Main

RIPK1 is a key mediator of apoptotic and necrotic cell death as well as inflammatory pathways1. Activation of RIPK1 promotes several cell death responses, including apoptosis and necroptosis, downstream of TNFR1. Caspase-8-mediated cleavage after Asp324 in human RIPK1 (or Asp325 in mouse RIPK1) separates the kinase domain in the N-terminal part of RIPK1 from its intermediate and death domains. The death domain is involved in mediating the activation of the N-terminal kinase by dimerization1,4,5. The D324A variant in human RIPK1 blocks cleavage by caspase-86. Homozygous D325A mutation in mouse RIPK1 sensitizes cells to both apoptosis and necroptosis induced by TNF and leads to embryonic lethality. The early demise of Ripk1D325A/D325A mice can be rescued by simultaneous deletion of Ripk3 and Fadd2, or Mlkl and Fadd3, but not of either gene alone. However, the functional importance of caspase-8-mediated cleavage of RIPK1 in humans is unknown. Here, we identified a human autoinflammatory disease caused by non-cleavable RIPK1 variants with mutations at D324. We show that disrupted cleavage of RIPK1 by caspase-8 in humans leads to a dominantly inherited condition by promoting the activation of RIPK1.

Patients with RIPK1 non-cleavable variants

The first patient (P1) is a two-year-old Chinese boy. His symptoms began at two months of age with periodic fever episodes occurring every eight to ten days and lasting for three to five days (Fig. 1a, b, Extended Data Table 1). His fevers were associated with increased levels of C-reactive protein and white blood cell counts, but no other accompanying symptoms. He developed lymphadenopathy at two years of age (Fig. 1c). The patient did not have a skin rash, arthritis, arthralgia or hepatosplenomegaly. Lymphocyte phenotyping revealed increased counts of both double-negative T cells and naive B cells (Extended Data Table 2).

Fig. 1: Heterozygous variants at the RIPK1 cleavage site cause autoinflammatory disease in humans.
figure 1

a, Pedigrees of two families with variants in RIPK1 at the caspase-8 cleavage site. b, Timeline of recurrent fever episodes in P1 over 4 months. Red dots denote increased temperatures during fever episodes. Blue boxes denote normal temperatures between flares. c, Computerized tomography scans of P1 (top) and sonographic image of P3 (bottom) show lymphadenopathy (arrows) and splenomegaly, respectively.

Full size image

The second family is of European Canadian ancestry. The proband (P2) is a 35-year-old female who experienced recurrent fevers from six months of age, and developed intermittent lymphadenopathy, hepatosplenomegaly and microcytic anaemia. Three of her four sons are affected. Her eldest (P3, 14 years of age) and youngest (P5, 10 years of age) sons have a similar history of recurrent fevers, intermittent lymphadenopathy, splenomegaly and microcytic anaemia. Her second son (P4, 12 years of age) has microcytic anaemia but no history of recurrent fevers (Fig. 1a, c, Extended Data Table 1).

Whole-exome sequencing (WES) of P1 and his parents revealed that P1 has a heterozygous de novo D324V mutation in RIPK1 (Fig. 1a, Extended Data Fig. 1a–c). For the second family, WES identified a single variant, D324H in RIPK1, which is de novo in the mother and inherited by her three affected sons (Fig. 1a, Extended Data Fig. 1c). No other mutations—including rare variants of unknown importance in genes known to cause periodic fever or autoinflammatory syndromes—were found (Supplementary Tables 1, 2). Copy number variant analysis based on WES data for the first family, and microarray analysis for the second family, did not identify any copy number variants among affected individuals. Both variants affected the caspase-8 cleavage site, D324, which is highly conserved in RIPK1 across species (Extended Data Fig. 1d, e). These variants were not reported in any public database of human exomes and were predicted to be deleterious (combined annotation-dependent depletion (CADD) score > 20) for protein function by computational in silico modelling (Extended Data Fig. 1f).

Expression of wild-type and mutant RIPK1 in HEK293T cells indicated that variants at residue D324, including D324V and D324H, blocked the cleavage of RIPK1 by caspase-8 (Extended Data Fig. 1g). D325A mutation in mouse RIPK1 (the equivalent residue for D324 in human RIPK1) did not affect its turnover (Extended Data Fig. 1h), or block its interaction with other proteins such as binding with caspase-8 into the FADDosome complex (Extended Data Fig. 1i). The variants in D324 resulted in non-cleavable RIPK1, which was directly demonstrated by incubating mutant RIPK1 generated by TNT cell-free protein expression with recombinant caspase-8 (Extended Data Fig. 1j). The inhibitory effect of the D324V variant on the cleavage of RIPK1 by caspase-8 was further confirmed in patient P1 fibroblasts after stimulation by TNF and cycloheximide (CHX) (Extended Data Fig. 1k).

Activation of inflammatory signalling in the patients

We detected markedly increased production of pro-inflammatory cytokines and chemokines such as IL-6, TNF and IFNγ, and anti-inflammatory cytokines such as IL-10 in serum from patients by cytometric bead array (Fig. 2a) or enzyme-linked immunosorbent assay (ELISA) (Extended Data Table 3). Serial sampling from P1 showed that activation of inflammatory responses was even more notable during fever episodes (Fig. 2a). Increased expression of IL-6, TNF and IL-8 in monocytes and IL-6 in T cells from P1 after stimulation by lipopolysaccharide (LPS) was detected by intracellular cytokine staining (Extended Data Fig. 2a, b). Moreover, phosphorylation of STAT3, the downstream marker of IL-6 signalling, was upregulated during fever episodes in patient monocytes at basal level when compared with unaffected controls (Extended Data Fig. 2c). We also observed increased phosphorylation of MAPK p38 in patient monocytes, B cells and T cells after LPS stimulation (Extended Data Fig. 2c).

Fig. 2: Strong activation of inflammatory signalling in patient P1.
figure 2

a, Serum levels of cytokines IL-6, TNF and IL-10 from patient P1 (red denotes serum during fever episodes; blue denotes serum during remission) determined by cytometric bead array. b, Left, uniform manifold approximation and projection (UMAP) of 18,928 cells, split between patient P1 and an age- and sex-matched unaffected control (C1) after alignment. Right, UMAP visualization and marker-based annotation of 16 cell subtypes, coloured by cluster identity. The patient P1 displayed higher percentage of monocytes (red frame). Eryth, erythrocytes; Mk, megakaryocyte; NK, natural killer cell; pDC, plasmacytoid dendritic cell. c, Visualization of expression of IL8 and IL1B (coloured single cells) on UMAP plot projecting PBMCs from patient P1 (n = 7,936 cells) and an age- and sex-matched unaffected control (n = 10,992 cells). d, RNA-sequencing analysis of cell death, NF-κB and type I IFN pathways in patient PBMCs compared with three paediatric unaffected controls (C1–C3). Analysis of each sample was performed in duplicate. For gene names, see Supplementary Fig. 2. e, qPCR analysis of cytokine and chemokine-related genes in PBMCs from P1 compared with five paediatric unaffected controls (C). Data are mean ± s.e.m. Circles correspond to each tested individual. Analysis of each sample was performed in triplicate. The PBMCs from patient P1 in b–e were obtained during fever episodes.

Source Data

Full size image

To study the transcriptional changes related to non-cleavable RIPK1 further, we performed single-cell RNA sequencing in patient peripheral blood mononuclear cells (PBMCs). The patient had a higher percentage of monocytes compared with an age- and sex-matched unaffected control (Fig. 2b, Extended Data Fig. 2d). We observed strong signals in both NF-κB and type-I IFN inflammatory pathways in the patient monocytes (Extended Data Fig. 2e, f). The patient monocytes highly expressed pro-inflammatory cytokines and chemokines, including IL8, IL1B and CCL3 (Fig. 2c, Extended Data Figs. 2g, 3a, b). In addition, RNA sequencing in PBMCs implicated different gene expression patterns in cell death pathways that include increased expression of RIPK3 and MLKL, suggesting increased levels of necroptosis machinery in the patient PBMCs (Fig. 2d). Quantitative PCR (qPCR) confirmed the increased expression of IL6, TNF, IL8, IL1B, CXCL2 and CXCL3 in patient PBMCs (Fig. 2e). Supporting a pathogenic role of excessive IL-6 production, P1 experienced clinical improvement and the PBMCs displayed normalized expression of inflammatory mediators after treatment with tocilizumab (monoclonal antibody against IL-6R) (Extended Data Fig. 4a).

Increased cell death and inflammatory response

We examined the response of patient PBMCs to TNF by measuring cell survival with the CellTiter-Glo assay, and quantified cell death by measuring the plasma membrane permeability with ToxiLight assay (Fig. 3a, Extended Data Fig. 4b). The PBMCs from patients P1, P2 and P3 showed increased sensitivity to both apoptosis induced by co-treatment with TNF and apoptosis-inducing SMAC mimetic SM-164, and necroptosis induced by co-treatment with SM-164 and the pan-caspase inhibitor Z-VAD-FMK (carbobenzoxy-valyl-alanyl-aspartyl-[O-methyl]-fluoromethylketone) compared with PBMCs from unaffected controls. Co-treatment was required as treatment with these compounds individually did not elicit cell death. Furthermore, both apoptosis and necroptosis of patient PBMCs were effectively suppressed by the RIPK1 inhibitor necrostatin-1s (Nec-1s) (Fig. 3a, Extended Data Fig. 4b). We found that levels of RIPK1 phosphorylated at S166 (p-S166-RIPK1)—a marker for the activation of RIPK17,8—were increased in the patient PBMCs treated with various combinations of these compound known to activate apoptosis or necroptosis, compared to that of controls (Fig. 3b, Extended Data Fig. 4c), which suggests that blocking the cleavage of RIPK1 sensitizes the activation of its kinase activity. We also found increased levels of p-S358-MLKL—a biomarker for necroptosis9—in the patient PBMCs treated with SM-164 plus Z-VAD-FMK compared with that of control cells (Fig. 3b). Notably, patient PBMCs treated with TNF and SM-164 showed increased levels of not only cleaved caspase-3, but also p-S358-MLKL, which were both effectively reduced by treatment with Nec-1s (Fig. 3b). These results suggest that the non-cleavable RIPK1 variant sensitized the patient PBMCs to both necroptosis and apoptosis in a RIPK1-dependent manner.

Fig. 3: RIPK1 cleavage site variants promote cell death and inflammatory response induced by TNF in patient PBMCs and MEFs.
figure 3

a, Cell viability (as measured by CellTiter-Glo assay) of PBMCs from patients and eight paediatric unaffected controls after treatment as indicated for 24 h. N, Nec-1s; S, SM-164; T, TNF; Z, Z-VAD-FMK. Data are mean ± s.e.m. Circles indicate one sample from each individual (P1 was sampled three times). Analysis of each sample was performed in triplicate. b, Western blots of PBMCs from patient P1 and a paediatric unaffected control after treatment as indicated for 24 h. −, untreated; cl, cleaved. For gel source data, see Supplementary Fig. 1. Results are representative of two independent experiments. c, Western blots of urine samples from P1 during a fever episode (red) and remission (blue) and three paediatric unaffected controls. Supernatant (sup.) of fibroblasts from an unaffected control stimulated with TNF, SM-164 and Z-VAD-FMK (TSZ) served as a positive control. For gel source data, see Supplementary Fig. 1. Results are representative of three independent experiments. d, NanoString analysis of PBMCs from patient P1 and three paediatric unaffected controls after stimulation as indicated. For gene names, see Supplementary Fig. 2. e, qPCR analysis of IL6 mRNA levels of PBMCs from patient P1 and four unaffected controls treated as indicated. Data are mean ± s.e.m. Circles correspond to each tested individual. Analysis of each sample was performed in triplicate. The PBMCs in a, b, d and e were obtained during remission. f, Cell viability of Ripk1-knockout MEFs complemented with: GFP; wild-type (WT) RIPK1; D325V, D325H or D138N mutant; or D138N/D325H or D138N/D325V double mutants, treated as indicated for 24 h. Data are mean ± s.e.m., n = 3. Circles correspond to each independent experiment. P values determined by unpaired two-tailed t-test (shown if P < 0.05). g, h, Western blots of Ripk1-knockout MEFs complemented with: GFP; wild-type RIPK1; D325V, D325H or D138N mutant; or D138N/D325H or D138N/D325V double mutants, treated as indicated. HA, haemagglutinin; LE, long exposure; SE, short exposure. For gel source data, see Supplementary Fig. 1. Results are representative of three independent experiments. i, Cell viability of Ripk1D325A/D325ARipk3−/− and Ripk1+/+Ripk3−/− MEFs treated as indicated for 24 h. Data are mean ± s.e.m., n = 3. Circles correspond to each independent experiment. P values determined by unpaired two-tailed t-test. j, Western blots of Ripk1D325A/D325ARipk3−/− and Ripk1+/+Ripk3−/− MEFs treated as indicated. For gel source data, see Supplementary Fig. 1. Results are representative of three independent experiments. k, Il6 mRNA expression of Ripk1-knockout MEFs complemented with: GFP; wild-type RIPK1; or D325V or D325H mutant, treated as indicated. Data are mean ± s.e.m., n = 3. Circles correspond to each independent experiment. P values determined by unpaired two-tailed t-test. l, m, qPCR analysis of Il6 and Cxcl2 (l) or Il6, Cxcl2 and Tnf (m) expression in Ripk1D325A/D325ARipk3−/− and Ripk1+/+Ripk3−/− MEFs treated as indicated for 2 or 4 h. Data are mean ± s.e.m., n = 3. Circles correspond to each independent experiment. P values determined by unpaired two-tailed t-test.

Source Data

Full size image

Release of cyclophilin A is a biomarker for necroptosis in cell-based assays and has also been implicated as a potential biomarker in human diseases10,11. We detected the presence of cyclophilin A in a urine sample from a patient during a fever episode but not in remission, which provides evidence for enhanced necrotic cell death in the setting of inflammation in vivo (Fig. 3c). These findings indicate that the non-cleavable RIPK1 variant may promote the activation of RIPK1, which leads to necrotic cell death in vivo.

Activation of necroptosis promotes a strong inflammatory response such as the production of pro-inflammatory cytokines12. Compared to that of control PBMCs, the patient PBMCs stimulated with TNF plus SM-164 showed an exacerbated inflammatory response, which was effectively inhibited by the RIPK1 inhibitor Nec-1s (Fig. 3d). Confirming the involvement of RIPK1 kinase activity in promoting the inflammatory responses, we found that the increased IL6 expression owing to cell death induced by TNF plus SM-164 stimulation in patient PBMCs was suppressed by Nec-1s (Fig. 3e).

The patient data raised the possibility that non-cleavable RIPK1 variants function directly in promoting its own activation, which in turn mediates apoptosis and necroptosis in a signal-dependent manner. To test this possibility experimentally, we expressed the cleavage site D325V and D325H RIPK1 mutants in Ripk1-knockout mouse embryonic fibroblasts (MEFs). Compared to that of Ripk1-knockout MEFs and Ripk1-knockout MEFs complemented with wild-type RIPK1, MEFs expressing the D325V or D325H RIPK1 mutant were consistently hypersensitive to cell death induced by TNF, which was inhibited by the addition of Nec-1s. The enhanced cell death could also be blocked by introducing a RIPK1 kinase inactivation mutation, D138N, in cis with D325V or D325H, providing direct evidence for the role of RIPK1 kinase activity in promoting cell death (Fig. 3f, Extended Data Fig. 5a, b). Similar to patient PBMCs, MEFs expressing D325V or D325H mutant RIPK1 stimulated by TNF alone or TNF plus SM-164 showed increased levels of p-S166-RIPK1 (Fig. 3g). By contrast, stimulation of Ripk1-knockout MEFs complemented with wild-type RIPK1 with TNF alone was not sufficient to promote the activation of RIPK1 (Fig. 3g). These data support the hypothesis that the non-cleavable variants of RIPK1 directly promote the activation of RIPK1.

Similar to that of patient PBMCs, RIPK1(D325V)- or RIPK1(D325H)-complemented Ripk1-knockout MEFs stimulated by TNF or TNF plus SM-164 showed increased levels of cleaved caspase-3 compared to that of wild-type-complemented MEFs, which was inhibited by Nec-1s and by the kinase inactivation mutation D138N in cis with D325V or D325H construct (Fig. 3g, h, Extended Data Fig. 5c). Also similar to that of patient PBMCs, the stimulation of Ripk1-knockout MEFs expressing D325V or D325H RIPK1 mutant with TNF alone or TNF plus SM-164 induced increased levels of p-S345-MLKL (Fig. 3g, Extended Data Fig. 5c). By contrast and as expected, stimulation of Ripk1-knockout MEFs or Ripk1-knockout MEFs complemented with wild-type RIPK1 with TNF alone or TNF plus SM-164 was not sufficient to promote the activation of necroptosis and appearance of p-S345-MLKL. TNF-induced cell death and the appearance of p-S345-MLKL in D325V- or D325H-complemented Ripk1-knockout MEFs were both blocked by Nec-1s and by the inactivation D138N mutation in cis with D325V or D325H (Fig. 3f–h, Extended Data Fig. 5c). These results suggest that D325V and D325H are gain-of-function mutations in RIPK1 that promote the activation of its kinase, which in turn mediates apoptosis and necroptosis.

Because the expression of non-cleavable RIPK1 promotes both apoptosis and necroptosis, we next determined whether these two forms of cell death might be independent of each other by examining Ripk1D325A/D325ARipk3−/− MEFs from Ripk1D325A/D325A knock-in mice crossed with necroptosis-deficient Ripk3−/− mice2. Notably, we found that Ripk1D325A/D325ARipk3−/− MEFs remained sensitized to apoptosis induced by TNF alone and TNF plus SM-164 and showed increased levels of p-S166-RIPK1 and cleaved caspase-3, which are both inhibited by Nec-1s (Fig. 3i, j, Extended Data Fig. 5d, e). Thus, the activated RIPK1 in cells expressing non-cleavable RIPK1 is able to drive RIPK1-dependent apoptosis, independently of necroptosis.

We also examined the effect of non-cleavable RIPK1 on ligands of other death receptors such as TRAIL13. We found that the cells expressing the non-cleavable mutant RIPK1 also showed increased sensitivity to TRAIL-induced cell death, which could be rescued by the addition of Nec-1s (Extended Data Fig. 5f). Levels of p-S166-RIPK1 were also increased after TRAIL stimulation in mutant RIPK1-complemented MEFs compared to wild-type RIPK1-complemented MEFs (Extended Data Fig. 5g). These data further illustrated that the non-cleavable mutations in RIPK1 increase RIPK1 kinase activity and sensitize the cells to cell death after stimulation by several stimuli.

We next characterized the effect of non-cleavable RIPK1 on cytokine production. NanoString analysis of the patient PBMCs stimulated with TNF alone exhibited upregulated gene expression in the inflammatory pathway, including IL6, which was reduced by Nec-1s (Extended Data Fig. 4d). Because patient P1 responded well to IL-6 blockade, we also compared the effects of Il6 expression in Ripk1-knockout MEFs expressing GFP alone, wild-type RIPK1 or mutant RIPK1 (D325V or D325H) (Fig. 3k). We found that MEFs expressing the D325V or D325H mutant showed distinctively enhanced transcription of Il6 compared to that of wild-type-complemented Ripk1-knockout MEFs in response to TNF alone. In addition, the transcriptional production of Il6, Cxcl2 and Tnf were also enhanced in Ripk1D325A/D325ARipk3−/− MEFs after stimulation by TNF or both TNF and SM-164. The enhancement was inhibited by the addition of Nec-1s (Fig. 3l, m). Together, these results suggest that the augmented inflammatory signals associated with the non-cleavable variants were dependent on RIPK1 kinase activity. In keeping with the patient’s therapeutic response to IL-6 blockade, these results demonstrate a pathogenic mechanism that relies on the activation of RIPK1 to mediate the production of IL-6.

Necroptosis and ferroptosis resistance in fibroblasts

Notably, we observed the opposite response to inducers of cell death in fibroblasts from patient P1 compared with MEFs and patient PBMCs. The patient fibroblasts showed resistance to cell death after stimulation with TNF or LPS plus SM-164 and Z-VAD-FMK (Fig. 4a, Extended Data Fig. 6a). The cell death resistance was further demonstrated by reduced phosphorylation of RIPK1 and MLKL in the patient fibroblasts (Fig. 4b, Extended Data Fig. 6b, c). Patient fibroblasts also showed diminished gene expression of IL6, IL1B and pro-inflammatory chemokines CXCL2 and CXCL3 in response to TNF, SM-164 and Z-VAD-FMK stimulation (Fig. 4c). Levels of RIPK1 protein under basal and stimulated conditions were lower in patient fibroblasts than that of controls (Fig. 4d), and the reduction was rescued by the Nec-1s (Extended Data Fig. 6d). We observed reduction of RIPK1 at the transcriptional level in fibroblasts (Extended Data Fig. 4e, Extended Data Fig. 6e). Together, these results suggest that decreased RIPK1 expression may compensate for the presence of the RIPK1-activating variant in patient fibroblasts. In addition, we found that patient fibroblasts exhibited reduced expression of TNFR1, which may provide a further mechanism for the decreased sensitivity to TNF (Fig. 4d, Extended Data Fig. 6f). Patient fibroblasts also showed decreased expression of genes involved in cell death pathways such as RIPK1 and RIPK3 and a different gene expression pattern (Extended Data Fig. 6g) compared to PBMCs (Fig. 2d). Together, these findings provide evidence of compensatory mechanisms to resist cell death in the patient fibroblasts.

Fig. 4: Necroptosis and ferroptosis are repressed in the patient fibroblasts.
figure 4

a, Cell viability of fibroblasts from P1 and seven paediatric unaffected controls after treatment with indicated stimulation for 24 h. N, Nec-1s; S, SM-164; T, TNF; Z, Z-VAD-FMK. Data are mean ± s.e.m. Circles correspond to each tested individual. Analysis of each sample was performed in triplicate. b, Western blots of fibroblasts from patient P1 and four paediatric unaffected controls after treatment with indicated stimulation for 6 h. For gel source data, see Supplementary Fig. 1. Results are representative of three independent experiments. VAD, Z-VAD-FMK. c, Patient and seven paediatric unaffected control fibroblasts were treated as indicated for 6 h. The mRNA levels of cytokines were measured by qPCR. Data are mean ± s.e.m. Circles correspond to each tested individual. Analysis of each sample was performed in triplicate. d, Western blots of RIPK1 and TNFR1 protein levels at basal state and after TNF stimulation in fibroblasts from P1 and four paediatric unaffected controls. For gel source data, see Supplementary Fig. 1. Results are representative of three independent experiments. e, Cell viability of patient fibroblasts treated with erastin, RSL3 or FIN56 compared with six paediatric unaffected controls. Data are mean ± s.e.m. Circles correspond to each tested individual. Analysis of each sample was performed in triplicate. f, Western blots of fibroblasts from P1 and three paediatric unaffected controls after treatment with erastin for 4 or 8 h. For gel source data, see Supplementary Fig. 1. Results are representative of three independent experiments. g, Expression patterns of genes involved in ferroptosis and antioxidant by RNA sequencing of fibroblasts from patient and three paediatric unaffected controls. Analysis of each sample was performed in duplicate. For gene names, see Supplementary Fig. 2. h, GSH concentrations in fibroblasts from P1 compared with six paediatric unaffected controls at baseline and after treatment with erastin or glutamate for 8 h. Data are mean ± s.e.m. Circles correspond to each tested individual. Analysis of each sample was performed in triplicate. i, Immunofluorescence (left) and relative fluorescent intensity (right) of cytosolic ROS (green foci) in patient fibroblasts after treatment with erastin for 8 h compared with that of three paediatric unaffected controls. Scale bar, 150 μm. Circles correspond to each tested individual sample. Analysis of each sample was performed in duplicate.

Source Data

Full size image

We also characterized the sensitivity of patient fibroblasts to other cell death stimuli. Notably, we found that the patient fibroblasts were highly protected against ferroptosis induced by erastin, RSL3 or FIN56 (Fig. 4e)—an effect that was not found in patient PBMCs or MEFs expressing the RIPK1 D325V or D325H mutant (Extended Data Fig. 4f, Extended Data Fig. 5h). Consistent with these findings, erastin-induced degradation of GPX414 was blocked in patient fibroblasts (Fig. 4f), but not in Ripk1−/− MEFs expressing mutant RIPK1 (Extended Data Fig. 5i). To explore the mechanism of ferroptosis resistance, we analysed gene expression in patient fibroblasts by RNA sequencing. We found that the expression of several genes involved in inhibiting ferroptosis—such as SLC7A11, CISD1 and CD4415—were upregulated in patient fibroblasts (Fig. 4g). This pattern was not observed in the patient PBMCs or MEFs (Extended Data Figs. 4g, h, 5j). Similarly, the concentration of the antioxidant glutathione (GSH) was much higher in patient fibroblasts than that of controls (Fig. 4h). By contrast, GSH levels were similar in PBMCs or MEFs expressing wild-type or mutant RIPK1 (Extended Data Figs. 4i, 5k). Consistent with increased levels of GSH, the amounts of reactive oxygen species (ROS) (as indicated by the cytosolic ROS sensor carboxy-H2DCFDA) were lower after erastin stimulation in patient fibroblasts (Fig. 4i), but not in Ripk1-knockout MEFs complemented with mutant RIPK1 (Extended Data Fig. 5l). These data suggest that restricted release of ROS by the patient fibroblasts may help to protect against ferroptosis, as ROS is known to be crucial for mediating ferroptosis16. Similarly, because ROS production can promote RIPK1 activation and necroptosis17,18, the high levels of antioxidant GSH in the patient fibroblasts may also contribute to the resistance to necroptosis.

Discussion

Our study identified a dominantly inherited autoinflammatory disease caused by impaired caspase-8 cleavage in RIPK1. This condition is distinct from the previously reported recessively inherited RIPK1-deficient condition that is characterized by immune deficiency19,20. By contrast, we show that patients with one copy of mutated RIPK1 in the caspase-8 cleavage site present with symptoms of immune dysfunction, including recurrent fevers and lymphadenopathy.

Our data highlight the role of RIPK1 kinase activity in promoting not only both apoptosis and necroptosis but also transcriptional production of pro-inflammatory cytokines, such as IL-6, which is a previously underappreciated aspect of RIPK1 biology. These results suggest that the periodic fevers of these patients may reflect the augmented production of cytokines such as IL-6 in response to what may be benign stimuli for normal individuals. Activated RIPK1 has been shown to mediate transcription of pro-inflammatory cytokines in myeloid lineages, independent of cell death, in neurodegenerative diseases21,22. In addition, cytokines such as TNF in turn can further promote cell death, thus establishing a vicious circle of inflammation that culminates in the development of an autoinflammatory disease.

We show that patient fibroblasts may have developed several compensatory mechanisms to protect against deleterious effects of activated RIPK1, including downregulating the expression of RIPK1 and TNFR1, as well as promoting anti-ROS mechanisms. These findings provide insights into the complex disease mechanisms behind non-cleavable RIPK1 variants in humans compared to that of the mouse models. Our study also linked an activating RIPK1 variant to ferroptosis, which sheds light on the diverse roles of RIPK1 in regulating several cell death pathways.

Methods

Patients

Patient P1 was evaluated under protocols approved by the Institutional Review Board by Children’s Hospital of Fudan University. Patients P2, P3, P4 and P5 and their unaffected family members were evaluated at McMaster Children’s Hospital, and the Hospital for Sick Children. Signed consent for their clinical information to be shared and for research samples to be sent to Boston Children’s Hospital was obtained. Ethics clearance was received from the Institutional Review Board at Boston Children’s Hospital and from Western Institutional Review Board. All relevant ethical regulations were followed. All patients and/or substitute decision markers provided written informed consent.

Unaffected controls

We used unaffected controls for functional assays. Paediatric unaffected controls are less than 10 years old and had no symptoms of inflammation when sampling.

WES

DNA from whole blood was extracted using a Maxwell RSC Whole Blood DNA Kit (Promega, AS1520). One microgram of DNA was used for whole-exome sequencing. For the first family, WES and data analysis were performed as previously described23,24,25. Variants were annotated by ANNOVAR (2018Apr16). Candidate variants were filtered to remove those presenting in the gnomAD, Kaviar, dbSNP and an in-house database. Variants were further filtered by de novo or dominant inheritance. For the second family, WES was performed and analysed concurrently for the proband, both parents and one affected son as previously described26. Other affected or unaffected family members were tested by Sanger sequencing for the presence or absence of the de novo variant identified in the proband.

Sanger sequencing

Sanger sequencing was used to confirm variants identified by exome sequencing as previously described23,24,25.

Cell preparation, culture and stimulation

The HEK293T cell line was from the American Type Culture Collection. Ripk1 gene knockout MEFs were established from Ripk1−/− mice. MEFs derived from D325A knock-in mice were provided by J. Zhang. PBMCs were separated by lymphocyte separation medium (LSM) and SepMate tubes (Stemcell) according to the manufacturer’s instructions. Fibroblasts were derived from skin biopsies of patient and control donors. HEK293T cells, MEFs and fibroblasts were grown in DMEM (Gibco) supplemented with 10% fetal bovine serum (FBS) (ExCell Bio) and penicillin/streptomycin (HyClone). PBMCs were grown in RPMI-1640 (Gibco) supplemented with 10% FBS and penicillin/streptomycin. All cell lines tested negative for mycoplasma contamination.

Recombinant human TNF (Peprotech, 300-01A) was used to stimulate PBMCs (50 ng ml−1, 100 ng ml−1), fibroblasts (20 ng ml−1, 50 ng ml−1) and MEFs (50 ng ml−1) for the indicated amount of time. LPS (Sigma, L6529) was used to stimulate PBMCs (1 μg ml−1), MEFs (1 μg ml−1) and fibroblasts (1 μg ml−1) for the indicated amount of time. TRAIL (R&D, 1121-TL) was used to stimulate MEFs (100 ng ml−1) for the indicated amount of time. Z-VAD-FMK (100 μM) and SM-164 (50 nM) (from Selleck) and Nec-1s (10 μM) (made by custom synthesis) were used to treat PBMCs, MEFs and fibroblasts. Erastin and RSL3 were used to induce cell ferroptosis in PBMCs (10 μM, 1 μM), MEFs (10 μM, 0.5 μM) and fibroblasts (10 μM, 0.5 μM).

RNA sequencing

One microgram of RNA was used for library preparation. Libraries were generated using NEBNext Ultra RNA Library Prep Kit for Illumina (NEB) following manufacturer’s recommendations and index codes were added to attribute sequences to each sample. Library quality was assessed on the Agilent Bioanalyzer 2100 system. The libraries were sequenced on Illumina Novaseq and 150-bp paired-end reads were generated. Sequenced reads were mapped against the human reference genome (GRCh38) or mouse reference genome (GRCm38) using HISAT2. featureCounts was used to count the reads numbers mapped to each gene. Differential expression analysis was performed using the DESeq2 R package.

Single-cell RNA sequencing

10X Genomics Chromium machine was used for 8,000–10,000 single-cell capture and cDNA preparation. The machine divided thousands of cells into nanolitre-scale Gel Bead-In-EMulsions for barcoding followed by clean up using the silane magnetic beads and Solid Phase Reversible Immobilization beads. Barcoded cDNA was then amplified by PCR. The library was constructed according to the manufacturer’s instruction. Sequencing was carried out on Illumina Novaseq. Sequence data were processed with Cell Ranger V3.0.1 (10X Genomics). The resulting count matrices followed the standard pipeline with default parameters. The UMAP plots were calculated based on the first 20 components of the CCA, and clusters were identified by Seurat R package (https://satijalab.org/seurat/).

NanoString assay

One-hundred nanograms of total RNA was used for NanoString assay and gene expression analysis was conducted using the nCounter Analysis System (NanoString Technologies) with a codeset designed to target 594 immunologically related genes. NanoString assay and data analysis were performed as previously described25.

Quantitative RT–PCR assay

Total RNA from fibroblasts, MEFs and PBMCs was extracted using the RNeasy Mini kit (Qiagen, 74104). cDNA was generated by the PrimeScript RT reagent kit with gDNA Eraser (Perfect Real Time) (Takara, RR047A), and qPCR was performed using TB Green Premix Ex Taq II (Tli RNaseH Plus) (Takara, RR820A). The reactions were run on Applied Biosystems 7500 Real-Time PCR System (Life Technologies) and ROCHE 480II. Relative mRNA expression was normalized to ACTB or GAPDH and analysed by the ΔΔCt method.

Antibodies and expression plasmids

The following antibodies were purchased from Cell Signaling Technology: β-actin (4970), β-tubulin (86298), GAPDH (5174), RIPK1 (3493), p-RIPK1 (Ser166) (65746), MLKL (14993), p-MLKL (Ser358) (91689), p-MLKL (Ser345) (37333), p65 (8242), p-p65 (Ser65) (3033), IKKα (11930), IKKβ (2370), p-IKKα/β (Ser176/180) (2697), IκBα (4814), p-IκBα (Ser32) (2859), p38 (8690), p-p38 (Thr180/Tyr182) (4511), TNFR1 (3736), caspase-8 (4790), cleaved-caspase-8 (8592), caspase-3 (9662), cleaved-caspase-3 (Asp175) (9661), HA-tag (3724), SLC7A11 (12691). Cyclophilin A (ab41684), GPX4 (ab125066), LAMP2A (ab125068), COX2 (ab15191) and p53 (ab32389) were purchased from Abcam. FADD (sc-6036) and ACSL4 (sc-365230) were purchased from Santa Cruz Biotechnology. p-RIPK1 (Ser166) (BX60008) was made by Biolyx. HSC70 (10654-1-AP) was purchased from Proteintech Group. HSP90 (BF9107) was purchased from Affinity. MLKL (reactivity for Mus musculus) was homemade27.

Human wild-type RIPK1 plasmid (RC216024) was from Origene, and the mutant RIPK1 plasmids (D324V, D324H and D324K) were constructed by site-directed mutagenesis. Mouse wild-type RIPK1 plasmid was generated by PCR amplification from the cDNAs of MEFs, and then cloned into the pMSCV vector made in-house, and the mutant mouse RIPK1 plasmids (D325V and D325H) were constructed by site-directed mutagenesis.

Immunoprecipitation and western blotting

Cells were lysed in cold cell lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% NP-40, protease and phosphatase inhibitor mixture (Thermo Fisher, 78442) and 10% glycerol) for 10 min and centrifuged at 20,000g for 10 min. Protein concentration was measured on the cleared lysates by BCA protein assay kit (Thermo Fisher, 23225). Immunoprecipitation and immunoblotting were conducted as described previously with specific antibodies23,25.

In vitro RIPK1 cleavage assay

Unlabelled in vitro transcription and translation (IVTT) of 1 μg wild-type and mutant RIPK1 constructs were performed in 50 μl reactions using the TNT T7 Quick Coupled Transcription/Translation System (Promega, L1170). The reaction was incubated with purified recombinant caspase-8 protein (R&D, 705-C8/CF) and then immunoblotted with RIPK1 antibody.

Cytokine detection in serum

The concentrations of cytokines in serum were measured by BD Cytometric Bead Array. Cytokine concentrations for IL-6, TNF and IL-10 in the serum were determined by BD Cytometric bead arrays (BD Bioscience). All data were analysed by FCAPArray V3 software (BD Biosciences).

Flow cytometry analysis of phosphorylation

For phos-flow staining, isolated PBMCs were treated with or without LPS (1 μg ml−1) for 6 h at 37 °C, with 5% CO2 and then permeabilized with Perm Buffer III according to the manufacturer’s instructions (BD Biosciences). Surface marker CD3, CD14 and CD19 (BD Biosciences) were used to gate total T cells, monocytes and total B cells. The expression of p-STAT3, p-p65 and p-p38 were analysed by flow cytometry. For phos-flow analysis, the following antibodies were used: Alexa Fluor 647-conjugated antibody against STAT3 phosphorylated at Y705 (BD Biosciences), Alexa Fluor 488-conjugated antibody against NF-κB p65 phosphorylated at S529 (BD Biosciences) and Alexa Fluor 488-conjugated antibody against p38 phosphorylated at T180/Y182 (BD Biosciences). Isotype control antibodies were used to normalize the background signals for intracellular staining. All events were acquired on a FACS Canto II cytometer (BD Biosciences) and analysed with FlowJo (Tree Star). Blue lines in the Extended Data Fig. 2c indicate basal levels, orange lines indicate LPS stimulation for 6 h and red lines indicate an isotype control. The numbers mark the percentage of cells displaying phosphorylation of STAT3 or p38 based on comparison with isotype control staining for each cell type.

Intracellular cytokine staining

Intracellular cytokine staining for IL-6, TNF and IL-8 were measured in PBMCs at baseline and following LPS stimulation. Cells were washed twice with PBS, then treated with LPS (1 μg ml−1 per 1 × 106 PBMCs) and Golgi plug (BD Biosciences) for 6 h at 37 °C, with 5% CO2 and then permeabilized with Perm/Fix for 30 min at 4 °C. Cells were stained by antibodies CD3-Percp-cy5.5 (BD Biosciences), CD14-PE-CY7 (BD Biosciences), CD4-FITC (BD Biosciences), CD19-APC (BD Biosciences), IL8-PE (Biolegend), IL10-BV421 (Biolegend), IL6-Percp-cy5.5 (Biolegend) and TNF-V450 (Biolegend). Isotype control antibodies were used to normalize the background signal for intracellular staining. All events were acquired on a FACS Canto II cytometer and analysed with FlowJo (Tree Star).

Cell viability assay

General cell survival was measured by the ATP luminescence assay CellTiter-Glo (Promega). The percentage of viability was normalized to readouts of untreated cells.

Cell death assay

Cell death was determined by ToxiLight Non-destructive Cytotoxicity BioAssay Kit (Lonza, LT07) or SYTOX Green Nucleic Acid Stain (Thermo Fisher, S7020). All experiments were conducted on 384-well plates with at least three biological replicates. Data were collected by the multimode plate reader (Bio Tek).

Intracellular ROS detection

Cells were seeded in 12-well plates and treated with the indicated stimuli for the indicated amount of time. After cell death induction, 5 μM carboxy-H2DCFDA was added to cells for 30 min at room temperature. Cells were then returned to warm growth medium and incubated for 15 min, followed by replacement of growth medium with PBS. Images were taken using a Leica fluorescence microscope.

Intracellular GSH detection

The GSH concentration in cells was assessed by GSH-Glo Glutathione Assay Kit (Promega, V6911) according to the manufacturer’s instructions.

Statistics

No statistical methods were used to predetermine sample size. For cell-based experiments, biological triplicates were performed in each single experiment in general, unless otherwise stated. All values were expressed as mean ± s.e.m. and calculated from the average of at least three independent biological replicates unless specifically stated. Statistical analysis was performed using GraphPad Prism 8 software (GraphPad Software). For comparisons between two groups, the Student’s t-test (unpaired and two-tailed) was applied. In all tests, a 95% confidence interval was used, for which P < 0.05 was considered a significant difference. Statistical analysis of single-cell RNA sequencing and RNA sequencing was performed using R Software (R v.3.5.2).

URLs

ANNOVAR, http://annovar.openbioinformatics.org/en/latest/user-guide/download/; CADD, https://cadd.gs.washington.edu/; gnomAD, https://gnomad.broadinstitute.org/; Kaviar genomic variant database (Kaviar), http://db.systemsbiology.net/kaviar/; Sorting Intolerant from Tolerant (SIFT), https://sift.bii.a-star.edu.sg/; PolyPhen-2, http://genetics.bwh.harvard.edu/pph2/; likelihood ratio test (LRT), http://www.genetics.wustl.edu/jflab/lrt_query.html; MutationTaster, http://www.mutationtaster.org/.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.