The RIPK3 Scaffold Regulates Lung Inflammation During Pseudomonas Aeruginosa Pneumonia

Abstract

RIPK3 (receptor-interacting protein kinase 3) activity triggers cell death via necroptosis, whereas scaffold function supports protein binding and cytokine production. To determine if RIPK3 kinase or scaffold domains mediate pathology during Pseudomonas aeruginosa infection, control mice and those with deletion or mutation of RIPK3 and associated signaling partners were subjected to Pseudomonas pneumonia and followed for survival or killed for biologic assays. Murine immune cells were studied in vitro for Pseudomonas-induced cytokine production and cell death, and RIPK3 binding interactions were blocked with the viral inhibitor M45. Human tissue effects were assayed by infecting airway epithelial cells with Pseudomonas and measuring cytokine production after siRNA inhibition of RIPK3. Deletion of RIPK3 reduced inflammation and decreased animal mortality after Pseudomonas pneumonia. RIPK3 kinase inactivation did neither. In cell culture, RIPK3 was dispensable for cell killing by Pseudomonas and instead drove cytokine production that required the RIPK3 scaffold domain but not kinase activity. Blocking the RIP homotypic interaction motif (RHIM) with M45 reduced the inflammatory response to infection in vitro. Similarly, siRNA knockdown of RIPK3 decreased infection-triggered inflammation in human airway epithelial cells. Thus, the RIPK3 scaffold drives deleterious pulmonary inflammation and mortality in a relevant clinical model of Pseudomonas pneumonia. This process is distinct from kinase-mediated necroptosis, requiring only the RIPK3 RHIM. Inhibition of RHIM signaling is a potential strategy to reduce lung inflammation during infection.

Cells respond to infection by activating cell death pathways (1). These pluripotent networks both directly eliminate infected cells and trigger cell death-independent inflammation, influencing the nature and intensity of the host immune response. Central to death pathway signaling is RIP (receptor-interacting protein) RIPK3 (kinase 3), a protein that coordinates cell signaling outcomes through both an N-terminal kinase domain and a central scaffold region that facilitates protein binding, the RHIM (RIP homotypic interaction motif) (2). The activity of the kinase domain prompts necroptosis, as RIPK3 phosphorylates mixed lineage kinase domain-like pseudokinase (MLKL) to cause cell membrane rupture and death (3). The RIPK3 RHIM displays a more diverse function. Although RHIM-mediated binding to RIPK1 and other RHIM-expressing proteins supports necroptotic signaling, the RIPK3 RHIM also stimulates inflammatory gene transcription that occurs independently of RIPK3 kinase activity or necroptosis (4–8). RIPK3 scaffold effects are thus ambiguous, facilitating cell demise under some conditions while spurring prosurvival inflammatory responses in others.

The relevance of RIPK3 signaling outcomes to human disease has become increasingly clear as knowledge of its role in infectious and noninfectious pathologies has grown. The lungs of patients with chronic obstructive pulmonary disease exhibit necroptotic cell death mediated by RIPK3 kinase activity, and several bacterial species trigger cell death via RIPK3, inciting tissue damage and aiding the spread of infection (9, 10). Conversely, cytokine-induction by the RIPK3 RHIM appears host-protective, spurring tissue regeneration during colitis and driving beneficial neuroinflammation to combat West Nile virus infection (8, 11).

Given the broad implications of RIPK3 signaling and its documented contribution to pulmonary pathology (9, 12), we investigated its function in the setting of Pseudomonas aeruginosa pneumonia. Pseudomonas continues to generate significant morbidity and mortality in hospitalized patients, causing frequent ventilator-associated pneumonias and other hospital-acquired infections, with isolated strains often being multidrug-resistant (13, 14). Further, Pseudomonas remains a constant threat in the lungs of patients with cystic fibrosis. Infection with Pseudomonas increases rates of organ failure and death in children, and by adulthood, most patients with cystic fibrosis remain chronically infected, subjecting lung tissue to ongoing inflammation (15, 16). Despite the cytotoxic and inflammatory nature of Pseudomonas infection, the contribution of RIPK3 in mediating pathology or host defense remains undefined (17–19). We, therefore, used genetic approaches to determine the impact of RIPK3 signaling in vitro and during the infection of experimental animals.

Here, we find that the RIPK3 scaffold is a central driver of lung inflammation and mortality after pulmonary infection with Pseudomonas. Deletion of RIPK3 greatly improves pneumonia survival in mice and broadly lowers tissue cytokine concentrations, whereas isolated kinase inactivation offers no survival benefit and does not reduce inflammation. Inflammatory RIPK3 effects are instead dependent on its protein scaffold function, and interrupting RHIM interactions dampens cytokine production from infected murine immune cells and human airway epithelium. These findings lend new weight to the importance of RIPK3 in regulating inflammation, broadening the understanding of RIPK3 biology and identifying RHIM inhibition as a novel therapeutic approach to reducing lung inflammation and mortality during infection.

Methods

Detailed materials and methods are provided in the data supplement.

Animal Experiments

Wild-type (WT) and mutant (3, 6, 20) male and female C57BL/6J mice were bred and housed at Emory University. Mice were 8–12 weeks old and sex-matched for each experiment. Bacterial pneumonia was modeled via direct tracheal injection of P. aeruginosa as previously described (21), and animals were either followed for survival or killed at specified time points for analyses. All animal experiments were conducted in accordance with the guidelines of the Emory University Institutional Animal Care and Use Committee.

Cell Culture Experiments

Murine immune cells and fibroblasts were derived from mouse tissues and maintained in culture. Primary human airway epithelial cells were maintained in air–liquid interface as previously described (22). In vitro infection was performed by inoculation of cultured cell lines with P. aeruginosa for specified durations before the performance of experimental assays. The use of human cell lines was in accordance with Emory University Institutional Review Board guidelines.

Data Analysis

Statistical significance was determined using an unpaired Student’s t test for comparisons of parametric data and a Mann-Whitney test for nonparametric data. Data sets too small to undergo normality testing were subjected to nonparametric analyses. Multiple comparisons were subjected to Benjamini-Yekutieli correction. Log-rank analysis was performed for survival studies with Bonferroni correction for multiple comparisons as needed. P < 0.05 were considered significant and indicated as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. All statistical analyses were performed using Graphpad Prism 8 (Graphpad Software Inc.).

Results

RIPK3 Mediates Mortality from P. aeruginosa Pneumonia

To screen for RIPK3-dependent phenotypes during Pseudomonas pneumonia, we first performed multiple survival analyses on mice with deletion or mutation of RIPK3 and signaling partners RIPK1 and CASP8 (caspase-8). Classically, upstream regulator RIPK1 binds to RIPK3 via the RHIM, allowing RIPK1 to phosphorylate RIPK3, which in turn phosphorylates MLKL to cause necroptosis (23). Under normal conditions, proapoptotic CASP8 inhibits RIPK3 kinase activity, preventing necroptosis, and Casp8^−/− mice are nonviable because of uncontrolled RIPK3-mediated cell death (3, 23).

Overall, we found that loss of RIPK3 produced a pronounced survival advantage. Although Pseudomonas pneumonia and the resulting sepsis proved lethal in over 70% of WT control subjects, Ripk3^−/− mice were largely protected from death, even with increasing bacterial inoculum (Figures 1A and E1A in the data supplement). This protection proved independent of CASP8, as Ripk3^−/−Casp8^−/− mice and littermate Ripk3^−/−Casp8^+/− control subjects displayed similar survival to Ripk3^−/− mice. The effect of RIPK3 in sepsis did not appear to stem from RIPK1–RIPK3 phosphorylation, as demonstrated by our findings in kinase-inactive animals. Ripk1^K45A/K45A mice contain a point knock-in mutation (K45A) in the catalytic region of the Ripk1 gene. Though viable and fertile, these animals express a RIPK1 protein that lacks kinase activity, permitting experimental isolation of kinase-dependent events in the RIPK1–RIPK3 signaling axis (24). A similar mutation in RIPK3 (Ripk3^K51A/K51A) prevents RIPK3 kinase activity but does not interrupt RIPK3 binding functions (20). Importantly, neither Ripk1^K45A/K45A nor Ripk3^K51A/K51A mice displayed a survival advantage over WT control subjects (Figures 1A and 1B). Thus, whereas deletion of RIPK3 generates resistance to Pseudomonas, loss of kinase activity in either RIPK3 or RIPK1 does not, indicating the protection observed in Ripk3^−/− mice does not result from the absence of kinase-mediated effects like necroptosis but rather from a lack of other signaling functions driven by RIPK3. These findings imply RHIM interactions may impact outcomes during Pseudomonas infections, and we, therefore, also examined survival in mice lacking other RHIM-containing proteins. Beyond RIPK3, RHIMs are found on RIPK1, TRIF (TIR domain-containing adaptor-inducing interferon-β), and ZBP1 (Z-DNA binding protein 1) (4, 5). Ripk1^−/− mice are not viable, and neither Trif^−/− nor Zbp1^−/− strains displayed improved survival after pneumonia (Figures E1B and E1C) (25).

Figure 1.

RIPK3 (receptor-interacting protein kinase 3) mediates mortality from Pseudomonas aeruginosa pneumonia. (A) Log-rank survival analysis after P. aeruginosa pneumonia in mice of indicated genotypes (n = 17–20 per group, indicators of significance reference comparisons to wild type [WT] survival). (B) Survival analysis comparing WT and kinase-inactive Ripk3^K51A/K51A mice after Pseudomonas pneumonia (n = 9 per group). (C and D) Quantitative analysis of colony-forming units CFU of P. aeruginosa isolated from (C) lung tissue and (D) serum of septic WT and Ripk3^−/− mice at 24 hpi (n = 12–14 mice per group). (E and F) ELISA quantification of cytokines (E) IL-1β and (F) TNF in lung tissue isolated from WT and Ripk3^−/− mice at 24 hpi (n = 8–13 per group). (G) Survival analysis for WT and Ripk3^−/− mice treated with anti–IL-1β antibody after Pseudomonas pneumonia (n = 10–15 per group). Bar heights and bars with scatter plots represent mean ± SD. For all figures, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. CFU = colony-forming units; n.s. = not significant.

As WT and Ripk3^−/− survival curves began to diverge roughly 24 hours after infection (hpi), we sampled murine tissues at this time point to identify differences associated with infection survival. Lung tissue, peripheral blood, and splenic tissue contained similar concentrations of Pseudomonas, suggesting Ripk3^−/− mice were not inherently resistant to local or disseminated infection (Figures 1C, 1D, and E1D). Similar concentrations of bacteria were also identified in the lungs and blood of WT and Ripk3^−/− mice treated with an increased inoculation of Pseudomonas (Figures E1E and E1F). However, despite the equivalent bacterial load, Ripk3^−/− lungs contained significantly lower concentrations of key inflammatory cytokines IL-1β and TNF (Figures 1E and 1F). To evaluate the potential functional impact of cytokine suppression associated with loss of RIPK3, we treated septic WT animals with IL-1β antibody (26). Anti–IL-1β therapy rescued septic WT mice, consistent with previous reports of IL-1β inhibition reducing Pseudomonas-induced pathology and suggesting cytokine reduction associated with RIPK3 knockout may directly impact sepsis survival (27, 28) (Figure 1G).

Notably, Ripk3^−/− lungs contained similar populations of innate immune cells after infection compared with WT control subjects. Analysis of single-cell lung suspensions from infected WT and Ripk3^−/− mice revealed comparable numbers of macrophages, dendritic cells, neutrophils, and natural killer cells (Figures 2A–2E and E2). Septic WT and Ripk3^−/− lung tissue also contained similar numbers of inflammatory monocytes, though the expression of CD62L, a key mediator of leukocyte tissue infiltration (29), was greater in mice lacking RIPK3 (Figure E3). Furthermore, lungs from unmanipulated mice contained significantly lower numbers of dendritic cells and natural killer cells (Figures 2D and 2E). Thus, RIPK3 may impact baseline homeostasis of tissue-resident immune cells and eventual immune cell trafficking during infection. However, the largely similar composition of immune cells in septic WT and Ripk3^−/− lungs implies RIPK3 drives the host inflammatory response via additional mechanisms beyond immune cell recruitment.

Figure 2.

Innate immune cell profiles are similar in infected WT and Ripk3^−/− lung tissue. (A) Composition of innate immune cells in WT and Ripk3^−/− lungs at 24 hpi as determined by flow cytometry (n = 9–10 per group). (B–E) Flow cytometric quantification of macrophages, neutrophils, dendritic cells, and NK cells in unmanipulated and infected lungs (24 hpi) taken from WT and Ripk3^−/− animals (n = 5–13 per group). NK = natural killer.

RIPK3 Drives Lung Inflammation without Activity of Its Kinase Domain

To more generally characterize the impact of RIPK3 on the inflammatory response to Pseudomonas, lung tissue homogenates from infected animals were pooled and subjected to immunoblot cytokine arrays. At 12 hpi, WT and Ripk3^−/− lung tissues exhibited comparable concentrations of cytokines and chemokines, establishing that RIPK3 does not influence the initial local response to bacteria. However, by 24 hpi, WT control and Ripk3^−/− inflammatory profiles appeared remarkably different, with Ripk3^−/− samples presenting a widely suppressed pattern of cytokine release in response to Pseudomonas (Figures 3A–3C). Notably, relative reductions were again apparent in TNF and IL-1β. In contrast to these results, RIPK3 kinase inactivation did not reduce inflammation, with a comparison of Ripk3^K51A/K51A and additional WT control lung samples revealing broadly similar cytokine profiles at both 12 and 24 hpi (Figures 3D and 3E). Together, these data display the inflammatory consequences of RIPK3 signaling in vivo and further emphasize the importance of kinase-independent RIPK3 effects.

Figure 3.

RIPK3 drives lung inflammation without the activity of its kinase domain. (A–C) Heatmaps comparing relative cytokine expression as detected by immunoblot array in homogenized lung tissue from WT and Ripk3^−/− mice after (A) sham operation and (B) 12 hpi and (C) 24 hpi. (D and E) Similar heatmaps for WT and kinase-inactive Ripk3^K51A/K51A lung tissue at (D) 12 hpi and (E) 24 hpi. For each comparison, n = 3 per genotype. Lung homogenates were pooled from mice within each group with comparable bacterial titers, as shown in Figure 1(C). Right scale bars on heatmaps indicate the range of dot pixel intensity of each cytokine on the immunoblots for the compared groups.

RIPK3 Is Dispensable for Pseudomonas-induced Cell Death

Our results in kinase-inactive Ripk3^K51A/K51A mice suggest kinase-mediated necroptosis does not drive inflammation during Pseudomonas infection, but RIPK3 also participates in cell death pathways beyond necroptosis (20, 30), and it remains possible that the reduced cytokine concentrations seen in Ripk3^−/− animals were derived from altered cell death outcomes. We, therefore, evaluated the impact of RIPK3 on overall Pseudomonas cytotoxicity. Macrophages readily engulf Pseudomonas, which induces macrophage inflammation and death (31–33). Mouse bone marrow-derived macrophages were thus cultured and infected in vitro and analyzed for cytokine production and cell death in the presence of Pseudomonas. Lower concentrations of IL-1β were evident in infected Ripk3^−/− macrophages at 4 hpi and persisted through 15 hpi, confirming the antiinflammatory effect of RIPK3 deletion we observed in live animals (Figure 4A). We then assayed cell death and cell viability during infection by quantifying the inclusion of cell-impermeable dye and ATP concentration relative to uninfected cells. Cell permeability assays are nonspecific with regard to cell death mode, and assessment of ATP content permits analysis of cell toxicity that may occur without overt cell death (34). Both Sytox-inclusion and ATP assays indicated that RIPK3 did not impact macrophage killing by Pseudomonas, with infected WT and Ripk3^−/− macrophages showing similar uptake of cell-impermeable dye and a similar loss of ATP during infection (Figures 4B–4D). Thus, whereas the proinflammatory function of RIPK3 during Pseudomonas infection is clear, RIPK3 does not direct infection-triggered cell death.

Figure 4.

RIPK3 promotes inflammation but is dispensable for Pseudomonas-induced cell death. (A) ELISA quantification of IL-1β in cell-free supernatants from WT and Ripk3^−/− bone marrow-derived macrophages (BMDMs) infected with Pseudomonas (multiplicity of infection [MOI] = 25) at indicated time points (n = 2–6 for each genotype/time point). (B and C) Time course of WT and Ripk3^−/− BMDM viability measured by (B) inclusion of the cell-impermeable fluorescent dye Sytox green (n = 3 experiments per time point) and (C) loss of intracellular ATP as compared with uninfected, untreated control cells (n = 6 experiments per time point) when cells were treated with media (mock) or infected with Pseudomonas (P. aeruginosa, MOI = 25). (D) Representative snapshot IncuCyte images of mock- or Pseudomonas-infected BMDM of indicated genotypes at 4 hpi and 15 hpi (stamped scale bars indicated in images).

RIPK3 Regulates Macrophage Cytokine Production in Conjunction with CASP8

RIPK3 participates as an inflammatory mediator in multiple distinct signaling pathways. Inflammasomes are cytosolic protein complexes that respond to cell stimuli by activating inflammatory caspases, resulting in the eventual release of IL-1β and IL-18 and the possibility of cell death by pyroptosis (35). RIPK3 supports inflammasome activation and pro–IL-1β processing through kinase-independent binding interactions with RIPK1, which in turn binds CASP8, and some RIPK3-deficient cells display a reduced IL-1β response to inflammatory stimuli (36, 37). RIPK3 also facilitates activation of the proinflammatory transcription factor NF-κβ, again in a kinase-independent fashion, leading to upregulation of inflammatory genes and production of cytokines like TNF and IL-6 in a process independent of inflammasome activity (7, 11, 38, 39). We thus used additional genetic studies in macrophages to further dissect RIPK3 signaling mechanisms in the context of Pseudomonas infection.

Confirming our findings in murine lung tissue, deletion of RIPK3 in macrophages reduced IL-1β production triggered by Pseudomonas, whereas isolated kinase inactivation (Ripk3^K51A/K51A) did not (Figure 5A). This result supports a role for RIPK3 in infection-induced inflammasome activity while also implying that RIPK3-mediated IL-1β production is not associated with kinase-dependent necroptosis. Importantly, whereas IL-1β production was normal in kinase-inactive Ripk3^K51A/K51A cells, additional deletion of CASP8 (Casp8^−/−Ripk3^K51A/K51A) was as effective as deletion of RIPK3 in reducing IL-1β (Figure 5A). Thus, RIPK3 and CASP8 are each implicated in IL-1β release from infected macrophages, and removal of either effectively suppresses cytokine production. This finding is in keeping with previous work in dendritic cells in which RIPK3 and RIPK1 directly support CASP8 activation and cleavage of pro–IL-1β in response to bacterial LPS (36). Consistent with dual control of RIP kinases and CASP8 over IL-1β production, minimal cytokine release was observed in infected Casp8^−/−Ripk3^−/− or Ripk1^−/−Casp8^−/−Ripk3^−/− macrophages (Figure 5A).

Figure 5.

RIPK3 regulates macrophage cytokine production in conjunction with CASP8 (caspase-8). (A and B) Quantification of (A) IL-1β and (B) TNF in cell-free supernatants secreted from bone BMDM of indicated genotypes after mock infection (t = 0) and at 2 and 4 hpi with P. aeruginosa (MOI = 25, n ⩾ 3 experiments per group).

TNF release from Pseudomonas-infected macrophages was less dependent on RIPK3, as the initial response at 2 hpi was not reduced in either Ripk3^K51A/K51A or Ripk3^−/− cells (Figure 5B). Rather, early TNF production required CASP8, as evidenced by minimal cytokine production at 2 hpi in Casp8^−/−Ripk3^K51A/K51A, Casp8^−/−Ripk3^−/−, and Ripk1^−/−Casp8^−/−Ripk3^−/− macrophages (Figure 5B). Interestingly, however, TNF concentrations in Casp8^−/−Ripk3^K51A/K51A cells rose by 4 hpi and were no longer significantly different from WT control subjects. These cells lack CASP8 but express a kinase-inactive RIPK3 protein, and the delayed cytokine elevation exhibited suggests the requirement for CASP8 in TNF production is not absolute and may be supported by RIPK3 (Figure 5B). It should be noted, though, that significant redundancy exists in the elaboration of TNF responses, and TNF release in Casp8^−/−Ripk3^K51A/K51A cells may be attributable to other signaling mechanisms that do not require RIPK3 (39). Furthermore, these in vitro findings contrast with the reduced TNF concentrations seen in lung tissues from infected Ripk3^−/− mice (Figures 1F and 3C). Variable signaling mechanisms in different tissues may account for this discrepancy, as the inflammatory function of RIPK3 varies in different cell populations (11). Kinetics must also be considered, as the TNF response in Ripk3^−/− animals was similarly RIPK3-independent at 12 hpi and did not appear reduced until 24 hpi (Figures 3B and 3C). Because inflammatory effects of RIPK3 may be cell compartment-specific (7, 8), we also examined cytokine production from murine fibroblasts, though neither deletion of RIPK3 (Ripk3^−/−) nor CASP8 (in Casp8^−/−Ripk3^K51A/K51A or Casp8^−/−Ripk3^−/− cells) significantly impaired TNF release (data not shown). Taken together, these data illustrate the relevance of kinase-independent RIPK3–CASP8 signaling in the context of Pseudomonas infection.

RHIM Inhibition Reduces the Inflammatory Response to Pseudomonas

The RIPK3 RHIM facilitates cytokine production in LPS-treated dendritic cells, supporting TNF and IL-1β production together with IL-22 and IL-23 (7, 11, 36). To confirm the importance of the RHIM signaling during live bacterial infection, we investigated the possibility of inhibiting RHIM interactions to reduce Pseudomonas-induced inflammation. The cytomegalovirus M45-encoded protein, a viral inhibitor of RIP activation, prevents the formation of RHIM-dependent complications, inhibiting binding interactions of RIPK3 and other RHIM-expressing proteins (6, 40). In the presence of Pseudomonas, M45 effectively reduced TNF production in murine macrophages in comparison to a nonfunctional RHIM construct at 6 hpi (M45mutRHIM) (Figure 6A). This finding again suggests a component of macrophage TNF production is RHIM-dependent (36). Notably, RHIM-inhibition with M45 did not impact cell viability in the presence of Pseudomonas but predictably restricted the RHIM-dependent, kinase-mediated necroptosis known to be induced by treatment with TNF + zVAD-fmk (Figure 6B) (23). Therefore, Pseudomonas triggers cell death signaling that does not require RHIM interactions, whereas inflammatory effects are RHIM-dependent and cell death-independent.

Figure 6.

RHIM (RIP homotypic interaction motif) inhibition reduces the inflammatory response to Pseudomonas. (A) Quantification of murine TNF by ELISA in cell-free supernatants secreted from infected (MOI = 25) BMDM at 6 hpi in the presence of the viral RHIM-inhibiting protein M45 or inactive M45mutRHIM (n = 6–8 per group). (B) Murine BMDM viability as determined by ATP content in cells infected with Pseudomonas (P. aeruginosa, MOI = 25) or treated with TNF + zVAD-fmk in the presence of viral proteins M45 or M45mutRHIM compared with uninfected, mock-treated cells (n = 3 experiments per group). (C) Quantification of TNF released from infected human lung epithelial cells with siRNA knockdown of RIPK3 or RIPK1 at 6 hpi (n = 3–4 per group).

Given these findings in murine cells, we further assessed whether impaired RHIM signaling might suppress cytokine production in infected human cells. To that end, we used primary human basal airway epithelial cells in an air–liquid interface culture (41, 42). In these cells, siRNA knockdown of RIPK3 also reduced TNF production during infection with Pseudomonas (Figures 6C and E4A). Therefore, RIPK3 facilitates inflammatory signaling in epithelial cells in addition to immune cells, and nonimmune inflammatory signaling may account for the discrepancy in TNF effects observed in Ripk3^−/− animals (Figure 3) and macrophages (Figure 5). To address the potential impact of RIPK1, we also treated cells with RIPK1 siRNA, finding a nonsignificant trend toward reduced TNF release (Figure 6C). It should be noted that knockdown of RIPK3 produced an increase in the epithelial release of lactate dehydrogenase during infection compared with treatment with control siRNA (Figure E4B). Thus, suppression of RIPK3 in human epithelial cells may reduce inflammation while also sensitizing them to infection-triggered cell death (43). Together, these data demonstrate a conserved proinflammatory role for RHIM signaling in multiple cell types during Pseudomonas infection and suggest RHIM inhibition may limit septic inflammation.

Discussion

We have here demonstrated that RIPK3 mediates mortality and pulmonary inflammation in an animal model of P. aeruginosa pneumonia, and we identify the conserved proinflammatory function of RIPK3 in infected human airway epithelial cells. Pseudomonas engages kinase-independent, cell death-independent RIPK3 signaling, and our data suggest RHIM inhibition may represent a novel therapeutic approach to limit inflammation and potentially reduce mortality from infection.

Our findings are unique in implicating RIPK3 as a prime driver of airway inflammation occurring in direct association with septic mortality, and the data should be considered in light of the existing RIPK3 literature. The concept of RIPK3 propelling cytokine responses in animals without triggering cell death is not novel, as kinase-independent or death-independent RIPK3 signaling has previously been described in models of West Nile virus infection and chemical colitis (7, 8, 11). However, in these models, RIPK3 supports host-protective inflammation, whereas we find RIPK3 to be an amplifier of the deleterious inflammation that characterizes bacterial sepsis (44, 45). After cecal ligation and puncture, a widely used model of polymicrobial sepsis, loss of RIPK3 signaling has not consistently proved beneficial (46–48), though we document robust protection from septic death triggered by Pseudomonas pneumonia. Our data thus highlights the context-dependent nature of RIPK3-based inflammation. Along this line, we also describe a number of findings that contrast known RIPK3 biology in response to LPS. Loss of RIPK3 does not improve survival after injection of LPS (49, 50), though we demonstrate Ripk3^−/− mice to be remarkably resistant to death from Pseudomonas pneumonia. Moreover, whereas RIPK3 may not support macrophage cytokine responses after LPS treatment (11), our data indicate the opposite is true when macrophages are infected with live bacteria. RIPK3 kinase activity drives IL-1β production stimulated by LPS (36), yet we find kinase activity to be completely dispensable in the inflammatory response to active Pseudomonas infection, and the sustained bacterial inflammation in Ripk3^K51A/K51A mice stands in stark contrast to the reduced inflammatory profiles previously documented in these animals when treated with LPS (51). Our results thus add nuance and scope to the complexities of RIPK3 function in the host inflammatory response, in which knowledge about the role of programmed cell death mediators is actively developing. As RIPK3 has variable defined roles during infection with Staph aureus, Yersinia pestis, Salmonella, and other species (10, 52–54), our data further emphasize the dynamic nature of bacterial interaction with RIPK3 signaling and highlight the need to better understand regulatory mechanisms.

We find that during Pseudomonas infection, RIPK3 drives inflammation without inducing cell death. Though RIPK3-mediated necroptosis has been proposed as a far-reaching inflammatory mediator in multiple human pathologies (55), our findings suggest kinase-independent inflammatory signaling may be of greater clinical relevance. RIPK3-dependent cell death has indeed been detailed during viral infection, and it appears to limit cytomegalovirus infections by clearing cells through necroptosis (6, 56), though widespread activation of RIPK3-dependent cell death has not been documented during bacterial sepsis. Given documented RIPK3 involvement in inflammasome activity and NF-κβ signaling across multiple models (10, 38), the death-independent inflammatory function of RIPK3 and other cell death proteins appears to feature more prominently in the host response to infection. In line with this concept, strains of Escherichia coli have been found to secrete a protease that cleaves RHIM domains, illustrating the importance of these mediators in the evolution of host–pathogen interactions (57).

Despite their importance, how these proteins interact in cells and tissues during live infection remains incompletely resolved, though our results offer some insights. Although both Ripk3^−/− and Ripk3^−/− Casp8^−/− mice were protected from mortality during Pseudomonas infection, additional deletion of RIPK1 (in Ripk1^−/− Casp8^−/−Ripk3^−/− animals) negated any survival benefit over WT control subjects. Further, the isolated effect of RIPK3 on macrophage inflammation was less than the near-complete suppression observed in Ripk1^−/−Casp8^−/−Ripk3^−/− macrophages (Figure 5). Thus, the mortality of triple-knockout animals may reflect an inadequate inflammatory response to infection, one equally as deadly as the hyperinflammatory response observed in WT control subjects. Some degree of host cytokine response is clearly needed to combat infection, and our data imply that while loss of RIPK3 tempers septic inflammation and improves survival, combined loss of RIPK1 and RIPK3 may go a step too far and negatively blunt an appropriate reaction to infection. Beyond RIPK1–RIPK3 interactions, we also find evidence of RIPK3 joining CASP8 in mediating the production of IL-1β after Pseudomonas infection, supporting results previously documented in the context of LPS (36). Both RIPK1–RIPK3 binding and RIPK3-mediated IL-1β processing are RHIM-dependent, and our results using M45 to reduce infection-triggered inflammation argue such RHIM interactions are relevant to Pseudomonas-induced inflammation and TNF production as well. However, the impact of RHIM inhibition on RHIM-containing proteins beyond RIPK3 must be more thoroughly characterized. Our data propose combined effects from RIPK3, CASP8, and possibly RIPK1, and it is unclear if RHIM inhibition observed in macrophages reduces inflammation via impacting specifically RIPK3 or RIPK1. Although knockout of additional RHIM-containing adapters TRIF and ZBP1 did not improve overall animal survival, these mediators may nonetheless contribute to RIPK3-dependent inflammation during infection, and currently, such contributions remain unresolved.

Our findings come with a number of limitations. First, animal models of bacterial infection and the resulting inflammatory response can only go so far toward recapitulating findings in human patients (58), and though we demonstrate some effect of RIPK3 in human cells, substantial work remains to determine if inflammatory findings in mice bear a resemblance to those human infections. Also, though our findings suggest RIPK3 drives inflammation and mortality during Pseudomonas sepsis, the precise mechanisms by which inflammation leads to death are unclear. This uncertainty remains a broader issue in the clinical understanding of how infection kills, as modern ICUs keep patients alive long after the development of multiorgan failure, and unraveling the means by which discrete signaling mechanisms contribute to a redundant inflammatory milieu remains extremely challenging (44). Furthermore, RIPK3 demonstrates variability in signaling outcomes across cell compartments and tissues, and a better understanding of RIPK3 effects and interactions with other RHIM partners throughout the infected host is required. Though we documented the inflammatory function of RIPK3 in macrophages, this was done at a single time point, and it remains to be determined how RIPK3 signaling may impact cell function or migration at other time points. Effects on nonimmune cells must also be considered. Kinase-independent RIPK3 signaling impacts vascular permeability (59), indicating genome-wide deletion of RIPK3 may alter numerous nonimmune functions that could mediate mortality during sepsis. Thus, a more complete understanding of RIPK3 biology and how it may drive outcomes in sepsis is still required.

Conclusions

Despite these limitations, our results clearly demonstrate the impact of RIPK3 inflammatory signaling in a relevant clinical model of infection. Pseudomonas drives host inflammation via RIPK3 in a process uncoupled from its kinase activity and cell death function. Restricting RIPK3 signaling suppresses host inflammation and improves survival, and targeting RIPK3 and RHIM–RHIM interactions represents a novel therapeutic approach to reduce morbidity and mortality from infection.