IL-12 is essential for protective T cell-mediated immunity against Salmonella infection. To characterize the role of the related cytokine IL-23, wild-type (WT) C57BL/6 and p19−/− mice were infected systemically with an attenuated strain of Salmonella enterica serovar Enteritidis (S. Enteritidis). IL-23-deficient mice controlled infection with S. Enteritidis similarly as WT mice. Similar IFN-γ production as compared with WT mice, but defective IL-17A and IL-22 production was found in the absence of IL-23. Nevertheless, although IL-23 is required for T cell-dependent cytokine responses, IL-23 is dispensable for protection against S. Enteritidis when IL-12 is present. To analyze the role of IL-23 in the absence of IL-12, low doses of S. Enteritidis were administered to p35−/− mice (lacking IL-12), p35/19−/− mice (lacking IL-12 and IL-23), p35/40−/− mice (lacking IL-12, IL-23, and homodimeric IL-12p40), or p35/IL-17A−/− mice (lacking IL-12 and IL-17A). We found survival of p35−/− and p35/IL-17A−/− mice, whereas p35/19−/− and p35/40−/− mice died within 3–6 wk and developed liver necrosis. This indicates that IL-23, but not homodimeric IL-12p40, is required for protection, which, surprisingly, is independent of IL-17A. Moreover, protection was associated with IL-22, but not IL-17F or IL-21 expression or with neutrophil recruitment. Finally, anti-IL-22 treatment of S. Enteritidis-infected p35−/− mice resulted in liver necrosis, indicating a central role of IL-22 in hepatocyte protection during salmonellosis. In conclusion, IL-23-dependent IL-22, but not IL-17 production is associated with protection against systemic infection with S. Enteritidis in the absence of IL-12.
It is very well documented that IL-12-dependent Th1 immunity is essential for control of Salmonella infection in mice and humans (1, 2). It remains to be studied whether Salmonella infection is able to activate IL-23-dependent Th17 responses associated with IL-17 and IL-22 production and whether this contributes to protective immunity. Th17 cells play a dominant role in host defense against bacterial infection at the mucosal body barriers (reviewed in Ref. 3). However, the significance of these cells during systemic infection is currently unknown. In particular, it is of interest whether IL-17A and IL-22 produced by Th17 cells have distinct or common functions.
Bacteriae of the genus Salmonella are Gram-negative Enterobacteriaceae that can infect humans and animals. The symptoms induced by Salmonella can range from acute self-limiting gastroenteritis (termed nontyphoidal salmonellosis) to systemic enteric fever (termed typhoid-like disease) (4, 5). Experimental systemic infection of mice with different serovars of Salmonella enterica mimics human typhoid-like disease. We established a model of systemic infection in mice with an attenuated strain of Salmonella enterica serovar Enteritidis (S. Enteritidis)3 (6). The pathogenicity of this auxotrophic strain of S. Enteritidis for mice depends on the size of inoculum used. It is sublethal for wild-type (WT) mice when given at inocula lower than 107 CFU. In addition, this strain of S. Enteritidis can be controlled even by Th1-compromised mice (e.g., IL-12-deficient mice) when given at low doses (up to 2.5 × 103 CFU/mouse). Thus, using an attenuated strain of S. Enteritidis in defined dose ranges allowed us to establish a long-term infection model for each of both immunocompetent and immunodeficient mice to analyze protective adaptive immune mechanisms. After i.p. application of a high sublethal infective dose of this attenuated S. Enteritidis strain (i.e., 2.5 × 106 CFU for WT C57BL/6 mice) and of a low sublethal infective dose (i.e., 2.5 × 103 CFU for IL-12-deficient C57BL/6 mice), either mouse strain survives and does not develop a typhoid-like disease. These two different experimental models are suitable for studies comparing immunoregulatory mechanisms in the presence and particularly in the absence of IL-12, as is the case in human patients with IL-12/IL-23 defects leading to recurrent salmonellosis (7).
IL-12 produced by Salmonella-infected dendritic cells and macrophages is the key cytokine for Th1 cell development. Th1-derived IFN-γ is responsible for activation of macrophages with inducible NO synthase-dependent production of NO (1, 8, 9). In the absence of IL-12, IFN-γ production is reduced and bacterial control is impaired (2). IL-12 is a heterodimeric cytokine composed of the two subunits p40 and p35. It shares the p40 subunit with the recently discovered IL-23 (10). The p40 subunit of IL-12 can also form homodimers (p40)2 that were shown to act as antagonists for IL-12 at the IL-12R (11) or as agonists (12, 13).
IL-23 was found to have functions different from IL-12. It acts on the CD4+ Th17 cell lineage (14, 15, 16) to induce proliferation and maintenance of these cells. Very recently, we were able to demonstrate that Salmonella infection, in addition to Th1 induction, activates Th17 cells and is associated with IL-17A production by classical Th17 cells, γδ T cells, and other CD4− lymphocytes (17). IL-17A produced by Th17 cells is a proinflammatory cytokine that induces the expression of IL-6, CXCL8, G-CSF, and TNF (18, 19, 20), and influences the activation and migration of polymorphonuclear cells (PMN) by induction of CXC chemokines (21, 22). The receptor for IL-17A is ubiquitously expressed in different tissues (23), leading to pleiotropic effects of the cytokine (reviewed in Refs. 24, 25, 26). IL-17-mediated recruitment of PMN is required for protective immunity after infection with several pathogens (27, 28, 29, 30). In addition, the IL-23/IL-17A pathway has been found to mediate pathophysiological processes and the development of organ-dependent autoimmunity, such as experimental autoimmune encephalomyelitis, collagen-induced arthritis, and inflammatory bowel disease (31, 32, 33, 34). IL-17F is another member of the IL-17 family and is most closely related to IL-17A. Its expression was found to be similarly regulated as IL-17A in activated Th17 cells (32, 35, 36, 37). Little is known yet about the biological function of IL-17F, but it was shown to share functions of IL-17A to a weaker extent (e.g., induction of various cytokines and pulmonary neutrophil recruitment) (38, 39).
Only recently, Th17 cells were found to coexpress IL-17A and IL-22 in response to IL-23 (40, 41). IL-22 is a novel IL-10 family member, originally termed IL-10-related T cell-derived inducible factor (42, 43). Neither resting nor activated immune cells express the functional IL-22R, and IL-22 does not have any effects on these cells (44, 45). In contrast, tissue cells of the skin, digestive, and respiratory system tracts are putative targets of this cytokine. IL-22 can induce the production of antimicrobial molecules, such as β-defensins and S100 proteins (45) and acute-phase proteins (e.g., serum amyloid A and LPS-binding protein) (46, 47). Furthermore, IL-22 expression has been linked to proinflammatory processes in psoriasis, rheumatoid arthritis, and inflammatory bowel disease (45, 47, 48, 49, 50, 51, 52). In contrast, IL-22 can protect hepatocytes from apoptosis after Con A-induced hepatitis (53, 54, 55). Therefore, IL-22 represents a novel type of immune mediator that, although produced by T cells, regulates tissue protection and homeostasis, and enhances the innate immunity of tissue cells (56).
The goal of this study was to define the role of IL-23 in immunity to systemic infection with an attenuated strain of S. Enteritidis. It was of interest to: 1) distinguish between the presence and absence of IL-12-dependent Th1 responses, and 2) characterize the role of IL-17A and IL-22 production stimulated by IL-23. Our data provide evidence for an essential role of a protective IL-23/IL-22 axis in a Th1-compromised state that may be relevant for patients with an inherited defect in the IL-12/IL-12R/IFN-γ pathway suffering from recurrent infections (reviewed in Ref. 7).
Materials and Methods
C57BL/6 WT, IL-23p19−/− (p19−/−) (31), IL-12p35−/− (p35−/−), p35/40−/− (57, 58), and p35/19−/− (59) mice were bred and kept at the animal facility of the Max Planck Institute of evolutionary Anthropology. The p35/IL-17A−/− mice were obtained by crossing p35−/− and IL-17A−/− mice (60). Only females at an initial age of 8–12 wk were used for experiments. Mice were housed under specific pathogen-free conditions. All experiments were conducted according to the German animal protection law (approved by the Regierungspräsidium Leipzig) and the safety guidelines for S1 organisms.
Bacteria and infection model
The attenuated strain of S. Enteritidis (ade−/his−; SALMOVAC SE) (6, 61, 62, 63) was provided by J. Selbitz (Impfstoffwerke Dessau-Tornau GmbH, Rosslau, Germany). The strain has been characterized by us before in a murine infection model (6) and in a murine vaccination model (63). Infection of mice with high inocula leads to typhoid-like disease associated with mortality depending on the infective dose. WT and p19−/− mice were i.p. infected with the highest sublethal infection dose for WT mice, which was determined to be 2.5 × 106 CFU. For infection of p35−/−, p35/40−/−, p35/19−/−, and p35/IL-17−/− mice, the highest sublethal inoculum for p35−/− mice (2.5 × 103 CFU) was used.
Bacteria were grown in Luria-Bertani medium for 5 h to the log phase, and aliquots with defined CFU of S. Enteritidis were suspended in FCS/10% DMSO and stored at −70°C. For infection, aliquots were thawed, washed twice, and diluted in PBS. The indicated inocula were administered in a volume of 500 μl. For ex vivo stimulation of splenocytes, heat-killed (hk) bacteria of the same batch were used (at 60°C for 60 min in a water bath).
In vivo neutralization of IL-22
The p35−/− mice were i.p. injected with 50 μg of purified goat anti-mouse-IL-22 IgG 1 day before infection (0 days postinfection (dpi)). Mice of the control group received 50 μg of normal goat IgG instead (both Abs from R&D Systems). At 7 dpi, all mice were sacrificed and the indicated analyses were performed.
Survival and bacterial counts in organs
Infected mice were monitored daily for survival until 80 dpi. For experiments with WT and p19−/− mice, seven mice per group were sacrificed by CO2 asphyxiation 20 and 80 days after infection. For experiments with p35−/−, p35/40−/−, p35/19−/−, and p35/IL-17−/− mice, five to seven mice per group were sacrificed 20 days after infection. For experiments for in vivo neutralization of IL-22 in p35−/− mice, four mice per group were sacrificed 7 days after infection. Blood was collected by cardiac puncture, and spleen and liver were removed under sterile conditions. Organs were weighed, and pieces of them were homogenized and diluted 1/10 in PBS (w/v). Log10 serial dilutions were plated onto xylose-lysine-desoxycholate agar (Sifin). After 24 h of incubation at 37°C, the number of CFU was determined and corrected for the whole organ weight.
Isolation and ex vivo restimulation of splenocytes
Single-cell suspensions of the removed spleens of individual mice were cleared from erythrocytes by treatment with Gey′s solution and washed and resuspended in ISCOVE′s medium (PAA Laboratories) supplemented with 10% heat-inactivated FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin. The cells of each group were pooled by taking equal cell numbers from individual mice, and adjusted to a concentration of 107 cells/ml. A volume of 500 μl of this splenocyte suspension was dispensed into each well of a 24-well plate. After 12 h of incubation at 37°C under a humidified atmosphere containing 5% CO2, the cells were restimulated for 48 h by addition of either 500 μl of Con A (final concentration 5 μg/ml) for polyclonal T cell stimulation, 108 hk CFU S. Enteritidis per ml for Ag-specific stimulation, or medium for negative control. Cell-free supernatants were harvested and stored at −20°C for cytokine and NO measurement.
ELISA for cytokine determination and colorimetric assay for detection of NO
Mouse IFN-γ, IL-17A, and IL-22 were quantified using Duo-Set ELISA (R&D Systems), according to recommended standard protocols. IL-12p40 was measured using the mAb 5C3 (5 μg/ml) as capture Ab and biotinylated polyclonal goat anti-mouse IL-12p40-purified IgG (1:1000; both Ab provided by M. Gately, Hoffmann-LaRoche, Nutley, NJ) as detection Ab, followed by addition of streptavidin-peroxidase (Southern Biotechnology Associates). Tetramethylbenzidine substrate was used for the colorimetric development (Kirkegaard & Perry Laboratories). Mouse rIL-12 was used as standard (provided by M. Gately, Hoffmann-LaRoche, Nutley, NJ). OD measurement of ELISA was performed with an ELISA reader (Spectramax 340 PC; Molecular Devices). NO was measured in cell culture supernatants with Griess reagent, as described (64). The cytokine and NO concentrations were estimated from the standard curves with SoftmaxPro (Molecular Devices).
Murine tissue samples, snap frozen in Invisorb lysing solution (Invitek), were homogenized during thawing by means of Ultraturrax tissue homogenizer (Jahnke and Kunkel) and then treated with 4 mg/ml proteinase K for 1 h (BD Clontech). Isolation of total cellular RNA was done by use of Invisorb RNA kit II (Invitek). mRNA was reverse transcribed and analyzed in triplicate assays by TaqMan PCR using the ABI Prism 7700 Sequence Detection System (Applied Biosystems), as described previously (44, 65). The appropriate assays, including double-fluorescent probes in combination with assay for the murine housekeeping gene hypoxanthine phosphoribosyltransferase (HPRT), were developed by ourselves (IL-22, IFN-γ, IL-10, and IL-22BP) or purchased from Applied Biosystems. Expression levels were calculated relative to the data for HPRT obtained with the every matching assay.
FACS analysis of splenocyte cell subsets
Single-cell suspensions of the spleens were prepared, as described above. A total of 105 cells per staining was pretreated with anti-CD16/CD32 Fc block (clone 2.4G2; BD Biocsiences) and subsequently stained with the indicated Abs and their isotype controls coupled to PE, FITC, or allophycocyanin. Cells were washed twice with FACS buffer (3% FCS, 0.1% NaN3 in PBS) and fixed with PBS/4% formaldehyde (v/v). Fluorescence was detected using flow cytometry (FACSCalibur flow cytometer; BD Biocsiences). Analyses were performed with the software CellQuestPro (BD Immunocytometry Systems).
Parts of livers from sacrificed mice were fixed in 4% buffered formalin, embedded in paraffin, processed routinely for light microscopy, and stained with H&E. Additional naphtol-AS-d-chloracetate esterase (NACE) staining was performed for the detection of PMN. PMN appear red in the stainings. For the analysis displayed in Fig. 7B (livers of anti-IL-22- and goat IgG-treated p35−/− mice), sections were prepared in a serial fashion. For that, four consecutive levels of two distant parts of each liver (eight levels/organ in total) were used. Livers and spleens were analyzed and assorted into groups in a single-blind trial by the pathologist. All studies were done with an Olympus BX51 microscope.
For comparison of two independent groups, the Mann-Whitney rank sum test was used. For statistical analyses of differences between more than two independent groups, a Kruskal-Wallis statistic followed by Dunn’s posttest was performed. Differences were considered to be significant at p < 0.05.
In the presence of IL-12, IL-23 is dispensable for survival and bacterial control following infection of mice with S. Enteritidis
IL-12 has been shown to play a major role for protective T cell-mediated immunity against intracellular pathogens such as S. Enteritidis (6, 8). As shown earlier, IL-12-deficient C57BL/6 p35−/− mice were highly susceptible to S. Enteritidis infection with a dose of ∼104 CFU leading to death within 4–5 wk (data not shown). Several studies demonstrated a major role of IL-23 for protective immunity after mucosal infection (3). To investigate whether IL-23 plays an essential role in immunity against intracellular pathogens such as S. Enteritidis during systemic infection, WT and p19−/− mice on a C57BL/6 background were infected with increasing doses of S. Enteritidis and monitored for at least 80 days. IL-23-deficient mice survived the largest sublethal inocula (up to 2.5 × 106 CFU S. Enteritidis) exactly like WT mice (100% survival; data not shown). This was associated with the same efficiency in bacterial control compared with WT mice at a low (2.5 × 103 CFU; data not shown) and the highest possible inoculum (2.5 × 106 CFU). In fact, there were no statistically significant differences in bacterial burden in spleen and liver at 20 and 80 dpi, and mice of both groups strongly reduced bacteria in these organs until 80 dpi (Fig. 1). Higher inocula of 2.5 × 107 CFU caused 100% mortality within 1 wk even in WT mice (data not shown). These data demonstrate that IL-23 is not required for protective immunity to S. Enteritidis when IL-12 is present.
Splenocytes from IL-23-deficient mice show normal IFN-γ, but abrogated Ag-specific IL-17A and IL-22 production at 20 dpi
A T cell-mediated immune response with the differentiation of IFN-γ-producing Th1 cells is essential for control of salmonellosis (1, 8). IFN-γ secretion causes NO production by activated macrophages, which leads to the killing of the pathogen (9). Because IL-23-deficient mice survived infection with S. Enteritidis exactly like WT mice, we were interested whether the secretion of IFN-γ and NO or other molecules secreted by T cells with a potential role in protective immunity is influenced when IL-23 is missing. Therefore, we restimulated splenocytes from naive and from S. Enteritidis-infected WT and p19−/− mice ex vivo with either hk S. Enteritidis or Con A for 48 h at 20 and 80 dpi, and analyzed the supernatants. We found that splenocytes from S. Enteritidis-infected WT mice responded strongly with IFN-γ production upon Ag-specific and polyclonal restimulation at 20 dpi as compared with cells of naive mice. Interestingly, splenocytes from p19−/− mice showed almost identical IFN-γ responses as WT splenocytes at 20 dpi (Fig. 2,A). Splenocytes of infected p19−/− mice also produced similar amounts of IL-12p40 and antibacterial NO as WT splenocytes (data not shown). These data indicate that IL-23 is dispensable for regulation of IFN-γ during the initial phase of adaptive immunity against infection with S. Enteritidis. However, at a late time point during infection, when most of the bacteria are already cleared from the organs (80 dpi, Fig. 1), IFN-γ production is ∼2-fold lower in splenocytes of p19−/− mice as compared with WT splenocytes. Ag-specific NO is not produced in both groups at 80 dpi (data not shown).
IL-23 has been shown to activate Th17 cells (37, 66) for production of IL-17A (67). Therefore, we also determined the Ag-specific and polyclonal production of IL-17A (Fig. 2,B). Splenocytes of infected WT mice produced significant amounts of Ag-specific and polyclonal IL-17A. We have shown earlier that S. Enteritidis induces a Th17 response in C57BL/6 mice with one-third of IL-17A-producing cells being CD4 positive (17). In contrast to WT splenocytes, splenocytes from infected p19−/− mice had a complete lack of Ag-specific IL-17A production and a partial reduction of polyclonally induced IL-17A production. Recently, Th17 cells were found to coexpress IL-17A and IL-22 in mice (40). Therefore, we investigated IL-22 production by splenocytes of p19−/− mice upon ex vivo stimulation. We found induction of Ag-specific IL-22 production in response to infection in WT as well as in p19−/− mice at 20 dpi. However, there is a ∼2-fold reduction of IL-22 production in splenocytes from p19−/− mice as compared with WT splenocytes. At the same time, IL-22 production of p19−/− splenocytes upon polyclonal T cell stimulation was hardly increased as compared with naive splenocytes. At 80 dpi, the Ag-specific IL-22 production of p19−/− splenocytes was back to baseline level, whereas WT splenocytes still responded to stimulation (Fig. 2 C). Therefore, we find reduced, but not abrogated Ag-specific IL-22 production in the spleen when IL-23 is absent. These data indicate that besides Th1 cells, also Th17 cells develop upon S. Enteritidis infection. The Th1-mediated production of cytokines is not affected by the absence of IL-23 at 20 dpi, but appears to be reduced at 80 dpi. Interestingly, during the infection with S. Enteritidis, IL-17A production was strictly and IL-22 production only partially IL-23 dependent.
We also measured IL-22 in the sera and found significantly reduced levels close to the detection limit (0.012 ng/ml) in the absence of IL-23 at 20 dpi (Fig. 2 D). At 80 dpi, the IL-22 levels of WT mice decreased compared with day 20 and reached the detection limit. Concentrations of IL-17A and IFN-γ were below detection limit in the sera (data not shown). Together, these in vivo data support the evidence of reduced Th17 responses in the absence of IL-23 following infection with S. Enteritidis.
In the absence of IL-12, IL-23, but not (p40)2, provides protection against S. Enteritidis infection that is independent of IL-17A
Absence of IL-12-dependent Th1 responses has been found in patients with an inherited defect in the IL-12/IL-12R/IFN-γ axis and was associated with recurrent nontuberculous mycobacterial or Salmonella infections (7). Thus, it was of interest to us to study the function of IL-23 in the absence of IL-12-dependent Th1 responses. Therefore, in contrast to the studies described above and in Figs. 1 and 2, the following studies were undertaken to analyze the role of IL-23 and/or (p40)2 in the absence of IL-12, i.e., in a Th1-compromised situation. In addition to IL-12, the homodimer (p40)2 can be protective in immunity to mycobacterial infection (12, 13). We reported previously that BALB/c and 129Sv/Ev mice lacking all p40-containing members of the IL-12 cytokine family show a higher susceptibility to infection with S. Enteritidis than mice only deficient in IL-12 (6). In accordance with these previous results, more recent experiments corroborate these findings in C57BL/6 mice. Mice lacking IL-12 (p35−/− mice) completely survived inocula of 2.5 × 102 CFU (data not shown), and 94% of all p35−/− mice (33 of 35 mice) survived infection with 2.5 × 103 CFU until the end of the experiment (80 dpi), whereas all p35/40−/− mice (lacking all p40-containing members of the IL-12 cytokine family) died after infection with either of those low inocula (Fig. 3,A). After infection with a 10-fold higher dose of 2.5 × 104 CFU, p35−/− mice died significantly later than p35/40−/− mice (data not shown). With even higher inocula, the difference in survival of mice lacking only IL-12 or all p40-dependent cytokines disappeared (data not shown). These data indicate that in the absence of IL-12, a p40-dependent cytokine is essential for survival of C57BL/6 mice after low-dose infection with S. Enteritidis. They leave open though whether IL-23 and/or (p40)2 account for the striking phenotypic difference, especially at an infective dose of 2.5 × 103 CFU. To analyze a potential protective role of IL-23 and/or (p40)2 in the absence of IL-12, p35−/− mice, p35/40−/− mice, and p35/19−/− mice (producing only free p40, but neither IL-12 nor IL-23), all on a C57BL/6 background, were infected with 2.5 × 103 CFU S. Enteritidis. This inoculum is sublethal for p35−/− mice and 1000-fold lower than the inocula used for WT and p19−/− mice. Strikingly, both p35/19−/− mice and p35/40−/− mice died within 4–6 wk after infection with S. Enteritidis (Fig. 3 A). This demonstrates that IL-23 is required for protective immunity to S. Enteritidis when IL-12 is absent, as has been shown earlier for mycobacterial and Toxoplasma infection (68, 69). Moreover, these data unambiguously exclude a protective role of monomeric and homodimeric IL-12p40 during Salmonella infection.
Because p19−/− mice showed significantly reduced IL-17A production (see Fig. 2,B), we suspected IL-17A to be responsible for protection of low-dose infected mice in the absence of IL-12. Therefore, we generated p35/IL-17A−/− mice and included them in the experiment. Surprisingly, these mice completely survived the low-dose infection (Fig. 3,A). This indicates that IL-17A-independent mechanisms are responsible for protection. After infection with 2.5 × 103 CFU S. Enteritidis, resistant p35−/− and p35/IL-17A−/− mice had significantly lower bacterial burden in spleen and liver 20 days after infection than the susceptible p35/40−/− and p35/19−/− mice (Fig. 3,B). To characterize the kinetics of bacterial development in these mice, the organ burden of moribund p35/40−/− and p35/19−/− mice was analyzed at a time point near death (∼35 dpi) and compared with the organ burden of long-term surviving p35−/− and p35/IL-17A−/− mice at the end of the experiment (80 dpi). At this time point, the p35−/− and p35/IL-17A−/− mice had low bacterial burden (101–103 CFU/organ; Fig. 3,C) and did not show any clinical phenotype indicating similar capability of clearing the infection. Whereas the p35−/− mice and the p35/IL-17A−/− mice were able to reduce the load of S. Enteritidis at 80 dpi, bacterial load in spleen and liver of p35/40−/− and p35/19−/− mice increased dramatically (108–109 CFU/organ) already at 35 dpi (Fig. 3 C). Therefore, bacterial control in the absence of IL-12 is IL-23 dependent, but not mediated by IL-17A.
IL-22, but neither IL-17A nor IFN-γ and NO production are associated with survival in the absence of IL-12
To find other potential mediators that can explain the surprising resistance of the p35/IL-17A−/− (and p35−/−) mice, we measured IL-22 in the serum and both IL-22 and IL-17A in the supernatants of restimulated splenocytes of infected mice. We found significantly reduced serum levels of IL-22 in the absence of IL-23 (p35/19−/− mice) as compared with p35−/− and p35/IL-17A−/− mice (Fig. 4,A). Serum IL-22 tended to be decreased in p35/IL-17A−/− mice as compared with p35−/− mice (without reaching statistical significance), suggesting a role of IL-17A in IL-22 production (Fig. 4,A). As expected, IL-22 production was strongly reduced in splenocytes of p35/19−/− mice. IL-17A production by splenocytes was found to be reduced, but not completely lost in the absence of IL-23 (Fig. 4 B). Therefore, in the absence of IL-12, IL-23 is absolutely required for Ag-specific production of IL-22, and to a lesser extent for IL-17A production in the spleen after S. Enteritidis infection. In addition, these findings support the assumption that IL-22-, but not IL-17A-mediated mechanisms are responsible for the protection after infection with Salmonella when IL-12 is absent, and that these mechanisms completely rely on IL-23.
The 100% survival rate and the low numbers of bacteria in spleen and liver of the p35/IL-17A−/− mice were striking (Fig. 3). To see whether this phenomenon was due to elevated IFN-γ production, we also measured IFN-γ in the spleen cell supernatants. Regarding the Th1-compromised immune status of mice lacking IL-12, we expected generally compromised IFN-γ levels as compared with WT mice. In line with this, Ag-specific IFN-γ production appeared to be ∼12- and 30-fold reduced in p35−/− and p35/19−/− mice, respectively (Fig. 4,C), as compared with WT mice (Fig. 2,A), and it was close to the detection level in p35/IL-17A−/− mice (Fig. 4,C). We found polyclonally induced IFN-γ production in splenocytes of p35−/− and p35/19−/− mice. Surprisingly, in the p35/IL-17A−/− mice, IFN-γ production was strongly reduced (Fig. 4,C). Data from p35/40−/− mice were similar to those of p35/19−/− mice (data not shown). We also measured the secretion of the effector molecule NO and found a strong reduction of Ag-specific and polyclonal production only in p35/IL-17A−/− mice similar to the low IFN-γ levels observed in these mice (Fig. 4 C). These data suggest that IL-17A contributes to IFN-γ and NO production of splenocytes. However, clearly IL-23-dependent protection must be mediated by other mechanisms than the well-known intracellular bacterial killing mechanism mediated by IFN-γ/NO.
Expression of IL-22 and IL-17A, but not IL-17F, strongly relies on the presence of IL-23
To determine whether other IL-23-dependent mechanisms are involved in protective immunity to S. Enteritidis, we measured the mRNA expression of other Th17-associated cytokines in the spleens of resistant p35−/− vs susceptible p35/19−/− mice at 20 dpi (Fig. 4,D). In particular, we wanted to clarify a potential role of IL-17F or IL-21, because these are yet additional cytokines produced by activated Th17 cells (70). Consistent with IL-22 and IL-17A protein production (see Fig. 4,B), we found significantly reduced mRNA expression of IL-22 (p = 0.002) and IL-17A (p = 0.03) when IL-23 was absent. In contrast, IL-17F as well as IL-21 expression was independent of IL-23 and did not differ between protected p35−/− and susceptible p35/19−/− mice. Therefore, IL-17F and IL-21 do not appear to be potential mediators of IL-23-dependent protection. In accordance with IFN-γ protein production (Fig. 4 C), there was also no difference in the expression of IFN-γ mRNA between infected p35−/− and p35/19−/− mice at 20 dpi. Furthermore, no difference was found for mRNA production of IL-27, IL-1ß, IL-6, TNF-α, or IL-10. These data strengthen the predominant association of IL-23-dependent IL-22 with protection.
Reduced CD11b+Gr1+ cells in the absence of IL-17A, but not in the absence of IL-23
Neutrophilic granulocytes have been shown to have a protective function during salmonellosis (71). IL-17A can contribute to recruitment of PMN to infected tissues (72). We analyzed the number of PMN in the spleens of p35−/−, p35/IL-17A−/−, and p35/19−/− mice by FACS and found significantly reduced numbers and percentages in the spleens of p35/IL-17A−/− mice. In contrast, the percentages (Fig. 5,A) and numbers (Fig. 5,B) of PMN in p35/19−/− mice were comparable to those of p35−/− mice. Therefore, this indicates that IL-23-dependent protection is independent of neutrophils. Moreover, PMN recruitment in systemic salmonellosis is not due to regulation of granulopoiesis by the IL-23/IL-17 axis, as it has been shown for the lungs of mice infected with Mycoplasma pneumoniae (29). The analysis of PMN recruitment to the spleen was confirmed by neutrophil-specific staining (NACE) of liver sections derived from resistant p35−/− and p35/IL-17A−/− vs susceptible p35/19−/− and p35/40−/− mice (Fig. 5,C). In livers of infected p35−/−, p35/19−/−, and p35/40−/− mice were similar frequencies of PMN, but fewer PMN were detectable in livers of infected p35/IL-17A−/− mice, indicating again IL-17A-, but not IL-23-dependent neutrophil recruitment in the absence of IL-12. These data also point to a role of IL-23-independent IL-17A in PMN recruitment (Figs. 4,B and 5).
In the absence of IL-23/IL-22, liver necrosis, fibrin thrombi, and aberrant granulomas develop
Histopathological analysis of the liver of infected resistant p35−/− and p35/IL-17A−/− mice and susceptible p35/19−/− and p35/40−/− mice was undertaken to characterize the inflammatory response after infection with S. Enteritidis. Resistant p35−/− and p35/IL-17A−/− mice showed granuloma formation, whereas susceptible p35/19−/− and p35/40−/− mice developed a more diffuse inflammation (Fig. 6,A). Susceptible p35/19−/− and p35/40−/− mice, but not resistant p35−/− and p35/IL-17A−/− mice, also showed the occurrence of fibrin thrombi (Fig. 6,A, bottom row) and necrotic tissue areas (Fig. 6,A, middle row) at 20 dpi. Interestingly, necrotic tissue damage of p35/19−/− and p35/40−/− mice was restricted to liver and did not occur in spleen, which is another site of Salmonella infection (data not shown). This is consistent with the expression of IL-22R1 in liver, but not spleen (45). Therefore, the absence of IL-23-dependent IL-22 is associated with the development of liver necrosis. Development of liver necrosis became very striking at a time point near death (35 dpi). Broad areas with confluent single cell necrosis of the liver cells had developed in p35/19−/− and p35/40−/− mice (Fig. 6,B). In contrast, we found only few well-defined granulomas and no necrotic tissue in the livers of resistant p35−/− and p35/IL-17A−/− mice at the end of the experiment at 80 dpi (Fig. 6 B), indicating that IL-17A is not involved in protection from liver damage.
To examine the role of IL-22 in the development of liver necrosis during Salmonella infection more directly, we treated p35−/− mice with neutralizing anti-IL-22 or normal goat IgG 1 day before S. Enteritidis infection. At 7 dpi, mice were sacrificed, bacterial organ burden was determined, and serial sections of the livers were prepared and analyzed in a single-blind fashion. Bacterial organ burden at 7 dpi was quite low (Fig. 7,A) and did not yet significantly differ between mice treated with anti-IL-22 or control IgG. However, already at this early time point, all mice were correctly grouped by histopathological examination in a single-blind approach. Anti-IL-22-treated mice developed single cell necrosis in the liver. In contrast, smaller and fewer inflammatory foci without necrosis were found in the livers of mice treated with normal goat IgG (Fig. 7 B). These data point to a role of IL-22 in the protection of hepatocytes in the Salmonella-infected liver.
In the present study, we show that the absence of IL-23 does not compromise protective immunity to systemic infection with even a high sublethal dose of S. Enteritidis inducing a typhoid-like disease. IL-23-deficient mice mounted normal IFN-γ responses, and, therefore, it can be concluded that the IL-12/IFN-γ pathway is still intact in the absence of IL-23. IL-23-dependent Th17-associated responses (e.g., production of IL-17A and IL-22) are dispensable in the presence of an effective Th1 response. In contrast, in the absence of IL-12-dependent Th1 responses, protection against low, yet sublethal doses of S. Enteritidis is associated with IL-23-dependent IL-22, but not IL-17A production, enabling protection from liver damage. Neither residual production of IL-12-independent IFN-γ/ΝΟ (see Fig. 4,C) nor IL-17A-dependent neutrophil recruitment (Fig. 5) is associated with protection.
In other models of intracellular infection (e.g., Toxoplasma gondii, Mycobacterium tuberculosis, Mycobacterium bovis bacillus Calmette-Guerin), IL-23-dependent protection was also found in the absence of IL-12 (59, 68, 69). Interestingly, IL-23-dependent generation of IL-17 responses was found, but did not contribute phenotypically to the course of the infection. IL-17A was experimentally excluded to have a functional role in immunity to infection with M. bovis bacillus Calmette-Guerin by neutralizing IL-17A in WT, p35−/− mice, and p40−/− mice (59). Previously, in the murine S. Enteritidis and M. tuberculosis infection models, IL-12p40/IL-23-dependent IFN-γ was shown in the absence of IL-12 (6, 68). With our present study, IL-23-dependent IL-22 comes into the focus for systemic Salmonella infection both for liver cell survival and for pathogen defense. However, in murine tuberculosis and toxoplasmosis, IL-22 remains to be studied. Interestingly, very recently in human tuberculosis patients, IL-17- and IL-22-producing CD4+ cells and bronchoalveolar lavage containing IL-22, but not IL-17, were found (73). In contrast for extracellularly multiplying pathogens (e.g., Klebsiella, Citrobacter), induction of Th17 responses was shown (74, 75). Extracellular multiplication of a pathogen may be a prerequisite for being efficiently rejected by the elements of Th17 responses. At the moment it seems, however, too early to draw conclusions as to a link between extracellularly and intracellularly replicating pathogens and the role of Th17 responses for protection against extracellular stages.
Recently, it was shown that Th17 cells producing IL-17A are required for protective immunity after mucosal infection of macaques and streptomycin-pretreated mice with S. Typhimurium (76). Evidence was obtained that in the absence of IL-17RA expression, mucosal barrier function of the gut is compromised and bacterial dissemination increased early on after oral infection. In contrast, IL-17A in systemic Salmonella infection is dispensable in the absence of IL-12. However, IL-22 becomes important in this situation because it is required for protection against hepatocyte necrosis during salmonellosis.
Recently, it was demonstrated that IL-22 did not influence the function of immune cells, but can induce antimicrobial molecules such as β-defensins and S100 proteins in epithelial cells of body barriers and enhance the innate immunity of tissues cells (45). An essential role of IL-23-dependent IL-22 during innate immunity has been found for protection against infection of mice with Citrobacter rodentium, a model of human infection with attaching and effacing (A/C) pathogens (74). In this study, IL-22-induced production of Reg family members acting as antimicrobial proteins appeared to contribute to protection. Thus, various mechanisms (e.g., stability of mucosal barrier function, induction of antimicrobial proteins) induced by IL-22 may be involved in innate protection against intestinal Gram-negative bacteria. This notion is further strengthened and extended by data from a pulmonary Gram-negative infection with Klebsiella pneumoniae (77). In this study, it was shown that although both IL-22 and IL-17A are involved in mucosal host defense, the role of IL-22 is more important than IL-17A. In particular, IL-22 enhanced the clonogenicity of human bronchial epithelial cells and recovery of resistance following experimental creation of a wound. In addition, IL-22 induced the expression of chemokines and other host defense genes such as lipocalin-2. Together, this points to a major function of IL-22 in innate immunity by mucosal epithelial cells (78). It remains to be shown which of these mechanisms are initiated by IL-22 in our study. Preliminary data suggest that production of acute-phase proteins, such as LPS-binding protein and serum amyloid A, is not the essential protective mechanism initiated by IL-22 after i.p. infection of mice with S. Enteritidis (data not shown).
Salmonella is able to invade both the liver and spleen following systemic or mucosal infection (as reviewed in Ref. 8). Already 7 days after i.p. infection, we found necrosis of liver cells after neutralizing IL-22. This is a striking result in light of the low bacterial loads at this early time point. Necrosis of the liver was accompanied at 20 dpi by the occurrence of fibrin thrombi in the absence of IL-22 (p35/19−/− and p35/40−/− mice; Fig. 6). All of this was not the case in spleen, indicating an organ-specific effect of IL-22. Liver necrosis may be causally involved in mortality of p35/19−/− and p35/40−/− mice unable to produce IL-22 after low-dose infection with S. Enteritidis. Survival of Salmonella-infected p35/IL-17A−/− mice indicates that IL-17A is not involved in protection of hepatocytes. This is consistent with a recent study of Con A-induced hepatitis in which also IL-22, but not IL-17, was shown to protect hepatocytes from the effects of the immune-mediated inflammation (55). Interestingly, C. rodentium infection was also found to be associated with hepatic embolic microabscesses in the absence of IL-22 (75). IL-22 was shown to protect hepatocytes from apoptosis involving STAT3 activation (54). The IL-22R can activate anti-inflammatory STAT3 activation indistinguishable from IL-10 (79).
Besides IL-23-dependent IL-17A and IL-22, IL-17F and IL-21 could have also been likely candidates to mediate the observed protection of p35/IL-17A−/− mice because they can also be expressed by activated Th17 cells (32, 35, 70). No differences of mRNA expression for IL-17F and IL-21 could, however, be found between resistant (p35−/−) and susceptible (p35/19−/−) mice, pointing once more to a unique role of IL-22 for protection.
We find low levels of IL-23-independent IL-17A in the absence of IL-12, but not in the presence of IL-12 (see Fig. 4 vs Fig. 2,B). In contrast, IL-22 production was only partially IL-23 dependent in the presence of IL-12 and completely IL-23 dependent in the absence of IL-12 (see Fig. 2,C vs Fig. 4,B). It is known that IL-17A can be produced independently of IL-23 (80). Therefore, it was to be expected that neutrophil recruitment can occur independent of IL-23 (see Fig. 5). Presently, it is unclear what the exact molecular basis is for the regulatory effect of IL-12 in IL-23-dependent and IL-23-independent IL-17 and IL-22 production. From our data it appears that IL-12 directly or indirectly suppresses IL-23-independent IL-17A production. The opposite holds true for IL-22 production. To fully answer this question, further studies have to be done.
Our findings from this experimental murine systemic Salmonella infection model may be relevant to the immune mechanisms that are active during human typhoid disease. We propose that the IL-12/IFN-γ axis is highly effective in control of high Salmonella inocula provoking otherwise liver damage. With a functional IL-12/IFN-γ pathway, the IL-23/IL-17/IL-22 axis is even dispensable. If, however, the IL-12/IFN-γ pathway is defective (as is the case for p35−/− mice), only low infective doses are tolerated, and this requires the IL-23/IL-22 cascade. In contrast to IL-22 production, IL-17A is functionally dispensable for protective immunity to S. Enteritidis infection in a compromised Th1 status, although IL-17A is also generated by the IL-23/IL-22 cascade and is associated with more effective PMN recruitment.
It is intriguing to note that in a Th1-compromised situation as seen in resistant p35/IL-17A−/− mice, the role of the IFN-γ/NO pathway appears to become dispensable (see Fig. 4 C). The underlying mechanism mediating the reduction of IFN-γ/NO levels during IL-17A deficiency is presently unclear. However, this phenomenon is interesting in light of the observation that Salmonella infections are more frequent in human patients with an IL-12/IL-23 defect than with an IFN-γ defect (81). IFN-γ-independent activities of IL-23 such as IL-22 induction could be protective immune mechanisms operative in these patients, as shown in this study in the murine Salmonella infection model of IL-23 deficiency.
We thank Petra Meier for preparing the histological sections; Sabine Siegemund and Juliane Richter for help with the dissection of animals and subsequent analysis; Eva Marquardt, Norman Kirchoff, and Rowina Voigtländer for help with the animal husbandry; and Tanja Sonntag for genotyping.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Research Grant Al 371/3-3 from the Deutsche Forschungsgemeinschaft (to G.A.) and Bundesministerium fur Bildung, Wissenschaft, Forschung und Technologie Grant EXC306 “Inflammation at Interfaces” (to C.H.).
Abbreviations used in this paper: S. Enteritidis, Salmonella enterica serovar Enteritidis; dpi, days postinfection; hk, heat killed; HPRT, hypoxanthine phosphoribosyltransferase; NACE, naphthol AS-d-chloracetate; PMN, polymorphonuclear cell; WT, wild type.