NINJ1 mediates plasma membrane rupture during lytic cell death

Plasma membrane rupture (PMR) is the final cataclysmic event in lytic cell death. PMR releases intracellular molecules known as damage-associated molecular patterns (DAMPs) that propagate the inflammatory response1–3. The underlying mechanism of PMR, however, is unknown. Here we show that the cell-surface NINJ1 protein4–8, which contains two transmembrane regions, has an essential role in the induction of PMR. A forward-genetic screen of randomly mutagenized mice linked NINJ1 to PMR. Ninj1−/− macrophages exhibited impaired PMR in response to diverse inducers of pyroptotic, necrotic and apoptotic cell death, and were unable to release numerous intracellular proteins including HMGB1 (a known DAMP) and LDH (a standard measure of PMR). Ninj1–/– macrophages died, but with a distinctive and persistent ballooned morphology, attributable to defective disintegration of bubble-like herniations. Ninj1–/– mice were more susceptible than wild-type mice to infection with Citrobacter rodentium, which suggests a role for PMR in anti-bacterial host defence. Mechanistically, NINJ1 used an evolutionarily conserved extracellular domain for oligomerization and subsequent PMR. The discovery of NINJ1 as a mediator of PMR overturns the long-held idea that cell death-related PMR is a passive event. The small transmembrane protein NINJ1 promotes plasma membrane rupture in lytic cell death associated with pyroptosis, necrosis and apoptosis.

Pyroptosis is a potent inflammatory mode of lytic cell death triggered by diverse infectious and sterile insults 2,3,9 . It is driven by the pore-forming fragment of gasdermin D (GSDMD) [10][11][12][13] and releases two exemplar proteins: the pro-inflammatory cytokine IL-1β, and LDH, a standard marker of PMR and lytic cell death. An early landmark study 14 predicted two sequential steps for pyroptosis: (1) initial formation of a small plasma membrane pore that causes the release of IL-1β and non-selective ionic fluxes, and (2) subsequent PMR attributable to oncotic cell swelling. PMR releases LDH (140 kDa) and large DAMPs. Although the predicted size of gasdermin pores (approximately 18 nm inner diameter 15 ) is large enough to release IL-1β (17 kDa, around 4.5 nm diameter), the underlying mechanism for subsequent PMR has been considered a passive osmotic lysis event.

A forward-genetic screen identifies NINJ1
To identify essential mediators of PMR, we performed a forward-genetic screen using bone marrow-derived macrophages (BMDMs) from mice mutagenized with N-ethyl-N-nitrosourea (ENU). Cytoplasmic lipopolysaccharide (LPS), a potent stimulator of the non-canonical inflammasome, was used to initiate caspase-11-and GSDMD-dependent release of LDH 10,11 from BMDMs that were primed with the Toll-like receptor 2 (TLR2) agonist Pam3CSK4. Mice derived from pedigree IGL03767 exhibited a Mendelian-recessive trait that compromised LDH release (Fig. 1a). Exome sequencing of the founder G1 male identified 19 single nucleotide variants (SNVs). Subsequent phenotyping and SNV genotyping showed that the trait correlated with inheritance of a point mutation in the gene encoding NINJ1 (Extended Data Fig. 1a, b, Extended Data Table 1).
NINJ1 is a 16-kDa cell-surface protein that has two transmembrane regions with N and C termini outside the cytoplasm (N out /C out topology) (Fig. 1b). NINJ1 is widely expressed, including in myeloid cells and the central nervous system, and reportedly functions as an adhesion molecule that is associated with inflammation and tumour suppression [4][5][6][7][8] . The SNV (A→T) occurred at the non-coding 3′ splice acceptor site of exon 2 (Extended Data Fig. 1c). Accordingly, the 16-kDa NINJ1 protein detected in wild-type BMDMs was absent in mutant Ninj1 mt/mt BMDMs (Fig. 1c). NINJ1 deficiency attenuated GSDMD-dependent LDH Article release in response to either LPS or nigericin (an ionophore activator of the NLRP3 canonical inflammasome) 10,11,16 (Fig. 1d).
IL-1β is a hallmark cytokine for pyroptosis that is thought to exit dying BMDMs through the GSDMD pore 2,3,9,14,15 . Accordingly, wild-type and Ninj1 −/− BMDMs released comparable levels of IL-1β in response to either nigericin or cytoplasmic LPS (Fig. 2d). NINJ1 was also dispensable for the release of IL-18, another pyroptosis-associated IL-1 family member (Extended Data Fig. 2d). Wild-type and Ninj1 -/-BMDMs were mostly indistinguishable in transcriptomic analyses, lipid composition, and TLR-induced production of IL-6 and TNF (Extended Data Fig. 2e-g). In summary, our data provide compelling genetic evidence that the release of IL-1β and IL-18 from macrophages is independent of PMR and probably occurs via the approximately 18-nm GSDMD pore.
Wild-type BMDMs stimulated with either nigericin or cytoplasmic LPS undergo characteristic morphological changes during pyroptosis; the cells cease moving, swell and develop bubble-like herniations that disintegrate abruptly to yield a shrunken corpse 2,14,17 . NINJ1 deficiency inhibited bubble disintegration while upstream events were unaltered (Fig. 2e, f, Supplementary Videos 1-3). Remarkably, even at 16 h after exposure to LPS or nigericin, Ninj1 −/− BMDMs retained prominent 'bubble' morphology ( Fig. 2e). The cells were dead based on their loss of ATP, mitochondrial membrane potential, and motility ( Fig. 2g, Extended Data Fig. 2h, Supplementary Videos 1, 2). Thus, PMR and related events, including LDH release and bubble disintegration, are genetically separable from GSDMD-driven cell death and IL-1β release. PMR is probably an event that occurs after cell death 17,18 . Of note, BMDMs ceased moving before bubble formation (Supplementary Videos 1, 2). NINJ1-independent loss of mitochondrial membrane potential also preceded PMR (as assessed by release of DD-150) (Extended Data Fig. 2h, i).
We confirmed that NINJ1-mediated PMR released more proteins than just LDH. Supernatants from wild-type BMDMs stimulated with nigericin or cytoplasmic LPS contained many proteins that were diminished in their Ninj1 −/− counterparts (Fig. 2h, Extended Data Fig. 2j). Subsequent secretome analysis detected approximately 780 molecules (including plectin) that were released in a NINJ1-dependent manner in response to cytoplasmic LPS (Fig. 2i, Extended Data Table 2). Of note, Ninj1 -/-BMDMs were unable to release HMGB1, a proinflammatory DAMP 19 , despite exhibiting normal GSDMD-dependent release of IL-1α (Fig. 2j, k). HMGB1 is a relatively small nuclear protein of approximately 28-kDa, but forms large complexes with nucleosomes and transcription factors 20 , which probably hinders its release through the approximately 18-nm GSDMD pore. Regardless, the NINJ1-dependent release of diverse intracellular proteins from pyroptotic cells, including HMGB1, suggests a pro-inflammatory role of NINJ1. Indeed, Ninj1 −/− mice were more susceptible than wild-type mice to infection with C. rodentium (Fig. 2l). Thus, NINJ1-dependent PMR may release DAMPs that are important for host defence against bacteria. Ninj1 −/− mice exhibited normal susceptibility to caspase-11-dependent 16 and GSDMD-dependent 11 acute septic shock induced by LPS (Extended Data Fig. 2k), which suggests that mechanisms that are driven by GSDMD but independent of PMR promote LPS-induced mortality.

NINJ1 has a global role in PMR
PMR is not exclusive to pyroptosis, and also occurs during necrosis or post-apoptosis (sometimes referred to as secondary necrosis) 21 . We examined the role of NINJ1 in non-pyroptotic PMR. In control experiments, pyroptotic stimuli (such as cytoplasmic LPS, nigericin and flagellin 10,11 ) caused NINJ1-and GSDMD-dependent PMR in BMDMs on the basis of LDH release, whereas PMR after freezing and thawing of cells did not require NINJ1 (Fig. 3a). Ninj1 −/− BMDMs released less  LDH than wild-type BMDMs in response to necrotic stimuli (bacterial pore-forming toxins) or apoptotic stimuli (including chemotherapeutic agents such as the DNA crosslinker cisplatin and BCL-2 antagonist venetoclax) 22 . The release of HMGB1 and other proteins from BMDMs exposed to cisplatin or venetoclax was also attenuated by NINJ1 deficiency (Fig. 3b, Extended Data Fig. 3a). Wild-type BMDMs treated with venetoclax exhibited typical apoptotic morphology 1 including cell shrinkage and bleb formation. Blebbing was not prevented by Ninj1 deficiency, but the dying cells developed a persistent ballooned morphology ( Fig. 3c, d, Extended Data Fig. 3b, Supplementary Video 4). These data underscore a global role for NINJ1 in inducing PMR. We confirmed that venetoclax-induced PMR in BMDMs was a gasdermin-independent, non-pyroptotic event. BMDMs express only Gsdmd and Gsdme (Extended Data Fig. 3c), but Gsdmd −/− Gsdme −/− BMDMs exhibited normal PMR in response to venetoclax, releasing both DD-150 and LDH (Fig. 3e, f). Mitochondrial dysfunction that leads to depletion of ATP and consequent cell swelling 22,23 may be responsible for apoptosis-related PMR. Oligomycin (an inhibitor of ATP synthase) also induced gasdermin-independent but NINJ1-dependent PMR ( Fig. 3f-h, Extended Data Fig. 3d, e).
NINJ1 is unlikely to be the only mediator of PMR because NINJ1 deficiency only partially attenuated the release of LDH and DD-150 from BMDMs undergoing MLKL-dependent necroptosis 24,25 after treatment with TNF plus the pan-caspase inhibitor zVAD (Fig. 3i, j). Overall protein release, including the release of HMGB1, was largely unaltered (Extended Data Fig. 3f, g). These data support the existence of a NINJ1-independent mechanism for PMR during necroptosis. It is possible that oligomerized MLKL 25 disrupts the plasma membrane to induce PMR directly, thereby bypassing the need for NINJ1.

NINJ1 oligomerizes to induce PMR
Ectopic expression of human or mouse NINJ1 in HEK293T cells caused marked cytotoxicity with concomitant release of LDH (Fig. 4a, Extended Data Fig. 4a-c). The Drosophila orthologues dNINJ-A and dNINJ-B were also cytotoxic, whereas dNINJ-C and NINJ2 were not (Fig. 4a, Extended Data Fig. 4a, c-e). Scanning mutagenesis identified a highly conserved, extracellular domain, which is predicted to be α-helical, as crucial for cell killing (Fig. 4b, Extended Data Fig. 4f-h). All NINJ1 mutants that contained five consecutive alanine substitutions within Article the putative α-helical domain (including A42NKKS→A42AAAA) exhibited impaired cytotoxic activity. Mutating a residue within this region to a helix-breaking proline (including A59P), or amino acid replacements that mimic the NINJ2 sequence (S62M/Q63R), also reduced NINJ1 killing activity (Extended Data Figs. 4e, 5a, b). These results confirm the importance of this conserved, putative α-helix domain. NINJ1 reportedly functions as a cell-cell adhesion molecule via homotypic binding of an adhesive segment (26-37 amino acids) 4,5 before the putative α-helix domain. However, mutagenesis of this evolutionarily variable segment did not reduce NINJ1 killing activity (Fig. 4b, Extended Data Fig. 4f). The previously reported adhesion-dead NINJ1(W29A) mutant 5 also restored LPS-induced PMR in Ninj1 −/− iMACs (Extended Data Fig. 5c). Thus, the adhesive segment of NINJ1 appears dispensable for PMR.
We investigated NINJ1 activation in BMDMs using standard biochemical approaches. By SDS-PAGE, with or without reductant (dithiothreitol), NINJ1 migrated as an approximately 16-kDa monomer regardless of stimulus (LPS, nigericin, TLRs or interferons) or treatment with the pan-kinase inhibitor staurosporine (Extended Data Fig. 5d-f). By blue-native-PAGE (BN-PAGE), which maintains native protein structures, endogenous NINJ1 was shifted from approximately 40 to 900 kDa in response to nigericin or cytoplasmic LPS (Fig. 4c). These data suggest that NINJ1 exists as a dimer or trimer in unstimulated BMDMs, and then further oligomerizes in response to death stimuli.
By immunofluorescence microscopy of unstimulated BMDMs, NINJ1 primarily localized at the plasma membrane alongside the surface marker CD44, although some NINJ1, probably in transit, co-stained with the Golgi marker GM130 (Extended Data Fig. 6). After stimulation with nigericin, NINJ1 formed several speck-like assemblies (Fig. 4d), consistent with NINJ1 oligomerization.
The putative extracellular α-helix domain of NINJ1 possesses characteristic hydrophilic and hydrophobic clusters that are reminiscent of amphipathic α-helices 26,27 (Extended Data Fig. 7a). Of note, proteins that contain amphipathic helices such as α-synuclein and anti-microbial peptides (including Melittin bee venom) disrupt phospholipid bilayer membranes by an unknown mechanism in which positively charged residues have crucial roles 28,29 . Mutagenesis of positively charged NINJ1 residues (H40, K44, K45 and K65) to the non-charged amino acid glutamine uncovered a K45Q mutation that impaired NINJ1 cytotoxicity in HEK293T cells (Fig. 4e, Extended Data Fig. 7b). The K45 residue is evolutionarily conserved and predicted to be near the N-terminal side of the α-helix (Extended Data Figs. 4f, 7a). Both NINJ1(K45Q) and NINJ1(A59P) were unable to restore LPS-induced PMR in Ninj1 −/− iMACs, and attenuated oligomerization of NINJ1 (Fig. 4f, g). These data support a model in which NINJ1 oligomerizes to induce PMR. A peptide corresponding to the putative α-helix domain of NINJ1 directly damaged synthetic liposome membranes to release encapsulated cargo (Extended Data Fig. 7c). Precisely how NINJ1 induces  PMR remains unclear and will probably require further structural insights. Given that NINJ1 mediates PMR during pyroptosis, necrosis and apoptosis (Fig. 3a), a common event such as an increase in cell volume may trigger activation of NINJ1. Some amphipathic helices sense lipid-packing defects and altered membrane curvature 26,30 , both of which may be caused by cell swelling. Although NINJ1 appears to be required for end-stage lysis, its mechanism of action remains undefined. The link between PMR and NINJ1 may well be indirect and not specific to programmed PMR. In summary, NINJ1 is an evolutionarily conserved cell-surface protein that mediates PMR and the release of DAMPs-key events in the propagation of inflammation. Given the potent and global role for NINJ1 in PMR induction and DAMP-release related to pyroptosis, apoptosis and necrosis, targeting NINJ1 may be of therapeutic benefit. Indeed, the addition of a monoclonal NINJ1 antibody to cell cultures expressing endogenous NINJ1 inhibited PMR (Extended Data Fig. 7d).

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-021-03218-7.

ENU-mutagenized mouse strains
C57BL/6NCrl G0 mice were treated with ENU and the resulting mutations were bred to homozygosity in G3 mice as previously described 31 .
All mice used in this study were cared for and used in experiments approved by the Australian National University Animal Experimentation Ethics Committee under protocol A2018/07. 129X1/SvJ strain was used as a Casp11 mutant 16 control.

Exome sequencing
Exon capture, sequencing and analysis were performed as previously described 32,33 . Ninj1 mutant genotyping primer sequences: F1: GAAGGTGACCAAGTTCATGCTCTGACCGCCTTGCTCCCACA; F2: GAAGGTCGGAGTCAACGGATTCTGACCGCCTTGCTCCCACT; R1: CGCTCTTCTTGTTGGCATAA. Groups were determined by genotype rather than treatment, therefore, randomization was not applicable. Mice were picked and treated by the same individual, therefore, blinding to genotype and treatment as well as during data collection and analysis was not possible. No statistical methods were used to predetermine sample size.

RNA-sequencing
Total RNA was extracted from primed or non-primed wild-type BMDMs and Ninj1 -/-BMDMs (n = 3 per genotype) using an RNeasy kit (Qiagen) with on-column DNase digestion. Quality control of total RNA was performed to determine sample quantity and quality. The concentration of RNA was determined using a NanoDrop 8000 (Thermo Fisher Scientific), and the integrity of the RNA was determined by Fragment Analyzer (Advanced Analytical Technologies). Total RNA (100 ng) was used as an input material for library preparation using the TruSeq Stranded Total RNA Library Prep Kit (Illumina). The sizes of the libraries were confirmed using High Sensitivity D1K screen tape (Agilent Technologies), and their concentrations were determined with a quantitative PCR-based method using a Library Quantification kit (KAPA). The libraries were multiplexed and sequenced on an Illumina HiSeq4000 (Illumina) to generate 30 million single-end, 50 bp reads. For RNA-seq analysis, the raw FASTQ reads were aligned to the mouse reference genome (GRCm38-mm10) using GSNAP (with parameters -M 2 -n 10 -B 2 -i 1 -N 1 -w 200000 -E 1 --pairmax-rna = 200000 --clip-overlap). Reads were filtered to include only the uniquely mapped reads. Differential expression analysis was performed using the voom/limma R package. Genes were considered to be differentially expressed if the log 2 -transformed fold change was > 1 or < −1 and the adjusted P value was < 0.05.
The mixture was vortexed and centrifuged at 1,000g for 10 min to achieve phase separation. The bottom layer was collected into a clean glass tube, and the upper layer was extracted once more by adding 1.8 ml of DCM. The bottom layer of the second extraction was combined with the first and dried under a gentle stream of nitrogen for subsequent LC-MS/MS analysis. Dried residue was reconstituted in 300 μl DCM:methanol (1:1), 10 mM ammonium acetate for direct infusion and analysis on a SelexION enabled 6500+ QTRAP mass spectrometry (SCIEX) by the method previously described 38 . For cholesterol analysis, dried residue was reconstituted in 200 μl DCM: methanol (1:1). HPLC separation of cholesterol from its metabolites was performed on a reverse-phase column (Luna Omega 1.6 μm C18 100A, LC column 100 × 2.1 mm). The temperatures of the column oven and auto sampler were set at 40 °C and 15 °C, respectively. The LC flow rate was set at 0.2 ml min −1 . Initial gradient conditions were 95% mobile phase A (3:1 water: methanol) and 5% mobile phase B (1:1 methanol: isopropanol). Mobile phase B was increased to 55% within 2 min, further increased to 65% over 12 min and then to 74% in 7 min. Mobile B was held at 74% for 11 min and increased to 100% in 4 min. Mobile phase B was returned to the initial conditions within 1 min and re-equilibrated for 5 min before the next injection. The liquid chromatograph was coupled to a 6500+ QTRAP mass spectrometer operated under positive ionization mode with the following source settings: turbo-ion-spray source at 500 °C, N 2 nebulization at 20 psi, N 2 heater gas at 20 psi, curtain gas at 30 psi, collision-activated dissociation gas pressure was held at medium, turbo ion-spray voltage at 5500 V, declustering potential at 60 V, entrance potential at 10 V and collision energy of 30 V. Sample analysis was performed in multiple reactions monitoring mode with a dwell time of 0.10 s. The transitions monitored for cholesterol and D7-cholesterol were 369.4/161 and 376.3/161, respectively. Cholesterol quantification was achieved by creating a standard curve using six concentration levels of cholesterol versus its normalized response to the internal standard (D7-cholesterol). Plots were generated using Prism.

Secretome analysis of BMDM supernatants
Approximately 5.0 × 10 6 primed BMDMs were neon-electroporated with 5 μg of LPS with 1,720 voltage, 10 width, 2 pulse settings. BMDMs were incubated in 250 μl of no-FBS high-glucose DMEM for 2 h, then pelleted by spinning at 300g for 10 min. Then, 20 μl of supernatants from two replicates of wild-type or Ninj1 -/-BMDMs were reduced with 10 mM dithiothreitol at 60 °C followed by alkylation with 20 mM iodoacetamide at room temperature. Proteins were digested with 0.2 μg trypsin (Promega) in ammonium bicarbonate pH 8.0 at 37 °C overnight. Digestion was quenched with formic acid and the supernatants were subjected to desalting on C18 PhyTips (PhyNexus), lyophilized, reconstituted to 25 μl in 0.1% formic acid containing 2% acetonitrile and analysed without further processing by reversed phase nano-LC-MS/MS on a Waters NanoAcquity HPLC system (Waters) interfaced to a Thermo Fisher Fusion Lumos (Thermo Fisher Scientific). Peptides were loaded onto a Symmetry C18 column (1.7 mm BEH-130, 0.1 × 100 mm, Waters) and separated with a 60 min gradient from 2% to 25% solvent B (0.1% formic acid, 98% acetonitrile) at 1 μl min −1 flow rate. Peptides were eluted directly into the mass spectrometer with a spray voltage of 1.2 kV. Full MS data were acquired in FT for 350-1,250 m/z with a 60,000 resolution. The most abundant ions found from full MS were selected for MS/MS through a 2-Da isolation window. Acquired tandem MS spectra were searched using the Mascot (Matrix Sciences) with trypsin enzyme specificity. Search criteria included a full MS tolerance of 50 ppm, MS/MS tolerance of 0.5 Da with oxidation (+15.9949 Da) of as variable modification and carbamidomethylation (+57.0215 Da) of cysteine as static modification. Data were searched against the mouse and contaminant subset of the Uniprot database that consists of the reverse protein sequences. Peptide assignments were first filtered to a 2% false discovery rate (FDR) at the peptide level and subsequently to a 2% FDR at the protein level. Label free quantification was performed using the Vista Algorithm 39 and peptide spectral matches (PSMs) per protein were visualized using Spotfire (TIBCO). Top hits were identified by multiplying the log 2 -transformed fold change (fold change comparing Ninj1 -/to wild-type BMDMs) by the log 10 P value.

Silver staining of total proteins in culture supernatant
For visualization of secreted proteins, primed BMDMs were washed with PBS three times and cultured in no-FBS high-glucose DMEM medium for 4 h after LPS electroporation or nigericin stimulation. Culture supernatant was collected after spinning for 10 min at 300g. Then, 10 μl of the supernatant was run on SDS-PAGE and proteins were silver stained by using SilverQuest (Thermo Fisher Scientific). For venetoclax and TNF plus zVAD stimulation, non-primed BMDMs were cultured with venetoclax or TNF plus zVAD for 16 h.

Live imaging of BMDMs
BMDMs were plated on glass-bottom SensoPlates. For LPS transfection on the glass-bottom plate, primed BMDMs were cultivated with 5 μg ml −1 LPS and 20 μg ml −1 CTB to deliver LPS inside place 37 . Where described, tetramethylrhodamine methyl ester perchlorate (TMRM, 200 μM, Thermo Fisher Scientific), HOECHST, and DD-150 were used. Plates were imaged either using a 60× Plan Fluor or 20× Super Plan Fluor ELWD objective on an ImageXpress Micro Confocal system equipped with an environmental controller and gas mixer to maintain cells at 37 °C and 5% CO 2 . Images of the bright-field and transmitted light and fluorescence channels were imaged every 5 min overnight. Images were subsequently processed and videos were generated using the scikit-image python package.

Lattice light-sheet microscopy
BMDMs were plated on 5 mm round coverslips in 6 well plates. Primed BMDMs were incubated with CellMask Deep Red Plasma membrane Stain (5 μg ml −1 Thermo Fisher Scientific) for 20 min. 4D datasets were generated using a lattice light-sheet microscope (Intelligent-Imaging Innovations). Coverslips were mounted to sample holders and placed in a 2.5 ml bath containing phenol red free DMEM (Thermo Fisher Scientific) with 160 μg ml −1 nigericin to stimulate bubble formation. A lattice light-sheet used for illumination was generated using a 640 nm laser and a 0.450 NAO/0.375 NAI annular mask. Entire cell volume was imaged using Z galvo sweep of the light sheet through the sample and a Flash 4 sCMOS camera (Hamamatsu). A total of 201 Z planes at 0.2-μm steps (0.1 post deskew), 10 ms exposure time and 5% laser power. Images were acquired and deskewed with Slidebook 6 (Intelligent-Imaging Innovations). Time-lapse images were imported into Imaris 9.5 (Oxford Instruments) to generate time-lapse videos for presentation.

Secondary structure prediction and helix modelling
Secondary structure prediction for mouse NINJ1 was performed using the JPred 4 server. Following prediction, the external alpha helical domain was modelled with SWISS-MODEL 42 and subsequently visualized with PyMOL (v.2.3.5). The structure was exported using ray-traced frames.

Conservation scores, sequence alignments, and phylogenetic tree generation
The conservation score for each amino acid residue of NINJ1 was calculated using the ConSurf 43,44 server and plotted using the Matplotlib python package. Individual orthologue NINJ1 and NINJ2 sequences obtained from the UniProt database 45 were aligned using the Clustal Omega multiple sequence alignment 46 . Phylogenetic trees were retrieved from the resulting Clustal Omega output and the alignment was visualized using ESPript 3.0 47 .

C. rodentium infection
Female 12-14-week-old Ninj1 -/and littermate wild-type control mice were infected perorally with 2 × 10 9 CFU of log-phase cultured C. rodentium (ATCC 51459) after overnight fasting. Infected mice were monitored for survival for 17 days after infection. Plot was generated using Prism.

LPS septic shock
Mouse model of LPS-induced acute septic shock was performed as previously described 11,16,37 . In brief, male mice aged 8 to 10 weeks were injected intraperitoneally with 54 mg kg −1 LPS (E. coli O111: B4, Sigma) and monitored eight times daily for a total of 6 days. Plots were generated using Prism.

Statistics and reproducibility
Unless otherwise specified, results are representative of two independent experiments and means are of at least three individual replicates.

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

Data availability
RNA-sequencing data are available through the Gene Expression Omnibus (GEO) database at accession number GSE156395. Datasets from UniProt database (https://www.uniprot.org/) including the mouse and contaminant subsets as well as the following accession numbers: Q70131, Q92982, P70617, F1PMB0, Q2TA30, R4GJU8, H9G4V3, Q66JI7, A0A0R4IDX9, Q9NZG7 and Q9JL89 were used. Other datasets generated during and/or analysed during the current study are available from the corresponding authors on reasonable request. Source data are provided with this paper.

Corresponding author(s): Nobuhiko Kayagaki and Vishva M. Dixit
Last updated by author(s): Dec 20, 2020 Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted

Software and code
Policy information about availability of computer code Data collection Imaging (live and fixed) data captured using MetaXpress 6.5.4.532, Slidebook 6, and Incucyte S3 2019A.

Data analysis
Mass spectrometry data analysed using Mascot 2.4.1, the Vista Algorithm 4.0.0, and TIBCO Spotfire 7.8.0 HF-003. Plots were generated either with Prism 7.0 or Matplotlib 3.1.3. Imaging data analysed and prepared using scikit-image 0.16.2 and Imaris 9.5. Secondary structure prediction was performed using the JPred 4 server. Structure modeling was performed with SWISS-MODEL 2018 and visualized with PyMOL 2.3.5. Conservation score was calculated using the ConSurf 2016 server. Sequence alignment and phylogenetic tree creation were performed using Clustal Omega 1.2.4. Sequence alignment was visualized with ESPript 3.0. RNA sequencing data analysed using GSNAP 2013-11-01 and voom/limma R package 3.44.3 (as described in the methods).
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability RNA-sequencing data is available through the GEO database (GSE156395). Source Data for Figs. 1-4 and Extended Data Figs. 2-8 are provided with the paper. Datasets from UniProt database (https://www.uniprot.org/) including the mouse and contaminant subsets as well as the following accession numbers: Q70131,