Introduction

Nearly 30 years ago, two hallmark papers studying acute peritoneal infections reported evidence for protective functions of mast cells in anti-bacterial defense^1,2. These experiments were performed in mice carrying inactivating mutations in the gene encoding the receptor tyrosine kinase Kit (Kit^W/Wv mice) which, at the time, were the widely used model of mast cell deficiency owing to the dependency of mast cell development on Kit expression³. Both reports showed that Kit^W/Wv mice were highly susceptible to peritoneal infections compared with Kit wild-type mice. Based on mast cell reconstitution and investigation of associated cytokines, the central conclusion was that the absence of mast cells in Kit^W/Wv mice, and in particular the absence of mast cell-derived tumor necrosis factor alfa (TNFα), was responsible for the observed immunodeficiency^1,2. Subsequently, mast cells were considered as key effectors in bacterial defense⁴.

At the time, it was well established that Kit mutations affect many cell lineages and tissues in developing and adult mice beyond the defect in mast cells, and hence it remained ambiguous whether or not a phenotype observed in Kit-mutant mice was indeed due to the absence of mast cells. With the advent of Kit mutation-independent mouse models of mast cell deficiency^5–8, which are by comparison to Kit-mutants highly specific experimental models^9–13, mast cell functions can be probed more conclusively in vivo. It has since become possible to independently confirm, or disprove, what had been concluded earlier on mast cell physiology based on work in Kit-mutants. Around 15 years into this process of re-evaluation of mast cell functions, a long list of suggested functions have not been reproduced when revisited in Kit-independent models (for discussions and references, see^9–13).

Here, we have re-addressed the question whether mast cells are involved in immunity to acute peritonitis and sepsis. We confirmed the susceptibility of Kit-mutant mice (Kit^W/Wv) to sepsis¹ in the cecal ligation and puncture (CLP) model¹⁴. However, comparison of Kit wild-type mice with (Cpa3^+/+) or without (Cpa3^Cre/+) mast cells⁶ showed that the outcome of such infections was independent of mast cells, i.e. mast cells were not required for resistance to polymicrobial sepsis in the CLP assay. Furthermore, Kit^W/Wv mice exhibited the same susceptibility as Kit^+/+ mice upon direct injection of intestinal bacteria, pointing at a defect in Kit^W/Wv mice related to the intestinal injury or the release of endogenous flora from the cecum in the CLP model. Finally, we identified markedly higher Escherichia coli (E. coli) colony forming units in the intestines of Kit^W/Wv mice compared to Kit^+/+ mice. The release of higher numbers of E. coli from the Kit^W/Wv mouse intestine during CLP sepsis induced a more lethal infection. Accordingly, co-housing neutralized the susceptibility of Kit-mutant mice to CLP-induced sepsis. Our data suggest that the dysbiosis of Kit^W/Wv mutants has been misinterpreted as immunodeficiency. Collectively, we present evidence that sepsis susceptibility of Kit-mutant mice is due to elevated microbiota pathogenicity, whereas mast cells do not play a role in protecting against polymicrobial sepsis.

Results

Mast cells are dispensable for survival of cecal ligation and puncture-induced polymicrobial sepsis

We subjected mast cell-deficient Cpa3^Cre/+ mice⁶ on the C57BL/6 background (Cpa3^Cre/+), and their wild-type littermates (Cpa3^+/+) to the surgical model of cecal ligation and puncture¹⁴. Under mild sepsis conditions, induced by a 50 % ligation of the cecum (tying at the middle between the base and the pole of the cecum) and a single puncture of the cecum with a 25-gauge (25 G) needle, all wild-type mice (Cpa3^+/+) survived (Fig. 1A). Interestingly, all mast cell-deficient Cpa3^Cre/+ mice also survived (Fig. 1A). To explore whether a role for mast cell would become evident under more severe conditions we induced CLP using a 22 G needle (Fig. 1B). The survival (Methods) within two weeks after surgical puncture dropped to about half of the animals for both the Cpa3^+/+ and Cpa3^Cre/+ mice (Fig. 1B). We analyzed large cohorts of about n = 50 for each genotype, and found no statistically significant difference between both groups (p = 0.23), indicating that the mortality in cecal ligation and puncture sepsis was independent of mast cells. These results were surprising given that previous publications using Kit-mutant, mast cell-deficient mice concluded that there was a fundamental requirement for mast cells in sepsis survival^1,2. As a control, we therefore included Kit^W/Wv mice in our studies. Even under mild sepsis conditions, which all Kit wild-type (Cpa3+/+ or Cpa3^Cre/+) mice survived, 75 % of Kit^W/Wv mice died (Fig. 1A), and in severe sepsis all Kit^W/Wv mice succumbed the treatment within two days (Fig. 1B). KitW/Wv mice are F1 offspring from WB-Kit^W/+ and B6-Kit^Wv/+ parents, hence they have a genetic F1 hybrid (WBB6F1) background¹⁵ (we refer here to WBB6F1-Kit^W/Wv mice as Kit^W/Wv mice). To exclude genetic background differences as a cause of the higher susceptibility of Kit^W/Wv mice compared to Cpa3^Cre/+ mice, we repeated the CLP experiments using all mice on the WBB6F1 background (WBB6F1-Cpa3^+/+, WBB6F1-Cpa3^Cre/+ and Kit^W/Wv). On this background, mast cell-proficient WBB6F1-Cpa3^+/+ and mast cell-deficient WBB6F1-Cpa3^Cre/+ mice were both even more resistant to CLP than on the B6 background (Fig. 1C), which is consistent with an earlier report¹⁶, and hybrid vigor. In total, 83 % of WBB6F1-Cpa3^+/+ and 74 % of WBB6F1-Cpa3^Cre/+ mice survived (p = 0.24), while again all KitW/Wv mice succumbed to sepsis. These experiments demonstrate that mast cells are not involved in sepsis resistance, while Kit-mutants are highly susceptible.

Figure 1.

Because absence of mast cells in Kit^W/Wv mice has been implicated in impaired cytokine responses in sepsis, notably in TNFα^1,2, we next analyzed the inflammatory cytokine response. Serum levels of TNFα, IL-6, and MCP-1 were measured at three different time points after CLP (Fig. 1D-F). Mast cell-proficient Cpa3^+/+ and mast cell-deficient Cpa3^Cre/+ mice showed very similar time course and amounts of cytokine production. The serum levels for each of the three cytokines was maximal after 8 h, and declined until 24 h without any significant further change at 48 h after cecal ligation and puncture. In contrast, Kit^W/Wv mice started at 8 h with cytokine levels similar to Cpa3^+/+ and Cpa3^Cre/+ mice, but mounted an almost 10-fold higher TNFα, IL-6 and MCP-1 response at 24 h after CLP (Fig. 1D-F). Sham-treated animals (dashed lines) had transiently increased levels of cytokines at 8 h, in response to the anesthesia and laparotomy (Fig. 1D-F). In Fig. 1G-I we plotted the 24 h serum levels of individual mice, comparing Cpa3^+/+, Cpa3Cre/+ and Kit^W/Wv mice. Median cytokine levels were similar in Cpa3+/+ and Cpa3^Cre/+ mice but elevated in Kit^W/Wv mice. Comparable cytokine responses in Cpa3^+/+ and Cpa3Cre/+ mice are consistent with their similar survival rates. Clearly, and in contrast to earlier reports^1,2, the TNFα response was independent of the presence of mast cells, excluding these cells as the main source for TNFα in experimental peritoneal sepsis. Elevate cytokine responses were associated with the Kit mutation. Rather than protection, the strongly elevated inflammatory cytokine levels 24 h after cecal ligation and puncture in most Kit^W/Wv, and in some Cpa3^+/+ and Cpa3^Cre/+ mice may reflect an exaggerated immune response before succumbing to sepsis.

Collectively, these experiments demonstrate that mast cells do not play a role in protecting against enteric bacterial sepsis in the CLP model. We also confirmed the severely impaired survival of Kit^W/Wv mice, and found that this was unrelated to deficiencies in production of TNFα, IL-6, or MCP-1.

Mast cell-deficient Kit^W/Wv mice resist intraperitoneal injection with cecal bacteria

Kit is essential for the development of interstitial cells of Cajal and for intestinal pacemaker activity, and hence autonomic gut motility is largely abrogated in Kit-mutants^17,18. Given this dysfunction of the intestinal physiology in Kit^W/Wv mice, we considered the possibility that the surgical CLP procedure, involving tissue ligation and perforation, could contribute to the sensitivity of Kit^W/Wv mice to sepsis. We therefore induced sepsis without injury of the colon by injection of a cecal bacterial slurry containing a defined amount of enteral bacteria. This is a reproducible, bacteria dose-controlled peritoneal infection model¹⁹. Donors for enteral slurry were C57BL/6 mice. More than 90 % of all three genotypes (Kit^W/Wv, Cpa3^+/+ and Cpa3^Cre/+ mice) survived injection of low dose (3 × 10⁸) bacteria (Fig. 2A). Survival was reduced to less than 50 % during the observation period of 10 days at an approximately 5-fold higher dose (14 × 10⁸ bacteria) (Fig. 2B). For both doses, the outcome for KitW/Wv mice was statistically comparable to that of wild-type control mice (Fig. 2A, B), ruling out immunodeficiency to explain the sensitivity of Kit^W/Wv mice in cecal ligation and puncture. Mast cell-deficient Cpa3^Cre/+ mice were no more susceptible than their mast cell bearing littermates (Fig. 2A, B), confirming the irrelevance of mast cells in this bacterial injection model.

Figure 2.

In order to measure early-stage inflammatory cytokine release, low dose cecal slurry infection was repeated, and mice were analyzed one or two hours later. Compared to NaCl control injected animals, TNFα, IL-6, MCP-1 and IL-10 were elevated in the peritoneal lavage already after one hour, and the cytokine concentrations further increased at two hours after bacteria injection, in particular for IL-6 and MCP-1 (Fig. 2C-F). We could not detect increases in IFN-γ or IL12p70 (data not shown). Comparison of Cpa3^+/+, Cpa3^Cre/+ and Kit^W/Wv mice revealed no significant difference in the cytokine responses, thus excluding mast cells as a major source of inflammatory cytokines, and indicating that this cytokine response is independent of Kit. Since mast cells have been implicated in the recruitment of innate immune cells, we measured neutrophils and macrophages in the peritoneal lavage fluid by flow cytometry one and two hours after bacterial injection (Fig. 2G+H). Neutrophils which are normally absent, or present only at very low numbers in the peritoneal cavity of naïve mice, were rapidly infiltrating the peritoneal cavity of infected animals (Fig. 2G). Of note, we counted very similar numbers of neutrophils (Gr1+CD11b+) in mast cell-proficient Cpa3+/+, and mast cell-deficient Cpa3Cre/+ mice. Neutrophil recruitment was slightly reduced in Kit^W/Wv mice which could be due to their global Kit-dependent neutropenia^20,21. Numbers of macrophages (F4/80+CD11b+) declined²² equivalently within the first hour of infection in Cpa3^+/+, Cpa3^Cre/+ and^KitW/Wv mice (Fig. 2H). Taken together, in contrast to the cecal ligation and puncture sepsis results, Kit^W/Wv mice show normal survival, cytokine response and neutrophil recruitment after bacterial slurry injection. This implies that the susceptibility of Kit^W/Wv mice to cecal ligation and puncture sepsis does not reflect defects in immunity, and that immunity is again unaffected by the absence of mast cells.

Moreover, our time course experiments show that, within the first hours of peritoneal bacterial infection, mast cells are not major producers of inflammatory cytokines.

Kit^W/Wv mice harbor intestinal microflora with increased pathogenic potential

In the CLP model, sepsis is driven by the mouse’s own intestinal flora. The selective susceptibility of Kit^W/Wv mice, but not Cpa3^Cre/+ mice, in this model raises the question whether impaired intestinal physiology, i.e. hypomotility of gut peristalsis and impaired gastrointestinal transit observed in Kit-mutants^18,23, may influence the composition and pathogenicity of the intestinal microflora. To directly compare the pathogenicity, we isolated enteral bacteria from either Kit^+/+.; or Kit^W/Wv mice (donor flora) and injected them into recipient cohorts of Cpa3^+/+ (Fig. 3A), Cpa3^Cre/+; (Fig. 3B), or Kit^W/Wv (Fig. 3C) mice (in this experiment all donor and recipient animals were on the WBB6F1 background). Cecal slurries with five different amounts of bacteria (2, 5, 10, 20, 40 × 108) were applied. This titration experiment revealed a clear correlation between bacterial load and mortality. Of note, we observed a marked leftward shift in the dose-response curves towards lower bacteria concentrations when injecting Kit^W/Wv compared to Kit^+/+ (Fig. 3A-C). Collectively, these data indicate a higher pathogenicity of the microflora harbored in Kit-mutants compared to Kit wild-type mice.

Figure 3.

Figure 4.

To investigate whether the increased pathogenicity of Kit^W/Wv microbiota is due to an altered bacterial composition of the cecal contents, we performed a basic microbiological examination and determined bacterial colony forming units (CFU) of the intestinal contents of Kit^+/+ and Kit^W/Wv mice, including the specimens used for the i.p. infection experiments shown in Fig. 3A-C. To this end, we prepared serial dilutions of the cecal slurries, cultured these on blood agar and McConkey agar plates, and counted bacterial colonies. While Lactobacilli colonies were obtained at similar frequencies from isolates of both mouse strains, CFU counts for Escherichia coli (E. coli) were more that 1000 times higher in Kit^W/Wv than in Kit^+/+ isolates (Fig. 3D). Staphylococcus sp. (most likely Staph. Xylosus) was sporadically found at low level in Kit^+/+ slurries (not shown). Hence, Kit^W/Wv microbiota contains high levels of E. coli, which may underlie the observed pathogenicity.

Cohousing transfers sepsis resistance to Kit^W/Wv mice

To independently prove that a pathogenic microbiota in Kit^W/Wv mice is responsible for the increased CLP-susceptibility of this strain independent of mast cells, we performed co-housing experiments. Kit^W/Wv and their wild-type Kit^+/+ littermates are F1 offspring from WB-Kit^W/+; and C57BL/6-Kit^Wv/+ parents. The phenotypically different KitW/Wv (white) and Kit+/+ (black) mice were normally separated at weaning, which may result in genotype-specific drifts of the enteric microbiota. To prevent such drift, we co-housed Kit^W/Wv and Kit^+/+ mice, and repeated cecal ligation and puncture experiments. In contrast to the previous CLP experiments with non-co-housed mice at which severe (22 G needle) conditions were lethal for all Kit^W/Wv mice (Fig. 1B), now with co-housed mice there remained no difference in survival (p = 0.3809) comparing Kit^W/Wv (74 % of n = 38) and Kit^+/+ (65 % of n = 25) mice (Fig. 3D). These experiments suggest that co-housing transfers CLP-resistance from Kit^+/+ to Kit^W/Wv mice.

Discussion

Kit^W/Wv mice have higher mortality rates compared to Kit^+/+ control mice¹ in the clinically relevant cecal ligation and puncture sepsis model²⁴. The absence of mast cells has been held responsible for this presumed deficiency of Kit^W/Wv mice to mount an adequate anti-bacterial response which would ensue normal survival. In this study, we re-addressed the roles of mast cells and Kit in cecal ligation and puncture experiments. Clearly, absence of mast cells in Kit wild-type mice had no impact on mortality, hence, mast cells play no role in protection from cecal ligation and puncture sepsis. The cecal ligation and puncture data shown in this manuscript are based on experiments conducted over eight years by three independent experimentalists in two institutions (University of Ulm, Germany and German Cancer Research Center in Heidelberg, Germany). Altogether, the sample sizes (n’s) are very large and we consider the data robust. Regarding Kit, we confirmed the susceptibility of Kit^W/Wv mice to cecal ligation and puncture sepsis. Remarkably, Kit^W/Wv mice were as protected as their wild-type controls when intestinal bacteria were injected (no puncture). Indeed, based on mortality, cytokine responses and neutrophil influx they were indistinguishable from Kit^+/+ controls. However, KitW/Wv mice differed from wild-type control mice in their microbiota. Based on in vivo lethality and in vitro colony formation, Kit^W/Wv intestines contained more pathogenic bacteria than Kit^+/+ intestines. This suggests that in cecal ligation and puncture, Kit^W/Wv mice release more pathogenic bacteria into their peritoneal cavity and are hence more severely challenged than Kit^+/+ mice. This would explain the stronger cytokine responses and the greater lethality of the Kit-mutant. In brief, the susceptibility of Kit^W/Wv mice to cecal ligation and puncture is unrelated to mast cells or even ‘immunology’. Our observation of incomparable intestinal microbiota comparing a mutant and its wild-type control may raise a note of caution on cecal ligation and puncture experiments comparing other mouse mutants and their wild-type counterparts.

There are conflicting reports regarding the role of mast cells in the cecal ligation and puncture assay, or in related bacterial infection models. Published results range from pro-pathogenic roles of mast cells²⁵ to enhanced vulnerablity of mouse mutants with loss of protease genes (Mcpt4; Mcpt6)^16,26. It seems difficult to reconcile these reports with our experiments which show unimpaired survival of mast cell-deficient (Cpa3^Cre/+) mice. In a peritonitis model induced by injection of a mouse-virulent strain of Klebsiella pneumoniae, Kit^W/Wv mice were reported to be impaired in bacterial clearance, neutrophil recruitment, and increased expression of TNFα in the peritoneal cavity². In our hands, using the ‘natural mixture’ of cecal bacteria for intraperitoneal infections, Kit^W/Wv mice showed comparable survival, TNFα secretion and neutrophil recruitment to the peritoneal cavity, all of which does not agree with the immunodeficiency that Malaviya et al. described². The reasons for these discrepancies are again unclear. Naturally, host-to-host transmission of Klebsiella pneumoniae requires close contact and generally occurs through the fecal-oral route²⁷ rather than experimental intraperitoneal infections. Rather than probing immunity against single bacterial strains, such as Klebsiella pneumoniae or E. coli, we chose a more natural polymicrobial model, mimicking the release of enteral bacteria following gastrointestinal rupture. In such pathology, we found no evidence for a protective role of mast cells. This conclusion aligns with recent proteome data obtained from primary mouse and human mast cells²⁸. By comparison with macrophages, ex vivo isolated mast cells lacked several different classes of pattern recognition receptors which suggests that peritoneal mast cells, at least at steady state, do not possess broad innate pattern recognition capabilities. Specifically, primary human and mouse mast cells do not express detectable amounts of TLRs, and their expression of MYD88 most likely does not represent a configuration for TLR but rather for IL-1R signaling. Mast cells share expression of the RNA sensor RIG-I and its downstream transcription factor IRF3 with most other immune cells, but do not express other RIG-I-like receptors or their downstream signaling components. Mast cells also do not express C-type lectin receptor components, or inflammasome components²⁸. Importantly, TNFα protein, reported to be stored in mast cell secretory granules in vitro²⁹, was also undetectable in primary mast cells²⁸. These data, together with our polymicrobial sepsis experiments, make it likely that mast cell play no role in protection against peritoneal sepsis. Instead, it appears that the principle innate responder cells in peritoneal sepsis are macrophages and neutrophils which are also the main cellular sources of TNFα.

The differences in composition and pathogenicity of Kit-mutant and wild-type enteral microbiota we report here can only directly refer to the mice bred and maintained during the experimental period in our mouse facilities. Breeding Kit+/+ and Kit^W/Wv mice from WB-Kit^W/+ and B6-Kit^Wv/+; parents, and separating the offspring at the time of weaning, led to the observed differences in microbiota within the same mouse facility. Performing sepsis experiments in mice generated in this manner revealed that lack of Kit can strongly enhance the pathogenicity and, in particular, hugely increase numbers of E. coli colony-forming units. Our data therefore raise the possibility that published work comparing Kit^+/+ and Kit^W/Wv mice may also have been based on separately kept mice. We cannot predict the outcome in microbiota composition of Kit^+/+ and Kit^W/Wv mice bred elsewhere, but we suggest that investigators may control for this. Of particular importance in this regard appear to be co-housing experiments between Kit-mutants and wild-type mice.

Methods

Mice

B6-Cpa3^Cre/+ mice were obtained after backcrossing the Cpa3^Cre allele (Cpa3^{tm3(icre)Hrr6} PMID: 22101159) for >20 generations onto C57BL/6J (B6). In all experiments, littermates from the breeding of B6-Cpa3^Cre/+ with wild-type B6 mice were used. Kit-mutant mast-cell-deficient WBB6F1-Kit^W/Wv mice and their WBB6F1-Kit^+/+ littermate controls were obtained by crossing WB-Kit^W/+ with B6-Kit^Wv/+ mice (both originally obtained from Japan-SLC, Shizuoka, Japan). Unless otherwise described in the text, WBB6F1-Kit^W/Wv mice and their WBB6F1-Kit^+/+ littermates were separated by coat color at weaning. For direct comparison, WBB6F1-Cpa3^Cre/+ and WBB6F1-Cpa3^+/+ littermates were generated in F1 intercrosses of WB-Kit^+/+ with B6-Cpa3^Cre/+ mice.

All animals were generated at the mouse breeding facilities of the University Clinics in Ulm or the center for preclinical research at the DKFZ in Heidelberg. All experiments were conducted in accordance to animal care guidelines pertaining to local animal committees (Regierungspräsidium Tübingen or Regierungspräsidium Karlsruhe) and to the institutional guidelines.

Sepsis models

Cecal ligation and puncture (CLP)

The surgical sepsis model was performed according to Rittirsch et al.³⁰ and Cuenca et al.³¹. In brief, mice were anesthetized (100 mg/kg ketamine and 16 mg/kg xylazine in saline, i.p.), belly shaved and placed onto a sterile-covered warming plate for the duration of the surgery. After disinfection with 70 % ethanol, the abdomen of the mice was opened with scissors by a 1-cm mid line cut into the skin, followed by a 1-cm incision of the peritoneum. The cecum was exteriorized and its stool content was gently palpated towards the distal end. The ligation with 4-0 sutures was placed at half the distance (‘50 % ligation’) between the distal pole and the base of the cecum. A single puncture was applied to the pole of the cecum (gauge sizes of the used needles are indicated for each experiment). Gentle pressure on the ligated cecum released a tiny drop of the cecal content while the needle was withdrawn to prevent immediate closure of the puncture hole. Afterwards the cecum was reposed and wound closure was performed with surgical clips. Sham control mice underwent the same surgical procedure, however without ligation and puncture of the cecum. Finally, mice received 1 ml saline s.c. for fluid replacement. For the entire observation period after surgery, all mice were closely monitored, and animals showing signs of severe distress or suffering were euthanized. Hence, statements in this manuscript referring to mortality all fell under this animal welfare protocol. For profiling of inflammatory cytokines, mice were sacrificed at the indicated time after surgery and blood for serum preparation was collected by heart puncture.

Cecal slurry injection

For the non-surgical sepsis model, suspensions of cecal bacteria were prepared from donor mice and injected intraperitoneally into the test mice. For preparation of the cecal slurry, four donor mice (B6-Cpa3^+/+; (Fig. 2), WBB6F1-Kit^+/+ or WBB6F1-Kit^W/Wv (Fig. 3)) were sacrificed and the content of their cecum was pooled and resuspended in 10 ml saline (0.9 M NaCl). The suspension was centrifuged for 1 min at 20 g (200 rpm) in a swing-out bucket and the upper 8 ml were subsequently passed through 100 μm and 40 μm filters. Bacteria counts (rod-shaped) were determined using a Neubauer counting chamber with 0.02 mm chamber depth. Injected bacterial doses are indicated in Figures 2 and 3. Injected mice were closely monitored and moribund mice were sacrificed (see ‘Cecal ligation and puncture’ paragraph above). For the analysis of cytokine concentrations and myeloid cell infiltrations in the peritoneal cavity, mice were sacrificed at the indicated time after cecal slurry injection and the peritoneal fluid was collected. Therefore, 700-800 μl sterile FACS buffer (PBS with 5 % FCS) were injected intraperitoneally, and after a short abdominal massage, the peritoneal cavity was opened by a 5 mm incision to retrieve about 500 μl of the peritoneal exudate suspension with a pipette. A small aliquot was used to determine the concentration of cells and the total amount of cells was calculated from the total injected volume. The peritoneal exudate suspension was then separated into a cellular and a soluble fraction by centrifugation for 5 min with 500 g (2200 rpm). The cell pellet was resuspended in FACS buffer for subsequent flow cytometric analysis. The soluble fraction was centrifuged for another 5 min at 16200 g (13000 rpm) and the supernatant was snap frozen in liquid nitrogen for later cytokine measurement.

CFU determination

CFU of the cecal slurry suspensions used in the sepsis experiment was determined by plating serial dilutions (10-3 – 10-10 in 0.9 M NaCl) of the slurry stock suspension in replicates on blood agar and McConkey agar. Lactobacillus and Staphylococcus colonies were counted on blood agar plates, Escherichia coli colonies on McConkey agar plates.

Flow cytometric assays

Cytometric bead array (CBA)

Cytokine concentrations of serum samples or peritoneal lavages were determined in a bead-based multiplex cytometric assay (Cytometric Bead Array (CBA) Mouse Inflammation Kit, BD Biosciences). Samples (undiluted and 1:20 dilutions) and cytokine standards (10 ng – 5 pg serial dilutions) were incubated with mixtures of fluorescent beads coated with capture antibodies against IL-6, IL-10, INF-γ, MCP-1, TNF-α and IL12p70. PE-labeled detection reagent (i.e. PE-conjugated antibodies specific for the respective mouse cytokines) was added, and, after incubation and wash, samples were measured at a BD FACSCanto™ or BD LSRFortessa™ instrument. Cytokine concentrations were calculated from mean fluorescence values using the BD FCAP Array Software.

Peritoneal exudate cells (PEC)

The cellular fraction of the peritoneal exudate suspensions was analyzed for CD117+ mast cells, Gr1⁺ CD11b⁺ granulocytes and Gr1–CD11b⁺ F4/80⁺ macrophages on a BD FACS Canto. Therefore, cells were first incubated with mouse IgG (300 μg/ml, Dianova) for blocking of Fc-receptors and then stained with titrated amounts of fluorescent-labeled antibodies in PBS with 5 % FCS. The following antibodies were used: CD117-APC (2B8), CD11b-FITC (M1/70), Gr1-PE (RB6-8C5) all from Pharmingen, and F4/80-APC-A780 (BM8) from ebioscience.

Acknowledgements

We thank Andrea Erles-Kemna, Katja Schmidt and Werner Nicklas, DKFZ, for expert microbiological analyses, and the staff of the animal facilities at the University Clinics Ulm and at DKFZ for expert mouse husbandry. We thank Bernd Echtenacher for the practical introduction to the CLP procedure, and Sven Schäfer for help with CLP experiments. We are grateful to Axel Roers and Thomas Plum for critical reading of the manuscript. This work was supported by the Deutsche Forschungsgemeinschaft (DFG) CRC156/TRR156 project A7 to T.B.F. and H.-R.R., and ERC Advanced grant 233074, HGF Project Immunology & Inflammation (ZT-0027), and the Leibniz program of the DFG (all to H.-R.R.)