Insights into Campylobacter jejuni colonization of the mammalian intestinal tract using a novel mouse model of infection

Abstract

A lack of relevant disease models for Campylobacter jejuni has long been an obstacle to research into this common enteric pathogen. We recently published that mice deficient in Single IgG Interleukin-1 related receptor (SIGIRR), a repressor of MyD88-dependent innate immune signaling, were highly susceptible to enteric infection by murine bacterial pathogens. Subsequently, we successfully employed these mice as an animal model for the human pathogen C. jejuni and gained substantial new insights into infection by this pathogen. The infected mice developed significant intestinal inflammation, primarily via TLR4 stimulation. Furthermore, the resulting gastroenteritis was dependent on C. jejuni pathogenesis as bacterial strains suffering mutations in key virulence factors were attenuated in causing disease. The ability to infect SIGIRR-deficient mice with C. jejuni sheds new light onto how these bacteria colonize the mucus layer of the intestinal tract, invade epithelial cells, and raises new prospects for studying the virulence strategies and pathogenesis of C. jejuni.

Introduction

Campylobacter jejuni is a common bacterial cause of foodborne gastrointestinal infection, afflicting children and adults in developing and developed nations alike.¹ Despite this, research into infection by this organism has significantly lagged behind other pathogens, partially due to limited relevant infection models with which to study Campylobacter infection.² The predominant reason for the paucity of relevant animal infection models is Campylobacter’s predisposition to colonize most animals, particularly birds, as a relatively harmless commensal.² This benign commensalism in most animals is in contrast to its effects in humans. Upon consumption of contaminated food or water, even a small infectious dose (<10³ organisms) is capable of colonizing and causing significant clinical disease.³ Campylobacteriosis manifests as watery diarrhea (often bloody), abdominal cramps, fever, and vomiting, although the symptoms and severity can be highly variable between individuals. The disease is self-limiting, lasting only a few days or weeks, but on rare occasions, it can lead to the development of Guillain-Barré syndrome. In poultry, one of C. jejuni's preferred hosts, C. jejuni colonizes the intestine to high numbers (>10⁹ CFUs/g), yet it has little, if any, detrimental impact on the health of the host bird.³ Other potential hosts, including cattle and pigs, can be transiently colonized to little ill effect, although colostrum-deprived, neonatal piglets are susceptible to infection, developing mild to severe gastroenteritis as a result.⁴

Another major reason behind the limited availability and usefulness of C. jejuni animal models has been the resistance of mice to both the development of disease, and even simple intestinal colonization by C. jejuni.⁵ Several recent studies have identified the intestinal microbiota in mice as being linked to their colonization resistance to C. jejuni.⁶ Although conventionally housed mice are highly resistant to C. jejuni colonization, germ-free mice or those with a limited-microflora are readily colonized by C. jejuni.^7,8 Interestingly, a study by Bereswill et al.⁶ found that when the microbiota of a conventionally housed mouse is replaced with a “humanized” microbiota, it facilitates colonization by C. jejuni. Although the precise differences between the conventional and “humanized” mouse microbiota that lead to this difference in colonization potential are unknown, the “humanized” microbiota presumably provides nutritional niches that facilitate C. jejuni colonization. Furthermore, in the recent study where we developed our novel Campylobacter mouse model, we found that disruption of the gut microbiota via a single pre-inoculation dose of vancomycin, was sufficient to facilitate high levels of C. jejuni colonization.⁹ The premise behind this antibiotic pre-treatment for C. jejuni is similar to the well-established model for Salmonella Typhimurium enteric infection using streptomycin.^10,11 In each case, whether the mouse is germ-free, “humanized” or antibiotic pre-treated, colonization is facilitated by the removal of competing commensal bacteria. It has become generally accepted that “colonization resistance” is a product of the competition between pathogens and commensals for limited resources in the gut. Under conventional conditions in mice, C. jejuni appears to be at a competitive disadvantage when attempting to establish itself, but upon disruption of the commensal microbial community, sufficient nutritional niches are made available, allowing for the establishment and proliferation of C. jejuni.⁵ In our experience with vancomycin pre-treatment, this disruption substantially reduced the total commensal microbial population and dramatically shifted the proportions of remaining bacteria, allowing C. jejuni to rapidly establish within the gut, reaching levels over 10⁹ CFUs/g, making C. jejuni one of the predominant microbes resident in the intestine during the course of infection.⁹

The SIGIRR Mouse Model

The second drawback of conventional mice as a model for C. jejuni infection is the absence of a significant inflammatory/immune response to C. jejuni, even after colonization has been established. The exact reasons for this disparity between human and murine responses to C. jejuni infection remain unclear, however research using several different mouse models suggests that the mouse innate immune system may be more tolerant toward C. jejuni. A number of studies have employed the IL-10 deficient (^−/−) mouse model ^8,12,13 where the absence of IL-10 impairs the immune system's ability to resolve inflammation, meaning that any induction of inflammation in the gut quickly becomes uncontrolled and ultimately terminal.⁸ The high susceptibility of these mice to C. jejuni colonization strongly suggests the mouse immune system can be stimulated by C. jejuni, but that inflammatory responses are kept tightly controlled under normal circumstances. Unfortunately, a limitation on the use of IL-10^−/− mice is their propensity to developing spontaneous colitis in response to their own intestinal microbiota, thereby limiting them to germ-free conditions.¹³

Sham et al. recently established SIGIRR ^−/− mice as a model for gastrointestinal infection by the mouse pathogens Citrobacter rodentium (a mouse-specific substitute for EPEC infection) and Salmonella enterica serovar Typhimurium, where in both cases, the mice displayed significantly increased intestinal inflammation in response to infection.¹⁴ In both humans and mice, SIGIRR acts as an inhibitor of MyD88-dependent signaling.¹⁵ In its absence, pro-inflammatory pathways involving MyD88, namely Toll-like receptors and IL-1R, exhibit substantially increased signaling in response to appropriate stimulation. Importantly, all other aspects of the immune system are kept intact and appropriately respond to the pathogen and these mice do not develop spontaneous intestinal inflammation in the absence of an infection.^15,16

Hypothesizing that SIGIRR deficiency might render the innate immune system of these mice more sensitive to a human pathogen such as C. jejuni, we inoculated, and successfully infected these mice with the C. jejuni 81–176 strain (a commonly used human isolate). As predicted, upon colonization, these mice quickly began displaying signs of enteritis, although this was an acute-self-limiting infection that resolved within a few weeks.⁹ Colonization in these mice was high (often over 10⁹ CFUs/g), but numbers began dropping by 10–14 days post-infection. At the height of infection, the mouse intestine displayed overt and histological signs of inflammation, including edema, immune-cell infiltration (particularly neutrophils), epithelial hyperplasia, and in the most severe instances, varying degrees of ulceration.⁹ Indeed, in many ways the pathology resembled acute human C. jejuni infection, so we further investigated the nature of C. jejuni colonization and infection in these mice.

Role of the Innate Immune System

One of the first questions we sought to answer was the avenue by which C. jejuni stimulated the immune system, and how it was affected by the SIGIRR deficiency of these mice. Previous work had already established both TLR2 and TLR4 as prominent receptors stimulated by C. jejuni, which we confirmed in vitro using reporter cell lines.^8,17 More interestingly, when we infected TLR4^−/−/SIGIRR^−/− double knockout mice, we determined that in the absence of TLR4, there were no signs of significant inflammation, indicating that TLR4 is one of the main drivers of inflammation in these mice.⁹ Mice deficient in TLR4, but still expressing SIGIRR were also completely unresponsive to the presence of C. jejuni. A previous study using MyD88^−/− mouse-derived macrophages also indicated that C. jejuni did not significantly stimulate an immune response in the absence of MyD88-dependent signaling,¹⁸ although a second study has suggested the existence of a TLR2-dependent, MyD88-independent response to C. jejuni in Caco-2 cells.¹⁹

Further insight into the role innate signaling plays during C. jejuni infection came from the infection of SIGIRR^−/− mice by C. jejuni mutants lacking the KpsM protein, the transporter protein necessary for the creation of the polysaccharide capsule. The capsule itself has been strongly linked to C. jejuni virulence, especially in an in vivo environment.^20,21 Several reports have indicated that these mutants trigger increased TLR signaling, and this was confirmed via our own in vitro reporter assays.^9,17 More interesting were the effects seen during in vivo infection of the SIGIRR^−/− and TLR2^−/−/SIGIRR^−/− mice by the ΔkpsM strain. Both knockout mouse strains displayed a very severe and rapid inflammatory response to colonization, which ultimately had only a modest negative effect on the bacteria, as indicated by a slight lag in intestinal colonization by the KpsM mutant as compared to WT C. jejuni.⁹ Since the polysaccharide capsule of C. jejuni has been previously shown to be an important feature of C. jejuni, its role as an immune modulator may be key to C. jejuni's virulence strategy.²¹ While live WT C. jejuni do stimulate TLR2 and TLR4 in vitro, they do so to a lesser degree than comparable numbers of heat-killed bacteria, indicating that viable bacteria obscure their PAMPs from innate receptors. This concealment was significantly reduced in the ΔkpsM mutant, with the mutant stimulating both receptors to a much higher degree.⁹ It would appear that one of the capsule's primary roles may be to limit the exposure of C. jejuni PAMPs to the innate immune system, thereby reducing the immune profile of the bacteria, effectively making C. jejuni more “stealthy” from an immunological perspective. Indeed, this is a strong indication that C. jejuni's main strategy for colonization may be to avoid immune detection as much as possible, despite its close association with the intestinal mucosa. C. jejuni's proposed interactions with the innate immune system are outlined in Fig. 1.

Figure 1.

A schematic outlining the steps of C. jejuni colonization and infection of the mammalian intestine, with details on how these steps relate to stimulation of TLR2 and TLR4. (A) Competition from competing commensal bacteria can prevent the initial colonization of C. jejuni. (B) If this is overcome, the bacteria come in contact with the intestinal epithelium, triggering an inflammatory response from the intestinal epithelium. (C) Pathogen-Associated Molecular Patterns (PAMPs), stimulate innate receptors, particularly TLR2 and TLR4. This stimulation is normally suppressed by SIGIRR, however, in its absence, the MyD88-dependent TLR signaling becomes elevated. C. jejuni itself can reduce this stimulation by masking its PAMPs with its polysaccharide capsule. What role its intracellular lifestyle may play in modulating inflammation currently remains unknown.

Crypt Colonization

Further insights into colonization and infection by C. jejuni were provided upon closer examination of the localization of this microbe within the GI tract. The highest numbers of C. jejuni, along with the majority of the inflammatory damage, were concentrated in the cecum and proximal colon. Numbers within the small intestine was relatively lower (<10⁷ CFUs/g), and relatively little inflammation was observed at this site.⁹ Surprisingly, immunofluorescent imaging of cecal and colonic cross-sections revealed relatively few C. jejuni within the central lumen. As illustrated in Fig. 2, Panel A, the vast majority of the labeled bacteria were concentrated within the relatively thin mucus layer lining the intestine and extending deep into the intestinal crypts. Indeed, the crypts themselves were often the principal site of colonization, with an extremely high density of C. jejuni often found within them. This finding is consistent with a number of previous in vitro and ex vivo studies, which suggested a close association between C. jejuni and the intestinal mucus layer.^22,23

Figure 2.

Diagram of C. jejuni colonization in our mouse model. C. jejuni (red, spiral-shaped bacteria), readily colonize the mucus layer and intestinal crypts, closer to the epithelium than the commensal microbiota. (A) An immunofluorescent image of a SIGIRR^−/− mouse cecum colonized with C. jejuni 81–176. C. jejuni (red) are visible within the mucus layer, and in large numbers in the crypts, but are sparse in the intestinal lumen. The mucus layer and associated goblet cells are made visible by labeling the mucin MUC2 (green) and the cell nuclei of the epithelium are stained using DAPI (Blue), 200x magnification. (B) Intracellular C. jejuni (red) are visible within epithelial cells along the top of the intestinal crypts, adjacent to the lumen. They are contrasted against Cytokeratin 19 (Green) and the nuclei (DAPI, Blue), 4000x magnification.

The mucus layer is a dynamic zone of large, interconnected mucin glycoproteins, meant to provide a barrier protecting the underlying epithelium from bacteria as well as physical debris in the intestinal lumen. To reach and infect the epithelium, pathogens must develop a means to traverse the mucus barrier. Conversely, most commensal bacteria lack the means to infiltrate this layer, so under normal circumstances the inner portions of the mucus layer are believed to be relatively sterile.^24,25 Lining the intestinal crypts are mucus-producing goblet cells, which release their contents into the crypt lumen, continually filling it with mucus. It is within this inner mucus layer, normally devoid of bacteria, where the majority of the C. jejuni were observed.⁹

The interrelationship between mucus and Campylobacter has been the subject of particular interest in recent years. Early work on C. jejuni, identified both crude mucus, mucins, and the mucin-associated sugar L-fucose as chemoattractants for C. jejuni.²⁶ Further investigations by Alemka et al. have described efficient growth of C. jejuni in the presence of mucus-secreting cell lines,²⁷ and other work has described growth of C. jejuni using crude mucus as a growth medium.²² Further elaboration into the nature of C. jejuni's growth in mucus has proven more elusive. It is known that C. jejuni prefers amino acids and organic acids as nutrient sources,²⁸ and some strains can utilize L-fucose in their metabolic pathways,²⁹ all of which would be plentiful within the intestinal mucus layer. Additionally, C. jejuni's need for a microaerobic environment containing low concentrations of oxygen might favor its growth in areas immediately adjacent to the host epithelium, as opposed to the mostly anaerobic environment of the intestinal lumen²⁸. Together, these findings illustrate that while C. jejuni may be relatively uncompetitive in relation to the normal gut microbiota, if given a chance, it can establish itself within the intestinal crypts and mucus layer. These sites lack the microbial competition of the intestinal lumen, while at the same time, are rich in many of C. jejuni's preferred nutrient sources and would provide it with a steady source of oxygen at low concentrations.

Cell Invasion in vivo

Another key aspect of C. jejuni infection is its ability to function as a facultative intracellular pathogen. Although C. jejuni cell invasion has been studied in in vitro cell-lines,^30-33 the role of cell invasion in vivo has remained a matter of much debate. Although the presence of intracellular C. jejuni has been indicated in infected human tissues,³⁴ little direct evidence or characterization has been carried out. Using immunofluorescent labeling for C. jejuni surface antigens to examine the infected cecal and colonic tissues of infected SIGIRR^−/− mice, we frequently observed intracellular C. jejuni within intestinal epithelial cells.⁹ These intracellular bacteria were confirmed to reside within LAMP-1 positive endosomal compartments, consistent with previous in vitro studies describing Campylobacter containing vacuoles (CCV).³⁰ Even more intriguing however, was the location of these bacteria. Although C. jejuni have been observed in polarized epithelial cell layers to pass through or between cells so they can invade the basolateral surface,³³ this process was not observed in our in vivo model. Rather, as illustrated in Fig. 1, panel B, CCVs were exclusively observed on the apical side of the nucleus, and no intracellular bacteria were observed in the basolateral regions of infected cells. Additionally, the epithelial cells containing CCVs were not evenly distributed, with virtually no intracellular bacteria found in epithelial cells deep within crypts, even when these crypts were highly colonized and C. jejuni were adherent to the epithelial surface. Rather, CCV-containing epithelial cells were found primarily among the most mature and superficial epithelial cells facing the intestinal lumen. Reasons for the differences between cell invasion in the in vitro and in vivo environments are not immediately evident, however, the substantial differences between in vitro cell monoculture and the complexities of the intact intestine are likely the leading factor.

More surprising was the lack of a direct correlation between intestinal inflammation and the presence of intracellular C. jejuni. Whereas high numbers of C. jejuni colonizing crypts often correlated with increasing signs of inflammation, CCV-containing epithelial cells were found in both asymptomatic WT and TLR4^−/−/SIGIRR^−/− mice, as well as inflamed SIGIRR^−/− and TLR2^−/−/SIGIRR^−/− mice.⁹ With the C. jejuni sensitive TLRs (TLR4 and TLR2) found primarily on the apical surface of epithelial cells or associated with immune cells,³⁵ it currently remains unclear whether epithelial cells are able to detect and respond to the intracellular C. jejuni found in their CCVs. Indeed, with CCV-containing epithelial cells showing no obvious signs of distress or damage, this may be another example by which C. jejuni remains undetected by the host innate immune system.

With confirmation now of C. jejuni's ability to invade epithelial cells under in vivo circumstances, we now need to define what role this invasion may play in infection. Intracellular replication has yet to be confirmed for C. jejuni, as does any role that intracellular bacteria may play in modulating or evading the immune system. To answer these questions a more prolonged and detailed examination of C. jejuni's intracellular lifecycle during the course of an infection will be necessary.

Implications for Human Campylobacter Infection

With these observations regarding C. jejuni colonization and infection in susceptible mice, the question remains as to how these results correlate with other known Campylobacter models of infection, and most importantly, how do they correlate with what we know about human infections. Crypt colonization has been described previously in germ-free mice using EM imaging,³⁶ and C. jejuni may also colonize the intestinal crypts of its avian hosts,³⁷ although evidence also suggests it does not associate closely with the avian gut epithelium in vivo.^37,38 Whether or not crypt colonization is a feature in other models of colonization is currently unknown. What is important now is further study regarding C. jejuni's survival within this niche, and whether these conditions differ between hosts, leading to differences in colonization. Current evidence would suggest differences in the mucus layer being another major factor in C. jejuni's lack of virulence in poultry, but details regarding a mechanism remain unknown.³⁸ Future work must establish C. jejuni's growth requirements within the mucus of the crypt, address how it survives there, and whether this can be manipulated to prevent the growth of C. jejuni.

Furthermore, the intracellular growth of C. jejuni requires significantly more study. Evidence suggests it takes place during human infection, but in vivo evidence of cell invasion in mice suggests that in vitro cell lines represent a poor tool for modeling this process. With so many subtle factors coming into play that seem to limit the scope of cell invasion to superficial epithelial cells at the top of crypts, we need more information regarding how cell-invasion works under in vivo conditions.

Overall, new in vivo animal models are shedding light on the nature of C. jejuni intestinal colonization and infection, but we are still only scratching the surface. There are substantial limitations to our in vitro tools, and the way forward is to study C. jejuni colonization and infection within the host directly.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.