Depletion of neutrophil extracellular traps in vivo results in hypersusceptibility to polymicrobial sepsis in mice

Peritonitis and sepsis are frequent complications of critically ill patients after trauma or undergoing abdominal surgery [1‐3]. Peritonitis and sepsis, frequently related to a polymicrobial infection with extracellular microorganisms from the intestinal flora, continue to cause high morbidity and mortality [1, 4‐6].

Neutrophils (PMN) represent the first line of innate immune defense against extracellular microorganisms [3, 7‐9]. Equipped with a full arsenal of antimicrobial proteins at their disposal, PMN have a well described role in phagocytic uptake and intracellular killing [10, 11]. Besides their ability to eliminate microorganisms by phagocytosis, it has recently been shown that activation of PMN causes the formation of neutrophil extracellular traps (NETs) [12‐15]. This novel mechanism, for the first time described by Brinkmann et al., consists of the release of web-like structures of DNA and proteins that bind, disarm and kill pathogens extracellularly [12].

NETs have been consistently shown to entrap and kill gram-negative and gram-positive bacteria, as well as fungi [11‐14, 16]. Moreover, the confinement of pathogens to a local site of infection might be an important function of NETs [16]. Nevertheless, it has been suggested that the antibacterial mechanism of NETs formation occurs at the expense of injury to the host [17]. Exposure of NETs-associated proteases and other granular proteins to the extracellular environment has been shown to damage endothelial cells in vitro [17‐19]. In models of lipopolysaccharide (LPS) induced endotoxemia, NETs have been identified in the microcirculation of the liver in the sinusoids resulting in an impaired perfusion and increased tissue damage measured by levels of alanine aminotransferase (ALT) [17].

The latter observations leave NETs as a potentially double-edged sword. On the one hand, NETs represent an important mechanism of bacterial killing and preventing the spread of pathogens from the initial site of infection. On the other hand, NET formation may have deleterious effects on the host due to the release of high levels of potentially noxious proteins like proteases inducing damage to the adjacent tissue [11].

Although NETs have been found to be abundant at sites of infection and observed in the circulation in infection and sepsis [12, 20‐22], the pathophysiogical relevance of NETs for in vivo conditions is still under debate. DNA is the main structural component and scaffold of NETs [12]. It has been shown that brief treatment of NETs with DNase abolishes microbial killing in vitro [12]. Of interest, bacterial trapping of Escherichia coli by activated PMN was reduced by prior exposure to DNase in vivo [17]. Recombinant human (rh)DNase catalyses the hydrolysis of extracellular DNA. Application of rhDNase has been shown to be safe and well tolerated in the treatment of patients with systemic lupus erythematosus and cystic fibrosis. Thus, administration of rhDNase offers the possibility to investigate the pathophysiological role of NETs for polymicrobial peritonitis and sepsis in vivo, which has not been investigated so far. The aim of this study was to evaluate the effects of the administration of rhDNase on the inflammatory response and bacterial spreading in a murine model of polymicrobial peritonitis and sepsis caused by cecal ligation and puncture (CLP).

This animal study shows that application of rhDNase leads to a profound and sustainable reduction of NETs-mediated bactericidal activity in vivo. More importantly, we could show that administration of rhDNase in a murine model of polymicrobial sepsis results in an advanced sepsis progression with temporarily increased mortality prevalence, enhanced bacterial dissemination as well as elevated neutrophil counts in sepsis-related organs, more severe organ injury, and higher levels of IL-6 in the circulation. Our data underline the important role of NETs as an accurate antibacterial mechanism in the early phase of infection in polymicrobial sepsis in vivo.

After an infection, PMN are the body's first line of defense and kill bacteria by phagocytosis or by the production of NETs. In a recent sepsis model, more bacteria have been found in NETs than were phagocytized by PMN or macrophages in vivo [17]. The authors concluded that snaring bacteria in NETs might be more effective than phagocytosis particularly under flow conditions. In the serious sepsis model presented here, the highest release of NETs was detectable within the first 24 hours after CLP. Ex vivo ability of PMN to release NETs was present over the entire observation period demonstrating that during sepsis there is no loss of PMN-mediated NET production. Interestingly, peak levels of induced NET release in PMN could be observed 24 hours after CLP showing good correlation with the highest cf-DNA values measured in vivo at this time. This suggests a possible priming of the NET-formation in the acute phase of sepsis. However, the above mentioned observations have some limitations since we studied PMN isolated from the bone marrow which might be functionally different from the circulating populations. The detailed signaling mechanisms leading to the shift from phagocytosis to NET formation during sepsis are poorly understood. Several factors have been described as being involved in the induction of PMN to form NETs. Observations in PMN of chronic granulomatous disease (CGD) patients, which exhibit severe defects in NADPH oxidase and fail to produce reactive oxygen species (ROS), revealed that ROS might contribute to the formation of NETs [26]. Furthermore, beside singlet oxygen, the Raf-MEK-ERK pathway, which is activated in many cells during sepsis, has been shown to be involved in NETs formation by promoting PMN activation [26, 27].

There is a growing body of evidence that the interaction of PMN and activated platelets is largely involved in sepsis and that activated platelets can mediate PMN to rapidly make NETs in vivo [11, 17]. It is well established that platelets, which are primarily involved in homeostasis, are decreased in sepsis. Thrombocytopenia is associated with severity and mortality during sepsis [28]. Clark et al. recently showed in a sepsis model in vivo that NETs formation induced by activated platelets was particularly evident in the liver sinusoids and the capillaries of the lung. Platelets become activated by LPS through the Toll-like receptor (TLR)-4 receptor and bind to PMN, thus triggering NET formation within a few minutes [17]. The authors suggested that the platelets function as a barometer in the blood, becoming activated only under very serious systemic infections to stimulate PMN to release the proteolytic and chromatin material, thereby increasing the capacity of the innate immune system to trap and kill circulating bacteria. These findings are in line with our data, showing an inverse correlation between NETs levels and platelets 24 hours after CLP. In mice treated with rhDNase, a strong decrease in cf-DNA/NETs levels could be determined accompanied by a significant raise in the number of platelets. It might be assumed that the dissociation of NETs by rhDNase provokes a platelet mobilization from the bone marrow to compensate reduced NET amounts. Thus, thrombocytopenia during sepsis might represent a sequel of an adherence of activated platelets to PMN during NET-formation. However, 48 hours after CLP the correlation between NETs amount and platelet counts was completely different with both decreasing dramatically over a period of 24 hours. Because we found PMN to be able to release NETs after ex vivo stimulation with PMA at this time, one might speculate that 48 hours after CLP, mice fail to produce enough NETs for disease control due to a lack of activated platelets.

Therapeutic administration of rhDNase is currently used to treat chronic pulmonary disease in cystic fibrosis and systemic lupus erythematosus [29]. There is evidence that treatment with rhDNase is associated with an improvement in lung function [12]. However, the influence of rhDNase treatment on bacterial colonization of the airways in cystic fibrosis has been investigated with conflicting results. While Frederisken et al. found rhDNase application to be beneficial to cystic fibrosis patients, in another more recent study the authors did not find any long-term effects of rhDNase on the pulmonary bacterial airway colonization [30, 31]. However, the most profound difference was found for Staphylococcus aureus and gram negative bacteria. Recently, the ability of important gram positive bacterial pathogens such as Streptococcus pneumoniae and also certain S. aureus strains to resist NET-dependent killing has been linked to their ability to secrete nucleases, which might also explain such a change in bacterial findings after rhDNase therapy. However, in this study we did not further differentiate bacterial infiltration in detail. Although macroscopic examination of CFU plates did not reveal any significant differences (data not shown), rhDNase therapy in sepsis potentially might lead to a shift of bacterial colonization to bacterial strains with a certain resistance to NET-dependent killing in vivo, which could influence the value of bacterial dissemination into organs and the outcome. Nevertheless, this speculative assumption should be addressed in further investigations.

Most pathogens are killed after they have been trapped by NETs [12, 16]. Pathogens entrapped in the NETs are dispatched possibly after being exposed to a high local concentration of antimicrobial proteins [12, 32]. On the other hand, it has recently been shown, that microorganisms captured by NETs and thought to be killed remain viable in the construct by undefined mechanisms [33]. Therefore, it has been suggested that confinement of bacteria to the local site of infection and thus prevention of further spread of bacteria, is another important function of NETs. Having shown that maximal release of NETs is achieved within 24 hours after CLP, this confinement of bacteria by NETs is obviously temporally restricted and their relevance in the late phase of sepsis seems to decline. Moreover, depending on the amount of bacteria, the physical barrier built by NETs may break down in time and allow bacteria to disseminate into sepsis-related organs.

During early sepsis, PMN become activated and lodge primarily in the capillaries of the lungs and the sinusoids of the liver [34]. In our study, neutrophil counts were significantly increased 24 hours after CLP in the lung and liver of mice treated with rhDNase and more pronounced injury has been found in these organs. IL-6, an important proinflammatory cytokine in response to infection, is increased after CLP in mice [35]. Indeed, in our model, blood levels of IL-6 increased significantly, reaching maximal levels 24 hours after CLP. In accordance with our previous data suggesting aggravation of infection by rhDNase treatment, IL-6 levels were found to rise faster in the rhDNase-treated group, and the highest cytokine concentrations were measured after six hours. These results indicate that the elimination of NETs exacerbates the inflammatory process early after CLP resulting in a shift of the inflammatory response to an earlier time point.

NETs formation plays an important role in bacterial control during polymicrobial sepsis in vivo. It becomes evident that depletion of NETs modulates the immune reaction and aggravates the pathology that follows a polymicrobial sepsis in vivo. Strategies of NETs induction rather than NETs destruction might be a promising therapeutic approach for septic patients in the future.