ABSTRACT
Infection with Helicobacter pylori cag pathogenicity island (cagPAI)-positive strains is associated with more destructive tissue damage and an increased risk of severe disease. The cagPAI encodes a type IV secretion system (TFSS) that delivers the bacterial effector molecules CagA and peptidoglycan into the host cell cytoplasm, thereby inducing responses in host cells. It was previously shown that interactions between CagL, present on the TFSS pilus, and host α5β1 integrin molecules were critical for CagA translocation and the induction of cytoskeletal rearrangements in epithelial cells. As the α5β1 integrin is found in cholesterol-rich microdomains (known as lipid rafts), we hypothesized that these domains may also be involved in the induction of proinflammatory responses mediated by NOD1 recognition of H. pylori peptidoglycan. Indeed, not only did methyl-β-cyclodextrin depletion of cholesterol from cultured epithelial cells have a significant effect on the levels of NF-κB and interleukin-8 (IL-8) responses induced by H. pylori bacteria with an intact TFSS (P < 0.05), but it also interfered with TFSS-mediated peptidoglycan delivery to cells. Both of these effects could be restored by cholesterol replenishment of the cells. Furthermore, we demonstrated for the first time the involvement of α5β1 integrin in the induction of proinflammatory responses by H. pylori. Taking the results together, we propose that α5β1 integrin, which is associated with cholesterol-rich microdomains at the host cell surface, is required for NOD1 recognition of peptidoglycan and subsequent induction of NF-κB-dependent responses to H. pylori. These data implicate cholesterol-rich microdomains as a novel platform for TFSS-dependent delivery of bacterial products to cytosolic pathogen recognition molecules.
Helicobacter pylori strains harboring the cytotoxin-associated gene pathogenicity island (cagPAI) are associated with the production of higher levels of interleukin-8 (IL-8), more destructive tissue damage, and an increased risk of severe disease (8, 13, 15, 43, 44). The cytotoxin-associated gene A product (CagA) is translocated into eukaryotic epithelial cells by the cagPAI-encoded type IV secretion system (TFSS) (7, 40). Once intracellular, CagA localizes on the inner surface of the plasma membrane and becomes phosphorylated on tyrosine residues by Src family kinases (5). The phosphorylated CagA subsequently induces a signaling cascade, ultimately resulting in the development of a “cell-scattering” phenotype in gastric epithelial cells (5, 50).
A large body of evidence from previous studies has suggested that CagA is not essential for IL-8 secretion from gastric epithelial cells, as ΔcagA mutants retained the ability to induce IL-8 production (11, 15, 18, 52). More recently, however, CagA was found to play a role in the potentiation of IL-8 induction upon extended coculture (>36 h) of epithelial cells with cagPAI-positive H. pylori strains (10). Another report recently described the rapid induction of nuclear factor kappa B (NF-κB) activation in epithelial cells via CagA (35). The reason for the lack of concordance of these more recent findings with those of previous works is not known, but it may be partly attributable to H. pylori strain differences.
Notwithstanding the ambiguity regarding the role of CagA in the induction of proinflammatory responses in epithelial cells, it is broadly accepted in the field that the H. pylori TFSS is essential for induction of NF-κB-dependent responses in these cells. Our group previously reported that induction of such responses was dependent on TFSS delivery of cell wall peptidoglycan (PG) to host cells (57). Once intracellular, PG was proposed to be detected by a cytosolic pathogen recognition molecule (PRM), nucleotide-binding oligomerization domain protein (NOD1) (19, 57), resulting in activation of NF-κB and the induction of IL-8 secretion by epithelial cells (57). Nevertheless, the exact mechanism by which PG may enter the host cell via the actions of the TFSS, so as to initiate these NOD1-dependent responses, has remained elusive.
It has recently been suggested that virulence factors of H. pylori associate with cholesterol-rich microdomains of the plasma membrane, commonly termed lipid rafts (34). These domains not only are enriched in cholesterol but also contain sphingolipids and proteins (53) and have been reported to be sites utilized by bacteria to interact with host cells (1) or as portals of entry by which bacteria enter these cells (29, 33). Wunder and colleagues demonstrated colocalization of H. pylori bacteria with GM1 ganglioside, a characteristic component of cholesterol-rich microdomains, and established a role for cholesterol in the attraction of H. pylori to host cells (62). H. pylori was reported to migrate toward and acquire exogenous cholesterol from the plasma membranes of host epithelial cells (62). Furthermore, disruption of cholesterol-rich microdomains using cholesterol-depleting agents such as methyl-β-cyclodextrin (MβCD) was shown to significantly reduce the internalization of the vacuolating cytotoxin (VacA) into target cells (49) and to inhibit the ability of the toxin to induce cell vacuolation (31, 48). Moreover, Lai et al. reported that cholesterol-rich microdomains are also crucial for efficient TFSS-mediated CagA translocation by H. pylori, as CagA-induced cellular responses, including the cell-scattering phenotype and IL-8 production, were found to be reduced in MβCD-treated cells (34).
The translocation and phosphorylation of CagA are triggered by direct interaction of the H. pylori TFSS with the α5β1 integrin, which is found on the surfaces of gastric epithelial cells (32). A number of viruses and bacteria are known to use integrin receptors to adhere to and invade host cells (9, 24). Indeed, Kwok et al. demonstrated that the H. pylori adhesin CagL is targeted to the surface of the secretory pilus encoded by the cagPAI, where it binds to and activates α5β1 integrin (32). This interaction triggers the delivery of CagA into the host as well as subsequent Src kinase activation (32). Given that α5β1 associates with cholesterol-rich microdomains (27, 37) and that these membrane microdomains have been shown to be important for CagA effects on host responses (34), we speculated that H. pylori strains may exploit a similar mechanism to induce proinflammatory responses in epithelial cells via delivery of PG to NOD1. We now show for the first time that H. pylori PG translocation into epithelial cells and the subsequent activation of NF-κB-dependent responses (a characteristic of NOD1 activation) are dependent on lipid rafts and more specifically on α5β1 integrin.
MATERIALS AND METHODS
Bacterial strains and isogenic mutants. H. pylori strains 251 (45), B128 7.13 (25), and P1 (7) and the isogenic 251 ΔcagA (57), 251 ΔcagM (57), 251 ΔlysA (57), 251 ΔlysA cagM (57) B128 7.13 ΔcagM (57), and P1 ΔcagA (5, 10, 32) mutants have been described previously. Briefly, H. pylori 251 ΔcagPAI (28) was constructed by natural transformation (59) using the DNA construct pJP46 (39), containing a kanamycin resistance cassette. The genotype of the H. pylori 251 ΔcagPAI mutant was confirmed by PCR using the primers 5′-CATCAGCTATACAAAGTGAAAACG-3′ and 5′-CATTCTGGCGATGCTATTGTG-3′, which anneal to the cagPAI flanking genes HP0519 and HP0549. H. pylori B128 7.13 ΔcagM was verified by PCR using the primers 5′-ACAAATACAAAAAAGAAAAAGAGGC-3′ and 5′-CGGTATGCAGAAACCACTG-3′. The insertion of the kanamycin cassette was further verified in both H. pylori 251 ΔcagPAI and B128 7.13 ΔcagM bacteria using the primers 5′-CGGTATAATCTTACCTATCACCTC-3′ and 5′-TTTGACTTACTGGGGATCAAGCCTG-3′. Bacteria were routinely cultured on horse blood agar or in liquid broth as described previously (17). Medium was supplemented with 20 μg/ml kanamycin or 4 μg/ml chloramphenicol as required (17). H. pylori was incubated with epithelial cells at a multiplicity of infection (MOI) of 10. Viable counts of H. pylori were determined by serial dilution and plating.
Cell culture assays.Human gastric adenocarcinoma (AGS) and human embryonic kidney (HEK293) cells were routinely cultured in RPMI medium or Dulbecco modified Eagle medium (DMEM), respectively, supplemented with 10% (vol/vol) fetal calf serum (FCS) (Gibco, Invitrogen, NY) (57). AGS cells stably expressing short hairpin RNA (shRNA) targeting β1 RNA or control cells expressing shRNA for firefly luciferase-GL2 RNA were generated by retroviral transduction and verified by fluorescence-activated cell sorter (FACS) analysis (32). The levels of NF-κB activation and IL-8 production in epithelial cells were measured using a luciferase reporter assay (2, 28, 57) and IL-8 enzyme-linked immunosorbent assay (ELISA), respectively. Briefly, cells were transfected with 300 ng/well of Ig-κ luciferase (20, 46) and 250 ng/well of dTK Renilla plasmid vector (Promega, WI) using either FuGene (Roche, IN) or polyethylenimine (PEI) (0.25 μM; Polysciences, Warrington, PA). Cells were cocultured with H. pylori for 4 h prior to cell lysis and measurement of luciferase activity. Transfection efficiencies were normalized using Renilla luciferase activity. For IL-8 ELISAs, cells were stimulated with H. pylori for 1 h, the medium was replaced, and the cells were further incubated for a total of 24 h. Cells were incubated with 160 nM phorbol myristate acetate (PMA) (Invitrogen) for 24 h. IL-8 levels in supernatants were determined using the BD OptEIA human IL-8 ELISA set according to the manufacturer's instructions (BD Biosciences).
Cholesterol depletion and replenishment of cultured cells.Cells were depleted of cholesterol and subsequently replenished as previously described (28). In brief, cells were incubated with 4 mM methyl-β-cyclodextrin (MβCD) (Sigma-Aldrich, St. Louis, MO) in serum-free medium at 37°C 5% CO2 for 30 min. This treatment was shown not to affect cell viability (28). Cholesterol was replenished on the membrane surface of AGS cells by the addition of 250 μM cholesterol (Sigma-Aldrich) with 4 mM MβCD for 1 h (28). Medium was replaced prior to stimulation. Cholesterol-rich microdomains were labeled using the Vybrant Alexa Fluor 555 lipid raft labeling kit according to the manufacturer's instructions (Molecular Probes, Invitrogen, OR). β1 integrin was labeled using an integrin β1 (M-106) antibody (rabbit polyclonal IgG; Santa Cruz, CA) followed by a goat anti-rabbit Alexa Fluor 488 secondary antibody (Molecular Probes). Cells were viewed using a Nikon C1 confocal microscope.
Isolation of cholesterol-rich microdomains using density gradient centrifugation.Cholesterol-rich microdomains were isolated as detergent-resistant membranes using modifications of a previously described method (23). Briefly, AGS cells (1 × 105 cells/ml) were cultured in 15-cm tissue culture dishes, grown to confluence and treated or not with 4 mM MβCD. Following incubation, cells were washed with phosphate-buffered saline (PBS), removed using a cell scraper, and washed twice. Cells were resuspended in lysis buffer (1% [vol/vol] Triton X-100, 25 mM Tris-HCl, 1 mM EDTA, 150 mM NaCl, Complete-Mini protease [Roche Diagnostics, Mannheim, Germany], and phosphatase inhibitors [Sigma-Aldrich]) and incubated at 4°C for 20 min. Cell lysates (1 ml) were passed 10 times through a 21-gauge needle and mixed with 2 ml of 60% (vol/vol) Optiprep solution. A four-step discontinuous Optiprep gradient was prepared by layering 3 ml of 35% (vol/vol) Optiprep in detergent-free lysis buffer, 3 ml of 30% (vol/vol) Optiprep in detergent-free lysis buffer, and 3 ml of lysis buffer. Gradients were centrifuged at 210,000 × g for 2 h at 4°C and 12 fractions collected. Proteins were precipitated with 10% (vol/vol) trichloroacetic acid (Merck, Darmstadt, Germany) and resuspended in 50 μl Laemmli buffer. Cholesterol-rich microdomains were identified using antibodies to lipid raft and non-lipid raft markers (flotillin-1 and transferrin receptor, respectively) (BD Biosciences, San Diego, CA). Antibody complexes were detected using ECL detection reagent (GE Healthcare, United Kingdom).
Measurement of peptidoglycan in cultured cells.Specific 3H radiolabeling of H. pylori PG was achieved by cultivation of H. pylori bacteria in brain heart infusion (BHI) broth supplemented with 20 μM 3H-labeled meso-diaminopimelate (mDAP) (American Radiolabeled Chemicals Inc., St. Louis, MO), as described previously (57). Because mDAP can be converted into either l-lysine (i.e., proteins) or PG, the gene encoding diaminopimelate decarboxylase (LysA) was inactivated to prevent conversion of mDAP into l-lysine (57). AGS cells were plated in either Labtek slides (Nalge Nunc International Corp, Naperville, IL) or six-well plates and cultured overnight. Prior to coculture with radiolabeled H. pylori, cells were either treated with MβCD, treated and cholesterol replenished, or left as controls. The presence of 3H-PG in AGS cells was detected by immersion of fixed slides in EM-1 emulsion (GE Healthcare) and counterstaining with Giemsa. The slides were processed and analyzed by bright-field photomicrography (36). Quantification of 3H-PG in AGS cells was determined using a Wallac 1409 scintillation counter. Prior to measurement, cells were washed three times with PBS, lysed using 500 μl 1% (vol/vol) Triton-X in PBS, and subsequently added to 4.5 ml of Optiphase Hisafe 3 scintillation fluid (Perkin-Elmer, Boston, MA). 3H-PG within AGS cells was further quantified by washing AGS cells that were cultured on Labtek slides with PBS, fixing with 4% (wt/vol) paraformaldehyde (BDH AnalaR, Merck, Australia) (57), and measuring radioactivity levels using a Microimager (Biospace, Paris, France) (36).
Treatment of cells with integrin-blocking antibodies.AGS cells were cultured overnight prior to incubation with either AIIB2 (rat anti-human β1 integrin, IgG1) or BIIG2 (rat anti-human α5 integrin, IgG2b κ) integrin-blocking antibodies (Developmental Studies Hybridoma Bank; 2.5 μg/ml) for 1 h at 37°C with 5% CO2 (32). As controls, cells either were left untreated or were pretreated with 2.5 μg/ml of a rat IgG1 isotype control (BD Pharmingen). Cells were then cocultured or not with H. pylori parental or ΔcagA mutant bacteria at an MOI of 10 for 4 to 6 h. Cells were subsequently fixed with 4% (wt/vol) paraformaldehyde (BDH AnalaR) and stained using Alexa Fluor 488 phalloidin (Molecular Probes). The percentages of AGS cells with the cell-scattering phenotype were enumerated in a blinded manner. All cells from at least five fields of view (magnification, ×40) for each condition were counted in three independent experiments. Elongated cells were defined as cells that had an elongated shape and thin, needle-like protrusions, as described in previous studies (6). Cells that were spherical were considered to be nonelongated cells.
Cell attachment assays.Cell attachment assays were performed according to procedures described previously (32). In brief, wells in a 96-well ELISA plate were coated with 50 μg/ml of fibronectin (Sigma) or vitronectin (Biosource) at 4°C overnight. Wells were washed with PBS and then blocked with 5% bovine serum albumin. After further washing of the wells with PBS, an aliquot of a single-cell suspension of AGS containing 4 × 104 cells was added to each well and incubated at 37°C for 2 h. After washing three times with RPMI, attached cells were fixed with 3.8% (wt/vol) paraformaldehyde (Sigma) and stained with 0.5% crystal violet (Sigma). Cells were then washed with PBS. The crystal violet bound to the attached cells was then extracted by incubation with 10% (vol/vol) acetic acid. The absorbance of the extract was measured at 590 nm for quantification of the amount of attached cells.
Statistical analysis.Data were analyzed using the Student's t test or the Mann-Whitney U test, as appropriate. Differences in data values were considered significant at a P value of <0.05.
RESULTS
Cholesterol depletion of epithelial cells results in decreased NF-κB activation and IL-8 production in response to H. pylori stimulation.It has previously been reported that the depletion of host cell cholesterol by MβCD reduced CagA translocation and CagA-induced responses in epithelial cells (34). As CagA translocation is dependent on a TFSS (7, 40) and this system is also required for delivery of PG to cells and the induction of proinflammatory responses (2, 21, 57), we wished to examine the effect of cholesterol depletion on NF-κB-dependent responses to H. pylori. As shown in Fig. 1, cells that had been pretreated with MβCD (Fig. 1B) showed substantially less reactivity with a lipid raft-specific stain than nontreated cells (Fig. 1A), suggesting disruption of lipid rafts by MβCD treatment. Additionally, immunofluorescent labeling of β1 integrin (Fig. 1C and D), which demonstrated colocalization of β1 integrin with lipid rafts (Fig. 1E, F, G, and H), confirmed earlier studies reporting the association of α5β1 integrin with cholesterol-rich microdomains (27, 37). The ability of MβCD to deplete cholesterol from epithelial cell membranes was further confirmed by density gradient isolation of detergent-resistant fractions, corresponding to lipid raft domains (23). Specifically, we showed that MβCD treatment of cells resulted in the redistribution of a lipid raft marker, flotillin-1, from detergent-resistant to detergent-soluble fractions (Fig. 1I).
Treatment of epithelial cells with MβCD results in disruption of cholesterol-rich microdomains. AGS cells were either not treated (A, C, E, and G) or treated with 4 mM MβCD (B, D, F, and H) prior to staining with a Vybrant Alexa Fluor 555 lipid raft stain (red) and an anti-β1 integrin antibody (green). (E and F) Merged images depicting areas of colocalization as indicated by arrows. (G and H) Magnified images of panels E and F as depicted by white boxes. Bars, 10 μm. (I) AGS cells were treated or not with MβCD, lysed with 1% Triton X-100, and separated by density gradient centrifugation (23). Twelve fractions were subjected to SDS-PAGE and Western blotting performed with antibodies to flotillin-1, a lipid raft marker, or transferrin receptor, a non-lipid raft marker. Data are representative of three independent experiments.
Importantly, pretreatment of either AGS or HEK293 cells with MβCD significantly reduced the levels of NF-κB activation induced by cagPAI-positive H. pylori strains compared to that in cells not treated with MβCD (P < 0.05) (Fig. 2A and B). MβCD pretreatment had a similar effect on IL-8 production in H. pylori-stimulated cells (Fig. 2C) (P < 0.001). To confirm the specific cholesterol-depleting effects of MβCD on lipid rafts, we replenished host cell membranes by the addition of exogenous cholesterol. Interestingly, we found that the decreased NF-κB responses observed in MβCD-treated HEK293 (Fig. 2D) or AGS (data not shown) cells could in fact be restored by cholesterol replenishment of cells. Furthermore, the ability of MβCD to specifically inhibit H. pylori-induced responses, but not other types of proinflammatory responses, was confirmed in cells that had been stimulated with the potent NF-κB agonist PMA (Fig. 2E).
Cholesterol-rich microdomains are required for H. pylori-induced NF-κB activation and IL-8 production. NF-κB reporter activity (A, B, and D) and IL-8 production (C and E) in epithelial cells that were either pretreated (open bars) or not (filled bars) with MβCD prior to bacterial stimulation are shown. (A and B) NF-κB reporter activity of HEK293 (A) or AGS (B) cells that were nonstimulated (NS) or stimulated with H. pylori 251 (A) or B128 7.13 (B) parental (wild type [WT]) bacteria or the respective isogenic cag strains. (C) IL-8 production in HEK cells stimulated with H. pylori 251 WT or isogenic mutant strains. (D) NF-κB reporter activity of HEK293 cells in which MβCD-treated cells were replenished with 250 μM cholesterol (shaded bars) prior to bacterial stimulation. (E) IL-8 production in AGS cells stimulated with either H. pylori 251 WT bacteria or PMA. Data correspond to the means ± standard errors of the means (SEM) (determined in triplicate) and are representative of at least three independent experiments, *, P < 0.05; ***, P < 0.001; Δ, P > 0.05.
Consistent with previous findings (2, 21, 57), isogenic H. pylori ΔcagPAI or ΔcagM strains, each lacking a functional TFSS, were poor inducers of NF-κB/IL-8 responses in cells, thus confirming the importance of a functional TFSS for these responses (Fig. 2A to C). In contrast, H. pylori ΔcagA bacteria, which retain a functional TFSS and its proinflammatory effects (18), consistently induced high levels of NF-κB activation (Fig. 2A). Although cagA inactivation appeared to have a small effect on H. pylori-mediated IL-8 responses (Fig. 2C), we suggest that the induction of proinflammatory responses in this particular model was mediated primarily by the TFSS rather than by the proinflammatory effects of CagA, as reported by some workers (10, 35). Taken together, these data suggest that cholesterol-rich microdomains are required for TFSS-dependent induction of NF-κB-dependent responses within epithelial cells.
H. pylori exploits cholesterol-rich microdomains for delivery of peptidoglycan into host cells.Previous studies from our group have identified the important role of NOD1 in the proinflammatory responses induced by the H. pylori TFSS in epithelial cells. (57) As these responses were shown here to be dependent on an association with cholesterol-rich microdomains and as NOD1 is known to be activated in response to H. pylori PG (57), we speculated that these domains may also be involved in H. pylori PG translocation to cells. To test this hypothesis, we specifically tritiated the PG in live H. pylori ΔlysA bacteria using a previously described technique (57) and then performed coculture experiments with these bacteria and AGS cells that had been either pretreated or not with MβCD prior to stimulation. Consistent with previous work (57), large accumulations of radiolabeled PG particles were observed in AGS cells incubated with H. pylori ΔlysA bacteria with a functional TFSS (Fig. 3B) compared with control noninfected cells (Fig. 3A). In contrast, MβCD treatment of AGS cells prior to coculture with H. pylori resulted in fewer deposits of radiolabeled PG particles than in nontreated, H. pylori-infected cells (Fig. 3C and data not shown). Furthermore, coculture of nontreated AGS cells with isogenic H. pylori bacteria without a functional TFSS (ΔlysA cagM) also resulted in reduced amounts of PG deposits in comparison to nontreated cells cocultured with H. pylori ΔlysA bacteria (Fig. 3D), thus confirming that the cagPAI is required for delivery of PG to AGS cells. The quantities of translocated PG were determined using scintillation counting and microimaging (Fig. 3E and F, respectively). In accordance with the observed reduction in NF-κB-dependent signaling following disruption of cholesterol-rich microdomains, MβCD treatment of AGS cells caused significant reductions in PG translocation compared with untreated cells (P < 0.05) (Fig. 3E and F). Conversely, the restoration of cellular cholesterol in MβCD-treated cells resulted in wild-type levels of PG delivery for H. pylori with an intact TFSS (Fig. 3E). As expected (57), cells cocultured with H. pylori ΔlysA cagM bacteria showed a reproducible trend toward a reduction in levels of radioactivity, although the reduction did not reach statistical significance (Fig. 3F). Collectively, these data suggest that cholesterol-rich microdomains are critical for TFSS-dependent delivery of PG into epithelial cells.
H. pylori interacts with cholesterol-rich microdomains for type IV delivery of PG into the host cell. (A to D) Presence of 3H-PG in nonstimulated AGS cells (A) or cells cocultured with H. pylori (B to D). Radioactive label is visualized by the deposition of silver particles (indicated by arrows). (A) Nontreated AGS cells; (B) nontreated cells cocultured with H. pylori 251 ΔlysA; (C) cells pretreated with MβCD prior to stimulation with H. pylori 251 ΔlysA; (D) nontreated cells stimulated with H. pylori 251 ΔlysA cagM. Cells were counterstained with Giemsa. Bars, 20 μm. (E) Radioactivity of AGS cells that were nonstimulated (NS) or stimulated with H. pylori 251 ΔlysA bacteria. Cells had been either pretreated (open bars) or not (filled bars) with MβCD or had been MβCD treated and subsequently replenished with 250 μM cholesterol (shaded bars) prior to bacterial stimulation. Radioactivity was measured by scintillation counting and normalized to the amount of radiation detected in nonstimulated, nontreated cells. Data correspond to the means ± SEM (determined in duplicate or triplicate) from three pooled experiments (*, P < 0.05). (F) Radioactivity of NS or H. pylori-stimulated AGS cells measured by microimaging. Data are expressed as the percentages of the means in counts per minute (cpm) per mm2 (calculated from duplicate readings) (***, P < 0.001; Δ, P > 0.05).
H. pylori interacts with α5β1 integrin to induce NF-κB-dependent signaling in epithelial cells. H. pylori has been reported to interact with α5β1 integrin for TFSS-dependent delivery of CagA into epithelial cells (32). Given that we have confirmed that this receptor associates with cholesterol-rich microdomains (Fig. 1A to H) (27, 37) and that cellular cholesterol is important for the induction of CagA-induced host responses (34), we wished to investigate whether H. pylori may also exploit α5β1 integrin to induce TFSS-dependent responses in epithelial cells. For this, we first assessed the ability of two integrin-blocking antibodies, AIIB2 and BIIG2, to inhibit H. pylori interactions with the host cell, by examining the effects of these antibodies on CagA-mediated cell scattering in AGS cells. In accordance with the work of Kwok et al. (32), pretreatment of the cells with either AIIB2 or BIIG2 antibodies caused an approximately 55% decrease in the proportions of cells exhibiting a cell-scattering phenotype induced by a wild-type H. pylori strain (Fig. 4A and B) (P < 0.05). An isogenic H. pylori ΔcagA mutant strain served as a negative control (3, 22, 38, 50, 51). To confirm the specificities of the AIIB2 and BIIG2 antibodies, which block integrin β1 and α5 functions, respectively, we performed cell attachment assays with fibronectin and vitronectin. Fibronectin mediates cell attachment by binding predominantly to integrin α5β1 (4, 47, 60), whereas vitronectin mediates cell attachment primarily through binding to integrins αvβ3 and αvβ5 (12, 61). Treatment of the cells with BIIG2 (Fig. 4C) or AIIB2 (Fig. 4D) antibodies abrogated cell attachment to fibronectin while causing only a slight inhibition of cell attachment to vitronectin, thus confirming the abilities of AIIB2 and BIIG2 to specifically inhibit integrin α5β1.
CagA requires α5β1 integrin for induction of the cell-scattering phenotype. (A) Induction of the cell-scattering phenotype in AGS cells, as determined by phalloidin staining of cells that were either nonstimulated (NS) or cocultured with H. pylori 251 parental (WT) or ΔcagA mutant bacteria. Cells had either been pretreated or not (control) with the integrin β1 and integrin α5 function-blocking antibodies AIIB2 and BIIG2, respectively. Bars, 20 μm. (B) The percentages of these cells (NS, filled bars; H. pylori 251 WT, open bars; H. pylori 251 ΔcagA, shaded bars) displaying a cell-scattering phenotype, characterized by cell elongation and the formation of spindles (indicated by arrows), were pooled from duplicate experiments and quantitated by blind counting and are presented as percentages. (C and D) AGS cells that had either been pretreated (open bars) or not (filled bars) with the antibody BIIG2 (C) or AIIB2 (D) were allowed to attach to immobilized fibronectin (50 μg/ml) or vitronectin (50 μg/ml). Net A590 refers to a quantification of attached cells. Data are triplicates from three independent experiments and are shown as means ± SEM. *, P < 0.05; ***, P < 0.001.
Next, we examined the ability of the integrin-blocking antibodies to affect NF-κB-dependent responses induced by the H. pylori TFSS. Importantly, cells that were preincubated with either AIIB2 or BIIG2 integrin-blocking antibodies and that were then cocultured with cagPAI-positive H. pylori bacteria demonstrated significant decreases in both NF-κB activation (Fig. 5A) (P < 0.05) and IL-8 production (Fig. 5B) (P < 0.001) compared with untreated cells. We further confirmed the specificity of antibody blocking by pretreating cells with either the AIIB2 antibody or an isotype control (rat IgG1) prior to stimulation with H. pylori 251 bacteria (Fig. 5C and D) (P < 0.05). This finding was further validated using β1-knockdown AGS cells in which β1 expression was transiently knocked down prior to stimulation with parental or ΔcagA H. pylori strains. As the β1 integrin subunit seemed to play a more important role than the α5 subunit for H. pylori interactions with AGS cells (Fig. 4B), we used AGS cells in which β1 integrin was knocked down. In these cells, a reproducible trend toward a reduction in NF-κB activation (Fig. 5E) (P > 0.05) was observed as well as a significant decrease in the levels of IL-8 synthesis (Fig. 5F) (P < 0.01) compared to those in control knockdown cells. Taken together, the data show that H. pylori-mediated NF-κB-dependent responses in AGS cells are dependent on bacterial interactions with α5β1 integrin in cholesterol-rich microdomains.
PG requires α5β1 integrin for the induction of NF-κB-dependent responses. (A and B) NF-κB reporter activity (A) and IL-8 production (B) in AGS cells that were nonstimulated (filled bars) or cocultured with H. pylori 251 bacteria (open bars) in the presence or absence (control) of the BIIG2 and AIIB2 antibodies. (C and D) To confirm the specificity of the AIIB2 antibody, NF-κB reporter activity (C) and IL-8 production (D) were measured in AGS cells that were not treated (filled bars) or pretreated with AIIB2 (open bars) or a rat IgG1 isotype control (shaded bars) prior to coculture with H. pylori 251 bacteria. (E and F) NF-κB reporter activity (E) and IL-8 production (F) in β1-knock down (open bars) or control (filled bars) AGS cells that were nonstimulated (NS) or cocultured with H. pylori P1 (WT) bacteria or the respective isogenic ΔcagA mutant. Data correspond to the means ± SEM (determined in triplicate) and are representative of two or three independent experiments (*, P < 0.05; **, P < 0.01; ***, P < 0.001; Δ, P > 0.05).
DISCUSSION
Although the role of the H. pylori TFSS in the induction of proinflammatory responses in epithelial cells is well established, the exact mechanism of action remains elusive. The current study provides evidence that cholesterol-rich microdomains, and more specifically α5β1 integrin, represent important points of interaction between the H. pylori TFSS and the host cell. Specifically, we demonstrate that these cholesterol-enriched microdomains are essential for not only the induction of NF-κB-dependent responses by the H. pylori TFSS but also its delivery of PG to epithelial cells. Indeed, the effects of the cholesterol-depleting agent MβCD on TFSS-dependent NF-κB activation and PG delivery could be reversed by the addition of exogenous cholesterol to host cells (Fig. 2D), thereby confirming the specificity of this pharmacological agent.
As reported previously (2, 57), PG delivered to the host cell by the TFSS activates NOD1 and, subsequently, NF-κB, leading to IL-8 release after only 1 h of contact of cells with H. pylori (2, 57). Furthermore, our group recently demonstrated that NOD1 was essential for the induction of both NF-κB and activator protein 1 (AP-1) activation by cagPAI-positive H. pylori (2). This was demonstrated using the electrophoretic mobility shift assay (EMSA), directly showing the physical translocation of NF-κB and AP-1 complexes to the nucleus in response to NOD1 activation (2). The nuclear translocation and activation of NF-κB and AP-1 in response to H. pylori stimulation was shown to be primarily dependent on NOD1 (2, 21). Consequently, it is likely that the NF-κB-dependent responses observed throughout this study are dependent on NOD1 activation and hence can be used as a measure of NOD1 activity.
Early studies reported that CagA was not required for H. pylori-mediated NF-κB and/or IL-8 responses in epithelial cells (11, 15, 18, 52). More recently, however, contradictory findings have been reported regarding the role of CagA in the induction of proinflammatory responses in host cells (10, 35). In at least one of these studies, it appeared that CagA-mediated IL-8 production occurred only after extended periods of cell stimulation with specific H. pylori strains (10). Thus, it is possible that strain-specific factors are responsible for the different findings. Alternatively, it is possible that distinct lineages of the AGS cell lines express different relative amounts of α5β1 integrin or indeed other, as-yet-unidentified, host molecules needed for optimal H. pylori TFSS-mediated responses.
From the data presented here, the fact that NF-κB and IL-8 responses were greatly reduced by MβCD pretreatment of cells stimulated with H. pylori 251 ΔcagA bacteria suggests that at least in our cell culture model, the observed lipid raft-dependent responses were independent of CagA (Fig. 2). Based on the accumulated evidence, we suggest that these responses were attributable to NOD1 recognition of PG. However, we do not rule out the possibility that part of the lipid raft-dependent NF-κB and IL-8 responses could be induced by H. pylori factors encoded by the cagPAI. Although CagA translocation via H. pylori interactions with cholesterol-rich microdomains has been reported previously (34), this is the first evidence suggesting that cholesterol-rich microdomains may also be important for the TFSS-dependent PG delivery and subsequent NF-κB activation and IL-8 production, mediated by the innate immune molecule NOD1.
Given that many host cell receptors and signaling molecules cluster within cholesterol-rich microdomains (27, 53, 54), we wished to determine whether a particular receptor may be involved in the NF-κB-dependent signaling induced by cagPAI-positive H. pylori bacteria. We have shown for the first time that the pretreatment of cells with α5β1 integrin-blocking antibodies, as well as the generation of β1-knockdown cells, resulted in reduced NF-κB and IL-8 responses (Fig. 5). In both instances, however, complete abolishment of these responses was not achieved. This could be attributed to the possibility that other host integrins may also play a role in the activation of NF-κB and IL-8 production in response to H. pylori or that these responses may be responding to non-integrin-mediated activation pathways. Interestingly, it was noted during this study that the AIIB2 antibody more frequently showed a stronger blocking effect on proinflammatory responses than the BIIG2 antibody, despite AGS cells expressing similar amounts of α5 and β1 integrin subunits on their surfaces (32). This observation could indicate a stronger reliance on the β1 integrin subunit for H. pylori interactions with AGS cells. Despite this, sufficiently high levels of disruption to α5β1 integrin signaling were achieved, which indicates that activation of NF-κB-dependent signaling by cagPAI-positive H. pylori bacteria is also likely to be dependent on α5β1 integrin. This hypothesis is consistent with earlier findings where an H. pylori ΔcagL mutant was unable to induce IL-8 responses in epithelial cells (11, 18). Studies are under way to investigate whether CagL, apart from being required for delivery of CagA into host cells (32), may also be involved in activation of the NOD1 signaling pathway.
In conclusion, the current work describes a possible mechanism whereby H. pylori induces proinflammatory responses in epithelial cells via the PRM NOD1. We suggest that the H. pylori adhesin CagL, which has been shown to be critical for induction of IL-8 responses (18), binds to and activates α5β1 integrin within cholesterol-rich microdomains. In a mechanism similar to that described for the roles of CagL (32) and CagY (26) in CagA delivery, we propose that CagL may interact with α5β1 integrin within cholesterol-rich microdomains to trigger delivery of PG across the host cell plasma membrane to cytosolic NOD1, resulting in downstream NF-κB activation and IL-8 production. Interestingly, NOD1 was shown in another study to be recruited to cholesterol-rich microdomains, which also serve as sites of bacterial entry for the invasive pathogen Shigella flexneri (30). Also, it was recently reported that bacterial outer membrane vesicles from Gram-negative bacteria enter epithelial cells through these domains to induce NOD1-dependent responses (28), albeit less effectively than via the TFSS. This could explain the significant reduction in NF-κB activation and IL-8 production seen in cells that were treated with MβCD prior to stimulation with TFSS-deficient mutant strains of H. pylori compared to untreated cells (Fig. 2A to C). Thus, we propose that bacterial PG may be delivered to cytosolic NOD1 in epithelial cells via multiple mechanisms, each involving cholesterol-rich microdomains or lipid rafts. Further work is therefore required to determine the precise mechanism(s) by which bacterial pathogens exploit cholesterol-rich microdomains (20, 55) to initiate NOD1 signaling in epithelial cells. The finding that α5β1 integrin is important for NF-κB activation and IL-8 induction by H. pylori, however, provides important new insights into how these microdomains may contribute to such responses. Nevertheless this finding cannot easily be tested in vivo because the reference mouse-colonizing H. pylori strain, SS1, lacks a functional TFSS (14, 16, 45, 56). Thus, it will be necessary to use H. pylori strains such as B128 7.13, which has a functional TFSS (25), but this strain does not colonize mice to the same high bacterial loads as H. pylori SS1 (R. L. Ferrero and M. Kaparakis-Liaskos, unpublished data). Alternatively, the Mongolian gerbil may be useful, because this model has been successfully used to study the in vivo functions of the H. pylori TFSS (25, 41, 42, 58). Finally, it remains to be elucidated whether other raft-located cellular components contribute to NF-κB-dependent responses in epithelial cells.
ACKNOWLEDGMENTS
This work was supported by the National Health and Medical Research Council (project grant 334127). M.L.H. is supported by a Monash University graduate scholarship and an Australian Postgraduate Award. The research at Monash Institute of Medical Research was supported by the Victorian Government's Operational Infrastructure Support Program.
The monoclonal antibodies AIIB2 and BIIG2 were generated by Caroline H. Damsky and were obtained from the Developmental Studies Hybridoma Bank, which was developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA. We thank Steffen Backert for the H. pylori P1 ΔcagA mutant, Tamas Hatfaludi for his help with scintillation counting, and Camden Lo for his assistance with confocal microscopy.
FOOTNOTES
- Received 28 April 2010.
- Returned for modification 31 May 2010.
- Accepted 6 August 2010.
- Copyright © 2010 American Society for Microbiology
