TAK1 Limits Death Receptor Fas-Induced Proinflammatory Cell Death in Macrophages Free

This article is featured in Top Reads, p.

The TNF receptor (TNFR) superfamily forms a large group of cytokine receptors, all of which contribute to exertion of diverse biological functions, and, for this reason, some of them now serve as biomarkers and therapeutic targets (1, 2). Among them, the death receptor family is well known to induce programmed cell death, essential for embryogenesis and adult organ homeostasis. Fas (also known as CD95), the best characterized member of the death receptor family, plays a central role in immune homeostasis. A molecular mechanism by which Fas signaling induces apoptosis, a form of programmed cell death, has been well described (3). Engagement of Fas with FasL, a specific ligand for Fas, facilitates assembly of a complex of proteins, termed the “death-inducing signaling complex,” at the cytoplasmic tail of Fas (4). The Fas death-inducing signaling complex is composed of different death domain– and death effector domain–containing proteins: Fas, Fas-associated death domain, caspase-8, and c-FLIP. Its formation leads to autoproteolytic processing of caspase-8, and then the ensuing active caspase-8 dimerization enables initiation of apoptosis via processing of target proteins, including caspase-3 (4, 5). Now it is widely appreciated that failure of Fas-induced cell death causes serious immune dysregulation. In fact, humans with genetic mutations in Fas, FasL, caspase-8, and caspase-10 develop autoimmune lymphoproliferative syndrome, which is mainly characterized by accumulation of TCRαβ⁺CD3⁺CD4^–CD8^– T cells, lymphadenopathy, splenomegaly, and autoimmunity, and mice with naturally occurring mutations in Fas (lpr mice) and FasL (gld mice) also show analogous phenotypes as in the case of human patients (6, 7). However, it has hitherto remained unknown whether a large excess of Fas-induced cell death gives rise to the immunological and pathophysiological impacts.

TGF-β–activated kinase 1 (TAK1), a member of the MAPK kinase kinase family encoded by Map3k7, participates in the various immunoreceptor-mediated signal transductions. It mainly functions as an important signaling factor to activate the MAPK and inhibitor of NF-κB kinase/NF-κB pathways, both of which are important to evoke innate and adaptive immune responses (8–11). We and other groups previously demonstrated that macrophages deficient in TAK1 or macrophages subjected to pharmacological inhibition of TAK1 underwent massive cell death upon TNFR1 or TLR stimulation, and, remarkably, these works highlighted the presence of novel cell death machinery dependent on the gasdermin family, executioners of pyroptosis, operating in downstream pathways of caspase-8 (12–16). The findings not only identified a novel aspect of TAK1 function to restrain the appearance of such proinflammatory cell death but allowed us to consider the following two possibilities. The first is whether Fas signaling in macrophages triggers caspase-8–dependent cell death, followed by inflammation, and the second is whether and how TAK1 is associated with the Fas signaling pathway.

Here we show that TAK1 potentially functions as a negative regulator of Fas signaling in macrophages. Mouse primary macrophages underwent cell death only when engaged with high concentrations of FasL and displayed proteolytic cleavage of gasdermin D (GSDMD) and GSDME, members of the gasdermin family, in a manner dependent on caspase enzymatic activity. Remarkably, TAK1-deficient macrophages exhibited high sensitivity to even low concentrations of FasL, which is attributed to increased kinase activity of receptor-interacting serine/threonine-protein kinase 1 (RIPK1). Surprisingly, mice deficient in TAK1 in macrophages spontaneously developed tissue inflammation, indicative of autoinflammation, and, importantly, the symptom was markedly diminished by crossing to mice harboring a catalytically inactive RIPK1. These in vivo data also provided novel insight into the possibility of perturbation of immune homeostasis driven by aberrant cell death.

Map3k7^flox/flox or Map3k7^flox/–, Lyz2^Cre/+, Tnf^−/−, Ripk1^D138N/D138N, and Ticam1^−/− mice, all of which are on a C57BL/6 background, have been described previously (8, 15, 17). All the mice were maintained at the Shinshu University animal facility under specific pathogen-free (SPF) conditions. The animal protocol for this research was approved by the committee for animal experiments of Shinshu University.

Generation of mouse bone marrow–derived macrophages (BMDMs) was performed according to a protocol described previously (15). The BMDMs were cultured overnight, and the culture medium was replaced with Opti-MEM (Thermo Fisher), followed by treatment with the indicated cytokines. For the cell death assay, BMDMs were seeded at 5 × 10⁴ cells per well in 96-well plates and were treated with FasL (MegaFasL, Adipogen). Six hours after stimulation, the proportion of dead cells was assessed by measuring lactate dehydrogenase (LDH) in culture supernatants using the Cytotoxicity LDH Assay Kit-WST (Dojindo) according to the manufacturer’s instructions. For immunoblot analysis, BMDMs were seeded at 1–1.5 × 10⁶ cells per well in 6-well plates and were treated with FasL or TNF-α (PeproTech) in the absence or presence of 50 μM Z-VAD-FMK (Selleck). To prepare cell lysates, BMDMs were extensively washed with ice-cold 1× PBS and were lysed with TNE-NS lysis buffer (20 mM Tris-Cl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Nonidet P-40, and 0.1% SDS) supplemented with a protease inhibitor mixture (Roche) and a phosphatase inhibitor mixture (nacalai tesque). For detection of phosphorylated mixed lineage kinase domain-like protein (MLKL), BMDMs were lysed with 1× SDS sample buffer, and cell lysates after sonication were boiled. Proteins in cell culture supernatants were precipitated with methanol/chloroform for concentration and were suspended in 1× SDS sample buffer. Crude proteins were subjected to immunoblotting using specific Abs to detect target proteins. Specific proteins were visualized with Immobilon Chemiluminescent HRP Substrate (Millipore). The Abs used in this study are as follows: p-RIPK1(Ser166) (catalog no. 31122, Cell Signaling Technology), cleaved caspase-8 (Asp387) (catalog no. 9429, Cell Signaling Technology), c-FLIP (catalog no. 56343, Cell Signaling Technology), caspase-3 (catalog no. 9662, Cell Signaling Technology), poly(ADP-ribose) polymerase (PARP; catalog no. 9542, Cell Signaling Technology), TAK1 (catalog no. 5206, Cell Signaling Technology), cleaved caspase-1 (Asp296) (catalog no. 89332, Cell Signaling Technology), p-MLKL (Ser345) (catalog no. 37333, Cell Signaling Technology), p-IκBα (Ser32/36) (catalog no. 9246, Cell Signaling Technology), phosphorylated stress-activated protein kinase/JNK (Thr183/Tyr185) (catalog no. 9251, Cell Signaling Technology), stress-activated protein kinase/JNK (catalog no. 9252, Cell Signaling Technology), p-p38 (Thr180/Tyr182) (catalog no. 9211, Cell Signaling Technology), IκBα (sc-371, Santa Cruz Biotechnology), p38α (sc-81621, Santa Cruz), GSDMD (ab209845, Abcam), GSDME (ab215191, Abcam), RIPK1 (catalog no. 610459, BD Biosciences), caspase-8 (cat ALX-804-447-C100, Enzo), MLKL (SAB1302339, Sigma-Aldrich), high mobility group box 1 (HMGB1; catalog no. 651402, BioLegend), and GAPDH (catalog no. M171-3, MBL).

Six- to 10-wk-old mice were used to isolate cells from each tissue. To collect cells in the peritoneal cavity, mice were injected and extensively washed with 5 ml of 1× PBS/2% FBS in the cavity. Peritoneal cells were sampled and used for experiments. To collect cells from the liver, lung, and visceral adipose tissue, the isolated tissues were chopped into little pieces with scalpels and were enzymatically digested for 30 min at 37°C with gentle shaking in RPMI 1640 supplemented with 0.5 mg/ml collagenase IV (Sigma-Aldrich) and 50 U/ml DNase I (Wako). Preparation of colonic lamina propria cells were performed as described previously (18). Briefly, the isolated colon was cut and opened longitudinally. After extensively washing out luminal contents with 1× PBS, the tissue was incubated for 30 min at 37°C with vigorous shaking in 1× PBS (calcium and magnesium free) supplemented with 5 mM EDTA and 2 mM DTT to remove the epithelial cell layer. The remaining colon was chopped into little pieces with scalpels, and these pieces were enzymatically digested for 30 min at 37°C with gentle shaking in RPMI 1640 supplemented with 2% FBS, 0.5 mg/ml collagenase IV, and 50 U/ml DNase I. Discrete cells from each tissue after the treatment were extensively washed with 1× PBS and suspended in 1× PBS/2% FBS, then stored at 4°C.

Single-cell suspensions prepared from each tissue were used for Ab staining. The following fluorochrome-conjugated Abs were used in this study: CD11b (M1/70), CD11c (N418), CD64 (X54-5/7.1), CD301a (LOM-8.7), CX3CR1 (SA011F11), I-A/I-E (MHC class II [MHC II], M5/114.15.2), Ly6C (HK1.4), Ly6G (1A8), F4/80 (BM8), and Tim4 (RMT4-54), all from BioLegend; Siglec-F (E50-24400) and CD16/32 (2.4G2) from BD Biosciences; and CD45 (30-F11) from eBioscience. Propidium iodide (Sigma-Aldrich) was used for dead cell staining. Stained cells were assessed by flow cytometry using the FACSCelesta device (BD Biosciences), and the data were analyzed with Kaluza software (Beckman Coulter).

For histopathological examinations, the livers from 6- to 10-wk-old mice were fixed by immersion in 10% buffered formalin and embedded in paraffin. Tissue sections with 4-μm thickness were stained with H&E. For serological analysis, blood was sampled from the retro-orbital sinus. Sera were collected by centrifugation and were assessed by measurement of aspartate aminotransferase, alanine aminotransferase, and LDH. As criteria for peritonitis, we evaluated the disappearance of peritoneal macrophages, the infiltration of inflammatory cells (neutrophils and monocytes), and MHC II expression in large peritoneal macrophages.

An unpaired two-tailed t test was used for evaluation of statistical significance. The p Values <0.05 were considered significant. Statistical analyses were performed using GraphPad Prism software.

We previously demonstrated that TAK1-deficient macrophages evoked cell death in response to TNF-α and TLR ligands, indicating that TAK1 is necessary for the maintenance of macrophage survival in TNFR1 and TLR signaling (15). However, a lot of research has demonstrated that Fas signaling can induce cell death in various types of cells, even in the presence of TAK1. As expected, ligation of Fas with recombinant hexameric FasL (19) induced cell death in C57BL/6-derived mouse BMDMs, but, curiously, higher concentrations of FasL were required to achieve this despite its potent cytotoxic activity (Fig. 1A). In addition, immunoblot analysis showed Fas-induced proteolytic cleavage of caspase-8, caspase-3, and PARP in mouse BMDMs, all of which play a critical role in inducing apoptosis. Interestingly, a minor but detectable level of Ser166 phosphorylation of RIPK1, an autophosphorylation site essential for modulation of its kinase activity (20, 21), was also observed (Fig. 1B). Surprisingly, upon ligation with FasL, proteolytic cleavage of the gasdermin family members GSDMD and GSDME was detected, which was almost completely blocked by pretreatment with Z-VAD-FMK, a pan-caspase inhibitor (Fig. 1B, 1C). This means that the Fas-induced gasdermin family processing takes place in a caspase enzymatic activity–dependent manner. Given that these unique characteristics resemble the former reports (12–16) and gasdermin processing is a well-defined feature for the induction of pyroptosis (22), our data demonstrate that Fas signaling possesses an ability to trigger pyroptosis in mouse macrophages, even though the strength of the Fas signal seems usually to be suppressed.

Because it is unknown whether and how TAK1 is associated with the Fas-mediated signaling pathway, we sought to explore a potential role of TAK1 in Fas signaling. To this end, we used BMDMs from mice with myeloid-specific deletion of TAK1 on a TNF-α–deficient background (Lyz2^Cre/+Map3k7^flox/floxTnf^−/− and Lyz2^Cre/+Map3k7^flox/–Tnf^−/− mice, referred to hereafter as TAK1^mKO mice) and Map3k7^flox/floxTnf^−/− mice as control mice (referred to hereafter as TAK1^WT mice) (15). Surprisingly, TAK1-deficient BMDMs from TAK1^mKO mice exhibited high sensitivity to even low concentrations of FasL compared with control BMDMs from TAK1^WT mice (Fig. 2A). Immunoblot analysis also demonstrated that TAK1-deficient BMDMs possessed increased Fas signaling strength, as illustrated by enhanced processing of caspase-8, GSDMD, and GSDME (Fig. 2B). Furthermore, culture supernatants in the FasL-treated TAK1-deficient BMDMs included intracellular proteins such as caspase-3, caspase-1, and HMGB1, a well-characterized inflammatory mediator belonging to damage-associated molecular pattern molecules (23) (Fig. 2C). Fas-mediated activation of the RIPK1/RIPK3/MLKL signaling pathway, a hallmark of programmed necrosis, termed “necroptosis” (24), was detectable in TAK1-deficient BMDMs only when in the presence of Z-VAD-FMK (Fig. 2D). Taken together, our data indicate a novel negative regulatory role of TAK1 to limit Fas-induced macrophage death.

We next sought to explore the molecular mechanism by which TAK1 negatively controls Fas-induced macrophage death. Through immunoblot-based screening, we discovered that ligation of Fas with a low concentration of FasL led to hyperphosphorylation of RIPK1 Ser166 in TAK1-deficient but not control BMDMs (Fig. 3A). Moreover, TAK1-deficient BMDMs showed accelerated protein reduction of full-length c-FLIP (c-FLIP_L), even though the cleaved c-FLIP was detected equally between both types of macrophages (Fig. 3A). These data suggest that TAK1 somehow regulates the dynamics of both RIPK1 and c-FLIP in Fas signaling. To further address this issue, we used TAK1^mKO mice harboring a catalytically inactive RIPK1 mutant (RIPK1D138N) (Lyz2^Cre/+Map3k7^flox/floxRipk1^D138N/D138NTnf^−/−, referred to hereafter as TAK1^mKORIPK1^D138N/D138N mice) (15, 17). Remarkably, TAK1-deficient BMDMs expressing the RIPK1D138N mutant showed resistance against cell death triggered by lower concentrations of FasL (Fig. 3B). Moreover, the presence of the RIPK1 mutant in TAK1-deficient macrophages effectively abrogated proteolytic cleavage of the effector molecules responding to Fas signaling and release of intracellular HMGB1 (Fig. 3C, 3D). Collectively, these data confirmed the role of TAK1 to target and antagonize RIPK1, whose kinase activity can affect the macrophage responsiveness to FasL.

From further observation of TAK1^mKO mice, we found that compared with TAK1^WT mice, 44% of TAK1^mKO mice among all we observed (22 out of 50 mice) were apparently ill from 6 wk after birth in the SPF environment, as characterized by growth retardation, ruffled fur, and lethargic behavior. These characteristic features of the mice strongly resembled those of mice with sepsis. Remarkably, flow cytometric analysis showed that peritoneal resident macrophages (defined as large peritoneal macrophages [LPMs]: CD11b⁺F4/80^hi) (25) were conspicuously absent in apparently ill TAK1^mKO mice (Fig. 4A, 4B). Alternatively, the CD11b⁺F4/80^−/dimLy6C⁺ cell population was abnormally accumulated in the peritoneal cavities of the TAK1^mKO mice (Fig. 4C) and was identified as a mixed population of monocytes and neutrophils from detailed analysis (Fig. 4C, 4D). These strongly resembled the symptoms of acute peritonitis, as exemplified by macrophage disappearance in parallel with the recruitment of neutrophils and monocytes in the peritoneal cavity (25–27). Consistent with this phenotype, histological analysis identified the frequent presence of inflammatory infiltration in the livers of sick TAK1^mKO mice (Fig. 4E–4H). In addition, flow cytometric analysis revealed increased numbers of CD45⁺ leukocytes, a decrease in the percentage of tissue-resident macrophages, mainly Kupffer cells, and then accumulation of neutrophils and monocytes in the liver (Fig. 4I and Supplemental Fig. 1). Furthermore, serological analysis showed significantly elevated levels of aspartate aminotransferase, alanine aminotransferase, and LDH (Fig. 4J–4L). Collectively, these data suggest that TAK1^mKO mice developed hepatic injury. In addition to the peritoneal cavity and liver, a lower percentage of tissue-resident macrophages was observed in the lungs and visceral adipose tissues but not the colons of the sick TAK1^mKO mice (Supplemental Fig. 2). Taken together, these data demonstrate that TAK1 in macrophages has an ability to protect mice from spontaneous tissue inflammation.

Intriguingly, even in visibly healthy TAK1^mKO mice, we have observed the complicated phenotypes in the peritoneal cavities: 32% of the mice (16 out of 50 mice) were normal, 34% of the mice (17 out of 50 mice) showed the accumulation of neutrophils and monocytes along with macrophage disappearance, and the remaining 34% of the mice (17 out of 50 mice) exhibited the presence of LPMs with increased expression of MHC II (Supplemental Fig. 3A, 3B). This implies that some event to induce the local inflammation naturally occurs in the peritoneal cavities of TAK1^mKO mice. In addition, the TAK1^mKO mice bearing MHC II⁺ LPMs tended to have infiltrating neutrophils in the peritoneal cavities (Supplemental Fig. 3A). As expected, neither MHC II⁺ LPMs nor infiltrating neutrophils was present in the TAK1^WT mice we tested (Fig. 4C and Supplemental Fig. 3B). On the basis of these objective criteria, we concluded that TAK1^mKO mice are prone to develop peritonitis spontaneously.

Given that special attention has been paid to a functional interplay between cell death and inflammation (28, 29), we hypothesized that the disappearance of tissue-resident macrophages in TAK1^mKO mice may result from cell death. Having observed the effectiveness of a kinase-defective RIPK1 mutant to block Fas-induced aberrant cell death in TAK1-deficient macrophages, we thus sought to assess whether the onset of peritonitis in TAK1^mKO mice was prevented by abrogating RIPK1 kinase activity. As mentioned above, inflammatory manifestations in the peritoneal cavity were evident in 68% of visibly healthy TAK1^mKO mice, and, remarkably, the sign of the disease was substantially reduced in TAK1^mKORIPK1^D138N/D138N mice (Fig. 5). On the other hand, TIR Toll/IL-1R (TIR) domain-containing adapter inducing IFN-β (TRIF) deficiency in TAK1^mKO mice (referred to as TAK1^mKOTRIF^−/− mice) (15) had only a minor impact on the relief of peritonitis compared with a catalytically inactive RIPK1 (Fig. 5). These data suggest that the emergence of spontaneous tissue inflammation in TAK1^mKO mice is mainly attributed to the hyperactivation of RIPK1 in macrophages.

A growing body of research has documented that the death receptor Fas induces apoptosis by facilitating proteolytic cleavage of caspase-8 and its downstream effector caspase-3 in various cell types (3). In the present study, we found that Fas stimulation promotes proteolytic cleavage of not only caspase-3 and PARP but also gasdermin family members GSDMD and GSDME in a caspase enzymatic activity–dependent mechanism. This result was surprising but was expected, to an extent, from previous reports (12–16, 30). Our data argue that Fas-to-caspase-8 signaling urges macrophages to gasdermin-mediated pyroptosis, even though we do not rule out any possibility of caspase-8/caspase-3–dependent apoptosis. Another important view is that in normal mouse macrophages, potent Fas engagement is required to induce cell death. In this respect, a past report may give insight into the current observation (31). We speculate that specific mechanisms not to readily kill macrophages in response to death receptor stimulation are present under physiological conditions; otherwise, the induction of unwanted macrophage death has the potential for breakdown of immune homeostasis (32). In support of this notion, to our knowledge, we demonstrated for the first time that TAK1-deficient macrophages underwent massive cell death in response to even low concentrations of FasL. This represents a negative regulatory role for TAK1 to modulate the Fas signaling threshold. Thus, these findings provide a new model to limit Fas-induced proinflammatory cell death.

Our data clearly showed that hyperphosphorylation of RIPK1 was crucial for an increased sensitivity to Fas-induced death of TAK1-deficient macrophages. This indicates the requirement of TAK1 in targeting and antagonizing RIPK1, as described previously (33–35). Because mouse BMDMs expressing a catalytically inactive RIPK1 mutant had no effect on the induction of cell death by Fas stimulation (Supplemental Fig. 4A), the data indicate that the RIPK1 kinase activity per se is dispensable for Fas-induced cell death in normal macrophages. In addition, we observed an accelerated reduction in c-FLIP_L protein in FasL-treated TAK1-deficient macrophages, and this resembles that of a past study (36). The mutual regulation between c-FLIP and caspase-8 is well appreciated, and c-FLIP is also known as one of prosurvival genes induced by NF-κB activation (5). However, Fas ligation did not lead to robust activation of NF-κB and MAPK signaling pathways even in TAK1-sufficient BMDMs (Supplemental Fig. 4B). For this reason, the dynamics of c-FLIP_L in TAK1-deficient macrophages is most likely affected by hyperactivation of RIPK1, thereby promoting caspase-8 activity that is directed to c-FLIP_L for proteolytic cleavage, whereas the involvement of NF-κB–dependent events is unlikely. Therefore, it is plausible that TAK1 limits Fas-to-caspase-8–driven macrophage death by restraining the RIPK1 kinase activity in a manner independent of NF-κB–induced de novo protein synthesis.

It is now known that the RIPK1 kinase activity is associated with the induction of not only caspase-8–mediated apoptosis but also RIPK3/MLKL-mediated necroptosis under certain circumstances (37, 38). Nevertheless, we failed to detect Fas-induced RIPK3/MLKL activation in TAK1-deficient macrophages in the absence of caspase inhibitor, indicating that necroptosis does not occur in this situation. One conceivable reason is that active caspase-8 targets and antagonizes RIPK3 as well as upstream RIPK1 to downregulate their activities, as supported by past works (39, 40).

In the present study, we provided strong evidence that macrophage TAK1 plays a unique role in suppressing tissue inflammation. Considering our previous report that Lyz2^Cre/+Map3k7^flox/flox mice showed neither gross abnormalities nor spontaneous tissue inflammation in the SPF environment (15), the unanticipated phenotypes of TAK1^mKO mice offered an opportunity for consideration of TAK1 function in macrophages in vivo. Our present views are as follows. TAK1-deficient macrophages in Lyz2^Cre/+Map3k7^flox/flox mice may die during fetal stages or soon after birth by the action of TNF-α. Meanwhile, TAK1-deficient macrophages in TAK1^mKO mice can persist in vivo after birth. However, as the mice grow up, they are threatened by some intrinsic factors that may lead the macrophages to death. Frequent macrophage disappearance in TAK1^mKO mice may reflect a result of macrophage death, and inflammation may be caused by damage-associated molecular patterns (e.g., HMGB1) derived from dead macrophages themselves. Taking our former and current in vitro studies into account (15), this scenario is feasible. Therefore, we focused on a causal relationship between aberrant cell death in TAK1-deficient macrophages after death receptor and/or TLR stimulation and spontaneous tissue inflammation in TAK1^mKO mice. Although we have yet to establish its direct link, the characteristic features of TAK1^mKO mice remind us of the relevance of death receptor–induced cell death and tissue inflammation. Indeed, our current data from TAK1^mKORIPK1^D138N/D138N mice are highly suggestive. In the meantime, our previous (15) and present studies suggest that there are only a few relationships of TLR-to-TRIF signaling with disease onset, if any. Future studies will be needed to unequivocally answer this question.

In summary, this study highlighted not only a potential role of TAK1 in regulating Fas-induced macrophage death but also the possibility of perturbation of immune homeostasis driven by aberrant proinflammatory cell death. On the basis of our results, we propose here that TAK1 deficiency in myeloid lineage cells is associated with autoinflammatory disease. Our findings may be applicable as supporting information to understand pathologies of unidentified autoinflammatory diseases and emerging infectious diseases (41, 42).

We are grateful to S. Akira (Osaka University) for providing Lyz2^Cre/+Map3k7^flox/flox mice and V. Dixit (Genentech) for providing Ripk1^D138N/D138N mice. We also thank the staff of the Division of Animal Research and Division of Instrumental Research in Research Center for Supports to Advanced Science at Shinshu University for helpful support.

The authors have no financial conflicts of interest.