Differences in Innate Defense Mechanisms in Endotoxemia and Polymicrobial Septic Peritonitis

Abstract

Loss, reduction, or enhancement of the ability to respond to bacterial lipopolysaccharide (LPS) has no influence on survival of mice in a model of postoperative polymicrobial septic peritonitis induced by cecal ligation and puncture (CLP). This was demonstrated by using either mice with a defective Tlr4 gene, which encodes the critical receptor molecule for LPS responses, or mice deficient for LPS binding protein (LBP) or mice sensitized to LPS by Propionibacterium acnes. Though interleukin-12 (IL-12) and gamma interferon (IFN-γ) play an important role in the sensitivity to LPS as well as in the resistance to several infections, loss of these cytokine pathways does not affect survival after CLP. Thus, neutralization of neither endogenous IL-12 nor IFN-γ altered mortality. In addition, IFN-γ receptor-deficient mice demonstrated the same sensitivity to CLP as mice with a functional IFN-γ receptor. However, administration of IFN-γ at the time of operation or pretreatment of both IFN-γ-sensitive and IFN-γ-resistant mice with IL-12 significantly enhanced mortality. This indicates that in the present infection model activation of innate defense mechanisms is not dependent on LPS recognition and does not require endogenous IL-12 or IFN-γ function. Indeed, exogenous application of these two mediators had deleterious effects.

During septic peritonitis gram-negative and gram-positive bacteria are transported via lymphatics and blood into vital organs. Lipopolysaccharide (LPS) is an important target molecule for recognition of gram-negative bacteria by the innate immune system and a widely used model substance which, depending on the dose, causes inflammation, shock, or death (7). LPS binding protein (LBP) is a plasma protein that accelerates binding of LPS to CD14 and thereby considerably enhances the host's sensitivity to LPS (23). Thus, the host perceives minute concentrations of LPS, which normally indicate a bacterial infection, and can mount an antibacterial response while the infection is still at an early stage. Inactivation of the LBP gene reduces the susceptibility of mice to LPS and increases their susceptibility to a Salmonella enterica serovar Typhimurium infection (12). Similarly, mutation or deletion of the Tlr4 gene which results in LPS unresponsiveness (22, 27) leads to a high susceptibility to infection with S. enterica serovar Typhimurium (21, 29) and encapsulated Escherichia coli (2).

The cytokines interleukin-12 (IL-12) and gamma interferon (IFN-γ) are important mediators of innate immune reactions, and both are released after bacterial infection or challenge with LPS. In a number of bacterial infection models survival or clearance of pathogens either requires IFN-γ or IL-12 or is enhanced when these cytokines are administered exogenously. IL-12 neutralization impaired the clearance of intraperitoneally (i.p.) instilled E. coli (33) and inhibited resistance against Yersinia enterocolitica (1), whereas treatment with IL-12 before or after infection with streptococci increased survival (19). Survival in a model of septic peritonitis (colon ascendens stent peritonitis [CASP]) required IFN-γ receptor (IFN-γR) activation (31), and IL-12 frequently exerts its protective effects through induction of IFN-γ (1, 32). Both cytokines, however, contribute to mortality after lethal LPS challenge, as demonstrated in mice pretreated with Propionibacterium acnes (5, 6) or infected with Mycobacterium bovis BCG (30). Recently, mice pretreated with P. acnes were shown to be highly sensitive to high-dose S. enterica serovar Typhimurium infection (M. Gumenscheimer and M. A. Freudenberg, unpublished data).

These findings raised the question of how LPS-insensitive (TLR4- and LBP-deficient) or LPS-hypersensitive (P. acnes-primed) mice would react in a more complex, clinically relevant model of septic peritonitis induced by cecal ligation and puncture (CLP). After CLP a postoperative, mixed, bacterial septic peritonitis, characterized by a septic focus and a protracted course of systemic infection, develops. The host becomes exposed to the whole range of intestinal flora, including gram-negative bacteria. Therefore, survival after CLP was expected to be improved by endogenous or exogenous IL-12 or IFN-γ. Surprisingly, however, we found that survival after CLP was not affected by loss of TLR4 or LBP, the presence of endogenous IL-12 or endogenous IFN-γ, or sensitization with P. acnes but was decreased by treatment with IL-12 or IFN-γ.

MATERIALS AND METHODS

Mice.

Male NMRI mice (25 to 30 g) were purchased from Charles River (Sulzfeld, Germany). IFN-γR-deficient (IFN-γR^−/−) mice (129/Sv) (11), LPS-nonresponder BALB/c/l mice carrying the mutated Tlr4 gene of C3H/HeJ mice (26), and the respective control mice were bred in the animal facilities of the Max-Planck-Institut für Immunbiologie (Freiburg, Germany). LBP-deficient (LBP^−/−) mice and their heterozygous littermates (12) were bred in the Institut für Immunologie (Greifswald, Germany).

CLP.

Mice were anesthetized by i.p. injection of 75 mg of Ketanest (Parke, Davis & Company, Münich, Germany)/kg of body weight and 16 mg of Rompun (Bayer AG, Leverkusen, Germany)/kg in 0.2 ml of sterile pyrogen-free saline (Fresenius AG, Bad Homburg, Germany). The cecum was exteriorized, and the distal end of the cecum (about 30% of the total cecal length for sublethal CLP and 70% for lethal CLP) was ligated and punctured once or twice (depending on the intended lethality) with a 0.9-mm-diameter needle as described previously (3). Mice were observed for 2 weeks.

Reagents.

Recombinant mouse IL-12, kindly donated by M. Gatley (Hoffman LaRoche Inc., Nutley, N.J.), was injected i.p. as indicated above or after CLP. Recombinant mouse IFN-γ was kindly donated by G. R. Adolf (Bender GmbH, Vienna, Austria), and 1 μg was injected i.p. immediately after CLP. P. acnes (strain ATCC 12930; American Type Culture Collection, Manassas, Va.) was grown and killed as described previously (13). For pretreatment with P. acnes mice were injected intravenously (i.v.) with 25 μg of heat-killed P. acnes/g 7 days before CLP. LPS W from Salmonella enterica serovar Minnesota 9700 was purchased from Difco Laboratories (Detroit, Mich.).

For neutralization of IFN-γ mice received 100 μg of protein A-purified rat anti-mouse IFN-γ monoclonal antibody R4-6A2 (24) i.p. immediately after CLP. To neutralize IL-12, mice received 1 ml of rabbit anti-mouse IL-12 antiserum (8) i.p. immediately after the operation. Control mice received 1 ml of normal rabbit serum. All sera were complement inactivated. The in vivo neutralization capacity of the antibody preparations has been demonstrated previously (8).

Quantitation of tumor necrosis factor (TNF) serum titers and IFN-γ serum titers was done using the bioassays for mouse TNF (3) and IFN (16), respectively, as described earlier.

RESULTS

Role of LPS sensitivity in septic peritonitis.

We compared the mortality of LPS-sensitive mice with that of LPS-resistant mice after CLP. BALB/c/l mice (26) are highly resistant to LPS due to a point mutation in the intracellular domain of TLR4, which is critical for transducing LPS responses (22). The sensitivity of these mice to CLP as measured by the mortality was not different from that of mice with normal LPS responsiveness (Fig. 1).

FIG. 1.

Mortality after CLP was not affected by TLR4 deficiency. BALB/c mice (n = 15) and BALB/c/l mice (n = 15) were subjected to CLP and mortality was recorded (P < 0.12; log rank statistic).

Furthermore, ablation of the gene for LBP, a molecule which facilitates LPS responses (12), did not alter the mortality in the CLP procedure (Fig. 2).

FIG. 2.

Mortality after CLP was not influenced by LBP gene deficiency. LBP^−/− mice (n = 11) and LBP^+/− mice (n = 11) were subjected to sublethal CLP, and LBP^−/− mice (n = 9) and LBP^+/− mice (n = 11) were subjected to lethal CLP. Mortality was recorded (P < 0.39 for sublethal CLP and P < 0.29 for lethal CLP; log rank statistic).

Role of IL-12 in septic peritonitis.

IL-12 has been shown on one hand to enhance mortality in LPS-induced shock (30) and on the other hand to be protective in a number of infection models (33). To determine whether endogenous IL-12 contributes to survival of mice after CLP-induced septic peritonitis, IL-12 was neutralized by treating mice with anti-IL-12 antiserum immediately after the operation. As can be seen in Fig. 3, there was no significant change in mortality due to this anti-IL-12 treatment, which has been shown in other experiments to effectively neutralize IL-12 in vivo (8).

FIG. 3.

Neutralization of IL-12 did not change survival after CLP. Groups of NMRI mice (n = 10) received either rabbit anti-IL-12 antiserum or normal rabbit serum immediately after CLP. Mortality was recorded (P < 0.46; log rank statistic). Ig, immunoglobulin.

Because exogenous IL-12 has been found to protect hosts from infections (19), we treated mice with recombinant IL-12 (rmIL-12). Administration of 10 to 100 ng of rmIL-12 immediately after CLP did not influence the outcome of CLP (Table 1). However, mice became more sensitive to CLP and the mortality increased when the animals were treated with 30 or 100 ng of rmIL-12 1 day prior to the operation or with 1 μg of rmIL-12 given two times before CLP.

TABLE 1.

Influence of rmIL-12 treatment on survival after sublethal CLP^a

Time of rmIL-12 treatment	No. of mice surviving/no. tested after treatment with rmIL-12 at (ng per mouse):
	0	10	30	100	1,000 (twice)
Immediately after CLP	4/5	3/5	5/5	5/5	nt
Day −1	5/5	5/5	2/4	1/5	nt
Days −2 and −1	11/12	nt	nt	nt	1/12

^aGroups of mice were treated with the indicated amounts of rmIL-12 i.p. immediately after sublethal CLP (day 0) or on day −1 or both days −2 and −1. The data are given as cumulative survival of the animals over a period of 7 days after CLP. nt: not tested. 

In mice pretreated in this way twice with 1 μg of rmIL-12, a low nonlethal dose of bacterial LPS induced IFN-γ production. Compared to results for nonpretreated animals the IL-12 treatment led to serum IFN-γ titers that were more than 10-fold enhanced (Table 2) 3 h after LPS injection. The pretreatment with rmIL-12 also enhanced LPS-induced TNF serum levels very strongly. Thus, rmIL-12-pretreated mice exhibited enhanced sensitivity to LPS and were more sensitive to CLP.

TABLE 2.

LPS-induced serum IFN-γ and TNF levels in untreated and IL-12 pretreated mice^a

Pretreatment	Level of:
	IFN-γ (U/ml) at (h):		TNF (U/ml) at (h):
	1	3	1	3
PBS	30	30	80	0
IL-12	120	480	5,120	2,560

^aGroups of mice (n = 3) which had been pretreated on day −2 and −1 with phosphate-buffered saline (PBS) or 1 μg of rmIL-12 i.p. received LPS (10 μg) from S. enterica serovar Minnesota i.p., and IFN-γ and TNF levels were determined in serum prepared from blood taken 1 or 3 h after LPS challenge. 

Role of IFN-γ in septic peritonitis.

To test whether IFN-γ is involved in the resistance to CLP-induced septic peritonitis, we compared the mortality of IFN-γR^−/− mice and that of the respective wild-type (wt) mice in this model. As shown in Fig. 4b the IFN-γR^−/− mice showed no change in mortality after CLP in comparison to control animals. Therefore we conclude that endogenous IFN-γ production is not essential for controlling the development of septic peritonitis after CLP. This was confirmed by the finding that neutralization of endogenously produced IFN-γ after CLP did not affect the outcome of CLP (Fig. 5).

FIG. 4.

No sensitization for CLP lethality by P. acnes pretreatment in normal or in IFN-γR^−/− mice. (a) Groups of IFN-γR^−/− mice and normal 129/Sv mice (n = 8) received either P. acnes or phosphate-buffered saline i.v. Seven days later a CLP of higher lethality was performed (P > 0.09 for each comparison of two Kaplan-Meier curves; log rank statistic). (b) Groups of IFN-γR^−/− mice and normal 129/Sv mice (n = 8) received P. acnes (25 μg i.v.). Seven days later a CLP of low lethality was performed. Mortality was recorded (P < 0.8; log rank statistic).

FIG. 5.

Postoperative neutralization of IFN-γ did not influence CLP lethality. Immediately after CLP of low lethality, groups of NMRI mice (n = 8) received either normal rat immunoglobulin G (IgG) or rat anti-mouse IFN-γ monoclonal antibody R4–6A2 (100 μg per mouse i.p.). Mortality was recorded (P < 0.9770; log rank statistic).

As shown above, pretreatment of mice with rmIL-12 enhanced mortality after CLP. IL-12 is a potent inducer of IFN-γ (1, 32). However, it is shown in Fig. 6 that pretreatment of IFN-γR^−/− mice with rmIL-12 also enhanced lethality after CLP. This indicates that IFN-γ plays no role in the enhanced susceptibility to septic peritonitis induced by IL-12. However, administration of 1 μg of recombinant IFN-γ (rmIFN-γ) at the time of the operation significantly enhanced mortality in wild-type mice. This could be seen both after sublethal CLP (Fig. 7a) and after CLP which led to 40% lethality (Fig. 7b). The amount of IFN-γ injected was not lethal per se for uninfected mice (data not shown).

FIG. 6.

Increased CLP susceptibility caused by IL-12 pretreatment did not depend on the IFN-γR. Groups of IFN-γR^−/− mice (n = 10) and normal 129/Sv mice (n = 7) received either rmIL-12 (100 ng per mouse) or phosphate-buffered saline (PBS) 24 h before CLP i.p. Mortality was recorded (P < 0.0014 for IFN-γR^−/− and IL-12 versus IFN-γR^−/− and PBS, and P < 0.0009 for normal 129/Sv mice treated with IL-12 versus normal 129/Sv mice treated with PBS; log rank statistic).

FIG. 7.

Postoperative administration of IFN-γ increased CLP lethality. (a) Immediately after sublethal CLP, groups of NMRI mice (n = 5) received either phosphate-buffered saline (PBS) or rmIFN-γ (1 μg per mouse i.p.). Mortality was recorded (P < 0.049; log rank statistic). (b) Immediately after CLP of low lethality, groups of NMRI mice (n = 5) received either PBS or rmIFN-γ (1 μg per mouse i.p.). Mortality was recorded (P < 0.004; log rank statistic).

Bacterium-induced IL-12 production and the subsequent IFN-γ response are the essential events involved in the sensitization of mice by P. acnes to LPS (5, 6) and some (but not all) of the other biologically active components of gram-negative and gram-positive bacteria (18). In this study we compared the sensitivities of P. acnes-primed and unprimed wild-type and IFN-γ^−/− mice to CLP. Surprisingly, no difference in susceptibility to CLP-caused death between control and sensitized mice and between wild-type and IFN-γR-deficient mice was observed, regardless of the severity of the CLP performed (Fig. 4).

DISCUSSION

The protective function of IL-12 and IFN-γ in a number of infection models has been reported (1, 32, 33), and impaired IL-12 production was correlated with increased susceptibility to postoperative sepsis in patients (10). At the same time IL-12 and IFN-γ have been found to be deleterious in LPS-induced shock (5, 30) or after high-dose infection with gram-negative bacteria (Gumenscheimer and Freudenberg, unpublished data). We therefore attempted to clarify the role of IL-12 and IFN-γ in a clinically relevant model of sepsis. The response to CLP, a model of a postoperative mixed sepsis, is clearly independent of the activation of the innate immune system via LPS, the principal endotoxic molecule from the outer cell walls of gram-negative bacteria. Mice (BALB/c/l) in which the sensing of LPS, and thus of gram-negative bacteria, is impaired were as sensitive in this CLP model as LPS-competent mice. This finding is in full agreement with the report of Mercer-Jones et al. (17), who obtained the same results with C3H/HeJ mice, which have the identical genetic defect. Furthermore, the fact that the presence or absence of LBP did not alter the outcome after CLP also indicated that the capacity to sense LPS as a danger molecule during infection with gram-negative bacteria (9) does not seem crucial for survival in this model. Even though gram-negative bacteria constitute part of the infection after CLP (28), it seems clear that constituents other than LPS dominate the initiation of the innate immune responses essential for localization of the septic focus and the induction of effective antibacterial mechanisms (4). In agreement, a strong enhancement of LPS susceptibility in mice by P. acnes did in no way influence the experimental outcome after CLP in this study. This is remarkable, since pretreatment with P. acnes does not sensitize animals to LPS and, thus, to gram-negative bacteria only. As shown recently, the pretreatment of mice with P. acnes increased the susceptibility to a not yet identified component of gram-positive Listeria monocytogenes and to a macrophage-activating lipopeptide (MALP-2) of Mycoplasma spp. (18).

Keeping in mind that LPS is not the key molecule determining survival after CLP, the relevant mechanisms for survival remain to be determined. Obviously, these mechanisms are different from those induced by LPS. Clearly, IFN-γR activation is not critical for survival in CLP-induced septic peritonitis. This is in sharp contrast to the situation after infection with several other microbial organisms. It is also in contrast with the finding of the protective role of IFN-γ in a different septic peritonitis model, called CASP (31). The explanation for the different IFN-γ requirements in the two peritonitis models with mixed bacterial infection might have its roots in the critical need for abscess formation after CLP in order to localize the septic focus (4). In contrast to the largely localized inflammatory response after CLP, a systemic response after CASP, which more closely resembles that developing after challenge with LPS, may be involved in survival. The differences between the systems is emphasized by the fact that, far from reducing mortality, exogenous application of IFN-γ directly after CLP was actually deleterious in our experiments, which is in agreement with the findings of Miles et al. (20).

IL-12 did not exert any protective effect in CLP-induced septic peritonitis. This is surprising in the light of the findings by Steinhauser et al. (25), who found that neutralization of endogenous IL-12 prevented the organization of the damaged cecum wall after CLP, and of the results from a study in which IL-12 treatment increased resistance of mice to a streptococcal skin infection (19). As shown here, IL-12 pretreatment, particularly after repeated bolus injection, had a rather deleterious effect in our model. The effect of IL-12 pretreatment in this study is suggestive of a sustained systemic inflammatory response which might contribute to toxicity (14). Our results show that the enhanced mortality after IL-12 pretreatment is not due to the endogenous production of IFN-γ. It is, however, entirely possible that enhanced LPS-elicited TNF induced by IL-12 pretreatment could contribute to the observed enhanced sensitivity. TNF has clearly been shown to be critically required for survival after CLP (3), but, at the same time, CLP-induced septic peritonitis makes mice exquisitely sensitive to TNF (15). The mechanism by which IL-12 and IFN-γ enhance mortality after CLP is not known and has to be elucidated in future studies. In this respect it is interesting that treatment of mice with P. acnes, which is known to induce elevated, long-lasting IL-12 and IFN-γ production, did not exert any deleterious effect in our infection model. It is possible that some positive, not yet identified effects of P. acnes treatment mask the negative effects of the above cytokines in the CLP model.

Over the years experiments with LPS have provided a large amount of valuable information on the mechanisms that the innate immune system sets in motion to mount an inflammatory reaction. However, since different pathogens employ very different strategies, it is clear that the host must be able to respond in ways optimized to combat each infecting agent depending on the infection route. No single host response will be able to cover all eventualities, and in this sense it is not surprising that different receptors and cytokine pathways seem to be crucial for survival in different modes of infection. CLP initiates a response more complex than that resulting from LPS or from infection with a single bacterial species. This clinically relevant polymicrobial infection may induce a wide range of innate immune mechanisms with overlapping effects. In this complex mixture each bacterial species may act as an adjuvant for the others. In this sense it is not altogether surprising that there is no discernible effect of disabling individual elements of the innate immune system (LBP, TLR4, IL-12, or IFN-γR). Following this line of thought, we tested whether sensitization with P. acnes would not perhaps also provide such an adjuvant effect. That this did not occur, although direct application of IFN-γ or of IL-12 did have a dramatic effect, may simply reflect the importance of carefully regulated cytokine responses in complex infections and in sepsis.