Sepsis continues to be a leading cause of in-hospital mortality despite carefully coordinated treatments, resulting in over 200,000 deaths per year in the United States (1). While the mortality rate of severe sepsis has decreased, the number of cases has continued to increase (2). Sepsis occurs when an infection induces a systemic inflammatory response syndrome (SIRS). This may lead to septic shock, characterized by extreme hypotension, or severe sepsis, characterized by multiple organ dysfunction. Current treatments include crystalloid fluid resuscitation, broad spectrum antibiotics, and vasopressors.
The innate immune response exerts both beneficial and harmful effects during sepsis. Neutrophils play a crucial role in clearing pathogens, but exert collateral damage on host tissues. Macrophages produce cytokines and chemokines that promote inflammation to help contain pathogens at the local site of infection. At the systemic level, however, excessive inflammation causes widespread vasodilation, leading to hypotensive shock, coagulopathy and hypo- and hyper-perfusion of vital organs. Additionally, SIRS is accompanied by a compensatory anti-inflammatory response syndrome (CARS), which leaves the host more vulnerable to primary and secondary infections (3). Even if the patient survives the early phase of sepsis, a persistent inflammation, immunosuppression, and catabolism syndrome (PICS) may ensue, where the patient fails to return to homeostasis over the long term (4).
While the inflammatory response may seem an obvious therapeutic target in sepsis, careful consideration must be given to the complex immune etiology of this disease. Broad immune suppression with corticosteroids only benefits patients with severe sepsis who fail to respond to fluid resuscitation and vasopressors. On the other end of the spectrum, mono-therapies that target individual cytokines failed to exert a benefit in clinical trials. However, therapies that modulate specific branches of the immune response could be advantageous to septic patients. Further clarification of which innate immune pathways are helpful versus harmful during sepsis will aid in the development of novel therapies that preserve host defense, while limiting the detrimental effects of SIRS, CARS, and PICS.
CLP is the most widely used animal model of sepsis (5, 6). In this model, the cecum is ligated and punctured, allowing commensal bacteria to escape into the peritoneal cavity, leading to polymicrobial peritonitis, bacteremia, and SIRS. The severity of CLP sepsis can be modified by changing the portion of the cecum that is ligated, size of the needle used, and number of punctures (5). Additionally, mice can be treated with therapeutic reagents, such as Lactated Ringer Solution and broad spectrum antibiotics to better recapitulate the standard of care for clinical sepsis and attenuate mouse mortality (7). The related colon ascendens stent peritonitis (CASP) model (8) and cecal slurry injection (CS) model (9) induce sepsis in a similar fashion.
Pattern recognition receptors (PRRs), such as Toll-like receptor (TLRs), play a vital role in early host defense. These receptors recognize pathogen-associated molecular patterns (PAMPs) and danger-associated molecular patterns (DAMPs) released from infected and injured tissues. TLR activation initiates signal transduction cascades through Myeloid Differentiation Factor 88 (MyD88) (downstream of TLRs 1–2 + 4–9) and TRIF (downstream of TLRs 3–4), culminating in activation of the transcription factors nuclear factor-κB, IRF3, and IRF7. These transcription factors induce inflammatory mediators including cytokines and type I interferons (10). In addition to TLRs, additional PRRs recognize nucleic acids in the cytosol. When bacterial or mammalian DNA gains access to the cytosol, it is detected by Cyclic GMP-AMP synthase (cGAS), which synthesizes the cyclic dinucleotide cGAMP (11). cGAMP and similar cyclic dinucleotides produced as bacterial second messengers bind to the STING adaptor to activate IRF3, inducing the production of type I interferons and cytokines (12, 13).
In the scenario of CLP-sepsis, commensal bacteria are released from the cecum and contain multiple PAMPs. Additionally, DAMPs, such as high-mobility group protein 1 (HMGB1) and cell-free DNA, are released into the bloodstream due to the surgical injury, infection, and organ damage (14–17). These activate multiple pathogen sensors that likely synergize to induce the dysregulated immune response during sepsis. Prior reports have generated conflicting results regarding the contribution of TLR pathways to sepsis pathogenesis. Several studies utilized mice genetically deficient in the MyD88 adaptor protein (18–20) or TRIF adaptor protein (20, 21) and obtained varying results using the CLP, CASP, and CS models of sepsis. These discrepant results are most likely due to differences in the disease model employed to recapitulate sepsis, outcomes measured (i.e., animal mortality, disease score, levels of inflammatory mediators) and animal attributes (i.e., age, gender, genetic modifications, and background strain).
We previously showed that IRF3-KO mice are protected from CLP-induced sepsis compared with wild-type (WT) mice, exhibiting improved survival, attenuated disease score and hypothermia, reduced bacterial load, and decreased levels of serum cytokines and creatine kinase (CK) (22). Additionally, we found that IRF3 is activated in macrophages following treatment with live and sonicated bacteria derived from the mouse cecum (22). As described above, IRF3 can be activated downstream of the TRIF-mediated TLR pathways, as well as the STING-mediated cytosolic DNA-sensing pathway. However, it is unclear which pathway activates IRF3 following CLP, especially given the conflicting evidence from prior reports investigating TRIF.
In this study, we investigated the role of STING and TRIF in sepsis pathogenesis. We used a severe version of the CLP model to induce sepsis in a similar manner to our prior report (22). Additionally, we performed a moderate version of the CLP model, administering Lactated Ringer Solution and imipenem/cilastatin antibiotics to test the influence of these pathways in the context of clinical treatments. Our data demonstrate that both the STING pathway and the TRIF pathway play a detrimental role in the severe CLP model employed in this study, while only the TRIF pathway contributes to the moderate CLP model of mouse sepsis. We detected bacteria in the peritoneum of mice that underwent severe and moderate CLP surgeries. In contrast, there was a trend toward increased cell-free DNA in mice that underwent severe CLP but not moderate CLP, relative to sham controls. IRF3 was activated in cultured macrophages treated with commensal bacteria or mouse genomic DNA/Lipofectamine 2000. IRF3 activation in response to bacteria required TRIF, whereas activation in response to genomic DNA required STING. Our results highlight how differences in the disease model utilized to study sepsis may affect the PAMPs and DAMPs that are released and consequently alter the contribution of innate immune pathways to disease pathogenesis.
MATERIALS AND METHODS
WT (C57BL/6J) and TRIF-KO (LPS2) mice were purchased from The Jackson Laboratory (Bar Harbor, Maine). STING-KO (Goldenticket) mice were a gift from Dr Russell Vance (University of California, Berkeley, Calif). STING-KO and TRIF-KO mice were interbred to generate STING/TRIF-DKO mice. These strains were bred in the TTUHSC El Paso Laboratory Animal Resources Center and Yale Animal Resources Center under specific pathogen-free conditions. Experimental animals included a mixture of males and females, 8 to 21 weeks of age (mice were matched for age and gender within each experiment). All experiments were approved by the Institutional Animal Care and Use Committee of the Texas Tech University Health Sciences Center or Yale University.
CLP was performed as previously described, with minor modifications (22, 23). For the severe version of CLP, a midline incision was made in the peritoneum and the cecum was exteriorized. Fifty percent of the cecum was ligated and pierced through and through with a 21G needle and small drop of cecal contents was extruded. The cecum was returned to the peritoneal cavity and the abdomen was closed in two layers. Mice were resuscitated with 1 mL warmed saline s.c. postoperatively and buprenorphine analgesia pre-op, at 4 to 9 h post-surgery and then at 12 h intervals, or buprenorphine SR (extended release) pre-op and at 48 h intervals. For the moderate version of CLP the following modifications were made. One centimeter of the cecum was ligated and a small drop of cecal contents was extruded. Mice were resuscitated with 1 mL warmed Lactated Ringers Solution s.c. postoperatively. Antibiotics (Primaxin; 25 mg/kg imipenem and 25 mg/kg cilastatin) were administered 1 to 6 h post-CLP and then once every 12 h for 5 d. We observed seasonal variations in the survival of WT mice in the moderate model during the course of the study; thus, we modified the timing of the first antibiotic dose (from 1 to 6 h post-surgery) to maintain a mortality rate of 30% to 50% in WT mice.
Serum and peritoneal lavage preparation
Blood was collected by retro-orbital bleed. Serum was prepared in serum separator tubes (Becton Dickinson, Franklin Lakes, NJ), according to manufacturer's instructions. For peritoneal lavage, mice were euthanized and 3 mL PBS was injected into the peritoneum. The abdomen was gently massaged for 2 min, and the fluid was recovered with a pipette through a small incision, subsequently peritoneal cells were separated from the liquid fraction via centrifugation.
Enzyme-Linked Immunosorbent Assay (ELISA), serum chemistry and CFU measurements
ELISA kits for IL-6, IL-12/23p40, and MCP-1 (BD Biosciences, San Jose, Calif) were used per the manufacturer's instructions. CK was measured with the CK reagent set (Pointe Scientific, Canton, Mich). Peritoneal lavage and blood samples were diluted in PBS, plated on BBL agar plates, and incubated for 18 h at 37°C. Bacterial colonies were then counted.
Quantification of cell-free DNA in the serum
DNA was purified from mouse serum using the QIAamp Blood DNA Mini Kit (Qiagen, Valencia, Calif), according to the manufacturer's instructions. To quantify the amount of mouse genomic DNA in the eluate, B-2microglobulin (B2 M) was amplified via qPCR, as described in a prior report (17). Briefly, 10 μL of iQ SYBR Green Supermix, 0.6 μL (600 ng) of B-2microglobulin (B2 M) forward and reverse primers: B2M-forward primer (B2 M exon4-F; 5′-CTTTTGGTAAAGCAAAGAGGCC-3′) and B2M-reverse primer (B2 M exon4-R; 5′-TTGGGGGTGAGAATTGCTAAG-3′), 6.8 μL of H20, and 2 μL of DNA were used per reaction. The reaction was amplified in a Realplex2 Mastercycler (Eppendorf, Hauppauge, NY), where the mixture was denatured at 95°C for 4 min, followed by 40 cycles of 95°C for 15 s, annealing at 60°C for 60 s, and extension at 72°C for 12 s. Results were recorded and analyzed using the Realplex software (Eppendorf, Hauppauge, NY). Fold-induction was determined using the ΔΔCT method relative to a standard of mouse genomic DNA (250 ng/mL; isolated from a mouse tail tip) and multiplied by the Qiagen input serum volume/eluate volume to determine the serum concentration.
Thioglycollate-elicted macrophages, preparation of commensal bacteria, and Western blot
Thioglycollate-elicted macrophages and commensal bacteria were prepared as previously described (22). Briefly, mice were injected with 1.5 mL 4% thioglycollate i.p. and peritoneal lavage was prepared 4 days later. Macrophages were enriched by plating 1–2e6 lavage cells on tissue culture plates for 60 min at 37°C, 5% CO2. The plate was washed to remove nonadherant cells. To prepare commensal bacteria, mouse cecal contents were inoculated into 10 mL Luria broth and cultured at 37°C on a shaking platform for 5 h. The suspension was filtered through 40-μm cell strainer to remove debris and centrifuged at 4000 rpm for 10 min. The bacterial pellet was then resuspended in 1 mL cell culture media and sonicated in 4 × 30-s bursts prior to addition to the macrophages at a 1/5 or 1/2.5 dilution. Macrophages were treated with 1 ug genomic DNA isolated from mouse tail clips complexed with Lipofectamine 2000 (per manufacturer's instructions). Cells were stimulated for 2 h and the Western blot was performed as previously described (22, 24), using antibodies against P-ser 396 IRF3, total IRF3 (both from Cell Signaling Technology, Danvers, Mass), and GAPDH (Santa Cruz Biotechnology, Santa Cruz, Calif). Western blot signal was detected with the chemiluminscent Amersham ECL Western blotting analysis system (GE Healthcare Life Sciences, Piscataway, NJ) followed by exposure using Syngene G Box (Syngene, Frederick, Md). Band density was quantified with GeneTools from Syngene.
Statistical support was provided by the Texas Tech Biostatistics and Epidemiology Consulting Lab. Survival rates were evaluated using Kaplan–Meier curves and compared using the log-rank test between groups. Normally distributed variables (surface temperature) were compared using repeated measures analysis of variance (ANOVA). Non-normally-distributed variables (serum and lavage cytokines, CK, bacterial counts, and peritoneal cell numbers) were log-transformed and then compared using either repeated measures ANOVA or an independent t test, depending on the number of comparative groups. Ordinal data (disease score) were compared using the Friedman test. Western blot results and serum DNA levels were evaluated by the Kruskal–Wallis test followed by a Dunn post hoc correction test. Each experiment was repeated at minimum in triplicate. The differences were considered to be statistically significant at P < 0.05.
STING-KO, TRIF-KO, and STING/TRIF-DKO mice are protected from severe CLP
We previously showed that IRF3-KO mice are protected from sepsis using a severe version of the CLP model (22). These data suggest that IRF3 plays a pathogenic role during the development of sepsis. We hypothesized that the cytosolic DNA-sensing pathway might activate IRF3 in the context of sepsis to induce this pathogenesis. Bacterial or mammalian DNA that gains access to the cytosol activates cGAS, which signals via the STING adaptor to activate IRF3. In order to test our hypothesis, we performed severe CLP surgeries on Goldenticket mice lacking functional STING protein (25) (henceforth referred to as STING-KO mice).
Following severe CLP, WT mice exhibited significant mortality: 85% died, with a median survival time (MST) of 43 h (Fig. 1A). In comparison, STING-KO mice had improved survival: only 66% of mice died and the MST was extended to 117 h (Fig. 1A). In WT mice, death was preceded by a rapid escalation in disease score (e.g., animal lethargy Fig. 1B) and hypothermia (Fig. 1C); however, these were attenuated in the STING-KO mice (Fig. 1, B and C). Control mice that underwent sham surgery did not exhibit mortality, lethargy, or hypothermia (Fig. 1, A–C).
In addition to the cytosolic DNA-sensing pathway, IRF3 is activated downstream of certain TLRs. TLR4 (activated by LPS) and TLR3 (activated by dsRNA) both signal via the adaptor TRIF to activate IRF3. In order to determine if TLR pathways induce IRF3-mediated sepsis pathogenesis, we performed severe CLP surgeries on the LPS2 strain lacking functional TRIF protein (26) (henceforth referred to as TRIF-KO mice).
As described above, WT mice displayed significant mortality: 88% died, with an MST of 64 h (Fig. 1D). However, TRIF-KO mice that underwent severe CLP had improved survival: only 63% of mice died and the MST was extended to 131 h (Fig. 1D). In WT mice, death was preceded by a rapid elevation in disease score (Fig. 1E) and hypothermia (Fig. 1F); however these were attenuated in the TRIF-KO mice (Fig. 1, E and F). Mice that underwent sham surgery did not show these signs of sepsis (Fig. 1, D–F).
To test the combined effects of the STING and TRIF pathways, we interbred these strains to generate STING/TRIF-DKO mice. Following severe CLP, 94% of the WT mice died with an MST of 47 h (Fig. 1G). In contrast, 61% of the STING/TRIF-DKO mice died with an MST of 86 h (Fig. 1G). Furthermore, disease score (Fig. 1H) and hypothermia (Fig. 1I) were attenuated in the STING/TRIF-DKO mice that underwent severe CLP surgery relative to WT counterparts. Control mice that underwent sham surgery did not exhibit signs of sepsis (Fig. 1, G–I). Collectively, these data suggest that both STING and TRIF contribute to disease pathogenesis in this severe model of CLP sepsis. STING/TRIF-DKO mice show a similar degree of protection to each single KO strain, suggesting that these two pathways may confer interrelated effects rather than additive effects.
STING-KO, TRIF-KO, and STING/TRIF-DKO mice have reduced levels of systemic inflammatory mediators following CLP
We next examined markers of systemic inflammation, indicative of SIRS. Following severe CLP, WT mice had substantially elevated levels of serum cytokines, including IL-6 (Fig. 2, A, D, and G) and IL-12/23p40 (Fig. 2, B, E, and H), and the chemokine MCP-1 (Fig. 2, C, F, and I). In comparison, there was a significant reduction in the levels of serum IL-6, IL-12/23p40, and MCP-1 in the STING-KO mice that underwent severe CLP (Fig. 2, A–C). Furthermore, there was a reduction in IL-6 and MCP-1 in the TRIF-KO mice (Fig. 2, D and F) and STING/TRIF-DKO mice (Fig. 2, G–I) that underwent severe CLP, whereas IL-12/23p40 levels remained similar to WT counterparts (Fig. 2, E and H). Mice that underwent sham surgeries exhibited low levels of inflammatory mediators throughout the experiment (Fig. 2, A–I). These data suggest that both STING and TRIF contribute to the systemic inflammatory response in this severe model of CLP sepsis.
STING-KO mice have reduced levels of CK following CLP
Organ damage is a hallmark of severe sepsis. When heart or skeletal muscles become damaged, CK is released into the serum. WT mice showed high levels of serum CK 19 to 23 h after severe CLP surgery (Fig. 3, A–C). The amount of serum CK was significantly lower in STING-KO mice that underwent severe CLP (Fig. 3A). However, TRIF-KO and STING/TRIF-DKO mice showed CK levels that were comparable to WT mice (Fig. 3, B and C). Mice that underwent sham surgeries exhibited low levels of CK (Fig. 3, A–C). These data suggest that the STING pathway contributes to muscle damage in this severe model of CLP sepsis.
STING-KO mice have a lower bacterial load following severe CLP
During CLP-sepsis, bacteria escape from the commensal-rich cecum into the normally sterile peritoneal cavity, causing peritonitis. If phagocytes are unable to clear these microbes, they subsequently translocate into the bloodstream, causing bacteremia.
We quantified bacterial load in the peritoneal lavage and blood of mice 19 h after severe CLP via aerobic culture on BBL agar plates. WT mice that underwent severe CLP had a high number of bacterial colony-forming units (CFU) in the lavage 19 h post-surgery (Fig. 4, A and C). In comparison, STING-KO mice had a lower number of bacterial CFU in the lavage (Fig. 4A). Likewise, TRIF-KO mice exhibited a trend toward reduced number of bacterial CFU in the lavage (Fig. 4C). Bacteria were also detected in the blood of some mice that underwent CLP; however, the levels were variable and we did not observe a trend among the genotypes (Fig. 4, B and D). In mice that underwent sham surgeries no bacteria were detected in the lavage or the blood (data not shown). These data suggest that STING contributes to an increased peritoneal bacterial load in this severe model of CLP sepsis.
STING-KO and TRIF-KO mice have lower levels of local inflammatory mediators following severe CLP
Macrophages reside in the peritoneum and play an important role in the early inflammatory response to CLP sepsis. Neutrophils and additional inflammatory macrophages rapidly infiltrate the peritoneum to help contain the infection. We sacrificed mice 19 h after severe CLP and obtained peritoneal lavage samples to examine the number of infiltrating cells and the local inflammatory response. In comparison to WT mice, STING-KO mice showed a trend toward fewer cells in the peritoneum at this time point (Fig. 5A); however, this difference was not statistically significant. We observed significantly reduced levels of local inflammatory mediators, including IL-6 (Fig. 5B) and MCP-1 (Fig. 5D) in the STING-KO mice that underwent CLP versus WT counterpart, as well as a trend toward a reduction in IL-12/23p40 (Fig. 5C).
In comparison to WT mice, TRIF-KO mice also displayed a trend toward a reduced number of peritoneal cells at the 19 h time point (Fig. 5E); however, the difference was not significant. We observed significantly reduced levels of local inflammatory mediators including IL-6 (Fig. 5F) and MCP-1 (Fig. 5H) in TRIF-KO mice versus WT mice that underwent severe CLP, as well as a trend toward a reduction in IL-12/23p40 (Fig. 5G). These data suggest that both STING and TRIF contribute to the local pro-inflammatory cytokine response in this severe model of CLP sepsis.
STING-KO and TRIF-KO response to moderate CLP sepsis
As part of their standard treatment regimen, patients with sepsis receive broad-spectrum antibiotics and crystalloid fluid resuscitation. These therapies have been successful in reducing patient mortality to ∼30% to 50% in different countries. We adapted the protocol described by Newcomb et al. (7), incorporating these treatments to recapitulate a moderate model of CLP sepsis that more closely resembles clinical sepsis (∼35% WT animal mortality). In this model, we ligated a smaller portion of the cecum (1 cm) and treated the mice with Lactated Ringers Solution and imipenem/cilastatin antibiotics (see Methods). In contrast to our results in the severe CLP model, STING-KO mice that underwent moderate CLP showed no significant difference in mortality relative to WT counterparts (39% vs. 35% Fig. 6A). STING-KO mice showed a similar disease score to WT mice following moderate CLP (Fig. 6B). In contrast to the severe CLP model, WT mice showed only a very subtle degree of hypothermia after moderate CLP, and STING-KO mice showed similar results (Fig. 6C).
TRIF-KO mice that underwent moderate CLP exhibited a trend toward reduced mortality relative to WT counterparts (21% vs. 37% Fig. 6D); however, this difference was not statistically significant. The disease score of TRIF-KO mice that underwent moderate CLP was significantly lower than WT counterparts (Fig. 6E). WT mice subject to the moderate model of CLP sepsis exhibited a subtle drop in temperature, which was significantly attenuated in TRIF-KO mice (Fig. 6F).
We also measured the production of pro-inflammatory mediators in WT, STING-KO, and TRIF-KO mice following moderate CLP surgeries. In STING-KO mice that underwent moderate CLP IL-12/23p40 was produced at slightly higher levels versus WT counterparts (Fig. 7B). There was a trend toward slightly higher levels of IL-6 in STING-KO mice that underwent moderate CLP versus WT counterparts at 18 h (Fig. 7A); however, this difference was not statistically significant, and MCP-1 levels were similar between STING-KO and WT counterparts (Fig. 7C). In TRIF-KO mice that underwent moderate CLP, there was a trend toward slightly lower levels of cytokines IL-6 at 18 h (Fig. 7D), IL-12/23p40 at 6 h (Fig. 7E), and MCP-1 at 6 h (Fig. 7F) relative to WT counterparts; however, these differences were not statistically significant. Hence, while STING and TRIF both contributed to sepsis pathogenesis in the severe model of CLP sepsis, these data suggest that STING does not play a significant role in this moderate model of CLP sepsis incorporating clinical therapies. Moreover, TRIF decreased the mouse disease score and elevated temperature but did not significantly improve mouse survival. Finally, serum cytokine levels showed only minor differences according to the mouse genotype in this moderate sepsis model.
Bacteria and cell-free genomic DNA are released during CLP sepsis
Both bacterial PAMPS and host DAMPs have been implicated in sepsis (8, 27, 28). Consistent with prior reports, we detected abundant bacteria in the peritoneum of mice that underwent both severe (22) and moderate CLP surgeries (29), although the levels were reduced by about two orders of magnitude in the moderate model incorporating antibiotics (Fig. 8A). In contrast, the peritoneum of mice that underwent sham surgeries was fairly sterile (one exudate produced two bacterial colonies that were likely lab contaminants; Fig. 8A). We found that serum levels of cell-free mouse genomic DNA appear to be slightly elevated after sham surgery, relative to naive mice (Fig. 8B); however, the difference was insignificant. Cell-free DNA was more substantially elevated in some of the mice that underwent severe CLP, while all of the mice that underwent moderate CLP exhibited similar cell-free DNA levels to sham controls (Fig. 8B).
Bacteria and genomic DNA activate IRF3 in cultured macrophages, via the TRIF and STING pathways respectively
In our prior report, we showed that IRF3 became phosphorylated (activated) in cells treated with bacteria derived from the mouse cecum. We reasoned that mouse genomic DNA may activate IRF3 in a similar manner. To test the effects of these ligands, we used an in vitro culture system where IRF3 activation can be readily detected. We prepared thioglycollate-elicited macrophages from WT mice and treated them with sonicated bacteria derived from the mouse cecum, or mouse genomic DNA complexed with Lipofectamine 2000 for 2 h. A Western blot showed evidence of phosphorylated IRF3 in cells treated with sonicated bacteria or DNA, but not in untreated cells (Fig. S1, https://links.lww.com/SHK/A490).
To determine the pathway of IRF3 activation, we prepared thioglycollate-elicited macrophages from WT, STING-KO, and TRIF-KO mice. These macrophages were treated with sonicated bacteria or genomic DNA complexed with Lipofectamine 2000 for 2 h, and then phosphorylation of IRF3 serine-396 was determined by Western blot. Cecal bacteria were able to induce IRF3 phosphorylation in WT and STING-KO macrophages, but this phosphorylation was reduced in TRIF-KO macrophages (Fig. 8C). Genomic DNA was able to induce IRF3 phosphorylation in WT and TRIF-KO macrophages, but this phosphorylation was reduced in STING-KO macrophages (Fig. 8D). The total amount of IRF3 remained similar in all samples, relative to GAPDH (Fig. 8, C and D). These data indicate that TRIF mediates IRF3 activation in macrophages treated with bacteria derived from the mouse cecum and STING mediates IRF3 activation in macrophages treated with genomic DNA. Hence, PAMPs and DAMPs that circulate during sepsis are sufficient to activate IRF3 via TRIF and STING.
Despite decades of research, we still lack an effective understanding of how we might modulate the innate immune response to benefit septic patients. Our prior study demonstrated that the IRF3 transcription factor contributes substantially to sepsis pathogenesis (22); however, it is unclear how this protein becomes activated in the context of polymicrobial infection. In this study, we investigated STING and TRIF, two innate immune adaptors that activate IRF3. No prior study has investigated the role of STING in sepsis pathogenesis, as this protein was only recently elucidated as a key adaptor in the cytosolic DNA-sensing pathway. Previous studies examining the role of TRIF during sepsis have obtained contradictory results.
We used a severe version of the CLP model to induce sepsis in a similar manner to our prior report (22). We found that both STING-KO (Goldenticket) mice (25) and TRIF-KO (LPS2) mice (30) were protected from sepsis in this severe CLP model, relative to WT counterparts. STING/TRIF-DKO mice showed a similar phenotype to each of the single KO strains. These data suggest that STING and TRIF both contribute to sepsis pathogenesis, but play interrelated rather than additive roles. Importantly, neither STING-KO, TRIF-KO nor STING/TRIF-DKO mice recapitulated the phenotype of IRF3-KO mice, which showed a much more substantial degree of protection from severe CLP-sepsis (75% survival), described in our prior report (22). We speculate that additional innate immune pathways activate IRF3 during sepsis, independent of STING and TRIF. For example, the role of the RIG-I-like receptors remains to be investigated. Additionally, we performed a moderate version of the CLP model, administering Lactated Ringer Solution and broad spectrum antibiotics to test the influence of these pathways in the context of clinical treatments. In this moderate model, TRIF-KO mice showed improvements in disease measurements relative to WT counterparts, while STING-KO mice were not protected.
Prior reports demonstrate that the serum of septic rodents contains bacterial PAMPS, such as LPS (8, 27) and host DAMPS, such as HMGB1 and cell-free DNA (15–17). We found significant numbers of bacteria in the peritoneum of mice following both severe and moderate CLP surgeries (Fig. 8A). Additionally, we found that cell-free mouse genomic DNA was slightly elevated after sham surgery relative to naive controls. Cell-free DNA was more significantly elevated in some mice that underwent severe CLP surgery, while the levels in mice that underwent moderate CLP were all similar to the sham controls (Fig. 8B). These data suggest that both bacteria and host DAMPS are released after sepsis; however, their abundance depends on the severity of the sepsis model.
Our in vitro data indicate that TRIF is a critical mediator of IRF3 activation in macrophages treated with sonicated bacteria derived from the mouse cecum. In contrast, STING is a critical mediator of IRF3 activation in macrophages treated with mouse genomic DNA. Although this in vitro culture system is somewhat artificial, it demonstrates proof of principal that TRIF and STING are capable of mediating IRF3 activation in response to bacterial PAMPs and host DAMPs that are present during sepsis. We attempted to detect IRF3 activation ex vivo in peritoneal cells obtained from septic mice; however, our efforts were not successful. This was likely due to technical issues: the antibody used for phosphorylated IRF3 has very low sensitivity in our hands and requires abundant innate immune activation for Western Blot detection, which likely exceeds physiological levels.
Our data indicate that there are important differences between the severe and moderate CLP sepsis models used in this study (Figs. 1–7). TRIF contributes to sepsis pathogenesis in both models, while STING contributes to the severe CLP model only. Bacteria are abundant in the peritoneum of mice who undergo both the severe and moderate CLP models (Fig. 8A). In contrast, cell-free genomic DNA becomes substantially elevated in the serum of some mice that undergo the severe CLP model, while all mice that underwent the moderate model showed levels of serum cell-free DNA that were similar to sham controls (Fig. 8B). In vitro, commensal bacteria activate IRF3 via TRIF, whereas mouse genomic DNA activates IRF3 via STING. These data are consistent with the notion that the milieu of PAMPs and DAMPs present in each model determine which innate immune pathways contribute to sepsis pathogenesis. Bacteria likely activate IRF3 via TRIF in both models, while cell-free DNA likely activates IRF3 via STING in the severe CLP model only. Clinical reports suggest that cell-free DNA is elevated in human patients with sepsis and is a useful biomarker to predict mortality (28, 31, 32). Hence, although our moderate CLP model aims to recapitulate clinical sepsis, some patients may have elevated levels of cell-free DNA that more closely resemble our severe CLP model.
IFNβ is the classical mediator of IRF3 activity and signals via IFNα/β receptor to induce IFNα. Notably, we could not detect IFNβ or IFNα in the serum or peritoneal lavage of mice that underwent CLP or sham surgeries (all values were below the minimum detection limit of 7.8 pg/mL). Interferons may exert their activity at very low levels (below the limits of ELISA detection) or may be present at a time point or anatomical site that we did not sample. Alternatively, STING and TRIF may exert their function independent of interferons during sepsis.
Chiswick et al. (29) recently studied the difference between mice predicted to die versus mice predicted to live after CLP (based on IL-6 levels) and found that mice predicted to die showed impaired phagocyte function. In our severe model of sepsis, we speculate that STING and TRIF may contribute to an excessive inflammatory response that impairs phagocyte function and bacterial killing. In support of this notion, we observed a significant reduction in the amount of bacteria present in the peritoneal cavity of STING-KO mice, and a trend toward reduced bacterial numbers in TRIF-KO mice versus WT counterparts (Fig. 4). This finding was not explained by an increased number of infiltrating phagocytes (Fig. 5; in fact, there was a trend toward a higher number of cells in the peritoneum of WT mice, paired with higher MCP-1 levels). Future studies examining phagocyte function may help to clarify if impaired bacterial killing may explain the increased mortality in WT versus STING and TRIF-KO mice, but are beyond the scope of this study.
Interestingly, IL-12/23p40 and CK were lower in STING-KO mice versus WT counterparts, but were not reduced in the TRIF-KO strain. A prior report indicates that IL-12 contributes to acute myocarditis during cocksackie-virus infection and the authors observed an IL-12-dependent elevation in serum CK values (33). We speculate that during CLP-sepsis, STING upregulates IL-12, contributing to heart damage. The increased mortality in WT mice versus TRIF-KO counterparts despite similar CK values may indicate that some mice are dying from a mechanism other than heart damage. Alternatively, IL-12 and CK may differ at a later time point in WT versus TRIF-KO mice, closer to impending death. Surprisingly, we observed similar IL-12p40 and CK levels in STING/TRIF-DKO mice versus WT counterparts; these data were unexpected given our results in STING-KO mice and we lack an explanation for this finding.
Previous studies examining the role of the TRIF pathway have garnered contradictory results. We observed that TRIF deficiency conferred protection from sepsis in both severe and moderate versions of the CLP model. A study by Feng et al. (21) tested a targeted TRIF-KO mouse strain (34) and found that these mice were not protected from CLP-induced severe sepsis, whereas these mice were significantly protected during LPS-induced shock. In contrast, Cuenca et al. (20) showed that TRIF-KO (LPS2 strain) adult mice were modestly protected from CS-induced moderate sepsis relative to WT counterparts, while TRIF-KO neonates had increased susceptibility. Our results in a severe model of CLP-sepsis conflict with the prior study by Feng et al. This discrepancy may stem from the use of a different genetically modified mouse to study the role of TRIF. Our results in the moderate CLP model are similar to the study by Cuenca et al. which used the same mouse strain.
In summary, our results, together with prior studies, suggest that the many innate immune pathways play a significant, detrimental role during severe models of sepsis. However, during moderate models of sepsis, some of these pathways remain neutral. The PAMPs and DAMPs released during severe versus moderate sepsis may help to explain this discrepancy. Further understanding of how these pathways contribute to sepsis pathogenesis, especially in the context of clinical treatments, may facilitate the development of novel therapies for human patients.
Support for biostatistical analysis was from the Biostatistics and Epidemiology Consulting Lab of the Paul L. Foster School of Medicine, TTUHSC El Paso. The authors thank Dr Russell Vance, University of California, Berkeley for providing the Goldenticket mouse strain, Alison Rembisz, Sara Ali, and Gabriela Segovia for technical assistance, and Kevin Liu, Christopher Dodoo, and Alok Dwivedi for their contributions to the statistical analyses.
1. Gaieski DF, Edwards JM, Kallan MJ, Carr BG. Benchmarking the incidence and mortality of severe sepsis in the United States. Crit Care Med
2013; 41 5:1167–1174.
2. Stoller J, Halpin L, Weis M, Aplin B, Qu W, Georgescu C, Nazzal M. Epidemiology of severe sepsis: 2008-2012. J Crit Care
2016; 31 1:58–62.
3. Ward NS, Casserly B, Ayala A. The compensatory anti-inflammatory response syndrome (CARS) in critically ill patients. Clin Chest Med
2008; 29 4:617–625.
4. Gentile LF, Cuenca AG, Efron PA, Ang D, Bihorac A, McKinley BA, Moldawer LL, Moore FA. Persistent inflammation and immunosuppression: a common syndrome and new horizon for surgical intensive care. J Trauma Acute Care Surg
2012; 72 6:1491–1501.
5. Baker CC, Chaudry IH, Gaines HO, Baue AE. Evaluation of factors affecting mortality rate after sepsis in a murine cecal ligation and puncture model. Surgery
1983; 94 2:331–335.
6. Ayala A, Herdon CD, Lehman DL, DeMaso CM, Ayala CA, Chaudry IH. The induction of accelerated thymic programmed cell death during polymicrobial sepsis: control by corticosteroids but not tumor necrosis factor. Shock
1995; 3 4:259–267.
7. Newcomb D, Bolgos G, Green L, Remick DG. Antibiotic treatment influences outcome in murine sepsis: mediators of increased morbidity. Shock
1998; 10 2:110–117.
8. Zantl N, Uebe A, Neumann B, Wagner H, Siewert JR, Holzmann B, Heidecke CD, Pfeffer K. Essential role of gamma interferon in survival of colon ascendens stent peritonitis, a novel murine model of abdominal sepsis. Infect Immun
1998; 66 5:2300–2309.
9. Wynn JL, Scumpia PO, Delano MJ, O’Malley KA, Ungaro R, Abouhamze A, Moldawer LL. Increased mortality and altered immunity in neonatal sepsis produced by generalized peritonitis. Shock
2007; 28 6:675–683.
10. Takeda K, Akira S. Toll-like receptors in innate immunity. Int Immunol
2005; 17 1:1–14.
11. Wu J, Sun L, Chen X, Du F, Shi H, Chen C, Chen ZJ. Cyclic GMP-AMP is an endogenous second messenger in innate immune signaling by cytosolic DNA. Science
2013; 339 6121:826–830.
12. Burdette DL, Monroe KM, Sotelo-Troha K, Iwig JS, Eckert B, Hyodo M, Hayakawa Y, Vance RE. STING is a direct innate immune sensor of cyclic di-GMP. Nature
2011; 478 7370:515–518.
13. Sun L, Wu J, Du F, Chen X, Chen ZJ. Cyclic GMP-AMP synthase is a cytosolic DNA sensor that activates the type I interferon pathway. Science
2013; 339 6121:786–791.
14. Wang H, Bloom O, Zhang M, Vishnubhakat JM, Ombrellino M, Che J, Frazier A, Yang H, Ivanova S, Borovikova L, et al. HMG-1 as a late mediator of endotoxin lethality in mice. Science
1999; 285 5425:248–251.
15. Meng W, Paunel-Gorgulu A, Flohe S, Hoffmann A, Witte I, MacKenzie C, Baldus SE, Windolf J, Logters TT. Depletion of neutrophil extracellular traps in vivo results in hypersusceptibility to polymicrobial sepsis in mice. Crit Care
2012; 16 4:R137.
16. Mai SH, Khan M, Dwivedi DJ, Ross CA, Zhou J, Gould TJ, Gross PL, Weitz JI, Fox-Robichaud AE, Liaw PC. Canadian Critical Care Translational Biology G. Delayed but not early treatment with dnase reduces organ damage and improves outcome in a murine model of sepsis. Shock
2015; 44 2:166–172.
17. Hamaguchi S, Akeda Y, Yamamoto N, Seki M, Yamamoto K, Oishi K, Tomono K. Origin of circulating free dna in sepsis: analysis of the clp mouse model. Mediators Inflamm
18. Weighardt H, Kaiser-Moore S, Vabulas RM, Kirschning CJ, Wagner H, Holzmann B. Cutting edge: myeloid differentiation factor 88 deficiency improves resistance against sepsis caused by polymicrobial infection. J Immunol
2002; 169 6:2823–2827.
19. Peck-Palmer OM, Unsinger J, Chang KC, Davis CG, McDunn JE, Hotchkiss RS. Deletion of MyD88 markedly attenuates sepsis-induced T and B lymphocyte apoptosis but worsens survival. J Leukoc Biol
2008; 83 4:1009–1018.
20. Cuenca AG, Joiner DN, Gentile LF, Cuenca AL, Wynn JL, Kelly-Scumpia KM, Scumpia PO, Behrns KE, Efron PA, Nacionales D, et al. TRIF-dependent innate immune activation is critical for survival to neonatal gram-negative sepsis. J Immunol
2015; 194 3:1169–1177.
21. Feng Y, Zou L, Zhang M, Li Y, Chen C, Chao W. MyD88 and Trif signaling play distinct roles in cardiac dysfunction and mortality during endotoxin shock and polymicrobial sepsis. Anesthesiology
2011; 115 3:555–567.
22. Walker WE, Bozzi AT, Goldstein DR. IRF3 contributes to sepsis pathogenesis in the mouse cecal ligation and puncture model. J Leukoc Biol
2012; 92 6:1261–1268.
23. Silver AC, Arjona A, Walker WE, Fikrig E. The circadian clock controls toll-like receptor 9-mediated innate and adaptive immunity. Immunity
2012; 36 2:251–261.
24. Walker WE, Booth CJ, Goldstein DR. TLR9 and IRF3 cooperate to induce a systemic inflammatory response in mice injected with liposome:DNA. Mol Ther
2010; 18 4:775–784.
25. Sauer JD, Sotelo-Troha K, von Moltke J, Monroe KM, Rae CS, Brubaker SW, Hyodo M, Hayakawa Y, Woodward JJ, Portnoy DA, et al. The N-ethyl-N-nitrosourea-induced Goldenticket mouse mutant reveals an essential function of Sting in the in vivo interferon response to Listeria monocytogenes and cyclic dinucleotides. Infect Immun
2011; 79 2:688–694.
26. Hoebe K, Du X, Georgel P, Janssen E, Tabeta K, Kim SO, Goode J, Lin P, Mann N, Mudd S, et al. Identification of Lps2 as a key transducer of MyD88-independent TIR signalling. Nature
2003; 424 6950:743–748.
27. Alexander HR, Sheppard BC, Jensen JC, Langstein HN, Buresh CM, Venzon D, Walker EC, Fraker DL, Stovroff MC, Norton JA. Treatment with recombinant human tumor necrosis factor-alpha protects rats against the lethality, hypotension, and hypothermia of gram-negative sepsis. J Clin Invest
1991; 88 1:34–39.
28. Margraf S, Logters T, Reipen J, Altrichter J, Scholz M, Windolf J. Neutrophil-derived circulating free DNA (cf-DNA/NETs): a potential prognostic marker for posttraumatic development of inflammatory second hit and sepsis. Shock
2008; 30 4:352–358.
29. Chiswick EL, Mella JR, Bernardo J, Remick DG. Acute-phase deaths from murine polymicrobial sepsis are characterized by innate immune suppression rather than exhaustion. J Immunol
2015; 195 8:3793–3802.
30. Hoebe K, Du X, Goode J, Mann N, Beutler B. Lps2: a new locus required for responses to lipopolysaccharide, revealed by germline mutagenesis and phenotypic screening. J Endotoxin Res
2003; 9 4:250–255.
31. Dwivedi DJ, Toltl LJ, Swystun LL, Pogue J, Liaw KL, Weitz JI, Cook DJ, Fox-Robichaud AE, Liaw PC. Canadian Critical Care Translational Biology G. Prognostic utility and characterization of cell-free DNA in patients with severe sepsis. Crit Care
2012; 16 4:R151.
32. Avriel A, Paryente Wiessman M, Almog Y, Perl Y, Novack V, Galante O, Klein M, Pencina MJ, Douvdevani A. Admission cell free DNA levels predict 28-day mortality in patients with severe sepsis in intensive care. PLoS One
2014; 9 6:e100514.
33. Fairweather D, Yusung S, Frisancho S, Barrett M, Gatewood S, Steele R, Rose NR. IL-12 receptor beta 1 and Toll-like receptor 4 increase IL-1 beta- and IL-18-associated myocarditis and coxsackievirus replication. J Immunol
2003; 170 9:4731–4737.
34. Yamamoto M, Sato S, Hemmi H, Hoshino K, Kaisho T, Sanjo H, Takeuchi O, Sugiyama M, Okabe M, Takeda K, et al. Role of adaptor TRIF in the MyD88-independent toll-like receptor signaling pathway. Science
2003; 301 5633:640–643.