MyD88-Dependent Signaling Is Required for HOIP Deficiency–Induced Autoinflammation

The linear ubiquitin chain assembly complex (LUBAC), composed of three core subunits, heme-oxidized IRP2 ubiquitin ligase 1 (HOIL-1), Shank-associated RH domain-interacting protein (SHARPIN), and the catalytically active HOIL-1–interacting protein (HOIP), is the only E3 ligase identified so far that can generate linear ubiquitin linkages (1–3). Linear ubiquitylation is crucial for multiple immune receptor signaling pathways, including TNF, TLRs, TRAIL, NOD-like receptors, and T and B cell receptors (4–9). The essence of LUBAC and linear ubiquitylation has been demonstrated by embryonic lethality of mice with deficiency of LUBAC components HOIP, HOIL-1, or LUBAC activity, which is caused by aberrant TNFR1-mediated signaling (10, 11). Defect in SHARPIN triggers severe dermatitis and systemic inflammation in adulthood (12, 13). In humans, patients with loss-of-function mutations in HOIP or HOIL-1 causes multiorgan autoinflammation, combined immunodeficiency, and recurrent viral and bacterial infections (14–16). At cellular level, B cell–specific ablation of HOIP results in B-1 cell development defect and defective Ab response to thymus-dependent and independent Ags in mice (8, 17). And mice with T cell–specific HOIP ablation display a strong defect in late thymocyte differentiation, regulatory T cell development, and homeostasis (7). Although defects in B cell and T cell development resulting from LUBAC deficiency may partially account for the immunodeficiency in patients with loss-of-function mutations in HOIP or HOIL-1, the source of autoinflammation is still not well understood, and dysregulated LUBAC activity is expected to have cell-specific effects on many cellular pathways and functions. Whether deficiency of LUBAC in innate immune cells, especially dendritic cells (DCs), is involved in the pathogenesis of autoinflammation or immunodeficiency is not clear.

DCs are potent APCs that initiate lymphocyte activation and critical for maintaining immune tolerance (18, 19). DCs are ubiquitously distributed in all tissues and surveil the environment to initiate immune response against foreign Ags or tolerance to self-antigens, especially in the intestine where microbiota and foreign Ags are ample. Therefore, strict regulations of DC activities are required for proper execution of immune response. The differentiation of DCs from common DC progenitors (CDPs) depends on the cytokines like fms-related tyrosine receptor kinase 3 ligand (Flt-3L) and GM-CSF (20), whereas homeostasis of DCs is maintained by processes like apoptosis to limit immune response after Ag presentation to T cells (21). Dysregulation of apoptosis in DCs results in the development of systemic autoimmunity (18, 22–24).

Given the indispensable role of DCs in linking innate and adaptive systems and dysregulation of DCs resulting in the development of systemic autoimmunity, we sought to examine the function of HOIP in DCs and explore whether loss of LUBAC function in DCs has any relationship with autoinflammation of patients. Using DC-specific conditional knockout mice, we found that HOIP was essential for maintenance of DC homeostasis and prevention of systemic inflammation. In knockout mice, DC numbers were decreased in the spleen because of uncontrolled DC death. Importantly, dysregulation of DC death resulted in the appearance of chronic systemic inflammation characterized by increased levels of serum proinflammatory cytokines and expansions in myeloid and B cells. We further identified a critical role for the intestinal microbiota and MyD88-dependent signals in the development of inflammation in mice with DC-specific HOIP deficiency. Our study uncovers that aberrant DC death via deficiency of LUBAC in DCs may underpin the pathogenesis of autoinflammation in LUBAC-mutated patients, and MyD88-dependent proinflammatory signal plays a substantial role in the autoinflammation associated with HOIP mutations.

All mice used in this study were on the C57BL/6 background at 7–9 wk of age unless otherwise noted. Hoip^{DC KO} mice were generated by crossing Itgax-Cre (CD11c-Cre) transgenic mice with Hoip^flox/flox mice. Hoip^flox/flox mice were generated by Crispr-Cas9 technique and backcrossed to C57BL/6 for at least 10 generations, as previously described (25). Itgax-Cre, Tnfrl^−/−, and Casp8^fl3-4 mice were purchased from The Jackson Laboratory and bred in the facility. Myd88^−/−mice and Mlkl^−/−mice were generously provided by Dr. Xiaoyu Hu at Tsinghua and Dr. Xiaodong Wang from the National Institute of Biological Sciences, respectively. For antibiotic treatment of Hoip^{DC KO} mice, broad-spectrum antibiotic mixture of 1 mg/ml ampicillin, 1 mg/ml neomycin, 0.5 mg/ml ciprofloxacin, and 0.5 mg/ml meropenem was added to the drinking water starting from day 2 of birth. Following weaning, mice were maintained on the same antibiotic mixture, except with 0.5 mg/ml vancomycin in place of ciprofloxacin until analysis (26). Mice treated with antibiotics were analyzed at ∼7 wk of age. All mice were housed in the specific pathogen–free animal facilities in Tsinghua University. All experiments used cohoused mice. All mouse experiments were performed in compliance with institutional guidelines and according to the protocol approved by the Institutional Animal Care and Use Committee of Tsinghua University.

Mouse spleens were fixed with 4% formaldehyde overnight and then dehydrated and embedded in paraffin. The 5-μm tissue sections were stained with H&E or Masson’s trichrome according to standard procedures. Stained sections were scanned using an Olympus microscope (IX73).

Spleens and lymph nodes were mechanically dissociated to obtain single-cell suspensions, and splenocytes were depleted of RBCs with ACK lysis buffer. Surface staining was completed in cold FACS buffer (PBS containing 1% FCS, 0.05% sodium azide). Samples were incubated with anti-CD16/CD32 prior to staining with the indicated fluorochrome-conjugated Abs. Samples were run on LSR Fortessa cytometers (BD Biosciences) and analyzed using FlowJo software (Tree Star). Cell populations were determined as follows: conventional DCs (cDCs), CD11c^high MHC class II⁺ (MHC II⁺); plasmacytoid DCs, CD11c^low Siglec-H⁺; cDC1 cells, CD11c^high MHC II⁺ CD8⁺ Sirp1α⁻; cDC2 cells, CD11c^high MHC II⁺ CD8⁻ Sirp1α⁺; pre-DCs, lin⁻ c-kit^int Flt-3⁺ CD11c⁺; CD115⁺ CDPs, lin⁻ c-kit^int Flt-3⁺ CD127⁻ CD115⁺; neutrophils, CD11b⁺ Ly6C^low Ly6G^high; monocytes, CD11b⁺ Ly6C^high Ly6G^low; B lymphocytes, CD19⁺ or B220⁺; follicular B lymphocytes, CD19⁺ CD93⁻ CD23⁺ CD21^low; germinal center B lymphocytes, CD19⁺ IgD^low CD95⁺ GL7⁺; CD4 T lymphocytes, CD4⁺ CD8⁻; CD8 T lymphocytes, CD8⁺ CD4⁻; and activated T lymphocytes, CD44^high CD62L^low. See Supplemental Table I for Abs used in this study.

Blood was collected from cardiac puncture, and different proteins in the serum were analyzed by sandwich ELISA. The levels of KC (MultiSciences), MCP-1 (MultiSciences), TNF-α (eBioscience), IFN-γ (eBioscience), IL-6 (eBioscience), IL-1β (eBioscience), and IL-1α (Abcam) were determined using ELISA kits. Abs to dsDNA (Thermo Fisher Scientific) were measured as previously described (27).

Bone marrow was flushed from femurs and tibias using a needle and syringe, and RBCs were lysed. Bone marrow–derived DCs (BMDCs) were generated by culturing bone marrow cell suspensions in complete RPMI 1640 supplemented with 10% FCS, l-glutamine, penicillin-streptomycin, sodium pyruvate, 2-ME, and 20 ng/ml recombinant GM-CSF (PeproTech) for 8 d. BMDCs were harvested and replated in complete RPMI 1640 for stimulation. Alternatively, CD11c⁺ BMDCs were enriched by sorting with FACS Aria (BD Biosciences).

Total RNA was extracted by TRIzol (Invitrogen), and reverse transcription was performed using RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. An ABI 7500 Real-Time PCR system (Applied Biosystems) and Power SYBR Green PCR Master Mix (Genestar) were used for real-time quantitative PCR. Results were normalized to Gapdh, and quantification was carried out using the 2^-ΔΔCt method. Melting curves were confirmed to ensure amplification of a single product. Primers for Hoip, which are located in exon3 and exon6, were designed. Primer pairs are as follows: Gapdh, sense, 5′-AAC AGC AAC TCC CAC TCT TC -3′, antisense, 5′-CCT GTT GCT GTA GCC GTA TT-3′; Il1a, sense, 5′-CGA AGA CTA CAG TTC TGC CAT T-3′, antisense, 5′-GAC GTT TCA GAG GTT CTC AGA G-3′; Il1b, sense, 5′-GCA ACT GTT CCT GAA CTC AAC T-3′, antisense, 5′-ATC TTT TGG GGT CCG TCA ACT-3′; and Hoip, sense, 5′-ACC GAG CTC AGT TTG CTG TT-3′, antisense, 5′-AGG AAA CAG GGA CCA GGA GT-3′.

A total of 5 × 10⁴ DCs per well in a 96-well format were treated with indicated stimulation. Effects on cell viability were measured using Cell Titer Glo, as described (28). Cell Titer Glo Luminescent Cell Viability Assay kit was from Promega.

Cells were collected with lysis buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 1% NP-40, 1 mM EDTA) containing 1 mM sodium orthovanadate, 1 mM sodium fluoride, 1 mM PMSF, and a protease inhibitor mixture (Roche). Total cell lysates were subjected to SDS-PAGE and transferred onto a PVDF membrane (Millipore). The membrane was sequentially probed with primary Abs and HRP-conjugated secondary Abs. ECL substrates (Pierce) were used to visualize the specific bands on the membrane.

For complex I immunoprecipitation, 2 × 10⁷ treated BMDCs were lysed in lysis buffer (30 mM Tris-HCl, pH 7.4, 120 mM NaCl, 2 mM EDTA, 2 mM KCl, 10% glycerol, and 1% Triton X-100) containing 1 mM sodium orthovanadate, 1 mM sodium fluoride, 1 mM PMSF, and a protease inhibitor mixture (Roche). Each sample was incubated with anti-FLAG M2 beads (A2220; Sigma-Aldrich) at 4°C overnight. Agarose beads were washed four times with cold lysis buffer, and the precipitates were boiled for 10 min in SDS loading buffer. The input and immunoprecipitation samples were then subjected to SDS-PAGE and then blotted using the indicated Abs. See Supplemental Table for Abs used in this study.

For bone marrow chimeras, 8-wk-old wild-type (WT) recipient mice were lethally irradiated by x-ray (5.5 Gy × 2) and received 5 × 10⁶ bone marrow cells from indicated donors by i.v. injection. Chimeras were used for further experiments 8 wk after initial reconstitution.

All statistical analysis was performed using Prism 6 (GraphPad Software). Statistical significance was evaluated using Student t test or one-way ANOVA. Log-rank (Mantel–Cox) test was used to compare survival curves. Differences were considered statistically significant if *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical parameters including number of biological replicates and repeat experiments, data dispersion, and precision measures (mean and SEM) are reported in the figure legends.

To investigate the role of HOIP in DC homeostasis and effector functions, we generated mice that lack HOIP in the DCs by crossing Hoip^flox/flox mice (Supplemental Fig. 1A) with mice expressing the Cre recombinase under the control of CD11c promoter (Itgax-Cre). Mice homozygous for DC-specific deletion of HOIP (Itgax-Cre Hoip^flox/flox, termed “Hoip^{DC KO}”) was confirmed by PCR (Supplemental Fig. 1B). At mRNA expression level, the specific deletion of HOIP was confirmed in sorted DCs by quantitative RT-PCR compared with other immune cells of hematopoietic origin (Supplemental Fig. 1C). At protein level, the specific loss of HOIP expression in enriched CD11c⁺ in vitro–generated BMDC was verified by Western blot analysis compared with Itgax-wt Hoip^flox/flox mice (termed “control”) (Supplemental Fig. 1D). In line with previous reports on other tissues and cells, analysis of TNFR1-signaling complex by TNF-α–Flag pulldown revealed that HOIP-deficient cells produced remarkably reduced levels of linear ubiquitination within the TNFR1 signaling complex (Supplemental Fig. 1E). Previous studies have reported that mouse embryonic fibroblasts with ablation of HOIP result in the impaired canonical NF-κB activation as exhibited by delayed phosphorylation and degradation of IκBα (10). Therefore, we examined NF-κB signaling in HOIP-deficient DCs. Surprisingly, this is not the case for Hoip^{DC KO} cells (Supplemental Fig. 1F). Consistently, NF-κB–targeted gene Ciap1 and A20 expression are not affected by HOIP deficiency in DCs (Supplemental Fig. 1G).

Hoip^{DC KO} mice were born at the expected Mendelian ratio and appeared healthy with no obvious abnormalities. However, by 6–8 wk of age, Hoip^{DC KO} mice spontaneously developed splenomegaly and lymphadenopathy (Fig. 1A–C). Histological analysis revealed that the splenic architecture was disrupted in Hoip^{DC KO} mice, and Masson trichrome staining revealed collagen deposits in the spleen, indicating fibrosis (Fig. 1D). Evidence of inflammation was observed in Hoip^{DC KO} mice, with elevated percentage and numbers of neutrophils and inflammatory monocytes in the spleens and mesenteric lymph nodes (mLNs) of Hoip^{DC KO} mice (Fig. 1E, 1F). We also detected elevated B cell numbers in the spleen and mLNs of Hoip^{DC KO} mice; further analysis revealed that follicular B cells, especially geminal center B cells, were accounted for the expansion of B cells (Supplemental Fig. 2A–C). Consistent with increased B cell expansion, a significant increase in serum anti-dsDNA IgG was observed in Hoip^{DC KO} mice compared with control mice (Fig. 1G), implicating autoimmunity was developed in Hoip^{DC KO} mice. Conversely, the absolute numbers of CD4 and CD8 T lymphocytes in spleen or mLNs have no differences, and there was no evidence of T cell activation in spleen or lymph nodes (Supplemental Fig. 2D, 2E). To further assess the inflammation in Hoip^{DC KO} mice, the levels of different proinflammatory cytokines and chemokines in the serum were measured. A significant increase in basal serum TNF-α, IFN-γ, KC, and MCP-1 was observed in Hoip^{DC KO} mice compared with control mice (Fig. 1H). To determine the ramifications of having increased inflammation and elevated cytokine levels, control or Hoip^{DC KO} mice were given a low dose of LPS (5 mg/kg). Under this circumstance, most of control mice survived, whereas all Hoip^{DC KO} mice succumbed to LPS-induced endotoxic shock within 48 h (Fig. 1I). Additionally, the levels of proinflammatory cytokines TNF-α, IL-1β, and IL-6 were significantly elevated in the sera of Hoip^{DC KO} mice compared with controls following LPS injection (Fig. 1J). Our data specifically demonstrate that Hoip^{DC KO} mice develop systemic inflammation characterized by increased numbers of neutrophils, inflammatory monocytes, B cells, and elevated levels of proinflammatory cytokines.

We further determined the effect of HOIP deficiency on DC population change. A significant decrease in the number of CD11c^hiMHC II⁺ cDCs was detected in the spleen of Hoip^{DC KO} mice compared with littermate control mice (Fig. 2A, 2B), whereas the number of plasmacytoid DCs in spleen was unchanged (Supplemental Fig. 3A, 3B). Further investigation of different subsets of splenic cDCs by examining CD8 and Sirp1α expression revealed that the percentage of CD8⁺Sirp1α⁻ DC subset (termed “cDC1”) was diminished, whereas the CD8⁻Sirp1α⁺ subset (termed “cDC2”) was increased (Supplemental Fig. 3A, 3B). Apart from spleen, CD103⁺ migratory DCs from mLNs were examined (29). Interestingly, significant reduction of migratory CD103⁺ DCs in mLN was observed in Hoip^{DC KO} mice (Fig. 2A, 2B). To investigate whether developmental impairment within DC precursor populations was responsible for the reduction of DC numbers in peripheral lymphoid organs, we examined pre-DCs and CDPs in the bone marrow of Hoip^{DC KO} mice. Hoip^{DC KO} mice still exhibited comparable number of both pre-DCs and CDPs in the bone marrow (Supplemental Fig. 3C, 3D), indicating that the origin of DCs is not responsible for the decreased DCs in peripheral lymphoid organs.

We next determined the cause of DC reduction and whether the decrease in DCs was responsible for the systemic inflammation observed in the Hoip^{DC KO} mice. It has been shown that LUBAC plays a critical role in TNF-α–induced cell death by switching prosurvival TNFR1 signaling into Caspase-8–dependent apoptosis or Caspase-8–independent, RIPK3/MLKL-mediated necroptosis (1, 10, 11). Therefore, we assessed the ability of HOIP-deficient DCs to undergo cell death in response TNF-α stimulation. Apoptosis can be classified into RIPK1-independent apoptosis (RIA) and RIPK1-dependent apoptosis (RDA) according to the dependence on RIPK1 kinase activity. Thus, we investigated the effect of HOIP loss on RIA. With TNF-α/CHX stimulation, cell death induction was comparable between HOIP-deficient DCs and control DCs (Fig. 2C, 2D), indicating HOIP is not required for prevention of TNF-α–induced RIA in DCs.

Second mitochondrial–derived activator of caspase (Smac) is a mitochondrial protein that can promote caspase activation to induce RDA through inhibition of cIAP1/2 (30). We then assessed the RDA induced by TNF-α plus Smac-mimetics (BV-6), and interestingly, we found that HOIP-deficient DCs showed a higher level of cell death than control DCs (Fig. 2C, 2E), suggesting HOIP is involved in negative regulation of TNF-α–induced RDA. Moreover, when TNF-α–induced apoptosis was blocked by pan-caspase inhibitor z-VAD, which induces necroptosis, the viability of HOIP-deficient DCs was lower compared with control cells (Fig. 2C, 2F), indicating HOIP is required for prevention of necroptosis induced by TNF-α. Necroptosis induced by TNF-α is dependent on the activities of RIPK1, RIPK3, and the pseudo-kinase MLKL. We also investigated the effect of inhibiting RIPK1 kinase activity on the caspase activation and necroptosis in HOIP-deficient DCs. Blocking of RDA and necroptosis using RIPK1 kinase inhibitor, Necrostatin-1, could increase the survival of DCs from Hoip^{DC KO} mice (Fig. 2C–F).

Because TNFR1 deficiency could prolong the survival of Hoip^−/− and Hoil-1^−/− embryos (10, 11) and HOIP-deficient DCs are sensitive to TNF-α–induced apoptosis and necroptosis, we hypothesized that TNF-α–induced cell death may cause the systemic inflammation in Hoip^{DC KO} mice. To address the role of TNFR1 signaling in inflammation in vivo, we crossed HOIP^{DC KO} mice with Tnfrl^−/− mice to generate Hoip^{DC KO} Tnfrl^−/− mice. Surprisingly, TNFR1 deficiency did not rescue inflammation but resulted in significant weight loss of Hoip^{DC KO} Tnfrl^−/− mice (Fig. 3A). In contrast to the accelerated TNF-α–induced cell death observed in vitro (Fig. 2E, 2F), Hoip^{DC KO} Tnfrl^−/− mice develop more severe systemic inflammation compared with Hoip^{DC KO} mice, with significant increases in percentage and numbers of inflammatory monocytes and neutrophils in spleen and mLN (Fig. 3C–F). Besides, absolute number of DCs of Hoip^{DC KO} Tnfrl^−/− mice were still reduced compared with Hoip^{DC KO} mice (Fig. 3G, 3H). Although HOIP is critical for restricting TNF-α–induced cell death in vitro, Hoip^{DC KO} Tnfrl^−/− mice develop inflammation and cachexia, indicating that signaling other than TNFR1 may contribute to HOIP deficiency–associated inflammation.

Because Hoip^{DC KO} mice have increased susceptibility to LPS-stimulated death (Fig. 1I) and multiple groups have reported that LPS can induce cell death (28, 31, 32), we examined whether a lower level of DCs in Hoip^{DC KO} mice was the result of HOIP-deficient DCs dying by LPS challenge. BMDCs generated from Hoip^{DC KO} or control mice were stimulated with LPS, cell death was examined by cell survival measurement, and increased cell death was observed in Hoip^{DC KO} cells (Fig. 4A). LPS has been reported as a strong inducer for both apoptosis and necroptosis; thus, we further investigated whether HOIP deficiency affects LPS-induced apoptosis. Because treatment with LPS in the presence of CHX, an inhibitor of protein synthesis, leads to activation of Caspase-3 and apoptosis via TLR4 (33), we stimulated BMDCs from Hoip^{DC KO} or control mice with LPS/CHX. With LPS/CHX stimulation, HOIP-deficient DCs exhibited higher level of cell death than control DCs (Fig. 4A, 4B), indicating that HOIP is required for prevention of apoptosis induced by LPS/CHX. Apart from LPS/CHX-induced apoptosis, apoptosis can also be induced by LPS in cells with the absence of inhibitor of apoptosis protein (IAP) (32). Thus, we used the LPS plus Smac-mimetics (BV-6) to inhibit IAP to induce apoptosis. With LPS/BV-6 stimulation, HOIP-deficient DCs exhibited decreased cell survival and enhanced Caspase-8 and Caspase-3 cleavage (Fig. 4A, 4C), suggesting that HOIP is involved in the negative regulation of apoptosis induced by LPS/BV-6. In addition to apoptosis, TLRs can activate necroptosis when Caspase-8 is inhibited by z-VAD (28, 34). Therefore, we used the LPS plus z-VAD to induce necroptosis. Remarkably, LPS/z-VAD induced more necroptosis in HOIP-deficient DCs (Fig. 4A, 4D), indicating that HOIP is required for prevention of necroptosis via TLR4. Interestingly, LPS-induced apoptosis and necroptosis were only partially rescued by the RIPK1 kinase inhibitor Necrostatin-1, implying that RIPK1 kinase-independent mechanism exists under LPS-induced cell death (Fig. 4C, 4D).

Because above data indicate that HOIP-deficient DCs were more sensitive to LPS-induced cell death, we hypothesized that reduced DCs in Hoip^{DC KO} mice could be the consequences of cell death induced directly by TLR ligands presented by commensal microbiota. To address the specific role of commensal microbiota in the development of inflammation in Hoip^{DC KO} mice, we depleted most of the commensal microbiota by treating 2-d-old mice with a broad spectrum of antibiotics for 7 wk. Following antibiotic treatment, Hoip^{DC KO} mice had reduced enlargement of spleens and mLNs (Fig. 4E). Furthermore, antibiotic treatment reduced the percentage and number of inflammatory monocytes and neutrophils in Hoip^{DC KO} mice (Fig. 4F, 4G). The reduction of DCs in Hoip^{DC KO} mice was partially recovered with depletion of commensal microbiota (Fig. 4H, 4I). Therefore, the systemic inflammation observed in untreated Hoip^{DC KO} mice was partially rescued in antibiotic-treated mice, suggesting the contribution of intestinal microbiota to the systemic inflammation in Hoip^{DC KO} mice.

To determine whether aberrant DC cell death is responsible for the systemic inflammation in Hoip^{DC KO} mice, we crossed Hoip^{DC KO} mice to Mlkl^−/− mice to generate Hoip^{DC KO} Mlkl^−/− mice, which would directly address the specific role of HOIP on MLKL-mediated necroptosis in DCs. We found that genetic ablation of MLKL alone failed to prevent systemic inflammation in Hoip^{DC KO} mice. Deletion of MLKL alone had little effect on inflammation of spleen and mLN, with Hoip^{DC KO} Mlkl^−/− mice exhibiting splenomegaly and lymphadenopathy (Supplemental Fig. 4A, 4B). Besides, inflammatory cell infiltration of spleens and mLNs observed in Hoip^{DC KO} Mlkl^−/− mice were comparable to Hoip^{DC KO} mice (Supplemental Fig. 4C, 4D).

Because Caspase-8–deficient mice are embryonically lethal because of sensitization to RIPK3- and MLKL-induced necroptosis (35–38), it is not possible to generate viable Casp8^−/− Hoip^{DC KO} mice to specifically address the role of apoptosis in Hoip^{DC KO} mice. We crossed Hoip^{DC KO} Mlkl^−/− to Casp8^+/−mice to generate Hoip^{DC KO} Casp8^−/− Mlkl^−/− mice to concurrently block both the Caspase-8–mediated apoptosis and MLKL-mediated necroptosis in Hoip^{DC KO} mice. Remarkably, although Hoip^{DC KO} Casp8^−/− Mlkl^−/− mice developed splenomegaly and lymphadenopathy like Casp8^−/− Mlkl^−/− mice as previously reported (39), the systemic inflammation associated with DC-HOIP deficiency was significantly alleviated (Supplemental Fig. 4E–G). Reduced numbers of inflammatory cell populations correlated with decreased level of inflammatory cytokine TNF-α in the sera of Hoip^{DC KO} Casp8^−/− Mlkl^−/− mice (Supplemental Fig. 4H). Because deletion of Caspase-8 and MLKL are assumed to prevent both apoptosis and necroptosis, we expected that cDC numbers would be rescued in Hoip^{DC KO} Casp8^−/− Mlkl^−/− mice. Indeed, decreased numbers of cDCs in Hoip^{DC KO} mice were reversed in the spleens of Hoip^{DC KO} Casp8^−/− Mlkl^−/− mice (Supplemental Fig. 4I), indicating DC death results in the systemic inflammation in Hoip^{DC KO} mice.

Because systemic deletion of Caspase-8 and MLKL may alleviate systemic inflammation of Hoip^{DC KO} mice by suppressing cell death of cells other than DCs, including both hematopoietic and nonhematopoietic cells, we further investigate whether Caspase-8 and MLKL in hematopoietic cells mediate systemic inflammation in Hoip^{DC KO} mice. We transferred bone marrow cells from CD45.2⁺ control mice, CD45.2⁺ Hoip^{DC KO} mice, CD45.2⁺ Casp8^−/− Mlkl^−/− mice, and CD45.2⁺ Hoip^{DC KO} Casp8^−/− Mlkl^−/− mice into irradiated CD45.1⁺ WT recipient mice. In contrast to mice transferred with bone marrow cells from Hoip^{DC KO} mice, the inflammation in mice transferred with bone marrow cells from Hoip^{DC KO} Casp8^−/− Mlkl^−/− mice was significantly alleviated, as exhibited by decreased infiltration of neutrophils and inflammatory monocytes in mice (Fig. 5B–E). Furthermore, the reduction of DCs in mice transferred with bone marrow cells from Hoip^{DC KO} mice was reversed in mice transferred with bone marrow cells from Hoip^{DC KO} Casp8^−/− Mlkl^−/− mice (Fig. 5F, 5G). Thus, aberrant apoptosis and necroptosis lead to systemic inflammation in Hoip^{DC KO} mice.

Cell death is predicted to result in the release damage-associated molecular patterns (DAMPs) and cytokines that are capable of activating other innate cells, resulting in inflammation (40). MyD88 is an essential adaptor molecule for TLRs and IL-1R, and activation of MyD88-dependent TLR signaling elicits the production of several proinflammatory cytokines, including TNF-α, IL-1, and IL-6 (41, 42). Besides, MyD88 signaling can also be activated by endogenous DAMPs (e.g., IL-1, IL-18) (43, 44). To investigate the involvement of MyD88 signaling in systemic inflammation, we crossed Hoip^{DC KO} mice to Myd88^−/−mice to generate Hoip^{DC KO} Myd88^−/− mice. Interestingly, Hoip^{DC KO} Myd88^−/− mice almost completely rescued the HOIP deficiency–associated enlargement of spleen and mLN (Fig. 6A). Furthermore, the inflammatory monocytes and neutrophils in the Hoip^{DC KO} Myd88^−/− mice also reduced to similar cell numbers as control mice (Fig. 6B). Accordingly, the elevated levels of TNF-α in Hoip^{DC KO} mice were reduced in Hoip^{DC KO} Myd88^−/− mice (Fig. 6C). However, analysis of DCs in Hoip^{DC KO} Myd88^−/− mice revealed that the numbers of DCs were still significantly decreased compared with those in control and Myd88^−/− mice (Fig. 6D). Consistent with decreased DCs in Hoip^{DC KO} Myd88^−/− mice, BMDCs from Hoip^{DC KO} Myd88^−/− mice are still susceptible to either TNF-α– or LPS-induced cell death (Fig. 6E). These results demonstrate that MyD88 is not responsible for the DC reduction because DCs still die in the absence of MyD88. However, MyD88 is required for the development of inflammation observed in Hoip^{DC KO} mice.

Because MyD88 deficiency is not responsible for DC reduction in Hoip^{DC KO} mice, we further assessed IL-1α in Hoip^{DC KO} mice and Hoip^{DC KO} Myd88^−/− mice considering the essential function of MyD88 in IL-1R signal. Interestingly, an increase in basal serum level of IL-1α was observed in Hoip^{DC KO} mice (Fig. 6F). Furthermore, LPS challenge could significantly enhance IL-1α serum level, which could be abrogated by MyD88 deficiency (Fig. 6F). Consistently, increased secretion of IL-1α and IL-1β was detected in BMDCs from Hoip^{DC KO} mice upon LPS challenge, which was diminished with MyD88 deficiency (Fig. 6G). Further analysis of mRNA of Il1a and Il1b revealed that the enhanced secretion of IL-1α and IL-1β was not due to an enhanced transcription in Hoip^{DC KO} cells (Fig. 6H), and enhanced cleavage of procaspase-1 and pro–IL-1β was detected in BMDC upon LPS challenge (Fig. 6I), indicating a negative regulation of IL-1α/IL-1β secretion by HOIP. Together, these results indicate that MyD88-dependent signaling plays crucial roles in the development of systemic inflammation in Hoip^{DC KO} mice.

Investigating the mechanism underpinning the autoinflammation is critical to understand the pathogenesis of patients with loss-of-function mutations in LUBAC components. Our study evaluated the impact of LUBAC deficiency on DC homeostasis by genetically ablating HOIP, the catalytically active component of LUBAC, in mouse DCs. Intriguingly, LUBAC deficiency in DCs results in the spontaneous development of systemic inflammation resembled to inflammation in patients. Additionally, this elevated systemic inflammation of Hoip^{DC KO} mice was distinct from other tissue-specific HOIP deficiency mice that exhibited no inflammation or localized inflammation (7, 8, 17, 25, 45, 46). In contrast to mice with T cell–specific or B cell–specific HOIP deficiency, both mice exhibit no obvious systemic inflammation; Hoip^{DC KO} mice spontaneously develop systemic inflammation at the age of 6–8 wk. Thus, LUBAC in DCs is indispensable for maintaining homeostasis and prevention of autoinflammation.

Previous studies showed that cell death can be the cause of inflammation and inflammation-associated diseases (47), and the absence of HOIP in other cell types caused marked sensitivity to cell death (8, 25, 45, 46). The results presented in this article provide additional evidence in support of cell death as an etiology of inflammation-associated diseases. Our study found that HOIP-deficient DCs were more susceptible to both apoptotic and necroptotic cell death in response to TNF-α or LPS stimulation.

Thus, the observed autoinflammatory phenotype in Hoip^{DC KO} mice presumably results from elevated levels of cell death, especially necroptosis (47, 48), upon LUBAC deficiency. In mice with HOIP deficiency, concomitant blocking of both the Caspase-8–mediated apoptosis and MLKL-mediated necroptosis by systemic deletion of both Caspase-8 and MLKL prevents cell death in Hoip^−/− embryos and embryonic lethality, resulting in viable Hoip^−/− Casp8^−/− Mlkl^−/− mice (11). In mice with keratinocyte-specific HOIP deficiency, systemic deletion of both Caspase-8 and MLKL could prevent cell death–associated dermatitis and postnatal lethality (46). Interestingly, our study found that concomitant blocking of both the Caspase-8–mediated apoptosis and MLKL-mediated necroptosis alleviates the systemic inflammation associated with a DC-HOIP deficiency, and DC numbers in Hoip^{DC KO} mice were reversed to the amount of WT mice by codeletion of Caspsase-8 and MLKL, suggesting that cell death of DCs is responsible for the inflammatory phenotype of Hoip^{DC KO} mice. Because various cell types from Casp8^−/− Mlkl^−/− mice were resistant to diverse necroptotic cell death stimuli, and Casp8^−/− Mlkl^−/− mice rapidly developed splenomegaly, lymphadenopathy, and autoimmune manifestations, bone marrow chimeric mice were employed to specifically delete Caspase-8 and MLKL in hematopoietic cells to confirm that preventing cell death in a hematopoietic system can rescue the inflammation in Hoip^{DC KO} mice. In the future, it would be better to delete Caspase-8 and MLKL in a DC-specific manner by using Itgax-Cre Hoip^flox/flox Casp8^flox/flox Mlkl^flox/flox to fully substantiate the involvement of DC cell death in systemic inflammation of Hoip^{DC KO} mice.

Apart from Caspase-8 and MLKL, a recent study reported that LUBAC interacts with Caspase-1 to generate linear ubiquitination on the CARD domain of Caspase-1 to regulate its function, and depletion of HOIP or Sharpin resulted in heightened Caspase-1 activation, IL-1β secretion, and pyroptosis (49, 50). Our study found that enhanced IL-1α and IL-1β secretion were detected in Hoip^{DC KO} cells upon LPS challenge, indicating the involvement of Caspase-1 in the regulation of cell death of Hoip^{DC KO} cells in response to LPS. Thus, it would be worth trying to explore the role of Caspase-1 in Hoip^{DC KO} cell death and following inflammation in Hoip^{DC KO} mice. Because simultaneous Ripk3 deletion and Casp8 heterozygosity suppressed dermatitis of Sharpin^cpdm/cpdm mice (51), and genetic ablation of Caspase-1 from Sharpin^cpdm/cpdm mice significantly reduced skin inflammation (52), concomitant prevention of pyroptosis, apoptosis, and necroptosis (PANoptosis) is likely to completely rescue inflammation in Hoip^{DC KO} mice.

Intriguingly, although both TNF-α and TLR4 stimulation can induce the cell death of HOIP-deficient DCs in vitro, the absence of TNFR1 signaling and TLR4 stimulation in vivo led to distinct results. Unexpectedly, unlike observations in keratinocyte-specific HOIP deficiency mice (25, 46) and HOIP-deficient embryos(10), the systemic inflammation in Hoip^{DC KO} mice could not be rescued by depleting TNFR1. One possibility is that TNFR1 signaling from non-DCs might also be involved in the development of inflammation in Hoip^{DC KO} mice. Therefore, it would be interesting to investigate the role of DC-specific TNFR1 signaling in aberrant cell death and inflammation driven by HOIP deficiency in DCs. Of note, although RIPK1-deficient DCs undergo TNF-α–mediated necroptosis in vitro, the absence of TNFR1 had no detectable effects on splenic inflammation in Ripk1^{DC KO} mice (53). Contrary to other findings that TNF-α promotes inflammation, however, our study shows that TNFR1 deficiency exacerbates the systemic inflammation of Hoip^{DC KO} mice as they developed symptoms of cachexia. Thus, the detailed molecular mechanism of TNFR1 in DC death and following inflammation needs to be further investigated.

Unlike TNFR1 signaling, TLR ligands from commensal microbiota are likely the source of signals to induce the cell death of HOIP-deficient DCs and inflammation of Hoip^{DC KO} mice, and depletion of microbiota by administration of broad-spectrum antibiotics partly abrogated the inflammation in Hoip^{DC KO} mice. In HOIP-deficient DCs, TLR4 stimulation can induce both enhanced apoptosis and necroptosis. Thus, the explanation for the involvement of intestinal microbiota in the pathogenesis of Hoip^{DC KO} mice could be that TLR ligands directly induce the cell death of HOIP-deficient DCs and consequently release DAMPs to induce inflammation. Of note, although a study by Sharma et al. (54) has demonstrated that depletion of gut microbiota by administration of antibiotics partially rescued skin inflammation and systemic inflammation of Sharpin^cpdm/cpdm mice, the detailed mechanism is not clear. Our results provide additional evidence in support of the role of commensal microbiota in the pathogenesis of inflammation associated with LUBAC component mutation and a possible mechanism for the regulation of TLR signal by LUBAC in DCs to prevent cell death and inflammation.

MyD88 is an essential adaptor for both TLRs and IL-1R, and previous studies have shown that MyD88 is involved in inflammation triggered by necroptotic cells (22, 55). Our data show that the inflammatory phenotype in Hoip^{DC KO} mice was absent in Hoip^{DC KO} Myd88^−/− mice. However, the absence of MyD88 did not rescue the death of DCs in Hoip^{DC KO} mice as shown by decreased splenic DCs and susceptible death of BMDCs generated from Hoip^{DC KO} Myd88^−/− mice. Thus, the abrogation of inflammation by MyD88 is not through prevention of DC cell death directly but relies on other indirect mechanisms.

Considering MyD88 mediates signal transduction for both most TLRs and IL-1R, the absence of inflammatory responses in the Hoip^{DC KO} Myd88^−/− mice may be the outcome of defective TLR signaling or IL-1R signaling. Because administration of broad-spectrum antibiotics partially rescued the inflammation of Hoip^{DC KO} mice and MyD88 deficiency abrogated the secretion of IL-1α/IL-1β induced by LPS, one explanation for the abrogation of inflammation in the Hoip^{DC KO} Myd88^−/− mice could be defective TLR signaling. Apart from TLR signaling, IL-1 signaling has been reported to be responsible for the inflammation induced by necrotic cells; thus mice deficiency of MyD88 failed to recruit neutrophils in response to dead cells (55). In our study, deficiency of MyD88 may abrogate the IL-1 signaling induced by cell death, contributing to the prevention of inflammation in Hoip^{DC KO} mice. In the case of Hoip^{DC KO} mice, we suggest that commensal microbes initiate innate immune signaling via the TLR/MyD88 axis, which leads to IL-1 production. Besides, TLR ligands from microbes can also induce cell death of DCs to release DAMPs like IL-1α. All above signaling via the IL-1R/MyD88 axis results in the development of chronic inflammation. Considering the essential function of MyD88 for both TLR4 and IL-1R, separately investigating the role of TLR4 deletion and IL-1R deletion in inflammation of Hoip^{DC KO} mice would be worth trying in the future.

It would therefore be tempting to speculate that HOIP deficiency in the DCs could mimic the conditions in the patients with loss-of-function mutations of LUBAC components and could possibly serve as a model for its etiology. The pathologies found in the Hoip^{DC KO} mice appear to be driven by aberrant cell death, which is often seen in LUBAC-mutated patients. However, blocking of TNFR1 signaling by anti-TNF treatment in LUBAC-mutated patients may not be as effective as several other autoinflammatory and autoimmune disorders, including rheumatoid arthritis and psoriasis (56, 57), as deficiency of TNFR1 signaling could not alleviate disease development. Our findings that commensal microbiota and MyD88-dependent signaling are required for development of inflammation in Hoip^{DC KO} mice suggest that modulation of microbiota may serve as a therapeutic method to attenuate the clinical autoinflammatory diseases because of mutations in the LUBAC complex. In conclusion, our findings suggest that LUBAC is essential for regulating DC survival, that disturbance of this regulation results in the development of systemic inflammation resembling autoinflammation in LUBAC-mutated patients, and that modulation of commensal microbiota may serve as a therapeutic method for these patients.

We thank Drs. Xiaodong Wang and Xiaoyu Hu for the generous sharing of Mlkl^−/−mice and Myd88^−/−mice, respectively.

The authors have no financial conflicts of interest.