Plasmacytoid Dendritic Cells Are Crucial in Bifidobacterium adolescentis-Mediated Inhibition of Yersinia enterocolitica Infection

In industrialized countries bacterial intestinal infections are commonly caused by enteropathogenic Enterobacteriaceae. The interaction of the microbiota with the host immune system determines the adequacy of an appropriate response against pathogens. In this study we addressed whether the probiotic Bifidobacterium adolescentis is protective during intestinal Yersinia enterocolitica infection. Female C57BL/6 mice were fed with B. adolescentis, infected with Yersinia enterocolitica, or B. adolescentis fed and subsequently infected with Yersinia enterocolitica. B. adolescentis fed and Yersinia infected mice were protected from Yersinia infection as indicated by a significantly reduced weight loss and splenic Yersinia load when compared to Yersinia infected mice. Moreover, protection from infection was associated with increased intestinal plasmacytoid dendritic cell and regulatory T-cell frequencies. Plasmacytoid dendritic cell function was investigated using depletion experiments by injecting B. adolescentis fed, Yersinia infected C57BL/6 mice with anti-mouse PDCA-1 antibody, to deplete plasmacytoid dendritic cells, or respective isotype control. The B. adolescentis-mediated protection from Yersinia dissemination to the spleen was abrogated after plasmacytoid dendritic cell depletion indicating a crucial function for pDC in control of intestinal Yersinia infection. We suggest that feeding of B. adolescentis modulates the intestinal immune system in terms of increased plasmacytoid dendritic cell and regulatory T-cell frequencies, which might account for the B. adolescentis-mediated protection from Yersinia enterocolitica infection.


Introduction
Infection with Yersinia enterocolitica e.g. by ingestion of contaminated food or drinking water can cause severe diarrhea, enterocolitis, and mesenteric lymphadenitis [1,2]. Yersinia enterocolitica is a facultative anaerobic, pleomorphic, gram-negative rod that belongs to the family of Enterobacteriaceae and its enteropathogenicity is associated with the presence of a 70-kb virulence plasmid (pYV) that encodes a type three secretion system, translocated effector proteins, and the trimeric autotransporter Yersinia adhesin A (YadA) [3,4].
Several studies demonstrate that the host's intestinal microbiota is crucial in defining the host's susceptibility towards intestinal infections. This is demonstrated by the significant influence of antibiotic treatment on the composition of the intestinal microbiota, in both, human subjects [5,6,7,8] and mice [9,10] where increased susceptibility towards enteropathogenic bacteria was shown [11,12]. The intestinal microbiota is thought to shape the innate immune system in different ways. It was demonstrated that antibiotic treatment of mice and subsequent alterations of the intestinal microbiota notably down-regulate the expression of Reg3c, a secreted C-type lectin which kills gram-positive bacteria including e.g. antibiotic-resistant bacteria such as vancomycin resistant Enterococcus (VRE) [13]. The secretion of Reg3c could be restored via stimulation of intestinal TLR4 thereby boosting the innate immune resistance of antibiotic-treated mice against infections with VRE [13]. In addition, antibiotic-induced disruption of the intestinal microbiota enhances the susceptibility of human hosts to infections with nontyphoidal Salmonellae [14], and is a prerequisite for infection of mice with Salmonella typhimurium [15,16].
Probiotics are defined as live microorganisms that, when administered in adequate amounts, confer a beneficial effect on the health of the host [17]. E. g. Lactobacillus rhamnosus GG, Bifidobacterium bifidum, Streptococcus thermophilus and Saccharomyces boulardii revealed a beneficial effect on children with rotavirus infections [18,19,20]. Furthermore, several studies summarized by Nomoto et al. report a decrease in the incidence of antibioticinduced diarrhea by administration of Saccharomyces boulardii, Lactobacillus rhamnosus GG, Bifidobacterium longum and Enterococcus faecium [21].
Dendritic cells (DCs) are essential initiators of immunity and link innate to adaptive antimicrobial immune responses [22]. In order to fulfill these tasks, first DC need to sample intestinal antigens. These are acquired by DC either in Peyer's patches, where M cells deliver luminal content by transcytosis [23,24], or DC can sample actively by extending dendrites into the lumen without disrupting epithelial tight junctions due to the expression of CX 3 CR1 [25]. However, this subset of DC was demonstrated to be non-migratory whereas CD103-expressing intestinal DC migrate to the mesenteric lymph nodes in a big way after antigen acquisition, for which they rely on other cell types [26]. In the intestine conventional DC (cDC) are made up by these two subsets which both can additionally express CD11b [26] and occur in the lamina propria as well as the Peyer's patches [25,27]. In addition to cDC, plasmacytoid DC exist and are defined by their intermediate expression of CD11c and additional expression of PDCA-1 and B220 [28]. They are found to a minor extent in secondary lymphoid organs [29] and the lamina propria but occur mainly in the intraepithelial compartment [28]. In response to TLR9 and TLR7 signalling pDCs produce high amounts of type I IFNs [26]. In addition, pDCs are thought to be of key importance for the regulation of tolerance and are considered to be inducers of regulatory T cells [30,31,32].
In previous work we identified a probiotic B. adolescentis strain that attenuated the course of Y. enterocolitica infection in mice by reducing clinical symptoms, dissemination of Yersinia, and Y. enterocolitica-induced mucosal inflammation [33]. In this study we demonstrate that feeding of viable B. adolescentis was associated with an increased number of PDCA-1-positive pDCs in the intestine and an attenuated course of Y. enterocolitica infection as indicated by reduced clinical symptoms and reduced dissemination of Yersinia.

Materials and Methods
Mice 6-10 weeks old female C57BL/6 mice were purchased from Harlan Laboratories, after transfer mice were housed under specific pathogen free conditions in isolated ventilated cages. Animal experiments were reviewed and approved by an appropriate institutional review committee (Genehmigung H2/05 Regierungsprä sidium Tübingen).

Cultivation of B. adolescentis and Oral Yersinia Infection of Mice
For feeding experiments, Bifidobacterium adolescentis Reuter 1963 (ATCC 15705) was anaerobically incubated in soy broth containing beef liver at 37uC for 48 h, then transferred into Bifidobacteria medium (formula according to M58-medium Leibniz Institute DSMZ-German Collection of Microorgansims and Cell Cultures) and cultivated under same conditions for additional 48 h. Bacteria were twice washed in phosphate buffered saline (PBS) (PAA) and thereafter resuspended in aseptic drinking water. 48 h prior to B. adolescentis feeding mice obtained drinking water containing streptomycin (20 g/l Sigma). Mice were intragastrically infected with 5610 8 plasmid harboring Y. enterocolitica WA-314 serotype O8 [34] as previously described [33]. Body weight development of mice was monitored daily, four days post infection mice were sacrificed by carbon dioxide asphyxiation, and subsequently the complete intestine, Peyer's patches, and the spleen were removed.

Antibiotic Treatment
Prior to B. adolescentis feeding, streptomycin containing drinking water (20 g/l Sigma) was administered to mice for 48 h in order to facilitate efficient B. adolescentis colonization. For B. adolescentis depletion experiments, mice received vancomycin (1 g/l Hexal) and metronidazole (1 g/l Sigma) containing drinking water for 48 h.

Determination of Colony Forming Units (CFU) of Yersinia
Fecal samples, Peyer's patches, and spleens were scaled. Tissue samples were homogenized by extruding through 40 mm cell strainer (BD Falcon). After, complete samples were resuspended in aseptic PBS, serial diluted until a factor of 10 24 , plated on CIN plates (Oxoid), and incubated for 48 h at 27uC. CFU were calculated per gram samples. The detection limit of CFU of Yersinia was 10.

Plasmacytoid Dendritic Cell Depletion
Mice were B. adolescentis fed and Yersinia infected as earlier described. In addition, mice were 1.5 days prior to and 0.5 days post Yersinia application intravenously injected with 150 mg of functional grade pure anti-mouse PDCA-1 antibody clone JF05-1C2.4.1 (Miltenyi Biotec) or functional grade purified Rat IgG2b isotype control (eBioscience).

Statistics
Statistical significances were calculated using unpaired t-test, when variances were statistically significantly different unpaired ttest with Welch's correction was used instead. For experiments with more than two investigated groups statistical significances were calculated using One-way analysis of variance followed by Tukey's Multiple Comparison Test. P values smaller than 0.05 were considered significant.

Results
The Bifidobacterium adolescentis-Mediated Protection from Yersinia enterocolitica Infection is Associated with an Increased Proportion of PDCA-1-Positive pDCs BY mice (3.2%67%) were significantly protected in terms of weight loss as compared to Y mice (29.3%63.4%; p = 0.0223) (Fig. 1A). Determination of CFU of Yersinia in the Peyer's patches revealed no differences between BY (log 10 3.761.8) and Y (log 10 3.461.8) mice (Fig. 1B). However, the BY mice were protected from dissemination of Yersinia during infection as indicated by the significantly reduced CFU of Yersinia in the spleens of B. adolescentis fed Y. enterocolitica infected mice (log 10 0.561.6) when compared to Y mice (log 10 2.562.5; p = 0.0036) (Fig. 1C). We were able to exclude that feeding of B. adolescentis reduced the number of intestinal enteropathogenic Yersinia in the BY mice, since BY (log 10 6.560.9) and Y mice (log 10 6.260.8) harbored comparable counts of Yersinia in the intestine (Fig. 1D).
These results are in line with previous results of our group demonstrating protective effects of B. adolescentis in both, inflammatory as well as infectious intestinal diseases [33].
Analysis of ieDC frequency in the intestine of differently treated mice, exhibited a significant increase in the percentage of ieDCs in Y mice (1.8%60.3%) as compared to remaining groups (M (0.9%60.1%) p,0.001, B (0.7%60.2%) p,0.001, BY (0.6%60.1%) p,0.001) ( Fig. 2A No significant differences in the percentage frequency of CD11c + CD11b + ieDC of Y mice were observed as compared to M mice (Fig. 2B). However, analysis of total numbers indicated a significant increase of CD11c + CD11b + ieDCs only in Y mice (Y (19. Table 1). The differences between absolute numbers of CD11c + CD11b + ieDCs between M mice and Y mice might be caused by an increased influx of CD11c + leukocytes to the Yersinia infected intestine ( Table 1).
Analysis concentrating on the percentage of CD11c + B220 + PDCA-1 + -expressing plasmacytoid (p)DCs demonstrated, however, that feeding of B. adolescentis The percentage frequency of pDC in BY mice might lead to the conclusion that BY mice need to have higher absolute cell counts as Y mice do. However, due to the fact that Y mice have in general a high absolute number of dendritic cells these mice even have with a low percentage frequency of pDC a high absolute frequency. Nevertheless, in Y mice the ratio between pDC and cDC is shifted in favor of cDC whereas in BY mice it is more balanced (Fig. 2, Table 1).
The analysis of the lpDCs gave comparable results: a significant decrease of CD11c + CD11b  Table S1, Table S2). To exclude that the enhanced frequency of pDCs in B and BY mice was due to streptomycin treatment prior to B. adolescentis administration C57BL/6 mice were treated with streptomycin (S) only and the frequency of intraepithelial pDCs was analyzed. Application of streptomycin did not result in an increase frequency of pDCs (M (8.1%62.9%, 1.1610 4 60.6610 4 ); S (9.5%63.6%, 1.2610 4 60.5610 4 ), therefore we can exclude the influence of streptomycin treatment or streptomycin caused alterations of the intestinal microbiota on the recruitment or onsite development of pDCs (Fig. 3A, Table 2). In order to demonstrate that the increased frequency of intraepithelial pDCs in B and BY mice is a result of B. adolescentis feeding, mice were either fed with B. adolescentis for 6 days or not and subsequently treated with vancomycin and metronidazole for 2 days. After 5 weeks the frequency of intraepithelial pDCs was determined. The antibiotics-induced reduction of B. adolescentis in the intestine resulted in comparable numbers of pDCs in antibiotic treated mock (28.4%610.9%, 16.8610 3 61.3610 3 ) as well as B. adolescentis fed and antibiotic treated mice (25.1%67%, 9.3610 3 63.1610 3 ). However, it remains elusive whether this is a direct effect of B. adolescentis or a secondary effect on the intestinal microbiota (Fig. 4A, Table 3).

B. adolescentis-Mediated Protection is Associated with an
Increase in FoxP3 + Regulatory T Cells Plasmacytoid DCs are described to induce both protective adaptive immunity as well as immune tolerance, depending on their localization and state of activation [22]. Moreover, this subset is associated with the induction of FoxP3 + CD4 + CD25 + -expressing regulatory T cells (T reg cells) [35] therefore we next analyzed the frequency of CD3e + CD4 + cells and of FoxP3 + CD4 + T reg cells.
From the above results, we conclude that feeding of B. adolescentis might promote development of T reg cells. In order to examine the inhibitory function of B. adolescentisinduced CD11c int B220 + PDCA-1 + plasmacytoid (p) DCs on Y. enterocolitica dissemination we performed pDC depletion experiments. Therefore we injected B. adolescentis fed C57BL/6 mice with anti-mouse PDCA-1 antibody or respective isotype control prior to and during Yersinia infection.
Based on the presented results we hypothesize that feeding of B. adolescentis might lead, by a yet unknown mechanism, to recruitment or increased onsite development of pDCs, which might promote development of regulatory T cells. We suggest that this immune response favors intestinal homoeostasis rather than inflammation, strengthens the intestinal barrier and thereby contributes to inhibition of Yersinia enterocolitica dissemination.

Discussion
Commensals are thought to modulate the susceptibility of the host towards intestinal infections. We recently showed that commensal B. adolescentis protects mice from Yersinia enterocolitica infection [33]. In line with these results the present study demonstrates that B. adolescentis-mediated protection is associated with increased frequency of T reg cells and pDCs. This DC subset seems to be crucial for protection since depletion of pDCs abrogated protection.
The intestinal microbiota exerts a variety of functions thereby strengthen the host resistance against intestinal infections. The high bacterial density and diversity in the mammalian intestine contributes to the prevention from infections with opportunistic and strict pathogens. This mechanism is known as colonization resistance and disturbances of microbial composition leads to an increased susceptibility to e.g. enteric infections [36] or Clostridium difficile. [37]. However, in this study colonization resistance seems not to be the underlying mechanism of B. adolescentis-mediated protection since Yersinia loads in Peyer's patches (PP) and fecal samples were comparable between groups.
In addition, the metabolism of the host depends on the microbiota to degrade otherwise indigestible nutritional carbohydrates which results in the release of short chain fatty acids providing an additional energy source. [38]. Metabolic byproducts fulfill additional tasks indicated by the fact that released short chain fatty acids attenuate intestinal inflammation [39] and protect from enteropathogenic infection [40] in mice.
Moreover, commensals are important to maintain and increase the hosts intestinal epithelial barrier function which helps to defend invasive pathogens. In intestinal epithelial cells (IEC) the recognition of commensal components and metabolites can result in the induction of tight junctions [41,42], cell growth [43], as well as mucus production [44,45]. Furthermore, the presence of commensals leads to secretion of antimicrobial compounds by IEC [46] enhancing the intestinal innate immune response to pathogens [13,47]. In addition to the host, bacteria secrete   antimicrobial compounds themselves, termed bacteriocins, able to change the microbial composition [48]. The presence of B. adolescentis might protect from Yersiniosis owing to an increase in epithelial barrier function either by direct interactions with IEC or by alterations of the microbiota composition. The host's resistance to pathogens relies on a fully functional gastrointestinal immune system, which develops from an immature state only after microbiota acquisition [49]. Moreover, the microbiota complexity directly influences intestinal dendritic cell subset composition. A less diverse microbiota resulted in reduced frequency of plasmacytoid dendritic cells in lymphoid tissues [50] and administration of the probiotic preparation VSL#3 decreased the pDC content in the lamina propria of mice. Our results additionally proof the influence of the microbiota diversity on DC composition since B. adolescentis feeding was associated with an increase in pDCs. These findings suggest that bacterial compounds might directly induce on-site development or the recruitment of plasmacytoid DCs into the intestinal compartment. The rise in pDCs might depend on invariant NKT (iNKT) cells since a recent study demonstrated that iNKT cells are responsible for the recruitment of pDCs to the pancreas during lymphocytic choriomeningitis virus infection [51]. In this study the release of glycolipids by B. adolescentis might result in the activation of iNKT cells which in turn may initiate the recruitment of pDCs to the intestine. The crucial function of pDCs is demonstrated by abolished B. adolescentis-mediated protection during Yersiniosis after depletion of pDCs. Depending on location the secretion of type I interferons (IFN) by pDCs can differ, since murine splenic pDCs secret high amounts of IFN-a upon TLR9 ligand stimulation whereas pDC of PP don't. In addition, treatment of splenic pDCs with IL-10, prostaglandine E2, and TGF-b, present at mucosal sites, prevents IFN induction, indicating that the mucosal microenvironment might condition pDCs for poor Type I IFN production [52]. In contrast, peripheral blood pDCs of IBD patients are impaired to secret IFN-a and simultaneously secret increased amounts of pro-inflammatory cytokines after TLR9 ligand challenge [30]. Moreover, IBD patients exhibit an increased frequency of pDCs in the intestinal mucosa when compared to controls [30]. A rise in pDC frequency may be supposed to attenuate inflammation which might be insufficient owing to reduced IFN-a secretion. In line with this is the fact that type I interferons secreted by DCs ameliorate DSS-induced colonic injury and inflammation in mice [53] and that ifnar 2/2 deficient mice are more susceptible to DSS [53,54]. However, IFN-b seems not to be protective in general since administration of IFN-b producing Lactobacilli to mice exacerbates DSS-induced disease [54]. In addition, might result the B. adolescentis-mediated increase in pDCs in a decreased migration rate of CD103-positive DC to the mesenteric lymph nodes due to a reduced expression of   CCR7. Since, in vitro stimulated bone marrow-derived DC exhibited a reduced expression of CCR7 after stimulation with cytokines when IFN-b was present [55]. CD103-expressing DC were shown to be responsible for the transport of living Salmonella typhimurium cells to the mesenteric lymph nodes in a CCR7 dependent manner [26]. Moreover, a recent publication demonstrated that CD103-positive DC are able to sample, in addition to CX 3 CR1 expressing DC, luminal Salmonella typhimurium cells by the production of transepithelial dendrites [56]. In this study Y. enterocolitica might also be transported by CD103-expressing DC to the mesenteric lymph nodes. Therefore B. adolescentis-induced pDCs might secret IFN-b which act on CD103-epxressing cells resulting in a lower expression of CCR7 and a decreased migration rate to the mesenteric lymph nodes. This in turn might prevent the transport of Y. enterocolitica to the mesenteric lymph nodes. The potential of pDCs to induce regulatory T ( Treg ) cells was demonstrated for human [35] and murine cells [57]. Constitutively present T reg are essential for the maintenance of intestinal homeostasis and to control inflammation [58]. Especially during inflammation the immunosuppressive function of T reg is crucial since a great deal of intestinal pathology originates not from the infection, instead is owing to an overwhelming immune response [59]. In addition, T reg cells are of importance since inflammation enhances the ability of enteric pathogens to establish infection [60]. Moreover, human IL-10 prevented pathology in H. pylori infected il-10-deficient mice and is accompanied by an increase in T reg cells [61]. In contrast, during H. pylori infection T reg cells can be of disadvantage since H. pylori manipulates DCs to become tolerant resulting in the induction of regulatory T cells which dampen the immune response and therefore enable long-lasting gastric colonization [62]. Furthermore, T reg cells are important in parasitic infections given that Toxoplasma gondii infection-induced pathology [63] and Schistosoma mansoni-caused colonic granulomatous [64] pathology are attenuated by T reg cells. B. adolescentis feeding-induced T reg cells might dampen intestinal inflammation resulting in unfavorable environmental conditions for Y. enterocolitica to establish a systemic infection. Yersinia pseudotuberculosis was demonstrated to disseminate directly from the intestinal lumen to the spleen and not to travel via PP and mesenteric lymph nodes [65]. In this study reduced inflammation caused by T reg cells might maintain intestinal barrier integrity and therefore prevent Y. enterocolitica dissemination.
Whether B. adolescentis directly interacts with the host mucosal immune system, or whether an indirect effect of B. adolescentis reduces the host susceptibility towards Yersinia infection by e.g. alteration of the microbiota remains to be elucidated.
Our data provide evidence that commensal B. adolescentis increases the frequency of pDC and T reg cells and that this might be essential for host resistance to intestinal infection.   Numbers indicate mean percentages 6 SD and mean cell counts 6 SD of B. adolescentis fed and Yersinia infected mice, either injected with anti-mouse PDCA-1 (n = 5) or respective isotype control (n = 3). a p = 0.0072.