Arcobacter butzleri Induce Colonic, Extra-Intestinal and Systemic Inflammatory Responses in Gnotobiotic IL-10 Deficient Mice in a Strain-Dependent Manner

Background The immunopathological impact of human Arcobacter (A.) infections is under current debate. Episodes of gastroenteritis with abdominal pain and acute or prolonged watery diarrhea were reported for A. butzleri infected patients. Whereas adhesive, invasive and cytotoxic capacities have been described for A. butzleri in vitro, only limited information is available about the immunopathogenic potential and mechanisms of infection in vivo. Methodology/Principal Findings Gnotobiotic IL-10-/- mice were generated by broad-spectrum antibiotic treatment and perorally infected with the A. butzleri strains CCUG 30485 and C1 shown to be invasive in cell culture assays. Bacterial colonization capacities, clinical conditions, intestinal, extra-intestinal and systemic immune responses were monitored at day six and 16 postinfection (p.i.). Despite stable intestinal A. butzleri colonization at high loads, gnotobiotic IL-10-/- mice were virtually unaffected and did not display any overt symptoms at either time point. Notably, A. butzleri infection induced apoptosis of colonic epithelial cells which was paralleled by increased abundance of proliferating cells. Furthermore A. butzleri infection caused a significant increase of distinct immune cell populations such as T and B cells, regulatory T cells, macrophages and monocytes in the colon which was accompanied by elevated colonic TNF, IFN-γ, nitric oxide (NO), IL-6, IL-12p70 and MCP-1 concentrations. Strikingly, A. butzleri induced extra-intestinal and systemic immune responses as indicated by higher NO concentrations in kidney and increased TNF, IFN-γ, IL-12p70 and IL-6 levels in serum samples of infected as compared to naive mice. Overall, inflammatory responses could be observed earlier in the course of infection by the CCUG 30485 as compared to the C1 strain. Conclusion/Significance Peroral A. butzleri infection induced not only intestinal but also extra-intestinal and systemic immune responses in gnotobiotic IL-10-/- mice in a strain-dependent manner. These findings point towards an immunopathogenic potential of A. butzleri in vertebrate hosts.


Introduction
The motile and spiral-shaped gram-negative Arcobacter (A.) species belong to the family of Campylobacteraceae and can be isolated from a broad range of habitats. In animals Arcobacter spp. are mostly reported as gastrointestinal commensals [1]. Among the 19 so far described Arcobacter spp., A. butzleri and A. cryaerophilus have been rated as serious hazards for human health by the International Commission on Microbiological Specifications for Foods [2]. Given that detection of Arcobacter spp. may fail in applied bacteriologic routine diagnostic procedures, the prevalence of Arcobacter associated human diseases is not known so far. On the other hand, a huge number of single clinical cases and few outbreaks reported in the literature point towards an important role of these bacteria in causing intestinal diseases [3,4]. Briefly, in retrospective studies several authors demonstrated that Arcobacter spp. are the fourth most common Campylobacterales species recovered from patients suffering from diarrhea [5][6][7]. Diseased patients have been shown to present symptoms of gastroenteritis including abdominal pain, acute diarrhea or prolonged watery diarrhea for up to two months [5,6]. So far, limited information is available about the underlying mechanisms of infection and the host immune responses. Results from phenotypic assays revealed adhesive, invasive and cytotoxic capabilities of A. butzleri on several cell lines in vitro [8][9][10][11][12][13][14]. The barrier dysfunction caused by A. butzleri infection in monolayers of the human colon cell line HT-29/B6 highlights potential mechanisms by which diarrhea is induced in susceptible human hosts [15]. In order to study the pathogenic potential of A. butzleri in more detail in vivo, we monitored the colonization properties of two A. butzleri strains, for which invasive capacities have been shown in in vitro assays [8], and the subsequent host responses following peroral infection. We first infected conventional wildtype mice with distinct A. butzleri strains in order to investigate whether the strains are capable of colonizing the mice at all. We then included gnotobiotic IL-10 -/mice in which the microbiota was virtually depleted by broad-spectrum antibiotic treatment into our experiments. In previous infection studies with enteric pathogens such as Campylobacter (C.) jejuni we could demonstrate that the physiological colonization resistance exerted by mice harboring a conventional microbiota prevented the animals from infection and could be overcome following eradication of the murine intestinal microbiota [16]. Furthermore, gnotobiotic IL-10 -/mice developed infection-induced immunopathological key features of human campylobacteriosis within six days following peroral C. jejuni infection [17][18][19][20]. Given that A. butzleri (formerly termed C. butzleri [21]) is related to C. jejuni, we further unraveled colonization and immunopathological features of A. butzleri infection in the gnotobiotic IL-10 -/mouse model. Corresponding results point towards an infectious and pro-inflammatory potential of A. butzleri which favours gnotobiotic murine models for the further study of pathogenicity factors of arcobacteriosis in vivo. By parallel investigation of two A. butzleri strains in the same experimental set-up also strain-dependent variances in the immunopathogenic potential could be observed.

Ethics statement
All animal experiments were conducted according to the European Guidelines for animal welfare (2010/63/EU) with approval of the commission for animal experiments headed by the "Landesamt für Gesundheit und Soziales" (LaGeSo, Berlin, registration number G0184/12). Animal welfare was monitored twice daily by assessment of clinical conditions. Mice C57BL/6j wildtype and IL-10 -/mice (in C57BL/6j background, B6) were bred and maintained in the facilities of the "Forschungseinrichtungen für Experimentelle Medizin" (FEM, Charité-Universitätsmedizin, Berlin, Germany) under specific pathogen-free (SPF) conditions. Gnotobiotic IL-10 -/mice (with a virtually depleted gastrointestinal microbiota) were generated by broad-spectrum antibiotic treatment as described earlier [22]. In brief, mice were transferred to sterile cages and treated by adding ampicillin/sulbactam (1 g/L; Pfizer, Berlin, Germany), vancomycin (500 mg/L; Hexal, Holzkirchen, Germany), ciprofloxacin (200 mg/L; Hexal), imipenem (250 mg/L; Fresenius Kabi, Graz, Austria), and metronidazole (1 g/L; Braun, Melsungen, Germany) to the drinking water ad libitum starting at three weeks of age immediately after weaning and continued for 3-4 months before the infection experiment [19]. Three days before infection, the antibiotic cocktail was replaced by sterile tap water (ad libitum). Mice were continuously kept in a sterile environment (autoclaved food and tap water) and handeled under strict antiseptic conditions.

Arcobacter butzleri infection of mice
Conventionally colonized wildtype and gnotobiotic IL-10 -/mice (all female) were infected with approximately 10 9 viable colony forming units (CFU) of either A. butzleri strain CCUG 30485 or strain C1 by gavage in a total volume of 0.3 mL PBS on two consecutive days (day 0 and day 1).
The A. butzleri reference strain CCUG 30485 was initially isolated from a fecal sample of a diarrheal patient [21], whereas the C1 strain was derived from fresh chicken meat [8]. Both A. butzleri strains were grown on Karmali-Agar (Oxoid, Wesel, Germany) for two days at 37°C under microaerobic conditions using CampyGen gas packs (Oxoid).

Sampling procedures
Mice were sacrificed by isofluran treatment (Abbott, Greifswald, Germany). Cardiac blood and tissue samples from colon, liver and kidney were removed under sterile conditions. Absolute large intestinal lengths were determined by measuring the distances from the ascending colon leaving the cecum to the rectum by a ruler. Intestinal samples from each mouse were collected in parallel for immunohistochemical, microbiological, and immunological analyses. Immunohistopathological changes were determined in colonic ex vivo biopsies immediately fixed in 5% formalin and embedded in paraffin. Sections (5 μm) were stained with hematoxylin and eosin (H&E) or respective antibodies for in situ immunohistochemistry as described earlier [19].

Quantitative analysis of Arcobacter butzleri colonization and translocation
Before infection of gnotobiotic mice, absence of commensal intestinal microbiota was confirmed as described previously [16][17][18][19][20]. Viable A. butzleri were detected in feces or at time of necropsy (day 6 or 16 postinfection; p.i.) in luminal samples taken from the colon, dissolved in sterile phosphate buffered saline (PBS) and serial dilutions cultured on Karmali-and Columbia-Agar supplemented with 5% sheep blood (Oxoid) for two days at 37°C under microaerobic conditions using CampyGen gas packs (Oxoid). To quantify bacterial translocation, liver and kidney ex vivo biopsies were homogenized in 1 mL sterile PBS, whereas cardiac blood (200 μL) was directly streaked onto Karmali agar and cultivated accordingly. The respective weights of fecal or tissue samples were determined by the difference of the sample weights before and after asservation. The detection limit of viable pathogens was 100 CFU per g.

Cytokine detection in serum and culture supernatants of intestinal and extra-intestinal ex vivo biopsies
Colonic ex vivo biopsies were cut longitudinally and washed in PBS. Kidney, liver or strips of approximately 1 cm 2 colonic tissue were placed in 24-flat-bottom well culture plates (Nunc, Wiesbaden, Germany) containing 500 μL serum-free RPMI 1640 medium (Gibco, life technologies, Paisley, UK) supplemented with penicillin (100 U/ mL) and streptomycin (100 μg/ mL; PAA Laboratories). After 18 h at 37°C, culture supernatants and serum samples were tested for IFN-γ, TNF, MCP-1, IL-6, IL-12p70 by the Mouse Inflammation Cytometric Bead Assay (CBA; BD Biosciences, San Jose, CA, USA) on a BD FACSCanto II flow cytometer (BD Biosciences). Nitric oxide (NO) was determined by Griess reaction as described earlier [22].

Statistical analysis
Medians and levels of significance were determined using Mann-Whitney test (GraphPad Prism v5, La Jolla, CA, USA) as indicated. Two-sided probability (P) values 0.05 were considered significant. Experiments were reproduced twice.

Results
Colonization capacities of A. butzleri in infected gnotobiotic IL-10 -/mice Given that mice can exhibit a strong colonization resistance against pathogens such as C. jejuni due to their host-and age-specific microbiota composition [16], we first addressed the question whether different A. butzleri strains (namely strain CCUG 30485 initially isolated from a diarrheal patient and strain C1 derived from fresh chicken meat) were able to stably establish intestinal colonization in wildtype mice with a conventional microbiota. Within 24 hours following the latest of two consecutive peroral infections with 10 9 viable A. butzleri of either strain by gavage, mice expelled either pathogen from their intestines as indicated by culture-negative fecal samples and did not display any infection-induced clinical symptoms (data not shown).
To overcome colonization resistance and investigate colonization as well as immunopathogenic properties of A. butzleri in vivo, we applied the gnotobiotic IL-10 -/mouse model. This animal infection model was chosen since within one week upon C. jejuni infection gnotobiotic IL-10 -/mice harbored high pathogenic loads and displayed non-selflimiting acute symptoms of infection-induced enterocolitis such as bloody diarrhea and wasting syndrome [17,19]. Daily survey of A. butzleri loads in fecal samples revealed that following infection with 10 9 viable A. butzleri strains CCUG 30485 or C1, gnotobiotic IL-10 -/mice could be stably colonized with pathogenic loads of 10 8 CFU per gram fecal sample (Fig 1). Despite high bacterial loads, however, mice did not display any overt infection-induced symptoms such as diarrhea or occurence of blood in feces at either time point (not shown).
Induction of apoptosis in the colon of A. butzleri infected gnotobiotic IL-10 -/mice We next raised the question whether A. butzleri was able to induce intestinal inflammatory responses in colonized, but clinically unaffected mice. To address this, gnotobiotic IL-10 -/mice were sacrificed as early as six days postinfection (p.i.) or to a later time point at day 16 p.i. Given that acute intestinal inflammation is accompanied by significant shortening of the intestinal tract [17,22], we determined absolute lengths of the large intestines, but did not observe differences between mice infected with either strain, neither at day six nor at day 16 p.i., as compared to age-and sex-matched naive control animals (not shown).
We further determined potential immunopathological responses in A. butzleri colonized animals. Histopathological changes in H&E stained colonic paraffin sections derived from with either A. butzleri strain colonized mice were rather subtle and did not differ at either time point as indicated by comparable histopathological scores (not shown). Since apoptosis is a commonly used diagnostic marker in the histopathological evaluation and grading of intestinal disease [16] and a hallmark of C. jejuni induced enterocolitis in gnotobiotic IL-10 -/mice [17], we quantitatively assessed numbers of caspase-3+ cells within the colonic mucosa of colonized mice. At day six p.i. with the A. butzleri strain CCUG 30485, but not strain C1, mice displayed significantly higher numbers of apoptotic cells in the colonic epithelium as compared to uninfected controls (p<0.005 ; Fig 2A). At day 16 p.i., however, animals infected with either strain exhibited more than twofold higher caspase-3+ cell numbers in their colon versus naive mice (p<0.05; Fig 2A). As Ki67 comprizes a nuclear protein associated with and necessary for cellular proliferation [24], we stained colonic paraffin sections against Ki67 to determine proliferative measures of the colonic epithelium counteracting apoptosis following A. butzleri infection. Colonic epithelial Ki67+ cell numbers increased in strain CCUG 30485 infected mice until day six (p<0.05; Fig 2B), but reached levels observed in naive mice at day 16 p.i., whereas following C1 strain infection colonic proliferating cells further increased during the course of infection (p<0.05; Fig 2B).

Large intestinal innate and adaptive immune cell responses in A. butzleri infected gnotobiotic IL-10 -/mice
Since recruitment of pro-inflammatory immune cells to sites of inflammation is a well known key feature of enteric pathogen infection (e.g. campylobacteriosis [16]), we next quantitatively assessed the influx of innate and adaptive immune as well as of effector cells into the colonic mucosa and lamina propria by in situ immunohistochemical staining of large intestinal paraffin sections. Within six days following A. butzleri CCUG 30485 strain infection, CD3+ cell numbers (i.e. T lymphocytes) increased by more than twofold (p<0.0005; Fig 3A), but declined back to numbers observed in naive mice until day 16 p.i. In C1 strain infected mice, however, an increase in colonic T cells of approximately 25% in gnotobiotic IL-10 -/mice could be observed until day six as well as day 16 p.i. (p<0.05 and p<0.0005, respectively; Fig 3A). Numbers of FOXP3+ regulatory T cells (Tregs) were higher in the large intestines of gnotobiotic IL-10 -/at either time point and irrespective of the A. butzleri strain (p<0.05-0.0001; Fig 3B). Whereas during C1 strain infection Tregs further increased until day 16 p.i., lower FOXP3+ cell numbers could be detected in colons of CCUG 30485 strain infected mice at day 16 as compared to day six p.i. (p<0.05; Fig 3B). When compared to naive mice, numbers of B220+ B lymphocytes were significantly higher in CCUG 30485 strain infected mice, both at days six and 16 p.i. (p<0.0001 and p<0.05, respectively; Fig 3C), and 16 days following C1 strain infection (p<0.0005; Fig 3C). Interestingly, kinetic changes of F4/80+ macrophages and monocytes in CCUG 30485 and C1 strain infected mice were comparable to those seen with CD3+ cells: Whereas in colons of C1 strain infected mice F4/80+ cell numbers increased by approximately 50% until day 6 following infection and remained on comparable levels until day 16 p.i. (p<0.0005 and p<0.05, respectively; Fig 3D), numbers of colonic macrophages and monocytes in CCUG 30485 strain infected animals peaked at day six p.i. and reached naive levels ten days thereafter (Fig 3D).

Induction of large intestinal pro-inflammatory immune responses in A.
butzleri infected gnotobiotic IL-10 -/mice In order to further assess large intestinal inflammatory sequelae of A. butzeri infection, we next measured pro-inflammatory cytokines in colonic ex vivo biospies. TNF protein levels increased upon A. butzleri infection with either strain and at both time points (p<0.05-0.0005; Fig 4A). Interestingly, TNF levels were more than 20 times higher in CCUG 30485 strain infected mice at day six p.i. as compared to naive mice, but declined significantly, still to an elevated level,  Table).
Taken together, despite absence of overt clinical symptom, A. butzleri infected mice exhibited colonic apoptosis accompanied by increased abundance of proliferating cells compensating for the intestinal damage. Furthermore an increase in immune cell populations in the large intestines could be observed which was accompanied by increased local pro-inflammatory mediator expression upon A. butzleri infection.

Lack of obvious bacterial translocation, but induction of extra-intestinal and systemic pro-inflammatory immune responses in A. butzleri infected gnotobiotic IL-10 -/mice
We next determined whether following A. butzleri infection bacterial strains might be able to translocate from the intestinal tract to extra-intestinal and systemic compartments. Irrespective  Table). of the A. butzleri strain, we were not able to detect any viable bacteria by direct plating (with a lower detection limit of 10 2 CFU/g sample) from liver, kidney or cardiac blood-neither at day six nor at day 16 p.i. (not shown). In the following, we raised the question whether intestinal A. butzleri infection might induce pro-inflammatory immune responses at distant sites such as liver, kidney and serum. In H&E stained paraffin sections derived from liver and kidney samples no histopathological changes could be observed. Moreover, pro-inflammatory cytokine levels in livers of infected and naive mice did not differ at either time point (not shown). In kidneys, however, NO levels increased until day six following A. butzleri strain CCUG 30485 and C1 infection (p<0.005 and p<0.05, respectively; Fig 5), whereas at day 16 p.i. NO secretion in renal ex vivo biopsies were higher following CCUG 30485 (p<0.05; Fig 5), but not C1 strain infection. Remarkably, further distinct systemic pro-inflammatory cytokine responses could be observed in serum samples derived from A. butzleri infected gnotobiotic IL-10 -/mice. As in colon, IFN-γ levels in serum increased multifold within six days following CCUG 30485 strain infection only, but decreased to naive levels until day 16 p.i. (p<0.005; Fig 6A). Furthermore, IL-12p70 serum levels increased following CCUG 30485, but not C1 strain infection peaking at day 6 p.i. (p<0.005; Fig 6B). Interestingly, serum levels of TNF and IL-6 increased rather late following CCUG 30485 and C1 strain infection as indicated by increased levels at day 16, but not day six p.i. (p<0.05-0.005; Fig 6C and 6D).  Table). In summary, peroral Arcobacter infections resulted not only in local (i.e. intestinal), but also in significant extra-intestinal and systemic immune responses in a strain dependent manner.

Discussion
The immunopathological impact of Arcobacter infections in vivo is under current debate. We here investigated the colonization and pro-inflammatory potential of peroral Arcobacter butzleri infection with two different strains in mice. As for other intestinal pathogens, mice harboring a conventional microbiota were protected from infection given that mice had expelled the respective A. butzleri strain within 24 hours p.i. despite dual peroral infection with high loads. This result is well in line with our previous C. jejuni infection studies showing that mice harboring a conventional microbiota were protected from colonization and expelled the pathogen within 48 hours p.i. [16]. To overcome physiological colonization resistance we applied gnotobiotic mice with a virtually depleted microbiota following broad-spectrum antibiotic treatment and chose the gnotobiotic IL-10 -/mouse model that was recently shown very well suitable to elucidate immunopathological mechanisms of C. jejuni infection mimicking key featues of human campylobacteriosis such as acute enterocolitis within six days p.i. [17,19,20]. Interestingly, despite high intestinal A. butzleri loads, however, gnotobiotic IL-10 -/did not display any overt symptoms such as diarrhea or occurence of blood in feces. Despite lack of clinical symptoms, A. butzleri induced colonic apoptosis which was paralleled by increased abundance of proliferating cells. This is well in line with an in vitro study showing a threefold increased epithelial caspase-3 dependent apoptosis induction in A. butzleri infected HT-29/B6 cells contributing to epithelial barrier dysfunction [15]. In another in vitro study, A. butzleri infection of THP-derived macrophages resulted in an increased activity of caspase-3 among caspase-7 and -8 which was accompanied by an upregulated expression of pro-inflammatory cytokines such as TNF, IL-6 and IL-12 [25] as shown in colon and serum of infected mice in our present in vivo study. Interestingly, despite initially increased caspase-3 levels, DNA damage was virtually absent suggesting potential counter-regulatory measures on the cellular level in THP-1 cells [25]. In line with this, we here demonstrate that A. butzleri induced large intestinal apoptosis was accompanied by increased numbers of Ki67+ proliferating cells in the colonic epithelium potentially counteracting the epithelial damage in the clinically unaffected infected mice. In previous in vivo studies, invasive capacities and virulence of A. butzleri were highly dependent on species and breed of the host and on the respective pathogenic strain [26]. In neonatal piglets, for instance, A. butzleri could be detected in small intestinal ex vivo biopsies and displayed rather invasive properties given that viable pathogens could be isloated from extra-intestinal organs such as liver, kidney, and even the brain after enrichment [27]. Eventhough we were unable to detect A. butzleri by direct plating from extra-intestinal compartments including liver and kidney in our study, we did detect higher NO levels in ex vivo biopsies derived from kidneys of infected mice underlining the potency of A. butzleri to induce not only intestinal but also extra-intestinal inflammatory collateral damages upon infection. Five days following oral infection with A. butzleri, albino rats were shown to present with diarrhea that was resolving within 21 days p.i., and small intestinal as well as hepatic necrosis [28]. The authors also described leukocytic infiltrates in the intestinal lamina propria, which is well  Table). doi:10.1371/journal.pone.0139402.g004 in line with our results given that not only increased numbers of T and B cells, but also of regulatory T cells, macrophages and monocytes could be detected in the large intestinal mucosa  Table).   Table). concentrations of leukocytes and neutrophils in serum samples following infection with an A. butzleri strain in a pathogen-load dependent manner [29]. A systemic pro-inflammatory response following A. butzleri infection was also evident in our study given that in serum samples of infected mice a plethora of pro-inflammatory cytokines such as TNF, IFN-γ, IL-6 and IL-12 were upregulated. Diarrhea in A. butzleri shedding rats was self-limiting and could be observed for up to five weeks p.i. pointing towards a potential etiologic role in human diarrhea [30]. In contrast, A. butzleri was uncapable of colonizing conventional chicken and turkey poults, whereas Beltsville white turkeys displayed highly variable, and A. butzleri strain dependent colonization and mortality rates [26].
In the present study we were also assessing the dynamics of inflammatory responses in the course of infection with two different A. butzleri strains. Whereas the A. butzleri reference strain CCUG 30485 had been initially isolated from a diseased patient [21], the C1 strain was derived from fresh chicken meat [8]. It is therefore not known if the latter is also able to induce disease in humans. Overall, the most distinct influx of immune cell populations such as T cells, Tregs, monocytes and macrophages into the colonic lamina propria could be observed six days following CCUG 30485 strain infection which was paralleled by peaking concentrations of pro-inflammatory mediators such as TNF, IFN-γ, NO and IL-6 in the large intestines, and IFN-γ and IL-12p70 in serum samples of infected mice declining therafter until day 16 p.i. Interestingly, other pro-inflammatory cytokines such as TNF and IL-6 increased later in the course of infection with either strain and were higher at day 16 as compared to day six p.i., pointing towards differentially regulated immune responses upon A. butzleri infection. It is hence highly likely that different A. butzleri strains might exert distinct host-dependent immune responses given that some strains cause overt disease in humans whereas others act as commensals like in chicken, for instance [31]. Furthermore, in vitro assays revealed differences in adhesive and invasive potentials of several A. butzleri strains [8,9,32]. However, no correlation between these phenotypes and the corresponding virulence gene pattern or functional domains of adhesion and invasion associated genes could be determined [8,9]. Nevertheless, the A. butzleri strains CCUG 30485 and C1 applied in our gnotobiotic IL-10 -/model, displayed a similar pattern of virulence genes and comparable adhesive and invasive capabilities in in vitro assays [8]. Taken together, eventhough both strains stably colonized the murine large intestines they induced slightly distinct local and systemic host responses in infected gnotobiotic IL-10 -/mice. This suggest that further, so far unraveled, virulence associated genes might be encoded by A. butzleri, contributing to different immunopathological potencies of respective A. butzleri strains.
In conclusion, A. butzleri induces not only intestinal but also extra-intestinal and systemic immune responses in gnotobiotic IL-10 -/mice following peroral infection. This points towards an immunopathogenic and differentially regulated role of A. butzleri in vertebrate hosts. Moreover, these results highlight gnotobiotic IL-10 -/mice as a valuable infection model to further unravel the underlying molecular mechanisms of A. butzleri induced pathogenicity in the near future.