Role of Gate-16 and Gabarap in Prevention of Caspase-11-Dependent Excess Inflammation and Lethal Endotoxic Shock

Sepsis is a life-threating multi-organ disease induced by host innate immunity to pathogen-derived endotoxins including lipopolysaccharide (LPS). Direct sensing of LPS by caspase-11 activates inflammasomes and causes lethal sepsis in mice. Inhibition of caspase-11 inflammasomes is important for the prevention of LPS-induced septic shock; however, whether a caspase-11 inflammasome-specific suppressive mechanism exists is unclear. Here we show that deficiency of GABARAP autophagy-related proteins results in over-activation of caspase-11 inflammasomes but not of canonical inflammasomes. Gate-16−/−Gabarap−/− macrophages exhibited elevated guanylate binding protein 2 (GBP2)-dependent caspase-11 activation and inflammatory responses. Deficiency of GABARAPs resulted in formation of GBP2-containing aggregates that promote IL-1β production. High mortality after low dose LPS challenge in Gate-16−/−Gabarap−/− mice primed with poly(I:C) or polymicrobial sepsis was ameliorated by compound GBP2 deficiency. These results reveal a critical function of Gate-16 and Gabarap to suppress GBP2-dependent caspase-11-induced inflammation and septic shock.


INTRODUCTION
Sepsis is defined as a life-threatening multi-organ dysfunction syndrome caused by the excessive induction of host innate immunity against microbial infection (1). Even in developed countries, mortality in patients with severe sepsis is 20-50% (2). Various microbes contain endotoxins that have critical roles in the induction of sepsis (3). Lipopolysaccharide (LPS), a cell-wall component of Gram-negative bacteria, is a major endotoxin that strongly stimulates host innate immunity (4). Extracellular LPS is recognized by cell surface receptor complexes containing Toll-like receptor 4 (TLR4) together with CD14, LPS binding protein (LBP), and Myeloid Differentiation factor 2 (MD-2). LBP binds to LPS and then transfers this complex to CD14 to promote the formation of a complex containing LPS and MD-2/TLR4 (5). Activation of TLR4 signaling cascades mediates the production of proinflammatory cytokines such as TNF-α, IL-6, and IL-12, and precursor (preform) proteins of IL-1β and IL-18 (proIL-1β and proIL-18), which are critical for tissue damage and high fever during sepsis (6).
Infection with Gram-negative bacteria such as Salmonella, Citrobacter, Chlamydia, and Escherichia into host innate immune cells activates the caspase-11 inflammasome (7). During the course of an infection, Gram-negative bacteria actively or passively invade into host cells and eventually form pathogencontaining vacuoles (PCVs) (33). Caspase-11 recognition of LPS from Gram-negative bacteria inside PCVs was dramatically enhanced through lysis of the PCVs by interferon-inducible GTPases such as guanylate binding proteins (GBPs), which normally function as interferon-inducible cell-autonomous effectors against various PCV-forming intracellular pathogens such as Toxoplasma and Gram-negative bacteria (33). Upon infection by vacuolar pathogens, GBPs are recruited onto the PCV membranes to destroy the structure (34,35). During Gram-negative bacterial infection, the accumulation of GBPs on PCVs is thought to promote the lysis of PCVs or destroy the bacterial cell wall, resulting in the exposure of LPS to the cytosol and its recognition by caspase-11 (33,36). Thus, GBP is involved in the activation of caspase-11 inflammasomes and acts as a hub for innate immunity and anti-microbial cellautonomous immunity.
Here we demonstrate that GABARAP autophagy proteins negatively regulate GBP2-dependent caspase-11 inflammasome activation to prevent sepsis. Depletion of the GABARAP subfamily, but not the LC3 subfamily, in macrophages resulted in enhanced IL-1β production and pyroptosis in response to LPS transfection, OMV treatment and Gram-negative bacterial infection. In contrast, the GBP2-independent LPS introduction-induced activation of caspase-11 inflammasome as well as the ATP-mediated activation of the canonical NLRP3 inflammasome were normal in Gate-16 −/− Gabarap −/− cells. Deficiency of Gate-16 and Gabarap resulted in formation of GBP2 aggregates also containing LPS. Moreover, Gate-16 −/− Gabarap −/− mice exhibited high susceptibility to LPSinduced and cecal ligation puncture (CLP)-induced septic shock, which was ameliorated by GBP2 deficiency. Taken together, these data demonstrate that GABARAP autophagy proteins specifically limit GBP2-dependent caspase-11 inflammasome activation and sepsis.

RESULTS
Elevated IL-1β Production and Pyroptosis in LPS-Transfected Gate-16 −/− Gabarap −/− Macrophages The lysis of bacteria-containing vacuoles by GBPs is important for the activation of caspase-11 inflammasomes (33,36). Furthermore, we recently showed that some essential autophagy proteins play a critical role in GBP-dependent anti-microbial cell-autonomous immunity (42,43). To analyze the roles of autophagy proteins in caspase-11-mediated immune responses, we measured LPS transfection-induced IL-1β production and pyroptosis in macrophages from Lysozyme M-Cre Atg12 fl/fl mice (Atg12 myeloid mice) (Figures 1A,B). Notably, Atg12 myeloid macrophages showed elevated IL-1β production and significantly increased LDH release compared with wildtype control cells (Figures 1A,B). Because Atg12 myeloid cells (and other cell types below) are particularly sensitive to LPS transfection, we modified the LPS transfection protocol slightly to avoid saturation, resulting in somewhat lower death rate than typically reported using standard protocol (33). Atg12 together with Atg5 and Atg16L1 are critical for the processing of Atg8 family proteins consisting of Lc3a, Lc3b, Gabarap, Gabarapl1, and Gate-16 in mice (35). Therefore, we examined which Atg8 proteins were responsible for the enhanced IL-1β production and pyroptosis in Atg12 myeloid macrophages (Figures 1C,D). Among macrophages lacking The data are combined data of more than three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, and N.S., not significant, two-tailed t-test. each of the Atg8 proteins, Gate-16 −/− cells showed slightly but significantly upregulated IL-1β production, whereas the other deficient cells did not ( Figure 1C). Furthermore, Gate-16 −/− Gabarap −/− macrophages exhibited much higher levels of IL-1β production and LDH release in response to LPS transfection with a magnitude comparable to Atg12 myeloid cells (Figures 1C,D). In contrast, macrophages lacking Lc3a and Lc3b (Lc3a −/− Lc3b −/− macrophages) had similar responses to wild-type cells after LPS transfection (Figures 1C,D), indicating that a lack of GABARAP subfamily proteins such as Gate-16 and Gabarap leads to elevated non-canonical inflammasome responses to LPS transfection. Similar to LPS transfection, Gate-16 −/− Gabarap −/− macrophages among cells lacking single or compound deficiency of GABARAP subfamily members were the most hyper-responsive to Citrobacter koseri, a Gram-negative bacterium whose infection causes caspase-11 inflammasome activation (33) (Figures 1E,F). Furthermore, excess IL-1β production and cell death in C. koseri-infected Gate-16 −/− Gabarap −/− macrophages were observed in manners dependent on bacterial dose and infection time (Figures 1G-J). Outer membrane vesicles (OMVs) derived from Gram-negative bacteria contain LPS that stimulates the caspase-11 inflammasome (44). OMV-induced production of IL-1β and induction of cell death were also enhanced in Gate-16 −/− Gabarap −/− macrophages (Figures 1K,L). Thus, caspase-11 inflammasome-mediated immune responses were increased by a deficiency of GABARAP subfamily proteins.

High Mortality in
We next examined the physiological relevance of the GABARAP subfamily-mediated negative regulation of non-canonical inflammasome activation. As previously reported (18,19), mice primed with the TLR3 agonist poly(I:C) exhibited LPS-triggered inflammation and mortality in a caspase-11-dependent manner (Supplementary Figures 3A,B). After poly(I:C) priming, wildtype and Gate-16 −/− Gabarap −/− mice were intraperitoneally challenged with a low dose of LPS and monitored for survival ( Figure 3A). Although 80% of wild-type mice survived, all Gate-16 −/− Gabarap −/− mice died within 20 h post-LPS injection ( Figure 3A). Next, we compared the levels of inflammatory cytokines in sera between wild-type and Gate-16 −/− Gabarap −/− mice following LPS challenge (Figures 3B-F). Levels of IL-1β and IL-18, TNF-α, IL-6, and IL-12 in the sera of LPSinjected Gate-16 −/− Gabarap −/− mice were significantly higher compared with wild-type mice (Figures 3B-F), suggesting hyper caspase-11-dependent inflammation in Gate-16 −/− Gabarap −/− mice following LPS challenge. LPS lethality can be driven via TLR-induced TNF-α/IFN-β and caspase-8 (46). Therefore, we examined the role of TLR4 in the high sensitivity to LPS in When wild-type mice were treated with TAK-242, a potent inhibitor of TLR4 signaling (47), and subsequently challenged with intraperitoneal LPS injection alone, LPS infection-induced mortality and elevation of proinflammatory cytokines were profoundly inhibited (Supplementary Figures 3C,D). Next we treated Gate-16 −/− Gabarap −/− mice with the dose of TAK-242, and challenged the mice with sublethal LPS injection after poly(I:C) priming (Supplementary Figures 3E,F). Mortality and levels of proinflammatory cytokines in sera from Gate-16 −/− Gabarap −/− mice in the absence or presence of TAK-242 were comparable (Supplementary Figures 3E,F), indicating that TLR4 is dispensable for the hyper sensitivity to the polyIC/LPS-induced septic shock in Gate-16 −/− Gabarap −/− mice. Taken together, these data indicate that GABARAP autophagy proteins are important for the suppression of caspase-11-dependent sepsis.

Mice in Sepsis Models
Finally we examined the physiological relevance of the GABARAP subfamily-mediated negative regulation of GBP2-dependent non-canonical inflammasome activation. As previously reported (51), mice primed with poly(I:C) exhibited LPS-triggered inflammation and mortality dependent on GBP2 and caspase-11 (Supplementary Figures 3A,B). Moreover, Gate-16 −/− Gabarap −/− Gbp2 −/− mice were more resistant to low dose LPS challenge ( Figure 6A) and exhibited lower levels of IL-1β and IL-18 in the sera (Figures 6B,C) compared with Gate-16 −/− Gabarap −/− mice. Next we tested the physiological significance of Gate-16 and Gabarap for inhibition of GBP2-dependent caspase-11 inflammasome activation during polymicrobial sepsis in the CLP model (52) (Figure 6D). Compared with wild-type mice, Gate-16 −/− Gabarap −/− mice showed increased mortality during the CLP and markedly higher production of IL-1β and IL-6 in the peritoneal fluids (Figures 6D-F). In contrast, the elevation of cytokines and mortality was significantly ameliorated in (Figures 6D-F). Taken together, these results suggest that GABARAP subfamily members specifically and physiologically downregulate GBP2dependent caspase-11-induced innate immune responses to prevent septic shock.

DISCUSSION
In this study, we demonstrated that deficiency of the GABARAP subfamily proteins such as Gate-16 and Gabarap enhances activation of caspase-11 inflammasome in response to specific LPS stimulation. We found how LPS was prepared for the stimulation of Gate-16 −/− Gabarap −/− or Gbp2 −/− macrophages determined whether caspase-11 inflammasome activation was upregulated or downregulated, respectively. Regarding the caspase-11-dependent response, the opposite phenotypes of Gate-16 −/− Gabarap −/− and Gbp2 −/− cells were only observed for LPS transfection, C. koseri infection and OMV stimulation but not for other stimuli such as LPS electroporation and CTB-LPS treatment. Electroporation by high electronic pulse forms pores on the plasma membranes (53), allowing LPS to be directly transferred into the cytoplasm. CTB physiologically transports cholera toxin A fragment from the plasma membrane into the cytoplasm (54). Thus, the co-internalization of LPS and CTB into the cytosol may result in the cytosolic exposure of LPS and subsequent activation of caspase-11 (7). However, LPS is included in liposomes or membranous vesicles in the case of LPS transfection, or bacterial infection and OMV (44). The liposomally transfected LPS may be in membranous structures as likely as OMV that is naturally generated by Gram negative bacteria (44). Whether CTB-or electroporation-mediated LPS transfer physiologically happens or not is uncertain, however, infection of Vibrio cholerae, which naturally possesses CTB (55), might activate caspase-11 inflammasome in GBP2-dependent and -independent pathways (33).
We found that Gate-16 −/− Gabarap −/− macrophages showed elevated immune responses to GBP2-dependent non-canonical NLRP3 inflammasome-dependent stimuli but not to canonical NLRP3 inflammasome-dependent stimulation. For canonical NLRP3 inflammasomes, various inhibitory mechanisms were previously reported. NLRP3 mRNA transcription is inhibited by Ahr and miR-233 (20,21) and NLRP3 mRNA translation is prevented by TTP (22). NLRP3 protein is ubiquitinated or nitrosylated by ARDH2 and nitric oxide, respectively (23,24). NLRP3 sensing of ROS is modulated by TRIM30, Tim-3 and Nrf2 (25)(26)(27). Moreover, the NLRP3 inflammasome is inhibited by type I interferon and IKKα by regulating IL-1β mRNA transcription and ASC proteins, respectively (28,29). Given that NLRP3 is shared by both the canonical and caspase-11 inflammasomes, these negative regulators might also suppress LPS-mediated excessive inflammation and septic shock. However, little is known about which inhibitory molecules are specific for caspase-11 inflammasomes, except for stearoyl lysophosphatidylcholine (56). Our study demonstrates that GABARAPs are specifically involved in caspase-11 inflammasomes upstream of caspase-11, but not in the canonical NLRP3, Aim2, and NLRC4 inflammasomes. Thus, GABARAPs are specific negative regulators of the caspase-11 inflammasome. However, a previous study showed that Gabarap −/− macrophages displayed increased responses of canonical NLRP3 inflammasome and high mortality after CLP (30). On the other hand, responses of canonical NLRP3 inflammasome in our Gabarap −/− macrophages were normal. The discrepancy might be caused by different strategies for generation of Gabarap −/− mice or different genetic background of the mice between the two studies. To check whether the phenotype found in a genetargeted cells is restored by reintroduction of the gene is important to prove that the phenotype is caused by loss of the gene. In the present study, we confirmed that the liposomally transfected LPS-mediated increased IL-1β production in Gate-16 −/− Gabarap −/− macrophages was restored by reintroduction of Gabarap to some extent, indicating the phenotype is due to loss of Gabarap. Further studies to compare the two different Gabarap −/− mouse lines will clarify the involvement of Gabarap in the negative regulation of canonical and non-canonical NLRP3 inflammasome activation. In addition, Nedd4 is very recently shown to mediate caspase-11 degradation by K48linked polyubiquitination (32), indicating that caspase-11 activity might be tightly controlled by the protein amounts and uniform distribution by Nedd4 and GABARAPs, respectively. Furthermore, SERPINB1 plays a role in inhibiting LPSinduced inflammasome activation (31). Thus, reports regarding inhibitory mechanisms on caspase-11 inflammasome activation are currently growing.
Atg12-deficient macrophages exhibited the over-activation of canonical and caspase-11 inflammasomes. A deficiency in GABARAPs resulted in the enhanced activation of caspase-11 inflammasomes whereas no effect was observed for the canonical inflammasome. However, a deficiency of LC3s led to the enhanced activation of canonical inflammasomes but not caspase-11 inflammasomes. Thus, downstream of Atg12, GABARAPs, and LC3s play different roles in the suppression of NLRP3 inflammasome activation. Atg12-deficient or Lc3a −/− Lc3b −/− macrophages contained more damaged mitochondria than wild-type cells in response to LPS plus ATP stimulation as previously reported (45). The accumulation of damaged mitochondria is caused by an impairment in autophagy termed mitophagy due to lack of Atg12 or LC3s (57). In contrast, the formation of GBP2 aggregates due to lack of Gate-16 and Gabarap is the direct cause of caspase-11 over-activation and is independent of autophagy because GBP aggregation was observed only in cells lacking Atg3, Atg5, Atg7, or Atg16L1, all of which are essential for Atg8 lipidation, but not in cells lacking Atg9, Atg14, or FIP200, all of which are essential for autophagy (39,42). Our current and previous study demonstrate that Arf1 inhibition by Brefeldin A in wild-type cell resulted in ubiquitin aggregates containing GBPs (42). Arf1 is a small GTPase important for generation of COPI-coated vesicles from Golgi during intra-Golgi transport (48). In addition, Gate-16 was originally identified as Golgiassociated ATPase enhance 16kD (58), whose function was not initially linked not with autophagy but only with intra-Golgi transport (59). Given that GBP2 is detected in microsomal fractions that also contain Golgi vesicles, Gate-16 (and also Gabarap) via interaction with Arf1 may regulate the generation of uniformly small size of Golgi-derived vesicles including GBP2 in an autophagy-independent fashion. Deficiency of the GABARAPs as likely as Arf1 inhibition might lead to failure of generation of the GBP2-containing vesicles, resulting in formation of large GBP2 aggregates which enhance GBP2mediated caspase-11-depenendent response. Thus, our data indicate that, although the Atg12-LC3 subfamily axis suppresses canonical NLRP3 inflammasomes by autophagy (mitophagy), the Atg12-GABARAP subfamily axis negates GBP2-dependent activation of non-canonical NLRP3 inflammasomes in an autophagy-independent manner.
We also demonstrated that poly(I:C)-primed Gate-16 −/− Gabarap −/− mice were highly susceptible to low dose LPS-induced and polymicrobial septic shock. However, the increased mortality of LPS-injected Gate-16 −/− Gabarap −/− mice was not completely prevented by additional GBP2 deficiency. In this regard, TLR4 may not play a role in the elevated immune responses in Gate-16 −/− Gabarap −/− mice since the pharmacological TLR4 inhibition did not reduce high mortality and levels of proinflammatory cytokines in sera. On the other hand, it is possible that increased canonical inflammasome activation and its inflammation by Gabarap deficiency could contribute to the high mortality in a manner independent on GBP2, since Gabarap has been shown as a negative regulator of canonical NLRP3 inflammasome and prevents LPS-induced lethality (30,60). Furthermore, we have found the physiological relevance of a negative regulatory mechanism for GBP2-dependent caspase-11 inflammasome activation, which is essential for the prevention of LPS-mediated and polymicrobial septic shock in vivo. OMVs deliver a number of membrane-bound antigens, OMV-based vaccines have attracted much attention (61). However, our current study indicates that we should be cautious of using OMV vaccine treatment in humans with a mutation in ATG16L1, in which GABARAPs are not lipidated and hence remain inactivated (62), because of the potential for strong non-inflammation responses and septic shock compared with normal individuals. Small compounds targeting the GTPase activity or the isoprenylation site of GBP2 might be helpful to attenuate the endotoxemia.
Before infection, macrophages were seeded into 6-, 24, 96 well-plates at a density of 2.5 × 10 6 , 3 × 10 5 , 1 × 10 5 cells and pre-stimulated with IFN-γ (10 ng/ml) for 24 h to induce GBP2. For infection with C. koseri, the bacteria was pre-cultured with LB medium for 4 h under aerobic conditions at 37 • C. The bacteria was subcultured (1:10) in fresh LB medium for 20 h to stationary phase. Bacterial density was calculated from the OD 600nm value, and the cells were resuspended into antibiotic-free medium at the indicated multiplicity of infection (MOI). BMDMs were infected with the bacteria (MOI = 10 or indicated MOI and centrifuged for 15 min at 500 × g and then BMDMs were incubated for 1 h at 37 • C. After 1 h, 100 µg/ml gentamycin (Invitrogen) was added to kill extracellular bacteria. After 1 h incubation, the cells were washed once with PBS and changed fresh macrophage medium containing 10 µg/ml gentamicin for the remainder of the infection. The cells or supernatants were collected at 16 h or indicated times after infection.

LPS and Cholera Toxin B Subunit (CTB) Treatment
BMDMs were pre-stimulated with 100 ng/ml Pam3CSK4 for 4 h. The cells were treated with non-treatment or 1 µg/ml LPS alone or 20 µg/ml CTB (List biological laboratories) or LPS and CTB treatment for indicated times.
LDH release was measured by using CytoTox96 Non-Radio Cytotoxicity Assay kit (Promega). To calculate % of LDH release, values of [(sample-untreated sample)/(total cell lysateuntreated sample)] × 100 were calculated in accordance with the manufacture's instruction.

Western Blot Analysis and Immunoprecipitation
The cells were lysed in a lysis buffer containing 1% Nonidet P-40, 150 mM NaCl, 20 mM Tris-HCl (pH 7.5), 1 mM EDTA and protease inhibitor cocktail (Nacalai Tesque). The cell lysates were separated by SDS-PAGE, transferred to polyvinyl difluoride membranes and subjected to immunoblotting using the indicated antibodies.
For caspase-1 and caspase-11 cleavage assay, Culture supernatant of BMDMs was added with 10% Trichloroacetic acid (TCA) and 10% acetone overnight at −20 • C. The supernatants were centrifuged for 30 min at 15,000 rpm, 4 • C and wash cold acetone two times. And then, Pellets were dried up and lysed in RIPA buffer. The lysates were detected by using Novex NuPAGE R SDS-PAGE Gel system (Thermo).
For immunoprecipitation, cell lysates were pre-cleared with Protein G-Sepharose TM (Amersham Pharmacia Biotech) for 2 h and then incubated with Protein G-Sepharose TM containing 1.0 µg of the indicated antibodies for 12 h with rotation at 4 • C. The immunoprecipitants were washed four times with lysis buffer, eluted by boiling with Laemmli sample buffer and subjected to immunoblot analysis using the indicated antibodies.
ASC Oligomerization Assay 2.5 × 10 6 BMDMs were seeded in 6 cm dish and pre-stimulated with 100 ng/ml Pam3CSK4 for 4 h and transfected with 2 µg/ml LPS for 6 h. The cells were collected and lysed with the 100 µl buffer A containing 20 mM HEPES-KOH (pH 7.5), 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, and 320 mM sucrose. The lysates were centrifuged for 8 min at 300 × g, 4 • C and the supernatants were mixed with the same volume of CHAPS buffer. The lysates were centrifuged for 8 min at 2,650 × g, 4 • C and the supernatants were removed and the pellets were incubated with 20 µl CHAPS buffer containing 2 mM DSS (Thermo) for 2 h on ice. And then 3 × sample buffer was mixed with the lysates and boiled for 5 min, 98 • C. ASC oligomerization was analyzed by western blot.

Isolation and Treatment of Bacterial OMVs
OMVs were isolated from E. coli BL21 as described previously (44). Briefly, E. coli (BL21) were cultured overnight, and subcultured 1/1000 in 500 ml LB broth media and cultured 37 • C overnight. The media were centrifuged at 10,000 × g for 10 min at 4 • C, then the supernatant were filtered through 0.45 µm filter and 0.22 µm to remove whole bacteria and debris. The solutions were centrifuged at 150,000 × g for 3 h at 4 • C to pellet OMVs. The pellets were re-suspended in sterile PBS. The concentration of OMVs was measured by protein assay and parts of OMVs solution were plated on the LB plate to confirm the bacterial free conditions. To stimulate BMDMs, BMDMs were seeded in the six well-plate for caspase-1 and caspase-11 cleavage assay and 96 well-plate for ELISA and LDH assay 1 day before stimulation. Isolated OMVs were treated with each 30 µg/well and 10 µg/ml in serum-free RPMI medium for 16 h.

Quantification of Activated Caspase-1
Pam3CSK4 pre-stimulated BMDMs were transfected with LPS for 4 h and added with FAM-FLICA (ImmunoChemistly technology) at 3 h. The cells were washed with FACS buffer containing 2% FBS, 0.009% NaN 3 , 2 mM EDTA and PBS two times. Activated caspase-1 was detected by using flow cytometry using FACS Verse (Becton Dickinson) and quantified using FlowJo Software (Tree Star).

Canonical Inflammasome Activation and Quantification of Damaged Mitochondria
To activate canonical inflammasome for the ELISA and WB, BMDMs were stimulated with 10 ng/ml LPS for 5 h and 5 mM ATP for 3 h. BMDMs were stimulated with 10 ng/ml LPS and 1 mM ATP for 30 min at 37 • C. To stain the mitochondria, the cells were stained with 25 nM of MitoTracker Green FM and MitoTracker Deep Red FM (Invivogen) for 15 min at 37 • C (67). The cells were washed with FACS buffer two times and the damaged mitochondria were analyzed by using flow cytometry.

Isolation of Cytosol Fraction From BMDMs and LPS Quantification
To isolate the cytosol fraction from BMDMs, Digitonin-based fractionation method were utilized as described previously (44). 7.5 × 10 5 BMDMs were seeded on the 24 well-plate and cultured overnight. The cells were transfected with 1 µg/well LPS for each time points and washed with cold PBS four times. Cells were treated with 180 µl of 0.005% digitonin extraction buffer for 8 min and the supernatant was collected for cytosol fraction. The non-cytosol fraction containing cell membrane, organelles and nucleus was collected in 180 µl of 0.1% CHAPS buffer. The quantity of LPS in each fraction was measured by using LAL Chromogenic Endotoxin Quantitation kit (Thermo).

LPS Challenge in vivo
Mice were intraperitoneally injected with 10 mg/kg poly(I:C) (GE Healthcare) for 7 h and then 0.1 mg/kg or 0.4 mg/kg LPS were injected. After 3 h, the sera were collected from each mice and survival rates were tested. To prevent TLR4 signaling, 20 mg/kg of TAK-242 (CS-0408; Chemscene) was intraperitoneally injected at 0.5 h prior to LPS injection and at 0 h or 0.5 h post-LPS injection. To make the graph, the survival rate and cytokine concentration were used Prism5 software (Graph Pad software).

Cecal Ligation Puncture (CLP)
The mice were anesthetized using pentobarbital by injecting intraperitoneally. Anesthetized mice were removed the hairs of the abdomen by shaver. The skin was disinfected with 75% Ethanol swab. The mouse made about 1 cm insection of the abdomen and took out the cecum from the cut line. Forty percent of the cecum was ligated and punctured once with a 21-gauge needle. Mice were monitored daily by the sign of a moribund state for lethality. For cytokine production, mice were sacrificed at 16 h post-CLP, then collected the peritoneal fluids and measured IL-1β and IL-6 by ELISA.

Statistical Analysis
All experiments were performed using randomly assigned mice without investigator blinding. All data points and n-values reflect biological replicates (from three or four independent experiments). No data were excluded. Sample sizes were chosen by standard methods to ensure adequate power. All statistical analysis was performed using Excel 2013 (Microsoft) or GraphPad Prism5 (GraphPad Software). Statistical analyses for differences between two groups were performed using an unpaired two-tailed Student's t-test or log-rank test for analysis of the survival. P < 0.05 was considered to be statistically significant. No statistical methods were used to pre-determine sample size.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by the Animal Research Committee of Research Institute for Microbial Diseases in Osaka University.

AUTHOR CONTRIBUTIONS
MY conceptualized and supervised this project. NS, MS, and HB performed experiments and analyzed the data. YL, AP, and JM prepared reagents, samples, and animals for this study. MY, NS, MS, and HB wrote this manuscript. All authors contributed to the article and approved the submitted version.