Cutting Edge: TAK1 Safeguards Macrophages against Proinflammatory Cell Death

Inflammatory responses are the first line of defense that acts against invading or endogenous pathogens, and their strength and duration need to be moderately balanced for maintaining tissue homeostasis (1). Monocytes and macrophages in the mononuclear phagocyte system play central roles in the sequential process from the initiation to the resolution in inflammation (2, 3). These cells express various receptors on the cell surface and in the cytosol, which are known as pattern recognition receptors (4), sensing foreign pathogen-derived products named pathogen-associated molecular patterns and host-derived products released from injured cells named damage-associated molecular patterns. The best-known example of pattern recognition receptors is TLRs, and it is now appreciated that engagement of TLRs triggers inflammatory responses that are essential for the activation of both the innate and adaptive immune systems (5).

TGF-β–activated kinase 1 (TAK1), a member of the MAPKKK family encoded by Map3k7, is now known as a key mediator downstream of TLRs and also cytokine receptor systems such as the TNF receptor superfamily. Indeed, genetic findings have revealed that TAK1 is obligatory for activation of NF-κB and MAPK cascades in these receptor-mediated signaling pathways and bolsters the thorough development of the innate and adaptive immune systems (6–11). In this study, we report a novel function of TAK1 in macrophages. We revealed that TAK1 is essential to protect macrophages from a novel cell death machinery in TLR and TNFR signaling, consisting of active Caspase8-mediated cleavage of gasdermin D (GSDMD) as an executioner of pyroptosis. We also discovered that this proinflammatory macrophage death is responsible for enhanced susceptibility to endotoxic shock in mice lacking TAK1 in macrophages. These findings uncovered a previously unidentified role for TAK1 in macrophages that ensures the maintenance of innate immune system by restricting the cell death–associated inflammatory responses.

Map3k7^flox/flox, Myd88^−/−, Lyz2^Cre/+, R26^tdRFP/+, Tnf^−/−, Casp8^DED/+, Ripk3^−/−, and Ripk1^D138N/D138N mice, all of which are C57BL/6 background, have been previously described (11–17). Toll/IL-1R domain-containing adapter inducing IFN-β (TRIF)–deficient mice were generated in house by targeting the mouse Ticam1 gene using CRISPR/Cas9-based genome editing. Animal protocol for this research was approved by the Committee for Animal Experiments of Shinshu University.

To generate bone marrow–derived macrophages (BMDMs), bone marrow cells were harvested from femurs and tibias of 6- to 12-wk-old mice and cultured for 6 or 7 d in DMEM supplemented with 10% FBS, 2 mM l-glutamine, 100 U/ml penicillin, and 100 U/ml streptomycin and 20% L929 culture supernatants as a source of M-CSF on petri dishes. For cell death assay, BMDMs were stimulated with 500 ng/ml LPS from Escherichia coli O55:B5 (Sigma-Aldrich), 100 ng/ml Pam3CSK4 (InvivoGen), 100 μg/ml poly(I:C) (Sigma-Aldrich), 500 nM CpG ODN1668 (InvivoGen), and 10 ng/ml mouse TNF-α (PeproTech) in the absence or presence of 200 nM (5Z)-7-Oxozeaenol (Calbiochem), 50 μM Z-VAD-FMK (Medical and Biological Laboratories [MBL]), 50 μM Z-IETD-FMK (MBL), 50 μM Z-DEVD-FMK (MBL), 50 μM Ac-YVAD-cmk (Sigma-Aldrich), 10 μM GSK’872 (Calbiochem), and 20 μM Necrostatin-1s (Calbiochem) for 16 h, and the proportion of dead cells was assessed by measuring lactate dehydrogenase (LDH) in culture supernatants using LDH Cytotoxicity Detection Kit (TaKaRa) according to the manufacturer’s instruction. For the detection of active Caspase3 and Caspase8 by flow cytometry, FITC-conjugated DEVD-FMK and IETD-FMK (both from eBioscience) were used according to the manufacturer’s instructions. DRAQ7 (Beckman Coulter) were used for dead cell staining. Stained cells were acquired on FACSCanto II (BD Biosciences). The data were analyzed with Kaluza software (Beckman Coulter). For ELISA, 8 h after stimulation, culture supernatants were harvested and analyzed using TNF-α and IL-6 ELISA kits (eBioscience). For immunoblot analysis, cell lysates or a mixture of cell lysates and precipitates from culture supernatants in TNE-NS buffer (20 mM Tris-Cl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, and 0.1% SDS) supplemented with protease inhibitor mixture (Roche) and phosphatase inhibitor mixture (Nacalai Tesque) were analyzed by the following Abs: TAK1 (catalog [cat] no. 5206; Cell Signaling Technology), p-RIPK1 Ser166 (cat no. 31122; Cell Signaling Technology), p-IRF3 (cat no. 4947; Cell Signaling Technology), Caspase8 (cat ALX-804-447-C100; Enzo Life Sciences), Caspase1 (cat 14-9832-80; eBioscience), GSDMD (ab209845; Abcam), RIPK1 (cat 610459; BD Biosciences), IRF3 (cat 11312-1-AP; proteintech), Tubulin (cat M175-3; MBL), and GAPDH (cat M171-3; MBL). Real-time PCR was performed as described previously (11).

Eight- to ten-week-old mice were injected i.p. with LPS from E. coli O55:B5 (Sigma-Aldrich) at 40 mg/kg, and the survival rate was monitored every 12 h.

Paired and unpaired two-tailed t tests were used for the evaluation of statistical significance. Regarding survival rate shown from Kaplan–Meier survival plots, log-rank test was used for statistics. The p values <0.05 were considered significant. Statistical analyses were performed using GraphPad Prism software.

Because of the failure of efficient TAK1 deletion in macrophages but not neutrophils in Lyz2^Cre/+Map3k7^FL/FL mice (Supplemental Fig. 1A–G), we reasoned that bona fide TAK1-deficient macrophages in our Lyz2^Cre/+Map3k7^FL/FL mice might undergo cell death in a TNF-α–dependent mechanism in vivo because TAK1 deficiency increased the sensitivity to TNF-α–induced cell death (6, 18). To test this possibility, we generated myeloid-specific deletion of TAK1 on a TNF-α–deficient background (Lyz2^Cre/+Map3k7^FL/FLTnf^−/− mice). Expectedly, TAK1 was largely absent in macrophages from Lyz2^Cre/+Map3k7^FL/FLTnf^−/− mice (Supplemental Fig. 1H, 1I), suggesting a crucial role of TAK1 for macrophage survival in TNFR signaling.

We next investigated the biological function of TAK1 in macrophages using Lyz2^Cre/+Map3k7^FL/FLTnf^−/− mice (referred to hereafter as Map3k7^ΔMTnf^−/− mice). Surprisingly, we observed that TAK1-deficient BMDMs from Map3k7^ΔMTnf^−/− mice underwent profound macrophage death by treatment with LPS, a TLR4 agonist, and poly(I:C), a TLR3 agonist as well as TNF-α (Fig. 1A). In addition, pharmacological inhibition of TAK1 induced cell death of TNF-α–deficient BMDMs only upon stimulation with LPS, poly(I:C), and TNF-α (Fig. 1B). We found that TAK1-deficient BMDMs produced detectable levels of active forms of both Caspase8 and Caspase3 as well as DRAQ7 nuclear staining in response to the indicated stimulants (Fig. 1C), suggesting that TAK1-deficient macrophages seem to die by apoptotic cell death. Immunoblot analysis showed that TAK1-deficient BMDMs generated active Caspase8 (p18) in response to the indicated TLR agonists as well as TNF-α, consistent with the above results (Fig. 1C), whereas phosphorylation of RIPK1 S166, which is responsible for necroptosis (19), was hardly detectable (Fig. 1D–F). However, the agonist-induced phosphorylation was clearly observed only in the presence of Z-VAD-FMK, a pan-caspase inhibitor (Fig. 1D–F). These data indicated that TAK1-deficient macrophages did not die by necroptosis. Yersinia pestis is a pathogenic bacterium that causes plague and is known to induce macrophage cell death upon infection by disturbing NF-κB and MAPK cascades and then by driving Caspase8 activation, which is definitely required for host innate immune responses (20–24). More recently, it has been reported that macrophages underwent cell death by triggering Caspase8-dependent cleavage of GSDMD and GSDME, both of which belong to a gasdermin family, by TAK1 blockade during Yersinia infection (25, 26). To confirm the mechanism using our genetic tools, we next investigated the status of GSDMD protein in TAK1-deficient macrophages upon TLR or TNFRI stimulation. Interestingly, in contrast to TAK1-sufficient BMDMs from Tnf^−/− mice, TAK1-deficient BMDMs showed GSDMD processing and the generation of active Caspase1 (p20) in response to LPS, poly(I:C), and TNF-α, and importantly, they were highly dependent on enzymatic activities of Caspase8, 3, and 1 (Fig. 1G). Given that the cleaved GSDMD protein is pivotal for the induction of pyroptosis (27), our data revealed the presence of a novel pyroptotic-like cell death machinery constituted by Caspase8/3/1-dependent GSDMD processing, which must be strictly regulated by TAK1.

We next sought to explore specific molecule(s) responsible for cell death in TAK1-deficient macrophages. We established Map3k7^ΔMMyd88^−/−Tnf^−/− mice and Map3k7^ΔMTicam1^−/−Tnf^−/− mice (Supplemental Fig. 2A–C), respectively, and analyzed their BMDMs. Interestingly, both TAK1- and TRIF-deficient but not both TAK1- and MyD88-deficient BMDMs showed a remarkable protection from cell death upon stimulation with the TLR agonists but not TNF-α (Fig. 2A, 2B). Despite the fact that Caspase8 is active in TAK1-deficient macrophages upon TLR or TNFRI stimulation, TAK1-deficient BMDMs underwent cell death in response to the indicated agonists in the presence of Z-VAD-FMK (Fig. 2C), presumably because of RIPK3-mediated necroptosis as described previously (28). As expected, macrophage death by TAK1 deficiency was dramatically blocked only when both Z-VAD-FMK and GSK’872, a RIPK3 inhibitor, were applied together (Fig. 2C). Furthermore, we confirmed this issue by investigating BMDMs from Lyz2^Cre/+Map3k7^FL/FL mice deficient for both Caspase8 and RIPK3 (hereafter as Map3k7^ΔMCasp8^DED/DEDRipk3^−/− mice) (Fig. 2D, Supplemental Fig. 2D, 2E). Several studies have described that pharmacological inhibition of TAK1 promotes TNF-α–induced active Caspase8-mediated cell death dependent on the kinase activity of RIPK1 (25, 26, 29–31). Interestingly, our extended analysis using genetic and pharmacological approaches provided evidence that the absence of RIPK1 kinase activity appeared to block cell death partially in TAK1-deficient BMDMs, but its effect was modest and insufficient compared with TRIF deficiency in TAK1-deficient BMDMs (Fig. 2E, 2F, Supplemental Fig. 2F–H), suggesting the presence of other mechanism(s) independent of RIPK1 kinase activity. Taken together, our data demonstrated that TAK1 in macrophages is crucial for the prevention of TLR/TRIF-induced active Caspase8/GSDMD-mediated cell death.

To investigate the role of TAK1 in macrophages for cytokine and chemokine expression in response to TLR agonists, we used macrophages from Map3k7^ΔMCasp8^DED/DEDRipk3^−/− mice because of their resistance to TLR-induced cell death (Fig. 2D). Whereas Casp8^DED/DED mice were embryonic lethal as described previously (14), both Casp8^DED/DEDRipk3^−/− and Map3k7^ΔMCasp8^DED/DEDRipk3^−/− mice were viable. Similarly to previous reports (32, 33), Casp8^DED/DEDRipk3^−/− mice developed an autoimmune lymphoproliferative syndrome–like phenotype with age, exemplified by the appearance of splenomegaly and lymphadenopathy (Supplemental Fig. 3A). TAK1-sufficient BMDMs from Casp8^DED/DEDRipk3^−/− mice as a control exhibited increased protein expressions of TNF-α and IL-6 in response to all TLR agonists we examined, even though the amount of cytokine was significantly decreased when compared with RIPK3-deficient BMDMs, akin to a previous report (22) (Supplemental Fig. 3B). Importantly, the cytokine expressions were markedly impaired in TAK1/Caspase8/RIPK3 triple mutant BMDMs (Fig. 3A, 3B). Next, to determine the contribution of TAK1 in the TRIF-dependent pathway, we compared the gene expression of Ifnb1, Ccl5, and Cxcl10. Real-time PCR analysis showed a slight but significant decrease of expression of the TRIF-dependent genes by LPS and poly(I:C) stimulation in TAK1-deficient BMDMs (Fig. 3C–E). Intriguingly, biochemical analysis revealed almost equal levels of phosphorylation of IRF3 in both types of BMDMs upon TLR3 and 4 ligations (Fig. 3F). Taken together, our data presented evidence that TAK1 in macrophages is required for the expression of proinflammatory cytokines and, to a lesser extent, IFN-β and chemokines in TLR signaling, whereas TAK1 is less involved in TLR/TRIF-mediated activation of IRF3.

We finally sought to define the impact of TAK1 deficiency in macrophages in TLR-driven inflammatory responses by using an experimental model of sepsis. When treated with a lethal dose of LPS, we observed that Tnf^−/− mice were resistant to LPS-induced endotoxic shock as expected (Fig. 4A), confirming the crucial role of TNF-α for the symptom. Surprisingly, we found that Map3k7^ΔMTnf^−/− mice were highly susceptible to LPS challenge (Fig. 4A). This result emphasizes that TAK1 deficiency in macrophages facilitates proinflammatory cell death via LPS-induced TLR4/TRIF/Caspase8/GSDMD signaling axis, resulting in enhanced lethality. Next, we compared the susceptibility of LPS-induced endotoxic shock between Casp8^DED/DEDRipk3^−/− and Map3k7^ΔMCasp8^DED/DEDRipk3^−/− mice. When treated with a lethal dose of LPS, we observed that all Map3k7^FL/FL as a control and Casp8^DED/DEDRipk3 ^−/− mice were sensitive to LPS-induced endotoxic shock. In sharp contrast, Map3k7^ΔM Casp8^DED/DEDRipk3^−/− mice were more resistant to LPS-induced endotoxic shock (Fig. 4B). The data indicated that even in the failure of some cell death machinery, proinflammatory cytokine expression from macrophages induced by LPS exposure, which needs activation of NF-κB and AP-1 via TAK1, is also responsible for the elicitation of LPS/TLR4-driven fatal septic shock. Collectively, these in vivo data revealed that TAK1 in macrophages encompasses more pleiotropic function as both a positive and negative regulator of the host inflammatory responses.

The present study aimed to characterize an unidentified role of TAK1 in macrophages. We revealed not only the presence of a novel proinflammatory cell death machinery dependent of Caspase8 but not RIPK3 in macrophages but also that TAK1 can limits the cell death signaling, providing a new mechanistic insight into TAK1 function in TLR signaling. It is surprising that TAK1 deficiency in macrophages is linked to Caspase8/3/1-dependent cleavage of GSDMD, an executioner of pyroptosis (27). Hence, it is plausible that TAK1-deficient macrophages die by pyroptosis upon TLR or TNFRI stimulation. Meanwhile, given that a growing body of evidence has implicated both Caspase8 and Caspase3 as the crucial factors for apoptosis, we cannot rule out the possibility that cell death in TAK1-deficient macrophages is because of the mixed phenotypic outcomes by both pyroptosis and apoptosis. At present, it remains unknown whether this is a highly specific mechanism for macrophages or not. But it should be noted that expression of GSDMD and pyroptosis are not restricted to macrophages. The same mechanism may exist in other types of TAK1-deficient cells. Additionally, it is noteworthy that TRIF plays multifaceted roles for TLR-induced biological responses including cell death, which are often subject to the activation status of regulatory molecules like TAK1 and Caspase8. This important feature also raises the question of how TAK1 controls TRIF-mediated signaling not to be directed to Caspase8/GSDMD-dependent proinflammatory cell death pathway. Future detailed analysis will be needed to help elucidate the precise mechanisms. From our in vivo study, we provided evidence that the unique cell death machinery caused by TAK1 deficiency in macrophages is associated with systemic inflammation. Meanwhile, a number of mouse studies with Yersinia infection have demonstrated the importance of Caspase8-mediated cell death pathway for the protection from pathogenic bacterial invasion (20–24). Consequently, the novel proinflammatory cell death pathway represents a double-edged sword in host defense.

In summary, our findings not only provided a striking feature of TAK1 to act as a signaling checkpoint to safeguard macrophages against TLR-induced proinflammatory cell death, which is a prerequisite for the proper control of inflammation and the maintenance of innate immune homeostasis, but also will shed a new light on effective strategies for the treatment of inflammatory diseases.

We thank both M. Takamoto and S. Kakuta (Shinshu University) for providing Tnf^−/− mice, both K. Sakamaki and S. Yonehara (Kyoto University) for providing Casp8^DED/+ mice, V. Dixit (Genentech) for providing Ripk3^−/− and Ripk1^D138N/D138N mice, and S. Yamasaki (Osaka University) for distributing R26^tdRFP/+ mice.

The authors have no financial conflicts of interest.