The intestinal epithelium is a single layer of polarized cells and is the primary barrier separating foreign antigen and underlying lymphoid tissue. IFNγ alters epithelial barrier function during inflammation by disrupting tight cell junctions and facilitating the paracellular transport of luminal antigens. The aim of this work was to determine whether Campylobacter infection of cells exposed to IFNγ would lead to greater disruption of cell monolayers and hence increased bacterial translocation.
Monolayers were polarized on Transwell polycarbonate membranes for 14 days and then cultured in the presence or absence of 100 U/mL IFNγ. Campylobacter was added to the apical side of the monolayer at an MOI of 30. Transepithelial electrical resistance (TEER) was recorded and bacteria in the basal well counted every 2 hours. Cells were stained for occludin, actin, and nuclear DNA, and cell viability determined by measurement of apoptosis.
In the presence of IFNγ, TEER dropped significantly after 18 hours, indicating a reduction in barrier function. A further significant decrease was seen in the presence of both IFNγ and Campylobacter, indicating a synergistic effect, and cellular morphology and viability were affected. Bacterial translocation across the monolayer was also significantly greater in the presence of IFNγ.
These combined effects indicate that Campylobacter infection concomitant with intestinal inflammation would result in a rapid and dramatic loss of epithelial barrier integrity, which may be a key event in the pathogenesis of Campylobacter-mediated colitis and the development of bloody diarrhea.
The incidence of inflammatory bowel disease (IBD) has increased markedly in Western countries in recent years. Chronic intestinal inflammation does not appear to have a single specific infectious or autoimmune cause, but is probably multifactorial and known to involve components of the intestinal microflora. Although bacterial pathogens are not thought to be a primary cause of IBD, superinfection with enteropathogens can cause exacerbations or relapse of both ulcerative colitis and Crohn's disease1 and Campylobacter spp. are frequently isolated from such patients.2,–4
Infection with Campylobacter spp. can cause a range of symptoms ranging from uncomplicated enteritis to profuse bloody diarrhea and chronic relapsing infection. Invasive infection can result in acute hemorrhagic inflammation, ulceration, and edema. Such invasion and tissue damage may occur throughout the jejunum, ileum, and colon. Infected tissues also show marked infiltration of polymorphonuclear leukocytes (PMNs) and lymphocytes, a picture similar to IBD. In acute disease immunologically active cells, erythrocytes, and frank blood will be present in the intestinal lumen due to ulceration of and damage to the mucosa.5
The healthy intestinal epithelium acts as a barrier between foreign antigen and underlying lymphoid tissue. Bacterial entry into cells, either by invasion of epithelial cells or uptake by phagocytic cells in the lamina propria, produces a cascade of signals causing the release of proinflammatory cytokines and chemokines from macrophages into the local environment, resulting in the induction of an inflammatory state. In the case of bacterial uptake by macrophages, the secreted chemokines recruit neutrophils and monocytes that destroy the invading organisms. When enterocytes are directly infected they release cytokines and chemokines, including tumor necrosis factor α (TNFα), interleukin-8, which is a potent recruiter of neutrophils, and CC chemokines, macrophage inflammatory protein-1α and β, monocyte chemoattractant protein-1, and RANTES, which are chemoattractants for eosinophils, monocytes, and T cells. Effector T cells are then recruited to the site of inflammation and produce proinflammatory cytokines including TNFα and interferon γ (IFNγ).
As part of the inflammatory response, IFNγ alone and in combination with other proinflammatory cytokines such as TNFα alters epithelial barrier function by disrupting the multiprotein tight junction complex between epithelial cells and facilitates the paracellular transport of luminal antigens, a process that has been examined extensively in vitro.6,–9 The reduction in transmembrane resistance and consequent breakdown in barrier function mediated by IFNγ is thought to be a key contributor to the pathogenesis of IBD,10,11 and upregulation of IFNγ has been demonstrated in mucosal biopsies from IBD patients.12,13 IFNγ influences epithelial barrier function independently of apoptosis and induces disassembly of tight junctions8 by inducing cellular internalization of occludin and claudin-1 by a macropinocytosis-like mechanism14,15 involving reorganization of the actin cytoskeleton.16
Occludin is expressed in nearly all tight junctions.17 Its loss results in an increase in mucosal permeability, but it may have a redundant role during development, with its absence being compensated for by other junctional components.18,19 Recent work has shown that Campylobacter spp. can alter and rearrange occludin so that they can cross the epithelium paracellularly via the tight junctions.20,21 The tight junctions are believed to repair, at least partially, behind the bacteria so that no decrease in transepithelial resistance (TEER) is seen over short time courses.20,22,23 Over longer periods of time, between 24 and 48 hours, the junctions are thought to gradually lose integrity, thus allowing the ingress of further bacteria into the lamina propria, potentially leading to the development of uncontrolled inflammation.20,21
The aim of this work was to determine whether the pathogenesis of infection with Campylobacter spp. would be exacerbated in the presence of IFNγ, which would be produced in the presence of a preexisting inflammatory state.
Materials and Methods
Growth of Polarized Caco-2 Monolayers and Exposure to IFNγ
Caco-2 human adenocarcinoma cells were grown in DMEM (Invitrogen, UK) + 10% FCS (PAA Labs, Germany) containing 2 mM glutamine and 100 U/mL penicillin/streptomycin on 12-mm Transwell membranes (12 mm diameter, 3 μm pore size; Corning Glass Works, Corning, NY) in 12-well tissue culture plates. A 1.5 mL volume of DMEM was added to each basal well of the plate and 3 × 105 cells in 0.5 mL DMEM added to the apical chamber of the insert. Plates were incubated at 37°C in 5% CO2 for 10–12 days until cells formed confluent monolayers and the TEER was greater than 300 Ω/cm2 as measured with an epithelial voltmeter. The medium was then changed for one that was antibiotic-free and the plates incubated for a further 24 hours in the presence or absence of 100 U/mL IFNγ (Sigma, St. Louis, MO) in the basolateral well.
Infection of Monolayers and Measurement of Electrical Resistance
Campylobacter jejuni NCTC 11168 cultures were grown in 1.5 mL volumes of Mueller–Hinton broth in a microaerobic atmosphere for 48 hours. These bacteria were applied to the apical chamber of the Transwell insert at an MOI of 30 and incubated with the monolayer for 24 hours in the presence or absence of 100 U/mL IFNγ. In parallel, 4 μg/mL mouse antihuman monoclonal neutralizing antibody to human IFNγ (Clone B27, BD Pharmingen, San Jose, CA), 2 μg/mL IFNγ receptor blocking antibody (goat polyclonal to IFNGR1, Abcam, Cambridge, UK) or an isotype-matched control antibody were added to the basolateral wells of monolayers exposed to either IFNγ, Campylobacter, or both. TEER of all monolayers was measured at 2-hour intervals and the significance of differences evaluated using a repeated-measures analysis of variance (ANOVA) (Microsoft Excel). Bacteria in the apical and basal wells were enumerated every 2 hours between 6 and 14 hours and differences evaluated statistically at each timepoint using a t-test. After 24 hours the medium was removed and the monolayers examined as outlined below.
Apoptosis
Transwell inserts were immersed in 4% paraformaldehyde in phosphate-buffered saline (PBS) and the cells fixed for 25 minutes at 4°C. Inserts were then washed in PBS, the cells were permeabilized in 0.2% Triton X-100 for 5 minutes, and washed again. Apoptotic cells were then stained using the DeadEnd fluorometric TUNEL system (Promega, Madison, WI) according to the manufacturer's instructions. The membranes were cut out of the inserts with a scalpel and mounted on plain glass slides with Vectashield containing DAPI (Vector Laboratories, Burlingame, CA) and coverslips sealed with nail varnish. Apoptosis and nuclear staining was examined by fluorescence microscopy on a Leica DMRA microscope equipped with a Hamamatsu Orca-ER monochrome camera. Ten fields of view per slide at 63× magnification were digitized using Leica Q-Fluoro software. Images were viewed using ImageJ software (http://rsb.info.nih.gov/ij/) and the number of apoptotic and nonapoptotic nuclei per field of view counted. Differences were analyzed using a t-test (SPSS, Chicago, IL).
Occludin and Actin
Transwell inserts were placed in ice-cold methanol and the cells fixed at 4°C for 45 minutes. Inserts were then washed in PBS, and the cells permeabilized in 0.1% Triton X-100 for 10 minutes, and washed again. Mouse anti-occludin monoclonal antibody (Zymed, San Francisco, CA), diluted 1:200, was added to the apical chamber of the insert and kept at room temperature for 45 minutes. Inserts were then washed in PBS and incubated for a further 45 minutes with FITC-conjugated isotype-specific goat antimouse antibody (Southern Biotech, Birmingham, AL) diluted 1:100 and 1.25% Texas Red-X phalloidin (Invitrogen) in the apical chamber of the insert. The filters were cut out, mounted, viewed, and digitized as above. Images were analyzed using ImageJ software and the area of positive staining for occludin and actin quantified.24 The significance of differences was examined using a t-test, as above.
Results
Monolayer Integrity and Occludin Distribution
Caco-2 cells formed a stable monolayer after 10–12 days culture. At this timepoint, cells were polarized and showed basal localization of the IFNγ receptor. Cells in the monolayer had a characteristic cobblestone-like appearance and occludin staining was visible around the periphery of all cells (Fig. 1a). TEER was maintained over 24 hours (Fig. 2). Monolayer integrity also appeared to be maintained in the presence of Campylobacter alone, as occludin staining was quantitatively similar to that seen on control cells (Fig. 3), and was continuous around the cell periphery (Fig. 1b). Qualitatively, however, junctional staining appeared more intense than in the control and areas of occludin were visible within cells. TEER was maintained (Fig. 2) and numbers of apoptotic cells were no greater than seen in the control monolayers. In monolayers preincubated with IFNγ, TEER dropped significantly during the 24 hours of the experiment, indicating a reduction in barrier function (P = 0.014) (Fig. 2), although the tight cell junctions were not visibly interrupted (Fig. 1c) and apoptotic cell numbers were not increased over the control. There was no significant loss of occludin staining (Fig. 3), but an increase in intensity of junctional staining and intracellular areas of occludin were visible in comparison to the control.
Effect of C. jejuni and IFNγ on occludin distribution in polarized Caco-2 monolayers. Monolayer not exposed to Campylobacter shown as control (a). Cell monolayers were co-incubated with Campylobacter for 24 hours (b), IFNγ for 48 hours (c) and a combination of IFNγ for 24 hours followed by a combination of IFNγ and Campylobacter for a further 24 hours (d).
Change of transepithelial resistance of polarized Caco-2 monolayers exposed to C. jejuni (diamonds), IFNγ (squares), or a combination of the 2 (triangles). TEER of control monolayers is indicated by circles. P-values are for repeated measures ANOVA between indicated lines. Other differences were not significant. Values are means of 3 experiments with bars showing standard deviations.
Area of occludin staining in fluorescence images relative to entire field of view. Positive pixels were enumerated using ImageJ software. Ten fields of view were examined for each experiment, and values are means of 3 experiments with bars showing standard deviation. P-value was for a t-test between values indicated. Other differences were not significant.
When cells were preincubated with interferon and then infected with Campylobacter, a marked breakdown of the tight cell junctions was visible (Fig. 1d) and TEER showed a significantly greater decrease than that of monolayers incubated with either Campylobacter or IFNγ alone (Fig. 2; P = 0.03, P < 0.001, respectively). The area of occludin staining was also significantly decreased (Fig. 3) and 10% of cells were apoptotic, significantly more than seen with either factor alone (P = 0.04; data not shown). Addition of either an IFNγ neutralizing antibody or an IFNγ receptor blocking antibody at the same time as monolayers were coinfected with Campylobacter and IFNγ resulted in TEER being maintained at a level similar to that seen with control monolayers, and no more apoptosis than seen in the control. Monolayers to which control antibodies were added showed a significant decrease in TEER, and significant apoptosis (data not shown).
Actin Distribution
The distribution of F-actin was investigated by quantitative immunofluorescence using Texas Red X-phalloidin. In uninfected cells fine flocculated F-actin was located centrally in the cells and linear areas of actin were present at the cell borders (Fig. 4a). In monolayers treated with IFNγ or Campylobacter, the staining pattern was more broken up, with some dense localized accumulations of F-actin (Fig. 4b,c, respectively). In the presence of both IFNγ and Campylobacter together there were focal accumulations of F-actin (Fig. 4d) and a significant increase in the percentage area of overall F-actin staining to 93.2 ± 2.3% of the total area compared to 60.7 ± 5.4% for the control, 51.6 ± 4.1% for monolayers exposed to interferon alone and 65.9 ± 1.1% for those exposed to Campylobacter alone (P < 0.005). The staining pattern of F-actin in the interferon pretreated Campylobacter-infected monolayers was examined at higher magnification (160×) and small round actin structures could be seen in cells (Fig. 4e). These were not visible in cells in monolayers exposed to single factors or the controls.
Effect of C. jejuni and IFNγ on actin distribution in polarized Caco-2 monolayers. Monolayer not exposed to Campylobacter shown as control (a). Cell monolayers were co-incubated with Campylobacter for 24 hours (b), IFNγ for 48 hours (c), and a combination of IFNγ for 24 hours followed by a combination of IFNγ and Campylobacter for a further 24 hours (d). Image (e) is a higher magnification of treatment (d).
Bacterial Translocation
Campylobacter translocation across the monolayer was below the level of detection until 8 hours. From 10 hours onward greater numbers of Campylobacter were detected in the basal well of monolayers treated with IFNγ than in those exposed to the bacteria alone (P = 0.02; Fig. 5). No multiplication of bacteria was seen in the apical or basal wells over 14 hours, but after 16 hours the experiment was discontinued as multiplication was seen in the apical well.
Numbers of Campylobacter cells recovered from basal wells of monolayers exposed to Campylobacter alone (diamonds) or Campylobacter and IFNγ (squares). P-value is for repeated-measures ANOVA between lines. Values are means of 3 experiments with bars showing standard deviations.
Discussion
Campylobacter can cause a range of diseases ranging from self-limiting enteritis to acute hemorrhagic diarrhea involving inflammation of the intestinal mucosa and shedding of the epithelium. The factors that govern the severity of infection are poorly understood, although we recently showed that the pathogenicity of Campylobacter jejuni in an intestinal model was increased in the presence of norepinephrine.25 The purpose of the work presented here was to determine whether a preexisting inflammatory state would predispose toward Campylobacter invasion of the epithelium, exacerbation of inflammation and loss of epithelial integrity, an outcome similar to that seen in patients with a Campylobacter infection superimposed over an existing inflammatory condition.2,–4
At the levels used here Campylobacter alone did not break down occludin, or disrupt the integrity of the cell monolayer over 24 hours. Previously published data have shown that, at high cell numbers, or over long time courses, C. jejuni can break down cellular tight junctions of Caco-2 cells.20 At the MOI of 30 and after the 24-hour time course used here this effect was not seen, although TEER appeared to decrease transiently during the experiment (Fig. 2). Significant breakdown of junctions and decrease in TEER has also been seen at a lower MOI of 10 after 24 hours in experiments by Chen et al21 using T84 cells. These authors also saw redistribution of occludin to intracellular compartments, but to a more extreme degree to that seen in this work. The differences between this work and the earlier work by Chen et al may be related to the cell lines used. T84 and Caco-2 cells are both immortal lines derived from colon adenocarcinoma, but polarized T84 cells are less differentiated than Caco-2 cells and do not form an organized brush border, but do form monolayers with a very high TEER. These differences may affect bacterial attachment to cells and the degree to which alteration of the tight junctions can be detected. Due to these differences in phenotype, T84 cells are commonly used as a model of intestinal crypt cells and Caco-2 cells as a model of villous enterocytes, the model chosen for examination in this work.
No evidence of increased apoptosis was seen in cells exposed to Campylobacter. Although C. jejuni is known to produce cytolethal distending toxin (CDT), which enters eukaryotic cells and breaks double-stranded DNA, and arrests the cell cycle at G2/M, this occurs over longer periods of time than those used in this work to cause apoptosis.26 Bacterial translocation across the monolayer was seen, but after 6 hours of incubation (Fig. 5). This may occur by invasion of epithelial cells and exocytosis or via movement between the cells.20,23
Although occludin staining retained a continuous appearance when cells were incubated with IFNγ (Fig. 1c) and there was no significant reduction in stained areas, there was a significant decrease in TEER (Fig. 2) in these monolayers at later timepoints. No increase was seen in the number of apoptotic cells. IFNγ is a potent proinflammatory cytokine, and has been found to increase the paracellular permeability of the intestinal epithelial cell barrier by modulating junctional proteins independently of apoptosis.8 These authors found that a number of apical junction complex proteins, including occludin, claudins 1 and 4, and junctional adhesion molecule 1 were internalized by epithelial cells over a 72-hour period of incubation of cells with 100 U/mL IFNγ, with a detectable decrease in TEER after 24 hours treatment and detectable loss of occludin and apoptosis after 72 hours. It is therefore possible that the 48-hour incubation of epithelial cells with IFNγ in this protocol may have had effects on the integrity of both the cells and the junctional proteins that were not detectable by the methods used in this work.
Marked effects on occludin distribution and significant reductions in the area of occludin staining and TEER were seen in the combined presence of Campylobacter and IFNγ. This was accompanied by an increase in the number of apoptotic cells, indicating widespread breakdown of the epithelial barrier by both junctional rearrangement and loss of cells. Further evidence for active rearrangement of junctions was seen in cells stained with phalloidin-Texas Red, in which rounded bodies of actin could be observed within cells (Fig. 4d,e). These appear similar to the F-actin-coated vacuoles described by Utech et al,16 who found that these bodies contained junctional proteins internalized by the cells. Their formation is associated with the loss of cellular polarity27 during junctional disassembly. This is thought to be part of a process during which epithelia become transiently permeable before the contents of these vacuoles are recycled back to the plasma membrane.16 This implies that a combination of IFNγ and Campylobacter are activating a programmed process of junctional disassembly rather than causing a loss of monolayer integrity and cell-to-cell contact entirely by inducing apoptosis. The loss of tight junction integrity allows neutrophils to access the intestinal lumen during combined inflammation and infection in vivo, a characteristic of severe Campylobacter infection,28 but also allows rapid and significantly greater translocation of Campylobacter across the monolayer in vitro (Fig. 5). The increased bacterial translocation is likely to further exacerbate inflammation in vivo, which would in turn result in a further loss of epithelial integrity, leading to the development of a feedback loop of inflammation and infection, a model propose d for Campylobacter interaction with epithelial cells by Chen et al.21 The addition of anti-IFNγ or IFNγ receptor-blocking antibodies in our system, though, resulted in the maintenance of monolayer integrity in the presence of Campylobacter. This intact monolayer also resulted in fewer bacteria accessing the basal surface of the cells (Fig. 5), which is thought to be a preferential site for Campylobacter invasion.29 The fact that an intact monolayer could be maintained and bacterial ingress limited by neutralization of the effects of IFNγ raises the possibility that antiinflammatory treatment could limit the development of disease. IFNγ is a T-cell cytokine and elevated expression is a hallmark of chronic inflammation. Our data therefore suggest that preexisting inflammation in the intestine might be important in accelerating the development of Campylobacter colitis and exacerbating its severity. The combined effects of IFNγ and Campylobacter would also lead the epithelium to become permeable to other bacteria and foreign antigens. The early control of intestinal inflammation during Campylobacter infection would seem to be the key step in preventing the development of complicated colitis. The first line of treatment for Campylobacter-associated bloody diarrhea/colitis is antibiotic therapy. Our data would suggest that at this point it may be prudent to consider treating the infection with a combined regimen of antibiotic to control the organism, and antiinflammatory therapy to control the inflammation associated with the infection, and therefore reduce the degree of tissue inflammation. Ideally, specific regulation of the source of IFNγ production (T cells and macrophages) during chronic inflammatory states overlaid by superinfection with Campylobact er would be the key to containing Campylobacter in the extracellular (luminal) compartment of the gut and retarding its entry into the tissues. This merits further investigation.
In conclusion, the presence of the proinflammatory cytokine IFNγ in combination with Campylobacter results in a rapid loss of epithelial barrier function and cellular viability, allowing greater bacterial movement across this barrier. The neutralization of this cytokine could limit the ingress of Campylobacter into tissues during bacterial infection overlaid over chronic inflammation, or could cut the feedback loop of inflammation and bacterial penetration believed to occur with Campylobacter infection
References
Author notes
From the Division of Veterinary Pathology, Infection and Immunity, School of Clinical Veterinary Science, University of Bristol, Bristol BS40 5DU, UK Email: tristan.cogan@bristol.ac.uk
Grant sponsor: University of Bristol.
