Crucial Role of Interferon Consensus Sequence Binding Protein, but neither of Interferon Regulatory Factor 1 nor of Nitric Oxide Synthesis for Protection Against Murine Listeriosis

Mice deficient for ICSBP (background C57BL6× 129Sv), IRF1(129Sv), IRF2 (C57BL/6), IFN I and II receptor (both 129Sv) were generated by gene targeting in embryonic stem cells as described (12, 26, 36, 39). IRF1-deficient mice were kindly provided by Prof. Charles Weissmann (Institute for Molecular Biology I, University of Zürich, Switzerland). IFN type I receptor^−/− (A129) and IFN type II receptor^−/− (G129) mice were obtained from the breeding colony of Prof. M. Aguet (Institute for Molecular Biology I, University of Zürich, Switzerland). Control C57BL/6 or 129Sv mice as well as RAG2^−/− mice were obtained from the Institute for Laboratory Animals (Veterinary Hospital, Zürich, Switzerland). Mice were used at 6–10 wk of age. The different breedings (except A129 and G129) and all the experiments were performed under conventional (non-SPF) conditions.

Listeria monocytogenes was originally obtained from B. Blanden (Canberra, Australia). It was cultured in trypticase soy broth (BBL Microbiology Systems, Cockeysville, MD), and overnight cultures were titrated on tryptose blood agar plates (Difco Laboratories, Detroit, MI). For injection, the original culture was diluted in BSS to inject the indicated dose in 200 μl for i.v. or 30 μl for injection into the footpad (i.f.).

On the indicated days after infection the whole spleen and one lobe of the liver were taken out and homogenized. Bacterial titers were determined by plating out four serial 10-fold dilutions of organ suspensions on tryptose blood agar plates.

On day 0, spleen single cell suspensions were let to adhere to plastic to deplete them from macrophages. After 2 h 3 × 10⁷ splenocytes were transferred into nonirradiated RAG2^−/− recipients. On day 1 the recipients were infected with 2 × 10⁵ CFU of Listeria, and on day 10 liver and spleen were taken out to determine bacterial titers.

Peritoneal macrophages (PEM) of different strains were elicited by injection of 2 ml of a starch solution (2%; Merck, Darmstadt, Germany) intraperitoneally on day −5 and harvested on day 0 by rinsing the peritoneal cavity with 10 ml of cold BSS. The macrophages were washed three times with BSS supplemented with albumin to prevent clumping and then plated on cover slips in 24-well plates. Cells were cultured in IMDM (Gibco, Basel, Switzerland) supplemented with 10% FCS, glutamine, and 50 μg/ml gentamicin, an only extracellularly effective antibiotic. After 2 h of adherence the cover slips were washed twice and put in 1 ml IMDM. The cultures were stimulated with 200 ng/ml LPS, with 200 U/ml recombinant murine IFN-γ (Genzyme, Cambridge, MA) or a combination of both for 42 h and then used for determination of nitric oxide (NO) production, of respiratory burst or of Listeria killing in vitro. In those cultures used for killing assays, the medium was changed to antibiotic-free after 24 h.

NO production was measured by determination of nitrite accumulation in PEM cultures with Griess reagent (0.05% N-1-naphthyl-ethylene-diamine-dihydrochloride/0.5% sulfanilamide/2.5% phosphoric acid; all from Fluka, Buchs, Switzerland) as described (41). In brief, 50 μl cell culture supernatant was added to 150 μl Griess reagent in 96well plates and incubated at room temperature for 10 min. Absorption was read with an ELISA reader at 570 and 630 nm.

Respiratory burst was measured as H₂O₂ production by cultured PEM upon PMA (Sigma, Buchs, Switzerland) stimulation as described (18). In brief, H₂O₂ secretion of macrophages was quantified by chemiluminescence under presence of horseradish peroxidase type I (Sigma) and 5-amino-2,3-dihydro-1,4-phthalazinedione (luminol; Sigma) after triggering with 50 ng/ml PMA. Light emission was discontinuously measured over 15 min in a LKB 1251 luminometer (LKB, Bromma, Sweden). Values in mV were converted into pmol H₂O₂ after calibration by the scopoletin method (42). The cells on the cover slips were counted, and values for NO and H₂O₂ calculated as nmol/10⁵ cells. In Fig. 4 B stimulation index of stimulated versus unstimulated cultures is shown, because absolute values of respiratory burst varied between the experiments. PMA was used to trigger respiratory burst because, as a chemically defined substance, it is the most reliable burst trigger. Also opsonized Listeria, BCG or zymosan could be used with similar capacities to trigger burst (23, 43), but more variability. Because we investigated mouse strains deficient for various IFNdependent transcription factors, the induction phase of NADPH oxidase during 2 d under IFN-γ stimulation is important in our experiment, whereas the effector phase of the burst trigger is only used as read-out to measure the enzyme activity by providing the best stimulator (PMA) and eccess of substrate (luminol).

PEM cultures in antibiotic-free medium as described above were infected with 10⁷ CFU of Listeria from an overnight culture, washed three times and opsonized with normal human serum. After 15 min of phagocytosis the infected cultures were washed thoroughly, and gentamicin-containing medium and the respective stimulators were added. To determine the infection rate at time point t₀, three cover slips were taken out. The remaining ones were further cultured for 7 h to allow digestion of Listeria by macrophages. After 7 h the cover slips were taken out, dried, and then stained according to May-Grünwald-Giemsa. For each mouse strain and each stimulation a total of 600 macrophages were counted under the microscope to determine the number of Listeria-infected cells. The change of infected macrophages was calculated in percentage of the infection rate at t₀. For details, see reference 18.

Mice infected with 5 × 10³ CFU of Listeria i.v. were sacrificed on day 5 or 6. Organs were immersed in Hank's BSS and frozen in liquid nitrogen. 5-μm cryosections were fixed with acetone for 10 min, immunostained for Listeria with a polyclonal rabbit anti-Listeria serum (diluted 1/2,000; kindly provided by Professor J. Bille, Institute of Microbiology, University Hospital of Lausanne, Switzerland) and for iNOS with a polyclonal rabbit anti-iNOS serum (diluted 1/1,500; Biomol, Plymouth, PA). Bound primary antibodies were detected using a sandwich staining procedure. Sections were incubated with alkaline phosphatase-labeled goat anti–rabbit Ig (diluted 1/80; Jackson Laboratories, Bar Harbor, Maine) followed by alkaline phosphataselabeled donkey anti–goat Ig (diluted 1/80; Tago). Dilutions of secondary reagents were made in TBS containing 5% normal mouse serum. All incubation steps were done for 30 min at room temperature. Alkaline phosphatase was visualized using naphthol ASBI phosphate and New Fuchsin (Sigma) as substrate, which yields a red color reaction product. Endogenous alkaline phosphatase was blocked by levamisole. Sections were counterstained with hemalum, and cover slips were mounted with glycerol/gelatin.

Gene targeted mice deficient for ICSBP, IRF1, or IRF2 were infected with various doses of Listeria intravenously or peripherally i.f., and survival was monitored daily (Table 2). All ICSBP^−/− mice died after injection of a dose as low as 50 CFU of Listeria, whereas five of six IRF2^−/− mice succumbed to a dose of 5 × 10³ CFU within 12 d. In contrast, IRF1^−/− and wild-type mice resisted to a dose of 5 × 10³ CFU injected intravenously. However, IRF1^−/− mice on C57BL/6 background and held under strict SPF conditions also showed enhanced susceptibility to Listeria, when injected with a 5–10-times higher dose intraperitoneally (Ferrick, D., and H.W. Mittrücker, personal communication). Listeria titers in liver and spleen were determined 24 h after a high dose (2 × 10⁵ CFU) and 5 d after an intermediate dose (5 × 10³ CFU) of Listeria injected intravenously. In the first 24 h, when neutrophils seem to play an important role (2), there was almost no titer difference between the three strains and only a 10-fold difference compared to control mice (data not shown). However, after 5 d when activated macrophages are essential for control of Listeria infection, ICSBP^−/− and IRF2^−/− showed between 10²- and 10⁶-fold higher titers in liver and spleen, whereas IRF1^−/− mice controlled Listeria replication comparable to controls (Fig. 1 A). In vitro gene regulation studies have revealed that ICSBP and IRF2 form complexes which then have a markedly enhanced DNA binding capacity to ISRE compared to the single factors (44). In contrast to IRF1, they are both negative regulators of classical IFN-induced genes. However, both transcription factors are obviously of major importance for early anti-Listeria immune responses. Since it has been shown that ICSBP^−/− mice do not express IRF2 (although the gene is intact [36]), this can explain the even more drastic phenotype of ICSBP^−/− compared to IRF2^−/− mice, because they represent functionally a double knock-out phenotype.

Competition of different transcription factors of the IRF family at the DNA binding level has been demonstrated in in vitro studies (45). It was therefore possible that lack of IFN type I–induced transcription factors would lead to increased activity of IFN type II–induced factors. To test this in vivo, we infected mice deficient for the type II (G129) or the type I (A129) IFN receptor and control mice (wt129) with 5 × 10³ CFU of Listeria and determined bacterial titers in liver and spleen on day 5 (Fig. 1,B). As demonstrated earlier (12), G129 mice showed drastically enhanced bacterial replication and lethality (Table 2), whereas A129 eliminated the pathogen even more efficiently than wt129 mice. This result suggests that competition between the two signaling pathways at the transcription factor level occurs. IFN type II–induced transcription factors (and among them especially ICSBP) may compensate for the lack of IFN type I–induced factors in the A129 mouse, thereby conferring even higher resistance to Listeria infection than in control mice. Because early Listeria clearance in nude mice (46) has been shown to be more efficient than in immunocompetent controls because their macrophages are preactivated (probably by LPS derived from normal intestinal bacteria leaking into circulation), this may be an additional factor explaining the results in A129 mice and also the difference between IRF1^−/− mice held under conventional versus SPF conditions.

Our results of anti-Listeria immune response in ICSBP^−/− mice suggested a major defect of IFN-γ-induced macrophage function, because lymphocytes, especially cytotoxic T cells, but also B cells, had been shown to function almost normally after viral infections (36, 39). Therefore macrophage functions were tested in vivo by adoptive transfer experiments and in vitro by PEM cultures.

RAG2^−/− mice are devoid of functional T and B cells, but have normal macrophages and natural killer cells (47). When infected with an intermediate dose of Listeria, they are able to control bacterial replication comparable to nude mice (46), but cannot eliminate the pathogen. To test whether ICSBP-deficient T cells could develop normal specific anti- Listeria immunity, we transferred on day 0 macrophage- depleted ICSBP^−/− spleen cells into RAG2^−/− mice, challenged them with a high dose of Listeria (2 × 10⁵ CFU i.v.) on day 1 and evaluated Listeria titers in liver and spleen on day 10 to look for efficiency of the specific immune response. As a positive control normal spleen cells and as a negative control no spleen cells were transferred. The result (Fig. 2) revealed no difference of Listeria counts between recipients of ICSBP^−/− and ICSBP^+/+ spleen cells; in contrast RAG2^−/− mice that did not receive spleen cells exhibited 100- (spleen) to 1,000-fold (liver) higher bacterial counts. This result indicates that ICSBP^−/− splenocytes (especially the mutant T cells) were able to promote elimination of Listeria as successfully as normal lymphocytes in cooperation with the intact macrophage compartment of the RAG2^−/− mouse.

To evaluate listericidal activity of macrophages of the different mutant mouse strains, we tested PEM in an in vitro killing assay. ICSBP^−/−, IRF1^−/−, and IRF2^−/− PEM were elicited by starch injection intraperitoneally, plated onto cover slips and cultured as described in Materials and Methods. After 42 h the cultures were infected with Listeria in vitro, and the number of infected cells determined at time point t₀ and after 7 h of infection. The results (Fig. 3) show that PEM of ICSBP^−/− and, to a lesser degree, IRF2^−/− mice allowed enhanced replication of Listeria, whereas PEM of IRF1^−/− and normal mice were able to reduce the bacterial load in these macrophage cultures. Also the number of bacteria per macrophage was higher in ICSBP^−/− mice (mostly more than 10 bacteria/cell) compared to their controls (0–4 bacteria/cell), revealing some macrophages with plenty of Listeria and typical comet tails (18). This finding confirms the defect in macrophage effector function, which correlates with the in vivo susceptibility of these mouse strains to Listeria (ICSBP^−/− >IRF2^−/− >IRF1^−/−).

The effector mechansim responsible for listericidal properties of macrophages is widely studied and still not clearly defined. ROI (14–19) as well as RNI (20–23) have been proposed to be of major importance. Therefore NO production (measured as nitrite accumulation in culture medium) and respiratory burst upon PMA stimulation (H₂O₂ production measured by chemiluminescence) were tested in PEM cultures stimulated with LPS and/or IFN-γ as described in Materials and Methods. NO production (Fig. 4,A) was absent in IRF1^−/− mice confirming earlier results that iNOS cannot be induced by IFN type I or II combined with LPS and/or TNF in the absence of IRF1 (48, 49). In contrast, iNOS activity was normal in ICSBP^−/− and in IRF2^−/− mice as well as in G129 mice (50). This finding in IRF2^−/− mice differs from recently published results (51). The high susceptibility of the ICSBP^−/− and G129 strains to Listeria infection in vivo and in vitro indicates that the NO effector mechanism does not play a limiting role in Listeria clearance. The phenotype of IRF1^−/− mice found here is compatible to the published results of Listeria infection in iNOS-deficient mice (23). These mice showed also a slightly enhanced bacterial replication (10–100-fold) and a higher lethality to Listeria infection, but only after injection of 6 × 10⁴ CFU i.v. The LD₅₀ of iNOS-deficient mice was only a factor 10 lower compared to their normal littermates, whereas in the case of ICSBP^−/− mice this difference is more than 10,000fold (Table 1; LD₅₀ for C57BL/6 mice is ∼3 × 10⁵ CFU [52]). The fact that iNOS induction in the susceptible strains (ICSBP^−/−, IRF2^−/−) was higher than in controls (Figs. 4,A, and 5) could even indicate that NO may have a toxic effect on infected cells during murine listeriosis.

With respect to respiratory burst upon PMA challenge, all mouse strains showed comparable basal activity of unstimulated cultures; however, in ICSBP^−/− and IRF2^−/− mice ROI production could not be stimulated by IFN-γ (Fig. 4 B) and was 3–5 min delayed compared to control mice (data not shown). This finding suggests that deficient ROI production might be partially responsible for the high susceptibility of ICSBP^−/− mice to Listeria, but it cannot fully explain the drastic phenotype of these mice. At least a third effector pathway not yet known may have to be evoked to explain this phenotype (see Discussion). In addition, the fact that LPS-induced respiratory burst was enhanced in IRF2^−/− mice correlates inversely with a recently published finding of high IRF2 levels in LPS-hyporesponsive mouse strains (53). Thus, IRF2 may mediate the macrophage deactivating effect of LPS (42).

Apart from macrophages hepatocytes are a major target cell in murine listeriosis. They are infected by direct cell-to-cell spread of the pathogen that is able to associate with actin filaments of the cytoskeleton (54). Hepatocytes can produce NO. Therefore, to see whether the phenotype of ICSBP^−/− and IRF2^−/− mice is due to a localized inability of iNOS expression in the liver, immunohistological analysis of iNOS expression in liver (Fig. 5) and spleen (not shown) after Listeria infection was performed. Mice of the three gene-targeted and control strains were infected with Listeria (5 × 10³ CFU i.v.). On day 5 or 6 liver and spleen were taken, cryosectioned, and then immunostained for Listeria and for iNOS with an appropriate polyclonal rabbit antiserum. Induction of iNOS comparable to wild-type mice could be demonstrated in all strains except IRF1^−/− (Fig. 5, C, F, K, M, O). It was abundant in regions where Listeria and abcesses were found (detail shown in Fig. 5, G and H). In addition fulminant Listeria proliferation in liver (Fig. 5,E) and spleen of ICSBP^−/− mice was found with accompanying tissue destruction (Fig. 5,D) correlating with the high bacterial titers (Fig 1). This analysis also shows that the susceptibility of ICSBP^−/− and IRF2^−/− mice to Listeria is not due to inefficient NO production in the liver. In contrast, IRF1^−/− mice were well protected and did not express iNOS in the liver. This cannot be explained by earlier decline of the bacterial load because these mice had equal Listeria titers in the liver as wild-type mice 5 d after infection (Fig. 1 A). The same analysis was performed on spleen sections with comparable results (not shown).

The type II IFN system has been shown to be of crucial importance for immunity against Listeria, because mice deficient for IFN-γ (11) or the type II IFN receptor (12) are highly susceptible to this bacterium. However, the intracellular signalling pathway and the final effector mechanisms involved in Listeria clearance are only incompletely understood. The presented analysis of three gene-targeted mouse strains deficient of ICSBP, IRF1, or IRF2 with respect to their capacity to survive and eliminate Listeria in vivo and in vitro and to the ability of their macrophages to respond with ROI or RNI production upon IFN-γ stimulation suggests three major conclusions: (a) ICSBP/IRF2 complex (but not IRF1) is of crucial importance for murine innate immunity to Listeria in vivo and in vitro, (b) iNOS induction and NO synthesis play no limiting role for anti- Listeria activity of macrophages upon IFN-γ stimulation, (c) stimulation of ROI production by IFN-γ together with a postulated third yet unknown effector pathway in macrophages may be responsible for protection in the early phase of primary Listeria infection. The role of NO for antimicrobial activity of macrophages has been tested in other infectious model systems. It seems to play an important role in leishmaniasis (55, 56) and tuberculosis (48), but has no limiting effect in toxoplasmosis (57) and listeriosis (this study, references 58, 59).

Our results of the analysis of mice deficient for IFN type I or II receptors revealed surprisingly that the type I IFN receptor-deficient mice (A129) were better protected than their normal littermates. This finding may reveal in vivo competition of transcription factors of both signaling pathways at the DNA binding level (45) suggesting that absence of the type I system enhances function of the type II system and concurrently Listeria protection. A potentiating effect of LPS leaking through from intestinal bacteria leading to macrophage preactivation may be involved.

From the analysis of TNF receptor 1-deficient (13) mice it is known that TNF-α is a second important cytokine for protection against Listeria. It is produced by macrophages upon infection with Listeria and may act via the following two pathways: (a) the SCID model revealed that TNF-α is necessary for activation of NK cells that then produce IFN-γ to further induce TNF receptor 1 and TNF-α expression (60, 61) and macrophage effector functions (4); (b) macrophage- or γδ T cell–derived TNF-α may act in an autocrine or paracrine fashion directly on macrophages to activate anti-Listeria effector molecules. Involvement of the ICSBP/ IRF2 complex in the signaling cascade of the TNF receptor could theoretically explain the described in vivo findings, but this has not been formally demonstrated so far. In addition, another transcription factor, NF-IL6, which can be upregulated by LPS/CD14 and also by TNF-α (62), is important for clearance of Listeria as demonstrated in the NF-IL6–deficient mice (63).

From our findings in three different mouse strains and from the published literature, the following model of signaling events in activation of anti-Listeria immunity may be proposed (Fig. 6): after activation of IFN receptors various tyrosine kinases are induced and STAT proteins phosphorylated (Table 1); they regulate the induction and activation of transcription factors of the IRF family among which the exclusively IFN-γ-dependent ICSBP mediates protection against Listeria. Two major questions remain open: (a) What molecules are involved in the signalling of the TNF receptor that could explain its importance for anti-Listeria immunity (ICSBP, NF-IL6, other transcription factors, indirect effect via NK cell activation)? (b) How do macrophages kill Listeria? Our results, but also the published ones on IFN type II receptor- and iNOS-deficient mice, rather argue against RNI production being a limiting factor. ROI may be involved since ICSBP- and NF-IL6–deficient mice had reduced respiratory burst, and this correlated with high susceptibility to Listeria infection. But still there may be a potential third mechanism involved to explain the drastic phenotype of ICSBP^−/− mice. Studies on iron metabolism of peritoneal macrophages (64) and murine β-thalassemia (65) suggested that iron scavengers lead to enhanced, and iron overload to reduced, resistance to Listeria by direct interference with the essential bacterial iron metabolism. Whether IFN-γ– and/or TNF-α–mediated enhancement of iron-binding proteins can explain resistance to murine listeriosis remains to be investigated.

We would like to thank A. Schaffner and H. Hengartner for expert advice and helpful discussion; C. Weissmann and M. Aguet for mutant mouse strains; J. Bille for anti-Listeria serum; A. Althage, L. Vlk, and H. Haber for excellent technical support; H. Neff and N. Wey for photographs; and E. Hörhager, and S. Kläusli for secretarial help.

This work was supported by the Swiss National Science Foundation grant no. 31-32195.91 to R.M. Zinkernagel, and grants no. 31-4577.95 and 32-42536.94 to G. Schoedon), the Kanton of Zürich and the German Science Foundation (DFG).