Crucial Role of TNF Receptors 1 and 2 in the Control of Polymicrobial Sepsis¹

Sepsis is generally defined as the result of a systemic inflammatory response caused by uncontrolled infection (1). This is still a major cause of morbidity and mortality in the intensive critical care unit (2). The most common causes are severe pneumonia and intra-abdominal infections, such as peritonitis and bacteremia, induced by surgical devices (3).

Upon bacterial infection or exposure to a large range of microbial products, two major types of systemic dysfunctions may be the basis of sepsis. On the one hand, an overwhelming inflammation may cause a profound suppression of the immune response. On the other hand, sepsis may induce a robust and overwhelming immune reaction leading to excessive production of mediators which, by inducing fever, cardiovascular dysfunction, and multiple organ failure, are harmful for the host (4). In the investigations of these complex pathophysiological syndromes, TNF was considered as a central mediator. Indeed, in vivo injection of TNF partially recapitulate symptoms of septic shock by inducing hypotension, cardiac dysfunction, and vascular leakage (5).

In this context, the role of TNF and its closest relative lymphotoxin α (LTα)³ have been extensively studied. TNF is produced by many cell types in vivo and it exerts numerous physiological functions by acting on specific receptors (6, 7). Upon cell activation, TNF is first expressed on the cellular membrane and then cleaved by the protease TNF-converting enzyme to soluble TNF trimer. Homotrimeric membrane and soluble TNF or LT mediate similar, but distinct functions (8, 9, 10) through their interaction with two receptors, TNFR1 (CD120α) and TNFR2 (CD120β). TNF has both beneficial and deleterious effects during infection. TNF enhances leukocytes recruitment and angiogenesis and accelerates the elimination of various pathogens such as Leishmania (11), Listeria (12), and Mycobacterium (13). In contrast, TNF causes mortality during sepsis and septic shock. TNF inhibitors protect mice from sepsis induced by LPS or bacteria inoculation (14, 15). TNFR1 deficiency protects mice from LPS/d-galactosamine shock as demonstrated using genetically modified mice (16) or soluble TNFR1 (17). In the colon ascendens stent peritonitis model, TNFR1-deficient mice present the same mortality rate as the B6 control mice (18). In the cecal ligation and puncture model, mortality is increased after TNF Ab neutralization or in TNFR1-deficient mice (15, 19). In contrast, few studies focused on the role of TNFR2 in the pathophysiology of sepsis, although a recent report has indicated that TNFR2 deficiency seems to increase susceptibility to sepsis (20).

After cell stimulation by various stimuli, including TNF-α itself, these two receptors can be proteolytically cleaved into two soluble forms, sTNFR1 and sTNFR2, which can be detected at high concentrations and for a prolonged period of time in the circulation of patients with various inflammatory diseases including LPS-induced sepsis (21, 22). These soluble receptors are involved in the control of cytokine activity by inhibiting their ability to bind their membrane receptors and generating a biological response.

For the first time and to better understand the role of the TNF-TNFR axis in polymicrobial sepsis and to understand why neutralizing TNF therapy has proved relatively unhelpful clinically (23), we investigated the immune and inflammatory responses in TNFR1/R2 double-deficient mice using a polymicrobial model of abdominal infection. We report here that TNFR1 and R2 deficiency is associated with protection as documented by neutrophil recruitment, bacterial clearance in the peritoneal cavity, and subsequent enhanced survival. Furthermore, TNFR deficiency leads to an attenuated hyperinflammation response associated with abdominal infection which prevents the induction of multiple organ failure and death. Neutrophils play a key role in the control of infection by limiting bacterial growth (24, 25). In the present study, we show that TNFR1 and R2 signaling pathways can modulate neutrophil homeostasis during bacterial infection by first enhancing chemotaxis and then by decreasing apoptosis. Finally, by contrast to results with TNFR1 deletion, pan-TNF neutralization using Enbrel, which blocks mouse TNF (26), revealed no increased protection from polymicrobial induced sepsis.

Adult male C57BL/6 (B6), TNFR1, TNFR2-, TNFR1/R2-, and TNF-deficient mice (obtained from The Jackson Laboratory) used in this study have been described before (27, 28, 29). Animals have been backcrossed at least 10 times on the C57BL/6 background. Animals were ∼20–25 g and 10–12 wk old. All mice were bred under specific pathogen-free conditions at the Transgenose Institute (Centre National de la Recherche Scientifique, Orleans, France). The animal experiments complied with the French government’s ethical and animal experiment regulations.

Sepsis was induced by i.p. injection of fecal preparation (0.200 ml/25 g body weight). This was prepared from the cecal contents of uninfected C57BL/6 mice bred under the same conditions and from the same batch. Cecal contents were suspended in saline and homogenized with the sterile disposable homogenization system Dispomix (Medic Tools) and were aliquoted and stored in 30% glycerol at −80°C. Control mice received 200 μl of normal saline. In some experiments, mice were treated with 30 or 50 mg/kg of Enbrel or with saline by i.p injection 1 h before and immediately after the injection of cecal contents.

Fecal samples were plated on brain-heart infusion-agar (Difco/BD Pharmingen) or COH-agar (Biomérieux) for total bacterial counts, Schaedler-agar to isolate the anaerobic flora, Pyocyanosel-agar to isolate Pseudomonas aeruginosa, Sabouraud to isolate yeast, Chapman to isolate Staphylococcus, Drigalski to isolate enterobacteria, and CNA to isolate enterococcus (Biomérieux). Plates were incubated at 37°C and CFU were observed after 24 h.

In survival studies, clinical score and rectal temperature were assessed every hour during the first 8 h after fecal peritonitis induction. Clinical score listed discernible symptoms such as ruffled fur, hunched posture, diarrhea, motor impairment, closed eyes, and coma and were each allotted one point. Rectal temperature was monitored using a rectal device (Physitemp Instruments). Mortality was assessed during 7 days.

Mice were sacrificed at 0, 1, 6, and 24 h after fecal peritonitis induction and a peritoneal lavage was performed with 3 ml of isotonic saline solution. Liver, spleen, and lung were harvested and separated into two parts. Blood was drawn from the lower cava vena. Sera was prepared by centrifugation at 3000 × g for 10 min at 4°C and then aliquoted and stocked at −80°C.

Peritoneal lavage fluid (PLF) was obtained by centrifugation at 400 × g for 10 min at 4°C. The supernatant (cell-free PLF) was stored at −80°C for cytokine analysis. An aliquot of the cell pellets was stained with Turk’s solution and counted, and 100,000 were cells centrifuged on microscopic slides (cytospin at 1000 rpm for 10 min at room temperature). Air-dried preparations were fixed and stained with Diff-Quik (Merz & Dade). Differential counts were made under light microscopy. One hundred cells were counted twice for the determination of the relative percentage of each cell type present in the PLF.

Peritoneal cells were collected 3 h after fecal peritonitis induction. Cells were washed once in PBS containing 0.5% BSA (PBS/BSA) and stained on ice at 10⁶ cells/100 μl with primary Abs: anti-GR-1 FITC or PE (clone 1A8), anti-CD11b PerCP (clone M1/70), anti-F4/80 PE (clone BM8), anti-Ly6C FITC (clone AL-21), and anti-CXCR2 (clone 242216) for 20 min in the dark. For the apoptosis assay, we used the Annexin V-FITC Apoptosis Detection Kit I. All Abs were from BD Pharmingen except CXCR2 Ab (R&D Systems). After washing with PBS/0.5% BSA, cells were analyzed on a BD Biosciences LSR I.

TNF, IL-1β, IL-6, keratinocyte-derived chemokine (KC), and MCP-1 in PLF and whole blood sera were measured by ELISA (Duoset Kit; R&D Systems) according to the manufacturer’s instructions (with detection limits at 50 pg/ml).

Ten-fold serial dilutions of PLF were plated on brain-heart infusion agar plates (Biovalley). Plates were incubated at 37°C and 5% CO₂, and the numbers of CFU were enumerated after 24 h. Organ homogenates were prepared in 3 ml of isotonic saline solution using a Dispomix tissue homogenizer (Medic Tools).

Liver homogenates were centrifuged at 10,000 × g for 10 min at 4°C and the supernatant was discarded. The pellets were resuspended in 1 ml of PBS containing 0.5% hexadecyltrimethyl ammonium bromide (HTAB) and 5 mM EDTA and then incubated for 2 h at 60°C to inactivate the endogenous catalases. PLF aliquots were pelleted and resuspended in 1 ml of PBS-HTAB-EDTA. Following a new centrifugation, 50 μl of supernatant was placed in test tubes with 200 μl of PBS-HTAB-EDTA, 1.6 ml of HBSS, 100 μl of o-dianisidine dihydrochloride (1.25 mg/ml), and 100 μl of 0.05% H₂O₂ After 15 min of incubation at 37°C under agitation, the reaction was stopped with 100 μl of 1% NaN₃. The MPO activity was determined as absorbance at 460 nm against blanck (reaction mixture with saline in place of sample).

Liver, spleen, and lung were fixed in 10% buffered formalin (Shandon). Tissues were dehydrated in ethanol and embedded in paraffin. Sections (3-μm thick) were stained with H&E for evaluation of pathological changes by two independent observers. All liver sections were graded as follows: grade 0, normal histomorphology; grade 1, minor inflammatory infiltrates with occasional liver cell necrosis; grade 2, moderate liver damage with inflammatory infiltrates and focal necroses; and grade 3, extensive infiltrates accompanied by diffusely distributed liver cell necroses. At least two separate sections were assessed per liver.

Murine bone marrow cells were isolated from femurs and tibiae in 2 ml of HBSS without Ca²⁺ and Mg²⁺ and laid on top of a two-layer Percoll gradient of 72 and 65% Percoll (Sigma-Aldrich) diluted in HBSS (100% Percoll was obtained by mixing nine parts of Percoll and one part of 10× HBSS) and centrifuged at 1200 × g for 30 min at room temperature without brake. The enriched neutrophil fraction was recovered at the interface between 65 and 72% Percoll. After washing twice with HBSS, 5 × 10⁻⁶ cells were obtained per mouse containing 95% of Gr-1-positive cells.

To examine the neutrophil chemoattractant response to MIP-2, a modified Boyden chamber assay was performed using a 48-well microchamber (NeuroProbe). Murine bone marrow neutrophils were isolated as above and resuspended in running buffer (HBSS 1× supplemented with 2 mg/ml BSA, 10 mM HEPES, 1 mM CaCl₂, and 1 mM MgCl2). Recombinant mouse MIP-2 (30 ng/ml) diluted in running buffer (for wells containing neutrophils) or appropriate buffer control was added to the lower chambers of the apparatus. A 5-μm pore polycarbonate membrane (NeuroProbe) was placed between the upper and lower chambers, and 5 × 10⁻⁴ cells in a volume of 50 μl were added to the top chambers of the apparatus. Cells were allowed to migrate into the membrane for 1 h at 37°C with 5% CO₂. Following incubation, the chamber was disassembled and the membrane was scraped and washed three times in PBS to remove nonadherent cells before being fixed and stained using Diff-Quik (Merz & Dade). Each well-associated membrane area was scored using light microscopy to count the intact cells present in five random fields. The results are expressed as the number of neutrophils per field.

Statistical evaluation of differences between the experimental groups was determined by using the log rank test for survival curves and both Student’s t test (comparing two groups) and one-way ANOVA followed by a Bonferroni post test (comparing more than two groups) for others data. All tests were performed with GraphPad Prism. All data are presented as mean ± SD. A p < 0.05 was considered significant: ∗, p < 0.05; ∗∗, p < 0.01; and ∗∗∗, p < 0.001.

Polymicrobial peritonitis and sepsis result in enhanced release of TNF and other proinflammatory cytokines causing a hyperinflammatory syndrome (30). To study the implication of the TNF pathway in polymicrobial sepsis-induced immune responses, we injected i.p. bacteria obtained from the cecum of control B6 mice. The inoculum represents mixed intestinal microbiota composed of aerobic and aero/anaerobic microbes and the main strains identified were Escherichia coli, Enterococcus, Staphylococcus, and Lactobacillus sp. (Fig. 1,A). First, we established a dose-response effect with the standard preparation of fecal preparation, which causes a dose-dependent hypothermia and death (Fig. 2).

The injection of 3.5 × 10⁵ CFU of these intestinal microbes induced a rapid local and systemic production of TNF which was maximal at 6 h (Fig. 1,B) and correlates with increased TNF serum levels (Fig. 1 C). Thus, polymicrobial sepsis leads to a marked local and systemic production of TNF.

Using TNF- or TNFR1-deficient mice, we asked whether TNF contributes to the development of the clinical signs of septic peritonitis. We observed reduced hypothermia in TNFR1- but not in TNF-deficient mice compared with feces-injected B6 mice (Fig. 1,D). These symptoms appear early after infection in B6 mice, predicting a poor prognosis (31). Thus, mice lacking the TNFR1 or the TNF gene are almost fully protected and had a survival rate around 70% after 7 days, whereas all B6 mice died within 2 days (Fig. 1 E). Therefore, these data suggested that TNF contributes to the negative outcome of acute polymicrobial septic peritonitis and that its incomplete inhibition will lead to a partial protection.

Following a peritoneal injection of 3.5 × 10⁵ CFU of intestinal microbes, TNFR1/R2 double-deficient mice were protected from clinical symptoms, which include reduced locomotion, diarrhea, and hunched posture (Fig. 3,A). Moreover, we observed reduced hypothermia in TNFR1/R2- deficient mice compared with B6 mice (Fig. 3,B). By contrast to results with TNFR1-only or TNF-deficient mice, mice double deficient in both TNFR1 and R2 were totally protected with 100% survival of polymicrobial sepsis at 7 days (Fig. 3 C). However, absence of TNFR2 unlike the combined absence of TNFR1 and R2 did not confer protection in this model of infection (data not shown). Together, these data showed that a complete abolition of the TNF-dependent signaling pathway can protect mice from an experimental polymicrobial sepsis.

Protection from polymicrobial sepsis and shock is associated with enhanced cell recruitment into the peritoneal cavity. Neutrophils and macrophages are the two main cell subtypes present in peritoneum after polymicrobial infection which are rapidly recruited and mediate the initial local inflammatory response (32). TNFR1/R2-deficient mice showed significantly higher numbers of neutrophils (Fig. 4,A) and macrophages (data not shown) in PLF, compared with B6 mice, upon infection. Both recruitment and neutrophil activation can be correlated with the quantification of the MPO activity (33). In fact, we observed that peritoneal MPO activity was significantly increased in TNFR1/R2-deficient mice compared with the B6 mice (Fig. 4 B).

Flow cytometric analysis of the inflammatory cells from PLF at 3 h after infection confirmed a dramatic increase of neutrophil recruitment in TNFR1/R2-deficient mice compared with B6 mice upon infection. Resident macrophages were diminished in B6 and absent in TNFR1/R2-deficient mice and inflammatory monocytes were increased upon infection in both groups (Fig. 4 C). Thus, neutrophil recruitment is clearly increased in the absence of TNF signaling.

Finally, the analysis of the local bacterial load in PLF showed that TNFR1/R2-deficient mice had significantly less viable bacteria at 24 h after infection than control B6 mice (Fig. 4 D). Moreover, cytospin preparations from the PLF at 24 h showed that neutrophils and macrophages from TNFR1/R2-deficient mice contained very few bacteria, whereas the B6 controls showed clearly detectable bacteria suggesting enhanced bacterial killing in the absence of TNF signaling (data not shown). Therefore, the innate host immune response to a polymicrobial infection is enhanced with more effective control of bacterial growth in the absence of TNF signaling.

TNF and also numerous cytokines are augmented in sepsis. They are known to activate cellular defense against infection while exaggerated cytokine production may lead to death (34, 35). Therefore, we investigated the local and systemic levels of inflammatory mediators upon polymicrobial infection in serum and PLF at 24 h after infection. This time point was chosen because TNF serum levels were still high. IL-6, IL-1β, KC, and MCP-1 in PLF (Fig. 5,A) and in serum (Fig. 5 B) were significantly lower in the absence of TNFR1/R2 compared with B6 mice. The overall increase of single mediators was not as dramatic as after endotoxin administration. However, the combined action of these mediators might cause hyperinflammation and tissue damage.

Collectively, these data indicate that cytokines and chemokines induced by polymicrobial sepsis, are in part dependent on TNF with reduced production in TNFR1/R2-deficient mice.

Microscopic examinations of the liver from B6 mice revealed focal inflammation with single-cell necrosis, focal neutrophil recruitment, and thrombotic lesions in the portal areas (Fig. 6,A). These hepatic changes, especially single-cell necrosis, were less pronounced in TNFR1/R2-deficient mice. The spleen was enlarged and showed congestion in the red pulp with lymphocyte depletion in the white pulp in B6 mice. These parameters were reduced in TNFR1/R2-deficient mice (Fig. 6,B). Lungs of B6 mice revealed a distinct alveolar congestion with cellular infiltrates around bronchi and capillaries and microthrombi in small vessels, while TNFR1/R2-deficient mice showed only minimal changes (Fig. 6 C).

The hepatic necrosis and neutrophil recruitment in the liver were assessed semiquantitatively. This analysis confirmed a significant protection from liver necrosis and inflammation in the absence of TNF signaling (Fig. 6,D), which correlated with the macroscopic changes. Since neutrophils contribute to liver injury, we quantified neutrophils by assessing MPO activity, which was augmented in B6 mice and significantly reduced in TNFR1/R2-deficient mice (Fig. 6 E). Therefore, the complete absence of TNF signaling reduced multiorgan failure associated with polymicrobial sepsis.

Reduction of neutrophil migration into the infectious focus during severe sepsis correlates with the severity of disease (36). This phenomenon may be the consequence of down-regulation of chemokine receptor CXCR2 on the surface of circulating neutrophils (37). Results in Fig. 3,B revealed that TNFR1/R2-deficient mice display increased neutrophil migration into the site of infection, suggesting that TNFR1/2 signaling might be involved in failure of neutrophil migration during severe sepsis. To test whether TNF negatively regulates neutrophil chemotaxis upon stimulation with CXCR2 ligand chemokines, we stimulated polymorphonuclear neutrophils from B6 mice with TNF at different concentrations and quantified the chemotaxis in response to MIP-2 in a modified Boyden chamber. TNF preincubation for 1 h diminished markedly neutrophil chemotaxis compared with the medium-treated cells. This reduction is similar to that induced by LPS (Fig. 7,A). Consistent with these first findings, we found that the chemotaxis of neutrophils from TNFR1-deficient mice was not down-regulated by LPS or lipoteichoic acid (Fig. 7 B), confirming that TNF leads to a marked reduction of neutrophil migration.

To further explore the role of TNF in neutrophil migration failure, we examined ex vivo the expression of CXCR2 on mice subjected to polymicrobial sepsis. Flow cytometric analysis of neutrophils (defined as Gr-1^highCD11b^high) showed that the absence of TNFR1/2 signaling significantly attenuated the reduction of CXCR2 expression observed in B6 neutrophils (Fig. 7 C).

TNF has a dual effect in neutrophil apoptosis. At low doses and coupled with others inflammatory stimuli, it prolongs the life span of circulating neutrophils. By contrast, a high dose and protracted exposure of TNF accelerates apoptosis (38). To test whether TNF will induce neutrophil apoptosis during polymicrobial sepsis, we measured ex vivo staining profiles for annexin V and propidium iodide (Fig. 7,D). We defined annexin V⁺ and propidium iodide⁻ as early apoptosis (R1), annexin V⁺ and propidium iodide medium as late apoptosis (R2), and annexin V^high propidium iodide^high as necrosis (R3). We observed that neutrophils from TNFR1/R2- deficient mice exhibit increased early apoptosis compared with the B6 controls. By contrast, neutrophils from B6 mice exhibited more late apoptosis (Fig. 7 E).

Since our investigations suggest that endogenous TNF may be a critical pathogenic factor of polymicrobial sepsis, we tested whether TNF neutralization with a soluble TNFR2 (Enbrel) modulates the outcome of polymicrobial inflammation. Mice were pretreated (−1 and 0 h) with Enbrel (30 mg/kg) or saline and then inoculated i.p. with the cecal contents. Enbrel slightly reduced the clinical symptoms of sepsis (Fig. 8,A) and hypothermia (Fig. 8,B) and partially enhanced survival (Fig. 8 C). Additional experiments with a higher dose of Enbrel (50 mg/kg) did not improve protection (data not shown). Furthermore, delayed injection (6 h after the infection) is ineffective since all of the treated mice died within 2 days (data not shown). Collectively, these data strongly suggest that complete and early TNF neutralization may be required to confer protection after sepsis onset.

This is the first report to demonstrate that combined loss of TNFR1 and R2 signaling pathways participate critically in the pathogenesis of lethal polymicrobial sepsis while TNF neutralization or single gene deletion does not provide the same level of protection.

To improve the understanding of the role of the TNF-TNFR axis in the pathophysiology encountered in sepsis, we used a standardized and highly reproducible murine model of polymicrobial sepsis also known as fecal peritonitis. We injected i.p. a mixed population of Gram-positive and -negative, aerobic, and anaerobic commensal bacteria. Using this model of polymicrobial peritonitis resulting in sepsis, we revealed that the host responds with a vigorous innate immune response with the release of proinflammatory cytokines, including TNF, and chemokines leading to monocyte and neutrophil recruitment. We further showed enhanced recruitment of immune cells in the absence of TNFR1/2. Control of the bacterial growth was enhanced, correlating with a total protection of TNFR1/R2- deficient mice compared with B6.

Then, we demonstrated that in the absence of TNFR1/R2, the late hyperinflammatory state was less pronounced than in the B6 controls. Systemic and local levels of IL-6 and IL-1β were reduced, which is consistent with the role of TNF on the production of these cytokines and with the fact that peritoneal sepsis induces a strong and rapid induction of proinflammatory mediators (39, 40). In addition, chemokine production was also dampened in TNFR1/R2-deficient mice. KC, the murine homolog of IL-8 which is known to be a poor prognosis marker in septic patients (41), was decreased during the late phase of septic peritonitis in TNFR1/R2-deficient mice while it was equivalent during the early part of the disease in both TNFR1/R2-deficient and B6 mice (data not shown). These findings were confirmed with MCP-1 which is also decreased when TNFR1 and R2 signaling is absent. These chemokines are known to induce the recruitment of neutrophils, particularly in septic peritonitis (42), but their accumulation will also lead to uncontrolled inflammation, organ failure, and lethal septic shock. Thus, our data highlighted the crucial role of TNFR1- and R2-mediated signaling pathways in the late hyperinflammation induced by sepsis, which is a poor prognosis for the host (43). These data were similar to those observed in IFNα-R1-deficient mice undergoing polymicrobial sepsis which identified type I IFN as a critical inducer of the late production of TNF, KC, and IP-10 (44). Furthermore, the present study showed that TNFR1/R2-deficient mice can maintain body temperature during the course of infection unlike B6 controls. This finding is in accordance with systemic organ dysfunctions that target liver, spleen, and lung during polymicrobial sepsis (45). Moreover, we observed that TNFR1/R2-deficient mice presented reduced central organ injuries compared with B6 controls. Multiple organ dysfunction syndrome (MODS) has been early defined as the main cause of morbidity in septic patient (46). It has been correlated with a loss of homeostasis in several interdependent organ systems including loss of control in body temperature. Septic patients who develop hypothermia have a significantly worse prognosis compared with those who have fever or maintain body temperature. In animal models, hypothermia is generally associated with immune dysfunction and poor outcome (47). Thus, TNFR1/R2-deficient mice were protected against systemic organ dysfunction which, in addition with a better bacterial clearance, will correlate the improved survival.

Massive neutrophil recruitment to the site of infection is an essential mechanism to control invading pathogens, especially in intra-abdominal bacterial infection (24, 48). In the present study, increased neutrophil accumulation in the peritoneal compartment was correlated with a better control of the bacteria. Thus, this mechanism may contribute to improved survival of TNFR1/R2-deficient mice. Chemokines are potent and specific chemoattractants for polymorphonuclear cells (49). The CXC family comprises several proteins such as MIP-2 or IP-10 that have been known for years to have a critical role in humans and animal models of disease (50, 51, 52). CXCR2 has been shown to mediate the responses to CXC chemokines in polymorphonuclear neutrophils (53). Under pathological conditions, surface expression of CXCR2 is down-regulated by 50% on neutrophils from septic patients (54, 55). Recent studies have shown that a proinflammatory environment can lead to a failure in neutrophil recruitment to the site of infection (37, 56). In the present study, we also demonstrated that TNF plays a role in CXCR2 down-regulation which leads to the failure of the neutrophils to migrate to the site of infection, as the neutrophils from TNFR1/R2-deficient exhibited an increased CXCR2 expression compared with the B6 controls.

One of the major pathways in limiting the inflammatory response is the clearance of neutrophils and their potentially cytotoxic content. Indeed, large numbers of apoptotic neutrophils or engulfed neutrophils in macrophages have been found in septic peritonitis (57) or in acute lung inflammation (58). Deregulation of immune cell apoptosis may be a component of the immune dysfunction and multiple organ failure that occur in sepsis (59, 60). Indeed, in sepsis, apoptosis inhibition protects animals from lethality (61). We therefore asked whether polymicrobial sepsis will modulate neutrophil accumulation by changing their ability to progress into apoptosis. The increased level of early apoptosis in TNFR1/R2-deficient mice may enhance bacterial and immune clearance and limit the inflammatory response. Furthermore, the decreased level of late apoptosis and, at the end, necrosis will prevent the generation of hazardous inflammatory components because B6 mice seem to exhibit more necrotic cells than TNFR1/R2 later after infection (data not shown).

Finally, the experiments with Enbrel revealed that this treatment partially protected mice from death without reducing temperature loss and clinical symptoms appearance. These findings are in accordance with the results obtained on TNFR1- or TNF-deficient mice that were only partially protected against the infection. These results can be explained by the fact that in TNF-deficient mice, LTα-dependent activation of the TNFR1/2 is present. Then, in TNFR1 single-deficient mice, TNF or LTα- dependent activation of TNFR2 is also effective. This would explain why the sole absence of TNFR2 is not protective. Finally, in both TNF- and TNFR1-deficient mice, the biological effect of sTNFR1 or sTNFR2, respectively, is still active. Moreover, a positive correlation exists between TNF-soluble receptors and simultaneously obtained sepsis score (62). Enbrel-treated mice submitted to an LPS shock only showed a partial cardiac function restoration and survival (63). Many studies in septic patients showed that TNF or IL-1 neutralization did not provide conclusive evidence of improving the outcome of clinical settings. Only one study has identified a small subset of critically septic patients that may benefit from TNF blockade, while in most others studies, no benefit was reported (64). In the context of a sepsis syndrome, our data suggest that an effective therapeutic neutralization of TNF effects should target both the TNFR1 and R2 signaling pathway and be initiated as soon as possible after the onset of the infection.

In conclusion, we demonstrate that the combined activation of TNFR1 and R2 signaling pathways play a detrimental role in an experimental model of polymicrobial sepsis. Moreover, the direct effect of TNF and the associated signaling pathways on neutrophil homeostasis during an ongoing infection was established and correlated with the induction of a delayed immune response and a hyperflammatory state which will lead to death. Blockade of both TNFR1/R2 may be required to confer substantial protection of polymicrobial sepsis.

We gladly thank Prof. Mauro Teixera and Prof. François Erard for assistance in manuscript preparation.

The authors have no financial conflict of interest.