Major injury is widely thought to predispose the injured host to opportunistic infections. This idea is supported by animal studies showing that major injury causes reduced resistance to polymicrobial sepsis induced by cecal ligation and puncture. Although cecal ligation and puncture represents a clinically relevant sepsis model, we wanted to test whether injury might also lead to greater susceptibility to peritoneal infection caused by a single common pathogen, Escherichia coli. Contrary to our expectation, we show herein that the LD50 for sham-injured mice was 103 CFU of E. coli, whereas the LD50 for burn-injured mice was 50 × 103 CFU at 7 days postinjury. This injury-associated enhanced resistance was apparent as early as 1 day after injury, and maximal resistance was observed at days 7 and 14. We found that burn-injured mice had higher numbers of circulating neutrophils and monocytes than did sham mice before infection and that injured mice were able to recruit greater numbers of neutrophils to the site of infection. Moreover, the peritoneal neutrophils in burn-injured mice were more highly activated than neutrophils from sham mice as determined by Mac-1 expression, superoxide generation, and bactericidal activity. Our findings suggest that the enhanced innate immune response that develops following injury, although it is commonly accepted as the mediator of the detrimental systemic inflammatory response syndrome, may also, in some cases, benefit the injured host by boosting innate immune antimicrobial defenses.
Severe injury can cause systemic inflammation with early multiple organ system dysfunction or can predispose the injured host to opportunistic infections as a consequence of perturbations of immune function. The complexity of the mammalian injury response is highlighted by the disparate inflammatory versus counter-inflammatory responses displayed by innate and adaptive immune cell types (1, 2, 3). The innate immune system shows a progressive increase in inflammatory reactivity, while cells of the adaptive immune system increasingly produce anti-inflammatory cytokines such as IL-10 and TGF-β (4, 5). This imbalance in immune regulation is thought to suppress antimicrobial host defenses while simultaneously promoting the development of the systemic inflammatory response syndrome (SIRS).3
A prominent feature of injury-induced changes in innate immune system function is increased responsiveness to a variety of TLR agonists by macrophages and dendritic cells. When exposed to TLR2 or TLR4 stimuli, these cells produce increased levels of proinflammatory cytokines such as IL-1β, TNF-α, and IL-6 (5). An increase in TLR reactivity suggests that injury might boost innate immune antimicrobial defenses. However, augmented TLR responsiveness might also lead to heightened and even lethal systemic inflammation if the injured host develops an infection (6, 7, 8). Thus, it is difficult to predict how the balance between potentially beneficial and detrimental effects of enhanced innate immune reactivity will influence the ability of the injured host to withstand an infection. Nevertheless, these injury-induced changes in immune system reactivity appear to be preserved across multiple different species, and they may represent an evolutionarily conserved response with survival benefits. The “danger” or “alarmin” theories of immune system reactivity would predict a beneficial role for the mammalian injury response because tissue injury should provide signals to enhance protective functions (9, 10).
In contrast, many clinical and animal studies have demonstrated that injured individuals are more susceptible to some types of infection. Work from our laboratory and others has shown that, compared with sham-injured animals, burn-injured mice are less resistant to cecal ligation and puncture (CLP)-induced sepsis (11, 12). CLP, however, is not a simple infection model. It not only causes polymicrobial sepsis with the multiple aerobic and anaerobic organisms normally found in bowel flora, but it also has the additional features of surgical trauma and tissue necrosis from the ligated cecum (13, 14). Consequently, it is difficult to determine whether mice succumb to infection, the stress of the surgery involved in performing the CLP procedure, the necrotic cecum, or a combination of all these insults.
Therefore, in this study, we wanted to examine the response of the injured host to a single infectious agent, specifically the common intestinal organism Escherichia coli. After a review of sepsis models found in the literature, we decided to use an i.p. infection procedure developed by Appelmelk et al. that uses a mucin and hemoglobin broth as the medium for the bacteria (15). The mucin/hemoglobin sepsis model avoids a problem associated with injecting organisms into the peritoneal cavity with a simple solution such as PBS, which usually requires a very large inoculum to cause mortality. When high numbers of bacteria are used to initiate an infection, it is difficult to distinguish between the response to bacterial toxins, such as endotoxin, versus the response to the bacteria themselves. The use of sterile mucin and hemoglobin broth to lower the bacterial inoculum required for a LD50 was first described in the 1930s (16). It is thought that the mucin impedes the host defenses from reaching the bacteria and that the hemoglobin provides an easy supply of iron to support bacterial growth. This model is characterized by robust in vivo multiplication of bacteria and is thought to mimic accurately a true i.p. infection (15).
Our initial hypothesis was that burn-injured mice would succumb more readily than sham-injured mice to the E. coli peritonitis, in concert with our findings with CLP-induced sepsis. Surprisingly, we demonstrate herein that burn-injured mice develop a much higher resistance to E. coli infection than do sham mice and that the degree of resistance to infection is a function of the time after injury. Resistance to i.p. E. coli infection was higher in burn-injured versus sham mice as early as 1 day after injury. This resistance peaked by 7 days and persisted for at least 14 days postinjury. These E. coli infection experiments also show that burn-injured mice are better able to contain the infection in the peritoneum and do not become bacteremic. Furthermore, we demonstrate that a potential mechanism for the enhanced resistance to infection following injury is an increase in the number of activated peritoneal neutrophils with heightened NADPH oxidase and bacteriocidal activity. Although these findings contradict the widely held idea that injury predisposes the host to opportunistic infections, we demonstrate in a mouse model that the enhanced responsiveness of the innate immune system that occurs after injury can in some instances help the host combat an infection caused by a common pathogen.
Materials and Methods
Male C57BL/6J mice were obtained from The Jackson Laboratory. Mice were maintained in an accredited virus-Ab-free animal facility in accordance with the guidelines of the National Institutes of Health and the Harvard Medical School Standing Committee on Animal Research. The mice were acclimated for at least 1 wk before being used in experiments at 6–9 wk of age.
E. coli strain no. 502501 (originally isolated in 1990 from an abscess in the Brigham and Women’s Hospital clinical lab) was provided by Dr. Andrew Onderdonk. This E. coli strain is ampicillin, cefazolin, mezlocillin, tetracycline, and trimethoprim resistant and has been used in other rodent infection studies (17). Bacteria were grown overnight at 37° C in brain heart infusion broth (Remel). Aliquots of the E. coli (1 ml) were flash frozen in liquid nitrogen and then stored in 1-ml aliquots at −70°C.
Mouse injury model
The mouse thermal injury protocol, approved by the Harvard Medical School Standing Committee on Animal Research, was performed as described previously (18). Mice were anesthetized by i.p. injection of ketamine (125 mg/kg) with xylazine (6.8 mg/kg). The dorsal fur was shaved, and the animal was placed in an insulated plastic mold to expose 25% total body surface area. This part of the dorsum was then immersed in 90°C water for 9 s to induce a well-demarcated, full-thickness, anesthetic burn injury. The exposed dorsum of sham animals was immersed in isothermic (24°C) water. All animals were resuscitated with an i.p. injection of 1 ml of 0.9% pyrogen-free saline. The mortality from burn injury was <5%.
Peritoneal E. coli infection model
Frozen aliquots of E. coli at ∼109 CFU of bacteria per aliquot were washed in Dulbecco’s PBS (DPBS) two times by centrifugation at 800 × g. A stock suspension of mucin (type II) (15%, w/v) (Sigma-Aldrich) and hemoglobin (4%, w/v) (Sigma-Aldrich) in PBS was prepared and autoclaved. Bacteria in a range of CFU were added to the mucin/hemoglobin stock to obtain a 6% mucin, 1.6% hemoglobin solution. Mice were then given 0.5 ml of this solution by i.p. injection.
Measurement of LD50
To determine the LD50, sham and burn mice 7 days after injury were injected i.p. with 0.5 ml of mucin/hemoglobin solution with a range of E. coli from 102 to 105 CFU. The infected mice were then monitored for survival for 7 days. Seven days was chosen as the endpoint for these studies because no deaths occurred after 5 days.
Sham and burn-injured mice at various time points after injury (ranging from 24 h to 14 days) were injected with 103 or 104 CFU of E. coli in the mucin/hemoglobin solution. The mice were then followed for survival for 7 days.
Peritoneal and blood cultures
Sham and burn-injured mice 7 days after injury were injected i.p. with 103 CFU of E. coli in mucin/hemoglobin solution. Twelve hours later, mice were sacrificed by CO2 asphyxiation, and whole blood samples were obtained by cardiac puncture using 1-ml syringes containing 50 μl of 169 mM EDTA. Peritoneal samples were obtained by peritoneal washings with 10 ml of warm, sterile Ca/Mg-free PBS containing 1 mM EDTA. The peritoneal and blood samples were then plated on blood agar plates in serial 1:10 dilutions. After overnight incubation at 37°C, CFU of E. coli were counted.
Peripheral blood and peritoneal cell counts
Sham and burn-injured mice 7 days after injury were injected i.p. with 104 CFU of E. coli in PBS. Immediately after infection and 4 h later, mice were sacrificed by CO2 asphyxiation. Whole blood samples were obtained by cardiac puncture using 1-ml syringes containing EDTA and transferred to sterile 2.0-ml polypropylene microcentrifuge tubes containing an additional 0.05 ml of EDTA. Peritoneal washings were performed by injecting 10 ml of warm Ca/Mg-free PBS with 1 mM EDTA into the peritoneal cavity under direct vision to avoid any bowel perforation and contamination. The mice were then agitated briefly to mix peritoneal washing fluid, and 5 ml of fluid was withdrawn. The blood was analyzed for leukocyte numbers and subsets using the Hemavet 850 instrument (Drew Scientific). Peritoneal cell counts were performed using a hemocytometer, and cell subsets were measured by flow cytometry. Briefly, cells were plated at 1 × 104 cells/well into a round-bottom 96-well plate and incubated with FcBlock reagent (BD Pharmingen) to prevent nonspecific binding of Abs. PE-labeled Abs specific for various cell-surface markers (F4/80, Gr-1, CD11b; BD Pharmingen) were added for 30 min. Cells were then washed in PBS with 1% BSA and 0.1% sodium azide and fixed in DPBS containing 0.3% paraformaldehyde. Flow cytometry was performed using a FACSCalibur instrument (BD Biosciences), and the results were analyzed using the accompanying CellQuest Pro software.
Measurement of peritoneal neutrophil activation status and phagocytic/binding function of peritoneal neutrophils and macrophages
Sham and day 7 burn-injured mice were injected i.p. with 1 × 104 CFU of E. coli/mouse in PBS, and peritoneal washings were obtained, as described above, immediately after injection and 4 h later. Peritoneal neutrophil activation status was measured by flow cytometry using PE-labeled anti-Gr-1 Ab to identify neutrophils, and APC-labeled anti-CD11b Ab, a component of the Mac-1 complex (CD11b/CD18), was used to detect neutrophil activation. Neutrophil NADPH oxidase activity was determined using the flow cytometric-based dihydrorhodamine 123 (DHR123) assay as described by Walrand et al. (19). In brief, peritoneal lavage cells (5 × 105 cells/well of a 96-well plate) were incubated for 25 min at 37°C in the presence or absence of 4 μg/ml DHR123. The reaction was stopped by centrifugation, followed by adding ice-cold PBS containing 0.3% paraformaldehyde to the cells. After a 10-min fixation, the cells were washed and surface-stained with APC-labeled anti-Gr-1 Ab to detect neutrophils or with anti-F4/80 Ab to detect macrophages. The levels of fluorescent DHR123 were detected by flow cytometry as green fluorescence in the fluorescence 1 channel on gated Gr-1+ and F4/80+ cells. FACS analysis was performed using the FACSCalibur instrument, and the results were analyzed using the CellQuest Pro software.
The adherence and phagocytic uptake of FITC-labeled, heat-killed E. coli (100°C for 1 h) were determined by flow cytometry as previously described by O’Brien et al. (20). In brief, heat-killed E. coli were thawed and incubated in 0.1 M carbonate buffer (pH 9.5) in the presence 0.1% FITC (Sigma-Aldrich) for 60 min at 37°C. The E. coli were then washed in PBS and stored at 4°C. Before use, the FITC-labeled E. coli were recovered by centrifugation at 12,000 rpm and incubated for 10 min in heat-inactivated mouse serum for opsonization. Peritoneal washout cells were harvested and washed, then incubated at 37°C for 20 min with FITC-labeled E. coli. After incubation, the peritoneal cells were washed with PBS to remove unbound E. coli and fixed in PBS containing 0.3% paraformaldehyde. Cells were then stained with PE-labeled Abs specific for macrophages (F4/80) or neutrophils (Gr-1). Flow cytometry was used to judge the percentage of neutrophils and macrophages that engulfed FITC-labeled E. coli.
Peritoneal cell E. coli killing assay
Peritoneal washout cells were harvested and resuspended in degassed DPBS containing 0.5% BSA and 2 mM EDTA. Neutrophil-enriched peritoneal cells were prepared by depleting peritoneal cells of F4/80+ cells using a customized MACS (Miltenyi Biotec) approach. F4/80+ cells (macrophages) were tagged with PE-labeled anti-F4/80 Ab and then removed from the cell suspension using anti-PE magnetic beads. This process depleted >98% of the F4/80+ cells from peritoneal cell suspensions. To measure bactericidal activity, 5 × 105 neutrophils or unseparated peritoneal cells were mixed with 2 × 103 CFU of E. coli in RPMI 1640 and incubated for 2 h at 37°C, 5% CO2. End-point E. coli CFU were determined by serial dilution of cell/bacteria mixtures in sterile water and drop-plating the dilutions onto tryptic soy agar plates. After overnight incubation at 37°C, colonies were counted to calculate CFU of E. coli.
Measurement of peritoneal and serum cytokines
At 7 days after injury, sham and burn-injured mice were given 107 CFU of E. coli by i.p injection. Two hours after infection, serum and peritoneal lavage samples were obtained as described above. Samples were analyzed for cytokine expression using cytometric bead arrays (BD Biosciences) according to the manufacturer’s instructions.
Results are presented as mean ± SEM. Statistical analysis was performed using the GraphPad Prism 4 software (GraphPad Software). Unpaired Student’s two-tailed t tests were used to generate p values. Survival statistics were determined by the log-rank test also performed with the GraphPad Prism 4 software (p < 0.05 was considered significant).
Burn injury markedly increases the resistance to an i.p. E. coli infection
To determine the number of E. coli required for a LD50 in uninjured C57BL/6 mice, we injected mice i.p. with a wide range of E. coli CFU mixed in the mucin/hemoglobin broth. We used the mucin/hemoglobin broth infection model for these mortality studies because it is well established that mice are relatively resistant to E. coli infection unless the bacteria are mixed with an adjuvant (e.g., mucin/hemoglobin or sterile cecal contents) that provides the initial nutrients and protection required to establish a low challenge dose infection in mice (15, 21). Giving mice the high numbers of bacteria needed to produce an LD50 without adjuvant (>109 CFU of E. coli) would convert this experimental model from an infection model to an acute inflammation and shock model. As shown in Fig. 1,A, the LD50 for male C57BL/6 mice was ∼103 CFU/mouse. Mortality occurred between 1 and 3 days after infectious challenge, with most animals dying within days 1 and 2 postinfection (Fig. 1,B). When we initially challenged mice with E. coli at 7 days after burn injury, we had expected that they would die more readily than sham mice. To our surprise, all but one burn-injured mouse (n = 14, three separate experiments) survived the 103 CFU/mouse challenge dose. Therefore, we injected an extended range of E. coli CFU to determine the LD50 dose for a burn-injured mouse. As shown in Fig. 1 A, the LD50 for day 7 burn-injured mice was between 50 × 103 and 100 × 103 CFU/mouse. These results indicate that burn-injured mice become 50–100 times more resistant to i.p. E. coli challenge than do sham mice at 7 days postinjury.
Resistance to E. coli infection peaks 7 days after injury and persists for at least 14 days
To study whether the observed increase in resistance to E. coli infection was a function of time after injury, we infected mice at days 1, 3, 7, 10, and 14 after burn or sham injury with 103 or 104 E. coli CFU/mouse. As shown in Fig. 2, challenging sham mice with 103 CFU caused 50% mortality, and 104 CFU caused 100% mortality. In contrast, even at 24 h after injury, the burn-injured mice displayed an increased resistance to E. coli infection. One day after injury, burn-injured mice showed a 66% survival when given the 103 CFU dose and a 33% survival with the 104 CFU dose. By day 3 after burn injury, all mice survived the 103 CFU dose and 66% of the mice survived the 104 CFU dose. From day 7 to 14 days after injury, there was a 100% survival with both doses (n = 6 in each dose and time group). These findings suggested to us that the increased resistance to infection in burn-injured mice was not a transient event and that it develops as early as 1 day after injury. Its time course appears similar to that of the innate immune system hyperactivity we have previously reported (5, 22).
Injured mice demonstrate rapid clearance and control of peritoneal E. coli
We hypothesized that the main cause of mortality in our mice was disseminated sepsis and that the burn-injured mice were more resistant because they were better able to contain the i.p. infection. To test this hypothesis, we cultured peritoneal fluid and blood from day 7 sham and burn-injured mice 12 hours after i.p. challenge with 103 CFU of E. coli. Fig. 3 confirms that the sham mice were less able to control the i.p. growth of E. coli with an average CFU count of 107 compared with 104 in the burn-injured mice. Additionally, none of the burn-injured mice became bacteremic, whereas 75% of the sham mice showed significant E. coli bacteremia, suggesting that the cause of death in E. coli-infected mice was likely disseminated sepsis.
Injury effects on blood and peritoneal cell populations
After establishing that burn-injured mice are more resistant to E. coli infection, we focused on examining how sham and burn-injured mice might differ in their responses to E. coli. Because the mucin/hemoglobin broth adjuvant interfered with our cellular assays and flow cytometry stains, we shifted to using live E. coli challenge in the absence of adjuvant for this portion of the study. To begin, we measured both peripheral blood and peritoneal leukocyte counts and subsets. We observed that burn-injured mice had significantly higher numbers of circulating peripheral blood leukocytes both at baseline and 4 h after E. coli challenge compared with sham mice (Fig. 4). At baseline, they had twice the number of circulating neutrophils, but this difference no longer existed 4 h after infection. Additionally, burn-injured mice had higher levels of circulating monocytes.
In the peritoneum, at baseline, both sham and burn-injured mice had similar total numbers of leukocytes (Fig. 5). However, 4 h after infection, burn-injured mice showed an almost two-fold increase in total leukocytes. There was no significant difference in F4/80+ macrophages numbers between burn-injured and sham mice at baseline, but there was a detectable difference at 4 h postinfection, although the overall percentage decreased in both groups. Moreover, at baseline there was a small, but significant increase in the percentage of peritoneal neutrophils in the burn-injured as compared with sham mice. This became much more pronounced 4 h after infection with both a greater percentage and absolute number of peritoneal neutrophils in burn-injured mice. These results indicate that injured mice had higher numbers of circulating neutrophils and monocytes at the time of E. coli infection and that the number of leukocytes recruited into the peritoneum of E. coli-challenged, burn-injured mice was much higher than in sham mice after infection.
Injury causes an increase in activated peritoneal neutrophils
The increased number of peritoneal neutrophils found in burn-injured as compared with sham mice suggested to us a possible mechanism responsible for the heightened control of peritoneal infection in burn-injured mice. Therefore, we examined the activation and functional status of the peritoneal neutrophils in day 7 burn-injured and sham mice. We used flow cytometry to measure cell-surface CD11b, a component of the Mac-1 complex, to judge neutrophil activation (23, 24). In burn-injured mice, nearly all the peritoneal neutrophils expressed CD11b at baseline and had a higher expression density of this activation marker compared with shams (Fig. 6). In comparison, only 25% percent of the peritoneal neutrophils from sham mice expressed CD11b. Nevertheless, 4 h after E. coli challenge, both sham and burn-injured mice showed similar evidence of neutrophil activation (≥90%).
We then measured neutrophil function at baseline by measuring the binding and uptake of FITC-labeled, heat-killed E. coli. Neutrophils from both sham and burn-injured mice showed similar levels of phagocytic uptake of E. coli, suggesting no difference in the capacity of neutrophils to engulf E. coli (Fig. 6,C). Because burn-injured mice have a higher percentage of activated resident neutrophils as judged by Mac-1 expression, we next asked whether peritoneal neutrophils might display enhanced antimicrobial activity as compared with sham neutrophils. To address this possibility, we first measured the level of NADPH oxidase activity in peritoneal neutrophils from burn-injured as compared with sham mice using a DHR123 assay. DHR123 becomes fluorescent when it reacts with superoxide anion. As shown in Fig. 7,A, we observed a marked difference in the level of DHR123-dependent fluorescence in neutrophils prepared from day 7 burn-injured as compared with sham mice. This increased NADPH oxidase activity seen in neutrophils of burn-injured but not sham mice suggests that neutrophils of burn-injured mice may kill E. coli more effectively once they are engulfed. We next compared the ability of neutrophil-enriched peritoneal cells from burn-injured and sham mice to kill E. coli in an in vitro assay. The results of these experiments indicate that neutrophils from burn-injured mice have significantly greater (two times) ability to kill E. coli than do neutrophils from sham mice (Fig. 7 B). Thus, the increase in the number of activated peritoneal neutrophils with enhanced antimicrobial activity found in burn-injured (but not sham-injured) mice appears to explain, at least in part, the enhanced resistance to E. coli challenge seen after burn injury.
Injured mice produce higher levels of inflammatory cytokines after E. coli challenge
In addition to examining the differences in cell subsets and neutrophil phenotype, we measured cytokine levels in both serum and peritoneal fluid in day 7 sham and burn-injured mice after infection with E. coli. For these experiments, we injected with E. coli in PBS rather than in mucin/hemoglobin broth because the mucin/hemoglobin particles interfered with our flow cytometry-based cytokine bead array assay. Accordingly, we used a larger inoculum (107 CFU) to achieve detectable levels of the cytokines. As shown in Fig. 8, burn-injured mice had significantly higher levels of the proinflammatory cytokines TNF-α and IL-6 in both peritoneal lavage fluid and serum compared with sham mice. Additionally, burn-injured mice had higher levels of the chemokine MCP-1 in the peritoneal fluid. These findings support the hypothesis that burn-injured mice show enhanced innate immune responsiveness to E. coli. This augmented innate reactivity to E. coli challenge may be responsible for the increased peritoneal influx of neutrophils and the more effective clearance of E. coli in burn-injured as compared with sham mice.
When we began these experiments, we anticipated that burn-injured mice would display a significantly reduced ability to control i.p. infection following E. coli inoculation. However, we were also aware that the heightened innate immune system reactivity that occurs after burn injury might boost antimicrobial defense mechanisms in vivo. These opposing assumptions were based on clinical observations and on many years of experience by us as well as others studying the influence of injury on immune function in animal models (22). Much of this prior research supported the conclusion that major injury led to diminished resistance to infection due to the development of the compensatory anti-inflammatory response syndrome noted clinically in critically ill patients and also in relevant rodent models of serious injury (25, 26, 27). We had also come to accept the idea that severe injury induces an initial SIRS that likely represents an overshoot of an evolutionarily conserved innate immune reaction to tissue injury. SIRS and compensatory anti-inflammatory response syndrome responses were thought to be salutary when confined to an area of a minor injury but to be inimical to host survival when expressed systemically. These systemic manifestations of the mammalian injury response purportedly included injurious inflammation on the one hand and compensatory loss of protective adaptive immunity on the other hand.
The specific injury-induced changes in immunity that might negatively impact the host’s ability to combat microbial infections include a reduction in Th1-type immune responses, which involve 1) suppressed delayed- or contact-type hypersensitivity responses, 2) reduced IFN-γ production capacity, 3) lowered Th1-type Ab formation, and 4) suppressed IL-12 production by dendritic cells and macrophages (28, 29, 30, 31, 32, 33, 34). In contrast, changes in innate immune function that have the potential to enhance antimicrobial immune function include 1) increased TLR2 and TLR4 responsiveness by macrophages and dendritic cells, 2) heightened neutrophil activation, and 3) increased production of proinflammatory cytokines and chemokines by activated innate immune cells (5, 35, 36, 37, 38, 39, 40). Although these changes in immune system function have been described in both injured patients and in animal models, controlled experiments to test changes in microbial defense after injury were of necessity largely confined to animal models. In these model systems, the most common septic challenge used to demonstrate resistance to infection was CLP, which was thought to be clinically relevant because it induced a localized peritonitis with mixed flora from the gut. Using CLP to test postinjury immune function, we and other groups have indeed shown that burn-injured mice readily succumb to CLP-induced sepsis, while sham-injured mice are better able to resist CLP challenge (18, 33, 41, 42).
In marked contrast, the results of the present studies using a single pathogen infection model clearly demonstrated that during the first week after burn injury, mice became increasingly more able to resist an i.p. inoculum of pathogenic E. coli than their sham counterparts. In fact, the resistance of burn-injured mice to E. coli challenge at 7 and 14 days after injury was 10–50 times greater than sham controls. Moreover, in almost all instances, burn injury prevented the development of bacteremia, suggesting that burn-injured mice have the ability to control E. coli infections locally. This was further supported by the observation that burn-injured mice were more able than sham mice to reduce the number of viable E. coli in the peritoneum. Thus, in a single pathogen model, injured mice show a greater capacity to combat a Gram-negative bacterial peritonitis.
In contrast, other groups have reported that challenging burn-injured rats at the time of injury leads to higher mortality. One of these studies showed that burn-injured rats had poorer day 1 and 2 postinjury survival rates than did sham rats when challenged intraperitoneally at the time of burn injury with fecal pellets containing Enterococcus faecalis (43). The authors performed this study to test the influence of bacterial translocation from the gut after burn injury on changes in gut permeability and not on microbial mortality, and thus only one time point was examined. Because the bacteria used and the timing of their bacterial challenge were markedly different from what we report here, it is difficult to determine whether the work of these authors truly is in contrast to our findings. However, we did perform a time-course study and showed that the increased resistance to E. coli infection was less dramatic at 1 day after injury than at later time points, suggesting that antimicrobial resistance is not as enhanced early after burn injury as it is later. Thus, it is possible that antimicrobial resistance is normal or less than normal when infection and injury occur simultaneously.
The findings from another study suggest that mice given Pseudomonas aeruginosa in the lungs after burn injury develop reduced resistance to secondary infections, but not to P. aeruginosa (44). Because these animals did not die of infections caused by the challenge organism, P. aerunginosa, this work neither supports nor contradicts the findings reported in this paper. Another single-pathogen challenge model used to investigate immune function in burn-injured mice is a wound infection model using P. aeruginosa as a challenge pathogen. The difficulty with using this model is that it is not possible to accurately compare sham and burn-injured mice because sham mice do not have a wound to infect. When challenged at the time of burn injury, burn-injured mice are susceptible to infection and will die with a moderate challenge dose of P. aeruginosa (103 CFU). However, mice developed resistance to burn-wound infection as early as 1 day after burn injury (45). The observations made using this wound infection model fully support our findings, which demonstrate that burn-injured mice develop increased resistance to bacterial infection when compared with sham mice.
The difference between CLP-induced mortality and E. coli-induced mortality was surprising. Possible explanations for this superior ability of burn-injured mice to contain and kill i.p. bacteria include the observation that at the time of bacterial inoculation the burn-injured mice had significantly higher total circulating leukocytes with increased percentages of neutrophils and monocytes, compared with sham controls. Furthermore, we observed that before bacterial challenge there were significantly more neutrophils in the peritoneal cavity in burn-injured as opposed to sham mice and that this was even more marked at 4 h after inoculation. Although it is logical to assume that the availability of more innate effector cells before and after bacteria challenge might be directly responsible for increased resistance to E. coli infection, we wanted to determine which particular cell type might be mediating this increase in antimicrobial defense. Thus, we tested the activation status of neutrophils and macrophages present in the peritoneal cavity of sham and burn-injured mice. The peritoneal neutrophils in sham and burn-injured mice showed clear differences. The resident peritoneal neutrophils in burn-injured mice were highly activated, as indicated by the expression of CD11b and increased NADPH oxidase activation potential. Interestingly, this enhancement of innate immune cell activation correlates well with prior observations showing that burn injury leads to augmented cellular reactivity to TLR stimuli (5). Because TLRs have been shown to be central to antimicrobial defenses, the innate cellular response to infection in the injured host would be expected to be more robust and rapid.
In this study, we also show that burn-injured mice have enhanced inflammatory cytokine reactivity and demonstrate early recruitment of higher numbers of activated neutrophils into the peritoneum. Moreover, these cells demonstrate a heightened ability to kill E. coli presumably via a mechanism involving the well-described NADPH oxidase-dependent killing pathway. These observations support the interpretation that higher numbers of activated neutrophils present in the peritoneal cavity with heightened ability to kill E. coli play a significant role in the rapid and effective clearance of bacteria to prevent the development of bacteremia and death.
The contrasting results of high CLP mortality but enhanced resistance to peritoneal E. coli infection after burn injury require further discussion. First, CLP represents the standard model for sepsis because it induces a peritoneal infection with mixed flora from the gut, which includes aerobic and anaerobic organisms. The E. coli inoculum, in contrast, induces a localized peritonitis with a single aerobic organism. Second, CLP adds the effects of a necrotic cecum to those of the peritoneal infection. Third, the CLP procedure is invariably performed under general anesthesia which, of itself, can be expected to supply a “second hit” in the animals after burn injury. The injury itself is obviously the first hit and the surgical procedure (CLP) 7 days later provides a second hit, which in clinical observations and animal studies often induces a heightened inflammatory response. Unfortunately, the complexity of the CLP model makes it impossible to determine which of these variables might explain the differences we observed between these two sepsis models. In contrast, the E. coli infection model allowed us to more precisely follow the host response to a common pathogen without contending with the other potential confounding variables associated with the CLP procedure.
The present results also appear to be at variance with recently reported results from our laboratory describing markedly increased mortality to an otherwise sublethal dose of E. coli endotoxin in burn-injured versus sham mice at 7 days, but not 1 day postinjury (6). However, because each E. coli cell contains a total of 10 femtograms of endotoxin, an inoculum of E. coli many times larger than the LD50 dose used in the present experiments would be necessary to provide sufficient endotoxin to mimic the 10 mg/kg dose of endotoxin used in the previously reported experiments. Thus, we think that the model used in this study represents an infection model rather than an endotoxin challenge model because the LD50 E. coli inoculum dose of 105 bacteria in burn-injured mice would result in a maximum endotoxin dose of ∼40 ng/kg.
In summary, the major contribution of the present studies is the demonstration that the immune perturbations caused by a major burn injury during the first 1–2 wk postinjury can be protective against infection with a common organism. The results support the concept that the proinflammatory state occurring after major injury, which includes augmented inflammatory cytokine production and rapid recruitment of neutrophils to the site of infection, can be beneficial to host survival. However, we fully acknowledge that critically injured patients are indeed predisposed to develop opportunistic infections. We also acknowledge that this infection model may not be directly relevant to what occurs to patients in intensive care units. Injured patients are likely exposed to multiple insults such as surgery or other related invasive procedures. This would in turn place the injured patient at higher risk of developing opportunistic infections. Moreover, critically injured patients are likely exposed to mixed pathogens in other organ systems that might be more damaging to the host, such as the lungs. Thus, the response to infection might be different in other organs and with other common pathogens. For these reasons, the influence of burn injury on the ability to withstand infection with other common pathogenic organisms (e.g., Staphylococcus sp., P. aeruginosa, or Streptococcus pneumoniae) and with inoculation at other sites (e.g., the lungs) will be the focus of future studies. Additionally, the adaptive immune system may play a significant role in determining the outcome of postinjury infection.
We, as well as other investigators, have reported that injury-induced perturbations of the adaptive immune system, especially loss of type 1 immune reactivity, are associated with increased mortality to CLP. A causal relationship has also been suggested by the fact that treatments that restore Th1 immune function (e.g., administration of the Th1-inducing cytokine IL-12) restore CLP resistance in injured animals (33). However, in the present experiments, loss of Th1 immune reactivity, invariably found in the burn-injured mouse model by 7 days postinjury, had no apparent negative effect on the ability of the injured animals to clear E. coli peritonitis. An increased understanding of the balance between innate and adaptive immune system function after injury will likely help guide the development of therapies to decrease infectious complications and improve the survival of critically injured patients.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grant funding from the National Institutes of Health (GM35633 and GM57664) and by the Julian and Eunice Cohen and Brook Family Funds for Surgical Research.
Abbreviations used in this paper: SIRS, systemic inflammatory response syndrome; CLP, cecal ligation and puncture; DHR123, dihydrorhodamine 123.