NADPH Oxidase Deficient Mice Develop Colitis and Bacteremia upon Infection with Normally Avirulent, TTSS-1- and TTSS-2-Deficient Salmonella Typhimurium

Infections, microbe sampling and occasional leakage of commensal microbiota and their products across the intestinal epithelial cell layer represent a permanent challenge to the intestinal immune system. The production of reactive oxygen species by NADPH oxidase is thought to be a key element of defense. Patients suffering from chronic granulomatous disease are deficient in one of the subunits of NADPH oxidase. They display a high incidence of Crohn’s disease-like intestinal inflammation and are hyper-susceptible to infection with fungi and bacteria, including a 10-fold increased risk of Salmonellosis. It is not completely understood which steps of the infection process are affected by the NADPH oxidase deficiency. We employed a mouse model for Salmonella diarrhea to study how NADPH oxidase deficiency (Cybb −/−) affects microbe handling by the large intestinal mucosa. In this animal model, wild type S. Typhimurium causes pronounced enteropathy in wild type mice. In contrast, an avirulent S. Typhimurium mutant (S.Tmavir; invGsseD), which lacks virulence factors boosting trans-epithelial penetration and growth in the lamina propria, cannot cause enteropathy in wild type mice. We found that Cybb −/− mice are efficiently infected by S.Tmavir and develop enteropathy by day 4 post infection. Cell depletion experiments and infections in Cybb −/− Myd88 −/− mice indicated that the S.Tmavir-inflicted disease in Cybb −/− mice hinges on CD11c+CX3CR1+ monocytic phagocytes mediating colonization of the cecal lamina propria and on Myd88-dependent proinflammatory immune responses. Interestingly, in mixed bone marrow chimeras a partial reconstitution of Cybb-proficiency in the bone marrow derived compartment was sufficient to ameliorate disease severity. Our data indicate that NADPH oxidase expression is of key importance for restricting the growth of S.Tmavir in the mucosal lamina propria. This provides important insights into microbe handling by the large intestinal mucosa and the role of NADPH oxidase in maintaining microbe-host mutualism at this exposed body surface.


Introduction
The intestinal immune system is capable of handling occasional breaches by the microbiota and by mucosal-invading pathogens. This is facilitated by efficient secondary barriers, such as the large number of specialized lymphoid and myeloid cells of the gutassociated immune system (e.g. Peyer's patches and isolated lymphoid follicles) and the lamina propria (LP) of the absorptive mucosa. Normally, commensals and pathogens which breach the epithelial layer are taken up, killed, processed and presented by diverse phagocytes, in particular by diverse mononuclear phagocyte populations and polymorphonuclear leukocytes/granulocytes (PMN). Therefore, these populations are thought to play an important role in limiting bacterial loads in the LP and preventing disease.
The antimicrobial repertoire of PMN includes proteases and reactive oxygen species (ROS) produced by the NADPH oxidase complex, containing CYBB [2]. Interestingly, NADPH oxidase deficiency leads to a pronounced susceptibility to bacterial infection and inflammatory disease [3,4]. This condition is termed chronic granulomatous disease (CGD) and is traceable to genetic disruptions of NADPH oxidase, i.e. in approximately 65% of cases to mutations of the Cybb gene encoding the cytochrome b-245 H chain catalytic subunit [5]. CGD patients are highly susceptible to systemic infection and/or granuloma formation by Staphylococcus spp., Mycobacterium spp., Salmonella spp., Aspergillus spp., Pseudomonas spp. and Burkholderia cepacia and chronic gut inflammation resembling inflammatory bowel diseases [5][6][7]. The latter indicates that NADPH oxidase is of significant importance for limiting microbe growth and/or access to the LP and/or regulation of inflammation in the intestine [8,9].
To analyze NADPH oxidase mediated defense in the intestinal mucosa, we have employed a mouse model for Salmonella enterica subspecies 1 serovar Typhimurium (S. Typhimurium) diarrhea. CGD patients display an approximately 10-fold increased rate of infection with Salmonella spp. than the normal population and Salmonella spp. have been isolated from stools of CGD patients with intestinal inflammation [6,7,10]. Similarly, S. Typhimurium grows in systemic sites in NADPH oxidase deficient and in PMNdepleted mice [3,[11][12][13][14][15][16][17]. However, the importance of Cybb expression by PMN in preventing mucosal infection has not been fully understood.
Two different versions of the streptomycin pretreated mouse model for S. Typhimurium diarrhea [18] were of particular interest for probing NADPH oxidase function in the gut. In the standard model [19], mice are infected with wild type S. Typhimurium and develop a pronounced gut inflammation in the cecum. In contrast, isogenic S. Typhimurium mutants lacking type three secretion system (TTSS)-1 and TTSS-2, responsible for the secretion of virulence factors boosting epithelial cell invasion and pathogen growth within LP phagocytes, do not cause disease. In a second version of this model, which employs S. Typhimurium mutants lacking a functional TTSS-1 (e.g. SL1344 DinvG, S. Tm invG ), the pathogen relies on CD11c + CX 3 CR1 + monocytic phagocytes to traverse the epithelial barrier, grows within CD11c 2 CX 3 CR1 2 monocytes of the LP and causes overt mucosal inflammation 3 days post infection (p.i.) [20,21]. This model allows analysis of pathogen virulence factors (e.g. TTSS-2; [20]) as well as the mechanisms used by the host to restrict pathogen growth within mucosal monocytic phagocytes [18].
Using these well-established mouse models for S. Typhimurium colitis, we have analyzed the role of NADPH oxidase in the infected mucosa. Our findings might be of general importance for understanding pathogen and commensal handling by the mucosal immune system and might help to understand the effects of a partial restoration of Cybb-functionality in CGD patients by gene therapy or bone marrow transfer.

Results
Cybb 2/2 Mice Fail to Control Infection with a Normally Avirulent S. Typhimurium Mutant To analyze the role of NADPH oxidase in mucosal defense, we have worked in the genetic background of S. Tm invG (lacking TTSS-1). S. Tm invG requires CD11c + CX 3 CR1 + monocytic phagocytes for traversing the epithelial barrier, grows within the LP and elicits enteropathy in a Myd88-dependent fashion by day 3 p.i. in wild type mice. This has been termed the ''alternative pathway'' [18,21]. We speculated that this pathway might be particularly sensitive for NADPH oxidase deficiency, as Cybb might help restricting bacterial growth in the LP.
We pretreated wild type and Cybb 2/2 mice with streptomycin and infected them for 4 days with S.Tm avir (5610 7 cfu by gavage) to analyze the role of Cybb in restricting the growth of S.Tm avir in the mucosal tissue. High S.Tm avir loads were detected in the gut lumen of wild type and Cybb 2/2 mice (Fig. 1A). Bacterial loads in the LP were significantly lower in the wild type than in the Cybb 2/2 animals (Fig. 1B) and only the latter developed pronounced mucosal inflammation by day 4 p.i. (Fig. 1C). Furthermore, the Cybb 2/2 mice displayed significantly increased loads of S.Tm avir in the mesenteric lymph nodes (mLNs, Fig. 1D), the livers (Fig. 1E) and the spleens (Fig. 1F) compared to C57BL/6 mice. This high susceptibility to systemic spread was expected as NADPH oxidase is known to be of key importance for limiting systemic infections [3,4]. Our data extended these findings by showing that NADPH oxidase is essential for restricting the growth of S.Tm avir not only at systemic sites, but also in the cecal LP.
iNOS does not Contribute Significantly to Mucosal Defense against S.Tm avir The inducible NO synthase (iNOS) is an important defense mechanism of monocytic macrophages [2] and can help restricting pathogen growth in various models [22][23][24][25][26][27]. In order to assess the role of iNOS in our infection model, we included iNOS-deficient (Nos2 2/2 ) animals and Cybb 2/2 Nos2 2/2 double KO mice into the infection experiments with S.Tm avir shown in Fig. 1. Neither cecum pathology (Fig. S1C) nor tissue loads (Fig. S1B, D-F) in the Nos2 2/2 mice differed from wild type C57BL/6 animals. Similarly, the cecum pathology (Fig. S1C) and the tissue loads in the cecal mucosa (Fig. S1B) and the livers (Fig. S1E) did not differ significantly between the Cybb 2/2 and the Cybb 2/2 Nos2 2/2 mice. The Cybb 2/2 Nos2 2/2 animals displayed slightly but significantly elevated S.Tm avir tissue loads only in the mLNs (Fig. S1D) and the spleens (Fig. S1F). However, even in these organs, Cybb-deficiency had a more pronounced effect than Nos2-deficiency and significant contributions of Nos2 were only detectable in the presence of Cybb, suggesting a possible synergistic role for Nos2 [16]. In conclusion, restriction of S.Tm avir in the cecal mucosa and the protection from enteropathy seems to hinge on NADPH oxidase while iNOS seems to contribute little (maximally in a synergistic manner) to mucosal defense, at least during the first 4 days of infection.

S. Typhimurium
The standard streptomycin model for murine S. Typhimurium diarrhea was used to assess Cybb expression in the infected mucosa of wild type C57BL/6 mice. Streptomycin pretreated animals were infected with wild type S. Typhimurium (S.Tm wt ; 5610 7 cfu by gavage) for 12 or 24 h. Samples of the cecum tissue (the site of the initial and most pronounced enteropathy [18,19]) were recovered to analyze Cybb expression by reverse transcription quantitative real-time PCR (RT-qPCR). In line with earlier data [28], the abundance of Cybb mRNA in the cecum increased by about 3-fold after 12 h and about 8-fold after 24 h of infection compared to streptomycin-treated animals (Fig. S2A). This went along with mucosal inflammation and infiltration of neutrophils and monocytic phagocytes into the cecal mucosa as observed by histopathology and flow cytometry analysis ( The role of Cybb and or PMN in the acute infection is not completely understood. Thus, we performed a number of control experiments to analyze the pathogenetic mechanism of S.Tm avir colitis in Cybb 2/2 mice. First, we analyzed uninfected and infected gut tissues with immuno-histopathological stainings for markers characteristic for a set of immune cells. The S.Tm avir infected mucosa of Cybb 2/2 mice at day 4 p.i. displayed a patchy pathology characterized by non-inflamed regions interspaced with pronounced inflammatory foci (Fig. S3). Such patchy pathology is characteristic for S.Tm invG (lacking only TTSS-1) infections at day 3 p.i. in wild type C57BL/6 mice ( Fig. S4; [20]). This provided first hints suggesting that Cybb-deficiency allows S.Tm avir to elicit enteropathy via the ''alternative pathway''. In line with this, RT-qPCR analysis of a panel of 27 genes for cytokines or antimicrobial defenses revealed similar mucosal gene expression profiles ( Fig.  S4G) in Cybb 2/2 mice (4 days p.i. with S.Tm avir ) and wild type C57BL/6 mice (3 days p.i. with S.Tm invG ) at their first day of overt enteropathy ( Fig. S4). In this experiment, the counts in the mLNs (Fig. S4D), livers (Fig. S4E) and spleens (Fig. S4F) were lower in the C57BL/6 mice compared to Cybb 2/2 mice. However, most importantly and in line with the RT-qPCR data, the degree of cecum inflammation was alike (Fig. S4B, C).
We then assessed the Myd88-dependency of the inflammatory response as a typical feature of the ''alternative pathway''. For this purpose we infected Cybb 2/2 Myd88 2/2 mice or Cybb 2/ 2 Myd88 +/2 littermate controls for 4 days with S.Tm avir (5610 7 cfu by gavage). High loads of S.Tm avir were observed in the gut lumen ( Fig (Fig. 2C). It is interesting to note that the Cybb 2/2 Myd88 2/2 animals displayed slightly but significantly elevated S.Tm avir loads in the cecal epithelium ( Fig. 2B; blue symbols). It is unclear whether this might be explained by reduced epithelial turnover rates of non-infected tissue in Myd88 2/2 animals [29]. Alternatively, this might be indicative of a Myd88-dependent, but Cybb-independent defense mechanism which may contribute to limiting bacterial growth in the enterocytes. Such mechanisms could be an interesting topic for future research. To this end, the data verified the Myd88dependency of enteropathy in S.Tm avir infected Cybb 2/2 mice.
Finally, we have assessed the dependency on mucosal monocytic phagocytes. Using transgenic mice expressing the diphtheria-toxin receptor under control of the CD11c promoter (DTR +/2 ; [30]) and diphtheria toxin-mediated (DTX) cell depletion, it has been previously established that S.Tm invG relies on mucosal CD11c + monocytic phagocytes for traversing the gut epithelium and colonizing the cecal LP [21]. Thus, Cybb 2/2 DTR +/2 mice or Cybb 2/2 littermates were treated with DTX and infected for 4 days with S.Tm avir . High loads of S.Tm avir were detected in the gut lumen of both groups (Fig. 3A). In contrast, the DTX-mediated cell depletion abolished mucosa tissue infection (Fig. 3B) and the elicitation of mucosal inflammation (Fig. 3C). Furthermore, it significantly reduced the infection of mLNs (Fig. 3D), livers ( Fig. 3E) and spleens (Fig. 3F). These data were all in line with the notion that S.Tm avir infection of Cybb 2/2 mice follows a similar pathogenetic mechanism as described earlier for the S.Tm invG infection in wild type mice. The key difference between both infections seems to reside in the failure of the Cybb 2/2 mice to control S.Tm avir infection/growth in the infected mucosa. This would suffice to explain the susceptibility of Cybb 2/2 (but not wild type) mice to S.Tm avir -triggered enteropathy and suggests that the model might be of interest for studying the role of Cybb in restricting bacterial growth in the cecal LP.

S.Tm avir Induces Intermediate Levels of Enteropathy in Mice Reconstituted with a Mix of Cybb 2/2 and Wild Type Bone Marrow
To further analyze how Cybb restricts S.Tm avir infection of the cecal mucosa, we performed an experiment with mixed bone marrow chimeric mice. The chimeras were generated by reconstituting irradiated Cybb 2/2 mice (congenic marker Ly5.2) with a mix of 50% Cybb 2/2 (congenic marker Ly5.2) and 50% C57BL/6 (congenic marker Ly5.1) bone marrow. After 8 weeks, these chimeras displayed 69% Cybb 2/2 (congenic marker Ly5.2) and 31% C57BL/6 (congenic marker Ly5.1) cells in the cecal LP as tested by flow cytometry at the end of the experiment (data not shown). In these mice, the stromal cells and all CD45.2 + cells (i.e. phagocytes, B-cells, T-cells, etc.) were Cybb-deficient, while the CD45.1 + cells were Cybb-proficient. Four days p.i. with S.Tm avir all chimeric mice (mixed BMC, Fig. 4) displayed high pathogen loads in the gut lumen (Fig. 4A) and significant amounts of bacteria in the mLNs (Fig. 4C) and spleens (Fig. 4E). The levels of mLN and spleen colonization in the mixed chimeras were lower than in the Cybb 2/2 mice, but higher than in the wild type C57BL/6 animals (Fig. 4C, E). The same intermediate phenotype was observed for the cecum pathology (Fig. 4B), whereas liver colonization was not distinguishable from wild type C57BL/6 animals (Fig. 4D). Interestingly, approximately one third of Cybb-proficient cells in a Cybb-knock out background lead to a reduction of bacterial counts in spleens and livers by 217-and 333-fold (ratio between the medians), respectively, if compared with mice completely deficient for Cybb. Additionally, RT-qPCR analysis of proinflammatory cytokines confirms the alleviated inflammatory phenotype (Fig. 4F). These data indicate that Cybb-proficiency in only 31% of the bone-marrow-derived compartment is sufficient to achieve a significant restriction of S.Tm avir colonization of the host tissue and enteropathy.

Discussion
Here, we have analyzed NADPH oxidase defenses of the intestinal mucosa. We established that NADPH oxidase deficient mice were not able to limit gut mucosa colonization and enteropathy by a normally avirulent S. Typhimurium strain. This demonstrated that disease via the ''alternative'' pathway hinges on a fine balance between microbe entry into the LP, microbe growth at this site and pathogen killing in the LP. Our data confirmed that LP access is controlled at least in part by dendritic cells (monocytic phagocytes), and demonstrated that microbe growth/killing is controlled by bacterial virulence factors (e.g. TTSS-2) and host defenses (e.g. NADPH oxidase-mediated killing in PMN).
While the central role of NADPH oxidase, i.e. CYBB, is well established in antimicrobial defense, the nature of the cell types facilitating the NADPH oxidase dependent defenses had remained less clear. The role of NADPH oxidase in the anti-microbial activity of neutrophils is well established [21], the same holds true for its role in dendritic cell-mediated antigen presentation and T- cell priming [31][32][33]. However, what is the mechanism activating NADPH oxidase in the mucosal phagocytes? Besides NADPH oxidase-deficiency/CGD, other primary immune deficiencies enhancing susceptibility to bacterial infection are deficiencies in Toll-like receptor-and IFNc-R-signalling [34,35]. In mouse models for systemic and intestinal Salmonella infection, Toll-like receptor -and IFNc-R-signalling were indeed found to restrict pathogen growth [20,[36][37][38][39][40][41][42]. NADPH oxidase (and iNOS) are known to be activated via both MyD88-and IFNc-signalling. However, S.Tm avir did not colonize the LP of MyD88 2/2 or IFNc-R 2/2 mice and did not cause enteropathy (this work and data not shown). This indicated that NADPH oxidase is activated by several redundant signalling pathways in LP cells. Deciphering such signalling pathways and the cell type mainly responsible for NADPH oxidase expression will be an interesting topic for future work. The S.Tm avir infection model would be well suited for such studies, because S. Tm avir offers well defined genetics and virulence factors. The removal of the latter from S. Tm wt still leads to disease in a mouse model of CGD. This indicates that even very low virulence is sufficient to cause enteropathy in mice deficient in a subunit of the NADPH oxidase, broadening our understanding of how commensals might induce enteropathy in CGD patients.
The diffusible nature of some of the ROS (i.e. hydrogen peroxide) has raised some interest, as neighboring cells might be affected, even if they are not by themself capable of expressing NADPH oxidase. This has in fact been demonstrated in vitro [43,44] and has thus complicated the interpretation of data from mouse experiments with cell-type specific NADPH oxidase deficiencies [45].
Our data demonstrate that the augmentation by neighboring cells (by ROS-signalling or by wild type cell mediated decreases of the pathogen loads) might be indeed of importance, at least in the S.Tm avir infected intestinal mucosa. The mixed Cybb-proficient and -deficient bone marrow chimeras displayed .2006 lower systemic S.Tm avir loads than the Cybb 2/2 controls. Apparently, 31% of Cybb-proficient CD45 + cells are sufficient for this. This is in line with other publications focusing on A. fumigatus infections. In vitro, A. fumigatus hyphae could be damaged by a mixture of normal and ''CGD neutrophils'' [46]. Furthermore, Cybb 2/2 mice with .92% Cybb-deficient and 4-8% Cybb-proficient cells were fully protected [46,47] to challenge with a dose of A. fumigatus sufficient to cause disease in Cybb 2/2 mice. Furthermore, the reported amount of Cybb-proficient cells necessary to respond similarly to an infection (i.e. survive) as wild type mice is 21-35% or 32-41% for challenge with S. aureus or B. cepacia, respectively [47]. Similarly, survival of CGD patients after entering adulthood was strongly associated with residual reactive oxygen intermediates production [48]. In extension, our data and the evidence from the other infection models discussed above indicate that even a partial therapy of CGD patients might be sufficient to significantly decrease their disease susceptibility far beyond the degree of achieved reconstitution. The need for less than 100% reconstitution (as typically observed in gene therapy [49]) might be of relevance for preclinical testing and the design of gene therapy approaches for treating CGD.
Up to date, it is unclear why a partial restoration of Cybb expression is sufficient to ameliorate the phenotype drastically. There are three possible explanations. Firstly, we showed recently that S. Tm invG is found exclusively in CD11c + cells at 1 day p.i. in our infection mouse model and only from 2 days p.i. on also in CD11c 2 cells [21,50]. This mechanism might also apply for the S. Tm avir infection mouse model, since the S. Tm invG infection in C57BL/6 mice seems to be phenotypically similar to S. Tm avir infection in Cybb 2/2 mice. The transition between CD11c + and CD11c 2 cells can possibly be the reason for an incidental exposure (and killing) of some S. Tm avir bacteria to Cybb-proficient phagocytes. Killing in the Cybb-proficient phagocyte populations could explain the reduced tissue loads and disease pathology of the mixed bone marrow chimeras. Secondly, ROS produced by Cybb-proficient cells play an important role in controlling signalling pathways. Here, the reversible oxidation and inactivation of protein tyrosine phosphatases and MAP kinase phosphatases by ROS are interesting examples [51]. As a measure for proinflammatory signalling levels, we quantified mRNA levels of 27 genes related to inflammation and defense against S. Typhimurium infection. However, although 6 out of 27 genes were expressed less in the S.Tm avir infections of mixed bone marrow chimeras compared to Cybb 2/2 deficient mice, 21 out of 27 gene expression levels were similarly induced, if induced at all. This might indicate that only a part of the signalling pathways are affected by Cybb-deficiency.
Thirdly, we cannot exclude that the, already discussed, diffusion of some of the ROS (i.e. hydrogen peroxide) from Cybb-proficient cells into neighboring Cybb-deficient cells described in vitro [43,44] may also occur in vivo. The current data is insufficient to tease apart these three mechanistic explanations. Nonetheless, the reduced disease severity of S. Tm avir infections in the mixed bone marrow chimeras provides a useful basis for addressing this issue.
In CGD patients, Crohn's disease-like inflammation of the intestinal mucosa is frequently observed [7,52]. Salmonella spp. can be isolated from the stools of some, but clearly not from all of these patients [7]. This indicates that, on the one hand, growth restriction of normally avirulent Salmonella by NADPH oxidase may be of relevance for CGD patients, but on the other hand that other microbial stimuli can also trigger enteropathy. In the ''non-Salmonella-related'' cases, inflammation might be attributable to insufficient restriction of commensal microbiota species which would not cause disease in NADPH oxidase proficient hosts. In this case, NADPH oxidase-mediated growth restriction by LP cells may function as an immunological barrier of general importance for maintaining homeostasis in the intestinal mucosa. Our findings might be of importance for understanding microbe handling by the intestinal immune system and for elucidating strategies employed by pathogens to overcome this defense.
Our findings may also contribute to our understanding of the evolution of S. Typhimurium as a successful enteropathogen. During its divergence from a commensal E. coli lineage, this pathogen has acquired two novel genetic loci of importance for enteropathogensis which encode the two TTSSs [53][54][55]. In wild type hosts, TTSS-2 was shown to enhance pathogen survival in LP phagocytes and thereby enhance mucosal inflammation [20,21,[56][57][58]. Tissue culture experiments suggested that this is attributable to TTSS-2 dependent interference with NADPH oxidase (or iNOS-) delivery to the Salmonella containing phagosome [59,60]. This is supported by our finding that S.Tm avir , which lacks a functional TTSS-2, is only capable of colonizing the LP and cause enteropathy in Cybb-deficient, but not in wild type mice. In conclusion, TTSS-2 may represent a pathogen-specific adaptation to overcome and subvert the NADPH oxidase mediated mucosal defense. This would explain how wild type S. Typhimurium colonizes these cells of the intestinal mucosa in wild type hosts [21,53].
Apparently, greatly reduced virulence of S. Typhimurium is sufficient to cause enteropathy in Cybb mice. This is clearly not due to deficiency in immune regulation, because bacterial species not recognized as pathogenic are capable of triggering enteropathy in CGD. S. Tm avir are a highly genetically amenable tool to study the mechanisms why. Direct and indirect susceptibility to ROS may be a determining feature of host microbiota species that permits their close relationship with the host.

Ethics Statement
All animal experiments and generation of new mouse-lines were approved by the legal authorities (licenses 201/2007 and 223/ 2010; Kantonales Veterinä ramt Zürich, Switzerland) and carried out in the legally required manner.

Flow Cytometry
Cecum and mLNs were chopped and digested in RPMI (Invitrogen) and Liberase TL (Roche) for 45 min at 37uC under vigorous shaking. The resulting cell suspension was filtered through a 100 mm nylon cell-strainer (Milian) and stained in buffer containing PBS, 5 mM EDTA, 10% FCS and 50 mg/ml streptomycin. All fluorophore-labeled monoclonal antibodies were purchased from BD Biosciences or Biolegend. The LP cells were analyzed on a LSR II cytometer (Becton Dickinson) and graphs were produced with FlowJo software (Tree Star, Inc.).

In vivo Dendritic Cell Depletion
DTX was injected i.p. (100 ng/25 g body weight; [64]) at 18 h before and 30 h after the infection. The depletion efficiency (.80%) and its negligible effect on other mucosal cell populations have been described before [21].

Histopathological Evaluation
Tissues were embedded in OCT (Sakura, Torrance, CA) and snap-frozen in liquid nitrogen. Five mm cryosections were stained with hematoxylin and eosin (H&E). The degree of cecal mucosal tissue inflammation, i.e. edema, PMN infiltration, reduced numbers of goblet cells containing visible mucus-filled vacuoles and epithelium disruption, was judged by a pathologist yielding to a score of inflammation between 0-13 points as described before [19,66].

RT-qPCR
The excised cecum tissue was washed in cold PBS, placed in 600 ml RNAlater (Qiagen) and subsequently frozen at 280uC. Total RNA extraction was done using the RNeasy mini kit (Qiagen) with RNase-free DNase digest (Qiagen). For reversetranscription of 1 mg mRNA aliquots, the RT 2 HT First Strand cDNA Kit (Qiagen) was used. Custom RT 2 Profiler PCR Arrays (Qiagen) were run with RT 2 SYBR Green ROX FAST (QIAGEN) on an Applied Biosystems 7900 HT Fast Real-Time PCR System to amplify the resulting cDNA. Relative mRNA levels (2 2DCq ) were determined by comparing the PCR quantification cycle (Cq, determined with the Software SDS 2.2.1) for 27 genes related to inflammation and defense against S. Typhimurium infection (the selection is based on Songhet et al., 2010) with the reference gene Actb. The differences in their Cq cycles were calculated (DCq). In all experiments, the upper limit of Cq was fixed to 35 cycles. Then, the fold-increase over streptomycintreated C57BL/6 mice was calculated and plotted. Each sample was controlled for mouse genomic DNA contamination. All DNApositive data were excluded from further analysis. Lastly, RNA quality was monitored with the Agilent RNA 6000 Nano Kit (Agilent Technologies) on a 2100 Bioanalyzer (Agilent Technologies) and only samples with a RNA integrity number (RIN) .9.90 were included.

Statistical Analysis
Statistical analysis was performed using the exact Mann-Whitney U test with the software GraphPad Prism 6. Values of p,0.05 (two tailed) were considered as significantly different between two groups. The minimal detectable bacterial colonization levels were set to 10 cfu/mLNs, 20 cfu/spleen, 60 cfu/liver ( Fig. 1-4) or 30 cfu/liver (Fig. S4) or 10 cfu/g cecum content in cases where no bacteria were detected by plating. Messenger RNA levels of two groups were compared using Mann-Whitney U tests with Hochberg corrections for multiple comparisons using R x64 3.0.1 (Fig. 4F, S4G). Figure S1 NADPH oxidase is expressed in the infected mucosa and PMNs increase in number by infection. C57BL/6 mice were pretreated with streptomycin and infected with S.Tm wt for 12 h or 24 h, as indicated. RT-qPCR for Cybb expression in cecal tissues (A). Representative H&E sections (contrast and brightness were adjusted, color was enhanced, scale bar: 50 mm, arrow indicates a PMN) (B). Quantity of PMNs/highpower field (C). FC of cecal LP (pregated on CD45 + cells) (D). *: p,0.05; ns: not significant; red line: median; dashed line: detection limit. (TIF) Figure S2 Cybb (but not iNOS) is important in mucosal defense against S.Tm avir infection. C57BL/6 mice (data replotted from Fig. 1), Nos2 2/2 mice (C57BL/6 background), Cybb 2/2 Nos2 2/2 mice (C57BL/6 background) or Cybb 2/2 mice (C57BL/6 background; data replotted from Fig. 1) were pretreated with streptomycin and infected for 4 days with S.Tm avir . The bacterial loads in the gut lumen (A), the LP (red (B)) or the epithelial cells of the cecum (blue (B)), the degree of mucosal inflammation (C) and bacterial loads in the mLNs (D), livers (E) and spleens (F) were analyzed. *: p,0.05; **: p,0.01; ns: not significant; red line: median; dashed line: minimal detectable value. (TIF) Figure S3 Immunohistology of S.TminvG infected wild type C57BL/6 mice and S.Tmavir infected Cybb 2/2 mice is similar. Cryo-sections of the cecal tissue from streptomycin pretreated wild type and Cybb 2/2 mice infected for 3 days with S.Tm invG or for 4 days with S.Tm avir , were stained with antibodies against CD11c (A), CD11b (B), CD68 (C), Gr-1 (D), CD3 (E) and CD8 (F) and imaged by bright field microscopy. The different times of infection are explained by the different disease kinetics of S.Tm invG and S.Tm avir . The former requires 3 days (in C57BL/6 mice) and the latter 4 days (in Cybb 2/2 mice) before overt inflammation of the cecal tissue is observed. The left panel shows representative pictures. The right panel shows the quantification. *: p,0.05; **: p,0.01; ns: not significant. Data is displayed as mean + SEM. S.Tm invG was able to elicit gut inflammation in wild type C57BL/6 and in Cybb 2/2 mice. In contrast, S.Tm avir triggered enteropathy only in the Cybb 2/2 mice, but not in wild type C57BL/6 animals. Please note that the inflammatory lesions in the S.Tm avir infected Cybb 2/2 mice displayed localized inflammatory lesions of equivalent immunohistopathology as the lesion triggered by S.Tm invG in C57BL/6 mice. (TIF) Figure S4 S.Tm invG infection in wild type C57BL/6 mice and S.Tm avir infection in Cybb 2/2 mice are similar. C57BL/6 mice were pretreated with streptomycin and infected with S.Tm invG for 3 days. Cybb 2/2 mice were pretreated with streptomycin and infected with S.Tm avir for 4 days. The bacterial loads in the gut lumen (A), the degree of mucosal inflammation (B), representative H&E pictures (contrast and brightness were adjusted and color was enhanced, scale bar: 200 mm, C) and bacterial loads in the mLNs (D), livers (E) and spleens (F) were analyzed. *: p,0.05; **: p,0.01; ns: not significant; red line: median; dashed line: minimal detectable value. Relative mRNA expression levels were compared between S.Tm invG infected C57BL/6 mice and S.Tm avir infected Cybb 2/2 mice, data replotted partly in Figure 4 (G). Data is displayed as mean + SEM, differences were not significant (G). (TIF)