Lethal Escherichia coli and Salmonella typhimurium endotoxemia is mediated through different pathways

2.1 Survival of TNF−/−LT−/− mice during endotoxemia

To assess the role of TNF in mediating LPS toxicity, mice deficient in TNF and LT and wild-type controls were injected with LPS from various Gram-negative bacteria. Death occurred during the first 5 days after LPS challenge, without further deaths thereafter. TNF−/−LT−/− were slightly more resistant to low and moderate doses (0.5 – 1.5 mg/mouse) of S. typhimurium LPS than TNF+/+LT+/+ controls, whereas higher doses (2 mg/ mouse) lead to 100 % mortality in both control and deficient mice (Fig. 1 A). In contrast, TNF−/−LT−/− mice were completely resistant to challenge with LPS derived from E. coli, irrespective of the dose injected (Fig. 1 B). Similarly, injection of LPS from K. pneumoniae into TNF−/−LT−/− mice resulted in 100 % survival at all LPS doses tested, whereas only 20 % of the TNF+/+LT+/+ control mice survived when injected with 1 or 1.5 mg, and there was no survival when injected with 2.0 mg K. pneumoniae LPS (p < 0.01). Thus, differential susceptibility of TNF−/−LT−/− to various LPS types is not a phenomenon restricted to E. coli LPS.

2.2 Cytokine concentrations in vivo

In TNF+/+LT+/+ mice, the highest concentrations of TNF were found 90 min after challenge with either 0.25 mg of S. typhimurium or E. coli LPS, whereas TNF was not present in the TNF−/−LT−/− mice (Fig. 2 A and B). After injection of S. typhimurium LPS into TNF+/+LT+/+ mice, maximal IL-1α and IL-1β concentrations were found between 90 min and 6 h after injection of LPS (Fig. 2 A). In TNF−/−LT−/− mice challenged with S. typhimurium LPS, there was an ongoing production of both IL-1α and IL-1β at 3 to 6 h after injection of LPS, and the levels were significantly higher than those recorded in the control animals (p < 0.01) (Fig. 2 A). In contrast, after challenge with E. coli LPS, IL-1α and IL-1β kinetics were identical in TNF−/−LT−/− and TNF+/+LT+/+ mice, but the deficient mice produced significantly lower amounts of these two cytokines compared to controls (p < 0.05) (Fig. 2 B).

The production of IFN-γ 6 h after challenge with LPS was significantly lower in the TNF−/−LT−/− mice than in TNF+/+ mice, independently of the LPS species (Fig. 3). It is, however, important to underscore that TNF−/−LT−/− mice produced 16-fold more IFN-γ after challenge with S. typhimurium LPS, then with E. coli LPS (23.4 ± 8.9 vs. 1.4 ± 0.5 pg/ml, p < 0.01) (Fig. 3). In contrast, the production of the antiinflammatory cytokine IL-10 was significantly lower in TNF−/−LT−/− mice challenged with both LPS species compared to wild-type mice (Fig. 3). However, mice stimulated with S. typhimurium LPS produced significantly less IL-10 then those challenged with E. coli LPS (Fig. 3). Similar data were obtained when mice were challenged with a lethal dose of LPS (1.5 mg/ mouse; Table 1).

2.3 Effect of cytokine neutralization

To investigate whether the up-regulation of IL-1α and/or IL-1β production in the TNF−/−LT−/− after challenge with S. typhimurium LPS is responsible for the mortality in the absence of TNF and LT, groups of deficient animals were infused with either IL-1Ra or saline through osmotic minipumps, before and during challenge with endotoxin. Saturation levels of IL-1Ra did not protect TNF−/−LT−/− mice against the lethal effects of S. typhimurium LPS (Fig. 4 A). However, treatment of TNF−/−LT−/− mice with anti-IFN-γ antibodies tended to have a beneficial effect against the lethal effects of S. typhimurium LPS (mean survival time 91 ± 23 vs. 48 ± 18 h), although the difference did not reach statistical significance (p > 0.05 log rank test; Fig. 4 B). The protective effect of the anti-IFN-γ antibodies was not potentiated by the concomitant infusion of IL-1Ra (mean survival time 72 ± 25 vs. 38 ± 10 h in controls, p > 0.05). Thus, IL-1α and/or IL-1β do not seem to play a major role in the pathogenesis of Salmonella LPS-induced shock and death, but IFN-γ is suggested to be involved.

2.4 Cytokine redundancy in mortality induced by S. typhimurium LPS

To investigate the role of IFN-γ in the pathogenesis of endotoxic shock induced by S. typhimurium LPS, IFN-γ−/− mice were challenged with a low dose (0.25 mg/mouse) of Salmonella LPS. The survival was 67 % in IFN-γ−/− mice vs. 0 % in wild-type mice (p < 0.05; Fig. 5 A). However, when IFN-γ−/− mice were challenged with higher (1 mg/mouse) doses of S. typhimurium LPS, no protection against endotoxic shock was apparent (Fig. 5 B). This observation suggests that IFN-γ contributes, but is not solely responsible for the mortality after challenge with S. typhimurium LPS.

Because the data presented above suggest that Salmonella LPS is able to induce several cytokines with overlapping activities that contribute to lethality, we challenged ICE−/− mice with LPS. ICE−/− mice are not able to process immature precursors into mature IL-1β and IL-18, and as a consequence are also deficient in the production of IL-1α, IFN-γ and TNF 18. When ICE−/− mice were challenged with 1 mg of LPS, they were completely resistant to both E. coli and S. typhimurium LPS (100 % survival in ICE−/− mice, compared with 0 % survival in control animals, p < 0.01). There was no difference in the course of mortality between E. coli LPS and S. typhimurium LPS in control mice.

2.5 The role of endotoxin-associated proteins in the effects of E. coli and S. typhimurium LPS

In our experiments, we have used commercial LPS with a protein content of less than 10 %. However, because even these small amounts of endotoxin-associated proteins may influence cytokine induction by LPS, as previously suggested by some 19, although not all 20 authors, we tested whether the mortality induced in TNF−/−LT−/− mice by S. typhimurium LPS was due to qualitative or quantitative differences in endotoxin-associated proteins. While commercial LPS preparations induced small, but detectable amounts of TNF in peritoneal macrophages of LPS-resistant C3H/HeJ mice (190 ± 74 pg/ml after E. coli LPS, and 221 ± 59 pg/ml after S. typhimurium LPS), indicating the presence of endotoxin-associated proteins, the extra-purified LPS preparations (with undetectable protein contamination) from either E. coli or S. typhimurium were not able to induce any detectable TNF, IL-1α and IL-1β synthesis (all values below detection limit). The role of endotoxin-associated proteins for mortality was tested by injecting either commercial or purified S. typhimurium LPS into TNF−/−LT−/− and TNF+/+LT+/+ mice. Both commercial and extra-purified Salmonella LPS induced similar mortality patterns in the knock-out and control mice (Fig. 6). The remaining 1 % contamination of purified LPS with bacterial DNA is unlikely to be important, as C3H/HeJ macrophages did not respond with any cytokine production upon stimulation with purified E. coli or Salmonella LPS. Since bacterial DNA stimulates cytokine production through TLR9 whichis functional in C3H/HeJ mice, significant contamination with bacterial DNA would have led to induction of cytokines in these mice, which was not the case with our purified LPS preparations.

In additional experiments, we investigated whether either LPS or endotoxin-associated proteins from the commercial preparations of E. coli or Salmonella LPS may differentially induce mortality in the absence of TLR4-dependent signals in the C3H/HeJ mice. While both preparations induced 80 to 100 % mortality in the susceptible C3H/HeN mice, no mortality was recorded inthe C3H/HeJ mice, independently of the LPS species injected in the mice (not shown).

4.1 Materials

LPS from E. coli (serotype O55 : B5), S. typhimurium and K. pneumoniae were obtained from Sigma Chemical Co (St. Louis, MO). Recombinant IL-1 receptor antagonist (IL-1Ra) was a kind gift from Amgen (Boulder, CO). Hybridoma cells producing rat anti-mouse IFN-γ monoclonal antibodies (clone B4-2A6) were obtained from the ATCC (Rockville, MD). Control rat IgG antibodies were obtained from Sigma.

4.2 Animals

TNF−/−/LT−/− mice were produced as previously described 23. IFN-γ−/− and the corresponding wild-type control mice were kindly provided by Organon (Oss, The Netherlands).The generation and background of ICE−/− mice were previously described 18, 41. C3H/HeJ and C3H/HeN mice were obtained from the Jackson Laboratory (Bar Harbour, ME). Specific pathogen-free knock-out mice and age- and weight-matched wild-type mice (20 – 25 g, 6 to 8 weeks old) were used. Mice were fed sterilized laboratory chow (Hope Farms, Woerden, The Netherlands) and water ad libitum.

4.3 Endotoxemia model

Groups of normal and deficient (TNF−/−LT−/−, IFN-γ−/−, or ICE−/−) mice were injected intraperitoneally (i.p.) with LPS in amounts varying from 0.25 to 2.0 mg/mouse. Survival was assessed during 1 week in groups of at least 10 mice. In addition, 30 min, 90 min, 3 h and 6 h after challenge with LPS, five animals per group injected with either 0.25 or 1.5 mg LPS/mouse were anesthetized with ether and bled from the retroorbital plexus for measurement of circulating cytokine concentrations.

In an additional experiment, the role of IL-1 and IFN-γ or the combination of both for the mortality induced by S. typhimurium LPS was assessed. IL-1Ra was administered to TNF−/−LT−/− mice using osmotic minipumps (model 1007D, Broekman Institute, Someren, The Netherlands). The pumps were loaded with 1.2 mg IL-1Ra and implanted 1 day before the LPS (2 mg/mouse) challenge. Thepumps give a sustained IL-1Ra release for 7 days, assuring an efficient neutralization of endogenous IL-1 42. A control group was implanted with minipumps loaded with sterile phosphate buffered saline. Anti-IFN-γ antibodies (0.5 mg/mouse) or control rat IgG were administered i.p. to the TNF−/−LT−/− mice, just before LPS (2 mg/mouse) administration. The survival was assessed for 1 week.

4.4 The role of endotoxin-associated proteins in the mediation of E. coli and S. typhimurium LPS efects

As "endotoxin associated proteins" have been shown to modulate LPS-induced cytokine production, we assessed whether protein contamination in commercial LPS preparations may be responsible for the possible differences between E. coli and S. typhimurium LPS. Commercial LPS preparations were purified as previously described by Manthey and colleagues 19. This method decreased protein contents of LPS preparations from 9 % (E. coli LPS) and 8 % (S. typhimurium LPS) to undetectable levels (as measured using the BCA assay), with a more than 90 % recovery of LPS 19. The DNA contamination (as detected using a Genequant spectrophotometer) decreased from 3.5 % and 3.1 % (E. coli and Salmonella LPS) to 1.3 % and 1.0 %, respectively. The purity of the newly extracted LPS preparations was tested by stimulation of cytokine production in peritoneal macropahges of LPS-resistant C3H/HeJ mice, as described 43. Briefly, resident peritoneal macrophages were harvested by injecting 4 ml of sterile PBS containing 0.38 % sodium citrate. After centrifugation and washing, thecells were resuspended in RPMI 1640 containing 1 mM pyruvate, 2 mM L-glutamine, 100 μg/ml gentamicin and 2 % fresh mouse plasma (culture medium). Cells were cultured in 96-well microtiter plates (Costar Corporation, Cambridge, MA) at 10⁵ cells/well, in a final volume of 200 μl. The cells were stimulated with 10 ng/ml of either commercial or extra-purified LPS preparations. After 24 h of incubation at 37 °C, the plates were centrifuged (500 × g, 10 min), and the supernatant was collected and stored at − 80 °C until cytokine assays were performed.

The role of endotoxin-associated proteins for the mortality induced by S. typhimurium LPS was tested by injecting TNF+/+LT+/+ and TNF−/−LT−/− mice with 1 mg of either the commercial or the extra-purified LPS preparation. In addition, a possible difference between E. coli and Salmonella LPS in inducing mortality through TLR4-independent pathways in C3H/HeJ mice was tested by injecting the animals with either 1 or 2 mg of the two LPS species. Mortality was assessed during 1 week after the challenge, as described above.

4.5 Cytokine measurements

TNF-α, IL-1α and IL-1β concentrations were determined using specific radioimmunoassays (RIA) developed in our laboratory 42. Murine IFN-γ ELISA was purchasedfrom Biosource (Camarillo, CA), and used according to the guidelines of the manufacturer. Murine IL-10 ELISA was purchased from CLB (Amsterdam, The Netherlands), and used according to the guidelines of the manufacturer.

4.6 Statistical analysis

Differences in concentrations of cytokines were analysed using the Mann-Whitney test, and by Kruskal-Wallis test where appropriate. Survival data were analyzed using the Kaplan-Meyer log rank test. Differences were considered significant at p < 0.05. All experiments employed at least 5 mice per group at each time point for the plasma cytokine concentrations, and 10 mice pergroup for survival experiments, unless otherwise indicated. Data are presented as means ± standard deviation.