IRF3 Signaling within the Mouse Stroma Influences Sepsis Pathogenesis

Erica L. Heipertz, Jourdan Harper, Dinesh G. Goswami, Charlie A. Lopez, Jose Nellikappallil, Ruben Zamora, Yoram Vodovotz and Wendy E. Walker
J Immunol January 15, 2021, 206 (2) 398-409; DOI: https://doi.org/10.4049/jimmunol.1900217

Key Points

  • IRF3 contributes to sepsis in a mouse model incorporating antibiotics and fluids.

  • IRF3 acts in stromal cells (nonleukocytes) to mediate this effect.

  • IRF3 indirectly alters macrophage behavior and the inflammatory network in sepsis.

Abstract

IFN regulatory factor 3 (IRF3) is a transcription factor that is activated by multiple pattern-recognition receptors. We demonstrated previously that IRF3 plays a detrimental role in a severe mouse model of sepsis, induced by cecal ligation and puncture. In this study, we found that IRF3–knockout (KO) mice were greatly protected from sepsis in a clinically relevant version of the cecal ligation and puncture model incorporating crystalloid fluids and antibiotics, exhibiting improved survival, reduced disease score, lower levels of serum cytokines, and improved phagocytic function relative to wild-type (WT) mice. Computational modeling revealed that the overall complexity of the systemic inflammatory/immune network was similar in IRF3-KO versus WT septic mice, although the tempo of connectivity differed. Furthermore, the mediators driving the network differed: TNF-α, IL-1β, and IL-6 predominated in WT mice, whereas MCP-1 and IL-6 predominated in IRF3-KO mice. Network analysis also suggested differential IL-6–related inflammatory programs in WT versus IRF3-KO mice. We created bone marrow chimeras to test the role of IRF3 within leukocytes versus stroma. Surprisingly, chimeras with IRF3-KO bone marrow showed little protection from sepsis, whereas chimeras with IRF3-KO stroma showed a substantial degree of protection. We found that WT and IRF3-KO macrophages had a similar capacity to produce IL-6 and phagocytose bacteria in vitro. Adoptive transfer experiments demonstrated that the genotype of the host environment affected the capacity of monocytes to produce IL-6 during sepsis. Thus, IRF3 acts principally within the stromal compartment to exacerbate sepsis pathogenesis via differential impacts on IL-6–related inflammatory programs.

This article is featured in Top Reads, p.243

Introduction

Sepsis is a medical emergency that requires prompt treatment to prevent death and disability. This syndrome arises when an infection induces a systemic inflammatory response, culminating in organ damage and dangerous levels of hypotension. Sepsis induces a 30–50% mortality rate, causes over 30% of in-hospital mortalities (1), and incurs the highest hospital costs among medical conditions in the United States (2). Over half of patients hospitalized with this condition are readmitted within 30 d (3). Additionally, patients who survive the acute phase of sepsis often experience impaired quality of life and a shortened lifespan (4).

Sepsis exhibits a complex immune/inflammatory etiology. The immune system is required to clear pathogens, but during sepsis, it becomes dysregulated as a function of elevated systemic inflammation (5). Overwhelming inflammation damages host tissues. High levels of serum cytokines increase vascular permeability. Neutrophils and monocytes extravasate, both at the local site of infection and in distal tissues, where they exert collateral damage. Conversely, the protective function of phagocytes and adaptive immune cells becomes impaired, compromising the host’s ability to clear the infection. As these inflammatory processes develop, lymphocytic processes become dysregulated, leading to immune exhaustion (5). Thus, many critically ill patients develop a persistent inflammation, immunosuppression, and catabolism syndrome, in which the immune system fails to return to homeostasis (6).

The inherent complexity of the intertwined infectious, inflammatory, and immune responses means that fighting sepsis is not as simple as broadly promoting or suppressing immunity. Many anti-inflammatory therapies that rescued mice from sepsis were ineffective in clinical trials [for example, soluble TNF-αR and IL-1R antagonist (7, 8)]. This caused investigators to rethink immunomodulatory therapies for sepsis (9). Recent approaches have focused on correcting immune deficits in sepsis, for example: suppressing lymphocyte apoptosis, supplementing Ig, and enhancing cellular immune responses (10–14). The limited success of early clinical trials also highlights the need to consider carefully the animal models of sepsis that we use. Many therapies that are effective in high-lethality models do not confer similar benefit in attenuated models of sepsis. Although no single animal model can recapitulate the heterogeneous conditions of human sepsis, a model with an infectious focus and a moderate mortality rate (30–50%) represents an advantageous system for preclinical studies (15).

Genetically manipulated mouse strains are a useful tool to elucidate how pattern-recognition receptors contribute to sepsis. IFN regulatory factor 3 (IRF3) was first identified as a transcription factor that is activated downstream of multiple pattern-recognition receptors, including TLRs, RIG-I–like receptors (RLRs), and cyclic GMP–AMP synthase (cGAS). Following its phosphorylation, IRF3 dimerizes, translocates to the nucleus, and binds to the promoter of IFN-β (16) as well as certain IFN-stimulated genes and cytokines (17). In addition to the classical function of IRF3, this protein influences diverse processes, including apoptosis (18), hepatocyte glucose regulation (19), and adipocyte browning (20). In a prior report, we used IRF3-deficient mice to demonstrate that IRF3 plays a significant, detrimental role during severe sepsis (21). IRF3–knockout (KO) mice that underwent a severe cecal ligation and puncture (CLP) sepsis model showed a substantial degree of protection relative to wild-type (WT) counterparts including reduced mortality, lower disease score, attenuated hypothermia, lower bacterial load, and reduced levels of serum cytokines (21). IRF3 played a similar role during the systemic inflammatory response induced by liposome:DNA (22). Further investigations using the CLP model and in vitro cell culture demonstrated that at least two pathways activate IRF3 during sepsis: the TIR domain–containing adapter-inducing IFN-β (TRIF) pathway (activated by bacteria) and the stimulator of IFN genes (STING) pathway (activated by cell-free genomic DNA) (23). The TRIF pathway appears to be relevant to sepsis pathogenesis both in a severe model of CLP sepsis (lacking clinical treatments) and in a moderate model of sepsis (incorporating antibiotics and lactated Ringer solution [LRS]) (23). In contrast, the STING pathway is only relevant in the severe CLP sepsis model (23), in which serum cell-free DNA is more abundant. Notably, mice lacking both TRIF and STING did not recapitulate the substantial degree of protection observed in IRF3-KO mice, suggesting that another factor must be at play.

In this study, we sought to further clarify how IRF3 influences sepsis pathogenesis in a more clinically realistic mouse model of sepsis. We found that IRF3-KO mice showed a substantial degree of protection in a moderate model of sepsis-incorporating clinical treatments (antibiotics and LRS) associated with reorganization of dynamic networks of systemic inflammation. Surprisingly, our data imply that the stromal compartment plays the predominant role in altering immune function in IRF3-KO mice, protecting them from sepsis.

Materials and Methods

Animal use

All animal work was approved by the Institutional Animal Care and Use Committee of the Texas Tech University Health Sciences Center system. This study used WT (C57BL/6J) mice and B6.SJL mice obtained from The Jackson Laboratory as well as IRF3-KO mice (17) obtained from Yale University. Experimental animals were bred in the animal facility at Texas Tech University Health Sciences Center El Paso. The study included both male and female animals. For intact mice, animals were used at 6–15 wk of age. For bone marrow chimeras, mice were used at 13–21 wk of age (8–13 wk postreconstitution with donor bone marrow). Within each experimental repeat, animals were matched for gender and age.

Sepsis model

We used a moderate CLP model of sepsis, described in our prior report (23). Following induction of isoflurane anesthesia, a midline incision was made in the peritoneum, and the cecum was located and exteriorized. One centimeter of the cecum was ligated with silk suture, and the organ was punctured once (through and through) with a 21-gauge (21G) needle. A small drop of fecal material was extruded, and the cecum was returned to the peritoneal cavity. The abdominal wound was closed in two layers. For sham surgery, the cecum was exteriorized and relocated in the peritoneum without ligation or puncture. All animals received buprenorphine sustained-release analgesia (0.5–1 mg/kg preoperative, then at 48-h intervals until alert or dead), 1 ml warm (37°C) LRS s.c. postoperative, and imipenem/cilastatin (25 mg/kg) i.p. at 12-h intervals for 5 d. As in our prior study, we observed periodic variations in the severity of this CLP model (23). We adjusted the timing of the first dose of imipenem/cilastatin from 1 to 6 h postoperative to maintain a mortality rate of 30–60% throughout the duration of the study. For the experiment comparing C57BL/6J, IRF3-KO (original), and IRF3-KO (backcrossed) mice, we used the severe model of sepsis described in our prior studies (21, 23–25).

Measurement of sepsis pathogenesis and serum cytokines via ELISA

Following CLP, mouse survival was monitored using inability to regain sternal recumbency as a humane end point (26). A disease score was assigned according to the degree of animal lethargy (0, alert; 1, slightly lethargic; 2, lethargic; 3, very lethargic; and 4, dead). Body surface temperature was monitored via infrared thermometer measurements taken on the mouse sternum. The animals were bled via the retro-orbital plexus, as previously described (27), and serum was prepared in gold-top microtainers (Becton Dickinson, Franklin Lakes, NJ).

The levels of proinflammatory cytokines IL-6 and IL-12/23p40 and the chemokine MCP-1 were measured by ELISA using BD OptEIA kits (Becton Dickinson), and samples were assayed for HMGB1 using a commercially available ELISA (Shino-Test, Kanagawa, Japan). We also measured a number of cytokines and chemokines that serve as biomarkers for the complex inflammatory response using the Luminex 100 IS system (Luminex, Austin, TX). The Ab bead kit includes the following: GM-CSF, IFN-γ, IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p40, IL-12p70, IL-13, IL-17, IP-10/CXCL10, keratinocyte-derived cytokine (KC)/CXCL1, MCP-1/CCL2, MIG/CXCL9, MIP-1α/CCL3, TNF-α, and vascular endothelial growth factor (VEGF).

Phagocytosis assay

Peritoneal lavage was collected from WT and IRF3-KO mice 21 h postsurgery (for ex vivo experiments) or 4 d after thioglycollate injection (for in vitro experiments), per our prior report (27). A total of 5 × 105 cells were resuspended in 1 ml complete RPMI 1640 (RPMI 1640 medium containing 1% penicillin–streptomycin and 10% FBS; all from Thermo Fisher Scientific, Waltham, MA) and were aliquoted into a cryotube and 50 μl of pHrodo Green Escherichia coli bioparticles was added. The suspension was incubated at 37°C in a stirring water bath for 1.5 h with manual agitation every 15 min. After completion of the assay, all samples were kept on ice to prevent further phagocytosis. Trypan blue was added to quench the fluorescent signal emitted by extracellular bioparticles. The bright green fluorescent signal emitted by phagocytosed bioparticles was acquired with a BD FACSCanto II flow cytometer (Becton Dickinson, Franklin Lakes, NJ) and analyzed using FlowJo software (FlowJo LLC, Ashland, OR).

Dynamic network analysis

Dynamic network analysis (DyNA) was carried out to define, in a granular fashion, the central inflammatory network nodes as a function of both time and experimental conditions. Using inflammatory mediator measurements, networks were created over three consecutive time periods (0–3, 3–6, and 6–20 h) using MATLAB software (28–31). Network connections, defined as the number of trajectories of serum inflammatory mediators that move in parallel, were created if the Pearson correlation coefficient between any two nodes (inflammatory mediators) at the same time interval was greater or equal to a threshold of 0.9. The network complexity for each time interval was calculated using the following formula: sum (N1 + N2 +…+ Nn)/n−1, where N represents the number of connections for each mediator, and n is the total number of mediators analyzed (28–31).

Generation of bone marrow chimeras

Recipient mice were irradiated with an RS 2000 x-ray irradiator (Rad Source Technologies, Buford, GA). Animals were exposed to two doses of 500 cGy, separated by a 4-h interval, to deplete endogenous leukocytes and bone marrow stem cells. Donor mice were euthanized, and their femurs and tibia were harvested aseptically. The bones were cleaned, and the marrow was flushed out of the medullary cavity with PBS using a needle and syringe. The bone marrow was filtered through a 70-μm nylon sieve, centrifuged, and resuspended in PBS. A total of 1 × 107 bone marrow cells were i.v. injected into each irradiated recipient mouse via the tail vein or retro-orbital plexus. Ten to twelve weeks after adoptive cell transfer, flow cytometry was used to confirm reconstitution of the leukocyte compartment with donor cells. Blood was collected by retro-orbital bleed or submandibular vein puncture from anesthetized mice, and PBMCs were prepared by Histopaque-1077 density centrifugation (Sigma-Aldrich, St. Louis, MO), according to the manufacturer’s instructions. The cells were stained with anti-CD45.1 and anti-CD45.2 Abs. Fluorescence data were acquired with a BD FACSCanto II flow cytometer (Becton Dickinson) and analyzed using FlowJo software (FlowJo, Ashland, OR), per our prior report (26).

Peritoneal thioglycollate-elicited macrophages elicitation and isolation

Mice were injected with 1.5 ml of 4% Brewer’s Thioglycollate Medium (pH 7.2; Sigma-Aldrich). Four days later, animals were euthanized, and peritoneal lavage was collected as previously described (27). The peritoneal cells were collected and cultured at 5 × 105 per well in a 24-well plate; nonadherent cells were removed to enrich macrophages. The macrophages were treated with sonicated bacteria cultured from the mouse cecum [prepared as described in our prior reports (21, 23)] at a 1:200 and 1:2000 final concentration incubated for 18 h, then an ELISA kit was used to measure IL-6 in the culture media.

Enrichment and adoptive transfer of monocytes

The tibia and femur were harvested from donor B6.SJL mice (carrying the CD45.1 allele), and bone marrow was flushed out as described above. Monocytes were enriched via magnetic selection with a mouse Monocyte Isolation Kit (BM) (Miltenyi Biotec, Bergisch Gladbach, Germany), per the manufacturer’s instructions. A total of 5 × 105 monocytes were injected i.p. into WT or IRF3-KO recipient mice (both carrying the CD45.2 allele) 3 h after CLP surgery. Eighteen hours after surgery, the recipient mice were euthanized. Peritoneal lavage was harvested as described above. Peritoneal cells were cultured in complete RPMI 1640 containing brefeldin A (GolgiPlug; BD Biosciences) for 4 h to block IL-6 secretion and then stained with cell surface Abs against CD45.1 and CD45.2 and intracellular IL-6 using the method described in our prior report (22).

Statistical analyses

Statistical analysis was performed using Prism 6 (GraphPad, San Diego, CA) and R software (The R Foundation). The experiments were repeated at minimum in triplicate; graphs show pooled results. Survival rates were compared using the Kaplan–Meier log-rank test. Surface temperatures and disease score approximated a normal distribution and were compared with a two-way repeated measures ANOVA. Serum inflammatory mediator data measured over time were log transformed to approximate a normal distribution and then compared with a two-way repeated measures ANOVA. Phagocytosis in peritoneal exudate cells were compared with a two-sample proportions test. The correlation between animal disease score and phagocytosis was assessed with Pearson test. In vitro IL-6 and phagocytosis data were compared with a Mann–Whitney U test. For the severe CLP model, levels of serum cytokines at 18 h were compared between multiple groups with a Kruskal–Wallis test. All p values <0.05 were considered statistically significant.

Results

IRF3 substantially contributes to sepsis pathogenesis in the context of clinical treatments for sepsis

We showed previously that IRF3 contributes to sepsis pathogenesis in a severe version of the CLP model (21). In the current study, we tested if IRF3 remains relevant in the context of a moderate CLP model incorporating clinical treatments to attenuate sepsis mortality. Mice underwent a smaller cecal ligation (1 cm) and single 21G needle puncture, then received LRS s.c. postoperative and imipenem/cilastatin antibiotics i.p. twice daily for 5 d. In this moderate CLP model, WT mice exhibited 61.5% survival (Fig. 1A). In comparison, IRF3-KO mice were significantly protected from sepsis, showing 92.3% survival. All mice that underwent sham surgery survived, regardless of genotype. In the WT mice, we observed a higher animal disease score (degree of lethargy) relative to IRF3-KO mice (Fig. 1B). In contrast to the severe model of CLP used in our prior study, we did not observe a substantial change in the mean surface temperature of mice in this moderate CLP model, regardless of genotype (Fig. 1C), although some individual mice who succumbed to sepsis did exhibit hypothermia prior to death. In both IRF3-KO and WT mice, the levels of the proinflammatory cytokines IL-6 (Fig. 1D) and IL-12/23p40 (Fig. 1E) as well as of the chemokine MCP-1 (Fig. 1F) reached their highest levels at 6 h post-CLP, followed by a decrease at 18 h post-CLP. However, the WT mice had significantly higher levels of circulating IL-6, IL-12/23p40, and MCP-1 when compared with the IRF3-KO mice (Fig. 1D–F). None of the sham-operated mice developed lethargy or elevated serum proinflammatory cytokine levels (Fig. 1D–F). These data suggest that IRF3 signaling contributes to sepsis pathogenesis in this moderate model of sepsis-incorporating clinical treatments.

FIGURE 1.
FIGURE 1.

IRF3 contributes to sepsis pathogenesis in the context of clinical treatments for sepsis. (A–F) IRF3-KO mice were subject to CLP (1-cm ligation, 1 × 21G needle puncture) and treated with LRS postoperative and imipenem/cilastatin twice per day for 5 d. We monitored (A) animal survival (B) disease score (degree of lethargy) and (C) animal surface temperature. Mice were bled at 0, 6, and 18 h postsurgery; serum was prepared; and the levels of (D) IL-6, (E) IL-12/23p40, and (F) MCP-1 were measured by ELISA. Graphs show Kaplan–Meier plots for animal survival and line graphs with mean and SE for the other measures.

IRF3 expression influences peritoneal cell phagocytosis in septic mice

In our prior study, we showed that IRF3-KO mice have a lower bacterial load than WT counterparts after CLP surgery (21). Prior reports demonstrate that mice predicted to die of sepsis exhibit impaired phagocytosis, associated with elevations in bacterial load (32). We envisioned that IRF3 could be exacerbating sepsis pathogenesis by inducing a defect in bacterial phagocytosis in peritoneal cells. We performed the moderate CLP procedure on WT and IRF3-KO mice and isolated peritoneal cells 21 h after CLP surgery. These peritoneal cells were incubated with fluorescently labeled E. coli bioparticles. Extracellular fluorescence was quenched with trypan blue, and then flow cytometry was performed to evaluate the percentage of cells phagocytosing these bioparticles. WT and IRF3-KO mice that underwent sham surgery exhibited a similar percentage of phagocytic cells (medians 40.2 and 35.9%, respectively, with a range of values between 26.4 and 47.8%; Fig. 2A). In contrast, mice that underwent CLP exhibited a biphasic response: some animals exhibited preserved (or even increased) phagocytosis, whereas others exhibited reduced phagocytosis (below the range in sham-operated mice [e.g., <26.4%]; Fig. 2A). Among the septic WT mice, 6/16 animals (37.5%) exhibited reduced phagocytosis, whereas only 1/17 (5.9%) septic IRF3-KO animals exhibited reduced phagocytosis (Fig. 2A), representing a significant difference in proportion. When we pooled all the surgical mice of both genotypes, we observed an inverse correlation between the percentage of phagocytic cells and the animal disease score, which was statistically significant (Fig. 2B). In summary, impaired phagocytosis was more prevalent in WT mice versus IRF3-KO counterparts, especially among mice with the highest disease score.

FIGURE 2.
FIGURE 2.

IRF3 influences the phagocytic capacity of peritoneal cells after CLP. (A and B) WT and IRF3-KO mice were subject to CLP (n = 16–17 per group) or sham surgeries (n = 7 per group); 21 h later, peritoneal cells were harvested and exposed to fluorescent E. coli bioparticles. (A) The percentage of FITC+ cells was analyzed via flow cytometry after cells were gated according to their forward scatter versus side scatter. Dashed lines denote the lower range of values obtained in the sham-operated mice (26.4%). (B) The disease score of mice 21 h after CLP was plotted against the percent of peritoneal cells undergoing phagocytosis. Each data point represents the value for one mouse.

IRF3 increases the complexity of the inflammatory network early after CLP

To further elucidate how IRF3 affects the dynamics of the inflammatory response, we next assayed a broad panel of systemic mediators that characterize a wide variety of immune/inflammatory pathways (damage-associated molecular patterns [DAMPs], chemokines, M1, M2, Th1, Th2, and Th17). These mediators were measured in the sera of WT and IRF3-KO mice at serial time points after CLP (0, 3, 6, and 20 h postsurgery). The levels of these mediators were analyzed by traditional statistical analyses (two-way ANOVA), as well as more advanced in silico modeling (DyNA) (28), to discern potential differences in the inflammatory programs of IRF3-KO mice versus WT controls. Based on these analyses, Table I shows the number of nodes (inflammatory mediators) and their interconnections (edges) at serial time intervals (0–3, 3–6, and 6–20 h). Fig. 3A presents a graph of the total number of network connections (indicative of network complexity), and Fig. 3B presents a circle plot showing these interconnections between individual mediators.

View this table:
Table I. Serum inflammatory/immune network connectivity
FIGURE 3.
FIGURE 3.

Modeling the inflammatory network in WT versus IRF3-KO mice. (A and B) A broad panel of systemic inflammatory/immune mediators was measured in the sera of WT and IRF3-KO mice (n = 4 per group) at 0, 3, 6, and 20 h after CLP surgery. Traditional statistical analysis and in silico (DyNA) modeling were performed to assess identify network connection at serial intervals (0–3, 3–6, and 6–20 h). (A) The overall degree of network complexity represented as total number of connections for each mouse strain was graphed over time. (B) A circle plot was generated to display the individual network connections.

Our analyses of dynamic networks suggested that the inflammatory response was roughly equally complex in WT versus IRF3-KO mice from a quantitative standpoint when all time points were considered (88 versus 80 total network connections, respectively; Table I). However, WT and IRF3-KO mice differed when considering individual time intervals: at the earliest interval (0–3 h), the systemic inflammatory response of WT mice was substantially more complex than that of IRF3-KO mice (68 versus 38 total network connections, respectively; Fig. 3, Table I). This pattern was reversed between 3 and 6 h, with the systemic inflammatory network in WT mice exhibiting less overall connectivity as compared with IRF3-KO mice (2 versus 20 network connection, respectively; Fig. 3, Table I). In the last time interval (6–20 h), both WT and IRF3-KO network connectivity were similar (18 versus 22 network connections, respectively; Fig. 3, Table I).

We next examined the degree of network connectivity of individual inflammatory mediators in WT versus IRF3-KO mice, hypothesizing that the greater the degree of network connectivity of a given mediator, the more central the role for that pathway in that strain of mouse. Considering the total number of network connections at all time points, our results suggest that TNF-α (12 connections), IL-1β (nine connections), and IL-6 (nine connections) are principal drivers of the inflammatory network in WT mice, whereas MCP-1 (16 connections) and IL-6 (10 connections) predominated in IRF3-KO mice (Fig. 3B, Table I).

Because IL-6 was a shared, highly connected node in dynamic networks of systemic inflammation for both genotypes, we considered the specific connections for this cytokine. This analysis showed substantial differences between WT and IRF3-KO mice (Fig. 3B, Table II). In WT mice, IL-6 showed early connectivity to the cytokines TNF-α, IL-12p40, and IL-17A as well as the chemokines IP-10, KC, and MIG (0–3 h); however, these connections were lost by the second interval (3–6 h). By the last interval (6–10 h), IL-6 formed a new connection with the DAMP HMGB1. In IRF3-KO mice, IL-6 was connected to the chemokine MCP-1 at all intervals and formed a connection with GM-CSF at the second and third intervals (3–6 and 6–20 h). Thus, we discerned distinct IL-6–related inflammatory programs in WT versus IRF3-KO mice.

View this table:
Table II. IL-6 connections

IRF3 signaling in the leukocyte compartment plays a minor role in sepsis pathogenesis

Our data suggest that IRF3 substantially changes the network of inflammatory mediators, particularly in regard to IL-6 (Fig. 3, Tables I, II). Because IL-6 is predominantly produced by immune cells, we reasoned that IRF3 likely acts within leukocytes to exert its major effect on sepsis pathogenesis. We used bone marrow chimeras to determine if this was the case. First, WT B6.SJL hosts (carrying the CD45.1 allele) were irradiated and reconstituted with WT C57BL/6J or IRF3-KO bone marrow (both strains carry the CD45.2 allele). Ten to twelve weeks later, the animals were bled, and PBMCs were stained to confirm reconstitution of the hematopoietic compartment with donor cells. We found that, on average, 94.3% of PBMCs in these chimeras were CD45.2+ (individual values ranged from 85.2 to 99.8%). These data indicate that the animals were successfully reconstituted with donor bone marrow.

These bone marrow chimeras underwent moderate CLP surgery to induce sepsis. In chimeras with WT bone marrow, the sepsis survival rate was 50.6% (Fig. 4A). Chimeras with IRF3-KO bone marrow exhibited only a marginal increase in their sepsis survival rate (67.5%), which was not statistically significant (Fig. 4A). Chimeras with IRF3-KO bone marrow also exhibited a trend toward improved animal disease score; however, this effect was minimal and not statistically significant (Fig. 4B). We did observe a small degree of hypothermia among the chimeras that underwent the moderate CLP model, relative to their counterparts that underwent sham surgery (Fig. 4C). However, we observed no differences in the surface temperatures of mice with WT versus IRF3-KO bone marrow (Fig. 4C). In both groups of chimeras that underwent CLP, the levels of IL-6 (Fig. 4D), IL-12/23p40 (Fig. 4E), and MCP-1 (Fig. 4F) reached peak levels at 6 h after CLP surgery and then declined at 18 h post-CLP. We observed a trend toward lower levels of serum IL-6 and MCP-1 in chimeras with IRF3-KO bone marrow versus WT bone marrow; however, this effect was minor, and the difference was not statistically significant (Fig. 4D–F). Control mice that underwent sham surgery survived and did not develop lethargy, hypothermia, or elevated proinflammatory mediators (Fig. 4). Surprisingly, these results indicate that the absence of IRF3 in hematopoietic cells had little effect on sepsis pathogenesis.

FIGURE 4.
FIGURE 4.

IRF3 signaling within mouse leukocytes plays a minor role in sepsis pathogenesis. (A–F) WT B6.SJL hosts were irradiated and reconstituted with WT C57BL/6J or IRF3-KO bone marrow. The resulting bone marrow chimeras were subject to CLP (1-cm ligation, 1 × 21G needle puncture) and treated with LRS postoperative and imipenem/cilastatin twice per day for 5 d. We monitored (A) animal survival (B) disease score (degree of lethargy) and (C) animal surface temperature. Mice were bled at 0, 6, and 18 h postsurgery; serum was prepared; and the levels of (D) IL-6, (E) IL-12/23p40, and (F) MCP-1 were measured by ELISA. Graphs show Kaplan–Meier plots for animal survival and line graphs with mean and SE for the other measures.

IRF3 signaling in the stromal compartment plays a major role in sepsis pathogenesis

Given our finding that mice lacking IRF3 in the leukocyte compartment do not recapitulate the phenotype of IRF3-KO mice, we hypothesized that IRF3 plays a role in the stromal compartment to modulate the inflammatory network and mouse survival. To test this hypothesis, IRF3-KO and WT C57BL/6J hosts (both carrying the CD45.2 allele) were irradiated and reconstituted with bone marrow from WT B6.SJL donors (carrying the CD45.1 allele). We found that, on average, 94.8% of PBMCs were CD45.1+ (individual values ranged from 90.9 to 97.1%). These data indicate that the animals were successfully reconstituted with donor bone marrow.

These chimeras underwent CLP to induce sepsis. Chimeras with a WT stroma exhibited a survival rate of 42.1%, whereas chimeras with an IRF3-KO stroma exhibited a substantially improved survival rate of 87.5% (Fig. 5A). Furthermore, animal disease score was higher in chimeric mice with a WT stroma, relative to chimeras with an IRF3-KO stroma (Fig. 5B). Additionally, chimeras with a WT stroma showed a greater drop in surface body temperature relative to counterparts with an IRF3-KO stroma (Fig. 5C). The levels of serum cytokines IL-6 (Fig. 5D) and IL-12/23p40 (Fig. 5E) and the chemokine MCP-1 (Fig. 5F) increased substantially after CLP, reaching maximum levels at 6 h post-CLP, followed by a decline in levels at 18 h post-CLP. The levels of IL-6, IL-12/23p40, and MCP-1 were higher in the chimeras with WT stroma when compared with the chimeras with IRF3-KO stroma (Fig. 5D–F). Control mice that underwent sham surgery survived and did not develop lethargy, hypothermia, or elevated proinflammatory mediators (Fig. 5).

FIGURE 5.
FIGURE 5.

IRF3 signaling within mouse stroma plays a major role in sepsis pathogenesis. (A–F) WT C57BL/6J or IRF3-KO hosts were irradiated and reconstituted with WT B6.SJL bone marrow. The resulting bone marrow chimeras were subject to CLP (1-cm ligation, 1 × 21G needle puncture) and treated with LRS postoperative and imipenem/cilastatin twice per day for 5 d. We monitored (A) animal survival (B) disease score (degree of lethargy) and (C) animal surface temperature. Mice were bled at 0, 6, and 18 h postsurgery; serum was prepared; and the levels of (D) IL-6, (E) IL-12/23p40, and (F) MCP-1 were measured by ELISA. Graphs show Kaplan–Meier plots for animal survival and line graphs with mean and SE for the other measures.

Thioglycollate-elicited macrophages from WT and IRF3-KO mice exhibit similar IL-6 production and phagocytosis in vitro

The results in Fig. 4 suggest that IRF3 signaling within leukocytes has little impact on serum cytokine production during sepsis. However, the inflammatory network was substantially reprogrammed in the presence versus absence of IRF3 (Fig. 2, Table I). Specifically, IRF3 increased the levels of serum IL-6 (Fig. 1D) and altered its connectivity (Fig. 3B, Table II). We set out to examine if IRF3 directly influences the capacity of inflammatory macrophages to produce IL-6 once they are removed from the host environment. For this purpose, we isolated thioglycollate-elicited macrophages from WT and IRF3-KO mice. These cells were stimulated in vitro with sonicated bacteria cultured from the mouse cecum at two dilutions (1/2000 and 1/200) for 18 h, and IL-6 levels were measured in the supernatant via ELISA. We found that IRF3-KO macrophages secreted slightly higher levels of IL-6 into the culture supernatant with both dilutions of bacteria (Fig. 6A); however, this effect was very slight and not statistically significant. The control cells without bacteria did not secrete IL-6 (Fig. 6A).

FIGURE 6.
FIGURE 6.

WT and IRF3-KO macrophages exhibit similar functions outside of the host environment. (A) Thioglycollate-elicited macrophages from WT and IRF3-KO mice were cultured alone (rest) or with sonicated bacteria cultured from the mouse cecum at a 1/2000 or 1/200 dilution. Eighteen hours later, the level of IL-6 was measured in the culture supernatant by ELISA. Graph shows pooled data from three independent experiments. n = 8 samples per group. (B) Thioglycollate-elicited macrophages from WT and IRF3-KO mice were subject to a phagocytosis assay using fluorescently labeled E. coli. The percentage of FITC+ macrophages was measured by FACS. Graph shows pooled data from three independent experiments, n = 9 samples per group. A Mann–Whitney U test was used for pairwise comparisons. Each data point represents the value for one mouse, and graphs show median and interquartile range.

Next, we tested the phagocytic capacity of thioglycollate-elicited macrophages from WT and IRF3-KO mice. Cells were incubated with pHrodo Green E. coli bioparticles (these emit a bright green fluorescent signal in the acidified endosome). Extracellular fluorescence was quenched with trypan blue, and then flow cytometry was performed to evaluate the percentage of cells phagocytosing these bioparticles. We observed a trend toward slightly lower levels of phagocytosis in the IRF3-KO cells versus WT cells (Fig. 6B); however, this difference was not statistically significant. Hence, the lower levels of IL-6 production and higher levels of phagocytosis that we observed in IRF3-KO mice that underwent CLP (Fig. 1) were not recapitulated during these in vitro experiments (Fig. 6); in fact, opposing trends were observed.

An IRF3-deficient host environment attenuates IL-6 production by adoptively transferred monocytes

These experiments above suggest that IRF3 does not impair the capacity of myeloid cells to produce IL-6 once they are removed from the host environment. We reasoned that the genotype of the host environment (WT versus IRF3-KO) may alter the capacity of myeloid cells to produce IL-6. To test this, we enriched monocytes from the bone marrow of WT B6.SJL (CD45.1) mice and injected them into the peritoneum of WT and IRF3-KO mice (both CD45.2) 3 h after the hosts underwent CLP or sham surgery. Eighteen hours later, the hosts were euthanized, and the peritoneal cells were recovered by lavage. We incubated the cells with brefeldin A for 4 h to block the secretory pathway and allow IL-6 to accumulate within cells producing this cytokine. Subsequently, the cells were stained for CD45.1 (to identify adoptively transferred cells) and intracellular IL-6. We observed a higher percentage of IL-6–producing cells when monocytes were adoptively transferred into WT versus IRF3-KO hosts that underwent CLP (Fig. 7). Adoptively transferred monocytes produced little IL-6 in WT hosts that underwent sham surgery (Fig. 7). In fact, the percentage of IL-6–producing cells was similar in the WT sham-operated hosts and the IRF3-KO CLP hosts. These data suggest that the genotype of the host environment modulates IL-6 production by myeloid cells located therein.

FIGURE 7.
FIGURE 7.

The host environment influences IL-6 production by adoptively transferred cells. Monocytes were isolated from the bone marrow of B6.SJL (CD45.1+) mice and injected into the peritoneum of WT versus IRF3-KO (CD45.2+) hosts after CLP or sham surgery. This graph shows the percentage of CD45.1+ cells producing IL-6, measured by intracellular cytokine staining. A Mann–Whitney U test was used for pairwise comparisons. Each data point represents the value for one mouse, and graph shows median and interquartile range.

The IRF3-KO mice used throughout this study exhibited a mixed C57BL/6J and 129S1/Svlm strain background; however, IRF3-KO mice retained their protected phenotype after de novo backcrossing to C57BL/6J

A recent report identified an issue with the congenicity of the IRF3-KO mouse strain (33). In that study, a high-density single-nucleotide polymorphism analysis (genome scan) revealed that the IRF3-KO mice carried substantial strain 129 contamination (15–25%), suggesting incomplete backcrossing (33). We employed a genome scan to determine the congenicity of our own IRF3-KO mice. Similar to the prior report, we found that our mice were 69.2% congenic to C57BL/6J and 30.8% congenic to 129S1/Svlm. These data are consistent with the notion that our IRF3-KO mice were incompletely backcrossed to C57BL/6J. The IRF3-KO mice were homozygous at all loci, suggesting that polymorphisms had become fixed at each locus in either the C57BL/6J or 129S1/Svlm form. Importantly, PCR-sequencing verified that our mice did not carry the inactivating caspase 11 gene mutation observed in 129 strain mice.

For scientific rigor, it was imperative to determine if the sepsis phenotype that we observed was caused by a contaminating 129S1/Svlm gene allele. We backcrossed our IRF3-KO (original) mice to the C57BL/6J strain for four generations and then intercrossed the offspring to create IRF3-KO (backcrossed) mice. A genome scan revealed that the backcrossed mice were 96.9% congenic to C57BL/6J strain. To get an expedient answer using the minimum number of animals, we performed the severe CLP model described in our prior reports (21, 23–25). We induced severe CLP in the IRF3-KO (backcrossed) mice alongside the IRF3-KO (original) mice and C57BL/6J controls. We found that the IRF3-KO (backcrossed) mice and the IRF3-KO (original) mice were substantially protected from sepsis relative to C57BL/6J controls, as evidenced by improved survival (Fig. 8A), reduced disease score (Fig. 8B), less-severe hypothermia (Fig. 8C), and lower levels of serum cytokines at 18 h postsurgery, including IL-6 (Fig. 8D), IL-12p40 (Fig. 8E,) and MCP-1 (Fig. 8F). Moreover, the phenotype of the IRF3-KO (backcrossed) mice was very similar to the IRF3-KO (original) mice (Fig. 8), suggesting that none of the 129S1/Svlm gene alleles present in the original strain significantly influence sepsis in this model. These data confirm that IRF3 deficiency protects mice from sepsis, regardless of the 129S1/Svlm genomic contamination.

FIGURE 8.
FIGURE 8.

IRF3-KO mice retain their protected phenotype after de novo backcrossing to C57BL/6J. Severe sepsis was induced in IRF3-KO (backcrossed) mice (96.9% congenic to C57BL/6J), the IRF3-KO (original) mice used throughout this study (69.2% congenic to C57BL/6J), and C57BL/6J mice. We monitored (A) animal survival (B) disease score (degree of lethargy) and (C) animal surface temperature. Mice were bled at 18 h postsurgery; serum was prepared; and the levels of (D) IL-6, (E) IL-12/23p40, and (F) MCP-1 were measured by ELISA (please note the logarithmic scale). Graphs show Kaplan–Meier plots for animal survival and dotplots with median for the other measures.

Discussion

A growing body of literature supports the notion that IRF3 plays an important role in sepsis and related conditions of sterile systemic inflammatory response. We previously showed that IRF3-KO mice are substantially protected in a severe version of the CLP sepsis model lacking clinical treatments (21). Additionally, we showed that IRF3-KO mice are protected from the systemic inflammatory response induced by lipsome:DNA (22). Two reports by the Taniguchi laboratory showed that IRF3-KO mice are substantially protected from LPS-induced endotoxemia (34, 35). Another study showed that hydroxystilbenes inhibit LPS-induced IRF3 activation and protect mice from LPS-induced endotoxemia (36).

In this report, we found that IRF3 plays a substantial role in a clinically relevant CLP sepsis model, employing LRS and broad-spectrum antibiotics. In this model, C57BL/6J mice showed 38.5% mortality (Fig. 1A), similar to the mortality rate for patients with severe sepsis (30–50%). In contrast to the severe model used in our prior study (21), WT mice did not exhibit hypothermia as a group (Fig. 1C), although the individual mice that succumbed to sepsis did exhibit hypothermia when they became very lethargic. IRF3-KO mice showed a substantially reduced mortality rate of 7.7% (Fig. 1A), paired with considerably lower disease score (Fig. 1B) and reduced levels of serum cytokines (Fig. 1D–F). Furthermore, a substantial number of WT mice exhibited a phagocytic defect after CLP, whereas this immune defect was rare in IRF3-KO mice (Fig. 2A). These data demonstrate that IRF3 contributes to sepsis pathogenesis in the context of a clinically relevant mouse model.

IRF3 is best known as a transcription factor that induces IFN-β and related mediators following activation of pattern-recognition receptors. Our prior report investigated the upstream pathways that activate IRF3 in the context of our severe and moderate models of sepsis (23). In our severe model of sepsis (lacking clinical treatments), we found that both TRIF and STING contribute to disease pathogenesis, as mice lacking these proteins showed improved survival, attenuated disease score, improved thermoregulation, and lower levels of serum cytokines (23). In our moderate model of sepsis (employing LRS and antibiotics), TRIF-deficient mice showed modest improvements in their disease score and thermoregulation and a trend toward improved survival that did not reach statistical significance (23). STING did not significantly impact sepsis pathogenesis in this moderate model employing clinical treatments (23). We determined that bacteria derived from the mouse cecum activated IRF3 via TRIF, whereas mouse genomic DNA activated IRF3 via STING in cultured macrophages (23). Bacteria were present in the peritoneum of mice 18 h after CLP in both the severe model and the moderate model (antibiotic treatment in the moderate model lowered the bacterial load, but did not ameliorate it) (23). We also looked at serum cell-free DNA (a DAMP that is released from host cells during sepsis). Some mice that underwent the severe CLP exhibited high levels of serum cell-free DNA, whereas levels in the moderate CLP model remained similar to sham-operated controls (26). Hence, the levels of the pathogen-associated molecular patterns and DAMPs in each model likely explain their dependence on TRIF and STING. Importantly, even mice lacking both TRIF and STING failed to recapitulate the substantial degree of protection observed in IRF3-KO mice, shown in this study (Fig. 1) and in our prior report (21). Hence, another factor must explain their phenotype. This factor could be an additional pattern-recognition receptor (such as an RIG-I–like receptor) or a separate function of IRF3.

To characterize how IRF3 influences the inflammatory network, we analyzed a wide array of immune/inflammatory mediators in the sera of mice after they underwent CLP surgery. Computational modeling was performed to determine the connectivity and complexity of the network. The network exhibited a similar complexity in WT and IRF3-KO mice when all time points were considered (Fig. 3A, Table I). However, the tempo of the connections differed between the genotypes. In WT mice, the inflammatory network was highly connected early on (0–3 h), rapidly lost this coordination (3–6 h), and then recovered (6–21 h). In contrast, for IRF3-KO mice, the network was less connected early on (0–3 h) but showed sustained connectivity over time (3–6 and 6–21 h).

Additionally, the specific mediators that predominated the response were different in each genotype (Fig. 3B, Table I), and IL-6 formed different connections (Table II). For WT mice, we observed a “canonical” response in sepsis, dominated by inflammatory cytokines (TNF-α, IL-1β, and IL-6), chemokines (IP-10, KC, and MIG), and the Th17 mediator IL-17A at early time points, followed by induction of HMGB1 later on. In contrast, for IRF3-KO mice, there is a persistent chemokine response (MCP-1), suggesting the system is “stuck” in an early phase, never progressing to drive Th17 activation and downstream HMGB1 release. IL-6 can orchestrate a shift from neutrophil to monocyte recruitment during acute inflammation (37) and, when combined with TGF-β1, promotes a Th17 response (38). Prior reports suggest that IL-17A and its receptor exert a detrimental effect in adult and neonatal mouse sepsis models (39, 40), although this finding is not unequivocal (41). HMGB1 is a DAMP that has been reported to perpetuate the inflammatory response during sepsis (42). Hence, the typical program that IL-6 coordinates (which could include a Th17 response) is absent in IRF3-KO mice. Interestingly, IL-6 induces GM-CSF in IRF3-KO mice, which may enhance the function of monocytes in this strain. GM-CSF is currently being tested in clinical trials as a therapy to restore monocyte immunocompetence in patients with severe sepsis and septic shock; however, its role and efficacy are not yet fully elucidated (43). Our finding that serum TNF-α is an early driver of inflammation in CLP sepsis in WT mice concurs with a recent report by two of our coauthors, investigating the principal drivers of inflammation in LPS endotoxemia (31). Future studies will investigate how IRF3 influences the inflammatory/immune network in other body sites after CLP (including the peritoneal cavity, lymphoid, and visceral organs) and determine how this relates to immune cell trafficking and behavior.

Our data suggest that IRF3 substantially changes the network of immune/inflammatory mediators, particularly in regard to IL-6 (Fig. 3, Tables I, II). Because these mediators are predominantly produced by immune cells, we hypothesized that IRF3 exerts its role in leukocytes. Our data in bone marrow chimeras, however, refute this hypothesis. Chimeras with IRF3-KO leukocytes exhibited only a slight and insignificant survival advantage over counterparts with WT leukocytes (Fig. 4A). Moreover, the disease score, temperature, and serum inflammatory mediators were not substantially different between the chimeras with IRF3-KO versus WT leukocytes (Fig. 4B–F). In contrast, chimeras with an IRF3-KO stroma exhibited a considerable survival advantage over counterparts with a WT stroma (Fig. 5A). The disease score and hypothermia were attenuated in chimeras with an IRF3-KO stroma, and their levels of serum cytokines were reduced relative to chimeras with a WT stroma (Fig. 5B–F). These data are consistent with the notion that IRF3 acts in nonleukocytic cells to influence sepsis pathogenesis, altering the behavior of innate immune cells in a paracrine fashion. An alternative explanation for our findings is that bone marrow chimeras possess a radiation-resistant immune cell type (such as tissue resident macrophages) derived from the host, which contributes to the improved innate immune response during sepsis. However, our subsequent experiments (discussed below) suggest that a paracrine effect is more likely to be the correct explanation.

In this study, bone marrow chimeras with WT leukocytes and stroma (Figs. 4, 5) exhibited a greater mortality rate than intact WT mice that underwent CLP (Fig. 1). Concomitantly, we observed hypothermia in the groups of chimeras (Figs. 4, 5) but not the groups of intact mice that underwent CLP (Fig. 1). Several factors could explain why the chimeras got sicker than intact mice. First, the chimeras underwent CLP at a more advanced age than the intact mice because of the time required for donor bone marrow to engraft the host. Second, the studies in bone marrow chimeras were executed in the winter when we observed a higher mortality rate in this model. Third, a change in the person performing the surgeries may have affected the severity of the sepsis model. Finally, a recent report described that B6.SJL mice have a point mutation in the NCR1 gene, conferring resistance to CMV infection and increased susceptibility to influenza infection as a consequence of an enhanced IFN-γ signaling (44). We used B6.SJL mice as hosts (Fig. 4) and bone marrow donors (Fig. 5) to confirm leukocyte reconstitution in our bone marrow chimeras. It is plausible that the NCR point mutation could confer increased susceptibility to sepsis in our bone marrow chimeras with B6.SJL bone marrow (Fig. 5) relative to intact C57BL/6J mice.

Our data suggest that IRF3 exacerbates sepsis through leukocyte-extrinsic effects that increase IL-6 and alter the immune/inflammatory network. We set out to directly test if IRF3 influences the capacity of myeloid cells to produce IL-6 or if the host environment alters this function. In tissue culture, we observed similar IL-6 production in IRF3-KO versus WT macrophages treated with bacteria (Fig. 6A). These data suggest that IRF3-KO macrophages do not have a lower capacity to produce IL-6 per se. When WT monocytes were transferred into septic hosts, however, they produced more IL-6 if the host was WT versus IRF3-KO (Fig. 7). These data demonstrate that the WT versus IRF3-KO host environment alter the cytokine-producing capacity of myeloid cells located therein via paracrine effects.

Our prior report showed that IRF3-KO mice subject to CLP have a reduced bacterial burden in the blood and peritoneal lavage relative to WT counterparts (21). A recent report by Remick’s group determined that mice predicted to die of sepsis (based on high levels of serum IL-6) exhibit reduced peritoneal cell phagocytosis relative to mice predicted to live (32). This impairs bacterial clearance in mice predicted to die of sepsis (32). Among the mice that underwent CLP, we observed variable levels of phagocytosis in their peritoneal cells. Some mice exhibited a preserved (or even increased) percentage of cells undergoing phagocytosis, whereas other mice exhibited a reduced percentage of cells undergoing phagocytosis. Six out of sixteen WT mice exhibited reduced phagocytosis after CLP (Fig. 2A); in contrast, only one out of seventeen mice lacking IRF3 exhibited reduced phagocytosis. When animals of both genotypes were pooled, we observed an inverse correlation between the animal disease score and the phagocytic index of their peritoneal cells (Fig. 2B). Our finding that advanced sepsis impairs the ability of phagocytic cells to clear bacteria agrees with Remick’s prior report (32); however, the phagocytic defect that we observed had a different character from their report. Remick’s group observed a reduced degree of phagocytosis per cell (32), whereas we observed a reduced percentage of cells undergoing phagocytosis (Fig. 2). This disparity likely stems from methodological differences between our studies.

The phagocytic defect and higher bacterial load that we observed in WT versus IRF3-KO mice [described in this study and in our prior report (21)] could theoretically explain the higher cytokine levels observed in WT mice. However, our data suggest that IRF3 does not directly affect the phagocytic capacity of macrophages. We observed a similar phagocytic index in WT and IRF3-KO sham-operated animals (Fig. 2A). Additionally, using an in vitro culture system, we found that the phagocytic capacity of WT macrophages was at least as high as IRF3-KO counterparts (Fig. 6B). Further experiments are required to elucidate the interplay between phagocytosis and the altered immune/inflammatory network in WT versus IRF3-KO mice; these processes are likely to be intertwined with bidirectional feedback.

One caveat to our data interpretation is that the IRF3-KO mouse strain used in our study exhibits functional inactivation of the adjacent gene Bcl2-like 12 (Bcl2L12) (45). BCL2L12 plays a proapoptotic role in mouse embryonic fibroblasts subject to DNA damage by promoting cleavage of caspase-3 and -9 (45). However, the proapoptotic role of BCL2L12 appears to be limited, as it does not affect apoptosis in thymocytes, nor does it influence apoptosis induced by Fas receptor ligation in mouse embryonic fibroblasts (45). Notably, the Taniguchi laboratory recently generated a cleaner IRF3-KO mouse in which BCL2L12 is unaffected (33). These mice were protected from LPS endotoxemia (33), demonstrating that IRF3, rather than BCL2L12, mediates this effect.

Importantly the recent report by Taniguchi’s group demonstrates that IRF3 contributes to LPS endotoxemia through its presence in macrophages and dendritic cells (34). These results are dissimilar to our own and highlight fundamental differences in the animal models employed in our studies. The authors detected a significant amount of IFN-β (∼1000 pg/ml) in the serum of mice 2 h after LPS administration (34). Mice lacking IRF3 globally or in myeloid cells (IRF3f/f × LysMCre) or dendritic cells (IRF3f/f × CD11c-cre) showed attenuated IFN-β production paired with improved animal survival following LPS administration (34). In our own prior study, we detected high levels of IFN-β (∼68,000 pg/ml) in the serum of mice 5 h after i.v. administration of liposome:DNA (22). Mice lacking IRF3 globally showed attenuated IFN-β production paired with improved survival following liposome:DNA challenge (22). In contrast, in the CLP model used in this report, we did not detect IFN-β in the serum of mice at any time point tested (0, 3, 6, or 20 h postsurgery; data not shown). We speculate that i.v. administration of LPS or liposome:DNA induces macrophages and dendritic cells to secrete IFN-β, whereas CLP does not. This could be a consequence of the specific pathogen-associated molecular patterns and DAMPs released in each model as well as their anatomical location, absolute levels, and kinetics. Importantly, two prior studies compared the LPS and CLP models of sepsis and found that the dynamics of cytokine production are substantially different despite similar mortality rates (46, 47). LPS induces higher acute levels of inflammatory cytokines that rapidly decline, whereas CLP induces lower acute levels, but the cytokine response is more prolonged (46, 47).

Finally, akin to a prior report (33), we found that the IRF3-KO mice used in our study retained a significant amount of genomic material (30.77%) from the 129S1/Svlm strain. It would appear that this IRF3-KO strain has been exchanged between multiple laboratories under the mistaken assumption that the mice were fully backcrossed to C57BL/6J. In our own study, de novo backcrossing verified that the phenotype of sepsis resistance stems from the IRF3-KO locus and contaminating 129S1/Svlm alleles have little effect (Fig. 8). In contrast, in a prior report, the authors observed a phenotype of retroviral resistance that was caused by an Fv1 allele from the 129S1/Svlm strain rather than the IRF3-KO locus (33). Other investigators using this strain are advised to check the congenicity of their mice for scientific rigor.

In summary, this report demonstrates a role for IRF3 in stromal cells, contributing to bacterial sepsis in manner that is associated with distinct inflammatory programs associated with IL-6. Future studies will determine the cell type that mediates this effect and examine if IFN-β or other mediators play a mechanistic role in the process. Armed with knowledge of the correct target tissues, we could begin to develop strategies to inhibit this protein for therapeutic benefit in septic patients.

Disclosures

The authors have no financial conflicts of interest.

Acknowledgments

We thank Indika Mallawaarachchi and Alok Dwivedi for statistical support. We acknowledge Devender Kumar for technical assistance (completing additional experimental repeats for Figs. 4 and 5 and generating preliminary data for Fig. 7).

Footnotes

  • This work was supported by Texas Tech University Health Sciences Center El Paso startup funds (to W.E.W.), the Shock Society Research Fellowship for Early Career Investigators (to W.E.W.), and National Institutes of Health/National Institute of General Medical Sciences Grant RO1-GM107231-01A1 (to R.Z. and Y.V.).

  • Abbreviations used in this article:

    CLP
    cecal ligation and puncture
    DAMP
    damage-associated molecular pattern
    DyNA
    dynamic network analysis
    21G
    21-gauge
    IRF3
    IFN regulatory factor 3
    KC
    keratinocyte-derived cytokine
    KO
    knockout
    LRS
    lactated Ringer solution
    STING
    stimulator of IFN genes
    TRIF
    TIR domain–containing adapter-inducing IFN-β
    VEGF
    vascular endothelial growth factor
    WT
    wild-type.

  • Received February 21, 2019.
  • Accepted October 20, 2020.

References

  1. ↵
    1. Liu, V.,
    2. G. J. Escobar,
    3. J. D. Greene,
    4. J. Soule,
    5. A. Whippy,
    6. D. C. Angus,
    7. T. J. Iwashyna
    . 2014. Hospital deaths in patients with sepsis from 2 independent cohorts. JAMA 312: 90–92.
    OpenUrlCrossRefPubMed
  2. ↵
    1. Torio, C. M.,
    2. B. J. Moore
    . 2016. National inpatient hospital costs: the most expensive conditions by payer, 2013. HCUP statistical brief #204. In Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Agency for Healthcare Research and Quality, Rockville, MD.
  3. ↵
    1. Fingar, K.,
    2. R. Washington
    . 2015. Trends in hospital readmissions for four high-volume conditions, 2009–2013. HCUP statistical brief #196. In Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Agency for Healthcare Research and Quality, Rockville, MD.
  4. ↵
    1. Winters, B. D.,
    2. M. Eberlein,
    3. J. Leung,
    4. D. M. Needham,
    5. P. J. Pronovost,
    6. J. E. Sevransky
    . 2010. Long-term mortality and quality of life in sepsis: a systematic review. Crit. Care Med. 38: 1276–1283.
    OpenUrlCrossRefPubMed
  5. ↵
    1. van der Poll, T.,
    2. F. L. van de Veerdonk,
    3. B. P. Scicluna,
    4. M. G. Netea
    . 2017. The immunopathology of sepsis and potential therapeutic targets. Nat. Rev. Immunol. 17: 407–420.
    OpenUrlCrossRefPubMed
  6. ↵
    1. Mira, J. C.,
    2. S. C. Brakenridge,
    3. L. L. Moldawer,
    4. F. A. Moore
    . 2017. Persistent inflammation, immunosuppression and catabolism syndrome. Crit. Care Clin. 33: 245–258.
    OpenUrlCrossRefPubMed
  7. ↵
    1. Fisher, C. J. Jr..,
    2. J. M. Agosti,
    3. S. M. Opal,
    4. S. F. Lowry,
    5. R. A. Balk,
    6. J. C. Sadoff,
    7. E. Abraham,
    8. R. M. Schein,
    9. E. Benjamin
    . 1996. Treatment of septic shock with the tumor necrosis factor receptor:Fc fusion protein. The soluble TNF receptor sepsis study group. N. Engl. J. Med. 334: 1697–1702.
    OpenUrlCrossRefPubMed
  8. ↵
    1. Opal, S. M.,
    2. C. J. Fisher Jr..,
    3. J. F. Dhainaut,
    4. J. L. Vincent,
    5. R. Brase,
    6. S. F. Lowry,
    7. J. C. Sadoff,
    8. G. J. Slotman,
    9. H. Levy,
    10. R. A. Balk, et al
    . 1997. Confirmatory interleukin-1 receptor antagonist trial in severe sepsis: a phase III, randomized, double-blind, placebo-controlled, multicenter trial. The interleukin-1 receptor antagonist sepsis investigator group. Crit. Care Med. 25: 1115–1124.
    OpenUrlCrossRefPubMed
  9. ↵
    1. Grimaldi, D.,
    2. J. L. Vincent
    . 2017. Clinical trial research in focus: rethinking trials in sepsis. Lancet Respir. Med. 5: 610–611.
    OpenUrl
  10. ↵
    1. Francois, B.,
    2. R. Jeannet,
    3. T. Daix,
    4. A. H. Walton,
    5. M. S. Shotwell,
    6. J. Unsinger,
    7. G. Monneret,
    8. T. Rimmelé,
    9. T. Blood,
    10. M. Morre, et al
    . 2018. Interleukin-7 restores lymphocytes in septic shock: the IRIS-7 randomized clinical trial. JCI Insight 3: e98960.
    1. Berlot, G.,
    2. M. C. Vassallo,
    3. N. Busetto,
    4. M. Nieto Yabar,
    5. T. Istrati,
    6. S. Baronio,
    7. G. Quarantotto,
    8. M. Bixio,
    9. G. Barbati,
    10. R. Dattola, et al
    . 2018. Effects of the timing of administration of IgM- and IgA-enriched intravenous polyclonal immunoglobulins on the outcome of septic shock patients. [Published erratum appears in 2019 Ann. Intensive Care 9: 33.] Ann. Intensive Care 8: 122.
    OpenUrl
    1. Wu, J.,
    2. L. Zhou,
    3. J. Liu,
    4. G. Ma,
    5. Q. Kou,
    6. Z. He,
    7. J. Chen,
    8. B. Ou-Yang,
    9. M. Chen,
    10. Y. Li, et al
    . 2013. The efficacy of thymosin alpha 1 for severe sepsis (ETASS): a multicenter, single-blind, randomized and controlled trial. Crit. Care 17: R8.
    OpenUrlPubMed
    1. Meisel, C.,
    2. J. C. Schefold,
    3. R. Pschowski,
    4. T. Baumann,
    5. K. Hetzger,
    6. J. Gregor,
    7. S. Weber-Carstens,
    8. D. Hasper,
    9. D. Keh,
    10. H. Zuckermann, et al
    . 2009. Granulocyte-macrophage colony-stimulating factor to reverse sepsis-associated immunosuppression: a double-blind, randomized, placebo-controlled multicenter trial. Am. J. Respir. Crit. Care Med. 180: 640–648.
    OpenUrlCrossRefPubMed
  14. ↵
    1. Marlow, N.,
    2. T. Morris,
    3. P. Brocklehurst,
    4. R. Carr,
    5. F. Cowan,
    6. N. Patel,
    7. S. Petrou,
    8. M. Redshaw,
    9. N. Modi,
    10. C. J. Doré
    . 2015. A randomised trial of granulocyte-macrophage colony-stimulating factor for neonatal sepsis: childhood outcomes at 5 years. Arch. Dis. Child. Fetal Neonatal Ed. 100: F320–F326.
    OpenUrlAbstract/FREE Full Text
  15. ↵
    1. Poli-de-Figueiredo, L. F.,
    2. A. G. Garrido,
    3. N. Nakagawa,
    4. P. Sannomiya
    . 2008. Experimental models of sepsis and their clinical relevance. Shock 30(Suppl. 1): 53–59.
    OpenUrlCrossRefPubMed
  16. ↵
    1. Honda, K.,
    2. A. Takaoka,
    3. T. Taniguchi
    . 2006. Type I interferon [corrected] gene induction by the interferon regulatory factor family of transcription factors. Immunity 25: 349–360.
    OpenUrlCrossRefPubMed
  17. ↵
    1. Sato, M.,
    2. H. Suemori,
    3. N. Hata,
    4. M. Asagiri,
    5. K. Ogasawara,
    6. K. Nakao,
    7. T. Nakaya,
    8. M. Katsuki,
    9. S. Noguchi,
    10. N. Tanaka,
    11. T. Taniguchi
    . 2000. Distinct and essential roles of transcription factors IRF-3 and IRF-7 in response to viruses for IFN-alpha/beta gene induction. Immunity 13: 539–548.
    OpenUrlCrossRefPubMed
  18. ↵
    1. Chattopadhyay, S.,
    2. T. Kuzmanovic,
    3. Y. Zhang,
    4. J. L. Wetzel,
    5. G. C. Sen
    . 2016. Ubiquitination of the transcription factor IRF-3 activates RIPA, the apoptotic pathway that protects mice from viral pathogenesis. Immunity 44: 1151–1161.
    OpenUrlCrossRefPubMed
  19. ↵
    1. Wang, X. A.,
    2. R. Zhang,
    3. Z. G. She,
    4. X. F. Zhang,
    5. D. S. Jiang,
    6. T. Wang,
    7. L. Gao,
    8. W. Deng,
    9. S. M. Zhang,
    10. L. H. Zhu, et al
    . 2014. Interferon regulatory factor 3 constrains IKKβ/NF-κB signaling to alleviate hepatic steatosis and insulin resistance. Hepatology 59: 870–885.
    OpenUrlCrossRefPubMed
  20. ↵
    1. Kumari, M.,
    2. X. Wang,
    3. L. Lantier,
    4. A. Lyubetskaya,
    5. J. Eguchi,
    6. S. Kang,
    7. D. Tenen,
    8. H. C. Roh,
    9. X. Kong,
    10. L. Kazak, et al
    . 2016. IRF3 promotes adipose inflammation and insulin resistance and represses browning. J. Clin. Invest. 126: 2839–2854.
    OpenUrlPubMed
  21. ↵
    1. Walker, W. E.,
    2. A. T. Bozzi,
    3. D. R. Goldstein
    . 2012. IRF3 contributes to sepsis pathogenesis in the mouse cecal ligation and puncture model. J. Leukoc. Biol. 92: 1261–1268.
    OpenUrlCrossRefPubMed
  22. ↵
    1. Walker, W. E.,
    2. C. J. Booth,
    3. D. R. Goldstein
    . 2010. TLR9 and IRF3 cooperate to induce a systemic inflammatory response in mice injected with liposome:DNA. Mol. Ther. 18: 775–784.
    OpenUrlCrossRefPubMed
  23. ↵
    1. Heipertz, E. L.,
    2. J. Harper,
    3. W. E. Walker
    . 2017. STING and TRIF contribute to mouse sepsis, depending on severity of the disease model. Shock 47: 621–631.
    OpenUrlCrossRef
    1. Heisig, M.,
    2. N. M. Abraham,
    3. L. Liu,
    4. G. Neelakanta,
    5. S. Mattessich,
    6. H. Sultana,
    7. Z. Shang,
    8. J. M. Ansari,
    9. C. Killiam,
    10. W. Walker, et al
    . 2014. Antivirulence properties of an antifreeze protein. Cell Rep. 9: 417–424.
    OpenUrlCrossRef
  25. ↵
    1. Silver, A. C.,
    2. A. Arjona,
    3. W. E. Walker,
    4. E. Fikrig
    . 2012. The circadian clock controls toll-like receptor 9-mediated innate and adaptive immunity. Immunity 36: 251–261.
    OpenUrlCrossRefPubMed
  26. ↵
    1. Heipertz, E. L.,
    2. J. Harper,
    3. C. A. Lopez,
    4. E. Fikrig,
    5. M. E. Hughes,
    6. W. E. Walker
    . 2018. Circadian rhythms influence the severity of sepsis in mice via a TLR2-dependent, leukocyte-intrinsic mechanism. J. Immunol. 201: 193–201.
    OpenUrlAbstract/FREE Full Text
  27. ↵
    1. Walker, W. E.
    2018. Methods to study the innate immune response to sepsis. Methods Mol. Biol. 1717: 189–206.
    OpenUrl
  28. ↵
    1. Mi, Q.,
    2. G. Constantine,
    3. C. Ziraldo,
    4. A. Solovyev,
    5. A. Torres,
    6. R. Namas,
    7. T. Bentley,
    8. T. R. Billiar,
    9. R. Zamora,
    10. J. C. Puyana,
    11. Y. Vodovotz
    . 2011. A dynamic view of trauma/hemorrhage-induced inflammation in mice: principal drivers and networks. PLoS One 6: e19424.
    1. Namas, R. A.,
    2. Y. Vodovotz,
    3. K. Almahmoud,
    4. O. Abdul-Malak,
    5. A. Zaaqoq,
    6. R. Namas,
    7. Q. Mi,
    8. D. Barclay,
    9. B. Zuckerbraun,
    10. A. B. Peitzman, et al
    . 2016. Temporal patterns of circulating inflammation biomarker networks differentiate susceptibility to nosocomial infection following blunt trauma in humans. Ann. Surg. 263: 191–198.
    OpenUrl
    1. Zamora, R.,
    2. Y. Vodovotz,
    3. Q. Mi,
    4. D. Barclay,
    5. J. Yin,
    6. S. Horslen,
    7. D. Rudnick,
    8. K. M. Loomes,
    9. R. H. Squires
    . 2017. Data-driven modeling for precision medicine in pediatric acute liver failure. Mol. Med. 22: 821–829.
    OpenUrl
  31. ↵
    1. Zamora, R.,
    2. S. Korff,
    3. Q. Mi,
    4. D. Barclay,
    5. L. Schimunek,
    6. R. Zucca,
    7. X. D. Arsiwalla,
    8. R. L. Simmons,
    9. P. Verschure,
    10. T. R. Billiar,
    11. Y. Vodovotz
    . 2018. A computational analysis of dynamic, multi-organ inflammatory crosstalk induced by endotoxin in mice. PLoS Comput. Biol. 14: e1006582.
  32. ↵
    1. Chiswick, E. L.,
    2. J. R. Mella,
    3. J. Bernardo,
    4. D. G. Remick
    . 2015. Acute-phase deaths from murine polymicrobial sepsis are characterized by innate immune suppression rather than exhaustion. J. Immunol. 195: 3793–3802.
    OpenUrlAbstract/FREE Full Text
  33. ↵
    1. O’Connor, M. A.,
    2. W. R. Green
    . 2014. Use of IRF-3 and/or IRF-7 knockout mice to study viral pathogenesis: lessons from a murine retrovirus-induced AIDS model. J. Virol. 88: 2349–2353.
    OpenUrlAbstract/FREE Full Text
  34. ↵
    1. Yanai, H.,
    2. S. Chiba,
    3. S. Hangai,
    4. K. Kometani,
    5. A. Inoue,
    6. Y. Kimura,
    7. T. Abe,
    8. H. Kiyonari,
    9. J. Nishio,
    10. N. Taguchi-Atarashi, et al
    . 2018. Revisiting the role of IRF3 in inflammation and immunity by conditional and specifically targeted gene ablation in mice. Proc. Natl. Acad. Sci. USA 115: 5253–5258.
    OpenUrlAbstract/FREE Full Text
  35. ↵
    1. Sakaguchi, S.,
    2. H. Negishi,
    3. M. Asagiri,
    4. C. Nakajima,
    5. T. Mizutani,
    6. A. Takaoka,
    7. K. Honda,
    8. T. Taniguchi
    . 2003. Essential role of IRF-3 in lipopolysaccharide-induced interferon-beta gene expression and endotoxin shock. Biochem. Biophys. Res. Commun. 306: 860–866.
    OpenUrlCrossRefPubMed
  36. ↵
    1. Dang, O.,
    2. L. Navarro,
    3. M. David
    . 2004. Inhibition of lipopolysaccharide-induced interferon regulatory factor 3 activation and protection from septic shock by hydroxystilbenes. Shock 21: 470–475.
    OpenUrlCrossRefPubMed
  37. ↵
    1. Hurst, S. M.,
    2. T. S. Wilkinson,
    3. R. M. McLoughlin,
    4. S. Jones,
    5. S. Horiuchi,
    6. N. Yamamoto,
    7. S. Rose-John,
    8. G. M. Fuller,
    9. N. Topley,
    10. S. A. Jones
    . 2001. Il-6 and its soluble receptor orchestrate a temporal switch in the pattern of leukocyte recruitment seen during acute inflammation. Immunity 14: 705–714.
    OpenUrlCrossRefPubMed
  38. ↵
    1. Bettelli, E.,
    2. Y. Carrier,
    3. W. Gao,
    4. T. Korn,
    5. T. B. Strom,
    6. M. Oukka,
    7. H. L. Weiner,
    8. V. K. Kuchroo
    . 2006. Reciprocal developmental pathways for the generation of pathogenic effector TH17 and regulatory T cells. Nature 441: 235–238.
    OpenUrlCrossRefPubMed
  39. ↵
    1. Flierl, M. A.,
    2. D. Rittirsch,
    3. H. Gao,
    4. L. M. Hoesel,
    5. B. A. Nadeau,
    6. D. E. Day,
    7. F. S. Zetoune,
    8. J. V. Sarma,
    9. M. S. Huber-Lang,
    10. J. L. M. Ferrara,
    11. P. A. Ward
    . 2008. Adverse functions of IL-17A in experimental sepsis. FASEB J. 22: 2198–2205.
    OpenUrlCrossRefPubMed
  40. ↵
    1. Wynn, J. L.,
    2. C. S. Wilson,
    3. J. Hawiger,
    4. P. O. Scumpia,
    5. A. F. Marshall,
    6. J. H. Liu,
    7. I. Zharkikh,
    8. H. R. Wong,
    9. P. Lahni,
    10. J. T. Benjamin, et al
    . 2016. Targeting IL-17A attenuates neonatal sepsis mortality induced by IL-18. Proc. Natl. Acad. Sci. USA 113: E2627–E2635.
    OpenUrlAbstract/FREE Full Text
  41. ↵
    1. Freitas, A.,
    2. J. C. Alves-Filho,
    3. T. Victoni,
    4. T. Secher,
    5. H. P. Lemos,
    6. F. Sônego,
    7. F. Q. Cunha,
    8. B. Ryffel
    . 2009. IL-17 receptor signaling is required to control polymicrobial sepsis. J. Immunol. 182: 7846–7854.
    OpenUrlAbstract/FREE Full Text
  42. ↵
    1. Andersson, U.,
    2. K. J. Tracey
    . 2011. HMGB1 is a therapeutic target for sterile inflammation and infection. Annu. Rev. Immunol. 29: 139–162.
    OpenUrlCrossRefPubMed
  43. ↵
    1. Bo, L.,
    2. F. Wang,
    3. J. Zhu,
    4. J. Li,
    5. X. Deng
    . 2011. Granulocyte-colony stimulating factor (G-CSF) and granulocyte-macrophage colony stimulating factor (GM-CSF) for sepsis: a meta-analysis. Crit. Care 15: R58.
    OpenUrlCrossRefPubMed
  44. ↵
    1. Jang, Y.,
    2. Z. J. Gerbec,
    3. T. Won,
    4. B. Choi,
    5. A. Podsiad,
    6. B. B Moore,
    7. S. Malarkannan,
    8. Y. Laouar
    . 2018. Cutting edge: check your mice-a point mutation in the Ncr1 locus identified in CD45.1 congenic mice with consequences in mouse susceptibility to infection. J. Immunol. 200: 1982–1987.
    OpenUrlAbstract/FREE Full Text
  45. ↵
    1. Nakajima, A.,
    2. K. Nishimura,
    3. Y. Nakaima,
    4. T. Oh,
    5. S. Noguchi,
    6. T. Taniguchi,
    7. T. Tamura
    . 2009. Cell type-dependent proapoptotic role of Bcl2L12 revealed by a mutation concomitant with the disruption of the juxtaposed Irf3 gene. Proc. Natl. Acad. Sci. USA 106: 12448–12452.
    OpenUrlAbstract/FREE Full Text
  46. ↵
    1. Remick, D. G.,
    2. D. E. Newcomb,
    3. G. L. Bolgos,
    4. D. R. Call
    . 2000. Comparison of the mortality and inflammatory response of two models of sepsis: lipopolysaccharide vs. cecal ligation and puncture. Shock 13: 110–116.
    OpenUrlCrossRefPubMed
  47. ↵
    1. Seemann, S.,
    2. F. Zohles,
    3. A. Lupp
    . 2017. Comprehensive comparison of three different animal models for systemic inflammation. J. Biomed. Sci. 24: 60.
    OpenUrl
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools

Jump to section