dispensable. These results are in agreement with previous reports showing that H. pylori CagA-deficient but T4SS-proficient strains induce NF-κB activation almost as efficiently as wild-type bacteria,5 ,47 while the exact mechanism of T4SS-dependent NF-κB activation is not fully understood.48 The presence of CagA has been linked to more severe gastric pathology in Western countries,49 yet, our results using CagA-negative H. pylori isolates indicate that important inflammatory signalling cascades, such as alternative NF-κB, can still be activated and contribute to gastric inflammation in the absence of CagA.

Our in vivo experiments blocking or activating LTβR signalling in mice during H. pylori infection showed an important role for this pathway in the control of the inflammatory response towards the bacterium. Reduction of chemokine expression in the stomach of LTβR-treated animals confirmed LTβR inhibition at the site of infection. Depletion of FDCs and secondary lymphoid organs was observed, suggesting an additional systemic extra-gastric effect. However, our experiments using the agonist of LTβR (ACH6) refute this possibility, since no effects on FDCs or lymph nodes have been reported with this antibody,35 but we still observed a reversed gastric phenotype (reduced bacterial load and increased gastric inflammation) compared with LTβR inhibition.

Previous studies using transgenic mouse models suggested the involvement of alternative NF-κB in gastric pathology: Knockout mice lacking the terminal ankyrin domain of NF-κB2 and constitutively expressing NF-κB2/p52 presented gastric hyperplasia in the absence of Helicobacter infection.50 In line, infection experiments of Nfkb2−/− mice suggested NF-κB2-mediated signalling to be required for the development of H. felis-induced gastric pathology.51 However, the specific pathways upstream of NF-κB2 remained elusive. Our in vivo studies using human pathogenic H. pylori reveal an essential function of LTβR signalling in gastric pathology upstream of NF-κB2.

Taken together, our data indicate that both NF-κB pathways are activated in the course of H. pylori infection. Our in vitro data show that activation of the canonical NF-κB signalling precedes alternative NF-κB, regulating a feed-forward loop through induction of LT expression (see online supplementary figure S7A). Ligand engagement to LTβR expressed on epithelial cells triggers p100/p52 processing and subsequent activation of the expression of chemokines important for immune cell recruitment (CXCL13, CXCL10 and CCL20). Recruitment of inflammatory cells (CD4+ T cells and B cells) to the stomach, which in turn also express LTs and LIGHT, leads to a feedback loop, maintaining H. pylori-induced gastric inflammation (see online supplementary figure S7B). This is confirmed by our observations in human gastric tissue, where expression of LTs and target genes is linked to the severity of the lesions in the presence of H. pylori.

Different studies have suggested inhibition of canonical NF-κB as a therapeutic strategy to reduce H. pylori-triggered gastric inflammation. Natural substances as capsaicin,52 antiulcer drugs as ecabet sodium53 or probiotics54 have been proven to inhibit in vitro NF-κB and IL-8 secretion in gastric epithelial cells. Moreover, inhibition of IKK2 in Mongolian gerbils was related to a reduced infiltration of neutrophils and mononuclear cells in the stomach, while no differences in bacterial load were detected,55 suggesting that inhibition of canonical NF-κB was not sufficient to control the infection in this model. In light of our results showing that blocking of canonical NF-κB reduces LT expression but induces LIGHT and CXCL13 upregulation upon H. pylori challenge, it is tempting to speculate that inhibition of canonical NF-κB might indeed not represent the best therapeutic option to reduce gastric inflammation and pathology elicited by H. pylori. On the contrary, it might imply other adverse effects that have not been considered until now, such as activation of LTβR signalling through LIGHT and concomitant exacerbation of inflammation.

This, in context with the self-sustaining inflammatory alternative NF-κB signalling loop, may aggravate the deleterious inflammatory effects induced by chronic H. pylori infection, especially in patients presenting with a high degree of gastritis. Having observed that blocking of LTβR signalling during H. pylori infection in mice reduces gastric inflammation and infiltration of inflammatory cells, we propose that such strategy could be considered as a supplement after antibiotic treatment or concomitant with inhibition of canonical NF-κB signalling.

Acknowledgments

The authors would like to thank Daniel Kull, Ruth Hillermann, Olga Seelbach and Andreas Wanisch for valuable technical assistance and Ivo Boneca for providing bacteria.

References

View Abstract

Footnotes

  • RM-L, JZ, MG and MH contributed equally.

  • Contributors Study concept and design: RM-L, JZ, MG, and MH. Acquisition of data: RM-L, JZ, FA, EL-G, TA, DBE and SU. Analysis and interpretation of data: RM-L, JZ, MG and MH. Drafting of the manuscript: RM-L. Critical revision of the manuscript: RM-L, JZ, AM, MG and MH. Administrative, technical or material support: MV, JLB and AM. Study supervision: MG and MH.

  • Funding This work was supported by an ERC starting grant (Liver Cancer Mechanism), the Stiftung für Biomedizinische Forschung (Hofschneider Foundation), the Helmholtz foundation through a young investigator group (YIG), the SFB-TR 36 and the Graduierten Kolleg GRK-42 to MH, and the Swiss National Science foundation (310030-143609 and BSCGIO 157841/1) to AM.

  • Competing interests None declared.

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

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.