Receptor interacting protein kinase 2-mediated mitophagy regulates inflammasome activation during virus infection

RIPK2 protects against severe IAV infection

To examine the potential role of NOD2 and RIPK2 in the immune response to IAV infection, wildtype (WT), Nod2^−/− and Ripk2^−/− mice were challenged intranasally (i.n.) with the virulent influenza A/PR/8/34 (PR8) H1N1 virus and survival was monitored over time. Relative to wildtype (WT) controls, Nod2^−/− and Ripk2^−/− mice were significantly more susceptible (Fig. 1a). While signs of pathological damage to the lungs of WT mice were minimal by day (d) 2 after PR8 inoculation, lung sections of both knockout mouse lines showed marked pathological changes characterized by greater pulmonary airway obstruction and increased cell death on d2 and d7 after infection (Fig. 1b–d and data not shown). Moreover, airway obstruction of IAV-infected Ripk2^−/− mice consisted mainly of neutrophils (Fig. 1e). While neutrophil counts were elevated in the Ripk2^−/− whole lung on d2 and d7, no increases were observed in the numbers of other myeloid cells (Fig. 1f–g). Despite the increased susceptibility of Ripk2^−/− mice to IAV infection, lung virus titers were comparable to those in the WT controls (Fig. 1h). Thus, NOD2 and RIPK2 protects against severe IAV infection, although this is not correlated with the extent of virus replication.

RIPK2 regulates IFN-γ and IL-18

Given that the inflammatory response to IAV is thought to contribute to mortality^{3, 4}, we investigated the role of RIPK2 in cytokine production. To this end, whole lung homogenates were taken on days 2 and 7 after i.n. challenge of WT and Ripk2^−/− mice with the PR8 virus, and the extent of inflammatory mediator production was determined by ELISA. Although RIPK2 regulates IL-6 and TNF production during bacterial infections²⁷, no significant differences in the levels of these cytokines were observed in lungs of virus-infected WT and Ripk2^−/− mice (Fig. 2a). In contrast, IFN-γ, a known mediator of inflammation in viral immunity²⁸, and IL-18 were significantly upregulated in Ripk2^−/− mice (Fig. 2b). IL-18 was originally identified as an IFN-γ-inducing cytokine²⁹, suggesting that IFN-γ production might be consequential to increased IL-18 release in IAV-infected Ripk2^−/− mice. Previous studies have shown the chemokines CCL2 (MCP-1) and CXCL10 (IP-10) were induced by IL-18 and IFN-γ³⁰ and accordingly, we observed elevated production of both CCL2 and CXCL10 in IAV-infected Ripk2^−/− mice (Fig. 2c). These results indicate that increased susceptibility to IAV infection in Ripk2^−/− mice is associated with exacerbated production of IL-18 and IFN-γ.

IL-18 antagonism in Ripk2^−/− mice improves survival

We hypothesized that increased IL-18 production was responsible for the elevated IAV susceptibility of Ripk2^−/− mice. To test this, Ripk2^−/− mice were injected intraperitoneally (i.p.) with anti-IL-18 neutralizing antibody 3h prior to PR8 infection. This treatment significantly reduced IL-18 and IFN-γ levels in the lungs of IAV-infected Ripk2^−/− mice (Fig. 3a). This was followed by significantly reduced CCL2 levels, while the concentration of the chemokines CXCL10, CXCL1 (KC) and CCL3 (MIP-1α) protein- albeit not statistically significant - showed a trend towards diminished expression (Supplementary Fig. 1a). Furthermore, IL-18 neutralization dampened virus-induced lung pathology and airway neutrophilia (Fig. 3b–c and Supplementary Fig. 1b), and significantly improved survival (Fig. 3d). These results suggest that increased IL-18 production drives elevated morbidity and mortality in IAV-infected Ripk2^−/− mice. To further confirm the critical role of IL-18 in this process, we generated mice lacking both RIPK2 and IL-18 (Ripk2^−/−Il18^−/− mice) and challenged these animals with a lethal dose of IAV. As with Ripk2^−/− mice receiving IL-18-neutralizing antibodies, genetic ablation of IL-18 rescued the hypersensitive response of Ripk2^−/− mice; tissue neutrophil infiltration and mortality rates returned to levels comparable to those of infected WT mice (Fig. 3e–f). Overall, these results demonstrate a critical role for elevated IL-18 production increasing morbidity and mortality in IAV-infected Ripk2^−/− mice.

Immune and lung cells contribute to elevated IL-18

We next set out to identify the cellular compartments that contribute to RIPK2-mediated modulation of IL-18. Bone marrow (BM) chimeras were generated that selectively expressed RIPK2 in the radiation-sensitive (donor BM) or radiation-resistant (recipient lung) compartments. These chimeric mice were subsequently infected i.n. with PR8, and IL-18 production was assessed by immunohistochemistry (IHC) and ELISA. Both Ripk2^−/− hematopoietic and non-immune (likely lung epithelial) cells contributed to increased production of IL-18 in the pulmonary tract of IAV-infected Ripk2^−/− mice (Fig. 4a–b, and Supplementary Fig. 2a).

Innate and adaptive lymphocytes are hyperactivated

We next examined the cellular populations that were responsible for the increased IFN-γ observed in Ripk2^−/− mice. Both NK cells and CD8⁺ T cells isolated from PR8-infected Ripk2^−/− mice produced larger quantities of IFN-γ on a per-cell basis and were more highly activated, although the absolute numbers of these cells remained similar to those in WT mice (Fig. 4c–f, and Supplementary Fig. 2b–d). These results indicate that increased IL-18 in Ripk2^−/− mice subsequently leads to increased IFN-γ production from innate and adaptive cell populations.

NOD2-RIPK2 axis regulates caspase-1 activation and IL-18

NOD2 was reported to regulate type I IFN during IAV infection^{25, 26}. We therefore examined type I IFN levels in IAV-infected Ripk2^−/− mice. Although we confirmed the reported defect in type I IFN production in Nod2^−/− alveolar macrophages, Ripk2^−/− macrophages produced type I IFN (IFN-β) protein in amounts similar to those of WT cells (data not shown), suggesting that differences in type I IFN signaling are unlikely to account for the observed hypersensitive phenotype of Ripk2^−/− mice. Similarly, we found no significant differences in activation of NF-κB (pIκBα) and MAPK (pERK) pathways in IAV-infected Ripk2^−/− bone marrow derived dendritic cells (BMDCs) (data not shown). We therefore analyzed processes directly regulating IL-18 maturation rather than IL-18 transcript abundance. Unlike most cytokines, IL-1β and IL-18 are produced as propeptides that require proteolytic maturation by the cysteine protease caspase-1 in order to be released in their biologically active forms³¹. Notably, both caspase-1 activation and release of mature IL-18 and IL-1β were markedly increased in PR8 infected Ripk2^−/− and Nod2^−/− BMDCs (Fig. 5a–b). Although we initially detected increased IL-18 levels in vivo, low amounts of IL-1β likely confounded detection of differences in this cytokine in vivo (Supplementary Fig. 3a). Next, we examined caspase-1 activation to other IAV infections. We observed that the ×31 low pathogenicity virus also elicited higher caspase-1 activation in Ripk2^−/− BMDCs (Supplementary Fig. 3b). We also examined inflammasome activation during infection with bacteria known to activate the RIPK2 signaling pathway. Infection with Listeria failed to augment caspase-1 processing and IL-18 production in Ripk2^−/− and Nod2^−/− BMDCs (Figs. 5c–d). This enhancement of inflammasome activation by IAV in Ripk2^−/− BMDCs was dependent on NLRP3, as the NLRP3 specific inhibitor glyburide³² significantly inhibited caspase-1 activation and IL-18 production (Fig. 5e–f). In the case of Listeria infection, the AIM2 inflammasome plays a predominant role in inflammasome activation³³. In agreement, Aim2^−/− BMDCs showed nearly complete impairment in caspase-1 activation following Listeria infection (Fig. 5g). These data indicate that enhanced NLRP3 inflammasome activation occurs in Ripk2^−/− BMDCs during IAV infection, but not Listeria infection, indicating that this pathway is not a global disregulation of inflammasome activation, but specific to the NLRP3 inflammasome.

RIPK2 regulates the NLRP3 inflammasome through autophagy

To determine the mechanism by which RIPK2 dampens inflammasome activation, we first assessed the effect of RIPK2 overexpression on caspase-1 maturation and IL-18 release in transfected 293T cells. In agreement with published reports³⁴, overexpressed RIPK2 interacted with caspase-1 but this interaction enhanced, rather than inhibited, caspase-1 processing and IL-18 release (data not shown) indicating that RIPK2 overexpression is not capable of directly inhibiting inflammasome activation. Because Ripk2^−/− BMDCs do not display a global defect in inflammasome activation (see Fig. 5), this suggests that RIPK2 may modulate caspase-1 indirectly in our model. In this regard, autophagy was recently described to negatively regulate inflammasome activation during bacterial infections^{17, 35, 36}. We therefore hypothesized that Ripk2^−/− and Nod2^−/− BMDCs might be defective in autophagy induction in response to IAV infection. LC3 lipidation is an important step in the formation of autophagosomes. Therefore, one method to examine autophagic activity is to examine the level of LC3-II (lipidated form). In agreement with our hypothesis, Ripk2^−/− and Nod2^−/− BMDCs had significantly lower amounts of LC3-II protein than WT BMDCs infected with either PR8 or ×31 (Fig. 6a and Supplementary Fig. 4a). It should be noted that turnover of autophagosomes, which results from their fusion with lysosomes, was not affected by IAV infection in WT or Ripk2^−/− BMDCs as treatment with the lysosomal fusion inhibitor chloroquine resulted in the expected accumulation of LC3-II⁺ puncta by immunofluorescence and LC3-II conversion by Western blot (Figure 6b–c and Supplementary Fig. 4b). Finally, we counted autophagosomes by electron microscopy and determined that Ripk2^−/− BMDCs had significantly fewer autophagosomes following IAV infection than WT BMDCs (Supplementary Fig. 4c–e). To establish the potential role of differential autophagy induction in caspase-1 activation, we compared WT, Ripk2^−/− and LysM-Cre⁺Atg7^flox/flox bone marrow-derived macrophages (BMDMs) which lack a critical component of the autophagy machinery in myeloid cells. Both Ripk2^−/−and Atg7^−/− BMDMs responded to IAV infection with enhanced caspase-1 activation relative to WT macrophages (Supplementary Fig. 4f). To further confirm the role of autophagy in Ripk2^−/− cells, we treated Ripk2^−/− BMDCs with rapamycin to force the induction of autophagy. Following rapamycin treatment, IL-18 production and caspase-1 activation in Ripk2^−/− BMDCs was diminished to WT levels (Fig. 6d–e). We next verified the role of autophagy in vivo and found that LC3-II protein was reduced in the lungs of IAV infected Ripk2^−/− mice as well (Fig. 6f). We also examined the effects of inducing autophagy in vivo with rapamycin treatment in IAV-infected Ripk2^−/− mice and observed that rapamycin treatment dampened neutrophil recruitment and inflammatory cytokine production in Ripk2^−/−mice (Fig. 6g–h). Together, these results demonstrate that RIPK2 negatively regulates inflammasome activation, inflammatory cytokine production and neutrophil recruitment by inducing autophagy in IAV-infected cells and mice.

Viral RNA genomes activate autophagy through NOD2-RIPK2

To examine the mechanism by which NOD2 and RIPK2 induce mitophagy, we expressed RIPK2 and NOD2 in the presence or absence of IAV infection and determined that infection induces an interaction between RIPK2 and NOD2 (Supplementary Fig. 5a). Next, we determined the viral ligand that triggers NOD2-RIPK2 signaling. The cytosolic delivery of viral RNA analogues (poly I:C + lyovec transfection) was capable of inducing LC3-II conversion in a RIPK2- and NOD2- dependent manner (Supplementary Fig. 5b). However, extracellular delivery of naked poly I:C induced LC3-II conversion independently of the RIPK2 pathway (Supplementary Fig. 5c). We also transfected BMDCs with purified viral RNA (vRNA) genomes and determined that IAV vRNA induces autophagy in a RIPK2 dependent manner (Supplementary Fig. 5d). Finally, transfection of IAV vRNA followed by ATP treatment resulted in enhanced IL-18 production in both Ripk2^−/− and Nod2^−/− BMDCs (Supplementary Fig. 5e). These data indicate that cytosolic viral genomes activate the cytosolic NOD2-RIPK2 signaling pathway to trigger autophagy.

NOD2-RIPK2 mediated mitophagy regulates the inflammasome

In addition to preserving cellular energy stores, autophagy is linked to the removal of damaged mitochondria. Notably, mitochondrial damage and the subsequent release of reactive oxygen species (ROS) is linked to activation of the NLRP3 inflammasome^34,37. We thus hypothesized that defective mitophagy induction in IAV-infected RIPK2-deficient cells triggered increased inflammasome activation due to the accumulation of damaged mitochondria. To investigate this possibility, we examined mitochondrial superoxide (SOX) production following IAV infection. IAV infection increased mitochondrial SOX production in Ripk2^−/− and Nod2^−/− BMDCs to a markedly higher extent than in WT BMDCs (Fig. 7a and Supplementary Fig. 6a–b). To further confirm that mitochondrial damage was responsible for our observed phenotype we also examined mitochondrial SOX during Listeria infection. Although Listeria infection generated a more robust production of SOX compared to IAV, there was no difference between WT and Ripk2^−/− cells (Fig. 7b and Supplementary Fig. 6c). We also counted damaged mitochondria (disrupted cristae) by electron microscopy and determined that IAV infected Ripk2^−/− BMDC have a higher ratio of damaged to healthy mitochondria than WT controls (Supplementary Fig. 6d). We further confirmed the role of mitophagy in our model by staining BMDCs infected with IAV or Listeria with MitoTracker Green and examined the fluorescence intensity by flow cytometry. Cells with higher geometric mean fluorescence intensity (MFI) indicate an accumulation of mitochondria. Although Ripk2^−/− cells tended to have more mitochondria even before infection, we observed that IAV infected Ripk2^−/− BMDC but not WT had an increase in mitochondria, and furthermore, this increase was not seen with Listeria infected Ripk2^−/− cells (Fig. 7c and Supplementary Fig. 7a–c). We also confirmed these findings by isolating DNA from BMDCs and comparing the ratio of nuclear DNA to mitochondrial DNA. Using this approach, we also determined that there are more mitochondria in IAV infected Ripk2^−/− BMDCs compared to WT BMDCs (Supplementary Fig. 7d). As mitophagy appears to be defective in Ripk2^−/− cells, resulting in greater production of mitochondrial SOX, we treated Ripk2^−/− BMDCs with N-acetyl cysteine (NAC), a general ROS inhibitor, and found significantly reduced IL-18 levels and caspase-1 activation relative to control-treated Ripk2^−/− BMDCs that were infected with IAV (Fig. 7d–e). Similar results were obtained with the specific mitochondrial SOX inhibitor mitoTEMPO (Fig. 7f). To confirm the specificity of these results, we also treated Ripk2^−/−Listeria-infected BMDCs with NAC. In agreement with the role for the AIM2 inflammasome in Listeria infection, we observed little or no effect of NAC treatment during Listeria infection (Supplementary Fig. 7e). Finally, we verified directly that mitophagy was reduced during IAV infection in Ripk2^−/− BMDCs by staining for LC3-II and mitochondria. The ratio of LC3-II puncta colocalized with mitochondria by confocal microscopy was significantly reduced in Ripk2^−/− compared to WT BMDCs (Fig. 7g). These results indicate that the induction of mitophagy during IAV infection, which is regulated by NOD2 and RIPK2, is essential for the clearance of damaged mitochondria. In the absence of RIPK2 or NOD2, elevated mitochondrial SOX production enhances NLRP3 inflammasome activation.

RIPK2 regulates mitophagy through its kinase activity

It was previously reported that RIPK2 regulates autophagy through the autophagy protein ATG16L1^{22, 23}. However, another group found no interaction between RIPK2 and ATG16L1³⁷. To examine how RIPK2 regulates mitophagy during viral infection, we first examined the possibility that RIPK2 might interact with ATG16L1. We expressed RIPK2 and the autophagy related protein ATG16L1 in the presence or absence of NOD2 and IAV infection. IAV infection was required for association of RIPK2, NOD2 and ATG16L1 (Supplementary Fig. 8a). However, we were unable to coimmunoprecipitate endogenous RIPK2 and ATG16L1 or detect any significant colocalization by immunofluorescence confocal microscopy in BMDCs during IAV infection (data not shown). Although the overexpression data above suggests the interaction with ATG16L1 may play an important role during IAV infection, this remains to be verified under endogenous conditions. Furthermore, an interaction between RIPK2 and ATG16L1 still does not address the upstream signaling which initiates mitophagy during infection. To begin to address the exact mechanism by which mitophagy is regulated by RIPK2 in our model, we examined the role of RIPK2's kinase activity during mitophagy induction. Treatment of WT BMDCs with the p38 and RIPK2 inhibitor SB203580 resulted in reduced LC3-II conversion and increased caspase-1 activation as well as increased IL-18 production (Fig. 8a–b). We confirmed that this effect was due to RIPK2 inhibition by examining autophagy in p38^Flox/Flox BMDCs. p38 deficiency did not impact LC3-II conversion during IAV infection, indicating the effect of SB203580 to be mediated by RIPK2 inhibition (Supplementary Fig. 8b). Next, we examined autophagy proteins known to be activated by phosphorylation and discovered that phosphorylation of ULK1 at Ser555 was reduced in Ripk2^−/− BMDCs during IAV infection although upregulation of total ULK1 was not affected (Fig. 8c). Phosphorylation at Ser555 of ULK1 is known to be important for its activation and ULK1 is known to play an important role in mitophagy, especially during hematopoietic development³⁸. We therefore examined Ulk1^−/− BMDCs and found that they also have increased caspase-1 activation following IAV infection and this increased caspase-1 activation was suppressed by NAC treatment (Fig. 8d). Furthermore, Ulk1^−/− BMDCs have more mitochondrial damage and increased mitochondrial mass compared to WT BMDCs (Fig. 8e–f and Supplementary Fig. 8c). Collectively, these data demonstrate that NOD2 and RIPK2 interact with each other in response to IAV vRNA genomes and regulate mitophagy in a RIPK2 kinase and ULK1 dependent manner. Mitophagy is then responsible for removing damaged mitochondria and preventing excessive NLRP3 activation and IL-18 production (Fig. 8g).

Virus, bacteria and ligands

The influenza A/Puerto Rico/8/34 H1N1 virus (PR8) was generated by an eight-plasmid reverse genetics system⁴⁶. X31 virus was prepared similarly. Stocks were propagated no more than twice by allantoic inoculation of 10-day-old embryonated hen's eggs with seed virus diluted 1:10⁴. Listeria monocytogenes was grown in brain heart infusion medium at 37 °C overnight and then subcultured to mid-log phase for infections. Poly I:C and Poly I:C lyovec were obtained from Invivogen and used at the indicated concentrations. IAV genomes, or cellular mRNA controls, were purified from viral stocks or BMDCs using Trizol LS and transfected using Lipofectamine 2000 according to manufactures instructions.

Measuring virus titers

Near confluent 9.6 cm² monolayers of Madin-Darby Canine Kidney (MDCK) cells were infected with 1 mL aliquots of 6×10-fold dilutions of lung homogenate, washed and overlayed with 3 ml minimum essential medium (MEM) containing 1 mg/ml L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Sigma Aldrich) and 1.0% Sea Plaque agarose (Lonza). After 3 days, plaques were visualized with crystal violet.

Mice

All mice were maintained at SJCRH and were fully on the C57BL/6J (B6) background. Nod2^−/−, LysM-Cre⁺ Atg7^flox/flox GFP-LC3⁺, p38^flox/flox (Mapk14^flox/flox Rosa26-Cre-ER^T2), Ulk1^−/− and Ripk2^{−/− 24, 32, 40, 47–49} mice have been reported previously and Ripk2^−/− × IL-18^−/− double deficient mice were generated by backcrossing. All mice were housed in a specific pathogen free (SPF) facility and experiments were conducted under protocols approved by the St. Jude Children's Research Hospital Committee on Use and Care of Animals.

Virus infection and sampling

Mice were anaesthetized with Avertin (2,2,2-tribromoethanol) and infected i.n. with the indicated dose of the PR8 virus in 30μL of endotoxin-free PBS. In the case of IL-18 neutralization, mice were injected as indicated with 200μL of rabbit antiserum against mouse IL-18 (C. A. Dinarello). For in vivo autophagy induction, mice were injected with 600μg/kg of rapamycin on D0, 1 and 2 and samples were collected 6h after the final injection. Mice were either weighed and monitored for severe illness daily for a period of 14–16d, or taken at various intervals for sampling the whole lung. The right lungs were processed by mincing and passing through a 70μm cell strainer using a total of 4ml HBSS. Following light centrifugation, total cell numbers per lung or lymph node were determined and stained for flow cytometry. In addition, HBSS supernatant was used for virus titration and cytokine analysis. The left lungs were used for histopathological analysis or ground in RIPA buffer + protease inhibitor + phosphatase inhibitor (Calbiochem) followed by boiling in sodium dodecyl sulfate (SDS) sample buffer and examination by Western blot.

Flow cytometry

Aliquots of whole lung populations (as above) were stained for myeloid cells with anti-CD11b, anti-MHC Class II, anti-GR1, anti-CD11c (Biolegend, M1/70, M5/114.15.2, RB6-8C5, N418) MAbs, and annexin V after blocking the Fc receptor with anti-CD32/CD16 MAb (from BD Pharmingen) at 4°C, and analyzed by flow cytometry. (macrophages=CD11b⁺,CD11c⁻,GR1^−,lo; granulocytes=CD11b^hi,CD11c⁻,GR1^hi; DCs=CD11c⁺, MHC Class II⁺) Additionally, lymphocytes were stained with anti-CD19, anti-TCRβ, anti-CD8, and anti-CD4 (Biolegend, 6D5, H57–597, 53–6.7, RM4–5). NK and NKT cells were stained with anti-CD3 and anti-NK1.1 (Biolegend, 145-2C11, PK136). For NK, NKT and CD8⁺ T cell activation experiments, anti-CD3 and anti-NK1.1 or anti-TCRβ, anti-CD8 and anti-CD44 (Biolegend, IM7) markers were used. Intracellular IFN-γ (Biolegned, XMG1.2) in CD8⁺ T cells was measured after stimulation of whole lung cells for 5h at 37°C and 5% CO₂ with a cocktail of IAV peptides (PB1, PB1-F2, NP, PA, M1) at 1ug/ml each with monensin. NK and NKT cells were similarly restimulated with 10ng/ml IL-12. Cell numbers are 1/10 of the total number present in the right lung lobes or 1/3 of the number present in the mediastinal lymph node

For staining mitochondria, BMDCs were uninfected as controls or infected for 6h with Listeria or 18–24h with PR8 influenza and then stained for 30 min in fresh complete RPMI1640 medium containing 5μM MitoSOX or 2μM MitoTracker Green. Cells were then washed in HBSS, resuspended in FACS buffer and analyzed immediately by flow cytometry.

Assaying for cytokines and chemokines

Mouse cytokines and chemokines in HBSS supernatants from whole lung homogenates were measured using the Millipore 22-multiplex assay following the manufacturer's instructions. In addition, lung samples and supernatants from BMDC cultured in vitro were assayed by ELISA for IL-18 (MBL International, Woburn, MA) and IL-1β or IFN-γ (eBiosciences). All cytokines from lung homogenates were normalized to the total protein in the homogenate by BCA protein assay (Pierce) and expressed as cytokine concentration per milligram total protein (pg/mg t.p.)

Histopathological analysis, transmission electron microscopy and confocal microscopy

Formalin-preserved left lungs were embedded in paraffin and processed by standard techniques. Longitudinal 5μm sections were stained with H&E and examined by an experienced pathologist blinded to the experimental groups. For immunohistochemistry, formalin-fixed paraffin-embedded tissues were cut into 4-μm sections and slides were stained with Abs to identify IL-18 (MBL International, Woburn, MA, clone 39–3F) and neutrophils (7/4 monoclonal antibody, Invitrogen).

For transmission electron microscopy, cells were fixed in a solution of 2.0% paraformaldahyde + 2.5% glutaraldehyde in 0.1 m cacodylate buffer (pH 7.4). Cells were then embedded and sectioned for transmission electron microscopy by the Cell and Tissue Imaging Core Facility of St. Jude Children's Research Hospital.

For examination of LC3⁺ puncta numbers or colocalization with mitochondria by confocal microscopy, BMDCs were infected with IAV at 10MOI and/or treated with 50μM chloroquine (Calbiochem) 1h after infection. After 18h, cells were treated with 100nM MitoTracker Orange CMTMRos for 30min. Cells were then fixed in 4% paraformaldehyde in PBS+100mM HEPES buffer. Cells were permeablized with methanol at −20°C for 5 min, washed 3× in PBS, and then blocked using 1× assay buffer (eBiosciences) + 0.1%Triton ×100 for 1h. Cells were stained overnight at 4°C with a 1:500 dilution of anti-LC3B antibody (Novus Biologicals, NB600–1384) in blocking buffer. Cells were washed 3× and then stained with Alexa488 donkey anti-rabbit secondary (Invitrogen, A-21206) for 2h in blocking buffer. Cells were washed an additional 3× and then mounted using Prolong GOLD with DAPI (Invitrogen) and examined on a Nikon C1Si laser scanning confocal microscope. Data analysis was performed using ImageJ.

Generation of bone marrow chimeras

WT (CD45.1) and Ripk2^−/− (CD45.2) mice, were lethally irradiated with a split dose of 1200 rads (800,400), then injected with 5 × 10⁶ bone marrow (BM) cells from the indicated donors. Mice were allowed to recover for 6 weeks to ensure successful engraftment. The extent of BM reconstitution was verified by staining lymphocytes with anti-CD45.1 APC and anti-CD45.2 FITC (Biolegend, A20, 104) for congenic markers, and was always > 90%.

Cell signaling, caspase-1 activation and IL-18 production in vitro

WT, Nod2^−/− and Ripk2^−/− bone marrow derived dendritic cells (BMDC) were differentiated in complete RPMI containing 10% heat-inactivated FBS and supplemented with 20ng/ml murine GM-CSF at 37°C in a humidified atmosphere containing 5% CO₂ for 7 days. Cells were mock infected, or infected with 10 MOI PR8 or 10 MOI ×31 for 24h or 10 MOI Listeria monocytogenes for 4–6h in antibiotic free medium. For N-acetyl cysteine (NAC, Sigma-Aldrich) or mitoTEMPO (Enzo Life Sciences) treatment, BMDCs were infected with 10 MOI PR8 for 1h and then fresh RPMI 1640 with 10% FBS containing the indicated concentrations of inhibitors were added. Other inhibitors, including glyburide (Sigma-Aldrich), rapamycin or SB203580 (Calbiochem) were also added 1h after infection. 18–24h later, supernatants were collected for ELISA and cells were lysed with RIPA buffer + protease inhibitor + phosphatase inhibitor (Calbiochem), followed by boiling in sodium dodecyl sulfate (SDS) sample buffer and examination by Western blot. Anti-caspase-1 (Adipogen, AG-20B-0042), anti-ULK1 (Sigma-Aldrich, A7481), anti-LC3 (Novus Biologicals, NB600–1384), and anti-pUlk1 Ser555 (Cell Signaling Technologies, D1H4), were used for Western blot detection and equal loading was verified by blotting with anti-GAPDH or anti-β-Tubulin (Cell signaling Technologies, D16H11, 9F3). HRP-labeled anti-rabbit antibodies were obtained from Jackson Immuno Research.

Overexpression and Coimmunoprecipitation

293T cells were transfected using Lipofectamine 2000 (Invitrogen) or Xfect (Clontech) according to the manufacturer's protocol and harvested 24h after transfection for examination by coimmunoprecipitation (CoIP) and/or by Western blot. For CoIP, 3–4 10cm wells each were transfected with pCMV6-Kan/Neo-RIPK2 (Origene), pBK-flag-NOD2, or pCMV-3xMyc-ATG16L1 and cells were lysed after 24h in DPBS with 1% NP40 + protease inhibitors + phosphatase inhibitors (Calbiochem). In some experiments, 293T cells were also infected with 10 MOI influenza A/PR/8/34 H1N1 1h prior to transfection. CoIP was performed overnight at 4°C with protein A/G PLUS agarose beads (SantaCruz) and rabbit anti-RIPK2 (SantaCruz, H300), mouse anti-Flag (Sigma, M2), and rabbit anti-Myc (SantaCruz, A14) antibodies. Western blot was then performed with rabbit anti-RIPK2, rabbit anti-Flag (Sigma, F7425), rabbit anti-Myc and Exactacruz HRP anti-rabbit secondary (SantaCruz, SC-45043).

Statistical analysis

Data are represented as mean ± SEM. Statistical significance was determined by a Student t-test, one-way ANOVA for multiple comparisons, and Kaplan-Meier Survival Plot and LogRank Test for survival data. Data for LC3⁺ puncta was not normally distributed and data were analyzed by the Mann Whitney test. P values ≤0.05 were considered statistically significant.